blob: 3acd9e53650c4f5da65c88c61a0e45ffff146420 [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>
drh75897232000-05-29 14:26:00 +000014
drh75897232000-05-29 14:26:00 +000015#ifndef __WIN32__
16# if defined(_WIN32) || defined(WIN32)
17# define __WIN32__
18# endif
19#endif
20
21/* #define PRIVATE static */
22#define PRIVATE
23
24#ifdef TEST
25#define MAXRHS 5 /* Set low to exercise exception code */
26#else
27#define MAXRHS 1000
28#endif
29
30char *msort();
31extern void *malloc();
32
33/******** From the file "action.h" *************************************/
34struct action *Action_new();
35struct action *Action_sort();
drh75897232000-05-29 14:26:00 +000036
37/********* From the file "assert.h" ************************************/
38void myassert();
39#ifndef NDEBUG
40# define assert(X) if(!(X))myassert(__FILE__,__LINE__)
41#else
42# define assert(X)
43#endif
44
45/********** From the file "build.h" ************************************/
46void FindRulePrecedences();
47void FindFirstSets();
48void FindStates();
49void FindLinks();
50void FindFollowSets();
51void FindActions();
52
53/********* From the file "configlist.h" *********************************/
54void Configlist_init(/* void */);
55struct config *Configlist_add(/* struct rule *, int */);
56struct config *Configlist_addbasis(/* struct rule *, int */);
57void Configlist_closure(/* void */);
58void Configlist_sort(/* void */);
59void Configlist_sortbasis(/* void */);
60struct config *Configlist_return(/* void */);
61struct config *Configlist_basis(/* void */);
62void Configlist_eat(/* struct config * */);
63void Configlist_reset(/* void */);
64
65/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +000066void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +000067
68/****** From the file "option.h" ******************************************/
69struct s_options {
70 enum { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
71 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR} type;
72 char *label;
73 char *arg;
74 char *message;
75};
drhb0c86772000-06-02 23:21:26 +000076int OptInit(/* char**,struct s_options*,FILE* */);
77int OptNArgs(/* void */);
78char *OptArg(/* int */);
79void OptErr(/* int */);
80void OptPrint(/* void */);
drh75897232000-05-29 14:26:00 +000081
82/******** From the file "parse.h" *****************************************/
83void Parse(/* struct lemon *lemp */);
84
85/********* From the file "plink.h" ***************************************/
86struct plink *Plink_new(/* void */);
87void Plink_add(/* struct plink **, struct config * */);
88void Plink_copy(/* struct plink **, struct plink * */);
89void Plink_delete(/* struct plink * */);
90
91/********** From the file "report.h" *************************************/
92void Reprint(/* struct lemon * */);
93void ReportOutput(/* struct lemon * */);
94void ReportTable(/* struct lemon * */);
95void ReportHeader(/* struct lemon * */);
96void CompressTables(/* struct lemon * */);
drhada354d2005-11-05 15:03:59 +000097void ResortStates(/* struct lemon * */);
drh75897232000-05-29 14:26:00 +000098
99/********** From the file "set.h" ****************************************/
100void SetSize(/* int N */); /* All sets will be of size N */
101char *SetNew(/* void */); /* A new set for element 0..N */
102void SetFree(/* char* */); /* Deallocate a set */
103
104int SetAdd(/* char*,int */); /* Add element to a set */
105int SetUnion(/* char *A,char *B */); /* A <- A U B, thru element N */
106
107#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
108
109/********** From the file "struct.h" *************************************/
110/*
111** Principal data structures for the LEMON parser generator.
112*/
113
drhb27b83a2002-08-14 23:18:57 +0000114typedef enum {B_FALSE=0, B_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000115
116/* Symbols (terminals and nonterminals) of the grammar are stored
117** in the following: */
118struct symbol {
119 char *name; /* Name of the symbol */
120 int index; /* Index number for this symbol */
121 enum {
122 TERMINAL,
drhfd405312005-11-06 04:06:59 +0000123 NONTERMINAL,
124 MULTITERMINAL
drh75897232000-05-29 14:26:00 +0000125 } type; /* Symbols are all either TERMINALS or NTs */
126 struct rule *rule; /* Linked list of rules of this (if an NT) */
drh0bd1f4e2002-06-06 18:54:39 +0000127 struct symbol *fallback; /* fallback token in case this token doesn't parse */
drh75897232000-05-29 14:26:00 +0000128 int prec; /* Precedence if defined (-1 otherwise) */
129 enum e_assoc {
130 LEFT,
131 RIGHT,
132 NONE,
133 UNK
134 } assoc; /* Associativity if predecence is defined */
135 char *firstset; /* First-set for all rules of this symbol */
136 Boolean lambda; /* True if NT and can generate an empty string */
137 char *destructor; /* Code which executes whenever this symbol is
138 ** popped from the stack during error processing */
139 int destructorln; /* Line number of destructor code */
140 char *datatype; /* The data type of information held by this
141 ** object. Only used if type==NONTERMINAL */
142 int dtnum; /* The data type number. In the parser, the value
143 ** stack is a union. The .yy%d element of this
144 ** union is the correct data type for this object */
drhfd405312005-11-06 04:06:59 +0000145 /* The following fields are used by MULTITERMINALs only */
146 int nsubsym; /* Number of constituent symbols in the MULTI */
147 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000148};
149
150/* Each production rule in the grammar is stored in the following
151** structure. */
152struct rule {
153 struct symbol *lhs; /* Left-hand side of the rule */
154 char *lhsalias; /* Alias for the LHS (NULL if none) */
155 int ruleline; /* Line number for the rule */
156 int nrhs; /* Number of RHS symbols */
157 struct symbol **rhs; /* The RHS symbols */
158 char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
159 int line; /* Line number at which code begins */
160 char *code; /* The code executed when this rule is reduced */
161 struct symbol *precsym; /* Precedence symbol for this rule */
162 int index; /* An index number for this rule */
163 Boolean canReduce; /* True if this rule is ever reduced */
164 struct rule *nextlhs; /* Next rule with the same LHS */
165 struct rule *next; /* Next rule in the global list */
166};
167
168/* A configuration is a production rule of the grammar together with
169** a mark (dot) showing how much of that rule has been processed so far.
170** Configurations also contain a follow-set which is a list of terminal
171** symbols which are allowed to immediately follow the end of the rule.
172** Every configuration is recorded as an instance of the following: */
173struct config {
174 struct rule *rp; /* The rule upon which the configuration is based */
175 int dot; /* The parse point */
176 char *fws; /* Follow-set for this configuration only */
177 struct plink *fplp; /* Follow-set forward propagation links */
178 struct plink *bplp; /* Follow-set backwards propagation links */
179 struct state *stp; /* Pointer to state which contains this */
180 enum {
181 COMPLETE, /* The status is used during followset and */
182 INCOMPLETE /* shift computations */
183 } status;
184 struct config *next; /* Next configuration in the state */
185 struct config *bp; /* The next basis configuration */
186};
187
188/* Every shift or reduce operation is stored as one of the following */
189struct action {
190 struct symbol *sp; /* The look-ahead symbol */
191 enum e_action {
192 SHIFT,
193 ACCEPT,
194 REDUCE,
195 ERROR,
196 CONFLICT, /* Was a reduce, but part of a conflict */
197 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
198 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
199 NOT_USED /* Deleted by compression */
200 } type;
201 union {
202 struct state *stp; /* The new state, if a shift */
203 struct rule *rp; /* The rule, if a reduce */
204 } x;
205 struct action *next; /* Next action for this state */
206 struct action *collide; /* Next action with the same hash */
207};
208
209/* Each state of the generated parser's finite state machine
210** is encoded as an instance of the following structure. */
211struct state {
212 struct config *bp; /* The basis configurations for this state */
213 struct config *cfp; /* All configurations in this set */
drhada354d2005-11-05 15:03:59 +0000214 int statenum; /* Sequencial number for this state */
drh75897232000-05-29 14:26:00 +0000215 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000216 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
217 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
218 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000219};
drh8b582012003-10-21 13:16:03 +0000220#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000221
222/* A followset propagation link indicates that the contents of one
223** configuration followset should be propagated to another whenever
224** the first changes. */
225struct plink {
226 struct config *cfp; /* The configuration to which linked */
227 struct plink *next; /* The next propagate link */
228};
229
230/* The state vector for the entire parser generator is recorded as
231** follows. (LEMON uses no global variables and makes little use of
232** static variables. Fields in the following structure can be thought
233** of as begin global variables in the program.) */
234struct lemon {
235 struct state **sorted; /* Table of states sorted by state number */
236 struct rule *rule; /* List of all rules */
237 int nstate; /* Number of states */
238 int nrule; /* Number of rules */
239 int nsymbol; /* Number of terminal and nonterminal symbols */
240 int nterminal; /* Number of terminal symbols */
241 struct symbol **symbols; /* Sorted array of pointers to symbols */
242 int errorcnt; /* Number of errors */
243 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000244 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000245 char *name; /* Name of the generated parser */
246 char *arg; /* Declaration of the 3th argument to parser */
247 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000248 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000249 char *start; /* Name of the start symbol for the grammar */
250 char *stacksize; /* Size of the parser stack */
251 char *include; /* Code to put at the start of the C file */
252 int includeln; /* Line number for start of include code */
253 char *error; /* Code to execute when an error is seen */
254 int errorln; /* Line number for start of error code */
255 char *overflow; /* Code to execute on a stack overflow */
256 int overflowln; /* Line number for start of overflow code */
257 char *failure; /* Code to execute on parser failure */
258 int failureln; /* Line number for start of failure code */
259 char *accept; /* Code to execute when the parser excepts */
260 int acceptln; /* Line number for the start of accept code */
261 char *extracode; /* Code appended to the generated file */
262 int extracodeln; /* Line number for the start of the extra code */
263 char *tokendest; /* Code to execute to destroy token data */
264 int tokendestln; /* Line number for token destroyer code */
drh960e8c62001-04-03 16:53:21 +0000265 char *vardest; /* Code for the default non-terminal destructor */
266 int vardestln; /* Line number for default non-term destructor code*/
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 */
drh0bd1f4e2002-06-06 18:54:39 +0000273 int has_fallback; /* True if any %fallback is seen in the grammer */
drh75897232000-05-29 14:26:00 +0000274 char *argv0; /* Name of the program */
275};
276
277#define MemoryCheck(X) if((X)==0){ \
278 extern void memory_error(); \
279 memory_error(); \
280}
281
282/**************** From the file "table.h" *********************************/
283/*
284** All code in this file has been automatically generated
285** from a specification in the file
286** "table.q"
287** by the associative array code building program "aagen".
288** Do not edit this file! Instead, edit the specification
289** file, then rerun aagen.
290*/
291/*
292** Code for processing tables in the LEMON parser generator.
293*/
294
295/* Routines for handling a strings */
296
297char *Strsafe();
298
299void Strsafe_init(/* void */);
300int Strsafe_insert(/* char * */);
301char *Strsafe_find(/* char * */);
302
303/* Routines for handling symbols of the grammar */
304
305struct symbol *Symbol_new();
306int Symbolcmpp(/* struct symbol **, struct symbol ** */);
307void Symbol_init(/* void */);
308int Symbol_insert(/* struct symbol *, char * */);
309struct symbol *Symbol_find(/* char * */);
310struct symbol *Symbol_Nth(/* int */);
311int Symbol_count(/* */);
312struct symbol **Symbol_arrayof(/* */);
313
314/* Routines to manage the state table */
315
316int Configcmp(/* struct config *, struct config * */);
317struct state *State_new();
318void State_init(/* void */);
319int State_insert(/* struct state *, struct config * */);
320struct state *State_find(/* struct config * */);
321struct state **State_arrayof(/* */);
322
323/* Routines used for efficiency in Configlist_add */
324
325void Configtable_init(/* void */);
326int Configtable_insert(/* struct config * */);
327struct config *Configtable_find(/* struct config * */);
328void Configtable_clear(/* int(*)(struct config *) */);
329/****************** From the file "action.c" *******************************/
330/*
331** Routines processing parser actions in the LEMON parser generator.
332*/
333
334/* Allocate a new parser action */
335struct action *Action_new(){
336 static struct action *freelist = 0;
337 struct action *new;
338
339 if( freelist==0 ){
340 int i;
341 int amt = 100;
342 freelist = (struct action *)malloc( sizeof(struct action)*amt );
343 if( freelist==0 ){
344 fprintf(stderr,"Unable to allocate memory for a new parser action.");
345 exit(1);
346 }
347 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
348 freelist[amt-1].next = 0;
349 }
350 new = freelist;
351 freelist = freelist->next;
352 return new;
353}
354
355/* Compare two actions */
356static int actioncmp(ap1,ap2)
357struct action *ap1;
358struct action *ap2;
359{
360 int rc;
361 rc = ap1->sp->index - ap2->sp->index;
362 if( rc==0 ) rc = (int)ap1->type - (int)ap2->type;
363 if( rc==0 ){
drh75897232000-05-29 14:26:00 +0000364 rc = ap1->x.rp->index - ap2->x.rp->index;
365 }
366 return rc;
367}
368
369/* Sort parser actions */
370struct action *Action_sort(ap)
371struct action *ap;
372{
drh218dc692004-05-31 23:13:45 +0000373 ap = (struct action *)msort((char *)ap,(char **)&ap->next,actioncmp);
drh75897232000-05-29 14:26:00 +0000374 return ap;
375}
376
377void Action_add(app,type,sp,arg)
378struct action **app;
379enum e_action type;
380struct symbol *sp;
381char *arg;
382{
383 struct action *new;
384 new = Action_new();
385 new->next = *app;
386 *app = new;
387 new->type = type;
388 new->sp = sp;
389 if( type==SHIFT ){
390 new->x.stp = (struct state *)arg;
391 }else{
392 new->x.rp = (struct rule *)arg;
393 }
394}
drh8b582012003-10-21 13:16:03 +0000395/********************** New code to implement the "acttab" module ***********/
396/*
397** This module implements routines use to construct the yy_action[] table.
398*/
399
400/*
401** The state of the yy_action table under construction is an instance of
402** the following structure
403*/
404typedef struct acttab acttab;
405struct acttab {
406 int nAction; /* Number of used slots in aAction[] */
407 int nActionAlloc; /* Slots allocated for aAction[] */
408 struct {
409 int lookahead; /* Value of the lookahead token */
410 int action; /* Action to take on the given lookahead */
411 } *aAction, /* The yy_action[] table under construction */
412 *aLookahead; /* A single new transaction set */
413 int mnLookahead; /* Minimum aLookahead[].lookahead */
414 int mnAction; /* Action associated with mnLookahead */
415 int mxLookahead; /* Maximum aLookahead[].lookahead */
416 int nLookahead; /* Used slots in aLookahead[] */
417 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
418};
419
420/* Return the number of entries in the yy_action table */
421#define acttab_size(X) ((X)->nAction)
422
423/* The value for the N-th entry in yy_action */
424#define acttab_yyaction(X,N) ((X)->aAction[N].action)
425
426/* The value for the N-th entry in yy_lookahead */
427#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
428
429/* Free all memory associated with the given acttab */
430void acttab_free(acttab *p){
431 free( p->aAction );
432 free( p->aLookahead );
433 free( p );
434}
435
436/* Allocate a new acttab structure */
437acttab *acttab_alloc(void){
438 acttab *p = malloc( sizeof(*p) );
439 if( p==0 ){
440 fprintf(stderr,"Unable to allocate memory for a new acttab.");
441 exit(1);
442 }
443 memset(p, 0, sizeof(*p));
444 return p;
445}
446
447/* Add a new action to the current transaction set
448*/
449void acttab_action(acttab *p, int lookahead, int action){
450 if( p->nLookahead>=p->nLookaheadAlloc ){
451 p->nLookaheadAlloc += 25;
452 p->aLookahead = realloc( p->aLookahead,
453 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
454 if( p->aLookahead==0 ){
455 fprintf(stderr,"malloc failed\n");
456 exit(1);
457 }
458 }
459 if( p->nLookahead==0 ){
460 p->mxLookahead = lookahead;
461 p->mnLookahead = lookahead;
462 p->mnAction = action;
463 }else{
464 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
465 if( p->mnLookahead>lookahead ){
466 p->mnLookahead = lookahead;
467 p->mnAction = action;
468 }
469 }
470 p->aLookahead[p->nLookahead].lookahead = lookahead;
471 p->aLookahead[p->nLookahead].action = action;
472 p->nLookahead++;
473}
474
475/*
476** Add the transaction set built up with prior calls to acttab_action()
477** into the current action table. Then reset the transaction set back
478** to an empty set in preparation for a new round of acttab_action() calls.
479**
480** Return the offset into the action table of the new transaction.
481*/
482int acttab_insert(acttab *p){
483 int i, j, k, n;
484 assert( p->nLookahead>0 );
485
486 /* Make sure we have enough space to hold the expanded action table
487 ** in the worst case. The worst case occurs if the transaction set
488 ** must be appended to the current action table
489 */
drh784d86f2004-02-19 18:41:53 +0000490 n = p->mxLookahead + 1;
drh8b582012003-10-21 13:16:03 +0000491 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000492 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000493 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
494 p->aAction = realloc( p->aAction,
495 sizeof(p->aAction[0])*p->nActionAlloc);
496 if( p->aAction==0 ){
497 fprintf(stderr,"malloc failed\n");
498 exit(1);
499 }
drhfdbf9282003-10-21 16:34:41 +0000500 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000501 p->aAction[i].lookahead = -1;
502 p->aAction[i].action = -1;
503 }
504 }
505
506 /* Scan the existing action table looking for an offset where we can
507 ** insert the current transaction set. Fall out of the loop when that
508 ** offset is found. In the worst case, we fall out of the loop when
509 ** i reaches p->nAction, which means we append the new transaction set.
510 **
511 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
512 */
drh784d86f2004-02-19 18:41:53 +0000513 for(i=0; i<p->nAction+p->mnLookahead; i++){
drh8b582012003-10-21 13:16:03 +0000514 if( p->aAction[i].lookahead<0 ){
515 for(j=0; j<p->nLookahead; j++){
516 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
517 if( k<0 ) break;
518 if( p->aAction[k].lookahead>=0 ) break;
519 }
drhfdbf9282003-10-21 16:34:41 +0000520 if( j<p->nLookahead ) continue;
521 for(j=0; j<p->nAction; j++){
522 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
523 }
524 if( j==p->nAction ){
525 break; /* Fits in empty slots */
526 }
drh8b582012003-10-21 13:16:03 +0000527 }else if( p->aAction[i].lookahead==p->mnLookahead ){
528 if( p->aAction[i].action!=p->mnAction ) continue;
529 for(j=0; j<p->nLookahead; j++){
530 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
531 if( k<0 || k>=p->nAction ) break;
532 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
533 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
534 }
535 if( j<p->nLookahead ) continue;
536 n = 0;
537 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000538 if( p->aAction[j].lookahead<0 ) continue;
539 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000540 }
drhfdbf9282003-10-21 16:34:41 +0000541 if( n==p->nLookahead ){
542 break; /* Same as a prior transaction set */
543 }
drh8b582012003-10-21 13:16:03 +0000544 }
545 }
546 /* Insert transaction set at index i. */
547 for(j=0; j<p->nLookahead; j++){
548 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
549 p->aAction[k] = p->aLookahead[j];
550 if( k>=p->nAction ) p->nAction = k+1;
551 }
552 p->nLookahead = 0;
553
554 /* Return the offset that is added to the lookahead in order to get the
555 ** index into yy_action of the action */
556 return i - p->mnLookahead;
557}
558
drh75897232000-05-29 14:26:00 +0000559/********************** From the file "assert.c" ****************************/
560/*
561** A more efficient way of handling assertions.
562*/
563void myassert(file,line)
564char *file;
565int line;
566{
567 fprintf(stderr,"Assertion failed on line %d of file \"%s\"\n",line,file);
568 exit(1);
569}
570/********************** From the file "build.c" *****************************/
571/*
572** Routines to construction the finite state machine for the LEMON
573** parser generator.
574*/
575
576/* Find a precedence symbol of every rule in the grammar.
577**
578** Those rules which have a precedence symbol coded in the input
579** grammar using the "[symbol]" construct will already have the
580** rp->precsym field filled. Other rules take as their precedence
581** symbol the first RHS symbol with a defined precedence. If there
582** are not RHS symbols with a defined precedence, the precedence
583** symbol field is left blank.
584*/
585void FindRulePrecedences(xp)
586struct lemon *xp;
587{
588 struct rule *rp;
589 for(rp=xp->rule; rp; rp=rp->next){
590 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000591 int i, j;
592 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
593 struct symbol *sp = rp->rhs[i];
594 if( sp->type==MULTITERMINAL ){
595 for(j=0; j<sp->nsubsym; j++){
596 if( sp->subsym[j]->prec>=0 ){
597 rp->precsym = sp->subsym[j];
598 break;
599 }
600 }
601 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000602 rp->precsym = rp->rhs[i];
drh75897232000-05-29 14:26:00 +0000603 }
604 }
605 }
606 }
607 return;
608}
609
610/* Find all nonterminals which will generate the empty string.
611** Then go back and compute the first sets of every nonterminal.
612** The first set is the set of all terminal symbols which can begin
613** a string generated by that nonterminal.
614*/
615void FindFirstSets(lemp)
616struct lemon *lemp;
617{
drhfd405312005-11-06 04:06:59 +0000618 int i, j;
drh75897232000-05-29 14:26:00 +0000619 struct rule *rp;
620 int progress;
621
622 for(i=0; i<lemp->nsymbol; i++){
drhb27b83a2002-08-14 23:18:57 +0000623 lemp->symbols[i]->lambda = B_FALSE;
drh75897232000-05-29 14:26:00 +0000624 }
625 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
626 lemp->symbols[i]->firstset = SetNew();
627 }
628
629 /* First compute all lambdas */
630 do{
631 progress = 0;
632 for(rp=lemp->rule; rp; rp=rp->next){
633 if( rp->lhs->lambda ) continue;
634 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000635 struct symbol *sp = rp->rhs[i];
636 if( sp->type!=TERMINAL || sp->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000637 }
638 if( i==rp->nrhs ){
drhb27b83a2002-08-14 23:18:57 +0000639 rp->lhs->lambda = B_TRUE;
drh75897232000-05-29 14:26:00 +0000640 progress = 1;
641 }
642 }
643 }while( progress );
644
645 /* Now compute all first sets */
646 do{
647 struct symbol *s1, *s2;
648 progress = 0;
649 for(rp=lemp->rule; rp; rp=rp->next){
650 s1 = rp->lhs;
651 for(i=0; i<rp->nrhs; i++){
652 s2 = rp->rhs[i];
653 if( s2->type==TERMINAL ){
654 progress += SetAdd(s1->firstset,s2->index);
655 break;
drhfd405312005-11-06 04:06:59 +0000656 }else if( s2->type==MULTITERMINAL ){
657 for(j=0; j<s2->nsubsym; j++){
658 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
659 }
660 break;
drh75897232000-05-29 14:26:00 +0000661 }else if( s1==s2 ){
drhb27b83a2002-08-14 23:18:57 +0000662 if( s1->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000663 }else{
664 progress += SetUnion(s1->firstset,s2->firstset);
drhb27b83a2002-08-14 23:18:57 +0000665 if( s2->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000666 }
667 }
668 }
669 }while( progress );
670 return;
671}
672
673/* Compute all LR(0) states for the grammar. Links
674** are added to between some states so that the LR(1) follow sets
675** can be computed later.
676*/
677PRIVATE struct state *getstate(/* struct lemon * */); /* forward reference */
678void FindStates(lemp)
679struct lemon *lemp;
680{
681 struct symbol *sp;
682 struct rule *rp;
683
684 Configlist_init();
685
686 /* Find the start symbol */
687 if( lemp->start ){
688 sp = Symbol_find(lemp->start);
689 if( sp==0 ){
690 ErrorMsg(lemp->filename,0,
691"The specified start symbol \"%s\" is not \
692in a nonterminal of the grammar. \"%s\" will be used as the start \
693symbol instead.",lemp->start,lemp->rule->lhs->name);
694 lemp->errorcnt++;
695 sp = lemp->rule->lhs;
696 }
697 }else{
698 sp = lemp->rule->lhs;
699 }
700
701 /* Make sure the start symbol doesn't occur on the right-hand side of
702 ** any rule. Report an error if it does. (YACC would generate a new
703 ** start symbol in this case.) */
704 for(rp=lemp->rule; rp; rp=rp->next){
705 int i;
706 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000707 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000708 ErrorMsg(lemp->filename,0,
709"The start symbol \"%s\" occurs on the \
710right-hand side of a rule. This will result in a parser which \
711does not work properly.",sp->name);
712 lemp->errorcnt++;
713 }
714 }
715 }
716
717 /* The basis configuration set for the first state
718 ** is all rules which have the start symbol as their
719 ** left-hand side */
720 for(rp=sp->rule; rp; rp=rp->nextlhs){
721 struct config *newcfp;
722 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];
973 assert( stp->ap );
974 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. */
drhb27b83a2002-08-14 23:18:57 +0000985 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = B_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){
drhb27b83a2002-08-14 23:18:57 +0000989 if( ap->type==REDUCE ) ap->x.rp->canReduce = B_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
1000** conflict can't be resolve, return non-zero.
1001**
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 ){
1021 apy->type = CONFLICT;
1022 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. */
1029 apy->type = CONFLICT;
1030 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 );
1041 apy->type = CONFLICT;
1042 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 ){
1049 apy->type = CONFLICT;
1050 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 ||
1060 apx->type==CONFLICT ||
1061 apy->type==SH_RESOLVED ||
1062 apy->type==RD_RESOLVED ||
1063 apy->type==CONFLICT
1064 );
1065 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1066 ** REDUCEs on the list. If we reach this point it must be because
1067 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001068 }
1069 return errcnt;
1070}
1071/********************* From the file "configlist.c" *************************/
1072/*
1073** Routines to processing a configuration list and building a state
1074** in the LEMON parser generator.
1075*/
1076
1077static struct config *freelist = 0; /* List of free configurations */
1078static struct config *current = 0; /* Top of list of configurations */
1079static struct config **currentend = 0; /* Last on list of configs */
1080static struct config *basis = 0; /* Top of list of basis configs */
1081static struct config **basisend = 0; /* End of list of basis configs */
1082
1083/* Return a pointer to a new configuration */
1084PRIVATE struct config *newconfig(){
1085 struct config *new;
1086 if( freelist==0 ){
1087 int i;
1088 int amt = 3;
1089 freelist = (struct config *)malloc( sizeof(struct config)*amt );
1090 if( freelist==0 ){
1091 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1092 exit(1);
1093 }
1094 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1095 freelist[amt-1].next = 0;
1096 }
1097 new = freelist;
1098 freelist = freelist->next;
1099 return new;
1100}
1101
1102/* The configuration "old" is no longer used */
1103PRIVATE void deleteconfig(old)
1104struct config *old;
1105{
1106 old->next = freelist;
1107 freelist = old;
1108}
1109
1110/* Initialized the configuration list builder */
1111void Configlist_init(){
1112 current = 0;
1113 currentend = &current;
1114 basis = 0;
1115 basisend = &basis;
1116 Configtable_init();
1117 return;
1118}
1119
1120/* Initialized the configuration list builder */
1121void Configlist_reset(){
1122 current = 0;
1123 currentend = &current;
1124 basis = 0;
1125 basisend = &basis;
1126 Configtable_clear(0);
1127 return;
1128}
1129
1130/* Add another configuration to the configuration list */
1131struct config *Configlist_add(rp,dot)
1132struct rule *rp; /* The rule */
1133int dot; /* Index into the RHS of the rule where the dot goes */
1134{
1135 struct config *cfp, model;
1136
1137 assert( currentend!=0 );
1138 model.rp = rp;
1139 model.dot = dot;
1140 cfp = Configtable_find(&model);
1141 if( cfp==0 ){
1142 cfp = newconfig();
1143 cfp->rp = rp;
1144 cfp->dot = dot;
1145 cfp->fws = SetNew();
1146 cfp->stp = 0;
1147 cfp->fplp = cfp->bplp = 0;
1148 cfp->next = 0;
1149 cfp->bp = 0;
1150 *currentend = cfp;
1151 currentend = &cfp->next;
1152 Configtable_insert(cfp);
1153 }
1154 return cfp;
1155}
1156
1157/* Add a basis configuration to the configuration list */
1158struct config *Configlist_addbasis(rp,dot)
1159struct rule *rp;
1160int dot;
1161{
1162 struct config *cfp, model;
1163
1164 assert( basisend!=0 );
1165 assert( currentend!=0 );
1166 model.rp = rp;
1167 model.dot = dot;
1168 cfp = Configtable_find(&model);
1169 if( cfp==0 ){
1170 cfp = newconfig();
1171 cfp->rp = rp;
1172 cfp->dot = dot;
1173 cfp->fws = SetNew();
1174 cfp->stp = 0;
1175 cfp->fplp = cfp->bplp = 0;
1176 cfp->next = 0;
1177 cfp->bp = 0;
1178 *currentend = cfp;
1179 currentend = &cfp->next;
1180 *basisend = cfp;
1181 basisend = &cfp->bp;
1182 Configtable_insert(cfp);
1183 }
1184 return cfp;
1185}
1186
1187/* Compute the closure of the configuration list */
1188void Configlist_closure(lemp)
1189struct lemon *lemp;
1190{
1191 struct config *cfp, *newcfp;
1192 struct rule *rp, *newrp;
1193 struct symbol *sp, *xsp;
1194 int i, dot;
1195
1196 assert( currentend!=0 );
1197 for(cfp=current; cfp; cfp=cfp->next){
1198 rp = cfp->rp;
1199 dot = cfp->dot;
1200 if( dot>=rp->nrhs ) continue;
1201 sp = rp->rhs[dot];
1202 if( sp->type==NONTERMINAL ){
1203 if( sp->rule==0 && sp!=lemp->errsym ){
1204 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1205 sp->name);
1206 lemp->errorcnt++;
1207 }
1208 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1209 newcfp = Configlist_add(newrp,0);
1210 for(i=dot+1; i<rp->nrhs; i++){
1211 xsp = rp->rhs[i];
1212 if( xsp->type==TERMINAL ){
1213 SetAdd(newcfp->fws,xsp->index);
1214 break;
drhfd405312005-11-06 04:06:59 +00001215 }else if( xsp->type==MULTITERMINAL ){
1216 int k;
1217 for(k=0; k<xsp->nsubsym; k++){
1218 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1219 }
1220 break;
drh75897232000-05-29 14:26:00 +00001221 }else{
1222 SetUnion(newcfp->fws,xsp->firstset);
drhb27b83a2002-08-14 23:18:57 +00001223 if( xsp->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +00001224 }
1225 }
1226 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1227 }
1228 }
1229 }
1230 return;
1231}
1232
1233/* Sort the configuration list */
1234void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001235 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001236 currentend = 0;
1237 return;
1238}
1239
1240/* Sort the basis configuration list */
1241void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001242 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001243 basisend = 0;
1244 return;
1245}
1246
1247/* Return a pointer to the head of the configuration list and
1248** reset the list */
1249struct config *Configlist_return(){
1250 struct config *old;
1251 old = current;
1252 current = 0;
1253 currentend = 0;
1254 return old;
1255}
1256
1257/* Return a pointer to the head of the configuration list and
1258** reset the list */
1259struct config *Configlist_basis(){
1260 struct config *old;
1261 old = basis;
1262 basis = 0;
1263 basisend = 0;
1264 return old;
1265}
1266
1267/* Free all elements of the given configuration list */
1268void Configlist_eat(cfp)
1269struct config *cfp;
1270{
1271 struct config *nextcfp;
1272 for(; cfp; cfp=nextcfp){
1273 nextcfp = cfp->next;
1274 assert( cfp->fplp==0 );
1275 assert( cfp->bplp==0 );
1276 if( cfp->fws ) SetFree(cfp->fws);
1277 deleteconfig(cfp);
1278 }
1279 return;
1280}
1281/***************** From the file "error.c" *********************************/
1282/*
1283** Code for printing error message.
1284*/
1285
1286/* Find a good place to break "msg" so that its length is at least "min"
1287** but no more than "max". Make the point as close to max as possible.
1288*/
1289static int findbreak(msg,min,max)
1290char *msg;
1291int min;
1292int max;
1293{
1294 int i,spot;
1295 char c;
1296 for(i=spot=min; i<=max; i++){
1297 c = msg[i];
1298 if( c=='\t' ) msg[i] = ' ';
1299 if( c=='\n' ){ msg[i] = ' '; spot = i; break; }
1300 if( c==0 ){ spot = i; break; }
1301 if( c=='-' && i<max-1 ) spot = i+1;
1302 if( c==' ' ) spot = i;
1303 }
1304 return spot;
1305}
1306
1307/*
1308** The error message is split across multiple lines if necessary. The
1309** splits occur at a space, if there is a space available near the end
1310** of the line.
1311*/
1312#define ERRMSGSIZE 10000 /* Hope this is big enough. No way to error check */
1313#define LINEWIDTH 79 /* Max width of any output line */
1314#define PREFIXLIMIT 30 /* Max width of the prefix on each line */
drhf9a2e7b2003-04-15 01:49:48 +00001315void ErrorMsg(const char *filename, int lineno, const char *format, ...){
drh75897232000-05-29 14:26:00 +00001316 char errmsg[ERRMSGSIZE];
1317 char prefix[PREFIXLIMIT+10];
1318 int errmsgsize;
1319 int prefixsize;
1320 int availablewidth;
1321 va_list ap;
1322 int end, restart, base;
1323
drhf9a2e7b2003-04-15 01:49:48 +00001324 va_start(ap, format);
drh75897232000-05-29 14:26:00 +00001325 /* Prepare a prefix to be prepended to every output line */
1326 if( lineno>0 ){
1327 sprintf(prefix,"%.*s:%d: ",PREFIXLIMIT-10,filename,lineno);
1328 }else{
1329 sprintf(prefix,"%.*s: ",PREFIXLIMIT-10,filename);
1330 }
1331 prefixsize = strlen(prefix);
1332 availablewidth = LINEWIDTH - prefixsize;
1333
1334 /* Generate the error message */
1335 vsprintf(errmsg,format,ap);
1336 va_end(ap);
1337 errmsgsize = strlen(errmsg);
1338 /* Remove trailing '\n's from the error message. */
1339 while( errmsgsize>0 && errmsg[errmsgsize-1]=='\n' ){
1340 errmsg[--errmsgsize] = 0;
1341 }
1342
1343 /* Print the error message */
1344 base = 0;
1345 while( errmsg[base]!=0 ){
1346 end = restart = findbreak(&errmsg[base],0,availablewidth);
1347 restart += base;
1348 while( errmsg[restart]==' ' ) restart++;
1349 fprintf(stdout,"%s%.*s\n",prefix,end,&errmsg[base]);
1350 base = restart;
1351 }
1352}
1353/**************** From the file "main.c" ************************************/
1354/*
1355** Main program file for the LEMON parser generator.
1356*/
1357
1358/* Report an out-of-memory condition and abort. This function
1359** is used mostly by the "MemoryCheck" macro in struct.h
1360*/
1361void memory_error(){
1362 fprintf(stderr,"Out of memory. Aborting...\n");
1363 exit(1);
1364}
1365
drh6d08b4d2004-07-20 12:45:22 +00001366static int nDefine = 0; /* Number of -D options on the command line */
1367static char **azDefine = 0; /* Name of the -D macros */
1368
1369/* This routine is called with the argument to each -D command-line option.
1370** Add the macro defined to the azDefine array.
1371*/
1372static void handle_D_option(char *z){
1373 char **paz;
1374 nDefine++;
1375 azDefine = realloc(azDefine, sizeof(azDefine[0])*nDefine);
1376 if( azDefine==0 ){
1377 fprintf(stderr,"out of memory\n");
1378 exit(1);
1379 }
1380 paz = &azDefine[nDefine-1];
1381 *paz = malloc( strlen(z)+1 );
1382 if( *paz==0 ){
1383 fprintf(stderr,"out of memory\n");
1384 exit(1);
1385 }
1386 strcpy(*paz, z);
1387 for(z=*paz; *z && *z!='='; z++){}
1388 *z = 0;
1389}
1390
drh75897232000-05-29 14:26:00 +00001391
1392/* The main program. Parse the command line and do it... */
1393int main(argc,argv)
1394int argc;
1395char **argv;
1396{
1397 static int version = 0;
1398 static int rpflag = 0;
1399 static int basisflag = 0;
1400 static int compress = 0;
1401 static int quiet = 0;
1402 static int statistics = 0;
1403 static int mhflag = 0;
1404 static struct s_options options[] = {
1405 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1406 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001407 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh75897232000-05-29 14:26:00 +00001408 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
1409 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file"},
1410 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drh6d08b4d2004-07-20 12:45:22 +00001411 {OPT_FLAG, "s", (char*)&statistics,
1412 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001413 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1414 {OPT_FLAG,0,0,0}
1415 };
1416 int i;
1417 struct lemon lem;
1418
drhb0c86772000-06-02 23:21:26 +00001419 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001420 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001421 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001422 exit(0);
1423 }
drhb0c86772000-06-02 23:21:26 +00001424 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001425 fprintf(stderr,"Exactly one filename argument is required.\n");
1426 exit(1);
1427 }
drh954f6b42006-06-13 13:27:46 +00001428 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001429 lem.errorcnt = 0;
1430
1431 /* Initialize the machine */
1432 Strsafe_init();
1433 Symbol_init();
1434 State_init();
1435 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001436 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001437 lem.basisflag = basisflag;
drh75897232000-05-29 14:26:00 +00001438 Symbol_new("$");
1439 lem.errsym = Symbol_new("error");
1440
1441 /* Parse the input file */
1442 Parse(&lem);
1443 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001444 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001445 fprintf(stderr,"Empty grammar.\n");
1446 exit(1);
1447 }
1448
1449 /* Count and index the symbols of the grammar */
1450 lem.nsymbol = Symbol_count();
1451 Symbol_new("{default}");
1452 lem.symbols = Symbol_arrayof();
drh60d31652004-02-22 00:08:04 +00001453 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
drh75897232000-05-29 14:26:00 +00001454 qsort(lem.symbols,lem.nsymbol+1,sizeof(struct symbol*),
1455 (int(*)())Symbolcmpp);
1456 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
1457 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1458 lem.nterminal = i;
1459
1460 /* Generate a reprint of the grammar, if requested on the command line */
1461 if( rpflag ){
1462 Reprint(&lem);
1463 }else{
1464 /* Initialize the size for all follow and first sets */
1465 SetSize(lem.nterminal);
1466
1467 /* Find the precedence for every production rule (that has one) */
1468 FindRulePrecedences(&lem);
1469
1470 /* Compute the lambda-nonterminals and the first-sets for every
1471 ** nonterminal */
1472 FindFirstSets(&lem);
1473
1474 /* Compute all LR(0) states. Also record follow-set propagation
1475 ** links so that the follow-set can be computed later */
1476 lem.nstate = 0;
1477 FindStates(&lem);
1478 lem.sorted = State_arrayof();
1479
1480 /* Tie up loose ends on the propagation links */
1481 FindLinks(&lem);
1482
1483 /* Compute the follow set of every reducible configuration */
1484 FindFollowSets(&lem);
1485
1486 /* Compute the action tables */
1487 FindActions(&lem);
1488
1489 /* Compress the action tables */
1490 if( compress==0 ) CompressTables(&lem);
1491
drhada354d2005-11-05 15:03:59 +00001492 /* Reorder and renumber the states so that states with fewer choices
1493 ** occur at the end. */
1494 ResortStates(&lem);
1495
drh75897232000-05-29 14:26:00 +00001496 /* Generate a report of the parser generated. (the "y.output" file) */
1497 if( !quiet ) ReportOutput(&lem);
1498
1499 /* Generate the source code for the parser */
1500 ReportTable(&lem, mhflag);
1501
1502 /* Produce a header file for use by the scanner. (This step is
1503 ** omitted if the "-m" option is used because makeheaders will
1504 ** generate the file for us.) */
1505 if( !mhflag ) ReportHeader(&lem);
1506 }
1507 if( statistics ){
1508 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1509 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1510 printf(" %d states, %d parser table entries, %d conflicts\n",
1511 lem.nstate, lem.tablesize, lem.nconflict);
1512 }
1513 if( lem.nconflict ){
1514 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
1515 }
1516 exit(lem.errorcnt + lem.nconflict);
drh218dc692004-05-31 23:13:45 +00001517 return (lem.errorcnt + lem.nconflict);
drh75897232000-05-29 14:26:00 +00001518}
1519/******************** From the file "msort.c" *******************************/
1520/*
1521** A generic merge-sort program.
1522**
1523** USAGE:
1524** Let "ptr" be a pointer to some structure which is at the head of
1525** a null-terminated list. Then to sort the list call:
1526**
1527** ptr = msort(ptr,&(ptr->next),cmpfnc);
1528**
1529** In the above, "cmpfnc" is a pointer to a function which compares
1530** two instances of the structure and returns an integer, as in
1531** strcmp. The second argument is a pointer to the pointer to the
1532** second element of the linked list. This address is used to compute
1533** the offset to the "next" field within the structure. The offset to
1534** the "next" field must be constant for all structures in the list.
1535**
1536** The function returns a new pointer which is the head of the list
1537** after sorting.
1538**
1539** ALGORITHM:
1540** Merge-sort.
1541*/
1542
1543/*
1544** Return a pointer to the next structure in the linked list.
1545*/
drhba99af52001-10-25 20:37:16 +00001546#define NEXT(A) (*(char**)(((unsigned long)A)+offset))
drh75897232000-05-29 14:26:00 +00001547
1548/*
1549** Inputs:
1550** a: A sorted, null-terminated linked list. (May be null).
1551** b: A sorted, null-terminated linked list. (May be null).
1552** cmp: A pointer to the comparison function.
1553** offset: Offset in the structure to the "next" field.
1554**
1555** Return Value:
1556** A pointer to the head of a sorted list containing the elements
1557** of both a and b.
1558**
1559** Side effects:
1560** The "next" pointers for elements in the lists a and b are
1561** changed.
1562*/
1563static char *merge(a,b,cmp,offset)
1564char *a;
1565char *b;
1566int (*cmp)();
1567int offset;
1568{
1569 char *ptr, *head;
1570
1571 if( a==0 ){
1572 head = b;
1573 }else if( b==0 ){
1574 head = a;
1575 }else{
1576 if( (*cmp)(a,b)<0 ){
1577 ptr = a;
1578 a = NEXT(a);
1579 }else{
1580 ptr = b;
1581 b = NEXT(b);
1582 }
1583 head = ptr;
1584 while( a && b ){
1585 if( (*cmp)(a,b)<0 ){
1586 NEXT(ptr) = a;
1587 ptr = a;
1588 a = NEXT(a);
1589 }else{
1590 NEXT(ptr) = b;
1591 ptr = b;
1592 b = NEXT(b);
1593 }
1594 }
1595 if( a ) NEXT(ptr) = a;
1596 else NEXT(ptr) = b;
1597 }
1598 return head;
1599}
1600
1601/*
1602** Inputs:
1603** list: Pointer to a singly-linked list of structures.
1604** next: Pointer to pointer to the second element of the list.
1605** cmp: A comparison function.
1606**
1607** Return Value:
1608** A pointer to the head of a sorted list containing the elements
1609** orginally in list.
1610**
1611** Side effects:
1612** The "next" pointers for elements in list are changed.
1613*/
1614#define LISTSIZE 30
1615char *msort(list,next,cmp)
1616char *list;
1617char **next;
1618int (*cmp)();
1619{
drhba99af52001-10-25 20:37:16 +00001620 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001621 char *ep;
1622 char *set[LISTSIZE];
1623 int i;
drhba99af52001-10-25 20:37:16 +00001624 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001625 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1626 while( list ){
1627 ep = list;
1628 list = NEXT(list);
1629 NEXT(ep) = 0;
1630 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1631 ep = merge(ep,set[i],cmp,offset);
1632 set[i] = 0;
1633 }
1634 set[i] = ep;
1635 }
1636 ep = 0;
1637 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(ep,set[i],cmp,offset);
1638 return ep;
1639}
1640/************************ From the file "option.c" **************************/
1641static char **argv;
1642static struct s_options *op;
1643static FILE *errstream;
1644
1645#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1646
1647/*
1648** Print the command line with a carrot pointing to the k-th character
1649** of the n-th field.
1650*/
1651static void errline(n,k,err)
1652int n;
1653int k;
1654FILE *err;
1655{
1656 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001657 if( argv[0] ) fprintf(err,"%s",argv[0]);
1658 spcnt = strlen(argv[0]) + 1;
1659 for(i=1; i<n && argv[i]; i++){
1660 fprintf(err," %s",argv[i]);
drhdc30dd32005-02-16 03:35:15 +00001661 spcnt += strlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001662 }
1663 spcnt += k;
1664 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1665 if( spcnt<20 ){
1666 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1667 }else{
1668 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1669 }
1670}
1671
1672/*
1673** Return the index of the N-th non-switch argument. Return -1
1674** if N is out of range.
1675*/
1676static int argindex(n)
1677int n;
1678{
1679 int i;
1680 int dashdash = 0;
1681 if( argv!=0 && *argv!=0 ){
1682 for(i=1; argv[i]; i++){
1683 if( dashdash || !ISOPT(argv[i]) ){
1684 if( n==0 ) return i;
1685 n--;
1686 }
1687 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1688 }
1689 }
1690 return -1;
1691}
1692
1693static char emsg[] = "Command line syntax error: ";
1694
1695/*
1696** Process a flag command line argument.
1697*/
1698static int handleflags(i,err)
1699int i;
1700FILE *err;
1701{
1702 int v;
1703 int errcnt = 0;
1704 int j;
1705 for(j=0; op[j].label; j++){
drh6d08b4d2004-07-20 12:45:22 +00001706 if( strncmp(&argv[i][1],op[j].label,strlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001707 }
1708 v = argv[i][0]=='-' ? 1 : 0;
1709 if( op[j].label==0 ){
1710 if( err ){
1711 fprintf(err,"%sundefined option.\n",emsg);
1712 errline(i,1,err);
1713 }
1714 errcnt++;
1715 }else if( op[j].type==OPT_FLAG ){
1716 *((int*)op[j].arg) = v;
1717 }else if( op[j].type==OPT_FFLAG ){
1718 (*(void(*)())(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001719 }else if( op[j].type==OPT_FSTR ){
1720 (*(void(*)())(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001721 }else{
1722 if( err ){
1723 fprintf(err,"%smissing argument on switch.\n",emsg);
1724 errline(i,1,err);
1725 }
1726 errcnt++;
1727 }
1728 return errcnt;
1729}
1730
1731/*
1732** Process a command line switch which has an argument.
1733*/
1734static int handleswitch(i,err)
1735int i;
1736FILE *err;
1737{
1738 int lv = 0;
1739 double dv = 0.0;
1740 char *sv = 0, *end;
1741 char *cp;
1742 int j;
1743 int errcnt = 0;
1744 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001745 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001746 *cp = 0;
1747 for(j=0; op[j].label; j++){
1748 if( strcmp(argv[i],op[j].label)==0 ) break;
1749 }
1750 *cp = '=';
1751 if( op[j].label==0 ){
1752 if( err ){
1753 fprintf(err,"%sundefined option.\n",emsg);
1754 errline(i,0,err);
1755 }
1756 errcnt++;
1757 }else{
1758 cp++;
1759 switch( op[j].type ){
1760 case OPT_FLAG:
1761 case OPT_FFLAG:
1762 if( err ){
1763 fprintf(err,"%soption requires an argument.\n",emsg);
1764 errline(i,0,err);
1765 }
1766 errcnt++;
1767 break;
1768 case OPT_DBL:
1769 case OPT_FDBL:
1770 dv = strtod(cp,&end);
1771 if( *end ){
1772 if( err ){
1773 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001774 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001775 }
1776 errcnt++;
1777 }
1778 break;
1779 case OPT_INT:
1780 case OPT_FINT:
1781 lv = strtol(cp,&end,0);
1782 if( *end ){
1783 if( err ){
1784 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001785 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001786 }
1787 errcnt++;
1788 }
1789 break;
1790 case OPT_STR:
1791 case OPT_FSTR:
1792 sv = cp;
1793 break;
1794 }
1795 switch( op[j].type ){
1796 case OPT_FLAG:
1797 case OPT_FFLAG:
1798 break;
1799 case OPT_DBL:
1800 *(double*)(op[j].arg) = dv;
1801 break;
1802 case OPT_FDBL:
1803 (*(void(*)())(op[j].arg))(dv);
1804 break;
1805 case OPT_INT:
1806 *(int*)(op[j].arg) = lv;
1807 break;
1808 case OPT_FINT:
1809 (*(void(*)())(op[j].arg))((int)lv);
1810 break;
1811 case OPT_STR:
1812 *(char**)(op[j].arg) = sv;
1813 break;
1814 case OPT_FSTR:
1815 (*(void(*)())(op[j].arg))(sv);
1816 break;
1817 }
1818 }
1819 return errcnt;
1820}
1821
drhb0c86772000-06-02 23:21:26 +00001822int OptInit(a,o,err)
drh75897232000-05-29 14:26:00 +00001823char **a;
1824struct s_options *o;
1825FILE *err;
1826{
1827 int errcnt = 0;
1828 argv = a;
1829 op = o;
1830 errstream = err;
1831 if( argv && *argv && op ){
1832 int i;
1833 for(i=1; argv[i]; i++){
1834 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1835 errcnt += handleflags(i,err);
1836 }else if( strchr(argv[i],'=') ){
1837 errcnt += handleswitch(i,err);
1838 }
1839 }
1840 }
1841 if( errcnt>0 ){
1842 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001843 OptPrint();
drh75897232000-05-29 14:26:00 +00001844 exit(1);
1845 }
1846 return 0;
1847}
1848
drhb0c86772000-06-02 23:21:26 +00001849int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001850 int cnt = 0;
1851 int dashdash = 0;
1852 int i;
1853 if( argv!=0 && argv[0]!=0 ){
1854 for(i=1; argv[i]; i++){
1855 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1856 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1857 }
1858 }
1859 return cnt;
1860}
1861
drhb0c86772000-06-02 23:21:26 +00001862char *OptArg(n)
drh75897232000-05-29 14:26:00 +00001863int n;
1864{
1865 int i;
1866 i = argindex(n);
1867 return i>=0 ? argv[i] : 0;
1868}
1869
drhb0c86772000-06-02 23:21:26 +00001870void OptErr(n)
drh75897232000-05-29 14:26:00 +00001871int n;
1872{
1873 int i;
1874 i = argindex(n);
1875 if( i>=0 ) errline(i,0,errstream);
1876}
1877
drhb0c86772000-06-02 23:21:26 +00001878void OptPrint(){
drh75897232000-05-29 14:26:00 +00001879 int i;
1880 int max, len;
1881 max = 0;
1882 for(i=0; op[i].label; i++){
1883 len = strlen(op[i].label) + 1;
1884 switch( op[i].type ){
1885 case OPT_FLAG:
1886 case OPT_FFLAG:
1887 break;
1888 case OPT_INT:
1889 case OPT_FINT:
1890 len += 9; /* length of "<integer>" */
1891 break;
1892 case OPT_DBL:
1893 case OPT_FDBL:
1894 len += 6; /* length of "<real>" */
1895 break;
1896 case OPT_STR:
1897 case OPT_FSTR:
1898 len += 8; /* length of "<string>" */
1899 break;
1900 }
1901 if( len>max ) max = len;
1902 }
1903 for(i=0; op[i].label; i++){
1904 switch( op[i].type ){
1905 case OPT_FLAG:
1906 case OPT_FFLAG:
1907 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
1908 break;
1909 case OPT_INT:
1910 case OPT_FINT:
1911 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001912 (int)(max-strlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001913 break;
1914 case OPT_DBL:
1915 case OPT_FDBL:
1916 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001917 (int)(max-strlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001918 break;
1919 case OPT_STR:
1920 case OPT_FSTR:
1921 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001922 (int)(max-strlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001923 break;
1924 }
1925 }
1926}
1927/*********************** From the file "parse.c" ****************************/
1928/*
1929** Input file parser for the LEMON parser generator.
1930*/
1931
1932/* The state of the parser */
1933struct pstate {
1934 char *filename; /* Name of the input file */
1935 int tokenlineno; /* Linenumber at which current token starts */
1936 int errorcnt; /* Number of errors so far */
1937 char *tokenstart; /* Text of current token */
1938 struct lemon *gp; /* Global state vector */
1939 enum e_state {
1940 INITIALIZE,
1941 WAITING_FOR_DECL_OR_RULE,
1942 WAITING_FOR_DECL_KEYWORD,
1943 WAITING_FOR_DECL_ARG,
1944 WAITING_FOR_PRECEDENCE_SYMBOL,
1945 WAITING_FOR_ARROW,
1946 IN_RHS,
1947 LHS_ALIAS_1,
1948 LHS_ALIAS_2,
1949 LHS_ALIAS_3,
1950 RHS_ALIAS_1,
1951 RHS_ALIAS_2,
1952 PRECEDENCE_MARK_1,
1953 PRECEDENCE_MARK_2,
1954 RESYNC_AFTER_RULE_ERROR,
1955 RESYNC_AFTER_DECL_ERROR,
1956 WAITING_FOR_DESTRUCTOR_SYMBOL,
drh0bd1f4e2002-06-06 18:54:39 +00001957 WAITING_FOR_DATATYPE_SYMBOL,
drhe09daa92006-06-10 13:29:31 +00001958 WAITING_FOR_FALLBACK_ID,
1959 WAITING_FOR_WILDCARD_ID
drh75897232000-05-29 14:26:00 +00001960 } state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00001961 struct symbol *fallback; /* The fallback token */
drh75897232000-05-29 14:26:00 +00001962 struct symbol *lhs; /* Left-hand side of current rule */
1963 char *lhsalias; /* Alias for the LHS */
1964 int nrhs; /* Number of right-hand side symbols seen */
1965 struct symbol *rhs[MAXRHS]; /* RHS symbols */
1966 char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
1967 struct rule *prevrule; /* Previous rule parsed */
1968 char *declkeyword; /* Keyword of a declaration */
1969 char **declargslot; /* Where the declaration argument should be put */
1970 int *decllnslot; /* Where the declaration linenumber is put */
1971 enum e_assoc declassoc; /* Assign this association to decl arguments */
1972 int preccounter; /* Assign this precedence to decl arguments */
1973 struct rule *firstrule; /* Pointer to first rule in the grammar */
1974 struct rule *lastrule; /* Pointer to the most recently parsed rule */
1975};
1976
1977/* Parse a single token */
1978static void parseonetoken(psp)
1979struct pstate *psp;
1980{
1981 char *x;
1982 x = Strsafe(psp->tokenstart); /* Save the token permanently */
1983#if 0
1984 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
1985 x,psp->state);
1986#endif
1987 switch( psp->state ){
1988 case INITIALIZE:
1989 psp->prevrule = 0;
1990 psp->preccounter = 0;
1991 psp->firstrule = psp->lastrule = 0;
1992 psp->gp->nrule = 0;
1993 /* Fall thru to next case */
1994 case WAITING_FOR_DECL_OR_RULE:
1995 if( x[0]=='%' ){
1996 psp->state = WAITING_FOR_DECL_KEYWORD;
1997 }else if( islower(x[0]) ){
1998 psp->lhs = Symbol_new(x);
1999 psp->nrhs = 0;
2000 psp->lhsalias = 0;
2001 psp->state = WAITING_FOR_ARROW;
2002 }else if( x[0]=='{' ){
2003 if( psp->prevrule==0 ){
2004 ErrorMsg(psp->filename,psp->tokenlineno,
2005"There is not prior rule opon which to attach the code \
2006fragment which begins on this line.");
2007 psp->errorcnt++;
2008 }else if( psp->prevrule->code!=0 ){
2009 ErrorMsg(psp->filename,psp->tokenlineno,
2010"Code fragment beginning on this line is not the first \
2011to follow the previous rule.");
2012 psp->errorcnt++;
2013 }else{
2014 psp->prevrule->line = psp->tokenlineno;
2015 psp->prevrule->code = &x[1];
2016 }
2017 }else if( x[0]=='[' ){
2018 psp->state = PRECEDENCE_MARK_1;
2019 }else{
2020 ErrorMsg(psp->filename,psp->tokenlineno,
2021 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2022 x);
2023 psp->errorcnt++;
2024 }
2025 break;
2026 case PRECEDENCE_MARK_1:
2027 if( !isupper(x[0]) ){
2028 ErrorMsg(psp->filename,psp->tokenlineno,
2029 "The precedence symbol must be a terminal.");
2030 psp->errorcnt++;
2031 }else if( psp->prevrule==0 ){
2032 ErrorMsg(psp->filename,psp->tokenlineno,
2033 "There is no prior rule to assign precedence \"[%s]\".",x);
2034 psp->errorcnt++;
2035 }else if( psp->prevrule->precsym!=0 ){
2036 ErrorMsg(psp->filename,psp->tokenlineno,
2037"Precedence mark on this line is not the first \
2038to follow the previous rule.");
2039 psp->errorcnt++;
2040 }else{
2041 psp->prevrule->precsym = Symbol_new(x);
2042 }
2043 psp->state = PRECEDENCE_MARK_2;
2044 break;
2045 case PRECEDENCE_MARK_2:
2046 if( x[0]!=']' ){
2047 ErrorMsg(psp->filename,psp->tokenlineno,
2048 "Missing \"]\" on precedence mark.");
2049 psp->errorcnt++;
2050 }
2051 psp->state = WAITING_FOR_DECL_OR_RULE;
2052 break;
2053 case WAITING_FOR_ARROW:
2054 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2055 psp->state = IN_RHS;
2056 }else if( x[0]=='(' ){
2057 psp->state = LHS_ALIAS_1;
2058 }else{
2059 ErrorMsg(psp->filename,psp->tokenlineno,
2060 "Expected to see a \":\" following the LHS symbol \"%s\".",
2061 psp->lhs->name);
2062 psp->errorcnt++;
2063 psp->state = RESYNC_AFTER_RULE_ERROR;
2064 }
2065 break;
2066 case LHS_ALIAS_1:
2067 if( isalpha(x[0]) ){
2068 psp->lhsalias = x;
2069 psp->state = LHS_ALIAS_2;
2070 }else{
2071 ErrorMsg(psp->filename,psp->tokenlineno,
2072 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2073 x,psp->lhs->name);
2074 psp->errorcnt++;
2075 psp->state = RESYNC_AFTER_RULE_ERROR;
2076 }
2077 break;
2078 case LHS_ALIAS_2:
2079 if( x[0]==')' ){
2080 psp->state = LHS_ALIAS_3;
2081 }else{
2082 ErrorMsg(psp->filename,psp->tokenlineno,
2083 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2084 psp->errorcnt++;
2085 psp->state = RESYNC_AFTER_RULE_ERROR;
2086 }
2087 break;
2088 case LHS_ALIAS_3:
2089 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2090 psp->state = IN_RHS;
2091 }else{
2092 ErrorMsg(psp->filename,psp->tokenlineno,
2093 "Missing \"->\" following: \"%s(%s)\".",
2094 psp->lhs->name,psp->lhsalias);
2095 psp->errorcnt++;
2096 psp->state = RESYNC_AFTER_RULE_ERROR;
2097 }
2098 break;
2099 case IN_RHS:
2100 if( x[0]=='.' ){
2101 struct rule *rp;
2102 rp = (struct rule *)malloc( sizeof(struct rule) +
2103 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs );
2104 if( rp==0 ){
2105 ErrorMsg(psp->filename,psp->tokenlineno,
2106 "Can't allocate enough memory for this rule.");
2107 psp->errorcnt++;
2108 psp->prevrule = 0;
2109 }else{
2110 int i;
2111 rp->ruleline = psp->tokenlineno;
2112 rp->rhs = (struct symbol**)&rp[1];
2113 rp->rhsalias = (char**)&(rp->rhs[psp->nrhs]);
2114 for(i=0; i<psp->nrhs; i++){
2115 rp->rhs[i] = psp->rhs[i];
2116 rp->rhsalias[i] = psp->alias[i];
2117 }
2118 rp->lhs = psp->lhs;
2119 rp->lhsalias = psp->lhsalias;
2120 rp->nrhs = psp->nrhs;
2121 rp->code = 0;
2122 rp->precsym = 0;
2123 rp->index = psp->gp->nrule++;
2124 rp->nextlhs = rp->lhs->rule;
2125 rp->lhs->rule = rp;
2126 rp->next = 0;
2127 if( psp->firstrule==0 ){
2128 psp->firstrule = psp->lastrule = rp;
2129 }else{
2130 psp->lastrule->next = rp;
2131 psp->lastrule = rp;
2132 }
2133 psp->prevrule = rp;
2134 }
2135 psp->state = WAITING_FOR_DECL_OR_RULE;
2136 }else if( isalpha(x[0]) ){
2137 if( psp->nrhs>=MAXRHS ){
2138 ErrorMsg(psp->filename,psp->tokenlineno,
drhfd405312005-11-06 04:06:59 +00002139 "Too many symbols on RHS or rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002140 x);
2141 psp->errorcnt++;
2142 psp->state = RESYNC_AFTER_RULE_ERROR;
2143 }else{
2144 psp->rhs[psp->nrhs] = Symbol_new(x);
2145 psp->alias[psp->nrhs] = 0;
2146 psp->nrhs++;
2147 }
drhfd405312005-11-06 04:06:59 +00002148 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2149 struct symbol *msp = psp->rhs[psp->nrhs-1];
2150 if( msp->type!=MULTITERMINAL ){
2151 struct symbol *origsp = msp;
2152 msp = malloc(sizeof(*msp));
2153 memset(msp, 0, sizeof(*msp));
2154 msp->type = MULTITERMINAL;
2155 msp->nsubsym = 1;
2156 msp->subsym = malloc(sizeof(struct symbol*));
2157 msp->subsym[0] = origsp;
2158 msp->name = origsp->name;
2159 psp->rhs[psp->nrhs-1] = msp;
2160 }
2161 msp->nsubsym++;
2162 msp->subsym = realloc(msp->subsym, sizeof(struct symbol*)*msp->nsubsym);
2163 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2164 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2165 ErrorMsg(psp->filename,psp->tokenlineno,
2166 "Cannot form a compound containing a non-terminal");
2167 psp->errorcnt++;
2168 }
drh75897232000-05-29 14:26:00 +00002169 }else if( x[0]=='(' && psp->nrhs>0 ){
2170 psp->state = RHS_ALIAS_1;
2171 }else{
2172 ErrorMsg(psp->filename,psp->tokenlineno,
2173 "Illegal character on RHS of rule: \"%s\".",x);
2174 psp->errorcnt++;
2175 psp->state = RESYNC_AFTER_RULE_ERROR;
2176 }
2177 break;
2178 case RHS_ALIAS_1:
2179 if( isalpha(x[0]) ){
2180 psp->alias[psp->nrhs-1] = x;
2181 psp->state = RHS_ALIAS_2;
2182 }else{
2183 ErrorMsg(psp->filename,psp->tokenlineno,
2184 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2185 x,psp->rhs[psp->nrhs-1]->name);
2186 psp->errorcnt++;
2187 psp->state = RESYNC_AFTER_RULE_ERROR;
2188 }
2189 break;
2190 case RHS_ALIAS_2:
2191 if( x[0]==')' ){
2192 psp->state = IN_RHS;
2193 }else{
2194 ErrorMsg(psp->filename,psp->tokenlineno,
2195 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2196 psp->errorcnt++;
2197 psp->state = RESYNC_AFTER_RULE_ERROR;
2198 }
2199 break;
2200 case WAITING_FOR_DECL_KEYWORD:
2201 if( isalpha(x[0]) ){
2202 psp->declkeyword = x;
2203 psp->declargslot = 0;
2204 psp->decllnslot = 0;
2205 psp->state = WAITING_FOR_DECL_ARG;
2206 if( strcmp(x,"name")==0 ){
2207 psp->declargslot = &(psp->gp->name);
2208 }else if( strcmp(x,"include")==0 ){
2209 psp->declargslot = &(psp->gp->include);
2210 psp->decllnslot = &psp->gp->includeln;
2211 }else if( strcmp(x,"code")==0 ){
2212 psp->declargslot = &(psp->gp->extracode);
2213 psp->decllnslot = &psp->gp->extracodeln;
2214 }else if( strcmp(x,"token_destructor")==0 ){
2215 psp->declargslot = &psp->gp->tokendest;
2216 psp->decllnslot = &psp->gp->tokendestln;
drh960e8c62001-04-03 16:53:21 +00002217 }else if( strcmp(x,"default_destructor")==0 ){
2218 psp->declargslot = &psp->gp->vardest;
2219 psp->decllnslot = &psp->gp->vardestln;
drh75897232000-05-29 14:26:00 +00002220 }else if( strcmp(x,"token_prefix")==0 ){
2221 psp->declargslot = &psp->gp->tokenprefix;
2222 }else if( strcmp(x,"syntax_error")==0 ){
2223 psp->declargslot = &(psp->gp->error);
2224 psp->decllnslot = &psp->gp->errorln;
2225 }else if( strcmp(x,"parse_accept")==0 ){
2226 psp->declargslot = &(psp->gp->accept);
2227 psp->decllnslot = &psp->gp->acceptln;
2228 }else if( strcmp(x,"parse_failure")==0 ){
2229 psp->declargslot = &(psp->gp->failure);
2230 psp->decllnslot = &psp->gp->failureln;
2231 }else if( strcmp(x,"stack_overflow")==0 ){
2232 psp->declargslot = &(psp->gp->overflow);
2233 psp->decllnslot = &psp->gp->overflowln;
2234 }else if( strcmp(x,"extra_argument")==0 ){
2235 psp->declargslot = &(psp->gp->arg);
2236 }else if( strcmp(x,"token_type")==0 ){
2237 psp->declargslot = &(psp->gp->tokentype);
drh960e8c62001-04-03 16:53:21 +00002238 }else if( strcmp(x,"default_type")==0 ){
2239 psp->declargslot = &(psp->gp->vartype);
drh75897232000-05-29 14:26:00 +00002240 }else if( strcmp(x,"stack_size")==0 ){
2241 psp->declargslot = &(psp->gp->stacksize);
2242 }else if( strcmp(x,"start_symbol")==0 ){
2243 psp->declargslot = &(psp->gp->start);
2244 }else if( strcmp(x,"left")==0 ){
2245 psp->preccounter++;
2246 psp->declassoc = LEFT;
2247 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2248 }else if( strcmp(x,"right")==0 ){
2249 psp->preccounter++;
2250 psp->declassoc = RIGHT;
2251 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2252 }else if( strcmp(x,"nonassoc")==0 ){
2253 psp->preccounter++;
2254 psp->declassoc = NONE;
2255 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2256 }else if( strcmp(x,"destructor")==0 ){
2257 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
2258 }else if( strcmp(x,"type")==0 ){
2259 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002260 }else if( strcmp(x,"fallback")==0 ){
2261 psp->fallback = 0;
2262 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002263 }else if( strcmp(x,"wildcard")==0 ){
2264 psp->state = WAITING_FOR_WILDCARD_ID;
drh75897232000-05-29 14:26:00 +00002265 }else{
2266 ErrorMsg(psp->filename,psp->tokenlineno,
2267 "Unknown declaration keyword: \"%%%s\".",x);
2268 psp->errorcnt++;
2269 psp->state = RESYNC_AFTER_DECL_ERROR;
2270 }
2271 }else{
2272 ErrorMsg(psp->filename,psp->tokenlineno,
2273 "Illegal declaration keyword: \"%s\".",x);
2274 psp->errorcnt++;
2275 psp->state = RESYNC_AFTER_DECL_ERROR;
2276 }
2277 break;
2278 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2279 if( !isalpha(x[0]) ){
2280 ErrorMsg(psp->filename,psp->tokenlineno,
2281 "Symbol name missing after %destructor keyword");
2282 psp->errorcnt++;
2283 psp->state = RESYNC_AFTER_DECL_ERROR;
2284 }else{
2285 struct symbol *sp = Symbol_new(x);
2286 psp->declargslot = &sp->destructor;
2287 psp->decllnslot = &sp->destructorln;
2288 psp->state = WAITING_FOR_DECL_ARG;
2289 }
2290 break;
2291 case WAITING_FOR_DATATYPE_SYMBOL:
2292 if( !isalpha(x[0]) ){
2293 ErrorMsg(psp->filename,psp->tokenlineno,
2294 "Symbol name missing after %destructor keyword");
2295 psp->errorcnt++;
2296 psp->state = RESYNC_AFTER_DECL_ERROR;
2297 }else{
2298 struct symbol *sp = Symbol_new(x);
2299 psp->declargslot = &sp->datatype;
2300 psp->decllnslot = 0;
2301 psp->state = WAITING_FOR_DECL_ARG;
2302 }
2303 break;
2304 case WAITING_FOR_PRECEDENCE_SYMBOL:
2305 if( x[0]=='.' ){
2306 psp->state = WAITING_FOR_DECL_OR_RULE;
2307 }else if( isupper(x[0]) ){
2308 struct symbol *sp;
2309 sp = Symbol_new(x);
2310 if( sp->prec>=0 ){
2311 ErrorMsg(psp->filename,psp->tokenlineno,
2312 "Symbol \"%s\" has already be given a precedence.",x);
2313 psp->errorcnt++;
2314 }else{
2315 sp->prec = psp->preccounter;
2316 sp->assoc = psp->declassoc;
2317 }
2318 }else{
2319 ErrorMsg(psp->filename,psp->tokenlineno,
2320 "Can't assign a precedence to \"%s\".",x);
2321 psp->errorcnt++;
2322 }
2323 break;
2324 case WAITING_FOR_DECL_ARG:
2325 if( (x[0]=='{' || x[0]=='\"' || isalnum(x[0])) ){
2326 if( *(psp->declargslot)!=0 ){
2327 ErrorMsg(psp->filename,psp->tokenlineno,
2328 "The argument \"%s\" to declaration \"%%%s\" is not the first.",
2329 x[0]=='\"' ? &x[1] : x,psp->declkeyword);
2330 psp->errorcnt++;
2331 psp->state = RESYNC_AFTER_DECL_ERROR;
2332 }else{
2333 *(psp->declargslot) = (x[0]=='\"' || x[0]=='{') ? &x[1] : x;
2334 if( psp->decllnslot ) *psp->decllnslot = psp->tokenlineno;
2335 psp->state = WAITING_FOR_DECL_OR_RULE;
2336 }
2337 }else{
2338 ErrorMsg(psp->filename,psp->tokenlineno,
2339 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2340 psp->errorcnt++;
2341 psp->state = RESYNC_AFTER_DECL_ERROR;
2342 }
2343 break;
drh0bd1f4e2002-06-06 18:54:39 +00002344 case WAITING_FOR_FALLBACK_ID:
2345 if( x[0]=='.' ){
2346 psp->state = WAITING_FOR_DECL_OR_RULE;
2347 }else if( !isupper(x[0]) ){
2348 ErrorMsg(psp->filename, psp->tokenlineno,
2349 "%%fallback argument \"%s\" should be a token", x);
2350 psp->errorcnt++;
2351 }else{
2352 struct symbol *sp = Symbol_new(x);
2353 if( psp->fallback==0 ){
2354 psp->fallback = sp;
2355 }else if( sp->fallback ){
2356 ErrorMsg(psp->filename, psp->tokenlineno,
2357 "More than one fallback assigned to token %s", x);
2358 psp->errorcnt++;
2359 }else{
2360 sp->fallback = psp->fallback;
2361 psp->gp->has_fallback = 1;
2362 }
2363 }
2364 break;
drhe09daa92006-06-10 13:29:31 +00002365 case WAITING_FOR_WILDCARD_ID:
2366 if( x[0]=='.' ){
2367 psp->state = WAITING_FOR_DECL_OR_RULE;
2368 }else if( !isupper(x[0]) ){
2369 ErrorMsg(psp->filename, psp->tokenlineno,
2370 "%%wildcard argument \"%s\" should be a token", x);
2371 psp->errorcnt++;
2372 }else{
2373 struct symbol *sp = Symbol_new(x);
2374 if( psp->gp->wildcard==0 ){
2375 psp->gp->wildcard = sp;
2376 }else{
2377 ErrorMsg(psp->filename, psp->tokenlineno,
2378 "Extra wildcard to token: %s", x);
2379 psp->errorcnt++;
2380 }
2381 }
2382 break;
drh75897232000-05-29 14:26:00 +00002383 case RESYNC_AFTER_RULE_ERROR:
2384/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2385** break; */
2386 case RESYNC_AFTER_DECL_ERROR:
2387 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2388 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2389 break;
2390 }
2391}
2392
drh6d08b4d2004-07-20 12:45:22 +00002393/* Run the proprocessor over the input file text. The global variables
2394** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2395** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2396** comments them out. Text in between is also commented out as appropriate.
2397*/
danielk1977940fac92005-01-23 22:41:37 +00002398static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002399 int i, j, k, n;
2400 int exclude = 0;
2401 int start;
2402 int lineno = 1;
2403 int start_lineno;
2404 for(i=0; z[i]; i++){
2405 if( z[i]=='\n' ) lineno++;
2406 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2407 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2408 if( exclude ){
2409 exclude--;
2410 if( exclude==0 ){
2411 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2412 }
2413 }
2414 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2415 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2416 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2417 if( exclude ){
2418 exclude++;
2419 }else{
2420 for(j=i+7; isspace(z[j]); j++){}
2421 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2422 exclude = 1;
2423 for(k=0; k<nDefine; k++){
2424 if( strncmp(azDefine[k],&z[j],n)==0 && strlen(azDefine[k])==n ){
2425 exclude = 0;
2426 break;
2427 }
2428 }
2429 if( z[i+3]=='n' ) exclude = !exclude;
2430 if( exclude ){
2431 start = i;
2432 start_lineno = lineno;
2433 }
2434 }
2435 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2436 }
2437 }
2438 if( exclude ){
2439 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2440 exit(1);
2441 }
2442}
2443
drh75897232000-05-29 14:26:00 +00002444/* In spite of its name, this function is really a scanner. It read
2445** in the entire input file (all at once) then tokenizes it. Each
2446** token is passed to the function "parseonetoken" which builds all
2447** the appropriate data structures in the global state vector "gp".
2448*/
2449void Parse(gp)
2450struct lemon *gp;
2451{
2452 struct pstate ps;
2453 FILE *fp;
2454 char *filebuf;
2455 int filesize;
2456 int lineno;
2457 int c;
2458 char *cp, *nextcp;
2459 int startline = 0;
2460
2461 ps.gp = gp;
2462 ps.filename = gp->filename;
2463 ps.errorcnt = 0;
2464 ps.state = INITIALIZE;
2465
2466 /* Begin by reading the input file */
2467 fp = fopen(ps.filename,"rb");
2468 if( fp==0 ){
2469 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2470 gp->errorcnt++;
2471 return;
2472 }
2473 fseek(fp,0,2);
2474 filesize = ftell(fp);
2475 rewind(fp);
2476 filebuf = (char *)malloc( filesize+1 );
2477 if( filebuf==0 ){
2478 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2479 filesize+1);
2480 gp->errorcnt++;
2481 return;
2482 }
2483 if( fread(filebuf,1,filesize,fp)!=filesize ){
2484 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2485 filesize);
2486 free(filebuf);
2487 gp->errorcnt++;
2488 return;
2489 }
2490 fclose(fp);
2491 filebuf[filesize] = 0;
2492
drh6d08b4d2004-07-20 12:45:22 +00002493 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2494 preprocess_input(filebuf);
2495
drh75897232000-05-29 14:26:00 +00002496 /* Now scan the text of the input file */
2497 lineno = 1;
2498 for(cp=filebuf; (c= *cp)!=0; ){
2499 if( c=='\n' ) lineno++; /* Keep track of the line number */
2500 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2501 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2502 cp+=2;
2503 while( (c= *cp)!=0 && c!='\n' ) cp++;
2504 continue;
2505 }
2506 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2507 cp+=2;
2508 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2509 if( c=='\n' ) lineno++;
2510 cp++;
2511 }
2512 if( c ) cp++;
2513 continue;
2514 }
2515 ps.tokenstart = cp; /* Mark the beginning of the token */
2516 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2517 if( c=='\"' ){ /* String literals */
2518 cp++;
2519 while( (c= *cp)!=0 && c!='\"' ){
2520 if( c=='\n' ) lineno++;
2521 cp++;
2522 }
2523 if( c==0 ){
2524 ErrorMsg(ps.filename,startline,
2525"String starting on this line is not terminated before the end of the file.");
2526 ps.errorcnt++;
2527 nextcp = cp;
2528 }else{
2529 nextcp = cp+1;
2530 }
2531 }else if( c=='{' ){ /* A block of C code */
2532 int level;
2533 cp++;
2534 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2535 if( c=='\n' ) lineno++;
2536 else if( c=='{' ) level++;
2537 else if( c=='}' ) level--;
2538 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2539 int prevc;
2540 cp = &cp[2];
2541 prevc = 0;
2542 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2543 if( c=='\n' ) lineno++;
2544 prevc = c;
2545 cp++;
2546 }
2547 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
2548 cp = &cp[2];
2549 while( (c= *cp)!=0 && c!='\n' ) cp++;
2550 if( c ) lineno++;
2551 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
2552 int startchar, prevc;
2553 startchar = c;
2554 prevc = 0;
2555 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2556 if( c=='\n' ) lineno++;
2557 if( prevc=='\\' ) prevc = 0;
2558 else prevc = c;
2559 }
2560 }
2561 }
2562 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002563 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002564"C code starting on this line is not terminated before the end of the file.");
2565 ps.errorcnt++;
2566 nextcp = cp;
2567 }else{
2568 nextcp = cp+1;
2569 }
2570 }else if( isalnum(c) ){ /* Identifiers */
2571 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2572 nextcp = cp;
2573 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2574 cp += 3;
2575 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002576 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2577 cp += 2;
2578 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2579 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002580 }else{ /* All other (one character) operators */
2581 cp++;
2582 nextcp = cp;
2583 }
2584 c = *cp;
2585 *cp = 0; /* Null terminate the token */
2586 parseonetoken(&ps); /* Parse the token */
2587 *cp = c; /* Restore the buffer */
2588 cp = nextcp;
2589 }
2590 free(filebuf); /* Release the buffer after parsing */
2591 gp->rule = ps.firstrule;
2592 gp->errorcnt = ps.errorcnt;
2593}
2594/*************************** From the file "plink.c" *********************/
2595/*
2596** Routines processing configuration follow-set propagation links
2597** in the LEMON parser generator.
2598*/
2599static struct plink *plink_freelist = 0;
2600
2601/* Allocate a new plink */
2602struct plink *Plink_new(){
2603 struct plink *new;
2604
2605 if( plink_freelist==0 ){
2606 int i;
2607 int amt = 100;
2608 plink_freelist = (struct plink *)malloc( sizeof(struct plink)*amt );
2609 if( plink_freelist==0 ){
2610 fprintf(stderr,
2611 "Unable to allocate memory for a new follow-set propagation link.\n");
2612 exit(1);
2613 }
2614 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2615 plink_freelist[amt-1].next = 0;
2616 }
2617 new = plink_freelist;
2618 plink_freelist = plink_freelist->next;
2619 return new;
2620}
2621
2622/* Add a plink to a plink list */
2623void Plink_add(plpp,cfp)
2624struct plink **plpp;
2625struct config *cfp;
2626{
2627 struct plink *new;
2628 new = Plink_new();
2629 new->next = *plpp;
2630 *plpp = new;
2631 new->cfp = cfp;
2632}
2633
2634/* Transfer every plink on the list "from" to the list "to" */
2635void Plink_copy(to,from)
2636struct plink **to;
2637struct plink *from;
2638{
2639 struct plink *nextpl;
2640 while( from ){
2641 nextpl = from->next;
2642 from->next = *to;
2643 *to = from;
2644 from = nextpl;
2645 }
2646}
2647
2648/* Delete every plink on the list */
2649void Plink_delete(plp)
2650struct plink *plp;
2651{
2652 struct plink *nextpl;
2653
2654 while( plp ){
2655 nextpl = plp->next;
2656 plp->next = plink_freelist;
2657 plink_freelist = plp;
2658 plp = nextpl;
2659 }
2660}
2661/*********************** From the file "report.c" **************************/
2662/*
2663** Procedures for generating reports and tables in the LEMON parser generator.
2664*/
2665
2666/* Generate a filename with the given suffix. Space to hold the
2667** name comes from malloc() and must be freed by the calling
2668** function.
2669*/
2670PRIVATE char *file_makename(lemp,suffix)
2671struct lemon *lemp;
2672char *suffix;
2673{
2674 char *name;
2675 char *cp;
2676
2677 name = malloc( strlen(lemp->filename) + strlen(suffix) + 5 );
2678 if( name==0 ){
2679 fprintf(stderr,"Can't allocate space for a filename.\n");
2680 exit(1);
2681 }
2682 strcpy(name,lemp->filename);
2683 cp = strrchr(name,'.');
2684 if( cp ) *cp = 0;
2685 strcat(name,suffix);
2686 return name;
2687}
2688
2689/* Open a file with a name based on the name of the input file,
2690** but with a different (specified) suffix, and return a pointer
2691** to the stream */
2692PRIVATE FILE *file_open(lemp,suffix,mode)
2693struct lemon *lemp;
2694char *suffix;
2695char *mode;
2696{
2697 FILE *fp;
2698
2699 if( lemp->outname ) free(lemp->outname);
2700 lemp->outname = file_makename(lemp, suffix);
2701 fp = fopen(lemp->outname,mode);
2702 if( fp==0 && *mode=='w' ){
2703 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2704 lemp->errorcnt++;
2705 return 0;
2706 }
2707 return fp;
2708}
2709
2710/* Duplicate the input file without comments and without actions
2711** on rules */
2712void Reprint(lemp)
2713struct lemon *lemp;
2714{
2715 struct rule *rp;
2716 struct symbol *sp;
2717 int i, j, maxlen, len, ncolumns, skip;
2718 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2719 maxlen = 10;
2720 for(i=0; i<lemp->nsymbol; i++){
2721 sp = lemp->symbols[i];
2722 len = strlen(sp->name);
2723 if( len>maxlen ) maxlen = len;
2724 }
2725 ncolumns = 76/(maxlen+5);
2726 if( ncolumns<1 ) ncolumns = 1;
2727 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2728 for(i=0; i<skip; i++){
2729 printf("//");
2730 for(j=i; j<lemp->nsymbol; j+=skip){
2731 sp = lemp->symbols[j];
2732 assert( sp->index==j );
2733 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2734 }
2735 printf("\n");
2736 }
2737 for(rp=lemp->rule; rp; rp=rp->next){
2738 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002739 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002740 printf(" ::=");
2741 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002742 sp = rp->rhs[i];
2743 printf(" %s", sp->name);
2744 if( sp->type==MULTITERMINAL ){
2745 for(j=1; j<sp->nsubsym; j++){
2746 printf("|%s", sp->subsym[j]->name);
2747 }
2748 }
2749 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002750 }
2751 printf(".");
2752 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002753 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002754 printf("\n");
2755 }
2756}
2757
2758void ConfigPrint(fp,cfp)
2759FILE *fp;
2760struct config *cfp;
2761{
2762 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002763 struct symbol *sp;
2764 int i, j;
drh75897232000-05-29 14:26:00 +00002765 rp = cfp->rp;
2766 fprintf(fp,"%s ::=",rp->lhs->name);
2767 for(i=0; i<=rp->nrhs; i++){
2768 if( i==cfp->dot ) fprintf(fp," *");
2769 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002770 sp = rp->rhs[i];
2771 fprintf(fp," %s", sp->name);
2772 if( sp->type==MULTITERMINAL ){
2773 for(j=1; j<sp->nsubsym; j++){
2774 fprintf(fp,"|%s",sp->subsym[j]->name);
2775 }
2776 }
drh75897232000-05-29 14:26:00 +00002777 }
2778}
2779
2780/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002781#if 0
drh75897232000-05-29 14:26:00 +00002782/* Print a set */
2783PRIVATE void SetPrint(out,set,lemp)
2784FILE *out;
2785char *set;
2786struct lemon *lemp;
2787{
2788 int i;
2789 char *spacer;
2790 spacer = "";
2791 fprintf(out,"%12s[","");
2792 for(i=0; i<lemp->nterminal; i++){
2793 if( SetFind(set,i) ){
2794 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2795 spacer = " ";
2796 }
2797 }
2798 fprintf(out,"]\n");
2799}
2800
2801/* Print a plink chain */
2802PRIVATE void PlinkPrint(out,plp,tag)
2803FILE *out;
2804struct plink *plp;
2805char *tag;
2806{
2807 while( plp ){
drhada354d2005-11-05 15:03:59 +00002808 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002809 ConfigPrint(out,plp->cfp);
2810 fprintf(out,"\n");
2811 plp = plp->next;
2812 }
2813}
2814#endif
2815
2816/* Print an action to the given file descriptor. Return FALSE if
2817** nothing was actually printed.
2818*/
2819int PrintAction(struct action *ap, FILE *fp, int indent){
2820 int result = 1;
2821 switch( ap->type ){
2822 case SHIFT:
drhada354d2005-11-05 15:03:59 +00002823 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00002824 break;
2825 case REDUCE:
2826 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
2827 break;
2828 case ACCEPT:
2829 fprintf(fp,"%*s accept",indent,ap->sp->name);
2830 break;
2831 case ERROR:
2832 fprintf(fp,"%*s error",indent,ap->sp->name);
2833 break;
2834 case CONFLICT:
2835 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
2836 indent,ap->sp->name,ap->x.rp->index);
2837 break;
2838 case SH_RESOLVED:
2839 case RD_RESOLVED:
2840 case NOT_USED:
2841 result = 0;
2842 break;
2843 }
2844 return result;
2845}
2846
2847/* Generate the "y.output" log file */
2848void ReportOutput(lemp)
2849struct lemon *lemp;
2850{
2851 int i;
2852 struct state *stp;
2853 struct config *cfp;
2854 struct action *ap;
2855 FILE *fp;
2856
drh2aa6ca42004-09-10 00:14:04 +00002857 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00002858 if( fp==0 ) return;
2859 fprintf(fp," \b");
2860 for(i=0; i<lemp->nstate; i++){
2861 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00002862 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00002863 if( lemp->basisflag ) cfp=stp->bp;
2864 else cfp=stp->cfp;
2865 while( cfp ){
2866 char buf[20];
2867 if( cfp->dot==cfp->rp->nrhs ){
2868 sprintf(buf,"(%d)",cfp->rp->index);
2869 fprintf(fp," %5s ",buf);
2870 }else{
2871 fprintf(fp," ");
2872 }
2873 ConfigPrint(fp,cfp);
2874 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00002875#if 0
drh75897232000-05-29 14:26:00 +00002876 SetPrint(fp,cfp->fws,lemp);
2877 PlinkPrint(fp,cfp->fplp,"To ");
2878 PlinkPrint(fp,cfp->bplp,"From");
2879#endif
2880 if( lemp->basisflag ) cfp=cfp->bp;
2881 else cfp=cfp->next;
2882 }
2883 fprintf(fp,"\n");
2884 for(ap=stp->ap; ap; ap=ap->next){
2885 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
2886 }
2887 fprintf(fp,"\n");
2888 }
2889 fclose(fp);
2890 return;
2891}
2892
2893/* Search for the file "name" which is in the same directory as
2894** the exacutable */
2895PRIVATE char *pathsearch(argv0,name,modemask)
2896char *argv0;
2897char *name;
2898int modemask;
2899{
2900 char *pathlist;
2901 char *path,*cp;
2902 char c;
2903 extern int access();
2904
2905#ifdef __WIN32__
2906 cp = strrchr(argv0,'\\');
2907#else
2908 cp = strrchr(argv0,'/');
2909#endif
2910 if( cp ){
2911 c = *cp;
2912 *cp = 0;
2913 path = (char *)malloc( strlen(argv0) + strlen(name) + 2 );
2914 if( path ) sprintf(path,"%s/%s",argv0,name);
2915 *cp = c;
2916 }else{
2917 extern char *getenv();
2918 pathlist = getenv("PATH");
2919 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
2920 path = (char *)malloc( strlen(pathlist)+strlen(name)+2 );
2921 if( path!=0 ){
2922 while( *pathlist ){
2923 cp = strchr(pathlist,':');
2924 if( cp==0 ) cp = &pathlist[strlen(pathlist)];
2925 c = *cp;
2926 *cp = 0;
2927 sprintf(path,"%s/%s",pathlist,name);
2928 *cp = c;
2929 if( c==0 ) pathlist = "";
2930 else pathlist = &cp[1];
2931 if( access(path,modemask)==0 ) break;
2932 }
2933 }
2934 }
2935 return path;
2936}
2937
2938/* Given an action, compute the integer value for that action
2939** which is to be put in the action table of the generated machine.
2940** Return negative if no action should be generated.
2941*/
2942PRIVATE int compute_action(lemp,ap)
2943struct lemon *lemp;
2944struct action *ap;
2945{
2946 int act;
2947 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00002948 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00002949 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
2950 case ERROR: act = lemp->nstate + lemp->nrule; break;
2951 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
2952 default: act = -1; break;
2953 }
2954 return act;
2955}
2956
2957#define LINESIZE 1000
2958/* The next cluster of routines are for reading the template file
2959** and writing the results to the generated parser */
2960/* The first function transfers data from "in" to "out" until
2961** a line is seen which begins with "%%". The line number is
2962** tracked.
2963**
2964** if name!=0, then any word that begin with "Parse" is changed to
2965** begin with *name instead.
2966*/
2967PRIVATE void tplt_xfer(name,in,out,lineno)
2968char *name;
2969FILE *in;
2970FILE *out;
2971int *lineno;
2972{
2973 int i, iStart;
2974 char line[LINESIZE];
2975 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
2976 (*lineno)++;
2977 iStart = 0;
2978 if( name ){
2979 for(i=0; line[i]; i++){
2980 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
2981 && (i==0 || !isalpha(line[i-1]))
2982 ){
2983 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
2984 fprintf(out,"%s",name);
2985 i += 4;
2986 iStart = i+1;
2987 }
2988 }
2989 }
2990 fprintf(out,"%s",&line[iStart]);
2991 }
2992}
2993
2994/* The next function finds the template file and opens it, returning
2995** a pointer to the opened file. */
2996PRIVATE FILE *tplt_open(lemp)
2997struct lemon *lemp;
2998{
2999 static char templatename[] = "lempar.c";
3000 char buf[1000];
3001 FILE *in;
3002 char *tpltname;
3003 char *cp;
3004
3005 cp = strrchr(lemp->filename,'.');
3006 if( cp ){
drh8b582012003-10-21 13:16:03 +00003007 sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003008 }else{
3009 sprintf(buf,"%s.lt",lemp->filename);
3010 }
3011 if( access(buf,004)==0 ){
3012 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003013 }else if( access(templatename,004)==0 ){
3014 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003015 }else{
3016 tpltname = pathsearch(lemp->argv0,templatename,0);
3017 }
3018 if( tpltname==0 ){
3019 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3020 templatename);
3021 lemp->errorcnt++;
3022 return 0;
3023 }
drh2aa6ca42004-09-10 00:14:04 +00003024 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003025 if( in==0 ){
3026 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3027 lemp->errorcnt++;
3028 return 0;
3029 }
3030 return in;
3031}
3032
drhaf805ca2004-09-07 11:28:25 +00003033/* Print a #line directive line to the output file. */
3034PRIVATE void tplt_linedir(out,lineno,filename)
3035FILE *out;
3036int lineno;
3037char *filename;
3038{
3039 fprintf(out,"#line %d \"",lineno);
3040 while( *filename ){
3041 if( *filename == '\\' ) putc('\\',out);
3042 putc(*filename,out);
3043 filename++;
3044 }
3045 fprintf(out,"\"\n");
3046}
3047
drh75897232000-05-29 14:26:00 +00003048/* Print a string to the file and keep the linenumber up to date */
3049PRIVATE void tplt_print(out,lemp,str,strln,lineno)
3050FILE *out;
3051struct lemon *lemp;
3052char *str;
3053int strln;
3054int *lineno;
3055{
3056 if( str==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00003057 tplt_linedir(out,strln,lemp->filename);
3058 (*lineno)++;
drh75897232000-05-29 14:26:00 +00003059 while( *str ){
3060 if( *str=='\n' ) (*lineno)++;
3061 putc(*str,out);
3062 str++;
3063 }
drh9db55df2004-09-09 14:01:21 +00003064 if( str[-1]!='\n' ){
3065 putc('\n',out);
3066 (*lineno)++;
3067 }
drhaf805ca2004-09-07 11:28:25 +00003068 tplt_linedir(out,*lineno+2,lemp->outname);
3069 (*lineno)+=2;
drh75897232000-05-29 14:26:00 +00003070 return;
3071}
3072
3073/*
3074** The following routine emits code for the destructor for the
3075** symbol sp
3076*/
3077void emit_destructor_code(out,sp,lemp,lineno)
3078FILE *out;
3079struct symbol *sp;
3080struct lemon *lemp;
3081int *lineno;
3082{
drhcc83b6e2004-04-23 23:38:42 +00003083 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003084
3085 int linecnt = 0;
3086 if( sp->type==TERMINAL ){
3087 cp = lemp->tokendest;
3088 if( cp==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00003089 tplt_linedir(out,lemp->tokendestln,lemp->filename);
3090 fprintf(out,"{");
drh960e8c62001-04-03 16:53:21 +00003091 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003092 cp = sp->destructor;
drhaf805ca2004-09-07 11:28:25 +00003093 tplt_linedir(out,sp->destructorln,lemp->filename);
3094 fprintf(out,"{");
drh960e8c62001-04-03 16:53:21 +00003095 }else if( lemp->vardest ){
3096 cp = lemp->vardest;
3097 if( cp==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00003098 tplt_linedir(out,lemp->vardestln,lemp->filename);
3099 fprintf(out,"{");
drhcc83b6e2004-04-23 23:38:42 +00003100 }else{
3101 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003102 }
3103 for(; *cp; cp++){
3104 if( *cp=='$' && cp[1]=='$' ){
3105 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3106 cp++;
3107 continue;
3108 }
3109 if( *cp=='\n' ) linecnt++;
3110 fputc(*cp,out);
3111 }
3112 (*lineno) += 3 + linecnt;
drhaf805ca2004-09-07 11:28:25 +00003113 fprintf(out,"}\n");
3114 tplt_linedir(out,*lineno,lemp->outname);
drh75897232000-05-29 14:26:00 +00003115 return;
3116}
3117
3118/*
drh960e8c62001-04-03 16:53:21 +00003119** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003120*/
3121int has_destructor(sp, lemp)
3122struct symbol *sp;
3123struct lemon *lemp;
3124{
3125 int ret;
3126 if( sp->type==TERMINAL ){
3127 ret = lemp->tokendest!=0;
3128 }else{
drh960e8c62001-04-03 16:53:21 +00003129 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003130 }
3131 return ret;
3132}
3133
drh0bb132b2004-07-20 14:06:51 +00003134/*
3135** Append text to a dynamically allocated string. If zText is 0 then
3136** reset the string to be empty again. Always return the complete text
3137** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003138**
3139** n bytes of zText are stored. If n==0 then all of zText up to the first
3140** \000 terminator is stored. zText can contain up to two instances of
3141** %d. The values of p1 and p2 are written into the first and second
3142** %d.
3143**
3144** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003145*/
3146PRIVATE char *append_str(char *zText, int n, int p1, int p2){
3147 static char *z = 0;
3148 static int alloced = 0;
3149 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003150 int c;
drh0bb132b2004-07-20 14:06:51 +00003151 char zInt[40];
3152
3153 if( zText==0 ){
3154 used = 0;
3155 return z;
3156 }
drh7ac25c72004-08-19 15:12:26 +00003157 if( n<=0 ){
3158 if( n<0 ){
3159 used += n;
3160 assert( used>=0 );
3161 }
3162 n = strlen(zText);
3163 }
drh0bb132b2004-07-20 14:06:51 +00003164 if( n+sizeof(zInt)*2+used >= alloced ){
3165 alloced = n + sizeof(zInt)*2 + used + 200;
3166 z = realloc(z, alloced);
3167 }
3168 if( z==0 ) return "";
3169 while( n-- > 0 ){
3170 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003171 if( c=='%' && n>0 && zText[0]=='d' ){
drh0bb132b2004-07-20 14:06:51 +00003172 sprintf(zInt, "%d", p1);
3173 p1 = p2;
3174 strcpy(&z[used], zInt);
3175 used += strlen(&z[used]);
3176 zText++;
3177 n--;
3178 }else{
3179 z[used++] = c;
3180 }
3181 }
3182 z[used] = 0;
3183 return z;
3184}
3185
3186/*
3187** zCode is a string that is the action associated with a rule. Expand
3188** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003189** stack.
drh0bb132b2004-07-20 14:06:51 +00003190*/
drhaf805ca2004-09-07 11:28:25 +00003191PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003192 char *cp, *xp;
3193 int i;
3194 char lhsused = 0; /* True if the LHS element has been used */
3195 char used[MAXRHS]; /* True for each RHS element which is used */
3196
3197 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3198 lhsused = 0;
3199
drh19c9e562007-03-29 20:13:53 +00003200 if( rp->code==0 ){
3201 rp->code = "\n";
3202 rp->line = rp->ruleline;
3203 }
3204
drh0bb132b2004-07-20 14:06:51 +00003205 append_str(0,0,0,0);
drh19c9e562007-03-29 20:13:53 +00003206 for(cp=rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003207 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3208 char saved;
3209 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3210 saved = *xp;
3211 *xp = 0;
3212 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003213 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003214 cp = xp;
3215 lhsused = 1;
3216 }else{
3217 for(i=0; i<rp->nrhs; i++){
3218 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003219 if( cp!=rp->code && cp[-1]=='@' ){
3220 /* If the argument is of the form @X then substituted
3221 ** the token number of X, not the value of X */
3222 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3223 }else{
drhfd405312005-11-06 04:06:59 +00003224 struct symbol *sp = rp->rhs[i];
3225 int dtnum;
3226 if( sp->type==MULTITERMINAL ){
3227 dtnum = sp->subsym[0]->dtnum;
3228 }else{
3229 dtnum = sp->dtnum;
3230 }
3231 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003232 }
drh0bb132b2004-07-20 14:06:51 +00003233 cp = xp;
3234 used[i] = 1;
3235 break;
3236 }
3237 }
3238 }
3239 *xp = saved;
3240 }
3241 append_str(cp, 1, 0, 0);
3242 } /* End loop */
3243
3244 /* Check to make sure the LHS has been used */
3245 if( rp->lhsalias && !lhsused ){
3246 ErrorMsg(lemp->filename,rp->ruleline,
3247 "Label \"%s\" for \"%s(%s)\" is never used.",
3248 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3249 lemp->errorcnt++;
3250 }
3251
3252 /* Generate destructor code for RHS symbols which are not used in the
3253 ** reduce code */
3254 for(i=0; i<rp->nrhs; i++){
3255 if( rp->rhsalias[i] && !used[i] ){
3256 ErrorMsg(lemp->filename,rp->ruleline,
3257 "Label %s for \"%s(%s)\" is never used.",
3258 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3259 lemp->errorcnt++;
3260 }else if( rp->rhsalias[i]==0 ){
3261 if( has_destructor(rp->rhs[i],lemp) ){
drh7ac25c72004-08-19 15:12:26 +00003262 append_str(" yy_destructor(%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003263 rp->rhs[i]->index,i-rp->nrhs+1);
3264 }else{
3265 /* No destructor defined for this term */
3266 }
3267 }
3268 }
drh61e339a2007-01-16 03:09:02 +00003269 if( rp->code ){
3270 cp = append_str(0,0,0,0);
3271 rp->code = Strsafe(cp?cp:"");
3272 }
drh0bb132b2004-07-20 14:06:51 +00003273}
3274
drh75897232000-05-29 14:26:00 +00003275/*
3276** Generate code which executes when the rule "rp" is reduced. Write
3277** the code to "out". Make sure lineno stays up-to-date.
3278*/
3279PRIVATE void emit_code(out,rp,lemp,lineno)
3280FILE *out;
3281struct rule *rp;
3282struct lemon *lemp;
3283int *lineno;
3284{
drh0bb132b2004-07-20 14:06:51 +00003285 char *cp;
drh75897232000-05-29 14:26:00 +00003286 int linecnt = 0;
drh75897232000-05-29 14:26:00 +00003287
3288 /* Generate code to do the reduce action */
3289 if( rp->code ){
drhaf805ca2004-09-07 11:28:25 +00003290 tplt_linedir(out,rp->line,lemp->filename);
3291 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003292 for(cp=rp->code; *cp; cp++){
drh75897232000-05-29 14:26:00 +00003293 if( *cp=='\n' ) linecnt++;
drh75897232000-05-29 14:26:00 +00003294 } /* End loop */
3295 (*lineno) += 3 + linecnt;
drhaf805ca2004-09-07 11:28:25 +00003296 fprintf(out,"}\n");
3297 tplt_linedir(out,*lineno,lemp->outname);
drh75897232000-05-29 14:26:00 +00003298 } /* End if( rp->code ) */
3299
drh75897232000-05-29 14:26:00 +00003300 return;
3301}
3302
3303/*
3304** Print the definition of the union used for the parser's data stack.
3305** This union contains fields for every possible data type for tokens
3306** and nonterminals. In the process of computing and printing this
3307** union, also set the ".dtnum" field of every terminal and nonterminal
3308** symbol.
3309*/
3310void print_stack_union(out,lemp,plineno,mhflag)
3311FILE *out; /* The output stream */
3312struct lemon *lemp; /* The main info structure for this parser */
3313int *plineno; /* Pointer to the line number */
3314int mhflag; /* True if generating makeheaders output */
3315{
3316 int lineno = *plineno; /* The line number of the output */
3317 char **types; /* A hash table of datatypes */
3318 int arraysize; /* Size of the "types" array */
3319 int maxdtlength; /* Maximum length of any ".datatype" field. */
3320 char *stddt; /* Standardized name for a datatype */
3321 int i,j; /* Loop counters */
3322 int hash; /* For hashing the name of a type */
3323 char *name; /* Name of the parser */
3324
3325 /* Allocate and initialize types[] and allocate stddt[] */
3326 arraysize = lemp->nsymbol * 2;
3327 types = (char**)malloc( arraysize * sizeof(char*) );
3328 for(i=0; i<arraysize; i++) types[i] = 0;
3329 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003330 if( lemp->vartype ){
3331 maxdtlength = strlen(lemp->vartype);
3332 }
drh75897232000-05-29 14:26:00 +00003333 for(i=0; i<lemp->nsymbol; i++){
3334 int len;
3335 struct symbol *sp = lemp->symbols[i];
3336 if( sp->datatype==0 ) continue;
3337 len = strlen(sp->datatype);
3338 if( len>maxdtlength ) maxdtlength = len;
3339 }
3340 stddt = (char*)malloc( maxdtlength*2 + 1 );
3341 if( types==0 || stddt==0 ){
3342 fprintf(stderr,"Out of memory.\n");
3343 exit(1);
3344 }
3345
3346 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3347 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003348 ** used for terminal symbols. If there is no %default_type defined then
3349 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3350 ** a datatype using the %type directive.
3351 */
drh75897232000-05-29 14:26:00 +00003352 for(i=0; i<lemp->nsymbol; i++){
3353 struct symbol *sp = lemp->symbols[i];
3354 char *cp;
3355 if( sp==lemp->errsym ){
3356 sp->dtnum = arraysize+1;
3357 continue;
3358 }
drh960e8c62001-04-03 16:53:21 +00003359 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003360 sp->dtnum = 0;
3361 continue;
3362 }
3363 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003364 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003365 j = 0;
3366 while( isspace(*cp) ) cp++;
3367 while( *cp ) stddt[j++] = *cp++;
3368 while( j>0 && isspace(stddt[j-1]) ) j--;
3369 stddt[j] = 0;
3370 hash = 0;
3371 for(j=0; stddt[j]; j++){
3372 hash = hash*53 + stddt[j];
3373 }
drh3b2129c2003-05-13 00:34:21 +00003374 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003375 while( types[hash] ){
3376 if( strcmp(types[hash],stddt)==0 ){
3377 sp->dtnum = hash + 1;
3378 break;
3379 }
3380 hash++;
3381 if( hash>=arraysize ) hash = 0;
3382 }
3383 if( types[hash]==0 ){
3384 sp->dtnum = hash + 1;
3385 types[hash] = (char*)malloc( strlen(stddt)+1 );
3386 if( types[hash]==0 ){
3387 fprintf(stderr,"Out of memory.\n");
3388 exit(1);
3389 }
3390 strcpy(types[hash],stddt);
3391 }
3392 }
3393
3394 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3395 name = lemp->name ? lemp->name : "Parse";
3396 lineno = *plineno;
3397 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3398 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3399 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3400 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3401 fprintf(out,"typedef union {\n"); lineno++;
3402 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3403 for(i=0; i<arraysize; i++){
3404 if( types[i]==0 ) continue;
3405 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3406 free(types[i]);
3407 }
3408 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3409 free(stddt);
3410 free(types);
3411 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3412 *plineno = lineno;
3413}
3414
drhb29b0a52002-02-23 19:39:46 +00003415/*
3416** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003417** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003418*/
drh8b582012003-10-21 13:16:03 +00003419static const char *minimum_size_type(int lwr, int upr){
3420 if( lwr>=0 ){
3421 if( upr<=255 ){
3422 return "unsigned char";
3423 }else if( upr<65535 ){
3424 return "unsigned short int";
3425 }else{
3426 return "unsigned int";
3427 }
3428 }else if( lwr>=-127 && upr<=127 ){
3429 return "signed char";
3430 }else if( lwr>=-32767 && upr<32767 ){
3431 return "short";
drhb29b0a52002-02-23 19:39:46 +00003432 }else{
drh8b582012003-10-21 13:16:03 +00003433 return "int";
drhb29b0a52002-02-23 19:39:46 +00003434 }
3435}
3436
drhfdbf9282003-10-21 16:34:41 +00003437/*
3438** Each state contains a set of token transaction and a set of
3439** nonterminal transactions. Each of these sets makes an instance
3440** of the following structure. An array of these structures is used
3441** to order the creation of entries in the yy_action[] table.
3442*/
3443struct axset {
3444 struct state *stp; /* A pointer to a state */
3445 int isTkn; /* True to use tokens. False for non-terminals */
3446 int nAction; /* Number of actions */
3447};
3448
3449/*
3450** Compare to axset structures for sorting purposes
3451*/
3452static int axset_compare(const void *a, const void *b){
3453 struct axset *p1 = (struct axset*)a;
3454 struct axset *p2 = (struct axset*)b;
3455 return p2->nAction - p1->nAction;
3456}
3457
drh75897232000-05-29 14:26:00 +00003458/* Generate C source code for the parser */
3459void ReportTable(lemp, mhflag)
3460struct lemon *lemp;
3461int mhflag; /* Output in makeheaders format if true */
3462{
3463 FILE *out, *in;
3464 char line[LINESIZE];
3465 int lineno;
3466 struct state *stp;
3467 struct action *ap;
3468 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003469 struct acttab *pActtab;
3470 int i, j, n;
drh75897232000-05-29 14:26:00 +00003471 char *name;
drh8b582012003-10-21 13:16:03 +00003472 int mnTknOfst, mxTknOfst;
3473 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003474 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003475
3476 in = tplt_open(lemp);
3477 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003478 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003479 if( out==0 ){
3480 fclose(in);
3481 return;
3482 }
3483 lineno = 1;
3484 tplt_xfer(lemp->name,in,out,&lineno);
3485
3486 /* Generate the include code, if any */
3487 tplt_print(out,lemp,lemp->include,lemp->includeln,&lineno);
3488 if( mhflag ){
3489 char *name = file_makename(lemp, ".h");
3490 fprintf(out,"#include \"%s\"\n", name); lineno++;
3491 free(name);
3492 }
3493 tplt_xfer(lemp->name,in,out,&lineno);
3494
3495 /* Generate #defines for all tokens */
3496 if( mhflag ){
3497 char *prefix;
3498 fprintf(out,"#if INTERFACE\n"); lineno++;
3499 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3500 else prefix = "";
3501 for(i=1; i<lemp->nterminal; i++){
3502 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3503 lineno++;
3504 }
3505 fprintf(out,"#endif\n"); lineno++;
3506 }
3507 tplt_xfer(lemp->name,in,out,&lineno);
3508
3509 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003510 fprintf(out,"#define YYCODETYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003511 minimum_size_type(0, lemp->nsymbol+5)); lineno++;
drh75897232000-05-29 14:26:00 +00003512 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3513 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003514 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003515 if( lemp->wildcard ){
3516 fprintf(out,"#define YYWILDCARD %d\n",
3517 lemp->wildcard->index); lineno++;
3518 }
drh75897232000-05-29 14:26:00 +00003519 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003520 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003521 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003522 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3523 }else{
3524 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3525 }
drhca44b5a2007-02-22 23:06:58 +00003526 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003527 if( mhflag ){
3528 fprintf(out,"#if INTERFACE\n"); lineno++;
3529 }
3530 name = lemp->name ? lemp->name : "Parse";
3531 if( lemp->arg && lemp->arg[0] ){
3532 int i;
3533 i = strlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003534 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3535 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003536 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3537 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3538 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3539 name,lemp->arg,&lemp->arg[i]); lineno++;
3540 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3541 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003542 }else{
drh1f245e42002-03-11 13:55:50 +00003543 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3544 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3545 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3546 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003547 }
3548 if( mhflag ){
3549 fprintf(out,"#endif\n"); lineno++;
3550 }
3551 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3552 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
3553 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3554 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
drh0bd1f4e2002-06-06 18:54:39 +00003555 if( lemp->has_fallback ){
3556 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3557 }
drh75897232000-05-29 14:26:00 +00003558 tplt_xfer(lemp->name,in,out,&lineno);
3559
drh8b582012003-10-21 13:16:03 +00003560 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003561 **
drh8b582012003-10-21 13:16:03 +00003562 ** yy_action[] A single table containing all actions.
3563 ** yy_lookahead[] A table containing the lookahead for each entry in
3564 ** yy_action. Used to detect hash collisions.
3565 ** yy_shift_ofst[] For each state, the offset into yy_action for
3566 ** shifting terminals.
3567 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3568 ** shifting non-terminals after a reduce.
3569 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003570 */
drh75897232000-05-29 14:26:00 +00003571
drh8b582012003-10-21 13:16:03 +00003572 /* Compute the actions on all states and count them up */
drhfdbf9282003-10-21 16:34:41 +00003573 ax = malloc( sizeof(ax[0])*lemp->nstate*2 );
3574 if( ax==0 ){
3575 fprintf(stderr,"malloc failed\n");
3576 exit(1);
3577 }
drh75897232000-05-29 14:26:00 +00003578 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003579 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003580 ax[i*2].stp = stp;
3581 ax[i*2].isTkn = 1;
3582 ax[i*2].nAction = stp->nTknAct;
3583 ax[i*2+1].stp = stp;
3584 ax[i*2+1].isTkn = 0;
3585 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003586 }
drh8b582012003-10-21 13:16:03 +00003587 mxTknOfst = mnTknOfst = 0;
3588 mxNtOfst = mnNtOfst = 0;
3589
drhfdbf9282003-10-21 16:34:41 +00003590 /* Compute the action table. In order to try to keep the size of the
3591 ** action table to a minimum, the heuristic of placing the largest action
3592 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003593 */
drhfdbf9282003-10-21 16:34:41 +00003594 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003595 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003596 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3597 stp = ax[i].stp;
3598 if( ax[i].isTkn ){
3599 for(ap=stp->ap; ap; ap=ap->next){
3600 int action;
3601 if( ap->sp->index>=lemp->nterminal ) continue;
3602 action = compute_action(lemp, ap);
3603 if( action<0 ) continue;
3604 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003605 }
drhfdbf9282003-10-21 16:34:41 +00003606 stp->iTknOfst = acttab_insert(pActtab);
3607 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3608 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3609 }else{
3610 for(ap=stp->ap; ap; ap=ap->next){
3611 int action;
3612 if( ap->sp->index<lemp->nterminal ) continue;
3613 if( ap->sp->index==lemp->nsymbol ) continue;
3614 action = compute_action(lemp, ap);
3615 if( action<0 ) continue;
3616 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003617 }
drhfdbf9282003-10-21 16:34:41 +00003618 stp->iNtOfst = acttab_insert(pActtab);
3619 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3620 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003621 }
3622 }
drhfdbf9282003-10-21 16:34:41 +00003623 free(ax);
drh8b582012003-10-21 13:16:03 +00003624
3625 /* Output the yy_action table */
drh57196282004-10-06 15:41:16 +00003626 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003627 n = acttab_size(pActtab);
3628 for(i=j=0; i<n; i++){
3629 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003630 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003631 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003632 fprintf(out, " %4d,", action);
3633 if( j==9 || i==n-1 ){
3634 fprintf(out, "\n"); lineno++;
3635 j = 0;
3636 }else{
3637 j++;
3638 }
3639 }
3640 fprintf(out, "};\n"); lineno++;
3641
3642 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003643 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003644 for(i=j=0; i<n; i++){
3645 int la = acttab_yylookahead(pActtab, i);
3646 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003647 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003648 fprintf(out, " %4d,", la);
3649 if( j==9 || i==n-1 ){
3650 fprintf(out, "\n"); lineno++;
3651 j = 0;
3652 }else{
3653 j++;
3654 }
3655 }
3656 fprintf(out, "};\n"); lineno++;
3657
3658 /* Output the yy_shift_ofst[] table */
3659 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003660 n = lemp->nstate;
3661 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
3662 fprintf(out, "#define YY_SHIFT_MAX %d\n", n-1); lineno++;
drh57196282004-10-06 15:41:16 +00003663 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003664 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003665 for(i=j=0; i<n; i++){
3666 int ofst;
3667 stp = lemp->sorted[i];
3668 ofst = stp->iTknOfst;
3669 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003670 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003671 fprintf(out, " %4d,", ofst);
3672 if( j==9 || i==n-1 ){
3673 fprintf(out, "\n"); lineno++;
3674 j = 0;
3675 }else{
3676 j++;
3677 }
3678 }
3679 fprintf(out, "};\n"); lineno++;
3680
3681 /* Output the yy_reduce_ofst[] table */
3682 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003683 n = lemp->nstate;
3684 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
3685 fprintf(out, "#define YY_REDUCE_MAX %d\n", n-1); lineno++;
drh57196282004-10-06 15:41:16 +00003686 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003687 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003688 for(i=j=0; i<n; i++){
3689 int ofst;
3690 stp = lemp->sorted[i];
3691 ofst = stp->iNtOfst;
3692 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003693 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003694 fprintf(out, " %4d,", ofst);
3695 if( j==9 || i==n-1 ){
3696 fprintf(out, "\n"); lineno++;
3697 j = 0;
3698 }else{
3699 j++;
3700 }
3701 }
3702 fprintf(out, "};\n"); lineno++;
3703
3704 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003705 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003706 n = lemp->nstate;
3707 for(i=j=0; i<n; i++){
3708 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003709 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003710 fprintf(out, " %4d,", stp->iDflt);
3711 if( j==9 || i==n-1 ){
3712 fprintf(out, "\n"); lineno++;
3713 j = 0;
3714 }else{
3715 j++;
3716 }
3717 }
3718 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003719 tplt_xfer(lemp->name,in,out,&lineno);
3720
drh0bd1f4e2002-06-06 18:54:39 +00003721 /* Generate the table of fallback tokens.
3722 */
3723 if( lemp->has_fallback ){
3724 for(i=0; i<lemp->nterminal; i++){
3725 struct symbol *p = lemp->symbols[i];
3726 if( p->fallback==0 ){
3727 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3728 }else{
3729 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3730 p->name, p->fallback->name);
3731 }
3732 lineno++;
3733 }
3734 }
3735 tplt_xfer(lemp->name, in, out, &lineno);
3736
3737 /* Generate a table containing the symbolic name of every symbol
3738 */
drh75897232000-05-29 14:26:00 +00003739 for(i=0; i<lemp->nsymbol; i++){
3740 sprintf(line,"\"%s\",",lemp->symbols[i]->name);
3741 fprintf(out," %-15s",line);
3742 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
3743 }
3744 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
3745 tplt_xfer(lemp->name,in,out,&lineno);
3746
drh0bd1f4e2002-06-06 18:54:39 +00003747 /* Generate a table containing a text string that describes every
3748 ** rule in the rule set of the grammer. This information is used
3749 ** when tracing REDUCE actions.
3750 */
3751 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
3752 assert( rp->index==i );
3753 fprintf(out," /* %3d */ \"%s ::=", i, rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00003754 for(j=0; j<rp->nrhs; j++){
3755 struct symbol *sp = rp->rhs[j];
3756 fprintf(out," %s", sp->name);
3757 if( sp->type==MULTITERMINAL ){
3758 int k;
3759 for(k=1; k<sp->nsubsym; k++){
3760 fprintf(out,"|%s",sp->subsym[k]->name);
3761 }
3762 }
3763 }
drh0bd1f4e2002-06-06 18:54:39 +00003764 fprintf(out,"\",\n"); lineno++;
3765 }
3766 tplt_xfer(lemp->name,in,out,&lineno);
3767
drh75897232000-05-29 14:26:00 +00003768 /* Generate code which executes every time a symbol is popped from
3769 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00003770 ** (In other words, generate the %destructor actions)
3771 */
drh75897232000-05-29 14:26:00 +00003772 if( lemp->tokendest ){
3773 for(i=0; i<lemp->nsymbol; i++){
3774 struct symbol *sp = lemp->symbols[i];
3775 if( sp==0 || sp->type!=TERMINAL ) continue;
3776 fprintf(out," case %d:\n",sp->index); lineno++;
3777 }
3778 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
3779 if( i<lemp->nsymbol ){
3780 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3781 fprintf(out," break;\n"); lineno++;
3782 }
3783 }
drh8d659732005-01-13 23:54:06 +00003784 if( lemp->vardest ){
3785 struct symbol *dflt_sp = 0;
3786 for(i=0; i<lemp->nsymbol; i++){
3787 struct symbol *sp = lemp->symbols[i];
3788 if( sp==0 || sp->type==TERMINAL ||
3789 sp->index<=0 || sp->destructor!=0 ) continue;
3790 fprintf(out," case %d:\n",sp->index); lineno++;
3791 dflt_sp = sp;
3792 }
3793 if( dflt_sp!=0 ){
3794 emit_destructor_code(out,dflt_sp,lemp,&lineno);
3795 fprintf(out," break;\n"); lineno++;
3796 }
3797 }
drh75897232000-05-29 14:26:00 +00003798 for(i=0; i<lemp->nsymbol; i++){
3799 struct symbol *sp = lemp->symbols[i];
3800 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
3801 fprintf(out," case %d:\n",sp->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003802
3803 /* Combine duplicate destructors into a single case */
3804 for(j=i+1; j<lemp->nsymbol; j++){
3805 struct symbol *sp2 = lemp->symbols[j];
3806 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
3807 && sp2->dtnum==sp->dtnum
3808 && strcmp(sp->destructor,sp2->destructor)==0 ){
3809 fprintf(out," case %d:\n",sp2->index); lineno++;
3810 sp2->destructor = 0;
3811 }
3812 }
3813
drh75897232000-05-29 14:26:00 +00003814 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3815 fprintf(out," break;\n"); lineno++;
3816 }
drh75897232000-05-29 14:26:00 +00003817 tplt_xfer(lemp->name,in,out,&lineno);
3818
3819 /* Generate code which executes whenever the parser stack overflows */
3820 tplt_print(out,lemp,lemp->overflow,lemp->overflowln,&lineno);
3821 tplt_xfer(lemp->name,in,out,&lineno);
3822
3823 /* Generate the table of rule information
3824 **
3825 ** Note: This code depends on the fact that rules are number
3826 ** sequentually beginning with 0.
3827 */
3828 for(rp=lemp->rule; rp; rp=rp->next){
3829 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
3830 }
3831 tplt_xfer(lemp->name,in,out,&lineno);
3832
3833 /* Generate code which execution during each REDUCE action */
3834 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00003835 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00003836 }
3837 for(rp=lemp->rule; rp; rp=rp->next){
3838 struct rule *rp2;
3839 if( rp->code==0 ) continue;
drh75897232000-05-29 14:26:00 +00003840 fprintf(out," case %d:\n",rp->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003841 for(rp2=rp->next; rp2; rp2=rp2->next){
3842 if( rp2->code==rp->code ){
3843 fprintf(out," case %d:\n",rp2->index); lineno++;
3844 rp2->code = 0;
3845 }
3846 }
drh75897232000-05-29 14:26:00 +00003847 emit_code(out,rp,lemp,&lineno);
3848 fprintf(out," break;\n"); lineno++;
3849 }
3850 tplt_xfer(lemp->name,in,out,&lineno);
3851
3852 /* Generate code which executes if a parse fails */
3853 tplt_print(out,lemp,lemp->failure,lemp->failureln,&lineno);
3854 tplt_xfer(lemp->name,in,out,&lineno);
3855
3856 /* Generate code which executes when a syntax error occurs */
3857 tplt_print(out,lemp,lemp->error,lemp->errorln,&lineno);
3858 tplt_xfer(lemp->name,in,out,&lineno);
3859
3860 /* Generate code which executes when the parser accepts its input */
3861 tplt_print(out,lemp,lemp->accept,lemp->acceptln,&lineno);
3862 tplt_xfer(lemp->name,in,out,&lineno);
3863
3864 /* Append any addition code the user desires */
3865 tplt_print(out,lemp,lemp->extracode,lemp->extracodeln,&lineno);
3866
3867 fclose(in);
3868 fclose(out);
3869 return;
3870}
3871
3872/* Generate a header file for the parser */
3873void ReportHeader(lemp)
3874struct lemon *lemp;
3875{
3876 FILE *out, *in;
3877 char *prefix;
3878 char line[LINESIZE];
3879 char pattern[LINESIZE];
3880 int i;
3881
3882 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3883 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00003884 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00003885 if( in ){
3886 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
3887 sprintf(pattern,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3888 if( strcmp(line,pattern) ) break;
3889 }
3890 fclose(in);
3891 if( i==lemp->nterminal ){
3892 /* No change in the file. Don't rewrite it. */
3893 return;
3894 }
3895 }
drh2aa6ca42004-09-10 00:14:04 +00003896 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00003897 if( out ){
3898 for(i=1; i<lemp->nterminal; i++){
3899 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3900 }
3901 fclose(out);
3902 }
3903 return;
3904}
3905
3906/* Reduce the size of the action tables, if possible, by making use
3907** of defaults.
3908**
drhb59499c2002-02-23 18:45:13 +00003909** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00003910** it the default. Except, there is no default if the wildcard token
3911** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00003912*/
3913void CompressTables(lemp)
3914struct lemon *lemp;
3915{
3916 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00003917 struct action *ap, *ap2;
3918 struct rule *rp, *rp2, *rbest;
3919 int nbest, n;
drh75897232000-05-29 14:26:00 +00003920 int i;
drhe09daa92006-06-10 13:29:31 +00003921 int usesWildcard;
drh75897232000-05-29 14:26:00 +00003922
3923 for(i=0; i<lemp->nstate; i++){
3924 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00003925 nbest = 0;
3926 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00003927 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00003928
drhb59499c2002-02-23 18:45:13 +00003929 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00003930 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
3931 usesWildcard = 1;
3932 }
drhb59499c2002-02-23 18:45:13 +00003933 if( ap->type!=REDUCE ) continue;
3934 rp = ap->x.rp;
3935 if( rp==rbest ) continue;
3936 n = 1;
3937 for(ap2=ap->next; ap2; ap2=ap2->next){
3938 if( ap2->type!=REDUCE ) continue;
3939 rp2 = ap2->x.rp;
3940 if( rp2==rbest ) continue;
3941 if( rp2==rp ) n++;
3942 }
3943 if( n>nbest ){
3944 nbest = n;
3945 rbest = rp;
drh75897232000-05-29 14:26:00 +00003946 }
3947 }
drhb59499c2002-02-23 18:45:13 +00003948
3949 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00003950 ** is not at least 1 or if the wildcard token is a possible
3951 ** lookahead.
3952 */
3953 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00003954
drhb59499c2002-02-23 18:45:13 +00003955
3956 /* Combine matching REDUCE actions into a single default */
3957 for(ap=stp->ap; ap; ap=ap->next){
3958 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
3959 }
drh75897232000-05-29 14:26:00 +00003960 assert( ap );
3961 ap->sp = Symbol_new("{default}");
3962 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00003963 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00003964 }
3965 stp->ap = Action_sort(stp->ap);
3966 }
3967}
drhb59499c2002-02-23 18:45:13 +00003968
drhada354d2005-11-05 15:03:59 +00003969
3970/*
3971** Compare two states for sorting purposes. The smaller state is the
3972** one with the most non-terminal actions. If they have the same number
3973** of non-terminal actions, then the smaller is the one with the most
3974** token actions.
3975*/
3976static int stateResortCompare(const void *a, const void *b){
3977 const struct state *pA = *(const struct state**)a;
3978 const struct state *pB = *(const struct state**)b;
3979 int n;
3980
3981 n = pB->nNtAct - pA->nNtAct;
3982 if( n==0 ){
3983 n = pB->nTknAct - pA->nTknAct;
3984 }
3985 return n;
3986}
3987
3988
3989/*
3990** Renumber and resort states so that states with fewer choices
3991** occur at the end. Except, keep state 0 as the first state.
3992*/
3993void ResortStates(lemp)
3994struct lemon *lemp;
3995{
3996 int i;
3997 struct state *stp;
3998 struct action *ap;
3999
4000 for(i=0; i<lemp->nstate; i++){
4001 stp = lemp->sorted[i];
4002 stp->nTknAct = stp->nNtAct = 0;
4003 stp->iDflt = lemp->nstate + lemp->nrule;
4004 stp->iTknOfst = NO_OFFSET;
4005 stp->iNtOfst = NO_OFFSET;
4006 for(ap=stp->ap; ap; ap=ap->next){
4007 if( compute_action(lemp,ap)>=0 ){
4008 if( ap->sp->index<lemp->nterminal ){
4009 stp->nTknAct++;
4010 }else if( ap->sp->index<lemp->nsymbol ){
4011 stp->nNtAct++;
4012 }else{
4013 stp->iDflt = compute_action(lemp, ap);
4014 }
4015 }
4016 }
4017 }
4018 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4019 stateResortCompare);
4020 for(i=0; i<lemp->nstate; i++){
4021 lemp->sorted[i]->statenum = i;
4022 }
4023}
4024
4025
drh75897232000-05-29 14:26:00 +00004026/***************** From the file "set.c" ************************************/
4027/*
4028** Set manipulation routines for the LEMON parser generator.
4029*/
4030
4031static int size = 0;
4032
4033/* Set the set size */
4034void SetSize(n)
4035int n;
4036{
4037 size = n+1;
4038}
4039
4040/* Allocate a new set */
4041char *SetNew(){
4042 char *s;
4043 int i;
4044 s = (char*)malloc( size );
4045 if( s==0 ){
4046 extern void memory_error();
4047 memory_error();
4048 }
4049 for(i=0; i<size; i++) s[i] = 0;
4050 return s;
4051}
4052
4053/* Deallocate a set */
4054void SetFree(s)
4055char *s;
4056{
4057 free(s);
4058}
4059
4060/* Add a new element to the set. Return TRUE if the element was added
4061** and FALSE if it was already there. */
4062int SetAdd(s,e)
4063char *s;
4064int e;
4065{
4066 int rv;
4067 rv = s[e];
4068 s[e] = 1;
4069 return !rv;
4070}
4071
4072/* Add every element of s2 to s1. Return TRUE if s1 changes. */
4073int SetUnion(s1,s2)
4074char *s1;
4075char *s2;
4076{
4077 int i, progress;
4078 progress = 0;
4079 for(i=0; i<size; i++){
4080 if( s2[i]==0 ) continue;
4081 if( s1[i]==0 ){
4082 progress = 1;
4083 s1[i] = 1;
4084 }
4085 }
4086 return progress;
4087}
4088/********************** From the file "table.c" ****************************/
4089/*
4090** All code in this file has been automatically generated
4091** from a specification in the file
4092** "table.q"
4093** by the associative array code building program "aagen".
4094** Do not edit this file! Instead, edit the specification
4095** file, then rerun aagen.
4096*/
4097/*
4098** Code for processing tables in the LEMON parser generator.
4099*/
4100
4101PRIVATE int strhash(x)
4102char *x;
4103{
4104 int h = 0;
4105 while( *x) h = h*13 + *(x++);
4106 return h;
4107}
4108
4109/* Works like strdup, sort of. Save a string in malloced memory, but
4110** keep strings in a table so that the same string is not in more
4111** than one place.
4112*/
4113char *Strsafe(y)
4114char *y;
4115{
4116 char *z;
4117
drh916f75f2006-07-17 00:19:39 +00004118 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004119 z = Strsafe_find(y);
4120 if( z==0 && (z=malloc( strlen(y)+1 ))!=0 ){
4121 strcpy(z,y);
4122 Strsafe_insert(z);
4123 }
4124 MemoryCheck(z);
4125 return z;
4126}
4127
4128/* There is one instance of the following structure for each
4129** associative array of type "x1".
4130*/
4131struct s_x1 {
4132 int size; /* The number of available slots. */
4133 /* Must be a power of 2 greater than or */
4134 /* equal to 1 */
4135 int count; /* Number of currently slots filled */
4136 struct s_x1node *tbl; /* The data stored here */
4137 struct s_x1node **ht; /* Hash table for lookups */
4138};
4139
4140/* There is one instance of this structure for every data element
4141** in an associative array of type "x1".
4142*/
4143typedef struct s_x1node {
4144 char *data; /* The data */
4145 struct s_x1node *next; /* Next entry with the same hash */
4146 struct s_x1node **from; /* Previous link */
4147} x1node;
4148
4149/* There is only one instance of the array, which is the following */
4150static struct s_x1 *x1a;
4151
4152/* Allocate a new associative array */
4153void Strsafe_init(){
4154 if( x1a ) return;
4155 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4156 if( x1a ){
4157 x1a->size = 1024;
4158 x1a->count = 0;
4159 x1a->tbl = (x1node*)malloc(
4160 (sizeof(x1node) + sizeof(x1node*))*1024 );
4161 if( x1a->tbl==0 ){
4162 free(x1a);
4163 x1a = 0;
4164 }else{
4165 int i;
4166 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4167 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4168 }
4169 }
4170}
4171/* Insert a new record into the array. Return TRUE if successful.
4172** Prior data with the same key is NOT overwritten */
4173int Strsafe_insert(data)
4174char *data;
4175{
4176 x1node *np;
4177 int h;
4178 int ph;
4179
4180 if( x1a==0 ) return 0;
4181 ph = strhash(data);
4182 h = ph & (x1a->size-1);
4183 np = x1a->ht[h];
4184 while( np ){
4185 if( strcmp(np->data,data)==0 ){
4186 /* An existing entry with the same key is found. */
4187 /* Fail because overwrite is not allows. */
4188 return 0;
4189 }
4190 np = np->next;
4191 }
4192 if( x1a->count>=x1a->size ){
4193 /* Need to make the hash table bigger */
4194 int i,size;
4195 struct s_x1 array;
4196 array.size = size = x1a->size*2;
4197 array.count = x1a->count;
4198 array.tbl = (x1node*)malloc(
4199 (sizeof(x1node) + sizeof(x1node*))*size );
4200 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4201 array.ht = (x1node**)&(array.tbl[size]);
4202 for(i=0; i<size; i++) array.ht[i] = 0;
4203 for(i=0; i<x1a->count; i++){
4204 x1node *oldnp, *newnp;
4205 oldnp = &(x1a->tbl[i]);
4206 h = strhash(oldnp->data) & (size-1);
4207 newnp = &(array.tbl[i]);
4208 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4209 newnp->next = array.ht[h];
4210 newnp->data = oldnp->data;
4211 newnp->from = &(array.ht[h]);
4212 array.ht[h] = newnp;
4213 }
4214 free(x1a->tbl);
4215 *x1a = array;
4216 }
4217 /* Insert the new data */
4218 h = ph & (x1a->size-1);
4219 np = &(x1a->tbl[x1a->count++]);
4220 np->data = data;
4221 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4222 np->next = x1a->ht[h];
4223 x1a->ht[h] = np;
4224 np->from = &(x1a->ht[h]);
4225 return 1;
4226}
4227
4228/* Return a pointer to data assigned to the given key. Return NULL
4229** if no such key. */
4230char *Strsafe_find(key)
4231char *key;
4232{
4233 int h;
4234 x1node *np;
4235
4236 if( x1a==0 ) return 0;
4237 h = strhash(key) & (x1a->size-1);
4238 np = x1a->ht[h];
4239 while( np ){
4240 if( strcmp(np->data,key)==0 ) break;
4241 np = np->next;
4242 }
4243 return np ? np->data : 0;
4244}
4245
4246/* Return a pointer to the (terminal or nonterminal) symbol "x".
4247** Create a new symbol if this is the first time "x" has been seen.
4248*/
4249struct symbol *Symbol_new(x)
4250char *x;
4251{
4252 struct symbol *sp;
4253
4254 sp = Symbol_find(x);
4255 if( sp==0 ){
4256 sp = (struct symbol *)malloc( sizeof(struct symbol) );
4257 MemoryCheck(sp);
4258 sp->name = Strsafe(x);
4259 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4260 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004261 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004262 sp->prec = -1;
4263 sp->assoc = UNK;
4264 sp->firstset = 0;
drhb27b83a2002-08-14 23:18:57 +00004265 sp->lambda = B_FALSE;
drh75897232000-05-29 14:26:00 +00004266 sp->destructor = 0;
4267 sp->datatype = 0;
4268 Symbol_insert(sp,sp->name);
4269 }
4270 return sp;
4271}
4272
drh60d31652004-02-22 00:08:04 +00004273/* Compare two symbols for working purposes
4274**
4275** Symbols that begin with upper case letters (terminals or tokens)
4276** must sort before symbols that begin with lower case letters
4277** (non-terminals). Other than that, the order does not matter.
4278**
4279** We find experimentally that leaving the symbols in their original
4280** order (the order they appeared in the grammar file) gives the
4281** smallest parser tables in SQLite.
4282*/
4283int Symbolcmpp(struct symbol **a, struct symbol **b){
4284 int i1 = (**a).index + 10000000*((**a).name[0]>'Z');
4285 int i2 = (**b).index + 10000000*((**b).name[0]>'Z');
4286 return i1-i2;
drh75897232000-05-29 14:26:00 +00004287}
4288
4289/* There is one instance of the following structure for each
4290** associative array of type "x2".
4291*/
4292struct s_x2 {
4293 int size; /* The number of available slots. */
4294 /* Must be a power of 2 greater than or */
4295 /* equal to 1 */
4296 int count; /* Number of currently slots filled */
4297 struct s_x2node *tbl; /* The data stored here */
4298 struct s_x2node **ht; /* Hash table for lookups */
4299};
4300
4301/* There is one instance of this structure for every data element
4302** in an associative array of type "x2".
4303*/
4304typedef struct s_x2node {
4305 struct symbol *data; /* The data */
4306 char *key; /* The key */
4307 struct s_x2node *next; /* Next entry with the same hash */
4308 struct s_x2node **from; /* Previous link */
4309} x2node;
4310
4311/* There is only one instance of the array, which is the following */
4312static struct s_x2 *x2a;
4313
4314/* Allocate a new associative array */
4315void Symbol_init(){
4316 if( x2a ) return;
4317 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4318 if( x2a ){
4319 x2a->size = 128;
4320 x2a->count = 0;
4321 x2a->tbl = (x2node*)malloc(
4322 (sizeof(x2node) + sizeof(x2node*))*128 );
4323 if( x2a->tbl==0 ){
4324 free(x2a);
4325 x2a = 0;
4326 }else{
4327 int i;
4328 x2a->ht = (x2node**)&(x2a->tbl[128]);
4329 for(i=0; i<128; i++) x2a->ht[i] = 0;
4330 }
4331 }
4332}
4333/* Insert a new record into the array. Return TRUE if successful.
4334** Prior data with the same key is NOT overwritten */
4335int Symbol_insert(data,key)
4336struct symbol *data;
4337char *key;
4338{
4339 x2node *np;
4340 int h;
4341 int ph;
4342
4343 if( x2a==0 ) return 0;
4344 ph = strhash(key);
4345 h = ph & (x2a->size-1);
4346 np = x2a->ht[h];
4347 while( np ){
4348 if( strcmp(np->key,key)==0 ){
4349 /* An existing entry with the same key is found. */
4350 /* Fail because overwrite is not allows. */
4351 return 0;
4352 }
4353 np = np->next;
4354 }
4355 if( x2a->count>=x2a->size ){
4356 /* Need to make the hash table bigger */
4357 int i,size;
4358 struct s_x2 array;
4359 array.size = size = x2a->size*2;
4360 array.count = x2a->count;
4361 array.tbl = (x2node*)malloc(
4362 (sizeof(x2node) + sizeof(x2node*))*size );
4363 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4364 array.ht = (x2node**)&(array.tbl[size]);
4365 for(i=0; i<size; i++) array.ht[i] = 0;
4366 for(i=0; i<x2a->count; i++){
4367 x2node *oldnp, *newnp;
4368 oldnp = &(x2a->tbl[i]);
4369 h = strhash(oldnp->key) & (size-1);
4370 newnp = &(array.tbl[i]);
4371 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4372 newnp->next = array.ht[h];
4373 newnp->key = oldnp->key;
4374 newnp->data = oldnp->data;
4375 newnp->from = &(array.ht[h]);
4376 array.ht[h] = newnp;
4377 }
4378 free(x2a->tbl);
4379 *x2a = array;
4380 }
4381 /* Insert the new data */
4382 h = ph & (x2a->size-1);
4383 np = &(x2a->tbl[x2a->count++]);
4384 np->key = key;
4385 np->data = data;
4386 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4387 np->next = x2a->ht[h];
4388 x2a->ht[h] = np;
4389 np->from = &(x2a->ht[h]);
4390 return 1;
4391}
4392
4393/* Return a pointer to data assigned to the given key. Return NULL
4394** if no such key. */
4395struct symbol *Symbol_find(key)
4396char *key;
4397{
4398 int h;
4399 x2node *np;
4400
4401 if( x2a==0 ) return 0;
4402 h = strhash(key) & (x2a->size-1);
4403 np = x2a->ht[h];
4404 while( np ){
4405 if( strcmp(np->key,key)==0 ) break;
4406 np = np->next;
4407 }
4408 return np ? np->data : 0;
4409}
4410
4411/* Return the n-th data. Return NULL if n is out of range. */
4412struct symbol *Symbol_Nth(n)
4413int n;
4414{
4415 struct symbol *data;
4416 if( x2a && n>0 && n<=x2a->count ){
4417 data = x2a->tbl[n-1].data;
4418 }else{
4419 data = 0;
4420 }
4421 return data;
4422}
4423
4424/* Return the size of the array */
4425int Symbol_count()
4426{
4427 return x2a ? x2a->count : 0;
4428}
4429
4430/* Return an array of pointers to all data in the table.
4431** The array is obtained from malloc. Return NULL if memory allocation
4432** problems, or if the array is empty. */
4433struct symbol **Symbol_arrayof()
4434{
4435 struct symbol **array;
4436 int i,size;
4437 if( x2a==0 ) return 0;
4438 size = x2a->count;
4439 array = (struct symbol **)malloc( sizeof(struct symbol *)*size );
4440 if( array ){
4441 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4442 }
4443 return array;
4444}
4445
4446/* Compare two configurations */
4447int Configcmp(a,b)
4448struct config *a;
4449struct config *b;
4450{
4451 int x;
4452 x = a->rp->index - b->rp->index;
4453 if( x==0 ) x = a->dot - b->dot;
4454 return x;
4455}
4456
4457/* Compare two states */
4458PRIVATE int statecmp(a,b)
4459struct config *a;
4460struct config *b;
4461{
4462 int rc;
4463 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4464 rc = a->rp->index - b->rp->index;
4465 if( rc==0 ) rc = a->dot - b->dot;
4466 }
4467 if( rc==0 ){
4468 if( a ) rc = 1;
4469 if( b ) rc = -1;
4470 }
4471 return rc;
4472}
4473
4474/* Hash a state */
4475PRIVATE int statehash(a)
4476struct config *a;
4477{
4478 int h=0;
4479 while( a ){
4480 h = h*571 + a->rp->index*37 + a->dot;
4481 a = a->bp;
4482 }
4483 return h;
4484}
4485
4486/* Allocate a new state structure */
4487struct state *State_new()
4488{
4489 struct state *new;
4490 new = (struct state *)malloc( sizeof(struct state) );
4491 MemoryCheck(new);
4492 return new;
4493}
4494
4495/* There is one instance of the following structure for each
4496** associative array of type "x3".
4497*/
4498struct s_x3 {
4499 int size; /* The number of available slots. */
4500 /* Must be a power of 2 greater than or */
4501 /* equal to 1 */
4502 int count; /* Number of currently slots filled */
4503 struct s_x3node *tbl; /* The data stored here */
4504 struct s_x3node **ht; /* Hash table for lookups */
4505};
4506
4507/* There is one instance of this structure for every data element
4508** in an associative array of type "x3".
4509*/
4510typedef struct s_x3node {
4511 struct state *data; /* The data */
4512 struct config *key; /* The key */
4513 struct s_x3node *next; /* Next entry with the same hash */
4514 struct s_x3node **from; /* Previous link */
4515} x3node;
4516
4517/* There is only one instance of the array, which is the following */
4518static struct s_x3 *x3a;
4519
4520/* Allocate a new associative array */
4521void State_init(){
4522 if( x3a ) return;
4523 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4524 if( x3a ){
4525 x3a->size = 128;
4526 x3a->count = 0;
4527 x3a->tbl = (x3node*)malloc(
4528 (sizeof(x3node) + sizeof(x3node*))*128 );
4529 if( x3a->tbl==0 ){
4530 free(x3a);
4531 x3a = 0;
4532 }else{
4533 int i;
4534 x3a->ht = (x3node**)&(x3a->tbl[128]);
4535 for(i=0; i<128; i++) x3a->ht[i] = 0;
4536 }
4537 }
4538}
4539/* Insert a new record into the array. Return TRUE if successful.
4540** Prior data with the same key is NOT overwritten */
4541int State_insert(data,key)
4542struct state *data;
4543struct config *key;
4544{
4545 x3node *np;
4546 int h;
4547 int ph;
4548
4549 if( x3a==0 ) return 0;
4550 ph = statehash(key);
4551 h = ph & (x3a->size-1);
4552 np = x3a->ht[h];
4553 while( np ){
4554 if( statecmp(np->key,key)==0 ){
4555 /* An existing entry with the same key is found. */
4556 /* Fail because overwrite is not allows. */
4557 return 0;
4558 }
4559 np = np->next;
4560 }
4561 if( x3a->count>=x3a->size ){
4562 /* Need to make the hash table bigger */
4563 int i,size;
4564 struct s_x3 array;
4565 array.size = size = x3a->size*2;
4566 array.count = x3a->count;
4567 array.tbl = (x3node*)malloc(
4568 (sizeof(x3node) + sizeof(x3node*))*size );
4569 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4570 array.ht = (x3node**)&(array.tbl[size]);
4571 for(i=0; i<size; i++) array.ht[i] = 0;
4572 for(i=0; i<x3a->count; i++){
4573 x3node *oldnp, *newnp;
4574 oldnp = &(x3a->tbl[i]);
4575 h = statehash(oldnp->key) & (size-1);
4576 newnp = &(array.tbl[i]);
4577 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4578 newnp->next = array.ht[h];
4579 newnp->key = oldnp->key;
4580 newnp->data = oldnp->data;
4581 newnp->from = &(array.ht[h]);
4582 array.ht[h] = newnp;
4583 }
4584 free(x3a->tbl);
4585 *x3a = array;
4586 }
4587 /* Insert the new data */
4588 h = ph & (x3a->size-1);
4589 np = &(x3a->tbl[x3a->count++]);
4590 np->key = key;
4591 np->data = data;
4592 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4593 np->next = x3a->ht[h];
4594 x3a->ht[h] = np;
4595 np->from = &(x3a->ht[h]);
4596 return 1;
4597}
4598
4599/* Return a pointer to data assigned to the given key. Return NULL
4600** if no such key. */
4601struct state *State_find(key)
4602struct config *key;
4603{
4604 int h;
4605 x3node *np;
4606
4607 if( x3a==0 ) return 0;
4608 h = statehash(key) & (x3a->size-1);
4609 np = x3a->ht[h];
4610 while( np ){
4611 if( statecmp(np->key,key)==0 ) break;
4612 np = np->next;
4613 }
4614 return np ? np->data : 0;
4615}
4616
4617/* Return an array of pointers to all data in the table.
4618** The array is obtained from malloc. Return NULL if memory allocation
4619** problems, or if the array is empty. */
4620struct state **State_arrayof()
4621{
4622 struct state **array;
4623 int i,size;
4624 if( x3a==0 ) return 0;
4625 size = x3a->count;
4626 array = (struct state **)malloc( sizeof(struct state *)*size );
4627 if( array ){
4628 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4629 }
4630 return array;
4631}
4632
4633/* Hash a configuration */
4634PRIVATE int confighash(a)
4635struct config *a;
4636{
4637 int h=0;
4638 h = h*571 + a->rp->index*37 + a->dot;
4639 return h;
4640}
4641
4642/* There is one instance of the following structure for each
4643** associative array of type "x4".
4644*/
4645struct s_x4 {
4646 int size; /* The number of available slots. */
4647 /* Must be a power of 2 greater than or */
4648 /* equal to 1 */
4649 int count; /* Number of currently slots filled */
4650 struct s_x4node *tbl; /* The data stored here */
4651 struct s_x4node **ht; /* Hash table for lookups */
4652};
4653
4654/* There is one instance of this structure for every data element
4655** in an associative array of type "x4".
4656*/
4657typedef struct s_x4node {
4658 struct config *data; /* The data */
4659 struct s_x4node *next; /* Next entry with the same hash */
4660 struct s_x4node **from; /* Previous link */
4661} x4node;
4662
4663/* There is only one instance of the array, which is the following */
4664static struct s_x4 *x4a;
4665
4666/* Allocate a new associative array */
4667void Configtable_init(){
4668 if( x4a ) return;
4669 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4670 if( x4a ){
4671 x4a->size = 64;
4672 x4a->count = 0;
4673 x4a->tbl = (x4node*)malloc(
4674 (sizeof(x4node) + sizeof(x4node*))*64 );
4675 if( x4a->tbl==0 ){
4676 free(x4a);
4677 x4a = 0;
4678 }else{
4679 int i;
4680 x4a->ht = (x4node**)&(x4a->tbl[64]);
4681 for(i=0; i<64; i++) x4a->ht[i] = 0;
4682 }
4683 }
4684}
4685/* Insert a new record into the array. Return TRUE if successful.
4686** Prior data with the same key is NOT overwritten */
4687int Configtable_insert(data)
4688struct config *data;
4689{
4690 x4node *np;
4691 int h;
4692 int ph;
4693
4694 if( x4a==0 ) return 0;
4695 ph = confighash(data);
4696 h = ph & (x4a->size-1);
4697 np = x4a->ht[h];
4698 while( np ){
4699 if( Configcmp(np->data,data)==0 ){
4700 /* An existing entry with the same key is found. */
4701 /* Fail because overwrite is not allows. */
4702 return 0;
4703 }
4704 np = np->next;
4705 }
4706 if( x4a->count>=x4a->size ){
4707 /* Need to make the hash table bigger */
4708 int i,size;
4709 struct s_x4 array;
4710 array.size = size = x4a->size*2;
4711 array.count = x4a->count;
4712 array.tbl = (x4node*)malloc(
4713 (sizeof(x4node) + sizeof(x4node*))*size );
4714 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4715 array.ht = (x4node**)&(array.tbl[size]);
4716 for(i=0; i<size; i++) array.ht[i] = 0;
4717 for(i=0; i<x4a->count; i++){
4718 x4node *oldnp, *newnp;
4719 oldnp = &(x4a->tbl[i]);
4720 h = confighash(oldnp->data) & (size-1);
4721 newnp = &(array.tbl[i]);
4722 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4723 newnp->next = array.ht[h];
4724 newnp->data = oldnp->data;
4725 newnp->from = &(array.ht[h]);
4726 array.ht[h] = newnp;
4727 }
4728 free(x4a->tbl);
4729 *x4a = array;
4730 }
4731 /* Insert the new data */
4732 h = ph & (x4a->size-1);
4733 np = &(x4a->tbl[x4a->count++]);
4734 np->data = data;
4735 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
4736 np->next = x4a->ht[h];
4737 x4a->ht[h] = np;
4738 np->from = &(x4a->ht[h]);
4739 return 1;
4740}
4741
4742/* Return a pointer to data assigned to the given key. Return NULL
4743** if no such key. */
4744struct config *Configtable_find(key)
4745struct config *key;
4746{
4747 int h;
4748 x4node *np;
4749
4750 if( x4a==0 ) return 0;
4751 h = confighash(key) & (x4a->size-1);
4752 np = x4a->ht[h];
4753 while( np ){
4754 if( Configcmp(np->data,key)==0 ) break;
4755 np = np->next;
4756 }
4757 return np ? np->data : 0;
4758}
4759
4760/* Remove all data from the table. Pass each data to the function "f"
4761** as it is removed. ("f" may be null to avoid this step.) */
4762void Configtable_clear(f)
4763int(*f)(/* struct config * */);
4764{
4765 int i;
4766 if( x4a==0 || x4a->count==0 ) return;
4767 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
4768 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
4769 x4a->count = 0;
4770 return;
4771}