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