blob: 0dfa1b9a61a1dc744d2924d206cf9c5f69bc1f40 [file] [log] [blame]
drh75897232000-05-29 14:26:00 +00001/*
drh75897232000-05-29 14:26:00 +00002** This file contains all sources (including headers) to the LEMON
3** LALR(1) parser generator. The sources have been combined into a
drh960e8c62001-04-03 16:53:21 +00004** single file to make it easy to include LEMON in the source tree
5** and Makefile of another program.
drh75897232000-05-29 14:26:00 +00006**
drhb19a2bc2001-09-16 00:13:26 +00007** The author of this program disclaims copyright.
drh75897232000-05-29 14:26:00 +00008*/
9#include <stdio.h>
drhf9a2e7b2003-04-15 01:49:48 +000010#include <stdarg.h>
drh75897232000-05-29 14:26:00 +000011#include <string.h>
12#include <ctype.h>
drh8b582012003-10-21 13:16:03 +000013#include <stdlib.h>
drhe9278182007-07-18 18:16:29 +000014#include <assert.h>
drh75897232000-05-29 14:26:00 +000015
drh75897232000-05-29 14:26:00 +000016#ifndef __WIN32__
17# if defined(_WIN32) || defined(WIN32)
18# define __WIN32__
19# endif
20#endif
21
rse8f304482007-07-30 18:31:53 +000022#ifdef __WIN32__
23extern int access();
24#else
25#include <unistd.h>
26#endif
27
drh75897232000-05-29 14:26:00 +000028/* #define PRIVATE static */
29#define PRIVATE
30
31#ifdef TEST
32#define MAXRHS 5 /* Set low to exercise exception code */
33#else
34#define MAXRHS 1000
35#endif
36
icculusf5ad8242010-02-14 05:34:42 +000037static const char **made_files = NULL;
38static int made_files_count = 0;
39static int successful_exit = 0;
40static void LemonAtExit(void)
41{
42 /* if we failed, delete (most) files we made, to unconfuse build tools. */
43 int i;
44 for (i = 0; i < made_files_count; i++) {
45 if (!successful_exit) {
46 remove(made_files[i]);
47 }
48 free((void *) made_files[i]);
49 }
50 free(made_files);
51 made_files_count = 0;
52 made_files = NULL;
53}
54
drhe9278182007-07-18 18:16:29 +000055static char *msort(char*,char**,int(*)(const char*,const char*));
drh75897232000-05-29 14:26:00 +000056
drh87cf1372008-08-13 20:09:06 +000057/*
58** Compilers are getting increasingly pedantic about type conversions
59** as C evolves ever closer to Ada.... To work around the latest problems
60** we have to define the following variant of strlen().
61*/
62#define lemonStrlen(X) ((int)strlen(X))
63
icculus9e44cf12010-02-14 17:14:22 +000064/* a few forward declarations... */
65struct rule;
66struct lemon;
67struct action;
68
drhe9278182007-07-18 18:16:29 +000069static struct action *Action_new(void);
70static struct action *Action_sort(struct action *);
drh75897232000-05-29 14:26:00 +000071
72/********** From the file "build.h" ************************************/
73void FindRulePrecedences();
74void FindFirstSets();
75void FindStates();
76void FindLinks();
77void FindFollowSets();
78void FindActions();
79
80/********* From the file "configlist.h" *********************************/
icculus9e44cf12010-02-14 17:14:22 +000081void Configlist_init(void);
82struct config *Configlist_add(struct rule *, int);
83struct config *Configlist_addbasis(struct rule *, int);
84void Configlist_closure(struct lemon *);
85void Configlist_sort(void);
86void Configlist_sortbasis(void);
87struct config *Configlist_return(void);
88struct config *Configlist_basis(void);
89void Configlist_eat(struct config *);
90void Configlist_reset(void);
drh75897232000-05-29 14:26:00 +000091
92/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +000093void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +000094
95/****** From the file "option.h" ******************************************/
icculus9e44cf12010-02-14 17:14:22 +000096enum option_type { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
97 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR};
drh75897232000-05-29 14:26:00 +000098struct s_options {
icculus9e44cf12010-02-14 17:14:22 +000099 enum option_type type;
100 const char *label;
drh75897232000-05-29 14:26:00 +0000101 char *arg;
icculus9e44cf12010-02-14 17:14:22 +0000102 const char *message;
drh75897232000-05-29 14:26:00 +0000103};
icculus9e44cf12010-02-14 17:14:22 +0000104int OptInit(char**,struct s_options*,FILE*);
105int OptNArgs(void);
106char *OptArg(int);
107void OptErr(int);
108void OptPrint(void);
drh75897232000-05-29 14:26:00 +0000109
110/******** From the file "parse.h" *****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000111void Parse(struct lemon *lemp);
drh75897232000-05-29 14:26:00 +0000112
113/********* From the file "plink.h" ***************************************/
icculus9e44cf12010-02-14 17:14:22 +0000114struct plink *Plink_new(void);
115void Plink_add(struct plink **, struct config *);
116void Plink_copy(struct plink **, struct plink *);
117void Plink_delete(struct plink *);
drh75897232000-05-29 14:26:00 +0000118
119/********** From the file "report.h" *************************************/
icculus9e44cf12010-02-14 17:14:22 +0000120void Reprint(struct lemon *);
121void ReportOutput(struct lemon *);
122void ReportTable(struct lemon *, int);
123void ReportHeader(struct lemon *);
124void CompressTables(struct lemon *);
125void ResortStates(struct lemon *);
drh75897232000-05-29 14:26:00 +0000126
127/********** From the file "set.h" ****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000128void SetSize(int); /* All sets will be of size N */
129char *SetNew(void); /* A new set for element 0..N */
130void SetFree(char*); /* Deallocate a set */
drh75897232000-05-29 14:26:00 +0000131
icculus9e44cf12010-02-14 17:14:22 +0000132char *SetNew(void); /* A new set for element 0..N */
133int SetAdd(char*,int); /* Add element to a set */
134int SetUnion(char *,char *); /* A <- A U B, thru element N */
drh75897232000-05-29 14:26:00 +0000135#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
136
137/********** From the file "struct.h" *************************************/
138/*
139** Principal data structures for the LEMON parser generator.
140*/
141
drhaa9f1122007-08-23 02:50:56 +0000142typedef enum {LEMON_FALSE=0, LEMON_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000143
144/* Symbols (terminals and nonterminals) of the grammar are stored
145** in the following: */
icculus9e44cf12010-02-14 17:14:22 +0000146enum symbol_type {
147 TERMINAL,
148 NONTERMINAL,
149 MULTITERMINAL
150};
151enum e_assoc {
drh75897232000-05-29 14:26:00 +0000152 LEFT,
153 RIGHT,
154 NONE,
155 UNK
icculus9e44cf12010-02-14 17:14:22 +0000156};
157struct symbol {
158 const char *name; /* Name of the symbol */
159 int index; /* Index number for this symbol */
160 enum symbol_type type; /* Symbols are all either TERMINALS or NTs */
161 struct rule *rule; /* Linked list of rules of this (if an NT) */
162 struct symbol *fallback; /* fallback token in case this token doesn't parse */
163 int prec; /* Precedence if defined (-1 otherwise) */
164 enum e_assoc assoc; /* Associativity if precedence is defined */
drh75897232000-05-29 14:26:00 +0000165 char *firstset; /* First-set for all rules of this symbol */
166 Boolean lambda; /* True if NT and can generate an empty string */
drhc4dd3fd2008-01-22 01:48:05 +0000167 int useCnt; /* Number of times used */
drh75897232000-05-29 14:26:00 +0000168 char *destructor; /* Code which executes whenever this symbol is
169 ** popped from the stack during error processing */
drh4dc8ef52008-07-01 17:13:57 +0000170 int destLineno; /* Line number for start of destructor */
drh75897232000-05-29 14:26:00 +0000171 char *datatype; /* The data type of information held by this
172 ** object. Only used if type==NONTERMINAL */
173 int dtnum; /* The data type number. In the parser, the value
174 ** stack is a union. The .yy%d element of this
175 ** union is the correct data type for this object */
drhfd405312005-11-06 04:06:59 +0000176 /* The following fields are used by MULTITERMINALs only */
177 int nsubsym; /* Number of constituent symbols in the MULTI */
178 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000179};
180
181/* Each production rule in the grammar is stored in the following
182** structure. */
183struct rule {
184 struct symbol *lhs; /* Left-hand side of the rule */
icculus9e44cf12010-02-14 17:14:22 +0000185 const char *lhsalias; /* Alias for the LHS (NULL if none) */
drhb4960992007-10-05 16:16:36 +0000186 int lhsStart; /* True if left-hand side is the start symbol */
drh75897232000-05-29 14:26:00 +0000187 int ruleline; /* Line number for the rule */
188 int nrhs; /* Number of RHS symbols */
189 struct symbol **rhs; /* The RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +0000190 const char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
drh75897232000-05-29 14:26:00 +0000191 int line; /* Line number at which code begins */
icculus9e44cf12010-02-14 17:14:22 +0000192 const char *code; /* The code executed when this rule is reduced */
drh75897232000-05-29 14:26:00 +0000193 struct symbol *precsym; /* Precedence symbol for this rule */
194 int index; /* An index number for this rule */
195 Boolean canReduce; /* True if this rule is ever reduced */
196 struct rule *nextlhs; /* Next rule with the same LHS */
197 struct rule *next; /* Next rule in the global list */
198};
199
200/* A configuration is a production rule of the grammar together with
201** a mark (dot) showing how much of that rule has been processed so far.
202** Configurations also contain a follow-set which is a list of terminal
203** symbols which are allowed to immediately follow the end of the rule.
204** Every configuration is recorded as an instance of the following: */
icculus9e44cf12010-02-14 17:14:22 +0000205enum cfgstatus {
206 COMPLETE,
207 INCOMPLETE
208};
drh75897232000-05-29 14:26:00 +0000209struct config {
210 struct rule *rp; /* The rule upon which the configuration is based */
211 int dot; /* The parse point */
212 char *fws; /* Follow-set for this configuration only */
213 struct plink *fplp; /* Follow-set forward propagation links */
214 struct plink *bplp; /* Follow-set backwards propagation links */
215 struct state *stp; /* Pointer to state which contains this */
icculus9e44cf12010-02-14 17:14:22 +0000216 enum cfgstatus status; /* used during followset and shift computations */
drh75897232000-05-29 14:26:00 +0000217 struct config *next; /* Next configuration in the state */
218 struct config *bp; /* The next basis configuration */
219};
220
icculus9e44cf12010-02-14 17:14:22 +0000221enum e_action {
222 SHIFT,
223 ACCEPT,
224 REDUCE,
225 ERROR,
226 SSCONFLICT, /* A shift/shift conflict */
227 SRCONFLICT, /* Was a reduce, but part of a conflict */
228 RRCONFLICT, /* Was a reduce, but part of a conflict */
229 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
230 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
231 NOT_USED /* Deleted by compression */
232};
233
drh75897232000-05-29 14:26:00 +0000234/* Every shift or reduce operation is stored as one of the following */
235struct action {
236 struct symbol *sp; /* The look-ahead symbol */
icculus9e44cf12010-02-14 17:14:22 +0000237 enum e_action type;
drh75897232000-05-29 14:26:00 +0000238 union {
239 struct state *stp; /* The new state, if a shift */
240 struct rule *rp; /* The rule, if a reduce */
241 } x;
242 struct action *next; /* Next action for this state */
243 struct action *collide; /* Next action with the same hash */
244};
245
246/* Each state of the generated parser's finite state machine
247** is encoded as an instance of the following structure. */
248struct state {
249 struct config *bp; /* The basis configurations for this state */
250 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000251 int statenum; /* Sequential number for this state */
drh75897232000-05-29 14:26:00 +0000252 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000253 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
254 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
255 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000256};
drh8b582012003-10-21 13:16:03 +0000257#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000258
259/* A followset propagation link indicates that the contents of one
260** configuration followset should be propagated to another whenever
261** the first changes. */
262struct plink {
263 struct config *cfp; /* The configuration to which linked */
264 struct plink *next; /* The next propagate link */
265};
266
267/* The state vector for the entire parser generator is recorded as
268** follows. (LEMON uses no global variables and makes little use of
269** static variables. Fields in the following structure can be thought
270** of as begin global variables in the program.) */
271struct lemon {
272 struct state **sorted; /* Table of states sorted by state number */
273 struct rule *rule; /* List of all rules */
274 int nstate; /* Number of states */
275 int nrule; /* Number of rules */
276 int nsymbol; /* Number of terminal and nonterminal symbols */
277 int nterminal; /* Number of terminal symbols */
278 struct symbol **symbols; /* Sorted array of pointers to symbols */
279 int errorcnt; /* Number of errors */
280 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000281 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000282 char *name; /* Name of the generated parser */
283 char *arg; /* Declaration of the 3th argument to parser */
284 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000285 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000286 char *start; /* Name of the start symbol for the grammar */
287 char *stacksize; /* Size of the parser stack */
288 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000289 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000290 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000291 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000292 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000293 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000294 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000295 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000296 char *filename; /* Name of the input file */
297 char *outname; /* Name of the current output file */
298 char *tokenprefix; /* A prefix added to token names in the .h file */
299 int nconflict; /* Number of parsing conflicts */
icculus42585cf2010-02-14 05:19:56 +0000300 int nexpected; /* Number of expected parsing conflicts */
drh75897232000-05-29 14:26:00 +0000301 int tablesize; /* Size of the parse tables */
302 int basisflag; /* Print only basis configurations */
drh34ff57b2008-07-14 12:27:51 +0000303 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000304 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000305 char *argv0; /* Name of the program */
306};
307
308#define MemoryCheck(X) if((X)==0){ \
309 extern void memory_error(); \
310 memory_error(); \
311}
312
313/**************** From the file "table.h" *********************************/
314/*
315** All code in this file has been automatically generated
316** from a specification in the file
317** "table.q"
318** by the associative array code building program "aagen".
319** Do not edit this file! Instead, edit the specification
320** file, then rerun aagen.
321*/
322/*
323** Code for processing tables in the LEMON parser generator.
324*/
drh75897232000-05-29 14:26:00 +0000325/* Routines for handling a strings */
326
icculus9e44cf12010-02-14 17:14:22 +0000327const char *Strsafe(const char *);
drh75897232000-05-29 14:26:00 +0000328
icculus9e44cf12010-02-14 17:14:22 +0000329void Strsafe_init(void);
330int Strsafe_insert(const char *);
331const char *Strsafe_find(const char *);
drh75897232000-05-29 14:26:00 +0000332
333/* Routines for handling symbols of the grammar */
334
icculus9e44cf12010-02-14 17:14:22 +0000335struct symbol *Symbol_new(const char *);
336int Symbolcmpp(const void *, const void *);
337void Symbol_init(void);
338int Symbol_insert(struct symbol *, const char *);
339struct symbol *Symbol_find(const char *);
340struct symbol *Symbol_Nth(int);
341int Symbol_count(void);
342struct symbol **Symbol_arrayof(void);
drh75897232000-05-29 14:26:00 +0000343
344/* Routines to manage the state table */
345
icculus9e44cf12010-02-14 17:14:22 +0000346int Configcmp(const char *, const char *);
347struct state *State_new(void);
348void State_init(void);
349int State_insert(struct state *, struct config *);
350struct state *State_find(struct config *);
drh75897232000-05-29 14:26:00 +0000351struct state **State_arrayof(/* */);
352
353/* Routines used for efficiency in Configlist_add */
354
icculus9e44cf12010-02-14 17:14:22 +0000355void Configtable_init(void);
356int Configtable_insert(struct config *);
357struct config *Configtable_find(struct config *);
358void Configtable_clear(int(*)(struct config *));
359
drh75897232000-05-29 14:26:00 +0000360/****************** From the file "action.c" *******************************/
361/*
362** Routines processing parser actions in the LEMON parser generator.
363*/
364
365/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000366static struct action *Action_new(void){
drh75897232000-05-29 14:26:00 +0000367 static struct action *freelist = 0;
icculus9e44cf12010-02-14 17:14:22 +0000368 struct action *newaction;
drh75897232000-05-29 14:26:00 +0000369
370 if( freelist==0 ){
371 int i;
372 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +0000373 freelist = (struct action *)calloc(amt, sizeof(struct action));
drh75897232000-05-29 14:26:00 +0000374 if( freelist==0 ){
375 fprintf(stderr,"Unable to allocate memory for a new parser action.");
376 exit(1);
377 }
378 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
379 freelist[amt-1].next = 0;
380 }
icculus9e44cf12010-02-14 17:14:22 +0000381 newaction = freelist;
drh75897232000-05-29 14:26:00 +0000382 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +0000383 return newaction;
drh75897232000-05-29 14:26:00 +0000384}
385
drhe9278182007-07-18 18:16:29 +0000386/* Compare two actions for sorting purposes. Return negative, zero, or
387** positive if the first action is less than, equal to, or greater than
388** the first
389*/
390static int actioncmp(
391 struct action *ap1,
392 struct action *ap2
393){
drh75897232000-05-29 14:26:00 +0000394 int rc;
395 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000396 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000397 rc = (int)ap1->type - (int)ap2->type;
398 }
399 if( rc==0 && ap1->type==REDUCE ){
drh75897232000-05-29 14:26:00 +0000400 rc = ap1->x.rp->index - ap2->x.rp->index;
401 }
drhe594bc32009-11-03 13:02:25 +0000402 if( rc==0 ){
403 rc = ap2 - ap1;
404 }
drh75897232000-05-29 14:26:00 +0000405 return rc;
406}
407
408/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000409static struct action *Action_sort(
410 struct action *ap
411){
412 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
413 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000414 return ap;
415}
416
icculus9e44cf12010-02-14 17:14:22 +0000417void Action_add(
418 struct action **app,
419 enum e_action type,
420 struct symbol *sp,
421 char *arg
422){
423 struct action *newaction;
424 newaction = Action_new();
425 newaction->next = *app;
426 *app = newaction;
427 newaction->type = type;
428 newaction->sp = sp;
drh75897232000-05-29 14:26:00 +0000429 if( type==SHIFT ){
icculus9e44cf12010-02-14 17:14:22 +0000430 newaction->x.stp = (struct state *)arg;
drh75897232000-05-29 14:26:00 +0000431 }else{
icculus9e44cf12010-02-14 17:14:22 +0000432 newaction->x.rp = (struct rule *)arg;
drh75897232000-05-29 14:26:00 +0000433 }
434}
drh8b582012003-10-21 13:16:03 +0000435/********************** New code to implement the "acttab" module ***********/
436/*
437** This module implements routines use to construct the yy_action[] table.
438*/
439
440/*
441** The state of the yy_action table under construction is an instance of
drh8dc3e8f2010-01-07 03:53:03 +0000442** the following structure.
443**
444** The yy_action table maps the pair (state_number, lookahead) into an
445** action_number. The table is an array of integers pairs. The state_number
446** determines an initial offset into the yy_action array. The lookahead
447** value is then added to this initial offset to get an index X into the
448** yy_action array. If the aAction[X].lookahead equals the value of the
449** of the lookahead input, then the value of the action_number output is
450** aAction[X].action. If the lookaheads do not match then the
451** default action for the state_number is returned.
452**
453** All actions associated with a single state_number are first entered
454** into aLookahead[] using multiple calls to acttab_action(). Then the
455** actions for that single state_number are placed into the aAction[]
456** array with a single call to acttab_insert(). The acttab_insert() call
457** also resets the aLookahead[] array in preparation for the next
458** state number.
drh8b582012003-10-21 13:16:03 +0000459*/
icculus9e44cf12010-02-14 17:14:22 +0000460struct lookahead_action {
461 int lookahead; /* Value of the lookahead token */
462 int action; /* Action to take on the given lookahead */
463};
drh8b582012003-10-21 13:16:03 +0000464typedef struct acttab acttab;
465struct acttab {
466 int nAction; /* Number of used slots in aAction[] */
467 int nActionAlloc; /* Slots allocated for aAction[] */
icculus9e44cf12010-02-14 17:14:22 +0000468 struct lookahead_action
469 *aAction, /* The yy_action[] table under construction */
drh8b582012003-10-21 13:16:03 +0000470 *aLookahead; /* A single new transaction set */
471 int mnLookahead; /* Minimum aLookahead[].lookahead */
472 int mnAction; /* Action associated with mnLookahead */
473 int mxLookahead; /* Maximum aLookahead[].lookahead */
474 int nLookahead; /* Used slots in aLookahead[] */
475 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
476};
477
478/* Return the number of entries in the yy_action table */
479#define acttab_size(X) ((X)->nAction)
480
481/* The value for the N-th entry in yy_action */
482#define acttab_yyaction(X,N) ((X)->aAction[N].action)
483
484/* The value for the N-th entry in yy_lookahead */
485#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
486
487/* Free all memory associated with the given acttab */
488void acttab_free(acttab *p){
489 free( p->aAction );
490 free( p->aLookahead );
491 free( p );
492}
493
494/* Allocate a new acttab structure */
495acttab *acttab_alloc(void){
icculus9e44cf12010-02-14 17:14:22 +0000496 acttab *p = (acttab *) calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000497 if( p==0 ){
498 fprintf(stderr,"Unable to allocate memory for a new acttab.");
499 exit(1);
500 }
501 memset(p, 0, sizeof(*p));
502 return p;
503}
504
drh8dc3e8f2010-01-07 03:53:03 +0000505/* Add a new action to the current transaction set.
506**
507** This routine is called once for each lookahead for a particular
508** state.
drh8b582012003-10-21 13:16:03 +0000509*/
510void acttab_action(acttab *p, int lookahead, int action){
511 if( p->nLookahead>=p->nLookaheadAlloc ){
512 p->nLookaheadAlloc += 25;
icculus9e44cf12010-02-14 17:14:22 +0000513 p->aLookahead = (struct lookahead_action *) realloc( p->aLookahead,
drh8b582012003-10-21 13:16:03 +0000514 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
515 if( p->aLookahead==0 ){
516 fprintf(stderr,"malloc failed\n");
517 exit(1);
518 }
519 }
520 if( p->nLookahead==0 ){
521 p->mxLookahead = lookahead;
522 p->mnLookahead = lookahead;
523 p->mnAction = action;
524 }else{
525 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
526 if( p->mnLookahead>lookahead ){
527 p->mnLookahead = lookahead;
528 p->mnAction = action;
529 }
530 }
531 p->aLookahead[p->nLookahead].lookahead = lookahead;
532 p->aLookahead[p->nLookahead].action = action;
533 p->nLookahead++;
534}
535
536/*
537** Add the transaction set built up with prior calls to acttab_action()
538** into the current action table. Then reset the transaction set back
539** to an empty set in preparation for a new round of acttab_action() calls.
540**
541** Return the offset into the action table of the new transaction.
542*/
543int acttab_insert(acttab *p){
544 int i, j, k, n;
545 assert( p->nLookahead>0 );
546
547 /* Make sure we have enough space to hold the expanded action table
548 ** in the worst case. The worst case occurs if the transaction set
549 ** must be appended to the current action table
550 */
drh784d86f2004-02-19 18:41:53 +0000551 n = p->mxLookahead + 1;
drh8dc3e8f2010-01-07 03:53:03 +0000552 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000553 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000554 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
icculus9e44cf12010-02-14 17:14:22 +0000555 p->aAction = (struct lookahead_action *) realloc( p->aAction,
drh8b582012003-10-21 13:16:03 +0000556 sizeof(p->aAction[0])*p->nActionAlloc);
557 if( p->aAction==0 ){
558 fprintf(stderr,"malloc failed\n");
559 exit(1);
560 }
drhfdbf9282003-10-21 16:34:41 +0000561 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000562 p->aAction[i].lookahead = -1;
563 p->aAction[i].action = -1;
564 }
565 }
566
drh8dc3e8f2010-01-07 03:53:03 +0000567 /* Scan the existing action table looking for an offset that is a
568 ** duplicate of the current transaction set. Fall out of the loop
569 ** if and when the duplicate is found.
drh8b582012003-10-21 13:16:03 +0000570 **
571 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
572 */
drh8dc3e8f2010-01-07 03:53:03 +0000573 for(i=p->nAction-1; i>=0; i--){
drhf16371d2009-11-03 19:18:31 +0000574 if( p->aAction[i].lookahead==p->mnLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000575 /* All lookaheads and actions in the aLookahead[] transaction
576 ** must match against the candidate aAction[i] entry. */
drh8b582012003-10-21 13:16:03 +0000577 if( p->aAction[i].action!=p->mnAction ) continue;
578 for(j=0; j<p->nLookahead; j++){
579 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
580 if( k<0 || k>=p->nAction ) break;
581 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
582 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
583 }
584 if( j<p->nLookahead ) continue;
drh8dc3e8f2010-01-07 03:53:03 +0000585
586 /* No possible lookahead value that is not in the aLookahead[]
587 ** transaction is allowed to match aAction[i] */
drh8b582012003-10-21 13:16:03 +0000588 n = 0;
589 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000590 if( p->aAction[j].lookahead<0 ) continue;
591 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000592 }
drhfdbf9282003-10-21 16:34:41 +0000593 if( n==p->nLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000594 break; /* An exact match is found at offset i */
drhfdbf9282003-10-21 16:34:41 +0000595 }
drh8b582012003-10-21 13:16:03 +0000596 }
597 }
drh8dc3e8f2010-01-07 03:53:03 +0000598
599 /* If no existing offsets exactly match the current transaction, find an
600 ** an empty offset in the aAction[] table in which we can add the
601 ** aLookahead[] transaction.
602 */
drhf16371d2009-11-03 19:18:31 +0000603 if( i<0 ){
drh8dc3e8f2010-01-07 03:53:03 +0000604 /* Look for holes in the aAction[] table that fit the current
605 ** aLookahead[] transaction. Leave i set to the offset of the hole.
606 ** If no holes are found, i is left at p->nAction, which means the
607 ** transaction will be appended. */
608 for(i=0; i<p->nActionAlloc - p->mxLookahead; i++){
drhf16371d2009-11-03 19:18:31 +0000609 if( p->aAction[i].lookahead<0 ){
610 for(j=0; j<p->nLookahead; j++){
611 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
612 if( k<0 ) break;
613 if( p->aAction[k].lookahead>=0 ) break;
614 }
615 if( j<p->nLookahead ) continue;
616 for(j=0; j<p->nAction; j++){
617 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
618 }
619 if( j==p->nAction ){
620 break; /* Fits in empty slots */
621 }
622 }
623 }
624 }
drh8b582012003-10-21 13:16:03 +0000625 /* Insert transaction set at index i. */
626 for(j=0; j<p->nLookahead; j++){
627 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
628 p->aAction[k] = p->aLookahead[j];
629 if( k>=p->nAction ) p->nAction = k+1;
630 }
631 p->nLookahead = 0;
632
633 /* Return the offset that is added to the lookahead in order to get the
634 ** index into yy_action of the action */
635 return i - p->mnLookahead;
636}
637
drh75897232000-05-29 14:26:00 +0000638/********************** From the file "build.c" *****************************/
639/*
640** Routines to construction the finite state machine for the LEMON
641** parser generator.
642*/
643
644/* Find a precedence symbol of every rule in the grammar.
645**
646** Those rules which have a precedence symbol coded in the input
647** grammar using the "[symbol]" construct will already have the
648** rp->precsym field filled. Other rules take as their precedence
649** symbol the first RHS symbol with a defined precedence. If there
650** are not RHS symbols with a defined precedence, the precedence
651** symbol field is left blank.
652*/
icculus9e44cf12010-02-14 17:14:22 +0000653void FindRulePrecedences(struct lemon *xp)
drh75897232000-05-29 14:26:00 +0000654{
655 struct rule *rp;
656 for(rp=xp->rule; rp; rp=rp->next){
657 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000658 int i, j;
659 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
660 struct symbol *sp = rp->rhs[i];
661 if( sp->type==MULTITERMINAL ){
662 for(j=0; j<sp->nsubsym; j++){
663 if( sp->subsym[j]->prec>=0 ){
664 rp->precsym = sp->subsym[j];
665 break;
666 }
667 }
668 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000669 rp->precsym = rp->rhs[i];
drh75897232000-05-29 14:26:00 +0000670 }
671 }
672 }
673 }
674 return;
675}
676
677/* Find all nonterminals which will generate the empty string.
678** Then go back and compute the first sets of every nonterminal.
679** The first set is the set of all terminal symbols which can begin
680** a string generated by that nonterminal.
681*/
icculus9e44cf12010-02-14 17:14:22 +0000682void FindFirstSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000683{
drhfd405312005-11-06 04:06:59 +0000684 int i, j;
drh75897232000-05-29 14:26:00 +0000685 struct rule *rp;
686 int progress;
687
688 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000689 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000690 }
691 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
692 lemp->symbols[i]->firstset = SetNew();
693 }
694
695 /* First compute all lambdas */
696 do{
697 progress = 0;
698 for(rp=lemp->rule; rp; rp=rp->next){
699 if( rp->lhs->lambda ) continue;
700 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000701 struct symbol *sp = rp->rhs[i];
drhaa9f1122007-08-23 02:50:56 +0000702 if( sp->type!=TERMINAL || sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000703 }
704 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000705 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000706 progress = 1;
707 }
708 }
709 }while( progress );
710
711 /* Now compute all first sets */
712 do{
713 struct symbol *s1, *s2;
714 progress = 0;
715 for(rp=lemp->rule; rp; rp=rp->next){
716 s1 = rp->lhs;
717 for(i=0; i<rp->nrhs; i++){
718 s2 = rp->rhs[i];
719 if( s2->type==TERMINAL ){
720 progress += SetAdd(s1->firstset,s2->index);
721 break;
drhfd405312005-11-06 04:06:59 +0000722 }else if( s2->type==MULTITERMINAL ){
723 for(j=0; j<s2->nsubsym; j++){
724 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
725 }
726 break;
drh75897232000-05-29 14:26:00 +0000727 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000728 if( s1->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000729 }else{
730 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000731 if( s2->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000732 }
733 }
734 }
735 }while( progress );
736 return;
737}
738
739/* Compute all LR(0) states for the grammar. Links
740** are added to between some states so that the LR(1) follow sets
741** can be computed later.
742*/
icculus9e44cf12010-02-14 17:14:22 +0000743PRIVATE struct state *getstate(struct lemon *); /* forward reference */
744void FindStates(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000745{
746 struct symbol *sp;
747 struct rule *rp;
748
749 Configlist_init();
750
751 /* Find the start symbol */
752 if( lemp->start ){
753 sp = Symbol_find(lemp->start);
754 if( sp==0 ){
755 ErrorMsg(lemp->filename,0,
756"The specified start symbol \"%s\" is not \
757in a nonterminal of the grammar. \"%s\" will be used as the start \
758symbol instead.",lemp->start,lemp->rule->lhs->name);
759 lemp->errorcnt++;
760 sp = lemp->rule->lhs;
761 }
762 }else{
763 sp = lemp->rule->lhs;
764 }
765
766 /* Make sure the start symbol doesn't occur on the right-hand side of
767 ** any rule. Report an error if it does. (YACC would generate a new
768 ** start symbol in this case.) */
769 for(rp=lemp->rule; rp; rp=rp->next){
770 int i;
771 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000772 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000773 ErrorMsg(lemp->filename,0,
774"The start symbol \"%s\" occurs on the \
775right-hand side of a rule. This will result in a parser which \
776does not work properly.",sp->name);
777 lemp->errorcnt++;
778 }
779 }
780 }
781
782 /* The basis configuration set for the first state
783 ** is all rules which have the start symbol as their
784 ** left-hand side */
785 for(rp=sp->rule; rp; rp=rp->nextlhs){
786 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000787 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000788 newcfp = Configlist_addbasis(rp,0);
789 SetAdd(newcfp->fws,0);
790 }
791
792 /* Compute the first state. All other states will be
793 ** computed automatically during the computation of the first one.
794 ** The returned pointer to the first state is not used. */
795 (void)getstate(lemp);
796 return;
797}
798
799/* Return a pointer to a state which is described by the configuration
800** list which has been built from calls to Configlist_add.
801*/
icculus9e44cf12010-02-14 17:14:22 +0000802PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
803PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000804{
805 struct config *cfp, *bp;
806 struct state *stp;
807
808 /* Extract the sorted basis of the new state. The basis was constructed
809 ** by prior calls to "Configlist_addbasis()". */
810 Configlist_sortbasis();
811 bp = Configlist_basis();
812
813 /* Get a state with the same basis */
814 stp = State_find(bp);
815 if( stp ){
816 /* A state with the same basis already exists! Copy all the follow-set
817 ** propagation links from the state under construction into the
818 ** preexisting state, then return a pointer to the preexisting state */
819 struct config *x, *y;
820 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
821 Plink_copy(&y->bplp,x->bplp);
822 Plink_delete(x->fplp);
823 x->fplp = x->bplp = 0;
824 }
825 cfp = Configlist_return();
826 Configlist_eat(cfp);
827 }else{
828 /* This really is a new state. Construct all the details */
829 Configlist_closure(lemp); /* Compute the configuration closure */
830 Configlist_sort(); /* Sort the configuration closure */
831 cfp = Configlist_return(); /* Get a pointer to the config list */
832 stp = State_new(); /* A new state structure */
833 MemoryCheck(stp);
834 stp->bp = bp; /* Remember the configuration basis */
835 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000836 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000837 stp->ap = 0; /* No actions, yet. */
838 State_insert(stp,stp->bp); /* Add to the state table */
839 buildshifts(lemp,stp); /* Recursively compute successor states */
840 }
841 return stp;
842}
843
drhfd405312005-11-06 04:06:59 +0000844/*
845** Return true if two symbols are the same.
846*/
icculus9e44cf12010-02-14 17:14:22 +0000847int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +0000848{
849 int i;
850 if( a==b ) return 1;
851 if( a->type!=MULTITERMINAL ) return 0;
852 if( b->type!=MULTITERMINAL ) return 0;
853 if( a->nsubsym!=b->nsubsym ) return 0;
854 for(i=0; i<a->nsubsym; i++){
855 if( a->subsym[i]!=b->subsym[i] ) return 0;
856 }
857 return 1;
858}
859
drh75897232000-05-29 14:26:00 +0000860/* Construct all successor states to the given state. A "successor"
861** state is any state which can be reached by a shift action.
862*/
icculus9e44cf12010-02-14 17:14:22 +0000863PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +0000864{
865 struct config *cfp; /* For looping thru the config closure of "stp" */
866 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +0000867 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +0000868 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
869 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
870 struct state *newstp; /* A pointer to a successor state */
871
872 /* Each configuration becomes complete after it contibutes to a successor
873 ** state. Initially, all configurations are incomplete */
874 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
875
876 /* Loop through all configurations of the state "stp" */
877 for(cfp=stp->cfp; cfp; cfp=cfp->next){
878 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
879 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
880 Configlist_reset(); /* Reset the new config set */
881 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
882
883 /* For every configuration in the state "stp" which has the symbol "sp"
884 ** following its dot, add the same configuration to the basis set under
885 ** construction but with the dot shifted one symbol to the right. */
886 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
887 if( bcfp->status==COMPLETE ) continue; /* Already used */
888 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
889 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +0000890 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +0000891 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +0000892 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
893 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +0000894 }
895
896 /* Get a pointer to the state described by the basis configuration set
897 ** constructed in the preceding loop */
898 newstp = getstate(lemp);
899
900 /* The state "newstp" is reached from the state "stp" by a shift action
901 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +0000902 if( sp->type==MULTITERMINAL ){
903 int i;
904 for(i=0; i<sp->nsubsym; i++){
905 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
906 }
907 }else{
908 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
909 }
drh75897232000-05-29 14:26:00 +0000910 }
911}
912
913/*
914** Construct the propagation links
915*/
icculus9e44cf12010-02-14 17:14:22 +0000916void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000917{
918 int i;
919 struct config *cfp, *other;
920 struct state *stp;
921 struct plink *plp;
922
923 /* Housekeeping detail:
924 ** Add to every propagate link a pointer back to the state to
925 ** which the link is attached. */
926 for(i=0; i<lemp->nstate; i++){
927 stp = lemp->sorted[i];
928 for(cfp=stp->cfp; cfp; cfp=cfp->next){
929 cfp->stp = stp;
930 }
931 }
932
933 /* Convert all backlinks into forward links. Only the forward
934 ** links are used in the follow-set computation. */
935 for(i=0; i<lemp->nstate; i++){
936 stp = lemp->sorted[i];
937 for(cfp=stp->cfp; cfp; cfp=cfp->next){
938 for(plp=cfp->bplp; plp; plp=plp->next){
939 other = plp->cfp;
940 Plink_add(&other->fplp,cfp);
941 }
942 }
943 }
944}
945
946/* Compute all followsets.
947**
948** A followset is the set of all symbols which can come immediately
949** after a configuration.
950*/
icculus9e44cf12010-02-14 17:14:22 +0000951void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000952{
953 int i;
954 struct config *cfp;
955 struct plink *plp;
956 int progress;
957 int change;
958
959 for(i=0; i<lemp->nstate; i++){
960 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
961 cfp->status = INCOMPLETE;
962 }
963 }
964
965 do{
966 progress = 0;
967 for(i=0; i<lemp->nstate; i++){
968 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
969 if( cfp->status==COMPLETE ) continue;
970 for(plp=cfp->fplp; plp; plp=plp->next){
971 change = SetUnion(plp->cfp->fws,cfp->fws);
972 if( change ){
973 plp->cfp->status = INCOMPLETE;
974 progress = 1;
975 }
976 }
977 cfp->status = COMPLETE;
978 }
979 }
980 }while( progress );
981}
982
icculus9e44cf12010-02-14 17:14:22 +0000983static int resolve_conflict(struct action *,struct action *, struct symbol *);
drh75897232000-05-29 14:26:00 +0000984
985/* Compute the reduce actions, and resolve conflicts.
986*/
icculus9e44cf12010-02-14 17:14:22 +0000987void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000988{
989 int i,j;
990 struct config *cfp;
991 struct state *stp;
992 struct symbol *sp;
993 struct rule *rp;
994
995 /* Add all of the reduce actions
996 ** A reduce action is added for each element of the followset of
997 ** a configuration which has its dot at the extreme right.
998 */
999 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
1000 stp = lemp->sorted[i];
1001 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
1002 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
1003 for(j=0; j<lemp->nterminal; j++){
1004 if( SetFind(cfp->fws,j) ){
1005 /* Add a reduce action to the state "stp" which will reduce by the
1006 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +00001007 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +00001008 }
1009 }
1010 }
1011 }
1012 }
1013
1014 /* Add the accepting token */
1015 if( lemp->start ){
1016 sp = Symbol_find(lemp->start);
1017 if( sp==0 ) sp = lemp->rule->lhs;
1018 }else{
1019 sp = lemp->rule->lhs;
1020 }
1021 /* Add to the first state (which is always the starting state of the
1022 ** finite state machine) an action to ACCEPT if the lookahead is the
1023 ** start nonterminal. */
1024 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1025
1026 /* Resolve conflicts */
1027 for(i=0; i<lemp->nstate; i++){
1028 struct action *ap, *nap;
1029 struct state *stp;
1030 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001031 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001032 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001033 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001034 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1035 /* The two actions "ap" and "nap" have the same lookahead.
1036 ** Figure out which one should be used */
1037 lemp->nconflict += resolve_conflict(ap,nap,lemp->errsym);
1038 }
1039 }
1040 }
1041
1042 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001043 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001044 for(i=0; i<lemp->nstate; i++){
1045 struct action *ap;
1046 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001047 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001048 }
1049 }
1050 for(rp=lemp->rule; rp; rp=rp->next){
1051 if( rp->canReduce ) continue;
1052 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1053 lemp->errorcnt++;
1054 }
1055}
1056
1057/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001058** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001059**
1060** NO LONGER TRUE:
1061** To resolve a conflict, first look to see if either action
1062** is on an error rule. In that case, take the action which
1063** is not associated with the error rule. If neither or both
1064** actions are associated with an error rule, then try to
1065** use precedence to resolve the conflict.
1066**
1067** If either action is a SHIFT, then it must be apx. This
1068** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1069*/
icculus9e44cf12010-02-14 17:14:22 +00001070static int resolve_conflict(
1071 struct action *apx,
1072 struct action *apy,
1073 struct symbol *errsym /* The error symbol (if defined. NULL otherwise) */
1074){
drh75897232000-05-29 14:26:00 +00001075 struct symbol *spx, *spy;
1076 int errcnt = 0;
1077 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001078 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001079 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001080 errcnt++;
1081 }
drh75897232000-05-29 14:26:00 +00001082 if( apx->type==SHIFT && apy->type==REDUCE ){
1083 spx = apx->sp;
1084 spy = apy->x.rp->precsym;
1085 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1086 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001087 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001088 errcnt++;
1089 }else if( spx->prec>spy->prec ){ /* Lower precedence wins */
1090 apy->type = RD_RESOLVED;
1091 }else if( spx->prec<spy->prec ){
1092 apx->type = SH_RESOLVED;
1093 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1094 apy->type = RD_RESOLVED; /* associativity */
1095 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1096 apx->type = SH_RESOLVED;
1097 }else{
1098 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh9892c5d2007-12-21 00:02:11 +00001099 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001100 errcnt++;
1101 }
1102 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1103 spx = apx->x.rp->precsym;
1104 spy = apy->x.rp->precsym;
1105 if( spx==0 || spy==0 || spx->prec<0 ||
1106 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001107 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001108 errcnt++;
1109 }else if( spx->prec>spy->prec ){
1110 apy->type = RD_RESOLVED;
1111 }else if( spx->prec<spy->prec ){
1112 apx->type = RD_RESOLVED;
1113 }
1114 }else{
drhb59499c2002-02-23 18:45:13 +00001115 assert(
1116 apx->type==SH_RESOLVED ||
1117 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001118 apx->type==SSCONFLICT ||
1119 apx->type==SRCONFLICT ||
1120 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001121 apy->type==SH_RESOLVED ||
1122 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001123 apy->type==SSCONFLICT ||
1124 apy->type==SRCONFLICT ||
1125 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001126 );
1127 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1128 ** REDUCEs on the list. If we reach this point it must be because
1129 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001130 }
1131 return errcnt;
1132}
1133/********************* From the file "configlist.c" *************************/
1134/*
1135** Routines to processing a configuration list and building a state
1136** in the LEMON parser generator.
1137*/
1138
1139static struct config *freelist = 0; /* List of free configurations */
1140static struct config *current = 0; /* Top of list of configurations */
1141static struct config **currentend = 0; /* Last on list of configs */
1142static struct config *basis = 0; /* Top of list of basis configs */
1143static struct config **basisend = 0; /* End of list of basis configs */
1144
1145/* Return a pointer to a new configuration */
1146PRIVATE struct config *newconfig(){
icculus9e44cf12010-02-14 17:14:22 +00001147 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001148 if( freelist==0 ){
1149 int i;
1150 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001151 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001152 if( freelist==0 ){
1153 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1154 exit(1);
1155 }
1156 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1157 freelist[amt-1].next = 0;
1158 }
icculus9e44cf12010-02-14 17:14:22 +00001159 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001160 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001161 return newcfg;
drh75897232000-05-29 14:26:00 +00001162}
1163
1164/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001165PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001166{
1167 old->next = freelist;
1168 freelist = old;
1169}
1170
1171/* Initialized the configuration list builder */
1172void Configlist_init(){
1173 current = 0;
1174 currentend = &current;
1175 basis = 0;
1176 basisend = &basis;
1177 Configtable_init();
1178 return;
1179}
1180
1181/* Initialized the configuration list builder */
1182void Configlist_reset(){
1183 current = 0;
1184 currentend = &current;
1185 basis = 0;
1186 basisend = &basis;
1187 Configtable_clear(0);
1188 return;
1189}
1190
1191/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001192struct config *Configlist_add(
1193 struct rule *rp, /* The rule */
1194 int dot /* Index into the RHS of the rule where the dot goes */
1195){
drh75897232000-05-29 14:26:00 +00001196 struct config *cfp, model;
1197
1198 assert( currentend!=0 );
1199 model.rp = rp;
1200 model.dot = dot;
1201 cfp = Configtable_find(&model);
1202 if( cfp==0 ){
1203 cfp = newconfig();
1204 cfp->rp = rp;
1205 cfp->dot = dot;
1206 cfp->fws = SetNew();
1207 cfp->stp = 0;
1208 cfp->fplp = cfp->bplp = 0;
1209 cfp->next = 0;
1210 cfp->bp = 0;
1211 *currentend = cfp;
1212 currentend = &cfp->next;
1213 Configtable_insert(cfp);
1214 }
1215 return cfp;
1216}
1217
1218/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001219struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001220{
1221 struct config *cfp, model;
1222
1223 assert( basisend!=0 );
1224 assert( currentend!=0 );
1225 model.rp = rp;
1226 model.dot = dot;
1227 cfp = Configtable_find(&model);
1228 if( cfp==0 ){
1229 cfp = newconfig();
1230 cfp->rp = rp;
1231 cfp->dot = dot;
1232 cfp->fws = SetNew();
1233 cfp->stp = 0;
1234 cfp->fplp = cfp->bplp = 0;
1235 cfp->next = 0;
1236 cfp->bp = 0;
1237 *currentend = cfp;
1238 currentend = &cfp->next;
1239 *basisend = cfp;
1240 basisend = &cfp->bp;
1241 Configtable_insert(cfp);
1242 }
1243 return cfp;
1244}
1245
1246/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001247void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001248{
1249 struct config *cfp, *newcfp;
1250 struct rule *rp, *newrp;
1251 struct symbol *sp, *xsp;
1252 int i, dot;
1253
1254 assert( currentend!=0 );
1255 for(cfp=current; cfp; cfp=cfp->next){
1256 rp = cfp->rp;
1257 dot = cfp->dot;
1258 if( dot>=rp->nrhs ) continue;
1259 sp = rp->rhs[dot];
1260 if( sp->type==NONTERMINAL ){
1261 if( sp->rule==0 && sp!=lemp->errsym ){
1262 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1263 sp->name);
1264 lemp->errorcnt++;
1265 }
1266 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1267 newcfp = Configlist_add(newrp,0);
1268 for(i=dot+1; i<rp->nrhs; i++){
1269 xsp = rp->rhs[i];
1270 if( xsp->type==TERMINAL ){
1271 SetAdd(newcfp->fws,xsp->index);
1272 break;
drhfd405312005-11-06 04:06:59 +00001273 }else if( xsp->type==MULTITERMINAL ){
1274 int k;
1275 for(k=0; k<xsp->nsubsym; k++){
1276 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1277 }
1278 break;
drh75897232000-05-29 14:26:00 +00001279 }else{
1280 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001281 if( xsp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +00001282 }
1283 }
1284 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1285 }
1286 }
1287 }
1288 return;
1289}
1290
1291/* Sort the configuration list */
1292void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001293 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001294 currentend = 0;
1295 return;
1296}
1297
1298/* Sort the basis configuration list */
1299void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001300 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001301 basisend = 0;
1302 return;
1303}
1304
1305/* Return a pointer to the head of the configuration list and
1306** reset the list */
1307struct config *Configlist_return(){
1308 struct config *old;
1309 old = current;
1310 current = 0;
1311 currentend = 0;
1312 return old;
1313}
1314
1315/* Return a pointer to the head of the configuration list and
1316** reset the list */
1317struct config *Configlist_basis(){
1318 struct config *old;
1319 old = basis;
1320 basis = 0;
1321 basisend = 0;
1322 return old;
1323}
1324
1325/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001326void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001327{
1328 struct config *nextcfp;
1329 for(; cfp; cfp=nextcfp){
1330 nextcfp = cfp->next;
1331 assert( cfp->fplp==0 );
1332 assert( cfp->bplp==0 );
1333 if( cfp->fws ) SetFree(cfp->fws);
1334 deleteconfig(cfp);
1335 }
1336 return;
1337}
1338/***************** From the file "error.c" *********************************/
1339/*
1340** Code for printing error message.
1341*/
1342
drhf9a2e7b2003-04-15 01:49:48 +00001343void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus1c11f742010-02-15 00:01:04 +00001344 fprintf(stderr, "%s:%d: ", filename, lineno);
1345 va_start(ap, format);
1346 vfprintf(stderr,format,ap);
1347 va_end(ap);
1348 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001349}
1350/**************** From the file "main.c" ************************************/
1351/*
1352** Main program file for the LEMON parser generator.
1353*/
1354
1355/* Report an out-of-memory condition and abort. This function
1356** is used mostly by the "MemoryCheck" macro in struct.h
1357*/
1358void memory_error(){
1359 fprintf(stderr,"Out of memory. Aborting...\n");
1360 exit(1);
1361}
1362
drh6d08b4d2004-07-20 12:45:22 +00001363static int nDefine = 0; /* Number of -D options on the command line */
1364static char **azDefine = 0; /* Name of the -D macros */
1365
1366/* This routine is called with the argument to each -D command-line option.
1367** Add the macro defined to the azDefine array.
1368*/
1369static void handle_D_option(char *z){
1370 char **paz;
1371 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001372 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001373 if( azDefine==0 ){
1374 fprintf(stderr,"out of memory\n");
1375 exit(1);
1376 }
1377 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001378 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001379 if( *paz==0 ){
1380 fprintf(stderr,"out of memory\n");
1381 exit(1);
1382 }
1383 strcpy(*paz, z);
1384 for(z=*paz; *z && *z!='='; z++){}
1385 *z = 0;
1386}
1387
icculus3e143bd2010-02-14 00:48:49 +00001388static char *user_templatename = NULL;
1389static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001390 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001391 if( user_templatename==0 ){
1392 memory_error();
1393 }
1394 strcpy(user_templatename, z);
1395}
drh75897232000-05-29 14:26:00 +00001396
1397/* The main program. Parse the command line and do it... */
icculus9e44cf12010-02-14 17:14:22 +00001398int main(int argc, char **argv)
drh75897232000-05-29 14:26:00 +00001399{
1400 static int version = 0;
1401 static int rpflag = 0;
1402 static int basisflag = 0;
1403 static int compress = 0;
1404 static int quiet = 0;
1405 static int statistics = 0;
1406 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001407 static int nolinenosflag = 0;
drh75897232000-05-29 14:26:00 +00001408 static struct s_options options[] = {
1409 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1410 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001411 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
icculus3e143bd2010-02-14 00:48:49 +00001412 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
drh75897232000-05-29 14:26:00 +00001413 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
shane58543932008-12-10 20:10:04 +00001414 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1415 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drh75897232000-05-29 14:26:00 +00001416 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drh6d08b4d2004-07-20 12:45:22 +00001417 {OPT_FLAG, "s", (char*)&statistics,
1418 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001419 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1420 {OPT_FLAG,0,0,0}
1421 };
1422 int i;
icculus42585cf2010-02-14 05:19:56 +00001423 int exitcode;
drh75897232000-05-29 14:26:00 +00001424 struct lemon lem;
1425
icculusf5ad8242010-02-14 05:34:42 +00001426 atexit(LemonAtExit);
1427
drhb0c86772000-06-02 23:21:26 +00001428 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001429 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001430 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001431 exit(0);
1432 }
drhb0c86772000-06-02 23:21:26 +00001433 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001434 fprintf(stderr,"Exactly one filename argument is required.\n");
1435 exit(1);
1436 }
drh954f6b42006-06-13 13:27:46 +00001437 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001438 lem.errorcnt = 0;
icculus42585cf2010-02-14 05:19:56 +00001439 lem.nexpected = -1;
drh75897232000-05-29 14:26:00 +00001440
1441 /* Initialize the machine */
1442 Strsafe_init();
1443 Symbol_init();
1444 State_init();
1445 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001446 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001447 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001448 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001449 Symbol_new("$");
1450 lem.errsym = Symbol_new("error");
drhc4dd3fd2008-01-22 01:48:05 +00001451 lem.errsym->useCnt = 0;
drh75897232000-05-29 14:26:00 +00001452
1453 /* Parse the input file */
1454 Parse(&lem);
1455 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001456 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001457 fprintf(stderr,"Empty grammar.\n");
1458 exit(1);
1459 }
1460
1461 /* Count and index the symbols of the grammar */
1462 lem.nsymbol = Symbol_count();
1463 Symbol_new("{default}");
1464 lem.symbols = Symbol_arrayof();
drh60d31652004-02-22 00:08:04 +00001465 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
icculus9e44cf12010-02-14 17:14:22 +00001466 qsort(lem.symbols,lem.nsymbol+1,sizeof(struct symbol*), Symbolcmpp);
drh75897232000-05-29 14:26:00 +00001467 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
1468 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1469 lem.nterminal = i;
1470
1471 /* Generate a reprint of the grammar, if requested on the command line */
1472 if( rpflag ){
1473 Reprint(&lem);
1474 }else{
1475 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001476 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001477
1478 /* Find the precedence for every production rule (that has one) */
1479 FindRulePrecedences(&lem);
1480
1481 /* Compute the lambda-nonterminals and the first-sets for every
1482 ** nonterminal */
1483 FindFirstSets(&lem);
1484
1485 /* Compute all LR(0) states. Also record follow-set propagation
1486 ** links so that the follow-set can be computed later */
1487 lem.nstate = 0;
1488 FindStates(&lem);
1489 lem.sorted = State_arrayof();
1490
1491 /* Tie up loose ends on the propagation links */
1492 FindLinks(&lem);
1493
1494 /* Compute the follow set of every reducible configuration */
1495 FindFollowSets(&lem);
1496
1497 /* Compute the action tables */
1498 FindActions(&lem);
1499
1500 /* Compress the action tables */
1501 if( compress==0 ) CompressTables(&lem);
1502
drhada354d2005-11-05 15:03:59 +00001503 /* Reorder and renumber the states so that states with fewer choices
1504 ** occur at the end. */
1505 ResortStates(&lem);
1506
drh75897232000-05-29 14:26:00 +00001507 /* Generate a report of the parser generated. (the "y.output" file) */
1508 if( !quiet ) ReportOutput(&lem);
1509
1510 /* Generate the source code for the parser */
1511 ReportTable(&lem, mhflag);
1512
1513 /* Produce a header file for use by the scanner. (This step is
1514 ** omitted if the "-m" option is used because makeheaders will
1515 ** generate the file for us.) */
1516 if( !mhflag ) ReportHeader(&lem);
1517 }
1518 if( statistics ){
1519 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1520 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1521 printf(" %d states, %d parser table entries, %d conflicts\n",
1522 lem.nstate, lem.tablesize, lem.nconflict);
1523 }
icculus42585cf2010-02-14 05:19:56 +00001524 if( lem.nexpected < 0 ) {
1525 lem.nexpected = 0; /* grammar didn't have an %expect declaration. */
drh75897232000-05-29 14:26:00 +00001526 }
icculus42585cf2010-02-14 05:19:56 +00001527 if( lem.nconflict != lem.nexpected ){
1528 fprintf(stderr,"%d parsing conflicts (%d expected).\n",lem.nconflict,lem.nexpected);
1529 }
1530
1531 /* return 0 on success, 1 on failure. */
1532 exitcode = ((lem.errorcnt > 0) || (lem.nconflict != lem.nexpected)) ? 1 : 0;
icculusf5ad8242010-02-14 05:34:42 +00001533 successful_exit = (exitcode == 0);
icculus42585cf2010-02-14 05:19:56 +00001534 exit(exitcode);
1535 return (exitcode);
drh75897232000-05-29 14:26:00 +00001536}
1537/******************** From the file "msort.c" *******************************/
1538/*
1539** A generic merge-sort program.
1540**
1541** USAGE:
1542** Let "ptr" be a pointer to some structure which is at the head of
1543** a null-terminated list. Then to sort the list call:
1544**
1545** ptr = msort(ptr,&(ptr->next),cmpfnc);
1546**
1547** In the above, "cmpfnc" is a pointer to a function which compares
1548** two instances of the structure and returns an integer, as in
1549** strcmp. The second argument is a pointer to the pointer to the
1550** second element of the linked list. This address is used to compute
1551** the offset to the "next" field within the structure. The offset to
1552** the "next" field must be constant for all structures in the list.
1553**
1554** The function returns a new pointer which is the head of the list
1555** after sorting.
1556**
1557** ALGORITHM:
1558** Merge-sort.
1559*/
1560
1561/*
1562** Return a pointer to the next structure in the linked list.
1563*/
drhba99af52001-10-25 20:37:16 +00001564#define NEXT(A) (*(char**)(((unsigned long)A)+offset))
drh75897232000-05-29 14:26:00 +00001565
1566/*
1567** Inputs:
1568** a: A sorted, null-terminated linked list. (May be null).
1569** b: A sorted, null-terminated linked list. (May be null).
1570** cmp: A pointer to the comparison function.
1571** offset: Offset in the structure to the "next" field.
1572**
1573** Return Value:
1574** A pointer to the head of a sorted list containing the elements
1575** of both a and b.
1576**
1577** Side effects:
1578** The "next" pointers for elements in the lists a and b are
1579** changed.
1580*/
drhe9278182007-07-18 18:16:29 +00001581static char *merge(
1582 char *a,
1583 char *b,
1584 int (*cmp)(const char*,const char*),
1585 int offset
1586){
drh75897232000-05-29 14:26:00 +00001587 char *ptr, *head;
1588
1589 if( a==0 ){
1590 head = b;
1591 }else if( b==0 ){
1592 head = a;
1593 }else{
drhe594bc32009-11-03 13:02:25 +00001594 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001595 ptr = a;
1596 a = NEXT(a);
1597 }else{
1598 ptr = b;
1599 b = NEXT(b);
1600 }
1601 head = ptr;
1602 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001603 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001604 NEXT(ptr) = a;
1605 ptr = a;
1606 a = NEXT(a);
1607 }else{
1608 NEXT(ptr) = b;
1609 ptr = b;
1610 b = NEXT(b);
1611 }
1612 }
1613 if( a ) NEXT(ptr) = a;
1614 else NEXT(ptr) = b;
1615 }
1616 return head;
1617}
1618
1619/*
1620** Inputs:
1621** list: Pointer to a singly-linked list of structures.
1622** next: Pointer to pointer to the second element of the list.
1623** cmp: A comparison function.
1624**
1625** Return Value:
1626** A pointer to the head of a sorted list containing the elements
1627** orginally in list.
1628**
1629** Side effects:
1630** The "next" pointers for elements in list are changed.
1631*/
1632#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001633static char *msort(
1634 char *list,
1635 char **next,
1636 int (*cmp)(const char*,const char*)
1637){
drhba99af52001-10-25 20:37:16 +00001638 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001639 char *ep;
1640 char *set[LISTSIZE];
1641 int i;
drhba99af52001-10-25 20:37:16 +00001642 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001643 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1644 while( list ){
1645 ep = list;
1646 list = NEXT(list);
1647 NEXT(ep) = 0;
1648 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1649 ep = merge(ep,set[i],cmp,offset);
1650 set[i] = 0;
1651 }
1652 set[i] = ep;
1653 }
1654 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001655 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001656 return ep;
1657}
1658/************************ From the file "option.c" **************************/
1659static char **argv;
1660static struct s_options *op;
1661static FILE *errstream;
1662
1663#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1664
1665/*
1666** Print the command line with a carrot pointing to the k-th character
1667** of the n-th field.
1668*/
icculus9e44cf12010-02-14 17:14:22 +00001669static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001670{
1671 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001672 if( argv[0] ) fprintf(err,"%s",argv[0]);
drh87cf1372008-08-13 20:09:06 +00001673 spcnt = lemonStrlen(argv[0]) + 1;
drh75897232000-05-29 14:26:00 +00001674 for(i=1; i<n && argv[i]; i++){
1675 fprintf(err," %s",argv[i]);
drh87cf1372008-08-13 20:09:06 +00001676 spcnt += lemonStrlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001677 }
1678 spcnt += k;
1679 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1680 if( spcnt<20 ){
1681 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1682 }else{
1683 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1684 }
1685}
1686
1687/*
1688** Return the index of the N-th non-switch argument. Return -1
1689** if N is out of range.
1690*/
icculus9e44cf12010-02-14 17:14:22 +00001691static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001692{
1693 int i;
1694 int dashdash = 0;
1695 if( argv!=0 && *argv!=0 ){
1696 for(i=1; argv[i]; i++){
1697 if( dashdash || !ISOPT(argv[i]) ){
1698 if( n==0 ) return i;
1699 n--;
1700 }
1701 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1702 }
1703 }
1704 return -1;
1705}
1706
1707static char emsg[] = "Command line syntax error: ";
1708
1709/*
1710** Process a flag command line argument.
1711*/
icculus9e44cf12010-02-14 17:14:22 +00001712static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001713{
1714 int v;
1715 int errcnt = 0;
1716 int j;
1717 for(j=0; op[j].label; j++){
drh87cf1372008-08-13 20:09:06 +00001718 if( strncmp(&argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001719 }
1720 v = argv[i][0]=='-' ? 1 : 0;
1721 if( op[j].label==0 ){
1722 if( err ){
1723 fprintf(err,"%sundefined option.\n",emsg);
1724 errline(i,1,err);
1725 }
1726 errcnt++;
1727 }else if( op[j].type==OPT_FLAG ){
1728 *((int*)op[j].arg) = v;
1729 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001730 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001731 }else if( op[j].type==OPT_FSTR ){
icculus9e44cf12010-02-14 17:14:22 +00001732 (*(void(*)(char *))(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001733 }else{
1734 if( err ){
1735 fprintf(err,"%smissing argument on switch.\n",emsg);
1736 errline(i,1,err);
1737 }
1738 errcnt++;
1739 }
1740 return errcnt;
1741}
1742
1743/*
1744** Process a command line switch which has an argument.
1745*/
icculus9e44cf12010-02-14 17:14:22 +00001746static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001747{
1748 int lv = 0;
1749 double dv = 0.0;
1750 char *sv = 0, *end;
1751 char *cp;
1752 int j;
1753 int errcnt = 0;
1754 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001755 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001756 *cp = 0;
1757 for(j=0; op[j].label; j++){
1758 if( strcmp(argv[i],op[j].label)==0 ) break;
1759 }
1760 *cp = '=';
1761 if( op[j].label==0 ){
1762 if( err ){
1763 fprintf(err,"%sundefined option.\n",emsg);
1764 errline(i,0,err);
1765 }
1766 errcnt++;
1767 }else{
1768 cp++;
1769 switch( op[j].type ){
1770 case OPT_FLAG:
1771 case OPT_FFLAG:
1772 if( err ){
1773 fprintf(err,"%soption requires an argument.\n",emsg);
1774 errline(i,0,err);
1775 }
1776 errcnt++;
1777 break;
1778 case OPT_DBL:
1779 case OPT_FDBL:
1780 dv = strtod(cp,&end);
1781 if( *end ){
1782 if( err ){
1783 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001784 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001785 }
1786 errcnt++;
1787 }
1788 break;
1789 case OPT_INT:
1790 case OPT_FINT:
1791 lv = strtol(cp,&end,0);
1792 if( *end ){
1793 if( err ){
1794 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001795 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001796 }
1797 errcnt++;
1798 }
1799 break;
1800 case OPT_STR:
1801 case OPT_FSTR:
1802 sv = cp;
1803 break;
1804 }
1805 switch( op[j].type ){
1806 case OPT_FLAG:
1807 case OPT_FFLAG:
1808 break;
1809 case OPT_DBL:
1810 *(double*)(op[j].arg) = dv;
1811 break;
1812 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00001813 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00001814 break;
1815 case OPT_INT:
1816 *(int*)(op[j].arg) = lv;
1817 break;
1818 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00001819 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00001820 break;
1821 case OPT_STR:
1822 *(char**)(op[j].arg) = sv;
1823 break;
1824 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00001825 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00001826 break;
1827 }
1828 }
1829 return errcnt;
1830}
1831
icculus9e44cf12010-02-14 17:14:22 +00001832int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00001833{
1834 int errcnt = 0;
1835 argv = a;
1836 op = o;
1837 errstream = err;
1838 if( argv && *argv && op ){
1839 int i;
1840 for(i=1; argv[i]; i++){
1841 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1842 errcnt += handleflags(i,err);
1843 }else if( strchr(argv[i],'=') ){
1844 errcnt += handleswitch(i,err);
1845 }
1846 }
1847 }
1848 if( errcnt>0 ){
1849 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001850 OptPrint();
drh75897232000-05-29 14:26:00 +00001851 exit(1);
1852 }
1853 return 0;
1854}
1855
drhb0c86772000-06-02 23:21:26 +00001856int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001857 int cnt = 0;
1858 int dashdash = 0;
1859 int i;
1860 if( argv!=0 && argv[0]!=0 ){
1861 for(i=1; argv[i]; i++){
1862 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1863 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1864 }
1865 }
1866 return cnt;
1867}
1868
icculus9e44cf12010-02-14 17:14:22 +00001869char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00001870{
1871 int i;
1872 i = argindex(n);
1873 return i>=0 ? argv[i] : 0;
1874}
1875
icculus9e44cf12010-02-14 17:14:22 +00001876void OptErr(int n)
drh75897232000-05-29 14:26:00 +00001877{
1878 int i;
1879 i = argindex(n);
1880 if( i>=0 ) errline(i,0,errstream);
1881}
1882
drhb0c86772000-06-02 23:21:26 +00001883void OptPrint(){
drh75897232000-05-29 14:26:00 +00001884 int i;
1885 int max, len;
1886 max = 0;
1887 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00001888 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00001889 switch( op[i].type ){
1890 case OPT_FLAG:
1891 case OPT_FFLAG:
1892 break;
1893 case OPT_INT:
1894 case OPT_FINT:
1895 len += 9; /* length of "<integer>" */
1896 break;
1897 case OPT_DBL:
1898 case OPT_FDBL:
1899 len += 6; /* length of "<real>" */
1900 break;
1901 case OPT_STR:
1902 case OPT_FSTR:
1903 len += 8; /* length of "<string>" */
1904 break;
1905 }
1906 if( len>max ) max = len;
1907 }
1908 for(i=0; op[i].label; i++){
1909 switch( op[i].type ){
1910 case OPT_FLAG:
1911 case OPT_FFLAG:
1912 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
1913 break;
1914 case OPT_INT:
1915 case OPT_FINT:
1916 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001917 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001918 break;
1919 case OPT_DBL:
1920 case OPT_FDBL:
1921 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001922 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001923 break;
1924 case OPT_STR:
1925 case OPT_FSTR:
1926 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001927 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001928 break;
1929 }
1930 }
1931}
1932/*********************** From the file "parse.c" ****************************/
1933/*
1934** Input file parser for the LEMON parser generator.
1935*/
1936
1937/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00001938enum e_state {
1939 INITIALIZE,
1940 WAITING_FOR_DECL_OR_RULE,
1941 WAITING_FOR_DECL_KEYWORD,
1942 WAITING_FOR_DECL_ARG,
1943 WAITING_FOR_PRECEDENCE_SYMBOL,
1944 WAITING_FOR_ARROW,
1945 IN_RHS,
1946 LHS_ALIAS_1,
1947 LHS_ALIAS_2,
1948 LHS_ALIAS_3,
1949 RHS_ALIAS_1,
1950 RHS_ALIAS_2,
1951 PRECEDENCE_MARK_1,
1952 PRECEDENCE_MARK_2,
1953 RESYNC_AFTER_RULE_ERROR,
1954 RESYNC_AFTER_DECL_ERROR,
1955 WAITING_FOR_DESTRUCTOR_SYMBOL,
1956 WAITING_FOR_DATATYPE_SYMBOL,
1957 WAITING_FOR_FALLBACK_ID,
1958 WAITING_FOR_EXPECT_VALUE,
1959 WAITING_FOR_WILDCARD_ID
1960};
drh75897232000-05-29 14:26:00 +00001961struct pstate {
1962 char *filename; /* Name of the input file */
1963 int tokenlineno; /* Linenumber at which current token starts */
1964 int errorcnt; /* Number of errors so far */
1965 char *tokenstart; /* Text of current token */
1966 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00001967 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00001968 struct symbol *fallback; /* The fallback token */
drh75897232000-05-29 14:26:00 +00001969 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00001970 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00001971 int nrhs; /* Number of right-hand side symbols seen */
1972 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00001973 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00001974 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00001975 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00001976 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00001977 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00001978 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00001979 enum e_assoc declassoc; /* Assign this association to decl arguments */
1980 int preccounter; /* Assign this precedence to decl arguments */
1981 struct rule *firstrule; /* Pointer to first rule in the grammar */
1982 struct rule *lastrule; /* Pointer to the most recently parsed rule */
1983};
1984
1985/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00001986static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00001987{
icculus42585cf2010-02-14 05:19:56 +00001988 char *endptr;
icculus9e44cf12010-02-14 17:14:22 +00001989 const char *x;
drh75897232000-05-29 14:26:00 +00001990 x = Strsafe(psp->tokenstart); /* Save the token permanently */
1991#if 0
1992 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
1993 x,psp->state);
1994#endif
1995 switch( psp->state ){
1996 case INITIALIZE:
1997 psp->prevrule = 0;
1998 psp->preccounter = 0;
1999 psp->firstrule = psp->lastrule = 0;
2000 psp->gp->nrule = 0;
2001 /* Fall thru to next case */
2002 case WAITING_FOR_DECL_OR_RULE:
2003 if( x[0]=='%' ){
2004 psp->state = WAITING_FOR_DECL_KEYWORD;
2005 }else if( islower(x[0]) ){
2006 psp->lhs = Symbol_new(x);
2007 psp->nrhs = 0;
2008 psp->lhsalias = 0;
2009 psp->state = WAITING_FOR_ARROW;
2010 }else if( x[0]=='{' ){
2011 if( psp->prevrule==0 ){
2012 ErrorMsg(psp->filename,psp->tokenlineno,
drh34ff57b2008-07-14 12:27:51 +00002013"There is no prior rule opon which to attach the code \
drh75897232000-05-29 14:26:00 +00002014fragment which begins on this line.");
2015 psp->errorcnt++;
2016 }else if( psp->prevrule->code!=0 ){
2017 ErrorMsg(psp->filename,psp->tokenlineno,
2018"Code fragment beginning on this line is not the first \
2019to follow the previous rule.");
2020 psp->errorcnt++;
2021 }else{
2022 psp->prevrule->line = psp->tokenlineno;
2023 psp->prevrule->code = &x[1];
2024 }
2025 }else if( x[0]=='[' ){
2026 psp->state = PRECEDENCE_MARK_1;
2027 }else{
2028 ErrorMsg(psp->filename,psp->tokenlineno,
2029 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2030 x);
2031 psp->errorcnt++;
2032 }
2033 break;
2034 case PRECEDENCE_MARK_1:
2035 if( !isupper(x[0]) ){
2036 ErrorMsg(psp->filename,psp->tokenlineno,
2037 "The precedence symbol must be a terminal.");
2038 psp->errorcnt++;
2039 }else if( psp->prevrule==0 ){
2040 ErrorMsg(psp->filename,psp->tokenlineno,
2041 "There is no prior rule to assign precedence \"[%s]\".",x);
2042 psp->errorcnt++;
2043 }else if( psp->prevrule->precsym!=0 ){
2044 ErrorMsg(psp->filename,psp->tokenlineno,
2045"Precedence mark on this line is not the first \
2046to follow the previous rule.");
2047 psp->errorcnt++;
2048 }else{
2049 psp->prevrule->precsym = Symbol_new(x);
2050 }
2051 psp->state = PRECEDENCE_MARK_2;
2052 break;
2053 case PRECEDENCE_MARK_2:
2054 if( x[0]!=']' ){
2055 ErrorMsg(psp->filename,psp->tokenlineno,
2056 "Missing \"]\" on precedence mark.");
2057 psp->errorcnt++;
2058 }
2059 psp->state = WAITING_FOR_DECL_OR_RULE;
2060 break;
2061 case WAITING_FOR_ARROW:
2062 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2063 psp->state = IN_RHS;
2064 }else if( x[0]=='(' ){
2065 psp->state = LHS_ALIAS_1;
2066 }else{
2067 ErrorMsg(psp->filename,psp->tokenlineno,
2068 "Expected to see a \":\" following the LHS symbol \"%s\".",
2069 psp->lhs->name);
2070 psp->errorcnt++;
2071 psp->state = RESYNC_AFTER_RULE_ERROR;
2072 }
2073 break;
2074 case LHS_ALIAS_1:
2075 if( isalpha(x[0]) ){
2076 psp->lhsalias = x;
2077 psp->state = LHS_ALIAS_2;
2078 }else{
2079 ErrorMsg(psp->filename,psp->tokenlineno,
2080 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2081 x,psp->lhs->name);
2082 psp->errorcnt++;
2083 psp->state = RESYNC_AFTER_RULE_ERROR;
2084 }
2085 break;
2086 case LHS_ALIAS_2:
2087 if( x[0]==')' ){
2088 psp->state = LHS_ALIAS_3;
2089 }else{
2090 ErrorMsg(psp->filename,psp->tokenlineno,
2091 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2092 psp->errorcnt++;
2093 psp->state = RESYNC_AFTER_RULE_ERROR;
2094 }
2095 break;
2096 case LHS_ALIAS_3:
2097 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2098 psp->state = IN_RHS;
2099 }else{
2100 ErrorMsg(psp->filename,psp->tokenlineno,
2101 "Missing \"->\" following: \"%s(%s)\".",
2102 psp->lhs->name,psp->lhsalias);
2103 psp->errorcnt++;
2104 psp->state = RESYNC_AFTER_RULE_ERROR;
2105 }
2106 break;
2107 case IN_RHS:
2108 if( x[0]=='.' ){
2109 struct rule *rp;
drh9892c5d2007-12-21 00:02:11 +00002110 rp = (struct rule *)calloc( sizeof(struct rule) +
2111 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002112 if( rp==0 ){
2113 ErrorMsg(psp->filename,psp->tokenlineno,
2114 "Can't allocate enough memory for this rule.");
2115 psp->errorcnt++;
2116 psp->prevrule = 0;
2117 }else{
2118 int i;
2119 rp->ruleline = psp->tokenlineno;
2120 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002121 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002122 for(i=0; i<psp->nrhs; i++){
2123 rp->rhs[i] = psp->rhs[i];
2124 rp->rhsalias[i] = psp->alias[i];
2125 }
2126 rp->lhs = psp->lhs;
2127 rp->lhsalias = psp->lhsalias;
2128 rp->nrhs = psp->nrhs;
2129 rp->code = 0;
2130 rp->precsym = 0;
2131 rp->index = psp->gp->nrule++;
2132 rp->nextlhs = rp->lhs->rule;
2133 rp->lhs->rule = rp;
2134 rp->next = 0;
2135 if( psp->firstrule==0 ){
2136 psp->firstrule = psp->lastrule = rp;
2137 }else{
2138 psp->lastrule->next = rp;
2139 psp->lastrule = rp;
2140 }
2141 psp->prevrule = rp;
2142 }
2143 psp->state = WAITING_FOR_DECL_OR_RULE;
2144 }else if( isalpha(x[0]) ){
2145 if( psp->nrhs>=MAXRHS ){
2146 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002147 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002148 x);
2149 psp->errorcnt++;
2150 psp->state = RESYNC_AFTER_RULE_ERROR;
2151 }else{
2152 psp->rhs[psp->nrhs] = Symbol_new(x);
2153 psp->alias[psp->nrhs] = 0;
2154 psp->nrhs++;
2155 }
drhfd405312005-11-06 04:06:59 +00002156 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2157 struct symbol *msp = psp->rhs[psp->nrhs-1];
2158 if( msp->type!=MULTITERMINAL ){
2159 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002160 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002161 memset(msp, 0, sizeof(*msp));
2162 msp->type = MULTITERMINAL;
2163 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002164 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002165 msp->subsym[0] = origsp;
2166 msp->name = origsp->name;
2167 psp->rhs[psp->nrhs-1] = msp;
2168 }
2169 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002170 msp->subsym = (struct symbol **) realloc(msp->subsym,
2171 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002172 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2173 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2174 ErrorMsg(psp->filename,psp->tokenlineno,
2175 "Cannot form a compound containing a non-terminal");
2176 psp->errorcnt++;
2177 }
drh75897232000-05-29 14:26:00 +00002178 }else if( x[0]=='(' && psp->nrhs>0 ){
2179 psp->state = RHS_ALIAS_1;
2180 }else{
2181 ErrorMsg(psp->filename,psp->tokenlineno,
2182 "Illegal character on RHS of rule: \"%s\".",x);
2183 psp->errorcnt++;
2184 psp->state = RESYNC_AFTER_RULE_ERROR;
2185 }
2186 break;
2187 case RHS_ALIAS_1:
2188 if( isalpha(x[0]) ){
2189 psp->alias[psp->nrhs-1] = x;
2190 psp->state = RHS_ALIAS_2;
2191 }else{
2192 ErrorMsg(psp->filename,psp->tokenlineno,
2193 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2194 x,psp->rhs[psp->nrhs-1]->name);
2195 psp->errorcnt++;
2196 psp->state = RESYNC_AFTER_RULE_ERROR;
2197 }
2198 break;
2199 case RHS_ALIAS_2:
2200 if( x[0]==')' ){
2201 psp->state = IN_RHS;
2202 }else{
2203 ErrorMsg(psp->filename,psp->tokenlineno,
2204 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2205 psp->errorcnt++;
2206 psp->state = RESYNC_AFTER_RULE_ERROR;
2207 }
2208 break;
2209 case WAITING_FOR_DECL_KEYWORD:
2210 if( isalpha(x[0]) ){
2211 psp->declkeyword = x;
2212 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002213 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002214 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002215 psp->state = WAITING_FOR_DECL_ARG;
2216 if( strcmp(x,"name")==0 ){
2217 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002218 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002219 }else if( strcmp(x,"include")==0 ){
2220 psp->declargslot = &(psp->gp->include);
drh75897232000-05-29 14:26:00 +00002221 }else if( strcmp(x,"code")==0 ){
2222 psp->declargslot = &(psp->gp->extracode);
drh75897232000-05-29 14:26:00 +00002223 }else if( strcmp(x,"token_destructor")==0 ){
2224 psp->declargslot = &psp->gp->tokendest;
drh960e8c62001-04-03 16:53:21 +00002225 }else if( strcmp(x,"default_destructor")==0 ){
2226 psp->declargslot = &psp->gp->vardest;
drh75897232000-05-29 14:26:00 +00002227 }else if( strcmp(x,"token_prefix")==0 ){
2228 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002229 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002230 }else if( strcmp(x,"syntax_error")==0 ){
2231 psp->declargslot = &(psp->gp->error);
drh75897232000-05-29 14:26:00 +00002232 }else if( strcmp(x,"parse_accept")==0 ){
2233 psp->declargslot = &(psp->gp->accept);
drh75897232000-05-29 14:26:00 +00002234 }else if( strcmp(x,"parse_failure")==0 ){
2235 psp->declargslot = &(psp->gp->failure);
drh75897232000-05-29 14:26:00 +00002236 }else if( strcmp(x,"stack_overflow")==0 ){
2237 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002238 }else if( strcmp(x,"extra_argument")==0 ){
2239 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002240 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002241 }else if( strcmp(x,"token_type")==0 ){
2242 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002243 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002244 }else if( strcmp(x,"default_type")==0 ){
2245 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002246 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002247 }else if( strcmp(x,"stack_size")==0 ){
2248 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002249 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002250 }else if( strcmp(x,"start_symbol")==0 ){
2251 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002252 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002253 }else if( strcmp(x,"left")==0 ){
2254 psp->preccounter++;
2255 psp->declassoc = LEFT;
2256 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2257 }else if( strcmp(x,"right")==0 ){
2258 psp->preccounter++;
2259 psp->declassoc = RIGHT;
2260 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2261 }else if( strcmp(x,"nonassoc")==0 ){
2262 psp->preccounter++;
2263 psp->declassoc = NONE;
2264 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2265 }else if( strcmp(x,"destructor")==0 ){
2266 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
2267 }else if( strcmp(x,"type")==0 ){
2268 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002269 }else if( strcmp(x,"fallback")==0 ){
2270 psp->fallback = 0;
2271 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002272 }else if( strcmp(x,"wildcard")==0 ){
2273 psp->state = WAITING_FOR_WILDCARD_ID;
icculus42585cf2010-02-14 05:19:56 +00002274 }else if( strcmp(x,"expect")==0 ){
2275 if (psp->gp->nexpected >= 0) {
2276 ErrorMsg(psp->filename,psp->tokenlineno, "Multiple %expect declarations.");
2277 psp->errorcnt++;
2278 psp->state = RESYNC_AFTER_DECL_ERROR;
2279 } else {
2280 psp->state = WAITING_FOR_EXPECT_VALUE;
2281 }
drh75897232000-05-29 14:26:00 +00002282 }else{
2283 ErrorMsg(psp->filename,psp->tokenlineno,
2284 "Unknown declaration keyword: \"%%%s\".",x);
2285 psp->errorcnt++;
2286 psp->state = RESYNC_AFTER_DECL_ERROR;
2287 }
2288 }else{
2289 ErrorMsg(psp->filename,psp->tokenlineno,
2290 "Illegal declaration keyword: \"%s\".",x);
2291 psp->errorcnt++;
2292 psp->state = RESYNC_AFTER_DECL_ERROR;
2293 }
2294 break;
2295 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2296 if( !isalpha(x[0]) ){
2297 ErrorMsg(psp->filename,psp->tokenlineno,
2298 "Symbol name missing after %destructor keyword");
2299 psp->errorcnt++;
2300 psp->state = RESYNC_AFTER_DECL_ERROR;
2301 }else{
2302 struct symbol *sp = Symbol_new(x);
2303 psp->declargslot = &sp->destructor;
drh4dc8ef52008-07-01 17:13:57 +00002304 psp->decllinenoslot = &sp->destLineno;
drha5808f32008-04-27 22:19:44 +00002305 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002306 psp->state = WAITING_FOR_DECL_ARG;
2307 }
2308 break;
icculus42585cf2010-02-14 05:19:56 +00002309 case WAITING_FOR_EXPECT_VALUE:
2310 psp->gp->nexpected = (int) strtol(x, &endptr, 10);
2311 if( (*endptr != '\0') || (endptr == x) ) {
2312 ErrorMsg(psp->filename,psp->tokenlineno,
2313 "Integer expected after %%expect keyword");
2314 psp->errorcnt++;
2315 } else if (psp->gp->nexpected < 0) {
2316 ErrorMsg(psp->filename,psp->tokenlineno,
2317 "Integer can't be negative after %%expect keyword");
2318 psp->errorcnt++;
2319 }
2320 psp->state = WAITING_FOR_DECL_OR_RULE;
2321 break;
drh75897232000-05-29 14:26:00 +00002322 case WAITING_FOR_DATATYPE_SYMBOL:
2323 if( !isalpha(x[0]) ){
2324 ErrorMsg(psp->filename,psp->tokenlineno,
2325 "Symbol name missing after %destructor keyword");
2326 psp->errorcnt++;
2327 psp->state = RESYNC_AFTER_DECL_ERROR;
2328 }else{
2329 struct symbol *sp = Symbol_new(x);
2330 psp->declargslot = &sp->datatype;
drha5808f32008-04-27 22:19:44 +00002331 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002332 psp->state = WAITING_FOR_DECL_ARG;
2333 }
2334 break;
2335 case WAITING_FOR_PRECEDENCE_SYMBOL:
2336 if( x[0]=='.' ){
2337 psp->state = WAITING_FOR_DECL_OR_RULE;
2338 }else if( isupper(x[0]) ){
2339 struct symbol *sp;
2340 sp = Symbol_new(x);
2341 if( sp->prec>=0 ){
2342 ErrorMsg(psp->filename,psp->tokenlineno,
2343 "Symbol \"%s\" has already be given a precedence.",x);
2344 psp->errorcnt++;
2345 }else{
2346 sp->prec = psp->preccounter;
2347 sp->assoc = psp->declassoc;
2348 }
2349 }else{
2350 ErrorMsg(psp->filename,psp->tokenlineno,
2351 "Can't assign a precedence to \"%s\".",x);
2352 psp->errorcnt++;
2353 }
2354 break;
2355 case WAITING_FOR_DECL_ARG:
drha5808f32008-04-27 22:19:44 +00002356 if( x[0]=='{' || x[0]=='\"' || isalnum(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002357 const char *zOld, *zNew;
2358 char *zBuf, *z;
drha5808f32008-04-27 22:19:44 +00002359 int nOld, n, nLine, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002360 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002361 char zLine[50];
2362 zNew = x;
2363 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002364 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002365 if( *psp->declargslot ){
2366 zOld = *psp->declargslot;
2367 }else{
2368 zOld = "";
2369 }
drh87cf1372008-08-13 20:09:06 +00002370 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002371 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002372 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002373 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2374 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002375 for(z=psp->filename, nBack=0; *z; z++){
2376 if( *z=='\\' ) nBack++;
2377 }
2378 sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002379 nLine = lemonStrlen(zLine);
2380 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002381 }
icculus9e44cf12010-02-14 17:14:22 +00002382 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2383 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002384 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002385 if( nOld && zBuf[-1]!='\n' ){
2386 *(zBuf++) = '\n';
2387 }
2388 memcpy(zBuf, zLine, nLine);
2389 zBuf += nLine;
2390 *(zBuf++) = '"';
2391 for(z=psp->filename; *z; z++){
2392 if( *z=='\\' ){
2393 *(zBuf++) = '\\';
2394 }
2395 *(zBuf++) = *z;
2396 }
2397 *(zBuf++) = '"';
2398 *(zBuf++) = '\n';
2399 }
drh4dc8ef52008-07-01 17:13:57 +00002400 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2401 psp->decllinenoslot[0] = psp->tokenlineno;
2402 }
drha5808f32008-04-27 22:19:44 +00002403 memcpy(zBuf, zNew, nNew);
2404 zBuf += nNew;
2405 *zBuf = 0;
2406 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002407 }else{
2408 ErrorMsg(psp->filename,psp->tokenlineno,
2409 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2410 psp->errorcnt++;
2411 psp->state = RESYNC_AFTER_DECL_ERROR;
2412 }
2413 break;
drh0bd1f4e2002-06-06 18:54:39 +00002414 case WAITING_FOR_FALLBACK_ID:
2415 if( x[0]=='.' ){
2416 psp->state = WAITING_FOR_DECL_OR_RULE;
2417 }else if( !isupper(x[0]) ){
2418 ErrorMsg(psp->filename, psp->tokenlineno,
2419 "%%fallback argument \"%s\" should be a token", x);
2420 psp->errorcnt++;
2421 }else{
2422 struct symbol *sp = Symbol_new(x);
2423 if( psp->fallback==0 ){
2424 psp->fallback = sp;
2425 }else if( sp->fallback ){
2426 ErrorMsg(psp->filename, psp->tokenlineno,
2427 "More than one fallback assigned to token %s", x);
2428 psp->errorcnt++;
2429 }else{
2430 sp->fallback = psp->fallback;
2431 psp->gp->has_fallback = 1;
2432 }
2433 }
2434 break;
drhe09daa92006-06-10 13:29:31 +00002435 case WAITING_FOR_WILDCARD_ID:
2436 if( x[0]=='.' ){
2437 psp->state = WAITING_FOR_DECL_OR_RULE;
2438 }else if( !isupper(x[0]) ){
2439 ErrorMsg(psp->filename, psp->tokenlineno,
2440 "%%wildcard argument \"%s\" should be a token", x);
2441 psp->errorcnt++;
2442 }else{
2443 struct symbol *sp = Symbol_new(x);
2444 if( psp->gp->wildcard==0 ){
2445 psp->gp->wildcard = sp;
2446 }else{
2447 ErrorMsg(psp->filename, psp->tokenlineno,
2448 "Extra wildcard to token: %s", x);
2449 psp->errorcnt++;
2450 }
2451 }
2452 break;
drh75897232000-05-29 14:26:00 +00002453 case RESYNC_AFTER_RULE_ERROR:
2454/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2455** break; */
2456 case RESYNC_AFTER_DECL_ERROR:
2457 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2458 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2459 break;
2460 }
2461}
2462
drh34ff57b2008-07-14 12:27:51 +00002463/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002464** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2465** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2466** comments them out. Text in between is also commented out as appropriate.
2467*/
danielk1977940fac92005-01-23 22:41:37 +00002468static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002469 int i, j, k, n;
2470 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002471 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002472 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002473 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002474 for(i=0; z[i]; i++){
2475 if( z[i]=='\n' ) lineno++;
2476 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2477 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2478 if( exclude ){
2479 exclude--;
2480 if( exclude==0 ){
2481 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2482 }
2483 }
2484 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2485 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2486 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2487 if( exclude ){
2488 exclude++;
2489 }else{
2490 for(j=i+7; isspace(z[j]); j++){}
2491 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2492 exclude = 1;
2493 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002494 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002495 exclude = 0;
2496 break;
2497 }
2498 }
2499 if( z[i+3]=='n' ) exclude = !exclude;
2500 if( exclude ){
2501 start = i;
2502 start_lineno = lineno;
2503 }
2504 }
2505 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2506 }
2507 }
2508 if( exclude ){
2509 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2510 exit(1);
2511 }
2512}
2513
drh75897232000-05-29 14:26:00 +00002514/* In spite of its name, this function is really a scanner. It read
2515** in the entire input file (all at once) then tokenizes it. Each
2516** token is passed to the function "parseonetoken" which builds all
2517** the appropriate data structures in the global state vector "gp".
2518*/
icculus9e44cf12010-02-14 17:14:22 +00002519void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002520{
2521 struct pstate ps;
2522 FILE *fp;
2523 char *filebuf;
2524 int filesize;
2525 int lineno;
2526 int c;
2527 char *cp, *nextcp;
2528 int startline = 0;
2529
rse38514a92007-09-20 11:34:17 +00002530 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002531 ps.gp = gp;
2532 ps.filename = gp->filename;
2533 ps.errorcnt = 0;
2534 ps.state = INITIALIZE;
2535
2536 /* Begin by reading the input file */
2537 fp = fopen(ps.filename,"rb");
2538 if( fp==0 ){
2539 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2540 gp->errorcnt++;
2541 return;
2542 }
2543 fseek(fp,0,2);
2544 filesize = ftell(fp);
2545 rewind(fp);
2546 filebuf = (char *)malloc( filesize+1 );
2547 if( filebuf==0 ){
2548 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2549 filesize+1);
2550 gp->errorcnt++;
2551 return;
2552 }
2553 if( fread(filebuf,1,filesize,fp)!=filesize ){
2554 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2555 filesize);
2556 free(filebuf);
2557 gp->errorcnt++;
2558 return;
2559 }
2560 fclose(fp);
2561 filebuf[filesize] = 0;
2562
drh6d08b4d2004-07-20 12:45:22 +00002563 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2564 preprocess_input(filebuf);
2565
drh75897232000-05-29 14:26:00 +00002566 /* Now scan the text of the input file */
2567 lineno = 1;
2568 for(cp=filebuf; (c= *cp)!=0; ){
2569 if( c=='\n' ) lineno++; /* Keep track of the line number */
2570 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2571 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2572 cp+=2;
2573 while( (c= *cp)!=0 && c!='\n' ) cp++;
2574 continue;
2575 }
2576 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2577 cp+=2;
2578 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2579 if( c=='\n' ) lineno++;
2580 cp++;
2581 }
2582 if( c ) cp++;
2583 continue;
2584 }
2585 ps.tokenstart = cp; /* Mark the beginning of the token */
2586 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2587 if( c=='\"' ){ /* String literals */
2588 cp++;
2589 while( (c= *cp)!=0 && c!='\"' ){
2590 if( c=='\n' ) lineno++;
2591 cp++;
2592 }
2593 if( c==0 ){
2594 ErrorMsg(ps.filename,startline,
2595"String starting on this line is not terminated before the end of the file.");
2596 ps.errorcnt++;
2597 nextcp = cp;
2598 }else{
2599 nextcp = cp+1;
2600 }
2601 }else if( c=='{' ){ /* A block of C code */
2602 int level;
2603 cp++;
2604 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2605 if( c=='\n' ) lineno++;
2606 else if( c=='{' ) level++;
2607 else if( c=='}' ) level--;
2608 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2609 int prevc;
2610 cp = &cp[2];
2611 prevc = 0;
2612 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2613 if( c=='\n' ) lineno++;
2614 prevc = c;
2615 cp++;
2616 }
2617 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
2618 cp = &cp[2];
2619 while( (c= *cp)!=0 && c!='\n' ) cp++;
2620 if( c ) lineno++;
2621 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
2622 int startchar, prevc;
2623 startchar = c;
2624 prevc = 0;
2625 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2626 if( c=='\n' ) lineno++;
2627 if( prevc=='\\' ) prevc = 0;
2628 else prevc = c;
2629 }
2630 }
2631 }
2632 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002633 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002634"C code starting on this line is not terminated before the end of the file.");
2635 ps.errorcnt++;
2636 nextcp = cp;
2637 }else{
2638 nextcp = cp+1;
2639 }
2640 }else if( isalnum(c) ){ /* Identifiers */
2641 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2642 nextcp = cp;
2643 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2644 cp += 3;
2645 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002646 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2647 cp += 2;
2648 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2649 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002650 }else{ /* All other (one character) operators */
2651 cp++;
2652 nextcp = cp;
2653 }
2654 c = *cp;
2655 *cp = 0; /* Null terminate the token */
2656 parseonetoken(&ps); /* Parse the token */
2657 *cp = c; /* Restore the buffer */
2658 cp = nextcp;
2659 }
2660 free(filebuf); /* Release the buffer after parsing */
2661 gp->rule = ps.firstrule;
2662 gp->errorcnt = ps.errorcnt;
2663}
2664/*************************** From the file "plink.c" *********************/
2665/*
2666** Routines processing configuration follow-set propagation links
2667** in the LEMON parser generator.
2668*/
2669static struct plink *plink_freelist = 0;
2670
2671/* Allocate a new plink */
2672struct plink *Plink_new(){
icculus9e44cf12010-02-14 17:14:22 +00002673 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00002674
2675 if( plink_freelist==0 ){
2676 int i;
2677 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002678 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002679 if( plink_freelist==0 ){
2680 fprintf(stderr,
2681 "Unable to allocate memory for a new follow-set propagation link.\n");
2682 exit(1);
2683 }
2684 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2685 plink_freelist[amt-1].next = 0;
2686 }
icculus9e44cf12010-02-14 17:14:22 +00002687 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00002688 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00002689 return newlink;
drh75897232000-05-29 14:26:00 +00002690}
2691
2692/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00002693void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002694{
icculus9e44cf12010-02-14 17:14:22 +00002695 struct plink *newlink;
2696 newlink = Plink_new();
2697 newlink->next = *plpp;
2698 *plpp = newlink;
2699 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00002700}
2701
2702/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00002703void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00002704{
2705 struct plink *nextpl;
2706 while( from ){
2707 nextpl = from->next;
2708 from->next = *to;
2709 *to = from;
2710 from = nextpl;
2711 }
2712}
2713
2714/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00002715void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00002716{
2717 struct plink *nextpl;
2718
2719 while( plp ){
2720 nextpl = plp->next;
2721 plp->next = plink_freelist;
2722 plink_freelist = plp;
2723 plp = nextpl;
2724 }
2725}
2726/*********************** From the file "report.c" **************************/
2727/*
2728** Procedures for generating reports and tables in the LEMON parser generator.
2729*/
2730
2731/* Generate a filename with the given suffix. Space to hold the
2732** name comes from malloc() and must be freed by the calling
2733** function.
2734*/
icculus9e44cf12010-02-14 17:14:22 +00002735PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00002736{
2737 char *name;
2738 char *cp;
2739
icculus9e44cf12010-02-14 17:14:22 +00002740 name = (char*)malloc( lemonStrlen(lemp->filename) + lemonStrlen(suffix) + 5 );
drh75897232000-05-29 14:26:00 +00002741 if( name==0 ){
2742 fprintf(stderr,"Can't allocate space for a filename.\n");
2743 exit(1);
2744 }
2745 strcpy(name,lemp->filename);
2746 cp = strrchr(name,'.');
2747 if( cp ) *cp = 0;
2748 strcat(name,suffix);
2749 return name;
2750}
2751
2752/* Open a file with a name based on the name of the input file,
2753** but with a different (specified) suffix, and return a pointer
2754** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00002755PRIVATE FILE *file_open(
2756 struct lemon *lemp,
2757 const char *suffix,
2758 const char *mode
2759){
drh75897232000-05-29 14:26:00 +00002760 FILE *fp;
2761
2762 if( lemp->outname ) free(lemp->outname);
2763 lemp->outname = file_makename(lemp, suffix);
2764 fp = fopen(lemp->outname,mode);
2765 if( fp==0 && *mode=='w' ){
2766 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2767 lemp->errorcnt++;
2768 return 0;
2769 }
icculusf5ad8242010-02-14 05:34:42 +00002770
2771 /* Add files we create to a list, so we can delete them if we fail. This
2772 ** is to keep makefiles from getting confused. We don't include .out files,
2773 ** though: this is debug information, and you don't want it deleted if there
2774 ** was an error you need to track down.
2775 */
2776 if(( *mode=='w' ) && (strcmp(suffix, ".out") != 0)){
2777 const char **ptr = (const char **)
2778 realloc(made_files, sizeof (const char **) * (made_files_count + 1));
2779 char *fname = strdup(lemp->outname);
2780 if ((ptr == NULL) || (fname == NULL)) {
2781 free(ptr);
2782 free(fname);
2783 memory_error();
2784 }
2785 made_files = ptr;
2786 made_files[made_files_count++] = fname;
2787 }
drh75897232000-05-29 14:26:00 +00002788 return fp;
2789}
2790
2791/* Duplicate the input file without comments and without actions
2792** on rules */
icculus9e44cf12010-02-14 17:14:22 +00002793void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002794{
2795 struct rule *rp;
2796 struct symbol *sp;
2797 int i, j, maxlen, len, ncolumns, skip;
2798 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2799 maxlen = 10;
2800 for(i=0; i<lemp->nsymbol; i++){
2801 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00002802 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00002803 if( len>maxlen ) maxlen = len;
2804 }
2805 ncolumns = 76/(maxlen+5);
2806 if( ncolumns<1 ) ncolumns = 1;
2807 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2808 for(i=0; i<skip; i++){
2809 printf("//");
2810 for(j=i; j<lemp->nsymbol; j+=skip){
2811 sp = lemp->symbols[j];
2812 assert( sp->index==j );
2813 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2814 }
2815 printf("\n");
2816 }
2817 for(rp=lemp->rule; rp; rp=rp->next){
2818 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002819 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002820 printf(" ::=");
2821 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002822 sp = rp->rhs[i];
2823 printf(" %s", sp->name);
2824 if( sp->type==MULTITERMINAL ){
2825 for(j=1; j<sp->nsubsym; j++){
2826 printf("|%s", sp->subsym[j]->name);
2827 }
2828 }
2829 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002830 }
2831 printf(".");
2832 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002833 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002834 printf("\n");
2835 }
2836}
2837
icculus9e44cf12010-02-14 17:14:22 +00002838void ConfigPrint(FILE *fp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002839{
2840 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002841 struct symbol *sp;
2842 int i, j;
drh75897232000-05-29 14:26:00 +00002843 rp = cfp->rp;
2844 fprintf(fp,"%s ::=",rp->lhs->name);
2845 for(i=0; i<=rp->nrhs; i++){
2846 if( i==cfp->dot ) fprintf(fp," *");
2847 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002848 sp = rp->rhs[i];
2849 fprintf(fp," %s", sp->name);
2850 if( sp->type==MULTITERMINAL ){
2851 for(j=1; j<sp->nsubsym; j++){
2852 fprintf(fp,"|%s",sp->subsym[j]->name);
2853 }
2854 }
drh75897232000-05-29 14:26:00 +00002855 }
2856}
2857
2858/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002859#if 0
drh75897232000-05-29 14:26:00 +00002860/* Print a set */
2861PRIVATE void SetPrint(out,set,lemp)
2862FILE *out;
2863char *set;
2864struct lemon *lemp;
2865{
2866 int i;
2867 char *spacer;
2868 spacer = "";
2869 fprintf(out,"%12s[","");
2870 for(i=0; i<lemp->nterminal; i++){
2871 if( SetFind(set,i) ){
2872 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2873 spacer = " ";
2874 }
2875 }
2876 fprintf(out,"]\n");
2877}
2878
2879/* Print a plink chain */
2880PRIVATE void PlinkPrint(out,plp,tag)
2881FILE *out;
2882struct plink *plp;
2883char *tag;
2884{
2885 while( plp ){
drhada354d2005-11-05 15:03:59 +00002886 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002887 ConfigPrint(out,plp->cfp);
2888 fprintf(out,"\n");
2889 plp = plp->next;
2890 }
2891}
2892#endif
2893
2894/* Print an action to the given file descriptor. Return FALSE if
2895** nothing was actually printed.
2896*/
2897int PrintAction(struct action *ap, FILE *fp, int indent){
2898 int result = 1;
2899 switch( ap->type ){
2900 case SHIFT:
drhada354d2005-11-05 15:03:59 +00002901 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00002902 break;
2903 case REDUCE:
2904 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
2905 break;
2906 case ACCEPT:
2907 fprintf(fp,"%*s accept",indent,ap->sp->name);
2908 break;
2909 case ERROR:
2910 fprintf(fp,"%*s error",indent,ap->sp->name);
2911 break;
drh9892c5d2007-12-21 00:02:11 +00002912 case SRCONFLICT:
2913 case RRCONFLICT:
drh75897232000-05-29 14:26:00 +00002914 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
2915 indent,ap->sp->name,ap->x.rp->index);
2916 break;
drh9892c5d2007-12-21 00:02:11 +00002917 case SSCONFLICT:
2918 fprintf(fp,"%*s shift %d ** Parsing conflict **",
2919 indent,ap->sp->name,ap->x.stp->statenum);
2920 break;
drh75897232000-05-29 14:26:00 +00002921 case SH_RESOLVED:
2922 case RD_RESOLVED:
2923 case NOT_USED:
2924 result = 0;
2925 break;
2926 }
2927 return result;
2928}
2929
2930/* Generate the "y.output" log file */
icculus9e44cf12010-02-14 17:14:22 +00002931void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002932{
2933 int i;
2934 struct state *stp;
2935 struct config *cfp;
2936 struct action *ap;
2937 FILE *fp;
2938
drh2aa6ca42004-09-10 00:14:04 +00002939 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00002940 if( fp==0 ) return;
drh75897232000-05-29 14:26:00 +00002941 for(i=0; i<lemp->nstate; i++){
2942 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00002943 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00002944 if( lemp->basisflag ) cfp=stp->bp;
2945 else cfp=stp->cfp;
2946 while( cfp ){
2947 char buf[20];
2948 if( cfp->dot==cfp->rp->nrhs ){
2949 sprintf(buf,"(%d)",cfp->rp->index);
2950 fprintf(fp," %5s ",buf);
2951 }else{
2952 fprintf(fp," ");
2953 }
2954 ConfigPrint(fp,cfp);
2955 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00002956#if 0
drh75897232000-05-29 14:26:00 +00002957 SetPrint(fp,cfp->fws,lemp);
2958 PlinkPrint(fp,cfp->fplp,"To ");
2959 PlinkPrint(fp,cfp->bplp,"From");
2960#endif
2961 if( lemp->basisflag ) cfp=cfp->bp;
2962 else cfp=cfp->next;
2963 }
2964 fprintf(fp,"\n");
2965 for(ap=stp->ap; ap; ap=ap->next){
2966 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
2967 }
2968 fprintf(fp,"\n");
2969 }
drhe9278182007-07-18 18:16:29 +00002970 fprintf(fp, "----------------------------------------------------\n");
2971 fprintf(fp, "Symbols:\n");
2972 for(i=0; i<lemp->nsymbol; i++){
2973 int j;
2974 struct symbol *sp;
2975
2976 sp = lemp->symbols[i];
2977 fprintf(fp, " %3d: %s", i, sp->name);
2978 if( sp->type==NONTERMINAL ){
2979 fprintf(fp, ":");
2980 if( sp->lambda ){
2981 fprintf(fp, " <lambda>");
2982 }
2983 for(j=0; j<lemp->nterminal; j++){
2984 if( sp->firstset && SetFind(sp->firstset, j) ){
2985 fprintf(fp, " %s", lemp->symbols[j]->name);
2986 }
2987 }
2988 }
2989 fprintf(fp, "\n");
2990 }
drh75897232000-05-29 14:26:00 +00002991 fclose(fp);
2992 return;
2993}
2994
2995/* Search for the file "name" which is in the same directory as
2996** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00002997PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00002998{
icculus9e44cf12010-02-14 17:14:22 +00002999 const char *pathlist;
3000 char *pathbufptr;
3001 char *pathbuf;
drh75897232000-05-29 14:26:00 +00003002 char *path,*cp;
3003 char c;
drh75897232000-05-29 14:26:00 +00003004
3005#ifdef __WIN32__
3006 cp = strrchr(argv0,'\\');
3007#else
3008 cp = strrchr(argv0,'/');
3009#endif
3010 if( cp ){
3011 c = *cp;
3012 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003013 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh75897232000-05-29 14:26:00 +00003014 if( path ) sprintf(path,"%s/%s",argv0,name);
3015 *cp = c;
3016 }else{
drh75897232000-05-29 14:26:00 +00003017 pathlist = getenv("PATH");
3018 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003019 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003020 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003021 if( (pathbuf != 0) && (path!=0) ){
3022 pathbufptr = pathbuf;
3023 strcpy(pathbuf, pathlist);
3024 while( *pathbuf ){
3025 cp = strchr(pathbuf,':');
3026 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003027 c = *cp;
3028 *cp = 0;
icculus9e44cf12010-02-14 17:14:22 +00003029 sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003030 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003031 if( c==0 ) pathbuf[0] = 0;
3032 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003033 if( access(path,modemask)==0 ) break;
3034 }
icculus9e44cf12010-02-14 17:14:22 +00003035 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003036 }
3037 }
3038 return path;
3039}
3040
3041/* Given an action, compute the integer value for that action
3042** which is to be put in the action table of the generated machine.
3043** Return negative if no action should be generated.
3044*/
icculus9e44cf12010-02-14 17:14:22 +00003045PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003046{
3047 int act;
3048 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00003049 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00003050 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
3051 case ERROR: act = lemp->nstate + lemp->nrule; break;
3052 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
3053 default: act = -1; break;
3054 }
3055 return act;
3056}
3057
3058#define LINESIZE 1000
3059/* The next cluster of routines are for reading the template file
3060** and writing the results to the generated parser */
3061/* The first function transfers data from "in" to "out" until
3062** a line is seen which begins with "%%". The line number is
3063** tracked.
3064**
3065** if name!=0, then any word that begin with "Parse" is changed to
3066** begin with *name instead.
3067*/
icculus9e44cf12010-02-14 17:14:22 +00003068PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003069{
3070 int i, iStart;
3071 char line[LINESIZE];
3072 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3073 (*lineno)++;
3074 iStart = 0;
3075 if( name ){
3076 for(i=0; line[i]; i++){
3077 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
3078 && (i==0 || !isalpha(line[i-1]))
3079 ){
3080 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3081 fprintf(out,"%s",name);
3082 i += 4;
3083 iStart = i+1;
3084 }
3085 }
3086 }
3087 fprintf(out,"%s",&line[iStart]);
3088 }
3089}
3090
3091/* The next function finds the template file and opens it, returning
3092** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003093PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003094{
3095 static char templatename[] = "lempar.c";
3096 char buf[1000];
3097 FILE *in;
3098 char *tpltname;
3099 char *cp;
3100
icculus3e143bd2010-02-14 00:48:49 +00003101 /* first, see if user specified a template filename on the command line. */
3102 if (user_templatename != 0) {
3103 if( access(user_templatename,004)==-1 ){
3104 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3105 user_templatename);
3106 lemp->errorcnt++;
3107 return 0;
3108 }
3109 in = fopen(user_templatename,"rb");
3110 if( in==0 ){
3111 fprintf(stderr,"Can't open the template file \"%s\".\n",user_templatename);
3112 lemp->errorcnt++;
3113 return 0;
3114 }
3115 return in;
3116 }
3117
drh75897232000-05-29 14:26:00 +00003118 cp = strrchr(lemp->filename,'.');
3119 if( cp ){
drh8b582012003-10-21 13:16:03 +00003120 sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003121 }else{
3122 sprintf(buf,"%s.lt",lemp->filename);
3123 }
3124 if( access(buf,004)==0 ){
3125 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003126 }else if( access(templatename,004)==0 ){
3127 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003128 }else{
3129 tpltname = pathsearch(lemp->argv0,templatename,0);
3130 }
3131 if( tpltname==0 ){
3132 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3133 templatename);
3134 lemp->errorcnt++;
3135 return 0;
3136 }
drh2aa6ca42004-09-10 00:14:04 +00003137 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003138 if( in==0 ){
3139 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3140 lemp->errorcnt++;
3141 return 0;
3142 }
3143 return in;
3144}
3145
drhaf805ca2004-09-07 11:28:25 +00003146/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003147PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003148{
3149 fprintf(out,"#line %d \"",lineno);
3150 while( *filename ){
3151 if( *filename == '\\' ) putc('\\',out);
3152 putc(*filename,out);
3153 filename++;
3154 }
3155 fprintf(out,"\"\n");
3156}
3157
drh75897232000-05-29 14:26:00 +00003158/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003159PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003160{
3161 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003162 while( *str ){
drh75897232000-05-29 14:26:00 +00003163 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003164 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003165 str++;
3166 }
drh9db55df2004-09-09 14:01:21 +00003167 if( str[-1]!='\n' ){
3168 putc('\n',out);
3169 (*lineno)++;
3170 }
shane58543932008-12-10 20:10:04 +00003171 if (!lemp->nolinenosflag) {
3172 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3173 }
drh75897232000-05-29 14:26:00 +00003174 return;
3175}
3176
3177/*
3178** The following routine emits code for the destructor for the
3179** symbol sp
3180*/
icculus9e44cf12010-02-14 17:14:22 +00003181void emit_destructor_code(
3182 FILE *out,
3183 struct symbol *sp,
3184 struct lemon *lemp,
3185 int *lineno
3186){
drhcc83b6e2004-04-23 23:38:42 +00003187 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003188
drh75897232000-05-29 14:26:00 +00003189 if( sp->type==TERMINAL ){
3190 cp = lemp->tokendest;
3191 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003192 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003193 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003194 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003195 fprintf(out,"{\n"); (*lineno)++;
shane58543932008-12-10 20:10:04 +00003196 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,sp->destLineno,lemp->filename); }
drh960e8c62001-04-03 16:53:21 +00003197 }else if( lemp->vardest ){
3198 cp = lemp->vardest;
3199 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003200 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003201 }else{
3202 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003203 }
3204 for(; *cp; cp++){
3205 if( *cp=='$' && cp[1]=='$' ){
3206 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3207 cp++;
3208 continue;
3209 }
shane58543932008-12-10 20:10:04 +00003210 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003211 fputc(*cp,out);
3212 }
shane58543932008-12-10 20:10:04 +00003213 fprintf(out,"\n"); (*lineno)++;
3214 if (!lemp->nolinenosflag) {
3215 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3216 }
3217 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003218 return;
3219}
3220
3221/*
drh960e8c62001-04-03 16:53:21 +00003222** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003223*/
icculus9e44cf12010-02-14 17:14:22 +00003224int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003225{
3226 int ret;
3227 if( sp->type==TERMINAL ){
3228 ret = lemp->tokendest!=0;
3229 }else{
drh960e8c62001-04-03 16:53:21 +00003230 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003231 }
3232 return ret;
3233}
3234
drh0bb132b2004-07-20 14:06:51 +00003235/*
3236** Append text to a dynamically allocated string. If zText is 0 then
3237** reset the string to be empty again. Always return the complete text
3238** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003239**
3240** n bytes of zText are stored. If n==0 then all of zText up to the first
3241** \000 terminator is stored. zText can contain up to two instances of
3242** %d. The values of p1 and p2 are written into the first and second
3243** %d.
3244**
3245** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003246*/
icculus9e44cf12010-02-14 17:14:22 +00003247PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3248 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003249 static char *z = 0;
3250 static int alloced = 0;
3251 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003252 int c;
drh0bb132b2004-07-20 14:06:51 +00003253 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003254 if( zText==0 ){
3255 used = 0;
3256 return z;
3257 }
drh7ac25c72004-08-19 15:12:26 +00003258 if( n<=0 ){
3259 if( n<0 ){
3260 used += n;
3261 assert( used>=0 );
3262 }
drh87cf1372008-08-13 20:09:06 +00003263 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003264 }
drh0bb132b2004-07-20 14:06:51 +00003265 if( n+sizeof(zInt)*2+used >= alloced ){
3266 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003267 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003268 }
icculus9e44cf12010-02-14 17:14:22 +00003269 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003270 while( n-- > 0 ){
3271 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003272 if( c=='%' && n>0 && zText[0]=='d' ){
drh0bb132b2004-07-20 14:06:51 +00003273 sprintf(zInt, "%d", p1);
3274 p1 = p2;
3275 strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003276 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003277 zText++;
3278 n--;
3279 }else{
3280 z[used++] = c;
3281 }
3282 }
3283 z[used] = 0;
3284 return z;
3285}
3286
3287/*
3288** zCode is a string that is the action associated with a rule. Expand
3289** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003290** stack.
drh0bb132b2004-07-20 14:06:51 +00003291*/
drhaf805ca2004-09-07 11:28:25 +00003292PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003293 char *cp, *xp;
3294 int i;
3295 char lhsused = 0; /* True if the LHS element has been used */
3296 char used[MAXRHS]; /* True for each RHS element which is used */
3297
3298 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3299 lhsused = 0;
3300
drh19c9e562007-03-29 20:13:53 +00003301 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003302 static char newlinestr[2] = { '\n', '\0' };
3303 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003304 rp->line = rp->ruleline;
3305 }
3306
drh0bb132b2004-07-20 14:06:51 +00003307 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003308
3309 /* This const cast is wrong but harmless, if we're careful. */
3310 for(cp=(char *)rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003311 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3312 char saved;
3313 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3314 saved = *xp;
3315 *xp = 0;
3316 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003317 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003318 cp = xp;
3319 lhsused = 1;
3320 }else{
3321 for(i=0; i<rp->nrhs; i++){
3322 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003323 if( cp!=rp->code && cp[-1]=='@' ){
3324 /* If the argument is of the form @X then substituted
3325 ** the token number of X, not the value of X */
3326 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3327 }else{
drhfd405312005-11-06 04:06:59 +00003328 struct symbol *sp = rp->rhs[i];
3329 int dtnum;
3330 if( sp->type==MULTITERMINAL ){
3331 dtnum = sp->subsym[0]->dtnum;
3332 }else{
3333 dtnum = sp->dtnum;
3334 }
3335 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003336 }
drh0bb132b2004-07-20 14:06:51 +00003337 cp = xp;
3338 used[i] = 1;
3339 break;
3340 }
3341 }
3342 }
3343 *xp = saved;
3344 }
3345 append_str(cp, 1, 0, 0);
3346 } /* End loop */
3347
3348 /* Check to make sure the LHS has been used */
3349 if( rp->lhsalias && !lhsused ){
3350 ErrorMsg(lemp->filename,rp->ruleline,
3351 "Label \"%s\" for \"%s(%s)\" is never used.",
3352 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3353 lemp->errorcnt++;
3354 }
3355
3356 /* Generate destructor code for RHS symbols which are not used in the
3357 ** reduce code */
3358 for(i=0; i<rp->nrhs; i++){
3359 if( rp->rhsalias[i] && !used[i] ){
3360 ErrorMsg(lemp->filename,rp->ruleline,
3361 "Label %s for \"%s(%s)\" is never used.",
3362 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3363 lemp->errorcnt++;
3364 }else if( rp->rhsalias[i]==0 ){
3365 if( has_destructor(rp->rhs[i],lemp) ){
drh639efd02008-08-13 20:04:29 +00003366 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003367 rp->rhs[i]->index,i-rp->nrhs+1);
3368 }else{
3369 /* No destructor defined for this term */
3370 }
3371 }
3372 }
drh61e339a2007-01-16 03:09:02 +00003373 if( rp->code ){
3374 cp = append_str(0,0,0,0);
3375 rp->code = Strsafe(cp?cp:"");
3376 }
drh0bb132b2004-07-20 14:06:51 +00003377}
3378
drh75897232000-05-29 14:26:00 +00003379/*
3380** Generate code which executes when the rule "rp" is reduced. Write
3381** the code to "out". Make sure lineno stays up-to-date.
3382*/
icculus9e44cf12010-02-14 17:14:22 +00003383PRIVATE void emit_code(
3384 FILE *out,
3385 struct rule *rp,
3386 struct lemon *lemp,
3387 int *lineno
3388){
3389 const char *cp;
drh75897232000-05-29 14:26:00 +00003390
3391 /* Generate code to do the reduce action */
3392 if( rp->code ){
shane58543932008-12-10 20:10:04 +00003393 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,rp->line,lemp->filename); }
drhaf805ca2004-09-07 11:28:25 +00003394 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003395 for(cp=rp->code; *cp; cp++){
shane58543932008-12-10 20:10:04 +00003396 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003397 } /* End loop */
shane58543932008-12-10 20:10:04 +00003398 fprintf(out,"}\n"); (*lineno)++;
3399 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,*lineno,lemp->outname); }
drh75897232000-05-29 14:26:00 +00003400 } /* End if( rp->code ) */
3401
drh75897232000-05-29 14:26:00 +00003402 return;
3403}
3404
3405/*
3406** Print the definition of the union used for the parser's data stack.
3407** This union contains fields for every possible data type for tokens
3408** and nonterminals. In the process of computing and printing this
3409** union, also set the ".dtnum" field of every terminal and nonterminal
3410** symbol.
3411*/
icculus9e44cf12010-02-14 17:14:22 +00003412void print_stack_union(
3413 FILE *out, /* The output stream */
3414 struct lemon *lemp, /* The main info structure for this parser */
3415 int *plineno, /* Pointer to the line number */
3416 int mhflag /* True if generating makeheaders output */
3417){
drh75897232000-05-29 14:26:00 +00003418 int lineno = *plineno; /* The line number of the output */
3419 char **types; /* A hash table of datatypes */
3420 int arraysize; /* Size of the "types" array */
3421 int maxdtlength; /* Maximum length of any ".datatype" field. */
3422 char *stddt; /* Standardized name for a datatype */
3423 int i,j; /* Loop counters */
3424 int hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00003425 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00003426
3427 /* Allocate and initialize types[] and allocate stddt[] */
3428 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003429 types = (char**)calloc( arraysize, sizeof(char*) );
drh75897232000-05-29 14:26:00 +00003430 for(i=0; i<arraysize; i++) types[i] = 0;
3431 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003432 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003433 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003434 }
drh75897232000-05-29 14:26:00 +00003435 for(i=0; i<lemp->nsymbol; i++){
3436 int len;
3437 struct symbol *sp = lemp->symbols[i];
3438 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003439 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003440 if( len>maxdtlength ) maxdtlength = len;
3441 }
3442 stddt = (char*)malloc( maxdtlength*2 + 1 );
3443 if( types==0 || stddt==0 ){
3444 fprintf(stderr,"Out of memory.\n");
3445 exit(1);
3446 }
3447
3448 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3449 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003450 ** used for terminal symbols. If there is no %default_type defined then
3451 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3452 ** a datatype using the %type directive.
3453 */
drh75897232000-05-29 14:26:00 +00003454 for(i=0; i<lemp->nsymbol; i++){
3455 struct symbol *sp = lemp->symbols[i];
3456 char *cp;
3457 if( sp==lemp->errsym ){
3458 sp->dtnum = arraysize+1;
3459 continue;
3460 }
drh960e8c62001-04-03 16:53:21 +00003461 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003462 sp->dtnum = 0;
3463 continue;
3464 }
3465 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003466 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003467 j = 0;
3468 while( isspace(*cp) ) cp++;
3469 while( *cp ) stddt[j++] = *cp++;
3470 while( j>0 && isspace(stddt[j-1]) ) j--;
3471 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00003472 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00003473 sp->dtnum = 0;
3474 continue;
3475 }
drh75897232000-05-29 14:26:00 +00003476 hash = 0;
3477 for(j=0; stddt[j]; j++){
3478 hash = hash*53 + stddt[j];
3479 }
drh3b2129c2003-05-13 00:34:21 +00003480 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003481 while( types[hash] ){
3482 if( strcmp(types[hash],stddt)==0 ){
3483 sp->dtnum = hash + 1;
3484 break;
3485 }
3486 hash++;
3487 if( hash>=arraysize ) hash = 0;
3488 }
3489 if( types[hash]==0 ){
3490 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00003491 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00003492 if( types[hash]==0 ){
3493 fprintf(stderr,"Out of memory.\n");
3494 exit(1);
3495 }
3496 strcpy(types[hash],stddt);
3497 }
3498 }
3499
3500 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3501 name = lemp->name ? lemp->name : "Parse";
3502 lineno = *plineno;
3503 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3504 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3505 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3506 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3507 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00003508 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003509 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3510 for(i=0; i<arraysize; i++){
3511 if( types[i]==0 ) continue;
3512 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3513 free(types[i]);
3514 }
drhc4dd3fd2008-01-22 01:48:05 +00003515 if( lemp->errsym->useCnt ){
3516 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3517 }
drh75897232000-05-29 14:26:00 +00003518 free(stddt);
3519 free(types);
3520 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3521 *plineno = lineno;
3522}
3523
drhb29b0a52002-02-23 19:39:46 +00003524/*
3525** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003526** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003527*/
drh8b582012003-10-21 13:16:03 +00003528static const char *minimum_size_type(int lwr, int upr){
3529 if( lwr>=0 ){
3530 if( upr<=255 ){
3531 return "unsigned char";
3532 }else if( upr<65535 ){
3533 return "unsigned short int";
3534 }else{
3535 return "unsigned int";
3536 }
3537 }else if( lwr>=-127 && upr<=127 ){
3538 return "signed char";
3539 }else if( lwr>=-32767 && upr<32767 ){
3540 return "short";
drhb29b0a52002-02-23 19:39:46 +00003541 }else{
drh8b582012003-10-21 13:16:03 +00003542 return "int";
drhb29b0a52002-02-23 19:39:46 +00003543 }
3544}
3545
drhfdbf9282003-10-21 16:34:41 +00003546/*
3547** Each state contains a set of token transaction and a set of
3548** nonterminal transactions. Each of these sets makes an instance
3549** of the following structure. An array of these structures is used
3550** to order the creation of entries in the yy_action[] table.
3551*/
3552struct axset {
3553 struct state *stp; /* A pointer to a state */
3554 int isTkn; /* True to use tokens. False for non-terminals */
3555 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00003556 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00003557};
3558
3559/*
3560** Compare to axset structures for sorting purposes
3561*/
3562static int axset_compare(const void *a, const void *b){
3563 struct axset *p1 = (struct axset*)a;
3564 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00003565 int c;
3566 c = p2->nAction - p1->nAction;
3567 if( c==0 ){
3568 c = p2->iOrder - p1->iOrder;
3569 }
3570 assert( c!=0 || p1==p2 );
3571 return c;
drhfdbf9282003-10-21 16:34:41 +00003572}
3573
drhc4dd3fd2008-01-22 01:48:05 +00003574/*
3575** Write text on "out" that describes the rule "rp".
3576*/
3577static void writeRuleText(FILE *out, struct rule *rp){
3578 int j;
3579 fprintf(out,"%s ::=", rp->lhs->name);
3580 for(j=0; j<rp->nrhs; j++){
3581 struct symbol *sp = rp->rhs[j];
3582 fprintf(out," %s", sp->name);
3583 if( sp->type==MULTITERMINAL ){
3584 int k;
3585 for(k=1; k<sp->nsubsym; k++){
3586 fprintf(out,"|%s",sp->subsym[k]->name);
3587 }
3588 }
3589 }
3590}
3591
3592
drh75897232000-05-29 14:26:00 +00003593/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00003594void ReportTable(
3595 struct lemon *lemp,
3596 int mhflag /* Output in makeheaders format if true */
3597){
drh75897232000-05-29 14:26:00 +00003598 FILE *out, *in;
3599 char line[LINESIZE];
3600 int lineno;
3601 struct state *stp;
3602 struct action *ap;
3603 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003604 struct acttab *pActtab;
drhf16371d2009-11-03 19:18:31 +00003605 int i, j, k, n;
icculus9e44cf12010-02-14 17:14:22 +00003606 const char *name;
drh8b582012003-10-21 13:16:03 +00003607 int mnTknOfst, mxTknOfst;
3608 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003609 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003610
3611 in = tplt_open(lemp);
3612 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003613 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003614 if( out==0 ){
3615 fclose(in);
3616 return;
3617 }
3618 lineno = 1;
3619 tplt_xfer(lemp->name,in,out,&lineno);
3620
3621 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00003622 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00003623 if( mhflag ){
3624 char *name = file_makename(lemp, ".h");
3625 fprintf(out,"#include \"%s\"\n", name); lineno++;
3626 free(name);
3627 }
3628 tplt_xfer(lemp->name,in,out,&lineno);
3629
3630 /* Generate #defines for all tokens */
3631 if( mhflag ){
icculus9e44cf12010-02-14 17:14:22 +00003632 const char *prefix;
drh75897232000-05-29 14:26:00 +00003633 fprintf(out,"#if INTERFACE\n"); lineno++;
3634 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3635 else prefix = "";
3636 for(i=1; i<lemp->nterminal; i++){
3637 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3638 lineno++;
3639 }
3640 fprintf(out,"#endif\n"); lineno++;
3641 }
3642 tplt_xfer(lemp->name,in,out,&lineno);
3643
3644 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003645 fprintf(out,"#define YYCODETYPE %s\n",
drh8a415d32009-06-12 13:53:51 +00003646 minimum_size_type(0, lemp->nsymbol+1)); lineno++;
drh75897232000-05-29 14:26:00 +00003647 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3648 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003649 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003650 if( lemp->wildcard ){
3651 fprintf(out,"#define YYWILDCARD %d\n",
3652 lemp->wildcard->index); lineno++;
3653 }
drh75897232000-05-29 14:26:00 +00003654 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003655 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003656 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003657 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3658 }else{
3659 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3660 }
drhca44b5a2007-02-22 23:06:58 +00003661 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003662 if( mhflag ){
3663 fprintf(out,"#if INTERFACE\n"); lineno++;
3664 }
3665 name = lemp->name ? lemp->name : "Parse";
3666 if( lemp->arg && lemp->arg[0] ){
3667 int i;
drh87cf1372008-08-13 20:09:06 +00003668 i = lemonStrlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003669 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3670 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003671 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3672 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3673 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3674 name,lemp->arg,&lemp->arg[i]); lineno++;
3675 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3676 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003677 }else{
drh1f245e42002-03-11 13:55:50 +00003678 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3679 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3680 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3681 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003682 }
3683 if( mhflag ){
3684 fprintf(out,"#endif\n"); lineno++;
3685 }
3686 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3687 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00003688 if( lemp->errsym->useCnt ){
3689 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3690 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
3691 }
drh0bd1f4e2002-06-06 18:54:39 +00003692 if( lemp->has_fallback ){
3693 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3694 }
drh75897232000-05-29 14:26:00 +00003695 tplt_xfer(lemp->name,in,out,&lineno);
3696
drh8b582012003-10-21 13:16:03 +00003697 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003698 **
drh8b582012003-10-21 13:16:03 +00003699 ** yy_action[] A single table containing all actions.
3700 ** yy_lookahead[] A table containing the lookahead for each entry in
3701 ** yy_action. Used to detect hash collisions.
3702 ** yy_shift_ofst[] For each state, the offset into yy_action for
3703 ** shifting terminals.
3704 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3705 ** shifting non-terminals after a reduce.
3706 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003707 */
drh75897232000-05-29 14:26:00 +00003708
drh8b582012003-10-21 13:16:03 +00003709 /* Compute the actions on all states and count them up */
icculus9e44cf12010-02-14 17:14:22 +00003710 ax = (struct axset *) calloc(lemp->nstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00003711 if( ax==0 ){
3712 fprintf(stderr,"malloc failed\n");
3713 exit(1);
3714 }
drh75897232000-05-29 14:26:00 +00003715 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003716 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003717 ax[i*2].stp = stp;
3718 ax[i*2].isTkn = 1;
3719 ax[i*2].nAction = stp->nTknAct;
3720 ax[i*2+1].stp = stp;
3721 ax[i*2+1].isTkn = 0;
3722 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003723 }
drh8b582012003-10-21 13:16:03 +00003724 mxTknOfst = mnTknOfst = 0;
3725 mxNtOfst = mnNtOfst = 0;
3726
drhfdbf9282003-10-21 16:34:41 +00003727 /* Compute the action table. In order to try to keep the size of the
3728 ** action table to a minimum, the heuristic of placing the largest action
3729 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003730 */
drhe594bc32009-11-03 13:02:25 +00003731 for(i=0; i<lemp->nstate*2; i++) ax[i].iOrder = i;
drhfdbf9282003-10-21 16:34:41 +00003732 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003733 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003734 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3735 stp = ax[i].stp;
3736 if( ax[i].isTkn ){
3737 for(ap=stp->ap; ap; ap=ap->next){
3738 int action;
3739 if( ap->sp->index>=lemp->nterminal ) continue;
3740 action = compute_action(lemp, ap);
3741 if( action<0 ) continue;
3742 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003743 }
drhfdbf9282003-10-21 16:34:41 +00003744 stp->iTknOfst = acttab_insert(pActtab);
3745 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3746 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3747 }else{
3748 for(ap=stp->ap; ap; ap=ap->next){
3749 int action;
3750 if( ap->sp->index<lemp->nterminal ) continue;
3751 if( ap->sp->index==lemp->nsymbol ) continue;
3752 action = compute_action(lemp, ap);
3753 if( action<0 ) continue;
3754 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003755 }
drhfdbf9282003-10-21 16:34:41 +00003756 stp->iNtOfst = acttab_insert(pActtab);
3757 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3758 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003759 }
3760 }
drhfdbf9282003-10-21 16:34:41 +00003761 free(ax);
drh8b582012003-10-21 13:16:03 +00003762
3763 /* Output the yy_action table */
drh8b582012003-10-21 13:16:03 +00003764 n = acttab_size(pActtab);
drhf16371d2009-11-03 19:18:31 +00003765 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
3766 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003767 for(i=j=0; i<n; i++){
3768 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003769 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003770 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003771 fprintf(out, " %4d,", action);
3772 if( j==9 || i==n-1 ){
3773 fprintf(out, "\n"); lineno++;
3774 j = 0;
3775 }else{
3776 j++;
3777 }
3778 }
3779 fprintf(out, "};\n"); lineno++;
3780
3781 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003782 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003783 for(i=j=0; i<n; i++){
3784 int la = acttab_yylookahead(pActtab, i);
3785 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003786 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003787 fprintf(out, " %4d,", la);
3788 if( j==9 || i==n-1 ){
3789 fprintf(out, "\n"); lineno++;
3790 j = 0;
3791 }else{
3792 j++;
3793 }
3794 }
3795 fprintf(out, "};\n"); lineno++;
3796
3797 /* Output the yy_shift_ofst[] table */
3798 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003799 n = lemp->nstate;
3800 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003801 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
3802 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
3803 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003804 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003805 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003806 for(i=j=0; i<n; i++){
3807 int ofst;
3808 stp = lemp->sorted[i];
3809 ofst = stp->iTknOfst;
3810 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003811 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003812 fprintf(out, " %4d,", ofst);
3813 if( j==9 || i==n-1 ){
3814 fprintf(out, "\n"); lineno++;
3815 j = 0;
3816 }else{
3817 j++;
3818 }
3819 }
3820 fprintf(out, "};\n"); lineno++;
3821
3822 /* Output the yy_reduce_ofst[] table */
3823 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003824 n = lemp->nstate;
3825 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003826 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
3827 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
3828 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003829 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003830 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003831 for(i=j=0; i<n; i++){
3832 int ofst;
3833 stp = lemp->sorted[i];
3834 ofst = stp->iNtOfst;
3835 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003836 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003837 fprintf(out, " %4d,", ofst);
3838 if( j==9 || i==n-1 ){
3839 fprintf(out, "\n"); lineno++;
3840 j = 0;
3841 }else{
3842 j++;
3843 }
3844 }
3845 fprintf(out, "};\n"); lineno++;
3846
3847 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003848 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003849 n = lemp->nstate;
3850 for(i=j=0; i<n; i++){
3851 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003852 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003853 fprintf(out, " %4d,", stp->iDflt);
3854 if( j==9 || i==n-1 ){
3855 fprintf(out, "\n"); lineno++;
3856 j = 0;
3857 }else{
3858 j++;
3859 }
3860 }
3861 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003862 tplt_xfer(lemp->name,in,out,&lineno);
3863
drh0bd1f4e2002-06-06 18:54:39 +00003864 /* Generate the table of fallback tokens.
3865 */
3866 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00003867 int mx = lemp->nterminal - 1;
3868 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
3869 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00003870 struct symbol *p = lemp->symbols[i];
3871 if( p->fallback==0 ){
3872 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3873 }else{
3874 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3875 p->name, p->fallback->name);
3876 }
3877 lineno++;
3878 }
3879 }
3880 tplt_xfer(lemp->name, in, out, &lineno);
3881
3882 /* Generate a table containing the symbolic name of every symbol
3883 */
drh75897232000-05-29 14:26:00 +00003884 for(i=0; i<lemp->nsymbol; i++){
3885 sprintf(line,"\"%s\",",lemp->symbols[i]->name);
3886 fprintf(out," %-15s",line);
3887 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
3888 }
3889 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
3890 tplt_xfer(lemp->name,in,out,&lineno);
3891
drh0bd1f4e2002-06-06 18:54:39 +00003892 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00003893 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00003894 ** when tracing REDUCE actions.
3895 */
3896 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
3897 assert( rp->index==i );
drhc4dd3fd2008-01-22 01:48:05 +00003898 fprintf(out," /* %3d */ \"", i);
3899 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00003900 fprintf(out,"\",\n"); lineno++;
3901 }
3902 tplt_xfer(lemp->name,in,out,&lineno);
3903
drh75897232000-05-29 14:26:00 +00003904 /* Generate code which executes every time a symbol is popped from
3905 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00003906 ** (In other words, generate the %destructor actions)
3907 */
drh75897232000-05-29 14:26:00 +00003908 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00003909 int once = 1;
drh75897232000-05-29 14:26:00 +00003910 for(i=0; i<lemp->nsymbol; i++){
3911 struct symbol *sp = lemp->symbols[i];
3912 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003913 if( once ){
3914 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
3915 once = 0;
3916 }
drhc53eed12009-06-12 17:46:19 +00003917 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00003918 }
3919 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
3920 if( i<lemp->nsymbol ){
3921 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3922 fprintf(out," break;\n"); lineno++;
3923 }
3924 }
drh8d659732005-01-13 23:54:06 +00003925 if( lemp->vardest ){
3926 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00003927 int once = 1;
drh8d659732005-01-13 23:54:06 +00003928 for(i=0; i<lemp->nsymbol; i++){
3929 struct symbol *sp = lemp->symbols[i];
3930 if( sp==0 || sp->type==TERMINAL ||
3931 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003932 if( once ){
3933 fprintf(out, " /* Default NON-TERMINAL Destructor */\n"); lineno++;
3934 once = 0;
3935 }
drhc53eed12009-06-12 17:46:19 +00003936 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00003937 dflt_sp = sp;
3938 }
3939 if( dflt_sp!=0 ){
3940 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00003941 }
drh4dc8ef52008-07-01 17:13:57 +00003942 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00003943 }
drh75897232000-05-29 14:26:00 +00003944 for(i=0; i<lemp->nsymbol; i++){
3945 struct symbol *sp = lemp->symbols[i];
3946 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh75013012009-06-12 15:47:34 +00003947 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003948
3949 /* Combine duplicate destructors into a single case */
3950 for(j=i+1; j<lemp->nsymbol; j++){
3951 struct symbol *sp2 = lemp->symbols[j];
3952 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
3953 && sp2->dtnum==sp->dtnum
3954 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00003955 fprintf(out," case %d: /* %s */\n",
3956 sp2->index, sp2->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003957 sp2->destructor = 0;
3958 }
3959 }
3960
drh75897232000-05-29 14:26:00 +00003961 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3962 fprintf(out," break;\n"); lineno++;
3963 }
drh75897232000-05-29 14:26:00 +00003964 tplt_xfer(lemp->name,in,out,&lineno);
3965
3966 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00003967 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00003968 tplt_xfer(lemp->name,in,out,&lineno);
3969
3970 /* Generate the table of rule information
3971 **
3972 ** Note: This code depends on the fact that rules are number
3973 ** sequentually beginning with 0.
3974 */
3975 for(rp=lemp->rule; rp; rp=rp->next){
3976 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
3977 }
3978 tplt_xfer(lemp->name,in,out,&lineno);
3979
3980 /* Generate code which execution during each REDUCE action */
3981 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00003982 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00003983 }
drhc53eed12009-06-12 17:46:19 +00003984 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00003985 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00003986 struct rule *rp2; /* Other rules with the same action */
drh0bb132b2004-07-20 14:06:51 +00003987 if( rp->code==0 ) continue;
drhc53eed12009-06-12 17:46:19 +00003988 if( rp->code[0]=='\n' && rp->code[1]==0 ) continue; /* Will be default: */
drhc4dd3fd2008-01-22 01:48:05 +00003989 fprintf(out," case %d: /* ", rp->index);
3990 writeRuleText(out, rp);
3991 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003992 for(rp2=rp->next; rp2; rp2=rp2->next){
3993 if( rp2->code==rp->code ){
drhc4dd3fd2008-01-22 01:48:05 +00003994 fprintf(out," case %d: /* ", rp2->index);
3995 writeRuleText(out, rp2);
drh8a415d32009-06-12 13:53:51 +00003996 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003997 rp2->code = 0;
3998 }
3999 }
drh75897232000-05-29 14:26:00 +00004000 emit_code(out,rp,lemp,&lineno);
4001 fprintf(out," break;\n"); lineno++;
drhc53eed12009-06-12 17:46:19 +00004002 rp->code = 0;
drh75897232000-05-29 14:26:00 +00004003 }
drhc53eed12009-06-12 17:46:19 +00004004 /* Finally, output the default: rule. We choose as the default: all
4005 ** empty actions. */
4006 fprintf(out," default:\n"); lineno++;
4007 for(rp=lemp->rule; rp; rp=rp->next){
4008 if( rp->code==0 ) continue;
4009 assert( rp->code[0]=='\n' && rp->code[1]==0 );
4010 fprintf(out," /* (%d) ", rp->index);
4011 writeRuleText(out, rp);
4012 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->index); lineno++;
4013 }
4014 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004015 tplt_xfer(lemp->name,in,out,&lineno);
4016
4017 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004018 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004019 tplt_xfer(lemp->name,in,out,&lineno);
4020
4021 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004022 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004023 tplt_xfer(lemp->name,in,out,&lineno);
4024
4025 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004026 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004027 tplt_xfer(lemp->name,in,out,&lineno);
4028
4029 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004030 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004031
4032 fclose(in);
4033 fclose(out);
4034 return;
4035}
4036
4037/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004038void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004039{
4040 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004041 const char *prefix;
drh75897232000-05-29 14:26:00 +00004042 char line[LINESIZE];
4043 char pattern[LINESIZE];
4044 int i;
4045
4046 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4047 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004048 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004049 if( in ){
4050 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
4051 sprintf(pattern,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4052 if( strcmp(line,pattern) ) break;
4053 }
4054 fclose(in);
4055 if( i==lemp->nterminal ){
4056 /* No change in the file. Don't rewrite it. */
4057 return;
4058 }
4059 }
drh2aa6ca42004-09-10 00:14:04 +00004060 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004061 if( out ){
4062 for(i=1; i<lemp->nterminal; i++){
4063 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4064 }
4065 fclose(out);
4066 }
4067 return;
4068}
4069
4070/* Reduce the size of the action tables, if possible, by making use
4071** of defaults.
4072**
drhb59499c2002-02-23 18:45:13 +00004073** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004074** it the default. Except, there is no default if the wildcard token
4075** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004076*/
icculus9e44cf12010-02-14 17:14:22 +00004077void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004078{
4079 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00004080 struct action *ap, *ap2;
4081 struct rule *rp, *rp2, *rbest;
4082 int nbest, n;
drh75897232000-05-29 14:26:00 +00004083 int i;
drhe09daa92006-06-10 13:29:31 +00004084 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004085
4086 for(i=0; i<lemp->nstate; i++){
4087 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004088 nbest = 0;
4089 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004090 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004091
drhb59499c2002-02-23 18:45:13 +00004092 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004093 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4094 usesWildcard = 1;
4095 }
drhb59499c2002-02-23 18:45:13 +00004096 if( ap->type!=REDUCE ) continue;
4097 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004098 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004099 if( rp==rbest ) continue;
4100 n = 1;
4101 for(ap2=ap->next; ap2; ap2=ap2->next){
4102 if( ap2->type!=REDUCE ) continue;
4103 rp2 = ap2->x.rp;
4104 if( rp2==rbest ) continue;
4105 if( rp2==rp ) n++;
4106 }
4107 if( n>nbest ){
4108 nbest = n;
4109 rbest = rp;
drh75897232000-05-29 14:26:00 +00004110 }
4111 }
drhb59499c2002-02-23 18:45:13 +00004112
4113 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004114 ** is not at least 1 or if the wildcard token is a possible
4115 ** lookahead.
4116 */
4117 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004118
drhb59499c2002-02-23 18:45:13 +00004119
4120 /* Combine matching REDUCE actions into a single default */
4121 for(ap=stp->ap; ap; ap=ap->next){
4122 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4123 }
drh75897232000-05-29 14:26:00 +00004124 assert( ap );
4125 ap->sp = Symbol_new("{default}");
4126 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004127 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004128 }
4129 stp->ap = Action_sort(stp->ap);
4130 }
4131}
drhb59499c2002-02-23 18:45:13 +00004132
drhada354d2005-11-05 15:03:59 +00004133
4134/*
4135** Compare two states for sorting purposes. The smaller state is the
4136** one with the most non-terminal actions. If they have the same number
4137** of non-terminal actions, then the smaller is the one with the most
4138** token actions.
4139*/
4140static int stateResortCompare(const void *a, const void *b){
4141 const struct state *pA = *(const struct state**)a;
4142 const struct state *pB = *(const struct state**)b;
4143 int n;
4144
4145 n = pB->nNtAct - pA->nNtAct;
4146 if( n==0 ){
4147 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004148 if( n==0 ){
4149 n = pB->statenum - pA->statenum;
4150 }
drhada354d2005-11-05 15:03:59 +00004151 }
drhe594bc32009-11-03 13:02:25 +00004152 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004153 return n;
4154}
4155
4156
4157/*
4158** Renumber and resort states so that states with fewer choices
4159** occur at the end. Except, keep state 0 as the first state.
4160*/
icculus9e44cf12010-02-14 17:14:22 +00004161void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00004162{
4163 int i;
4164 struct state *stp;
4165 struct action *ap;
4166
4167 for(i=0; i<lemp->nstate; i++){
4168 stp = lemp->sorted[i];
4169 stp->nTknAct = stp->nNtAct = 0;
4170 stp->iDflt = lemp->nstate + lemp->nrule;
4171 stp->iTknOfst = NO_OFFSET;
4172 stp->iNtOfst = NO_OFFSET;
4173 for(ap=stp->ap; ap; ap=ap->next){
4174 if( compute_action(lemp,ap)>=0 ){
4175 if( ap->sp->index<lemp->nterminal ){
4176 stp->nTknAct++;
4177 }else if( ap->sp->index<lemp->nsymbol ){
4178 stp->nNtAct++;
4179 }else{
4180 stp->iDflt = compute_action(lemp, ap);
4181 }
4182 }
4183 }
4184 }
4185 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4186 stateResortCompare);
4187 for(i=0; i<lemp->nstate; i++){
4188 lemp->sorted[i]->statenum = i;
4189 }
4190}
4191
4192
drh75897232000-05-29 14:26:00 +00004193/***************** From the file "set.c" ************************************/
4194/*
4195** Set manipulation routines for the LEMON parser generator.
4196*/
4197
4198static int size = 0;
4199
4200/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00004201void SetSize(int n)
drh75897232000-05-29 14:26:00 +00004202{
4203 size = n+1;
4204}
4205
4206/* Allocate a new set */
4207char *SetNew(){
4208 char *s;
drh9892c5d2007-12-21 00:02:11 +00004209 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004210 if( s==0 ){
4211 extern void memory_error();
4212 memory_error();
4213 }
drh75897232000-05-29 14:26:00 +00004214 return s;
4215}
4216
4217/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00004218void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00004219{
4220 free(s);
4221}
4222
4223/* Add a new element to the set. Return TRUE if the element was added
4224** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00004225int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00004226{
4227 int rv;
drh9892c5d2007-12-21 00:02:11 +00004228 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004229 rv = s[e];
4230 s[e] = 1;
4231 return !rv;
4232}
4233
4234/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00004235int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00004236{
4237 int i, progress;
4238 progress = 0;
4239 for(i=0; i<size; i++){
4240 if( s2[i]==0 ) continue;
4241 if( s1[i]==0 ){
4242 progress = 1;
4243 s1[i] = 1;
4244 }
4245 }
4246 return progress;
4247}
4248/********************** From the file "table.c" ****************************/
4249/*
4250** All code in this file has been automatically generated
4251** from a specification in the file
4252** "table.q"
4253** by the associative array code building program "aagen".
4254** Do not edit this file! Instead, edit the specification
4255** file, then rerun aagen.
4256*/
4257/*
4258** Code for processing tables in the LEMON parser generator.
4259*/
4260
icculus9e44cf12010-02-14 17:14:22 +00004261PRIVATE int strhash(const char *x)
drh75897232000-05-29 14:26:00 +00004262{
4263 int h = 0;
4264 while( *x) h = h*13 + *(x++);
4265 return h;
4266}
4267
4268/* Works like strdup, sort of. Save a string in malloced memory, but
4269** keep strings in a table so that the same string is not in more
4270** than one place.
4271*/
icculus9e44cf12010-02-14 17:14:22 +00004272const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00004273{
icculus9e44cf12010-02-14 17:14:22 +00004274 const char *z;
4275 char *cpy;
drh75897232000-05-29 14:26:00 +00004276
drh916f75f2006-07-17 00:19:39 +00004277 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004278 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00004279 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
4280 strcpy(cpy,y);
4281 z = cpy;
drh75897232000-05-29 14:26:00 +00004282 Strsafe_insert(z);
4283 }
4284 MemoryCheck(z);
4285 return z;
4286}
4287
4288/* There is one instance of the following structure for each
4289** associative array of type "x1".
4290*/
4291struct s_x1 {
4292 int size; /* The number of available slots. */
4293 /* Must be a power of 2 greater than or */
4294 /* equal to 1 */
4295 int count; /* Number of currently slots filled */
4296 struct s_x1node *tbl; /* The data stored here */
4297 struct s_x1node **ht; /* Hash table for lookups */
4298};
4299
4300/* There is one instance of this structure for every data element
4301** in an associative array of type "x1".
4302*/
4303typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00004304 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00004305 struct s_x1node *next; /* Next entry with the same hash */
4306 struct s_x1node **from; /* Previous link */
4307} x1node;
4308
4309/* There is only one instance of the array, which is the following */
4310static struct s_x1 *x1a;
4311
4312/* Allocate a new associative array */
4313void Strsafe_init(){
4314 if( x1a ) return;
4315 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4316 if( x1a ){
4317 x1a->size = 1024;
4318 x1a->count = 0;
4319 x1a->tbl = (x1node*)malloc(
4320 (sizeof(x1node) + sizeof(x1node*))*1024 );
4321 if( x1a->tbl==0 ){
4322 free(x1a);
4323 x1a = 0;
4324 }else{
4325 int i;
4326 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4327 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4328 }
4329 }
4330}
4331/* Insert a new record into the array. Return TRUE if successful.
4332** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004333int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00004334{
4335 x1node *np;
4336 int h;
4337 int ph;
4338
4339 if( x1a==0 ) return 0;
4340 ph = strhash(data);
4341 h = ph & (x1a->size-1);
4342 np = x1a->ht[h];
4343 while( np ){
4344 if( strcmp(np->data,data)==0 ){
4345 /* An existing entry with the same key is found. */
4346 /* Fail because overwrite is not allows. */
4347 return 0;
4348 }
4349 np = np->next;
4350 }
4351 if( x1a->count>=x1a->size ){
4352 /* Need to make the hash table bigger */
4353 int i,size;
4354 struct s_x1 array;
4355 array.size = size = x1a->size*2;
4356 array.count = x1a->count;
4357 array.tbl = (x1node*)malloc(
4358 (sizeof(x1node) + sizeof(x1node*))*size );
4359 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4360 array.ht = (x1node**)&(array.tbl[size]);
4361 for(i=0; i<size; i++) array.ht[i] = 0;
4362 for(i=0; i<x1a->count; i++){
4363 x1node *oldnp, *newnp;
4364 oldnp = &(x1a->tbl[i]);
4365 h = strhash(oldnp->data) & (size-1);
4366 newnp = &(array.tbl[i]);
4367 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4368 newnp->next = array.ht[h];
4369 newnp->data = oldnp->data;
4370 newnp->from = &(array.ht[h]);
4371 array.ht[h] = newnp;
4372 }
4373 free(x1a->tbl);
4374 *x1a = array;
4375 }
4376 /* Insert the new data */
4377 h = ph & (x1a->size-1);
4378 np = &(x1a->tbl[x1a->count++]);
4379 np->data = data;
4380 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4381 np->next = x1a->ht[h];
4382 x1a->ht[h] = np;
4383 np->from = &(x1a->ht[h]);
4384 return 1;
4385}
4386
4387/* Return a pointer to data assigned to the given key. Return NULL
4388** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004389const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00004390{
4391 int h;
4392 x1node *np;
4393
4394 if( x1a==0 ) return 0;
4395 h = strhash(key) & (x1a->size-1);
4396 np = x1a->ht[h];
4397 while( np ){
4398 if( strcmp(np->data,key)==0 ) break;
4399 np = np->next;
4400 }
4401 return np ? np->data : 0;
4402}
4403
4404/* Return a pointer to the (terminal or nonterminal) symbol "x".
4405** Create a new symbol if this is the first time "x" has been seen.
4406*/
icculus9e44cf12010-02-14 17:14:22 +00004407struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00004408{
4409 struct symbol *sp;
4410
4411 sp = Symbol_find(x);
4412 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00004413 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00004414 MemoryCheck(sp);
4415 sp->name = Strsafe(x);
4416 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4417 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004418 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004419 sp->prec = -1;
4420 sp->assoc = UNK;
4421 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00004422 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00004423 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00004424 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00004425 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00004426 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00004427 Symbol_insert(sp,sp->name);
4428 }
drhc4dd3fd2008-01-22 01:48:05 +00004429 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00004430 return sp;
4431}
4432
drh60d31652004-02-22 00:08:04 +00004433/* Compare two symbols for working purposes
4434**
4435** Symbols that begin with upper case letters (terminals or tokens)
4436** must sort before symbols that begin with lower case letters
4437** (non-terminals). Other than that, the order does not matter.
4438**
4439** We find experimentally that leaving the symbols in their original
4440** order (the order they appeared in the grammar file) gives the
4441** smallest parser tables in SQLite.
4442*/
icculus9e44cf12010-02-14 17:14:22 +00004443int Symbolcmpp(const void *_a, const void *_b)
4444{
4445 const struct symbol **a = (const struct symbol **) _a;
4446 const struct symbol **b = (const struct symbol **) _b;
drh60d31652004-02-22 00:08:04 +00004447 int i1 = (**a).index + 10000000*((**a).name[0]>'Z');
4448 int i2 = (**b).index + 10000000*((**b).name[0]>'Z');
drhe594bc32009-11-03 13:02:25 +00004449 assert( i1!=i2 || strcmp((**a).name,(**b).name)==0 );
drh60d31652004-02-22 00:08:04 +00004450 return i1-i2;
drh75897232000-05-29 14:26:00 +00004451}
4452
4453/* There is one instance of the following structure for each
4454** associative array of type "x2".
4455*/
4456struct s_x2 {
4457 int size; /* The number of available slots. */
4458 /* Must be a power of 2 greater than or */
4459 /* equal to 1 */
4460 int count; /* Number of currently slots filled */
4461 struct s_x2node *tbl; /* The data stored here */
4462 struct s_x2node **ht; /* Hash table for lookups */
4463};
4464
4465/* There is one instance of this structure for every data element
4466** in an associative array of type "x2".
4467*/
4468typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00004469 struct symbol *data; /* The data */
4470 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00004471 struct s_x2node *next; /* Next entry with the same hash */
4472 struct s_x2node **from; /* Previous link */
4473} x2node;
4474
4475/* There is only one instance of the array, which is the following */
4476static struct s_x2 *x2a;
4477
4478/* Allocate a new associative array */
4479void Symbol_init(){
4480 if( x2a ) return;
4481 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4482 if( x2a ){
4483 x2a->size = 128;
4484 x2a->count = 0;
4485 x2a->tbl = (x2node*)malloc(
4486 (sizeof(x2node) + sizeof(x2node*))*128 );
4487 if( x2a->tbl==0 ){
4488 free(x2a);
4489 x2a = 0;
4490 }else{
4491 int i;
4492 x2a->ht = (x2node**)&(x2a->tbl[128]);
4493 for(i=0; i<128; i++) x2a->ht[i] = 0;
4494 }
4495 }
4496}
4497/* Insert a new record into the array. Return TRUE if successful.
4498** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004499int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00004500{
4501 x2node *np;
4502 int h;
4503 int ph;
4504
4505 if( x2a==0 ) return 0;
4506 ph = strhash(key);
4507 h = ph & (x2a->size-1);
4508 np = x2a->ht[h];
4509 while( np ){
4510 if( strcmp(np->key,key)==0 ){
4511 /* An existing entry with the same key is found. */
4512 /* Fail because overwrite is not allows. */
4513 return 0;
4514 }
4515 np = np->next;
4516 }
4517 if( x2a->count>=x2a->size ){
4518 /* Need to make the hash table bigger */
4519 int i,size;
4520 struct s_x2 array;
4521 array.size = size = x2a->size*2;
4522 array.count = x2a->count;
4523 array.tbl = (x2node*)malloc(
4524 (sizeof(x2node) + sizeof(x2node*))*size );
4525 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4526 array.ht = (x2node**)&(array.tbl[size]);
4527 for(i=0; i<size; i++) array.ht[i] = 0;
4528 for(i=0; i<x2a->count; i++){
4529 x2node *oldnp, *newnp;
4530 oldnp = &(x2a->tbl[i]);
4531 h = strhash(oldnp->key) & (size-1);
4532 newnp = &(array.tbl[i]);
4533 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4534 newnp->next = array.ht[h];
4535 newnp->key = oldnp->key;
4536 newnp->data = oldnp->data;
4537 newnp->from = &(array.ht[h]);
4538 array.ht[h] = newnp;
4539 }
4540 free(x2a->tbl);
4541 *x2a = array;
4542 }
4543 /* Insert the new data */
4544 h = ph & (x2a->size-1);
4545 np = &(x2a->tbl[x2a->count++]);
4546 np->key = key;
4547 np->data = data;
4548 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4549 np->next = x2a->ht[h];
4550 x2a->ht[h] = np;
4551 np->from = &(x2a->ht[h]);
4552 return 1;
4553}
4554
4555/* Return a pointer to data assigned to the given key. Return NULL
4556** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004557struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00004558{
4559 int h;
4560 x2node *np;
4561
4562 if( x2a==0 ) return 0;
4563 h = strhash(key) & (x2a->size-1);
4564 np = x2a->ht[h];
4565 while( np ){
4566 if( strcmp(np->key,key)==0 ) break;
4567 np = np->next;
4568 }
4569 return np ? np->data : 0;
4570}
4571
4572/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00004573struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00004574{
4575 struct symbol *data;
4576 if( x2a && n>0 && n<=x2a->count ){
4577 data = x2a->tbl[n-1].data;
4578 }else{
4579 data = 0;
4580 }
4581 return data;
4582}
4583
4584/* Return the size of the array */
4585int Symbol_count()
4586{
4587 return x2a ? x2a->count : 0;
4588}
4589
4590/* Return an array of pointers to all data in the table.
4591** The array is obtained from malloc. Return NULL if memory allocation
4592** problems, or if the array is empty. */
4593struct symbol **Symbol_arrayof()
4594{
4595 struct symbol **array;
4596 int i,size;
4597 if( x2a==0 ) return 0;
4598 size = x2a->count;
drh9892c5d2007-12-21 00:02:11 +00004599 array = (struct symbol **)calloc(size, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00004600 if( array ){
4601 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4602 }
4603 return array;
4604}
4605
4606/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00004607int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00004608{
icculus9e44cf12010-02-14 17:14:22 +00004609 const struct config *a = (struct config *) _a;
4610 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00004611 int x;
4612 x = a->rp->index - b->rp->index;
4613 if( x==0 ) x = a->dot - b->dot;
4614 return x;
4615}
4616
4617/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00004618PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00004619{
4620 int rc;
4621 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4622 rc = a->rp->index - b->rp->index;
4623 if( rc==0 ) rc = a->dot - b->dot;
4624 }
4625 if( rc==0 ){
4626 if( a ) rc = 1;
4627 if( b ) rc = -1;
4628 }
4629 return rc;
4630}
4631
4632/* Hash a state */
icculus9e44cf12010-02-14 17:14:22 +00004633PRIVATE int statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00004634{
4635 int h=0;
4636 while( a ){
4637 h = h*571 + a->rp->index*37 + a->dot;
4638 a = a->bp;
4639 }
4640 return h;
4641}
4642
4643/* Allocate a new state structure */
4644struct state *State_new()
4645{
icculus9e44cf12010-02-14 17:14:22 +00004646 struct state *newstate;
4647 newstate = (struct state *)calloc(1, sizeof(struct state) );
4648 MemoryCheck(newstate);
4649 return newstate;
drh75897232000-05-29 14:26:00 +00004650}
4651
4652/* There is one instance of the following structure for each
4653** associative array of type "x3".
4654*/
4655struct s_x3 {
4656 int size; /* The number of available slots. */
4657 /* Must be a power of 2 greater than or */
4658 /* equal to 1 */
4659 int count; /* Number of currently slots filled */
4660 struct s_x3node *tbl; /* The data stored here */
4661 struct s_x3node **ht; /* Hash table for lookups */
4662};
4663
4664/* There is one instance of this structure for every data element
4665** in an associative array of type "x3".
4666*/
4667typedef struct s_x3node {
4668 struct state *data; /* The data */
4669 struct config *key; /* The key */
4670 struct s_x3node *next; /* Next entry with the same hash */
4671 struct s_x3node **from; /* Previous link */
4672} x3node;
4673
4674/* There is only one instance of the array, which is the following */
4675static struct s_x3 *x3a;
4676
4677/* Allocate a new associative array */
4678void State_init(){
4679 if( x3a ) return;
4680 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4681 if( x3a ){
4682 x3a->size = 128;
4683 x3a->count = 0;
4684 x3a->tbl = (x3node*)malloc(
4685 (sizeof(x3node) + sizeof(x3node*))*128 );
4686 if( x3a->tbl==0 ){
4687 free(x3a);
4688 x3a = 0;
4689 }else{
4690 int i;
4691 x3a->ht = (x3node**)&(x3a->tbl[128]);
4692 for(i=0; i<128; i++) x3a->ht[i] = 0;
4693 }
4694 }
4695}
4696/* Insert a new record into the array. Return TRUE if successful.
4697** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004698int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00004699{
4700 x3node *np;
4701 int h;
4702 int ph;
4703
4704 if( x3a==0 ) return 0;
4705 ph = statehash(key);
4706 h = ph & (x3a->size-1);
4707 np = x3a->ht[h];
4708 while( np ){
4709 if( statecmp(np->key,key)==0 ){
4710 /* An existing entry with the same key is found. */
4711 /* Fail because overwrite is not allows. */
4712 return 0;
4713 }
4714 np = np->next;
4715 }
4716 if( x3a->count>=x3a->size ){
4717 /* Need to make the hash table bigger */
4718 int i,size;
4719 struct s_x3 array;
4720 array.size = size = x3a->size*2;
4721 array.count = x3a->count;
4722 array.tbl = (x3node*)malloc(
4723 (sizeof(x3node) + sizeof(x3node*))*size );
4724 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4725 array.ht = (x3node**)&(array.tbl[size]);
4726 for(i=0; i<size; i++) array.ht[i] = 0;
4727 for(i=0; i<x3a->count; i++){
4728 x3node *oldnp, *newnp;
4729 oldnp = &(x3a->tbl[i]);
4730 h = statehash(oldnp->key) & (size-1);
4731 newnp = &(array.tbl[i]);
4732 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4733 newnp->next = array.ht[h];
4734 newnp->key = oldnp->key;
4735 newnp->data = oldnp->data;
4736 newnp->from = &(array.ht[h]);
4737 array.ht[h] = newnp;
4738 }
4739 free(x3a->tbl);
4740 *x3a = array;
4741 }
4742 /* Insert the new data */
4743 h = ph & (x3a->size-1);
4744 np = &(x3a->tbl[x3a->count++]);
4745 np->key = key;
4746 np->data = data;
4747 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4748 np->next = x3a->ht[h];
4749 x3a->ht[h] = np;
4750 np->from = &(x3a->ht[h]);
4751 return 1;
4752}
4753
4754/* Return a pointer to data assigned to the given key. Return NULL
4755** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004756struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004757{
4758 int h;
4759 x3node *np;
4760
4761 if( x3a==0 ) return 0;
4762 h = statehash(key) & (x3a->size-1);
4763 np = x3a->ht[h];
4764 while( np ){
4765 if( statecmp(np->key,key)==0 ) break;
4766 np = np->next;
4767 }
4768 return np ? np->data : 0;
4769}
4770
4771/* Return an array of pointers to all data in the table.
4772** The array is obtained from malloc. Return NULL if memory allocation
4773** problems, or if the array is empty. */
4774struct state **State_arrayof()
4775{
4776 struct state **array;
4777 int i,size;
4778 if( x3a==0 ) return 0;
4779 size = x3a->count;
4780 array = (struct state **)malloc( sizeof(struct state *)*size );
4781 if( array ){
4782 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4783 }
4784 return array;
4785}
4786
4787/* Hash a configuration */
icculus9e44cf12010-02-14 17:14:22 +00004788PRIVATE int confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00004789{
4790 int h=0;
4791 h = h*571 + a->rp->index*37 + a->dot;
4792 return h;
4793}
4794
4795/* There is one instance of the following structure for each
4796** associative array of type "x4".
4797*/
4798struct s_x4 {
4799 int size; /* The number of available slots. */
4800 /* Must be a power of 2 greater than or */
4801 /* equal to 1 */
4802 int count; /* Number of currently slots filled */
4803 struct s_x4node *tbl; /* The data stored here */
4804 struct s_x4node **ht; /* Hash table for lookups */
4805};
4806
4807/* There is one instance of this structure for every data element
4808** in an associative array of type "x4".
4809*/
4810typedef struct s_x4node {
4811 struct config *data; /* The data */
4812 struct s_x4node *next; /* Next entry with the same hash */
4813 struct s_x4node **from; /* Previous link */
4814} x4node;
4815
4816/* There is only one instance of the array, which is the following */
4817static struct s_x4 *x4a;
4818
4819/* Allocate a new associative array */
4820void Configtable_init(){
4821 if( x4a ) return;
4822 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4823 if( x4a ){
4824 x4a->size = 64;
4825 x4a->count = 0;
4826 x4a->tbl = (x4node*)malloc(
4827 (sizeof(x4node) + sizeof(x4node*))*64 );
4828 if( x4a->tbl==0 ){
4829 free(x4a);
4830 x4a = 0;
4831 }else{
4832 int i;
4833 x4a->ht = (x4node**)&(x4a->tbl[64]);
4834 for(i=0; i<64; i++) x4a->ht[i] = 0;
4835 }
4836 }
4837}
4838/* Insert a new record into the array. Return TRUE if successful.
4839** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004840int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00004841{
4842 x4node *np;
4843 int h;
4844 int ph;
4845
4846 if( x4a==0 ) return 0;
4847 ph = confighash(data);
4848 h = ph & (x4a->size-1);
4849 np = x4a->ht[h];
4850 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004851 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00004852 /* An existing entry with the same key is found. */
4853 /* Fail because overwrite is not allows. */
4854 return 0;
4855 }
4856 np = np->next;
4857 }
4858 if( x4a->count>=x4a->size ){
4859 /* Need to make the hash table bigger */
4860 int i,size;
4861 struct s_x4 array;
4862 array.size = size = x4a->size*2;
4863 array.count = x4a->count;
4864 array.tbl = (x4node*)malloc(
4865 (sizeof(x4node) + sizeof(x4node*))*size );
4866 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4867 array.ht = (x4node**)&(array.tbl[size]);
4868 for(i=0; i<size; i++) array.ht[i] = 0;
4869 for(i=0; i<x4a->count; i++){
4870 x4node *oldnp, *newnp;
4871 oldnp = &(x4a->tbl[i]);
4872 h = confighash(oldnp->data) & (size-1);
4873 newnp = &(array.tbl[i]);
4874 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4875 newnp->next = array.ht[h];
4876 newnp->data = oldnp->data;
4877 newnp->from = &(array.ht[h]);
4878 array.ht[h] = newnp;
4879 }
4880 free(x4a->tbl);
4881 *x4a = array;
4882 }
4883 /* Insert the new data */
4884 h = ph & (x4a->size-1);
4885 np = &(x4a->tbl[x4a->count++]);
4886 np->data = data;
4887 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
4888 np->next = x4a->ht[h];
4889 x4a->ht[h] = np;
4890 np->from = &(x4a->ht[h]);
4891 return 1;
4892}
4893
4894/* Return a pointer to data assigned to the given key. Return NULL
4895** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004896struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004897{
4898 int h;
4899 x4node *np;
4900
4901 if( x4a==0 ) return 0;
4902 h = confighash(key) & (x4a->size-1);
4903 np = x4a->ht[h];
4904 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004905 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00004906 np = np->next;
4907 }
4908 return np ? np->data : 0;
4909}
4910
4911/* Remove all data from the table. Pass each data to the function "f"
4912** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00004913void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00004914{
4915 int i;
4916 if( x4a==0 || x4a->count==0 ) return;
4917 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
4918 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
4919 x4a->count = 0;
4920 return;
4921}