blob: 25787abf8db1861c1bcbd328ceac27ee47a30752 [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)
drhf2f105d2012-08-20 15:53:54 +000018# define __WIN32__
drh75897232000-05-29 14:26:00 +000019# endif
20#endif
21
rse8f304482007-07-30 18:31:53 +000022#ifdef __WIN32__
drhdf609712010-11-23 20:55:27 +000023#ifdef __cplusplus
24extern "C" {
25#endif
26extern int access(const char *path, int mode);
27#ifdef __cplusplus
28}
29#endif
rse8f304482007-07-30 18:31:53 +000030#else
31#include <unistd.h>
32#endif
33
drh75897232000-05-29 14:26:00 +000034/* #define PRIVATE static */
35#define PRIVATE
36
37#ifdef TEST
38#define MAXRHS 5 /* Set low to exercise exception code */
39#else
40#define MAXRHS 1000
41#endif
42
drhf5c4e0f2010-07-18 11:35:53 +000043static int showPrecedenceConflict = 0;
drhe9278182007-07-18 18:16:29 +000044static char *msort(char*,char**,int(*)(const char*,const char*));
drh75897232000-05-29 14:26:00 +000045
drh87cf1372008-08-13 20:09:06 +000046/*
47** Compilers are getting increasingly pedantic about type conversions
48** as C evolves ever closer to Ada.... To work around the latest problems
49** we have to define the following variant of strlen().
50*/
51#define lemonStrlen(X) ((int)strlen(X))
52
drh898799f2014-01-10 23:21:00 +000053/*
54** Compilers are starting to complain about the use of sprintf() and strcpy(),
55** saying they are unsafe. So we define our own versions of those routines too.
56**
57** There are three routines here: lemon_sprintf(), lemon_vsprintf(), and
58** lemon_addtext(). The first two are replacements for sprintf() and vsprintf().
59** The third is a helper routine for vsnprintf() that adds texts to the end of a
60** buffer, making sure the buffer is always zero-terminated.
61**
62** The string formatter is a minimal subset of stdlib sprintf() supporting only
63** a few simply conversions:
64**
65** %d
66** %s
67** %.*s
68**
69*/
70static void lemon_addtext(
71 char *zBuf, /* The buffer to which text is added */
72 int *pnUsed, /* Slots of the buffer used so far */
73 const char *zIn, /* Text to add */
drh61f92cd2014-01-11 03:06:18 +000074 int nIn, /* Bytes of text to add. -1 to use strlen() */
75 int iWidth /* Field width. Negative to left justify */
drh898799f2014-01-10 23:21:00 +000076){
77 if( nIn<0 ) for(nIn=0; zIn[nIn]; nIn++){}
drh61f92cd2014-01-11 03:06:18 +000078 while( iWidth>nIn ){ zBuf[*(pnUsed++)] = ' '; iWidth--; }
drh898799f2014-01-10 23:21:00 +000079 if( nIn==0 ) return;
80 memcpy(&zBuf[*pnUsed], zIn, nIn);
81 *pnUsed += nIn;
drh61f92cd2014-01-11 03:06:18 +000082 while( (-iWidth)>nIn ){ zBuf[*(pnUsed++)] = ' '; iWidth++; }
drh898799f2014-01-10 23:21:00 +000083 zBuf[*pnUsed] = 0;
84}
85static int lemon_vsprintf(char *str, const char *zFormat, va_list ap){
86 int i, j, k, c, size;
87 int nUsed = 0;
88 const char *z;
89 char zTemp[50];
90 str[0] = 0;
91 for(i=j=0; (c = zFormat[i])!=0; i++){
92 if( c=='%' ){
drh61f92cd2014-01-11 03:06:18 +000093 int iWidth = 0;
94 lemon_addtext(str, &nUsed, &zFormat[j], i-j, 0);
drh898799f2014-01-10 23:21:00 +000095 c = zFormat[++i];
drh61f92cd2014-01-11 03:06:18 +000096 if( isdigit(c) || (c=='-' && isdigit(zFormat[i+1])) ){
97 if( c=='-' ) i++;
98 while( isdigit(zFormat[i]) ) iWidth = iWidth*10 + zFormat[i++] - '0';
99 if( c=='-' ) iWidth = -iWidth;
100 c = zFormat[i];
101 }
drh898799f2014-01-10 23:21:00 +0000102 if( c=='d' ){
103 int v = va_arg(ap, int);
104 if( v<0 ){
drh61f92cd2014-01-11 03:06:18 +0000105 lemon_addtext(str, &nUsed, "-", 1, iWidth);
drh898799f2014-01-10 23:21:00 +0000106 v = -v;
107 }else if( v==0 ){
drh61f92cd2014-01-11 03:06:18 +0000108 lemon_addtext(str, &nUsed, "0", 1, iWidth);
drh898799f2014-01-10 23:21:00 +0000109 }
110 k = 0;
111 while( v>0 ){
112 k++;
113 zTemp[sizeof(zTemp)-k] = (v%10) + '0';
114 v /= 10;
115 }
drh61f92cd2014-01-11 03:06:18 +0000116 lemon_addtext(str, &nUsed, &zTemp[sizeof(zTemp)-k], k, iWidth);
drh898799f2014-01-10 23:21:00 +0000117 }else if( c=='s' ){
118 z = va_arg(ap, const char*);
drh61f92cd2014-01-11 03:06:18 +0000119 lemon_addtext(str, &nUsed, z, -1, iWidth);
drh898799f2014-01-10 23:21:00 +0000120 }else if( c=='.' && memcmp(&zFormat[i], ".*s", 3)==0 ){
121 i += 2;
122 k = va_arg(ap, int);
123 z = va_arg(ap, const char*);
drh61f92cd2014-01-11 03:06:18 +0000124 lemon_addtext(str, &nUsed, z, k, iWidth);
drh898799f2014-01-10 23:21:00 +0000125 }else if( c=='%' ){
drh61f92cd2014-01-11 03:06:18 +0000126 lemon_addtext(str, &nUsed, "%", 1, 0);
drh898799f2014-01-10 23:21:00 +0000127 }else{
128 fprintf(stderr, "illegal format\n");
129 exit(1);
130 }
131 j = i+1;
132 }
133 }
drh61f92cd2014-01-11 03:06:18 +0000134 lemon_addtext(str, &nUsed, &zFormat[j], i-j, 0);
drh898799f2014-01-10 23:21:00 +0000135 return nUsed;
136}
137static int lemon_sprintf(char *str, const char *format, ...){
138 va_list ap;
139 int rc;
140 va_start(ap, format);
141 rc = lemon_vsprintf(str, format, ap);
142 va_end(ap);
143 return rc;
144}
145static void lemon_strcpy(char *dest, const char *src){
146 while( (*(dest++) = *(src++))!=0 ){}
147}
148static void lemon_strcat(char *dest, const char *src){
149 while( *dest ) dest++;
150 lemon_strcpy(dest, src);
151}
152
153
icculus9e44cf12010-02-14 17:14:22 +0000154/* a few forward declarations... */
155struct rule;
156struct lemon;
157struct action;
158
drhe9278182007-07-18 18:16:29 +0000159static struct action *Action_new(void);
160static struct action *Action_sort(struct action *);
drh75897232000-05-29 14:26:00 +0000161
162/********** From the file "build.h" ************************************/
163void FindRulePrecedences();
164void FindFirstSets();
165void FindStates();
166void FindLinks();
167void FindFollowSets();
168void FindActions();
169
170/********* From the file "configlist.h" *********************************/
icculus9e44cf12010-02-14 17:14:22 +0000171void Configlist_init(void);
172struct config *Configlist_add(struct rule *, int);
173struct config *Configlist_addbasis(struct rule *, int);
174void Configlist_closure(struct lemon *);
175void Configlist_sort(void);
176void Configlist_sortbasis(void);
177struct config *Configlist_return(void);
178struct config *Configlist_basis(void);
179void Configlist_eat(struct config *);
180void Configlist_reset(void);
drh75897232000-05-29 14:26:00 +0000181
182/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +0000183void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +0000184
185/****** From the file "option.h" ******************************************/
icculus9e44cf12010-02-14 17:14:22 +0000186enum option_type { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
187 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR};
drh75897232000-05-29 14:26:00 +0000188struct s_options {
icculus9e44cf12010-02-14 17:14:22 +0000189 enum option_type type;
190 const char *label;
drh75897232000-05-29 14:26:00 +0000191 char *arg;
icculus9e44cf12010-02-14 17:14:22 +0000192 const char *message;
drh75897232000-05-29 14:26:00 +0000193};
icculus9e44cf12010-02-14 17:14:22 +0000194int OptInit(char**,struct s_options*,FILE*);
195int OptNArgs(void);
196char *OptArg(int);
197void OptErr(int);
198void OptPrint(void);
drh75897232000-05-29 14:26:00 +0000199
200/******** From the file "parse.h" *****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000201void Parse(struct lemon *lemp);
drh75897232000-05-29 14:26:00 +0000202
203/********* From the file "plink.h" ***************************************/
icculus9e44cf12010-02-14 17:14:22 +0000204struct plink *Plink_new(void);
205void Plink_add(struct plink **, struct config *);
206void Plink_copy(struct plink **, struct plink *);
207void Plink_delete(struct plink *);
drh75897232000-05-29 14:26:00 +0000208
209/********** From the file "report.h" *************************************/
icculus9e44cf12010-02-14 17:14:22 +0000210void Reprint(struct lemon *);
211void ReportOutput(struct lemon *);
212void ReportTable(struct lemon *, int);
213void ReportHeader(struct lemon *);
214void CompressTables(struct lemon *);
215void ResortStates(struct lemon *);
drh75897232000-05-29 14:26:00 +0000216
217/********** From the file "set.h" ****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000218void SetSize(int); /* All sets will be of size N */
219char *SetNew(void); /* A new set for element 0..N */
220void SetFree(char*); /* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +0000221int SetAdd(char*,int); /* Add element to a set */
222int SetUnion(char *,char *); /* A <- A U B, thru element N */
drh75897232000-05-29 14:26:00 +0000223#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
224
225/********** From the file "struct.h" *************************************/
226/*
227** Principal data structures for the LEMON parser generator.
228*/
229
drhaa9f1122007-08-23 02:50:56 +0000230typedef enum {LEMON_FALSE=0, LEMON_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000231
232/* Symbols (terminals and nonterminals) of the grammar are stored
233** in the following: */
icculus9e44cf12010-02-14 17:14:22 +0000234enum symbol_type {
235 TERMINAL,
236 NONTERMINAL,
237 MULTITERMINAL
238};
239enum e_assoc {
drh75897232000-05-29 14:26:00 +0000240 LEFT,
241 RIGHT,
242 NONE,
243 UNK
icculus9e44cf12010-02-14 17:14:22 +0000244};
245struct symbol {
246 const char *name; /* Name of the symbol */
247 int index; /* Index number for this symbol */
248 enum symbol_type type; /* Symbols are all either TERMINALS or NTs */
249 struct rule *rule; /* Linked list of rules of this (if an NT) */
250 struct symbol *fallback; /* fallback token in case this token doesn't parse */
251 int prec; /* Precedence if defined (-1 otherwise) */
252 enum e_assoc assoc; /* Associativity if precedence is defined */
drh75897232000-05-29 14:26:00 +0000253 char *firstset; /* First-set for all rules of this symbol */
254 Boolean lambda; /* True if NT and can generate an empty string */
drhc4dd3fd2008-01-22 01:48:05 +0000255 int useCnt; /* Number of times used */
drh75897232000-05-29 14:26:00 +0000256 char *destructor; /* Code which executes whenever this symbol is
257 ** popped from the stack during error processing */
drh4dc8ef52008-07-01 17:13:57 +0000258 int destLineno; /* Line number for start of destructor */
drh75897232000-05-29 14:26:00 +0000259 char *datatype; /* The data type of information held by this
260 ** object. Only used if type==NONTERMINAL */
261 int dtnum; /* The data type number. In the parser, the value
262 ** stack is a union. The .yy%d element of this
263 ** union is the correct data type for this object */
drhfd405312005-11-06 04:06:59 +0000264 /* The following fields are used by MULTITERMINALs only */
265 int nsubsym; /* Number of constituent symbols in the MULTI */
266 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000267};
268
269/* Each production rule in the grammar is stored in the following
270** structure. */
271struct rule {
272 struct symbol *lhs; /* Left-hand side of the rule */
icculus9e44cf12010-02-14 17:14:22 +0000273 const char *lhsalias; /* Alias for the LHS (NULL if none) */
drhb4960992007-10-05 16:16:36 +0000274 int lhsStart; /* True if left-hand side is the start symbol */
drh75897232000-05-29 14:26:00 +0000275 int ruleline; /* Line number for the rule */
276 int nrhs; /* Number of RHS symbols */
277 struct symbol **rhs; /* The RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +0000278 const char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
drh75897232000-05-29 14:26:00 +0000279 int line; /* Line number at which code begins */
icculus9e44cf12010-02-14 17:14:22 +0000280 const char *code; /* The code executed when this rule is reduced */
drh75897232000-05-29 14:26:00 +0000281 struct symbol *precsym; /* Precedence symbol for this rule */
282 int index; /* An index number for this rule */
283 Boolean canReduce; /* True if this rule is ever reduced */
284 struct rule *nextlhs; /* Next rule with the same LHS */
285 struct rule *next; /* Next rule in the global list */
286};
287
288/* A configuration is a production rule of the grammar together with
289** a mark (dot) showing how much of that rule has been processed so far.
290** Configurations also contain a follow-set which is a list of terminal
291** symbols which are allowed to immediately follow the end of the rule.
292** Every configuration is recorded as an instance of the following: */
icculus9e44cf12010-02-14 17:14:22 +0000293enum cfgstatus {
294 COMPLETE,
295 INCOMPLETE
296};
drh75897232000-05-29 14:26:00 +0000297struct config {
298 struct rule *rp; /* The rule upon which the configuration is based */
299 int dot; /* The parse point */
300 char *fws; /* Follow-set for this configuration only */
301 struct plink *fplp; /* Follow-set forward propagation links */
302 struct plink *bplp; /* Follow-set backwards propagation links */
303 struct state *stp; /* Pointer to state which contains this */
icculus9e44cf12010-02-14 17:14:22 +0000304 enum cfgstatus status; /* used during followset and shift computations */
drh75897232000-05-29 14:26:00 +0000305 struct config *next; /* Next configuration in the state */
306 struct config *bp; /* The next basis configuration */
307};
308
icculus9e44cf12010-02-14 17:14:22 +0000309enum e_action {
310 SHIFT,
311 ACCEPT,
312 REDUCE,
313 ERROR,
314 SSCONFLICT, /* A shift/shift conflict */
315 SRCONFLICT, /* Was a reduce, but part of a conflict */
316 RRCONFLICT, /* Was a reduce, but part of a conflict */
317 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
318 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
319 NOT_USED /* Deleted by compression */
320};
321
drh75897232000-05-29 14:26:00 +0000322/* Every shift or reduce operation is stored as one of the following */
323struct action {
324 struct symbol *sp; /* The look-ahead symbol */
icculus9e44cf12010-02-14 17:14:22 +0000325 enum e_action type;
drh75897232000-05-29 14:26:00 +0000326 union {
327 struct state *stp; /* The new state, if a shift */
328 struct rule *rp; /* The rule, if a reduce */
329 } x;
330 struct action *next; /* Next action for this state */
331 struct action *collide; /* Next action with the same hash */
332};
333
334/* Each state of the generated parser's finite state machine
335** is encoded as an instance of the following structure. */
336struct state {
337 struct config *bp; /* The basis configurations for this state */
338 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000339 int statenum; /* Sequential number for this state */
drh75897232000-05-29 14:26:00 +0000340 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000341 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
342 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
343 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000344};
drh8b582012003-10-21 13:16:03 +0000345#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000346
347/* A followset propagation link indicates that the contents of one
348** configuration followset should be propagated to another whenever
349** the first changes. */
350struct plink {
351 struct config *cfp; /* The configuration to which linked */
352 struct plink *next; /* The next propagate link */
353};
354
355/* The state vector for the entire parser generator is recorded as
356** follows. (LEMON uses no global variables and makes little use of
357** static variables. Fields in the following structure can be thought
358** of as begin global variables in the program.) */
359struct lemon {
360 struct state **sorted; /* Table of states sorted by state number */
361 struct rule *rule; /* List of all rules */
362 int nstate; /* Number of states */
363 int nrule; /* Number of rules */
364 int nsymbol; /* Number of terminal and nonterminal symbols */
365 int nterminal; /* Number of terminal symbols */
366 struct symbol **symbols; /* Sorted array of pointers to symbols */
367 int errorcnt; /* Number of errors */
368 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000369 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000370 char *name; /* Name of the generated parser */
371 char *arg; /* Declaration of the 3th argument to parser */
372 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000373 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000374 char *start; /* Name of the start symbol for the grammar */
375 char *stacksize; /* Size of the parser stack */
376 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000377 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000378 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000379 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000380 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000381 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000382 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000383 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000384 char *filename; /* Name of the input file */
385 char *outname; /* Name of the current output file */
386 char *tokenprefix; /* A prefix added to token names in the .h file */
387 int nconflict; /* Number of parsing conflicts */
388 int tablesize; /* Size of the parse tables */
389 int basisflag; /* Print only basis configurations */
drh34ff57b2008-07-14 12:27:51 +0000390 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000391 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000392 char *argv0; /* Name of the program */
393};
394
395#define MemoryCheck(X) if((X)==0){ \
396 extern void memory_error(); \
397 memory_error(); \
398}
399
400/**************** From the file "table.h" *********************************/
401/*
402** All code in this file has been automatically generated
403** from a specification in the file
404** "table.q"
405** by the associative array code building program "aagen".
406** Do not edit this file! Instead, edit the specification
407** file, then rerun aagen.
408*/
409/*
410** Code for processing tables in the LEMON parser generator.
411*/
drh75897232000-05-29 14:26:00 +0000412/* Routines for handling a strings */
413
icculus9e44cf12010-02-14 17:14:22 +0000414const char *Strsafe(const char *);
drh75897232000-05-29 14:26:00 +0000415
icculus9e44cf12010-02-14 17:14:22 +0000416void Strsafe_init(void);
417int Strsafe_insert(const char *);
418const char *Strsafe_find(const char *);
drh75897232000-05-29 14:26:00 +0000419
420/* Routines for handling symbols of the grammar */
421
icculus9e44cf12010-02-14 17:14:22 +0000422struct symbol *Symbol_new(const char *);
423int Symbolcmpp(const void *, const void *);
424void Symbol_init(void);
425int Symbol_insert(struct symbol *, const char *);
426struct symbol *Symbol_find(const char *);
427struct symbol *Symbol_Nth(int);
428int Symbol_count(void);
429struct symbol **Symbol_arrayof(void);
drh75897232000-05-29 14:26:00 +0000430
431/* Routines to manage the state table */
432
icculus9e44cf12010-02-14 17:14:22 +0000433int Configcmp(const char *, const char *);
434struct state *State_new(void);
435void State_init(void);
436int State_insert(struct state *, struct config *);
437struct state *State_find(struct config *);
drh75897232000-05-29 14:26:00 +0000438struct state **State_arrayof(/* */);
439
440/* Routines used for efficiency in Configlist_add */
441
icculus9e44cf12010-02-14 17:14:22 +0000442void Configtable_init(void);
443int Configtable_insert(struct config *);
444struct config *Configtable_find(struct config *);
445void Configtable_clear(int(*)(struct config *));
446
drh75897232000-05-29 14:26:00 +0000447/****************** From the file "action.c" *******************************/
448/*
449** Routines processing parser actions in the LEMON parser generator.
450*/
451
452/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000453static struct action *Action_new(void){
drh75897232000-05-29 14:26:00 +0000454 static struct action *freelist = 0;
icculus9e44cf12010-02-14 17:14:22 +0000455 struct action *newaction;
drh75897232000-05-29 14:26:00 +0000456
457 if( freelist==0 ){
458 int i;
459 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +0000460 freelist = (struct action *)calloc(amt, sizeof(struct action));
drh75897232000-05-29 14:26:00 +0000461 if( freelist==0 ){
462 fprintf(stderr,"Unable to allocate memory for a new parser action.");
463 exit(1);
464 }
465 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
466 freelist[amt-1].next = 0;
467 }
icculus9e44cf12010-02-14 17:14:22 +0000468 newaction = freelist;
drh75897232000-05-29 14:26:00 +0000469 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +0000470 return newaction;
drh75897232000-05-29 14:26:00 +0000471}
472
drhe9278182007-07-18 18:16:29 +0000473/* Compare two actions for sorting purposes. Return negative, zero, or
474** positive if the first action is less than, equal to, or greater than
475** the first
476*/
477static int actioncmp(
478 struct action *ap1,
479 struct action *ap2
480){
drh75897232000-05-29 14:26:00 +0000481 int rc;
482 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000483 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000484 rc = (int)ap1->type - (int)ap2->type;
485 }
486 if( rc==0 && ap1->type==REDUCE ){
drh75897232000-05-29 14:26:00 +0000487 rc = ap1->x.rp->index - ap2->x.rp->index;
488 }
drhe594bc32009-11-03 13:02:25 +0000489 if( rc==0 ){
icculus7b429aa2010-03-03 17:09:01 +0000490 rc = (int) (ap2 - ap1);
drhe594bc32009-11-03 13:02:25 +0000491 }
drh75897232000-05-29 14:26:00 +0000492 return rc;
493}
494
495/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000496static struct action *Action_sort(
497 struct action *ap
498){
499 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
500 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000501 return ap;
502}
503
icculus9e44cf12010-02-14 17:14:22 +0000504void Action_add(
505 struct action **app,
506 enum e_action type,
507 struct symbol *sp,
508 char *arg
509){
510 struct action *newaction;
511 newaction = Action_new();
512 newaction->next = *app;
513 *app = newaction;
514 newaction->type = type;
515 newaction->sp = sp;
drh75897232000-05-29 14:26:00 +0000516 if( type==SHIFT ){
icculus9e44cf12010-02-14 17:14:22 +0000517 newaction->x.stp = (struct state *)arg;
drh75897232000-05-29 14:26:00 +0000518 }else{
icculus9e44cf12010-02-14 17:14:22 +0000519 newaction->x.rp = (struct rule *)arg;
drh75897232000-05-29 14:26:00 +0000520 }
521}
drh8b582012003-10-21 13:16:03 +0000522/********************** New code to implement the "acttab" module ***********/
523/*
524** This module implements routines use to construct the yy_action[] table.
525*/
526
527/*
528** The state of the yy_action table under construction is an instance of
drh8dc3e8f2010-01-07 03:53:03 +0000529** the following structure.
530**
531** The yy_action table maps the pair (state_number, lookahead) into an
532** action_number. The table is an array of integers pairs. The state_number
533** determines an initial offset into the yy_action array. The lookahead
534** value is then added to this initial offset to get an index X into the
535** yy_action array. If the aAction[X].lookahead equals the value of the
536** of the lookahead input, then the value of the action_number output is
537** aAction[X].action. If the lookaheads do not match then the
538** default action for the state_number is returned.
539**
540** All actions associated with a single state_number are first entered
541** into aLookahead[] using multiple calls to acttab_action(). Then the
542** actions for that single state_number are placed into the aAction[]
543** array with a single call to acttab_insert(). The acttab_insert() call
544** also resets the aLookahead[] array in preparation for the next
545** state number.
drh8b582012003-10-21 13:16:03 +0000546*/
icculus9e44cf12010-02-14 17:14:22 +0000547struct lookahead_action {
548 int lookahead; /* Value of the lookahead token */
549 int action; /* Action to take on the given lookahead */
550};
drh8b582012003-10-21 13:16:03 +0000551typedef struct acttab acttab;
552struct acttab {
553 int nAction; /* Number of used slots in aAction[] */
554 int nActionAlloc; /* Slots allocated for aAction[] */
icculus9e44cf12010-02-14 17:14:22 +0000555 struct lookahead_action
556 *aAction, /* The yy_action[] table under construction */
drh8b582012003-10-21 13:16:03 +0000557 *aLookahead; /* A single new transaction set */
558 int mnLookahead; /* Minimum aLookahead[].lookahead */
559 int mnAction; /* Action associated with mnLookahead */
560 int mxLookahead; /* Maximum aLookahead[].lookahead */
561 int nLookahead; /* Used slots in aLookahead[] */
562 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
563};
564
565/* Return the number of entries in the yy_action table */
566#define acttab_size(X) ((X)->nAction)
567
568/* The value for the N-th entry in yy_action */
569#define acttab_yyaction(X,N) ((X)->aAction[N].action)
570
571/* The value for the N-th entry in yy_lookahead */
572#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
573
574/* Free all memory associated with the given acttab */
575void acttab_free(acttab *p){
576 free( p->aAction );
577 free( p->aLookahead );
578 free( p );
579}
580
581/* Allocate a new acttab structure */
582acttab *acttab_alloc(void){
icculus9e44cf12010-02-14 17:14:22 +0000583 acttab *p = (acttab *) calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000584 if( p==0 ){
585 fprintf(stderr,"Unable to allocate memory for a new acttab.");
586 exit(1);
587 }
588 memset(p, 0, sizeof(*p));
589 return p;
590}
591
drh8dc3e8f2010-01-07 03:53:03 +0000592/* Add a new action to the current transaction set.
593**
594** This routine is called once for each lookahead for a particular
595** state.
drh8b582012003-10-21 13:16:03 +0000596*/
597void acttab_action(acttab *p, int lookahead, int action){
598 if( p->nLookahead>=p->nLookaheadAlloc ){
599 p->nLookaheadAlloc += 25;
icculus9e44cf12010-02-14 17:14:22 +0000600 p->aLookahead = (struct lookahead_action *) realloc( p->aLookahead,
drh8b582012003-10-21 13:16:03 +0000601 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
602 if( p->aLookahead==0 ){
603 fprintf(stderr,"malloc failed\n");
604 exit(1);
605 }
606 }
607 if( p->nLookahead==0 ){
608 p->mxLookahead = lookahead;
609 p->mnLookahead = lookahead;
610 p->mnAction = action;
611 }else{
612 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
613 if( p->mnLookahead>lookahead ){
614 p->mnLookahead = lookahead;
615 p->mnAction = action;
616 }
617 }
618 p->aLookahead[p->nLookahead].lookahead = lookahead;
619 p->aLookahead[p->nLookahead].action = action;
620 p->nLookahead++;
621}
622
623/*
624** Add the transaction set built up with prior calls to acttab_action()
625** into the current action table. Then reset the transaction set back
626** to an empty set in preparation for a new round of acttab_action() calls.
627**
628** Return the offset into the action table of the new transaction.
629*/
630int acttab_insert(acttab *p){
631 int i, j, k, n;
632 assert( p->nLookahead>0 );
633
634 /* Make sure we have enough space to hold the expanded action table
635 ** in the worst case. The worst case occurs if the transaction set
636 ** must be appended to the current action table
637 */
drh784d86f2004-02-19 18:41:53 +0000638 n = p->mxLookahead + 1;
drh8dc3e8f2010-01-07 03:53:03 +0000639 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000640 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000641 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
icculus9e44cf12010-02-14 17:14:22 +0000642 p->aAction = (struct lookahead_action *) realloc( p->aAction,
drh8b582012003-10-21 13:16:03 +0000643 sizeof(p->aAction[0])*p->nActionAlloc);
644 if( p->aAction==0 ){
645 fprintf(stderr,"malloc failed\n");
646 exit(1);
647 }
drhfdbf9282003-10-21 16:34:41 +0000648 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000649 p->aAction[i].lookahead = -1;
650 p->aAction[i].action = -1;
651 }
652 }
653
drh8dc3e8f2010-01-07 03:53:03 +0000654 /* Scan the existing action table looking for an offset that is a
655 ** duplicate of the current transaction set. Fall out of the loop
656 ** if and when the duplicate is found.
drh8b582012003-10-21 13:16:03 +0000657 **
658 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
659 */
drh8dc3e8f2010-01-07 03:53:03 +0000660 for(i=p->nAction-1; i>=0; i--){
drhf16371d2009-11-03 19:18:31 +0000661 if( p->aAction[i].lookahead==p->mnLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000662 /* All lookaheads and actions in the aLookahead[] transaction
663 ** must match against the candidate aAction[i] entry. */
drh8b582012003-10-21 13:16:03 +0000664 if( p->aAction[i].action!=p->mnAction ) continue;
665 for(j=0; j<p->nLookahead; j++){
666 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
667 if( k<0 || k>=p->nAction ) break;
668 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
669 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
670 }
671 if( j<p->nLookahead ) continue;
drh8dc3e8f2010-01-07 03:53:03 +0000672
673 /* No possible lookahead value that is not in the aLookahead[]
674 ** transaction is allowed to match aAction[i] */
drh8b582012003-10-21 13:16:03 +0000675 n = 0;
676 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000677 if( p->aAction[j].lookahead<0 ) continue;
678 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000679 }
drhfdbf9282003-10-21 16:34:41 +0000680 if( n==p->nLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000681 break; /* An exact match is found at offset i */
drhfdbf9282003-10-21 16:34:41 +0000682 }
drh8b582012003-10-21 13:16:03 +0000683 }
684 }
drh8dc3e8f2010-01-07 03:53:03 +0000685
686 /* If no existing offsets exactly match the current transaction, find an
687 ** an empty offset in the aAction[] table in which we can add the
688 ** aLookahead[] transaction.
689 */
drhf16371d2009-11-03 19:18:31 +0000690 if( i<0 ){
drh8dc3e8f2010-01-07 03:53:03 +0000691 /* Look for holes in the aAction[] table that fit the current
692 ** aLookahead[] transaction. Leave i set to the offset of the hole.
693 ** If no holes are found, i is left at p->nAction, which means the
694 ** transaction will be appended. */
695 for(i=0; i<p->nActionAlloc - p->mxLookahead; i++){
drhf16371d2009-11-03 19:18:31 +0000696 if( p->aAction[i].lookahead<0 ){
697 for(j=0; j<p->nLookahead; j++){
698 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
699 if( k<0 ) break;
700 if( p->aAction[k].lookahead>=0 ) break;
701 }
702 if( j<p->nLookahead ) continue;
703 for(j=0; j<p->nAction; j++){
704 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
705 }
706 if( j==p->nAction ){
707 break; /* Fits in empty slots */
708 }
709 }
710 }
711 }
drh8b582012003-10-21 13:16:03 +0000712 /* Insert transaction set at index i. */
713 for(j=0; j<p->nLookahead; j++){
714 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
715 p->aAction[k] = p->aLookahead[j];
716 if( k>=p->nAction ) p->nAction = k+1;
717 }
718 p->nLookahead = 0;
719
720 /* Return the offset that is added to the lookahead in order to get the
721 ** index into yy_action of the action */
722 return i - p->mnLookahead;
723}
724
drh75897232000-05-29 14:26:00 +0000725/********************** From the file "build.c" *****************************/
726/*
727** Routines to construction the finite state machine for the LEMON
728** parser generator.
729*/
730
731/* Find a precedence symbol of every rule in the grammar.
732**
733** Those rules which have a precedence symbol coded in the input
734** grammar using the "[symbol]" construct will already have the
735** rp->precsym field filled. Other rules take as their precedence
736** symbol the first RHS symbol with a defined precedence. If there
737** are not RHS symbols with a defined precedence, the precedence
738** symbol field is left blank.
739*/
icculus9e44cf12010-02-14 17:14:22 +0000740void FindRulePrecedences(struct lemon *xp)
drh75897232000-05-29 14:26:00 +0000741{
742 struct rule *rp;
743 for(rp=xp->rule; rp; rp=rp->next){
744 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000745 int i, j;
746 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
747 struct symbol *sp = rp->rhs[i];
748 if( sp->type==MULTITERMINAL ){
749 for(j=0; j<sp->nsubsym; j++){
750 if( sp->subsym[j]->prec>=0 ){
751 rp->precsym = sp->subsym[j];
752 break;
753 }
754 }
755 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000756 rp->precsym = rp->rhs[i];
drhf2f105d2012-08-20 15:53:54 +0000757 }
drh75897232000-05-29 14:26:00 +0000758 }
759 }
760 }
761 return;
762}
763
764/* Find all nonterminals which will generate the empty string.
765** Then go back and compute the first sets of every nonterminal.
766** The first set is the set of all terminal symbols which can begin
767** a string generated by that nonterminal.
768*/
icculus9e44cf12010-02-14 17:14:22 +0000769void FindFirstSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000770{
drhfd405312005-11-06 04:06:59 +0000771 int i, j;
drh75897232000-05-29 14:26:00 +0000772 struct rule *rp;
773 int progress;
774
775 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000776 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000777 }
778 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
779 lemp->symbols[i]->firstset = SetNew();
780 }
781
782 /* First compute all lambdas */
783 do{
784 progress = 0;
785 for(rp=lemp->rule; rp; rp=rp->next){
786 if( rp->lhs->lambda ) continue;
787 for(i=0; i<rp->nrhs; i++){
drh7dd1ac62012-01-07 15:17:18 +0000788 struct symbol *sp = rp->rhs[i];
789 assert( sp->type==NONTERMINAL || sp->lambda==LEMON_FALSE );
790 if( sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000791 }
792 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000793 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000794 progress = 1;
795 }
796 }
797 }while( progress );
798
799 /* Now compute all first sets */
800 do{
801 struct symbol *s1, *s2;
802 progress = 0;
803 for(rp=lemp->rule; rp; rp=rp->next){
804 s1 = rp->lhs;
805 for(i=0; i<rp->nrhs; i++){
806 s2 = rp->rhs[i];
807 if( s2->type==TERMINAL ){
808 progress += SetAdd(s1->firstset,s2->index);
809 break;
drhfd405312005-11-06 04:06:59 +0000810 }else if( s2->type==MULTITERMINAL ){
811 for(j=0; j<s2->nsubsym; j++){
812 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
813 }
814 break;
drhf2f105d2012-08-20 15:53:54 +0000815 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000816 if( s1->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000817 }else{
drh75897232000-05-29 14:26:00 +0000818 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000819 if( s2->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000820 }
drh75897232000-05-29 14:26:00 +0000821 }
822 }
823 }while( progress );
824 return;
825}
826
827/* Compute all LR(0) states for the grammar. Links
828** are added to between some states so that the LR(1) follow sets
829** can be computed later.
830*/
icculus9e44cf12010-02-14 17:14:22 +0000831PRIVATE struct state *getstate(struct lemon *); /* forward reference */
832void FindStates(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000833{
834 struct symbol *sp;
835 struct rule *rp;
836
837 Configlist_init();
838
839 /* Find the start symbol */
840 if( lemp->start ){
841 sp = Symbol_find(lemp->start);
842 if( sp==0 ){
843 ErrorMsg(lemp->filename,0,
844"The specified start symbol \"%s\" is not \
845in a nonterminal of the grammar. \"%s\" will be used as the start \
846symbol instead.",lemp->start,lemp->rule->lhs->name);
847 lemp->errorcnt++;
848 sp = lemp->rule->lhs;
849 }
850 }else{
851 sp = lemp->rule->lhs;
852 }
853
854 /* Make sure the start symbol doesn't occur on the right-hand side of
855 ** any rule. Report an error if it does. (YACC would generate a new
856 ** start symbol in this case.) */
857 for(rp=lemp->rule; rp; rp=rp->next){
858 int i;
859 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000860 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000861 ErrorMsg(lemp->filename,0,
862"The start symbol \"%s\" occurs on the \
863right-hand side of a rule. This will result in a parser which \
864does not work properly.",sp->name);
865 lemp->errorcnt++;
866 }
867 }
868 }
869
870 /* The basis configuration set for the first state
871 ** is all rules which have the start symbol as their
872 ** left-hand side */
873 for(rp=sp->rule; rp; rp=rp->nextlhs){
874 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000875 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000876 newcfp = Configlist_addbasis(rp,0);
877 SetAdd(newcfp->fws,0);
878 }
879
880 /* Compute the first state. All other states will be
881 ** computed automatically during the computation of the first one.
882 ** The returned pointer to the first state is not used. */
883 (void)getstate(lemp);
884 return;
885}
886
887/* Return a pointer to a state which is described by the configuration
888** list which has been built from calls to Configlist_add.
889*/
icculus9e44cf12010-02-14 17:14:22 +0000890PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
891PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000892{
893 struct config *cfp, *bp;
894 struct state *stp;
895
896 /* Extract the sorted basis of the new state. The basis was constructed
897 ** by prior calls to "Configlist_addbasis()". */
898 Configlist_sortbasis();
899 bp = Configlist_basis();
900
901 /* Get a state with the same basis */
902 stp = State_find(bp);
903 if( stp ){
904 /* A state with the same basis already exists! Copy all the follow-set
905 ** propagation links from the state under construction into the
906 ** preexisting state, then return a pointer to the preexisting state */
907 struct config *x, *y;
908 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
909 Plink_copy(&y->bplp,x->bplp);
910 Plink_delete(x->fplp);
911 x->fplp = x->bplp = 0;
912 }
913 cfp = Configlist_return();
914 Configlist_eat(cfp);
915 }else{
916 /* This really is a new state. Construct all the details */
917 Configlist_closure(lemp); /* Compute the configuration closure */
918 Configlist_sort(); /* Sort the configuration closure */
919 cfp = Configlist_return(); /* Get a pointer to the config list */
920 stp = State_new(); /* A new state structure */
921 MemoryCheck(stp);
922 stp->bp = bp; /* Remember the configuration basis */
923 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000924 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000925 stp->ap = 0; /* No actions, yet. */
926 State_insert(stp,stp->bp); /* Add to the state table */
927 buildshifts(lemp,stp); /* Recursively compute successor states */
928 }
929 return stp;
930}
931
drhfd405312005-11-06 04:06:59 +0000932/*
933** Return true if two symbols are the same.
934*/
icculus9e44cf12010-02-14 17:14:22 +0000935int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +0000936{
937 int i;
938 if( a==b ) return 1;
939 if( a->type!=MULTITERMINAL ) return 0;
940 if( b->type!=MULTITERMINAL ) return 0;
941 if( a->nsubsym!=b->nsubsym ) return 0;
942 for(i=0; i<a->nsubsym; i++){
943 if( a->subsym[i]!=b->subsym[i] ) return 0;
944 }
945 return 1;
946}
947
drh75897232000-05-29 14:26:00 +0000948/* Construct all successor states to the given state. A "successor"
949** state is any state which can be reached by a shift action.
950*/
icculus9e44cf12010-02-14 17:14:22 +0000951PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +0000952{
953 struct config *cfp; /* For looping thru the config closure of "stp" */
954 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +0000955 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +0000956 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
957 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
958 struct state *newstp; /* A pointer to a successor state */
959
960 /* Each configuration becomes complete after it contibutes to a successor
961 ** state. Initially, all configurations are incomplete */
962 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
963
964 /* Loop through all configurations of the state "stp" */
965 for(cfp=stp->cfp; cfp; cfp=cfp->next){
966 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
967 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
968 Configlist_reset(); /* Reset the new config set */
969 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
970
971 /* For every configuration in the state "stp" which has the symbol "sp"
972 ** following its dot, add the same configuration to the basis set under
973 ** construction but with the dot shifted one symbol to the right. */
974 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
975 if( bcfp->status==COMPLETE ) continue; /* Already used */
976 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
977 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +0000978 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +0000979 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +0000980 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
981 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +0000982 }
983
984 /* Get a pointer to the state described by the basis configuration set
985 ** constructed in the preceding loop */
986 newstp = getstate(lemp);
987
988 /* The state "newstp" is reached from the state "stp" by a shift action
989 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +0000990 if( sp->type==MULTITERMINAL ){
991 int i;
992 for(i=0; i<sp->nsubsym; i++){
993 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
994 }
995 }else{
996 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
997 }
drh75897232000-05-29 14:26:00 +0000998 }
999}
1000
1001/*
1002** Construct the propagation links
1003*/
icculus9e44cf12010-02-14 17:14:22 +00001004void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001005{
1006 int i;
1007 struct config *cfp, *other;
1008 struct state *stp;
1009 struct plink *plp;
1010
1011 /* Housekeeping detail:
1012 ** Add to every propagate link a pointer back to the state to
1013 ** which the link is attached. */
1014 for(i=0; i<lemp->nstate; i++){
1015 stp = lemp->sorted[i];
1016 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1017 cfp->stp = stp;
1018 }
1019 }
1020
1021 /* Convert all backlinks into forward links. Only the forward
1022 ** links are used in the follow-set computation. */
1023 for(i=0; i<lemp->nstate; i++){
1024 stp = lemp->sorted[i];
1025 for(cfp=stp->cfp; cfp; cfp=cfp->next){
1026 for(plp=cfp->bplp; plp; plp=plp->next){
1027 other = plp->cfp;
1028 Plink_add(&other->fplp,cfp);
1029 }
1030 }
1031 }
1032}
1033
1034/* Compute all followsets.
1035**
1036** A followset is the set of all symbols which can come immediately
1037** after a configuration.
1038*/
icculus9e44cf12010-02-14 17:14:22 +00001039void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001040{
1041 int i;
1042 struct config *cfp;
1043 struct plink *plp;
1044 int progress;
1045 int change;
1046
1047 for(i=0; i<lemp->nstate; i++){
1048 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1049 cfp->status = INCOMPLETE;
1050 }
1051 }
1052
1053 do{
1054 progress = 0;
1055 for(i=0; i<lemp->nstate; i++){
1056 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
1057 if( cfp->status==COMPLETE ) continue;
1058 for(plp=cfp->fplp; plp; plp=plp->next){
1059 change = SetUnion(plp->cfp->fws,cfp->fws);
1060 if( change ){
1061 plp->cfp->status = INCOMPLETE;
1062 progress = 1;
drhf2f105d2012-08-20 15:53:54 +00001063 }
1064 }
drh75897232000-05-29 14:26:00 +00001065 cfp->status = COMPLETE;
1066 }
1067 }
1068 }while( progress );
1069}
1070
drh3cb2f6e2012-01-09 14:19:05 +00001071static int resolve_conflict(struct action *,struct action *);
drh75897232000-05-29 14:26:00 +00001072
1073/* Compute the reduce actions, and resolve conflicts.
1074*/
icculus9e44cf12010-02-14 17:14:22 +00001075void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001076{
1077 int i,j;
1078 struct config *cfp;
1079 struct state *stp;
1080 struct symbol *sp;
1081 struct rule *rp;
1082
1083 /* Add all of the reduce actions
1084 ** A reduce action is added for each element of the followset of
1085 ** a configuration which has its dot at the extreme right.
1086 */
1087 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
1088 stp = lemp->sorted[i];
1089 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
1090 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
1091 for(j=0; j<lemp->nterminal; j++){
1092 if( SetFind(cfp->fws,j) ){
1093 /* Add a reduce action to the state "stp" which will reduce by the
1094 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +00001095 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +00001096 }
drhf2f105d2012-08-20 15:53:54 +00001097 }
drh75897232000-05-29 14:26:00 +00001098 }
1099 }
1100 }
1101
1102 /* Add the accepting token */
1103 if( lemp->start ){
1104 sp = Symbol_find(lemp->start);
1105 if( sp==0 ) sp = lemp->rule->lhs;
1106 }else{
1107 sp = lemp->rule->lhs;
1108 }
1109 /* Add to the first state (which is always the starting state of the
1110 ** finite state machine) an action to ACCEPT if the lookahead is the
1111 ** start nonterminal. */
1112 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1113
1114 /* Resolve conflicts */
1115 for(i=0; i<lemp->nstate; i++){
1116 struct action *ap, *nap;
1117 struct state *stp;
1118 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001119 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001120 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001121 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001122 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1123 /* The two actions "ap" and "nap" have the same lookahead.
1124 ** Figure out which one should be used */
drh3cb2f6e2012-01-09 14:19:05 +00001125 lemp->nconflict += resolve_conflict(ap,nap);
drh75897232000-05-29 14:26:00 +00001126 }
1127 }
1128 }
1129
1130 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001131 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001132 for(i=0; i<lemp->nstate; i++){
1133 struct action *ap;
1134 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001135 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001136 }
1137 }
1138 for(rp=lemp->rule; rp; rp=rp->next){
1139 if( rp->canReduce ) continue;
1140 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1141 lemp->errorcnt++;
1142 }
1143}
1144
1145/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001146** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001147**
1148** NO LONGER TRUE:
1149** To resolve a conflict, first look to see if either action
1150** is on an error rule. In that case, take the action which
1151** is not associated with the error rule. If neither or both
1152** actions are associated with an error rule, then try to
1153** use precedence to resolve the conflict.
1154**
1155** If either action is a SHIFT, then it must be apx. This
1156** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1157*/
icculus9e44cf12010-02-14 17:14:22 +00001158static int resolve_conflict(
1159 struct action *apx,
drh3cb2f6e2012-01-09 14:19:05 +00001160 struct action *apy
icculus9e44cf12010-02-14 17:14:22 +00001161){
drh75897232000-05-29 14:26:00 +00001162 struct symbol *spx, *spy;
1163 int errcnt = 0;
1164 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001165 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001166 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001167 errcnt++;
1168 }
drh75897232000-05-29 14:26:00 +00001169 if( apx->type==SHIFT && apy->type==REDUCE ){
1170 spx = apx->sp;
1171 spy = apy->x.rp->precsym;
1172 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1173 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001174 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001175 errcnt++;
drhdd7e9db2010-07-19 01:52:07 +00001176 }else if( spx->prec>spy->prec ){ /* higher precedence wins */
drh75897232000-05-29 14:26:00 +00001177 apy->type = RD_RESOLVED;
1178 }else if( spx->prec<spy->prec ){
1179 apx->type = SH_RESOLVED;
1180 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1181 apy->type = RD_RESOLVED; /* associativity */
1182 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1183 apx->type = SH_RESOLVED;
1184 }else{
1185 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh9892c5d2007-12-21 00:02:11 +00001186 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001187 errcnt++;
1188 }
1189 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1190 spx = apx->x.rp->precsym;
1191 spy = apy->x.rp->precsym;
1192 if( spx==0 || spy==0 || spx->prec<0 ||
1193 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001194 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001195 errcnt++;
1196 }else if( spx->prec>spy->prec ){
1197 apy->type = RD_RESOLVED;
1198 }else if( spx->prec<spy->prec ){
1199 apx->type = RD_RESOLVED;
1200 }
1201 }else{
drhb59499c2002-02-23 18:45:13 +00001202 assert(
1203 apx->type==SH_RESOLVED ||
1204 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001205 apx->type==SSCONFLICT ||
1206 apx->type==SRCONFLICT ||
1207 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001208 apy->type==SH_RESOLVED ||
1209 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001210 apy->type==SSCONFLICT ||
1211 apy->type==SRCONFLICT ||
1212 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001213 );
1214 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1215 ** REDUCEs on the list. If we reach this point it must be because
1216 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001217 }
1218 return errcnt;
1219}
1220/********************* From the file "configlist.c" *************************/
1221/*
1222** Routines to processing a configuration list and building a state
1223** in the LEMON parser generator.
1224*/
1225
1226static struct config *freelist = 0; /* List of free configurations */
1227static struct config *current = 0; /* Top of list of configurations */
1228static struct config **currentend = 0; /* Last on list of configs */
1229static struct config *basis = 0; /* Top of list of basis configs */
1230static struct config **basisend = 0; /* End of list of basis configs */
1231
1232/* Return a pointer to a new configuration */
1233PRIVATE struct config *newconfig(){
icculus9e44cf12010-02-14 17:14:22 +00001234 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001235 if( freelist==0 ){
1236 int i;
1237 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001238 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001239 if( freelist==0 ){
1240 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1241 exit(1);
1242 }
1243 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1244 freelist[amt-1].next = 0;
1245 }
icculus9e44cf12010-02-14 17:14:22 +00001246 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001247 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001248 return newcfg;
drh75897232000-05-29 14:26:00 +00001249}
1250
1251/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001252PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001253{
1254 old->next = freelist;
1255 freelist = old;
1256}
1257
1258/* Initialized the configuration list builder */
1259void Configlist_init(){
1260 current = 0;
1261 currentend = &current;
1262 basis = 0;
1263 basisend = &basis;
1264 Configtable_init();
1265 return;
1266}
1267
1268/* Initialized the configuration list builder */
1269void Configlist_reset(){
1270 current = 0;
1271 currentend = &current;
1272 basis = 0;
1273 basisend = &basis;
1274 Configtable_clear(0);
1275 return;
1276}
1277
1278/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001279struct config *Configlist_add(
1280 struct rule *rp, /* The rule */
1281 int dot /* Index into the RHS of the rule where the dot goes */
1282){
drh75897232000-05-29 14:26:00 +00001283 struct config *cfp, model;
1284
1285 assert( currentend!=0 );
1286 model.rp = rp;
1287 model.dot = dot;
1288 cfp = Configtable_find(&model);
1289 if( cfp==0 ){
1290 cfp = newconfig();
1291 cfp->rp = rp;
1292 cfp->dot = dot;
1293 cfp->fws = SetNew();
1294 cfp->stp = 0;
1295 cfp->fplp = cfp->bplp = 0;
1296 cfp->next = 0;
1297 cfp->bp = 0;
1298 *currentend = cfp;
1299 currentend = &cfp->next;
1300 Configtable_insert(cfp);
1301 }
1302 return cfp;
1303}
1304
1305/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001306struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001307{
1308 struct config *cfp, model;
1309
1310 assert( basisend!=0 );
1311 assert( currentend!=0 );
1312 model.rp = rp;
1313 model.dot = dot;
1314 cfp = Configtable_find(&model);
1315 if( cfp==0 ){
1316 cfp = newconfig();
1317 cfp->rp = rp;
1318 cfp->dot = dot;
1319 cfp->fws = SetNew();
1320 cfp->stp = 0;
1321 cfp->fplp = cfp->bplp = 0;
1322 cfp->next = 0;
1323 cfp->bp = 0;
1324 *currentend = cfp;
1325 currentend = &cfp->next;
1326 *basisend = cfp;
1327 basisend = &cfp->bp;
1328 Configtable_insert(cfp);
1329 }
1330 return cfp;
1331}
1332
1333/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001334void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001335{
1336 struct config *cfp, *newcfp;
1337 struct rule *rp, *newrp;
1338 struct symbol *sp, *xsp;
1339 int i, dot;
1340
1341 assert( currentend!=0 );
1342 for(cfp=current; cfp; cfp=cfp->next){
1343 rp = cfp->rp;
1344 dot = cfp->dot;
1345 if( dot>=rp->nrhs ) continue;
1346 sp = rp->rhs[dot];
1347 if( sp->type==NONTERMINAL ){
1348 if( sp->rule==0 && sp!=lemp->errsym ){
1349 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1350 sp->name);
1351 lemp->errorcnt++;
1352 }
1353 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1354 newcfp = Configlist_add(newrp,0);
1355 for(i=dot+1; i<rp->nrhs; i++){
1356 xsp = rp->rhs[i];
1357 if( xsp->type==TERMINAL ){
1358 SetAdd(newcfp->fws,xsp->index);
1359 break;
drhfd405312005-11-06 04:06:59 +00001360 }else if( xsp->type==MULTITERMINAL ){
1361 int k;
1362 for(k=0; k<xsp->nsubsym; k++){
1363 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1364 }
1365 break;
drhf2f105d2012-08-20 15:53:54 +00001366 }else{
drh75897232000-05-29 14:26:00 +00001367 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001368 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001369 }
1370 }
drh75897232000-05-29 14:26:00 +00001371 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1372 }
1373 }
1374 }
1375 return;
1376}
1377
1378/* Sort the configuration list */
1379void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001380 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001381 currentend = 0;
1382 return;
1383}
1384
1385/* Sort the basis configuration list */
1386void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001387 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001388 basisend = 0;
1389 return;
1390}
1391
1392/* Return a pointer to the head of the configuration list and
1393** reset the list */
1394struct config *Configlist_return(){
1395 struct config *old;
1396 old = current;
1397 current = 0;
1398 currentend = 0;
1399 return old;
1400}
1401
1402/* Return a pointer to the head of the configuration list and
1403** reset the list */
1404struct config *Configlist_basis(){
1405 struct config *old;
1406 old = basis;
1407 basis = 0;
1408 basisend = 0;
1409 return old;
1410}
1411
1412/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001413void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001414{
1415 struct config *nextcfp;
1416 for(; cfp; cfp=nextcfp){
1417 nextcfp = cfp->next;
1418 assert( cfp->fplp==0 );
1419 assert( cfp->bplp==0 );
1420 if( cfp->fws ) SetFree(cfp->fws);
1421 deleteconfig(cfp);
1422 }
1423 return;
1424}
1425/***************** From the file "error.c" *********************************/
1426/*
1427** Code for printing error message.
1428*/
1429
drhf9a2e7b2003-04-15 01:49:48 +00001430void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001431 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001432 fprintf(stderr, "%s:%d: ", filename, lineno);
1433 va_start(ap, format);
1434 vfprintf(stderr,format,ap);
1435 va_end(ap);
1436 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001437}
1438/**************** From the file "main.c" ************************************/
1439/*
1440** Main program file for the LEMON parser generator.
1441*/
1442
1443/* Report an out-of-memory condition and abort. This function
1444** is used mostly by the "MemoryCheck" macro in struct.h
1445*/
1446void memory_error(){
1447 fprintf(stderr,"Out of memory. Aborting...\n");
1448 exit(1);
1449}
1450
drh6d08b4d2004-07-20 12:45:22 +00001451static int nDefine = 0; /* Number of -D options on the command line */
1452static char **azDefine = 0; /* Name of the -D macros */
1453
1454/* This routine is called with the argument to each -D command-line option.
1455** Add the macro defined to the azDefine array.
1456*/
1457static void handle_D_option(char *z){
1458 char **paz;
1459 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001460 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001461 if( azDefine==0 ){
1462 fprintf(stderr,"out of memory\n");
1463 exit(1);
1464 }
1465 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001466 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001467 if( *paz==0 ){
1468 fprintf(stderr,"out of memory\n");
1469 exit(1);
1470 }
drh898799f2014-01-10 23:21:00 +00001471 lemon_strcpy(*paz, z);
drh6d08b4d2004-07-20 12:45:22 +00001472 for(z=*paz; *z && *z!='='; z++){}
1473 *z = 0;
1474}
1475
icculus3e143bd2010-02-14 00:48:49 +00001476static char *user_templatename = NULL;
1477static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001478 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001479 if( user_templatename==0 ){
1480 memory_error();
1481 }
drh898799f2014-01-10 23:21:00 +00001482 lemon_strcpy(user_templatename, z);
icculus3e143bd2010-02-14 00:48:49 +00001483}
drh75897232000-05-29 14:26:00 +00001484
1485/* The main program. Parse the command line and do it... */
icculus9e44cf12010-02-14 17:14:22 +00001486int main(int argc, char **argv)
drh75897232000-05-29 14:26:00 +00001487{
1488 static int version = 0;
1489 static int rpflag = 0;
1490 static int basisflag = 0;
1491 static int compress = 0;
1492 static int quiet = 0;
1493 static int statistics = 0;
1494 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001495 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001496 static int noResort = 0;
drh75897232000-05-29 14:26:00 +00001497 static struct s_options options[] = {
1498 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1499 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001500 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
icculus3e143bd2010-02-14 00:48:49 +00001501 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
drh75897232000-05-29 14:26:00 +00001502 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
shane58543932008-12-10 20:10:04 +00001503 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1504 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drhf5c4e0f2010-07-18 11:35:53 +00001505 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1506 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001507 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001508 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001509 {OPT_FLAG, "s", (char*)&statistics,
1510 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001511 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1512 {OPT_FLAG,0,0,0}
1513 };
1514 int i;
icculus42585cf2010-02-14 05:19:56 +00001515 int exitcode;
drh75897232000-05-29 14:26:00 +00001516 struct lemon lem;
1517
drhb0c86772000-06-02 23:21:26 +00001518 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001519 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001520 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001521 exit(0);
1522 }
drhb0c86772000-06-02 23:21:26 +00001523 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001524 fprintf(stderr,"Exactly one filename argument is required.\n");
1525 exit(1);
1526 }
drh954f6b42006-06-13 13:27:46 +00001527 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001528 lem.errorcnt = 0;
1529
1530 /* Initialize the machine */
1531 Strsafe_init();
1532 Symbol_init();
1533 State_init();
1534 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001535 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001536 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001537 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001538 Symbol_new("$");
1539 lem.errsym = Symbol_new("error");
drhc4dd3fd2008-01-22 01:48:05 +00001540 lem.errsym->useCnt = 0;
drh75897232000-05-29 14:26:00 +00001541
1542 /* Parse the input file */
1543 Parse(&lem);
1544 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001545 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001546 fprintf(stderr,"Empty grammar.\n");
1547 exit(1);
1548 }
1549
1550 /* Count and index the symbols of the grammar */
drh75897232000-05-29 14:26:00 +00001551 Symbol_new("{default}");
drh61f92cd2014-01-11 03:06:18 +00001552 lem.nsymbol = Symbol_count();
drh75897232000-05-29 14:26:00 +00001553 lem.symbols = Symbol_arrayof();
drh61f92cd2014-01-11 03:06:18 +00001554 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1555 qsort(lem.symbols,lem.nsymbol,sizeof(struct symbol*), Symbolcmpp);
1556 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1557 while( lem.symbols[i-1]->type==MULTITERMINAL ){ i--; }
1558 assert( strcmp(lem.symbols[i-1]->name,"{default}")==0 );
1559 lem.nsymbol = i - 1;
drh75897232000-05-29 14:26:00 +00001560 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1561 lem.nterminal = i;
1562
1563 /* Generate a reprint of the grammar, if requested on the command line */
1564 if( rpflag ){
1565 Reprint(&lem);
1566 }else{
1567 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001568 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001569
1570 /* Find the precedence for every production rule (that has one) */
1571 FindRulePrecedences(&lem);
1572
1573 /* Compute the lambda-nonterminals and the first-sets for every
1574 ** nonterminal */
1575 FindFirstSets(&lem);
1576
1577 /* Compute all LR(0) states. Also record follow-set propagation
1578 ** links so that the follow-set can be computed later */
1579 lem.nstate = 0;
1580 FindStates(&lem);
1581 lem.sorted = State_arrayof();
1582
1583 /* Tie up loose ends on the propagation links */
1584 FindLinks(&lem);
1585
1586 /* Compute the follow set of every reducible configuration */
1587 FindFollowSets(&lem);
1588
1589 /* Compute the action tables */
1590 FindActions(&lem);
1591
1592 /* Compress the action tables */
1593 if( compress==0 ) CompressTables(&lem);
1594
drhada354d2005-11-05 15:03:59 +00001595 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001596 ** occur at the end. This is an optimization that helps make the
1597 ** generated parser tables smaller. */
1598 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001599
drh75897232000-05-29 14:26:00 +00001600 /* Generate a report of the parser generated. (the "y.output" file) */
1601 if( !quiet ) ReportOutput(&lem);
1602
1603 /* Generate the source code for the parser */
1604 ReportTable(&lem, mhflag);
1605
1606 /* Produce a header file for use by the scanner. (This step is
1607 ** omitted if the "-m" option is used because makeheaders will
1608 ** generate the file for us.) */
1609 if( !mhflag ) ReportHeader(&lem);
1610 }
1611 if( statistics ){
1612 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1613 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1614 printf(" %d states, %d parser table entries, %d conflicts\n",
1615 lem.nstate, lem.tablesize, lem.nconflict);
1616 }
icculus8e158022010-02-16 16:09:03 +00001617 if( lem.nconflict > 0 ){
1618 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001619 }
1620
1621 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001622 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001623 exit(exitcode);
1624 return (exitcode);
drh75897232000-05-29 14:26:00 +00001625}
1626/******************** From the file "msort.c" *******************************/
1627/*
1628** A generic merge-sort program.
1629**
1630** USAGE:
1631** Let "ptr" be a pointer to some structure which is at the head of
1632** a null-terminated list. Then to sort the list call:
1633**
1634** ptr = msort(ptr,&(ptr->next),cmpfnc);
1635**
1636** In the above, "cmpfnc" is a pointer to a function which compares
1637** two instances of the structure and returns an integer, as in
1638** strcmp. The second argument is a pointer to the pointer to the
1639** second element of the linked list. This address is used to compute
1640** the offset to the "next" field within the structure. The offset to
1641** the "next" field must be constant for all structures in the list.
1642**
1643** The function returns a new pointer which is the head of the list
1644** after sorting.
1645**
1646** ALGORITHM:
1647** Merge-sort.
1648*/
1649
1650/*
1651** Return a pointer to the next structure in the linked list.
1652*/
drhd25d6922012-04-18 09:59:56 +00001653#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001654
1655/*
1656** Inputs:
1657** a: A sorted, null-terminated linked list. (May be null).
1658** b: A sorted, null-terminated linked list. (May be null).
1659** cmp: A pointer to the comparison function.
1660** offset: Offset in the structure to the "next" field.
1661**
1662** Return Value:
1663** A pointer to the head of a sorted list containing the elements
1664** of both a and b.
1665**
1666** Side effects:
1667** The "next" pointers for elements in the lists a and b are
1668** changed.
1669*/
drhe9278182007-07-18 18:16:29 +00001670static char *merge(
1671 char *a,
1672 char *b,
1673 int (*cmp)(const char*,const char*),
1674 int offset
1675){
drh75897232000-05-29 14:26:00 +00001676 char *ptr, *head;
1677
1678 if( a==0 ){
1679 head = b;
1680 }else if( b==0 ){
1681 head = a;
1682 }else{
drhe594bc32009-11-03 13:02:25 +00001683 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001684 ptr = a;
1685 a = NEXT(a);
1686 }else{
1687 ptr = b;
1688 b = NEXT(b);
1689 }
1690 head = ptr;
1691 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001692 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001693 NEXT(ptr) = a;
1694 ptr = a;
1695 a = NEXT(a);
1696 }else{
1697 NEXT(ptr) = b;
1698 ptr = b;
1699 b = NEXT(b);
1700 }
1701 }
1702 if( a ) NEXT(ptr) = a;
1703 else NEXT(ptr) = b;
1704 }
1705 return head;
1706}
1707
1708/*
1709** Inputs:
1710** list: Pointer to a singly-linked list of structures.
1711** next: Pointer to pointer to the second element of the list.
1712** cmp: A comparison function.
1713**
1714** Return Value:
1715** A pointer to the head of a sorted list containing the elements
1716** orginally in list.
1717**
1718** Side effects:
1719** The "next" pointers for elements in list are changed.
1720*/
1721#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001722static char *msort(
1723 char *list,
1724 char **next,
1725 int (*cmp)(const char*,const char*)
1726){
drhba99af52001-10-25 20:37:16 +00001727 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001728 char *ep;
1729 char *set[LISTSIZE];
1730 int i;
drhba99af52001-10-25 20:37:16 +00001731 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001732 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1733 while( list ){
1734 ep = list;
1735 list = NEXT(list);
1736 NEXT(ep) = 0;
1737 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1738 ep = merge(ep,set[i],cmp,offset);
1739 set[i] = 0;
1740 }
1741 set[i] = ep;
1742 }
1743 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001744 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001745 return ep;
1746}
1747/************************ From the file "option.c" **************************/
1748static char **argv;
1749static struct s_options *op;
1750static FILE *errstream;
1751
1752#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1753
1754/*
1755** Print the command line with a carrot pointing to the k-th character
1756** of the n-th field.
1757*/
icculus9e44cf12010-02-14 17:14:22 +00001758static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001759{
1760 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001761 if( argv[0] ) fprintf(err,"%s",argv[0]);
drh87cf1372008-08-13 20:09:06 +00001762 spcnt = lemonStrlen(argv[0]) + 1;
drh75897232000-05-29 14:26:00 +00001763 for(i=1; i<n && argv[i]; i++){
1764 fprintf(err," %s",argv[i]);
drh87cf1372008-08-13 20:09:06 +00001765 spcnt += lemonStrlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001766 }
1767 spcnt += k;
1768 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1769 if( spcnt<20 ){
1770 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1771 }else{
1772 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1773 }
1774}
1775
1776/*
1777** Return the index of the N-th non-switch argument. Return -1
1778** if N is out of range.
1779*/
icculus9e44cf12010-02-14 17:14:22 +00001780static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001781{
1782 int i;
1783 int dashdash = 0;
1784 if( argv!=0 && *argv!=0 ){
1785 for(i=1; argv[i]; i++){
1786 if( dashdash || !ISOPT(argv[i]) ){
1787 if( n==0 ) return i;
1788 n--;
1789 }
1790 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1791 }
1792 }
1793 return -1;
1794}
1795
1796static char emsg[] = "Command line syntax error: ";
1797
1798/*
1799** Process a flag command line argument.
1800*/
icculus9e44cf12010-02-14 17:14:22 +00001801static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001802{
1803 int v;
1804 int errcnt = 0;
1805 int j;
1806 for(j=0; op[j].label; j++){
drh87cf1372008-08-13 20:09:06 +00001807 if( strncmp(&argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001808 }
1809 v = argv[i][0]=='-' ? 1 : 0;
1810 if( op[j].label==0 ){
1811 if( err ){
1812 fprintf(err,"%sundefined option.\n",emsg);
1813 errline(i,1,err);
1814 }
1815 errcnt++;
1816 }else if( op[j].type==OPT_FLAG ){
1817 *((int*)op[j].arg) = v;
1818 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001819 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001820 }else if( op[j].type==OPT_FSTR ){
icculus9e44cf12010-02-14 17:14:22 +00001821 (*(void(*)(char *))(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001822 }else{
1823 if( err ){
1824 fprintf(err,"%smissing argument on switch.\n",emsg);
1825 errline(i,1,err);
1826 }
1827 errcnt++;
1828 }
1829 return errcnt;
1830}
1831
1832/*
1833** Process a command line switch which has an argument.
1834*/
icculus9e44cf12010-02-14 17:14:22 +00001835static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001836{
1837 int lv = 0;
1838 double dv = 0.0;
1839 char *sv = 0, *end;
1840 char *cp;
1841 int j;
1842 int errcnt = 0;
1843 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001844 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001845 *cp = 0;
1846 for(j=0; op[j].label; j++){
1847 if( strcmp(argv[i],op[j].label)==0 ) break;
1848 }
1849 *cp = '=';
1850 if( op[j].label==0 ){
1851 if( err ){
1852 fprintf(err,"%sundefined option.\n",emsg);
1853 errline(i,0,err);
1854 }
1855 errcnt++;
1856 }else{
1857 cp++;
1858 switch( op[j].type ){
1859 case OPT_FLAG:
1860 case OPT_FFLAG:
1861 if( err ){
1862 fprintf(err,"%soption requires an argument.\n",emsg);
1863 errline(i,0,err);
1864 }
1865 errcnt++;
1866 break;
1867 case OPT_DBL:
1868 case OPT_FDBL:
1869 dv = strtod(cp,&end);
1870 if( *end ){
1871 if( err ){
1872 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001873 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001874 }
1875 errcnt++;
1876 }
1877 break;
1878 case OPT_INT:
1879 case OPT_FINT:
1880 lv = strtol(cp,&end,0);
1881 if( *end ){
1882 if( err ){
1883 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001884 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001885 }
1886 errcnt++;
1887 }
1888 break;
1889 case OPT_STR:
1890 case OPT_FSTR:
1891 sv = cp;
1892 break;
1893 }
1894 switch( op[j].type ){
1895 case OPT_FLAG:
1896 case OPT_FFLAG:
1897 break;
1898 case OPT_DBL:
1899 *(double*)(op[j].arg) = dv;
1900 break;
1901 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00001902 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00001903 break;
1904 case OPT_INT:
1905 *(int*)(op[j].arg) = lv;
1906 break;
1907 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00001908 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00001909 break;
1910 case OPT_STR:
1911 *(char**)(op[j].arg) = sv;
1912 break;
1913 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00001914 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00001915 break;
1916 }
1917 }
1918 return errcnt;
1919}
1920
icculus9e44cf12010-02-14 17:14:22 +00001921int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00001922{
1923 int errcnt = 0;
1924 argv = a;
1925 op = o;
1926 errstream = err;
1927 if( argv && *argv && op ){
1928 int i;
1929 for(i=1; argv[i]; i++){
1930 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1931 errcnt += handleflags(i,err);
1932 }else if( strchr(argv[i],'=') ){
1933 errcnt += handleswitch(i,err);
1934 }
1935 }
1936 }
1937 if( errcnt>0 ){
1938 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001939 OptPrint();
drh75897232000-05-29 14:26:00 +00001940 exit(1);
1941 }
1942 return 0;
1943}
1944
drhb0c86772000-06-02 23:21:26 +00001945int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001946 int cnt = 0;
1947 int dashdash = 0;
1948 int i;
1949 if( argv!=0 && argv[0]!=0 ){
1950 for(i=1; argv[i]; i++){
1951 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1952 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1953 }
1954 }
1955 return cnt;
1956}
1957
icculus9e44cf12010-02-14 17:14:22 +00001958char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00001959{
1960 int i;
1961 i = argindex(n);
1962 return i>=0 ? argv[i] : 0;
1963}
1964
icculus9e44cf12010-02-14 17:14:22 +00001965void OptErr(int n)
drh75897232000-05-29 14:26:00 +00001966{
1967 int i;
1968 i = argindex(n);
1969 if( i>=0 ) errline(i,0,errstream);
1970}
1971
drhb0c86772000-06-02 23:21:26 +00001972void OptPrint(){
drh75897232000-05-29 14:26:00 +00001973 int i;
1974 int max, len;
1975 max = 0;
1976 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00001977 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00001978 switch( op[i].type ){
1979 case OPT_FLAG:
1980 case OPT_FFLAG:
1981 break;
1982 case OPT_INT:
1983 case OPT_FINT:
1984 len += 9; /* length of "<integer>" */
1985 break;
1986 case OPT_DBL:
1987 case OPT_FDBL:
1988 len += 6; /* length of "<real>" */
1989 break;
1990 case OPT_STR:
1991 case OPT_FSTR:
1992 len += 8; /* length of "<string>" */
1993 break;
1994 }
1995 if( len>max ) max = len;
1996 }
1997 for(i=0; op[i].label; i++){
1998 switch( op[i].type ){
1999 case OPT_FLAG:
2000 case OPT_FFLAG:
2001 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
2002 break;
2003 case OPT_INT:
2004 case OPT_FINT:
2005 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002006 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002007 break;
2008 case OPT_DBL:
2009 case OPT_FDBL:
2010 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002011 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002012 break;
2013 case OPT_STR:
2014 case OPT_FSTR:
2015 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002016 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002017 break;
2018 }
2019 }
2020}
2021/*********************** From the file "parse.c" ****************************/
2022/*
2023** Input file parser for the LEMON parser generator.
2024*/
2025
2026/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00002027enum e_state {
2028 INITIALIZE,
2029 WAITING_FOR_DECL_OR_RULE,
2030 WAITING_FOR_DECL_KEYWORD,
2031 WAITING_FOR_DECL_ARG,
2032 WAITING_FOR_PRECEDENCE_SYMBOL,
2033 WAITING_FOR_ARROW,
2034 IN_RHS,
2035 LHS_ALIAS_1,
2036 LHS_ALIAS_2,
2037 LHS_ALIAS_3,
2038 RHS_ALIAS_1,
2039 RHS_ALIAS_2,
2040 PRECEDENCE_MARK_1,
2041 PRECEDENCE_MARK_2,
2042 RESYNC_AFTER_RULE_ERROR,
2043 RESYNC_AFTER_DECL_ERROR,
2044 WAITING_FOR_DESTRUCTOR_SYMBOL,
2045 WAITING_FOR_DATATYPE_SYMBOL,
2046 WAITING_FOR_FALLBACK_ID,
drh61f92cd2014-01-11 03:06:18 +00002047 WAITING_FOR_WILDCARD_ID,
2048 WAITING_FOR_CLASS_ID,
2049 WAITING_FOR_CLASS_TOKEN
icculus9e44cf12010-02-14 17:14:22 +00002050};
drh75897232000-05-29 14:26:00 +00002051struct pstate {
2052 char *filename; /* Name of the input file */
2053 int tokenlineno; /* Linenumber at which current token starts */
2054 int errorcnt; /* Number of errors so far */
2055 char *tokenstart; /* Text of current token */
2056 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00002057 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00002058 struct symbol *fallback; /* The fallback token */
drh61f92cd2014-01-11 03:06:18 +00002059 struct symbol *tkclass; /* Token class symbol */
drh75897232000-05-29 14:26:00 +00002060 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00002061 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00002062 int nrhs; /* Number of right-hand side symbols seen */
2063 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00002064 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00002065 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00002066 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00002067 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00002068 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00002069 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00002070 enum e_assoc declassoc; /* Assign this association to decl arguments */
2071 int preccounter; /* Assign this precedence to decl arguments */
2072 struct rule *firstrule; /* Pointer to first rule in the grammar */
2073 struct rule *lastrule; /* Pointer to the most recently parsed rule */
2074};
2075
2076/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00002077static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00002078{
icculus9e44cf12010-02-14 17:14:22 +00002079 const char *x;
drh75897232000-05-29 14:26:00 +00002080 x = Strsafe(psp->tokenstart); /* Save the token permanently */
2081#if 0
2082 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
2083 x,psp->state);
2084#endif
2085 switch( psp->state ){
2086 case INITIALIZE:
2087 psp->prevrule = 0;
2088 psp->preccounter = 0;
2089 psp->firstrule = psp->lastrule = 0;
2090 psp->gp->nrule = 0;
2091 /* Fall thru to next case */
2092 case WAITING_FOR_DECL_OR_RULE:
2093 if( x[0]=='%' ){
2094 psp->state = WAITING_FOR_DECL_KEYWORD;
2095 }else if( islower(x[0]) ){
2096 psp->lhs = Symbol_new(x);
2097 psp->nrhs = 0;
2098 psp->lhsalias = 0;
2099 psp->state = WAITING_FOR_ARROW;
2100 }else if( x[0]=='{' ){
2101 if( psp->prevrule==0 ){
2102 ErrorMsg(psp->filename,psp->tokenlineno,
drh3cb2f6e2012-01-09 14:19:05 +00002103"There is no prior rule upon which to attach the code \
drh75897232000-05-29 14:26:00 +00002104fragment which begins on this line.");
2105 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002106 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002107 ErrorMsg(psp->filename,psp->tokenlineno,
2108"Code fragment beginning on this line is not the first \
2109to follow the previous rule.");
2110 psp->errorcnt++;
2111 }else{
2112 psp->prevrule->line = psp->tokenlineno;
2113 psp->prevrule->code = &x[1];
drhf2f105d2012-08-20 15:53:54 +00002114 }
drh75897232000-05-29 14:26:00 +00002115 }else if( x[0]=='[' ){
2116 psp->state = PRECEDENCE_MARK_1;
2117 }else{
2118 ErrorMsg(psp->filename,psp->tokenlineno,
2119 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2120 x);
2121 psp->errorcnt++;
2122 }
2123 break;
2124 case PRECEDENCE_MARK_1:
2125 if( !isupper(x[0]) ){
2126 ErrorMsg(psp->filename,psp->tokenlineno,
2127 "The precedence symbol must be a terminal.");
2128 psp->errorcnt++;
2129 }else if( psp->prevrule==0 ){
2130 ErrorMsg(psp->filename,psp->tokenlineno,
2131 "There is no prior rule to assign precedence \"[%s]\".",x);
2132 psp->errorcnt++;
2133 }else if( psp->prevrule->precsym!=0 ){
2134 ErrorMsg(psp->filename,psp->tokenlineno,
2135"Precedence mark on this line is not the first \
2136to follow the previous rule.");
2137 psp->errorcnt++;
2138 }else{
2139 psp->prevrule->precsym = Symbol_new(x);
2140 }
2141 psp->state = PRECEDENCE_MARK_2;
2142 break;
2143 case PRECEDENCE_MARK_2:
2144 if( x[0]!=']' ){
2145 ErrorMsg(psp->filename,psp->tokenlineno,
2146 "Missing \"]\" on precedence mark.");
2147 psp->errorcnt++;
2148 }
2149 psp->state = WAITING_FOR_DECL_OR_RULE;
2150 break;
2151 case WAITING_FOR_ARROW:
2152 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2153 psp->state = IN_RHS;
2154 }else if( x[0]=='(' ){
2155 psp->state = LHS_ALIAS_1;
2156 }else{
2157 ErrorMsg(psp->filename,psp->tokenlineno,
2158 "Expected to see a \":\" following the LHS symbol \"%s\".",
2159 psp->lhs->name);
2160 psp->errorcnt++;
2161 psp->state = RESYNC_AFTER_RULE_ERROR;
2162 }
2163 break;
2164 case LHS_ALIAS_1:
2165 if( isalpha(x[0]) ){
2166 psp->lhsalias = x;
2167 psp->state = LHS_ALIAS_2;
2168 }else{
2169 ErrorMsg(psp->filename,psp->tokenlineno,
2170 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2171 x,psp->lhs->name);
2172 psp->errorcnt++;
2173 psp->state = RESYNC_AFTER_RULE_ERROR;
2174 }
2175 break;
2176 case LHS_ALIAS_2:
2177 if( x[0]==')' ){
2178 psp->state = LHS_ALIAS_3;
2179 }else{
2180 ErrorMsg(psp->filename,psp->tokenlineno,
2181 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2182 psp->errorcnt++;
2183 psp->state = RESYNC_AFTER_RULE_ERROR;
2184 }
2185 break;
2186 case LHS_ALIAS_3:
2187 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2188 psp->state = IN_RHS;
2189 }else{
2190 ErrorMsg(psp->filename,psp->tokenlineno,
2191 "Missing \"->\" following: \"%s(%s)\".",
2192 psp->lhs->name,psp->lhsalias);
2193 psp->errorcnt++;
2194 psp->state = RESYNC_AFTER_RULE_ERROR;
2195 }
2196 break;
2197 case IN_RHS:
2198 if( x[0]=='.' ){
2199 struct rule *rp;
drh9892c5d2007-12-21 00:02:11 +00002200 rp = (struct rule *)calloc( sizeof(struct rule) +
2201 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002202 if( rp==0 ){
2203 ErrorMsg(psp->filename,psp->tokenlineno,
2204 "Can't allocate enough memory for this rule.");
2205 psp->errorcnt++;
2206 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002207 }else{
drh75897232000-05-29 14:26:00 +00002208 int i;
2209 rp->ruleline = psp->tokenlineno;
2210 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002211 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002212 for(i=0; i<psp->nrhs; i++){
2213 rp->rhs[i] = psp->rhs[i];
2214 rp->rhsalias[i] = psp->alias[i];
drhf2f105d2012-08-20 15:53:54 +00002215 }
drh75897232000-05-29 14:26:00 +00002216 rp->lhs = psp->lhs;
2217 rp->lhsalias = psp->lhsalias;
2218 rp->nrhs = psp->nrhs;
2219 rp->code = 0;
2220 rp->precsym = 0;
2221 rp->index = psp->gp->nrule++;
2222 rp->nextlhs = rp->lhs->rule;
2223 rp->lhs->rule = rp;
2224 rp->next = 0;
2225 if( psp->firstrule==0 ){
2226 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002227 }else{
drh75897232000-05-29 14:26:00 +00002228 psp->lastrule->next = rp;
2229 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002230 }
drh75897232000-05-29 14:26:00 +00002231 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002232 }
drh75897232000-05-29 14:26:00 +00002233 psp->state = WAITING_FOR_DECL_OR_RULE;
2234 }else if( isalpha(x[0]) ){
2235 if( psp->nrhs>=MAXRHS ){
2236 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002237 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002238 x);
2239 psp->errorcnt++;
2240 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002241 }else{
drh75897232000-05-29 14:26:00 +00002242 psp->rhs[psp->nrhs] = Symbol_new(x);
2243 psp->alias[psp->nrhs] = 0;
2244 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002245 }
drhfd405312005-11-06 04:06:59 +00002246 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2247 struct symbol *msp = psp->rhs[psp->nrhs-1];
2248 if( msp->type!=MULTITERMINAL ){
2249 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002250 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002251 memset(msp, 0, sizeof(*msp));
2252 msp->type = MULTITERMINAL;
2253 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002254 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002255 msp->subsym[0] = origsp;
2256 msp->name = origsp->name;
2257 psp->rhs[psp->nrhs-1] = msp;
2258 }
2259 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002260 msp->subsym = (struct symbol **) realloc(msp->subsym,
2261 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002262 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2263 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2264 ErrorMsg(psp->filename,psp->tokenlineno,
2265 "Cannot form a compound containing a non-terminal");
2266 psp->errorcnt++;
2267 }
drh75897232000-05-29 14:26:00 +00002268 }else if( x[0]=='(' && psp->nrhs>0 ){
2269 psp->state = RHS_ALIAS_1;
2270 }else{
2271 ErrorMsg(psp->filename,psp->tokenlineno,
2272 "Illegal character on RHS of rule: \"%s\".",x);
2273 psp->errorcnt++;
2274 psp->state = RESYNC_AFTER_RULE_ERROR;
2275 }
2276 break;
2277 case RHS_ALIAS_1:
2278 if( isalpha(x[0]) ){
2279 psp->alias[psp->nrhs-1] = x;
2280 psp->state = RHS_ALIAS_2;
2281 }else{
2282 ErrorMsg(psp->filename,psp->tokenlineno,
2283 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2284 x,psp->rhs[psp->nrhs-1]->name);
2285 psp->errorcnt++;
2286 psp->state = RESYNC_AFTER_RULE_ERROR;
2287 }
2288 break;
2289 case RHS_ALIAS_2:
2290 if( x[0]==')' ){
2291 psp->state = IN_RHS;
2292 }else{
2293 ErrorMsg(psp->filename,psp->tokenlineno,
2294 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2295 psp->errorcnt++;
2296 psp->state = RESYNC_AFTER_RULE_ERROR;
2297 }
2298 break;
2299 case WAITING_FOR_DECL_KEYWORD:
2300 if( isalpha(x[0]) ){
2301 psp->declkeyword = x;
2302 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002303 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002304 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002305 psp->state = WAITING_FOR_DECL_ARG;
2306 if( strcmp(x,"name")==0 ){
2307 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002308 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002309 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002310 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002311 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002312 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002313 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002314 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002315 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002316 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002317 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002318 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002319 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002320 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002321 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002322 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002323 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002324 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002325 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002326 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002327 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002328 }else if( strcmp(x,"extra_argument")==0 ){
2329 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002330 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002331 }else if( strcmp(x,"token_type")==0 ){
2332 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002333 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002334 }else if( strcmp(x,"default_type")==0 ){
2335 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002336 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002337 }else if( strcmp(x,"stack_size")==0 ){
2338 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002339 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002340 }else if( strcmp(x,"start_symbol")==0 ){
2341 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002342 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002343 }else if( strcmp(x,"left")==0 ){
2344 psp->preccounter++;
2345 psp->declassoc = LEFT;
2346 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2347 }else if( strcmp(x,"right")==0 ){
2348 psp->preccounter++;
2349 psp->declassoc = RIGHT;
2350 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2351 }else if( strcmp(x,"nonassoc")==0 ){
2352 psp->preccounter++;
2353 psp->declassoc = NONE;
2354 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002355 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002356 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002357 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002358 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002359 }else if( strcmp(x,"fallback")==0 ){
2360 psp->fallback = 0;
2361 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002362 }else if( strcmp(x,"wildcard")==0 ){
2363 psp->state = WAITING_FOR_WILDCARD_ID;
drh61f92cd2014-01-11 03:06:18 +00002364 }else if( strcmp(x,"token_class")==0 ){
2365 psp->state = WAITING_FOR_CLASS_ID;
drh75897232000-05-29 14:26:00 +00002366 }else{
2367 ErrorMsg(psp->filename,psp->tokenlineno,
2368 "Unknown declaration keyword: \"%%%s\".",x);
2369 psp->errorcnt++;
2370 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002371 }
drh75897232000-05-29 14:26:00 +00002372 }else{
2373 ErrorMsg(psp->filename,psp->tokenlineno,
2374 "Illegal declaration keyword: \"%s\".",x);
2375 psp->errorcnt++;
2376 psp->state = RESYNC_AFTER_DECL_ERROR;
2377 }
2378 break;
2379 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2380 if( !isalpha(x[0]) ){
2381 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002382 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002383 psp->errorcnt++;
2384 psp->state = RESYNC_AFTER_DECL_ERROR;
2385 }else{
icculusd286fa62010-03-03 17:06:32 +00002386 struct symbol *sp = Symbol_new(x);
2387 psp->declargslot = &sp->destructor;
2388 psp->decllinenoslot = &sp->destLineno;
2389 psp->insertLineMacro = 1;
2390 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002391 }
2392 break;
2393 case WAITING_FOR_DATATYPE_SYMBOL:
2394 if( !isalpha(x[0]) ){
2395 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002396 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002397 psp->errorcnt++;
2398 psp->state = RESYNC_AFTER_DECL_ERROR;
2399 }else{
icculus866bf1e2010-02-17 20:31:32 +00002400 struct symbol *sp = Symbol_find(x);
2401 if((sp) && (sp->datatype)){
2402 ErrorMsg(psp->filename,psp->tokenlineno,
2403 "Symbol %%type \"%s\" already defined", x);
2404 psp->errorcnt++;
2405 psp->state = RESYNC_AFTER_DECL_ERROR;
2406 }else{
2407 if (!sp){
2408 sp = Symbol_new(x);
2409 }
2410 psp->declargslot = &sp->datatype;
2411 psp->insertLineMacro = 0;
2412 psp->state = WAITING_FOR_DECL_ARG;
2413 }
drh75897232000-05-29 14:26:00 +00002414 }
2415 break;
2416 case WAITING_FOR_PRECEDENCE_SYMBOL:
2417 if( x[0]=='.' ){
2418 psp->state = WAITING_FOR_DECL_OR_RULE;
2419 }else if( isupper(x[0]) ){
2420 struct symbol *sp;
2421 sp = Symbol_new(x);
2422 if( sp->prec>=0 ){
2423 ErrorMsg(psp->filename,psp->tokenlineno,
2424 "Symbol \"%s\" has already be given a precedence.",x);
2425 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002426 }else{
drh75897232000-05-29 14:26:00 +00002427 sp->prec = psp->preccounter;
2428 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002429 }
drh75897232000-05-29 14:26:00 +00002430 }else{
2431 ErrorMsg(psp->filename,psp->tokenlineno,
2432 "Can't assign a precedence to \"%s\".",x);
2433 psp->errorcnt++;
2434 }
2435 break;
2436 case WAITING_FOR_DECL_ARG:
drha5808f32008-04-27 22:19:44 +00002437 if( x[0]=='{' || x[0]=='\"' || isalnum(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002438 const char *zOld, *zNew;
2439 char *zBuf, *z;
drha5808f32008-04-27 22:19:44 +00002440 int nOld, n, nLine, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002441 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002442 char zLine[50];
2443 zNew = x;
2444 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002445 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002446 if( *psp->declargslot ){
2447 zOld = *psp->declargslot;
2448 }else{
2449 zOld = "";
2450 }
drh87cf1372008-08-13 20:09:06 +00002451 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002452 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002453 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002454 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2455 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002456 for(z=psp->filename, nBack=0; *z; z++){
2457 if( *z=='\\' ) nBack++;
2458 }
drh898799f2014-01-10 23:21:00 +00002459 lemon_sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002460 nLine = lemonStrlen(zLine);
2461 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002462 }
icculus9e44cf12010-02-14 17:14:22 +00002463 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2464 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002465 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002466 if( nOld && zBuf[-1]!='\n' ){
2467 *(zBuf++) = '\n';
2468 }
2469 memcpy(zBuf, zLine, nLine);
2470 zBuf += nLine;
2471 *(zBuf++) = '"';
2472 for(z=psp->filename; *z; z++){
2473 if( *z=='\\' ){
2474 *(zBuf++) = '\\';
2475 }
2476 *(zBuf++) = *z;
2477 }
2478 *(zBuf++) = '"';
2479 *(zBuf++) = '\n';
2480 }
drh4dc8ef52008-07-01 17:13:57 +00002481 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2482 psp->decllinenoslot[0] = psp->tokenlineno;
2483 }
drha5808f32008-04-27 22:19:44 +00002484 memcpy(zBuf, zNew, nNew);
2485 zBuf += nNew;
2486 *zBuf = 0;
2487 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002488 }else{
2489 ErrorMsg(psp->filename,psp->tokenlineno,
2490 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2491 psp->errorcnt++;
2492 psp->state = RESYNC_AFTER_DECL_ERROR;
2493 }
2494 break;
drh0bd1f4e2002-06-06 18:54:39 +00002495 case WAITING_FOR_FALLBACK_ID:
2496 if( x[0]=='.' ){
2497 psp->state = WAITING_FOR_DECL_OR_RULE;
2498 }else if( !isupper(x[0]) ){
2499 ErrorMsg(psp->filename, psp->tokenlineno,
2500 "%%fallback argument \"%s\" should be a token", x);
2501 psp->errorcnt++;
2502 }else{
2503 struct symbol *sp = Symbol_new(x);
2504 if( psp->fallback==0 ){
2505 psp->fallback = sp;
2506 }else if( sp->fallback ){
2507 ErrorMsg(psp->filename, psp->tokenlineno,
2508 "More than one fallback assigned to token %s", x);
2509 psp->errorcnt++;
2510 }else{
2511 sp->fallback = psp->fallback;
2512 psp->gp->has_fallback = 1;
2513 }
2514 }
2515 break;
drhe09daa92006-06-10 13:29:31 +00002516 case WAITING_FOR_WILDCARD_ID:
2517 if( x[0]=='.' ){
2518 psp->state = WAITING_FOR_DECL_OR_RULE;
2519 }else if( !isupper(x[0]) ){
2520 ErrorMsg(psp->filename, psp->tokenlineno,
2521 "%%wildcard argument \"%s\" should be a token", x);
2522 psp->errorcnt++;
2523 }else{
2524 struct symbol *sp = Symbol_new(x);
2525 if( psp->gp->wildcard==0 ){
2526 psp->gp->wildcard = sp;
2527 }else{
2528 ErrorMsg(psp->filename, psp->tokenlineno,
2529 "Extra wildcard to token: %s", x);
2530 psp->errorcnt++;
2531 }
2532 }
2533 break;
drh61f92cd2014-01-11 03:06:18 +00002534 case WAITING_FOR_CLASS_ID:
2535 if( !islower(x[0]) ){
2536 ErrorMsg(psp->filename, psp->tokenlineno,
2537 "%%token_class must be followed by an identifier: ", x);
2538 psp->errorcnt++;
2539 psp->state = RESYNC_AFTER_DECL_ERROR;
2540 }else if( Symbol_find(x) ){
2541 ErrorMsg(psp->filename, psp->tokenlineno,
2542 "Symbol \"%s\" already used", x);
2543 psp->errorcnt++;
2544 psp->state = RESYNC_AFTER_DECL_ERROR;
2545 }else{
2546 psp->tkclass = Symbol_new(x);
2547 psp->tkclass->type = MULTITERMINAL;
2548 psp->state = WAITING_FOR_CLASS_TOKEN;
2549 }
2550 break;
2551 case WAITING_FOR_CLASS_TOKEN:
2552 if( x[0]=='.' ){
2553 psp->state = WAITING_FOR_DECL_OR_RULE;
2554 }else if( isupper(x[0]) || ((x[0]=='|' || x[0]=='/') && isupper(x[1])) ){
2555 struct symbol *msp = psp->tkclass;
2556 msp->nsubsym++;
2557 msp->subsym = (struct symbol **) realloc(msp->subsym,
2558 sizeof(struct symbol*)*msp->nsubsym);
2559 if( !isupper(x[0]) ) x++;
2560 msp->subsym[msp->nsubsym-1] = Symbol_new(x);
2561 }else{
2562 ErrorMsg(psp->filename, psp->tokenlineno,
2563 "%%token_class argument \"%s\" should be a token", x);
2564 psp->errorcnt++;
2565 psp->state = RESYNC_AFTER_DECL_ERROR;
2566 }
2567 break;
drh75897232000-05-29 14:26:00 +00002568 case RESYNC_AFTER_RULE_ERROR:
2569/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2570** break; */
2571 case RESYNC_AFTER_DECL_ERROR:
2572 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2573 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2574 break;
2575 }
2576}
2577
drh34ff57b2008-07-14 12:27:51 +00002578/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002579** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2580** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2581** comments them out. Text in between is also commented out as appropriate.
2582*/
danielk1977940fac92005-01-23 22:41:37 +00002583static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002584 int i, j, k, n;
2585 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002586 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002587 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002588 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002589 for(i=0; z[i]; i++){
2590 if( z[i]=='\n' ) lineno++;
2591 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2592 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2593 if( exclude ){
2594 exclude--;
2595 if( exclude==0 ){
2596 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2597 }
2598 }
2599 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2600 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2601 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2602 if( exclude ){
2603 exclude++;
2604 }else{
2605 for(j=i+7; isspace(z[j]); j++){}
2606 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2607 exclude = 1;
2608 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002609 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002610 exclude = 0;
2611 break;
2612 }
2613 }
2614 if( z[i+3]=='n' ) exclude = !exclude;
2615 if( exclude ){
2616 start = i;
2617 start_lineno = lineno;
2618 }
2619 }
2620 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2621 }
2622 }
2623 if( exclude ){
2624 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2625 exit(1);
2626 }
2627}
2628
drh75897232000-05-29 14:26:00 +00002629/* In spite of its name, this function is really a scanner. It read
2630** in the entire input file (all at once) then tokenizes it. Each
2631** token is passed to the function "parseonetoken" which builds all
2632** the appropriate data structures in the global state vector "gp".
2633*/
icculus9e44cf12010-02-14 17:14:22 +00002634void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002635{
2636 struct pstate ps;
2637 FILE *fp;
2638 char *filebuf;
2639 int filesize;
2640 int lineno;
2641 int c;
2642 char *cp, *nextcp;
2643 int startline = 0;
2644
rse38514a92007-09-20 11:34:17 +00002645 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002646 ps.gp = gp;
2647 ps.filename = gp->filename;
2648 ps.errorcnt = 0;
2649 ps.state = INITIALIZE;
2650
2651 /* Begin by reading the input file */
2652 fp = fopen(ps.filename,"rb");
2653 if( fp==0 ){
2654 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2655 gp->errorcnt++;
2656 return;
2657 }
2658 fseek(fp,0,2);
2659 filesize = ftell(fp);
2660 rewind(fp);
2661 filebuf = (char *)malloc( filesize+1 );
2662 if( filebuf==0 ){
2663 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2664 filesize+1);
2665 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002666 fclose(fp);
drh75897232000-05-29 14:26:00 +00002667 return;
2668 }
2669 if( fread(filebuf,1,filesize,fp)!=filesize ){
2670 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2671 filesize);
2672 free(filebuf);
2673 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002674 fclose(fp);
drh75897232000-05-29 14:26:00 +00002675 return;
2676 }
2677 fclose(fp);
2678 filebuf[filesize] = 0;
2679
drh6d08b4d2004-07-20 12:45:22 +00002680 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2681 preprocess_input(filebuf);
2682
drh75897232000-05-29 14:26:00 +00002683 /* Now scan the text of the input file */
2684 lineno = 1;
2685 for(cp=filebuf; (c= *cp)!=0; ){
2686 if( c=='\n' ) lineno++; /* Keep track of the line number */
2687 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2688 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2689 cp+=2;
2690 while( (c= *cp)!=0 && c!='\n' ) cp++;
2691 continue;
2692 }
2693 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2694 cp+=2;
2695 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2696 if( c=='\n' ) lineno++;
2697 cp++;
2698 }
2699 if( c ) cp++;
2700 continue;
2701 }
2702 ps.tokenstart = cp; /* Mark the beginning of the token */
2703 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2704 if( c=='\"' ){ /* String literals */
2705 cp++;
2706 while( (c= *cp)!=0 && c!='\"' ){
2707 if( c=='\n' ) lineno++;
2708 cp++;
2709 }
2710 if( c==0 ){
2711 ErrorMsg(ps.filename,startline,
2712"String starting on this line is not terminated before the end of the file.");
2713 ps.errorcnt++;
2714 nextcp = cp;
2715 }else{
2716 nextcp = cp+1;
2717 }
2718 }else if( c=='{' ){ /* A block of C code */
2719 int level;
2720 cp++;
2721 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2722 if( c=='\n' ) lineno++;
2723 else if( c=='{' ) level++;
2724 else if( c=='}' ) level--;
2725 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2726 int prevc;
2727 cp = &cp[2];
2728 prevc = 0;
2729 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2730 if( c=='\n' ) lineno++;
2731 prevc = c;
2732 cp++;
drhf2f105d2012-08-20 15:53:54 +00002733 }
2734 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00002735 cp = &cp[2];
2736 while( (c= *cp)!=0 && c!='\n' ) cp++;
2737 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00002738 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00002739 int startchar, prevc;
2740 startchar = c;
2741 prevc = 0;
2742 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2743 if( c=='\n' ) lineno++;
2744 if( prevc=='\\' ) prevc = 0;
2745 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00002746 }
2747 }
drh75897232000-05-29 14:26:00 +00002748 }
2749 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002750 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002751"C code starting on this line is not terminated before the end of the file.");
2752 ps.errorcnt++;
2753 nextcp = cp;
2754 }else{
2755 nextcp = cp+1;
2756 }
2757 }else if( isalnum(c) ){ /* Identifiers */
2758 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2759 nextcp = cp;
2760 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2761 cp += 3;
2762 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002763 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2764 cp += 2;
2765 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2766 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002767 }else{ /* All other (one character) operators */
2768 cp++;
2769 nextcp = cp;
2770 }
2771 c = *cp;
2772 *cp = 0; /* Null terminate the token */
2773 parseonetoken(&ps); /* Parse the token */
2774 *cp = c; /* Restore the buffer */
2775 cp = nextcp;
2776 }
2777 free(filebuf); /* Release the buffer after parsing */
2778 gp->rule = ps.firstrule;
2779 gp->errorcnt = ps.errorcnt;
2780}
2781/*************************** From the file "plink.c" *********************/
2782/*
2783** Routines processing configuration follow-set propagation links
2784** in the LEMON parser generator.
2785*/
2786static struct plink *plink_freelist = 0;
2787
2788/* Allocate a new plink */
2789struct plink *Plink_new(){
icculus9e44cf12010-02-14 17:14:22 +00002790 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00002791
2792 if( plink_freelist==0 ){
2793 int i;
2794 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002795 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002796 if( plink_freelist==0 ){
2797 fprintf(stderr,
2798 "Unable to allocate memory for a new follow-set propagation link.\n");
2799 exit(1);
2800 }
2801 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2802 plink_freelist[amt-1].next = 0;
2803 }
icculus9e44cf12010-02-14 17:14:22 +00002804 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00002805 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00002806 return newlink;
drh75897232000-05-29 14:26:00 +00002807}
2808
2809/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00002810void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002811{
icculus9e44cf12010-02-14 17:14:22 +00002812 struct plink *newlink;
2813 newlink = Plink_new();
2814 newlink->next = *plpp;
2815 *plpp = newlink;
2816 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00002817}
2818
2819/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00002820void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00002821{
2822 struct plink *nextpl;
2823 while( from ){
2824 nextpl = from->next;
2825 from->next = *to;
2826 *to = from;
2827 from = nextpl;
2828 }
2829}
2830
2831/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00002832void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00002833{
2834 struct plink *nextpl;
2835
2836 while( plp ){
2837 nextpl = plp->next;
2838 plp->next = plink_freelist;
2839 plink_freelist = plp;
2840 plp = nextpl;
2841 }
2842}
2843/*********************** From the file "report.c" **************************/
2844/*
2845** Procedures for generating reports and tables in the LEMON parser generator.
2846*/
2847
2848/* Generate a filename with the given suffix. Space to hold the
2849** name comes from malloc() and must be freed by the calling
2850** function.
2851*/
icculus9e44cf12010-02-14 17:14:22 +00002852PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00002853{
2854 char *name;
2855 char *cp;
2856
icculus9e44cf12010-02-14 17:14:22 +00002857 name = (char*)malloc( lemonStrlen(lemp->filename) + lemonStrlen(suffix) + 5 );
drh75897232000-05-29 14:26:00 +00002858 if( name==0 ){
2859 fprintf(stderr,"Can't allocate space for a filename.\n");
2860 exit(1);
2861 }
drh898799f2014-01-10 23:21:00 +00002862 lemon_strcpy(name,lemp->filename);
drh75897232000-05-29 14:26:00 +00002863 cp = strrchr(name,'.');
2864 if( cp ) *cp = 0;
drh898799f2014-01-10 23:21:00 +00002865 lemon_strcat(name,suffix);
drh75897232000-05-29 14:26:00 +00002866 return name;
2867}
2868
2869/* Open a file with a name based on the name of the input file,
2870** but with a different (specified) suffix, and return a pointer
2871** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00002872PRIVATE FILE *file_open(
2873 struct lemon *lemp,
2874 const char *suffix,
2875 const char *mode
2876){
drh75897232000-05-29 14:26:00 +00002877 FILE *fp;
2878
2879 if( lemp->outname ) free(lemp->outname);
2880 lemp->outname = file_makename(lemp, suffix);
2881 fp = fopen(lemp->outname,mode);
2882 if( fp==0 && *mode=='w' ){
2883 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2884 lemp->errorcnt++;
2885 return 0;
2886 }
2887 return fp;
2888}
2889
2890/* Duplicate the input file without comments and without actions
2891** on rules */
icculus9e44cf12010-02-14 17:14:22 +00002892void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002893{
2894 struct rule *rp;
2895 struct symbol *sp;
2896 int i, j, maxlen, len, ncolumns, skip;
2897 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2898 maxlen = 10;
2899 for(i=0; i<lemp->nsymbol; i++){
2900 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00002901 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00002902 if( len>maxlen ) maxlen = len;
2903 }
2904 ncolumns = 76/(maxlen+5);
2905 if( ncolumns<1 ) ncolumns = 1;
2906 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2907 for(i=0; i<skip; i++){
2908 printf("//");
2909 for(j=i; j<lemp->nsymbol; j+=skip){
2910 sp = lemp->symbols[j];
2911 assert( sp->index==j );
2912 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2913 }
2914 printf("\n");
2915 }
2916 for(rp=lemp->rule; rp; rp=rp->next){
2917 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002918 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002919 printf(" ::=");
2920 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002921 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00002922 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00002923 printf(" %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00002924 for(j=1; j<sp->nsubsym; j++){
2925 printf("|%s", sp->subsym[j]->name);
2926 }
drh61f92cd2014-01-11 03:06:18 +00002927 }else{
2928 printf(" %s", sp->name);
drhfd405312005-11-06 04:06:59 +00002929 }
2930 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002931 }
2932 printf(".");
2933 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002934 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002935 printf("\n");
2936 }
2937}
2938
icculus9e44cf12010-02-14 17:14:22 +00002939void ConfigPrint(FILE *fp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002940{
2941 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002942 struct symbol *sp;
2943 int i, j;
drh75897232000-05-29 14:26:00 +00002944 rp = cfp->rp;
2945 fprintf(fp,"%s ::=",rp->lhs->name);
2946 for(i=0; i<=rp->nrhs; i++){
2947 if( i==cfp->dot ) fprintf(fp," *");
2948 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002949 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00002950 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00002951 fprintf(fp," %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00002952 for(j=1; j<sp->nsubsym; j++){
2953 fprintf(fp,"|%s",sp->subsym[j]->name);
2954 }
drh61f92cd2014-01-11 03:06:18 +00002955 }else{
2956 fprintf(fp," %s", sp->name);
drhfd405312005-11-06 04:06:59 +00002957 }
drh75897232000-05-29 14:26:00 +00002958 }
2959}
2960
2961/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002962#if 0
drh75897232000-05-29 14:26:00 +00002963/* Print a set */
2964PRIVATE void SetPrint(out,set,lemp)
2965FILE *out;
2966char *set;
2967struct lemon *lemp;
2968{
2969 int i;
2970 char *spacer;
2971 spacer = "";
2972 fprintf(out,"%12s[","");
2973 for(i=0; i<lemp->nterminal; i++){
2974 if( SetFind(set,i) ){
2975 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2976 spacer = " ";
2977 }
2978 }
2979 fprintf(out,"]\n");
2980}
2981
2982/* Print a plink chain */
2983PRIVATE void PlinkPrint(out,plp,tag)
2984FILE *out;
2985struct plink *plp;
2986char *tag;
2987{
2988 while( plp ){
drhada354d2005-11-05 15:03:59 +00002989 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002990 ConfigPrint(out,plp->cfp);
2991 fprintf(out,"\n");
2992 plp = plp->next;
2993 }
2994}
2995#endif
2996
2997/* Print an action to the given file descriptor. Return FALSE if
2998** nothing was actually printed.
2999*/
3000int PrintAction(struct action *ap, FILE *fp, int indent){
3001 int result = 1;
3002 switch( ap->type ){
3003 case SHIFT:
drhada354d2005-11-05 15:03:59 +00003004 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00003005 break;
3006 case REDUCE:
3007 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
3008 break;
3009 case ACCEPT:
3010 fprintf(fp,"%*s accept",indent,ap->sp->name);
3011 break;
3012 case ERROR:
3013 fprintf(fp,"%*s error",indent,ap->sp->name);
3014 break;
drh9892c5d2007-12-21 00:02:11 +00003015 case SRCONFLICT:
3016 case RRCONFLICT:
drh75897232000-05-29 14:26:00 +00003017 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
3018 indent,ap->sp->name,ap->x.rp->index);
3019 break;
drh9892c5d2007-12-21 00:02:11 +00003020 case SSCONFLICT:
drhdd7e9db2010-07-19 01:52:07 +00003021 fprintf(fp,"%*s shift %-3d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00003022 indent,ap->sp->name,ap->x.stp->statenum);
3023 break;
drh75897232000-05-29 14:26:00 +00003024 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003025 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00003026 fprintf(fp,"%*s shift %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003027 indent,ap->sp->name,ap->x.stp->statenum);
3028 }else{
3029 result = 0;
3030 }
3031 break;
drh75897232000-05-29 14:26:00 +00003032 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003033 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00003034 fprintf(fp,"%*s reduce %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003035 indent,ap->sp->name,ap->x.rp->index);
3036 }else{
3037 result = 0;
3038 }
3039 break;
drh75897232000-05-29 14:26:00 +00003040 case NOT_USED:
3041 result = 0;
3042 break;
3043 }
3044 return result;
3045}
3046
3047/* Generate the "y.output" log file */
icculus9e44cf12010-02-14 17:14:22 +00003048void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003049{
3050 int i;
3051 struct state *stp;
3052 struct config *cfp;
3053 struct action *ap;
3054 FILE *fp;
3055
drh2aa6ca42004-09-10 00:14:04 +00003056 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00003057 if( fp==0 ) return;
drh75897232000-05-29 14:26:00 +00003058 for(i=0; i<lemp->nstate; i++){
3059 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00003060 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00003061 if( lemp->basisflag ) cfp=stp->bp;
3062 else cfp=stp->cfp;
3063 while( cfp ){
3064 char buf[20];
3065 if( cfp->dot==cfp->rp->nrhs ){
drh898799f2014-01-10 23:21:00 +00003066 lemon_sprintf(buf,"(%d)",cfp->rp->index);
drh75897232000-05-29 14:26:00 +00003067 fprintf(fp," %5s ",buf);
3068 }else{
3069 fprintf(fp," ");
3070 }
3071 ConfigPrint(fp,cfp);
3072 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00003073#if 0
drh75897232000-05-29 14:26:00 +00003074 SetPrint(fp,cfp->fws,lemp);
3075 PlinkPrint(fp,cfp->fplp,"To ");
3076 PlinkPrint(fp,cfp->bplp,"From");
3077#endif
3078 if( lemp->basisflag ) cfp=cfp->bp;
3079 else cfp=cfp->next;
3080 }
3081 fprintf(fp,"\n");
3082 for(ap=stp->ap; ap; ap=ap->next){
3083 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
3084 }
3085 fprintf(fp,"\n");
3086 }
drhe9278182007-07-18 18:16:29 +00003087 fprintf(fp, "----------------------------------------------------\n");
3088 fprintf(fp, "Symbols:\n");
3089 for(i=0; i<lemp->nsymbol; i++){
3090 int j;
3091 struct symbol *sp;
3092
3093 sp = lemp->symbols[i];
3094 fprintf(fp, " %3d: %s", i, sp->name);
3095 if( sp->type==NONTERMINAL ){
3096 fprintf(fp, ":");
3097 if( sp->lambda ){
3098 fprintf(fp, " <lambda>");
3099 }
3100 for(j=0; j<lemp->nterminal; j++){
3101 if( sp->firstset && SetFind(sp->firstset, j) ){
3102 fprintf(fp, " %s", lemp->symbols[j]->name);
3103 }
3104 }
3105 }
3106 fprintf(fp, "\n");
3107 }
drh75897232000-05-29 14:26:00 +00003108 fclose(fp);
3109 return;
3110}
3111
3112/* Search for the file "name" which is in the same directory as
3113** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00003114PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00003115{
icculus9e44cf12010-02-14 17:14:22 +00003116 const char *pathlist;
3117 char *pathbufptr;
3118 char *pathbuf;
drh75897232000-05-29 14:26:00 +00003119 char *path,*cp;
3120 char c;
drh75897232000-05-29 14:26:00 +00003121
3122#ifdef __WIN32__
3123 cp = strrchr(argv0,'\\');
3124#else
3125 cp = strrchr(argv0,'/');
3126#endif
3127 if( cp ){
3128 c = *cp;
3129 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003130 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh898799f2014-01-10 23:21:00 +00003131 if( path ) lemon_sprintf(path,"%s/%s",argv0,name);
drh75897232000-05-29 14:26:00 +00003132 *cp = c;
3133 }else{
drh75897232000-05-29 14:26:00 +00003134 pathlist = getenv("PATH");
3135 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003136 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003137 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003138 if( (pathbuf != 0) && (path!=0) ){
3139 pathbufptr = pathbuf;
drh898799f2014-01-10 23:21:00 +00003140 lemon_strcpy(pathbuf, pathlist);
icculus9e44cf12010-02-14 17:14:22 +00003141 while( *pathbuf ){
3142 cp = strchr(pathbuf,':');
3143 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003144 c = *cp;
3145 *cp = 0;
drh898799f2014-01-10 23:21:00 +00003146 lemon_sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003147 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003148 if( c==0 ) pathbuf[0] = 0;
3149 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003150 if( access(path,modemask)==0 ) break;
3151 }
icculus9e44cf12010-02-14 17:14:22 +00003152 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003153 }
3154 }
3155 return path;
3156}
3157
3158/* Given an action, compute the integer value for that action
3159** which is to be put in the action table of the generated machine.
3160** Return negative if no action should be generated.
3161*/
icculus9e44cf12010-02-14 17:14:22 +00003162PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003163{
3164 int act;
3165 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00003166 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00003167 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
3168 case ERROR: act = lemp->nstate + lemp->nrule; break;
3169 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
3170 default: act = -1; break;
3171 }
3172 return act;
3173}
3174
3175#define LINESIZE 1000
3176/* The next cluster of routines are for reading the template file
3177** and writing the results to the generated parser */
3178/* The first function transfers data from "in" to "out" until
3179** a line is seen which begins with "%%". The line number is
3180** tracked.
3181**
3182** if name!=0, then any word that begin with "Parse" is changed to
3183** begin with *name instead.
3184*/
icculus9e44cf12010-02-14 17:14:22 +00003185PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003186{
3187 int i, iStart;
3188 char line[LINESIZE];
3189 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3190 (*lineno)++;
3191 iStart = 0;
3192 if( name ){
3193 for(i=0; line[i]; i++){
3194 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
3195 && (i==0 || !isalpha(line[i-1]))
3196 ){
3197 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3198 fprintf(out,"%s",name);
3199 i += 4;
3200 iStart = i+1;
3201 }
3202 }
3203 }
3204 fprintf(out,"%s",&line[iStart]);
3205 }
3206}
3207
3208/* The next function finds the template file and opens it, returning
3209** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003210PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003211{
3212 static char templatename[] = "lempar.c";
3213 char buf[1000];
3214 FILE *in;
3215 char *tpltname;
3216 char *cp;
3217
icculus3e143bd2010-02-14 00:48:49 +00003218 /* first, see if user specified a template filename on the command line. */
3219 if (user_templatename != 0) {
3220 if( access(user_templatename,004)==-1 ){
3221 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3222 user_templatename);
3223 lemp->errorcnt++;
3224 return 0;
3225 }
3226 in = fopen(user_templatename,"rb");
3227 if( in==0 ){
3228 fprintf(stderr,"Can't open the template file \"%s\".\n",user_templatename);
3229 lemp->errorcnt++;
3230 return 0;
3231 }
3232 return in;
3233 }
3234
drh75897232000-05-29 14:26:00 +00003235 cp = strrchr(lemp->filename,'.');
3236 if( cp ){
drh898799f2014-01-10 23:21:00 +00003237 lemon_sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003238 }else{
drh898799f2014-01-10 23:21:00 +00003239 lemon_sprintf(buf,"%s.lt",lemp->filename);
drh75897232000-05-29 14:26:00 +00003240 }
3241 if( access(buf,004)==0 ){
3242 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003243 }else if( access(templatename,004)==0 ){
3244 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003245 }else{
3246 tpltname = pathsearch(lemp->argv0,templatename,0);
3247 }
3248 if( tpltname==0 ){
3249 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3250 templatename);
3251 lemp->errorcnt++;
3252 return 0;
3253 }
drh2aa6ca42004-09-10 00:14:04 +00003254 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003255 if( in==0 ){
3256 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3257 lemp->errorcnt++;
3258 return 0;
3259 }
3260 return in;
3261}
3262
drhaf805ca2004-09-07 11:28:25 +00003263/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003264PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003265{
3266 fprintf(out,"#line %d \"",lineno);
3267 while( *filename ){
3268 if( *filename == '\\' ) putc('\\',out);
3269 putc(*filename,out);
3270 filename++;
3271 }
3272 fprintf(out,"\"\n");
3273}
3274
drh75897232000-05-29 14:26:00 +00003275/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003276PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003277{
3278 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003279 while( *str ){
drh75897232000-05-29 14:26:00 +00003280 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003281 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003282 str++;
3283 }
drh9db55df2004-09-09 14:01:21 +00003284 if( str[-1]!='\n' ){
3285 putc('\n',out);
3286 (*lineno)++;
3287 }
shane58543932008-12-10 20:10:04 +00003288 if (!lemp->nolinenosflag) {
3289 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3290 }
drh75897232000-05-29 14:26:00 +00003291 return;
3292}
3293
3294/*
3295** The following routine emits code for the destructor for the
3296** symbol sp
3297*/
icculus9e44cf12010-02-14 17:14:22 +00003298void emit_destructor_code(
3299 FILE *out,
3300 struct symbol *sp,
3301 struct lemon *lemp,
3302 int *lineno
3303){
drhcc83b6e2004-04-23 23:38:42 +00003304 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003305
drh75897232000-05-29 14:26:00 +00003306 if( sp->type==TERMINAL ){
3307 cp = lemp->tokendest;
3308 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003309 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003310 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003311 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003312 fprintf(out,"{\n"); (*lineno)++;
shane58543932008-12-10 20:10:04 +00003313 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,sp->destLineno,lemp->filename); }
drh960e8c62001-04-03 16:53:21 +00003314 }else if( lemp->vardest ){
3315 cp = lemp->vardest;
3316 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003317 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003318 }else{
3319 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003320 }
3321 for(; *cp; cp++){
3322 if( *cp=='$' && cp[1]=='$' ){
3323 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3324 cp++;
3325 continue;
3326 }
shane58543932008-12-10 20:10:04 +00003327 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003328 fputc(*cp,out);
3329 }
shane58543932008-12-10 20:10:04 +00003330 fprintf(out,"\n"); (*lineno)++;
3331 if (!lemp->nolinenosflag) {
3332 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3333 }
3334 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003335 return;
3336}
3337
3338/*
drh960e8c62001-04-03 16:53:21 +00003339** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003340*/
icculus9e44cf12010-02-14 17:14:22 +00003341int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003342{
3343 int ret;
3344 if( sp->type==TERMINAL ){
3345 ret = lemp->tokendest!=0;
3346 }else{
drh960e8c62001-04-03 16:53:21 +00003347 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003348 }
3349 return ret;
3350}
3351
drh0bb132b2004-07-20 14:06:51 +00003352/*
3353** Append text to a dynamically allocated string. If zText is 0 then
3354** reset the string to be empty again. Always return the complete text
3355** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003356**
3357** n bytes of zText are stored. If n==0 then all of zText up to the first
3358** \000 terminator is stored. zText can contain up to two instances of
3359** %d. The values of p1 and p2 are written into the first and second
3360** %d.
3361**
3362** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003363*/
icculus9e44cf12010-02-14 17:14:22 +00003364PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3365 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003366 static char *z = 0;
3367 static int alloced = 0;
3368 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003369 int c;
drh0bb132b2004-07-20 14:06:51 +00003370 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003371 if( zText==0 ){
3372 used = 0;
3373 return z;
3374 }
drh7ac25c72004-08-19 15:12:26 +00003375 if( n<=0 ){
3376 if( n<0 ){
3377 used += n;
3378 assert( used>=0 );
3379 }
drh87cf1372008-08-13 20:09:06 +00003380 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003381 }
drhdf609712010-11-23 20:55:27 +00003382 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003383 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003384 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003385 }
icculus9e44cf12010-02-14 17:14:22 +00003386 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003387 while( n-- > 0 ){
3388 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003389 if( c=='%' && n>0 && zText[0]=='d' ){
drh898799f2014-01-10 23:21:00 +00003390 lemon_sprintf(zInt, "%d", p1);
drh0bb132b2004-07-20 14:06:51 +00003391 p1 = p2;
drh898799f2014-01-10 23:21:00 +00003392 lemon_strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003393 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003394 zText++;
3395 n--;
3396 }else{
3397 z[used++] = c;
3398 }
3399 }
3400 z[used] = 0;
3401 return z;
3402}
3403
3404/*
3405** zCode is a string that is the action associated with a rule. Expand
3406** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003407** stack.
drh0bb132b2004-07-20 14:06:51 +00003408*/
drhaf805ca2004-09-07 11:28:25 +00003409PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003410 char *cp, *xp;
3411 int i;
3412 char lhsused = 0; /* True if the LHS element has been used */
3413 char used[MAXRHS]; /* True for each RHS element which is used */
3414
3415 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3416 lhsused = 0;
3417
drh19c9e562007-03-29 20:13:53 +00003418 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003419 static char newlinestr[2] = { '\n', '\0' };
3420 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003421 rp->line = rp->ruleline;
3422 }
3423
drh0bb132b2004-07-20 14:06:51 +00003424 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003425
3426 /* This const cast is wrong but harmless, if we're careful. */
3427 for(cp=(char *)rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003428 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3429 char saved;
3430 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3431 saved = *xp;
3432 *xp = 0;
3433 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003434 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003435 cp = xp;
3436 lhsused = 1;
3437 }else{
3438 for(i=0; i<rp->nrhs; i++){
3439 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003440 if( cp!=rp->code && cp[-1]=='@' ){
3441 /* If the argument is of the form @X then substituted
3442 ** the token number of X, not the value of X */
3443 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3444 }else{
drhfd405312005-11-06 04:06:59 +00003445 struct symbol *sp = rp->rhs[i];
3446 int dtnum;
3447 if( sp->type==MULTITERMINAL ){
3448 dtnum = sp->subsym[0]->dtnum;
3449 }else{
3450 dtnum = sp->dtnum;
3451 }
3452 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003453 }
drh0bb132b2004-07-20 14:06:51 +00003454 cp = xp;
3455 used[i] = 1;
3456 break;
3457 }
3458 }
3459 }
3460 *xp = saved;
3461 }
3462 append_str(cp, 1, 0, 0);
3463 } /* End loop */
3464
3465 /* Check to make sure the LHS has been used */
3466 if( rp->lhsalias && !lhsused ){
3467 ErrorMsg(lemp->filename,rp->ruleline,
3468 "Label \"%s\" for \"%s(%s)\" is never used.",
3469 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3470 lemp->errorcnt++;
3471 }
3472
3473 /* Generate destructor code for RHS symbols which are not used in the
3474 ** reduce code */
3475 for(i=0; i<rp->nrhs; i++){
3476 if( rp->rhsalias[i] && !used[i] ){
3477 ErrorMsg(lemp->filename,rp->ruleline,
3478 "Label %s for \"%s(%s)\" is never used.",
3479 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3480 lemp->errorcnt++;
3481 }else if( rp->rhsalias[i]==0 ){
3482 if( has_destructor(rp->rhs[i],lemp) ){
drh639efd02008-08-13 20:04:29 +00003483 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003484 rp->rhs[i]->index,i-rp->nrhs+1);
3485 }else{
3486 /* No destructor defined for this term */
3487 }
3488 }
3489 }
drh61e339a2007-01-16 03:09:02 +00003490 if( rp->code ){
3491 cp = append_str(0,0,0,0);
3492 rp->code = Strsafe(cp?cp:"");
3493 }
drh0bb132b2004-07-20 14:06:51 +00003494}
3495
drh75897232000-05-29 14:26:00 +00003496/*
3497** Generate code which executes when the rule "rp" is reduced. Write
3498** the code to "out". Make sure lineno stays up-to-date.
3499*/
icculus9e44cf12010-02-14 17:14:22 +00003500PRIVATE void emit_code(
3501 FILE *out,
3502 struct rule *rp,
3503 struct lemon *lemp,
3504 int *lineno
3505){
3506 const char *cp;
drh75897232000-05-29 14:26:00 +00003507
3508 /* Generate code to do the reduce action */
3509 if( rp->code ){
shane58543932008-12-10 20:10:04 +00003510 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,rp->line,lemp->filename); }
drhaf805ca2004-09-07 11:28:25 +00003511 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003512 for(cp=rp->code; *cp; cp++){
shane58543932008-12-10 20:10:04 +00003513 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003514 } /* End loop */
shane58543932008-12-10 20:10:04 +00003515 fprintf(out,"}\n"); (*lineno)++;
3516 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,*lineno,lemp->outname); }
drh75897232000-05-29 14:26:00 +00003517 } /* End if( rp->code ) */
3518
drh75897232000-05-29 14:26:00 +00003519 return;
3520}
3521
3522/*
3523** Print the definition of the union used for the parser's data stack.
3524** This union contains fields for every possible data type for tokens
3525** and nonterminals. In the process of computing and printing this
3526** union, also set the ".dtnum" field of every terminal and nonterminal
3527** symbol.
3528*/
icculus9e44cf12010-02-14 17:14:22 +00003529void print_stack_union(
3530 FILE *out, /* The output stream */
3531 struct lemon *lemp, /* The main info structure for this parser */
3532 int *plineno, /* Pointer to the line number */
3533 int mhflag /* True if generating makeheaders output */
3534){
drh75897232000-05-29 14:26:00 +00003535 int lineno = *plineno; /* The line number of the output */
3536 char **types; /* A hash table of datatypes */
3537 int arraysize; /* Size of the "types" array */
3538 int maxdtlength; /* Maximum length of any ".datatype" field. */
3539 char *stddt; /* Standardized name for a datatype */
3540 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00003541 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00003542 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00003543
3544 /* Allocate and initialize types[] and allocate stddt[] */
3545 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003546 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00003547 if( types==0 ){
3548 fprintf(stderr,"Out of memory.\n");
3549 exit(1);
3550 }
drh75897232000-05-29 14:26:00 +00003551 for(i=0; i<arraysize; i++) types[i] = 0;
3552 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003553 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003554 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003555 }
drh75897232000-05-29 14:26:00 +00003556 for(i=0; i<lemp->nsymbol; i++){
3557 int len;
3558 struct symbol *sp = lemp->symbols[i];
3559 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003560 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003561 if( len>maxdtlength ) maxdtlength = len;
3562 }
3563 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00003564 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00003565 fprintf(stderr,"Out of memory.\n");
3566 exit(1);
3567 }
3568
3569 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3570 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003571 ** used for terminal symbols. If there is no %default_type defined then
3572 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3573 ** a datatype using the %type directive.
3574 */
drh75897232000-05-29 14:26:00 +00003575 for(i=0; i<lemp->nsymbol; i++){
3576 struct symbol *sp = lemp->symbols[i];
3577 char *cp;
3578 if( sp==lemp->errsym ){
3579 sp->dtnum = arraysize+1;
3580 continue;
3581 }
drh960e8c62001-04-03 16:53:21 +00003582 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003583 sp->dtnum = 0;
3584 continue;
3585 }
3586 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003587 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003588 j = 0;
3589 while( isspace(*cp) ) cp++;
3590 while( *cp ) stddt[j++] = *cp++;
3591 while( j>0 && isspace(stddt[j-1]) ) j--;
3592 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00003593 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00003594 sp->dtnum = 0;
3595 continue;
3596 }
drh75897232000-05-29 14:26:00 +00003597 hash = 0;
3598 for(j=0; stddt[j]; j++){
3599 hash = hash*53 + stddt[j];
3600 }
drh3b2129c2003-05-13 00:34:21 +00003601 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003602 while( types[hash] ){
3603 if( strcmp(types[hash],stddt)==0 ){
3604 sp->dtnum = hash + 1;
3605 break;
3606 }
3607 hash++;
drh2b51f212013-10-11 23:01:02 +00003608 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00003609 }
3610 if( types[hash]==0 ){
3611 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00003612 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00003613 if( types[hash]==0 ){
3614 fprintf(stderr,"Out of memory.\n");
3615 exit(1);
3616 }
drh898799f2014-01-10 23:21:00 +00003617 lemon_strcpy(types[hash],stddt);
drh75897232000-05-29 14:26:00 +00003618 }
3619 }
3620
3621 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3622 name = lemp->name ? lemp->name : "Parse";
3623 lineno = *plineno;
3624 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3625 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3626 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3627 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3628 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00003629 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003630 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3631 for(i=0; i<arraysize; i++){
3632 if( types[i]==0 ) continue;
3633 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3634 free(types[i]);
3635 }
drhc4dd3fd2008-01-22 01:48:05 +00003636 if( lemp->errsym->useCnt ){
3637 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3638 }
drh75897232000-05-29 14:26:00 +00003639 free(stddt);
3640 free(types);
3641 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3642 *plineno = lineno;
3643}
3644
drhb29b0a52002-02-23 19:39:46 +00003645/*
3646** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003647** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003648*/
drh8b582012003-10-21 13:16:03 +00003649static const char *minimum_size_type(int lwr, int upr){
3650 if( lwr>=0 ){
3651 if( upr<=255 ){
3652 return "unsigned char";
3653 }else if( upr<65535 ){
3654 return "unsigned short int";
3655 }else{
3656 return "unsigned int";
3657 }
3658 }else if( lwr>=-127 && upr<=127 ){
3659 return "signed char";
3660 }else if( lwr>=-32767 && upr<32767 ){
3661 return "short";
drhb29b0a52002-02-23 19:39:46 +00003662 }else{
drh8b582012003-10-21 13:16:03 +00003663 return "int";
drhb29b0a52002-02-23 19:39:46 +00003664 }
3665}
3666
drhfdbf9282003-10-21 16:34:41 +00003667/*
3668** Each state contains a set of token transaction and a set of
3669** nonterminal transactions. Each of these sets makes an instance
3670** of the following structure. An array of these structures is used
3671** to order the creation of entries in the yy_action[] table.
3672*/
3673struct axset {
3674 struct state *stp; /* A pointer to a state */
3675 int isTkn; /* True to use tokens. False for non-terminals */
3676 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00003677 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00003678};
3679
3680/*
3681** Compare to axset structures for sorting purposes
3682*/
3683static int axset_compare(const void *a, const void *b){
3684 struct axset *p1 = (struct axset*)a;
3685 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00003686 int c;
3687 c = p2->nAction - p1->nAction;
3688 if( c==0 ){
3689 c = p2->iOrder - p1->iOrder;
3690 }
3691 assert( c!=0 || p1==p2 );
3692 return c;
drhfdbf9282003-10-21 16:34:41 +00003693}
3694
drhc4dd3fd2008-01-22 01:48:05 +00003695/*
3696** Write text on "out" that describes the rule "rp".
3697*/
3698static void writeRuleText(FILE *out, struct rule *rp){
3699 int j;
3700 fprintf(out,"%s ::=", rp->lhs->name);
3701 for(j=0; j<rp->nrhs; j++){
3702 struct symbol *sp = rp->rhs[j];
drh61f92cd2014-01-11 03:06:18 +00003703 if( sp->type!=MULTITERMINAL ){
3704 fprintf(out," %s", sp->name);
3705 }else{
drhc4dd3fd2008-01-22 01:48:05 +00003706 int k;
drh61f92cd2014-01-11 03:06:18 +00003707 fprintf(out," %s", sp->subsym[0]->name);
drhc4dd3fd2008-01-22 01:48:05 +00003708 for(k=1; k<sp->nsubsym; k++){
3709 fprintf(out,"|%s",sp->subsym[k]->name);
3710 }
3711 }
3712 }
3713}
3714
3715
drh75897232000-05-29 14:26:00 +00003716/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00003717void ReportTable(
3718 struct lemon *lemp,
3719 int mhflag /* Output in makeheaders format if true */
3720){
drh75897232000-05-29 14:26:00 +00003721 FILE *out, *in;
3722 char line[LINESIZE];
3723 int lineno;
3724 struct state *stp;
3725 struct action *ap;
3726 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003727 struct acttab *pActtab;
icculusa3191192010-02-17 05:40:34 +00003728 int i, j, n;
icculus9e44cf12010-02-14 17:14:22 +00003729 const char *name;
drh8b582012003-10-21 13:16:03 +00003730 int mnTknOfst, mxTknOfst;
3731 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003732 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003733
3734 in = tplt_open(lemp);
3735 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003736 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003737 if( out==0 ){
3738 fclose(in);
3739 return;
3740 }
3741 lineno = 1;
3742 tplt_xfer(lemp->name,in,out,&lineno);
3743
3744 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00003745 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00003746 if( mhflag ){
3747 char *name = file_makename(lemp, ".h");
3748 fprintf(out,"#include \"%s\"\n", name); lineno++;
3749 free(name);
3750 }
3751 tplt_xfer(lemp->name,in,out,&lineno);
3752
3753 /* Generate #defines for all tokens */
3754 if( mhflag ){
icculus9e44cf12010-02-14 17:14:22 +00003755 const char *prefix;
drh75897232000-05-29 14:26:00 +00003756 fprintf(out,"#if INTERFACE\n"); lineno++;
3757 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3758 else prefix = "";
3759 for(i=1; i<lemp->nterminal; i++){
3760 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3761 lineno++;
3762 }
3763 fprintf(out,"#endif\n"); lineno++;
3764 }
3765 tplt_xfer(lemp->name,in,out,&lineno);
3766
3767 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003768 fprintf(out,"#define YYCODETYPE %s\n",
drh8a415d32009-06-12 13:53:51 +00003769 minimum_size_type(0, lemp->nsymbol+1)); lineno++;
drh75897232000-05-29 14:26:00 +00003770 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3771 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003772 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003773 if( lemp->wildcard ){
3774 fprintf(out,"#define YYWILDCARD %d\n",
3775 lemp->wildcard->index); lineno++;
3776 }
drh75897232000-05-29 14:26:00 +00003777 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003778 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003779 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003780 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3781 }else{
3782 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3783 }
drhca44b5a2007-02-22 23:06:58 +00003784 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003785 if( mhflag ){
3786 fprintf(out,"#if INTERFACE\n"); lineno++;
3787 }
3788 name = lemp->name ? lemp->name : "Parse";
3789 if( lemp->arg && lemp->arg[0] ){
3790 int i;
drh87cf1372008-08-13 20:09:06 +00003791 i = lemonStrlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003792 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3793 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003794 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3795 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3796 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3797 name,lemp->arg,&lemp->arg[i]); lineno++;
3798 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3799 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003800 }else{
drh1f245e42002-03-11 13:55:50 +00003801 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3802 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3803 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3804 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003805 }
3806 if( mhflag ){
3807 fprintf(out,"#endif\n"); lineno++;
3808 }
3809 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3810 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00003811 if( lemp->errsym->useCnt ){
3812 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3813 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
3814 }
drh0bd1f4e2002-06-06 18:54:39 +00003815 if( lemp->has_fallback ){
3816 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3817 }
drh75897232000-05-29 14:26:00 +00003818 tplt_xfer(lemp->name,in,out,&lineno);
3819
drh8b582012003-10-21 13:16:03 +00003820 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003821 **
drh8b582012003-10-21 13:16:03 +00003822 ** yy_action[] A single table containing all actions.
3823 ** yy_lookahead[] A table containing the lookahead for each entry in
3824 ** yy_action. Used to detect hash collisions.
3825 ** yy_shift_ofst[] For each state, the offset into yy_action for
3826 ** shifting terminals.
3827 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3828 ** shifting non-terminals after a reduce.
3829 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003830 */
drh75897232000-05-29 14:26:00 +00003831
drh8b582012003-10-21 13:16:03 +00003832 /* Compute the actions on all states and count them up */
icculus9e44cf12010-02-14 17:14:22 +00003833 ax = (struct axset *) calloc(lemp->nstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00003834 if( ax==0 ){
3835 fprintf(stderr,"malloc failed\n");
3836 exit(1);
3837 }
drh75897232000-05-29 14:26:00 +00003838 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003839 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003840 ax[i*2].stp = stp;
3841 ax[i*2].isTkn = 1;
3842 ax[i*2].nAction = stp->nTknAct;
3843 ax[i*2+1].stp = stp;
3844 ax[i*2+1].isTkn = 0;
3845 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003846 }
drh8b582012003-10-21 13:16:03 +00003847 mxTknOfst = mnTknOfst = 0;
3848 mxNtOfst = mnNtOfst = 0;
3849
drhfdbf9282003-10-21 16:34:41 +00003850 /* Compute the action table. In order to try to keep the size of the
3851 ** action table to a minimum, the heuristic of placing the largest action
3852 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003853 */
drhe594bc32009-11-03 13:02:25 +00003854 for(i=0; i<lemp->nstate*2; i++) ax[i].iOrder = i;
drhfdbf9282003-10-21 16:34:41 +00003855 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003856 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003857 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3858 stp = ax[i].stp;
3859 if( ax[i].isTkn ){
3860 for(ap=stp->ap; ap; ap=ap->next){
3861 int action;
3862 if( ap->sp->index>=lemp->nterminal ) continue;
3863 action = compute_action(lemp, ap);
3864 if( action<0 ) continue;
3865 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003866 }
drhfdbf9282003-10-21 16:34:41 +00003867 stp->iTknOfst = acttab_insert(pActtab);
3868 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3869 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3870 }else{
3871 for(ap=stp->ap; ap; ap=ap->next){
3872 int action;
3873 if( ap->sp->index<lemp->nterminal ) continue;
3874 if( ap->sp->index==lemp->nsymbol ) continue;
3875 action = compute_action(lemp, ap);
3876 if( action<0 ) continue;
3877 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003878 }
drhfdbf9282003-10-21 16:34:41 +00003879 stp->iNtOfst = acttab_insert(pActtab);
3880 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3881 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003882 }
3883 }
drhfdbf9282003-10-21 16:34:41 +00003884 free(ax);
drh8b582012003-10-21 13:16:03 +00003885
3886 /* Output the yy_action table */
drh8b582012003-10-21 13:16:03 +00003887 n = acttab_size(pActtab);
drhf16371d2009-11-03 19:18:31 +00003888 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
3889 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003890 for(i=j=0; i<n; i++){
3891 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003892 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003893 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003894 fprintf(out, " %4d,", action);
3895 if( j==9 || i==n-1 ){
3896 fprintf(out, "\n"); lineno++;
3897 j = 0;
3898 }else{
3899 j++;
3900 }
3901 }
3902 fprintf(out, "};\n"); lineno++;
3903
3904 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003905 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003906 for(i=j=0; i<n; i++){
3907 int la = acttab_yylookahead(pActtab, i);
3908 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003909 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003910 fprintf(out, " %4d,", la);
3911 if( j==9 || i==n-1 ){
3912 fprintf(out, "\n"); lineno++;
3913 j = 0;
3914 }else{
3915 j++;
3916 }
3917 }
3918 fprintf(out, "};\n"); lineno++;
3919
3920 /* Output the yy_shift_ofst[] table */
3921 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003922 n = lemp->nstate;
3923 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003924 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
3925 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
3926 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003927 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003928 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003929 for(i=j=0; i<n; i++){
3930 int ofst;
3931 stp = lemp->sorted[i];
3932 ofst = stp->iTknOfst;
3933 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003934 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003935 fprintf(out, " %4d,", ofst);
3936 if( j==9 || i==n-1 ){
3937 fprintf(out, "\n"); lineno++;
3938 j = 0;
3939 }else{
3940 j++;
3941 }
3942 }
3943 fprintf(out, "};\n"); lineno++;
3944
3945 /* Output the yy_reduce_ofst[] table */
3946 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003947 n = lemp->nstate;
3948 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003949 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
3950 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
3951 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003952 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003953 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003954 for(i=j=0; i<n; i++){
3955 int ofst;
3956 stp = lemp->sorted[i];
3957 ofst = stp->iNtOfst;
3958 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003959 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003960 fprintf(out, " %4d,", ofst);
3961 if( j==9 || i==n-1 ){
3962 fprintf(out, "\n"); lineno++;
3963 j = 0;
3964 }else{
3965 j++;
3966 }
3967 }
3968 fprintf(out, "};\n"); lineno++;
3969
3970 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003971 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003972 n = lemp->nstate;
3973 for(i=j=0; i<n; i++){
3974 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003975 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003976 fprintf(out, " %4d,", stp->iDflt);
3977 if( j==9 || i==n-1 ){
3978 fprintf(out, "\n"); lineno++;
3979 j = 0;
3980 }else{
3981 j++;
3982 }
3983 }
3984 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003985 tplt_xfer(lemp->name,in,out,&lineno);
3986
drh0bd1f4e2002-06-06 18:54:39 +00003987 /* Generate the table of fallback tokens.
3988 */
3989 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00003990 int mx = lemp->nterminal - 1;
3991 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
3992 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00003993 struct symbol *p = lemp->symbols[i];
3994 if( p->fallback==0 ){
3995 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3996 }else{
3997 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3998 p->name, p->fallback->name);
3999 }
4000 lineno++;
4001 }
4002 }
4003 tplt_xfer(lemp->name, in, out, &lineno);
4004
4005 /* Generate a table containing the symbolic name of every symbol
4006 */
drh75897232000-05-29 14:26:00 +00004007 for(i=0; i<lemp->nsymbol; i++){
drh898799f2014-01-10 23:21:00 +00004008 lemon_sprintf(line,"\"%s\",",lemp->symbols[i]->name);
drh75897232000-05-29 14:26:00 +00004009 fprintf(out," %-15s",line);
4010 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
4011 }
4012 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
4013 tplt_xfer(lemp->name,in,out,&lineno);
4014
drh0bd1f4e2002-06-06 18:54:39 +00004015 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00004016 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00004017 ** when tracing REDUCE actions.
4018 */
4019 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4020 assert( rp->index==i );
drhc4dd3fd2008-01-22 01:48:05 +00004021 fprintf(out," /* %3d */ \"", i);
4022 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00004023 fprintf(out,"\",\n"); lineno++;
4024 }
4025 tplt_xfer(lemp->name,in,out,&lineno);
4026
drh75897232000-05-29 14:26:00 +00004027 /* Generate code which executes every time a symbol is popped from
4028 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00004029 ** (In other words, generate the %destructor actions)
4030 */
drh75897232000-05-29 14:26:00 +00004031 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00004032 int once = 1;
drh75897232000-05-29 14:26:00 +00004033 for(i=0; i<lemp->nsymbol; i++){
4034 struct symbol *sp = lemp->symbols[i];
4035 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004036 if( once ){
4037 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
4038 once = 0;
4039 }
drhc53eed12009-06-12 17:46:19 +00004040 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00004041 }
4042 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
4043 if( i<lemp->nsymbol ){
4044 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4045 fprintf(out," break;\n"); lineno++;
4046 }
4047 }
drh8d659732005-01-13 23:54:06 +00004048 if( lemp->vardest ){
4049 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00004050 int once = 1;
drh8d659732005-01-13 23:54:06 +00004051 for(i=0; i<lemp->nsymbol; i++){
4052 struct symbol *sp = lemp->symbols[i];
4053 if( sp==0 || sp->type==TERMINAL ||
4054 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004055 if( once ){
4056 fprintf(out, " /* Default NON-TERMINAL Destructor */\n"); lineno++;
4057 once = 0;
4058 }
drhc53eed12009-06-12 17:46:19 +00004059 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00004060 dflt_sp = sp;
4061 }
4062 if( dflt_sp!=0 ){
4063 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00004064 }
drh4dc8ef52008-07-01 17:13:57 +00004065 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00004066 }
drh75897232000-05-29 14:26:00 +00004067 for(i=0; i<lemp->nsymbol; i++){
4068 struct symbol *sp = lemp->symbols[i];
4069 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh75013012009-06-12 15:47:34 +00004070 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004071
4072 /* Combine duplicate destructors into a single case */
4073 for(j=i+1; j<lemp->nsymbol; j++){
4074 struct symbol *sp2 = lemp->symbols[j];
4075 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
4076 && sp2->dtnum==sp->dtnum
4077 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00004078 fprintf(out," case %d: /* %s */\n",
4079 sp2->index, sp2->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004080 sp2->destructor = 0;
4081 }
4082 }
4083
drh75897232000-05-29 14:26:00 +00004084 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4085 fprintf(out," break;\n"); lineno++;
4086 }
drh75897232000-05-29 14:26:00 +00004087 tplt_xfer(lemp->name,in,out,&lineno);
4088
4089 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00004090 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00004091 tplt_xfer(lemp->name,in,out,&lineno);
4092
4093 /* Generate the table of rule information
4094 **
4095 ** Note: This code depends on the fact that rules are number
4096 ** sequentually beginning with 0.
4097 */
4098 for(rp=lemp->rule; rp; rp=rp->next){
4099 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
4100 }
4101 tplt_xfer(lemp->name,in,out,&lineno);
4102
4103 /* Generate code which execution during each REDUCE action */
4104 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00004105 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00004106 }
drhc53eed12009-06-12 17:46:19 +00004107 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00004108 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00004109 struct rule *rp2; /* Other rules with the same action */
drh0bb132b2004-07-20 14:06:51 +00004110 if( rp->code==0 ) continue;
drhc53eed12009-06-12 17:46:19 +00004111 if( rp->code[0]=='\n' && rp->code[1]==0 ) continue; /* Will be default: */
drhc4dd3fd2008-01-22 01:48:05 +00004112 fprintf(out," case %d: /* ", rp->index);
4113 writeRuleText(out, rp);
4114 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004115 for(rp2=rp->next; rp2; rp2=rp2->next){
4116 if( rp2->code==rp->code ){
drhc4dd3fd2008-01-22 01:48:05 +00004117 fprintf(out," case %d: /* ", rp2->index);
4118 writeRuleText(out, rp2);
drh8a415d32009-06-12 13:53:51 +00004119 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004120 rp2->code = 0;
4121 }
4122 }
drh75897232000-05-29 14:26:00 +00004123 emit_code(out,rp,lemp,&lineno);
4124 fprintf(out," break;\n"); lineno++;
drhc53eed12009-06-12 17:46:19 +00004125 rp->code = 0;
drh75897232000-05-29 14:26:00 +00004126 }
drhc53eed12009-06-12 17:46:19 +00004127 /* Finally, output the default: rule. We choose as the default: all
4128 ** empty actions. */
4129 fprintf(out," default:\n"); lineno++;
4130 for(rp=lemp->rule; rp; rp=rp->next){
4131 if( rp->code==0 ) continue;
4132 assert( rp->code[0]=='\n' && rp->code[1]==0 );
4133 fprintf(out," /* (%d) ", rp->index);
4134 writeRuleText(out, rp);
4135 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->index); lineno++;
4136 }
4137 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004138 tplt_xfer(lemp->name,in,out,&lineno);
4139
4140 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004141 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004142 tplt_xfer(lemp->name,in,out,&lineno);
4143
4144 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004145 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004146 tplt_xfer(lemp->name,in,out,&lineno);
4147
4148 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004149 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004150 tplt_xfer(lemp->name,in,out,&lineno);
4151
4152 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004153 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004154
4155 fclose(in);
4156 fclose(out);
4157 return;
4158}
4159
4160/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004161void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004162{
4163 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004164 const char *prefix;
drh75897232000-05-29 14:26:00 +00004165 char line[LINESIZE];
4166 char pattern[LINESIZE];
4167 int i;
4168
4169 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4170 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004171 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004172 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004173 int nextChar;
drh75897232000-05-29 14:26:00 +00004174 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
drh61f92cd2014-01-11 03:06:18 +00004175 lemon_sprintf(pattern,"#define %s%-30s %3d\n",
4176 prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004177 if( strcmp(line,pattern) ) break;
4178 }
drh8ba0d1c2012-06-16 15:26:31 +00004179 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004180 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004181 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004182 /* No change in the file. Don't rewrite it. */
4183 return;
4184 }
4185 }
drh2aa6ca42004-09-10 00:14:04 +00004186 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004187 if( out ){
4188 for(i=1; i<lemp->nterminal; i++){
drh61f92cd2014-01-11 03:06:18 +00004189 fprintf(out,"#define %s%-30s %3d\n",prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004190 }
4191 fclose(out);
4192 }
4193 return;
4194}
4195
4196/* Reduce the size of the action tables, if possible, by making use
4197** of defaults.
4198**
drhb59499c2002-02-23 18:45:13 +00004199** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004200** it the default. Except, there is no default if the wildcard token
4201** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004202*/
icculus9e44cf12010-02-14 17:14:22 +00004203void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004204{
4205 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00004206 struct action *ap, *ap2;
4207 struct rule *rp, *rp2, *rbest;
4208 int nbest, n;
drh75897232000-05-29 14:26:00 +00004209 int i;
drhe09daa92006-06-10 13:29:31 +00004210 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004211
4212 for(i=0; i<lemp->nstate; i++){
4213 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004214 nbest = 0;
4215 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004216 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004217
drhb59499c2002-02-23 18:45:13 +00004218 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004219 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4220 usesWildcard = 1;
4221 }
drhb59499c2002-02-23 18:45:13 +00004222 if( ap->type!=REDUCE ) continue;
4223 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004224 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004225 if( rp==rbest ) continue;
4226 n = 1;
4227 for(ap2=ap->next; ap2; ap2=ap2->next){
4228 if( ap2->type!=REDUCE ) continue;
4229 rp2 = ap2->x.rp;
4230 if( rp2==rbest ) continue;
4231 if( rp2==rp ) n++;
4232 }
4233 if( n>nbest ){
4234 nbest = n;
4235 rbest = rp;
drh75897232000-05-29 14:26:00 +00004236 }
4237 }
drhb59499c2002-02-23 18:45:13 +00004238
4239 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004240 ** is not at least 1 or if the wildcard token is a possible
4241 ** lookahead.
4242 */
4243 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004244
drhb59499c2002-02-23 18:45:13 +00004245
4246 /* Combine matching REDUCE actions into a single default */
4247 for(ap=stp->ap; ap; ap=ap->next){
4248 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4249 }
drh75897232000-05-29 14:26:00 +00004250 assert( ap );
4251 ap->sp = Symbol_new("{default}");
4252 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004253 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004254 }
4255 stp->ap = Action_sort(stp->ap);
4256 }
4257}
drhb59499c2002-02-23 18:45:13 +00004258
drhada354d2005-11-05 15:03:59 +00004259
4260/*
4261** Compare two states for sorting purposes. The smaller state is the
4262** one with the most non-terminal actions. If they have the same number
4263** of non-terminal actions, then the smaller is the one with the most
4264** token actions.
4265*/
4266static int stateResortCompare(const void *a, const void *b){
4267 const struct state *pA = *(const struct state**)a;
4268 const struct state *pB = *(const struct state**)b;
4269 int n;
4270
4271 n = pB->nNtAct - pA->nNtAct;
4272 if( n==0 ){
4273 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004274 if( n==0 ){
4275 n = pB->statenum - pA->statenum;
4276 }
drhada354d2005-11-05 15:03:59 +00004277 }
drhe594bc32009-11-03 13:02:25 +00004278 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004279 return n;
4280}
4281
4282
4283/*
4284** Renumber and resort states so that states with fewer choices
4285** occur at the end. Except, keep state 0 as the first state.
4286*/
icculus9e44cf12010-02-14 17:14:22 +00004287void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00004288{
4289 int i;
4290 struct state *stp;
4291 struct action *ap;
4292
4293 for(i=0; i<lemp->nstate; i++){
4294 stp = lemp->sorted[i];
4295 stp->nTknAct = stp->nNtAct = 0;
4296 stp->iDflt = lemp->nstate + lemp->nrule;
4297 stp->iTknOfst = NO_OFFSET;
4298 stp->iNtOfst = NO_OFFSET;
4299 for(ap=stp->ap; ap; ap=ap->next){
4300 if( compute_action(lemp,ap)>=0 ){
4301 if( ap->sp->index<lemp->nterminal ){
4302 stp->nTknAct++;
4303 }else if( ap->sp->index<lemp->nsymbol ){
4304 stp->nNtAct++;
4305 }else{
4306 stp->iDflt = compute_action(lemp, ap);
4307 }
4308 }
4309 }
4310 }
4311 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4312 stateResortCompare);
4313 for(i=0; i<lemp->nstate; i++){
4314 lemp->sorted[i]->statenum = i;
4315 }
4316}
4317
4318
drh75897232000-05-29 14:26:00 +00004319/***************** From the file "set.c" ************************************/
4320/*
4321** Set manipulation routines for the LEMON parser generator.
4322*/
4323
4324static int size = 0;
4325
4326/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00004327void SetSize(int n)
drh75897232000-05-29 14:26:00 +00004328{
4329 size = n+1;
4330}
4331
4332/* Allocate a new set */
4333char *SetNew(){
4334 char *s;
drh9892c5d2007-12-21 00:02:11 +00004335 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004336 if( s==0 ){
4337 extern void memory_error();
4338 memory_error();
4339 }
drh75897232000-05-29 14:26:00 +00004340 return s;
4341}
4342
4343/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00004344void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00004345{
4346 free(s);
4347}
4348
4349/* Add a new element to the set. Return TRUE if the element was added
4350** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00004351int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00004352{
4353 int rv;
drh9892c5d2007-12-21 00:02:11 +00004354 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004355 rv = s[e];
4356 s[e] = 1;
4357 return !rv;
4358}
4359
4360/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00004361int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00004362{
4363 int i, progress;
4364 progress = 0;
4365 for(i=0; i<size; i++){
4366 if( s2[i]==0 ) continue;
4367 if( s1[i]==0 ){
4368 progress = 1;
4369 s1[i] = 1;
4370 }
4371 }
4372 return progress;
4373}
4374/********************** From the file "table.c" ****************************/
4375/*
4376** All code in this file has been automatically generated
4377** from a specification in the file
4378** "table.q"
4379** by the associative array code building program "aagen".
4380** Do not edit this file! Instead, edit the specification
4381** file, then rerun aagen.
4382*/
4383/*
4384** Code for processing tables in the LEMON parser generator.
4385*/
4386
drh01f75f22013-10-02 20:46:30 +00004387PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00004388{
drh01f75f22013-10-02 20:46:30 +00004389 unsigned h = 0;
4390 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00004391 return h;
4392}
4393
4394/* Works like strdup, sort of. Save a string in malloced memory, but
4395** keep strings in a table so that the same string is not in more
4396** than one place.
4397*/
icculus9e44cf12010-02-14 17:14:22 +00004398const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00004399{
icculus9e44cf12010-02-14 17:14:22 +00004400 const char *z;
4401 char *cpy;
drh75897232000-05-29 14:26:00 +00004402
drh916f75f2006-07-17 00:19:39 +00004403 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004404 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00004405 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
drh898799f2014-01-10 23:21:00 +00004406 lemon_strcpy(cpy,y);
icculus9e44cf12010-02-14 17:14:22 +00004407 z = cpy;
drh75897232000-05-29 14:26:00 +00004408 Strsafe_insert(z);
4409 }
4410 MemoryCheck(z);
4411 return z;
4412}
4413
4414/* There is one instance of the following structure for each
4415** associative array of type "x1".
4416*/
4417struct s_x1 {
4418 int size; /* The number of available slots. */
4419 /* Must be a power of 2 greater than or */
4420 /* equal to 1 */
4421 int count; /* Number of currently slots filled */
4422 struct s_x1node *tbl; /* The data stored here */
4423 struct s_x1node **ht; /* Hash table for lookups */
4424};
4425
4426/* There is one instance of this structure for every data element
4427** in an associative array of type "x1".
4428*/
4429typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00004430 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00004431 struct s_x1node *next; /* Next entry with the same hash */
4432 struct s_x1node **from; /* Previous link */
4433} x1node;
4434
4435/* There is only one instance of the array, which is the following */
4436static struct s_x1 *x1a;
4437
4438/* Allocate a new associative array */
4439void Strsafe_init(){
4440 if( x1a ) return;
4441 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4442 if( x1a ){
4443 x1a->size = 1024;
4444 x1a->count = 0;
4445 x1a->tbl = (x1node*)malloc(
4446 (sizeof(x1node) + sizeof(x1node*))*1024 );
4447 if( x1a->tbl==0 ){
4448 free(x1a);
4449 x1a = 0;
4450 }else{
4451 int i;
4452 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4453 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4454 }
4455 }
4456}
4457/* Insert a new record into the array. Return TRUE if successful.
4458** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004459int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00004460{
4461 x1node *np;
drh01f75f22013-10-02 20:46:30 +00004462 unsigned h;
4463 unsigned ph;
drh75897232000-05-29 14:26:00 +00004464
4465 if( x1a==0 ) return 0;
4466 ph = strhash(data);
4467 h = ph & (x1a->size-1);
4468 np = x1a->ht[h];
4469 while( np ){
4470 if( strcmp(np->data,data)==0 ){
4471 /* An existing entry with the same key is found. */
4472 /* Fail because overwrite is not allows. */
4473 return 0;
4474 }
4475 np = np->next;
4476 }
4477 if( x1a->count>=x1a->size ){
4478 /* Need to make the hash table bigger */
4479 int i,size;
4480 struct s_x1 array;
4481 array.size = size = x1a->size*2;
4482 array.count = x1a->count;
4483 array.tbl = (x1node*)malloc(
4484 (sizeof(x1node) + sizeof(x1node*))*size );
4485 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4486 array.ht = (x1node**)&(array.tbl[size]);
4487 for(i=0; i<size; i++) array.ht[i] = 0;
4488 for(i=0; i<x1a->count; i++){
4489 x1node *oldnp, *newnp;
4490 oldnp = &(x1a->tbl[i]);
4491 h = strhash(oldnp->data) & (size-1);
4492 newnp = &(array.tbl[i]);
4493 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4494 newnp->next = array.ht[h];
4495 newnp->data = oldnp->data;
4496 newnp->from = &(array.ht[h]);
4497 array.ht[h] = newnp;
4498 }
4499 free(x1a->tbl);
4500 *x1a = array;
4501 }
4502 /* Insert the new data */
4503 h = ph & (x1a->size-1);
4504 np = &(x1a->tbl[x1a->count++]);
4505 np->data = data;
4506 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4507 np->next = x1a->ht[h];
4508 x1a->ht[h] = np;
4509 np->from = &(x1a->ht[h]);
4510 return 1;
4511}
4512
4513/* Return a pointer to data assigned to the given key. Return NULL
4514** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004515const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00004516{
drh01f75f22013-10-02 20:46:30 +00004517 unsigned h;
drh75897232000-05-29 14:26:00 +00004518 x1node *np;
4519
4520 if( x1a==0 ) return 0;
4521 h = strhash(key) & (x1a->size-1);
4522 np = x1a->ht[h];
4523 while( np ){
4524 if( strcmp(np->data,key)==0 ) break;
4525 np = np->next;
4526 }
4527 return np ? np->data : 0;
4528}
4529
4530/* Return a pointer to the (terminal or nonterminal) symbol "x".
4531** Create a new symbol if this is the first time "x" has been seen.
4532*/
icculus9e44cf12010-02-14 17:14:22 +00004533struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00004534{
4535 struct symbol *sp;
4536
4537 sp = Symbol_find(x);
4538 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00004539 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00004540 MemoryCheck(sp);
4541 sp->name = Strsafe(x);
4542 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4543 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004544 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004545 sp->prec = -1;
4546 sp->assoc = UNK;
4547 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00004548 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00004549 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00004550 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00004551 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00004552 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00004553 Symbol_insert(sp,sp->name);
4554 }
drhc4dd3fd2008-01-22 01:48:05 +00004555 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00004556 return sp;
4557}
4558
drh61f92cd2014-01-11 03:06:18 +00004559/* Compare two symbols for sorting purposes. Return negative,
4560** zero, or positive if a is less then, equal to, or greater
4561** than b.
drh60d31652004-02-22 00:08:04 +00004562**
4563** Symbols that begin with upper case letters (terminals or tokens)
4564** must sort before symbols that begin with lower case letters
drh61f92cd2014-01-11 03:06:18 +00004565** (non-terminals). And MULTITERMINAL symbols (created using the
4566** %token_class directive) must sort at the very end. Other than
4567** that, the order does not matter.
drh60d31652004-02-22 00:08:04 +00004568**
4569** We find experimentally that leaving the symbols in their original
4570** order (the order they appeared in the grammar file) gives the
4571** smallest parser tables in SQLite.
4572*/
icculus9e44cf12010-02-14 17:14:22 +00004573int Symbolcmpp(const void *_a, const void *_b)
4574{
drh61f92cd2014-01-11 03:06:18 +00004575 const struct symbol *a = *(const struct symbol **) _a;
4576 const struct symbol *b = *(const struct symbol **) _b;
4577 int i1 = a->type==MULTITERMINAL ? 3 : a->name[0]>'Z' ? 2 : 1;
4578 int i2 = b->type==MULTITERMINAL ? 3 : b->name[0]>'Z' ? 2 : 1;
4579 return i1==i2 ? a->index - b->index : i1 - i2;
drh75897232000-05-29 14:26:00 +00004580}
4581
4582/* There is one instance of the following structure for each
4583** associative array of type "x2".
4584*/
4585struct s_x2 {
4586 int size; /* The number of available slots. */
4587 /* Must be a power of 2 greater than or */
4588 /* equal to 1 */
4589 int count; /* Number of currently slots filled */
4590 struct s_x2node *tbl; /* The data stored here */
4591 struct s_x2node **ht; /* Hash table for lookups */
4592};
4593
4594/* There is one instance of this structure for every data element
4595** in an associative array of type "x2".
4596*/
4597typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00004598 struct symbol *data; /* The data */
4599 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00004600 struct s_x2node *next; /* Next entry with the same hash */
4601 struct s_x2node **from; /* Previous link */
4602} x2node;
4603
4604/* There is only one instance of the array, which is the following */
4605static struct s_x2 *x2a;
4606
4607/* Allocate a new associative array */
4608void Symbol_init(){
4609 if( x2a ) return;
4610 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4611 if( x2a ){
4612 x2a->size = 128;
4613 x2a->count = 0;
4614 x2a->tbl = (x2node*)malloc(
4615 (sizeof(x2node) + sizeof(x2node*))*128 );
4616 if( x2a->tbl==0 ){
4617 free(x2a);
4618 x2a = 0;
4619 }else{
4620 int i;
4621 x2a->ht = (x2node**)&(x2a->tbl[128]);
4622 for(i=0; i<128; i++) x2a->ht[i] = 0;
4623 }
4624 }
4625}
4626/* Insert a new record into the array. Return TRUE if successful.
4627** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004628int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00004629{
4630 x2node *np;
drh01f75f22013-10-02 20:46:30 +00004631 unsigned h;
4632 unsigned ph;
drh75897232000-05-29 14:26:00 +00004633
4634 if( x2a==0 ) return 0;
4635 ph = strhash(key);
4636 h = ph & (x2a->size-1);
4637 np = x2a->ht[h];
4638 while( np ){
4639 if( strcmp(np->key,key)==0 ){
4640 /* An existing entry with the same key is found. */
4641 /* Fail because overwrite is not allows. */
4642 return 0;
4643 }
4644 np = np->next;
4645 }
4646 if( x2a->count>=x2a->size ){
4647 /* Need to make the hash table bigger */
4648 int i,size;
4649 struct s_x2 array;
4650 array.size = size = x2a->size*2;
4651 array.count = x2a->count;
4652 array.tbl = (x2node*)malloc(
4653 (sizeof(x2node) + sizeof(x2node*))*size );
4654 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4655 array.ht = (x2node**)&(array.tbl[size]);
4656 for(i=0; i<size; i++) array.ht[i] = 0;
4657 for(i=0; i<x2a->count; i++){
4658 x2node *oldnp, *newnp;
4659 oldnp = &(x2a->tbl[i]);
4660 h = strhash(oldnp->key) & (size-1);
4661 newnp = &(array.tbl[i]);
4662 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4663 newnp->next = array.ht[h];
4664 newnp->key = oldnp->key;
4665 newnp->data = oldnp->data;
4666 newnp->from = &(array.ht[h]);
4667 array.ht[h] = newnp;
4668 }
4669 free(x2a->tbl);
4670 *x2a = array;
4671 }
4672 /* Insert the new data */
4673 h = ph & (x2a->size-1);
4674 np = &(x2a->tbl[x2a->count++]);
4675 np->key = key;
4676 np->data = data;
4677 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4678 np->next = x2a->ht[h];
4679 x2a->ht[h] = np;
4680 np->from = &(x2a->ht[h]);
4681 return 1;
4682}
4683
4684/* Return a pointer to data assigned to the given key. Return NULL
4685** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004686struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00004687{
drh01f75f22013-10-02 20:46:30 +00004688 unsigned h;
drh75897232000-05-29 14:26:00 +00004689 x2node *np;
4690
4691 if( x2a==0 ) return 0;
4692 h = strhash(key) & (x2a->size-1);
4693 np = x2a->ht[h];
4694 while( np ){
4695 if( strcmp(np->key,key)==0 ) break;
4696 np = np->next;
4697 }
4698 return np ? np->data : 0;
4699}
4700
4701/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00004702struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00004703{
4704 struct symbol *data;
4705 if( x2a && n>0 && n<=x2a->count ){
4706 data = x2a->tbl[n-1].data;
4707 }else{
4708 data = 0;
4709 }
4710 return data;
4711}
4712
4713/* Return the size of the array */
4714int Symbol_count()
4715{
4716 return x2a ? x2a->count : 0;
4717}
4718
4719/* Return an array of pointers to all data in the table.
4720** The array is obtained from malloc. Return NULL if memory allocation
4721** problems, or if the array is empty. */
4722struct symbol **Symbol_arrayof()
4723{
4724 struct symbol **array;
4725 int i,size;
4726 if( x2a==0 ) return 0;
4727 size = x2a->count;
drh9892c5d2007-12-21 00:02:11 +00004728 array = (struct symbol **)calloc(size, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00004729 if( array ){
4730 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4731 }
4732 return array;
4733}
4734
4735/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00004736int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00004737{
icculus9e44cf12010-02-14 17:14:22 +00004738 const struct config *a = (struct config *) _a;
4739 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00004740 int x;
4741 x = a->rp->index - b->rp->index;
4742 if( x==0 ) x = a->dot - b->dot;
4743 return x;
4744}
4745
4746/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00004747PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00004748{
4749 int rc;
4750 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4751 rc = a->rp->index - b->rp->index;
4752 if( rc==0 ) rc = a->dot - b->dot;
4753 }
4754 if( rc==0 ){
4755 if( a ) rc = 1;
4756 if( b ) rc = -1;
4757 }
4758 return rc;
4759}
4760
4761/* Hash a state */
drh01f75f22013-10-02 20:46:30 +00004762PRIVATE unsigned statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00004763{
drh01f75f22013-10-02 20:46:30 +00004764 unsigned h=0;
drh75897232000-05-29 14:26:00 +00004765 while( a ){
4766 h = h*571 + a->rp->index*37 + a->dot;
4767 a = a->bp;
4768 }
4769 return h;
4770}
4771
4772/* Allocate a new state structure */
4773struct state *State_new()
4774{
icculus9e44cf12010-02-14 17:14:22 +00004775 struct state *newstate;
4776 newstate = (struct state *)calloc(1, sizeof(struct state) );
4777 MemoryCheck(newstate);
4778 return newstate;
drh75897232000-05-29 14:26:00 +00004779}
4780
4781/* There is one instance of the following structure for each
4782** associative array of type "x3".
4783*/
4784struct s_x3 {
4785 int size; /* The number of available slots. */
4786 /* Must be a power of 2 greater than or */
4787 /* equal to 1 */
4788 int count; /* Number of currently slots filled */
4789 struct s_x3node *tbl; /* The data stored here */
4790 struct s_x3node **ht; /* Hash table for lookups */
4791};
4792
4793/* There is one instance of this structure for every data element
4794** in an associative array of type "x3".
4795*/
4796typedef struct s_x3node {
4797 struct state *data; /* The data */
4798 struct config *key; /* The key */
4799 struct s_x3node *next; /* Next entry with the same hash */
4800 struct s_x3node **from; /* Previous link */
4801} x3node;
4802
4803/* There is only one instance of the array, which is the following */
4804static struct s_x3 *x3a;
4805
4806/* Allocate a new associative array */
4807void State_init(){
4808 if( x3a ) return;
4809 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4810 if( x3a ){
4811 x3a->size = 128;
4812 x3a->count = 0;
4813 x3a->tbl = (x3node*)malloc(
4814 (sizeof(x3node) + sizeof(x3node*))*128 );
4815 if( x3a->tbl==0 ){
4816 free(x3a);
4817 x3a = 0;
4818 }else{
4819 int i;
4820 x3a->ht = (x3node**)&(x3a->tbl[128]);
4821 for(i=0; i<128; i++) x3a->ht[i] = 0;
4822 }
4823 }
4824}
4825/* Insert a new record into the array. Return TRUE if successful.
4826** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004827int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00004828{
4829 x3node *np;
drh01f75f22013-10-02 20:46:30 +00004830 unsigned h;
4831 unsigned ph;
drh75897232000-05-29 14:26:00 +00004832
4833 if( x3a==0 ) return 0;
4834 ph = statehash(key);
4835 h = ph & (x3a->size-1);
4836 np = x3a->ht[h];
4837 while( np ){
4838 if( statecmp(np->key,key)==0 ){
4839 /* An existing entry with the same key is found. */
4840 /* Fail because overwrite is not allows. */
4841 return 0;
4842 }
4843 np = np->next;
4844 }
4845 if( x3a->count>=x3a->size ){
4846 /* Need to make the hash table bigger */
4847 int i,size;
4848 struct s_x3 array;
4849 array.size = size = x3a->size*2;
4850 array.count = x3a->count;
4851 array.tbl = (x3node*)malloc(
4852 (sizeof(x3node) + sizeof(x3node*))*size );
4853 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4854 array.ht = (x3node**)&(array.tbl[size]);
4855 for(i=0; i<size; i++) array.ht[i] = 0;
4856 for(i=0; i<x3a->count; i++){
4857 x3node *oldnp, *newnp;
4858 oldnp = &(x3a->tbl[i]);
4859 h = statehash(oldnp->key) & (size-1);
4860 newnp = &(array.tbl[i]);
4861 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4862 newnp->next = array.ht[h];
4863 newnp->key = oldnp->key;
4864 newnp->data = oldnp->data;
4865 newnp->from = &(array.ht[h]);
4866 array.ht[h] = newnp;
4867 }
4868 free(x3a->tbl);
4869 *x3a = array;
4870 }
4871 /* Insert the new data */
4872 h = ph & (x3a->size-1);
4873 np = &(x3a->tbl[x3a->count++]);
4874 np->key = key;
4875 np->data = data;
4876 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4877 np->next = x3a->ht[h];
4878 x3a->ht[h] = np;
4879 np->from = &(x3a->ht[h]);
4880 return 1;
4881}
4882
4883/* Return a pointer to data assigned to the given key. Return NULL
4884** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004885struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004886{
drh01f75f22013-10-02 20:46:30 +00004887 unsigned h;
drh75897232000-05-29 14:26:00 +00004888 x3node *np;
4889
4890 if( x3a==0 ) return 0;
4891 h = statehash(key) & (x3a->size-1);
4892 np = x3a->ht[h];
4893 while( np ){
4894 if( statecmp(np->key,key)==0 ) break;
4895 np = np->next;
4896 }
4897 return np ? np->data : 0;
4898}
4899
4900/* Return an array of pointers to all data in the table.
4901** The array is obtained from malloc. Return NULL if memory allocation
4902** problems, or if the array is empty. */
4903struct state **State_arrayof()
4904{
4905 struct state **array;
4906 int i,size;
4907 if( x3a==0 ) return 0;
4908 size = x3a->count;
4909 array = (struct state **)malloc( sizeof(struct state *)*size );
4910 if( array ){
4911 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4912 }
4913 return array;
4914}
4915
4916/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00004917PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00004918{
drh01f75f22013-10-02 20:46:30 +00004919 unsigned h=0;
drh75897232000-05-29 14:26:00 +00004920 h = h*571 + a->rp->index*37 + a->dot;
4921 return h;
4922}
4923
4924/* There is one instance of the following structure for each
4925** associative array of type "x4".
4926*/
4927struct s_x4 {
4928 int size; /* The number of available slots. */
4929 /* Must be a power of 2 greater than or */
4930 /* equal to 1 */
4931 int count; /* Number of currently slots filled */
4932 struct s_x4node *tbl; /* The data stored here */
4933 struct s_x4node **ht; /* Hash table for lookups */
4934};
4935
4936/* There is one instance of this structure for every data element
4937** in an associative array of type "x4".
4938*/
4939typedef struct s_x4node {
4940 struct config *data; /* The data */
4941 struct s_x4node *next; /* Next entry with the same hash */
4942 struct s_x4node **from; /* Previous link */
4943} x4node;
4944
4945/* There is only one instance of the array, which is the following */
4946static struct s_x4 *x4a;
4947
4948/* Allocate a new associative array */
4949void Configtable_init(){
4950 if( x4a ) return;
4951 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4952 if( x4a ){
4953 x4a->size = 64;
4954 x4a->count = 0;
4955 x4a->tbl = (x4node*)malloc(
4956 (sizeof(x4node) + sizeof(x4node*))*64 );
4957 if( x4a->tbl==0 ){
4958 free(x4a);
4959 x4a = 0;
4960 }else{
4961 int i;
4962 x4a->ht = (x4node**)&(x4a->tbl[64]);
4963 for(i=0; i<64; i++) x4a->ht[i] = 0;
4964 }
4965 }
4966}
4967/* Insert a new record into the array. Return TRUE if successful.
4968** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004969int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00004970{
4971 x4node *np;
drh01f75f22013-10-02 20:46:30 +00004972 unsigned h;
4973 unsigned ph;
drh75897232000-05-29 14:26:00 +00004974
4975 if( x4a==0 ) return 0;
4976 ph = confighash(data);
4977 h = ph & (x4a->size-1);
4978 np = x4a->ht[h];
4979 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004980 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00004981 /* An existing entry with the same key is found. */
4982 /* Fail because overwrite is not allows. */
4983 return 0;
4984 }
4985 np = np->next;
4986 }
4987 if( x4a->count>=x4a->size ){
4988 /* Need to make the hash table bigger */
4989 int i,size;
4990 struct s_x4 array;
4991 array.size = size = x4a->size*2;
4992 array.count = x4a->count;
4993 array.tbl = (x4node*)malloc(
4994 (sizeof(x4node) + sizeof(x4node*))*size );
4995 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4996 array.ht = (x4node**)&(array.tbl[size]);
4997 for(i=0; i<size; i++) array.ht[i] = 0;
4998 for(i=0; i<x4a->count; i++){
4999 x4node *oldnp, *newnp;
5000 oldnp = &(x4a->tbl[i]);
5001 h = confighash(oldnp->data) & (size-1);
5002 newnp = &(array.tbl[i]);
5003 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5004 newnp->next = array.ht[h];
5005 newnp->data = oldnp->data;
5006 newnp->from = &(array.ht[h]);
5007 array.ht[h] = newnp;
5008 }
5009 free(x4a->tbl);
5010 *x4a = array;
5011 }
5012 /* Insert the new data */
5013 h = ph & (x4a->size-1);
5014 np = &(x4a->tbl[x4a->count++]);
5015 np->data = data;
5016 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
5017 np->next = x4a->ht[h];
5018 x4a->ht[h] = np;
5019 np->from = &(x4a->ht[h]);
5020 return 1;
5021}
5022
5023/* Return a pointer to data assigned to the given key. Return NULL
5024** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005025struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005026{
5027 int h;
5028 x4node *np;
5029
5030 if( x4a==0 ) return 0;
5031 h = confighash(key) & (x4a->size-1);
5032 np = x4a->ht[h];
5033 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005034 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00005035 np = np->next;
5036 }
5037 return np ? np->data : 0;
5038}
5039
5040/* Remove all data from the table. Pass each data to the function "f"
5041** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00005042void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00005043{
5044 int i;
5045 if( x4a==0 || x4a->count==0 ) return;
5046 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
5047 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
5048 x4a->count = 0;
5049 return;
5050}