Annotation of mandoc/main.c, Revision 1.136
1.136 ! kristaps 1: /* $Id: main.c,v 1.135 2011/01/04 15:02:00 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",
195: "skipping text before the first section header",
196: "skipping unknown macro",
1.110 kristaps 197: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 198: "line scope broken",
199: "argument count wrong",
1.121 kristaps 200: "skipping end of block that is not open",
201: "missing end of block",
1.106 schwarze 202: "scope open on exit",
1.117 kristaps 203: "uname(3) system call failed",
1.79 kristaps 204: "macro requires line argument(s)",
205: "macro requires body argument(s)",
206: "macro requires argument(s)",
1.82 kristaps 207: "missing list type",
1.79 kristaps 208: "line argument(s) will be lost",
209: "body argument(s) will be lost",
1.94 schwarze 210:
211: "generic fatal error",
212:
1.80 kristaps 213: "column syntax is inconsistent",
1.121 kristaps 214: "NOT IMPLEMENTED: .Bd -file",
1.79 kristaps 215: "line scope broken, syntax violated",
216: "argument count wrong, violates syntax",
217: "child violates parent syntax",
218: "argument count wrong, violates syntax",
1.113 kristaps 219: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 220: "no document body",
221: "no document prologue",
1.103 schwarze 222: "static buffer exhausted",
1.5 kristaps 223: };
1.1 kristaps 224:
1.111 kristaps 225: static void parsebuf(struct curparse *, struct buf, int);
226: static void pdesc(struct curparse *);
1.66 kristaps 227: static void fdesc(struct curparse *);
228: static void ffile(const char *, struct curparse *);
1.111 kristaps 229: static int pfile(const char *, struct curparse *);
1.10 kristaps 230: static int moptions(enum intt *, char *);
1.71 kristaps 231: static int mmsg(enum mandocerr, void *,
232: int, int, const char *);
1.112 kristaps 233: static void pset(const char *, int, struct curparse *);
1.66 kristaps 234: static int toptions(struct curparse *, char *);
235: static void usage(void) __attribute__((noreturn));
1.55 kristaps 236: static void version(void) __attribute__((noreturn));
1.103 schwarze 237: static int woptions(struct curparse *, char *);
1.1 kristaps 238:
1.54 kristaps 239: static const char *progname;
1.115 schwarze 240: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 241: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 242:
243: int
244: main(int argc, char *argv[])
245: {
1.64 kristaps 246: int c;
1.5 kristaps 247: struct curparse curp;
1.1 kristaps 248:
1.54 kristaps 249: progname = strrchr(argv[0], '/');
250: if (progname == NULL)
251: progname = argv[0];
252: else
253: ++progname;
254:
1.51 kristaps 255: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 256:
1.19 kristaps 257: curp.inttype = INTT_AUTO;
1.22 kristaps 258: curp.outtype = OUTT_ASCII;
1.103 schwarze 259: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 260:
1.1 kristaps 261: /* LINTED */
1.103 schwarze 262: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 263: switch (c) {
1.10 kristaps 264: case ('m'):
1.19 kristaps 265: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 266: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 267: break;
1.48 kristaps 268: case ('O'):
1.49 kristaps 269: (void)strlcat(curp.outopts, optarg, BUFSIZ);
270: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 271: break;
1.1 kristaps 272: case ('T'):
1.60 kristaps 273: if ( ! toptions(&curp, optarg))
1.105 kristaps 274: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 275: break;
276: case ('W'):
1.103 schwarze 277: if ( ! woptions(&curp, optarg))
1.105 kristaps 278: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 279: break;
280: case ('V'):
281: version();
282: /* NOTREACHED */
283: default:
284: usage();
285: /* NOTREACHED */
286: }
287:
288: argc -= optind;
289: argv += optind;
290:
1.30 kristaps 291: if (NULL == *argv) {
292: curp.file = "<stdin>";
293: curp.fd = STDIN_FILENO;
1.39 kristaps 294:
1.66 kristaps 295: fdesc(&curp);
1.64 kristaps 296: }
297:
298: while (*argv) {
1.66 kristaps 299: ffile(*argv, &curp);
1.103 schwarze 300: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 301: break;
302: ++argv;
1.1 kristaps 303: }
304:
1.22 kristaps 305: if (curp.outfree)
306: (*curp.outfree)(curp.outdata);
1.112 kristaps 307: if (curp.pmdoc)
308: mdoc_free(curp.pmdoc);
309: if (curp.pman)
310: man_free(curp.pman);
1.71 kristaps 311: if (curp.roff)
312: roff_free(curp.roff);
1.1 kristaps 313:
1.105 kristaps 314: return((int)exit_status);
1.1 kristaps 315: }
316:
317:
1.55 kristaps 318: static void
1.1 kristaps 319: version(void)
320: {
321:
1.54 kristaps 322: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 323: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 324: }
325:
326:
1.55 kristaps 327: static void
1.1 kristaps 328: usage(void)
329: {
330:
1.112 kristaps 331: (void)fprintf(stderr, "usage: %s "
332: "[-V] "
333: "[-foption] "
334: "[-mformat] "
335: "[-Ooption] "
336: "[-Toutput] "
337: "[-Werr] "
338: "[file...]\n",
339: progname);
340:
1.105 kristaps 341: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 342: }
343:
1.64 kristaps 344: static void
1.66 kristaps 345: ffile(const char *file, struct curparse *curp)
1.1 kristaps 346: {
347:
1.112 kristaps 348: /*
349: * Called once per input file. Get the file ready for reading,
350: * pass it through to the parser-driver, then close it out.
351: * XXX: don't do anything special as this is only called for
352: * files; stdin goes directly to fdesc().
353: */
354:
1.19 kristaps 355: curp->file = file;
1.112 kristaps 356:
1.19 kristaps 357: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 358: perror(curp->file);
1.103 schwarze 359: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 360: return;
1.1 kristaps 361: }
362:
1.66 kristaps 363: fdesc(curp);
1.1 kristaps 364:
1.19 kristaps 365: if (-1 == close(curp->fd))
1.53 kristaps 366: perror(curp->file);
1.1 kristaps 367: }
368:
1.111 kristaps 369: static int
370: pfile(const char *file, struct curparse *curp)
371: {
372: const char *savefile;
373: int fd, savefd;
374:
375: if (-1 == (fd = open(file, O_RDONLY, 0))) {
376: perror(file);
1.115 schwarze 377: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 378: return(0);
379: }
380:
381: savefile = curp->file;
382: savefd = curp->fd;
383:
384: curp->file = file;
385: curp->fd = fd;
386:
387: pdesc(curp);
388:
389: curp->file = savefile;
390: curp->fd = savefd;
391:
392: if (-1 == close(fd))
393: perror(file);
394:
1.115 schwarze 395: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 396: }
397:
1.1 kristaps 398:
1.103 schwarze 399: static void
1.68 joerg 400: resize_buf(struct buf *buf, size_t initial)
401: {
402:
1.124 schwarze 403: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.103 schwarze 404: buf->buf = realloc(buf->buf, buf->sz);
405: if (NULL == buf->buf) {
1.68 joerg 406: perror(NULL);
1.105 kristaps 407: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 408: }
409: }
410:
411:
412: static int
1.66 kristaps 413: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
414: {
415: struct stat st;
1.68 joerg 416: size_t off;
1.66 kristaps 417: ssize_t ssz;
418:
419: if (-1 == fstat(curp->fd, &st)) {
420: perror(curp->file);
421: return(0);
422: }
423:
424: /*
425: * If we're a regular file, try just reading in the whole entry
426: * via mmap(). This is faster than reading it into blocks, and
427: * since each file is only a few bytes to begin with, I'm not
428: * concerned that this is going to tank any machines.
429: */
430:
431: if (S_ISREG(st.st_mode)) {
432: if (st.st_size >= (1U << 31)) {
433: fprintf(stderr, "%s: input too large\n",
434: curp->file);
435: return(0);
436: }
437: *with_mmap = 1;
1.71 kristaps 438: fb->sz = (size_t)st.st_size;
1.66 kristaps 439: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 440: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 441: if (fb->buf != MAP_FAILED)
442: return(1);
443: }
444:
445: /*
446: * If this isn't a regular file (like, say, stdin), then we must
447: * go the old way and just read things in bit by bit.
448: */
449:
450: *with_mmap = 0;
451: off = 0;
452: fb->sz = 0;
453: fb->buf = NULL;
454: for (;;) {
455: if (off == fb->sz) {
456: if (fb->sz == (1U << 31)) {
457: fprintf(stderr, "%s: input too large\n",
458: curp->file);
459: break;
460: }
1.103 schwarze 461: resize_buf(fb, 65536);
1.66 kristaps 462: }
1.71 kristaps 463: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 464: if (ssz == 0) {
465: fb->sz = off;
466: return(1);
467: }
468: if (ssz == -1) {
469: perror(curp->file);
470: break;
471: }
1.71 kristaps 472: off += (size_t)ssz;
1.66 kristaps 473: }
474:
475: free(fb->buf);
476: fb->buf = NULL;
477: return(0);
478: }
479:
480:
1.64 kristaps 481: static void
1.66 kristaps 482: fdesc(struct curparse *curp)
1.1 kristaps 483: {
1.112 kristaps 484:
485: /*
486: * Called once per file with an opened file descriptor. All
487: * pre-file-parse operations (whether stdin or a file) should go
488: * here.
489: *
490: * This calls down into the nested parser, which drills down and
491: * fully parses a file and all its dependences (i.e., `so'). It
492: * then runs the cleanup validators and pushes to output.
493: */
494:
495: /* Zero the parse type. */
496:
497: curp->mdoc = NULL;
498: curp->man = NULL;
1.115 schwarze 499: file_status = MANDOCLEVEL_OK;
1.112 kristaps 500:
501: /* Make sure the mandotory roff parser is initialised. */
502:
503: if (NULL == curp->roff) {
504: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
505: assert(curp->roff);
506: }
507:
508: /* Fully parse the file. */
1.10 kristaps 509:
1.111 kristaps 510: pdesc(curp);
1.92 kristaps 511:
1.115 schwarze 512: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 513: goto cleanup;
514:
515: /* NOTE a parser may not have been assigned, yet. */
516:
1.112 kristaps 517: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 518: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 519: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 520: goto cleanup;
521: }
522:
523: /* Clean up the parse routine ASTs. */
524:
1.112 kristaps 525: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 526: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 527: goto cleanup;
528: }
1.112 kristaps 529:
530: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 531: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 532: goto cleanup;
533: }
1.112 kristaps 534:
535: assert(curp->roff);
1.130 kristaps 536: roff_endparse(curp->roff);
1.1 kristaps 537:
538: /*
1.111 kristaps 539: * With -Wstop and warnings or errors of at least
540: * the requested level, do not produce output.
1.1 kristaps 541: */
542:
1.115 schwarze 543: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 544: goto cleanup;
545:
546: /* If unset, allocate output dev now (if applicable). */
547:
548: if ( ! (curp->outman && curp->outmdoc)) {
549: switch (curp->outtype) {
550: case (OUTT_XHTML):
551: curp->outdata = xhtml_alloc(curp->outopts);
552: break;
553: case (OUTT_HTML):
554: curp->outdata = html_alloc(curp->outopts);
555: break;
556: case (OUTT_ASCII):
557: curp->outdata = ascii_alloc(curp->outopts);
558: curp->outfree = ascii_free;
559: break;
560: case (OUTT_PDF):
561: curp->outdata = pdf_alloc(curp->outopts);
562: curp->outfree = pspdf_free;
563: break;
564: case (OUTT_PS):
565: curp->outdata = ps_alloc(curp->outopts);
566: curp->outfree = pspdf_free;
567: break;
568: default:
569: break;
570: }
571:
572: switch (curp->outtype) {
573: case (OUTT_HTML):
574: /* FALLTHROUGH */
575: case (OUTT_XHTML):
576: curp->outman = html_man;
577: curp->outmdoc = html_mdoc;
578: curp->outfree = html_free;
579: break;
580: case (OUTT_TREE):
581: curp->outman = tree_man;
582: curp->outmdoc = tree_mdoc;
583: break;
584: case (OUTT_PDF):
585: /* FALLTHROUGH */
586: case (OUTT_ASCII):
587: /* FALLTHROUGH */
588: case (OUTT_PS):
589: curp->outman = terminal_man;
590: curp->outmdoc = terminal_mdoc;
591: break;
592: default:
593: break;
594: }
595: }
596:
597: /* Execute the out device, if it exists. */
598:
1.112 kristaps 599: if (curp->man && curp->outman)
600: (*curp->outman)(curp->outdata, curp->man);
601: if (curp->mdoc && curp->outmdoc)
602: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 603:
604: cleanup:
1.112 kristaps 605:
1.111 kristaps 606: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 607:
608: /* Reset the current-parse compilers. */
609:
610: if (curp->mdoc)
611: mdoc_reset(curp->mdoc);
612: if (curp->man)
613: man_reset(curp->man);
614:
615: assert(curp->roff);
616: roff_reset(curp->roff);
1.111 kristaps 617:
1.115 schwarze 618: if (exit_status < file_status)
619: exit_status = file_status;
620:
1.111 kristaps 621: return;
622: }
623:
624: static void
625: pdesc(struct curparse *curp)
626: {
627: struct buf blk;
628: int with_mmap;
629:
1.112 kristaps 630: /*
631: * Run for each opened file; may be called more than once for
632: * each full parse sequence if the opened file is nested (i.e.,
633: * from `so'). Simply sucks in the whole file and moves into
634: * the parse phase for the file.
635: */
636:
1.103 schwarze 637: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 638: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 639: return;
1.103 schwarze 640: }
1.1 kristaps 641:
1.112 kristaps 642: /* Line number is per-file. */
1.111 kristaps 643:
644: curp->line = 1;
1.112 kristaps 645:
1.111 kristaps 646: parsebuf(curp, blk, 1);
647:
648: if (with_mmap)
649: munmap(blk.buf, blk.sz);
650: else
651: free(blk.buf);
652: }
653:
654: static void
655: parsebuf(struct curparse *curp, struct buf blk, int start)
656: {
657: struct buf ln;
1.114 kristaps 658: enum rofferr rr;
1.112 kristaps 659: int i, of, rc;
660: int pos; /* byte number in the ln buffer */
661: int lnn; /* line number in the real file */
1.111 kristaps 662: unsigned char c;
1.112 kristaps 663:
664: /*
665: * Main parse routine for an opened file. This is called for
666: * each opened file and simply loops around the full input file,
667: * possibly nesting (i.e., with `so').
668: */
1.71 kristaps 669:
1.111 kristaps 670: memset(&ln, 0, sizeof(struct buf));
671:
1.112 kristaps 672: lnn = curp->line;
673: pos = 0;
1.111 kristaps 674:
1.112 kristaps 675: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 676: if (0 == pos && '\0' == blk.buf[i])
677: break;
1.112 kristaps 678:
1.122 schwarze 679: if (start) {
1.111 kristaps 680: curp->line = lnn;
1.122 schwarze 681: curp->reparse_count = 0;
682: }
1.111 kristaps 683:
684: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 685: if ('\n' == blk.buf[i]) {
686: ++i;
687: ++lnn;
688: break;
689: }
1.99 kristaps 690:
691: /*
692: * Warn about bogus characters. If you're using
693: * non-ASCII encoding, you're screwing your
694: * readers. Since I'd rather this not happen,
695: * I'll be helpful and drop these characters so
696: * we don't display gibberish. Note to manual
697: * writers: use special characters.
698: */
699:
1.102 schwarze 700: c = (unsigned char) blk.buf[i];
1.112 kristaps 701:
702: if ( ! (isascii(c) &&
703: (isgraph(c) || isblank(c)))) {
1.103 schwarze 704: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 705: curp->line, pos, "ignoring byte");
1.99 kristaps 706: i++;
707: continue;
708: }
709:
1.112 kristaps 710: /* Trailing backslash = a plain char. */
711:
1.70 joerg 712: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
713: if (pos >= (int)ln.sz)
1.103 schwarze 714: resize_buf(&ln, 256);
1.70 joerg 715: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 716: continue;
1.70 joerg 717: }
1.112 kristaps 718:
719: /* Found escape & at least one other char. */
720:
1.70 joerg 721: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 722: i += 2;
1.70 joerg 723: /* Escaped newlines are skipped over */
724: ++lnn;
1.67 joerg 725: continue;
726: }
1.112 kristaps 727:
1.70 joerg 728: if ('"' == blk.buf[i + 1]) {
729: i += 2;
730: /* Comment, skip to end of line */
731: for (; i < (int)blk.sz; ++i) {
732: if ('\n' == blk.buf[i]) {
733: ++i;
734: ++lnn;
735: break;
736: }
737: }
1.112 kristaps 738:
1.70 joerg 739: /* Backout trailing whitespaces */
740: for (; pos > 0; --pos) {
741: if (ln.buf[pos - 1] != ' ')
742: break;
743: if (pos > 2 && ln.buf[pos - 2] == '\\')
744: break;
745: }
746: break;
747: }
1.112 kristaps 748:
749: /* Some other escape sequence, copy & cont. */
750:
1.70 joerg 751: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 752: resize_buf(&ln, 256);
1.1 kristaps 753:
1.70 joerg 754: ln.buf[pos++] = blk.buf[i++];
755: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 756: }
1.1 kristaps 757:
1.70 joerg 758: if (pos >= (int)ln.sz)
1.103 schwarze 759: resize_buf(&ln, 256);
1.112 kristaps 760:
1.71 kristaps 761: ln.buf[pos] = '\0';
762:
1.76 kristaps 763: /*
764: * A significant amount of complexity is contained by
765: * the roff preprocessor. It's line-oriented but can be
766: * expressed on one line, so we need at times to
767: * readjust our starting point and re-run it. The roff
768: * preprocessor can also readjust the buffers with new
769: * data, so we pass them in wholesale.
770: */
771:
772: of = 0;
1.112 kristaps 773:
1.111 kristaps 774: rerun:
1.114 kristaps 775: rr = roff_parseln
1.112 kristaps 776: (curp->roff, curp->line,
777: &ln.buf, &ln.sz, of, &of);
778:
1.114 kristaps 779: switch (rr) {
1.111 kristaps 780: case (ROFF_REPARSE):
1.122 schwarze 781: if (REPARSE_LIMIT >= ++curp->reparse_count)
782: parsebuf(curp, ln, 0);
783: else
784: mmsg(MANDOCERR_ROFFLOOP, curp,
785: curp->line, pos, NULL);
1.111 kristaps 786: pos = 0;
787: continue;
788: case (ROFF_APPEND):
789: pos = strlen(ln.buf);
790: continue;
791: case (ROFF_RERUN):
792: goto rerun;
793: case (ROFF_IGN):
794: pos = 0;
1.71 kristaps 795: continue;
1.111 kristaps 796: case (ROFF_ERR):
1.115 schwarze 797: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 798: break;
799: case (ROFF_SO):
800: if (pfile(ln.buf + of, curp)) {
801: pos = 0;
802: continue;
803: } else
804: break;
1.128 kristaps 805: default:
1.111 kristaps 806: break;
1.103 schwarze 807: }
1.132 kristaps 808:
809: /*
810: * If we encounter errors in the recursive parsebuf()
811: * call, make sure we don't continue parsing.
812: */
813:
814: if (MANDOCLEVEL_FATAL <= file_status)
815: break;
1.29 kristaps 816:
1.76 kristaps 817: /*
818: * If input parsers have not been allocated, do so now.
819: * We keep these instanced betwen parsers, but set them
820: * locally per parse routine since we can use different
821: * parsers with each one.
822: */
1.19 kristaps 823:
1.112 kristaps 824: if ( ! (curp->man || curp->mdoc))
825: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 826:
1.112 kristaps 827: /*
828: * Lastly, push down into the parsers themselves. One
829: * of these will have already been set in the pset()
830: * routine.
1.128 kristaps 831: * If libroff returns ROFF_TBL, then add it to the
832: * currently open parse. Since we only get here if
833: * there does exist data (see tbl_data.c), we're
834: * guaranteed that something's been allocated.
1.112 kristaps 835: */
836:
1.128 kristaps 837: if (ROFF_TBL == rr) {
838: assert(curp->man || curp->mdoc);
839: if (curp->man)
840: man_addspan(curp->man, roff_span(curp->roff));
841: else
842: mdoc_addspan(curp->mdoc, roff_span(curp->roff));
843:
844: } else if (curp->man || curp->mdoc) {
1.112 kristaps 845: rc = curp->man ?
846: man_parseln(curp->man,
847: curp->line, ln.buf, of) :
848: mdoc_parseln(curp->mdoc,
849: curp->line, ln.buf, of);
1.30 kristaps 850:
1.112 kristaps 851: if ( ! rc) {
1.115 schwarze 852: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 853: break;
854: }
1.103 schwarze 855: }
1.19 kristaps 856:
1.111 kristaps 857: /* Temporary buffers typically are not full. */
1.112 kristaps 858:
1.111 kristaps 859: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 860: break;
1.85 kristaps 861:
1.111 kristaps 862: /* Start the next input line. */
1.112 kristaps 863:
1.111 kristaps 864: pos = 0;
1.22 kristaps 865: }
866:
1.111 kristaps 867: free(ln.buf);
1.19 kristaps 868: }
869:
1.103 schwarze 870: static void
1.112 kristaps 871: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 872: {
1.29 kristaps 873: int i;
1.19 kristaps 874:
875: /*
876: * Try to intuit which kind of manual parser should be used. If
877: * passed in by command-line (-man, -mdoc), then use that
878: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 879: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 880: * default to -man, which is more lenient.
1.112 kristaps 881: *
882: * Separate out pmdoc/pman from mdoc/man: the first persists
883: * through all parsers, while the latter is used per-parse.
1.19 kristaps 884: */
885:
1.75 kristaps 886: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 887: for (i = 1; buf[i]; i++)
888: if (' ' != buf[i] && '\t' != buf[i])
889: break;
1.103 schwarze 890: if ('\0' == buf[i])
891: return;
1.29 kristaps 892: }
1.10 kristaps 893:
1.19 kristaps 894: switch (curp->inttype) {
895: case (INTT_MDOC):
1.112 kristaps 896: if (NULL == curp->pmdoc)
897: curp->pmdoc = mdoc_alloc
898: (&curp->regs, curp, mmsg);
899: assert(curp->pmdoc);
900: curp->mdoc = curp->pmdoc;
1.103 schwarze 901: return;
1.19 kristaps 902: case (INTT_MAN):
1.112 kristaps 903: if (NULL == curp->pman)
904: curp->pman = man_alloc
905: (&curp->regs, curp, mmsg);
906: assert(curp->pman);
907: curp->man = curp->pman;
1.103 schwarze 908: return;
1.19 kristaps 909: default:
910: break;
911: }
912:
913: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 914: if (NULL == curp->pmdoc)
915: curp->pmdoc = mdoc_alloc
916: (&curp->regs, curp, mmsg);
917: assert(curp->pmdoc);
918: curp->mdoc = curp->pmdoc;
1.103 schwarze 919: return;
1.19 kristaps 920: }
921:
1.112 kristaps 922: if (NULL == curp->pman)
923: curp->pman = man_alloc(&curp->regs, curp, mmsg);
924: assert(curp->pman);
925: curp->man = curp->pman;
1.10 kristaps 926: }
927:
928: static int
929: moptions(enum intt *tflags, char *arg)
930: {
931:
1.17 kristaps 932: if (0 == strcmp(arg, "doc"))
1.10 kristaps 933: *tflags = INTT_MDOC;
1.19 kristaps 934: else if (0 == strcmp(arg, "andoc"))
935: *tflags = INTT_AUTO;
1.17 kristaps 936: else if (0 == strcmp(arg, "an"))
1.10 kristaps 937: *tflags = INTT_MAN;
938: else {
1.57 kristaps 939: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 940: return(0);
941: }
942:
943: return(1);
1.1 kristaps 944: }
945:
946: static int
1.60 kristaps 947: toptions(struct curparse *curp, char *arg)
1.1 kristaps 948: {
949:
950: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 951: curp->outtype = OUTT_ASCII;
952: else if (0 == strcmp(arg, "lint")) {
953: curp->outtype = OUTT_LINT;
1.103 schwarze 954: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 955: }
1.1 kristaps 956: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 957: curp->outtype = OUTT_TREE;
1.43 kristaps 958: else if (0 == strcmp(arg, "html"))
1.60 kristaps 959: curp->outtype = OUTT_HTML;
1.59 kristaps 960: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 961: curp->outtype = OUTT_XHTML;
1.85 kristaps 962: else if (0 == strcmp(arg, "ps"))
963: curp->outtype = OUTT_PS;
1.100 kristaps 964: else if (0 == strcmp(arg, "pdf"))
965: curp->outtype = OUTT_PDF;
1.1 kristaps 966: else {
1.57 kristaps 967: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 968: return(0);
969: }
970:
971: return(1);
972: }
973:
974: static int
1.103 schwarze 975: woptions(struct curparse *curp, char *arg)
1.1 kristaps 976: {
1.34 kristaps 977: char *v, *o;
1.103 schwarze 978: const char *toks[6];
1.1 kristaps 979:
1.103 schwarze 980: toks[0] = "stop";
981: toks[1] = "all";
982: toks[2] = "warning";
983: toks[3] = "error";
984: toks[4] = "fatal";
985: toks[5] = NULL;
1.1 kristaps 986:
1.34 kristaps 987: while (*arg) {
988: o = arg;
1.45 kristaps 989: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 990: case (0):
1.103 schwarze 991: curp->wstop = 1;
1.1 kristaps 992: break;
993: case (1):
1.103 schwarze 994: /* FALLTHROUGH */
1.1 kristaps 995: case (2):
1.103 schwarze 996: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 997: break;
1.20 kristaps 998: case (3):
1.103 schwarze 999: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 1000: break;
1001: case (4):
1.103 schwarze 1002: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 1003: break;
1.1 kristaps 1004: default:
1.103 schwarze 1005: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 1006: return(0);
1007: }
1.34 kristaps 1008: }
1.1 kristaps 1009:
1010: return(1);
1011: }
1012:
1.71 kristaps 1013: static int
1014: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
1015: {
1016: struct curparse *cp;
1.103 schwarze 1017: enum mandoclevel level;
1018:
1019: level = MANDOCLEVEL_FATAL;
1020: while (t < mandoclimits[level])
1.105 kristaps 1021: /* LINTED */
1.103 schwarze 1022: level--;
1.71 kristaps 1023:
1024: cp = (struct curparse *)arg;
1.103 schwarze 1025: if (level < cp->wlevel)
1026: return(1);
1.73 kristaps 1027:
1.103 schwarze 1028: fprintf(stderr, "%s:%d:%d: %s: %s",
1029: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 1030: if (msg)
1031: fprintf(stderr, ": %s", msg);
1032: fputc('\n', stderr);
1.79 kristaps 1033:
1.115 schwarze 1034: if (file_status < level)
1035: file_status = level;
1.103 schwarze 1036:
1037: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 1038: }
CVSweb