blob: 11e25c8a1c4e68135b5f32a941bb8e8be4b66618 [file] [log] [blame]
drh75897232000-05-29 14:26:00 +00001/*
drh75897232000-05-29 14:26:00 +00002** This file contains all sources (including headers) to the LEMON
3** LALR(1) parser generator. The sources have been combined into a
drh960e8c62001-04-03 16:53:21 +00004** single file to make it easy to include LEMON in the source tree
5** and Makefile of another program.
drh75897232000-05-29 14:26:00 +00006**
drhb19a2bc2001-09-16 00:13:26 +00007** The author of this program disclaims copyright.
drh75897232000-05-29 14:26:00 +00008*/
9#include <stdio.h>
drhf9a2e7b2003-04-15 01:49:48 +000010#include <stdarg.h>
drh75897232000-05-29 14:26:00 +000011#include <string.h>
12#include <ctype.h>
drh8b582012003-10-21 13:16:03 +000013#include <stdlib.h>
drhe9278182007-07-18 18:16:29 +000014#include <assert.h>
drh75897232000-05-29 14:26:00 +000015
drhc56fac72015-10-29 13:48:15 +000016#define ISSPACE(X) isspace((unsigned char)(X))
17#define ISDIGIT(X) isdigit((unsigned char)(X))
18#define ISALNUM(X) isalnum((unsigned char)(X))
19#define ISALPHA(X) isalpha((unsigned char)(X))
20#define ISUPPER(X) isupper((unsigned char)(X))
21#define ISLOWER(X) islower((unsigned char)(X))
22
23
drh75897232000-05-29 14:26:00 +000024#ifndef __WIN32__
25# if defined(_WIN32) || defined(WIN32)
drhf2f105d2012-08-20 15:53:54 +000026# define __WIN32__
drh75897232000-05-29 14:26:00 +000027# endif
28#endif
29
rse8f304482007-07-30 18:31:53 +000030#ifdef __WIN32__
drhdf609712010-11-23 20:55:27 +000031#ifdef __cplusplus
32extern "C" {
33#endif
34extern int access(const char *path, int mode);
35#ifdef __cplusplus
36}
37#endif
rse8f304482007-07-30 18:31:53 +000038#else
39#include <unistd.h>
40#endif
41
drh75897232000-05-29 14:26:00 +000042/* #define PRIVATE static */
43#define PRIVATE
44
45#ifdef TEST
46#define MAXRHS 5 /* Set low to exercise exception code */
47#else
48#define MAXRHS 1000
49#endif
50
mistachkin17df7612019-06-03 15:09:25 +000051extern void memory_error();
drhf5c4e0f2010-07-18 11:35:53 +000052static int showPrecedenceConflict = 0;
drhe9278182007-07-18 18:16:29 +000053static char *msort(char*,char**,int(*)(const char*,const char*));
drh75897232000-05-29 14:26:00 +000054
drh87cf1372008-08-13 20:09:06 +000055/*
56** Compilers are getting increasingly pedantic about type conversions
57** as C evolves ever closer to Ada.... To work around the latest problems
58** we have to define the following variant of strlen().
59*/
60#define lemonStrlen(X) ((int)strlen(X))
61
drh898799f2014-01-10 23:21:00 +000062/*
63** Compilers are starting to complain about the use of sprintf() and strcpy(),
64** saying they are unsafe. So we define our own versions of those routines too.
65**
66** There are three routines here: lemon_sprintf(), lemon_vsprintf(), and
drh25473362015-09-04 18:03:45 +000067** lemon_addtext(). The first two are replacements for sprintf() and vsprintf().
drh898799f2014-01-10 23:21:00 +000068** The third is a helper routine for vsnprintf() that adds texts to the end of a
69** buffer, making sure the buffer is always zero-terminated.
70**
71** The string formatter is a minimal subset of stdlib sprintf() supporting only
72** a few simply conversions:
73**
74** %d
75** %s
76** %.*s
77**
78*/
79static void lemon_addtext(
80 char *zBuf, /* The buffer to which text is added */
81 int *pnUsed, /* Slots of the buffer used so far */
82 const char *zIn, /* Text to add */
drh61f92cd2014-01-11 03:06:18 +000083 int nIn, /* Bytes of text to add. -1 to use strlen() */
84 int iWidth /* Field width. Negative to left justify */
drh898799f2014-01-10 23:21:00 +000085){
86 if( nIn<0 ) for(nIn=0; zIn[nIn]; nIn++){}
drhecaa9d32014-01-11 03:27:37 +000087 while( iWidth>nIn ){ zBuf[(*pnUsed)++] = ' '; iWidth--; }
drh898799f2014-01-10 23:21:00 +000088 if( nIn==0 ) return;
89 memcpy(&zBuf[*pnUsed], zIn, nIn);
90 *pnUsed += nIn;
drhecaa9d32014-01-11 03:27:37 +000091 while( (-iWidth)>nIn ){ zBuf[(*pnUsed)++] = ' '; iWidth++; }
drh898799f2014-01-10 23:21:00 +000092 zBuf[*pnUsed] = 0;
93}
94static int lemon_vsprintf(char *str, const char *zFormat, va_list ap){
mistachkin7a429652014-01-14 10:17:21 +000095 int i, j, k, c;
drh898799f2014-01-10 23:21:00 +000096 int nUsed = 0;
97 const char *z;
98 char zTemp[50];
99 str[0] = 0;
100 for(i=j=0; (c = zFormat[i])!=0; i++){
101 if( c=='%' ){
drh61f92cd2014-01-11 03:06:18 +0000102 int iWidth = 0;
103 lemon_addtext(str, &nUsed, &zFormat[j], i-j, 0);
drh898799f2014-01-10 23:21:00 +0000104 c = zFormat[++i];
drhc56fac72015-10-29 13:48:15 +0000105 if( ISDIGIT(c) || (c=='-' && ISDIGIT(zFormat[i+1])) ){
drh61f92cd2014-01-11 03:06:18 +0000106 if( c=='-' ) i++;
drhc56fac72015-10-29 13:48:15 +0000107 while( ISDIGIT(zFormat[i]) ) iWidth = iWidth*10 + zFormat[i++] - '0';
drh61f92cd2014-01-11 03:06:18 +0000108 if( c=='-' ) iWidth = -iWidth;
109 c = zFormat[i];
110 }
drh898799f2014-01-10 23:21:00 +0000111 if( c=='d' ){
112 int v = va_arg(ap, int);
113 if( v<0 ){
drh61f92cd2014-01-11 03:06:18 +0000114 lemon_addtext(str, &nUsed, "-", 1, iWidth);
drh898799f2014-01-10 23:21:00 +0000115 v = -v;
116 }else if( v==0 ){
drh61f92cd2014-01-11 03:06:18 +0000117 lemon_addtext(str, &nUsed, "0", 1, iWidth);
drh898799f2014-01-10 23:21:00 +0000118 }
119 k = 0;
120 while( v>0 ){
121 k++;
122 zTemp[sizeof(zTemp)-k] = (v%10) + '0';
123 v /= 10;
124 }
drh61f92cd2014-01-11 03:06:18 +0000125 lemon_addtext(str, &nUsed, &zTemp[sizeof(zTemp)-k], k, iWidth);
drh898799f2014-01-10 23:21:00 +0000126 }else if( c=='s' ){
127 z = va_arg(ap, const char*);
drh61f92cd2014-01-11 03:06:18 +0000128 lemon_addtext(str, &nUsed, z, -1, iWidth);
drh898799f2014-01-10 23:21:00 +0000129 }else if( c=='.' && memcmp(&zFormat[i], ".*s", 3)==0 ){
130 i += 2;
131 k = va_arg(ap, int);
132 z = va_arg(ap, const char*);
drh61f92cd2014-01-11 03:06:18 +0000133 lemon_addtext(str, &nUsed, z, k, iWidth);
drh898799f2014-01-10 23:21:00 +0000134 }else if( c=='%' ){
drh61f92cd2014-01-11 03:06:18 +0000135 lemon_addtext(str, &nUsed, "%", 1, 0);
drh898799f2014-01-10 23:21:00 +0000136 }else{
137 fprintf(stderr, "illegal format\n");
138 exit(1);
139 }
140 j = i+1;
141 }
142 }
drh61f92cd2014-01-11 03:06:18 +0000143 lemon_addtext(str, &nUsed, &zFormat[j], i-j, 0);
drh898799f2014-01-10 23:21:00 +0000144 return nUsed;
145}
146static int lemon_sprintf(char *str, const char *format, ...){
147 va_list ap;
148 int rc;
149 va_start(ap, format);
150 rc = lemon_vsprintf(str, format, ap);
151 va_end(ap);
152 return rc;
153}
154static void lemon_strcpy(char *dest, const char *src){
155 while( (*(dest++) = *(src++))!=0 ){}
156}
157static void lemon_strcat(char *dest, const char *src){
158 while( *dest ) dest++;
159 lemon_strcpy(dest, src);
160}
161
162
icculus9e44cf12010-02-14 17:14:22 +0000163/* a few forward declarations... */
164struct rule;
165struct lemon;
166struct action;
167
drhe9278182007-07-18 18:16:29 +0000168static struct action *Action_new(void);
169static struct action *Action_sort(struct action *);
drh75897232000-05-29 14:26:00 +0000170
171/********** From the file "build.h" ************************************/
drh14d88552017-04-14 19:44:15 +0000172void FindRulePrecedences(struct lemon*);
173void FindFirstSets(struct lemon*);
174void FindStates(struct lemon*);
175void FindLinks(struct lemon*);
176void FindFollowSets(struct lemon*);
177void FindActions(struct lemon*);
drh75897232000-05-29 14:26:00 +0000178
179/********* From the file "configlist.h" *********************************/
icculus9e44cf12010-02-14 17:14:22 +0000180void Configlist_init(void);
181struct config *Configlist_add(struct rule *, int);
182struct config *Configlist_addbasis(struct rule *, int);
183void Configlist_closure(struct lemon *);
184void Configlist_sort(void);
185void Configlist_sortbasis(void);
186struct config *Configlist_return(void);
187struct config *Configlist_basis(void);
188void Configlist_eat(struct config *);
189void Configlist_reset(void);
drh75897232000-05-29 14:26:00 +0000190
191/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +0000192void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +0000193
194/****** From the file "option.h" ******************************************/
icculus9e44cf12010-02-14 17:14:22 +0000195enum option_type { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
196 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR};
drh75897232000-05-29 14:26:00 +0000197struct s_options {
icculus9e44cf12010-02-14 17:14:22 +0000198 enum option_type type;
199 const char *label;
drh75897232000-05-29 14:26:00 +0000200 char *arg;
icculus9e44cf12010-02-14 17:14:22 +0000201 const char *message;
drh75897232000-05-29 14:26:00 +0000202};
icculus9e44cf12010-02-14 17:14:22 +0000203int OptInit(char**,struct s_options*,FILE*);
204int OptNArgs(void);
205char *OptArg(int);
206void OptErr(int);
207void OptPrint(void);
drh75897232000-05-29 14:26:00 +0000208
209/******** From the file "parse.h" *****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000210void Parse(struct lemon *lemp);
drh75897232000-05-29 14:26:00 +0000211
212/********* From the file "plink.h" ***************************************/
icculus9e44cf12010-02-14 17:14:22 +0000213struct plink *Plink_new(void);
214void Plink_add(struct plink **, struct config *);
215void Plink_copy(struct plink **, struct plink *);
216void Plink_delete(struct plink *);
drh75897232000-05-29 14:26:00 +0000217
218/********** From the file "report.h" *************************************/
icculus9e44cf12010-02-14 17:14:22 +0000219void Reprint(struct lemon *);
220void ReportOutput(struct lemon *);
drhfe03dac2019-11-26 02:22:39 +0000221void ReportTable(struct lemon *, int, int);
icculus9e44cf12010-02-14 17:14:22 +0000222void ReportHeader(struct lemon *);
223void CompressTables(struct lemon *);
224void ResortStates(struct lemon *);
drh75897232000-05-29 14:26:00 +0000225
226/********** From the file "set.h" ****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000227void SetSize(int); /* All sets will be of size N */
228char *SetNew(void); /* A new set for element 0..N */
229void SetFree(char*); /* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +0000230int SetAdd(char*,int); /* Add element to a set */
231int SetUnion(char *,char *); /* A <- A U B, thru element N */
drh75897232000-05-29 14:26:00 +0000232#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
233
234/********** From the file "struct.h" *************************************/
235/*
236** Principal data structures for the LEMON parser generator.
237*/
238
drhaa9f1122007-08-23 02:50:56 +0000239typedef enum {LEMON_FALSE=0, LEMON_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000240
241/* Symbols (terminals and nonterminals) of the grammar are stored
242** in the following: */
icculus9e44cf12010-02-14 17:14:22 +0000243enum symbol_type {
244 TERMINAL,
245 NONTERMINAL,
246 MULTITERMINAL
247};
248enum e_assoc {
drh75897232000-05-29 14:26:00 +0000249 LEFT,
250 RIGHT,
251 NONE,
252 UNK
icculus9e44cf12010-02-14 17:14:22 +0000253};
254struct symbol {
255 const char *name; /* Name of the symbol */
256 int index; /* Index number for this symbol */
257 enum symbol_type type; /* Symbols are all either TERMINALS or NTs */
258 struct rule *rule; /* Linked list of rules of this (if an NT) */
259 struct symbol *fallback; /* fallback token in case this token doesn't parse */
260 int prec; /* Precedence if defined (-1 otherwise) */
261 enum e_assoc assoc; /* Associativity if precedence is defined */
drh75897232000-05-29 14:26:00 +0000262 char *firstset; /* First-set for all rules of this symbol */
263 Boolean lambda; /* True if NT and can generate an empty string */
drhc4dd3fd2008-01-22 01:48:05 +0000264 int useCnt; /* Number of times used */
drh75897232000-05-29 14:26:00 +0000265 char *destructor; /* Code which executes whenever this symbol is
266 ** popped from the stack during error processing */
drh0f832dd2016-08-16 16:46:40 +0000267 int destLineno; /* Line number for start of destructor. Set to
268 ** -1 for duplicate destructors. */
drh75897232000-05-29 14:26:00 +0000269 char *datatype; /* The data type of information held by this
270 ** object. Only used if type==NONTERMINAL */
271 int dtnum; /* The data type number. In the parser, the value
272 ** stack is a union. The .yy%d element of this
273 ** union is the correct data type for this object */
drh539e7412018-04-21 20:24:19 +0000274 int bContent; /* True if this symbol ever carries content - if
275 ** it is ever more than just syntax */
drhfd405312005-11-06 04:06:59 +0000276 /* The following fields are used by MULTITERMINALs only */
277 int nsubsym; /* Number of constituent symbols in the MULTI */
278 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000279};
280
281/* Each production rule in the grammar is stored in the following
282** structure. */
283struct rule {
284 struct symbol *lhs; /* Left-hand side of the rule */
icculus9e44cf12010-02-14 17:14:22 +0000285 const char *lhsalias; /* Alias for the LHS (NULL if none) */
drhb4960992007-10-05 16:16:36 +0000286 int lhsStart; /* True if left-hand side is the start symbol */
drh75897232000-05-29 14:26:00 +0000287 int ruleline; /* Line number for the rule */
288 int nrhs; /* Number of RHS symbols */
289 struct symbol **rhs; /* The RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +0000290 const char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
drh75897232000-05-29 14:26:00 +0000291 int line; /* Line number at which code begins */
icculus9e44cf12010-02-14 17:14:22 +0000292 const char *code; /* The code executed when this rule is reduced */
drh4dd0d3f2016-02-17 01:18:33 +0000293 const char *codePrefix; /* Setup code before code[] above */
294 const char *codeSuffix; /* Breakdown code after code[] above */
drh75897232000-05-29 14:26:00 +0000295 struct symbol *precsym; /* Precedence symbol for this rule */
296 int index; /* An index number for this rule */
drh4ef07702016-03-16 19:45:54 +0000297 int iRule; /* Rule number as used in the generated tables */
drhe94006e2019-12-10 20:41:48 +0000298 Boolean noCode; /* True if this rule has no associated C code */
299 Boolean codeEmitted; /* True if the code has been emitted already */
drh75897232000-05-29 14:26:00 +0000300 Boolean canReduce; /* True if this rule is ever reduced */
drh756b41e2016-05-24 18:55:08 +0000301 Boolean doesReduce; /* Reduce actions occur after optimization */
drhe94006e2019-12-10 20:41:48 +0000302 Boolean neverReduce; /* Reduce is theoretically possible, but prevented
303 ** by actions or other outside implementation */
drh75897232000-05-29 14:26:00 +0000304 struct rule *nextlhs; /* Next rule with the same LHS */
305 struct rule *next; /* Next rule in the global list */
306};
307
308/* A configuration is a production rule of the grammar together with
309** a mark (dot) showing how much of that rule has been processed so far.
310** Configurations also contain a follow-set which is a list of terminal
311** symbols which are allowed to immediately follow the end of the rule.
312** Every configuration is recorded as an instance of the following: */
icculus9e44cf12010-02-14 17:14:22 +0000313enum cfgstatus {
314 COMPLETE,
315 INCOMPLETE
316};
drh75897232000-05-29 14:26:00 +0000317struct config {
318 struct rule *rp; /* The rule upon which the configuration is based */
319 int dot; /* The parse point */
320 char *fws; /* Follow-set for this configuration only */
321 struct plink *fplp; /* Follow-set forward propagation links */
322 struct plink *bplp; /* Follow-set backwards propagation links */
323 struct state *stp; /* Pointer to state which contains this */
icculus9e44cf12010-02-14 17:14:22 +0000324 enum cfgstatus status; /* used during followset and shift computations */
drh75897232000-05-29 14:26:00 +0000325 struct config *next; /* Next configuration in the state */
326 struct config *bp; /* The next basis configuration */
327};
328
icculus9e44cf12010-02-14 17:14:22 +0000329enum e_action {
330 SHIFT,
331 ACCEPT,
332 REDUCE,
333 ERROR,
334 SSCONFLICT, /* A shift/shift conflict */
335 SRCONFLICT, /* Was a reduce, but part of a conflict */
336 RRCONFLICT, /* Was a reduce, but part of a conflict */
337 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
338 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
drh3bd48ab2015-09-07 18:23:37 +0000339 NOT_USED, /* Deleted by compression */
340 SHIFTREDUCE /* Shift first, then reduce */
icculus9e44cf12010-02-14 17:14:22 +0000341};
342
drh75897232000-05-29 14:26:00 +0000343/* Every shift or reduce operation is stored as one of the following */
344struct action {
345 struct symbol *sp; /* The look-ahead symbol */
icculus9e44cf12010-02-14 17:14:22 +0000346 enum e_action type;
drh75897232000-05-29 14:26:00 +0000347 union {
348 struct state *stp; /* The new state, if a shift */
349 struct rule *rp; /* The rule, if a reduce */
350 } x;
drhc173ad82016-05-23 16:15:02 +0000351 struct symbol *spOpt; /* SHIFTREDUCE optimization to this symbol */
drh75897232000-05-29 14:26:00 +0000352 struct action *next; /* Next action for this state */
353 struct action *collide; /* Next action with the same hash */
354};
355
356/* Each state of the generated parser's finite state machine
357** is encoded as an instance of the following structure. */
358struct state {
359 struct config *bp; /* The basis configurations for this state */
360 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000361 int statenum; /* Sequential number for this state */
drh711c9812016-05-23 14:24:31 +0000362 struct action *ap; /* List of actions for this state */
drh8b582012003-10-21 13:16:03 +0000363 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
364 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
drh3bd48ab2015-09-07 18:23:37 +0000365 int iDfltReduce; /* Default action is to REDUCE by this rule */
366 struct rule *pDfltReduce;/* The default REDUCE rule. */
367 int autoReduce; /* True if this is an auto-reduce state */
drh75897232000-05-29 14:26:00 +0000368};
drh8b582012003-10-21 13:16:03 +0000369#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000370
371/* A followset propagation link indicates that the contents of one
372** configuration followset should be propagated to another whenever
373** the first changes. */
374struct plink {
375 struct config *cfp; /* The configuration to which linked */
376 struct plink *next; /* The next propagate link */
377};
378
379/* The state vector for the entire parser generator is recorded as
380** follows. (LEMON uses no global variables and makes little use of
381** static variables. Fields in the following structure can be thought
382** of as begin global variables in the program.) */
383struct lemon {
384 struct state **sorted; /* Table of states sorted by state number */
385 struct rule *rule; /* List of all rules */
drh4ef07702016-03-16 19:45:54 +0000386 struct rule *startRule; /* First rule */
drh75897232000-05-29 14:26:00 +0000387 int nstate; /* Number of states */
drh3bd48ab2015-09-07 18:23:37 +0000388 int nxstate; /* nstate with tail degenerate states removed */
drh75897232000-05-29 14:26:00 +0000389 int nrule; /* Number of rules */
drhce678c22019-12-11 18:53:51 +0000390 int nruleWithAction; /* Number of rules with actions */
drh75897232000-05-29 14:26:00 +0000391 int nsymbol; /* Number of terminal and nonterminal symbols */
392 int nterminal; /* Number of terminal symbols */
drh5c8241b2017-12-24 23:38:10 +0000393 int minShiftReduce; /* Minimum shift-reduce action value */
394 int errAction; /* Error action value */
395 int accAction; /* Accept action value */
396 int noAction; /* No-op action value */
397 int minReduce; /* Minimum reduce action */
398 int maxAction; /* Maximum action value of any kind */
drh75897232000-05-29 14:26:00 +0000399 struct symbol **symbols; /* Sorted array of pointers to symbols */
400 int errorcnt; /* Number of errors */
401 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000402 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000403 char *name; /* Name of the generated parser */
drhc2dbf352021-01-07 16:10:14 +0000404 char *arg; /* Declaration of the 3rd argument to parser */
drhfb32c442018-04-21 13:51:42 +0000405 char *ctx; /* Declaration of 2nd argument to constructor */
drh75897232000-05-29 14:26:00 +0000406 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000407 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000408 char *start; /* Name of the start symbol for the grammar */
409 char *stacksize; /* Size of the parser stack */
410 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000411 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000412 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000413 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000414 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000415 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000416 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000417 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000418 char *filename; /* Name of the input file */
419 char *outname; /* Name of the current output file */
420 char *tokenprefix; /* A prefix added to token names in the .h file */
421 int nconflict; /* Number of parsing conflicts */
drhc75e0162015-09-07 02:23:02 +0000422 int nactiontab; /* Number of entries in the yy_action[] table */
drh3a9d6c72017-12-25 04:15:38 +0000423 int nlookaheadtab; /* Number of entries in yy_lookahead[] */
drhc75e0162015-09-07 02:23:02 +0000424 int tablesize; /* Total table size of all tables in bytes */
drh75897232000-05-29 14:26:00 +0000425 int basisflag; /* Print only basis configurations */
drh0a34cf52020-07-03 15:41:08 +0000426 int printPreprocessed; /* Show preprocessor output on stdout */
drh34ff57b2008-07-14 12:27:51 +0000427 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000428 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000429 char *argv0; /* Name of the program */
430};
431
432#define MemoryCheck(X) if((X)==0){ \
433 extern void memory_error(); \
434 memory_error(); \
435}
436
437/**************** From the file "table.h" *********************************/
438/*
439** All code in this file has been automatically generated
440** from a specification in the file
441** "table.q"
442** by the associative array code building program "aagen".
443** Do not edit this file! Instead, edit the specification
444** file, then rerun aagen.
445*/
446/*
447** Code for processing tables in the LEMON parser generator.
448*/
drh75897232000-05-29 14:26:00 +0000449/* Routines for handling a strings */
450
icculus9e44cf12010-02-14 17:14:22 +0000451const char *Strsafe(const char *);
drh75897232000-05-29 14:26:00 +0000452
icculus9e44cf12010-02-14 17:14:22 +0000453void Strsafe_init(void);
454int Strsafe_insert(const char *);
455const char *Strsafe_find(const char *);
drh75897232000-05-29 14:26:00 +0000456
457/* Routines for handling symbols of the grammar */
458
icculus9e44cf12010-02-14 17:14:22 +0000459struct symbol *Symbol_new(const char *);
460int Symbolcmpp(const void *, const void *);
461void Symbol_init(void);
462int Symbol_insert(struct symbol *, const char *);
463struct symbol *Symbol_find(const char *);
464struct symbol *Symbol_Nth(int);
465int Symbol_count(void);
466struct symbol **Symbol_arrayof(void);
drh75897232000-05-29 14:26:00 +0000467
468/* Routines to manage the state table */
469
icculus9e44cf12010-02-14 17:14:22 +0000470int Configcmp(const char *, const char *);
471struct state *State_new(void);
472void State_init(void);
473int State_insert(struct state *, struct config *);
474struct state *State_find(struct config *);
drh14d88552017-04-14 19:44:15 +0000475struct state **State_arrayof(void);
drh75897232000-05-29 14:26:00 +0000476
477/* Routines used for efficiency in Configlist_add */
478
icculus9e44cf12010-02-14 17:14:22 +0000479void Configtable_init(void);
480int Configtable_insert(struct config *);
481struct config *Configtable_find(struct config *);
482void Configtable_clear(int(*)(struct config *));
483
drh75897232000-05-29 14:26:00 +0000484/****************** From the file "action.c" *******************************/
485/*
486** Routines processing parser actions in the LEMON parser generator.
487*/
488
489/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000490static struct action *Action_new(void){
mistachkind9bc6e82019-05-10 16:16:19 +0000491 static struct action *actionfreelist = 0;
icculus9e44cf12010-02-14 17:14:22 +0000492 struct action *newaction;
drh75897232000-05-29 14:26:00 +0000493
mistachkind9bc6e82019-05-10 16:16:19 +0000494 if( actionfreelist==0 ){
drh75897232000-05-29 14:26:00 +0000495 int i;
496 int amt = 100;
mistachkind9bc6e82019-05-10 16:16:19 +0000497 actionfreelist = (struct action *)calloc(amt, sizeof(struct action));
498 if( actionfreelist==0 ){
drh75897232000-05-29 14:26:00 +0000499 fprintf(stderr,"Unable to allocate memory for a new parser action.");
500 exit(1);
501 }
mistachkind9bc6e82019-05-10 16:16:19 +0000502 for(i=0; i<amt-1; i++) actionfreelist[i].next = &actionfreelist[i+1];
503 actionfreelist[amt-1].next = 0;
drh75897232000-05-29 14:26:00 +0000504 }
mistachkind9bc6e82019-05-10 16:16:19 +0000505 newaction = actionfreelist;
506 actionfreelist = actionfreelist->next;
icculus9e44cf12010-02-14 17:14:22 +0000507 return newaction;
drh75897232000-05-29 14:26:00 +0000508}
509
drhe9278182007-07-18 18:16:29 +0000510/* Compare two actions for sorting purposes. Return negative, zero, or
511** positive if the first action is less than, equal to, or greater than
512** the first
513*/
514static int actioncmp(
515 struct action *ap1,
516 struct action *ap2
517){
drh75897232000-05-29 14:26:00 +0000518 int rc;
519 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000520 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000521 rc = (int)ap1->type - (int)ap2->type;
522 }
drh3bd48ab2015-09-07 18:23:37 +0000523 if( rc==0 && (ap1->type==REDUCE || ap1->type==SHIFTREDUCE) ){
drh75897232000-05-29 14:26:00 +0000524 rc = ap1->x.rp->index - ap2->x.rp->index;
525 }
drhe594bc32009-11-03 13:02:25 +0000526 if( rc==0 ){
icculus7b429aa2010-03-03 17:09:01 +0000527 rc = (int) (ap2 - ap1);
drhe594bc32009-11-03 13:02:25 +0000528 }
drh75897232000-05-29 14:26:00 +0000529 return rc;
530}
531
532/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000533static struct action *Action_sort(
534 struct action *ap
535){
536 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
537 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000538 return ap;
539}
540
icculus9e44cf12010-02-14 17:14:22 +0000541void Action_add(
542 struct action **app,
543 enum e_action type,
544 struct symbol *sp,
545 char *arg
546){
547 struct action *newaction;
548 newaction = Action_new();
549 newaction->next = *app;
550 *app = newaction;
551 newaction->type = type;
552 newaction->sp = sp;
drhc173ad82016-05-23 16:15:02 +0000553 newaction->spOpt = 0;
drh75897232000-05-29 14:26:00 +0000554 if( type==SHIFT ){
icculus9e44cf12010-02-14 17:14:22 +0000555 newaction->x.stp = (struct state *)arg;
drh75897232000-05-29 14:26:00 +0000556 }else{
icculus9e44cf12010-02-14 17:14:22 +0000557 newaction->x.rp = (struct rule *)arg;
drh75897232000-05-29 14:26:00 +0000558 }
559}
drh8b582012003-10-21 13:16:03 +0000560/********************** New code to implement the "acttab" module ***********/
561/*
562** This module implements routines use to construct the yy_action[] table.
563*/
564
565/*
566** The state of the yy_action table under construction is an instance of
drh8dc3e8f2010-01-07 03:53:03 +0000567** the following structure.
568**
569** The yy_action table maps the pair (state_number, lookahead) into an
570** action_number. The table is an array of integers pairs. The state_number
571** determines an initial offset into the yy_action array. The lookahead
572** value is then added to this initial offset to get an index X into the
573** yy_action array. If the aAction[X].lookahead equals the value of the
574** of the lookahead input, then the value of the action_number output is
575** aAction[X].action. If the lookaheads do not match then the
576** default action for the state_number is returned.
577**
578** All actions associated with a single state_number are first entered
drh06f60d82017-04-14 19:46:12 +0000579** into aLookahead[] using multiple calls to acttab_action(). Then the
580** actions for that single state_number are placed into the aAction[]
drh8dc3e8f2010-01-07 03:53:03 +0000581** array with a single call to acttab_insert(). The acttab_insert() call
582** also resets the aLookahead[] array in preparation for the next
583** state number.
drh8b582012003-10-21 13:16:03 +0000584*/
icculus9e44cf12010-02-14 17:14:22 +0000585struct lookahead_action {
586 int lookahead; /* Value of the lookahead token */
587 int action; /* Action to take on the given lookahead */
588};
drh8b582012003-10-21 13:16:03 +0000589typedef struct acttab acttab;
590struct acttab {
591 int nAction; /* Number of used slots in aAction[] */
592 int nActionAlloc; /* Slots allocated for aAction[] */
icculus9e44cf12010-02-14 17:14:22 +0000593 struct lookahead_action
594 *aAction, /* The yy_action[] table under construction */
drh8b582012003-10-21 13:16:03 +0000595 *aLookahead; /* A single new transaction set */
596 int mnLookahead; /* Minimum aLookahead[].lookahead */
597 int mnAction; /* Action associated with mnLookahead */
598 int mxLookahead; /* Maximum aLookahead[].lookahead */
599 int nLookahead; /* Used slots in aLookahead[] */
600 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
drh3a9d6c72017-12-25 04:15:38 +0000601 int nterminal; /* Number of terminal symbols */
602 int nsymbol; /* total number of symbols */
drh8b582012003-10-21 13:16:03 +0000603};
604
605/* Return the number of entries in the yy_action table */
drh3a9d6c72017-12-25 04:15:38 +0000606#define acttab_lookahead_size(X) ((X)->nAction)
drh8b582012003-10-21 13:16:03 +0000607
608/* The value for the N-th entry in yy_action */
609#define acttab_yyaction(X,N) ((X)->aAction[N].action)
610
611/* The value for the N-th entry in yy_lookahead */
612#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
613
614/* Free all memory associated with the given acttab */
615void acttab_free(acttab *p){
616 free( p->aAction );
617 free( p->aLookahead );
618 free( p );
619}
620
621/* Allocate a new acttab structure */
drh3a9d6c72017-12-25 04:15:38 +0000622acttab *acttab_alloc(int nsymbol, int nterminal){
icculus9e44cf12010-02-14 17:14:22 +0000623 acttab *p = (acttab *) calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000624 if( p==0 ){
625 fprintf(stderr,"Unable to allocate memory for a new acttab.");
626 exit(1);
627 }
628 memset(p, 0, sizeof(*p));
drh3a9d6c72017-12-25 04:15:38 +0000629 p->nsymbol = nsymbol;
630 p->nterminal = nterminal;
drh8b582012003-10-21 13:16:03 +0000631 return p;
632}
633
drh06f60d82017-04-14 19:46:12 +0000634/* Add a new action to the current transaction set.
drh8dc3e8f2010-01-07 03:53:03 +0000635**
636** This routine is called once for each lookahead for a particular
637** state.
drh8b582012003-10-21 13:16:03 +0000638*/
639void acttab_action(acttab *p, int lookahead, int action){
640 if( p->nLookahead>=p->nLookaheadAlloc ){
641 p->nLookaheadAlloc += 25;
icculus9e44cf12010-02-14 17:14:22 +0000642 p->aLookahead = (struct lookahead_action *) realloc( p->aLookahead,
drh8b582012003-10-21 13:16:03 +0000643 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
644 if( p->aLookahead==0 ){
645 fprintf(stderr,"malloc failed\n");
646 exit(1);
647 }
648 }
649 if( p->nLookahead==0 ){
650 p->mxLookahead = lookahead;
651 p->mnLookahead = lookahead;
652 p->mnAction = action;
653 }else{
654 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
655 if( p->mnLookahead>lookahead ){
656 p->mnLookahead = lookahead;
657 p->mnAction = action;
658 }
659 }
660 p->aLookahead[p->nLookahead].lookahead = lookahead;
661 p->aLookahead[p->nLookahead].action = action;
662 p->nLookahead++;
663}
664
665/*
666** Add the transaction set built up with prior calls to acttab_action()
667** into the current action table. Then reset the transaction set back
668** to an empty set in preparation for a new round of acttab_action() calls.
669**
670** Return the offset into the action table of the new transaction.
drh3a9d6c72017-12-25 04:15:38 +0000671**
672** If the makeItSafe parameter is true, then the offset is chosen so that
673** it is impossible to overread the yy_lookaside[] table regardless of
674** the lookaside token. This is done for the terminal symbols, as they
675** come from external inputs and can contain syntax errors. When makeItSafe
676** is false, there is more flexibility in selecting offsets, resulting in
677** a smaller table. For non-terminal symbols, which are never syntax errors,
678** makeItSafe can be false.
drh8b582012003-10-21 13:16:03 +0000679*/
drh3a9d6c72017-12-25 04:15:38 +0000680int acttab_insert(acttab *p, int makeItSafe){
681 int i, j, k, n, end;
drh8b582012003-10-21 13:16:03 +0000682 assert( p->nLookahead>0 );
683
684 /* Make sure we have enough space to hold the expanded action table
685 ** in the worst case. The worst case occurs if the transaction set
686 ** must be appended to the current action table
687 */
drh3a9d6c72017-12-25 04:15:38 +0000688 n = p->nsymbol + 1;
drh8dc3e8f2010-01-07 03:53:03 +0000689 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000690 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000691 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
icculus9e44cf12010-02-14 17:14:22 +0000692 p->aAction = (struct lookahead_action *) realloc( p->aAction,
drh8b582012003-10-21 13:16:03 +0000693 sizeof(p->aAction[0])*p->nActionAlloc);
694 if( p->aAction==0 ){
695 fprintf(stderr,"malloc failed\n");
696 exit(1);
697 }
drhfdbf9282003-10-21 16:34:41 +0000698 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000699 p->aAction[i].lookahead = -1;
700 p->aAction[i].action = -1;
701 }
702 }
703
drh06f60d82017-04-14 19:46:12 +0000704 /* Scan the existing action table looking for an offset that is a
drh8dc3e8f2010-01-07 03:53:03 +0000705 ** duplicate of the current transaction set. Fall out of the loop
706 ** if and when the duplicate is found.
drh8b582012003-10-21 13:16:03 +0000707 **
708 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
709 */
drh3a9d6c72017-12-25 04:15:38 +0000710 end = makeItSafe ? p->mnLookahead : 0;
711 for(i=p->nAction-1; i>=end; i--){
drhf16371d2009-11-03 19:18:31 +0000712 if( p->aAction[i].lookahead==p->mnLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000713 /* All lookaheads and actions in the aLookahead[] transaction
714 ** must match against the candidate aAction[i] entry. */
drh8b582012003-10-21 13:16:03 +0000715 if( p->aAction[i].action!=p->mnAction ) continue;
716 for(j=0; j<p->nLookahead; j++){
717 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
718 if( k<0 || k>=p->nAction ) break;
719 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
720 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
721 }
722 if( j<p->nLookahead ) continue;
drh8dc3e8f2010-01-07 03:53:03 +0000723
724 /* No possible lookahead value that is not in the aLookahead[]
725 ** transaction is allowed to match aAction[i] */
drh8b582012003-10-21 13:16:03 +0000726 n = 0;
727 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000728 if( p->aAction[j].lookahead<0 ) continue;
729 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000730 }
drhfdbf9282003-10-21 16:34:41 +0000731 if( n==p->nLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000732 break; /* An exact match is found at offset i */
drhfdbf9282003-10-21 16:34:41 +0000733 }
drh8b582012003-10-21 13:16:03 +0000734 }
735 }
drh8dc3e8f2010-01-07 03:53:03 +0000736
737 /* If no existing offsets exactly match the current transaction, find an
738 ** an empty offset in the aAction[] table in which we can add the
739 ** aLookahead[] transaction.
740 */
drh3a9d6c72017-12-25 04:15:38 +0000741 if( i<end ){
drh8dc3e8f2010-01-07 03:53:03 +0000742 /* Look for holes in the aAction[] table that fit the current
743 ** aLookahead[] transaction. Leave i set to the offset of the hole.
744 ** If no holes are found, i is left at p->nAction, which means the
745 ** transaction will be appended. */
drh3a9d6c72017-12-25 04:15:38 +0000746 i = makeItSafe ? p->mnLookahead : 0;
747 for(; i<p->nActionAlloc - p->mxLookahead; i++){
drhf16371d2009-11-03 19:18:31 +0000748 if( p->aAction[i].lookahead<0 ){
749 for(j=0; j<p->nLookahead; j++){
750 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
751 if( k<0 ) break;
752 if( p->aAction[k].lookahead>=0 ) break;
753 }
754 if( j<p->nLookahead ) continue;
755 for(j=0; j<p->nAction; j++){
756 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
757 }
758 if( j==p->nAction ){
759 break; /* Fits in empty slots */
760 }
761 }
762 }
763 }
drh8b582012003-10-21 13:16:03 +0000764 /* Insert transaction set at index i. */
drh3a9d6c72017-12-25 04:15:38 +0000765#if 0
766 printf("Acttab:");
767 for(j=0; j<p->nLookahead; j++){
768 printf(" %d", p->aLookahead[j].lookahead);
769 }
770 printf(" inserted at %d\n", i);
771#endif
drh8b582012003-10-21 13:16:03 +0000772 for(j=0; j<p->nLookahead; j++){
773 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
774 p->aAction[k] = p->aLookahead[j];
775 if( k>=p->nAction ) p->nAction = k+1;
776 }
drh4396c612017-12-27 15:21:16 +0000777 if( makeItSafe && i+p->nterminal>=p->nAction ) p->nAction = i+p->nterminal+1;
drh8b582012003-10-21 13:16:03 +0000778 p->nLookahead = 0;
779
780 /* Return the offset that is added to the lookahead in order to get the
781 ** index into yy_action of the action */
782 return i - p->mnLookahead;
783}
784
drh3a9d6c72017-12-25 04:15:38 +0000785/*
786** Return the size of the action table without the trailing syntax error
787** entries.
788*/
789int acttab_action_size(acttab *p){
790 int n = p->nAction;
791 while( n>0 && p->aAction[n-1].lookahead<0 ){ n--; }
792 return n;
793}
794
drh75897232000-05-29 14:26:00 +0000795/********************** From the file "build.c" *****************************/
796/*
797** Routines to construction the finite state machine for the LEMON
798** parser generator.
799*/
800
801/* Find a precedence symbol of every rule in the grammar.
drh06f60d82017-04-14 19:46:12 +0000802**
drh75897232000-05-29 14:26:00 +0000803** Those rules which have a precedence symbol coded in the input
804** grammar using the "[symbol]" construct will already have the
805** rp->precsym field filled. Other rules take as their precedence
806** symbol the first RHS symbol with a defined precedence. If there
807** are not RHS symbols with a defined precedence, the precedence
808** symbol field is left blank.
809*/
icculus9e44cf12010-02-14 17:14:22 +0000810void FindRulePrecedences(struct lemon *xp)
drh75897232000-05-29 14:26:00 +0000811{
812 struct rule *rp;
813 for(rp=xp->rule; rp; rp=rp->next){
814 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000815 int i, j;
816 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
817 struct symbol *sp = rp->rhs[i];
818 if( sp->type==MULTITERMINAL ){
819 for(j=0; j<sp->nsubsym; j++){
820 if( sp->subsym[j]->prec>=0 ){
821 rp->precsym = sp->subsym[j];
822 break;
823 }
824 }
825 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000826 rp->precsym = rp->rhs[i];
drhf2f105d2012-08-20 15:53:54 +0000827 }
drh75897232000-05-29 14:26:00 +0000828 }
829 }
830 }
831 return;
832}
833
834/* Find all nonterminals which will generate the empty string.
835** Then go back and compute the first sets of every nonterminal.
836** The first set is the set of all terminal symbols which can begin
837** a string generated by that nonterminal.
838*/
icculus9e44cf12010-02-14 17:14:22 +0000839void FindFirstSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000840{
drhfd405312005-11-06 04:06:59 +0000841 int i, j;
drh75897232000-05-29 14:26:00 +0000842 struct rule *rp;
843 int progress;
844
845 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000846 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000847 }
848 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
849 lemp->symbols[i]->firstset = SetNew();
850 }
851
852 /* First compute all lambdas */
853 do{
854 progress = 0;
855 for(rp=lemp->rule; rp; rp=rp->next){
856 if( rp->lhs->lambda ) continue;
857 for(i=0; i<rp->nrhs; i++){
drh7dd1ac62012-01-07 15:17:18 +0000858 struct symbol *sp = rp->rhs[i];
859 assert( sp->type==NONTERMINAL || sp->lambda==LEMON_FALSE );
860 if( sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000861 }
862 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000863 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000864 progress = 1;
865 }
866 }
867 }while( progress );
868
869 /* Now compute all first sets */
870 do{
871 struct symbol *s1, *s2;
872 progress = 0;
873 for(rp=lemp->rule; rp; rp=rp->next){
874 s1 = rp->lhs;
875 for(i=0; i<rp->nrhs; i++){
876 s2 = rp->rhs[i];
877 if( s2->type==TERMINAL ){
878 progress += SetAdd(s1->firstset,s2->index);
879 break;
drhfd405312005-11-06 04:06:59 +0000880 }else if( s2->type==MULTITERMINAL ){
881 for(j=0; j<s2->nsubsym; j++){
882 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
883 }
884 break;
drhf2f105d2012-08-20 15:53:54 +0000885 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000886 if( s1->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000887 }else{
drh75897232000-05-29 14:26:00 +0000888 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000889 if( s2->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000890 }
drh75897232000-05-29 14:26:00 +0000891 }
892 }
893 }while( progress );
894 return;
895}
896
897/* Compute all LR(0) states for the grammar. Links
898** are added to between some states so that the LR(1) follow sets
899** can be computed later.
900*/
icculus9e44cf12010-02-14 17:14:22 +0000901PRIVATE struct state *getstate(struct lemon *); /* forward reference */
902void FindStates(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000903{
904 struct symbol *sp;
905 struct rule *rp;
906
907 Configlist_init();
908
909 /* Find the start symbol */
910 if( lemp->start ){
911 sp = Symbol_find(lemp->start);
912 if( sp==0 ){
913 ErrorMsg(lemp->filename,0,
drh3ecc05b2019-12-12 00:20:40 +0000914 "The specified start symbol \"%s\" is not "
915 "in a nonterminal of the grammar. \"%s\" will be used as the start "
916 "symbol instead.",lemp->start,lemp->startRule->lhs->name);
drh75897232000-05-29 14:26:00 +0000917 lemp->errorcnt++;
drh4ef07702016-03-16 19:45:54 +0000918 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +0000919 }
drh10761302021-10-04 16:14:51 +0000920 }else if( lemp->startRule ){
drh4ef07702016-03-16 19:45:54 +0000921 sp = lemp->startRule->lhs;
drh10761302021-10-04 16:14:51 +0000922 }else{
923 ErrorMsg(lemp->filename,0,"Internal error - no start rule\n");
924 exit(1);
drh75897232000-05-29 14:26:00 +0000925 }
926
927 /* Make sure the start symbol doesn't occur on the right-hand side of
928 ** any rule. Report an error if it does. (YACC would generate a new
929 ** start symbol in this case.) */
930 for(rp=lemp->rule; rp; rp=rp->next){
931 int i;
932 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000933 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000934 ErrorMsg(lemp->filename,0,
drh3ecc05b2019-12-12 00:20:40 +0000935 "The start symbol \"%s\" occurs on the "
936 "right-hand side of a rule. This will result in a parser which "
937 "does not work properly.",sp->name);
drh75897232000-05-29 14:26:00 +0000938 lemp->errorcnt++;
939 }
940 }
941 }
942
943 /* The basis configuration set for the first state
944 ** is all rules which have the start symbol as their
945 ** left-hand side */
946 for(rp=sp->rule; rp; rp=rp->nextlhs){
947 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000948 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000949 newcfp = Configlist_addbasis(rp,0);
950 SetAdd(newcfp->fws,0);
951 }
952
953 /* Compute the first state. All other states will be
954 ** computed automatically during the computation of the first one.
955 ** The returned pointer to the first state is not used. */
956 (void)getstate(lemp);
957 return;
958}
959
960/* Return a pointer to a state which is described by the configuration
961** list which has been built from calls to Configlist_add.
962*/
icculus9e44cf12010-02-14 17:14:22 +0000963PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
964PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000965{
966 struct config *cfp, *bp;
967 struct state *stp;
968
969 /* Extract the sorted basis of the new state. The basis was constructed
970 ** by prior calls to "Configlist_addbasis()". */
971 Configlist_sortbasis();
972 bp = Configlist_basis();
973
974 /* Get a state with the same basis */
975 stp = State_find(bp);
976 if( stp ){
977 /* A state with the same basis already exists! Copy all the follow-set
978 ** propagation links from the state under construction into the
979 ** preexisting state, then return a pointer to the preexisting state */
980 struct config *x, *y;
981 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
982 Plink_copy(&y->bplp,x->bplp);
983 Plink_delete(x->fplp);
984 x->fplp = x->bplp = 0;
985 }
986 cfp = Configlist_return();
987 Configlist_eat(cfp);
988 }else{
989 /* This really is a new state. Construct all the details */
990 Configlist_closure(lemp); /* Compute the configuration closure */
991 Configlist_sort(); /* Sort the configuration closure */
992 cfp = Configlist_return(); /* Get a pointer to the config list */
993 stp = State_new(); /* A new state structure */
994 MemoryCheck(stp);
995 stp->bp = bp; /* Remember the configuration basis */
996 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000997 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000998 stp->ap = 0; /* No actions, yet. */
999 State_insert(stp,stp->bp); /* Add to the state table */
1000 buildshifts(lemp,stp); /* Recursively compute successor states */
1001 }
1002 return stp;
1003}
1004
drhfd405312005-11-06 04:06:59 +00001005/*
1006** Return true if two symbols are the same.
1007*/
icculus9e44cf12010-02-14 17:14:22 +00001008int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +00001009{
1010 int i;
1011 if( a==b ) return 1;
1012 if( a->type!=MULTITERMINAL ) return 0;
1013 if( b->type!=MULTITERMINAL ) return 0;
1014 if( a->nsubsym!=b->nsubsym ) return 0;
1015 for(i=0; i<a->nsubsym; i++){
1016 if( a->subsym[i]!=b->subsym[i] ) return 0;
1017 }
1018 return 1;
1019}
1020
drh75897232000-05-29 14:26:00 +00001021/* Construct all successor states to the given state. A "successor"
1022** state is any state which can be reached by a shift action.
1023*/
icculus9e44cf12010-02-14 17:14:22 +00001024PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +00001025{
1026 struct config *cfp; /* For looping thru the config closure of "stp" */
1027 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +00001028 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +00001029 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
1030 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
1031 struct state *newstp; /* A pointer to a successor state */
1032
drhc2dbf352021-01-07 16:10:14 +00001033 /* Each configuration becomes complete after it contributes to a successor
drh75897232000-05-29 14:26:00 +00001034 ** state. Initially, all configurations are incomplete */
1035 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
1036
1037 /* Loop through all configurations of the state "stp" */
1038 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1039 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
1040 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
1041 Configlist_reset(); /* Reset the new config set */
1042 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
1043
1044 /* For every configuration in the state "stp" which has the symbol "sp"
1045 ** following its dot, add the same configuration to the basis set under
1046 ** construction but with the dot shifted one symbol to the right. */
1047 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
1048 if( bcfp->status==COMPLETE ) continue; /* Already used */
1049 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
1050 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +00001051 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +00001052 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +00001053 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
1054 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +00001055 }
1056
1057 /* Get a pointer to the state described by the basis configuration set
1058 ** constructed in the preceding loop */
1059 newstp = getstate(lemp);
1060
1061 /* The state "newstp" is reached from the state "stp" by a shift action
1062 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +00001063 if( sp->type==MULTITERMINAL ){
1064 int i;
1065 for(i=0; i<sp->nsubsym; i++){
1066 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
1067 }
1068 }else{
1069 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
1070 }
drh75897232000-05-29 14:26:00 +00001071 }
1072}
1073
1074/*
1075** Construct the propagation links
1076*/
icculus9e44cf12010-02-14 17:14:22 +00001077void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001078{
1079 int i;
1080 struct config *cfp, *other;
1081 struct state *stp;
1082 struct plink *plp;
1083
1084 /* Housekeeping detail:
1085 ** Add to every propagate link a pointer back to the state to
1086 ** which the link is attached. */
1087 for(i=0; i<lemp->nstate; i++){
1088 stp = lemp->sorted[i];
drh10761302021-10-04 16:14:51 +00001089 for(cfp=stp?stp->cfp:0; cfp; cfp=cfp->next){
drh75897232000-05-29 14:26:00 +00001090 cfp->stp = stp;
1091 }
1092 }
1093
1094 /* Convert all backlinks into forward links. Only the forward
1095 ** links are used in the follow-set computation. */
1096 for(i=0; i<lemp->nstate; i++){
1097 stp = lemp->sorted[i];
drh10761302021-10-04 16:14:51 +00001098 for(cfp=stp?stp->cfp:0; cfp; cfp=cfp->next){
drh75897232000-05-29 14:26:00 +00001099 for(plp=cfp->bplp; plp; plp=plp->next){
1100 other = plp->cfp;
1101 Plink_add(&other->fplp,cfp);
1102 }
1103 }
1104 }
1105}
1106
1107/* Compute all followsets.
1108**
1109** A followset is the set of all symbols which can come immediately
1110** after a configuration.
1111*/
icculus9e44cf12010-02-14 17:14:22 +00001112void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001113{
1114 int i;
1115 struct config *cfp;
1116 struct plink *plp;
1117 int progress;
1118 int change;
1119
1120 for(i=0; i<lemp->nstate; i++){
drh10761302021-10-04 16:14:51 +00001121 assert( lemp->sorted[i]!=0 );
drh75897232000-05-29 14:26:00 +00001122 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1123 cfp->status = INCOMPLETE;
1124 }
1125 }
drh06f60d82017-04-14 19:46:12 +00001126
drh75897232000-05-29 14:26:00 +00001127 do{
1128 progress = 0;
1129 for(i=0; i<lemp->nstate; i++){
drh10761302021-10-04 16:14:51 +00001130 assert( lemp->sorted[i]!=0 );
drh75897232000-05-29 14:26:00 +00001131 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1132 if( cfp->status==COMPLETE ) continue;
1133 for(plp=cfp->fplp; plp; plp=plp->next){
1134 change = SetUnion(plp->cfp->fws,cfp->fws);
1135 if( change ){
1136 plp->cfp->status = INCOMPLETE;
1137 progress = 1;
drhf2f105d2012-08-20 15:53:54 +00001138 }
1139 }
drh75897232000-05-29 14:26:00 +00001140 cfp->status = COMPLETE;
1141 }
1142 }
1143 }while( progress );
1144}
1145
drh3cb2f6e2012-01-09 14:19:05 +00001146static int resolve_conflict(struct action *,struct action *);
drh75897232000-05-29 14:26:00 +00001147
1148/* Compute the reduce actions, and resolve conflicts.
1149*/
icculus9e44cf12010-02-14 17:14:22 +00001150void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001151{
1152 int i,j;
1153 struct config *cfp;
1154 struct state *stp;
1155 struct symbol *sp;
1156 struct rule *rp;
1157
drh06f60d82017-04-14 19:46:12 +00001158 /* Add all of the reduce actions
drh75897232000-05-29 14:26:00 +00001159 ** A reduce action is added for each element of the followset of
1160 ** a configuration which has its dot at the extreme right.
1161 */
1162 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
1163 stp = lemp->sorted[i];
1164 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
1165 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
1166 for(j=0; j<lemp->nterminal; j++){
1167 if( SetFind(cfp->fws,j) ){
1168 /* Add a reduce action to the state "stp" which will reduce by the
1169 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +00001170 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +00001171 }
drhf2f105d2012-08-20 15:53:54 +00001172 }
drh75897232000-05-29 14:26:00 +00001173 }
1174 }
1175 }
1176
1177 /* Add the accepting token */
1178 if( lemp->start ){
1179 sp = Symbol_find(lemp->start);
drh10761302021-10-04 16:14:51 +00001180 if( sp==0 ){
1181 if( lemp->startRule==0 ){
1182 fprintf(stderr, "internal error on source line %d: no start rule\n",
1183 __LINE__);
1184 exit(1);
1185 }
1186 sp = lemp->startRule->lhs;
1187 }
drh75897232000-05-29 14:26:00 +00001188 }else{
drh4ef07702016-03-16 19:45:54 +00001189 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +00001190 }
1191 /* Add to the first state (which is always the starting state of the
1192 ** finite state machine) an action to ACCEPT if the lookahead is the
1193 ** start nonterminal. */
1194 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1195
1196 /* Resolve conflicts */
1197 for(i=0; i<lemp->nstate; i++){
1198 struct action *ap, *nap;
drh75897232000-05-29 14:26:00 +00001199 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001200 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001201 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001202 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001203 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1204 /* The two actions "ap" and "nap" have the same lookahead.
1205 ** Figure out which one should be used */
drh3cb2f6e2012-01-09 14:19:05 +00001206 lemp->nconflict += resolve_conflict(ap,nap);
drh75897232000-05-29 14:26:00 +00001207 }
1208 }
1209 }
1210
1211 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001212 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001213 for(i=0; i<lemp->nstate; i++){
1214 struct action *ap;
1215 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001216 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001217 }
1218 }
1219 for(rp=lemp->rule; rp; rp=rp->next){
1220 if( rp->canReduce ) continue;
1221 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1222 lemp->errorcnt++;
1223 }
1224}
1225
1226/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001227** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001228**
1229** NO LONGER TRUE:
1230** To resolve a conflict, first look to see if either action
1231** is on an error rule. In that case, take the action which
1232** is not associated with the error rule. If neither or both
1233** actions are associated with an error rule, then try to
1234** use precedence to resolve the conflict.
1235**
1236** If either action is a SHIFT, then it must be apx. This
1237** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1238*/
icculus9e44cf12010-02-14 17:14:22 +00001239static int resolve_conflict(
1240 struct action *apx,
drh3cb2f6e2012-01-09 14:19:05 +00001241 struct action *apy
icculus9e44cf12010-02-14 17:14:22 +00001242){
drh75897232000-05-29 14:26:00 +00001243 struct symbol *spx, *spy;
1244 int errcnt = 0;
1245 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001246 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001247 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001248 errcnt++;
1249 }
drh75897232000-05-29 14:26:00 +00001250 if( apx->type==SHIFT && apy->type==REDUCE ){
1251 spx = apx->sp;
1252 spy = apy->x.rp->precsym;
1253 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1254 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001255 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001256 errcnt++;
drhdd7e9db2010-07-19 01:52:07 +00001257 }else if( spx->prec>spy->prec ){ /* higher precedence wins */
drh75897232000-05-29 14:26:00 +00001258 apy->type = RD_RESOLVED;
1259 }else if( spx->prec<spy->prec ){
1260 apx->type = SH_RESOLVED;
1261 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1262 apy->type = RD_RESOLVED; /* associativity */
1263 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1264 apx->type = SH_RESOLVED;
1265 }else{
1266 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh62a223e2014-06-09 13:11:40 +00001267 apx->type = ERROR;
drh75897232000-05-29 14:26:00 +00001268 }
1269 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1270 spx = apx->x.rp->precsym;
1271 spy = apy->x.rp->precsym;
1272 if( spx==0 || spy==0 || spx->prec<0 ||
1273 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001274 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001275 errcnt++;
1276 }else if( spx->prec>spy->prec ){
1277 apy->type = RD_RESOLVED;
1278 }else if( spx->prec<spy->prec ){
1279 apx->type = RD_RESOLVED;
1280 }
1281 }else{
drh06f60d82017-04-14 19:46:12 +00001282 assert(
drhb59499c2002-02-23 18:45:13 +00001283 apx->type==SH_RESOLVED ||
1284 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001285 apx->type==SSCONFLICT ||
1286 apx->type==SRCONFLICT ||
1287 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001288 apy->type==SH_RESOLVED ||
1289 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001290 apy->type==SSCONFLICT ||
1291 apy->type==SRCONFLICT ||
1292 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001293 );
1294 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1295 ** REDUCEs on the list. If we reach this point it must be because
1296 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001297 }
1298 return errcnt;
1299}
1300/********************* From the file "configlist.c" *************************/
1301/*
1302** Routines to processing a configuration list and building a state
1303** in the LEMON parser generator.
1304*/
1305
1306static struct config *freelist = 0; /* List of free configurations */
1307static struct config *current = 0; /* Top of list of configurations */
1308static struct config **currentend = 0; /* Last on list of configs */
1309static struct config *basis = 0; /* Top of list of basis configs */
1310static struct config **basisend = 0; /* End of list of basis configs */
1311
1312/* Return a pointer to a new configuration */
drh14d88552017-04-14 19:44:15 +00001313PRIVATE struct config *newconfig(void){
drh10761302021-10-04 16:14:51 +00001314 return (struct config*)calloc(1, sizeof(struct config));
drh75897232000-05-29 14:26:00 +00001315}
1316
1317/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001318PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001319{
1320 old->next = freelist;
1321 freelist = old;
1322}
1323
1324/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001325void Configlist_init(void){
drh75897232000-05-29 14:26:00 +00001326 current = 0;
1327 currentend = &current;
1328 basis = 0;
1329 basisend = &basis;
1330 Configtable_init();
1331 return;
1332}
1333
1334/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001335void Configlist_reset(void){
drh75897232000-05-29 14:26:00 +00001336 current = 0;
1337 currentend = &current;
1338 basis = 0;
1339 basisend = &basis;
1340 Configtable_clear(0);
1341 return;
1342}
1343
1344/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001345struct config *Configlist_add(
1346 struct rule *rp, /* The rule */
1347 int dot /* Index into the RHS of the rule where the dot goes */
1348){
drh75897232000-05-29 14:26:00 +00001349 struct config *cfp, model;
1350
1351 assert( currentend!=0 );
1352 model.rp = rp;
1353 model.dot = dot;
1354 cfp = Configtable_find(&model);
1355 if( cfp==0 ){
1356 cfp = newconfig();
1357 cfp->rp = rp;
1358 cfp->dot = dot;
1359 cfp->fws = SetNew();
1360 cfp->stp = 0;
1361 cfp->fplp = cfp->bplp = 0;
1362 cfp->next = 0;
1363 cfp->bp = 0;
1364 *currentend = cfp;
1365 currentend = &cfp->next;
1366 Configtable_insert(cfp);
1367 }
1368 return cfp;
1369}
1370
1371/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001372struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001373{
1374 struct config *cfp, model;
1375
1376 assert( basisend!=0 );
1377 assert( currentend!=0 );
1378 model.rp = rp;
1379 model.dot = dot;
1380 cfp = Configtable_find(&model);
1381 if( cfp==0 ){
1382 cfp = newconfig();
1383 cfp->rp = rp;
1384 cfp->dot = dot;
1385 cfp->fws = SetNew();
1386 cfp->stp = 0;
1387 cfp->fplp = cfp->bplp = 0;
1388 cfp->next = 0;
1389 cfp->bp = 0;
1390 *currentend = cfp;
1391 currentend = &cfp->next;
1392 *basisend = cfp;
1393 basisend = &cfp->bp;
1394 Configtable_insert(cfp);
1395 }
1396 return cfp;
1397}
1398
1399/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001400void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001401{
1402 struct config *cfp, *newcfp;
1403 struct rule *rp, *newrp;
1404 struct symbol *sp, *xsp;
1405 int i, dot;
1406
1407 assert( currentend!=0 );
1408 for(cfp=current; cfp; cfp=cfp->next){
1409 rp = cfp->rp;
1410 dot = cfp->dot;
1411 if( dot>=rp->nrhs ) continue;
1412 sp = rp->rhs[dot];
1413 if( sp->type==NONTERMINAL ){
1414 if( sp->rule==0 && sp!=lemp->errsym ){
1415 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1416 sp->name);
1417 lemp->errorcnt++;
1418 }
1419 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1420 newcfp = Configlist_add(newrp,0);
1421 for(i=dot+1; i<rp->nrhs; i++){
1422 xsp = rp->rhs[i];
1423 if( xsp->type==TERMINAL ){
1424 SetAdd(newcfp->fws,xsp->index);
1425 break;
drhfd405312005-11-06 04:06:59 +00001426 }else if( xsp->type==MULTITERMINAL ){
1427 int k;
1428 for(k=0; k<xsp->nsubsym; k++){
1429 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1430 }
1431 break;
drhf2f105d2012-08-20 15:53:54 +00001432 }else{
drh75897232000-05-29 14:26:00 +00001433 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001434 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001435 }
1436 }
drh75897232000-05-29 14:26:00 +00001437 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1438 }
1439 }
1440 }
1441 return;
1442}
1443
1444/* Sort the configuration list */
drh14d88552017-04-14 19:44:15 +00001445void Configlist_sort(void){
drh25473362015-09-04 18:03:45 +00001446 current = (struct config*)msort((char*)current,(char**)&(current->next),
1447 Configcmp);
drh75897232000-05-29 14:26:00 +00001448 currentend = 0;
1449 return;
1450}
1451
1452/* Sort the basis configuration list */
drh14d88552017-04-14 19:44:15 +00001453void Configlist_sortbasis(void){
drh25473362015-09-04 18:03:45 +00001454 basis = (struct config*)msort((char*)current,(char**)&(current->bp),
1455 Configcmp);
drh75897232000-05-29 14:26:00 +00001456 basisend = 0;
1457 return;
1458}
1459
1460/* Return a pointer to the head of the configuration list and
1461** reset the list */
drh14d88552017-04-14 19:44:15 +00001462struct config *Configlist_return(void){
drh75897232000-05-29 14:26:00 +00001463 struct config *old;
1464 old = current;
1465 current = 0;
1466 currentend = 0;
1467 return old;
1468}
1469
1470/* Return a pointer to the head of the configuration list and
1471** reset the list */
drh14d88552017-04-14 19:44:15 +00001472struct config *Configlist_basis(void){
drh75897232000-05-29 14:26:00 +00001473 struct config *old;
1474 old = basis;
1475 basis = 0;
1476 basisend = 0;
1477 return old;
1478}
1479
1480/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001481void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001482{
1483 struct config *nextcfp;
1484 for(; cfp; cfp=nextcfp){
1485 nextcfp = cfp->next;
1486 assert( cfp->fplp==0 );
1487 assert( cfp->bplp==0 );
1488 if( cfp->fws ) SetFree(cfp->fws);
1489 deleteconfig(cfp);
1490 }
1491 return;
1492}
1493/***************** From the file "error.c" *********************************/
1494/*
1495** Code for printing error message.
1496*/
1497
drhf9a2e7b2003-04-15 01:49:48 +00001498void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001499 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001500 fprintf(stderr, "%s:%d: ", filename, lineno);
1501 va_start(ap, format);
1502 vfprintf(stderr,format,ap);
1503 va_end(ap);
1504 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001505}
1506/**************** From the file "main.c" ************************************/
1507/*
1508** Main program file for the LEMON parser generator.
1509*/
1510
1511/* Report an out-of-memory condition and abort. This function
1512** is used mostly by the "MemoryCheck" macro in struct.h
1513*/
drh14d88552017-04-14 19:44:15 +00001514void memory_error(void){
drh75897232000-05-29 14:26:00 +00001515 fprintf(stderr,"Out of memory. Aborting...\n");
1516 exit(1);
1517}
1518
drh6d08b4d2004-07-20 12:45:22 +00001519static int nDefine = 0; /* Number of -D options on the command line */
1520static char **azDefine = 0; /* Name of the -D macros */
1521
1522/* This routine is called with the argument to each -D command-line option.
1523** Add the macro defined to the azDefine array.
1524*/
1525static void handle_D_option(char *z){
1526 char **paz;
1527 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001528 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001529 if( azDefine==0 ){
1530 fprintf(stderr,"out of memory\n");
1531 exit(1);
1532 }
1533 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001534 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001535 if( *paz==0 ){
1536 fprintf(stderr,"out of memory\n");
1537 exit(1);
1538 }
drh898799f2014-01-10 23:21:00 +00001539 lemon_strcpy(*paz, z);
drh6d08b4d2004-07-20 12:45:22 +00001540 for(z=*paz; *z && *z!='='; z++){}
1541 *z = 0;
1542}
1543
drh9f88e6d2018-04-20 20:47:49 +00001544/* Rember the name of the output directory
1545*/
1546static char *outputDir = NULL;
1547static void handle_d_option(char *z){
1548 outputDir = (char *) malloc( lemonStrlen(z)+1 );
1549 if( outputDir==0 ){
1550 fprintf(stderr,"out of memory\n");
1551 exit(1);
1552 }
1553 lemon_strcpy(outputDir, z);
1554}
1555
icculus3e143bd2010-02-14 00:48:49 +00001556static char *user_templatename = NULL;
1557static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001558 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001559 if( user_templatename==0 ){
1560 memory_error();
1561 }
drh898799f2014-01-10 23:21:00 +00001562 lemon_strcpy(user_templatename, z);
icculus3e143bd2010-02-14 00:48:49 +00001563}
drh75897232000-05-29 14:26:00 +00001564
drh711c9812016-05-23 14:24:31 +00001565/* Merge together to lists of rules ordered by rule.iRule */
drh4ef07702016-03-16 19:45:54 +00001566static struct rule *Rule_merge(struct rule *pA, struct rule *pB){
1567 struct rule *pFirst = 0;
1568 struct rule **ppPrev = &pFirst;
1569 while( pA && pB ){
1570 if( pA->iRule<pB->iRule ){
1571 *ppPrev = pA;
1572 ppPrev = &pA->next;
1573 pA = pA->next;
1574 }else{
1575 *ppPrev = pB;
1576 ppPrev = &pB->next;
1577 pB = pB->next;
1578 }
1579 }
1580 if( pA ){
1581 *ppPrev = pA;
1582 }else{
1583 *ppPrev = pB;
1584 }
1585 return pFirst;
1586}
1587
1588/*
1589** Sort a list of rules in order of increasing iRule value
1590*/
1591static struct rule *Rule_sort(struct rule *rp){
drh18ef40e2020-09-16 16:55:56 +00001592 unsigned int i;
drh4ef07702016-03-16 19:45:54 +00001593 struct rule *pNext;
1594 struct rule *x[32];
1595 memset(x, 0, sizeof(x));
1596 while( rp ){
1597 pNext = rp->next;
1598 rp->next = 0;
drh79d9af92020-09-20 12:10:28 +00001599 for(i=0; i<sizeof(x)/sizeof(x[0])-1 && x[i]; i++){
drh4ef07702016-03-16 19:45:54 +00001600 rp = Rule_merge(x[i], rp);
1601 x[i] = 0;
1602 }
1603 x[i] = rp;
1604 rp = pNext;
1605 }
1606 rp = 0;
1607 for(i=0; i<sizeof(x)/sizeof(x[0]); i++){
1608 rp = Rule_merge(x[i], rp);
1609 }
1610 return rp;
1611}
1612
drhc75e0162015-09-07 02:23:02 +00001613/* forward reference */
1614static const char *minimum_size_type(int lwr, int upr, int *pnByte);
1615
1616/* Print a single line of the "Parser Stats" output
1617*/
1618static void stats_line(const char *zLabel, int iValue){
1619 int nLabel = lemonStrlen(zLabel);
1620 printf(" %s%.*s %5d\n", zLabel,
1621 35-nLabel, "................................",
1622 iValue);
1623}
1624
drh75897232000-05-29 14:26:00 +00001625/* The main program. Parse the command line and do it... */
drh18ef40e2020-09-16 16:55:56 +00001626int main(int argc, char **argv){
drh75897232000-05-29 14:26:00 +00001627 static int version = 0;
1628 static int rpflag = 0;
1629 static int basisflag = 0;
1630 static int compress = 0;
1631 static int quiet = 0;
1632 static int statistics = 0;
1633 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001634 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001635 static int noResort = 0;
drhfe03dac2019-11-26 02:22:39 +00001636 static int sqlFlag = 0;
drh0a34cf52020-07-03 15:41:08 +00001637 static int printPP = 0;
drh9f88e6d2018-04-20 20:47:49 +00001638
drh75897232000-05-29 14:26:00 +00001639 static struct s_options options[] = {
1640 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1641 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh9f88e6d2018-04-20 20:47:49 +00001642 {OPT_FSTR, "d", (char*)&handle_d_option, "Output directory. Default '.'"},
drh6d08b4d2004-07-20 12:45:22 +00001643 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh0a34cf52020-07-03 15:41:08 +00001644 {OPT_FLAG, "E", (char*)&printPP, "Print input file after preprocessing."},
drh0325d392015-01-01 19:11:22 +00001645 {OPT_FSTR, "f", 0, "Ignored. (Placeholder for -f compiler options.)"},
drh75897232000-05-29 14:26:00 +00001646 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
drh0325d392015-01-01 19:11:22 +00001647 {OPT_FSTR, "I", 0, "Ignored. (Placeholder for '-I' compiler options.)"},
shane58543932008-12-10 20:10:04 +00001648 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1649 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drh0325d392015-01-01 19:11:22 +00001650 {OPT_FSTR, "O", 0, "Ignored. (Placeholder for '-O' compiler options.)"},
drhf5c4e0f2010-07-18 11:35:53 +00001651 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1652 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001653 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001654 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001655 {OPT_FLAG, "s", (char*)&statistics,
1656 "Print parser stats to standard output."},
drhfe03dac2019-11-26 02:22:39 +00001657 {OPT_FLAG, "S", (char*)&sqlFlag,
1658 "Generate the *.sql file describing the parser tables."},
drh75897232000-05-29 14:26:00 +00001659 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
drh0325d392015-01-01 19:11:22 +00001660 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
1661 {OPT_FSTR, "W", 0, "Ignored. (Placeholder for '-W' compiler options.)"},
drh75897232000-05-29 14:26:00 +00001662 {OPT_FLAG,0,0,0}
1663 };
1664 int i;
icculus42585cf2010-02-14 05:19:56 +00001665 int exitcode;
drh75897232000-05-29 14:26:00 +00001666 struct lemon lem;
drh4ef07702016-03-16 19:45:54 +00001667 struct rule *rp;
drh75897232000-05-29 14:26:00 +00001668
drh18ef40e2020-09-16 16:55:56 +00001669 (void)argc;
drhb0c86772000-06-02 23:21:26 +00001670 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001671 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001672 printf("Lemon version 1.0\n");
drh06f60d82017-04-14 19:46:12 +00001673 exit(0);
drh75897232000-05-29 14:26:00 +00001674 }
drhb0c86772000-06-02 23:21:26 +00001675 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001676 fprintf(stderr,"Exactly one filename argument is required.\n");
1677 exit(1);
1678 }
drh954f6b42006-06-13 13:27:46 +00001679 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001680 lem.errorcnt = 0;
1681
1682 /* Initialize the machine */
1683 Strsafe_init();
1684 Symbol_init();
1685 State_init();
1686 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001687 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001688 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001689 lem.nolinenosflag = nolinenosflag;
drh0a34cf52020-07-03 15:41:08 +00001690 lem.printPreprocessed = printPP;
drh75897232000-05-29 14:26:00 +00001691 Symbol_new("$");
drh75897232000-05-29 14:26:00 +00001692
1693 /* Parse the input file */
1694 Parse(&lem);
drh0a34cf52020-07-03 15:41:08 +00001695 if( lem.printPreprocessed || lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001696 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001697 fprintf(stderr,"Empty grammar.\n");
1698 exit(1);
1699 }
drhed0c15b2018-04-16 14:31:34 +00001700 lem.errsym = Symbol_find("error");
drh75897232000-05-29 14:26:00 +00001701
1702 /* Count and index the symbols of the grammar */
drh75897232000-05-29 14:26:00 +00001703 Symbol_new("{default}");
drh61f92cd2014-01-11 03:06:18 +00001704 lem.nsymbol = Symbol_count();
drh75897232000-05-29 14:26:00 +00001705 lem.symbols = Symbol_arrayof();
drh61f92cd2014-01-11 03:06:18 +00001706 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1707 qsort(lem.symbols,lem.nsymbol,sizeof(struct symbol*), Symbolcmpp);
1708 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1709 while( lem.symbols[i-1]->type==MULTITERMINAL ){ i--; }
1710 assert( strcmp(lem.symbols[i-1]->name,"{default}")==0 );
1711 lem.nsymbol = i - 1;
drhc56fac72015-10-29 13:48:15 +00001712 for(i=1; ISUPPER(lem.symbols[i]->name[0]); i++);
drh75897232000-05-29 14:26:00 +00001713 lem.nterminal = i;
1714
drh711c9812016-05-23 14:24:31 +00001715 /* Assign sequential rule numbers. Start with 0. Put rules that have no
1716 ** reduce action C-code associated with them last, so that the switch()
1717 ** statement that selects reduction actions will have a smaller jump table.
1718 */
drh4ef07702016-03-16 19:45:54 +00001719 for(i=0, rp=lem.rule; rp; rp=rp->next){
1720 rp->iRule = rp->code ? i++ : -1;
1721 }
drhce678c22019-12-11 18:53:51 +00001722 lem.nruleWithAction = i;
drh4ef07702016-03-16 19:45:54 +00001723 for(rp=lem.rule; rp; rp=rp->next){
1724 if( rp->iRule<0 ) rp->iRule = i++;
1725 }
1726 lem.startRule = lem.rule;
1727 lem.rule = Rule_sort(lem.rule);
1728
drh75897232000-05-29 14:26:00 +00001729 /* Generate a reprint of the grammar, if requested on the command line */
1730 if( rpflag ){
1731 Reprint(&lem);
1732 }else{
1733 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001734 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001735
1736 /* Find the precedence for every production rule (that has one) */
1737 FindRulePrecedences(&lem);
1738
1739 /* Compute the lambda-nonterminals and the first-sets for every
1740 ** nonterminal */
1741 FindFirstSets(&lem);
1742
1743 /* Compute all LR(0) states. Also record follow-set propagation
1744 ** links so that the follow-set can be computed later */
1745 lem.nstate = 0;
1746 FindStates(&lem);
1747 lem.sorted = State_arrayof();
1748
1749 /* Tie up loose ends on the propagation links */
1750 FindLinks(&lem);
1751
1752 /* Compute the follow set of every reducible configuration */
1753 FindFollowSets(&lem);
1754
1755 /* Compute the action tables */
1756 FindActions(&lem);
1757
1758 /* Compress the action tables */
1759 if( compress==0 ) CompressTables(&lem);
1760
drhada354d2005-11-05 15:03:59 +00001761 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001762 ** occur at the end. This is an optimization that helps make the
1763 ** generated parser tables smaller. */
1764 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001765
drh75897232000-05-29 14:26:00 +00001766 /* Generate a report of the parser generated. (the "y.output" file) */
1767 if( !quiet ) ReportOutput(&lem);
1768
1769 /* Generate the source code for the parser */
drhfe03dac2019-11-26 02:22:39 +00001770 ReportTable(&lem, mhflag, sqlFlag);
drh75897232000-05-29 14:26:00 +00001771
1772 /* Produce a header file for use by the scanner. (This step is
1773 ** omitted if the "-m" option is used because makeheaders will
1774 ** generate the file for us.) */
1775 if( !mhflag ) ReportHeader(&lem);
1776 }
1777 if( statistics ){
drhc75e0162015-09-07 02:23:02 +00001778 printf("Parser statistics:\n");
1779 stats_line("terminal symbols", lem.nterminal);
1780 stats_line("non-terminal symbols", lem.nsymbol - lem.nterminal);
1781 stats_line("total symbols", lem.nsymbol);
1782 stats_line("rules", lem.nrule);
drh3bd48ab2015-09-07 18:23:37 +00001783 stats_line("states", lem.nxstate);
drhc75e0162015-09-07 02:23:02 +00001784 stats_line("conflicts", lem.nconflict);
1785 stats_line("action table entries", lem.nactiontab);
drh3a9d6c72017-12-25 04:15:38 +00001786 stats_line("lookahead table entries", lem.nlookaheadtab);
drhc75e0162015-09-07 02:23:02 +00001787 stats_line("total table size (bytes)", lem.tablesize);
drh75897232000-05-29 14:26:00 +00001788 }
icculus8e158022010-02-16 16:09:03 +00001789 if( lem.nconflict > 0 ){
1790 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001791 }
1792
1793 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001794 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001795 exit(exitcode);
1796 return (exitcode);
drh75897232000-05-29 14:26:00 +00001797}
1798/******************** From the file "msort.c" *******************************/
1799/*
1800** A generic merge-sort program.
1801**
1802** USAGE:
1803** Let "ptr" be a pointer to some structure which is at the head of
1804** a null-terminated list. Then to sort the list call:
1805**
1806** ptr = msort(ptr,&(ptr->next),cmpfnc);
1807**
1808** In the above, "cmpfnc" is a pointer to a function which compares
1809** two instances of the structure and returns an integer, as in
1810** strcmp. The second argument is a pointer to the pointer to the
1811** second element of the linked list. This address is used to compute
1812** the offset to the "next" field within the structure. The offset to
1813** the "next" field must be constant for all structures in the list.
1814**
1815** The function returns a new pointer which is the head of the list
1816** after sorting.
1817**
1818** ALGORITHM:
1819** Merge-sort.
1820*/
1821
1822/*
1823** Return a pointer to the next structure in the linked list.
1824*/
drhd25d6922012-04-18 09:59:56 +00001825#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001826
1827/*
1828** Inputs:
1829** a: A sorted, null-terminated linked list. (May be null).
1830** b: A sorted, null-terminated linked list. (May be null).
1831** cmp: A pointer to the comparison function.
1832** offset: Offset in the structure to the "next" field.
1833**
1834** Return Value:
1835** A pointer to the head of a sorted list containing the elements
1836** of both a and b.
1837**
1838** Side effects:
1839** The "next" pointers for elements in the lists a and b are
1840** changed.
1841*/
drhe9278182007-07-18 18:16:29 +00001842static char *merge(
1843 char *a,
1844 char *b,
1845 int (*cmp)(const char*,const char*),
1846 int offset
1847){
drh75897232000-05-29 14:26:00 +00001848 char *ptr, *head;
1849
1850 if( a==0 ){
1851 head = b;
1852 }else if( b==0 ){
1853 head = a;
1854 }else{
drhe594bc32009-11-03 13:02:25 +00001855 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001856 ptr = a;
1857 a = NEXT(a);
1858 }else{
1859 ptr = b;
1860 b = NEXT(b);
1861 }
1862 head = ptr;
1863 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001864 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001865 NEXT(ptr) = a;
1866 ptr = a;
1867 a = NEXT(a);
1868 }else{
1869 NEXT(ptr) = b;
1870 ptr = b;
1871 b = NEXT(b);
1872 }
1873 }
1874 if( a ) NEXT(ptr) = a;
1875 else NEXT(ptr) = b;
1876 }
1877 return head;
1878}
1879
1880/*
1881** Inputs:
1882** list: Pointer to a singly-linked list of structures.
1883** next: Pointer to pointer to the second element of the list.
1884** cmp: A comparison function.
1885**
1886** Return Value:
1887** A pointer to the head of a sorted list containing the elements
drhc2dbf352021-01-07 16:10:14 +00001888** originally in list.
drh75897232000-05-29 14:26:00 +00001889**
1890** Side effects:
1891** The "next" pointers for elements in list are changed.
1892*/
1893#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001894static char *msort(
1895 char *list,
1896 char **next,
1897 int (*cmp)(const char*,const char*)
1898){
drhba99af52001-10-25 20:37:16 +00001899 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001900 char *ep;
1901 char *set[LISTSIZE];
1902 int i;
drh1cc0d112015-03-31 15:15:48 +00001903 offset = (unsigned long)((char*)next - (char*)list);
drh75897232000-05-29 14:26:00 +00001904 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1905 while( list ){
1906 ep = list;
1907 list = NEXT(list);
1908 NEXT(ep) = 0;
1909 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1910 ep = merge(ep,set[i],cmp,offset);
1911 set[i] = 0;
1912 }
1913 set[i] = ep;
1914 }
1915 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001916 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001917 return ep;
1918}
1919/************************ From the file "option.c" **************************/
mistachkind9bc6e82019-05-10 16:16:19 +00001920static char **g_argv;
drh75897232000-05-29 14:26:00 +00001921static struct s_options *op;
1922static FILE *errstream;
1923
1924#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1925
1926/*
1927** Print the command line with a carrot pointing to the k-th character
1928** of the n-th field.
1929*/
icculus9e44cf12010-02-14 17:14:22 +00001930static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001931{
1932 int spcnt, i;
drh10761302021-10-04 16:14:51 +00001933 if( g_argv[0] ){
1934 fprintf(err,"%s",g_argv[0]);
1935 spcnt = lemonStrlen(g_argv[0]) + 1;
1936 }else{
1937 spcnt = 0;
1938 }
mistachkind9bc6e82019-05-10 16:16:19 +00001939 for(i=1; i<n && g_argv[i]; i++){
1940 fprintf(err," %s",g_argv[i]);
1941 spcnt += lemonStrlen(g_argv[i])+1;
drh75897232000-05-29 14:26:00 +00001942 }
1943 spcnt += k;
mistachkind9bc6e82019-05-10 16:16:19 +00001944 for(; g_argv[i]; i++) fprintf(err," %s",g_argv[i]);
drh75897232000-05-29 14:26:00 +00001945 if( spcnt<20 ){
1946 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1947 }else{
1948 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1949 }
1950}
1951
1952/*
1953** Return the index of the N-th non-switch argument. Return -1
1954** if N is out of range.
1955*/
icculus9e44cf12010-02-14 17:14:22 +00001956static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001957{
1958 int i;
1959 int dashdash = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00001960 if( g_argv!=0 && *g_argv!=0 ){
1961 for(i=1; g_argv[i]; i++){
1962 if( dashdash || !ISOPT(g_argv[i]) ){
drh75897232000-05-29 14:26:00 +00001963 if( n==0 ) return i;
1964 n--;
1965 }
mistachkind9bc6e82019-05-10 16:16:19 +00001966 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00001967 }
1968 }
1969 return -1;
1970}
1971
1972static char emsg[] = "Command line syntax error: ";
1973
1974/*
1975** Process a flag command line argument.
1976*/
icculus9e44cf12010-02-14 17:14:22 +00001977static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001978{
1979 int v;
1980 int errcnt = 0;
1981 int j;
1982 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00001983 if( strncmp(&g_argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001984 }
mistachkind9bc6e82019-05-10 16:16:19 +00001985 v = g_argv[i][0]=='-' ? 1 : 0;
drh75897232000-05-29 14:26:00 +00001986 if( op[j].label==0 ){
1987 if( err ){
1988 fprintf(err,"%sundefined option.\n",emsg);
1989 errline(i,1,err);
1990 }
1991 errcnt++;
drh0325d392015-01-01 19:11:22 +00001992 }else if( op[j].arg==0 ){
1993 /* Ignore this option */
drh75897232000-05-29 14:26:00 +00001994 }else if( op[j].type==OPT_FLAG ){
1995 *((int*)op[j].arg) = v;
1996 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001997 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001998 }else if( op[j].type==OPT_FSTR ){
mistachkind9bc6e82019-05-10 16:16:19 +00001999 (*(void(*)(char *))(op[j].arg))(&g_argv[i][2]);
drh75897232000-05-29 14:26:00 +00002000 }else{
2001 if( err ){
2002 fprintf(err,"%smissing argument on switch.\n",emsg);
2003 errline(i,1,err);
2004 }
2005 errcnt++;
2006 }
2007 return errcnt;
2008}
2009
2010/*
2011** Process a command line switch which has an argument.
2012*/
icculus9e44cf12010-02-14 17:14:22 +00002013static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00002014{
2015 int lv = 0;
2016 double dv = 0.0;
2017 char *sv = 0, *end;
2018 char *cp;
2019 int j;
2020 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002021 cp = strchr(g_argv[i],'=');
drh43617e92006-03-06 20:55:46 +00002022 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00002023 *cp = 0;
2024 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00002025 if( strcmp(g_argv[i],op[j].label)==0 ) break;
drh75897232000-05-29 14:26:00 +00002026 }
2027 *cp = '=';
2028 if( op[j].label==0 ){
2029 if( err ){
2030 fprintf(err,"%sundefined option.\n",emsg);
2031 errline(i,0,err);
2032 }
2033 errcnt++;
2034 }else{
2035 cp++;
2036 switch( op[j].type ){
2037 case OPT_FLAG:
2038 case OPT_FFLAG:
2039 if( err ){
2040 fprintf(err,"%soption requires an argument.\n",emsg);
2041 errline(i,0,err);
2042 }
2043 errcnt++;
2044 break;
2045 case OPT_DBL:
2046 case OPT_FDBL:
2047 dv = strtod(cp,&end);
2048 if( *end ){
2049 if( err ){
drh25473362015-09-04 18:03:45 +00002050 fprintf(err,
2051 "%sillegal character in floating-point argument.\n",emsg);
mistachkind9bc6e82019-05-10 16:16:19 +00002052 errline(i,(int)((char*)end-(char*)g_argv[i]),err);
drh75897232000-05-29 14:26:00 +00002053 }
2054 errcnt++;
2055 }
2056 break;
2057 case OPT_INT:
2058 case OPT_FINT:
2059 lv = strtol(cp,&end,0);
2060 if( *end ){
2061 if( err ){
2062 fprintf(err,"%sillegal character in integer argument.\n",emsg);
mistachkind9bc6e82019-05-10 16:16:19 +00002063 errline(i,(int)((char*)end-(char*)g_argv[i]),err);
drh75897232000-05-29 14:26:00 +00002064 }
2065 errcnt++;
2066 }
2067 break;
2068 case OPT_STR:
2069 case OPT_FSTR:
2070 sv = cp;
2071 break;
2072 }
2073 switch( op[j].type ){
2074 case OPT_FLAG:
2075 case OPT_FFLAG:
2076 break;
2077 case OPT_DBL:
2078 *(double*)(op[j].arg) = dv;
2079 break;
2080 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00002081 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00002082 break;
2083 case OPT_INT:
2084 *(int*)(op[j].arg) = lv;
2085 break;
2086 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00002087 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00002088 break;
2089 case OPT_STR:
2090 *(char**)(op[j].arg) = sv;
2091 break;
2092 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00002093 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00002094 break;
2095 }
2096 }
2097 return errcnt;
2098}
2099
icculus9e44cf12010-02-14 17:14:22 +00002100int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00002101{
2102 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002103 g_argv = a;
drh75897232000-05-29 14:26:00 +00002104 op = o;
2105 errstream = err;
mistachkind9bc6e82019-05-10 16:16:19 +00002106 if( g_argv && *g_argv && op ){
drh75897232000-05-29 14:26:00 +00002107 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002108 for(i=1; g_argv[i]; i++){
2109 if( g_argv[i][0]=='+' || g_argv[i][0]=='-' ){
drh75897232000-05-29 14:26:00 +00002110 errcnt += handleflags(i,err);
mistachkind9bc6e82019-05-10 16:16:19 +00002111 }else if( strchr(g_argv[i],'=') ){
drh75897232000-05-29 14:26:00 +00002112 errcnt += handleswitch(i,err);
2113 }
2114 }
2115 }
2116 if( errcnt>0 ){
2117 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00002118 OptPrint();
drh75897232000-05-29 14:26:00 +00002119 exit(1);
2120 }
2121 return 0;
2122}
2123
drh14d88552017-04-14 19:44:15 +00002124int OptNArgs(void){
drh75897232000-05-29 14:26:00 +00002125 int cnt = 0;
2126 int dashdash = 0;
2127 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002128 if( g_argv!=0 && g_argv[0]!=0 ){
2129 for(i=1; g_argv[i]; i++){
2130 if( dashdash || !ISOPT(g_argv[i]) ) cnt++;
2131 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00002132 }
2133 }
2134 return cnt;
2135}
2136
icculus9e44cf12010-02-14 17:14:22 +00002137char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00002138{
2139 int i;
2140 i = argindex(n);
mistachkind9bc6e82019-05-10 16:16:19 +00002141 return i>=0 ? g_argv[i] : 0;
drh75897232000-05-29 14:26:00 +00002142}
2143
icculus9e44cf12010-02-14 17:14:22 +00002144void OptErr(int n)
drh75897232000-05-29 14:26:00 +00002145{
2146 int i;
2147 i = argindex(n);
2148 if( i>=0 ) errline(i,0,errstream);
2149}
2150
drh14d88552017-04-14 19:44:15 +00002151void OptPrint(void){
drh75897232000-05-29 14:26:00 +00002152 int i;
2153 int max, len;
2154 max = 0;
2155 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00002156 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00002157 switch( op[i].type ){
2158 case OPT_FLAG:
2159 case OPT_FFLAG:
2160 break;
2161 case OPT_INT:
2162 case OPT_FINT:
2163 len += 9; /* length of "<integer>" */
2164 break;
2165 case OPT_DBL:
2166 case OPT_FDBL:
2167 len += 6; /* length of "<real>" */
2168 break;
2169 case OPT_STR:
2170 case OPT_FSTR:
2171 len += 8; /* length of "<string>" */
2172 break;
2173 }
2174 if( len>max ) max = len;
2175 }
2176 for(i=0; op[i].label; i++){
2177 switch( op[i].type ){
2178 case OPT_FLAG:
2179 case OPT_FFLAG:
2180 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
2181 break;
2182 case OPT_INT:
2183 case OPT_FINT:
drh0325d392015-01-01 19:11:22 +00002184 fprintf(errstream," -%s<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002185 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002186 break;
2187 case OPT_DBL:
2188 case OPT_FDBL:
drh0325d392015-01-01 19:11:22 +00002189 fprintf(errstream," -%s<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002190 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002191 break;
2192 case OPT_STR:
2193 case OPT_FSTR:
drh0325d392015-01-01 19:11:22 +00002194 fprintf(errstream," -%s<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002195 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002196 break;
2197 }
2198 }
2199}
2200/*********************** From the file "parse.c" ****************************/
2201/*
2202** Input file parser for the LEMON parser generator.
2203*/
2204
2205/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00002206enum e_state {
2207 INITIALIZE,
2208 WAITING_FOR_DECL_OR_RULE,
2209 WAITING_FOR_DECL_KEYWORD,
2210 WAITING_FOR_DECL_ARG,
2211 WAITING_FOR_PRECEDENCE_SYMBOL,
2212 WAITING_FOR_ARROW,
2213 IN_RHS,
2214 LHS_ALIAS_1,
2215 LHS_ALIAS_2,
2216 LHS_ALIAS_3,
2217 RHS_ALIAS_1,
2218 RHS_ALIAS_2,
2219 PRECEDENCE_MARK_1,
2220 PRECEDENCE_MARK_2,
2221 RESYNC_AFTER_RULE_ERROR,
2222 RESYNC_AFTER_DECL_ERROR,
2223 WAITING_FOR_DESTRUCTOR_SYMBOL,
2224 WAITING_FOR_DATATYPE_SYMBOL,
2225 WAITING_FOR_FALLBACK_ID,
drh61f92cd2014-01-11 03:06:18 +00002226 WAITING_FOR_WILDCARD_ID,
2227 WAITING_FOR_CLASS_ID,
drh59c435a2017-08-02 03:21:11 +00002228 WAITING_FOR_CLASS_TOKEN,
2229 WAITING_FOR_TOKEN_NAME
icculus9e44cf12010-02-14 17:14:22 +00002230};
drh75897232000-05-29 14:26:00 +00002231struct pstate {
2232 char *filename; /* Name of the input file */
2233 int tokenlineno; /* Linenumber at which current token starts */
2234 int errorcnt; /* Number of errors so far */
2235 char *tokenstart; /* Text of current token */
2236 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00002237 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00002238 struct symbol *fallback; /* The fallback token */
drh61f92cd2014-01-11 03:06:18 +00002239 struct symbol *tkclass; /* Token class symbol */
drh75897232000-05-29 14:26:00 +00002240 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00002241 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00002242 int nrhs; /* Number of right-hand side symbols seen */
2243 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00002244 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00002245 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00002246 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00002247 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00002248 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00002249 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00002250 enum e_assoc declassoc; /* Assign this association to decl arguments */
2251 int preccounter; /* Assign this precedence to decl arguments */
2252 struct rule *firstrule; /* Pointer to first rule in the grammar */
2253 struct rule *lastrule; /* Pointer to the most recently parsed rule */
2254};
2255
2256/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00002257static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00002258{
icculus9e44cf12010-02-14 17:14:22 +00002259 const char *x;
drh75897232000-05-29 14:26:00 +00002260 x = Strsafe(psp->tokenstart); /* Save the token permanently */
2261#if 0
2262 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
2263 x,psp->state);
2264#endif
2265 switch( psp->state ){
2266 case INITIALIZE:
2267 psp->prevrule = 0;
2268 psp->preccounter = 0;
2269 psp->firstrule = psp->lastrule = 0;
2270 psp->gp->nrule = 0;
drh18ef40e2020-09-16 16:55:56 +00002271 /* fall through */
drh75897232000-05-29 14:26:00 +00002272 case WAITING_FOR_DECL_OR_RULE:
2273 if( x[0]=='%' ){
2274 psp->state = WAITING_FOR_DECL_KEYWORD;
drhc56fac72015-10-29 13:48:15 +00002275 }else if( ISLOWER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002276 psp->lhs = Symbol_new(x);
2277 psp->nrhs = 0;
2278 psp->lhsalias = 0;
2279 psp->state = WAITING_FOR_ARROW;
2280 }else if( x[0]=='{' ){
2281 if( psp->prevrule==0 ){
2282 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002283 "There is no prior rule upon which to attach the code "
2284 "fragment which begins on this line.");
drh75897232000-05-29 14:26:00 +00002285 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002286 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002287 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002288 "Code fragment beginning on this line is not the first "
2289 "to follow the previous rule.");
drh75897232000-05-29 14:26:00 +00002290 psp->errorcnt++;
drhe94006e2019-12-10 20:41:48 +00002291 }else if( strcmp(x, "{NEVER-REDUCE")==0 ){
2292 psp->prevrule->neverReduce = 1;
drh75897232000-05-29 14:26:00 +00002293 }else{
2294 psp->prevrule->line = psp->tokenlineno;
2295 psp->prevrule->code = &x[1];
drh711c9812016-05-23 14:24:31 +00002296 psp->prevrule->noCode = 0;
drhf2f105d2012-08-20 15:53:54 +00002297 }
drh75897232000-05-29 14:26:00 +00002298 }else if( x[0]=='[' ){
2299 psp->state = PRECEDENCE_MARK_1;
2300 }else{
2301 ErrorMsg(psp->filename,psp->tokenlineno,
2302 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2303 x);
2304 psp->errorcnt++;
2305 }
2306 break;
2307 case PRECEDENCE_MARK_1:
drhc56fac72015-10-29 13:48:15 +00002308 if( !ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002309 ErrorMsg(psp->filename,psp->tokenlineno,
2310 "The precedence symbol must be a terminal.");
2311 psp->errorcnt++;
2312 }else if( psp->prevrule==0 ){
2313 ErrorMsg(psp->filename,psp->tokenlineno,
2314 "There is no prior rule to assign precedence \"[%s]\".",x);
2315 psp->errorcnt++;
2316 }else if( psp->prevrule->precsym!=0 ){
2317 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002318 "Precedence mark on this line is not the first "
2319 "to follow the previous rule.");
drh75897232000-05-29 14:26:00 +00002320 psp->errorcnt++;
2321 }else{
2322 psp->prevrule->precsym = Symbol_new(x);
2323 }
2324 psp->state = PRECEDENCE_MARK_2;
2325 break;
2326 case PRECEDENCE_MARK_2:
2327 if( x[0]!=']' ){
2328 ErrorMsg(psp->filename,psp->tokenlineno,
2329 "Missing \"]\" on precedence mark.");
2330 psp->errorcnt++;
2331 }
2332 psp->state = WAITING_FOR_DECL_OR_RULE;
2333 break;
2334 case WAITING_FOR_ARROW:
2335 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2336 psp->state = IN_RHS;
2337 }else if( x[0]=='(' ){
2338 psp->state = LHS_ALIAS_1;
2339 }else{
2340 ErrorMsg(psp->filename,psp->tokenlineno,
2341 "Expected to see a \":\" following the LHS symbol \"%s\".",
2342 psp->lhs->name);
2343 psp->errorcnt++;
2344 psp->state = RESYNC_AFTER_RULE_ERROR;
2345 }
2346 break;
2347 case LHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002348 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002349 psp->lhsalias = x;
2350 psp->state = LHS_ALIAS_2;
2351 }else{
2352 ErrorMsg(psp->filename,psp->tokenlineno,
2353 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2354 x,psp->lhs->name);
2355 psp->errorcnt++;
2356 psp->state = RESYNC_AFTER_RULE_ERROR;
2357 }
2358 break;
2359 case LHS_ALIAS_2:
2360 if( x[0]==')' ){
2361 psp->state = LHS_ALIAS_3;
2362 }else{
2363 ErrorMsg(psp->filename,psp->tokenlineno,
2364 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2365 psp->errorcnt++;
2366 psp->state = RESYNC_AFTER_RULE_ERROR;
2367 }
2368 break;
2369 case LHS_ALIAS_3:
2370 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2371 psp->state = IN_RHS;
2372 }else{
2373 ErrorMsg(psp->filename,psp->tokenlineno,
2374 "Missing \"->\" following: \"%s(%s)\".",
2375 psp->lhs->name,psp->lhsalias);
2376 psp->errorcnt++;
2377 psp->state = RESYNC_AFTER_RULE_ERROR;
2378 }
2379 break;
2380 case IN_RHS:
2381 if( x[0]=='.' ){
2382 struct rule *rp;
drh06f60d82017-04-14 19:46:12 +00002383 rp = (struct rule *)calloc( sizeof(struct rule) +
drh9892c5d2007-12-21 00:02:11 +00002384 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002385 if( rp==0 ){
2386 ErrorMsg(psp->filename,psp->tokenlineno,
2387 "Can't allocate enough memory for this rule.");
2388 psp->errorcnt++;
2389 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002390 }else{
drh75897232000-05-29 14:26:00 +00002391 int i;
2392 rp->ruleline = psp->tokenlineno;
2393 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002394 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002395 for(i=0; i<psp->nrhs; i++){
2396 rp->rhs[i] = psp->rhs[i];
2397 rp->rhsalias[i] = psp->alias[i];
drh539e7412018-04-21 20:24:19 +00002398 if( rp->rhsalias[i]!=0 ){ rp->rhs[i]->bContent = 1; }
drhf2f105d2012-08-20 15:53:54 +00002399 }
drh75897232000-05-29 14:26:00 +00002400 rp->lhs = psp->lhs;
2401 rp->lhsalias = psp->lhsalias;
2402 rp->nrhs = psp->nrhs;
2403 rp->code = 0;
drh711c9812016-05-23 14:24:31 +00002404 rp->noCode = 1;
drh75897232000-05-29 14:26:00 +00002405 rp->precsym = 0;
2406 rp->index = psp->gp->nrule++;
2407 rp->nextlhs = rp->lhs->rule;
2408 rp->lhs->rule = rp;
2409 rp->next = 0;
2410 if( psp->firstrule==0 ){
2411 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002412 }else{
drh75897232000-05-29 14:26:00 +00002413 psp->lastrule->next = rp;
2414 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002415 }
drh75897232000-05-29 14:26:00 +00002416 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002417 }
drh75897232000-05-29 14:26:00 +00002418 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002419 }else if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002420 if( psp->nrhs>=MAXRHS ){
2421 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002422 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002423 x);
2424 psp->errorcnt++;
2425 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002426 }else{
drh75897232000-05-29 14:26:00 +00002427 psp->rhs[psp->nrhs] = Symbol_new(x);
2428 psp->alias[psp->nrhs] = 0;
2429 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002430 }
drh2b75d3b2020-09-05 06:21:54 +00002431 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 && ISUPPER(x[1]) ){
drhfd405312005-11-06 04:06:59 +00002432 struct symbol *msp = psp->rhs[psp->nrhs-1];
2433 if( msp->type!=MULTITERMINAL ){
2434 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002435 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002436 memset(msp, 0, sizeof(*msp));
2437 msp->type = MULTITERMINAL;
2438 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002439 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002440 msp->subsym[0] = origsp;
2441 msp->name = origsp->name;
2442 psp->rhs[psp->nrhs-1] = msp;
2443 }
2444 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002445 msp->subsym = (struct symbol **) realloc(msp->subsym,
2446 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002447 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
drhc56fac72015-10-29 13:48:15 +00002448 if( ISLOWER(x[1]) || ISLOWER(msp->subsym[0]->name[0]) ){
drhfd405312005-11-06 04:06:59 +00002449 ErrorMsg(psp->filename,psp->tokenlineno,
2450 "Cannot form a compound containing a non-terminal");
2451 psp->errorcnt++;
2452 }
drh75897232000-05-29 14:26:00 +00002453 }else if( x[0]=='(' && psp->nrhs>0 ){
2454 psp->state = RHS_ALIAS_1;
2455 }else{
2456 ErrorMsg(psp->filename,psp->tokenlineno,
2457 "Illegal character on RHS of rule: \"%s\".",x);
2458 psp->errorcnt++;
2459 psp->state = RESYNC_AFTER_RULE_ERROR;
2460 }
2461 break;
2462 case RHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002463 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002464 psp->alias[psp->nrhs-1] = x;
2465 psp->state = RHS_ALIAS_2;
2466 }else{
2467 ErrorMsg(psp->filename,psp->tokenlineno,
2468 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2469 x,psp->rhs[psp->nrhs-1]->name);
2470 psp->errorcnt++;
2471 psp->state = RESYNC_AFTER_RULE_ERROR;
2472 }
2473 break;
2474 case RHS_ALIAS_2:
2475 if( x[0]==')' ){
2476 psp->state = IN_RHS;
2477 }else{
2478 ErrorMsg(psp->filename,psp->tokenlineno,
2479 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2480 psp->errorcnt++;
2481 psp->state = RESYNC_AFTER_RULE_ERROR;
2482 }
2483 break;
2484 case WAITING_FOR_DECL_KEYWORD:
drhc56fac72015-10-29 13:48:15 +00002485 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002486 psp->declkeyword = x;
2487 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002488 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002489 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002490 psp->state = WAITING_FOR_DECL_ARG;
2491 if( strcmp(x,"name")==0 ){
2492 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002493 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002494 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002495 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002496 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002497 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002498 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002499 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002500 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002501 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002502 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002503 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002504 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002505 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002506 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002507 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002508 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002509 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002510 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002511 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002512 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002513 }else if( strcmp(x,"extra_argument")==0 ){
2514 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002515 psp->insertLineMacro = 0;
drhfb32c442018-04-21 13:51:42 +00002516 }else if( strcmp(x,"extra_context")==0 ){
2517 psp->declargslot = &(psp->gp->ctx);
2518 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002519 }else if( strcmp(x,"token_type")==0 ){
2520 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002521 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002522 }else if( strcmp(x,"default_type")==0 ){
2523 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002524 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002525 }else if( strcmp(x,"stack_size")==0 ){
2526 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002527 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002528 }else if( strcmp(x,"start_symbol")==0 ){
2529 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002530 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002531 }else if( strcmp(x,"left")==0 ){
2532 psp->preccounter++;
2533 psp->declassoc = LEFT;
2534 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2535 }else if( strcmp(x,"right")==0 ){
2536 psp->preccounter++;
2537 psp->declassoc = RIGHT;
2538 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2539 }else if( strcmp(x,"nonassoc")==0 ){
2540 psp->preccounter++;
2541 psp->declassoc = NONE;
2542 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002543 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002544 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002545 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002546 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002547 }else if( strcmp(x,"fallback")==0 ){
2548 psp->fallback = 0;
2549 psp->state = WAITING_FOR_FALLBACK_ID;
drh59c435a2017-08-02 03:21:11 +00002550 }else if( strcmp(x,"token")==0 ){
2551 psp->state = WAITING_FOR_TOKEN_NAME;
drhe09daa92006-06-10 13:29:31 +00002552 }else if( strcmp(x,"wildcard")==0 ){
2553 psp->state = WAITING_FOR_WILDCARD_ID;
drh61f92cd2014-01-11 03:06:18 +00002554 }else if( strcmp(x,"token_class")==0 ){
2555 psp->state = WAITING_FOR_CLASS_ID;
drh75897232000-05-29 14:26:00 +00002556 }else{
2557 ErrorMsg(psp->filename,psp->tokenlineno,
2558 "Unknown declaration keyword: \"%%%s\".",x);
2559 psp->errorcnt++;
2560 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002561 }
drh75897232000-05-29 14:26:00 +00002562 }else{
2563 ErrorMsg(psp->filename,psp->tokenlineno,
2564 "Illegal declaration keyword: \"%s\".",x);
2565 psp->errorcnt++;
2566 psp->state = RESYNC_AFTER_DECL_ERROR;
2567 }
2568 break;
2569 case WAITING_FOR_DESTRUCTOR_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002570 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002571 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002572 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002573 psp->errorcnt++;
2574 psp->state = RESYNC_AFTER_DECL_ERROR;
2575 }else{
icculusd286fa62010-03-03 17:06:32 +00002576 struct symbol *sp = Symbol_new(x);
2577 psp->declargslot = &sp->destructor;
2578 psp->decllinenoslot = &sp->destLineno;
2579 psp->insertLineMacro = 1;
2580 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002581 }
2582 break;
2583 case WAITING_FOR_DATATYPE_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002584 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002585 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002586 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002587 psp->errorcnt++;
2588 psp->state = RESYNC_AFTER_DECL_ERROR;
2589 }else{
icculus866bf1e2010-02-17 20:31:32 +00002590 struct symbol *sp = Symbol_find(x);
2591 if((sp) && (sp->datatype)){
2592 ErrorMsg(psp->filename,psp->tokenlineno,
2593 "Symbol %%type \"%s\" already defined", x);
2594 psp->errorcnt++;
2595 psp->state = RESYNC_AFTER_DECL_ERROR;
2596 }else{
2597 if (!sp){
2598 sp = Symbol_new(x);
2599 }
2600 psp->declargslot = &sp->datatype;
2601 psp->insertLineMacro = 0;
2602 psp->state = WAITING_FOR_DECL_ARG;
2603 }
drh75897232000-05-29 14:26:00 +00002604 }
2605 break;
2606 case WAITING_FOR_PRECEDENCE_SYMBOL:
2607 if( x[0]=='.' ){
2608 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002609 }else if( ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002610 struct symbol *sp;
2611 sp = Symbol_new(x);
2612 if( sp->prec>=0 ){
2613 ErrorMsg(psp->filename,psp->tokenlineno,
2614 "Symbol \"%s\" has already be given a precedence.",x);
2615 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002616 }else{
drh75897232000-05-29 14:26:00 +00002617 sp->prec = psp->preccounter;
2618 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002619 }
drh75897232000-05-29 14:26:00 +00002620 }else{
2621 ErrorMsg(psp->filename,psp->tokenlineno,
2622 "Can't assign a precedence to \"%s\".",x);
2623 psp->errorcnt++;
2624 }
2625 break;
2626 case WAITING_FOR_DECL_ARG:
drhc56fac72015-10-29 13:48:15 +00002627 if( x[0]=='{' || x[0]=='\"' || ISALNUM(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002628 const char *zOld, *zNew;
2629 char *zBuf, *z;
mistachkin2318d332015-01-12 18:02:52 +00002630 int nOld, n, nLine = 0, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002631 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002632 char zLine[50];
2633 zNew = x;
2634 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002635 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002636 if( *psp->declargslot ){
2637 zOld = *psp->declargslot;
2638 }else{
2639 zOld = "";
2640 }
drh87cf1372008-08-13 20:09:06 +00002641 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002642 n = nOld + nNew + 20;
drh60c71b02020-09-01 11:20:03 +00002643 addLineMacro = !psp->gp->nolinenosflag
2644 && psp->insertLineMacro
2645 && psp->tokenlineno>1
2646 && (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
drhb5bd49e2008-07-14 12:21:08 +00002647 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002648 for(z=psp->filename, nBack=0; *z; z++){
2649 if( *z=='\\' ) nBack++;
2650 }
drh898799f2014-01-10 23:21:00 +00002651 lemon_sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002652 nLine = lemonStrlen(zLine);
2653 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002654 }
icculus9e44cf12010-02-14 17:14:22 +00002655 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2656 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002657 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002658 if( nOld && zBuf[-1]!='\n' ){
2659 *(zBuf++) = '\n';
2660 }
2661 memcpy(zBuf, zLine, nLine);
2662 zBuf += nLine;
2663 *(zBuf++) = '"';
2664 for(z=psp->filename; *z; z++){
2665 if( *z=='\\' ){
2666 *(zBuf++) = '\\';
2667 }
2668 *(zBuf++) = *z;
2669 }
2670 *(zBuf++) = '"';
2671 *(zBuf++) = '\n';
2672 }
drh4dc8ef52008-07-01 17:13:57 +00002673 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2674 psp->decllinenoslot[0] = psp->tokenlineno;
2675 }
drha5808f32008-04-27 22:19:44 +00002676 memcpy(zBuf, zNew, nNew);
2677 zBuf += nNew;
2678 *zBuf = 0;
2679 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002680 }else{
2681 ErrorMsg(psp->filename,psp->tokenlineno,
2682 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2683 psp->errorcnt++;
2684 psp->state = RESYNC_AFTER_DECL_ERROR;
2685 }
2686 break;
drh0bd1f4e2002-06-06 18:54:39 +00002687 case WAITING_FOR_FALLBACK_ID:
2688 if( x[0]=='.' ){
2689 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002690 }else if( !ISUPPER(x[0]) ){
drh0bd1f4e2002-06-06 18:54:39 +00002691 ErrorMsg(psp->filename, psp->tokenlineno,
2692 "%%fallback argument \"%s\" should be a token", x);
2693 psp->errorcnt++;
2694 }else{
2695 struct symbol *sp = Symbol_new(x);
2696 if( psp->fallback==0 ){
2697 psp->fallback = sp;
2698 }else if( sp->fallback ){
2699 ErrorMsg(psp->filename, psp->tokenlineno,
2700 "More than one fallback assigned to token %s", x);
2701 psp->errorcnt++;
2702 }else{
2703 sp->fallback = psp->fallback;
2704 psp->gp->has_fallback = 1;
2705 }
2706 }
2707 break;
drh59c435a2017-08-02 03:21:11 +00002708 case WAITING_FOR_TOKEN_NAME:
2709 /* Tokens do not have to be declared before use. But they can be
2710 ** in order to control their assigned integer number. The number for
2711 ** each token is assigned when it is first seen. So by including
2712 **
drh9cffb0f2021-03-28 20:44:01 +00002713 ** %token ONE TWO THREE.
drh59c435a2017-08-02 03:21:11 +00002714 **
2715 ** early in the grammar file, that assigns small consecutive values
2716 ** to each of the tokens ONE TWO and THREE.
2717 */
2718 if( x[0]=='.' ){
2719 psp->state = WAITING_FOR_DECL_OR_RULE;
2720 }else if( !ISUPPER(x[0]) ){
2721 ErrorMsg(psp->filename, psp->tokenlineno,
2722 "%%token argument \"%s\" should be a token", x);
2723 psp->errorcnt++;
2724 }else{
2725 (void)Symbol_new(x);
2726 }
2727 break;
drhe09daa92006-06-10 13:29:31 +00002728 case WAITING_FOR_WILDCARD_ID:
2729 if( x[0]=='.' ){
2730 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002731 }else if( !ISUPPER(x[0]) ){
drhe09daa92006-06-10 13:29:31 +00002732 ErrorMsg(psp->filename, psp->tokenlineno,
2733 "%%wildcard argument \"%s\" should be a token", x);
2734 psp->errorcnt++;
2735 }else{
2736 struct symbol *sp = Symbol_new(x);
2737 if( psp->gp->wildcard==0 ){
2738 psp->gp->wildcard = sp;
2739 }else{
2740 ErrorMsg(psp->filename, psp->tokenlineno,
2741 "Extra wildcard to token: %s", x);
2742 psp->errorcnt++;
2743 }
2744 }
2745 break;
drh61f92cd2014-01-11 03:06:18 +00002746 case WAITING_FOR_CLASS_ID:
drhc56fac72015-10-29 13:48:15 +00002747 if( !ISLOWER(x[0]) ){
drh61f92cd2014-01-11 03:06:18 +00002748 ErrorMsg(psp->filename, psp->tokenlineno,
mistachkind9bc6e82019-05-10 16:16:19 +00002749 "%%token_class must be followed by an identifier: %s", x);
drh61f92cd2014-01-11 03:06:18 +00002750 psp->errorcnt++;
2751 psp->state = RESYNC_AFTER_DECL_ERROR;
2752 }else if( Symbol_find(x) ){
2753 ErrorMsg(psp->filename, psp->tokenlineno,
2754 "Symbol \"%s\" already used", x);
2755 psp->errorcnt++;
2756 psp->state = RESYNC_AFTER_DECL_ERROR;
2757 }else{
2758 psp->tkclass = Symbol_new(x);
2759 psp->tkclass->type = MULTITERMINAL;
2760 psp->state = WAITING_FOR_CLASS_TOKEN;
2761 }
2762 break;
2763 case WAITING_FOR_CLASS_TOKEN:
2764 if( x[0]=='.' ){
2765 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002766 }else if( ISUPPER(x[0]) || ((x[0]=='|' || x[0]=='/') && ISUPPER(x[1])) ){
drh61f92cd2014-01-11 03:06:18 +00002767 struct symbol *msp = psp->tkclass;
2768 msp->nsubsym++;
2769 msp->subsym = (struct symbol **) realloc(msp->subsym,
2770 sizeof(struct symbol*)*msp->nsubsym);
drhc56fac72015-10-29 13:48:15 +00002771 if( !ISUPPER(x[0]) ) x++;
drh61f92cd2014-01-11 03:06:18 +00002772 msp->subsym[msp->nsubsym-1] = Symbol_new(x);
2773 }else{
2774 ErrorMsg(psp->filename, psp->tokenlineno,
2775 "%%token_class argument \"%s\" should be a token", x);
2776 psp->errorcnt++;
2777 psp->state = RESYNC_AFTER_DECL_ERROR;
2778 }
2779 break;
drh75897232000-05-29 14:26:00 +00002780 case RESYNC_AFTER_RULE_ERROR:
2781/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2782** break; */
2783 case RESYNC_AFTER_DECL_ERROR:
2784 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2785 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2786 break;
2787 }
2788}
2789
drh0a34cf52020-07-03 15:41:08 +00002790/* The text in the input is part of the argument to an %ifdef or %ifndef.
2791** Evaluate the text as a boolean expression. Return true or false.
2792*/
2793static int eval_preprocessor_boolean(char *z, int lineno){
2794 int neg = 0;
2795 int res = 0;
2796 int okTerm = 1;
2797 int i;
2798 for(i=0; z[i]!=0; i++){
2799 if( ISSPACE(z[i]) ) continue;
2800 if( z[i]=='!' ){
2801 if( !okTerm ) goto pp_syntax_error;
2802 neg = !neg;
2803 continue;
2804 }
2805 if( z[i]=='|' && z[i+1]=='|' ){
2806 if( okTerm ) goto pp_syntax_error;
2807 if( res ) return 1;
2808 i++;
2809 okTerm = 1;
2810 continue;
2811 }
2812 if( z[i]=='&' && z[i+1]=='&' ){
2813 if( okTerm ) goto pp_syntax_error;
2814 if( !res ) return 0;
2815 i++;
2816 okTerm = 1;
2817 continue;
2818 }
2819 if( z[i]=='(' ){
2820 int k;
2821 int n = 1;
2822 if( !okTerm ) goto pp_syntax_error;
2823 for(k=i+1; z[k]; k++){
2824 if( z[k]==')' ){
2825 n--;
2826 if( n==0 ){
2827 z[k] = 0;
2828 res = eval_preprocessor_boolean(&z[i+1], -1);
2829 z[k] = ')';
2830 if( res<0 ){
2831 i = i-res;
2832 goto pp_syntax_error;
2833 }
2834 i = k;
2835 break;
2836 }
2837 }else if( z[k]=='(' ){
2838 n++;
2839 }else if( z[k]==0 ){
2840 i = k;
2841 goto pp_syntax_error;
2842 }
2843 }
2844 if( neg ){
2845 res = !res;
2846 neg = 0;
2847 }
2848 okTerm = 0;
2849 continue;
2850 }
2851 if( ISALPHA(z[i]) ){
2852 int j, k, n;
2853 if( !okTerm ) goto pp_syntax_error;
2854 for(k=i+1; ISALNUM(z[k]) || z[k]=='_'; k++){}
2855 n = k - i;
2856 res = 0;
2857 for(j=0; j<nDefine; j++){
2858 if( strncmp(azDefine[j],&z[i],n)==0 && azDefine[j][n]==0 ){
2859 res = 1;
2860 break;
2861 }
2862 }
2863 i = k-1;
2864 if( neg ){
2865 res = !res;
2866 neg = 0;
2867 }
2868 okTerm = 0;
2869 continue;
2870 }
2871 goto pp_syntax_error;
2872 }
2873 return res;
2874
2875pp_syntax_error:
2876 if( lineno>0 ){
2877 fprintf(stderr, "%%if syntax error on line %d.\n", lineno);
2878 fprintf(stderr, " %.*s <-- syntax error here\n", i+1, z);
2879 exit(1);
2880 }else{
2881 return -(i+1);
2882 }
2883}
2884
drh34ff57b2008-07-14 12:27:51 +00002885/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002886** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2887** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2888** comments them out. Text in between is also commented out as appropriate.
2889*/
danielk1977940fac92005-01-23 22:41:37 +00002890static void preprocess_input(char *z){
drh0a34cf52020-07-03 15:41:08 +00002891 int i, j, k;
drh6d08b4d2004-07-20 12:45:22 +00002892 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002893 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002894 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002895 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002896 for(i=0; z[i]; i++){
2897 if( z[i]=='\n' ) lineno++;
2898 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
drhc56fac72015-10-29 13:48:15 +00002899 if( strncmp(&z[i],"%endif",6)==0 && ISSPACE(z[i+6]) ){
drh6d08b4d2004-07-20 12:45:22 +00002900 if( exclude ){
2901 exclude--;
2902 if( exclude==0 ){
2903 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2904 }
2905 }
2906 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
drh0a34cf52020-07-03 15:41:08 +00002907 }else if( strncmp(&z[i],"%else",5)==0 && ISSPACE(z[i+5]) ){
2908 if( exclude==1){
2909 exclude = 0;
2910 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2911 }else if( exclude==0 ){
2912 exclude = 1;
2913 start = i;
2914 start_lineno = lineno;
2915 }
2916 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2917 }else if( strncmp(&z[i],"%ifdef ",7)==0
2918 || strncmp(&z[i],"%if ",4)==0
2919 || strncmp(&z[i],"%ifndef ",8)==0 ){
drh6d08b4d2004-07-20 12:45:22 +00002920 if( exclude ){
2921 exclude++;
2922 }else{
drh0a34cf52020-07-03 15:41:08 +00002923 int isNot;
2924 int iBool;
2925 for(j=i; z[j] && !ISSPACE(z[j]); j++){}
2926 iBool = j;
2927 isNot = (j==i+7);
2928 while( z[j] && z[j]!='\n' ){ j++; }
2929 k = z[j];
2930 z[j] = 0;
2931 exclude = eval_preprocessor_boolean(&z[iBool], lineno);
2932 z[j] = k;
2933 if( !isNot ) exclude = !exclude;
drh6d08b4d2004-07-20 12:45:22 +00002934 if( exclude ){
2935 start = i;
2936 start_lineno = lineno;
2937 }
2938 }
2939 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2940 }
2941 }
2942 if( exclude ){
2943 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2944 exit(1);
2945 }
2946}
2947
drh75897232000-05-29 14:26:00 +00002948/* In spite of its name, this function is really a scanner. It read
2949** in the entire input file (all at once) then tokenizes it. Each
2950** token is passed to the function "parseonetoken" which builds all
2951** the appropriate data structures in the global state vector "gp".
2952*/
icculus9e44cf12010-02-14 17:14:22 +00002953void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002954{
2955 struct pstate ps;
2956 FILE *fp;
2957 char *filebuf;
mistachkin2318d332015-01-12 18:02:52 +00002958 unsigned int filesize;
drh75897232000-05-29 14:26:00 +00002959 int lineno;
2960 int c;
2961 char *cp, *nextcp;
2962 int startline = 0;
2963
rse38514a92007-09-20 11:34:17 +00002964 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002965 ps.gp = gp;
2966 ps.filename = gp->filename;
2967 ps.errorcnt = 0;
2968 ps.state = INITIALIZE;
2969
2970 /* Begin by reading the input file */
2971 fp = fopen(ps.filename,"rb");
2972 if( fp==0 ){
2973 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2974 gp->errorcnt++;
2975 return;
2976 }
2977 fseek(fp,0,2);
2978 filesize = ftell(fp);
2979 rewind(fp);
2980 filebuf = (char *)malloc( filesize+1 );
drh03e1b1f2014-01-11 12:52:25 +00002981 if( filesize>100000000 || filebuf==0 ){
2982 ErrorMsg(ps.filename,0,"Input file too large.");
mistachkinc93c6142018-09-08 16:55:18 +00002983 free(filebuf);
drh75897232000-05-29 14:26:00 +00002984 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002985 fclose(fp);
drh75897232000-05-29 14:26:00 +00002986 return;
2987 }
2988 if( fread(filebuf,1,filesize,fp)!=filesize ){
2989 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2990 filesize);
2991 free(filebuf);
2992 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002993 fclose(fp);
drh75897232000-05-29 14:26:00 +00002994 return;
2995 }
2996 fclose(fp);
2997 filebuf[filesize] = 0;
2998
drh6d08b4d2004-07-20 12:45:22 +00002999 /* Make an initial pass through the file to handle %ifdef and %ifndef */
3000 preprocess_input(filebuf);
drh0a34cf52020-07-03 15:41:08 +00003001 if( gp->printPreprocessed ){
3002 printf("%s\n", filebuf);
3003 return;
3004 }
drh6d08b4d2004-07-20 12:45:22 +00003005
drh75897232000-05-29 14:26:00 +00003006 /* Now scan the text of the input file */
3007 lineno = 1;
3008 for(cp=filebuf; (c= *cp)!=0; ){
3009 if( c=='\n' ) lineno++; /* Keep track of the line number */
drhc56fac72015-10-29 13:48:15 +00003010 if( ISSPACE(c) ){ cp++; continue; } /* Skip all white space */
drh75897232000-05-29 14:26:00 +00003011 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
3012 cp+=2;
3013 while( (c= *cp)!=0 && c!='\n' ) cp++;
3014 continue;
3015 }
3016 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
3017 cp+=2;
3018 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
3019 if( c=='\n' ) lineno++;
3020 cp++;
3021 }
3022 if( c ) cp++;
3023 continue;
3024 }
3025 ps.tokenstart = cp; /* Mark the beginning of the token */
3026 ps.tokenlineno = lineno; /* Linenumber on which token begins */
3027 if( c=='\"' ){ /* String literals */
3028 cp++;
3029 while( (c= *cp)!=0 && c!='\"' ){
3030 if( c=='\n' ) lineno++;
3031 cp++;
3032 }
3033 if( c==0 ){
3034 ErrorMsg(ps.filename,startline,
drh3ecc05b2019-12-12 00:20:40 +00003035 "String starting on this line is not terminated before "
3036 "the end of the file.");
drh75897232000-05-29 14:26:00 +00003037 ps.errorcnt++;
3038 nextcp = cp;
3039 }else{
3040 nextcp = cp+1;
3041 }
3042 }else if( c=='{' ){ /* A block of C code */
3043 int level;
3044 cp++;
3045 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
3046 if( c=='\n' ) lineno++;
3047 else if( c=='{' ) level++;
3048 else if( c=='}' ) level--;
3049 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
3050 int prevc;
3051 cp = &cp[2];
3052 prevc = 0;
3053 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
3054 if( c=='\n' ) lineno++;
3055 prevc = c;
3056 cp++;
drhf2f105d2012-08-20 15:53:54 +00003057 }
3058 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00003059 cp = &cp[2];
3060 while( (c= *cp)!=0 && c!='\n' ) cp++;
3061 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00003062 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00003063 int startchar, prevc;
3064 startchar = c;
3065 prevc = 0;
3066 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
3067 if( c=='\n' ) lineno++;
3068 if( prevc=='\\' ) prevc = 0;
3069 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00003070 }
3071 }
drh75897232000-05-29 14:26:00 +00003072 }
3073 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00003074 ErrorMsg(ps.filename,ps.tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00003075 "C code starting on this line is not terminated before "
3076 "the end of the file.");
drh75897232000-05-29 14:26:00 +00003077 ps.errorcnt++;
3078 nextcp = cp;
3079 }else{
3080 nextcp = cp+1;
3081 }
drhc56fac72015-10-29 13:48:15 +00003082 }else if( ISALNUM(c) ){ /* Identifiers */
3083 while( (c= *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drh75897232000-05-29 14:26:00 +00003084 nextcp = cp;
3085 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
3086 cp += 3;
3087 nextcp = cp;
drhc56fac72015-10-29 13:48:15 +00003088 }else if( (c=='/' || c=='|') && ISALPHA(cp[1]) ){
drhfd405312005-11-06 04:06:59 +00003089 cp += 2;
drhc56fac72015-10-29 13:48:15 +00003090 while( (c = *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drhfd405312005-11-06 04:06:59 +00003091 nextcp = cp;
drh75897232000-05-29 14:26:00 +00003092 }else{ /* All other (one character) operators */
3093 cp++;
3094 nextcp = cp;
3095 }
3096 c = *cp;
3097 *cp = 0; /* Null terminate the token */
3098 parseonetoken(&ps); /* Parse the token */
mistachkin2318d332015-01-12 18:02:52 +00003099 *cp = (char)c; /* Restore the buffer */
drh75897232000-05-29 14:26:00 +00003100 cp = nextcp;
3101 }
3102 free(filebuf); /* Release the buffer after parsing */
3103 gp->rule = ps.firstrule;
3104 gp->errorcnt = ps.errorcnt;
3105}
3106/*************************** From the file "plink.c" *********************/
3107/*
3108** Routines processing configuration follow-set propagation links
3109** in the LEMON parser generator.
3110*/
3111static struct plink *plink_freelist = 0;
3112
3113/* Allocate a new plink */
drh14d88552017-04-14 19:44:15 +00003114struct plink *Plink_new(void){
icculus9e44cf12010-02-14 17:14:22 +00003115 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00003116
3117 if( plink_freelist==0 ){
3118 int i;
3119 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00003120 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00003121 if( plink_freelist==0 ){
3122 fprintf(stderr,
3123 "Unable to allocate memory for a new follow-set propagation link.\n");
3124 exit(1);
3125 }
3126 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
3127 plink_freelist[amt-1].next = 0;
3128 }
icculus9e44cf12010-02-14 17:14:22 +00003129 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00003130 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00003131 return newlink;
drh75897232000-05-29 14:26:00 +00003132}
3133
3134/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00003135void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00003136{
icculus9e44cf12010-02-14 17:14:22 +00003137 struct plink *newlink;
3138 newlink = Plink_new();
3139 newlink->next = *plpp;
3140 *plpp = newlink;
3141 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00003142}
3143
3144/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00003145void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00003146{
3147 struct plink *nextpl;
3148 while( from ){
3149 nextpl = from->next;
3150 from->next = *to;
3151 *to = from;
3152 from = nextpl;
3153 }
3154}
3155
3156/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00003157void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00003158{
3159 struct plink *nextpl;
3160
3161 while( plp ){
3162 nextpl = plp->next;
3163 plp->next = plink_freelist;
3164 plink_freelist = plp;
3165 plp = nextpl;
3166 }
3167}
3168/*********************** From the file "report.c" **************************/
3169/*
3170** Procedures for generating reports and tables in the LEMON parser generator.
3171*/
3172
3173/* Generate a filename with the given suffix. Space to hold the
3174** name comes from malloc() and must be freed by the calling
3175** function.
3176*/
icculus9e44cf12010-02-14 17:14:22 +00003177PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00003178{
3179 char *name;
3180 char *cp;
drh9f88e6d2018-04-20 20:47:49 +00003181 char *filename = lemp->filename;
3182 int sz;
drh75897232000-05-29 14:26:00 +00003183
drh9f88e6d2018-04-20 20:47:49 +00003184 if( outputDir ){
3185 cp = strrchr(filename, '/');
3186 if( cp ) filename = cp + 1;
3187 }
3188 sz = lemonStrlen(filename);
3189 sz += lemonStrlen(suffix);
3190 if( outputDir ) sz += lemonStrlen(outputDir) + 1;
3191 sz += 5;
3192 name = (char*)malloc( sz );
drh75897232000-05-29 14:26:00 +00003193 if( name==0 ){
3194 fprintf(stderr,"Can't allocate space for a filename.\n");
3195 exit(1);
3196 }
drh9f88e6d2018-04-20 20:47:49 +00003197 name[0] = 0;
3198 if( outputDir ){
3199 lemon_strcpy(name, outputDir);
3200 lemon_strcat(name, "/");
3201 }
3202 lemon_strcat(name,filename);
drh75897232000-05-29 14:26:00 +00003203 cp = strrchr(name,'.');
3204 if( cp ) *cp = 0;
drh898799f2014-01-10 23:21:00 +00003205 lemon_strcat(name,suffix);
drh75897232000-05-29 14:26:00 +00003206 return name;
3207}
3208
3209/* Open a file with a name based on the name of the input file,
3210** but with a different (specified) suffix, and return a pointer
3211** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00003212PRIVATE FILE *file_open(
3213 struct lemon *lemp,
3214 const char *suffix,
3215 const char *mode
3216){
drh75897232000-05-29 14:26:00 +00003217 FILE *fp;
3218
3219 if( lemp->outname ) free(lemp->outname);
3220 lemp->outname = file_makename(lemp, suffix);
3221 fp = fopen(lemp->outname,mode);
3222 if( fp==0 && *mode=='w' ){
3223 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
3224 lemp->errorcnt++;
3225 return 0;
3226 }
3227 return fp;
3228}
3229
drh5c8241b2017-12-24 23:38:10 +00003230/* Print the text of a rule
3231*/
3232void rule_print(FILE *out, struct rule *rp){
3233 int i, j;
3234 fprintf(out, "%s",rp->lhs->name);
3235 /* if( rp->lhsalias ) fprintf(out,"(%s)",rp->lhsalias); */
3236 fprintf(out," ::=");
3237 for(i=0; i<rp->nrhs; i++){
3238 struct symbol *sp = rp->rhs[i];
3239 if( sp->type==MULTITERMINAL ){
3240 fprintf(out," %s", sp->subsym[0]->name);
3241 for(j=1; j<sp->nsubsym; j++){
3242 fprintf(out,"|%s", sp->subsym[j]->name);
3243 }
3244 }else{
3245 fprintf(out," %s", sp->name);
3246 }
3247 /* if( rp->rhsalias[i] ) fprintf(out,"(%s)",rp->rhsalias[i]); */
3248 }
3249}
3250
drh06f60d82017-04-14 19:46:12 +00003251/* Duplicate the input file without comments and without actions
drh75897232000-05-29 14:26:00 +00003252** on rules */
icculus9e44cf12010-02-14 17:14:22 +00003253void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003254{
3255 struct rule *rp;
3256 struct symbol *sp;
3257 int i, j, maxlen, len, ncolumns, skip;
3258 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
3259 maxlen = 10;
3260 for(i=0; i<lemp->nsymbol; i++){
3261 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00003262 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00003263 if( len>maxlen ) maxlen = len;
3264 }
3265 ncolumns = 76/(maxlen+5);
3266 if( ncolumns<1 ) ncolumns = 1;
3267 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
3268 for(i=0; i<skip; i++){
3269 printf("//");
3270 for(j=i; j<lemp->nsymbol; j+=skip){
3271 sp = lemp->symbols[j];
3272 assert( sp->index==j );
3273 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
3274 }
3275 printf("\n");
3276 }
3277 for(rp=lemp->rule; rp; rp=rp->next){
drh5c8241b2017-12-24 23:38:10 +00003278 rule_print(stdout, rp);
drh75897232000-05-29 14:26:00 +00003279 printf(".");
3280 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00003281 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00003282 printf("\n");
3283 }
3284}
3285
drh7e698e92015-09-07 14:22:24 +00003286/* Print a single rule.
3287*/
3288void RulePrint(FILE *fp, struct rule *rp, int iCursor){
drhfd405312005-11-06 04:06:59 +00003289 struct symbol *sp;
3290 int i, j;
drh75897232000-05-29 14:26:00 +00003291 fprintf(fp,"%s ::=",rp->lhs->name);
3292 for(i=0; i<=rp->nrhs; i++){
drh7e698e92015-09-07 14:22:24 +00003293 if( i==iCursor ) fprintf(fp," *");
drh75897232000-05-29 14:26:00 +00003294 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00003295 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00003296 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00003297 fprintf(fp," %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00003298 for(j=1; j<sp->nsubsym; j++){
3299 fprintf(fp,"|%s",sp->subsym[j]->name);
3300 }
drh61f92cd2014-01-11 03:06:18 +00003301 }else{
3302 fprintf(fp," %s", sp->name);
drhfd405312005-11-06 04:06:59 +00003303 }
drh75897232000-05-29 14:26:00 +00003304 }
3305}
3306
drh7e698e92015-09-07 14:22:24 +00003307/* Print the rule for a configuration.
3308*/
3309void ConfigPrint(FILE *fp, struct config *cfp){
3310 RulePrint(fp, cfp->rp, cfp->dot);
3311}
3312
drh75897232000-05-29 14:26:00 +00003313/* #define TEST */
drhfd405312005-11-06 04:06:59 +00003314#if 0
drh75897232000-05-29 14:26:00 +00003315/* Print a set */
3316PRIVATE void SetPrint(out,set,lemp)
3317FILE *out;
3318char *set;
3319struct lemon *lemp;
3320{
3321 int i;
3322 char *spacer;
3323 spacer = "";
3324 fprintf(out,"%12s[","");
3325 for(i=0; i<lemp->nterminal; i++){
3326 if( SetFind(set,i) ){
3327 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
3328 spacer = " ";
3329 }
3330 }
3331 fprintf(out,"]\n");
3332}
3333
3334/* Print a plink chain */
3335PRIVATE void PlinkPrint(out,plp,tag)
3336FILE *out;
3337struct plink *plp;
3338char *tag;
3339{
3340 while( plp ){
drhada354d2005-11-05 15:03:59 +00003341 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00003342 ConfigPrint(out,plp->cfp);
3343 fprintf(out,"\n");
3344 plp = plp->next;
3345 }
3346}
3347#endif
3348
3349/* Print an action to the given file descriptor. Return FALSE if
3350** nothing was actually printed.
3351*/
drh7e698e92015-09-07 14:22:24 +00003352int PrintAction(
3353 struct action *ap, /* The action to print */
3354 FILE *fp, /* Print the action here */
drh3bd48ab2015-09-07 18:23:37 +00003355 int indent /* Indent by this amount */
drh7e698e92015-09-07 14:22:24 +00003356){
drh75897232000-05-29 14:26:00 +00003357 int result = 1;
3358 switch( ap->type ){
drh7e698e92015-09-07 14:22:24 +00003359 case SHIFT: {
3360 struct state *stp = ap->x.stp;
drh3bd48ab2015-09-07 18:23:37 +00003361 fprintf(fp,"%*s shift %-7d",indent,ap->sp->name,stp->statenum);
drh75897232000-05-29 14:26:00 +00003362 break;
drh7e698e92015-09-07 14:22:24 +00003363 }
3364 case REDUCE: {
3365 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003366 fprintf(fp,"%*s reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003367 RulePrint(fp, rp, -1);
3368 break;
3369 }
3370 case SHIFTREDUCE: {
3371 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003372 fprintf(fp,"%*s shift-reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003373 RulePrint(fp, rp, -1);
drh75897232000-05-29 14:26:00 +00003374 break;
drh7e698e92015-09-07 14:22:24 +00003375 }
drh75897232000-05-29 14:26:00 +00003376 case ACCEPT:
3377 fprintf(fp,"%*s accept",indent,ap->sp->name);
3378 break;
3379 case ERROR:
3380 fprintf(fp,"%*s error",indent,ap->sp->name);
3381 break;
drh9892c5d2007-12-21 00:02:11 +00003382 case SRCONFLICT:
3383 case RRCONFLICT:
drh3bd48ab2015-09-07 18:23:37 +00003384 fprintf(fp,"%*s reduce %-7d ** Parsing conflict **",
drh4ef07702016-03-16 19:45:54 +00003385 indent,ap->sp->name,ap->x.rp->iRule);
drh75897232000-05-29 14:26:00 +00003386 break;
drh9892c5d2007-12-21 00:02:11 +00003387 case SSCONFLICT:
drh06f60d82017-04-14 19:46:12 +00003388 fprintf(fp,"%*s shift %-7d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00003389 indent,ap->sp->name,ap->x.stp->statenum);
3390 break;
drh75897232000-05-29 14:26:00 +00003391 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003392 if( showPrecedenceConflict ){
drh3bd48ab2015-09-07 18:23:37 +00003393 fprintf(fp,"%*s shift %-7d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003394 indent,ap->sp->name,ap->x.stp->statenum);
3395 }else{
3396 result = 0;
3397 }
3398 break;
drh75897232000-05-29 14:26:00 +00003399 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003400 if( showPrecedenceConflict ){
drh7e698e92015-09-07 14:22:24 +00003401 fprintf(fp,"%*s reduce %-7d -- dropped by precedence",
drh4ef07702016-03-16 19:45:54 +00003402 indent,ap->sp->name,ap->x.rp->iRule);
drhf5c4e0f2010-07-18 11:35:53 +00003403 }else{
3404 result = 0;
3405 }
3406 break;
drh75897232000-05-29 14:26:00 +00003407 case NOT_USED:
3408 result = 0;
3409 break;
3410 }
drhc173ad82016-05-23 16:15:02 +00003411 if( result && ap->spOpt ){
3412 fprintf(fp," /* because %s==%s */", ap->sp->name, ap->spOpt->name);
3413 }
drh75897232000-05-29 14:26:00 +00003414 return result;
3415}
3416
drh3bd48ab2015-09-07 18:23:37 +00003417/* Generate the "*.out" log file */
icculus9e44cf12010-02-14 17:14:22 +00003418void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003419{
drh539e7412018-04-21 20:24:19 +00003420 int i, n;
drh75897232000-05-29 14:26:00 +00003421 struct state *stp;
3422 struct config *cfp;
3423 struct action *ap;
drh12f68392018-04-06 19:12:55 +00003424 struct rule *rp;
drh75897232000-05-29 14:26:00 +00003425 FILE *fp;
3426
drh2aa6ca42004-09-10 00:14:04 +00003427 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00003428 if( fp==0 ) return;
drh3bd48ab2015-09-07 18:23:37 +00003429 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00003430 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00003431 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00003432 if( lemp->basisflag ) cfp=stp->bp;
3433 else cfp=stp->cfp;
3434 while( cfp ){
3435 char buf[20];
3436 if( cfp->dot==cfp->rp->nrhs ){
drh4ef07702016-03-16 19:45:54 +00003437 lemon_sprintf(buf,"(%d)",cfp->rp->iRule);
drh75897232000-05-29 14:26:00 +00003438 fprintf(fp," %5s ",buf);
3439 }else{
3440 fprintf(fp," ");
3441 }
3442 ConfigPrint(fp,cfp);
3443 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00003444#if 0
drh75897232000-05-29 14:26:00 +00003445 SetPrint(fp,cfp->fws,lemp);
3446 PlinkPrint(fp,cfp->fplp,"To ");
3447 PlinkPrint(fp,cfp->bplp,"From");
3448#endif
3449 if( lemp->basisflag ) cfp=cfp->bp;
3450 else cfp=cfp->next;
3451 }
3452 fprintf(fp,"\n");
3453 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00003454 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
drh75897232000-05-29 14:26:00 +00003455 }
3456 fprintf(fp,"\n");
3457 }
drhe9278182007-07-18 18:16:29 +00003458 fprintf(fp, "----------------------------------------------------\n");
3459 fprintf(fp, "Symbols:\n");
drh539e7412018-04-21 20:24:19 +00003460 fprintf(fp, "The first-set of non-terminals is shown after the name.\n\n");
drhe9278182007-07-18 18:16:29 +00003461 for(i=0; i<lemp->nsymbol; i++){
3462 int j;
3463 struct symbol *sp;
3464
3465 sp = lemp->symbols[i];
3466 fprintf(fp, " %3d: %s", i, sp->name);
3467 if( sp->type==NONTERMINAL ){
3468 fprintf(fp, ":");
3469 if( sp->lambda ){
3470 fprintf(fp, " <lambda>");
3471 }
3472 for(j=0; j<lemp->nterminal; j++){
3473 if( sp->firstset && SetFind(sp->firstset, j) ){
3474 fprintf(fp, " %s", lemp->symbols[j]->name);
3475 }
3476 }
3477 }
drh12f68392018-04-06 19:12:55 +00003478 if( sp->prec>=0 ) fprintf(fp," (precedence=%d)", sp->prec);
drhe9278182007-07-18 18:16:29 +00003479 fprintf(fp, "\n");
3480 }
drh12f68392018-04-06 19:12:55 +00003481 fprintf(fp, "----------------------------------------------------\n");
drh539e7412018-04-21 20:24:19 +00003482 fprintf(fp, "Syntax-only Symbols:\n");
3483 fprintf(fp, "The following symbols never carry semantic content.\n\n");
3484 for(i=n=0; i<lemp->nsymbol; i++){
3485 int w;
3486 struct symbol *sp = lemp->symbols[i];
3487 if( sp->bContent ) continue;
3488 w = (int)strlen(sp->name);
3489 if( n>0 && n+w>75 ){
3490 fprintf(fp,"\n");
3491 n = 0;
3492 }
3493 if( n>0 ){
3494 fprintf(fp, " ");
3495 n++;
3496 }
3497 fprintf(fp, "%s", sp->name);
3498 n += w;
3499 }
3500 if( n>0 ) fprintf(fp, "\n");
3501 fprintf(fp, "----------------------------------------------------\n");
drh12f68392018-04-06 19:12:55 +00003502 fprintf(fp, "Rules:\n");
3503 for(rp=lemp->rule; rp; rp=rp->next){
3504 fprintf(fp, "%4d: ", rp->iRule);
3505 rule_print(fp, rp);
3506 fprintf(fp,".");
3507 if( rp->precsym ){
3508 fprintf(fp," [%s precedence=%d]",
3509 rp->precsym->name, rp->precsym->prec);
3510 }
3511 fprintf(fp,"\n");
3512 }
drh75897232000-05-29 14:26:00 +00003513 fclose(fp);
3514 return;
3515}
3516
3517/* Search for the file "name" which is in the same directory as
drhc2dbf352021-01-07 16:10:14 +00003518** the executable */
icculus9e44cf12010-02-14 17:14:22 +00003519PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00003520{
icculus9e44cf12010-02-14 17:14:22 +00003521 const char *pathlist;
mistachkin273ee152020-09-21 20:18:44 +00003522 char *pathbufptr = 0;
drh79d9af92020-09-20 12:10:28 +00003523 char *pathbuf = 0;
drh75897232000-05-29 14:26:00 +00003524 char *path,*cp;
3525 char c;
drh75897232000-05-29 14:26:00 +00003526
3527#ifdef __WIN32__
3528 cp = strrchr(argv0,'\\');
3529#else
3530 cp = strrchr(argv0,'/');
3531#endif
3532 if( cp ){
3533 c = *cp;
3534 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003535 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh898799f2014-01-10 23:21:00 +00003536 if( path ) lemon_sprintf(path,"%s/%s",argv0,name);
drh75897232000-05-29 14:26:00 +00003537 *cp = c;
3538 }else{
drh75897232000-05-29 14:26:00 +00003539 pathlist = getenv("PATH");
3540 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003541 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003542 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003543 if( (pathbuf != 0) && (path!=0) ){
3544 pathbufptr = pathbuf;
drh898799f2014-01-10 23:21:00 +00003545 lemon_strcpy(pathbuf, pathlist);
icculus9e44cf12010-02-14 17:14:22 +00003546 while( *pathbuf ){
3547 cp = strchr(pathbuf,':');
3548 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003549 c = *cp;
3550 *cp = 0;
drh898799f2014-01-10 23:21:00 +00003551 lemon_sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003552 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003553 if( c==0 ) pathbuf[0] = 0;
3554 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003555 if( access(path,modemask)==0 ) break;
3556 }
3557 }
drh79d9af92020-09-20 12:10:28 +00003558 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003559 }
3560 return path;
3561}
3562
3563/* Given an action, compute the integer value for that action
3564** which is to be put in the action table of the generated machine.
3565** Return negative if no action should be generated.
3566*/
icculus9e44cf12010-02-14 17:14:22 +00003567PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003568{
3569 int act;
3570 switch( ap->type ){
drh3bd48ab2015-09-07 18:23:37 +00003571 case SHIFT: act = ap->x.stp->statenum; break;
drhbd8fcc12017-06-28 11:56:18 +00003572 case SHIFTREDUCE: {
drhbd8fcc12017-06-28 11:56:18 +00003573 /* Since a SHIFT is inherient after a prior REDUCE, convert any
3574 ** SHIFTREDUCE action with a nonterminal on the LHS into a simple
3575 ** REDUCE action: */
drh6d77a292021-08-27 11:26:37 +00003576 if( ap->sp->index>=lemp->nterminal
3577 && (lemp->errsym==0 || ap->sp->index!=lemp->errsym->index)
3578 ){
drh5c8241b2017-12-24 23:38:10 +00003579 act = lemp->minReduce + ap->x.rp->iRule;
3580 }else{
3581 act = lemp->minShiftReduce + ap->x.rp->iRule;
3582 }
drhbd8fcc12017-06-28 11:56:18 +00003583 break;
3584 }
drh5c8241b2017-12-24 23:38:10 +00003585 case REDUCE: act = lemp->minReduce + ap->x.rp->iRule; break;
3586 case ERROR: act = lemp->errAction; break;
3587 case ACCEPT: act = lemp->accAction; break;
drh75897232000-05-29 14:26:00 +00003588 default: act = -1; break;
3589 }
3590 return act;
3591}
3592
3593#define LINESIZE 1000
3594/* The next cluster of routines are for reading the template file
3595** and writing the results to the generated parser */
3596/* The first function transfers data from "in" to "out" until
3597** a line is seen which begins with "%%". The line number is
3598** tracked.
3599**
3600** if name!=0, then any word that begin with "Parse" is changed to
3601** begin with *name instead.
3602*/
icculus9e44cf12010-02-14 17:14:22 +00003603PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003604{
3605 int i, iStart;
3606 char line[LINESIZE];
3607 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3608 (*lineno)++;
3609 iStart = 0;
3610 if( name ){
3611 for(i=0; line[i]; i++){
3612 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
drhc56fac72015-10-29 13:48:15 +00003613 && (i==0 || !ISALPHA(line[i-1]))
drh75897232000-05-29 14:26:00 +00003614 ){
3615 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3616 fprintf(out,"%s",name);
3617 i += 4;
3618 iStart = i+1;
3619 }
3620 }
3621 }
3622 fprintf(out,"%s",&line[iStart]);
3623 }
3624}
3625
drh60c71b02020-09-01 11:20:03 +00003626/* Skip forward past the header of the template file to the first "%%"
3627*/
3628PRIVATE void tplt_skip_header(FILE *in, int *lineno)
3629{
3630 char line[LINESIZE];
3631 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3632 (*lineno)++;
3633 }
3634}
3635
drh75897232000-05-29 14:26:00 +00003636/* The next function finds the template file and opens it, returning
3637** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003638PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003639{
3640 static char templatename[] = "lempar.c";
3641 char buf[1000];
3642 FILE *in;
3643 char *tpltname;
drh79d9af92020-09-20 12:10:28 +00003644 char *toFree = 0;
drh75897232000-05-29 14:26:00 +00003645 char *cp;
3646
icculus3e143bd2010-02-14 00:48:49 +00003647 /* first, see if user specified a template filename on the command line. */
3648 if (user_templatename != 0) {
3649 if( access(user_templatename,004)==-1 ){
3650 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3651 user_templatename);
3652 lemp->errorcnt++;
3653 return 0;
3654 }
3655 in = fopen(user_templatename,"rb");
3656 if( in==0 ){
drh25473362015-09-04 18:03:45 +00003657 fprintf(stderr,"Can't open the template file \"%s\".\n",
3658 user_templatename);
icculus3e143bd2010-02-14 00:48:49 +00003659 lemp->errorcnt++;
3660 return 0;
3661 }
3662 return in;
3663 }
3664
drh75897232000-05-29 14:26:00 +00003665 cp = strrchr(lemp->filename,'.');
3666 if( cp ){
drh898799f2014-01-10 23:21:00 +00003667 lemon_sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003668 }else{
drh898799f2014-01-10 23:21:00 +00003669 lemon_sprintf(buf,"%s.lt",lemp->filename);
drh75897232000-05-29 14:26:00 +00003670 }
3671 if( access(buf,004)==0 ){
3672 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003673 }else if( access(templatename,004)==0 ){
3674 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003675 }else{
drh79d9af92020-09-20 12:10:28 +00003676 toFree = tpltname = pathsearch(lemp->argv0,templatename,0);
drh75897232000-05-29 14:26:00 +00003677 }
3678 if( tpltname==0 ){
3679 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3680 templatename);
3681 lemp->errorcnt++;
3682 return 0;
3683 }
drh2aa6ca42004-09-10 00:14:04 +00003684 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003685 if( in==0 ){
drh79d9af92020-09-20 12:10:28 +00003686 fprintf(stderr,"Can't open the template file \"%s\".\n",tpltname);
drh75897232000-05-29 14:26:00 +00003687 lemp->errorcnt++;
drh75897232000-05-29 14:26:00 +00003688 }
drh79d9af92020-09-20 12:10:28 +00003689 free(toFree);
drh75897232000-05-29 14:26:00 +00003690 return in;
3691}
3692
drhaf805ca2004-09-07 11:28:25 +00003693/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003694PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003695{
3696 fprintf(out,"#line %d \"",lineno);
3697 while( *filename ){
3698 if( *filename == '\\' ) putc('\\',out);
3699 putc(*filename,out);
3700 filename++;
3701 }
3702 fprintf(out,"\"\n");
3703}
3704
drh75897232000-05-29 14:26:00 +00003705/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003706PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003707{
3708 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003709 while( *str ){
drh75897232000-05-29 14:26:00 +00003710 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003711 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003712 str++;
3713 }
drh9db55df2004-09-09 14:01:21 +00003714 if( str[-1]!='\n' ){
3715 putc('\n',out);
3716 (*lineno)++;
3717 }
shane58543932008-12-10 20:10:04 +00003718 if (!lemp->nolinenosflag) {
drh06f60d82017-04-14 19:46:12 +00003719 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003720 }
drh75897232000-05-29 14:26:00 +00003721 return;
3722}
3723
3724/*
3725** The following routine emits code for the destructor for the
3726** symbol sp
3727*/
icculus9e44cf12010-02-14 17:14:22 +00003728void emit_destructor_code(
3729 FILE *out,
3730 struct symbol *sp,
3731 struct lemon *lemp,
3732 int *lineno
3733){
drhcc83b6e2004-04-23 23:38:42 +00003734 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003735
drh75897232000-05-29 14:26:00 +00003736 if( sp->type==TERMINAL ){
3737 cp = lemp->tokendest;
3738 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003739 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003740 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003741 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003742 fprintf(out,"{\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00003743 if( !lemp->nolinenosflag ){
3744 (*lineno)++;
3745 tplt_linedir(out,sp->destLineno,lemp->filename);
3746 }
drh960e8c62001-04-03 16:53:21 +00003747 }else if( lemp->vardest ){
3748 cp = lemp->vardest;
3749 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003750 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003751 }else{
3752 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003753 }
3754 for(; *cp; cp++){
3755 if( *cp=='$' && cp[1]=='$' ){
3756 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3757 cp++;
3758 continue;
3759 }
shane58543932008-12-10 20:10:04 +00003760 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003761 fputc(*cp,out);
3762 }
shane58543932008-12-10 20:10:04 +00003763 fprintf(out,"\n"); (*lineno)++;
drh06f60d82017-04-14 19:46:12 +00003764 if (!lemp->nolinenosflag) {
3765 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003766 }
3767 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003768 return;
3769}
3770
3771/*
drh960e8c62001-04-03 16:53:21 +00003772** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003773*/
icculus9e44cf12010-02-14 17:14:22 +00003774int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003775{
3776 int ret;
3777 if( sp->type==TERMINAL ){
3778 ret = lemp->tokendest!=0;
3779 }else{
drh960e8c62001-04-03 16:53:21 +00003780 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003781 }
3782 return ret;
3783}
3784
drh0bb132b2004-07-20 14:06:51 +00003785/*
3786** Append text to a dynamically allocated string. If zText is 0 then
3787** reset the string to be empty again. Always return the complete text
3788** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003789**
3790** n bytes of zText are stored. If n==0 then all of zText up to the first
3791** \000 terminator is stored. zText can contain up to two instances of
3792** %d. The values of p1 and p2 are written into the first and second
3793** %d.
3794**
3795** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003796*/
icculus9e44cf12010-02-14 17:14:22 +00003797PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3798 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003799 static char *z = 0;
3800 static int alloced = 0;
3801 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003802 int c;
drh0bb132b2004-07-20 14:06:51 +00003803 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003804 if( zText==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003805 if( used==0 && z!=0 ) z[0] = 0;
drh0bb132b2004-07-20 14:06:51 +00003806 used = 0;
3807 return z;
3808 }
drh7ac25c72004-08-19 15:12:26 +00003809 if( n<=0 ){
3810 if( n<0 ){
3811 used += n;
3812 assert( used>=0 );
3813 }
drh87cf1372008-08-13 20:09:06 +00003814 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003815 }
drhdf609712010-11-23 20:55:27 +00003816 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003817 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003818 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003819 }
icculus9e44cf12010-02-14 17:14:22 +00003820 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003821 while( n-- > 0 ){
3822 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003823 if( c=='%' && n>0 && zText[0]=='d' ){
drh898799f2014-01-10 23:21:00 +00003824 lemon_sprintf(zInt, "%d", p1);
drh0bb132b2004-07-20 14:06:51 +00003825 p1 = p2;
drh898799f2014-01-10 23:21:00 +00003826 lemon_strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003827 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003828 zText++;
3829 n--;
3830 }else{
mistachkin2318d332015-01-12 18:02:52 +00003831 z[used++] = (char)c;
drh0bb132b2004-07-20 14:06:51 +00003832 }
3833 }
3834 z[used] = 0;
3835 return z;
3836}
3837
3838/*
drh711c9812016-05-23 14:24:31 +00003839** Write and transform the rp->code string so that symbols are expanded.
3840** Populate the rp->codePrefix and rp->codeSuffix strings, as appropriate.
drhdabd04c2016-02-17 01:46:19 +00003841**
3842** Return 1 if the expanded code requires that "yylhsminor" local variable
3843** to be defined.
drh0bb132b2004-07-20 14:06:51 +00003844*/
drhdabd04c2016-02-17 01:46:19 +00003845PRIVATE int translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003846 char *cp, *xp;
3847 int i;
drhcf82f0d2016-02-17 04:33:10 +00003848 int rc = 0; /* True if yylhsminor is used */
drh43303de2016-02-17 12:34:03 +00003849 int dontUseRhs0 = 0; /* If true, use of left-most RHS label is illegal */
drhcf82f0d2016-02-17 04:33:10 +00003850 const char *zSkip = 0; /* The zOvwrt comment within rp->code, or NULL */
3851 char lhsused = 0; /* True if the LHS element has been used */
3852 char lhsdirect; /* True if LHS writes directly into stack */
3853 char used[MAXRHS]; /* True for each RHS element which is used */
3854 char zLhs[50]; /* Convert the LHS symbol into this string */
3855 char zOvwrt[900]; /* Comment that to allow LHS to overwrite RHS */
drh0bb132b2004-07-20 14:06:51 +00003856
3857 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3858 lhsused = 0;
3859
drh19c9e562007-03-29 20:13:53 +00003860 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003861 static char newlinestr[2] = { '\n', '\0' };
3862 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003863 rp->line = rp->ruleline;
drh711c9812016-05-23 14:24:31 +00003864 rp->noCode = 1;
3865 }else{
3866 rp->noCode = 0;
drh19c9e562007-03-29 20:13:53 +00003867 }
3868
drh4dd0d3f2016-02-17 01:18:33 +00003869
drh2e55b042016-04-30 17:19:30 +00003870 if( rp->nrhs==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003871 /* If there are no RHS symbols, then writing directly to the LHS is ok */
3872 lhsdirect = 1;
3873 }else if( rp->rhsalias[0]==0 ){
drh2e55b042016-04-30 17:19:30 +00003874 /* The left-most RHS symbol has no value. LHS direct is ok. But
drhc2dbf352021-01-07 16:10:14 +00003875 ** we have to call the destructor on the RHS symbol first. */
drh4dd0d3f2016-02-17 01:18:33 +00003876 lhsdirect = 1;
3877 if( has_destructor(rp->rhs[0],lemp) ){
3878 append_str(0,0,0,0);
3879 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
3880 rp->rhs[0]->index,1-rp->nrhs);
3881 rp->codePrefix = Strsafe(append_str(0,0,0,0));
drh711c9812016-05-23 14:24:31 +00003882 rp->noCode = 0;
drh4dd0d3f2016-02-17 01:18:33 +00003883 }
drh2e55b042016-04-30 17:19:30 +00003884 }else if( rp->lhsalias==0 ){
3885 /* There is no LHS value symbol. */
3886 lhsdirect = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003887 }else if( strcmp(rp->lhsalias,rp->rhsalias[0])==0 ){
drh06f60d82017-04-14 19:46:12 +00003888 /* The LHS symbol and the left-most RHS symbol are the same, so
drh4dd0d3f2016-02-17 01:18:33 +00003889 ** direct writing is allowed */
3890 lhsdirect = 1;
3891 lhsused = 1;
3892 used[0] = 1;
3893 if( rp->lhs->dtnum!=rp->rhs[0]->dtnum ){
3894 ErrorMsg(lemp->filename,rp->ruleline,
3895 "%s(%s) and %s(%s) share the same label but have "
3896 "different datatypes.",
3897 rp->lhs->name, rp->lhsalias, rp->rhs[0]->name, rp->rhsalias[0]);
3898 lemp->errorcnt++;
drh06f60d82017-04-14 19:46:12 +00003899 }
drh4dd0d3f2016-02-17 01:18:33 +00003900 }else{
drhcf82f0d2016-02-17 04:33:10 +00003901 lemon_sprintf(zOvwrt, "/*%s-overwrites-%s*/",
3902 rp->lhsalias, rp->rhsalias[0]);
3903 zSkip = strstr(rp->code, zOvwrt);
3904 if( zSkip!=0 ){
3905 /* The code contains a special comment that indicates that it is safe
3906 ** for the LHS label to overwrite left-most RHS label. */
3907 lhsdirect = 1;
3908 }else{
3909 lhsdirect = 0;
3910 }
drh4dd0d3f2016-02-17 01:18:33 +00003911 }
3912 if( lhsdirect ){
3913 sprintf(zLhs, "yymsp[%d].minor.yy%d",1-rp->nrhs,rp->lhs->dtnum);
3914 }else{
drhdabd04c2016-02-17 01:46:19 +00003915 rc = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003916 sprintf(zLhs, "yylhsminor.yy%d",rp->lhs->dtnum);
3917 }
3918
drh0bb132b2004-07-20 14:06:51 +00003919 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003920
3921 /* This const cast is wrong but harmless, if we're careful. */
3922 for(cp=(char *)rp->code; *cp; cp++){
drhcf82f0d2016-02-17 04:33:10 +00003923 if( cp==zSkip ){
3924 append_str(zOvwrt,0,0,0);
3925 cp += lemonStrlen(zOvwrt)-1;
drh43303de2016-02-17 12:34:03 +00003926 dontUseRhs0 = 1;
drhcf82f0d2016-02-17 04:33:10 +00003927 continue;
3928 }
drhc56fac72015-10-29 13:48:15 +00003929 if( ISALPHA(*cp) && (cp==rp->code || (!ISALNUM(cp[-1]) && cp[-1]!='_')) ){
drh0bb132b2004-07-20 14:06:51 +00003930 char saved;
drhc56fac72015-10-29 13:48:15 +00003931 for(xp= &cp[1]; ISALNUM(*xp) || *xp=='_'; xp++);
drh0bb132b2004-07-20 14:06:51 +00003932 saved = *xp;
3933 *xp = 0;
3934 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003935 append_str(zLhs,0,0,0);
drh0bb132b2004-07-20 14:06:51 +00003936 cp = xp;
3937 lhsused = 1;
3938 }else{
3939 for(i=0; i<rp->nrhs; i++){
3940 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh43303de2016-02-17 12:34:03 +00003941 if( i==0 && dontUseRhs0 ){
3942 ErrorMsg(lemp->filename,rp->ruleline,
3943 "Label %s used after '%s'.",
3944 rp->rhsalias[0], zOvwrt);
3945 lemp->errorcnt++;
3946 }else if( cp!=rp->code && cp[-1]=='@' ){
drh7ac25c72004-08-19 15:12:26 +00003947 /* If the argument is of the form @X then substituted
3948 ** the token number of X, not the value of X */
3949 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3950 }else{
drhfd405312005-11-06 04:06:59 +00003951 struct symbol *sp = rp->rhs[i];
3952 int dtnum;
3953 if( sp->type==MULTITERMINAL ){
3954 dtnum = sp->subsym[0]->dtnum;
3955 }else{
3956 dtnum = sp->dtnum;
3957 }
3958 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003959 }
drh0bb132b2004-07-20 14:06:51 +00003960 cp = xp;
3961 used[i] = 1;
3962 break;
3963 }
3964 }
3965 }
3966 *xp = saved;
3967 }
3968 append_str(cp, 1, 0, 0);
3969 } /* End loop */
3970
drh4dd0d3f2016-02-17 01:18:33 +00003971 /* Main code generation completed */
3972 cp = append_str(0,0,0,0);
3973 if( cp && cp[0] ) rp->code = Strsafe(cp);
3974 append_str(0,0,0,0);
3975
drh0bb132b2004-07-20 14:06:51 +00003976 /* Check to make sure the LHS has been used */
3977 if( rp->lhsalias && !lhsused ){
3978 ErrorMsg(lemp->filename,rp->ruleline,
3979 "Label \"%s\" for \"%s(%s)\" is never used.",
3980 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3981 lemp->errorcnt++;
3982 }
3983
drh4dd0d3f2016-02-17 01:18:33 +00003984 /* Generate destructor code for RHS minor values which are not referenced.
3985 ** Generate error messages for unused labels and duplicate labels.
3986 */
drh0bb132b2004-07-20 14:06:51 +00003987 for(i=0; i<rp->nrhs; i++){
drh4dd0d3f2016-02-17 01:18:33 +00003988 if( rp->rhsalias[i] ){
3989 if( i>0 ){
3990 int j;
3991 if( rp->lhsalias && strcmp(rp->lhsalias,rp->rhsalias[i])==0 ){
3992 ErrorMsg(lemp->filename,rp->ruleline,
3993 "%s(%s) has the same label as the LHS but is not the left-most "
3994 "symbol on the RHS.",
drhf135cb72019-04-30 14:26:31 +00003995 rp->rhs[i]->name, rp->rhsalias[i]);
drh4dd0d3f2016-02-17 01:18:33 +00003996 lemp->errorcnt++;
3997 }
3998 for(j=0; j<i; j++){
3999 if( rp->rhsalias[j] && strcmp(rp->rhsalias[j],rp->rhsalias[i])==0 ){
4000 ErrorMsg(lemp->filename,rp->ruleline,
4001 "Label %s used for multiple symbols on the RHS of a rule.",
4002 rp->rhsalias[i]);
4003 lemp->errorcnt++;
4004 break;
4005 }
4006 }
drh0bb132b2004-07-20 14:06:51 +00004007 }
drh4dd0d3f2016-02-17 01:18:33 +00004008 if( !used[i] ){
4009 ErrorMsg(lemp->filename,rp->ruleline,
4010 "Label %s for \"%s(%s)\" is never used.",
4011 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
4012 lemp->errorcnt++;
4013 }
4014 }else if( i>0 && has_destructor(rp->rhs[i],lemp) ){
4015 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
4016 rp->rhs[i]->index,i-rp->nrhs+1);
drh0bb132b2004-07-20 14:06:51 +00004017 }
4018 }
drh4dd0d3f2016-02-17 01:18:33 +00004019
4020 /* If unable to write LHS values directly into the stack, write the
4021 ** saved LHS value now. */
4022 if( lhsdirect==0 ){
4023 append_str(" yymsp[%d].minor.yy%d = ", 0, 1-rp->nrhs, rp->lhs->dtnum);
4024 append_str(zLhs, 0, 0, 0);
4025 append_str(";\n", 0, 0, 0);
drh61e339a2007-01-16 03:09:02 +00004026 }
drh4dd0d3f2016-02-17 01:18:33 +00004027
4028 /* Suffix code generation complete */
4029 cp = append_str(0,0,0,0);
drh711c9812016-05-23 14:24:31 +00004030 if( cp && cp[0] ){
4031 rp->codeSuffix = Strsafe(cp);
4032 rp->noCode = 0;
4033 }
drhdabd04c2016-02-17 01:46:19 +00004034
4035 return rc;
drh0bb132b2004-07-20 14:06:51 +00004036}
4037
drh06f60d82017-04-14 19:46:12 +00004038/*
drh75897232000-05-29 14:26:00 +00004039** Generate code which executes when the rule "rp" is reduced. Write
4040** the code to "out". Make sure lineno stays up-to-date.
4041*/
icculus9e44cf12010-02-14 17:14:22 +00004042PRIVATE void emit_code(
4043 FILE *out,
4044 struct rule *rp,
4045 struct lemon *lemp,
4046 int *lineno
4047){
4048 const char *cp;
drh75897232000-05-29 14:26:00 +00004049
drh4dd0d3f2016-02-17 01:18:33 +00004050 /* Setup code prior to the #line directive */
4051 if( rp->codePrefix && rp->codePrefix[0] ){
4052 fprintf(out, "{%s", rp->codePrefix);
4053 for(cp=rp->codePrefix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
4054 }
4055
drh75897232000-05-29 14:26:00 +00004056 /* Generate code to do the reduce action */
4057 if( rp->code ){
drh25473362015-09-04 18:03:45 +00004058 if( !lemp->nolinenosflag ){
4059 (*lineno)++;
4060 tplt_linedir(out,rp->line,lemp->filename);
4061 }
drhaf805ca2004-09-07 11:28:25 +00004062 fprintf(out,"{%s",rp->code);
drh4dd0d3f2016-02-17 01:18:33 +00004063 for(cp=rp->code; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
shane58543932008-12-10 20:10:04 +00004064 fprintf(out,"}\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00004065 if( !lemp->nolinenosflag ){
4066 (*lineno)++;
4067 tplt_linedir(out,*lineno,lemp->outname);
4068 }
drh4dd0d3f2016-02-17 01:18:33 +00004069 }
4070
4071 /* Generate breakdown code that occurs after the #line directive */
4072 if( rp->codeSuffix && rp->codeSuffix[0] ){
4073 fprintf(out, "%s", rp->codeSuffix);
4074 for(cp=rp->codeSuffix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
4075 }
4076
4077 if( rp->codePrefix ){
4078 fprintf(out, "}\n"); (*lineno)++;
4079 }
drh75897232000-05-29 14:26:00 +00004080
drh75897232000-05-29 14:26:00 +00004081 return;
4082}
4083
4084/*
4085** Print the definition of the union used for the parser's data stack.
4086** This union contains fields for every possible data type for tokens
4087** and nonterminals. In the process of computing and printing this
4088** union, also set the ".dtnum" field of every terminal and nonterminal
4089** symbol.
4090*/
icculus9e44cf12010-02-14 17:14:22 +00004091void print_stack_union(
4092 FILE *out, /* The output stream */
4093 struct lemon *lemp, /* The main info structure for this parser */
4094 int *plineno, /* Pointer to the line number */
4095 int mhflag /* True if generating makeheaders output */
4096){
drh10761302021-10-04 16:14:51 +00004097 int lineno; /* The line number of the output */
drh75897232000-05-29 14:26:00 +00004098 char **types; /* A hash table of datatypes */
4099 int arraysize; /* Size of the "types" array */
4100 int maxdtlength; /* Maximum length of any ".datatype" field. */
4101 char *stddt; /* Standardized name for a datatype */
4102 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00004103 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00004104 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00004105
4106 /* Allocate and initialize types[] and allocate stddt[] */
4107 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00004108 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00004109 if( types==0 ){
4110 fprintf(stderr,"Out of memory.\n");
4111 exit(1);
4112 }
drh75897232000-05-29 14:26:00 +00004113 for(i=0; i<arraysize; i++) types[i] = 0;
4114 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00004115 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00004116 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00004117 }
drh75897232000-05-29 14:26:00 +00004118 for(i=0; i<lemp->nsymbol; i++){
4119 int len;
4120 struct symbol *sp = lemp->symbols[i];
4121 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00004122 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00004123 if( len>maxdtlength ) maxdtlength = len;
4124 }
4125 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00004126 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00004127 fprintf(stderr,"Out of memory.\n");
4128 exit(1);
4129 }
4130
4131 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
4132 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00004133 ** used for terminal symbols. If there is no %default_type defined then
4134 ** 0 is also used as the .dtnum value for nonterminals which do not specify
4135 ** a datatype using the %type directive.
4136 */
drh75897232000-05-29 14:26:00 +00004137 for(i=0; i<lemp->nsymbol; i++){
4138 struct symbol *sp = lemp->symbols[i];
4139 char *cp;
4140 if( sp==lemp->errsym ){
4141 sp->dtnum = arraysize+1;
4142 continue;
4143 }
drh960e8c62001-04-03 16:53:21 +00004144 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00004145 sp->dtnum = 0;
4146 continue;
4147 }
4148 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00004149 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00004150 j = 0;
drhc56fac72015-10-29 13:48:15 +00004151 while( ISSPACE(*cp) ) cp++;
drh75897232000-05-29 14:26:00 +00004152 while( *cp ) stddt[j++] = *cp++;
drhc56fac72015-10-29 13:48:15 +00004153 while( j>0 && ISSPACE(stddt[j-1]) ) j--;
drh75897232000-05-29 14:26:00 +00004154 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00004155 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00004156 sp->dtnum = 0;
4157 continue;
4158 }
drh75897232000-05-29 14:26:00 +00004159 hash = 0;
4160 for(j=0; stddt[j]; j++){
4161 hash = hash*53 + stddt[j];
4162 }
drh3b2129c2003-05-13 00:34:21 +00004163 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00004164 while( types[hash] ){
4165 if( strcmp(types[hash],stddt)==0 ){
4166 sp->dtnum = hash + 1;
4167 break;
4168 }
4169 hash++;
drh2b51f212013-10-11 23:01:02 +00004170 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00004171 }
4172 if( types[hash]==0 ){
4173 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00004174 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00004175 if( types[hash]==0 ){
4176 fprintf(stderr,"Out of memory.\n");
4177 exit(1);
4178 }
drh898799f2014-01-10 23:21:00 +00004179 lemon_strcpy(types[hash],stddt);
drh75897232000-05-29 14:26:00 +00004180 }
4181 }
4182
4183 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
4184 name = lemp->name ? lemp->name : "Parse";
4185 lineno = *plineno;
4186 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
4187 fprintf(out,"#define %sTOKENTYPE %s\n",name,
4188 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
4189 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
4190 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00004191 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004192 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
4193 for(i=0; i<arraysize; i++){
4194 if( types[i]==0 ) continue;
4195 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
4196 free(types[i]);
4197 }
drhed0c15b2018-04-16 14:31:34 +00004198 if( lemp->errsym && lemp->errsym->useCnt ){
drhc4dd3fd2008-01-22 01:48:05 +00004199 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
4200 }
drh75897232000-05-29 14:26:00 +00004201 free(stddt);
4202 free(types);
4203 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
4204 *plineno = lineno;
4205}
4206
drhb29b0a52002-02-23 19:39:46 +00004207/*
4208** Return the name of a C datatype able to represent values between
drhc75e0162015-09-07 02:23:02 +00004209** lwr and upr, inclusive. If pnByte!=NULL then also write the sizeof
4210** for that type (1, 2, or 4) into *pnByte.
drhb29b0a52002-02-23 19:39:46 +00004211*/
drhc75e0162015-09-07 02:23:02 +00004212static const char *minimum_size_type(int lwr, int upr, int *pnByte){
4213 const char *zType = "int";
4214 int nByte = 4;
drh8b582012003-10-21 13:16:03 +00004215 if( lwr>=0 ){
4216 if( upr<=255 ){
drhc75e0162015-09-07 02:23:02 +00004217 zType = "unsigned char";
4218 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004219 }else if( upr<65535 ){
drhc75e0162015-09-07 02:23:02 +00004220 zType = "unsigned short int";
4221 nByte = 2;
drh8b582012003-10-21 13:16:03 +00004222 }else{
drhc75e0162015-09-07 02:23:02 +00004223 zType = "unsigned int";
4224 nByte = 4;
drh8b582012003-10-21 13:16:03 +00004225 }
4226 }else if( lwr>=-127 && upr<=127 ){
drhc75e0162015-09-07 02:23:02 +00004227 zType = "signed char";
4228 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004229 }else if( lwr>=-32767 && upr<32767 ){
drhc75e0162015-09-07 02:23:02 +00004230 zType = "short";
4231 nByte = 2;
drhb29b0a52002-02-23 19:39:46 +00004232 }
drhc75e0162015-09-07 02:23:02 +00004233 if( pnByte ) *pnByte = nByte;
4234 return zType;
drhb29b0a52002-02-23 19:39:46 +00004235}
4236
drhfdbf9282003-10-21 16:34:41 +00004237/*
4238** Each state contains a set of token transaction and a set of
4239** nonterminal transactions. Each of these sets makes an instance
4240** of the following structure. An array of these structures is used
4241** to order the creation of entries in the yy_action[] table.
4242*/
4243struct axset {
4244 struct state *stp; /* A pointer to a state */
4245 int isTkn; /* True to use tokens. False for non-terminals */
4246 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00004247 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00004248};
4249
4250/*
4251** Compare to axset structures for sorting purposes
4252*/
4253static int axset_compare(const void *a, const void *b){
4254 struct axset *p1 = (struct axset*)a;
4255 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00004256 int c;
4257 c = p2->nAction - p1->nAction;
4258 if( c==0 ){
drh337cd0d2015-09-07 23:40:42 +00004259 c = p1->iOrder - p2->iOrder;
drhe594bc32009-11-03 13:02:25 +00004260 }
4261 assert( c!=0 || p1==p2 );
4262 return c;
drhfdbf9282003-10-21 16:34:41 +00004263}
4264
drhc4dd3fd2008-01-22 01:48:05 +00004265/*
4266** Write text on "out" that describes the rule "rp".
4267*/
4268static void writeRuleText(FILE *out, struct rule *rp){
4269 int j;
4270 fprintf(out,"%s ::=", rp->lhs->name);
4271 for(j=0; j<rp->nrhs; j++){
4272 struct symbol *sp = rp->rhs[j];
drh61f92cd2014-01-11 03:06:18 +00004273 if( sp->type!=MULTITERMINAL ){
4274 fprintf(out," %s", sp->name);
4275 }else{
drhc4dd3fd2008-01-22 01:48:05 +00004276 int k;
drh61f92cd2014-01-11 03:06:18 +00004277 fprintf(out," %s", sp->subsym[0]->name);
drhc4dd3fd2008-01-22 01:48:05 +00004278 for(k=1; k<sp->nsubsym; k++){
4279 fprintf(out,"|%s",sp->subsym[k]->name);
4280 }
4281 }
4282 }
4283}
4284
4285
drh75897232000-05-29 14:26:00 +00004286/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004287void ReportTable(
4288 struct lemon *lemp,
drhfe03dac2019-11-26 02:22:39 +00004289 int mhflag, /* Output in makeheaders format if true */
4290 int sqlFlag /* Generate the *.sql file too */
icculus9e44cf12010-02-14 17:14:22 +00004291){
drhfe03dac2019-11-26 02:22:39 +00004292 FILE *out, *in, *sql;
drh75897232000-05-29 14:26:00 +00004293 int lineno;
4294 struct state *stp;
4295 struct action *ap;
4296 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00004297 struct acttab *pActtab;
drhc75e0162015-09-07 02:23:02 +00004298 int i, j, n, sz;
drh2e517162019-08-28 02:09:47 +00004299 int nLookAhead;
drhc75e0162015-09-07 02:23:02 +00004300 int szActionType; /* sizeof(YYACTIONTYPE) */
4301 int szCodeType; /* sizeof(YYCODETYPE) */
icculus9e44cf12010-02-14 17:14:22 +00004302 const char *name;
drh8b582012003-10-21 13:16:03 +00004303 int mnTknOfst, mxTknOfst;
4304 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00004305 struct axset *ax;
drh60c71b02020-09-01 11:20:03 +00004306 char *prefix;
drh75897232000-05-29 14:26:00 +00004307
drh5c8241b2017-12-24 23:38:10 +00004308 lemp->minShiftReduce = lemp->nstate;
4309 lemp->errAction = lemp->minShiftReduce + lemp->nrule;
4310 lemp->accAction = lemp->errAction + 1;
4311 lemp->noAction = lemp->accAction + 1;
4312 lemp->minReduce = lemp->noAction + 1;
4313 lemp->maxAction = lemp->minReduce + lemp->nrule;
4314
drh75897232000-05-29 14:26:00 +00004315 in = tplt_open(lemp);
4316 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00004317 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00004318 if( out==0 ){
4319 fclose(in);
4320 return;
4321 }
drhfe03dac2019-11-26 02:22:39 +00004322 if( sqlFlag==0 ){
4323 sql = 0;
4324 }else{
4325 sql = file_open(lemp, ".sql", "wb");
4326 if( sql==0 ){
4327 fclose(in);
4328 fclose(out);
4329 return;
4330 }
4331 fprintf(sql,
drh1417c2f2019-11-29 12:51:00 +00004332 "BEGIN;\n"
drhfe03dac2019-11-26 02:22:39 +00004333 "CREATE TABLE symbol(\n"
4334 " id INTEGER PRIMARY KEY,\n"
4335 " name TEXT NOT NULL,\n"
4336 " isTerminal BOOLEAN NOT NULL,\n"
drh1417c2f2019-11-29 12:51:00 +00004337 " fallback INTEGER REFERENCES symbol"
4338 " DEFERRABLE INITIALLY DEFERRED\n"
drhfe03dac2019-11-26 02:22:39 +00004339 ");\n"
4340 );
4341 for(i=0; i<lemp->nsymbol; i++){
4342 fprintf(sql,
4343 "INSERT INTO symbol(id,name,isTerminal,fallback)"
4344 "VALUES(%d,'%s',%s",
4345 i, lemp->symbols[i]->name,
4346 i<lemp->nterminal ? "TRUE" : "FALSE"
4347 );
4348 if( lemp->symbols[i]->fallback ){
4349 fprintf(sql, ",%d);\n", lemp->symbols[i]->fallback->index);
4350 }else{
4351 fprintf(sql, ",NULL);\n");
4352 }
4353 }
4354 fprintf(sql,
4355 "CREATE TABLE rule(\n"
4356 " ruleid INTEGER PRIMARY KEY,\n"
drh3e5f7fe2019-12-19 12:29:31 +00004357 " lhs INTEGER REFERENCES symbol(id),\n"
4358 " txt TEXT\n"
drhfe03dac2019-11-26 02:22:39 +00004359 ");\n"
4360 "CREATE TABLE rulerhs(\n"
4361 " ruleid INTEGER REFERENCES rule(ruleid),\n"
4362 " pos INTEGER,\n"
4363 " sym INTEGER REFERENCES symbol(id)\n"
4364 ");\n"
4365 );
4366 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4367 assert( i==rp->iRule );
4368 fprintf(sql,
drh59c56792019-12-19 13:17:07 +00004369 "INSERT INTO rule(ruleid,lhs,txt)VALUES(%d,%d,'",
drhfe03dac2019-11-26 02:22:39 +00004370 rp->iRule, rp->lhs->index
4371 );
drh3e5f7fe2019-12-19 12:29:31 +00004372 writeRuleText(sql, rp);
4373 fprintf(sql,"');\n");
drhfe03dac2019-11-26 02:22:39 +00004374 for(j=0; j<rp->nrhs; j++){
4375 struct symbol *sp = rp->rhs[j];
4376 if( sp->type!=MULTITERMINAL ){
4377 fprintf(sql,
4378 "INSERT INTO rulerhs(ruleid,pos,sym)VALUES(%d,%d,%d);\n",
4379 i,j,sp->index
4380 );
4381 }else{
4382 int k;
4383 for(k=0; k<sp->nsubsym; k++){
4384 fprintf(sql,
4385 "INSERT INTO rulerhs(ruleid,pos,sym)VALUES(%d,%d,%d);\n",
4386 i,j,sp->subsym[k]->index
4387 );
4388 }
4389 }
4390 }
4391 }
drh1417c2f2019-11-29 12:51:00 +00004392 fprintf(sql, "COMMIT;\n");
drhfe03dac2019-11-26 02:22:39 +00004393 }
drh75897232000-05-29 14:26:00 +00004394 lineno = 1;
drh60c71b02020-09-01 11:20:03 +00004395
drh512aa782020-09-01 12:26:55 +00004396 fprintf(out,
4397 "/* This file is automatically generated by Lemon from input grammar\n"
4398 "** source file \"%s\". */\n", lemp->filename); lineno += 2;
4399
drh60c71b02020-09-01 11:20:03 +00004400 /* The first %include directive begins with a C-language comment,
4401 ** then skip over the header comment of the template file
4402 */
4403 if( lemp->include==0 ) lemp->include = "";
4404 for(i=0; ISSPACE(lemp->include[i]); i++){
4405 if( lemp->include[i]=='\n' ){
4406 lemp->include += i+1;
4407 i = -1;
4408 }
4409 }
4410 if( lemp->include[0]=='/' ){
4411 tplt_skip_header(in,&lineno);
4412 }else{
4413 tplt_xfer(lemp->name,in,out,&lineno);
4414 }
drh75897232000-05-29 14:26:00 +00004415
4416 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00004417 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00004418 if( mhflag ){
mistachkin8e189222015-04-19 21:43:16 +00004419 char *incName = file_makename(lemp, ".h");
4420 fprintf(out,"#include \"%s\"\n", incName); lineno++;
4421 free(incName);
drh75897232000-05-29 14:26:00 +00004422 }
4423 tplt_xfer(lemp->name,in,out,&lineno);
4424
4425 /* Generate #defines for all tokens */
drh60c71b02020-09-01 11:20:03 +00004426 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4427 else prefix = "";
drh75897232000-05-29 14:26:00 +00004428 if( mhflag ){
drh75897232000-05-29 14:26:00 +00004429 fprintf(out,"#if INTERFACE\n"); lineno++;
drh60c71b02020-09-01 11:20:03 +00004430 }else{
4431 fprintf(out,"#ifndef %s%s\n", prefix, lemp->symbols[1]->name);
drh75897232000-05-29 14:26:00 +00004432 }
drh60c71b02020-09-01 11:20:03 +00004433 for(i=1; i<lemp->nterminal; i++){
4434 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4435 lineno++;
4436 }
4437 fprintf(out,"#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004438 tplt_xfer(lemp->name,in,out,&lineno);
4439
4440 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00004441 fprintf(out,"#define YYCODETYPE %s\n",
drhed0c15b2018-04-16 14:31:34 +00004442 minimum_size_type(0, lemp->nsymbol, &szCodeType)); lineno++;
4443 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol); lineno++;
drh75897232000-05-29 14:26:00 +00004444 fprintf(out,"#define YYACTIONTYPE %s\n",
drh5c8241b2017-12-24 23:38:10 +00004445 minimum_size_type(0,lemp->maxAction,&szActionType)); lineno++;
drhe09daa92006-06-10 13:29:31 +00004446 if( lemp->wildcard ){
4447 fprintf(out,"#define YYWILDCARD %d\n",
4448 lemp->wildcard->index); lineno++;
4449 }
drh75897232000-05-29 14:26:00 +00004450 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00004451 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004452 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00004453 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
4454 }else{
4455 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
4456 }
drhca44b5a2007-02-22 23:06:58 +00004457 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004458 if( mhflag ){
4459 fprintf(out,"#if INTERFACE\n"); lineno++;
4460 }
4461 name = lemp->name ? lemp->name : "Parse";
4462 if( lemp->arg && lemp->arg[0] ){
drh87cf1372008-08-13 20:09:06 +00004463 i = lemonStrlen(lemp->arg);
drhc56fac72015-10-29 13:48:15 +00004464 while( i>=1 && ISSPACE(lemp->arg[i-1]) ) i--;
4465 while( i>=1 && (ISALNUM(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00004466 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
4467 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
drhfb32c442018-04-21 13:51:42 +00004468 fprintf(out,"#define %sARG_PARAM ,%s\n",name,&lemp->arg[i]); lineno++;
4469 fprintf(out,"#define %sARG_FETCH %s=yypParser->%s;\n",
drh1f245e42002-03-11 13:55:50 +00004470 name,lemp->arg,&lemp->arg[i]); lineno++;
drhfb32c442018-04-21 13:51:42 +00004471 fprintf(out,"#define %sARG_STORE yypParser->%s=%s;\n",
drh1f245e42002-03-11 13:55:50 +00004472 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00004473 }else{
drhfb32c442018-04-21 13:51:42 +00004474 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
4475 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
4476 fprintf(out,"#define %sARG_PARAM\n",name); lineno++;
drh1f245e42002-03-11 13:55:50 +00004477 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
4478 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00004479 }
drhfb32c442018-04-21 13:51:42 +00004480 if( lemp->ctx && lemp->ctx[0] ){
4481 i = lemonStrlen(lemp->ctx);
4482 while( i>=1 && ISSPACE(lemp->ctx[i-1]) ) i--;
4483 while( i>=1 && (ISALNUM(lemp->ctx[i-1]) || lemp->ctx[i-1]=='_') ) i--;
4484 fprintf(out,"#define %sCTX_SDECL %s;\n",name,lemp->ctx); lineno++;
4485 fprintf(out,"#define %sCTX_PDECL ,%s\n",name,lemp->ctx); lineno++;
4486 fprintf(out,"#define %sCTX_PARAM ,%s\n",name,&lemp->ctx[i]); lineno++;
4487 fprintf(out,"#define %sCTX_FETCH %s=yypParser->%s;\n",
4488 name,lemp->ctx,&lemp->ctx[i]); lineno++;
4489 fprintf(out,"#define %sCTX_STORE yypParser->%s=%s;\n",
4490 name,&lemp->ctx[i],&lemp->ctx[i]); lineno++;
4491 }else{
4492 fprintf(out,"#define %sCTX_SDECL\n",name); lineno++;
4493 fprintf(out,"#define %sCTX_PDECL\n",name); lineno++;
4494 fprintf(out,"#define %sCTX_PARAM\n",name); lineno++;
4495 fprintf(out,"#define %sCTX_FETCH\n",name); lineno++;
4496 fprintf(out,"#define %sCTX_STORE\n",name); lineno++;
4497 }
drh75897232000-05-29 14:26:00 +00004498 if( mhflag ){
4499 fprintf(out,"#endif\n"); lineno++;
4500 }
drhed0c15b2018-04-16 14:31:34 +00004501 if( lemp->errsym && lemp->errsym->useCnt ){
drh3bd48ab2015-09-07 18:23:37 +00004502 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
4503 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00004504 }
drh0bd1f4e2002-06-06 18:54:39 +00004505 if( lemp->has_fallback ){
4506 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
4507 }
drh75897232000-05-29 14:26:00 +00004508
drh3bd48ab2015-09-07 18:23:37 +00004509 /* Compute the action table, but do not output it yet. The action
4510 ** table must be computed before generating the YYNSTATE macro because
4511 ** we need to know how many states can be eliminated.
drh75897232000-05-29 14:26:00 +00004512 */
drh3bd48ab2015-09-07 18:23:37 +00004513 ax = (struct axset *) calloc(lemp->nxstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00004514 if( ax==0 ){
4515 fprintf(stderr,"malloc failed\n");
4516 exit(1);
4517 }
drh3bd48ab2015-09-07 18:23:37 +00004518 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00004519 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004520 ax[i*2].stp = stp;
4521 ax[i*2].isTkn = 1;
4522 ax[i*2].nAction = stp->nTknAct;
4523 ax[i*2+1].stp = stp;
4524 ax[i*2+1].isTkn = 0;
4525 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00004526 }
drh8b582012003-10-21 13:16:03 +00004527 mxTknOfst = mnTknOfst = 0;
4528 mxNtOfst = mnNtOfst = 0;
drh3bd48ab2015-09-07 18:23:37 +00004529 /* In an effort to minimize the action table size, use the heuristic
4530 ** of placing the largest action sets first */
4531 for(i=0; i<lemp->nxstate*2; i++) ax[i].iOrder = i;
4532 qsort(ax, lemp->nxstate*2, sizeof(ax[0]), axset_compare);
drh3a9d6c72017-12-25 04:15:38 +00004533 pActtab = acttab_alloc(lemp->nsymbol, lemp->nterminal);
drh3bd48ab2015-09-07 18:23:37 +00004534 for(i=0; i<lemp->nxstate*2 && ax[i].nAction>0; i++){
drhfdbf9282003-10-21 16:34:41 +00004535 stp = ax[i].stp;
4536 if( ax[i].isTkn ){
4537 for(ap=stp->ap; ap; ap=ap->next){
4538 int action;
4539 if( ap->sp->index>=lemp->nterminal ) continue;
4540 action = compute_action(lemp, ap);
4541 if( action<0 ) continue;
4542 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004543 }
drh3a9d6c72017-12-25 04:15:38 +00004544 stp->iTknOfst = acttab_insert(pActtab, 1);
drhfdbf9282003-10-21 16:34:41 +00004545 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
4546 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
4547 }else{
4548 for(ap=stp->ap; ap; ap=ap->next){
4549 int action;
4550 if( ap->sp->index<lemp->nterminal ) continue;
4551 if( ap->sp->index==lemp->nsymbol ) continue;
4552 action = compute_action(lemp, ap);
4553 if( action<0 ) continue;
4554 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004555 }
drh3a9d6c72017-12-25 04:15:38 +00004556 stp->iNtOfst = acttab_insert(pActtab, 0);
drhfdbf9282003-10-21 16:34:41 +00004557 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
4558 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00004559 }
drh337cd0d2015-09-07 23:40:42 +00004560#if 0 /* Uncomment for a trace of how the yy_action[] table fills out */
4561 { int jj, nn;
4562 for(jj=nn=0; jj<pActtab->nAction; jj++){
4563 if( pActtab->aAction[jj].action<0 ) nn++;
4564 }
4565 printf("%4d: State %3d %s n: %2d size: %5d freespace: %d\n",
4566 i, stp->statenum, ax[i].isTkn ? "Token" : "Var ",
4567 ax[i].nAction, pActtab->nAction, nn);
4568 }
4569#endif
drh8b582012003-10-21 13:16:03 +00004570 }
drhfdbf9282003-10-21 16:34:41 +00004571 free(ax);
drh8b582012003-10-21 13:16:03 +00004572
drh756b41e2016-05-24 18:55:08 +00004573 /* Mark rules that are actually used for reduce actions after all
4574 ** optimizations have been applied
4575 */
4576 for(rp=lemp->rule; rp; rp=rp->next) rp->doesReduce = LEMON_FALSE;
4577 for(i=0; i<lemp->nxstate; i++){
drh756b41e2016-05-24 18:55:08 +00004578 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
4579 if( ap->type==REDUCE || ap->type==SHIFTREDUCE ){
drh69bfa552017-04-26 04:32:17 +00004580 ap->x.rp->doesReduce = 1;
drh756b41e2016-05-24 18:55:08 +00004581 }
4582 }
4583 }
4584
drh3bd48ab2015-09-07 18:23:37 +00004585 /* Finish rendering the constants now that the action table has
4586 ** been computed */
4587 fprintf(out,"#define YYNSTATE %d\n",lemp->nxstate); lineno++;
4588 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhce678c22019-12-11 18:53:51 +00004589 fprintf(out,"#define YYNRULE_WITH_ACTION %d\n",lemp->nruleWithAction);
4590 lineno++;
drh0d9de992017-12-26 18:04:23 +00004591 fprintf(out,"#define YYNTOKEN %d\n",lemp->nterminal); lineno++;
drh337cd0d2015-09-07 23:40:42 +00004592 fprintf(out,"#define YY_MAX_SHIFT %d\n",lemp->nxstate-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004593 i = lemp->minShiftReduce;
4594 fprintf(out,"#define YY_MIN_SHIFTREDUCE %d\n",i); lineno++;
4595 i += lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004596 fprintf(out,"#define YY_MAX_SHIFTREDUCE %d\n", i-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004597 fprintf(out,"#define YY_ERROR_ACTION %d\n", lemp->errAction); lineno++;
4598 fprintf(out,"#define YY_ACCEPT_ACTION %d\n", lemp->accAction); lineno++;
4599 fprintf(out,"#define YY_NO_ACTION %d\n", lemp->noAction); lineno++;
4600 fprintf(out,"#define YY_MIN_REDUCE %d\n", lemp->minReduce); lineno++;
4601 i = lemp->minReduce + lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004602 fprintf(out,"#define YY_MAX_REDUCE %d\n", i-1); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004603 tplt_xfer(lemp->name,in,out,&lineno);
4604
4605 /* Now output the action table and its associates:
4606 **
4607 ** yy_action[] A single table containing all actions.
4608 ** yy_lookahead[] A table containing the lookahead for each entry in
4609 ** yy_action. Used to detect hash collisions.
4610 ** yy_shift_ofst[] For each state, the offset into yy_action for
4611 ** shifting terminals.
4612 ** yy_reduce_ofst[] For each state, the offset into yy_action for
4613 ** shifting non-terminals after a reduce.
4614 ** yy_default[] Default action for each state.
4615 */
4616
drh8b582012003-10-21 13:16:03 +00004617 /* Output the yy_action table */
drh3a9d6c72017-12-25 04:15:38 +00004618 lemp->nactiontab = n = acttab_action_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004619 lemp->tablesize += n*szActionType;
drhf16371d2009-11-03 19:18:31 +00004620 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
4621 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004622 for(i=j=0; i<n; i++){
4623 int action = acttab_yyaction(pActtab, i);
drh5c8241b2017-12-24 23:38:10 +00004624 if( action<0 ) action = lemp->noAction;
drhfdbf9282003-10-21 16:34:41 +00004625 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004626 fprintf(out, " %4d,", action);
4627 if( j==9 || i==n-1 ){
4628 fprintf(out, "\n"); lineno++;
4629 j = 0;
4630 }else{
4631 j++;
4632 }
4633 }
4634 fprintf(out, "};\n"); lineno++;
4635
4636 /* Output the yy_lookahead table */
drh3a9d6c72017-12-25 04:15:38 +00004637 lemp->nlookaheadtab = n = acttab_lookahead_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004638 lemp->tablesize += n*szCodeType;
drh57196282004-10-06 15:41:16 +00004639 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004640 for(i=j=0; i<n; i++){
4641 int la = acttab_yylookahead(pActtab, i);
4642 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00004643 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004644 fprintf(out, " %4d,", la);
drh2e517162019-08-28 02:09:47 +00004645 if( j==9 ){
drh8b582012003-10-21 13:16:03 +00004646 fprintf(out, "\n"); lineno++;
4647 j = 0;
4648 }else{
4649 j++;
4650 }
4651 }
drh2e517162019-08-28 02:09:47 +00004652 /* Add extra entries to the end of the yy_lookahead[] table so that
4653 ** yy_shift_ofst[]+iToken will always be a valid index into the array,
4654 ** even for the largest possible value of yy_shift_ofst[] and iToken. */
4655 nLookAhead = lemp->nterminal + lemp->nactiontab;
4656 while( i<nLookAhead ){
4657 if( j==0 ) fprintf(out," /* %5d */ ", i);
4658 fprintf(out, " %4d,", lemp->nterminal);
4659 if( j==9 ){
4660 fprintf(out, "\n"); lineno++;
4661 j = 0;
4662 }else{
4663 j++;
4664 }
4665 i++;
4666 }
mistachkinacf6e082019-09-11 15:25:26 +00004667 if( j>0 ){ fprintf(out, "\n"); lineno++; }
drh8b582012003-10-21 13:16:03 +00004668 fprintf(out, "};\n"); lineno++;
4669
4670 /* Output the yy_shift_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004671 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004672 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drh701b6882016-08-10 13:30:43 +00004673 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
4674 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
4675 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004676 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh701b6882016-08-10 13:30:43 +00004677 minimum_size_type(mnTknOfst, lemp->nterminal+lemp->nactiontab, &sz));
4678 lineno++;
drhc75e0162015-09-07 02:23:02 +00004679 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004680 for(i=j=0; i<n; i++){
4681 int ofst;
4682 stp = lemp->sorted[i];
4683 ofst = stp->iTknOfst;
drh701b6882016-08-10 13:30:43 +00004684 if( ofst==NO_OFFSET ) ofst = lemp->nactiontab;
drhfdbf9282003-10-21 16:34:41 +00004685 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004686 fprintf(out, " %4d,", ofst);
4687 if( j==9 || i==n-1 ){
4688 fprintf(out, "\n"); lineno++;
4689 j = 0;
4690 }else{
4691 j++;
4692 }
4693 }
4694 fprintf(out, "};\n"); lineno++;
4695
4696 /* Output the yy_reduce_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004697 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004698 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00004699 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
4700 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
4701 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004702 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drhc75e0162015-09-07 02:23:02 +00004703 minimum_size_type(mnNtOfst-1, mxNtOfst, &sz)); lineno++;
4704 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004705 for(i=j=0; i<n; i++){
4706 int ofst;
4707 stp = lemp->sorted[i];
4708 ofst = stp->iNtOfst;
4709 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00004710 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004711 fprintf(out, " %4d,", ofst);
4712 if( j==9 || i==n-1 ){
4713 fprintf(out, "\n"); lineno++;
4714 j = 0;
4715 }else{
4716 j++;
4717 }
4718 }
4719 fprintf(out, "};\n"); lineno++;
4720
4721 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00004722 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004723 n = lemp->nxstate;
drhc75e0162015-09-07 02:23:02 +00004724 lemp->tablesize += n*szActionType;
drh8b582012003-10-21 13:16:03 +00004725 for(i=j=0; i<n; i++){
4726 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004727 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh5c8241b2017-12-24 23:38:10 +00004728 if( stp->iDfltReduce<0 ){
4729 fprintf(out, " %4d,", lemp->errAction);
4730 }else{
4731 fprintf(out, " %4d,", stp->iDfltReduce + lemp->minReduce);
4732 }
drh8b582012003-10-21 13:16:03 +00004733 if( j==9 || i==n-1 ){
4734 fprintf(out, "\n"); lineno++;
4735 j = 0;
4736 }else{
4737 j++;
4738 }
4739 }
4740 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004741 tplt_xfer(lemp->name,in,out,&lineno);
4742
drh0bd1f4e2002-06-06 18:54:39 +00004743 /* Generate the table of fallback tokens.
4744 */
4745 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00004746 int mx = lemp->nterminal - 1;
drh010bdb42019-08-28 11:31:11 +00004747 /* 2019-08-28: Generate fallback entries for every token to avoid
4748 ** having to do a range check on the index */
4749 /* while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; } */
drhc75e0162015-09-07 02:23:02 +00004750 lemp->tablesize += (mx+1)*szCodeType;
drh1441f3e2009-06-12 12:50:50 +00004751 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00004752 struct symbol *p = lemp->symbols[i];
4753 if( p->fallback==0 ){
4754 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
4755 }else{
4756 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
4757 p->name, p->fallback->name);
4758 }
4759 lineno++;
4760 }
4761 }
4762 tplt_xfer(lemp->name, in, out, &lineno);
4763
4764 /* Generate a table containing the symbolic name of every symbol
4765 */
drh75897232000-05-29 14:26:00 +00004766 for(i=0; i<lemp->nsymbol; i++){
drh3a9d6c72017-12-25 04:15:38 +00004767 fprintf(out," /* %4d */ \"%s\",\n",i, lemp->symbols[i]->name); lineno++;
drh75897232000-05-29 14:26:00 +00004768 }
drh75897232000-05-29 14:26:00 +00004769 tplt_xfer(lemp->name,in,out,&lineno);
4770
drh0bd1f4e2002-06-06 18:54:39 +00004771 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00004772 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00004773 ** when tracing REDUCE actions.
4774 */
4775 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drh4ef07702016-03-16 19:45:54 +00004776 assert( rp->iRule==i );
drhc4dd3fd2008-01-22 01:48:05 +00004777 fprintf(out," /* %3d */ \"", i);
4778 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00004779 fprintf(out,"\",\n"); lineno++;
4780 }
4781 tplt_xfer(lemp->name,in,out,&lineno);
4782
drh75897232000-05-29 14:26:00 +00004783 /* Generate code which executes every time a symbol is popped from
drh06f60d82017-04-14 19:46:12 +00004784 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00004785 ** (In other words, generate the %destructor actions)
4786 */
drh75897232000-05-29 14:26:00 +00004787 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00004788 int once = 1;
drh75897232000-05-29 14:26:00 +00004789 for(i=0; i<lemp->nsymbol; i++){
4790 struct symbol *sp = lemp->symbols[i];
4791 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004792 if( once ){
4793 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
4794 once = 0;
4795 }
drhc53eed12009-06-12 17:46:19 +00004796 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00004797 }
4798 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
4799 if( i<lemp->nsymbol ){
4800 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4801 fprintf(out," break;\n"); lineno++;
4802 }
4803 }
drh8d659732005-01-13 23:54:06 +00004804 if( lemp->vardest ){
4805 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00004806 int once = 1;
drh8d659732005-01-13 23:54:06 +00004807 for(i=0; i<lemp->nsymbol; i++){
4808 struct symbol *sp = lemp->symbols[i];
4809 if( sp==0 || sp->type==TERMINAL ||
4810 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004811 if( once ){
drh5c8241b2017-12-24 23:38:10 +00004812 fprintf(out, " /* Default NON-TERMINAL Destructor */\n");lineno++;
drh4dc8ef52008-07-01 17:13:57 +00004813 once = 0;
4814 }
drhc53eed12009-06-12 17:46:19 +00004815 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00004816 dflt_sp = sp;
4817 }
4818 if( dflt_sp!=0 ){
4819 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00004820 }
drh4dc8ef52008-07-01 17:13:57 +00004821 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00004822 }
drh75897232000-05-29 14:26:00 +00004823 for(i=0; i<lemp->nsymbol; i++){
4824 struct symbol *sp = lemp->symbols[i];
4825 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh0f832dd2016-08-16 16:46:40 +00004826 if( sp->destLineno<0 ) continue; /* Already emitted */
drh75013012009-06-12 15:47:34 +00004827 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004828
4829 /* Combine duplicate destructors into a single case */
4830 for(j=i+1; j<lemp->nsymbol; j++){
4831 struct symbol *sp2 = lemp->symbols[j];
4832 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
4833 && sp2->dtnum==sp->dtnum
4834 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00004835 fprintf(out," case %d: /* %s */\n",
4836 sp2->index, sp2->name); lineno++;
drh0f832dd2016-08-16 16:46:40 +00004837 sp2->destLineno = -1; /* Avoid emitting this destructor again */
drh0bb132b2004-07-20 14:06:51 +00004838 }
4839 }
4840
drh75897232000-05-29 14:26:00 +00004841 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4842 fprintf(out," break;\n"); lineno++;
4843 }
drh75897232000-05-29 14:26:00 +00004844 tplt_xfer(lemp->name,in,out,&lineno);
4845
4846 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00004847 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00004848 tplt_xfer(lemp->name,in,out,&lineno);
4849
drhcfc45b12018-12-03 23:57:27 +00004850 /* Generate the tables of rule information. yyRuleInfoLhs[] and
4851 ** yyRuleInfoNRhs[].
drh75897232000-05-29 14:26:00 +00004852 **
4853 ** Note: This code depends on the fact that rules are number
drhc2dbf352021-01-07 16:10:14 +00004854 ** sequentially beginning with 0.
drh75897232000-05-29 14:26:00 +00004855 */
drh5c8241b2017-12-24 23:38:10 +00004856 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drhcfc45b12018-12-03 23:57:27 +00004857 fprintf(out," %4d, /* (%d) ", rp->lhs->index, i);
4858 rule_print(out, rp);
4859 fprintf(out," */\n"); lineno++;
4860 }
4861 tplt_xfer(lemp->name,in,out,&lineno);
4862 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4863 fprintf(out," %3d, /* (%d) ", -rp->nrhs, i);
drh5c8241b2017-12-24 23:38:10 +00004864 rule_print(out, rp);
4865 fprintf(out," */\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004866 }
4867 tplt_xfer(lemp->name,in,out,&lineno);
4868
4869 /* Generate code which execution during each REDUCE action */
drhdabd04c2016-02-17 01:46:19 +00004870 i = 0;
drh75897232000-05-29 14:26:00 +00004871 for(rp=lemp->rule; rp; rp=rp->next){
drhdabd04c2016-02-17 01:46:19 +00004872 i += translate_code(lemp, rp);
4873 }
4874 if( i ){
4875 fprintf(out," YYMINORTYPE yylhsminor;\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004876 }
drhc53eed12009-06-12 17:46:19 +00004877 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00004878 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00004879 struct rule *rp2; /* Other rules with the same action */
drh711c9812016-05-23 14:24:31 +00004880 if( rp->codeEmitted ) continue;
4881 if( rp->noCode ){
4882 /* No C code actions, so this will be part of the "default:" rule */
drh2e55b042016-04-30 17:19:30 +00004883 continue;
4884 }
drh4ef07702016-03-16 19:45:54 +00004885 fprintf(out," case %d: /* ", rp->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004886 writeRuleText(out, rp);
4887 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004888 for(rp2=rp->next; rp2; rp2=rp2->next){
drhafb8cd92016-04-29 11:28:35 +00004889 if( rp2->code==rp->code && rp2->codePrefix==rp->codePrefix
4890 && rp2->codeSuffix==rp->codeSuffix ){
drh4ef07702016-03-16 19:45:54 +00004891 fprintf(out," case %d: /* ", rp2->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004892 writeRuleText(out, rp2);
drh4ef07702016-03-16 19:45:54 +00004893 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->iRule); lineno++;
drh711c9812016-05-23 14:24:31 +00004894 rp2->codeEmitted = 1;
drh0bb132b2004-07-20 14:06:51 +00004895 }
4896 }
drh75897232000-05-29 14:26:00 +00004897 emit_code(out,rp,lemp,&lineno);
4898 fprintf(out," break;\n"); lineno++;
drh711c9812016-05-23 14:24:31 +00004899 rp->codeEmitted = 1;
drh75897232000-05-29 14:26:00 +00004900 }
drhc53eed12009-06-12 17:46:19 +00004901 /* Finally, output the default: rule. We choose as the default: all
4902 ** empty actions. */
4903 fprintf(out," default:\n"); lineno++;
4904 for(rp=lemp->rule; rp; rp=rp->next){
drh711c9812016-05-23 14:24:31 +00004905 if( rp->codeEmitted ) continue;
4906 assert( rp->noCode );
drh4ef07702016-03-16 19:45:54 +00004907 fprintf(out," /* (%d) ", rp->iRule);
drhc53eed12009-06-12 17:46:19 +00004908 writeRuleText(out, rp);
drhe94006e2019-12-10 20:41:48 +00004909 if( rp->neverReduce ){
4910 fprintf(out, " (NEVER REDUCES) */ assert(yyruleno!=%d);\n",
4911 rp->iRule); lineno++;
4912 }else if( rp->doesReduce ){
drh756b41e2016-05-24 18:55:08 +00004913 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->iRule); lineno++;
4914 }else{
4915 fprintf(out, " (OPTIMIZED OUT) */ assert(yyruleno!=%d);\n",
4916 rp->iRule); lineno++;
4917 }
drhc53eed12009-06-12 17:46:19 +00004918 }
4919 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004920 tplt_xfer(lemp->name,in,out,&lineno);
4921
4922 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004923 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004924 tplt_xfer(lemp->name,in,out,&lineno);
4925
4926 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004927 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004928 tplt_xfer(lemp->name,in,out,&lineno);
4929
4930 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004931 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004932 tplt_xfer(lemp->name,in,out,&lineno);
4933
4934 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004935 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004936
drhe2dcc422019-01-15 14:44:23 +00004937 acttab_free(pActtab);
drh75897232000-05-29 14:26:00 +00004938 fclose(in);
4939 fclose(out);
drhfe03dac2019-11-26 02:22:39 +00004940 if( sql ) fclose(sql);
drh75897232000-05-29 14:26:00 +00004941 return;
4942}
4943
4944/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004945void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004946{
4947 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004948 const char *prefix;
drh75897232000-05-29 14:26:00 +00004949 char line[LINESIZE];
4950 char pattern[LINESIZE];
4951 int i;
4952
4953 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4954 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004955 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004956 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004957 int nextChar;
drh75897232000-05-29 14:26:00 +00004958 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
drh61f92cd2014-01-11 03:06:18 +00004959 lemon_sprintf(pattern,"#define %s%-30s %3d\n",
4960 prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004961 if( strcmp(line,pattern) ) break;
4962 }
drh8ba0d1c2012-06-16 15:26:31 +00004963 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004964 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004965 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004966 /* No change in the file. Don't rewrite it. */
4967 return;
4968 }
4969 }
drh2aa6ca42004-09-10 00:14:04 +00004970 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004971 if( out ){
4972 for(i=1; i<lemp->nterminal; i++){
drh61f92cd2014-01-11 03:06:18 +00004973 fprintf(out,"#define %s%-30s %3d\n",prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004974 }
drh06f60d82017-04-14 19:46:12 +00004975 fclose(out);
drh75897232000-05-29 14:26:00 +00004976 }
4977 return;
4978}
4979
4980/* Reduce the size of the action tables, if possible, by making use
4981** of defaults.
4982**
drhb59499c2002-02-23 18:45:13 +00004983** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004984** it the default. Except, there is no default if the wildcard token
4985** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004986*/
icculus9e44cf12010-02-14 17:14:22 +00004987void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004988{
4989 struct state *stp;
drhc173ad82016-05-23 16:15:02 +00004990 struct action *ap, *ap2, *nextap;
drhb59499c2002-02-23 18:45:13 +00004991 struct rule *rp, *rp2, *rbest;
drh0c6dfaa2015-09-08 21:16:46 +00004992 int nbest, n;
drh75897232000-05-29 14:26:00 +00004993 int i;
drhe09daa92006-06-10 13:29:31 +00004994 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004995
4996 for(i=0; i<lemp->nstate; i++){
4997 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004998 nbest = 0;
4999 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00005000 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00005001
drhb59499c2002-02-23 18:45:13 +00005002 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00005003 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
5004 usesWildcard = 1;
5005 }
drhb59499c2002-02-23 18:45:13 +00005006 if( ap->type!=REDUCE ) continue;
5007 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00005008 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00005009 if( rp==rbest ) continue;
5010 n = 1;
5011 for(ap2=ap->next; ap2; ap2=ap2->next){
5012 if( ap2->type!=REDUCE ) continue;
5013 rp2 = ap2->x.rp;
5014 if( rp2==rbest ) continue;
5015 if( rp2==rp ) n++;
5016 }
5017 if( n>nbest ){
5018 nbest = n;
5019 rbest = rp;
drh75897232000-05-29 14:26:00 +00005020 }
5021 }
drh06f60d82017-04-14 19:46:12 +00005022
drhb59499c2002-02-23 18:45:13 +00005023 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00005024 ** is not at least 1 or if the wildcard token is a possible
5025 ** lookahead.
5026 */
5027 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00005028
drhb59499c2002-02-23 18:45:13 +00005029
5030 /* Combine matching REDUCE actions into a single default */
5031 for(ap=stp->ap; ap; ap=ap->next){
5032 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
5033 }
drh75897232000-05-29 14:26:00 +00005034 assert( ap );
5035 ap->sp = Symbol_new("{default}");
5036 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00005037 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00005038 }
5039 stp->ap = Action_sort(stp->ap);
drh3bd48ab2015-09-07 18:23:37 +00005040
5041 for(ap=stp->ap; ap; ap=ap->next){
5042 if( ap->type==SHIFT ) break;
5043 if( ap->type==REDUCE && ap->x.rp!=rbest ) break;
5044 }
5045 if( ap==0 ){
5046 stp->autoReduce = 1;
5047 stp->pDfltReduce = rbest;
5048 }
5049 }
5050
5051 /* Make a second pass over all states and actions. Convert
5052 ** every action that is a SHIFT to an autoReduce state into
5053 ** a SHIFTREDUCE action.
5054 */
5055 for(i=0; i<lemp->nstate; i++){
5056 stp = lemp->sorted[i];
5057 for(ap=stp->ap; ap; ap=ap->next){
5058 struct state *pNextState;
5059 if( ap->type!=SHIFT ) continue;
5060 pNextState = ap->x.stp;
5061 if( pNextState->autoReduce && pNextState->pDfltReduce!=0 ){
5062 ap->type = SHIFTREDUCE;
5063 ap->x.rp = pNextState->pDfltReduce;
5064 }
5065 }
drh75897232000-05-29 14:26:00 +00005066 }
drhc173ad82016-05-23 16:15:02 +00005067
5068 /* If a SHIFTREDUCE action specifies a rule that has a single RHS term
5069 ** (meaning that the SHIFTREDUCE will land back in the state where it
5070 ** started) and if there is no C-code associated with the reduce action,
5071 ** then we can go ahead and convert the action to be the same as the
5072 ** action for the RHS of the rule.
5073 */
5074 for(i=0; i<lemp->nstate; i++){
5075 stp = lemp->sorted[i];
5076 for(ap=stp->ap; ap; ap=nextap){
5077 nextap = ap->next;
5078 if( ap->type!=SHIFTREDUCE ) continue;
5079 rp = ap->x.rp;
5080 if( rp->noCode==0 ) continue;
5081 if( rp->nrhs!=1 ) continue;
5082#if 1
5083 /* Only apply this optimization to non-terminals. It would be OK to
5084 ** apply it to terminal symbols too, but that makes the parser tables
5085 ** larger. */
5086 if( ap->sp->index<lemp->nterminal ) continue;
5087#endif
5088 /* If we reach this point, it means the optimization can be applied */
5089 nextap = ap;
5090 for(ap2=stp->ap; ap2 && (ap2==ap || ap2->sp!=rp->lhs); ap2=ap2->next){}
5091 assert( ap2!=0 );
5092 ap->spOpt = ap2->sp;
5093 ap->type = ap2->type;
5094 ap->x = ap2->x;
5095 }
5096 }
drh75897232000-05-29 14:26:00 +00005097}
drhb59499c2002-02-23 18:45:13 +00005098
drhada354d2005-11-05 15:03:59 +00005099
5100/*
5101** Compare two states for sorting purposes. The smaller state is the
5102** one with the most non-terminal actions. If they have the same number
5103** of non-terminal actions, then the smaller is the one with the most
5104** token actions.
5105*/
5106static int stateResortCompare(const void *a, const void *b){
5107 const struct state *pA = *(const struct state**)a;
5108 const struct state *pB = *(const struct state**)b;
5109 int n;
5110
5111 n = pB->nNtAct - pA->nNtAct;
5112 if( n==0 ){
5113 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00005114 if( n==0 ){
5115 n = pB->statenum - pA->statenum;
5116 }
drhada354d2005-11-05 15:03:59 +00005117 }
drhe594bc32009-11-03 13:02:25 +00005118 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00005119 return n;
5120}
5121
5122
5123/*
5124** Renumber and resort states so that states with fewer choices
5125** occur at the end. Except, keep state 0 as the first state.
5126*/
icculus9e44cf12010-02-14 17:14:22 +00005127void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00005128{
5129 int i;
5130 struct state *stp;
5131 struct action *ap;
5132
5133 for(i=0; i<lemp->nstate; i++){
5134 stp = lemp->sorted[i];
5135 stp->nTknAct = stp->nNtAct = 0;
drh5c8241b2017-12-24 23:38:10 +00005136 stp->iDfltReduce = -1; /* Init dflt action to "syntax error" */
drhada354d2005-11-05 15:03:59 +00005137 stp->iTknOfst = NO_OFFSET;
5138 stp->iNtOfst = NO_OFFSET;
5139 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00005140 int iAction = compute_action(lemp,ap);
5141 if( iAction>=0 ){
drhada354d2005-11-05 15:03:59 +00005142 if( ap->sp->index<lemp->nterminal ){
5143 stp->nTknAct++;
5144 }else if( ap->sp->index<lemp->nsymbol ){
5145 stp->nNtAct++;
5146 }else{
drh3bd48ab2015-09-07 18:23:37 +00005147 assert( stp->autoReduce==0 || stp->pDfltReduce==ap->x.rp );
drh5c8241b2017-12-24 23:38:10 +00005148 stp->iDfltReduce = iAction;
drhada354d2005-11-05 15:03:59 +00005149 }
5150 }
5151 }
5152 }
5153 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
5154 stateResortCompare);
5155 for(i=0; i<lemp->nstate; i++){
5156 lemp->sorted[i]->statenum = i;
5157 }
drh3bd48ab2015-09-07 18:23:37 +00005158 lemp->nxstate = lemp->nstate;
5159 while( lemp->nxstate>1 && lemp->sorted[lemp->nxstate-1]->autoReduce ){
5160 lemp->nxstate--;
5161 }
drhada354d2005-11-05 15:03:59 +00005162}
5163
5164
drh75897232000-05-29 14:26:00 +00005165/***************** From the file "set.c" ************************************/
5166/*
5167** Set manipulation routines for the LEMON parser generator.
5168*/
5169
5170static int size = 0;
5171
5172/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00005173void SetSize(int n)
drh75897232000-05-29 14:26:00 +00005174{
5175 size = n+1;
5176}
5177
5178/* Allocate a new set */
drh14d88552017-04-14 19:44:15 +00005179char *SetNew(void){
drh75897232000-05-29 14:26:00 +00005180 char *s;
drh9892c5d2007-12-21 00:02:11 +00005181 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00005182 if( s==0 ){
drh75897232000-05-29 14:26:00 +00005183 memory_error();
5184 }
drh75897232000-05-29 14:26:00 +00005185 return s;
5186}
5187
5188/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00005189void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00005190{
5191 free(s);
5192}
5193
5194/* Add a new element to the set. Return TRUE if the element was added
5195** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00005196int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00005197{
5198 int rv;
drh9892c5d2007-12-21 00:02:11 +00005199 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00005200 rv = s[e];
5201 s[e] = 1;
5202 return !rv;
5203}
5204
5205/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00005206int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00005207{
5208 int i, progress;
5209 progress = 0;
5210 for(i=0; i<size; i++){
5211 if( s2[i]==0 ) continue;
5212 if( s1[i]==0 ){
5213 progress = 1;
5214 s1[i] = 1;
5215 }
5216 }
5217 return progress;
5218}
5219/********************** From the file "table.c" ****************************/
5220/*
5221** All code in this file has been automatically generated
5222** from a specification in the file
5223** "table.q"
5224** by the associative array code building program "aagen".
5225** Do not edit this file! Instead, edit the specification
5226** file, then rerun aagen.
5227*/
5228/*
5229** Code for processing tables in the LEMON parser generator.
5230*/
5231
drh01f75f22013-10-02 20:46:30 +00005232PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00005233{
drh01f75f22013-10-02 20:46:30 +00005234 unsigned h = 0;
5235 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00005236 return h;
5237}
5238
5239/* Works like strdup, sort of. Save a string in malloced memory, but
5240** keep strings in a table so that the same string is not in more
5241** than one place.
5242*/
icculus9e44cf12010-02-14 17:14:22 +00005243const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00005244{
icculus9e44cf12010-02-14 17:14:22 +00005245 const char *z;
5246 char *cpy;
drh75897232000-05-29 14:26:00 +00005247
drh916f75f2006-07-17 00:19:39 +00005248 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00005249 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00005250 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
drh898799f2014-01-10 23:21:00 +00005251 lemon_strcpy(cpy,y);
icculus9e44cf12010-02-14 17:14:22 +00005252 z = cpy;
drh75897232000-05-29 14:26:00 +00005253 Strsafe_insert(z);
5254 }
5255 MemoryCheck(z);
5256 return z;
5257}
5258
5259/* There is one instance of the following structure for each
5260** associative array of type "x1".
5261*/
5262struct s_x1 {
5263 int size; /* The number of available slots. */
5264 /* Must be a power of 2 greater than or */
5265 /* equal to 1 */
5266 int count; /* Number of currently slots filled */
5267 struct s_x1node *tbl; /* The data stored here */
5268 struct s_x1node **ht; /* Hash table for lookups */
5269};
5270
5271/* There is one instance of this structure for every data element
5272** in an associative array of type "x1".
5273*/
5274typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00005275 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00005276 struct s_x1node *next; /* Next entry with the same hash */
5277 struct s_x1node **from; /* Previous link */
5278} x1node;
5279
5280/* There is only one instance of the array, which is the following */
5281static struct s_x1 *x1a;
5282
5283/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005284void Strsafe_init(void){
drh75897232000-05-29 14:26:00 +00005285 if( x1a ) return;
5286 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
5287 if( x1a ){
5288 x1a->size = 1024;
5289 x1a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005290 x1a->tbl = (x1node*)calloc(1024, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005291 if( x1a->tbl==0 ){
5292 free(x1a);
5293 x1a = 0;
5294 }else{
5295 int i;
5296 x1a->ht = (x1node**)&(x1a->tbl[1024]);
5297 for(i=0; i<1024; i++) x1a->ht[i] = 0;
5298 }
5299 }
5300}
5301/* Insert a new record into the array. Return TRUE if successful.
5302** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005303int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00005304{
5305 x1node *np;
drh01f75f22013-10-02 20:46:30 +00005306 unsigned h;
5307 unsigned ph;
drh75897232000-05-29 14:26:00 +00005308
5309 if( x1a==0 ) return 0;
5310 ph = strhash(data);
5311 h = ph & (x1a->size-1);
5312 np = x1a->ht[h];
5313 while( np ){
5314 if( strcmp(np->data,data)==0 ){
5315 /* An existing entry with the same key is found. */
5316 /* Fail because overwrite is not allows. */
5317 return 0;
5318 }
5319 np = np->next;
5320 }
5321 if( x1a->count>=x1a->size ){
5322 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005323 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005324 struct s_x1 array;
mistachkin8e189222015-04-19 21:43:16 +00005325 array.size = arrSize = x1a->size*2;
drh75897232000-05-29 14:26:00 +00005326 array.count = x1a->count;
mistachkin8e189222015-04-19 21:43:16 +00005327 array.tbl = (x1node*)calloc(arrSize, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005328 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005329 array.ht = (x1node**)&(array.tbl[arrSize]);
5330 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005331 for(i=0; i<x1a->count; i++){
5332 x1node *oldnp, *newnp;
5333 oldnp = &(x1a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005334 h = strhash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005335 newnp = &(array.tbl[i]);
5336 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5337 newnp->next = array.ht[h];
5338 newnp->data = oldnp->data;
5339 newnp->from = &(array.ht[h]);
5340 array.ht[h] = newnp;
5341 }
drh10761302021-10-04 16:14:51 +00005342 /* free(x1a->tbl); // This program was originally for 16-bit machines.
5343 ** Don't worry about freeing memory on modern platforms. */
drh75897232000-05-29 14:26:00 +00005344 *x1a = array;
5345 }
5346 /* Insert the new data */
5347 h = ph & (x1a->size-1);
5348 np = &(x1a->tbl[x1a->count++]);
5349 np->data = data;
5350 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
5351 np->next = x1a->ht[h];
5352 x1a->ht[h] = np;
5353 np->from = &(x1a->ht[h]);
5354 return 1;
5355}
5356
5357/* Return a pointer to data assigned to the given key. Return NULL
5358** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005359const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00005360{
drh01f75f22013-10-02 20:46:30 +00005361 unsigned h;
drh75897232000-05-29 14:26:00 +00005362 x1node *np;
5363
5364 if( x1a==0 ) return 0;
5365 h = strhash(key) & (x1a->size-1);
5366 np = x1a->ht[h];
5367 while( np ){
5368 if( strcmp(np->data,key)==0 ) break;
5369 np = np->next;
5370 }
5371 return np ? np->data : 0;
5372}
5373
5374/* Return a pointer to the (terminal or nonterminal) symbol "x".
5375** Create a new symbol if this is the first time "x" has been seen.
5376*/
icculus9e44cf12010-02-14 17:14:22 +00005377struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00005378{
5379 struct symbol *sp;
5380
5381 sp = Symbol_find(x);
5382 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00005383 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00005384 MemoryCheck(sp);
5385 sp->name = Strsafe(x);
drhc56fac72015-10-29 13:48:15 +00005386 sp->type = ISUPPER(*x) ? TERMINAL : NONTERMINAL;
drh75897232000-05-29 14:26:00 +00005387 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00005388 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00005389 sp->prec = -1;
5390 sp->assoc = UNK;
5391 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00005392 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00005393 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00005394 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00005395 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00005396 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00005397 Symbol_insert(sp,sp->name);
5398 }
drhc4dd3fd2008-01-22 01:48:05 +00005399 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00005400 return sp;
5401}
5402
drh61f92cd2014-01-11 03:06:18 +00005403/* Compare two symbols for sorting purposes. Return negative,
5404** zero, or positive if a is less then, equal to, or greater
5405** than b.
drh60d31652004-02-22 00:08:04 +00005406**
5407** Symbols that begin with upper case letters (terminals or tokens)
5408** must sort before symbols that begin with lower case letters
drh61f92cd2014-01-11 03:06:18 +00005409** (non-terminals). And MULTITERMINAL symbols (created using the
5410** %token_class directive) must sort at the very end. Other than
5411** that, the order does not matter.
drh60d31652004-02-22 00:08:04 +00005412**
5413** We find experimentally that leaving the symbols in their original
5414** order (the order they appeared in the grammar file) gives the
5415** smallest parser tables in SQLite.
5416*/
icculus9e44cf12010-02-14 17:14:22 +00005417int Symbolcmpp(const void *_a, const void *_b)
5418{
drh61f92cd2014-01-11 03:06:18 +00005419 const struct symbol *a = *(const struct symbol **) _a;
5420 const struct symbol *b = *(const struct symbol **) _b;
5421 int i1 = a->type==MULTITERMINAL ? 3 : a->name[0]>'Z' ? 2 : 1;
5422 int i2 = b->type==MULTITERMINAL ? 3 : b->name[0]>'Z' ? 2 : 1;
5423 return i1==i2 ? a->index - b->index : i1 - i2;
drh75897232000-05-29 14:26:00 +00005424}
5425
5426/* There is one instance of the following structure for each
5427** associative array of type "x2".
5428*/
5429struct s_x2 {
5430 int size; /* The number of available slots. */
5431 /* Must be a power of 2 greater than or */
5432 /* equal to 1 */
5433 int count; /* Number of currently slots filled */
5434 struct s_x2node *tbl; /* The data stored here */
5435 struct s_x2node **ht; /* Hash table for lookups */
5436};
5437
5438/* There is one instance of this structure for every data element
5439** in an associative array of type "x2".
5440*/
5441typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00005442 struct symbol *data; /* The data */
5443 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00005444 struct s_x2node *next; /* Next entry with the same hash */
5445 struct s_x2node **from; /* Previous link */
5446} x2node;
5447
5448/* There is only one instance of the array, which is the following */
5449static struct s_x2 *x2a;
5450
5451/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005452void Symbol_init(void){
drh75897232000-05-29 14:26:00 +00005453 if( x2a ) return;
5454 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
5455 if( x2a ){
5456 x2a->size = 128;
5457 x2a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005458 x2a->tbl = (x2node*)calloc(128, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005459 if( x2a->tbl==0 ){
5460 free(x2a);
5461 x2a = 0;
5462 }else{
5463 int i;
5464 x2a->ht = (x2node**)&(x2a->tbl[128]);
5465 for(i=0; i<128; i++) x2a->ht[i] = 0;
5466 }
5467 }
5468}
5469/* Insert a new record into the array. Return TRUE if successful.
5470** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005471int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00005472{
5473 x2node *np;
drh01f75f22013-10-02 20:46:30 +00005474 unsigned h;
5475 unsigned ph;
drh75897232000-05-29 14:26:00 +00005476
5477 if( x2a==0 ) return 0;
5478 ph = strhash(key);
5479 h = ph & (x2a->size-1);
5480 np = x2a->ht[h];
5481 while( np ){
5482 if( strcmp(np->key,key)==0 ){
5483 /* An existing entry with the same key is found. */
5484 /* Fail because overwrite is not allows. */
5485 return 0;
5486 }
5487 np = np->next;
5488 }
5489 if( x2a->count>=x2a->size ){
5490 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005491 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005492 struct s_x2 array;
mistachkin8e189222015-04-19 21:43:16 +00005493 array.size = arrSize = x2a->size*2;
drh75897232000-05-29 14:26:00 +00005494 array.count = x2a->count;
mistachkin8e189222015-04-19 21:43:16 +00005495 array.tbl = (x2node*)calloc(arrSize, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005496 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005497 array.ht = (x2node**)&(array.tbl[arrSize]);
5498 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005499 for(i=0; i<x2a->count; i++){
5500 x2node *oldnp, *newnp;
5501 oldnp = &(x2a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005502 h = strhash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005503 newnp = &(array.tbl[i]);
5504 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5505 newnp->next = array.ht[h];
5506 newnp->key = oldnp->key;
5507 newnp->data = oldnp->data;
5508 newnp->from = &(array.ht[h]);
5509 array.ht[h] = newnp;
5510 }
drh10761302021-10-04 16:14:51 +00005511 /* free(x2a->tbl); // This program was originally written for 16-bit
5512 ** machines. Don't worry about freeing this trivial amount of memory
5513 ** on modern platforms. Just leak it. */
drh75897232000-05-29 14:26:00 +00005514 *x2a = array;
5515 }
5516 /* Insert the new data */
5517 h = ph & (x2a->size-1);
5518 np = &(x2a->tbl[x2a->count++]);
5519 np->key = key;
5520 np->data = data;
5521 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
5522 np->next = x2a->ht[h];
5523 x2a->ht[h] = np;
5524 np->from = &(x2a->ht[h]);
5525 return 1;
5526}
5527
5528/* Return a pointer to data assigned to the given key. Return NULL
5529** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005530struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00005531{
drh01f75f22013-10-02 20:46:30 +00005532 unsigned h;
drh75897232000-05-29 14:26:00 +00005533 x2node *np;
5534
5535 if( x2a==0 ) return 0;
5536 h = strhash(key) & (x2a->size-1);
5537 np = x2a->ht[h];
5538 while( np ){
5539 if( strcmp(np->key,key)==0 ) break;
5540 np = np->next;
5541 }
5542 return np ? np->data : 0;
5543}
5544
5545/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00005546struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00005547{
5548 struct symbol *data;
5549 if( x2a && n>0 && n<=x2a->count ){
5550 data = x2a->tbl[n-1].data;
5551 }else{
5552 data = 0;
5553 }
5554 return data;
5555}
5556
5557/* Return the size of the array */
5558int Symbol_count()
5559{
5560 return x2a ? x2a->count : 0;
5561}
5562
5563/* Return an array of pointers to all data in the table.
5564** The array is obtained from malloc. Return NULL if memory allocation
5565** problems, or if the array is empty. */
5566struct symbol **Symbol_arrayof()
5567{
5568 struct symbol **array;
mistachkin8e189222015-04-19 21:43:16 +00005569 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005570 if( x2a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005571 arrSize = x2a->count;
5572 array = (struct symbol **)calloc(arrSize, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00005573 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005574 for(i=0; i<arrSize; i++) array[i] = x2a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005575 }
5576 return array;
5577}
5578
5579/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00005580int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00005581{
icculus9e44cf12010-02-14 17:14:22 +00005582 const struct config *a = (struct config *) _a;
5583 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00005584 int x;
5585 x = a->rp->index - b->rp->index;
5586 if( x==0 ) x = a->dot - b->dot;
5587 return x;
5588}
5589
5590/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00005591PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00005592{
5593 int rc;
5594 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
5595 rc = a->rp->index - b->rp->index;
5596 if( rc==0 ) rc = a->dot - b->dot;
5597 }
5598 if( rc==0 ){
5599 if( a ) rc = 1;
5600 if( b ) rc = -1;
5601 }
5602 return rc;
5603}
5604
5605/* Hash a state */
drh01f75f22013-10-02 20:46:30 +00005606PRIVATE unsigned statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00005607{
drh01f75f22013-10-02 20:46:30 +00005608 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005609 while( a ){
5610 h = h*571 + a->rp->index*37 + a->dot;
5611 a = a->bp;
5612 }
5613 return h;
5614}
5615
5616/* Allocate a new state structure */
5617struct state *State_new()
5618{
icculus9e44cf12010-02-14 17:14:22 +00005619 struct state *newstate;
5620 newstate = (struct state *)calloc(1, sizeof(struct state) );
5621 MemoryCheck(newstate);
5622 return newstate;
drh75897232000-05-29 14:26:00 +00005623}
5624
5625/* There is one instance of the following structure for each
5626** associative array of type "x3".
5627*/
5628struct s_x3 {
5629 int size; /* The number of available slots. */
5630 /* Must be a power of 2 greater than or */
5631 /* equal to 1 */
5632 int count; /* Number of currently slots filled */
5633 struct s_x3node *tbl; /* The data stored here */
5634 struct s_x3node **ht; /* Hash table for lookups */
5635};
5636
5637/* There is one instance of this structure for every data element
5638** in an associative array of type "x3".
5639*/
5640typedef struct s_x3node {
5641 struct state *data; /* The data */
5642 struct config *key; /* The key */
5643 struct s_x3node *next; /* Next entry with the same hash */
5644 struct s_x3node **from; /* Previous link */
5645} x3node;
5646
5647/* There is only one instance of the array, which is the following */
5648static struct s_x3 *x3a;
5649
5650/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005651void State_init(void){
drh75897232000-05-29 14:26:00 +00005652 if( x3a ) return;
5653 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
5654 if( x3a ){
5655 x3a->size = 128;
5656 x3a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005657 x3a->tbl = (x3node*)calloc(128, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005658 if( x3a->tbl==0 ){
5659 free(x3a);
5660 x3a = 0;
5661 }else{
5662 int i;
5663 x3a->ht = (x3node**)&(x3a->tbl[128]);
5664 for(i=0; i<128; i++) x3a->ht[i] = 0;
5665 }
5666 }
5667}
5668/* Insert a new record into the array. Return TRUE if successful.
5669** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005670int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00005671{
5672 x3node *np;
drh01f75f22013-10-02 20:46:30 +00005673 unsigned h;
5674 unsigned ph;
drh75897232000-05-29 14:26:00 +00005675
5676 if( x3a==0 ) return 0;
5677 ph = statehash(key);
5678 h = ph & (x3a->size-1);
5679 np = x3a->ht[h];
5680 while( np ){
5681 if( statecmp(np->key,key)==0 ){
5682 /* An existing entry with the same key is found. */
5683 /* Fail because overwrite is not allows. */
5684 return 0;
5685 }
5686 np = np->next;
5687 }
5688 if( x3a->count>=x3a->size ){
5689 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005690 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005691 struct s_x3 array;
mistachkin8e189222015-04-19 21:43:16 +00005692 array.size = arrSize = x3a->size*2;
drh75897232000-05-29 14:26:00 +00005693 array.count = x3a->count;
mistachkin8e189222015-04-19 21:43:16 +00005694 array.tbl = (x3node*)calloc(arrSize, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005695 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005696 array.ht = (x3node**)&(array.tbl[arrSize]);
5697 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005698 for(i=0; i<x3a->count; i++){
5699 x3node *oldnp, *newnp;
5700 oldnp = &(x3a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005701 h = statehash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005702 newnp = &(array.tbl[i]);
5703 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5704 newnp->next = array.ht[h];
5705 newnp->key = oldnp->key;
5706 newnp->data = oldnp->data;
5707 newnp->from = &(array.ht[h]);
5708 array.ht[h] = newnp;
5709 }
5710 free(x3a->tbl);
5711 *x3a = array;
5712 }
5713 /* Insert the new data */
5714 h = ph & (x3a->size-1);
5715 np = &(x3a->tbl[x3a->count++]);
5716 np->key = key;
5717 np->data = data;
5718 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
5719 np->next = x3a->ht[h];
5720 x3a->ht[h] = np;
5721 np->from = &(x3a->ht[h]);
5722 return 1;
5723}
5724
5725/* Return a pointer to data assigned to the given key. Return NULL
5726** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005727struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005728{
drh01f75f22013-10-02 20:46:30 +00005729 unsigned h;
drh75897232000-05-29 14:26:00 +00005730 x3node *np;
5731
5732 if( x3a==0 ) return 0;
5733 h = statehash(key) & (x3a->size-1);
5734 np = x3a->ht[h];
5735 while( np ){
5736 if( statecmp(np->key,key)==0 ) break;
5737 np = np->next;
5738 }
5739 return np ? np->data : 0;
5740}
5741
5742/* Return an array of pointers to all data in the table.
5743** The array is obtained from malloc. Return NULL if memory allocation
5744** problems, or if the array is empty. */
drh14d88552017-04-14 19:44:15 +00005745struct state **State_arrayof(void)
drh75897232000-05-29 14:26:00 +00005746{
5747 struct state **array;
mistachkin8e189222015-04-19 21:43:16 +00005748 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005749 if( x3a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005750 arrSize = x3a->count;
5751 array = (struct state **)calloc(arrSize, sizeof(struct state *));
drh75897232000-05-29 14:26:00 +00005752 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005753 for(i=0; i<arrSize; i++) array[i] = x3a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005754 }
5755 return array;
5756}
5757
5758/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00005759PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00005760{
drh01f75f22013-10-02 20:46:30 +00005761 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005762 h = h*571 + a->rp->index*37 + a->dot;
5763 return h;
5764}
5765
5766/* There is one instance of the following structure for each
5767** associative array of type "x4".
5768*/
5769struct s_x4 {
5770 int size; /* The number of available slots. */
5771 /* Must be a power of 2 greater than or */
5772 /* equal to 1 */
5773 int count; /* Number of currently slots filled */
5774 struct s_x4node *tbl; /* The data stored here */
5775 struct s_x4node **ht; /* Hash table for lookups */
5776};
5777
5778/* There is one instance of this structure for every data element
5779** in an associative array of type "x4".
5780*/
5781typedef struct s_x4node {
5782 struct config *data; /* The data */
5783 struct s_x4node *next; /* Next entry with the same hash */
5784 struct s_x4node **from; /* Previous link */
5785} x4node;
5786
5787/* There is only one instance of the array, which is the following */
5788static struct s_x4 *x4a;
5789
5790/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005791void Configtable_init(void){
drh75897232000-05-29 14:26:00 +00005792 if( x4a ) return;
5793 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
5794 if( x4a ){
5795 x4a->size = 64;
5796 x4a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005797 x4a->tbl = (x4node*)calloc(64, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005798 if( x4a->tbl==0 ){
5799 free(x4a);
5800 x4a = 0;
5801 }else{
5802 int i;
5803 x4a->ht = (x4node**)&(x4a->tbl[64]);
5804 for(i=0; i<64; i++) x4a->ht[i] = 0;
5805 }
5806 }
5807}
5808/* Insert a new record into the array. Return TRUE if successful.
5809** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005810int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00005811{
5812 x4node *np;
drh01f75f22013-10-02 20:46:30 +00005813 unsigned h;
5814 unsigned ph;
drh75897232000-05-29 14:26:00 +00005815
5816 if( x4a==0 ) return 0;
5817 ph = confighash(data);
5818 h = ph & (x4a->size-1);
5819 np = x4a->ht[h];
5820 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005821 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00005822 /* An existing entry with the same key is found. */
5823 /* Fail because overwrite is not allows. */
5824 return 0;
5825 }
5826 np = np->next;
5827 }
5828 if( x4a->count>=x4a->size ){
5829 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005830 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005831 struct s_x4 array;
mistachkin8e189222015-04-19 21:43:16 +00005832 array.size = arrSize = x4a->size*2;
drh75897232000-05-29 14:26:00 +00005833 array.count = x4a->count;
mistachkin8e189222015-04-19 21:43:16 +00005834 array.tbl = (x4node*)calloc(arrSize, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005835 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005836 array.ht = (x4node**)&(array.tbl[arrSize]);
5837 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005838 for(i=0; i<x4a->count; i++){
5839 x4node *oldnp, *newnp;
5840 oldnp = &(x4a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005841 h = confighash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005842 newnp = &(array.tbl[i]);
5843 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5844 newnp->next = array.ht[h];
5845 newnp->data = oldnp->data;
5846 newnp->from = &(array.ht[h]);
5847 array.ht[h] = newnp;
5848 }
drh10761302021-10-04 16:14:51 +00005849 /* free(x4a->tbl); // This code was originall written for 16-bit machines.
5850 ** on modern machines, don't worry about freeing this trival amount of
5851 ** memory. */
drh75897232000-05-29 14:26:00 +00005852 *x4a = array;
5853 }
5854 /* Insert the new data */
5855 h = ph & (x4a->size-1);
5856 np = &(x4a->tbl[x4a->count++]);
5857 np->data = data;
5858 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
5859 np->next = x4a->ht[h];
5860 x4a->ht[h] = np;
5861 np->from = &(x4a->ht[h]);
5862 return 1;
5863}
5864
5865/* Return a pointer to data assigned to the given key. Return NULL
5866** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005867struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005868{
5869 int h;
5870 x4node *np;
5871
5872 if( x4a==0 ) return 0;
5873 h = confighash(key) & (x4a->size-1);
5874 np = x4a->ht[h];
5875 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005876 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00005877 np = np->next;
5878 }
5879 return np ? np->data : 0;
5880}
5881
5882/* Remove all data from the table. Pass each data to the function "f"
5883** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00005884void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00005885{
5886 int i;
5887 if( x4a==0 || x4a->count==0 ) return;
5888 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
5889 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
5890 x4a->count = 0;
5891 return;
5892}