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