Annotation of mandoc/main.c, Revision 1.147
1.147 ! schwarze 1: /* $Id: main.c,v 1.146 2011/02/09 09:33:43 kristaps Exp $ */
1.1 kristaps 2: /*
1.133 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.140 schwarze 4: * Copyright (c) 2010, 2011 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",
1.147 ! schwarze 131: "date missing, using today's date",
! 132: "cannot parse date, using it verbatim",
1.121 kristaps 133: "prologue macros out of order",
134: "duplicate prologue macro",
135: "macro not allowed in prologue",
136: "macro not allowed in body",
137:
138: /* related to document structure */
1.113 kristaps 139: ".so is fragile, better use ln(1)",
1.121 kristaps 140: "NAME section must come first",
141: "bad NAME section contents",
142: "manual name not yet set",
1.81 kristaps 143: "sections out of conventional order",
1.121 kristaps 144: "duplicate section name",
1.79 kristaps 145: "section not in conventional manual section",
1.121 kristaps 146:
147: /* related to macros and nesting */
148: "skipping obsolete macro",
1.125 kristaps 149: "skipping paragraph macro",
1.142 kristaps 150: "skipping no-space macro",
1.95 schwarze 151: "blocks badly nested",
1.121 kristaps 152: "child violates parent syntax",
153: "nested displays are not portable",
154: "already in literal mode",
1.94 schwarze 155:
1.121 kristaps 156: /* related to missing macro arguments */
157: "skipping empty macro",
1.133 schwarze 158: "argument count wrong",
1.121 kristaps 159: "missing display type",
160: "list type must come first",
161: "tag lists require a width argument",
162: "missing font type",
1.138 kristaps 163: "skipping end of block that is not open",
1.94 schwarze 164:
1.121 kristaps 165: /* related to bad macro arguments */
166: "skipping argument",
167: "duplicate argument",
168: "duplicate display type",
169: "duplicate list type",
170: "unknown AT&T UNIX version",
1.79 kristaps 171: "bad Boolean value",
1.120 kristaps 172: "unknown font",
1.121 kristaps 173: "unknown standard specifier",
174: "bad width argument",
175:
176: /* related to plain text */
177: "blank line in non-literal context",
178: "tab in non-literal context",
179: "end of line whitespace",
1.79 kristaps 180: "bad comment style",
1.121 kristaps 181: "unknown escape sequence",
182: "unterminated quoted string",
1.131 kristaps 183:
1.121 kristaps 184: "generic error",
185:
1.131 kristaps 186: /* related to tables */
1.126 kristaps 187: "bad table syntax",
188: "bad table option",
1.127 kristaps 189: "bad table layout",
190: "no table layout cells specified",
1.129 kristaps 191: "no table data cells specified",
1.134 kristaps 192: "ignore data in cell",
1.135 kristaps 193: "data block still open",
1.136 kristaps 194: "ignoring extra data cells",
1.134 kristaps 195:
1.122 schwarze 196: "input stack limit exceeded, infinite loop?",
1.121 kristaps 197: "skipping bad character",
1.137 schwarze 198: "escaped character not allowed in a name",
1.121 kristaps 199: "skipping text before the first section header",
200: "skipping unknown macro",
1.139 schwarze 201: "NOT IMPLEMENTED, please use groff: skipping request",
1.79 kristaps 202: "line scope broken",
203: "argument count wrong",
1.121 kristaps 204: "skipping end of block that is not open",
205: "missing end of block",
1.106 schwarze 206: "scope open on exit",
1.117 kristaps 207: "uname(3) system call failed",
1.79 kristaps 208: "macro requires line argument(s)",
209: "macro requires body argument(s)",
210: "macro requires argument(s)",
1.82 kristaps 211: "missing list type",
1.79 kristaps 212: "line argument(s) will be lost",
213: "body argument(s) will be lost",
1.94 schwarze 214:
215: "generic fatal error",
216:
1.80 kristaps 217: "column syntax is inconsistent",
1.121 kristaps 218: "NOT IMPLEMENTED: .Bd -file",
1.79 kristaps 219: "line scope broken, syntax violated",
220: "argument count wrong, violates syntax",
221: "child violates parent syntax",
222: "argument count wrong, violates syntax",
1.113 kristaps 223: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 224: "no document body",
225: "no document prologue",
1.103 schwarze 226: "static buffer exhausted",
1.5 kristaps 227: };
1.1 kristaps 228:
1.111 kristaps 229: static void parsebuf(struct curparse *, struct buf, int);
230: static void pdesc(struct curparse *);
1.66 kristaps 231: static void fdesc(struct curparse *);
232: static void ffile(const char *, struct curparse *);
1.111 kristaps 233: static int pfile(const char *, struct curparse *);
1.10 kristaps 234: static int moptions(enum intt *, char *);
1.71 kristaps 235: static int mmsg(enum mandocerr, void *,
236: int, int, const char *);
1.112 kristaps 237: static void pset(const char *, int, struct curparse *);
1.66 kristaps 238: static int toptions(struct curparse *, char *);
239: static void usage(void) __attribute__((noreturn));
1.55 kristaps 240: static void version(void) __attribute__((noreturn));
1.103 schwarze 241: static int woptions(struct curparse *, char *);
1.1 kristaps 242:
1.54 kristaps 243: static const char *progname;
1.115 schwarze 244: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 245: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 246:
247: int
248: main(int argc, char *argv[])
249: {
1.64 kristaps 250: int c;
1.5 kristaps 251: struct curparse curp;
1.1 kristaps 252:
1.54 kristaps 253: progname = strrchr(argv[0], '/');
254: if (progname == NULL)
255: progname = argv[0];
256: else
257: ++progname;
258:
1.51 kristaps 259: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 260:
1.19 kristaps 261: curp.inttype = INTT_AUTO;
1.22 kristaps 262: curp.outtype = OUTT_ASCII;
1.103 schwarze 263: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 264:
1.1 kristaps 265: /* LINTED */
1.103 schwarze 266: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 267: switch (c) {
1.10 kristaps 268: case ('m'):
1.19 kristaps 269: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 270: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 271: break;
1.48 kristaps 272: case ('O'):
1.49 kristaps 273: (void)strlcat(curp.outopts, optarg, BUFSIZ);
274: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 275: break;
1.1 kristaps 276: case ('T'):
1.60 kristaps 277: if ( ! toptions(&curp, optarg))
1.105 kristaps 278: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 279: break;
280: case ('W'):
1.103 schwarze 281: if ( ! woptions(&curp, optarg))
1.105 kristaps 282: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 283: break;
284: case ('V'):
285: version();
286: /* NOTREACHED */
287: default:
288: usage();
289: /* NOTREACHED */
290: }
291:
292: argc -= optind;
293: argv += optind;
294:
1.30 kristaps 295: if (NULL == *argv) {
296: curp.file = "<stdin>";
297: curp.fd = STDIN_FILENO;
1.39 kristaps 298:
1.66 kristaps 299: fdesc(&curp);
1.64 kristaps 300: }
301:
302: while (*argv) {
1.66 kristaps 303: ffile(*argv, &curp);
1.103 schwarze 304: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 305: break;
306: ++argv;
1.1 kristaps 307: }
308:
1.22 kristaps 309: if (curp.outfree)
310: (*curp.outfree)(curp.outdata);
1.112 kristaps 311: if (curp.pmdoc)
312: mdoc_free(curp.pmdoc);
313: if (curp.pman)
314: man_free(curp.pman);
1.71 kristaps 315: if (curp.roff)
316: roff_free(curp.roff);
1.1 kristaps 317:
1.105 kristaps 318: return((int)exit_status);
1.1 kristaps 319: }
320:
321:
1.55 kristaps 322: static void
1.1 kristaps 323: version(void)
324: {
325:
1.54 kristaps 326: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 327: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 328: }
329:
330:
1.55 kristaps 331: static void
1.1 kristaps 332: usage(void)
333: {
334:
1.112 kristaps 335: (void)fprintf(stderr, "usage: %s "
336: "[-V] "
337: "[-foption] "
338: "[-mformat] "
339: "[-Ooption] "
340: "[-Toutput] "
341: "[-Werr] "
342: "[file...]\n",
343: progname);
344:
1.105 kristaps 345: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 346: }
347:
1.64 kristaps 348: static void
1.66 kristaps 349: ffile(const char *file, struct curparse *curp)
1.1 kristaps 350: {
351:
1.112 kristaps 352: /*
353: * Called once per input file. Get the file ready for reading,
354: * pass it through to the parser-driver, then close it out.
355: * XXX: don't do anything special as this is only called for
356: * files; stdin goes directly to fdesc().
357: */
358:
1.19 kristaps 359: curp->file = file;
1.112 kristaps 360:
1.19 kristaps 361: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 362: perror(curp->file);
1.103 schwarze 363: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 364: return;
1.1 kristaps 365: }
366:
1.66 kristaps 367: fdesc(curp);
1.1 kristaps 368:
1.19 kristaps 369: if (-1 == close(curp->fd))
1.53 kristaps 370: perror(curp->file);
1.1 kristaps 371: }
372:
1.111 kristaps 373: static int
374: pfile(const char *file, struct curparse *curp)
375: {
376: const char *savefile;
377: int fd, savefd;
378:
379: if (-1 == (fd = open(file, O_RDONLY, 0))) {
380: perror(file);
1.115 schwarze 381: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 382: return(0);
383: }
384:
385: savefile = curp->file;
386: savefd = curp->fd;
387:
388: curp->file = file;
389: curp->fd = fd;
390:
391: pdesc(curp);
392:
393: curp->file = savefile;
394: curp->fd = savefd;
395:
396: if (-1 == close(fd))
397: perror(file);
398:
1.115 schwarze 399: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 400: }
401:
1.1 kristaps 402:
1.103 schwarze 403: static void
1.68 joerg 404: resize_buf(struct buf *buf, size_t initial)
405: {
406:
1.124 schwarze 407: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.103 schwarze 408: buf->buf = realloc(buf->buf, buf->sz);
409: if (NULL == buf->buf) {
1.68 joerg 410: perror(NULL);
1.105 kristaps 411: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 412: }
413: }
414:
415:
416: static int
1.66 kristaps 417: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
418: {
419: struct stat st;
1.68 joerg 420: size_t off;
1.66 kristaps 421: ssize_t ssz;
422:
423: if (-1 == fstat(curp->fd, &st)) {
424: perror(curp->file);
425: return(0);
426: }
427:
428: /*
429: * If we're a regular file, try just reading in the whole entry
430: * via mmap(). This is faster than reading it into blocks, and
431: * since each file is only a few bytes to begin with, I'm not
432: * concerned that this is going to tank any machines.
433: */
434:
435: if (S_ISREG(st.st_mode)) {
436: if (st.st_size >= (1U << 31)) {
437: fprintf(stderr, "%s: input too large\n",
438: curp->file);
439: return(0);
440: }
441: *with_mmap = 1;
1.71 kristaps 442: fb->sz = (size_t)st.st_size;
1.66 kristaps 443: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 444: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 445: if (fb->buf != MAP_FAILED)
446: return(1);
447: }
448:
449: /*
450: * If this isn't a regular file (like, say, stdin), then we must
451: * go the old way and just read things in bit by bit.
452: */
453:
454: *with_mmap = 0;
455: off = 0;
456: fb->sz = 0;
457: fb->buf = NULL;
458: for (;;) {
459: if (off == fb->sz) {
460: if (fb->sz == (1U << 31)) {
461: fprintf(stderr, "%s: input too large\n",
462: curp->file);
463: break;
464: }
1.103 schwarze 465: resize_buf(fb, 65536);
1.66 kristaps 466: }
1.71 kristaps 467: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 468: if (ssz == 0) {
469: fb->sz = off;
470: return(1);
471: }
472: if (ssz == -1) {
473: perror(curp->file);
474: break;
475: }
1.71 kristaps 476: off += (size_t)ssz;
1.66 kristaps 477: }
478:
479: free(fb->buf);
480: fb->buf = NULL;
481: return(0);
482: }
483:
484:
1.64 kristaps 485: static void
1.66 kristaps 486: fdesc(struct curparse *curp)
1.1 kristaps 487: {
1.112 kristaps 488:
489: /*
490: * Called once per file with an opened file descriptor. All
491: * pre-file-parse operations (whether stdin or a file) should go
492: * here.
493: *
494: * This calls down into the nested parser, which drills down and
495: * fully parses a file and all its dependences (i.e., `so'). It
496: * then runs the cleanup validators and pushes to output.
497: */
498:
499: /* Zero the parse type. */
500:
501: curp->mdoc = NULL;
502: curp->man = NULL;
1.115 schwarze 503: file_status = MANDOCLEVEL_OK;
1.112 kristaps 504:
505: /* Make sure the mandotory roff parser is initialised. */
506:
507: if (NULL == curp->roff) {
508: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
509: assert(curp->roff);
510: }
511:
512: /* Fully parse the file. */
1.10 kristaps 513:
1.111 kristaps 514: pdesc(curp);
1.92 kristaps 515:
1.115 schwarze 516: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 517: goto cleanup;
518:
519: /* NOTE a parser may not have been assigned, yet. */
520:
1.112 kristaps 521: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 522: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 523: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 524: goto cleanup;
525: }
526:
527: /* Clean up the parse routine ASTs. */
528:
1.112 kristaps 529: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 530: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 531: goto cleanup;
532: }
1.112 kristaps 533:
534: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 535: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 536: goto cleanup;
537: }
1.112 kristaps 538:
539: assert(curp->roff);
1.130 kristaps 540: roff_endparse(curp->roff);
1.1 kristaps 541:
542: /*
1.111 kristaps 543: * With -Wstop and warnings or errors of at least
544: * the requested level, do not produce output.
1.1 kristaps 545: */
546:
1.115 schwarze 547: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 548: goto cleanup;
549:
550: /* If unset, allocate output dev now (if applicable). */
551:
552: if ( ! (curp->outman && curp->outmdoc)) {
553: switch (curp->outtype) {
554: case (OUTT_XHTML):
555: curp->outdata = xhtml_alloc(curp->outopts);
556: break;
557: case (OUTT_HTML):
558: curp->outdata = html_alloc(curp->outopts);
559: break;
560: case (OUTT_ASCII):
561: curp->outdata = ascii_alloc(curp->outopts);
562: curp->outfree = ascii_free;
563: break;
564: case (OUTT_PDF):
565: curp->outdata = pdf_alloc(curp->outopts);
566: curp->outfree = pspdf_free;
567: break;
568: case (OUTT_PS):
569: curp->outdata = ps_alloc(curp->outopts);
570: curp->outfree = pspdf_free;
571: break;
572: default:
573: break;
574: }
575:
576: switch (curp->outtype) {
577: case (OUTT_HTML):
578: /* FALLTHROUGH */
579: case (OUTT_XHTML):
580: curp->outman = html_man;
581: curp->outmdoc = html_mdoc;
582: curp->outfree = html_free;
583: break;
584: case (OUTT_TREE):
585: curp->outman = tree_man;
586: curp->outmdoc = tree_mdoc;
587: break;
588: case (OUTT_PDF):
589: /* FALLTHROUGH */
590: case (OUTT_ASCII):
591: /* FALLTHROUGH */
592: case (OUTT_PS):
593: curp->outman = terminal_man;
594: curp->outmdoc = terminal_mdoc;
595: break;
596: default:
597: break;
598: }
599: }
600:
601: /* Execute the out device, if it exists. */
602:
1.112 kristaps 603: if (curp->man && curp->outman)
604: (*curp->outman)(curp->outdata, curp->man);
605: if (curp->mdoc && curp->outmdoc)
606: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 607:
608: cleanup:
1.112 kristaps 609:
1.111 kristaps 610: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 611:
612: /* Reset the current-parse compilers. */
613:
614: if (curp->mdoc)
615: mdoc_reset(curp->mdoc);
616: if (curp->man)
617: man_reset(curp->man);
618:
619: assert(curp->roff);
620: roff_reset(curp->roff);
1.111 kristaps 621:
1.115 schwarze 622: if (exit_status < file_status)
623: exit_status = file_status;
624:
1.111 kristaps 625: return;
626: }
627:
628: static void
629: pdesc(struct curparse *curp)
630: {
631: struct buf blk;
632: int with_mmap;
633:
1.112 kristaps 634: /*
635: * Run for each opened file; may be called more than once for
636: * each full parse sequence if the opened file is nested (i.e.,
637: * from `so'). Simply sucks in the whole file and moves into
638: * the parse phase for the file.
639: */
640:
1.103 schwarze 641: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 642: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 643: return;
1.103 schwarze 644: }
1.1 kristaps 645:
1.112 kristaps 646: /* Line number is per-file. */
1.111 kristaps 647:
648: curp->line = 1;
1.112 kristaps 649:
1.111 kristaps 650: parsebuf(curp, blk, 1);
651:
652: if (with_mmap)
653: munmap(blk.buf, blk.sz);
654: else
655: free(blk.buf);
656: }
657:
658: static void
659: parsebuf(struct curparse *curp, struct buf blk, int start)
660: {
1.141 schwarze 661: const struct tbl_span *span;
1.111 kristaps 662: struct buf ln;
1.114 kristaps 663: enum rofferr rr;
1.112 kristaps 664: int i, of, rc;
665: int pos; /* byte number in the ln buffer */
666: int lnn; /* line number in the real file */
1.111 kristaps 667: unsigned char c;
1.112 kristaps 668:
669: /*
670: * Main parse routine for an opened file. This is called for
671: * each opened file and simply loops around the full input file,
672: * possibly nesting (i.e., with `so').
673: */
1.71 kristaps 674:
1.111 kristaps 675: memset(&ln, 0, sizeof(struct buf));
676:
1.112 kristaps 677: lnn = curp->line;
678: pos = 0;
1.111 kristaps 679:
1.112 kristaps 680: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 681: if (0 == pos && '\0' == blk.buf[i])
682: break;
1.112 kristaps 683:
1.122 schwarze 684: if (start) {
1.111 kristaps 685: curp->line = lnn;
1.122 schwarze 686: curp->reparse_count = 0;
687: }
1.111 kristaps 688:
689: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.140 schwarze 690:
691: /*
692: * When finding an unescaped newline character,
693: * leave the character loop to process the line.
694: * Skip a preceding carriage return, if any.
695: */
696:
697: if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
698: '\n' == blk.buf[i + 1])
699: ++i;
1.70 joerg 700: if ('\n' == blk.buf[i]) {
701: ++i;
702: ++lnn;
703: break;
704: }
1.99 kristaps 705:
706: /*
707: * Warn about bogus characters. If you're using
708: * non-ASCII encoding, you're screwing your
709: * readers. Since I'd rather this not happen,
710: * I'll be helpful and drop these characters so
711: * we don't display gibberish. Note to manual
712: * writers: use special characters.
713: */
714:
1.102 schwarze 715: c = (unsigned char) blk.buf[i];
1.112 kristaps 716:
717: if ( ! (isascii(c) &&
718: (isgraph(c) || isblank(c)))) {
1.103 schwarze 719: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 720: curp->line, pos, "ignoring byte");
1.99 kristaps 721: i++;
722: continue;
723: }
724:
1.112 kristaps 725: /* Trailing backslash = a plain char. */
726:
1.70 joerg 727: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
728: if (pos >= (int)ln.sz)
1.103 schwarze 729: resize_buf(&ln, 256);
1.70 joerg 730: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 731: continue;
1.70 joerg 732: }
1.112 kristaps 733:
1.140 schwarze 734: /*
735: * Found escape and at least one other character.
736: * When it's a newline character, skip it.
737: * When there is a carriage return in between,
738: * skip that one as well.
739: */
1.112 kristaps 740:
1.140 schwarze 741: if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
742: '\n' == blk.buf[i + 2])
743: ++i;
1.70 joerg 744: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 745: i += 2;
1.70 joerg 746: ++lnn;
1.67 joerg 747: continue;
748: }
1.112 kristaps 749:
1.70 joerg 750: if ('"' == blk.buf[i + 1]) {
751: i += 2;
752: /* Comment, skip to end of line */
753: for (; i < (int)blk.sz; ++i) {
754: if ('\n' == blk.buf[i]) {
755: ++i;
756: ++lnn;
757: break;
758: }
759: }
1.112 kristaps 760:
1.70 joerg 761: /* Backout trailing whitespaces */
762: for (; pos > 0; --pos) {
763: if (ln.buf[pos - 1] != ' ')
764: break;
765: if (pos > 2 && ln.buf[pos - 2] == '\\')
766: break;
767: }
768: break;
769: }
1.112 kristaps 770:
771: /* Some other escape sequence, copy & cont. */
772:
1.70 joerg 773: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 774: resize_buf(&ln, 256);
1.1 kristaps 775:
1.70 joerg 776: ln.buf[pos++] = blk.buf[i++];
777: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 778: }
1.1 kristaps 779:
1.70 joerg 780: if (pos >= (int)ln.sz)
1.103 schwarze 781: resize_buf(&ln, 256);
1.112 kristaps 782:
1.71 kristaps 783: ln.buf[pos] = '\0';
784:
1.76 kristaps 785: /*
786: * A significant amount of complexity is contained by
787: * the roff preprocessor. It's line-oriented but can be
788: * expressed on one line, so we need at times to
789: * readjust our starting point and re-run it. The roff
790: * preprocessor can also readjust the buffers with new
791: * data, so we pass them in wholesale.
792: */
793:
794: of = 0;
1.112 kristaps 795:
1.111 kristaps 796: rerun:
1.114 kristaps 797: rr = roff_parseln
1.112 kristaps 798: (curp->roff, curp->line,
799: &ln.buf, &ln.sz, of, &of);
800:
1.114 kristaps 801: switch (rr) {
1.111 kristaps 802: case (ROFF_REPARSE):
1.122 schwarze 803: if (REPARSE_LIMIT >= ++curp->reparse_count)
804: parsebuf(curp, ln, 0);
805: else
806: mmsg(MANDOCERR_ROFFLOOP, curp,
807: curp->line, pos, NULL);
1.111 kristaps 808: pos = 0;
809: continue;
810: case (ROFF_APPEND):
811: pos = strlen(ln.buf);
812: continue;
813: case (ROFF_RERUN):
814: goto rerun;
815: case (ROFF_IGN):
816: pos = 0;
1.71 kristaps 817: continue;
1.111 kristaps 818: case (ROFF_ERR):
1.115 schwarze 819: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 820: break;
821: case (ROFF_SO):
822: if (pfile(ln.buf + of, curp)) {
823: pos = 0;
824: continue;
825: } else
826: break;
1.128 kristaps 827: default:
1.111 kristaps 828: break;
1.103 schwarze 829: }
1.132 kristaps 830:
831: /*
832: * If we encounter errors in the recursive parsebuf()
833: * call, make sure we don't continue parsing.
834: */
835:
836: if (MANDOCLEVEL_FATAL <= file_status)
837: break;
1.29 kristaps 838:
1.76 kristaps 839: /*
840: * If input parsers have not been allocated, do so now.
841: * We keep these instanced betwen parsers, but set them
842: * locally per parse routine since we can use different
843: * parsers with each one.
844: */
1.19 kristaps 845:
1.112 kristaps 846: if ( ! (curp->man || curp->mdoc))
847: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 848:
1.112 kristaps 849: /*
850: * Lastly, push down into the parsers themselves. One
851: * of these will have already been set in the pset()
852: * routine.
1.128 kristaps 853: * If libroff returns ROFF_TBL, then add it to the
854: * currently open parse. Since we only get here if
855: * there does exist data (see tbl_data.c), we're
856: * guaranteed that something's been allocated.
1.146 kristaps 857: * Do the same for ROFF_EQN.
1.112 kristaps 858: */
859:
1.146 kristaps 860: rc = -1;
861:
862: if (ROFF_TBL == rr)
1.141 schwarze 863: while (NULL != (span = roff_span(curp->roff))) {
1.146 kristaps 864: rc = curp->man ?
865: man_addspan(curp->man, span) :
1.141 schwarze 866: mdoc_addspan(curp->mdoc, span);
1.146 kristaps 867: if (0 == rc)
868: break;
1.141 schwarze 869: }
1.146 kristaps 870: else if (ROFF_EQN == rr)
1.145 kristaps 871: rc = curp->mdoc ?
872: mdoc_addeqn(curp->mdoc,
873: roff_eqn(curp->roff)) :
874: man_addeqn(curp->man,
875: roff_eqn(curp->roff));
1.146 kristaps 876: else if (curp->man || curp->mdoc)
1.112 kristaps 877: rc = curp->man ?
878: man_parseln(curp->man,
879: curp->line, ln.buf, of) :
880: mdoc_parseln(curp->mdoc,
881: curp->line, ln.buf, of);
1.30 kristaps 882:
1.146 kristaps 883: if (0 == rc) {
884: assert(MANDOCLEVEL_FATAL <= file_status);
885: break;
1.103 schwarze 886: }
1.19 kristaps 887:
1.111 kristaps 888: /* Temporary buffers typically are not full. */
1.112 kristaps 889:
1.111 kristaps 890: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 891: break;
1.85 kristaps 892:
1.111 kristaps 893: /* Start the next input line. */
1.112 kristaps 894:
1.111 kristaps 895: pos = 0;
1.22 kristaps 896: }
897:
1.111 kristaps 898: free(ln.buf);
1.19 kristaps 899: }
900:
1.103 schwarze 901: static void
1.112 kristaps 902: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 903: {
1.29 kristaps 904: int i;
1.19 kristaps 905:
906: /*
907: * Try to intuit which kind of manual parser should be used. If
908: * passed in by command-line (-man, -mdoc), then use that
909: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 910: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 911: * default to -man, which is more lenient.
1.112 kristaps 912: *
913: * Separate out pmdoc/pman from mdoc/man: the first persists
914: * through all parsers, while the latter is used per-parse.
1.19 kristaps 915: */
916:
1.75 kristaps 917: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 918: for (i = 1; buf[i]; i++)
919: if (' ' != buf[i] && '\t' != buf[i])
920: break;
1.103 schwarze 921: if ('\0' == buf[i])
922: return;
1.29 kristaps 923: }
1.10 kristaps 924:
1.19 kristaps 925: switch (curp->inttype) {
926: case (INTT_MDOC):
1.112 kristaps 927: if (NULL == curp->pmdoc)
928: curp->pmdoc = mdoc_alloc
929: (&curp->regs, curp, mmsg);
930: assert(curp->pmdoc);
931: curp->mdoc = curp->pmdoc;
1.103 schwarze 932: return;
1.19 kristaps 933: case (INTT_MAN):
1.112 kristaps 934: if (NULL == curp->pman)
935: curp->pman = man_alloc
936: (&curp->regs, curp, mmsg);
937: assert(curp->pman);
938: curp->man = curp->pman;
1.103 schwarze 939: return;
1.19 kristaps 940: default:
941: break;
942: }
943:
944: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 945: if (NULL == curp->pmdoc)
946: curp->pmdoc = mdoc_alloc
947: (&curp->regs, curp, mmsg);
948: assert(curp->pmdoc);
949: curp->mdoc = curp->pmdoc;
1.103 schwarze 950: return;
1.19 kristaps 951: }
952:
1.112 kristaps 953: if (NULL == curp->pman)
954: curp->pman = man_alloc(&curp->regs, curp, mmsg);
955: assert(curp->pman);
956: curp->man = curp->pman;
1.10 kristaps 957: }
958:
959: static int
960: moptions(enum intt *tflags, char *arg)
961: {
962:
1.17 kristaps 963: if (0 == strcmp(arg, "doc"))
1.10 kristaps 964: *tflags = INTT_MDOC;
1.19 kristaps 965: else if (0 == strcmp(arg, "andoc"))
966: *tflags = INTT_AUTO;
1.17 kristaps 967: else if (0 == strcmp(arg, "an"))
1.10 kristaps 968: *tflags = INTT_MAN;
969: else {
1.57 kristaps 970: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 971: return(0);
972: }
973:
974: return(1);
1.1 kristaps 975: }
976:
977: static int
1.60 kristaps 978: toptions(struct curparse *curp, char *arg)
1.1 kristaps 979: {
980:
981: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 982: curp->outtype = OUTT_ASCII;
983: else if (0 == strcmp(arg, "lint")) {
984: curp->outtype = OUTT_LINT;
1.103 schwarze 985: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 986: }
1.1 kristaps 987: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 988: curp->outtype = OUTT_TREE;
1.43 kristaps 989: else if (0 == strcmp(arg, "html"))
1.60 kristaps 990: curp->outtype = OUTT_HTML;
1.59 kristaps 991: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 992: curp->outtype = OUTT_XHTML;
1.85 kristaps 993: else if (0 == strcmp(arg, "ps"))
994: curp->outtype = OUTT_PS;
1.100 kristaps 995: else if (0 == strcmp(arg, "pdf"))
996: curp->outtype = OUTT_PDF;
1.1 kristaps 997: else {
1.57 kristaps 998: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 999: return(0);
1000: }
1001:
1002: return(1);
1003: }
1004:
1005: static int
1.103 schwarze 1006: woptions(struct curparse *curp, char *arg)
1.1 kristaps 1007: {
1.34 kristaps 1008: char *v, *o;
1.103 schwarze 1009: const char *toks[6];
1.1 kristaps 1010:
1.103 schwarze 1011: toks[0] = "stop";
1012: toks[1] = "all";
1013: toks[2] = "warning";
1014: toks[3] = "error";
1015: toks[4] = "fatal";
1016: toks[5] = NULL;
1.1 kristaps 1017:
1.34 kristaps 1018: while (*arg) {
1019: o = arg;
1.45 kristaps 1020: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 1021: case (0):
1.103 schwarze 1022: curp->wstop = 1;
1.1 kristaps 1023: break;
1024: case (1):
1.103 schwarze 1025: /* FALLTHROUGH */
1.1 kristaps 1026: case (2):
1.103 schwarze 1027: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 1028: break;
1.20 kristaps 1029: case (3):
1.103 schwarze 1030: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 1031: break;
1032: case (4):
1.103 schwarze 1033: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 1034: break;
1.1 kristaps 1035: default:
1.103 schwarze 1036: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 1037: return(0);
1038: }
1.34 kristaps 1039: }
1.1 kristaps 1040:
1041: return(1);
1042: }
1043:
1.71 kristaps 1044: static int
1045: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
1046: {
1047: struct curparse *cp;
1.103 schwarze 1048: enum mandoclevel level;
1049:
1050: level = MANDOCLEVEL_FATAL;
1051: while (t < mandoclimits[level])
1.105 kristaps 1052: /* LINTED */
1.103 schwarze 1053: level--;
1.71 kristaps 1054:
1055: cp = (struct curparse *)arg;
1.103 schwarze 1056: if (level < cp->wlevel)
1057: return(1);
1.73 kristaps 1058:
1.103 schwarze 1059: fprintf(stderr, "%s:%d:%d: %s: %s",
1060: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 1061: if (msg)
1062: fprintf(stderr, ": %s", msg);
1063: fputc('\n', stderr);
1.79 kristaps 1064:
1.115 schwarze 1065: if (file_status < level)
1066: file_status = level;
1.103 schwarze 1067:
1068: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 1069: }
CVSweb