blob: 82531362e70eb0c1c1e568f446579ca5bf3a9343 [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 */
404 char *arg; /* Declaration of the 3th 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 }
920 }else{
drh4ef07702016-03-16 19:45:54 +0000921 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +0000922 }
923
924 /* Make sure the start symbol doesn't occur on the right-hand side of
925 ** any rule. Report an error if it does. (YACC would generate a new
926 ** start symbol in this case.) */
927 for(rp=lemp->rule; rp; rp=rp->next){
928 int i;
929 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000930 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000931 ErrorMsg(lemp->filename,0,
drh3ecc05b2019-12-12 00:20:40 +0000932 "The start symbol \"%s\" occurs on the "
933 "right-hand side of a rule. This will result in a parser which "
934 "does not work properly.",sp->name);
drh75897232000-05-29 14:26:00 +0000935 lemp->errorcnt++;
936 }
937 }
938 }
939
940 /* The basis configuration set for the first state
941 ** is all rules which have the start symbol as their
942 ** left-hand side */
943 for(rp=sp->rule; rp; rp=rp->nextlhs){
944 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000945 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000946 newcfp = Configlist_addbasis(rp,0);
947 SetAdd(newcfp->fws,0);
948 }
949
950 /* Compute the first state. All other states will be
951 ** computed automatically during the computation of the first one.
952 ** The returned pointer to the first state is not used. */
953 (void)getstate(lemp);
954 return;
955}
956
957/* Return a pointer to a state which is described by the configuration
958** list which has been built from calls to Configlist_add.
959*/
icculus9e44cf12010-02-14 17:14:22 +0000960PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
961PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000962{
963 struct config *cfp, *bp;
964 struct state *stp;
965
966 /* Extract the sorted basis of the new state. The basis was constructed
967 ** by prior calls to "Configlist_addbasis()". */
968 Configlist_sortbasis();
969 bp = Configlist_basis();
970
971 /* Get a state with the same basis */
972 stp = State_find(bp);
973 if( stp ){
974 /* A state with the same basis already exists! Copy all the follow-set
975 ** propagation links from the state under construction into the
976 ** preexisting state, then return a pointer to the preexisting state */
977 struct config *x, *y;
978 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
979 Plink_copy(&y->bplp,x->bplp);
980 Plink_delete(x->fplp);
981 x->fplp = x->bplp = 0;
982 }
983 cfp = Configlist_return();
984 Configlist_eat(cfp);
985 }else{
986 /* This really is a new state. Construct all the details */
987 Configlist_closure(lemp); /* Compute the configuration closure */
988 Configlist_sort(); /* Sort the configuration closure */
989 cfp = Configlist_return(); /* Get a pointer to the config list */
990 stp = State_new(); /* A new state structure */
991 MemoryCheck(stp);
992 stp->bp = bp; /* Remember the configuration basis */
993 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000994 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000995 stp->ap = 0; /* No actions, yet. */
996 State_insert(stp,stp->bp); /* Add to the state table */
997 buildshifts(lemp,stp); /* Recursively compute successor states */
998 }
999 return stp;
1000}
1001
drhfd405312005-11-06 04:06:59 +00001002/*
1003** Return true if two symbols are the same.
1004*/
icculus9e44cf12010-02-14 17:14:22 +00001005int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +00001006{
1007 int i;
1008 if( a==b ) return 1;
1009 if( a->type!=MULTITERMINAL ) return 0;
1010 if( b->type!=MULTITERMINAL ) return 0;
1011 if( a->nsubsym!=b->nsubsym ) return 0;
1012 for(i=0; i<a->nsubsym; i++){
1013 if( a->subsym[i]!=b->subsym[i] ) return 0;
1014 }
1015 return 1;
1016}
1017
drh75897232000-05-29 14:26:00 +00001018/* Construct all successor states to the given state. A "successor"
1019** state is any state which can be reached by a shift action.
1020*/
icculus9e44cf12010-02-14 17:14:22 +00001021PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +00001022{
1023 struct config *cfp; /* For looping thru the config closure of "stp" */
1024 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +00001025 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +00001026 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
1027 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
1028 struct state *newstp; /* A pointer to a successor state */
1029
1030 /* Each configuration becomes complete after it contibutes to a successor
1031 ** state. Initially, all configurations are incomplete */
1032 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
1033
1034 /* Loop through all configurations of the state "stp" */
1035 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1036 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
1037 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
1038 Configlist_reset(); /* Reset the new config set */
1039 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
1040
1041 /* For every configuration in the state "stp" which has the symbol "sp"
1042 ** following its dot, add the same configuration to the basis set under
1043 ** construction but with the dot shifted one symbol to the right. */
1044 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
1045 if( bcfp->status==COMPLETE ) continue; /* Already used */
1046 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
1047 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +00001048 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +00001049 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +00001050 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
1051 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +00001052 }
1053
1054 /* Get a pointer to the state described by the basis configuration set
1055 ** constructed in the preceding loop */
1056 newstp = getstate(lemp);
1057
1058 /* The state "newstp" is reached from the state "stp" by a shift action
1059 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +00001060 if( sp->type==MULTITERMINAL ){
1061 int i;
1062 for(i=0; i<sp->nsubsym; i++){
1063 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
1064 }
1065 }else{
1066 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
1067 }
drh75897232000-05-29 14:26:00 +00001068 }
1069}
1070
1071/*
1072** Construct the propagation links
1073*/
icculus9e44cf12010-02-14 17:14:22 +00001074void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001075{
1076 int i;
1077 struct config *cfp, *other;
1078 struct state *stp;
1079 struct plink *plp;
1080
1081 /* Housekeeping detail:
1082 ** Add to every propagate link a pointer back to the state to
1083 ** which the link is attached. */
1084 for(i=0; i<lemp->nstate; i++){
1085 stp = lemp->sorted[i];
1086 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1087 cfp->stp = stp;
1088 }
1089 }
1090
1091 /* Convert all backlinks into forward links. Only the forward
1092 ** links are used in the follow-set computation. */
1093 for(i=0; i<lemp->nstate; i++){
1094 stp = lemp->sorted[i];
1095 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1096 for(plp=cfp->bplp; plp; plp=plp->next){
1097 other = plp->cfp;
1098 Plink_add(&other->fplp,cfp);
1099 }
1100 }
1101 }
1102}
1103
1104/* Compute all followsets.
1105**
1106** A followset is the set of all symbols which can come immediately
1107** after a configuration.
1108*/
icculus9e44cf12010-02-14 17:14:22 +00001109void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001110{
1111 int i;
1112 struct config *cfp;
1113 struct plink *plp;
1114 int progress;
1115 int change;
1116
1117 for(i=0; i<lemp->nstate; i++){
1118 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1119 cfp->status = INCOMPLETE;
1120 }
1121 }
drh06f60d82017-04-14 19:46:12 +00001122
drh75897232000-05-29 14:26:00 +00001123 do{
1124 progress = 0;
1125 for(i=0; i<lemp->nstate; i++){
1126 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1127 if( cfp->status==COMPLETE ) continue;
1128 for(plp=cfp->fplp; plp; plp=plp->next){
1129 change = SetUnion(plp->cfp->fws,cfp->fws);
1130 if( change ){
1131 plp->cfp->status = INCOMPLETE;
1132 progress = 1;
drhf2f105d2012-08-20 15:53:54 +00001133 }
1134 }
drh75897232000-05-29 14:26:00 +00001135 cfp->status = COMPLETE;
1136 }
1137 }
1138 }while( progress );
1139}
1140
drh3cb2f6e2012-01-09 14:19:05 +00001141static int resolve_conflict(struct action *,struct action *);
drh75897232000-05-29 14:26:00 +00001142
1143/* Compute the reduce actions, and resolve conflicts.
1144*/
icculus9e44cf12010-02-14 17:14:22 +00001145void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001146{
1147 int i,j;
1148 struct config *cfp;
1149 struct state *stp;
1150 struct symbol *sp;
1151 struct rule *rp;
1152
drh06f60d82017-04-14 19:46:12 +00001153 /* Add all of the reduce actions
drh75897232000-05-29 14:26:00 +00001154 ** A reduce action is added for each element of the followset of
1155 ** a configuration which has its dot at the extreme right.
1156 */
1157 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
1158 stp = lemp->sorted[i];
1159 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
1160 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
1161 for(j=0; j<lemp->nterminal; j++){
1162 if( SetFind(cfp->fws,j) ){
1163 /* Add a reduce action to the state "stp" which will reduce by the
1164 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +00001165 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +00001166 }
drhf2f105d2012-08-20 15:53:54 +00001167 }
drh75897232000-05-29 14:26:00 +00001168 }
1169 }
1170 }
1171
1172 /* Add the accepting token */
1173 if( lemp->start ){
1174 sp = Symbol_find(lemp->start);
drh4ef07702016-03-16 19:45:54 +00001175 if( sp==0 ) sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +00001176 }else{
drh4ef07702016-03-16 19:45:54 +00001177 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +00001178 }
1179 /* Add to the first state (which is always the starting state of the
1180 ** finite state machine) an action to ACCEPT if the lookahead is the
1181 ** start nonterminal. */
1182 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1183
1184 /* Resolve conflicts */
1185 for(i=0; i<lemp->nstate; i++){
1186 struct action *ap, *nap;
drh75897232000-05-29 14:26:00 +00001187 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001188 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001189 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001190 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001191 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1192 /* The two actions "ap" and "nap" have the same lookahead.
1193 ** Figure out which one should be used */
drh3cb2f6e2012-01-09 14:19:05 +00001194 lemp->nconflict += resolve_conflict(ap,nap);
drh75897232000-05-29 14:26:00 +00001195 }
1196 }
1197 }
1198
1199 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001200 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001201 for(i=0; i<lemp->nstate; i++){
1202 struct action *ap;
1203 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001204 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001205 }
1206 }
1207 for(rp=lemp->rule; rp; rp=rp->next){
1208 if( rp->canReduce ) continue;
1209 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1210 lemp->errorcnt++;
1211 }
1212}
1213
1214/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001215** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001216**
1217** NO LONGER TRUE:
1218** To resolve a conflict, first look to see if either action
1219** is on an error rule. In that case, take the action which
1220** is not associated with the error rule. If neither or both
1221** actions are associated with an error rule, then try to
1222** use precedence to resolve the conflict.
1223**
1224** If either action is a SHIFT, then it must be apx. This
1225** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1226*/
icculus9e44cf12010-02-14 17:14:22 +00001227static int resolve_conflict(
1228 struct action *apx,
drh3cb2f6e2012-01-09 14:19:05 +00001229 struct action *apy
icculus9e44cf12010-02-14 17:14:22 +00001230){
drh75897232000-05-29 14:26:00 +00001231 struct symbol *spx, *spy;
1232 int errcnt = 0;
1233 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001234 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001235 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001236 errcnt++;
1237 }
drh75897232000-05-29 14:26:00 +00001238 if( apx->type==SHIFT && apy->type==REDUCE ){
1239 spx = apx->sp;
1240 spy = apy->x.rp->precsym;
1241 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1242 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001243 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001244 errcnt++;
drhdd7e9db2010-07-19 01:52:07 +00001245 }else if( spx->prec>spy->prec ){ /* higher precedence wins */
drh75897232000-05-29 14:26:00 +00001246 apy->type = RD_RESOLVED;
1247 }else if( spx->prec<spy->prec ){
1248 apx->type = SH_RESOLVED;
1249 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1250 apy->type = RD_RESOLVED; /* associativity */
1251 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1252 apx->type = SH_RESOLVED;
1253 }else{
1254 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh62a223e2014-06-09 13:11:40 +00001255 apx->type = ERROR;
drh75897232000-05-29 14:26:00 +00001256 }
1257 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1258 spx = apx->x.rp->precsym;
1259 spy = apy->x.rp->precsym;
1260 if( spx==0 || spy==0 || spx->prec<0 ||
1261 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001262 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001263 errcnt++;
1264 }else if( spx->prec>spy->prec ){
1265 apy->type = RD_RESOLVED;
1266 }else if( spx->prec<spy->prec ){
1267 apx->type = RD_RESOLVED;
1268 }
1269 }else{
drh06f60d82017-04-14 19:46:12 +00001270 assert(
drhb59499c2002-02-23 18:45:13 +00001271 apx->type==SH_RESOLVED ||
1272 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001273 apx->type==SSCONFLICT ||
1274 apx->type==SRCONFLICT ||
1275 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001276 apy->type==SH_RESOLVED ||
1277 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001278 apy->type==SSCONFLICT ||
1279 apy->type==SRCONFLICT ||
1280 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001281 );
1282 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1283 ** REDUCEs on the list. If we reach this point it must be because
1284 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001285 }
1286 return errcnt;
1287}
1288/********************* From the file "configlist.c" *************************/
1289/*
1290** Routines to processing a configuration list and building a state
1291** in the LEMON parser generator.
1292*/
1293
1294static struct config *freelist = 0; /* List of free configurations */
1295static struct config *current = 0; /* Top of list of configurations */
1296static struct config **currentend = 0; /* Last on list of configs */
1297static struct config *basis = 0; /* Top of list of basis configs */
1298static struct config **basisend = 0; /* End of list of basis configs */
1299
1300/* Return a pointer to a new configuration */
drh14d88552017-04-14 19:44:15 +00001301PRIVATE struct config *newconfig(void){
icculus9e44cf12010-02-14 17:14:22 +00001302 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001303 if( freelist==0 ){
1304 int i;
1305 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001306 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001307 if( freelist==0 ){
1308 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1309 exit(1);
1310 }
1311 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1312 freelist[amt-1].next = 0;
1313 }
icculus9e44cf12010-02-14 17:14:22 +00001314 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001315 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001316 return newcfg;
drh75897232000-05-29 14:26:00 +00001317}
1318
1319/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001320PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001321{
1322 old->next = freelist;
1323 freelist = old;
1324}
1325
1326/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001327void Configlist_init(void){
drh75897232000-05-29 14:26:00 +00001328 current = 0;
1329 currentend = &current;
1330 basis = 0;
1331 basisend = &basis;
1332 Configtable_init();
1333 return;
1334}
1335
1336/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001337void Configlist_reset(void){
drh75897232000-05-29 14:26:00 +00001338 current = 0;
1339 currentend = &current;
1340 basis = 0;
1341 basisend = &basis;
1342 Configtable_clear(0);
1343 return;
1344}
1345
1346/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001347struct config *Configlist_add(
1348 struct rule *rp, /* The rule */
1349 int dot /* Index into the RHS of the rule where the dot goes */
1350){
drh75897232000-05-29 14:26:00 +00001351 struct config *cfp, model;
1352
1353 assert( currentend!=0 );
1354 model.rp = rp;
1355 model.dot = dot;
1356 cfp = Configtable_find(&model);
1357 if( cfp==0 ){
1358 cfp = newconfig();
1359 cfp->rp = rp;
1360 cfp->dot = dot;
1361 cfp->fws = SetNew();
1362 cfp->stp = 0;
1363 cfp->fplp = cfp->bplp = 0;
1364 cfp->next = 0;
1365 cfp->bp = 0;
1366 *currentend = cfp;
1367 currentend = &cfp->next;
1368 Configtable_insert(cfp);
1369 }
1370 return cfp;
1371}
1372
1373/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001374struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001375{
1376 struct config *cfp, model;
1377
1378 assert( basisend!=0 );
1379 assert( currentend!=0 );
1380 model.rp = rp;
1381 model.dot = dot;
1382 cfp = Configtable_find(&model);
1383 if( cfp==0 ){
1384 cfp = newconfig();
1385 cfp->rp = rp;
1386 cfp->dot = dot;
1387 cfp->fws = SetNew();
1388 cfp->stp = 0;
1389 cfp->fplp = cfp->bplp = 0;
1390 cfp->next = 0;
1391 cfp->bp = 0;
1392 *currentend = cfp;
1393 currentend = &cfp->next;
1394 *basisend = cfp;
1395 basisend = &cfp->bp;
1396 Configtable_insert(cfp);
1397 }
1398 return cfp;
1399}
1400
1401/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001402void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001403{
1404 struct config *cfp, *newcfp;
1405 struct rule *rp, *newrp;
1406 struct symbol *sp, *xsp;
1407 int i, dot;
1408
1409 assert( currentend!=0 );
1410 for(cfp=current; cfp; cfp=cfp->next){
1411 rp = cfp->rp;
1412 dot = cfp->dot;
1413 if( dot>=rp->nrhs ) continue;
1414 sp = rp->rhs[dot];
1415 if( sp->type==NONTERMINAL ){
1416 if( sp->rule==0 && sp!=lemp->errsym ){
1417 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1418 sp->name);
1419 lemp->errorcnt++;
1420 }
1421 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1422 newcfp = Configlist_add(newrp,0);
1423 for(i=dot+1; i<rp->nrhs; i++){
1424 xsp = rp->rhs[i];
1425 if( xsp->type==TERMINAL ){
1426 SetAdd(newcfp->fws,xsp->index);
1427 break;
drhfd405312005-11-06 04:06:59 +00001428 }else if( xsp->type==MULTITERMINAL ){
1429 int k;
1430 for(k=0; k<xsp->nsubsym; k++){
1431 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1432 }
1433 break;
drhf2f105d2012-08-20 15:53:54 +00001434 }else{
drh75897232000-05-29 14:26:00 +00001435 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001436 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001437 }
1438 }
drh75897232000-05-29 14:26:00 +00001439 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1440 }
1441 }
1442 }
1443 return;
1444}
1445
1446/* Sort the configuration list */
drh14d88552017-04-14 19:44:15 +00001447void Configlist_sort(void){
drh25473362015-09-04 18:03:45 +00001448 current = (struct config*)msort((char*)current,(char**)&(current->next),
1449 Configcmp);
drh75897232000-05-29 14:26:00 +00001450 currentend = 0;
1451 return;
1452}
1453
1454/* Sort the basis configuration list */
drh14d88552017-04-14 19:44:15 +00001455void Configlist_sortbasis(void){
drh25473362015-09-04 18:03:45 +00001456 basis = (struct config*)msort((char*)current,(char**)&(current->bp),
1457 Configcmp);
drh75897232000-05-29 14:26:00 +00001458 basisend = 0;
1459 return;
1460}
1461
1462/* Return a pointer to the head of the configuration list and
1463** reset the list */
drh14d88552017-04-14 19:44:15 +00001464struct config *Configlist_return(void){
drh75897232000-05-29 14:26:00 +00001465 struct config *old;
1466 old = current;
1467 current = 0;
1468 currentend = 0;
1469 return old;
1470}
1471
1472/* Return a pointer to the head of the configuration list and
1473** reset the list */
drh14d88552017-04-14 19:44:15 +00001474struct config *Configlist_basis(void){
drh75897232000-05-29 14:26:00 +00001475 struct config *old;
1476 old = basis;
1477 basis = 0;
1478 basisend = 0;
1479 return old;
1480}
1481
1482/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001483void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001484{
1485 struct config *nextcfp;
1486 for(; cfp; cfp=nextcfp){
1487 nextcfp = cfp->next;
1488 assert( cfp->fplp==0 );
1489 assert( cfp->bplp==0 );
1490 if( cfp->fws ) SetFree(cfp->fws);
1491 deleteconfig(cfp);
1492 }
1493 return;
1494}
1495/***************** From the file "error.c" *********************************/
1496/*
1497** Code for printing error message.
1498*/
1499
drhf9a2e7b2003-04-15 01:49:48 +00001500void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001501 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001502 fprintf(stderr, "%s:%d: ", filename, lineno);
1503 va_start(ap, format);
1504 vfprintf(stderr,format,ap);
1505 va_end(ap);
1506 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001507}
1508/**************** From the file "main.c" ************************************/
1509/*
1510** Main program file for the LEMON parser generator.
1511*/
1512
1513/* Report an out-of-memory condition and abort. This function
1514** is used mostly by the "MemoryCheck" macro in struct.h
1515*/
drh14d88552017-04-14 19:44:15 +00001516void memory_error(void){
drh75897232000-05-29 14:26:00 +00001517 fprintf(stderr,"Out of memory. Aborting...\n");
1518 exit(1);
1519}
1520
drh6d08b4d2004-07-20 12:45:22 +00001521static int nDefine = 0; /* Number of -D options on the command line */
1522static char **azDefine = 0; /* Name of the -D macros */
1523
1524/* This routine is called with the argument to each -D command-line option.
1525** Add the macro defined to the azDefine array.
1526*/
1527static void handle_D_option(char *z){
1528 char **paz;
1529 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001530 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001531 if( azDefine==0 ){
1532 fprintf(stderr,"out of memory\n");
1533 exit(1);
1534 }
1535 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001536 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001537 if( *paz==0 ){
1538 fprintf(stderr,"out of memory\n");
1539 exit(1);
1540 }
drh898799f2014-01-10 23:21:00 +00001541 lemon_strcpy(*paz, z);
drh6d08b4d2004-07-20 12:45:22 +00001542 for(z=*paz; *z && *z!='='; z++){}
1543 *z = 0;
1544}
1545
drh9f88e6d2018-04-20 20:47:49 +00001546/* Rember the name of the output directory
1547*/
1548static char *outputDir = NULL;
1549static void handle_d_option(char *z){
1550 outputDir = (char *) malloc( lemonStrlen(z)+1 );
1551 if( outputDir==0 ){
1552 fprintf(stderr,"out of memory\n");
1553 exit(1);
1554 }
1555 lemon_strcpy(outputDir, z);
1556}
1557
icculus3e143bd2010-02-14 00:48:49 +00001558static char *user_templatename = NULL;
1559static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001560 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001561 if( user_templatename==0 ){
1562 memory_error();
1563 }
drh898799f2014-01-10 23:21:00 +00001564 lemon_strcpy(user_templatename, z);
icculus3e143bd2010-02-14 00:48:49 +00001565}
drh75897232000-05-29 14:26:00 +00001566
drh711c9812016-05-23 14:24:31 +00001567/* Merge together to lists of rules ordered by rule.iRule */
drh4ef07702016-03-16 19:45:54 +00001568static struct rule *Rule_merge(struct rule *pA, struct rule *pB){
1569 struct rule *pFirst = 0;
1570 struct rule **ppPrev = &pFirst;
1571 while( pA && pB ){
1572 if( pA->iRule<pB->iRule ){
1573 *ppPrev = pA;
1574 ppPrev = &pA->next;
1575 pA = pA->next;
1576 }else{
1577 *ppPrev = pB;
1578 ppPrev = &pB->next;
1579 pB = pB->next;
1580 }
1581 }
1582 if( pA ){
1583 *ppPrev = pA;
1584 }else{
1585 *ppPrev = pB;
1586 }
1587 return pFirst;
1588}
1589
1590/*
1591** Sort a list of rules in order of increasing iRule value
1592*/
1593static struct rule *Rule_sort(struct rule *rp){
drh18ef40e2020-09-16 16:55:56 +00001594 unsigned int i;
drh4ef07702016-03-16 19:45:54 +00001595 struct rule *pNext;
1596 struct rule *x[32];
1597 memset(x, 0, sizeof(x));
1598 while( rp ){
1599 pNext = rp->next;
1600 rp->next = 0;
1601 for(i=0; i<sizeof(x)/sizeof(x[0]) && x[i]; i++){
1602 rp = Rule_merge(x[i], rp);
1603 x[i] = 0;
1604 }
1605 x[i] = rp;
1606 rp = pNext;
1607 }
1608 rp = 0;
1609 for(i=0; i<sizeof(x)/sizeof(x[0]); i++){
1610 rp = Rule_merge(x[i], rp);
1611 }
1612 return rp;
1613}
1614
drhc75e0162015-09-07 02:23:02 +00001615/* forward reference */
1616static const char *minimum_size_type(int lwr, int upr, int *pnByte);
1617
1618/* Print a single line of the "Parser Stats" output
1619*/
1620static void stats_line(const char *zLabel, int iValue){
1621 int nLabel = lemonStrlen(zLabel);
1622 printf(" %s%.*s %5d\n", zLabel,
1623 35-nLabel, "................................",
1624 iValue);
1625}
1626
drh75897232000-05-29 14:26:00 +00001627/* The main program. Parse the command line and do it... */
drh18ef40e2020-09-16 16:55:56 +00001628int main(int argc, char **argv){
drh75897232000-05-29 14:26:00 +00001629 static int version = 0;
1630 static int rpflag = 0;
1631 static int basisflag = 0;
1632 static int compress = 0;
1633 static int quiet = 0;
1634 static int statistics = 0;
1635 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001636 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001637 static int noResort = 0;
drhfe03dac2019-11-26 02:22:39 +00001638 static int sqlFlag = 0;
drh0a34cf52020-07-03 15:41:08 +00001639 static int printPP = 0;
drh9f88e6d2018-04-20 20:47:49 +00001640
drh75897232000-05-29 14:26:00 +00001641 static struct s_options options[] = {
1642 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1643 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh9f88e6d2018-04-20 20:47:49 +00001644 {OPT_FSTR, "d", (char*)&handle_d_option, "Output directory. Default '.'"},
drh6d08b4d2004-07-20 12:45:22 +00001645 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh0a34cf52020-07-03 15:41:08 +00001646 {OPT_FLAG, "E", (char*)&printPP, "Print input file after preprocessing."},
drh0325d392015-01-01 19:11:22 +00001647 {OPT_FSTR, "f", 0, "Ignored. (Placeholder for -f compiler options.)"},
drh75897232000-05-29 14:26:00 +00001648 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
drh0325d392015-01-01 19:11:22 +00001649 {OPT_FSTR, "I", 0, "Ignored. (Placeholder for '-I' compiler options.)"},
shane58543932008-12-10 20:10:04 +00001650 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1651 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drh0325d392015-01-01 19:11:22 +00001652 {OPT_FSTR, "O", 0, "Ignored. (Placeholder for '-O' compiler options.)"},
drhf5c4e0f2010-07-18 11:35:53 +00001653 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1654 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001655 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001656 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001657 {OPT_FLAG, "s", (char*)&statistics,
1658 "Print parser stats to standard output."},
drhfe03dac2019-11-26 02:22:39 +00001659 {OPT_FLAG, "S", (char*)&sqlFlag,
1660 "Generate the *.sql file describing the parser tables."},
drh75897232000-05-29 14:26:00 +00001661 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
drh0325d392015-01-01 19:11:22 +00001662 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
1663 {OPT_FSTR, "W", 0, "Ignored. (Placeholder for '-W' compiler options.)"},
drh75897232000-05-29 14:26:00 +00001664 {OPT_FLAG,0,0,0}
1665 };
1666 int i;
icculus42585cf2010-02-14 05:19:56 +00001667 int exitcode;
drh75897232000-05-29 14:26:00 +00001668 struct lemon lem;
drh4ef07702016-03-16 19:45:54 +00001669 struct rule *rp;
drh75897232000-05-29 14:26:00 +00001670
drh18ef40e2020-09-16 16:55:56 +00001671 (void)argc;
drhb0c86772000-06-02 23:21:26 +00001672 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001673 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001674 printf("Lemon version 1.0\n");
drh06f60d82017-04-14 19:46:12 +00001675 exit(0);
drh75897232000-05-29 14:26:00 +00001676 }
drhb0c86772000-06-02 23:21:26 +00001677 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001678 fprintf(stderr,"Exactly one filename argument is required.\n");
1679 exit(1);
1680 }
drh954f6b42006-06-13 13:27:46 +00001681 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001682 lem.errorcnt = 0;
1683
1684 /* Initialize the machine */
1685 Strsafe_init();
1686 Symbol_init();
1687 State_init();
1688 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001689 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001690 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001691 lem.nolinenosflag = nolinenosflag;
drh0a34cf52020-07-03 15:41:08 +00001692 lem.printPreprocessed = printPP;
drh75897232000-05-29 14:26:00 +00001693 Symbol_new("$");
drh75897232000-05-29 14:26:00 +00001694
1695 /* Parse the input file */
1696 Parse(&lem);
drh0a34cf52020-07-03 15:41:08 +00001697 if( lem.printPreprocessed || lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001698 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001699 fprintf(stderr,"Empty grammar.\n");
1700 exit(1);
1701 }
drhed0c15b2018-04-16 14:31:34 +00001702 lem.errsym = Symbol_find("error");
drh75897232000-05-29 14:26:00 +00001703
1704 /* Count and index the symbols of the grammar */
drh75897232000-05-29 14:26:00 +00001705 Symbol_new("{default}");
drh61f92cd2014-01-11 03:06:18 +00001706 lem.nsymbol = Symbol_count();
drh75897232000-05-29 14:26:00 +00001707 lem.symbols = Symbol_arrayof();
drh61f92cd2014-01-11 03:06:18 +00001708 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1709 qsort(lem.symbols,lem.nsymbol,sizeof(struct symbol*), Symbolcmpp);
1710 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1711 while( lem.symbols[i-1]->type==MULTITERMINAL ){ i--; }
1712 assert( strcmp(lem.symbols[i-1]->name,"{default}")==0 );
1713 lem.nsymbol = i - 1;
drhc56fac72015-10-29 13:48:15 +00001714 for(i=1; ISUPPER(lem.symbols[i]->name[0]); i++);
drh75897232000-05-29 14:26:00 +00001715 lem.nterminal = i;
1716
drh711c9812016-05-23 14:24:31 +00001717 /* Assign sequential rule numbers. Start with 0. Put rules that have no
1718 ** reduce action C-code associated with them last, so that the switch()
1719 ** statement that selects reduction actions will have a smaller jump table.
1720 */
drh4ef07702016-03-16 19:45:54 +00001721 for(i=0, rp=lem.rule; rp; rp=rp->next){
1722 rp->iRule = rp->code ? i++ : -1;
1723 }
drhce678c22019-12-11 18:53:51 +00001724 lem.nruleWithAction = i;
drh4ef07702016-03-16 19:45:54 +00001725 for(rp=lem.rule; rp; rp=rp->next){
1726 if( rp->iRule<0 ) rp->iRule = i++;
1727 }
1728 lem.startRule = lem.rule;
1729 lem.rule = Rule_sort(lem.rule);
1730
drh75897232000-05-29 14:26:00 +00001731 /* Generate a reprint of the grammar, if requested on the command line */
1732 if( rpflag ){
1733 Reprint(&lem);
1734 }else{
1735 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001736 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001737
1738 /* Find the precedence for every production rule (that has one) */
1739 FindRulePrecedences(&lem);
1740
1741 /* Compute the lambda-nonterminals and the first-sets for every
1742 ** nonterminal */
1743 FindFirstSets(&lem);
1744
1745 /* Compute all LR(0) states. Also record follow-set propagation
1746 ** links so that the follow-set can be computed later */
1747 lem.nstate = 0;
1748 FindStates(&lem);
1749 lem.sorted = State_arrayof();
1750
1751 /* Tie up loose ends on the propagation links */
1752 FindLinks(&lem);
1753
1754 /* Compute the follow set of every reducible configuration */
1755 FindFollowSets(&lem);
1756
1757 /* Compute the action tables */
1758 FindActions(&lem);
1759
1760 /* Compress the action tables */
1761 if( compress==0 ) CompressTables(&lem);
1762
drhada354d2005-11-05 15:03:59 +00001763 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001764 ** occur at the end. This is an optimization that helps make the
1765 ** generated parser tables smaller. */
1766 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001767
drh75897232000-05-29 14:26:00 +00001768 /* Generate a report of the parser generated. (the "y.output" file) */
1769 if( !quiet ) ReportOutput(&lem);
1770
1771 /* Generate the source code for the parser */
drhfe03dac2019-11-26 02:22:39 +00001772 ReportTable(&lem, mhflag, sqlFlag);
drh75897232000-05-29 14:26:00 +00001773
1774 /* Produce a header file for use by the scanner. (This step is
1775 ** omitted if the "-m" option is used because makeheaders will
1776 ** generate the file for us.) */
1777 if( !mhflag ) ReportHeader(&lem);
1778 }
1779 if( statistics ){
drhc75e0162015-09-07 02:23:02 +00001780 printf("Parser statistics:\n");
1781 stats_line("terminal symbols", lem.nterminal);
1782 stats_line("non-terminal symbols", lem.nsymbol - lem.nterminal);
1783 stats_line("total symbols", lem.nsymbol);
1784 stats_line("rules", lem.nrule);
drh3bd48ab2015-09-07 18:23:37 +00001785 stats_line("states", lem.nxstate);
drhc75e0162015-09-07 02:23:02 +00001786 stats_line("conflicts", lem.nconflict);
1787 stats_line("action table entries", lem.nactiontab);
drh3a9d6c72017-12-25 04:15:38 +00001788 stats_line("lookahead table entries", lem.nlookaheadtab);
drhc75e0162015-09-07 02:23:02 +00001789 stats_line("total table size (bytes)", lem.tablesize);
drh75897232000-05-29 14:26:00 +00001790 }
icculus8e158022010-02-16 16:09:03 +00001791 if( lem.nconflict > 0 ){
1792 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001793 }
1794
1795 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001796 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001797 exit(exitcode);
1798 return (exitcode);
drh75897232000-05-29 14:26:00 +00001799}
1800/******************** From the file "msort.c" *******************************/
1801/*
1802** A generic merge-sort program.
1803**
1804** USAGE:
1805** Let "ptr" be a pointer to some structure which is at the head of
1806** a null-terminated list. Then to sort the list call:
1807**
1808** ptr = msort(ptr,&(ptr->next),cmpfnc);
1809**
1810** In the above, "cmpfnc" is a pointer to a function which compares
1811** two instances of the structure and returns an integer, as in
1812** strcmp. The second argument is a pointer to the pointer to the
1813** second element of the linked list. This address is used to compute
1814** the offset to the "next" field within the structure. The offset to
1815** the "next" field must be constant for all structures in the list.
1816**
1817** The function returns a new pointer which is the head of the list
1818** after sorting.
1819**
1820** ALGORITHM:
1821** Merge-sort.
1822*/
1823
1824/*
1825** Return a pointer to the next structure in the linked list.
1826*/
drhd25d6922012-04-18 09:59:56 +00001827#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001828
1829/*
1830** Inputs:
1831** a: A sorted, null-terminated linked list. (May be null).
1832** b: A sorted, null-terminated linked list. (May be null).
1833** cmp: A pointer to the comparison function.
1834** offset: Offset in the structure to the "next" field.
1835**
1836** Return Value:
1837** A pointer to the head of a sorted list containing the elements
1838** of both a and b.
1839**
1840** Side effects:
1841** The "next" pointers for elements in the lists a and b are
1842** changed.
1843*/
drhe9278182007-07-18 18:16:29 +00001844static char *merge(
1845 char *a,
1846 char *b,
1847 int (*cmp)(const char*,const char*),
1848 int offset
1849){
drh75897232000-05-29 14:26:00 +00001850 char *ptr, *head;
1851
1852 if( a==0 ){
1853 head = b;
1854 }else if( b==0 ){
1855 head = a;
1856 }else{
drhe594bc32009-11-03 13:02:25 +00001857 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001858 ptr = a;
1859 a = NEXT(a);
1860 }else{
1861 ptr = b;
1862 b = NEXT(b);
1863 }
1864 head = ptr;
1865 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001866 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001867 NEXT(ptr) = a;
1868 ptr = a;
1869 a = NEXT(a);
1870 }else{
1871 NEXT(ptr) = b;
1872 ptr = b;
1873 b = NEXT(b);
1874 }
1875 }
1876 if( a ) NEXT(ptr) = a;
1877 else NEXT(ptr) = b;
1878 }
1879 return head;
1880}
1881
1882/*
1883** Inputs:
1884** list: Pointer to a singly-linked list of structures.
1885** next: Pointer to pointer to the second element of the list.
1886** cmp: A comparison function.
1887**
1888** Return Value:
1889** A pointer to the head of a sorted list containing the elements
1890** orginally in list.
1891**
1892** Side effects:
1893** The "next" pointers for elements in list are changed.
1894*/
1895#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001896static char *msort(
1897 char *list,
1898 char **next,
1899 int (*cmp)(const char*,const char*)
1900){
drhba99af52001-10-25 20:37:16 +00001901 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001902 char *ep;
1903 char *set[LISTSIZE];
1904 int i;
drh1cc0d112015-03-31 15:15:48 +00001905 offset = (unsigned long)((char*)next - (char*)list);
drh75897232000-05-29 14:26:00 +00001906 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1907 while( list ){
1908 ep = list;
1909 list = NEXT(list);
1910 NEXT(ep) = 0;
1911 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1912 ep = merge(ep,set[i],cmp,offset);
1913 set[i] = 0;
1914 }
1915 set[i] = ep;
1916 }
1917 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001918 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001919 return ep;
1920}
1921/************************ From the file "option.c" **************************/
mistachkind9bc6e82019-05-10 16:16:19 +00001922static char **g_argv;
drh75897232000-05-29 14:26:00 +00001923static struct s_options *op;
1924static FILE *errstream;
1925
1926#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1927
1928/*
1929** Print the command line with a carrot pointing to the k-th character
1930** of the n-th field.
1931*/
icculus9e44cf12010-02-14 17:14:22 +00001932static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001933{
1934 int spcnt, i;
mistachkind9bc6e82019-05-10 16:16:19 +00001935 if( g_argv[0] ) fprintf(err,"%s",g_argv[0]);
1936 spcnt = lemonStrlen(g_argv[0]) + 1;
1937 for(i=1; i<n && g_argv[i]; i++){
1938 fprintf(err," %s",g_argv[i]);
1939 spcnt += lemonStrlen(g_argv[i])+1;
drh75897232000-05-29 14:26:00 +00001940 }
1941 spcnt += k;
mistachkind9bc6e82019-05-10 16:16:19 +00001942 for(; g_argv[i]; i++) fprintf(err," %s",g_argv[i]);
drh75897232000-05-29 14:26:00 +00001943 if( spcnt<20 ){
1944 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1945 }else{
1946 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1947 }
1948}
1949
1950/*
1951** Return the index of the N-th non-switch argument. Return -1
1952** if N is out of range.
1953*/
icculus9e44cf12010-02-14 17:14:22 +00001954static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001955{
1956 int i;
1957 int dashdash = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00001958 if( g_argv!=0 && *g_argv!=0 ){
1959 for(i=1; g_argv[i]; i++){
1960 if( dashdash || !ISOPT(g_argv[i]) ){
drh75897232000-05-29 14:26:00 +00001961 if( n==0 ) return i;
1962 n--;
1963 }
mistachkind9bc6e82019-05-10 16:16:19 +00001964 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00001965 }
1966 }
1967 return -1;
1968}
1969
1970static char emsg[] = "Command line syntax error: ";
1971
1972/*
1973** Process a flag command line argument.
1974*/
icculus9e44cf12010-02-14 17:14:22 +00001975static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001976{
1977 int v;
1978 int errcnt = 0;
1979 int j;
1980 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00001981 if( strncmp(&g_argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001982 }
mistachkind9bc6e82019-05-10 16:16:19 +00001983 v = g_argv[i][0]=='-' ? 1 : 0;
drh75897232000-05-29 14:26:00 +00001984 if( op[j].label==0 ){
1985 if( err ){
1986 fprintf(err,"%sundefined option.\n",emsg);
1987 errline(i,1,err);
1988 }
1989 errcnt++;
drh0325d392015-01-01 19:11:22 +00001990 }else if( op[j].arg==0 ){
1991 /* Ignore this option */
drh75897232000-05-29 14:26:00 +00001992 }else if( op[j].type==OPT_FLAG ){
1993 *((int*)op[j].arg) = v;
1994 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001995 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001996 }else if( op[j].type==OPT_FSTR ){
mistachkind9bc6e82019-05-10 16:16:19 +00001997 (*(void(*)(char *))(op[j].arg))(&g_argv[i][2]);
drh75897232000-05-29 14:26:00 +00001998 }else{
1999 if( err ){
2000 fprintf(err,"%smissing argument on switch.\n",emsg);
2001 errline(i,1,err);
2002 }
2003 errcnt++;
2004 }
2005 return errcnt;
2006}
2007
2008/*
2009** Process a command line switch which has an argument.
2010*/
icculus9e44cf12010-02-14 17:14:22 +00002011static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00002012{
2013 int lv = 0;
2014 double dv = 0.0;
2015 char *sv = 0, *end;
2016 char *cp;
2017 int j;
2018 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002019 cp = strchr(g_argv[i],'=');
drh43617e92006-03-06 20:55:46 +00002020 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00002021 *cp = 0;
2022 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00002023 if( strcmp(g_argv[i],op[j].label)==0 ) break;
drh75897232000-05-29 14:26:00 +00002024 }
2025 *cp = '=';
2026 if( op[j].label==0 ){
2027 if( err ){
2028 fprintf(err,"%sundefined option.\n",emsg);
2029 errline(i,0,err);
2030 }
2031 errcnt++;
2032 }else{
2033 cp++;
2034 switch( op[j].type ){
2035 case OPT_FLAG:
2036 case OPT_FFLAG:
2037 if( err ){
2038 fprintf(err,"%soption requires an argument.\n",emsg);
2039 errline(i,0,err);
2040 }
2041 errcnt++;
2042 break;
2043 case OPT_DBL:
2044 case OPT_FDBL:
2045 dv = strtod(cp,&end);
2046 if( *end ){
2047 if( err ){
drh25473362015-09-04 18:03:45 +00002048 fprintf(err,
2049 "%sillegal character in floating-point argument.\n",emsg);
mistachkind9bc6e82019-05-10 16:16:19 +00002050 errline(i,(int)((char*)end-(char*)g_argv[i]),err);
drh75897232000-05-29 14:26:00 +00002051 }
2052 errcnt++;
2053 }
2054 break;
2055 case OPT_INT:
2056 case OPT_FINT:
2057 lv = strtol(cp,&end,0);
2058 if( *end ){
2059 if( err ){
2060 fprintf(err,"%sillegal character in integer argument.\n",emsg);
mistachkind9bc6e82019-05-10 16:16:19 +00002061 errline(i,(int)((char*)end-(char*)g_argv[i]),err);
drh75897232000-05-29 14:26:00 +00002062 }
2063 errcnt++;
2064 }
2065 break;
2066 case OPT_STR:
2067 case OPT_FSTR:
2068 sv = cp;
2069 break;
2070 }
2071 switch( op[j].type ){
2072 case OPT_FLAG:
2073 case OPT_FFLAG:
2074 break;
2075 case OPT_DBL:
2076 *(double*)(op[j].arg) = dv;
2077 break;
2078 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00002079 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00002080 break;
2081 case OPT_INT:
2082 *(int*)(op[j].arg) = lv;
2083 break;
2084 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00002085 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00002086 break;
2087 case OPT_STR:
2088 *(char**)(op[j].arg) = sv;
2089 break;
2090 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00002091 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00002092 break;
2093 }
2094 }
2095 return errcnt;
2096}
2097
icculus9e44cf12010-02-14 17:14:22 +00002098int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00002099{
2100 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002101 g_argv = a;
drh75897232000-05-29 14:26:00 +00002102 op = o;
2103 errstream = err;
mistachkind9bc6e82019-05-10 16:16:19 +00002104 if( g_argv && *g_argv && op ){
drh75897232000-05-29 14:26:00 +00002105 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002106 for(i=1; g_argv[i]; i++){
2107 if( g_argv[i][0]=='+' || g_argv[i][0]=='-' ){
drh75897232000-05-29 14:26:00 +00002108 errcnt += handleflags(i,err);
mistachkind9bc6e82019-05-10 16:16:19 +00002109 }else if( strchr(g_argv[i],'=') ){
drh75897232000-05-29 14:26:00 +00002110 errcnt += handleswitch(i,err);
2111 }
2112 }
2113 }
2114 if( errcnt>0 ){
2115 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00002116 OptPrint();
drh75897232000-05-29 14:26:00 +00002117 exit(1);
2118 }
2119 return 0;
2120}
2121
drh14d88552017-04-14 19:44:15 +00002122int OptNArgs(void){
drh75897232000-05-29 14:26:00 +00002123 int cnt = 0;
2124 int dashdash = 0;
2125 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002126 if( g_argv!=0 && g_argv[0]!=0 ){
2127 for(i=1; g_argv[i]; i++){
2128 if( dashdash || !ISOPT(g_argv[i]) ) cnt++;
2129 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00002130 }
2131 }
2132 return cnt;
2133}
2134
icculus9e44cf12010-02-14 17:14:22 +00002135char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00002136{
2137 int i;
2138 i = argindex(n);
mistachkind9bc6e82019-05-10 16:16:19 +00002139 return i>=0 ? g_argv[i] : 0;
drh75897232000-05-29 14:26:00 +00002140}
2141
icculus9e44cf12010-02-14 17:14:22 +00002142void OptErr(int n)
drh75897232000-05-29 14:26:00 +00002143{
2144 int i;
2145 i = argindex(n);
2146 if( i>=0 ) errline(i,0,errstream);
2147}
2148
drh14d88552017-04-14 19:44:15 +00002149void OptPrint(void){
drh75897232000-05-29 14:26:00 +00002150 int i;
2151 int max, len;
2152 max = 0;
2153 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00002154 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00002155 switch( op[i].type ){
2156 case OPT_FLAG:
2157 case OPT_FFLAG:
2158 break;
2159 case OPT_INT:
2160 case OPT_FINT:
2161 len += 9; /* length of "<integer>" */
2162 break;
2163 case OPT_DBL:
2164 case OPT_FDBL:
2165 len += 6; /* length of "<real>" */
2166 break;
2167 case OPT_STR:
2168 case OPT_FSTR:
2169 len += 8; /* length of "<string>" */
2170 break;
2171 }
2172 if( len>max ) max = len;
2173 }
2174 for(i=0; op[i].label; i++){
2175 switch( op[i].type ){
2176 case OPT_FLAG:
2177 case OPT_FFLAG:
2178 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
2179 break;
2180 case OPT_INT:
2181 case OPT_FINT:
drh0325d392015-01-01 19:11:22 +00002182 fprintf(errstream," -%s<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002183 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002184 break;
2185 case OPT_DBL:
2186 case OPT_FDBL:
drh0325d392015-01-01 19:11:22 +00002187 fprintf(errstream," -%s<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002188 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002189 break;
2190 case OPT_STR:
2191 case OPT_FSTR:
drh0325d392015-01-01 19:11:22 +00002192 fprintf(errstream," -%s<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002193 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002194 break;
2195 }
2196 }
2197}
2198/*********************** From the file "parse.c" ****************************/
2199/*
2200** Input file parser for the LEMON parser generator.
2201*/
2202
2203/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00002204enum e_state {
2205 INITIALIZE,
2206 WAITING_FOR_DECL_OR_RULE,
2207 WAITING_FOR_DECL_KEYWORD,
2208 WAITING_FOR_DECL_ARG,
2209 WAITING_FOR_PRECEDENCE_SYMBOL,
2210 WAITING_FOR_ARROW,
2211 IN_RHS,
2212 LHS_ALIAS_1,
2213 LHS_ALIAS_2,
2214 LHS_ALIAS_3,
2215 RHS_ALIAS_1,
2216 RHS_ALIAS_2,
2217 PRECEDENCE_MARK_1,
2218 PRECEDENCE_MARK_2,
2219 RESYNC_AFTER_RULE_ERROR,
2220 RESYNC_AFTER_DECL_ERROR,
2221 WAITING_FOR_DESTRUCTOR_SYMBOL,
2222 WAITING_FOR_DATATYPE_SYMBOL,
2223 WAITING_FOR_FALLBACK_ID,
drh61f92cd2014-01-11 03:06:18 +00002224 WAITING_FOR_WILDCARD_ID,
2225 WAITING_FOR_CLASS_ID,
drh59c435a2017-08-02 03:21:11 +00002226 WAITING_FOR_CLASS_TOKEN,
2227 WAITING_FOR_TOKEN_NAME
icculus9e44cf12010-02-14 17:14:22 +00002228};
drh75897232000-05-29 14:26:00 +00002229struct pstate {
2230 char *filename; /* Name of the input file */
2231 int tokenlineno; /* Linenumber at which current token starts */
2232 int errorcnt; /* Number of errors so far */
2233 char *tokenstart; /* Text of current token */
2234 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00002235 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00002236 struct symbol *fallback; /* The fallback token */
drh61f92cd2014-01-11 03:06:18 +00002237 struct symbol *tkclass; /* Token class symbol */
drh75897232000-05-29 14:26:00 +00002238 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00002239 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00002240 int nrhs; /* Number of right-hand side symbols seen */
2241 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00002242 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00002243 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00002244 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00002245 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00002246 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00002247 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00002248 enum e_assoc declassoc; /* Assign this association to decl arguments */
2249 int preccounter; /* Assign this precedence to decl arguments */
2250 struct rule *firstrule; /* Pointer to first rule in the grammar */
2251 struct rule *lastrule; /* Pointer to the most recently parsed rule */
2252};
2253
2254/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00002255static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00002256{
icculus9e44cf12010-02-14 17:14:22 +00002257 const char *x;
drh75897232000-05-29 14:26:00 +00002258 x = Strsafe(psp->tokenstart); /* Save the token permanently */
2259#if 0
2260 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
2261 x,psp->state);
2262#endif
2263 switch( psp->state ){
2264 case INITIALIZE:
2265 psp->prevrule = 0;
2266 psp->preccounter = 0;
2267 psp->firstrule = psp->lastrule = 0;
2268 psp->gp->nrule = 0;
drh18ef40e2020-09-16 16:55:56 +00002269 /* fall through */
drh75897232000-05-29 14:26:00 +00002270 case WAITING_FOR_DECL_OR_RULE:
2271 if( x[0]=='%' ){
2272 psp->state = WAITING_FOR_DECL_KEYWORD;
drhc56fac72015-10-29 13:48:15 +00002273 }else if( ISLOWER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002274 psp->lhs = Symbol_new(x);
2275 psp->nrhs = 0;
2276 psp->lhsalias = 0;
2277 psp->state = WAITING_FOR_ARROW;
2278 }else if( x[0]=='{' ){
2279 if( psp->prevrule==0 ){
2280 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002281 "There is no prior rule upon which to attach the code "
2282 "fragment which begins on this line.");
drh75897232000-05-29 14:26:00 +00002283 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002284 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002285 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002286 "Code fragment beginning on this line is not the first "
2287 "to follow the previous rule.");
drh75897232000-05-29 14:26:00 +00002288 psp->errorcnt++;
drhe94006e2019-12-10 20:41:48 +00002289 }else if( strcmp(x, "{NEVER-REDUCE")==0 ){
2290 psp->prevrule->neverReduce = 1;
drh75897232000-05-29 14:26:00 +00002291 }else{
2292 psp->prevrule->line = psp->tokenlineno;
2293 psp->prevrule->code = &x[1];
drh711c9812016-05-23 14:24:31 +00002294 psp->prevrule->noCode = 0;
drhf2f105d2012-08-20 15:53:54 +00002295 }
drh75897232000-05-29 14:26:00 +00002296 }else if( x[0]=='[' ){
2297 psp->state = PRECEDENCE_MARK_1;
2298 }else{
2299 ErrorMsg(psp->filename,psp->tokenlineno,
2300 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2301 x);
2302 psp->errorcnt++;
2303 }
2304 break;
2305 case PRECEDENCE_MARK_1:
drhc56fac72015-10-29 13:48:15 +00002306 if( !ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002307 ErrorMsg(psp->filename,psp->tokenlineno,
2308 "The precedence symbol must be a terminal.");
2309 psp->errorcnt++;
2310 }else if( psp->prevrule==0 ){
2311 ErrorMsg(psp->filename,psp->tokenlineno,
2312 "There is no prior rule to assign precedence \"[%s]\".",x);
2313 psp->errorcnt++;
2314 }else if( psp->prevrule->precsym!=0 ){
2315 ErrorMsg(psp->filename,psp->tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00002316 "Precedence mark on this line is not the first "
2317 "to follow the previous rule.");
drh75897232000-05-29 14:26:00 +00002318 psp->errorcnt++;
2319 }else{
2320 psp->prevrule->precsym = Symbol_new(x);
2321 }
2322 psp->state = PRECEDENCE_MARK_2;
2323 break;
2324 case PRECEDENCE_MARK_2:
2325 if( x[0]!=']' ){
2326 ErrorMsg(psp->filename,psp->tokenlineno,
2327 "Missing \"]\" on precedence mark.");
2328 psp->errorcnt++;
2329 }
2330 psp->state = WAITING_FOR_DECL_OR_RULE;
2331 break;
2332 case WAITING_FOR_ARROW:
2333 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2334 psp->state = IN_RHS;
2335 }else if( x[0]=='(' ){
2336 psp->state = LHS_ALIAS_1;
2337 }else{
2338 ErrorMsg(psp->filename,psp->tokenlineno,
2339 "Expected to see a \":\" following the LHS symbol \"%s\".",
2340 psp->lhs->name);
2341 psp->errorcnt++;
2342 psp->state = RESYNC_AFTER_RULE_ERROR;
2343 }
2344 break;
2345 case LHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002346 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002347 psp->lhsalias = x;
2348 psp->state = LHS_ALIAS_2;
2349 }else{
2350 ErrorMsg(psp->filename,psp->tokenlineno,
2351 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2352 x,psp->lhs->name);
2353 psp->errorcnt++;
2354 psp->state = RESYNC_AFTER_RULE_ERROR;
2355 }
2356 break;
2357 case LHS_ALIAS_2:
2358 if( x[0]==')' ){
2359 psp->state = LHS_ALIAS_3;
2360 }else{
2361 ErrorMsg(psp->filename,psp->tokenlineno,
2362 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2363 psp->errorcnt++;
2364 psp->state = RESYNC_AFTER_RULE_ERROR;
2365 }
2366 break;
2367 case LHS_ALIAS_3:
2368 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2369 psp->state = IN_RHS;
2370 }else{
2371 ErrorMsg(psp->filename,psp->tokenlineno,
2372 "Missing \"->\" following: \"%s(%s)\".",
2373 psp->lhs->name,psp->lhsalias);
2374 psp->errorcnt++;
2375 psp->state = RESYNC_AFTER_RULE_ERROR;
2376 }
2377 break;
2378 case IN_RHS:
2379 if( x[0]=='.' ){
2380 struct rule *rp;
drh06f60d82017-04-14 19:46:12 +00002381 rp = (struct rule *)calloc( sizeof(struct rule) +
drh9892c5d2007-12-21 00:02:11 +00002382 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002383 if( rp==0 ){
2384 ErrorMsg(psp->filename,psp->tokenlineno,
2385 "Can't allocate enough memory for this rule.");
2386 psp->errorcnt++;
2387 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002388 }else{
drh75897232000-05-29 14:26:00 +00002389 int i;
2390 rp->ruleline = psp->tokenlineno;
2391 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002392 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002393 for(i=0; i<psp->nrhs; i++){
2394 rp->rhs[i] = psp->rhs[i];
2395 rp->rhsalias[i] = psp->alias[i];
drh539e7412018-04-21 20:24:19 +00002396 if( rp->rhsalias[i]!=0 ){ rp->rhs[i]->bContent = 1; }
drhf2f105d2012-08-20 15:53:54 +00002397 }
drh75897232000-05-29 14:26:00 +00002398 rp->lhs = psp->lhs;
2399 rp->lhsalias = psp->lhsalias;
2400 rp->nrhs = psp->nrhs;
2401 rp->code = 0;
drh711c9812016-05-23 14:24:31 +00002402 rp->noCode = 1;
drh75897232000-05-29 14:26:00 +00002403 rp->precsym = 0;
2404 rp->index = psp->gp->nrule++;
2405 rp->nextlhs = rp->lhs->rule;
2406 rp->lhs->rule = rp;
2407 rp->next = 0;
2408 if( psp->firstrule==0 ){
2409 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002410 }else{
drh75897232000-05-29 14:26:00 +00002411 psp->lastrule->next = rp;
2412 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002413 }
drh75897232000-05-29 14:26:00 +00002414 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002415 }
drh75897232000-05-29 14:26:00 +00002416 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002417 }else if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002418 if( psp->nrhs>=MAXRHS ){
2419 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002420 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002421 x);
2422 psp->errorcnt++;
2423 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002424 }else{
drh75897232000-05-29 14:26:00 +00002425 psp->rhs[psp->nrhs] = Symbol_new(x);
2426 psp->alias[psp->nrhs] = 0;
2427 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002428 }
drh2b75d3b2020-09-05 06:21:54 +00002429 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 && ISUPPER(x[1]) ){
drhfd405312005-11-06 04:06:59 +00002430 struct symbol *msp = psp->rhs[psp->nrhs-1];
2431 if( msp->type!=MULTITERMINAL ){
2432 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002433 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002434 memset(msp, 0, sizeof(*msp));
2435 msp->type = MULTITERMINAL;
2436 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002437 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002438 msp->subsym[0] = origsp;
2439 msp->name = origsp->name;
2440 psp->rhs[psp->nrhs-1] = msp;
2441 }
2442 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002443 msp->subsym = (struct symbol **) realloc(msp->subsym,
2444 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002445 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
drhc56fac72015-10-29 13:48:15 +00002446 if( ISLOWER(x[1]) || ISLOWER(msp->subsym[0]->name[0]) ){
drhfd405312005-11-06 04:06:59 +00002447 ErrorMsg(psp->filename,psp->tokenlineno,
2448 "Cannot form a compound containing a non-terminal");
2449 psp->errorcnt++;
2450 }
drh75897232000-05-29 14:26:00 +00002451 }else if( x[0]=='(' && psp->nrhs>0 ){
2452 psp->state = RHS_ALIAS_1;
2453 }else{
2454 ErrorMsg(psp->filename,psp->tokenlineno,
2455 "Illegal character on RHS of rule: \"%s\".",x);
2456 psp->errorcnt++;
2457 psp->state = RESYNC_AFTER_RULE_ERROR;
2458 }
2459 break;
2460 case RHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002461 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002462 psp->alias[psp->nrhs-1] = x;
2463 psp->state = RHS_ALIAS_2;
2464 }else{
2465 ErrorMsg(psp->filename,psp->tokenlineno,
2466 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2467 x,psp->rhs[psp->nrhs-1]->name);
2468 psp->errorcnt++;
2469 psp->state = RESYNC_AFTER_RULE_ERROR;
2470 }
2471 break;
2472 case RHS_ALIAS_2:
2473 if( x[0]==')' ){
2474 psp->state = IN_RHS;
2475 }else{
2476 ErrorMsg(psp->filename,psp->tokenlineno,
2477 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2478 psp->errorcnt++;
2479 psp->state = RESYNC_AFTER_RULE_ERROR;
2480 }
2481 break;
2482 case WAITING_FOR_DECL_KEYWORD:
drhc56fac72015-10-29 13:48:15 +00002483 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002484 psp->declkeyword = x;
2485 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002486 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002487 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002488 psp->state = WAITING_FOR_DECL_ARG;
2489 if( strcmp(x,"name")==0 ){
2490 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002491 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002492 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002493 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002494 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002495 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002496 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002497 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002498 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002499 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002500 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002501 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002502 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002503 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002504 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002505 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002506 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002507 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002508 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002509 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002510 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002511 }else if( strcmp(x,"extra_argument")==0 ){
2512 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002513 psp->insertLineMacro = 0;
drhfb32c442018-04-21 13:51:42 +00002514 }else if( strcmp(x,"extra_context")==0 ){
2515 psp->declargslot = &(psp->gp->ctx);
2516 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002517 }else if( strcmp(x,"token_type")==0 ){
2518 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002519 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002520 }else if( strcmp(x,"default_type")==0 ){
2521 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002522 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002523 }else if( strcmp(x,"stack_size")==0 ){
2524 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002525 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002526 }else if( strcmp(x,"start_symbol")==0 ){
2527 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002528 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002529 }else if( strcmp(x,"left")==0 ){
2530 psp->preccounter++;
2531 psp->declassoc = LEFT;
2532 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2533 }else if( strcmp(x,"right")==0 ){
2534 psp->preccounter++;
2535 psp->declassoc = RIGHT;
2536 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2537 }else if( strcmp(x,"nonassoc")==0 ){
2538 psp->preccounter++;
2539 psp->declassoc = NONE;
2540 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002541 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002542 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002543 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002544 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002545 }else if( strcmp(x,"fallback")==0 ){
2546 psp->fallback = 0;
2547 psp->state = WAITING_FOR_FALLBACK_ID;
drh59c435a2017-08-02 03:21:11 +00002548 }else if( strcmp(x,"token")==0 ){
2549 psp->state = WAITING_FOR_TOKEN_NAME;
drhe09daa92006-06-10 13:29:31 +00002550 }else if( strcmp(x,"wildcard")==0 ){
2551 psp->state = WAITING_FOR_WILDCARD_ID;
drh61f92cd2014-01-11 03:06:18 +00002552 }else if( strcmp(x,"token_class")==0 ){
2553 psp->state = WAITING_FOR_CLASS_ID;
drh75897232000-05-29 14:26:00 +00002554 }else{
2555 ErrorMsg(psp->filename,psp->tokenlineno,
2556 "Unknown declaration keyword: \"%%%s\".",x);
2557 psp->errorcnt++;
2558 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002559 }
drh75897232000-05-29 14:26:00 +00002560 }else{
2561 ErrorMsg(psp->filename,psp->tokenlineno,
2562 "Illegal declaration keyword: \"%s\".",x);
2563 psp->errorcnt++;
2564 psp->state = RESYNC_AFTER_DECL_ERROR;
2565 }
2566 break;
2567 case WAITING_FOR_DESTRUCTOR_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002568 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002569 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002570 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002571 psp->errorcnt++;
2572 psp->state = RESYNC_AFTER_DECL_ERROR;
2573 }else{
icculusd286fa62010-03-03 17:06:32 +00002574 struct symbol *sp = Symbol_new(x);
2575 psp->declargslot = &sp->destructor;
2576 psp->decllinenoslot = &sp->destLineno;
2577 psp->insertLineMacro = 1;
2578 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002579 }
2580 break;
2581 case WAITING_FOR_DATATYPE_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002582 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002583 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002584 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002585 psp->errorcnt++;
2586 psp->state = RESYNC_AFTER_DECL_ERROR;
2587 }else{
icculus866bf1e2010-02-17 20:31:32 +00002588 struct symbol *sp = Symbol_find(x);
2589 if((sp) && (sp->datatype)){
2590 ErrorMsg(psp->filename,psp->tokenlineno,
2591 "Symbol %%type \"%s\" already defined", x);
2592 psp->errorcnt++;
2593 psp->state = RESYNC_AFTER_DECL_ERROR;
2594 }else{
2595 if (!sp){
2596 sp = Symbol_new(x);
2597 }
2598 psp->declargslot = &sp->datatype;
2599 psp->insertLineMacro = 0;
2600 psp->state = WAITING_FOR_DECL_ARG;
2601 }
drh75897232000-05-29 14:26:00 +00002602 }
2603 break;
2604 case WAITING_FOR_PRECEDENCE_SYMBOL:
2605 if( x[0]=='.' ){
2606 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002607 }else if( ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002608 struct symbol *sp;
2609 sp = Symbol_new(x);
2610 if( sp->prec>=0 ){
2611 ErrorMsg(psp->filename,psp->tokenlineno,
2612 "Symbol \"%s\" has already be given a precedence.",x);
2613 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002614 }else{
drh75897232000-05-29 14:26:00 +00002615 sp->prec = psp->preccounter;
2616 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002617 }
drh75897232000-05-29 14:26:00 +00002618 }else{
2619 ErrorMsg(psp->filename,psp->tokenlineno,
2620 "Can't assign a precedence to \"%s\".",x);
2621 psp->errorcnt++;
2622 }
2623 break;
2624 case WAITING_FOR_DECL_ARG:
drhc56fac72015-10-29 13:48:15 +00002625 if( x[0]=='{' || x[0]=='\"' || ISALNUM(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002626 const char *zOld, *zNew;
2627 char *zBuf, *z;
mistachkin2318d332015-01-12 18:02:52 +00002628 int nOld, n, nLine = 0, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002629 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002630 char zLine[50];
2631 zNew = x;
2632 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002633 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002634 if( *psp->declargslot ){
2635 zOld = *psp->declargslot;
2636 }else{
2637 zOld = "";
2638 }
drh87cf1372008-08-13 20:09:06 +00002639 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002640 n = nOld + nNew + 20;
drh60c71b02020-09-01 11:20:03 +00002641 addLineMacro = !psp->gp->nolinenosflag
2642 && psp->insertLineMacro
2643 && psp->tokenlineno>1
2644 && (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
drhb5bd49e2008-07-14 12:21:08 +00002645 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002646 for(z=psp->filename, nBack=0; *z; z++){
2647 if( *z=='\\' ) nBack++;
2648 }
drh898799f2014-01-10 23:21:00 +00002649 lemon_sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002650 nLine = lemonStrlen(zLine);
2651 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002652 }
icculus9e44cf12010-02-14 17:14:22 +00002653 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2654 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002655 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002656 if( nOld && zBuf[-1]!='\n' ){
2657 *(zBuf++) = '\n';
2658 }
2659 memcpy(zBuf, zLine, nLine);
2660 zBuf += nLine;
2661 *(zBuf++) = '"';
2662 for(z=psp->filename; *z; z++){
2663 if( *z=='\\' ){
2664 *(zBuf++) = '\\';
2665 }
2666 *(zBuf++) = *z;
2667 }
2668 *(zBuf++) = '"';
2669 *(zBuf++) = '\n';
2670 }
drh4dc8ef52008-07-01 17:13:57 +00002671 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2672 psp->decllinenoslot[0] = psp->tokenlineno;
2673 }
drha5808f32008-04-27 22:19:44 +00002674 memcpy(zBuf, zNew, nNew);
2675 zBuf += nNew;
2676 *zBuf = 0;
2677 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002678 }else{
2679 ErrorMsg(psp->filename,psp->tokenlineno,
2680 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2681 psp->errorcnt++;
2682 psp->state = RESYNC_AFTER_DECL_ERROR;
2683 }
2684 break;
drh0bd1f4e2002-06-06 18:54:39 +00002685 case WAITING_FOR_FALLBACK_ID:
2686 if( x[0]=='.' ){
2687 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002688 }else if( !ISUPPER(x[0]) ){
drh0bd1f4e2002-06-06 18:54:39 +00002689 ErrorMsg(psp->filename, psp->tokenlineno,
2690 "%%fallback argument \"%s\" should be a token", x);
2691 psp->errorcnt++;
2692 }else{
2693 struct symbol *sp = Symbol_new(x);
2694 if( psp->fallback==0 ){
2695 psp->fallback = sp;
2696 }else if( sp->fallback ){
2697 ErrorMsg(psp->filename, psp->tokenlineno,
2698 "More than one fallback assigned to token %s", x);
2699 psp->errorcnt++;
2700 }else{
2701 sp->fallback = psp->fallback;
2702 psp->gp->has_fallback = 1;
2703 }
2704 }
2705 break;
drh59c435a2017-08-02 03:21:11 +00002706 case WAITING_FOR_TOKEN_NAME:
2707 /* Tokens do not have to be declared before use. But they can be
2708 ** in order to control their assigned integer number. The number for
2709 ** each token is assigned when it is first seen. So by including
2710 **
2711 ** %token ONE TWO THREE
2712 **
2713 ** early in the grammar file, that assigns small consecutive values
2714 ** to each of the tokens ONE TWO and THREE.
2715 */
2716 if( x[0]=='.' ){
2717 psp->state = WAITING_FOR_DECL_OR_RULE;
2718 }else if( !ISUPPER(x[0]) ){
2719 ErrorMsg(psp->filename, psp->tokenlineno,
2720 "%%token argument \"%s\" should be a token", x);
2721 psp->errorcnt++;
2722 }else{
2723 (void)Symbol_new(x);
2724 }
2725 break;
drhe09daa92006-06-10 13:29:31 +00002726 case WAITING_FOR_WILDCARD_ID:
2727 if( x[0]=='.' ){
2728 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002729 }else if( !ISUPPER(x[0]) ){
drhe09daa92006-06-10 13:29:31 +00002730 ErrorMsg(psp->filename, psp->tokenlineno,
2731 "%%wildcard argument \"%s\" should be a token", x);
2732 psp->errorcnt++;
2733 }else{
2734 struct symbol *sp = Symbol_new(x);
2735 if( psp->gp->wildcard==0 ){
2736 psp->gp->wildcard = sp;
2737 }else{
2738 ErrorMsg(psp->filename, psp->tokenlineno,
2739 "Extra wildcard to token: %s", x);
2740 psp->errorcnt++;
2741 }
2742 }
2743 break;
drh61f92cd2014-01-11 03:06:18 +00002744 case WAITING_FOR_CLASS_ID:
drhc56fac72015-10-29 13:48:15 +00002745 if( !ISLOWER(x[0]) ){
drh61f92cd2014-01-11 03:06:18 +00002746 ErrorMsg(psp->filename, psp->tokenlineno,
mistachkind9bc6e82019-05-10 16:16:19 +00002747 "%%token_class must be followed by an identifier: %s", x);
drh61f92cd2014-01-11 03:06:18 +00002748 psp->errorcnt++;
2749 psp->state = RESYNC_AFTER_DECL_ERROR;
2750 }else if( Symbol_find(x) ){
2751 ErrorMsg(psp->filename, psp->tokenlineno,
2752 "Symbol \"%s\" already used", x);
2753 psp->errorcnt++;
2754 psp->state = RESYNC_AFTER_DECL_ERROR;
2755 }else{
2756 psp->tkclass = Symbol_new(x);
2757 psp->tkclass->type = MULTITERMINAL;
2758 psp->state = WAITING_FOR_CLASS_TOKEN;
2759 }
2760 break;
2761 case WAITING_FOR_CLASS_TOKEN:
2762 if( x[0]=='.' ){
2763 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002764 }else if( ISUPPER(x[0]) || ((x[0]=='|' || x[0]=='/') && ISUPPER(x[1])) ){
drh61f92cd2014-01-11 03:06:18 +00002765 struct symbol *msp = psp->tkclass;
2766 msp->nsubsym++;
2767 msp->subsym = (struct symbol **) realloc(msp->subsym,
2768 sizeof(struct symbol*)*msp->nsubsym);
drhc56fac72015-10-29 13:48:15 +00002769 if( !ISUPPER(x[0]) ) x++;
drh61f92cd2014-01-11 03:06:18 +00002770 msp->subsym[msp->nsubsym-1] = Symbol_new(x);
2771 }else{
2772 ErrorMsg(psp->filename, psp->tokenlineno,
2773 "%%token_class argument \"%s\" should be a token", x);
2774 psp->errorcnt++;
2775 psp->state = RESYNC_AFTER_DECL_ERROR;
2776 }
2777 break;
drh75897232000-05-29 14:26:00 +00002778 case RESYNC_AFTER_RULE_ERROR:
2779/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2780** break; */
2781 case RESYNC_AFTER_DECL_ERROR:
2782 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2783 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2784 break;
2785 }
2786}
2787
drh0a34cf52020-07-03 15:41:08 +00002788/* The text in the input is part of the argument to an %ifdef or %ifndef.
2789** Evaluate the text as a boolean expression. Return true or false.
2790*/
2791static int eval_preprocessor_boolean(char *z, int lineno){
2792 int neg = 0;
2793 int res = 0;
2794 int okTerm = 1;
2795 int i;
2796 for(i=0; z[i]!=0; i++){
2797 if( ISSPACE(z[i]) ) continue;
2798 if( z[i]=='!' ){
2799 if( !okTerm ) goto pp_syntax_error;
2800 neg = !neg;
2801 continue;
2802 }
2803 if( z[i]=='|' && z[i+1]=='|' ){
2804 if( okTerm ) goto pp_syntax_error;
2805 if( res ) return 1;
2806 i++;
2807 okTerm = 1;
2808 continue;
2809 }
2810 if( z[i]=='&' && z[i+1]=='&' ){
2811 if( okTerm ) goto pp_syntax_error;
2812 if( !res ) return 0;
2813 i++;
2814 okTerm = 1;
2815 continue;
2816 }
2817 if( z[i]=='(' ){
2818 int k;
2819 int n = 1;
2820 if( !okTerm ) goto pp_syntax_error;
2821 for(k=i+1; z[k]; k++){
2822 if( z[k]==')' ){
2823 n--;
2824 if( n==0 ){
2825 z[k] = 0;
2826 res = eval_preprocessor_boolean(&z[i+1], -1);
2827 z[k] = ')';
2828 if( res<0 ){
2829 i = i-res;
2830 goto pp_syntax_error;
2831 }
2832 i = k;
2833 break;
2834 }
2835 }else if( z[k]=='(' ){
2836 n++;
2837 }else if( z[k]==0 ){
2838 i = k;
2839 goto pp_syntax_error;
2840 }
2841 }
2842 if( neg ){
2843 res = !res;
2844 neg = 0;
2845 }
2846 okTerm = 0;
2847 continue;
2848 }
2849 if( ISALPHA(z[i]) ){
2850 int j, k, n;
2851 if( !okTerm ) goto pp_syntax_error;
2852 for(k=i+1; ISALNUM(z[k]) || z[k]=='_'; k++){}
2853 n = k - i;
2854 res = 0;
2855 for(j=0; j<nDefine; j++){
2856 if( strncmp(azDefine[j],&z[i],n)==0 && azDefine[j][n]==0 ){
2857 res = 1;
2858 break;
2859 }
2860 }
2861 i = k-1;
2862 if( neg ){
2863 res = !res;
2864 neg = 0;
2865 }
2866 okTerm = 0;
2867 continue;
2868 }
2869 goto pp_syntax_error;
2870 }
2871 return res;
2872
2873pp_syntax_error:
2874 if( lineno>0 ){
2875 fprintf(stderr, "%%if syntax error on line %d.\n", lineno);
2876 fprintf(stderr, " %.*s <-- syntax error here\n", i+1, z);
2877 exit(1);
2878 }else{
2879 return -(i+1);
2880 }
2881}
2882
drh34ff57b2008-07-14 12:27:51 +00002883/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002884** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2885** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2886** comments them out. Text in between is also commented out as appropriate.
2887*/
danielk1977940fac92005-01-23 22:41:37 +00002888static void preprocess_input(char *z){
drh0a34cf52020-07-03 15:41:08 +00002889 int i, j, k;
drh6d08b4d2004-07-20 12:45:22 +00002890 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002891 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002892 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002893 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002894 for(i=0; z[i]; i++){
2895 if( z[i]=='\n' ) lineno++;
2896 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
drhc56fac72015-10-29 13:48:15 +00002897 if( strncmp(&z[i],"%endif",6)==0 && ISSPACE(z[i+6]) ){
drh6d08b4d2004-07-20 12:45:22 +00002898 if( exclude ){
2899 exclude--;
2900 if( exclude==0 ){
2901 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2902 }
2903 }
2904 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
drh0a34cf52020-07-03 15:41:08 +00002905 }else if( strncmp(&z[i],"%else",5)==0 && ISSPACE(z[i+5]) ){
2906 if( exclude==1){
2907 exclude = 0;
2908 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2909 }else if( exclude==0 ){
2910 exclude = 1;
2911 start = i;
2912 start_lineno = lineno;
2913 }
2914 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2915 }else if( strncmp(&z[i],"%ifdef ",7)==0
2916 || strncmp(&z[i],"%if ",4)==0
2917 || strncmp(&z[i],"%ifndef ",8)==0 ){
drh6d08b4d2004-07-20 12:45:22 +00002918 if( exclude ){
2919 exclude++;
2920 }else{
drh0a34cf52020-07-03 15:41:08 +00002921 int isNot;
2922 int iBool;
2923 for(j=i; z[j] && !ISSPACE(z[j]); j++){}
2924 iBool = j;
2925 isNot = (j==i+7);
2926 while( z[j] && z[j]!='\n' ){ j++; }
2927 k = z[j];
2928 z[j] = 0;
2929 exclude = eval_preprocessor_boolean(&z[iBool], lineno);
2930 z[j] = k;
2931 if( !isNot ) exclude = !exclude;
drh6d08b4d2004-07-20 12:45:22 +00002932 if( exclude ){
2933 start = i;
2934 start_lineno = lineno;
2935 }
2936 }
2937 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2938 }
2939 }
2940 if( exclude ){
2941 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2942 exit(1);
2943 }
2944}
2945
drh75897232000-05-29 14:26:00 +00002946/* In spite of its name, this function is really a scanner. It read
2947** in the entire input file (all at once) then tokenizes it. Each
2948** token is passed to the function "parseonetoken" which builds all
2949** the appropriate data structures in the global state vector "gp".
2950*/
icculus9e44cf12010-02-14 17:14:22 +00002951void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002952{
2953 struct pstate ps;
2954 FILE *fp;
2955 char *filebuf;
mistachkin2318d332015-01-12 18:02:52 +00002956 unsigned int filesize;
drh75897232000-05-29 14:26:00 +00002957 int lineno;
2958 int c;
2959 char *cp, *nextcp;
2960 int startline = 0;
2961
rse38514a92007-09-20 11:34:17 +00002962 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002963 ps.gp = gp;
2964 ps.filename = gp->filename;
2965 ps.errorcnt = 0;
2966 ps.state = INITIALIZE;
2967
2968 /* Begin by reading the input file */
2969 fp = fopen(ps.filename,"rb");
2970 if( fp==0 ){
2971 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2972 gp->errorcnt++;
2973 return;
2974 }
2975 fseek(fp,0,2);
2976 filesize = ftell(fp);
2977 rewind(fp);
2978 filebuf = (char *)malloc( filesize+1 );
drh03e1b1f2014-01-11 12:52:25 +00002979 if( filesize>100000000 || filebuf==0 ){
2980 ErrorMsg(ps.filename,0,"Input file too large.");
mistachkinc93c6142018-09-08 16:55:18 +00002981 free(filebuf);
drh75897232000-05-29 14:26:00 +00002982 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002983 fclose(fp);
drh75897232000-05-29 14:26:00 +00002984 return;
2985 }
2986 if( fread(filebuf,1,filesize,fp)!=filesize ){
2987 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2988 filesize);
2989 free(filebuf);
2990 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002991 fclose(fp);
drh75897232000-05-29 14:26:00 +00002992 return;
2993 }
2994 fclose(fp);
2995 filebuf[filesize] = 0;
2996
drh6d08b4d2004-07-20 12:45:22 +00002997 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2998 preprocess_input(filebuf);
drh0a34cf52020-07-03 15:41:08 +00002999 if( gp->printPreprocessed ){
3000 printf("%s\n", filebuf);
3001 return;
3002 }
drh6d08b4d2004-07-20 12:45:22 +00003003
drh75897232000-05-29 14:26:00 +00003004 /* Now scan the text of the input file */
3005 lineno = 1;
3006 for(cp=filebuf; (c= *cp)!=0; ){
3007 if( c=='\n' ) lineno++; /* Keep track of the line number */
drhc56fac72015-10-29 13:48:15 +00003008 if( ISSPACE(c) ){ cp++; continue; } /* Skip all white space */
drh75897232000-05-29 14:26:00 +00003009 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
3010 cp+=2;
3011 while( (c= *cp)!=0 && c!='\n' ) cp++;
3012 continue;
3013 }
3014 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
3015 cp+=2;
3016 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
3017 if( c=='\n' ) lineno++;
3018 cp++;
3019 }
3020 if( c ) cp++;
3021 continue;
3022 }
3023 ps.tokenstart = cp; /* Mark the beginning of the token */
3024 ps.tokenlineno = lineno; /* Linenumber on which token begins */
3025 if( c=='\"' ){ /* String literals */
3026 cp++;
3027 while( (c= *cp)!=0 && c!='\"' ){
3028 if( c=='\n' ) lineno++;
3029 cp++;
3030 }
3031 if( c==0 ){
3032 ErrorMsg(ps.filename,startline,
drh3ecc05b2019-12-12 00:20:40 +00003033 "String starting on this line is not terminated before "
3034 "the end of the file.");
drh75897232000-05-29 14:26:00 +00003035 ps.errorcnt++;
3036 nextcp = cp;
3037 }else{
3038 nextcp = cp+1;
3039 }
3040 }else if( c=='{' ){ /* A block of C code */
3041 int level;
3042 cp++;
3043 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
3044 if( c=='\n' ) lineno++;
3045 else if( c=='{' ) level++;
3046 else if( c=='}' ) level--;
3047 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
3048 int prevc;
3049 cp = &cp[2];
3050 prevc = 0;
3051 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
3052 if( c=='\n' ) lineno++;
3053 prevc = c;
3054 cp++;
drhf2f105d2012-08-20 15:53:54 +00003055 }
3056 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00003057 cp = &cp[2];
3058 while( (c= *cp)!=0 && c!='\n' ) cp++;
3059 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00003060 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00003061 int startchar, prevc;
3062 startchar = c;
3063 prevc = 0;
3064 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
3065 if( c=='\n' ) lineno++;
3066 if( prevc=='\\' ) prevc = 0;
3067 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00003068 }
3069 }
drh75897232000-05-29 14:26:00 +00003070 }
3071 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00003072 ErrorMsg(ps.filename,ps.tokenlineno,
drh3ecc05b2019-12-12 00:20:40 +00003073 "C code starting on this line is not terminated before "
3074 "the end of the file.");
drh75897232000-05-29 14:26:00 +00003075 ps.errorcnt++;
3076 nextcp = cp;
3077 }else{
3078 nextcp = cp+1;
3079 }
drhc56fac72015-10-29 13:48:15 +00003080 }else if( ISALNUM(c) ){ /* Identifiers */
3081 while( (c= *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drh75897232000-05-29 14:26:00 +00003082 nextcp = cp;
3083 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
3084 cp += 3;
3085 nextcp = cp;
drhc56fac72015-10-29 13:48:15 +00003086 }else if( (c=='/' || c=='|') && ISALPHA(cp[1]) ){
drhfd405312005-11-06 04:06:59 +00003087 cp += 2;
drhc56fac72015-10-29 13:48:15 +00003088 while( (c = *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drhfd405312005-11-06 04:06:59 +00003089 nextcp = cp;
drh75897232000-05-29 14:26:00 +00003090 }else{ /* All other (one character) operators */
3091 cp++;
3092 nextcp = cp;
3093 }
3094 c = *cp;
3095 *cp = 0; /* Null terminate the token */
3096 parseonetoken(&ps); /* Parse the token */
mistachkin2318d332015-01-12 18:02:52 +00003097 *cp = (char)c; /* Restore the buffer */
drh75897232000-05-29 14:26:00 +00003098 cp = nextcp;
3099 }
3100 free(filebuf); /* Release the buffer after parsing */
3101 gp->rule = ps.firstrule;
3102 gp->errorcnt = ps.errorcnt;
3103}
3104/*************************** From the file "plink.c" *********************/
3105/*
3106** Routines processing configuration follow-set propagation links
3107** in the LEMON parser generator.
3108*/
3109static struct plink *plink_freelist = 0;
3110
3111/* Allocate a new plink */
drh14d88552017-04-14 19:44:15 +00003112struct plink *Plink_new(void){
icculus9e44cf12010-02-14 17:14:22 +00003113 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00003114
3115 if( plink_freelist==0 ){
3116 int i;
3117 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00003118 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00003119 if( plink_freelist==0 ){
3120 fprintf(stderr,
3121 "Unable to allocate memory for a new follow-set propagation link.\n");
3122 exit(1);
3123 }
3124 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
3125 plink_freelist[amt-1].next = 0;
3126 }
icculus9e44cf12010-02-14 17:14:22 +00003127 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00003128 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00003129 return newlink;
drh75897232000-05-29 14:26:00 +00003130}
3131
3132/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00003133void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00003134{
icculus9e44cf12010-02-14 17:14:22 +00003135 struct plink *newlink;
3136 newlink = Plink_new();
3137 newlink->next = *plpp;
3138 *plpp = newlink;
3139 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00003140}
3141
3142/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00003143void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00003144{
3145 struct plink *nextpl;
3146 while( from ){
3147 nextpl = from->next;
3148 from->next = *to;
3149 *to = from;
3150 from = nextpl;
3151 }
3152}
3153
3154/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00003155void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00003156{
3157 struct plink *nextpl;
3158
3159 while( plp ){
3160 nextpl = plp->next;
3161 plp->next = plink_freelist;
3162 plink_freelist = plp;
3163 plp = nextpl;
3164 }
3165}
3166/*********************** From the file "report.c" **************************/
3167/*
3168** Procedures for generating reports and tables in the LEMON parser generator.
3169*/
3170
3171/* Generate a filename with the given suffix. Space to hold the
3172** name comes from malloc() and must be freed by the calling
3173** function.
3174*/
icculus9e44cf12010-02-14 17:14:22 +00003175PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00003176{
3177 char *name;
3178 char *cp;
drh9f88e6d2018-04-20 20:47:49 +00003179 char *filename = lemp->filename;
3180 int sz;
drh75897232000-05-29 14:26:00 +00003181
drh9f88e6d2018-04-20 20:47:49 +00003182 if( outputDir ){
3183 cp = strrchr(filename, '/');
3184 if( cp ) filename = cp + 1;
3185 }
3186 sz = lemonStrlen(filename);
3187 sz += lemonStrlen(suffix);
3188 if( outputDir ) sz += lemonStrlen(outputDir) + 1;
3189 sz += 5;
3190 name = (char*)malloc( sz );
drh75897232000-05-29 14:26:00 +00003191 if( name==0 ){
3192 fprintf(stderr,"Can't allocate space for a filename.\n");
3193 exit(1);
3194 }
drh9f88e6d2018-04-20 20:47:49 +00003195 name[0] = 0;
3196 if( outputDir ){
3197 lemon_strcpy(name, outputDir);
3198 lemon_strcat(name, "/");
3199 }
3200 lemon_strcat(name,filename);
drh75897232000-05-29 14:26:00 +00003201 cp = strrchr(name,'.');
3202 if( cp ) *cp = 0;
drh898799f2014-01-10 23:21:00 +00003203 lemon_strcat(name,suffix);
drh75897232000-05-29 14:26:00 +00003204 return name;
3205}
3206
3207/* Open a file with a name based on the name of the input file,
3208** but with a different (specified) suffix, and return a pointer
3209** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00003210PRIVATE FILE *file_open(
3211 struct lemon *lemp,
3212 const char *suffix,
3213 const char *mode
3214){
drh75897232000-05-29 14:26:00 +00003215 FILE *fp;
3216
3217 if( lemp->outname ) free(lemp->outname);
3218 lemp->outname = file_makename(lemp, suffix);
3219 fp = fopen(lemp->outname,mode);
3220 if( fp==0 && *mode=='w' ){
3221 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
3222 lemp->errorcnt++;
3223 return 0;
3224 }
3225 return fp;
3226}
3227
drh5c8241b2017-12-24 23:38:10 +00003228/* Print the text of a rule
3229*/
3230void rule_print(FILE *out, struct rule *rp){
3231 int i, j;
3232 fprintf(out, "%s",rp->lhs->name);
3233 /* if( rp->lhsalias ) fprintf(out,"(%s)",rp->lhsalias); */
3234 fprintf(out," ::=");
3235 for(i=0; i<rp->nrhs; i++){
3236 struct symbol *sp = rp->rhs[i];
3237 if( sp->type==MULTITERMINAL ){
3238 fprintf(out," %s", sp->subsym[0]->name);
3239 for(j=1; j<sp->nsubsym; j++){
3240 fprintf(out,"|%s", sp->subsym[j]->name);
3241 }
3242 }else{
3243 fprintf(out," %s", sp->name);
3244 }
3245 /* if( rp->rhsalias[i] ) fprintf(out,"(%s)",rp->rhsalias[i]); */
3246 }
3247}
3248
drh06f60d82017-04-14 19:46:12 +00003249/* Duplicate the input file without comments and without actions
drh75897232000-05-29 14:26:00 +00003250** on rules */
icculus9e44cf12010-02-14 17:14:22 +00003251void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003252{
3253 struct rule *rp;
3254 struct symbol *sp;
3255 int i, j, maxlen, len, ncolumns, skip;
3256 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
3257 maxlen = 10;
3258 for(i=0; i<lemp->nsymbol; i++){
3259 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00003260 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00003261 if( len>maxlen ) maxlen = len;
3262 }
3263 ncolumns = 76/(maxlen+5);
3264 if( ncolumns<1 ) ncolumns = 1;
3265 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
3266 for(i=0; i<skip; i++){
3267 printf("//");
3268 for(j=i; j<lemp->nsymbol; j+=skip){
3269 sp = lemp->symbols[j];
3270 assert( sp->index==j );
3271 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
3272 }
3273 printf("\n");
3274 }
3275 for(rp=lemp->rule; rp; rp=rp->next){
drh5c8241b2017-12-24 23:38:10 +00003276 rule_print(stdout, rp);
drh75897232000-05-29 14:26:00 +00003277 printf(".");
3278 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00003279 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00003280 printf("\n");
3281 }
3282}
3283
drh7e698e92015-09-07 14:22:24 +00003284/* Print a single rule.
3285*/
3286void RulePrint(FILE *fp, struct rule *rp, int iCursor){
drhfd405312005-11-06 04:06:59 +00003287 struct symbol *sp;
3288 int i, j;
drh75897232000-05-29 14:26:00 +00003289 fprintf(fp,"%s ::=",rp->lhs->name);
3290 for(i=0; i<=rp->nrhs; i++){
drh7e698e92015-09-07 14:22:24 +00003291 if( i==iCursor ) fprintf(fp," *");
drh75897232000-05-29 14:26:00 +00003292 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00003293 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00003294 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00003295 fprintf(fp," %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00003296 for(j=1; j<sp->nsubsym; j++){
3297 fprintf(fp,"|%s",sp->subsym[j]->name);
3298 }
drh61f92cd2014-01-11 03:06:18 +00003299 }else{
3300 fprintf(fp," %s", sp->name);
drhfd405312005-11-06 04:06:59 +00003301 }
drh75897232000-05-29 14:26:00 +00003302 }
3303}
3304
drh7e698e92015-09-07 14:22:24 +00003305/* Print the rule for a configuration.
3306*/
3307void ConfigPrint(FILE *fp, struct config *cfp){
3308 RulePrint(fp, cfp->rp, cfp->dot);
3309}
3310
drh75897232000-05-29 14:26:00 +00003311/* #define TEST */
drhfd405312005-11-06 04:06:59 +00003312#if 0
drh75897232000-05-29 14:26:00 +00003313/* Print a set */
3314PRIVATE void SetPrint(out,set,lemp)
3315FILE *out;
3316char *set;
3317struct lemon *lemp;
3318{
3319 int i;
3320 char *spacer;
3321 spacer = "";
3322 fprintf(out,"%12s[","");
3323 for(i=0; i<lemp->nterminal; i++){
3324 if( SetFind(set,i) ){
3325 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
3326 spacer = " ";
3327 }
3328 }
3329 fprintf(out,"]\n");
3330}
3331
3332/* Print a plink chain */
3333PRIVATE void PlinkPrint(out,plp,tag)
3334FILE *out;
3335struct plink *plp;
3336char *tag;
3337{
3338 while( plp ){
drhada354d2005-11-05 15:03:59 +00003339 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00003340 ConfigPrint(out,plp->cfp);
3341 fprintf(out,"\n");
3342 plp = plp->next;
3343 }
3344}
3345#endif
3346
3347/* Print an action to the given file descriptor. Return FALSE if
3348** nothing was actually printed.
3349*/
drh7e698e92015-09-07 14:22:24 +00003350int PrintAction(
3351 struct action *ap, /* The action to print */
3352 FILE *fp, /* Print the action here */
drh3bd48ab2015-09-07 18:23:37 +00003353 int indent /* Indent by this amount */
drh7e698e92015-09-07 14:22:24 +00003354){
drh75897232000-05-29 14:26:00 +00003355 int result = 1;
3356 switch( ap->type ){
drh7e698e92015-09-07 14:22:24 +00003357 case SHIFT: {
3358 struct state *stp = ap->x.stp;
drh3bd48ab2015-09-07 18:23:37 +00003359 fprintf(fp,"%*s shift %-7d",indent,ap->sp->name,stp->statenum);
drh75897232000-05-29 14:26:00 +00003360 break;
drh7e698e92015-09-07 14:22:24 +00003361 }
3362 case REDUCE: {
3363 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003364 fprintf(fp,"%*s reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003365 RulePrint(fp, rp, -1);
3366 break;
3367 }
3368 case SHIFTREDUCE: {
3369 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003370 fprintf(fp,"%*s shift-reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003371 RulePrint(fp, rp, -1);
drh75897232000-05-29 14:26:00 +00003372 break;
drh7e698e92015-09-07 14:22:24 +00003373 }
drh75897232000-05-29 14:26:00 +00003374 case ACCEPT:
3375 fprintf(fp,"%*s accept",indent,ap->sp->name);
3376 break;
3377 case ERROR:
3378 fprintf(fp,"%*s error",indent,ap->sp->name);
3379 break;
drh9892c5d2007-12-21 00:02:11 +00003380 case SRCONFLICT:
3381 case RRCONFLICT:
drh3bd48ab2015-09-07 18:23:37 +00003382 fprintf(fp,"%*s reduce %-7d ** Parsing conflict **",
drh4ef07702016-03-16 19:45:54 +00003383 indent,ap->sp->name,ap->x.rp->iRule);
drh75897232000-05-29 14:26:00 +00003384 break;
drh9892c5d2007-12-21 00:02:11 +00003385 case SSCONFLICT:
drh06f60d82017-04-14 19:46:12 +00003386 fprintf(fp,"%*s shift %-7d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00003387 indent,ap->sp->name,ap->x.stp->statenum);
3388 break;
drh75897232000-05-29 14:26:00 +00003389 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003390 if( showPrecedenceConflict ){
drh3bd48ab2015-09-07 18:23:37 +00003391 fprintf(fp,"%*s shift %-7d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003392 indent,ap->sp->name,ap->x.stp->statenum);
3393 }else{
3394 result = 0;
3395 }
3396 break;
drh75897232000-05-29 14:26:00 +00003397 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003398 if( showPrecedenceConflict ){
drh7e698e92015-09-07 14:22:24 +00003399 fprintf(fp,"%*s reduce %-7d -- dropped by precedence",
drh4ef07702016-03-16 19:45:54 +00003400 indent,ap->sp->name,ap->x.rp->iRule);
drhf5c4e0f2010-07-18 11:35:53 +00003401 }else{
3402 result = 0;
3403 }
3404 break;
drh75897232000-05-29 14:26:00 +00003405 case NOT_USED:
3406 result = 0;
3407 break;
3408 }
drhc173ad82016-05-23 16:15:02 +00003409 if( result && ap->spOpt ){
3410 fprintf(fp," /* because %s==%s */", ap->sp->name, ap->spOpt->name);
3411 }
drh75897232000-05-29 14:26:00 +00003412 return result;
3413}
3414
drh3bd48ab2015-09-07 18:23:37 +00003415/* Generate the "*.out" log file */
icculus9e44cf12010-02-14 17:14:22 +00003416void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003417{
drh539e7412018-04-21 20:24:19 +00003418 int i, n;
drh75897232000-05-29 14:26:00 +00003419 struct state *stp;
3420 struct config *cfp;
3421 struct action *ap;
drh12f68392018-04-06 19:12:55 +00003422 struct rule *rp;
drh75897232000-05-29 14:26:00 +00003423 FILE *fp;
3424
drh2aa6ca42004-09-10 00:14:04 +00003425 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00003426 if( fp==0 ) return;
drh3bd48ab2015-09-07 18:23:37 +00003427 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00003428 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00003429 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00003430 if( lemp->basisflag ) cfp=stp->bp;
3431 else cfp=stp->cfp;
3432 while( cfp ){
3433 char buf[20];
3434 if( cfp->dot==cfp->rp->nrhs ){
drh4ef07702016-03-16 19:45:54 +00003435 lemon_sprintf(buf,"(%d)",cfp->rp->iRule);
drh75897232000-05-29 14:26:00 +00003436 fprintf(fp," %5s ",buf);
3437 }else{
3438 fprintf(fp," ");
3439 }
3440 ConfigPrint(fp,cfp);
3441 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00003442#if 0
drh75897232000-05-29 14:26:00 +00003443 SetPrint(fp,cfp->fws,lemp);
3444 PlinkPrint(fp,cfp->fplp,"To ");
3445 PlinkPrint(fp,cfp->bplp,"From");
3446#endif
3447 if( lemp->basisflag ) cfp=cfp->bp;
3448 else cfp=cfp->next;
3449 }
3450 fprintf(fp,"\n");
3451 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00003452 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
drh75897232000-05-29 14:26:00 +00003453 }
3454 fprintf(fp,"\n");
3455 }
drhe9278182007-07-18 18:16:29 +00003456 fprintf(fp, "----------------------------------------------------\n");
3457 fprintf(fp, "Symbols:\n");
drh539e7412018-04-21 20:24:19 +00003458 fprintf(fp, "The first-set of non-terminals is shown after the name.\n\n");
drhe9278182007-07-18 18:16:29 +00003459 for(i=0; i<lemp->nsymbol; i++){
3460 int j;
3461 struct symbol *sp;
3462
3463 sp = lemp->symbols[i];
3464 fprintf(fp, " %3d: %s", i, sp->name);
3465 if( sp->type==NONTERMINAL ){
3466 fprintf(fp, ":");
3467 if( sp->lambda ){
3468 fprintf(fp, " <lambda>");
3469 }
3470 for(j=0; j<lemp->nterminal; j++){
3471 if( sp->firstset && SetFind(sp->firstset, j) ){
3472 fprintf(fp, " %s", lemp->symbols[j]->name);
3473 }
3474 }
3475 }
drh12f68392018-04-06 19:12:55 +00003476 if( sp->prec>=0 ) fprintf(fp," (precedence=%d)", sp->prec);
drhe9278182007-07-18 18:16:29 +00003477 fprintf(fp, "\n");
3478 }
drh12f68392018-04-06 19:12:55 +00003479 fprintf(fp, "----------------------------------------------------\n");
drh539e7412018-04-21 20:24:19 +00003480 fprintf(fp, "Syntax-only Symbols:\n");
3481 fprintf(fp, "The following symbols never carry semantic content.\n\n");
3482 for(i=n=0; i<lemp->nsymbol; i++){
3483 int w;
3484 struct symbol *sp = lemp->symbols[i];
3485 if( sp->bContent ) continue;
3486 w = (int)strlen(sp->name);
3487 if( n>0 && n+w>75 ){
3488 fprintf(fp,"\n");
3489 n = 0;
3490 }
3491 if( n>0 ){
3492 fprintf(fp, " ");
3493 n++;
3494 }
3495 fprintf(fp, "%s", sp->name);
3496 n += w;
3497 }
3498 if( n>0 ) fprintf(fp, "\n");
3499 fprintf(fp, "----------------------------------------------------\n");
drh12f68392018-04-06 19:12:55 +00003500 fprintf(fp, "Rules:\n");
3501 for(rp=lemp->rule; rp; rp=rp->next){
3502 fprintf(fp, "%4d: ", rp->iRule);
3503 rule_print(fp, rp);
3504 fprintf(fp,".");
3505 if( rp->precsym ){
3506 fprintf(fp," [%s precedence=%d]",
3507 rp->precsym->name, rp->precsym->prec);
3508 }
3509 fprintf(fp,"\n");
3510 }
drh75897232000-05-29 14:26:00 +00003511 fclose(fp);
3512 return;
3513}
3514
3515/* Search for the file "name" which is in the same directory as
3516** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00003517PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00003518{
icculus9e44cf12010-02-14 17:14:22 +00003519 const char *pathlist;
3520 char *pathbufptr;
3521 char *pathbuf;
drh75897232000-05-29 14:26:00 +00003522 char *path,*cp;
3523 char c;
drh75897232000-05-29 14:26:00 +00003524
3525#ifdef __WIN32__
3526 cp = strrchr(argv0,'\\');
3527#else
3528 cp = strrchr(argv0,'/');
3529#endif
3530 if( cp ){
3531 c = *cp;
3532 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003533 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh898799f2014-01-10 23:21:00 +00003534 if( path ) lemon_sprintf(path,"%s/%s",argv0,name);
drh75897232000-05-29 14:26:00 +00003535 *cp = c;
3536 }else{
drh75897232000-05-29 14:26:00 +00003537 pathlist = getenv("PATH");
3538 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003539 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003540 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003541 if( (pathbuf != 0) && (path!=0) ){
3542 pathbufptr = pathbuf;
drh898799f2014-01-10 23:21:00 +00003543 lemon_strcpy(pathbuf, pathlist);
icculus9e44cf12010-02-14 17:14:22 +00003544 while( *pathbuf ){
3545 cp = strchr(pathbuf,':');
3546 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003547 c = *cp;
3548 *cp = 0;
drh898799f2014-01-10 23:21:00 +00003549 lemon_sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003550 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003551 if( c==0 ) pathbuf[0] = 0;
3552 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003553 if( access(path,modemask)==0 ) break;
3554 }
icculus9e44cf12010-02-14 17:14:22 +00003555 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003556 }
3557 }
3558 return path;
3559}
3560
3561/* Given an action, compute the integer value for that action
3562** which is to be put in the action table of the generated machine.
3563** Return negative if no action should be generated.
3564*/
icculus9e44cf12010-02-14 17:14:22 +00003565PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003566{
3567 int act;
3568 switch( ap->type ){
drh3bd48ab2015-09-07 18:23:37 +00003569 case SHIFT: act = ap->x.stp->statenum; break;
drhbd8fcc12017-06-28 11:56:18 +00003570 case SHIFTREDUCE: {
drhbd8fcc12017-06-28 11:56:18 +00003571 /* Since a SHIFT is inherient after a prior REDUCE, convert any
3572 ** SHIFTREDUCE action with a nonterminal on the LHS into a simple
3573 ** REDUCE action: */
drh5c8241b2017-12-24 23:38:10 +00003574 if( ap->sp->index>=lemp->nterminal ){
3575 act = lemp->minReduce + ap->x.rp->iRule;
3576 }else{
3577 act = lemp->minShiftReduce + ap->x.rp->iRule;
3578 }
drhbd8fcc12017-06-28 11:56:18 +00003579 break;
3580 }
drh5c8241b2017-12-24 23:38:10 +00003581 case REDUCE: act = lemp->minReduce + ap->x.rp->iRule; break;
3582 case ERROR: act = lemp->errAction; break;
3583 case ACCEPT: act = lemp->accAction; break;
drh75897232000-05-29 14:26:00 +00003584 default: act = -1; break;
3585 }
3586 return act;
3587}
3588
3589#define LINESIZE 1000
3590/* The next cluster of routines are for reading the template file
3591** and writing the results to the generated parser */
3592/* The first function transfers data from "in" to "out" until
3593** a line is seen which begins with "%%". The line number is
3594** tracked.
3595**
3596** if name!=0, then any word that begin with "Parse" is changed to
3597** begin with *name instead.
3598*/
icculus9e44cf12010-02-14 17:14:22 +00003599PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003600{
3601 int i, iStart;
3602 char line[LINESIZE];
3603 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3604 (*lineno)++;
3605 iStart = 0;
3606 if( name ){
3607 for(i=0; line[i]; i++){
3608 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
drhc56fac72015-10-29 13:48:15 +00003609 && (i==0 || !ISALPHA(line[i-1]))
drh75897232000-05-29 14:26:00 +00003610 ){
3611 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3612 fprintf(out,"%s",name);
3613 i += 4;
3614 iStart = i+1;
3615 }
3616 }
3617 }
3618 fprintf(out,"%s",&line[iStart]);
3619 }
3620}
3621
drh60c71b02020-09-01 11:20:03 +00003622/* Skip forward past the header of the template file to the first "%%"
3623*/
3624PRIVATE void tplt_skip_header(FILE *in, int *lineno)
3625{
3626 char line[LINESIZE];
3627 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3628 (*lineno)++;
3629 }
3630}
3631
drh75897232000-05-29 14:26:00 +00003632/* The next function finds the template file and opens it, returning
3633** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003634PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003635{
3636 static char templatename[] = "lempar.c";
3637 char buf[1000];
3638 FILE *in;
3639 char *tpltname;
3640 char *cp;
3641
icculus3e143bd2010-02-14 00:48:49 +00003642 /* first, see if user specified a template filename on the command line. */
3643 if (user_templatename != 0) {
3644 if( access(user_templatename,004)==-1 ){
3645 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3646 user_templatename);
3647 lemp->errorcnt++;
3648 return 0;
3649 }
3650 in = fopen(user_templatename,"rb");
3651 if( in==0 ){
drh25473362015-09-04 18:03:45 +00003652 fprintf(stderr,"Can't open the template file \"%s\".\n",
3653 user_templatename);
icculus3e143bd2010-02-14 00:48:49 +00003654 lemp->errorcnt++;
3655 return 0;
3656 }
3657 return in;
3658 }
3659
drh75897232000-05-29 14:26:00 +00003660 cp = strrchr(lemp->filename,'.');
3661 if( cp ){
drh898799f2014-01-10 23:21:00 +00003662 lemon_sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003663 }else{
drh898799f2014-01-10 23:21:00 +00003664 lemon_sprintf(buf,"%s.lt",lemp->filename);
drh75897232000-05-29 14:26:00 +00003665 }
3666 if( access(buf,004)==0 ){
3667 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003668 }else if( access(templatename,004)==0 ){
3669 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003670 }else{
3671 tpltname = pathsearch(lemp->argv0,templatename,0);
3672 }
3673 if( tpltname==0 ){
3674 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3675 templatename);
3676 lemp->errorcnt++;
3677 return 0;
3678 }
drh2aa6ca42004-09-10 00:14:04 +00003679 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003680 if( in==0 ){
3681 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3682 lemp->errorcnt++;
3683 return 0;
3684 }
3685 return in;
3686}
3687
drhaf805ca2004-09-07 11:28:25 +00003688/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003689PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003690{
3691 fprintf(out,"#line %d \"",lineno);
3692 while( *filename ){
3693 if( *filename == '\\' ) putc('\\',out);
3694 putc(*filename,out);
3695 filename++;
3696 }
3697 fprintf(out,"\"\n");
3698}
3699
drh75897232000-05-29 14:26:00 +00003700/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003701PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003702{
3703 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003704 while( *str ){
drh75897232000-05-29 14:26:00 +00003705 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003706 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003707 str++;
3708 }
drh9db55df2004-09-09 14:01:21 +00003709 if( str[-1]!='\n' ){
3710 putc('\n',out);
3711 (*lineno)++;
3712 }
shane58543932008-12-10 20:10:04 +00003713 if (!lemp->nolinenosflag) {
drh06f60d82017-04-14 19:46:12 +00003714 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003715 }
drh75897232000-05-29 14:26:00 +00003716 return;
3717}
3718
3719/*
3720** The following routine emits code for the destructor for the
3721** symbol sp
3722*/
icculus9e44cf12010-02-14 17:14:22 +00003723void emit_destructor_code(
3724 FILE *out,
3725 struct symbol *sp,
3726 struct lemon *lemp,
3727 int *lineno
3728){
drhcc83b6e2004-04-23 23:38:42 +00003729 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003730
drh75897232000-05-29 14:26:00 +00003731 if( sp->type==TERMINAL ){
3732 cp = lemp->tokendest;
3733 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003734 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003735 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003736 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003737 fprintf(out,"{\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00003738 if( !lemp->nolinenosflag ){
3739 (*lineno)++;
3740 tplt_linedir(out,sp->destLineno,lemp->filename);
3741 }
drh960e8c62001-04-03 16:53:21 +00003742 }else if( lemp->vardest ){
3743 cp = lemp->vardest;
3744 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003745 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003746 }else{
3747 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003748 }
3749 for(; *cp; cp++){
3750 if( *cp=='$' && cp[1]=='$' ){
3751 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3752 cp++;
3753 continue;
3754 }
shane58543932008-12-10 20:10:04 +00003755 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003756 fputc(*cp,out);
3757 }
shane58543932008-12-10 20:10:04 +00003758 fprintf(out,"\n"); (*lineno)++;
drh06f60d82017-04-14 19:46:12 +00003759 if (!lemp->nolinenosflag) {
3760 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003761 }
3762 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003763 return;
3764}
3765
3766/*
drh960e8c62001-04-03 16:53:21 +00003767** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003768*/
icculus9e44cf12010-02-14 17:14:22 +00003769int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003770{
3771 int ret;
3772 if( sp->type==TERMINAL ){
3773 ret = lemp->tokendest!=0;
3774 }else{
drh960e8c62001-04-03 16:53:21 +00003775 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003776 }
3777 return ret;
3778}
3779
drh0bb132b2004-07-20 14:06:51 +00003780/*
3781** Append text to a dynamically allocated string. If zText is 0 then
3782** reset the string to be empty again. Always return the complete text
3783** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003784**
3785** n bytes of zText are stored. If n==0 then all of zText up to the first
3786** \000 terminator is stored. zText can contain up to two instances of
3787** %d. The values of p1 and p2 are written into the first and second
3788** %d.
3789**
3790** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003791*/
icculus9e44cf12010-02-14 17:14:22 +00003792PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3793 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003794 static char *z = 0;
3795 static int alloced = 0;
3796 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003797 int c;
drh0bb132b2004-07-20 14:06:51 +00003798 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003799 if( zText==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003800 if( used==0 && z!=0 ) z[0] = 0;
drh0bb132b2004-07-20 14:06:51 +00003801 used = 0;
3802 return z;
3803 }
drh7ac25c72004-08-19 15:12:26 +00003804 if( n<=0 ){
3805 if( n<0 ){
3806 used += n;
3807 assert( used>=0 );
3808 }
drh87cf1372008-08-13 20:09:06 +00003809 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003810 }
drhdf609712010-11-23 20:55:27 +00003811 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003812 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003813 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003814 }
icculus9e44cf12010-02-14 17:14:22 +00003815 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003816 while( n-- > 0 ){
3817 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003818 if( c=='%' && n>0 && zText[0]=='d' ){
drh898799f2014-01-10 23:21:00 +00003819 lemon_sprintf(zInt, "%d", p1);
drh0bb132b2004-07-20 14:06:51 +00003820 p1 = p2;
drh898799f2014-01-10 23:21:00 +00003821 lemon_strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003822 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003823 zText++;
3824 n--;
3825 }else{
mistachkin2318d332015-01-12 18:02:52 +00003826 z[used++] = (char)c;
drh0bb132b2004-07-20 14:06:51 +00003827 }
3828 }
3829 z[used] = 0;
3830 return z;
3831}
3832
3833/*
drh711c9812016-05-23 14:24:31 +00003834** Write and transform the rp->code string so that symbols are expanded.
3835** Populate the rp->codePrefix and rp->codeSuffix strings, as appropriate.
drhdabd04c2016-02-17 01:46:19 +00003836**
3837** Return 1 if the expanded code requires that "yylhsminor" local variable
3838** to be defined.
drh0bb132b2004-07-20 14:06:51 +00003839*/
drhdabd04c2016-02-17 01:46:19 +00003840PRIVATE int translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003841 char *cp, *xp;
3842 int i;
drhcf82f0d2016-02-17 04:33:10 +00003843 int rc = 0; /* True if yylhsminor is used */
drh43303de2016-02-17 12:34:03 +00003844 int dontUseRhs0 = 0; /* If true, use of left-most RHS label is illegal */
drhcf82f0d2016-02-17 04:33:10 +00003845 const char *zSkip = 0; /* The zOvwrt comment within rp->code, or NULL */
3846 char lhsused = 0; /* True if the LHS element has been used */
3847 char lhsdirect; /* True if LHS writes directly into stack */
3848 char used[MAXRHS]; /* True for each RHS element which is used */
3849 char zLhs[50]; /* Convert the LHS symbol into this string */
3850 char zOvwrt[900]; /* Comment that to allow LHS to overwrite RHS */
drh0bb132b2004-07-20 14:06:51 +00003851
3852 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3853 lhsused = 0;
3854
drh19c9e562007-03-29 20:13:53 +00003855 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003856 static char newlinestr[2] = { '\n', '\0' };
3857 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003858 rp->line = rp->ruleline;
drh711c9812016-05-23 14:24:31 +00003859 rp->noCode = 1;
3860 }else{
3861 rp->noCode = 0;
drh19c9e562007-03-29 20:13:53 +00003862 }
3863
drh4dd0d3f2016-02-17 01:18:33 +00003864
drh2e55b042016-04-30 17:19:30 +00003865 if( rp->nrhs==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003866 /* If there are no RHS symbols, then writing directly to the LHS is ok */
3867 lhsdirect = 1;
3868 }else if( rp->rhsalias[0]==0 ){
drh2e55b042016-04-30 17:19:30 +00003869 /* The left-most RHS symbol has no value. LHS direct is ok. But
drh4dd0d3f2016-02-17 01:18:33 +00003870 ** we have to call the distructor on the RHS symbol first. */
3871 lhsdirect = 1;
3872 if( has_destructor(rp->rhs[0],lemp) ){
3873 append_str(0,0,0,0);
3874 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
3875 rp->rhs[0]->index,1-rp->nrhs);
3876 rp->codePrefix = Strsafe(append_str(0,0,0,0));
drh711c9812016-05-23 14:24:31 +00003877 rp->noCode = 0;
drh4dd0d3f2016-02-17 01:18:33 +00003878 }
drh2e55b042016-04-30 17:19:30 +00003879 }else if( rp->lhsalias==0 ){
3880 /* There is no LHS value symbol. */
3881 lhsdirect = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003882 }else if( strcmp(rp->lhsalias,rp->rhsalias[0])==0 ){
drh06f60d82017-04-14 19:46:12 +00003883 /* The LHS symbol and the left-most RHS symbol are the same, so
drh4dd0d3f2016-02-17 01:18:33 +00003884 ** direct writing is allowed */
3885 lhsdirect = 1;
3886 lhsused = 1;
3887 used[0] = 1;
3888 if( rp->lhs->dtnum!=rp->rhs[0]->dtnum ){
3889 ErrorMsg(lemp->filename,rp->ruleline,
3890 "%s(%s) and %s(%s) share the same label but have "
3891 "different datatypes.",
3892 rp->lhs->name, rp->lhsalias, rp->rhs[0]->name, rp->rhsalias[0]);
3893 lemp->errorcnt++;
drh06f60d82017-04-14 19:46:12 +00003894 }
drh4dd0d3f2016-02-17 01:18:33 +00003895 }else{
drhcf82f0d2016-02-17 04:33:10 +00003896 lemon_sprintf(zOvwrt, "/*%s-overwrites-%s*/",
3897 rp->lhsalias, rp->rhsalias[0]);
3898 zSkip = strstr(rp->code, zOvwrt);
3899 if( zSkip!=0 ){
3900 /* The code contains a special comment that indicates that it is safe
3901 ** for the LHS label to overwrite left-most RHS label. */
3902 lhsdirect = 1;
3903 }else{
3904 lhsdirect = 0;
3905 }
drh4dd0d3f2016-02-17 01:18:33 +00003906 }
3907 if( lhsdirect ){
3908 sprintf(zLhs, "yymsp[%d].minor.yy%d",1-rp->nrhs,rp->lhs->dtnum);
3909 }else{
drhdabd04c2016-02-17 01:46:19 +00003910 rc = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003911 sprintf(zLhs, "yylhsminor.yy%d",rp->lhs->dtnum);
3912 }
3913
drh0bb132b2004-07-20 14:06:51 +00003914 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003915
3916 /* This const cast is wrong but harmless, if we're careful. */
3917 for(cp=(char *)rp->code; *cp; cp++){
drhcf82f0d2016-02-17 04:33:10 +00003918 if( cp==zSkip ){
3919 append_str(zOvwrt,0,0,0);
3920 cp += lemonStrlen(zOvwrt)-1;
drh43303de2016-02-17 12:34:03 +00003921 dontUseRhs0 = 1;
drhcf82f0d2016-02-17 04:33:10 +00003922 continue;
3923 }
drhc56fac72015-10-29 13:48:15 +00003924 if( ISALPHA(*cp) && (cp==rp->code || (!ISALNUM(cp[-1]) && cp[-1]!='_')) ){
drh0bb132b2004-07-20 14:06:51 +00003925 char saved;
drhc56fac72015-10-29 13:48:15 +00003926 for(xp= &cp[1]; ISALNUM(*xp) || *xp=='_'; xp++);
drh0bb132b2004-07-20 14:06:51 +00003927 saved = *xp;
3928 *xp = 0;
3929 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003930 append_str(zLhs,0,0,0);
drh0bb132b2004-07-20 14:06:51 +00003931 cp = xp;
3932 lhsused = 1;
3933 }else{
3934 for(i=0; i<rp->nrhs; i++){
3935 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh43303de2016-02-17 12:34:03 +00003936 if( i==0 && dontUseRhs0 ){
3937 ErrorMsg(lemp->filename,rp->ruleline,
3938 "Label %s used after '%s'.",
3939 rp->rhsalias[0], zOvwrt);
3940 lemp->errorcnt++;
3941 }else if( cp!=rp->code && cp[-1]=='@' ){
drh7ac25c72004-08-19 15:12:26 +00003942 /* If the argument is of the form @X then substituted
3943 ** the token number of X, not the value of X */
3944 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3945 }else{
drhfd405312005-11-06 04:06:59 +00003946 struct symbol *sp = rp->rhs[i];
3947 int dtnum;
3948 if( sp->type==MULTITERMINAL ){
3949 dtnum = sp->subsym[0]->dtnum;
3950 }else{
3951 dtnum = sp->dtnum;
3952 }
3953 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003954 }
drh0bb132b2004-07-20 14:06:51 +00003955 cp = xp;
3956 used[i] = 1;
3957 break;
3958 }
3959 }
3960 }
3961 *xp = saved;
3962 }
3963 append_str(cp, 1, 0, 0);
3964 } /* End loop */
3965
drh4dd0d3f2016-02-17 01:18:33 +00003966 /* Main code generation completed */
3967 cp = append_str(0,0,0,0);
3968 if( cp && cp[0] ) rp->code = Strsafe(cp);
3969 append_str(0,0,0,0);
3970
drh0bb132b2004-07-20 14:06:51 +00003971 /* Check to make sure the LHS has been used */
3972 if( rp->lhsalias && !lhsused ){
3973 ErrorMsg(lemp->filename,rp->ruleline,
3974 "Label \"%s\" for \"%s(%s)\" is never used.",
3975 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3976 lemp->errorcnt++;
3977 }
3978
drh4dd0d3f2016-02-17 01:18:33 +00003979 /* Generate destructor code for RHS minor values which are not referenced.
3980 ** Generate error messages for unused labels and duplicate labels.
3981 */
drh0bb132b2004-07-20 14:06:51 +00003982 for(i=0; i<rp->nrhs; i++){
drh4dd0d3f2016-02-17 01:18:33 +00003983 if( rp->rhsalias[i] ){
3984 if( i>0 ){
3985 int j;
3986 if( rp->lhsalias && strcmp(rp->lhsalias,rp->rhsalias[i])==0 ){
3987 ErrorMsg(lemp->filename,rp->ruleline,
3988 "%s(%s) has the same label as the LHS but is not the left-most "
3989 "symbol on the RHS.",
drhf135cb72019-04-30 14:26:31 +00003990 rp->rhs[i]->name, rp->rhsalias[i]);
drh4dd0d3f2016-02-17 01:18:33 +00003991 lemp->errorcnt++;
3992 }
3993 for(j=0; j<i; j++){
3994 if( rp->rhsalias[j] && strcmp(rp->rhsalias[j],rp->rhsalias[i])==0 ){
3995 ErrorMsg(lemp->filename,rp->ruleline,
3996 "Label %s used for multiple symbols on the RHS of a rule.",
3997 rp->rhsalias[i]);
3998 lemp->errorcnt++;
3999 break;
4000 }
4001 }
drh0bb132b2004-07-20 14:06:51 +00004002 }
drh4dd0d3f2016-02-17 01:18:33 +00004003 if( !used[i] ){
4004 ErrorMsg(lemp->filename,rp->ruleline,
4005 "Label %s for \"%s(%s)\" is never used.",
4006 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
4007 lemp->errorcnt++;
4008 }
4009 }else if( i>0 && has_destructor(rp->rhs[i],lemp) ){
4010 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
4011 rp->rhs[i]->index,i-rp->nrhs+1);
drh0bb132b2004-07-20 14:06:51 +00004012 }
4013 }
drh4dd0d3f2016-02-17 01:18:33 +00004014
4015 /* If unable to write LHS values directly into the stack, write the
4016 ** saved LHS value now. */
4017 if( lhsdirect==0 ){
4018 append_str(" yymsp[%d].minor.yy%d = ", 0, 1-rp->nrhs, rp->lhs->dtnum);
4019 append_str(zLhs, 0, 0, 0);
4020 append_str(";\n", 0, 0, 0);
drh61e339a2007-01-16 03:09:02 +00004021 }
drh4dd0d3f2016-02-17 01:18:33 +00004022
4023 /* Suffix code generation complete */
4024 cp = append_str(0,0,0,0);
drh711c9812016-05-23 14:24:31 +00004025 if( cp && cp[0] ){
4026 rp->codeSuffix = Strsafe(cp);
4027 rp->noCode = 0;
4028 }
drhdabd04c2016-02-17 01:46:19 +00004029
4030 return rc;
drh0bb132b2004-07-20 14:06:51 +00004031}
4032
drh06f60d82017-04-14 19:46:12 +00004033/*
drh75897232000-05-29 14:26:00 +00004034** Generate code which executes when the rule "rp" is reduced. Write
4035** the code to "out". Make sure lineno stays up-to-date.
4036*/
icculus9e44cf12010-02-14 17:14:22 +00004037PRIVATE void emit_code(
4038 FILE *out,
4039 struct rule *rp,
4040 struct lemon *lemp,
4041 int *lineno
4042){
4043 const char *cp;
drh75897232000-05-29 14:26:00 +00004044
drh4dd0d3f2016-02-17 01:18:33 +00004045 /* Setup code prior to the #line directive */
4046 if( rp->codePrefix && rp->codePrefix[0] ){
4047 fprintf(out, "{%s", rp->codePrefix);
4048 for(cp=rp->codePrefix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
4049 }
4050
drh75897232000-05-29 14:26:00 +00004051 /* Generate code to do the reduce action */
4052 if( rp->code ){
drh25473362015-09-04 18:03:45 +00004053 if( !lemp->nolinenosflag ){
4054 (*lineno)++;
4055 tplt_linedir(out,rp->line,lemp->filename);
4056 }
drhaf805ca2004-09-07 11:28:25 +00004057 fprintf(out,"{%s",rp->code);
drh4dd0d3f2016-02-17 01:18:33 +00004058 for(cp=rp->code; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
shane58543932008-12-10 20:10:04 +00004059 fprintf(out,"}\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00004060 if( !lemp->nolinenosflag ){
4061 (*lineno)++;
4062 tplt_linedir(out,*lineno,lemp->outname);
4063 }
drh4dd0d3f2016-02-17 01:18:33 +00004064 }
4065
4066 /* Generate breakdown code that occurs after the #line directive */
4067 if( rp->codeSuffix && rp->codeSuffix[0] ){
4068 fprintf(out, "%s", rp->codeSuffix);
4069 for(cp=rp->codeSuffix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
4070 }
4071
4072 if( rp->codePrefix ){
4073 fprintf(out, "}\n"); (*lineno)++;
4074 }
drh75897232000-05-29 14:26:00 +00004075
drh75897232000-05-29 14:26:00 +00004076 return;
4077}
4078
4079/*
4080** Print the definition of the union used for the parser's data stack.
4081** This union contains fields for every possible data type for tokens
4082** and nonterminals. In the process of computing and printing this
4083** union, also set the ".dtnum" field of every terminal and nonterminal
4084** symbol.
4085*/
icculus9e44cf12010-02-14 17:14:22 +00004086void print_stack_union(
4087 FILE *out, /* The output stream */
4088 struct lemon *lemp, /* The main info structure for this parser */
4089 int *plineno, /* Pointer to the line number */
4090 int mhflag /* True if generating makeheaders output */
4091){
drh75897232000-05-29 14:26:00 +00004092 int lineno = *plineno; /* The line number of the output */
4093 char **types; /* A hash table of datatypes */
4094 int arraysize; /* Size of the "types" array */
4095 int maxdtlength; /* Maximum length of any ".datatype" field. */
4096 char *stddt; /* Standardized name for a datatype */
4097 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00004098 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00004099 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00004100
4101 /* Allocate and initialize types[] and allocate stddt[] */
4102 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00004103 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00004104 if( types==0 ){
4105 fprintf(stderr,"Out of memory.\n");
4106 exit(1);
4107 }
drh75897232000-05-29 14:26:00 +00004108 for(i=0; i<arraysize; i++) types[i] = 0;
4109 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00004110 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00004111 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00004112 }
drh75897232000-05-29 14:26:00 +00004113 for(i=0; i<lemp->nsymbol; i++){
4114 int len;
4115 struct symbol *sp = lemp->symbols[i];
4116 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00004117 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00004118 if( len>maxdtlength ) maxdtlength = len;
4119 }
4120 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00004121 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00004122 fprintf(stderr,"Out of memory.\n");
4123 exit(1);
4124 }
4125
4126 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
4127 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00004128 ** used for terminal symbols. If there is no %default_type defined then
4129 ** 0 is also used as the .dtnum value for nonterminals which do not specify
4130 ** a datatype using the %type directive.
4131 */
drh75897232000-05-29 14:26:00 +00004132 for(i=0; i<lemp->nsymbol; i++){
4133 struct symbol *sp = lemp->symbols[i];
4134 char *cp;
4135 if( sp==lemp->errsym ){
4136 sp->dtnum = arraysize+1;
4137 continue;
4138 }
drh960e8c62001-04-03 16:53:21 +00004139 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00004140 sp->dtnum = 0;
4141 continue;
4142 }
4143 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00004144 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00004145 j = 0;
drhc56fac72015-10-29 13:48:15 +00004146 while( ISSPACE(*cp) ) cp++;
drh75897232000-05-29 14:26:00 +00004147 while( *cp ) stddt[j++] = *cp++;
drhc56fac72015-10-29 13:48:15 +00004148 while( j>0 && ISSPACE(stddt[j-1]) ) j--;
drh75897232000-05-29 14:26:00 +00004149 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00004150 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00004151 sp->dtnum = 0;
4152 continue;
4153 }
drh75897232000-05-29 14:26:00 +00004154 hash = 0;
4155 for(j=0; stddt[j]; j++){
4156 hash = hash*53 + stddt[j];
4157 }
drh3b2129c2003-05-13 00:34:21 +00004158 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00004159 while( types[hash] ){
4160 if( strcmp(types[hash],stddt)==0 ){
4161 sp->dtnum = hash + 1;
4162 break;
4163 }
4164 hash++;
drh2b51f212013-10-11 23:01:02 +00004165 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00004166 }
4167 if( types[hash]==0 ){
4168 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00004169 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00004170 if( types[hash]==0 ){
4171 fprintf(stderr,"Out of memory.\n");
4172 exit(1);
4173 }
drh898799f2014-01-10 23:21:00 +00004174 lemon_strcpy(types[hash],stddt);
drh75897232000-05-29 14:26:00 +00004175 }
4176 }
4177
4178 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
4179 name = lemp->name ? lemp->name : "Parse";
4180 lineno = *plineno;
4181 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
4182 fprintf(out,"#define %sTOKENTYPE %s\n",name,
4183 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
4184 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
4185 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00004186 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004187 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
4188 for(i=0; i<arraysize; i++){
4189 if( types[i]==0 ) continue;
4190 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
4191 free(types[i]);
4192 }
drhed0c15b2018-04-16 14:31:34 +00004193 if( lemp->errsym && lemp->errsym->useCnt ){
drhc4dd3fd2008-01-22 01:48:05 +00004194 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
4195 }
drh75897232000-05-29 14:26:00 +00004196 free(stddt);
4197 free(types);
4198 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
4199 *plineno = lineno;
4200}
4201
drhb29b0a52002-02-23 19:39:46 +00004202/*
4203** Return the name of a C datatype able to represent values between
drhc75e0162015-09-07 02:23:02 +00004204** lwr and upr, inclusive. If pnByte!=NULL then also write the sizeof
4205** for that type (1, 2, or 4) into *pnByte.
drhb29b0a52002-02-23 19:39:46 +00004206*/
drhc75e0162015-09-07 02:23:02 +00004207static const char *minimum_size_type(int lwr, int upr, int *pnByte){
4208 const char *zType = "int";
4209 int nByte = 4;
drh8b582012003-10-21 13:16:03 +00004210 if( lwr>=0 ){
4211 if( upr<=255 ){
drhc75e0162015-09-07 02:23:02 +00004212 zType = "unsigned char";
4213 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004214 }else if( upr<65535 ){
drhc75e0162015-09-07 02:23:02 +00004215 zType = "unsigned short int";
4216 nByte = 2;
drh8b582012003-10-21 13:16:03 +00004217 }else{
drhc75e0162015-09-07 02:23:02 +00004218 zType = "unsigned int";
4219 nByte = 4;
drh8b582012003-10-21 13:16:03 +00004220 }
4221 }else if( lwr>=-127 && upr<=127 ){
drhc75e0162015-09-07 02:23:02 +00004222 zType = "signed char";
4223 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004224 }else if( lwr>=-32767 && upr<32767 ){
drhc75e0162015-09-07 02:23:02 +00004225 zType = "short";
4226 nByte = 2;
drhb29b0a52002-02-23 19:39:46 +00004227 }
drhc75e0162015-09-07 02:23:02 +00004228 if( pnByte ) *pnByte = nByte;
4229 return zType;
drhb29b0a52002-02-23 19:39:46 +00004230}
4231
drhfdbf9282003-10-21 16:34:41 +00004232/*
4233** Each state contains a set of token transaction and a set of
4234** nonterminal transactions. Each of these sets makes an instance
4235** of the following structure. An array of these structures is used
4236** to order the creation of entries in the yy_action[] table.
4237*/
4238struct axset {
4239 struct state *stp; /* A pointer to a state */
4240 int isTkn; /* True to use tokens. False for non-terminals */
4241 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00004242 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00004243};
4244
4245/*
4246** Compare to axset structures for sorting purposes
4247*/
4248static int axset_compare(const void *a, const void *b){
4249 struct axset *p1 = (struct axset*)a;
4250 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00004251 int c;
4252 c = p2->nAction - p1->nAction;
4253 if( c==0 ){
drh337cd0d2015-09-07 23:40:42 +00004254 c = p1->iOrder - p2->iOrder;
drhe594bc32009-11-03 13:02:25 +00004255 }
4256 assert( c!=0 || p1==p2 );
4257 return c;
drhfdbf9282003-10-21 16:34:41 +00004258}
4259
drhc4dd3fd2008-01-22 01:48:05 +00004260/*
4261** Write text on "out" that describes the rule "rp".
4262*/
4263static void writeRuleText(FILE *out, struct rule *rp){
4264 int j;
4265 fprintf(out,"%s ::=", rp->lhs->name);
4266 for(j=0; j<rp->nrhs; j++){
4267 struct symbol *sp = rp->rhs[j];
drh61f92cd2014-01-11 03:06:18 +00004268 if( sp->type!=MULTITERMINAL ){
4269 fprintf(out," %s", sp->name);
4270 }else{
drhc4dd3fd2008-01-22 01:48:05 +00004271 int k;
drh61f92cd2014-01-11 03:06:18 +00004272 fprintf(out," %s", sp->subsym[0]->name);
drhc4dd3fd2008-01-22 01:48:05 +00004273 for(k=1; k<sp->nsubsym; k++){
4274 fprintf(out,"|%s",sp->subsym[k]->name);
4275 }
4276 }
4277 }
4278}
4279
4280
drh75897232000-05-29 14:26:00 +00004281/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004282void ReportTable(
4283 struct lemon *lemp,
drhfe03dac2019-11-26 02:22:39 +00004284 int mhflag, /* Output in makeheaders format if true */
4285 int sqlFlag /* Generate the *.sql file too */
icculus9e44cf12010-02-14 17:14:22 +00004286){
drhfe03dac2019-11-26 02:22:39 +00004287 FILE *out, *in, *sql;
drh75897232000-05-29 14:26:00 +00004288 char line[LINESIZE];
4289 int lineno;
4290 struct state *stp;
4291 struct action *ap;
4292 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00004293 struct acttab *pActtab;
drhc75e0162015-09-07 02:23:02 +00004294 int i, j, n, sz;
drh2e517162019-08-28 02:09:47 +00004295 int nLookAhead;
drhc75e0162015-09-07 02:23:02 +00004296 int szActionType; /* sizeof(YYACTIONTYPE) */
4297 int szCodeType; /* sizeof(YYCODETYPE) */
icculus9e44cf12010-02-14 17:14:22 +00004298 const char *name;
drh8b582012003-10-21 13:16:03 +00004299 int mnTknOfst, mxTknOfst;
4300 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00004301 struct axset *ax;
drh60c71b02020-09-01 11:20:03 +00004302 char *prefix;
drh75897232000-05-29 14:26:00 +00004303
drh5c8241b2017-12-24 23:38:10 +00004304 lemp->minShiftReduce = lemp->nstate;
4305 lemp->errAction = lemp->minShiftReduce + lemp->nrule;
4306 lemp->accAction = lemp->errAction + 1;
4307 lemp->noAction = lemp->accAction + 1;
4308 lemp->minReduce = lemp->noAction + 1;
4309 lemp->maxAction = lemp->minReduce + lemp->nrule;
4310
drh75897232000-05-29 14:26:00 +00004311 in = tplt_open(lemp);
4312 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00004313 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00004314 if( out==0 ){
4315 fclose(in);
4316 return;
4317 }
drhfe03dac2019-11-26 02:22:39 +00004318 if( sqlFlag==0 ){
4319 sql = 0;
4320 }else{
4321 sql = file_open(lemp, ".sql", "wb");
4322 if( sql==0 ){
4323 fclose(in);
4324 fclose(out);
4325 return;
4326 }
4327 fprintf(sql,
drh1417c2f2019-11-29 12:51:00 +00004328 "BEGIN;\n"
drhfe03dac2019-11-26 02:22:39 +00004329 "CREATE TABLE symbol(\n"
4330 " id INTEGER PRIMARY KEY,\n"
4331 " name TEXT NOT NULL,\n"
4332 " isTerminal BOOLEAN NOT NULL,\n"
drh1417c2f2019-11-29 12:51:00 +00004333 " fallback INTEGER REFERENCES symbol"
4334 " DEFERRABLE INITIALLY DEFERRED\n"
drhfe03dac2019-11-26 02:22:39 +00004335 ");\n"
4336 );
4337 for(i=0; i<lemp->nsymbol; i++){
4338 fprintf(sql,
4339 "INSERT INTO symbol(id,name,isTerminal,fallback)"
4340 "VALUES(%d,'%s',%s",
4341 i, lemp->symbols[i]->name,
4342 i<lemp->nterminal ? "TRUE" : "FALSE"
4343 );
4344 if( lemp->symbols[i]->fallback ){
4345 fprintf(sql, ",%d);\n", lemp->symbols[i]->fallback->index);
4346 }else{
4347 fprintf(sql, ",NULL);\n");
4348 }
4349 }
4350 fprintf(sql,
4351 "CREATE TABLE rule(\n"
4352 " ruleid INTEGER PRIMARY KEY,\n"
drh3e5f7fe2019-12-19 12:29:31 +00004353 " lhs INTEGER REFERENCES symbol(id),\n"
4354 " txt TEXT\n"
drhfe03dac2019-11-26 02:22:39 +00004355 ");\n"
4356 "CREATE TABLE rulerhs(\n"
4357 " ruleid INTEGER REFERENCES rule(ruleid),\n"
4358 " pos INTEGER,\n"
4359 " sym INTEGER REFERENCES symbol(id)\n"
4360 ");\n"
4361 );
4362 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4363 assert( i==rp->iRule );
4364 fprintf(sql,
drh59c56792019-12-19 13:17:07 +00004365 "INSERT INTO rule(ruleid,lhs,txt)VALUES(%d,%d,'",
drhfe03dac2019-11-26 02:22:39 +00004366 rp->iRule, rp->lhs->index
4367 );
drh3e5f7fe2019-12-19 12:29:31 +00004368 writeRuleText(sql, rp);
4369 fprintf(sql,"');\n");
drhfe03dac2019-11-26 02:22:39 +00004370 for(j=0; j<rp->nrhs; j++){
4371 struct symbol *sp = rp->rhs[j];
4372 if( sp->type!=MULTITERMINAL ){
4373 fprintf(sql,
4374 "INSERT INTO rulerhs(ruleid,pos,sym)VALUES(%d,%d,%d);\n",
4375 i,j,sp->index
4376 );
4377 }else{
4378 int k;
4379 for(k=0; k<sp->nsubsym; k++){
4380 fprintf(sql,
4381 "INSERT INTO rulerhs(ruleid,pos,sym)VALUES(%d,%d,%d);\n",
4382 i,j,sp->subsym[k]->index
4383 );
4384 }
4385 }
4386 }
4387 }
drh1417c2f2019-11-29 12:51:00 +00004388 fprintf(sql, "COMMIT;\n");
drhfe03dac2019-11-26 02:22:39 +00004389 }
drh75897232000-05-29 14:26:00 +00004390 lineno = 1;
drh60c71b02020-09-01 11:20:03 +00004391
drh512aa782020-09-01 12:26:55 +00004392 fprintf(out,
4393 "/* This file is automatically generated by Lemon from input grammar\n"
4394 "** source file \"%s\". */\n", lemp->filename); lineno += 2;
4395
drh60c71b02020-09-01 11:20:03 +00004396 /* The first %include directive begins with a C-language comment,
4397 ** then skip over the header comment of the template file
4398 */
4399 if( lemp->include==0 ) lemp->include = "";
4400 for(i=0; ISSPACE(lemp->include[i]); i++){
4401 if( lemp->include[i]=='\n' ){
4402 lemp->include += i+1;
4403 i = -1;
4404 }
4405 }
4406 if( lemp->include[0]=='/' ){
4407 tplt_skip_header(in,&lineno);
4408 }else{
4409 tplt_xfer(lemp->name,in,out,&lineno);
4410 }
drh75897232000-05-29 14:26:00 +00004411
4412 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00004413 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00004414 if( mhflag ){
mistachkin8e189222015-04-19 21:43:16 +00004415 char *incName = file_makename(lemp, ".h");
4416 fprintf(out,"#include \"%s\"\n", incName); lineno++;
4417 free(incName);
drh75897232000-05-29 14:26:00 +00004418 }
4419 tplt_xfer(lemp->name,in,out,&lineno);
4420
4421 /* Generate #defines for all tokens */
drh60c71b02020-09-01 11:20:03 +00004422 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4423 else prefix = "";
drh75897232000-05-29 14:26:00 +00004424 if( mhflag ){
drh75897232000-05-29 14:26:00 +00004425 fprintf(out,"#if INTERFACE\n"); lineno++;
drh60c71b02020-09-01 11:20:03 +00004426 }else{
4427 fprintf(out,"#ifndef %s%s\n", prefix, lemp->symbols[1]->name);
drh75897232000-05-29 14:26:00 +00004428 }
drh60c71b02020-09-01 11:20:03 +00004429 for(i=1; i<lemp->nterminal; i++){
4430 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4431 lineno++;
4432 }
4433 fprintf(out,"#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004434 tplt_xfer(lemp->name,in,out,&lineno);
4435
4436 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00004437 fprintf(out,"#define YYCODETYPE %s\n",
drhed0c15b2018-04-16 14:31:34 +00004438 minimum_size_type(0, lemp->nsymbol, &szCodeType)); lineno++;
4439 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol); lineno++;
drh75897232000-05-29 14:26:00 +00004440 fprintf(out,"#define YYACTIONTYPE %s\n",
drh5c8241b2017-12-24 23:38:10 +00004441 minimum_size_type(0,lemp->maxAction,&szActionType)); lineno++;
drhe09daa92006-06-10 13:29:31 +00004442 if( lemp->wildcard ){
4443 fprintf(out,"#define YYWILDCARD %d\n",
4444 lemp->wildcard->index); lineno++;
4445 }
drh75897232000-05-29 14:26:00 +00004446 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00004447 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004448 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00004449 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
4450 }else{
4451 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
4452 }
drhca44b5a2007-02-22 23:06:58 +00004453 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004454 if( mhflag ){
4455 fprintf(out,"#if INTERFACE\n"); lineno++;
4456 }
4457 name = lemp->name ? lemp->name : "Parse";
4458 if( lemp->arg && lemp->arg[0] ){
drh87cf1372008-08-13 20:09:06 +00004459 i = lemonStrlen(lemp->arg);
drhc56fac72015-10-29 13:48:15 +00004460 while( i>=1 && ISSPACE(lemp->arg[i-1]) ) i--;
4461 while( i>=1 && (ISALNUM(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00004462 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
4463 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
drhfb32c442018-04-21 13:51:42 +00004464 fprintf(out,"#define %sARG_PARAM ,%s\n",name,&lemp->arg[i]); lineno++;
4465 fprintf(out,"#define %sARG_FETCH %s=yypParser->%s;\n",
drh1f245e42002-03-11 13:55:50 +00004466 name,lemp->arg,&lemp->arg[i]); lineno++;
drhfb32c442018-04-21 13:51:42 +00004467 fprintf(out,"#define %sARG_STORE yypParser->%s=%s;\n",
drh1f245e42002-03-11 13:55:50 +00004468 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00004469 }else{
drhfb32c442018-04-21 13:51:42 +00004470 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
4471 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
4472 fprintf(out,"#define %sARG_PARAM\n",name); lineno++;
drh1f245e42002-03-11 13:55:50 +00004473 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
4474 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00004475 }
drhfb32c442018-04-21 13:51:42 +00004476 if( lemp->ctx && lemp->ctx[0] ){
4477 i = lemonStrlen(lemp->ctx);
4478 while( i>=1 && ISSPACE(lemp->ctx[i-1]) ) i--;
4479 while( i>=1 && (ISALNUM(lemp->ctx[i-1]) || lemp->ctx[i-1]=='_') ) i--;
4480 fprintf(out,"#define %sCTX_SDECL %s;\n",name,lemp->ctx); lineno++;
4481 fprintf(out,"#define %sCTX_PDECL ,%s\n",name,lemp->ctx); lineno++;
4482 fprintf(out,"#define %sCTX_PARAM ,%s\n",name,&lemp->ctx[i]); lineno++;
4483 fprintf(out,"#define %sCTX_FETCH %s=yypParser->%s;\n",
4484 name,lemp->ctx,&lemp->ctx[i]); lineno++;
4485 fprintf(out,"#define %sCTX_STORE yypParser->%s=%s;\n",
4486 name,&lemp->ctx[i],&lemp->ctx[i]); lineno++;
4487 }else{
4488 fprintf(out,"#define %sCTX_SDECL\n",name); lineno++;
4489 fprintf(out,"#define %sCTX_PDECL\n",name); lineno++;
4490 fprintf(out,"#define %sCTX_PARAM\n",name); lineno++;
4491 fprintf(out,"#define %sCTX_FETCH\n",name); lineno++;
4492 fprintf(out,"#define %sCTX_STORE\n",name); lineno++;
4493 }
drh75897232000-05-29 14:26:00 +00004494 if( mhflag ){
4495 fprintf(out,"#endif\n"); lineno++;
4496 }
drhed0c15b2018-04-16 14:31:34 +00004497 if( lemp->errsym && lemp->errsym->useCnt ){
drh3bd48ab2015-09-07 18:23:37 +00004498 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
4499 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00004500 }
drh0bd1f4e2002-06-06 18:54:39 +00004501 if( lemp->has_fallback ){
4502 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
4503 }
drh75897232000-05-29 14:26:00 +00004504
drh3bd48ab2015-09-07 18:23:37 +00004505 /* Compute the action table, but do not output it yet. The action
4506 ** table must be computed before generating the YYNSTATE macro because
4507 ** we need to know how many states can be eliminated.
drh75897232000-05-29 14:26:00 +00004508 */
drh3bd48ab2015-09-07 18:23:37 +00004509 ax = (struct axset *) calloc(lemp->nxstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00004510 if( ax==0 ){
4511 fprintf(stderr,"malloc failed\n");
4512 exit(1);
4513 }
drh3bd48ab2015-09-07 18:23:37 +00004514 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00004515 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004516 ax[i*2].stp = stp;
4517 ax[i*2].isTkn = 1;
4518 ax[i*2].nAction = stp->nTknAct;
4519 ax[i*2+1].stp = stp;
4520 ax[i*2+1].isTkn = 0;
4521 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00004522 }
drh8b582012003-10-21 13:16:03 +00004523 mxTknOfst = mnTknOfst = 0;
4524 mxNtOfst = mnNtOfst = 0;
drh3bd48ab2015-09-07 18:23:37 +00004525 /* In an effort to minimize the action table size, use the heuristic
4526 ** of placing the largest action sets first */
4527 for(i=0; i<lemp->nxstate*2; i++) ax[i].iOrder = i;
4528 qsort(ax, lemp->nxstate*2, sizeof(ax[0]), axset_compare);
drh3a9d6c72017-12-25 04:15:38 +00004529 pActtab = acttab_alloc(lemp->nsymbol, lemp->nterminal);
drh3bd48ab2015-09-07 18:23:37 +00004530 for(i=0; i<lemp->nxstate*2 && ax[i].nAction>0; i++){
drhfdbf9282003-10-21 16:34:41 +00004531 stp = ax[i].stp;
4532 if( ax[i].isTkn ){
4533 for(ap=stp->ap; ap; ap=ap->next){
4534 int action;
4535 if( ap->sp->index>=lemp->nterminal ) continue;
4536 action = compute_action(lemp, ap);
4537 if( action<0 ) continue;
4538 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004539 }
drh3a9d6c72017-12-25 04:15:38 +00004540 stp->iTknOfst = acttab_insert(pActtab, 1);
drhfdbf9282003-10-21 16:34:41 +00004541 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
4542 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
4543 }else{
4544 for(ap=stp->ap; ap; ap=ap->next){
4545 int action;
4546 if( ap->sp->index<lemp->nterminal ) continue;
4547 if( ap->sp->index==lemp->nsymbol ) continue;
4548 action = compute_action(lemp, ap);
4549 if( action<0 ) continue;
4550 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004551 }
drh3a9d6c72017-12-25 04:15:38 +00004552 stp->iNtOfst = acttab_insert(pActtab, 0);
drhfdbf9282003-10-21 16:34:41 +00004553 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
4554 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00004555 }
drh337cd0d2015-09-07 23:40:42 +00004556#if 0 /* Uncomment for a trace of how the yy_action[] table fills out */
4557 { int jj, nn;
4558 for(jj=nn=0; jj<pActtab->nAction; jj++){
4559 if( pActtab->aAction[jj].action<0 ) nn++;
4560 }
4561 printf("%4d: State %3d %s n: %2d size: %5d freespace: %d\n",
4562 i, stp->statenum, ax[i].isTkn ? "Token" : "Var ",
4563 ax[i].nAction, pActtab->nAction, nn);
4564 }
4565#endif
drh8b582012003-10-21 13:16:03 +00004566 }
drhfdbf9282003-10-21 16:34:41 +00004567 free(ax);
drh8b582012003-10-21 13:16:03 +00004568
drh756b41e2016-05-24 18:55:08 +00004569 /* Mark rules that are actually used for reduce actions after all
4570 ** optimizations have been applied
4571 */
4572 for(rp=lemp->rule; rp; rp=rp->next) rp->doesReduce = LEMON_FALSE;
4573 for(i=0; i<lemp->nxstate; i++){
drh756b41e2016-05-24 18:55:08 +00004574 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
4575 if( ap->type==REDUCE || ap->type==SHIFTREDUCE ){
drh69bfa552017-04-26 04:32:17 +00004576 ap->x.rp->doesReduce = 1;
drh756b41e2016-05-24 18:55:08 +00004577 }
4578 }
4579 }
4580
drh3bd48ab2015-09-07 18:23:37 +00004581 /* Finish rendering the constants now that the action table has
4582 ** been computed */
4583 fprintf(out,"#define YYNSTATE %d\n",lemp->nxstate); lineno++;
4584 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhce678c22019-12-11 18:53:51 +00004585 fprintf(out,"#define YYNRULE_WITH_ACTION %d\n",lemp->nruleWithAction);
4586 lineno++;
drh0d9de992017-12-26 18:04:23 +00004587 fprintf(out,"#define YYNTOKEN %d\n",lemp->nterminal); lineno++;
drh337cd0d2015-09-07 23:40:42 +00004588 fprintf(out,"#define YY_MAX_SHIFT %d\n",lemp->nxstate-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004589 i = lemp->minShiftReduce;
4590 fprintf(out,"#define YY_MIN_SHIFTREDUCE %d\n",i); lineno++;
4591 i += lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004592 fprintf(out,"#define YY_MAX_SHIFTREDUCE %d\n", i-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004593 fprintf(out,"#define YY_ERROR_ACTION %d\n", lemp->errAction); lineno++;
4594 fprintf(out,"#define YY_ACCEPT_ACTION %d\n", lemp->accAction); lineno++;
4595 fprintf(out,"#define YY_NO_ACTION %d\n", lemp->noAction); lineno++;
4596 fprintf(out,"#define YY_MIN_REDUCE %d\n", lemp->minReduce); lineno++;
4597 i = lemp->minReduce + lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004598 fprintf(out,"#define YY_MAX_REDUCE %d\n", i-1); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004599 tplt_xfer(lemp->name,in,out,&lineno);
4600
4601 /* Now output the action table and its associates:
4602 **
4603 ** yy_action[] A single table containing all actions.
4604 ** yy_lookahead[] A table containing the lookahead for each entry in
4605 ** yy_action. Used to detect hash collisions.
4606 ** yy_shift_ofst[] For each state, the offset into yy_action for
4607 ** shifting terminals.
4608 ** yy_reduce_ofst[] For each state, the offset into yy_action for
4609 ** shifting non-terminals after a reduce.
4610 ** yy_default[] Default action for each state.
4611 */
4612
drh8b582012003-10-21 13:16:03 +00004613 /* Output the yy_action table */
drh3a9d6c72017-12-25 04:15:38 +00004614 lemp->nactiontab = n = acttab_action_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004615 lemp->tablesize += n*szActionType;
drhf16371d2009-11-03 19:18:31 +00004616 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
4617 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004618 for(i=j=0; i<n; i++){
4619 int action = acttab_yyaction(pActtab, i);
drh5c8241b2017-12-24 23:38:10 +00004620 if( action<0 ) action = lemp->noAction;
drhfdbf9282003-10-21 16:34:41 +00004621 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004622 fprintf(out, " %4d,", action);
4623 if( j==9 || i==n-1 ){
4624 fprintf(out, "\n"); lineno++;
4625 j = 0;
4626 }else{
4627 j++;
4628 }
4629 }
4630 fprintf(out, "};\n"); lineno++;
4631
4632 /* Output the yy_lookahead table */
drh3a9d6c72017-12-25 04:15:38 +00004633 lemp->nlookaheadtab = n = acttab_lookahead_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004634 lemp->tablesize += n*szCodeType;
drh57196282004-10-06 15:41:16 +00004635 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004636 for(i=j=0; i<n; i++){
4637 int la = acttab_yylookahead(pActtab, i);
4638 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00004639 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004640 fprintf(out, " %4d,", la);
drh2e517162019-08-28 02:09:47 +00004641 if( j==9 ){
drh8b582012003-10-21 13:16:03 +00004642 fprintf(out, "\n"); lineno++;
4643 j = 0;
4644 }else{
4645 j++;
4646 }
4647 }
drh2e517162019-08-28 02:09:47 +00004648 /* Add extra entries to the end of the yy_lookahead[] table so that
4649 ** yy_shift_ofst[]+iToken will always be a valid index into the array,
4650 ** even for the largest possible value of yy_shift_ofst[] and iToken. */
4651 nLookAhead = lemp->nterminal + lemp->nactiontab;
4652 while( i<nLookAhead ){
4653 if( j==0 ) fprintf(out," /* %5d */ ", i);
4654 fprintf(out, " %4d,", lemp->nterminal);
4655 if( j==9 ){
4656 fprintf(out, "\n"); lineno++;
4657 j = 0;
4658 }else{
4659 j++;
4660 }
4661 i++;
4662 }
mistachkinacf6e082019-09-11 15:25:26 +00004663 if( j>0 ){ fprintf(out, "\n"); lineno++; }
drh8b582012003-10-21 13:16:03 +00004664 fprintf(out, "};\n"); lineno++;
4665
4666 /* Output the yy_shift_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004667 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004668 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drh701b6882016-08-10 13:30:43 +00004669 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
4670 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
4671 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004672 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh701b6882016-08-10 13:30:43 +00004673 minimum_size_type(mnTknOfst, lemp->nterminal+lemp->nactiontab, &sz));
4674 lineno++;
drhc75e0162015-09-07 02:23:02 +00004675 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004676 for(i=j=0; i<n; i++){
4677 int ofst;
4678 stp = lemp->sorted[i];
4679 ofst = stp->iTknOfst;
drh701b6882016-08-10 13:30:43 +00004680 if( ofst==NO_OFFSET ) ofst = lemp->nactiontab;
drhfdbf9282003-10-21 16:34:41 +00004681 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004682 fprintf(out, " %4d,", ofst);
4683 if( j==9 || i==n-1 ){
4684 fprintf(out, "\n"); lineno++;
4685 j = 0;
4686 }else{
4687 j++;
4688 }
4689 }
4690 fprintf(out, "};\n"); lineno++;
4691
4692 /* Output the yy_reduce_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004693 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004694 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00004695 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
4696 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
4697 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004698 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drhc75e0162015-09-07 02:23:02 +00004699 minimum_size_type(mnNtOfst-1, mxNtOfst, &sz)); lineno++;
4700 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004701 for(i=j=0; i<n; i++){
4702 int ofst;
4703 stp = lemp->sorted[i];
4704 ofst = stp->iNtOfst;
4705 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00004706 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004707 fprintf(out, " %4d,", ofst);
4708 if( j==9 || i==n-1 ){
4709 fprintf(out, "\n"); lineno++;
4710 j = 0;
4711 }else{
4712 j++;
4713 }
4714 }
4715 fprintf(out, "};\n"); lineno++;
4716
4717 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00004718 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004719 n = lemp->nxstate;
drhc75e0162015-09-07 02:23:02 +00004720 lemp->tablesize += n*szActionType;
drh8b582012003-10-21 13:16:03 +00004721 for(i=j=0; i<n; i++){
4722 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004723 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh5c8241b2017-12-24 23:38:10 +00004724 if( stp->iDfltReduce<0 ){
4725 fprintf(out, " %4d,", lemp->errAction);
4726 }else{
4727 fprintf(out, " %4d,", stp->iDfltReduce + lemp->minReduce);
4728 }
drh8b582012003-10-21 13:16:03 +00004729 if( j==9 || i==n-1 ){
4730 fprintf(out, "\n"); lineno++;
4731 j = 0;
4732 }else{
4733 j++;
4734 }
4735 }
4736 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004737 tplt_xfer(lemp->name,in,out,&lineno);
4738
drh0bd1f4e2002-06-06 18:54:39 +00004739 /* Generate the table of fallback tokens.
4740 */
4741 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00004742 int mx = lemp->nterminal - 1;
drh010bdb42019-08-28 11:31:11 +00004743 /* 2019-08-28: Generate fallback entries for every token to avoid
4744 ** having to do a range check on the index */
4745 /* while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; } */
drhc75e0162015-09-07 02:23:02 +00004746 lemp->tablesize += (mx+1)*szCodeType;
drh1441f3e2009-06-12 12:50:50 +00004747 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00004748 struct symbol *p = lemp->symbols[i];
4749 if( p->fallback==0 ){
4750 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
4751 }else{
4752 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
4753 p->name, p->fallback->name);
4754 }
4755 lineno++;
4756 }
4757 }
4758 tplt_xfer(lemp->name, in, out, &lineno);
4759
4760 /* Generate a table containing the symbolic name of every symbol
4761 */
drh75897232000-05-29 14:26:00 +00004762 for(i=0; i<lemp->nsymbol; i++){
drh898799f2014-01-10 23:21:00 +00004763 lemon_sprintf(line,"\"%s\",",lemp->symbols[i]->name);
drh3a9d6c72017-12-25 04:15:38 +00004764 fprintf(out," /* %4d */ \"%s\",\n",i, lemp->symbols[i]->name); lineno++;
drh75897232000-05-29 14:26:00 +00004765 }
drh75897232000-05-29 14:26:00 +00004766 tplt_xfer(lemp->name,in,out,&lineno);
4767
drh0bd1f4e2002-06-06 18:54:39 +00004768 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00004769 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00004770 ** when tracing REDUCE actions.
4771 */
4772 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drh4ef07702016-03-16 19:45:54 +00004773 assert( rp->iRule==i );
drhc4dd3fd2008-01-22 01:48:05 +00004774 fprintf(out," /* %3d */ \"", i);
4775 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00004776 fprintf(out,"\",\n"); lineno++;
4777 }
4778 tplt_xfer(lemp->name,in,out,&lineno);
4779
drh75897232000-05-29 14:26:00 +00004780 /* Generate code which executes every time a symbol is popped from
drh06f60d82017-04-14 19:46:12 +00004781 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00004782 ** (In other words, generate the %destructor actions)
4783 */
drh75897232000-05-29 14:26:00 +00004784 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00004785 int once = 1;
drh75897232000-05-29 14:26:00 +00004786 for(i=0; i<lemp->nsymbol; i++){
4787 struct symbol *sp = lemp->symbols[i];
4788 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004789 if( once ){
4790 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
4791 once = 0;
4792 }
drhc53eed12009-06-12 17:46:19 +00004793 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00004794 }
4795 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
4796 if( i<lemp->nsymbol ){
4797 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4798 fprintf(out," break;\n"); lineno++;
4799 }
4800 }
drh8d659732005-01-13 23:54:06 +00004801 if( lemp->vardest ){
4802 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00004803 int once = 1;
drh8d659732005-01-13 23:54:06 +00004804 for(i=0; i<lemp->nsymbol; i++){
4805 struct symbol *sp = lemp->symbols[i];
4806 if( sp==0 || sp->type==TERMINAL ||
4807 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004808 if( once ){
drh5c8241b2017-12-24 23:38:10 +00004809 fprintf(out, " /* Default NON-TERMINAL Destructor */\n");lineno++;
drh4dc8ef52008-07-01 17:13:57 +00004810 once = 0;
4811 }
drhc53eed12009-06-12 17:46:19 +00004812 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00004813 dflt_sp = sp;
4814 }
4815 if( dflt_sp!=0 ){
4816 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00004817 }
drh4dc8ef52008-07-01 17:13:57 +00004818 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00004819 }
drh75897232000-05-29 14:26:00 +00004820 for(i=0; i<lemp->nsymbol; i++){
4821 struct symbol *sp = lemp->symbols[i];
4822 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh0f832dd2016-08-16 16:46:40 +00004823 if( sp->destLineno<0 ) continue; /* Already emitted */
drh75013012009-06-12 15:47:34 +00004824 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004825
4826 /* Combine duplicate destructors into a single case */
4827 for(j=i+1; j<lemp->nsymbol; j++){
4828 struct symbol *sp2 = lemp->symbols[j];
4829 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
4830 && sp2->dtnum==sp->dtnum
4831 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00004832 fprintf(out," case %d: /* %s */\n",
4833 sp2->index, sp2->name); lineno++;
drh0f832dd2016-08-16 16:46:40 +00004834 sp2->destLineno = -1; /* Avoid emitting this destructor again */
drh0bb132b2004-07-20 14:06:51 +00004835 }
4836 }
4837
drh75897232000-05-29 14:26:00 +00004838 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4839 fprintf(out," break;\n"); lineno++;
4840 }
drh75897232000-05-29 14:26:00 +00004841 tplt_xfer(lemp->name,in,out,&lineno);
4842
4843 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00004844 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00004845 tplt_xfer(lemp->name,in,out,&lineno);
4846
drhcfc45b12018-12-03 23:57:27 +00004847 /* Generate the tables of rule information. yyRuleInfoLhs[] and
4848 ** yyRuleInfoNRhs[].
drh75897232000-05-29 14:26:00 +00004849 **
4850 ** Note: This code depends on the fact that rules are number
4851 ** sequentually beginning with 0.
4852 */
drh5c8241b2017-12-24 23:38:10 +00004853 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drhcfc45b12018-12-03 23:57:27 +00004854 fprintf(out," %4d, /* (%d) ", rp->lhs->index, i);
4855 rule_print(out, rp);
4856 fprintf(out," */\n"); lineno++;
4857 }
4858 tplt_xfer(lemp->name,in,out,&lineno);
4859 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4860 fprintf(out," %3d, /* (%d) ", -rp->nrhs, i);
drh5c8241b2017-12-24 23:38:10 +00004861 rule_print(out, rp);
4862 fprintf(out," */\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004863 }
4864 tplt_xfer(lemp->name,in,out,&lineno);
4865
4866 /* Generate code which execution during each REDUCE action */
drhdabd04c2016-02-17 01:46:19 +00004867 i = 0;
drh75897232000-05-29 14:26:00 +00004868 for(rp=lemp->rule; rp; rp=rp->next){
drhdabd04c2016-02-17 01:46:19 +00004869 i += translate_code(lemp, rp);
4870 }
4871 if( i ){
4872 fprintf(out," YYMINORTYPE yylhsminor;\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004873 }
drhc53eed12009-06-12 17:46:19 +00004874 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00004875 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00004876 struct rule *rp2; /* Other rules with the same action */
drh711c9812016-05-23 14:24:31 +00004877 if( rp->codeEmitted ) continue;
4878 if( rp->noCode ){
4879 /* No C code actions, so this will be part of the "default:" rule */
drh2e55b042016-04-30 17:19:30 +00004880 continue;
4881 }
drh4ef07702016-03-16 19:45:54 +00004882 fprintf(out," case %d: /* ", rp->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004883 writeRuleText(out, rp);
4884 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004885 for(rp2=rp->next; rp2; rp2=rp2->next){
drhafb8cd92016-04-29 11:28:35 +00004886 if( rp2->code==rp->code && rp2->codePrefix==rp->codePrefix
4887 && rp2->codeSuffix==rp->codeSuffix ){
drh4ef07702016-03-16 19:45:54 +00004888 fprintf(out," case %d: /* ", rp2->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004889 writeRuleText(out, rp2);
drh4ef07702016-03-16 19:45:54 +00004890 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->iRule); lineno++;
drh711c9812016-05-23 14:24:31 +00004891 rp2->codeEmitted = 1;
drh0bb132b2004-07-20 14:06:51 +00004892 }
4893 }
drh75897232000-05-29 14:26:00 +00004894 emit_code(out,rp,lemp,&lineno);
4895 fprintf(out," break;\n"); lineno++;
drh711c9812016-05-23 14:24:31 +00004896 rp->codeEmitted = 1;
drh75897232000-05-29 14:26:00 +00004897 }
drhc53eed12009-06-12 17:46:19 +00004898 /* Finally, output the default: rule. We choose as the default: all
4899 ** empty actions. */
4900 fprintf(out," default:\n"); lineno++;
4901 for(rp=lemp->rule; rp; rp=rp->next){
drh711c9812016-05-23 14:24:31 +00004902 if( rp->codeEmitted ) continue;
4903 assert( rp->noCode );
drh4ef07702016-03-16 19:45:54 +00004904 fprintf(out," /* (%d) ", rp->iRule);
drhc53eed12009-06-12 17:46:19 +00004905 writeRuleText(out, rp);
drhe94006e2019-12-10 20:41:48 +00004906 if( rp->neverReduce ){
4907 fprintf(out, " (NEVER REDUCES) */ assert(yyruleno!=%d);\n",
4908 rp->iRule); lineno++;
4909 }else if( rp->doesReduce ){
drh756b41e2016-05-24 18:55:08 +00004910 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->iRule); lineno++;
4911 }else{
4912 fprintf(out, " (OPTIMIZED OUT) */ assert(yyruleno!=%d);\n",
4913 rp->iRule); lineno++;
4914 }
drhc53eed12009-06-12 17:46:19 +00004915 }
4916 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004917 tplt_xfer(lemp->name,in,out,&lineno);
4918
4919 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004920 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004921 tplt_xfer(lemp->name,in,out,&lineno);
4922
4923 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004924 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004925 tplt_xfer(lemp->name,in,out,&lineno);
4926
4927 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004928 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004929 tplt_xfer(lemp->name,in,out,&lineno);
4930
4931 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004932 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004933
drhe2dcc422019-01-15 14:44:23 +00004934 acttab_free(pActtab);
drh75897232000-05-29 14:26:00 +00004935 fclose(in);
4936 fclose(out);
drhfe03dac2019-11-26 02:22:39 +00004937 if( sql ) fclose(sql);
drh75897232000-05-29 14:26:00 +00004938 return;
4939}
4940
4941/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004942void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004943{
4944 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004945 const char *prefix;
drh75897232000-05-29 14:26:00 +00004946 char line[LINESIZE];
4947 char pattern[LINESIZE];
4948 int i;
4949
4950 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4951 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004952 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004953 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004954 int nextChar;
drh75897232000-05-29 14:26:00 +00004955 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
drh61f92cd2014-01-11 03:06:18 +00004956 lemon_sprintf(pattern,"#define %s%-30s %3d\n",
4957 prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004958 if( strcmp(line,pattern) ) break;
4959 }
drh8ba0d1c2012-06-16 15:26:31 +00004960 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004961 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004962 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004963 /* No change in the file. Don't rewrite it. */
4964 return;
4965 }
4966 }
drh2aa6ca42004-09-10 00:14:04 +00004967 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004968 if( out ){
4969 for(i=1; i<lemp->nterminal; i++){
drh61f92cd2014-01-11 03:06:18 +00004970 fprintf(out,"#define %s%-30s %3d\n",prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004971 }
drh06f60d82017-04-14 19:46:12 +00004972 fclose(out);
drh75897232000-05-29 14:26:00 +00004973 }
4974 return;
4975}
4976
4977/* Reduce the size of the action tables, if possible, by making use
4978** of defaults.
4979**
drhb59499c2002-02-23 18:45:13 +00004980** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004981** it the default. Except, there is no default if the wildcard token
4982** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004983*/
icculus9e44cf12010-02-14 17:14:22 +00004984void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004985{
4986 struct state *stp;
drhc173ad82016-05-23 16:15:02 +00004987 struct action *ap, *ap2, *nextap;
drhb59499c2002-02-23 18:45:13 +00004988 struct rule *rp, *rp2, *rbest;
drh0c6dfaa2015-09-08 21:16:46 +00004989 int nbest, n;
drh75897232000-05-29 14:26:00 +00004990 int i;
drhe09daa92006-06-10 13:29:31 +00004991 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004992
4993 for(i=0; i<lemp->nstate; i++){
4994 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004995 nbest = 0;
4996 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004997 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004998
drhb59499c2002-02-23 18:45:13 +00004999 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00005000 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
5001 usesWildcard = 1;
5002 }
drhb59499c2002-02-23 18:45:13 +00005003 if( ap->type!=REDUCE ) continue;
5004 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00005005 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00005006 if( rp==rbest ) continue;
5007 n = 1;
5008 for(ap2=ap->next; ap2; ap2=ap2->next){
5009 if( ap2->type!=REDUCE ) continue;
5010 rp2 = ap2->x.rp;
5011 if( rp2==rbest ) continue;
5012 if( rp2==rp ) n++;
5013 }
5014 if( n>nbest ){
5015 nbest = n;
5016 rbest = rp;
drh75897232000-05-29 14:26:00 +00005017 }
5018 }
drh06f60d82017-04-14 19:46:12 +00005019
drhb59499c2002-02-23 18:45:13 +00005020 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00005021 ** is not at least 1 or if the wildcard token is a possible
5022 ** lookahead.
5023 */
5024 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00005025
drhb59499c2002-02-23 18:45:13 +00005026
5027 /* Combine matching REDUCE actions into a single default */
5028 for(ap=stp->ap; ap; ap=ap->next){
5029 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
5030 }
drh75897232000-05-29 14:26:00 +00005031 assert( ap );
5032 ap->sp = Symbol_new("{default}");
5033 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00005034 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00005035 }
5036 stp->ap = Action_sort(stp->ap);
drh3bd48ab2015-09-07 18:23:37 +00005037
5038 for(ap=stp->ap; ap; ap=ap->next){
5039 if( ap->type==SHIFT ) break;
5040 if( ap->type==REDUCE && ap->x.rp!=rbest ) break;
5041 }
5042 if( ap==0 ){
5043 stp->autoReduce = 1;
5044 stp->pDfltReduce = rbest;
5045 }
5046 }
5047
5048 /* Make a second pass over all states and actions. Convert
5049 ** every action that is a SHIFT to an autoReduce state into
5050 ** a SHIFTREDUCE action.
5051 */
5052 for(i=0; i<lemp->nstate; i++){
5053 stp = lemp->sorted[i];
5054 for(ap=stp->ap; ap; ap=ap->next){
5055 struct state *pNextState;
5056 if( ap->type!=SHIFT ) continue;
5057 pNextState = ap->x.stp;
5058 if( pNextState->autoReduce && pNextState->pDfltReduce!=0 ){
5059 ap->type = SHIFTREDUCE;
5060 ap->x.rp = pNextState->pDfltReduce;
5061 }
5062 }
drh75897232000-05-29 14:26:00 +00005063 }
drhc173ad82016-05-23 16:15:02 +00005064
5065 /* If a SHIFTREDUCE action specifies a rule that has a single RHS term
5066 ** (meaning that the SHIFTREDUCE will land back in the state where it
5067 ** started) and if there is no C-code associated with the reduce action,
5068 ** then we can go ahead and convert the action to be the same as the
5069 ** action for the RHS of the rule.
5070 */
5071 for(i=0; i<lemp->nstate; i++){
5072 stp = lemp->sorted[i];
5073 for(ap=stp->ap; ap; ap=nextap){
5074 nextap = ap->next;
5075 if( ap->type!=SHIFTREDUCE ) continue;
5076 rp = ap->x.rp;
5077 if( rp->noCode==0 ) continue;
5078 if( rp->nrhs!=1 ) continue;
5079#if 1
5080 /* Only apply this optimization to non-terminals. It would be OK to
5081 ** apply it to terminal symbols too, but that makes the parser tables
5082 ** larger. */
5083 if( ap->sp->index<lemp->nterminal ) continue;
5084#endif
5085 /* If we reach this point, it means the optimization can be applied */
5086 nextap = ap;
5087 for(ap2=stp->ap; ap2 && (ap2==ap || ap2->sp!=rp->lhs); ap2=ap2->next){}
5088 assert( ap2!=0 );
5089 ap->spOpt = ap2->sp;
5090 ap->type = ap2->type;
5091 ap->x = ap2->x;
5092 }
5093 }
drh75897232000-05-29 14:26:00 +00005094}
drhb59499c2002-02-23 18:45:13 +00005095
drhada354d2005-11-05 15:03:59 +00005096
5097/*
5098** Compare two states for sorting purposes. The smaller state is the
5099** one with the most non-terminal actions. If they have the same number
5100** of non-terminal actions, then the smaller is the one with the most
5101** token actions.
5102*/
5103static int stateResortCompare(const void *a, const void *b){
5104 const struct state *pA = *(const struct state**)a;
5105 const struct state *pB = *(const struct state**)b;
5106 int n;
5107
5108 n = pB->nNtAct - pA->nNtAct;
5109 if( n==0 ){
5110 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00005111 if( n==0 ){
5112 n = pB->statenum - pA->statenum;
5113 }
drhada354d2005-11-05 15:03:59 +00005114 }
drhe594bc32009-11-03 13:02:25 +00005115 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00005116 return n;
5117}
5118
5119
5120/*
5121** Renumber and resort states so that states with fewer choices
5122** occur at the end. Except, keep state 0 as the first state.
5123*/
icculus9e44cf12010-02-14 17:14:22 +00005124void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00005125{
5126 int i;
5127 struct state *stp;
5128 struct action *ap;
5129
5130 for(i=0; i<lemp->nstate; i++){
5131 stp = lemp->sorted[i];
5132 stp->nTknAct = stp->nNtAct = 0;
drh5c8241b2017-12-24 23:38:10 +00005133 stp->iDfltReduce = -1; /* Init dflt action to "syntax error" */
drhada354d2005-11-05 15:03:59 +00005134 stp->iTknOfst = NO_OFFSET;
5135 stp->iNtOfst = NO_OFFSET;
5136 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00005137 int iAction = compute_action(lemp,ap);
5138 if( iAction>=0 ){
drhada354d2005-11-05 15:03:59 +00005139 if( ap->sp->index<lemp->nterminal ){
5140 stp->nTknAct++;
5141 }else if( ap->sp->index<lemp->nsymbol ){
5142 stp->nNtAct++;
5143 }else{
drh3bd48ab2015-09-07 18:23:37 +00005144 assert( stp->autoReduce==0 || stp->pDfltReduce==ap->x.rp );
drh5c8241b2017-12-24 23:38:10 +00005145 stp->iDfltReduce = iAction;
drhada354d2005-11-05 15:03:59 +00005146 }
5147 }
5148 }
5149 }
5150 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
5151 stateResortCompare);
5152 for(i=0; i<lemp->nstate; i++){
5153 lemp->sorted[i]->statenum = i;
5154 }
drh3bd48ab2015-09-07 18:23:37 +00005155 lemp->nxstate = lemp->nstate;
5156 while( lemp->nxstate>1 && lemp->sorted[lemp->nxstate-1]->autoReduce ){
5157 lemp->nxstate--;
5158 }
drhada354d2005-11-05 15:03:59 +00005159}
5160
5161
drh75897232000-05-29 14:26:00 +00005162/***************** From the file "set.c" ************************************/
5163/*
5164** Set manipulation routines for the LEMON parser generator.
5165*/
5166
5167static int size = 0;
5168
5169/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00005170void SetSize(int n)
drh75897232000-05-29 14:26:00 +00005171{
5172 size = n+1;
5173}
5174
5175/* Allocate a new set */
drh14d88552017-04-14 19:44:15 +00005176char *SetNew(void){
drh75897232000-05-29 14:26:00 +00005177 char *s;
drh9892c5d2007-12-21 00:02:11 +00005178 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00005179 if( s==0 ){
drh75897232000-05-29 14:26:00 +00005180 memory_error();
5181 }
drh75897232000-05-29 14:26:00 +00005182 return s;
5183}
5184
5185/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00005186void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00005187{
5188 free(s);
5189}
5190
5191/* Add a new element to the set. Return TRUE if the element was added
5192** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00005193int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00005194{
5195 int rv;
drh9892c5d2007-12-21 00:02:11 +00005196 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00005197 rv = s[e];
5198 s[e] = 1;
5199 return !rv;
5200}
5201
5202/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00005203int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00005204{
5205 int i, progress;
5206 progress = 0;
5207 for(i=0; i<size; i++){
5208 if( s2[i]==0 ) continue;
5209 if( s1[i]==0 ){
5210 progress = 1;
5211 s1[i] = 1;
5212 }
5213 }
5214 return progress;
5215}
5216/********************** From the file "table.c" ****************************/
5217/*
5218** All code in this file has been automatically generated
5219** from a specification in the file
5220** "table.q"
5221** by the associative array code building program "aagen".
5222** Do not edit this file! Instead, edit the specification
5223** file, then rerun aagen.
5224*/
5225/*
5226** Code for processing tables in the LEMON parser generator.
5227*/
5228
drh01f75f22013-10-02 20:46:30 +00005229PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00005230{
drh01f75f22013-10-02 20:46:30 +00005231 unsigned h = 0;
5232 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00005233 return h;
5234}
5235
5236/* Works like strdup, sort of. Save a string in malloced memory, but
5237** keep strings in a table so that the same string is not in more
5238** than one place.
5239*/
icculus9e44cf12010-02-14 17:14:22 +00005240const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00005241{
icculus9e44cf12010-02-14 17:14:22 +00005242 const char *z;
5243 char *cpy;
drh75897232000-05-29 14:26:00 +00005244
drh916f75f2006-07-17 00:19:39 +00005245 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00005246 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00005247 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
drh898799f2014-01-10 23:21:00 +00005248 lemon_strcpy(cpy,y);
icculus9e44cf12010-02-14 17:14:22 +00005249 z = cpy;
drh75897232000-05-29 14:26:00 +00005250 Strsafe_insert(z);
5251 }
5252 MemoryCheck(z);
5253 return z;
5254}
5255
5256/* There is one instance of the following structure for each
5257** associative array of type "x1".
5258*/
5259struct s_x1 {
5260 int size; /* The number of available slots. */
5261 /* Must be a power of 2 greater than or */
5262 /* equal to 1 */
5263 int count; /* Number of currently slots filled */
5264 struct s_x1node *tbl; /* The data stored here */
5265 struct s_x1node **ht; /* Hash table for lookups */
5266};
5267
5268/* There is one instance of this structure for every data element
5269** in an associative array of type "x1".
5270*/
5271typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00005272 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00005273 struct s_x1node *next; /* Next entry with the same hash */
5274 struct s_x1node **from; /* Previous link */
5275} x1node;
5276
5277/* There is only one instance of the array, which is the following */
5278static struct s_x1 *x1a;
5279
5280/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005281void Strsafe_init(void){
drh75897232000-05-29 14:26:00 +00005282 if( x1a ) return;
5283 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
5284 if( x1a ){
5285 x1a->size = 1024;
5286 x1a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005287 x1a->tbl = (x1node*)calloc(1024, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005288 if( x1a->tbl==0 ){
5289 free(x1a);
5290 x1a = 0;
5291 }else{
5292 int i;
5293 x1a->ht = (x1node**)&(x1a->tbl[1024]);
5294 for(i=0; i<1024; i++) x1a->ht[i] = 0;
5295 }
5296 }
5297}
5298/* Insert a new record into the array. Return TRUE if successful.
5299** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005300int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00005301{
5302 x1node *np;
drh01f75f22013-10-02 20:46:30 +00005303 unsigned h;
5304 unsigned ph;
drh75897232000-05-29 14:26:00 +00005305
5306 if( x1a==0 ) return 0;
5307 ph = strhash(data);
5308 h = ph & (x1a->size-1);
5309 np = x1a->ht[h];
5310 while( np ){
5311 if( strcmp(np->data,data)==0 ){
5312 /* An existing entry with the same key is found. */
5313 /* Fail because overwrite is not allows. */
5314 return 0;
5315 }
5316 np = np->next;
5317 }
5318 if( x1a->count>=x1a->size ){
5319 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005320 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005321 struct s_x1 array;
mistachkin8e189222015-04-19 21:43:16 +00005322 array.size = arrSize = x1a->size*2;
drh75897232000-05-29 14:26:00 +00005323 array.count = x1a->count;
mistachkin8e189222015-04-19 21:43:16 +00005324 array.tbl = (x1node*)calloc(arrSize, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005325 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005326 array.ht = (x1node**)&(array.tbl[arrSize]);
5327 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005328 for(i=0; i<x1a->count; i++){
5329 x1node *oldnp, *newnp;
5330 oldnp = &(x1a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005331 h = strhash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005332 newnp = &(array.tbl[i]);
5333 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5334 newnp->next = array.ht[h];
5335 newnp->data = oldnp->data;
5336 newnp->from = &(array.ht[h]);
5337 array.ht[h] = newnp;
5338 }
5339 free(x1a->tbl);
5340 *x1a = array;
5341 }
5342 /* Insert the new data */
5343 h = ph & (x1a->size-1);
5344 np = &(x1a->tbl[x1a->count++]);
5345 np->data = data;
5346 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
5347 np->next = x1a->ht[h];
5348 x1a->ht[h] = np;
5349 np->from = &(x1a->ht[h]);
5350 return 1;
5351}
5352
5353/* Return a pointer to data assigned to the given key. Return NULL
5354** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005355const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00005356{
drh01f75f22013-10-02 20:46:30 +00005357 unsigned h;
drh75897232000-05-29 14:26:00 +00005358 x1node *np;
5359
5360 if( x1a==0 ) return 0;
5361 h = strhash(key) & (x1a->size-1);
5362 np = x1a->ht[h];
5363 while( np ){
5364 if( strcmp(np->data,key)==0 ) break;
5365 np = np->next;
5366 }
5367 return np ? np->data : 0;
5368}
5369
5370/* Return a pointer to the (terminal or nonterminal) symbol "x".
5371** Create a new symbol if this is the first time "x" has been seen.
5372*/
icculus9e44cf12010-02-14 17:14:22 +00005373struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00005374{
5375 struct symbol *sp;
5376
5377 sp = Symbol_find(x);
5378 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00005379 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00005380 MemoryCheck(sp);
5381 sp->name = Strsafe(x);
drhc56fac72015-10-29 13:48:15 +00005382 sp->type = ISUPPER(*x) ? TERMINAL : NONTERMINAL;
drh75897232000-05-29 14:26:00 +00005383 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00005384 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00005385 sp->prec = -1;
5386 sp->assoc = UNK;
5387 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00005388 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00005389 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00005390 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00005391 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00005392 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00005393 Symbol_insert(sp,sp->name);
5394 }
drhc4dd3fd2008-01-22 01:48:05 +00005395 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00005396 return sp;
5397}
5398
drh61f92cd2014-01-11 03:06:18 +00005399/* Compare two symbols for sorting purposes. Return negative,
5400** zero, or positive if a is less then, equal to, or greater
5401** than b.
drh60d31652004-02-22 00:08:04 +00005402**
5403** Symbols that begin with upper case letters (terminals or tokens)
5404** must sort before symbols that begin with lower case letters
drh61f92cd2014-01-11 03:06:18 +00005405** (non-terminals). And MULTITERMINAL symbols (created using the
5406** %token_class directive) must sort at the very end. Other than
5407** that, the order does not matter.
drh60d31652004-02-22 00:08:04 +00005408**
5409** We find experimentally that leaving the symbols in their original
5410** order (the order they appeared in the grammar file) gives the
5411** smallest parser tables in SQLite.
5412*/
icculus9e44cf12010-02-14 17:14:22 +00005413int Symbolcmpp(const void *_a, const void *_b)
5414{
drh61f92cd2014-01-11 03:06:18 +00005415 const struct symbol *a = *(const struct symbol **) _a;
5416 const struct symbol *b = *(const struct symbol **) _b;
5417 int i1 = a->type==MULTITERMINAL ? 3 : a->name[0]>'Z' ? 2 : 1;
5418 int i2 = b->type==MULTITERMINAL ? 3 : b->name[0]>'Z' ? 2 : 1;
5419 return i1==i2 ? a->index - b->index : i1 - i2;
drh75897232000-05-29 14:26:00 +00005420}
5421
5422/* There is one instance of the following structure for each
5423** associative array of type "x2".
5424*/
5425struct s_x2 {
5426 int size; /* The number of available slots. */
5427 /* Must be a power of 2 greater than or */
5428 /* equal to 1 */
5429 int count; /* Number of currently slots filled */
5430 struct s_x2node *tbl; /* The data stored here */
5431 struct s_x2node **ht; /* Hash table for lookups */
5432};
5433
5434/* There is one instance of this structure for every data element
5435** in an associative array of type "x2".
5436*/
5437typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00005438 struct symbol *data; /* The data */
5439 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00005440 struct s_x2node *next; /* Next entry with the same hash */
5441 struct s_x2node **from; /* Previous link */
5442} x2node;
5443
5444/* There is only one instance of the array, which is the following */
5445static struct s_x2 *x2a;
5446
5447/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005448void Symbol_init(void){
drh75897232000-05-29 14:26:00 +00005449 if( x2a ) return;
5450 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
5451 if( x2a ){
5452 x2a->size = 128;
5453 x2a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005454 x2a->tbl = (x2node*)calloc(128, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005455 if( x2a->tbl==0 ){
5456 free(x2a);
5457 x2a = 0;
5458 }else{
5459 int i;
5460 x2a->ht = (x2node**)&(x2a->tbl[128]);
5461 for(i=0; i<128; i++) x2a->ht[i] = 0;
5462 }
5463 }
5464}
5465/* Insert a new record into the array. Return TRUE if successful.
5466** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005467int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00005468{
5469 x2node *np;
drh01f75f22013-10-02 20:46:30 +00005470 unsigned h;
5471 unsigned ph;
drh75897232000-05-29 14:26:00 +00005472
5473 if( x2a==0 ) return 0;
5474 ph = strhash(key);
5475 h = ph & (x2a->size-1);
5476 np = x2a->ht[h];
5477 while( np ){
5478 if( strcmp(np->key,key)==0 ){
5479 /* An existing entry with the same key is found. */
5480 /* Fail because overwrite is not allows. */
5481 return 0;
5482 }
5483 np = np->next;
5484 }
5485 if( x2a->count>=x2a->size ){
5486 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005487 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005488 struct s_x2 array;
mistachkin8e189222015-04-19 21:43:16 +00005489 array.size = arrSize = x2a->size*2;
drh75897232000-05-29 14:26:00 +00005490 array.count = x2a->count;
mistachkin8e189222015-04-19 21:43:16 +00005491 array.tbl = (x2node*)calloc(arrSize, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005492 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005493 array.ht = (x2node**)&(array.tbl[arrSize]);
5494 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005495 for(i=0; i<x2a->count; i++){
5496 x2node *oldnp, *newnp;
5497 oldnp = &(x2a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005498 h = strhash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005499 newnp = &(array.tbl[i]);
5500 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5501 newnp->next = array.ht[h];
5502 newnp->key = oldnp->key;
5503 newnp->data = oldnp->data;
5504 newnp->from = &(array.ht[h]);
5505 array.ht[h] = newnp;
5506 }
5507 free(x2a->tbl);
5508 *x2a = array;
5509 }
5510 /* Insert the new data */
5511 h = ph & (x2a->size-1);
5512 np = &(x2a->tbl[x2a->count++]);
5513 np->key = key;
5514 np->data = data;
5515 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
5516 np->next = x2a->ht[h];
5517 x2a->ht[h] = np;
5518 np->from = &(x2a->ht[h]);
5519 return 1;
5520}
5521
5522/* Return a pointer to data assigned to the given key. Return NULL
5523** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005524struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00005525{
drh01f75f22013-10-02 20:46:30 +00005526 unsigned h;
drh75897232000-05-29 14:26:00 +00005527 x2node *np;
5528
5529 if( x2a==0 ) return 0;
5530 h = strhash(key) & (x2a->size-1);
5531 np = x2a->ht[h];
5532 while( np ){
5533 if( strcmp(np->key,key)==0 ) break;
5534 np = np->next;
5535 }
5536 return np ? np->data : 0;
5537}
5538
5539/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00005540struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00005541{
5542 struct symbol *data;
5543 if( x2a && n>0 && n<=x2a->count ){
5544 data = x2a->tbl[n-1].data;
5545 }else{
5546 data = 0;
5547 }
5548 return data;
5549}
5550
5551/* Return the size of the array */
5552int Symbol_count()
5553{
5554 return x2a ? x2a->count : 0;
5555}
5556
5557/* Return an array of pointers to all data in the table.
5558** The array is obtained from malloc. Return NULL if memory allocation
5559** problems, or if the array is empty. */
5560struct symbol **Symbol_arrayof()
5561{
5562 struct symbol **array;
mistachkin8e189222015-04-19 21:43:16 +00005563 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005564 if( x2a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005565 arrSize = x2a->count;
5566 array = (struct symbol **)calloc(arrSize, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00005567 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005568 for(i=0; i<arrSize; i++) array[i] = x2a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005569 }
5570 return array;
5571}
5572
5573/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00005574int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00005575{
icculus9e44cf12010-02-14 17:14:22 +00005576 const struct config *a = (struct config *) _a;
5577 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00005578 int x;
5579 x = a->rp->index - b->rp->index;
5580 if( x==0 ) x = a->dot - b->dot;
5581 return x;
5582}
5583
5584/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00005585PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00005586{
5587 int rc;
5588 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
5589 rc = a->rp->index - b->rp->index;
5590 if( rc==0 ) rc = a->dot - b->dot;
5591 }
5592 if( rc==0 ){
5593 if( a ) rc = 1;
5594 if( b ) rc = -1;
5595 }
5596 return rc;
5597}
5598
5599/* Hash a state */
drh01f75f22013-10-02 20:46:30 +00005600PRIVATE unsigned statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00005601{
drh01f75f22013-10-02 20:46:30 +00005602 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005603 while( a ){
5604 h = h*571 + a->rp->index*37 + a->dot;
5605 a = a->bp;
5606 }
5607 return h;
5608}
5609
5610/* Allocate a new state structure */
5611struct state *State_new()
5612{
icculus9e44cf12010-02-14 17:14:22 +00005613 struct state *newstate;
5614 newstate = (struct state *)calloc(1, sizeof(struct state) );
5615 MemoryCheck(newstate);
5616 return newstate;
drh75897232000-05-29 14:26:00 +00005617}
5618
5619/* There is one instance of the following structure for each
5620** associative array of type "x3".
5621*/
5622struct s_x3 {
5623 int size; /* The number of available slots. */
5624 /* Must be a power of 2 greater than or */
5625 /* equal to 1 */
5626 int count; /* Number of currently slots filled */
5627 struct s_x3node *tbl; /* The data stored here */
5628 struct s_x3node **ht; /* Hash table for lookups */
5629};
5630
5631/* There is one instance of this structure for every data element
5632** in an associative array of type "x3".
5633*/
5634typedef struct s_x3node {
5635 struct state *data; /* The data */
5636 struct config *key; /* The key */
5637 struct s_x3node *next; /* Next entry with the same hash */
5638 struct s_x3node **from; /* Previous link */
5639} x3node;
5640
5641/* There is only one instance of the array, which is the following */
5642static struct s_x3 *x3a;
5643
5644/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005645void State_init(void){
drh75897232000-05-29 14:26:00 +00005646 if( x3a ) return;
5647 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
5648 if( x3a ){
5649 x3a->size = 128;
5650 x3a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005651 x3a->tbl = (x3node*)calloc(128, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005652 if( x3a->tbl==0 ){
5653 free(x3a);
5654 x3a = 0;
5655 }else{
5656 int i;
5657 x3a->ht = (x3node**)&(x3a->tbl[128]);
5658 for(i=0; i<128; i++) x3a->ht[i] = 0;
5659 }
5660 }
5661}
5662/* Insert a new record into the array. Return TRUE if successful.
5663** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005664int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00005665{
5666 x3node *np;
drh01f75f22013-10-02 20:46:30 +00005667 unsigned h;
5668 unsigned ph;
drh75897232000-05-29 14:26:00 +00005669
5670 if( x3a==0 ) return 0;
5671 ph = statehash(key);
5672 h = ph & (x3a->size-1);
5673 np = x3a->ht[h];
5674 while( np ){
5675 if( statecmp(np->key,key)==0 ){
5676 /* An existing entry with the same key is found. */
5677 /* Fail because overwrite is not allows. */
5678 return 0;
5679 }
5680 np = np->next;
5681 }
5682 if( x3a->count>=x3a->size ){
5683 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005684 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005685 struct s_x3 array;
mistachkin8e189222015-04-19 21:43:16 +00005686 array.size = arrSize = x3a->size*2;
drh75897232000-05-29 14:26:00 +00005687 array.count = x3a->count;
mistachkin8e189222015-04-19 21:43:16 +00005688 array.tbl = (x3node*)calloc(arrSize, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005689 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005690 array.ht = (x3node**)&(array.tbl[arrSize]);
5691 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005692 for(i=0; i<x3a->count; i++){
5693 x3node *oldnp, *newnp;
5694 oldnp = &(x3a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005695 h = statehash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005696 newnp = &(array.tbl[i]);
5697 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5698 newnp->next = array.ht[h];
5699 newnp->key = oldnp->key;
5700 newnp->data = oldnp->data;
5701 newnp->from = &(array.ht[h]);
5702 array.ht[h] = newnp;
5703 }
5704 free(x3a->tbl);
5705 *x3a = array;
5706 }
5707 /* Insert the new data */
5708 h = ph & (x3a->size-1);
5709 np = &(x3a->tbl[x3a->count++]);
5710 np->key = key;
5711 np->data = data;
5712 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
5713 np->next = x3a->ht[h];
5714 x3a->ht[h] = np;
5715 np->from = &(x3a->ht[h]);
5716 return 1;
5717}
5718
5719/* Return a pointer to data assigned to the given key. Return NULL
5720** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005721struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005722{
drh01f75f22013-10-02 20:46:30 +00005723 unsigned h;
drh75897232000-05-29 14:26:00 +00005724 x3node *np;
5725
5726 if( x3a==0 ) return 0;
5727 h = statehash(key) & (x3a->size-1);
5728 np = x3a->ht[h];
5729 while( np ){
5730 if( statecmp(np->key,key)==0 ) break;
5731 np = np->next;
5732 }
5733 return np ? np->data : 0;
5734}
5735
5736/* Return an array of pointers to all data in the table.
5737** The array is obtained from malloc. Return NULL if memory allocation
5738** problems, or if the array is empty. */
drh14d88552017-04-14 19:44:15 +00005739struct state **State_arrayof(void)
drh75897232000-05-29 14:26:00 +00005740{
5741 struct state **array;
mistachkin8e189222015-04-19 21:43:16 +00005742 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005743 if( x3a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005744 arrSize = x3a->count;
5745 array = (struct state **)calloc(arrSize, sizeof(struct state *));
drh75897232000-05-29 14:26:00 +00005746 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005747 for(i=0; i<arrSize; i++) array[i] = x3a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005748 }
5749 return array;
5750}
5751
5752/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00005753PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00005754{
drh01f75f22013-10-02 20:46:30 +00005755 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005756 h = h*571 + a->rp->index*37 + a->dot;
5757 return h;
5758}
5759
5760/* There is one instance of the following structure for each
5761** associative array of type "x4".
5762*/
5763struct s_x4 {
5764 int size; /* The number of available slots. */
5765 /* Must be a power of 2 greater than or */
5766 /* equal to 1 */
5767 int count; /* Number of currently slots filled */
5768 struct s_x4node *tbl; /* The data stored here */
5769 struct s_x4node **ht; /* Hash table for lookups */
5770};
5771
5772/* There is one instance of this structure for every data element
5773** in an associative array of type "x4".
5774*/
5775typedef struct s_x4node {
5776 struct config *data; /* The data */
5777 struct s_x4node *next; /* Next entry with the same hash */
5778 struct s_x4node **from; /* Previous link */
5779} x4node;
5780
5781/* There is only one instance of the array, which is the following */
5782static struct s_x4 *x4a;
5783
5784/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005785void Configtable_init(void){
drh75897232000-05-29 14:26:00 +00005786 if( x4a ) return;
5787 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
5788 if( x4a ){
5789 x4a->size = 64;
5790 x4a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005791 x4a->tbl = (x4node*)calloc(64, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005792 if( x4a->tbl==0 ){
5793 free(x4a);
5794 x4a = 0;
5795 }else{
5796 int i;
5797 x4a->ht = (x4node**)&(x4a->tbl[64]);
5798 for(i=0; i<64; i++) x4a->ht[i] = 0;
5799 }
5800 }
5801}
5802/* Insert a new record into the array. Return TRUE if successful.
5803** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005804int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00005805{
5806 x4node *np;
drh01f75f22013-10-02 20:46:30 +00005807 unsigned h;
5808 unsigned ph;
drh75897232000-05-29 14:26:00 +00005809
5810 if( x4a==0 ) return 0;
5811 ph = confighash(data);
5812 h = ph & (x4a->size-1);
5813 np = x4a->ht[h];
5814 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005815 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00005816 /* An existing entry with the same key is found. */
5817 /* Fail because overwrite is not allows. */
5818 return 0;
5819 }
5820 np = np->next;
5821 }
5822 if( x4a->count>=x4a->size ){
5823 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005824 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005825 struct s_x4 array;
mistachkin8e189222015-04-19 21:43:16 +00005826 array.size = arrSize = x4a->size*2;
drh75897232000-05-29 14:26:00 +00005827 array.count = x4a->count;
mistachkin8e189222015-04-19 21:43:16 +00005828 array.tbl = (x4node*)calloc(arrSize, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005829 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005830 array.ht = (x4node**)&(array.tbl[arrSize]);
5831 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005832 for(i=0; i<x4a->count; i++){
5833 x4node *oldnp, *newnp;
5834 oldnp = &(x4a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005835 h = confighash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005836 newnp = &(array.tbl[i]);
5837 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5838 newnp->next = array.ht[h];
5839 newnp->data = oldnp->data;
5840 newnp->from = &(array.ht[h]);
5841 array.ht[h] = newnp;
5842 }
5843 free(x4a->tbl);
5844 *x4a = array;
5845 }
5846 /* Insert the new data */
5847 h = ph & (x4a->size-1);
5848 np = &(x4a->tbl[x4a->count++]);
5849 np->data = data;
5850 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
5851 np->next = x4a->ht[h];
5852 x4a->ht[h] = np;
5853 np->from = &(x4a->ht[h]);
5854 return 1;
5855}
5856
5857/* Return a pointer to data assigned to the given key. Return NULL
5858** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005859struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005860{
5861 int h;
5862 x4node *np;
5863
5864 if( x4a==0 ) return 0;
5865 h = confighash(key) & (x4a->size-1);
5866 np = x4a->ht[h];
5867 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005868 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00005869 np = np->next;
5870 }
5871 return np ? np->data : 0;
5872}
5873
5874/* Remove all data from the table. Pass each data to the function "f"
5875** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00005876void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00005877{
5878 int i;
5879 if( x4a==0 || x4a->count==0 ) return;
5880 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
5881 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
5882 x4a->count = 0;
5883 return;
5884}