Annotation of src/usr.bin/mandoc/main.c, Revision 1.53
1.53 ! schwarze 1: /* $Id: main.c,v 1.52 2010/10/26 22:28:56 schwarze Exp $ */
1.1 kristaps 2: /*
1.42 schwarze 3: * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.2 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.27 schwarze 18: #include <sys/types.h>
19: #include <sys/mman.h>
1.1 kristaps 20: #include <sys/stat.h>
21:
22: #include <assert.h>
1.43 schwarze 23: #include <ctype.h>
1.1 kristaps 24: #include <fcntl.h>
25: #include <stdio.h>
1.17 schwarze 26: #include <stdint.h>
1.1 kristaps 27: #include <stdlib.h>
28: #include <string.h>
29: #include <unistd.h>
30:
1.30 schwarze 31: #include "mandoc.h"
1.38 schwarze 32: #include "main.h"
1.1 kristaps 33: #include "mdoc.h"
34: #include "man.h"
1.30 schwarze 35: #include "roff.h"
1.17 schwarze 36:
37: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
1.1 kristaps 38:
1.16 schwarze 39: typedef void (*out_mdoc)(void *, const struct mdoc *);
40: typedef void (*out_man)(void *, const struct man *);
1.1 kristaps 41: typedef void (*out_free)(void *);
42:
43: struct buf {
44: char *buf;
45: size_t sz;
46: };
47:
48: enum intt {
49: INTT_AUTO,
50: INTT_MDOC,
51: INTT_MAN
52: };
53:
54: enum outt {
55: OUTT_ASCII = 0,
56: OUTT_TREE,
1.17 schwarze 57: OUTT_HTML,
1.21 schwarze 58: OUTT_XHTML,
1.36 schwarze 59: OUTT_LINT,
1.43 schwarze 60: OUTT_PS,
61: OUTT_PDF
1.1 kristaps 62: };
63:
64: struct curparse {
65: const char *file; /* Current parse. */
66: int fd; /* Current parse. */
1.45 schwarze 67: enum mandoclevel wlevel; /* Ignore messages below this. */
68: int wstop; /* Stop after a file with a warning. */
1.33 schwarze 69: enum intt inttype; /* which parser to use */
70: struct man *man; /* man parser */
71: struct mdoc *mdoc; /* mdoc parser */
72: struct roff *roff; /* roff parser (!NULL) */
1.38 schwarze 73: struct regset regs; /* roff registers */
1.33 schwarze 74: enum outt outtype; /* which output to use */
75: out_mdoc outmdoc; /* mdoc output ptr */
76: out_man outman; /* man output ptr */
77: out_free outfree; /* free output ptr */
78: void *outdata; /* data for output */
79: char outopts[BUFSIZ]; /* buf of output opts */
80: };
81:
1.45 schwarze 82: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
83: "SUCCESS",
84: "RESERVED",
85: "WARNING",
86: "ERROR",
87: "FATAL",
88: "BADARG",
89: "SYSERR"
90: };
91:
92: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
93: MANDOCERR_OK,
94: MANDOCERR_WARNING,
95: MANDOCERR_WARNING,
96: MANDOCERR_ERROR,
97: MANDOCERR_FATAL,
98: MANDOCERR_MAX,
99: MANDOCERR_MAX
100: };
101:
1.33 schwarze 102: static const char * const mandocerrs[MANDOCERR_MAX] = {
103: "ok",
1.40 schwarze 104:
105: "generic warning",
106:
1.53 ! schwarze 107: /* related to the prologue */
! 108: "no title in document",
! 109: "document title should be all caps",
! 110: "unknown manual section",
! 111: "cannot parse date argument",
! 112: "prologue macros out of order",
! 113: "duplicate prologue macro",
! 114: "macro not allowed in prologue",
! 115: "macro not allowed in body",
! 116:
! 117: /* related to document structure */
! 118: "NAME section must come first",
! 119: "bad NAME section contents",
! 120: "manual name not yet set",
1.34 schwarze 121: "sections out of conventional order",
1.53 ! schwarze 122: "duplicate section name",
! 123: "section not in conventional manual section",
! 124:
! 125: /* related to macros and nesting */
! 126: "skipping obsolete macro",
! 127: "skipping paragraph macro",
! 128: "blocks badly nested",
! 129: "child violates parent syntax",
! 130: "nested displays are not portable",
! 131: "already in literal mode",
! 132:
! 133: /* related to missing macro arguments */
! 134: "skipping empty macro",
! 135: "missing display type",
1.33 schwarze 136: "list type must come first",
1.53 ! schwarze 137: "tag lists require a width argument",
! 138: "missing font type",
! 139:
! 140: /* related to bad macro arguments */
! 141: "skipping argument",
! 142: "duplicate argument",
! 143: "duplicate display type",
! 144: "duplicate list type",
! 145: "unknown AT&T UNIX version",
! 146: "bad Boolean value",
! 147: "unknown library specifier",
! 148: "unknown standard specifier",
! 149: "bad width argument",
! 150:
! 151: /* related to plain text */
! 152: "blank line in non-literal context",
1.43 schwarze 153: "tab in non-literal context",
1.53 ! schwarze 154: "end of line whitespace",
! 155: "bad comment style",
! 156: "unknown escape sequence",
1.33 schwarze 157: "unterminated quoted string",
1.40 schwarze 158:
159: "generic error",
160:
1.53 ! schwarze 161: "skipping bad character",
! 162: "skipping text before the first section header",
! 163: "skipping unknown macro",
1.33 schwarze 164: "line scope broken",
165: "argument count wrong",
1.53 ! schwarze 166: "skipping end of block that is not open",
1.50 schwarze 167: "missing end of block",
1.47 schwarze 168: "scope open on exit",
1.50 schwarze 169: "uname(3) system call failed",
1.33 schwarze 170: "macro requires line argument(s)",
171: "macro requires body argument(s)",
172: "macro requires argument(s)",
1.34 schwarze 173: "missing list type",
1.33 schwarze 174: "line argument(s) will be lost",
175: "body argument(s) will be lost",
1.49 schwarze 176: "tbl(1) error",
1.40 schwarze 177:
178: "generic fatal error",
179:
1.34 schwarze 180: "column syntax is inconsistent",
1.37 schwarze 181: "unsupported display type",
1.33 schwarze 182: "line scope broken, syntax violated",
183: "argument count wrong, violates syntax",
184: "child violates parent syntax",
185: "argument count wrong, violates syntax",
1.52 schwarze 186: "invalid path in include directive",
1.33 schwarze 187: "no document body",
188: "no document prologue",
1.45 schwarze 189: "static buffer exhausted",
1.1 kristaps 190: };
191:
1.51 schwarze 192: static void pdesc(struct curparse *);
1.27 schwarze 193: static void fdesc(struct curparse *);
194: static void ffile(const char *, struct curparse *);
1.52 schwarze 195: static int pfile(const char *, struct curparse *, int);
1.1 kristaps 196: static int moptions(enum intt *, char *);
1.30 schwarze 197: static int mmsg(enum mandocerr, void *,
198: int, int, const char *);
1.45 schwarze 199: static void pset(const char *, int, struct curparse *,
1.1 kristaps 200: struct man **, struct mdoc **);
1.27 schwarze 201: static int toptions(struct curparse *, char *);
202: static void usage(void) __attribute__((noreturn));
1.20 schwarze 203: static void version(void) __attribute__((noreturn));
1.45 schwarze 204: static int woptions(struct curparse *, char *);
1.1 kristaps 205:
1.19 schwarze 206: static const char *progname;
1.45 schwarze 207: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 208:
209: int
210: main(int argc, char *argv[])
211: {
1.27 schwarze 212: int c;
1.1 kristaps 213: struct curparse curp;
214:
1.19 schwarze 215: progname = strrchr(argv[0], '/');
216: if (progname == NULL)
217: progname = argv[0];
218: else
219: ++progname;
220:
221: memset(&curp, 0, sizeof(struct curparse));
1.1 kristaps 222:
223: curp.inttype = INTT_AUTO;
224: curp.outtype = OUTT_ASCII;
1.45 schwarze 225: curp.wlevel = MANDOCLEVEL_FATAL;
1.1 kristaps 226:
227: /* LINTED */
1.45 schwarze 228: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 229: switch (c) {
230: case ('m'):
231: if ( ! moptions(&curp.inttype, optarg))
1.47 schwarze 232: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 233: break;
1.18 schwarze 234: case ('O'):
235: (void)strlcat(curp.outopts, optarg, BUFSIZ);
236: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.17 schwarze 237: break;
1.1 kristaps 238: case ('T'):
1.22 schwarze 239: if ( ! toptions(&curp, optarg))
1.47 schwarze 240: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 241: break;
242: case ('W'):
1.45 schwarze 243: if ( ! woptions(&curp, optarg))
1.47 schwarze 244: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 245: break;
1.3 schwarze 246: case ('V'):
247: version();
248: /* NOTREACHED */
1.1 kristaps 249: default:
250: usage();
251: /* NOTREACHED */
252: }
253:
254: argc -= optind;
255: argv += optind;
256:
1.7 schwarze 257: if (NULL == *argv) {
258: curp.file = "<stdin>";
259: curp.fd = STDIN_FILENO;
1.14 schwarze 260:
1.27 schwarze 261: fdesc(&curp);
262: }
263:
264: while (*argv) {
265: ffile(*argv, &curp);
1.45 schwarze 266: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.27 schwarze 267: break;
268: ++argv;
1.1 kristaps 269: }
270:
271: if (curp.outfree)
272: (*curp.outfree)(curp.outdata);
1.30 schwarze 273: if (curp.mdoc)
274: mdoc_free(curp.mdoc);
275: if (curp.man)
276: man_free(curp.man);
277: if (curp.roff)
278: roff_free(curp.roff);
1.1 kristaps 279:
1.47 schwarze 280: return((int)exit_status);
1.1 kristaps 281: }
282:
283:
1.20 schwarze 284: static void
1.3 schwarze 285: version(void)
286: {
287:
1.19 schwarze 288: (void)printf("%s %s\n", progname, VERSION);
1.47 schwarze 289: exit((int)MANDOCLEVEL_OK);
1.3 schwarze 290: }
291:
292:
1.20 schwarze 293: static void
1.1 kristaps 294: usage(void)
295: {
296:
1.23 jmc 297: (void)fprintf(stderr, "usage: %s [-V] [-foption] "
1.18 schwarze 298: "[-mformat] [-Ooption] [-Toutput] "
1.23 jmc 299: "[-Werr] [file...]\n", progname);
1.47 schwarze 300: exit((int)MANDOCLEVEL_BADARG);
1.1 kristaps 301: }
302:
303:
1.27 schwarze 304: static void
305: ffile(const char *file, struct curparse *curp)
1.1 kristaps 306: {
307:
308: curp->file = file;
309: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.19 schwarze 310: perror(curp->file);
1.45 schwarze 311: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 312: return;
1.1 kristaps 313: }
314:
1.27 schwarze 315: fdesc(curp);
1.1 kristaps 316:
317: if (-1 == close(curp->fd))
1.19 schwarze 318: perror(curp->file);
1.27 schwarze 319: }
1.1 kristaps 320:
1.52 schwarze 321: static int
322: pfile(const char *file, struct curparse *curp, int ln)
323: {
324: const char *savefile;
325: int fd, savefd;
326:
327: if (-1 == (fd = open(file, O_RDONLY, 0))) {
328: perror(file);
329: exit_status = MANDOCLEVEL_SYSERR;
330: return(0);
331: }
332:
333: savefile = curp->file;
334: savefd = curp->fd;
335:
336: curp->file = file;
337: curp->fd = fd;
338:
339: pdesc(curp);
340:
341: curp->file = savefile;
342: curp->fd = savefd;
343:
344: if (-1 == close(fd))
345: perror(file);
346:
347: return(MANDOCLEVEL_FATAL > exit_status ? 1 : 0);
348: }
349:
1.27 schwarze 350:
1.45 schwarze 351: static void
1.27 schwarze 352: resize_buf(struct buf *buf, size_t initial)
353: {
354:
1.45 schwarze 355: buf->sz = buf->sz ? 2 * buf->sz : initial;
356: buf->buf = realloc(buf->buf, buf->sz);
357: if (NULL == buf->buf) {
1.27 schwarze 358: perror(NULL);
1.47 schwarze 359: exit((int)MANDOCLEVEL_SYSERR);
1.27 schwarze 360: }
1.1 kristaps 361: }
362:
363:
364: static int
1.27 schwarze 365: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
1.1 kristaps 366: {
1.27 schwarze 367: struct stat st;
368: size_t off;
1.1 kristaps 369: ssize_t ssz;
1.27 schwarze 370:
371: if (-1 == fstat(curp->fd, &st)) {
372: perror(curp->file);
373: return(0);
374: }
375:
376: /*
377: * If we're a regular file, try just reading in the whole entry
378: * via mmap(). This is faster than reading it into blocks, and
379: * since each file is only a few bytes to begin with, I'm not
380: * concerned that this is going to tank any machines.
381: */
382:
383: if (S_ISREG(st.st_mode)) {
384: if (st.st_size >= (1U << 31)) {
385: fprintf(stderr, "%s: input too large\n",
386: curp->file);
387: return(0);
388: }
389: *with_mmap = 1;
1.30 schwarze 390: fb->sz = (size_t)st.st_size;
1.27 schwarze 391: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.35 schwarze 392: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.27 schwarze 393: if (fb->buf != MAP_FAILED)
394: return(1);
395: }
396:
397: /*
398: * If this isn't a regular file (like, say, stdin), then we must
399: * go the old way and just read things in bit by bit.
400: */
401:
402: *with_mmap = 0;
403: off = 0;
404: fb->sz = 0;
405: fb->buf = NULL;
406: for (;;) {
407: if (off == fb->sz) {
408: if (fb->sz == (1U << 31)) {
409: fprintf(stderr, "%s: input too large\n",
410: curp->file);
411: break;
412: }
1.45 schwarze 413: resize_buf(fb, 65536);
1.27 schwarze 414: }
1.30 schwarze 415: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.27 schwarze 416: if (ssz == 0) {
417: fb->sz = off;
418: return(1);
419: }
420: if (ssz == -1) {
421: perror(curp->file);
422: break;
423: }
1.30 schwarze 424: off += (size_t)ssz;
1.27 schwarze 425: }
426:
427: free(fb->buf);
428: fb->buf = NULL;
429: return(0);
430: }
431:
432:
433: static void
434: fdesc(struct curparse *curp)
435: {
1.51 schwarze 436: struct man *man;
437: struct mdoc *mdoc;
438: struct roff *roff;
439:
440: pdesc(curp);
441:
442: man = curp->man;
443: mdoc = curp->mdoc;
444: roff = curp->roff;
445:
446: if (MANDOCLEVEL_FATAL <= exit_status)
447: goto cleanup;
448:
449: /* NOTE a parser may not have been assigned, yet. */
450:
451: if ( ! (man || mdoc)) {
452: fprintf(stderr, "%s: Not a manual\n", curp->file);
453: exit_status = MANDOCLEVEL_FATAL;
454: goto cleanup;
455: }
456:
457: /* Clean up the parse routine ASTs. */
458:
459: if (mdoc && ! mdoc_endparse(mdoc)) {
460: assert(MANDOCLEVEL_FATAL <= exit_status);
461: goto cleanup;
462: }
463: if (man && ! man_endparse(man)) {
464: assert(MANDOCLEVEL_FATAL <= exit_status);
465: goto cleanup;
466: }
467: if (roff && ! roff_endparse(roff)) {
468: assert(MANDOCLEVEL_FATAL <= exit_status);
469: goto cleanup;
470: }
471:
472: /*
473: * With -Wstop and warnings or errors of at least
474: * the requested level, do not produce output.
475: */
476:
477: if (MANDOCLEVEL_OK != exit_status && curp->wstop)
478: goto cleanup;
479:
480: /* If unset, allocate output dev now (if applicable). */
481:
482: if ( ! (curp->outman && curp->outmdoc)) {
483: switch (curp->outtype) {
484: case (OUTT_XHTML):
485: curp->outdata = xhtml_alloc(curp->outopts);
486: break;
487: case (OUTT_HTML):
488: curp->outdata = html_alloc(curp->outopts);
489: break;
490: case (OUTT_ASCII):
491: curp->outdata = ascii_alloc(curp->outopts);
492: curp->outfree = ascii_free;
493: break;
494: case (OUTT_PDF):
495: curp->outdata = pdf_alloc(curp->outopts);
496: curp->outfree = pspdf_free;
497: break;
498: case (OUTT_PS):
499: curp->outdata = ps_alloc(curp->outopts);
500: curp->outfree = pspdf_free;
501: break;
502: default:
503: break;
504: }
505:
506: switch (curp->outtype) {
507: case (OUTT_HTML):
508: /* FALLTHROUGH */
509: case (OUTT_XHTML):
510: curp->outman = html_man;
511: curp->outmdoc = html_mdoc;
512: curp->outfree = html_free;
513: break;
514: case (OUTT_TREE):
515: curp->outman = tree_man;
516: curp->outmdoc = tree_mdoc;
517: break;
518: case (OUTT_PDF):
519: /* FALLTHROUGH */
520: case (OUTT_ASCII):
521: /* FALLTHROUGH */
522: case (OUTT_PS):
523: curp->outman = terminal_man;
524: curp->outmdoc = terminal_mdoc;
525: break;
526: default:
527: break;
528: }
529: }
530:
531: /* Execute the out device, if it exists. */
532:
533: if (man && curp->outman)
534: (*curp->outman)(curp->outdata, man);
535: if (mdoc && curp->outmdoc)
536: (*curp->outmdoc)(curp->outdata, mdoc);
537:
538: cleanup:
539: memset(&curp->regs, 0, sizeof(struct regset));
540: if (mdoc)
541: mdoc_reset(mdoc);
542: if (man)
543: man_reset(man);
544: if (roff)
545: roff_reset(roff);
546:
547: return;
548: }
549:
550:
551: static void
552: pdesc(struct curparse *curp)
553: {
1.27 schwarze 554: struct buf ln, blk;
1.32 schwarze 555: int i, pos, lnn, lnn_start, with_mmap, of;
1.30 schwarze 556: enum rofferr re;
1.44 schwarze 557: unsigned char c;
1.1 kristaps 558: struct man *man;
559: struct mdoc *mdoc;
1.30 schwarze 560: struct roff *roff;
1.1 kristaps 561:
1.27 schwarze 562: memset(&ln, 0, sizeof(struct buf));
1.1 kristaps 563:
564: /*
1.29 schwarze 565: * Two buffers: ln and buf. buf is the input file and may be
566: * memory mapped. ln is a line buffer and grows on-demand.
1.1 kristaps 567: */
568:
1.45 schwarze 569: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
570: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 571: return;
1.45 schwarze 572: }
1.27 schwarze 573:
1.30 schwarze 574: if (NULL == curp->roff)
1.45 schwarze 575: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
576: assert(curp->roff);
577: roff = curp->roff;
1.51 schwarze 578: mdoc = curp->mdoc;
579: man = curp->man;
1.30 schwarze 580:
1.29 schwarze 581: for (i = 0, lnn = 1; i < (int)blk.sz;) {
582: pos = 0;
583: lnn_start = lnn;
584: while (i < (int)blk.sz) {
585: if ('\n' == blk.buf[i]) {
586: ++i;
587: ++lnn;
588: break;
589: }
1.43 schwarze 590:
591: /*
592: * Warn about bogus characters. If you're using
593: * non-ASCII encoding, you're screwing your
594: * readers. Since I'd rather this not happen,
595: * I'll be helpful and drop these characters so
596: * we don't display gibberish. Note to manual
597: * writers: use special characters.
598: */
599:
1.44 schwarze 600: c = (unsigned char) blk.buf[i];
601: if ( ! (isascii(c) && (isgraph(c) || isblank(c)))) {
1.45 schwarze 602: mmsg(MANDOCERR_BADCHAR, curp,
603: lnn_start, pos, "ignoring byte");
1.43 schwarze 604: i++;
605: continue;
606: }
607:
1.29 schwarze 608: /* Trailing backslash is like a plain character. */
609: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
610: if (pos >= (int)ln.sz)
1.45 schwarze 611: resize_buf(&ln, 256);
1.29 schwarze 612: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 613: continue;
1.29 schwarze 614: }
615: /* Found an escape and at least one other character. */
616: if ('\n' == blk.buf[i + 1]) {
617: /* Escaped newlines are skipped over */
618: i += 2;
619: ++lnn;
1.27 schwarze 620: continue;
621: }
1.29 schwarze 622: if ('"' == blk.buf[i + 1]) {
623: i += 2;
624: /* Comment, skip to end of line */
625: for (; i < (int)blk.sz; ++i) {
626: if ('\n' == blk.buf[i]) {
627: ++i;
628: ++lnn;
629: break;
630: }
631: }
632: /* Backout trailing whitespaces */
633: for (; pos > 0; --pos) {
634: if (ln.buf[pos - 1] != ' ')
635: break;
636: if (pos > 2 && ln.buf[pos - 2] == '\\')
637: break;
638: }
639: break;
640: }
641: /* Some other escape sequence, copy and continue. */
642: if (pos + 1 >= (int)ln.sz)
1.45 schwarze 643: resize_buf(&ln, 256);
1.1 kristaps 644:
1.29 schwarze 645: ln.buf[pos++] = blk.buf[i++];
646: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 647: }
1.1 kristaps 648:
1.29 schwarze 649: if (pos >= (int)ln.sz)
1.45 schwarze 650: resize_buf(&ln, 256);
1.30 schwarze 651: ln.buf[pos] = '\0';
652:
1.32 schwarze 653: /*
654: * A significant amount of complexity is contained by
655: * the roff preprocessor. It's line-oriented but can be
656: * expressed on one line, so we need at times to
657: * readjust our starting point and re-run it. The roff
658: * preprocessor can also readjust the buffers with new
659: * data, so we pass them in wholesale.
660: */
661:
662: of = 0;
663: do {
664: re = roff_parseln(roff, lnn_start,
665: &ln.buf, &ln.sz, of, &of);
666: } while (ROFF_RERUN == re);
667:
1.45 schwarze 668: if (ROFF_IGN == re) {
1.30 schwarze 669: continue;
1.45 schwarze 670: } else if (ROFF_ERR == re) {
671: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 672: break;
1.52 schwarze 673: } else if (ROFF_SO == re) {
674: if (pfile(ln.buf + of, curp, lnn_start))
675: continue;
676: else
677: break;
1.45 schwarze 678: }
1.5 schwarze 679:
1.32 schwarze 680: /*
681: * If input parsers have not been allocated, do so now.
682: * We keep these instanced betwen parsers, but set them
683: * locally per parse routine since we can use different
684: * parsers with each one.
685: */
1.1 kristaps 686:
1.32 schwarze 687: if ( ! (man || mdoc))
1.45 schwarze 688: pset(ln.buf + of, pos - of, curp, &man, &mdoc);
1.5 schwarze 689:
1.32 schwarze 690: /* Lastly, push down into the parsers themselves. */
1.1 kristaps 691:
1.45 schwarze 692: if (man && ! man_parseln(man, lnn_start, ln.buf, of)) {
693: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 694: break;
1.45 schwarze 695: }
696: if (mdoc && ! mdoc_parseln(mdoc, lnn_start, ln.buf, of)) {
697: assert(MANDOCLEVEL_FATAL <= exit_status);
1.1 kristaps 698: break;
699: }
700: }
701:
1.51 schwarze 702: free(ln.buf);
1.27 schwarze 703: if (with_mmap)
704: munmap(blk.buf, blk.sz);
705: else
706: free(blk.buf);
1.1 kristaps 707: }
708:
709:
1.45 schwarze 710: static void
1.1 kristaps 711: pset(const char *buf, int pos, struct curparse *curp,
712: struct man **man, struct mdoc **mdoc)
713: {
1.5 schwarze 714: int i;
1.1 kristaps 715:
716: /*
717: * Try to intuit which kind of manual parser should be used. If
718: * passed in by command-line (-man, -mdoc), then use that
719: * explicitly. If passed as -mandoc, then try to guess from the
1.5 schwarze 720: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.1 kristaps 721: * default to -man, which is more lenient.
722: */
723:
1.31 schwarze 724: if ('.' == buf[0] || '\'' == buf[0]) {
1.5 schwarze 725: for (i = 1; buf[i]; i++)
726: if (' ' != buf[i] && '\t' != buf[i])
727: break;
1.45 schwarze 728: if ('\0' == buf[i])
729: return;
1.5 schwarze 730: }
1.1 kristaps 731:
732: switch (curp->inttype) {
733: case (INTT_MDOC):
734: if (NULL == curp->mdoc)
1.45 schwarze 735: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
736: assert(curp->mdoc);
737: *mdoc = curp->mdoc;
738: return;
1.1 kristaps 739: case (INTT_MAN):
740: if (NULL == curp->man)
1.45 schwarze 741: curp->man = man_alloc(&curp->regs, curp, mmsg);
742: assert(curp->man);
743: *man = curp->man;
744: return;
1.1 kristaps 745: default:
746: break;
747: }
748:
749: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
750: if (NULL == curp->mdoc)
1.45 schwarze 751: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
752: assert(curp->mdoc);
753: *mdoc = curp->mdoc;
754: return;
1.1 kristaps 755: }
756:
757: if (NULL == curp->man)
1.45 schwarze 758: curp->man = man_alloc(&curp->regs, curp, mmsg);
759: assert(curp->man);
760: *man = curp->man;
1.1 kristaps 761: }
762:
763:
764: static int
765: moptions(enum intt *tflags, char *arg)
766: {
767:
768: if (0 == strcmp(arg, "doc"))
769: *tflags = INTT_MDOC;
770: else if (0 == strcmp(arg, "andoc"))
771: *tflags = INTT_AUTO;
772: else if (0 == strcmp(arg, "an"))
773: *tflags = INTT_MAN;
774: else {
1.20 schwarze 775: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 776: return(0);
777: }
778:
779: return(1);
780: }
781:
782:
783: static int
1.22 schwarze 784: toptions(struct curparse *curp, char *arg)
1.1 kristaps 785: {
786:
787: if (0 == strcmp(arg, "ascii"))
1.22 schwarze 788: curp->outtype = OUTT_ASCII;
789: else if (0 == strcmp(arg, "lint")) {
790: curp->outtype = OUTT_LINT;
1.45 schwarze 791: curp->wlevel = MANDOCLEVEL_WARNING;
1.22 schwarze 792: }
1.1 kristaps 793: else if (0 == strcmp(arg, "tree"))
1.22 schwarze 794: curp->outtype = OUTT_TREE;
1.17 schwarze 795: else if (0 == strcmp(arg, "html"))
1.22 schwarze 796: curp->outtype = OUTT_HTML;
1.21 schwarze 797: else if (0 == strcmp(arg, "xhtml"))
1.22 schwarze 798: curp->outtype = OUTT_XHTML;
1.36 schwarze 799: else if (0 == strcmp(arg, "ps"))
800: curp->outtype = OUTT_PS;
1.43 schwarze 801: else if (0 == strcmp(arg, "pdf"))
802: curp->outtype = OUTT_PDF;
1.1 kristaps 803: else {
1.20 schwarze 804: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 805: return(0);
806: }
807:
808: return(1);
809: }
810:
811:
812: static int
1.45 schwarze 813: woptions(struct curparse *curp, char *arg)
1.1 kristaps 814: {
1.10 schwarze 815: char *v, *o;
1.45 schwarze 816: const char *toks[6];
1.1 kristaps 817:
1.45 schwarze 818: toks[0] = "stop";
819: toks[1] = "all";
820: toks[2] = "warning";
821: toks[3] = "error";
822: toks[4] = "fatal";
823: toks[5] = NULL;
1.1 kristaps 824:
1.10 schwarze 825: while (*arg) {
826: o = arg;
1.17 schwarze 827: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 828: case (0):
1.45 schwarze 829: curp->wstop = 1;
1.1 kristaps 830: break;
831: case (1):
1.45 schwarze 832: /* FALLTHROUGH */
1.1 kristaps 833: case (2):
1.45 schwarze 834: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 835: break;
836: case (3):
1.45 schwarze 837: curp->wlevel = MANDOCLEVEL_ERROR;
1.1 kristaps 838: break;
839: case (4):
1.45 schwarze 840: curp->wlevel = MANDOCLEVEL_FATAL;
1.19 schwarze 841: break;
1.1 kristaps 842: default:
1.45 schwarze 843: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 844: return(0);
845: }
1.10 schwarze 846: }
1.1 kristaps 847:
848: return(1);
849: }
850:
851:
1.30 schwarze 852: static int
853: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
854: {
855: struct curparse *cp;
1.45 schwarze 856: enum mandoclevel level;
857:
858: level = MANDOCLEVEL_FATAL;
859: while (t < mandoclimits[level])
1.47 schwarze 860: /* LINTED */
1.45 schwarze 861: level--;
1.30 schwarze 862:
863: cp = (struct curparse *)arg;
1.45 schwarze 864: if (level < cp->wlevel)
865: return(1);
1.30 schwarze 866:
1.45 schwarze 867: fprintf(stderr, "%s:%d:%d: %s: %s",
868: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.30 schwarze 869: if (msg)
870: fprintf(stderr, ": %s", msg);
871: fputc('\n', stderr);
1.33 schwarze 872:
1.45 schwarze 873: if (exit_status < level)
874: exit_status = level;
875:
876: return(level < MANDOCLEVEL_FATAL);
1.30 schwarze 877: }