Annotation of src/usr.bin/mandoc/main.c, Revision 1.55
1.55 ! schwarze 1: /* $Id: main.c,v 1.54 2010/10/26 23:34:38 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",
149: "unknown library specifier",
150: "unknown standard specifier",
151: "bad width argument",
152:
153: /* related to plain text */
154: "blank line in non-literal context",
1.43 schwarze 155: "tab in non-literal context",
1.53 schwarze 156: "end of line whitespace",
157: "bad comment style",
158: "unknown escape sequence",
1.33 schwarze 159: "unterminated quoted string",
1.40 schwarze 160:
161: "generic error",
162:
1.53 schwarze 163: "skipping bad character",
164: "skipping text before the first section header",
165: "skipping unknown macro",
1.33 schwarze 166: "line scope broken",
167: "argument count wrong",
1.53 schwarze 168: "skipping end of block that is not open",
1.50 schwarze 169: "missing end of block",
1.47 schwarze 170: "scope open on exit",
1.50 schwarze 171: "uname(3) system call failed",
1.33 schwarze 172: "macro requires line argument(s)",
173: "macro requires body argument(s)",
174: "macro requires argument(s)",
1.34 schwarze 175: "missing list type",
1.33 schwarze 176: "line argument(s) will be lost",
177: "body argument(s) will be lost",
1.49 schwarze 178: "tbl(1) error",
1.40 schwarze 179:
180: "generic fatal error",
181:
1.34 schwarze 182: "column syntax is inconsistent",
1.37 schwarze 183: "unsupported display type",
1.33 schwarze 184: "line scope broken, syntax violated",
185: "argument count wrong, violates syntax",
186: "child violates parent syntax",
187: "argument count wrong, violates syntax",
1.52 schwarze 188: "invalid path in include directive",
1.33 schwarze 189: "no document body",
190: "no document prologue",
1.45 schwarze 191: "static buffer exhausted",
1.1 kristaps 192: };
193:
1.55 ! schwarze 194: static void parsebuf(struct curparse *, struct buf, int);
1.51 schwarze 195: static void pdesc(struct curparse *);
1.27 schwarze 196: static void fdesc(struct curparse *);
197: static void ffile(const char *, struct curparse *);
1.55 ! schwarze 198: static int pfile(const char *, struct curparse *);
1.1 kristaps 199: static int moptions(enum intt *, char *);
1.30 schwarze 200: static int mmsg(enum mandocerr, void *,
201: int, int, const char *);
1.45 schwarze 202: static void pset(const char *, int, struct curparse *,
1.1 kristaps 203: struct man **, struct mdoc **);
1.27 schwarze 204: static int toptions(struct curparse *, char *);
205: static void usage(void) __attribute__((noreturn));
1.20 schwarze 206: static void version(void) __attribute__((noreturn));
1.45 schwarze 207: static int woptions(struct curparse *, char *);
1.1 kristaps 208:
1.19 schwarze 209: static const char *progname;
1.45 schwarze 210: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 211:
212: int
213: main(int argc, char *argv[])
214: {
1.27 schwarze 215: int c;
1.1 kristaps 216: struct curparse curp;
217:
1.19 schwarze 218: progname = strrchr(argv[0], '/');
219: if (progname == NULL)
220: progname = argv[0];
221: else
222: ++progname;
223:
224: memset(&curp, 0, sizeof(struct curparse));
1.1 kristaps 225:
226: curp.inttype = INTT_AUTO;
227: curp.outtype = OUTT_ASCII;
1.45 schwarze 228: curp.wlevel = MANDOCLEVEL_FATAL;
1.1 kristaps 229:
230: /* LINTED */
1.45 schwarze 231: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 232: switch (c) {
233: case ('m'):
234: if ( ! moptions(&curp.inttype, optarg))
1.47 schwarze 235: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 236: break;
1.18 schwarze 237: case ('O'):
238: (void)strlcat(curp.outopts, optarg, BUFSIZ);
239: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.17 schwarze 240: break;
1.1 kristaps 241: case ('T'):
1.22 schwarze 242: if ( ! toptions(&curp, optarg))
1.47 schwarze 243: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 244: break;
245: case ('W'):
1.45 schwarze 246: if ( ! woptions(&curp, optarg))
1.47 schwarze 247: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 248: break;
1.3 schwarze 249: case ('V'):
250: version();
251: /* NOTREACHED */
1.1 kristaps 252: default:
253: usage();
254: /* NOTREACHED */
255: }
256:
257: argc -= optind;
258: argv += optind;
259:
1.7 schwarze 260: if (NULL == *argv) {
261: curp.file = "<stdin>";
262: curp.fd = STDIN_FILENO;
1.14 schwarze 263:
1.27 schwarze 264: fdesc(&curp);
265: }
266:
267: while (*argv) {
268: ffile(*argv, &curp);
1.45 schwarze 269: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.27 schwarze 270: break;
271: ++argv;
1.1 kristaps 272: }
273:
274: if (curp.outfree)
275: (*curp.outfree)(curp.outdata);
1.30 schwarze 276: if (curp.mdoc)
277: mdoc_free(curp.mdoc);
278: if (curp.man)
279: man_free(curp.man);
280: if (curp.roff)
281: roff_free(curp.roff);
1.1 kristaps 282:
1.47 schwarze 283: return((int)exit_status);
1.1 kristaps 284: }
285:
286:
1.20 schwarze 287: static void
1.3 schwarze 288: version(void)
289: {
290:
1.19 schwarze 291: (void)printf("%s %s\n", progname, VERSION);
1.47 schwarze 292: exit((int)MANDOCLEVEL_OK);
1.3 schwarze 293: }
294:
295:
1.20 schwarze 296: static void
1.1 kristaps 297: usage(void)
298: {
299:
1.23 jmc 300: (void)fprintf(stderr, "usage: %s [-V] [-foption] "
1.18 schwarze 301: "[-mformat] [-Ooption] [-Toutput] "
1.23 jmc 302: "[-Werr] [file...]\n", progname);
1.47 schwarze 303: exit((int)MANDOCLEVEL_BADARG);
1.1 kristaps 304: }
305:
306:
1.27 schwarze 307: static void
308: ffile(const char *file, struct curparse *curp)
1.1 kristaps 309: {
310:
311: curp->file = file;
312: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.19 schwarze 313: perror(curp->file);
1.45 schwarze 314: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 315: return;
1.1 kristaps 316: }
317:
1.27 schwarze 318: fdesc(curp);
1.1 kristaps 319:
320: if (-1 == close(curp->fd))
1.19 schwarze 321: perror(curp->file);
1.27 schwarze 322: }
1.1 kristaps 323:
1.52 schwarze 324: static int
1.55 ! schwarze 325: pfile(const char *file, struct curparse *curp)
1.52 schwarze 326: {
327: const char *savefile;
328: int fd, savefd;
329:
330: if (-1 == (fd = open(file, O_RDONLY, 0))) {
331: perror(file);
332: exit_status = MANDOCLEVEL_SYSERR;
333: return(0);
334: }
335:
336: savefile = curp->file;
337: savefd = curp->fd;
338:
339: curp->file = file;
340: curp->fd = fd;
341:
342: pdesc(curp);
343:
344: curp->file = savefile;
345: curp->fd = savefd;
346:
347: if (-1 == close(fd))
348: perror(file);
349:
350: return(MANDOCLEVEL_FATAL > exit_status ? 1 : 0);
351: }
352:
1.27 schwarze 353:
1.45 schwarze 354: static void
1.27 schwarze 355: resize_buf(struct buf *buf, size_t initial)
356: {
357:
1.45 schwarze 358: buf->sz = buf->sz ? 2 * buf->sz : initial;
359: buf->buf = realloc(buf->buf, buf->sz);
360: if (NULL == buf->buf) {
1.27 schwarze 361: perror(NULL);
1.47 schwarze 362: exit((int)MANDOCLEVEL_SYSERR);
1.27 schwarze 363: }
1.1 kristaps 364: }
365:
366:
367: static int
1.27 schwarze 368: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
1.1 kristaps 369: {
1.27 schwarze 370: struct stat st;
371: size_t off;
1.1 kristaps 372: ssize_t ssz;
1.27 schwarze 373:
374: if (-1 == fstat(curp->fd, &st)) {
375: perror(curp->file);
376: return(0);
377: }
378:
379: /*
380: * If we're a regular file, try just reading in the whole entry
381: * via mmap(). This is faster than reading it into blocks, and
382: * since each file is only a few bytes to begin with, I'm not
383: * concerned that this is going to tank any machines.
384: */
385:
386: if (S_ISREG(st.st_mode)) {
387: if (st.st_size >= (1U << 31)) {
388: fprintf(stderr, "%s: input too large\n",
389: curp->file);
390: return(0);
391: }
392: *with_mmap = 1;
1.30 schwarze 393: fb->sz = (size_t)st.st_size;
1.27 schwarze 394: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.35 schwarze 395: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.27 schwarze 396: if (fb->buf != MAP_FAILED)
397: return(1);
398: }
399:
400: /*
401: * If this isn't a regular file (like, say, stdin), then we must
402: * go the old way and just read things in bit by bit.
403: */
404:
405: *with_mmap = 0;
406: off = 0;
407: fb->sz = 0;
408: fb->buf = NULL;
409: for (;;) {
410: if (off == fb->sz) {
411: if (fb->sz == (1U << 31)) {
412: fprintf(stderr, "%s: input too large\n",
413: curp->file);
414: break;
415: }
1.45 schwarze 416: resize_buf(fb, 65536);
1.27 schwarze 417: }
1.30 schwarze 418: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.27 schwarze 419: if (ssz == 0) {
420: fb->sz = off;
421: return(1);
422: }
423: if (ssz == -1) {
424: perror(curp->file);
425: break;
426: }
1.30 schwarze 427: off += (size_t)ssz;
1.27 schwarze 428: }
429:
430: free(fb->buf);
431: fb->buf = NULL;
432: return(0);
433: }
434:
435:
436: static void
437: fdesc(struct curparse *curp)
438: {
1.51 schwarze 439: struct man *man;
440: struct mdoc *mdoc;
441: struct roff *roff;
442:
443: pdesc(curp);
444:
445: man = curp->man;
446: mdoc = curp->mdoc;
447: roff = curp->roff;
448:
449: if (MANDOCLEVEL_FATAL <= exit_status)
450: goto cleanup;
451:
452: /* NOTE a parser may not have been assigned, yet. */
453:
454: if ( ! (man || mdoc)) {
455: fprintf(stderr, "%s: Not a manual\n", curp->file);
456: exit_status = MANDOCLEVEL_FATAL;
457: goto cleanup;
458: }
459:
460: /* Clean up the parse routine ASTs. */
461:
462: if (mdoc && ! mdoc_endparse(mdoc)) {
463: assert(MANDOCLEVEL_FATAL <= exit_status);
464: goto cleanup;
465: }
466: if (man && ! man_endparse(man)) {
467: assert(MANDOCLEVEL_FATAL <= exit_status);
468: goto cleanup;
469: }
470: if (roff && ! roff_endparse(roff)) {
471: assert(MANDOCLEVEL_FATAL <= exit_status);
472: goto cleanup;
473: }
474:
475: /*
476: * With -Wstop and warnings or errors of at least
477: * the requested level, do not produce output.
478: */
479:
480: if (MANDOCLEVEL_OK != exit_status && curp->wstop)
481: goto cleanup;
482:
483: /* If unset, allocate output dev now (if applicable). */
484:
485: if ( ! (curp->outman && curp->outmdoc)) {
486: switch (curp->outtype) {
487: case (OUTT_XHTML):
488: curp->outdata = xhtml_alloc(curp->outopts);
489: break;
490: case (OUTT_HTML):
491: curp->outdata = html_alloc(curp->outopts);
492: break;
493: case (OUTT_ASCII):
494: curp->outdata = ascii_alloc(curp->outopts);
495: curp->outfree = ascii_free;
496: break;
497: case (OUTT_PDF):
498: curp->outdata = pdf_alloc(curp->outopts);
499: curp->outfree = pspdf_free;
500: break;
501: case (OUTT_PS):
502: curp->outdata = ps_alloc(curp->outopts);
503: curp->outfree = pspdf_free;
504: break;
505: default:
506: break;
507: }
508:
509: switch (curp->outtype) {
510: case (OUTT_HTML):
511: /* FALLTHROUGH */
512: case (OUTT_XHTML):
513: curp->outman = html_man;
514: curp->outmdoc = html_mdoc;
515: curp->outfree = html_free;
516: break;
517: case (OUTT_TREE):
518: curp->outman = tree_man;
519: curp->outmdoc = tree_mdoc;
520: break;
521: case (OUTT_PDF):
522: /* FALLTHROUGH */
523: case (OUTT_ASCII):
524: /* FALLTHROUGH */
525: case (OUTT_PS):
526: curp->outman = terminal_man;
527: curp->outmdoc = terminal_mdoc;
528: break;
529: default:
530: break;
531: }
532: }
533:
534: /* Execute the out device, if it exists. */
535:
536: if (man && curp->outman)
537: (*curp->outman)(curp->outdata, man);
538: if (mdoc && curp->outmdoc)
539: (*curp->outmdoc)(curp->outdata, mdoc);
540:
541: cleanup:
542: memset(&curp->regs, 0, sizeof(struct regset));
543: if (mdoc)
544: mdoc_reset(mdoc);
545: if (man)
546: man_reset(man);
547: if (roff)
548: roff_reset(roff);
549:
550: return;
551: }
552:
553:
554: static void
555: pdesc(struct curparse *curp)
556: {
1.55 ! schwarze 557: struct buf blk;
! 558: int with_mmap;
1.1 kristaps 559:
1.45 schwarze 560: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
561: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 562: return;
1.45 schwarze 563: }
1.27 schwarze 564:
1.30 schwarze 565: if (NULL == curp->roff)
1.45 schwarze 566: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
567: assert(curp->roff);
1.55 ! schwarze 568:
! 569: curp->line = 1;
! 570: parsebuf(curp, blk, 1);
! 571:
! 572: if (with_mmap)
! 573: munmap(blk.buf, blk.sz);
! 574: else
! 575: free(blk.buf);
! 576: }
! 577:
! 578: static void
! 579: parsebuf(struct curparse *curp, struct buf blk, int start)
! 580: {
! 581: struct buf ln;
! 582: int i, pos, lnn, of;
! 583: unsigned char c;
! 584: struct man *man;
! 585: struct mdoc *mdoc;
! 586: struct roff *roff;
! 587:
! 588: man = curp->man;
! 589: mdoc = curp->mdoc;
1.45 schwarze 590: roff = curp->roff;
1.30 schwarze 591:
1.55 ! schwarze 592: memset(&ln, 0, sizeof(struct buf));
! 593:
! 594: lnn = curp->line; /* line number in the real file */
! 595: pos = 0; /* byte number in the ln buffer */
! 596:
! 597: for (i = 0; i < (int)blk.sz;) {
! 598: if (0 == pos && '\0' == blk.buf[i])
! 599: break;
! 600: if (start)
! 601: curp->line = lnn;
! 602:
! 603: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.29 schwarze 604: if ('\n' == blk.buf[i]) {
605: ++i;
606: ++lnn;
607: break;
608: }
1.43 schwarze 609:
610: /*
611: * Warn about bogus characters. If you're using
612: * non-ASCII encoding, you're screwing your
613: * readers. Since I'd rather this not happen,
614: * I'll be helpful and drop these characters so
615: * we don't display gibberish. Note to manual
616: * writers: use special characters.
617: */
618:
1.44 schwarze 619: c = (unsigned char) blk.buf[i];
620: if ( ! (isascii(c) && (isgraph(c) || isblank(c)))) {
1.45 schwarze 621: mmsg(MANDOCERR_BADCHAR, curp,
1.55 ! schwarze 622: curp->line, pos, "ignoring byte");
1.43 schwarze 623: i++;
624: continue;
625: }
626:
1.29 schwarze 627: /* Trailing backslash is like a plain character. */
628: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
629: if (pos >= (int)ln.sz)
1.45 schwarze 630: resize_buf(&ln, 256);
1.29 schwarze 631: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 632: continue;
1.29 schwarze 633: }
634: /* Found an escape and at least one other character. */
635: if ('\n' == blk.buf[i + 1]) {
636: /* Escaped newlines are skipped over */
637: i += 2;
638: ++lnn;
1.27 schwarze 639: continue;
640: }
1.29 schwarze 641: if ('"' == blk.buf[i + 1]) {
642: i += 2;
643: /* Comment, skip to end of line */
644: for (; i < (int)blk.sz; ++i) {
645: if ('\n' == blk.buf[i]) {
646: ++i;
647: ++lnn;
648: break;
649: }
650: }
651: /* Backout trailing whitespaces */
652: for (; pos > 0; --pos) {
653: if (ln.buf[pos - 1] != ' ')
654: break;
655: if (pos > 2 && ln.buf[pos - 2] == '\\')
656: break;
657: }
658: break;
659: }
660: /* Some other escape sequence, copy and continue. */
661: if (pos + 1 >= (int)ln.sz)
1.45 schwarze 662: resize_buf(&ln, 256);
1.1 kristaps 663:
1.29 schwarze 664: ln.buf[pos++] = blk.buf[i++];
665: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 666: }
1.1 kristaps 667:
1.29 schwarze 668: if (pos >= (int)ln.sz)
1.45 schwarze 669: resize_buf(&ln, 256);
1.30 schwarze 670: ln.buf[pos] = '\0';
671:
1.32 schwarze 672: /*
673: * A significant amount of complexity is contained by
674: * the roff preprocessor. It's line-oriented but can be
675: * expressed on one line, so we need at times to
676: * readjust our starting point and re-run it. The roff
677: * preprocessor can also readjust the buffers with new
678: * data, so we pass them in wholesale.
679: */
680:
681: of = 0;
1.55 ! schwarze 682: rerun:
! 683: switch (roff_parseln(roff, curp->line, &ln.buf, &ln.sz,
! 684: of, &of)) {
! 685: case (ROFF_REPARSE):
! 686: parsebuf(curp, ln, 0);
! 687: pos = 0;
! 688: continue;
! 689: case (ROFF_APPEND):
! 690: pos = strlen(ln.buf);
! 691: continue;
! 692: case (ROFF_RERUN):
! 693: goto rerun;
! 694: case (ROFF_IGN):
! 695: pos = 0;
1.30 schwarze 696: continue;
1.55 ! schwarze 697: case (ROFF_ERR):
1.45 schwarze 698: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 699: break;
1.55 ! schwarze 700: case (ROFF_SO):
! 701: if (pfile(ln.buf + of, curp)) {
! 702: pos = 0;
1.52 schwarze 703: continue;
1.55 ! schwarze 704: } else
1.52 schwarze 705: break;
1.55 ! schwarze 706: case (ROFF_CONT):
! 707: break;
1.45 schwarze 708: }
1.5 schwarze 709:
1.32 schwarze 710: /*
711: * If input parsers have not been allocated, do so now.
712: * We keep these instanced betwen parsers, but set them
713: * locally per parse routine since we can use different
714: * parsers with each one.
715: */
1.1 kristaps 716:
1.32 schwarze 717: if ( ! (man || mdoc))
1.45 schwarze 718: pset(ln.buf + of, pos - of, curp, &man, &mdoc);
1.5 schwarze 719:
1.32 schwarze 720: /* Lastly, push down into the parsers themselves. */
1.1 kristaps 721:
1.55 ! schwarze 722: if (man && ! man_parseln(man, curp->line, ln.buf, of)) {
1.45 schwarze 723: assert(MANDOCLEVEL_FATAL <= exit_status);
1.51 schwarze 724: break;
1.45 schwarze 725: }
1.55 ! schwarze 726: if (mdoc && ! mdoc_parseln(mdoc, curp->line, ln.buf, of)) {
1.45 schwarze 727: assert(MANDOCLEVEL_FATAL <= exit_status);
1.1 kristaps 728: break;
729: }
1.55 ! schwarze 730:
! 731: /* Temporary buffers typically are not full. */
! 732: if (0 == start && '\0' == blk.buf[i])
! 733: break;
! 734:
! 735: /* Start the next input line. */
! 736: pos = 0;
1.1 kristaps 737: }
738:
1.51 schwarze 739: free(ln.buf);
1.1 kristaps 740: }
741:
742:
1.45 schwarze 743: static void
1.1 kristaps 744: pset(const char *buf, int pos, struct curparse *curp,
745: struct man **man, struct mdoc **mdoc)
746: {
1.5 schwarze 747: int i;
1.1 kristaps 748:
749: /*
750: * Try to intuit which kind of manual parser should be used. If
751: * passed in by command-line (-man, -mdoc), then use that
752: * explicitly. If passed as -mandoc, then try to guess from the
1.5 schwarze 753: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.1 kristaps 754: * default to -man, which is more lenient.
755: */
756:
1.31 schwarze 757: if ('.' == buf[0] || '\'' == buf[0]) {
1.5 schwarze 758: for (i = 1; buf[i]; i++)
759: if (' ' != buf[i] && '\t' != buf[i])
760: break;
1.45 schwarze 761: if ('\0' == buf[i])
762: return;
1.5 schwarze 763: }
1.1 kristaps 764:
765: switch (curp->inttype) {
766: case (INTT_MDOC):
767: if (NULL == curp->mdoc)
1.45 schwarze 768: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
769: assert(curp->mdoc);
770: *mdoc = curp->mdoc;
771: return;
1.1 kristaps 772: case (INTT_MAN):
773: if (NULL == curp->man)
1.45 schwarze 774: curp->man = man_alloc(&curp->regs, curp, mmsg);
775: assert(curp->man);
776: *man = curp->man;
777: return;
1.1 kristaps 778: default:
779: break;
780: }
781:
782: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
783: if (NULL == curp->mdoc)
1.45 schwarze 784: curp->mdoc = mdoc_alloc(&curp->regs, curp, mmsg);
785: assert(curp->mdoc);
786: *mdoc = curp->mdoc;
787: return;
1.1 kristaps 788: }
789:
790: if (NULL == curp->man)
1.45 schwarze 791: curp->man = man_alloc(&curp->regs, curp, mmsg);
792: assert(curp->man);
793: *man = curp->man;
1.1 kristaps 794: }
795:
796:
797: static int
798: moptions(enum intt *tflags, char *arg)
799: {
800:
801: if (0 == strcmp(arg, "doc"))
802: *tflags = INTT_MDOC;
803: else if (0 == strcmp(arg, "andoc"))
804: *tflags = INTT_AUTO;
805: else if (0 == strcmp(arg, "an"))
806: *tflags = INTT_MAN;
807: else {
1.20 schwarze 808: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 809: return(0);
810: }
811:
812: return(1);
813: }
814:
815:
816: static int
1.22 schwarze 817: toptions(struct curparse *curp, char *arg)
1.1 kristaps 818: {
819:
820: if (0 == strcmp(arg, "ascii"))
1.22 schwarze 821: curp->outtype = OUTT_ASCII;
822: else if (0 == strcmp(arg, "lint")) {
823: curp->outtype = OUTT_LINT;
1.45 schwarze 824: curp->wlevel = MANDOCLEVEL_WARNING;
1.22 schwarze 825: }
1.1 kristaps 826: else if (0 == strcmp(arg, "tree"))
1.22 schwarze 827: curp->outtype = OUTT_TREE;
1.17 schwarze 828: else if (0 == strcmp(arg, "html"))
1.22 schwarze 829: curp->outtype = OUTT_HTML;
1.21 schwarze 830: else if (0 == strcmp(arg, "xhtml"))
1.22 schwarze 831: curp->outtype = OUTT_XHTML;
1.36 schwarze 832: else if (0 == strcmp(arg, "ps"))
833: curp->outtype = OUTT_PS;
1.43 schwarze 834: else if (0 == strcmp(arg, "pdf"))
835: curp->outtype = OUTT_PDF;
1.1 kristaps 836: else {
1.20 schwarze 837: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 838: return(0);
839: }
840:
841: return(1);
842: }
843:
844:
845: static int
1.45 schwarze 846: woptions(struct curparse *curp, char *arg)
1.1 kristaps 847: {
1.10 schwarze 848: char *v, *o;
1.45 schwarze 849: const char *toks[6];
1.1 kristaps 850:
1.45 schwarze 851: toks[0] = "stop";
852: toks[1] = "all";
853: toks[2] = "warning";
854: toks[3] = "error";
855: toks[4] = "fatal";
856: toks[5] = NULL;
1.1 kristaps 857:
1.10 schwarze 858: while (*arg) {
859: o = arg;
1.17 schwarze 860: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 861: case (0):
1.45 schwarze 862: curp->wstop = 1;
1.1 kristaps 863: break;
864: case (1):
1.45 schwarze 865: /* FALLTHROUGH */
1.1 kristaps 866: case (2):
1.45 schwarze 867: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 868: break;
869: case (3):
1.45 schwarze 870: curp->wlevel = MANDOCLEVEL_ERROR;
1.1 kristaps 871: break;
872: case (4):
1.45 schwarze 873: curp->wlevel = MANDOCLEVEL_FATAL;
1.19 schwarze 874: break;
1.1 kristaps 875: default:
1.45 schwarze 876: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 877: return(0);
878: }
1.10 schwarze 879: }
1.1 kristaps 880:
881: return(1);
882: }
883:
884:
1.30 schwarze 885: static int
886: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
887: {
888: struct curparse *cp;
1.45 schwarze 889: enum mandoclevel level;
890:
891: level = MANDOCLEVEL_FATAL;
892: while (t < mandoclimits[level])
1.47 schwarze 893: /* LINTED */
1.45 schwarze 894: level--;
1.30 schwarze 895:
896: cp = (struct curparse *)arg;
1.45 schwarze 897: if (level < cp->wlevel)
898: return(1);
1.30 schwarze 899:
1.45 schwarze 900: fprintf(stderr, "%s:%d:%d: %s: %s",
901: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.30 schwarze 902: if (msg)
903: fprintf(stderr, ": %s", msg);
904: fputc('\n', stderr);
1.33 schwarze 905:
1.45 schwarze 906: if (exit_status < level)
907: exit_status = level;
908:
909: return(level < MANDOCLEVEL_FATAL);
1.30 schwarze 910: }