Annotation of texi2mdoc/util.c, Revision 1.26
1.26 ! kristaps 1: /* $Id: util.c,v 1.25 2015/03/03 22:37:24 kristaps Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2015 Kristaps Dzonsons <kristaps@bsd.lv>
4: *
5: * Permission to use, copy, modify, and distribute this software for any
6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
8: *
9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16: */
17: #include <sys/mman.h>
18: #include <sys/stat.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
22: #include <fcntl.h>
23: #include <limits.h>
24: #include <stdarg.h>
25: #include <stdio.h>
26: #include <stdlib.h>
27: #include <string.h>
28: #include <time.h>
29: #include <unistd.h>
30:
31: #include "extern.h"
32:
33: /*
34: * Unmap the top-most file in the stack of files currently opened (that
35: * is, nested calls to parsefile()).
36: */
37: void
38: texifilepop(struct texi *p)
39: {
40: struct texifile *f;
41:
42: assert(p->filepos > 0);
43: f = &p->files[--p->filepos];
1.14 kristaps 44: free(f->map);
1.1 kristaps 45: }
46:
1.7 kristaps 47: static void
48: teximacrofree(struct teximacro *p)
49: {
50: size_t i;
51:
52: for (i = 0; i < p->argsz; i++)
53: free(p->args[i]);
54:
55: free(p->args);
56: free(p->key);
57: free(p->value);
58: }
59:
60: static void
61: texivaluefree(struct texivalue *p)
62: {
63:
64: free(p->key);
65: free(p->value);
66: }
67:
1.1 kristaps 68: /*
69: * Unmap all files that we're currently using and free all resources
70: * that we've allocated during the parse.
71: * The utility should exit(...) after this is called.
72: */
73: void
74: texiexit(struct texi *p)
75: {
76: size_t i;
77:
78: /* Make sure we're newline-terminated. */
79: if (p->outcol)
1.20 kristaps 80: fputc('\n', p->outfile);
81: if (NULL != p->chapters)
82: teximdocclose(p, 1);
1.1 kristaps 83:
84: /* Unmap all files. */
85: while (p->filepos > 0)
86: texifilepop(p);
87:
1.7 kristaps 88: for (i = 0; i < p->macrosz; i++)
89: teximacrofree(&p->macros[i]);
1.1 kristaps 90: for (i = 0; i < p->dirsz; i++)
91: free(p->dirs[i]);
1.4 kristaps 92: for (i = 0; i < p->indexsz; i++)
93: free(p->indexs[i]);
1.7 kristaps 94: for (i = 0; i < p->valsz; i++)
95: texivaluefree(&p->vals[i]);
1.4 kristaps 96:
1.7 kristaps 97: free(p->macros);
1.1 kristaps 98: free(p->vals);
1.4 kristaps 99: free(p->indexs);
1.1 kristaps 100: free(p->dirs);
101: free(p->subtitle);
102: free(p->title);
1.26 ! kristaps 103: free(p->copying);
1.1 kristaps 104: }
105:
106: /*
107: * Fatal error: unmap all files and exit.
108: * The "errstring" is passed to perror(3).
109: */
110: void
111: texiabort(struct texi *p, const char *errstring)
112: {
113:
114: perror(errstring);
115: texiexit(p);
116: exit(EXIT_FAILURE);
117: }
118:
119: /*
120: * Print a generic warning message (to stderr) tied to our current
121: * location in the parse sequence.
122: */
123: void
124: texiwarn(const struct texi *p, const char *fmt, ...)
125: {
1.15 kristaps 126: va_list ap;
127: const struct texifile *f;
128:
129: f = &p->files[p->filepos - 1];
130:
131: if (f->insplice)
132: fprintf(stderr, "%s:%zu:%zu (%zuB left in splice): "
133: "warning: ", f->name, f->line + 1,
134: f->col + 1, f->insplice);
135: else
136: fprintf(stderr, "%s:%zu:%zu: warning: ",
137: f->name, f->line + 1, f->col + 1);
1.1 kristaps 138:
139: va_start(ap, fmt);
140: vfprintf(stderr, fmt, ap);
141: va_end(ap);
142: fputc('\n', stderr);
143: }
144:
145: /*
146: * Print an error message (to stderr) tied to our current location in
147: * the parse sequence, invoke texiexit(), then die.
148: */
149: void
150: texierr(struct texi *p, const char *fmt, ...)
151: {
1.15 kristaps 152: va_list ap;
153: struct texifile *f;
154:
155: f = &p->files[p->filepos - 1];
156:
157: if (f->insplice)
158: fprintf(stderr, "%s:%zu:%zu: (%zuB left in splice): "
159: "error: ", f->name, f->line + 1,
160: f->col + 1, f->insplice);
161: else
162: fprintf(stderr, "%s:%zu:%zu: error: ",
163: f->name, f->line + 1, f->col + 1);
1.1 kristaps 164:
165: va_start(ap, fmt);
166: vfprintf(stderr, fmt, ap);
167: va_end(ap);
168: fputc('\n', stderr);
169: texiexit(p);
170: exit(EXIT_FAILURE);
171: }
172:
173: /*
174: * Put a single data character to the output if we're not ignoring.
1.13 kristaps 175: * Escape starting a line with a control character and slashes.
1.1 kristaps 176: */
177: void
178: texiputchar(struct texi *p, char c)
179: {
180:
181: if (p->ign)
182: return;
183: if ('.' == c && 0 == p->outcol)
1.20 kristaps 184: fputs("\\&", p->outfile);
1.10 kristaps 185: if ('\'' == c && 0 == p->outcol)
1.20 kristaps 186: fputs("\\&", p->outfile);
1.1 kristaps 187:
1.23 kristaps 188: if (p->uppercase)
189: fputc(toupper((unsigned int)c), p->outfile);
190: else
191: fputc(c, p->outfile);
1.13 kristaps 192: if ('\\' == c)
1.20 kristaps 193: fputc('e', p->outfile);
1.1 kristaps 194: p->seenvs = 0;
195: if ('\n' == c) {
196: p->outcol = 0;
197: p->seenws = 0;
198: } else
199: p->outcol++;
200: }
201:
202: /*
1.13 kristaps 203: * Put an opaque series of characters.
204: * Characters starting a line with a control character are escaped, but
205: * that's it, so don't use this for non-controlled sequences of text.
1.1 kristaps 206: */
207: void
208: texiputchars(struct texi *p, const char *s)
209: {
210:
1.13 kristaps 211: if (p->ign)
212: return;
213: if ('.' == *s && 0 == p->outcol)
1.20 kristaps 214: fputs("\\&", p->outfile);
1.13 kristaps 215: if ('\'' == *s && 0 == p->outcol)
1.20 kristaps 216: fputs("\\&", p->outfile);
1.23 kristaps 217: if (p->uppercase)
218: for ( ; '\0' != *s; s++)
219: p->outcol += fputc(toupper
220: ((unsigned int)*s), p->outfile);
221: else
222: p->outcol += fputs(s, p->outfile);
1.13 kristaps 223: p->seenvs = 0;
1.9 kristaps 224: }
225:
226: /*
227: * This puts all characters onto the output stream but makes sure to
228: * escape mdoc(7) slashes.
1.14 kristaps 229: * FIXME: useless.
1.9 kristaps 230: */
231: void
1.14 kristaps 232: texiputbuf(struct texi *p, size_t start, size_t end)
1.9 kristaps 233: {
234:
1.14 kristaps 235: for ( ; start < end; start++)
236: texiputchar(p, BUF(p)[start]);
1.1 kristaps 237: }
238:
239: /*
240: * Close an mdoc(7) macro opened with teximacroopen().
241: * If there are no more macros on the line, prints a newline.
242: */
243: void
244: teximacroclose(struct texi *p)
245: {
246:
247: if (p->ign)
248: return;
249:
250: if (0 == --p->outmacro) {
1.20 kristaps 251: fputc('\n', p->outfile);
1.1 kristaps 252: p->outcol = p->seenws = 0;
253: }
254: }
255:
256: /*
257: * Open a mdoc(7) macro.
258: * This is used for line macros, e.g., Qq [foo bar baz].
259: * It can be invoked for nested macros, e.g., Qq Li foo .
260: * TODO: flush-right punctuation (e.g., parenthesis).
261: */
262: void
263: teximacroopen(struct texi *p, const char *s)
264: {
265: int rc;
266:
267: if (p->ign)
268: return;
269:
270: if (p->outcol && 0 == p->outmacro) {
1.20 kristaps 271: fputc('\n', p->outfile);
1.1 kristaps 272: p->outcol = 0;
273: }
274:
275: if (0 == p->outmacro)
1.20 kristaps 276: fputc('.', p->outfile);
1.1 kristaps 277: else
1.20 kristaps 278: fputc(' ', p->outfile);
1.1 kristaps 279:
1.20 kristaps 280: if (EOF != (rc = fputs(s, p->outfile)))
1.1 kristaps 281: p->outcol += rc;
282:
1.20 kristaps 283: fputc(' ', p->outfile);
1.1 kristaps 284: p->outcol++;
285: p->outmacro++;
286: p->seenws = 0;
287: }
288:
289: /*
290: * Put a stadnalone mdoc(7) command with the trailing newline.
291: */
292: void
293: teximacro(struct texi *p, const char *s)
294: {
295:
296: if (p->ign)
297: return;
298:
299: if (p->outmacro)
300: texierr(p, "\"%s\" in open line scope!?", s);
301: if (p->literal)
302: texierr(p, "\"%s\" in a literal scope!?", s);
303:
304: if (p->outcol)
1.20 kristaps 305: fputc('\n', p->outfile);
1.1 kristaps 306:
1.20 kristaps 307: fputc('.', p->outfile);
308: fputs(s, p->outfile);
309: fputc('\n', p->outfile);
1.1 kristaps 310: p->outcol = p->seenws = 0;
311: }
312:
313: /*
314: * Introduce vertical space during normal (non-macro) input.
315: */
316: void
317: texivspace(struct texi *p)
318: {
319:
1.5 kristaps 320: if (p->seenvs || TEXILIST_TABLE == p->list)
1.1 kristaps 321: return;
322: teximacro(p, "Pp");
323: p->seenvs = 1;
324: }
325:
326: /*
327: * Advance by a single byte in the input stream, adjusting our location
328: * in the current input file.
329: */
330: void
1.14 kristaps 331: advance(struct texi *p, size_t *pos)
1.1 kristaps 332: {
1.15 kristaps 333: struct texifile *f;
1.1 kristaps 334:
1.15 kristaps 335: f = &p->files[p->filepos - 1];
336:
337: if (0 == f->insplice) {
338: if ('\n' == BUF(p)[*pos]) {
339: f->line++;
340: f->col = 0;
341: } else
342: f->col++;
1.17 kristaps 343: } else {
1.15 kristaps 344: --f->insplice;
1.17 kristaps 345: if (0 == f->insplice)
346: f->depth = 0;
347: }
1.1 kristaps 348:
349: (*pos)++;
350: }
351:
352: /*
353: * It's common to wait punctuation to float on the right side of macro
354: * lines in mdoc(7), e.g., ".Em hello ) ."
355: * This function does so, and should be called before teximacroclose().
356: * It will detect that it's the last in the nested macros and
357: * appropriately flush-left punctuation alongside the macro.
358: */
359: void
1.14 kristaps 360: texipunctuate(struct texi *p, size_t *pos)
1.1 kristaps 361: {
362: size_t start, end;
363:
364: if (1 != p->outmacro)
365: return;
366:
1.14 kristaps 367: for (start = end = *pos; end < BUFSZ(p); end++) {
368: switch (BUF(p)[end]) {
1.1 kristaps 369: case (','):
370: case (')'):
371: case ('.'):
372: case ('"'):
373: case (':'):
1.22 kristaps 374: case (';'):
1.1 kristaps 375: case ('!'):
376: case ('?'):
377: continue;
378: default:
379: break;
380: }
381: break;
382: }
383: if (end == *pos)
384: return;
1.14 kristaps 385: if (end + 1 == BUFSZ(p) || ' ' == BUF(p)[end] ||
386: '\n' == BUF(p)[end]) {
1.1 kristaps 387: for ( ; start < end; start++) {
388: texiputchar(p, ' ');
1.14 kristaps 389: texiputchar(p, BUF(p)[start]);
390: advance(p, pos);
1.1 kristaps 391: }
392: }
393: }
394:
395: /*
396: * Advance to the next non-whitespace word in the input stream.
397: * If we're in literal mode, then print all of the whitespace as we're
398: * doing so.
399: */
400: static size_t
1.14 kristaps 401: advancenext(struct texi *p, size_t *pos)
1.1 kristaps 402: {
403:
404: if (p->literal) {
1.14 kristaps 405: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) {
406: texiputchar(p, BUF(p)[*pos]);
407: advance(p, pos);
1.1 kristaps 408: }
409: return(*pos);
410: }
411:
1.14 kristaps 412: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) {
1.1 kristaps 413: p->seenws = 1;
414: /*
415: * If it looks like we've printed a double-line, then
416: * output a paragraph.
417: * FIXME: this is stupid.
418: */
1.14 kristaps 419: if (*pos && '\n' == BUF(p)[*pos] && '\n' == BUF(p)[*pos - 1])
1.1 kristaps 420: texivspace(p);
1.14 kristaps 421: advance(p, pos);
1.1 kristaps 422: }
423: return(*pos);
424: }
425:
426: /*
427: * Advance to the EOLN in the input stream.
1.22 kristaps 428: * This will skip over '@' markers in an effort to ignore escaped
429: * newlines.
1.1 kristaps 430: */
431: size_t
1.14 kristaps 432: advanceeoln(struct texi *p, size_t *pos, int consumenl)
1.1 kristaps 433: {
434:
1.22 kristaps 435: while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) {
436: if ('@' == BUF(p)[*pos])
437: advance(p, pos);
1.14 kristaps 438: advance(p, pos);
1.22 kristaps 439: }
1.14 kristaps 440: if (*pos < BUFSZ(p) && consumenl)
441: advance(p, pos);
1.1 kristaps 442: return(*pos);
443: }
444:
445: /*
446: * Advance to position "end", which is an absolute position in the
447: * current buffer greater than or equal to the current position.
448: */
449: void
1.14 kristaps 450: advanceto(struct texi *p, size_t *pos, size_t end)
1.1 kristaps 451: {
452:
453: assert(*pos <= end);
454: while (*pos < end)
1.14 kristaps 455: advance(p, pos);
1.1 kristaps 456: }
457:
1.7 kristaps 458: static void
1.17 kristaps 459: texiexecmacro(struct texi *p, struct teximacro *m, size_t sv, size_t *pos)
1.7 kristaps 460: {
1.11 kristaps 461: size_t valsz, realsz, aasz, asz,
462: ssz, i, j, k, start, end;
463: char *val;
464: char **args;
465: const char *cp;
1.7 kristaps 466:
1.17 kristaps 467: /* Disregard empty macros. */
1.22 kristaps 468: if (0 == (valsz = realsz = strlen(m->value))) {
469: args = argparse(p, pos, &asz, m->argsz);
470: for (i = 0; i < asz; i++)
471: free(args[i]);
472: free(args);
1.17 kristaps 473: return;
1.22 kristaps 474: }
1.17 kristaps 475:
476: /*
477: * This is important: it protect us from macros that invoke more
478: * macros, possibly going on infinitely.
479: * We use "sv" instead of the current position because we might
480: * be invoked at the end of the macro (i.e., insplice == 0).
481: * The "sv" value was initialised at the start of the macro.
482: */
483: if (sv > 0)
1.24 kristaps 484: if (++p->files[p->filepos - 1].depth > 64)
1.17 kristaps 485: texierr(p, "maximium recursive depth");
486:
1.14 kristaps 487: args = argparse(p, pos, &asz, m->argsz);
1.7 kristaps 488: if (asz != m->argsz)
489: texiwarn(p, "invalid macro argument length");
490: aasz = asz < m->argsz ? asz : m->argsz;
491:
492: if (0 == aasz) {
1.21 kristaps 493: texisplice(p, m->value, valsz, *pos);
1.7 kristaps 494: return;
495: }
496:
497: val = strdup(m->value);
498:
499: for (i = j = 0; i < realsz; i++) {
500: /* Parse blindly til the backslash delimiter. */
501: if ('\\' != m->value[i]) {
502: val[j++] = m->value[i];
503: val[j] = '\0';
504: continue;
505: } else if (i == realsz - 1)
506: texierr(p, "trailing argument name delimiter");
507:
508: /* Double-backslash is escaped. */
509: if ('\\' == m->value[i + 1]) {
510: val[j++] = m->value[i++];
511: val[j] = '\0';
512: continue;
513: }
514:
515: assert('\\' == m->value[i] && i < realsz - 1);
516:
517: /* Parse to terminating delimiter. */
518: /* FIXME: embedded, escaped delimiters? */
519: for (start = end = i + 1; end < realsz; end++)
520: if ('\\' == m->value[end])
521: break;
522: if (end == realsz)
523: texierr(p, "unterminated argument name");
524:
525: for (k = 0; k < aasz; k++) {
526: if ((ssz = strlen(m->args[k])) != (end - start))
527: continue;
528: if (strncmp(&m->value[start], m->args[k], ssz))
529: continue;
530: break;
531: }
532:
533: /*
534: * Argument didn't exist in argument table.
1.14 kristaps 535: * Just ignore it.
1.7 kristaps 536: */
537: if (k == aasz) {
1.14 kristaps 538: i = end;
1.7 kristaps 539: continue;
540: }
541:
542: if (strlen(args[k]) > ssz) {
543: valsz += strlen(args[k]);
544: val = realloc(val, valsz + 1);
545: if (NULL == val)
546: texiabort(p, NULL);
547: }
548:
1.11 kristaps 549: for (cp = args[k]; '\0' != *cp; cp++)
550: val[j++] = *cp;
551:
552: val[j] = '\0';
1.7 kristaps 553: i = end;
554: }
555:
1.21 kristaps 556: texisplice(p, val, strlen(val), *pos);
1.7 kristaps 557:
558: for (i = 0; i < asz; i++)
559: free(args[i]);
560: free(args);
561: free(val);
562: }
563:
1.1 kristaps 564: /*
565: * Output a free-form word in the input stream, progressing to the next
566: * command or white-space.
567: * This also will advance the input stream.
568: */
569: static void
1.14 kristaps 570: parseword(struct texi *p, size_t *pos, char extra)
1.1 kristaps 571: {
572:
1.25 kristaps 573: /*
574: * Some line control: if we (non-macro, non-literal) already
575: * have more than 72 characters written to the screen, then
576: * output a newline before getting started.
577: */
1.1 kristaps 578: if (p->seenws && 0 == p->outmacro &&
579: p->outcol > 72 && 0 == p->literal)
580: texiputchar(p, '\n');
1.25 kristaps 581:
582: /* Usual padding in the case of seen whitespace. */
1.1 kristaps 583: if (p->seenws && p->outcol && 0 == p->literal)
584: texiputchar(p, ' ');
585:
586: p->seenws = 0;
587:
1.14 kristaps 588: while (*pos < BUFSZ(p) && ! ismspace(BUF(p)[*pos])) {
589: switch (BUF(p)[*pos]) {
1.1 kristaps 590: case ('@'):
591: case ('}'):
592: case ('{'):
593: return;
594: }
1.14 kristaps 595: if ('\0' != extra && BUF(p)[*pos] == extra)
1.1 kristaps 596: return;
1.14 kristaps 597: if (*pos < BUFSZ(p) - 1 &&
598: '`' == BUF(p)[*pos] &&
599: '`' == BUF(p)[*pos + 1]) {
1.1 kristaps 600: texiputchars(p, "\\(lq");
1.14 kristaps 601: advance(p, pos);
602: } else if (*pos < BUFSZ(p) - 1 &&
603: '\'' == BUF(p)[*pos] &&
604: '\'' == BUF(p)[*pos + 1]) {
1.1 kristaps 605: texiputchars(p, "\\(rq");
1.14 kristaps 606: advance(p, pos);
1.1 kristaps 607: } else
1.14 kristaps 608: texiputchar(p, BUF(p)[*pos]);
609: advance(p, pos);
1.1 kristaps 610: }
1.25 kristaps 611:
612: /*
613: * New sentence, new line:if we (non-macro, non-literal) see a
614: * period at the end of the last printed word, then open a
615: * newline.
616: */
617: if (0 == p->literal && 0 == p->outmacro &&
618: *pos < BUFSZ(p) && '.' == BUF(p)[*pos - 1])
619: texiputchar(p, '\n');
1.1 kristaps 620: }
621:
622: /*
623: * Look up the command at position "pos" in the buffer, returning it (or
624: * TEXICMD__MAX if none found) and setting "end" to be the absolute
625: * index after the command name.
626: */
627: enum texicmd
1.19 kristaps 628: texicmd(const struct texi *p, size_t pos, size_t *end, struct teximacro **macro)
1.1 kristaps 629: {
1.4 kristaps 630: size_t i, len, toksz;
1.1 kristaps 631:
1.14 kristaps 632: assert('@' == BUF(p)[pos]);
1.1 kristaps 633:
1.7 kristaps 634: if (NULL != macro)
635: *macro = NULL;
636:
1.14 kristaps 637: if ((*end = pos) == BUFSZ(p))
1.1 kristaps 638: return(TEXICMD__MAX);
1.14 kristaps 639: else if ((*end = ++pos) == BUFSZ(p))
1.1 kristaps 640: return(TEXICMD__MAX);
641:
642: /* Alphabetic commands are special. */
1.23 kristaps 643: if ( ! isalpha((unsigned int)BUF(p)[pos])) {
1.14 kristaps 644: if ((*end = pos + 1) == BUFSZ(p))
1.1 kristaps 645: return(TEXICMD__MAX);
646: for (i = 0; i < TEXICMD__MAX; i++) {
647: if (1 != texitoks[i].len)
648: continue;
1.14 kristaps 649: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], 1))
1.1 kristaps 650: return(i);
651: }
1.14 kristaps 652: texiwarn(p, "bad command: @%c", BUF(p)[pos]);
1.1 kristaps 653: return(TEXICMD__MAX);
654: }
655:
1.4 kristaps 656: /* Scan to the end of the possible command name. */
1.14 kristaps 657: for (*end = pos; *end < BUFSZ(p) && ! ismspace(BUF(p)[*end]); (*end)++)
658: if ((*end > pos && ('@' == BUF(p)[*end] ||
659: '{' == BUF(p)[*end] || '}' == BUF(p)[*end])))
1.1 kristaps 660: break;
661:
1.4 kristaps 662: /* Look for the command. */
1.1 kristaps 663: len = *end - pos;
664: for (i = 0; i < TEXICMD__MAX; i++) {
665: if (len != texitoks[i].len)
666: continue;
1.14 kristaps 667: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], len))
1.1 kristaps 668: return(i);
669: }
670:
1.4 kristaps 671: /* Look for it in our indices. */
672: for (i = 0; i < p->indexsz; i++) {
673: toksz = strlen(p->indexs[i]);
674: if (len != 5 + toksz)
675: continue;
1.14 kristaps 676: if (strncmp(&BUF(p)[pos], p->indexs[i], toksz))
1.4 kristaps 677: continue;
1.14 kristaps 678: if (0 == strncmp(&BUF(p)[pos + toksz], "index", 5))
1.7 kristaps 679: return(TEXICMD_USER_INDEX);
680: }
681:
682: for (i = 0; i < p->macrosz; i++) {
683: if (len != strlen(p->macros[i].key))
684: continue;
1.14 kristaps 685: if (strncmp(&BUF(p)[pos], p->macros[i].key, len))
1.7 kristaps 686: continue;
687: if (NULL != macro)
688: *macro = &p->macros[i];
689: return(TEXICMD__MAX);
1.4 kristaps 690: }
691:
1.14 kristaps 692: texiwarn(p, "bad command: @%.*s", (int)len, &BUF(p)[pos]);
1.1 kristaps 693: return(TEXICMD__MAX);
694: }
695:
696: /*
697: * Parse an argument from a bracketed command, e.g., @url{foo, baz}.
698: * Num should be set to the argument we're currently parsing, although
699: * it suffixes for it to be zero or non-zero.
700: * This will return 1 if there are more arguments, 0 otherwise.
701: * This will stop (returning 0) in the event of EOF or if we're not at a
702: * bracket for the zeroth parse.
703: */
704: int
1.14 kristaps 705: parsearg(struct texi *p, size_t *pos, size_t num)
1.1 kristaps 706: {
1.17 kristaps 707: size_t end, sv;
1.7 kristaps 708: enum texicmd cmd;
709: struct teximacro *macro;
1.1 kristaps 710:
1.14 kristaps 711: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
712: advance(p, pos);
713: if (*pos == BUFSZ(p) || (0 == num && '{' != BUF(p)[*pos]))
1.1 kristaps 714: return(0);
715: if (0 == num)
1.14 kristaps 716: advance(p, pos);
1.1 kristaps 717:
1.14 kristaps 718: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
719: switch (BUF(p)[*pos]) {
1.1 kristaps 720: case (','):
1.14 kristaps 721: advance(p, pos);
1.1 kristaps 722: return(1);
723: case ('}'):
1.14 kristaps 724: advance(p, pos);
1.1 kristaps 725: return(0);
726: case ('{'):
727: if (0 == p->ign)
728: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 729: advance(p, pos);
1.1 kristaps 730: continue;
731: case ('@'):
732: break;
733: default:
1.14 kristaps 734: parseword(p, pos, ',');
1.1 kristaps 735: continue;
736: }
737:
1.17 kristaps 738: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 739: cmd = texicmd(p, *pos, &end, ¯o);
740: advanceto(p, pos, end);
1.7 kristaps 741: if (NULL != macro)
1.17 kristaps 742: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 743: if (TEXICMD__MAX == cmd)
744: continue;
745: if (NULL != texitoks[cmd].fp)
1.14 kristaps 746: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 747: }
748: return(0);
749: }
750:
751: /*
752: * Parse until the end of a bracketed statement, e.g., @foo{bar baz}.
753: * This will stop in the event of EOF or if we're not at a bracket.
754: */
755: void
1.18 kristaps 756: parsebracket(struct texi *p, size_t *pos, int dostack)
1.1 kristaps 757: {
1.18 kristaps 758: size_t end, sv, stack;
1.7 kristaps 759: enum texicmd cmd;
760: struct teximacro *macro;
1.1 kristaps 761:
1.14 kristaps 762: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
763: advance(p, pos);
1.1 kristaps 764:
1.14 kristaps 765: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.1 kristaps 766: return;
1.14 kristaps 767: advance(p, pos);
1.1 kristaps 768:
1.18 kristaps 769: stack = 0;
1.14 kristaps 770: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
771: switch (BUF(p)[*pos]) {
1.1 kristaps 772: case ('}'):
1.18 kristaps 773: if (stack > 0) {
774: stack--;
775: advance(p, pos);
776: texiputchar(p, '}');
777: continue;
778: }
1.14 kristaps 779: advance(p, pos);
1.1 kristaps 780: return;
781: case ('{'):
1.18 kristaps 782: if (dostack) {
783: stack++;
784: advance(p, pos);
785: texiputchar(p, '{');
786: continue;
787: }
1.1 kristaps 788: if (0 == p->ign)
789: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 790: advance(p, pos);
1.1 kristaps 791: continue;
792: case ('@'):
793: break;
794: default:
1.14 kristaps 795: parseword(p, pos, '\0');
1.1 kristaps 796: continue;
797: }
798:
1.17 kristaps 799: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 800: cmd = texicmd(p, *pos, &end, ¯o);
801: advanceto(p, pos, end);
1.7 kristaps 802: if (NULL != macro)
1.17 kristaps 803: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 804: if (TEXICMD__MAX == cmd)
805: continue;
806: if (NULL != texitoks[cmd].fp)
1.14 kristaps 807: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 808: }
809: }
810:
811: /*
812: * This should be invoked when we're on a macro line and want to process
813: * to the end of the current input line, doing all of our macros along
814: * the way.
815: */
816: void
1.14 kristaps 817: parseeoln(struct texi *p, size_t *pos)
1.1 kristaps 818: {
1.17 kristaps 819: size_t end, sv;
1.7 kristaps 820: enum texicmd cmd;
821: struct teximacro *macro;
1.1 kristaps 822:
1.14 kristaps 823: while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) {
824: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 825: p->seenws = 1;
826: if (p->literal)
1.14 kristaps 827: texiputchar(p, BUF(p)[*pos]);
828: advance(p, pos);
1.1 kristaps 829: }
1.14 kristaps 830: switch (BUF(p)[*pos]) {
1.1 kristaps 831: case ('}'):
832: if (0 == p->ign)
833: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 834: advance(p, pos);
1.1 kristaps 835: continue;
836: case ('{'):
837: if (0 == p->ign)
838: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 839: advance(p, pos);
1.1 kristaps 840: continue;
841: case ('@'):
842: break;
843: default:
1.14 kristaps 844: parseword(p, pos, '\0');
1.1 kristaps 845: continue;
846: }
847:
1.17 kristaps 848: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 849: cmd = texicmd(p, *pos, &end, ¯o);
850: advanceto(p, pos, end);
1.7 kristaps 851: if (NULL != macro)
1.17 kristaps 852: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 853: if (TEXICMD__MAX == cmd)
854: continue;
855: if (NULL != texitoks[cmd].fp)
1.14 kristaps 856: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 857: }
1.14 kristaps 858:
859: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
860: advance(p, pos);
1.19 kristaps 861: }
862:
863: /*
864: * Peek to see if there's a command after subsequent whitespace.
865: * If so, return the macro identifier.
866: * This DOES NOT work with user-defined macros.
867: */
868: enum texicmd
869: peekcmd(const struct texi *p, size_t pos)
870: {
871: size_t end;
872:
873: while (pos < BUFSZ(p) && ismspace(BUF(p)[pos]))
874: pos++;
875: if (pos == BUFSZ(p) || '@' != BUF(p)[pos])
876: return(TEXICMD__MAX);
877: return(texicmd(p, pos, &end, NULL));
1.1 kristaps 878: }
879:
880: /*
881: * Parse a single word or command.
882: * This will return immediately at the EOF.
883: */
1.14 kristaps 884: static void
885: parsesingle(struct texi *p, size_t *pos)
1.1 kristaps 886: {
1.17 kristaps 887: size_t end, sv;
1.7 kristaps 888: enum texicmd cmd;
889: struct teximacro *macro;
1.1 kristaps 890:
1.14 kristaps 891: if ((*pos = advancenext(p, pos)) >= BUFSZ(p))
1.1 kristaps 892: return;
893:
1.14 kristaps 894: switch (BUF(p)[*pos]) {
1.1 kristaps 895: case ('}'):
896: if (0 == p->ign)
897: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 898: advance(p, pos);
1.1 kristaps 899: return;
900: case ('{'):
901: if (0 == p->ign)
902: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 903: advance(p, pos);
1.1 kristaps 904: return;
905: case ('@'):
906: break;
907: default:
1.14 kristaps 908: parseword(p, pos, '\0');
1.1 kristaps 909: return;
910: }
911:
1.17 kristaps 912: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 913: cmd = texicmd(p, *pos, &end, ¯o);
914: advanceto(p, pos, end);
1.7 kristaps 915: if (NULL != macro)
1.17 kristaps 916: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 917: if (TEXICMD__MAX == cmd)
918: return;
919: if (NULL != texitoks[cmd].fp)
1.14 kristaps 920: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 921: }
922:
923: /*
924: * This is used in the @deffn type of command.
925: * These have an arbitrary number of line arguments; however, these
926: * arguments may or may not be surrounded by brackets.
927: * In this function, we parse each one as either a bracketed or
928: * non-bracketed argument, returning 0 when we've reached the end of
929: * line or 1 otherwise.
930: */
931: int
1.14 kristaps 932: parselinearg(struct texi *p, size_t *pos)
1.1 kristaps 933: {
934:
1.14 kristaps 935: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 936: p->seenws = 1;
1.14 kristaps 937: advance(p, pos);
1.1 kristaps 938: }
939:
1.14 kristaps 940: if (*pos < BUFSZ(p) && '{' == BUF(p)[*pos])
1.18 kristaps 941: parsebracket(p, pos, 0);
1.14 kristaps 942: else if (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos])
943: parsesingle(p, pos);
1.1 kristaps 944: else
945: return(0);
946:
947: return(1);
948: }
949:
950: /*
951: * Parse til the end of the buffer.
952: */
1.14 kristaps 953: static void
954: parseeof(struct texi *p)
1.1 kristaps 955: {
956: size_t pos;
957:
1.14 kristaps 958: for (pos = 0; pos < BUFSZ(p); )
959: parsesingle(p, &pos);
1.1 kristaps 960: }
961:
1.8 kristaps 962: void
1.21 kristaps 963: texisplice(struct texi *p, const char *buf, size_t sz, size_t pos)
1.8 kristaps 964: {
1.14 kristaps 965: char *cp;
966: struct texifile *f;
1.8 kristaps 967:
1.14 kristaps 968: assert(p->filepos > 0);
969: f = &p->files[p->filepos - 1];
1.8 kristaps 970:
1.14 kristaps 971: if (f->mapsz + sz > f->mapmaxsz) {
972: f->mapmaxsz = f->mapsz + sz + 1024;
973: cp = realloc(f->map, f->mapmaxsz);
974: if (NULL == cp)
975: texiabort(p, NULL);
976: f->map = cp;
977: }
1.8 kristaps 978:
1.15 kristaps 979: f->insplice += sz;
1.21 kristaps 980: memmove(f->map + pos + sz, f->map + pos, f->mapsz - pos);
981: memcpy(f->map + pos, buf, sz);
1.14 kristaps 982: f->mapsz += sz;
1.8 kristaps 983: }
984:
985: /*
1.1 kristaps 986: * Parse a block sequence until we have the "@end endtoken" command
987: * invocation.
988: * This will return immediately at EOF.
989: */
990: void
1.14 kristaps 991: parseto(struct texi *p, size_t *pos, const char *endtoken)
1.1 kristaps 992: {
1.17 kristaps 993: size_t end, sv;
1.7 kristaps 994: enum texicmd cmd;
995: size_t endtoksz;
996: struct teximacro *macro;
1.1 kristaps 997:
998: endtoksz = strlen(endtoken);
999: assert(endtoksz > 0);
1000:
1.14 kristaps 1001: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
1002: switch (BUF(p)[*pos]) {
1.1 kristaps 1003: case ('}'):
1004: if (0 == p->ign)
1005: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 1006: advance(p, pos);
1.1 kristaps 1007: continue;
1008: case ('{'):
1009: if (0 == p->ign)
1010: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 1011: advance(p, pos);
1.1 kristaps 1012: continue;
1013: case ('@'):
1014: break;
1015: default:
1.14 kristaps 1016: parseword(p, pos, '\0');
1.1 kristaps 1017: continue;
1018: }
1019:
1.17 kristaps 1020: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 1021: cmd = texicmd(p, *pos, &end, ¯o);
1022: advanceto(p, pos, end);
1.1 kristaps 1023: if (TEXICMD_END == cmd) {
1.14 kristaps 1024: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1025: advance(p, pos);
1.1 kristaps 1026: /*
1027: * FIXME: check the full word, not just its
1028: * initial substring!
1029: */
1.14 kristaps 1030: if (BUFSZ(p) - *pos >= endtoksz && 0 == strncmp
1031: (&BUF(p)[*pos], endtoken, endtoksz)) {
1032: advanceeoln(p, pos, 0);
1.1 kristaps 1033: break;
1034: }
1035: if (0 == p->ign)
1036: texiwarn(p, "unexpected \"end\"");
1.14 kristaps 1037: advanceeoln(p, pos, 0);
1.1 kristaps 1038: continue;
1.7 kristaps 1039: }
1040: if (NULL != macro)
1.17 kristaps 1041: texiexecmacro(p, macro, sv, pos);
1.7 kristaps 1042: if (TEXICMD__MAX == cmd)
1043: continue;
1044: if (NULL != texitoks[cmd].fp)
1.14 kristaps 1045: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 1046: }
1047: }
1048:
1049: /*
1.12 kristaps 1050: * Like parsefile() but used for reading from stdandard input.
1051: * This can only be called for the first file!
1052: */
1053: void
1054: parsestdin(struct texi *p)
1055: {
1056: struct texifile *f;
1057: ssize_t ssz;
1058:
1059: assert(0 == p->filepos);
1060: f = &p->files[p->filepos];
1061: memset(f, 0, sizeof(struct texifile));
1062:
1063: f->type = TEXISRC_STDIN;
1064: f->name = "<stdin>";
1065:
1.14 kristaps 1066: for (f->mapsz = 0; ; f->mapsz += (size_t)ssz) {
1067: if (f->mapsz == f->mapmaxsz) {
1068: if (f->mapmaxsz == (1U << 31))
1.12 kristaps 1069: texierr(p, "stdin buffer too long");
1.14 kristaps 1070: f->mapmaxsz = f->mapmaxsz > 65536 / 2 ?
1071: 2 * f->mapmaxsz : 65536;
1072: f->map = realloc(f->map, f->mapmaxsz);
1.12 kristaps 1073: if (NULL == f->map)
1074: texiabort(p, NULL);
1075: }
1.14 kristaps 1076: ssz = read(STDIN_FILENO, f->map +
1077: (int)f->mapsz, f->mapmaxsz - f->mapsz);
1.12 kristaps 1078: if (0 == ssz)
1079: break;
1080: else if (-1 == ssz)
1081: texiabort(p, NULL);
1082: }
1083:
1084: p->filepos++;
1.14 kristaps 1085: parseeof(p);
1.12 kristaps 1086: texifilepop(p);
1087: }
1088:
1089: /*
1.1 kristaps 1090: * Memory-map the file "fname" and begin parsing it unless "parse" is
1091: * zero, in which case we just dump the file to stdout (making sure it
1092: * doesn't trip up mdoc(7) along the way).
1093: * This can be called in a nested context.
1094: */
1095: void
1096: parsefile(struct texi *p, const char *fname, int parse)
1097: {
1098: struct texifile *f;
1099: int fd;
1100: struct stat st;
1101: size_t i;
1.14 kristaps 1102: char *map;
1.1 kristaps 1103:
1.5 kristaps 1104: if (64 == p->filepos)
1.6 kristaps 1105: texierr(p, "too many open files");
1.1 kristaps 1106: f = &p->files[p->filepos];
1107: memset(f, 0, sizeof(struct texifile));
1108:
1.12 kristaps 1109: f->type = TEXISRC_FILE;
1.1 kristaps 1110: f->name = fname;
1111: if (-1 == (fd = open(fname, O_RDONLY, 0))) {
1112: texiabort(p, fname);
1113: } else if (-1 == fstat(fd, &st)) {
1114: close(fd);
1115: texiabort(p, fname);
1116: }
1117:
1.14 kristaps 1118: f->mapsz = f->mapmaxsz = st.st_size;
1119: map = mmap(NULL, f->mapsz,
1.1 kristaps 1120: PROT_READ, MAP_SHARED, fd, 0);
1121: close(fd);
1122:
1.14 kristaps 1123: if (MAP_FAILED == map)
1.1 kristaps 1124: texiabort(p, fname);
1125:
1126: if ( ! parse) {
1.13 kristaps 1127: for (i = 0; i < f->mapsz; i++)
1.14 kristaps 1128: texiputchar(p, map[i]);
1.13 kristaps 1129: if (p->outcol)
1130: texiputchar(p, '\n');
1.14 kristaps 1131: munmap(map, f->mapsz);
1132: return;
1133: }
1134:
1135: p->filepos++;
1136: f->map = malloc(f->mapsz);
1137: memcpy(f->map, map, f->mapsz);
1138: munmap(map, f->mapsz);
1139: parseeof(p);
1.1 kristaps 1140: texifilepop(p);
1141: }
1142:
1.2 kristaps 1143: /*
1144: * Look up the value to a stored pair's value starting in "buf" from
1145: * start to end.
1146: * Return the pointer to the value memory, which can be NULL if the
1147: * pointer key does not exist.
1148: * The pointer can point to NULL if the value has been unset.
1149: */
1150: static char **
1.14 kristaps 1151: valuequery(const struct texi *p, size_t start, size_t end)
1.2 kristaps 1152: {
1153: size_t i, sz, len;
1154:
1155: assert(end >= start);
1156: /* Ignore zero-length. */
1157: if (0 == (len = (end - start)))
1158: return(NULL);
1159: for (i = 0; i < p->valsz; i++) {
1160: sz = strlen(p->vals[i].key);
1161: if (sz != len)
1162: continue;
1.14 kristaps 1163: if (0 == strncmp(p->vals[i].key, &BUF(p)[start], len))
1.2 kristaps 1164: return(&p->vals[i].value);
1165: }
1166: return(NULL);
1167: }
1168:
1169: /*
1170: * Parse a key until the end of line, e.g., @clear foo\n, and return the
1171: * pointer to its value via valuequery().
1172: */
1173: static char **
1.14 kristaps 1174: valuelquery(struct texi *p, size_t *pos)
1.2 kristaps 1175: {
1176: size_t start, end;
1177: char **ret;
1178:
1.14 kristaps 1179: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1180: advance(p, pos);
1181: if (*pos == BUFSZ(p))
1.2 kristaps 1182: return(NULL);
1.14 kristaps 1183: for (start = end = *pos; end < BUFSZ(p); end++)
1184: if ('\n' == BUF(p)[end])
1.2 kristaps 1185: break;
1.14 kristaps 1186: advanceto(p, pos, end);
1187: if (*pos < BUFSZ(p)) {
1188: assert('\n' == BUF(p)[*pos]);
1189: advance(p, pos);
1.2 kristaps 1190: }
1.14 kristaps 1191: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1192: return(NULL);
1193: return(ret);
1194: }
1195:
1196: void
1.14 kristaps 1197: valuelclear(struct texi *p, size_t *pos)
1.2 kristaps 1198: {
1199: char **ret;
1200:
1.14 kristaps 1201: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1202: return;
1203: free(*ret);
1204: *ret = NULL;
1205: }
1206:
1207: const char *
1.14 kristaps 1208: valuellookup(struct texi *p, size_t *pos)
1.2 kristaps 1209: {
1210: char **ret;
1211:
1.14 kristaps 1212: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1213: return(NULL);
1214: return(*ret);
1215: }
1216:
1217: /*
1218: * Parse a key from a bracketed string, e.g., @value{foo}, and return
1219: * the pointer to its value.
1220: * If the returned pointer is NULL, either there was no string within
1221: * the brackets (or no brackets), or the value was not found, or the
1222: * value had previously been unset.
1223: */
1224: const char *
1.14 kristaps 1225: valueblookup(struct texi *p, size_t *pos)
1.2 kristaps 1226: {
1227: size_t start, end;
1228: char **ret;
1229:
1.14 kristaps 1230: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1231: advance(p, pos);
1232: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.2 kristaps 1233: return(NULL);
1.14 kristaps 1234: advance(p, pos);
1235: for (start = end = *pos; end < BUFSZ(p); end++)
1236: if ('}' == BUF(p)[end])
1.2 kristaps 1237: break;
1.14 kristaps 1238: advanceto(p, pos, end);
1239: if (*pos < BUFSZ(p)) {
1240: assert('}' == BUF(p)[*pos]);
1241: advance(p, pos);
1.2 kristaps 1242: }
1.14 kristaps 1243: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1244: return(NULL);
1245: return(*ret);
1246: }
1247:
1248: void
1249: valueadd(struct texi *p, char *key, char *val)
1250: {
1251: size_t i;
1252:
1253: assert(NULL != key);
1254: assert(NULL != val);
1255:
1256: for (i = 0; i < p->valsz; i++)
1257: if (0 == strcmp(p->vals[i].key, key))
1258: break;
1259:
1260: if (i < p->valsz) {
1261: free(key);
1262: free(p->vals[i].value);
1263: p->vals[i].value = val;
1264: } else {
1.4 kristaps 1265: /* FIXME: reallocarray() */
1.2 kristaps 1266: p->vals = realloc(p->vals,
1267: (p->valsz + 1) *
1268: sizeof(struct texivalue));
1.4 kristaps 1269: if (NULL == p->vals)
1270: texiabort(p, NULL);
1.2 kristaps 1271: p->vals[p->valsz].key = key;
1272: p->vals[p->valsz].value = val;
1273: p->valsz++;
1274: }
1.7 kristaps 1275: }
1276:
1277: /*
1278: * Take the arguments to a macro, e.g., @foo{bar, baz, xyzzy} (or the
1279: * declaration form, @macro foo {arg1, ...}) and textually convert it to
1280: * an array of arguments of size "argsz".
1281: * These need to be freed individually and as a whole.
1282: * NOTE: this will puke on @, or @} macros, which can trick it into
1283: * stopping argument parsing earlier.
1284: * Ergo, textual: this doesn't interpret the arguments in any way.
1285: */
1286: char **
1.14 kristaps 1287: argparse(struct texi *p, size_t *pos, size_t *argsz, size_t hint)
1.7 kristaps 1288: {
1289: char **args;
1290: size_t start, end, stack;
1291:
1.14 kristaps 1292: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1293: advance(p, pos);
1.7 kristaps 1294:
1295: args = NULL;
1296: *argsz = 0;
1297:
1.17 kristaps 1298: if (*pos == BUFSZ(p))
1299: return(args);
1300:
1.14 kristaps 1301: if ('{' != BUF(p)[*pos] && hint) {
1.10 kristaps 1302: /*
1303: * Special case: if we encounter an unbracketed argument
1304: * and we're being invoked with non-zero arguments
1305: * (versus being set, i.e., hint>0), then parse until
1306: * the end of line.
1307: */
1308: *argsz = 1;
1309: args = calloc(1, sizeof(char *));
1310: if (NULL == args)
1311: texiabort(p, NULL);
1312: start = *pos;
1.14 kristaps 1313: while (*pos < BUFSZ(p)) {
1314: if ('\n' == BUF(p)[*pos])
1.10 kristaps 1315: break;
1.14 kristaps 1316: advance(p, pos);
1.10 kristaps 1317: }
1318: args[0] = malloc(*pos - start + 1);
1.14 kristaps 1319: memcpy(args[0], &BUF(p)[start], *pos - start);
1.10 kristaps 1320: args[0][*pos - start] = '\0';
1.14 kristaps 1321: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
1322: advance(p, pos);
1.10 kristaps 1323: return(args);
1.14 kristaps 1324: } else if ('{' != BUF(p)[*pos])
1.7 kristaps 1325: return(args);
1.17 kristaps 1326:
1327: assert('{' == BUF(p)[*pos]);
1.7 kristaps 1328:
1329: /* Parse til the closing '}', putting into the array. */
1.14 kristaps 1330: advance(p, pos);
1331: while (*pos < BUFSZ(p)) {
1332: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1333: advance(p, pos);
1.7 kristaps 1334: start = *pos;
1335: stack = 0;
1.14 kristaps 1336: while (*pos < BUFSZ(p)) {
1.7 kristaps 1337: /*
1338: * According to the manual, commas within
1339: * embedded commands are escaped.
1340: * We keep track of embedded-ness in the "stack"
1341: * state anyway, so this is free.
1342: */
1.14 kristaps 1343: if (',' == BUF(p)[*pos] && 0 == stack && 1 != hint)
1.7 kristaps 1344: break;
1.14 kristaps 1345: else if (0 == stack && '}' == BUF(p)[*pos])
1.7 kristaps 1346: break;
1.14 kristaps 1347: else if (0 != stack && '}' == BUF(p)[*pos])
1.7 kristaps 1348: stack--;
1.14 kristaps 1349: else if ('{' == BUF(p)[*pos])
1.7 kristaps 1350: stack++;
1.14 kristaps 1351: advance(p, pos);
1.7 kristaps 1352: }
1353: if (stack)
1354: texiwarn(p, "unterminated macro "
1355: "in macro arguments");
1.14 kristaps 1356: if ((end = *pos) == BUFSZ(p))
1.7 kristaps 1357: break;
1358: /* Test for zero-length '{ }'. */
1.14 kristaps 1359: if (start == end && '}' == BUF(p)[*pos] && 0 == *argsz)
1.7 kristaps 1360: break;
1361: /* FIXME: use reallocarray. */
1362: args = realloc
1363: (args, sizeof(char *) *
1364: (*argsz + 1));
1365: if (NULL == args)
1366: texiabort(p, NULL);
1367: args[*argsz] = malloc(end - start + 1);
1368: if (NULL == args[*argsz])
1369: texiabort(p, NULL);
1370: memcpy(args[*argsz],
1.14 kristaps 1371: &BUF(p)[start], end - start);
1.7 kristaps 1372: args[*argsz][end - start] = '\0';
1373: (*argsz)++;
1.14 kristaps 1374: if ('}' == BUF(p)[*pos])
1.7 kristaps 1375: break;
1.14 kristaps 1376: advance(p, pos);
1.7 kristaps 1377: }
1378:
1.14 kristaps 1379: if (*pos == BUFSZ(p))
1.7 kristaps 1380: texierr(p, "unterminated arguments");
1.14 kristaps 1381: assert('}' == BUF(p)[*pos]);
1382: advance(p, pos);
1.7 kristaps 1383: return(args);
1.2 kristaps 1384: }
1.20 kristaps 1385:
1386: /*
1387: * If we're printing chapters, then do some naviation here and then
1388: * close our outfile.
1389: * I want to call this the SEE ALSO section, but that's not really what
1390: * it is: we'll refer to the "initial" (top) node and the next and
1391: * previous chapters.
1392: */
1393: void
1394: teximdocclose(struct texi *p, int last)
1395: {
1396: char buf[PATH_MAX];
1397:
1398: if (NULL == p->chapters || 0 == p->chapnum)
1399: return;
1400:
1401: teximacro(p, "Sh INFO NAVIGATION");
1402:
1403: /* Print a reference to the "top" node. */
1404: if (p->chapnum > 1) {
1.22 kristaps 1405: texiputchars(p, "Top node,");
1.20 kristaps 1406: snprintf(buf, sizeof(buf), "node1 7");
1407: teximacroopen(p, "Xr ");
1408: texiputchars(p, buf);
1.22 kristaps 1409: texiputchars(p, " ;");
1.20 kristaps 1410: teximacroclose(p);
1411: }
1412:
1413: /* Print a reference to the previous node. */
1414: if (p->chapnum > 2) {
1.22 kristaps 1415: texiputchars(p, "previous node,");
1.20 kristaps 1416: snprintf(buf, sizeof(buf),
1417: "node%zu 7", p->chapnum - 1);
1418: teximacroopen(p, "Xr ");
1419: texiputchars(p, buf);
1420: if ( ! last)
1.22 kristaps 1421: texiputchars(p, " ;");
1.20 kristaps 1422: teximacroclose(p);
1423: }
1424:
1425: /* Print a reference to the next node. */
1426: if ( ! last) {
1.22 kristaps 1427: if (1 == p->chapnum)
1428: texiputchars(p, "Next node,");
1429: else
1430: texiputchars(p, "next node,");
1.20 kristaps 1431: snprintf(buf, sizeof(buf),
1432: "node%zu 7", p->chapnum + 1);
1433: teximacroopen(p, "Xr ");
1434: texiputchars(p, buf);
1435: teximacroclose(p);
1436: }
1437:
1438: fclose(p->outfile);
1439: }
1440:
1441: /*
1442: * Open a mdoc(7) context.
1443: * If we're printing chapters, then open the outfile here, too.
1444: * Otherwise just print the mdoc(7) prologue.
1445: */
1446: void
1.21 kristaps 1447: teximdocopen(struct texi *p, size_t *pos)
1.20 kristaps 1448: {
1449: const char *cp;
1450: time_t t;
1451: char date[32];
1452: char fname[PATH_MAX];
1453:
1454: if (NULL != p->chapters) {
1455: snprintf(fname, sizeof(fname), "%s/node%zu.7",
1456: p->chapters, ++p->chapnum);
1457: p->outfile = fopen(fname, "w");
1458: if (NULL == p->outfile)
1459: texiabort(p, fname);
1460: }
1461:
1462: /*
1463: * Here we print our standard mdoc(7) prologue.
1464: * We use the title set with @settitle for the `Nd' description
1465: * and the source document filename (the first one as invoked on
1466: * the command line) for the title.
1467: * The date is set to the current date.
1468: */
1469: t = time(NULL);
1470: strftime(date, sizeof(date), "%F", localtime(&t));
1471:
1472: teximacroopen(p, "Dd");
1473: texiputchars(p, date);
1474: teximacroclose(p);
1475: teximacroopen(p, "Dt");
1476: for (cp = p->title; '\0' != *cp; cp++)
1477: texiputchar(p, toupper((unsigned int)*cp));
1478: texiputchars(p, " 7");
1479: teximacroclose(p);
1480: teximacro(p, "Os");
1481: teximacro(p, "Sh NAME");
1482: teximacroopen(p, "Nm");
1483: for (cp = p->title; '\0' != *cp; cp++)
1484: texiputchar(p, *cp);
1485: teximacroclose(p);
1486: teximacroopen(p, "Nd");
1.21 kristaps 1487: /*
1488: * The subtitle `Nd' can consist of arbitrary macros, so paste
1489: * it and parse to the end of the line.
1490: */
1491: if (NULL != p->subtitle) {
1492: texisplice(p, p->subtitle, strlen(p->subtitle), *pos);
1493: parseeoln(p, pos);
1494: } else
1.20 kristaps 1495: texiputchars(p, "Unknown description");
1496: teximacroclose(p);
1497: p->seenvs = 1;
1498: }
1499:
CVSweb