Annotation of mandoc/main.c, Revision 1.137
1.137 ! schwarze 1: /* $Id: main.c,v 1.136 2011/01/10 14:40:30 kristaps Exp $ */
1.1 kristaps 2: /*
1.133 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.97 schwarze 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",
1.133 schwarze 156: "argument count wrong",
1.121 kristaps 157: "missing display type",
158: "list type must come first",
159: "tag lists require a width argument",
160: "missing font type",
1.94 schwarze 161:
1.121 kristaps 162: /* related to bad macro arguments */
163: "skipping argument",
164: "duplicate argument",
165: "duplicate display type",
166: "duplicate list type",
167: "unknown AT&T UNIX version",
1.79 kristaps 168: "bad Boolean value",
1.120 kristaps 169: "unknown font",
1.121 kristaps 170: "unknown standard specifier",
171: "bad width argument",
172:
173: /* related to plain text */
174: "blank line in non-literal context",
175: "tab in non-literal context",
176: "end of line whitespace",
1.79 kristaps 177: "bad comment style",
1.121 kristaps 178: "unknown escape sequence",
179: "unterminated quoted string",
1.131 kristaps 180:
1.121 kristaps 181: "generic error",
182:
1.131 kristaps 183: /* related to tables */
1.126 kristaps 184: "bad table syntax",
185: "bad table option",
1.127 kristaps 186: "bad table layout",
187: "no table layout cells specified",
1.129 kristaps 188: "no table data cells specified",
1.134 kristaps 189: "ignore data in cell",
1.135 kristaps 190: "data block still open",
1.136 kristaps 191: "ignoring extra data cells",
1.134 kristaps 192:
1.122 schwarze 193: "input stack limit exceeded, infinite loop?",
1.121 kristaps 194: "skipping bad character",
1.137 ! schwarze 195: "escaped character not allowed in a name",
1.121 kristaps 196: "skipping text before the first section header",
197: "skipping unknown macro",
1.110 kristaps 198: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 199: "line scope broken",
200: "argument count wrong",
1.121 kristaps 201: "skipping end of block that is not open",
202: "missing end of block",
1.106 schwarze 203: "scope open on exit",
1.117 kristaps 204: "uname(3) system call failed",
1.79 kristaps 205: "macro requires line argument(s)",
206: "macro requires body argument(s)",
207: "macro requires argument(s)",
1.82 kristaps 208: "missing list type",
1.79 kristaps 209: "line argument(s) will be lost",
210: "body argument(s) will be lost",
1.94 schwarze 211:
212: "generic fatal error",
213:
1.80 kristaps 214: "column syntax is inconsistent",
1.121 kristaps 215: "NOT IMPLEMENTED: .Bd -file",
1.79 kristaps 216: "line scope broken, syntax violated",
217: "argument count wrong, violates syntax",
218: "child violates parent syntax",
219: "argument count wrong, violates syntax",
1.113 kristaps 220: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 221: "no document body",
222: "no document prologue",
1.103 schwarze 223: "static buffer exhausted",
1.5 kristaps 224: };
1.1 kristaps 225:
1.111 kristaps 226: static void parsebuf(struct curparse *, struct buf, int);
227: static void pdesc(struct curparse *);
1.66 kristaps 228: static void fdesc(struct curparse *);
229: static void ffile(const char *, struct curparse *);
1.111 kristaps 230: static int pfile(const char *, struct curparse *);
1.10 kristaps 231: static int moptions(enum intt *, char *);
1.71 kristaps 232: static int mmsg(enum mandocerr, void *,
233: int, int, const char *);
1.112 kristaps 234: static void pset(const char *, int, struct curparse *);
1.66 kristaps 235: static int toptions(struct curparse *, char *);
236: static void usage(void) __attribute__((noreturn));
1.55 kristaps 237: static void version(void) __attribute__((noreturn));
1.103 schwarze 238: static int woptions(struct curparse *, char *);
1.1 kristaps 239:
1.54 kristaps 240: static const char *progname;
1.115 schwarze 241: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 242: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 243:
244: int
245: main(int argc, char *argv[])
246: {
1.64 kristaps 247: int c;
1.5 kristaps 248: struct curparse curp;
1.1 kristaps 249:
1.54 kristaps 250: progname = strrchr(argv[0], '/');
251: if (progname == NULL)
252: progname = argv[0];
253: else
254: ++progname;
255:
1.51 kristaps 256: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 257:
1.19 kristaps 258: curp.inttype = INTT_AUTO;
1.22 kristaps 259: curp.outtype = OUTT_ASCII;
1.103 schwarze 260: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 261:
1.1 kristaps 262: /* LINTED */
1.103 schwarze 263: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 264: switch (c) {
1.10 kristaps 265: case ('m'):
1.19 kristaps 266: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 267: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 268: break;
1.48 kristaps 269: case ('O'):
1.49 kristaps 270: (void)strlcat(curp.outopts, optarg, BUFSIZ);
271: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 272: break;
1.1 kristaps 273: case ('T'):
1.60 kristaps 274: if ( ! toptions(&curp, optarg))
1.105 kristaps 275: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 276: break;
277: case ('W'):
1.103 schwarze 278: if ( ! woptions(&curp, optarg))
1.105 kristaps 279: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 280: break;
281: case ('V'):
282: version();
283: /* NOTREACHED */
284: default:
285: usage();
286: /* NOTREACHED */
287: }
288:
289: argc -= optind;
290: argv += optind;
291:
1.30 kristaps 292: if (NULL == *argv) {
293: curp.file = "<stdin>";
294: curp.fd = STDIN_FILENO;
1.39 kristaps 295:
1.66 kristaps 296: fdesc(&curp);
1.64 kristaps 297: }
298:
299: while (*argv) {
1.66 kristaps 300: ffile(*argv, &curp);
1.103 schwarze 301: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 302: break;
303: ++argv;
1.1 kristaps 304: }
305:
1.22 kristaps 306: if (curp.outfree)
307: (*curp.outfree)(curp.outdata);
1.112 kristaps 308: if (curp.pmdoc)
309: mdoc_free(curp.pmdoc);
310: if (curp.pman)
311: man_free(curp.pman);
1.71 kristaps 312: if (curp.roff)
313: roff_free(curp.roff);
1.1 kristaps 314:
1.105 kristaps 315: return((int)exit_status);
1.1 kristaps 316: }
317:
318:
1.55 kristaps 319: static void
1.1 kristaps 320: version(void)
321: {
322:
1.54 kristaps 323: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 324: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 325: }
326:
327:
1.55 kristaps 328: static void
1.1 kristaps 329: usage(void)
330: {
331:
1.112 kristaps 332: (void)fprintf(stderr, "usage: %s "
333: "[-V] "
334: "[-foption] "
335: "[-mformat] "
336: "[-Ooption] "
337: "[-Toutput] "
338: "[-Werr] "
339: "[file...]\n",
340: progname);
341:
1.105 kristaps 342: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 343: }
344:
1.64 kristaps 345: static void
1.66 kristaps 346: ffile(const char *file, struct curparse *curp)
1.1 kristaps 347: {
348:
1.112 kristaps 349: /*
350: * Called once per input file. Get the file ready for reading,
351: * pass it through to the parser-driver, then close it out.
352: * XXX: don't do anything special as this is only called for
353: * files; stdin goes directly to fdesc().
354: */
355:
1.19 kristaps 356: curp->file = file;
1.112 kristaps 357:
1.19 kristaps 358: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 359: perror(curp->file);
1.103 schwarze 360: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 361: return;
1.1 kristaps 362: }
363:
1.66 kristaps 364: fdesc(curp);
1.1 kristaps 365:
1.19 kristaps 366: if (-1 == close(curp->fd))
1.53 kristaps 367: perror(curp->file);
1.1 kristaps 368: }
369:
1.111 kristaps 370: static int
371: pfile(const char *file, struct curparse *curp)
372: {
373: const char *savefile;
374: int fd, savefd;
375:
376: if (-1 == (fd = open(file, O_RDONLY, 0))) {
377: perror(file);
1.115 schwarze 378: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 379: return(0);
380: }
381:
382: savefile = curp->file;
383: savefd = curp->fd;
384:
385: curp->file = file;
386: curp->fd = fd;
387:
388: pdesc(curp);
389:
390: curp->file = savefile;
391: curp->fd = savefd;
392:
393: if (-1 == close(fd))
394: perror(file);
395:
1.115 schwarze 396: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 397: }
398:
1.1 kristaps 399:
1.103 schwarze 400: static void
1.68 joerg 401: resize_buf(struct buf *buf, size_t initial)
402: {
403:
1.124 schwarze 404: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.103 schwarze 405: buf->buf = realloc(buf->buf, buf->sz);
406: if (NULL == buf->buf) {
1.68 joerg 407: perror(NULL);
1.105 kristaps 408: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 409: }
410: }
411:
412:
413: static int
1.66 kristaps 414: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
415: {
416: struct stat st;
1.68 joerg 417: size_t off;
1.66 kristaps 418: ssize_t ssz;
419:
420: if (-1 == fstat(curp->fd, &st)) {
421: perror(curp->file);
422: return(0);
423: }
424:
425: /*
426: * If we're a regular file, try just reading in the whole entry
427: * via mmap(). This is faster than reading it into blocks, and
428: * since each file is only a few bytes to begin with, I'm not
429: * concerned that this is going to tank any machines.
430: */
431:
432: if (S_ISREG(st.st_mode)) {
433: if (st.st_size >= (1U << 31)) {
434: fprintf(stderr, "%s: input too large\n",
435: curp->file);
436: return(0);
437: }
438: *with_mmap = 1;
1.71 kristaps 439: fb->sz = (size_t)st.st_size;
1.66 kristaps 440: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 441: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 442: if (fb->buf != MAP_FAILED)
443: return(1);
444: }
445:
446: /*
447: * If this isn't a regular file (like, say, stdin), then we must
448: * go the old way and just read things in bit by bit.
449: */
450:
451: *with_mmap = 0;
452: off = 0;
453: fb->sz = 0;
454: fb->buf = NULL;
455: for (;;) {
456: if (off == fb->sz) {
457: if (fb->sz == (1U << 31)) {
458: fprintf(stderr, "%s: input too large\n",
459: curp->file);
460: break;
461: }
1.103 schwarze 462: resize_buf(fb, 65536);
1.66 kristaps 463: }
1.71 kristaps 464: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 465: if (ssz == 0) {
466: fb->sz = off;
467: return(1);
468: }
469: if (ssz == -1) {
470: perror(curp->file);
471: break;
472: }
1.71 kristaps 473: off += (size_t)ssz;
1.66 kristaps 474: }
475:
476: free(fb->buf);
477: fb->buf = NULL;
478: return(0);
479: }
480:
481:
1.64 kristaps 482: static void
1.66 kristaps 483: fdesc(struct curparse *curp)
1.1 kristaps 484: {
1.112 kristaps 485:
486: /*
487: * Called once per file with an opened file descriptor. All
488: * pre-file-parse operations (whether stdin or a file) should go
489: * here.
490: *
491: * This calls down into the nested parser, which drills down and
492: * fully parses a file and all its dependences (i.e., `so'). It
493: * then runs the cleanup validators and pushes to output.
494: */
495:
496: /* Zero the parse type. */
497:
498: curp->mdoc = NULL;
499: curp->man = NULL;
1.115 schwarze 500: file_status = MANDOCLEVEL_OK;
1.112 kristaps 501:
502: /* Make sure the mandotory roff parser is initialised. */
503:
504: if (NULL == curp->roff) {
505: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
506: assert(curp->roff);
507: }
508:
509: /* Fully parse the file. */
1.10 kristaps 510:
1.111 kristaps 511: pdesc(curp);
1.92 kristaps 512:
1.115 schwarze 513: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 514: goto cleanup;
515:
516: /* NOTE a parser may not have been assigned, yet. */
517:
1.112 kristaps 518: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 519: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 520: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 521: goto cleanup;
522: }
523:
524: /* Clean up the parse routine ASTs. */
525:
1.112 kristaps 526: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 527: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 528: goto cleanup;
529: }
1.112 kristaps 530:
531: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 532: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 533: goto cleanup;
534: }
1.112 kristaps 535:
536: assert(curp->roff);
1.130 kristaps 537: roff_endparse(curp->roff);
1.1 kristaps 538:
539: /*
1.111 kristaps 540: * With -Wstop and warnings or errors of at least
541: * the requested level, do not produce output.
1.1 kristaps 542: */
543:
1.115 schwarze 544: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 545: goto cleanup;
546:
547: /* If unset, allocate output dev now (if applicable). */
548:
549: if ( ! (curp->outman && curp->outmdoc)) {
550: switch (curp->outtype) {
551: case (OUTT_XHTML):
552: curp->outdata = xhtml_alloc(curp->outopts);
553: break;
554: case (OUTT_HTML):
555: curp->outdata = html_alloc(curp->outopts);
556: break;
557: case (OUTT_ASCII):
558: curp->outdata = ascii_alloc(curp->outopts);
559: curp->outfree = ascii_free;
560: break;
561: case (OUTT_PDF):
562: curp->outdata = pdf_alloc(curp->outopts);
563: curp->outfree = pspdf_free;
564: break;
565: case (OUTT_PS):
566: curp->outdata = ps_alloc(curp->outopts);
567: curp->outfree = pspdf_free;
568: break;
569: default:
570: break;
571: }
572:
573: switch (curp->outtype) {
574: case (OUTT_HTML):
575: /* FALLTHROUGH */
576: case (OUTT_XHTML):
577: curp->outman = html_man;
578: curp->outmdoc = html_mdoc;
579: curp->outfree = html_free;
580: break;
581: case (OUTT_TREE):
582: curp->outman = tree_man;
583: curp->outmdoc = tree_mdoc;
584: break;
585: case (OUTT_PDF):
586: /* FALLTHROUGH */
587: case (OUTT_ASCII):
588: /* FALLTHROUGH */
589: case (OUTT_PS):
590: curp->outman = terminal_man;
591: curp->outmdoc = terminal_mdoc;
592: break;
593: default:
594: break;
595: }
596: }
597:
598: /* Execute the out device, if it exists. */
599:
1.112 kristaps 600: if (curp->man && curp->outman)
601: (*curp->outman)(curp->outdata, curp->man);
602: if (curp->mdoc && curp->outmdoc)
603: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 604:
605: cleanup:
1.112 kristaps 606:
1.111 kristaps 607: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 608:
609: /* Reset the current-parse compilers. */
610:
611: if (curp->mdoc)
612: mdoc_reset(curp->mdoc);
613: if (curp->man)
614: man_reset(curp->man);
615:
616: assert(curp->roff);
617: roff_reset(curp->roff);
1.111 kristaps 618:
1.115 schwarze 619: if (exit_status < file_status)
620: exit_status = file_status;
621:
1.111 kristaps 622: return;
623: }
624:
625: static void
626: pdesc(struct curparse *curp)
627: {
628: struct buf blk;
629: int with_mmap;
630:
1.112 kristaps 631: /*
632: * Run for each opened file; may be called more than once for
633: * each full parse sequence if the opened file is nested (i.e.,
634: * from `so'). Simply sucks in the whole file and moves into
635: * the parse phase for the file.
636: */
637:
1.103 schwarze 638: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 639: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 640: return;
1.103 schwarze 641: }
1.1 kristaps 642:
1.112 kristaps 643: /* Line number is per-file. */
1.111 kristaps 644:
645: curp->line = 1;
1.112 kristaps 646:
1.111 kristaps 647: parsebuf(curp, blk, 1);
648:
649: if (with_mmap)
650: munmap(blk.buf, blk.sz);
651: else
652: free(blk.buf);
653: }
654:
655: static void
656: parsebuf(struct curparse *curp, struct buf blk, int start)
657: {
658: struct buf ln;
1.114 kristaps 659: enum rofferr rr;
1.112 kristaps 660: int i, of, rc;
661: int pos; /* byte number in the ln buffer */
662: int lnn; /* line number in the real file */
1.111 kristaps 663: unsigned char c;
1.112 kristaps 664:
665: /*
666: * Main parse routine for an opened file. This is called for
667: * each opened file and simply loops around the full input file,
668: * possibly nesting (i.e., with `so').
669: */
1.71 kristaps 670:
1.111 kristaps 671: memset(&ln, 0, sizeof(struct buf));
672:
1.112 kristaps 673: lnn = curp->line;
674: pos = 0;
1.111 kristaps 675:
1.112 kristaps 676: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 677: if (0 == pos && '\0' == blk.buf[i])
678: break;
1.112 kristaps 679:
1.122 schwarze 680: if (start) {
1.111 kristaps 681: curp->line = lnn;
1.122 schwarze 682: curp->reparse_count = 0;
683: }
1.111 kristaps 684:
685: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 686: if ('\n' == blk.buf[i]) {
687: ++i;
688: ++lnn;
689: break;
690: }
1.99 kristaps 691:
692: /*
693: * Warn about bogus characters. If you're using
694: * non-ASCII encoding, you're screwing your
695: * readers. Since I'd rather this not happen,
696: * I'll be helpful and drop these characters so
697: * we don't display gibberish. Note to manual
698: * writers: use special characters.
699: */
700:
1.102 schwarze 701: c = (unsigned char) blk.buf[i];
1.112 kristaps 702:
703: if ( ! (isascii(c) &&
704: (isgraph(c) || isblank(c)))) {
1.103 schwarze 705: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 706: curp->line, pos, "ignoring byte");
1.99 kristaps 707: i++;
708: continue;
709: }
710:
1.112 kristaps 711: /* Trailing backslash = a plain char. */
712:
1.70 joerg 713: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
714: if (pos >= (int)ln.sz)
1.103 schwarze 715: resize_buf(&ln, 256);
1.70 joerg 716: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 717: continue;
1.70 joerg 718: }
1.112 kristaps 719:
720: /* Found escape & at least one other char. */
721:
1.70 joerg 722: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 723: i += 2;
1.70 joerg 724: /* Escaped newlines are skipped over */
725: ++lnn;
1.67 joerg 726: continue;
727: }
1.112 kristaps 728:
1.70 joerg 729: if ('"' == blk.buf[i + 1]) {
730: i += 2;
731: /* Comment, skip to end of line */
732: for (; i < (int)blk.sz; ++i) {
733: if ('\n' == blk.buf[i]) {
734: ++i;
735: ++lnn;
736: break;
737: }
738: }
1.112 kristaps 739:
1.70 joerg 740: /* Backout trailing whitespaces */
741: for (; pos > 0; --pos) {
742: if (ln.buf[pos - 1] != ' ')
743: break;
744: if (pos > 2 && ln.buf[pos - 2] == '\\')
745: break;
746: }
747: break;
748: }
1.112 kristaps 749:
750: /* Some other escape sequence, copy & cont. */
751:
1.70 joerg 752: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 753: resize_buf(&ln, 256);
1.1 kristaps 754:
1.70 joerg 755: ln.buf[pos++] = blk.buf[i++];
756: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 757: }
1.1 kristaps 758:
1.70 joerg 759: if (pos >= (int)ln.sz)
1.103 schwarze 760: resize_buf(&ln, 256);
1.112 kristaps 761:
1.71 kristaps 762: ln.buf[pos] = '\0';
763:
1.76 kristaps 764: /*
765: * A significant amount of complexity is contained by
766: * the roff preprocessor. It's line-oriented but can be
767: * expressed on one line, so we need at times to
768: * readjust our starting point and re-run it. The roff
769: * preprocessor can also readjust the buffers with new
770: * data, so we pass them in wholesale.
771: */
772:
773: of = 0;
1.112 kristaps 774:
1.111 kristaps 775: rerun:
1.114 kristaps 776: rr = roff_parseln
1.112 kristaps 777: (curp->roff, curp->line,
778: &ln.buf, &ln.sz, of, &of);
779:
1.114 kristaps 780: switch (rr) {
1.111 kristaps 781: case (ROFF_REPARSE):
1.122 schwarze 782: if (REPARSE_LIMIT >= ++curp->reparse_count)
783: parsebuf(curp, ln, 0);
784: else
785: mmsg(MANDOCERR_ROFFLOOP, curp,
786: curp->line, pos, NULL);
1.111 kristaps 787: pos = 0;
788: continue;
789: case (ROFF_APPEND):
790: pos = strlen(ln.buf);
791: continue;
792: case (ROFF_RERUN):
793: goto rerun;
794: case (ROFF_IGN):
795: pos = 0;
1.71 kristaps 796: continue;
1.111 kristaps 797: case (ROFF_ERR):
1.115 schwarze 798: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 799: break;
800: case (ROFF_SO):
801: if (pfile(ln.buf + of, curp)) {
802: pos = 0;
803: continue;
804: } else
805: break;
1.128 kristaps 806: default:
1.111 kristaps 807: break;
1.103 schwarze 808: }
1.132 kristaps 809:
810: /*
811: * If we encounter errors in the recursive parsebuf()
812: * call, make sure we don't continue parsing.
813: */
814:
815: if (MANDOCLEVEL_FATAL <= file_status)
816: break;
1.29 kristaps 817:
1.76 kristaps 818: /*
819: * If input parsers have not been allocated, do so now.
820: * We keep these instanced betwen parsers, but set them
821: * locally per parse routine since we can use different
822: * parsers with each one.
823: */
1.19 kristaps 824:
1.112 kristaps 825: if ( ! (curp->man || curp->mdoc))
826: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 827:
1.112 kristaps 828: /*
829: * Lastly, push down into the parsers themselves. One
830: * of these will have already been set in the pset()
831: * routine.
1.128 kristaps 832: * If libroff returns ROFF_TBL, then add it to the
833: * currently open parse. Since we only get here if
834: * there does exist data (see tbl_data.c), we're
835: * guaranteed that something's been allocated.
1.112 kristaps 836: */
837:
1.128 kristaps 838: if (ROFF_TBL == rr) {
839: assert(curp->man || curp->mdoc);
840: if (curp->man)
841: man_addspan(curp->man, roff_span(curp->roff));
842: else
843: mdoc_addspan(curp->mdoc, roff_span(curp->roff));
844:
845: } else if (curp->man || curp->mdoc) {
1.112 kristaps 846: rc = curp->man ?
847: man_parseln(curp->man,
848: curp->line, ln.buf, of) :
849: mdoc_parseln(curp->mdoc,
850: curp->line, ln.buf, of);
1.30 kristaps 851:
1.112 kristaps 852: if ( ! rc) {
1.115 schwarze 853: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 854: break;
855: }
1.103 schwarze 856: }
1.19 kristaps 857:
1.111 kristaps 858: /* Temporary buffers typically are not full. */
1.112 kristaps 859:
1.111 kristaps 860: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 861: break;
1.85 kristaps 862:
1.111 kristaps 863: /* Start the next input line. */
1.112 kristaps 864:
1.111 kristaps 865: pos = 0;
1.22 kristaps 866: }
867:
1.111 kristaps 868: free(ln.buf);
1.19 kristaps 869: }
870:
1.103 schwarze 871: static void
1.112 kristaps 872: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 873: {
1.29 kristaps 874: int i;
1.19 kristaps 875:
876: /*
877: * Try to intuit which kind of manual parser should be used. If
878: * passed in by command-line (-man, -mdoc), then use that
879: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 880: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 881: * default to -man, which is more lenient.
1.112 kristaps 882: *
883: * Separate out pmdoc/pman from mdoc/man: the first persists
884: * through all parsers, while the latter is used per-parse.
1.19 kristaps 885: */
886:
1.75 kristaps 887: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 888: for (i = 1; buf[i]; i++)
889: if (' ' != buf[i] && '\t' != buf[i])
890: break;
1.103 schwarze 891: if ('\0' == buf[i])
892: return;
1.29 kristaps 893: }
1.10 kristaps 894:
1.19 kristaps 895: switch (curp->inttype) {
896: case (INTT_MDOC):
1.112 kristaps 897: if (NULL == curp->pmdoc)
898: curp->pmdoc = mdoc_alloc
899: (&curp->regs, curp, mmsg);
900: assert(curp->pmdoc);
901: curp->mdoc = curp->pmdoc;
1.103 schwarze 902: return;
1.19 kristaps 903: case (INTT_MAN):
1.112 kristaps 904: if (NULL == curp->pman)
905: curp->pman = man_alloc
906: (&curp->regs, curp, mmsg);
907: assert(curp->pman);
908: curp->man = curp->pman;
1.103 schwarze 909: return;
1.19 kristaps 910: default:
911: break;
912: }
913:
914: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 915: if (NULL == curp->pmdoc)
916: curp->pmdoc = mdoc_alloc
917: (&curp->regs, curp, mmsg);
918: assert(curp->pmdoc);
919: curp->mdoc = curp->pmdoc;
1.103 schwarze 920: return;
1.19 kristaps 921: }
922:
1.112 kristaps 923: if (NULL == curp->pman)
924: curp->pman = man_alloc(&curp->regs, curp, mmsg);
925: assert(curp->pman);
926: curp->man = curp->pman;
1.10 kristaps 927: }
928:
929: static int
930: moptions(enum intt *tflags, char *arg)
931: {
932:
1.17 kristaps 933: if (0 == strcmp(arg, "doc"))
1.10 kristaps 934: *tflags = INTT_MDOC;
1.19 kristaps 935: else if (0 == strcmp(arg, "andoc"))
936: *tflags = INTT_AUTO;
1.17 kristaps 937: else if (0 == strcmp(arg, "an"))
1.10 kristaps 938: *tflags = INTT_MAN;
939: else {
1.57 kristaps 940: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 941: return(0);
942: }
943:
944: return(1);
1.1 kristaps 945: }
946:
947: static int
1.60 kristaps 948: toptions(struct curparse *curp, char *arg)
1.1 kristaps 949: {
950:
951: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 952: curp->outtype = OUTT_ASCII;
953: else if (0 == strcmp(arg, "lint")) {
954: curp->outtype = OUTT_LINT;
1.103 schwarze 955: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 956: }
1.1 kristaps 957: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 958: curp->outtype = OUTT_TREE;
1.43 kristaps 959: else if (0 == strcmp(arg, "html"))
1.60 kristaps 960: curp->outtype = OUTT_HTML;
1.59 kristaps 961: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 962: curp->outtype = OUTT_XHTML;
1.85 kristaps 963: else if (0 == strcmp(arg, "ps"))
964: curp->outtype = OUTT_PS;
1.100 kristaps 965: else if (0 == strcmp(arg, "pdf"))
966: curp->outtype = OUTT_PDF;
1.1 kristaps 967: else {
1.57 kristaps 968: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 969: return(0);
970: }
971:
972: return(1);
973: }
974:
975: static int
1.103 schwarze 976: woptions(struct curparse *curp, char *arg)
1.1 kristaps 977: {
1.34 kristaps 978: char *v, *o;
1.103 schwarze 979: const char *toks[6];
1.1 kristaps 980:
1.103 schwarze 981: toks[0] = "stop";
982: toks[1] = "all";
983: toks[2] = "warning";
984: toks[3] = "error";
985: toks[4] = "fatal";
986: toks[5] = NULL;
1.1 kristaps 987:
1.34 kristaps 988: while (*arg) {
989: o = arg;
1.45 kristaps 990: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 991: case (0):
1.103 schwarze 992: curp->wstop = 1;
1.1 kristaps 993: break;
994: case (1):
1.103 schwarze 995: /* FALLTHROUGH */
1.1 kristaps 996: case (2):
1.103 schwarze 997: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 998: break;
1.20 kristaps 999: case (3):
1.103 schwarze 1000: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 1001: break;
1002: case (4):
1.103 schwarze 1003: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 1004: break;
1.1 kristaps 1005: default:
1.103 schwarze 1006: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 1007: return(0);
1008: }
1.34 kristaps 1009: }
1.1 kristaps 1010:
1011: return(1);
1012: }
1013:
1.71 kristaps 1014: static int
1015: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
1016: {
1017: struct curparse *cp;
1.103 schwarze 1018: enum mandoclevel level;
1019:
1020: level = MANDOCLEVEL_FATAL;
1021: while (t < mandoclimits[level])
1.105 kristaps 1022: /* LINTED */
1.103 schwarze 1023: level--;
1.71 kristaps 1024:
1025: cp = (struct curparse *)arg;
1.103 schwarze 1026: if (level < cp->wlevel)
1027: return(1);
1.73 kristaps 1028:
1.103 schwarze 1029: fprintf(stderr, "%s:%d:%d: %s: %s",
1030: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 1031: if (msg)
1032: fprintf(stderr, ": %s", msg);
1033: fputc('\n', stderr);
1.79 kristaps 1034:
1.115 schwarze 1035: if (file_status < level)
1036: file_status = level;
1.103 schwarze 1037:
1038: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 1039: }
CVSweb