blob: 4ada425d6931d8af2df8eba63a053a66d3ac8cda [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++){}
drhecaa9d32014-01-11 03:27:37 +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;
drhecaa9d32014-01-11 03:27:37 +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){
mistachkin7a429652014-01-14 10:17:21 +000086 int i, j, k, c;
drh898799f2014-01-10 23:21:00 +000087 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 );
drh62a223e2014-06-09 13:11:40 +00001186 apx->type = ERROR;
drh75897232000-05-29 14:26:00 +00001187 }
1188 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1189 spx = apx->x.rp->precsym;
1190 spy = apy->x.rp->precsym;
1191 if( spx==0 || spy==0 || spx->prec<0 ||
1192 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001193 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001194 errcnt++;
1195 }else if( spx->prec>spy->prec ){
1196 apy->type = RD_RESOLVED;
1197 }else if( spx->prec<spy->prec ){
1198 apx->type = RD_RESOLVED;
1199 }
1200 }else{
drhb59499c2002-02-23 18:45:13 +00001201 assert(
1202 apx->type==SH_RESOLVED ||
1203 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001204 apx->type==SSCONFLICT ||
1205 apx->type==SRCONFLICT ||
1206 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001207 apy->type==SH_RESOLVED ||
1208 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001209 apy->type==SSCONFLICT ||
1210 apy->type==SRCONFLICT ||
1211 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001212 );
1213 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1214 ** REDUCEs on the list. If we reach this point it must be because
1215 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001216 }
1217 return errcnt;
1218}
1219/********************* From the file "configlist.c" *************************/
1220/*
1221** Routines to processing a configuration list and building a state
1222** in the LEMON parser generator.
1223*/
1224
1225static struct config *freelist = 0; /* List of free configurations */
1226static struct config *current = 0; /* Top of list of configurations */
1227static struct config **currentend = 0; /* Last on list of configs */
1228static struct config *basis = 0; /* Top of list of basis configs */
1229static struct config **basisend = 0; /* End of list of basis configs */
1230
1231/* Return a pointer to a new configuration */
1232PRIVATE struct config *newconfig(){
icculus9e44cf12010-02-14 17:14:22 +00001233 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001234 if( freelist==0 ){
1235 int i;
1236 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001237 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001238 if( freelist==0 ){
1239 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1240 exit(1);
1241 }
1242 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1243 freelist[amt-1].next = 0;
1244 }
icculus9e44cf12010-02-14 17:14:22 +00001245 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001246 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001247 return newcfg;
drh75897232000-05-29 14:26:00 +00001248}
1249
1250/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001251PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001252{
1253 old->next = freelist;
1254 freelist = old;
1255}
1256
1257/* Initialized the configuration list builder */
1258void Configlist_init(){
1259 current = 0;
1260 currentend = &current;
1261 basis = 0;
1262 basisend = &basis;
1263 Configtable_init();
1264 return;
1265}
1266
1267/* Initialized the configuration list builder */
1268void Configlist_reset(){
1269 current = 0;
1270 currentend = &current;
1271 basis = 0;
1272 basisend = &basis;
1273 Configtable_clear(0);
1274 return;
1275}
1276
1277/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001278struct config *Configlist_add(
1279 struct rule *rp, /* The rule */
1280 int dot /* Index into the RHS of the rule where the dot goes */
1281){
drh75897232000-05-29 14:26:00 +00001282 struct config *cfp, model;
1283
1284 assert( currentend!=0 );
1285 model.rp = rp;
1286 model.dot = dot;
1287 cfp = Configtable_find(&model);
1288 if( cfp==0 ){
1289 cfp = newconfig();
1290 cfp->rp = rp;
1291 cfp->dot = dot;
1292 cfp->fws = SetNew();
1293 cfp->stp = 0;
1294 cfp->fplp = cfp->bplp = 0;
1295 cfp->next = 0;
1296 cfp->bp = 0;
1297 *currentend = cfp;
1298 currentend = &cfp->next;
1299 Configtable_insert(cfp);
1300 }
1301 return cfp;
1302}
1303
1304/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001305struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001306{
1307 struct config *cfp, model;
1308
1309 assert( basisend!=0 );
1310 assert( currentend!=0 );
1311 model.rp = rp;
1312 model.dot = dot;
1313 cfp = Configtable_find(&model);
1314 if( cfp==0 ){
1315 cfp = newconfig();
1316 cfp->rp = rp;
1317 cfp->dot = dot;
1318 cfp->fws = SetNew();
1319 cfp->stp = 0;
1320 cfp->fplp = cfp->bplp = 0;
1321 cfp->next = 0;
1322 cfp->bp = 0;
1323 *currentend = cfp;
1324 currentend = &cfp->next;
1325 *basisend = cfp;
1326 basisend = &cfp->bp;
1327 Configtable_insert(cfp);
1328 }
1329 return cfp;
1330}
1331
1332/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001333void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001334{
1335 struct config *cfp, *newcfp;
1336 struct rule *rp, *newrp;
1337 struct symbol *sp, *xsp;
1338 int i, dot;
1339
1340 assert( currentend!=0 );
1341 for(cfp=current; cfp; cfp=cfp->next){
1342 rp = cfp->rp;
1343 dot = cfp->dot;
1344 if( dot>=rp->nrhs ) continue;
1345 sp = rp->rhs[dot];
1346 if( sp->type==NONTERMINAL ){
1347 if( sp->rule==0 && sp!=lemp->errsym ){
1348 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1349 sp->name);
1350 lemp->errorcnt++;
1351 }
1352 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1353 newcfp = Configlist_add(newrp,0);
1354 for(i=dot+1; i<rp->nrhs; i++){
1355 xsp = rp->rhs[i];
1356 if( xsp->type==TERMINAL ){
1357 SetAdd(newcfp->fws,xsp->index);
1358 break;
drhfd405312005-11-06 04:06:59 +00001359 }else if( xsp->type==MULTITERMINAL ){
1360 int k;
1361 for(k=0; k<xsp->nsubsym; k++){
1362 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1363 }
1364 break;
drhf2f105d2012-08-20 15:53:54 +00001365 }else{
drh75897232000-05-29 14:26:00 +00001366 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001367 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001368 }
1369 }
drh75897232000-05-29 14:26:00 +00001370 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1371 }
1372 }
1373 }
1374 return;
1375}
1376
1377/* Sort the configuration list */
1378void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001379 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001380 currentend = 0;
1381 return;
1382}
1383
1384/* Sort the basis configuration list */
1385void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001386 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001387 basisend = 0;
1388 return;
1389}
1390
1391/* Return a pointer to the head of the configuration list and
1392** reset the list */
1393struct config *Configlist_return(){
1394 struct config *old;
1395 old = current;
1396 current = 0;
1397 currentend = 0;
1398 return old;
1399}
1400
1401/* Return a pointer to the head of the configuration list and
1402** reset the list */
1403struct config *Configlist_basis(){
1404 struct config *old;
1405 old = basis;
1406 basis = 0;
1407 basisend = 0;
1408 return old;
1409}
1410
1411/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001412void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001413{
1414 struct config *nextcfp;
1415 for(; cfp; cfp=nextcfp){
1416 nextcfp = cfp->next;
1417 assert( cfp->fplp==0 );
1418 assert( cfp->bplp==0 );
1419 if( cfp->fws ) SetFree(cfp->fws);
1420 deleteconfig(cfp);
1421 }
1422 return;
1423}
1424/***************** From the file "error.c" *********************************/
1425/*
1426** Code for printing error message.
1427*/
1428
drhf9a2e7b2003-04-15 01:49:48 +00001429void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001430 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001431 fprintf(stderr, "%s:%d: ", filename, lineno);
1432 va_start(ap, format);
1433 vfprintf(stderr,format,ap);
1434 va_end(ap);
1435 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001436}
1437/**************** From the file "main.c" ************************************/
1438/*
1439** Main program file for the LEMON parser generator.
1440*/
1441
1442/* Report an out-of-memory condition and abort. This function
1443** is used mostly by the "MemoryCheck" macro in struct.h
1444*/
1445void memory_error(){
1446 fprintf(stderr,"Out of memory. Aborting...\n");
1447 exit(1);
1448}
1449
drh6d08b4d2004-07-20 12:45:22 +00001450static int nDefine = 0; /* Number of -D options on the command line */
1451static char **azDefine = 0; /* Name of the -D macros */
1452
1453/* This routine is called with the argument to each -D command-line option.
1454** Add the macro defined to the azDefine array.
1455*/
1456static void handle_D_option(char *z){
1457 char **paz;
1458 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001459 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001460 if( azDefine==0 ){
1461 fprintf(stderr,"out of memory\n");
1462 exit(1);
1463 }
1464 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001465 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001466 if( *paz==0 ){
1467 fprintf(stderr,"out of memory\n");
1468 exit(1);
1469 }
drh898799f2014-01-10 23:21:00 +00001470 lemon_strcpy(*paz, z);
drh6d08b4d2004-07-20 12:45:22 +00001471 for(z=*paz; *z && *z!='='; z++){}
1472 *z = 0;
1473}
1474
icculus3e143bd2010-02-14 00:48:49 +00001475static char *user_templatename = NULL;
1476static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001477 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001478 if( user_templatename==0 ){
1479 memory_error();
1480 }
drh898799f2014-01-10 23:21:00 +00001481 lemon_strcpy(user_templatename, z);
icculus3e143bd2010-02-14 00:48:49 +00001482}
drh75897232000-05-29 14:26:00 +00001483
1484/* The main program. Parse the command line and do it... */
icculus9e44cf12010-02-14 17:14:22 +00001485int main(int argc, char **argv)
drh75897232000-05-29 14:26:00 +00001486{
1487 static int version = 0;
1488 static int rpflag = 0;
1489 static int basisflag = 0;
1490 static int compress = 0;
1491 static int quiet = 0;
1492 static int statistics = 0;
1493 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001494 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001495 static int noResort = 0;
drh75897232000-05-29 14:26:00 +00001496 static struct s_options options[] = {
1497 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1498 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001499 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
drh0325d392015-01-01 19:11:22 +00001500 {OPT_FSTR, "f", 0, "Ignored. (Placeholder for -f compiler options.)"},
drh75897232000-05-29 14:26:00 +00001501 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
drh0325d392015-01-01 19:11:22 +00001502 {OPT_FSTR, "I", 0, "Ignored. (Placeholder for '-I' compiler options.)"},
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."},
drh0325d392015-01-01 19:11:22 +00001505 {OPT_FSTR, "O", 0, "Ignored. (Placeholder for '-O' compiler options.)"},
drhf5c4e0f2010-07-18 11:35:53 +00001506 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1507 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001508 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001509 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001510 {OPT_FLAG, "s", (char*)&statistics,
1511 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001512 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
drh0325d392015-01-01 19:11:22 +00001513 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
1514 {OPT_FSTR, "W", 0, "Ignored. (Placeholder for '-W' compiler options.)"},
drh75897232000-05-29 14:26:00 +00001515 {OPT_FLAG,0,0,0}
1516 };
1517 int i;
icculus42585cf2010-02-14 05:19:56 +00001518 int exitcode;
drh75897232000-05-29 14:26:00 +00001519 struct lemon lem;
1520
drhb0c86772000-06-02 23:21:26 +00001521 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001522 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001523 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001524 exit(0);
1525 }
drhb0c86772000-06-02 23:21:26 +00001526 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001527 fprintf(stderr,"Exactly one filename argument is required.\n");
1528 exit(1);
1529 }
drh954f6b42006-06-13 13:27:46 +00001530 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001531 lem.errorcnt = 0;
1532
1533 /* Initialize the machine */
1534 Strsafe_init();
1535 Symbol_init();
1536 State_init();
1537 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001538 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001539 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001540 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001541 Symbol_new("$");
1542 lem.errsym = Symbol_new("error");
drhc4dd3fd2008-01-22 01:48:05 +00001543 lem.errsym->useCnt = 0;
drh75897232000-05-29 14:26:00 +00001544
1545 /* Parse the input file */
1546 Parse(&lem);
1547 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001548 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001549 fprintf(stderr,"Empty grammar.\n");
1550 exit(1);
1551 }
1552
1553 /* Count and index the symbols of the grammar */
drh75897232000-05-29 14:26:00 +00001554 Symbol_new("{default}");
drh61f92cd2014-01-11 03:06:18 +00001555 lem.nsymbol = Symbol_count();
drh75897232000-05-29 14:26:00 +00001556 lem.symbols = Symbol_arrayof();
drh61f92cd2014-01-11 03:06:18 +00001557 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1558 qsort(lem.symbols,lem.nsymbol,sizeof(struct symbol*), Symbolcmpp);
1559 for(i=0; i<lem.nsymbol; i++) lem.symbols[i]->index = i;
1560 while( lem.symbols[i-1]->type==MULTITERMINAL ){ i--; }
1561 assert( strcmp(lem.symbols[i-1]->name,"{default}")==0 );
1562 lem.nsymbol = i - 1;
drh75897232000-05-29 14:26:00 +00001563 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1564 lem.nterminal = i;
1565
1566 /* Generate a reprint of the grammar, if requested on the command line */
1567 if( rpflag ){
1568 Reprint(&lem);
1569 }else{
1570 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001571 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001572
1573 /* Find the precedence for every production rule (that has one) */
1574 FindRulePrecedences(&lem);
1575
1576 /* Compute the lambda-nonterminals and the first-sets for every
1577 ** nonterminal */
1578 FindFirstSets(&lem);
1579
1580 /* Compute all LR(0) states. Also record follow-set propagation
1581 ** links so that the follow-set can be computed later */
1582 lem.nstate = 0;
1583 FindStates(&lem);
1584 lem.sorted = State_arrayof();
1585
1586 /* Tie up loose ends on the propagation links */
1587 FindLinks(&lem);
1588
1589 /* Compute the follow set of every reducible configuration */
1590 FindFollowSets(&lem);
1591
1592 /* Compute the action tables */
1593 FindActions(&lem);
1594
1595 /* Compress the action tables */
1596 if( compress==0 ) CompressTables(&lem);
1597
drhada354d2005-11-05 15:03:59 +00001598 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001599 ** occur at the end. This is an optimization that helps make the
1600 ** generated parser tables smaller. */
1601 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001602
drh75897232000-05-29 14:26:00 +00001603 /* Generate a report of the parser generated. (the "y.output" file) */
1604 if( !quiet ) ReportOutput(&lem);
1605
1606 /* Generate the source code for the parser */
1607 ReportTable(&lem, mhflag);
1608
1609 /* Produce a header file for use by the scanner. (This step is
1610 ** omitted if the "-m" option is used because makeheaders will
1611 ** generate the file for us.) */
1612 if( !mhflag ) ReportHeader(&lem);
1613 }
1614 if( statistics ){
1615 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1616 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1617 printf(" %d states, %d parser table entries, %d conflicts\n",
1618 lem.nstate, lem.tablesize, lem.nconflict);
1619 }
icculus8e158022010-02-16 16:09:03 +00001620 if( lem.nconflict > 0 ){
1621 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001622 }
1623
1624 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001625 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001626 exit(exitcode);
1627 return (exitcode);
drh75897232000-05-29 14:26:00 +00001628}
1629/******************** From the file "msort.c" *******************************/
1630/*
1631** A generic merge-sort program.
1632**
1633** USAGE:
1634** Let "ptr" be a pointer to some structure which is at the head of
1635** a null-terminated list. Then to sort the list call:
1636**
1637** ptr = msort(ptr,&(ptr->next),cmpfnc);
1638**
1639** In the above, "cmpfnc" is a pointer to a function which compares
1640** two instances of the structure and returns an integer, as in
1641** strcmp. The second argument is a pointer to the pointer to the
1642** second element of the linked list. This address is used to compute
1643** the offset to the "next" field within the structure. The offset to
1644** the "next" field must be constant for all structures in the list.
1645**
1646** The function returns a new pointer which is the head of the list
1647** after sorting.
1648**
1649** ALGORITHM:
1650** Merge-sort.
1651*/
1652
1653/*
1654** Return a pointer to the next structure in the linked list.
1655*/
drhd25d6922012-04-18 09:59:56 +00001656#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001657
1658/*
1659** Inputs:
1660** a: A sorted, null-terminated linked list. (May be null).
1661** b: A sorted, null-terminated linked list. (May be null).
1662** cmp: A pointer to the comparison function.
1663** offset: Offset in the structure to the "next" field.
1664**
1665** Return Value:
1666** A pointer to the head of a sorted list containing the elements
1667** of both a and b.
1668**
1669** Side effects:
1670** The "next" pointers for elements in the lists a and b are
1671** changed.
1672*/
drhe9278182007-07-18 18:16:29 +00001673static char *merge(
1674 char *a,
1675 char *b,
1676 int (*cmp)(const char*,const char*),
1677 int offset
1678){
drh75897232000-05-29 14:26:00 +00001679 char *ptr, *head;
1680
1681 if( a==0 ){
1682 head = b;
1683 }else if( b==0 ){
1684 head = a;
1685 }else{
drhe594bc32009-11-03 13:02:25 +00001686 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001687 ptr = a;
1688 a = NEXT(a);
1689 }else{
1690 ptr = b;
1691 b = NEXT(b);
1692 }
1693 head = ptr;
1694 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001695 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001696 NEXT(ptr) = a;
1697 ptr = a;
1698 a = NEXT(a);
1699 }else{
1700 NEXT(ptr) = b;
1701 ptr = b;
1702 b = NEXT(b);
1703 }
1704 }
1705 if( a ) NEXT(ptr) = a;
1706 else NEXT(ptr) = b;
1707 }
1708 return head;
1709}
1710
1711/*
1712** Inputs:
1713** list: Pointer to a singly-linked list of structures.
1714** next: Pointer to pointer to the second element of the list.
1715** cmp: A comparison function.
1716**
1717** Return Value:
1718** A pointer to the head of a sorted list containing the elements
1719** orginally in list.
1720**
1721** Side effects:
1722** The "next" pointers for elements in list are changed.
1723*/
1724#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001725static char *msort(
1726 char *list,
1727 char **next,
1728 int (*cmp)(const char*,const char*)
1729){
drhba99af52001-10-25 20:37:16 +00001730 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001731 char *ep;
1732 char *set[LISTSIZE];
1733 int i;
drhba99af52001-10-25 20:37:16 +00001734 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001735 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1736 while( list ){
1737 ep = list;
1738 list = NEXT(list);
1739 NEXT(ep) = 0;
1740 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1741 ep = merge(ep,set[i],cmp,offset);
1742 set[i] = 0;
1743 }
1744 set[i] = ep;
1745 }
1746 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001747 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001748 return ep;
1749}
1750/************************ From the file "option.c" **************************/
1751static char **argv;
1752static struct s_options *op;
1753static FILE *errstream;
1754
1755#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1756
1757/*
1758** Print the command line with a carrot pointing to the k-th character
1759** of the n-th field.
1760*/
icculus9e44cf12010-02-14 17:14:22 +00001761static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001762{
1763 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001764 if( argv[0] ) fprintf(err,"%s",argv[0]);
drh87cf1372008-08-13 20:09:06 +00001765 spcnt = lemonStrlen(argv[0]) + 1;
drh75897232000-05-29 14:26:00 +00001766 for(i=1; i<n && argv[i]; i++){
1767 fprintf(err," %s",argv[i]);
drh87cf1372008-08-13 20:09:06 +00001768 spcnt += lemonStrlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001769 }
1770 spcnt += k;
1771 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1772 if( spcnt<20 ){
1773 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1774 }else{
1775 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1776 }
1777}
1778
1779/*
1780** Return the index of the N-th non-switch argument. Return -1
1781** if N is out of range.
1782*/
icculus9e44cf12010-02-14 17:14:22 +00001783static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001784{
1785 int i;
1786 int dashdash = 0;
1787 if( argv!=0 && *argv!=0 ){
1788 for(i=1; argv[i]; i++){
1789 if( dashdash || !ISOPT(argv[i]) ){
1790 if( n==0 ) return i;
1791 n--;
1792 }
1793 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1794 }
1795 }
1796 return -1;
1797}
1798
1799static char emsg[] = "Command line syntax error: ";
1800
1801/*
1802** Process a flag command line argument.
1803*/
icculus9e44cf12010-02-14 17:14:22 +00001804static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001805{
1806 int v;
1807 int errcnt = 0;
1808 int j;
1809 for(j=0; op[j].label; j++){
drh87cf1372008-08-13 20:09:06 +00001810 if( strncmp(&argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001811 }
1812 v = argv[i][0]=='-' ? 1 : 0;
1813 if( op[j].label==0 ){
1814 if( err ){
1815 fprintf(err,"%sundefined option.\n",emsg);
1816 errline(i,1,err);
1817 }
1818 errcnt++;
drh0325d392015-01-01 19:11:22 +00001819 }else if( op[j].arg==0 ){
1820 /* Ignore this option */
drh75897232000-05-29 14:26:00 +00001821 }else if( op[j].type==OPT_FLAG ){
1822 *((int*)op[j].arg) = v;
1823 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001824 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001825 }else if( op[j].type==OPT_FSTR ){
icculus9e44cf12010-02-14 17:14:22 +00001826 (*(void(*)(char *))(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001827 }else{
1828 if( err ){
1829 fprintf(err,"%smissing argument on switch.\n",emsg);
1830 errline(i,1,err);
1831 }
1832 errcnt++;
1833 }
1834 return errcnt;
1835}
1836
1837/*
1838** Process a command line switch which has an argument.
1839*/
icculus9e44cf12010-02-14 17:14:22 +00001840static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001841{
1842 int lv = 0;
1843 double dv = 0.0;
1844 char *sv = 0, *end;
1845 char *cp;
1846 int j;
1847 int errcnt = 0;
1848 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001849 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001850 *cp = 0;
1851 for(j=0; op[j].label; j++){
1852 if( strcmp(argv[i],op[j].label)==0 ) break;
1853 }
1854 *cp = '=';
1855 if( op[j].label==0 ){
1856 if( err ){
1857 fprintf(err,"%sundefined option.\n",emsg);
1858 errline(i,0,err);
1859 }
1860 errcnt++;
1861 }else{
1862 cp++;
1863 switch( op[j].type ){
1864 case OPT_FLAG:
1865 case OPT_FFLAG:
1866 if( err ){
1867 fprintf(err,"%soption requires an argument.\n",emsg);
1868 errline(i,0,err);
1869 }
1870 errcnt++;
1871 break;
1872 case OPT_DBL:
1873 case OPT_FDBL:
1874 dv = strtod(cp,&end);
1875 if( *end ){
1876 if( err ){
1877 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001878 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001879 }
1880 errcnt++;
1881 }
1882 break;
1883 case OPT_INT:
1884 case OPT_FINT:
1885 lv = strtol(cp,&end,0);
1886 if( *end ){
1887 if( err ){
1888 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001889 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001890 }
1891 errcnt++;
1892 }
1893 break;
1894 case OPT_STR:
1895 case OPT_FSTR:
1896 sv = cp;
1897 break;
1898 }
1899 switch( op[j].type ){
1900 case OPT_FLAG:
1901 case OPT_FFLAG:
1902 break;
1903 case OPT_DBL:
1904 *(double*)(op[j].arg) = dv;
1905 break;
1906 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00001907 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00001908 break;
1909 case OPT_INT:
1910 *(int*)(op[j].arg) = lv;
1911 break;
1912 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00001913 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00001914 break;
1915 case OPT_STR:
1916 *(char**)(op[j].arg) = sv;
1917 break;
1918 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00001919 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00001920 break;
1921 }
1922 }
1923 return errcnt;
1924}
1925
icculus9e44cf12010-02-14 17:14:22 +00001926int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00001927{
1928 int errcnt = 0;
1929 argv = a;
1930 op = o;
1931 errstream = err;
1932 if( argv && *argv && op ){
1933 int i;
1934 for(i=1; argv[i]; i++){
1935 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1936 errcnt += handleflags(i,err);
1937 }else if( strchr(argv[i],'=') ){
1938 errcnt += handleswitch(i,err);
1939 }
1940 }
1941 }
1942 if( errcnt>0 ){
1943 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001944 OptPrint();
drh75897232000-05-29 14:26:00 +00001945 exit(1);
1946 }
1947 return 0;
1948}
1949
drhb0c86772000-06-02 23:21:26 +00001950int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001951 int cnt = 0;
1952 int dashdash = 0;
1953 int i;
1954 if( argv!=0 && argv[0]!=0 ){
1955 for(i=1; argv[i]; i++){
1956 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1957 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1958 }
1959 }
1960 return cnt;
1961}
1962
icculus9e44cf12010-02-14 17:14:22 +00001963char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00001964{
1965 int i;
1966 i = argindex(n);
1967 return i>=0 ? argv[i] : 0;
1968}
1969
icculus9e44cf12010-02-14 17:14:22 +00001970void OptErr(int n)
drh75897232000-05-29 14:26:00 +00001971{
1972 int i;
1973 i = argindex(n);
1974 if( i>=0 ) errline(i,0,errstream);
1975}
1976
drhb0c86772000-06-02 23:21:26 +00001977void OptPrint(){
drh75897232000-05-29 14:26:00 +00001978 int i;
1979 int max, len;
1980 max = 0;
1981 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00001982 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00001983 switch( op[i].type ){
1984 case OPT_FLAG:
1985 case OPT_FFLAG:
1986 break;
1987 case OPT_INT:
1988 case OPT_FINT:
1989 len += 9; /* length of "<integer>" */
1990 break;
1991 case OPT_DBL:
1992 case OPT_FDBL:
1993 len += 6; /* length of "<real>" */
1994 break;
1995 case OPT_STR:
1996 case OPT_FSTR:
1997 len += 8; /* length of "<string>" */
1998 break;
1999 }
2000 if( len>max ) max = len;
2001 }
2002 for(i=0; op[i].label; i++){
2003 switch( op[i].type ){
2004 case OPT_FLAG:
2005 case OPT_FFLAG:
2006 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
2007 break;
2008 case OPT_INT:
2009 case OPT_FINT:
drh0325d392015-01-01 19:11:22 +00002010 fprintf(errstream," -%s<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002011 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002012 break;
2013 case OPT_DBL:
2014 case OPT_FDBL:
drh0325d392015-01-01 19:11:22 +00002015 fprintf(errstream," -%s<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002016 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002017 break;
2018 case OPT_STR:
2019 case OPT_FSTR:
drh0325d392015-01-01 19:11:22 +00002020 fprintf(errstream," -%s<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00002021 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00002022 break;
2023 }
2024 }
2025}
2026/*********************** From the file "parse.c" ****************************/
2027/*
2028** Input file parser for the LEMON parser generator.
2029*/
2030
2031/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00002032enum e_state {
2033 INITIALIZE,
2034 WAITING_FOR_DECL_OR_RULE,
2035 WAITING_FOR_DECL_KEYWORD,
2036 WAITING_FOR_DECL_ARG,
2037 WAITING_FOR_PRECEDENCE_SYMBOL,
2038 WAITING_FOR_ARROW,
2039 IN_RHS,
2040 LHS_ALIAS_1,
2041 LHS_ALIAS_2,
2042 LHS_ALIAS_3,
2043 RHS_ALIAS_1,
2044 RHS_ALIAS_2,
2045 PRECEDENCE_MARK_1,
2046 PRECEDENCE_MARK_2,
2047 RESYNC_AFTER_RULE_ERROR,
2048 RESYNC_AFTER_DECL_ERROR,
2049 WAITING_FOR_DESTRUCTOR_SYMBOL,
2050 WAITING_FOR_DATATYPE_SYMBOL,
2051 WAITING_FOR_FALLBACK_ID,
drh61f92cd2014-01-11 03:06:18 +00002052 WAITING_FOR_WILDCARD_ID,
2053 WAITING_FOR_CLASS_ID,
2054 WAITING_FOR_CLASS_TOKEN
icculus9e44cf12010-02-14 17:14:22 +00002055};
drh75897232000-05-29 14:26:00 +00002056struct pstate {
2057 char *filename; /* Name of the input file */
2058 int tokenlineno; /* Linenumber at which current token starts */
2059 int errorcnt; /* Number of errors so far */
2060 char *tokenstart; /* Text of current token */
2061 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00002062 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00002063 struct symbol *fallback; /* The fallback token */
drh61f92cd2014-01-11 03:06:18 +00002064 struct symbol *tkclass; /* Token class symbol */
drh75897232000-05-29 14:26:00 +00002065 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00002066 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00002067 int nrhs; /* Number of right-hand side symbols seen */
2068 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00002069 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00002070 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00002071 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00002072 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00002073 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00002074 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00002075 enum e_assoc declassoc; /* Assign this association to decl arguments */
2076 int preccounter; /* Assign this precedence to decl arguments */
2077 struct rule *firstrule; /* Pointer to first rule in the grammar */
2078 struct rule *lastrule; /* Pointer to the most recently parsed rule */
2079};
2080
2081/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00002082static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00002083{
icculus9e44cf12010-02-14 17:14:22 +00002084 const char *x;
drh75897232000-05-29 14:26:00 +00002085 x = Strsafe(psp->tokenstart); /* Save the token permanently */
2086#if 0
2087 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
2088 x,psp->state);
2089#endif
2090 switch( psp->state ){
2091 case INITIALIZE:
2092 psp->prevrule = 0;
2093 psp->preccounter = 0;
2094 psp->firstrule = psp->lastrule = 0;
2095 psp->gp->nrule = 0;
2096 /* Fall thru to next case */
2097 case WAITING_FOR_DECL_OR_RULE:
2098 if( x[0]=='%' ){
2099 psp->state = WAITING_FOR_DECL_KEYWORD;
2100 }else if( islower(x[0]) ){
2101 psp->lhs = Symbol_new(x);
2102 psp->nrhs = 0;
2103 psp->lhsalias = 0;
2104 psp->state = WAITING_FOR_ARROW;
2105 }else if( x[0]=='{' ){
2106 if( psp->prevrule==0 ){
2107 ErrorMsg(psp->filename,psp->tokenlineno,
drh3cb2f6e2012-01-09 14:19:05 +00002108"There is no prior rule upon which to attach the code \
drh75897232000-05-29 14:26:00 +00002109fragment which begins on this line.");
2110 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002111 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002112 ErrorMsg(psp->filename,psp->tokenlineno,
2113"Code fragment beginning on this line is not the first \
2114to follow the previous rule.");
2115 psp->errorcnt++;
2116 }else{
2117 psp->prevrule->line = psp->tokenlineno;
2118 psp->prevrule->code = &x[1];
drhf2f105d2012-08-20 15:53:54 +00002119 }
drh75897232000-05-29 14:26:00 +00002120 }else if( x[0]=='[' ){
2121 psp->state = PRECEDENCE_MARK_1;
2122 }else{
2123 ErrorMsg(psp->filename,psp->tokenlineno,
2124 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2125 x);
2126 psp->errorcnt++;
2127 }
2128 break;
2129 case PRECEDENCE_MARK_1:
2130 if( !isupper(x[0]) ){
2131 ErrorMsg(psp->filename,psp->tokenlineno,
2132 "The precedence symbol must be a terminal.");
2133 psp->errorcnt++;
2134 }else if( psp->prevrule==0 ){
2135 ErrorMsg(psp->filename,psp->tokenlineno,
2136 "There is no prior rule to assign precedence \"[%s]\".",x);
2137 psp->errorcnt++;
2138 }else if( psp->prevrule->precsym!=0 ){
2139 ErrorMsg(psp->filename,psp->tokenlineno,
2140"Precedence mark on this line is not the first \
2141to follow the previous rule.");
2142 psp->errorcnt++;
2143 }else{
2144 psp->prevrule->precsym = Symbol_new(x);
2145 }
2146 psp->state = PRECEDENCE_MARK_2;
2147 break;
2148 case PRECEDENCE_MARK_2:
2149 if( x[0]!=']' ){
2150 ErrorMsg(psp->filename,psp->tokenlineno,
2151 "Missing \"]\" on precedence mark.");
2152 psp->errorcnt++;
2153 }
2154 psp->state = WAITING_FOR_DECL_OR_RULE;
2155 break;
2156 case WAITING_FOR_ARROW:
2157 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2158 psp->state = IN_RHS;
2159 }else if( x[0]=='(' ){
2160 psp->state = LHS_ALIAS_1;
2161 }else{
2162 ErrorMsg(psp->filename,psp->tokenlineno,
2163 "Expected to see a \":\" following the LHS symbol \"%s\".",
2164 psp->lhs->name);
2165 psp->errorcnt++;
2166 psp->state = RESYNC_AFTER_RULE_ERROR;
2167 }
2168 break;
2169 case LHS_ALIAS_1:
2170 if( isalpha(x[0]) ){
2171 psp->lhsalias = x;
2172 psp->state = LHS_ALIAS_2;
2173 }else{
2174 ErrorMsg(psp->filename,psp->tokenlineno,
2175 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2176 x,psp->lhs->name);
2177 psp->errorcnt++;
2178 psp->state = RESYNC_AFTER_RULE_ERROR;
2179 }
2180 break;
2181 case LHS_ALIAS_2:
2182 if( x[0]==')' ){
2183 psp->state = LHS_ALIAS_3;
2184 }else{
2185 ErrorMsg(psp->filename,psp->tokenlineno,
2186 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2187 psp->errorcnt++;
2188 psp->state = RESYNC_AFTER_RULE_ERROR;
2189 }
2190 break;
2191 case LHS_ALIAS_3:
2192 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2193 psp->state = IN_RHS;
2194 }else{
2195 ErrorMsg(psp->filename,psp->tokenlineno,
2196 "Missing \"->\" following: \"%s(%s)\".",
2197 psp->lhs->name,psp->lhsalias);
2198 psp->errorcnt++;
2199 psp->state = RESYNC_AFTER_RULE_ERROR;
2200 }
2201 break;
2202 case IN_RHS:
2203 if( x[0]=='.' ){
2204 struct rule *rp;
drh9892c5d2007-12-21 00:02:11 +00002205 rp = (struct rule *)calloc( sizeof(struct rule) +
2206 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002207 if( rp==0 ){
2208 ErrorMsg(psp->filename,psp->tokenlineno,
2209 "Can't allocate enough memory for this rule.");
2210 psp->errorcnt++;
2211 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002212 }else{
drh75897232000-05-29 14:26:00 +00002213 int i;
2214 rp->ruleline = psp->tokenlineno;
2215 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002216 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002217 for(i=0; i<psp->nrhs; i++){
2218 rp->rhs[i] = psp->rhs[i];
2219 rp->rhsalias[i] = psp->alias[i];
drhf2f105d2012-08-20 15:53:54 +00002220 }
drh75897232000-05-29 14:26:00 +00002221 rp->lhs = psp->lhs;
2222 rp->lhsalias = psp->lhsalias;
2223 rp->nrhs = psp->nrhs;
2224 rp->code = 0;
2225 rp->precsym = 0;
2226 rp->index = psp->gp->nrule++;
2227 rp->nextlhs = rp->lhs->rule;
2228 rp->lhs->rule = rp;
2229 rp->next = 0;
2230 if( psp->firstrule==0 ){
2231 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002232 }else{
drh75897232000-05-29 14:26:00 +00002233 psp->lastrule->next = rp;
2234 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002235 }
drh75897232000-05-29 14:26:00 +00002236 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002237 }
drh75897232000-05-29 14:26:00 +00002238 psp->state = WAITING_FOR_DECL_OR_RULE;
2239 }else if( isalpha(x[0]) ){
2240 if( psp->nrhs>=MAXRHS ){
2241 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002242 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002243 x);
2244 psp->errorcnt++;
2245 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002246 }else{
drh75897232000-05-29 14:26:00 +00002247 psp->rhs[psp->nrhs] = Symbol_new(x);
2248 psp->alias[psp->nrhs] = 0;
2249 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002250 }
drhfd405312005-11-06 04:06:59 +00002251 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2252 struct symbol *msp = psp->rhs[psp->nrhs-1];
2253 if( msp->type!=MULTITERMINAL ){
2254 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002255 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002256 memset(msp, 0, sizeof(*msp));
2257 msp->type = MULTITERMINAL;
2258 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002259 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002260 msp->subsym[0] = origsp;
2261 msp->name = origsp->name;
2262 psp->rhs[psp->nrhs-1] = msp;
2263 }
2264 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002265 msp->subsym = (struct symbol **) realloc(msp->subsym,
2266 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002267 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2268 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2269 ErrorMsg(psp->filename,psp->tokenlineno,
2270 "Cannot form a compound containing a non-terminal");
2271 psp->errorcnt++;
2272 }
drh75897232000-05-29 14:26:00 +00002273 }else if( x[0]=='(' && psp->nrhs>0 ){
2274 psp->state = RHS_ALIAS_1;
2275 }else{
2276 ErrorMsg(psp->filename,psp->tokenlineno,
2277 "Illegal character on RHS of rule: \"%s\".",x);
2278 psp->errorcnt++;
2279 psp->state = RESYNC_AFTER_RULE_ERROR;
2280 }
2281 break;
2282 case RHS_ALIAS_1:
2283 if( isalpha(x[0]) ){
2284 psp->alias[psp->nrhs-1] = x;
2285 psp->state = RHS_ALIAS_2;
2286 }else{
2287 ErrorMsg(psp->filename,psp->tokenlineno,
2288 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2289 x,psp->rhs[psp->nrhs-1]->name);
2290 psp->errorcnt++;
2291 psp->state = RESYNC_AFTER_RULE_ERROR;
2292 }
2293 break;
2294 case RHS_ALIAS_2:
2295 if( x[0]==')' ){
2296 psp->state = IN_RHS;
2297 }else{
2298 ErrorMsg(psp->filename,psp->tokenlineno,
2299 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2300 psp->errorcnt++;
2301 psp->state = RESYNC_AFTER_RULE_ERROR;
2302 }
2303 break;
2304 case WAITING_FOR_DECL_KEYWORD:
2305 if( isalpha(x[0]) ){
2306 psp->declkeyword = x;
2307 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002308 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002309 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002310 psp->state = WAITING_FOR_DECL_ARG;
2311 if( strcmp(x,"name")==0 ){
2312 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002313 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002314 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002315 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002316 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002317 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002318 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002319 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002320 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002321 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002322 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002323 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002324 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002325 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002326 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002327 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002328 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002329 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002330 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002331 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002332 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002333 }else if( strcmp(x,"extra_argument")==0 ){
2334 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002335 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002336 }else if( strcmp(x,"token_type")==0 ){
2337 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002338 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002339 }else if( strcmp(x,"default_type")==0 ){
2340 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002341 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002342 }else if( strcmp(x,"stack_size")==0 ){
2343 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002344 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002345 }else if( strcmp(x,"start_symbol")==0 ){
2346 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002347 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002348 }else if( strcmp(x,"left")==0 ){
2349 psp->preccounter++;
2350 psp->declassoc = LEFT;
2351 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2352 }else if( strcmp(x,"right")==0 ){
2353 psp->preccounter++;
2354 psp->declassoc = RIGHT;
2355 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2356 }else if( strcmp(x,"nonassoc")==0 ){
2357 psp->preccounter++;
2358 psp->declassoc = NONE;
2359 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002360 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002361 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002362 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002363 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002364 }else if( strcmp(x,"fallback")==0 ){
2365 psp->fallback = 0;
2366 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002367 }else if( strcmp(x,"wildcard")==0 ){
2368 psp->state = WAITING_FOR_WILDCARD_ID;
drh61f92cd2014-01-11 03:06:18 +00002369 }else if( strcmp(x,"token_class")==0 ){
2370 psp->state = WAITING_FOR_CLASS_ID;
drh75897232000-05-29 14:26:00 +00002371 }else{
2372 ErrorMsg(psp->filename,psp->tokenlineno,
2373 "Unknown declaration keyword: \"%%%s\".",x);
2374 psp->errorcnt++;
2375 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002376 }
drh75897232000-05-29 14:26:00 +00002377 }else{
2378 ErrorMsg(psp->filename,psp->tokenlineno,
2379 "Illegal declaration keyword: \"%s\".",x);
2380 psp->errorcnt++;
2381 psp->state = RESYNC_AFTER_DECL_ERROR;
2382 }
2383 break;
2384 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2385 if( !isalpha(x[0]) ){
2386 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002387 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002388 psp->errorcnt++;
2389 psp->state = RESYNC_AFTER_DECL_ERROR;
2390 }else{
icculusd286fa62010-03-03 17:06:32 +00002391 struct symbol *sp = Symbol_new(x);
2392 psp->declargslot = &sp->destructor;
2393 psp->decllinenoslot = &sp->destLineno;
2394 psp->insertLineMacro = 1;
2395 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002396 }
2397 break;
2398 case WAITING_FOR_DATATYPE_SYMBOL:
2399 if( !isalpha(x[0]) ){
2400 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002401 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002402 psp->errorcnt++;
2403 psp->state = RESYNC_AFTER_DECL_ERROR;
2404 }else{
icculus866bf1e2010-02-17 20:31:32 +00002405 struct symbol *sp = Symbol_find(x);
2406 if((sp) && (sp->datatype)){
2407 ErrorMsg(psp->filename,psp->tokenlineno,
2408 "Symbol %%type \"%s\" already defined", x);
2409 psp->errorcnt++;
2410 psp->state = RESYNC_AFTER_DECL_ERROR;
2411 }else{
2412 if (!sp){
2413 sp = Symbol_new(x);
2414 }
2415 psp->declargslot = &sp->datatype;
2416 psp->insertLineMacro = 0;
2417 psp->state = WAITING_FOR_DECL_ARG;
2418 }
drh75897232000-05-29 14:26:00 +00002419 }
2420 break;
2421 case WAITING_FOR_PRECEDENCE_SYMBOL:
2422 if( x[0]=='.' ){
2423 psp->state = WAITING_FOR_DECL_OR_RULE;
2424 }else if( isupper(x[0]) ){
2425 struct symbol *sp;
2426 sp = Symbol_new(x);
2427 if( sp->prec>=0 ){
2428 ErrorMsg(psp->filename,psp->tokenlineno,
2429 "Symbol \"%s\" has already be given a precedence.",x);
2430 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002431 }else{
drh75897232000-05-29 14:26:00 +00002432 sp->prec = psp->preccounter;
2433 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002434 }
drh75897232000-05-29 14:26:00 +00002435 }else{
2436 ErrorMsg(psp->filename,psp->tokenlineno,
2437 "Can't assign a precedence to \"%s\".",x);
2438 psp->errorcnt++;
2439 }
2440 break;
2441 case WAITING_FOR_DECL_ARG:
drha5808f32008-04-27 22:19:44 +00002442 if( x[0]=='{' || x[0]=='\"' || isalnum(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002443 const char *zOld, *zNew;
2444 char *zBuf, *z;
mistachkin2318d332015-01-12 18:02:52 +00002445 int nOld, n, nLine = 0, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002446 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002447 char zLine[50];
2448 zNew = x;
2449 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002450 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002451 if( *psp->declargslot ){
2452 zOld = *psp->declargslot;
2453 }else{
2454 zOld = "";
2455 }
drh87cf1372008-08-13 20:09:06 +00002456 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002457 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002458 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002459 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2460 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002461 for(z=psp->filename, nBack=0; *z; z++){
2462 if( *z=='\\' ) nBack++;
2463 }
drh898799f2014-01-10 23:21:00 +00002464 lemon_sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002465 nLine = lemonStrlen(zLine);
2466 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002467 }
icculus9e44cf12010-02-14 17:14:22 +00002468 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2469 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002470 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002471 if( nOld && zBuf[-1]!='\n' ){
2472 *(zBuf++) = '\n';
2473 }
2474 memcpy(zBuf, zLine, nLine);
2475 zBuf += nLine;
2476 *(zBuf++) = '"';
2477 for(z=psp->filename; *z; z++){
2478 if( *z=='\\' ){
2479 *(zBuf++) = '\\';
2480 }
2481 *(zBuf++) = *z;
2482 }
2483 *(zBuf++) = '"';
2484 *(zBuf++) = '\n';
2485 }
drh4dc8ef52008-07-01 17:13:57 +00002486 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2487 psp->decllinenoslot[0] = psp->tokenlineno;
2488 }
drha5808f32008-04-27 22:19:44 +00002489 memcpy(zBuf, zNew, nNew);
2490 zBuf += nNew;
2491 *zBuf = 0;
2492 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002493 }else{
2494 ErrorMsg(psp->filename,psp->tokenlineno,
2495 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2496 psp->errorcnt++;
2497 psp->state = RESYNC_AFTER_DECL_ERROR;
2498 }
2499 break;
drh0bd1f4e2002-06-06 18:54:39 +00002500 case WAITING_FOR_FALLBACK_ID:
2501 if( x[0]=='.' ){
2502 psp->state = WAITING_FOR_DECL_OR_RULE;
2503 }else if( !isupper(x[0]) ){
2504 ErrorMsg(psp->filename, psp->tokenlineno,
2505 "%%fallback argument \"%s\" should be a token", x);
2506 psp->errorcnt++;
2507 }else{
2508 struct symbol *sp = Symbol_new(x);
2509 if( psp->fallback==0 ){
2510 psp->fallback = sp;
2511 }else if( sp->fallback ){
2512 ErrorMsg(psp->filename, psp->tokenlineno,
2513 "More than one fallback assigned to token %s", x);
2514 psp->errorcnt++;
2515 }else{
2516 sp->fallback = psp->fallback;
2517 psp->gp->has_fallback = 1;
2518 }
2519 }
2520 break;
drhe09daa92006-06-10 13:29:31 +00002521 case WAITING_FOR_WILDCARD_ID:
2522 if( x[0]=='.' ){
2523 psp->state = WAITING_FOR_DECL_OR_RULE;
2524 }else if( !isupper(x[0]) ){
2525 ErrorMsg(psp->filename, psp->tokenlineno,
2526 "%%wildcard argument \"%s\" should be a token", x);
2527 psp->errorcnt++;
2528 }else{
2529 struct symbol *sp = Symbol_new(x);
2530 if( psp->gp->wildcard==0 ){
2531 psp->gp->wildcard = sp;
2532 }else{
2533 ErrorMsg(psp->filename, psp->tokenlineno,
2534 "Extra wildcard to token: %s", x);
2535 psp->errorcnt++;
2536 }
2537 }
2538 break;
drh61f92cd2014-01-11 03:06:18 +00002539 case WAITING_FOR_CLASS_ID:
2540 if( !islower(x[0]) ){
2541 ErrorMsg(psp->filename, psp->tokenlineno,
2542 "%%token_class must be followed by an identifier: ", x);
2543 psp->errorcnt++;
2544 psp->state = RESYNC_AFTER_DECL_ERROR;
2545 }else if( Symbol_find(x) ){
2546 ErrorMsg(psp->filename, psp->tokenlineno,
2547 "Symbol \"%s\" already used", x);
2548 psp->errorcnt++;
2549 psp->state = RESYNC_AFTER_DECL_ERROR;
2550 }else{
2551 psp->tkclass = Symbol_new(x);
2552 psp->tkclass->type = MULTITERMINAL;
2553 psp->state = WAITING_FOR_CLASS_TOKEN;
2554 }
2555 break;
2556 case WAITING_FOR_CLASS_TOKEN:
2557 if( x[0]=='.' ){
2558 psp->state = WAITING_FOR_DECL_OR_RULE;
2559 }else if( isupper(x[0]) || ((x[0]=='|' || x[0]=='/') && isupper(x[1])) ){
2560 struct symbol *msp = psp->tkclass;
2561 msp->nsubsym++;
2562 msp->subsym = (struct symbol **) realloc(msp->subsym,
2563 sizeof(struct symbol*)*msp->nsubsym);
2564 if( !isupper(x[0]) ) x++;
2565 msp->subsym[msp->nsubsym-1] = Symbol_new(x);
2566 }else{
2567 ErrorMsg(psp->filename, psp->tokenlineno,
2568 "%%token_class argument \"%s\" should be a token", x);
2569 psp->errorcnt++;
2570 psp->state = RESYNC_AFTER_DECL_ERROR;
2571 }
2572 break;
drh75897232000-05-29 14:26:00 +00002573 case RESYNC_AFTER_RULE_ERROR:
2574/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2575** break; */
2576 case RESYNC_AFTER_DECL_ERROR:
2577 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2578 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2579 break;
2580 }
2581}
2582
drh34ff57b2008-07-14 12:27:51 +00002583/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002584** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2585** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2586** comments them out. Text in between is also commented out as appropriate.
2587*/
danielk1977940fac92005-01-23 22:41:37 +00002588static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002589 int i, j, k, n;
2590 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002591 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002592 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002593 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002594 for(i=0; z[i]; i++){
2595 if( z[i]=='\n' ) lineno++;
2596 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2597 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2598 if( exclude ){
2599 exclude--;
2600 if( exclude==0 ){
2601 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2602 }
2603 }
2604 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2605 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2606 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2607 if( exclude ){
2608 exclude++;
2609 }else{
2610 for(j=i+7; isspace(z[j]); j++){}
2611 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2612 exclude = 1;
2613 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002614 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002615 exclude = 0;
2616 break;
2617 }
2618 }
2619 if( z[i+3]=='n' ) exclude = !exclude;
2620 if( exclude ){
2621 start = i;
2622 start_lineno = lineno;
2623 }
2624 }
2625 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2626 }
2627 }
2628 if( exclude ){
2629 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2630 exit(1);
2631 }
2632}
2633
drh75897232000-05-29 14:26:00 +00002634/* In spite of its name, this function is really a scanner. It read
2635** in the entire input file (all at once) then tokenizes it. Each
2636** token is passed to the function "parseonetoken" which builds all
2637** the appropriate data structures in the global state vector "gp".
2638*/
icculus9e44cf12010-02-14 17:14:22 +00002639void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002640{
2641 struct pstate ps;
2642 FILE *fp;
2643 char *filebuf;
mistachkin2318d332015-01-12 18:02:52 +00002644 unsigned int filesize;
drh75897232000-05-29 14:26:00 +00002645 int lineno;
2646 int c;
2647 char *cp, *nextcp;
2648 int startline = 0;
2649
rse38514a92007-09-20 11:34:17 +00002650 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002651 ps.gp = gp;
2652 ps.filename = gp->filename;
2653 ps.errorcnt = 0;
2654 ps.state = INITIALIZE;
2655
2656 /* Begin by reading the input file */
2657 fp = fopen(ps.filename,"rb");
2658 if( fp==0 ){
2659 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2660 gp->errorcnt++;
2661 return;
2662 }
2663 fseek(fp,0,2);
2664 filesize = ftell(fp);
2665 rewind(fp);
2666 filebuf = (char *)malloc( filesize+1 );
drh03e1b1f2014-01-11 12:52:25 +00002667 if( filesize>100000000 || filebuf==0 ){
2668 ErrorMsg(ps.filename,0,"Input file too large.");
drh75897232000-05-29 14:26:00 +00002669 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002670 fclose(fp);
drh75897232000-05-29 14:26:00 +00002671 return;
2672 }
2673 if( fread(filebuf,1,filesize,fp)!=filesize ){
2674 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2675 filesize);
2676 free(filebuf);
2677 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002678 fclose(fp);
drh75897232000-05-29 14:26:00 +00002679 return;
2680 }
2681 fclose(fp);
2682 filebuf[filesize] = 0;
2683
drh6d08b4d2004-07-20 12:45:22 +00002684 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2685 preprocess_input(filebuf);
2686
drh75897232000-05-29 14:26:00 +00002687 /* Now scan the text of the input file */
2688 lineno = 1;
2689 for(cp=filebuf; (c= *cp)!=0; ){
2690 if( c=='\n' ) lineno++; /* Keep track of the line number */
2691 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2692 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2693 cp+=2;
2694 while( (c= *cp)!=0 && c!='\n' ) cp++;
2695 continue;
2696 }
2697 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2698 cp+=2;
2699 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2700 if( c=='\n' ) lineno++;
2701 cp++;
2702 }
2703 if( c ) cp++;
2704 continue;
2705 }
2706 ps.tokenstart = cp; /* Mark the beginning of the token */
2707 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2708 if( c=='\"' ){ /* String literals */
2709 cp++;
2710 while( (c= *cp)!=0 && c!='\"' ){
2711 if( c=='\n' ) lineno++;
2712 cp++;
2713 }
2714 if( c==0 ){
2715 ErrorMsg(ps.filename,startline,
2716"String starting on this line is not terminated before the end of the file.");
2717 ps.errorcnt++;
2718 nextcp = cp;
2719 }else{
2720 nextcp = cp+1;
2721 }
2722 }else if( c=='{' ){ /* A block of C code */
2723 int level;
2724 cp++;
2725 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2726 if( c=='\n' ) lineno++;
2727 else if( c=='{' ) level++;
2728 else if( c=='}' ) level--;
2729 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2730 int prevc;
2731 cp = &cp[2];
2732 prevc = 0;
2733 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2734 if( c=='\n' ) lineno++;
2735 prevc = c;
2736 cp++;
drhf2f105d2012-08-20 15:53:54 +00002737 }
2738 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00002739 cp = &cp[2];
2740 while( (c= *cp)!=0 && c!='\n' ) cp++;
2741 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00002742 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00002743 int startchar, prevc;
2744 startchar = c;
2745 prevc = 0;
2746 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2747 if( c=='\n' ) lineno++;
2748 if( prevc=='\\' ) prevc = 0;
2749 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00002750 }
2751 }
drh75897232000-05-29 14:26:00 +00002752 }
2753 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002754 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002755"C code starting on this line is not terminated before the end of the file.");
2756 ps.errorcnt++;
2757 nextcp = cp;
2758 }else{
2759 nextcp = cp+1;
2760 }
2761 }else if( isalnum(c) ){ /* Identifiers */
2762 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2763 nextcp = cp;
2764 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2765 cp += 3;
2766 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002767 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2768 cp += 2;
2769 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2770 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002771 }else{ /* All other (one character) operators */
2772 cp++;
2773 nextcp = cp;
2774 }
2775 c = *cp;
2776 *cp = 0; /* Null terminate the token */
2777 parseonetoken(&ps); /* Parse the token */
mistachkin2318d332015-01-12 18:02:52 +00002778 *cp = (char)c; /* Restore the buffer */
drh75897232000-05-29 14:26:00 +00002779 cp = nextcp;
2780 }
2781 free(filebuf); /* Release the buffer after parsing */
2782 gp->rule = ps.firstrule;
2783 gp->errorcnt = ps.errorcnt;
2784}
2785/*************************** From the file "plink.c" *********************/
2786/*
2787** Routines processing configuration follow-set propagation links
2788** in the LEMON parser generator.
2789*/
2790static struct plink *plink_freelist = 0;
2791
2792/* Allocate a new plink */
2793struct plink *Plink_new(){
icculus9e44cf12010-02-14 17:14:22 +00002794 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00002795
2796 if( plink_freelist==0 ){
2797 int i;
2798 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002799 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002800 if( plink_freelist==0 ){
2801 fprintf(stderr,
2802 "Unable to allocate memory for a new follow-set propagation link.\n");
2803 exit(1);
2804 }
2805 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2806 plink_freelist[amt-1].next = 0;
2807 }
icculus9e44cf12010-02-14 17:14:22 +00002808 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00002809 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00002810 return newlink;
drh75897232000-05-29 14:26:00 +00002811}
2812
2813/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00002814void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002815{
icculus9e44cf12010-02-14 17:14:22 +00002816 struct plink *newlink;
2817 newlink = Plink_new();
2818 newlink->next = *plpp;
2819 *plpp = newlink;
2820 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00002821}
2822
2823/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00002824void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00002825{
2826 struct plink *nextpl;
2827 while( from ){
2828 nextpl = from->next;
2829 from->next = *to;
2830 *to = from;
2831 from = nextpl;
2832 }
2833}
2834
2835/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00002836void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00002837{
2838 struct plink *nextpl;
2839
2840 while( plp ){
2841 nextpl = plp->next;
2842 plp->next = plink_freelist;
2843 plink_freelist = plp;
2844 plp = nextpl;
2845 }
2846}
2847/*********************** From the file "report.c" **************************/
2848/*
2849** Procedures for generating reports and tables in the LEMON parser generator.
2850*/
2851
2852/* Generate a filename with the given suffix. Space to hold the
2853** name comes from malloc() and must be freed by the calling
2854** function.
2855*/
icculus9e44cf12010-02-14 17:14:22 +00002856PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00002857{
2858 char *name;
2859 char *cp;
2860
icculus9e44cf12010-02-14 17:14:22 +00002861 name = (char*)malloc( lemonStrlen(lemp->filename) + lemonStrlen(suffix) + 5 );
drh75897232000-05-29 14:26:00 +00002862 if( name==0 ){
2863 fprintf(stderr,"Can't allocate space for a filename.\n");
2864 exit(1);
2865 }
drh898799f2014-01-10 23:21:00 +00002866 lemon_strcpy(name,lemp->filename);
drh75897232000-05-29 14:26:00 +00002867 cp = strrchr(name,'.');
2868 if( cp ) *cp = 0;
drh898799f2014-01-10 23:21:00 +00002869 lemon_strcat(name,suffix);
drh75897232000-05-29 14:26:00 +00002870 return name;
2871}
2872
2873/* Open a file with a name based on the name of the input file,
2874** but with a different (specified) suffix, and return a pointer
2875** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00002876PRIVATE FILE *file_open(
2877 struct lemon *lemp,
2878 const char *suffix,
2879 const char *mode
2880){
drh75897232000-05-29 14:26:00 +00002881 FILE *fp;
2882
2883 if( lemp->outname ) free(lemp->outname);
2884 lemp->outname = file_makename(lemp, suffix);
2885 fp = fopen(lemp->outname,mode);
2886 if( fp==0 && *mode=='w' ){
2887 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2888 lemp->errorcnt++;
2889 return 0;
2890 }
2891 return fp;
2892}
2893
2894/* Duplicate the input file without comments and without actions
2895** on rules */
icculus9e44cf12010-02-14 17:14:22 +00002896void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002897{
2898 struct rule *rp;
2899 struct symbol *sp;
2900 int i, j, maxlen, len, ncolumns, skip;
2901 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2902 maxlen = 10;
2903 for(i=0; i<lemp->nsymbol; i++){
2904 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00002905 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00002906 if( len>maxlen ) maxlen = len;
2907 }
2908 ncolumns = 76/(maxlen+5);
2909 if( ncolumns<1 ) ncolumns = 1;
2910 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2911 for(i=0; i<skip; i++){
2912 printf("//");
2913 for(j=i; j<lemp->nsymbol; j+=skip){
2914 sp = lemp->symbols[j];
2915 assert( sp->index==j );
2916 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2917 }
2918 printf("\n");
2919 }
2920 for(rp=lemp->rule; rp; rp=rp->next){
2921 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002922 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002923 printf(" ::=");
2924 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002925 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00002926 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00002927 printf(" %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00002928 for(j=1; j<sp->nsubsym; j++){
2929 printf("|%s", sp->subsym[j]->name);
2930 }
drh61f92cd2014-01-11 03:06:18 +00002931 }else{
2932 printf(" %s", sp->name);
drhfd405312005-11-06 04:06:59 +00002933 }
2934 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002935 }
2936 printf(".");
2937 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002938 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002939 printf("\n");
2940 }
2941}
2942
icculus9e44cf12010-02-14 17:14:22 +00002943void ConfigPrint(FILE *fp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002944{
2945 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002946 struct symbol *sp;
2947 int i, j;
drh75897232000-05-29 14:26:00 +00002948 rp = cfp->rp;
2949 fprintf(fp,"%s ::=",rp->lhs->name);
2950 for(i=0; i<=rp->nrhs; i++){
2951 if( i==cfp->dot ) fprintf(fp," *");
2952 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002953 sp = rp->rhs[i];
drhfd405312005-11-06 04:06:59 +00002954 if( sp->type==MULTITERMINAL ){
drh61f92cd2014-01-11 03:06:18 +00002955 fprintf(fp," %s", sp->subsym[0]->name);
drhfd405312005-11-06 04:06:59 +00002956 for(j=1; j<sp->nsubsym; j++){
2957 fprintf(fp,"|%s",sp->subsym[j]->name);
2958 }
drh61f92cd2014-01-11 03:06:18 +00002959 }else{
2960 fprintf(fp," %s", sp->name);
drhfd405312005-11-06 04:06:59 +00002961 }
drh75897232000-05-29 14:26:00 +00002962 }
2963}
2964
2965/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002966#if 0
drh75897232000-05-29 14:26:00 +00002967/* Print a set */
2968PRIVATE void SetPrint(out,set,lemp)
2969FILE *out;
2970char *set;
2971struct lemon *lemp;
2972{
2973 int i;
2974 char *spacer;
2975 spacer = "";
2976 fprintf(out,"%12s[","");
2977 for(i=0; i<lemp->nterminal; i++){
2978 if( SetFind(set,i) ){
2979 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2980 spacer = " ";
2981 }
2982 }
2983 fprintf(out,"]\n");
2984}
2985
2986/* Print a plink chain */
2987PRIVATE void PlinkPrint(out,plp,tag)
2988FILE *out;
2989struct plink *plp;
2990char *tag;
2991{
2992 while( plp ){
drhada354d2005-11-05 15:03:59 +00002993 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002994 ConfigPrint(out,plp->cfp);
2995 fprintf(out,"\n");
2996 plp = plp->next;
2997 }
2998}
2999#endif
3000
3001/* Print an action to the given file descriptor. Return FALSE if
3002** nothing was actually printed.
3003*/
3004int PrintAction(struct action *ap, FILE *fp, int indent){
3005 int result = 1;
3006 switch( ap->type ){
3007 case SHIFT:
drhada354d2005-11-05 15:03:59 +00003008 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00003009 break;
3010 case REDUCE:
3011 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
3012 break;
3013 case ACCEPT:
3014 fprintf(fp,"%*s accept",indent,ap->sp->name);
3015 break;
3016 case ERROR:
3017 fprintf(fp,"%*s error",indent,ap->sp->name);
3018 break;
drh9892c5d2007-12-21 00:02:11 +00003019 case SRCONFLICT:
3020 case RRCONFLICT:
drh75897232000-05-29 14:26:00 +00003021 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
3022 indent,ap->sp->name,ap->x.rp->index);
3023 break;
drh9892c5d2007-12-21 00:02:11 +00003024 case SSCONFLICT:
drhdd7e9db2010-07-19 01:52:07 +00003025 fprintf(fp,"%*s shift %-3d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00003026 indent,ap->sp->name,ap->x.stp->statenum);
3027 break;
drh75897232000-05-29 14:26:00 +00003028 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003029 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00003030 fprintf(fp,"%*s shift %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003031 indent,ap->sp->name,ap->x.stp->statenum);
3032 }else{
3033 result = 0;
3034 }
3035 break;
drh75897232000-05-29 14:26:00 +00003036 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00003037 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00003038 fprintf(fp,"%*s reduce %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00003039 indent,ap->sp->name,ap->x.rp->index);
3040 }else{
3041 result = 0;
3042 }
3043 break;
drh75897232000-05-29 14:26:00 +00003044 case NOT_USED:
3045 result = 0;
3046 break;
3047 }
3048 return result;
3049}
3050
3051/* Generate the "y.output" log file */
icculus9e44cf12010-02-14 17:14:22 +00003052void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003053{
3054 int i;
3055 struct state *stp;
3056 struct config *cfp;
3057 struct action *ap;
3058 FILE *fp;
3059
drh2aa6ca42004-09-10 00:14:04 +00003060 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00003061 if( fp==0 ) return;
drh75897232000-05-29 14:26:00 +00003062 for(i=0; i<lemp->nstate; i++){
3063 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00003064 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00003065 if( lemp->basisflag ) cfp=stp->bp;
3066 else cfp=stp->cfp;
3067 while( cfp ){
3068 char buf[20];
3069 if( cfp->dot==cfp->rp->nrhs ){
drh898799f2014-01-10 23:21:00 +00003070 lemon_sprintf(buf,"(%d)",cfp->rp->index);
drh75897232000-05-29 14:26:00 +00003071 fprintf(fp," %5s ",buf);
3072 }else{
3073 fprintf(fp," ");
3074 }
3075 ConfigPrint(fp,cfp);
3076 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00003077#if 0
drh75897232000-05-29 14:26:00 +00003078 SetPrint(fp,cfp->fws,lemp);
3079 PlinkPrint(fp,cfp->fplp,"To ");
3080 PlinkPrint(fp,cfp->bplp,"From");
3081#endif
3082 if( lemp->basisflag ) cfp=cfp->bp;
3083 else cfp=cfp->next;
3084 }
3085 fprintf(fp,"\n");
3086 for(ap=stp->ap; ap; ap=ap->next){
3087 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
3088 }
3089 fprintf(fp,"\n");
3090 }
drhe9278182007-07-18 18:16:29 +00003091 fprintf(fp, "----------------------------------------------------\n");
3092 fprintf(fp, "Symbols:\n");
3093 for(i=0; i<lemp->nsymbol; i++){
3094 int j;
3095 struct symbol *sp;
3096
3097 sp = lemp->symbols[i];
3098 fprintf(fp, " %3d: %s", i, sp->name);
3099 if( sp->type==NONTERMINAL ){
3100 fprintf(fp, ":");
3101 if( sp->lambda ){
3102 fprintf(fp, " <lambda>");
3103 }
3104 for(j=0; j<lemp->nterminal; j++){
3105 if( sp->firstset && SetFind(sp->firstset, j) ){
3106 fprintf(fp, " %s", lemp->symbols[j]->name);
3107 }
3108 }
3109 }
3110 fprintf(fp, "\n");
3111 }
drh75897232000-05-29 14:26:00 +00003112 fclose(fp);
3113 return;
3114}
3115
3116/* Search for the file "name" which is in the same directory as
3117** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00003118PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00003119{
icculus9e44cf12010-02-14 17:14:22 +00003120 const char *pathlist;
3121 char *pathbufptr;
3122 char *pathbuf;
drh75897232000-05-29 14:26:00 +00003123 char *path,*cp;
3124 char c;
drh75897232000-05-29 14:26:00 +00003125
3126#ifdef __WIN32__
3127 cp = strrchr(argv0,'\\');
3128#else
3129 cp = strrchr(argv0,'/');
3130#endif
3131 if( cp ){
3132 c = *cp;
3133 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00003134 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh898799f2014-01-10 23:21:00 +00003135 if( path ) lemon_sprintf(path,"%s/%s",argv0,name);
drh75897232000-05-29 14:26:00 +00003136 *cp = c;
3137 }else{
drh75897232000-05-29 14:26:00 +00003138 pathlist = getenv("PATH");
3139 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00003140 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00003141 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00003142 if( (pathbuf != 0) && (path!=0) ){
3143 pathbufptr = pathbuf;
drh898799f2014-01-10 23:21:00 +00003144 lemon_strcpy(pathbuf, pathlist);
icculus9e44cf12010-02-14 17:14:22 +00003145 while( *pathbuf ){
3146 cp = strchr(pathbuf,':');
3147 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00003148 c = *cp;
3149 *cp = 0;
drh898799f2014-01-10 23:21:00 +00003150 lemon_sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003151 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003152 if( c==0 ) pathbuf[0] = 0;
3153 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003154 if( access(path,modemask)==0 ) break;
3155 }
icculus9e44cf12010-02-14 17:14:22 +00003156 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003157 }
3158 }
3159 return path;
3160}
3161
3162/* Given an action, compute the integer value for that action
3163** which is to be put in the action table of the generated machine.
3164** Return negative if no action should be generated.
3165*/
icculus9e44cf12010-02-14 17:14:22 +00003166PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003167{
3168 int act;
3169 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00003170 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00003171 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
3172 case ERROR: act = lemp->nstate + lemp->nrule; break;
3173 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
3174 default: act = -1; break;
3175 }
3176 return act;
3177}
3178
3179#define LINESIZE 1000
3180/* The next cluster of routines are for reading the template file
3181** and writing the results to the generated parser */
3182/* The first function transfers data from "in" to "out" until
3183** a line is seen which begins with "%%". The line number is
3184** tracked.
3185**
3186** if name!=0, then any word that begin with "Parse" is changed to
3187** begin with *name instead.
3188*/
icculus9e44cf12010-02-14 17:14:22 +00003189PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003190{
3191 int i, iStart;
3192 char line[LINESIZE];
3193 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3194 (*lineno)++;
3195 iStart = 0;
3196 if( name ){
3197 for(i=0; line[i]; i++){
3198 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
3199 && (i==0 || !isalpha(line[i-1]))
3200 ){
3201 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3202 fprintf(out,"%s",name);
3203 i += 4;
3204 iStart = i+1;
3205 }
3206 }
3207 }
3208 fprintf(out,"%s",&line[iStart]);
3209 }
3210}
3211
3212/* The next function finds the template file and opens it, returning
3213** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003214PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003215{
3216 static char templatename[] = "lempar.c";
3217 char buf[1000];
3218 FILE *in;
3219 char *tpltname;
3220 char *cp;
3221
icculus3e143bd2010-02-14 00:48:49 +00003222 /* first, see if user specified a template filename on the command line. */
3223 if (user_templatename != 0) {
3224 if( access(user_templatename,004)==-1 ){
3225 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3226 user_templatename);
3227 lemp->errorcnt++;
3228 return 0;
3229 }
3230 in = fopen(user_templatename,"rb");
3231 if( in==0 ){
3232 fprintf(stderr,"Can't open the template file \"%s\".\n",user_templatename);
3233 lemp->errorcnt++;
3234 return 0;
3235 }
3236 return in;
3237 }
3238
drh75897232000-05-29 14:26:00 +00003239 cp = strrchr(lemp->filename,'.');
3240 if( cp ){
drh898799f2014-01-10 23:21:00 +00003241 lemon_sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003242 }else{
drh898799f2014-01-10 23:21:00 +00003243 lemon_sprintf(buf,"%s.lt",lemp->filename);
drh75897232000-05-29 14:26:00 +00003244 }
3245 if( access(buf,004)==0 ){
3246 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003247 }else if( access(templatename,004)==0 ){
3248 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003249 }else{
3250 tpltname = pathsearch(lemp->argv0,templatename,0);
3251 }
3252 if( tpltname==0 ){
3253 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3254 templatename);
3255 lemp->errorcnt++;
3256 return 0;
3257 }
drh2aa6ca42004-09-10 00:14:04 +00003258 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003259 if( in==0 ){
3260 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3261 lemp->errorcnt++;
3262 return 0;
3263 }
3264 return in;
3265}
3266
drhaf805ca2004-09-07 11:28:25 +00003267/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003268PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003269{
3270 fprintf(out,"#line %d \"",lineno);
3271 while( *filename ){
3272 if( *filename == '\\' ) putc('\\',out);
3273 putc(*filename,out);
3274 filename++;
3275 }
3276 fprintf(out,"\"\n");
3277}
3278
drh75897232000-05-29 14:26:00 +00003279/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003280PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003281{
3282 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003283 while( *str ){
drh75897232000-05-29 14:26:00 +00003284 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003285 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003286 str++;
3287 }
drh9db55df2004-09-09 14:01:21 +00003288 if( str[-1]!='\n' ){
3289 putc('\n',out);
3290 (*lineno)++;
3291 }
shane58543932008-12-10 20:10:04 +00003292 if (!lemp->nolinenosflag) {
3293 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3294 }
drh75897232000-05-29 14:26:00 +00003295 return;
3296}
3297
3298/*
3299** The following routine emits code for the destructor for the
3300** symbol sp
3301*/
icculus9e44cf12010-02-14 17:14:22 +00003302void emit_destructor_code(
3303 FILE *out,
3304 struct symbol *sp,
3305 struct lemon *lemp,
3306 int *lineno
3307){
drhcc83b6e2004-04-23 23:38:42 +00003308 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003309
drh75897232000-05-29 14:26:00 +00003310 if( sp->type==TERMINAL ){
3311 cp = lemp->tokendest;
3312 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003313 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003314 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003315 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003316 fprintf(out,"{\n"); (*lineno)++;
shane58543932008-12-10 20:10:04 +00003317 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,sp->destLineno,lemp->filename); }
drh960e8c62001-04-03 16:53:21 +00003318 }else if( lemp->vardest ){
3319 cp = lemp->vardest;
3320 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003321 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003322 }else{
3323 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003324 }
3325 for(; *cp; cp++){
3326 if( *cp=='$' && cp[1]=='$' ){
3327 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3328 cp++;
3329 continue;
3330 }
shane58543932008-12-10 20:10:04 +00003331 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003332 fputc(*cp,out);
3333 }
shane58543932008-12-10 20:10:04 +00003334 fprintf(out,"\n"); (*lineno)++;
3335 if (!lemp->nolinenosflag) {
3336 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3337 }
3338 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003339 return;
3340}
3341
3342/*
drh960e8c62001-04-03 16:53:21 +00003343** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003344*/
icculus9e44cf12010-02-14 17:14:22 +00003345int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003346{
3347 int ret;
3348 if( sp->type==TERMINAL ){
3349 ret = lemp->tokendest!=0;
3350 }else{
drh960e8c62001-04-03 16:53:21 +00003351 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003352 }
3353 return ret;
3354}
3355
drh0bb132b2004-07-20 14:06:51 +00003356/*
3357** Append text to a dynamically allocated string. If zText is 0 then
3358** reset the string to be empty again. Always return the complete text
3359** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003360**
3361** n bytes of zText are stored. If n==0 then all of zText up to the first
3362** \000 terminator is stored. zText can contain up to two instances of
3363** %d. The values of p1 and p2 are written into the first and second
3364** %d.
3365**
3366** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003367*/
icculus9e44cf12010-02-14 17:14:22 +00003368PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3369 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003370 static char *z = 0;
3371 static int alloced = 0;
3372 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003373 int c;
drh0bb132b2004-07-20 14:06:51 +00003374 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003375 if( zText==0 ){
3376 used = 0;
3377 return z;
3378 }
drh7ac25c72004-08-19 15:12:26 +00003379 if( n<=0 ){
3380 if( n<0 ){
3381 used += n;
3382 assert( used>=0 );
3383 }
drh87cf1372008-08-13 20:09:06 +00003384 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003385 }
drhdf609712010-11-23 20:55:27 +00003386 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003387 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003388 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003389 }
icculus9e44cf12010-02-14 17:14:22 +00003390 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003391 while( n-- > 0 ){
3392 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003393 if( c=='%' && n>0 && zText[0]=='d' ){
drh898799f2014-01-10 23:21:00 +00003394 lemon_sprintf(zInt, "%d", p1);
drh0bb132b2004-07-20 14:06:51 +00003395 p1 = p2;
drh898799f2014-01-10 23:21:00 +00003396 lemon_strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003397 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003398 zText++;
3399 n--;
3400 }else{
mistachkin2318d332015-01-12 18:02:52 +00003401 z[used++] = (char)c;
drh0bb132b2004-07-20 14:06:51 +00003402 }
3403 }
3404 z[used] = 0;
3405 return z;
3406}
3407
3408/*
3409** zCode is a string that is the action associated with a rule. Expand
3410** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003411** stack.
drh0bb132b2004-07-20 14:06:51 +00003412*/
drhaf805ca2004-09-07 11:28:25 +00003413PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003414 char *cp, *xp;
3415 int i;
3416 char lhsused = 0; /* True if the LHS element has been used */
3417 char used[MAXRHS]; /* True for each RHS element which is used */
3418
3419 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3420 lhsused = 0;
3421
drh19c9e562007-03-29 20:13:53 +00003422 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003423 static char newlinestr[2] = { '\n', '\0' };
3424 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003425 rp->line = rp->ruleline;
3426 }
3427
drh0bb132b2004-07-20 14:06:51 +00003428 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003429
3430 /* This const cast is wrong but harmless, if we're careful. */
3431 for(cp=(char *)rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003432 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3433 char saved;
3434 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3435 saved = *xp;
3436 *xp = 0;
3437 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003438 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003439 cp = xp;
3440 lhsused = 1;
3441 }else{
3442 for(i=0; i<rp->nrhs; i++){
3443 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003444 if( cp!=rp->code && cp[-1]=='@' ){
3445 /* If the argument is of the form @X then substituted
3446 ** the token number of X, not the value of X */
3447 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3448 }else{
drhfd405312005-11-06 04:06:59 +00003449 struct symbol *sp = rp->rhs[i];
3450 int dtnum;
3451 if( sp->type==MULTITERMINAL ){
3452 dtnum = sp->subsym[0]->dtnum;
3453 }else{
3454 dtnum = sp->dtnum;
3455 }
3456 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003457 }
drh0bb132b2004-07-20 14:06:51 +00003458 cp = xp;
3459 used[i] = 1;
3460 break;
3461 }
3462 }
3463 }
3464 *xp = saved;
3465 }
3466 append_str(cp, 1, 0, 0);
3467 } /* End loop */
3468
3469 /* Check to make sure the LHS has been used */
3470 if( rp->lhsalias && !lhsused ){
3471 ErrorMsg(lemp->filename,rp->ruleline,
3472 "Label \"%s\" for \"%s(%s)\" is never used.",
3473 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3474 lemp->errorcnt++;
3475 }
3476
3477 /* Generate destructor code for RHS symbols which are not used in the
3478 ** reduce code */
3479 for(i=0; i<rp->nrhs; i++){
3480 if( rp->rhsalias[i] && !used[i] ){
3481 ErrorMsg(lemp->filename,rp->ruleline,
3482 "Label %s for \"%s(%s)\" is never used.",
3483 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3484 lemp->errorcnt++;
3485 }else if( rp->rhsalias[i]==0 ){
3486 if( has_destructor(rp->rhs[i],lemp) ){
drh639efd02008-08-13 20:04:29 +00003487 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003488 rp->rhs[i]->index,i-rp->nrhs+1);
3489 }else{
3490 /* No destructor defined for this term */
3491 }
3492 }
3493 }
drh61e339a2007-01-16 03:09:02 +00003494 if( rp->code ){
3495 cp = append_str(0,0,0,0);
3496 rp->code = Strsafe(cp?cp:"");
3497 }
drh0bb132b2004-07-20 14:06:51 +00003498}
3499
drh75897232000-05-29 14:26:00 +00003500/*
3501** Generate code which executes when the rule "rp" is reduced. Write
3502** the code to "out". Make sure lineno stays up-to-date.
3503*/
icculus9e44cf12010-02-14 17:14:22 +00003504PRIVATE void emit_code(
3505 FILE *out,
3506 struct rule *rp,
3507 struct lemon *lemp,
3508 int *lineno
3509){
3510 const char *cp;
drh75897232000-05-29 14:26:00 +00003511
3512 /* Generate code to do the reduce action */
3513 if( rp->code ){
shane58543932008-12-10 20:10:04 +00003514 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,rp->line,lemp->filename); }
drhaf805ca2004-09-07 11:28:25 +00003515 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003516 for(cp=rp->code; *cp; cp++){
shane58543932008-12-10 20:10:04 +00003517 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003518 } /* End loop */
shane58543932008-12-10 20:10:04 +00003519 fprintf(out,"}\n"); (*lineno)++;
3520 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,*lineno,lemp->outname); }
drh75897232000-05-29 14:26:00 +00003521 } /* End if( rp->code ) */
3522
drh75897232000-05-29 14:26:00 +00003523 return;
3524}
3525
3526/*
3527** Print the definition of the union used for the parser's data stack.
3528** This union contains fields for every possible data type for tokens
3529** and nonterminals. In the process of computing and printing this
3530** union, also set the ".dtnum" field of every terminal and nonterminal
3531** symbol.
3532*/
icculus9e44cf12010-02-14 17:14:22 +00003533void print_stack_union(
3534 FILE *out, /* The output stream */
3535 struct lemon *lemp, /* The main info structure for this parser */
3536 int *plineno, /* Pointer to the line number */
3537 int mhflag /* True if generating makeheaders output */
3538){
drh75897232000-05-29 14:26:00 +00003539 int lineno = *plineno; /* The line number of the output */
3540 char **types; /* A hash table of datatypes */
3541 int arraysize; /* Size of the "types" array */
3542 int maxdtlength; /* Maximum length of any ".datatype" field. */
3543 char *stddt; /* Standardized name for a datatype */
3544 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00003545 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00003546 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00003547
3548 /* Allocate and initialize types[] and allocate stddt[] */
3549 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003550 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00003551 if( types==0 ){
3552 fprintf(stderr,"Out of memory.\n");
3553 exit(1);
3554 }
drh75897232000-05-29 14:26:00 +00003555 for(i=0; i<arraysize; i++) types[i] = 0;
3556 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003557 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003558 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003559 }
drh75897232000-05-29 14:26:00 +00003560 for(i=0; i<lemp->nsymbol; i++){
3561 int len;
3562 struct symbol *sp = lemp->symbols[i];
3563 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003564 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003565 if( len>maxdtlength ) maxdtlength = len;
3566 }
3567 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00003568 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00003569 fprintf(stderr,"Out of memory.\n");
3570 exit(1);
3571 }
3572
3573 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3574 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003575 ** used for terminal symbols. If there is no %default_type defined then
3576 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3577 ** a datatype using the %type directive.
3578 */
drh75897232000-05-29 14:26:00 +00003579 for(i=0; i<lemp->nsymbol; i++){
3580 struct symbol *sp = lemp->symbols[i];
3581 char *cp;
3582 if( sp==lemp->errsym ){
3583 sp->dtnum = arraysize+1;
3584 continue;
3585 }
drh960e8c62001-04-03 16:53:21 +00003586 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003587 sp->dtnum = 0;
3588 continue;
3589 }
3590 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003591 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003592 j = 0;
3593 while( isspace(*cp) ) cp++;
3594 while( *cp ) stddt[j++] = *cp++;
3595 while( j>0 && isspace(stddt[j-1]) ) j--;
3596 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00003597 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00003598 sp->dtnum = 0;
3599 continue;
3600 }
drh75897232000-05-29 14:26:00 +00003601 hash = 0;
3602 for(j=0; stddt[j]; j++){
3603 hash = hash*53 + stddt[j];
3604 }
drh3b2129c2003-05-13 00:34:21 +00003605 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003606 while( types[hash] ){
3607 if( strcmp(types[hash],stddt)==0 ){
3608 sp->dtnum = hash + 1;
3609 break;
3610 }
3611 hash++;
drh2b51f212013-10-11 23:01:02 +00003612 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00003613 }
3614 if( types[hash]==0 ){
3615 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00003616 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00003617 if( types[hash]==0 ){
3618 fprintf(stderr,"Out of memory.\n");
3619 exit(1);
3620 }
drh898799f2014-01-10 23:21:00 +00003621 lemon_strcpy(types[hash],stddt);
drh75897232000-05-29 14:26:00 +00003622 }
3623 }
3624
3625 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3626 name = lemp->name ? lemp->name : "Parse";
3627 lineno = *plineno;
3628 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3629 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3630 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3631 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3632 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00003633 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003634 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3635 for(i=0; i<arraysize; i++){
3636 if( types[i]==0 ) continue;
3637 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3638 free(types[i]);
3639 }
drhc4dd3fd2008-01-22 01:48:05 +00003640 if( lemp->errsym->useCnt ){
3641 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3642 }
drh75897232000-05-29 14:26:00 +00003643 free(stddt);
3644 free(types);
3645 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3646 *plineno = lineno;
3647}
3648
drhb29b0a52002-02-23 19:39:46 +00003649/*
3650** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003651** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003652*/
drh8b582012003-10-21 13:16:03 +00003653static const char *minimum_size_type(int lwr, int upr){
3654 if( lwr>=0 ){
3655 if( upr<=255 ){
3656 return "unsigned char";
3657 }else if( upr<65535 ){
3658 return "unsigned short int";
3659 }else{
3660 return "unsigned int";
3661 }
3662 }else if( lwr>=-127 && upr<=127 ){
3663 return "signed char";
3664 }else if( lwr>=-32767 && upr<32767 ){
3665 return "short";
drhb29b0a52002-02-23 19:39:46 +00003666 }else{
drh8b582012003-10-21 13:16:03 +00003667 return "int";
drhb29b0a52002-02-23 19:39:46 +00003668 }
3669}
3670
drhfdbf9282003-10-21 16:34:41 +00003671/*
3672** Each state contains a set of token transaction and a set of
3673** nonterminal transactions. Each of these sets makes an instance
3674** of the following structure. An array of these structures is used
3675** to order the creation of entries in the yy_action[] table.
3676*/
3677struct axset {
3678 struct state *stp; /* A pointer to a state */
3679 int isTkn; /* True to use tokens. False for non-terminals */
3680 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00003681 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00003682};
3683
3684/*
3685** Compare to axset structures for sorting purposes
3686*/
3687static int axset_compare(const void *a, const void *b){
3688 struct axset *p1 = (struct axset*)a;
3689 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00003690 int c;
3691 c = p2->nAction - p1->nAction;
3692 if( c==0 ){
3693 c = p2->iOrder - p1->iOrder;
3694 }
3695 assert( c!=0 || p1==p2 );
3696 return c;
drhfdbf9282003-10-21 16:34:41 +00003697}
3698
drhc4dd3fd2008-01-22 01:48:05 +00003699/*
3700** Write text on "out" that describes the rule "rp".
3701*/
3702static void writeRuleText(FILE *out, struct rule *rp){
3703 int j;
3704 fprintf(out,"%s ::=", rp->lhs->name);
3705 for(j=0; j<rp->nrhs; j++){
3706 struct symbol *sp = rp->rhs[j];
drh61f92cd2014-01-11 03:06:18 +00003707 if( sp->type!=MULTITERMINAL ){
3708 fprintf(out," %s", sp->name);
3709 }else{
drhc4dd3fd2008-01-22 01:48:05 +00003710 int k;
drh61f92cd2014-01-11 03:06:18 +00003711 fprintf(out," %s", sp->subsym[0]->name);
drhc4dd3fd2008-01-22 01:48:05 +00003712 for(k=1; k<sp->nsubsym; k++){
3713 fprintf(out,"|%s",sp->subsym[k]->name);
3714 }
3715 }
3716 }
3717}
3718
3719
drh75897232000-05-29 14:26:00 +00003720/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00003721void ReportTable(
3722 struct lemon *lemp,
3723 int mhflag /* Output in makeheaders format if true */
3724){
drh75897232000-05-29 14:26:00 +00003725 FILE *out, *in;
3726 char line[LINESIZE];
3727 int lineno;
3728 struct state *stp;
3729 struct action *ap;
3730 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003731 struct acttab *pActtab;
icculusa3191192010-02-17 05:40:34 +00003732 int i, j, n;
icculus9e44cf12010-02-14 17:14:22 +00003733 const char *name;
drh8b582012003-10-21 13:16:03 +00003734 int mnTknOfst, mxTknOfst;
3735 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003736 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003737
3738 in = tplt_open(lemp);
3739 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003740 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003741 if( out==0 ){
3742 fclose(in);
3743 return;
3744 }
3745 lineno = 1;
3746 tplt_xfer(lemp->name,in,out,&lineno);
3747
3748 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00003749 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00003750 if( mhflag ){
3751 char *name = file_makename(lemp, ".h");
3752 fprintf(out,"#include \"%s\"\n", name); lineno++;
3753 free(name);
3754 }
3755 tplt_xfer(lemp->name,in,out,&lineno);
3756
3757 /* Generate #defines for all tokens */
3758 if( mhflag ){
icculus9e44cf12010-02-14 17:14:22 +00003759 const char *prefix;
drh75897232000-05-29 14:26:00 +00003760 fprintf(out,"#if INTERFACE\n"); lineno++;
3761 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3762 else prefix = "";
3763 for(i=1; i<lemp->nterminal; i++){
3764 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3765 lineno++;
3766 }
3767 fprintf(out,"#endif\n"); lineno++;
3768 }
3769 tplt_xfer(lemp->name,in,out,&lineno);
3770
3771 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003772 fprintf(out,"#define YYCODETYPE %s\n",
drh8a415d32009-06-12 13:53:51 +00003773 minimum_size_type(0, lemp->nsymbol+1)); lineno++;
drh75897232000-05-29 14:26:00 +00003774 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3775 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003776 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003777 if( lemp->wildcard ){
3778 fprintf(out,"#define YYWILDCARD %d\n",
3779 lemp->wildcard->index); lineno++;
3780 }
drh75897232000-05-29 14:26:00 +00003781 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003782 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003783 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003784 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3785 }else{
3786 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3787 }
drhca44b5a2007-02-22 23:06:58 +00003788 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003789 if( mhflag ){
3790 fprintf(out,"#if INTERFACE\n"); lineno++;
3791 }
3792 name = lemp->name ? lemp->name : "Parse";
3793 if( lemp->arg && lemp->arg[0] ){
3794 int i;
drh87cf1372008-08-13 20:09:06 +00003795 i = lemonStrlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003796 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3797 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003798 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3799 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3800 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3801 name,lemp->arg,&lemp->arg[i]); lineno++;
3802 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3803 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003804 }else{
drh1f245e42002-03-11 13:55:50 +00003805 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3806 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3807 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3808 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003809 }
3810 if( mhflag ){
3811 fprintf(out,"#endif\n"); lineno++;
3812 }
3813 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3814 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00003815 if( lemp->errsym->useCnt ){
3816 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3817 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
3818 }
drh0bd1f4e2002-06-06 18:54:39 +00003819 if( lemp->has_fallback ){
3820 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3821 }
drh75897232000-05-29 14:26:00 +00003822 tplt_xfer(lemp->name,in,out,&lineno);
3823
drh8b582012003-10-21 13:16:03 +00003824 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003825 **
drh8b582012003-10-21 13:16:03 +00003826 ** yy_action[] A single table containing all actions.
3827 ** yy_lookahead[] A table containing the lookahead for each entry in
3828 ** yy_action. Used to detect hash collisions.
3829 ** yy_shift_ofst[] For each state, the offset into yy_action for
3830 ** shifting terminals.
3831 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3832 ** shifting non-terminals after a reduce.
3833 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003834 */
drh75897232000-05-29 14:26:00 +00003835
drh8b582012003-10-21 13:16:03 +00003836 /* Compute the actions on all states and count them up */
icculus9e44cf12010-02-14 17:14:22 +00003837 ax = (struct axset *) calloc(lemp->nstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00003838 if( ax==0 ){
3839 fprintf(stderr,"malloc failed\n");
3840 exit(1);
3841 }
drh75897232000-05-29 14:26:00 +00003842 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003843 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003844 ax[i*2].stp = stp;
3845 ax[i*2].isTkn = 1;
3846 ax[i*2].nAction = stp->nTknAct;
3847 ax[i*2+1].stp = stp;
3848 ax[i*2+1].isTkn = 0;
3849 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003850 }
drh8b582012003-10-21 13:16:03 +00003851 mxTknOfst = mnTknOfst = 0;
3852 mxNtOfst = mnNtOfst = 0;
3853
drhfdbf9282003-10-21 16:34:41 +00003854 /* Compute the action table. In order to try to keep the size of the
3855 ** action table to a minimum, the heuristic of placing the largest action
3856 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003857 */
drhe594bc32009-11-03 13:02:25 +00003858 for(i=0; i<lemp->nstate*2; i++) ax[i].iOrder = i;
drhfdbf9282003-10-21 16:34:41 +00003859 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003860 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003861 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3862 stp = ax[i].stp;
3863 if( ax[i].isTkn ){
3864 for(ap=stp->ap; ap; ap=ap->next){
3865 int action;
3866 if( ap->sp->index>=lemp->nterminal ) continue;
3867 action = compute_action(lemp, ap);
3868 if( action<0 ) continue;
3869 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003870 }
drhfdbf9282003-10-21 16:34:41 +00003871 stp->iTknOfst = acttab_insert(pActtab);
3872 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3873 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3874 }else{
3875 for(ap=stp->ap; ap; ap=ap->next){
3876 int action;
3877 if( ap->sp->index<lemp->nterminal ) continue;
3878 if( ap->sp->index==lemp->nsymbol ) continue;
3879 action = compute_action(lemp, ap);
3880 if( action<0 ) continue;
3881 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003882 }
drhfdbf9282003-10-21 16:34:41 +00003883 stp->iNtOfst = acttab_insert(pActtab);
3884 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3885 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003886 }
3887 }
drhfdbf9282003-10-21 16:34:41 +00003888 free(ax);
drh8b582012003-10-21 13:16:03 +00003889
3890 /* Output the yy_action table */
drh8b582012003-10-21 13:16:03 +00003891 n = acttab_size(pActtab);
drhf16371d2009-11-03 19:18:31 +00003892 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
3893 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003894 for(i=j=0; i<n; i++){
3895 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003896 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003897 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003898 fprintf(out, " %4d,", action);
3899 if( j==9 || i==n-1 ){
3900 fprintf(out, "\n"); lineno++;
3901 j = 0;
3902 }else{
3903 j++;
3904 }
3905 }
3906 fprintf(out, "};\n"); lineno++;
3907
3908 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003909 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003910 for(i=j=0; i<n; i++){
3911 int la = acttab_yylookahead(pActtab, i);
3912 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003913 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003914 fprintf(out, " %4d,", la);
3915 if( j==9 || i==n-1 ){
3916 fprintf(out, "\n"); lineno++;
3917 j = 0;
3918 }else{
3919 j++;
3920 }
3921 }
3922 fprintf(out, "};\n"); lineno++;
3923
3924 /* Output the yy_shift_ofst[] table */
3925 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003926 n = lemp->nstate;
3927 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003928 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
3929 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
3930 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003931 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003932 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003933 for(i=j=0; i<n; i++){
3934 int ofst;
3935 stp = lemp->sorted[i];
3936 ofst = stp->iTknOfst;
3937 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003938 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003939 fprintf(out, " %4d,", ofst);
3940 if( j==9 || i==n-1 ){
3941 fprintf(out, "\n"); lineno++;
3942 j = 0;
3943 }else{
3944 j++;
3945 }
3946 }
3947 fprintf(out, "};\n"); lineno++;
3948
3949 /* Output the yy_reduce_ofst[] table */
3950 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003951 n = lemp->nstate;
3952 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003953 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
3954 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
3955 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003956 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003957 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003958 for(i=j=0; i<n; i++){
3959 int ofst;
3960 stp = lemp->sorted[i];
3961 ofst = stp->iNtOfst;
3962 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003963 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003964 fprintf(out, " %4d,", ofst);
3965 if( j==9 || i==n-1 ){
3966 fprintf(out, "\n"); lineno++;
3967 j = 0;
3968 }else{
3969 j++;
3970 }
3971 }
3972 fprintf(out, "};\n"); lineno++;
3973
3974 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003975 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003976 n = lemp->nstate;
3977 for(i=j=0; i<n; i++){
3978 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003979 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003980 fprintf(out, " %4d,", stp->iDflt);
3981 if( j==9 || i==n-1 ){
3982 fprintf(out, "\n"); lineno++;
3983 j = 0;
3984 }else{
3985 j++;
3986 }
3987 }
3988 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003989 tplt_xfer(lemp->name,in,out,&lineno);
3990
drh0bd1f4e2002-06-06 18:54:39 +00003991 /* Generate the table of fallback tokens.
3992 */
3993 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00003994 int mx = lemp->nterminal - 1;
3995 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
3996 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00003997 struct symbol *p = lemp->symbols[i];
3998 if( p->fallback==0 ){
3999 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
4000 }else{
4001 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
4002 p->name, p->fallback->name);
4003 }
4004 lineno++;
4005 }
4006 }
4007 tplt_xfer(lemp->name, in, out, &lineno);
4008
4009 /* Generate a table containing the symbolic name of every symbol
4010 */
drh75897232000-05-29 14:26:00 +00004011 for(i=0; i<lemp->nsymbol; i++){
drh898799f2014-01-10 23:21:00 +00004012 lemon_sprintf(line,"\"%s\",",lemp->symbols[i]->name);
drh75897232000-05-29 14:26:00 +00004013 fprintf(out," %-15s",line);
4014 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
4015 }
4016 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
4017 tplt_xfer(lemp->name,in,out,&lineno);
4018
drh0bd1f4e2002-06-06 18:54:39 +00004019 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00004020 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00004021 ** when tracing REDUCE actions.
4022 */
4023 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
4024 assert( rp->index==i );
drhc4dd3fd2008-01-22 01:48:05 +00004025 fprintf(out," /* %3d */ \"", i);
4026 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00004027 fprintf(out,"\",\n"); lineno++;
4028 }
4029 tplt_xfer(lemp->name,in,out,&lineno);
4030
drh75897232000-05-29 14:26:00 +00004031 /* Generate code which executes every time a symbol is popped from
4032 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00004033 ** (In other words, generate the %destructor actions)
4034 */
drh75897232000-05-29 14:26:00 +00004035 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00004036 int once = 1;
drh75897232000-05-29 14:26:00 +00004037 for(i=0; i<lemp->nsymbol; i++){
4038 struct symbol *sp = lemp->symbols[i];
4039 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004040 if( once ){
4041 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
4042 once = 0;
4043 }
drhc53eed12009-06-12 17:46:19 +00004044 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00004045 }
4046 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
4047 if( i<lemp->nsymbol ){
4048 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4049 fprintf(out," break;\n"); lineno++;
4050 }
4051 }
drh8d659732005-01-13 23:54:06 +00004052 if( lemp->vardest ){
4053 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00004054 int once = 1;
drh8d659732005-01-13 23:54:06 +00004055 for(i=0; i<lemp->nsymbol; i++){
4056 struct symbol *sp = lemp->symbols[i];
4057 if( sp==0 || sp->type==TERMINAL ||
4058 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00004059 if( once ){
4060 fprintf(out, " /* Default NON-TERMINAL Destructor */\n"); lineno++;
4061 once = 0;
4062 }
drhc53eed12009-06-12 17:46:19 +00004063 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00004064 dflt_sp = sp;
4065 }
4066 if( dflt_sp!=0 ){
4067 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00004068 }
drh4dc8ef52008-07-01 17:13:57 +00004069 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00004070 }
drh75897232000-05-29 14:26:00 +00004071 for(i=0; i<lemp->nsymbol; i++){
4072 struct symbol *sp = lemp->symbols[i];
4073 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh75013012009-06-12 15:47:34 +00004074 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004075
4076 /* Combine duplicate destructors into a single case */
4077 for(j=i+1; j<lemp->nsymbol; j++){
4078 struct symbol *sp2 = lemp->symbols[j];
4079 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
4080 && sp2->dtnum==sp->dtnum
4081 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00004082 fprintf(out," case %d: /* %s */\n",
4083 sp2->index, sp2->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004084 sp2->destructor = 0;
4085 }
4086 }
4087
drh75897232000-05-29 14:26:00 +00004088 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
4089 fprintf(out," break;\n"); lineno++;
4090 }
drh75897232000-05-29 14:26:00 +00004091 tplt_xfer(lemp->name,in,out,&lineno);
4092
4093 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00004094 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00004095 tplt_xfer(lemp->name,in,out,&lineno);
4096
4097 /* Generate the table of rule information
4098 **
4099 ** Note: This code depends on the fact that rules are number
4100 ** sequentually beginning with 0.
4101 */
4102 for(rp=lemp->rule; rp; rp=rp->next){
4103 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
4104 }
4105 tplt_xfer(lemp->name,in,out,&lineno);
4106
4107 /* Generate code which execution during each REDUCE action */
4108 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00004109 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00004110 }
drhc53eed12009-06-12 17:46:19 +00004111 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00004112 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00004113 struct rule *rp2; /* Other rules with the same action */
drh0bb132b2004-07-20 14:06:51 +00004114 if( rp->code==0 ) continue;
drhc53eed12009-06-12 17:46:19 +00004115 if( rp->code[0]=='\n' && rp->code[1]==0 ) continue; /* Will be default: */
drhc4dd3fd2008-01-22 01:48:05 +00004116 fprintf(out," case %d: /* ", rp->index);
4117 writeRuleText(out, rp);
4118 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004119 for(rp2=rp->next; rp2; rp2=rp2->next){
4120 if( rp2->code==rp->code ){
drhc4dd3fd2008-01-22 01:48:05 +00004121 fprintf(out," case %d: /* ", rp2->index);
4122 writeRuleText(out, rp2);
drh8a415d32009-06-12 13:53:51 +00004123 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00004124 rp2->code = 0;
4125 }
4126 }
drh75897232000-05-29 14:26:00 +00004127 emit_code(out,rp,lemp,&lineno);
4128 fprintf(out," break;\n"); lineno++;
drhc53eed12009-06-12 17:46:19 +00004129 rp->code = 0;
drh75897232000-05-29 14:26:00 +00004130 }
drhc53eed12009-06-12 17:46:19 +00004131 /* Finally, output the default: rule. We choose as the default: all
4132 ** empty actions. */
4133 fprintf(out," default:\n"); lineno++;
4134 for(rp=lemp->rule; rp; rp=rp->next){
4135 if( rp->code==0 ) continue;
4136 assert( rp->code[0]=='\n' && rp->code[1]==0 );
4137 fprintf(out," /* (%d) ", rp->index);
4138 writeRuleText(out, rp);
4139 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->index); lineno++;
4140 }
4141 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00004142 tplt_xfer(lemp->name,in,out,&lineno);
4143
4144 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00004145 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00004146 tplt_xfer(lemp->name,in,out,&lineno);
4147
4148 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00004149 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00004150 tplt_xfer(lemp->name,in,out,&lineno);
4151
4152 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004153 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004154 tplt_xfer(lemp->name,in,out,&lineno);
4155
4156 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004157 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004158
4159 fclose(in);
4160 fclose(out);
4161 return;
4162}
4163
4164/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004165void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004166{
4167 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004168 const char *prefix;
drh75897232000-05-29 14:26:00 +00004169 char line[LINESIZE];
4170 char pattern[LINESIZE];
4171 int i;
4172
4173 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4174 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004175 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004176 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004177 int nextChar;
drh75897232000-05-29 14:26:00 +00004178 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
drh61f92cd2014-01-11 03:06:18 +00004179 lemon_sprintf(pattern,"#define %s%-30s %3d\n",
4180 prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004181 if( strcmp(line,pattern) ) break;
4182 }
drh8ba0d1c2012-06-16 15:26:31 +00004183 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004184 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004185 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004186 /* No change in the file. Don't rewrite it. */
4187 return;
4188 }
4189 }
drh2aa6ca42004-09-10 00:14:04 +00004190 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004191 if( out ){
4192 for(i=1; i<lemp->nterminal; i++){
drh61f92cd2014-01-11 03:06:18 +00004193 fprintf(out,"#define %s%-30s %3d\n",prefix,lemp->symbols[i]->name,i);
drh75897232000-05-29 14:26:00 +00004194 }
4195 fclose(out);
4196 }
4197 return;
4198}
4199
4200/* Reduce the size of the action tables, if possible, by making use
4201** of defaults.
4202**
drhb59499c2002-02-23 18:45:13 +00004203** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004204** it the default. Except, there is no default if the wildcard token
4205** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004206*/
icculus9e44cf12010-02-14 17:14:22 +00004207void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004208{
4209 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00004210 struct action *ap, *ap2;
4211 struct rule *rp, *rp2, *rbest;
4212 int nbest, n;
drh75897232000-05-29 14:26:00 +00004213 int i;
drhe09daa92006-06-10 13:29:31 +00004214 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004215
4216 for(i=0; i<lemp->nstate; i++){
4217 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004218 nbest = 0;
4219 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004220 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004221
drhb59499c2002-02-23 18:45:13 +00004222 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004223 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4224 usesWildcard = 1;
4225 }
drhb59499c2002-02-23 18:45:13 +00004226 if( ap->type!=REDUCE ) continue;
4227 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004228 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004229 if( rp==rbest ) continue;
4230 n = 1;
4231 for(ap2=ap->next; ap2; ap2=ap2->next){
4232 if( ap2->type!=REDUCE ) continue;
4233 rp2 = ap2->x.rp;
4234 if( rp2==rbest ) continue;
4235 if( rp2==rp ) n++;
4236 }
4237 if( n>nbest ){
4238 nbest = n;
4239 rbest = rp;
drh75897232000-05-29 14:26:00 +00004240 }
4241 }
drhb59499c2002-02-23 18:45:13 +00004242
4243 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004244 ** is not at least 1 or if the wildcard token is a possible
4245 ** lookahead.
4246 */
4247 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004248
drhb59499c2002-02-23 18:45:13 +00004249
4250 /* Combine matching REDUCE actions into a single default */
4251 for(ap=stp->ap; ap; ap=ap->next){
4252 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4253 }
drh75897232000-05-29 14:26:00 +00004254 assert( ap );
4255 ap->sp = Symbol_new("{default}");
4256 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004257 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004258 }
4259 stp->ap = Action_sort(stp->ap);
4260 }
4261}
drhb59499c2002-02-23 18:45:13 +00004262
drhada354d2005-11-05 15:03:59 +00004263
4264/*
4265** Compare two states for sorting purposes. The smaller state is the
4266** one with the most non-terminal actions. If they have the same number
4267** of non-terminal actions, then the smaller is the one with the most
4268** token actions.
4269*/
4270static int stateResortCompare(const void *a, const void *b){
4271 const struct state *pA = *(const struct state**)a;
4272 const struct state *pB = *(const struct state**)b;
4273 int n;
4274
4275 n = pB->nNtAct - pA->nNtAct;
4276 if( n==0 ){
4277 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004278 if( n==0 ){
4279 n = pB->statenum - pA->statenum;
4280 }
drhada354d2005-11-05 15:03:59 +00004281 }
drhe594bc32009-11-03 13:02:25 +00004282 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004283 return n;
4284}
4285
4286
4287/*
4288** Renumber and resort states so that states with fewer choices
4289** occur at the end. Except, keep state 0 as the first state.
4290*/
icculus9e44cf12010-02-14 17:14:22 +00004291void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00004292{
4293 int i;
4294 struct state *stp;
4295 struct action *ap;
4296
4297 for(i=0; i<lemp->nstate; i++){
4298 stp = lemp->sorted[i];
4299 stp->nTknAct = stp->nNtAct = 0;
4300 stp->iDflt = lemp->nstate + lemp->nrule;
4301 stp->iTknOfst = NO_OFFSET;
4302 stp->iNtOfst = NO_OFFSET;
4303 for(ap=stp->ap; ap; ap=ap->next){
4304 if( compute_action(lemp,ap)>=0 ){
4305 if( ap->sp->index<lemp->nterminal ){
4306 stp->nTknAct++;
4307 }else if( ap->sp->index<lemp->nsymbol ){
4308 stp->nNtAct++;
4309 }else{
4310 stp->iDflt = compute_action(lemp, ap);
4311 }
4312 }
4313 }
4314 }
4315 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4316 stateResortCompare);
4317 for(i=0; i<lemp->nstate; i++){
4318 lemp->sorted[i]->statenum = i;
4319 }
4320}
4321
4322
drh75897232000-05-29 14:26:00 +00004323/***************** From the file "set.c" ************************************/
4324/*
4325** Set manipulation routines for the LEMON parser generator.
4326*/
4327
4328static int size = 0;
4329
4330/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00004331void SetSize(int n)
drh75897232000-05-29 14:26:00 +00004332{
4333 size = n+1;
4334}
4335
4336/* Allocate a new set */
4337char *SetNew(){
4338 char *s;
drh9892c5d2007-12-21 00:02:11 +00004339 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004340 if( s==0 ){
4341 extern void memory_error();
4342 memory_error();
4343 }
drh75897232000-05-29 14:26:00 +00004344 return s;
4345}
4346
4347/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00004348void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00004349{
4350 free(s);
4351}
4352
4353/* Add a new element to the set. Return TRUE if the element was added
4354** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00004355int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00004356{
4357 int rv;
drh9892c5d2007-12-21 00:02:11 +00004358 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004359 rv = s[e];
4360 s[e] = 1;
4361 return !rv;
4362}
4363
4364/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00004365int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00004366{
4367 int i, progress;
4368 progress = 0;
4369 for(i=0; i<size; i++){
4370 if( s2[i]==0 ) continue;
4371 if( s1[i]==0 ){
4372 progress = 1;
4373 s1[i] = 1;
4374 }
4375 }
4376 return progress;
4377}
4378/********************** From the file "table.c" ****************************/
4379/*
4380** All code in this file has been automatically generated
4381** from a specification in the file
4382** "table.q"
4383** by the associative array code building program "aagen".
4384** Do not edit this file! Instead, edit the specification
4385** file, then rerun aagen.
4386*/
4387/*
4388** Code for processing tables in the LEMON parser generator.
4389*/
4390
drh01f75f22013-10-02 20:46:30 +00004391PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00004392{
drh01f75f22013-10-02 20:46:30 +00004393 unsigned h = 0;
4394 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00004395 return h;
4396}
4397
4398/* Works like strdup, sort of. Save a string in malloced memory, but
4399** keep strings in a table so that the same string is not in more
4400** than one place.
4401*/
icculus9e44cf12010-02-14 17:14:22 +00004402const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00004403{
icculus9e44cf12010-02-14 17:14:22 +00004404 const char *z;
4405 char *cpy;
drh75897232000-05-29 14:26:00 +00004406
drh916f75f2006-07-17 00:19:39 +00004407 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004408 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00004409 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
drh898799f2014-01-10 23:21:00 +00004410 lemon_strcpy(cpy,y);
icculus9e44cf12010-02-14 17:14:22 +00004411 z = cpy;
drh75897232000-05-29 14:26:00 +00004412 Strsafe_insert(z);
4413 }
4414 MemoryCheck(z);
4415 return z;
4416}
4417
4418/* There is one instance of the following structure for each
4419** associative array of type "x1".
4420*/
4421struct s_x1 {
4422 int size; /* The number of available slots. */
4423 /* Must be a power of 2 greater than or */
4424 /* equal to 1 */
4425 int count; /* Number of currently slots filled */
4426 struct s_x1node *tbl; /* The data stored here */
4427 struct s_x1node **ht; /* Hash table for lookups */
4428};
4429
4430/* There is one instance of this structure for every data element
4431** in an associative array of type "x1".
4432*/
4433typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00004434 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00004435 struct s_x1node *next; /* Next entry with the same hash */
4436 struct s_x1node **from; /* Previous link */
4437} x1node;
4438
4439/* There is only one instance of the array, which is the following */
4440static struct s_x1 *x1a;
4441
4442/* Allocate a new associative array */
4443void Strsafe_init(){
4444 if( x1a ) return;
4445 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4446 if( x1a ){
4447 x1a->size = 1024;
4448 x1a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00004449 x1a->tbl = (x1node*)calloc(1024, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00004450 if( x1a->tbl==0 ){
4451 free(x1a);
4452 x1a = 0;
4453 }else{
4454 int i;
4455 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4456 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4457 }
4458 }
4459}
4460/* Insert a new record into the array. Return TRUE if successful.
4461** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004462int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00004463{
4464 x1node *np;
drh01f75f22013-10-02 20:46:30 +00004465 unsigned h;
4466 unsigned ph;
drh75897232000-05-29 14:26:00 +00004467
4468 if( x1a==0 ) return 0;
4469 ph = strhash(data);
4470 h = ph & (x1a->size-1);
4471 np = x1a->ht[h];
4472 while( np ){
4473 if( strcmp(np->data,data)==0 ){
4474 /* An existing entry with the same key is found. */
4475 /* Fail because overwrite is not allows. */
4476 return 0;
4477 }
4478 np = np->next;
4479 }
4480 if( x1a->count>=x1a->size ){
4481 /* Need to make the hash table bigger */
4482 int i,size;
4483 struct s_x1 array;
4484 array.size = size = x1a->size*2;
4485 array.count = x1a->count;
drh03e1b1f2014-01-11 12:52:25 +00004486 array.tbl = (x1node*)calloc(size, sizeof(x1node) + sizeof(x1node*));
drh75897232000-05-29 14:26:00 +00004487 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4488 array.ht = (x1node**)&(array.tbl[size]);
4489 for(i=0; i<size; i++) array.ht[i] = 0;
4490 for(i=0; i<x1a->count; i++){
4491 x1node *oldnp, *newnp;
4492 oldnp = &(x1a->tbl[i]);
4493 h = strhash(oldnp->data) & (size-1);
4494 newnp = &(array.tbl[i]);
4495 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4496 newnp->next = array.ht[h];
4497 newnp->data = oldnp->data;
4498 newnp->from = &(array.ht[h]);
4499 array.ht[h] = newnp;
4500 }
4501 free(x1a->tbl);
4502 *x1a = array;
4503 }
4504 /* Insert the new data */
4505 h = ph & (x1a->size-1);
4506 np = &(x1a->tbl[x1a->count++]);
4507 np->data = data;
4508 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4509 np->next = x1a->ht[h];
4510 x1a->ht[h] = np;
4511 np->from = &(x1a->ht[h]);
4512 return 1;
4513}
4514
4515/* Return a pointer to data assigned to the given key. Return NULL
4516** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004517const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00004518{
drh01f75f22013-10-02 20:46:30 +00004519 unsigned h;
drh75897232000-05-29 14:26:00 +00004520 x1node *np;
4521
4522 if( x1a==0 ) return 0;
4523 h = strhash(key) & (x1a->size-1);
4524 np = x1a->ht[h];
4525 while( np ){
4526 if( strcmp(np->data,key)==0 ) break;
4527 np = np->next;
4528 }
4529 return np ? np->data : 0;
4530}
4531
4532/* Return a pointer to the (terminal or nonterminal) symbol "x".
4533** Create a new symbol if this is the first time "x" has been seen.
4534*/
icculus9e44cf12010-02-14 17:14:22 +00004535struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00004536{
4537 struct symbol *sp;
4538
4539 sp = Symbol_find(x);
4540 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00004541 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00004542 MemoryCheck(sp);
4543 sp->name = Strsafe(x);
4544 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4545 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004546 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004547 sp->prec = -1;
4548 sp->assoc = UNK;
4549 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00004550 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00004551 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00004552 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00004553 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00004554 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00004555 Symbol_insert(sp,sp->name);
4556 }
drhc4dd3fd2008-01-22 01:48:05 +00004557 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00004558 return sp;
4559}
4560
drh61f92cd2014-01-11 03:06:18 +00004561/* Compare two symbols for sorting purposes. Return negative,
4562** zero, or positive if a is less then, equal to, or greater
4563** than b.
drh60d31652004-02-22 00:08:04 +00004564**
4565** Symbols that begin with upper case letters (terminals or tokens)
4566** must sort before symbols that begin with lower case letters
drh61f92cd2014-01-11 03:06:18 +00004567** (non-terminals). And MULTITERMINAL symbols (created using the
4568** %token_class directive) must sort at the very end. Other than
4569** that, the order does not matter.
drh60d31652004-02-22 00:08:04 +00004570**
4571** We find experimentally that leaving the symbols in their original
4572** order (the order they appeared in the grammar file) gives the
4573** smallest parser tables in SQLite.
4574*/
icculus9e44cf12010-02-14 17:14:22 +00004575int Symbolcmpp(const void *_a, const void *_b)
4576{
drh61f92cd2014-01-11 03:06:18 +00004577 const struct symbol *a = *(const struct symbol **) _a;
4578 const struct symbol *b = *(const struct symbol **) _b;
4579 int i1 = a->type==MULTITERMINAL ? 3 : a->name[0]>'Z' ? 2 : 1;
4580 int i2 = b->type==MULTITERMINAL ? 3 : b->name[0]>'Z' ? 2 : 1;
4581 return i1==i2 ? a->index - b->index : i1 - i2;
drh75897232000-05-29 14:26:00 +00004582}
4583
4584/* There is one instance of the following structure for each
4585** associative array of type "x2".
4586*/
4587struct s_x2 {
4588 int size; /* The number of available slots. */
4589 /* Must be a power of 2 greater than or */
4590 /* equal to 1 */
4591 int count; /* Number of currently slots filled */
4592 struct s_x2node *tbl; /* The data stored here */
4593 struct s_x2node **ht; /* Hash table for lookups */
4594};
4595
4596/* There is one instance of this structure for every data element
4597** in an associative array of type "x2".
4598*/
4599typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00004600 struct symbol *data; /* The data */
4601 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00004602 struct s_x2node *next; /* Next entry with the same hash */
4603 struct s_x2node **from; /* Previous link */
4604} x2node;
4605
4606/* There is only one instance of the array, which is the following */
4607static struct s_x2 *x2a;
4608
4609/* Allocate a new associative array */
4610void Symbol_init(){
4611 if( x2a ) return;
4612 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4613 if( x2a ){
4614 x2a->size = 128;
4615 x2a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00004616 x2a->tbl = (x2node*)calloc(128, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00004617 if( x2a->tbl==0 ){
4618 free(x2a);
4619 x2a = 0;
4620 }else{
4621 int i;
4622 x2a->ht = (x2node**)&(x2a->tbl[128]);
4623 for(i=0; i<128; i++) x2a->ht[i] = 0;
4624 }
4625 }
4626}
4627/* Insert a new record into the array. Return TRUE if successful.
4628** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004629int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00004630{
4631 x2node *np;
drh01f75f22013-10-02 20:46:30 +00004632 unsigned h;
4633 unsigned ph;
drh75897232000-05-29 14:26:00 +00004634
4635 if( x2a==0 ) return 0;
4636 ph = strhash(key);
4637 h = ph & (x2a->size-1);
4638 np = x2a->ht[h];
4639 while( np ){
4640 if( strcmp(np->key,key)==0 ){
4641 /* An existing entry with the same key is found. */
4642 /* Fail because overwrite is not allows. */
4643 return 0;
4644 }
4645 np = np->next;
4646 }
4647 if( x2a->count>=x2a->size ){
4648 /* Need to make the hash table bigger */
4649 int i,size;
4650 struct s_x2 array;
4651 array.size = size = x2a->size*2;
4652 array.count = x2a->count;
drh03e1b1f2014-01-11 12:52:25 +00004653 array.tbl = (x2node*)calloc(size, sizeof(x2node) + sizeof(x2node*));
drh75897232000-05-29 14:26:00 +00004654 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;
drh03e1b1f2014-01-11 12:52:25 +00004813 x3a->tbl = (x3node*)calloc(128, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00004814 if( x3a->tbl==0 ){
4815 free(x3a);
4816 x3a = 0;
4817 }else{
4818 int i;
4819 x3a->ht = (x3node**)&(x3a->tbl[128]);
4820 for(i=0; i<128; i++) x3a->ht[i] = 0;
4821 }
4822 }
4823}
4824/* Insert a new record into the array. Return TRUE if successful.
4825** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004826int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00004827{
4828 x3node *np;
drh01f75f22013-10-02 20:46:30 +00004829 unsigned h;
4830 unsigned ph;
drh75897232000-05-29 14:26:00 +00004831
4832 if( x3a==0 ) return 0;
4833 ph = statehash(key);
4834 h = ph & (x3a->size-1);
4835 np = x3a->ht[h];
4836 while( np ){
4837 if( statecmp(np->key,key)==0 ){
4838 /* An existing entry with the same key is found. */
4839 /* Fail because overwrite is not allows. */
4840 return 0;
4841 }
4842 np = np->next;
4843 }
4844 if( x3a->count>=x3a->size ){
4845 /* Need to make the hash table bigger */
4846 int i,size;
4847 struct s_x3 array;
4848 array.size = size = x3a->size*2;
4849 array.count = x3a->count;
drh03e1b1f2014-01-11 12:52:25 +00004850 array.tbl = (x3node*)calloc(size, sizeof(x3node) + sizeof(x3node*));
drh75897232000-05-29 14:26:00 +00004851 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4852 array.ht = (x3node**)&(array.tbl[size]);
4853 for(i=0; i<size; i++) array.ht[i] = 0;
4854 for(i=0; i<x3a->count; i++){
4855 x3node *oldnp, *newnp;
4856 oldnp = &(x3a->tbl[i]);
4857 h = statehash(oldnp->key) & (size-1);
4858 newnp = &(array.tbl[i]);
4859 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4860 newnp->next = array.ht[h];
4861 newnp->key = oldnp->key;
4862 newnp->data = oldnp->data;
4863 newnp->from = &(array.ht[h]);
4864 array.ht[h] = newnp;
4865 }
4866 free(x3a->tbl);
4867 *x3a = array;
4868 }
4869 /* Insert the new data */
4870 h = ph & (x3a->size-1);
4871 np = &(x3a->tbl[x3a->count++]);
4872 np->key = key;
4873 np->data = data;
4874 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4875 np->next = x3a->ht[h];
4876 x3a->ht[h] = np;
4877 np->from = &(x3a->ht[h]);
4878 return 1;
4879}
4880
4881/* Return a pointer to data assigned to the given key. Return NULL
4882** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004883struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004884{
drh01f75f22013-10-02 20:46:30 +00004885 unsigned h;
drh75897232000-05-29 14:26:00 +00004886 x3node *np;
4887
4888 if( x3a==0 ) return 0;
4889 h = statehash(key) & (x3a->size-1);
4890 np = x3a->ht[h];
4891 while( np ){
4892 if( statecmp(np->key,key)==0 ) break;
4893 np = np->next;
4894 }
4895 return np ? np->data : 0;
4896}
4897
4898/* Return an array of pointers to all data in the table.
4899** The array is obtained from malloc. Return NULL if memory allocation
4900** problems, or if the array is empty. */
4901struct state **State_arrayof()
4902{
4903 struct state **array;
4904 int i,size;
4905 if( x3a==0 ) return 0;
4906 size = x3a->count;
drh03e1b1f2014-01-11 12:52:25 +00004907 array = (struct state **)calloc(size, sizeof(struct state *));
drh75897232000-05-29 14:26:00 +00004908 if( array ){
4909 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4910 }
4911 return array;
4912}
4913
4914/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00004915PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00004916{
drh01f75f22013-10-02 20:46:30 +00004917 unsigned h=0;
drh75897232000-05-29 14:26:00 +00004918 h = h*571 + a->rp->index*37 + a->dot;
4919 return h;
4920}
4921
4922/* There is one instance of the following structure for each
4923** associative array of type "x4".
4924*/
4925struct s_x4 {
4926 int size; /* The number of available slots. */
4927 /* Must be a power of 2 greater than or */
4928 /* equal to 1 */
4929 int count; /* Number of currently slots filled */
4930 struct s_x4node *tbl; /* The data stored here */
4931 struct s_x4node **ht; /* Hash table for lookups */
4932};
4933
4934/* There is one instance of this structure for every data element
4935** in an associative array of type "x4".
4936*/
4937typedef struct s_x4node {
4938 struct config *data; /* The data */
4939 struct s_x4node *next; /* Next entry with the same hash */
4940 struct s_x4node **from; /* Previous link */
4941} x4node;
4942
4943/* There is only one instance of the array, which is the following */
4944static struct s_x4 *x4a;
4945
4946/* Allocate a new associative array */
4947void Configtable_init(){
4948 if( x4a ) return;
4949 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4950 if( x4a ){
4951 x4a->size = 64;
4952 x4a->count = 0;
drh03e1b1f2014-01-11 12:52:25 +00004953 x4a->tbl = (x4node*)calloc(64, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00004954 if( x4a->tbl==0 ){
4955 free(x4a);
4956 x4a = 0;
4957 }else{
4958 int i;
4959 x4a->ht = (x4node**)&(x4a->tbl[64]);
4960 for(i=0; i<64; i++) x4a->ht[i] = 0;
4961 }
4962 }
4963}
4964/* Insert a new record into the array. Return TRUE if successful.
4965** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004966int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00004967{
4968 x4node *np;
drh01f75f22013-10-02 20:46:30 +00004969 unsigned h;
4970 unsigned ph;
drh75897232000-05-29 14:26:00 +00004971
4972 if( x4a==0 ) return 0;
4973 ph = confighash(data);
4974 h = ph & (x4a->size-1);
4975 np = x4a->ht[h];
4976 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004977 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00004978 /* An existing entry with the same key is found. */
4979 /* Fail because overwrite is not allows. */
4980 return 0;
4981 }
4982 np = np->next;
4983 }
4984 if( x4a->count>=x4a->size ){
4985 /* Need to make the hash table bigger */
4986 int i,size;
4987 struct s_x4 array;
4988 array.size = size = x4a->size*2;
4989 array.count = x4a->count;
drh03e1b1f2014-01-11 12:52:25 +00004990 array.tbl = (x4node*)calloc(size, sizeof(x4node) + sizeof(x4node*));
drh75897232000-05-29 14:26:00 +00004991 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4992 array.ht = (x4node**)&(array.tbl[size]);
4993 for(i=0; i<size; i++) array.ht[i] = 0;
4994 for(i=0; i<x4a->count; i++){
4995 x4node *oldnp, *newnp;
4996 oldnp = &(x4a->tbl[i]);
4997 h = confighash(oldnp->data) & (size-1);
4998 newnp = &(array.tbl[i]);
4999 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
5000 newnp->next = array.ht[h];
5001 newnp->data = oldnp->data;
5002 newnp->from = &(array.ht[h]);
5003 array.ht[h] = newnp;
5004 }
5005 free(x4a->tbl);
5006 *x4a = array;
5007 }
5008 /* Insert the new data */
5009 h = ph & (x4a->size-1);
5010 np = &(x4a->tbl[x4a->count++]);
5011 np->data = data;
5012 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
5013 np->next = x4a->ht[h];
5014 x4a->ht[h] = np;
5015 np->from = &(x4a->ht[h]);
5016 return 1;
5017}
5018
5019/* Return a pointer to data assigned to the given key. Return NULL
5020** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00005021struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00005022{
5023 int h;
5024 x4node *np;
5025
5026 if( x4a==0 ) return 0;
5027 h = confighash(key) & (x4a->size-1);
5028 np = x4a->ht[h];
5029 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00005030 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00005031 np = np->next;
5032 }
5033 return np ? np->data : 0;
5034}
5035
5036/* Remove all data from the table. Pass each data to the function "f"
5037** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00005038void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00005039{
5040 int i;
5041 if( x4a==0 || x4a->count==0 ) return;
5042 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
5043 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
5044 x4a->count = 0;
5045 return;
5046}