d7482997 |
1 | /* |
2 | * input.c: read the source form |
3 | */ |
4 | |
5 | #include <stdio.h> |
6 | #include <assert.h> |
7 | #include <time.h> |
8 | #include "halibut.h" |
9 | |
10 | #define TAB_STOP 8 /* for column number tracking */ |
11 | |
12 | static void setpos(input *in, char *fname) { |
13 | in->pos.filename = fname; |
14 | in->pos.line = 1; |
15 | in->pos.col = (in->reportcols ? 1 : -1); |
16 | } |
17 | |
18 | static void unget(input *in, int c, filepos *pos) { |
19 | if (in->npushback >= in->pushbacksize) { |
20 | in->pushbacksize = in->npushback + 16; |
f1530049 |
21 | in->pushback = sresize(in->pushback, in->pushbacksize, pushback); |
d7482997 |
22 | } |
23 | in->pushback[in->npushback].chr = c; |
24 | in->pushback[in->npushback].pos = *pos; /* structure copy */ |
25 | in->npushback++; |
26 | } |
27 | |
28 | /* ---------------------------------------------------------------------- */ |
29 | /* |
30 | * Macro subsystem |
31 | */ |
32 | typedef struct macro_Tag macro; |
33 | struct macro_Tag { |
34 | wchar_t *name, *text; |
35 | }; |
36 | struct macrostack_Tag { |
37 | macrostack *next; |
38 | wchar_t *text; |
39 | int ptr, npushback; |
40 | filepos pos; |
41 | }; |
42 | static int macrocmp(void *av, void *bv) { |
43 | macro *a = (macro *)av, *b = (macro *)bv; |
44 | return ustrcmp(a->name, b->name); |
45 | } |
46 | static void macrodef(tree234 *macros, wchar_t *name, wchar_t *text, |
47 | filepos fpos) { |
f1530049 |
48 | macro *m = snew(macro); |
d7482997 |
49 | m->name = name; |
50 | m->text = text; |
51 | if (add234(macros, m) != m) { |
52 | error(err_macroexists, &fpos, name); |
53 | sfree(name); |
54 | sfree(text); |
55 | } |
56 | } |
57 | static int macrolookup(tree234 *macros, input *in, wchar_t *name, |
58 | filepos *pos) { |
59 | macro m, *gotit; |
60 | m.name = name; |
61 | gotit = find234(macros, &m, NULL); |
62 | if (gotit) { |
f1530049 |
63 | macrostack *expansion = snew(macrostack); |
d7482997 |
64 | expansion->next = in->stack; |
65 | expansion->text = gotit->text; |
66 | expansion->pos = *pos; /* structure copy */ |
67 | expansion->ptr = 0; |
68 | expansion->npushback = in->npushback; |
69 | in->stack = expansion; |
70 | return TRUE; |
71 | } else |
72 | return FALSE; |
73 | } |
74 | static void macrocleanup(tree234 *macros) { |
75 | int ti; |
76 | macro *m; |
77 | for (ti = 0; (m = (macro *)index234(macros, ti)) != NULL; ti++) { |
78 | sfree(m->name); |
79 | sfree(m->text); |
80 | sfree(m); |
81 | } |
82 | freetree234(macros); |
83 | } |
84 | |
e34ba5c3 |
85 | static void input_configure(input *in, paragraph *cfg) { |
86 | assert(cfg->type == para_Config); |
87 | |
88 | if (!ustricmp(cfg->keyword, L"input-charset")) { |
0960a3d8 |
89 | in->charset = charset_from_ustr(&cfg->fpos, uadv(cfg->keyword)); |
e34ba5c3 |
90 | } |
91 | } |
92 | |
d7482997 |
93 | /* |
94 | * Can return EOF |
95 | */ |
e4ea58f8 |
96 | static int get(input *in, filepos *pos, rdstringc *rsc) { |
d7482997 |
97 | int pushbackpt = in->stack ? in->stack->npushback : 0; |
98 | if (in->npushback > pushbackpt) { |
99 | --in->npushback; |
100 | if (pos) |
101 | *pos = in->pushback[in->npushback].pos; /* structure copy */ |
102 | return in->pushback[in->npushback].chr; |
103 | } |
104 | else if (in->stack) { |
105 | wchar_t c = in->stack->text[in->stack->ptr]; |
106 | if (in->stack->text[++in->stack->ptr] == L'\0') { |
107 | macrostack *tmp = in->stack; |
108 | in->stack = tmp->next; |
109 | sfree(tmp); |
110 | } |
111 | return c; |
112 | } |
113 | else if (in->currfp) { |
d7482997 |
114 | |
e34ba5c3 |
115 | while (in->wcpos >= in->nwc) { |
116 | |
117 | int c = getc(in->currfp); |
118 | |
119 | if (c == EOF) { |
120 | fclose(in->currfp); |
121 | in->currfp = NULL; |
122 | return EOF; |
123 | } |
e4ea58f8 |
124 | |
125 | if (rsc) |
126 | rdaddc(rsc, c); |
127 | |
e34ba5c3 |
128 | /* Track line numbers, for error reporting */ |
129 | if (pos) |
130 | *pos = in->pos; |
131 | if (in->reportcols) { |
132 | switch (c) { |
133 | case '\t': |
134 | in->pos.col = 1 + (in->pos.col + TAB_STOP-1) % TAB_STOP; |
135 | break; |
136 | case '\n': |
137 | in->pos.col = 1; |
138 | in->pos.line++; |
139 | break; |
140 | default: |
141 | in->pos.col++; |
142 | break; |
143 | } |
144 | } else { |
145 | in->pos.col = -1; |
146 | if (c == '\n') |
147 | in->pos.line++; |
148 | } |
149 | |
150 | /* |
151 | * Do input character set translation, so that we return |
152 | * Unicode. |
153 | */ |
154 | { |
155 | char buf[1]; |
156 | char const *p; |
157 | int inlen; |
158 | |
159 | buf[0] = (char)c; |
160 | p = buf; |
161 | inlen = 1; |
162 | |
163 | in->nwc = charset_to_unicode(&p, &inlen, |
164 | in->wc, lenof(in->wc), |
165 | in->charset, &in->csstate, |
166 | NULL, 0); |
167 | assert(p == buf+1 && inlen == 0); |
168 | |
169 | in->wcpos = 0; |
d7482997 |
170 | } |
d7482997 |
171 | } |
e34ba5c3 |
172 | |
173 | return in->wc[in->wcpos++]; |
174 | |
d7482997 |
175 | } else |
176 | return EOF; |
177 | } |
178 | |
179 | /* |
180 | * Lexical analysis of source files. |
181 | */ |
182 | typedef struct token_Tag token; |
183 | struct token_Tag { |
184 | int type; |
185 | int cmd, aux; |
186 | wchar_t *text; |
e4ea58f8 |
187 | char *origtext; |
d7482997 |
188 | filepos pos; |
189 | }; |
190 | enum { |
191 | tok_eof, /* end of file */ |
192 | tok_eop, /* end of paragraph */ |
193 | tok_white, /* whitespace */ |
194 | tok_word, /* a word or word fragment */ |
195 | tok_cmd, /* \command */ |
196 | tok_lbrace, /* { */ |
197 | tok_rbrace /* } */ |
198 | }; |
199 | |
200 | /* Halibut command keywords. */ |
201 | enum { |
202 | c__invalid, /* invalid command */ |
203 | c__comment, /* comment command (\#) */ |
204 | c__escaped, /* escaped character */ |
672e4626 |
205 | c__nop, /* no-op */ |
d7482997 |
206 | c__nbsp, /* nonbreaking space */ |
207 | c_A, /* appendix heading */ |
208 | c_B, /* bibliography entry */ |
209 | c_BR, /* bibliography rewrite */ |
210 | c_C, /* chapter heading */ |
211 | c_H, /* heading */ |
212 | c_I, /* invisible index mark */ |
213 | c_IM, /* index merge/rewrite */ |
214 | c_K, /* capitalised cross-reference */ |
215 | c_S, /* aux field is 0, 1, 2, ... */ |
216 | c_U, /* unnumbered-chapter heading */ |
217 | c_W, /* Web hyperlink */ |
218 | c_b, /* bulletted list */ |
219 | c_c, /* code */ |
220 | c_cfg, /* configuration directive */ |
221 | c_copyright, /* copyright statement */ |
b7747050 |
222 | c_cq, /* quoted code (sugar for \q{\cw{x}}) */ |
d7482997 |
223 | c_cw, /* weak code */ |
224 | c_date, /* document processing date */ |
7136a6c7 |
225 | c_dd, /* description list: description */ |
d7482997 |
226 | c_define, /* macro definition */ |
7136a6c7 |
227 | c_dt, /* description list: described thing */ |
d7482997 |
228 | c_e, /* emphasis */ |
229 | c_i, /* visible index mark */ |
230 | c_ii, /* uncapitalised visible index mark */ |
231 | c_k, /* uncapitalised cross-reference */ |
7136a6c7 |
232 | c_lcont, /* continuation para(s) for list item */ |
d7482997 |
233 | c_n, /* numbered list */ |
234 | c_nocite, /* bibliography trickery */ |
8902e0ed |
235 | c_preamble, /* (obsolete) preamble text */ |
d7482997 |
236 | c_q, /* quote marks */ |
2614b01d |
237 | c_quote, /* block-quoted paragraphs */ |
d7482997 |
238 | c_rule, /* horizontal rule */ |
239 | c_title, /* document title */ |
240 | c_u, /* aux field is char code */ |
241 | c_versionid /* document RCS id */ |
242 | }; |
243 | |
244 | /* Perhaps whitespace should be defined in a more Unicode-friendly way? */ |
245 | #define iswhite(c) ( (c)==32 || (c)==9 || (c)==13 || (c)==10 ) |
246 | #define isnl(c) ( (c)==10 ) |
247 | #define isdec(c) ( ((c)>='0'&&(c)<='9') ) |
248 | #define fromdec(c) ( (c)-'0' ) |
249 | #define ishex(c) ( ((c)>='0'&&(c)<='9') || ((c)>='A'&&(c)<='F') || ((c)>='a'&&(c)<='f')) |
250 | #define fromhex(c) ( (c)<='9' ? (c)-'0' : ((c)&0xDF) - ('A'-10) ) |
251 | #define iscmd(c) ( ((c)>='0'&&(c)<='9') || ((c)>='A'&&(c)<='Z') || ((c)>='a'&&(c)<='z')) |
252 | |
253 | /* |
254 | * Keyword comparison function. Like strcmp, but between a wchar_t * |
255 | * and a char *. |
256 | */ |
257 | static int kwcmp(wchar_t const *p, char const *q) { |
258 | int i; |
259 | do { |
260 | i = *p - *q; |
261 | } while (*p++ && *q++ && !i); |
262 | return i; |
263 | } |
264 | |
265 | /* |
266 | * Match a keyword. |
267 | */ |
268 | static void match_kw(token *tok) { |
269 | /* |
270 | * FIXME. The ids are explicit in here so as to allow long-name |
271 | * equivalents to the various very short keywords. |
272 | */ |
273 | static const struct { char const *name; int id; } keywords[] = { |
274 | {"#", c__comment}, /* comment command (\#) */ |
275 | {"-", c__escaped}, /* nonbreaking hyphen */ |
672e4626 |
276 | {".", c__nop}, /* no-op */ |
d7482997 |
277 | {"A", c_A}, /* appendix heading */ |
278 | {"B", c_B}, /* bibliography entry */ |
279 | {"BR", c_BR}, /* bibliography rewrite */ |
280 | {"C", c_C}, /* chapter heading */ |
281 | {"H", c_H}, /* heading */ |
282 | {"I", c_I}, /* invisible index mark */ |
283 | {"IM", c_IM}, /* index merge/rewrite */ |
284 | {"K", c_K}, /* capitalised cross-reference */ |
285 | {"U", c_U}, /* unnumbered-chapter heading */ |
286 | {"W", c_W}, /* Web hyperlink */ |
287 | {"\\", c__escaped}, /* escaped backslash (\\) */ |
288 | {"_", c__nbsp}, /* nonbreaking space (\_) */ |
289 | {"b", c_b}, /* bulletted list */ |
290 | {"c", c_c}, /* code */ |
291 | {"cfg", c_cfg}, /* configuration directive */ |
292 | {"copyright", c_copyright}, /* copyright statement */ |
b7747050 |
293 | {"cq", c_cq}, /* quoted code (sugar for \q{\cw{x}}) */ |
d7482997 |
294 | {"cw", c_cw}, /* weak code */ |
295 | {"date", c_date}, /* document processing date */ |
7136a6c7 |
296 | {"dd", c_dd}, /* description list: description */ |
d7482997 |
297 | {"define", c_define}, /* macro definition */ |
7136a6c7 |
298 | {"dt", c_dt}, /* description list: described thing */ |
d7482997 |
299 | {"e", c_e}, /* emphasis */ |
300 | {"i", c_i}, /* visible index mark */ |
301 | {"ii", c_ii}, /* uncapitalised visible index mark */ |
302 | {"k", c_k}, /* uncapitalised cross-reference */ |
7136a6c7 |
303 | {"lcont", c_lcont}, /* continuation para(s) for list item */ |
d7482997 |
304 | {"n", c_n}, /* numbered list */ |
305 | {"nocite", c_nocite}, /* bibliography trickery */ |
8902e0ed |
306 | {"preamble", c_preamble}, /* (obsolete) preamble text */ |
d7482997 |
307 | {"q", c_q}, /* quote marks */ |
2614b01d |
308 | {"quote", c_quote}, /* block-quoted paragraphs */ |
d7482997 |
309 | {"rule", c_rule}, /* horizontal rule */ |
310 | {"title", c_title}, /* document title */ |
311 | {"versionid", c_versionid}, /* document RCS id */ |
312 | {"{", c__escaped}, /* escaped lbrace (\{) */ |
313 | {"}", c__escaped}, /* escaped rbrace (\}) */ |
314 | }; |
315 | int i, j, k, c; |
316 | |
317 | /* |
318 | * Special cases: \S{0,1,2,...} and \uABCD. If the syntax |
319 | * doesn't match correctly, we just fall through to the |
320 | * binary-search phase. |
321 | */ |
322 | if (tok->text[0] == 'S') { |
323 | /* We expect numeric characters thereafter. */ |
324 | wchar_t *p = tok->text+1; |
325 | int n; |
326 | if (!*p) |
327 | n = 1; |
328 | else { |
329 | n = 0; |
330 | while (*p && isdec(*p)) { |
331 | n = 10 * n + fromdec(*p); |
332 | p++; |
333 | } |
334 | } |
335 | if (!*p) { |
336 | tok->cmd = c_S; |
337 | tok->aux = n; |
338 | return; |
339 | } |
340 | } else if (tok->text[0] == 'u') { |
341 | /* We expect hex characters thereafter. */ |
342 | wchar_t *p = tok->text+1; |
343 | int n = 0; |
344 | while (*p && ishex(*p)) { |
345 | n = 16 * n + fromhex(*p); |
346 | p++; |
347 | } |
348 | if (!*p) { |
349 | tok->cmd = c_u; |
350 | tok->aux = n; |
351 | return; |
352 | } |
353 | } |
354 | |
355 | i = -1; |
356 | j = sizeof(keywords)/sizeof(*keywords); |
357 | while (j-i > 1) { |
358 | k = (i+j)/2; |
359 | c = kwcmp(tok->text, keywords[k].name); |
360 | if (c < 0) |
361 | j = k; |
362 | else if (c > 0) |
363 | i = k; |
364 | else /* c == 0 */ { |
365 | tok->cmd = keywords[k].id; |
366 | return; |
367 | } |
368 | } |
369 | |
370 | tok->cmd = c__invalid; |
371 | } |
372 | |
373 | |
374 | /* |
375 | * Read a token from the input file, in the normal way (`normal' in |
376 | * the sense that code paragraphs work a different way). |
377 | */ |
378 | token get_token(input *in) { |
379 | int c; |
380 | int nls; |
e4ea58f8 |
381 | int prevpos; |
d7482997 |
382 | token ret; |
383 | rdstring rs = { 0, 0, NULL }; |
e4ea58f8 |
384 | rdstringc rsc = { 0, 0, NULL }; |
d7482997 |
385 | filepos cpos; |
386 | |
387 | ret.text = NULL; /* default */ |
e4ea58f8 |
388 | ret.origtext = NULL; /* default */ |
389 | if (in->pushback_chars) { |
390 | rdaddsc(&rsc, in->pushback_chars); |
391 | sfree(in->pushback_chars); |
392 | in->pushback_chars = NULL; |
393 | } |
394 | c = get(in, &cpos, &rsc); |
d7482997 |
395 | ret.pos = cpos; |
396 | if (iswhite(c)) { /* tok_white or tok_eop */ |
397 | nls = 0; |
e4ea58f8 |
398 | prevpos = 0; |
d7482997 |
399 | do { |
400 | if (isnl(c)) |
401 | nls++; |
e4ea58f8 |
402 | prevpos = rsc.pos; |
403 | } while ((c = get(in, &cpos, &rsc)) != EOF && iswhite(c)); |
d7482997 |
404 | if (c == EOF) { |
405 | ret.type = tok_eof; |
e4ea58f8 |
406 | sfree(rsc.text); |
d7482997 |
407 | return ret; |
408 | } |
e4ea58f8 |
409 | if (rsc.text) { |
410 | in->pushback_chars = dupstr(rsc.text + prevpos); |
411 | sfree(rsc.text); |
412 | } |
d7482997 |
413 | unget(in, c, &cpos); |
414 | ret.type = (nls > 1 ? tok_eop : tok_white); |
415 | return ret; |
416 | } else if (c == EOF) { /* tok_eof */ |
417 | ret.type = tok_eof; |
e4ea58f8 |
418 | sfree(rsc.text); |
d7482997 |
419 | return ret; |
420 | } else if (c == '\\') { /* tok_cmd */ |
e4ea58f8 |
421 | rsc.pos = prevpos = 0; |
422 | c = get(in, &cpos, &rsc); |
d7482997 |
423 | if (c == '-' || c == '\\' || c == '_' || |
672e4626 |
424 | c == '#' || c == '{' || c == '}' || c == '.') { |
d7482997 |
425 | /* single-char command */ |
426 | rdadd(&rs, c); |
427 | } else if (c == 'u') { |
428 | int len = 0; |
429 | do { |
430 | rdadd(&rs, c); |
431 | len++; |
e4ea58f8 |
432 | prevpos = rsc.pos; |
433 | c = get(in, &cpos, &rsc); |
d7482997 |
434 | } while (ishex(c) && len < 5); |
435 | unget(in, c, &cpos); |
436 | } else if (iscmd(c)) { |
437 | do { |
438 | rdadd(&rs, c); |
e4ea58f8 |
439 | prevpos = rsc.pos; |
440 | c = get(in, &cpos, &rsc); |
d7482997 |
441 | } while (iscmd(c)); |
442 | unget(in, c, &cpos); |
443 | } |
444 | /* |
445 | * Now match the command against the list of available |
446 | * ones. |
447 | */ |
448 | ret.type = tok_cmd; |
449 | ret.text = ustrdup(rs.text); |
e4ea58f8 |
450 | if (rsc.text) { |
451 | in->pushback_chars = dupstr(rsc.text + prevpos); |
452 | rsc.text[prevpos] = '\0'; |
453 | ret.origtext = dupstr(rsc.text); |
454 | } else { |
455 | ret.origtext = dupstr(""); |
456 | } |
d7482997 |
457 | match_kw(&ret); |
458 | sfree(rs.text); |
e4ea58f8 |
459 | sfree(rsc.text); |
d7482997 |
460 | return ret; |
461 | } else if (c == '{') { /* tok_lbrace */ |
462 | ret.type = tok_lbrace; |
e4ea58f8 |
463 | sfree(rsc.text); |
d7482997 |
464 | return ret; |
465 | } else if (c == '}') { /* tok_rbrace */ |
466 | ret.type = tok_rbrace; |
e4ea58f8 |
467 | sfree(rsc.text); |
d7482997 |
468 | return ret; |
469 | } else { /* tok_word */ |
470 | /* |
471 | * Read a word: the longest possible contiguous sequence of |
472 | * things other than whitespace, backslash, braces and |
473 | * hyphen. A hyphen terminates the word but is returned as |
474 | * part of it; everything else is pushed back for the next |
475 | * token. The `aux' field contains TRUE if the word ends in |
476 | * a hyphen. |
477 | */ |
478 | ret.aux = FALSE; /* assumed for now */ |
e4ea58f8 |
479 | prevpos = 0; |
d7482997 |
480 | while (1) { |
481 | if (iswhite(c) || c=='{' || c=='}' || c=='\\' || c==EOF) { |
482 | /* Put back the character that caused termination */ |
483 | unget(in, c, &cpos); |
484 | break; |
485 | } else { |
486 | rdadd(&rs, c); |
487 | if (c == '-') { |
e4ea58f8 |
488 | prevpos = rsc.pos; |
d7482997 |
489 | ret.aux = TRUE; |
490 | break; /* hyphen terminates word */ |
491 | } |
492 | } |
e4ea58f8 |
493 | prevpos = rsc.pos; |
494 | c = get(in, &cpos, &rsc); |
d7482997 |
495 | } |
496 | ret.type = tok_word; |
497 | ret.text = ustrdup(rs.text); |
e4ea58f8 |
498 | if (rsc.text) { |
499 | in->pushback_chars = dupstr(rsc.text + prevpos); |
500 | rsc.text[prevpos] = '\0'; |
501 | ret.origtext = dupstr(rsc.text); |
502 | } else { |
503 | ret.origtext = dupstr(""); |
504 | } |
d7482997 |
505 | sfree(rs.text); |
e4ea58f8 |
506 | sfree(rsc.text); |
d7482997 |
507 | return ret; |
508 | } |
509 | } |
510 | |
511 | /* |
512 | * Determine whether the next input character is an open brace (for |
513 | * telling code paragraphs from paragraphs which merely start with |
514 | * code). |
515 | */ |
516 | int isbrace(input *in) { |
517 | int c; |
518 | filepos cpos; |
519 | |
e4ea58f8 |
520 | c = get(in, &cpos, NULL); |
d7482997 |
521 | unget(in, c, &cpos); |
522 | return (c == '{'); |
523 | } |
524 | |
525 | /* |
526 | * Read the rest of a line that starts `\c'. Including nothing at |
527 | * all (tok_word with empty text). |
528 | */ |
529 | token get_codepar_token(input *in) { |
530 | int c; |
531 | token ret; |
532 | rdstring rs = { 0, 0, NULL }; |
533 | filepos cpos; |
534 | |
535 | ret.type = tok_word; |
e4ea58f8 |
536 | ret.origtext = NULL; |
537 | c = get(in, &cpos, NULL); /* expect (and discard) one space */ |
d7482997 |
538 | ret.pos = cpos; |
539 | if (c == ' ') { |
e4ea58f8 |
540 | c = get(in, &cpos, NULL); |
d7482997 |
541 | ret.pos = cpos; |
542 | } |
543 | while (!isnl(c) && c != EOF) { |
544 | int c2 = c; |
e4ea58f8 |
545 | c = get(in, &cpos, NULL); |
d7482997 |
546 | /* Discard \r just before \n. */ |
547 | if (c2 != 13 || !isnl(c)) |
548 | rdadd(&rs, c2); |
549 | } |
550 | unget(in, c, &cpos); |
551 | ret.text = ustrdup(rs.text); |
552 | sfree(rs.text); |
553 | return ret; |
554 | } |
555 | |
556 | /* |
557 | * Adds a new word to a linked list |
558 | */ |
559 | static word *addword(word newword, word ***hptrptr) { |
560 | word *mnewword; |
561 | if (!hptrptr) |
562 | return NULL; |
f1530049 |
563 | mnewword = snew(word); |
d7482997 |
564 | *mnewword = newword; /* structure copy */ |
565 | mnewword->next = NULL; |
566 | **hptrptr = mnewword; |
567 | *hptrptr = &mnewword->next; |
568 | return mnewword; |
569 | } |
570 | |
571 | /* |
572 | * Adds a new paragraph to a linked list |
573 | */ |
574 | static paragraph *addpara(paragraph newpara, paragraph ***hptrptr) { |
f1530049 |
575 | paragraph *mnewpara = snew(paragraph); |
d7482997 |
576 | *mnewpara = newpara; /* structure copy */ |
577 | mnewpara->next = NULL; |
578 | **hptrptr = mnewpara; |
579 | *hptrptr = &mnewpara->next; |
580 | return mnewpara; |
581 | } |
582 | |
583 | /* |
584 | * Destructor before token is reassigned; should catch most memory |
585 | * leaks |
586 | */ |
e4ea58f8 |
587 | #define dtor(t) ( sfree(t.text), sfree(t.origtext) ) |
d7482997 |
588 | |
589 | /* |
590 | * Reads a single file (ie until get() returns EOF) |
591 | */ |
592 | static void read_file(paragraph ***ret, input *in, indexdata *idx) { |
593 | token t; |
594 | paragraph par; |
595 | word wd, **whptr, **idximplicit; |
596 | tree234 *macros; |
597 | wchar_t utext[2], *wdtext; |
598 | int style, spcstyle; |
599 | int already; |
600 | int iswhite, seenwhite; |
601 | int type; |
7136a6c7 |
602 | int prev_para_type; |
d7482997 |
603 | struct stack_item { |
604 | enum { |
605 | stack_nop = 0, /* do nothing (for error recovery) */ |
606 | stack_ualt = 1, /* \u alternative */ |
607 | stack_style = 2, /* \e, \c, \cw */ |
608 | stack_idx = 4, /* \I, \i, \ii */ |
609 | stack_hyper = 8, /* \W */ |
610 | stack_quote = 16, /* \q */ |
611 | } type; |
612 | word **whptr; /* to restore from \u alternatives */ |
613 | word **idximplicit; /* to restore from \u alternatives */ |
f4551933 |
614 | filepos fpos; |
6ff15f2b |
615 | int in_code; |
d7482997 |
616 | } *sitem; |
617 | stack parsestk; |
7136a6c7 |
618 | struct crossparaitem { |
2614b01d |
619 | int type; /* currently c_lcont, c_quote or -1 */ |
620 | int seen_lcont, seen_quote; |
7136a6c7 |
621 | }; |
622 | stack crossparastk; |
d7482997 |
623 | word *indexword, *uword, *iword; |
624 | word *idxwordlist; |
625 | rdstring indexstr; |
626 | int index_downcase, index_visible, indexing; |
627 | const rdstring nullrs = { 0, 0, NULL }; |
628 | wchar_t uchr; |
629 | |
630 | t.text = NULL; |
e4ea58f8 |
631 | t.origtext = NULL; |
d7482997 |
632 | macros = newtree234(macrocmp); |
b3c80b25 |
633 | already = FALSE; |
d7482997 |
634 | |
7136a6c7 |
635 | crossparastk = stk_new(); |
636 | |
d7482997 |
637 | /* |
638 | * Loop on each paragraph. |
639 | */ |
640 | while (1) { |
b3c80b25 |
641 | int start_cmd = c__invalid; |
d7482997 |
642 | par.words = NULL; |
643 | par.keyword = NULL; |
e4ea58f8 |
644 | par.origkeyword = NULL; |
d7482997 |
645 | whptr = &par.words; |
646 | |
647 | /* |
648 | * Get a token. |
649 | */ |
7136a6c7 |
650 | do { |
651 | if (!already) { |
652 | dtor(t), t = get_token(in); |
653 | } |
654 | already = FALSE; |
655 | } while (t.type == tok_eop); |
d7482997 |
656 | if (t.type == tok_eof) |
677e18a2 |
657 | break; |
d7482997 |
658 | |
659 | /* |
660 | * Parse code paragraphs separately. |
661 | */ |
662 | if (t.type == tok_cmd && t.cmd == c_c && !isbrace(in)) { |
4b3c5afb |
663 | int wtype = word_WeakCode; |
664 | |
d7482997 |
665 | par.type = para_Code; |
666 | par.fpos = t.pos; |
667 | while (1) { |
668 | dtor(t), t = get_codepar_token(in); |
4b3c5afb |
669 | wd.type = wtype; |
d7482997 |
670 | wd.breaks = FALSE; /* shouldn't need this... */ |
671 | wd.text = ustrdup(t.text); |
672 | wd.alt = NULL; |
673 | wd.fpos = t.pos; |
674 | addword(wd, &whptr); |
675 | dtor(t), t = get_token(in); |
676 | if (t.type == tok_white) { |
677 | /* |
678 | * The newline after a code-paragraph line |
679 | */ |
680 | dtor(t), t = get_token(in); |
681 | } |
780f45d4 |
682 | if (t.type == tok_eop || t.type == tok_eof || |
683 | t.type == tok_rbrace) { /* might be } terminating \lcont */ |
684 | if (t.type == tok_rbrace) |
685 | already = TRUE; |
d7482997 |
686 | break; |
780f45d4 |
687 | } else if (t.type == tok_cmd && t.cmd == c_c) { |
4b3c5afb |
688 | wtype = word_WeakCode; |
780f45d4 |
689 | } else if (t.type == tok_cmd && t.cmd == c_e && |
690 | wtype == word_WeakCode) { |
4b3c5afb |
691 | wtype = word_Emph; |
780f45d4 |
692 | } else { |
d7482997 |
693 | error(err_brokencodepara, &t.pos); |
7136a6c7 |
694 | prev_para_type = par.type; |
d7482997 |
695 | addpara(par, ret); |
696 | while (t.type != tok_eop) /* error recovery: */ |
697 | dtor(t), t = get_token(in); /* eat rest of paragraph */ |
698 | goto codeparabroken; /* ick, but such is life */ |
699 | } |
700 | } |
7136a6c7 |
701 | prev_para_type = par.type; |
d7482997 |
702 | addpara(par, ret); |
703 | codeparabroken: |
704 | continue; |
705 | } |
706 | |
707 | /* |
7136a6c7 |
708 | * Spot the special commands that define a grouping of more |
709 | * than one paragraph, and also the closing braces that |
710 | * finish them. |
711 | */ |
712 | if (t.type == tok_cmd && |
2614b01d |
713 | (t.cmd == c_lcont || t.cmd == c_quote)) { |
7136a6c7 |
714 | struct crossparaitem *sitem, *stop; |
2614b01d |
715 | int cmd = t.cmd; |
7136a6c7 |
716 | |
717 | /* |
718 | * Expect, and swallow, an open brace. |
719 | */ |
720 | dtor(t), t = get_token(in); |
721 | if (t.type != tok_lbrace) { |
722 | error(err_explbr, &t.pos); |
723 | continue; |
724 | } |
725 | |
780f45d4 |
726 | /* |
727 | * Also expect, and swallow, any whitespace after that |
728 | * (a newline before a code paragraph wouldn't be |
729 | * surprising). |
730 | */ |
731 | do { |
732 | dtor(t), t = get_token(in); |
733 | } while (t.type == tok_white); |
734 | already = TRUE; |
735 | |
2614b01d |
736 | if (cmd == c_lcont) { |
737 | /* |
738 | * \lcont causes a continuation of a list item into |
739 | * multiple paragraphs (which may in turn contain |
740 | * nested lists, code paras etc). Hence, the previous |
741 | * paragraph must be of a list type. |
742 | */ |
f1530049 |
743 | sitem = snew(struct crossparaitem); |
2614b01d |
744 | stop = (struct crossparaitem *)stk_top(crossparastk); |
745 | if (stop) |
746 | *sitem = *stop; |
747 | else |
748 | sitem->seen_quote = sitem->seen_lcont = 0; |
749 | |
750 | if (prev_para_type == para_Bullet || |
751 | prev_para_type == para_NumberedList || |
752 | prev_para_type == para_Description) { |
753 | sitem->type = c_lcont; |
754 | sitem->seen_lcont = 1; |
755 | par.type = para_LcontPush; |
756 | prev_para_type = par.type; |
757 | addpara(par, ret); |
758 | } else { |
759 | /* |
760 | * Push a null item on the cross-para stack so that |
761 | * when we see the corresponding closing brace we |
762 | * don't give a cascade error. |
763 | */ |
764 | sitem->type = -1; |
765 | error(err_misplacedlcont, &t.pos); |
766 | } |
7136a6c7 |
767 | } else { |
768 | /* |
2614b01d |
769 | * \quote causes a group of paragraphs to be |
770 | * block-quoted (typically they will be indented a |
771 | * bit). |
7136a6c7 |
772 | */ |
f1530049 |
773 | sitem = snew(struct crossparaitem); |
2614b01d |
774 | stop = (struct crossparaitem *)stk_top(crossparastk); |
775 | if (stop) |
776 | *sitem = *stop; |
777 | else |
778 | sitem->seen_quote = sitem->seen_lcont = 0; |
779 | sitem->type = c_quote; |
780 | sitem->seen_quote = 1; |
781 | par.type = para_QuotePush; |
782 | prev_para_type = par.type; |
783 | addpara(par, ret); |
7136a6c7 |
784 | } |
785 | stk_push(crossparastk, sitem); |
786 | continue; |
787 | } else if (t.type == tok_rbrace) { |
788 | struct crossparaitem *sitem = stk_pop(crossparastk); |
789 | if (!sitem) |
790 | error(err_unexbrace, &t.pos); |
791 | else { |
792 | switch (sitem->type) { |
793 | case c_lcont: |
794 | par.type = para_LcontPop; |
795 | prev_para_type = par.type; |
796 | addpara(par, ret); |
797 | break; |
2614b01d |
798 | case c_quote: |
799 | par.type = para_QuotePop; |
800 | prev_para_type = par.type; |
801 | addpara(par, ret); |
802 | break; |
7136a6c7 |
803 | } |
804 | sfree(sitem); |
805 | } |
806 | continue; |
807 | } |
808 | |
809 | /* |
d7482997 |
810 | * This token begins a paragraph. See if it's one of the |
811 | * special commands that define a paragraph type. |
812 | * |
813 | * (note that \# is special in a way, and \nocite takes no |
814 | * text) |
815 | */ |
816 | par.type = para_Normal; |
817 | if (t.type == tok_cmd) { |
818 | int needkw; |
819 | int is_macro = FALSE; |
820 | |
821 | par.fpos = t.pos; |
822 | switch (t.cmd) { |
823 | default: |
824 | needkw = -1; |
825 | break; |
826 | case c__invalid: |
827 | error(err_badparatype, t.text, &t.pos); |
828 | needkw = 4; |
829 | break; |
830 | case c__comment: |
831 | if (isbrace(in)) |
832 | break; /* `\#{': isn't a comment para */ |
833 | do { |
834 | dtor(t), t = get_token(in); |
835 | } while (t.type != tok_eop && t.type != tok_eof); |
836 | continue; /* next paragraph */ |
837 | /* |
838 | * `needkw' values: |
839 | * |
840 | * 1 -- exactly one keyword |
841 | * 2 -- at least one keyword |
842 | * 4 -- any number of keywords including zero |
843 | * 8 -- at least one keyword and then nothing else |
844 | * 16 -- nothing at all! no keywords, no body |
845 | * 32 -- no keywords at all |
846 | */ |
847 | case c_A: needkw = 2; par.type = para_Appendix; break; |
848 | case c_B: needkw = 2; par.type = para_Biblio; break; |
b3c80b25 |
849 | case c_BR: needkw = 1; par.type = para_BR; |
850 | start_cmd = c_BR; break; |
d7482997 |
851 | case c_C: needkw = 2; par.type = para_Chapter; break; |
852 | case c_H: needkw = 2; par.type = para_Heading; |
853 | par.aux = 0; |
854 | break; |
b3c80b25 |
855 | case c_IM: needkw = 2; par.type = para_IM; |
856 | start_cmd = c_IM; break; |
d7482997 |
857 | case c_S: needkw = 2; par.type = para_Subsect; |
858 | par.aux = t.aux; break; |
859 | case c_U: needkw = 32; par.type = para_UnnumberedChapter; break; |
860 | /* For \b and \n the keyword is optional */ |
861 | case c_b: needkw = 4; par.type = para_Bullet; break; |
7136a6c7 |
862 | case c_dt: needkw = 4; par.type = para_DescribedThing; break; |
863 | case c_dd: needkw = 4; par.type = para_Description; break; |
d7482997 |
864 | case c_n: needkw = 4; par.type = para_NumberedList; break; |
b3c80b25 |
865 | case c_cfg: needkw = 8; par.type = para_Config; |
866 | start_cmd = c_cfg; break; |
d7482997 |
867 | case c_copyright: needkw = 32; par.type = para_Copyright; break; |
868 | case c_define: is_macro = TRUE; needkw = 1; break; |
869 | /* For \nocite the keyword is _everything_ */ |
870 | case c_nocite: needkw = 8; par.type = para_NoCite; break; |
8902e0ed |
871 | case c_preamble: needkw = 32; par.type = para_Normal; break; |
d7482997 |
872 | case c_rule: needkw = 16; par.type = para_Rule; break; |
873 | case c_title: needkw = 32; par.type = para_Title; break; |
874 | case c_versionid: needkw = 32; par.type = para_VersionID; break; |
875 | } |
876 | |
7136a6c7 |
877 | if (par.type == para_Chapter || |
878 | par.type == para_Heading || |
879 | par.type == para_Subsect || |
880 | par.type == para_Appendix || |
881 | par.type == para_UnnumberedChapter) { |
882 | struct crossparaitem *sitem = stk_top(crossparastk); |
2614b01d |
883 | if (sitem && (sitem->seen_lcont || sitem->seen_quote)) { |
884 | error(err_sectmarkerinblock, |
885 | &t.pos, |
886 | (sitem->seen_lcont ? "lcont" : "quote")); |
7136a6c7 |
887 | } |
888 | } |
889 | |
d7482997 |
890 | if (needkw > 0) { |
891 | rdstring rs = { 0, 0, NULL }; |
e4ea58f8 |
892 | rdstringc rsc = { 0, 0, NULL }; |
d7482997 |
893 | int nkeys = 0; |
894 | filepos fp; |
895 | |
896 | /* Get keywords. */ |
897 | dtor(t), t = get_token(in); |
898 | fp = t.pos; |
899 | while (t.type == tok_lbrace) { |
900 | /* This is a keyword. */ |
901 | nkeys++; |
902 | /* FIXME: there will be bugs if anyone specifies an |
903 | * empty keyword (\foo{}), so trap this case. */ |
904 | while (dtor(t), t = get_token(in), |
905 | t.type == tok_word || |
906 | t.type == tok_white || |
907 | (t.type == tok_cmd && t.cmd == c__nbsp) || |
db662ca1 |
908 | (t.type == tok_cmd && t.cmd == c__escaped) || |
909 | (t.type == tok_cmd && t.cmd == c_u)) { |
d7482997 |
910 | if (t.type == tok_white || |
e4ea58f8 |
911 | (t.type == tok_cmd && t.cmd == c__nbsp)) { |
d7482997 |
912 | rdadd(&rs, ' '); |
e4ea58f8 |
913 | rdaddc(&rsc, ' '); |
db662ca1 |
914 | } else if (t.type == tok_cmd && t.cmd == c_u) { |
915 | rdadd(&rs, t.aux); |
916 | rdaddc(&rsc, '\\'); |
917 | rdaddsc(&rsc, t.origtext); |
e4ea58f8 |
918 | } else { |
d7482997 |
919 | rdadds(&rs, t.text); |
e4ea58f8 |
920 | rdaddsc(&rsc, t.origtext); |
921 | } |
d7482997 |
922 | } |
923 | if (t.type != tok_rbrace) { |
924 | error(err_kwunclosed, &t.pos); |
925 | continue; |
926 | } |
927 | rdadd(&rs, 0); /* add string terminator */ |
e4ea58f8 |
928 | rdaddc(&rsc, 0); /* add string terminator */ |
d7482997 |
929 | dtor(t), t = get_token(in); /* eat right brace */ |
930 | } |
931 | |
e4ea58f8 |
932 | rdadd(&rs, 0); /* add string terminator */ |
933 | rdaddc(&rsc, 0); /* add string terminator */ |
d7482997 |
934 | |
935 | /* See whether we have the right number of keywords. */ |
936 | if ((needkw & 48) && nkeys > 0) |
937 | error(err_kwillegal, &fp); |
938 | if ((needkw & 11) && nkeys == 0) |
939 | error(err_kwexpected, &fp); |
940 | if ((needkw & 5) && nkeys > 1) |
941 | error(err_kwtoomany, &fp); |
942 | |
943 | if (is_macro) { |
944 | /* |
945 | * Macro definition. Get the rest of the line |
946 | * as a code-paragraph token, repeatedly until |
947 | * there's nothing more left of it. Separate |
948 | * with newlines. |
949 | */ |
950 | rdstring macrotext = { 0, 0, NULL }; |
951 | while (1) { |
952 | dtor(t), t = get_codepar_token(in); |
953 | if (macrotext.pos > 0) |
954 | rdadd(¯otext, L'\n'); |
955 | rdadds(¯otext, t.text); |
956 | dtor(t), t = get_token(in); |
957 | if (t.type == tok_eop) break; |
958 | } |
959 | macrodef(macros, rs.text, macrotext.text, fp); |
960 | continue; /* next paragraph */ |
961 | } |
962 | |
963 | par.keyword = rdtrim(&rs); |
e4ea58f8 |
964 | par.origkeyword = rdtrimc(&rsc); |
d7482997 |
965 | |
966 | /* Move to EOP in case of needkw==8 or 16 (no body) */ |
967 | if (needkw & 24) { |
b3c80b25 |
968 | /* We allow whitespace even when we expect no para body */ |
969 | while (t.type == tok_white) |
970 | dtor(t), t = get_token(in); |
971 | if (t.type != tok_eop && t.type != tok_eof && |
972 | (start_cmd == c__invalid || |
973 | t.type != tok_cmd || t.cmd != start_cmd)) { |
d7482997 |
974 | error(err_bodyillegal, &t.pos); |
975 | /* Error recovery: eat the rest of the paragraph */ |
b3c80b25 |
976 | while (t.type != tok_eop && t.type != tok_eof && |
977 | (start_cmd == c__invalid || |
978 | t.type != tok_cmd || t.cmd != start_cmd)) |
d7482997 |
979 | dtor(t), t = get_token(in); |
980 | } |
b3c80b25 |
981 | if (t.type == tok_cmd) |
982 | already = TRUE;/* inhibit get_token at top of loop */ |
7136a6c7 |
983 | prev_para_type = par.type; |
d7482997 |
984 | addpara(par, ret); |
e34ba5c3 |
985 | |
986 | if (par.type == para_Config) { |
987 | input_configure(in, &par); |
988 | } |
d7482997 |
989 | continue; /* next paragraph */ |
990 | } |
991 | } |
992 | } |
993 | |
994 | /* |
995 | * Now read the actual paragraph, word by word, adding to |
996 | * the paragraph list. |
997 | * |
998 | * Mid-paragraph commands: |
999 | * |
1000 | * \K \k |
b7747050 |
1001 | * \c \cw \cq |
d7482997 |
1002 | * \e |
1003 | * \i \ii |
1004 | * \I |
b7747050 |
1005 | * \q |
d7482997 |
1006 | * \u |
1007 | * \W |
1008 | * \date |
1009 | * \\ \{ \} |
1010 | */ |
1011 | parsestk = stk_new(); |
1012 | style = word_Normal; |
1013 | spcstyle = word_WhiteSpace; |
1014 | indexing = FALSE; |
1015 | seenwhite = TRUE; |
1016 | while (t.type != tok_eop && t.type != tok_eof) { |
1017 | iswhite = FALSE; |
1018 | already = FALSE; |
b3c80b25 |
1019 | |
1020 | /* Handle implicit paragraph breaks after \IM, \BR etc */ |
1021 | if (start_cmd != c__invalid && |
1022 | t.type == tok_cmd && t.cmd == start_cmd) { |
1023 | already = TRUE; /* inhibit get_token at top of loop */ |
1024 | break; |
1025 | } |
1026 | |
672e4626 |
1027 | if (t.type == tok_cmd && t.cmd == c__nop) { |
1028 | dtor(t), t = get_token(in); |
1029 | continue; /* do nothing! */ |
1030 | } |
1031 | |
d7482997 |
1032 | if (t.type == tok_cmd && t.cmd == c__escaped) { |
1033 | t.type = tok_word; /* nice and simple */ |
1034 | t.aux = 0; /* even if `\-' - nonbreaking! */ |
1035 | } |
1036 | if (t.type == tok_cmd && t.cmd == c__nbsp) { |
1037 | t.type = tok_word; /* nice and simple */ |
1038 | sfree(t.text); |
1039 | t.text = ustrdup(L" "); /* text is ` ' not `_' */ |
1040 | t.aux = 0; /* (nonbreaking) */ |
1041 | } |
1042 | switch (t.type) { |
1043 | case tok_white: |
1044 | if (whptr == &par.words) |
1045 | break; /* strip whitespace at start of para */ |
1046 | wd.text = NULL; |
1047 | wd.type = spcstyle; |
1048 | wd.alt = NULL; |
1049 | wd.aux = 0; |
1050 | wd.fpos = t.pos; |
1051 | wd.breaks = FALSE; |
b3c80b25 |
1052 | |
1053 | /* |
1054 | * Inhibit use of whitespace if it's (probably the |
1055 | * newline) before a repeat \IM / \BR type |
1056 | * directive. |
1057 | */ |
1058 | if (start_cmd != c__invalid) { |
1059 | dtor(t), t = get_token(in); |
1060 | already = TRUE; |
1061 | if (t.type == tok_cmd && t.cmd == start_cmd) |
1062 | break; |
1063 | } |
1064 | |
d7482997 |
1065 | if (indexing) |
1066 | rdadd(&indexstr, ' '); |
1067 | if (!indexing || index_visible) |
1068 | addword(wd, &whptr); |
1069 | if (indexing) |
1070 | addword(wd, &idximplicit); |
1071 | iswhite = TRUE; |
1072 | break; |
1073 | case tok_word: |
1074 | if (indexing) |
1075 | rdadds(&indexstr, t.text); |
1076 | wd.type = style; |
1077 | wd.alt = NULL; |
1078 | wd.aux = 0; |
1079 | wd.fpos = t.pos; |
1080 | wd.breaks = t.aux; |
1081 | if (!indexing || index_visible) { |
1082 | wd.text = ustrdup(t.text); |
1083 | addword(wd, &whptr); |
1084 | } |
1085 | if (indexing) { |
1086 | wd.text = ustrdup(t.text); |
1087 | addword(wd, &idximplicit); |
1088 | } |
1089 | break; |
1090 | case tok_lbrace: |
1091 | error(err_unexbrace, &t.pos); |
1092 | /* Error recovery: push nop */ |
f1530049 |
1093 | sitem = snew(struct stack_item); |
d7482997 |
1094 | sitem->type = stack_nop; |
f4551933 |
1095 | sitem->fpos = t.pos; |
d7482997 |
1096 | stk_push(parsestk, sitem); |
1097 | break; |
1098 | case tok_rbrace: |
1099 | sitem = stk_pop(parsestk); |
7136a6c7 |
1100 | if (!sitem) { |
1101 | /* |
1102 | * This closing brace could have been an |
1103 | * indication that the cross-paragraph stack |
1104 | * wants popping. Accordingly, we treat it here |
1105 | * as an indication that the paragraph is over. |
1106 | */ |
1107 | already = TRUE; |
1108 | goto finished_para; |
1109 | } else { |
d7482997 |
1110 | if (sitem->type & stack_ualt) { |
1111 | whptr = sitem->whptr; |
1112 | idximplicit = sitem->idximplicit; |
1113 | } |
1114 | if (sitem->type & stack_style) { |
1115 | style = word_Normal; |
1116 | spcstyle = word_WhiteSpace; |
1117 | } |
1118 | if (sitem->type & stack_idx) { |
1119 | indexword->text = ustrdup(indexstr.text); |
8b7d993d |
1120 | if (index_downcase) { |
1121 | word *w; |
1122 | |
d7482997 |
1123 | ustrlow(indexword->text); |
8b7d993d |
1124 | ustrlow(indexstr.text); |
1125 | |
1126 | for (w = idxwordlist; w; w = w->next) |
1127 | if (w->text) |
1128 | ustrlow(w->text); |
1129 | } |
d7482997 |
1130 | indexing = FALSE; |
1131 | rdadd(&indexstr, L'\0'); |
f4551933 |
1132 | index_merge(idx, FALSE, indexstr.text, |
1133 | idxwordlist, &sitem->fpos); |
d7482997 |
1134 | sfree(indexstr.text); |
1135 | } |
1136 | if (sitem->type & stack_hyper) { |
1137 | wd.text = NULL; |
1138 | wd.type = word_HyperEnd; |
1139 | wd.alt = NULL; |
1140 | wd.aux = 0; |
1141 | wd.fpos = t.pos; |
1142 | wd.breaks = FALSE; |
1143 | if (!indexing || index_visible) |
1144 | addword(wd, &whptr); |
1145 | if (indexing) |
1146 | addword(wd, &idximplicit); |
1147 | } |
1148 | if (sitem->type & stack_quote) { |
1149 | wd.text = NULL; |
1150 | wd.type = toquotestyle(style); |
1151 | wd.alt = NULL; |
1152 | wd.aux = quote_Close; |
1153 | wd.fpos = t.pos; |
1154 | wd.breaks = FALSE; |
1155 | if (!indexing || index_visible) |
1156 | addword(wd, &whptr); |
1157 | if (indexing) { |
1158 | rdadd(&indexstr, L'"'); |
1159 | addword(wd, &idximplicit); |
1160 | } |
1161 | } |
1162 | } |
1163 | sfree(sitem); |
1164 | break; |
1165 | case tok_cmd: |
1166 | switch (t.cmd) { |
1167 | case c__comment: |
1168 | /* |
1169 | * In-paragraph comment: \#{ balanced braces } |
1170 | * |
1171 | * Anything goes here; even tok_eop. We should |
1172 | * eat whitespace after the close brace _if_ |
1173 | * there was whitespace before the \#. |
1174 | */ |
1175 | dtor(t), t = get_token(in); |
1176 | if (t.type != tok_lbrace) { |
1177 | error(err_explbr, &t.pos); |
1178 | } else { |
1179 | int braces = 1; |
1180 | while (braces > 0) { |
1181 | dtor(t), t = get_token(in); |
1182 | if (t.type == tok_lbrace) |
1183 | braces++; |
1184 | else if (t.type == tok_rbrace) |
1185 | braces--; |
1186 | else if (t.type == tok_eof) { |
1187 | error(err_commenteof, &t.pos); |
1188 | break; |
1189 | } |
1190 | } |
1191 | } |
1192 | if (seenwhite) { |
1193 | already = TRUE; |
1194 | dtor(t), t = get_token(in); |
1195 | if (t.type == tok_white) { |
1196 | iswhite = TRUE; |
1197 | already = FALSE; |
1198 | } |
1199 | } |
1200 | break; |
1201 | case c_q: |
b7747050 |
1202 | case c_cq: |
1203 | type = t.cmd; |
d7482997 |
1204 | dtor(t), t = get_token(in); |
1205 | if (t.type != tok_lbrace) { |
1206 | error(err_explbr, &t.pos); |
1207 | } else { |
6ff15f2b |
1208 | /* |
1209 | * Enforce that \q may not be used anywhere |
1210 | * within \c. (It shouldn't be necessary |
1211 | * since the whole point of \c should be |
1212 | * that the user wants to exercise exact |
1213 | * control over the glyphs used, and |
1214 | * forbidding it has the useful effect of |
1215 | * relieving some backends of having to |
1216 | * make difficult decisions.) |
1217 | */ |
1218 | int stype; |
1219 | |
1220 | if (style != word_Code && style != word_WeakCode) { |
1221 | wd.text = NULL; |
1222 | wd.type = toquotestyle(style); |
1223 | wd.alt = NULL; |
1224 | wd.aux = quote_Open; |
1225 | wd.fpos = t.pos; |
1226 | wd.breaks = FALSE; |
1227 | if (!indexing || index_visible) |
1228 | addword(wd, &whptr); |
1229 | if (indexing) { |
1230 | rdadd(&indexstr, L'"'); |
1231 | addword(wd, &idximplicit); |
1232 | } |
1233 | stype = stack_quote; |
1234 | } else { |
1235 | error(err_codequote, &t.pos); |
1236 | stype = stack_nop; |
d7482997 |
1237 | } |
f1530049 |
1238 | sitem = snew(struct stack_item); |
f4551933 |
1239 | sitem->fpos = t.pos; |
6ff15f2b |
1240 | sitem->type = stype; |
b7747050 |
1241 | if (type == c_cq) { |
1242 | if (style != word_Normal) { |
1243 | error(err_nestedstyles, &t.pos); |
1244 | } else { |
1245 | style = word_WeakCode; |
1246 | spcstyle = tospacestyle(style); |
1247 | sitem->type |= stack_style; |
1248 | } |
1249 | } |
d7482997 |
1250 | stk_push(parsestk, sitem); |
1251 | } |
1252 | break; |
1253 | case c_K: |
1254 | case c_k: |
1255 | case c_W: |
1256 | case c_date: |
1257 | /* |
1258 | * Keyword, hyperlink, or \date. We expect a |
1259 | * left brace, some text, and then a right |
1260 | * brace. No nesting; no arguments. |
1261 | */ |
1262 | wd.fpos = t.pos; |
1263 | wd.breaks = FALSE; |
1264 | if (t.cmd == c_K) |
1265 | wd.type = word_UpperXref; |
1266 | else if (t.cmd == c_k) |
1267 | wd.type = word_LowerXref; |
1268 | else if (t.cmd == c_W) |
1269 | wd.type = word_HyperLink; |
1270 | else |
1271 | wd.type = word_Normal; |
1272 | dtor(t), t = get_token(in); |
1273 | if (t.type != tok_lbrace) { |
1274 | if (wd.type == word_Normal) { |
1275 | time_t thetime = time(NULL); |
1276 | struct tm *broken = localtime(&thetime); |
1277 | already = TRUE; |
1278 | wdtext = ustrftime(NULL, broken); |
1279 | wd.type = style; |
1280 | } else { |
1281 | error(err_explbr, &t.pos); |
1282 | wdtext = NULL; |
1283 | } |
1284 | } else { |
1285 | rdstring rs = { 0, 0, NULL }; |
1286 | while (dtor(t), t = get_token(in), |
1287 | t.type == tok_word || t.type == tok_white) { |
1288 | if (t.type == tok_white) |
1289 | rdadd(&rs, ' '); |
1290 | else |
1291 | rdadds(&rs, t.text); |
1292 | } |
1293 | if (wd.type == word_Normal) { |
1294 | time_t thetime = time(NULL); |
1295 | struct tm *broken = localtime(&thetime); |
1296 | wdtext = ustrftime(rs.text, broken); |
1297 | wd.type = style; |
1298 | } else { |
1299 | wdtext = ustrdup(rs.text); |
1300 | } |
1301 | sfree(rs.text); |
1302 | if (t.type != tok_rbrace) { |
1303 | error(err_kwexprbr, &t.pos); |
1304 | } |
1305 | } |
1306 | wd.alt = NULL; |
1307 | wd.aux = 0; |
1308 | if (!indexing || index_visible) { |
1309 | wd.text = ustrdup(wdtext); |
1310 | addword(wd, &whptr); |
1311 | } |
1312 | if (indexing) { |
1313 | wd.text = ustrdup(wdtext); |
1314 | addword(wd, &idximplicit); |
1315 | } |
1316 | sfree(wdtext); |
1317 | if (wd.type == word_HyperLink) { |
1318 | /* |
1319 | * Hyperlinks are different: they then |
1320 | * expect another left brace, to begin |
1321 | * delimiting the text marked by the link. |
1322 | */ |
1323 | dtor(t), t = get_token(in); |
f1530049 |
1324 | sitem = snew(struct stack_item); |
f4551933 |
1325 | sitem->fpos = wd.fpos; |
d7482997 |
1326 | sitem->type = stack_hyper; |
b80802ba |
1327 | /* |
1328 | * Special cases: \W{}\i, \W{}\ii |
1329 | */ |
1330 | if (t.type == tok_cmd && |
1331 | (t.cmd == c_i || t.cmd == c_ii)) { |
1332 | if (indexing) { |
1333 | error(err_nestedindex, &t.pos); |
1334 | } else { |
1335 | /* Add an index-reference word with no |
1336 | * text as yet */ |
1337 | wd.type = word_IndexRef; |
1338 | wd.text = NULL; |
1339 | wd.alt = NULL; |
1340 | wd.aux = 0; |
1341 | wd.breaks = FALSE; |
1342 | indexword = addword(wd, &whptr); |
1343 | /* Set up a rdstring to read the |
1344 | * index text */ |
1345 | indexstr = nullrs; |
1346 | /* Flags so that we do the Right |
1347 | * Things with text */ |
1348 | index_visible = (type != c_I); |
1349 | index_downcase = (type == c_ii); |
1350 | indexing = TRUE; |
1351 | idxwordlist = NULL; |
1352 | idximplicit = &idxwordlist; |
1353 | |
1354 | sitem->type |= stack_idx; |
1355 | } |
1356 | dtor(t), t = get_token(in); |
1357 | } |
1358 | /* |
1359 | * Special cases: \W{}\c, \W{}\e, \W{}\cw |
1360 | */ |
d7482997 |
1361 | if (t.type == tok_cmd && |
1362 | (t.cmd == c_e || t.cmd == c_c || t.cmd == c_cw)) { |
1363 | if (style != word_Normal) |
1364 | error(err_nestedstyles, &t.pos); |
1365 | else { |
1366 | style = (t.cmd == c_c ? word_Code : |
1367 | t.cmd == c_cw ? word_WeakCode : |
1368 | word_Emph); |
1369 | spcstyle = tospacestyle(style); |
1370 | sitem->type |= stack_style; |
1371 | } |
1372 | dtor(t), t = get_token(in); |
1373 | } |
1374 | if (t.type != tok_lbrace) { |
1375 | error(err_explbr, &t.pos); |
1376 | sfree(sitem); |
1377 | } else { |
1378 | stk_push(parsestk, sitem); |
1379 | } |
1380 | } |
1381 | break; |
1382 | case c_c: |
1383 | case c_cw: |
1384 | case c_e: |
1385 | type = t.cmd; |
1386 | if (style != word_Normal) { |
1387 | error(err_nestedstyles, &t.pos); |
1388 | /* Error recovery: eat lbrace, push nop. */ |
1389 | dtor(t), t = get_token(in); |
f1530049 |
1390 | sitem = snew(struct stack_item); |
f4551933 |
1391 | sitem->fpos = t.pos; |
d7482997 |
1392 | sitem->type = stack_nop; |
1393 | stk_push(parsestk, sitem); |
1394 | } |
1395 | dtor(t), t = get_token(in); |
1396 | if (t.type != tok_lbrace) { |
1397 | error(err_explbr, &t.pos); |
1398 | } else { |
1399 | style = (type == c_c ? word_Code : |
1400 | type == c_cw ? word_WeakCode : |
1401 | word_Emph); |
1402 | spcstyle = tospacestyle(style); |
f1530049 |
1403 | sitem = snew(struct stack_item); |
f4551933 |
1404 | sitem->fpos = t.pos; |
d7482997 |
1405 | sitem->type = stack_style; |
1406 | stk_push(parsestk, sitem); |
1407 | } |
1408 | break; |
1409 | case c_i: |
1410 | case c_ii: |
1411 | case c_I: |
1412 | type = t.cmd; |
1413 | if (indexing) { |
1414 | error(err_nestedindex, &t.pos); |
1415 | /* Error recovery: eat lbrace, push nop. */ |
1416 | dtor(t), t = get_token(in); |
f1530049 |
1417 | sitem = snew(struct stack_item); |
f4551933 |
1418 | sitem->fpos = t.pos; |
d7482997 |
1419 | sitem->type = stack_nop; |
1420 | stk_push(parsestk, sitem); |
1421 | } |
f1530049 |
1422 | sitem = snew(struct stack_item); |
f4551933 |
1423 | sitem->fpos = t.pos; |
d7482997 |
1424 | sitem->type = stack_idx; |
1425 | dtor(t), t = get_token(in); |
1426 | /* |
1427 | * Special cases: \i\c, \i\e, \i\cw |
1428 | */ |
1429 | wd.fpos = t.pos; |
1430 | if (t.type == tok_cmd && |
1431 | (t.cmd == c_e || t.cmd == c_c || t.cmd == c_cw)) { |
1432 | if (style != word_Normal) |
1433 | error(err_nestedstyles, &t.pos); |
1434 | else { |
1435 | style = (t.cmd == c_c ? word_Code : |
1436 | t.cmd == c_cw ? word_WeakCode : |
1437 | word_Emph); |
1438 | spcstyle = tospacestyle(style); |
1439 | sitem->type |= stack_style; |
1440 | } |
1441 | dtor(t), t = get_token(in); |
1442 | } |
1443 | if (t.type != tok_lbrace) { |
1444 | sfree(sitem); |
1445 | error(err_explbr, &t.pos); |
1446 | } else { |
1447 | /* Add an index-reference word with no text as yet */ |
1448 | wd.type = word_IndexRef; |
1449 | wd.text = NULL; |
1450 | wd.alt = NULL; |
1451 | wd.aux = 0; |
1452 | wd.breaks = FALSE; |
1453 | indexword = addword(wd, &whptr); |
1454 | /* Set up a rdstring to read the index text */ |
1455 | indexstr = nullrs; |
1456 | /* Flags so that we do the Right Things with text */ |
1457 | index_visible = (type != c_I); |
1458 | index_downcase = (type == c_ii); |
1459 | indexing = TRUE; |
1460 | idxwordlist = NULL; |
1461 | idximplicit = &idxwordlist; |
1462 | /* Stack item to close the indexing on exit */ |
1463 | stk_push(parsestk, sitem); |
1464 | } |
1465 | break; |
1466 | case c_u: |
1467 | uchr = t.aux; |
1468 | utext[0] = uchr; utext[1] = 0; |
1469 | wd.type = style; |
1470 | wd.breaks = FALSE; |
1471 | wd.alt = NULL; |
1472 | wd.aux = 0; |
1473 | wd.fpos = t.pos; |
1474 | if (!indexing || index_visible) { |
1475 | wd.text = ustrdup(utext); |
1476 | uword = addword(wd, &whptr); |
1477 | } else |
1478 | uword = NULL; |
1479 | if (indexing) { |
1480 | wd.text = ustrdup(utext); |
1481 | iword = addword(wd, &idximplicit); |
1482 | } else |
1483 | iword = NULL; |
1484 | dtor(t), t = get_token(in); |
1485 | if (t.type == tok_lbrace) { |
1486 | /* |
1487 | * \u with a left brace. Until the brace |
1488 | * closes, all further words go on a |
1489 | * sidetrack from the main thread of the |
1490 | * paragraph. |
1491 | */ |
f1530049 |
1492 | sitem = snew(struct stack_item); |
f4551933 |
1493 | sitem->fpos = t.pos; |
d7482997 |
1494 | sitem->type = stack_ualt; |
1495 | sitem->whptr = whptr; |
1496 | sitem->idximplicit = idximplicit; |
1497 | stk_push(parsestk, sitem); |
1498 | whptr = uword ? &uword->alt : NULL; |
1499 | idximplicit = iword ? &iword->alt : NULL; |
1500 | } else { |
1501 | if (indexing) |
1502 | rdadd(&indexstr, uchr); |
1503 | already = TRUE; |
1504 | } |
1505 | break; |
1506 | default: |
1507 | if (!macrolookup(macros, in, t.text, &t.pos)) |
1508 | error(err_badmidcmd, t.text, &t.pos); |
1509 | break; |
1510 | } |
1511 | } |
1512 | if (!already) |
1513 | dtor(t), t = get_token(in); |
1514 | seenwhite = iswhite; |
1515 | } |
7136a6c7 |
1516 | finished_para: |
d7482997 |
1517 | /* Check the stack is empty */ |
7136a6c7 |
1518 | if (stk_top(parsestk)) { |
1519 | while ((sitem = stk_pop(parsestk))) |
d7482997 |
1520 | sfree(sitem); |
d7482997 |
1521 | error(err_missingrbrace, &t.pos); |
1522 | } |
1523 | stk_free(parsestk); |
7136a6c7 |
1524 | prev_para_type = par.type; |
d7482997 |
1525 | addpara(par, ret); |
7136a6c7 |
1526 | if (t.type == tok_eof) |
1527 | already = TRUE; |
1528 | } |
1529 | |
1530 | if (stk_top(crossparastk)) { |
1531 | void *p; |
1532 | |
1533 | error(err_missingrbrace2, &t.pos); |
1534 | while ((p = stk_pop(crossparastk))) |
1535 | sfree(p); |
d7482997 |
1536 | } |
677e18a2 |
1537 | |
1538 | /* |
1539 | * We break to here rather than returning, because otherwise |
1540 | * this cleanup doesn't happen. |
1541 | */ |
d7482997 |
1542 | dtor(t); |
1543 | macrocleanup(macros); |
7136a6c7 |
1544 | |
1545 | stk_free(crossparastk); |
d7482997 |
1546 | } |
1547 | |
1548 | paragraph *read_input(input *in, indexdata *idx) { |
1549 | paragraph *head = NULL; |
1550 | paragraph **hptr = &head; |
1551 | |
1552 | while (in->currindex < in->nfiles) { |
1553 | in->currfp = fopen(in->filenames[in->currindex], "r"); |
1554 | if (in->currfp) { |
1555 | setpos(in, in->filenames[in->currindex]); |
e34ba5c3 |
1556 | in->charset = in->defcharset; |
1557 | in->csstate = charset_init_state; |
e4ea58f8 |
1558 | in->wcpos = in->nwc = 0; |
1559 | in->pushback_chars = NULL; |
d7482997 |
1560 | read_file(&hptr, in, idx); |
1561 | } |
1562 | in->currindex++; |
1563 | } |
1564 | |
1565 | return head; |
1566 | } |