Annotation of pod2mdoc/pod2mdoc.c, Revision 1.33
1.33 ! schwarze 1: /* $Id: pod2mdoc.c,v 1.32 2014/07/18 05:09:32 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2014 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 AUTHORS DISCLAIM ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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/stat.h>
18: #include <sys/time.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
22: #include <fcntl.h>
23: #include <getopt.h>
24: #include <stdio.h>
25: #include <stdlib.h>
26: #include <string.h>
27: #include <unistd.h>
28:
1.10 kristaps 29: /*
1.19 kristaps 30: * In what section can we find Perl module manuals?
31: * Sometimes (Mac OS X) it's 3pm, sometimes (OpenBSD, etc.) 3p.
32: * XXX IF YOU CHANGE THIS, CHANGE POD2MDOC.1 AS WELL.
1.10 kristaps 33: */
34: #define PERL_SECTION "3p"
35:
1.1 schwarze 36: struct args {
37: const char *title; /* override "Dt" title */
38: const char *date; /* override "Dd" date */
39: const char *section; /* override "Dt" section */
40: };
41:
1.4 schwarze 42: enum list {
43: LIST_BULLET = 0,
44: LIST_ENUM,
45: LIST_TAG,
46: LIST__MAX
47: };
48:
1.11 kristaps 49: enum sect {
50: SECT_NONE = 0,
51: SECT_NAME, /* NAME section */
52: SECT_SYNOPSIS, /* SYNOPSIS section */
53: };
54:
1.32 schwarze 55: enum outstate {
56: OUST_NL = 0, /* just started a new output line */
57: OUST_TXT, /* text line output in progress */
58: OUST_MAC /* macro line output in progress */
59: };
60:
1.1 schwarze 61: struct state {
1.31 schwarze 62: const char *fname; /* file being parsed */
1.1 schwarze 63: int parsing; /* after =cut of before command */
64: int paused; /* in =begin and before =end */
1.11 kristaps 65: enum sect sect; /* which section are we in? */
1.4 schwarze 66: #define LIST_STACKSZ 128
67: enum list lstack[LIST_STACKSZ]; /* open lists */
68: size_t lpos; /* where in list stack */
1.31 schwarze 69: int haspar; /* in paragraph: do we need Pp? */
1.32 schwarze 70: enum outstate oust; /* state of the mdoc output stream */
71: int wantws; /* let mdoc(7) output whitespace here */
1.31 schwarze 72: char *outbuf; /* text buffered for output */
73: size_t outbufsz; /* allocated size of outbuf */
74: size_t outbuflen; /* current length of outbuf */
1.1 schwarze 75: };
76:
77: enum fmt {
78: FMT_ITALIC,
79: FMT_BOLD,
80: FMT_CODE,
81: FMT_LINK,
82: FMT_ESCAPE,
83: FMT_FILE,
84: FMT_NBSP,
85: FMT_INDEX,
86: FMT_NULL,
87: FMT__MAX
88: };
89:
90: enum cmd {
91: CMD_POD = 0,
92: CMD_HEAD1,
93: CMD_HEAD2,
94: CMD_HEAD3,
95: CMD_HEAD4,
96: CMD_OVER,
97: CMD_ITEM,
98: CMD_BACK,
99: CMD_BEGIN,
100: CMD_END,
101: CMD_FOR,
102: CMD_ENCODING,
103: CMD_CUT,
104: CMD__MAX
105: };
106:
107: static const char *const cmds[CMD__MAX] = {
108: "pod", /* CMD_POD */
109: "head1", /* CMD_HEAD1 */
110: "head2", /* CMD_HEAD2 */
111: "head3", /* CMD_HEAD3 */
112: "head4", /* CMD_HEAD4 */
113: "over", /* CMD_OVER */
114: "item", /* CMD_ITEM */
115: "back", /* CMD_BACK */
116: "begin", /* CMD_BEGIN */
117: "end", /* CMD_END */
118: "for", /* CMD_FOR */
119: "encoding", /* CMD_ENCODING */
120: "cut" /* CMD_CUT */
121: };
122:
123: static const char fmts[FMT__MAX] = {
124: 'I', /* FMT_ITALIC */
125: 'B', /* FMT_BOLD */
126: 'C', /* FMT_CODE */
127: 'L', /* FMT_LINK */
128: 'E', /* FMT_ESCAPE */
129: 'F', /* FMT_FILE */
130: 'S', /* FMT_NBSP */
131: 'X', /* FMT_INDEX */
132: 'Z' /* FMT_NULL */
133: };
134:
1.6 kristaps 135: static int last;
136:
1.31 schwarze 137:
138: static void
139: outbuf_grow(struct state *st, size_t by)
140: {
141:
142: st->outbufsz += (by / 128 + 1) * 128;
143: st->outbuf = realloc(st->outbuf, st->outbufsz);
144: if (NULL == st->outbuf) {
145: perror(NULL);
146: exit(EXIT_FAILURE);
147: }
148: }
149:
150: static void
151: outbuf_addchar(struct state *st)
152: {
153:
154: if (st->outbuflen + 2 >= st->outbufsz)
155: outbuf_grow(st, 1);
156: st->outbuf[st->outbuflen++] = last;
157: if ('\\' == last)
158: st->outbuf[st->outbuflen++] = 'e';
159: st->outbuf[st->outbuflen] = '\0';
1.32 schwarze 160: st->wantws = 0;
1.31 schwarze 161: }
162:
163: static void
164: outbuf_addstr(struct state *st, const char *str)
165: {
166: size_t slen;
167:
168: slen = strlen(str);
169: if (st->outbuflen + slen >= st->outbufsz)
170: outbuf_grow(st, slen);
171: memcpy(st->outbuf + st->outbuflen, str, slen+1);
1.33 ! schwarze 172: st->outbuflen += slen;
1.31 schwarze 173: last = str[slen - 1];
1.32 schwarze 174: st->wantws = 0;
1.31 schwarze 175: }
176:
177: static void
178: outbuf_flush(struct state *st)
179: {
180:
181: if (0 == st->outbuflen)
182: return;
183:
184: fputs(st->outbuf, stdout);
185: *st->outbuf = '\0';
186: st->outbuflen = 0;
1.32 schwarze 187:
188: if (OUST_NL == st->oust)
189: st->oust = OUST_TXT;
1.31 schwarze 190: }
191:
192: static void
1.32 schwarze 193: mdoc_newln(struct state *st)
1.31 schwarze 194: {
195:
1.32 schwarze 196: if (OUST_NL == st->oust)
1.31 schwarze 197: return;
1.32 schwarze 198:
1.31 schwarze 199: putchar('\n');
200: last = '\n';
1.32 schwarze 201: st->oust = OUST_NL;
202: st->wantws = 1;
1.31 schwarze 203: }
204:
1.1 schwarze 205: /*
206: * Given buf[*start] is at the start of an escape name, read til the end
207: * of the escape ('>') then try to do something with it.
208: * Sets start to be one after the '>'.
1.32 schwarze 209: *
210: * This function does not care about output modes,
211: * it merely appends text to the output buffer,
212: * which can then be used in any mode.
1.1 schwarze 213: */
214: static void
1.31 schwarze 215: formatescape(struct state *st, const char *buf, size_t *start, size_t end)
1.1 schwarze 216: {
217: char esc[16]; /* no more needed */
218: size_t i, max;
219:
220: max = sizeof(esc) - 1;
221: i = 0;
222: /* Read til our buffer is full. */
223: while (*start < end && '>' != buf[*start] && i < max)
224: esc[i++] = buf[(*start)++];
225: esc[i] = '\0';
226:
227: if (i == max) {
228: /* Too long... skip til we end. */
229: while (*start < end && '>' != buf[*start])
230: (*start)++;
231: return;
232: } else if (*start >= end)
233: return;
234:
235: assert('>' == buf[*start]);
236: (*start)++;
237:
238: /*
239: * TODO: right now, we only recognise the named escapes.
240: * Just let the rest of them go.
241: */
1.6 kristaps 242: if (0 == strcmp(esc, "lt"))
1.31 schwarze 243: outbuf_addstr(st, "\\(la");
1.1 schwarze 244: else if (0 == strcmp(esc, "gt"))
1.31 schwarze 245: outbuf_addstr(st, "\\(ra");
1.33 ! schwarze 246: else if (0 == strcmp(esc, "verbar"))
1.31 schwarze 247: outbuf_addstr(st, "\\(ba");
1.1 schwarze 248: else if (0 == strcmp(esc, "sol"))
1.31 schwarze 249: outbuf_addstr(st, "\\(sl");
1.1 schwarze 250: }
251:
252: /*
1.9 kristaps 253: * Run some heuristics to intuit a link format.
1.19 kristaps 254: * I set "start" to be the end of the sequence (last right-carrot) so
1.9 kristaps 255: * that the caller can safely just continue processing.
1.19 kristaps 256: * If this is just an empty tag, I'll return 0.
1.32 schwarze 257: *
258: * Always operates in OUST_MAC mode.
259: * Mode handling is done by the caller.
1.9 kristaps 260: */
261: static int
262: trylink(const char *buf, size_t *start, size_t end, size_t dsz)
263: {
1.21 kristaps 264: size_t linkstart, realend, linkend,
265: i, j, textsz, stack;
1.9 kristaps 266:
267: /*
268: * Scan to the start of the terminus.
269: * This function is more or less replicated in the formatcode()
270: * for null or index formatting codes.
1.23 kristaps 271: * However, we're slightly different because we might have
272: * nested escapes we need to ignore.
1.9 kristaps 273: */
1.21 kristaps 274: stack = 0;
1.19 kristaps 275: for (linkstart = realend = *start; realend < end; realend++) {
1.23 kristaps 276: if ('<' == buf[realend])
277: stack++;
1.19 kristaps 278: if ('>' != buf[realend])
1.9 kristaps 279: continue;
1.23 kristaps 280: else if (stack-- > 0)
281: continue;
282: if (dsz == 1)
1.9 kristaps 283: break;
1.19 kristaps 284: assert(realend > 0);
285: if (' ' != buf[realend - 1])
1.9 kristaps 286: continue;
1.19 kristaps 287: for (i = realend, j = 0; i < end && j < dsz; j++)
1.9 kristaps 288: if ('>' != buf[i++])
289: break;
290: if (dsz == j)
291: break;
292: }
1.19 kristaps 293:
294: /* Ignore stubs. */
295: if (realend == end || realend == *start)
1.9 kristaps 296: return(0);
297:
1.19 kristaps 298: /* Set linkend to the end of content. */
299: linkend = dsz > 1 ? realend - 1 : realend;
1.18 kristaps 300:
1.19 kristaps 301: /* Re-scan to see if we have a title or section. */
302: for (textsz = *start; textsz < linkend; textsz++)
303: if ('|' == buf[textsz] || '/' == buf[textsz])
1.18 kristaps 304: break;
305:
1.19 kristaps 306: if (textsz < linkend && '|' == buf[textsz]) {
1.20 kristaps 307: /* With title: set start, then end at section. */
1.19 kristaps 308: linkstart = textsz + 1;
1.18 kristaps 309: textsz = textsz - *start;
1.19 kristaps 310: for (i = linkstart; i < linkend; i++)
311: if ('/' == buf[i])
312: break;
313: if (i < linkend)
314: linkend = i;
1.20 kristaps 315: } else if (textsz < linkend && '/' == buf[textsz]) {
316: /* With section: set end at section. */
317: linkend = textsz;
318: textsz = 0;
319: } else
320: /* No title, no section. */
1.18 kristaps 321: textsz = 0;
1.19 kristaps 322:
323: *start = realend;
324: j = linkend - linkstart;
325:
1.20 kristaps 326: /* Do we have only subsection material? */
327: if (0 == j && '/' == buf[linkend]) {
328: linkstart = linkend + 1;
329: linkend = dsz > 1 ? realend - 1 : realend;
330: if (0 == (j = linkend - linkstart))
331: return(0);
332: printf("Sx %.*s", (int)j, &buf[linkstart]);
333: return(1);
334: } else if (0 == j)
1.19 kristaps 335: return(0);
336:
337: /* See if we qualify as being a link or not. */
1.20 kristaps 338: if ((j > 4 && 0 == memcmp("http:", &buf[linkstart], j)) ||
339: (j > 5 && 0 == memcmp("https:", &buf[linkstart], j)) ||
340: (j > 3 && 0 == memcmp("ftp:", &buf[linkstart], j)) ||
341: (j > 4 && 0 == memcmp("sftp:", &buf[linkstart], j)) ||
342: (j > 3 && 0 == memcmp("smb:", &buf[linkstart], j)) ||
343: (j > 3 && 0 == memcmp("afs:", &buf[linkstart], j))) {
344: /* Gross. */
345: printf("Lk %.*s", (int)((dsz > 1 ? realend - 1 :
346: realend) - linkstart), &buf[linkstart]);
1.19 kristaps 347: return(1);
348: }
349:
350: /* See if we qualify as a mailto. */
1.20 kristaps 351: if (j > 6 && 0 == memcmp("mailto:", &buf[linkstart], j)) {
1.19 kristaps 352: printf("Mt %.*s", (int)j, &buf[linkstart]);
353: return(1);
354: }
355:
356: /* See if we're a foo(5), foo(5x), or foo(5xx) manpage. */
357: if ((j > 3 && ')' == buf[linkend - 1]) &&
358: ('(' == buf[linkend - 3])) {
359: printf("Xr %.*s %c", (int)(j - 3),
360: &buf[linkstart], buf[linkend - 2]);
361: return(1);
362: } else if ((j > 4 && ')' == buf[linkend - 1]) &&
363: ('(' == buf[linkend - 4])) {
364: printf("Xr %.*s %.*s", (int)(j - 4),
365: &buf[linkstart], 2, &buf[linkend - 3]);
366: return(1);
367: } else if ((j > 5 && ')' == buf[linkend - 1]) &&
368: ('(' == buf[linkend - 5])) {
369: printf("Xr %.*s %.*s", (int)(j - 5),
370: &buf[linkstart], 3, &buf[linkend - 4]);
371: return(1);
372: }
373:
374: /* Last try: do we have a double-colon? */
375: for (i = linkstart + 1; i < linkend; i++)
376: if (':' == buf[i] && ':' == buf[i - 1])
1.18 kristaps 377: break;
1.9 kristaps 378:
1.19 kristaps 379: if (i < linkend)
1.10 kristaps 380: printf("Xr %.*s " PERL_SECTION,
1.19 kristaps 381: (int)j, &buf[linkstart]);
1.9 kristaps 382: else
1.19 kristaps 383: printf("Xr %.*s 1", (int)j, &buf[linkstart]);
1.9 kristaps 384:
385: return(1);
386: }
387:
1.13 kristaps 388: /*
389: * Doclifting: if we're a bold "-xx" and we're in the SYNOPSIS section,
390: * then it's likely that we're a flag.
391: * Our flag might be followed by an argument, so make sure that we're
392: * accounting for that, too.
393: * If we don't have a flag at all, however, then assume we're an "Ar".
1.32 schwarze 394: *
395: * Always operates in OUST_MAC mode.
396: * Mode handlinf is done by the caller.
1.13 kristaps 397: */
398: static void
399: dosynopsisfl(const char *buf, size_t *start, size_t end)
400: {
401: size_t i;
402: again:
1.14 kristaps 403: assert(*start + 1 < end);
404: assert('-' == buf[*start]);
405:
406: if ( ! isalnum((int)buf[*start + 1]) &&
407: '?' != buf[*start + 1] &&
408: '-' != buf[*start + 1]) {
409: (*start)--;
410: fputs("Ar ", stdout);
411: return;
412: }
413:
1.13 kristaps 414: (*start)++;
415: for (i = *start; i < end; i++)
416: if (isalnum((int)buf[i]))
417: continue;
1.14 kristaps 418: else if ('?' == buf[i])
419: continue;
1.13 kristaps 420: else if ('-' == buf[i])
421: continue;
422: else if ('_' == buf[i])
423: continue;
424: else
425: break;
426:
427: assert(i < end);
428:
429: if ( ! (' ' == buf[i] || '>' == buf[i])) {
430: printf("Ar ");
431: return;
432: }
433:
434: printf("Fl ");
435: if (end - *start > 1 &&
436: isupper((int)buf[*start]) &&
437: islower((int)buf[*start + 1]) &&
438: (end - *start == 2 ||
439: ' ' == buf[*start + 2]))
440: printf("\\&");
441: printf("%.*s ", (int)(i - *start), &buf[*start]);
442: *start = i;
443:
444: if (' ' == buf[i]) {
445: while (i < end && ' ' == buf[i])
446: i++;
447: assert(i < end);
448: if ('-' == buf[i]) {
449: *start = i;
450: goto again;
451: }
452: printf("Ar ");
453: *start = i;
454: }
455: }
456:
1.9 kristaps 457: /*
1.1 schwarze 458: * We're at the character in front of a format code, which is structured
459: * like X<...> and can contain nested format codes.
460: * This consumes the whole format code, and any nested format codes, til
461: * the end of matched production.
1.6 kristaps 462: * If "nomacro", then we don't print any macros, just contained data
463: * (e.g., following "Sh" or "Nm").
1.15 kristaps 464: * "pos" is only significant in SYNOPSIS, and should be 0 when invoked
465: * as the first format code on a line (for decoration as an "Nm"),
466: * non-zero otherwise.
1.32 schwarze 467: *
468: * Output mode handling is most complicated here.
469: * We may enter in any mode.
470: * We usually exit in OUST_MAC mode, except when
471: * entering without OUST_MAC and the code is invalid.
1.1 schwarze 472: */
1.33 ! schwarze 473: static int
1.15 kristaps 474: formatcode(struct state *st, const char *buf, size_t *start,
1.32 schwarze 475: size_t end, int nomacro, int pos)
1.1 schwarze 476: {
477: enum fmt fmt;
1.5 kristaps 478: size_t i, j, dsz;
1.1 schwarze 479:
480: assert(*start + 1 < end);
481: assert('<' == buf[*start + 1]);
482:
1.6 kristaps 483: /*
484: * First, look up the format code.
1.30 schwarze 485: * If it's not valid, treat it as a NOOP.
1.6 kristaps 486: */
487: for (fmt = 0; fmt < FMT__MAX; fmt++)
488: if (buf[*start] == fmts[fmt])
489: break;
490:
1.5 kristaps 491: /*
492: * Determine whether we're overriding our delimiter.
493: * According to POD, if we have more than one '<' followed by a
494: * space, then we need a space followed by matching '>' to close
495: * the expression.
496: * Otherwise we use the usual '<' and '>' matched pair.
497: */
498: i = *start + 1;
499: while (i < end && '<' == buf[i])
500: i++;
501: assert(i > *start + 1);
502: dsz = i - (*start + 1);
503: if (dsz > 1 && (i >= end || ' ' != buf[i]))
504: dsz = 1;
505:
506: /* Remember, if dsz>1, to jump the trailing space. */
507: *start += dsz + 1 + (dsz > 1 ? 1 : 0);
1.1 schwarze 508:
509: /*
1.6 kristaps 510: * Escapes and ignored codes (NULL and INDEX) don't print macro
511: * sequences, so just output them like normal text before
512: * processing for real macros.
1.1 schwarze 513: */
514: if (FMT_ESCAPE == fmt) {
1.31 schwarze 515: formatescape(st, buf, start, end);
1.33 ! schwarze 516: return(0);
1.1 schwarze 517: } else if (FMT_NULL == fmt || FMT_INDEX == fmt) {
1.5 kristaps 518: /*
1.6 kristaps 519: * Just consume til the end delimiter, accounting for
520: * whether it's a custom one.
1.5 kristaps 521: */
522: for ( ; *start < end; (*start)++) {
523: if ('>' != buf[*start])
524: continue;
525: else if (dsz == 1)
526: break;
527: assert(*start > 0);
528: if (' ' != buf[*start - 1])
529: continue;
530: i = *start;
531: for (j = 0; i < end && j < dsz; j++)
532: if ('>' != buf[i++])
533: break;
534: if (dsz != j)
535: continue;
536: (*start) += dsz;
537: break;
538: }
1.24 kristaps 539: if (*start < end) {
540: assert('>' == buf[*start]);
541: (*start)++;
542: }
543: if (isspace(last))
544: while (*start < end && isspace((int)buf[*start]))
545: (*start)++;
1.33 ! schwarze 546: return(0);
1.1 schwarze 547: }
548:
1.6 kristaps 549: /*
550: * Check whether we're supposed to print macro stuff (this is
551: * suppressed in, e.g., "Nm" and "Sh" macros).
552: */
1.30 schwarze 553: if (FMT__MAX != fmt && !nomacro) {
1.32 schwarze 554:
555: /*
556: * We may already have wantws if there was whitespace
557: * before the code ("text B<text"), but initial
558: * whitespace inside our scope ("textB< text")
559: * allows to break at this point as well.
560: */
561:
562: st->wantws |= ' ' == buf[*start];
1.31 schwarze 563:
1.1 schwarze 564: /*
1.31 schwarze 565: * If we are on a text line and there is no
566: * whitespace before our content, we have to make
567: * the previous word a prefix to the macro line.
1.32 schwarze 568: * In the following, mdoc_newln() must not be used
569: * lest we clobber out output state.
1.1 schwarze 570: */
1.31 schwarze 571:
1.32 schwarze 572: if (OUST_MAC != st->oust && !st->wantws) {
573: if (OUST_NL != st->oust)
1.31 schwarze 574: putchar('\n');
575: printf(".Pf ");
576: }
577:
578: outbuf_flush(st);
579:
580: /* Whitespace is easier to suppress on macro lines. */
581:
1.32 schwarze 582: if (OUST_MAC == st->oust && !st->wantws)
583: printf(" Ns ");
1.31 schwarze 584:
585: /* Unless we are on a macro line, start one. */
586:
1.32 schwarze 587: if (OUST_MAC != st->oust && st->wantws) {
588: if (OUST_NL != st->oust)
1.6 kristaps 589: putchar('\n');
1.1 schwarze 590: putchar('.');
1.31 schwarze 591: } else
1.1 schwarze 592: putchar(' ');
1.31 schwarze 593:
1.32 schwarze 594: /*
595: * Print the macro corresponding to this format code,
596: * and update the output state afterwards.
597: */
1.6 kristaps 598:
1.1 schwarze 599: switch (fmt) {
600: case (FMT_ITALIC):
601: printf("Em ");
602: break;
603: case (FMT_BOLD):
1.14 kristaps 604: if (SECT_SYNOPSIS == st->sect) {
605: if (1 == dsz && '-' == buf[*start])
606: dosynopsisfl(buf, start, end);
1.15 kristaps 607: else if (0 == pos)
608: printf("Nm ");
1.14 kristaps 609: else
610: printf("Ar ");
611: break;
612: }
1.27 schwarze 613: if (0 == strncmp(buf + *start, "NULL", 4) &&
614: ('=' == buf[*start + 4] ||
615: '>' == buf[*start + 4]))
616: printf("Dv ");
617: else
618: printf("Sy ");
1.1 schwarze 619: break;
620: case (FMT_CODE):
1.2 schwarze 621: printf("Qo Li ");
1.1 schwarze 622: break;
623: case (FMT_LINK):
1.19 kristaps 624: /* Try to link; use "No" if it's empty. */
1.9 kristaps 625: if ( ! trylink(buf, start, end, dsz))
626: printf("No ");
1.1 schwarze 627: break;
628: case (FMT_FILE):
629: printf("Pa ");
630: break;
631: case (FMT_NBSP):
632: printf("No ");
633: break;
634: default:
635: abort();
636: }
1.32 schwarze 637: st->oust = OUST_MAC;
638: st->wantws = 1;
1.31 schwarze 639: } else
640: outbuf_flush(st);
1.1 schwarze 641:
642: /*
1.6 kristaps 643: * Process until we reach the end marker (e.g., '>') or until we
1.5 kristaps 644: * find a nested format code.
1.1 schwarze 645: * Don't emit any newlines: since we're on a macro line, we
646: * don't want to break the line.
647: */
648: while (*start < end) {
1.5 kristaps 649: if ('>' == buf[*start] && 1 == dsz) {
1.1 schwarze 650: (*start)++;
651: break;
1.5 kristaps 652: } else if ('>' == buf[*start] &&
653: ' ' == buf[*start - 1]) {
654: /*
655: * Handle custom delimiters.
656: * These require a certain number of
657: * space-preceded carrots before we're really at
658: * the end.
659: */
660: i = *start;
661: for (j = 0; i < end && j < dsz; j++)
662: if ('>' != buf[i++])
663: break;
664: if (dsz == j) {
665: *start += dsz;
666: break;
667: }
1.1 schwarze 668: }
669: if (*start + 1 < end && '<' == buf[*start + 1]) {
1.32 schwarze 670: formatcode(st, buf, start, end, nomacro, 1);
1.1 schwarze 671: continue;
672: }
1.3 schwarze 673:
1.32 schwarze 674: /* Suppress newlines and multiple spaces. */
675:
676: last = buf[(*start)++];
677: if (' ' == last || '\n' == last) {
678: putchar(' ');
679: while (*start < end && ' ' == buf[*start])
680: (*start)++;
681: continue;
682: }
683:
1.33 ! schwarze 684: if (OUST_MAC == st->oust && FMT__MAX != fmt) {
1.32 schwarze 685: if ( ! st->wantws) {
686: printf(" Ns ");
687: st->wantws = 1;
688: }
689:
690: /*
691: * Escape macro-like words.
692: * This matches "Xx " and "XxEOLN".
693: */
694:
695: if (end - *start > 0 &&
696: isupper((unsigned char)last) &&
697: islower((unsigned char)buf[*start]) &&
698: (end - *start == 1 ||
699: ' ' == buf[*start + 1] ||
700: '>' == buf[*start + 1]))
701: printf("\\&");
702: }
1.3 schwarze 703:
1.32 schwarze 704: putchar(last);
1.4 schwarze 705:
1.8 kristaps 706: /* Protect against character escapes. */
1.32 schwarze 707:
1.8 kristaps 708: if ('\\' == last)
709: putchar('e');
1.1 schwarze 710: }
1.2 schwarze 711:
1.33 ! schwarze 712: if (FMT__MAX == fmt)
! 713: return(0);
! 714:
1.2 schwarze 715: if ( ! nomacro && FMT_CODE == fmt)
716: printf(" Qc ");
1.1 schwarze 717:
1.33 ! schwarze 718: st->wantws = ' ' == last;
! 719: return(1);
1.1 schwarze 720: }
721:
722: /*
723: * Calls formatcode() til the end of a paragraph.
1.32 schwarze 724: * Goes to OUST_MAC mode and stays there when returning,
725: * such that the caller can add arguments to the macro line
726: * before closing it out.
1.1 schwarze 727: */
728: static void
1.32 schwarze 729: formatcodeln(struct state *st, const char *linemac,
730: const char *buf, size_t *start, size_t end, int nomacro)
1.1 schwarze 731: {
1.33 ! schwarze 732: int gotmacro, wantws;
1.1 schwarze 733:
1.32 schwarze 734: assert(OUST_NL == st->oust);
735: assert(st->wantws);
736: printf(".%s ", linemac);
737: st->oust = OUST_MAC;
738:
1.33 ! schwarze 739: gotmacro = 0;
1.1 schwarze 740: while (*start < end) {
1.33 ! schwarze 741: wantws = ' ' == buf[*start] || '\n' == buf[*start];
! 742: if (wantws) {
! 743: last = ' ';
! 744: do {
! 745: (*start)++;
! 746: } while (*start < end && ' ' == buf[*start]);
! 747: }
! 748:
1.1 schwarze 749: if (*start + 1 < end && '<' == buf[*start + 1]) {
1.33 ! schwarze 750: st->wantws |= wantws;
! 751: gotmacro = formatcode(st, buf,
! 752: start, end, nomacro, 1);
1.1 schwarze 753: continue;
754: }
1.32 schwarze 755:
1.33 ! schwarze 756: if (gotmacro) {
! 757: if (*start < end || st->outbuflen) {
! 758: if (st->wantws ||
! 759: (wantws && !st->outbuflen))
! 760: printf(" No ");
! 761: else
! 762: printf(" Ns ");
! 763: }
! 764: gotmacro = 0;
! 765: }
! 766: outbuf_flush(st);
! 767: st->wantws = wantws;
! 768:
! 769: if (*start >= end)
! 770: break;
! 771:
! 772: if (st->wantws) {
! 773: putchar(' ');
! 774: st->wantws = 0;
1.32 schwarze 775: }
776:
1.4 schwarze 777: /*
778: * Since we're already on a macro line, we want to make
779: * sure that we don't inadvertently invoke a macro.
780: * We need to do this carefully because section names
781: * are used in troff and we don't want to escape
782: * something that needn't be escaped.
783: */
784: if (' ' == last && end - *start > 1 &&
1.33 ! schwarze 785: isupper((unsigned char)buf[*start]) &&
! 786: islower((unsigned char)buf[*start + 1]) &&
! 787: (end - *start == 2 || ' ' == buf[*start + 2]))
1.4 schwarze 788: printf("\\&");
789:
1.33 ! schwarze 790: putchar(last = buf[*start]);
1.8 kristaps 791:
792: /* Protect against character escapes. */
1.33 ! schwarze 793:
1.8 kristaps 794: if ('\\' == last)
795: putchar('e');
796:
1.1 schwarze 797: (*start)++;
798: }
799: }
800:
801: /*
1.4 schwarze 802: * Guess at what kind of list we are.
803: * These are taken straight from the POD manual.
804: * I don't know what people do in real life.
805: */
806: static enum list
807: listguess(const char *buf, size_t start, size_t end)
808: {
809: size_t len = end - start;
810:
811: assert(end >= start);
812:
813: if (len == 1 && '*' == buf[start])
814: return(LIST_BULLET);
815: if (len == 2 && '1' == buf[start] && '.' == buf[start + 1])
816: return(LIST_ENUM);
817: else if (len == 1 && '1' == buf[start])
818: return(LIST_ENUM);
819: else
820: return(LIST_TAG);
821: }
822:
823: /*
1.1 schwarze 824: * A command paragraph, as noted in the perlpod manual, just indicates
825: * that we should do something, optionally with some text to print as
826: * well.
1.32 schwarze 827: * From the perspective of external callers,
828: * always stays in OUST_NL/wantws mode,
829: * but its children do use OUST_MAC.
1.1 schwarze 830: */
831: static void
832: command(struct state *st, const char *buf, size_t start, size_t end)
833: {
834: size_t len, csz;
835: enum cmd cmd;
836:
837: assert('=' == buf[start]);
838: start++;
839: len = end - start;
840:
841: for (cmd = 0; cmd < CMD__MAX; cmd++) {
842: csz = strlen(cmds[cmd]);
843: if (len < csz)
844: continue;
845: if (0 == memcmp(&buf[start], cmd[cmds], csz))
846: break;
847: }
848:
849: /* Ignore bogus commands. */
850:
851: if (CMD__MAX == cmd)
852: return;
853:
854: start += csz;
1.8 kristaps 855: while (start < end && ' ' == buf[start])
856: start++;
857:
1.1 schwarze 858: len = end - start;
859:
860: if (st->paused) {
861: st->paused = CMD_END != cmd;
862: return;
863: }
864:
865: switch (cmd) {
866: case (CMD_POD):
867: break;
868: case (CMD_HEAD1):
869: /*
870: * The behaviour of head= follows from a quick glance at
871: * how pod2man handles it.
872: */
1.11 kristaps 873: st->sect = SECT_NONE;
874: if (end - start == 4) {
1.1 schwarze 875: if (0 == memcmp(&buf[start], "NAME", 4))
1.11 kristaps 876: st->sect = SECT_NAME;
877: } else if (end - start == 8) {
878: if (0 == memcmp(&buf[start], "SYNOPSIS", 8))
879: st->sect = SECT_SYNOPSIS;
880: }
1.32 schwarze 881: formatcodeln(st, "Sh", buf, &start, end, 1);
882: mdoc_newln(st);
1.1 schwarze 883: st->haspar = 1;
884: break;
885: case (CMD_HEAD2):
1.32 schwarze 886: formatcodeln(st, "Ss", buf, &start, end, 1);
887: mdoc_newln(st);
1.1 schwarze 888: st->haspar = 1;
889: break;
890: case (CMD_HEAD3):
891: puts(".Pp");
1.32 schwarze 892: formatcodeln(st, "Em", buf, &start, end, 0);
893: mdoc_newln(st);
1.1 schwarze 894: puts(".Pp");
895: st->haspar = 1;
896: break;
897: case (CMD_HEAD4):
898: puts(".Pp");
1.32 schwarze 899: formatcodeln(st, "No", buf, &start, end, 0);
900: mdoc_newln(st);
1.1 schwarze 901: puts(".Pp");
902: st->haspar = 1;
903: break;
904: case (CMD_OVER):
1.4 schwarze 905: /*
906: * If we have an existing list that hasn't had an =item
907: * yet, then make sure that we open it now.
908: * We use the default list type, but that can't be
909: * helped (we haven't seen any items yet).
1.1 schwarze 910: */
1.4 schwarze 911: if (st->lpos > 0)
912: if (LIST__MAX == st->lstack[st->lpos - 1]) {
913: st->lstack[st->lpos - 1] = LIST_TAG;
914: puts(".Bl -tag -width Ds");
915: }
916: st->lpos++;
917: assert(st->lpos < LIST_STACKSZ);
918: st->lstack[st->lpos - 1] = LIST__MAX;
1.1 schwarze 919: break;
920: case (CMD_ITEM):
1.6 kristaps 921: if (0 == st->lpos) {
922: /*
923: * Bad markup.
924: * Try to compensate.
925: */
926: st->lstack[st->lpos] = LIST__MAX;
927: st->lpos++;
928: }
1.4 schwarze 929: assert(st->lpos > 0);
930: /*
931: * If we're the first =item, guess at what our content
932: * will be: "*" is a bullet list, "1." is a numbered
933: * list, and everything is tagged.
934: */
935: if (LIST__MAX == st->lstack[st->lpos - 1]) {
936: st->lstack[st->lpos - 1] =
937: listguess(buf, start, end);
938: switch (st->lstack[st->lpos - 1]) {
939: case (LIST_BULLET):
940: puts(".Bl -bullet");
941: break;
942: case (LIST_ENUM):
943: puts(".Bl -enum");
944: break;
945: default:
946: puts(".Bl -tag -width Ds");
947: break;
948: }
949: }
950: switch (st->lstack[st->lpos - 1]) {
951: case (LIST_TAG):
1.32 schwarze 952: formatcodeln(st, "It", buf, &start, end, 0);
953: mdoc_newln(st);
1.4 schwarze 954: break;
955: case (LIST_ENUM):
956: /* FALLTHROUGH */
957: case (LIST_BULLET):
958: /*
959: * Abandon the remainder of the paragraph
960: * because we're going to be a bulletted or
961: * numbered list.
962: */
963: puts(".It");
964: break;
965: default:
966: abort();
967: }
1.1 schwarze 968: st->haspar = 1;
969: break;
970: case (CMD_BACK):
1.4 schwarze 971: /* Make sure we don't back over the stack. */
972: if (st->lpos > 0) {
973: st->lpos--;
974: puts(".El");
975: }
1.1 schwarze 976: break;
977: case (CMD_BEGIN):
978: /*
979: * We disregard all types for now.
980: * TODO: process at least "text" in a -literal block.
981: */
982: st->paused = 1;
983: break;
984: case (CMD_FOR):
985: /*
986: * We ignore all types of encodings and formats
987: * unilaterally.
988: */
989: break;
990: case (CMD_ENCODING):
991: break;
992: case (CMD_CUT):
993: st->parsing = 0;
994: return;
995: default:
996: abort();
997: }
998:
999: /* Any command (but =cut) makes us start parsing. */
1000: st->parsing = 1;
1001: }
1002:
1003: /*
1004: * Just pump out the line in a verbatim block.
1.32 schwarze 1005: * From the perspective of external callers,
1006: * always stays in OUST_NL/wantws mode.
1.1 schwarze 1007: */
1008: static void
1009: verbatim(struct state *st, const char *buf, size_t start, size_t end)
1010: {
1.22 kristaps 1011: size_t i;
1.1 schwarze 1012:
1013: if ( ! st->parsing || st->paused)
1014: return;
1.22 kristaps 1015: again:
1016: /*
1017: * If we're in the SYNOPSIS, see if we're an #include block.
1018: * If we are, then print the "In" macro and re-loop.
1019: * This handles any number of inclusions, but only when they
1020: * come before the remaining parts...
1021: */
1022: if (SECT_SYNOPSIS == st->sect) {
1023: i = start;
1024: for (i = start; i < end && ' ' == buf[i]; i++)
1025: /* Spin. */ ;
1026: if (i == end)
1027: return;
1028: /* We're an include block! */
1029: if (end - i > 10 &&
1030: 0 == memcmp(&buf[i], "#include <", 10)) {
1031: start = i + 10;
1032: while (start < end && ' ' == buf[start])
1033: start++;
1034: fputs(".In ", stdout);
1035: /* Stop til the '>' marker or we hit eoln. */
1036: while (start < end &&
1037: '>' != buf[start] && '\n' != buf[start])
1038: putchar(buf[start++]);
1039: putchar('\n');
1040: if (start < end && '>' == buf[start])
1041: start++;
1042: if (start < end && '\n' == buf[start])
1043: start++;
1044: if (start < end)
1045: goto again;
1046: return;
1047: }
1048: }
1049:
1050: if (start == end)
1051: return;
1.1 schwarze 1052: puts(".Bd -literal");
1.8 kristaps 1053: for (last = ' '; start < end; start++) {
1054: /*
1055: * Handle accidental macros (newline starting with
1056: * control character) and escapes.
1057: */
1058: if ('\n' == last)
1.7 kristaps 1059: if ('.' == buf[start] || '\'' == buf[start])
1060: printf("\\&");
1.8 kristaps 1061: putchar(last = buf[start]);
1062: if ('\\' == buf[start])
1063: printf("e");
1.7 kristaps 1064: }
1.31 schwarze 1065: putchar(last = '\n');
1.1 schwarze 1066: puts(".Ed");
1067: }
1068:
1069: /*
1.13 kristaps 1070: * See dosynopsisop().
1071: */
1072: static int
1073: hasmatch(const char *buf, size_t start, size_t end)
1074: {
1075: size_t stack;
1076:
1077: for (stack = 0; start < end; start++)
1078: if (buf[start] == '[')
1079: stack++;
1080: else if (buf[start] == ']' && 0 == stack)
1081: return(1);
1082: else if (buf[start] == ']')
1083: stack--;
1084: return(0);
1085: }
1086:
1087: /*
1088: * If we're in the SYNOPSIS section and we've encounter braces in an
1089: * ordinary paragraph, then try to see whether we're an [-option].
1090: * Do this, if we're an opening bracket, by first seeing if we have a
1091: * matching end via hasmatch().
1092: * If we're an ending bracket, see if we have a stack already.
1093: */
1094: static int
1.32 schwarze 1095: dosynopsisop(struct state *st, const char *buf,
1096: size_t *start, size_t end, size_t *opstack)
1.13 kristaps 1097: {
1098:
1099: assert('[' == buf[*start] || ']' == buf[*start]);
1100:
1101: if ('[' == buf[*start] && hasmatch(buf, *start + 1, end)) {
1.32 schwarze 1102: mdoc_newln(st);
1.13 kristaps 1103: puts(".Oo");
1104: (*opstack)++;
1105: } else if ('[' == buf[*start])
1106: return(0);
1107:
1108: if (']' == buf[*start] && *opstack > 0) {
1.32 schwarze 1109: mdoc_newln(st);
1.13 kristaps 1110: puts(".Oc");
1111: (*opstack)--;
1112: } else if (']' == buf[*start])
1113: return(0);
1114:
1115: (*start)++;
1.31 schwarze 1116: last = '\n';
1.13 kristaps 1117: while (' ' == buf[*start])
1118: (*start)++;
1119: return(1);
1120: }
1121:
1122: /*
1.17 kristaps 1123: * Format multiple "Nm" manpage names in the NAME section.
1.32 schwarze 1124: * From the perspective of external callers,
1125: * always stays in OUST_NL/wantws mode,
1126: * but its children do use OUST_MAC.
1.17 kristaps 1127: */
1128: static void
1129: donamenm(struct state *st, const char *buf, size_t *start, size_t end)
1130: {
1131: size_t word;
1132:
1.32 schwarze 1133: assert(OUST_NL == st->oust);
1134: assert(st->wantws);
1135:
1.17 kristaps 1136: while (*start < end && ' ' == buf[*start])
1137: (*start)++;
1138:
1139: if (end == *start) {
1140: puts(".Nm unknown");
1141: return;
1142: }
1143:
1144: while (*start < end) {
1145: for (word = *start; word < end; word++)
1146: if (',' == buf[word])
1147: break;
1.32 schwarze 1148: formatcodeln(st, "Nm", buf, start, word, 1);
1.17 kristaps 1149: if (*start == end) {
1.32 schwarze 1150: mdoc_newln(st);
1151: break;
1.17 kristaps 1152: }
1153: assert(',' == buf[*start]);
1.32 schwarze 1154: printf(" ,");
1155: mdoc_newln(st);
1.17 kristaps 1156: (*start)++;
1157: while (*start < end && ' ' == buf[*start])
1158: (*start)++;
1159: }
1160: }
1161:
1162: /*
1.1 schwarze 1163: * Ordinary paragraph.
1164: * Well, this is really the hardest--POD seems to assume that, for
1165: * example, a leading space implies a newline, and so on.
1166: * Lots of other snakes in the grass: escaping a newline followed by a
1167: * period (accidental mdoc(7) control), double-newlines after macro
1168: * passages, etc.
1.32 schwarze 1169: *
1170: * Uses formatcode() to go to OUST_MAC mode
1171: * and outbuf_flush() to go to OUST_TXT mode.
1172: * Main text mode wantws handling is in this function.
1173: * Must make sure to go back to OUST_NL/wantws mode before returning.
1.1 schwarze 1174: */
1175: static void
1176: ordinary(struct state *st, const char *buf, size_t start, size_t end)
1177: {
1.13 kristaps 1178: size_t i, j, opstack;
1.15 kristaps 1179: int seq;
1.1 schwarze 1180:
1181: if ( ! st->parsing || st->paused)
1182: return;
1183:
1184: /*
1185: * Special-case: the NAME section.
1186: * If we find a "-" when searching from the end, assume that
1187: * we're in "name - description" format.
1188: * To wit, print out a "Nm" and "Nd" in that format.
1189: */
1.11 kristaps 1190: if (SECT_NAME == st->sect) {
1.15 kristaps 1191: for (i = end - 2; i > start; i--)
1192: if ('-' == buf[i] && ' ' == buf[i + 1])
1.1 schwarze 1193: break;
1194: if ('-' == buf[i]) {
1195: j = i;
1196: /* Roll over multiple "-". */
1197: for ( ; i > start; i--)
1198: if ('-' != buf[i])
1199: break;
1.17 kristaps 1200: donamenm(st, buf, &start, i + 1);
1.5 kristaps 1201: start = j + 1;
1.17 kristaps 1202: while (start < end && ' ' == buf[start])
1203: start++;
1.32 schwarze 1204: formatcodeln(st, "Nd", buf, &start, end, 1);
1205: mdoc_newln(st);
1.1 schwarze 1206: return;
1207: }
1208: }
1209:
1210: if ( ! st->haspar)
1211: puts(".Pp");
1212:
1213: st->haspar = 0;
1214: last = '\n';
1.13 kristaps 1215: opstack = 0;
1.1 schwarze 1216:
1.15 kristaps 1217: for (seq = 0; start < end; seq++) {
1.1 schwarze 1218: /*
1219: * Loop til we get either to a newline or escape.
1220: * Escape initial control characters.
1221: */
1222: while (start < end) {
1223: if (start < end - 1 && '<' == buf[start + 1])
1224: break;
1225: else if ('\n' == buf[start])
1226: break;
1227: else if ('\n' == last && '.' == buf[start])
1.31 schwarze 1228: outbuf_addstr(st, "\\&");
1.1 schwarze 1229: else if ('\n' == last && '\'' == buf[start])
1.31 schwarze 1230: outbuf_addstr(st, "\\&");
1.12 kristaps 1231: /*
1232: * If we're in the SYNOPSIS, have square
1233: * brackets indicate that we're opening and
1234: * closing an optional context.
1235: */
1.32 schwarze 1236:
1.13 kristaps 1237: if (SECT_SYNOPSIS == st->sect &&
1238: ('[' == buf[start] ||
1239: ']' == buf[start]) &&
1.32 schwarze 1240: dosynopsisop(st, buf,
1241: &start, end, &opstack))
1.13 kristaps 1242: continue;
1.32 schwarze 1243:
1244: /*
1245: * On whitespace, flush the output buffer
1246: * and allow breaking to a macro line.
1247: * Otherwise, buffer text and clear wantws.
1248: */
1249:
1.31 schwarze 1250: last = buf[start++];
1251: if (' ' == last) {
1252: outbuf_flush(st);
1253: putchar(' ');
1.32 schwarze 1254: st->wantws = 1;
1.31 schwarze 1255: } else
1256: outbuf_addchar(st);
1.1 schwarze 1257: }
1258:
1259: if (start < end - 1 && '<' == buf[start + 1]) {
1.32 schwarze 1260: formatcode(st, buf, &start, end, 0, seq);
1261: if (OUST_MAC == st->oust) {
1.30 schwarze 1262: /*
1263: * Let mdoc(7) handle trailing punctuation.
1264: * XXX Some punctuation characters
1265: * are not handled yet.
1266: */
1.16 kristaps 1267: if ((start == end - 1 ||
1268: (start < end - 1 &&
1269: (' ' == buf[start + 1] ||
1270: '\n' == buf[start + 1]))) &&
1271: ('.' == buf[start] ||
1272: ',' == buf[start])) {
1273: putchar(' ');
1274: putchar(buf[start++]);
1275: }
1.32 schwarze 1276:
1277: if (st->wantws ||
1278: ' ' == buf[start] ||
1279: '\n' == buf[start])
1280: mdoc_newln(st);
1281:
1.30 schwarze 1282: /*
1283: * Consume all whitespace
1284: * so we don't accidentally start
1285: * an implicit literal line.
1286: */
1.32 schwarze 1287:
1.6 kristaps 1288: while (start < end && ' ' == buf[start])
1289: start++;
1.32 schwarze 1290:
1291: /*
1292: * Some text is following.
1293: * Implement requested spacing.
1294: */
1295:
1296: if ( ! st->wantws && start < end &&
1297: '<' != buf[start + 1]) {
1298: printf(" Ns ");
1299: st->wantws = 1;
1300: }
1.6 kristaps 1301: }
1.1 schwarze 1302: } else if (start < end && '\n' == buf[start]) {
1.32 schwarze 1303: outbuf_flush(st);
1304: mdoc_newln(st);
1.1 schwarze 1305: if (++start >= end)
1306: continue;
1307: /*
1308: * If we have whitespace next, eat it to prevent
1309: * mdoc(7) from thinking that it's meant for
1310: * verbatim text.
1311: * It is--but if we start with that, we can't
1312: * have a macro subsequent it, which may be
1313: * possible if we have an escape next.
1314: */
1.31 schwarze 1315: if (' ' == buf[start] || '\t' == buf[start])
1.1 schwarze 1316: puts(".br");
1317: for ( ; start < end; start++)
1318: if (' ' != buf[start] && '\t' != buf[start])
1319: break;
1.12 kristaps 1320: }
1.1 schwarze 1321: }
1.32 schwarze 1322: outbuf_flush(st);
1323: mdoc_newln(st);
1.1 schwarze 1324: }
1325:
1326: /*
1327: * There are three kinds of paragraphs: verbatim (starts with whitespace
1328: * of some sort), ordinary (starts without "=" marker), or a command
1329: * (default: starts with "=").
1330: */
1331: static void
1332: dopar(struct state *st, const char *buf, size_t start, size_t end)
1333: {
1334:
1.32 schwarze 1335: assert(OUST_NL == st->oust);
1336: assert(st->wantws);
1337:
1.1 schwarze 1338: if (end == start)
1339: return;
1340: if (' ' == buf[start] || '\t' == buf[start])
1341: verbatim(st, buf, start, end);
1342: else if ('=' != buf[start])
1343: ordinary(st, buf, start, end);
1344: else
1345: command(st, buf, start, end);
1346: }
1347:
1348: /*
1349: * Loop around paragraphs within a document, processing each one in the
1350: * POD way.
1351: */
1352: static void
1353: dofile(const struct args *args, const char *fname,
1354: const struct tm *tm, const char *buf, size_t sz)
1355: {
1.29 schwarze 1356: char datebuf[64];
1.1 schwarze 1357: struct state st;
1.29 schwarze 1358: const char *fbase, *fext, *section, *date;
1.1 schwarze 1359: char *title, *cp;
1.29 schwarze 1360: size_t sup, end, i, cur = 0;
1.1 schwarze 1361:
1362: if (0 == sz)
1363: return;
1364:
1.29 schwarze 1365: /*
1366: * Parsing the filename is almost always required,
1367: * except when both the title and the section
1368: * are provided on the command line.
1369: */
1370:
1371: if (NULL == args->title || NULL == args->section) {
1372: fbase = strrchr(fname, '/');
1373: if (NULL == fbase)
1374: fbase = fname;
1375: else
1376: fbase++;
1377: fext = strrchr(fbase, '.');
1378: } else
1379: fext = NULL;
1380:
1381: /*
1382: * The title will be converted to uppercase,
1383: * so it needs to be copied.
1384: */
1385:
1386: title = (NULL != args->title) ? strdup(args->title) :
1387: (NULL != fext) ? strndup(fbase, fext - fbase) :
1388: strdup(fbase);
1.1 schwarze 1389:
1390: if (NULL == title) {
1391: perror(NULL);
1392: exit(EXIT_FAILURE);
1393: }
1394:
1395: /* Section is 1 unless suffix is "pm". */
1396:
1.29 schwarze 1397: section = (NULL != args->section) ? args->section :
1398: (NULL == fext || strcmp(fext + 1, "pm")) ? "1" :
1399: PERL_SECTION;
1.1 schwarze 1400:
1401: /* Date. Or the given "tm" if not supplied. */
1402:
1403: if (NULL == (date = args->date)) {
1404: strftime(datebuf, sizeof(datebuf), "%B %d, %Y", tm);
1405: date = datebuf;
1406: }
1407:
1408: for (cp = title; '\0' != *cp; cp++)
1409: *cp = toupper((int)*cp);
1410:
1411: /* The usual mdoc(7) preamble. */
1412:
1413: printf(".Dd %s\n", date);
1414: printf(".Dt %s %s\n", title, section);
1415: puts(".Os");
1416:
1417: free(title);
1418:
1419: memset(&st, 0, sizeof(struct state));
1.32 schwarze 1420: st.oust = OUST_NL;
1421: st.wantws = 1;
1422:
1.1 schwarze 1423: assert(sz > 0);
1424:
1425: /* Main loop over file contents. */
1426:
1427: while (cur < sz) {
1428: /* Read until next paragraph. */
1429: for (i = cur + 1; i < sz; i++)
1430: if ('\n' == buf[i] && '\n' == buf[i - 1]) {
1431: /* Consume blank paragraphs. */
1432: while (i + 1 < sz && '\n' == buf[i + 1])
1433: i++;
1434: break;
1435: }
1436:
1437: /* Adjust end marker for EOF. */
1438: end = i < sz ? i - 1 :
1439: ('\n' == buf[sz - 1] ? sz - 1 : sz);
1440: sup = i < sz ? end + 2 : sz;
1441:
1442: /* Process paragraph and adjust start. */
1443: dopar(&st, buf, cur, end);
1444: cur = sup;
1445: }
1446: }
1447:
1448: /*
1449: * Read a single file fully into memory.
1450: * If the file is "-", do it from stdin.
1451: * If successfully read, send the input buffer to dofile() for further
1452: * processing.
1453: */
1454: static int
1455: readfile(const struct args *args, const char *fname)
1456: {
1457: int fd;
1458: char *buf;
1459: size_t bufsz, cur;
1460: ssize_t ssz;
1461: struct tm *tm;
1462: time_t ttm;
1463: struct stat st;
1464:
1465: fd = 0 != strcmp("-", fname) ?
1466: open(fname, O_RDONLY, 0) : STDIN_FILENO;
1467:
1468: if (-1 == fd) {
1469: perror(fname);
1470: return(0);
1471: }
1472:
1473: if (STDIN_FILENO == fd || -1 == fstat(fd, &st)) {
1474: ttm = time(NULL);
1475: tm = localtime(&ttm);
1476: } else
1477: tm = localtime(&st.st_mtime);
1478:
1479: /*
1480: * Arbitrarily-sized initial buffer.
1481: * Should be big enough for most files...
1482: */
1483: cur = 0;
1484: bufsz = 1 << 14;
1485: if (NULL == (buf = malloc(bufsz))) {
1486: perror(NULL);
1487: exit(EXIT_FAILURE);
1488: }
1489:
1490: while ((ssz = read(fd, buf + cur, bufsz - cur)) > 0) {
1491: /* Double buffer size on fill. */
1492: if ((size_t)ssz == bufsz - cur) {
1493: bufsz *= 2;
1494: if (NULL == (buf = realloc(buf, bufsz))) {
1495: perror(NULL);
1496: exit(EXIT_FAILURE);
1497: }
1498: }
1499: cur += (size_t)ssz;
1500: }
1501: if (ssz < 0) {
1502: perror(fname);
1503: free(buf);
1504: return(0);
1505: }
1506:
1507: dofile(args, STDIN_FILENO == fd ?
1508: "STDIN" : fname, tm, buf, cur);
1509: free(buf);
1510: if (STDIN_FILENO != fd)
1511: close(fd);
1512: return(1);
1513: }
1514:
1515: int
1516: main(int argc, char *argv[])
1517: {
1518: const char *fname, *name;
1519: struct args args;
1520: int c;
1521:
1522: name = strrchr(argv[0], '/');
1523: if (name == NULL)
1524: name = argv[0];
1525: else
1526: ++name;
1527:
1528: memset(&args, 0, sizeof(struct args));
1529: fname = "-";
1530:
1531: /* Accept no arguments for now. */
1532:
1533: while (-1 != (c = getopt(argc, argv, "c:d:hln:oq:rs:uv")))
1534: switch (c) {
1535: case ('h'):
1536: /* FALLTHROUGH */
1537: case ('l'):
1538: /* FALLTHROUGH */
1539: case ('c'):
1540: /* FALLTHROUGH */
1541: case ('o'):
1542: /* FALLTHROUGH */
1543: case ('q'):
1544: /* FALLTHROUGH */
1545: case ('r'):
1546: /* FALLTHROUGH */
1547: case ('u'):
1548: /* FALLTHROUGH */
1549: case ('v'):
1550: /* Ignore these. */
1551: break;
1552: case ('d'):
1553: args.date = optarg;
1554: break;
1555: case ('n'):
1556: args.title = optarg;
1557: break;
1558: case ('s'):
1559: args.section = optarg;
1560: break;
1561: default:
1562: goto usage;
1563: }
1564:
1565: argc -= optind;
1566: argv += optind;
1567:
1568: /* Accept only a single input file. */
1569:
1.25 schwarze 1570: if (argc > 1)
1571: goto usage;
1.1 schwarze 1572: else if (1 == argc)
1573: fname = *argv;
1574:
1575: return(readfile(&args, fname) ?
1576: EXIT_SUCCESS : EXIT_FAILURE);
1577:
1578: usage:
1579: fprintf(stderr, "usage: %s [-d date] "
1.25 schwarze 1580: "[-n title] [-s section] [file]\n", name);
1.1 schwarze 1581:
1582: return(EXIT_FAILURE);
1583: }
CVSweb
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to pod2mdoc.c CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [cvsweb.bsd.lv] / pod2mdoc