Annotation of src/usr.bin/mandoc/main.c, Revision 1.57
1.57 ! schwarze 1: /* $Id: main.c,v 1.56 2010/11/28 19:35:33 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. */
1.55 schwarze 67: int line; /* Line number in the file. */
1.45 schwarze 68: enum mandoclevel wlevel; /* Ignore messages below this. */
69: int wstop; /* Stop after a file with a warning. */
1.33 schwarze 70: enum intt inttype; /* which parser to use */
71: struct man *man; /* man parser */
72: struct mdoc *mdoc; /* mdoc parser */
73: struct roff *roff; /* roff parser (!NULL) */
1.38 schwarze 74: struct regset regs; /* roff registers */
1.33 schwarze 75: enum outt outtype; /* which output to use */
76: out_mdoc outmdoc; /* mdoc output ptr */
77: out_man outman; /* man output ptr */
78: out_free outfree; /* free output ptr */
79: void *outdata; /* data for output */
80: char outopts[BUFSIZ]; /* buf of output opts */
81: };
82:
1.45 schwarze 83: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
84: "SUCCESS",
85: "RESERVED",
86: "WARNING",
87: "ERROR",
88: "FATAL",
89: "BADARG",
90: "SYSERR"
91: };
92:
93: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
94: MANDOCERR_OK,
95: MANDOCERR_WARNING,
96: MANDOCERR_WARNING,
97: MANDOCERR_ERROR,
98: MANDOCERR_FATAL,
99: MANDOCERR_MAX,
100: MANDOCERR_MAX
101: };
102:
1.33 schwarze 103: static const char * const mandocerrs[MANDOCERR_MAX] = {
104: "ok",
1.40 schwarze 105:
106: "generic warning",
107:
1.53 schwarze 108: /* related to the prologue */
109: "no title in document",
110: "document title should be all caps",
111: "unknown manual section",
112: "cannot parse date argument",
113: "prologue macros out of order",
114: "duplicate prologue macro",
115: "macro not allowed in prologue",
116: "macro not allowed in body",
117:
118: /* related to document structure */
1.54 schwarze 119: ".so is fragile, better use ln(1)",
1.53 schwarze 120: "NAME section must come first",
121: "bad NAME section contents",
122: "manual name not yet set",
1.34 schwarze 123: "sections out of conventional order",
1.53 schwarze 124: "duplicate section name",
125: "section not in conventional manual section",
126:
127: /* related to macros and nesting */
128: "skipping obsolete macro",
129: "skipping paragraph macro",
130: "blocks badly nested",
131: "child violates parent syntax",
132: "nested displays are not portable",
133: "already in literal mode",
134:
135: /* related to missing macro arguments */
136: "skipping empty macro",
137: "missing display type",
1.33 schwarze 138: "list type must come first",
1.53 schwarze 139: "tag lists require a width argument",
140: "missing font type",
141:
142: /* related to bad macro arguments */
143: "skipping argument",
144: "duplicate argument",
145: "duplicate display type",
146: "duplicate list type",
147: "unknown AT&T UNIX version",
148: "bad Boolean value",
1.57 ! schwarze 149: "unknown font",
1.53 schwarze 150: "unknown library specifier",
151: "unknown standard specifier",
152: "bad width argument",
153:
154: /* related to plain text */
155: "blank line in non-literal context",
1.43 schwarze 156: "tab in non-literal context",
1.53 schwarze 157: "end of line whitespace",
158: "bad comment style",
159: "unknown escape sequence",
1.33 schwarze 160: "unterminated quoted string",
1.40 schwarze 161:
162: "generic error",
163:
1.53 schwarze 164: "skipping bad character",
165: "skipping text before the first section header",
166: "skipping unknown macro",
1.56 schwarze 167: "NOT IMPLEMENTED: skipping request",
1.33 schwarze 168: "line scope broken",
169: "argument count wrong",
1.53 schwarze 170: "skipping end of block that is not open",
1.50 schwarze 171: "missing end of block",
1.47 schwarze 172: "scope open on exit",
1.50 schwarze 173: "uname(3) system call failed",
1.33 schwarze 174: "macro requires line argument(s)",
175: "macro requires body argument(s)",
176: "macro requires argument(s)",
1.34 schwarze 177: "missing list type",
1.33 schwarze 178: "line argument(s) will be lost",
179: "body argument(s) will be lost",
1.49 schwarze 180: "tbl(1) error",
1.40 schwarze 181:
182: "generic fatal error",
183:
1.34 schwarze 184: "column syntax is inconsistent",
1.56 schwarze 185: "NOT IMPLEMENTED: .Bd -file",
1.33 schwarze 186: "line scope broken, syntax violated",
187: "argument count wrong, violates syntax",
188: "child violates parent syntax",
189: "argument count wrong, violates syntax",
1.56 schwarze 190: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.33 schwarze 191: "no document body",
192: "no document prologue",
1.45 schwarze 193: "static buffer exhausted",
1.1 kristaps 194: };
195:
1.55 schwarze 196: static void parsebuf(struct curparse *, struct buf, int);
1.51 schwarze 197: static void pdesc(struct curparse *);
1.27 schwarze 198: static void fdesc(struct curparse *);
199: static void ffile(const char *, struct curparse *);
1.55 schwarze 200: static int pfile(const char *, struct curparse *);
1.1 kristaps 201: static int moptions(enum intt *, char *);
1.30 schwarze 202: static int mmsg(enum mandocerr, void *,
203: int, int, const char *);
1.45 schwarze 204: static void pset(const char *, int, struct curparse *,
1.1 kristaps 205: struct man **, struct mdoc **);
1.27 schwarze 206: static int toptions(struct curparse *, char *);
207: static void usage(void) __attribute__((noreturn));
1.20 schwarze 208: static void version(void) __attribute__((noreturn));
1.45 schwarze 209: static int woptions(struct curparse *, char *);
1.1 kristaps 210:
1.19 schwarze 211: static const char *progname;
1.45 schwarze 212: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 213:
214: int
215: main(int argc, char *argv[])
216: {
1.27 schwarze 217: int c;
1.1 kristaps 218: struct curparse curp;
219:
1.19 schwarze 220: progname = strrchr(argv[0], '/');
221: if (progname == NULL)
222: progname = argv[0];
223: else
224: ++progname;
225:
226: memset(&curp, 0, sizeof(struct curparse));
1.1 kristaps 227:
228: curp.inttype = INTT_AUTO;
229: curp.outtype = OUTT_ASCII;
1.45 schwarze 230: curp.wlevel = MANDOCLEVEL_FATAL;
1.1 kristaps 231:
232: /* LINTED */
1.45 schwarze 233: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 234: switch (c) {
235: case ('m'):
236: if ( ! moptions(&curp.inttype, optarg))
1.47 schwarze 237: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 238: break;
1.18 schwarze 239: case ('O'):
240: (void)strlcat(curp.outopts, optarg, BUFSIZ);
241: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.17 schwarze 242: break;
1.1 kristaps 243: case ('T'):
1.22 schwarze 244: if ( ! toptions(&curp, optarg))
1.47 schwarze 245: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 246: break;
247: case ('W'):
1.45 schwarze 248: if ( ! woptions(&curp, optarg))
1.47 schwarze 249: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 250: break;
1.3 schwarze 251: case ('V'):
252: version();
253: /* NOTREACHED */
1.1 kristaps 254: default:
255: usage();
256: /* NOTREACHED */
257: }
258:
259: argc -= optind;
260: argv += optind;
261:
1.7 schwarze 262: if (NULL == *argv) {
263: curp.file = "<stdin>";
264: curp.fd = STDIN_FILENO;
1.14 schwarze 265:
1.27 schwarze 266: fdesc(&curp);
267: }
268:
269: while (*argv) {
270: ffile(*argv, &curp);
1.45 schwarze 271: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.27 schwarze 272: break;
273: ++argv;
1.1 kristaps 274: }
275:
276: if (curp.outfree)
277: (*curp.outfree)(curp.outdata);
1.30 schwarze 278: if (curp.mdoc)
279: mdoc_free(curp.mdoc);
280: if (curp.man)
281: man_free(curp.man);
282: if (curp.roff)
283: roff_free(curp.roff);
1.1 kristaps 284:
1.47 schwarze 285: return((int)exit_status);
1.1 kristaps 286: }
287:
288:
1.20 schwarze 289: static void
1.3 schwarze 290: version(void)
291: {
292:
1.19 schwarze 293: (void)printf("%s %s\n", progname, VERSION);
1.47 schwarze 294: exit((int)MANDOCLEVEL_OK);
1.3 schwarze 295: }
296:
297:
1.20 schwarze 298: static void
1.1 kristaps 299: usage(void)
300: {
301:
1.23 jmc 302: (void)fprintf(stderr, "usage: %s [-V] [-foption] "
1.18 schwarze 303: "[-mformat] [-Ooption] [-Toutput] "
1.23 jmc 304: "[-Werr] [file...]\n", progname);
1.47 schwarze 305: exit((int)MANDOCLEVEL_BADARG);
1.1 kristaps 306: }
307:
308:
1.27 schwarze 309: static void
310: ffile(const char *file, struct curparse *curp)
1.1 kristaps 311: {
312:
313: curp->file = file;
314: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.19 schwarze 315: perror(curp->file);
1.45 schwarze 316: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 317: return;
1.1 kristaps 318: }
319:
1.27 schwarze 320: fdesc(curp);
1.1 kristaps 321:
322: if (-1 == close(curp->fd))
1.19 schwarze 323: perror(curp->file);
1.27 schwarze 324: }
1.1 kristaps 325:
1.52 schwarze 326: static int
1.55 schwarze 327: pfile(const char *file, struct curparse *curp)
1.52 schwarze 328: {
329: const char *savefile;
330: int fd, savefd;
331:
332: if (-1 == (fd = open(file, O_RDONLY, 0))) {
333: perror(file);
334: exit_status = MANDOCLEVEL_SYSERR;
335: return(0);
336: }
337:
338: savefile = curp->file;
339: savefd = curp->fd;
340:
341: curp->file = file;
342: curp->fd = fd;
343:
344: pdesc(curp);
345:
346: curp->file = savefile;
347: curp->fd = savefd;
348:
349: if (-1 == close(fd))
350: perror(file);
351:
352: return(MANDOCLEVEL_FATAL > exit_status ? 1 : 0);
353: }
354:
1.27 schwarze 355:
1.45 schwarze 356: static void
1.27 schwarze 357: resize_buf(struct buf *buf, size_t initial)
358: {
359:
1.45 schwarze 360: buf->sz = buf->sz ? 2 * buf->sz : initial;
361: buf->buf = realloc(buf->buf, buf->sz);
362: if (NULL == buf->buf) {
1.27 schwarze 363: perror(NULL);
1.47 schwarze 364: exit((int)MANDOCLEVEL_SYSERR);
1.27 schwarze 365: }
1.1 kristaps 366: }
367:
368:
369: static int
1.27 schwarze 370: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
1.1 kristaps 371: {
1.27 schwarze 372: struct stat st;
373: size_t off;
1.1 kristaps 374: ssize_t ssz;
1.27 schwarze 375:
376: if (-1 == fstat(curp->fd, &st)) {
377: perror(curp->file);
378: return(0);
379: }
380:
381: /*
382: * If we're a regular file, try just reading in the whole entry
383: * via mmap(). This is faster than reading it into blocks, and
384: * since each file is only a few bytes to begin with, I'm not
385: * concerned that this is going to tank any machines.
386: */
387:
388: if (S_ISREG(st.st_mode)) {
389: if (st.st_size >= (1U << 31)) {
390: fprintf(stderr, "%s: input too large\n",
391: curp->file);
392: return(0);
393: }
394: *with_mmap = 1;
1.30 schwarze 395: fb->sz = (size_t)st.st_size;
1.27 schwarze 396: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.35 schwarze 397: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.27 schwarze 398: if (fb->buf != MAP_FAILED)
399: return(1);
400: }
401:
402: /*
403: * If this isn't a regular file (like, say, stdin), then we must
404: * go the old way and just read things in bit by bit.
405: */
406:
407: *with_mmap = 0;
408: off = 0;
409: fb->sz = 0;
410: fb->buf = NULL;
411: for (;;) {
412: if (off == fb->sz) {
413: if (fb->sz == (1U << 31)) {
414: fprintf(stderr, "%s: input too large\n",
415: curp->file);
416: break;
417: }
1.45 schwarze 418: resize_buf(fb, 65536);
1.27 schwarze 419: }
1.30 schwarze 420: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.27 schwarze 421: if (ssz == 0) {
422: fb->sz = off;
423: return(1);
424: }
425: if (ssz == -1) {
426: perror(curp->file);
427: break;
428: }
1.30 schwarze 429: off += (size_t)ssz;
1.27 schwarze 430: }
431:
432: free(fb->buf);
433: fb->buf = NULL;
434: return(0);
435: }
436:
437:
438: static void
439: fdesc(struct curparse *curp)
440: {
1.51 schwarze 441: struct man *man;
442: struct mdoc *mdoc;
443: struct roff *roff;
444:
445: pdesc(curp);
446:
447: man = curp->man;
448: mdoc = curp->mdoc;
449: roff = curp->roff;
450:
451: if (MANDOCLEVEL_FATAL <= exit_status)
452: goto cleanup;
453:
454: /* NOTE a parser may not have been assigned, yet. */
455:
456: if ( ! (man || mdoc)) {
457: fprintf(stderr, "%s: Not a manual\n", curp->file);
458: exit_status = MANDOCLEVEL_FATAL;
459: goto cleanup;
460: }
461:
462: /* Clean up the parse routine ASTs. */
463:
464: if (mdoc && ! mdoc_endparse(mdoc)) {
465: assert(MANDOCLEVEL_FATAL <= exit_status);
466: goto cleanup;
467: }
468: if (man && ! man_endparse(man)) {
469: assert(MANDOCLEVEL_FATAL <= exit_status);
470: goto cleanup;
471: }
472: if (roff && ! roff_endparse(roff)) {
473: assert(MANDOCLEVEL_FATAL <= exit_status);
474: goto cleanup;
475: }
476:
477: /*
478: * With -Wstop and warnings or errors of at least
479: * the requested level, do not produce output.
480: */
481:
482: if (MANDOCLEVEL_OK != exit_status && curp->wstop)
483: goto cleanup;
484:
485: /* If unset, allocate output dev now (if applicable). */
486:
487: if ( ! (curp->outman && curp->outmdoc)) {
488: switch (curp->outtype) {
489: case (OUTT_XHTML):
490: curp->outdata = xhtml_alloc(curp->outopts);
491: break;
492: case (OUTT_HTML):
493: curp->outdata = html_alloc(curp->outopts);
494: break;
495: case (OUTT_ASCII):
496: curp->outdata = ascii_alloc(curp->outopts);
497: curp->outfree = ascii_free;
498: break;
499: case (OUTT_PDF):
500: curp->outdata = pdf_alloc(curp->outopts);
501: curp->outfree = pspdf_free;
502: break;
503: case (OUTT_PS):
504: curp->outdata = ps_alloc(curp->outopts);
505: curp->outfree = pspdf_free;
506: break;
507: default:
508: break;
509: }
510:
511: switch (curp->outtype) {
512: case (OUTT_HTML):
513: /* FALLTHROUGH */
514: case (OUTT_XHTML):
515: curp->outman = html_man;
516: curp->outmdoc = html_mdoc;
517: curp->outfree = html_free;
518: break;
519: case (OUTT_TREE):
520: curp->outman = tree_man;
521: curp->outmdoc = tree_mdoc;
522: break;
523: case (OUTT_PDF):
524: /* FALLTHROUGH */
525: case (OUTT_ASCII):
526: /* FALLTHROUGH */
527: case (OUTT_PS):
528: curp->outman = terminal_man;
529: curp->outmdoc = terminal_mdoc;
530: break;
531: default:
532: break;
533: }
534: }
535:
536: /* Execute the out device, if it exists. */
537:
538: if (man && curp->outman)
539: (*curp->outman)(curp->outdata, man);
540: if (mdoc && curp->outmdoc)
541: (*curp->outmdoc)(curp->outdata, mdoc);
542:
543: cleanup:
544: memset(&curp->regs, 0, sizeof(struct regset));
545: if (mdoc)
546: mdoc_reset(mdoc);
547: if (man)
548: man_reset(man);
549: if (roff)
550: roff_reset(roff);
551:
552: return;
553: }
554:
555:
556: static void
557: pdesc(struct curparse *curp)
558: {
1.55 schwarze 559: struct buf blk;
560: int with_mmap;
1.1 kristaps 561:
1.45 schwarze 562: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
563: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 564: return;
1.45 schwarze 565: }
1.27 schwarze 566:
1.30 schwarze 567: if (NULL == curp->roff)
1.45 schwarze 568: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
569: assert(curp->roff);
1.55 schwarze 570:
571: curp->line = 1;
572: parsebuf(curp, blk, 1);
573:
574: if (with_mmap)
575: munmap(blk.buf, blk.sz);
576: else
577: free(blk.buf);
578: }
579:
580: static void
581: parsebuf(struct curparse *curp, struct buf blk, int start)
582: {
583: struct buf ln;
584: int i, pos, lnn, of;
585: unsigned char c;
586: struct man *man;
587: struct mdoc *mdoc;
588: struct roff *roff;
589:
590: man = curp->man;
591: mdoc = curp->mdoc;
1.45 schwarze 592: roff = curp->roff;
1.30 schwarze 593:
1.55 schwarze 594: memset(&ln, 0, sizeof(struct buf));
595:
596: lnn = curp->line; /* line number in the real file */
597: pos = 0; /* byte number in the ln buffer */
598:
599: for (i = 0; i < (int)blk.sz;) {
600: if (0 == pos && '\0' == blk.buf[i])
601: break;
602: if (start)
603: curp->line = lnn;
604:
605: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.29 schwarze 606: if ('\n' == blk.buf[i]) {
607: ++i;
608: ++lnn;
609: break;
610: }
1.43 schwarze 611:
612: /*
613: * Warn about bogus characters. If you're using
614: * non-ASCII encoding, you're screwing your
615: * readers. Since I'd rather this not happen,
616: * I'll be helpful and drop these characters so
617: * we don't display gibberish. Note to manual
618: * writers: use special characters.
619: */
620:
1.44 schwarze 621: c = (unsigned char) blk.buf[i];
622: if ( ! (isascii(c) && (isgraph(c) || isblank(c)))) {
1.45 schwarze 623: mmsg(MANDOCERR_BADCHAR, curp,
1.55 schwarze 624: curp->line, pos, "ignoring byte");
1.43 schwarze 625: i++;
626: continue;
627: }
628:
1.29 schwarze 629: /* Trailing backslash is like a plain character. */
630: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
631: if (pos >= (int)ln.sz)
1.45 schwarze 632: resize_buf(&ln, 256);
1.29 schwarze 633: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 634: continue;
1.29 schwarze 635: }
636: /* Found an escape and at least one other character. */
637: if ('\n' == blk.buf[i + 1]) {
638: /* Escaped newlines are skipped over */
639: i += 2;
640: ++lnn;
1.27 schwarze 641: continue;
642: }
1.29 schwarze 643: if ('"' == blk.buf[i + 1]) {
644: i += 2;
645: /* Comment, skip to end of line */
646: for (; i < (int)blk.sz; ++i) {
647: if ('\n' == blk.buf[i]) {
648: ++i;
649: ++lnn;
650: break;
651: }
652: }
653: /* Backout trailing whitespaces */
654: for (; pos > 0; --pos) {
655: if (ln.buf[pos - 1] != ' ')
656: break;
657: if (pos > 2 && ln.buf[pos - 2] == '\\')
658: break;
659: }
660: break;
661: }
662: /* Some other escape sequence, copy and continue. */
663: if (pos + 1 >= (int)ln.sz)
1.45 schwarze 664: resize_buf(&ln, 256);
1.1 kristaps 665:
1.29 schwarze 666: ln.buf[pos++] = blk.buf[i++];
667: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 668: }
1.1 kristaps 669:
1.29 schwarze 670: if (pos >= (int)ln.sz)
1.45 schwarze 671: resize_buf(&ln, 256);
1.30 schwarze 672: ln.buf[pos] = '\0';
673:
1.32 schwarze 674: /*
675: * A significant amount of complexity is contained by
676: * the roff preprocessor. It's line-oriented but can be
677: * expressed on one line, so we need at times to
678: * readjust our starting point and re-run it. The roff
679: * preprocessor can also readjust the buffers with new
680: * data, so we pass them in wholesale.
681: */
682:
683: of = 0;
1.55 schwarze 684: rerun:
685: switch (roff_parseln(roff, curp->line, &ln.buf, &ln.sz,
686: of, &of)) {
687: case (ROFF_REPARSE):
688: parsebuf(curp, ln, 0);
689: pos = 0;
690: continue;
691: case (ROFF_APPEND):
692: pos = strlen(ln.buf);
693: continue;
694: case (ROFF_RERUN):
695: goto rerun;
696: case (ROFF_IGN):
697: pos = 0;
1.30 schwarze 698: continue;
1.55 schwarze 699: case (ROFF_ERR):
1.45 schwarze 700: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 701: break;
1.55 schwarze 702: case (ROFF_SO):
703: if (pfile(ln.buf + of, curp)) {
704: pos = 0;
1.52 schwarze 705: continue;
1.55 schwarze 706: } else
1.52 schwarze 707: break;
1.55 schwarze 708: case (ROFF_CONT):
709: break;
1.45 schwarze 710: }
1.5 schwarze 711:
1.32 schwarze 712: /*
713: * If input parsers have not been allocated, do so now.
714: * We keep these instanced betwen parsers, but set them
715: * locally per parse routine since we can use different
716: * parsers with each one.
717: */
1.1 kristaps 718:
1.32 schwarze 719: if ( ! (man || mdoc))
1.45 schwarze 720: pset(ln.buf + of, pos - of, curp, &man, &mdoc);
1.5 schwarze 721:
1.32 schwarze 722: /* Lastly, push down into the parsers themselves. */
1.1 kristaps 723:
1.55 schwarze 724: if (man && ! man_parseln(man, curp->line, ln.buf, of)) {
1.45 schwarze 725: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 726: break;
1.45 schwarze 727: }
1.55 schwarze 728: if (mdoc && ! mdoc_parseln(mdoc, curp->line, ln.buf, of)) {
1.45 schwarze 729: assert(MANDOCLEVEL_FATAL <= exit_status);
1.1 kristaps 730: break;
731: }
1.55 schwarze 732:
733: /* Temporary buffers typically are not full. */
734: if (0 == start && '\0' == blk.buf[i])
735: break;
736:
737: /* Start the next input line. */
738: pos = 0;
1.1 kristaps 739: }
740:
1.51 schwarze 741: free(ln.buf);
1.1 kristaps 742: }
743:
744:
1.45 schwarze 745: static void
1.1 kristaps 746: pset(const char *buf, int pos, struct curparse *curp,
747: struct man **man, struct mdoc **mdoc)
748: {
1.5 schwarze 749: int i;
1.1 kristaps 750:
751: /*
752: * Try to intuit which kind of manual parser should be used. If
753: * passed in by command-line (-man, -mdoc), then use that
754: * explicitly. If passed as -mandoc, then try to guess from the
1.5 schwarze 755: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.1 kristaps 756: * default to -man, which is more lenient.
757: */
758:
1.31 schwarze 759: if ('.' == buf[0] || '\'' == buf[0]) {
1.5 schwarze 760: for (i = 1; buf[i]; i++)
761: if (' ' != buf[i] && '\t' != buf[i])
762: break;
1.45 schwarze 763: if ('\0' == buf[i])
764: return;
1.5 schwarze 765: }
1.1 kristaps 766:
767: switch (curp->inttype) {
768: case (INTT_MDOC):
769: if (NULL == curp->mdoc)
1.45 schwarze 770: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
771: assert(curp->mdoc);
772: *mdoc = curp->mdoc;
773: return;
1.1 kristaps 774: case (INTT_MAN):
775: if (NULL == curp->man)
1.45 schwarze 776: curp->man = man_alloc(&curp->regs, curp, mmsg);
777: assert(curp->man);
778: *man = curp->man;
779: return;
1.1 kristaps 780: default:
781: break;
782: }
783:
784: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
785: if (NULL == curp->mdoc)
1.45 schwarze 786: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
787: assert(curp->mdoc);
788: *mdoc = curp->mdoc;
789: return;
1.1 kristaps 790: }
791:
792: if (NULL == curp->man)
1.45 schwarze 793: curp->man = man_alloc(&curp->regs, curp, mmsg);
794: assert(curp->man);
795: *man = curp->man;
1.1 kristaps 796: }
797:
798:
799: static int
800: moptions(enum intt *tflags, char *arg)
801: {
802:
803: if (0 == strcmp(arg, "doc"))
804: *tflags = INTT_MDOC;
805: else if (0 == strcmp(arg, "andoc"))
806: *tflags = INTT_AUTO;
807: else if (0 == strcmp(arg, "an"))
808: *tflags = INTT_MAN;
809: else {
1.20 schwarze 810: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 811: return(0);
812: }
813:
814: return(1);
815: }
816:
817:
818: static int
1.22 schwarze 819: toptions(struct curparse *curp, char *arg)
1.1 kristaps 820: {
821:
822: if (0 == strcmp(arg, "ascii"))
1.22 schwarze 823: curp->outtype = OUTT_ASCII;
824: else if (0 == strcmp(arg, "lint")) {
825: curp->outtype = OUTT_LINT;
1.45 schwarze 826: curp->wlevel = MANDOCLEVEL_WARNING;
1.22 schwarze 827: }
1.1 kristaps 828: else if (0 == strcmp(arg, "tree"))
1.22 schwarze 829: curp->outtype = OUTT_TREE;
1.17 schwarze 830: else if (0 == strcmp(arg, "html"))
1.22 schwarze 831: curp->outtype = OUTT_HTML;
1.21 schwarze 832: else if (0 == strcmp(arg, "xhtml"))
1.22 schwarze 833: curp->outtype = OUTT_XHTML;
1.36 schwarze 834: else if (0 == strcmp(arg, "ps"))
835: curp->outtype = OUTT_PS;
1.43 schwarze 836: else if (0 == strcmp(arg, "pdf"))
837: curp->outtype = OUTT_PDF;
1.1 kristaps 838: else {
1.20 schwarze 839: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 840: return(0);
841: }
842:
843: return(1);
844: }
845:
846:
847: static int
1.45 schwarze 848: woptions(struct curparse *curp, char *arg)
1.1 kristaps 849: {
1.10 schwarze 850: char *v, *o;
1.45 schwarze 851: const char *toks[6];
1.1 kristaps 852:
1.45 schwarze 853: toks[0] = "stop";
854: toks[1] = "all";
855: toks[2] = "warning";
856: toks[3] = "error";
857: toks[4] = "fatal";
858: toks[5] = NULL;
1.1 kristaps 859:
1.10 schwarze 860: while (*arg) {
861: o = arg;
1.17 schwarze 862: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 863: case (0):
1.45 schwarze 864: curp->wstop = 1;
1.1 kristaps 865: break;
866: case (1):
1.45 schwarze 867: /* FALLTHROUGH */
1.1 kristaps 868: case (2):
1.45 schwarze 869: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 870: break;
871: case (3):
1.45 schwarze 872: curp->wlevel = MANDOCLEVEL_ERROR;
1.1 kristaps 873: break;
874: case (4):
1.45 schwarze 875: curp->wlevel = MANDOCLEVEL_FATAL;
1.19 schwarze 876: break;
1.1 kristaps 877: default:
1.45 schwarze 878: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 879: return(0);
880: }
1.10 schwarze 881: }
1.1 kristaps 882:
883: return(1);
884: }
885:
886:
1.30 schwarze 887: static int
888: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
889: {
890: struct curparse *cp;
1.45 schwarze 891: enum mandoclevel level;
892:
893: level = MANDOCLEVEL_FATAL;
894: while (t < mandoclimits[level])
1.47 schwarze 895: /* LINTED */
1.45 schwarze 896: level--;
1.30 schwarze 897:
898: cp = (struct curparse *)arg;
1.45 schwarze 899: if (level < cp->wlevel)
900: return(1);
1.30 schwarze 901:
1.45 schwarze 902: fprintf(stderr, "%s:%d:%d: %s: %s",
903: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.30 schwarze 904: if (msg)
905: fprintf(stderr, ": %s", msg);
906: fputc('\n', stderr);
1.33 schwarze 907:
1.45 schwarze 908: if (exit_status < level)
909: exit_status = level;
910:
911: return(level < MANDOCLEVEL_FATAL);
1.30 schwarze 912: }