blob: 58f13880f0e9d5689491b542407cb54aacdc654a [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
icculus9e44cf12010-02-14 17:14:22 +000053/* a few forward declarations... */
54struct rule;
55struct lemon;
56struct action;
57
drhe9278182007-07-18 18:16:29 +000058static struct action *Action_new(void);
59static struct action *Action_sort(struct action *);
drh75897232000-05-29 14:26:00 +000060
61/********** From the file "build.h" ************************************/
62void FindRulePrecedences();
63void FindFirstSets();
64void FindStates();
65void FindLinks();
66void FindFollowSets();
67void FindActions();
68
69/********* From the file "configlist.h" *********************************/
icculus9e44cf12010-02-14 17:14:22 +000070void Configlist_init(void);
71struct config *Configlist_add(struct rule *, int);
72struct config *Configlist_addbasis(struct rule *, int);
73void Configlist_closure(struct lemon *);
74void Configlist_sort(void);
75void Configlist_sortbasis(void);
76struct config *Configlist_return(void);
77struct config *Configlist_basis(void);
78void Configlist_eat(struct config *);
79void Configlist_reset(void);
drh75897232000-05-29 14:26:00 +000080
81/********* From the file "error.h" ***************************************/
drhf9a2e7b2003-04-15 01:49:48 +000082void ErrorMsg(const char *, int,const char *, ...);
drh75897232000-05-29 14:26:00 +000083
84/****** From the file "option.h" ******************************************/
icculus9e44cf12010-02-14 17:14:22 +000085enum option_type { OPT_FLAG=1, OPT_INT, OPT_DBL, OPT_STR,
86 OPT_FFLAG, OPT_FINT, OPT_FDBL, OPT_FSTR};
drh75897232000-05-29 14:26:00 +000087struct s_options {
icculus9e44cf12010-02-14 17:14:22 +000088 enum option_type type;
89 const char *label;
drh75897232000-05-29 14:26:00 +000090 char *arg;
icculus9e44cf12010-02-14 17:14:22 +000091 const char *message;
drh75897232000-05-29 14:26:00 +000092};
icculus9e44cf12010-02-14 17:14:22 +000093int OptInit(char**,struct s_options*,FILE*);
94int OptNArgs(void);
95char *OptArg(int);
96void OptErr(int);
97void OptPrint(void);
drh75897232000-05-29 14:26:00 +000098
99/******** From the file "parse.h" *****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000100void Parse(struct lemon *lemp);
drh75897232000-05-29 14:26:00 +0000101
102/********* From the file "plink.h" ***************************************/
icculus9e44cf12010-02-14 17:14:22 +0000103struct plink *Plink_new(void);
104void Plink_add(struct plink **, struct config *);
105void Plink_copy(struct plink **, struct plink *);
106void Plink_delete(struct plink *);
drh75897232000-05-29 14:26:00 +0000107
108/********** From the file "report.h" *************************************/
icculus9e44cf12010-02-14 17:14:22 +0000109void Reprint(struct lemon *);
110void ReportOutput(struct lemon *);
111void ReportTable(struct lemon *, int);
112void ReportHeader(struct lemon *);
113void CompressTables(struct lemon *);
114void ResortStates(struct lemon *);
drh75897232000-05-29 14:26:00 +0000115
116/********** From the file "set.h" ****************************************/
icculus9e44cf12010-02-14 17:14:22 +0000117void SetSize(int); /* All sets will be of size N */
118char *SetNew(void); /* A new set for element 0..N */
119void SetFree(char*); /* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +0000120int SetAdd(char*,int); /* Add element to a set */
121int SetUnion(char *,char *); /* A <- A U B, thru element N */
drh75897232000-05-29 14:26:00 +0000122#define SetFind(X,Y) (X[Y]) /* True if Y is in set X */
123
124/********** From the file "struct.h" *************************************/
125/*
126** Principal data structures for the LEMON parser generator.
127*/
128
drhaa9f1122007-08-23 02:50:56 +0000129typedef enum {LEMON_FALSE=0, LEMON_TRUE} Boolean;
drh75897232000-05-29 14:26:00 +0000130
131/* Symbols (terminals and nonterminals) of the grammar are stored
132** in the following: */
icculus9e44cf12010-02-14 17:14:22 +0000133enum symbol_type {
134 TERMINAL,
135 NONTERMINAL,
136 MULTITERMINAL
137};
138enum e_assoc {
drh75897232000-05-29 14:26:00 +0000139 LEFT,
140 RIGHT,
141 NONE,
142 UNK
icculus9e44cf12010-02-14 17:14:22 +0000143};
144struct symbol {
145 const char *name; /* Name of the symbol */
146 int index; /* Index number for this symbol */
147 enum symbol_type type; /* Symbols are all either TERMINALS or NTs */
148 struct rule *rule; /* Linked list of rules of this (if an NT) */
149 struct symbol *fallback; /* fallback token in case this token doesn't parse */
150 int prec; /* Precedence if defined (-1 otherwise) */
151 enum e_assoc assoc; /* Associativity if precedence is defined */
drh75897232000-05-29 14:26:00 +0000152 char *firstset; /* First-set for all rules of this symbol */
153 Boolean lambda; /* True if NT and can generate an empty string */
drhc4dd3fd2008-01-22 01:48:05 +0000154 int useCnt; /* Number of times used */
drh75897232000-05-29 14:26:00 +0000155 char *destructor; /* Code which executes whenever this symbol is
156 ** popped from the stack during error processing */
drh4dc8ef52008-07-01 17:13:57 +0000157 int destLineno; /* Line number for start of destructor */
drh75897232000-05-29 14:26:00 +0000158 char *datatype; /* The data type of information held by this
159 ** object. Only used if type==NONTERMINAL */
160 int dtnum; /* The data type number. In the parser, the value
161 ** stack is a union. The .yy%d element of this
162 ** union is the correct data type for this object */
drhfd405312005-11-06 04:06:59 +0000163 /* The following fields are used by MULTITERMINALs only */
164 int nsubsym; /* Number of constituent symbols in the MULTI */
165 struct symbol **subsym; /* Array of constituent symbols */
drh75897232000-05-29 14:26:00 +0000166};
167
168/* Each production rule in the grammar is stored in the following
169** structure. */
170struct rule {
171 struct symbol *lhs; /* Left-hand side of the rule */
icculus9e44cf12010-02-14 17:14:22 +0000172 const char *lhsalias; /* Alias for the LHS (NULL if none) */
drhb4960992007-10-05 16:16:36 +0000173 int lhsStart; /* True if left-hand side is the start symbol */
drh75897232000-05-29 14:26:00 +0000174 int ruleline; /* Line number for the rule */
175 int nrhs; /* Number of RHS symbols */
176 struct symbol **rhs; /* The RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +0000177 const char **rhsalias; /* An alias for each RHS symbol (NULL if none) */
drh75897232000-05-29 14:26:00 +0000178 int line; /* Line number at which code begins */
icculus9e44cf12010-02-14 17:14:22 +0000179 const char *code; /* The code executed when this rule is reduced */
drh75897232000-05-29 14:26:00 +0000180 struct symbol *precsym; /* Precedence symbol for this rule */
181 int index; /* An index number for this rule */
182 Boolean canReduce; /* True if this rule is ever reduced */
183 struct rule *nextlhs; /* Next rule with the same LHS */
184 struct rule *next; /* Next rule in the global list */
185};
186
187/* A configuration is a production rule of the grammar together with
188** a mark (dot) showing how much of that rule has been processed so far.
189** Configurations also contain a follow-set which is a list of terminal
190** symbols which are allowed to immediately follow the end of the rule.
191** Every configuration is recorded as an instance of the following: */
icculus9e44cf12010-02-14 17:14:22 +0000192enum cfgstatus {
193 COMPLETE,
194 INCOMPLETE
195};
drh75897232000-05-29 14:26:00 +0000196struct config {
197 struct rule *rp; /* The rule upon which the configuration is based */
198 int dot; /* The parse point */
199 char *fws; /* Follow-set for this configuration only */
200 struct plink *fplp; /* Follow-set forward propagation links */
201 struct plink *bplp; /* Follow-set backwards propagation links */
202 struct state *stp; /* Pointer to state which contains this */
icculus9e44cf12010-02-14 17:14:22 +0000203 enum cfgstatus status; /* used during followset and shift computations */
drh75897232000-05-29 14:26:00 +0000204 struct config *next; /* Next configuration in the state */
205 struct config *bp; /* The next basis configuration */
206};
207
icculus9e44cf12010-02-14 17:14:22 +0000208enum e_action {
209 SHIFT,
210 ACCEPT,
211 REDUCE,
212 ERROR,
213 SSCONFLICT, /* A shift/shift conflict */
214 SRCONFLICT, /* Was a reduce, but part of a conflict */
215 RRCONFLICT, /* Was a reduce, but part of a conflict */
216 SH_RESOLVED, /* Was a shift. Precedence resolved conflict */
217 RD_RESOLVED, /* Was reduce. Precedence resolved conflict */
218 NOT_USED /* Deleted by compression */
219};
220
drh75897232000-05-29 14:26:00 +0000221/* Every shift or reduce operation is stored as one of the following */
222struct action {
223 struct symbol *sp; /* The look-ahead symbol */
icculus9e44cf12010-02-14 17:14:22 +0000224 enum e_action type;
drh75897232000-05-29 14:26:00 +0000225 union {
226 struct state *stp; /* The new state, if a shift */
227 struct rule *rp; /* The rule, if a reduce */
228 } x;
229 struct action *next; /* Next action for this state */
230 struct action *collide; /* Next action with the same hash */
231};
232
233/* Each state of the generated parser's finite state machine
234** is encoded as an instance of the following structure. */
235struct state {
236 struct config *bp; /* The basis configurations for this state */
237 struct config *cfp; /* All configurations in this set */
drh34ff57b2008-07-14 12:27:51 +0000238 int statenum; /* Sequential number for this state */
drh75897232000-05-29 14:26:00 +0000239 struct action *ap; /* Array of actions for this state */
drh8b582012003-10-21 13:16:03 +0000240 int nTknAct, nNtAct; /* Number of actions on terminals and nonterminals */
241 int iTknOfst, iNtOfst; /* yy_action[] offset for terminals and nonterms */
242 int iDflt; /* Default action */
drh75897232000-05-29 14:26:00 +0000243};
drh8b582012003-10-21 13:16:03 +0000244#define NO_OFFSET (-2147483647)
drh75897232000-05-29 14:26:00 +0000245
246/* A followset propagation link indicates that the contents of one
247** configuration followset should be propagated to another whenever
248** the first changes. */
249struct plink {
250 struct config *cfp; /* The configuration to which linked */
251 struct plink *next; /* The next propagate link */
252};
253
254/* The state vector for the entire parser generator is recorded as
255** follows. (LEMON uses no global variables and makes little use of
256** static variables. Fields in the following structure can be thought
257** of as begin global variables in the program.) */
258struct lemon {
259 struct state **sorted; /* Table of states sorted by state number */
260 struct rule *rule; /* List of all rules */
261 int nstate; /* Number of states */
262 int nrule; /* Number of rules */
263 int nsymbol; /* Number of terminal and nonterminal symbols */
264 int nterminal; /* Number of terminal symbols */
265 struct symbol **symbols; /* Sorted array of pointers to symbols */
266 int errorcnt; /* Number of errors */
267 struct symbol *errsym; /* The error symbol */
drhe09daa92006-06-10 13:29:31 +0000268 struct symbol *wildcard; /* Token that matches anything */
drh75897232000-05-29 14:26:00 +0000269 char *name; /* Name of the generated parser */
270 char *arg; /* Declaration of the 3th argument to parser */
271 char *tokentype; /* Type of terminal symbols in the parser stack */
drh960e8c62001-04-03 16:53:21 +0000272 char *vartype; /* The default type of non-terminal symbols */
drh75897232000-05-29 14:26:00 +0000273 char *start; /* Name of the start symbol for the grammar */
274 char *stacksize; /* Size of the parser stack */
275 char *include; /* Code to put at the start of the C file */
drh75897232000-05-29 14:26:00 +0000276 char *error; /* Code to execute when an error is seen */
drh75897232000-05-29 14:26:00 +0000277 char *overflow; /* Code to execute on a stack overflow */
drh75897232000-05-29 14:26:00 +0000278 char *failure; /* Code to execute on parser failure */
drh75897232000-05-29 14:26:00 +0000279 char *accept; /* Code to execute when the parser excepts */
drh75897232000-05-29 14:26:00 +0000280 char *extracode; /* Code appended to the generated file */
drh75897232000-05-29 14:26:00 +0000281 char *tokendest; /* Code to execute to destroy token data */
drh960e8c62001-04-03 16:53:21 +0000282 char *vardest; /* Code for the default non-terminal destructor */
drh75897232000-05-29 14:26:00 +0000283 char *filename; /* Name of the input file */
284 char *outname; /* Name of the current output file */
285 char *tokenprefix; /* A prefix added to token names in the .h file */
286 int nconflict; /* Number of parsing conflicts */
287 int tablesize; /* Size of the parse tables */
288 int basisflag; /* Print only basis configurations */
drh34ff57b2008-07-14 12:27:51 +0000289 int has_fallback; /* True if any %fallback is seen in the grammar */
shane58543932008-12-10 20:10:04 +0000290 int nolinenosflag; /* True if #line statements should not be printed */
drh75897232000-05-29 14:26:00 +0000291 char *argv0; /* Name of the program */
292};
293
294#define MemoryCheck(X) if((X)==0){ \
295 extern void memory_error(); \
296 memory_error(); \
297}
298
299/**************** From the file "table.h" *********************************/
300/*
301** All code in this file has been automatically generated
302** from a specification in the file
303** "table.q"
304** by the associative array code building program "aagen".
305** Do not edit this file! Instead, edit the specification
306** file, then rerun aagen.
307*/
308/*
309** Code for processing tables in the LEMON parser generator.
310*/
drh75897232000-05-29 14:26:00 +0000311/* Routines for handling a strings */
312
icculus9e44cf12010-02-14 17:14:22 +0000313const char *Strsafe(const char *);
drh75897232000-05-29 14:26:00 +0000314
icculus9e44cf12010-02-14 17:14:22 +0000315void Strsafe_init(void);
316int Strsafe_insert(const char *);
317const char *Strsafe_find(const char *);
drh75897232000-05-29 14:26:00 +0000318
319/* Routines for handling symbols of the grammar */
320
icculus9e44cf12010-02-14 17:14:22 +0000321struct symbol *Symbol_new(const char *);
322int Symbolcmpp(const void *, const void *);
323void Symbol_init(void);
324int Symbol_insert(struct symbol *, const char *);
325struct symbol *Symbol_find(const char *);
326struct symbol *Symbol_Nth(int);
327int Symbol_count(void);
328struct symbol **Symbol_arrayof(void);
drh75897232000-05-29 14:26:00 +0000329
330/* Routines to manage the state table */
331
icculus9e44cf12010-02-14 17:14:22 +0000332int Configcmp(const char *, const char *);
333struct state *State_new(void);
334void State_init(void);
335int State_insert(struct state *, struct config *);
336struct state *State_find(struct config *);
drh75897232000-05-29 14:26:00 +0000337struct state **State_arrayof(/* */);
338
339/* Routines used for efficiency in Configlist_add */
340
icculus9e44cf12010-02-14 17:14:22 +0000341void Configtable_init(void);
342int Configtable_insert(struct config *);
343struct config *Configtable_find(struct config *);
344void Configtable_clear(int(*)(struct config *));
345
drh75897232000-05-29 14:26:00 +0000346/****************** From the file "action.c" *******************************/
347/*
348** Routines processing parser actions in the LEMON parser generator.
349*/
350
351/* Allocate a new parser action */
drhe9278182007-07-18 18:16:29 +0000352static struct action *Action_new(void){
drh75897232000-05-29 14:26:00 +0000353 static struct action *freelist = 0;
icculus9e44cf12010-02-14 17:14:22 +0000354 struct action *newaction;
drh75897232000-05-29 14:26:00 +0000355
356 if( freelist==0 ){
357 int i;
358 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +0000359 freelist = (struct action *)calloc(amt, sizeof(struct action));
drh75897232000-05-29 14:26:00 +0000360 if( freelist==0 ){
361 fprintf(stderr,"Unable to allocate memory for a new parser action.");
362 exit(1);
363 }
364 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
365 freelist[amt-1].next = 0;
366 }
icculus9e44cf12010-02-14 17:14:22 +0000367 newaction = freelist;
drh75897232000-05-29 14:26:00 +0000368 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +0000369 return newaction;
drh75897232000-05-29 14:26:00 +0000370}
371
drhe9278182007-07-18 18:16:29 +0000372/* Compare two actions for sorting purposes. Return negative, zero, or
373** positive if the first action is less than, equal to, or greater than
374** the first
375*/
376static int actioncmp(
377 struct action *ap1,
378 struct action *ap2
379){
drh75897232000-05-29 14:26:00 +0000380 int rc;
381 rc = ap1->sp->index - ap2->sp->index;
drh75897232000-05-29 14:26:00 +0000382 if( rc==0 ){
drh9892c5d2007-12-21 00:02:11 +0000383 rc = (int)ap1->type - (int)ap2->type;
384 }
385 if( rc==0 && ap1->type==REDUCE ){
drh75897232000-05-29 14:26:00 +0000386 rc = ap1->x.rp->index - ap2->x.rp->index;
387 }
drhe594bc32009-11-03 13:02:25 +0000388 if( rc==0 ){
icculus7b429aa2010-03-03 17:09:01 +0000389 rc = (int) (ap2 - ap1);
drhe594bc32009-11-03 13:02:25 +0000390 }
drh75897232000-05-29 14:26:00 +0000391 return rc;
392}
393
394/* Sort parser actions */
drhe9278182007-07-18 18:16:29 +0000395static struct action *Action_sort(
396 struct action *ap
397){
398 ap = (struct action *)msort((char *)ap,(char **)&ap->next,
399 (int(*)(const char*,const char*))actioncmp);
drh75897232000-05-29 14:26:00 +0000400 return ap;
401}
402
icculus9e44cf12010-02-14 17:14:22 +0000403void Action_add(
404 struct action **app,
405 enum e_action type,
406 struct symbol *sp,
407 char *arg
408){
409 struct action *newaction;
410 newaction = Action_new();
411 newaction->next = *app;
412 *app = newaction;
413 newaction->type = type;
414 newaction->sp = sp;
drh75897232000-05-29 14:26:00 +0000415 if( type==SHIFT ){
icculus9e44cf12010-02-14 17:14:22 +0000416 newaction->x.stp = (struct state *)arg;
drh75897232000-05-29 14:26:00 +0000417 }else{
icculus9e44cf12010-02-14 17:14:22 +0000418 newaction->x.rp = (struct rule *)arg;
drh75897232000-05-29 14:26:00 +0000419 }
420}
drh8b582012003-10-21 13:16:03 +0000421/********************** New code to implement the "acttab" module ***********/
422/*
423** This module implements routines use to construct the yy_action[] table.
424*/
425
426/*
427** The state of the yy_action table under construction is an instance of
drh8dc3e8f2010-01-07 03:53:03 +0000428** the following structure.
429**
430** The yy_action table maps the pair (state_number, lookahead) into an
431** action_number. The table is an array of integers pairs. The state_number
432** determines an initial offset into the yy_action array. The lookahead
433** value is then added to this initial offset to get an index X into the
434** yy_action array. If the aAction[X].lookahead equals the value of the
435** of the lookahead input, then the value of the action_number output is
436** aAction[X].action. If the lookaheads do not match then the
437** default action for the state_number is returned.
438**
439** All actions associated with a single state_number are first entered
440** into aLookahead[] using multiple calls to acttab_action(). Then the
441** actions for that single state_number are placed into the aAction[]
442** array with a single call to acttab_insert(). The acttab_insert() call
443** also resets the aLookahead[] array in preparation for the next
444** state number.
drh8b582012003-10-21 13:16:03 +0000445*/
icculus9e44cf12010-02-14 17:14:22 +0000446struct lookahead_action {
447 int lookahead; /* Value of the lookahead token */
448 int action; /* Action to take on the given lookahead */
449};
drh8b582012003-10-21 13:16:03 +0000450typedef struct acttab acttab;
451struct acttab {
452 int nAction; /* Number of used slots in aAction[] */
453 int nActionAlloc; /* Slots allocated for aAction[] */
icculus9e44cf12010-02-14 17:14:22 +0000454 struct lookahead_action
455 *aAction, /* The yy_action[] table under construction */
drh8b582012003-10-21 13:16:03 +0000456 *aLookahead; /* A single new transaction set */
457 int mnLookahead; /* Minimum aLookahead[].lookahead */
458 int mnAction; /* Action associated with mnLookahead */
459 int mxLookahead; /* Maximum aLookahead[].lookahead */
460 int nLookahead; /* Used slots in aLookahead[] */
461 int nLookaheadAlloc; /* Slots allocated in aLookahead[] */
462};
463
464/* Return the number of entries in the yy_action table */
465#define acttab_size(X) ((X)->nAction)
466
467/* The value for the N-th entry in yy_action */
468#define acttab_yyaction(X,N) ((X)->aAction[N].action)
469
470/* The value for the N-th entry in yy_lookahead */
471#define acttab_yylookahead(X,N) ((X)->aAction[N].lookahead)
472
473/* Free all memory associated with the given acttab */
474void acttab_free(acttab *p){
475 free( p->aAction );
476 free( p->aLookahead );
477 free( p );
478}
479
480/* Allocate a new acttab structure */
481acttab *acttab_alloc(void){
icculus9e44cf12010-02-14 17:14:22 +0000482 acttab *p = (acttab *) calloc( 1, sizeof(*p) );
drh8b582012003-10-21 13:16:03 +0000483 if( p==0 ){
484 fprintf(stderr,"Unable to allocate memory for a new acttab.");
485 exit(1);
486 }
487 memset(p, 0, sizeof(*p));
488 return p;
489}
490
drh8dc3e8f2010-01-07 03:53:03 +0000491/* Add a new action to the current transaction set.
492**
493** This routine is called once for each lookahead for a particular
494** state.
drh8b582012003-10-21 13:16:03 +0000495*/
496void acttab_action(acttab *p, int lookahead, int action){
497 if( p->nLookahead>=p->nLookaheadAlloc ){
498 p->nLookaheadAlloc += 25;
icculus9e44cf12010-02-14 17:14:22 +0000499 p->aLookahead = (struct lookahead_action *) realloc( p->aLookahead,
drh8b582012003-10-21 13:16:03 +0000500 sizeof(p->aLookahead[0])*p->nLookaheadAlloc );
501 if( p->aLookahead==0 ){
502 fprintf(stderr,"malloc failed\n");
503 exit(1);
504 }
505 }
506 if( p->nLookahead==0 ){
507 p->mxLookahead = lookahead;
508 p->mnLookahead = lookahead;
509 p->mnAction = action;
510 }else{
511 if( p->mxLookahead<lookahead ) p->mxLookahead = lookahead;
512 if( p->mnLookahead>lookahead ){
513 p->mnLookahead = lookahead;
514 p->mnAction = action;
515 }
516 }
517 p->aLookahead[p->nLookahead].lookahead = lookahead;
518 p->aLookahead[p->nLookahead].action = action;
519 p->nLookahead++;
520}
521
522/*
523** Add the transaction set built up with prior calls to acttab_action()
524** into the current action table. Then reset the transaction set back
525** to an empty set in preparation for a new round of acttab_action() calls.
526**
527** Return the offset into the action table of the new transaction.
528*/
529int acttab_insert(acttab *p){
530 int i, j, k, n;
531 assert( p->nLookahead>0 );
532
533 /* Make sure we have enough space to hold the expanded action table
534 ** in the worst case. The worst case occurs if the transaction set
535 ** must be appended to the current action table
536 */
drh784d86f2004-02-19 18:41:53 +0000537 n = p->mxLookahead + 1;
drh8dc3e8f2010-01-07 03:53:03 +0000538 if( p->nAction + n >= p->nActionAlloc ){
drhfdbf9282003-10-21 16:34:41 +0000539 int oldAlloc = p->nActionAlloc;
drh8b582012003-10-21 13:16:03 +0000540 p->nActionAlloc = p->nAction + n + p->nActionAlloc + 20;
icculus9e44cf12010-02-14 17:14:22 +0000541 p->aAction = (struct lookahead_action *) realloc( p->aAction,
drh8b582012003-10-21 13:16:03 +0000542 sizeof(p->aAction[0])*p->nActionAlloc);
543 if( p->aAction==0 ){
544 fprintf(stderr,"malloc failed\n");
545 exit(1);
546 }
drhfdbf9282003-10-21 16:34:41 +0000547 for(i=oldAlloc; i<p->nActionAlloc; i++){
drh8b582012003-10-21 13:16:03 +0000548 p->aAction[i].lookahead = -1;
549 p->aAction[i].action = -1;
550 }
551 }
552
drh8dc3e8f2010-01-07 03:53:03 +0000553 /* Scan the existing action table looking for an offset that is a
554 ** duplicate of the current transaction set. Fall out of the loop
555 ** if and when the duplicate is found.
drh8b582012003-10-21 13:16:03 +0000556 **
557 ** i is the index in p->aAction[] where p->mnLookahead is inserted.
558 */
drh8dc3e8f2010-01-07 03:53:03 +0000559 for(i=p->nAction-1; i>=0; i--){
drhf16371d2009-11-03 19:18:31 +0000560 if( p->aAction[i].lookahead==p->mnLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000561 /* All lookaheads and actions in the aLookahead[] transaction
562 ** must match against the candidate aAction[i] entry. */
drh8b582012003-10-21 13:16:03 +0000563 if( p->aAction[i].action!=p->mnAction ) continue;
564 for(j=0; j<p->nLookahead; j++){
565 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
566 if( k<0 || k>=p->nAction ) break;
567 if( p->aLookahead[j].lookahead!=p->aAction[k].lookahead ) break;
568 if( p->aLookahead[j].action!=p->aAction[k].action ) break;
569 }
570 if( j<p->nLookahead ) continue;
drh8dc3e8f2010-01-07 03:53:03 +0000571
572 /* No possible lookahead value that is not in the aLookahead[]
573 ** transaction is allowed to match aAction[i] */
drh8b582012003-10-21 13:16:03 +0000574 n = 0;
575 for(j=0; j<p->nAction; j++){
drhfdbf9282003-10-21 16:34:41 +0000576 if( p->aAction[j].lookahead<0 ) continue;
577 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) n++;
drh8b582012003-10-21 13:16:03 +0000578 }
drhfdbf9282003-10-21 16:34:41 +0000579 if( n==p->nLookahead ){
drh8dc3e8f2010-01-07 03:53:03 +0000580 break; /* An exact match is found at offset i */
drhfdbf9282003-10-21 16:34:41 +0000581 }
drh8b582012003-10-21 13:16:03 +0000582 }
583 }
drh8dc3e8f2010-01-07 03:53:03 +0000584
585 /* If no existing offsets exactly match the current transaction, find an
586 ** an empty offset in the aAction[] table in which we can add the
587 ** aLookahead[] transaction.
588 */
drhf16371d2009-11-03 19:18:31 +0000589 if( i<0 ){
drh8dc3e8f2010-01-07 03:53:03 +0000590 /* Look for holes in the aAction[] table that fit the current
591 ** aLookahead[] transaction. Leave i set to the offset of the hole.
592 ** If no holes are found, i is left at p->nAction, which means the
593 ** transaction will be appended. */
594 for(i=0; i<p->nActionAlloc - p->mxLookahead; i++){
drhf16371d2009-11-03 19:18:31 +0000595 if( p->aAction[i].lookahead<0 ){
596 for(j=0; j<p->nLookahead; j++){
597 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
598 if( k<0 ) break;
599 if( p->aAction[k].lookahead>=0 ) break;
600 }
601 if( j<p->nLookahead ) continue;
602 for(j=0; j<p->nAction; j++){
603 if( p->aAction[j].lookahead==j+p->mnLookahead-i ) break;
604 }
605 if( j==p->nAction ){
606 break; /* Fits in empty slots */
607 }
608 }
609 }
610 }
drh8b582012003-10-21 13:16:03 +0000611 /* Insert transaction set at index i. */
612 for(j=0; j<p->nLookahead; j++){
613 k = p->aLookahead[j].lookahead - p->mnLookahead + i;
614 p->aAction[k] = p->aLookahead[j];
615 if( k>=p->nAction ) p->nAction = k+1;
616 }
617 p->nLookahead = 0;
618
619 /* Return the offset that is added to the lookahead in order to get the
620 ** index into yy_action of the action */
621 return i - p->mnLookahead;
622}
623
drh75897232000-05-29 14:26:00 +0000624/********************** From the file "build.c" *****************************/
625/*
626** Routines to construction the finite state machine for the LEMON
627** parser generator.
628*/
629
630/* Find a precedence symbol of every rule in the grammar.
631**
632** Those rules which have a precedence symbol coded in the input
633** grammar using the "[symbol]" construct will already have the
634** rp->precsym field filled. Other rules take as their precedence
635** symbol the first RHS symbol with a defined precedence. If there
636** are not RHS symbols with a defined precedence, the precedence
637** symbol field is left blank.
638*/
icculus9e44cf12010-02-14 17:14:22 +0000639void FindRulePrecedences(struct lemon *xp)
drh75897232000-05-29 14:26:00 +0000640{
641 struct rule *rp;
642 for(rp=xp->rule; rp; rp=rp->next){
643 if( rp->precsym==0 ){
drhfd405312005-11-06 04:06:59 +0000644 int i, j;
645 for(i=0; i<rp->nrhs && rp->precsym==0; i++){
646 struct symbol *sp = rp->rhs[i];
647 if( sp->type==MULTITERMINAL ){
648 for(j=0; j<sp->nsubsym; j++){
649 if( sp->subsym[j]->prec>=0 ){
650 rp->precsym = sp->subsym[j];
651 break;
652 }
653 }
654 }else if( sp->prec>=0 ){
drh75897232000-05-29 14:26:00 +0000655 rp->precsym = rp->rhs[i];
drhf2f105d2012-08-20 15:53:54 +0000656 }
drh75897232000-05-29 14:26:00 +0000657 }
658 }
659 }
660 return;
661}
662
663/* Find all nonterminals which will generate the empty string.
664** Then go back and compute the first sets of every nonterminal.
665** The first set is the set of all terminal symbols which can begin
666** a string generated by that nonterminal.
667*/
icculus9e44cf12010-02-14 17:14:22 +0000668void FindFirstSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000669{
drhfd405312005-11-06 04:06:59 +0000670 int i, j;
drh75897232000-05-29 14:26:00 +0000671 struct rule *rp;
672 int progress;
673
674 for(i=0; i<lemp->nsymbol; i++){
drhaa9f1122007-08-23 02:50:56 +0000675 lemp->symbols[i]->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +0000676 }
677 for(i=lemp->nterminal; i<lemp->nsymbol; i++){
678 lemp->symbols[i]->firstset = SetNew();
679 }
680
681 /* First compute all lambdas */
682 do{
683 progress = 0;
684 for(rp=lemp->rule; rp; rp=rp->next){
685 if( rp->lhs->lambda ) continue;
686 for(i=0; i<rp->nrhs; i++){
drh7dd1ac62012-01-07 15:17:18 +0000687 struct symbol *sp = rp->rhs[i];
688 assert( sp->type==NONTERMINAL || sp->lambda==LEMON_FALSE );
689 if( sp->lambda==LEMON_FALSE ) break;
drh75897232000-05-29 14:26:00 +0000690 }
691 if( i==rp->nrhs ){
drhaa9f1122007-08-23 02:50:56 +0000692 rp->lhs->lambda = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +0000693 progress = 1;
694 }
695 }
696 }while( progress );
697
698 /* Now compute all first sets */
699 do{
700 struct symbol *s1, *s2;
701 progress = 0;
702 for(rp=lemp->rule; rp; rp=rp->next){
703 s1 = rp->lhs;
704 for(i=0; i<rp->nrhs; i++){
705 s2 = rp->rhs[i];
706 if( s2->type==TERMINAL ){
707 progress += SetAdd(s1->firstset,s2->index);
708 break;
drhfd405312005-11-06 04:06:59 +0000709 }else if( s2->type==MULTITERMINAL ){
710 for(j=0; j<s2->nsubsym; j++){
711 progress += SetAdd(s1->firstset,s2->subsym[j]->index);
712 }
713 break;
drhf2f105d2012-08-20 15:53:54 +0000714 }else if( s1==s2 ){
drhaa9f1122007-08-23 02:50:56 +0000715 if( s1->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000716 }else{
drh75897232000-05-29 14:26:00 +0000717 progress += SetUnion(s1->firstset,s2->firstset);
drhaa9f1122007-08-23 02:50:56 +0000718 if( s2->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +0000719 }
drh75897232000-05-29 14:26:00 +0000720 }
721 }
722 }while( progress );
723 return;
724}
725
726/* Compute all LR(0) states for the grammar. Links
727** are added to between some states so that the LR(1) follow sets
728** can be computed later.
729*/
icculus9e44cf12010-02-14 17:14:22 +0000730PRIVATE struct state *getstate(struct lemon *); /* forward reference */
731void FindStates(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000732{
733 struct symbol *sp;
734 struct rule *rp;
735
736 Configlist_init();
737
738 /* Find the start symbol */
739 if( lemp->start ){
740 sp = Symbol_find(lemp->start);
741 if( sp==0 ){
742 ErrorMsg(lemp->filename,0,
743"The specified start symbol \"%s\" is not \
744in a nonterminal of the grammar. \"%s\" will be used as the start \
745symbol instead.",lemp->start,lemp->rule->lhs->name);
746 lemp->errorcnt++;
747 sp = lemp->rule->lhs;
748 }
749 }else{
750 sp = lemp->rule->lhs;
751 }
752
753 /* Make sure the start symbol doesn't occur on the right-hand side of
754 ** any rule. Report an error if it does. (YACC would generate a new
755 ** start symbol in this case.) */
756 for(rp=lemp->rule; rp; rp=rp->next){
757 int i;
758 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +0000759 if( rp->rhs[i]==sp ){ /* FIX ME: Deal with multiterminals */
drh75897232000-05-29 14:26:00 +0000760 ErrorMsg(lemp->filename,0,
761"The start symbol \"%s\" occurs on the \
762right-hand side of a rule. This will result in a parser which \
763does not work properly.",sp->name);
764 lemp->errorcnt++;
765 }
766 }
767 }
768
769 /* The basis configuration set for the first state
770 ** is all rules which have the start symbol as their
771 ** left-hand side */
772 for(rp=sp->rule; rp; rp=rp->nextlhs){
773 struct config *newcfp;
drhb4960992007-10-05 16:16:36 +0000774 rp->lhsStart = 1;
drh75897232000-05-29 14:26:00 +0000775 newcfp = Configlist_addbasis(rp,0);
776 SetAdd(newcfp->fws,0);
777 }
778
779 /* Compute the first state. All other states will be
780 ** computed automatically during the computation of the first one.
781 ** The returned pointer to the first state is not used. */
782 (void)getstate(lemp);
783 return;
784}
785
786/* Return a pointer to a state which is described by the configuration
787** list which has been built from calls to Configlist_add.
788*/
icculus9e44cf12010-02-14 17:14:22 +0000789PRIVATE void buildshifts(struct lemon *, struct state *); /* Forwd ref */
790PRIVATE struct state *getstate(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000791{
792 struct config *cfp, *bp;
793 struct state *stp;
794
795 /* Extract the sorted basis of the new state. The basis was constructed
796 ** by prior calls to "Configlist_addbasis()". */
797 Configlist_sortbasis();
798 bp = Configlist_basis();
799
800 /* Get a state with the same basis */
801 stp = State_find(bp);
802 if( stp ){
803 /* A state with the same basis already exists! Copy all the follow-set
804 ** propagation links from the state under construction into the
805 ** preexisting state, then return a pointer to the preexisting state */
806 struct config *x, *y;
807 for(x=bp, y=stp->bp; x && y; x=x->bp, y=y->bp){
808 Plink_copy(&y->bplp,x->bplp);
809 Plink_delete(x->fplp);
810 x->fplp = x->bplp = 0;
811 }
812 cfp = Configlist_return();
813 Configlist_eat(cfp);
814 }else{
815 /* This really is a new state. Construct all the details */
816 Configlist_closure(lemp); /* Compute the configuration closure */
817 Configlist_sort(); /* Sort the configuration closure */
818 cfp = Configlist_return(); /* Get a pointer to the config list */
819 stp = State_new(); /* A new state structure */
820 MemoryCheck(stp);
821 stp->bp = bp; /* Remember the configuration basis */
822 stp->cfp = cfp; /* Remember the configuration closure */
drhada354d2005-11-05 15:03:59 +0000823 stp->statenum = lemp->nstate++; /* Every state gets a sequence number */
drh75897232000-05-29 14:26:00 +0000824 stp->ap = 0; /* No actions, yet. */
825 State_insert(stp,stp->bp); /* Add to the state table */
826 buildshifts(lemp,stp); /* Recursively compute successor states */
827 }
828 return stp;
829}
830
drhfd405312005-11-06 04:06:59 +0000831/*
832** Return true if two symbols are the same.
833*/
icculus9e44cf12010-02-14 17:14:22 +0000834int same_symbol(struct symbol *a, struct symbol *b)
drhfd405312005-11-06 04:06:59 +0000835{
836 int i;
837 if( a==b ) return 1;
838 if( a->type!=MULTITERMINAL ) return 0;
839 if( b->type!=MULTITERMINAL ) return 0;
840 if( a->nsubsym!=b->nsubsym ) return 0;
841 for(i=0; i<a->nsubsym; i++){
842 if( a->subsym[i]!=b->subsym[i] ) return 0;
843 }
844 return 1;
845}
846
drh75897232000-05-29 14:26:00 +0000847/* Construct all successor states to the given state. A "successor"
848** state is any state which can be reached by a shift action.
849*/
icculus9e44cf12010-02-14 17:14:22 +0000850PRIVATE void buildshifts(struct lemon *lemp, struct state *stp)
drh75897232000-05-29 14:26:00 +0000851{
852 struct config *cfp; /* For looping thru the config closure of "stp" */
853 struct config *bcfp; /* For the inner loop on config closure of "stp" */
icculus9e44cf12010-02-14 17:14:22 +0000854 struct config *newcfg; /* */
drh75897232000-05-29 14:26:00 +0000855 struct symbol *sp; /* Symbol following the dot in configuration "cfp" */
856 struct symbol *bsp; /* Symbol following the dot in configuration "bcfp" */
857 struct state *newstp; /* A pointer to a successor state */
858
859 /* Each configuration becomes complete after it contibutes to a successor
860 ** state. Initially, all configurations are incomplete */
861 for(cfp=stp->cfp; cfp; cfp=cfp->next) cfp->status = INCOMPLETE;
862
863 /* Loop through all configurations of the state "stp" */
864 for(cfp=stp->cfp; cfp; cfp=cfp->next){
865 if( cfp->status==COMPLETE ) continue; /* Already used by inner loop */
866 if( cfp->dot>=cfp->rp->nrhs ) continue; /* Can't shift this config */
867 Configlist_reset(); /* Reset the new config set */
868 sp = cfp->rp->rhs[cfp->dot]; /* Symbol after the dot */
869
870 /* For every configuration in the state "stp" which has the symbol "sp"
871 ** following its dot, add the same configuration to the basis set under
872 ** construction but with the dot shifted one symbol to the right. */
873 for(bcfp=cfp; bcfp; bcfp=bcfp->next){
874 if( bcfp->status==COMPLETE ) continue; /* Already used */
875 if( bcfp->dot>=bcfp->rp->nrhs ) continue; /* Can't shift this one */
876 bsp = bcfp->rp->rhs[bcfp->dot]; /* Get symbol after dot */
drhfd405312005-11-06 04:06:59 +0000877 if( !same_symbol(bsp,sp) ) continue; /* Must be same as for "cfp" */
drh75897232000-05-29 14:26:00 +0000878 bcfp->status = COMPLETE; /* Mark this config as used */
icculus9e44cf12010-02-14 17:14:22 +0000879 newcfg = Configlist_addbasis(bcfp->rp,bcfp->dot+1);
880 Plink_add(&newcfg->bplp,bcfp);
drh75897232000-05-29 14:26:00 +0000881 }
882
883 /* Get a pointer to the state described by the basis configuration set
884 ** constructed in the preceding loop */
885 newstp = getstate(lemp);
886
887 /* The state "newstp" is reached from the state "stp" by a shift action
888 ** on the symbol "sp" */
drhfd405312005-11-06 04:06:59 +0000889 if( sp->type==MULTITERMINAL ){
890 int i;
891 for(i=0; i<sp->nsubsym; i++){
892 Action_add(&stp->ap,SHIFT,sp->subsym[i],(char*)newstp);
893 }
894 }else{
895 Action_add(&stp->ap,SHIFT,sp,(char *)newstp);
896 }
drh75897232000-05-29 14:26:00 +0000897 }
898}
899
900/*
901** Construct the propagation links
902*/
icculus9e44cf12010-02-14 17:14:22 +0000903void FindLinks(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000904{
905 int i;
906 struct config *cfp, *other;
907 struct state *stp;
908 struct plink *plp;
909
910 /* Housekeeping detail:
911 ** Add to every propagate link a pointer back to the state to
912 ** which the link is attached. */
913 for(i=0; i<lemp->nstate; i++){
914 stp = lemp->sorted[i];
915 for(cfp=stp->cfp; cfp; cfp=cfp->next){
916 cfp->stp = stp;
917 }
918 }
919
920 /* Convert all backlinks into forward links. Only the forward
921 ** links are used in the follow-set computation. */
922 for(i=0; i<lemp->nstate; i++){
923 stp = lemp->sorted[i];
924 for(cfp=stp->cfp; cfp; cfp=cfp->next){
925 for(plp=cfp->bplp; plp; plp=plp->next){
926 other = plp->cfp;
927 Plink_add(&other->fplp,cfp);
928 }
929 }
930 }
931}
932
933/* Compute all followsets.
934**
935** A followset is the set of all symbols which can come immediately
936** after a configuration.
937*/
icculus9e44cf12010-02-14 17:14:22 +0000938void FindFollowSets(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000939{
940 int i;
941 struct config *cfp;
942 struct plink *plp;
943 int progress;
944 int change;
945
946 for(i=0; i<lemp->nstate; i++){
947 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
948 cfp->status = INCOMPLETE;
949 }
950 }
951
952 do{
953 progress = 0;
954 for(i=0; i<lemp->nstate; i++){
955 for(cfp=lemp->sorted[i]->cfp; cfp; cfp=cfp->next){
956 if( cfp->status==COMPLETE ) continue;
957 for(plp=cfp->fplp; plp; plp=plp->next){
958 change = SetUnion(plp->cfp->fws,cfp->fws);
959 if( change ){
960 plp->cfp->status = INCOMPLETE;
961 progress = 1;
drhf2f105d2012-08-20 15:53:54 +0000962 }
963 }
drh75897232000-05-29 14:26:00 +0000964 cfp->status = COMPLETE;
965 }
966 }
967 }while( progress );
968}
969
drh3cb2f6e2012-01-09 14:19:05 +0000970static int resolve_conflict(struct action *,struct action *);
drh75897232000-05-29 14:26:00 +0000971
972/* Compute the reduce actions, and resolve conflicts.
973*/
icculus9e44cf12010-02-14 17:14:22 +0000974void FindActions(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +0000975{
976 int i,j;
977 struct config *cfp;
978 struct state *stp;
979 struct symbol *sp;
980 struct rule *rp;
981
982 /* Add all of the reduce actions
983 ** A reduce action is added for each element of the followset of
984 ** a configuration which has its dot at the extreme right.
985 */
986 for(i=0; i<lemp->nstate; i++){ /* Loop over all states */
987 stp = lemp->sorted[i];
988 for(cfp=stp->cfp; cfp; cfp=cfp->next){ /* Loop over all configurations */
989 if( cfp->rp->nrhs==cfp->dot ){ /* Is dot at extreme right? */
990 for(j=0; j<lemp->nterminal; j++){
991 if( SetFind(cfp->fws,j) ){
992 /* Add a reduce action to the state "stp" which will reduce by the
993 ** rule "cfp->rp" if the lookahead symbol is "lemp->symbols[j]" */
drh218dc692004-05-31 23:13:45 +0000994 Action_add(&stp->ap,REDUCE,lemp->symbols[j],(char *)cfp->rp);
drh75897232000-05-29 14:26:00 +0000995 }
drhf2f105d2012-08-20 15:53:54 +0000996 }
drh75897232000-05-29 14:26:00 +0000997 }
998 }
999 }
1000
1001 /* Add the accepting token */
1002 if( lemp->start ){
1003 sp = Symbol_find(lemp->start);
1004 if( sp==0 ) sp = lemp->rule->lhs;
1005 }else{
1006 sp = lemp->rule->lhs;
1007 }
1008 /* Add to the first state (which is always the starting state of the
1009 ** finite state machine) an action to ACCEPT if the lookahead is the
1010 ** start nonterminal. */
1011 Action_add(&lemp->sorted[0]->ap,ACCEPT,sp,0);
1012
1013 /* Resolve conflicts */
1014 for(i=0; i<lemp->nstate; i++){
1015 struct action *ap, *nap;
1016 struct state *stp;
1017 stp = lemp->sorted[i];
drhe9278182007-07-18 18:16:29 +00001018 /* assert( stp->ap ); */
drh75897232000-05-29 14:26:00 +00001019 stp->ap = Action_sort(stp->ap);
drhb59499c2002-02-23 18:45:13 +00001020 for(ap=stp->ap; ap && ap->next; ap=ap->next){
drh75897232000-05-29 14:26:00 +00001021 for(nap=ap->next; nap && nap->sp==ap->sp; nap=nap->next){
1022 /* The two actions "ap" and "nap" have the same lookahead.
1023 ** Figure out which one should be used */
drh3cb2f6e2012-01-09 14:19:05 +00001024 lemp->nconflict += resolve_conflict(ap,nap);
drh75897232000-05-29 14:26:00 +00001025 }
1026 }
1027 }
1028
1029 /* Report an error for each rule that can never be reduced. */
drhaa9f1122007-08-23 02:50:56 +00001030 for(rp=lemp->rule; rp; rp=rp->next) rp->canReduce = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00001031 for(i=0; i<lemp->nstate; i++){
1032 struct action *ap;
1033 for(ap=lemp->sorted[i]->ap; ap; ap=ap->next){
drhaa9f1122007-08-23 02:50:56 +00001034 if( ap->type==REDUCE ) ap->x.rp->canReduce = LEMON_TRUE;
drh75897232000-05-29 14:26:00 +00001035 }
1036 }
1037 for(rp=lemp->rule; rp; rp=rp->next){
1038 if( rp->canReduce ) continue;
1039 ErrorMsg(lemp->filename,rp->ruleline,"This rule can not be reduced.\n");
1040 lemp->errorcnt++;
1041 }
1042}
1043
1044/* Resolve a conflict between the two given actions. If the
drh34ff57b2008-07-14 12:27:51 +00001045** conflict can't be resolved, return non-zero.
drh75897232000-05-29 14:26:00 +00001046**
1047** NO LONGER TRUE:
1048** To resolve a conflict, first look to see if either action
1049** is on an error rule. In that case, take the action which
1050** is not associated with the error rule. If neither or both
1051** actions are associated with an error rule, then try to
1052** use precedence to resolve the conflict.
1053**
1054** If either action is a SHIFT, then it must be apx. This
1055** function won't work if apx->type==REDUCE and apy->type==SHIFT.
1056*/
icculus9e44cf12010-02-14 17:14:22 +00001057static int resolve_conflict(
1058 struct action *apx,
drh3cb2f6e2012-01-09 14:19:05 +00001059 struct action *apy
icculus9e44cf12010-02-14 17:14:22 +00001060){
drh75897232000-05-29 14:26:00 +00001061 struct symbol *spx, *spy;
1062 int errcnt = 0;
1063 assert( apx->sp==apy->sp ); /* Otherwise there would be no conflict */
drhf0fa1c12006-12-14 01:06:22 +00001064 if( apx->type==SHIFT && apy->type==SHIFT ){
drh9892c5d2007-12-21 00:02:11 +00001065 apy->type = SSCONFLICT;
drhf0fa1c12006-12-14 01:06:22 +00001066 errcnt++;
1067 }
drh75897232000-05-29 14:26:00 +00001068 if( apx->type==SHIFT && apy->type==REDUCE ){
1069 spx = apx->sp;
1070 spy = apy->x.rp->precsym;
1071 if( spy==0 || spx->prec<0 || spy->prec<0 ){
1072 /* Not enough precedence information. */
drh9892c5d2007-12-21 00:02:11 +00001073 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001074 errcnt++;
drhdd7e9db2010-07-19 01:52:07 +00001075 }else if( spx->prec>spy->prec ){ /* higher precedence wins */
drh75897232000-05-29 14:26:00 +00001076 apy->type = RD_RESOLVED;
1077 }else if( spx->prec<spy->prec ){
1078 apx->type = SH_RESOLVED;
1079 }else if( spx->prec==spy->prec && spx->assoc==RIGHT ){ /* Use operator */
1080 apy->type = RD_RESOLVED; /* associativity */
1081 }else if( spx->prec==spy->prec && spx->assoc==LEFT ){ /* to break tie */
1082 apx->type = SH_RESOLVED;
1083 }else{
1084 assert( spx->prec==spy->prec && spx->assoc==NONE );
drh9892c5d2007-12-21 00:02:11 +00001085 apy->type = SRCONFLICT;
drh75897232000-05-29 14:26:00 +00001086 errcnt++;
1087 }
1088 }else if( apx->type==REDUCE && apy->type==REDUCE ){
1089 spx = apx->x.rp->precsym;
1090 spy = apy->x.rp->precsym;
1091 if( spx==0 || spy==0 || spx->prec<0 ||
1092 spy->prec<0 || spx->prec==spy->prec ){
drh9892c5d2007-12-21 00:02:11 +00001093 apy->type = RRCONFLICT;
drh75897232000-05-29 14:26:00 +00001094 errcnt++;
1095 }else if( spx->prec>spy->prec ){
1096 apy->type = RD_RESOLVED;
1097 }else if( spx->prec<spy->prec ){
1098 apx->type = RD_RESOLVED;
1099 }
1100 }else{
drhb59499c2002-02-23 18:45:13 +00001101 assert(
1102 apx->type==SH_RESOLVED ||
1103 apx->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001104 apx->type==SSCONFLICT ||
1105 apx->type==SRCONFLICT ||
1106 apx->type==RRCONFLICT ||
drhb59499c2002-02-23 18:45:13 +00001107 apy->type==SH_RESOLVED ||
1108 apy->type==RD_RESOLVED ||
drh9892c5d2007-12-21 00:02:11 +00001109 apy->type==SSCONFLICT ||
1110 apy->type==SRCONFLICT ||
1111 apy->type==RRCONFLICT
drhb59499c2002-02-23 18:45:13 +00001112 );
1113 /* The REDUCE/SHIFT case cannot happen because SHIFTs come before
1114 ** REDUCEs on the list. If we reach this point it must be because
1115 ** the parser conflict had already been resolved. */
drh75897232000-05-29 14:26:00 +00001116 }
1117 return errcnt;
1118}
1119/********************* From the file "configlist.c" *************************/
1120/*
1121** Routines to processing a configuration list and building a state
1122** in the LEMON parser generator.
1123*/
1124
1125static struct config *freelist = 0; /* List of free configurations */
1126static struct config *current = 0; /* Top of list of configurations */
1127static struct config **currentend = 0; /* Last on list of configs */
1128static struct config *basis = 0; /* Top of list of basis configs */
1129static struct config **basisend = 0; /* End of list of basis configs */
1130
1131/* Return a pointer to a new configuration */
1132PRIVATE struct config *newconfig(){
icculus9e44cf12010-02-14 17:14:22 +00001133 struct config *newcfg;
drh75897232000-05-29 14:26:00 +00001134 if( freelist==0 ){
1135 int i;
1136 int amt = 3;
drh9892c5d2007-12-21 00:02:11 +00001137 freelist = (struct config *)calloc( amt, sizeof(struct config) );
drh75897232000-05-29 14:26:00 +00001138 if( freelist==0 ){
1139 fprintf(stderr,"Unable to allocate memory for a new configuration.");
1140 exit(1);
1141 }
1142 for(i=0; i<amt-1; i++) freelist[i].next = &freelist[i+1];
1143 freelist[amt-1].next = 0;
1144 }
icculus9e44cf12010-02-14 17:14:22 +00001145 newcfg = freelist;
drh75897232000-05-29 14:26:00 +00001146 freelist = freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00001147 return newcfg;
drh75897232000-05-29 14:26:00 +00001148}
1149
1150/* The configuration "old" is no longer used */
icculus9e44cf12010-02-14 17:14:22 +00001151PRIVATE void deleteconfig(struct config *old)
drh75897232000-05-29 14:26:00 +00001152{
1153 old->next = freelist;
1154 freelist = old;
1155}
1156
1157/* Initialized the configuration list builder */
1158void Configlist_init(){
1159 current = 0;
1160 currentend = &current;
1161 basis = 0;
1162 basisend = &basis;
1163 Configtable_init();
1164 return;
1165}
1166
1167/* Initialized the configuration list builder */
1168void Configlist_reset(){
1169 current = 0;
1170 currentend = &current;
1171 basis = 0;
1172 basisend = &basis;
1173 Configtable_clear(0);
1174 return;
1175}
1176
1177/* Add another configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001178struct config *Configlist_add(
1179 struct rule *rp, /* The rule */
1180 int dot /* Index into the RHS of the rule where the dot goes */
1181){
drh75897232000-05-29 14:26:00 +00001182 struct config *cfp, model;
1183
1184 assert( currentend!=0 );
1185 model.rp = rp;
1186 model.dot = dot;
1187 cfp = Configtable_find(&model);
1188 if( cfp==0 ){
1189 cfp = newconfig();
1190 cfp->rp = rp;
1191 cfp->dot = dot;
1192 cfp->fws = SetNew();
1193 cfp->stp = 0;
1194 cfp->fplp = cfp->bplp = 0;
1195 cfp->next = 0;
1196 cfp->bp = 0;
1197 *currentend = cfp;
1198 currentend = &cfp->next;
1199 Configtable_insert(cfp);
1200 }
1201 return cfp;
1202}
1203
1204/* Add a basis configuration to the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001205struct config *Configlist_addbasis(struct rule *rp, int dot)
drh75897232000-05-29 14:26:00 +00001206{
1207 struct config *cfp, model;
1208
1209 assert( basisend!=0 );
1210 assert( currentend!=0 );
1211 model.rp = rp;
1212 model.dot = dot;
1213 cfp = Configtable_find(&model);
1214 if( cfp==0 ){
1215 cfp = newconfig();
1216 cfp->rp = rp;
1217 cfp->dot = dot;
1218 cfp->fws = SetNew();
1219 cfp->stp = 0;
1220 cfp->fplp = cfp->bplp = 0;
1221 cfp->next = 0;
1222 cfp->bp = 0;
1223 *currentend = cfp;
1224 currentend = &cfp->next;
1225 *basisend = cfp;
1226 basisend = &cfp->bp;
1227 Configtable_insert(cfp);
1228 }
1229 return cfp;
1230}
1231
1232/* Compute the closure of the configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001233void Configlist_closure(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00001234{
1235 struct config *cfp, *newcfp;
1236 struct rule *rp, *newrp;
1237 struct symbol *sp, *xsp;
1238 int i, dot;
1239
1240 assert( currentend!=0 );
1241 for(cfp=current; cfp; cfp=cfp->next){
1242 rp = cfp->rp;
1243 dot = cfp->dot;
1244 if( dot>=rp->nrhs ) continue;
1245 sp = rp->rhs[dot];
1246 if( sp->type==NONTERMINAL ){
1247 if( sp->rule==0 && sp!=lemp->errsym ){
1248 ErrorMsg(lemp->filename,rp->line,"Nonterminal \"%s\" has no rules.",
1249 sp->name);
1250 lemp->errorcnt++;
1251 }
1252 for(newrp=sp->rule; newrp; newrp=newrp->nextlhs){
1253 newcfp = Configlist_add(newrp,0);
1254 for(i=dot+1; i<rp->nrhs; i++){
1255 xsp = rp->rhs[i];
1256 if( xsp->type==TERMINAL ){
1257 SetAdd(newcfp->fws,xsp->index);
1258 break;
drhfd405312005-11-06 04:06:59 +00001259 }else if( xsp->type==MULTITERMINAL ){
1260 int k;
1261 for(k=0; k<xsp->nsubsym; k++){
1262 SetAdd(newcfp->fws, xsp->subsym[k]->index);
1263 }
1264 break;
drhf2f105d2012-08-20 15:53:54 +00001265 }else{
drh75897232000-05-29 14:26:00 +00001266 SetUnion(newcfp->fws,xsp->firstset);
drhaa9f1122007-08-23 02:50:56 +00001267 if( xsp->lambda==LEMON_FALSE ) break;
drhf2f105d2012-08-20 15:53:54 +00001268 }
1269 }
drh75897232000-05-29 14:26:00 +00001270 if( i==rp->nrhs ) Plink_add(&cfp->fplp,newcfp);
1271 }
1272 }
1273 }
1274 return;
1275}
1276
1277/* Sort the configuration list */
1278void Configlist_sort(){
drh218dc692004-05-31 23:13:45 +00001279 current = (struct config *)msort((char *)current,(char **)&(current->next),Configcmp);
drh75897232000-05-29 14:26:00 +00001280 currentend = 0;
1281 return;
1282}
1283
1284/* Sort the basis configuration list */
1285void Configlist_sortbasis(){
drh218dc692004-05-31 23:13:45 +00001286 basis = (struct config *)msort((char *)current,(char **)&(current->bp),Configcmp);
drh75897232000-05-29 14:26:00 +00001287 basisend = 0;
1288 return;
1289}
1290
1291/* Return a pointer to the head of the configuration list and
1292** reset the list */
1293struct config *Configlist_return(){
1294 struct config *old;
1295 old = current;
1296 current = 0;
1297 currentend = 0;
1298 return old;
1299}
1300
1301/* Return a pointer to the head of the configuration list and
1302** reset the list */
1303struct config *Configlist_basis(){
1304 struct config *old;
1305 old = basis;
1306 basis = 0;
1307 basisend = 0;
1308 return old;
1309}
1310
1311/* Free all elements of the given configuration list */
icculus9e44cf12010-02-14 17:14:22 +00001312void Configlist_eat(struct config *cfp)
drh75897232000-05-29 14:26:00 +00001313{
1314 struct config *nextcfp;
1315 for(; cfp; cfp=nextcfp){
1316 nextcfp = cfp->next;
1317 assert( cfp->fplp==0 );
1318 assert( cfp->bplp==0 );
1319 if( cfp->fws ) SetFree(cfp->fws);
1320 deleteconfig(cfp);
1321 }
1322 return;
1323}
1324/***************** From the file "error.c" *********************************/
1325/*
1326** Code for printing error message.
1327*/
1328
drhf9a2e7b2003-04-15 01:49:48 +00001329void ErrorMsg(const char *filename, int lineno, const char *format, ...){
icculus15a2cec2010-02-16 16:07:28 +00001330 va_list ap;
icculus1c11f742010-02-15 00:01:04 +00001331 fprintf(stderr, "%s:%d: ", filename, lineno);
1332 va_start(ap, format);
1333 vfprintf(stderr,format,ap);
1334 va_end(ap);
1335 fprintf(stderr, "\n");
drh75897232000-05-29 14:26:00 +00001336}
1337/**************** From the file "main.c" ************************************/
1338/*
1339** Main program file for the LEMON parser generator.
1340*/
1341
1342/* Report an out-of-memory condition and abort. This function
1343** is used mostly by the "MemoryCheck" macro in struct.h
1344*/
1345void memory_error(){
1346 fprintf(stderr,"Out of memory. Aborting...\n");
1347 exit(1);
1348}
1349
drh6d08b4d2004-07-20 12:45:22 +00001350static int nDefine = 0; /* Number of -D options on the command line */
1351static char **azDefine = 0; /* Name of the -D macros */
1352
1353/* This routine is called with the argument to each -D command-line option.
1354** Add the macro defined to the azDefine array.
1355*/
1356static void handle_D_option(char *z){
1357 char **paz;
1358 nDefine++;
icculus9e44cf12010-02-14 17:14:22 +00001359 azDefine = (char **) realloc(azDefine, sizeof(azDefine[0])*nDefine);
drh6d08b4d2004-07-20 12:45:22 +00001360 if( azDefine==0 ){
1361 fprintf(stderr,"out of memory\n");
1362 exit(1);
1363 }
1364 paz = &azDefine[nDefine-1];
icculus9e44cf12010-02-14 17:14:22 +00001365 *paz = (char *) malloc( lemonStrlen(z)+1 );
drh6d08b4d2004-07-20 12:45:22 +00001366 if( *paz==0 ){
1367 fprintf(stderr,"out of memory\n");
1368 exit(1);
1369 }
1370 strcpy(*paz, z);
1371 for(z=*paz; *z && *z!='='; z++){}
1372 *z = 0;
1373}
1374
icculus3e143bd2010-02-14 00:48:49 +00001375static char *user_templatename = NULL;
1376static void handle_T_option(char *z){
icculus9e44cf12010-02-14 17:14:22 +00001377 user_templatename = (char *) malloc( lemonStrlen(z)+1 );
icculus3e143bd2010-02-14 00:48:49 +00001378 if( user_templatename==0 ){
1379 memory_error();
1380 }
1381 strcpy(user_templatename, z);
1382}
drh75897232000-05-29 14:26:00 +00001383
1384/* The main program. Parse the command line and do it... */
icculus9e44cf12010-02-14 17:14:22 +00001385int main(int argc, char **argv)
drh75897232000-05-29 14:26:00 +00001386{
1387 static int version = 0;
1388 static int rpflag = 0;
1389 static int basisflag = 0;
1390 static int compress = 0;
1391 static int quiet = 0;
1392 static int statistics = 0;
1393 static int mhflag = 0;
shane58543932008-12-10 20:10:04 +00001394 static int nolinenosflag = 0;
drhdd7e9db2010-07-19 01:52:07 +00001395 static int noResort = 0;
drh75897232000-05-29 14:26:00 +00001396 static struct s_options options[] = {
1397 {OPT_FLAG, "b", (char*)&basisflag, "Print only the basis in report."},
1398 {OPT_FLAG, "c", (char*)&compress, "Don't compress the action table."},
drh6d08b4d2004-07-20 12:45:22 +00001399 {OPT_FSTR, "D", (char*)handle_D_option, "Define an %ifdef macro."},
icculus3e143bd2010-02-14 00:48:49 +00001400 {OPT_FSTR, "T", (char*)handle_T_option, "Specify a template file."},
drh75897232000-05-29 14:26:00 +00001401 {OPT_FLAG, "g", (char*)&rpflag, "Print grammar without actions."},
shane58543932008-12-10 20:10:04 +00001402 {OPT_FLAG, "m", (char*)&mhflag, "Output a makeheaders compatible file."},
1403 {OPT_FLAG, "l", (char*)&nolinenosflag, "Do not print #line statements."},
drhf5c4e0f2010-07-18 11:35:53 +00001404 {OPT_FLAG, "p", (char*)&showPrecedenceConflict,
1405 "Show conflicts resolved by precedence rules"},
drh75897232000-05-29 14:26:00 +00001406 {OPT_FLAG, "q", (char*)&quiet, "(Quiet) Don't print the report file."},
drhdd7e9db2010-07-19 01:52:07 +00001407 {OPT_FLAG, "r", (char*)&noResort, "Do not sort or renumber states"},
drh6d08b4d2004-07-20 12:45:22 +00001408 {OPT_FLAG, "s", (char*)&statistics,
1409 "Print parser stats to standard output."},
drh75897232000-05-29 14:26:00 +00001410 {OPT_FLAG, "x", (char*)&version, "Print the version number."},
1411 {OPT_FLAG,0,0,0}
1412 };
1413 int i;
icculus42585cf2010-02-14 05:19:56 +00001414 int exitcode;
drh75897232000-05-29 14:26:00 +00001415 struct lemon lem;
1416
drhb0c86772000-06-02 23:21:26 +00001417 OptInit(argv,options,stderr);
drh75897232000-05-29 14:26:00 +00001418 if( version ){
drhb19a2bc2001-09-16 00:13:26 +00001419 printf("Lemon version 1.0\n");
drh75897232000-05-29 14:26:00 +00001420 exit(0);
1421 }
drhb0c86772000-06-02 23:21:26 +00001422 if( OptNArgs()!=1 ){
drh75897232000-05-29 14:26:00 +00001423 fprintf(stderr,"Exactly one filename argument is required.\n");
1424 exit(1);
1425 }
drh954f6b42006-06-13 13:27:46 +00001426 memset(&lem, 0, sizeof(lem));
drh75897232000-05-29 14:26:00 +00001427 lem.errorcnt = 0;
1428
1429 /* Initialize the machine */
1430 Strsafe_init();
1431 Symbol_init();
1432 State_init();
1433 lem.argv0 = argv[0];
drhb0c86772000-06-02 23:21:26 +00001434 lem.filename = OptArg(0);
drh75897232000-05-29 14:26:00 +00001435 lem.basisflag = basisflag;
shane58543932008-12-10 20:10:04 +00001436 lem.nolinenosflag = nolinenosflag;
drh75897232000-05-29 14:26:00 +00001437 Symbol_new("$");
1438 lem.errsym = Symbol_new("error");
drhc4dd3fd2008-01-22 01:48:05 +00001439 lem.errsym->useCnt = 0;
drh75897232000-05-29 14:26:00 +00001440
1441 /* Parse the input file */
1442 Parse(&lem);
1443 if( lem.errorcnt ) exit(lem.errorcnt);
drh954f6b42006-06-13 13:27:46 +00001444 if( lem.nrule==0 ){
drh75897232000-05-29 14:26:00 +00001445 fprintf(stderr,"Empty grammar.\n");
1446 exit(1);
1447 }
1448
1449 /* Count and index the symbols of the grammar */
1450 lem.nsymbol = Symbol_count();
1451 Symbol_new("{default}");
1452 lem.symbols = Symbol_arrayof();
drh60d31652004-02-22 00:08:04 +00001453 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
icculus9e44cf12010-02-14 17:14:22 +00001454 qsort(lem.symbols,lem.nsymbol+1,sizeof(struct symbol*), Symbolcmpp);
drh75897232000-05-29 14:26:00 +00001455 for(i=0; i<=lem.nsymbol; i++) lem.symbols[i]->index = i;
1456 for(i=1; isupper(lem.symbols[i]->name[0]); i++);
1457 lem.nterminal = i;
1458
1459 /* Generate a reprint of the grammar, if requested on the command line */
1460 if( rpflag ){
1461 Reprint(&lem);
1462 }else{
1463 /* Initialize the size for all follow and first sets */
drh9892c5d2007-12-21 00:02:11 +00001464 SetSize(lem.nterminal+1);
drh75897232000-05-29 14:26:00 +00001465
1466 /* Find the precedence for every production rule (that has one) */
1467 FindRulePrecedences(&lem);
1468
1469 /* Compute the lambda-nonterminals and the first-sets for every
1470 ** nonterminal */
1471 FindFirstSets(&lem);
1472
1473 /* Compute all LR(0) states. Also record follow-set propagation
1474 ** links so that the follow-set can be computed later */
1475 lem.nstate = 0;
1476 FindStates(&lem);
1477 lem.sorted = State_arrayof();
1478
1479 /* Tie up loose ends on the propagation links */
1480 FindLinks(&lem);
1481
1482 /* Compute the follow set of every reducible configuration */
1483 FindFollowSets(&lem);
1484
1485 /* Compute the action tables */
1486 FindActions(&lem);
1487
1488 /* Compress the action tables */
1489 if( compress==0 ) CompressTables(&lem);
1490
drhada354d2005-11-05 15:03:59 +00001491 /* Reorder and renumber the states so that states with fewer choices
drhdd7e9db2010-07-19 01:52:07 +00001492 ** occur at the end. This is an optimization that helps make the
1493 ** generated parser tables smaller. */
1494 if( noResort==0 ) ResortStates(&lem);
drhada354d2005-11-05 15:03:59 +00001495
drh75897232000-05-29 14:26:00 +00001496 /* Generate a report of the parser generated. (the "y.output" file) */
1497 if( !quiet ) ReportOutput(&lem);
1498
1499 /* Generate the source code for the parser */
1500 ReportTable(&lem, mhflag);
1501
1502 /* Produce a header file for use by the scanner. (This step is
1503 ** omitted if the "-m" option is used because makeheaders will
1504 ** generate the file for us.) */
1505 if( !mhflag ) ReportHeader(&lem);
1506 }
1507 if( statistics ){
1508 printf("Parser statistics: %d terminals, %d nonterminals, %d rules\n",
1509 lem.nterminal, lem.nsymbol - lem.nterminal, lem.nrule);
1510 printf(" %d states, %d parser table entries, %d conflicts\n",
1511 lem.nstate, lem.tablesize, lem.nconflict);
1512 }
icculus8e158022010-02-16 16:09:03 +00001513 if( lem.nconflict > 0 ){
1514 fprintf(stderr,"%d parsing conflicts.\n",lem.nconflict);
icculus42585cf2010-02-14 05:19:56 +00001515 }
1516
1517 /* return 0 on success, 1 on failure. */
icculus8e158022010-02-16 16:09:03 +00001518 exitcode = ((lem.errorcnt > 0) || (lem.nconflict > 0)) ? 1 : 0;
icculus42585cf2010-02-14 05:19:56 +00001519 exit(exitcode);
1520 return (exitcode);
drh75897232000-05-29 14:26:00 +00001521}
1522/******************** From the file "msort.c" *******************************/
1523/*
1524** A generic merge-sort program.
1525**
1526** USAGE:
1527** Let "ptr" be a pointer to some structure which is at the head of
1528** a null-terminated list. Then to sort the list call:
1529**
1530** ptr = msort(ptr,&(ptr->next),cmpfnc);
1531**
1532** In the above, "cmpfnc" is a pointer to a function which compares
1533** two instances of the structure and returns an integer, as in
1534** strcmp. The second argument is a pointer to the pointer to the
1535** second element of the linked list. This address is used to compute
1536** the offset to the "next" field within the structure. The offset to
1537** the "next" field must be constant for all structures in the list.
1538**
1539** The function returns a new pointer which is the head of the list
1540** after sorting.
1541**
1542** ALGORITHM:
1543** Merge-sort.
1544*/
1545
1546/*
1547** Return a pointer to the next structure in the linked list.
1548*/
drhd25d6922012-04-18 09:59:56 +00001549#define NEXT(A) (*(char**)(((char*)A)+offset))
drh75897232000-05-29 14:26:00 +00001550
1551/*
1552** Inputs:
1553** a: A sorted, null-terminated linked list. (May be null).
1554** b: A sorted, null-terminated linked list. (May be null).
1555** cmp: A pointer to the comparison function.
1556** offset: Offset in the structure to the "next" field.
1557**
1558** Return Value:
1559** A pointer to the head of a sorted list containing the elements
1560** of both a and b.
1561**
1562** Side effects:
1563** The "next" pointers for elements in the lists a and b are
1564** changed.
1565*/
drhe9278182007-07-18 18:16:29 +00001566static char *merge(
1567 char *a,
1568 char *b,
1569 int (*cmp)(const char*,const char*),
1570 int offset
1571){
drh75897232000-05-29 14:26:00 +00001572 char *ptr, *head;
1573
1574 if( a==0 ){
1575 head = b;
1576 }else if( b==0 ){
1577 head = a;
1578 }else{
drhe594bc32009-11-03 13:02:25 +00001579 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001580 ptr = a;
1581 a = NEXT(a);
1582 }else{
1583 ptr = b;
1584 b = NEXT(b);
1585 }
1586 head = ptr;
1587 while( a && b ){
drhe594bc32009-11-03 13:02:25 +00001588 if( (*cmp)(a,b)<=0 ){
drh75897232000-05-29 14:26:00 +00001589 NEXT(ptr) = a;
1590 ptr = a;
1591 a = NEXT(a);
1592 }else{
1593 NEXT(ptr) = b;
1594 ptr = b;
1595 b = NEXT(b);
1596 }
1597 }
1598 if( a ) NEXT(ptr) = a;
1599 else NEXT(ptr) = b;
1600 }
1601 return head;
1602}
1603
1604/*
1605** Inputs:
1606** list: Pointer to a singly-linked list of structures.
1607** next: Pointer to pointer to the second element of the list.
1608** cmp: A comparison function.
1609**
1610** Return Value:
1611** A pointer to the head of a sorted list containing the elements
1612** orginally in list.
1613**
1614** Side effects:
1615** The "next" pointers for elements in list are changed.
1616*/
1617#define LISTSIZE 30
drhe9278182007-07-18 18:16:29 +00001618static char *msort(
1619 char *list,
1620 char **next,
1621 int (*cmp)(const char*,const char*)
1622){
drhba99af52001-10-25 20:37:16 +00001623 unsigned long offset;
drh75897232000-05-29 14:26:00 +00001624 char *ep;
1625 char *set[LISTSIZE];
1626 int i;
drhba99af52001-10-25 20:37:16 +00001627 offset = (unsigned long)next - (unsigned long)list;
drh75897232000-05-29 14:26:00 +00001628 for(i=0; i<LISTSIZE; i++) set[i] = 0;
1629 while( list ){
1630 ep = list;
1631 list = NEXT(list);
1632 NEXT(ep) = 0;
1633 for(i=0; i<LISTSIZE-1 && set[i]!=0; i++){
1634 ep = merge(ep,set[i],cmp,offset);
1635 set[i] = 0;
1636 }
1637 set[i] = ep;
1638 }
1639 ep = 0;
drhe594bc32009-11-03 13:02:25 +00001640 for(i=0; i<LISTSIZE; i++) if( set[i] ) ep = merge(set[i],ep,cmp,offset);
drh75897232000-05-29 14:26:00 +00001641 return ep;
1642}
1643/************************ From the file "option.c" **************************/
1644static char **argv;
1645static struct s_options *op;
1646static FILE *errstream;
1647
1648#define ISOPT(X) ((X)[0]=='-'||(X)[0]=='+'||strchr((X),'=')!=0)
1649
1650/*
1651** Print the command line with a carrot pointing to the k-th character
1652** of the n-th field.
1653*/
icculus9e44cf12010-02-14 17:14:22 +00001654static void errline(int n, int k, FILE *err)
drh75897232000-05-29 14:26:00 +00001655{
1656 int spcnt, i;
drh75897232000-05-29 14:26:00 +00001657 if( argv[0] ) fprintf(err,"%s",argv[0]);
drh87cf1372008-08-13 20:09:06 +00001658 spcnt = lemonStrlen(argv[0]) + 1;
drh75897232000-05-29 14:26:00 +00001659 for(i=1; i<n && argv[i]; i++){
1660 fprintf(err," %s",argv[i]);
drh87cf1372008-08-13 20:09:06 +00001661 spcnt += lemonStrlen(argv[i])+1;
drh75897232000-05-29 14:26:00 +00001662 }
1663 spcnt += k;
1664 for(; argv[i]; i++) fprintf(err," %s",argv[i]);
1665 if( spcnt<20 ){
1666 fprintf(err,"\n%*s^-- here\n",spcnt,"");
1667 }else{
1668 fprintf(err,"\n%*shere --^\n",spcnt-7,"");
1669 }
1670}
1671
1672/*
1673** Return the index of the N-th non-switch argument. Return -1
1674** if N is out of range.
1675*/
icculus9e44cf12010-02-14 17:14:22 +00001676static int argindex(int n)
drh75897232000-05-29 14:26:00 +00001677{
1678 int i;
1679 int dashdash = 0;
1680 if( argv!=0 && *argv!=0 ){
1681 for(i=1; argv[i]; i++){
1682 if( dashdash || !ISOPT(argv[i]) ){
1683 if( n==0 ) return i;
1684 n--;
1685 }
1686 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1687 }
1688 }
1689 return -1;
1690}
1691
1692static char emsg[] = "Command line syntax error: ";
1693
1694/*
1695** Process a flag command line argument.
1696*/
icculus9e44cf12010-02-14 17:14:22 +00001697static int handleflags(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001698{
1699 int v;
1700 int errcnt = 0;
1701 int j;
1702 for(j=0; op[j].label; j++){
drh87cf1372008-08-13 20:09:06 +00001703 if( strncmp(&argv[i][1],op[j].label,lemonStrlen(op[j].label))==0 ) break;
drh75897232000-05-29 14:26:00 +00001704 }
1705 v = argv[i][0]=='-' ? 1 : 0;
1706 if( op[j].label==0 ){
1707 if( err ){
1708 fprintf(err,"%sundefined option.\n",emsg);
1709 errline(i,1,err);
1710 }
1711 errcnt++;
1712 }else if( op[j].type==OPT_FLAG ){
1713 *((int*)op[j].arg) = v;
1714 }else if( op[j].type==OPT_FFLAG ){
icculus9e44cf12010-02-14 17:14:22 +00001715 (*(void(*)(int))(op[j].arg))(v);
drh6d08b4d2004-07-20 12:45:22 +00001716 }else if( op[j].type==OPT_FSTR ){
icculus9e44cf12010-02-14 17:14:22 +00001717 (*(void(*)(char *))(op[j].arg))(&argv[i][2]);
drh75897232000-05-29 14:26:00 +00001718 }else{
1719 if( err ){
1720 fprintf(err,"%smissing argument on switch.\n",emsg);
1721 errline(i,1,err);
1722 }
1723 errcnt++;
1724 }
1725 return errcnt;
1726}
1727
1728/*
1729** Process a command line switch which has an argument.
1730*/
icculus9e44cf12010-02-14 17:14:22 +00001731static int handleswitch(int i, FILE *err)
drh75897232000-05-29 14:26:00 +00001732{
1733 int lv = 0;
1734 double dv = 0.0;
1735 char *sv = 0, *end;
1736 char *cp;
1737 int j;
1738 int errcnt = 0;
1739 cp = strchr(argv[i],'=');
drh43617e92006-03-06 20:55:46 +00001740 assert( cp!=0 );
drh75897232000-05-29 14:26:00 +00001741 *cp = 0;
1742 for(j=0; op[j].label; j++){
1743 if( strcmp(argv[i],op[j].label)==0 ) break;
1744 }
1745 *cp = '=';
1746 if( op[j].label==0 ){
1747 if( err ){
1748 fprintf(err,"%sundefined option.\n",emsg);
1749 errline(i,0,err);
1750 }
1751 errcnt++;
1752 }else{
1753 cp++;
1754 switch( op[j].type ){
1755 case OPT_FLAG:
1756 case OPT_FFLAG:
1757 if( err ){
1758 fprintf(err,"%soption requires an argument.\n",emsg);
1759 errline(i,0,err);
1760 }
1761 errcnt++;
1762 break;
1763 case OPT_DBL:
1764 case OPT_FDBL:
1765 dv = strtod(cp,&end);
1766 if( *end ){
1767 if( err ){
1768 fprintf(err,"%sillegal character in floating-point argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001769 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001770 }
1771 errcnt++;
1772 }
1773 break;
1774 case OPT_INT:
1775 case OPT_FINT:
1776 lv = strtol(cp,&end,0);
1777 if( *end ){
1778 if( err ){
1779 fprintf(err,"%sillegal character in integer argument.\n",emsg);
drhba99af52001-10-25 20:37:16 +00001780 errline(i,((unsigned long)end)-(unsigned long)argv[i],err);
drh75897232000-05-29 14:26:00 +00001781 }
1782 errcnt++;
1783 }
1784 break;
1785 case OPT_STR:
1786 case OPT_FSTR:
1787 sv = cp;
1788 break;
1789 }
1790 switch( op[j].type ){
1791 case OPT_FLAG:
1792 case OPT_FFLAG:
1793 break;
1794 case OPT_DBL:
1795 *(double*)(op[j].arg) = dv;
1796 break;
1797 case OPT_FDBL:
icculus9e44cf12010-02-14 17:14:22 +00001798 (*(void(*)(double))(op[j].arg))(dv);
drh75897232000-05-29 14:26:00 +00001799 break;
1800 case OPT_INT:
1801 *(int*)(op[j].arg) = lv;
1802 break;
1803 case OPT_FINT:
icculus9e44cf12010-02-14 17:14:22 +00001804 (*(void(*)(int))(op[j].arg))((int)lv);
drh75897232000-05-29 14:26:00 +00001805 break;
1806 case OPT_STR:
1807 *(char**)(op[j].arg) = sv;
1808 break;
1809 case OPT_FSTR:
icculus9e44cf12010-02-14 17:14:22 +00001810 (*(void(*)(char *))(op[j].arg))(sv);
drh75897232000-05-29 14:26:00 +00001811 break;
1812 }
1813 }
1814 return errcnt;
1815}
1816
icculus9e44cf12010-02-14 17:14:22 +00001817int OptInit(char **a, struct s_options *o, FILE *err)
drh75897232000-05-29 14:26:00 +00001818{
1819 int errcnt = 0;
1820 argv = a;
1821 op = o;
1822 errstream = err;
1823 if( argv && *argv && op ){
1824 int i;
1825 for(i=1; argv[i]; i++){
1826 if( argv[i][0]=='+' || argv[i][0]=='-' ){
1827 errcnt += handleflags(i,err);
1828 }else if( strchr(argv[i],'=') ){
1829 errcnt += handleswitch(i,err);
1830 }
1831 }
1832 }
1833 if( errcnt>0 ){
1834 fprintf(err,"Valid command line options for \"%s\" are:\n",*a);
drhb0c86772000-06-02 23:21:26 +00001835 OptPrint();
drh75897232000-05-29 14:26:00 +00001836 exit(1);
1837 }
1838 return 0;
1839}
1840
drhb0c86772000-06-02 23:21:26 +00001841int OptNArgs(){
drh75897232000-05-29 14:26:00 +00001842 int cnt = 0;
1843 int dashdash = 0;
1844 int i;
1845 if( argv!=0 && argv[0]!=0 ){
1846 for(i=1; argv[i]; i++){
1847 if( dashdash || !ISOPT(argv[i]) ) cnt++;
1848 if( strcmp(argv[i],"--")==0 ) dashdash = 1;
1849 }
1850 }
1851 return cnt;
1852}
1853
icculus9e44cf12010-02-14 17:14:22 +00001854char *OptArg(int n)
drh75897232000-05-29 14:26:00 +00001855{
1856 int i;
1857 i = argindex(n);
1858 return i>=0 ? argv[i] : 0;
1859}
1860
icculus9e44cf12010-02-14 17:14:22 +00001861void OptErr(int n)
drh75897232000-05-29 14:26:00 +00001862{
1863 int i;
1864 i = argindex(n);
1865 if( i>=0 ) errline(i,0,errstream);
1866}
1867
drhb0c86772000-06-02 23:21:26 +00001868void OptPrint(){
drh75897232000-05-29 14:26:00 +00001869 int i;
1870 int max, len;
1871 max = 0;
1872 for(i=0; op[i].label; i++){
drh87cf1372008-08-13 20:09:06 +00001873 len = lemonStrlen(op[i].label) + 1;
drh75897232000-05-29 14:26:00 +00001874 switch( op[i].type ){
1875 case OPT_FLAG:
1876 case OPT_FFLAG:
1877 break;
1878 case OPT_INT:
1879 case OPT_FINT:
1880 len += 9; /* length of "<integer>" */
1881 break;
1882 case OPT_DBL:
1883 case OPT_FDBL:
1884 len += 6; /* length of "<real>" */
1885 break;
1886 case OPT_STR:
1887 case OPT_FSTR:
1888 len += 8; /* length of "<string>" */
1889 break;
1890 }
1891 if( len>max ) max = len;
1892 }
1893 for(i=0; op[i].label; i++){
1894 switch( op[i].type ){
1895 case OPT_FLAG:
1896 case OPT_FFLAG:
1897 fprintf(errstream," -%-*s %s\n",max,op[i].label,op[i].message);
1898 break;
1899 case OPT_INT:
1900 case OPT_FINT:
1901 fprintf(errstream," %s=<integer>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001902 (int)(max-lemonStrlen(op[i].label)-9),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001903 break;
1904 case OPT_DBL:
1905 case OPT_FDBL:
1906 fprintf(errstream," %s=<real>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001907 (int)(max-lemonStrlen(op[i].label)-6),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001908 break;
1909 case OPT_STR:
1910 case OPT_FSTR:
1911 fprintf(errstream," %s=<string>%*s %s\n",op[i].label,
drh87cf1372008-08-13 20:09:06 +00001912 (int)(max-lemonStrlen(op[i].label)-8),"",op[i].message);
drh75897232000-05-29 14:26:00 +00001913 break;
1914 }
1915 }
1916}
1917/*********************** From the file "parse.c" ****************************/
1918/*
1919** Input file parser for the LEMON parser generator.
1920*/
1921
1922/* The state of the parser */
icculus9e44cf12010-02-14 17:14:22 +00001923enum e_state {
1924 INITIALIZE,
1925 WAITING_FOR_DECL_OR_RULE,
1926 WAITING_FOR_DECL_KEYWORD,
1927 WAITING_FOR_DECL_ARG,
1928 WAITING_FOR_PRECEDENCE_SYMBOL,
1929 WAITING_FOR_ARROW,
1930 IN_RHS,
1931 LHS_ALIAS_1,
1932 LHS_ALIAS_2,
1933 LHS_ALIAS_3,
1934 RHS_ALIAS_1,
1935 RHS_ALIAS_2,
1936 PRECEDENCE_MARK_1,
1937 PRECEDENCE_MARK_2,
1938 RESYNC_AFTER_RULE_ERROR,
1939 RESYNC_AFTER_DECL_ERROR,
1940 WAITING_FOR_DESTRUCTOR_SYMBOL,
1941 WAITING_FOR_DATATYPE_SYMBOL,
1942 WAITING_FOR_FALLBACK_ID,
icculus9e44cf12010-02-14 17:14:22 +00001943 WAITING_FOR_WILDCARD_ID
1944};
drh75897232000-05-29 14:26:00 +00001945struct pstate {
1946 char *filename; /* Name of the input file */
1947 int tokenlineno; /* Linenumber at which current token starts */
1948 int errorcnt; /* Number of errors so far */
1949 char *tokenstart; /* Text of current token */
1950 struct lemon *gp; /* Global state vector */
icculus9e44cf12010-02-14 17:14:22 +00001951 enum e_state state; /* The state of the parser */
drh0bd1f4e2002-06-06 18:54:39 +00001952 struct symbol *fallback; /* The fallback token */
drh75897232000-05-29 14:26:00 +00001953 struct symbol *lhs; /* Left-hand side of current rule */
icculus9e44cf12010-02-14 17:14:22 +00001954 const char *lhsalias; /* Alias for the LHS */
drh75897232000-05-29 14:26:00 +00001955 int nrhs; /* Number of right-hand side symbols seen */
1956 struct symbol *rhs[MAXRHS]; /* RHS symbols */
icculus9e44cf12010-02-14 17:14:22 +00001957 const char *alias[MAXRHS]; /* Aliases for each RHS symbol (or NULL) */
drh75897232000-05-29 14:26:00 +00001958 struct rule *prevrule; /* Previous rule parsed */
icculus9e44cf12010-02-14 17:14:22 +00001959 const char *declkeyword; /* Keyword of a declaration */
drh75897232000-05-29 14:26:00 +00001960 char **declargslot; /* Where the declaration argument should be put */
drha5808f32008-04-27 22:19:44 +00001961 int insertLineMacro; /* Add #line before declaration insert */
drh4dc8ef52008-07-01 17:13:57 +00001962 int *decllinenoslot; /* Where to write declaration line number */
drh75897232000-05-29 14:26:00 +00001963 enum e_assoc declassoc; /* Assign this association to decl arguments */
1964 int preccounter; /* Assign this precedence to decl arguments */
1965 struct rule *firstrule; /* Pointer to first rule in the grammar */
1966 struct rule *lastrule; /* Pointer to the most recently parsed rule */
1967};
1968
1969/* Parse a single token */
icculus9e44cf12010-02-14 17:14:22 +00001970static void parseonetoken(struct pstate *psp)
drh75897232000-05-29 14:26:00 +00001971{
icculus9e44cf12010-02-14 17:14:22 +00001972 const char *x;
drh75897232000-05-29 14:26:00 +00001973 x = Strsafe(psp->tokenstart); /* Save the token permanently */
1974#if 0
1975 printf("%s:%d: Token=[%s] state=%d\n",psp->filename,psp->tokenlineno,
1976 x,psp->state);
1977#endif
1978 switch( psp->state ){
1979 case INITIALIZE:
1980 psp->prevrule = 0;
1981 psp->preccounter = 0;
1982 psp->firstrule = psp->lastrule = 0;
1983 psp->gp->nrule = 0;
1984 /* Fall thru to next case */
1985 case WAITING_FOR_DECL_OR_RULE:
1986 if( x[0]=='%' ){
1987 psp->state = WAITING_FOR_DECL_KEYWORD;
1988 }else if( islower(x[0]) ){
1989 psp->lhs = Symbol_new(x);
1990 psp->nrhs = 0;
1991 psp->lhsalias = 0;
1992 psp->state = WAITING_FOR_ARROW;
1993 }else if( x[0]=='{' ){
1994 if( psp->prevrule==0 ){
1995 ErrorMsg(psp->filename,psp->tokenlineno,
drh3cb2f6e2012-01-09 14:19:05 +00001996"There is no prior rule upon which to attach the code \
drh75897232000-05-29 14:26:00 +00001997fragment which begins on this line.");
1998 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00001999 }else if( psp->prevrule->code!=0 ){
drh75897232000-05-29 14:26:00 +00002000 ErrorMsg(psp->filename,psp->tokenlineno,
2001"Code fragment beginning on this line is not the first \
2002to follow the previous rule.");
2003 psp->errorcnt++;
2004 }else{
2005 psp->prevrule->line = psp->tokenlineno;
2006 psp->prevrule->code = &x[1];
drhf2f105d2012-08-20 15:53:54 +00002007 }
drh75897232000-05-29 14:26:00 +00002008 }else if( x[0]=='[' ){
2009 psp->state = PRECEDENCE_MARK_1;
2010 }else{
2011 ErrorMsg(psp->filename,psp->tokenlineno,
2012 "Token \"%s\" should be either \"%%\" or a nonterminal name.",
2013 x);
2014 psp->errorcnt++;
2015 }
2016 break;
2017 case PRECEDENCE_MARK_1:
2018 if( !isupper(x[0]) ){
2019 ErrorMsg(psp->filename,psp->tokenlineno,
2020 "The precedence symbol must be a terminal.");
2021 psp->errorcnt++;
2022 }else if( psp->prevrule==0 ){
2023 ErrorMsg(psp->filename,psp->tokenlineno,
2024 "There is no prior rule to assign precedence \"[%s]\".",x);
2025 psp->errorcnt++;
2026 }else if( psp->prevrule->precsym!=0 ){
2027 ErrorMsg(psp->filename,psp->tokenlineno,
2028"Precedence mark on this line is not the first \
2029to follow the previous rule.");
2030 psp->errorcnt++;
2031 }else{
2032 psp->prevrule->precsym = Symbol_new(x);
2033 }
2034 psp->state = PRECEDENCE_MARK_2;
2035 break;
2036 case PRECEDENCE_MARK_2:
2037 if( x[0]!=']' ){
2038 ErrorMsg(psp->filename,psp->tokenlineno,
2039 "Missing \"]\" on precedence mark.");
2040 psp->errorcnt++;
2041 }
2042 psp->state = WAITING_FOR_DECL_OR_RULE;
2043 break;
2044 case WAITING_FOR_ARROW:
2045 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2046 psp->state = IN_RHS;
2047 }else if( x[0]=='(' ){
2048 psp->state = LHS_ALIAS_1;
2049 }else{
2050 ErrorMsg(psp->filename,psp->tokenlineno,
2051 "Expected to see a \":\" following the LHS symbol \"%s\".",
2052 psp->lhs->name);
2053 psp->errorcnt++;
2054 psp->state = RESYNC_AFTER_RULE_ERROR;
2055 }
2056 break;
2057 case LHS_ALIAS_1:
2058 if( isalpha(x[0]) ){
2059 psp->lhsalias = x;
2060 psp->state = LHS_ALIAS_2;
2061 }else{
2062 ErrorMsg(psp->filename,psp->tokenlineno,
2063 "\"%s\" is not a valid alias for the LHS \"%s\"\n",
2064 x,psp->lhs->name);
2065 psp->errorcnt++;
2066 psp->state = RESYNC_AFTER_RULE_ERROR;
2067 }
2068 break;
2069 case LHS_ALIAS_2:
2070 if( x[0]==')' ){
2071 psp->state = LHS_ALIAS_3;
2072 }else{
2073 ErrorMsg(psp->filename,psp->tokenlineno,
2074 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2075 psp->errorcnt++;
2076 psp->state = RESYNC_AFTER_RULE_ERROR;
2077 }
2078 break;
2079 case LHS_ALIAS_3:
2080 if( x[0]==':' && x[1]==':' && x[2]=='=' ){
2081 psp->state = IN_RHS;
2082 }else{
2083 ErrorMsg(psp->filename,psp->tokenlineno,
2084 "Missing \"->\" following: \"%s(%s)\".",
2085 psp->lhs->name,psp->lhsalias);
2086 psp->errorcnt++;
2087 psp->state = RESYNC_AFTER_RULE_ERROR;
2088 }
2089 break;
2090 case IN_RHS:
2091 if( x[0]=='.' ){
2092 struct rule *rp;
drh9892c5d2007-12-21 00:02:11 +00002093 rp = (struct rule *)calloc( sizeof(struct rule) +
2094 sizeof(struct symbol*)*psp->nrhs + sizeof(char*)*psp->nrhs, 1);
drh75897232000-05-29 14:26:00 +00002095 if( rp==0 ){
2096 ErrorMsg(psp->filename,psp->tokenlineno,
2097 "Can't allocate enough memory for this rule.");
2098 psp->errorcnt++;
2099 psp->prevrule = 0;
drhf2f105d2012-08-20 15:53:54 +00002100 }else{
drh75897232000-05-29 14:26:00 +00002101 int i;
2102 rp->ruleline = psp->tokenlineno;
2103 rp->rhs = (struct symbol**)&rp[1];
icculus9e44cf12010-02-14 17:14:22 +00002104 rp->rhsalias = (const char**)&(rp->rhs[psp->nrhs]);
drh75897232000-05-29 14:26:00 +00002105 for(i=0; i<psp->nrhs; i++){
2106 rp->rhs[i] = psp->rhs[i];
2107 rp->rhsalias[i] = psp->alias[i];
drhf2f105d2012-08-20 15:53:54 +00002108 }
drh75897232000-05-29 14:26:00 +00002109 rp->lhs = psp->lhs;
2110 rp->lhsalias = psp->lhsalias;
2111 rp->nrhs = psp->nrhs;
2112 rp->code = 0;
2113 rp->precsym = 0;
2114 rp->index = psp->gp->nrule++;
2115 rp->nextlhs = rp->lhs->rule;
2116 rp->lhs->rule = rp;
2117 rp->next = 0;
2118 if( psp->firstrule==0 ){
2119 psp->firstrule = psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002120 }else{
drh75897232000-05-29 14:26:00 +00002121 psp->lastrule->next = rp;
2122 psp->lastrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002123 }
drh75897232000-05-29 14:26:00 +00002124 psp->prevrule = rp;
drhf2f105d2012-08-20 15:53:54 +00002125 }
drh75897232000-05-29 14:26:00 +00002126 psp->state = WAITING_FOR_DECL_OR_RULE;
2127 }else if( isalpha(x[0]) ){
2128 if( psp->nrhs>=MAXRHS ){
2129 ErrorMsg(psp->filename,psp->tokenlineno,
drhc4dd3fd2008-01-22 01:48:05 +00002130 "Too many symbols on RHS of rule beginning at \"%s\".",
drh75897232000-05-29 14:26:00 +00002131 x);
2132 psp->errorcnt++;
2133 psp->state = RESYNC_AFTER_RULE_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002134 }else{
drh75897232000-05-29 14:26:00 +00002135 psp->rhs[psp->nrhs] = Symbol_new(x);
2136 psp->alias[psp->nrhs] = 0;
2137 psp->nrhs++;
drhf2f105d2012-08-20 15:53:54 +00002138 }
drhfd405312005-11-06 04:06:59 +00002139 }else if( (x[0]=='|' || x[0]=='/') && psp->nrhs>0 ){
2140 struct symbol *msp = psp->rhs[psp->nrhs-1];
2141 if( msp->type!=MULTITERMINAL ){
2142 struct symbol *origsp = msp;
icculus9e44cf12010-02-14 17:14:22 +00002143 msp = (struct symbol *) calloc(1,sizeof(*msp));
drhfd405312005-11-06 04:06:59 +00002144 memset(msp, 0, sizeof(*msp));
2145 msp->type = MULTITERMINAL;
2146 msp->nsubsym = 1;
icculus9e44cf12010-02-14 17:14:22 +00002147 msp->subsym = (struct symbol **) calloc(1,sizeof(struct symbol*));
drhfd405312005-11-06 04:06:59 +00002148 msp->subsym[0] = origsp;
2149 msp->name = origsp->name;
2150 psp->rhs[psp->nrhs-1] = msp;
2151 }
2152 msp->nsubsym++;
icculus9e44cf12010-02-14 17:14:22 +00002153 msp->subsym = (struct symbol **) realloc(msp->subsym,
2154 sizeof(struct symbol*)*msp->nsubsym);
drhfd405312005-11-06 04:06:59 +00002155 msp->subsym[msp->nsubsym-1] = Symbol_new(&x[1]);
2156 if( islower(x[1]) || islower(msp->subsym[0]->name[0]) ){
2157 ErrorMsg(psp->filename,psp->tokenlineno,
2158 "Cannot form a compound containing a non-terminal");
2159 psp->errorcnt++;
2160 }
drh75897232000-05-29 14:26:00 +00002161 }else if( x[0]=='(' && psp->nrhs>0 ){
2162 psp->state = RHS_ALIAS_1;
2163 }else{
2164 ErrorMsg(psp->filename,psp->tokenlineno,
2165 "Illegal character on RHS of rule: \"%s\".",x);
2166 psp->errorcnt++;
2167 psp->state = RESYNC_AFTER_RULE_ERROR;
2168 }
2169 break;
2170 case RHS_ALIAS_1:
2171 if( isalpha(x[0]) ){
2172 psp->alias[psp->nrhs-1] = x;
2173 psp->state = RHS_ALIAS_2;
2174 }else{
2175 ErrorMsg(psp->filename,psp->tokenlineno,
2176 "\"%s\" is not a valid alias for the RHS symbol \"%s\"\n",
2177 x,psp->rhs[psp->nrhs-1]->name);
2178 psp->errorcnt++;
2179 psp->state = RESYNC_AFTER_RULE_ERROR;
2180 }
2181 break;
2182 case RHS_ALIAS_2:
2183 if( x[0]==')' ){
2184 psp->state = IN_RHS;
2185 }else{
2186 ErrorMsg(psp->filename,psp->tokenlineno,
2187 "Missing \")\" following LHS alias name \"%s\".",psp->lhsalias);
2188 psp->errorcnt++;
2189 psp->state = RESYNC_AFTER_RULE_ERROR;
2190 }
2191 break;
2192 case WAITING_FOR_DECL_KEYWORD:
2193 if( isalpha(x[0]) ){
2194 psp->declkeyword = x;
2195 psp->declargslot = 0;
drh4dc8ef52008-07-01 17:13:57 +00002196 psp->decllinenoslot = 0;
drha5808f32008-04-27 22:19:44 +00002197 psp->insertLineMacro = 1;
drh75897232000-05-29 14:26:00 +00002198 psp->state = WAITING_FOR_DECL_ARG;
2199 if( strcmp(x,"name")==0 ){
2200 psp->declargslot = &(psp->gp->name);
drha5808f32008-04-27 22:19:44 +00002201 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002202 }else if( strcmp(x,"include")==0 ){
drh75897232000-05-29 14:26:00 +00002203 psp->declargslot = &(psp->gp->include);
drhf2f105d2012-08-20 15:53:54 +00002204 }else if( strcmp(x,"code")==0 ){
drh75897232000-05-29 14:26:00 +00002205 psp->declargslot = &(psp->gp->extracode);
drhf2f105d2012-08-20 15:53:54 +00002206 }else if( strcmp(x,"token_destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002207 psp->declargslot = &psp->gp->tokendest;
drhf2f105d2012-08-20 15:53:54 +00002208 }else if( strcmp(x,"default_destructor")==0 ){
drh960e8c62001-04-03 16:53:21 +00002209 psp->declargslot = &psp->gp->vardest;
drhf2f105d2012-08-20 15:53:54 +00002210 }else if( strcmp(x,"token_prefix")==0 ){
drh75897232000-05-29 14:26:00 +00002211 psp->declargslot = &psp->gp->tokenprefix;
drha5808f32008-04-27 22:19:44 +00002212 psp->insertLineMacro = 0;
drhf2f105d2012-08-20 15:53:54 +00002213 }else if( strcmp(x,"syntax_error")==0 ){
drh75897232000-05-29 14:26:00 +00002214 psp->declargslot = &(psp->gp->error);
drhf2f105d2012-08-20 15:53:54 +00002215 }else if( strcmp(x,"parse_accept")==0 ){
drh75897232000-05-29 14:26:00 +00002216 psp->declargslot = &(psp->gp->accept);
drhf2f105d2012-08-20 15:53:54 +00002217 }else if( strcmp(x,"parse_failure")==0 ){
drh75897232000-05-29 14:26:00 +00002218 psp->declargslot = &(psp->gp->failure);
drhf2f105d2012-08-20 15:53:54 +00002219 }else if( strcmp(x,"stack_overflow")==0 ){
drh75897232000-05-29 14:26:00 +00002220 psp->declargslot = &(psp->gp->overflow);
drh75897232000-05-29 14:26:00 +00002221 }else if( strcmp(x,"extra_argument")==0 ){
2222 psp->declargslot = &(psp->gp->arg);
drha5808f32008-04-27 22:19:44 +00002223 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002224 }else if( strcmp(x,"token_type")==0 ){
2225 psp->declargslot = &(psp->gp->tokentype);
drha5808f32008-04-27 22:19:44 +00002226 psp->insertLineMacro = 0;
drh960e8c62001-04-03 16:53:21 +00002227 }else if( strcmp(x,"default_type")==0 ){
2228 psp->declargslot = &(psp->gp->vartype);
drha5808f32008-04-27 22:19:44 +00002229 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002230 }else if( strcmp(x,"stack_size")==0 ){
2231 psp->declargslot = &(psp->gp->stacksize);
drha5808f32008-04-27 22:19:44 +00002232 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002233 }else if( strcmp(x,"start_symbol")==0 ){
2234 psp->declargslot = &(psp->gp->start);
drha5808f32008-04-27 22:19:44 +00002235 psp->insertLineMacro = 0;
drh75897232000-05-29 14:26:00 +00002236 }else if( strcmp(x,"left")==0 ){
2237 psp->preccounter++;
2238 psp->declassoc = LEFT;
2239 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2240 }else if( strcmp(x,"right")==0 ){
2241 psp->preccounter++;
2242 psp->declassoc = RIGHT;
2243 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
2244 }else if( strcmp(x,"nonassoc")==0 ){
2245 psp->preccounter++;
2246 psp->declassoc = NONE;
2247 psp->state = WAITING_FOR_PRECEDENCE_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002248 }else if( strcmp(x,"destructor")==0 ){
drh75897232000-05-29 14:26:00 +00002249 psp->state = WAITING_FOR_DESTRUCTOR_SYMBOL;
drhf2f105d2012-08-20 15:53:54 +00002250 }else if( strcmp(x,"type")==0 ){
drh75897232000-05-29 14:26:00 +00002251 psp->state = WAITING_FOR_DATATYPE_SYMBOL;
drh0bd1f4e2002-06-06 18:54:39 +00002252 }else if( strcmp(x,"fallback")==0 ){
2253 psp->fallback = 0;
2254 psp->state = WAITING_FOR_FALLBACK_ID;
drhe09daa92006-06-10 13:29:31 +00002255 }else if( strcmp(x,"wildcard")==0 ){
2256 psp->state = WAITING_FOR_WILDCARD_ID;
drh75897232000-05-29 14:26:00 +00002257 }else{
2258 ErrorMsg(psp->filename,psp->tokenlineno,
2259 "Unknown declaration keyword: \"%%%s\".",x);
2260 psp->errorcnt++;
2261 psp->state = RESYNC_AFTER_DECL_ERROR;
drhf2f105d2012-08-20 15:53:54 +00002262 }
drh75897232000-05-29 14:26:00 +00002263 }else{
2264 ErrorMsg(psp->filename,psp->tokenlineno,
2265 "Illegal declaration keyword: \"%s\".",x);
2266 psp->errorcnt++;
2267 psp->state = RESYNC_AFTER_DECL_ERROR;
2268 }
2269 break;
2270 case WAITING_FOR_DESTRUCTOR_SYMBOL:
2271 if( !isalpha(x[0]) ){
2272 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002273 "Symbol name missing after %%destructor keyword");
drh75897232000-05-29 14:26:00 +00002274 psp->errorcnt++;
2275 psp->state = RESYNC_AFTER_DECL_ERROR;
2276 }else{
icculusd286fa62010-03-03 17:06:32 +00002277 struct symbol *sp = Symbol_new(x);
2278 psp->declargslot = &sp->destructor;
2279 psp->decllinenoslot = &sp->destLineno;
2280 psp->insertLineMacro = 1;
2281 psp->state = WAITING_FOR_DECL_ARG;
drh75897232000-05-29 14:26:00 +00002282 }
2283 break;
2284 case WAITING_FOR_DATATYPE_SYMBOL:
2285 if( !isalpha(x[0]) ){
2286 ErrorMsg(psp->filename,psp->tokenlineno,
icculusd0d97b02010-02-17 20:22:10 +00002287 "Symbol name missing after %%type keyword");
drh75897232000-05-29 14:26:00 +00002288 psp->errorcnt++;
2289 psp->state = RESYNC_AFTER_DECL_ERROR;
2290 }else{
icculus866bf1e2010-02-17 20:31:32 +00002291 struct symbol *sp = Symbol_find(x);
2292 if((sp) && (sp->datatype)){
2293 ErrorMsg(psp->filename,psp->tokenlineno,
2294 "Symbol %%type \"%s\" already defined", x);
2295 psp->errorcnt++;
2296 psp->state = RESYNC_AFTER_DECL_ERROR;
2297 }else{
2298 if (!sp){
2299 sp = Symbol_new(x);
2300 }
2301 psp->declargslot = &sp->datatype;
2302 psp->insertLineMacro = 0;
2303 psp->state = WAITING_FOR_DECL_ARG;
2304 }
drh75897232000-05-29 14:26:00 +00002305 }
2306 break;
2307 case WAITING_FOR_PRECEDENCE_SYMBOL:
2308 if( x[0]=='.' ){
2309 psp->state = WAITING_FOR_DECL_OR_RULE;
2310 }else if( isupper(x[0]) ){
2311 struct symbol *sp;
2312 sp = Symbol_new(x);
2313 if( sp->prec>=0 ){
2314 ErrorMsg(psp->filename,psp->tokenlineno,
2315 "Symbol \"%s\" has already be given a precedence.",x);
2316 psp->errorcnt++;
drhf2f105d2012-08-20 15:53:54 +00002317 }else{
drh75897232000-05-29 14:26:00 +00002318 sp->prec = psp->preccounter;
2319 sp->assoc = psp->declassoc;
drhf2f105d2012-08-20 15:53:54 +00002320 }
drh75897232000-05-29 14:26:00 +00002321 }else{
2322 ErrorMsg(psp->filename,psp->tokenlineno,
2323 "Can't assign a precedence to \"%s\".",x);
2324 psp->errorcnt++;
2325 }
2326 break;
2327 case WAITING_FOR_DECL_ARG:
drha5808f32008-04-27 22:19:44 +00002328 if( x[0]=='{' || x[0]=='\"' || isalnum(x[0]) ){
icculus9e44cf12010-02-14 17:14:22 +00002329 const char *zOld, *zNew;
2330 char *zBuf, *z;
drha5808f32008-04-27 22:19:44 +00002331 int nOld, n, nLine, nNew, nBack;
drhb5bd49e2008-07-14 12:21:08 +00002332 int addLineMacro;
drha5808f32008-04-27 22:19:44 +00002333 char zLine[50];
2334 zNew = x;
2335 if( zNew[0]=='"' || zNew[0]=='{' ) zNew++;
drh87cf1372008-08-13 20:09:06 +00002336 nNew = lemonStrlen(zNew);
drha5808f32008-04-27 22:19:44 +00002337 if( *psp->declargslot ){
2338 zOld = *psp->declargslot;
2339 }else{
2340 zOld = "";
2341 }
drh87cf1372008-08-13 20:09:06 +00002342 nOld = lemonStrlen(zOld);
drha5808f32008-04-27 22:19:44 +00002343 n = nOld + nNew + 20;
shane58543932008-12-10 20:10:04 +00002344 addLineMacro = !psp->gp->nolinenosflag && psp->insertLineMacro &&
drhb5bd49e2008-07-14 12:21:08 +00002345 (psp->decllinenoslot==0 || psp->decllinenoslot[0]!=0);
2346 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002347 for(z=psp->filename, nBack=0; *z; z++){
2348 if( *z=='\\' ) nBack++;
2349 }
2350 sprintf(zLine, "#line %d ", psp->tokenlineno);
drh87cf1372008-08-13 20:09:06 +00002351 nLine = lemonStrlen(zLine);
2352 n += nLine + lemonStrlen(psp->filename) + nBack;
drha5808f32008-04-27 22:19:44 +00002353 }
icculus9e44cf12010-02-14 17:14:22 +00002354 *psp->declargslot = (char *) realloc(*psp->declargslot, n);
2355 zBuf = *psp->declargslot + nOld;
drhb5bd49e2008-07-14 12:21:08 +00002356 if( addLineMacro ){
drha5808f32008-04-27 22:19:44 +00002357 if( nOld && zBuf[-1]!='\n' ){
2358 *(zBuf++) = '\n';
2359 }
2360 memcpy(zBuf, zLine, nLine);
2361 zBuf += nLine;
2362 *(zBuf++) = '"';
2363 for(z=psp->filename; *z; z++){
2364 if( *z=='\\' ){
2365 *(zBuf++) = '\\';
2366 }
2367 *(zBuf++) = *z;
2368 }
2369 *(zBuf++) = '"';
2370 *(zBuf++) = '\n';
2371 }
drh4dc8ef52008-07-01 17:13:57 +00002372 if( psp->decllinenoslot && psp->decllinenoslot[0]==0 ){
2373 psp->decllinenoslot[0] = psp->tokenlineno;
2374 }
drha5808f32008-04-27 22:19:44 +00002375 memcpy(zBuf, zNew, nNew);
2376 zBuf += nNew;
2377 *zBuf = 0;
2378 psp->state = WAITING_FOR_DECL_OR_RULE;
drh75897232000-05-29 14:26:00 +00002379 }else{
2380 ErrorMsg(psp->filename,psp->tokenlineno,
2381 "Illegal argument to %%%s: %s",psp->declkeyword,x);
2382 psp->errorcnt++;
2383 psp->state = RESYNC_AFTER_DECL_ERROR;
2384 }
2385 break;
drh0bd1f4e2002-06-06 18:54:39 +00002386 case WAITING_FOR_FALLBACK_ID:
2387 if( x[0]=='.' ){
2388 psp->state = WAITING_FOR_DECL_OR_RULE;
2389 }else if( !isupper(x[0]) ){
2390 ErrorMsg(psp->filename, psp->tokenlineno,
2391 "%%fallback argument \"%s\" should be a token", x);
2392 psp->errorcnt++;
2393 }else{
2394 struct symbol *sp = Symbol_new(x);
2395 if( psp->fallback==0 ){
2396 psp->fallback = sp;
2397 }else if( sp->fallback ){
2398 ErrorMsg(psp->filename, psp->tokenlineno,
2399 "More than one fallback assigned to token %s", x);
2400 psp->errorcnt++;
2401 }else{
2402 sp->fallback = psp->fallback;
2403 psp->gp->has_fallback = 1;
2404 }
2405 }
2406 break;
drhe09daa92006-06-10 13:29:31 +00002407 case WAITING_FOR_WILDCARD_ID:
2408 if( x[0]=='.' ){
2409 psp->state = WAITING_FOR_DECL_OR_RULE;
2410 }else if( !isupper(x[0]) ){
2411 ErrorMsg(psp->filename, psp->tokenlineno,
2412 "%%wildcard argument \"%s\" should be a token", x);
2413 psp->errorcnt++;
2414 }else{
2415 struct symbol *sp = Symbol_new(x);
2416 if( psp->gp->wildcard==0 ){
2417 psp->gp->wildcard = sp;
2418 }else{
2419 ErrorMsg(psp->filename, psp->tokenlineno,
2420 "Extra wildcard to token: %s", x);
2421 psp->errorcnt++;
2422 }
2423 }
2424 break;
drh75897232000-05-29 14:26:00 +00002425 case RESYNC_AFTER_RULE_ERROR:
2426/* if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2427** break; */
2428 case RESYNC_AFTER_DECL_ERROR:
2429 if( x[0]=='.' ) psp->state = WAITING_FOR_DECL_OR_RULE;
2430 if( x[0]=='%' ) psp->state = WAITING_FOR_DECL_KEYWORD;
2431 break;
2432 }
2433}
2434
drh34ff57b2008-07-14 12:27:51 +00002435/* Run the preprocessor over the input file text. The global variables
drh6d08b4d2004-07-20 12:45:22 +00002436** azDefine[0] through azDefine[nDefine-1] contains the names of all defined
2437** macros. This routine looks for "%ifdef" and "%ifndef" and "%endif" and
2438** comments them out. Text in between is also commented out as appropriate.
2439*/
danielk1977940fac92005-01-23 22:41:37 +00002440static void preprocess_input(char *z){
drh6d08b4d2004-07-20 12:45:22 +00002441 int i, j, k, n;
2442 int exclude = 0;
rse38514a92007-09-20 11:34:17 +00002443 int start = 0;
drh6d08b4d2004-07-20 12:45:22 +00002444 int lineno = 1;
rse38514a92007-09-20 11:34:17 +00002445 int start_lineno = 1;
drh6d08b4d2004-07-20 12:45:22 +00002446 for(i=0; z[i]; i++){
2447 if( z[i]=='\n' ) lineno++;
2448 if( z[i]!='%' || (i>0 && z[i-1]!='\n') ) continue;
2449 if( strncmp(&z[i],"%endif",6)==0 && isspace(z[i+6]) ){
2450 if( exclude ){
2451 exclude--;
2452 if( exclude==0 ){
2453 for(j=start; j<i; j++) if( z[j]!='\n' ) z[j] = ' ';
2454 }
2455 }
2456 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2457 }else if( (strncmp(&z[i],"%ifdef",6)==0 && isspace(z[i+6]))
2458 || (strncmp(&z[i],"%ifndef",7)==0 && isspace(z[i+7])) ){
2459 if( exclude ){
2460 exclude++;
2461 }else{
2462 for(j=i+7; isspace(z[j]); j++){}
2463 for(n=0; z[j+n] && !isspace(z[j+n]); n++){}
2464 exclude = 1;
2465 for(k=0; k<nDefine; k++){
drh87cf1372008-08-13 20:09:06 +00002466 if( strncmp(azDefine[k],&z[j],n)==0 && lemonStrlen(azDefine[k])==n ){
drh6d08b4d2004-07-20 12:45:22 +00002467 exclude = 0;
2468 break;
2469 }
2470 }
2471 if( z[i+3]=='n' ) exclude = !exclude;
2472 if( exclude ){
2473 start = i;
2474 start_lineno = lineno;
2475 }
2476 }
2477 for(j=i; z[j] && z[j]!='\n'; j++) z[j] = ' ';
2478 }
2479 }
2480 if( exclude ){
2481 fprintf(stderr,"unterminated %%ifdef starting on line %d\n", start_lineno);
2482 exit(1);
2483 }
2484}
2485
drh75897232000-05-29 14:26:00 +00002486/* In spite of its name, this function is really a scanner. It read
2487** in the entire input file (all at once) then tokenizes it. Each
2488** token is passed to the function "parseonetoken" which builds all
2489** the appropriate data structures in the global state vector "gp".
2490*/
icculus9e44cf12010-02-14 17:14:22 +00002491void Parse(struct lemon *gp)
drh75897232000-05-29 14:26:00 +00002492{
2493 struct pstate ps;
2494 FILE *fp;
2495 char *filebuf;
2496 int filesize;
2497 int lineno;
2498 int c;
2499 char *cp, *nextcp;
2500 int startline = 0;
2501
rse38514a92007-09-20 11:34:17 +00002502 memset(&ps, '\0', sizeof(ps));
drh75897232000-05-29 14:26:00 +00002503 ps.gp = gp;
2504 ps.filename = gp->filename;
2505 ps.errorcnt = 0;
2506 ps.state = INITIALIZE;
2507
2508 /* Begin by reading the input file */
2509 fp = fopen(ps.filename,"rb");
2510 if( fp==0 ){
2511 ErrorMsg(ps.filename,0,"Can't open this file for reading.");
2512 gp->errorcnt++;
2513 return;
2514 }
2515 fseek(fp,0,2);
2516 filesize = ftell(fp);
2517 rewind(fp);
2518 filebuf = (char *)malloc( filesize+1 );
2519 if( filebuf==0 ){
2520 ErrorMsg(ps.filename,0,"Can't allocate %d of memory to hold this file.",
2521 filesize+1);
2522 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002523 fclose(fp);
drh75897232000-05-29 14:26:00 +00002524 return;
2525 }
2526 if( fread(filebuf,1,filesize,fp)!=filesize ){
2527 ErrorMsg(ps.filename,0,"Can't read in all %d bytes of this file.",
2528 filesize);
2529 free(filebuf);
2530 gp->errorcnt++;
drhe0a59cf2011-08-30 00:58:58 +00002531 fclose(fp);
drh75897232000-05-29 14:26:00 +00002532 return;
2533 }
2534 fclose(fp);
2535 filebuf[filesize] = 0;
2536
drh6d08b4d2004-07-20 12:45:22 +00002537 /* Make an initial pass through the file to handle %ifdef and %ifndef */
2538 preprocess_input(filebuf);
2539
drh75897232000-05-29 14:26:00 +00002540 /* Now scan the text of the input file */
2541 lineno = 1;
2542 for(cp=filebuf; (c= *cp)!=0; ){
2543 if( c=='\n' ) lineno++; /* Keep track of the line number */
2544 if( isspace(c) ){ cp++; continue; } /* Skip all white space */
2545 if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments */
2546 cp+=2;
2547 while( (c= *cp)!=0 && c!='\n' ) cp++;
2548 continue;
2549 }
2550 if( c=='/' && cp[1]=='*' ){ /* Skip C style comments */
2551 cp+=2;
2552 while( (c= *cp)!=0 && (c!='/' || cp[-1]!='*') ){
2553 if( c=='\n' ) lineno++;
2554 cp++;
2555 }
2556 if( c ) cp++;
2557 continue;
2558 }
2559 ps.tokenstart = cp; /* Mark the beginning of the token */
2560 ps.tokenlineno = lineno; /* Linenumber on which token begins */
2561 if( c=='\"' ){ /* String literals */
2562 cp++;
2563 while( (c= *cp)!=0 && c!='\"' ){
2564 if( c=='\n' ) lineno++;
2565 cp++;
2566 }
2567 if( c==0 ){
2568 ErrorMsg(ps.filename,startline,
2569"String starting on this line is not terminated before the end of the file.");
2570 ps.errorcnt++;
2571 nextcp = cp;
2572 }else{
2573 nextcp = cp+1;
2574 }
2575 }else if( c=='{' ){ /* A block of C code */
2576 int level;
2577 cp++;
2578 for(level=1; (c= *cp)!=0 && (level>1 || c!='}'); cp++){
2579 if( c=='\n' ) lineno++;
2580 else if( c=='{' ) level++;
2581 else if( c=='}' ) level--;
2582 else if( c=='/' && cp[1]=='*' ){ /* Skip comments */
2583 int prevc;
2584 cp = &cp[2];
2585 prevc = 0;
2586 while( (c= *cp)!=0 && (c!='/' || prevc!='*') ){
2587 if( c=='\n' ) lineno++;
2588 prevc = c;
2589 cp++;
drhf2f105d2012-08-20 15:53:54 +00002590 }
2591 }else if( c=='/' && cp[1]=='/' ){ /* Skip C++ style comments too */
drh75897232000-05-29 14:26:00 +00002592 cp = &cp[2];
2593 while( (c= *cp)!=0 && c!='\n' ) cp++;
2594 if( c ) lineno++;
drhf2f105d2012-08-20 15:53:54 +00002595 }else if( c=='\'' || c=='\"' ){ /* String a character literals */
drh75897232000-05-29 14:26:00 +00002596 int startchar, prevc;
2597 startchar = c;
2598 prevc = 0;
2599 for(cp++; (c= *cp)!=0 && (c!=startchar || prevc=='\\'); cp++){
2600 if( c=='\n' ) lineno++;
2601 if( prevc=='\\' ) prevc = 0;
2602 else prevc = c;
drhf2f105d2012-08-20 15:53:54 +00002603 }
2604 }
drh75897232000-05-29 14:26:00 +00002605 }
2606 if( c==0 ){
drh960e8c62001-04-03 16:53:21 +00002607 ErrorMsg(ps.filename,ps.tokenlineno,
drh75897232000-05-29 14:26:00 +00002608"C code starting on this line is not terminated before the end of the file.");
2609 ps.errorcnt++;
2610 nextcp = cp;
2611 }else{
2612 nextcp = cp+1;
2613 }
2614 }else if( isalnum(c) ){ /* Identifiers */
2615 while( (c= *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2616 nextcp = cp;
2617 }else if( c==':' && cp[1]==':' && cp[2]=='=' ){ /* The operator "::=" */
2618 cp += 3;
2619 nextcp = cp;
drhfd405312005-11-06 04:06:59 +00002620 }else if( (c=='/' || c=='|') && isalpha(cp[1]) ){
2621 cp += 2;
2622 while( (c = *cp)!=0 && (isalnum(c) || c=='_') ) cp++;
2623 nextcp = cp;
drh75897232000-05-29 14:26:00 +00002624 }else{ /* All other (one character) operators */
2625 cp++;
2626 nextcp = cp;
2627 }
2628 c = *cp;
2629 *cp = 0; /* Null terminate the token */
2630 parseonetoken(&ps); /* Parse the token */
2631 *cp = c; /* Restore the buffer */
2632 cp = nextcp;
2633 }
2634 free(filebuf); /* Release the buffer after parsing */
2635 gp->rule = ps.firstrule;
2636 gp->errorcnt = ps.errorcnt;
2637}
2638/*************************** From the file "plink.c" *********************/
2639/*
2640** Routines processing configuration follow-set propagation links
2641** in the LEMON parser generator.
2642*/
2643static struct plink *plink_freelist = 0;
2644
2645/* Allocate a new plink */
2646struct plink *Plink_new(){
icculus9e44cf12010-02-14 17:14:22 +00002647 struct plink *newlink;
drh75897232000-05-29 14:26:00 +00002648
2649 if( plink_freelist==0 ){
2650 int i;
2651 int amt = 100;
drh9892c5d2007-12-21 00:02:11 +00002652 plink_freelist = (struct plink *)calloc( amt, sizeof(struct plink) );
drh75897232000-05-29 14:26:00 +00002653 if( plink_freelist==0 ){
2654 fprintf(stderr,
2655 "Unable to allocate memory for a new follow-set propagation link.\n");
2656 exit(1);
2657 }
2658 for(i=0; i<amt-1; i++) plink_freelist[i].next = &plink_freelist[i+1];
2659 plink_freelist[amt-1].next = 0;
2660 }
icculus9e44cf12010-02-14 17:14:22 +00002661 newlink = plink_freelist;
drh75897232000-05-29 14:26:00 +00002662 plink_freelist = plink_freelist->next;
icculus9e44cf12010-02-14 17:14:22 +00002663 return newlink;
drh75897232000-05-29 14:26:00 +00002664}
2665
2666/* Add a plink to a plink list */
icculus9e44cf12010-02-14 17:14:22 +00002667void Plink_add(struct plink **plpp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002668{
icculus9e44cf12010-02-14 17:14:22 +00002669 struct plink *newlink;
2670 newlink = Plink_new();
2671 newlink->next = *plpp;
2672 *plpp = newlink;
2673 newlink->cfp = cfp;
drh75897232000-05-29 14:26:00 +00002674}
2675
2676/* Transfer every plink on the list "from" to the list "to" */
icculus9e44cf12010-02-14 17:14:22 +00002677void Plink_copy(struct plink **to, struct plink *from)
drh75897232000-05-29 14:26:00 +00002678{
2679 struct plink *nextpl;
2680 while( from ){
2681 nextpl = from->next;
2682 from->next = *to;
2683 *to = from;
2684 from = nextpl;
2685 }
2686}
2687
2688/* Delete every plink on the list */
icculus9e44cf12010-02-14 17:14:22 +00002689void Plink_delete(struct plink *plp)
drh75897232000-05-29 14:26:00 +00002690{
2691 struct plink *nextpl;
2692
2693 while( plp ){
2694 nextpl = plp->next;
2695 plp->next = plink_freelist;
2696 plink_freelist = plp;
2697 plp = nextpl;
2698 }
2699}
2700/*********************** From the file "report.c" **************************/
2701/*
2702** Procedures for generating reports and tables in the LEMON parser generator.
2703*/
2704
2705/* Generate a filename with the given suffix. Space to hold the
2706** name comes from malloc() and must be freed by the calling
2707** function.
2708*/
icculus9e44cf12010-02-14 17:14:22 +00002709PRIVATE char *file_makename(struct lemon *lemp, const char *suffix)
drh75897232000-05-29 14:26:00 +00002710{
2711 char *name;
2712 char *cp;
2713
icculus9e44cf12010-02-14 17:14:22 +00002714 name = (char*)malloc( lemonStrlen(lemp->filename) + lemonStrlen(suffix) + 5 );
drh75897232000-05-29 14:26:00 +00002715 if( name==0 ){
2716 fprintf(stderr,"Can't allocate space for a filename.\n");
2717 exit(1);
2718 }
2719 strcpy(name,lemp->filename);
2720 cp = strrchr(name,'.');
2721 if( cp ) *cp = 0;
2722 strcat(name,suffix);
2723 return name;
2724}
2725
2726/* Open a file with a name based on the name of the input file,
2727** but with a different (specified) suffix, and return a pointer
2728** to the stream */
icculus9e44cf12010-02-14 17:14:22 +00002729PRIVATE FILE *file_open(
2730 struct lemon *lemp,
2731 const char *suffix,
2732 const char *mode
2733){
drh75897232000-05-29 14:26:00 +00002734 FILE *fp;
2735
2736 if( lemp->outname ) free(lemp->outname);
2737 lemp->outname = file_makename(lemp, suffix);
2738 fp = fopen(lemp->outname,mode);
2739 if( fp==0 && *mode=='w' ){
2740 fprintf(stderr,"Can't open file \"%s\".\n",lemp->outname);
2741 lemp->errorcnt++;
2742 return 0;
2743 }
2744 return fp;
2745}
2746
2747/* Duplicate the input file without comments and without actions
2748** on rules */
icculus9e44cf12010-02-14 17:14:22 +00002749void Reprint(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002750{
2751 struct rule *rp;
2752 struct symbol *sp;
2753 int i, j, maxlen, len, ncolumns, skip;
2754 printf("// Reprint of input file \"%s\".\n// Symbols:\n",lemp->filename);
2755 maxlen = 10;
2756 for(i=0; i<lemp->nsymbol; i++){
2757 sp = lemp->symbols[i];
drh87cf1372008-08-13 20:09:06 +00002758 len = lemonStrlen(sp->name);
drh75897232000-05-29 14:26:00 +00002759 if( len>maxlen ) maxlen = len;
2760 }
2761 ncolumns = 76/(maxlen+5);
2762 if( ncolumns<1 ) ncolumns = 1;
2763 skip = (lemp->nsymbol + ncolumns - 1)/ncolumns;
2764 for(i=0; i<skip; i++){
2765 printf("//");
2766 for(j=i; j<lemp->nsymbol; j+=skip){
2767 sp = lemp->symbols[j];
2768 assert( sp->index==j );
2769 printf(" %3d %-*.*s",j,maxlen,maxlen,sp->name);
2770 }
2771 printf("\n");
2772 }
2773 for(rp=lemp->rule; rp; rp=rp->next){
2774 printf("%s",rp->lhs->name);
drhfd405312005-11-06 04:06:59 +00002775 /* if( rp->lhsalias ) printf("(%s)",rp->lhsalias); */
drh75897232000-05-29 14:26:00 +00002776 printf(" ::=");
2777 for(i=0; i<rp->nrhs; i++){
drhfd405312005-11-06 04:06:59 +00002778 sp = rp->rhs[i];
2779 printf(" %s", sp->name);
2780 if( sp->type==MULTITERMINAL ){
2781 for(j=1; j<sp->nsubsym; j++){
2782 printf("|%s", sp->subsym[j]->name);
2783 }
2784 }
2785 /* if( rp->rhsalias[i] ) printf("(%s)",rp->rhsalias[i]); */
drh75897232000-05-29 14:26:00 +00002786 }
2787 printf(".");
2788 if( rp->precsym ) printf(" [%s]",rp->precsym->name);
drhfd405312005-11-06 04:06:59 +00002789 /* if( rp->code ) printf("\n %s",rp->code); */
drh75897232000-05-29 14:26:00 +00002790 printf("\n");
2791 }
2792}
2793
icculus9e44cf12010-02-14 17:14:22 +00002794void ConfigPrint(FILE *fp, struct config *cfp)
drh75897232000-05-29 14:26:00 +00002795{
2796 struct rule *rp;
drhfd405312005-11-06 04:06:59 +00002797 struct symbol *sp;
2798 int i, j;
drh75897232000-05-29 14:26:00 +00002799 rp = cfp->rp;
2800 fprintf(fp,"%s ::=",rp->lhs->name);
2801 for(i=0; i<=rp->nrhs; i++){
2802 if( i==cfp->dot ) fprintf(fp," *");
2803 if( i==rp->nrhs ) break;
drhfd405312005-11-06 04:06:59 +00002804 sp = rp->rhs[i];
2805 fprintf(fp," %s", sp->name);
2806 if( sp->type==MULTITERMINAL ){
2807 for(j=1; j<sp->nsubsym; j++){
2808 fprintf(fp,"|%s",sp->subsym[j]->name);
2809 }
2810 }
drh75897232000-05-29 14:26:00 +00002811 }
2812}
2813
2814/* #define TEST */
drhfd405312005-11-06 04:06:59 +00002815#if 0
drh75897232000-05-29 14:26:00 +00002816/* Print a set */
2817PRIVATE void SetPrint(out,set,lemp)
2818FILE *out;
2819char *set;
2820struct lemon *lemp;
2821{
2822 int i;
2823 char *spacer;
2824 spacer = "";
2825 fprintf(out,"%12s[","");
2826 for(i=0; i<lemp->nterminal; i++){
2827 if( SetFind(set,i) ){
2828 fprintf(out,"%s%s",spacer,lemp->symbols[i]->name);
2829 spacer = " ";
2830 }
2831 }
2832 fprintf(out,"]\n");
2833}
2834
2835/* Print a plink chain */
2836PRIVATE void PlinkPrint(out,plp,tag)
2837FILE *out;
2838struct plink *plp;
2839char *tag;
2840{
2841 while( plp ){
drhada354d2005-11-05 15:03:59 +00002842 fprintf(out,"%12s%s (state %2d) ","",tag,plp->cfp->stp->statenum);
drh75897232000-05-29 14:26:00 +00002843 ConfigPrint(out,plp->cfp);
2844 fprintf(out,"\n");
2845 plp = plp->next;
2846 }
2847}
2848#endif
2849
2850/* Print an action to the given file descriptor. Return FALSE if
2851** nothing was actually printed.
2852*/
2853int PrintAction(struct action *ap, FILE *fp, int indent){
2854 int result = 1;
2855 switch( ap->type ){
2856 case SHIFT:
drhada354d2005-11-05 15:03:59 +00002857 fprintf(fp,"%*s shift %d",indent,ap->sp->name,ap->x.stp->statenum);
drh75897232000-05-29 14:26:00 +00002858 break;
2859 case REDUCE:
2860 fprintf(fp,"%*s reduce %d",indent,ap->sp->name,ap->x.rp->index);
2861 break;
2862 case ACCEPT:
2863 fprintf(fp,"%*s accept",indent,ap->sp->name);
2864 break;
2865 case ERROR:
2866 fprintf(fp,"%*s error",indent,ap->sp->name);
2867 break;
drh9892c5d2007-12-21 00:02:11 +00002868 case SRCONFLICT:
2869 case RRCONFLICT:
drh75897232000-05-29 14:26:00 +00002870 fprintf(fp,"%*s reduce %-3d ** Parsing conflict **",
2871 indent,ap->sp->name,ap->x.rp->index);
2872 break;
drh9892c5d2007-12-21 00:02:11 +00002873 case SSCONFLICT:
drhdd7e9db2010-07-19 01:52:07 +00002874 fprintf(fp,"%*s shift %-3d ** Parsing conflict **",
drh9892c5d2007-12-21 00:02:11 +00002875 indent,ap->sp->name,ap->x.stp->statenum);
2876 break;
drh75897232000-05-29 14:26:00 +00002877 case SH_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00002878 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00002879 fprintf(fp,"%*s shift %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00002880 indent,ap->sp->name,ap->x.stp->statenum);
2881 }else{
2882 result = 0;
2883 }
2884 break;
drh75897232000-05-29 14:26:00 +00002885 case RD_RESOLVED:
drhf5c4e0f2010-07-18 11:35:53 +00002886 if( showPrecedenceConflict ){
drhdd7e9db2010-07-19 01:52:07 +00002887 fprintf(fp,"%*s reduce %-3d -- dropped by precedence",
drhf5c4e0f2010-07-18 11:35:53 +00002888 indent,ap->sp->name,ap->x.rp->index);
2889 }else{
2890 result = 0;
2891 }
2892 break;
drh75897232000-05-29 14:26:00 +00002893 case NOT_USED:
2894 result = 0;
2895 break;
2896 }
2897 return result;
2898}
2899
2900/* Generate the "y.output" log file */
icculus9e44cf12010-02-14 17:14:22 +00002901void ReportOutput(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00002902{
2903 int i;
2904 struct state *stp;
2905 struct config *cfp;
2906 struct action *ap;
2907 FILE *fp;
2908
drh2aa6ca42004-09-10 00:14:04 +00002909 fp = file_open(lemp,".out","wb");
drh75897232000-05-29 14:26:00 +00002910 if( fp==0 ) return;
drh75897232000-05-29 14:26:00 +00002911 for(i=0; i<lemp->nstate; i++){
2912 stp = lemp->sorted[i];
drhada354d2005-11-05 15:03:59 +00002913 fprintf(fp,"State %d:\n",stp->statenum);
drh75897232000-05-29 14:26:00 +00002914 if( lemp->basisflag ) cfp=stp->bp;
2915 else cfp=stp->cfp;
2916 while( cfp ){
2917 char buf[20];
2918 if( cfp->dot==cfp->rp->nrhs ){
2919 sprintf(buf,"(%d)",cfp->rp->index);
2920 fprintf(fp," %5s ",buf);
2921 }else{
2922 fprintf(fp," ");
2923 }
2924 ConfigPrint(fp,cfp);
2925 fprintf(fp,"\n");
drhfd405312005-11-06 04:06:59 +00002926#if 0
drh75897232000-05-29 14:26:00 +00002927 SetPrint(fp,cfp->fws,lemp);
2928 PlinkPrint(fp,cfp->fplp,"To ");
2929 PlinkPrint(fp,cfp->bplp,"From");
2930#endif
2931 if( lemp->basisflag ) cfp=cfp->bp;
2932 else cfp=cfp->next;
2933 }
2934 fprintf(fp,"\n");
2935 for(ap=stp->ap; ap; ap=ap->next){
2936 if( PrintAction(ap,fp,30) ) fprintf(fp,"\n");
2937 }
2938 fprintf(fp,"\n");
2939 }
drhe9278182007-07-18 18:16:29 +00002940 fprintf(fp, "----------------------------------------------------\n");
2941 fprintf(fp, "Symbols:\n");
2942 for(i=0; i<lemp->nsymbol; i++){
2943 int j;
2944 struct symbol *sp;
2945
2946 sp = lemp->symbols[i];
2947 fprintf(fp, " %3d: %s", i, sp->name);
2948 if( sp->type==NONTERMINAL ){
2949 fprintf(fp, ":");
2950 if( sp->lambda ){
2951 fprintf(fp, " <lambda>");
2952 }
2953 for(j=0; j<lemp->nterminal; j++){
2954 if( sp->firstset && SetFind(sp->firstset, j) ){
2955 fprintf(fp, " %s", lemp->symbols[j]->name);
2956 }
2957 }
2958 }
2959 fprintf(fp, "\n");
2960 }
drh75897232000-05-29 14:26:00 +00002961 fclose(fp);
2962 return;
2963}
2964
2965/* Search for the file "name" which is in the same directory as
2966** the exacutable */
icculus9e44cf12010-02-14 17:14:22 +00002967PRIVATE char *pathsearch(char *argv0, char *name, int modemask)
drh75897232000-05-29 14:26:00 +00002968{
icculus9e44cf12010-02-14 17:14:22 +00002969 const char *pathlist;
2970 char *pathbufptr;
2971 char *pathbuf;
drh75897232000-05-29 14:26:00 +00002972 char *path,*cp;
2973 char c;
drh75897232000-05-29 14:26:00 +00002974
2975#ifdef __WIN32__
2976 cp = strrchr(argv0,'\\');
2977#else
2978 cp = strrchr(argv0,'/');
2979#endif
2980 if( cp ){
2981 c = *cp;
2982 *cp = 0;
drh87cf1372008-08-13 20:09:06 +00002983 path = (char *)malloc( lemonStrlen(argv0) + lemonStrlen(name) + 2 );
drh75897232000-05-29 14:26:00 +00002984 if( path ) sprintf(path,"%s/%s",argv0,name);
2985 *cp = c;
2986 }else{
drh75897232000-05-29 14:26:00 +00002987 pathlist = getenv("PATH");
2988 if( pathlist==0 ) pathlist = ".:/bin:/usr/bin";
icculus9e44cf12010-02-14 17:14:22 +00002989 pathbuf = (char *) malloc( lemonStrlen(pathlist) + 1 );
drh87cf1372008-08-13 20:09:06 +00002990 path = (char *)malloc( lemonStrlen(pathlist)+lemonStrlen(name)+2 );
icculus9e44cf12010-02-14 17:14:22 +00002991 if( (pathbuf != 0) && (path!=0) ){
2992 pathbufptr = pathbuf;
2993 strcpy(pathbuf, pathlist);
2994 while( *pathbuf ){
2995 cp = strchr(pathbuf,':');
2996 if( cp==0 ) cp = &pathbuf[lemonStrlen(pathbuf)];
drh75897232000-05-29 14:26:00 +00002997 c = *cp;
2998 *cp = 0;
icculus9e44cf12010-02-14 17:14:22 +00002999 sprintf(path,"%s/%s",pathbuf,name);
drh75897232000-05-29 14:26:00 +00003000 *cp = c;
icculus9e44cf12010-02-14 17:14:22 +00003001 if( c==0 ) pathbuf[0] = 0;
3002 else pathbuf = &cp[1];
drh75897232000-05-29 14:26:00 +00003003 if( access(path,modemask)==0 ) break;
3004 }
icculus9e44cf12010-02-14 17:14:22 +00003005 free(pathbufptr);
drh75897232000-05-29 14:26:00 +00003006 }
3007 }
3008 return path;
3009}
3010
3011/* Given an action, compute the integer value for that action
3012** which is to be put in the action table of the generated machine.
3013** Return negative if no action should be generated.
3014*/
icculus9e44cf12010-02-14 17:14:22 +00003015PRIVATE int compute_action(struct lemon *lemp, struct action *ap)
drh75897232000-05-29 14:26:00 +00003016{
3017 int act;
3018 switch( ap->type ){
drhada354d2005-11-05 15:03:59 +00003019 case SHIFT: act = ap->x.stp->statenum; break;
drh75897232000-05-29 14:26:00 +00003020 case REDUCE: act = ap->x.rp->index + lemp->nstate; break;
3021 case ERROR: act = lemp->nstate + lemp->nrule; break;
3022 case ACCEPT: act = lemp->nstate + lemp->nrule + 1; break;
3023 default: act = -1; break;
3024 }
3025 return act;
3026}
3027
3028#define LINESIZE 1000
3029/* The next cluster of routines are for reading the template file
3030** and writing the results to the generated parser */
3031/* The first function transfers data from "in" to "out" until
3032** a line is seen which begins with "%%". The line number is
3033** tracked.
3034**
3035** if name!=0, then any word that begin with "Parse" is changed to
3036** begin with *name instead.
3037*/
icculus9e44cf12010-02-14 17:14:22 +00003038PRIVATE void tplt_xfer(char *name, FILE *in, FILE *out, int *lineno)
drh75897232000-05-29 14:26:00 +00003039{
3040 int i, iStart;
3041 char line[LINESIZE];
3042 while( fgets(line,LINESIZE,in) && (line[0]!='%' || line[1]!='%') ){
3043 (*lineno)++;
3044 iStart = 0;
3045 if( name ){
3046 for(i=0; line[i]; i++){
3047 if( line[i]=='P' && strncmp(&line[i],"Parse",5)==0
3048 && (i==0 || !isalpha(line[i-1]))
3049 ){
3050 if( i>iStart ) fprintf(out,"%.*s",i-iStart,&line[iStart]);
3051 fprintf(out,"%s",name);
3052 i += 4;
3053 iStart = i+1;
3054 }
3055 }
3056 }
3057 fprintf(out,"%s",&line[iStart]);
3058 }
3059}
3060
3061/* The next function finds the template file and opens it, returning
3062** a pointer to the opened file. */
icculus9e44cf12010-02-14 17:14:22 +00003063PRIVATE FILE *tplt_open(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003064{
3065 static char templatename[] = "lempar.c";
3066 char buf[1000];
3067 FILE *in;
3068 char *tpltname;
3069 char *cp;
3070
icculus3e143bd2010-02-14 00:48:49 +00003071 /* first, see if user specified a template filename on the command line. */
3072 if (user_templatename != 0) {
3073 if( access(user_templatename,004)==-1 ){
3074 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3075 user_templatename);
3076 lemp->errorcnt++;
3077 return 0;
3078 }
3079 in = fopen(user_templatename,"rb");
3080 if( in==0 ){
3081 fprintf(stderr,"Can't open the template file \"%s\".\n",user_templatename);
3082 lemp->errorcnt++;
3083 return 0;
3084 }
3085 return in;
3086 }
3087
drh75897232000-05-29 14:26:00 +00003088 cp = strrchr(lemp->filename,'.');
3089 if( cp ){
drh8b582012003-10-21 13:16:03 +00003090 sprintf(buf,"%.*s.lt",(int)(cp-lemp->filename),lemp->filename);
drh75897232000-05-29 14:26:00 +00003091 }else{
3092 sprintf(buf,"%s.lt",lemp->filename);
3093 }
3094 if( access(buf,004)==0 ){
3095 tpltname = buf;
drh960e8c62001-04-03 16:53:21 +00003096 }else if( access(templatename,004)==0 ){
3097 tpltname = templatename;
drh75897232000-05-29 14:26:00 +00003098 }else{
3099 tpltname = pathsearch(lemp->argv0,templatename,0);
3100 }
3101 if( tpltname==0 ){
3102 fprintf(stderr,"Can't find the parser driver template file \"%s\".\n",
3103 templatename);
3104 lemp->errorcnt++;
3105 return 0;
3106 }
drh2aa6ca42004-09-10 00:14:04 +00003107 in = fopen(tpltname,"rb");
drh75897232000-05-29 14:26:00 +00003108 if( in==0 ){
3109 fprintf(stderr,"Can't open the template file \"%s\".\n",templatename);
3110 lemp->errorcnt++;
3111 return 0;
3112 }
3113 return in;
3114}
3115
drhaf805ca2004-09-07 11:28:25 +00003116/* Print a #line directive line to the output file. */
icculus9e44cf12010-02-14 17:14:22 +00003117PRIVATE void tplt_linedir(FILE *out, int lineno, char *filename)
drhaf805ca2004-09-07 11:28:25 +00003118{
3119 fprintf(out,"#line %d \"",lineno);
3120 while( *filename ){
3121 if( *filename == '\\' ) putc('\\',out);
3122 putc(*filename,out);
3123 filename++;
3124 }
3125 fprintf(out,"\"\n");
3126}
3127
drh75897232000-05-29 14:26:00 +00003128/* Print a string to the file and keep the linenumber up to date */
icculus9e44cf12010-02-14 17:14:22 +00003129PRIVATE void tplt_print(FILE *out, struct lemon *lemp, char *str, int *lineno)
drh75897232000-05-29 14:26:00 +00003130{
3131 if( str==0 ) return;
drh75897232000-05-29 14:26:00 +00003132 while( *str ){
drh75897232000-05-29 14:26:00 +00003133 putc(*str,out);
shane58543932008-12-10 20:10:04 +00003134 if( *str=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003135 str++;
3136 }
drh9db55df2004-09-09 14:01:21 +00003137 if( str[-1]!='\n' ){
3138 putc('\n',out);
3139 (*lineno)++;
3140 }
shane58543932008-12-10 20:10:04 +00003141 if (!lemp->nolinenosflag) {
3142 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3143 }
drh75897232000-05-29 14:26:00 +00003144 return;
3145}
3146
3147/*
3148** The following routine emits code for the destructor for the
3149** symbol sp
3150*/
icculus9e44cf12010-02-14 17:14:22 +00003151void emit_destructor_code(
3152 FILE *out,
3153 struct symbol *sp,
3154 struct lemon *lemp,
3155 int *lineno
3156){
drhcc83b6e2004-04-23 23:38:42 +00003157 char *cp = 0;
drh75897232000-05-29 14:26:00 +00003158
drh75897232000-05-29 14:26:00 +00003159 if( sp->type==TERMINAL ){
3160 cp = lemp->tokendest;
3161 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003162 fprintf(out,"{\n"); (*lineno)++;
drh960e8c62001-04-03 16:53:21 +00003163 }else if( sp->destructor ){
drh75897232000-05-29 14:26:00 +00003164 cp = sp->destructor;
drha5808f32008-04-27 22:19:44 +00003165 fprintf(out,"{\n"); (*lineno)++;
shane58543932008-12-10 20:10:04 +00003166 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,sp->destLineno,lemp->filename); }
drh960e8c62001-04-03 16:53:21 +00003167 }else if( lemp->vardest ){
3168 cp = lemp->vardest;
3169 if( cp==0 ) return;
drha5808f32008-04-27 22:19:44 +00003170 fprintf(out,"{\n"); (*lineno)++;
drhcc83b6e2004-04-23 23:38:42 +00003171 }else{
3172 assert( 0 ); /* Cannot happen */
drh75897232000-05-29 14:26:00 +00003173 }
3174 for(; *cp; cp++){
3175 if( *cp=='$' && cp[1]=='$' ){
3176 fprintf(out,"(yypminor->yy%d)",sp->dtnum);
3177 cp++;
3178 continue;
3179 }
shane58543932008-12-10 20:10:04 +00003180 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003181 fputc(*cp,out);
3182 }
shane58543932008-12-10 20:10:04 +00003183 fprintf(out,"\n"); (*lineno)++;
3184 if (!lemp->nolinenosflag) {
3185 (*lineno)++; tplt_linedir(out,*lineno,lemp->outname);
3186 }
3187 fprintf(out,"}\n"); (*lineno)++;
drh75897232000-05-29 14:26:00 +00003188 return;
3189}
3190
3191/*
drh960e8c62001-04-03 16:53:21 +00003192** Return TRUE (non-zero) if the given symbol has a destructor.
drh75897232000-05-29 14:26:00 +00003193*/
icculus9e44cf12010-02-14 17:14:22 +00003194int has_destructor(struct symbol *sp, struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00003195{
3196 int ret;
3197 if( sp->type==TERMINAL ){
3198 ret = lemp->tokendest!=0;
3199 }else{
drh960e8c62001-04-03 16:53:21 +00003200 ret = lemp->vardest!=0 || sp->destructor!=0;
drh75897232000-05-29 14:26:00 +00003201 }
3202 return ret;
3203}
3204
drh0bb132b2004-07-20 14:06:51 +00003205/*
3206** Append text to a dynamically allocated string. If zText is 0 then
3207** reset the string to be empty again. Always return the complete text
3208** of the string (which is overwritten with each call).
drh7ac25c72004-08-19 15:12:26 +00003209**
3210** n bytes of zText are stored. If n==0 then all of zText up to the first
3211** \000 terminator is stored. zText can contain up to two instances of
3212** %d. The values of p1 and p2 are written into the first and second
3213** %d.
3214**
3215** If n==-1, then the previous character is overwritten.
drh0bb132b2004-07-20 14:06:51 +00003216*/
icculus9e44cf12010-02-14 17:14:22 +00003217PRIVATE char *append_str(const char *zText, int n, int p1, int p2){
3218 static char empty[1] = { 0 };
drh0bb132b2004-07-20 14:06:51 +00003219 static char *z = 0;
3220 static int alloced = 0;
3221 static int used = 0;
drhaf805ca2004-09-07 11:28:25 +00003222 int c;
drh0bb132b2004-07-20 14:06:51 +00003223 char zInt[40];
drh0bb132b2004-07-20 14:06:51 +00003224 if( zText==0 ){
3225 used = 0;
3226 return z;
3227 }
drh7ac25c72004-08-19 15:12:26 +00003228 if( n<=0 ){
3229 if( n<0 ){
3230 used += n;
3231 assert( used>=0 );
3232 }
drh87cf1372008-08-13 20:09:06 +00003233 n = lemonStrlen(zText);
drh7ac25c72004-08-19 15:12:26 +00003234 }
drhdf609712010-11-23 20:55:27 +00003235 if( (int) (n+sizeof(zInt)*2+used) >= alloced ){
drh0bb132b2004-07-20 14:06:51 +00003236 alloced = n + sizeof(zInt)*2 + used + 200;
icculus9e44cf12010-02-14 17:14:22 +00003237 z = (char *) realloc(z, alloced);
drh0bb132b2004-07-20 14:06:51 +00003238 }
icculus9e44cf12010-02-14 17:14:22 +00003239 if( z==0 ) return empty;
drh0bb132b2004-07-20 14:06:51 +00003240 while( n-- > 0 ){
3241 c = *(zText++);
drh50489622006-10-13 12:25:29 +00003242 if( c=='%' && n>0 && zText[0]=='d' ){
drh0bb132b2004-07-20 14:06:51 +00003243 sprintf(zInt, "%d", p1);
3244 p1 = p2;
3245 strcpy(&z[used], zInt);
drh87cf1372008-08-13 20:09:06 +00003246 used += lemonStrlen(&z[used]);
drh0bb132b2004-07-20 14:06:51 +00003247 zText++;
3248 n--;
3249 }else{
3250 z[used++] = c;
3251 }
3252 }
3253 z[used] = 0;
3254 return z;
3255}
3256
3257/*
3258** zCode is a string that is the action associated with a rule. Expand
3259** the symbols in this string so that the refer to elements of the parser
drhaf805ca2004-09-07 11:28:25 +00003260** stack.
drh0bb132b2004-07-20 14:06:51 +00003261*/
drhaf805ca2004-09-07 11:28:25 +00003262PRIVATE void translate_code(struct lemon *lemp, struct rule *rp){
drh0bb132b2004-07-20 14:06:51 +00003263 char *cp, *xp;
3264 int i;
3265 char lhsused = 0; /* True if the LHS element has been used */
3266 char used[MAXRHS]; /* True for each RHS element which is used */
3267
3268 for(i=0; i<rp->nrhs; i++) used[i] = 0;
3269 lhsused = 0;
3270
drh19c9e562007-03-29 20:13:53 +00003271 if( rp->code==0 ){
icculus9e44cf12010-02-14 17:14:22 +00003272 static char newlinestr[2] = { '\n', '\0' };
3273 rp->code = newlinestr;
drh19c9e562007-03-29 20:13:53 +00003274 rp->line = rp->ruleline;
3275 }
3276
drh0bb132b2004-07-20 14:06:51 +00003277 append_str(0,0,0,0);
icculus9e44cf12010-02-14 17:14:22 +00003278
3279 /* This const cast is wrong but harmless, if we're careful. */
3280 for(cp=(char *)rp->code; *cp; cp++){
drh0bb132b2004-07-20 14:06:51 +00003281 if( isalpha(*cp) && (cp==rp->code || (!isalnum(cp[-1]) && cp[-1]!='_')) ){
3282 char saved;
3283 for(xp= &cp[1]; isalnum(*xp) || *xp=='_'; xp++);
3284 saved = *xp;
3285 *xp = 0;
3286 if( rp->lhsalias && strcmp(cp,rp->lhsalias)==0 ){
drh7ac25c72004-08-19 15:12:26 +00003287 append_str("yygotominor.yy%d",0,rp->lhs->dtnum,0);
drh0bb132b2004-07-20 14:06:51 +00003288 cp = xp;
3289 lhsused = 1;
3290 }else{
3291 for(i=0; i<rp->nrhs; i++){
3292 if( rp->rhsalias[i] && strcmp(cp,rp->rhsalias[i])==0 ){
drh7ac25c72004-08-19 15:12:26 +00003293 if( cp!=rp->code && cp[-1]=='@' ){
3294 /* If the argument is of the form @X then substituted
3295 ** the token number of X, not the value of X */
3296 append_str("yymsp[%d].major",-1,i-rp->nrhs+1,0);
3297 }else{
drhfd405312005-11-06 04:06:59 +00003298 struct symbol *sp = rp->rhs[i];
3299 int dtnum;
3300 if( sp->type==MULTITERMINAL ){
3301 dtnum = sp->subsym[0]->dtnum;
3302 }else{
3303 dtnum = sp->dtnum;
3304 }
3305 append_str("yymsp[%d].minor.yy%d",0,i-rp->nrhs+1, dtnum);
drh7ac25c72004-08-19 15:12:26 +00003306 }
drh0bb132b2004-07-20 14:06:51 +00003307 cp = xp;
3308 used[i] = 1;
3309 break;
3310 }
3311 }
3312 }
3313 *xp = saved;
3314 }
3315 append_str(cp, 1, 0, 0);
3316 } /* End loop */
3317
3318 /* Check to make sure the LHS has been used */
3319 if( rp->lhsalias && !lhsused ){
3320 ErrorMsg(lemp->filename,rp->ruleline,
3321 "Label \"%s\" for \"%s(%s)\" is never used.",
3322 rp->lhsalias,rp->lhs->name,rp->lhsalias);
3323 lemp->errorcnt++;
3324 }
3325
3326 /* Generate destructor code for RHS symbols which are not used in the
3327 ** reduce code */
3328 for(i=0; i<rp->nrhs; i++){
3329 if( rp->rhsalias[i] && !used[i] ){
3330 ErrorMsg(lemp->filename,rp->ruleline,
3331 "Label %s for \"%s(%s)\" is never used.",
3332 rp->rhsalias[i],rp->rhs[i]->name,rp->rhsalias[i]);
3333 lemp->errorcnt++;
3334 }else if( rp->rhsalias[i]==0 ){
3335 if( has_destructor(rp->rhs[i],lemp) ){
drh639efd02008-08-13 20:04:29 +00003336 append_str(" yy_destructor(yypParser,%d,&yymsp[%d].minor);\n", 0,
drh0bb132b2004-07-20 14:06:51 +00003337 rp->rhs[i]->index,i-rp->nrhs+1);
3338 }else{
3339 /* No destructor defined for this term */
3340 }
3341 }
3342 }
drh61e339a2007-01-16 03:09:02 +00003343 if( rp->code ){
3344 cp = append_str(0,0,0,0);
3345 rp->code = Strsafe(cp?cp:"");
3346 }
drh0bb132b2004-07-20 14:06:51 +00003347}
3348
drh75897232000-05-29 14:26:00 +00003349/*
3350** Generate code which executes when the rule "rp" is reduced. Write
3351** the code to "out". Make sure lineno stays up-to-date.
3352*/
icculus9e44cf12010-02-14 17:14:22 +00003353PRIVATE void emit_code(
3354 FILE *out,
3355 struct rule *rp,
3356 struct lemon *lemp,
3357 int *lineno
3358){
3359 const char *cp;
drh75897232000-05-29 14:26:00 +00003360
3361 /* Generate code to do the reduce action */
3362 if( rp->code ){
shane58543932008-12-10 20:10:04 +00003363 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,rp->line,lemp->filename); }
drhaf805ca2004-09-07 11:28:25 +00003364 fprintf(out,"{%s",rp->code);
drh75897232000-05-29 14:26:00 +00003365 for(cp=rp->code; *cp; cp++){
shane58543932008-12-10 20:10:04 +00003366 if( *cp=='\n' ) (*lineno)++;
drh75897232000-05-29 14:26:00 +00003367 } /* End loop */
shane58543932008-12-10 20:10:04 +00003368 fprintf(out,"}\n"); (*lineno)++;
3369 if (!lemp->nolinenosflag) { (*lineno)++; tplt_linedir(out,*lineno,lemp->outname); }
drh75897232000-05-29 14:26:00 +00003370 } /* End if( rp->code ) */
3371
drh75897232000-05-29 14:26:00 +00003372 return;
3373}
3374
3375/*
3376** Print the definition of the union used for the parser's data stack.
3377** This union contains fields for every possible data type for tokens
3378** and nonterminals. In the process of computing and printing this
3379** union, also set the ".dtnum" field of every terminal and nonterminal
3380** symbol.
3381*/
icculus9e44cf12010-02-14 17:14:22 +00003382void print_stack_union(
3383 FILE *out, /* The output stream */
3384 struct lemon *lemp, /* The main info structure for this parser */
3385 int *plineno, /* Pointer to the line number */
3386 int mhflag /* True if generating makeheaders output */
3387){
drh75897232000-05-29 14:26:00 +00003388 int lineno = *plineno; /* The line number of the output */
3389 char **types; /* A hash table of datatypes */
3390 int arraysize; /* Size of the "types" array */
3391 int maxdtlength; /* Maximum length of any ".datatype" field. */
3392 char *stddt; /* Standardized name for a datatype */
3393 int i,j; /* Loop counters */
drh01f75f22013-10-02 20:46:30 +00003394 unsigned hash; /* For hashing the name of a type */
icculus9e44cf12010-02-14 17:14:22 +00003395 const char *name; /* Name of the parser */
drh75897232000-05-29 14:26:00 +00003396
3397 /* Allocate and initialize types[] and allocate stddt[] */
3398 arraysize = lemp->nsymbol * 2;
drh9892c5d2007-12-21 00:02:11 +00003399 types = (char**)calloc( arraysize, sizeof(char*) );
drh070d4222011-06-02 15:48:51 +00003400 if( types==0 ){
3401 fprintf(stderr,"Out of memory.\n");
3402 exit(1);
3403 }
drh75897232000-05-29 14:26:00 +00003404 for(i=0; i<arraysize; i++) types[i] = 0;
3405 maxdtlength = 0;
drh960e8c62001-04-03 16:53:21 +00003406 if( lemp->vartype ){
drh87cf1372008-08-13 20:09:06 +00003407 maxdtlength = lemonStrlen(lemp->vartype);
drh960e8c62001-04-03 16:53:21 +00003408 }
drh75897232000-05-29 14:26:00 +00003409 for(i=0; i<lemp->nsymbol; i++){
3410 int len;
3411 struct symbol *sp = lemp->symbols[i];
3412 if( sp->datatype==0 ) continue;
drh87cf1372008-08-13 20:09:06 +00003413 len = lemonStrlen(sp->datatype);
drh75897232000-05-29 14:26:00 +00003414 if( len>maxdtlength ) maxdtlength = len;
3415 }
3416 stddt = (char*)malloc( maxdtlength*2 + 1 );
drh070d4222011-06-02 15:48:51 +00003417 if( stddt==0 ){
drh75897232000-05-29 14:26:00 +00003418 fprintf(stderr,"Out of memory.\n");
3419 exit(1);
3420 }
3421
3422 /* Build a hash table of datatypes. The ".dtnum" field of each symbol
3423 ** is filled in with the hash index plus 1. A ".dtnum" value of 0 is
drh960e8c62001-04-03 16:53:21 +00003424 ** used for terminal symbols. If there is no %default_type defined then
3425 ** 0 is also used as the .dtnum value for nonterminals which do not specify
3426 ** a datatype using the %type directive.
3427 */
drh75897232000-05-29 14:26:00 +00003428 for(i=0; i<lemp->nsymbol; i++){
3429 struct symbol *sp = lemp->symbols[i];
3430 char *cp;
3431 if( sp==lemp->errsym ){
3432 sp->dtnum = arraysize+1;
3433 continue;
3434 }
drh960e8c62001-04-03 16:53:21 +00003435 if( sp->type!=NONTERMINAL || (sp->datatype==0 && lemp->vartype==0) ){
drh75897232000-05-29 14:26:00 +00003436 sp->dtnum = 0;
3437 continue;
3438 }
3439 cp = sp->datatype;
drh960e8c62001-04-03 16:53:21 +00003440 if( cp==0 ) cp = lemp->vartype;
drh75897232000-05-29 14:26:00 +00003441 j = 0;
3442 while( isspace(*cp) ) cp++;
3443 while( *cp ) stddt[j++] = *cp++;
3444 while( j>0 && isspace(stddt[j-1]) ) j--;
3445 stddt[j] = 0;
drh02368c92009-04-05 15:18:02 +00003446 if( lemp->tokentype && strcmp(stddt, lemp->tokentype)==0 ){
drh32c4d742008-07-01 16:34:49 +00003447 sp->dtnum = 0;
3448 continue;
3449 }
drh75897232000-05-29 14:26:00 +00003450 hash = 0;
3451 for(j=0; stddt[j]; j++){
3452 hash = hash*53 + stddt[j];
3453 }
drh3b2129c2003-05-13 00:34:21 +00003454 hash = (hash & 0x7fffffff)%arraysize;
drh75897232000-05-29 14:26:00 +00003455 while( types[hash] ){
3456 if( strcmp(types[hash],stddt)==0 ){
3457 sp->dtnum = hash + 1;
3458 break;
3459 }
3460 hash++;
drh2b51f212013-10-11 23:01:02 +00003461 if( hash>=(unsigned)arraysize ) hash = 0;
drh75897232000-05-29 14:26:00 +00003462 }
3463 if( types[hash]==0 ){
3464 sp->dtnum = hash + 1;
drh87cf1372008-08-13 20:09:06 +00003465 types[hash] = (char*)malloc( lemonStrlen(stddt)+1 );
drh75897232000-05-29 14:26:00 +00003466 if( types[hash]==0 ){
3467 fprintf(stderr,"Out of memory.\n");
3468 exit(1);
3469 }
3470 strcpy(types[hash],stddt);
3471 }
3472 }
3473
3474 /* Print out the definition of YYTOKENTYPE and YYMINORTYPE */
3475 name = lemp->name ? lemp->name : "Parse";
3476 lineno = *plineno;
3477 if( mhflag ){ fprintf(out,"#if INTERFACE\n"); lineno++; }
3478 fprintf(out,"#define %sTOKENTYPE %s\n",name,
3479 lemp->tokentype?lemp->tokentype:"void*"); lineno++;
3480 if( mhflag ){ fprintf(out,"#endif\n"); lineno++; }
3481 fprintf(out,"typedef union {\n"); lineno++;
drh15b024c2008-12-11 02:20:43 +00003482 fprintf(out," int yyinit;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003483 fprintf(out," %sTOKENTYPE yy0;\n",name); lineno++;
3484 for(i=0; i<arraysize; i++){
3485 if( types[i]==0 ) continue;
3486 fprintf(out," %s yy%d;\n",types[i],i+1); lineno++;
3487 free(types[i]);
3488 }
drhc4dd3fd2008-01-22 01:48:05 +00003489 if( lemp->errsym->useCnt ){
3490 fprintf(out," int yy%d;\n",lemp->errsym->dtnum); lineno++;
3491 }
drh75897232000-05-29 14:26:00 +00003492 free(stddt);
3493 free(types);
3494 fprintf(out,"} YYMINORTYPE;\n"); lineno++;
3495 *plineno = lineno;
3496}
3497
drhb29b0a52002-02-23 19:39:46 +00003498/*
3499** Return the name of a C datatype able to represent values between
drh8b582012003-10-21 13:16:03 +00003500** lwr and upr, inclusive.
drhb29b0a52002-02-23 19:39:46 +00003501*/
drh8b582012003-10-21 13:16:03 +00003502static const char *minimum_size_type(int lwr, int upr){
3503 if( lwr>=0 ){
3504 if( upr<=255 ){
3505 return "unsigned char";
3506 }else if( upr<65535 ){
3507 return "unsigned short int";
3508 }else{
3509 return "unsigned int";
3510 }
3511 }else if( lwr>=-127 && upr<=127 ){
3512 return "signed char";
3513 }else if( lwr>=-32767 && upr<32767 ){
3514 return "short";
drhb29b0a52002-02-23 19:39:46 +00003515 }else{
drh8b582012003-10-21 13:16:03 +00003516 return "int";
drhb29b0a52002-02-23 19:39:46 +00003517 }
3518}
3519
drhfdbf9282003-10-21 16:34:41 +00003520/*
3521** Each state contains a set of token transaction and a set of
3522** nonterminal transactions. Each of these sets makes an instance
3523** of the following structure. An array of these structures is used
3524** to order the creation of entries in the yy_action[] table.
3525*/
3526struct axset {
3527 struct state *stp; /* A pointer to a state */
3528 int isTkn; /* True to use tokens. False for non-terminals */
3529 int nAction; /* Number of actions */
drhe594bc32009-11-03 13:02:25 +00003530 int iOrder; /* Original order of action sets */
drhfdbf9282003-10-21 16:34:41 +00003531};
3532
3533/*
3534** Compare to axset structures for sorting purposes
3535*/
3536static int axset_compare(const void *a, const void *b){
3537 struct axset *p1 = (struct axset*)a;
3538 struct axset *p2 = (struct axset*)b;
drhe594bc32009-11-03 13:02:25 +00003539 int c;
3540 c = p2->nAction - p1->nAction;
3541 if( c==0 ){
3542 c = p2->iOrder - p1->iOrder;
3543 }
3544 assert( c!=0 || p1==p2 );
3545 return c;
drhfdbf9282003-10-21 16:34:41 +00003546}
3547
drhc4dd3fd2008-01-22 01:48:05 +00003548/*
3549** Write text on "out" that describes the rule "rp".
3550*/
3551static void writeRuleText(FILE *out, struct rule *rp){
3552 int j;
3553 fprintf(out,"%s ::=", rp->lhs->name);
3554 for(j=0; j<rp->nrhs; j++){
3555 struct symbol *sp = rp->rhs[j];
3556 fprintf(out," %s", sp->name);
3557 if( sp->type==MULTITERMINAL ){
3558 int k;
3559 for(k=1; k<sp->nsubsym; k++){
3560 fprintf(out,"|%s",sp->subsym[k]->name);
3561 }
3562 }
3563 }
3564}
3565
3566
drh75897232000-05-29 14:26:00 +00003567/* Generate C source code for the parser */
icculus9e44cf12010-02-14 17:14:22 +00003568void ReportTable(
3569 struct lemon *lemp,
3570 int mhflag /* Output in makeheaders format if true */
3571){
drh75897232000-05-29 14:26:00 +00003572 FILE *out, *in;
3573 char line[LINESIZE];
3574 int lineno;
3575 struct state *stp;
3576 struct action *ap;
3577 struct rule *rp;
drh8b582012003-10-21 13:16:03 +00003578 struct acttab *pActtab;
icculusa3191192010-02-17 05:40:34 +00003579 int i, j, n;
icculus9e44cf12010-02-14 17:14:22 +00003580 const char *name;
drh8b582012003-10-21 13:16:03 +00003581 int mnTknOfst, mxTknOfst;
3582 int mnNtOfst, mxNtOfst;
drhfdbf9282003-10-21 16:34:41 +00003583 struct axset *ax;
drh75897232000-05-29 14:26:00 +00003584
3585 in = tplt_open(lemp);
3586 if( in==0 ) return;
drh2aa6ca42004-09-10 00:14:04 +00003587 out = file_open(lemp,".c","wb");
drh75897232000-05-29 14:26:00 +00003588 if( out==0 ){
3589 fclose(in);
3590 return;
3591 }
3592 lineno = 1;
3593 tplt_xfer(lemp->name,in,out,&lineno);
3594
3595 /* Generate the include code, if any */
drha5808f32008-04-27 22:19:44 +00003596 tplt_print(out,lemp,lemp->include,&lineno);
drh75897232000-05-29 14:26:00 +00003597 if( mhflag ){
3598 char *name = file_makename(lemp, ".h");
3599 fprintf(out,"#include \"%s\"\n", name); lineno++;
3600 free(name);
3601 }
3602 tplt_xfer(lemp->name,in,out,&lineno);
3603
3604 /* Generate #defines for all tokens */
3605 if( mhflag ){
icculus9e44cf12010-02-14 17:14:22 +00003606 const char *prefix;
drh75897232000-05-29 14:26:00 +00003607 fprintf(out,"#if INTERFACE\n"); lineno++;
3608 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
3609 else prefix = "";
3610 for(i=1; i<lemp->nterminal; i++){
3611 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
3612 lineno++;
3613 }
3614 fprintf(out,"#endif\n"); lineno++;
3615 }
3616 tplt_xfer(lemp->name,in,out,&lineno);
3617
3618 /* Generate the defines */
drh75897232000-05-29 14:26:00 +00003619 fprintf(out,"#define YYCODETYPE %s\n",
drh8a415d32009-06-12 13:53:51 +00003620 minimum_size_type(0, lemp->nsymbol+1)); lineno++;
drh75897232000-05-29 14:26:00 +00003621 fprintf(out,"#define YYNOCODE %d\n",lemp->nsymbol+1); lineno++;
3622 fprintf(out,"#define YYACTIONTYPE %s\n",
drh8b582012003-10-21 13:16:03 +00003623 minimum_size_type(0, lemp->nstate+lemp->nrule+5)); lineno++;
drhe09daa92006-06-10 13:29:31 +00003624 if( lemp->wildcard ){
3625 fprintf(out,"#define YYWILDCARD %d\n",
3626 lemp->wildcard->index); lineno++;
3627 }
drh75897232000-05-29 14:26:00 +00003628 print_stack_union(out,lemp,&lineno,mhflag);
drhca44b5a2007-02-22 23:06:58 +00003629 fprintf(out, "#ifndef YYSTACKDEPTH\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003630 if( lemp->stacksize ){
drh75897232000-05-29 14:26:00 +00003631 fprintf(out,"#define YYSTACKDEPTH %s\n",lemp->stacksize); lineno++;
3632 }else{
3633 fprintf(out,"#define YYSTACKDEPTH 100\n"); lineno++;
3634 }
drhca44b5a2007-02-22 23:06:58 +00003635 fprintf(out, "#endif\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003636 if( mhflag ){
3637 fprintf(out,"#if INTERFACE\n"); lineno++;
3638 }
3639 name = lemp->name ? lemp->name : "Parse";
3640 if( lemp->arg && lemp->arg[0] ){
3641 int i;
drh87cf1372008-08-13 20:09:06 +00003642 i = lemonStrlen(lemp->arg);
drhb1edd012000-06-02 18:52:12 +00003643 while( i>=1 && isspace(lemp->arg[i-1]) ) i--;
3644 while( i>=1 && (isalnum(lemp->arg[i-1]) || lemp->arg[i-1]=='_') ) i--;
drh1f245e42002-03-11 13:55:50 +00003645 fprintf(out,"#define %sARG_SDECL %s;\n",name,lemp->arg); lineno++;
3646 fprintf(out,"#define %sARG_PDECL ,%s\n",name,lemp->arg); lineno++;
3647 fprintf(out,"#define %sARG_FETCH %s = yypParser->%s\n",
3648 name,lemp->arg,&lemp->arg[i]); lineno++;
3649 fprintf(out,"#define %sARG_STORE yypParser->%s = %s\n",
3650 name,&lemp->arg[i],&lemp->arg[i]); lineno++;
drh75897232000-05-29 14:26:00 +00003651 }else{
drh1f245e42002-03-11 13:55:50 +00003652 fprintf(out,"#define %sARG_SDECL\n",name); lineno++;
3653 fprintf(out,"#define %sARG_PDECL\n",name); lineno++;
3654 fprintf(out,"#define %sARG_FETCH\n",name); lineno++;
3655 fprintf(out,"#define %sARG_STORE\n",name); lineno++;
drh75897232000-05-29 14:26:00 +00003656 }
3657 if( mhflag ){
3658 fprintf(out,"#endif\n"); lineno++;
3659 }
3660 fprintf(out,"#define YYNSTATE %d\n",lemp->nstate); lineno++;
3661 fprintf(out,"#define YYNRULE %d\n",lemp->nrule); lineno++;
drhc4dd3fd2008-01-22 01:48:05 +00003662 if( lemp->errsym->useCnt ){
3663 fprintf(out,"#define YYERRORSYMBOL %d\n",lemp->errsym->index); lineno++;
3664 fprintf(out,"#define YYERRSYMDT yy%d\n",lemp->errsym->dtnum); lineno++;
3665 }
drh0bd1f4e2002-06-06 18:54:39 +00003666 if( lemp->has_fallback ){
3667 fprintf(out,"#define YYFALLBACK 1\n"); lineno++;
3668 }
drh75897232000-05-29 14:26:00 +00003669 tplt_xfer(lemp->name,in,out,&lineno);
3670
drh8b582012003-10-21 13:16:03 +00003671 /* Generate the action table and its associates:
drh75897232000-05-29 14:26:00 +00003672 **
drh8b582012003-10-21 13:16:03 +00003673 ** yy_action[] A single table containing all actions.
3674 ** yy_lookahead[] A table containing the lookahead for each entry in
3675 ** yy_action. Used to detect hash collisions.
3676 ** yy_shift_ofst[] For each state, the offset into yy_action for
3677 ** shifting terminals.
3678 ** yy_reduce_ofst[] For each state, the offset into yy_action for
3679 ** shifting non-terminals after a reduce.
3680 ** yy_default[] Default action for each state.
drh75897232000-05-29 14:26:00 +00003681 */
drh75897232000-05-29 14:26:00 +00003682
drh8b582012003-10-21 13:16:03 +00003683 /* Compute the actions on all states and count them up */
icculus9e44cf12010-02-14 17:14:22 +00003684 ax = (struct axset *) calloc(lemp->nstate*2, sizeof(ax[0]));
drhfdbf9282003-10-21 16:34:41 +00003685 if( ax==0 ){
3686 fprintf(stderr,"malloc failed\n");
3687 exit(1);
3688 }
drh75897232000-05-29 14:26:00 +00003689 for(i=0; i<lemp->nstate; i++){
drh75897232000-05-29 14:26:00 +00003690 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003691 ax[i*2].stp = stp;
3692 ax[i*2].isTkn = 1;
3693 ax[i*2].nAction = stp->nTknAct;
3694 ax[i*2+1].stp = stp;
3695 ax[i*2+1].isTkn = 0;
3696 ax[i*2+1].nAction = stp->nNtAct;
drh75897232000-05-29 14:26:00 +00003697 }
drh8b582012003-10-21 13:16:03 +00003698 mxTknOfst = mnTknOfst = 0;
3699 mxNtOfst = mnNtOfst = 0;
3700
drhfdbf9282003-10-21 16:34:41 +00003701 /* Compute the action table. In order to try to keep the size of the
3702 ** action table to a minimum, the heuristic of placing the largest action
3703 ** sets first is used.
drh8b582012003-10-21 13:16:03 +00003704 */
drhe594bc32009-11-03 13:02:25 +00003705 for(i=0; i<lemp->nstate*2; i++) ax[i].iOrder = i;
drhfdbf9282003-10-21 16:34:41 +00003706 qsort(ax, lemp->nstate*2, sizeof(ax[0]), axset_compare);
drh8b582012003-10-21 13:16:03 +00003707 pActtab = acttab_alloc();
drhfdbf9282003-10-21 16:34:41 +00003708 for(i=0; i<lemp->nstate*2 && ax[i].nAction>0; i++){
3709 stp = ax[i].stp;
3710 if( ax[i].isTkn ){
3711 for(ap=stp->ap; ap; ap=ap->next){
3712 int action;
3713 if( ap->sp->index>=lemp->nterminal ) continue;
3714 action = compute_action(lemp, ap);
3715 if( action<0 ) continue;
3716 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003717 }
drhfdbf9282003-10-21 16:34:41 +00003718 stp->iTknOfst = acttab_insert(pActtab);
3719 if( stp->iTknOfst<mnTknOfst ) mnTknOfst = stp->iTknOfst;
3720 if( stp->iTknOfst>mxTknOfst ) mxTknOfst = stp->iTknOfst;
3721 }else{
3722 for(ap=stp->ap; ap; ap=ap->next){
3723 int action;
3724 if( ap->sp->index<lemp->nterminal ) continue;
3725 if( ap->sp->index==lemp->nsymbol ) continue;
3726 action = compute_action(lemp, ap);
3727 if( action<0 ) continue;
3728 acttab_action(pActtab, ap->sp->index, action);
drh8b582012003-10-21 13:16:03 +00003729 }
drhfdbf9282003-10-21 16:34:41 +00003730 stp->iNtOfst = acttab_insert(pActtab);
3731 if( stp->iNtOfst<mnNtOfst ) mnNtOfst = stp->iNtOfst;
3732 if( stp->iNtOfst>mxNtOfst ) mxNtOfst = stp->iNtOfst;
drh8b582012003-10-21 13:16:03 +00003733 }
3734 }
drhfdbf9282003-10-21 16:34:41 +00003735 free(ax);
drh8b582012003-10-21 13:16:03 +00003736
3737 /* Output the yy_action table */
drh8b582012003-10-21 13:16:03 +00003738 n = acttab_size(pActtab);
drhf16371d2009-11-03 19:18:31 +00003739 fprintf(out,"#define YY_ACTTAB_COUNT (%d)\n", n); lineno++;
3740 fprintf(out,"static const YYACTIONTYPE yy_action[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003741 for(i=j=0; i<n; i++){
3742 int action = acttab_yyaction(pActtab, i);
drhe0479212007-01-12 23:09:23 +00003743 if( action<0 ) action = lemp->nstate + lemp->nrule + 2;
drhfdbf9282003-10-21 16:34:41 +00003744 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003745 fprintf(out, " %4d,", action);
3746 if( j==9 || i==n-1 ){
3747 fprintf(out, "\n"); lineno++;
3748 j = 0;
3749 }else{
3750 j++;
3751 }
3752 }
3753 fprintf(out, "};\n"); lineno++;
3754
3755 /* Output the yy_lookahead table */
drh57196282004-10-06 15:41:16 +00003756 fprintf(out,"static const YYCODETYPE yy_lookahead[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003757 for(i=j=0; i<n; i++){
3758 int la = acttab_yylookahead(pActtab, i);
3759 if( la<0 ) la = lemp->nsymbol;
drhfdbf9282003-10-21 16:34:41 +00003760 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003761 fprintf(out, " %4d,", la);
3762 if( j==9 || i==n-1 ){
3763 fprintf(out, "\n"); lineno++;
3764 j = 0;
3765 }else{
3766 j++;
3767 }
3768 }
3769 fprintf(out, "};\n"); lineno++;
3770
3771 /* Output the yy_shift_ofst[] table */
3772 fprintf(out, "#define YY_SHIFT_USE_DFLT (%d)\n", mnTknOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003773 n = lemp->nstate;
3774 while( n>0 && lemp->sorted[n-1]->iTknOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003775 fprintf(out, "#define YY_SHIFT_COUNT (%d)\n", n-1); lineno++;
3776 fprintf(out, "#define YY_SHIFT_MIN (%d)\n", mnTknOfst); lineno++;
3777 fprintf(out, "#define YY_SHIFT_MAX (%d)\n", mxTknOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003778 fprintf(out, "static const %s yy_shift_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003779 minimum_size_type(mnTknOfst-1, mxTknOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003780 for(i=j=0; i<n; i++){
3781 int ofst;
3782 stp = lemp->sorted[i];
3783 ofst = stp->iTknOfst;
3784 if( ofst==NO_OFFSET ) ofst = mnTknOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003785 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003786 fprintf(out, " %4d,", ofst);
3787 if( j==9 || i==n-1 ){
3788 fprintf(out, "\n"); lineno++;
3789 j = 0;
3790 }else{
3791 j++;
3792 }
3793 }
3794 fprintf(out, "};\n"); lineno++;
3795
3796 /* Output the yy_reduce_ofst[] table */
3797 fprintf(out, "#define YY_REDUCE_USE_DFLT (%d)\n", mnNtOfst-1); lineno++;
drhada354d2005-11-05 15:03:59 +00003798 n = lemp->nstate;
3799 while( n>0 && lemp->sorted[n-1]->iNtOfst==NO_OFFSET ) n--;
drhf16371d2009-11-03 19:18:31 +00003800 fprintf(out, "#define YY_REDUCE_COUNT (%d)\n", n-1); lineno++;
3801 fprintf(out, "#define YY_REDUCE_MIN (%d)\n", mnNtOfst); lineno++;
3802 fprintf(out, "#define YY_REDUCE_MAX (%d)\n", mxNtOfst); lineno++;
drh57196282004-10-06 15:41:16 +00003803 fprintf(out, "static const %s yy_reduce_ofst[] = {\n",
drh8b582012003-10-21 13:16:03 +00003804 minimum_size_type(mnNtOfst-1, mxNtOfst)); lineno++;
drh8b582012003-10-21 13:16:03 +00003805 for(i=j=0; i<n; i++){
3806 int ofst;
3807 stp = lemp->sorted[i];
3808 ofst = stp->iNtOfst;
3809 if( ofst==NO_OFFSET ) ofst = mnNtOfst - 1;
drhfdbf9282003-10-21 16:34:41 +00003810 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003811 fprintf(out, " %4d,", ofst);
3812 if( j==9 || i==n-1 ){
3813 fprintf(out, "\n"); lineno++;
3814 j = 0;
3815 }else{
3816 j++;
3817 }
3818 }
3819 fprintf(out, "};\n"); lineno++;
3820
3821 /* Output the default action table */
drh57196282004-10-06 15:41:16 +00003822 fprintf(out, "static const YYACTIONTYPE yy_default[] = {\n"); lineno++;
drh8b582012003-10-21 13:16:03 +00003823 n = lemp->nstate;
3824 for(i=j=0; i<n; i++){
3825 stp = lemp->sorted[i];
drhfdbf9282003-10-21 16:34:41 +00003826 if( j==0 ) fprintf(out," /* %5d */ ", i);
drh8b582012003-10-21 13:16:03 +00003827 fprintf(out, " %4d,", stp->iDflt);
3828 if( j==9 || i==n-1 ){
3829 fprintf(out, "\n"); lineno++;
3830 j = 0;
3831 }else{
3832 j++;
3833 }
3834 }
3835 fprintf(out, "};\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003836 tplt_xfer(lemp->name,in,out,&lineno);
3837
drh0bd1f4e2002-06-06 18:54:39 +00003838 /* Generate the table of fallback tokens.
3839 */
3840 if( lemp->has_fallback ){
drh1441f3e2009-06-12 12:50:50 +00003841 int mx = lemp->nterminal - 1;
3842 while( mx>0 && lemp->symbols[mx]->fallback==0 ){ mx--; }
3843 for(i=0; i<=mx; i++){
drh0bd1f4e2002-06-06 18:54:39 +00003844 struct symbol *p = lemp->symbols[i];
3845 if( p->fallback==0 ){
3846 fprintf(out, " 0, /* %10s => nothing */\n", p->name);
3847 }else{
3848 fprintf(out, " %3d, /* %10s => %s */\n", p->fallback->index,
3849 p->name, p->fallback->name);
3850 }
3851 lineno++;
3852 }
3853 }
3854 tplt_xfer(lemp->name, in, out, &lineno);
3855
3856 /* Generate a table containing the symbolic name of every symbol
3857 */
drh75897232000-05-29 14:26:00 +00003858 for(i=0; i<lemp->nsymbol; i++){
3859 sprintf(line,"\"%s\",",lemp->symbols[i]->name);
3860 fprintf(out," %-15s",line);
3861 if( (i&3)==3 ){ fprintf(out,"\n"); lineno++; }
3862 }
3863 if( (i&3)!=0 ){ fprintf(out,"\n"); lineno++; }
3864 tplt_xfer(lemp->name,in,out,&lineno);
3865
drh0bd1f4e2002-06-06 18:54:39 +00003866 /* Generate a table containing a text string that describes every
drh34ff57b2008-07-14 12:27:51 +00003867 ** rule in the rule set of the grammar. This information is used
drh0bd1f4e2002-06-06 18:54:39 +00003868 ** when tracing REDUCE actions.
3869 */
3870 for(i=0, rp=lemp->rule; rp; rp=rp->next, i++){
3871 assert( rp->index==i );
drhc4dd3fd2008-01-22 01:48:05 +00003872 fprintf(out," /* %3d */ \"", i);
3873 writeRuleText(out, rp);
drh0bd1f4e2002-06-06 18:54:39 +00003874 fprintf(out,"\",\n"); lineno++;
3875 }
3876 tplt_xfer(lemp->name,in,out,&lineno);
3877
drh75897232000-05-29 14:26:00 +00003878 /* Generate code which executes every time a symbol is popped from
3879 ** the stack while processing errors or while destroying the parser.
drh0bd1f4e2002-06-06 18:54:39 +00003880 ** (In other words, generate the %destructor actions)
3881 */
drh75897232000-05-29 14:26:00 +00003882 if( lemp->tokendest ){
drh4dc8ef52008-07-01 17:13:57 +00003883 int once = 1;
drh75897232000-05-29 14:26:00 +00003884 for(i=0; i<lemp->nsymbol; i++){
3885 struct symbol *sp = lemp->symbols[i];
3886 if( sp==0 || sp->type!=TERMINAL ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003887 if( once ){
3888 fprintf(out, " /* TERMINAL Destructor */\n"); lineno++;
3889 once = 0;
3890 }
drhc53eed12009-06-12 17:46:19 +00003891 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh75897232000-05-29 14:26:00 +00003892 }
3893 for(i=0; i<lemp->nsymbol && lemp->symbols[i]->type!=TERMINAL; i++);
3894 if( i<lemp->nsymbol ){
3895 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3896 fprintf(out," break;\n"); lineno++;
3897 }
3898 }
drh8d659732005-01-13 23:54:06 +00003899 if( lemp->vardest ){
3900 struct symbol *dflt_sp = 0;
drh4dc8ef52008-07-01 17:13:57 +00003901 int once = 1;
drh8d659732005-01-13 23:54:06 +00003902 for(i=0; i<lemp->nsymbol; i++){
3903 struct symbol *sp = lemp->symbols[i];
3904 if( sp==0 || sp->type==TERMINAL ||
3905 sp->index<=0 || sp->destructor!=0 ) continue;
drh4dc8ef52008-07-01 17:13:57 +00003906 if( once ){
3907 fprintf(out, " /* Default NON-TERMINAL Destructor */\n"); lineno++;
3908 once = 0;
3909 }
drhc53eed12009-06-12 17:46:19 +00003910 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh8d659732005-01-13 23:54:06 +00003911 dflt_sp = sp;
3912 }
3913 if( dflt_sp!=0 ){
3914 emit_destructor_code(out,dflt_sp,lemp,&lineno);
drh8d659732005-01-13 23:54:06 +00003915 }
drh4dc8ef52008-07-01 17:13:57 +00003916 fprintf(out," break;\n"); lineno++;
drh8d659732005-01-13 23:54:06 +00003917 }
drh75897232000-05-29 14:26:00 +00003918 for(i=0; i<lemp->nsymbol; i++){
3919 struct symbol *sp = lemp->symbols[i];
3920 if( sp==0 || sp->type==TERMINAL || sp->destructor==0 ) continue;
drh75013012009-06-12 15:47:34 +00003921 fprintf(out," case %d: /* %s */\n", sp->index, sp->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003922
3923 /* Combine duplicate destructors into a single case */
3924 for(j=i+1; j<lemp->nsymbol; j++){
3925 struct symbol *sp2 = lemp->symbols[j];
3926 if( sp2 && sp2->type!=TERMINAL && sp2->destructor
3927 && sp2->dtnum==sp->dtnum
3928 && strcmp(sp->destructor,sp2->destructor)==0 ){
drhc53eed12009-06-12 17:46:19 +00003929 fprintf(out," case %d: /* %s */\n",
3930 sp2->index, sp2->name); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003931 sp2->destructor = 0;
3932 }
3933 }
3934
drh75897232000-05-29 14:26:00 +00003935 emit_destructor_code(out,lemp->symbols[i],lemp,&lineno);
3936 fprintf(out," break;\n"); lineno++;
3937 }
drh75897232000-05-29 14:26:00 +00003938 tplt_xfer(lemp->name,in,out,&lineno);
3939
3940 /* Generate code which executes whenever the parser stack overflows */
drha5808f32008-04-27 22:19:44 +00003941 tplt_print(out,lemp,lemp->overflow,&lineno);
drh75897232000-05-29 14:26:00 +00003942 tplt_xfer(lemp->name,in,out,&lineno);
3943
3944 /* Generate the table of rule information
3945 **
3946 ** Note: This code depends on the fact that rules are number
3947 ** sequentually beginning with 0.
3948 */
3949 for(rp=lemp->rule; rp; rp=rp->next){
3950 fprintf(out," { %d, %d },\n",rp->lhs->index,rp->nrhs); lineno++;
3951 }
3952 tplt_xfer(lemp->name,in,out,&lineno);
3953
3954 /* Generate code which execution during each REDUCE action */
3955 for(rp=lemp->rule; rp; rp=rp->next){
drh61e339a2007-01-16 03:09:02 +00003956 translate_code(lemp, rp);
drh0bb132b2004-07-20 14:06:51 +00003957 }
drhc53eed12009-06-12 17:46:19 +00003958 /* First output rules other than the default: rule */
drh0bb132b2004-07-20 14:06:51 +00003959 for(rp=lemp->rule; rp; rp=rp->next){
drhc53eed12009-06-12 17:46:19 +00003960 struct rule *rp2; /* Other rules with the same action */
drh0bb132b2004-07-20 14:06:51 +00003961 if( rp->code==0 ) continue;
drhc53eed12009-06-12 17:46:19 +00003962 if( rp->code[0]=='\n' && rp->code[1]==0 ) continue; /* Will be default: */
drhc4dd3fd2008-01-22 01:48:05 +00003963 fprintf(out," case %d: /* ", rp->index);
3964 writeRuleText(out, rp);
3965 fprintf(out, " */\n"); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003966 for(rp2=rp->next; rp2; rp2=rp2->next){
3967 if( rp2->code==rp->code ){
drhc4dd3fd2008-01-22 01:48:05 +00003968 fprintf(out," case %d: /* ", rp2->index);
3969 writeRuleText(out, rp2);
drh8a415d32009-06-12 13:53:51 +00003970 fprintf(out," */ yytestcase(yyruleno==%d);\n", rp2->index); lineno++;
drh0bb132b2004-07-20 14:06:51 +00003971 rp2->code = 0;
3972 }
3973 }
drh75897232000-05-29 14:26:00 +00003974 emit_code(out,rp,lemp,&lineno);
3975 fprintf(out," break;\n"); lineno++;
drhc53eed12009-06-12 17:46:19 +00003976 rp->code = 0;
drh75897232000-05-29 14:26:00 +00003977 }
drhc53eed12009-06-12 17:46:19 +00003978 /* Finally, output the default: rule. We choose as the default: all
3979 ** empty actions. */
3980 fprintf(out," default:\n"); lineno++;
3981 for(rp=lemp->rule; rp; rp=rp->next){
3982 if( rp->code==0 ) continue;
3983 assert( rp->code[0]=='\n' && rp->code[1]==0 );
3984 fprintf(out," /* (%d) ", rp->index);
3985 writeRuleText(out, rp);
3986 fprintf(out, " */ yytestcase(yyruleno==%d);\n", rp->index); lineno++;
3987 }
3988 fprintf(out," break;\n"); lineno++;
drh75897232000-05-29 14:26:00 +00003989 tplt_xfer(lemp->name,in,out,&lineno);
3990
3991 /* Generate code which executes if a parse fails */
drha5808f32008-04-27 22:19:44 +00003992 tplt_print(out,lemp,lemp->failure,&lineno);
drh75897232000-05-29 14:26:00 +00003993 tplt_xfer(lemp->name,in,out,&lineno);
3994
3995 /* Generate code which executes when a syntax error occurs */
drha5808f32008-04-27 22:19:44 +00003996 tplt_print(out,lemp,lemp->error,&lineno);
drh75897232000-05-29 14:26:00 +00003997 tplt_xfer(lemp->name,in,out,&lineno);
3998
3999 /* Generate code which executes when the parser accepts its input */
drha5808f32008-04-27 22:19:44 +00004000 tplt_print(out,lemp,lemp->accept,&lineno);
drh75897232000-05-29 14:26:00 +00004001 tplt_xfer(lemp->name,in,out,&lineno);
4002
4003 /* Append any addition code the user desires */
drha5808f32008-04-27 22:19:44 +00004004 tplt_print(out,lemp,lemp->extracode,&lineno);
drh75897232000-05-29 14:26:00 +00004005
4006 fclose(in);
4007 fclose(out);
4008 return;
4009}
4010
4011/* Generate a header file for the parser */
icculus9e44cf12010-02-14 17:14:22 +00004012void ReportHeader(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004013{
4014 FILE *out, *in;
icculus9e44cf12010-02-14 17:14:22 +00004015 const char *prefix;
drh75897232000-05-29 14:26:00 +00004016 char line[LINESIZE];
4017 char pattern[LINESIZE];
4018 int i;
4019
4020 if( lemp->tokenprefix ) prefix = lemp->tokenprefix;
4021 else prefix = "";
drh2aa6ca42004-09-10 00:14:04 +00004022 in = file_open(lemp,".h","rb");
drh75897232000-05-29 14:26:00 +00004023 if( in ){
drh8ba0d1c2012-06-16 15:26:31 +00004024 int nextChar;
drh75897232000-05-29 14:26:00 +00004025 for(i=1; i<lemp->nterminal && fgets(line,LINESIZE,in); i++){
4026 sprintf(pattern,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4027 if( strcmp(line,pattern) ) break;
4028 }
drh8ba0d1c2012-06-16 15:26:31 +00004029 nextChar = fgetc(in);
drh75897232000-05-29 14:26:00 +00004030 fclose(in);
drh8ba0d1c2012-06-16 15:26:31 +00004031 if( i==lemp->nterminal && nextChar==EOF ){
drh75897232000-05-29 14:26:00 +00004032 /* No change in the file. Don't rewrite it. */
4033 return;
4034 }
4035 }
drh2aa6ca42004-09-10 00:14:04 +00004036 out = file_open(lemp,".h","wb");
drh75897232000-05-29 14:26:00 +00004037 if( out ){
4038 for(i=1; i<lemp->nterminal; i++){
4039 fprintf(out,"#define %s%-30s %2d\n",prefix,lemp->symbols[i]->name,i);
4040 }
4041 fclose(out);
4042 }
4043 return;
4044}
4045
4046/* Reduce the size of the action tables, if possible, by making use
4047** of defaults.
4048**
drhb59499c2002-02-23 18:45:13 +00004049** In this version, we take the most frequent REDUCE action and make
drhe09daa92006-06-10 13:29:31 +00004050** it the default. Except, there is no default if the wildcard token
4051** is a possible look-ahead.
drh75897232000-05-29 14:26:00 +00004052*/
icculus9e44cf12010-02-14 17:14:22 +00004053void CompressTables(struct lemon *lemp)
drh75897232000-05-29 14:26:00 +00004054{
4055 struct state *stp;
drhb59499c2002-02-23 18:45:13 +00004056 struct action *ap, *ap2;
4057 struct rule *rp, *rp2, *rbest;
4058 int nbest, n;
drh75897232000-05-29 14:26:00 +00004059 int i;
drhe09daa92006-06-10 13:29:31 +00004060 int usesWildcard;
drh75897232000-05-29 14:26:00 +00004061
4062 for(i=0; i<lemp->nstate; i++){
4063 stp = lemp->sorted[i];
drhb59499c2002-02-23 18:45:13 +00004064 nbest = 0;
4065 rbest = 0;
drhe09daa92006-06-10 13:29:31 +00004066 usesWildcard = 0;
drh75897232000-05-29 14:26:00 +00004067
drhb59499c2002-02-23 18:45:13 +00004068 for(ap=stp->ap; ap; ap=ap->next){
drhe09daa92006-06-10 13:29:31 +00004069 if( ap->type==SHIFT && ap->sp==lemp->wildcard ){
4070 usesWildcard = 1;
4071 }
drhb59499c2002-02-23 18:45:13 +00004072 if( ap->type!=REDUCE ) continue;
4073 rp = ap->x.rp;
drhb4960992007-10-05 16:16:36 +00004074 if( rp->lhsStart ) continue;
drhb59499c2002-02-23 18:45:13 +00004075 if( rp==rbest ) continue;
4076 n = 1;
4077 for(ap2=ap->next; ap2; ap2=ap2->next){
4078 if( ap2->type!=REDUCE ) continue;
4079 rp2 = ap2->x.rp;
4080 if( rp2==rbest ) continue;
4081 if( rp2==rp ) n++;
4082 }
4083 if( n>nbest ){
4084 nbest = n;
4085 rbest = rp;
drh75897232000-05-29 14:26:00 +00004086 }
4087 }
drhb59499c2002-02-23 18:45:13 +00004088
4089 /* Do not make a default if the number of rules to default
drhe09daa92006-06-10 13:29:31 +00004090 ** is not at least 1 or if the wildcard token is a possible
4091 ** lookahead.
4092 */
4093 if( nbest<1 || usesWildcard ) continue;
drh75897232000-05-29 14:26:00 +00004094
drhb59499c2002-02-23 18:45:13 +00004095
4096 /* Combine matching REDUCE actions into a single default */
4097 for(ap=stp->ap; ap; ap=ap->next){
4098 if( ap->type==REDUCE && ap->x.rp==rbest ) break;
4099 }
drh75897232000-05-29 14:26:00 +00004100 assert( ap );
4101 ap->sp = Symbol_new("{default}");
4102 for(ap=ap->next; ap; ap=ap->next){
drhb59499c2002-02-23 18:45:13 +00004103 if( ap->type==REDUCE && ap->x.rp==rbest ) ap->type = NOT_USED;
drh75897232000-05-29 14:26:00 +00004104 }
4105 stp->ap = Action_sort(stp->ap);
4106 }
4107}
drhb59499c2002-02-23 18:45:13 +00004108
drhada354d2005-11-05 15:03:59 +00004109
4110/*
4111** Compare two states for sorting purposes. The smaller state is the
4112** one with the most non-terminal actions. If they have the same number
4113** of non-terminal actions, then the smaller is the one with the most
4114** token actions.
4115*/
4116static int stateResortCompare(const void *a, const void *b){
4117 const struct state *pA = *(const struct state**)a;
4118 const struct state *pB = *(const struct state**)b;
4119 int n;
4120
4121 n = pB->nNtAct - pA->nNtAct;
4122 if( n==0 ){
4123 n = pB->nTknAct - pA->nTknAct;
drhe594bc32009-11-03 13:02:25 +00004124 if( n==0 ){
4125 n = pB->statenum - pA->statenum;
4126 }
drhada354d2005-11-05 15:03:59 +00004127 }
drhe594bc32009-11-03 13:02:25 +00004128 assert( n!=0 );
drhada354d2005-11-05 15:03:59 +00004129 return n;
4130}
4131
4132
4133/*
4134** Renumber and resort states so that states with fewer choices
4135** occur at the end. Except, keep state 0 as the first state.
4136*/
icculus9e44cf12010-02-14 17:14:22 +00004137void ResortStates(struct lemon *lemp)
drhada354d2005-11-05 15:03:59 +00004138{
4139 int i;
4140 struct state *stp;
4141 struct action *ap;
4142
4143 for(i=0; i<lemp->nstate; i++){
4144 stp = lemp->sorted[i];
4145 stp->nTknAct = stp->nNtAct = 0;
4146 stp->iDflt = lemp->nstate + lemp->nrule;
4147 stp->iTknOfst = NO_OFFSET;
4148 stp->iNtOfst = NO_OFFSET;
4149 for(ap=stp->ap; ap; ap=ap->next){
4150 if( compute_action(lemp,ap)>=0 ){
4151 if( ap->sp->index<lemp->nterminal ){
4152 stp->nTknAct++;
4153 }else if( ap->sp->index<lemp->nsymbol ){
4154 stp->nNtAct++;
4155 }else{
4156 stp->iDflt = compute_action(lemp, ap);
4157 }
4158 }
4159 }
4160 }
4161 qsort(&lemp->sorted[1], lemp->nstate-1, sizeof(lemp->sorted[0]),
4162 stateResortCompare);
4163 for(i=0; i<lemp->nstate; i++){
4164 lemp->sorted[i]->statenum = i;
4165 }
4166}
4167
4168
drh75897232000-05-29 14:26:00 +00004169/***************** From the file "set.c" ************************************/
4170/*
4171** Set manipulation routines for the LEMON parser generator.
4172*/
4173
4174static int size = 0;
4175
4176/* Set the set size */
icculus9e44cf12010-02-14 17:14:22 +00004177void SetSize(int n)
drh75897232000-05-29 14:26:00 +00004178{
4179 size = n+1;
4180}
4181
4182/* Allocate a new set */
4183char *SetNew(){
4184 char *s;
drh9892c5d2007-12-21 00:02:11 +00004185 s = (char*)calloc( size, 1);
drh75897232000-05-29 14:26:00 +00004186 if( s==0 ){
4187 extern void memory_error();
4188 memory_error();
4189 }
drh75897232000-05-29 14:26:00 +00004190 return s;
4191}
4192
4193/* Deallocate a set */
icculus9e44cf12010-02-14 17:14:22 +00004194void SetFree(char *s)
drh75897232000-05-29 14:26:00 +00004195{
4196 free(s);
4197}
4198
4199/* Add a new element to the set. Return TRUE if the element was added
4200** and FALSE if it was already there. */
icculus9e44cf12010-02-14 17:14:22 +00004201int SetAdd(char *s, int e)
drh75897232000-05-29 14:26:00 +00004202{
4203 int rv;
drh9892c5d2007-12-21 00:02:11 +00004204 assert( e>=0 && e<size );
drh75897232000-05-29 14:26:00 +00004205 rv = s[e];
4206 s[e] = 1;
4207 return !rv;
4208}
4209
4210/* Add every element of s2 to s1. Return TRUE if s1 changes. */
icculus9e44cf12010-02-14 17:14:22 +00004211int SetUnion(char *s1, char *s2)
drh75897232000-05-29 14:26:00 +00004212{
4213 int i, progress;
4214 progress = 0;
4215 for(i=0; i<size; i++){
4216 if( s2[i]==0 ) continue;
4217 if( s1[i]==0 ){
4218 progress = 1;
4219 s1[i] = 1;
4220 }
4221 }
4222 return progress;
4223}
4224/********************** From the file "table.c" ****************************/
4225/*
4226** All code in this file has been automatically generated
4227** from a specification in the file
4228** "table.q"
4229** by the associative array code building program "aagen".
4230** Do not edit this file! Instead, edit the specification
4231** file, then rerun aagen.
4232*/
4233/*
4234** Code for processing tables in the LEMON parser generator.
4235*/
4236
drh01f75f22013-10-02 20:46:30 +00004237PRIVATE unsigned strhash(const char *x)
drh75897232000-05-29 14:26:00 +00004238{
drh01f75f22013-10-02 20:46:30 +00004239 unsigned h = 0;
4240 while( *x ) h = h*13 + *(x++);
drh75897232000-05-29 14:26:00 +00004241 return h;
4242}
4243
4244/* Works like strdup, sort of. Save a string in malloced memory, but
4245** keep strings in a table so that the same string is not in more
4246** than one place.
4247*/
icculus9e44cf12010-02-14 17:14:22 +00004248const char *Strsafe(const char *y)
drh75897232000-05-29 14:26:00 +00004249{
icculus9e44cf12010-02-14 17:14:22 +00004250 const char *z;
4251 char *cpy;
drh75897232000-05-29 14:26:00 +00004252
drh916f75f2006-07-17 00:19:39 +00004253 if( y==0 ) return 0;
drh75897232000-05-29 14:26:00 +00004254 z = Strsafe_find(y);
icculus9e44cf12010-02-14 17:14:22 +00004255 if( z==0 && (cpy=(char *)malloc( lemonStrlen(y)+1 ))!=0 ){
4256 strcpy(cpy,y);
4257 z = cpy;
drh75897232000-05-29 14:26:00 +00004258 Strsafe_insert(z);
4259 }
4260 MemoryCheck(z);
4261 return z;
4262}
4263
4264/* There is one instance of the following structure for each
4265** associative array of type "x1".
4266*/
4267struct s_x1 {
4268 int size; /* The number of available slots. */
4269 /* Must be a power of 2 greater than or */
4270 /* equal to 1 */
4271 int count; /* Number of currently slots filled */
4272 struct s_x1node *tbl; /* The data stored here */
4273 struct s_x1node **ht; /* Hash table for lookups */
4274};
4275
4276/* There is one instance of this structure for every data element
4277** in an associative array of type "x1".
4278*/
4279typedef struct s_x1node {
icculus9e44cf12010-02-14 17:14:22 +00004280 const char *data; /* The data */
drh75897232000-05-29 14:26:00 +00004281 struct s_x1node *next; /* Next entry with the same hash */
4282 struct s_x1node **from; /* Previous link */
4283} x1node;
4284
4285/* There is only one instance of the array, which is the following */
4286static struct s_x1 *x1a;
4287
4288/* Allocate a new associative array */
4289void Strsafe_init(){
4290 if( x1a ) return;
4291 x1a = (struct s_x1*)malloc( sizeof(struct s_x1) );
4292 if( x1a ){
4293 x1a->size = 1024;
4294 x1a->count = 0;
4295 x1a->tbl = (x1node*)malloc(
4296 (sizeof(x1node) + sizeof(x1node*))*1024 );
4297 if( x1a->tbl==0 ){
4298 free(x1a);
4299 x1a = 0;
4300 }else{
4301 int i;
4302 x1a->ht = (x1node**)&(x1a->tbl[1024]);
4303 for(i=0; i<1024; i++) x1a->ht[i] = 0;
4304 }
4305 }
4306}
4307/* Insert a new record into the array. Return TRUE if successful.
4308** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004309int Strsafe_insert(const char *data)
drh75897232000-05-29 14:26:00 +00004310{
4311 x1node *np;
drh01f75f22013-10-02 20:46:30 +00004312 unsigned h;
4313 unsigned ph;
drh75897232000-05-29 14:26:00 +00004314
4315 if( x1a==0 ) return 0;
4316 ph = strhash(data);
4317 h = ph & (x1a->size-1);
4318 np = x1a->ht[h];
4319 while( np ){
4320 if( strcmp(np->data,data)==0 ){
4321 /* An existing entry with the same key is found. */
4322 /* Fail because overwrite is not allows. */
4323 return 0;
4324 }
4325 np = np->next;
4326 }
4327 if( x1a->count>=x1a->size ){
4328 /* Need to make the hash table bigger */
4329 int i,size;
4330 struct s_x1 array;
4331 array.size = size = x1a->size*2;
4332 array.count = x1a->count;
4333 array.tbl = (x1node*)malloc(
4334 (sizeof(x1node) + sizeof(x1node*))*size );
4335 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4336 array.ht = (x1node**)&(array.tbl[size]);
4337 for(i=0; i<size; i++) array.ht[i] = 0;
4338 for(i=0; i<x1a->count; i++){
4339 x1node *oldnp, *newnp;
4340 oldnp = &(x1a->tbl[i]);
4341 h = strhash(oldnp->data) & (size-1);
4342 newnp = &(array.tbl[i]);
4343 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4344 newnp->next = array.ht[h];
4345 newnp->data = oldnp->data;
4346 newnp->from = &(array.ht[h]);
4347 array.ht[h] = newnp;
4348 }
4349 free(x1a->tbl);
4350 *x1a = array;
4351 }
4352 /* Insert the new data */
4353 h = ph & (x1a->size-1);
4354 np = &(x1a->tbl[x1a->count++]);
4355 np->data = data;
4356 if( x1a->ht[h] ) x1a->ht[h]->from = &(np->next);
4357 np->next = x1a->ht[h];
4358 x1a->ht[h] = np;
4359 np->from = &(x1a->ht[h]);
4360 return 1;
4361}
4362
4363/* Return a pointer to data assigned to the given key. Return NULL
4364** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004365const char *Strsafe_find(const char *key)
drh75897232000-05-29 14:26:00 +00004366{
drh01f75f22013-10-02 20:46:30 +00004367 unsigned h;
drh75897232000-05-29 14:26:00 +00004368 x1node *np;
4369
4370 if( x1a==0 ) return 0;
4371 h = strhash(key) & (x1a->size-1);
4372 np = x1a->ht[h];
4373 while( np ){
4374 if( strcmp(np->data,key)==0 ) break;
4375 np = np->next;
4376 }
4377 return np ? np->data : 0;
4378}
4379
4380/* Return a pointer to the (terminal or nonterminal) symbol "x".
4381** Create a new symbol if this is the first time "x" has been seen.
4382*/
icculus9e44cf12010-02-14 17:14:22 +00004383struct symbol *Symbol_new(const char *x)
drh75897232000-05-29 14:26:00 +00004384{
4385 struct symbol *sp;
4386
4387 sp = Symbol_find(x);
4388 if( sp==0 ){
drh9892c5d2007-12-21 00:02:11 +00004389 sp = (struct symbol *)calloc(1, sizeof(struct symbol) );
drh75897232000-05-29 14:26:00 +00004390 MemoryCheck(sp);
4391 sp->name = Strsafe(x);
4392 sp->type = isupper(*x) ? TERMINAL : NONTERMINAL;
4393 sp->rule = 0;
drh0bd1f4e2002-06-06 18:54:39 +00004394 sp->fallback = 0;
drh75897232000-05-29 14:26:00 +00004395 sp->prec = -1;
4396 sp->assoc = UNK;
4397 sp->firstset = 0;
drhaa9f1122007-08-23 02:50:56 +00004398 sp->lambda = LEMON_FALSE;
drh75897232000-05-29 14:26:00 +00004399 sp->destructor = 0;
drh4dc8ef52008-07-01 17:13:57 +00004400 sp->destLineno = 0;
drh75897232000-05-29 14:26:00 +00004401 sp->datatype = 0;
drhc4dd3fd2008-01-22 01:48:05 +00004402 sp->useCnt = 0;
drh75897232000-05-29 14:26:00 +00004403 Symbol_insert(sp,sp->name);
4404 }
drhc4dd3fd2008-01-22 01:48:05 +00004405 sp->useCnt++;
drh75897232000-05-29 14:26:00 +00004406 return sp;
4407}
4408
drh60d31652004-02-22 00:08:04 +00004409/* Compare two symbols for working purposes
4410**
4411** Symbols that begin with upper case letters (terminals or tokens)
4412** must sort before symbols that begin with lower case letters
4413** (non-terminals). Other than that, the order does not matter.
4414**
4415** We find experimentally that leaving the symbols in their original
4416** order (the order they appeared in the grammar file) gives the
4417** smallest parser tables in SQLite.
4418*/
icculus9e44cf12010-02-14 17:14:22 +00004419int Symbolcmpp(const void *_a, const void *_b)
4420{
4421 const struct symbol **a = (const struct symbol **) _a;
4422 const struct symbol **b = (const struct symbol **) _b;
drh60d31652004-02-22 00:08:04 +00004423 int i1 = (**a).index + 10000000*((**a).name[0]>'Z');
4424 int i2 = (**b).index + 10000000*((**b).name[0]>'Z');
drhe594bc32009-11-03 13:02:25 +00004425 assert( i1!=i2 || strcmp((**a).name,(**b).name)==0 );
drh60d31652004-02-22 00:08:04 +00004426 return i1-i2;
drh75897232000-05-29 14:26:00 +00004427}
4428
4429/* There is one instance of the following structure for each
4430** associative array of type "x2".
4431*/
4432struct s_x2 {
4433 int size; /* The number of available slots. */
4434 /* Must be a power of 2 greater than or */
4435 /* equal to 1 */
4436 int count; /* Number of currently slots filled */
4437 struct s_x2node *tbl; /* The data stored here */
4438 struct s_x2node **ht; /* Hash table for lookups */
4439};
4440
4441/* There is one instance of this structure for every data element
4442** in an associative array of type "x2".
4443*/
4444typedef struct s_x2node {
icculus9e44cf12010-02-14 17:14:22 +00004445 struct symbol *data; /* The data */
4446 const char *key; /* The key */
drh75897232000-05-29 14:26:00 +00004447 struct s_x2node *next; /* Next entry with the same hash */
4448 struct s_x2node **from; /* Previous link */
4449} x2node;
4450
4451/* There is only one instance of the array, which is the following */
4452static struct s_x2 *x2a;
4453
4454/* Allocate a new associative array */
4455void Symbol_init(){
4456 if( x2a ) return;
4457 x2a = (struct s_x2*)malloc( sizeof(struct s_x2) );
4458 if( x2a ){
4459 x2a->size = 128;
4460 x2a->count = 0;
4461 x2a->tbl = (x2node*)malloc(
4462 (sizeof(x2node) + sizeof(x2node*))*128 );
4463 if( x2a->tbl==0 ){
4464 free(x2a);
4465 x2a = 0;
4466 }else{
4467 int i;
4468 x2a->ht = (x2node**)&(x2a->tbl[128]);
4469 for(i=0; i<128; i++) x2a->ht[i] = 0;
4470 }
4471 }
4472}
4473/* Insert a new record into the array. Return TRUE if successful.
4474** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004475int Symbol_insert(struct symbol *data, const char *key)
drh75897232000-05-29 14:26:00 +00004476{
4477 x2node *np;
drh01f75f22013-10-02 20:46:30 +00004478 unsigned h;
4479 unsigned ph;
drh75897232000-05-29 14:26:00 +00004480
4481 if( x2a==0 ) return 0;
4482 ph = strhash(key);
4483 h = ph & (x2a->size-1);
4484 np = x2a->ht[h];
4485 while( np ){
4486 if( strcmp(np->key,key)==0 ){
4487 /* An existing entry with the same key is found. */
4488 /* Fail because overwrite is not allows. */
4489 return 0;
4490 }
4491 np = np->next;
4492 }
4493 if( x2a->count>=x2a->size ){
4494 /* Need to make the hash table bigger */
4495 int i,size;
4496 struct s_x2 array;
4497 array.size = size = x2a->size*2;
4498 array.count = x2a->count;
4499 array.tbl = (x2node*)malloc(
4500 (sizeof(x2node) + sizeof(x2node*))*size );
4501 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4502 array.ht = (x2node**)&(array.tbl[size]);
4503 for(i=0; i<size; i++) array.ht[i] = 0;
4504 for(i=0; i<x2a->count; i++){
4505 x2node *oldnp, *newnp;
4506 oldnp = &(x2a->tbl[i]);
4507 h = strhash(oldnp->key) & (size-1);
4508 newnp = &(array.tbl[i]);
4509 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4510 newnp->next = array.ht[h];
4511 newnp->key = oldnp->key;
4512 newnp->data = oldnp->data;
4513 newnp->from = &(array.ht[h]);
4514 array.ht[h] = newnp;
4515 }
4516 free(x2a->tbl);
4517 *x2a = array;
4518 }
4519 /* Insert the new data */
4520 h = ph & (x2a->size-1);
4521 np = &(x2a->tbl[x2a->count++]);
4522 np->key = key;
4523 np->data = data;
4524 if( x2a->ht[h] ) x2a->ht[h]->from = &(np->next);
4525 np->next = x2a->ht[h];
4526 x2a->ht[h] = np;
4527 np->from = &(x2a->ht[h]);
4528 return 1;
4529}
4530
4531/* Return a pointer to data assigned to the given key. Return NULL
4532** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004533struct symbol *Symbol_find(const char *key)
drh75897232000-05-29 14:26:00 +00004534{
drh01f75f22013-10-02 20:46:30 +00004535 unsigned h;
drh75897232000-05-29 14:26:00 +00004536 x2node *np;
4537
4538 if( x2a==0 ) return 0;
4539 h = strhash(key) & (x2a->size-1);
4540 np = x2a->ht[h];
4541 while( np ){
4542 if( strcmp(np->key,key)==0 ) break;
4543 np = np->next;
4544 }
4545 return np ? np->data : 0;
4546}
4547
4548/* Return the n-th data. Return NULL if n is out of range. */
icculus9e44cf12010-02-14 17:14:22 +00004549struct symbol *Symbol_Nth(int n)
drh75897232000-05-29 14:26:00 +00004550{
4551 struct symbol *data;
4552 if( x2a && n>0 && n<=x2a->count ){
4553 data = x2a->tbl[n-1].data;
4554 }else{
4555 data = 0;
4556 }
4557 return data;
4558}
4559
4560/* Return the size of the array */
4561int Symbol_count()
4562{
4563 return x2a ? x2a->count : 0;
4564}
4565
4566/* Return an array of pointers to all data in the table.
4567** The array is obtained from malloc. Return NULL if memory allocation
4568** problems, or if the array is empty. */
4569struct symbol **Symbol_arrayof()
4570{
4571 struct symbol **array;
4572 int i,size;
4573 if( x2a==0 ) return 0;
4574 size = x2a->count;
drh9892c5d2007-12-21 00:02:11 +00004575 array = (struct symbol **)calloc(size, sizeof(struct symbol *));
drh75897232000-05-29 14:26:00 +00004576 if( array ){
4577 for(i=0; i<size; i++) array[i] = x2a->tbl[i].data;
4578 }
4579 return array;
4580}
4581
4582/* Compare two configurations */
icculus9e44cf12010-02-14 17:14:22 +00004583int Configcmp(const char *_a,const char *_b)
drh75897232000-05-29 14:26:00 +00004584{
icculus9e44cf12010-02-14 17:14:22 +00004585 const struct config *a = (struct config *) _a;
4586 const struct config *b = (struct config *) _b;
drh75897232000-05-29 14:26:00 +00004587 int x;
4588 x = a->rp->index - b->rp->index;
4589 if( x==0 ) x = a->dot - b->dot;
4590 return x;
4591}
4592
4593/* Compare two states */
icculus9e44cf12010-02-14 17:14:22 +00004594PRIVATE int statecmp(struct config *a, struct config *b)
drh75897232000-05-29 14:26:00 +00004595{
4596 int rc;
4597 for(rc=0; rc==0 && a && b; a=a->bp, b=b->bp){
4598 rc = a->rp->index - b->rp->index;
4599 if( rc==0 ) rc = a->dot - b->dot;
4600 }
4601 if( rc==0 ){
4602 if( a ) rc = 1;
4603 if( b ) rc = -1;
4604 }
4605 return rc;
4606}
4607
4608/* Hash a state */
drh01f75f22013-10-02 20:46:30 +00004609PRIVATE unsigned statehash(struct config *a)
drh75897232000-05-29 14:26:00 +00004610{
drh01f75f22013-10-02 20:46:30 +00004611 unsigned h=0;
drh75897232000-05-29 14:26:00 +00004612 while( a ){
4613 h = h*571 + a->rp->index*37 + a->dot;
4614 a = a->bp;
4615 }
4616 return h;
4617}
4618
4619/* Allocate a new state structure */
4620struct state *State_new()
4621{
icculus9e44cf12010-02-14 17:14:22 +00004622 struct state *newstate;
4623 newstate = (struct state *)calloc(1, sizeof(struct state) );
4624 MemoryCheck(newstate);
4625 return newstate;
drh75897232000-05-29 14:26:00 +00004626}
4627
4628/* There is one instance of the following structure for each
4629** associative array of type "x3".
4630*/
4631struct s_x3 {
4632 int size; /* The number of available slots. */
4633 /* Must be a power of 2 greater than or */
4634 /* equal to 1 */
4635 int count; /* Number of currently slots filled */
4636 struct s_x3node *tbl; /* The data stored here */
4637 struct s_x3node **ht; /* Hash table for lookups */
4638};
4639
4640/* There is one instance of this structure for every data element
4641** in an associative array of type "x3".
4642*/
4643typedef struct s_x3node {
4644 struct state *data; /* The data */
4645 struct config *key; /* The key */
4646 struct s_x3node *next; /* Next entry with the same hash */
4647 struct s_x3node **from; /* Previous link */
4648} x3node;
4649
4650/* There is only one instance of the array, which is the following */
4651static struct s_x3 *x3a;
4652
4653/* Allocate a new associative array */
4654void State_init(){
4655 if( x3a ) return;
4656 x3a = (struct s_x3*)malloc( sizeof(struct s_x3) );
4657 if( x3a ){
4658 x3a->size = 128;
4659 x3a->count = 0;
4660 x3a->tbl = (x3node*)malloc(
4661 (sizeof(x3node) + sizeof(x3node*))*128 );
4662 if( x3a->tbl==0 ){
4663 free(x3a);
4664 x3a = 0;
4665 }else{
4666 int i;
4667 x3a->ht = (x3node**)&(x3a->tbl[128]);
4668 for(i=0; i<128; i++) x3a->ht[i] = 0;
4669 }
4670 }
4671}
4672/* Insert a new record into the array. Return TRUE if successful.
4673** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004674int State_insert(struct state *data, struct config *key)
drh75897232000-05-29 14:26:00 +00004675{
4676 x3node *np;
drh01f75f22013-10-02 20:46:30 +00004677 unsigned h;
4678 unsigned ph;
drh75897232000-05-29 14:26:00 +00004679
4680 if( x3a==0 ) return 0;
4681 ph = statehash(key);
4682 h = ph & (x3a->size-1);
4683 np = x3a->ht[h];
4684 while( np ){
4685 if( statecmp(np->key,key)==0 ){
4686 /* An existing entry with the same key is found. */
4687 /* Fail because overwrite is not allows. */
4688 return 0;
4689 }
4690 np = np->next;
4691 }
4692 if( x3a->count>=x3a->size ){
4693 /* Need to make the hash table bigger */
4694 int i,size;
4695 struct s_x3 array;
4696 array.size = size = x3a->size*2;
4697 array.count = x3a->count;
4698 array.tbl = (x3node*)malloc(
4699 (sizeof(x3node) + sizeof(x3node*))*size );
4700 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4701 array.ht = (x3node**)&(array.tbl[size]);
4702 for(i=0; i<size; i++) array.ht[i] = 0;
4703 for(i=0; i<x3a->count; i++){
4704 x3node *oldnp, *newnp;
4705 oldnp = &(x3a->tbl[i]);
4706 h = statehash(oldnp->key) & (size-1);
4707 newnp = &(array.tbl[i]);
4708 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4709 newnp->next = array.ht[h];
4710 newnp->key = oldnp->key;
4711 newnp->data = oldnp->data;
4712 newnp->from = &(array.ht[h]);
4713 array.ht[h] = newnp;
4714 }
4715 free(x3a->tbl);
4716 *x3a = array;
4717 }
4718 /* Insert the new data */
4719 h = ph & (x3a->size-1);
4720 np = &(x3a->tbl[x3a->count++]);
4721 np->key = key;
4722 np->data = data;
4723 if( x3a->ht[h] ) x3a->ht[h]->from = &(np->next);
4724 np->next = x3a->ht[h];
4725 x3a->ht[h] = np;
4726 np->from = &(x3a->ht[h]);
4727 return 1;
4728}
4729
4730/* Return a pointer to data assigned to the given key. Return NULL
4731** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004732struct state *State_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004733{
drh01f75f22013-10-02 20:46:30 +00004734 unsigned h;
drh75897232000-05-29 14:26:00 +00004735 x3node *np;
4736
4737 if( x3a==0 ) return 0;
4738 h = statehash(key) & (x3a->size-1);
4739 np = x3a->ht[h];
4740 while( np ){
4741 if( statecmp(np->key,key)==0 ) break;
4742 np = np->next;
4743 }
4744 return np ? np->data : 0;
4745}
4746
4747/* Return an array of pointers to all data in the table.
4748** The array is obtained from malloc. Return NULL if memory allocation
4749** problems, or if the array is empty. */
4750struct state **State_arrayof()
4751{
4752 struct state **array;
4753 int i,size;
4754 if( x3a==0 ) return 0;
4755 size = x3a->count;
4756 array = (struct state **)malloc( sizeof(struct state *)*size );
4757 if( array ){
4758 for(i=0; i<size; i++) array[i] = x3a->tbl[i].data;
4759 }
4760 return array;
4761}
4762
4763/* Hash a configuration */
drh01f75f22013-10-02 20:46:30 +00004764PRIVATE unsigned confighash(struct config *a)
drh75897232000-05-29 14:26:00 +00004765{
drh01f75f22013-10-02 20:46:30 +00004766 unsigned h=0;
drh75897232000-05-29 14:26:00 +00004767 h = h*571 + a->rp->index*37 + a->dot;
4768 return h;
4769}
4770
4771/* There is one instance of the following structure for each
4772** associative array of type "x4".
4773*/
4774struct s_x4 {
4775 int size; /* The number of available slots. */
4776 /* Must be a power of 2 greater than or */
4777 /* equal to 1 */
4778 int count; /* Number of currently slots filled */
4779 struct s_x4node *tbl; /* The data stored here */
4780 struct s_x4node **ht; /* Hash table for lookups */
4781};
4782
4783/* There is one instance of this structure for every data element
4784** in an associative array of type "x4".
4785*/
4786typedef struct s_x4node {
4787 struct config *data; /* The data */
4788 struct s_x4node *next; /* Next entry with the same hash */
4789 struct s_x4node **from; /* Previous link */
4790} x4node;
4791
4792/* There is only one instance of the array, which is the following */
4793static struct s_x4 *x4a;
4794
4795/* Allocate a new associative array */
4796void Configtable_init(){
4797 if( x4a ) return;
4798 x4a = (struct s_x4*)malloc( sizeof(struct s_x4) );
4799 if( x4a ){
4800 x4a->size = 64;
4801 x4a->count = 0;
4802 x4a->tbl = (x4node*)malloc(
4803 (sizeof(x4node) + sizeof(x4node*))*64 );
4804 if( x4a->tbl==0 ){
4805 free(x4a);
4806 x4a = 0;
4807 }else{
4808 int i;
4809 x4a->ht = (x4node**)&(x4a->tbl[64]);
4810 for(i=0; i<64; i++) x4a->ht[i] = 0;
4811 }
4812 }
4813}
4814/* Insert a new record into the array. Return TRUE if successful.
4815** Prior data with the same key is NOT overwritten */
icculus9e44cf12010-02-14 17:14:22 +00004816int Configtable_insert(struct config *data)
drh75897232000-05-29 14:26:00 +00004817{
4818 x4node *np;
drh01f75f22013-10-02 20:46:30 +00004819 unsigned h;
4820 unsigned ph;
drh75897232000-05-29 14:26:00 +00004821
4822 if( x4a==0 ) return 0;
4823 ph = confighash(data);
4824 h = ph & (x4a->size-1);
4825 np = x4a->ht[h];
4826 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004827 if( Configcmp((const char *) np->data,(const char *) data)==0 ){
drh75897232000-05-29 14:26:00 +00004828 /* An existing entry with the same key is found. */
4829 /* Fail because overwrite is not allows. */
4830 return 0;
4831 }
4832 np = np->next;
4833 }
4834 if( x4a->count>=x4a->size ){
4835 /* Need to make the hash table bigger */
4836 int i,size;
4837 struct s_x4 array;
4838 array.size = size = x4a->size*2;
4839 array.count = x4a->count;
4840 array.tbl = (x4node*)malloc(
4841 (sizeof(x4node) + sizeof(x4node*))*size );
4842 if( array.tbl==0 ) return 0; /* Fail due to malloc failure */
4843 array.ht = (x4node**)&(array.tbl[size]);
4844 for(i=0; i<size; i++) array.ht[i] = 0;
4845 for(i=0; i<x4a->count; i++){
4846 x4node *oldnp, *newnp;
4847 oldnp = &(x4a->tbl[i]);
4848 h = confighash(oldnp->data) & (size-1);
4849 newnp = &(array.tbl[i]);
4850 if( array.ht[h] ) array.ht[h]->from = &(newnp->next);
4851 newnp->next = array.ht[h];
4852 newnp->data = oldnp->data;
4853 newnp->from = &(array.ht[h]);
4854 array.ht[h] = newnp;
4855 }
4856 free(x4a->tbl);
4857 *x4a = array;
4858 }
4859 /* Insert the new data */
4860 h = ph & (x4a->size-1);
4861 np = &(x4a->tbl[x4a->count++]);
4862 np->data = data;
4863 if( x4a->ht[h] ) x4a->ht[h]->from = &(np->next);
4864 np->next = x4a->ht[h];
4865 x4a->ht[h] = np;
4866 np->from = &(x4a->ht[h]);
4867 return 1;
4868}
4869
4870/* Return a pointer to data assigned to the given key. Return NULL
4871** if no such key. */
icculus9e44cf12010-02-14 17:14:22 +00004872struct config *Configtable_find(struct config *key)
drh75897232000-05-29 14:26:00 +00004873{
4874 int h;
4875 x4node *np;
4876
4877 if( x4a==0 ) return 0;
4878 h = confighash(key) & (x4a->size-1);
4879 np = x4a->ht[h];
4880 while( np ){
icculus9e44cf12010-02-14 17:14:22 +00004881 if( Configcmp((const char *) np->data,(const char *) key)==0 ) break;
drh75897232000-05-29 14:26:00 +00004882 np = np->next;
4883 }
4884 return np ? np->data : 0;
4885}
4886
4887/* Remove all data from the table. Pass each data to the function "f"
4888** as it is removed. ("f" may be null to avoid this step.) */
icculus9e44cf12010-02-14 17:14:22 +00004889void Configtable_clear(int(*f)(struct config *))
drh75897232000-05-29 14:26:00 +00004890{
4891 int i;
4892 if( x4a==0 || x4a->count==0 ) return;
4893 if( f ) for(i=0; i<x4a->count; i++) (*f)(x4a->tbl[i].data);
4894 for(i=0; i<x4a->size; i++) x4a->ht[i] = 0;
4895 x4a->count = 0;
4896 return;
4897}