Annotation of mandoc/main.c, Revision 1.131
1.131 ! kristaps 1: /* $Id: main.c,v 1.130 2011/01/01 16:18:39 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",
1.131 ! kristaps 179:
! 180: /* related to tables */
! 181: "extra data cells",
1.121 kristaps 182:
183: "generic error",
184:
1.131 ! kristaps 185: /* related to tables */
1.126 kristaps 186: "bad table syntax",
187: "bad table option",
1.127 kristaps 188: "bad table layout",
189: "no table layout cells specified",
1.129 kristaps 190: "no table data cells specified",
1.122 schwarze 191: "input stack limit exceeded, infinite loop?",
1.121 kristaps 192: "skipping bad character",
193: "skipping text before the first section header",
194: "skipping unknown macro",
1.110 kristaps 195: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 196: "line scope broken",
197: "argument count wrong",
1.121 kristaps 198: "skipping end of block that is not open",
199: "missing end of block",
1.106 schwarze 200: "scope open on exit",
1.117 kristaps 201: "uname(3) system call failed",
1.79 kristaps 202: "macro requires line argument(s)",
203: "macro requires body argument(s)",
204: "macro requires argument(s)",
1.82 kristaps 205: "missing list type",
1.79 kristaps 206: "line argument(s) will be lost",
207: "body argument(s) will be lost",
1.94 schwarze 208:
209: "generic fatal error",
210:
1.80 kristaps 211: "column syntax is inconsistent",
1.121 kristaps 212: "NOT IMPLEMENTED: .Bd -file",
1.79 kristaps 213: "line scope broken, syntax violated",
214: "argument count wrong, violates syntax",
215: "child violates parent syntax",
216: "argument count wrong, violates syntax",
1.113 kristaps 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 218: "no document body",
219: "no document prologue",
1.103 schwarze 220: "static buffer exhausted",
1.5 kristaps 221: };
1.1 kristaps 222:
1.111 kristaps 223: static void parsebuf(struct curparse *, struct buf, int);
224: static void pdesc(struct curparse *);
1.66 kristaps 225: static void fdesc(struct curparse *);
226: static void ffile(const char *, struct curparse *);
1.111 kristaps 227: static int pfile(const char *, struct curparse *);
1.10 kristaps 228: static int moptions(enum intt *, char *);
1.71 kristaps 229: static int mmsg(enum mandocerr, void *,
230: int, int, const char *);
1.112 kristaps 231: static void pset(const char *, int, struct curparse *);
1.66 kristaps 232: static int toptions(struct curparse *, char *);
233: static void usage(void) __attribute__((noreturn));
1.55 kristaps 234: static void version(void) __attribute__((noreturn));
1.103 schwarze 235: static int woptions(struct curparse *, char *);
1.1 kristaps 236:
1.54 kristaps 237: static const char *progname;
1.115 schwarze 238: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 239: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 240:
241: int
242: main(int argc, char *argv[])
243: {
1.64 kristaps 244: int c;
1.5 kristaps 245: struct curparse curp;
1.1 kristaps 246:
1.54 kristaps 247: progname = strrchr(argv[0], '/');
248: if (progname == NULL)
249: progname = argv[0];
250: else
251: ++progname;
252:
1.51 kristaps 253: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 254:
1.19 kristaps 255: curp.inttype = INTT_AUTO;
1.22 kristaps 256: curp.outtype = OUTT_ASCII;
1.103 schwarze 257: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 258:
1.1 kristaps 259: /* LINTED */
1.103 schwarze 260: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 261: switch (c) {
1.10 kristaps 262: case ('m'):
1.19 kristaps 263: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 264: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 265: break;
1.48 kristaps 266: case ('O'):
1.49 kristaps 267: (void)strlcat(curp.outopts, optarg, BUFSIZ);
268: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 269: break;
1.1 kristaps 270: case ('T'):
1.60 kristaps 271: if ( ! toptions(&curp, optarg))
1.105 kristaps 272: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 273: break;
274: case ('W'):
1.103 schwarze 275: if ( ! woptions(&curp, optarg))
1.105 kristaps 276: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 277: break;
278: case ('V'):
279: version();
280: /* NOTREACHED */
281: default:
282: usage();
283: /* NOTREACHED */
284: }
285:
286: argc -= optind;
287: argv += optind;
288:
1.30 kristaps 289: if (NULL == *argv) {
290: curp.file = "<stdin>";
291: curp.fd = STDIN_FILENO;
1.39 kristaps 292:
1.66 kristaps 293: fdesc(&curp);
1.64 kristaps 294: }
295:
296: while (*argv) {
1.66 kristaps 297: ffile(*argv, &curp);
1.103 schwarze 298: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 299: break;
300: ++argv;
1.1 kristaps 301: }
302:
1.22 kristaps 303: if (curp.outfree)
304: (*curp.outfree)(curp.outdata);
1.112 kristaps 305: if (curp.pmdoc)
306: mdoc_free(curp.pmdoc);
307: if (curp.pman)
308: man_free(curp.pman);
1.71 kristaps 309: if (curp.roff)
310: roff_free(curp.roff);
1.1 kristaps 311:
1.105 kristaps 312: return((int)exit_status);
1.1 kristaps 313: }
314:
315:
1.55 kristaps 316: static void
1.1 kristaps 317: version(void)
318: {
319:
1.54 kristaps 320: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 321: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 322: }
323:
324:
1.55 kristaps 325: static void
1.1 kristaps 326: usage(void)
327: {
328:
1.112 kristaps 329: (void)fprintf(stderr, "usage: %s "
330: "[-V] "
331: "[-foption] "
332: "[-mformat] "
333: "[-Ooption] "
334: "[-Toutput] "
335: "[-Werr] "
336: "[file...]\n",
337: progname);
338:
1.105 kristaps 339: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 340: }
341:
1.64 kristaps 342: static void
1.66 kristaps 343: ffile(const char *file, struct curparse *curp)
1.1 kristaps 344: {
345:
1.112 kristaps 346: /*
347: * Called once per input file. Get the file ready for reading,
348: * pass it through to the parser-driver, then close it out.
349: * XXX: don't do anything special as this is only called for
350: * files; stdin goes directly to fdesc().
351: */
352:
1.19 kristaps 353: curp->file = file;
1.112 kristaps 354:
1.19 kristaps 355: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 356: perror(curp->file);
1.103 schwarze 357: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 358: return;
1.1 kristaps 359: }
360:
1.66 kristaps 361: fdesc(curp);
1.1 kristaps 362:
1.19 kristaps 363: if (-1 == close(curp->fd))
1.53 kristaps 364: perror(curp->file);
1.1 kristaps 365: }
366:
1.111 kristaps 367: static int
368: pfile(const char *file, struct curparse *curp)
369: {
370: const char *savefile;
371: int fd, savefd;
372:
373: if (-1 == (fd = open(file, O_RDONLY, 0))) {
374: perror(file);
1.115 schwarze 375: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 376: return(0);
377: }
378:
379: savefile = curp->file;
380: savefd = curp->fd;
381:
382: curp->file = file;
383: curp->fd = fd;
384:
385: pdesc(curp);
386:
387: curp->file = savefile;
388: curp->fd = savefd;
389:
390: if (-1 == close(fd))
391: perror(file);
392:
1.115 schwarze 393: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 394: }
395:
1.1 kristaps 396:
1.103 schwarze 397: static void
1.68 joerg 398: resize_buf(struct buf *buf, size_t initial)
399: {
400:
1.124 schwarze 401: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.103 schwarze 402: buf->buf = realloc(buf->buf, buf->sz);
403: if (NULL == buf->buf) {
1.68 joerg 404: perror(NULL);
1.105 kristaps 405: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 406: }
407: }
408:
409:
410: static int
1.66 kristaps 411: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
412: {
413: struct stat st;
1.68 joerg 414: size_t off;
1.66 kristaps 415: ssize_t ssz;
416:
417: if (-1 == fstat(curp->fd, &st)) {
418: perror(curp->file);
419: return(0);
420: }
421:
422: /*
423: * If we're a regular file, try just reading in the whole entry
424: * via mmap(). This is faster than reading it into blocks, and
425: * since each file is only a few bytes to begin with, I'm not
426: * concerned that this is going to tank any machines.
427: */
428:
429: if (S_ISREG(st.st_mode)) {
430: if (st.st_size >= (1U << 31)) {
431: fprintf(stderr, "%s: input too large\n",
432: curp->file);
433: return(0);
434: }
435: *with_mmap = 1;
1.71 kristaps 436: fb->sz = (size_t)st.st_size;
1.66 kristaps 437: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 438: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 439: if (fb->buf != MAP_FAILED)
440: return(1);
441: }
442:
443: /*
444: * If this isn't a regular file (like, say, stdin), then we must
445: * go the old way and just read things in bit by bit.
446: */
447:
448: *with_mmap = 0;
449: off = 0;
450: fb->sz = 0;
451: fb->buf = NULL;
452: for (;;) {
453: if (off == fb->sz) {
454: if (fb->sz == (1U << 31)) {
455: fprintf(stderr, "%s: input too large\n",
456: curp->file);
457: break;
458: }
1.103 schwarze 459: resize_buf(fb, 65536);
1.66 kristaps 460: }
1.71 kristaps 461: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 462: if (ssz == 0) {
463: fb->sz = off;
464: return(1);
465: }
466: if (ssz == -1) {
467: perror(curp->file);
468: break;
469: }
1.71 kristaps 470: off += (size_t)ssz;
1.66 kristaps 471: }
472:
473: free(fb->buf);
474: fb->buf = NULL;
475: return(0);
476: }
477:
478:
1.64 kristaps 479: static void
1.66 kristaps 480: fdesc(struct curparse *curp)
1.1 kristaps 481: {
1.112 kristaps 482:
483: /*
484: * Called once per file with an opened file descriptor. All
485: * pre-file-parse operations (whether stdin or a file) should go
486: * here.
487: *
488: * This calls down into the nested parser, which drills down and
489: * fully parses a file and all its dependences (i.e., `so'). It
490: * then runs the cleanup validators and pushes to output.
491: */
492:
493: /* Zero the parse type. */
494:
495: curp->mdoc = NULL;
496: curp->man = NULL;
1.115 schwarze 497: file_status = MANDOCLEVEL_OK;
1.112 kristaps 498:
499: /* Make sure the mandotory roff parser is initialised. */
500:
501: if (NULL == curp->roff) {
502: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
503: assert(curp->roff);
504: }
505:
506: /* Fully parse the file. */
1.10 kristaps 507:
1.111 kristaps 508: pdesc(curp);
1.92 kristaps 509:
1.115 schwarze 510: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 511: goto cleanup;
512:
513: /* NOTE a parser may not have been assigned, yet. */
514:
1.112 kristaps 515: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 516: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 517: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 518: goto cleanup;
519: }
520:
521: /* Clean up the parse routine ASTs. */
522:
1.112 kristaps 523: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 524: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 525: goto cleanup;
526: }
1.112 kristaps 527:
528: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 529: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 530: goto cleanup;
531: }
1.112 kristaps 532:
533: assert(curp->roff);
1.130 kristaps 534: roff_endparse(curp->roff);
1.1 kristaps 535:
536: /*
1.111 kristaps 537: * With -Wstop and warnings or errors of at least
538: * the requested level, do not produce output.
1.1 kristaps 539: */
540:
1.115 schwarze 541: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 542: goto cleanup;
543:
544: /* If unset, allocate output dev now (if applicable). */
545:
546: if ( ! (curp->outman && curp->outmdoc)) {
547: switch (curp->outtype) {
548: case (OUTT_XHTML):
549: curp->outdata = xhtml_alloc(curp->outopts);
550: break;
551: case (OUTT_HTML):
552: curp->outdata = html_alloc(curp->outopts);
553: break;
554: case (OUTT_ASCII):
555: curp->outdata = ascii_alloc(curp->outopts);
556: curp->outfree = ascii_free;
557: break;
558: case (OUTT_PDF):
559: curp->outdata = pdf_alloc(curp->outopts);
560: curp->outfree = pspdf_free;
561: break;
562: case (OUTT_PS):
563: curp->outdata = ps_alloc(curp->outopts);
564: curp->outfree = pspdf_free;
565: break;
566: default:
567: break;
568: }
569:
570: switch (curp->outtype) {
571: case (OUTT_HTML):
572: /* FALLTHROUGH */
573: case (OUTT_XHTML):
574: curp->outman = html_man;
575: curp->outmdoc = html_mdoc;
576: curp->outfree = html_free;
577: break;
578: case (OUTT_TREE):
579: curp->outman = tree_man;
580: curp->outmdoc = tree_mdoc;
581: break;
582: case (OUTT_PDF):
583: /* FALLTHROUGH */
584: case (OUTT_ASCII):
585: /* FALLTHROUGH */
586: case (OUTT_PS):
587: curp->outman = terminal_man;
588: curp->outmdoc = terminal_mdoc;
589: break;
590: default:
591: break;
592: }
593: }
594:
595: /* Execute the out device, if it exists. */
596:
1.112 kristaps 597: if (curp->man && curp->outman)
598: (*curp->outman)(curp->outdata, curp->man);
599: if (curp->mdoc && curp->outmdoc)
600: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 601:
602: cleanup:
1.112 kristaps 603:
1.111 kristaps 604: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 605:
606: /* Reset the current-parse compilers. */
607:
608: if (curp->mdoc)
609: mdoc_reset(curp->mdoc);
610: if (curp->man)
611: man_reset(curp->man);
612:
613: assert(curp->roff);
614: roff_reset(curp->roff);
1.111 kristaps 615:
1.115 schwarze 616: if (exit_status < file_status)
617: exit_status = file_status;
618:
1.111 kristaps 619: return;
620: }
621:
622: static void
623: pdesc(struct curparse *curp)
624: {
625: struct buf blk;
626: int with_mmap;
627:
1.112 kristaps 628: /*
629: * Run for each opened file; may be called more than once for
630: * each full parse sequence if the opened file is nested (i.e.,
631: * from `so'). Simply sucks in the whole file and moves into
632: * the parse phase for the file.
633: */
634:
1.103 schwarze 635: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 636: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 637: return;
1.103 schwarze 638: }
1.1 kristaps 639:
1.112 kristaps 640: /* Line number is per-file. */
1.111 kristaps 641:
642: curp->line = 1;
1.112 kristaps 643:
1.111 kristaps 644: parsebuf(curp, blk, 1);
645:
646: if (with_mmap)
647: munmap(blk.buf, blk.sz);
648: else
649: free(blk.buf);
650: }
651:
652: static void
653: parsebuf(struct curparse *curp, struct buf blk, int start)
654: {
655: struct buf ln;
1.114 kristaps 656: enum rofferr rr;
1.112 kristaps 657: int i, of, rc;
658: int pos; /* byte number in the ln buffer */
659: int lnn; /* line number in the real file */
1.111 kristaps 660: unsigned char c;
1.112 kristaps 661:
662: /*
663: * Main parse routine for an opened file. This is called for
664: * each opened file and simply loops around the full input file,
665: * possibly nesting (i.e., with `so').
666: */
1.71 kristaps 667:
1.111 kristaps 668: memset(&ln, 0, sizeof(struct buf));
669:
1.112 kristaps 670: lnn = curp->line;
671: pos = 0;
1.111 kristaps 672:
1.112 kristaps 673: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 674: if (0 == pos && '\0' == blk.buf[i])
675: break;
1.112 kristaps 676:
1.122 schwarze 677: if (start) {
1.111 kristaps 678: curp->line = lnn;
1.122 schwarze 679: curp->reparse_count = 0;
680: }
1.111 kristaps 681:
682: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 683: if ('\n' == blk.buf[i]) {
684: ++i;
685: ++lnn;
686: break;
687: }
1.99 kristaps 688:
689: /*
690: * Warn about bogus characters. If you're using
691: * non-ASCII encoding, you're screwing your
692: * readers. Since I'd rather this not happen,
693: * I'll be helpful and drop these characters so
694: * we don't display gibberish. Note to manual
695: * writers: use special characters.
696: */
697:
1.102 schwarze 698: c = (unsigned char) blk.buf[i];
1.112 kristaps 699:
700: if ( ! (isascii(c) &&
701: (isgraph(c) || isblank(c)))) {
1.103 schwarze 702: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 703: curp->line, pos, "ignoring byte");
1.99 kristaps 704: i++;
705: continue;
706: }
707:
1.112 kristaps 708: /* Trailing backslash = a plain char. */
709:
1.70 joerg 710: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
711: if (pos >= (int)ln.sz)
1.103 schwarze 712: resize_buf(&ln, 256);
1.70 joerg 713: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 714: continue;
1.70 joerg 715: }
1.112 kristaps 716:
717: /* Found escape & at least one other char. */
718:
1.70 joerg 719: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 720: i += 2;
1.70 joerg 721: /* Escaped newlines are skipped over */
722: ++lnn;
1.67 joerg 723: continue;
724: }
1.112 kristaps 725:
1.70 joerg 726: if ('"' == blk.buf[i + 1]) {
727: i += 2;
728: /* Comment, skip to end of line */
729: for (; i < (int)blk.sz; ++i) {
730: if ('\n' == blk.buf[i]) {
731: ++i;
732: ++lnn;
733: break;
734: }
735: }
1.112 kristaps 736:
1.70 joerg 737: /* Backout trailing whitespaces */
738: for (; pos > 0; --pos) {
739: if (ln.buf[pos - 1] != ' ')
740: break;
741: if (pos > 2 && ln.buf[pos - 2] == '\\')
742: break;
743: }
744: break;
745: }
1.112 kristaps 746:
747: /* Some other escape sequence, copy & cont. */
748:
1.70 joerg 749: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 750: resize_buf(&ln, 256);
1.1 kristaps 751:
1.70 joerg 752: ln.buf[pos++] = blk.buf[i++];
753: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 754: }
1.1 kristaps 755:
1.70 joerg 756: if (pos >= (int)ln.sz)
1.103 schwarze 757: resize_buf(&ln, 256);
1.112 kristaps 758:
1.71 kristaps 759: ln.buf[pos] = '\0';
760:
1.76 kristaps 761: /*
762: * A significant amount of complexity is contained by
763: * the roff preprocessor. It's line-oriented but can be
764: * expressed on one line, so we need at times to
765: * readjust our starting point and re-run it. The roff
766: * preprocessor can also readjust the buffers with new
767: * data, so we pass them in wholesale.
768: */
769:
770: of = 0;
1.112 kristaps 771:
1.111 kristaps 772: rerun:
1.114 kristaps 773: rr = roff_parseln
1.112 kristaps 774: (curp->roff, curp->line,
775: &ln.buf, &ln.sz, of, &of);
776:
1.114 kristaps 777: switch (rr) {
1.111 kristaps 778: case (ROFF_REPARSE):
1.122 schwarze 779: if (REPARSE_LIMIT >= ++curp->reparse_count)
780: parsebuf(curp, ln, 0);
781: else
782: mmsg(MANDOCERR_ROFFLOOP, curp,
783: curp->line, pos, NULL);
1.111 kristaps 784: pos = 0;
785: continue;
786: case (ROFF_APPEND):
787: pos = strlen(ln.buf);
788: continue;
789: case (ROFF_RERUN):
790: goto rerun;
791: case (ROFF_IGN):
792: pos = 0;
1.71 kristaps 793: continue;
1.111 kristaps 794: case (ROFF_ERR):
1.115 schwarze 795: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 796: break;
797: case (ROFF_SO):
798: if (pfile(ln.buf + of, curp)) {
799: pos = 0;
800: continue;
801: } else
802: break;
1.128 kristaps 803: default:
1.111 kristaps 804: break;
1.103 schwarze 805: }
1.29 kristaps 806:
1.76 kristaps 807: /*
808: * If input parsers have not been allocated, do so now.
809: * We keep these instanced betwen parsers, but set them
810: * locally per parse routine since we can use different
811: * parsers with each one.
812: */
1.19 kristaps 813:
1.112 kristaps 814: if ( ! (curp->man || curp->mdoc))
815: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 816:
1.112 kristaps 817: /*
818: * Lastly, push down into the parsers themselves. One
819: * of these will have already been set in the pset()
820: * routine.
1.128 kristaps 821: * If libroff returns ROFF_TBL, then add it to the
822: * currently open parse. Since we only get here if
823: * there does exist data (see tbl_data.c), we're
824: * guaranteed that something's been allocated.
1.112 kristaps 825: */
826:
1.128 kristaps 827: if (ROFF_TBL == rr) {
828: assert(curp->man || curp->mdoc);
829: if (curp->man)
830: man_addspan(curp->man, roff_span(curp->roff));
831: else
832: mdoc_addspan(curp->mdoc, roff_span(curp->roff));
833:
834: } else if (curp->man || curp->mdoc) {
1.112 kristaps 835: rc = curp->man ?
836: man_parseln(curp->man,
837: curp->line, ln.buf, of) :
838: mdoc_parseln(curp->mdoc,
839: curp->line, ln.buf, of);
1.30 kristaps 840:
1.112 kristaps 841: if ( ! rc) {
1.115 schwarze 842: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 843: break;
844: }
1.103 schwarze 845: }
1.19 kristaps 846:
1.111 kristaps 847: /* Temporary buffers typically are not full. */
1.112 kristaps 848:
1.111 kristaps 849: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 850: break;
1.85 kristaps 851:
1.111 kristaps 852: /* Start the next input line. */
1.112 kristaps 853:
1.111 kristaps 854: pos = 0;
1.22 kristaps 855: }
856:
1.111 kristaps 857: free(ln.buf);
1.19 kristaps 858: }
859:
1.103 schwarze 860: static void
1.112 kristaps 861: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 862: {
1.29 kristaps 863: int i;
1.19 kristaps 864:
865: /*
866: * Try to intuit which kind of manual parser should be used. If
867: * passed in by command-line (-man, -mdoc), then use that
868: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 869: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 870: * default to -man, which is more lenient.
1.112 kristaps 871: *
872: * Separate out pmdoc/pman from mdoc/man: the first persists
873: * through all parsers, while the latter is used per-parse.
1.19 kristaps 874: */
875:
1.75 kristaps 876: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 877: for (i = 1; buf[i]; i++)
878: if (' ' != buf[i] && '\t' != buf[i])
879: break;
1.103 schwarze 880: if ('\0' == buf[i])
881: return;
1.29 kristaps 882: }
1.10 kristaps 883:
1.19 kristaps 884: switch (curp->inttype) {
885: case (INTT_MDOC):
1.112 kristaps 886: if (NULL == curp->pmdoc)
887: curp->pmdoc = mdoc_alloc
888: (&curp->regs, curp, mmsg);
889: assert(curp->pmdoc);
890: curp->mdoc = curp->pmdoc;
1.103 schwarze 891: return;
1.19 kristaps 892: case (INTT_MAN):
1.112 kristaps 893: if (NULL == curp->pman)
894: curp->pman = man_alloc
895: (&curp->regs, curp, mmsg);
896: assert(curp->pman);
897: curp->man = curp->pman;
1.103 schwarze 898: return;
1.19 kristaps 899: default:
900: break;
901: }
902:
903: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 904: if (NULL == curp->pmdoc)
905: curp->pmdoc = mdoc_alloc
906: (&curp->regs, curp, mmsg);
907: assert(curp->pmdoc);
908: curp->mdoc = curp->pmdoc;
1.103 schwarze 909: return;
1.19 kristaps 910: }
911:
1.112 kristaps 912: if (NULL == curp->pman)
913: curp->pman = man_alloc(&curp->regs, curp, mmsg);
914: assert(curp->pman);
915: curp->man = curp->pman;
1.10 kristaps 916: }
917:
918: static int
919: moptions(enum intt *tflags, char *arg)
920: {
921:
1.17 kristaps 922: if (0 == strcmp(arg, "doc"))
1.10 kristaps 923: *tflags = INTT_MDOC;
1.19 kristaps 924: else if (0 == strcmp(arg, "andoc"))
925: *tflags = INTT_AUTO;
1.17 kristaps 926: else if (0 == strcmp(arg, "an"))
1.10 kristaps 927: *tflags = INTT_MAN;
928: else {
1.57 kristaps 929: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 930: return(0);
931: }
932:
933: return(1);
1.1 kristaps 934: }
935:
936: static int
1.60 kristaps 937: toptions(struct curparse *curp, char *arg)
1.1 kristaps 938: {
939:
940: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 941: curp->outtype = OUTT_ASCII;
942: else if (0 == strcmp(arg, "lint")) {
943: curp->outtype = OUTT_LINT;
1.103 schwarze 944: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 945: }
1.1 kristaps 946: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 947: curp->outtype = OUTT_TREE;
1.43 kristaps 948: else if (0 == strcmp(arg, "html"))
1.60 kristaps 949: curp->outtype = OUTT_HTML;
1.59 kristaps 950: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 951: curp->outtype = OUTT_XHTML;
1.85 kristaps 952: else if (0 == strcmp(arg, "ps"))
953: curp->outtype = OUTT_PS;
1.100 kristaps 954: else if (0 == strcmp(arg, "pdf"))
955: curp->outtype = OUTT_PDF;
1.1 kristaps 956: else {
1.57 kristaps 957: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 958: return(0);
959: }
960:
961: return(1);
962: }
963:
964: static int
1.103 schwarze 965: woptions(struct curparse *curp, char *arg)
1.1 kristaps 966: {
1.34 kristaps 967: char *v, *o;
1.103 schwarze 968: const char *toks[6];
1.1 kristaps 969:
1.103 schwarze 970: toks[0] = "stop";
971: toks[1] = "all";
972: toks[2] = "warning";
973: toks[3] = "error";
974: toks[4] = "fatal";
975: toks[5] = NULL;
1.1 kristaps 976:
1.34 kristaps 977: while (*arg) {
978: o = arg;
1.45 kristaps 979: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 980: case (0):
1.103 schwarze 981: curp->wstop = 1;
1.1 kristaps 982: break;
983: case (1):
1.103 schwarze 984: /* FALLTHROUGH */
1.1 kristaps 985: case (2):
1.103 schwarze 986: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 987: break;
1.20 kristaps 988: case (3):
1.103 schwarze 989: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 990: break;
991: case (4):
1.103 schwarze 992: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 993: break;
1.1 kristaps 994: default:
1.103 schwarze 995: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 996: return(0);
997: }
1.34 kristaps 998: }
1.1 kristaps 999:
1000: return(1);
1001: }
1002:
1.71 kristaps 1003: static int
1004: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
1005: {
1006: struct curparse *cp;
1.103 schwarze 1007: enum mandoclevel level;
1008:
1009: level = MANDOCLEVEL_FATAL;
1010: while (t < mandoclimits[level])
1.105 kristaps 1011: /* LINTED */
1.103 schwarze 1012: level--;
1.71 kristaps 1013:
1014: cp = (struct curparse *)arg;
1.103 schwarze 1015: if (level < cp->wlevel)
1016: return(1);
1.73 kristaps 1017:
1.103 schwarze 1018: fprintf(stderr, "%s:%d:%d: %s: %s",
1019: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 1020: if (msg)
1021: fprintf(stderr, ": %s", msg);
1022: fputc('\n', stderr);
1.79 kristaps 1023:
1.115 schwarze 1024: if (file_status < level)
1025: file_status = level;
1.103 schwarze 1026:
1027: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 1028: }
CVSweb