blob: bde65362bc9693195f7ee60d5eab4b0023ed344f [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
drh75897232000-05-29 14:26:00 +000016#ifndef __WIN32__
17# if defined(_WIN32) || defined(WIN32)
18# define __WIN32__
19# endif
20#endif
21
rse8f304482007-07-30 18:31:53 +000022#ifdef __WIN32__
23extern int access();
24#else
25#include <unistd.h>
26#endif
27
drh75897232000-05-29 14:26:00 +000028/* #define PRIVATE static */
29#define PRIVATE
30
31#ifdef TEST
32#define MAXRHS 5 /* Set low to exercise exception code */
33#else
34#define MAXRHS 1000
35#endif
36
drhe9278182007-07-18 18:16:29 +000037static char *msort(char*,char**,int(*)(const char*,const char*));
drh75897232000-05-29 14:26:00 +000038
drh87cf1372008-08-13 20:09:06 +000039/*
40** Compilers are getting increasingly pedantic about type conversions
41** as C evolves ever closer to Ada.... To work around the latest problems
42** we have to define the following variant of strlen().
43*/
44#define lemonStrlen(X) ((int)strlen(X))
45
drhe9278182007-07-18 18:16:29 +000046static struct action *Action_new(void);
47static struct action *Action_sort(struct action *);
drh75897232000-05-29 14:26:00 +000048
49/********** From the file "build.h" ************************************/
50void FindRulePrecedences();
51void FindFirstSets();
52void FindStates();
53void FindLinks();
54void FindFollowSets();
55void FindActions();
56
57/********* From the file "configlist.h" *********************************/
58void Configlist_init(/* void */);
59struct config *Configlist_add(/* struct rule *, int */);
60struct config *Configlist_addbasis(/* struct rule *, int */);
61void Configlist_closure(/* void */);
62void Configlist_sort(/* void */);
63void Configlist_sortbasis(/* void */);
64struct config *Configlist_return(/* void */);
65struct config *Configlist_basis(/* void */);
66void Configlist_eat(/* struct config * */);
67void Configlist_reset(/* void */);
68
69/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +000070void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +000071
72/****** From the file "option.h" ******************************************/
73struct s_options {
74 enum { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
75 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR} type;
76 char *label;
77 char *arg;
78 char *message;
79};
drhb0c86772000-06-02 23:21:26 +000080int OptInit(/* char**,struct s_options*,FILE* */);
81int OptNArgs(/* void */);
82char *OptArg(/* int */);
83void OptErr(/* int */);
84void OptPrint(/* void */);
drh75897232000-05-29 14:26:00 +000085
86/******** From the file "parse.h" *****************************************/
87void Parse(/* struct lemon *lemp */);
88
89/********* From the file "plink.h" ***************************************/
90struct plink *Plink_new(/* void */);
91void Plink_add(/* struct plink **, struct config * */);
92void Plink_copy(/* struct plink **, struct plink * */);
93void Plink_delete(/* struct plink * */);
94
95/********** From the file "report.h" *************************************/
96void Reprint(/* struct lemon * */);
97void ReportOutput(/* struct lemon * */);
98void ReportTable(/* struct lemon * */);
99void ReportHeader(/* struct lemon * */);
100void CompressTables(/* struct lemon * */);
drhada354d2005-11-05 15:03:59 +0000101void ResortStates(/* struct lemon * */);
drh75897232000-05-29 14:26:00 +0000102
103/********** From the file "set.h" ****************************************/
104void SetSize(/* int N */); /* All sets will be of size N */
105char *SetNew(/* void */); /* A new set for element 0..N */
106void SetFree(/* char* */); /* Deallocate a set */
107
108int SetAdd(/* char*,int */); /* Add element to a set */
109int SetUnion(/* char *A,char *B */); /* A <- A U B, thru element N */
110
111#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
112
113/********** From the file "struct.h" *************************************/
114/*
115** Principal data structures for the LEMON parser generator.
116*/
117
drhaa9f1122007-08-23 02:50:56 +0000118typedef enum {LEMON_FALSE=0, LEMON_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000119
120/* Symbols (terminals and nonterminals) of the grammar are stored
121** in the following: */
122struct symbol {
123 char *name; /* Name of the symbol */
124 int index; /* Index number for this symbol */
125 enum {
126 TERMINAL,
drhfd405312005-11-06 04:06:59 +0000127 NONTERMINAL,
128 MULTITERMINAL
drh75897232000-05-29 14:26:00 +0000129 } type; /* Symbols are all either TERMINALS or NTs */
130 struct rule *rule; /* Linked list of rules of this (if an NT) */
drh0bd1f4e2002-06-06 18:54:39 +0000131 struct symbol *fallback; /* fallback token in case this token doesn't parse */
drh75897232000-05-29 14:26:00 +0000132 int prec; /* Precedence if defined (-1 otherwise) */
133 enum e_assoc {
134 LEFT,
135 RIGHT,
136 NONE,
137 UNK
drh34ff57b2008-07-14 12:27:51 +0000138 } assoc; /* Associativity if precedence is defined */
drh75897232000-05-29 14:26:00 +0000139 char *firstset; /* First-set for all rules of this symbol */
140 Boolean lambda; /* True if NT and can generate an empty string */
drhc4dd3fd2008-01-22 01:48:05 +0000141 int useCnt; /* Number of times used */
drh75897232000-05-29 14:26:00 +0000142 char *destructor; /* Code which executes whenever this symbol is
143 ** popped from the stack during error processing */
drh4dc8ef52008-07-01 17:13:57 +0000144 int destLineno; /* Line number for start of destructor */
drh75897232000-05-29 14:26:00 +0000145 char *datatype; /* The data type of information held by this
146 ** object. Only used if type==NONTERMINAL */
147 int dtnum; /* The data type number. In the parser, the value
148 ** stack is a union. The .yy%d element of this
149 ** union is the correct data type for this object */
drhfd405312005-11-06 04:06:59 +0000150 /* The following fields are used by MULTITERMINALs only */
151 int nsubsym; /* Number of constituent symbols in the MULTI */
152 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000153};
154
155/* Each production rule in the grammar is stored in the following
156** structure. */
157struct rule {
158 struct symbol *lhs; /* Left-hand side of the rule */
159 char *lhsalias; /* Alias for the LHS (NULL if none) */
drhb4960992007-10-05 16:16:36 +0000160 int lhsStart; /* True if left-hand side is the start symbol */
drh75897232000-05-29 14:26:00 +0000161 int ruleline; /* Line number for the rule */
162 int nrhs; /* Number of RHS symbols */
163 struct symbol **rhs; /* The RHS symbols */
164 char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
165 int line; /* Line number at which code begins */
166 char *code; /* The code executed when this rule is reduced */
167 struct symbol *precsym; /* Precedence symbol for this rule */
168 int index; /* An index number for this rule */
169 Boolean canReduce; /* True if this rule is ever reduced */
170 struct rule *nextlhs; /* Next rule with the same LHS */
171 struct rule *next; /* Next rule in the global list */
172};
173
174/* A configuration is a production rule of the grammar together with
175** a mark (dot) showing how much of that rule has been processed so far.
176** Configurations also contain a follow-set which is a list of terminal
177** symbols which are allowed to immediately follow the end of the rule.
178** Every configuration is recorded as an instance of the following: */
179struct config {
180 struct rule *rp; /* The rule upon which the configuration is based */
181 int dot; /* The parse point */
182 char *fws; /* Follow-set for this configuration only */
183 struct plink *fplp; /* Follow-set forward propagation links */
184 struct plink *bplp; /* Follow-set backwards propagation links */
185 struct state *stp; /* Pointer to state which contains this */
186 enum {
187 COMPLETE, /* The status is used during followset and */
188 INCOMPLETE /* shift computations */
189 } status;
190 struct config *next; /* Next configuration in the state */
191 struct config *bp; /* The next basis configuration */
192};
193
194/* Every shift or reduce operation is stored as one of the following */
195struct action {
196 struct symbol *sp; /* The look-ahead symbol */
197 enum e_action {
198 SHIFT,
199 ACCEPT,
200 REDUCE,
201 ERROR,
drh9892c5d2007-12-21 00:02:11 +0000202 SSCONFLICT, /* A shift/shift conflict */
203 SRCONFLICT, /* Was a reduce, but part of a conflict */
204 RRCONFLICT, /* Was a reduce, but part of a conflict */
drh75897232000-05-29 14:26:00 +0000205 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
206 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
207 NOT_USED /* Deleted by compression */
208 } type;
209 union {
210 struct state *stp; /* The new state, if a shift */
211 struct rule *rp; /* The rule, if a reduce */
212 } x;
213 struct action *next; /* Next action for this state */
214 struct action *collide; /* Next action with the same hash */
215};
216
217/* Each state of the generated parser's finite state machine
218** is encoded as an instance of the following structure. */
219struct state {
220 struct config *bp; /* The basis configurations for this state */
221 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000222 int statenum; /* Sequential number for this state */
drh75897232000-05-29 14:26:00 +0000223 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000224 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
225 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
226 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000227};
drh8b582012003-10-21 13:16:03 +0000228#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000229
230/* A followset propagation link indicates that the contents of one
231** configuration followset should be propagated to another whenever
232** the first changes. */
233struct plink {
234 struct config *cfp; /* The configuration to which linked */
235 struct plink *next; /* The next propagate link */
236};
237
238/* The state vector for the entire parser generator is recorded as
239** follows. (LEMON uses no global variables and makes little use of
240** static variables. Fields in the following structure can be thought
241** of as begin global variables in the program.) */
242struct lemon {
243 struct state **sorted; /* Table of states sorted by state number */
244 struct rule *rule; /* List of all rules */
245 int nstate; /* Number of states */
246 int nrule; /* Number of rules */
247 int nsymbol; /* Number of terminal and nonterminal symbols */
248 int nterminal; /* Number of terminal symbols */
249 struct symbol **symbols; /* Sorted array of pointers to symbols */
250 int errorcnt; /* Number of errors */
251 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000252 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000253 char *name; /* Name of the generated parser */
254 char *arg; /* Declaration of the 3th argument to parser */
255 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000256 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000257 char *start; /* Name of the start symbol for the grammar */
258 char *stacksize; /* Size of the parser stack */
259 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000260 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000261 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000262 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000263 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000264 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000265 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000266 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000267 char *filename; /* Name of the input file */
268 char *outname; /* Name of the current output file */
269 char *tokenprefix; /* A prefix added to token names in the .h file */
270 int nconflict; /* Number of parsing conflicts */
271 int tablesize; /* Size of the parse tables */
272 int basisflag; /* Print only basis configurations */
drh34ff57b2008-07-14 12:27:51 +0000273 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000274 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000275 char *argv0; /* Name of the program */
276};
277
278#define MemoryCheck(X) if((X)==0){ \
279 extern void memory_error(); \
280 memory_error(); \
281}
282
283/**************** From the file "table.h" *********************************/
284/*
285** All code in this file has been automatically generated
286** from a specification in the file
287** "table.q"
288** by the associative array code building program "aagen".
289** Do not edit this file! Instead, edit the specification
290** file, then rerun aagen.
291*/
292/*
293** Code for processing tables in the LEMON parser generator.
294*/
295
296/* Routines for handling a strings */
297
298char *Strsafe();
299
300void Strsafe_init(/* void */);
301int Strsafe_insert(/* char * */);
302char *Strsafe_find(/* char * */);
303
304/* Routines for handling symbols of the grammar */
305
306struct symbol *Symbol_new();
307int Symbolcmpp(/* struct symbol **, struct symbol ** */);
308void Symbol_init(/* void */);
309int Symbol_insert(/* struct symbol *, char * */);
310struct symbol *Symbol_find(/* char * */);
311struct symbol *Symbol_Nth(/* int */);
312int Symbol_count(/* */);
313struct symbol **Symbol_arrayof(/* */);
314
315/* Routines to manage the state table */
316
317int Configcmp(/* struct config *, struct config * */);
318struct state *State_new();
319void State_init(/* void */);
320int State_insert(/* struct state *, struct config * */);
321struct state *State_find(/* struct config * */);
322struct state **State_arrayof(/* */);
323
324/* Routines used for efficiency in Configlist_add */
325
326void Configtable_init(/* void */);
327int Configtable_insert(/* struct config * */);
328struct config *Configtable_find(/* struct config * */);
329void Configtable_clear(/* int(*)(struct config *) */);
330/****************** From the file "action.c" *******************************/
331/*
332** Routines processing parser actions in the LEMON parser generator.
333*/
334
335/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000336static struct action *Action_new(void){
drh75897232000-05-29 14:26:00 +0000337 static struct action *freelist = 0;
338 struct action *new;
339
340 if( freelist==0 ){
341 int i;
342 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +0000343 freelist = (struct action *)calloc(amt, sizeof(struct action));
drh75897232000-05-29 14:26:00 +0000344 if( freelist==0 ){
345 fprintf(stderr,"Unable to allocate memory for a new parser action.");
346 exit(1);
347 }
348 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
349 freelist[amt-1].next = 0;
350 }
351 new = freelist;
352 freelist = freelist->next;
353 return new;
354}
355
drhe9278182007-07-18 18:16:29 +0000356/* Compare two actions for sorting purposes. Return negative, zero, or
357** positive if the first action is less than, equal to, or greater than
358** the first
359*/
360static int actioncmp(
361 struct action *ap1,
362 struct action *ap2
363){
drh75897232000-05-29 14:26:00 +0000364 int rc;
365 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000366 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000367 rc = (int)ap1->type - (int)ap2->type;
368 }
369 if( rc==0 && ap1->type==REDUCE ){
drh75897232000-05-29 14:26:00 +0000370 rc = ap1->x.rp->index - ap2->x.rp->index;
371 }
drhe594bc32009-11-03 13:02:25 +0000372 if( rc==0 ){
373 rc = ap2 - ap1;
374 }
drh75897232000-05-29 14:26:00 +0000375 return rc;
376}
377
378/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000379static struct action *Action_sort(
380 struct action *ap
381){
382 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
383 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000384 return ap;
385}
386
387void Action_add(app,type,sp,arg)
388struct action **app;
389enum e_action type;
390struct symbol *sp;
391char *arg;
392{
393 struct action *new;
394 new = Action_new();
395 new->next = *app;
396 *app = new;
397 new->type = type;
398 new->sp = sp;
399 if( type==SHIFT ){
400 new->x.stp = (struct state *)arg;
401 }else{
402 new->x.rp = (struct rule *)arg;
403 }
404}
drh8b582012003-10-21 13:16:03 +0000405/********************** New code to implement the "acttab" module ***********/
406/*
407** This module implements routines use to construct the yy_action[] table.
408*/
409
410/*
411** The state of the yy_action table under construction is an instance of
412** the following structure
413*/
414typedef struct acttab acttab;
415struct acttab {
416 int nAction; /* Number of used slots in aAction[] */
417 int nActionAlloc; /* Slots allocated for aAction[] */
418 struct {
419 int lookahead; /* Value of the lookahead token */
420 int action; /* Action to take on the given lookahead */
421 } *aAction, /* The yy_action[] table under construction */
422 *aLookahead; /* A single new transaction set */
423 int mnLookahead; /* Minimum aLookahead[].lookahead */
424 int mnAction; /* Action associated with mnLookahead */
425 int mxLookahead; /* Maximum aLookahead[].lookahead */
426 int nLookahead; /* Used slots in aLookahead[] */
427 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
428};
429
430/* Return the number of entries in the yy_action table */
431#define acttab_size(X) ((X)->nAction)
432
433/* The value for the N-th entry in yy_action */
434#define acttab_yyaction(X,N) ((X)->aAction[N].action)
435
436/* The value for the N-th entry in yy_lookahead */
437#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
438
439/* Free all memory associated with the given acttab */
440void acttab_free(acttab *p){
441 free( p->aAction );
442 free( p->aLookahead );
443 free( p );
444}
445
446/* Allocate a new acttab structure */
447acttab *acttab_alloc(void){
drh9892c5d2007-12-21 00:02:11 +0000448 acttab *p = calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000449 if( p==0 ){
450 fprintf(stderr,"Unable to allocate memory for a new acttab.");
451 exit(1);
452 }
453 memset(p, 0, sizeof(*p));
454 return p;
455}
456
457/* Add a new action to the current transaction set
458*/
459void acttab_action(acttab *p, int lookahead, int action){
460 if( p->nLookahead>=p->nLookaheadAlloc ){
461 p->nLookaheadAlloc += 25;
462 p->aLookahead = realloc( p->aLookahead,
463 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
464 if( p->aLookahead==0 ){
465 fprintf(stderr,"malloc failed\n");
466 exit(1);
467 }
468 }
469 if( p->nLookahead==0 ){
470 p->mxLookahead = lookahead;
471 p->mnLookahead = lookahead;
472 p->mnAction = action;
473 }else{
474 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
475 if( p->mnLookahead>lookahead ){
476 p->mnLookahead = lookahead;
477 p->mnAction = action;
478 }
479 }
480 p->aLookahead[p->nLookahead].lookahead = lookahead;
481 p->aLookahead[p->nLookahead].action = action;
482 p->nLookahead++;
483}
484
485/*
486** Add the transaction set built up with prior calls to acttab_action()
487** into the current action table. Then reset the transaction set back
488** to an empty set in preparation for a new round of acttab_action() calls.
489**
490** Return the offset into the action table of the new transaction.
491*/
492int acttab_insert(acttab *p){
493 int i, j, k, n;
494 assert( p->nLookahead>0 );
495
496 /* Make sure we have enough space to hold the expanded action table
497 ** in the worst case. The worst case occurs if the transaction set
498 ** must be appended to the current action table
499 */
drh784d86f2004-02-19 18:41:53 +0000500 n = p->mxLookahead + 1;
drh8b582012003-10-21 13:16:03 +0000501 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000502 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000503 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
504 p->aAction = realloc( p->aAction,
505 sizeof(p->aAction[0])*p->nActionAlloc);
506 if( p->aAction==0 ){
507 fprintf(stderr,"malloc failed\n");
508 exit(1);
509 }
drhfdbf9282003-10-21 16:34:41 +0000510 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000511 p->aAction[i].lookahead = -1;
512 p->aAction[i].action = -1;
513 }
514 }
515
516 /* Scan the existing action table looking for an offset where we can
517 ** insert the current transaction set. Fall out of the loop when that
518 ** offset is found. In the worst case, we fall out of the loop when
519 ** i reaches p->nAction, which means we append the new transaction set.
520 **
521 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
522 */
drh784d86f2004-02-19 18:41:53 +0000523 for(i=0; i<p->nAction+p->mnLookahead; i++){
drh8b582012003-10-21 13:16:03 +0000524 if( p->aAction[i].lookahead<0 ){
525 for(j=0; j<p->nLookahead; j++){
526 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
527 if( k<0 ) break;
528 if( p->aAction[k].lookahead>=0 ) break;
529 }
drhfdbf9282003-10-21 16:34:41 +0000530 if( j<p->nLookahead ) continue;
531 for(j=0; j<p->nAction; j++){
532 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
533 }
534 if( j==p->nAction ){
535 break; /* Fits in empty slots */
536 }
drh8b582012003-10-21 13:16:03 +0000537 }else if( p->aAction[i].lookahead==p->mnLookahead ){
538 if( p->aAction[i].action!=p->mnAction ) continue;
539 for(j=0; j<p->nLookahead; j++){
540 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
541 if( k<0 || k>=p->nAction ) break;
542 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
543 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
544 }
545 if( j<p->nLookahead ) continue;
546 n = 0;
547 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000548 if( p->aAction[j].lookahead<0 ) continue;
549 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000550 }
drhfdbf9282003-10-21 16:34:41 +0000551 if( n==p->nLookahead ){
552 break; /* Same as a prior transaction set */
553 }
drh8b582012003-10-21 13:16:03 +0000554 }
555 }
556 /* Insert transaction set at index i. */
557 for(j=0; j<p->nLookahead; j++){
558 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
559 p->aAction[k] = p->aLookahead[j];
560 if( k>=p->nAction ) p->nAction = k+1;
561 }
562 p->nLookahead = 0;
563
564 /* Return the offset that is added to the lookahead in order to get the
565 ** index into yy_action of the action */
566 return i - p->mnLookahead;
567}
568
drh75897232000-05-29 14:26:00 +0000569/********************** From the file "build.c" *****************************/
570/*
571** Routines to construction the finite state machine for the LEMON
572** parser generator.
573*/
574
575/* Find a precedence symbol of every rule in the grammar.
576**
577** Those rules which have a precedence symbol coded in the input
578** grammar using the "[symbol]" construct will already have the
579** rp->precsym field filled. Other rules take as their precedence
580** symbol the first RHS symbol with a defined precedence. If there
581** are not RHS symbols with a defined precedence, the precedence
582** symbol field is left blank.
583*/
584void FindRulePrecedences(xp)
585struct lemon *xp;
586{
587 struct rule *rp;
588 for(rp=xp->rule; rp; rp=rp->next){
589 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000590 int i, j;
591 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
592 struct symbol *sp = rp->rhs[i];
593 if( sp->type==MULTITERMINAL ){
594 for(j=0; j<sp->nsubsym; j++){
595 if( sp->subsym[j]->prec>=0 ){
596 rp->precsym = sp->subsym[j];
597 break;
598 }
599 }
600 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000601 rp->precsym = rp->rhs[i];
drh75897232000-05-29 14:26:00 +0000602 }
603 }
604 }
605 }
606 return;
607}
608
609/* Find all nonterminals which will generate the empty string.
610** Then go back and compute the first sets of every nonterminal.
611** The first set is the set of all terminal symbols which can begin
612** a string generated by that nonterminal.
613*/
614void FindFirstSets(lemp)
615struct lemon *lemp;
616{
drhfd405312005-11-06 04:06:59 +0000617 int i, j;
drh75897232000-05-29 14:26:00 +0000618 struct rule *rp;
619 int progress;
620
621 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000622 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000623 }
624 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
625 lemp->symbols[i]->firstset = SetNew();
626 }
627
628 /* First compute all lambdas */
629 do{
630 progress = 0;
631 for(rp=lemp->rule; rp; rp=rp->next){
632 if( rp->lhs->lambda ) continue;
633 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000634 struct symbol *sp = rp->rhs[i];
drhaa9f1122007-08-23 02:50:56 +0000635 if( sp->type!=TERMINAL || sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000636 }
637 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000638 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000639 progress = 1;
640 }
641 }
642 }while( progress );
643
644 /* Now compute all first sets */
645 do{
646 struct symbol *s1, *s2;
647 progress = 0;
648 for(rp=lemp->rule; rp; rp=rp->next){
649 s1 = rp->lhs;
650 for(i=0; i<rp->nrhs; i++){
651 s2 = rp->rhs[i];
652 if( s2->type==TERMINAL ){
653 progress += SetAdd(s1->firstset,s2->index);
654 break;
drhfd405312005-11-06 04:06:59 +0000655 }else if( s2->type==MULTITERMINAL ){
656 for(j=0; j<s2->nsubsym; j++){
657 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
658 }
659 break;
drh75897232000-05-29 14:26:00 +0000660 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000661 if( s1->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000662 }else{
663 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000664 if( s2->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000665 }
666 }
667 }
668 }while( progress );
669 return;
670}
671
672/* Compute all LR(0) states for the grammar. Links
673** are added to between some states so that the LR(1) follow sets
674** can be computed later.
675*/
676PRIVATE struct state *getstate(/* struct lemon * */); /* forward reference */
677void FindStates(lemp)
678struct lemon *lemp;
679{
680 struct symbol *sp;
681 struct rule *rp;
682
683 Configlist_init();
684
685 /* Find the start symbol */
686 if( lemp->start ){
687 sp = Symbol_find(lemp->start);
688 if( sp==0 ){
689 ErrorMsg(lemp->filename,0,
690"The specified start symbol \"%s\" is not \
691in a nonterminal of the grammar. \"%s\" will be used as the start \
692symbol instead.",lemp->start,lemp->rule->lhs->name);
693 lemp->errorcnt++;
694 sp = lemp->rule->lhs;
695 }
696 }else{
697 sp = lemp->rule->lhs;
698 }
699
700 /* Make sure the start symbol doesn't occur on the right-hand side of
701 ** any rule. Report an error if it does. (YACC would generate a new
702 ** start symbol in this case.) */
703 for(rp=lemp->rule; rp; rp=rp->next){
704 int i;
705 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000706 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000707 ErrorMsg(lemp->filename,0,
708"The start symbol \"%s\" occurs on the \
709right-hand side of a rule. This will result in a parser which \
710does not work properly.",sp->name);
711 lemp->errorcnt++;
712 }
713 }
714 }
715
716 /* The basis configuration set for the first state
717 ** is all rules which have the start symbol as their
718 ** left-hand side */
719 for(rp=sp->rule; rp; rp=rp->nextlhs){
720 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000721 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000722 newcfp = Configlist_addbasis(rp,0);
723 SetAdd(newcfp->fws,0);
724 }
725
726 /* Compute the first state. All other states will be
727 ** computed automatically during the computation of the first one.
728 ** The returned pointer to the first state is not used. */
729 (void)getstate(lemp);
730 return;
731}
732
733/* Return a pointer to a state which is described by the configuration
734** list which has been built from calls to Configlist_add.
735*/
736PRIVATE void buildshifts(/* struct lemon *, struct state * */); /* Forwd ref */
737PRIVATE struct state *getstate(lemp)
738struct lemon *lemp;
739{
740 struct config *cfp, *bp;
741 struct state *stp;
742
743 /* Extract the sorted basis of the new state. The basis was constructed
744 ** by prior calls to "Configlist_addbasis()". */
745 Configlist_sortbasis();
746 bp = Configlist_basis();
747
748 /* Get a state with the same basis */
749 stp = State_find(bp);
750 if( stp ){
751 /* A state with the same basis already exists! Copy all the follow-set
752 ** propagation links from the state under construction into the
753 ** preexisting state, then return a pointer to the preexisting state */
754 struct config *x, *y;
755 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
756 Plink_copy(&y->bplp,x->bplp);
757 Plink_delete(x->fplp);
758 x->fplp = x->bplp = 0;
759 }
760 cfp = Configlist_return();
761 Configlist_eat(cfp);
762 }else{
763 /* This really is a new state. Construct all the details */
764 Configlist_closure(lemp); /* Compute the configuration closure */
765 Configlist_sort(); /* Sort the configuration closure */
766 cfp = Configlist_return(); /* Get a pointer to the config list */
767 stp = State_new(); /* A new state structure */
768 MemoryCheck(stp);
769 stp->bp = bp; /* Remember the configuration basis */
770 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000771 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000772 stp->ap = 0; /* No actions, yet. */
773 State_insert(stp,stp->bp); /* Add to the state table */
774 buildshifts(lemp,stp); /* Recursively compute successor states */
775 }
776 return stp;
777}
778
drhfd405312005-11-06 04:06:59 +0000779/*
780** Return true if two symbols are the same.
781*/
782int same_symbol(a,b)
783struct symbol *a;
784struct symbol *b;
785{
786 int i;
787 if( a==b ) return 1;
788 if( a->type!=MULTITERMINAL ) return 0;
789 if( b->type!=MULTITERMINAL ) return 0;
790 if( a->nsubsym!=b->nsubsym ) return 0;
791 for(i=0; i<a->nsubsym; i++){
792 if( a->subsym[i]!=b->subsym[i] ) return 0;
793 }
794 return 1;
795}
796
drh75897232000-05-29 14:26:00 +0000797/* Construct all successor states to the given state. A "successor"
798** state is any state which can be reached by a shift action.
799*/
800PRIVATE void buildshifts(lemp,stp)
801struct lemon *lemp;
802struct state *stp; /* The state from which successors are computed */
803{
804 struct config *cfp; /* For looping thru the config closure of "stp" */
805 struct config *bcfp; /* For the inner loop on config closure of "stp" */
806 struct config *new; /* */
807 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
808 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
809 struct state *newstp; /* A pointer to a successor state */
810
811 /* Each configuration becomes complete after it contibutes to a successor
812 ** state. Initially, all configurations are incomplete */
813 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
814
815 /* Loop through all configurations of the state "stp" */
816 for(cfp=stp->cfp; cfp; cfp=cfp->next){
817 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
818 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
819 Configlist_reset(); /* Reset the new config set */
820 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
821
822 /* For every configuration in the state "stp" which has the symbol "sp"
823 ** following its dot, add the same configuration to the basis set under
824 ** construction but with the dot shifted one symbol to the right. */
825 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
826 if( bcfp->status==COMPLETE ) continue; /* Already used */
827 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
828 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +0000829 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +0000830 bcfp->status = COMPLETE; /* Mark this config as used */
831 new = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
832 Plink_add(&new->bplp,bcfp);
833 }
834
835 /* Get a pointer to the state described by the basis configuration set
836 ** constructed in the preceding loop */
837 newstp = getstate(lemp);
838
839 /* The state "newstp" is reached from the state "stp" by a shift action
840 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +0000841 if( sp->type==MULTITERMINAL ){
842 int i;
843 for(i=0; i<sp->nsubsym; i++){
844 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
845 }
846 }else{
847 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
848 }
drh75897232000-05-29 14:26:00 +0000849 }
850}
851
852/*
853** Construct the propagation links
854*/
855void FindLinks(lemp)
856struct lemon *lemp;
857{
858 int i;
859 struct config *cfp, *other;
860 struct state *stp;
861 struct plink *plp;
862
863 /* Housekeeping detail:
864 ** Add to every propagate link a pointer back to the state to
865 ** which the link is attached. */
866 for(i=0; i<lemp->nstate; i++){
867 stp = lemp->sorted[i];
868 for(cfp=stp->cfp; cfp; cfp=cfp->next){
869 cfp->stp = stp;
870 }
871 }
872
873 /* Convert all backlinks into forward links. Only the forward
874 ** links are used in the follow-set computation. */
875 for(i=0; i<lemp->nstate; i++){
876 stp = lemp->sorted[i];
877 for(cfp=stp->cfp; cfp; cfp=cfp->next){
878 for(plp=cfp->bplp; plp; plp=plp->next){
879 other = plp->cfp;
880 Plink_add(&other->fplp,cfp);
881 }
882 }
883 }
884}
885
886/* Compute all followsets.
887**
888** A followset is the set of all symbols which can come immediately
889** after a configuration.
890*/
891void FindFollowSets(lemp)
892struct lemon *lemp;
893{
894 int i;
895 struct config *cfp;
896 struct plink *plp;
897 int progress;
898 int change;
899
900 for(i=0; i<lemp->nstate; i++){
901 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
902 cfp->status = INCOMPLETE;
903 }
904 }
905
906 do{
907 progress = 0;
908 for(i=0; i<lemp->nstate; i++){
909 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
910 if( cfp->status==COMPLETE ) continue;
911 for(plp=cfp->fplp; plp; plp=plp->next){
912 change = SetUnion(plp->cfp->fws,cfp->fws);
913 if( change ){
914 plp->cfp->status = INCOMPLETE;
915 progress = 1;
916 }
917 }
918 cfp->status = COMPLETE;
919 }
920 }
921 }while( progress );
922}
923
924static int resolve_conflict();
925
926/* Compute the reduce actions, and resolve conflicts.
927*/
928void FindActions(lemp)
929struct lemon *lemp;
930{
931 int i,j;
932 struct config *cfp;
933 struct state *stp;
934 struct symbol *sp;
935 struct rule *rp;
936
937 /* Add all of the reduce actions
938 ** A reduce action is added for each element of the followset of
939 ** a configuration which has its dot at the extreme right.
940 */
941 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
942 stp = lemp->sorted[i];
943 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
944 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
945 for(j=0; j<lemp->nterminal; j++){
946 if( SetFind(cfp->fws,j) ){
947 /* Add a reduce action to the state "stp" which will reduce by the
948 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +0000949 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +0000950 }
951 }
952 }
953 }
954 }
955
956 /* Add the accepting token */
957 if( lemp->start ){
958 sp = Symbol_find(lemp->start);
959 if( sp==0 ) sp = lemp->rule->lhs;
960 }else{
961 sp = lemp->rule->lhs;
962 }
963 /* Add to the first state (which is always the starting state of the
964 ** finite state machine) an action to ACCEPT if the lookahead is the
965 ** start nonterminal. */
966 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
967
968 /* Resolve conflicts */
969 for(i=0; i<lemp->nstate; i++){
970 struct action *ap, *nap;
971 struct state *stp;
972 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +0000973 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +0000974 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +0000975 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +0000976 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
977 /* The two actions "ap" and "nap" have the same lookahead.
978 ** Figure out which one should be used */
979 lemp->nconflict += resolve_conflict(ap,nap,lemp->errsym);
980 }
981 }
982 }
983
984 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +0000985 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000986 for(i=0; i<lemp->nstate; i++){
987 struct action *ap;
988 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +0000989 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000990 }
991 }
992 for(rp=lemp->rule; rp; rp=rp->next){
993 if( rp->canReduce ) continue;
994 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
995 lemp->errorcnt++;
996 }
997}
998
999/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001000** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001001**
1002** NO LONGER TRUE:
1003** To resolve a conflict, first look to see if either action
1004** is on an error rule. In that case, take the action which
1005** is not associated with the error rule. If neither or both
1006** actions are associated with an error rule, then try to
1007** use precedence to resolve the conflict.
1008**
1009** If either action is a SHIFT, then it must be apx. This
1010** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1011*/
1012static int resolve_conflict(apx,apy,errsym)
1013struct action *apx;
1014struct action *apy;
1015struct symbol *errsym; /* The error symbol (if defined. NULL otherwise) */
1016{
1017 struct symbol *spx, *spy;
1018 int errcnt = 0;
1019 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001020 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001021 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001022 errcnt++;
1023 }
drh75897232000-05-29 14:26:00 +00001024 if( apx->type==SHIFT && apy->type==REDUCE ){
1025 spx = apx->sp;
1026 spy = apy->x.rp->precsym;
1027 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1028 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001029 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001030 errcnt++;
1031 }else if( spx->prec>spy->prec ){ /* Lower precedence wins */
1032 apy->type = RD_RESOLVED;
1033 }else if( spx->prec<spy->prec ){
1034 apx->type = SH_RESOLVED;
1035 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1036 apy->type = RD_RESOLVED; /* associativity */
1037 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1038 apx->type = SH_RESOLVED;
1039 }else{
1040 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh9892c5d2007-12-21 00:02:11 +00001041 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001042 errcnt++;
1043 }
1044 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1045 spx = apx->x.rp->precsym;
1046 spy = apy->x.rp->precsym;
1047 if( spx==0 || spy==0 || spx->prec<0 ||
1048 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001049 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001050 errcnt++;
1051 }else if( spx->prec>spy->prec ){
1052 apy->type = RD_RESOLVED;
1053 }else if( spx->prec<spy->prec ){
1054 apx->type = RD_RESOLVED;
1055 }
1056 }else{
drhb59499c2002-02-23 18:45:13 +00001057 assert(
1058 apx->type==SH_RESOLVED ||
1059 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001060 apx->type==SSCONFLICT ||
1061 apx->type==SRCONFLICT ||
1062 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001063 apy->type==SH_RESOLVED ||
1064 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001065 apy->type==SSCONFLICT ||
1066 apy->type==SRCONFLICT ||
1067 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001068 );
1069 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1070 ** REDUCEs on the list. If we reach this point it must be because
1071 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001072 }
1073 return errcnt;
1074}
1075/********************* From the file "configlist.c" *************************/
1076/*
1077** Routines to processing a configuration list and building a state
1078** in the LEMON parser generator.
1079*/
1080
1081static struct config *freelist = 0; /* List of free configurations */
1082static struct config *current = 0; /* Top of list of configurations */
1083static struct config **currentend = 0; /* Last on list of configs */
1084static struct config *basis = 0; /* Top of list of basis configs */
1085static struct config **basisend = 0; /* End of list of basis configs */
1086
1087/* Return a pointer to a new configuration */
1088PRIVATE struct config *newconfig(){
1089 struct config *new;
1090 if( freelist==0 ){
1091 int i;
1092 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001093 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001094 if( freelist==0 ){
1095 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1096 exit(1);
1097 }
1098 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1099 freelist[amt-1].next = 0;
1100 }
1101 new = freelist;
1102 freelist = freelist->next;
1103 return new;
1104}
1105
1106/* The configuration "old" is no longer used */
1107PRIVATE void deleteconfig(old)
1108struct config *old;
1109{
1110 old->next = freelist;
1111 freelist = old;
1112}
1113
1114/* Initialized the configuration list builder */
1115void Configlist_init(){
1116 current = 0;
1117 currentend = &current;
1118 basis = 0;
1119 basisend = &basis;
1120 Configtable_init();
1121 return;
1122}
1123
1124/* Initialized the configuration list builder */
1125void Configlist_reset(){
1126 current = 0;
1127 currentend = &current;
1128 basis = 0;
1129 basisend = &basis;
1130 Configtable_clear(0);
1131 return;
1132}
1133
1134/* Add another configuration to the configuration list */
1135struct config *Configlist_add(rp,dot)
1136struct rule *rp; /* The rule */
1137int dot; /* Index into the RHS of the rule where the dot goes */
1138{
1139 struct config *cfp, model;
1140
1141 assert( currentend!=0 );
1142 model.rp = rp;
1143 model.dot = dot;
1144 cfp = Configtable_find(&model);
1145 if( cfp==0 ){
1146 cfp = newconfig();
1147 cfp->rp = rp;
1148 cfp->dot = dot;
1149 cfp->fws = SetNew();
1150 cfp->stp = 0;
1151 cfp->fplp = cfp->bplp = 0;
1152 cfp->next = 0;
1153 cfp->bp = 0;
1154 *currentend = cfp;
1155 currentend = &cfp->next;
1156 Configtable_insert(cfp);
1157 }
1158 return cfp;
1159}
1160
1161/* Add a basis configuration to the configuration list */
1162struct config *Configlist_addbasis(rp,dot)
1163struct rule *rp;
1164int dot;
1165{
1166 struct config *cfp, model;
1167
1168 assert( basisend!=0 );
1169 assert( currentend!=0 );
1170 model.rp = rp;
1171 model.dot = dot;
1172 cfp = Configtable_find(&model);
1173 if( cfp==0 ){
1174 cfp = newconfig();
1175 cfp->rp = rp;
1176 cfp->dot = dot;
1177 cfp->fws = SetNew();
1178 cfp->stp = 0;
1179 cfp->fplp = cfp->bplp = 0;
1180 cfp->next = 0;
1181 cfp->bp = 0;
1182 *currentend = cfp;
1183 currentend = &cfp->next;
1184 *basisend = cfp;
1185 basisend = &cfp->bp;
1186 Configtable_insert(cfp);
1187 }
1188 return cfp;
1189}
1190
1191/* Compute the closure of the configuration list */
1192void Configlist_closure(lemp)
1193struct lemon *lemp;
1194{
1195 struct config *cfp, *newcfp;
1196 struct rule *rp, *newrp;
1197 struct symbol *sp, *xsp;
1198 int i, dot;
1199
1200 assert( currentend!=0 );
1201 for(cfp=current; cfp; cfp=cfp->next){
1202 rp = cfp->rp;
1203 dot = cfp->dot;
1204 if( dot>=rp->nrhs ) continue;
1205 sp = rp->rhs[dot];
1206 if( sp->type==NONTERMINAL ){
1207 if( sp->rule==0 && sp!=lemp->errsym ){
1208 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1209 sp->name);
1210 lemp->errorcnt++;
1211 }
1212 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1213 newcfp = Configlist_add(newrp,0);
1214 for(i=dot+1; i<rp->nrhs; i++){
1215 xsp = rp->rhs[i];
1216 if( xsp->type==TERMINAL ){
1217 SetAdd(newcfp->fws,xsp->index);
1218 break;
drhfd405312005-11-06 04:06:59 +00001219 }else if( xsp->type==MULTITERMINAL ){
1220 int k;
1221 for(k=0; k<xsp->nsubsym; k++){
1222 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1223 }
1224 break;
drh75897232000-05-29 14:26:00 +00001225 }else{
1226 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001227 if( xsp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +00001228 }
1229 }
1230 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1231 }
1232 }
1233 }
1234 return;
1235}
1236
1237/* Sort the configuration list */
1238void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001239 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001240 currentend = 0;
1241 return;
1242}
1243
1244/* Sort the basis configuration list */
1245void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001246 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001247 basisend = 0;
1248 return;
1249}
1250
1251/* Return a pointer to the head of the configuration list and
1252** reset the list */
1253struct config *Configlist_return(){
1254 struct config *old;
1255 old = current;
1256 current = 0;
1257 currentend = 0;
1258 return old;
1259}
1260
1261/* Return a pointer to the head of the configuration list and
1262** reset the list */
1263struct config *Configlist_basis(){
1264 struct config *old;
1265 old = basis;
1266 basis = 0;
1267 basisend = 0;
1268 return old;
1269}
1270
1271/* Free all elements of the given configuration list */
1272void Configlist_eat(cfp)
1273struct config *cfp;
1274{
1275 struct config *nextcfp;
1276 for(; cfp; cfp=nextcfp){
1277 nextcfp = cfp->next;
1278 assert( cfp->fplp==0 );
1279 assert( cfp->bplp==0 );
1280 if( cfp->fws ) SetFree(cfp->fws);
1281 deleteconfig(cfp);
1282 }
1283 return;
1284}
1285/***************** From the file "error.c" *********************************/
1286/*
1287** Code for printing error message.
1288*/
1289
1290/* Find a good place to break "msg" so that its length is at least "min"
1291** but no more than "max". Make the point as close to max as possible.
1292*/
1293static int findbreak(msg,min,max)
1294char *msg;
1295int min;
1296int max;
1297{
1298 int i,spot;
1299 char c;
1300 for(i=spot=min; i<=max; i++){
1301 c = msg[i];
1302 if( c=='\t' ) msg[i] = ' ';
1303 if( c=='\n' ){ msg[i] = ' '; spot = i; break; }
1304 if( c==0 ){ spot = i; break; }
1305 if( c=='-' && i<max-1 ) spot = i+1;
1306 if( c==' ' ) spot = i;
1307 }
1308 return spot;
1309}
1310
1311/*
1312** The error message is split across multiple lines if necessary. The
1313** splits occur at a space, if there is a space available near the end
1314** of the line.
1315*/
1316#define ERRMSGSIZE 10000 /* Hope this is big enough. No way to error check */
1317#define LINEWIDTH 79 /* Max width of any output line */
1318#define PREFIXLIMIT 30 /* Max width of the prefix on each line */
drhf9a2e7b2003-04-15 01:49:48 +00001319void ErrorMsg(const char *filename, int lineno, const char *format, ...){
drh75897232000-05-29 14:26:00 +00001320 char errmsg[ERRMSGSIZE];
1321 char prefix[PREFIXLIMIT+10];
1322 int errmsgsize;
1323 int prefixsize;
1324 int availablewidth;
1325 va_list ap;
1326 int end, restart, base;
1327
drhf9a2e7b2003-04-15 01:49:48 +00001328 va_start(ap, format);
drh75897232000-05-29 14:26:00 +00001329 /* Prepare a prefix to be prepended to every output line */
1330 if( lineno>0 ){
1331 sprintf(prefix,"%.*s:%d: ",PREFIXLIMIT-10,filename,lineno);
1332 }else{
1333 sprintf(prefix,"%.*s: ",PREFIXLIMIT-10,filename);
1334 }
drh87cf1372008-08-13 20:09:06 +00001335 prefixsize = lemonStrlen(prefix);
drh75897232000-05-29 14:26:00 +00001336 availablewidth = LINEWIDTH - prefixsize;
1337
1338 /* Generate the error message */
1339 vsprintf(errmsg,format,ap);
1340 va_end(ap);
drh87cf1372008-08-13 20:09:06 +00001341 errmsgsize = lemonStrlen(errmsg);
drh75897232000-05-29 14:26:00 +00001342 /* Remove trailing '\n's from the error message. */
1343 while( errmsgsize>0 && errmsg[errmsgsize-1]=='\n' ){
1344 errmsg[--errmsgsize] = 0;
1345 }
1346
1347 /* Print the error message */
1348 base = 0;
1349 while( errmsg[base]!=0 ){
1350 end = restart = findbreak(&errmsg[base],0,availablewidth);
1351 restart += base;
1352 while( errmsg[restart]==' ' ) restart++;
1353 fprintf(stdout,"%s%.*s\n",prefix,end,&errmsg[base]);
1354 base = restart;
1355 }
1356}
1357/**************** From the file "main.c" ************************************/
1358/*
1359** Main program file for the LEMON parser generator.
1360*/
1361
1362/* Report an out-of-memory condition and abort. This function
1363** is used mostly by the "MemoryCheck" macro in struct.h
1364*/
1365void memory_error(){
1366 fprintf(stderr,"Out of memory. Aborting...\n");
1367 exit(1);
1368}
1369
drh6d08b4d2004-07-20 12:45:22 +00001370static int nDefine = 0; /* Number of -D options on the command line */
1371static char **azDefine = 0; /* Name of the -D macros */
1372
1373/* This routine is called with the argument to each -D command-line option.
1374** Add the macro defined to the azDefine array.
1375*/
1376static void handle_D_option(char *z){
1377 char **paz;
1378 nDefine++;
1379 azDefine = realloc(azDefine, sizeof(azDefine[0])*nDefine);
1380 if( azDefine==0 ){
1381 fprintf(stderr,"out of memory\n");
1382 exit(1);
1383 }
1384 paz = &azDefine[nDefine-1];
drh87cf1372008-08-13 20:09:06 +00001385 *paz = malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001386 if( *paz==0 ){
1387 fprintf(stderr,"out of memory\n");
1388 exit(1);
1389 }
1390 strcpy(*paz, z);
1391 for(z=*paz; *z && *z!='='; z++){}
1392 *z = 0;
1393}
1394
drh75897232000-05-29 14:26:00 +00001395
1396/* The main program. Parse the command line and do it... */
1397int main(argc,argv)
1398int argc;
1399char **argv;
1400{
1401 static int version = 0;
1402 static int rpflag = 0;
1403 static int basisflag = 0;
1404 static int compress = 0;
1405 static int quiet = 0;
1406 static int statistics = 0;
1407 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001408 static int nolinenosflag = 0;
drh75897232000-05-29 14:26:00 +00001409 static struct s_options options[] = {
1410 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1411 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001412 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh75897232000-05-29 14:26:00 +00001413 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
shane58543932008-12-10 20:10:04 +00001414 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1415 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drh75897232000-05-29 14:26:00 +00001416 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drh6d08b4d2004-07-20 12:45:22 +00001417 {OPT_FLAG, "s", (char*)&statistics,
1418 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001419 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1420 {OPT_FLAG,0,0,0}
1421 };
1422 int i;
1423 struct lemon lem;
1424
drhb0c86772000-06-02 23:21:26 +00001425 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001426 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001427 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001428 exit(0);
1429 }
drhb0c86772000-06-02 23:21:26 +00001430 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001431 fprintf(stderr,"Exactly one filename argument is required.\n");
1432 exit(1);
1433 }
drh954f6b42006-06-13 13:27:46 +00001434 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001435 lem.errorcnt = 0;
1436
1437 /* Initialize the machine */
1438 Strsafe_init();
1439 Symbol_init();
1440 State_init();
1441 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001442 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001443 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001444 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001445 Symbol_new("$");
1446 lem.errsym = Symbol_new("error");
drhc4dd3fd2008-01-22 01:48:05 +00001447 lem.errsym->useCnt = 0;
drh75897232000-05-29 14:26:00 +00001448
1449 /* Parse the input file */
1450 Parse(&lem);
1451 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001452 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001453 fprintf(stderr,"Empty grammar.\n");
1454 exit(1);
1455 }
1456
1457 /* Count and index the symbols of the grammar */
1458 lem.nsymbol = Symbol_count();
1459 Symbol_new("{default}");
1460 lem.symbols = Symbol_arrayof();
drh60d31652004-02-22 00:08:04 +00001461 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
drh75897232000-05-29 14:26:00 +00001462 qsort(lem.symbols,lem.nsymbol+1,sizeof(struct symbol*),
1463 (int(*)())Symbolcmpp);
1464 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
1465 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1466 lem.nterminal = i;
1467
1468 /* Generate a reprint of the grammar, if requested on the command line */
1469 if( rpflag ){
1470 Reprint(&lem);
1471 }else{
1472 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001473 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001474
1475 /* Find the precedence for every production rule (that has one) */
1476 FindRulePrecedences(&lem);
1477
1478 /* Compute the lambda-nonterminals and the first-sets for every
1479 ** nonterminal */
1480 FindFirstSets(&lem);
1481
1482 /* Compute all LR(0) states. Also record follow-set propagation
1483 ** links so that the follow-set can be computed later */
1484 lem.nstate = 0;
1485 FindStates(&lem);
1486 lem.sorted = State_arrayof();
1487
1488 /* Tie up loose ends on the propagation links */
1489 FindLinks(&lem);
1490
1491 /* Compute the follow set of every reducible configuration */
1492 FindFollowSets(&lem);
1493
1494 /* Compute the action tables */
1495 FindActions(&lem);
1496
1497 /* Compress the action tables */
1498 if( compress==0 ) CompressTables(&lem);
1499
drhada354d2005-11-05 15:03:59 +00001500 /* Reorder and renumber the states so that states with fewer choices
1501 ** occur at the end. */
1502 ResortStates(&lem);
1503
drh75897232000-05-29 14:26:00 +00001504 /* Generate a report of the parser generated. (the "y.output" file) */
1505 if( !quiet ) ReportOutput(&lem);
1506
1507 /* Generate the source code for the parser */
1508 ReportTable(&lem, mhflag);
1509
1510 /* Produce a header file for use by the scanner. (This step is
1511 ** omitted if the "-m" option is used because makeheaders will
1512 ** generate the file for us.) */
1513 if( !mhflag ) ReportHeader(&lem);
1514 }
1515 if( statistics ){
1516 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1517 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1518 printf(" %d states, %d parser table entries, %d conflicts\n",
1519 lem.nstate, lem.tablesize, lem.nconflict);
1520 }
1521 if( lem.nconflict ){
1522 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
1523 }
1524 exit(lem.errorcnt + lem.nconflict);
drh218dc692004-05-31 23:13:45 +00001525 return (lem.errorcnt + lem.nconflict);
drh75897232000-05-29 14:26:00 +00001526}
1527/******************** From the file "msort.c" *******************************/
1528/*
1529** A generic merge-sort program.
1530**
1531** USAGE:
1532** Let "ptr" be a pointer to some structure which is at the head of
1533** a null-terminated list. Then to sort the list call:
1534**
1535** ptr = msort(ptr,&(ptr->next),cmpfnc);
1536**
1537** In the above, "cmpfnc" is a pointer to a function which compares
1538** two instances of the structure and returns an integer, as in
1539** strcmp. The second argument is a pointer to the pointer to the
1540** second element of the linked list. This address is used to compute
1541** the offset to the "next" field within the structure. The offset to
1542** the "next" field must be constant for all structures in the list.
1543**
1544** The function returns a new pointer which is the head of the list
1545** after sorting.
1546**
1547** ALGORITHM:
1548** Merge-sort.
1549*/
1550
1551/*
1552** Return a pointer to the next structure in the linked list.
1553*/
drhba99af52001-10-25 20:37:16 +00001554#define NEXT(A) (*(char**)(((unsigned long)A)+offset))
drh75897232000-05-29 14:26:00 +00001555
1556/*
1557** Inputs:
1558** a: A sorted, null-terminated linked list. (May be null).
1559** b: A sorted, null-terminated linked list. (May be null).
1560** cmp: A pointer to the comparison function.
1561** offset: Offset in the structure to the "next" field.
1562**
1563** Return Value:
1564** A pointer to the head of a sorted list containing the elements
1565** of both a and b.
1566**
1567** Side effects:
1568** The "next" pointers for elements in the lists a and b are
1569** changed.
1570*/
drhe9278182007-07-18 18:16:29 +00001571static char *merge(
1572 char *a,
1573 char *b,
1574 int (*cmp)(const char*,const char*),
1575 int offset
1576){
drh75897232000-05-29 14:26:00 +00001577 char *ptr, *head;
1578
1579 if( a==0 ){
1580 head = b;
1581 }else if( b==0 ){
1582 head = a;
1583 }else{
drhe594bc32009-11-03 13:02:25 +00001584 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001585 ptr = a;
1586 a = NEXT(a);
1587 }else{
1588 ptr = b;
1589 b = NEXT(b);
1590 }
1591 head = ptr;
1592 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001593 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001594 NEXT(ptr) = a;
1595 ptr = a;
1596 a = NEXT(a);
1597 }else{
1598 NEXT(ptr) = b;
1599 ptr = b;
1600 b = NEXT(b);
1601 }
1602 }
1603 if( a ) NEXT(ptr) = a;
1604 else NEXT(ptr) = b;
1605 }
1606 return head;
1607}
1608
1609/*
1610** Inputs:
1611** list: Pointer to a singly-linked list of structures.
1612** next: Pointer to pointer to the second element of the list.
1613** cmp: A comparison function.
1614**
1615** Return Value:
1616** A pointer to the head of a sorted list containing the elements
1617** orginally in list.
1618**
1619** Side effects:
1620** The "next" pointers for elements in list are changed.
1621*/
1622#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001623static char *msort(
1624 char *list,
1625 char **next,
1626 int (*cmp)(const char*,const char*)
1627){
drhba99af52001-10-25 20:37:16 +00001628 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001629 char *ep;
1630 char *set[LISTSIZE];
1631 int i;
drhba99af52001-10-25 20:37:16 +00001632 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001633 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1634 while( list ){
1635 ep = list;
1636 list = NEXT(list);
1637 NEXT(ep) = 0;
1638 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1639 ep = merge(ep,set[i],cmp,offset);
1640 set[i] = 0;
1641 }
1642 set[i] = ep;
1643 }
1644 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001645 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001646 return ep;
1647}
1648/************************ From the file "option.c" **************************/
1649static char **argv;
1650static struct s_options *op;
1651static FILE *errstream;
1652
1653#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1654
1655/*
1656** Print the command line with a carrot pointing to the k-th character
1657** of the n-th field.
1658*/
1659static void errline(n,k,err)
1660int n;
1661int k;
1662FILE *err;
1663{
1664 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001665 if( argv[0] ) fprintf(err,"%s",argv[0]);
drh87cf1372008-08-13 20:09:06 +00001666 spcnt = lemonStrlen(argv[0]) + 1;
drh75897232000-05-29 14:26:00 +00001667 for(i=1; i<n && argv[i]; i++){
1668 fprintf(err," %s",argv[i]);
drh87cf1372008-08-13 20:09:06 +00001669 spcnt += lemonStrlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001670 }
1671 spcnt += k;
1672 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1673 if( spcnt<20 ){
1674 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1675 }else{
1676 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1677 }
1678}
1679
1680/*
1681** Return the index of the N-th non-switch argument. Return -1
1682** if N is out of range.
1683*/
1684static int argindex(n)
1685int n;
1686{
1687 int i;
1688 int dashdash = 0;
1689 if( argv!=0 && *argv!=0 ){
1690 for(i=1; argv[i]; i++){
1691 if( dashdash || !ISOPT(argv[i]) ){
1692 if( n==0 ) return i;
1693 n--;
1694 }
1695 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1696 }
1697 }
1698 return -1;
1699}
1700
1701static char emsg[] = "Command line syntax error: ";
1702
1703/*
1704** Process a flag command line argument.
1705*/
1706static int handleflags(i,err)
1707int i;
1708FILE *err;
1709{
1710 int v;
1711 int errcnt = 0;
1712 int j;
1713 for(j=0; op[j].label; j++){
drh87cf1372008-08-13 20:09:06 +00001714 if( strncmp(&argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001715 }
1716 v = argv[i][0]=='-' ? 1 : 0;
1717 if( op[j].label==0 ){
1718 if( err ){
1719 fprintf(err,"%sundefined option.\n",emsg);
1720 errline(i,1,err);
1721 }
1722 errcnt++;
1723 }else if( op[j].type==OPT_FLAG ){
1724 *((int*)op[j].arg) = v;
1725 }else if( op[j].type==OPT_FFLAG ){
1726 (*(void(*)())(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001727 }else if( op[j].type==OPT_FSTR ){
1728 (*(void(*)())(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001729 }else{
1730 if( err ){
1731 fprintf(err,"%smissing argument on switch.\n",emsg);
1732 errline(i,1,err);
1733 }
1734 errcnt++;
1735 }
1736 return errcnt;
1737}
1738
1739/*
1740** Process a command line switch which has an argument.
1741*/
1742static int handleswitch(i,err)
1743int i;
1744FILE *err;
1745{
1746 int lv = 0;
1747 double dv = 0.0;
1748 char *sv = 0, *end;
1749 char *cp;
1750 int j;
1751 int errcnt = 0;
1752 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001753 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001754 *cp = 0;
1755 for(j=0; op[j].label; j++){
1756 if( strcmp(argv[i],op[j].label)==0 ) break;
1757 }
1758 *cp = '=';
1759 if( op[j].label==0 ){
1760 if( err ){
1761 fprintf(err,"%sundefined option.\n",emsg);
1762 errline(i,0,err);
1763 }
1764 errcnt++;
1765 }else{
1766 cp++;
1767 switch( op[j].type ){
1768 case OPT_FLAG:
1769 case OPT_FFLAG:
1770 if( err ){
1771 fprintf(err,"%soption requires an argument.\n",emsg);
1772 errline(i,0,err);
1773 }
1774 errcnt++;
1775 break;
1776 case OPT_DBL:
1777 case OPT_FDBL:
1778 dv = strtod(cp,&end);
1779 if( *end ){
1780 if( err ){
1781 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001782 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001783 }
1784 errcnt++;
1785 }
1786 break;
1787 case OPT_INT:
1788 case OPT_FINT:
1789 lv = strtol(cp,&end,0);
1790 if( *end ){
1791 if( err ){
1792 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001793 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001794 }
1795 errcnt++;
1796 }
1797 break;
1798 case OPT_STR:
1799 case OPT_FSTR:
1800 sv = cp;
1801 break;
1802 }
1803 switch( op[j].type ){
1804 case OPT_FLAG:
1805 case OPT_FFLAG:
1806 break;
1807 case OPT_DBL:
1808 *(double*)(op[j].arg) = dv;
1809 break;
1810 case OPT_FDBL:
1811 (*(void(*)())(op[j].arg))(dv);
1812 break;
1813 case OPT_INT:
1814 *(int*)(op[j].arg) = lv;
1815 break;
1816 case OPT_FINT:
1817 (*(void(*)())(op[j].arg))((int)lv);
1818 break;
1819 case OPT_STR:
1820 *(char**)(op[j].arg) = sv;
1821 break;
1822 case OPT_FSTR:
1823 (*(void(*)())(op[j].arg))(sv);
1824 break;
1825 }
1826 }
1827 return errcnt;
1828}
1829
drhb0c86772000-06-02 23:21:26 +00001830int OptInit(a,o,err)
drh75897232000-05-29 14:26:00 +00001831char **a;
1832struct s_options *o;
1833FILE *err;
1834{
1835 int errcnt = 0;
1836 argv = a;
1837 op = o;
1838 errstream = err;
1839 if( argv && *argv && op ){
1840 int i;
1841 for(i=1; argv[i]; i++){
1842 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1843 errcnt += handleflags(i,err);
1844 }else if( strchr(argv[i],'=') ){
1845 errcnt += handleswitch(i,err);
1846 }
1847 }
1848 }
1849 if( errcnt>0 ){
1850 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001851 OptPrint();
drh75897232000-05-29 14:26:00 +00001852 exit(1);
1853 }
1854 return 0;
1855}
1856
drhb0c86772000-06-02 23:21:26 +00001857int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001858 int cnt = 0;
1859 int dashdash = 0;
1860 int i;
1861 if( argv!=0 && argv[0]!=0 ){
1862 for(i=1; argv[i]; i++){
1863 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1864 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1865 }
1866 }
1867 return cnt;
1868}
1869
drhb0c86772000-06-02 23:21:26 +00001870char *OptArg(n)
drh75897232000-05-29 14:26:00 +00001871int n;
1872{
1873 int i;
1874 i = argindex(n);
1875 return i>=0 ? argv[i] : 0;
1876}
1877
drhb0c86772000-06-02 23:21:26 +00001878void OptErr(n)
drh75897232000-05-29 14:26:00 +00001879int n;
1880{
1881 int i;
1882 i = argindex(n);
1883 if( i>=0 ) errline(i,0,errstream);
1884}
1885
drhb0c86772000-06-02 23:21:26 +00001886void OptPrint(){
drh75897232000-05-29 14:26:00 +00001887 int i;
1888 int max, len;
1889 max = 0;
1890 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00001891 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00001892 switch( op[i].type ){
1893 case OPT_FLAG:
1894 case OPT_FFLAG:
1895 break;
1896 case OPT_INT:
1897 case OPT_FINT:
1898 len += 9; /* length of "<integer>" */
1899 break;
1900 case OPT_DBL:
1901 case OPT_FDBL:
1902 len += 6; /* length of "<real>" */
1903 break;
1904 case OPT_STR:
1905 case OPT_FSTR:
1906 len += 8; /* length of "<string>" */
1907 break;
1908 }
1909 if( len>max ) max = len;
1910 }
1911 for(i=0; op[i].label; i++){
1912 switch( op[i].type ){
1913 case OPT_FLAG:
1914 case OPT_FFLAG:
1915 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
1916 break;
1917 case OPT_INT:
1918 case OPT_FINT:
1919 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001920 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001921 break;
1922 case OPT_DBL:
1923 case OPT_FDBL:
1924 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001925 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001926 break;
1927 case OPT_STR:
1928 case OPT_FSTR:
1929 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001930 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001931 break;
1932 }
1933 }
1934}
1935/*********************** From the file "parse.c" ****************************/
1936/*
1937** Input file parser for the LEMON parser generator.
1938*/
1939
1940/* The state of the parser */
1941struct pstate {
1942 char *filename; /* Name of the input file */
1943 int tokenlineno; /* Linenumber at which current token starts */
1944 int errorcnt; /* Number of errors so far */
1945 char *tokenstart; /* Text of current token */
1946 struct lemon *gp; /* Global state vector */
1947 enum e_state {
1948 INITIALIZE,
1949 WAITING_FOR_DECL_OR_RULE,
1950 WAITING_FOR_DECL_KEYWORD,
1951 WAITING_FOR_DECL_ARG,
1952 WAITING_FOR_PRECEDENCE_SYMBOL,
1953 WAITING_FOR_ARROW,
1954 IN_RHS,
1955 LHS_ALIAS_1,
1956 LHS_ALIAS_2,
1957 LHS_ALIAS_3,
1958 RHS_ALIAS_1,
1959 RHS_ALIAS_2,
1960 PRECEDENCE_MARK_1,
1961 PRECEDENCE_MARK_2,
1962 RESYNC_AFTER_RULE_ERROR,
1963 RESYNC_AFTER_DECL_ERROR,
1964 WAITING_FOR_DESTRUCTOR_SYMBOL,
drh0bd1f4e2002-06-06 18:54:39 +00001965 WAITING_FOR_DATATYPE_SYMBOL,
drhe09daa92006-06-10 13:29:31 +00001966 WAITING_FOR_FALLBACK_ID,
1967 WAITING_FOR_WILDCARD_ID
drh75897232000-05-29 14:26:00 +00001968 } state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00001969 struct symbol *fallback; /* The fallback token */
drh75897232000-05-29 14:26:00 +00001970 struct symbol *lhs; /* Left-hand side of current rule */
1971 char *lhsalias; /* Alias for the LHS */
1972 int nrhs; /* Number of right-hand side symbols seen */
1973 struct symbol *rhs[MAXRHS]; /* RHS symbols */
1974 char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
1975 struct rule *prevrule; /* Previous rule parsed */
1976 char *declkeyword; /* Keyword of a declaration */
1977 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00001978 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00001979 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00001980 enum e_assoc declassoc; /* Assign this association to decl arguments */
1981 int preccounter; /* Assign this precedence to decl arguments */
1982 struct rule *firstrule; /* Pointer to first rule in the grammar */
1983 struct rule *lastrule; /* Pointer to the most recently parsed rule */
1984};
1985
1986/* Parse a single token */
1987static void parseonetoken(psp)
1988struct pstate *psp;
1989{
1990 char *x;
1991 x = Strsafe(psp->tokenstart); /* Save the token permanently */
1992#if 0
1993 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
1994 x,psp->state);
1995#endif
1996 switch( psp->state ){
1997 case INITIALIZE:
1998 psp->prevrule = 0;
1999 psp->preccounter = 0;
2000 psp->firstrule = psp->lastrule = 0;
2001 psp->gp->nrule = 0;
2002 /* Fall thru to next case */
2003 case WAITING_FOR_DECL_OR_RULE:
2004 if( x[0]=='%' ){
2005 psp->state = WAITING_FOR_DECL_KEYWORD;
2006 }else if( islower(x[0]) ){
2007 psp->lhs = Symbol_new(x);
2008 psp->nrhs = 0;
2009 psp->lhsalias = 0;
2010 psp->state = WAITING_FOR_ARROW;
2011 }else if( x[0]=='{' ){
2012 if( psp->prevrule==0 ){
2013 ErrorMsg(psp->filename,psp->tokenlineno,
drh34ff57b2008-07-14 12:27:51 +00002014"There is no prior rule opon which to attach the code \
drh75897232000-05-29 14:26:00 +00002015fragment which begins on this line.");
2016 psp->errorcnt++;
2017 }else if( psp->prevrule->code!=0 ){
2018 ErrorMsg(psp->filename,psp->tokenlineno,
2019"Code fragment beginning on this line is not the first \
2020to follow the previous rule.");
2021 psp->errorcnt++;
2022 }else{
2023 psp->prevrule->line = psp->tokenlineno;
2024 psp->prevrule->code = &x[1];
2025 }
2026 }else if( x[0]=='[' ){
2027 psp->state = PRECEDENCE_MARK_1;
2028 }else{
2029 ErrorMsg(psp->filename,psp->tokenlineno,
2030 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2031 x);
2032 psp->errorcnt++;
2033 }
2034 break;
2035 case PRECEDENCE_MARK_1:
2036 if( !isupper(x[0]) ){
2037 ErrorMsg(psp->filename,psp->tokenlineno,
2038 "The precedence symbol must be a terminal.");
2039 psp->errorcnt++;
2040 }else if( psp->prevrule==0 ){
2041 ErrorMsg(psp->filename,psp->tokenlineno,
2042 "There is no prior rule to assign precedence \"[%s]\".",x);
2043 psp->errorcnt++;
2044 }else if( psp->prevrule->precsym!=0 ){
2045 ErrorMsg(psp->filename,psp->tokenlineno,
2046"Precedence mark on this line is not the first \
2047to follow the previous rule.");
2048 psp->errorcnt++;
2049 }else{
2050 psp->prevrule->precsym = Symbol_new(x);
2051 }
2052 psp->state = PRECEDENCE_MARK_2;
2053 break;
2054 case PRECEDENCE_MARK_2:
2055 if( x[0]!=']' ){
2056 ErrorMsg(psp->filename,psp->tokenlineno,
2057 "Missing \"]\" on precedence mark.");
2058 psp->errorcnt++;
2059 }
2060 psp->state = WAITING_FOR_DECL_OR_RULE;
2061 break;
2062 case WAITING_FOR_ARROW:
2063 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2064 psp->state = IN_RHS;
2065 }else if( x[0]=='(' ){
2066 psp->state = LHS_ALIAS_1;
2067 }else{
2068 ErrorMsg(psp->filename,psp->tokenlineno,
2069 "Expected to see a \":\" following the LHS symbol \"%s\".",
2070 psp->lhs->name);
2071 psp->errorcnt++;
2072 psp->state = RESYNC_AFTER_RULE_ERROR;
2073 }
2074 break;
2075 case LHS_ALIAS_1:
2076 if( isalpha(x[0]) ){
2077 psp->lhsalias = x;
2078 psp->state = LHS_ALIAS_2;
2079 }else{
2080 ErrorMsg(psp->filename,psp->tokenlineno,
2081 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2082 x,psp->lhs->name);
2083 psp->errorcnt++;
2084 psp->state = RESYNC_AFTER_RULE_ERROR;
2085 }
2086 break;
2087 case LHS_ALIAS_2:
2088 if( x[0]==')' ){
2089 psp->state = LHS_ALIAS_3;
2090 }else{
2091 ErrorMsg(psp->filename,psp->tokenlineno,
2092 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2093 psp->errorcnt++;
2094 psp->state = RESYNC_AFTER_RULE_ERROR;
2095 }
2096 break;
2097 case LHS_ALIAS_3:
2098 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2099 psp->state = IN_RHS;
2100 }else{
2101 ErrorMsg(psp->filename,psp->tokenlineno,
2102 "Missing \"->\" following: \"%s(%s)\".",
2103 psp->lhs->name,psp->lhsalias);
2104 psp->errorcnt++;
2105 psp->state = RESYNC_AFTER_RULE_ERROR;
2106 }
2107 break;
2108 case IN_RHS:
2109 if( x[0]=='.' ){
2110 struct rule *rp;
drh9892c5d2007-12-21 00:02:11 +00002111 rp = (struct rule *)calloc( sizeof(struct rule) +
2112 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002113 if( rp==0 ){
2114 ErrorMsg(psp->filename,psp->tokenlineno,
2115 "Can't allocate enough memory for this rule.");
2116 psp->errorcnt++;
2117 psp->prevrule = 0;
2118 }else{
2119 int i;
2120 rp->ruleline = psp->tokenlineno;
2121 rp->rhs = (struct symbol**)&rp[1];
2122 rp->rhsalias = (char**)&(rp->rhs[psp->nrhs]);
2123 for(i=0; i<psp->nrhs; i++){
2124 rp->rhs[i] = psp->rhs[i];
2125 rp->rhsalias[i] = psp->alias[i];
2126 }
2127 rp->lhs = psp->lhs;
2128 rp->lhsalias = psp->lhsalias;
2129 rp->nrhs = psp->nrhs;
2130 rp->code = 0;
2131 rp->precsym = 0;
2132 rp->index = psp->gp->nrule++;
2133 rp->nextlhs = rp->lhs->rule;
2134 rp->lhs->rule = rp;
2135 rp->next = 0;
2136 if( psp->firstrule==0 ){
2137 psp->firstrule = psp->lastrule = rp;
2138 }else{
2139 psp->lastrule->next = rp;
2140 psp->lastrule = rp;
2141 }
2142 psp->prevrule = rp;
2143 }
2144 psp->state = WAITING_FOR_DECL_OR_RULE;
2145 }else if( isalpha(x[0]) ){
2146 if( psp->nrhs>=MAXRHS ){
2147 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002148 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002149 x);
2150 psp->errorcnt++;
2151 psp->state = RESYNC_AFTER_RULE_ERROR;
2152 }else{
2153 psp->rhs[psp->nrhs] = Symbol_new(x);
2154 psp->alias[psp->nrhs] = 0;
2155 psp->nrhs++;
2156 }
drhfd405312005-11-06 04:06:59 +00002157 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2158 struct symbol *msp = psp->rhs[psp->nrhs-1];
2159 if( msp->type!=MULTITERMINAL ){
2160 struct symbol *origsp = msp;
drh9892c5d2007-12-21 00:02:11 +00002161 msp = calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002162 memset(msp, 0, sizeof(*msp));
2163 msp->type = MULTITERMINAL;
2164 msp->nsubsym = 1;
drh9892c5d2007-12-21 00:02:11 +00002165 msp->subsym = calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002166 msp->subsym[0] = origsp;
2167 msp->name = origsp->name;
2168 psp->rhs[psp->nrhs-1] = msp;
2169 }
2170 msp->nsubsym++;
2171 msp->subsym = realloc(msp->subsym, sizeof(struct symbol*)*msp->nsubsym);
2172 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2173 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2174 ErrorMsg(psp->filename,psp->tokenlineno,
2175 "Cannot form a compound containing a non-terminal");
2176 psp->errorcnt++;
2177 }
drh75897232000-05-29 14:26:00 +00002178 }else if( x[0]=='(' && psp->nrhs>0 ){
2179 psp->state = RHS_ALIAS_1;
2180 }else{
2181 ErrorMsg(psp->filename,psp->tokenlineno,
2182 "Illegal character on RHS of rule: \"%s\".",x);
2183 psp->errorcnt++;
2184 psp->state = RESYNC_AFTER_RULE_ERROR;
2185 }
2186 break;
2187 case RHS_ALIAS_1:
2188 if( isalpha(x[0]) ){
2189 psp->alias[psp->nrhs-1] = x;
2190 psp->state = RHS_ALIAS_2;
2191 }else{
2192 ErrorMsg(psp->filename,psp->tokenlineno,
2193 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2194 x,psp->rhs[psp->nrhs-1]->name);
2195 psp->errorcnt++;
2196 psp->state = RESYNC_AFTER_RULE_ERROR;
2197 }
2198 break;
2199 case RHS_ALIAS_2:
2200 if( x[0]==')' ){
2201 psp->state = IN_RHS;
2202 }else{
2203 ErrorMsg(psp->filename,psp->tokenlineno,
2204 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2205 psp->errorcnt++;
2206 psp->state = RESYNC_AFTER_RULE_ERROR;
2207 }
2208 break;
2209 case WAITING_FOR_DECL_KEYWORD:
2210 if( isalpha(x[0]) ){
2211 psp->declkeyword = x;
2212 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002213 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002214 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002215 psp->state = WAITING_FOR_DECL_ARG;
2216 if( strcmp(x,"name")==0 ){
2217 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002218 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002219 }else if( strcmp(x,"include")==0 ){
2220 psp->declargslot = &(psp->gp->include);
drh75897232000-05-29 14:26:00 +00002221 }else if( strcmp(x,"code")==0 ){
2222 psp->declargslot = &(psp->gp->extracode);
drh75897232000-05-29 14:26:00 +00002223 }else if( strcmp(x,"token_destructor")==0 ){
2224 psp->declargslot = &psp->gp->tokendest;
drh960e8c62001-04-03 16:53:21 +00002225 }else if( strcmp(x,"default_destructor")==0 ){
2226 psp->declargslot = &psp->gp->vardest;
drh75897232000-05-29 14:26:00 +00002227 }else if( strcmp(x,"token_prefix")==0 ){
2228 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002229 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002230 }else if( strcmp(x,"syntax_error")==0 ){
2231 psp->declargslot = &(psp->gp->error);
drh75897232000-05-29 14:26:00 +00002232 }else if( strcmp(x,"parse_accept")==0 ){
2233 psp->declargslot = &(psp->gp->accept);
drh75897232000-05-29 14:26:00 +00002234 }else if( strcmp(x,"parse_failure")==0 ){
2235 psp->declargslot = &(psp->gp->failure);
drh75897232000-05-29 14:26:00 +00002236 }else if( strcmp(x,"stack_overflow")==0 ){
2237 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002238 }else if( strcmp(x,"extra_argument")==0 ){
2239 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002240 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002241 }else if( strcmp(x,"token_type")==0 ){
2242 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002243 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002244 }else if( strcmp(x,"default_type")==0 ){
2245 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002246 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002247 }else if( strcmp(x,"stack_size")==0 ){
2248 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002249 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002250 }else if( strcmp(x,"start_symbol")==0 ){
2251 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002252 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002253 }else if( strcmp(x,"left")==0 ){
2254 psp->preccounter++;
2255 psp->declassoc = LEFT;
2256 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2257 }else if( strcmp(x,"right")==0 ){
2258 psp->preccounter++;
2259 psp->declassoc = RIGHT;
2260 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2261 }else if( strcmp(x,"nonassoc")==0 ){
2262 psp->preccounter++;
2263 psp->declassoc = NONE;
2264 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2265 }else if( strcmp(x,"destructor")==0 ){
2266 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
2267 }else if( strcmp(x,"type")==0 ){
2268 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002269 }else if( strcmp(x,"fallback")==0 ){
2270 psp->fallback = 0;
2271 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002272 }else if( strcmp(x,"wildcard")==0 ){
2273 psp->state = WAITING_FOR_WILDCARD_ID;
drh75897232000-05-29 14:26:00 +00002274 }else{
2275 ErrorMsg(psp->filename,psp->tokenlineno,
2276 "Unknown declaration keyword: \"%%%s\".",x);
2277 psp->errorcnt++;
2278 psp->state = RESYNC_AFTER_DECL_ERROR;
2279 }
2280 }else{
2281 ErrorMsg(psp->filename,psp->tokenlineno,
2282 "Illegal declaration keyword: \"%s\".",x);
2283 psp->errorcnt++;
2284 psp->state = RESYNC_AFTER_DECL_ERROR;
2285 }
2286 break;
2287 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2288 if( !isalpha(x[0]) ){
2289 ErrorMsg(psp->filename,psp->tokenlineno,
2290 "Symbol name missing after %destructor keyword");
2291 psp->errorcnt++;
2292 psp->state = RESYNC_AFTER_DECL_ERROR;
2293 }else{
2294 struct symbol *sp = Symbol_new(x);
2295 psp->declargslot = &sp->destructor;
drh4dc8ef52008-07-01 17:13:57 +00002296 psp->decllinenoslot = &sp->destLineno;
drha5808f32008-04-27 22:19:44 +00002297 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002298 psp->state = WAITING_FOR_DECL_ARG;
2299 }
2300 break;
2301 case WAITING_FOR_DATATYPE_SYMBOL:
2302 if( !isalpha(x[0]) ){
2303 ErrorMsg(psp->filename,psp->tokenlineno,
2304 "Symbol name missing after %destructor keyword");
2305 psp->errorcnt++;
2306 psp->state = RESYNC_AFTER_DECL_ERROR;
2307 }else{
2308 struct symbol *sp = Symbol_new(x);
2309 psp->declargslot = &sp->datatype;
drha5808f32008-04-27 22:19:44 +00002310 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002311 psp->state = WAITING_FOR_DECL_ARG;
2312 }
2313 break;
2314 case WAITING_FOR_PRECEDENCE_SYMBOL:
2315 if( x[0]=='.' ){
2316 psp->state = WAITING_FOR_DECL_OR_RULE;
2317 }else if( isupper(x[0]) ){
2318 struct symbol *sp;
2319 sp = Symbol_new(x);
2320 if( sp->prec>=0 ){
2321 ErrorMsg(psp->filename,psp->tokenlineno,
2322 "Symbol \"%s\" has already be given a precedence.",x);
2323 psp->errorcnt++;
2324 }else{
2325 sp->prec = psp->preccounter;
2326 sp->assoc = psp->declassoc;
2327 }
2328 }else{
2329 ErrorMsg(psp->filename,psp->tokenlineno,
2330 "Can't assign a precedence to \"%s\".",x);
2331 psp->errorcnt++;
2332 }
2333 break;
2334 case WAITING_FOR_DECL_ARG:
drha5808f32008-04-27 22:19:44 +00002335 if( x[0]=='{' || x[0]=='\"' || isalnum(x[0]) ){
2336 char *zOld, *zNew, *zBuf, *z;
2337 int nOld, n, nLine, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002338 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002339 char zLine[50];
2340 zNew = x;
2341 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002342 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002343 if( *psp->declargslot ){
2344 zOld = *psp->declargslot;
2345 }else{
2346 zOld = "";
2347 }
drh87cf1372008-08-13 20:09:06 +00002348 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002349 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002350 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002351 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2352 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002353 for(z=psp->filename, nBack=0; *z; z++){
2354 if( *z=='\\' ) nBack++;
2355 }
2356 sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002357 nLine = lemonStrlen(zLine);
2358 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002359 }
2360 *psp->declargslot = zBuf = realloc(*psp->declargslot, n);
2361 zBuf += nOld;
drhb5bd49e2008-07-14 12:21:08 +00002362 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002363 if( nOld && zBuf[-1]!='\n' ){
2364 *(zBuf++) = '\n';
2365 }
2366 memcpy(zBuf, zLine, nLine);
2367 zBuf += nLine;
2368 *(zBuf++) = '"';
2369 for(z=psp->filename; *z; z++){
2370 if( *z=='\\' ){
2371 *(zBuf++) = '\\';
2372 }
2373 *(zBuf++) = *z;
2374 }
2375 *(zBuf++) = '"';
2376 *(zBuf++) = '\n';
2377 }
drh4dc8ef52008-07-01 17:13:57 +00002378 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2379 psp->decllinenoslot[0] = psp->tokenlineno;
2380 }
drha5808f32008-04-27 22:19:44 +00002381 memcpy(zBuf, zNew, nNew);
2382 zBuf += nNew;
2383 *zBuf = 0;
2384 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002385 }else{
2386 ErrorMsg(psp->filename,psp->tokenlineno,
2387 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2388 psp->errorcnt++;
2389 psp->state = RESYNC_AFTER_DECL_ERROR;
2390 }
2391 break;
drh0bd1f4e2002-06-06 18:54:39 +00002392 case WAITING_FOR_FALLBACK_ID:
2393 if( x[0]=='.' ){
2394 psp->state = WAITING_FOR_DECL_OR_RULE;
2395 }else if( !isupper(x[0]) ){
2396 ErrorMsg(psp->filename, psp->tokenlineno,
2397 "%%fallback argument \"%s\" should be a token", x);
2398 psp->errorcnt++;
2399 }else{
2400 struct symbol *sp = Symbol_new(x);
2401 if( psp->fallback==0 ){
2402 psp->fallback = sp;
2403 }else if( sp->fallback ){
2404 ErrorMsg(psp->filename, psp->tokenlineno,
2405 "More than one fallback assigned to token %s", x);
2406 psp->errorcnt++;
2407 }else{
2408 sp->fallback = psp->fallback;
2409 psp->gp->has_fallback = 1;
2410 }
2411 }
2412 break;
drhe09daa92006-06-10 13:29:31 +00002413 case WAITING_FOR_WILDCARD_ID:
2414 if( x[0]=='.' ){
2415 psp->state = WAITING_FOR_DECL_OR_RULE;
2416 }else if( !isupper(x[0]) ){
2417 ErrorMsg(psp->filename, psp->tokenlineno,
2418 "%%wildcard argument \"%s\" should be a token", x);
2419 psp->errorcnt++;
2420 }else{
2421 struct symbol *sp = Symbol_new(x);
2422 if( psp->gp->wildcard==0 ){
2423 psp->gp->wildcard = sp;
2424 }else{
2425 ErrorMsg(psp->filename, psp->tokenlineno,
2426 "Extra wildcard to token: %s", x);
2427 psp->errorcnt++;
2428 }
2429 }
2430 break;
drh75897232000-05-29 14:26:00 +00002431 case RESYNC_AFTER_RULE_ERROR:
2432/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2433** break; */
2434 case RESYNC_AFTER_DECL_ERROR:
2435 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2436 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2437 break;
2438 }
2439}
2440
drh34ff57b2008-07-14 12:27:51 +00002441/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002442** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2443** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2444** comments them out. Text in between is also commented out as appropriate.
2445*/
danielk1977940fac92005-01-23 22:41:37 +00002446static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002447 int i, j, k, n;
2448 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002449 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002450 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002451 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002452 for(i=0; z[i]; i++){
2453 if( z[i]=='\n' ) lineno++;
2454 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2455 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2456 if( exclude ){
2457 exclude--;
2458 if( exclude==0 ){
2459 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2460 }
2461 }
2462 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2463 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2464 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2465 if( exclude ){
2466 exclude++;
2467 }else{
2468 for(j=i+7; isspace(z[j]); j++){}
2469 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2470 exclude = 1;
2471 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002472 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002473 exclude = 0;
2474 break;
2475 }
2476 }
2477 if( z[i+3]=='n' ) exclude = !exclude;
2478 if( exclude ){
2479 start = i;
2480 start_lineno = lineno;
2481 }
2482 }
2483 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2484 }
2485 }
2486 if( exclude ){
2487 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2488 exit(1);
2489 }
2490}
2491
drh75897232000-05-29 14:26:00 +00002492/* In spite of its name, this function is really a scanner. It read
2493** in the entire input file (all at once) then tokenizes it. Each
2494** token is passed to the function "parseonetoken" which builds all
2495** the appropriate data structures in the global state vector "gp".
2496*/
2497void Parse(gp)
2498struct lemon *gp;
2499{
2500 struct pstate ps;
2501 FILE *fp;
2502 char *filebuf;
2503 int filesize;
2504 int lineno;
2505 int c;
2506 char *cp, *nextcp;
2507 int startline = 0;
2508
rse38514a92007-09-20 11:34:17 +00002509 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002510 ps.gp = gp;
2511 ps.filename = gp->filename;
2512 ps.errorcnt = 0;
2513 ps.state = INITIALIZE;
2514
2515 /* Begin by reading the input file */
2516 fp = fopen(ps.filename,"rb");
2517 if( fp==0 ){
2518 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2519 gp->errorcnt++;
2520 return;
2521 }
2522 fseek(fp,0,2);
2523 filesize = ftell(fp);
2524 rewind(fp);
2525 filebuf = (char *)malloc( filesize+1 );
2526 if( filebuf==0 ){
2527 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2528 filesize+1);
2529 gp->errorcnt++;
2530 return;
2531 }
2532 if( fread(filebuf,1,filesize,fp)!=filesize ){
2533 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2534 filesize);
2535 free(filebuf);
2536 gp->errorcnt++;
2537 return;
2538 }
2539 fclose(fp);
2540 filebuf[filesize] = 0;
2541
drh6d08b4d2004-07-20 12:45:22 +00002542 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2543 preprocess_input(filebuf);
2544
drh75897232000-05-29 14:26:00 +00002545 /* Now scan the text of the input file */
2546 lineno = 1;
2547 for(cp=filebuf; (c= *cp)!=0; ){
2548 if( c=='\n' ) lineno++; /* Keep track of the line number */
2549 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2550 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2551 cp+=2;
2552 while( (c= *cp)!=0 && c!='\n' ) cp++;
2553 continue;
2554 }
2555 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2556 cp+=2;
2557 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2558 if( c=='\n' ) lineno++;
2559 cp++;
2560 }
2561 if( c ) cp++;
2562 continue;
2563 }
2564 ps.tokenstart = cp; /* Mark the beginning of the token */
2565 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2566 if( c=='\"' ){ /* String literals */
2567 cp++;
2568 while( (c= *cp)!=0 && c!='\"' ){
2569 if( c=='\n' ) lineno++;
2570 cp++;
2571 }
2572 if( c==0 ){
2573 ErrorMsg(ps.filename,startline,
2574"String starting on this line is not terminated before the end of the file.");
2575 ps.errorcnt++;
2576 nextcp = cp;
2577 }else{
2578 nextcp = cp+1;
2579 }
2580 }else if( c=='{' ){ /* A block of C code */
2581 int level;
2582 cp++;
2583 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2584 if( c=='\n' ) lineno++;
2585 else if( c=='{' ) level++;
2586 else if( c=='}' ) level--;
2587 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2588 int prevc;
2589 cp = &cp[2];
2590 prevc = 0;
2591 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2592 if( c=='\n' ) lineno++;
2593 prevc = c;
2594 cp++;
2595 }
2596 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
2597 cp = &cp[2];
2598 while( (c= *cp)!=0 && c!='\n' ) cp++;
2599 if( c ) lineno++;
2600 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
2601 int startchar, prevc;
2602 startchar = c;
2603 prevc = 0;
2604 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2605 if( c=='\n' ) lineno++;
2606 if( prevc=='\\' ) prevc = 0;
2607 else prevc = c;
2608 }
2609 }
2610 }
2611 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002612 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002613"C code starting on this line is not terminated before the end of the file.");
2614 ps.errorcnt++;
2615 nextcp = cp;
2616 }else{
2617 nextcp = cp+1;
2618 }
2619 }else if( isalnum(c) ){ /* Identifiers */
2620 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2621 nextcp = cp;
2622 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2623 cp += 3;
2624 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002625 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2626 cp += 2;
2627 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2628 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002629 }else{ /* All other (one character) operators */
2630 cp++;
2631 nextcp = cp;
2632 }
2633 c = *cp;
2634 *cp = 0; /* Null terminate the token */
2635 parseonetoken(&ps); /* Parse the token */
2636 *cp = c; /* Restore the buffer */
2637 cp = nextcp;
2638 }
2639 free(filebuf); /* Release the buffer after parsing */
2640 gp->rule = ps.firstrule;
2641 gp->errorcnt = ps.errorcnt;
2642}
2643/*************************** From the file "plink.c" *********************/
2644/*
2645** Routines processing configuration follow-set propagation links
2646** in the LEMON parser generator.
2647*/
2648static struct plink *plink_freelist = 0;
2649
2650/* Allocate a new plink */
2651struct plink *Plink_new(){
2652 struct plink *new;
2653
2654 if( plink_freelist==0 ){
2655 int i;
2656 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002657 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002658 if( plink_freelist==0 ){
2659 fprintf(stderr,
2660 "Unable to allocate memory for a new follow-set propagation link.\n");
2661 exit(1);
2662 }
2663 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2664 plink_freelist[amt-1].next = 0;
2665 }
2666 new = plink_freelist;
2667 plink_freelist = plink_freelist->next;
2668 return new;
2669}
2670
2671/* Add a plink to a plink list */
2672void Plink_add(plpp,cfp)
2673struct plink **plpp;
2674struct config *cfp;
2675{
2676 struct plink *new;
2677 new = Plink_new();
2678 new->next = *plpp;
2679 *plpp = new;
2680 new->cfp = cfp;
2681}
2682
2683/* Transfer every plink on the list "from" to the list "to" */
2684void Plink_copy(to,from)
2685struct plink **to;
2686struct plink *from;
2687{
2688 struct plink *nextpl;
2689 while( from ){
2690 nextpl = from->next;
2691 from->next = *to;
2692 *to = from;
2693 from = nextpl;
2694 }
2695}
2696
2697/* Delete every plink on the list */
2698void Plink_delete(plp)
2699struct plink *plp;
2700{
2701 struct plink *nextpl;
2702
2703 while( plp ){
2704 nextpl = plp->next;
2705 plp->next = plink_freelist;
2706 plink_freelist = plp;
2707 plp = nextpl;
2708 }
2709}
2710/*********************** From the file "report.c" **************************/
2711/*
2712** Procedures for generating reports and tables in the LEMON parser generator.
2713*/
2714
2715/* Generate a filename with the given suffix. Space to hold the
2716** name comes from malloc() and must be freed by the calling
2717** function.
2718*/
2719PRIVATE char *file_makename(lemp,suffix)
2720struct lemon *lemp;
2721char *suffix;
2722{
2723 char *name;
2724 char *cp;
2725
drh87cf1372008-08-13 20:09:06 +00002726 name = malloc( lemonStrlen(lemp->filename) + lemonStrlen(suffix) + 5 );
drh75897232000-05-29 14:26:00 +00002727 if( name==0 ){
2728 fprintf(stderr,"Can't allocate space for a filename.\n");
2729 exit(1);
2730 }
2731 strcpy(name,lemp->filename);
2732 cp = strrchr(name,'.');
2733 if( cp ) *cp = 0;
2734 strcat(name,suffix);
2735 return name;
2736}
2737
2738/* Open a file with a name based on the name of the input file,
2739** but with a different (specified) suffix, and return a pointer
2740** to the stream */
2741PRIVATE FILE *file_open(lemp,suffix,mode)
2742struct lemon *lemp;
2743char *suffix;
2744char *mode;
2745{
2746 FILE *fp;
2747
2748 if( lemp->outname ) free(lemp->outname);
2749 lemp->outname = file_makename(lemp, suffix);
2750 fp = fopen(lemp->outname,mode);
2751 if( fp==0 && *mode=='w' ){
2752 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2753 lemp->errorcnt++;
2754 return 0;
2755 }
2756 return fp;
2757}
2758
2759/* Duplicate the input file without comments and without actions
2760** on rules */
2761void Reprint(lemp)
2762struct lemon *lemp;
2763{
2764 struct rule *rp;
2765 struct symbol *sp;
2766 int i, j, maxlen, len, ncolumns, skip;
2767 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2768 maxlen = 10;
2769 for(i=0; i<lemp->nsymbol; i++){
2770 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00002771 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00002772 if( len>maxlen ) maxlen = len;
2773 }
2774 ncolumns = 76/(maxlen+5);
2775 if( ncolumns<1 ) ncolumns = 1;
2776 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2777 for(i=0; i<skip; i++){
2778 printf("//");
2779 for(j=i; j<lemp->nsymbol; j+=skip){
2780 sp = lemp->symbols[j];
2781 assert( sp->index==j );
2782 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2783 }
2784 printf("\n");
2785 }
2786 for(rp=lemp->rule; rp; rp=rp->next){
2787 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002788 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002789 printf(" ::=");
2790 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002791 sp = rp->rhs[i];
2792 printf(" %s", sp->name);
2793 if( sp->type==MULTITERMINAL ){
2794 for(j=1; j<sp->nsubsym; j++){
2795 printf("|%s", sp->subsym[j]->name);
2796 }
2797 }
2798 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002799 }
2800 printf(".");
2801 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002802 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002803 printf("\n");
2804 }
2805}
2806
2807void ConfigPrint(fp,cfp)
2808FILE *fp;
2809struct config *cfp;
2810{
2811 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002812 struct symbol *sp;
2813 int i, j;
drh75897232000-05-29 14:26:00 +00002814 rp = cfp->rp;
2815 fprintf(fp,"%s ::=",rp->lhs->name);
2816 for(i=0; i<=rp->nrhs; i++){
2817 if( i==cfp->dot ) fprintf(fp," *");
2818 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002819 sp = rp->rhs[i];
2820 fprintf(fp," %s", sp->name);
2821 if( sp->type==MULTITERMINAL ){
2822 for(j=1; j<sp->nsubsym; j++){
2823 fprintf(fp,"|%s",sp->subsym[j]->name);
2824 }
2825 }
drh75897232000-05-29 14:26:00 +00002826 }
2827}
2828
2829/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002830#if 0
drh75897232000-05-29 14:26:00 +00002831/* Print a set */
2832PRIVATE void SetPrint(out,set,lemp)
2833FILE *out;
2834char *set;
2835struct lemon *lemp;
2836{
2837 int i;
2838 char *spacer;
2839 spacer = "";
2840 fprintf(out,"%12s[","");
2841 for(i=0; i<lemp->nterminal; i++){
2842 if( SetFind(set,i) ){
2843 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2844 spacer = " ";
2845 }
2846 }
2847 fprintf(out,"]\n");
2848}
2849
2850/* Print a plink chain */
2851PRIVATE void PlinkPrint(out,plp,tag)
2852FILE *out;
2853struct plink *plp;
2854char *tag;
2855{
2856 while( plp ){
drhada354d2005-11-05 15:03:59 +00002857 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002858 ConfigPrint(out,plp->cfp);
2859 fprintf(out,"\n");
2860 plp = plp->next;
2861 }
2862}
2863#endif
2864
2865/* Print an action to the given file descriptor. Return FALSE if
2866** nothing was actually printed.
2867*/
2868int PrintAction(struct action *ap, FILE *fp, int indent){
2869 int result = 1;
2870 switch( ap->type ){
2871 case SHIFT:
drhada354d2005-11-05 15:03:59 +00002872 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00002873 break;
2874 case REDUCE:
2875 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
2876 break;
2877 case ACCEPT:
2878 fprintf(fp,"%*s accept",indent,ap->sp->name);
2879 break;
2880 case ERROR:
2881 fprintf(fp,"%*s error",indent,ap->sp->name);
2882 break;
drh9892c5d2007-12-21 00:02:11 +00002883 case SRCONFLICT:
2884 case RRCONFLICT:
drh75897232000-05-29 14:26:00 +00002885 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
2886 indent,ap->sp->name,ap->x.rp->index);
2887 break;
drh9892c5d2007-12-21 00:02:11 +00002888 case SSCONFLICT:
2889 fprintf(fp,"%*s shift %d ** Parsing conflict **",
2890 indent,ap->sp->name,ap->x.stp->statenum);
2891 break;
drh75897232000-05-29 14:26:00 +00002892 case SH_RESOLVED:
2893 case RD_RESOLVED:
2894 case NOT_USED:
2895 result = 0;
2896 break;
2897 }
2898 return result;
2899}
2900
2901/* Generate the "y.output" log file */
2902void ReportOutput(lemp)
2903struct lemon *lemp;
2904{
2905 int i;
2906 struct state *stp;
2907 struct config *cfp;
2908 struct action *ap;
2909 FILE *fp;
2910
drh2aa6ca42004-09-10 00:14:04 +00002911 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00002912 if( fp==0 ) return;
drh75897232000-05-29 14:26:00 +00002913 for(i=0; i<lemp->nstate; i++){
2914 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00002915 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00002916 if( lemp->basisflag ) cfp=stp->bp;
2917 else cfp=stp->cfp;
2918 while( cfp ){
2919 char buf[20];
2920 if( cfp->dot==cfp->rp->nrhs ){
2921 sprintf(buf,"(%d)",cfp->rp->index);
2922 fprintf(fp," %5s ",buf);
2923 }else{
2924 fprintf(fp," ");
2925 }
2926 ConfigPrint(fp,cfp);
2927 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00002928#if 0
drh75897232000-05-29 14:26:00 +00002929 SetPrint(fp,cfp->fws,lemp);
2930 PlinkPrint(fp,cfp->fplp,"To ");
2931 PlinkPrint(fp,cfp->bplp,"From");
2932#endif
2933 if( lemp->basisflag ) cfp=cfp->bp;
2934 else cfp=cfp->next;
2935 }
2936 fprintf(fp,"\n");
2937 for(ap=stp->ap; ap; ap=ap->next){
2938 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
2939 }
2940 fprintf(fp,"\n");
2941 }
drhe9278182007-07-18 18:16:29 +00002942 fprintf(fp, "----------------------------------------------------\n");
2943 fprintf(fp, "Symbols:\n");
2944 for(i=0; i<lemp->nsymbol; i++){
2945 int j;
2946 struct symbol *sp;
2947
2948 sp = lemp->symbols[i];
2949 fprintf(fp, " %3d: %s", i, sp->name);
2950 if( sp->type==NONTERMINAL ){
2951 fprintf(fp, ":");
2952 if( sp->lambda ){
2953 fprintf(fp, " <lambda>");
2954 }
2955 for(j=0; j<lemp->nterminal; j++){
2956 if( sp->firstset && SetFind(sp->firstset, j) ){
2957 fprintf(fp, " %s", lemp->symbols[j]->name);
2958 }
2959 }
2960 }
2961 fprintf(fp, "\n");
2962 }
drh75897232000-05-29 14:26:00 +00002963 fclose(fp);
2964 return;
2965}
2966
2967/* Search for the file "name" which is in the same directory as
2968** the exacutable */
2969PRIVATE char *pathsearch(argv0,name,modemask)
2970char *argv0;
2971char *name;
2972int modemask;
2973{
2974 char *pathlist;
2975 char *path,*cp;
2976 char c;
drh75897232000-05-29 14:26:00 +00002977
2978#ifdef __WIN32__
2979 cp = strrchr(argv0,'\\');
2980#else
2981 cp = strrchr(argv0,'/');
2982#endif
2983 if( cp ){
2984 c = *cp;
2985 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00002986 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh75897232000-05-29 14:26:00 +00002987 if( path ) sprintf(path,"%s/%s",argv0,name);
2988 *cp = c;
2989 }else{
2990 extern char *getenv();
2991 pathlist = getenv("PATH");
2992 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
drh87cf1372008-08-13 20:09:06 +00002993 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
drh75897232000-05-29 14:26:00 +00002994 if( path!=0 ){
2995 while( *pathlist ){
2996 cp = strchr(pathlist,':');
drh87cf1372008-08-13 20:09:06 +00002997 if( cp==0 ) cp = &pathlist[lemonStrlen(pathlist)];
drh75897232000-05-29 14:26:00 +00002998 c = *cp;
2999 *cp = 0;
3000 sprintf(path,"%s/%s",pathlist,name);
3001 *cp = c;
3002 if( c==0 ) pathlist = "";
3003 else pathlist = &cp[1];
3004 if( access(path,modemask)==0 ) break;
3005 }
3006 }
3007 }
3008 return path;
3009}
3010
3011/* Given an action, compute the integer value for that action
3012** which is to be put in the action table of the generated machine.
3013** Return negative if no action should be generated.
3014*/
3015PRIVATE int compute_action(lemp,ap)
3016struct lemon *lemp;
3017struct action *ap;
3018{
3019 int act;
3020 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00003021 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00003022 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
3023 case ERROR: act = lemp->nstate + lemp->nrule; break;
3024 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
3025 default: act = -1; break;
3026 }
3027 return act;
3028}
3029
3030#define LINESIZE 1000
3031/* The next cluster of routines are for reading the template file
3032** and writing the results to the generated parser */
3033/* The first function transfers data from "in" to "out" until
3034** a line is seen which begins with "%%". The line number is
3035** tracked.
3036**
3037** if name!=0, then any word that begin with "Parse" is changed to
3038** begin with *name instead.
3039*/
3040PRIVATE void tplt_xfer(name,in,out,lineno)
3041char *name;
3042FILE *in;
3043FILE *out;
3044int *lineno;
3045{
3046 int i, iStart;
3047 char line[LINESIZE];
3048 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3049 (*lineno)++;
3050 iStart = 0;
3051 if( name ){
3052 for(i=0; line[i]; i++){
3053 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
3054 && (i==0 || !isalpha(line[i-1]))
3055 ){
3056 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3057 fprintf(out,"%s",name);
3058 i += 4;
3059 iStart = i+1;
3060 }
3061 }
3062 }
3063 fprintf(out,"%s",&line[iStart]);
3064 }
3065}
3066
3067/* The next function finds the template file and opens it, returning
3068** a pointer to the opened file. */
3069PRIVATE FILE *tplt_open(lemp)
3070struct lemon *lemp;
3071{
3072 static char templatename[] = "lempar.c";
3073 char buf[1000];
3074 FILE *in;
3075 char *tpltname;
3076 char *cp;
3077
3078 cp = strrchr(lemp->filename,'.');
3079 if( cp ){
drh8b582012003-10-21 13:16:03 +00003080 sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003081 }else{
3082 sprintf(buf,"%s.lt",lemp->filename);
3083 }
3084 if( access(buf,004)==0 ){
3085 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003086 }else if( access(templatename,004)==0 ){
3087 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003088 }else{
3089 tpltname = pathsearch(lemp->argv0,templatename,0);
3090 }
3091 if( tpltname==0 ){
3092 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3093 templatename);
3094 lemp->errorcnt++;
3095 return 0;
3096 }
drh2aa6ca42004-09-10 00:14:04 +00003097 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003098 if( in==0 ){
3099 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3100 lemp->errorcnt++;
3101 return 0;
3102 }
3103 return in;
3104}
3105
drhaf805ca2004-09-07 11:28:25 +00003106/* Print a #line directive line to the output file. */
3107PRIVATE void tplt_linedir(out,lineno,filename)
3108FILE *out;
3109int lineno;
3110char *filename;
3111{
3112 fprintf(out,"#line %d \"",lineno);
3113 while( *filename ){
3114 if( *filename == '\\' ) putc('\\',out);
3115 putc(*filename,out);
3116 filename++;
3117 }
3118 fprintf(out,"\"\n");
3119}
3120
drh75897232000-05-29 14:26:00 +00003121/* Print a string to the file and keep the linenumber up to date */
drha5808f32008-04-27 22:19:44 +00003122PRIVATE void tplt_print(out,lemp,str,lineno)
drh75897232000-05-29 14:26:00 +00003123FILE *out;
3124struct lemon *lemp;
3125char *str;
drh75897232000-05-29 14:26:00 +00003126int *lineno;
3127{
3128 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003129 while( *str ){
drh75897232000-05-29 14:26:00 +00003130 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003131 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003132 str++;
3133 }
drh9db55df2004-09-09 14:01:21 +00003134 if( str[-1]!='\n' ){
3135 putc('\n',out);
3136 (*lineno)++;
3137 }
shane58543932008-12-10 20:10:04 +00003138 if (!lemp->nolinenosflag) {
3139 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3140 }
drh75897232000-05-29 14:26:00 +00003141 return;
3142}
3143
3144/*
3145** The following routine emits code for the destructor for the
3146** symbol sp
3147*/
3148void emit_destructor_code(out,sp,lemp,lineno)
3149FILE *out;
3150struct symbol *sp;
3151struct lemon *lemp;
3152int *lineno;
3153{
drhcc83b6e2004-04-23 23:38:42 +00003154 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003155
drh75897232000-05-29 14:26:00 +00003156 if( sp->type==TERMINAL ){
3157 cp = lemp->tokendest;
3158 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003159 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003160 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003161 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003162 fprintf(out,"{\n"); (*lineno)++;
shane58543932008-12-10 20:10:04 +00003163 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,sp->destLineno,lemp->filename); }
drh960e8c62001-04-03 16:53:21 +00003164 }else if( lemp->vardest ){
3165 cp = lemp->vardest;
3166 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003167 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003168 }else{
3169 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003170 }
3171 for(; *cp; cp++){
3172 if( *cp=='$' && cp[1]=='$' ){
3173 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3174 cp++;
3175 continue;
3176 }
shane58543932008-12-10 20:10:04 +00003177 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003178 fputc(*cp,out);
3179 }
shane58543932008-12-10 20:10:04 +00003180 fprintf(out,"\n"); (*lineno)++;
3181 if (!lemp->nolinenosflag) {
3182 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3183 }
3184 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003185 return;
3186}
3187
3188/*
drh960e8c62001-04-03 16:53:21 +00003189** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003190*/
3191int has_destructor(sp, lemp)
3192struct symbol *sp;
3193struct lemon *lemp;
3194{
3195 int ret;
3196 if( sp->type==TERMINAL ){
3197 ret = lemp->tokendest!=0;
3198 }else{
drh960e8c62001-04-03 16:53:21 +00003199 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003200 }
3201 return ret;
3202}
3203
drh0bb132b2004-07-20 14:06:51 +00003204/*
3205** Append text to a dynamically allocated string. If zText is 0 then
3206** reset the string to be empty again. Always return the complete text
3207** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003208**
3209** n bytes of zText are stored. If n==0 then all of zText up to the first
3210** \000 terminator is stored. zText can contain up to two instances of
3211** %d. The values of p1 and p2 are written into the first and second
3212** %d.
3213**
3214** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003215*/
3216PRIVATE char *append_str(char *zText, int n, int p1, int p2){
3217 static char *z = 0;
3218 static int alloced = 0;
3219 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003220 int c;
drh0bb132b2004-07-20 14:06:51 +00003221 char zInt[40];
3222
3223 if( zText==0 ){
3224 used = 0;
3225 return z;
3226 }
drh7ac25c72004-08-19 15:12:26 +00003227 if( n<=0 ){
3228 if( n<0 ){
3229 used += n;
3230 assert( used>=0 );
3231 }
drh87cf1372008-08-13 20:09:06 +00003232 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003233 }
drh0bb132b2004-07-20 14:06:51 +00003234 if( n+sizeof(zInt)*2+used >= alloced ){
3235 alloced = n + sizeof(zInt)*2 + used + 200;
3236 z = realloc(z, alloced);
3237 }
3238 if( z==0 ) return "";
3239 while( n-- > 0 ){
3240 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003241 if( c=='%' && n>0 && zText[0]=='d' ){
drh0bb132b2004-07-20 14:06:51 +00003242 sprintf(zInt, "%d", p1);
3243 p1 = p2;
3244 strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003245 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003246 zText++;
3247 n--;
3248 }else{
3249 z[used++] = c;
3250 }
3251 }
3252 z[used] = 0;
3253 return z;
3254}
3255
3256/*
3257** zCode is a string that is the action associated with a rule. Expand
3258** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003259** stack.
drh0bb132b2004-07-20 14:06:51 +00003260*/
drhaf805ca2004-09-07 11:28:25 +00003261PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003262 char *cp, *xp;
3263 int i;
3264 char lhsused = 0; /* True if the LHS element has been used */
3265 char used[MAXRHS]; /* True for each RHS element which is used */
3266
3267 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3268 lhsused = 0;
3269
drh19c9e562007-03-29 20:13:53 +00003270 if( rp->code==0 ){
3271 rp->code = "\n";
3272 rp->line = rp->ruleline;
3273 }
3274
drh0bb132b2004-07-20 14:06:51 +00003275 append_str(0,0,0,0);
drh19c9e562007-03-29 20:13:53 +00003276 for(cp=rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003277 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3278 char saved;
3279 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3280 saved = *xp;
3281 *xp = 0;
3282 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003283 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003284 cp = xp;
3285 lhsused = 1;
3286 }else{
3287 for(i=0; i<rp->nrhs; i++){
3288 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003289 if( cp!=rp->code && cp[-1]=='@' ){
3290 /* If the argument is of the form @X then substituted
3291 ** the token number of X, not the value of X */
3292 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3293 }else{
drhfd405312005-11-06 04:06:59 +00003294 struct symbol *sp = rp->rhs[i];
3295 int dtnum;
3296 if( sp->type==MULTITERMINAL ){
3297 dtnum = sp->subsym[0]->dtnum;
3298 }else{
3299 dtnum = sp->dtnum;
3300 }
3301 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003302 }
drh0bb132b2004-07-20 14:06:51 +00003303 cp = xp;
3304 used[i] = 1;
3305 break;
3306 }
3307 }
3308 }
3309 *xp = saved;
3310 }
3311 append_str(cp, 1, 0, 0);
3312 } /* End loop */
3313
3314 /* Check to make sure the LHS has been used */
3315 if( rp->lhsalias && !lhsused ){
3316 ErrorMsg(lemp->filename,rp->ruleline,
3317 "Label \"%s\" for \"%s(%s)\" is never used.",
3318 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3319 lemp->errorcnt++;
3320 }
3321
3322 /* Generate destructor code for RHS symbols which are not used in the
3323 ** reduce code */
3324 for(i=0; i<rp->nrhs; i++){
3325 if( rp->rhsalias[i] && !used[i] ){
3326 ErrorMsg(lemp->filename,rp->ruleline,
3327 "Label %s for \"%s(%s)\" is never used.",
3328 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3329 lemp->errorcnt++;
3330 }else if( rp->rhsalias[i]==0 ){
3331 if( has_destructor(rp->rhs[i],lemp) ){
drh639efd02008-08-13 20:04:29 +00003332 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003333 rp->rhs[i]->index,i-rp->nrhs+1);
3334 }else{
3335 /* No destructor defined for this term */
3336 }
3337 }
3338 }
drh61e339a2007-01-16 03:09:02 +00003339 if( rp->code ){
3340 cp = append_str(0,0,0,0);
3341 rp->code = Strsafe(cp?cp:"");
3342 }
drh0bb132b2004-07-20 14:06:51 +00003343}
3344
drh75897232000-05-29 14:26:00 +00003345/*
3346** Generate code which executes when the rule "rp" is reduced. Write
3347** the code to "out". Make sure lineno stays up-to-date.
3348*/
3349PRIVATE void emit_code(out,rp,lemp,lineno)
3350FILE *out;
3351struct rule *rp;
3352struct lemon *lemp;
3353int *lineno;
3354{
drh0bb132b2004-07-20 14:06:51 +00003355 char *cp;
drh75897232000-05-29 14:26:00 +00003356
3357 /* Generate code to do the reduce action */
3358 if( rp->code ){
shane58543932008-12-10 20:10:04 +00003359 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,rp->line,lemp->filename); }
drhaf805ca2004-09-07 11:28:25 +00003360 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003361 for(cp=rp->code; *cp; cp++){
shane58543932008-12-10 20:10:04 +00003362 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003363 } /* End loop */
shane58543932008-12-10 20:10:04 +00003364 fprintf(out,"}\n"); (*lineno)++;
3365 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,*lineno,lemp->outname); }
drh75897232000-05-29 14:26:00 +00003366 } /* End if( rp->code ) */
3367
drh75897232000-05-29 14:26:00 +00003368 return;
3369}
3370
3371/*
3372** Print the definition of the union used for the parser's data stack.
3373** This union contains fields for every possible data type for tokens
3374** and nonterminals. In the process of computing and printing this
3375** union, also set the ".dtnum" field of every terminal and nonterminal
3376** symbol.
3377*/
3378void print_stack_union(out,lemp,plineno,mhflag)
3379FILE *out; /* The output stream */
3380struct lemon *lemp; /* The main info structure for this parser */
3381int *plineno; /* Pointer to the line number */
3382int mhflag; /* True if generating makeheaders output */
3383{
3384 int lineno = *plineno; /* The line number of the output */
3385 char **types; /* A hash table of datatypes */
3386 int arraysize; /* Size of the "types" array */
3387 int maxdtlength; /* Maximum length of any ".datatype" field. */
3388 char *stddt; /* Standardized name for a datatype */
3389 int i,j; /* Loop counters */
3390 int hash; /* For hashing the name of a type */
3391 char *name; /* Name of the parser */
3392
3393 /* Allocate and initialize types[] and allocate stddt[] */
3394 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003395 types = (char**)calloc( arraysize, sizeof(char*) );
drh75897232000-05-29 14:26:00 +00003396 for(i=0; i<arraysize; i++) types[i] = 0;
3397 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003398 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003399 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003400 }
drh75897232000-05-29 14:26:00 +00003401 for(i=0; i<lemp->nsymbol; i++){
3402 int len;
3403 struct symbol *sp = lemp->symbols[i];
3404 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003405 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003406 if( len>maxdtlength ) maxdtlength = len;
3407 }
3408 stddt = (char*)malloc( maxdtlength*2 + 1 );
3409 if( types==0 || stddt==0 ){
3410 fprintf(stderr,"Out of memory.\n");
3411 exit(1);
3412 }
3413
3414 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3415 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003416 ** used for terminal symbols. If there is no %default_type defined then
3417 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3418 ** a datatype using the %type directive.
3419 */
drh75897232000-05-29 14:26:00 +00003420 for(i=0; i<lemp->nsymbol; i++){
3421 struct symbol *sp = lemp->symbols[i];
3422 char *cp;
3423 if( sp==lemp->errsym ){
3424 sp->dtnum = arraysize+1;
3425 continue;
3426 }
drh960e8c62001-04-03 16:53:21 +00003427 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003428 sp->dtnum = 0;
3429 continue;
3430 }
3431 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003432 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003433 j = 0;
3434 while( isspace(*cp) ) cp++;
3435 while( *cp ) stddt[j++] = *cp++;
3436 while( j>0 && isspace(stddt[j-1]) ) j--;
3437 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00003438 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00003439 sp->dtnum = 0;
3440 continue;
3441 }
drh75897232000-05-29 14:26:00 +00003442 hash = 0;
3443 for(j=0; stddt[j]; j++){
3444 hash = hash*53 + stddt[j];
3445 }
drh3b2129c2003-05-13 00:34:21 +00003446 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003447 while( types[hash] ){
3448 if( strcmp(types[hash],stddt)==0 ){
3449 sp->dtnum = hash + 1;
3450 break;
3451 }
3452 hash++;
3453 if( hash>=arraysize ) hash = 0;
3454 }
3455 if( types[hash]==0 ){
3456 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00003457 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00003458 if( types[hash]==0 ){
3459 fprintf(stderr,"Out of memory.\n");
3460 exit(1);
3461 }
3462 strcpy(types[hash],stddt);
3463 }
3464 }
3465
3466 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3467 name = lemp->name ? lemp->name : "Parse";
3468 lineno = *plineno;
3469 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3470 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3471 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3472 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3473 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00003474 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003475 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3476 for(i=0; i<arraysize; i++){
3477 if( types[i]==0 ) continue;
3478 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3479 free(types[i]);
3480 }
drhc4dd3fd2008-01-22 01:48:05 +00003481 if( lemp->errsym->useCnt ){
3482 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3483 }
drh75897232000-05-29 14:26:00 +00003484 free(stddt);
3485 free(types);
3486 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3487 *plineno = lineno;
3488}
3489
drhb29b0a52002-02-23 19:39:46 +00003490/*
3491** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003492** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003493*/
drh8b582012003-10-21 13:16:03 +00003494static const char *minimum_size_type(int lwr, int upr){
3495 if( lwr>=0 ){
3496 if( upr<=255 ){
3497 return "unsigned char";
3498 }else if( upr<65535 ){
3499 return "unsigned short int";
3500 }else{
3501 return "unsigned int";
3502 }
3503 }else if( lwr>=-127 && upr<=127 ){
3504 return "signed char";
3505 }else if( lwr>=-32767 && upr<32767 ){
3506 return "short";
drhb29b0a52002-02-23 19:39:46 +00003507 }else{
drh8b582012003-10-21 13:16:03 +00003508 return "int";
drhb29b0a52002-02-23 19:39:46 +00003509 }
3510}
3511
drhfdbf9282003-10-21 16:34:41 +00003512/*
3513** Each state contains a set of token transaction and a set of
3514** nonterminal transactions. Each of these sets makes an instance
3515** of the following structure. An array of these structures is used
3516** to order the creation of entries in the yy_action[] table.
3517*/
3518struct axset {
3519 struct state *stp; /* A pointer to a state */
3520 int isTkn; /* True to use tokens. False for non-terminals */
3521 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00003522 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00003523};
3524
3525/*
3526** Compare to axset structures for sorting purposes
3527*/
3528static int axset_compare(const void *a, const void *b){
3529 struct axset *p1 = (struct axset*)a;
3530 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00003531 int c;
3532 c = p2->nAction - p1->nAction;
3533 if( c==0 ){
3534 c = p2->iOrder - p1->iOrder;
3535 }
3536 assert( c!=0 || p1==p2 );
3537 return c;
drhfdbf9282003-10-21 16:34:41 +00003538}
3539
drhc4dd3fd2008-01-22 01:48:05 +00003540/*
3541** Write text on "out" that describes the rule "rp".
3542*/
3543static void writeRuleText(FILE *out, struct rule *rp){
3544 int j;
3545 fprintf(out,"%s ::=", rp->lhs->name);
3546 for(j=0; j<rp->nrhs; j++){
3547 struct symbol *sp = rp->rhs[j];
3548 fprintf(out," %s", sp->name);
3549 if( sp->type==MULTITERMINAL ){
3550 int k;
3551 for(k=1; k<sp->nsubsym; k++){
3552 fprintf(out,"|%s",sp->subsym[k]->name);
3553 }
3554 }
3555 }
3556}
3557
3558
drh75897232000-05-29 14:26:00 +00003559/* Generate C source code for the parser */
3560void ReportTable(lemp, mhflag)
3561struct lemon *lemp;
3562int mhflag; /* Output in makeheaders format if true */
3563{
3564 FILE *out, *in;
3565 char line[LINESIZE];
3566 int lineno;
3567 struct state *stp;
3568 struct action *ap;
3569 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003570 struct acttab *pActtab;
3571 int i, j, n;
drh75897232000-05-29 14:26:00 +00003572 char *name;
drh8b582012003-10-21 13:16:03 +00003573 int mnTknOfst, mxTknOfst;
3574 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003575 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003576
3577 in = tplt_open(lemp);
3578 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003579 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003580 if( out==0 ){
3581 fclose(in);
3582 return;
3583 }
3584 lineno = 1;
3585 tplt_xfer(lemp->name,in,out,&lineno);
3586
3587 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00003588 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00003589 if( mhflag ){
3590 char *name = file_makename(lemp, ".h");
3591 fprintf(out,"#include \"%s\"\n", name); lineno++;
3592 free(name);
3593 }
3594 tplt_xfer(lemp->name,in,out,&lineno);
3595
3596 /* Generate #defines for all tokens */
3597 if( mhflag ){
3598 char *prefix;
3599 fprintf(out,"#if INTERFACE\n"); lineno++;
3600 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3601 else prefix = "";
3602 for(i=1; i<lemp->nterminal; i++){
3603 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3604 lineno++;
3605 }
3606 fprintf(out,"#endif\n"); lineno++;
3607 }
3608 tplt_xfer(lemp->name,in,out,&lineno);
3609
3610 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003611 fprintf(out,"#define YYCODETYPE %s\n",
drh8a415d32009-06-12 13:53:51 +00003612 minimum_size_type(0, lemp->nsymbol+1)); lineno++;
drh75897232000-05-29 14:26:00 +00003613 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3614 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003615 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003616 if( lemp->wildcard ){
3617 fprintf(out,"#define YYWILDCARD %d\n",
3618 lemp->wildcard->index); lineno++;
3619 }
drh75897232000-05-29 14:26:00 +00003620 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003621 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003622 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003623 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3624 }else{
3625 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3626 }
drhca44b5a2007-02-22 23:06:58 +00003627 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003628 if( mhflag ){
3629 fprintf(out,"#if INTERFACE\n"); lineno++;
3630 }
3631 name = lemp->name ? lemp->name : "Parse";
3632 if( lemp->arg && lemp->arg[0] ){
3633 int i;
drh87cf1372008-08-13 20:09:06 +00003634 i = lemonStrlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003635 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3636 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003637 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3638 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3639 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3640 name,lemp->arg,&lemp->arg[i]); lineno++;
3641 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3642 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003643 }else{
drh1f245e42002-03-11 13:55:50 +00003644 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3645 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3646 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3647 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003648 }
3649 if( mhflag ){
3650 fprintf(out,"#endif\n"); lineno++;
3651 }
3652 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3653 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00003654 if( lemp->errsym->useCnt ){
3655 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3656 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
3657 }
drh0bd1f4e2002-06-06 18:54:39 +00003658 if( lemp->has_fallback ){
3659 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3660 }
drh75897232000-05-29 14:26:00 +00003661 tplt_xfer(lemp->name,in,out,&lineno);
3662
drh8b582012003-10-21 13:16:03 +00003663 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003664 **
drh8b582012003-10-21 13:16:03 +00003665 ** yy_action[] A single table containing all actions.
3666 ** yy_lookahead[] A table containing the lookahead for each entry in
3667 ** yy_action. Used to detect hash collisions.
3668 ** yy_shift_ofst[] For each state, the offset into yy_action for
3669 ** shifting terminals.
3670 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3671 ** shifting non-terminals after a reduce.
3672 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003673 */
drh75897232000-05-29 14:26:00 +00003674
drh8b582012003-10-21 13:16:03 +00003675 /* Compute the actions on all states and count them up */
drh9892c5d2007-12-21 00:02:11 +00003676 ax = calloc(lemp->nstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00003677 if( ax==0 ){
3678 fprintf(stderr,"malloc failed\n");
3679 exit(1);
3680 }
drh75897232000-05-29 14:26:00 +00003681 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003682 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003683 ax[i*2].stp = stp;
3684 ax[i*2].isTkn = 1;
3685 ax[i*2].nAction = stp->nTknAct;
3686 ax[i*2+1].stp = stp;
3687 ax[i*2+1].isTkn = 0;
3688 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003689 }
drh8b582012003-10-21 13:16:03 +00003690 mxTknOfst = mnTknOfst = 0;
3691 mxNtOfst = mnNtOfst = 0;
3692
drhfdbf9282003-10-21 16:34:41 +00003693 /* Compute the action table. In order to try to keep the size of the
3694 ** action table to a minimum, the heuristic of placing the largest action
3695 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003696 */
drhe594bc32009-11-03 13:02:25 +00003697 for(i=0; i<lemp->nstate*2; i++) ax[i].iOrder = i;
drhfdbf9282003-10-21 16:34:41 +00003698 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003699 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003700 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3701 stp = ax[i].stp;
3702 if( ax[i].isTkn ){
3703 for(ap=stp->ap; ap; ap=ap->next){
3704 int action;
3705 if( ap->sp->index>=lemp->nterminal ) continue;
3706 action = compute_action(lemp, ap);
3707 if( action<0 ) continue;
3708 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003709 }
drhfdbf9282003-10-21 16:34:41 +00003710 stp->iTknOfst = acttab_insert(pActtab);
3711 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3712 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3713 }else{
3714 for(ap=stp->ap; ap; ap=ap->next){
3715 int action;
3716 if( ap->sp->index<lemp->nterminal ) continue;
3717 if( ap->sp->index==lemp->nsymbol ) continue;
3718 action = compute_action(lemp, ap);
3719 if( action<0 ) continue;
3720 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003721 }
drhfdbf9282003-10-21 16:34:41 +00003722 stp->iNtOfst = acttab_insert(pActtab);
3723 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3724 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003725 }
3726 }
drhfdbf9282003-10-21 16:34:41 +00003727 free(ax);
drh8b582012003-10-21 13:16:03 +00003728
3729 /* Output the yy_action table */
drh57196282004-10-06 15:41:16 +00003730 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003731 n = acttab_size(pActtab);
3732 for(i=j=0; i<n; i++){
3733 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003734 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003735 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003736 fprintf(out, " %4d,", action);
3737 if( j==9 || i==n-1 ){
3738 fprintf(out, "\n"); lineno++;
3739 j = 0;
3740 }else{
3741 j++;
3742 }
3743 }
3744 fprintf(out, "};\n"); lineno++;
3745
3746 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003747 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003748 for(i=j=0; i<n; i++){
3749 int la = acttab_yylookahead(pActtab, i);
3750 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003751 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003752 fprintf(out, " %4d,", la);
3753 if( j==9 || i==n-1 ){
3754 fprintf(out, "\n"); lineno++;
3755 j = 0;
3756 }else{
3757 j++;
3758 }
3759 }
3760 fprintf(out, "};\n"); lineno++;
3761
3762 /* Output the yy_shift_ofst[] table */
3763 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003764 n = lemp->nstate;
3765 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
3766 fprintf(out, "#define YY_SHIFT_MAX %d\n", n-1); lineno++;
drh57196282004-10-06 15:41:16 +00003767 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003768 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003769 for(i=j=0; i<n; i++){
3770 int ofst;
3771 stp = lemp->sorted[i];
3772 ofst = stp->iTknOfst;
3773 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003774 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003775 fprintf(out, " %4d,", ofst);
3776 if( j==9 || i==n-1 ){
3777 fprintf(out, "\n"); lineno++;
3778 j = 0;
3779 }else{
3780 j++;
3781 }
3782 }
3783 fprintf(out, "};\n"); lineno++;
3784
3785 /* Output the yy_reduce_ofst[] table */
3786 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003787 n = lemp->nstate;
3788 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
3789 fprintf(out, "#define YY_REDUCE_MAX %d\n", n-1); lineno++;
drh57196282004-10-06 15:41:16 +00003790 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003791 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003792 for(i=j=0; i<n; i++){
3793 int ofst;
3794 stp = lemp->sorted[i];
3795 ofst = stp->iNtOfst;
3796 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003797 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003798 fprintf(out, " %4d,", ofst);
3799 if( j==9 || i==n-1 ){
3800 fprintf(out, "\n"); lineno++;
3801 j = 0;
3802 }else{
3803 j++;
3804 }
3805 }
3806 fprintf(out, "};\n"); lineno++;
3807
3808 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003809 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003810 n = lemp->nstate;
3811 for(i=j=0; i<n; i++){
3812 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003813 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003814 fprintf(out, " %4d,", stp->iDflt);
3815 if( j==9 || i==n-1 ){
3816 fprintf(out, "\n"); lineno++;
3817 j = 0;
3818 }else{
3819 j++;
3820 }
3821 }
3822 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003823 tplt_xfer(lemp->name,in,out,&lineno);
3824
drh0bd1f4e2002-06-06 18:54:39 +00003825 /* Generate the table of fallback tokens.
3826 */
3827 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00003828 int mx = lemp->nterminal - 1;
3829 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
3830 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00003831 struct symbol *p = lemp->symbols[i];
3832 if( p->fallback==0 ){
3833 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3834 }else{
3835 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3836 p->name, p->fallback->name);
3837 }
3838 lineno++;
3839 }
3840 }
3841 tplt_xfer(lemp->name, in, out, &lineno);
3842
3843 /* Generate a table containing the symbolic name of every symbol
3844 */
drh75897232000-05-29 14:26:00 +00003845 for(i=0; i<lemp->nsymbol; i++){
3846 sprintf(line,"\"%s\",",lemp->symbols[i]->name);
3847 fprintf(out," %-15s",line);
3848 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
3849 }
3850 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
3851 tplt_xfer(lemp->name,in,out,&lineno);
3852
drh0bd1f4e2002-06-06 18:54:39 +00003853 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00003854 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00003855 ** when tracing REDUCE actions.
3856 */
3857 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
3858 assert( rp->index==i );
drhc4dd3fd2008-01-22 01:48:05 +00003859 fprintf(out," /* %3d */ \"", i);
3860 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00003861 fprintf(out,"\",\n"); lineno++;
3862 }
3863 tplt_xfer(lemp->name,in,out,&lineno);
3864
drh75897232000-05-29 14:26:00 +00003865 /* Generate code which executes every time a symbol is popped from
3866 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00003867 ** (In other words, generate the %destructor actions)
3868 */
drh75897232000-05-29 14:26:00 +00003869 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00003870 int once = 1;
drh75897232000-05-29 14:26:00 +00003871 for(i=0; i<lemp->nsymbol; i++){
3872 struct symbol *sp = lemp->symbols[i];
3873 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003874 if( once ){
3875 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
3876 once = 0;
3877 }
drhc53eed12009-06-12 17:46:19 +00003878 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00003879 }
3880 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
3881 if( i<lemp->nsymbol ){
3882 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3883 fprintf(out," break;\n"); lineno++;
3884 }
3885 }
drh8d659732005-01-13 23:54:06 +00003886 if( lemp->vardest ){
3887 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00003888 int once = 1;
drh8d659732005-01-13 23:54:06 +00003889 for(i=0; i<lemp->nsymbol; i++){
3890 struct symbol *sp = lemp->symbols[i];
3891 if( sp==0 || sp->type==TERMINAL ||
3892 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003893 if( once ){
3894 fprintf(out, " /* Default NON-TERMINAL Destructor */\n"); lineno++;
3895 once = 0;
3896 }
drhc53eed12009-06-12 17:46:19 +00003897 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00003898 dflt_sp = sp;
3899 }
3900 if( dflt_sp!=0 ){
3901 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00003902 }
drh4dc8ef52008-07-01 17:13:57 +00003903 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00003904 }
drh75897232000-05-29 14:26:00 +00003905 for(i=0; i<lemp->nsymbol; i++){
3906 struct symbol *sp = lemp->symbols[i];
3907 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh75013012009-06-12 15:47:34 +00003908 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003909
3910 /* Combine duplicate destructors into a single case */
3911 for(j=i+1; j<lemp->nsymbol; j++){
3912 struct symbol *sp2 = lemp->symbols[j];
3913 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
3914 && sp2->dtnum==sp->dtnum
3915 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00003916 fprintf(out," case %d: /* %s */\n",
3917 sp2->index, sp2->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003918 sp2->destructor = 0;
3919 }
3920 }
3921
drh75897232000-05-29 14:26:00 +00003922 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3923 fprintf(out," break;\n"); lineno++;
3924 }
drh75897232000-05-29 14:26:00 +00003925 tplt_xfer(lemp->name,in,out,&lineno);
3926
3927 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00003928 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00003929 tplt_xfer(lemp->name,in,out,&lineno);
3930
3931 /* Generate the table of rule information
3932 **
3933 ** Note: This code depends on the fact that rules are number
3934 ** sequentually beginning with 0.
3935 */
3936 for(rp=lemp->rule; rp; rp=rp->next){
3937 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
3938 }
3939 tplt_xfer(lemp->name,in,out,&lineno);
3940
3941 /* Generate code which execution during each REDUCE action */
3942 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00003943 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00003944 }
drhc53eed12009-06-12 17:46:19 +00003945 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00003946 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00003947 struct rule *rp2; /* Other rules with the same action */
drh0bb132b2004-07-20 14:06:51 +00003948 if( rp->code==0 ) continue;
drhc53eed12009-06-12 17:46:19 +00003949 if( rp->code[0]=='\n' && rp->code[1]==0 ) continue; /* Will be default: */
drhc4dd3fd2008-01-22 01:48:05 +00003950 fprintf(out," case %d: /* ", rp->index);
3951 writeRuleText(out, rp);
3952 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003953 for(rp2=rp->next; rp2; rp2=rp2->next){
3954 if( rp2->code==rp->code ){
drhc4dd3fd2008-01-22 01:48:05 +00003955 fprintf(out," case %d: /* ", rp2->index);
3956 writeRuleText(out, rp2);
drh8a415d32009-06-12 13:53:51 +00003957 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003958 rp2->code = 0;
3959 }
3960 }
drh75897232000-05-29 14:26:00 +00003961 emit_code(out,rp,lemp,&lineno);
3962 fprintf(out," break;\n"); lineno++;
drhc53eed12009-06-12 17:46:19 +00003963 rp->code = 0;
drh75897232000-05-29 14:26:00 +00003964 }
drhc53eed12009-06-12 17:46:19 +00003965 /* Finally, output the default: rule. We choose as the default: all
3966 ** empty actions. */
3967 fprintf(out," default:\n"); lineno++;
3968 for(rp=lemp->rule; rp; rp=rp->next){
3969 if( rp->code==0 ) continue;
3970 assert( rp->code[0]=='\n' && rp->code[1]==0 );
3971 fprintf(out," /* (%d) ", rp->index);
3972 writeRuleText(out, rp);
3973 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->index); lineno++;
3974 }
3975 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003976 tplt_xfer(lemp->name,in,out,&lineno);
3977
3978 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00003979 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00003980 tplt_xfer(lemp->name,in,out,&lineno);
3981
3982 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00003983 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00003984 tplt_xfer(lemp->name,in,out,&lineno);
3985
3986 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00003987 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00003988 tplt_xfer(lemp->name,in,out,&lineno);
3989
3990 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00003991 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00003992
3993 fclose(in);
3994 fclose(out);
3995 return;
3996}
3997
3998/* Generate a header file for the parser */
3999void ReportHeader(lemp)
4000struct lemon *lemp;
4001{
4002 FILE *out, *in;
4003 char *prefix;
4004 char line[LINESIZE];
4005 char pattern[LINESIZE];
4006 int i;
4007
4008 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4009 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004010 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004011 if( in ){
4012 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
4013 sprintf(pattern,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4014 if( strcmp(line,pattern) ) break;
4015 }
4016 fclose(in);
4017 if( i==lemp->nterminal ){
4018 /* No change in the file. Don't rewrite it. */
4019 return;
4020 }
4021 }
drh2aa6ca42004-09-10 00:14:04 +00004022 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004023 if( out ){
4024 for(i=1; i<lemp->nterminal; i++){
4025 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4026 }
4027 fclose(out);
4028 }
4029 return;
4030}
4031
4032/* Reduce the size of the action tables, if possible, by making use
4033** of defaults.
4034**
drhb59499c2002-02-23 18:45:13 +00004035** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004036** it the default. Except, there is no default if the wildcard token
4037** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004038*/
4039void CompressTables(lemp)
4040struct lemon *lemp;
4041{
4042 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00004043 struct action *ap, *ap2;
4044 struct rule *rp, *rp2, *rbest;
4045 int nbest, n;
drh75897232000-05-29 14:26:00 +00004046 int i;
drhe09daa92006-06-10 13:29:31 +00004047 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004048
4049 for(i=0; i<lemp->nstate; i++){
4050 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004051 nbest = 0;
4052 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004053 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004054
drhb59499c2002-02-23 18:45:13 +00004055 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004056 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4057 usesWildcard = 1;
4058 }
drhb59499c2002-02-23 18:45:13 +00004059 if( ap->type!=REDUCE ) continue;
4060 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004061 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004062 if( rp==rbest ) continue;
4063 n = 1;
4064 for(ap2=ap->next; ap2; ap2=ap2->next){
4065 if( ap2->type!=REDUCE ) continue;
4066 rp2 = ap2->x.rp;
4067 if( rp2==rbest ) continue;
4068 if( rp2==rp ) n++;
4069 }
4070 if( n>nbest ){
4071 nbest = n;
4072 rbest = rp;
drh75897232000-05-29 14:26:00 +00004073 }
4074 }
drhb59499c2002-02-23 18:45:13 +00004075
4076 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004077 ** is not at least 1 or if the wildcard token is a possible
4078 ** lookahead.
4079 */
4080 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004081
drhb59499c2002-02-23 18:45:13 +00004082
4083 /* Combine matching REDUCE actions into a single default */
4084 for(ap=stp->ap; ap; ap=ap->next){
4085 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4086 }
drh75897232000-05-29 14:26:00 +00004087 assert( ap );
4088 ap->sp = Symbol_new("{default}");
4089 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004090 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004091 }
4092 stp->ap = Action_sort(stp->ap);
4093 }
4094}
drhb59499c2002-02-23 18:45:13 +00004095
drhada354d2005-11-05 15:03:59 +00004096
4097/*
4098** Compare two states for sorting purposes. The smaller state is the
4099** one with the most non-terminal actions. If they have the same number
4100** of non-terminal actions, then the smaller is the one with the most
4101** token actions.
4102*/
4103static int stateResortCompare(const void *a, const void *b){
4104 const struct state *pA = *(const struct state**)a;
4105 const struct state *pB = *(const struct state**)b;
4106 int n;
4107
4108 n = pB->nNtAct - pA->nNtAct;
4109 if( n==0 ){
4110 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004111 if( n==0 ){
4112 n = pB->statenum - pA->statenum;
4113 }
drhada354d2005-11-05 15:03:59 +00004114 }
drhe594bc32009-11-03 13:02:25 +00004115 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004116 return n;
4117}
4118
4119
4120/*
4121** Renumber and resort states so that states with fewer choices
4122** occur at the end. Except, keep state 0 as the first state.
4123*/
4124void ResortStates(lemp)
4125struct lemon *lemp;
4126{
4127 int i;
4128 struct state *stp;
4129 struct action *ap;
4130
4131 for(i=0; i<lemp->nstate; i++){
4132 stp = lemp->sorted[i];
4133 stp->nTknAct = stp->nNtAct = 0;
4134 stp->iDflt = lemp->nstate + lemp->nrule;
4135 stp->iTknOfst = NO_OFFSET;
4136 stp->iNtOfst = NO_OFFSET;
4137 for(ap=stp->ap; ap; ap=ap->next){
4138 if( compute_action(lemp,ap)>=0 ){
4139 if( ap->sp->index<lemp->nterminal ){
4140 stp->nTknAct++;
4141 }else if( ap->sp->index<lemp->nsymbol ){
4142 stp->nNtAct++;
4143 }else{
4144 stp->iDflt = compute_action(lemp, ap);
4145 }
4146 }
4147 }
4148 }
4149 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4150 stateResortCompare);
4151 for(i=0; i<lemp->nstate; i++){
4152 lemp->sorted[i]->statenum = i;
4153 }
4154}
4155
4156
drh75897232000-05-29 14:26:00 +00004157/***************** From the file "set.c" ************************************/
4158/*
4159** Set manipulation routines for the LEMON parser generator.
4160*/
4161
4162static int size = 0;
4163
4164/* Set the set size */
4165void SetSize(n)
4166int n;
4167{
4168 size = n+1;
4169}
4170
4171/* Allocate a new set */
4172char *SetNew(){
4173 char *s;
drh9892c5d2007-12-21 00:02:11 +00004174 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004175 if( s==0 ){
4176 extern void memory_error();
4177 memory_error();
4178 }
drh75897232000-05-29 14:26:00 +00004179 return s;
4180}
4181
4182/* Deallocate a set */
4183void SetFree(s)
4184char *s;
4185{
4186 free(s);
4187}
4188
4189/* Add a new element to the set. Return TRUE if the element was added
4190** and FALSE if it was already there. */
4191int SetAdd(s,e)
4192char *s;
4193int e;
4194{
4195 int rv;
drh9892c5d2007-12-21 00:02:11 +00004196 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004197 rv = s[e];
4198 s[e] = 1;
4199 return !rv;
4200}
4201
4202/* Add every element of s2 to s1. Return TRUE if s1 changes. */
4203int SetUnion(s1,s2)
4204char *s1;
4205char *s2;
4206{
4207 int i, progress;
4208 progress = 0;
4209 for(i=0; i<size; i++){
4210 if( s2[i]==0 ) continue;
4211 if( s1[i]==0 ){
4212 progress = 1;
4213 s1[i] = 1;
4214 }
4215 }
4216 return progress;
4217}
4218/********************** From the file "table.c" ****************************/
4219/*
4220** All code in this file has been automatically generated
4221** from a specification in the file
4222** "table.q"
4223** by the associative array code building program "aagen".
4224** Do not edit this file! Instead, edit the specification
4225** file, then rerun aagen.
4226*/
4227/*
4228** Code for processing tables in the LEMON parser generator.
4229*/
4230
4231PRIVATE int strhash(x)
4232char *x;
4233{
4234 int h = 0;
4235 while( *x) h = h*13 + *(x++);
4236 return h;
4237}
4238
4239/* Works like strdup, sort of. Save a string in malloced memory, but
4240** keep strings in a table so that the same string is not in more
4241** than one place.
4242*/
4243char *Strsafe(y)
4244char *y;
4245{
4246 char *z;
4247
drh916f75f2006-07-17 00:19:39 +00004248 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004249 z = Strsafe_find(y);
drh87cf1372008-08-13 20:09:06 +00004250 if( z==0 && (z=malloc( lemonStrlen(y)+1 ))!=0 ){
drh75897232000-05-29 14:26:00 +00004251 strcpy(z,y);
4252 Strsafe_insert(z);
4253 }
4254 MemoryCheck(z);
4255 return z;
4256}
4257
4258/* There is one instance of the following structure for each
4259** associative array of type "x1".
4260*/
4261struct s_x1 {
4262 int size; /* The number of available slots. */
4263 /* Must be a power of 2 greater than or */
4264 /* equal to 1 */
4265 int count; /* Number of currently slots filled */
4266 struct s_x1node *tbl; /* The data stored here */
4267 struct s_x1node **ht; /* Hash table for lookups */
4268};
4269
4270/* There is one instance of this structure for every data element
4271** in an associative array of type "x1".
4272*/
4273typedef struct s_x1node {
4274 char *data; /* The data */
4275 struct s_x1node *next; /* Next entry with the same hash */
4276 struct s_x1node **from; /* Previous link */
4277} x1node;
4278
4279/* There is only one instance of the array, which is the following */
4280static struct s_x1 *x1a;
4281
4282/* Allocate a new associative array */
4283void Strsafe_init(){
4284 if( x1a ) return;
4285 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4286 if( x1a ){
4287 x1a->size = 1024;
4288 x1a->count = 0;
4289 x1a->tbl = (x1node*)malloc(
4290 (sizeof(x1node) + sizeof(x1node*))*1024 );
4291 if( x1a->tbl==0 ){
4292 free(x1a);
4293 x1a = 0;
4294 }else{
4295 int i;
4296 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4297 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4298 }
4299 }
4300}
4301/* Insert a new record into the array. Return TRUE if successful.
4302** Prior data with the same key is NOT overwritten */
4303int Strsafe_insert(data)
4304char *data;
4305{
4306 x1node *np;
4307 int h;
4308 int ph;
4309
4310 if( x1a==0 ) return 0;
4311 ph = strhash(data);
4312 h = ph & (x1a->size-1);
4313 np = x1a->ht[h];
4314 while( np ){
4315 if( strcmp(np->data,data)==0 ){
4316 /* An existing entry with the same key is found. */
4317 /* Fail because overwrite is not allows. */
4318 return 0;
4319 }
4320 np = np->next;
4321 }
4322 if( x1a->count>=x1a->size ){
4323 /* Need to make the hash table bigger */
4324 int i,size;
4325 struct s_x1 array;
4326 array.size = size = x1a->size*2;
4327 array.count = x1a->count;
4328 array.tbl = (x1node*)malloc(
4329 (sizeof(x1node) + sizeof(x1node*))*size );
4330 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4331 array.ht = (x1node**)&(array.tbl[size]);
4332 for(i=0; i<size; i++) array.ht[i] = 0;
4333 for(i=0; i<x1a->count; i++){
4334 x1node *oldnp, *newnp;
4335 oldnp = &(x1a->tbl[i]);
4336 h = strhash(oldnp->data) & (size-1);
4337 newnp = &(array.tbl[i]);
4338 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4339 newnp->next = array.ht[h];
4340 newnp->data = oldnp->data;
4341 newnp->from = &(array.ht[h]);
4342 array.ht[h] = newnp;
4343 }
4344 free(x1a->tbl);
4345 *x1a = array;
4346 }
4347 /* Insert the new data */
4348 h = ph & (x1a->size-1);
4349 np = &(x1a->tbl[x1a->count++]);
4350 np->data = data;
4351 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4352 np->next = x1a->ht[h];
4353 x1a->ht[h] = np;
4354 np->from = &(x1a->ht[h]);
4355 return 1;
4356}
4357
4358/* Return a pointer to data assigned to the given key. Return NULL
4359** if no such key. */
4360char *Strsafe_find(key)
4361char *key;
4362{
4363 int h;
4364 x1node *np;
4365
4366 if( x1a==0 ) return 0;
4367 h = strhash(key) & (x1a->size-1);
4368 np = x1a->ht[h];
4369 while( np ){
4370 if( strcmp(np->data,key)==0 ) break;
4371 np = np->next;
4372 }
4373 return np ? np->data : 0;
4374}
4375
4376/* Return a pointer to the (terminal or nonterminal) symbol "x".
4377** Create a new symbol if this is the first time "x" has been seen.
4378*/
4379struct symbol *Symbol_new(x)
4380char *x;
4381{
4382 struct symbol *sp;
4383
4384 sp = Symbol_find(x);
4385 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00004386 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00004387 MemoryCheck(sp);
4388 sp->name = Strsafe(x);
4389 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4390 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004391 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004392 sp->prec = -1;
4393 sp->assoc = UNK;
4394 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00004395 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00004396 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00004397 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00004398 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00004399 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00004400 Symbol_insert(sp,sp->name);
4401 }
drhc4dd3fd2008-01-22 01:48:05 +00004402 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00004403 return sp;
4404}
4405
drh60d31652004-02-22 00:08:04 +00004406/* Compare two symbols for working purposes
4407**
4408** Symbols that begin with upper case letters (terminals or tokens)
4409** must sort before symbols that begin with lower case letters
4410** (non-terminals). Other than that, the order does not matter.
4411**
4412** We find experimentally that leaving the symbols in their original
4413** order (the order they appeared in the grammar file) gives the
4414** smallest parser tables in SQLite.
4415*/
4416int Symbolcmpp(struct symbol **a, struct symbol **b){
4417 int i1 = (**a).index + 10000000*((**a).name[0]>'Z');
4418 int i2 = (**b).index + 10000000*((**b).name[0]>'Z');
drhe594bc32009-11-03 13:02:25 +00004419 assert( i1!=i2 || strcmp((**a).name,(**b).name)==0 );
drh60d31652004-02-22 00:08:04 +00004420 return i1-i2;
drh75897232000-05-29 14:26:00 +00004421}
4422
4423/* There is one instance of the following structure for each
4424** associative array of type "x2".
4425*/
4426struct s_x2 {
4427 int size; /* The number of available slots. */
4428 /* Must be a power of 2 greater than or */
4429 /* equal to 1 */
4430 int count; /* Number of currently slots filled */
4431 struct s_x2node *tbl; /* The data stored here */
4432 struct s_x2node **ht; /* Hash table for lookups */
4433};
4434
4435/* There is one instance of this structure for every data element
4436** in an associative array of type "x2".
4437*/
4438typedef struct s_x2node {
4439 struct symbol *data; /* The data */
4440 char *key; /* The key */
4441 struct s_x2node *next; /* Next entry with the same hash */
4442 struct s_x2node **from; /* Previous link */
4443} x2node;
4444
4445/* There is only one instance of the array, which is the following */
4446static struct s_x2 *x2a;
4447
4448/* Allocate a new associative array */
4449void Symbol_init(){
4450 if( x2a ) return;
4451 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4452 if( x2a ){
4453 x2a->size = 128;
4454 x2a->count = 0;
4455 x2a->tbl = (x2node*)malloc(
4456 (sizeof(x2node) + sizeof(x2node*))*128 );
4457 if( x2a->tbl==0 ){
4458 free(x2a);
4459 x2a = 0;
4460 }else{
4461 int i;
4462 x2a->ht = (x2node**)&(x2a->tbl[128]);
4463 for(i=0; i<128; i++) x2a->ht[i] = 0;
4464 }
4465 }
4466}
4467/* Insert a new record into the array. Return TRUE if successful.
4468** Prior data with the same key is NOT overwritten */
4469int Symbol_insert(data,key)
4470struct symbol *data;
4471char *key;
4472{
4473 x2node *np;
4474 int h;
4475 int ph;
4476
4477 if( x2a==0 ) return 0;
4478 ph = strhash(key);
4479 h = ph & (x2a->size-1);
4480 np = x2a->ht[h];
4481 while( np ){
4482 if( strcmp(np->key,key)==0 ){
4483 /* An existing entry with the same key is found. */
4484 /* Fail because overwrite is not allows. */
4485 return 0;
4486 }
4487 np = np->next;
4488 }
4489 if( x2a->count>=x2a->size ){
4490 /* Need to make the hash table bigger */
4491 int i,size;
4492 struct s_x2 array;
4493 array.size = size = x2a->size*2;
4494 array.count = x2a->count;
4495 array.tbl = (x2node*)malloc(
4496 (sizeof(x2node) + sizeof(x2node*))*size );
4497 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4498 array.ht = (x2node**)&(array.tbl[size]);
4499 for(i=0; i<size; i++) array.ht[i] = 0;
4500 for(i=0; i<x2a->count; i++){
4501 x2node *oldnp, *newnp;
4502 oldnp = &(x2a->tbl[i]);
4503 h = strhash(oldnp->key) & (size-1);
4504 newnp = &(array.tbl[i]);
4505 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4506 newnp->next = array.ht[h];
4507 newnp->key = oldnp->key;
4508 newnp->data = oldnp->data;
4509 newnp->from = &(array.ht[h]);
4510 array.ht[h] = newnp;
4511 }
4512 free(x2a->tbl);
4513 *x2a = array;
4514 }
4515 /* Insert the new data */
4516 h = ph & (x2a->size-1);
4517 np = &(x2a->tbl[x2a->count++]);
4518 np->key = key;
4519 np->data = data;
4520 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4521 np->next = x2a->ht[h];
4522 x2a->ht[h] = np;
4523 np->from = &(x2a->ht[h]);
4524 return 1;
4525}
4526
4527/* Return a pointer to data assigned to the given key. Return NULL
4528** if no such key. */
4529struct symbol *Symbol_find(key)
4530char *key;
4531{
4532 int h;
4533 x2node *np;
4534
4535 if( x2a==0 ) return 0;
4536 h = strhash(key) & (x2a->size-1);
4537 np = x2a->ht[h];
4538 while( np ){
4539 if( strcmp(np->key,key)==0 ) break;
4540 np = np->next;
4541 }
4542 return np ? np->data : 0;
4543}
4544
4545/* Return the n-th data. Return NULL if n is out of range. */
4546struct symbol *Symbol_Nth(n)
4547int n;
4548{
4549 struct symbol *data;
4550 if( x2a && n>0 && n<=x2a->count ){
4551 data = x2a->tbl[n-1].data;
4552 }else{
4553 data = 0;
4554 }
4555 return data;
4556}
4557
4558/* Return the size of the array */
4559int Symbol_count()
4560{
4561 return x2a ? x2a->count : 0;
4562}
4563
4564/* Return an array of pointers to all data in the table.
4565** The array is obtained from malloc. Return NULL if memory allocation
4566** problems, or if the array is empty. */
4567struct symbol **Symbol_arrayof()
4568{
4569 struct symbol **array;
4570 int i,size;
4571 if( x2a==0 ) return 0;
4572 size = x2a->count;
drh9892c5d2007-12-21 00:02:11 +00004573 array = (struct symbol **)calloc(size, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00004574 if( array ){
4575 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4576 }
4577 return array;
4578}
4579
4580/* Compare two configurations */
4581int Configcmp(a,b)
4582struct config *a;
4583struct config *b;
4584{
4585 int x;
4586 x = a->rp->index - b->rp->index;
4587 if( x==0 ) x = a->dot - b->dot;
4588 return x;
4589}
4590
4591/* Compare two states */
4592PRIVATE int statecmp(a,b)
4593struct config *a;
4594struct config *b;
4595{
4596 int rc;
4597 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4598 rc = a->rp->index - b->rp->index;
4599 if( rc==0 ) rc = a->dot - b->dot;
4600 }
4601 if( rc==0 ){
4602 if( a ) rc = 1;
4603 if( b ) rc = -1;
4604 }
4605 return rc;
4606}
4607
4608/* Hash a state */
4609PRIVATE int statehash(a)
4610struct config *a;
4611{
4612 int h=0;
4613 while( a ){
4614 h = h*571 + a->rp->index*37 + a->dot;
4615 a = a->bp;
4616 }
4617 return h;
4618}
4619
4620/* Allocate a new state structure */
4621struct state *State_new()
4622{
4623 struct state *new;
drh9892c5d2007-12-21 00:02:11 +00004624 new = (struct state *)calloc(1, sizeof(struct state) );
drh75897232000-05-29 14:26:00 +00004625 MemoryCheck(new);
4626 return new;
4627}
4628
4629/* There is one instance of the following structure for each
4630** associative array of type "x3".
4631*/
4632struct s_x3 {
4633 int size; /* The number of available slots. */
4634 /* Must be a power of 2 greater than or */
4635 /* equal to 1 */
4636 int count; /* Number of currently slots filled */
4637 struct s_x3node *tbl; /* The data stored here */
4638 struct s_x3node **ht; /* Hash table for lookups */
4639};
4640
4641/* There is one instance of this structure for every data element
4642** in an associative array of type "x3".
4643*/
4644typedef struct s_x3node {
4645 struct state *data; /* The data */
4646 struct config *key; /* The key */
4647 struct s_x3node *next; /* Next entry with the same hash */
4648 struct s_x3node **from; /* Previous link */
4649} x3node;
4650
4651/* There is only one instance of the array, which is the following */
4652static struct s_x3 *x3a;
4653
4654/* Allocate a new associative array */
4655void State_init(){
4656 if( x3a ) return;
4657 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4658 if( x3a ){
4659 x3a->size = 128;
4660 x3a->count = 0;
4661 x3a->tbl = (x3node*)malloc(
4662 (sizeof(x3node) + sizeof(x3node*))*128 );
4663 if( x3a->tbl==0 ){
4664 free(x3a);
4665 x3a = 0;
4666 }else{
4667 int i;
4668 x3a->ht = (x3node**)&(x3a->tbl[128]);
4669 for(i=0; i<128; i++) x3a->ht[i] = 0;
4670 }
4671 }
4672}
4673/* Insert a new record into the array. Return TRUE if successful.
4674** Prior data with the same key is NOT overwritten */
4675int State_insert(data,key)
4676struct state *data;
4677struct config *key;
4678{
4679 x3node *np;
4680 int h;
4681 int ph;
4682
4683 if( x3a==0 ) return 0;
4684 ph = statehash(key);
4685 h = ph & (x3a->size-1);
4686 np = x3a->ht[h];
4687 while( np ){
4688 if( statecmp(np->key,key)==0 ){
4689 /* An existing entry with the same key is found. */
4690 /* Fail because overwrite is not allows. */
4691 return 0;
4692 }
4693 np = np->next;
4694 }
4695 if( x3a->count>=x3a->size ){
4696 /* Need to make the hash table bigger */
4697 int i,size;
4698 struct s_x3 array;
4699 array.size = size = x3a->size*2;
4700 array.count = x3a->count;
4701 array.tbl = (x3node*)malloc(
4702 (sizeof(x3node) + sizeof(x3node*))*size );
4703 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4704 array.ht = (x3node**)&(array.tbl[size]);
4705 for(i=0; i<size; i++) array.ht[i] = 0;
4706 for(i=0; i<x3a->count; i++){
4707 x3node *oldnp, *newnp;
4708 oldnp = &(x3a->tbl[i]);
4709 h = statehash(oldnp->key) & (size-1);
4710 newnp = &(array.tbl[i]);
4711 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4712 newnp->next = array.ht[h];
4713 newnp->key = oldnp->key;
4714 newnp->data = oldnp->data;
4715 newnp->from = &(array.ht[h]);
4716 array.ht[h] = newnp;
4717 }
4718 free(x3a->tbl);
4719 *x3a = array;
4720 }
4721 /* Insert the new data */
4722 h = ph & (x3a->size-1);
4723 np = &(x3a->tbl[x3a->count++]);
4724 np->key = key;
4725 np->data = data;
4726 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4727 np->next = x3a->ht[h];
4728 x3a->ht[h] = np;
4729 np->from = &(x3a->ht[h]);
4730 return 1;
4731}
4732
4733/* Return a pointer to data assigned to the given key. Return NULL
4734** if no such key. */
4735struct state *State_find(key)
4736struct config *key;
4737{
4738 int h;
4739 x3node *np;
4740
4741 if( x3a==0 ) return 0;
4742 h = statehash(key) & (x3a->size-1);
4743 np = x3a->ht[h];
4744 while( np ){
4745 if( statecmp(np->key,key)==0 ) break;
4746 np = np->next;
4747 }
4748 return np ? np->data : 0;
4749}
4750
4751/* Return an array of pointers to all data in the table.
4752** The array is obtained from malloc. Return NULL if memory allocation
4753** problems, or if the array is empty. */
4754struct state **State_arrayof()
4755{
4756 struct state **array;
4757 int i,size;
4758 if( x3a==0 ) return 0;
4759 size = x3a->count;
4760 array = (struct state **)malloc( sizeof(struct state *)*size );
4761 if( array ){
4762 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4763 }
4764 return array;
4765}
4766
4767/* Hash a configuration */
4768PRIVATE int confighash(a)
4769struct config *a;
4770{
4771 int h=0;
4772 h = h*571 + a->rp->index*37 + a->dot;
4773 return h;
4774}
4775
4776/* There is one instance of the following structure for each
4777** associative array of type "x4".
4778*/
4779struct s_x4 {
4780 int size; /* The number of available slots. */
4781 /* Must be a power of 2 greater than or */
4782 /* equal to 1 */
4783 int count; /* Number of currently slots filled */
4784 struct s_x4node *tbl; /* The data stored here */
4785 struct s_x4node **ht; /* Hash table for lookups */
4786};
4787
4788/* There is one instance of this structure for every data element
4789** in an associative array of type "x4".
4790*/
4791typedef struct s_x4node {
4792 struct config *data; /* The data */
4793 struct s_x4node *next; /* Next entry with the same hash */
4794 struct s_x4node **from; /* Previous link */
4795} x4node;
4796
4797/* There is only one instance of the array, which is the following */
4798static struct s_x4 *x4a;
4799
4800/* Allocate a new associative array */
4801void Configtable_init(){
4802 if( x4a ) return;
4803 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4804 if( x4a ){
4805 x4a->size = 64;
4806 x4a->count = 0;
4807 x4a->tbl = (x4node*)malloc(
4808 (sizeof(x4node) + sizeof(x4node*))*64 );
4809 if( x4a->tbl==0 ){
4810 free(x4a);
4811 x4a = 0;
4812 }else{
4813 int i;
4814 x4a->ht = (x4node**)&(x4a->tbl[64]);
4815 for(i=0; i<64; i++) x4a->ht[i] = 0;
4816 }
4817 }
4818}
4819/* Insert a new record into the array. Return TRUE if successful.
4820** Prior data with the same key is NOT overwritten */
4821int Configtable_insert(data)
4822struct config *data;
4823{
4824 x4node *np;
4825 int h;
4826 int ph;
4827
4828 if( x4a==0 ) return 0;
4829 ph = confighash(data);
4830 h = ph & (x4a->size-1);
4831 np = x4a->ht[h];
4832 while( np ){
4833 if( Configcmp(np->data,data)==0 ){
4834 /* An existing entry with the same key is found. */
4835 /* Fail because overwrite is not allows. */
4836 return 0;
4837 }
4838 np = np->next;
4839 }
4840 if( x4a->count>=x4a->size ){
4841 /* Need to make the hash table bigger */
4842 int i,size;
4843 struct s_x4 array;
4844 array.size = size = x4a->size*2;
4845 array.count = x4a->count;
4846 array.tbl = (x4node*)malloc(
4847 (sizeof(x4node) + sizeof(x4node*))*size );
4848 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4849 array.ht = (x4node**)&(array.tbl[size]);
4850 for(i=0; i<size; i++) array.ht[i] = 0;
4851 for(i=0; i<x4a->count; i++){
4852 x4node *oldnp, *newnp;
4853 oldnp = &(x4a->tbl[i]);
4854 h = confighash(oldnp->data) & (size-1);
4855 newnp = &(array.tbl[i]);
4856 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4857 newnp->next = array.ht[h];
4858 newnp->data = oldnp->data;
4859 newnp->from = &(array.ht[h]);
4860 array.ht[h] = newnp;
4861 }
4862 free(x4a->tbl);
4863 *x4a = array;
4864 }
4865 /* Insert the new data */
4866 h = ph & (x4a->size-1);
4867 np = &(x4a->tbl[x4a->count++]);
4868 np->data = data;
4869 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
4870 np->next = x4a->ht[h];
4871 x4a->ht[h] = np;
4872 np->from = &(x4a->ht[h]);
4873 return 1;
4874}
4875
4876/* Return a pointer to data assigned to the given key. Return NULL
4877** if no such key. */
4878struct config *Configtable_find(key)
4879struct config *key;
4880{
4881 int h;
4882 x4node *np;
4883
4884 if( x4a==0 ) return 0;
4885 h = confighash(key) & (x4a->size-1);
4886 np = x4a->ht[h];
4887 while( np ){
4888 if( Configcmp(np->data,key)==0 ) break;
4889 np = np->next;
4890 }
4891 return np ? np->data : 0;
4892}
4893
4894/* Remove all data from the table. Pass each data to the function "f"
4895** as it is removed. ("f" may be null to avoid this step.) */
4896void Configtable_clear(f)
4897int(*f)(/* struct config * */);
4898{
4899 int i;
4900 if( x4a==0 || x4a->count==0 ) return;
4901 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
4902 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
4903 x4a->count = 0;
4904 return;
4905}