Annotation of texi2mdoc/util.c, Revision 1.28
1.28 ! kristaps 1: /* $Id: util.c,v 1.27 2015/03/05 09:36:41 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: if ('\n' == c) {
195: p->outcol = 0;
196: p->seenws = 0;
197: } else
198: p->outcol++;
199: }
200:
201: /*
1.13 kristaps 202: * Put an opaque series of characters.
203: * Characters starting a line with a control character are escaped, but
204: * that's it, so don't use this for non-controlled sequences of text.
1.1 kristaps 205: */
206: void
207: texiputchars(struct texi *p, const char *s)
208: {
209:
1.13 kristaps 210: if (p->ign)
211: return;
212: if ('.' == *s && 0 == p->outcol)
1.20 kristaps 213: fputs("\\&", p->outfile);
1.13 kristaps 214: if ('\'' == *s && 0 == p->outcol)
1.20 kristaps 215: fputs("\\&", p->outfile);
1.23 kristaps 216: if (p->uppercase)
217: for ( ; '\0' != *s; s++)
218: p->outcol += fputc(toupper
219: ((unsigned int)*s), p->outfile);
220: else
221: p->outcol += fputs(s, p->outfile);
1.9 kristaps 222: }
223:
224: /*
225: * This puts all characters onto the output stream but makes sure to
226: * escape mdoc(7) slashes.
1.14 kristaps 227: * FIXME: useless.
1.9 kristaps 228: */
229: void
1.14 kristaps 230: texiputbuf(struct texi *p, size_t start, size_t end)
1.9 kristaps 231: {
232:
1.14 kristaps 233: for ( ; start < end; start++)
234: texiputchar(p, BUF(p)[start]);
1.1 kristaps 235: }
236:
237: /*
238: * Close an mdoc(7) macro opened with teximacroopen().
239: * If there are no more macros on the line, prints a newline.
240: */
241: void
242: teximacroclose(struct texi *p)
243: {
244:
245: if (p->ign)
246: return;
247:
248: if (0 == --p->outmacro) {
1.20 kristaps 249: fputc('\n', p->outfile);
1.1 kristaps 250: p->outcol = p->seenws = 0;
251: }
1.27 kristaps 252: p->seenvs = 0;
1.1 kristaps 253: }
254:
255: /*
256: * Open a mdoc(7) macro.
257: * This is used for line macros, e.g., Qq [foo bar baz].
258: * It can be invoked for nested macros, e.g., Qq Li foo .
259: * TODO: flush-right punctuation (e.g., parenthesis).
260: */
261: void
262: teximacroopen(struct texi *p, const char *s)
263: {
264: int rc;
265:
266: if (p->ign)
267: return;
268:
269: if (p->outcol && 0 == p->outmacro) {
1.20 kristaps 270: fputc('\n', p->outfile);
1.1 kristaps 271: p->outcol = 0;
272: }
273:
274: if (0 == p->outmacro)
1.20 kristaps 275: fputc('.', p->outfile);
1.1 kristaps 276: else
1.20 kristaps 277: fputc(' ', p->outfile);
1.1 kristaps 278:
1.20 kristaps 279: if (EOF != (rc = fputs(s, p->outfile)))
1.1 kristaps 280: p->outcol += rc;
281:
1.20 kristaps 282: fputc(' ', p->outfile);
1.1 kristaps 283: p->outcol++;
284: p->outmacro++;
285: p->seenws = 0;
1.27 kristaps 286: p->seenvs = 0;
1.1 kristaps 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;
1.27 kristaps 311: p->seenvs = 0;
1.1 kristaps 312: }
313:
314: /*
315: * Introduce vertical space during normal (non-macro) input.
316: */
317: void
318: texivspace(struct texi *p)
319: {
320:
1.27 kristaps 321: if (TEXILIST_TABLE != p->list)
322: teximacro(p, "Pp");
1.1 kristaps 323: }
324:
325: /*
326: * Advance by a single byte in the input stream, adjusting our location
327: * in the current input file.
328: */
329: void
1.14 kristaps 330: advance(struct texi *p, size_t *pos)
1.1 kristaps 331: {
1.15 kristaps 332: struct texifile *f;
1.1 kristaps 333:
1.15 kristaps 334: f = &p->files[p->filepos - 1];
335:
336: if (0 == f->insplice) {
337: if ('\n' == BUF(p)[*pos]) {
338: f->line++;
339: f->col = 0;
340: } else
341: f->col++;
1.17 kristaps 342: } else {
1.15 kristaps 343: --f->insplice;
1.17 kristaps 344: if (0 == f->insplice)
345: f->depth = 0;
346: }
1.1 kristaps 347:
348: (*pos)++;
349: }
350:
351: /*
352: * It's common to wait punctuation to float on the right side of macro
353: * lines in mdoc(7), e.g., ".Em hello ) ."
354: * This function does so, and should be called before teximacroclose().
355: * It will detect that it's the last in the nested macros and
356: * appropriately flush-left punctuation alongside the macro.
357: */
358: void
1.14 kristaps 359: texipunctuate(struct texi *p, size_t *pos)
1.1 kristaps 360: {
361: size_t start, end;
362:
363: if (1 != p->outmacro)
364: return;
365:
1.14 kristaps 366: for (start = end = *pos; end < BUFSZ(p); end++) {
367: switch (BUF(p)[end]) {
1.1 kristaps 368: case (','):
369: case (')'):
370: case ('.'):
371: case ('"'):
372: case (':'):
1.22 kristaps 373: case (';'):
1.1 kristaps 374: case ('!'):
375: case ('?'):
376: continue;
377: default:
378: break;
379: }
380: break;
381: }
382: if (end == *pos)
383: return;
1.14 kristaps 384: if (end + 1 == BUFSZ(p) || ' ' == BUF(p)[end] ||
385: '\n' == BUF(p)[end]) {
1.1 kristaps 386: for ( ; start < end; start++) {
387: texiputchar(p, ' ');
1.14 kristaps 388: texiputchar(p, BUF(p)[start]);
389: advance(p, pos);
1.1 kristaps 390: }
391: }
392: }
393:
394: /*
395: * Advance to the next non-whitespace word in the input stream.
396: * If we're in literal mode, then print all of the whitespace as we're
397: * doing so.
398: */
399: static size_t
1.14 kristaps 400: advancenext(struct texi *p, size_t *pos)
1.1 kristaps 401: {
402:
403: if (p->literal) {
1.14 kristaps 404: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) {
405: texiputchar(p, BUF(p)[*pos]);
406: advance(p, pos);
1.1 kristaps 407: }
408: return(*pos);
409: }
410:
1.14 kristaps 411: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) {
1.1 kristaps 412: p->seenws = 1;
1.14 kristaps 413: advance(p, pos);
1.1 kristaps 414: }
415: return(*pos);
416: }
417:
418: /*
419: * Advance to the EOLN in the input stream.
1.22 kristaps 420: * This will skip over '@' markers in an effort to ignore escaped
421: * newlines.
1.1 kristaps 422: */
423: size_t
1.14 kristaps 424: advanceeoln(struct texi *p, size_t *pos, int consumenl)
1.1 kristaps 425: {
426:
1.22 kristaps 427: while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) {
428: if ('@' == BUF(p)[*pos])
429: advance(p, pos);
1.14 kristaps 430: advance(p, pos);
1.22 kristaps 431: }
1.14 kristaps 432: if (*pos < BUFSZ(p) && consumenl)
433: advance(p, pos);
1.1 kristaps 434: return(*pos);
435: }
436:
437: /*
438: * Advance to position "end", which is an absolute position in the
439: * current buffer greater than or equal to the current position.
440: */
441: void
1.14 kristaps 442: advanceto(struct texi *p, size_t *pos, size_t end)
1.1 kristaps 443: {
444:
445: assert(*pos <= end);
446: while (*pos < end)
1.14 kristaps 447: advance(p, pos);
1.1 kristaps 448: }
449:
1.7 kristaps 450: static void
1.17 kristaps 451: texiexecmacro(struct texi *p, struct teximacro *m, size_t sv, size_t *pos)
1.7 kristaps 452: {
1.11 kristaps 453: size_t valsz, realsz, aasz, asz,
454: ssz, i, j, k, start, end;
455: char *val;
456: char **args;
457: const char *cp;
1.7 kristaps 458:
1.17 kristaps 459: /* Disregard empty macros. */
1.22 kristaps 460: if (0 == (valsz = realsz = strlen(m->value))) {
461: args = argparse(p, pos, &asz, m->argsz);
462: for (i = 0; i < asz; i++)
463: free(args[i]);
464: free(args);
1.17 kristaps 465: return;
1.22 kristaps 466: }
1.17 kristaps 467:
468: /*
469: * This is important: it protect us from macros that invoke more
470: * macros, possibly going on infinitely.
471: * We use "sv" instead of the current position because we might
472: * be invoked at the end of the macro (i.e., insplice == 0).
473: * The "sv" value was initialised at the start of the macro.
474: */
475: if (sv > 0)
1.24 kristaps 476: if (++p->files[p->filepos - 1].depth > 64)
1.17 kristaps 477: texierr(p, "maximium recursive depth");
478:
1.14 kristaps 479: args = argparse(p, pos, &asz, m->argsz);
1.7 kristaps 480: if (asz != m->argsz)
481: texiwarn(p, "invalid macro argument length");
482: aasz = asz < m->argsz ? asz : m->argsz;
483:
484: if (0 == aasz) {
1.21 kristaps 485: texisplice(p, m->value, valsz, *pos);
1.7 kristaps 486: return;
487: }
488:
489: val = strdup(m->value);
490:
491: for (i = j = 0; i < realsz; i++) {
492: /* Parse blindly til the backslash delimiter. */
493: if ('\\' != m->value[i]) {
494: val[j++] = m->value[i];
495: val[j] = '\0';
496: continue;
497: } else if (i == realsz - 1)
498: texierr(p, "trailing argument name delimiter");
499:
500: /* Double-backslash is escaped. */
501: if ('\\' == m->value[i + 1]) {
502: val[j++] = m->value[i++];
503: val[j] = '\0';
504: continue;
505: }
506:
507: assert('\\' == m->value[i] && i < realsz - 1);
508:
509: /* Parse to terminating delimiter. */
510: /* FIXME: embedded, escaped delimiters? */
511: for (start = end = i + 1; end < realsz; end++)
512: if ('\\' == m->value[end])
513: break;
514: if (end == realsz)
515: texierr(p, "unterminated argument name");
516:
517: for (k = 0; k < aasz; k++) {
518: if ((ssz = strlen(m->args[k])) != (end - start))
519: continue;
520: if (strncmp(&m->value[start], m->args[k], ssz))
521: continue;
522: break;
523: }
524:
525: /*
526: * Argument didn't exist in argument table.
1.14 kristaps 527: * Just ignore it.
1.7 kristaps 528: */
529: if (k == aasz) {
1.14 kristaps 530: i = end;
1.7 kristaps 531: continue;
532: }
533:
534: if (strlen(args[k]) > ssz) {
535: valsz += strlen(args[k]);
536: val = realloc(val, valsz + 1);
537: if (NULL == val)
538: texiabort(p, NULL);
539: }
540:
1.11 kristaps 541: for (cp = args[k]; '\0' != *cp; cp++)
542: val[j++] = *cp;
543:
544: val[j] = '\0';
1.7 kristaps 545: i = end;
546: }
547:
1.21 kristaps 548: texisplice(p, val, strlen(val), *pos);
1.7 kristaps 549:
550: for (i = 0; i < asz; i++)
551: free(args[i]);
552: free(args);
553: free(val);
554: }
555:
1.1 kristaps 556: /*
557: * Output a free-form word in the input stream, progressing to the next
558: * command or white-space.
559: * This also will advance the input stream.
560: */
561: static void
1.14 kristaps 562: parseword(struct texi *p, size_t *pos, char extra)
1.1 kristaps 563: {
564:
1.25 kristaps 565: /*
1.27 kristaps 566: * If a prior word had a terminating double-newline, then begin
567: * this text block with a `Pp'.
568: * We don't do this if we're in a literal context (we'll print
569: * out the newlines themselves) nor in a `TS' table.
570: */
571: if (p->seenvs && 0 == p->literal && TEXILIST_TABLE != p->list)
572: teximacro(p, "Pp");
573:
574: p->seenvs = 0;
575:
576: /*
1.25 kristaps 577: * Some line control: if we (non-macro, non-literal) already
578: * have more than 72 characters written to the screen, then
579: * output a newline before getting started.
580: */
1.1 kristaps 581: if (p->seenws && 0 == p->outmacro &&
582: p->outcol > 72 && 0 == p->literal)
583: texiputchar(p, '\n');
1.25 kristaps 584:
585: /* Usual padding in the case of seen whitespace. */
1.1 kristaps 586: if (p->seenws && p->outcol && 0 == p->literal)
587: texiputchar(p, ' ');
588:
589: p->seenws = 0;
590:
1.14 kristaps 591: while (*pos < BUFSZ(p) && ! ismspace(BUF(p)[*pos])) {
592: switch (BUF(p)[*pos]) {
1.1 kristaps 593: case ('@'):
594: case ('}'):
595: case ('{'):
596: return;
597: }
1.14 kristaps 598: if ('\0' != extra && BUF(p)[*pos] == extra)
1.1 kristaps 599: return;
1.28 ! kristaps 600:
! 601: if (p->literal) {
! 602: texiputchar(p, BUF(p)[*pos]);
! 603: advance(p, pos);
! 604: continue;
! 605: }
! 606:
! 607: if (*pos < BUFSZ(p) - 2 &&
! 608: '-' == BUF(p)[*pos] &&
! 609: '-' == BUF(p)[*pos + 1] &&
! 610: '-' == BUF(p)[*pos + 2]) {
! 611: texiputchars(p, "\\(em");
! 612: advance(p, pos);
! 613: advance(p, pos);
! 614: } else if (*pos < BUFSZ(p) - 1 &&
! 615: '-' == BUF(p)[*pos] &&
! 616: '-' == BUF(p)[*pos + 1]) {
! 617: texiputchars(p, "\\(en");
! 618: advance(p, pos);
! 619: } else if (*pos < BUFSZ(p) - 1 &&
1.14 kristaps 620: '`' == BUF(p)[*pos] &&
621: '`' == BUF(p)[*pos + 1]) {
1.1 kristaps 622: texiputchars(p, "\\(lq");
1.14 kristaps 623: advance(p, pos);
624: } else if (*pos < BUFSZ(p) - 1 &&
625: '\'' == BUF(p)[*pos] &&
626: '\'' == BUF(p)[*pos + 1]) {
1.1 kristaps 627: texiputchars(p, "\\(rq");
1.14 kristaps 628: advance(p, pos);
1.1 kristaps 629: } else
1.14 kristaps 630: texiputchar(p, BUF(p)[*pos]);
1.28 ! kristaps 631:
1.14 kristaps 632: advance(p, pos);
1.1 kristaps 633: }
1.25 kristaps 634:
1.27 kristaps 635: if (*pos + 1 < BUFSZ(p) &&
636: '\n' == BUF(p)[*pos] &&
637: '\n' == BUF(p)[*pos + 1])
638: p->seenvs = 1;
639:
1.25 kristaps 640: /*
641: * New sentence, new line:if we (non-macro, non-literal) see a
642: * period at the end of the last printed word, then open a
643: * newline.
644: */
645: if (0 == p->literal && 0 == p->outmacro &&
646: *pos < BUFSZ(p) && '.' == BUF(p)[*pos - 1])
647: texiputchar(p, '\n');
1.1 kristaps 648: }
649:
650: /*
651: * Look up the command at position "pos" in the buffer, returning it (or
652: * TEXICMD__MAX if none found) and setting "end" to be the absolute
653: * index after the command name.
654: */
655: enum texicmd
1.19 kristaps 656: texicmd(const struct texi *p, size_t pos, size_t *end, struct teximacro **macro)
1.1 kristaps 657: {
1.4 kristaps 658: size_t i, len, toksz;
1.1 kristaps 659:
1.14 kristaps 660: assert('@' == BUF(p)[pos]);
1.1 kristaps 661:
1.7 kristaps 662: if (NULL != macro)
663: *macro = NULL;
664:
1.14 kristaps 665: if ((*end = pos) == BUFSZ(p))
1.1 kristaps 666: return(TEXICMD__MAX);
1.14 kristaps 667: else if ((*end = ++pos) == BUFSZ(p))
1.1 kristaps 668: return(TEXICMD__MAX);
669:
670: /* Alphabetic commands are special. */
1.23 kristaps 671: if ( ! isalpha((unsigned int)BUF(p)[pos])) {
1.14 kristaps 672: if ((*end = pos + 1) == BUFSZ(p))
1.1 kristaps 673: return(TEXICMD__MAX);
674: for (i = 0; i < TEXICMD__MAX; i++) {
675: if (1 != texitoks[i].len)
676: continue;
1.14 kristaps 677: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], 1))
1.1 kristaps 678: return(i);
679: }
1.14 kristaps 680: texiwarn(p, "bad command: @%c", BUF(p)[pos]);
1.1 kristaps 681: return(TEXICMD__MAX);
682: }
683:
1.4 kristaps 684: /* Scan to the end of the possible command name. */
1.14 kristaps 685: for (*end = pos; *end < BUFSZ(p) && ! ismspace(BUF(p)[*end]); (*end)++)
686: if ((*end > pos && ('@' == BUF(p)[*end] ||
687: '{' == BUF(p)[*end] || '}' == BUF(p)[*end])))
1.1 kristaps 688: break;
689:
1.4 kristaps 690: /* Look for the command. */
1.1 kristaps 691: len = *end - pos;
692: for (i = 0; i < TEXICMD__MAX; i++) {
693: if (len != texitoks[i].len)
694: continue;
1.14 kristaps 695: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], len))
1.1 kristaps 696: return(i);
697: }
698:
1.4 kristaps 699: /* Look for it in our indices. */
700: for (i = 0; i < p->indexsz; i++) {
701: toksz = strlen(p->indexs[i]);
702: if (len != 5 + toksz)
703: continue;
1.14 kristaps 704: if (strncmp(&BUF(p)[pos], p->indexs[i], toksz))
1.4 kristaps 705: continue;
1.14 kristaps 706: if (0 == strncmp(&BUF(p)[pos + toksz], "index", 5))
1.7 kristaps 707: return(TEXICMD_USER_INDEX);
708: }
709:
710: for (i = 0; i < p->macrosz; i++) {
711: if (len != strlen(p->macros[i].key))
712: continue;
1.14 kristaps 713: if (strncmp(&BUF(p)[pos], p->macros[i].key, len))
1.7 kristaps 714: continue;
715: if (NULL != macro)
716: *macro = &p->macros[i];
717: return(TEXICMD__MAX);
1.4 kristaps 718: }
719:
1.14 kristaps 720: texiwarn(p, "bad command: @%.*s", (int)len, &BUF(p)[pos]);
1.1 kristaps 721: return(TEXICMD__MAX);
722: }
723:
724: /*
725: * Parse an argument from a bracketed command, e.g., @url{foo, baz}.
726: * Num should be set to the argument we're currently parsing, although
727: * it suffixes for it to be zero or non-zero.
728: * This will return 1 if there are more arguments, 0 otherwise.
729: * This will stop (returning 0) in the event of EOF or if we're not at a
730: * bracket for the zeroth parse.
731: */
732: int
1.14 kristaps 733: parsearg(struct texi *p, size_t *pos, size_t num)
1.1 kristaps 734: {
1.17 kristaps 735: size_t end, sv;
1.7 kristaps 736: enum texicmd cmd;
737: struct teximacro *macro;
1.1 kristaps 738:
1.14 kristaps 739: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
740: advance(p, pos);
741: if (*pos == BUFSZ(p) || (0 == num && '{' != BUF(p)[*pos]))
1.1 kristaps 742: return(0);
743: if (0 == num)
1.14 kristaps 744: advance(p, pos);
1.1 kristaps 745:
1.14 kristaps 746: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
747: switch (BUF(p)[*pos]) {
1.1 kristaps 748: case (','):
1.14 kristaps 749: advance(p, pos);
1.1 kristaps 750: return(1);
751: case ('}'):
1.14 kristaps 752: advance(p, pos);
1.1 kristaps 753: return(0);
754: case ('{'):
755: if (0 == p->ign)
756: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 757: advance(p, pos);
1.1 kristaps 758: continue;
759: case ('@'):
760: break;
761: default:
1.14 kristaps 762: parseword(p, pos, ',');
1.1 kristaps 763: continue;
764: }
765:
1.17 kristaps 766: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 767: cmd = texicmd(p, *pos, &end, ¯o);
768: advanceto(p, pos, end);
1.7 kristaps 769: if (NULL != macro)
1.17 kristaps 770: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 771: if (TEXICMD__MAX == cmd)
772: continue;
773: if (NULL != texitoks[cmd].fp)
1.14 kristaps 774: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 775: }
776: return(0);
777: }
778:
779: /*
780: * Parse until the end of a bracketed statement, e.g., @foo{bar baz}.
781: * This will stop in the event of EOF or if we're not at a bracket.
782: */
783: void
1.18 kristaps 784: parsebracket(struct texi *p, size_t *pos, int dostack)
1.1 kristaps 785: {
1.18 kristaps 786: size_t end, sv, stack;
1.7 kristaps 787: enum texicmd cmd;
788: struct teximacro *macro;
1.1 kristaps 789:
1.14 kristaps 790: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
791: advance(p, pos);
1.1 kristaps 792:
1.14 kristaps 793: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.1 kristaps 794: return;
1.14 kristaps 795: advance(p, pos);
1.1 kristaps 796:
1.18 kristaps 797: stack = 0;
1.14 kristaps 798: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
799: switch (BUF(p)[*pos]) {
1.1 kristaps 800: case ('}'):
1.18 kristaps 801: if (stack > 0) {
802: stack--;
803: advance(p, pos);
804: texiputchar(p, '}');
805: continue;
806: }
1.14 kristaps 807: advance(p, pos);
1.1 kristaps 808: return;
809: case ('{'):
1.18 kristaps 810: if (dostack) {
811: stack++;
812: advance(p, pos);
813: texiputchar(p, '{');
814: continue;
815: }
1.1 kristaps 816: if (0 == p->ign)
817: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 818: advance(p, pos);
1.1 kristaps 819: continue;
820: case ('@'):
821: break;
822: default:
1.14 kristaps 823: parseword(p, pos, '\0');
1.1 kristaps 824: continue;
825: }
826:
1.17 kristaps 827: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 828: cmd = texicmd(p, *pos, &end, ¯o);
829: advanceto(p, pos, end);
1.7 kristaps 830: if (NULL != macro)
1.17 kristaps 831: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 832: if (TEXICMD__MAX == cmd)
833: continue;
834: if (NULL != texitoks[cmd].fp)
1.14 kristaps 835: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 836: }
837: }
838:
839: /*
840: * This should be invoked when we're on a macro line and want to process
841: * to the end of the current input line, doing all of our macros along
842: * the way.
843: */
844: void
1.14 kristaps 845: parseeoln(struct texi *p, size_t *pos)
1.1 kristaps 846: {
1.17 kristaps 847: size_t end, sv;
1.7 kristaps 848: enum texicmd cmd;
849: struct teximacro *macro;
1.1 kristaps 850:
1.14 kristaps 851: while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) {
852: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 853: p->seenws = 1;
854: if (p->literal)
1.14 kristaps 855: texiputchar(p, BUF(p)[*pos]);
856: advance(p, pos);
1.1 kristaps 857: }
1.14 kristaps 858: switch (BUF(p)[*pos]) {
1.1 kristaps 859: case ('}'):
860: if (0 == p->ign)
861: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 862: advance(p, pos);
1.1 kristaps 863: continue;
864: case ('{'):
865: if (0 == p->ign)
866: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 867: advance(p, pos);
1.1 kristaps 868: continue;
869: case ('@'):
870: break;
871: default:
1.14 kristaps 872: parseword(p, pos, '\0');
1.1 kristaps 873: continue;
874: }
875:
1.17 kristaps 876: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 877: cmd = texicmd(p, *pos, &end, ¯o);
878: advanceto(p, pos, end);
1.7 kristaps 879: if (NULL != macro)
1.17 kristaps 880: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 881: if (TEXICMD__MAX == cmd)
882: continue;
883: if (NULL != texitoks[cmd].fp)
1.14 kristaps 884: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 885: }
1.14 kristaps 886:
887: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
888: advance(p, pos);
1.19 kristaps 889: }
890:
891: /*
892: * Peek to see if there's a command after subsequent whitespace.
893: * If so, return the macro identifier.
894: * This DOES NOT work with user-defined macros.
895: */
896: enum texicmd
897: peekcmd(const struct texi *p, size_t pos)
898: {
899: size_t end;
900:
901: while (pos < BUFSZ(p) && ismspace(BUF(p)[pos]))
902: pos++;
903: if (pos == BUFSZ(p) || '@' != BUF(p)[pos])
904: return(TEXICMD__MAX);
905: return(texicmd(p, pos, &end, NULL));
1.1 kristaps 906: }
907:
908: /*
909: * Parse a single word or command.
910: * This will return immediately at the EOF.
911: */
1.14 kristaps 912: static void
913: parsesingle(struct texi *p, size_t *pos)
1.1 kristaps 914: {
1.17 kristaps 915: size_t end, sv;
1.7 kristaps 916: enum texicmd cmd;
917: struct teximacro *macro;
1.1 kristaps 918:
1.14 kristaps 919: if ((*pos = advancenext(p, pos)) >= BUFSZ(p))
1.1 kristaps 920: return;
921:
1.14 kristaps 922: switch (BUF(p)[*pos]) {
1.1 kristaps 923: case ('}'):
924: if (0 == p->ign)
925: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 926: advance(p, pos);
1.1 kristaps 927: return;
928: case ('{'):
929: if (0 == p->ign)
930: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 931: advance(p, pos);
1.1 kristaps 932: return;
933: case ('@'):
934: break;
935: default:
1.14 kristaps 936: parseword(p, pos, '\0');
1.1 kristaps 937: return;
938: }
939:
1.17 kristaps 940: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 941: cmd = texicmd(p, *pos, &end, ¯o);
942: advanceto(p, pos, end);
1.7 kristaps 943: if (NULL != macro)
1.17 kristaps 944: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 945: if (TEXICMD__MAX == cmd)
946: return;
947: if (NULL != texitoks[cmd].fp)
1.14 kristaps 948: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 949: }
950:
951: /*
952: * This is used in the @deffn type of command.
953: * These have an arbitrary number of line arguments; however, these
954: * arguments may or may not be surrounded by brackets.
955: * In this function, we parse each one as either a bracketed or
956: * non-bracketed argument, returning 0 when we've reached the end of
957: * line or 1 otherwise.
958: */
959: int
1.14 kristaps 960: parselinearg(struct texi *p, size_t *pos)
1.1 kristaps 961: {
962:
1.14 kristaps 963: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 964: p->seenws = 1;
1.14 kristaps 965: advance(p, pos);
1.1 kristaps 966: }
967:
1.14 kristaps 968: if (*pos < BUFSZ(p) && '{' == BUF(p)[*pos])
1.18 kristaps 969: parsebracket(p, pos, 0);
1.14 kristaps 970: else if (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos])
971: parsesingle(p, pos);
1.1 kristaps 972: else
973: return(0);
974:
975: return(1);
976: }
977:
978: /*
979: * Parse til the end of the buffer.
980: */
1.14 kristaps 981: static void
982: parseeof(struct texi *p)
1.1 kristaps 983: {
984: size_t pos;
985:
1.14 kristaps 986: for (pos = 0; pos < BUFSZ(p); )
987: parsesingle(p, &pos);
1.1 kristaps 988: }
989:
1.8 kristaps 990: void
1.21 kristaps 991: texisplice(struct texi *p, const char *buf, size_t sz, size_t pos)
1.8 kristaps 992: {
1.14 kristaps 993: char *cp;
994: struct texifile *f;
1.8 kristaps 995:
1.14 kristaps 996: assert(p->filepos > 0);
997: f = &p->files[p->filepos - 1];
1.8 kristaps 998:
1.14 kristaps 999: if (f->mapsz + sz > f->mapmaxsz) {
1000: f->mapmaxsz = f->mapsz + sz + 1024;
1001: cp = realloc(f->map, f->mapmaxsz);
1002: if (NULL == cp)
1003: texiabort(p, NULL);
1004: f->map = cp;
1005: }
1.8 kristaps 1006:
1.15 kristaps 1007: f->insplice += sz;
1.21 kristaps 1008: memmove(f->map + pos + sz, f->map + pos, f->mapsz - pos);
1009: memcpy(f->map + pos, buf, sz);
1.14 kristaps 1010: f->mapsz += sz;
1.8 kristaps 1011: }
1012:
1013: /*
1.1 kristaps 1014: * Parse a block sequence until we have the "@end endtoken" command
1015: * invocation.
1016: * This will return immediately at EOF.
1017: */
1018: void
1.14 kristaps 1019: parseto(struct texi *p, size_t *pos, const char *endtoken)
1.1 kristaps 1020: {
1.17 kristaps 1021: size_t end, sv;
1.7 kristaps 1022: enum texicmd cmd;
1023: size_t endtoksz;
1024: struct teximacro *macro;
1.1 kristaps 1025:
1026: endtoksz = strlen(endtoken);
1027: assert(endtoksz > 0);
1028:
1.14 kristaps 1029: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
1030: switch (BUF(p)[*pos]) {
1.1 kristaps 1031: case ('}'):
1032: if (0 == p->ign)
1033: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 1034: advance(p, pos);
1.1 kristaps 1035: continue;
1036: case ('{'):
1037: if (0 == p->ign)
1038: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 1039: advance(p, pos);
1.1 kristaps 1040: continue;
1041: case ('@'):
1042: break;
1043: default:
1.14 kristaps 1044: parseword(p, pos, '\0');
1.1 kristaps 1045: continue;
1046: }
1047:
1.17 kristaps 1048: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 1049: cmd = texicmd(p, *pos, &end, ¯o);
1050: advanceto(p, pos, end);
1.1 kristaps 1051: if (TEXICMD_END == cmd) {
1.14 kristaps 1052: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1053: advance(p, pos);
1.1 kristaps 1054: /*
1055: * FIXME: check the full word, not just its
1056: * initial substring!
1057: */
1.14 kristaps 1058: if (BUFSZ(p) - *pos >= endtoksz && 0 == strncmp
1059: (&BUF(p)[*pos], endtoken, endtoksz)) {
1060: advanceeoln(p, pos, 0);
1.1 kristaps 1061: break;
1062: }
1063: if (0 == p->ign)
1064: texiwarn(p, "unexpected \"end\"");
1.14 kristaps 1065: advanceeoln(p, pos, 0);
1.1 kristaps 1066: continue;
1.7 kristaps 1067: }
1068: if (NULL != macro)
1.17 kristaps 1069: texiexecmacro(p, macro, sv, pos);
1.7 kristaps 1070: if (TEXICMD__MAX == cmd)
1071: continue;
1072: if (NULL != texitoks[cmd].fp)
1.14 kristaps 1073: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 1074: }
1075: }
1076:
1077: /*
1.12 kristaps 1078: * Like parsefile() but used for reading from stdandard input.
1079: * This can only be called for the first file!
1080: */
1081: void
1082: parsestdin(struct texi *p)
1083: {
1084: struct texifile *f;
1085: ssize_t ssz;
1086:
1087: assert(0 == p->filepos);
1088: f = &p->files[p->filepos];
1089: memset(f, 0, sizeof(struct texifile));
1090:
1091: f->type = TEXISRC_STDIN;
1092: f->name = "<stdin>";
1093:
1.14 kristaps 1094: for (f->mapsz = 0; ; f->mapsz += (size_t)ssz) {
1095: if (f->mapsz == f->mapmaxsz) {
1096: if (f->mapmaxsz == (1U << 31))
1.12 kristaps 1097: texierr(p, "stdin buffer too long");
1.14 kristaps 1098: f->mapmaxsz = f->mapmaxsz > 65536 / 2 ?
1099: 2 * f->mapmaxsz : 65536;
1100: f->map = realloc(f->map, f->mapmaxsz);
1.12 kristaps 1101: if (NULL == f->map)
1102: texiabort(p, NULL);
1103: }
1.14 kristaps 1104: ssz = read(STDIN_FILENO, f->map +
1105: (int)f->mapsz, f->mapmaxsz - f->mapsz);
1.12 kristaps 1106: if (0 == ssz)
1107: break;
1108: else if (-1 == ssz)
1109: texiabort(p, NULL);
1110: }
1111:
1112: p->filepos++;
1.14 kristaps 1113: parseeof(p);
1.12 kristaps 1114: texifilepop(p);
1115: }
1116:
1117: /*
1.1 kristaps 1118: * Memory-map the file "fname" and begin parsing it unless "parse" is
1119: * zero, in which case we just dump the file to stdout (making sure it
1120: * doesn't trip up mdoc(7) along the way).
1121: * This can be called in a nested context.
1122: */
1123: void
1124: parsefile(struct texi *p, const char *fname, int parse)
1125: {
1126: struct texifile *f;
1127: int fd;
1128: struct stat st;
1129: size_t i;
1.14 kristaps 1130: char *map;
1.1 kristaps 1131:
1.5 kristaps 1132: if (64 == p->filepos)
1.6 kristaps 1133: texierr(p, "too many open files");
1.1 kristaps 1134: f = &p->files[p->filepos];
1135: memset(f, 0, sizeof(struct texifile));
1136:
1.12 kristaps 1137: f->type = TEXISRC_FILE;
1.1 kristaps 1138: f->name = fname;
1139: if (-1 == (fd = open(fname, O_RDONLY, 0))) {
1140: texiabort(p, fname);
1141: } else if (-1 == fstat(fd, &st)) {
1142: close(fd);
1143: texiabort(p, fname);
1144: }
1145:
1.14 kristaps 1146: f->mapsz = f->mapmaxsz = st.st_size;
1147: map = mmap(NULL, f->mapsz,
1.1 kristaps 1148: PROT_READ, MAP_SHARED, fd, 0);
1149: close(fd);
1150:
1.14 kristaps 1151: if (MAP_FAILED == map)
1.1 kristaps 1152: texiabort(p, fname);
1153:
1154: if ( ! parse) {
1.13 kristaps 1155: for (i = 0; i < f->mapsz; i++)
1.14 kristaps 1156: texiputchar(p, map[i]);
1.13 kristaps 1157: if (p->outcol)
1158: texiputchar(p, '\n');
1.14 kristaps 1159: munmap(map, f->mapsz);
1160: return;
1161: }
1162:
1163: p->filepos++;
1164: f->map = malloc(f->mapsz);
1165: memcpy(f->map, map, f->mapsz);
1166: munmap(map, f->mapsz);
1167: parseeof(p);
1.1 kristaps 1168: texifilepop(p);
1169: }
1170:
1.2 kristaps 1171: /*
1172: * Look up the value to a stored pair's value starting in "buf" from
1173: * start to end.
1174: * Return the pointer to the value memory, which can be NULL if the
1175: * pointer key does not exist.
1176: * The pointer can point to NULL if the value has been unset.
1177: */
1178: static char **
1.14 kristaps 1179: valuequery(const struct texi *p, size_t start, size_t end)
1.2 kristaps 1180: {
1181: size_t i, sz, len;
1182:
1183: assert(end >= start);
1184: /* Ignore zero-length. */
1185: if (0 == (len = (end - start)))
1186: return(NULL);
1187: for (i = 0; i < p->valsz; i++) {
1188: sz = strlen(p->vals[i].key);
1189: if (sz != len)
1190: continue;
1.14 kristaps 1191: if (0 == strncmp(p->vals[i].key, &BUF(p)[start], len))
1.2 kristaps 1192: return(&p->vals[i].value);
1193: }
1194: return(NULL);
1195: }
1196:
1197: /*
1198: * Parse a key until the end of line, e.g., @clear foo\n, and return the
1199: * pointer to its value via valuequery().
1200: */
1201: static char **
1.14 kristaps 1202: valuelquery(struct texi *p, size_t *pos)
1.2 kristaps 1203: {
1204: size_t start, end;
1205: char **ret;
1206:
1.14 kristaps 1207: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1208: advance(p, pos);
1209: if (*pos == BUFSZ(p))
1.2 kristaps 1210: return(NULL);
1.14 kristaps 1211: for (start = end = *pos; end < BUFSZ(p); end++)
1212: if ('\n' == BUF(p)[end])
1.2 kristaps 1213: break;
1.14 kristaps 1214: advanceto(p, pos, end);
1215: if (*pos < BUFSZ(p)) {
1216: assert('\n' == BUF(p)[*pos]);
1217: advance(p, pos);
1.2 kristaps 1218: }
1.14 kristaps 1219: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1220: return(NULL);
1221: return(ret);
1222: }
1223:
1224: void
1.14 kristaps 1225: valuelclear(struct texi *p, size_t *pos)
1.2 kristaps 1226: {
1227: char **ret;
1228:
1.14 kristaps 1229: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1230: return;
1231: free(*ret);
1232: *ret = NULL;
1233: }
1234:
1235: const char *
1.14 kristaps 1236: valuellookup(struct texi *p, size_t *pos)
1.2 kristaps 1237: {
1238: char **ret;
1239:
1.14 kristaps 1240: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1241: return(NULL);
1242: return(*ret);
1243: }
1244:
1245: /*
1246: * Parse a key from a bracketed string, e.g., @value{foo}, and return
1247: * the pointer to its value.
1248: * If the returned pointer is NULL, either there was no string within
1249: * the brackets (or no brackets), or the value was not found, or the
1250: * value had previously been unset.
1251: */
1252: const char *
1.14 kristaps 1253: valueblookup(struct texi *p, size_t *pos)
1.2 kristaps 1254: {
1255: size_t start, end;
1256: char **ret;
1257:
1.14 kristaps 1258: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1259: advance(p, pos);
1260: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.2 kristaps 1261: return(NULL);
1.14 kristaps 1262: advance(p, pos);
1263: for (start = end = *pos; end < BUFSZ(p); end++)
1264: if ('}' == BUF(p)[end])
1.2 kristaps 1265: break;
1.14 kristaps 1266: advanceto(p, pos, end);
1267: if (*pos < BUFSZ(p)) {
1268: assert('}' == BUF(p)[*pos]);
1269: advance(p, pos);
1.2 kristaps 1270: }
1.14 kristaps 1271: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1272: return(NULL);
1273: return(*ret);
1274: }
1275:
1276: void
1277: valueadd(struct texi *p, char *key, char *val)
1278: {
1279: size_t i;
1280:
1281: assert(NULL != key);
1282: assert(NULL != val);
1283:
1284: for (i = 0; i < p->valsz; i++)
1285: if (0 == strcmp(p->vals[i].key, key))
1286: break;
1287:
1288: if (i < p->valsz) {
1289: free(key);
1290: free(p->vals[i].value);
1291: p->vals[i].value = val;
1292: } else {
1.4 kristaps 1293: /* FIXME: reallocarray() */
1.2 kristaps 1294: p->vals = realloc(p->vals,
1295: (p->valsz + 1) *
1296: sizeof(struct texivalue));
1.4 kristaps 1297: if (NULL == p->vals)
1298: texiabort(p, NULL);
1.2 kristaps 1299: p->vals[p->valsz].key = key;
1300: p->vals[p->valsz].value = val;
1301: p->valsz++;
1302: }
1.7 kristaps 1303: }
1304:
1305: /*
1306: * Take the arguments to a macro, e.g., @foo{bar, baz, xyzzy} (or the
1307: * declaration form, @macro foo {arg1, ...}) and textually convert it to
1308: * an array of arguments of size "argsz".
1309: * These need to be freed individually and as a whole.
1310: * NOTE: this will puke on @, or @} macros, which can trick it into
1311: * stopping argument parsing earlier.
1312: * Ergo, textual: this doesn't interpret the arguments in any way.
1313: */
1314: char **
1.14 kristaps 1315: argparse(struct texi *p, size_t *pos, size_t *argsz, size_t hint)
1.7 kristaps 1316: {
1317: char **args;
1318: size_t start, end, stack;
1319:
1.14 kristaps 1320: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1321: advance(p, pos);
1.7 kristaps 1322:
1323: args = NULL;
1324: *argsz = 0;
1325:
1.17 kristaps 1326: if (*pos == BUFSZ(p))
1327: return(args);
1328:
1.14 kristaps 1329: if ('{' != BUF(p)[*pos] && hint) {
1.10 kristaps 1330: /*
1331: * Special case: if we encounter an unbracketed argument
1332: * and we're being invoked with non-zero arguments
1333: * (versus being set, i.e., hint>0), then parse until
1334: * the end of line.
1335: */
1336: *argsz = 1;
1337: args = calloc(1, sizeof(char *));
1338: if (NULL == args)
1339: texiabort(p, NULL);
1340: start = *pos;
1.14 kristaps 1341: while (*pos < BUFSZ(p)) {
1342: if ('\n' == BUF(p)[*pos])
1.10 kristaps 1343: break;
1.14 kristaps 1344: advance(p, pos);
1.10 kristaps 1345: }
1346: args[0] = malloc(*pos - start + 1);
1.14 kristaps 1347: memcpy(args[0], &BUF(p)[start], *pos - start);
1.10 kristaps 1348: args[0][*pos - start] = '\0';
1.14 kristaps 1349: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
1350: advance(p, pos);
1.10 kristaps 1351: return(args);
1.14 kristaps 1352: } else if ('{' != BUF(p)[*pos])
1.7 kristaps 1353: return(args);
1.17 kristaps 1354:
1355: assert('{' == BUF(p)[*pos]);
1.7 kristaps 1356:
1357: /* Parse til the closing '}', putting into the array. */
1.14 kristaps 1358: advance(p, pos);
1359: while (*pos < BUFSZ(p)) {
1360: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1361: advance(p, pos);
1.7 kristaps 1362: start = *pos;
1363: stack = 0;
1.14 kristaps 1364: while (*pos < BUFSZ(p)) {
1.7 kristaps 1365: /*
1366: * According to the manual, commas within
1367: * embedded commands are escaped.
1368: * We keep track of embedded-ness in the "stack"
1369: * state anyway, so this is free.
1370: */
1.14 kristaps 1371: if (',' == BUF(p)[*pos] && 0 == stack && 1 != hint)
1.7 kristaps 1372: break;
1.14 kristaps 1373: else if (0 == stack && '}' == BUF(p)[*pos])
1.7 kristaps 1374: break;
1.14 kristaps 1375: else if (0 != stack && '}' == BUF(p)[*pos])
1.7 kristaps 1376: stack--;
1.14 kristaps 1377: else if ('{' == BUF(p)[*pos])
1.7 kristaps 1378: stack++;
1.14 kristaps 1379: advance(p, pos);
1.7 kristaps 1380: }
1381: if (stack)
1382: texiwarn(p, "unterminated macro "
1383: "in macro arguments");
1.14 kristaps 1384: if ((end = *pos) == BUFSZ(p))
1.7 kristaps 1385: break;
1386: /* Test for zero-length '{ }'. */
1.14 kristaps 1387: if (start == end && '}' == BUF(p)[*pos] && 0 == *argsz)
1.7 kristaps 1388: break;
1389: /* FIXME: use reallocarray. */
1390: args = realloc
1391: (args, sizeof(char *) *
1392: (*argsz + 1));
1393: if (NULL == args)
1394: texiabort(p, NULL);
1395: args[*argsz] = malloc(end - start + 1);
1396: if (NULL == args[*argsz])
1397: texiabort(p, NULL);
1398: memcpy(args[*argsz],
1.14 kristaps 1399: &BUF(p)[start], end - start);
1.7 kristaps 1400: args[*argsz][end - start] = '\0';
1401: (*argsz)++;
1.14 kristaps 1402: if ('}' == BUF(p)[*pos])
1.7 kristaps 1403: break;
1.14 kristaps 1404: advance(p, pos);
1.7 kristaps 1405: }
1406:
1.14 kristaps 1407: if (*pos == BUFSZ(p))
1.7 kristaps 1408: texierr(p, "unterminated arguments");
1.14 kristaps 1409: assert('}' == BUF(p)[*pos]);
1410: advance(p, pos);
1.7 kristaps 1411: return(args);
1.2 kristaps 1412: }
1.20 kristaps 1413:
1414: /*
1415: * If we're printing chapters, then do some naviation here and then
1416: * close our outfile.
1417: * I want to call this the SEE ALSO section, but that's not really what
1418: * it is: we'll refer to the "initial" (top) node and the next and
1419: * previous chapters.
1420: */
1421: void
1422: teximdocclose(struct texi *p, int last)
1423: {
1424: char buf[PATH_MAX];
1425:
1426: if (NULL == p->chapters || 0 == p->chapnum)
1427: return;
1428:
1429: teximacro(p, "Sh INFO NAVIGATION");
1430:
1431: /* Print a reference to the "top" node. */
1432: if (p->chapnum > 1) {
1.22 kristaps 1433: texiputchars(p, "Top node,");
1.20 kristaps 1434: snprintf(buf, sizeof(buf), "node1 7");
1435: teximacroopen(p, "Xr ");
1436: texiputchars(p, buf);
1.22 kristaps 1437: texiputchars(p, " ;");
1.20 kristaps 1438: teximacroclose(p);
1439: }
1440:
1441: /* Print a reference to the previous node. */
1442: if (p->chapnum > 2) {
1.22 kristaps 1443: texiputchars(p, "previous node,");
1.20 kristaps 1444: snprintf(buf, sizeof(buf),
1445: "node%zu 7", p->chapnum - 1);
1446: teximacroopen(p, "Xr ");
1447: texiputchars(p, buf);
1448: if ( ! last)
1.22 kristaps 1449: texiputchars(p, " ;");
1.20 kristaps 1450: teximacroclose(p);
1451: }
1452:
1453: /* Print a reference to the next node. */
1454: if ( ! last) {
1.22 kristaps 1455: if (1 == p->chapnum)
1456: texiputchars(p, "Next node,");
1457: else
1458: texiputchars(p, "next node,");
1.20 kristaps 1459: snprintf(buf, sizeof(buf),
1460: "node%zu 7", p->chapnum + 1);
1461: teximacroopen(p, "Xr ");
1462: texiputchars(p, buf);
1463: teximacroclose(p);
1464: }
1465:
1466: fclose(p->outfile);
1467: }
1468:
1469: /*
1470: * Open a mdoc(7) context.
1471: * If we're printing chapters, then open the outfile here, too.
1472: * Otherwise just print the mdoc(7) prologue.
1473: */
1474: void
1.21 kristaps 1475: teximdocopen(struct texi *p, size_t *pos)
1.20 kristaps 1476: {
1477: const char *cp;
1478: time_t t;
1479: char date[32];
1480: char fname[PATH_MAX];
1481:
1482: if (NULL != p->chapters) {
1483: snprintf(fname, sizeof(fname), "%s/node%zu.7",
1484: p->chapters, ++p->chapnum);
1485: p->outfile = fopen(fname, "w");
1486: if (NULL == p->outfile)
1487: texiabort(p, fname);
1488: }
1489:
1490: /*
1491: * Here we print our standard mdoc(7) prologue.
1492: * We use the title set with @settitle for the `Nd' description
1493: * and the source document filename (the first one as invoked on
1494: * the command line) for the title.
1495: * The date is set to the current date.
1496: */
1497: t = time(NULL);
1498: strftime(date, sizeof(date), "%F", localtime(&t));
1499:
1500: teximacroopen(p, "Dd");
1501: texiputchars(p, date);
1502: teximacroclose(p);
1503: teximacroopen(p, "Dt");
1504: for (cp = p->title; '\0' != *cp; cp++)
1505: texiputchar(p, toupper((unsigned int)*cp));
1506: texiputchars(p, " 7");
1507: teximacroclose(p);
1508: teximacro(p, "Os");
1509: teximacro(p, "Sh NAME");
1510: teximacroopen(p, "Nm");
1511: for (cp = p->title; '\0' != *cp; cp++)
1512: texiputchar(p, *cp);
1513: teximacroclose(p);
1514: teximacroopen(p, "Nd");
1.21 kristaps 1515: /*
1516: * The subtitle `Nd' can consist of arbitrary macros, so paste
1517: * it and parse to the end of the line.
1518: */
1519: if (NULL != p->subtitle) {
1520: texisplice(p, p->subtitle, strlen(p->subtitle), *pos);
1521: parseeoln(p, pos);
1522: } else
1.20 kristaps 1523: texiputchars(p, "Unknown description");
1524: teximacroclose(p);
1525: }
1526:
CVSweb