Annotation of texi2mdoc/util.c, Revision 1.27
1.27 ! kristaps 1: /* $Id: util.c,v 1.26 2015/03/05 08:35:35 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;
1.14 kristaps 600: if (*pos < BUFSZ(p) - 1 &&
601: '`' == BUF(p)[*pos] &&
602: '`' == BUF(p)[*pos + 1]) {
1.1 kristaps 603: texiputchars(p, "\\(lq");
1.14 kristaps 604: advance(p, pos);
605: } else if (*pos < BUFSZ(p) - 1 &&
606: '\'' == BUF(p)[*pos] &&
607: '\'' == BUF(p)[*pos + 1]) {
1.1 kristaps 608: texiputchars(p, "\\(rq");
1.14 kristaps 609: advance(p, pos);
1.1 kristaps 610: } else
1.14 kristaps 611: texiputchar(p, BUF(p)[*pos]);
612: advance(p, pos);
1.1 kristaps 613: }
1.25 kristaps 614:
1.27 ! kristaps 615: if (*pos + 1 < BUFSZ(p) &&
! 616: '\n' == BUF(p)[*pos] &&
! 617: '\n' == BUF(p)[*pos + 1])
! 618: p->seenvs = 1;
! 619:
1.25 kristaps 620: /*
621: * New sentence, new line:if we (non-macro, non-literal) see a
622: * period at the end of the last printed word, then open a
623: * newline.
624: */
625: if (0 == p->literal && 0 == p->outmacro &&
626: *pos < BUFSZ(p) && '.' == BUF(p)[*pos - 1])
627: texiputchar(p, '\n');
1.1 kristaps 628: }
629:
630: /*
631: * Look up the command at position "pos" in the buffer, returning it (or
632: * TEXICMD__MAX if none found) and setting "end" to be the absolute
633: * index after the command name.
634: */
635: enum texicmd
1.19 kristaps 636: texicmd(const struct texi *p, size_t pos, size_t *end, struct teximacro **macro)
1.1 kristaps 637: {
1.4 kristaps 638: size_t i, len, toksz;
1.1 kristaps 639:
1.14 kristaps 640: assert('@' == BUF(p)[pos]);
1.1 kristaps 641:
1.7 kristaps 642: if (NULL != macro)
643: *macro = NULL;
644:
1.14 kristaps 645: if ((*end = pos) == BUFSZ(p))
1.1 kristaps 646: return(TEXICMD__MAX);
1.14 kristaps 647: else if ((*end = ++pos) == BUFSZ(p))
1.1 kristaps 648: return(TEXICMD__MAX);
649:
650: /* Alphabetic commands are special. */
1.23 kristaps 651: if ( ! isalpha((unsigned int)BUF(p)[pos])) {
1.14 kristaps 652: if ((*end = pos + 1) == BUFSZ(p))
1.1 kristaps 653: return(TEXICMD__MAX);
654: for (i = 0; i < TEXICMD__MAX; i++) {
655: if (1 != texitoks[i].len)
656: continue;
1.14 kristaps 657: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], 1))
1.1 kristaps 658: return(i);
659: }
1.14 kristaps 660: texiwarn(p, "bad command: @%c", BUF(p)[pos]);
1.1 kristaps 661: return(TEXICMD__MAX);
662: }
663:
1.4 kristaps 664: /* Scan to the end of the possible command name. */
1.14 kristaps 665: for (*end = pos; *end < BUFSZ(p) && ! ismspace(BUF(p)[*end]); (*end)++)
666: if ((*end > pos && ('@' == BUF(p)[*end] ||
667: '{' == BUF(p)[*end] || '}' == BUF(p)[*end])))
1.1 kristaps 668: break;
669:
1.4 kristaps 670: /* Look for the command. */
1.1 kristaps 671: len = *end - pos;
672: for (i = 0; i < TEXICMD__MAX; i++) {
673: if (len != texitoks[i].len)
674: continue;
1.14 kristaps 675: if (0 == strncmp(texitoks[i].tok, &BUF(p)[pos], len))
1.1 kristaps 676: return(i);
677: }
678:
1.4 kristaps 679: /* Look for it in our indices. */
680: for (i = 0; i < p->indexsz; i++) {
681: toksz = strlen(p->indexs[i]);
682: if (len != 5 + toksz)
683: continue;
1.14 kristaps 684: if (strncmp(&BUF(p)[pos], p->indexs[i], toksz))
1.4 kristaps 685: continue;
1.14 kristaps 686: if (0 == strncmp(&BUF(p)[pos + toksz], "index", 5))
1.7 kristaps 687: return(TEXICMD_USER_INDEX);
688: }
689:
690: for (i = 0; i < p->macrosz; i++) {
691: if (len != strlen(p->macros[i].key))
692: continue;
1.14 kristaps 693: if (strncmp(&BUF(p)[pos], p->macros[i].key, len))
1.7 kristaps 694: continue;
695: if (NULL != macro)
696: *macro = &p->macros[i];
697: return(TEXICMD__MAX);
1.4 kristaps 698: }
699:
1.14 kristaps 700: texiwarn(p, "bad command: @%.*s", (int)len, &BUF(p)[pos]);
1.1 kristaps 701: return(TEXICMD__MAX);
702: }
703:
704: /*
705: * Parse an argument from a bracketed command, e.g., @url{foo, baz}.
706: * Num should be set to the argument we're currently parsing, although
707: * it suffixes for it to be zero or non-zero.
708: * This will return 1 if there are more arguments, 0 otherwise.
709: * This will stop (returning 0) in the event of EOF or if we're not at a
710: * bracket for the zeroth parse.
711: */
712: int
1.14 kristaps 713: parsearg(struct texi *p, size_t *pos, size_t num)
1.1 kristaps 714: {
1.17 kristaps 715: size_t end, sv;
1.7 kristaps 716: enum texicmd cmd;
717: struct teximacro *macro;
1.1 kristaps 718:
1.14 kristaps 719: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
720: advance(p, pos);
721: if (*pos == BUFSZ(p) || (0 == num && '{' != BUF(p)[*pos]))
1.1 kristaps 722: return(0);
723: if (0 == num)
1.14 kristaps 724: advance(p, pos);
1.1 kristaps 725:
1.14 kristaps 726: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
727: switch (BUF(p)[*pos]) {
1.1 kristaps 728: case (','):
1.14 kristaps 729: advance(p, pos);
1.1 kristaps 730: return(1);
731: case ('}'):
1.14 kristaps 732: advance(p, pos);
1.1 kristaps 733: return(0);
734: case ('{'):
735: if (0 == p->ign)
736: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 737: advance(p, pos);
1.1 kristaps 738: continue;
739: case ('@'):
740: break;
741: default:
1.14 kristaps 742: parseword(p, pos, ',');
1.1 kristaps 743: continue;
744: }
745:
1.17 kristaps 746: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 747: cmd = texicmd(p, *pos, &end, ¯o);
748: advanceto(p, pos, end);
1.7 kristaps 749: if (NULL != macro)
1.17 kristaps 750: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 751: if (TEXICMD__MAX == cmd)
752: continue;
753: if (NULL != texitoks[cmd].fp)
1.14 kristaps 754: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 755: }
756: return(0);
757: }
758:
759: /*
760: * Parse until the end of a bracketed statement, e.g., @foo{bar baz}.
761: * This will stop in the event of EOF or if we're not at a bracket.
762: */
763: void
1.18 kristaps 764: parsebracket(struct texi *p, size_t *pos, int dostack)
1.1 kristaps 765: {
1.18 kristaps 766: size_t end, sv, stack;
1.7 kristaps 767: enum texicmd cmd;
768: struct teximacro *macro;
1.1 kristaps 769:
1.14 kristaps 770: while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos]))
771: advance(p, pos);
1.1 kristaps 772:
1.14 kristaps 773: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.1 kristaps 774: return;
1.14 kristaps 775: advance(p, pos);
1.1 kristaps 776:
1.18 kristaps 777: stack = 0;
1.14 kristaps 778: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
779: switch (BUF(p)[*pos]) {
1.1 kristaps 780: case ('}'):
1.18 kristaps 781: if (stack > 0) {
782: stack--;
783: advance(p, pos);
784: texiputchar(p, '}');
785: continue;
786: }
1.14 kristaps 787: advance(p, pos);
1.1 kristaps 788: return;
789: case ('{'):
1.18 kristaps 790: if (dostack) {
791: stack++;
792: advance(p, pos);
793: texiputchar(p, '{');
794: continue;
795: }
1.1 kristaps 796: if (0 == p->ign)
797: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 798: advance(p, pos);
1.1 kristaps 799: continue;
800: case ('@'):
801: break;
802: default:
1.14 kristaps 803: parseword(p, pos, '\0');
1.1 kristaps 804: continue;
805: }
806:
1.17 kristaps 807: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 808: cmd = texicmd(p, *pos, &end, ¯o);
809: advanceto(p, pos, end);
1.7 kristaps 810: if (NULL != macro)
1.17 kristaps 811: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 812: if (TEXICMD__MAX == cmd)
813: continue;
814: if (NULL != texitoks[cmd].fp)
1.14 kristaps 815: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 816: }
817: }
818:
819: /*
820: * This should be invoked when we're on a macro line and want to process
821: * to the end of the current input line, doing all of our macros along
822: * the way.
823: */
824: void
1.14 kristaps 825: parseeoln(struct texi *p, size_t *pos)
1.1 kristaps 826: {
1.17 kristaps 827: size_t end, sv;
1.7 kristaps 828: enum texicmd cmd;
829: struct teximacro *macro;
1.1 kristaps 830:
1.14 kristaps 831: while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) {
832: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 833: p->seenws = 1;
834: if (p->literal)
1.14 kristaps 835: texiputchar(p, BUF(p)[*pos]);
836: advance(p, pos);
1.1 kristaps 837: }
1.14 kristaps 838: switch (BUF(p)[*pos]) {
1.1 kristaps 839: case ('}'):
840: if (0 == p->ign)
841: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 842: advance(p, pos);
1.1 kristaps 843: continue;
844: case ('{'):
845: if (0 == p->ign)
846: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 847: advance(p, pos);
1.1 kristaps 848: continue;
849: case ('@'):
850: break;
851: default:
1.14 kristaps 852: parseword(p, pos, '\0');
1.1 kristaps 853: continue;
854: }
855:
1.17 kristaps 856: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 857: cmd = texicmd(p, *pos, &end, ¯o);
858: advanceto(p, pos, end);
1.7 kristaps 859: if (NULL != macro)
1.17 kristaps 860: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 861: if (TEXICMD__MAX == cmd)
862: continue;
863: if (NULL != texitoks[cmd].fp)
1.14 kristaps 864: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 865: }
1.14 kristaps 866:
867: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
868: advance(p, pos);
1.19 kristaps 869: }
870:
871: /*
872: * Peek to see if there's a command after subsequent whitespace.
873: * If so, return the macro identifier.
874: * This DOES NOT work with user-defined macros.
875: */
876: enum texicmd
877: peekcmd(const struct texi *p, size_t pos)
878: {
879: size_t end;
880:
881: while (pos < BUFSZ(p) && ismspace(BUF(p)[pos]))
882: pos++;
883: if (pos == BUFSZ(p) || '@' != BUF(p)[pos])
884: return(TEXICMD__MAX);
885: return(texicmd(p, pos, &end, NULL));
1.1 kristaps 886: }
887:
888: /*
889: * Parse a single word or command.
890: * This will return immediately at the EOF.
891: */
1.14 kristaps 892: static void
893: parsesingle(struct texi *p, size_t *pos)
1.1 kristaps 894: {
1.17 kristaps 895: size_t end, sv;
1.7 kristaps 896: enum texicmd cmd;
897: struct teximacro *macro;
1.1 kristaps 898:
1.14 kristaps 899: if ((*pos = advancenext(p, pos)) >= BUFSZ(p))
1.1 kristaps 900: return;
901:
1.14 kristaps 902: switch (BUF(p)[*pos]) {
1.1 kristaps 903: case ('}'):
904: if (0 == p->ign)
905: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 906: advance(p, pos);
1.1 kristaps 907: return;
908: case ('{'):
909: if (0 == p->ign)
910: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 911: advance(p, pos);
1.1 kristaps 912: return;
913: case ('@'):
914: break;
915: default:
1.14 kristaps 916: parseword(p, pos, '\0');
1.1 kristaps 917: return;
918: }
919:
1.17 kristaps 920: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 921: cmd = texicmd(p, *pos, &end, ¯o);
922: advanceto(p, pos, end);
1.7 kristaps 923: if (NULL != macro)
1.17 kristaps 924: texiexecmacro(p, macro, sv, pos);
1.1 kristaps 925: if (TEXICMD__MAX == cmd)
926: return;
927: if (NULL != texitoks[cmd].fp)
1.14 kristaps 928: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 929: }
930:
931: /*
932: * This is used in the @deffn type of command.
933: * These have an arbitrary number of line arguments; however, these
934: * arguments may or may not be surrounded by brackets.
935: * In this function, we parse each one as either a bracketed or
936: * non-bracketed argument, returning 0 when we've reached the end of
937: * line or 1 otherwise.
938: */
939: int
1.14 kristaps 940: parselinearg(struct texi *p, size_t *pos)
1.1 kristaps 941: {
942:
1.14 kristaps 943: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) {
1.1 kristaps 944: p->seenws = 1;
1.14 kristaps 945: advance(p, pos);
1.1 kristaps 946: }
947:
1.14 kristaps 948: if (*pos < BUFSZ(p) && '{' == BUF(p)[*pos])
1.18 kristaps 949: parsebracket(p, pos, 0);
1.14 kristaps 950: else if (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos])
951: parsesingle(p, pos);
1.1 kristaps 952: else
953: return(0);
954:
955: return(1);
956: }
957:
958: /*
959: * Parse til the end of the buffer.
960: */
1.14 kristaps 961: static void
962: parseeof(struct texi *p)
1.1 kristaps 963: {
964: size_t pos;
965:
1.14 kristaps 966: for (pos = 0; pos < BUFSZ(p); )
967: parsesingle(p, &pos);
1.1 kristaps 968: }
969:
1.8 kristaps 970: void
1.21 kristaps 971: texisplice(struct texi *p, const char *buf, size_t sz, size_t pos)
1.8 kristaps 972: {
1.14 kristaps 973: char *cp;
974: struct texifile *f;
1.8 kristaps 975:
1.14 kristaps 976: assert(p->filepos > 0);
977: f = &p->files[p->filepos - 1];
1.8 kristaps 978:
1.14 kristaps 979: if (f->mapsz + sz > f->mapmaxsz) {
980: f->mapmaxsz = f->mapsz + sz + 1024;
981: cp = realloc(f->map, f->mapmaxsz);
982: if (NULL == cp)
983: texiabort(p, NULL);
984: f->map = cp;
985: }
1.8 kristaps 986:
1.15 kristaps 987: f->insplice += sz;
1.21 kristaps 988: memmove(f->map + pos + sz, f->map + pos, f->mapsz - pos);
989: memcpy(f->map + pos, buf, sz);
1.14 kristaps 990: f->mapsz += sz;
1.8 kristaps 991: }
992:
993: /*
1.1 kristaps 994: * Parse a block sequence until we have the "@end endtoken" command
995: * invocation.
996: * This will return immediately at EOF.
997: */
998: void
1.14 kristaps 999: parseto(struct texi *p, size_t *pos, const char *endtoken)
1.1 kristaps 1000: {
1.17 kristaps 1001: size_t end, sv;
1.7 kristaps 1002: enum texicmd cmd;
1003: size_t endtoksz;
1004: struct teximacro *macro;
1.1 kristaps 1005:
1006: endtoksz = strlen(endtoken);
1007: assert(endtoksz > 0);
1008:
1.14 kristaps 1009: while ((*pos = advancenext(p, pos)) < BUFSZ(p)) {
1010: switch (BUF(p)[*pos]) {
1.1 kristaps 1011: case ('}'):
1012: if (0 == p->ign)
1013: texiwarn(p, "unexpected \"}\"");
1.14 kristaps 1014: advance(p, pos);
1.1 kristaps 1015: continue;
1016: case ('{'):
1017: if (0 == p->ign)
1018: texiwarn(p, "unexpected \"{\"");
1.14 kristaps 1019: advance(p, pos);
1.1 kristaps 1020: continue;
1021: case ('@'):
1022: break;
1023: default:
1.14 kristaps 1024: parseword(p, pos, '\0');
1.1 kristaps 1025: continue;
1026: }
1027:
1.17 kristaps 1028: sv = p->files[p->filepos - 1].insplice;
1.14 kristaps 1029: cmd = texicmd(p, *pos, &end, ¯o);
1030: advanceto(p, pos, end);
1.1 kristaps 1031: if (TEXICMD_END == cmd) {
1.14 kristaps 1032: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1033: advance(p, pos);
1.1 kristaps 1034: /*
1035: * FIXME: check the full word, not just its
1036: * initial substring!
1037: */
1.14 kristaps 1038: if (BUFSZ(p) - *pos >= endtoksz && 0 == strncmp
1039: (&BUF(p)[*pos], endtoken, endtoksz)) {
1040: advanceeoln(p, pos, 0);
1.1 kristaps 1041: break;
1042: }
1043: if (0 == p->ign)
1044: texiwarn(p, "unexpected \"end\"");
1.14 kristaps 1045: advanceeoln(p, pos, 0);
1.1 kristaps 1046: continue;
1.7 kristaps 1047: }
1048: if (NULL != macro)
1.17 kristaps 1049: texiexecmacro(p, macro, sv, pos);
1.7 kristaps 1050: if (TEXICMD__MAX == cmd)
1051: continue;
1052: if (NULL != texitoks[cmd].fp)
1.14 kristaps 1053: (*texitoks[cmd].fp)(p, cmd, pos);
1.1 kristaps 1054: }
1055: }
1056:
1057: /*
1.12 kristaps 1058: * Like parsefile() but used for reading from stdandard input.
1059: * This can only be called for the first file!
1060: */
1061: void
1062: parsestdin(struct texi *p)
1063: {
1064: struct texifile *f;
1065: ssize_t ssz;
1066:
1067: assert(0 == p->filepos);
1068: f = &p->files[p->filepos];
1069: memset(f, 0, sizeof(struct texifile));
1070:
1071: f->type = TEXISRC_STDIN;
1072: f->name = "<stdin>";
1073:
1.14 kristaps 1074: for (f->mapsz = 0; ; f->mapsz += (size_t)ssz) {
1075: if (f->mapsz == f->mapmaxsz) {
1076: if (f->mapmaxsz == (1U << 31))
1.12 kristaps 1077: texierr(p, "stdin buffer too long");
1.14 kristaps 1078: f->mapmaxsz = f->mapmaxsz > 65536 / 2 ?
1079: 2 * f->mapmaxsz : 65536;
1080: f->map = realloc(f->map, f->mapmaxsz);
1.12 kristaps 1081: if (NULL == f->map)
1082: texiabort(p, NULL);
1083: }
1.14 kristaps 1084: ssz = read(STDIN_FILENO, f->map +
1085: (int)f->mapsz, f->mapmaxsz - f->mapsz);
1.12 kristaps 1086: if (0 == ssz)
1087: break;
1088: else if (-1 == ssz)
1089: texiabort(p, NULL);
1090: }
1091:
1092: p->filepos++;
1.14 kristaps 1093: parseeof(p);
1.12 kristaps 1094: texifilepop(p);
1095: }
1096:
1097: /*
1.1 kristaps 1098: * Memory-map the file "fname" and begin parsing it unless "parse" is
1099: * zero, in which case we just dump the file to stdout (making sure it
1100: * doesn't trip up mdoc(7) along the way).
1101: * This can be called in a nested context.
1102: */
1103: void
1104: parsefile(struct texi *p, const char *fname, int parse)
1105: {
1106: struct texifile *f;
1107: int fd;
1108: struct stat st;
1109: size_t i;
1.14 kristaps 1110: char *map;
1.1 kristaps 1111:
1.5 kristaps 1112: if (64 == p->filepos)
1.6 kristaps 1113: texierr(p, "too many open files");
1.1 kristaps 1114: f = &p->files[p->filepos];
1115: memset(f, 0, sizeof(struct texifile));
1116:
1.12 kristaps 1117: f->type = TEXISRC_FILE;
1.1 kristaps 1118: f->name = fname;
1119: if (-1 == (fd = open(fname, O_RDONLY, 0))) {
1120: texiabort(p, fname);
1121: } else if (-1 == fstat(fd, &st)) {
1122: close(fd);
1123: texiabort(p, fname);
1124: }
1125:
1.14 kristaps 1126: f->mapsz = f->mapmaxsz = st.st_size;
1127: map = mmap(NULL, f->mapsz,
1.1 kristaps 1128: PROT_READ, MAP_SHARED, fd, 0);
1129: close(fd);
1130:
1.14 kristaps 1131: if (MAP_FAILED == map)
1.1 kristaps 1132: texiabort(p, fname);
1133:
1134: if ( ! parse) {
1.13 kristaps 1135: for (i = 0; i < f->mapsz; i++)
1.14 kristaps 1136: texiputchar(p, map[i]);
1.13 kristaps 1137: if (p->outcol)
1138: texiputchar(p, '\n');
1.14 kristaps 1139: munmap(map, f->mapsz);
1140: return;
1141: }
1142:
1143: p->filepos++;
1144: f->map = malloc(f->mapsz);
1145: memcpy(f->map, map, f->mapsz);
1146: munmap(map, f->mapsz);
1147: parseeof(p);
1.1 kristaps 1148: texifilepop(p);
1149: }
1150:
1.2 kristaps 1151: /*
1152: * Look up the value to a stored pair's value starting in "buf" from
1153: * start to end.
1154: * Return the pointer to the value memory, which can be NULL if the
1155: * pointer key does not exist.
1156: * The pointer can point to NULL if the value has been unset.
1157: */
1158: static char **
1.14 kristaps 1159: valuequery(const struct texi *p, size_t start, size_t end)
1.2 kristaps 1160: {
1161: size_t i, sz, len;
1162:
1163: assert(end >= start);
1164: /* Ignore zero-length. */
1165: if (0 == (len = (end - start)))
1166: return(NULL);
1167: for (i = 0; i < p->valsz; i++) {
1168: sz = strlen(p->vals[i].key);
1169: if (sz != len)
1170: continue;
1.14 kristaps 1171: if (0 == strncmp(p->vals[i].key, &BUF(p)[start], len))
1.2 kristaps 1172: return(&p->vals[i].value);
1173: }
1174: return(NULL);
1175: }
1176:
1177: /*
1178: * Parse a key until the end of line, e.g., @clear foo\n, and return the
1179: * pointer to its value via valuequery().
1180: */
1181: static char **
1.14 kristaps 1182: valuelquery(struct texi *p, size_t *pos)
1.2 kristaps 1183: {
1184: size_t start, end;
1185: char **ret;
1186:
1.14 kristaps 1187: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1188: advance(p, pos);
1189: if (*pos == BUFSZ(p))
1.2 kristaps 1190: return(NULL);
1.14 kristaps 1191: for (start = end = *pos; end < BUFSZ(p); end++)
1192: if ('\n' == BUF(p)[end])
1.2 kristaps 1193: break;
1.14 kristaps 1194: advanceto(p, pos, end);
1195: if (*pos < BUFSZ(p)) {
1196: assert('\n' == BUF(p)[*pos]);
1197: advance(p, pos);
1.2 kristaps 1198: }
1.14 kristaps 1199: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1200: return(NULL);
1201: return(ret);
1202: }
1203:
1204: void
1.14 kristaps 1205: valuelclear(struct texi *p, size_t *pos)
1.2 kristaps 1206: {
1207: char **ret;
1208:
1.14 kristaps 1209: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1210: return;
1211: free(*ret);
1212: *ret = NULL;
1213: }
1214:
1215: const char *
1.14 kristaps 1216: valuellookup(struct texi *p, size_t *pos)
1.2 kristaps 1217: {
1218: char **ret;
1219:
1.14 kristaps 1220: if (NULL == (ret = valuelquery(p, pos)))
1.2 kristaps 1221: return(NULL);
1222: return(*ret);
1223: }
1224:
1225: /*
1226: * Parse a key from a bracketed string, e.g., @value{foo}, and return
1227: * the pointer to its value.
1228: * If the returned pointer is NULL, either there was no string within
1229: * the brackets (or no brackets), or the value was not found, or the
1230: * value had previously been unset.
1231: */
1232: const char *
1.14 kristaps 1233: valueblookup(struct texi *p, size_t *pos)
1.2 kristaps 1234: {
1235: size_t start, end;
1236: char **ret;
1237:
1.14 kristaps 1238: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1239: advance(p, pos);
1240: if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos])
1.2 kristaps 1241: return(NULL);
1.14 kristaps 1242: advance(p, pos);
1243: for (start = end = *pos; end < BUFSZ(p); end++)
1244: if ('}' == BUF(p)[end])
1.2 kristaps 1245: break;
1.14 kristaps 1246: advanceto(p, pos, end);
1247: if (*pos < BUFSZ(p)) {
1248: assert('}' == BUF(p)[*pos]);
1249: advance(p, pos);
1.2 kristaps 1250: }
1.14 kristaps 1251: if (NULL == (ret = valuequery(p, start, end)))
1.2 kristaps 1252: return(NULL);
1253: return(*ret);
1254: }
1255:
1256: void
1257: valueadd(struct texi *p, char *key, char *val)
1258: {
1259: size_t i;
1260:
1261: assert(NULL != key);
1262: assert(NULL != val);
1263:
1264: for (i = 0; i < p->valsz; i++)
1265: if (0 == strcmp(p->vals[i].key, key))
1266: break;
1267:
1268: if (i < p->valsz) {
1269: free(key);
1270: free(p->vals[i].value);
1271: p->vals[i].value = val;
1272: } else {
1.4 kristaps 1273: /* FIXME: reallocarray() */
1.2 kristaps 1274: p->vals = realloc(p->vals,
1275: (p->valsz + 1) *
1276: sizeof(struct texivalue));
1.4 kristaps 1277: if (NULL == p->vals)
1278: texiabort(p, NULL);
1.2 kristaps 1279: p->vals[p->valsz].key = key;
1280: p->vals[p->valsz].value = val;
1281: p->valsz++;
1282: }
1.7 kristaps 1283: }
1284:
1285: /*
1286: * Take the arguments to a macro, e.g., @foo{bar, baz, xyzzy} (or the
1287: * declaration form, @macro foo {arg1, ...}) and textually convert it to
1288: * an array of arguments of size "argsz".
1289: * These need to be freed individually and as a whole.
1290: * NOTE: this will puke on @, or @} macros, which can trick it into
1291: * stopping argument parsing earlier.
1292: * Ergo, textual: this doesn't interpret the arguments in any way.
1293: */
1294: char **
1.14 kristaps 1295: argparse(struct texi *p, size_t *pos, size_t *argsz, size_t hint)
1.7 kristaps 1296: {
1297: char **args;
1298: size_t start, end, stack;
1299:
1.14 kristaps 1300: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1301: advance(p, pos);
1.7 kristaps 1302:
1303: args = NULL;
1304: *argsz = 0;
1305:
1.17 kristaps 1306: if (*pos == BUFSZ(p))
1307: return(args);
1308:
1.14 kristaps 1309: if ('{' != BUF(p)[*pos] && hint) {
1.10 kristaps 1310: /*
1311: * Special case: if we encounter an unbracketed argument
1312: * and we're being invoked with non-zero arguments
1313: * (versus being set, i.e., hint>0), then parse until
1314: * the end of line.
1315: */
1316: *argsz = 1;
1317: args = calloc(1, sizeof(char *));
1318: if (NULL == args)
1319: texiabort(p, NULL);
1320: start = *pos;
1.14 kristaps 1321: while (*pos < BUFSZ(p)) {
1322: if ('\n' == BUF(p)[*pos])
1.10 kristaps 1323: break;
1.14 kristaps 1324: advance(p, pos);
1.10 kristaps 1325: }
1326: args[0] = malloc(*pos - start + 1);
1.14 kristaps 1327: memcpy(args[0], &BUF(p)[start], *pos - start);
1.10 kristaps 1328: args[0][*pos - start] = '\0';
1.14 kristaps 1329: if (*pos < BUFSZ(p) && '\n' == BUF(p)[*pos])
1330: advance(p, pos);
1.10 kristaps 1331: return(args);
1.14 kristaps 1332: } else if ('{' != BUF(p)[*pos])
1.7 kristaps 1333: return(args);
1.17 kristaps 1334:
1335: assert('{' == BUF(p)[*pos]);
1.7 kristaps 1336:
1337: /* Parse til the closing '}', putting into the array. */
1.14 kristaps 1338: advance(p, pos);
1339: while (*pos < BUFSZ(p)) {
1340: while (*pos < BUFSZ(p) && isws(BUF(p)[*pos]))
1341: advance(p, pos);
1.7 kristaps 1342: start = *pos;
1343: stack = 0;
1.14 kristaps 1344: while (*pos < BUFSZ(p)) {
1.7 kristaps 1345: /*
1346: * According to the manual, commas within
1347: * embedded commands are escaped.
1348: * We keep track of embedded-ness in the "stack"
1349: * state anyway, so this is free.
1350: */
1.14 kristaps 1351: if (',' == BUF(p)[*pos] && 0 == stack && 1 != hint)
1.7 kristaps 1352: break;
1.14 kristaps 1353: else if (0 == stack && '}' == BUF(p)[*pos])
1.7 kristaps 1354: break;
1.14 kristaps 1355: else if (0 != stack && '}' == BUF(p)[*pos])
1.7 kristaps 1356: stack--;
1.14 kristaps 1357: else if ('{' == BUF(p)[*pos])
1.7 kristaps 1358: stack++;
1.14 kristaps 1359: advance(p, pos);
1.7 kristaps 1360: }
1361: if (stack)
1362: texiwarn(p, "unterminated macro "
1363: "in macro arguments");
1.14 kristaps 1364: if ((end = *pos) == BUFSZ(p))
1.7 kristaps 1365: break;
1366: /* Test for zero-length '{ }'. */
1.14 kristaps 1367: if (start == end && '}' == BUF(p)[*pos] && 0 == *argsz)
1.7 kristaps 1368: break;
1369: /* FIXME: use reallocarray. */
1370: args = realloc
1371: (args, sizeof(char *) *
1372: (*argsz + 1));
1373: if (NULL == args)
1374: texiabort(p, NULL);
1375: args[*argsz] = malloc(end - start + 1);
1376: if (NULL == args[*argsz])
1377: texiabort(p, NULL);
1378: memcpy(args[*argsz],
1.14 kristaps 1379: &BUF(p)[start], end - start);
1.7 kristaps 1380: args[*argsz][end - start] = '\0';
1381: (*argsz)++;
1.14 kristaps 1382: if ('}' == BUF(p)[*pos])
1.7 kristaps 1383: break;
1.14 kristaps 1384: advance(p, pos);
1.7 kristaps 1385: }
1386:
1.14 kristaps 1387: if (*pos == BUFSZ(p))
1.7 kristaps 1388: texierr(p, "unterminated arguments");
1.14 kristaps 1389: assert('}' == BUF(p)[*pos]);
1390: advance(p, pos);
1.7 kristaps 1391: return(args);
1.2 kristaps 1392: }
1.20 kristaps 1393:
1394: /*
1395: * If we're printing chapters, then do some naviation here and then
1396: * close our outfile.
1397: * I want to call this the SEE ALSO section, but that's not really what
1398: * it is: we'll refer to the "initial" (top) node and the next and
1399: * previous chapters.
1400: */
1401: void
1402: teximdocclose(struct texi *p, int last)
1403: {
1404: char buf[PATH_MAX];
1405:
1406: if (NULL == p->chapters || 0 == p->chapnum)
1407: return;
1408:
1409: teximacro(p, "Sh INFO NAVIGATION");
1410:
1411: /* Print a reference to the "top" node. */
1412: if (p->chapnum > 1) {
1.22 kristaps 1413: texiputchars(p, "Top node,");
1.20 kristaps 1414: snprintf(buf, sizeof(buf), "node1 7");
1415: teximacroopen(p, "Xr ");
1416: texiputchars(p, buf);
1.22 kristaps 1417: texiputchars(p, " ;");
1.20 kristaps 1418: teximacroclose(p);
1419: }
1420:
1421: /* Print a reference to the previous node. */
1422: if (p->chapnum > 2) {
1.22 kristaps 1423: texiputchars(p, "previous node,");
1.20 kristaps 1424: snprintf(buf, sizeof(buf),
1425: "node%zu 7", p->chapnum - 1);
1426: teximacroopen(p, "Xr ");
1427: texiputchars(p, buf);
1428: if ( ! last)
1.22 kristaps 1429: texiputchars(p, " ;");
1.20 kristaps 1430: teximacroclose(p);
1431: }
1432:
1433: /* Print a reference to the next node. */
1434: if ( ! last) {
1.22 kristaps 1435: if (1 == p->chapnum)
1436: texiputchars(p, "Next node,");
1437: else
1438: texiputchars(p, "next node,");
1.20 kristaps 1439: snprintf(buf, sizeof(buf),
1440: "node%zu 7", p->chapnum + 1);
1441: teximacroopen(p, "Xr ");
1442: texiputchars(p, buf);
1443: teximacroclose(p);
1444: }
1445:
1446: fclose(p->outfile);
1447: }
1448:
1449: /*
1450: * Open a mdoc(7) context.
1451: * If we're printing chapters, then open the outfile here, too.
1452: * Otherwise just print the mdoc(7) prologue.
1453: */
1454: void
1.21 kristaps 1455: teximdocopen(struct texi *p, size_t *pos)
1.20 kristaps 1456: {
1457: const char *cp;
1458: time_t t;
1459: char date[32];
1460: char fname[PATH_MAX];
1461:
1462: if (NULL != p->chapters) {
1463: snprintf(fname, sizeof(fname), "%s/node%zu.7",
1464: p->chapters, ++p->chapnum);
1465: p->outfile = fopen(fname, "w");
1466: if (NULL == p->outfile)
1467: texiabort(p, fname);
1468: }
1469:
1470: /*
1471: * Here we print our standard mdoc(7) prologue.
1472: * We use the title set with @settitle for the `Nd' description
1473: * and the source document filename (the first one as invoked on
1474: * the command line) for the title.
1475: * The date is set to the current date.
1476: */
1477: t = time(NULL);
1478: strftime(date, sizeof(date), "%F", localtime(&t));
1479:
1480: teximacroopen(p, "Dd");
1481: texiputchars(p, date);
1482: teximacroclose(p);
1483: teximacroopen(p, "Dt");
1484: for (cp = p->title; '\0' != *cp; cp++)
1485: texiputchar(p, toupper((unsigned int)*cp));
1486: texiputchars(p, " 7");
1487: teximacroclose(p);
1488: teximacro(p, "Os");
1489: teximacro(p, "Sh NAME");
1490: teximacroopen(p, "Nm");
1491: for (cp = p->title; '\0' != *cp; cp++)
1492: texiputchar(p, *cp);
1493: teximacroclose(p);
1494: teximacroopen(p, "Nd");
1.21 kristaps 1495: /*
1496: * The subtitle `Nd' can consist of arbitrary macros, so paste
1497: * it and parse to the end of the line.
1498: */
1499: if (NULL != p->subtitle) {
1500: texisplice(p, p->subtitle, strlen(p->subtitle), *pos);
1501: parseeoln(p, pos);
1502: } else
1.20 kristaps 1503: texiputchars(p, "Unknown description");
1504: teximacroclose(p);
1505: }
1506:
CVSweb