blob: 4416f67bb2ee2e097100b6eb05d1860f42484096 [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 *);
221void ReportTable(struct lemon *, int);
222void 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 */
drh711c9812016-05-23 14:24:31 +0000295 int noCode; /* True if this rule has no associated C code */
296 int codeEmitted; /* True if the code has been emitted already */
drh75897232000-05-29 14:26:00 +0000297 struct symbol *precsym; /* Precedence symbol for this rule */
298 int index; /* An index number for this rule */
drh4ef07702016-03-16 19:45:54 +0000299 int iRule; /* Rule number as used in the generated tables */
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 */
drh75897232000-05-29 14:26:00 +0000302 struct rule *nextlhs; /* Next rule with the same LHS */
303 struct rule *next; /* Next rule in the global list */
304};
305
306/* A configuration is a production rule of the grammar together with
307** a mark (dot) showing how much of that rule has been processed so far.
308** Configurations also contain a follow-set which is a list of terminal
309** symbols which are allowed to immediately follow the end of the rule.
310** Every configuration is recorded as an instance of the following: */
icculus9e44cf12010-02-14 17:14:22 +0000311enum cfgstatus {
312 COMPLETE,
313 INCOMPLETE
314};
drh75897232000-05-29 14:26:00 +0000315struct config {
316 struct rule *rp; /* The rule upon which the configuration is based */
317 int dot; /* The parse point */
318 char *fws; /* Follow-set for this configuration only */
319 struct plink *fplp; /* Follow-set forward propagation links */
320 struct plink *bplp; /* Follow-set backwards propagation links */
321 struct state *stp; /* Pointer to state which contains this */
icculus9e44cf12010-02-14 17:14:22 +0000322 enum cfgstatus status; /* used during followset and shift computations */
drh75897232000-05-29 14:26:00 +0000323 struct config *next; /* Next configuration in the state */
324 struct config *bp; /* The next basis configuration */
325};
326
icculus9e44cf12010-02-14 17:14:22 +0000327enum e_action {
328 SHIFT,
329 ACCEPT,
330 REDUCE,
331 ERROR,
332 SSCONFLICT, /* A shift/shift conflict */
333 SRCONFLICT, /* Was a reduce, but part of a conflict */
334 RRCONFLICT, /* Was a reduce, but part of a conflict */
335 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
336 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
drh3bd48ab2015-09-07 18:23:37 +0000337 NOT_USED, /* Deleted by compression */
338 SHIFTREDUCE /* Shift first, then reduce */
icculus9e44cf12010-02-14 17:14:22 +0000339};
340
drh75897232000-05-29 14:26:00 +0000341/* Every shift or reduce operation is stored as one of the following */
342struct action {
343 struct symbol *sp; /* The look-ahead symbol */
icculus9e44cf12010-02-14 17:14:22 +0000344 enum e_action type;
drh75897232000-05-29 14:26:00 +0000345 union {
346 struct state *stp; /* The new state, if a shift */
347 struct rule *rp; /* The rule, if a reduce */
348 } x;
drhc173ad82016-05-23 16:15:02 +0000349 struct symbol *spOpt; /* SHIFTREDUCE optimization to this symbol */
drh75897232000-05-29 14:26:00 +0000350 struct action *next; /* Next action for this state */
351 struct action *collide; /* Next action with the same hash */
352};
353
354/* Each state of the generated parser's finite state machine
355** is encoded as an instance of the following structure. */
356struct state {
357 struct config *bp; /* The basis configurations for this state */
358 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000359 int statenum; /* Sequential number for this state */
drh711c9812016-05-23 14:24:31 +0000360 struct action *ap; /* List of actions for this state */
drh8b582012003-10-21 13:16:03 +0000361 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
362 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
drh3bd48ab2015-09-07 18:23:37 +0000363 int iDfltReduce; /* Default action is to REDUCE by this rule */
364 struct rule *pDfltReduce;/* The default REDUCE rule. */
365 int autoReduce; /* True if this is an auto-reduce state */
drh75897232000-05-29 14:26:00 +0000366};
drh8b582012003-10-21 13:16:03 +0000367#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000368
369/* A followset propagation link indicates that the contents of one
370** configuration followset should be propagated to another whenever
371** the first changes. */
372struct plink {
373 struct config *cfp; /* The configuration to which linked */
374 struct plink *next; /* The next propagate link */
375};
376
377/* The state vector for the entire parser generator is recorded as
378** follows. (LEMON uses no global variables and makes little use of
379** static variables. Fields in the following structure can be thought
380** of as begin global variables in the program.) */
381struct lemon {
382 struct state **sorted; /* Table of states sorted by state number */
383 struct rule *rule; /* List of all rules */
drh4ef07702016-03-16 19:45:54 +0000384 struct rule *startRule; /* First rule */
drh75897232000-05-29 14:26:00 +0000385 int nstate; /* Number of states */
drh3bd48ab2015-09-07 18:23:37 +0000386 int nxstate; /* nstate with tail degenerate states removed */
drh75897232000-05-29 14:26:00 +0000387 int nrule; /* Number of rules */
388 int nsymbol; /* Number of terminal and nonterminal symbols */
389 int nterminal; /* Number of terminal symbols */
drh5c8241b2017-12-24 23:38:10 +0000390 int minShiftReduce; /* Minimum shift-reduce action value */
391 int errAction; /* Error action value */
392 int accAction; /* Accept action value */
393 int noAction; /* No-op action value */
394 int minReduce; /* Minimum reduce action */
395 int maxAction; /* Maximum action value of any kind */
drh75897232000-05-29 14:26:00 +0000396 struct symbol **symbols; /* Sorted array of pointers to symbols */
397 int errorcnt; /* Number of errors */
398 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000399 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000400 char *name; /* Name of the generated parser */
401 char *arg; /* Declaration of the 3th argument to parser */
drhfb32c442018-04-21 13:51:42 +0000402 char *ctx; /* Declaration of 2nd argument to constructor */
drh75897232000-05-29 14:26:00 +0000403 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000404 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000405 char *start; /* Name of the start symbol for the grammar */
406 char *stacksize; /* Size of the parser stack */
407 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000408 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000409 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000410 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000411 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000412 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000413 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000414 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000415 char *filename; /* Name of the input file */
416 char *outname; /* Name of the current output file */
417 char *tokenprefix; /* A prefix added to token names in the .h file */
418 int nconflict; /* Number of parsing conflicts */
drhc75e0162015-09-07 02:23:02 +0000419 int nactiontab; /* Number of entries in the yy_action[] table */
drh3a9d6c72017-12-25 04:15:38 +0000420 int nlookaheadtab; /* Number of entries in yy_lookahead[] */
drhc75e0162015-09-07 02:23:02 +0000421 int tablesize; /* Total table size of all tables in bytes */
drh75897232000-05-29 14:26:00 +0000422 int basisflag; /* Print only basis configurations */
drh34ff57b2008-07-14 12:27:51 +0000423 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000424 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000425 char *argv0; /* Name of the program */
426};
427
428#define MemoryCheck(X) if((X)==0){ \
429 extern void memory_error(); \
430 memory_error(); \
431}
432
433/**************** From the file "table.h" *********************************/
434/*
435** All code in this file has been automatically generated
436** from a specification in the file
437** "table.q"
438** by the associative array code building program "aagen".
439** Do not edit this file! Instead, edit the specification
440** file, then rerun aagen.
441*/
442/*
443** Code for processing tables in the LEMON parser generator.
444*/
drh75897232000-05-29 14:26:00 +0000445/* Routines for handling a strings */
446
icculus9e44cf12010-02-14 17:14:22 +0000447const char *Strsafe(const char *);
drh75897232000-05-29 14:26:00 +0000448
icculus9e44cf12010-02-14 17:14:22 +0000449void Strsafe_init(void);
450int Strsafe_insert(const char *);
451const char *Strsafe_find(const char *);
drh75897232000-05-29 14:26:00 +0000452
453/* Routines for handling symbols of the grammar */
454
icculus9e44cf12010-02-14 17:14:22 +0000455struct symbol *Symbol_new(const char *);
456int Symbolcmpp(const void *, const void *);
457void Symbol_init(void);
458int Symbol_insert(struct symbol *, const char *);
459struct symbol *Symbol_find(const char *);
460struct symbol *Symbol_Nth(int);
461int Symbol_count(void);
462struct symbol **Symbol_arrayof(void);
drh75897232000-05-29 14:26:00 +0000463
464/* Routines to manage the state table */
465
icculus9e44cf12010-02-14 17:14:22 +0000466int Configcmp(const char *, const char *);
467struct state *State_new(void);
468void State_init(void);
469int State_insert(struct state *, struct config *);
470struct state *State_find(struct config *);
drh14d88552017-04-14 19:44:15 +0000471struct state **State_arrayof(void);
drh75897232000-05-29 14:26:00 +0000472
473/* Routines used for efficiency in Configlist_add */
474
icculus9e44cf12010-02-14 17:14:22 +0000475void Configtable_init(void);
476int Configtable_insert(struct config *);
477struct config *Configtable_find(struct config *);
478void Configtable_clear(int(*)(struct config *));
479
drh75897232000-05-29 14:26:00 +0000480/****************** From the file "action.c" *******************************/
481/*
482** Routines processing parser actions in the LEMON parser generator.
483*/
484
485/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000486static struct action *Action_new(void){
mistachkind9bc6e82019-05-10 16:16:19 +0000487 static struct action *actionfreelist = 0;
icculus9e44cf12010-02-14 17:14:22 +0000488 struct action *newaction;
drh75897232000-05-29 14:26:00 +0000489
mistachkind9bc6e82019-05-10 16:16:19 +0000490 if( actionfreelist==0 ){
drh75897232000-05-29 14:26:00 +0000491 int i;
492 int amt = 100;
mistachkind9bc6e82019-05-10 16:16:19 +0000493 actionfreelist = (struct action *)calloc(amt, sizeof(struct action));
494 if( actionfreelist==0 ){
drh75897232000-05-29 14:26:00 +0000495 fprintf(stderr,"Unable to allocate memory for a new parser action.");
496 exit(1);
497 }
mistachkind9bc6e82019-05-10 16:16:19 +0000498 for(i=0; i<amt-1; i++) actionfreelist[i].next = &actionfreelist[i+1];
499 actionfreelist[amt-1].next = 0;
drh75897232000-05-29 14:26:00 +0000500 }
mistachkind9bc6e82019-05-10 16:16:19 +0000501 newaction = actionfreelist;
502 actionfreelist = actionfreelist->next;
icculus9e44cf12010-02-14 17:14:22 +0000503 return newaction;
drh75897232000-05-29 14:26:00 +0000504}
505
drhe9278182007-07-18 18:16:29 +0000506/* Compare two actions for sorting purposes. Return negative, zero, or
507** positive if the first action is less than, equal to, or greater than
508** the first
509*/
510static int actioncmp(
511 struct action *ap1,
512 struct action *ap2
513){
drh75897232000-05-29 14:26:00 +0000514 int rc;
515 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000516 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000517 rc = (int)ap1->type - (int)ap2->type;
518 }
drh3bd48ab2015-09-07 18:23:37 +0000519 if( rc==0 && (ap1->type==REDUCE || ap1->type==SHIFTREDUCE) ){
drh75897232000-05-29 14:26:00 +0000520 rc = ap1->x.rp->index - ap2->x.rp->index;
521 }
drhe594bc32009-11-03 13:02:25 +0000522 if( rc==0 ){
icculus7b429aa2010-03-03 17:09:01 +0000523 rc = (int) (ap2 - ap1);
drhe594bc32009-11-03 13:02:25 +0000524 }
drh75897232000-05-29 14:26:00 +0000525 return rc;
526}
527
528/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000529static struct action *Action_sort(
530 struct action *ap
531){
532 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
533 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000534 return ap;
535}
536
icculus9e44cf12010-02-14 17:14:22 +0000537void Action_add(
538 struct action **app,
539 enum e_action type,
540 struct symbol *sp,
541 char *arg
542){
543 struct action *newaction;
544 newaction = Action_new();
545 newaction->next = *app;
546 *app = newaction;
547 newaction->type = type;
548 newaction->sp = sp;
drhc173ad82016-05-23 16:15:02 +0000549 newaction->spOpt = 0;
drh75897232000-05-29 14:26:00 +0000550 if( type==SHIFT ){
icculus9e44cf12010-02-14 17:14:22 +0000551 newaction->x.stp = (struct state *)arg;
drh75897232000-05-29 14:26:00 +0000552 }else{
icculus9e44cf12010-02-14 17:14:22 +0000553 newaction->x.rp = (struct rule *)arg;
drh75897232000-05-29 14:26:00 +0000554 }
555}
drh8b582012003-10-21 13:16:03 +0000556/********************** New code to implement the "acttab" module ***********/
557/*
558** This module implements routines use to construct the yy_action[] table.
559*/
560
561/*
562** The state of the yy_action table under construction is an instance of
drh8dc3e8f2010-01-07 03:53:03 +0000563** the following structure.
564**
565** The yy_action table maps the pair (state_number, lookahead) into an
566** action_number. The table is an array of integers pairs. The state_number
567** determines an initial offset into the yy_action array. The lookahead
568** value is then added to this initial offset to get an index X into the
569** yy_action array. If the aAction[X].lookahead equals the value of the
570** of the lookahead input, then the value of the action_number output is
571** aAction[X].action. If the lookaheads do not match then the
572** default action for the state_number is returned.
573**
574** All actions associated with a single state_number are first entered
drh06f60d82017-04-14 19:46:12 +0000575** into aLookahead[] using multiple calls to acttab_action(). Then the
576** actions for that single state_number are placed into the aAction[]
drh8dc3e8f2010-01-07 03:53:03 +0000577** array with a single call to acttab_insert(). The acttab_insert() call
578** also resets the aLookahead[] array in preparation for the next
579** state number.
drh8b582012003-10-21 13:16:03 +0000580*/
icculus9e44cf12010-02-14 17:14:22 +0000581struct lookahead_action {
582 int lookahead; /* Value of the lookahead token */
583 int action; /* Action to take on the given lookahead */
584};
drh8b582012003-10-21 13:16:03 +0000585typedef struct acttab acttab;
586struct acttab {
587 int nAction; /* Number of used slots in aAction[] */
588 int nActionAlloc; /* Slots allocated for aAction[] */
icculus9e44cf12010-02-14 17:14:22 +0000589 struct lookahead_action
590 *aAction, /* The yy_action[] table under construction */
drh8b582012003-10-21 13:16:03 +0000591 *aLookahead; /* A single new transaction set */
592 int mnLookahead; /* Minimum aLookahead[].lookahead */
593 int mnAction; /* Action associated with mnLookahead */
594 int mxLookahead; /* Maximum aLookahead[].lookahead */
595 int nLookahead; /* Used slots in aLookahead[] */
596 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
drh3a9d6c72017-12-25 04:15:38 +0000597 int nterminal; /* Number of terminal symbols */
598 int nsymbol; /* total number of symbols */
drh8b582012003-10-21 13:16:03 +0000599};
600
601/* Return the number of entries in the yy_action table */
drh3a9d6c72017-12-25 04:15:38 +0000602#define acttab_lookahead_size(X) ((X)->nAction)
drh8b582012003-10-21 13:16:03 +0000603
604/* The value for the N-th entry in yy_action */
605#define acttab_yyaction(X,N) ((X)->aAction[N].action)
606
607/* The value for the N-th entry in yy_lookahead */
608#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
609
610/* Free all memory associated with the given acttab */
611void acttab_free(acttab *p){
612 free( p->aAction );
613 free( p->aLookahead );
614 free( p );
615}
616
617/* Allocate a new acttab structure */
drh3a9d6c72017-12-25 04:15:38 +0000618acttab *acttab_alloc(int nsymbol, int nterminal){
icculus9e44cf12010-02-14 17:14:22 +0000619 acttab *p = (acttab *) calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000620 if( p==0 ){
621 fprintf(stderr,"Unable to allocate memory for a new acttab.");
622 exit(1);
623 }
624 memset(p, 0, sizeof(*p));
drh3a9d6c72017-12-25 04:15:38 +0000625 p->nsymbol = nsymbol;
626 p->nterminal = nterminal;
drh8b582012003-10-21 13:16:03 +0000627 return p;
628}
629
drh06f60d82017-04-14 19:46:12 +0000630/* Add a new action to the current transaction set.
drh8dc3e8f2010-01-07 03:53:03 +0000631**
632** This routine is called once for each lookahead for a particular
633** state.
drh8b582012003-10-21 13:16:03 +0000634*/
635void acttab_action(acttab *p, int lookahead, int action){
636 if( p->nLookahead>=p->nLookaheadAlloc ){
637 p->nLookaheadAlloc += 25;
icculus9e44cf12010-02-14 17:14:22 +0000638 p->aLookahead = (struct lookahead_action *) realloc( p->aLookahead,
drh8b582012003-10-21 13:16:03 +0000639 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
640 if( p->aLookahead==0 ){
641 fprintf(stderr,"malloc failed\n");
642 exit(1);
643 }
644 }
645 if( p->nLookahead==0 ){
646 p->mxLookahead = lookahead;
647 p->mnLookahead = lookahead;
648 p->mnAction = action;
649 }else{
650 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
651 if( p->mnLookahead>lookahead ){
652 p->mnLookahead = lookahead;
653 p->mnAction = action;
654 }
655 }
656 p->aLookahead[p->nLookahead].lookahead = lookahead;
657 p->aLookahead[p->nLookahead].action = action;
658 p->nLookahead++;
659}
660
661/*
662** Add the transaction set built up with prior calls to acttab_action()
663** into the current action table. Then reset the transaction set back
664** to an empty set in preparation for a new round of acttab_action() calls.
665**
666** Return the offset into the action table of the new transaction.
drh3a9d6c72017-12-25 04:15:38 +0000667**
668** If the makeItSafe parameter is true, then the offset is chosen so that
669** it is impossible to overread the yy_lookaside[] table regardless of
670** the lookaside token. This is done for the terminal symbols, as they
671** come from external inputs and can contain syntax errors. When makeItSafe
672** is false, there is more flexibility in selecting offsets, resulting in
673** a smaller table. For non-terminal symbols, which are never syntax errors,
674** makeItSafe can be false.
drh8b582012003-10-21 13:16:03 +0000675*/
drh3a9d6c72017-12-25 04:15:38 +0000676int acttab_insert(acttab *p, int makeItSafe){
677 int i, j, k, n, end;
drh8b582012003-10-21 13:16:03 +0000678 assert( p->nLookahead>0 );
679
680 /* Make sure we have enough space to hold the expanded action table
681 ** in the worst case. The worst case occurs if the transaction set
682 ** must be appended to the current action table
683 */
drh3a9d6c72017-12-25 04:15:38 +0000684 n = p->nsymbol + 1;
drh8dc3e8f2010-01-07 03:53:03 +0000685 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000686 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000687 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
icculus9e44cf12010-02-14 17:14:22 +0000688 p->aAction = (struct lookahead_action *) realloc( p->aAction,
drh8b582012003-10-21 13:16:03 +0000689 sizeof(p->aAction[0])*p->nActionAlloc);
690 if( p->aAction==0 ){
691 fprintf(stderr,"malloc failed\n");
692 exit(1);
693 }
drhfdbf9282003-10-21 16:34:41 +0000694 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000695 p->aAction[i].lookahead = -1;
696 p->aAction[i].action = -1;
697 }
698 }
699
drh06f60d82017-04-14 19:46:12 +0000700 /* Scan the existing action table looking for an offset that is a
drh8dc3e8f2010-01-07 03:53:03 +0000701 ** duplicate of the current transaction set. Fall out of the loop
702 ** if and when the duplicate is found.
drh8b582012003-10-21 13:16:03 +0000703 **
704 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
705 */
drh3a9d6c72017-12-25 04:15:38 +0000706 end = makeItSafe ? p->mnLookahead : 0;
707 for(i=p->nAction-1; i>=end; i--){
drhf16371d2009-11-03 19:18:31 +0000708 if( p->aAction[i].lookahead==p->mnLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000709 /* All lookaheads and actions in the aLookahead[] transaction
710 ** must match against the candidate aAction[i] entry. */
drh8b582012003-10-21 13:16:03 +0000711 if( p->aAction[i].action!=p->mnAction ) continue;
712 for(j=0; j<p->nLookahead; j++){
713 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
714 if( k<0 || k>=p->nAction ) break;
715 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
716 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
717 }
718 if( j<p->nLookahead ) continue;
drh8dc3e8f2010-01-07 03:53:03 +0000719
720 /* No possible lookahead value that is not in the aLookahead[]
721 ** transaction is allowed to match aAction[i] */
drh8b582012003-10-21 13:16:03 +0000722 n = 0;
723 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000724 if( p->aAction[j].lookahead<0 ) continue;
725 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000726 }
drhfdbf9282003-10-21 16:34:41 +0000727 if( n==p->nLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000728 break; /* An exact match is found at offset i */
drhfdbf9282003-10-21 16:34:41 +0000729 }
drh8b582012003-10-21 13:16:03 +0000730 }
731 }
drh8dc3e8f2010-01-07 03:53:03 +0000732
733 /* If no existing offsets exactly match the current transaction, find an
734 ** an empty offset in the aAction[] table in which we can add the
735 ** aLookahead[] transaction.
736 */
drh3a9d6c72017-12-25 04:15:38 +0000737 if( i<end ){
drh8dc3e8f2010-01-07 03:53:03 +0000738 /* Look for holes in the aAction[] table that fit the current
739 ** aLookahead[] transaction. Leave i set to the offset of the hole.
740 ** If no holes are found, i is left at p->nAction, which means the
741 ** transaction will be appended. */
drh3a9d6c72017-12-25 04:15:38 +0000742 i = makeItSafe ? p->mnLookahead : 0;
743 for(; i<p->nActionAlloc - p->mxLookahead; i++){
drhf16371d2009-11-03 19:18:31 +0000744 if( p->aAction[i].lookahead<0 ){
745 for(j=0; j<p->nLookahead; j++){
746 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
747 if( k<0 ) break;
748 if( p->aAction[k].lookahead>=0 ) break;
749 }
750 if( j<p->nLookahead ) continue;
751 for(j=0; j<p->nAction; j++){
752 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
753 }
754 if( j==p->nAction ){
755 break; /* Fits in empty slots */
756 }
757 }
758 }
759 }
drh8b582012003-10-21 13:16:03 +0000760 /* Insert transaction set at index i. */
drh3a9d6c72017-12-25 04:15:38 +0000761#if 0
762 printf("Acttab:");
763 for(j=0; j<p->nLookahead; j++){
764 printf(" %d", p->aLookahead[j].lookahead);
765 }
766 printf(" inserted at %d\n", i);
767#endif
drh8b582012003-10-21 13:16:03 +0000768 for(j=0; j<p->nLookahead; j++){
769 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
770 p->aAction[k] = p->aLookahead[j];
771 if( k>=p->nAction ) p->nAction = k+1;
772 }
drh4396c612017-12-27 15:21:16 +0000773 if( makeItSafe && i+p->nterminal>=p->nAction ) p->nAction = i+p->nterminal+1;
drh8b582012003-10-21 13:16:03 +0000774 p->nLookahead = 0;
775
776 /* Return the offset that is added to the lookahead in order to get the
777 ** index into yy_action of the action */
778 return i - p->mnLookahead;
779}
780
drh3a9d6c72017-12-25 04:15:38 +0000781/*
782** Return the size of the action table without the trailing syntax error
783** entries.
784*/
785int acttab_action_size(acttab *p){
786 int n = p->nAction;
787 while( n>0 && p->aAction[n-1].lookahead<0 ){ n--; }
788 return n;
789}
790
drh75897232000-05-29 14:26:00 +0000791/********************** From the file "build.c" *****************************/
792/*
793** Routines to construction the finite state machine for the LEMON
794** parser generator.
795*/
796
797/* Find a precedence symbol of every rule in the grammar.
drh06f60d82017-04-14 19:46:12 +0000798**
drh75897232000-05-29 14:26:00 +0000799** Those rules which have a precedence symbol coded in the input
800** grammar using the "[symbol]" construct will already have the
801** rp->precsym field filled. Other rules take as their precedence
802** symbol the first RHS symbol with a defined precedence. If there
803** are not RHS symbols with a defined precedence, the precedence
804** symbol field is left blank.
805*/
icculus9e44cf12010-02-14 17:14:22 +0000806void FindRulePrecedences(struct lemon *xp)
drh75897232000-05-29 14:26:00 +0000807{
808 struct rule *rp;
809 for(rp=xp->rule; rp; rp=rp->next){
810 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000811 int i, j;
812 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
813 struct symbol *sp = rp->rhs[i];
814 if( sp->type==MULTITERMINAL ){
815 for(j=0; j<sp->nsubsym; j++){
816 if( sp->subsym[j]->prec>=0 ){
817 rp->precsym = sp->subsym[j];
818 break;
819 }
820 }
821 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000822 rp->precsym = rp->rhs[i];
drhf2f105d2012-08-20 15:53:54 +0000823 }
drh75897232000-05-29 14:26:00 +0000824 }
825 }
826 }
827 return;
828}
829
830/* Find all nonterminals which will generate the empty string.
831** Then go back and compute the first sets of every nonterminal.
832** The first set is the set of all terminal symbols which can begin
833** a string generated by that nonterminal.
834*/
icculus9e44cf12010-02-14 17:14:22 +0000835void FindFirstSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000836{
drhfd405312005-11-06 04:06:59 +0000837 int i, j;
drh75897232000-05-29 14:26:00 +0000838 struct rule *rp;
839 int progress;
840
841 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000842 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000843 }
844 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
845 lemp->symbols[i]->firstset = SetNew();
846 }
847
848 /* First compute all lambdas */
849 do{
850 progress = 0;
851 for(rp=lemp->rule; rp; rp=rp->next){
852 if( rp->lhs->lambda ) continue;
853 for(i=0; i<rp->nrhs; i++){
drh7dd1ac62012-01-07 15:17:18 +0000854 struct symbol *sp = rp->rhs[i];
855 assert( sp->type==NONTERMINAL || sp->lambda==LEMON_FALSE );
856 if( sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000857 }
858 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000859 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000860 progress = 1;
861 }
862 }
863 }while( progress );
864
865 /* Now compute all first sets */
866 do{
867 struct symbol *s1, *s2;
868 progress = 0;
869 for(rp=lemp->rule; rp; rp=rp->next){
870 s1 = rp->lhs;
871 for(i=0; i<rp->nrhs; i++){
872 s2 = rp->rhs[i];
873 if( s2->type==TERMINAL ){
874 progress += SetAdd(s1->firstset,s2->index);
875 break;
drhfd405312005-11-06 04:06:59 +0000876 }else if( s2->type==MULTITERMINAL ){
877 for(j=0; j<s2->nsubsym; j++){
878 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
879 }
880 break;
drhf2f105d2012-08-20 15:53:54 +0000881 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000882 if( s1->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000883 }else{
drh75897232000-05-29 14:26:00 +0000884 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000885 if( s2->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000886 }
drh75897232000-05-29 14:26:00 +0000887 }
888 }
889 }while( progress );
890 return;
891}
892
893/* Compute all LR(0) states for the grammar. Links
894** are added to between some states so that the LR(1) follow sets
895** can be computed later.
896*/
icculus9e44cf12010-02-14 17:14:22 +0000897PRIVATE struct state *getstate(struct lemon *); /* forward reference */
898void FindStates(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000899{
900 struct symbol *sp;
901 struct rule *rp;
902
903 Configlist_init();
904
905 /* Find the start symbol */
906 if( lemp->start ){
907 sp = Symbol_find(lemp->start);
908 if( sp==0 ){
909 ErrorMsg(lemp->filename,0,
910"The specified start symbol \"%s\" is not \
911in a nonterminal of the grammar. \"%s\" will be used as the start \
drh4ef07702016-03-16 19:45:54 +0000912symbol instead.",lemp->start,lemp->startRule->lhs->name);
drh75897232000-05-29 14:26:00 +0000913 lemp->errorcnt++;
drh4ef07702016-03-16 19:45:54 +0000914 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +0000915 }
916 }else{
drh4ef07702016-03-16 19:45:54 +0000917 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +0000918 }
919
920 /* Make sure the start symbol doesn't occur on the right-hand side of
921 ** any rule. Report an error if it does. (YACC would generate a new
922 ** start symbol in this case.) */
923 for(rp=lemp->rule; rp; rp=rp->next){
924 int i;
925 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000926 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000927 ErrorMsg(lemp->filename,0,
928"The start symbol \"%s\" occurs on the \
929right-hand side of a rule. This will result in a parser which \
930does not work properly.",sp->name);
931 lemp->errorcnt++;
932 }
933 }
934 }
935
936 /* The basis configuration set for the first state
937 ** is all rules which have the start symbol as their
938 ** left-hand side */
939 for(rp=sp->rule; rp; rp=rp->nextlhs){
940 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000941 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000942 newcfp = Configlist_addbasis(rp,0);
943 SetAdd(newcfp->fws,0);
944 }
945
946 /* Compute the first state. All other states will be
947 ** computed automatically during the computation of the first one.
948 ** The returned pointer to the first state is not used. */
949 (void)getstate(lemp);
950 return;
951}
952
953/* Return a pointer to a state which is described by the configuration
954** list which has been built from calls to Configlist_add.
955*/
icculus9e44cf12010-02-14 17:14:22 +0000956PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
957PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000958{
959 struct config *cfp, *bp;
960 struct state *stp;
961
962 /* Extract the sorted basis of the new state. The basis was constructed
963 ** by prior calls to "Configlist_addbasis()". */
964 Configlist_sortbasis();
965 bp = Configlist_basis();
966
967 /* Get a state with the same basis */
968 stp = State_find(bp);
969 if( stp ){
970 /* A state with the same basis already exists! Copy all the follow-set
971 ** propagation links from the state under construction into the
972 ** preexisting state, then return a pointer to the preexisting state */
973 struct config *x, *y;
974 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
975 Plink_copy(&y->bplp,x->bplp);
976 Plink_delete(x->fplp);
977 x->fplp = x->bplp = 0;
978 }
979 cfp = Configlist_return();
980 Configlist_eat(cfp);
981 }else{
982 /* This really is a new state. Construct all the details */
983 Configlist_closure(lemp); /* Compute the configuration closure */
984 Configlist_sort(); /* Sort the configuration closure */
985 cfp = Configlist_return(); /* Get a pointer to the config list */
986 stp = State_new(); /* A new state structure */
987 MemoryCheck(stp);
988 stp->bp = bp; /* Remember the configuration basis */
989 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000990 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000991 stp->ap = 0; /* No actions, yet. */
992 State_insert(stp,stp->bp); /* Add to the state table */
993 buildshifts(lemp,stp); /* Recursively compute successor states */
994 }
995 return stp;
996}
997
drhfd405312005-11-06 04:06:59 +0000998/*
999** Return true if two symbols are the same.
1000*/
icculus9e44cf12010-02-14 17:14:22 +00001001int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +00001002{
1003 int i;
1004 if( a==b ) return 1;
1005 if( a->type!=MULTITERMINAL ) return 0;
1006 if( b->type!=MULTITERMINAL ) return 0;
1007 if( a->nsubsym!=b->nsubsym ) return 0;
1008 for(i=0; i<a->nsubsym; i++){
1009 if( a->subsym[i]!=b->subsym[i] ) return 0;
1010 }
1011 return 1;
1012}
1013
drh75897232000-05-29 14:26:00 +00001014/* Construct all successor states to the given state. A "successor"
1015** state is any state which can be reached by a shift action.
1016*/
icculus9e44cf12010-02-14 17:14:22 +00001017PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +00001018{
1019 struct config *cfp; /* For looping thru the config closure of "stp" */
1020 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +00001021 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +00001022 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
1023 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
1024 struct state *newstp; /* A pointer to a successor state */
1025
1026 /* Each configuration becomes complete after it contibutes to a successor
1027 ** state. Initially, all configurations are incomplete */
1028 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
1029
1030 /* Loop through all configurations of the state "stp" */
1031 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1032 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
1033 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
1034 Configlist_reset(); /* Reset the new config set */
1035 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
1036
1037 /* For every configuration in the state "stp" which has the symbol "sp"
1038 ** following its dot, add the same configuration to the basis set under
1039 ** construction but with the dot shifted one symbol to the right. */
1040 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
1041 if( bcfp->status==COMPLETE ) continue; /* Already used */
1042 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
1043 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +00001044 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +00001045 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +00001046 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
1047 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +00001048 }
1049
1050 /* Get a pointer to the state described by the basis configuration set
1051 ** constructed in the preceding loop */
1052 newstp = getstate(lemp);
1053
1054 /* The state "newstp" is reached from the state "stp" by a shift action
1055 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +00001056 if( sp->type==MULTITERMINAL ){
1057 int i;
1058 for(i=0; i<sp->nsubsym; i++){
1059 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
1060 }
1061 }else{
1062 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
1063 }
drh75897232000-05-29 14:26:00 +00001064 }
1065}
1066
1067/*
1068** Construct the propagation links
1069*/
icculus9e44cf12010-02-14 17:14:22 +00001070void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001071{
1072 int i;
1073 struct config *cfp, *other;
1074 struct state *stp;
1075 struct plink *plp;
1076
1077 /* Housekeeping detail:
1078 ** Add to every propagate link a pointer back to the state to
1079 ** which the link is attached. */
1080 for(i=0; i<lemp->nstate; i++){
1081 stp = lemp->sorted[i];
1082 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1083 cfp->stp = stp;
1084 }
1085 }
1086
1087 /* Convert all backlinks into forward links. Only the forward
1088 ** links are used in the follow-set computation. */
1089 for(i=0; i<lemp->nstate; i++){
1090 stp = lemp->sorted[i];
1091 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1092 for(plp=cfp->bplp; plp; plp=plp->next){
1093 other = plp->cfp;
1094 Plink_add(&other->fplp,cfp);
1095 }
1096 }
1097 }
1098}
1099
1100/* Compute all followsets.
1101**
1102** A followset is the set of all symbols which can come immediately
1103** after a configuration.
1104*/
icculus9e44cf12010-02-14 17:14:22 +00001105void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001106{
1107 int i;
1108 struct config *cfp;
1109 struct plink *plp;
1110 int progress;
1111 int change;
1112
1113 for(i=0; i<lemp->nstate; i++){
1114 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1115 cfp->status = INCOMPLETE;
1116 }
1117 }
drh06f60d82017-04-14 19:46:12 +00001118
drh75897232000-05-29 14:26:00 +00001119 do{
1120 progress = 0;
1121 for(i=0; i<lemp->nstate; i++){
1122 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1123 if( cfp->status==COMPLETE ) continue;
1124 for(plp=cfp->fplp; plp; plp=plp->next){
1125 change = SetUnion(plp->cfp->fws,cfp->fws);
1126 if( change ){
1127 plp->cfp->status = INCOMPLETE;
1128 progress = 1;
drhf2f105d2012-08-20 15:53:54 +00001129 }
1130 }
drh75897232000-05-29 14:26:00 +00001131 cfp->status = COMPLETE;
1132 }
1133 }
1134 }while( progress );
1135}
1136
drh3cb2f6e2012-01-09 14:19:05 +00001137static int resolve_conflict(struct action *,struct action *);
drh75897232000-05-29 14:26:00 +00001138
1139/* Compute the reduce actions, and resolve conflicts.
1140*/
icculus9e44cf12010-02-14 17:14:22 +00001141void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001142{
1143 int i,j;
1144 struct config *cfp;
1145 struct state *stp;
1146 struct symbol *sp;
1147 struct rule *rp;
1148
drh06f60d82017-04-14 19:46:12 +00001149 /* Add all of the reduce actions
drh75897232000-05-29 14:26:00 +00001150 ** A reduce action is added for each element of the followset of
1151 ** a configuration which has its dot at the extreme right.
1152 */
1153 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
1154 stp = lemp->sorted[i];
1155 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
1156 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
1157 for(j=0; j<lemp->nterminal; j++){
1158 if( SetFind(cfp->fws,j) ){
1159 /* Add a reduce action to the state "stp" which will reduce by the
1160 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +00001161 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +00001162 }
drhf2f105d2012-08-20 15:53:54 +00001163 }
drh75897232000-05-29 14:26:00 +00001164 }
1165 }
1166 }
1167
1168 /* Add the accepting token */
1169 if( lemp->start ){
1170 sp = Symbol_find(lemp->start);
drh4ef07702016-03-16 19:45:54 +00001171 if( sp==0 ) sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +00001172 }else{
drh4ef07702016-03-16 19:45:54 +00001173 sp = lemp->startRule->lhs;
drh75897232000-05-29 14:26:00 +00001174 }
1175 /* Add to the first state (which is always the starting state of the
1176 ** finite state machine) an action to ACCEPT if the lookahead is the
1177 ** start nonterminal. */
1178 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1179
1180 /* Resolve conflicts */
1181 for(i=0; i<lemp->nstate; i++){
1182 struct action *ap, *nap;
drh75897232000-05-29 14:26:00 +00001183 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001184 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001185 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001186 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001187 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1188 /* The two actions "ap" and "nap" have the same lookahead.
1189 ** Figure out which one should be used */
drh3cb2f6e2012-01-09 14:19:05 +00001190 lemp->nconflict += resolve_conflict(ap,nap);
drh75897232000-05-29 14:26:00 +00001191 }
1192 }
1193 }
1194
1195 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001196 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001197 for(i=0; i<lemp->nstate; i++){
1198 struct action *ap;
1199 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001200 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001201 }
1202 }
1203 for(rp=lemp->rule; rp; rp=rp->next){
1204 if( rp->canReduce ) continue;
1205 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1206 lemp->errorcnt++;
1207 }
1208}
1209
1210/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001211** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001212**
1213** NO LONGER TRUE:
1214** To resolve a conflict, first look to see if either action
1215** is on an error rule. In that case, take the action which
1216** is not associated with the error rule. If neither or both
1217** actions are associated with an error rule, then try to
1218** use precedence to resolve the conflict.
1219**
1220** If either action is a SHIFT, then it must be apx. This
1221** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1222*/
icculus9e44cf12010-02-14 17:14:22 +00001223static int resolve_conflict(
1224 struct action *apx,
drh3cb2f6e2012-01-09 14:19:05 +00001225 struct action *apy
icculus9e44cf12010-02-14 17:14:22 +00001226){
drh75897232000-05-29 14:26:00 +00001227 struct symbol *spx, *spy;
1228 int errcnt = 0;
1229 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001230 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001231 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001232 errcnt++;
1233 }
drh75897232000-05-29 14:26:00 +00001234 if( apx->type==SHIFT && apy->type==REDUCE ){
1235 spx = apx->sp;
1236 spy = apy->x.rp->precsym;
1237 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1238 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001239 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001240 errcnt++;
drhdd7e9db2010-07-19 01:52:07 +00001241 }else if( spx->prec>spy->prec ){ /* higher precedence wins */
drh75897232000-05-29 14:26:00 +00001242 apy->type = RD_RESOLVED;
1243 }else if( spx->prec<spy->prec ){
1244 apx->type = SH_RESOLVED;
1245 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1246 apy->type = RD_RESOLVED; /* associativity */
1247 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1248 apx->type = SH_RESOLVED;
1249 }else{
1250 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh62a223e2014-06-09 13:11:40 +00001251 apx->type = ERROR;
drh75897232000-05-29 14:26:00 +00001252 }
1253 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1254 spx = apx->x.rp->precsym;
1255 spy = apy->x.rp->precsym;
1256 if( spx==0 || spy==0 || spx->prec<0 ||
1257 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001258 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001259 errcnt++;
1260 }else if( spx->prec>spy->prec ){
1261 apy->type = RD_RESOLVED;
1262 }else if( spx->prec<spy->prec ){
1263 apx->type = RD_RESOLVED;
1264 }
1265 }else{
drh06f60d82017-04-14 19:46:12 +00001266 assert(
drhb59499c2002-02-23 18:45:13 +00001267 apx->type==SH_RESOLVED ||
1268 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001269 apx->type==SSCONFLICT ||
1270 apx->type==SRCONFLICT ||
1271 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001272 apy->type==SH_RESOLVED ||
1273 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001274 apy->type==SSCONFLICT ||
1275 apy->type==SRCONFLICT ||
1276 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001277 );
1278 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1279 ** REDUCEs on the list. If we reach this point it must be because
1280 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001281 }
1282 return errcnt;
1283}
1284/********************* From the file "configlist.c" *************************/
1285/*
1286** Routines to processing a configuration list and building a state
1287** in the LEMON parser generator.
1288*/
1289
1290static struct config *freelist = 0; /* List of free configurations */
1291static struct config *current = 0; /* Top of list of configurations */
1292static struct config **currentend = 0; /* Last on list of configs */
1293static struct config *basis = 0; /* Top of list of basis configs */
1294static struct config **basisend = 0; /* End of list of basis configs */
1295
1296/* Return a pointer to a new configuration */
drh14d88552017-04-14 19:44:15 +00001297PRIVATE struct config *newconfig(void){
icculus9e44cf12010-02-14 17:14:22 +00001298 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001299 if( freelist==0 ){
1300 int i;
1301 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001302 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001303 if( freelist==0 ){
1304 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1305 exit(1);
1306 }
1307 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1308 freelist[amt-1].next = 0;
1309 }
icculus9e44cf12010-02-14 17:14:22 +00001310 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001311 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001312 return newcfg;
drh75897232000-05-29 14:26:00 +00001313}
1314
1315/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001316PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001317{
1318 old->next = freelist;
1319 freelist = old;
1320}
1321
1322/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001323void Configlist_init(void){
drh75897232000-05-29 14:26:00 +00001324 current = 0;
1325 currentend = &current;
1326 basis = 0;
1327 basisend = &basis;
1328 Configtable_init();
1329 return;
1330}
1331
1332/* Initialized the configuration list builder */
drh14d88552017-04-14 19:44:15 +00001333void Configlist_reset(void){
drh75897232000-05-29 14:26:00 +00001334 current = 0;
1335 currentend = &current;
1336 basis = 0;
1337 basisend = &basis;
1338 Configtable_clear(0);
1339 return;
1340}
1341
1342/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001343struct config *Configlist_add(
1344 struct rule *rp, /* The rule */
1345 int dot /* Index into the RHS of the rule where the dot goes */
1346){
drh75897232000-05-29 14:26:00 +00001347 struct config *cfp, model;
1348
1349 assert( currentend!=0 );
1350 model.rp = rp;
1351 model.dot = dot;
1352 cfp = Configtable_find(&model);
1353 if( cfp==0 ){
1354 cfp = newconfig();
1355 cfp->rp = rp;
1356 cfp->dot = dot;
1357 cfp->fws = SetNew();
1358 cfp->stp = 0;
1359 cfp->fplp = cfp->bplp = 0;
1360 cfp->next = 0;
1361 cfp->bp = 0;
1362 *currentend = cfp;
1363 currentend = &cfp->next;
1364 Configtable_insert(cfp);
1365 }
1366 return cfp;
1367}
1368
1369/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001370struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001371{
1372 struct config *cfp, model;
1373
1374 assert( basisend!=0 );
1375 assert( currentend!=0 );
1376 model.rp = rp;
1377 model.dot = dot;
1378 cfp = Configtable_find(&model);
1379 if( cfp==0 ){
1380 cfp = newconfig();
1381 cfp->rp = rp;
1382 cfp->dot = dot;
1383 cfp->fws = SetNew();
1384 cfp->stp = 0;
1385 cfp->fplp = cfp->bplp = 0;
1386 cfp->next = 0;
1387 cfp->bp = 0;
1388 *currentend = cfp;
1389 currentend = &cfp->next;
1390 *basisend = cfp;
1391 basisend = &cfp->bp;
1392 Configtable_insert(cfp);
1393 }
1394 return cfp;
1395}
1396
1397/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001398void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001399{
1400 struct config *cfp, *newcfp;
1401 struct rule *rp, *newrp;
1402 struct symbol *sp, *xsp;
1403 int i, dot;
1404
1405 assert( currentend!=0 );
1406 for(cfp=current; cfp; cfp=cfp->next){
1407 rp = cfp->rp;
1408 dot = cfp->dot;
1409 if( dot>=rp->nrhs ) continue;
1410 sp = rp->rhs[dot];
1411 if( sp->type==NONTERMINAL ){
1412 if( sp->rule==0 && sp!=lemp->errsym ){
1413 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1414 sp->name);
1415 lemp->errorcnt++;
1416 }
1417 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1418 newcfp = Configlist_add(newrp,0);
1419 for(i=dot+1; i<rp->nrhs; i++){
1420 xsp = rp->rhs[i];
1421 if( xsp->type==TERMINAL ){
1422 SetAdd(newcfp->fws,xsp->index);
1423 break;
drhfd405312005-11-06 04:06:59 +00001424 }else if( xsp->type==MULTITERMINAL ){
1425 int k;
1426 for(k=0; k<xsp->nsubsym; k++){
1427 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1428 }
1429 break;
drhf2f105d2012-08-20 15:53:54 +00001430 }else{
drh75897232000-05-29 14:26:00 +00001431 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001432 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001433 }
1434 }
drh75897232000-05-29 14:26:00 +00001435 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1436 }
1437 }
1438 }
1439 return;
1440}
1441
1442/* Sort the configuration list */
drh14d88552017-04-14 19:44:15 +00001443void Configlist_sort(void){
drh25473362015-09-04 18:03:45 +00001444 current = (struct config*)msort((char*)current,(char**)&(current->next),
1445 Configcmp);
drh75897232000-05-29 14:26:00 +00001446 currentend = 0;
1447 return;
1448}
1449
1450/* Sort the basis configuration list */
drh14d88552017-04-14 19:44:15 +00001451void Configlist_sortbasis(void){
drh25473362015-09-04 18:03:45 +00001452 basis = (struct config*)msort((char*)current,(char**)&(current->bp),
1453 Configcmp);
drh75897232000-05-29 14:26:00 +00001454 basisend = 0;
1455 return;
1456}
1457
1458/* Return a pointer to the head of the configuration list and
1459** reset the list */
drh14d88552017-04-14 19:44:15 +00001460struct config *Configlist_return(void){
drh75897232000-05-29 14:26:00 +00001461 struct config *old;
1462 old = current;
1463 current = 0;
1464 currentend = 0;
1465 return old;
1466}
1467
1468/* Return a pointer to the head of the configuration list and
1469** reset the list */
drh14d88552017-04-14 19:44:15 +00001470struct config *Configlist_basis(void){
drh75897232000-05-29 14:26:00 +00001471 struct config *old;
1472 old = basis;
1473 basis = 0;
1474 basisend = 0;
1475 return old;
1476}
1477
1478/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001479void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001480{
1481 struct config *nextcfp;
1482 for(; cfp; cfp=nextcfp){
1483 nextcfp = cfp->next;
1484 assert( cfp->fplp==0 );
1485 assert( cfp->bplp==0 );
1486 if( cfp->fws ) SetFree(cfp->fws);
1487 deleteconfig(cfp);
1488 }
1489 return;
1490}
1491/***************** From the file "error.c" *********************************/
1492/*
1493** Code for printing error message.
1494*/
1495
drhf9a2e7b2003-04-15 01:49:48 +00001496void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001497 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001498 fprintf(stderr, "%s:%d: ", filename, lineno);
1499 va_start(ap, format);
1500 vfprintf(stderr,format,ap);
1501 va_end(ap);
1502 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001503}
1504/**************** From the file "main.c" ************************************/
1505/*
1506** Main program file for the LEMON parser generator.
1507*/
1508
1509/* Report an out-of-memory condition and abort. This function
1510** is used mostly by the "MemoryCheck" macro in struct.h
1511*/
drh14d88552017-04-14 19:44:15 +00001512void memory_error(void){
drh75897232000-05-29 14:26:00 +00001513 fprintf(stderr,"Out of memory. Aborting...\n");
1514 exit(1);
1515}
1516
drh6d08b4d2004-07-20 12:45:22 +00001517static int nDefine = 0; /* Number of -D options on the command line */
1518static char **azDefine = 0; /* Name of the -D macros */
1519
1520/* This routine is called with the argument to each -D command-line option.
1521** Add the macro defined to the azDefine array.
1522*/
1523static void handle_D_option(char *z){
1524 char **paz;
1525 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001526 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001527 if( azDefine==0 ){
1528 fprintf(stderr,"out of memory\n");
1529 exit(1);
1530 }
1531 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001532 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001533 if( *paz==0 ){
1534 fprintf(stderr,"out of memory\n");
1535 exit(1);
1536 }
drh898799f2014-01-10 23:21:00 +00001537 lemon_strcpy(*paz, z);
drh6d08b4d2004-07-20 12:45:22 +00001538 for(z=*paz; *z && *z!='='; z++){}
1539 *z = 0;
1540}
1541
drh9f88e6d2018-04-20 20:47:49 +00001542/* Rember the name of the output directory
1543*/
1544static char *outputDir = NULL;
1545static void handle_d_option(char *z){
1546 outputDir = (char *) malloc( lemonStrlen(z)+1 );
1547 if( outputDir==0 ){
1548 fprintf(stderr,"out of memory\n");
1549 exit(1);
1550 }
1551 lemon_strcpy(outputDir, z);
1552}
1553
icculus3e143bd2010-02-14 00:48:49 +00001554static char *user_templatename = NULL;
1555static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001556 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001557 if( user_templatename==0 ){
1558 memory_error();
1559 }
drh898799f2014-01-10 23:21:00 +00001560 lemon_strcpy(user_templatename, z);
icculus3e143bd2010-02-14 00:48:49 +00001561}
drh75897232000-05-29 14:26:00 +00001562
drh711c9812016-05-23 14:24:31 +00001563/* Merge together to lists of rules ordered by rule.iRule */
drh4ef07702016-03-16 19:45:54 +00001564static struct rule *Rule_merge(struct rule *pA, struct rule *pB){
1565 struct rule *pFirst = 0;
1566 struct rule **ppPrev = &pFirst;
1567 while( pA && pB ){
1568 if( pA->iRule<pB->iRule ){
1569 *ppPrev = pA;
1570 ppPrev = &pA->next;
1571 pA = pA->next;
1572 }else{
1573 *ppPrev = pB;
1574 ppPrev = &pB->next;
1575 pB = pB->next;
1576 }
1577 }
1578 if( pA ){
1579 *ppPrev = pA;
1580 }else{
1581 *ppPrev = pB;
1582 }
1583 return pFirst;
1584}
1585
1586/*
1587** Sort a list of rules in order of increasing iRule value
1588*/
1589static struct rule *Rule_sort(struct rule *rp){
1590 int i;
1591 struct rule *pNext;
1592 struct rule *x[32];
1593 memset(x, 0, sizeof(x));
1594 while( rp ){
1595 pNext = rp->next;
1596 rp->next = 0;
1597 for(i=0; i<sizeof(x)/sizeof(x[0]) && x[i]; i++){
1598 rp = Rule_merge(x[i], rp);
1599 x[i] = 0;
1600 }
1601 x[i] = rp;
1602 rp = pNext;
1603 }
1604 rp = 0;
1605 for(i=0; i<sizeof(x)/sizeof(x[0]); i++){
1606 rp = Rule_merge(x[i], rp);
1607 }
1608 return rp;
1609}
1610
drhc75e0162015-09-07 02:23:02 +00001611/* forward reference */
1612static const char *minimum_size_type(int lwr, int upr, int *pnByte);
1613
1614/* Print a single line of the "Parser Stats" output
1615*/
1616static void stats_line(const char *zLabel, int iValue){
1617 int nLabel = lemonStrlen(zLabel);
1618 printf(" %s%.*s %5d\n", zLabel,
1619 35-nLabel, "................................",
1620 iValue);
1621}
1622
drh75897232000-05-29 14:26:00 +00001623/* The main program. Parse the command line and do it... */
icculus9e44cf12010-02-14 17:14:22 +00001624int main(int argc, char **argv)
drh75897232000-05-29 14:26:00 +00001625{
1626 static int version = 0;
1627 static int rpflag = 0;
1628 static int basisflag = 0;
1629 static int compress = 0;
1630 static int quiet = 0;
1631 static int statistics = 0;
1632 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001633 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001634 static int noResort = 0;
drh9f88e6d2018-04-20 20:47:49 +00001635
drh75897232000-05-29 14:26:00 +00001636 static struct s_options options[] = {
1637 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1638 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh9f88e6d2018-04-20 20:47:49 +00001639 {OPT_FSTR, "d", (char*)&handle_d_option, "Output directory. Default '.'"},
drh6d08b4d2004-07-20 12:45:22 +00001640 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh0325d392015-01-01 19:11:22 +00001641 {OPT_FSTR, "f", 0, "Ignored. (Placeholder for -f compiler options.)"},
drh75897232000-05-29 14:26:00 +00001642 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
drh0325d392015-01-01 19:11:22 +00001643 {OPT_FSTR, "I", 0, "Ignored. (Placeholder for '-I' compiler options.)"},
shane58543932008-12-10 20:10:04 +00001644 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1645 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drh0325d392015-01-01 19:11:22 +00001646 {OPT_FSTR, "O", 0, "Ignored. (Placeholder for '-O' compiler options.)"},
drhf5c4e0f2010-07-18 11:35:53 +00001647 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1648 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001649 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001650 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001651 {OPT_FLAG, "s", (char*)&statistics,
1652 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001653 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
drh0325d392015-01-01 19:11:22 +00001654 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
1655 {OPT_FSTR, "W", 0, "Ignored. (Placeholder for '-W' compiler options.)"},
drh75897232000-05-29 14:26:00 +00001656 {OPT_FLAG,0,0,0}
1657 };
1658 int i;
icculus42585cf2010-02-14 05:19:56 +00001659 int exitcode;
drh75897232000-05-29 14:26:00 +00001660 struct lemon lem;
drh4ef07702016-03-16 19:45:54 +00001661 struct rule *rp;
drh75897232000-05-29 14:26:00 +00001662
drhb0c86772000-06-02 23:21:26 +00001663 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001664 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001665 printf("Lemon version 1.0\n");
drh06f60d82017-04-14 19:46:12 +00001666 exit(0);
drh75897232000-05-29 14:26:00 +00001667 }
drhb0c86772000-06-02 23:21:26 +00001668 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001669 fprintf(stderr,"Exactly one filename argument is required.\n");
1670 exit(1);
1671 }
drh954f6b42006-06-13 13:27:46 +00001672 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001673 lem.errorcnt = 0;
1674
1675 /* Initialize the machine */
1676 Strsafe_init();
1677 Symbol_init();
1678 State_init();
1679 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001680 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001681 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001682 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001683 Symbol_new("$");
drh75897232000-05-29 14:26:00 +00001684
1685 /* Parse the input file */
1686 Parse(&lem);
1687 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001688 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001689 fprintf(stderr,"Empty grammar.\n");
1690 exit(1);
1691 }
drhed0c15b2018-04-16 14:31:34 +00001692 lem.errsym = Symbol_find("error");
drh75897232000-05-29 14:26:00 +00001693
1694 /* Count and index the symbols of the grammar */
drh75897232000-05-29 14:26:00 +00001695 Symbol_new("{default}");
drh61f92cd2014-01-11 03:06:18 +00001696 lem.nsymbol = Symbol_count();
drh75897232000-05-29 14:26:00 +00001697 lem.symbols = Symbol_arrayof();
drh61f92cd2014-01-11 03:06:18 +00001698 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1699 qsort(lem.symbols,lem.nsymbol,sizeof(struct symbol*), Symbolcmpp);
1700 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1701 while( lem.symbols[i-1]->type==MULTITERMINAL ){ i--; }
1702 assert( strcmp(lem.symbols[i-1]->name,"{default}")==0 );
1703 lem.nsymbol = i - 1;
drhc56fac72015-10-29 13:48:15 +00001704 for(i=1; ISUPPER(lem.symbols[i]->name[0]); i++);
drh75897232000-05-29 14:26:00 +00001705 lem.nterminal = i;
1706
drh711c9812016-05-23 14:24:31 +00001707 /* Assign sequential rule numbers. Start with 0. Put rules that have no
1708 ** reduce action C-code associated with them last, so that the switch()
1709 ** statement that selects reduction actions will have a smaller jump table.
1710 */
drh4ef07702016-03-16 19:45:54 +00001711 for(i=0, rp=lem.rule; rp; rp=rp->next){
1712 rp->iRule = rp->code ? i++ : -1;
1713 }
1714 for(rp=lem.rule; rp; rp=rp->next){
1715 if( rp->iRule<0 ) rp->iRule = i++;
1716 }
1717 lem.startRule = lem.rule;
1718 lem.rule = Rule_sort(lem.rule);
1719
drh75897232000-05-29 14:26:00 +00001720 /* Generate a reprint of the grammar, if requested on the command line */
1721 if( rpflag ){
1722 Reprint(&lem);
1723 }else{
1724 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001725 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001726
1727 /* Find the precedence for every production rule (that has one) */
1728 FindRulePrecedences(&lem);
1729
1730 /* Compute the lambda-nonterminals and the first-sets for every
1731 ** nonterminal */
1732 FindFirstSets(&lem);
1733
1734 /* Compute all LR(0) states. Also record follow-set propagation
1735 ** links so that the follow-set can be computed later */
1736 lem.nstate = 0;
1737 FindStates(&lem);
1738 lem.sorted = State_arrayof();
1739
1740 /* Tie up loose ends on the propagation links */
1741 FindLinks(&lem);
1742
1743 /* Compute the follow set of every reducible configuration */
1744 FindFollowSets(&lem);
1745
1746 /* Compute the action tables */
1747 FindActions(&lem);
1748
1749 /* Compress the action tables */
1750 if( compress==0 ) CompressTables(&lem);
1751
drhada354d2005-11-05 15:03:59 +00001752 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001753 ** occur at the end. This is an optimization that helps make the
1754 ** generated parser tables smaller. */
1755 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001756
drh75897232000-05-29 14:26:00 +00001757 /* Generate a report of the parser generated. (the "y.output" file) */
1758 if( !quiet ) ReportOutput(&lem);
1759
1760 /* Generate the source code for the parser */
1761 ReportTable(&lem, mhflag);
1762
1763 /* Produce a header file for use by the scanner. (This step is
1764 ** omitted if the "-m" option is used because makeheaders will
1765 ** generate the file for us.) */
1766 if( !mhflag ) ReportHeader(&lem);
1767 }
1768 if( statistics ){
drhc75e0162015-09-07 02:23:02 +00001769 printf("Parser statistics:\n");
1770 stats_line("terminal symbols", lem.nterminal);
1771 stats_line("non-terminal symbols", lem.nsymbol - lem.nterminal);
1772 stats_line("total symbols", lem.nsymbol);
1773 stats_line("rules", lem.nrule);
drh3bd48ab2015-09-07 18:23:37 +00001774 stats_line("states", lem.nxstate);
drhc75e0162015-09-07 02:23:02 +00001775 stats_line("conflicts", lem.nconflict);
1776 stats_line("action table entries", lem.nactiontab);
drh3a9d6c72017-12-25 04:15:38 +00001777 stats_line("lookahead table entries", lem.nlookaheadtab);
drhc75e0162015-09-07 02:23:02 +00001778 stats_line("total table size (bytes)", lem.tablesize);
drh75897232000-05-29 14:26:00 +00001779 }
icculus8e158022010-02-16 16:09:03 +00001780 if( lem.nconflict > 0 ){
1781 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001782 }
1783
1784 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001785 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001786 exit(exitcode);
1787 return (exitcode);
drh75897232000-05-29 14:26:00 +00001788}
1789/******************** From the file "msort.c" *******************************/
1790/*
1791** A generic merge-sort program.
1792**
1793** USAGE:
1794** Let "ptr" be a pointer to some structure which is at the head of
1795** a null-terminated list. Then to sort the list call:
1796**
1797** ptr = msort(ptr,&(ptr->next),cmpfnc);
1798**
1799** In the above, "cmpfnc" is a pointer to a function which compares
1800** two instances of the structure and returns an integer, as in
1801** strcmp. The second argument is a pointer to the pointer to the
1802** second element of the linked list. This address is used to compute
1803** the offset to the "next" field within the structure. The offset to
1804** the "next" field must be constant for all structures in the list.
1805**
1806** The function returns a new pointer which is the head of the list
1807** after sorting.
1808**
1809** ALGORITHM:
1810** Merge-sort.
1811*/
1812
1813/*
1814** Return a pointer to the next structure in the linked list.
1815*/
drhd25d6922012-04-18 09:59:56 +00001816#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001817
1818/*
1819** Inputs:
1820** a: A sorted, null-terminated linked list. (May be null).
1821** b: A sorted, null-terminated linked list. (May be null).
1822** cmp: A pointer to the comparison function.
1823** offset: Offset in the structure to the "next" field.
1824**
1825** Return Value:
1826** A pointer to the head of a sorted list containing the elements
1827** of both a and b.
1828**
1829** Side effects:
1830** The "next" pointers for elements in the lists a and b are
1831** changed.
1832*/
drhe9278182007-07-18 18:16:29 +00001833static char *merge(
1834 char *a,
1835 char *b,
1836 int (*cmp)(const char*,const char*),
1837 int offset
1838){
drh75897232000-05-29 14:26:00 +00001839 char *ptr, *head;
1840
1841 if( a==0 ){
1842 head = b;
1843 }else if( b==0 ){
1844 head = a;
1845 }else{
drhe594bc32009-11-03 13:02:25 +00001846 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001847 ptr = a;
1848 a = NEXT(a);
1849 }else{
1850 ptr = b;
1851 b = NEXT(b);
1852 }
1853 head = ptr;
1854 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001855 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001856 NEXT(ptr) = a;
1857 ptr = a;
1858 a = NEXT(a);
1859 }else{
1860 NEXT(ptr) = b;
1861 ptr = b;
1862 b = NEXT(b);
1863 }
1864 }
1865 if( a ) NEXT(ptr) = a;
1866 else NEXT(ptr) = b;
1867 }
1868 return head;
1869}
1870
1871/*
1872** Inputs:
1873** list: Pointer to a singly-linked list of structures.
1874** next: Pointer to pointer to the second element of the list.
1875** cmp: A comparison function.
1876**
1877** Return Value:
1878** A pointer to the head of a sorted list containing the elements
1879** orginally in list.
1880**
1881** Side effects:
1882** The "next" pointers for elements in list are changed.
1883*/
1884#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001885static char *msort(
1886 char *list,
1887 char **next,
1888 int (*cmp)(const char*,const char*)
1889){
drhba99af52001-10-25 20:37:16 +00001890 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001891 char *ep;
1892 char *set[LISTSIZE];
1893 int i;
drh1cc0d112015-03-31 15:15:48 +00001894 offset = (unsigned long)((char*)next - (char*)list);
drh75897232000-05-29 14:26:00 +00001895 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1896 while( list ){
1897 ep = list;
1898 list = NEXT(list);
1899 NEXT(ep) = 0;
1900 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1901 ep = merge(ep,set[i],cmp,offset);
1902 set[i] = 0;
1903 }
1904 set[i] = ep;
1905 }
1906 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001907 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001908 return ep;
1909}
1910/************************ From the file "option.c" **************************/
mistachkind9bc6e82019-05-10 16:16:19 +00001911static char **g_argv;
drh75897232000-05-29 14:26:00 +00001912static struct s_options *op;
1913static FILE *errstream;
1914
1915#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1916
1917/*
1918** Print the command line with a carrot pointing to the k-th character
1919** of the n-th field.
1920*/
icculus9e44cf12010-02-14 17:14:22 +00001921static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001922{
1923 int spcnt, i;
mistachkind9bc6e82019-05-10 16:16:19 +00001924 if( g_argv[0] ) fprintf(err,"%s",g_argv[0]);
1925 spcnt = lemonStrlen(g_argv[0]) + 1;
1926 for(i=1; i<n && g_argv[i]; i++){
1927 fprintf(err," %s",g_argv[i]);
1928 spcnt += lemonStrlen(g_argv[i])+1;
drh75897232000-05-29 14:26:00 +00001929 }
1930 spcnt += k;
mistachkind9bc6e82019-05-10 16:16:19 +00001931 for(; g_argv[i]; i++) fprintf(err," %s",g_argv[i]);
drh75897232000-05-29 14:26:00 +00001932 if( spcnt<20 ){
1933 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1934 }else{
1935 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1936 }
1937}
1938
1939/*
1940** Return the index of the N-th non-switch argument. Return -1
1941** if N is out of range.
1942*/
icculus9e44cf12010-02-14 17:14:22 +00001943static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001944{
1945 int i;
1946 int dashdash = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00001947 if( g_argv!=0 && *g_argv!=0 ){
1948 for(i=1; g_argv[i]; i++){
1949 if( dashdash || !ISOPT(g_argv[i]) ){
drh75897232000-05-29 14:26:00 +00001950 if( n==0 ) return i;
1951 n--;
1952 }
mistachkind9bc6e82019-05-10 16:16:19 +00001953 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00001954 }
1955 }
1956 return -1;
1957}
1958
1959static char emsg[] = "Command line syntax error: ";
1960
1961/*
1962** Process a flag command line argument.
1963*/
icculus9e44cf12010-02-14 17:14:22 +00001964static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001965{
1966 int v;
1967 int errcnt = 0;
1968 int j;
1969 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00001970 if( strncmp(&g_argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001971 }
mistachkind9bc6e82019-05-10 16:16:19 +00001972 v = g_argv[i][0]=='-' ? 1 : 0;
drh75897232000-05-29 14:26:00 +00001973 if( op[j].label==0 ){
1974 if( err ){
1975 fprintf(err,"%sundefined option.\n",emsg);
1976 errline(i,1,err);
1977 }
1978 errcnt++;
drh0325d392015-01-01 19:11:22 +00001979 }else if( op[j].arg==0 ){
1980 /* Ignore this option */
drh75897232000-05-29 14:26:00 +00001981 }else if( op[j].type==OPT_FLAG ){
1982 *((int*)op[j].arg) = v;
1983 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001984 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001985 }else if( op[j].type==OPT_FSTR ){
mistachkind9bc6e82019-05-10 16:16:19 +00001986 (*(void(*)(char *))(op[j].arg))(&g_argv[i][2]);
drh75897232000-05-29 14:26:00 +00001987 }else{
1988 if( err ){
1989 fprintf(err,"%smissing argument on switch.\n",emsg);
1990 errline(i,1,err);
1991 }
1992 errcnt++;
1993 }
1994 return errcnt;
1995}
1996
1997/*
1998** Process a command line switch which has an argument.
1999*/
icculus9e44cf12010-02-14 17:14:22 +00002000static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00002001{
2002 int lv = 0;
2003 double dv = 0.0;
2004 char *sv = 0, *end;
2005 char *cp;
2006 int j;
2007 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002008 cp = strchr(g_argv[i],'=');
drh43617e92006-03-06 20:55:46 +00002009 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00002010 *cp = 0;
2011 for(j=0; op[j].label; j++){
mistachkind9bc6e82019-05-10 16:16:19 +00002012 if( strcmp(g_argv[i],op[j].label)==0 ) break;
drh75897232000-05-29 14:26:00 +00002013 }
2014 *cp = '=';
2015 if( op[j].label==0 ){
2016 if( err ){
2017 fprintf(err,"%sundefined option.\n",emsg);
2018 errline(i,0,err);
2019 }
2020 errcnt++;
2021 }else{
2022 cp++;
2023 switch( op[j].type ){
2024 case OPT_FLAG:
2025 case OPT_FFLAG:
2026 if( err ){
2027 fprintf(err,"%soption requires an argument.\n",emsg);
2028 errline(i,0,err);
2029 }
2030 errcnt++;
2031 break;
2032 case OPT_DBL:
2033 case OPT_FDBL:
2034 dv = strtod(cp,&end);
2035 if( *end ){
2036 if( err ){
drh25473362015-09-04 18:03:45 +00002037 fprintf(err,
2038 "%sillegal character in floating-point argument.\n",emsg);
mistachkind9bc6e82019-05-10 16:16:19 +00002039 errline(i,(int)((char*)end-(char*)g_argv[i]),err);
drh75897232000-05-29 14:26:00 +00002040 }
2041 errcnt++;
2042 }
2043 break;
2044 case OPT_INT:
2045 case OPT_FINT:
2046 lv = strtol(cp,&end,0);
2047 if( *end ){
2048 if( err ){
2049 fprintf(err,"%sillegal character in integer 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_STR:
2056 case OPT_FSTR:
2057 sv = cp;
2058 break;
2059 }
2060 switch( op[j].type ){
2061 case OPT_FLAG:
2062 case OPT_FFLAG:
2063 break;
2064 case OPT_DBL:
2065 *(double*)(op[j].arg) = dv;
2066 break;
2067 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00002068 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00002069 break;
2070 case OPT_INT:
2071 *(int*)(op[j].arg) = lv;
2072 break;
2073 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00002074 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00002075 break;
2076 case OPT_STR:
2077 *(char**)(op[j].arg) = sv;
2078 break;
2079 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00002080 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00002081 break;
2082 }
2083 }
2084 return errcnt;
2085}
2086
icculus9e44cf12010-02-14 17:14:22 +00002087int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00002088{
2089 int errcnt = 0;
mistachkind9bc6e82019-05-10 16:16:19 +00002090 g_argv = a;
drh75897232000-05-29 14:26:00 +00002091 op = o;
2092 errstream = err;
mistachkind9bc6e82019-05-10 16:16:19 +00002093 if( g_argv && *g_argv && op ){
drh75897232000-05-29 14:26:00 +00002094 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002095 for(i=1; g_argv[i]; i++){
2096 if( g_argv[i][0]=='+' || g_argv[i][0]=='-' ){
drh75897232000-05-29 14:26:00 +00002097 errcnt += handleflags(i,err);
mistachkind9bc6e82019-05-10 16:16:19 +00002098 }else if( strchr(g_argv[i],'=') ){
drh75897232000-05-29 14:26:00 +00002099 errcnt += handleswitch(i,err);
2100 }
2101 }
2102 }
2103 if( errcnt>0 ){
2104 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00002105 OptPrint();
drh75897232000-05-29 14:26:00 +00002106 exit(1);
2107 }
2108 return 0;
2109}
2110
drh14d88552017-04-14 19:44:15 +00002111int OptNArgs(void){
drh75897232000-05-29 14:26:00 +00002112 int cnt = 0;
2113 int dashdash = 0;
2114 int i;
mistachkind9bc6e82019-05-10 16:16:19 +00002115 if( g_argv!=0 && g_argv[0]!=0 ){
2116 for(i=1; g_argv[i]; i++){
2117 if( dashdash || !ISOPT(g_argv[i]) ) cnt++;
2118 if( strcmp(g_argv[i],"--")==0 ) dashdash = 1;
drh75897232000-05-29 14:26:00 +00002119 }
2120 }
2121 return cnt;
2122}
2123
icculus9e44cf12010-02-14 17:14:22 +00002124char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00002125{
2126 int i;
2127 i = argindex(n);
mistachkind9bc6e82019-05-10 16:16:19 +00002128 return i>=0 ? g_argv[i] : 0;
drh75897232000-05-29 14:26:00 +00002129}
2130
icculus9e44cf12010-02-14 17:14:22 +00002131void OptErr(int n)
drh75897232000-05-29 14:26:00 +00002132{
2133 int i;
2134 i = argindex(n);
2135 if( i>=0 ) errline(i,0,errstream);
2136}
2137
drh14d88552017-04-14 19:44:15 +00002138void OptPrint(void){
drh75897232000-05-29 14:26:00 +00002139 int i;
2140 int max, len;
2141 max = 0;
2142 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00002143 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00002144 switch( op[i].type ){
2145 case OPT_FLAG:
2146 case OPT_FFLAG:
2147 break;
2148 case OPT_INT:
2149 case OPT_FINT:
2150 len += 9; /* length of "<integer>" */
2151 break;
2152 case OPT_DBL:
2153 case OPT_FDBL:
2154 len += 6; /* length of "<real>" */
2155 break;
2156 case OPT_STR:
2157 case OPT_FSTR:
2158 len += 8; /* length of "<string>" */
2159 break;
2160 }
2161 if( len>max ) max = len;
2162 }
2163 for(i=0; op[i].label; i++){
2164 switch( op[i].type ){
2165 case OPT_FLAG:
2166 case OPT_FFLAG:
2167 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
2168 break;
2169 case OPT_INT:
2170 case OPT_FINT:
drh0325d392015-01-01 19:11:22 +00002171 fprintf(errstream," -%s<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002172 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002173 break;
2174 case OPT_DBL:
2175 case OPT_FDBL:
drh0325d392015-01-01 19:11:22 +00002176 fprintf(errstream," -%s<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002177 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002178 break;
2179 case OPT_STR:
2180 case OPT_FSTR:
drh0325d392015-01-01 19:11:22 +00002181 fprintf(errstream," -%s<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002182 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002183 break;
2184 }
2185 }
2186}
2187/*********************** From the file "parse.c" ****************************/
2188/*
2189** Input file parser for the LEMON parser generator.
2190*/
2191
2192/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00002193enum e_state {
2194 INITIALIZE,
2195 WAITING_FOR_DECL_OR_RULE,
2196 WAITING_FOR_DECL_KEYWORD,
2197 WAITING_FOR_DECL_ARG,
2198 WAITING_FOR_PRECEDENCE_SYMBOL,
2199 WAITING_FOR_ARROW,
2200 IN_RHS,
2201 LHS_ALIAS_1,
2202 LHS_ALIAS_2,
2203 LHS_ALIAS_3,
2204 RHS_ALIAS_1,
2205 RHS_ALIAS_2,
2206 PRECEDENCE_MARK_1,
2207 PRECEDENCE_MARK_2,
2208 RESYNC_AFTER_RULE_ERROR,
2209 RESYNC_AFTER_DECL_ERROR,
2210 WAITING_FOR_DESTRUCTOR_SYMBOL,
2211 WAITING_FOR_DATATYPE_SYMBOL,
2212 WAITING_FOR_FALLBACK_ID,
drh61f92cd2014-01-11 03:06:18 +00002213 WAITING_FOR_WILDCARD_ID,
2214 WAITING_FOR_CLASS_ID,
drh59c435a2017-08-02 03:21:11 +00002215 WAITING_FOR_CLASS_TOKEN,
2216 WAITING_FOR_TOKEN_NAME
icculus9e44cf12010-02-14 17:14:22 +00002217};
drh75897232000-05-29 14:26:00 +00002218struct pstate {
2219 char *filename; /* Name of the input file */
2220 int tokenlineno; /* Linenumber at which current token starts */
2221 int errorcnt; /* Number of errors so far */
2222 char *tokenstart; /* Text of current token */
2223 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00002224 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00002225 struct symbol *fallback; /* The fallback token */
drh61f92cd2014-01-11 03:06:18 +00002226 struct symbol *tkclass; /* Token class symbol */
drh75897232000-05-29 14:26:00 +00002227 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00002228 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00002229 int nrhs; /* Number of right-hand side symbols seen */
2230 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00002231 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00002232 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00002233 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00002234 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00002235 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00002236 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00002237 enum e_assoc declassoc; /* Assign this association to decl arguments */
2238 int preccounter; /* Assign this precedence to decl arguments */
2239 struct rule *firstrule; /* Pointer to first rule in the grammar */
2240 struct rule *lastrule; /* Pointer to the most recently parsed rule */
2241};
2242
2243/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00002244static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00002245{
icculus9e44cf12010-02-14 17:14:22 +00002246 const char *x;
drh75897232000-05-29 14:26:00 +00002247 x = Strsafe(psp->tokenstart); /* Save the token permanently */
2248#if 0
2249 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
2250 x,psp->state);
2251#endif
2252 switch( psp->state ){
2253 case INITIALIZE:
2254 psp->prevrule = 0;
2255 psp->preccounter = 0;
2256 psp->firstrule = psp->lastrule = 0;
2257 psp->gp->nrule = 0;
2258 /* Fall thru to next case */
2259 case WAITING_FOR_DECL_OR_RULE:
2260 if( x[0]=='%' ){
2261 psp->state = WAITING_FOR_DECL_KEYWORD;
drhc56fac72015-10-29 13:48:15 +00002262 }else if( ISLOWER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002263 psp->lhs = Symbol_new(x);
2264 psp->nrhs = 0;
2265 psp->lhsalias = 0;
2266 psp->state = WAITING_FOR_ARROW;
2267 }else if( x[0]=='{' ){
2268 if( psp->prevrule==0 ){
2269 ErrorMsg(psp->filename,psp->tokenlineno,
drh3cb2f6e2012-01-09 14:19:05 +00002270"There is no prior rule upon which to attach the code \
drh75897232000-05-29 14:26:00 +00002271fragment which begins on this line.");
2272 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002273 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002274 ErrorMsg(psp->filename,psp->tokenlineno,
2275"Code fragment beginning on this line is not the first \
2276to follow the previous rule.");
2277 psp->errorcnt++;
2278 }else{
2279 psp->prevrule->line = psp->tokenlineno;
2280 psp->prevrule->code = &x[1];
drh711c9812016-05-23 14:24:31 +00002281 psp->prevrule->noCode = 0;
drhf2f105d2012-08-20 15:53:54 +00002282 }
drh75897232000-05-29 14:26:00 +00002283 }else if( x[0]=='[' ){
2284 psp->state = PRECEDENCE_MARK_1;
2285 }else{
2286 ErrorMsg(psp->filename,psp->tokenlineno,
2287 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2288 x);
2289 psp->errorcnt++;
2290 }
2291 break;
2292 case PRECEDENCE_MARK_1:
drhc56fac72015-10-29 13:48:15 +00002293 if( !ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002294 ErrorMsg(psp->filename,psp->tokenlineno,
2295 "The precedence symbol must be a terminal.");
2296 psp->errorcnt++;
2297 }else if( psp->prevrule==0 ){
2298 ErrorMsg(psp->filename,psp->tokenlineno,
2299 "There is no prior rule to assign precedence \"[%s]\".",x);
2300 psp->errorcnt++;
2301 }else if( psp->prevrule->precsym!=0 ){
2302 ErrorMsg(psp->filename,psp->tokenlineno,
2303"Precedence mark on this line is not the first \
2304to follow the previous rule.");
2305 psp->errorcnt++;
2306 }else{
2307 psp->prevrule->precsym = Symbol_new(x);
2308 }
2309 psp->state = PRECEDENCE_MARK_2;
2310 break;
2311 case PRECEDENCE_MARK_2:
2312 if( x[0]!=']' ){
2313 ErrorMsg(psp->filename,psp->tokenlineno,
2314 "Missing \"]\" on precedence mark.");
2315 psp->errorcnt++;
2316 }
2317 psp->state = WAITING_FOR_DECL_OR_RULE;
2318 break;
2319 case WAITING_FOR_ARROW:
2320 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2321 psp->state = IN_RHS;
2322 }else if( x[0]=='(' ){
2323 psp->state = LHS_ALIAS_1;
2324 }else{
2325 ErrorMsg(psp->filename,psp->tokenlineno,
2326 "Expected to see a \":\" following the LHS symbol \"%s\".",
2327 psp->lhs->name);
2328 psp->errorcnt++;
2329 psp->state = RESYNC_AFTER_RULE_ERROR;
2330 }
2331 break;
2332 case LHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002333 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002334 psp->lhsalias = x;
2335 psp->state = LHS_ALIAS_2;
2336 }else{
2337 ErrorMsg(psp->filename,psp->tokenlineno,
2338 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2339 x,psp->lhs->name);
2340 psp->errorcnt++;
2341 psp->state = RESYNC_AFTER_RULE_ERROR;
2342 }
2343 break;
2344 case LHS_ALIAS_2:
2345 if( x[0]==')' ){
2346 psp->state = LHS_ALIAS_3;
2347 }else{
2348 ErrorMsg(psp->filename,psp->tokenlineno,
2349 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2350 psp->errorcnt++;
2351 psp->state = RESYNC_AFTER_RULE_ERROR;
2352 }
2353 break;
2354 case LHS_ALIAS_3:
2355 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2356 psp->state = IN_RHS;
2357 }else{
2358 ErrorMsg(psp->filename,psp->tokenlineno,
2359 "Missing \"->\" following: \"%s(%s)\".",
2360 psp->lhs->name,psp->lhsalias);
2361 psp->errorcnt++;
2362 psp->state = RESYNC_AFTER_RULE_ERROR;
2363 }
2364 break;
2365 case IN_RHS:
2366 if( x[0]=='.' ){
2367 struct rule *rp;
drh06f60d82017-04-14 19:46:12 +00002368 rp = (struct rule *)calloc( sizeof(struct rule) +
drh9892c5d2007-12-21 00:02:11 +00002369 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002370 if( rp==0 ){
2371 ErrorMsg(psp->filename,psp->tokenlineno,
2372 "Can't allocate enough memory for this rule.");
2373 psp->errorcnt++;
2374 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002375 }else{
drh75897232000-05-29 14:26:00 +00002376 int i;
2377 rp->ruleline = psp->tokenlineno;
2378 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002379 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002380 for(i=0; i<psp->nrhs; i++){
2381 rp->rhs[i] = psp->rhs[i];
2382 rp->rhsalias[i] = psp->alias[i];
drh539e7412018-04-21 20:24:19 +00002383 if( rp->rhsalias[i]!=0 ){ rp->rhs[i]->bContent = 1; }
drhf2f105d2012-08-20 15:53:54 +00002384 }
drh75897232000-05-29 14:26:00 +00002385 rp->lhs = psp->lhs;
2386 rp->lhsalias = psp->lhsalias;
2387 rp->nrhs = psp->nrhs;
2388 rp->code = 0;
drh711c9812016-05-23 14:24:31 +00002389 rp->noCode = 1;
drh75897232000-05-29 14:26:00 +00002390 rp->precsym = 0;
2391 rp->index = psp->gp->nrule++;
2392 rp->nextlhs = rp->lhs->rule;
2393 rp->lhs->rule = rp;
2394 rp->next = 0;
2395 if( psp->firstrule==0 ){
2396 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002397 }else{
drh75897232000-05-29 14:26:00 +00002398 psp->lastrule->next = rp;
2399 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002400 }
drh75897232000-05-29 14:26:00 +00002401 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002402 }
drh75897232000-05-29 14:26:00 +00002403 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002404 }else if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002405 if( psp->nrhs>=MAXRHS ){
2406 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002407 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002408 x);
2409 psp->errorcnt++;
2410 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002411 }else{
drh75897232000-05-29 14:26:00 +00002412 psp->rhs[psp->nrhs] = Symbol_new(x);
2413 psp->alias[psp->nrhs] = 0;
2414 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002415 }
drhfd405312005-11-06 04:06:59 +00002416 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2417 struct symbol *msp = psp->rhs[psp->nrhs-1];
2418 if( msp->type!=MULTITERMINAL ){
2419 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002420 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002421 memset(msp, 0, sizeof(*msp));
2422 msp->type = MULTITERMINAL;
2423 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002424 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002425 msp->subsym[0] = origsp;
2426 msp->name = origsp->name;
2427 psp->rhs[psp->nrhs-1] = msp;
2428 }
2429 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002430 msp->subsym = (struct symbol **) realloc(msp->subsym,
2431 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002432 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
drhc56fac72015-10-29 13:48:15 +00002433 if( ISLOWER(x[1]) || ISLOWER(msp->subsym[0]->name[0]) ){
drhfd405312005-11-06 04:06:59 +00002434 ErrorMsg(psp->filename,psp->tokenlineno,
2435 "Cannot form a compound containing a non-terminal");
2436 psp->errorcnt++;
2437 }
drh75897232000-05-29 14:26:00 +00002438 }else if( x[0]=='(' && psp->nrhs>0 ){
2439 psp->state = RHS_ALIAS_1;
2440 }else{
2441 ErrorMsg(psp->filename,psp->tokenlineno,
2442 "Illegal character on RHS of rule: \"%s\".",x);
2443 psp->errorcnt++;
2444 psp->state = RESYNC_AFTER_RULE_ERROR;
2445 }
2446 break;
2447 case RHS_ALIAS_1:
drhc56fac72015-10-29 13:48:15 +00002448 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002449 psp->alias[psp->nrhs-1] = x;
2450 psp->state = RHS_ALIAS_2;
2451 }else{
2452 ErrorMsg(psp->filename,psp->tokenlineno,
2453 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2454 x,psp->rhs[psp->nrhs-1]->name);
2455 psp->errorcnt++;
2456 psp->state = RESYNC_AFTER_RULE_ERROR;
2457 }
2458 break;
2459 case RHS_ALIAS_2:
2460 if( x[0]==')' ){
2461 psp->state = IN_RHS;
2462 }else{
2463 ErrorMsg(psp->filename,psp->tokenlineno,
2464 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2465 psp->errorcnt++;
2466 psp->state = RESYNC_AFTER_RULE_ERROR;
2467 }
2468 break;
2469 case WAITING_FOR_DECL_KEYWORD:
drhc56fac72015-10-29 13:48:15 +00002470 if( ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002471 psp->declkeyword = x;
2472 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002473 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002474 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002475 psp->state = WAITING_FOR_DECL_ARG;
2476 if( strcmp(x,"name")==0 ){
2477 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002478 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002479 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002480 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002481 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002482 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002483 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002484 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002485 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002486 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002487 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002488 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002489 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002490 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002491 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002492 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002493 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002494 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002495 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002496 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002497 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002498 }else if( strcmp(x,"extra_argument")==0 ){
2499 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002500 psp->insertLineMacro = 0;
drhfb32c442018-04-21 13:51:42 +00002501 }else if( strcmp(x,"extra_context")==0 ){
2502 psp->declargslot = &(psp->gp->ctx);
2503 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002504 }else if( strcmp(x,"token_type")==0 ){
2505 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002506 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002507 }else if( strcmp(x,"default_type")==0 ){
2508 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002509 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002510 }else if( strcmp(x,"stack_size")==0 ){
2511 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002512 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002513 }else if( strcmp(x,"start_symbol")==0 ){
2514 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002515 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002516 }else if( strcmp(x,"left")==0 ){
2517 psp->preccounter++;
2518 psp->declassoc = LEFT;
2519 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2520 }else if( strcmp(x,"right")==0 ){
2521 psp->preccounter++;
2522 psp->declassoc = RIGHT;
2523 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2524 }else if( strcmp(x,"nonassoc")==0 ){
2525 psp->preccounter++;
2526 psp->declassoc = NONE;
2527 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002528 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002529 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002530 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002531 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002532 }else if( strcmp(x,"fallback")==0 ){
2533 psp->fallback = 0;
2534 psp->state = WAITING_FOR_FALLBACK_ID;
drh59c435a2017-08-02 03:21:11 +00002535 }else if( strcmp(x,"token")==0 ){
2536 psp->state = WAITING_FOR_TOKEN_NAME;
drhe09daa92006-06-10 13:29:31 +00002537 }else if( strcmp(x,"wildcard")==0 ){
2538 psp->state = WAITING_FOR_WILDCARD_ID;
drh61f92cd2014-01-11 03:06:18 +00002539 }else if( strcmp(x,"token_class")==0 ){
2540 psp->state = WAITING_FOR_CLASS_ID;
drh75897232000-05-29 14:26:00 +00002541 }else{
2542 ErrorMsg(psp->filename,psp->tokenlineno,
2543 "Unknown declaration keyword: \"%%%s\".",x);
2544 psp->errorcnt++;
2545 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002546 }
drh75897232000-05-29 14:26:00 +00002547 }else{
2548 ErrorMsg(psp->filename,psp->tokenlineno,
2549 "Illegal declaration keyword: \"%s\".",x);
2550 psp->errorcnt++;
2551 psp->state = RESYNC_AFTER_DECL_ERROR;
2552 }
2553 break;
2554 case WAITING_FOR_DESTRUCTOR_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002555 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002556 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002557 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002558 psp->errorcnt++;
2559 psp->state = RESYNC_AFTER_DECL_ERROR;
2560 }else{
icculusd286fa62010-03-03 17:06:32 +00002561 struct symbol *sp = Symbol_new(x);
2562 psp->declargslot = &sp->destructor;
2563 psp->decllinenoslot = &sp->destLineno;
2564 psp->insertLineMacro = 1;
2565 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002566 }
2567 break;
2568 case WAITING_FOR_DATATYPE_SYMBOL:
drhc56fac72015-10-29 13:48:15 +00002569 if( !ISALPHA(x[0]) ){
drh75897232000-05-29 14:26:00 +00002570 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002571 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002572 psp->errorcnt++;
2573 psp->state = RESYNC_AFTER_DECL_ERROR;
2574 }else{
icculus866bf1e2010-02-17 20:31:32 +00002575 struct symbol *sp = Symbol_find(x);
2576 if((sp) && (sp->datatype)){
2577 ErrorMsg(psp->filename,psp->tokenlineno,
2578 "Symbol %%type \"%s\" already defined", x);
2579 psp->errorcnt++;
2580 psp->state = RESYNC_AFTER_DECL_ERROR;
2581 }else{
2582 if (!sp){
2583 sp = Symbol_new(x);
2584 }
2585 psp->declargslot = &sp->datatype;
2586 psp->insertLineMacro = 0;
2587 psp->state = WAITING_FOR_DECL_ARG;
2588 }
drh75897232000-05-29 14:26:00 +00002589 }
2590 break;
2591 case WAITING_FOR_PRECEDENCE_SYMBOL:
2592 if( x[0]=='.' ){
2593 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002594 }else if( ISUPPER(x[0]) ){
drh75897232000-05-29 14:26:00 +00002595 struct symbol *sp;
2596 sp = Symbol_new(x);
2597 if( sp->prec>=0 ){
2598 ErrorMsg(psp->filename,psp->tokenlineno,
2599 "Symbol \"%s\" has already be given a precedence.",x);
2600 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002601 }else{
drh75897232000-05-29 14:26:00 +00002602 sp->prec = psp->preccounter;
2603 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002604 }
drh75897232000-05-29 14:26:00 +00002605 }else{
2606 ErrorMsg(psp->filename,psp->tokenlineno,
2607 "Can't assign a precedence to \"%s\".",x);
2608 psp->errorcnt++;
2609 }
2610 break;
2611 case WAITING_FOR_DECL_ARG:
drhc56fac72015-10-29 13:48:15 +00002612 if( x[0]=='{' || x[0]=='\"' || ISALNUM(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002613 const char *zOld, *zNew;
2614 char *zBuf, *z;
mistachkin2318d332015-01-12 18:02:52 +00002615 int nOld, n, nLine = 0, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002616 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002617 char zLine[50];
2618 zNew = x;
2619 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002620 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002621 if( *psp->declargslot ){
2622 zOld = *psp->declargslot;
2623 }else{
2624 zOld = "";
2625 }
drh87cf1372008-08-13 20:09:06 +00002626 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002627 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002628 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002629 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2630 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002631 for(z=psp->filename, nBack=0; *z; z++){
2632 if( *z=='\\' ) nBack++;
2633 }
drh898799f2014-01-10 23:21:00 +00002634 lemon_sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002635 nLine = lemonStrlen(zLine);
2636 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002637 }
icculus9e44cf12010-02-14 17:14:22 +00002638 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2639 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002640 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002641 if( nOld && zBuf[-1]!='\n' ){
2642 *(zBuf++) = '\n';
2643 }
2644 memcpy(zBuf, zLine, nLine);
2645 zBuf += nLine;
2646 *(zBuf++) = '"';
2647 for(z=psp->filename; *z; z++){
2648 if( *z=='\\' ){
2649 *(zBuf++) = '\\';
2650 }
2651 *(zBuf++) = *z;
2652 }
2653 *(zBuf++) = '"';
2654 *(zBuf++) = '\n';
2655 }
drh4dc8ef52008-07-01 17:13:57 +00002656 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2657 psp->decllinenoslot[0] = psp->tokenlineno;
2658 }
drha5808f32008-04-27 22:19:44 +00002659 memcpy(zBuf, zNew, nNew);
2660 zBuf += nNew;
2661 *zBuf = 0;
2662 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002663 }else{
2664 ErrorMsg(psp->filename,psp->tokenlineno,
2665 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2666 psp->errorcnt++;
2667 psp->state = RESYNC_AFTER_DECL_ERROR;
2668 }
2669 break;
drh0bd1f4e2002-06-06 18:54:39 +00002670 case WAITING_FOR_FALLBACK_ID:
2671 if( x[0]=='.' ){
2672 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002673 }else if( !ISUPPER(x[0]) ){
drh0bd1f4e2002-06-06 18:54:39 +00002674 ErrorMsg(psp->filename, psp->tokenlineno,
2675 "%%fallback argument \"%s\" should be a token", x);
2676 psp->errorcnt++;
2677 }else{
2678 struct symbol *sp = Symbol_new(x);
2679 if( psp->fallback==0 ){
2680 psp->fallback = sp;
2681 }else if( sp->fallback ){
2682 ErrorMsg(psp->filename, psp->tokenlineno,
2683 "More than one fallback assigned to token %s", x);
2684 psp->errorcnt++;
2685 }else{
2686 sp->fallback = psp->fallback;
2687 psp->gp->has_fallback = 1;
2688 }
2689 }
2690 break;
drh59c435a2017-08-02 03:21:11 +00002691 case WAITING_FOR_TOKEN_NAME:
2692 /* Tokens do not have to be declared before use. But they can be
2693 ** in order to control their assigned integer number. The number for
2694 ** each token is assigned when it is first seen. So by including
2695 **
2696 ** %token ONE TWO THREE
2697 **
2698 ** early in the grammar file, that assigns small consecutive values
2699 ** to each of the tokens ONE TWO and THREE.
2700 */
2701 if( x[0]=='.' ){
2702 psp->state = WAITING_FOR_DECL_OR_RULE;
2703 }else if( !ISUPPER(x[0]) ){
2704 ErrorMsg(psp->filename, psp->tokenlineno,
2705 "%%token argument \"%s\" should be a token", x);
2706 psp->errorcnt++;
2707 }else{
2708 (void)Symbol_new(x);
2709 }
2710 break;
drhe09daa92006-06-10 13:29:31 +00002711 case WAITING_FOR_WILDCARD_ID:
2712 if( x[0]=='.' ){
2713 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002714 }else if( !ISUPPER(x[0]) ){
drhe09daa92006-06-10 13:29:31 +00002715 ErrorMsg(psp->filename, psp->tokenlineno,
2716 "%%wildcard argument \"%s\" should be a token", x);
2717 psp->errorcnt++;
2718 }else{
2719 struct symbol *sp = Symbol_new(x);
2720 if( psp->gp->wildcard==0 ){
2721 psp->gp->wildcard = sp;
2722 }else{
2723 ErrorMsg(psp->filename, psp->tokenlineno,
2724 "Extra wildcard to token: %s", x);
2725 psp->errorcnt++;
2726 }
2727 }
2728 break;
drh61f92cd2014-01-11 03:06:18 +00002729 case WAITING_FOR_CLASS_ID:
drhc56fac72015-10-29 13:48:15 +00002730 if( !ISLOWER(x[0]) ){
drh61f92cd2014-01-11 03:06:18 +00002731 ErrorMsg(psp->filename, psp->tokenlineno,
mistachkind9bc6e82019-05-10 16:16:19 +00002732 "%%token_class must be followed by an identifier: %s", x);
drh61f92cd2014-01-11 03:06:18 +00002733 psp->errorcnt++;
2734 psp->state = RESYNC_AFTER_DECL_ERROR;
2735 }else if( Symbol_find(x) ){
2736 ErrorMsg(psp->filename, psp->tokenlineno,
2737 "Symbol \"%s\" already used", x);
2738 psp->errorcnt++;
2739 psp->state = RESYNC_AFTER_DECL_ERROR;
2740 }else{
2741 psp->tkclass = Symbol_new(x);
2742 psp->tkclass->type = MULTITERMINAL;
2743 psp->state = WAITING_FOR_CLASS_TOKEN;
2744 }
2745 break;
2746 case WAITING_FOR_CLASS_TOKEN:
2747 if( x[0]=='.' ){
2748 psp->state = WAITING_FOR_DECL_OR_RULE;
drhc56fac72015-10-29 13:48:15 +00002749 }else if( ISUPPER(x[0]) || ((x[0]=='|' || x[0]=='/') && ISUPPER(x[1])) ){
drh61f92cd2014-01-11 03:06:18 +00002750 struct symbol *msp = psp->tkclass;
2751 msp->nsubsym++;
2752 msp->subsym = (struct symbol **) realloc(msp->subsym,
2753 sizeof(struct symbol*)*msp->nsubsym);
drhc56fac72015-10-29 13:48:15 +00002754 if( !ISUPPER(x[0]) ) x++;
drh61f92cd2014-01-11 03:06:18 +00002755 msp->subsym[msp->nsubsym-1] = Symbol_new(x);
2756 }else{
2757 ErrorMsg(psp->filename, psp->tokenlineno,
2758 "%%token_class argument \"%s\" should be a token", x);
2759 psp->errorcnt++;
2760 psp->state = RESYNC_AFTER_DECL_ERROR;
2761 }
2762 break;
drh75897232000-05-29 14:26:00 +00002763 case RESYNC_AFTER_RULE_ERROR:
2764/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2765** break; */
2766 case RESYNC_AFTER_DECL_ERROR:
2767 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2768 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2769 break;
2770 }
2771}
2772
drh34ff57b2008-07-14 12:27:51 +00002773/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002774** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2775** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2776** comments them out. Text in between is also commented out as appropriate.
2777*/
danielk1977940fac92005-01-23 22:41:37 +00002778static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002779 int i, j, k, n;
2780 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002781 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002782 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002783 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002784 for(i=0; z[i]; i++){
2785 if( z[i]=='\n' ) lineno++;
2786 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
drhc56fac72015-10-29 13:48:15 +00002787 if( strncmp(&z[i],"%endif",6)==0 && ISSPACE(z[i+6]) ){
drh6d08b4d2004-07-20 12:45:22 +00002788 if( exclude ){
2789 exclude--;
2790 if( exclude==0 ){
2791 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2792 }
2793 }
2794 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
drhc56fac72015-10-29 13:48:15 +00002795 }else if( (strncmp(&z[i],"%ifdef",6)==0 && ISSPACE(z[i+6]))
2796 || (strncmp(&z[i],"%ifndef",7)==0 && ISSPACE(z[i+7])) ){
drh6d08b4d2004-07-20 12:45:22 +00002797 if( exclude ){
2798 exclude++;
2799 }else{
drhc56fac72015-10-29 13:48:15 +00002800 for(j=i+7; ISSPACE(z[j]); j++){}
2801 for(n=0; z[j+n] && !ISSPACE(z[j+n]); n++){}
drh6d08b4d2004-07-20 12:45:22 +00002802 exclude = 1;
2803 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002804 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002805 exclude = 0;
2806 break;
2807 }
2808 }
2809 if( z[i+3]=='n' ) exclude = !exclude;
2810 if( exclude ){
2811 start = i;
2812 start_lineno = lineno;
2813 }
2814 }
2815 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2816 }
2817 }
2818 if( exclude ){
2819 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2820 exit(1);
2821 }
2822}
2823
drh75897232000-05-29 14:26:00 +00002824/* In spite of its name, this function is really a scanner. It read
2825** in the entire input file (all at once) then tokenizes it. Each
2826** token is passed to the function "parseonetoken" which builds all
2827** the appropriate data structures in the global state vector "gp".
2828*/
icculus9e44cf12010-02-14 17:14:22 +00002829void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002830{
2831 struct pstate ps;
2832 FILE *fp;
2833 char *filebuf;
mistachkin2318d332015-01-12 18:02:52 +00002834 unsigned int filesize;
drh75897232000-05-29 14:26:00 +00002835 int lineno;
2836 int c;
2837 char *cp, *nextcp;
2838 int startline = 0;
2839
rse38514a92007-09-20 11:34:17 +00002840 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002841 ps.gp = gp;
2842 ps.filename = gp->filename;
2843 ps.errorcnt = 0;
2844 ps.state = INITIALIZE;
2845
2846 /* Begin by reading the input file */
2847 fp = fopen(ps.filename,"rb");
2848 if( fp==0 ){
2849 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2850 gp->errorcnt++;
2851 return;
2852 }
2853 fseek(fp,0,2);
2854 filesize = ftell(fp);
2855 rewind(fp);
2856 filebuf = (char *)malloc( filesize+1 );
drh03e1b1f2014-01-11 12:52:25 +00002857 if( filesize>100000000 || filebuf==0 ){
2858 ErrorMsg(ps.filename,0,"Input file too large.");
mistachkinc93c6142018-09-08 16:55:18 +00002859 free(filebuf);
drh75897232000-05-29 14:26:00 +00002860 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002861 fclose(fp);
drh75897232000-05-29 14:26:00 +00002862 return;
2863 }
2864 if( fread(filebuf,1,filesize,fp)!=filesize ){
2865 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2866 filesize);
2867 free(filebuf);
2868 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002869 fclose(fp);
drh75897232000-05-29 14:26:00 +00002870 return;
2871 }
2872 fclose(fp);
2873 filebuf[filesize] = 0;
2874
drh6d08b4d2004-07-20 12:45:22 +00002875 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2876 preprocess_input(filebuf);
2877
drh75897232000-05-29 14:26:00 +00002878 /* Now scan the text of the input file */
2879 lineno = 1;
2880 for(cp=filebuf; (c= *cp)!=0; ){
2881 if( c=='\n' ) lineno++; /* Keep track of the line number */
drhc56fac72015-10-29 13:48:15 +00002882 if( ISSPACE(c) ){ cp++; continue; } /* Skip all white space */
drh75897232000-05-29 14:26:00 +00002883 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2884 cp+=2;
2885 while( (c= *cp)!=0 && c!='\n' ) cp++;
2886 continue;
2887 }
2888 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2889 cp+=2;
2890 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2891 if( c=='\n' ) lineno++;
2892 cp++;
2893 }
2894 if( c ) cp++;
2895 continue;
2896 }
2897 ps.tokenstart = cp; /* Mark the beginning of the token */
2898 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2899 if( c=='\"' ){ /* String literals */
2900 cp++;
2901 while( (c= *cp)!=0 && c!='\"' ){
2902 if( c=='\n' ) lineno++;
2903 cp++;
2904 }
2905 if( c==0 ){
2906 ErrorMsg(ps.filename,startline,
2907"String starting on this line is not terminated before the end of the file.");
2908 ps.errorcnt++;
2909 nextcp = cp;
2910 }else{
2911 nextcp = cp+1;
2912 }
2913 }else if( c=='{' ){ /* A block of C code */
2914 int level;
2915 cp++;
2916 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2917 if( c=='\n' ) lineno++;
2918 else if( c=='{' ) level++;
2919 else if( c=='}' ) level--;
2920 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2921 int prevc;
2922 cp = &cp[2];
2923 prevc = 0;
2924 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2925 if( c=='\n' ) lineno++;
2926 prevc = c;
2927 cp++;
drhf2f105d2012-08-20 15:53:54 +00002928 }
2929 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00002930 cp = &cp[2];
2931 while( (c= *cp)!=0 && c!='\n' ) cp++;
2932 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00002933 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00002934 int startchar, prevc;
2935 startchar = c;
2936 prevc = 0;
2937 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2938 if( c=='\n' ) lineno++;
2939 if( prevc=='\\' ) prevc = 0;
2940 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00002941 }
2942 }
drh75897232000-05-29 14:26:00 +00002943 }
2944 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002945 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002946"C code starting on this line is not terminated before the end of the file.");
2947 ps.errorcnt++;
2948 nextcp = cp;
2949 }else{
2950 nextcp = cp+1;
2951 }
drhc56fac72015-10-29 13:48:15 +00002952 }else if( ISALNUM(c) ){ /* Identifiers */
2953 while( (c= *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drh75897232000-05-29 14:26:00 +00002954 nextcp = cp;
2955 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2956 cp += 3;
2957 nextcp = cp;
drhc56fac72015-10-29 13:48:15 +00002958 }else if( (c=='/' || c=='|') && ISALPHA(cp[1]) ){
drhfd405312005-11-06 04:06:59 +00002959 cp += 2;
drhc56fac72015-10-29 13:48:15 +00002960 while( (c = *cp)!=0 && (ISALNUM(c) || c=='_') ) cp++;
drhfd405312005-11-06 04:06:59 +00002961 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002962 }else{ /* All other (one character) operators */
2963 cp++;
2964 nextcp = cp;
2965 }
2966 c = *cp;
2967 *cp = 0; /* Null terminate the token */
2968 parseonetoken(&ps); /* Parse the token */
mistachkin2318d332015-01-12 18:02:52 +00002969 *cp = (char)c; /* Restore the buffer */
drh75897232000-05-29 14:26:00 +00002970 cp = nextcp;
2971 }
2972 free(filebuf); /* Release the buffer after parsing */
2973 gp->rule = ps.firstrule;
2974 gp->errorcnt = ps.errorcnt;
2975}
2976/*************************** From the file "plink.c" *********************/
2977/*
2978** Routines processing configuration follow-set propagation links
2979** in the LEMON parser generator.
2980*/
2981static struct plink *plink_freelist = 0;
2982
2983/* Allocate a new plink */
drh14d88552017-04-14 19:44:15 +00002984struct plink *Plink_new(void){
icculus9e44cf12010-02-14 17:14:22 +00002985 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00002986
2987 if( plink_freelist==0 ){
2988 int i;
2989 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002990 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002991 if( plink_freelist==0 ){
2992 fprintf(stderr,
2993 "Unable to allocate memory for a new follow-set propagation link.\n");
2994 exit(1);
2995 }
2996 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2997 plink_freelist[amt-1].next = 0;
2998 }
icculus9e44cf12010-02-14 17:14:22 +00002999 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00003000 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00003001 return newlink;
drh75897232000-05-29 14:26:00 +00003002}
3003
3004/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00003005void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00003006{
icculus9e44cf12010-02-14 17:14:22 +00003007 struct plink *newlink;
3008 newlink = Plink_new();
3009 newlink->next = *plpp;
3010 *plpp = newlink;
3011 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00003012}
3013
3014/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00003015void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00003016{
3017 struct plink *nextpl;
3018 while( from ){
3019 nextpl = from->next;
3020 from->next = *to;
3021 *to = from;
3022 from = nextpl;
3023 }
3024}
3025
3026/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00003027void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00003028{
3029 struct plink *nextpl;
3030
3031 while( plp ){
3032 nextpl = plp->next;
3033 plp->next = plink_freelist;
3034 plink_freelist = plp;
3035 plp = nextpl;
3036 }
3037}
3038/*********************** From the file "report.c" **************************/
3039/*
3040** Procedures for generating reports and tables in the LEMON parser generator.
3041*/
3042
3043/* Generate a filename with the given suffix. Space to hold the
3044** name comes from malloc() and must be freed by the calling
3045** function.
3046*/
icculus9e44cf12010-02-14 17:14:22 +00003047PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00003048{
3049 char *name;
3050 char *cp;
drh9f88e6d2018-04-20 20:47:49 +00003051 char *filename = lemp->filename;
3052 int sz;
drh75897232000-05-29 14:26:00 +00003053
drh9f88e6d2018-04-20 20:47:49 +00003054 if( outputDir ){
3055 cp = strrchr(filename, '/');
3056 if( cp ) filename = cp + 1;
3057 }
3058 sz = lemonStrlen(filename);
3059 sz += lemonStrlen(suffix);
3060 if( outputDir ) sz += lemonStrlen(outputDir) + 1;
3061 sz += 5;
3062 name = (char*)malloc( sz );
drh75897232000-05-29 14:26:00 +00003063 if( name==0 ){
3064 fprintf(stderr,"Can't allocate space for a filename.\n");
3065 exit(1);
3066 }
drh9f88e6d2018-04-20 20:47:49 +00003067 name[0] = 0;
3068 if( outputDir ){
3069 lemon_strcpy(name, outputDir);
3070 lemon_strcat(name, "/");
3071 }
3072 lemon_strcat(name,filename);
drh75897232000-05-29 14:26:00 +00003073 cp = strrchr(name,'.');
3074 if( cp ) *cp = 0;
drh898799f2014-01-10 23:21:00 +00003075 lemon_strcat(name,suffix);
drh75897232000-05-29 14:26:00 +00003076 return name;
3077}
3078
3079/* Open a file with a name based on the name of the input file,
3080** but with a different (specified) suffix, and return a pointer
3081** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00003082PRIVATE FILE *file_open(
3083 struct lemon *lemp,
3084 const char *suffix,
3085 const char *mode
3086){
drh75897232000-05-29 14:26:00 +00003087 FILE *fp;
3088
3089 if( lemp->outname ) free(lemp->outname);
3090 lemp->outname = file_makename(lemp, suffix);
3091 fp = fopen(lemp->outname,mode);
3092 if( fp==0 && *mode=='w' ){
3093 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
3094 lemp->errorcnt++;
3095 return 0;
3096 }
3097 return fp;
3098}
3099
drh5c8241b2017-12-24 23:38:10 +00003100/* Print the text of a rule
3101*/
3102void rule_print(FILE *out, struct rule *rp){
3103 int i, j;
3104 fprintf(out, "%s",rp->lhs->name);
3105 /* if( rp->lhsalias ) fprintf(out,"(%s)",rp->lhsalias); */
3106 fprintf(out," ::=");
3107 for(i=0; i<rp->nrhs; i++){
3108 struct symbol *sp = rp->rhs[i];
3109 if( sp->type==MULTITERMINAL ){
3110 fprintf(out," %s", sp->subsym[0]->name);
3111 for(j=1; j<sp->nsubsym; j++){
3112 fprintf(out,"|%s", sp->subsym[j]->name);
3113 }
3114 }else{
3115 fprintf(out," %s", sp->name);
3116 }
3117 /* if( rp->rhsalias[i] ) fprintf(out,"(%s)",rp->rhsalias[i]); */
3118 }
3119}
3120
drh06f60d82017-04-14 19:46:12 +00003121/* Duplicate the input file without comments and without actions
drh75897232000-05-29 14:26:00 +00003122** on rules */
icculus9e44cf12010-02-14 17:14:22 +00003123void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003124{
3125 struct rule *rp;
3126 struct symbol *sp;
3127 int i, j, maxlen, len, ncolumns, skip;
3128 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
3129 maxlen = 10;
3130 for(i=0; i<lemp->nsymbol; i++){
3131 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00003132 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00003133 if( len>maxlen ) maxlen = len;
3134 }
3135 ncolumns = 76/(maxlen+5);
3136 if( ncolumns<1 ) ncolumns = 1;
3137 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
3138 for(i=0; i<skip; i++){
3139 printf("//");
3140 for(j=i; j<lemp->nsymbol; j+=skip){
3141 sp = lemp->symbols[j];
3142 assert( sp->index==j );
3143 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
3144 }
3145 printf("\n");
3146 }
3147 for(rp=lemp->rule; rp; rp=rp->next){
drh5c8241b2017-12-24 23:38:10 +00003148 rule_print(stdout, rp);
drh75897232000-05-29 14:26:00 +00003149 printf(".");
3150 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00003151 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00003152 printf("\n");
3153 }
3154}
3155
drh7e698e92015-09-07 14:22:24 +00003156/* Print a single rule.
3157*/
3158void RulePrint(FILE *fp, struct rule *rp, int iCursor){
drhfd405312005-11-06 04:06:59 +00003159 struct symbol *sp;
3160 int i, j;
drh75897232000-05-29 14:26:00 +00003161 fprintf(fp,"%s ::=",rp->lhs->name);
3162 for(i=0; i<=rp->nrhs; i++){
drh7e698e92015-09-07 14:22:24 +00003163 if( i==iCursor ) fprintf(fp," *");
drh75897232000-05-29 14:26:00 +00003164 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00003165 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00003166 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00003167 fprintf(fp," %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00003168 for(j=1; j<sp->nsubsym; j++){
3169 fprintf(fp,"|%s",sp->subsym[j]->name);
3170 }
drh61f92cd2014-01-11 03:06:18 +00003171 }else{
3172 fprintf(fp," %s", sp->name);
drhfd405312005-11-06 04:06:59 +00003173 }
drh75897232000-05-29 14:26:00 +00003174 }
3175}
3176
drh7e698e92015-09-07 14:22:24 +00003177/* Print the rule for a configuration.
3178*/
3179void ConfigPrint(FILE *fp, struct config *cfp){
3180 RulePrint(fp, cfp->rp, cfp->dot);
3181}
3182
drh75897232000-05-29 14:26:00 +00003183/* #define TEST */
drhfd405312005-11-06 04:06:59 +00003184#if 0
drh75897232000-05-29 14:26:00 +00003185/* Print a set */
3186PRIVATE void SetPrint(out,set,lemp)
3187FILE *out;
3188char *set;
3189struct lemon *lemp;
3190{
3191 int i;
3192 char *spacer;
3193 spacer = "";
3194 fprintf(out,"%12s[","");
3195 for(i=0; i<lemp->nterminal; i++){
3196 if( SetFind(set,i) ){
3197 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
3198 spacer = " ";
3199 }
3200 }
3201 fprintf(out,"]\n");
3202}
3203
3204/* Print a plink chain */
3205PRIVATE void PlinkPrint(out,plp,tag)
3206FILE *out;
3207struct plink *plp;
3208char *tag;
3209{
3210 while( plp ){
drhada354d2005-11-05 15:03:59 +00003211 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00003212 ConfigPrint(out,plp->cfp);
3213 fprintf(out,"\n");
3214 plp = plp->next;
3215 }
3216}
3217#endif
3218
3219/* Print an action to the given file descriptor. Return FALSE if
3220** nothing was actually printed.
3221*/
drh7e698e92015-09-07 14:22:24 +00003222int PrintAction(
3223 struct action *ap, /* The action to print */
3224 FILE *fp, /* Print the action here */
drh3bd48ab2015-09-07 18:23:37 +00003225 int indent /* Indent by this amount */
drh7e698e92015-09-07 14:22:24 +00003226){
drh75897232000-05-29 14:26:00 +00003227 int result = 1;
3228 switch( ap->type ){
drh7e698e92015-09-07 14:22:24 +00003229 case SHIFT: {
3230 struct state *stp = ap->x.stp;
drh3bd48ab2015-09-07 18:23:37 +00003231 fprintf(fp,"%*s shift %-7d",indent,ap->sp->name,stp->statenum);
drh75897232000-05-29 14:26:00 +00003232 break;
drh7e698e92015-09-07 14:22:24 +00003233 }
3234 case REDUCE: {
3235 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003236 fprintf(fp,"%*s reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003237 RulePrint(fp, rp, -1);
3238 break;
3239 }
3240 case SHIFTREDUCE: {
3241 struct rule *rp = ap->x.rp;
drh4ef07702016-03-16 19:45:54 +00003242 fprintf(fp,"%*s shift-reduce %-7d",indent,ap->sp->name,rp->iRule);
drh3bd48ab2015-09-07 18:23:37 +00003243 RulePrint(fp, rp, -1);
drh75897232000-05-29 14:26:00 +00003244 break;
drh7e698e92015-09-07 14:22:24 +00003245 }
drh75897232000-05-29 14:26:00 +00003246 case ACCEPT:
3247 fprintf(fp,"%*s accept",indent,ap->sp->name);
3248 break;
3249 case ERROR:
3250 fprintf(fp,"%*s error",indent,ap->sp->name);
3251 break;
drh9892c5d2007-12-21 00:02:11 +00003252 case SRCONFLICT:
3253 case RRCONFLICT:
drh3bd48ab2015-09-07 18:23:37 +00003254 fprintf(fp,"%*s reduce %-7d ** Parsing conflict **",
drh4ef07702016-03-16 19:45:54 +00003255 indent,ap->sp->name,ap->x.rp->iRule);
drh75897232000-05-29 14:26:00 +00003256 break;
drh9892c5d2007-12-21 00:02:11 +00003257 case SSCONFLICT:
drh06f60d82017-04-14 19:46:12 +00003258 fprintf(fp,"%*s shift %-7d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00003259 indent,ap->sp->name,ap->x.stp->statenum);
3260 break;
drh75897232000-05-29 14:26:00 +00003261 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003262 if( showPrecedenceConflict ){
drh3bd48ab2015-09-07 18:23:37 +00003263 fprintf(fp,"%*s shift %-7d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003264 indent,ap->sp->name,ap->x.stp->statenum);
3265 }else{
3266 result = 0;
3267 }
3268 break;
drh75897232000-05-29 14:26:00 +00003269 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003270 if( showPrecedenceConflict ){
drh7e698e92015-09-07 14:22:24 +00003271 fprintf(fp,"%*s reduce %-7d -- dropped by precedence",
drh4ef07702016-03-16 19:45:54 +00003272 indent,ap->sp->name,ap->x.rp->iRule);
drhf5c4e0f2010-07-18 11:35:53 +00003273 }else{
3274 result = 0;
3275 }
3276 break;
drh75897232000-05-29 14:26:00 +00003277 case NOT_USED:
3278 result = 0;
3279 break;
3280 }
drhc173ad82016-05-23 16:15:02 +00003281 if( result && ap->spOpt ){
3282 fprintf(fp," /* because %s==%s */", ap->sp->name, ap->spOpt->name);
3283 }
drh75897232000-05-29 14:26:00 +00003284 return result;
3285}
3286
drh3bd48ab2015-09-07 18:23:37 +00003287/* Generate the "*.out" log file */
icculus9e44cf12010-02-14 17:14:22 +00003288void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003289{
drh539e7412018-04-21 20:24:19 +00003290 int i, n;
drh75897232000-05-29 14:26:00 +00003291 struct state *stp;
3292 struct config *cfp;
3293 struct action *ap;
drh12f68392018-04-06 19:12:55 +00003294 struct rule *rp;
drh75897232000-05-29 14:26:00 +00003295 FILE *fp;
3296
drh2aa6ca42004-09-10 00:14:04 +00003297 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00003298 if( fp==0 ) return;
drh3bd48ab2015-09-07 18:23:37 +00003299 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00003300 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00003301 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00003302 if( lemp->basisflag ) cfp=stp->bp;
3303 else cfp=stp->cfp;
3304 while( cfp ){
3305 char buf[20];
3306 if( cfp->dot==cfp->rp->nrhs ){
drh4ef07702016-03-16 19:45:54 +00003307 lemon_sprintf(buf,"(%d)",cfp->rp->iRule);
drh75897232000-05-29 14:26:00 +00003308 fprintf(fp," %5s ",buf);
3309 }else{
3310 fprintf(fp," ");
3311 }
3312 ConfigPrint(fp,cfp);
3313 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00003314#if 0
drh75897232000-05-29 14:26:00 +00003315 SetPrint(fp,cfp->fws,lemp);
3316 PlinkPrint(fp,cfp->fplp,"To ");
3317 PlinkPrint(fp,cfp->bplp,"From");
3318#endif
3319 if( lemp->basisflag ) cfp=cfp->bp;
3320 else cfp=cfp->next;
3321 }
3322 fprintf(fp,"\n");
3323 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00003324 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
drh75897232000-05-29 14:26:00 +00003325 }
3326 fprintf(fp,"\n");
3327 }
drhe9278182007-07-18 18:16:29 +00003328 fprintf(fp, "----------------------------------------------------\n");
3329 fprintf(fp, "Symbols:\n");
drh539e7412018-04-21 20:24:19 +00003330 fprintf(fp, "The first-set of non-terminals is shown after the name.\n\n");
drhe9278182007-07-18 18:16:29 +00003331 for(i=0; i<lemp->nsymbol; i++){
3332 int j;
3333 struct symbol *sp;
3334
3335 sp = lemp->symbols[i];
3336 fprintf(fp, " %3d: %s", i, sp->name);
3337 if( sp->type==NONTERMINAL ){
3338 fprintf(fp, ":");
3339 if( sp->lambda ){
3340 fprintf(fp, " <lambda>");
3341 }
3342 for(j=0; j<lemp->nterminal; j++){
3343 if( sp->firstset && SetFind(sp->firstset, j) ){
3344 fprintf(fp, " %s", lemp->symbols[j]->name);
3345 }
3346 }
3347 }
drh12f68392018-04-06 19:12:55 +00003348 if( sp->prec>=0 ) fprintf(fp," (precedence=%d)", sp->prec);
drhe9278182007-07-18 18:16:29 +00003349 fprintf(fp, "\n");
3350 }
drh12f68392018-04-06 19:12:55 +00003351 fprintf(fp, "----------------------------------------------------\n");
drh539e7412018-04-21 20:24:19 +00003352 fprintf(fp, "Syntax-only Symbols:\n");
3353 fprintf(fp, "The following symbols never carry semantic content.\n\n");
3354 for(i=n=0; i<lemp->nsymbol; i++){
3355 int w;
3356 struct symbol *sp = lemp->symbols[i];
3357 if( sp->bContent ) continue;
3358 w = (int)strlen(sp->name);
3359 if( n>0 && n+w>75 ){
3360 fprintf(fp,"\n");
3361 n = 0;
3362 }
3363 if( n>0 ){
3364 fprintf(fp, " ");
3365 n++;
3366 }
3367 fprintf(fp, "%s", sp->name);
3368 n += w;
3369 }
3370 if( n>0 ) fprintf(fp, "\n");
3371 fprintf(fp, "----------------------------------------------------\n");
drh12f68392018-04-06 19:12:55 +00003372 fprintf(fp, "Rules:\n");
3373 for(rp=lemp->rule; rp; rp=rp->next){
3374 fprintf(fp, "%4d: ", rp->iRule);
3375 rule_print(fp, rp);
3376 fprintf(fp,".");
3377 if( rp->precsym ){
3378 fprintf(fp," [%s precedence=%d]",
3379 rp->precsym->name, rp->precsym->prec);
3380 }
3381 fprintf(fp,"\n");
3382 }
drh75897232000-05-29 14:26:00 +00003383 fclose(fp);
3384 return;
3385}
3386
3387/* Search for the file "name" which is in the same directory as
3388** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00003389PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00003390{
icculus9e44cf12010-02-14 17:14:22 +00003391 const char *pathlist;
3392 char *pathbufptr;
3393 char *pathbuf;
drh75897232000-05-29 14:26:00 +00003394 char *path,*cp;
3395 char c;
drh75897232000-05-29 14:26:00 +00003396
3397#ifdef __WIN32__
3398 cp = strrchr(argv0,'\\');
3399#else
3400 cp = strrchr(argv0,'/');
3401#endif
3402 if( cp ){
3403 c = *cp;
3404 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003405 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh898799f2014-01-10 23:21:00 +00003406 if( path ) lemon_sprintf(path,"%s/%s",argv0,name);
drh75897232000-05-29 14:26:00 +00003407 *cp = c;
3408 }else{
drh75897232000-05-29 14:26:00 +00003409 pathlist = getenv("PATH");
3410 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003411 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003412 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003413 if( (pathbuf != 0) && (path!=0) ){
3414 pathbufptr = pathbuf;
drh898799f2014-01-10 23:21:00 +00003415 lemon_strcpy(pathbuf, pathlist);
icculus9e44cf12010-02-14 17:14:22 +00003416 while( *pathbuf ){
3417 cp = strchr(pathbuf,':');
3418 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003419 c = *cp;
3420 *cp = 0;
drh898799f2014-01-10 23:21:00 +00003421 lemon_sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003422 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003423 if( c==0 ) pathbuf[0] = 0;
3424 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003425 if( access(path,modemask)==0 ) break;
3426 }
icculus9e44cf12010-02-14 17:14:22 +00003427 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003428 }
3429 }
3430 return path;
3431}
3432
3433/* Given an action, compute the integer value for that action
3434** which is to be put in the action table of the generated machine.
3435** Return negative if no action should be generated.
3436*/
icculus9e44cf12010-02-14 17:14:22 +00003437PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003438{
3439 int act;
3440 switch( ap->type ){
drh3bd48ab2015-09-07 18:23:37 +00003441 case SHIFT: act = ap->x.stp->statenum; break;
drhbd8fcc12017-06-28 11:56:18 +00003442 case SHIFTREDUCE: {
drhbd8fcc12017-06-28 11:56:18 +00003443 /* Since a SHIFT is inherient after a prior REDUCE, convert any
3444 ** SHIFTREDUCE action with a nonterminal on the LHS into a simple
3445 ** REDUCE action: */
drh5c8241b2017-12-24 23:38:10 +00003446 if( ap->sp->index>=lemp->nterminal ){
3447 act = lemp->minReduce + ap->x.rp->iRule;
3448 }else{
3449 act = lemp->minShiftReduce + ap->x.rp->iRule;
3450 }
drhbd8fcc12017-06-28 11:56:18 +00003451 break;
3452 }
drh5c8241b2017-12-24 23:38:10 +00003453 case REDUCE: act = lemp->minReduce + ap->x.rp->iRule; break;
3454 case ERROR: act = lemp->errAction; break;
3455 case ACCEPT: act = lemp->accAction; break;
drh75897232000-05-29 14:26:00 +00003456 default: act = -1; break;
3457 }
3458 return act;
3459}
3460
3461#define LINESIZE 1000
3462/* The next cluster of routines are for reading the template file
3463** and writing the results to the generated parser */
3464/* The first function transfers data from "in" to "out" until
3465** a line is seen which begins with "%%". The line number is
3466** tracked.
3467**
3468** if name!=0, then any word that begin with "Parse" is changed to
3469** begin with *name instead.
3470*/
icculus9e44cf12010-02-14 17:14:22 +00003471PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003472{
3473 int i, iStart;
3474 char line[LINESIZE];
3475 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3476 (*lineno)++;
3477 iStart = 0;
3478 if( name ){
3479 for(i=0; line[i]; i++){
3480 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
drhc56fac72015-10-29 13:48:15 +00003481 && (i==0 || !ISALPHA(line[i-1]))
drh75897232000-05-29 14:26:00 +00003482 ){
3483 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3484 fprintf(out,"%s",name);
3485 i += 4;
3486 iStart = i+1;
3487 }
3488 }
3489 }
3490 fprintf(out,"%s",&line[iStart]);
3491 }
3492}
3493
3494/* The next function finds the template file and opens it, returning
3495** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003496PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003497{
3498 static char templatename[] = "lempar.c";
3499 char buf[1000];
3500 FILE *in;
3501 char *tpltname;
3502 char *cp;
3503
icculus3e143bd2010-02-14 00:48:49 +00003504 /* first, see if user specified a template filename on the command line. */
3505 if (user_templatename != 0) {
3506 if( access(user_templatename,004)==-1 ){
3507 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3508 user_templatename);
3509 lemp->errorcnt++;
3510 return 0;
3511 }
3512 in = fopen(user_templatename,"rb");
3513 if( in==0 ){
drh25473362015-09-04 18:03:45 +00003514 fprintf(stderr,"Can't open the template file \"%s\".\n",
3515 user_templatename);
icculus3e143bd2010-02-14 00:48:49 +00003516 lemp->errorcnt++;
3517 return 0;
3518 }
3519 return in;
3520 }
3521
drh75897232000-05-29 14:26:00 +00003522 cp = strrchr(lemp->filename,'.');
3523 if( cp ){
drh898799f2014-01-10 23:21:00 +00003524 lemon_sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003525 }else{
drh898799f2014-01-10 23:21:00 +00003526 lemon_sprintf(buf,"%s.lt",lemp->filename);
drh75897232000-05-29 14:26:00 +00003527 }
3528 if( access(buf,004)==0 ){
3529 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003530 }else if( access(templatename,004)==0 ){
3531 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003532 }else{
3533 tpltname = pathsearch(lemp->argv0,templatename,0);
3534 }
3535 if( tpltname==0 ){
3536 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3537 templatename);
3538 lemp->errorcnt++;
3539 return 0;
3540 }
drh2aa6ca42004-09-10 00:14:04 +00003541 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003542 if( in==0 ){
3543 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3544 lemp->errorcnt++;
3545 return 0;
3546 }
3547 return in;
3548}
3549
drhaf805ca2004-09-07 11:28:25 +00003550/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003551PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003552{
3553 fprintf(out,"#line %d \"",lineno);
3554 while( *filename ){
3555 if( *filename == '\\' ) putc('\\',out);
3556 putc(*filename,out);
3557 filename++;
3558 }
3559 fprintf(out,"\"\n");
3560}
3561
drh75897232000-05-29 14:26:00 +00003562/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003563PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003564{
3565 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003566 while( *str ){
drh75897232000-05-29 14:26:00 +00003567 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003568 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003569 str++;
3570 }
drh9db55df2004-09-09 14:01:21 +00003571 if( str[-1]!='\n' ){
3572 putc('\n',out);
3573 (*lineno)++;
3574 }
shane58543932008-12-10 20:10:04 +00003575 if (!lemp->nolinenosflag) {
drh06f60d82017-04-14 19:46:12 +00003576 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003577 }
drh75897232000-05-29 14:26:00 +00003578 return;
3579}
3580
3581/*
3582** The following routine emits code for the destructor for the
3583** symbol sp
3584*/
icculus9e44cf12010-02-14 17:14:22 +00003585void emit_destructor_code(
3586 FILE *out,
3587 struct symbol *sp,
3588 struct lemon *lemp,
3589 int *lineno
3590){
drhcc83b6e2004-04-23 23:38:42 +00003591 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003592
drh75897232000-05-29 14:26:00 +00003593 if( sp->type==TERMINAL ){
3594 cp = lemp->tokendest;
3595 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003596 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003597 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003598 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003599 fprintf(out,"{\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00003600 if( !lemp->nolinenosflag ){
3601 (*lineno)++;
3602 tplt_linedir(out,sp->destLineno,lemp->filename);
3603 }
drh960e8c62001-04-03 16:53:21 +00003604 }else if( lemp->vardest ){
3605 cp = lemp->vardest;
3606 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003607 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003608 }else{
3609 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003610 }
3611 for(; *cp; cp++){
3612 if( *cp=='$' && cp[1]=='$' ){
3613 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3614 cp++;
3615 continue;
3616 }
shane58543932008-12-10 20:10:04 +00003617 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003618 fputc(*cp,out);
3619 }
shane58543932008-12-10 20:10:04 +00003620 fprintf(out,"\n"); (*lineno)++;
drh06f60d82017-04-14 19:46:12 +00003621 if (!lemp->nolinenosflag) {
3622 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
shane58543932008-12-10 20:10:04 +00003623 }
3624 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003625 return;
3626}
3627
3628/*
drh960e8c62001-04-03 16:53:21 +00003629** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003630*/
icculus9e44cf12010-02-14 17:14:22 +00003631int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003632{
3633 int ret;
3634 if( sp->type==TERMINAL ){
3635 ret = lemp->tokendest!=0;
3636 }else{
drh960e8c62001-04-03 16:53:21 +00003637 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003638 }
3639 return ret;
3640}
3641
drh0bb132b2004-07-20 14:06:51 +00003642/*
3643** Append text to a dynamically allocated string. If zText is 0 then
3644** reset the string to be empty again. Always return the complete text
3645** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003646**
3647** n bytes of zText are stored. If n==0 then all of zText up to the first
3648** \000 terminator is stored. zText can contain up to two instances of
3649** %d. The values of p1 and p2 are written into the first and second
3650** %d.
3651**
3652** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003653*/
icculus9e44cf12010-02-14 17:14:22 +00003654PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3655 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003656 static char *z = 0;
3657 static int alloced = 0;
3658 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003659 int c;
drh0bb132b2004-07-20 14:06:51 +00003660 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003661 if( zText==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003662 if( used==0 && z!=0 ) z[0] = 0;
drh0bb132b2004-07-20 14:06:51 +00003663 used = 0;
3664 return z;
3665 }
drh7ac25c72004-08-19 15:12:26 +00003666 if( n<=0 ){
3667 if( n<0 ){
3668 used += n;
3669 assert( used>=0 );
3670 }
drh87cf1372008-08-13 20:09:06 +00003671 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003672 }
drhdf609712010-11-23 20:55:27 +00003673 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003674 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003675 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003676 }
icculus9e44cf12010-02-14 17:14:22 +00003677 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003678 while( n-- > 0 ){
3679 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003680 if( c=='%' && n>0 && zText[0]=='d' ){
drh898799f2014-01-10 23:21:00 +00003681 lemon_sprintf(zInt, "%d", p1);
drh0bb132b2004-07-20 14:06:51 +00003682 p1 = p2;
drh898799f2014-01-10 23:21:00 +00003683 lemon_strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003684 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003685 zText++;
3686 n--;
3687 }else{
mistachkin2318d332015-01-12 18:02:52 +00003688 z[used++] = (char)c;
drh0bb132b2004-07-20 14:06:51 +00003689 }
3690 }
3691 z[used] = 0;
3692 return z;
3693}
3694
3695/*
drh711c9812016-05-23 14:24:31 +00003696** Write and transform the rp->code string so that symbols are expanded.
3697** Populate the rp->codePrefix and rp->codeSuffix strings, as appropriate.
drhdabd04c2016-02-17 01:46:19 +00003698**
3699** Return 1 if the expanded code requires that "yylhsminor" local variable
3700** to be defined.
drh0bb132b2004-07-20 14:06:51 +00003701*/
drhdabd04c2016-02-17 01:46:19 +00003702PRIVATE int translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003703 char *cp, *xp;
3704 int i;
drhcf82f0d2016-02-17 04:33:10 +00003705 int rc = 0; /* True if yylhsminor is used */
drh43303de2016-02-17 12:34:03 +00003706 int dontUseRhs0 = 0; /* If true, use of left-most RHS label is illegal */
drhcf82f0d2016-02-17 04:33:10 +00003707 const char *zSkip = 0; /* The zOvwrt comment within rp->code, or NULL */
3708 char lhsused = 0; /* True if the LHS element has been used */
3709 char lhsdirect; /* True if LHS writes directly into stack */
3710 char used[MAXRHS]; /* True for each RHS element which is used */
3711 char zLhs[50]; /* Convert the LHS symbol into this string */
3712 char zOvwrt[900]; /* Comment that to allow LHS to overwrite RHS */
drh0bb132b2004-07-20 14:06:51 +00003713
3714 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3715 lhsused = 0;
3716
drh19c9e562007-03-29 20:13:53 +00003717 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003718 static char newlinestr[2] = { '\n', '\0' };
3719 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003720 rp->line = rp->ruleline;
drh711c9812016-05-23 14:24:31 +00003721 rp->noCode = 1;
3722 }else{
3723 rp->noCode = 0;
drh19c9e562007-03-29 20:13:53 +00003724 }
3725
drh4dd0d3f2016-02-17 01:18:33 +00003726
drh2e55b042016-04-30 17:19:30 +00003727 if( rp->nrhs==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003728 /* If there are no RHS symbols, then writing directly to the LHS is ok */
3729 lhsdirect = 1;
3730 }else if( rp->rhsalias[0]==0 ){
drh2e55b042016-04-30 17:19:30 +00003731 /* The left-most RHS symbol has no value. LHS direct is ok. But
drh4dd0d3f2016-02-17 01:18:33 +00003732 ** we have to call the distructor on the RHS symbol first. */
3733 lhsdirect = 1;
3734 if( has_destructor(rp->rhs[0],lemp) ){
3735 append_str(0,0,0,0);
3736 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
3737 rp->rhs[0]->index,1-rp->nrhs);
3738 rp->codePrefix = Strsafe(append_str(0,0,0,0));
drh711c9812016-05-23 14:24:31 +00003739 rp->noCode = 0;
drh4dd0d3f2016-02-17 01:18:33 +00003740 }
drh2e55b042016-04-30 17:19:30 +00003741 }else if( rp->lhsalias==0 ){
3742 /* There is no LHS value symbol. */
3743 lhsdirect = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003744 }else if( strcmp(rp->lhsalias,rp->rhsalias[0])==0 ){
drh06f60d82017-04-14 19:46:12 +00003745 /* The LHS symbol and the left-most RHS symbol are the same, so
drh4dd0d3f2016-02-17 01:18:33 +00003746 ** direct writing is allowed */
3747 lhsdirect = 1;
3748 lhsused = 1;
3749 used[0] = 1;
3750 if( rp->lhs->dtnum!=rp->rhs[0]->dtnum ){
3751 ErrorMsg(lemp->filename,rp->ruleline,
3752 "%s(%s) and %s(%s) share the same label but have "
3753 "different datatypes.",
3754 rp->lhs->name, rp->lhsalias, rp->rhs[0]->name, rp->rhsalias[0]);
3755 lemp->errorcnt++;
drh06f60d82017-04-14 19:46:12 +00003756 }
drh4dd0d3f2016-02-17 01:18:33 +00003757 }else{
drhcf82f0d2016-02-17 04:33:10 +00003758 lemon_sprintf(zOvwrt, "/*%s-overwrites-%s*/",
3759 rp->lhsalias, rp->rhsalias[0]);
3760 zSkip = strstr(rp->code, zOvwrt);
3761 if( zSkip!=0 ){
3762 /* The code contains a special comment that indicates that it is safe
3763 ** for the LHS label to overwrite left-most RHS label. */
3764 lhsdirect = 1;
3765 }else{
3766 lhsdirect = 0;
3767 }
drh4dd0d3f2016-02-17 01:18:33 +00003768 }
3769 if( lhsdirect ){
3770 sprintf(zLhs, "yymsp[%d].minor.yy%d",1-rp->nrhs,rp->lhs->dtnum);
3771 }else{
drhdabd04c2016-02-17 01:46:19 +00003772 rc = 1;
drh4dd0d3f2016-02-17 01:18:33 +00003773 sprintf(zLhs, "yylhsminor.yy%d",rp->lhs->dtnum);
3774 }
3775
drh0bb132b2004-07-20 14:06:51 +00003776 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003777
3778 /* This const cast is wrong but harmless, if we're careful. */
3779 for(cp=(char *)rp->code; *cp; cp++){
drhcf82f0d2016-02-17 04:33:10 +00003780 if( cp==zSkip ){
3781 append_str(zOvwrt,0,0,0);
3782 cp += lemonStrlen(zOvwrt)-1;
drh43303de2016-02-17 12:34:03 +00003783 dontUseRhs0 = 1;
drhcf82f0d2016-02-17 04:33:10 +00003784 continue;
3785 }
drhc56fac72015-10-29 13:48:15 +00003786 if( ISALPHA(*cp) && (cp==rp->code || (!ISALNUM(cp[-1]) && cp[-1]!='_')) ){
drh0bb132b2004-07-20 14:06:51 +00003787 char saved;
drhc56fac72015-10-29 13:48:15 +00003788 for(xp= &cp[1]; ISALNUM(*xp) || *xp=='_'; xp++);
drh0bb132b2004-07-20 14:06:51 +00003789 saved = *xp;
3790 *xp = 0;
3791 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh4dd0d3f2016-02-17 01:18:33 +00003792 append_str(zLhs,0,0,0);
drh0bb132b2004-07-20 14:06:51 +00003793 cp = xp;
3794 lhsused = 1;
3795 }else{
3796 for(i=0; i<rp->nrhs; i++){
3797 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh43303de2016-02-17 12:34:03 +00003798 if( i==0 && dontUseRhs0 ){
3799 ErrorMsg(lemp->filename,rp->ruleline,
3800 "Label %s used after '%s'.",
3801 rp->rhsalias[0], zOvwrt);
3802 lemp->errorcnt++;
3803 }else if( cp!=rp->code && cp[-1]=='@' ){
drh7ac25c72004-08-19 15:12:26 +00003804 /* If the argument is of the form @X then substituted
3805 ** the token number of X, not the value of X */
3806 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3807 }else{
drhfd405312005-11-06 04:06:59 +00003808 struct symbol *sp = rp->rhs[i];
3809 int dtnum;
3810 if( sp->type==MULTITERMINAL ){
3811 dtnum = sp->subsym[0]->dtnum;
3812 }else{
3813 dtnum = sp->dtnum;
3814 }
3815 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003816 }
drh0bb132b2004-07-20 14:06:51 +00003817 cp = xp;
3818 used[i] = 1;
3819 break;
3820 }
3821 }
3822 }
3823 *xp = saved;
3824 }
3825 append_str(cp, 1, 0, 0);
3826 } /* End loop */
3827
drh4dd0d3f2016-02-17 01:18:33 +00003828 /* Main code generation completed */
3829 cp = append_str(0,0,0,0);
3830 if( cp && cp[0] ) rp->code = Strsafe(cp);
3831 append_str(0,0,0,0);
3832
drh0bb132b2004-07-20 14:06:51 +00003833 /* Check to make sure the LHS has been used */
3834 if( rp->lhsalias && !lhsused ){
3835 ErrorMsg(lemp->filename,rp->ruleline,
3836 "Label \"%s\" for \"%s(%s)\" is never used.",
3837 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3838 lemp->errorcnt++;
3839 }
3840
drh4dd0d3f2016-02-17 01:18:33 +00003841 /* Generate destructor code for RHS minor values which are not referenced.
3842 ** Generate error messages for unused labels and duplicate labels.
3843 */
drh0bb132b2004-07-20 14:06:51 +00003844 for(i=0; i<rp->nrhs; i++){
drh4dd0d3f2016-02-17 01:18:33 +00003845 if( rp->rhsalias[i] ){
3846 if( i>0 ){
3847 int j;
3848 if( rp->lhsalias && strcmp(rp->lhsalias,rp->rhsalias[i])==0 ){
3849 ErrorMsg(lemp->filename,rp->ruleline,
3850 "%s(%s) has the same label as the LHS but is not the left-most "
3851 "symbol on the RHS.",
drhf135cb72019-04-30 14:26:31 +00003852 rp->rhs[i]->name, rp->rhsalias[i]);
drh4dd0d3f2016-02-17 01:18:33 +00003853 lemp->errorcnt++;
3854 }
3855 for(j=0; j<i; j++){
3856 if( rp->rhsalias[j] && strcmp(rp->rhsalias[j],rp->rhsalias[i])==0 ){
3857 ErrorMsg(lemp->filename,rp->ruleline,
3858 "Label %s used for multiple symbols on the RHS of a rule.",
3859 rp->rhsalias[i]);
3860 lemp->errorcnt++;
3861 break;
3862 }
3863 }
drh0bb132b2004-07-20 14:06:51 +00003864 }
drh4dd0d3f2016-02-17 01:18:33 +00003865 if( !used[i] ){
3866 ErrorMsg(lemp->filename,rp->ruleline,
3867 "Label %s for \"%s(%s)\" is never used.",
3868 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3869 lemp->errorcnt++;
3870 }
3871 }else if( i>0 && has_destructor(rp->rhs[i],lemp) ){
3872 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
3873 rp->rhs[i]->index,i-rp->nrhs+1);
drh0bb132b2004-07-20 14:06:51 +00003874 }
3875 }
drh4dd0d3f2016-02-17 01:18:33 +00003876
3877 /* If unable to write LHS values directly into the stack, write the
3878 ** saved LHS value now. */
3879 if( lhsdirect==0 ){
3880 append_str(" yymsp[%d].minor.yy%d = ", 0, 1-rp->nrhs, rp->lhs->dtnum);
3881 append_str(zLhs, 0, 0, 0);
3882 append_str(";\n", 0, 0, 0);
drh61e339a2007-01-16 03:09:02 +00003883 }
drh4dd0d3f2016-02-17 01:18:33 +00003884
3885 /* Suffix code generation complete */
3886 cp = append_str(0,0,0,0);
drh711c9812016-05-23 14:24:31 +00003887 if( cp && cp[0] ){
3888 rp->codeSuffix = Strsafe(cp);
3889 rp->noCode = 0;
3890 }
drhdabd04c2016-02-17 01:46:19 +00003891
3892 return rc;
drh0bb132b2004-07-20 14:06:51 +00003893}
3894
drh06f60d82017-04-14 19:46:12 +00003895/*
drh75897232000-05-29 14:26:00 +00003896** Generate code which executes when the rule "rp" is reduced. Write
3897** the code to "out". Make sure lineno stays up-to-date.
3898*/
icculus9e44cf12010-02-14 17:14:22 +00003899PRIVATE void emit_code(
3900 FILE *out,
3901 struct rule *rp,
3902 struct lemon *lemp,
3903 int *lineno
3904){
3905 const char *cp;
drh75897232000-05-29 14:26:00 +00003906
drh4dd0d3f2016-02-17 01:18:33 +00003907 /* Setup code prior to the #line directive */
3908 if( rp->codePrefix && rp->codePrefix[0] ){
3909 fprintf(out, "{%s", rp->codePrefix);
3910 for(cp=rp->codePrefix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
3911 }
3912
drh75897232000-05-29 14:26:00 +00003913 /* Generate code to do the reduce action */
3914 if( rp->code ){
drh25473362015-09-04 18:03:45 +00003915 if( !lemp->nolinenosflag ){
3916 (*lineno)++;
3917 tplt_linedir(out,rp->line,lemp->filename);
3918 }
drhaf805ca2004-09-07 11:28:25 +00003919 fprintf(out,"{%s",rp->code);
drh4dd0d3f2016-02-17 01:18:33 +00003920 for(cp=rp->code; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
shane58543932008-12-10 20:10:04 +00003921 fprintf(out,"}\n"); (*lineno)++;
drh25473362015-09-04 18:03:45 +00003922 if( !lemp->nolinenosflag ){
3923 (*lineno)++;
3924 tplt_linedir(out,*lineno,lemp->outname);
3925 }
drh4dd0d3f2016-02-17 01:18:33 +00003926 }
3927
3928 /* Generate breakdown code that occurs after the #line directive */
3929 if( rp->codeSuffix && rp->codeSuffix[0] ){
3930 fprintf(out, "%s", rp->codeSuffix);
3931 for(cp=rp->codeSuffix; *cp; cp++){ if( *cp=='\n' ) (*lineno)++; }
3932 }
3933
3934 if( rp->codePrefix ){
3935 fprintf(out, "}\n"); (*lineno)++;
3936 }
drh75897232000-05-29 14:26:00 +00003937
drh75897232000-05-29 14:26:00 +00003938 return;
3939}
3940
3941/*
3942** Print the definition of the union used for the parser's data stack.
3943** This union contains fields for every possible data type for tokens
3944** and nonterminals. In the process of computing and printing this
3945** union, also set the ".dtnum" field of every terminal and nonterminal
3946** symbol.
3947*/
icculus9e44cf12010-02-14 17:14:22 +00003948void print_stack_union(
3949 FILE *out, /* The output stream */
3950 struct lemon *lemp, /* The main info structure for this parser */
3951 int *plineno, /* Pointer to the line number */
3952 int mhflag /* True if generating makeheaders output */
3953){
drh75897232000-05-29 14:26:00 +00003954 int lineno = *plineno; /* The line number of the output */
3955 char **types; /* A hash table of datatypes */
3956 int arraysize; /* Size of the "types" array */
3957 int maxdtlength; /* Maximum length of any ".datatype" field. */
3958 char *stddt; /* Standardized name for a datatype */
3959 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00003960 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00003961 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00003962
3963 /* Allocate and initialize types[] and allocate stddt[] */
3964 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003965 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00003966 if( types==0 ){
3967 fprintf(stderr,"Out of memory.\n");
3968 exit(1);
3969 }
drh75897232000-05-29 14:26:00 +00003970 for(i=0; i<arraysize; i++) types[i] = 0;
3971 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003972 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003973 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003974 }
drh75897232000-05-29 14:26:00 +00003975 for(i=0; i<lemp->nsymbol; i++){
3976 int len;
3977 struct symbol *sp = lemp->symbols[i];
3978 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003979 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003980 if( len>maxdtlength ) maxdtlength = len;
3981 }
3982 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00003983 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00003984 fprintf(stderr,"Out of memory.\n");
3985 exit(1);
3986 }
3987
3988 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3989 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003990 ** used for terminal symbols. If there is no %default_type defined then
3991 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3992 ** a datatype using the %type directive.
3993 */
drh75897232000-05-29 14:26:00 +00003994 for(i=0; i<lemp->nsymbol; i++){
3995 struct symbol *sp = lemp->symbols[i];
3996 char *cp;
3997 if( sp==lemp->errsym ){
3998 sp->dtnum = arraysize+1;
3999 continue;
4000 }
drh960e8c62001-04-03 16:53:21 +00004001 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00004002 sp->dtnum = 0;
4003 continue;
4004 }
4005 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00004006 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00004007 j = 0;
drhc56fac72015-10-29 13:48:15 +00004008 while( ISSPACE(*cp) ) cp++;
drh75897232000-05-29 14:26:00 +00004009 while( *cp ) stddt[j++] = *cp++;
drhc56fac72015-10-29 13:48:15 +00004010 while( j>0 && ISSPACE(stddt[j-1]) ) j--;
drh75897232000-05-29 14:26:00 +00004011 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00004012 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00004013 sp->dtnum = 0;
4014 continue;
4015 }
drh75897232000-05-29 14:26:00 +00004016 hash = 0;
4017 for(j=0; stddt[j]; j++){
4018 hash = hash*53 + stddt[j];
4019 }
drh3b2129c2003-05-13 00:34:21 +00004020 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00004021 while( types[hash] ){
4022 if( strcmp(types[hash],stddt)==0 ){
4023 sp->dtnum = hash + 1;
4024 break;
4025 }
4026 hash++;
drh2b51f212013-10-11 23:01:02 +00004027 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00004028 }
4029 if( types[hash]==0 ){
4030 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00004031 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00004032 if( types[hash]==0 ){
4033 fprintf(stderr,"Out of memory.\n");
4034 exit(1);
4035 }
drh898799f2014-01-10 23:21:00 +00004036 lemon_strcpy(types[hash],stddt);
drh75897232000-05-29 14:26:00 +00004037 }
4038 }
4039
4040 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
4041 name = lemp->name ? lemp->name : "Parse";
4042 lineno = *plineno;
4043 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
4044 fprintf(out,"#define %sTOKENTYPE %s\n",name,
4045 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
4046 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
4047 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00004048 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004049 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
4050 for(i=0; i<arraysize; i++){
4051 if( types[i]==0 ) continue;
4052 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
4053 free(types[i]);
4054 }
drhed0c15b2018-04-16 14:31:34 +00004055 if( lemp->errsym && lemp->errsym->useCnt ){
drhc4dd3fd2008-01-22 01:48:05 +00004056 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
4057 }
drh75897232000-05-29 14:26:00 +00004058 free(stddt);
4059 free(types);
4060 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
4061 *plineno = lineno;
4062}
4063
drhb29b0a52002-02-23 19:39:46 +00004064/*
4065** Return the name of a C datatype able to represent values between
drhc75e0162015-09-07 02:23:02 +00004066** lwr and upr, inclusive. If pnByte!=NULL then also write the sizeof
4067** for that type (1, 2, or 4) into *pnByte.
drhb29b0a52002-02-23 19:39:46 +00004068*/
drhc75e0162015-09-07 02:23:02 +00004069static const char *minimum_size_type(int lwr, int upr, int *pnByte){
4070 const char *zType = "int";
4071 int nByte = 4;
drh8b582012003-10-21 13:16:03 +00004072 if( lwr>=0 ){
4073 if( upr<=255 ){
drhc75e0162015-09-07 02:23:02 +00004074 zType = "unsigned char";
4075 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004076 }else if( upr<65535 ){
drhc75e0162015-09-07 02:23:02 +00004077 zType = "unsigned short int";
4078 nByte = 2;
drh8b582012003-10-21 13:16:03 +00004079 }else{
drhc75e0162015-09-07 02:23:02 +00004080 zType = "unsigned int";
4081 nByte = 4;
drh8b582012003-10-21 13:16:03 +00004082 }
4083 }else if( lwr>=-127 && upr<=127 ){
drhc75e0162015-09-07 02:23:02 +00004084 zType = "signed char";
4085 nByte = 1;
drh8b582012003-10-21 13:16:03 +00004086 }else if( lwr>=-32767 && upr<32767 ){
drhc75e0162015-09-07 02:23:02 +00004087 zType = "short";
4088 nByte = 2;
drhb29b0a52002-02-23 19:39:46 +00004089 }
drhc75e0162015-09-07 02:23:02 +00004090 if( pnByte ) *pnByte = nByte;
4091 return zType;
drhb29b0a52002-02-23 19:39:46 +00004092}
4093
drhfdbf9282003-10-21 16:34:41 +00004094/*
4095** Each state contains a set of token transaction and a set of
4096** nonterminal transactions. Each of these sets makes an instance
4097** of the following structure. An array of these structures is used
4098** to order the creation of entries in the yy_action[] table.
4099*/
4100struct axset {
4101 struct state *stp; /* A pointer to a state */
4102 int isTkn; /* True to use tokens. False for non-terminals */
4103 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00004104 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00004105};
4106
4107/*
4108** Compare to axset structures for sorting purposes
4109*/
4110static int axset_compare(const void *a, const void *b){
4111 struct axset *p1 = (struct axset*)a;
4112 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00004113 int c;
4114 c = p2->nAction - p1->nAction;
4115 if( c==0 ){
drh337cd0d2015-09-07 23:40:42 +00004116 c = p1->iOrder - p2->iOrder;
drhe594bc32009-11-03 13:02:25 +00004117 }
4118 assert( c!=0 || p1==p2 );
4119 return c;
drhfdbf9282003-10-21 16:34:41 +00004120}
4121
drhc4dd3fd2008-01-22 01:48:05 +00004122/*
4123** Write text on "out" that describes the rule "rp".
4124*/
4125static void writeRuleText(FILE *out, struct rule *rp){
4126 int j;
4127 fprintf(out,"%s ::=", rp->lhs->name);
4128 for(j=0; j<rp->nrhs; j++){
4129 struct symbol *sp = rp->rhs[j];
drh61f92cd2014-01-11 03:06:18 +00004130 if( sp->type!=MULTITERMINAL ){
4131 fprintf(out," %s", sp->name);
4132 }else{
drhc4dd3fd2008-01-22 01:48:05 +00004133 int k;
drh61f92cd2014-01-11 03:06:18 +00004134 fprintf(out," %s", sp->subsym[0]->name);
drhc4dd3fd2008-01-22 01:48:05 +00004135 for(k=1; k<sp->nsubsym; k++){
4136 fprintf(out,"|%s",sp->subsym[k]->name);
4137 }
4138 }
4139 }
4140}
4141
4142
drh75897232000-05-29 14:26:00 +00004143/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004144void ReportTable(
4145 struct lemon *lemp,
4146 int mhflag /* Output in makeheaders format if true */
4147){
drh75897232000-05-29 14:26:00 +00004148 FILE *out, *in;
4149 char line[LINESIZE];
4150 int lineno;
4151 struct state *stp;
4152 struct action *ap;
4153 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00004154 struct acttab *pActtab;
drhc75e0162015-09-07 02:23:02 +00004155 int i, j, n, sz;
4156 int szActionType; /* sizeof(YYACTIONTYPE) */
4157 int szCodeType; /* sizeof(YYCODETYPE) */
icculus9e44cf12010-02-14 17:14:22 +00004158 const char *name;
drh8b582012003-10-21 13:16:03 +00004159 int mnTknOfst, mxTknOfst;
4160 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00004161 struct axset *ax;
drh75897232000-05-29 14:26:00 +00004162
drh5c8241b2017-12-24 23:38:10 +00004163 lemp->minShiftReduce = lemp->nstate;
4164 lemp->errAction = lemp->minShiftReduce + lemp->nrule;
4165 lemp->accAction = lemp->errAction + 1;
4166 lemp->noAction = lemp->accAction + 1;
4167 lemp->minReduce = lemp->noAction + 1;
4168 lemp->maxAction = lemp->minReduce + lemp->nrule;
4169
drh75897232000-05-29 14:26:00 +00004170 in = tplt_open(lemp);
4171 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00004172 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00004173 if( out==0 ){
4174 fclose(in);
4175 return;
4176 }
4177 lineno = 1;
4178 tplt_xfer(lemp->name,in,out,&lineno);
4179
4180 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00004181 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00004182 if( mhflag ){
mistachkin8e189222015-04-19 21:43:16 +00004183 char *incName = file_makename(lemp, ".h");
4184 fprintf(out,"#include \"%s\"\n", incName); lineno++;
4185 free(incName);
drh75897232000-05-29 14:26:00 +00004186 }
4187 tplt_xfer(lemp->name,in,out,&lineno);
4188
4189 /* Generate #defines for all tokens */
4190 if( mhflag ){
icculus9e44cf12010-02-14 17:14:22 +00004191 const char *prefix;
drh75897232000-05-29 14:26:00 +00004192 fprintf(out,"#if INTERFACE\n"); lineno++;
4193 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4194 else prefix = "";
4195 for(i=1; i<lemp->nterminal; i++){
4196 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4197 lineno++;
4198 }
4199 fprintf(out,"#endif\n"); lineno++;
4200 }
4201 tplt_xfer(lemp->name,in,out,&lineno);
4202
4203 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00004204 fprintf(out,"#define YYCODETYPE %s\n",
drhed0c15b2018-04-16 14:31:34 +00004205 minimum_size_type(0, lemp->nsymbol, &szCodeType)); lineno++;
4206 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol); lineno++;
drh75897232000-05-29 14:26:00 +00004207 fprintf(out,"#define YYACTIONTYPE %s\n",
drh5c8241b2017-12-24 23:38:10 +00004208 minimum_size_type(0,lemp->maxAction,&szActionType)); lineno++;
drhe09daa92006-06-10 13:29:31 +00004209 if( lemp->wildcard ){
4210 fprintf(out,"#define YYWILDCARD %d\n",
4211 lemp->wildcard->index); lineno++;
4212 }
drh75897232000-05-29 14:26:00 +00004213 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00004214 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004215 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00004216 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
4217 }else{
4218 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
4219 }
drhca44b5a2007-02-22 23:06:58 +00004220 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004221 if( mhflag ){
4222 fprintf(out,"#if INTERFACE\n"); lineno++;
4223 }
4224 name = lemp->name ? lemp->name : "Parse";
4225 if( lemp->arg && lemp->arg[0] ){
drh87cf1372008-08-13 20:09:06 +00004226 i = lemonStrlen(lemp->arg);
drhc56fac72015-10-29 13:48:15 +00004227 while( i>=1 && ISSPACE(lemp->arg[i-1]) ) i--;
4228 while( i>=1 && (ISALNUM(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00004229 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
4230 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
drhfb32c442018-04-21 13:51:42 +00004231 fprintf(out,"#define %sARG_PARAM ,%s\n",name,&lemp->arg[i]); lineno++;
4232 fprintf(out,"#define %sARG_FETCH %s=yypParser->%s;\n",
drh1f245e42002-03-11 13:55:50 +00004233 name,lemp->arg,&lemp->arg[i]); lineno++;
drhfb32c442018-04-21 13:51:42 +00004234 fprintf(out,"#define %sARG_STORE yypParser->%s=%s;\n",
drh1f245e42002-03-11 13:55:50 +00004235 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00004236 }else{
drhfb32c442018-04-21 13:51:42 +00004237 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
4238 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
4239 fprintf(out,"#define %sARG_PARAM\n",name); lineno++;
drh1f245e42002-03-11 13:55:50 +00004240 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
4241 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00004242 }
drhfb32c442018-04-21 13:51:42 +00004243 if( lemp->ctx && lemp->ctx[0] ){
4244 i = lemonStrlen(lemp->ctx);
4245 while( i>=1 && ISSPACE(lemp->ctx[i-1]) ) i--;
4246 while( i>=1 && (ISALNUM(lemp->ctx[i-1]) || lemp->ctx[i-1]=='_') ) i--;
4247 fprintf(out,"#define %sCTX_SDECL %s;\n",name,lemp->ctx); lineno++;
4248 fprintf(out,"#define %sCTX_PDECL ,%s\n",name,lemp->ctx); lineno++;
4249 fprintf(out,"#define %sCTX_PARAM ,%s\n",name,&lemp->ctx[i]); lineno++;
4250 fprintf(out,"#define %sCTX_FETCH %s=yypParser->%s;\n",
4251 name,lemp->ctx,&lemp->ctx[i]); lineno++;
4252 fprintf(out,"#define %sCTX_STORE yypParser->%s=%s;\n",
4253 name,&lemp->ctx[i],&lemp->ctx[i]); lineno++;
4254 }else{
4255 fprintf(out,"#define %sCTX_SDECL\n",name); lineno++;
4256 fprintf(out,"#define %sCTX_PDECL\n",name); lineno++;
4257 fprintf(out,"#define %sCTX_PARAM\n",name); lineno++;
4258 fprintf(out,"#define %sCTX_FETCH\n",name); lineno++;
4259 fprintf(out,"#define %sCTX_STORE\n",name); lineno++;
4260 }
drh75897232000-05-29 14:26:00 +00004261 if( mhflag ){
4262 fprintf(out,"#endif\n"); lineno++;
4263 }
drhed0c15b2018-04-16 14:31:34 +00004264 if( lemp->errsym && lemp->errsym->useCnt ){
drh3bd48ab2015-09-07 18:23:37 +00004265 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
4266 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00004267 }
drh0bd1f4e2002-06-06 18:54:39 +00004268 if( lemp->has_fallback ){
4269 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
4270 }
drh75897232000-05-29 14:26:00 +00004271
drh3bd48ab2015-09-07 18:23:37 +00004272 /* Compute the action table, but do not output it yet. The action
4273 ** table must be computed before generating the YYNSTATE macro because
4274 ** we need to know how many states can be eliminated.
drh75897232000-05-29 14:26:00 +00004275 */
drh3bd48ab2015-09-07 18:23:37 +00004276 ax = (struct axset *) calloc(lemp->nxstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00004277 if( ax==0 ){
4278 fprintf(stderr,"malloc failed\n");
4279 exit(1);
4280 }
drh3bd48ab2015-09-07 18:23:37 +00004281 for(i=0; i<lemp->nxstate; i++){
drh75897232000-05-29 14:26:00 +00004282 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004283 ax[i*2].stp = stp;
4284 ax[i*2].isTkn = 1;
4285 ax[i*2].nAction = stp->nTknAct;
4286 ax[i*2+1].stp = stp;
4287 ax[i*2+1].isTkn = 0;
4288 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00004289 }
drh8b582012003-10-21 13:16:03 +00004290 mxTknOfst = mnTknOfst = 0;
4291 mxNtOfst = mnNtOfst = 0;
drh3bd48ab2015-09-07 18:23:37 +00004292 /* In an effort to minimize the action table size, use the heuristic
4293 ** of placing the largest action sets first */
4294 for(i=0; i<lemp->nxstate*2; i++) ax[i].iOrder = i;
4295 qsort(ax, lemp->nxstate*2, sizeof(ax[0]), axset_compare);
drh3a9d6c72017-12-25 04:15:38 +00004296 pActtab = acttab_alloc(lemp->nsymbol, lemp->nterminal);
drh3bd48ab2015-09-07 18:23:37 +00004297 for(i=0; i<lemp->nxstate*2 && ax[i].nAction>0; i++){
drhfdbf9282003-10-21 16:34:41 +00004298 stp = ax[i].stp;
4299 if( ax[i].isTkn ){
4300 for(ap=stp->ap; ap; ap=ap->next){
4301 int action;
4302 if( ap->sp->index>=lemp->nterminal ) continue;
4303 action = compute_action(lemp, ap);
4304 if( action<0 ) continue;
4305 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004306 }
drh3a9d6c72017-12-25 04:15:38 +00004307 stp->iTknOfst = acttab_insert(pActtab, 1);
drhfdbf9282003-10-21 16:34:41 +00004308 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
4309 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
4310 }else{
4311 for(ap=stp->ap; ap; ap=ap->next){
4312 int action;
4313 if( ap->sp->index<lemp->nterminal ) continue;
4314 if( ap->sp->index==lemp->nsymbol ) continue;
4315 action = compute_action(lemp, ap);
4316 if( action<0 ) continue;
4317 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00004318 }
drh3a9d6c72017-12-25 04:15:38 +00004319 stp->iNtOfst = acttab_insert(pActtab, 0);
drhfdbf9282003-10-21 16:34:41 +00004320 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
4321 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00004322 }
drh337cd0d2015-09-07 23:40:42 +00004323#if 0 /* Uncomment for a trace of how the yy_action[] table fills out */
4324 { int jj, nn;
4325 for(jj=nn=0; jj<pActtab->nAction; jj++){
4326 if( pActtab->aAction[jj].action<0 ) nn++;
4327 }
4328 printf("%4d: State %3d %s n: %2d size: %5d freespace: %d\n",
4329 i, stp->statenum, ax[i].isTkn ? "Token" : "Var ",
4330 ax[i].nAction, pActtab->nAction, nn);
4331 }
4332#endif
drh8b582012003-10-21 13:16:03 +00004333 }
drhfdbf9282003-10-21 16:34:41 +00004334 free(ax);
drh8b582012003-10-21 13:16:03 +00004335
drh756b41e2016-05-24 18:55:08 +00004336 /* Mark rules that are actually used for reduce actions after all
4337 ** optimizations have been applied
4338 */
4339 for(rp=lemp->rule; rp; rp=rp->next) rp->doesReduce = LEMON_FALSE;
4340 for(i=0; i<lemp->nxstate; i++){
drh756b41e2016-05-24 18:55:08 +00004341 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
4342 if( ap->type==REDUCE || ap->type==SHIFTREDUCE ){
drh69bfa552017-04-26 04:32:17 +00004343 ap->x.rp->doesReduce = 1;
drh756b41e2016-05-24 18:55:08 +00004344 }
4345 }
4346 }
4347
drh3bd48ab2015-09-07 18:23:37 +00004348 /* Finish rendering the constants now that the action table has
4349 ** been computed */
4350 fprintf(out,"#define YYNSTATE %d\n",lemp->nxstate); lineno++;
4351 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drh0d9de992017-12-26 18:04:23 +00004352 fprintf(out,"#define YYNTOKEN %d\n",lemp->nterminal); lineno++;
drh337cd0d2015-09-07 23:40:42 +00004353 fprintf(out,"#define YY_MAX_SHIFT %d\n",lemp->nxstate-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004354 i = lemp->minShiftReduce;
4355 fprintf(out,"#define YY_MIN_SHIFTREDUCE %d\n",i); lineno++;
4356 i += lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004357 fprintf(out,"#define YY_MAX_SHIFTREDUCE %d\n", i-1); lineno++;
drh5c8241b2017-12-24 23:38:10 +00004358 fprintf(out,"#define YY_ERROR_ACTION %d\n", lemp->errAction); lineno++;
4359 fprintf(out,"#define YY_ACCEPT_ACTION %d\n", lemp->accAction); lineno++;
4360 fprintf(out,"#define YY_NO_ACTION %d\n", lemp->noAction); lineno++;
4361 fprintf(out,"#define YY_MIN_REDUCE %d\n", lemp->minReduce); lineno++;
4362 i = lemp->minReduce + lemp->nrule;
drh3bd48ab2015-09-07 18:23:37 +00004363 fprintf(out,"#define YY_MAX_REDUCE %d\n", i-1); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004364 tplt_xfer(lemp->name,in,out,&lineno);
4365
4366 /* Now output the action table and its associates:
4367 **
4368 ** yy_action[] A single table containing all actions.
4369 ** yy_lookahead[] A table containing the lookahead for each entry in
4370 ** yy_action. Used to detect hash collisions.
4371 ** yy_shift_ofst[] For each state, the offset into yy_action for
4372 ** shifting terminals.
4373 ** yy_reduce_ofst[] For each state, the offset into yy_action for
4374 ** shifting non-terminals after a reduce.
4375 ** yy_default[] Default action for each state.
4376 */
4377
drh8b582012003-10-21 13:16:03 +00004378 /* Output the yy_action table */
drh3a9d6c72017-12-25 04:15:38 +00004379 lemp->nactiontab = n = acttab_action_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004380 lemp->tablesize += n*szActionType;
drhf16371d2009-11-03 19:18:31 +00004381 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
4382 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004383 for(i=j=0; i<n; i++){
4384 int action = acttab_yyaction(pActtab, i);
drh5c8241b2017-12-24 23:38:10 +00004385 if( action<0 ) action = lemp->noAction;
drhfdbf9282003-10-21 16:34:41 +00004386 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004387 fprintf(out, " %4d,", action);
4388 if( j==9 || i==n-1 ){
4389 fprintf(out, "\n"); lineno++;
4390 j = 0;
4391 }else{
4392 j++;
4393 }
4394 }
4395 fprintf(out, "};\n"); lineno++;
4396
4397 /* Output the yy_lookahead table */
drh3a9d6c72017-12-25 04:15:38 +00004398 lemp->nlookaheadtab = n = acttab_lookahead_size(pActtab);
drhc75e0162015-09-07 02:23:02 +00004399 lemp->tablesize += n*szCodeType;
drh57196282004-10-06 15:41:16 +00004400 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00004401 for(i=j=0; i<n; i++){
4402 int la = acttab_yylookahead(pActtab, i);
4403 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00004404 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004405 fprintf(out, " %4d,", la);
4406 if( j==9 || i==n-1 ){
4407 fprintf(out, "\n"); lineno++;
4408 j = 0;
4409 }else{
4410 j++;
4411 }
4412 }
4413 fprintf(out, "};\n"); lineno++;
4414
4415 /* Output the yy_shift_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004416 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004417 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drh701b6882016-08-10 13:30:43 +00004418 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
4419 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
4420 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004421 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh701b6882016-08-10 13:30:43 +00004422 minimum_size_type(mnTknOfst, lemp->nterminal+lemp->nactiontab, &sz));
4423 lineno++;
drhc75e0162015-09-07 02:23:02 +00004424 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004425 for(i=j=0; i<n; i++){
4426 int ofst;
4427 stp = lemp->sorted[i];
4428 ofst = stp->iTknOfst;
drh701b6882016-08-10 13:30:43 +00004429 if( ofst==NO_OFFSET ) ofst = lemp->nactiontab;
drhfdbf9282003-10-21 16:34:41 +00004430 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004431 fprintf(out, " %4d,", ofst);
4432 if( j==9 || i==n-1 ){
4433 fprintf(out, "\n"); lineno++;
4434 j = 0;
4435 }else{
4436 j++;
4437 }
4438 }
4439 fprintf(out, "};\n"); lineno++;
4440
4441 /* Output the yy_reduce_ofst[] table */
drh3bd48ab2015-09-07 18:23:37 +00004442 n = lemp->nxstate;
drhada354d2005-11-05 15:03:59 +00004443 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00004444 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
4445 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
4446 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh06f60d82017-04-14 19:46:12 +00004447 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drhc75e0162015-09-07 02:23:02 +00004448 minimum_size_type(mnNtOfst-1, mxNtOfst, &sz)); lineno++;
4449 lemp->tablesize += n*sz;
drh8b582012003-10-21 13:16:03 +00004450 for(i=j=0; i<n; i++){
4451 int ofst;
4452 stp = lemp->sorted[i];
4453 ofst = stp->iNtOfst;
4454 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00004455 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00004456 fprintf(out, " %4d,", ofst);
4457 if( j==9 || i==n-1 ){
4458 fprintf(out, "\n"); lineno++;
4459 j = 0;
4460 }else{
4461 j++;
4462 }
4463 }
4464 fprintf(out, "};\n"); lineno++;
4465
4466 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00004467 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh3bd48ab2015-09-07 18:23:37 +00004468 n = lemp->nxstate;
drhc75e0162015-09-07 02:23:02 +00004469 lemp->tablesize += n*szActionType;
drh8b582012003-10-21 13:16:03 +00004470 for(i=j=0; i<n; i++){
4471 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00004472 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh5c8241b2017-12-24 23:38:10 +00004473 if( stp->iDfltReduce<0 ){
4474 fprintf(out, " %4d,", lemp->errAction);
4475 }else{
4476 fprintf(out, " %4d,", stp->iDfltReduce + lemp->minReduce);
4477 }
drh8b582012003-10-21 13:16:03 +00004478 if( j==9 || i==n-1 ){
4479 fprintf(out, "\n"); lineno++;
4480 j = 0;
4481 }else{
4482 j++;
4483 }
4484 }
4485 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004486 tplt_xfer(lemp->name,in,out,&lineno);
4487
drh0bd1f4e2002-06-06 18:54:39 +00004488 /* Generate the table of fallback tokens.
4489 */
4490 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00004491 int mx = lemp->nterminal - 1;
4492 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
drhc75e0162015-09-07 02:23:02 +00004493 lemp->tablesize += (mx+1)*szCodeType;
drh1441f3e2009-06-12 12:50:50 +00004494 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00004495 struct symbol *p = lemp->symbols[i];
4496 if( p->fallback==0 ){
4497 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
4498 }else{
4499 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
4500 p->name, p->fallback->name);
4501 }
4502 lineno++;
4503 }
4504 }
4505 tplt_xfer(lemp->name, in, out, &lineno);
4506
4507 /* Generate a table containing the symbolic name of every symbol
4508 */
drh75897232000-05-29 14:26:00 +00004509 for(i=0; i<lemp->nsymbol; i++){
drh898799f2014-01-10 23:21:00 +00004510 lemon_sprintf(line,"\"%s\",",lemp->symbols[i]->name);
drh3a9d6c72017-12-25 04:15:38 +00004511 fprintf(out," /* %4d */ \"%s\",\n",i, lemp->symbols[i]->name); lineno++;
drh75897232000-05-29 14:26:00 +00004512 }
drh75897232000-05-29 14:26:00 +00004513 tplt_xfer(lemp->name,in,out,&lineno);
4514
drh0bd1f4e2002-06-06 18:54:39 +00004515 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00004516 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00004517 ** when tracing REDUCE actions.
4518 */
4519 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drh4ef07702016-03-16 19:45:54 +00004520 assert( rp->iRule==i );
drhc4dd3fd2008-01-22 01:48:05 +00004521 fprintf(out," /* %3d */ \"", i);
4522 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00004523 fprintf(out,"\",\n"); lineno++;
4524 }
4525 tplt_xfer(lemp->name,in,out,&lineno);
4526
drh75897232000-05-29 14:26:00 +00004527 /* Generate code which executes every time a symbol is popped from
drh06f60d82017-04-14 19:46:12 +00004528 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00004529 ** (In other words, generate the %destructor actions)
4530 */
drh75897232000-05-29 14:26:00 +00004531 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00004532 int once = 1;
drh75897232000-05-29 14:26:00 +00004533 for(i=0; i<lemp->nsymbol; i++){
4534 struct symbol *sp = lemp->symbols[i];
4535 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004536 if( once ){
4537 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
4538 once = 0;
4539 }
drhc53eed12009-06-12 17:46:19 +00004540 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00004541 }
4542 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
4543 if( i<lemp->nsymbol ){
4544 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4545 fprintf(out," break;\n"); lineno++;
4546 }
4547 }
drh8d659732005-01-13 23:54:06 +00004548 if( lemp->vardest ){
4549 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00004550 int once = 1;
drh8d659732005-01-13 23:54:06 +00004551 for(i=0; i<lemp->nsymbol; i++){
4552 struct symbol *sp = lemp->symbols[i];
4553 if( sp==0 || sp->type==TERMINAL ||
4554 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004555 if( once ){
drh5c8241b2017-12-24 23:38:10 +00004556 fprintf(out, " /* Default NON-TERMINAL Destructor */\n");lineno++;
drh4dc8ef52008-07-01 17:13:57 +00004557 once = 0;
4558 }
drhc53eed12009-06-12 17:46:19 +00004559 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00004560 dflt_sp = sp;
4561 }
4562 if( dflt_sp!=0 ){
4563 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00004564 }
drh4dc8ef52008-07-01 17:13:57 +00004565 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00004566 }
drh75897232000-05-29 14:26:00 +00004567 for(i=0; i<lemp->nsymbol; i++){
4568 struct symbol *sp = lemp->symbols[i];
4569 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh0f832dd2016-08-16 16:46:40 +00004570 if( sp->destLineno<0 ) continue; /* Already emitted */
drh75013012009-06-12 15:47:34 +00004571 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004572
4573 /* Combine duplicate destructors into a single case */
4574 for(j=i+1; j<lemp->nsymbol; j++){
4575 struct symbol *sp2 = lemp->symbols[j];
4576 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
4577 && sp2->dtnum==sp->dtnum
4578 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00004579 fprintf(out," case %d: /* %s */\n",
4580 sp2->index, sp2->name); lineno++;
drh0f832dd2016-08-16 16:46:40 +00004581 sp2->destLineno = -1; /* Avoid emitting this destructor again */
drh0bb132b2004-07-20 14:06:51 +00004582 }
4583 }
4584
drh75897232000-05-29 14:26:00 +00004585 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4586 fprintf(out," break;\n"); lineno++;
4587 }
drh75897232000-05-29 14:26:00 +00004588 tplt_xfer(lemp->name,in,out,&lineno);
4589
4590 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00004591 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00004592 tplt_xfer(lemp->name,in,out,&lineno);
4593
drhcfc45b12018-12-03 23:57:27 +00004594 /* Generate the tables of rule information. yyRuleInfoLhs[] and
4595 ** yyRuleInfoNRhs[].
drh75897232000-05-29 14:26:00 +00004596 **
4597 ** Note: This code depends on the fact that rules are number
4598 ** sequentually beginning with 0.
4599 */
drh5c8241b2017-12-24 23:38:10 +00004600 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
drhcfc45b12018-12-03 23:57:27 +00004601 fprintf(out," %4d, /* (%d) ", rp->lhs->index, i);
4602 rule_print(out, rp);
4603 fprintf(out," */\n"); lineno++;
4604 }
4605 tplt_xfer(lemp->name,in,out,&lineno);
4606 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4607 fprintf(out," %3d, /* (%d) ", -rp->nrhs, i);
drh5c8241b2017-12-24 23:38:10 +00004608 rule_print(out, rp);
4609 fprintf(out," */\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004610 }
4611 tplt_xfer(lemp->name,in,out,&lineno);
4612
4613 /* Generate code which execution during each REDUCE action */
drhdabd04c2016-02-17 01:46:19 +00004614 i = 0;
drh75897232000-05-29 14:26:00 +00004615 for(rp=lemp->rule; rp; rp=rp->next){
drhdabd04c2016-02-17 01:46:19 +00004616 i += translate_code(lemp, rp);
4617 }
4618 if( i ){
4619 fprintf(out," YYMINORTYPE yylhsminor;\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004620 }
drhc53eed12009-06-12 17:46:19 +00004621 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00004622 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00004623 struct rule *rp2; /* Other rules with the same action */
drh711c9812016-05-23 14:24:31 +00004624 if( rp->codeEmitted ) continue;
4625 if( rp->noCode ){
4626 /* No C code actions, so this will be part of the "default:" rule */
drh2e55b042016-04-30 17:19:30 +00004627 continue;
4628 }
drh4ef07702016-03-16 19:45:54 +00004629 fprintf(out," case %d: /* ", rp->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004630 writeRuleText(out, rp);
4631 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004632 for(rp2=rp->next; rp2; rp2=rp2->next){
drhafb8cd92016-04-29 11:28:35 +00004633 if( rp2->code==rp->code && rp2->codePrefix==rp->codePrefix
4634 && rp2->codeSuffix==rp->codeSuffix ){
drh4ef07702016-03-16 19:45:54 +00004635 fprintf(out," case %d: /* ", rp2->iRule);
drhc4dd3fd2008-01-22 01:48:05 +00004636 writeRuleText(out, rp2);
drh4ef07702016-03-16 19:45:54 +00004637 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->iRule); lineno++;
drh711c9812016-05-23 14:24:31 +00004638 rp2->codeEmitted = 1;
drh0bb132b2004-07-20 14:06:51 +00004639 }
4640 }
drh75897232000-05-29 14:26:00 +00004641 emit_code(out,rp,lemp,&lineno);
4642 fprintf(out," break;\n"); lineno++;
drh711c9812016-05-23 14:24:31 +00004643 rp->codeEmitted = 1;
drh75897232000-05-29 14:26:00 +00004644 }
drhc53eed12009-06-12 17:46:19 +00004645 /* Finally, output the default: rule. We choose as the default: all
4646 ** empty actions. */
4647 fprintf(out," default:\n"); lineno++;
4648 for(rp=lemp->rule; rp; rp=rp->next){
drh711c9812016-05-23 14:24:31 +00004649 if( rp->codeEmitted ) continue;
4650 assert( rp->noCode );
drh4ef07702016-03-16 19:45:54 +00004651 fprintf(out," /* (%d) ", rp->iRule);
drhc53eed12009-06-12 17:46:19 +00004652 writeRuleText(out, rp);
drh756b41e2016-05-24 18:55:08 +00004653 if( rp->doesReduce ){
4654 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->iRule); lineno++;
4655 }else{
4656 fprintf(out, " (OPTIMIZED OUT) */ assert(yyruleno!=%d);\n",
4657 rp->iRule); lineno++;
4658 }
drhc53eed12009-06-12 17:46:19 +00004659 }
4660 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004661 tplt_xfer(lemp->name,in,out,&lineno);
4662
4663 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004664 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004665 tplt_xfer(lemp->name,in,out,&lineno);
4666
4667 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004668 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004669 tplt_xfer(lemp->name,in,out,&lineno);
4670
4671 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004672 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004673 tplt_xfer(lemp->name,in,out,&lineno);
4674
4675 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004676 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004677
drhe2dcc422019-01-15 14:44:23 +00004678 acttab_free(pActtab);
drh75897232000-05-29 14:26:00 +00004679 fclose(in);
4680 fclose(out);
4681 return;
4682}
4683
4684/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004685void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004686{
4687 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004688 const char *prefix;
drh75897232000-05-29 14:26:00 +00004689 char line[LINESIZE];
4690 char pattern[LINESIZE];
4691 int i;
4692
4693 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4694 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004695 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004696 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004697 int nextChar;
drh75897232000-05-29 14:26:00 +00004698 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
drh61f92cd2014-01-11 03:06:18 +00004699 lemon_sprintf(pattern,"#define %s%-30s %3d\n",
4700 prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004701 if( strcmp(line,pattern) ) break;
4702 }
drh8ba0d1c2012-06-16 15:26:31 +00004703 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004704 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004705 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004706 /* No change in the file. Don't rewrite it. */
4707 return;
4708 }
4709 }
drh2aa6ca42004-09-10 00:14:04 +00004710 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004711 if( out ){
4712 for(i=1; i<lemp->nterminal; i++){
drh61f92cd2014-01-11 03:06:18 +00004713 fprintf(out,"#define %s%-30s %3d\n",prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004714 }
drh06f60d82017-04-14 19:46:12 +00004715 fclose(out);
drh75897232000-05-29 14:26:00 +00004716 }
4717 return;
4718}
4719
4720/* Reduce the size of the action tables, if possible, by making use
4721** of defaults.
4722**
drhb59499c2002-02-23 18:45:13 +00004723** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004724** it the default. Except, there is no default if the wildcard token
4725** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004726*/
icculus9e44cf12010-02-14 17:14:22 +00004727void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004728{
4729 struct state *stp;
drhc173ad82016-05-23 16:15:02 +00004730 struct action *ap, *ap2, *nextap;
drhb59499c2002-02-23 18:45:13 +00004731 struct rule *rp, *rp2, *rbest;
drh0c6dfaa2015-09-08 21:16:46 +00004732 int nbest, n;
drh75897232000-05-29 14:26:00 +00004733 int i;
drhe09daa92006-06-10 13:29:31 +00004734 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004735
4736 for(i=0; i<lemp->nstate; i++){
4737 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004738 nbest = 0;
4739 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004740 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004741
drhb59499c2002-02-23 18:45:13 +00004742 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004743 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4744 usesWildcard = 1;
4745 }
drhb59499c2002-02-23 18:45:13 +00004746 if( ap->type!=REDUCE ) continue;
4747 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004748 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004749 if( rp==rbest ) continue;
4750 n = 1;
4751 for(ap2=ap->next; ap2; ap2=ap2->next){
4752 if( ap2->type!=REDUCE ) continue;
4753 rp2 = ap2->x.rp;
4754 if( rp2==rbest ) continue;
4755 if( rp2==rp ) n++;
4756 }
4757 if( n>nbest ){
4758 nbest = n;
4759 rbest = rp;
drh75897232000-05-29 14:26:00 +00004760 }
4761 }
drh06f60d82017-04-14 19:46:12 +00004762
drhb59499c2002-02-23 18:45:13 +00004763 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004764 ** is not at least 1 or if the wildcard token is a possible
4765 ** lookahead.
4766 */
4767 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004768
drhb59499c2002-02-23 18:45:13 +00004769
4770 /* Combine matching REDUCE actions into a single default */
4771 for(ap=stp->ap; ap; ap=ap->next){
4772 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4773 }
drh75897232000-05-29 14:26:00 +00004774 assert( ap );
4775 ap->sp = Symbol_new("{default}");
4776 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004777 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004778 }
4779 stp->ap = Action_sort(stp->ap);
drh3bd48ab2015-09-07 18:23:37 +00004780
4781 for(ap=stp->ap; ap; ap=ap->next){
4782 if( ap->type==SHIFT ) break;
4783 if( ap->type==REDUCE && ap->x.rp!=rbest ) break;
4784 }
4785 if( ap==0 ){
4786 stp->autoReduce = 1;
4787 stp->pDfltReduce = rbest;
4788 }
4789 }
4790
4791 /* Make a second pass over all states and actions. Convert
4792 ** every action that is a SHIFT to an autoReduce state into
4793 ** a SHIFTREDUCE action.
4794 */
4795 for(i=0; i<lemp->nstate; i++){
4796 stp = lemp->sorted[i];
4797 for(ap=stp->ap; ap; ap=ap->next){
4798 struct state *pNextState;
4799 if( ap->type!=SHIFT ) continue;
4800 pNextState = ap->x.stp;
4801 if( pNextState->autoReduce && pNextState->pDfltReduce!=0 ){
4802 ap->type = SHIFTREDUCE;
4803 ap->x.rp = pNextState->pDfltReduce;
4804 }
4805 }
drh75897232000-05-29 14:26:00 +00004806 }
drhc173ad82016-05-23 16:15:02 +00004807
4808 /* If a SHIFTREDUCE action specifies a rule that has a single RHS term
4809 ** (meaning that the SHIFTREDUCE will land back in the state where it
4810 ** started) and if there is no C-code associated with the reduce action,
4811 ** then we can go ahead and convert the action to be the same as the
4812 ** action for the RHS of the rule.
4813 */
4814 for(i=0; i<lemp->nstate; i++){
4815 stp = lemp->sorted[i];
4816 for(ap=stp->ap; ap; ap=nextap){
4817 nextap = ap->next;
4818 if( ap->type!=SHIFTREDUCE ) continue;
4819 rp = ap->x.rp;
4820 if( rp->noCode==0 ) continue;
4821 if( rp->nrhs!=1 ) continue;
4822#if 1
4823 /* Only apply this optimization to non-terminals. It would be OK to
4824 ** apply it to terminal symbols too, but that makes the parser tables
4825 ** larger. */
4826 if( ap->sp->index<lemp->nterminal ) continue;
4827#endif
4828 /* If we reach this point, it means the optimization can be applied */
4829 nextap = ap;
4830 for(ap2=stp->ap; ap2 && (ap2==ap || ap2->sp!=rp->lhs); ap2=ap2->next){}
4831 assert( ap2!=0 );
4832 ap->spOpt = ap2->sp;
4833 ap->type = ap2->type;
4834 ap->x = ap2->x;
4835 }
4836 }
drh75897232000-05-29 14:26:00 +00004837}
drhb59499c2002-02-23 18:45:13 +00004838
drhada354d2005-11-05 15:03:59 +00004839
4840/*
4841** Compare two states for sorting purposes. The smaller state is the
4842** one with the most non-terminal actions. If they have the same number
4843** of non-terminal actions, then the smaller is the one with the most
4844** token actions.
4845*/
4846static int stateResortCompare(const void *a, const void *b){
4847 const struct state *pA = *(const struct state**)a;
4848 const struct state *pB = *(const struct state**)b;
4849 int n;
4850
4851 n = pB->nNtAct - pA->nNtAct;
4852 if( n==0 ){
4853 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004854 if( n==0 ){
4855 n = pB->statenum - pA->statenum;
4856 }
drhada354d2005-11-05 15:03:59 +00004857 }
drhe594bc32009-11-03 13:02:25 +00004858 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004859 return n;
4860}
4861
4862
4863/*
4864** Renumber and resort states so that states with fewer choices
4865** occur at the end. Except, keep state 0 as the first state.
4866*/
icculus9e44cf12010-02-14 17:14:22 +00004867void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00004868{
4869 int i;
4870 struct state *stp;
4871 struct action *ap;
4872
4873 for(i=0; i<lemp->nstate; i++){
4874 stp = lemp->sorted[i];
4875 stp->nTknAct = stp->nNtAct = 0;
drh5c8241b2017-12-24 23:38:10 +00004876 stp->iDfltReduce = -1; /* Init dflt action to "syntax error" */
drhada354d2005-11-05 15:03:59 +00004877 stp->iTknOfst = NO_OFFSET;
4878 stp->iNtOfst = NO_OFFSET;
4879 for(ap=stp->ap; ap; ap=ap->next){
drh3bd48ab2015-09-07 18:23:37 +00004880 int iAction = compute_action(lemp,ap);
4881 if( iAction>=0 ){
drhada354d2005-11-05 15:03:59 +00004882 if( ap->sp->index<lemp->nterminal ){
4883 stp->nTknAct++;
4884 }else if( ap->sp->index<lemp->nsymbol ){
4885 stp->nNtAct++;
4886 }else{
drh3bd48ab2015-09-07 18:23:37 +00004887 assert( stp->autoReduce==0 || stp->pDfltReduce==ap->x.rp );
drh5c8241b2017-12-24 23:38:10 +00004888 stp->iDfltReduce = iAction;
drhada354d2005-11-05 15:03:59 +00004889 }
4890 }
4891 }
4892 }
4893 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4894 stateResortCompare);
4895 for(i=0; i<lemp->nstate; i++){
4896 lemp->sorted[i]->statenum = i;
4897 }
drh3bd48ab2015-09-07 18:23:37 +00004898 lemp->nxstate = lemp->nstate;
4899 while( lemp->nxstate>1 && lemp->sorted[lemp->nxstate-1]->autoReduce ){
4900 lemp->nxstate--;
4901 }
drhada354d2005-11-05 15:03:59 +00004902}
4903
4904
drh75897232000-05-29 14:26:00 +00004905/***************** From the file "set.c" ************************************/
4906/*
4907** Set manipulation routines for the LEMON parser generator.
4908*/
4909
4910static int size = 0;
4911
4912/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00004913void SetSize(int n)
drh75897232000-05-29 14:26:00 +00004914{
4915 size = n+1;
4916}
4917
4918/* Allocate a new set */
drh14d88552017-04-14 19:44:15 +00004919char *SetNew(void){
drh75897232000-05-29 14:26:00 +00004920 char *s;
drh9892c5d2007-12-21 00:02:11 +00004921 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004922 if( s==0 ){
drh75897232000-05-29 14:26:00 +00004923 memory_error();
4924 }
drh75897232000-05-29 14:26:00 +00004925 return s;
4926}
4927
4928/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00004929void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00004930{
4931 free(s);
4932}
4933
4934/* Add a new element to the set. Return TRUE if the element was added
4935** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00004936int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00004937{
4938 int rv;
drh9892c5d2007-12-21 00:02:11 +00004939 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004940 rv = s[e];
4941 s[e] = 1;
4942 return !rv;
4943}
4944
4945/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00004946int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00004947{
4948 int i, progress;
4949 progress = 0;
4950 for(i=0; i<size; i++){
4951 if( s2[i]==0 ) continue;
4952 if( s1[i]==0 ){
4953 progress = 1;
4954 s1[i] = 1;
4955 }
4956 }
4957 return progress;
4958}
4959/********************** From the file "table.c" ****************************/
4960/*
4961** All code in this file has been automatically generated
4962** from a specification in the file
4963** "table.q"
4964** by the associative array code building program "aagen".
4965** Do not edit this file! Instead, edit the specification
4966** file, then rerun aagen.
4967*/
4968/*
4969** Code for processing tables in the LEMON parser generator.
4970*/
4971
drh01f75f22013-10-02 20:46:30 +00004972PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00004973{
drh01f75f22013-10-02 20:46:30 +00004974 unsigned h = 0;
4975 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00004976 return h;
4977}
4978
4979/* Works like strdup, sort of. Save a string in malloced memory, but
4980** keep strings in a table so that the same string is not in more
4981** than one place.
4982*/
icculus9e44cf12010-02-14 17:14:22 +00004983const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00004984{
icculus9e44cf12010-02-14 17:14:22 +00004985 const char *z;
4986 char *cpy;
drh75897232000-05-29 14:26:00 +00004987
drh916f75f2006-07-17 00:19:39 +00004988 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004989 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00004990 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
drh898799f2014-01-10 23:21:00 +00004991 lemon_strcpy(cpy,y);
icculus9e44cf12010-02-14 17:14:22 +00004992 z = cpy;
drh75897232000-05-29 14:26:00 +00004993 Strsafe_insert(z);
4994 }
4995 MemoryCheck(z);
4996 return z;
4997}
4998
4999/* There is one instance of the following structure for each
5000** associative array of type "x1".
5001*/
5002struct s_x1 {
5003 int size; /* The number of available slots. */
5004 /* Must be a power of 2 greater than or */
5005 /* equal to 1 */
5006 int count; /* Number of currently slots filled */
5007 struct s_x1node *tbl; /* The data stored here */
5008 struct s_x1node **ht; /* Hash table for lookups */
5009};
5010
5011/* There is one instance of this structure for every data element
5012** in an associative array of type "x1".
5013*/
5014typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00005015 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00005016 struct s_x1node *next; /* Next entry with the same hash */
5017 struct s_x1node **from; /* Previous link */
5018} x1node;
5019
5020/* There is only one instance of the array, which is the following */
5021static struct s_x1 *x1a;
5022
5023/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005024void Strsafe_init(void){
drh75897232000-05-29 14:26:00 +00005025 if( x1a ) return;
5026 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
5027 if( x1a ){
5028 x1a->size = 1024;
5029 x1a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005030 x1a->tbl = (x1node*)calloc(1024, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005031 if( x1a->tbl==0 ){
5032 free(x1a);
5033 x1a = 0;
5034 }else{
5035 int i;
5036 x1a->ht = (x1node**)&(x1a->tbl[1024]);
5037 for(i=0; i<1024; i++) x1a->ht[i] = 0;
5038 }
5039 }
5040}
5041/* Insert a new record into the array. Return TRUE if successful.
5042** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005043int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00005044{
5045 x1node *np;
drh01f75f22013-10-02 20:46:30 +00005046 unsigned h;
5047 unsigned ph;
drh75897232000-05-29 14:26:00 +00005048
5049 if( x1a==0 ) return 0;
5050 ph = strhash(data);
5051 h = ph & (x1a->size-1);
5052 np = x1a->ht[h];
5053 while( np ){
5054 if( strcmp(np->data,data)==0 ){
5055 /* An existing entry with the same key is found. */
5056 /* Fail because overwrite is not allows. */
5057 return 0;
5058 }
5059 np = np->next;
5060 }
5061 if( x1a->count>=x1a->size ){
5062 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005063 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005064 struct s_x1 array;
mistachkin8e189222015-04-19 21:43:16 +00005065 array.size = arrSize = x1a->size*2;
drh75897232000-05-29 14:26:00 +00005066 array.count = x1a->count;
mistachkin8e189222015-04-19 21:43:16 +00005067 array.tbl = (x1node*)calloc(arrSize, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00005068 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005069 array.ht = (x1node**)&(array.tbl[arrSize]);
5070 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005071 for(i=0; i<x1a->count; i++){
5072 x1node *oldnp, *newnp;
5073 oldnp = &(x1a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005074 h = strhash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005075 newnp = &(array.tbl[i]);
5076 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5077 newnp->next = array.ht[h];
5078 newnp->data = oldnp->data;
5079 newnp->from = &(array.ht[h]);
5080 array.ht[h] = newnp;
5081 }
5082 free(x1a->tbl);
5083 *x1a = array;
5084 }
5085 /* Insert the new data */
5086 h = ph & (x1a->size-1);
5087 np = &(x1a->tbl[x1a->count++]);
5088 np->data = data;
5089 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
5090 np->next = x1a->ht[h];
5091 x1a->ht[h] = np;
5092 np->from = &(x1a->ht[h]);
5093 return 1;
5094}
5095
5096/* Return a pointer to data assigned to the given key. Return NULL
5097** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005098const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00005099{
drh01f75f22013-10-02 20:46:30 +00005100 unsigned h;
drh75897232000-05-29 14:26:00 +00005101 x1node *np;
5102
5103 if( x1a==0 ) return 0;
5104 h = strhash(key) & (x1a->size-1);
5105 np = x1a->ht[h];
5106 while( np ){
5107 if( strcmp(np->data,key)==0 ) break;
5108 np = np->next;
5109 }
5110 return np ? np->data : 0;
5111}
5112
5113/* Return a pointer to the (terminal or nonterminal) symbol "x".
5114** Create a new symbol if this is the first time "x" has been seen.
5115*/
icculus9e44cf12010-02-14 17:14:22 +00005116struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00005117{
5118 struct symbol *sp;
5119
5120 sp = Symbol_find(x);
5121 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00005122 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00005123 MemoryCheck(sp);
5124 sp->name = Strsafe(x);
drhc56fac72015-10-29 13:48:15 +00005125 sp->type = ISUPPER(*x) ? TERMINAL : NONTERMINAL;
drh75897232000-05-29 14:26:00 +00005126 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00005127 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00005128 sp->prec = -1;
5129 sp->assoc = UNK;
5130 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00005131 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00005132 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00005133 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00005134 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00005135 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00005136 Symbol_insert(sp,sp->name);
5137 }
drhc4dd3fd2008-01-22 01:48:05 +00005138 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00005139 return sp;
5140}
5141
drh61f92cd2014-01-11 03:06:18 +00005142/* Compare two symbols for sorting purposes. Return negative,
5143** zero, or positive if a is less then, equal to, or greater
5144** than b.
drh60d31652004-02-22 00:08:04 +00005145**
5146** Symbols that begin with upper case letters (terminals or tokens)
5147** must sort before symbols that begin with lower case letters
drh61f92cd2014-01-11 03:06:18 +00005148** (non-terminals). And MULTITERMINAL symbols (created using the
5149** %token_class directive) must sort at the very end. Other than
5150** that, the order does not matter.
drh60d31652004-02-22 00:08:04 +00005151**
5152** We find experimentally that leaving the symbols in their original
5153** order (the order they appeared in the grammar file) gives the
5154** smallest parser tables in SQLite.
5155*/
icculus9e44cf12010-02-14 17:14:22 +00005156int Symbolcmpp(const void *_a, const void *_b)
5157{
drh61f92cd2014-01-11 03:06:18 +00005158 const struct symbol *a = *(const struct symbol **) _a;
5159 const struct symbol *b = *(const struct symbol **) _b;
5160 int i1 = a->type==MULTITERMINAL ? 3 : a->name[0]>'Z' ? 2 : 1;
5161 int i2 = b->type==MULTITERMINAL ? 3 : b->name[0]>'Z' ? 2 : 1;
5162 return i1==i2 ? a->index - b->index : i1 - i2;
drh75897232000-05-29 14:26:00 +00005163}
5164
5165/* There is one instance of the following structure for each
5166** associative array of type "x2".
5167*/
5168struct s_x2 {
5169 int size; /* The number of available slots. */
5170 /* Must be a power of 2 greater than or */
5171 /* equal to 1 */
5172 int count; /* Number of currently slots filled */
5173 struct s_x2node *tbl; /* The data stored here */
5174 struct s_x2node **ht; /* Hash table for lookups */
5175};
5176
5177/* There is one instance of this structure for every data element
5178** in an associative array of type "x2".
5179*/
5180typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00005181 struct symbol *data; /* The data */
5182 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00005183 struct s_x2node *next; /* Next entry with the same hash */
5184 struct s_x2node **from; /* Previous link */
5185} x2node;
5186
5187/* There is only one instance of the array, which is the following */
5188static struct s_x2 *x2a;
5189
5190/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005191void Symbol_init(void){
drh75897232000-05-29 14:26:00 +00005192 if( x2a ) return;
5193 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
5194 if( x2a ){
5195 x2a->size = 128;
5196 x2a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005197 x2a->tbl = (x2node*)calloc(128, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005198 if( x2a->tbl==0 ){
5199 free(x2a);
5200 x2a = 0;
5201 }else{
5202 int i;
5203 x2a->ht = (x2node**)&(x2a->tbl[128]);
5204 for(i=0; i<128; i++) x2a->ht[i] = 0;
5205 }
5206 }
5207}
5208/* Insert a new record into the array. Return TRUE if successful.
5209** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005210int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00005211{
5212 x2node *np;
drh01f75f22013-10-02 20:46:30 +00005213 unsigned h;
5214 unsigned ph;
drh75897232000-05-29 14:26:00 +00005215
5216 if( x2a==0 ) return 0;
5217 ph = strhash(key);
5218 h = ph & (x2a->size-1);
5219 np = x2a->ht[h];
5220 while( np ){
5221 if( strcmp(np->key,key)==0 ){
5222 /* An existing entry with the same key is found. */
5223 /* Fail because overwrite is not allows. */
5224 return 0;
5225 }
5226 np = np->next;
5227 }
5228 if( x2a->count>=x2a->size ){
5229 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005230 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005231 struct s_x2 array;
mistachkin8e189222015-04-19 21:43:16 +00005232 array.size = arrSize = x2a->size*2;
drh75897232000-05-29 14:26:00 +00005233 array.count = x2a->count;
mistachkin8e189222015-04-19 21:43:16 +00005234 array.tbl = (x2node*)calloc(arrSize, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00005235 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005236 array.ht = (x2node**)&(array.tbl[arrSize]);
5237 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005238 for(i=0; i<x2a->count; i++){
5239 x2node *oldnp, *newnp;
5240 oldnp = &(x2a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005241 h = strhash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005242 newnp = &(array.tbl[i]);
5243 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5244 newnp->next = array.ht[h];
5245 newnp->key = oldnp->key;
5246 newnp->data = oldnp->data;
5247 newnp->from = &(array.ht[h]);
5248 array.ht[h] = newnp;
5249 }
5250 free(x2a->tbl);
5251 *x2a = array;
5252 }
5253 /* Insert the new data */
5254 h = ph & (x2a->size-1);
5255 np = &(x2a->tbl[x2a->count++]);
5256 np->key = key;
5257 np->data = data;
5258 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
5259 np->next = x2a->ht[h];
5260 x2a->ht[h] = np;
5261 np->from = &(x2a->ht[h]);
5262 return 1;
5263}
5264
5265/* Return a pointer to data assigned to the given key. Return NULL
5266** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005267struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00005268{
drh01f75f22013-10-02 20:46:30 +00005269 unsigned h;
drh75897232000-05-29 14:26:00 +00005270 x2node *np;
5271
5272 if( x2a==0 ) return 0;
5273 h = strhash(key) & (x2a->size-1);
5274 np = x2a->ht[h];
5275 while( np ){
5276 if( strcmp(np->key,key)==0 ) break;
5277 np = np->next;
5278 }
5279 return np ? np->data : 0;
5280}
5281
5282/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00005283struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00005284{
5285 struct symbol *data;
5286 if( x2a && n>0 && n<=x2a->count ){
5287 data = x2a->tbl[n-1].data;
5288 }else{
5289 data = 0;
5290 }
5291 return data;
5292}
5293
5294/* Return the size of the array */
5295int Symbol_count()
5296{
5297 return x2a ? x2a->count : 0;
5298}
5299
5300/* Return an array of pointers to all data in the table.
5301** The array is obtained from malloc. Return NULL if memory allocation
5302** problems, or if the array is empty. */
5303struct symbol **Symbol_arrayof()
5304{
5305 struct symbol **array;
mistachkin8e189222015-04-19 21:43:16 +00005306 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005307 if( x2a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005308 arrSize = x2a->count;
5309 array = (struct symbol **)calloc(arrSize, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00005310 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005311 for(i=0; i<arrSize; i++) array[i] = x2a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005312 }
5313 return array;
5314}
5315
5316/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00005317int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00005318{
icculus9e44cf12010-02-14 17:14:22 +00005319 const struct config *a = (struct config *) _a;
5320 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00005321 int x;
5322 x = a->rp->index - b->rp->index;
5323 if( x==0 ) x = a->dot - b->dot;
5324 return x;
5325}
5326
5327/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00005328PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00005329{
5330 int rc;
5331 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
5332 rc = a->rp->index - b->rp->index;
5333 if( rc==0 ) rc = a->dot - b->dot;
5334 }
5335 if( rc==0 ){
5336 if( a ) rc = 1;
5337 if( b ) rc = -1;
5338 }
5339 return rc;
5340}
5341
5342/* Hash a state */
drh01f75f22013-10-02 20:46:30 +00005343PRIVATE unsigned statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00005344{
drh01f75f22013-10-02 20:46:30 +00005345 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005346 while( a ){
5347 h = h*571 + a->rp->index*37 + a->dot;
5348 a = a->bp;
5349 }
5350 return h;
5351}
5352
5353/* Allocate a new state structure */
5354struct state *State_new()
5355{
icculus9e44cf12010-02-14 17:14:22 +00005356 struct state *newstate;
5357 newstate = (struct state *)calloc(1, sizeof(struct state) );
5358 MemoryCheck(newstate);
5359 return newstate;
drh75897232000-05-29 14:26:00 +00005360}
5361
5362/* There is one instance of the following structure for each
5363** associative array of type "x3".
5364*/
5365struct s_x3 {
5366 int size; /* The number of available slots. */
5367 /* Must be a power of 2 greater than or */
5368 /* equal to 1 */
5369 int count; /* Number of currently slots filled */
5370 struct s_x3node *tbl; /* The data stored here */
5371 struct s_x3node **ht; /* Hash table for lookups */
5372};
5373
5374/* There is one instance of this structure for every data element
5375** in an associative array of type "x3".
5376*/
5377typedef struct s_x3node {
5378 struct state *data; /* The data */
5379 struct config *key; /* The key */
5380 struct s_x3node *next; /* Next entry with the same hash */
5381 struct s_x3node **from; /* Previous link */
5382} x3node;
5383
5384/* There is only one instance of the array, which is the following */
5385static struct s_x3 *x3a;
5386
5387/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005388void State_init(void){
drh75897232000-05-29 14:26:00 +00005389 if( x3a ) return;
5390 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
5391 if( x3a ){
5392 x3a->size = 128;
5393 x3a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005394 x3a->tbl = (x3node*)calloc(128, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005395 if( x3a->tbl==0 ){
5396 free(x3a);
5397 x3a = 0;
5398 }else{
5399 int i;
5400 x3a->ht = (x3node**)&(x3a->tbl[128]);
5401 for(i=0; i<128; i++) x3a->ht[i] = 0;
5402 }
5403 }
5404}
5405/* Insert a new record into the array. Return TRUE if successful.
5406** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005407int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00005408{
5409 x3node *np;
drh01f75f22013-10-02 20:46:30 +00005410 unsigned h;
5411 unsigned ph;
drh75897232000-05-29 14:26:00 +00005412
5413 if( x3a==0 ) return 0;
5414 ph = statehash(key);
5415 h = ph & (x3a->size-1);
5416 np = x3a->ht[h];
5417 while( np ){
5418 if( statecmp(np->key,key)==0 ){
5419 /* An existing entry with the same key is found. */
5420 /* Fail because overwrite is not allows. */
5421 return 0;
5422 }
5423 np = np->next;
5424 }
5425 if( x3a->count>=x3a->size ){
5426 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005427 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005428 struct s_x3 array;
mistachkin8e189222015-04-19 21:43:16 +00005429 array.size = arrSize = x3a->size*2;
drh75897232000-05-29 14:26:00 +00005430 array.count = x3a->count;
mistachkin8e189222015-04-19 21:43:16 +00005431 array.tbl = (x3node*)calloc(arrSize, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00005432 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005433 array.ht = (x3node**)&(array.tbl[arrSize]);
5434 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005435 for(i=0; i<x3a->count; i++){
5436 x3node *oldnp, *newnp;
5437 oldnp = &(x3a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005438 h = statehash(oldnp->key) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005439 newnp = &(array.tbl[i]);
5440 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5441 newnp->next = array.ht[h];
5442 newnp->key = oldnp->key;
5443 newnp->data = oldnp->data;
5444 newnp->from = &(array.ht[h]);
5445 array.ht[h] = newnp;
5446 }
5447 free(x3a->tbl);
5448 *x3a = array;
5449 }
5450 /* Insert the new data */
5451 h = ph & (x3a->size-1);
5452 np = &(x3a->tbl[x3a->count++]);
5453 np->key = key;
5454 np->data = data;
5455 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
5456 np->next = x3a->ht[h];
5457 x3a->ht[h] = np;
5458 np->from = &(x3a->ht[h]);
5459 return 1;
5460}
5461
5462/* Return a pointer to data assigned to the given key. Return NULL
5463** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005464struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005465{
drh01f75f22013-10-02 20:46:30 +00005466 unsigned h;
drh75897232000-05-29 14:26:00 +00005467 x3node *np;
5468
5469 if( x3a==0 ) return 0;
5470 h = statehash(key) & (x3a->size-1);
5471 np = x3a->ht[h];
5472 while( np ){
5473 if( statecmp(np->key,key)==0 ) break;
5474 np = np->next;
5475 }
5476 return np ? np->data : 0;
5477}
5478
5479/* Return an array of pointers to all data in the table.
5480** The array is obtained from malloc. Return NULL if memory allocation
5481** problems, or if the array is empty. */
drh14d88552017-04-14 19:44:15 +00005482struct state **State_arrayof(void)
drh75897232000-05-29 14:26:00 +00005483{
5484 struct state **array;
mistachkin8e189222015-04-19 21:43:16 +00005485 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005486 if( x3a==0 ) return 0;
mistachkin8e189222015-04-19 21:43:16 +00005487 arrSize = x3a->count;
5488 array = (struct state **)calloc(arrSize, sizeof(struct state *));
drh75897232000-05-29 14:26:00 +00005489 if( array ){
mistachkin8e189222015-04-19 21:43:16 +00005490 for(i=0; i<arrSize; i++) array[i] = x3a->tbl[i].data;
drh75897232000-05-29 14:26:00 +00005491 }
5492 return array;
5493}
5494
5495/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00005496PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00005497{
drh01f75f22013-10-02 20:46:30 +00005498 unsigned h=0;
drh75897232000-05-29 14:26:00 +00005499 h = h*571 + a->rp->index*37 + a->dot;
5500 return h;
5501}
5502
5503/* There is one instance of the following structure for each
5504** associative array of type "x4".
5505*/
5506struct s_x4 {
5507 int size; /* The number of available slots. */
5508 /* Must be a power of 2 greater than or */
5509 /* equal to 1 */
5510 int count; /* Number of currently slots filled */
5511 struct s_x4node *tbl; /* The data stored here */
5512 struct s_x4node **ht; /* Hash table for lookups */
5513};
5514
5515/* There is one instance of this structure for every data element
5516** in an associative array of type "x4".
5517*/
5518typedef struct s_x4node {
5519 struct config *data; /* The data */
5520 struct s_x4node *next; /* Next entry with the same hash */
5521 struct s_x4node **from; /* Previous link */
5522} x4node;
5523
5524/* There is only one instance of the array, which is the following */
5525static struct s_x4 *x4a;
5526
5527/* Allocate a new associative array */
drh14d88552017-04-14 19:44:15 +00005528void Configtable_init(void){
drh75897232000-05-29 14:26:00 +00005529 if( x4a ) return;
5530 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
5531 if( x4a ){
5532 x4a->size = 64;
5533 x4a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00005534 x4a->tbl = (x4node*)calloc(64, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005535 if( x4a->tbl==0 ){
5536 free(x4a);
5537 x4a = 0;
5538 }else{
5539 int i;
5540 x4a->ht = (x4node**)&(x4a->tbl[64]);
5541 for(i=0; i<64; i++) x4a->ht[i] = 0;
5542 }
5543 }
5544}
5545/* Insert a new record into the array. Return TRUE if successful.
5546** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00005547int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00005548{
5549 x4node *np;
drh01f75f22013-10-02 20:46:30 +00005550 unsigned h;
5551 unsigned ph;
drh75897232000-05-29 14:26:00 +00005552
5553 if( x4a==0 ) return 0;
5554 ph = confighash(data);
5555 h = ph & (x4a->size-1);
5556 np = x4a->ht[h];
5557 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005558 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00005559 /* An existing entry with the same key is found. */
5560 /* Fail because overwrite is not allows. */
5561 return 0;
5562 }
5563 np = np->next;
5564 }
5565 if( x4a->count>=x4a->size ){
5566 /* Need to make the hash table bigger */
mistachkin8e189222015-04-19 21:43:16 +00005567 int i,arrSize;
drh75897232000-05-29 14:26:00 +00005568 struct s_x4 array;
mistachkin8e189222015-04-19 21:43:16 +00005569 array.size = arrSize = x4a->size*2;
drh75897232000-05-29 14:26:00 +00005570 array.count = x4a->count;
mistachkin8e189222015-04-19 21:43:16 +00005571 array.tbl = (x4node*)calloc(arrSize, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00005572 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
mistachkin8e189222015-04-19 21:43:16 +00005573 array.ht = (x4node**)&(array.tbl[arrSize]);
5574 for(i=0; i<arrSize; i++) array.ht[i] = 0;
drh75897232000-05-29 14:26:00 +00005575 for(i=0; i<x4a->count; i++){
5576 x4node *oldnp, *newnp;
5577 oldnp = &(x4a->tbl[i]);
mistachkin8e189222015-04-19 21:43:16 +00005578 h = confighash(oldnp->data) & (arrSize-1);
drh75897232000-05-29 14:26:00 +00005579 newnp = &(array.tbl[i]);
5580 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5581 newnp->next = array.ht[h];
5582 newnp->data = oldnp->data;
5583 newnp->from = &(array.ht[h]);
5584 array.ht[h] = newnp;
5585 }
5586 free(x4a->tbl);
5587 *x4a = array;
5588 }
5589 /* Insert the new data */
5590 h = ph & (x4a->size-1);
5591 np = &(x4a->tbl[x4a->count++]);
5592 np->data = data;
5593 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
5594 np->next = x4a->ht[h];
5595 x4a->ht[h] = np;
5596 np->from = &(x4a->ht[h]);
5597 return 1;
5598}
5599
5600/* Return a pointer to data assigned to the given key. Return NULL
5601** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005602struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005603{
5604 int h;
5605 x4node *np;
5606
5607 if( x4a==0 ) return 0;
5608 h = confighash(key) & (x4a->size-1);
5609 np = x4a->ht[h];
5610 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005611 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00005612 np = np->next;
5613 }
5614 return np ? np->data : 0;
5615}
5616
5617/* Remove all data from the table. Pass each data to the function "f"
5618** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00005619void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00005620{
5621 int i;
5622 if( x4a==0 || x4a->count==0 ) return;
5623 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
5624 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
5625 x4a->count = 0;
5626 return;
5627}