blob: 8f6e87330a53107940d7f586afb7c1b3c200f502 [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 * */);
97
98/********** From the file "set.h" ****************************************/
99void SetSize(/* int N */); /* All sets will be of size N */
100char *SetNew(/* void */); /* A new set for element 0..N */
101void SetFree(/* char* */); /* Deallocate a set */
102
103int SetAdd(/* char*,int */); /* Add element to a set */
104int SetUnion(/* char *A,char *B */); /* A <- A U B, thru element N */
105
106#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
107
108/********** From the file "struct.h" *************************************/
109/*
110** Principal data structures for the LEMON parser generator.
111*/
112
drhb27b83a2002-08-14 23:18:57 +0000113typedef enum {B_FALSE=0, B_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000114
115/* Symbols (terminals and nonterminals) of the grammar are stored
116** in the following: */
117struct symbol {
118 char *name; /* Name of the symbol */
119 int index; /* Index number for this symbol */
120 enum {
121 TERMINAL,
122 NONTERMINAL
123 } type; /* Symbols are all either TERMINALS or NTs */
124 struct rule *rule; /* Linked list of rules of this (if an NT) */
drh0bd1f4e2002-06-06 18:54:39 +0000125 struct symbol *fallback; /* fallback token in case this token doesn't parse */
drh75897232000-05-29 14:26:00 +0000126 int prec; /* Precedence if defined (-1 otherwise) */
127 enum e_assoc {
128 LEFT,
129 RIGHT,
130 NONE,
131 UNK
132 } assoc; /* Associativity if predecence is defined */
133 char *firstset; /* First-set for all rules of this symbol */
134 Boolean lambda; /* True if NT and can generate an empty string */
135 char *destructor; /* Code which executes whenever this symbol is
136 ** popped from the stack during error processing */
137 int destructorln; /* Line number of destructor code */
138 char *datatype; /* The data type of information held by this
139 ** object. Only used if type==NONTERMINAL */
140 int dtnum; /* The data type number. In the parser, the value
141 ** stack is a union. The .yy%d element of this
142 ** union is the correct data type for this object */
143};
144
145/* Each production rule in the grammar is stored in the following
146** structure. */
147struct rule {
148 struct symbol *lhs; /* Left-hand side of the rule */
149 char *lhsalias; /* Alias for the LHS (NULL if none) */
150 int ruleline; /* Line number for the rule */
151 int nrhs; /* Number of RHS symbols */
152 struct symbol **rhs; /* The RHS symbols */
153 char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
154 int line; /* Line number at which code begins */
155 char *code; /* The code executed when this rule is reduced */
156 struct symbol *precsym; /* Precedence symbol for this rule */
157 int index; /* An index number for this rule */
158 Boolean canReduce; /* True if this rule is ever reduced */
159 struct rule *nextlhs; /* Next rule with the same LHS */
160 struct rule *next; /* Next rule in the global list */
161};
162
163/* A configuration is a production rule of the grammar together with
164** a mark (dot) showing how much of that rule has been processed so far.
165** Configurations also contain a follow-set which is a list of terminal
166** symbols which are allowed to immediately follow the end of the rule.
167** Every configuration is recorded as an instance of the following: */
168struct config {
169 struct rule *rp; /* The rule upon which the configuration is based */
170 int dot; /* The parse point */
171 char *fws; /* Follow-set for this configuration only */
172 struct plink *fplp; /* Follow-set forward propagation links */
173 struct plink *bplp; /* Follow-set backwards propagation links */
174 struct state *stp; /* Pointer to state which contains this */
175 enum {
176 COMPLETE, /* The status is used during followset and */
177 INCOMPLETE /* shift computations */
178 } status;
179 struct config *next; /* Next configuration in the state */
180 struct config *bp; /* The next basis configuration */
181};
182
183/* Every shift or reduce operation is stored as one of the following */
184struct action {
185 struct symbol *sp; /* The look-ahead symbol */
186 enum e_action {
187 SHIFT,
188 ACCEPT,
189 REDUCE,
190 ERROR,
191 CONFLICT, /* Was a reduce, but part of a conflict */
192 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
193 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
194 NOT_USED /* Deleted by compression */
195 } type;
196 union {
197 struct state *stp; /* The new state, if a shift */
198 struct rule *rp; /* The rule, if a reduce */
199 } x;
200 struct action *next; /* Next action for this state */
201 struct action *collide; /* Next action with the same hash */
202};
203
204/* Each state of the generated parser's finite state machine
205** is encoded as an instance of the following structure. */
206struct state {
207 struct config *bp; /* The basis configurations for this state */
208 struct config *cfp; /* All configurations in this set */
209 int index; /* Sequencial number for this state */
210 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000211 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
212 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
213 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000214};
drh8b582012003-10-21 13:16:03 +0000215#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000216
217/* A followset propagation link indicates that the contents of one
218** configuration followset should be propagated to another whenever
219** the first changes. */
220struct plink {
221 struct config *cfp; /* The configuration to which linked */
222 struct plink *next; /* The next propagate link */
223};
224
225/* The state vector for the entire parser generator is recorded as
226** follows. (LEMON uses no global variables and makes little use of
227** static variables. Fields in the following structure can be thought
228** of as begin global variables in the program.) */
229struct lemon {
230 struct state **sorted; /* Table of states sorted by state number */
231 struct rule *rule; /* List of all rules */
232 int nstate; /* Number of states */
233 int nrule; /* Number of rules */
234 int nsymbol; /* Number of terminal and nonterminal symbols */
235 int nterminal; /* Number of terminal symbols */
236 struct symbol **symbols; /* Sorted array of pointers to symbols */
237 int errorcnt; /* Number of errors */
238 struct symbol *errsym; /* The error symbol */
239 char *name; /* Name of the generated parser */
240 char *arg; /* Declaration of the 3th argument to parser */
241 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000242 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000243 char *start; /* Name of the start symbol for the grammar */
244 char *stacksize; /* Size of the parser stack */
245 char *include; /* Code to put at the start of the C file */
246 int includeln; /* Line number for start of include code */
247 char *error; /* Code to execute when an error is seen */
248 int errorln; /* Line number for start of error code */
249 char *overflow; /* Code to execute on a stack overflow */
250 int overflowln; /* Line number for start of overflow code */
251 char *failure; /* Code to execute on parser failure */
252 int failureln; /* Line number for start of failure code */
253 char *accept; /* Code to execute when the parser excepts */
254 int acceptln; /* Line number for the start of accept code */
255 char *extracode; /* Code appended to the generated file */
256 int extracodeln; /* Line number for the start of the extra code */
257 char *tokendest; /* Code to execute to destroy token data */
258 int tokendestln; /* Line number for token destroyer code */
drh960e8c62001-04-03 16:53:21 +0000259 char *vardest; /* Code for the default non-terminal destructor */
260 int vardestln; /* Line number for default non-term destructor code*/
drh75897232000-05-29 14:26:00 +0000261 char *filename; /* Name of the input file */
262 char *outname; /* Name of the current output file */
263 char *tokenprefix; /* A prefix added to token names in the .h file */
264 int nconflict; /* Number of parsing conflicts */
265 int tablesize; /* Size of the parse tables */
266 int basisflag; /* Print only basis configurations */
drh0bd1f4e2002-06-06 18:54:39 +0000267 int has_fallback; /* True if any %fallback is seen in the grammer */
drh75897232000-05-29 14:26:00 +0000268 char *argv0; /* Name of the program */
269};
270
271#define MemoryCheck(X) if((X)==0){ \
272 extern void memory_error(); \
273 memory_error(); \
274}
275
276/**************** From the file "table.h" *********************************/
277/*
278** All code in this file has been automatically generated
279** from a specification in the file
280** "table.q"
281** by the associative array code building program "aagen".
282** Do not edit this file! Instead, edit the specification
283** file, then rerun aagen.
284*/
285/*
286** Code for processing tables in the LEMON parser generator.
287*/
288
289/* Routines for handling a strings */
290
291char *Strsafe();
292
293void Strsafe_init(/* void */);
294int Strsafe_insert(/* char * */);
295char *Strsafe_find(/* char * */);
296
297/* Routines for handling symbols of the grammar */
298
299struct symbol *Symbol_new();
300int Symbolcmpp(/* struct symbol **, struct symbol ** */);
301void Symbol_init(/* void */);
302int Symbol_insert(/* struct symbol *, char * */);
303struct symbol *Symbol_find(/* char * */);
304struct symbol *Symbol_Nth(/* int */);
305int Symbol_count(/* */);
306struct symbol **Symbol_arrayof(/* */);
307
308/* Routines to manage the state table */
309
310int Configcmp(/* struct config *, struct config * */);
311struct state *State_new();
312void State_init(/* void */);
313int State_insert(/* struct state *, struct config * */);
314struct state *State_find(/* struct config * */);
315struct state **State_arrayof(/* */);
316
317/* Routines used for efficiency in Configlist_add */
318
319void Configtable_init(/* void */);
320int Configtable_insert(/* struct config * */);
321struct config *Configtable_find(/* struct config * */);
322void Configtable_clear(/* int(*)(struct config *) */);
323/****************** From the file "action.c" *******************************/
324/*
325** Routines processing parser actions in the LEMON parser generator.
326*/
327
328/* Allocate a new parser action */
329struct action *Action_new(){
330 static struct action *freelist = 0;
331 struct action *new;
332
333 if( freelist==0 ){
334 int i;
335 int amt = 100;
336 freelist = (struct action *)malloc( sizeof(struct action)*amt );
337 if( freelist==0 ){
338 fprintf(stderr,"Unable to allocate memory for a new parser action.");
339 exit(1);
340 }
341 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
342 freelist[amt-1].next = 0;
343 }
344 new = freelist;
345 freelist = freelist->next;
346 return new;
347}
348
349/* Compare two actions */
350static int actioncmp(ap1,ap2)
351struct action *ap1;
352struct action *ap2;
353{
354 int rc;
355 rc = ap1->sp->index - ap2->sp->index;
356 if( rc==0 ) rc = (int)ap1->type - (int)ap2->type;
357 if( rc==0 ){
drh61bc2722000-08-20 11:42:46 +0000358 assert( ap1->type==REDUCE || ap1->type==RD_RESOLVED || ap1->type==CONFLICT);
359 assert( ap2->type==REDUCE || ap2->type==RD_RESOLVED || ap2->type==CONFLICT);
drh75897232000-05-29 14:26:00 +0000360 rc = ap1->x.rp->index - ap2->x.rp->index;
361 }
362 return rc;
363}
364
365/* Sort parser actions */
366struct action *Action_sort(ap)
367struct action *ap;
368{
drh218dc692004-05-31 23:13:45 +0000369 ap = (struct action *)msort((char *)ap,(char **)&ap->next,actioncmp);
drh75897232000-05-29 14:26:00 +0000370 return ap;
371}
372
373void Action_add(app,type,sp,arg)
374struct action **app;
375enum e_action type;
376struct symbol *sp;
377char *arg;
378{
379 struct action *new;
380 new = Action_new();
381 new->next = *app;
382 *app = new;
383 new->type = type;
384 new->sp = sp;
385 if( type==SHIFT ){
386 new->x.stp = (struct state *)arg;
387 }else{
388 new->x.rp = (struct rule *)arg;
389 }
390}
drh8b582012003-10-21 13:16:03 +0000391/********************** New code to implement the "acttab" module ***********/
392/*
393** This module implements routines use to construct the yy_action[] table.
394*/
395
396/*
397** The state of the yy_action table under construction is an instance of
398** the following structure
399*/
400typedef struct acttab acttab;
401struct acttab {
402 int nAction; /* Number of used slots in aAction[] */
403 int nActionAlloc; /* Slots allocated for aAction[] */
404 struct {
405 int lookahead; /* Value of the lookahead token */
406 int action; /* Action to take on the given lookahead */
407 } *aAction, /* The yy_action[] table under construction */
408 *aLookahead; /* A single new transaction set */
409 int mnLookahead; /* Minimum aLookahead[].lookahead */
410 int mnAction; /* Action associated with mnLookahead */
411 int mxLookahead; /* Maximum aLookahead[].lookahead */
412 int nLookahead; /* Used slots in aLookahead[] */
413 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
414};
415
416/* Return the number of entries in the yy_action table */
417#define acttab_size(X) ((X)->nAction)
418
419/* The value for the N-th entry in yy_action */
420#define acttab_yyaction(X,N) ((X)->aAction[N].action)
421
422/* The value for the N-th entry in yy_lookahead */
423#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
424
425/* Free all memory associated with the given acttab */
426void acttab_free(acttab *p){
427 free( p->aAction );
428 free( p->aLookahead );
429 free( p );
430}
431
432/* Allocate a new acttab structure */
433acttab *acttab_alloc(void){
434 acttab *p = malloc( sizeof(*p) );
435 if( p==0 ){
436 fprintf(stderr,"Unable to allocate memory for a new acttab.");
437 exit(1);
438 }
439 memset(p, 0, sizeof(*p));
440 return p;
441}
442
443/* Add a new action to the current transaction set
444*/
445void acttab_action(acttab *p, int lookahead, int action){
446 if( p->nLookahead>=p->nLookaheadAlloc ){
447 p->nLookaheadAlloc += 25;
448 p->aLookahead = realloc( p->aLookahead,
449 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
450 if( p->aLookahead==0 ){
451 fprintf(stderr,"malloc failed\n");
452 exit(1);
453 }
454 }
455 if( p->nLookahead==0 ){
456 p->mxLookahead = lookahead;
457 p->mnLookahead = lookahead;
458 p->mnAction = action;
459 }else{
460 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
461 if( p->mnLookahead>lookahead ){
462 p->mnLookahead = lookahead;
463 p->mnAction = action;
464 }
465 }
466 p->aLookahead[p->nLookahead].lookahead = lookahead;
467 p->aLookahead[p->nLookahead].action = action;
468 p->nLookahead++;
469}
470
471/*
472** Add the transaction set built up with prior calls to acttab_action()
473** into the current action table. Then reset the transaction set back
474** to an empty set in preparation for a new round of acttab_action() calls.
475**
476** Return the offset into the action table of the new transaction.
477*/
478int acttab_insert(acttab *p){
479 int i, j, k, n;
480 assert( p->nLookahead>0 );
481
482 /* Make sure we have enough space to hold the expanded action table
483 ** in the worst case. The worst case occurs if the transaction set
484 ** must be appended to the current action table
485 */
drh784d86f2004-02-19 18:41:53 +0000486 n = p->mxLookahead + 1;
drh8b582012003-10-21 13:16:03 +0000487 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000488 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000489 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
490 p->aAction = realloc( p->aAction,
491 sizeof(p->aAction[0])*p->nActionAlloc);
492 if( p->aAction==0 ){
493 fprintf(stderr,"malloc failed\n");
494 exit(1);
495 }
drhfdbf9282003-10-21 16:34:41 +0000496 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000497 p->aAction[i].lookahead = -1;
498 p->aAction[i].action = -1;
499 }
500 }
501
502 /* Scan the existing action table looking for an offset where we can
503 ** insert the current transaction set. Fall out of the loop when that
504 ** offset is found. In the worst case, we fall out of the loop when
505 ** i reaches p->nAction, which means we append the new transaction set.
506 **
507 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
508 */
drh784d86f2004-02-19 18:41:53 +0000509 for(i=0; i<p->nAction+p->mnLookahead; i++){
drh8b582012003-10-21 13:16:03 +0000510 if( p->aAction[i].lookahead<0 ){
511 for(j=0; j<p->nLookahead; j++){
512 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
513 if( k<0 ) break;
514 if( p->aAction[k].lookahead>=0 ) break;
515 }
drhfdbf9282003-10-21 16:34:41 +0000516 if( j<p->nLookahead ) continue;
517 for(j=0; j<p->nAction; j++){
518 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
519 }
520 if( j==p->nAction ){
521 break; /* Fits in empty slots */
522 }
drh8b582012003-10-21 13:16:03 +0000523 }else if( p->aAction[i].lookahead==p->mnLookahead ){
524 if( p->aAction[i].action!=p->mnAction ) continue;
525 for(j=0; j<p->nLookahead; j++){
526 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
527 if( k<0 || k>=p->nAction ) break;
528 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
529 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
530 }
531 if( j<p->nLookahead ) continue;
532 n = 0;
533 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000534 if( p->aAction[j].lookahead<0 ) continue;
535 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000536 }
drhfdbf9282003-10-21 16:34:41 +0000537 if( n==p->nLookahead ){
538 break; /* Same as a prior transaction set */
539 }
drh8b582012003-10-21 13:16:03 +0000540 }
541 }
542 /* Insert transaction set at index i. */
543 for(j=0; j<p->nLookahead; j++){
544 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
545 p->aAction[k] = p->aLookahead[j];
546 if( k>=p->nAction ) p->nAction = k+1;
547 }
548 p->nLookahead = 0;
549
550 /* Return the offset that is added to the lookahead in order to get the
551 ** index into yy_action of the action */
552 return i - p->mnLookahead;
553}
554
drh75897232000-05-29 14:26:00 +0000555/********************** From the file "assert.c" ****************************/
556/*
557** A more efficient way of handling assertions.
558*/
559void myassert(file,line)
560char *file;
561int line;
562{
563 fprintf(stderr,"Assertion failed on line %d of file \"%s\"\n",line,file);
564 exit(1);
565}
566/********************** From the file "build.c" *****************************/
567/*
568** Routines to construction the finite state machine for the LEMON
569** parser generator.
570*/
571
572/* Find a precedence symbol of every rule in the grammar.
573**
574** Those rules which have a precedence symbol coded in the input
575** grammar using the "[symbol]" construct will already have the
576** rp->precsym field filled. Other rules take as their precedence
577** symbol the first RHS symbol with a defined precedence. If there
578** are not RHS symbols with a defined precedence, the precedence
579** symbol field is left blank.
580*/
581void FindRulePrecedences(xp)
582struct lemon *xp;
583{
584 struct rule *rp;
585 for(rp=xp->rule; rp; rp=rp->next){
586 if( rp->precsym==0 ){
587 int i;
588 for(i=0; i<rp->nrhs; i++){
589 if( rp->rhs[i]->prec>=0 ){
590 rp->precsym = rp->rhs[i];
591 break;
592 }
593 }
594 }
595 }
596 return;
597}
598
599/* Find all nonterminals which will generate the empty string.
600** Then go back and compute the first sets of every nonterminal.
601** The first set is the set of all terminal symbols which can begin
602** a string generated by that nonterminal.
603*/
604void FindFirstSets(lemp)
605struct lemon *lemp;
606{
607 int i;
608 struct rule *rp;
609 int progress;
610
611 for(i=0; i<lemp->nsymbol; i++){
drhb27b83a2002-08-14 23:18:57 +0000612 lemp->symbols[i]->lambda = B_FALSE;
drh75897232000-05-29 14:26:00 +0000613 }
614 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
615 lemp->symbols[i]->firstset = SetNew();
616 }
617
618 /* First compute all lambdas */
619 do{
620 progress = 0;
621 for(rp=lemp->rule; rp; rp=rp->next){
622 if( rp->lhs->lambda ) continue;
623 for(i=0; i<rp->nrhs; i++){
drhb27b83a2002-08-14 23:18:57 +0000624 if( rp->rhs[i]->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000625 }
626 if( i==rp->nrhs ){
drhb27b83a2002-08-14 23:18:57 +0000627 rp->lhs->lambda = B_TRUE;
drh75897232000-05-29 14:26:00 +0000628 progress = 1;
629 }
630 }
631 }while( progress );
632
633 /* Now compute all first sets */
634 do{
635 struct symbol *s1, *s2;
636 progress = 0;
637 for(rp=lemp->rule; rp; rp=rp->next){
638 s1 = rp->lhs;
639 for(i=0; i<rp->nrhs; i++){
640 s2 = rp->rhs[i];
641 if( s2->type==TERMINAL ){
642 progress += SetAdd(s1->firstset,s2->index);
643 break;
644 }else if( s1==s2 ){
drhb27b83a2002-08-14 23:18:57 +0000645 if( s1->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000646 }else{
647 progress += SetUnion(s1->firstset,s2->firstset);
drhb27b83a2002-08-14 23:18:57 +0000648 if( s2->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000649 }
650 }
651 }
652 }while( progress );
653 return;
654}
655
656/* Compute all LR(0) states for the grammar. Links
657** are added to between some states so that the LR(1) follow sets
658** can be computed later.
659*/
660PRIVATE struct state *getstate(/* struct lemon * */); /* forward reference */
661void FindStates(lemp)
662struct lemon *lemp;
663{
664 struct symbol *sp;
665 struct rule *rp;
666
667 Configlist_init();
668
669 /* Find the start symbol */
670 if( lemp->start ){
671 sp = Symbol_find(lemp->start);
672 if( sp==0 ){
673 ErrorMsg(lemp->filename,0,
674"The specified start symbol \"%s\" is not \
675in a nonterminal of the grammar. \"%s\" will be used as the start \
676symbol instead.",lemp->start,lemp->rule->lhs->name);
677 lemp->errorcnt++;
678 sp = lemp->rule->lhs;
679 }
680 }else{
681 sp = lemp->rule->lhs;
682 }
683
684 /* Make sure the start symbol doesn't occur on the right-hand side of
685 ** any rule. Report an error if it does. (YACC would generate a new
686 ** start symbol in this case.) */
687 for(rp=lemp->rule; rp; rp=rp->next){
688 int i;
689 for(i=0; i<rp->nrhs; i++){
690 if( rp->rhs[i]==sp ){
691 ErrorMsg(lemp->filename,0,
692"The start symbol \"%s\" occurs on the \
693right-hand side of a rule. This will result in a parser which \
694does not work properly.",sp->name);
695 lemp->errorcnt++;
696 }
697 }
698 }
699
700 /* The basis configuration set for the first state
701 ** is all rules which have the start symbol as their
702 ** left-hand side */
703 for(rp=sp->rule; rp; rp=rp->nextlhs){
704 struct config *newcfp;
705 newcfp = Configlist_addbasis(rp,0);
706 SetAdd(newcfp->fws,0);
707 }
708
709 /* Compute the first state. All other states will be
710 ** computed automatically during the computation of the first one.
711 ** The returned pointer to the first state is not used. */
712 (void)getstate(lemp);
713 return;
714}
715
716/* Return a pointer to a state which is described by the configuration
717** list which has been built from calls to Configlist_add.
718*/
719PRIVATE void buildshifts(/* struct lemon *, struct state * */); /* Forwd ref */
720PRIVATE struct state *getstate(lemp)
721struct lemon *lemp;
722{
723 struct config *cfp, *bp;
724 struct state *stp;
725
726 /* Extract the sorted basis of the new state. The basis was constructed
727 ** by prior calls to "Configlist_addbasis()". */
728 Configlist_sortbasis();
729 bp = Configlist_basis();
730
731 /* Get a state with the same basis */
732 stp = State_find(bp);
733 if( stp ){
734 /* A state with the same basis already exists! Copy all the follow-set
735 ** propagation links from the state under construction into the
736 ** preexisting state, then return a pointer to the preexisting state */
737 struct config *x, *y;
738 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
739 Plink_copy(&y->bplp,x->bplp);
740 Plink_delete(x->fplp);
741 x->fplp = x->bplp = 0;
742 }
743 cfp = Configlist_return();
744 Configlist_eat(cfp);
745 }else{
746 /* This really is a new state. Construct all the details */
747 Configlist_closure(lemp); /* Compute the configuration closure */
748 Configlist_sort(); /* Sort the configuration closure */
749 cfp = Configlist_return(); /* Get a pointer to the config list */
750 stp = State_new(); /* A new state structure */
751 MemoryCheck(stp);
752 stp->bp = bp; /* Remember the configuration basis */
753 stp->cfp = cfp; /* Remember the configuration closure */
754 stp->index = lemp->nstate++; /* Every state gets a sequence number */
755 stp->ap = 0; /* No actions, yet. */
756 State_insert(stp,stp->bp); /* Add to the state table */
757 buildshifts(lemp,stp); /* Recursively compute successor states */
758 }
759 return stp;
760}
761
762/* Construct all successor states to the given state. A "successor"
763** state is any state which can be reached by a shift action.
764*/
765PRIVATE void buildshifts(lemp,stp)
766struct lemon *lemp;
767struct state *stp; /* The state from which successors are computed */
768{
769 struct config *cfp; /* For looping thru the config closure of "stp" */
770 struct config *bcfp; /* For the inner loop on config closure of "stp" */
771 struct config *new; /* */
772 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
773 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
774 struct state *newstp; /* A pointer to a successor state */
775
776 /* Each configuration becomes complete after it contibutes to a successor
777 ** state. Initially, all configurations are incomplete */
778 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
779
780 /* Loop through all configurations of the state "stp" */
781 for(cfp=stp->cfp; cfp; cfp=cfp->next){
782 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
783 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
784 Configlist_reset(); /* Reset the new config set */
785 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
786
787 /* For every configuration in the state "stp" which has the symbol "sp"
788 ** following its dot, add the same configuration to the basis set under
789 ** construction but with the dot shifted one symbol to the right. */
790 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
791 if( bcfp->status==COMPLETE ) continue; /* Already used */
792 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
793 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
794 if( bsp!=sp ) continue; /* Must be same as for "cfp" */
795 bcfp->status = COMPLETE; /* Mark this config as used */
796 new = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
797 Plink_add(&new->bplp,bcfp);
798 }
799
800 /* Get a pointer to the state described by the basis configuration set
801 ** constructed in the preceding loop */
802 newstp = getstate(lemp);
803
804 /* The state "newstp" is reached from the state "stp" by a shift action
805 ** on the symbol "sp" */
drh218dc692004-05-31 23:13:45 +0000806 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
drh75897232000-05-29 14:26:00 +0000807 }
808}
809
810/*
811** Construct the propagation links
812*/
813void FindLinks(lemp)
814struct lemon *lemp;
815{
816 int i;
817 struct config *cfp, *other;
818 struct state *stp;
819 struct plink *plp;
820
821 /* Housekeeping detail:
822 ** Add to every propagate link a pointer back to the state to
823 ** which the link is attached. */
824 for(i=0; i<lemp->nstate; i++){
825 stp = lemp->sorted[i];
826 for(cfp=stp->cfp; cfp; cfp=cfp->next){
827 cfp->stp = stp;
828 }
829 }
830
831 /* Convert all backlinks into forward links. Only the forward
832 ** links are used in the follow-set computation. */
833 for(i=0; i<lemp->nstate; i++){
834 stp = lemp->sorted[i];
835 for(cfp=stp->cfp; cfp; cfp=cfp->next){
836 for(plp=cfp->bplp; plp; plp=plp->next){
837 other = plp->cfp;
838 Plink_add(&other->fplp,cfp);
839 }
840 }
841 }
842}
843
844/* Compute all followsets.
845**
846** A followset is the set of all symbols which can come immediately
847** after a configuration.
848*/
849void FindFollowSets(lemp)
850struct lemon *lemp;
851{
852 int i;
853 struct config *cfp;
854 struct plink *plp;
855 int progress;
856 int change;
857
858 for(i=0; i<lemp->nstate; i++){
859 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
860 cfp->status = INCOMPLETE;
861 }
862 }
863
864 do{
865 progress = 0;
866 for(i=0; i<lemp->nstate; i++){
867 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
868 if( cfp->status==COMPLETE ) continue;
869 for(plp=cfp->fplp; plp; plp=plp->next){
870 change = SetUnion(plp->cfp->fws,cfp->fws);
871 if( change ){
872 plp->cfp->status = INCOMPLETE;
873 progress = 1;
874 }
875 }
876 cfp->status = COMPLETE;
877 }
878 }
879 }while( progress );
880}
881
882static int resolve_conflict();
883
884/* Compute the reduce actions, and resolve conflicts.
885*/
886void FindActions(lemp)
887struct lemon *lemp;
888{
889 int i,j;
890 struct config *cfp;
891 struct state *stp;
892 struct symbol *sp;
893 struct rule *rp;
894
895 /* Add all of the reduce actions
896 ** A reduce action is added for each element of the followset of
897 ** a configuration which has its dot at the extreme right.
898 */
899 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
900 stp = lemp->sorted[i];
901 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
902 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
903 for(j=0; j<lemp->nterminal; j++){
904 if( SetFind(cfp->fws,j) ){
905 /* Add a reduce action to the state "stp" which will reduce by the
906 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +0000907 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +0000908 }
909 }
910 }
911 }
912 }
913
914 /* Add the accepting token */
915 if( lemp->start ){
916 sp = Symbol_find(lemp->start);
917 if( sp==0 ) sp = lemp->rule->lhs;
918 }else{
919 sp = lemp->rule->lhs;
920 }
921 /* Add to the first state (which is always the starting state of the
922 ** finite state machine) an action to ACCEPT if the lookahead is the
923 ** start nonterminal. */
924 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
925
926 /* Resolve conflicts */
927 for(i=0; i<lemp->nstate; i++){
928 struct action *ap, *nap;
929 struct state *stp;
930 stp = lemp->sorted[i];
931 assert( stp->ap );
932 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +0000933 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +0000934 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
935 /* The two actions "ap" and "nap" have the same lookahead.
936 ** Figure out which one should be used */
937 lemp->nconflict += resolve_conflict(ap,nap,lemp->errsym);
938 }
939 }
940 }
941
942 /* Report an error for each rule that can never be reduced. */
drhb27b83a2002-08-14 23:18:57 +0000943 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = B_FALSE;
drh75897232000-05-29 14:26:00 +0000944 for(i=0; i<lemp->nstate; i++){
945 struct action *ap;
946 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhb27b83a2002-08-14 23:18:57 +0000947 if( ap->type==REDUCE ) ap->x.rp->canReduce = B_TRUE;
drh75897232000-05-29 14:26:00 +0000948 }
949 }
950 for(rp=lemp->rule; rp; rp=rp->next){
951 if( rp->canReduce ) continue;
952 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
953 lemp->errorcnt++;
954 }
955}
956
957/* Resolve a conflict between the two given actions. If the
958** conflict can't be resolve, return non-zero.
959**
960** NO LONGER TRUE:
961** To resolve a conflict, first look to see if either action
962** is on an error rule. In that case, take the action which
963** is not associated with the error rule. If neither or both
964** actions are associated with an error rule, then try to
965** use precedence to resolve the conflict.
966**
967** If either action is a SHIFT, then it must be apx. This
968** function won't work if apx->type==REDUCE and apy->type==SHIFT.
969*/
970static int resolve_conflict(apx,apy,errsym)
971struct action *apx;
972struct action *apy;
973struct symbol *errsym; /* The error symbol (if defined. NULL otherwise) */
974{
975 struct symbol *spx, *spy;
976 int errcnt = 0;
977 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
978 if( apx->type==SHIFT && apy->type==REDUCE ){
979 spx = apx->sp;
980 spy = apy->x.rp->precsym;
981 if( spy==0 || spx->prec<0 || spy->prec<0 ){
982 /* Not enough precedence information. */
983 apy->type = CONFLICT;
984 errcnt++;
985 }else if( spx->prec>spy->prec ){ /* Lower precedence wins */
986 apy->type = RD_RESOLVED;
987 }else if( spx->prec<spy->prec ){
988 apx->type = SH_RESOLVED;
989 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
990 apy->type = RD_RESOLVED; /* associativity */
991 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
992 apx->type = SH_RESOLVED;
993 }else{
994 assert( spx->prec==spy->prec && spx->assoc==NONE );
995 apy->type = CONFLICT;
996 errcnt++;
997 }
998 }else if( apx->type==REDUCE && apy->type==REDUCE ){
999 spx = apx->x.rp->precsym;
1000 spy = apy->x.rp->precsym;
1001 if( spx==0 || spy==0 || spx->prec<0 ||
1002 spy->prec<0 || spx->prec==spy->prec ){
1003 apy->type = CONFLICT;
1004 errcnt++;
1005 }else if( spx->prec>spy->prec ){
1006 apy->type = RD_RESOLVED;
1007 }else if( spx->prec<spy->prec ){
1008 apx->type = RD_RESOLVED;
1009 }
1010 }else{
drhb59499c2002-02-23 18:45:13 +00001011 assert(
1012 apx->type==SH_RESOLVED ||
1013 apx->type==RD_RESOLVED ||
1014 apx->type==CONFLICT ||
1015 apy->type==SH_RESOLVED ||
1016 apy->type==RD_RESOLVED ||
1017 apy->type==CONFLICT
1018 );
1019 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1020 ** REDUCEs on the list. If we reach this point it must be because
1021 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001022 }
1023 return errcnt;
1024}
1025/********************* From the file "configlist.c" *************************/
1026/*
1027** Routines to processing a configuration list and building a state
1028** in the LEMON parser generator.
1029*/
1030
1031static struct config *freelist = 0; /* List of free configurations */
1032static struct config *current = 0; /* Top of list of configurations */
1033static struct config **currentend = 0; /* Last on list of configs */
1034static struct config *basis = 0; /* Top of list of basis configs */
1035static struct config **basisend = 0; /* End of list of basis configs */
1036
1037/* Return a pointer to a new configuration */
1038PRIVATE struct config *newconfig(){
1039 struct config *new;
1040 if( freelist==0 ){
1041 int i;
1042 int amt = 3;
1043 freelist = (struct config *)malloc( sizeof(struct config)*amt );
1044 if( freelist==0 ){
1045 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1046 exit(1);
1047 }
1048 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1049 freelist[amt-1].next = 0;
1050 }
1051 new = freelist;
1052 freelist = freelist->next;
1053 return new;
1054}
1055
1056/* The configuration "old" is no longer used */
1057PRIVATE void deleteconfig(old)
1058struct config *old;
1059{
1060 old->next = freelist;
1061 freelist = old;
1062}
1063
1064/* Initialized the configuration list builder */
1065void Configlist_init(){
1066 current = 0;
1067 currentend = &current;
1068 basis = 0;
1069 basisend = &basis;
1070 Configtable_init();
1071 return;
1072}
1073
1074/* Initialized the configuration list builder */
1075void Configlist_reset(){
1076 current = 0;
1077 currentend = &current;
1078 basis = 0;
1079 basisend = &basis;
1080 Configtable_clear(0);
1081 return;
1082}
1083
1084/* Add another configuration to the configuration list */
1085struct config *Configlist_add(rp,dot)
1086struct rule *rp; /* The rule */
1087int dot; /* Index into the RHS of the rule where the dot goes */
1088{
1089 struct config *cfp, model;
1090
1091 assert( currentend!=0 );
1092 model.rp = rp;
1093 model.dot = dot;
1094 cfp = Configtable_find(&model);
1095 if( cfp==0 ){
1096 cfp = newconfig();
1097 cfp->rp = rp;
1098 cfp->dot = dot;
1099 cfp->fws = SetNew();
1100 cfp->stp = 0;
1101 cfp->fplp = cfp->bplp = 0;
1102 cfp->next = 0;
1103 cfp->bp = 0;
1104 *currentend = cfp;
1105 currentend = &cfp->next;
1106 Configtable_insert(cfp);
1107 }
1108 return cfp;
1109}
1110
1111/* Add a basis configuration to the configuration list */
1112struct config *Configlist_addbasis(rp,dot)
1113struct rule *rp;
1114int dot;
1115{
1116 struct config *cfp, model;
1117
1118 assert( basisend!=0 );
1119 assert( currentend!=0 );
1120 model.rp = rp;
1121 model.dot = dot;
1122 cfp = Configtable_find(&model);
1123 if( cfp==0 ){
1124 cfp = newconfig();
1125 cfp->rp = rp;
1126 cfp->dot = dot;
1127 cfp->fws = SetNew();
1128 cfp->stp = 0;
1129 cfp->fplp = cfp->bplp = 0;
1130 cfp->next = 0;
1131 cfp->bp = 0;
1132 *currentend = cfp;
1133 currentend = &cfp->next;
1134 *basisend = cfp;
1135 basisend = &cfp->bp;
1136 Configtable_insert(cfp);
1137 }
1138 return cfp;
1139}
1140
1141/* Compute the closure of the configuration list */
1142void Configlist_closure(lemp)
1143struct lemon *lemp;
1144{
1145 struct config *cfp, *newcfp;
1146 struct rule *rp, *newrp;
1147 struct symbol *sp, *xsp;
1148 int i, dot;
1149
1150 assert( currentend!=0 );
1151 for(cfp=current; cfp; cfp=cfp->next){
1152 rp = cfp->rp;
1153 dot = cfp->dot;
1154 if( dot>=rp->nrhs ) continue;
1155 sp = rp->rhs[dot];
1156 if( sp->type==NONTERMINAL ){
1157 if( sp->rule==0 && sp!=lemp->errsym ){
1158 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1159 sp->name);
1160 lemp->errorcnt++;
1161 }
1162 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1163 newcfp = Configlist_add(newrp,0);
1164 for(i=dot+1; i<rp->nrhs; i++){
1165 xsp = rp->rhs[i];
1166 if( xsp->type==TERMINAL ){
1167 SetAdd(newcfp->fws,xsp->index);
1168 break;
1169 }else{
1170 SetUnion(newcfp->fws,xsp->firstset);
drhb27b83a2002-08-14 23:18:57 +00001171 if( xsp->lambda==B_FALSE ) break;
drh75897232000-05-29 14:26:00 +00001172 }
1173 }
1174 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1175 }
1176 }
1177 }
1178 return;
1179}
1180
1181/* Sort the configuration list */
1182void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001183 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001184 currentend = 0;
1185 return;
1186}
1187
1188/* Sort the basis configuration list */
1189void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001190 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001191 basisend = 0;
1192 return;
1193}
1194
1195/* Return a pointer to the head of the configuration list and
1196** reset the list */
1197struct config *Configlist_return(){
1198 struct config *old;
1199 old = current;
1200 current = 0;
1201 currentend = 0;
1202 return old;
1203}
1204
1205/* Return a pointer to the head of the configuration list and
1206** reset the list */
1207struct config *Configlist_basis(){
1208 struct config *old;
1209 old = basis;
1210 basis = 0;
1211 basisend = 0;
1212 return old;
1213}
1214
1215/* Free all elements of the given configuration list */
1216void Configlist_eat(cfp)
1217struct config *cfp;
1218{
1219 struct config *nextcfp;
1220 for(; cfp; cfp=nextcfp){
1221 nextcfp = cfp->next;
1222 assert( cfp->fplp==0 );
1223 assert( cfp->bplp==0 );
1224 if( cfp->fws ) SetFree(cfp->fws);
1225 deleteconfig(cfp);
1226 }
1227 return;
1228}
1229/***************** From the file "error.c" *********************************/
1230/*
1231** Code for printing error message.
1232*/
1233
1234/* Find a good place to break "msg" so that its length is at least "min"
1235** but no more than "max". Make the point as close to max as possible.
1236*/
1237static int findbreak(msg,min,max)
1238char *msg;
1239int min;
1240int max;
1241{
1242 int i,spot;
1243 char c;
1244 for(i=spot=min; i<=max; i++){
1245 c = msg[i];
1246 if( c=='\t' ) msg[i] = ' ';
1247 if( c=='\n' ){ msg[i] = ' '; spot = i; break; }
1248 if( c==0 ){ spot = i; break; }
1249 if( c=='-' && i<max-1 ) spot = i+1;
1250 if( c==' ' ) spot = i;
1251 }
1252 return spot;
1253}
1254
1255/*
1256** The error message is split across multiple lines if necessary. The
1257** splits occur at a space, if there is a space available near the end
1258** of the line.
1259*/
1260#define ERRMSGSIZE 10000 /* Hope this is big enough. No way to error check */
1261#define LINEWIDTH 79 /* Max width of any output line */
1262#define PREFIXLIMIT 30 /* Max width of the prefix on each line */
drhf9a2e7b2003-04-15 01:49:48 +00001263void ErrorMsg(const char *filename, int lineno, const char *format, ...){
drh75897232000-05-29 14:26:00 +00001264 char errmsg[ERRMSGSIZE];
1265 char prefix[PREFIXLIMIT+10];
1266 int errmsgsize;
1267 int prefixsize;
1268 int availablewidth;
1269 va_list ap;
1270 int end, restart, base;
1271
drhf9a2e7b2003-04-15 01:49:48 +00001272 va_start(ap, format);
drh75897232000-05-29 14:26:00 +00001273 /* Prepare a prefix to be prepended to every output line */
1274 if( lineno>0 ){
1275 sprintf(prefix,"%.*s:%d: ",PREFIXLIMIT-10,filename,lineno);
1276 }else{
1277 sprintf(prefix,"%.*s: ",PREFIXLIMIT-10,filename);
1278 }
1279 prefixsize = strlen(prefix);
1280 availablewidth = LINEWIDTH - prefixsize;
1281
1282 /* Generate the error message */
1283 vsprintf(errmsg,format,ap);
1284 va_end(ap);
1285 errmsgsize = strlen(errmsg);
1286 /* Remove trailing '\n's from the error message. */
1287 while( errmsgsize>0 && errmsg[errmsgsize-1]=='\n' ){
1288 errmsg[--errmsgsize] = 0;
1289 }
1290
1291 /* Print the error message */
1292 base = 0;
1293 while( errmsg[base]!=0 ){
1294 end = restart = findbreak(&errmsg[base],0,availablewidth);
1295 restart += base;
1296 while( errmsg[restart]==' ' ) restart++;
1297 fprintf(stdout,"%s%.*s\n",prefix,end,&errmsg[base]);
1298 base = restart;
1299 }
1300}
1301/**************** From the file "main.c" ************************************/
1302/*
1303** Main program file for the LEMON parser generator.
1304*/
1305
1306/* Report an out-of-memory condition and abort. This function
1307** is used mostly by the "MemoryCheck" macro in struct.h
1308*/
1309void memory_error(){
1310 fprintf(stderr,"Out of memory. Aborting...\n");
1311 exit(1);
1312}
1313
drh6d08b4d2004-07-20 12:45:22 +00001314static int nDefine = 0; /* Number of -D options on the command line */
1315static char **azDefine = 0; /* Name of the -D macros */
1316
1317/* This routine is called with the argument to each -D command-line option.
1318** Add the macro defined to the azDefine array.
1319*/
1320static void handle_D_option(char *z){
1321 char **paz;
1322 nDefine++;
1323 azDefine = realloc(azDefine, sizeof(azDefine[0])*nDefine);
1324 if( azDefine==0 ){
1325 fprintf(stderr,"out of memory\n");
1326 exit(1);
1327 }
1328 paz = &azDefine[nDefine-1];
1329 *paz = malloc( strlen(z)+1 );
1330 if( *paz==0 ){
1331 fprintf(stderr,"out of memory\n");
1332 exit(1);
1333 }
1334 strcpy(*paz, z);
1335 for(z=*paz; *z && *z!='='; z++){}
1336 *z = 0;
1337}
1338
drh75897232000-05-29 14:26:00 +00001339
1340/* The main program. Parse the command line and do it... */
1341int main(argc,argv)
1342int argc;
1343char **argv;
1344{
1345 static int version = 0;
1346 static int rpflag = 0;
1347 static int basisflag = 0;
1348 static int compress = 0;
1349 static int quiet = 0;
1350 static int statistics = 0;
1351 static int mhflag = 0;
1352 static struct s_options options[] = {
1353 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1354 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001355 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh75897232000-05-29 14:26:00 +00001356 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
1357 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file"},
1358 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drh6d08b4d2004-07-20 12:45:22 +00001359 {OPT_FLAG, "s", (char*)&statistics,
1360 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001361 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1362 {OPT_FLAG,0,0,0}
1363 };
1364 int i;
1365 struct lemon lem;
1366
drhb0c86772000-06-02 23:21:26 +00001367 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001368 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001369 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001370 exit(0);
1371 }
drhb0c86772000-06-02 23:21:26 +00001372 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001373 fprintf(stderr,"Exactly one filename argument is required.\n");
1374 exit(1);
1375 }
1376 lem.errorcnt = 0;
1377
1378 /* Initialize the machine */
1379 Strsafe_init();
1380 Symbol_init();
1381 State_init();
1382 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001383 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001384 lem.basisflag = basisflag;
drh0bd1f4e2002-06-06 18:54:39 +00001385 lem.has_fallback = 0;
drh75897232000-05-29 14:26:00 +00001386 lem.nconflict = 0;
1387 lem.name = lem.include = lem.arg = lem.tokentype = lem.start = 0;
drh960e8c62001-04-03 16:53:21 +00001388 lem.vartype = 0;
drh75897232000-05-29 14:26:00 +00001389 lem.stacksize = 0;
1390 lem.error = lem.overflow = lem.failure = lem.accept = lem.tokendest =
1391 lem.tokenprefix = lem.outname = lem.extracode = 0;
drh960e8c62001-04-03 16:53:21 +00001392 lem.vardest = 0;
drh75897232000-05-29 14:26:00 +00001393 lem.tablesize = 0;
1394 Symbol_new("$");
1395 lem.errsym = Symbol_new("error");
1396
1397 /* Parse the input file */
1398 Parse(&lem);
1399 if( lem.errorcnt ) exit(lem.errorcnt);
1400 if( lem.rule==0 ){
1401 fprintf(stderr,"Empty grammar.\n");
1402 exit(1);
1403 }
1404
1405 /* Count and index the symbols of the grammar */
1406 lem.nsymbol = Symbol_count();
1407 Symbol_new("{default}");
1408 lem.symbols = Symbol_arrayof();
drh60d31652004-02-22 00:08:04 +00001409 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
drh75897232000-05-29 14:26:00 +00001410 qsort(lem.symbols,lem.nsymbol+1,sizeof(struct symbol*),
1411 (int(*)())Symbolcmpp);
1412 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
1413 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1414 lem.nterminal = i;
1415
1416 /* Generate a reprint of the grammar, if requested on the command line */
1417 if( rpflag ){
1418 Reprint(&lem);
1419 }else{
1420 /* Initialize the size for all follow and first sets */
1421 SetSize(lem.nterminal);
1422
1423 /* Find the precedence for every production rule (that has one) */
1424 FindRulePrecedences(&lem);
1425
1426 /* Compute the lambda-nonterminals and the first-sets for every
1427 ** nonterminal */
1428 FindFirstSets(&lem);
1429
1430 /* Compute all LR(0) states. Also record follow-set propagation
1431 ** links so that the follow-set can be computed later */
1432 lem.nstate = 0;
1433 FindStates(&lem);
1434 lem.sorted = State_arrayof();
1435
1436 /* Tie up loose ends on the propagation links */
1437 FindLinks(&lem);
1438
1439 /* Compute the follow set of every reducible configuration */
1440 FindFollowSets(&lem);
1441
1442 /* Compute the action tables */
1443 FindActions(&lem);
1444
1445 /* Compress the action tables */
1446 if( compress==0 ) CompressTables(&lem);
1447
1448 /* Generate a report of the parser generated. (the "y.output" file) */
1449 if( !quiet ) ReportOutput(&lem);
1450
1451 /* Generate the source code for the parser */
1452 ReportTable(&lem, mhflag);
1453
1454 /* Produce a header file for use by the scanner. (This step is
1455 ** omitted if the "-m" option is used because makeheaders will
1456 ** generate the file for us.) */
1457 if( !mhflag ) ReportHeader(&lem);
1458 }
1459 if( statistics ){
1460 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1461 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1462 printf(" %d states, %d parser table entries, %d conflicts\n",
1463 lem.nstate, lem.tablesize, lem.nconflict);
1464 }
1465 if( lem.nconflict ){
1466 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
1467 }
1468 exit(lem.errorcnt + lem.nconflict);
drh218dc692004-05-31 23:13:45 +00001469 return (lem.errorcnt + lem.nconflict);
drh75897232000-05-29 14:26:00 +00001470}
1471/******************** From the file "msort.c" *******************************/
1472/*
1473** A generic merge-sort program.
1474**
1475** USAGE:
1476** Let "ptr" be a pointer to some structure which is at the head of
1477** a null-terminated list. Then to sort the list call:
1478**
1479** ptr = msort(ptr,&(ptr->next),cmpfnc);
1480**
1481** In the above, "cmpfnc" is a pointer to a function which compares
1482** two instances of the structure and returns an integer, as in
1483** strcmp. The second argument is a pointer to the pointer to the
1484** second element of the linked list. This address is used to compute
1485** the offset to the "next" field within the structure. The offset to
1486** the "next" field must be constant for all structures in the list.
1487**
1488** The function returns a new pointer which is the head of the list
1489** after sorting.
1490**
1491** ALGORITHM:
1492** Merge-sort.
1493*/
1494
1495/*
1496** Return a pointer to the next structure in the linked list.
1497*/
drhba99af52001-10-25 20:37:16 +00001498#define NEXT(A) (*(char**)(((unsigned long)A)+offset))
drh75897232000-05-29 14:26:00 +00001499
1500/*
1501** Inputs:
1502** a: A sorted, null-terminated linked list. (May be null).
1503** b: A sorted, null-terminated linked list. (May be null).
1504** cmp: A pointer to the comparison function.
1505** offset: Offset in the structure to the "next" field.
1506**
1507** Return Value:
1508** A pointer to the head of a sorted list containing the elements
1509** of both a and b.
1510**
1511** Side effects:
1512** The "next" pointers for elements in the lists a and b are
1513** changed.
1514*/
1515static char *merge(a,b,cmp,offset)
1516char *a;
1517char *b;
1518int (*cmp)();
1519int offset;
1520{
1521 char *ptr, *head;
1522
1523 if( a==0 ){
1524 head = b;
1525 }else if( b==0 ){
1526 head = a;
1527 }else{
1528 if( (*cmp)(a,b)<0 ){
1529 ptr = a;
1530 a = NEXT(a);
1531 }else{
1532 ptr = b;
1533 b = NEXT(b);
1534 }
1535 head = ptr;
1536 while( a && b ){
1537 if( (*cmp)(a,b)<0 ){
1538 NEXT(ptr) = a;
1539 ptr = a;
1540 a = NEXT(a);
1541 }else{
1542 NEXT(ptr) = b;
1543 ptr = b;
1544 b = NEXT(b);
1545 }
1546 }
1547 if( a ) NEXT(ptr) = a;
1548 else NEXT(ptr) = b;
1549 }
1550 return head;
1551}
1552
1553/*
1554** Inputs:
1555** list: Pointer to a singly-linked list of structures.
1556** next: Pointer to pointer to the second element of the list.
1557** cmp: A comparison function.
1558**
1559** Return Value:
1560** A pointer to the head of a sorted list containing the elements
1561** orginally in list.
1562**
1563** Side effects:
1564** The "next" pointers for elements in list are changed.
1565*/
1566#define LISTSIZE 30
1567char *msort(list,next,cmp)
1568char *list;
1569char **next;
1570int (*cmp)();
1571{
drhba99af52001-10-25 20:37:16 +00001572 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001573 char *ep;
1574 char *set[LISTSIZE];
1575 int i;
drhba99af52001-10-25 20:37:16 +00001576 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001577 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1578 while( list ){
1579 ep = list;
1580 list = NEXT(list);
1581 NEXT(ep) = 0;
1582 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1583 ep = merge(ep,set[i],cmp,offset);
1584 set[i] = 0;
1585 }
1586 set[i] = ep;
1587 }
1588 ep = 0;
1589 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(ep,set[i],cmp,offset);
1590 return ep;
1591}
1592/************************ From the file "option.c" **************************/
1593static char **argv;
1594static struct s_options *op;
1595static FILE *errstream;
1596
1597#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1598
1599/*
1600** Print the command line with a carrot pointing to the k-th character
1601** of the n-th field.
1602*/
1603static void errline(n,k,err)
1604int n;
1605int k;
1606FILE *err;
1607{
1608 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001609 if( argv[0] ) fprintf(err,"%s",argv[0]);
1610 spcnt = strlen(argv[0]) + 1;
1611 for(i=1; i<n && argv[i]; i++){
1612 fprintf(err," %s",argv[i]);
drhdc30dd32005-02-16 03:35:15 +00001613 spcnt += strlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001614 }
1615 spcnt += k;
1616 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1617 if( spcnt<20 ){
1618 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1619 }else{
1620 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1621 }
1622}
1623
1624/*
1625** Return the index of the N-th non-switch argument. Return -1
1626** if N is out of range.
1627*/
1628static int argindex(n)
1629int n;
1630{
1631 int i;
1632 int dashdash = 0;
1633 if( argv!=0 && *argv!=0 ){
1634 for(i=1; argv[i]; i++){
1635 if( dashdash || !ISOPT(argv[i]) ){
1636 if( n==0 ) return i;
1637 n--;
1638 }
1639 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1640 }
1641 }
1642 return -1;
1643}
1644
1645static char emsg[] = "Command line syntax error: ";
1646
1647/*
1648** Process a flag command line argument.
1649*/
1650static int handleflags(i,err)
1651int i;
1652FILE *err;
1653{
1654 int v;
1655 int errcnt = 0;
1656 int j;
1657 for(j=0; op[j].label; j++){
drh6d08b4d2004-07-20 12:45:22 +00001658 if( strncmp(&argv[i][1],op[j].label,strlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001659 }
1660 v = argv[i][0]=='-' ? 1 : 0;
1661 if( op[j].label==0 ){
1662 if( err ){
1663 fprintf(err,"%sundefined option.\n",emsg);
1664 errline(i,1,err);
1665 }
1666 errcnt++;
1667 }else if( op[j].type==OPT_FLAG ){
1668 *((int*)op[j].arg) = v;
1669 }else if( op[j].type==OPT_FFLAG ){
1670 (*(void(*)())(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001671 }else if( op[j].type==OPT_FSTR ){
1672 (*(void(*)())(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001673 }else{
1674 if( err ){
1675 fprintf(err,"%smissing argument on switch.\n",emsg);
1676 errline(i,1,err);
1677 }
1678 errcnt++;
1679 }
1680 return errcnt;
1681}
1682
1683/*
1684** Process a command line switch which has an argument.
1685*/
1686static int handleswitch(i,err)
1687int i;
1688FILE *err;
1689{
1690 int lv = 0;
1691 double dv = 0.0;
1692 char *sv = 0, *end;
1693 char *cp;
1694 int j;
1695 int errcnt = 0;
1696 cp = strchr(argv[i],'=');
1697 *cp = 0;
1698 for(j=0; op[j].label; j++){
1699 if( strcmp(argv[i],op[j].label)==0 ) break;
1700 }
1701 *cp = '=';
1702 if( op[j].label==0 ){
1703 if( err ){
1704 fprintf(err,"%sundefined option.\n",emsg);
1705 errline(i,0,err);
1706 }
1707 errcnt++;
1708 }else{
1709 cp++;
1710 switch( op[j].type ){
1711 case OPT_FLAG:
1712 case OPT_FFLAG:
1713 if( err ){
1714 fprintf(err,"%soption requires an argument.\n",emsg);
1715 errline(i,0,err);
1716 }
1717 errcnt++;
1718 break;
1719 case OPT_DBL:
1720 case OPT_FDBL:
1721 dv = strtod(cp,&end);
1722 if( *end ){
1723 if( err ){
1724 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001725 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001726 }
1727 errcnt++;
1728 }
1729 break;
1730 case OPT_INT:
1731 case OPT_FINT:
1732 lv = strtol(cp,&end,0);
1733 if( *end ){
1734 if( err ){
1735 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001736 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001737 }
1738 errcnt++;
1739 }
1740 break;
1741 case OPT_STR:
1742 case OPT_FSTR:
1743 sv = cp;
1744 break;
1745 }
1746 switch( op[j].type ){
1747 case OPT_FLAG:
1748 case OPT_FFLAG:
1749 break;
1750 case OPT_DBL:
1751 *(double*)(op[j].arg) = dv;
1752 break;
1753 case OPT_FDBL:
1754 (*(void(*)())(op[j].arg))(dv);
1755 break;
1756 case OPT_INT:
1757 *(int*)(op[j].arg) = lv;
1758 break;
1759 case OPT_FINT:
1760 (*(void(*)())(op[j].arg))((int)lv);
1761 break;
1762 case OPT_STR:
1763 *(char**)(op[j].arg) = sv;
1764 break;
1765 case OPT_FSTR:
1766 (*(void(*)())(op[j].arg))(sv);
1767 break;
1768 }
1769 }
1770 return errcnt;
1771}
1772
drhb0c86772000-06-02 23:21:26 +00001773int OptInit(a,o,err)
drh75897232000-05-29 14:26:00 +00001774char **a;
1775struct s_options *o;
1776FILE *err;
1777{
1778 int errcnt = 0;
1779 argv = a;
1780 op = o;
1781 errstream = err;
1782 if( argv && *argv && op ){
1783 int i;
1784 for(i=1; argv[i]; i++){
1785 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1786 errcnt += handleflags(i,err);
1787 }else if( strchr(argv[i],'=') ){
1788 errcnt += handleswitch(i,err);
1789 }
1790 }
1791 }
1792 if( errcnt>0 ){
1793 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001794 OptPrint();
drh75897232000-05-29 14:26:00 +00001795 exit(1);
1796 }
1797 return 0;
1798}
1799
drhb0c86772000-06-02 23:21:26 +00001800int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001801 int cnt = 0;
1802 int dashdash = 0;
1803 int i;
1804 if( argv!=0 && argv[0]!=0 ){
1805 for(i=1; argv[i]; i++){
1806 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1807 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1808 }
1809 }
1810 return cnt;
1811}
1812
drhb0c86772000-06-02 23:21:26 +00001813char *OptArg(n)
drh75897232000-05-29 14:26:00 +00001814int n;
1815{
1816 int i;
1817 i = argindex(n);
1818 return i>=0 ? argv[i] : 0;
1819}
1820
drhb0c86772000-06-02 23:21:26 +00001821void OptErr(n)
drh75897232000-05-29 14:26:00 +00001822int n;
1823{
1824 int i;
1825 i = argindex(n);
1826 if( i>=0 ) errline(i,0,errstream);
1827}
1828
drhb0c86772000-06-02 23:21:26 +00001829void OptPrint(){
drh75897232000-05-29 14:26:00 +00001830 int i;
1831 int max, len;
1832 max = 0;
1833 for(i=0; op[i].label; i++){
1834 len = strlen(op[i].label) + 1;
1835 switch( op[i].type ){
1836 case OPT_FLAG:
1837 case OPT_FFLAG:
1838 break;
1839 case OPT_INT:
1840 case OPT_FINT:
1841 len += 9; /* length of "<integer>" */
1842 break;
1843 case OPT_DBL:
1844 case OPT_FDBL:
1845 len += 6; /* length of "<real>" */
1846 break;
1847 case OPT_STR:
1848 case OPT_FSTR:
1849 len += 8; /* length of "<string>" */
1850 break;
1851 }
1852 if( len>max ) max = len;
1853 }
1854 for(i=0; op[i].label; i++){
1855 switch( op[i].type ){
1856 case OPT_FLAG:
1857 case OPT_FFLAG:
1858 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
1859 break;
1860 case OPT_INT:
1861 case OPT_FINT:
1862 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001863 (int)(max-strlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001864 break;
1865 case OPT_DBL:
1866 case OPT_FDBL:
1867 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001868 (int)(max-strlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001869 break;
1870 case OPT_STR:
1871 case OPT_FSTR:
1872 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh8b582012003-10-21 13:16:03 +00001873 (int)(max-strlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001874 break;
1875 }
1876 }
1877}
1878/*********************** From the file "parse.c" ****************************/
1879/*
1880** Input file parser for the LEMON parser generator.
1881*/
1882
1883/* The state of the parser */
1884struct pstate {
1885 char *filename; /* Name of the input file */
1886 int tokenlineno; /* Linenumber at which current token starts */
1887 int errorcnt; /* Number of errors so far */
1888 char *tokenstart; /* Text of current token */
1889 struct lemon *gp; /* Global state vector */
1890 enum e_state {
1891 INITIALIZE,
1892 WAITING_FOR_DECL_OR_RULE,
1893 WAITING_FOR_DECL_KEYWORD,
1894 WAITING_FOR_DECL_ARG,
1895 WAITING_FOR_PRECEDENCE_SYMBOL,
1896 WAITING_FOR_ARROW,
1897 IN_RHS,
1898 LHS_ALIAS_1,
1899 LHS_ALIAS_2,
1900 LHS_ALIAS_3,
1901 RHS_ALIAS_1,
1902 RHS_ALIAS_2,
1903 PRECEDENCE_MARK_1,
1904 PRECEDENCE_MARK_2,
1905 RESYNC_AFTER_RULE_ERROR,
1906 RESYNC_AFTER_DECL_ERROR,
1907 WAITING_FOR_DESTRUCTOR_SYMBOL,
drh0bd1f4e2002-06-06 18:54:39 +00001908 WAITING_FOR_DATATYPE_SYMBOL,
1909 WAITING_FOR_FALLBACK_ID
drh75897232000-05-29 14:26:00 +00001910 } state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00001911 struct symbol *fallback; /* The fallback token */
drh75897232000-05-29 14:26:00 +00001912 struct symbol *lhs; /* Left-hand side of current rule */
1913 char *lhsalias; /* Alias for the LHS */
1914 int nrhs; /* Number of right-hand side symbols seen */
1915 struct symbol *rhs[MAXRHS]; /* RHS symbols */
1916 char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
1917 struct rule *prevrule; /* Previous rule parsed */
1918 char *declkeyword; /* Keyword of a declaration */
1919 char **declargslot; /* Where the declaration argument should be put */
1920 int *decllnslot; /* Where the declaration linenumber is put */
1921 enum e_assoc declassoc; /* Assign this association to decl arguments */
1922 int preccounter; /* Assign this precedence to decl arguments */
1923 struct rule *firstrule; /* Pointer to first rule in the grammar */
1924 struct rule *lastrule; /* Pointer to the most recently parsed rule */
1925};
1926
1927/* Parse a single token */
1928static void parseonetoken(psp)
1929struct pstate *psp;
1930{
1931 char *x;
1932 x = Strsafe(psp->tokenstart); /* Save the token permanently */
1933#if 0
1934 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
1935 x,psp->state);
1936#endif
1937 switch( psp->state ){
1938 case INITIALIZE:
1939 psp->prevrule = 0;
1940 psp->preccounter = 0;
1941 psp->firstrule = psp->lastrule = 0;
1942 psp->gp->nrule = 0;
1943 /* Fall thru to next case */
1944 case WAITING_FOR_DECL_OR_RULE:
1945 if( x[0]=='%' ){
1946 psp->state = WAITING_FOR_DECL_KEYWORD;
1947 }else if( islower(x[0]) ){
1948 psp->lhs = Symbol_new(x);
1949 psp->nrhs = 0;
1950 psp->lhsalias = 0;
1951 psp->state = WAITING_FOR_ARROW;
1952 }else if( x[0]=='{' ){
1953 if( psp->prevrule==0 ){
1954 ErrorMsg(psp->filename,psp->tokenlineno,
1955"There is not prior rule opon which to attach the code \
1956fragment which begins on this line.");
1957 psp->errorcnt++;
1958 }else if( psp->prevrule->code!=0 ){
1959 ErrorMsg(psp->filename,psp->tokenlineno,
1960"Code fragment beginning on this line is not the first \
1961to follow the previous rule.");
1962 psp->errorcnt++;
1963 }else{
1964 psp->prevrule->line = psp->tokenlineno;
1965 psp->prevrule->code = &x[1];
1966 }
1967 }else if( x[0]=='[' ){
1968 psp->state = PRECEDENCE_MARK_1;
1969 }else{
1970 ErrorMsg(psp->filename,psp->tokenlineno,
1971 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
1972 x);
1973 psp->errorcnt++;
1974 }
1975 break;
1976 case PRECEDENCE_MARK_1:
1977 if( !isupper(x[0]) ){
1978 ErrorMsg(psp->filename,psp->tokenlineno,
1979 "The precedence symbol must be a terminal.");
1980 psp->errorcnt++;
1981 }else if( psp->prevrule==0 ){
1982 ErrorMsg(psp->filename,psp->tokenlineno,
1983 "There is no prior rule to assign precedence \"[%s]\".",x);
1984 psp->errorcnt++;
1985 }else if( psp->prevrule->precsym!=0 ){
1986 ErrorMsg(psp->filename,psp->tokenlineno,
1987"Precedence mark on this line is not the first \
1988to follow the previous rule.");
1989 psp->errorcnt++;
1990 }else{
1991 psp->prevrule->precsym = Symbol_new(x);
1992 }
1993 psp->state = PRECEDENCE_MARK_2;
1994 break;
1995 case PRECEDENCE_MARK_2:
1996 if( x[0]!=']' ){
1997 ErrorMsg(psp->filename,psp->tokenlineno,
1998 "Missing \"]\" on precedence mark.");
1999 psp->errorcnt++;
2000 }
2001 psp->state = WAITING_FOR_DECL_OR_RULE;
2002 break;
2003 case WAITING_FOR_ARROW:
2004 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2005 psp->state = IN_RHS;
2006 }else if( x[0]=='(' ){
2007 psp->state = LHS_ALIAS_1;
2008 }else{
2009 ErrorMsg(psp->filename,psp->tokenlineno,
2010 "Expected to see a \":\" following the LHS symbol \"%s\".",
2011 psp->lhs->name);
2012 psp->errorcnt++;
2013 psp->state = RESYNC_AFTER_RULE_ERROR;
2014 }
2015 break;
2016 case LHS_ALIAS_1:
2017 if( isalpha(x[0]) ){
2018 psp->lhsalias = x;
2019 psp->state = LHS_ALIAS_2;
2020 }else{
2021 ErrorMsg(psp->filename,psp->tokenlineno,
2022 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2023 x,psp->lhs->name);
2024 psp->errorcnt++;
2025 psp->state = RESYNC_AFTER_RULE_ERROR;
2026 }
2027 break;
2028 case LHS_ALIAS_2:
2029 if( x[0]==')' ){
2030 psp->state = LHS_ALIAS_3;
2031 }else{
2032 ErrorMsg(psp->filename,psp->tokenlineno,
2033 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2034 psp->errorcnt++;
2035 psp->state = RESYNC_AFTER_RULE_ERROR;
2036 }
2037 break;
2038 case LHS_ALIAS_3:
2039 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2040 psp->state = IN_RHS;
2041 }else{
2042 ErrorMsg(psp->filename,psp->tokenlineno,
2043 "Missing \"->\" following: \"%s(%s)\".",
2044 psp->lhs->name,psp->lhsalias);
2045 psp->errorcnt++;
2046 psp->state = RESYNC_AFTER_RULE_ERROR;
2047 }
2048 break;
2049 case IN_RHS:
2050 if( x[0]=='.' ){
2051 struct rule *rp;
2052 rp = (struct rule *)malloc( sizeof(struct rule) +
2053 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs );
2054 if( rp==0 ){
2055 ErrorMsg(psp->filename,psp->tokenlineno,
2056 "Can't allocate enough memory for this rule.");
2057 psp->errorcnt++;
2058 psp->prevrule = 0;
2059 }else{
2060 int i;
2061 rp->ruleline = psp->tokenlineno;
2062 rp->rhs = (struct symbol**)&rp[1];
2063 rp->rhsalias = (char**)&(rp->rhs[psp->nrhs]);
2064 for(i=0; i<psp->nrhs; i++){
2065 rp->rhs[i] = psp->rhs[i];
2066 rp->rhsalias[i] = psp->alias[i];
2067 }
2068 rp->lhs = psp->lhs;
2069 rp->lhsalias = psp->lhsalias;
2070 rp->nrhs = psp->nrhs;
2071 rp->code = 0;
2072 rp->precsym = 0;
2073 rp->index = psp->gp->nrule++;
2074 rp->nextlhs = rp->lhs->rule;
2075 rp->lhs->rule = rp;
2076 rp->next = 0;
2077 if( psp->firstrule==0 ){
2078 psp->firstrule = psp->lastrule = rp;
2079 }else{
2080 psp->lastrule->next = rp;
2081 psp->lastrule = rp;
2082 }
2083 psp->prevrule = rp;
2084 }
2085 psp->state = WAITING_FOR_DECL_OR_RULE;
2086 }else if( isalpha(x[0]) ){
2087 if( psp->nrhs>=MAXRHS ){
2088 ErrorMsg(psp->filename,psp->tokenlineno,
2089 "Too many symbol on RHS or rule beginning at \"%s\".",
2090 x);
2091 psp->errorcnt++;
2092 psp->state = RESYNC_AFTER_RULE_ERROR;
2093 }else{
2094 psp->rhs[psp->nrhs] = Symbol_new(x);
2095 psp->alias[psp->nrhs] = 0;
2096 psp->nrhs++;
2097 }
2098 }else if( x[0]=='(' && psp->nrhs>0 ){
2099 psp->state = RHS_ALIAS_1;
2100 }else{
2101 ErrorMsg(psp->filename,psp->tokenlineno,
2102 "Illegal character on RHS of rule: \"%s\".",x);
2103 psp->errorcnt++;
2104 psp->state = RESYNC_AFTER_RULE_ERROR;
2105 }
2106 break;
2107 case RHS_ALIAS_1:
2108 if( isalpha(x[0]) ){
2109 psp->alias[psp->nrhs-1] = x;
2110 psp->state = RHS_ALIAS_2;
2111 }else{
2112 ErrorMsg(psp->filename,psp->tokenlineno,
2113 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2114 x,psp->rhs[psp->nrhs-1]->name);
2115 psp->errorcnt++;
2116 psp->state = RESYNC_AFTER_RULE_ERROR;
2117 }
2118 break;
2119 case RHS_ALIAS_2:
2120 if( x[0]==')' ){
2121 psp->state = IN_RHS;
2122 }else{
2123 ErrorMsg(psp->filename,psp->tokenlineno,
2124 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2125 psp->errorcnt++;
2126 psp->state = RESYNC_AFTER_RULE_ERROR;
2127 }
2128 break;
2129 case WAITING_FOR_DECL_KEYWORD:
2130 if( isalpha(x[0]) ){
2131 psp->declkeyword = x;
2132 psp->declargslot = 0;
2133 psp->decllnslot = 0;
2134 psp->state = WAITING_FOR_DECL_ARG;
2135 if( strcmp(x,"name")==0 ){
2136 psp->declargslot = &(psp->gp->name);
2137 }else if( strcmp(x,"include")==0 ){
2138 psp->declargslot = &(psp->gp->include);
2139 psp->decllnslot = &psp->gp->includeln;
2140 }else if( strcmp(x,"code")==0 ){
2141 psp->declargslot = &(psp->gp->extracode);
2142 psp->decllnslot = &psp->gp->extracodeln;
2143 }else if( strcmp(x,"token_destructor")==0 ){
2144 psp->declargslot = &psp->gp->tokendest;
2145 psp->decllnslot = &psp->gp->tokendestln;
drh960e8c62001-04-03 16:53:21 +00002146 }else if( strcmp(x,"default_destructor")==0 ){
2147 psp->declargslot = &psp->gp->vardest;
2148 psp->decllnslot = &psp->gp->vardestln;
drh75897232000-05-29 14:26:00 +00002149 }else if( strcmp(x,"token_prefix")==0 ){
2150 psp->declargslot = &psp->gp->tokenprefix;
2151 }else if( strcmp(x,"syntax_error")==0 ){
2152 psp->declargslot = &(psp->gp->error);
2153 psp->decllnslot = &psp->gp->errorln;
2154 }else if( strcmp(x,"parse_accept")==0 ){
2155 psp->declargslot = &(psp->gp->accept);
2156 psp->decllnslot = &psp->gp->acceptln;
2157 }else if( strcmp(x,"parse_failure")==0 ){
2158 psp->declargslot = &(psp->gp->failure);
2159 psp->decllnslot = &psp->gp->failureln;
2160 }else if( strcmp(x,"stack_overflow")==0 ){
2161 psp->declargslot = &(psp->gp->overflow);
2162 psp->decllnslot = &psp->gp->overflowln;
2163 }else if( strcmp(x,"extra_argument")==0 ){
2164 psp->declargslot = &(psp->gp->arg);
2165 }else if( strcmp(x,"token_type")==0 ){
2166 psp->declargslot = &(psp->gp->tokentype);
drh960e8c62001-04-03 16:53:21 +00002167 }else if( strcmp(x,"default_type")==0 ){
2168 psp->declargslot = &(psp->gp->vartype);
drh75897232000-05-29 14:26:00 +00002169 }else if( strcmp(x,"stack_size")==0 ){
2170 psp->declargslot = &(psp->gp->stacksize);
2171 }else if( strcmp(x,"start_symbol")==0 ){
2172 psp->declargslot = &(psp->gp->start);
2173 }else if( strcmp(x,"left")==0 ){
2174 psp->preccounter++;
2175 psp->declassoc = LEFT;
2176 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2177 }else if( strcmp(x,"right")==0 ){
2178 psp->preccounter++;
2179 psp->declassoc = RIGHT;
2180 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2181 }else if( strcmp(x,"nonassoc")==0 ){
2182 psp->preccounter++;
2183 psp->declassoc = NONE;
2184 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2185 }else if( strcmp(x,"destructor")==0 ){
2186 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
2187 }else if( strcmp(x,"type")==0 ){
2188 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002189 }else if( strcmp(x,"fallback")==0 ){
2190 psp->fallback = 0;
2191 psp->state = WAITING_FOR_FALLBACK_ID;
drh75897232000-05-29 14:26:00 +00002192 }else{
2193 ErrorMsg(psp->filename,psp->tokenlineno,
2194 "Unknown declaration keyword: \"%%%s\".",x);
2195 psp->errorcnt++;
2196 psp->state = RESYNC_AFTER_DECL_ERROR;
2197 }
2198 }else{
2199 ErrorMsg(psp->filename,psp->tokenlineno,
2200 "Illegal declaration keyword: \"%s\".",x);
2201 psp->errorcnt++;
2202 psp->state = RESYNC_AFTER_DECL_ERROR;
2203 }
2204 break;
2205 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2206 if( !isalpha(x[0]) ){
2207 ErrorMsg(psp->filename,psp->tokenlineno,
2208 "Symbol name missing after %destructor keyword");
2209 psp->errorcnt++;
2210 psp->state = RESYNC_AFTER_DECL_ERROR;
2211 }else{
2212 struct symbol *sp = Symbol_new(x);
2213 psp->declargslot = &sp->destructor;
2214 psp->decllnslot = &sp->destructorln;
2215 psp->state = WAITING_FOR_DECL_ARG;
2216 }
2217 break;
2218 case WAITING_FOR_DATATYPE_SYMBOL:
2219 if( !isalpha(x[0]) ){
2220 ErrorMsg(psp->filename,psp->tokenlineno,
2221 "Symbol name missing after %destructor keyword");
2222 psp->errorcnt++;
2223 psp->state = RESYNC_AFTER_DECL_ERROR;
2224 }else{
2225 struct symbol *sp = Symbol_new(x);
2226 psp->declargslot = &sp->datatype;
2227 psp->decllnslot = 0;
2228 psp->state = WAITING_FOR_DECL_ARG;
2229 }
2230 break;
2231 case WAITING_FOR_PRECEDENCE_SYMBOL:
2232 if( x[0]=='.' ){
2233 psp->state = WAITING_FOR_DECL_OR_RULE;
2234 }else if( isupper(x[0]) ){
2235 struct symbol *sp;
2236 sp = Symbol_new(x);
2237 if( sp->prec>=0 ){
2238 ErrorMsg(psp->filename,psp->tokenlineno,
2239 "Symbol \"%s\" has already be given a precedence.",x);
2240 psp->errorcnt++;
2241 }else{
2242 sp->prec = psp->preccounter;
2243 sp->assoc = psp->declassoc;
2244 }
2245 }else{
2246 ErrorMsg(psp->filename,psp->tokenlineno,
2247 "Can't assign a precedence to \"%s\".",x);
2248 psp->errorcnt++;
2249 }
2250 break;
2251 case WAITING_FOR_DECL_ARG:
2252 if( (x[0]=='{' || x[0]=='\"' || isalnum(x[0])) ){
2253 if( *(psp->declargslot)!=0 ){
2254 ErrorMsg(psp->filename,psp->tokenlineno,
2255 "The argument \"%s\" to declaration \"%%%s\" is not the first.",
2256 x[0]=='\"' ? &x[1] : x,psp->declkeyword);
2257 psp->errorcnt++;
2258 psp->state = RESYNC_AFTER_DECL_ERROR;
2259 }else{
2260 *(psp->declargslot) = (x[0]=='\"' || x[0]=='{') ? &x[1] : x;
2261 if( psp->decllnslot ) *psp->decllnslot = psp->tokenlineno;
2262 psp->state = WAITING_FOR_DECL_OR_RULE;
2263 }
2264 }else{
2265 ErrorMsg(psp->filename,psp->tokenlineno,
2266 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2267 psp->errorcnt++;
2268 psp->state = RESYNC_AFTER_DECL_ERROR;
2269 }
2270 break;
drh0bd1f4e2002-06-06 18:54:39 +00002271 case WAITING_FOR_FALLBACK_ID:
2272 if( x[0]=='.' ){
2273 psp->state = WAITING_FOR_DECL_OR_RULE;
2274 }else if( !isupper(x[0]) ){
2275 ErrorMsg(psp->filename, psp->tokenlineno,
2276 "%%fallback argument \"%s\" should be a token", x);
2277 psp->errorcnt++;
2278 }else{
2279 struct symbol *sp = Symbol_new(x);
2280 if( psp->fallback==0 ){
2281 psp->fallback = sp;
2282 }else if( sp->fallback ){
2283 ErrorMsg(psp->filename, psp->tokenlineno,
2284 "More than one fallback assigned to token %s", x);
2285 psp->errorcnt++;
2286 }else{
2287 sp->fallback = psp->fallback;
2288 psp->gp->has_fallback = 1;
2289 }
2290 }
2291 break;
drh75897232000-05-29 14:26:00 +00002292 case RESYNC_AFTER_RULE_ERROR:
2293/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2294** break; */
2295 case RESYNC_AFTER_DECL_ERROR:
2296 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2297 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2298 break;
2299 }
2300}
2301
drh6d08b4d2004-07-20 12:45:22 +00002302/* Run the proprocessor over the input file text. The global variables
2303** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2304** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2305** comments them out. Text in between is also commented out as appropriate.
2306*/
danielk1977940fac92005-01-23 22:41:37 +00002307static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002308 int i, j, k, n;
2309 int exclude = 0;
2310 int start;
2311 int lineno = 1;
2312 int start_lineno;
2313 for(i=0; z[i]; i++){
2314 if( z[i]=='\n' ) lineno++;
2315 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2316 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2317 if( exclude ){
2318 exclude--;
2319 if( exclude==0 ){
2320 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2321 }
2322 }
2323 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2324 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2325 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2326 if( exclude ){
2327 exclude++;
2328 }else{
2329 for(j=i+7; isspace(z[j]); j++){}
2330 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2331 exclude = 1;
2332 for(k=0; k<nDefine; k++){
2333 if( strncmp(azDefine[k],&z[j],n)==0 && strlen(azDefine[k])==n ){
2334 exclude = 0;
2335 break;
2336 }
2337 }
2338 if( z[i+3]=='n' ) exclude = !exclude;
2339 if( exclude ){
2340 start = i;
2341 start_lineno = lineno;
2342 }
2343 }
2344 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2345 }
2346 }
2347 if( exclude ){
2348 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2349 exit(1);
2350 }
2351}
2352
drh75897232000-05-29 14:26:00 +00002353/* In spite of its name, this function is really a scanner. It read
2354** in the entire input file (all at once) then tokenizes it. Each
2355** token is passed to the function "parseonetoken" which builds all
2356** the appropriate data structures in the global state vector "gp".
2357*/
2358void Parse(gp)
2359struct lemon *gp;
2360{
2361 struct pstate ps;
2362 FILE *fp;
2363 char *filebuf;
2364 int filesize;
2365 int lineno;
2366 int c;
2367 char *cp, *nextcp;
2368 int startline = 0;
2369
2370 ps.gp = gp;
2371 ps.filename = gp->filename;
2372 ps.errorcnt = 0;
2373 ps.state = INITIALIZE;
2374
2375 /* Begin by reading the input file */
2376 fp = fopen(ps.filename,"rb");
2377 if( fp==0 ){
2378 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2379 gp->errorcnt++;
2380 return;
2381 }
2382 fseek(fp,0,2);
2383 filesize = ftell(fp);
2384 rewind(fp);
2385 filebuf = (char *)malloc( filesize+1 );
2386 if( filebuf==0 ){
2387 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2388 filesize+1);
2389 gp->errorcnt++;
2390 return;
2391 }
2392 if( fread(filebuf,1,filesize,fp)!=filesize ){
2393 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2394 filesize);
2395 free(filebuf);
2396 gp->errorcnt++;
2397 return;
2398 }
2399 fclose(fp);
2400 filebuf[filesize] = 0;
2401
drh6d08b4d2004-07-20 12:45:22 +00002402 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2403 preprocess_input(filebuf);
2404
drh75897232000-05-29 14:26:00 +00002405 /* Now scan the text of the input file */
2406 lineno = 1;
2407 for(cp=filebuf; (c= *cp)!=0; ){
2408 if( c=='\n' ) lineno++; /* Keep track of the line number */
2409 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2410 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2411 cp+=2;
2412 while( (c= *cp)!=0 && c!='\n' ) cp++;
2413 continue;
2414 }
2415 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2416 cp+=2;
2417 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2418 if( c=='\n' ) lineno++;
2419 cp++;
2420 }
2421 if( c ) cp++;
2422 continue;
2423 }
2424 ps.tokenstart = cp; /* Mark the beginning of the token */
2425 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2426 if( c=='\"' ){ /* String literals */
2427 cp++;
2428 while( (c= *cp)!=0 && c!='\"' ){
2429 if( c=='\n' ) lineno++;
2430 cp++;
2431 }
2432 if( c==0 ){
2433 ErrorMsg(ps.filename,startline,
2434"String starting on this line is not terminated before the end of the file.");
2435 ps.errorcnt++;
2436 nextcp = cp;
2437 }else{
2438 nextcp = cp+1;
2439 }
2440 }else if( c=='{' ){ /* A block of C code */
2441 int level;
2442 cp++;
2443 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2444 if( c=='\n' ) lineno++;
2445 else if( c=='{' ) level++;
2446 else if( c=='}' ) level--;
2447 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2448 int prevc;
2449 cp = &cp[2];
2450 prevc = 0;
2451 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2452 if( c=='\n' ) lineno++;
2453 prevc = c;
2454 cp++;
2455 }
2456 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
2457 cp = &cp[2];
2458 while( (c= *cp)!=0 && c!='\n' ) cp++;
2459 if( c ) lineno++;
2460 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
2461 int startchar, prevc;
2462 startchar = c;
2463 prevc = 0;
2464 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2465 if( c=='\n' ) lineno++;
2466 if( prevc=='\\' ) prevc = 0;
2467 else prevc = c;
2468 }
2469 }
2470 }
2471 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002472 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002473"C code starting on this line is not terminated before the end of the file.");
2474 ps.errorcnt++;
2475 nextcp = cp;
2476 }else{
2477 nextcp = cp+1;
2478 }
2479 }else if( isalnum(c) ){ /* Identifiers */
2480 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2481 nextcp = cp;
2482 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2483 cp += 3;
2484 nextcp = cp;
2485 }else{ /* All other (one character) operators */
2486 cp++;
2487 nextcp = cp;
2488 }
2489 c = *cp;
2490 *cp = 0; /* Null terminate the token */
2491 parseonetoken(&ps); /* Parse the token */
2492 *cp = c; /* Restore the buffer */
2493 cp = nextcp;
2494 }
2495 free(filebuf); /* Release the buffer after parsing */
2496 gp->rule = ps.firstrule;
2497 gp->errorcnt = ps.errorcnt;
2498}
2499/*************************** From the file "plink.c" *********************/
2500/*
2501** Routines processing configuration follow-set propagation links
2502** in the LEMON parser generator.
2503*/
2504static struct plink *plink_freelist = 0;
2505
2506/* Allocate a new plink */
2507struct plink *Plink_new(){
2508 struct plink *new;
2509
2510 if( plink_freelist==0 ){
2511 int i;
2512 int amt = 100;
2513 plink_freelist = (struct plink *)malloc( sizeof(struct plink)*amt );
2514 if( plink_freelist==0 ){
2515 fprintf(stderr,
2516 "Unable to allocate memory for a new follow-set propagation link.\n");
2517 exit(1);
2518 }
2519 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2520 plink_freelist[amt-1].next = 0;
2521 }
2522 new = plink_freelist;
2523 plink_freelist = plink_freelist->next;
2524 return new;
2525}
2526
2527/* Add a plink to a plink list */
2528void Plink_add(plpp,cfp)
2529struct plink **plpp;
2530struct config *cfp;
2531{
2532 struct plink *new;
2533 new = Plink_new();
2534 new->next = *plpp;
2535 *plpp = new;
2536 new->cfp = cfp;
2537}
2538
2539/* Transfer every plink on the list "from" to the list "to" */
2540void Plink_copy(to,from)
2541struct plink **to;
2542struct plink *from;
2543{
2544 struct plink *nextpl;
2545 while( from ){
2546 nextpl = from->next;
2547 from->next = *to;
2548 *to = from;
2549 from = nextpl;
2550 }
2551}
2552
2553/* Delete every plink on the list */
2554void Plink_delete(plp)
2555struct plink *plp;
2556{
2557 struct plink *nextpl;
2558
2559 while( plp ){
2560 nextpl = plp->next;
2561 plp->next = plink_freelist;
2562 plink_freelist = plp;
2563 plp = nextpl;
2564 }
2565}
2566/*********************** From the file "report.c" **************************/
2567/*
2568** Procedures for generating reports and tables in the LEMON parser generator.
2569*/
2570
2571/* Generate a filename with the given suffix. Space to hold the
2572** name comes from malloc() and must be freed by the calling
2573** function.
2574*/
2575PRIVATE char *file_makename(lemp,suffix)
2576struct lemon *lemp;
2577char *suffix;
2578{
2579 char *name;
2580 char *cp;
2581
2582 name = malloc( strlen(lemp->filename) + strlen(suffix) + 5 );
2583 if( name==0 ){
2584 fprintf(stderr,"Can't allocate space for a filename.\n");
2585 exit(1);
2586 }
2587 strcpy(name,lemp->filename);
2588 cp = strrchr(name,'.');
2589 if( cp ) *cp = 0;
2590 strcat(name,suffix);
2591 return name;
2592}
2593
2594/* Open a file with a name based on the name of the input file,
2595** but with a different (specified) suffix, and return a pointer
2596** to the stream */
2597PRIVATE FILE *file_open(lemp,suffix,mode)
2598struct lemon *lemp;
2599char *suffix;
2600char *mode;
2601{
2602 FILE *fp;
2603
2604 if( lemp->outname ) free(lemp->outname);
2605 lemp->outname = file_makename(lemp, suffix);
2606 fp = fopen(lemp->outname,mode);
2607 if( fp==0 && *mode=='w' ){
2608 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2609 lemp->errorcnt++;
2610 return 0;
2611 }
2612 return fp;
2613}
2614
2615/* Duplicate the input file without comments and without actions
2616** on rules */
2617void Reprint(lemp)
2618struct lemon *lemp;
2619{
2620 struct rule *rp;
2621 struct symbol *sp;
2622 int i, j, maxlen, len, ncolumns, skip;
2623 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2624 maxlen = 10;
2625 for(i=0; i<lemp->nsymbol; i++){
2626 sp = lemp->symbols[i];
2627 len = strlen(sp->name);
2628 if( len>maxlen ) maxlen = len;
2629 }
2630 ncolumns = 76/(maxlen+5);
2631 if( ncolumns<1 ) ncolumns = 1;
2632 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2633 for(i=0; i<skip; i++){
2634 printf("//");
2635 for(j=i; j<lemp->nsymbol; j+=skip){
2636 sp = lemp->symbols[j];
2637 assert( sp->index==j );
2638 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2639 }
2640 printf("\n");
2641 }
2642 for(rp=lemp->rule; rp; rp=rp->next){
2643 printf("%s",rp->lhs->name);
2644/* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
2645 printf(" ::=");
2646 for(i=0; i<rp->nrhs; i++){
2647 printf(" %s",rp->rhs[i]->name);
2648/* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
2649 }
2650 printf(".");
2651 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
2652/* if( rp->code ) printf("\n %s",rp->code); */
2653 printf("\n");
2654 }
2655}
2656
2657void ConfigPrint(fp,cfp)
2658FILE *fp;
2659struct config *cfp;
2660{
2661 struct rule *rp;
2662 int i;
2663 rp = cfp->rp;
2664 fprintf(fp,"%s ::=",rp->lhs->name);
2665 for(i=0; i<=rp->nrhs; i++){
2666 if( i==cfp->dot ) fprintf(fp," *");
2667 if( i==rp->nrhs ) break;
2668 fprintf(fp," %s",rp->rhs[i]->name);
2669 }
2670}
2671
2672/* #define TEST */
2673#ifdef TEST
2674/* Print a set */
2675PRIVATE void SetPrint(out,set,lemp)
2676FILE *out;
2677char *set;
2678struct lemon *lemp;
2679{
2680 int i;
2681 char *spacer;
2682 spacer = "";
2683 fprintf(out,"%12s[","");
2684 for(i=0; i<lemp->nterminal; i++){
2685 if( SetFind(set,i) ){
2686 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2687 spacer = " ";
2688 }
2689 }
2690 fprintf(out,"]\n");
2691}
2692
2693/* Print a plink chain */
2694PRIVATE void PlinkPrint(out,plp,tag)
2695FILE *out;
2696struct plink *plp;
2697char *tag;
2698{
2699 while( plp ){
2700 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->index);
2701 ConfigPrint(out,plp->cfp);
2702 fprintf(out,"\n");
2703 plp = plp->next;
2704 }
2705}
2706#endif
2707
2708/* Print an action to the given file descriptor. Return FALSE if
2709** nothing was actually printed.
2710*/
2711int PrintAction(struct action *ap, FILE *fp, int indent){
2712 int result = 1;
2713 switch( ap->type ){
2714 case SHIFT:
2715 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->index);
2716 break;
2717 case REDUCE:
2718 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
2719 break;
2720 case ACCEPT:
2721 fprintf(fp,"%*s accept",indent,ap->sp->name);
2722 break;
2723 case ERROR:
2724 fprintf(fp,"%*s error",indent,ap->sp->name);
2725 break;
2726 case CONFLICT:
2727 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
2728 indent,ap->sp->name,ap->x.rp->index);
2729 break;
2730 case SH_RESOLVED:
2731 case RD_RESOLVED:
2732 case NOT_USED:
2733 result = 0;
2734 break;
2735 }
2736 return result;
2737}
2738
2739/* Generate the "y.output" log file */
2740void ReportOutput(lemp)
2741struct lemon *lemp;
2742{
2743 int i;
2744 struct state *stp;
2745 struct config *cfp;
2746 struct action *ap;
2747 FILE *fp;
2748
drh2aa6ca42004-09-10 00:14:04 +00002749 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00002750 if( fp==0 ) return;
2751 fprintf(fp," \b");
2752 for(i=0; i<lemp->nstate; i++){
2753 stp = lemp->sorted[i];
2754 fprintf(fp,"State %d:\n",stp->index);
2755 if( lemp->basisflag ) cfp=stp->bp;
2756 else cfp=stp->cfp;
2757 while( cfp ){
2758 char buf[20];
2759 if( cfp->dot==cfp->rp->nrhs ){
2760 sprintf(buf,"(%d)",cfp->rp->index);
2761 fprintf(fp," %5s ",buf);
2762 }else{
2763 fprintf(fp," ");
2764 }
2765 ConfigPrint(fp,cfp);
2766 fprintf(fp,"\n");
2767#ifdef TEST
2768 SetPrint(fp,cfp->fws,lemp);
2769 PlinkPrint(fp,cfp->fplp,"To ");
2770 PlinkPrint(fp,cfp->bplp,"From");
2771#endif
2772 if( lemp->basisflag ) cfp=cfp->bp;
2773 else cfp=cfp->next;
2774 }
2775 fprintf(fp,"\n");
2776 for(ap=stp->ap; ap; ap=ap->next){
2777 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
2778 }
2779 fprintf(fp,"\n");
2780 }
2781 fclose(fp);
2782 return;
2783}
2784
2785/* Search for the file "name" which is in the same directory as
2786** the exacutable */
2787PRIVATE char *pathsearch(argv0,name,modemask)
2788char *argv0;
2789char *name;
2790int modemask;
2791{
2792 char *pathlist;
2793 char *path,*cp;
2794 char c;
2795 extern int access();
2796
2797#ifdef __WIN32__
2798 cp = strrchr(argv0,'\\');
2799#else
2800 cp = strrchr(argv0,'/');
2801#endif
2802 if( cp ){
2803 c = *cp;
2804 *cp = 0;
2805 path = (char *)malloc( strlen(argv0) + strlen(name) + 2 );
2806 if( path ) sprintf(path,"%s/%s",argv0,name);
2807 *cp = c;
2808 }else{
2809 extern char *getenv();
2810 pathlist = getenv("PATH");
2811 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
2812 path = (char *)malloc( strlen(pathlist)+strlen(name)+2 );
2813 if( path!=0 ){
2814 while( *pathlist ){
2815 cp = strchr(pathlist,':');
2816 if( cp==0 ) cp = &pathlist[strlen(pathlist)];
2817 c = *cp;
2818 *cp = 0;
2819 sprintf(path,"%s/%s",pathlist,name);
2820 *cp = c;
2821 if( c==0 ) pathlist = "";
2822 else pathlist = &cp[1];
2823 if( access(path,modemask)==0 ) break;
2824 }
2825 }
2826 }
2827 return path;
2828}
2829
2830/* Given an action, compute the integer value for that action
2831** which is to be put in the action table of the generated machine.
2832** Return negative if no action should be generated.
2833*/
2834PRIVATE int compute_action(lemp,ap)
2835struct lemon *lemp;
2836struct action *ap;
2837{
2838 int act;
2839 switch( ap->type ){
2840 case SHIFT: act = ap->x.stp->index; break;
2841 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
2842 case ERROR: act = lemp->nstate + lemp->nrule; break;
2843 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
2844 default: act = -1; break;
2845 }
2846 return act;
2847}
2848
2849#define LINESIZE 1000
2850/* The next cluster of routines are for reading the template file
2851** and writing the results to the generated parser */
2852/* The first function transfers data from "in" to "out" until
2853** a line is seen which begins with "%%". The line number is
2854** tracked.
2855**
2856** if name!=0, then any word that begin with "Parse" is changed to
2857** begin with *name instead.
2858*/
2859PRIVATE void tplt_xfer(name,in,out,lineno)
2860char *name;
2861FILE *in;
2862FILE *out;
2863int *lineno;
2864{
2865 int i, iStart;
2866 char line[LINESIZE];
2867 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
2868 (*lineno)++;
2869 iStart = 0;
2870 if( name ){
2871 for(i=0; line[i]; i++){
2872 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
2873 && (i==0 || !isalpha(line[i-1]))
2874 ){
2875 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
2876 fprintf(out,"%s",name);
2877 i += 4;
2878 iStart = i+1;
2879 }
2880 }
2881 }
2882 fprintf(out,"%s",&line[iStart]);
2883 }
2884}
2885
2886/* The next function finds the template file and opens it, returning
2887** a pointer to the opened file. */
2888PRIVATE FILE *tplt_open(lemp)
2889struct lemon *lemp;
2890{
2891 static char templatename[] = "lempar.c";
2892 char buf[1000];
2893 FILE *in;
2894 char *tpltname;
2895 char *cp;
2896
2897 cp = strrchr(lemp->filename,'.');
2898 if( cp ){
drh8b582012003-10-21 13:16:03 +00002899 sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00002900 }else{
2901 sprintf(buf,"%s.lt",lemp->filename);
2902 }
2903 if( access(buf,004)==0 ){
2904 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00002905 }else if( access(templatename,004)==0 ){
2906 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00002907 }else{
2908 tpltname = pathsearch(lemp->argv0,templatename,0);
2909 }
2910 if( tpltname==0 ){
2911 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
2912 templatename);
2913 lemp->errorcnt++;
2914 return 0;
2915 }
drh2aa6ca42004-09-10 00:14:04 +00002916 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00002917 if( in==0 ){
2918 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
2919 lemp->errorcnt++;
2920 return 0;
2921 }
2922 return in;
2923}
2924
drhaf805ca2004-09-07 11:28:25 +00002925/* Print a #line directive line to the output file. */
2926PRIVATE void tplt_linedir(out,lineno,filename)
2927FILE *out;
2928int lineno;
2929char *filename;
2930{
2931 fprintf(out,"#line %d \"",lineno);
2932 while( *filename ){
2933 if( *filename == '\\' ) putc('\\',out);
2934 putc(*filename,out);
2935 filename++;
2936 }
2937 fprintf(out,"\"\n");
2938}
2939
drh75897232000-05-29 14:26:00 +00002940/* Print a string to the file and keep the linenumber up to date */
2941PRIVATE void tplt_print(out,lemp,str,strln,lineno)
2942FILE *out;
2943struct lemon *lemp;
2944char *str;
2945int strln;
2946int *lineno;
2947{
2948 if( str==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00002949 tplt_linedir(out,strln,lemp->filename);
2950 (*lineno)++;
drh75897232000-05-29 14:26:00 +00002951 while( *str ){
2952 if( *str=='\n' ) (*lineno)++;
2953 putc(*str,out);
2954 str++;
2955 }
drh9db55df2004-09-09 14:01:21 +00002956 if( str[-1]!='\n' ){
2957 putc('\n',out);
2958 (*lineno)++;
2959 }
drhaf805ca2004-09-07 11:28:25 +00002960 tplt_linedir(out,*lineno+2,lemp->outname);
2961 (*lineno)+=2;
drh75897232000-05-29 14:26:00 +00002962 return;
2963}
2964
2965/*
2966** The following routine emits code for the destructor for the
2967** symbol sp
2968*/
2969void emit_destructor_code(out,sp,lemp,lineno)
2970FILE *out;
2971struct symbol *sp;
2972struct lemon *lemp;
2973int *lineno;
2974{
drhcc83b6e2004-04-23 23:38:42 +00002975 char *cp = 0;
drh75897232000-05-29 14:26:00 +00002976
2977 int linecnt = 0;
2978 if( sp->type==TERMINAL ){
2979 cp = lemp->tokendest;
2980 if( cp==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00002981 tplt_linedir(out,lemp->tokendestln,lemp->filename);
2982 fprintf(out,"{");
drh960e8c62001-04-03 16:53:21 +00002983 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00002984 cp = sp->destructor;
drhaf805ca2004-09-07 11:28:25 +00002985 tplt_linedir(out,sp->destructorln,lemp->filename);
2986 fprintf(out,"{");
drh960e8c62001-04-03 16:53:21 +00002987 }else if( lemp->vardest ){
2988 cp = lemp->vardest;
2989 if( cp==0 ) return;
drhaf805ca2004-09-07 11:28:25 +00002990 tplt_linedir(out,lemp->vardestln,lemp->filename);
2991 fprintf(out,"{");
drhcc83b6e2004-04-23 23:38:42 +00002992 }else{
2993 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00002994 }
2995 for(; *cp; cp++){
2996 if( *cp=='$' && cp[1]=='$' ){
2997 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
2998 cp++;
2999 continue;
3000 }
3001 if( *cp=='\n' ) linecnt++;
3002 fputc(*cp,out);
3003 }
3004 (*lineno) += 3 + linecnt;
drhaf805ca2004-09-07 11:28:25 +00003005 fprintf(out,"}\n");
3006 tplt_linedir(out,*lineno,lemp->outname);
drh75897232000-05-29 14:26:00 +00003007 return;
3008}
3009
3010/*
drh960e8c62001-04-03 16:53:21 +00003011** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003012*/
3013int has_destructor(sp, lemp)
3014struct symbol *sp;
3015struct lemon *lemp;
3016{
3017 int ret;
3018 if( sp->type==TERMINAL ){
3019 ret = lemp->tokendest!=0;
3020 }else{
drh960e8c62001-04-03 16:53:21 +00003021 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003022 }
3023 return ret;
3024}
3025
drh0bb132b2004-07-20 14:06:51 +00003026/*
3027** Append text to a dynamically allocated string. If zText is 0 then
3028** reset the string to be empty again. Always return the complete text
3029** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003030**
3031** n bytes of zText are stored. If n==0 then all of zText up to the first
3032** \000 terminator is stored. zText can contain up to two instances of
3033** %d. The values of p1 and p2 are written into the first and second
3034** %d.
3035**
3036** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003037*/
3038PRIVATE char *append_str(char *zText, int n, int p1, int p2){
3039 static char *z = 0;
3040 static int alloced = 0;
3041 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003042 int c;
drh0bb132b2004-07-20 14:06:51 +00003043 char zInt[40];
3044
3045 if( zText==0 ){
3046 used = 0;
3047 return z;
3048 }
drh7ac25c72004-08-19 15:12:26 +00003049 if( n<=0 ){
3050 if( n<0 ){
3051 used += n;
3052 assert( used>=0 );
3053 }
3054 n = strlen(zText);
3055 }
drh0bb132b2004-07-20 14:06:51 +00003056 if( n+sizeof(zInt)*2+used >= alloced ){
3057 alloced = n + sizeof(zInt)*2 + used + 200;
3058 z = realloc(z, alloced);
3059 }
3060 if( z==0 ) return "";
3061 while( n-- > 0 ){
3062 c = *(zText++);
3063 if( c=='%' && zText[0]=='d' ){
3064 sprintf(zInt, "%d", p1);
3065 p1 = p2;
3066 strcpy(&z[used], zInt);
3067 used += strlen(&z[used]);
3068 zText++;
3069 n--;
3070 }else{
3071 z[used++] = c;
3072 }
3073 }
3074 z[used] = 0;
3075 return z;
3076}
3077
3078/*
3079** zCode is a string that is the action associated with a rule. Expand
3080** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003081** stack.
drh0bb132b2004-07-20 14:06:51 +00003082*/
drhaf805ca2004-09-07 11:28:25 +00003083PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003084 char *cp, *xp;
3085 int i;
3086 char lhsused = 0; /* True if the LHS element has been used */
3087 char used[MAXRHS]; /* True for each RHS element which is used */
3088
3089 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3090 lhsused = 0;
3091
3092 append_str(0,0,0,0);
3093 for(cp=rp->code; *cp; cp++){
3094 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3095 char saved;
3096 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3097 saved = *xp;
3098 *xp = 0;
3099 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003100 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003101 cp = xp;
3102 lhsused = 1;
3103 }else{
3104 for(i=0; i<rp->nrhs; i++){
3105 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003106 if( cp!=rp->code && cp[-1]=='@' ){
3107 /* If the argument is of the form @X then substituted
3108 ** the token number of X, not the value of X */
3109 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3110 }else{
3111 append_str("yymsp[%d].minor.yy%d",0,
3112 i-rp->nrhs+1,rp->rhs[i]->dtnum);
3113 }
drh0bb132b2004-07-20 14:06:51 +00003114 cp = xp;
3115 used[i] = 1;
3116 break;
3117 }
3118 }
3119 }
3120 *xp = saved;
3121 }
3122 append_str(cp, 1, 0, 0);
3123 } /* End loop */
3124
3125 /* Check to make sure the LHS has been used */
3126 if( rp->lhsalias && !lhsused ){
3127 ErrorMsg(lemp->filename,rp->ruleline,
3128 "Label \"%s\" for \"%s(%s)\" is never used.",
3129 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3130 lemp->errorcnt++;
3131 }
3132
3133 /* Generate destructor code for RHS symbols which are not used in the
3134 ** reduce code */
3135 for(i=0; i<rp->nrhs; i++){
3136 if( rp->rhsalias[i] && !used[i] ){
3137 ErrorMsg(lemp->filename,rp->ruleline,
3138 "Label %s for \"%s(%s)\" is never used.",
3139 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3140 lemp->errorcnt++;
3141 }else if( rp->rhsalias[i]==0 ){
3142 if( has_destructor(rp->rhs[i],lemp) ){
drh7ac25c72004-08-19 15:12:26 +00003143 append_str(" yy_destructor(%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003144 rp->rhs[i]->index,i-rp->nrhs+1);
3145 }else{
3146 /* No destructor defined for this term */
3147 }
3148 }
3149 }
3150 cp = append_str(0,0,0,0);
3151 rp->code = Strsafe(cp);
3152}
3153
drh75897232000-05-29 14:26:00 +00003154/*
3155** Generate code which executes when the rule "rp" is reduced. Write
3156** the code to "out". Make sure lineno stays up-to-date.
3157*/
3158PRIVATE void emit_code(out,rp,lemp,lineno)
3159FILE *out;
3160struct rule *rp;
3161struct lemon *lemp;
3162int *lineno;
3163{
drh0bb132b2004-07-20 14:06:51 +00003164 char *cp;
drh75897232000-05-29 14:26:00 +00003165 int linecnt = 0;
drh75897232000-05-29 14:26:00 +00003166
3167 /* Generate code to do the reduce action */
3168 if( rp->code ){
drhaf805ca2004-09-07 11:28:25 +00003169 tplt_linedir(out,rp->line,lemp->filename);
3170 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003171 for(cp=rp->code; *cp; cp++){
drh75897232000-05-29 14:26:00 +00003172 if( *cp=='\n' ) linecnt++;
drh75897232000-05-29 14:26:00 +00003173 } /* End loop */
3174 (*lineno) += 3 + linecnt;
drhaf805ca2004-09-07 11:28:25 +00003175 fprintf(out,"}\n");
3176 tplt_linedir(out,*lineno,lemp->outname);
drh75897232000-05-29 14:26:00 +00003177 } /* End if( rp->code ) */
3178
drh75897232000-05-29 14:26:00 +00003179 return;
3180}
3181
3182/*
3183** Print the definition of the union used for the parser's data stack.
3184** This union contains fields for every possible data type for tokens
3185** and nonterminals. In the process of computing and printing this
3186** union, also set the ".dtnum" field of every terminal and nonterminal
3187** symbol.
3188*/
3189void print_stack_union(out,lemp,plineno,mhflag)
3190FILE *out; /* The output stream */
3191struct lemon *lemp; /* The main info structure for this parser */
3192int *plineno; /* Pointer to the line number */
3193int mhflag; /* True if generating makeheaders output */
3194{
3195 int lineno = *plineno; /* The line number of the output */
3196 char **types; /* A hash table of datatypes */
3197 int arraysize; /* Size of the "types" array */
3198 int maxdtlength; /* Maximum length of any ".datatype" field. */
3199 char *stddt; /* Standardized name for a datatype */
3200 int i,j; /* Loop counters */
3201 int hash; /* For hashing the name of a type */
3202 char *name; /* Name of the parser */
3203
3204 /* Allocate and initialize types[] and allocate stddt[] */
3205 arraysize = lemp->nsymbol * 2;
3206 types = (char**)malloc( arraysize * sizeof(char*) );
3207 for(i=0; i<arraysize; i++) types[i] = 0;
3208 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003209 if( lemp->vartype ){
3210 maxdtlength = strlen(lemp->vartype);
3211 }
drh75897232000-05-29 14:26:00 +00003212 for(i=0; i<lemp->nsymbol; i++){
3213 int len;
3214 struct symbol *sp = lemp->symbols[i];
3215 if( sp->datatype==0 ) continue;
3216 len = strlen(sp->datatype);
3217 if( len>maxdtlength ) maxdtlength = len;
3218 }
3219 stddt = (char*)malloc( maxdtlength*2 + 1 );
3220 if( types==0 || stddt==0 ){
3221 fprintf(stderr,"Out of memory.\n");
3222 exit(1);
3223 }
3224
3225 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3226 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003227 ** used for terminal symbols. If there is no %default_type defined then
3228 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3229 ** a datatype using the %type directive.
3230 */
drh75897232000-05-29 14:26:00 +00003231 for(i=0; i<lemp->nsymbol; i++){
3232 struct symbol *sp = lemp->symbols[i];
3233 char *cp;
3234 if( sp==lemp->errsym ){
3235 sp->dtnum = arraysize+1;
3236 continue;
3237 }
drh960e8c62001-04-03 16:53:21 +00003238 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003239 sp->dtnum = 0;
3240 continue;
3241 }
3242 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003243 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003244 j = 0;
3245 while( isspace(*cp) ) cp++;
3246 while( *cp ) stddt[j++] = *cp++;
3247 while( j>0 && isspace(stddt[j-1]) ) j--;
3248 stddt[j] = 0;
3249 hash = 0;
3250 for(j=0; stddt[j]; j++){
3251 hash = hash*53 + stddt[j];
3252 }
drh3b2129c2003-05-13 00:34:21 +00003253 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003254 while( types[hash] ){
3255 if( strcmp(types[hash],stddt)==0 ){
3256 sp->dtnum = hash + 1;
3257 break;
3258 }
3259 hash++;
3260 if( hash>=arraysize ) hash = 0;
3261 }
3262 if( types[hash]==0 ){
3263 sp->dtnum = hash + 1;
3264 types[hash] = (char*)malloc( strlen(stddt)+1 );
3265 if( types[hash]==0 ){
3266 fprintf(stderr,"Out of memory.\n");
3267 exit(1);
3268 }
3269 strcpy(types[hash],stddt);
3270 }
3271 }
3272
3273 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3274 name = lemp->name ? lemp->name : "Parse";
3275 lineno = *plineno;
3276 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3277 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3278 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3279 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3280 fprintf(out,"typedef union {\n"); lineno++;
3281 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3282 for(i=0; i<arraysize; i++){
3283 if( types[i]==0 ) continue;
3284 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3285 free(types[i]);
3286 }
3287 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3288 free(stddt);
3289 free(types);
3290 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3291 *plineno = lineno;
3292}
3293
drhb29b0a52002-02-23 19:39:46 +00003294/*
3295** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003296** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003297*/
drh8b582012003-10-21 13:16:03 +00003298static const char *minimum_size_type(int lwr, int upr){
3299 if( lwr>=0 ){
3300 if( upr<=255 ){
3301 return "unsigned char";
3302 }else if( upr<65535 ){
3303 return "unsigned short int";
3304 }else{
3305 return "unsigned int";
3306 }
3307 }else if( lwr>=-127 && upr<=127 ){
3308 return "signed char";
3309 }else if( lwr>=-32767 && upr<32767 ){
3310 return "short";
drhb29b0a52002-02-23 19:39:46 +00003311 }else{
drh8b582012003-10-21 13:16:03 +00003312 return "int";
drhb29b0a52002-02-23 19:39:46 +00003313 }
3314}
3315
drhfdbf9282003-10-21 16:34:41 +00003316/*
3317** Each state contains a set of token transaction and a set of
3318** nonterminal transactions. Each of these sets makes an instance
3319** of the following structure. An array of these structures is used
3320** to order the creation of entries in the yy_action[] table.
3321*/
3322struct axset {
3323 struct state *stp; /* A pointer to a state */
3324 int isTkn; /* True to use tokens. False for non-terminals */
3325 int nAction; /* Number of actions */
3326};
3327
3328/*
3329** Compare to axset structures for sorting purposes
3330*/
3331static int axset_compare(const void *a, const void *b){
3332 struct axset *p1 = (struct axset*)a;
3333 struct axset *p2 = (struct axset*)b;
3334 return p2->nAction - p1->nAction;
3335}
3336
drh75897232000-05-29 14:26:00 +00003337/* Generate C source code for the parser */
3338void ReportTable(lemp, mhflag)
3339struct lemon *lemp;
3340int mhflag; /* Output in makeheaders format if true */
3341{
3342 FILE *out, *in;
3343 char line[LINESIZE];
3344 int lineno;
3345 struct state *stp;
3346 struct action *ap;
3347 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003348 struct acttab *pActtab;
3349 int i, j, n;
drh75897232000-05-29 14:26:00 +00003350 char *name;
drh8b582012003-10-21 13:16:03 +00003351 int mnTknOfst, mxTknOfst;
3352 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003353 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003354
3355 in = tplt_open(lemp);
3356 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003357 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003358 if( out==0 ){
3359 fclose(in);
3360 return;
3361 }
3362 lineno = 1;
3363 tplt_xfer(lemp->name,in,out,&lineno);
3364
3365 /* Generate the include code, if any */
3366 tplt_print(out,lemp,lemp->include,lemp->includeln,&lineno);
3367 if( mhflag ){
3368 char *name = file_makename(lemp, ".h");
3369 fprintf(out,"#include \"%s\"\n", name); lineno++;
3370 free(name);
3371 }
3372 tplt_xfer(lemp->name,in,out,&lineno);
3373
3374 /* Generate #defines for all tokens */
3375 if( mhflag ){
3376 char *prefix;
3377 fprintf(out,"#if INTERFACE\n"); lineno++;
3378 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3379 else prefix = "";
3380 for(i=1; i<lemp->nterminal; i++){
3381 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3382 lineno++;
3383 }
3384 fprintf(out,"#endif\n"); lineno++;
3385 }
3386 tplt_xfer(lemp->name,in,out,&lineno);
3387
3388 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003389 fprintf(out,"#define YYCODETYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003390 minimum_size_type(0, lemp->nsymbol+5)); lineno++;
drh75897232000-05-29 14:26:00 +00003391 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3392 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003393 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drh75897232000-05-29 14:26:00 +00003394 print_stack_union(out,lemp,&lineno,mhflag);
3395 if( lemp->stacksize ){
3396 if( atoi(lemp->stacksize)<=0 ){
3397 ErrorMsg(lemp->filename,0,
3398"Illegal stack size: [%s]. The stack size should be an integer constant.",
3399 lemp->stacksize);
3400 lemp->errorcnt++;
3401 lemp->stacksize = "100";
3402 }
3403 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3404 }else{
3405 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3406 }
3407 if( mhflag ){
3408 fprintf(out,"#if INTERFACE\n"); lineno++;
3409 }
3410 name = lemp->name ? lemp->name : "Parse";
3411 if( lemp->arg && lemp->arg[0] ){
3412 int i;
3413 i = strlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003414 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3415 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003416 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3417 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3418 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3419 name,lemp->arg,&lemp->arg[i]); lineno++;
3420 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3421 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003422 }else{
drh1f245e42002-03-11 13:55:50 +00003423 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3424 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3425 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3426 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003427 }
3428 if( mhflag ){
3429 fprintf(out,"#endif\n"); lineno++;
3430 }
3431 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3432 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
3433 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3434 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
drh0bd1f4e2002-06-06 18:54:39 +00003435 if( lemp->has_fallback ){
3436 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3437 }
drh75897232000-05-29 14:26:00 +00003438 tplt_xfer(lemp->name,in,out,&lineno);
3439
drh8b582012003-10-21 13:16:03 +00003440 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003441 **
drh8b582012003-10-21 13:16:03 +00003442 ** yy_action[] A single table containing all actions.
3443 ** yy_lookahead[] A table containing the lookahead for each entry in
3444 ** yy_action. Used to detect hash collisions.
3445 ** yy_shift_ofst[] For each state, the offset into yy_action for
3446 ** shifting terminals.
3447 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3448 ** shifting non-terminals after a reduce.
3449 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003450 */
drh75897232000-05-29 14:26:00 +00003451
drh8b582012003-10-21 13:16:03 +00003452 /* Compute the actions on all states and count them up */
drhfdbf9282003-10-21 16:34:41 +00003453 ax = malloc( sizeof(ax[0])*lemp->nstate*2 );
3454 if( ax==0 ){
3455 fprintf(stderr,"malloc failed\n");
3456 exit(1);
3457 }
drh75897232000-05-29 14:26:00 +00003458 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003459 stp = lemp->sorted[i];
drh8b582012003-10-21 13:16:03 +00003460 stp->nTknAct = stp->nNtAct = 0;
3461 stp->iDflt = lemp->nstate + lemp->nrule;
3462 stp->iTknOfst = NO_OFFSET;
3463 stp->iNtOfst = NO_OFFSET;
3464 for(ap=stp->ap; ap; ap=ap->next){
3465 if( compute_action(lemp,ap)>=0 ){
3466 if( ap->sp->index<lemp->nterminal ){
3467 stp->nTknAct++;
3468 }else if( ap->sp->index<lemp->nsymbol ){
3469 stp->nNtAct++;
3470 }else{
3471 stp->iDflt = compute_action(lemp, ap);
3472 }
3473 }
3474 }
drhfdbf9282003-10-21 16:34:41 +00003475 ax[i*2].stp = stp;
3476 ax[i*2].isTkn = 1;
3477 ax[i*2].nAction = stp->nTknAct;
3478 ax[i*2+1].stp = stp;
3479 ax[i*2+1].isTkn = 0;
3480 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003481 }
drh8b582012003-10-21 13:16:03 +00003482 mxTknOfst = mnTknOfst = 0;
3483 mxNtOfst = mnNtOfst = 0;
3484
drhfdbf9282003-10-21 16:34:41 +00003485 /* Compute the action table. In order to try to keep the size of the
3486 ** action table to a minimum, the heuristic of placing the largest action
3487 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003488 */
drhfdbf9282003-10-21 16:34:41 +00003489 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003490 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003491 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3492 stp = ax[i].stp;
3493 if( ax[i].isTkn ){
3494 for(ap=stp->ap; ap; ap=ap->next){
3495 int action;
3496 if( ap->sp->index>=lemp->nterminal ) continue;
3497 action = compute_action(lemp, ap);
3498 if( action<0 ) continue;
3499 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003500 }
drhfdbf9282003-10-21 16:34:41 +00003501 stp->iTknOfst = acttab_insert(pActtab);
3502 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3503 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3504 }else{
3505 for(ap=stp->ap; ap; ap=ap->next){
3506 int action;
3507 if( ap->sp->index<lemp->nterminal ) continue;
3508 if( ap->sp->index==lemp->nsymbol ) continue;
3509 action = compute_action(lemp, ap);
3510 if( action<0 ) continue;
3511 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003512 }
drhfdbf9282003-10-21 16:34:41 +00003513 stp->iNtOfst = acttab_insert(pActtab);
3514 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3515 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003516 }
3517 }
drhfdbf9282003-10-21 16:34:41 +00003518 free(ax);
drh8b582012003-10-21 13:16:03 +00003519
3520 /* Output the yy_action table */
drh57196282004-10-06 15:41:16 +00003521 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003522 n = acttab_size(pActtab);
3523 for(i=j=0; i<n; i++){
3524 int action = acttab_yyaction(pActtab, i);
3525 if( action<0 ) action = lemp->nsymbol + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003526 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003527 fprintf(out, " %4d,", action);
3528 if( j==9 || i==n-1 ){
3529 fprintf(out, "\n"); lineno++;
3530 j = 0;
3531 }else{
3532 j++;
3533 }
3534 }
3535 fprintf(out, "};\n"); lineno++;
3536
3537 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003538 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003539 for(i=j=0; i<n; i++){
3540 int la = acttab_yylookahead(pActtab, i);
3541 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003542 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003543 fprintf(out, " %4d,", la);
3544 if( j==9 || i==n-1 ){
3545 fprintf(out, "\n"); lineno++;
3546 j = 0;
3547 }else{
3548 j++;
3549 }
3550 }
3551 fprintf(out, "};\n"); lineno++;
3552
3553 /* Output the yy_shift_ofst[] table */
3554 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drh57196282004-10-06 15:41:16 +00003555 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003556 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
3557 n = lemp->nstate;
3558 for(i=j=0; i<n; i++){
3559 int ofst;
3560 stp = lemp->sorted[i];
3561 ofst = stp->iTknOfst;
3562 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003563 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003564 fprintf(out, " %4d,", ofst);
3565 if( j==9 || i==n-1 ){
3566 fprintf(out, "\n"); lineno++;
3567 j = 0;
3568 }else{
3569 j++;
3570 }
3571 }
3572 fprintf(out, "};\n"); lineno++;
3573
3574 /* Output the yy_reduce_ofst[] table */
3575 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drh57196282004-10-06 15:41:16 +00003576 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003577 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
3578 n = lemp->nstate;
3579 for(i=j=0; i<n; i++){
3580 int ofst;
3581 stp = lemp->sorted[i];
3582 ofst = stp->iNtOfst;
3583 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003584 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003585 fprintf(out, " %4d,", ofst);
3586 if( j==9 || i==n-1 ){
3587 fprintf(out, "\n"); lineno++;
3588 j = 0;
3589 }else{
3590 j++;
3591 }
3592 }
3593 fprintf(out, "};\n"); lineno++;
3594
3595 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003596 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003597 n = lemp->nstate;
3598 for(i=j=0; i<n; i++){
3599 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003600 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003601 fprintf(out, " %4d,", stp->iDflt);
3602 if( j==9 || i==n-1 ){
3603 fprintf(out, "\n"); lineno++;
3604 j = 0;
3605 }else{
3606 j++;
3607 }
3608 }
3609 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003610 tplt_xfer(lemp->name,in,out,&lineno);
3611
drh0bd1f4e2002-06-06 18:54:39 +00003612 /* Generate the table of fallback tokens.
3613 */
3614 if( lemp->has_fallback ){
3615 for(i=0; i<lemp->nterminal; i++){
3616 struct symbol *p = lemp->symbols[i];
3617 if( p->fallback==0 ){
3618 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3619 }else{
3620 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3621 p->name, p->fallback->name);
3622 }
3623 lineno++;
3624 }
3625 }
3626 tplt_xfer(lemp->name, in, out, &lineno);
3627
3628 /* Generate a table containing the symbolic name of every symbol
3629 */
drh75897232000-05-29 14:26:00 +00003630 for(i=0; i<lemp->nsymbol; i++){
3631 sprintf(line,"\"%s\",",lemp->symbols[i]->name);
3632 fprintf(out," %-15s",line);
3633 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
3634 }
3635 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
3636 tplt_xfer(lemp->name,in,out,&lineno);
3637
drh0bd1f4e2002-06-06 18:54:39 +00003638 /* Generate a table containing a text string that describes every
3639 ** rule in the rule set of the grammer. This information is used
3640 ** when tracing REDUCE actions.
3641 */
3642 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
3643 assert( rp->index==i );
3644 fprintf(out," /* %3d */ \"%s ::=", i, rp->lhs->name);
3645 for(j=0; j<rp->nrhs; j++) fprintf(out," %s",rp->rhs[j]->name);
3646 fprintf(out,"\",\n"); lineno++;
3647 }
3648 tplt_xfer(lemp->name,in,out,&lineno);
3649
drh75897232000-05-29 14:26:00 +00003650 /* Generate code which executes every time a symbol is popped from
3651 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00003652 ** (In other words, generate the %destructor actions)
3653 */
drh75897232000-05-29 14:26:00 +00003654 if( lemp->tokendest ){
3655 for(i=0; i<lemp->nsymbol; i++){
3656 struct symbol *sp = lemp->symbols[i];
3657 if( sp==0 || sp->type!=TERMINAL ) continue;
3658 fprintf(out," case %d:\n",sp->index); lineno++;
3659 }
3660 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
3661 if( i<lemp->nsymbol ){
3662 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3663 fprintf(out," break;\n"); lineno++;
3664 }
3665 }
drh8d659732005-01-13 23:54:06 +00003666 if( lemp->vardest ){
3667 struct symbol *dflt_sp = 0;
3668 for(i=0; i<lemp->nsymbol; i++){
3669 struct symbol *sp = lemp->symbols[i];
3670 if( sp==0 || sp->type==TERMINAL ||
3671 sp->index<=0 || sp->destructor!=0 ) continue;
3672 fprintf(out," case %d:\n",sp->index); lineno++;
3673 dflt_sp = sp;
3674 }
3675 if( dflt_sp!=0 ){
3676 emit_destructor_code(out,dflt_sp,lemp,&lineno);
3677 fprintf(out," break;\n"); lineno++;
3678 }
3679 }
drh75897232000-05-29 14:26:00 +00003680 for(i=0; i<lemp->nsymbol; i++){
3681 struct symbol *sp = lemp->symbols[i];
3682 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
3683 fprintf(out," case %d:\n",sp->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003684
3685 /* Combine duplicate destructors into a single case */
3686 for(j=i+1; j<lemp->nsymbol; j++){
3687 struct symbol *sp2 = lemp->symbols[j];
3688 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
3689 && sp2->dtnum==sp->dtnum
3690 && strcmp(sp->destructor,sp2->destructor)==0 ){
3691 fprintf(out," case %d:\n",sp2->index); lineno++;
3692 sp2->destructor = 0;
3693 }
3694 }
3695
drh75897232000-05-29 14:26:00 +00003696 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3697 fprintf(out," break;\n"); lineno++;
3698 }
drh75897232000-05-29 14:26:00 +00003699 tplt_xfer(lemp->name,in,out,&lineno);
3700
3701 /* Generate code which executes whenever the parser stack overflows */
3702 tplt_print(out,lemp,lemp->overflow,lemp->overflowln,&lineno);
3703 tplt_xfer(lemp->name,in,out,&lineno);
3704
3705 /* Generate the table of rule information
3706 **
3707 ** Note: This code depends on the fact that rules are number
3708 ** sequentually beginning with 0.
3709 */
3710 for(rp=lemp->rule; rp; rp=rp->next){
3711 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
3712 }
3713 tplt_xfer(lemp->name,in,out,&lineno);
3714
3715 /* Generate code which execution during each REDUCE action */
3716 for(rp=lemp->rule; rp; rp=rp->next){
drh0bb132b2004-07-20 14:06:51 +00003717 if( rp->code ) translate_code(lemp, rp);
3718 }
3719 for(rp=lemp->rule; rp; rp=rp->next){
3720 struct rule *rp2;
3721 if( rp->code==0 ) continue;
drh75897232000-05-29 14:26:00 +00003722 fprintf(out," case %d:\n",rp->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003723 for(rp2=rp->next; rp2; rp2=rp2->next){
3724 if( rp2->code==rp->code ){
3725 fprintf(out," case %d:\n",rp2->index); lineno++;
3726 rp2->code = 0;
3727 }
3728 }
drh75897232000-05-29 14:26:00 +00003729 emit_code(out,rp,lemp,&lineno);
3730 fprintf(out," break;\n"); lineno++;
3731 }
3732 tplt_xfer(lemp->name,in,out,&lineno);
3733
3734 /* Generate code which executes if a parse fails */
3735 tplt_print(out,lemp,lemp->failure,lemp->failureln,&lineno);
3736 tplt_xfer(lemp->name,in,out,&lineno);
3737
3738 /* Generate code which executes when a syntax error occurs */
3739 tplt_print(out,lemp,lemp->error,lemp->errorln,&lineno);
3740 tplt_xfer(lemp->name,in,out,&lineno);
3741
3742 /* Generate code which executes when the parser accepts its input */
3743 tplt_print(out,lemp,lemp->accept,lemp->acceptln,&lineno);
3744 tplt_xfer(lemp->name,in,out,&lineno);
3745
3746 /* Append any addition code the user desires */
3747 tplt_print(out,lemp,lemp->extracode,lemp->extracodeln,&lineno);
3748
3749 fclose(in);
3750 fclose(out);
3751 return;
3752}
3753
3754/* Generate a header file for the parser */
3755void ReportHeader(lemp)
3756struct lemon *lemp;
3757{
3758 FILE *out, *in;
3759 char *prefix;
3760 char line[LINESIZE];
3761 char pattern[LINESIZE];
3762 int i;
3763
3764 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3765 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00003766 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00003767 if( in ){
3768 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
3769 sprintf(pattern,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3770 if( strcmp(line,pattern) ) break;
3771 }
3772 fclose(in);
3773 if( i==lemp->nterminal ){
3774 /* No change in the file. Don't rewrite it. */
3775 return;
3776 }
3777 }
drh2aa6ca42004-09-10 00:14:04 +00003778 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00003779 if( out ){
3780 for(i=1; i<lemp->nterminal; i++){
3781 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3782 }
3783 fclose(out);
3784 }
3785 return;
3786}
3787
3788/* Reduce the size of the action tables, if possible, by making use
3789** of defaults.
3790**
drhb59499c2002-02-23 18:45:13 +00003791** In this version, we take the most frequent REDUCE action and make
3792** it the default. Only default a reduce if there are more than one.
drh75897232000-05-29 14:26:00 +00003793*/
3794void CompressTables(lemp)
3795struct lemon *lemp;
3796{
3797 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00003798 struct action *ap, *ap2;
3799 struct rule *rp, *rp2, *rbest;
3800 int nbest, n;
drh75897232000-05-29 14:26:00 +00003801 int i;
drh75897232000-05-29 14:26:00 +00003802
3803 for(i=0; i<lemp->nstate; i++){
3804 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00003805 nbest = 0;
3806 rbest = 0;
drh75897232000-05-29 14:26:00 +00003807
drhb59499c2002-02-23 18:45:13 +00003808 for(ap=stp->ap; ap; ap=ap->next){
3809 if( ap->type!=REDUCE ) continue;
3810 rp = ap->x.rp;
3811 if( rp==rbest ) continue;
3812 n = 1;
3813 for(ap2=ap->next; ap2; ap2=ap2->next){
3814 if( ap2->type!=REDUCE ) continue;
3815 rp2 = ap2->x.rp;
3816 if( rp2==rbest ) continue;
3817 if( rp2==rp ) n++;
3818 }
3819 if( n>nbest ){
3820 nbest = n;
3821 rbest = rp;
drh75897232000-05-29 14:26:00 +00003822 }
3823 }
drhb59499c2002-02-23 18:45:13 +00003824
3825 /* Do not make a default if the number of rules to default
3826 ** is not at least 2 */
3827 if( nbest<2 ) continue;
drh75897232000-05-29 14:26:00 +00003828
drhb59499c2002-02-23 18:45:13 +00003829
3830 /* Combine matching REDUCE actions into a single default */
3831 for(ap=stp->ap; ap; ap=ap->next){
3832 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
3833 }
drh75897232000-05-29 14:26:00 +00003834 assert( ap );
3835 ap->sp = Symbol_new("{default}");
3836 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00003837 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00003838 }
3839 stp->ap = Action_sort(stp->ap);
3840 }
3841}
drhb59499c2002-02-23 18:45:13 +00003842
drh75897232000-05-29 14:26:00 +00003843/***************** From the file "set.c" ************************************/
3844/*
3845** Set manipulation routines for the LEMON parser generator.
3846*/
3847
3848static int size = 0;
3849
3850/* Set the set size */
3851void SetSize(n)
3852int n;
3853{
3854 size = n+1;
3855}
3856
3857/* Allocate a new set */
3858char *SetNew(){
3859 char *s;
3860 int i;
3861 s = (char*)malloc( size );
3862 if( s==0 ){
3863 extern void memory_error();
3864 memory_error();
3865 }
3866 for(i=0; i<size; i++) s[i] = 0;
3867 return s;
3868}
3869
3870/* Deallocate a set */
3871void SetFree(s)
3872char *s;
3873{
3874 free(s);
3875}
3876
3877/* Add a new element to the set. Return TRUE if the element was added
3878** and FALSE if it was already there. */
3879int SetAdd(s,e)
3880char *s;
3881int e;
3882{
3883 int rv;
3884 rv = s[e];
3885 s[e] = 1;
3886 return !rv;
3887}
3888
3889/* Add every element of s2 to s1. Return TRUE if s1 changes. */
3890int SetUnion(s1,s2)
3891char *s1;
3892char *s2;
3893{
3894 int i, progress;
3895 progress = 0;
3896 for(i=0; i<size; i++){
3897 if( s2[i]==0 ) continue;
3898 if( s1[i]==0 ){
3899 progress = 1;
3900 s1[i] = 1;
3901 }
3902 }
3903 return progress;
3904}
3905/********************** From the file "table.c" ****************************/
3906/*
3907** All code in this file has been automatically generated
3908** from a specification in the file
3909** "table.q"
3910** by the associative array code building program "aagen".
3911** Do not edit this file! Instead, edit the specification
3912** file, then rerun aagen.
3913*/
3914/*
3915** Code for processing tables in the LEMON parser generator.
3916*/
3917
3918PRIVATE int strhash(x)
3919char *x;
3920{
3921 int h = 0;
3922 while( *x) h = h*13 + *(x++);
3923 return h;
3924}
3925
3926/* Works like strdup, sort of. Save a string in malloced memory, but
3927** keep strings in a table so that the same string is not in more
3928** than one place.
3929*/
3930char *Strsafe(y)
3931char *y;
3932{
3933 char *z;
3934
3935 z = Strsafe_find(y);
3936 if( z==0 && (z=malloc( strlen(y)+1 ))!=0 ){
3937 strcpy(z,y);
3938 Strsafe_insert(z);
3939 }
3940 MemoryCheck(z);
3941 return z;
3942}
3943
3944/* There is one instance of the following structure for each
3945** associative array of type "x1".
3946*/
3947struct s_x1 {
3948 int size; /* The number of available slots. */
3949 /* Must be a power of 2 greater than or */
3950 /* equal to 1 */
3951 int count; /* Number of currently slots filled */
3952 struct s_x1node *tbl; /* The data stored here */
3953 struct s_x1node **ht; /* Hash table for lookups */
3954};
3955
3956/* There is one instance of this structure for every data element
3957** in an associative array of type "x1".
3958*/
3959typedef struct s_x1node {
3960 char *data; /* The data */
3961 struct s_x1node *next; /* Next entry with the same hash */
3962 struct s_x1node **from; /* Previous link */
3963} x1node;
3964
3965/* There is only one instance of the array, which is the following */
3966static struct s_x1 *x1a;
3967
3968/* Allocate a new associative array */
3969void Strsafe_init(){
3970 if( x1a ) return;
3971 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
3972 if( x1a ){
3973 x1a->size = 1024;
3974 x1a->count = 0;
3975 x1a->tbl = (x1node*)malloc(
3976 (sizeof(x1node) + sizeof(x1node*))*1024 );
3977 if( x1a->tbl==0 ){
3978 free(x1a);
3979 x1a = 0;
3980 }else{
3981 int i;
3982 x1a->ht = (x1node**)&(x1a->tbl[1024]);
3983 for(i=0; i<1024; i++) x1a->ht[i] = 0;
3984 }
3985 }
3986}
3987/* Insert a new record into the array. Return TRUE if successful.
3988** Prior data with the same key is NOT overwritten */
3989int Strsafe_insert(data)
3990char *data;
3991{
3992 x1node *np;
3993 int h;
3994 int ph;
3995
3996 if( x1a==0 ) return 0;
3997 ph = strhash(data);
3998 h = ph & (x1a->size-1);
3999 np = x1a->ht[h];
4000 while( np ){
4001 if( strcmp(np->data,data)==0 ){
4002 /* An existing entry with the same key is found. */
4003 /* Fail because overwrite is not allows. */
4004 return 0;
4005 }
4006 np = np->next;
4007 }
4008 if( x1a->count>=x1a->size ){
4009 /* Need to make the hash table bigger */
4010 int i,size;
4011 struct s_x1 array;
4012 array.size = size = x1a->size*2;
4013 array.count = x1a->count;
4014 array.tbl = (x1node*)malloc(
4015 (sizeof(x1node) + sizeof(x1node*))*size );
4016 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4017 array.ht = (x1node**)&(array.tbl[size]);
4018 for(i=0; i<size; i++) array.ht[i] = 0;
4019 for(i=0; i<x1a->count; i++){
4020 x1node *oldnp, *newnp;
4021 oldnp = &(x1a->tbl[i]);
4022 h = strhash(oldnp->data) & (size-1);
4023 newnp = &(array.tbl[i]);
4024 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4025 newnp->next = array.ht[h];
4026 newnp->data = oldnp->data;
4027 newnp->from = &(array.ht[h]);
4028 array.ht[h] = newnp;
4029 }
4030 free(x1a->tbl);
4031 *x1a = array;
4032 }
4033 /* Insert the new data */
4034 h = ph & (x1a->size-1);
4035 np = &(x1a->tbl[x1a->count++]);
4036 np->data = data;
4037 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4038 np->next = x1a->ht[h];
4039 x1a->ht[h] = np;
4040 np->from = &(x1a->ht[h]);
4041 return 1;
4042}
4043
4044/* Return a pointer to data assigned to the given key. Return NULL
4045** if no such key. */
4046char *Strsafe_find(key)
4047char *key;
4048{
4049 int h;
4050 x1node *np;
4051
4052 if( x1a==0 ) return 0;
4053 h = strhash(key) & (x1a->size-1);
4054 np = x1a->ht[h];
4055 while( np ){
4056 if( strcmp(np->data,key)==0 ) break;
4057 np = np->next;
4058 }
4059 return np ? np->data : 0;
4060}
4061
4062/* Return a pointer to the (terminal or nonterminal) symbol "x".
4063** Create a new symbol if this is the first time "x" has been seen.
4064*/
4065struct symbol *Symbol_new(x)
4066char *x;
4067{
4068 struct symbol *sp;
4069
4070 sp = Symbol_find(x);
4071 if( sp==0 ){
4072 sp = (struct symbol *)malloc( sizeof(struct symbol) );
4073 MemoryCheck(sp);
4074 sp->name = Strsafe(x);
4075 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4076 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004077 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004078 sp->prec = -1;
4079 sp->assoc = UNK;
4080 sp->firstset = 0;
drhb27b83a2002-08-14 23:18:57 +00004081 sp->lambda = B_FALSE;
drh75897232000-05-29 14:26:00 +00004082 sp->destructor = 0;
4083 sp->datatype = 0;
4084 Symbol_insert(sp,sp->name);
4085 }
4086 return sp;
4087}
4088
drh60d31652004-02-22 00:08:04 +00004089/* Compare two symbols for working purposes
4090**
4091** Symbols that begin with upper case letters (terminals or tokens)
4092** must sort before symbols that begin with lower case letters
4093** (non-terminals). Other than that, the order does not matter.
4094**
4095** We find experimentally that leaving the symbols in their original
4096** order (the order they appeared in the grammar file) gives the
4097** smallest parser tables in SQLite.
4098*/
4099int Symbolcmpp(struct symbol **a, struct symbol **b){
4100 int i1 = (**a).index + 10000000*((**a).name[0]>'Z');
4101 int i2 = (**b).index + 10000000*((**b).name[0]>'Z');
4102 return i1-i2;
drh75897232000-05-29 14:26:00 +00004103}
4104
4105/* There is one instance of the following structure for each
4106** associative array of type "x2".
4107*/
4108struct s_x2 {
4109 int size; /* The number of available slots. */
4110 /* Must be a power of 2 greater than or */
4111 /* equal to 1 */
4112 int count; /* Number of currently slots filled */
4113 struct s_x2node *tbl; /* The data stored here */
4114 struct s_x2node **ht; /* Hash table for lookups */
4115};
4116
4117/* There is one instance of this structure for every data element
4118** in an associative array of type "x2".
4119*/
4120typedef struct s_x2node {
4121 struct symbol *data; /* The data */
4122 char *key; /* The key */
4123 struct s_x2node *next; /* Next entry with the same hash */
4124 struct s_x2node **from; /* Previous link */
4125} x2node;
4126
4127/* There is only one instance of the array, which is the following */
4128static struct s_x2 *x2a;
4129
4130/* Allocate a new associative array */
4131void Symbol_init(){
4132 if( x2a ) return;
4133 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4134 if( x2a ){
4135 x2a->size = 128;
4136 x2a->count = 0;
4137 x2a->tbl = (x2node*)malloc(
4138 (sizeof(x2node) + sizeof(x2node*))*128 );
4139 if( x2a->tbl==0 ){
4140 free(x2a);
4141 x2a = 0;
4142 }else{
4143 int i;
4144 x2a->ht = (x2node**)&(x2a->tbl[128]);
4145 for(i=0; i<128; i++) x2a->ht[i] = 0;
4146 }
4147 }
4148}
4149/* Insert a new record into the array. Return TRUE if successful.
4150** Prior data with the same key is NOT overwritten */
4151int Symbol_insert(data,key)
4152struct symbol *data;
4153char *key;
4154{
4155 x2node *np;
4156 int h;
4157 int ph;
4158
4159 if( x2a==0 ) return 0;
4160 ph = strhash(key);
4161 h = ph & (x2a->size-1);
4162 np = x2a->ht[h];
4163 while( np ){
4164 if( strcmp(np->key,key)==0 ){
4165 /* An existing entry with the same key is found. */
4166 /* Fail because overwrite is not allows. */
4167 return 0;
4168 }
4169 np = np->next;
4170 }
4171 if( x2a->count>=x2a->size ){
4172 /* Need to make the hash table bigger */
4173 int i,size;
4174 struct s_x2 array;
4175 array.size = size = x2a->size*2;
4176 array.count = x2a->count;
4177 array.tbl = (x2node*)malloc(
4178 (sizeof(x2node) + sizeof(x2node*))*size );
4179 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4180 array.ht = (x2node**)&(array.tbl[size]);
4181 for(i=0; i<size; i++) array.ht[i] = 0;
4182 for(i=0; i<x2a->count; i++){
4183 x2node *oldnp, *newnp;
4184 oldnp = &(x2a->tbl[i]);
4185 h = strhash(oldnp->key) & (size-1);
4186 newnp = &(array.tbl[i]);
4187 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4188 newnp->next = array.ht[h];
4189 newnp->key = oldnp->key;
4190 newnp->data = oldnp->data;
4191 newnp->from = &(array.ht[h]);
4192 array.ht[h] = newnp;
4193 }
4194 free(x2a->tbl);
4195 *x2a = array;
4196 }
4197 /* Insert the new data */
4198 h = ph & (x2a->size-1);
4199 np = &(x2a->tbl[x2a->count++]);
4200 np->key = key;
4201 np->data = data;
4202 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4203 np->next = x2a->ht[h];
4204 x2a->ht[h] = np;
4205 np->from = &(x2a->ht[h]);
4206 return 1;
4207}
4208
4209/* Return a pointer to data assigned to the given key. Return NULL
4210** if no such key. */
4211struct symbol *Symbol_find(key)
4212char *key;
4213{
4214 int h;
4215 x2node *np;
4216
4217 if( x2a==0 ) return 0;
4218 h = strhash(key) & (x2a->size-1);
4219 np = x2a->ht[h];
4220 while( np ){
4221 if( strcmp(np->key,key)==0 ) break;
4222 np = np->next;
4223 }
4224 return np ? np->data : 0;
4225}
4226
4227/* Return the n-th data. Return NULL if n is out of range. */
4228struct symbol *Symbol_Nth(n)
4229int n;
4230{
4231 struct symbol *data;
4232 if( x2a && n>0 && n<=x2a->count ){
4233 data = x2a->tbl[n-1].data;
4234 }else{
4235 data = 0;
4236 }
4237 return data;
4238}
4239
4240/* Return the size of the array */
4241int Symbol_count()
4242{
4243 return x2a ? x2a->count : 0;
4244}
4245
4246/* Return an array of pointers to all data in the table.
4247** The array is obtained from malloc. Return NULL if memory allocation
4248** problems, or if the array is empty. */
4249struct symbol **Symbol_arrayof()
4250{
4251 struct symbol **array;
4252 int i,size;
4253 if( x2a==0 ) return 0;
4254 size = x2a->count;
4255 array = (struct symbol **)malloc( sizeof(struct symbol *)*size );
4256 if( array ){
4257 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4258 }
4259 return array;
4260}
4261
4262/* Compare two configurations */
4263int Configcmp(a,b)
4264struct config *a;
4265struct config *b;
4266{
4267 int x;
4268 x = a->rp->index - b->rp->index;
4269 if( x==0 ) x = a->dot - b->dot;
4270 return x;
4271}
4272
4273/* Compare two states */
4274PRIVATE int statecmp(a,b)
4275struct config *a;
4276struct config *b;
4277{
4278 int rc;
4279 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4280 rc = a->rp->index - b->rp->index;
4281 if( rc==0 ) rc = a->dot - b->dot;
4282 }
4283 if( rc==0 ){
4284 if( a ) rc = 1;
4285 if( b ) rc = -1;
4286 }
4287 return rc;
4288}
4289
4290/* Hash a state */
4291PRIVATE int statehash(a)
4292struct config *a;
4293{
4294 int h=0;
4295 while( a ){
4296 h = h*571 + a->rp->index*37 + a->dot;
4297 a = a->bp;
4298 }
4299 return h;
4300}
4301
4302/* Allocate a new state structure */
4303struct state *State_new()
4304{
4305 struct state *new;
4306 new = (struct state *)malloc( sizeof(struct state) );
4307 MemoryCheck(new);
4308 return new;
4309}
4310
4311/* There is one instance of the following structure for each
4312** associative array of type "x3".
4313*/
4314struct s_x3 {
4315 int size; /* The number of available slots. */
4316 /* Must be a power of 2 greater than or */
4317 /* equal to 1 */
4318 int count; /* Number of currently slots filled */
4319 struct s_x3node *tbl; /* The data stored here */
4320 struct s_x3node **ht; /* Hash table for lookups */
4321};
4322
4323/* There is one instance of this structure for every data element
4324** in an associative array of type "x3".
4325*/
4326typedef struct s_x3node {
4327 struct state *data; /* The data */
4328 struct config *key; /* The key */
4329 struct s_x3node *next; /* Next entry with the same hash */
4330 struct s_x3node **from; /* Previous link */
4331} x3node;
4332
4333/* There is only one instance of the array, which is the following */
4334static struct s_x3 *x3a;
4335
4336/* Allocate a new associative array */
4337void State_init(){
4338 if( x3a ) return;
4339 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4340 if( x3a ){
4341 x3a->size = 128;
4342 x3a->count = 0;
4343 x3a->tbl = (x3node*)malloc(
4344 (sizeof(x3node) + sizeof(x3node*))*128 );
4345 if( x3a->tbl==0 ){
4346 free(x3a);
4347 x3a = 0;
4348 }else{
4349 int i;
4350 x3a->ht = (x3node**)&(x3a->tbl[128]);
4351 for(i=0; i<128; i++) x3a->ht[i] = 0;
4352 }
4353 }
4354}
4355/* Insert a new record into the array. Return TRUE if successful.
4356** Prior data with the same key is NOT overwritten */
4357int State_insert(data,key)
4358struct state *data;
4359struct config *key;
4360{
4361 x3node *np;
4362 int h;
4363 int ph;
4364
4365 if( x3a==0 ) return 0;
4366 ph = statehash(key);
4367 h = ph & (x3a->size-1);
4368 np = x3a->ht[h];
4369 while( np ){
4370 if( statecmp(np->key,key)==0 ){
4371 /* An existing entry with the same key is found. */
4372 /* Fail because overwrite is not allows. */
4373 return 0;
4374 }
4375 np = np->next;
4376 }
4377 if( x3a->count>=x3a->size ){
4378 /* Need to make the hash table bigger */
4379 int i,size;
4380 struct s_x3 array;
4381 array.size = size = x3a->size*2;
4382 array.count = x3a->count;
4383 array.tbl = (x3node*)malloc(
4384 (sizeof(x3node) + sizeof(x3node*))*size );
4385 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4386 array.ht = (x3node**)&(array.tbl[size]);
4387 for(i=0; i<size; i++) array.ht[i] = 0;
4388 for(i=0; i<x3a->count; i++){
4389 x3node *oldnp, *newnp;
4390 oldnp = &(x3a->tbl[i]);
4391 h = statehash(oldnp->key) & (size-1);
4392 newnp = &(array.tbl[i]);
4393 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4394 newnp->next = array.ht[h];
4395 newnp->key = oldnp->key;
4396 newnp->data = oldnp->data;
4397 newnp->from = &(array.ht[h]);
4398 array.ht[h] = newnp;
4399 }
4400 free(x3a->tbl);
4401 *x3a = array;
4402 }
4403 /* Insert the new data */
4404 h = ph & (x3a->size-1);
4405 np = &(x3a->tbl[x3a->count++]);
4406 np->key = key;
4407 np->data = data;
4408 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4409 np->next = x3a->ht[h];
4410 x3a->ht[h] = np;
4411 np->from = &(x3a->ht[h]);
4412 return 1;
4413}
4414
4415/* Return a pointer to data assigned to the given key. Return NULL
4416** if no such key. */
4417struct state *State_find(key)
4418struct config *key;
4419{
4420 int h;
4421 x3node *np;
4422
4423 if( x3a==0 ) return 0;
4424 h = statehash(key) & (x3a->size-1);
4425 np = x3a->ht[h];
4426 while( np ){
4427 if( statecmp(np->key,key)==0 ) break;
4428 np = np->next;
4429 }
4430 return np ? np->data : 0;
4431}
4432
4433/* Return an array of pointers to all data in the table.
4434** The array is obtained from malloc. Return NULL if memory allocation
4435** problems, or if the array is empty. */
4436struct state **State_arrayof()
4437{
4438 struct state **array;
4439 int i,size;
4440 if( x3a==0 ) return 0;
4441 size = x3a->count;
4442 array = (struct state **)malloc( sizeof(struct state *)*size );
4443 if( array ){
4444 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4445 }
4446 return array;
4447}
4448
4449/* Hash a configuration */
4450PRIVATE int confighash(a)
4451struct config *a;
4452{
4453 int h=0;
4454 h = h*571 + a->rp->index*37 + a->dot;
4455 return h;
4456}
4457
4458/* There is one instance of the following structure for each
4459** associative array of type "x4".
4460*/
4461struct s_x4 {
4462 int size; /* The number of available slots. */
4463 /* Must be a power of 2 greater than or */
4464 /* equal to 1 */
4465 int count; /* Number of currently slots filled */
4466 struct s_x4node *tbl; /* The data stored here */
4467 struct s_x4node **ht; /* Hash table for lookups */
4468};
4469
4470/* There is one instance of this structure for every data element
4471** in an associative array of type "x4".
4472*/
4473typedef struct s_x4node {
4474 struct config *data; /* The data */
4475 struct s_x4node *next; /* Next entry with the same hash */
4476 struct s_x4node **from; /* Previous link */
4477} x4node;
4478
4479/* There is only one instance of the array, which is the following */
4480static struct s_x4 *x4a;
4481
4482/* Allocate a new associative array */
4483void Configtable_init(){
4484 if( x4a ) return;
4485 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4486 if( x4a ){
4487 x4a->size = 64;
4488 x4a->count = 0;
4489 x4a->tbl = (x4node*)malloc(
4490 (sizeof(x4node) + sizeof(x4node*))*64 );
4491 if( x4a->tbl==0 ){
4492 free(x4a);
4493 x4a = 0;
4494 }else{
4495 int i;
4496 x4a->ht = (x4node**)&(x4a->tbl[64]);
4497 for(i=0; i<64; i++) x4a->ht[i] = 0;
4498 }
4499 }
4500}
4501/* Insert a new record into the array. Return TRUE if successful.
4502** Prior data with the same key is NOT overwritten */
4503int Configtable_insert(data)
4504struct config *data;
4505{
4506 x4node *np;
4507 int h;
4508 int ph;
4509
4510 if( x4a==0 ) return 0;
4511 ph = confighash(data);
4512 h = ph & (x4a->size-1);
4513 np = x4a->ht[h];
4514 while( np ){
4515 if( Configcmp(np->data,data)==0 ){
4516 /* An existing entry with the same key is found. */
4517 /* Fail because overwrite is not allows. */
4518 return 0;
4519 }
4520 np = np->next;
4521 }
4522 if( x4a->count>=x4a->size ){
4523 /* Need to make the hash table bigger */
4524 int i,size;
4525 struct s_x4 array;
4526 array.size = size = x4a->size*2;
4527 array.count = x4a->count;
4528 array.tbl = (x4node*)malloc(
4529 (sizeof(x4node) + sizeof(x4node*))*size );
4530 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4531 array.ht = (x4node**)&(array.tbl[size]);
4532 for(i=0; i<size; i++) array.ht[i] = 0;
4533 for(i=0; i<x4a->count; i++){
4534 x4node *oldnp, *newnp;
4535 oldnp = &(x4a->tbl[i]);
4536 h = confighash(oldnp->data) & (size-1);
4537 newnp = &(array.tbl[i]);
4538 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4539 newnp->next = array.ht[h];
4540 newnp->data = oldnp->data;
4541 newnp->from = &(array.ht[h]);
4542 array.ht[h] = newnp;
4543 }
4544 free(x4a->tbl);
4545 *x4a = array;
4546 }
4547 /* Insert the new data */
4548 h = ph & (x4a->size-1);
4549 np = &(x4a->tbl[x4a->count++]);
4550 np->data = data;
4551 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
4552 np->next = x4a->ht[h];
4553 x4a->ht[h] = np;
4554 np->from = &(x4a->ht[h]);
4555 return 1;
4556}
4557
4558/* Return a pointer to data assigned to the given key. Return NULL
4559** if no such key. */
4560struct config *Configtable_find(key)
4561struct config *key;
4562{
4563 int h;
4564 x4node *np;
4565
4566 if( x4a==0 ) return 0;
4567 h = confighash(key) & (x4a->size-1);
4568 np = x4a->ht[h];
4569 while( np ){
4570 if( Configcmp(np->data,key)==0 ) break;
4571 np = np->next;
4572 }
4573 return np ? np->data : 0;
4574}
4575
4576/* Remove all data from the table. Pass each data to the function "f"
4577** as it is removed. ("f" may be null to avoid this step.) */
4578void Configtable_clear(f)
4579int(*f)(/* struct config * */);
4580{
4581 int i;
4582 if( x4a==0 || x4a->count==0 ) return;
4583 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
4584 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
4585 x4a->count = 0;
4586 return;
4587}