Annotation of mandoc/main.c, Revision 1.126
1.126 ! kristaps 1: /* $Id: main.c,v 1.125 2010/12/22 11:38:17 kristaps Exp $ */
1.1 kristaps 2: /*
1.97 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
1.25 kristaps 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: *
1.25 kristaps 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: */
1.58 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.66 kristaps 22: #include <sys/mman.h>
1.1 kristaps 23: #include <sys/stat.h>
24:
25: #include <assert.h>
1.99 kristaps 26: #include <ctype.h>
1.1 kristaps 27: #include <fcntl.h>
28: #include <stdio.h>
1.45 kristaps 29: #include <stdint.h>
1.1 kristaps 30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
33:
1.71 kristaps 34: #include "mandoc.h"
1.90 kristaps 35: #include "main.h"
1.1 kristaps 36: #include "mdoc.h"
1.10 kristaps 37: #include "man.h"
1.71 kristaps 38: #include "roff.h"
1.101 joerg 39:
40: #ifndef MAP_FILE
41: #define MAP_FILE 0
42: #endif
1.1 kristaps 43:
1.122 schwarze 44: #define REPARSE_LIMIT 1000
1.45 kristaps 45: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
46:
1.55 kristaps 47: /* FIXME: Intel's compiler? LLVM? pcc? */
48:
49: #if !defined(__GNUC__) || (__GNUC__ < 2)
1.56 kristaps 50: # if !defined(lint)
51: # define __attribute__(x)
52: # endif
1.55 kristaps 53: #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
1.16 kristaps 54:
1.42 kristaps 55: typedef void (*out_mdoc)(void *, const struct mdoc *);
56: typedef void (*out_man)(void *, const struct man *);
1.22 kristaps 57: typedef void (*out_free)(void *);
58:
1.5 kristaps 59: struct buf {
60: char *buf;
61: size_t sz;
62: };
63:
1.19 kristaps 64: enum intt {
65: INTT_AUTO,
66: INTT_MDOC,
67: INTT_MAN
68: };
69:
70: enum outt {
71: OUTT_ASCII = 0,
72: OUTT_TREE,
1.43 kristaps 73: OUTT_HTML,
1.59 kristaps 74: OUTT_XHTML,
1.85 kristaps 75: OUTT_LINT,
1.100 kristaps 76: OUTT_PS,
77: OUTT_PDF
1.19 kristaps 78: };
79:
1.5 kristaps 80: struct curparse {
1.22 kristaps 81: const char *file; /* Current parse. */
82: int fd; /* Current parse. */
1.111 kristaps 83: int line; /* Line number in the file. */
1.103 schwarze 84: enum mandoclevel wlevel; /* Ignore messages below this. */
85: int wstop; /* Stop after a file with a warning. */
1.79 kristaps 86: enum intt inttype; /* which parser to use */
1.112 kristaps 87: struct man *pman; /* persistent man parser */
88: struct mdoc *pmdoc; /* persistent mdoc parser */
1.79 kristaps 89: struct man *man; /* man parser */
90: struct mdoc *mdoc; /* mdoc parser */
91: struct roff *roff; /* roff parser (!NULL) */
1.92 kristaps 92: struct regset regs; /* roff registers */
1.122 schwarze 93: int reparse_count; /* finite interpolation stack */
1.79 kristaps 94: enum outt outtype; /* which output to use */
95: out_mdoc outmdoc; /* mdoc output ptr */
96: out_man outman; /* man output ptr */
97: out_free outfree; /* free output ptr */
98: void *outdata; /* data for output */
99: char outopts[BUFSIZ]; /* buf of output opts */
100: };
101:
1.103 schwarze 102: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
103: "SUCCESS",
104: "RESERVED",
105: "WARNING",
106: "ERROR",
107: "FATAL",
108: "BADARG",
109: "SYSERR"
110: };
111:
112: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
113: MANDOCERR_OK,
114: MANDOCERR_WARNING,
115: MANDOCERR_WARNING,
116: MANDOCERR_ERROR,
117: MANDOCERR_FATAL,
118: MANDOCERR_MAX,
119: MANDOCERR_MAX
120: };
121:
1.79 kristaps 122: static const char * const mandocerrs[MANDOCERR_MAX] = {
123: "ok",
1.94 schwarze 124:
125: "generic warning",
126:
1.121 kristaps 127: /* related to the prologue */
128: "no title in document",
129: "document title should be all caps",
130: "unknown manual section",
131: "cannot parse date argument",
132: "prologue macros out of order",
133: "duplicate prologue macro",
134: "macro not allowed in prologue",
135: "macro not allowed in body",
136:
137: /* related to document structure */
1.113 kristaps 138: ".so is fragile, better use ln(1)",
1.121 kristaps 139: "NAME section must come first",
140: "bad NAME section contents",
141: "manual name not yet set",
1.81 kristaps 142: "sections out of conventional order",
1.121 kristaps 143: "duplicate section name",
1.79 kristaps 144: "section not in conventional manual section",
1.121 kristaps 145:
146: /* related to macros and nesting */
147: "skipping obsolete macro",
1.125 kristaps 148: "skipping paragraph macro",
1.95 schwarze 149: "blocks badly nested",
1.121 kristaps 150: "child violates parent syntax",
151: "nested displays are not portable",
152: "already in literal mode",
1.94 schwarze 153:
1.121 kristaps 154: /* related to missing macro arguments */
155: "skipping empty macro",
156: "missing display type",
157: "list type must come first",
158: "tag lists require a width argument",
159: "missing font type",
1.94 schwarze 160:
1.121 kristaps 161: /* related to bad macro arguments */
162: "skipping argument",
163: "duplicate argument",
164: "duplicate display type",
165: "duplicate list type",
166: "unknown AT&T UNIX version",
1.79 kristaps 167: "bad Boolean value",
1.120 kristaps 168: "unknown font",
1.121 kristaps 169: "unknown standard specifier",
170: "bad width argument",
171:
172: /* related to plain text */
173: "blank line in non-literal context",
174: "tab in non-literal context",
175: "end of line whitespace",
1.79 kristaps 176: "bad comment style",
1.121 kristaps 177: "unknown escape sequence",
178: "unterminated quoted string",
179:
180: "generic error",
181:
1.126 ! kristaps 182: "bad table syntax",
! 183: "bad table option",
1.122 schwarze 184: "input stack limit exceeded, infinite loop?",
1.121 kristaps 185: "skipping bad character",
186: "skipping text before the first section header",
187: "skipping unknown macro",
1.110 kristaps 188: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 189: "line scope broken",
190: "argument count wrong",
1.121 kristaps 191: "skipping end of block that is not open",
192: "missing end of block",
1.106 schwarze 193: "scope open on exit",
1.117 kristaps 194: "uname(3) system call failed",
1.79 kristaps 195: "macro requires line argument(s)",
196: "macro requires body argument(s)",
197: "macro requires argument(s)",
1.82 kristaps 198: "missing list type",
1.79 kristaps 199: "line argument(s) will be lost",
200: "body argument(s) will be lost",
1.94 schwarze 201:
202: "generic fatal error",
203:
1.80 kristaps 204: "column syntax is inconsistent",
1.121 kristaps 205: "NOT IMPLEMENTED: .Bd -file",
1.79 kristaps 206: "line scope broken, syntax violated",
207: "argument count wrong, violates syntax",
208: "child violates parent syntax",
209: "argument count wrong, violates syntax",
1.113 kristaps 210: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 211: "no document body",
212: "no document prologue",
1.103 schwarze 213: "static buffer exhausted",
1.5 kristaps 214: };
1.1 kristaps 215:
1.111 kristaps 216: static void parsebuf(struct curparse *, struct buf, int);
217: static void pdesc(struct curparse *);
1.66 kristaps 218: static void fdesc(struct curparse *);
219: static void ffile(const char *, struct curparse *);
1.111 kristaps 220: static int pfile(const char *, struct curparse *);
1.10 kristaps 221: static int moptions(enum intt *, char *);
1.71 kristaps 222: static int mmsg(enum mandocerr, void *,
223: int, int, const char *);
1.112 kristaps 224: static void pset(const char *, int, struct curparse *);
1.66 kristaps 225: static int toptions(struct curparse *, char *);
226: static void usage(void) __attribute__((noreturn));
1.55 kristaps 227: static void version(void) __attribute__((noreturn));
1.103 schwarze 228: static int woptions(struct curparse *, char *);
1.1 kristaps 229:
1.54 kristaps 230: static const char *progname;
1.115 schwarze 231: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 232: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 233:
234: int
235: main(int argc, char *argv[])
236: {
1.64 kristaps 237: int c;
1.5 kristaps 238: struct curparse curp;
1.1 kristaps 239:
1.54 kristaps 240: progname = strrchr(argv[0], '/');
241: if (progname == NULL)
242: progname = argv[0];
243: else
244: ++progname;
245:
1.51 kristaps 246: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 247:
1.19 kristaps 248: curp.inttype = INTT_AUTO;
1.22 kristaps 249: curp.outtype = OUTT_ASCII;
1.103 schwarze 250: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 251:
1.1 kristaps 252: /* LINTED */
1.103 schwarze 253: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 254: switch (c) {
1.10 kristaps 255: case ('m'):
1.19 kristaps 256: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 257: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 258: break;
1.48 kristaps 259: case ('O'):
1.49 kristaps 260: (void)strlcat(curp.outopts, optarg, BUFSIZ);
261: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 262: break;
1.1 kristaps 263: case ('T'):
1.60 kristaps 264: if ( ! toptions(&curp, optarg))
1.105 kristaps 265: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 266: break;
267: case ('W'):
1.103 schwarze 268: if ( ! woptions(&curp, optarg))
1.105 kristaps 269: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 270: break;
271: case ('V'):
272: version();
273: /* NOTREACHED */
274: default:
275: usage();
276: /* NOTREACHED */
277: }
278:
279: argc -= optind;
280: argv += optind;
281:
1.30 kristaps 282: if (NULL == *argv) {
283: curp.file = "<stdin>";
284: curp.fd = STDIN_FILENO;
1.39 kristaps 285:
1.66 kristaps 286: fdesc(&curp);
1.64 kristaps 287: }
288:
289: while (*argv) {
1.66 kristaps 290: ffile(*argv, &curp);
1.103 schwarze 291: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 292: break;
293: ++argv;
1.1 kristaps 294: }
295:
1.22 kristaps 296: if (curp.outfree)
297: (*curp.outfree)(curp.outdata);
1.112 kristaps 298: if (curp.pmdoc)
299: mdoc_free(curp.pmdoc);
300: if (curp.pman)
301: man_free(curp.pman);
1.71 kristaps 302: if (curp.roff)
303: roff_free(curp.roff);
1.1 kristaps 304:
1.105 kristaps 305: return((int)exit_status);
1.1 kristaps 306: }
307:
308:
1.55 kristaps 309: static void
1.1 kristaps 310: version(void)
311: {
312:
1.54 kristaps 313: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 314: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 315: }
316:
317:
1.55 kristaps 318: static void
1.1 kristaps 319: usage(void)
320: {
321:
1.112 kristaps 322: (void)fprintf(stderr, "usage: %s "
323: "[-V] "
324: "[-foption] "
325: "[-mformat] "
326: "[-Ooption] "
327: "[-Toutput] "
328: "[-Werr] "
329: "[file...]\n",
330: progname);
331:
1.105 kristaps 332: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 333: }
334:
1.64 kristaps 335: static void
1.66 kristaps 336: ffile(const char *file, struct curparse *curp)
1.1 kristaps 337: {
338:
1.112 kristaps 339: /*
340: * Called once per input file. Get the file ready for reading,
341: * pass it through to the parser-driver, then close it out.
342: * XXX: don't do anything special as this is only called for
343: * files; stdin goes directly to fdesc().
344: */
345:
1.19 kristaps 346: curp->file = file;
1.112 kristaps 347:
1.19 kristaps 348: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 349: perror(curp->file);
1.103 schwarze 350: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 351: return;
1.1 kristaps 352: }
353:
1.66 kristaps 354: fdesc(curp);
1.1 kristaps 355:
1.19 kristaps 356: if (-1 == close(curp->fd))
1.53 kristaps 357: perror(curp->file);
1.1 kristaps 358: }
359:
1.111 kristaps 360: static int
361: pfile(const char *file, struct curparse *curp)
362: {
363: const char *savefile;
364: int fd, savefd;
365:
366: if (-1 == (fd = open(file, O_RDONLY, 0))) {
367: perror(file);
1.115 schwarze 368: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 369: return(0);
370: }
371:
372: savefile = curp->file;
373: savefd = curp->fd;
374:
375: curp->file = file;
376: curp->fd = fd;
377:
378: pdesc(curp);
379:
380: curp->file = savefile;
381: curp->fd = savefd;
382:
383: if (-1 == close(fd))
384: perror(file);
385:
1.115 schwarze 386: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 387: }
388:
1.1 kristaps 389:
1.103 schwarze 390: static void
1.68 joerg 391: resize_buf(struct buf *buf, size_t initial)
392: {
393:
1.124 schwarze 394: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.103 schwarze 395: buf->buf = realloc(buf->buf, buf->sz);
396: if (NULL == buf->buf) {
1.68 joerg 397: perror(NULL);
1.105 kristaps 398: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 399: }
400: }
401:
402:
403: static int
1.66 kristaps 404: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
405: {
406: struct stat st;
1.68 joerg 407: size_t off;
1.66 kristaps 408: ssize_t ssz;
409:
410: if (-1 == fstat(curp->fd, &st)) {
411: perror(curp->file);
412: return(0);
413: }
414:
415: /*
416: * If we're a regular file, try just reading in the whole entry
417: * via mmap(). This is faster than reading it into blocks, and
418: * since each file is only a few bytes to begin with, I'm not
419: * concerned that this is going to tank any machines.
420: */
421:
422: if (S_ISREG(st.st_mode)) {
423: if (st.st_size >= (1U << 31)) {
424: fprintf(stderr, "%s: input too large\n",
425: curp->file);
426: return(0);
427: }
428: *with_mmap = 1;
1.71 kristaps 429: fb->sz = (size_t)st.st_size;
1.66 kristaps 430: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 431: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 432: if (fb->buf != MAP_FAILED)
433: return(1);
434: }
435:
436: /*
437: * If this isn't a regular file (like, say, stdin), then we must
438: * go the old way and just read things in bit by bit.
439: */
440:
441: *with_mmap = 0;
442: off = 0;
443: fb->sz = 0;
444: fb->buf = NULL;
445: for (;;) {
446: if (off == fb->sz) {
447: if (fb->sz == (1U << 31)) {
448: fprintf(stderr, "%s: input too large\n",
449: curp->file);
450: break;
451: }
1.103 schwarze 452: resize_buf(fb, 65536);
1.66 kristaps 453: }
1.71 kristaps 454: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 455: if (ssz == 0) {
456: fb->sz = off;
457: return(1);
458: }
459: if (ssz == -1) {
460: perror(curp->file);
461: break;
462: }
1.71 kristaps 463: off += (size_t)ssz;
1.66 kristaps 464: }
465:
466: free(fb->buf);
467: fb->buf = NULL;
468: return(0);
469: }
470:
471:
1.64 kristaps 472: static void
1.66 kristaps 473: fdesc(struct curparse *curp)
1.1 kristaps 474: {
1.112 kristaps 475:
476: /*
477: * Called once per file with an opened file descriptor. All
478: * pre-file-parse operations (whether stdin or a file) should go
479: * here.
480: *
481: * This calls down into the nested parser, which drills down and
482: * fully parses a file and all its dependences (i.e., `so'). It
483: * then runs the cleanup validators and pushes to output.
484: */
485:
486: /* Zero the parse type. */
487:
488: curp->mdoc = NULL;
489: curp->man = NULL;
1.115 schwarze 490: file_status = MANDOCLEVEL_OK;
1.112 kristaps 491:
492: /* Make sure the mandotory roff parser is initialised. */
493:
494: if (NULL == curp->roff) {
495: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
496: assert(curp->roff);
497: }
498:
499: /* Fully parse the file. */
1.10 kristaps 500:
1.111 kristaps 501: pdesc(curp);
1.92 kristaps 502:
1.115 schwarze 503: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 504: goto cleanup;
505:
506: /* NOTE a parser may not have been assigned, yet. */
507:
1.112 kristaps 508: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 509: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 510: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 511: goto cleanup;
512: }
513:
514: /* Clean up the parse routine ASTs. */
515:
1.112 kristaps 516: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 517: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 518: goto cleanup;
519: }
1.112 kristaps 520:
521: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 522: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 523: goto cleanup;
524: }
1.112 kristaps 525:
526: assert(curp->roff);
527: if ( ! roff_endparse(curp->roff)) {
1.115 schwarze 528: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 529: goto cleanup;
530: }
1.1 kristaps 531:
532: /*
1.111 kristaps 533: * With -Wstop and warnings or errors of at least
534: * the requested level, do not produce output.
1.1 kristaps 535: */
536:
1.115 schwarze 537: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 538: goto cleanup;
539:
540: /* If unset, allocate output dev now (if applicable). */
541:
542: if ( ! (curp->outman && curp->outmdoc)) {
543: switch (curp->outtype) {
544: case (OUTT_XHTML):
545: curp->outdata = xhtml_alloc(curp->outopts);
546: break;
547: case (OUTT_HTML):
548: curp->outdata = html_alloc(curp->outopts);
549: break;
550: case (OUTT_ASCII):
551: curp->outdata = ascii_alloc(curp->outopts);
552: curp->outfree = ascii_free;
553: break;
554: case (OUTT_PDF):
555: curp->outdata = pdf_alloc(curp->outopts);
556: curp->outfree = pspdf_free;
557: break;
558: case (OUTT_PS):
559: curp->outdata = ps_alloc(curp->outopts);
560: curp->outfree = pspdf_free;
561: break;
562: default:
563: break;
564: }
565:
566: switch (curp->outtype) {
567: case (OUTT_HTML):
568: /* FALLTHROUGH */
569: case (OUTT_XHTML):
570: curp->outman = html_man;
571: curp->outmdoc = html_mdoc;
572: curp->outfree = html_free;
573: break;
574: case (OUTT_TREE):
575: curp->outman = tree_man;
576: curp->outmdoc = tree_mdoc;
577: break;
578: case (OUTT_PDF):
579: /* FALLTHROUGH */
580: case (OUTT_ASCII):
581: /* FALLTHROUGH */
582: case (OUTT_PS):
583: curp->outman = terminal_man;
584: curp->outmdoc = terminal_mdoc;
585: break;
586: default:
587: break;
588: }
589: }
590:
591: /* Execute the out device, if it exists. */
592:
1.112 kristaps 593: if (curp->man && curp->outman)
594: (*curp->outman)(curp->outdata, curp->man);
595: if (curp->mdoc && curp->outmdoc)
596: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 597:
598: cleanup:
1.112 kristaps 599:
1.111 kristaps 600: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 601:
602: /* Reset the current-parse compilers. */
603:
604: if (curp->mdoc)
605: mdoc_reset(curp->mdoc);
606: if (curp->man)
607: man_reset(curp->man);
608:
609: assert(curp->roff);
610: roff_reset(curp->roff);
1.111 kristaps 611:
1.115 schwarze 612: if (exit_status < file_status)
613: exit_status = file_status;
614:
1.111 kristaps 615: return;
616: }
617:
618: static void
619: pdesc(struct curparse *curp)
620: {
621: struct buf blk;
622: int with_mmap;
623:
1.112 kristaps 624: /*
625: * Run for each opened file; may be called more than once for
626: * each full parse sequence if the opened file is nested (i.e.,
627: * from `so'). Simply sucks in the whole file and moves into
628: * the parse phase for the file.
629: */
630:
1.103 schwarze 631: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 632: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 633: return;
1.103 schwarze 634: }
1.1 kristaps 635:
1.112 kristaps 636: /* Line number is per-file. */
1.111 kristaps 637:
638: curp->line = 1;
1.112 kristaps 639:
1.111 kristaps 640: parsebuf(curp, blk, 1);
641:
642: if (with_mmap)
643: munmap(blk.buf, blk.sz);
644: else
645: free(blk.buf);
646: }
647:
648: static void
649: parsebuf(struct curparse *curp, struct buf blk, int start)
650: {
651: struct buf ln;
1.114 kristaps 652: enum rofferr rr;
1.112 kristaps 653: int i, of, rc;
654: int pos; /* byte number in the ln buffer */
655: int lnn; /* line number in the real file */
1.111 kristaps 656: unsigned char c;
1.112 kristaps 657:
658: /*
659: * Main parse routine for an opened file. This is called for
660: * each opened file and simply loops around the full input file,
661: * possibly nesting (i.e., with `so').
662: */
1.71 kristaps 663:
1.111 kristaps 664: memset(&ln, 0, sizeof(struct buf));
665:
1.112 kristaps 666: lnn = curp->line;
667: pos = 0;
1.111 kristaps 668:
1.112 kristaps 669: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 670: if (0 == pos && '\0' == blk.buf[i])
671: break;
1.112 kristaps 672:
1.122 schwarze 673: if (start) {
1.111 kristaps 674: curp->line = lnn;
1.122 schwarze 675: curp->reparse_count = 0;
676: }
1.111 kristaps 677:
678: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 679: if ('\n' == blk.buf[i]) {
680: ++i;
681: ++lnn;
682: break;
683: }
1.99 kristaps 684:
685: /*
686: * Warn about bogus characters. If you're using
687: * non-ASCII encoding, you're screwing your
688: * readers. Since I'd rather this not happen,
689: * I'll be helpful and drop these characters so
690: * we don't display gibberish. Note to manual
691: * writers: use special characters.
692: */
693:
1.102 schwarze 694: c = (unsigned char) blk.buf[i];
1.112 kristaps 695:
696: if ( ! (isascii(c) &&
697: (isgraph(c) || isblank(c)))) {
1.103 schwarze 698: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 699: curp->line, pos, "ignoring byte");
1.99 kristaps 700: i++;
701: continue;
702: }
703:
1.112 kristaps 704: /* Trailing backslash = a plain char. */
705:
1.70 joerg 706: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
707: if (pos >= (int)ln.sz)
1.103 schwarze 708: resize_buf(&ln, 256);
1.70 joerg 709: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 710: continue;
1.70 joerg 711: }
1.112 kristaps 712:
713: /* Found escape & at least one other char. */
714:
1.70 joerg 715: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 716: i += 2;
1.70 joerg 717: /* Escaped newlines are skipped over */
718: ++lnn;
1.67 joerg 719: continue;
720: }
1.112 kristaps 721:
1.70 joerg 722: if ('"' == blk.buf[i + 1]) {
723: i += 2;
724: /* Comment, skip to end of line */
725: for (; i < (int)blk.sz; ++i) {
726: if ('\n' == blk.buf[i]) {
727: ++i;
728: ++lnn;
729: break;
730: }
731: }
1.112 kristaps 732:
1.70 joerg 733: /* Backout trailing whitespaces */
734: for (; pos > 0; --pos) {
735: if (ln.buf[pos - 1] != ' ')
736: break;
737: if (pos > 2 && ln.buf[pos - 2] == '\\')
738: break;
739: }
740: break;
741: }
1.112 kristaps 742:
743: /* Some other escape sequence, copy & cont. */
744:
1.70 joerg 745: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 746: resize_buf(&ln, 256);
1.1 kristaps 747:
1.70 joerg 748: ln.buf[pos++] = blk.buf[i++];
749: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 750: }
1.1 kristaps 751:
1.70 joerg 752: if (pos >= (int)ln.sz)
1.103 schwarze 753: resize_buf(&ln, 256);
1.112 kristaps 754:
1.71 kristaps 755: ln.buf[pos] = '\0';
756:
1.76 kristaps 757: /*
758: * A significant amount of complexity is contained by
759: * the roff preprocessor. It's line-oriented but can be
760: * expressed on one line, so we need at times to
761: * readjust our starting point and re-run it. The roff
762: * preprocessor can also readjust the buffers with new
763: * data, so we pass them in wholesale.
764: */
765:
766: of = 0;
1.112 kristaps 767:
1.111 kristaps 768: rerun:
1.114 kristaps 769: rr = roff_parseln
1.112 kristaps 770: (curp->roff, curp->line,
771: &ln.buf, &ln.sz, of, &of);
772:
1.114 kristaps 773: switch (rr) {
1.111 kristaps 774: case (ROFF_REPARSE):
1.122 schwarze 775: if (REPARSE_LIMIT >= ++curp->reparse_count)
776: parsebuf(curp, ln, 0);
777: else
778: mmsg(MANDOCERR_ROFFLOOP, curp,
779: curp->line, pos, NULL);
1.111 kristaps 780: pos = 0;
781: continue;
782: case (ROFF_APPEND):
783: pos = strlen(ln.buf);
784: continue;
785: case (ROFF_RERUN):
786: goto rerun;
787: case (ROFF_IGN):
788: pos = 0;
1.71 kristaps 789: continue;
1.111 kristaps 790: case (ROFF_ERR):
1.115 schwarze 791: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 792: break;
793: case (ROFF_SO):
794: if (pfile(ln.buf + of, curp)) {
795: pos = 0;
796: continue;
797: } else
798: break;
799: case (ROFF_CONT):
800: break;
1.103 schwarze 801: }
1.29 kristaps 802:
1.76 kristaps 803: /*
804: * If input parsers have not been allocated, do so now.
805: * We keep these instanced betwen parsers, but set them
806: * locally per parse routine since we can use different
807: * parsers with each one.
808: */
1.19 kristaps 809:
1.112 kristaps 810: if ( ! (curp->man || curp->mdoc))
811: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 812:
1.112 kristaps 813: /*
814: * Lastly, push down into the parsers themselves. One
815: * of these will have already been set in the pset()
816: * routine.
817: */
818:
819: if (curp->man || curp->mdoc) {
820: rc = curp->man ?
821: man_parseln(curp->man,
822: curp->line, ln.buf, of) :
823: mdoc_parseln(curp->mdoc,
824: curp->line, ln.buf, of);
1.30 kristaps 825:
1.112 kristaps 826: if ( ! rc) {
1.115 schwarze 827: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 828: break;
829: }
1.103 schwarze 830: }
1.19 kristaps 831:
1.111 kristaps 832: /* Temporary buffers typically are not full. */
1.112 kristaps 833:
1.111 kristaps 834: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 835: break;
1.85 kristaps 836:
1.111 kristaps 837: /* Start the next input line. */
1.112 kristaps 838:
1.111 kristaps 839: pos = 0;
1.22 kristaps 840: }
841:
1.111 kristaps 842: free(ln.buf);
1.19 kristaps 843: }
844:
1.103 schwarze 845: static void
1.112 kristaps 846: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 847: {
1.29 kristaps 848: int i;
1.19 kristaps 849:
850: /*
851: * Try to intuit which kind of manual parser should be used. If
852: * passed in by command-line (-man, -mdoc), then use that
853: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 854: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 855: * default to -man, which is more lenient.
1.112 kristaps 856: *
857: * Separate out pmdoc/pman from mdoc/man: the first persists
858: * through all parsers, while the latter is used per-parse.
1.19 kristaps 859: */
860:
1.75 kristaps 861: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 862: for (i = 1; buf[i]; i++)
863: if (' ' != buf[i] && '\t' != buf[i])
864: break;
1.103 schwarze 865: if ('\0' == buf[i])
866: return;
1.29 kristaps 867: }
1.10 kristaps 868:
1.19 kristaps 869: switch (curp->inttype) {
870: case (INTT_MDOC):
1.112 kristaps 871: if (NULL == curp->pmdoc)
872: curp->pmdoc = mdoc_alloc
873: (&curp->regs, curp, mmsg);
874: assert(curp->pmdoc);
875: curp->mdoc = curp->pmdoc;
1.103 schwarze 876: return;
1.19 kristaps 877: case (INTT_MAN):
1.112 kristaps 878: if (NULL == curp->pman)
879: curp->pman = man_alloc
880: (&curp->regs, curp, mmsg);
881: assert(curp->pman);
882: curp->man = curp->pman;
1.103 schwarze 883: return;
1.19 kristaps 884: default:
885: break;
886: }
887:
888: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 889: if (NULL == curp->pmdoc)
890: curp->pmdoc = mdoc_alloc
891: (&curp->regs, curp, mmsg);
892: assert(curp->pmdoc);
893: curp->mdoc = curp->pmdoc;
1.103 schwarze 894: return;
1.19 kristaps 895: }
896:
1.112 kristaps 897: if (NULL == curp->pman)
898: curp->pman = man_alloc(&curp->regs, curp, mmsg);
899: assert(curp->pman);
900: curp->man = curp->pman;
1.10 kristaps 901: }
902:
903: static int
904: moptions(enum intt *tflags, char *arg)
905: {
906:
1.17 kristaps 907: if (0 == strcmp(arg, "doc"))
1.10 kristaps 908: *tflags = INTT_MDOC;
1.19 kristaps 909: else if (0 == strcmp(arg, "andoc"))
910: *tflags = INTT_AUTO;
1.17 kristaps 911: else if (0 == strcmp(arg, "an"))
1.10 kristaps 912: *tflags = INTT_MAN;
913: else {
1.57 kristaps 914: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 915: return(0);
916: }
917:
918: return(1);
1.1 kristaps 919: }
920:
921: static int
1.60 kristaps 922: toptions(struct curparse *curp, char *arg)
1.1 kristaps 923: {
924:
925: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 926: curp->outtype = OUTT_ASCII;
927: else if (0 == strcmp(arg, "lint")) {
928: curp->outtype = OUTT_LINT;
1.103 schwarze 929: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 930: }
1.1 kristaps 931: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 932: curp->outtype = OUTT_TREE;
1.43 kristaps 933: else if (0 == strcmp(arg, "html"))
1.60 kristaps 934: curp->outtype = OUTT_HTML;
1.59 kristaps 935: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 936: curp->outtype = OUTT_XHTML;
1.85 kristaps 937: else if (0 == strcmp(arg, "ps"))
938: curp->outtype = OUTT_PS;
1.100 kristaps 939: else if (0 == strcmp(arg, "pdf"))
940: curp->outtype = OUTT_PDF;
1.1 kristaps 941: else {
1.57 kristaps 942: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 943: return(0);
944: }
945:
946: return(1);
947: }
948:
949: static int
1.103 schwarze 950: woptions(struct curparse *curp, char *arg)
1.1 kristaps 951: {
1.34 kristaps 952: char *v, *o;
1.103 schwarze 953: const char *toks[6];
1.1 kristaps 954:
1.103 schwarze 955: toks[0] = "stop";
956: toks[1] = "all";
957: toks[2] = "warning";
958: toks[3] = "error";
959: toks[4] = "fatal";
960: toks[5] = NULL;
1.1 kristaps 961:
1.34 kristaps 962: while (*arg) {
963: o = arg;
1.45 kristaps 964: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 965: case (0):
1.103 schwarze 966: curp->wstop = 1;
1.1 kristaps 967: break;
968: case (1):
1.103 schwarze 969: /* FALLTHROUGH */
1.1 kristaps 970: case (2):
1.103 schwarze 971: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 972: break;
1.20 kristaps 973: case (3):
1.103 schwarze 974: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 975: break;
976: case (4):
1.103 schwarze 977: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 978: break;
1.1 kristaps 979: default:
1.103 schwarze 980: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 981: return(0);
982: }
1.34 kristaps 983: }
1.1 kristaps 984:
985: return(1);
986: }
987:
1.71 kristaps 988: static int
989: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
990: {
991: struct curparse *cp;
1.103 schwarze 992: enum mandoclevel level;
993:
994: level = MANDOCLEVEL_FATAL;
995: while (t < mandoclimits[level])
1.105 kristaps 996: /* LINTED */
1.103 schwarze 997: level--;
1.71 kristaps 998:
999: cp = (struct curparse *)arg;
1.103 schwarze 1000: if (level < cp->wlevel)
1001: return(1);
1.73 kristaps 1002:
1.103 schwarze 1003: fprintf(stderr, "%s:%d:%d: %s: %s",
1004: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 1005: if (msg)
1006: fprintf(stderr, ": %s", msg);
1007: fputc('\n', stderr);
1.79 kristaps 1008:
1.115 schwarze 1009: if (file_status < level)
1010: file_status = level;
1.103 schwarze 1011:
1012: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 1013: }
CVSweb