ntdll: Directly access the syscall table variable on ARM64.
[wine.git] / tools / wmc / mcy.y
blob52f28ca6b74ab40fe017ef71a06bc7e76a73cdcb
1 /*
2 * Wine Message Compiler parser
4 * Copyright 2000 Bertho A. Stultiens (BS)
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
20 * NOTES:
22 * The basic grammar of the file is yet another example of, humpf,
23 * design. There is a mix of context-insensitive and -sensitive
24 * stuff, which makes it rather complicated.
25 * The header definitions are all context-insensitive because they have
26 * delimited arguments, whereas the message headers are (semi-) context-
27 * sensitive and the messages themselves are, well, RFC82[12] delimited.
28 * This mixture seems to originate from the time that ms and ibm were
29 * good friends and developing os/2 according to the "compatibility"
30 * switch and reading some comments here and there.
32 * I'll ignore most of the complications and concentrate on the concept
33 * which allows me to use yacc. Basically, everything is context-
34 * insensitive now, with the exception of the message-text itself and
35 * the preceding language declaration.
41 #include "config.h"
43 #include <stdio.h>
44 #include <stdlib.h>
45 #include <assert.h>
47 #include "wmc.h"
48 #include "utils.h"
49 #include "lang.h"
51 static const char err_syntax[] = "Syntax error\n";
52 static const char err_number[] = "Number expected\n";
53 static const char err_ident[] = "Identifier expected\n";
54 static const char err_assign[] = "'=' expected\n";
55 static const char err_popen[] = "'(' expected\n";
56 static const char err_pclose[] = "')' expected\n";
57 static const char err_colon[] = "':' expected\n";
58 static const char err_msg[] = "Message expected\n";
60 /* Scanner switches */
61 int want_nl = 0; /* Request next newlinw */
62 int want_line = 0; /* Request next complete line */
63 int want_file = 0; /* Request next ident as filename */
65 struct node *nodehead = NULL; /* The list of all parsed elements */
66 static struct node *nodetail;
67 struct lan_blk *lanblockhead; /* List of parsed elements transposed */
69 static int base = 16; /* Current printout base to use (8, 10 or 16) */
70 static WCHAR *cast = NULL; /* Current typecast to use */
72 static int last_id = 0; /* The last message ID parsed */
73 static int last_sev = 0; /* Last severity code parsed */
74 static int last_fac = 0; /* Last facility code parsed */
75 static WCHAR *last_sym = NULL;/* Last alias symbol parsed */
76 static int have_sev; /* Set if severity parsed for current message */
77 static int have_fac; /* Set if facility parsed for current message */
78 static int have_sym; /* Set is symbol parsed for current message */
80 static struct cp_xlat *cpxlattab = NULL; /* Codepage translation table */
81 static int ncpxlattab = 0;
83 /* Prototypes */
84 static WCHAR *merge(WCHAR *s1, WCHAR *s2);
85 static struct lanmsg *new_lanmsg(struct lan_cp *lcp, WCHAR *msg);
86 static struct msg *add_lanmsg(struct msg *msg, struct lanmsg *lanmsg);
87 static struct msg *complete_msg(struct msg *msg, int id);
88 static void add_node(enum node_type type, void *p);
89 static void do_add_token(enum tok_enum type, struct token *tok, const char *code);
90 static void test_id(int id);
91 static int check_languages(struct node *head);
92 static struct lan_blk *block_messages(struct node *head);
93 static void add_cpxlat(int lan, int cpin, int cpout);
94 static struct cp_xlat *find_cpxlat(int lan);
98 %define api.prefix {mcy_}
100 %union {
101 WCHAR *str;
102 unsigned num;
103 struct token *tok;
104 struct lanmsg *lmp;
105 struct msg *msg;
106 struct lan_cp lcp;
110 %token tSEVNAMES tFACNAMES tLANNAMES tBASE tCODEPAGE
111 %token tTYPEDEF tNL tSYMNAME tMSGEND
112 %token tSEVERITY tFACILITY tLANGUAGE tMSGID
113 %token <str> tIDENT tLINE tFILE tCOMMENT
114 %token <num> tNUMBER
115 %token <tok> tTOKEN
117 %type <str> alias lines
118 %type <num> optcp id msgid clan
119 %type <tok> token
120 %type <lmp> body
121 %type <msg> bodies msg
122 %type <lcp> lang
125 file : items {
126 if(!check_languages(nodehead))
127 xyyerror("No messages defined\n");
128 lanblockhead = block_messages(nodehead);
132 items : decl
133 | items decl
136 decl : global
137 | msg { add_node(nd_msg, $1); }
138 | tCOMMENT { add_node(nd_comment, $1); }
139 | error { xyyerror(err_syntax); /* `Catch all' error */ }
142 global : tSEVNAMES '=' '(' smaps ')'
143 | tSEVNAMES '=' '(' smaps error { xyyerror(err_pclose); }
144 | tSEVNAMES '=' error { xyyerror(err_popen); }
145 | tSEVNAMES error { xyyerror(err_assign); }
146 | tFACNAMES '=' '(' fmaps ')'
147 | tFACNAMES '=' '(' fmaps error { xyyerror(err_pclose); }
148 | tFACNAMES '=' error { xyyerror(err_popen); }
149 | tFACNAMES error { xyyerror(err_assign); }
150 | tLANNAMES '=' '(' lmaps ')'
151 | tLANNAMES '=' '(' lmaps error { xyyerror(err_pclose); }
152 | tLANNAMES '=' error { xyyerror(err_popen); }
153 | tLANNAMES error { xyyerror(err_assign); }
154 | tCODEPAGE '=' '(' cmaps ')'
155 | tCODEPAGE '=' '(' cmaps error { xyyerror(err_pclose); }
156 | tCODEPAGE '=' error { xyyerror(err_popen); }
157 | tCODEPAGE error { xyyerror(err_assign); }
158 | tTYPEDEF '=' tIDENT { cast = $3; }
159 | tTYPEDEF '=' error { xyyerror(err_number); }
160 | tTYPEDEF error { xyyerror(err_assign); }
161 | tBASE '=' tNUMBER {
162 switch(base)
164 case 8:
165 case 10:
166 case 16:
167 base = $3;
168 break;
169 default:
170 xyyerror("Numberbase must be 8, 10 or 16\n");
173 | tBASE '=' error { xyyerror(err_number); }
174 | tBASE error { xyyerror(err_assign); }
177 /*----------------------------------------------------------------------
178 * SeverityNames mapping
180 smaps : smap
181 | smaps smap
182 | error { xyyerror(err_ident); }
185 smap : token '=' tNUMBER alias {
186 $1->token = $3;
187 $1->alias = $4;
188 if($3 & (~0x3))
189 xyyerror("Severity value out of range (0x%08x > 0x3)\n", $3);
190 do_add_token(tok_severity, $1, "severity");
192 | token '=' error { xyyerror(err_number); }
193 | token error { xyyerror(err_assign); }
196 /*----------------------------------------------------------------------
197 * FacilityNames mapping
199 fmaps : fmap
200 | fmaps fmap
201 | error { xyyerror(err_ident); }
204 fmap : token '=' tNUMBER alias {
205 $1->token = $3;
206 $1->alias = $4;
207 if($3 & (~0xfff))
208 xyyerror("Facility value out of range (0x%08x > 0xfff)\n", $3);
209 do_add_token(tok_facility, $1, "facility");
211 | token '=' error { xyyerror(err_number); }
212 | token error { xyyerror(err_assign); }
215 alias : /* Empty */ { $$ = NULL; }
216 | ':' tIDENT { $$ = $2; }
217 | ':' error { xyyerror(err_ident); }
220 /*----------------------------------------------------------------------
221 * LanguageNames mapping
223 lmaps : lmap
224 | lmaps lmap
225 | error { xyyerror(err_ident); }
228 lmap : token '=' tNUMBER setfile ':' tFILE optcp {
229 $1->token = $3;
230 $1->alias = $6;
231 $1->codepage = $7;
232 do_add_token(tok_language, $1, "language");
233 if(!find_language($3) && !find_cpxlat($3))
234 mcy_warning("Language 0x%x not built-in, using codepage %d; use explicit codepage to override\n", $3, WMC_DEFAULT_CODEPAGE);
236 | token '=' tNUMBER setfile ':' error { xyyerror("Filename expected\n"); }
237 | token '=' tNUMBER error { xyyerror(err_colon); }
238 | token '=' error { xyyerror(err_number); }
239 | token error { xyyerror(err_assign); }
242 optcp : /* Empty */ { $$ = 0; }
243 | ':' tNUMBER { $$ = $2; }
244 | ':' error { xyyerror("Codepage-number expected\n"); }
247 /*----------------------------------------------------------------------
248 * Codepages mapping
250 cmaps : cmap
251 | cmaps cmap
252 | error { xyyerror(err_ident); }
255 cmap : clan '=' tNUMBER ':' tNUMBER {
256 static const char err_nocp[] = "Codepage %d not builtin; cannot convert\n";
257 if(find_cpxlat($1))
258 xyyerror("Codepage translation already defined for language 0x%x\n", $1);
259 if($3 && !is_valid_codepage($3))
260 xyyerror(err_nocp, $3);
261 if($5 && !is_valid_codepage($5))
262 xyyerror(err_nocp, $5);
263 add_cpxlat($1, $3, $5);
265 | clan '=' tNUMBER ':' error { xyyerror(err_number); }
266 | clan '=' tNUMBER error { xyyerror(err_colon); }
267 | clan '=' error { xyyerror(err_number); }
268 | clan error { xyyerror(err_assign); }
271 clan : tNUMBER { $$ = $1; }
272 | tTOKEN {
273 if($1->type != tok_language)
274 xyyerror("Language name or code expected\n");
275 $$ = $1->token;
279 /*----------------------------------------------------------------------
280 * Message-definition parsing
282 msg : msgid sevfacsym { test_id($1); } bodies { $$ = complete_msg($4, $1); }
285 msgid : tMSGID '=' id {
286 if($3 & (~0xffff))
287 xyyerror("Message ID value out of range (0x%08x > 0xffff)\n", $3);
288 $$ = $3;
290 | tMSGID error { xyyerror(err_assign); }
293 id : /* Empty */ { $$ = ++last_id; }
294 | tNUMBER { $$ = last_id = $1; }
295 | '+' tNUMBER { $$ = last_id += $2; }
296 | '+' error { xyyerror(err_number); }
299 sevfacsym: /* Empty */ { have_sev = have_fac = have_sym = 0; }
300 | sevfacsym sev { if(have_sev) xyyerror("Severity already defined\n"); have_sev = 1; }
301 | sevfacsym fac { if(have_fac) xyyerror("Facility already defined\n"); have_fac = 1; }
302 | sevfacsym sym { if(have_sym) xyyerror("Symbolname already defined\n"); have_sym = 1; }
305 sym : tSYMNAME '=' tIDENT { last_sym = $3; }
306 | tSYMNAME '=' error { xyyerror(err_ident); }
307 | tSYMNAME error { xyyerror(err_assign); }
310 sev : tSEVERITY '=' token {
311 struct token *tok = lookup_token($3->name);
312 if(!tok)
313 xyyerror("Undefined severityname\n");
314 if(tok->type != tok_severity)
315 xyyerror("Identifier is not of class 'severity'\n");
316 last_sev = tok->token;
318 | tSEVERITY '=' error { xyyerror(err_ident); }
319 | tSEVERITY error { xyyerror(err_assign); }
322 fac : tFACILITY '=' token {
323 struct token *tok = lookup_token($3->name);
324 if(!tok)
325 xyyerror("Undefined facilityname\n");
326 if(tok->type != tok_facility)
327 xyyerror("Identifier is not of class 'facility'\n");
328 last_fac = tok->token;
330 | tFACILITY '=' error { xyyerror(err_ident); }
331 | tFACILITY error { xyyerror(err_assign); }
334 /*----------------------------------------------------------------------
335 * Message-text parsing
337 bodies : body { $$ = add_lanmsg(NULL, $1); }
338 | bodies body { $$ = add_lanmsg($1, $2); }
339 | error { xyyerror("'Language=...' (start of message text-definition) expected\n"); }
342 body : lang setline lines tMSGEND { $$ = new_lanmsg(&$1, $3); }
346 * The newline is to be able to set the codepage
347 * to the language based codepage for the next
348 * message to be parsed.
350 lang : tLANGUAGE setnl '=' token tNL {
351 struct token *tok = lookup_token($4->name);
352 struct cp_xlat *cpx;
353 if(!tok)
354 xyyerror("Undefined language\n");
355 if(tok->type != tok_language)
356 xyyerror("Identifier is not of class 'language'\n");
357 if((cpx = find_cpxlat(tok->token)))
359 set_codepage($$.codepage = cpx->cpin);
361 else if(!tok->codepage)
363 const struct language *lan = find_language(tok->token);
364 if(!lan)
366 /* Just set default; warning was given while parsing languagenames */
367 set_codepage($$.codepage = WMC_DEFAULT_CODEPAGE);
369 else
371 /* The default seems to be to use the DOS codepage... */
372 set_codepage($$.codepage = lan->doscp);
375 else
376 set_codepage($$.codepage = tok->codepage);
377 $$.language = tok->token;
379 | tLANGUAGE setnl '=' token error { xyyerror("Missing newline\n"); }
380 | tLANGUAGE setnl '=' error { xyyerror(err_ident); }
381 | tLANGUAGE error { xyyerror(err_assign); }
384 lines : tLINE { $$ = $1; }
385 | lines tLINE { $$ = merge($1, $2); }
386 | error { xyyerror(err_msg); }
387 | lines error { xyyerror(err_msg); }
390 /*----------------------------------------------------------------------
391 * Helper rules
393 token : tIDENT { $$ = xmalloc(sizeof(struct token)); memset($$,0,sizeof(*$$)); $$->name = $1; }
394 | tTOKEN { $$ = $1; }
397 setnl : /* Empty */ { want_nl = 1; }
400 setline : /* Empty */ { want_line = 1; }
403 setfile : /* Empty */ { want_file = 1; }
408 static WCHAR *merge(WCHAR *s1, WCHAR *s2)
410 int l1 = unistrlen(s1);
411 int l2 = unistrlen(s2);
412 s1 = xrealloc(s1, (l1 + l2 + 1) * sizeof(*s1));
413 unistrcpy(s1+l1, s2);
414 free(s2);
415 return s1;
418 static void do_add_token(enum tok_enum type, struct token *tok, const char *code)
420 struct token *tp = lookup_token(tok->name);
421 if(tp)
423 if(tok->type != type)
424 mcy_warning("Type change in token\n");
425 if(tp != tok)
426 xyyerror("Overlapping token not the same\n");
427 /* else its already defined and changed */
428 if(tok->fixed)
429 xyyerror("Redefinition of %s\n", code);
430 tok->fixed = 1;
432 else
434 add_token(type, tok->name, tok->token, tok->codepage, tok->alias, 1);
435 free(tok);
439 static struct lanmsg *new_lanmsg(struct lan_cp *lcp, WCHAR *msg)
441 struct lanmsg *lmp = xmalloc(sizeof(*lmp));
442 lmp->lan = lcp->language;
443 lmp->cp = lcp->codepage;
444 lmp->msg = msg;
445 lmp->len = unistrlen(msg) + 1; /* Include termination */
446 lmp->file = input_name;
447 lmp->line = line_number;
448 if(lmp->len > 4096)
449 mcy_warning("Message exceptionally long; might be a missing termination\n");
450 return lmp;
453 static struct msg *add_lanmsg(struct msg *msg, struct lanmsg *lanmsg)
455 int i;
456 if(!msg)
458 msg = xmalloc(sizeof(*msg));
459 memset( msg, 0, sizeof(*msg) );
461 msg->msgs = xrealloc(msg->msgs, (msg->nmsgs+1) * sizeof(*(msg->msgs)));
462 msg->msgs[msg->nmsgs] = lanmsg;
463 msg->nmsgs++;
464 for(i = 0; i < msg->nmsgs-1; i++)
466 if(msg->msgs[i]->lan == lanmsg->lan)
467 xyyerror("Message for language 0x%x already defined\n", lanmsg->lan);
469 return msg;
472 static int sort_lanmsg(const void *p1, const void *p2)
474 return (*(const struct lanmsg * const *)p1)->lan - (*(const struct lanmsg * const*)p2)->lan;
477 static struct msg *complete_msg(struct msg *mp, int id)
479 assert(mp != NULL);
480 mp->id = id;
481 if(have_sym)
482 mp->sym = last_sym;
483 else
484 xyyerror("No symbolic name defined for message id %d\n", id);
485 mp->sev = last_sev;
486 mp->fac = last_fac;
487 qsort(mp->msgs, mp->nmsgs, sizeof(*(mp->msgs)), sort_lanmsg);
488 mp->realid = id | (last_sev << 30) | (last_fac << 16);
489 if(custombit)
490 mp->realid |= 1 << 29;
491 mp->base = base;
492 mp->cast = cast;
493 return mp;
496 static void add_node(enum node_type type, void *p)
498 struct node *ndp = xmalloc(sizeof(*ndp));
499 memset( ndp, 0, sizeof(*ndp) );
500 ndp->type = type;
501 ndp->u.all = p;
503 if(nodetail)
505 ndp->prev = nodetail;
506 nodetail->next = ndp;
507 nodetail = ndp;
509 else
511 nodehead = nodetail = ndp;
515 static void test_id(int id)
517 struct node *ndp;
518 for(ndp = nodehead; ndp; ndp = ndp->next)
520 if(ndp->type != nd_msg)
521 continue;
522 if(ndp->u.msg->id == id && ndp->u.msg->sev == last_sev && ndp->u.msg->fac == last_fac)
523 xyyerror("MessageId %d with facility 0x%x and severity 0x%x already defined\n", id, last_fac, last_sev);
527 static int check_languages(struct node *head)
529 static const char err_missing[] = "Missing definition for language 0x%x; MessageID %d, facility 0x%x, severity 0x%x\n";
530 struct node *ndp;
531 int nm = 0;
532 struct msg *msg = NULL;
534 for(ndp = head; ndp; ndp = ndp->next)
536 if(ndp->type != nd_msg)
537 continue;
538 if(!nm)
540 msg = ndp->u.msg;
542 else
544 int i;
545 struct msg *m1;
546 struct msg *m2;
547 if(ndp->u.msg->nmsgs > msg->nmsgs)
549 m1 = ndp->u.msg;
550 m2 = msg;
552 else
554 m1 = msg;
555 m2 = ndp->u.msg;
558 for(i = 0; i < m1->nmsgs; i++)
560 if(i > m2->nmsgs)
561 error(err_missing, m1->msgs[i]->lan, m2->id, m2->fac, m2->sev);
562 else if(m1->msgs[i]->lan < m2->msgs[i]->lan)
563 error(err_missing, m1->msgs[i]->lan, m2->id, m2->fac, m2->sev);
564 else if(m1->msgs[i]->lan > m2->msgs[i]->lan)
565 error(err_missing, m2->msgs[i]->lan, m1->id, m1->fac, m1->sev);
568 nm++;
570 return nm;
573 #define MSGRID(x) ((*(const struct msg * const*)(x))->realid)
574 static int sort_msg(const void *p1, const void *p2)
576 return MSGRID(p1) > MSGRID(p2) ? 1 : (MSGRID(p1) == MSGRID(p2) ? 0 : -1);
580 * block_messages() basically transposes the messages
581 * from ID/language based list to a language/ID
582 * based list.
584 static struct lan_blk *block_messages(struct node *head)
586 struct lan_blk *lbp;
587 struct lan_blk *lblktail = NULL;
588 struct lan_blk *lblkhead = NULL;
589 struct msg **msgtab = NULL;
590 struct node *ndp;
591 int nmsg = 0;
592 int i;
593 int nl;
594 int factor = 2;
596 for(ndp = head; ndp; ndp = ndp->next)
598 if(ndp->type != nd_msg)
599 continue;
600 msgtab = xrealloc(msgtab, (nmsg+1) * sizeof(*msgtab));
601 msgtab[nmsg++] = ndp->u.msg;
604 assert(nmsg != 0);
605 qsort(msgtab, nmsg, sizeof(*msgtab), sort_msg);
607 for(nl = 0; nl < msgtab[0]->nmsgs; nl++) /* This should be equal for all after check_languages() */
609 lbp = xmalloc(sizeof(*lbp));
610 memset( lbp, 0, sizeof(*lbp) );
611 if(!lblktail)
613 lblkhead = lblktail = lbp;
615 else
617 lblktail->next = lbp;
618 lbp->prev = lblktail;
619 lblktail = lbp;
621 lbp->nblk = 1;
622 lbp->blks = xmalloc(sizeof(*lbp->blks));
623 lbp->blks[0].idlo = msgtab[0]->realid;
624 lbp->blks[0].idhi = msgtab[0]->realid;
625 /* The plus 4 is the entry header; (+3)&~3 is DWORD alignment */
626 lbp->blks[0].size = ((factor * msgtab[0]->msgs[nl]->len + 3) & ~3) + 4;
627 lbp->blks[0].msgs = xmalloc(sizeof(*lbp->blks[0].msgs));
628 lbp->blks[0].nmsg = 1;
629 lbp->blks[0].msgs[0] = msgtab[0]->msgs[nl];
630 lbp->lan = msgtab[0]->msgs[nl]->lan;
632 for(i = 1; i < nmsg; i++)
634 struct block *blk = &(lbp->blks[lbp->nblk-1]);
635 if(msgtab[i]->realid == blk->idhi+1)
637 blk->size += ((factor * msgtab[i]->msgs[nl]->len + 3) & ~3) + 4;
638 blk->idhi++;
639 blk->msgs = xrealloc(blk->msgs, (blk->nmsg+1) * sizeof(*blk->msgs));
640 blk->msgs[blk->nmsg++] = msgtab[i]->msgs[nl];
642 else
644 lbp->nblk++;
645 lbp->blks = xrealloc(lbp->blks, lbp->nblk * sizeof(*lbp->blks));
646 blk = &(lbp->blks[lbp->nblk-1]);
647 blk->idlo = msgtab[i]->realid;
648 blk->idhi = msgtab[i]->realid;
649 blk->size = ((factor * msgtab[i]->msgs[nl]->len + 3) & ~3) + 4;
650 blk->msgs = xmalloc(sizeof(*blk->msgs));
651 blk->nmsg = 1;
652 blk->msgs[0] = msgtab[i]->msgs[nl];
656 free(msgtab);
657 return lblkhead;
660 static int sc_xlat(const void *p1, const void *p2)
662 return ((const struct cp_xlat *)p1)->lan - ((const struct cp_xlat *)p2)->lan;
665 static void add_cpxlat(int lan, int cpin, int cpout)
667 cpxlattab = xrealloc(cpxlattab, (ncpxlattab+1) * sizeof(*cpxlattab));
668 cpxlattab[ncpxlattab].lan = lan;
669 cpxlattab[ncpxlattab].cpin = cpin;
670 cpxlattab[ncpxlattab].cpout = cpout;
671 ncpxlattab++;
672 qsort(cpxlattab, ncpxlattab, sizeof(*cpxlattab), sc_xlat);
675 static struct cp_xlat *find_cpxlat(int lan)
677 struct cp_xlat t;
679 if(!cpxlattab) return NULL;
681 t.lan = lan;
682 return (struct cp_xlat *)bsearch(&t, cpxlattab, ncpxlattab, sizeof(*cpxlattab), sc_xlat);