Annotation of src/usr.bin/mandoc/main.c, Revision 1.74
1.74 ! schwarze 1: /* $Id: main.c,v 1.73 2011/03/07 01:35:33 schwarze Exp $ */
1.1 kristaps 2: /*
1.64 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.70 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
1.2 schwarze 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: *
1.2 schwarze 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: */
1.27 schwarze 18: #include <sys/types.h>
19: #include <sys/mman.h>
1.1 kristaps 20: #include <sys/stat.h>
21:
22: #include <assert.h>
1.43 schwarze 23: #include <ctype.h>
1.1 kristaps 24: #include <fcntl.h>
25: #include <stdio.h>
1.17 schwarze 26: #include <stdint.h>
1.1 kristaps 27: #include <stdlib.h>
28: #include <string.h>
29: #include <unistd.h>
30:
1.30 schwarze 31: #include "mandoc.h"
1.38 schwarze 32: #include "main.h"
1.1 kristaps 33: #include "mdoc.h"
34: #include "man.h"
1.30 schwarze 35: #include "roff.h"
1.17 schwarze 36:
1.61 schwarze 37: #define REPARSE_LIMIT 1000
1.17 schwarze 38: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
1.1 kristaps 39:
1.16 schwarze 40: typedef void (*out_mdoc)(void *, const struct mdoc *);
41: typedef void (*out_man)(void *, const struct man *);
1.1 kristaps 42: typedef void (*out_free)(void *);
43:
44: struct buf {
45: char *buf;
46: size_t sz;
47: };
48:
49: enum intt {
50: INTT_AUTO,
51: INTT_MDOC,
52: INTT_MAN
53: };
54:
55: enum outt {
56: OUTT_ASCII = 0,
57: OUTT_TREE,
1.17 schwarze 58: OUTT_HTML,
1.21 schwarze 59: OUTT_XHTML,
1.36 schwarze 60: OUTT_LINT,
1.43 schwarze 61: OUTT_PS,
62: OUTT_PDF
1.1 kristaps 63: };
64:
65: struct curparse {
66: const char *file; /* Current parse. */
67: int fd; /* Current parse. */
1.55 schwarze 68: int line; /* Line number in the file. */
1.45 schwarze 69: enum mandoclevel wlevel; /* Ignore messages below this. */
70: int wstop; /* Stop after a file with a warning. */
1.33 schwarze 71: enum intt inttype; /* which parser to use */
1.59 schwarze 72: struct man *pman; /* persistent man parser */
73: struct mdoc *pmdoc; /* persistent mdoc parser */
1.33 schwarze 74: struct man *man; /* man parser */
75: struct mdoc *mdoc; /* mdoc parser */
76: struct roff *roff; /* roff parser (!NULL) */
1.38 schwarze 77: struct regset regs; /* roff registers */
1.61 schwarze 78: int reparse_count; /* finite interpolation stack */
1.33 schwarze 79: enum outt outtype; /* which output to use */
80: out_mdoc outmdoc; /* mdoc output ptr */
81: out_man outman; /* man output ptr */
82: out_free outfree; /* free output ptr */
83: void *outdata; /* data for output */
84: char outopts[BUFSIZ]; /* buf of output opts */
85: };
86:
1.45 schwarze 87: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
88: "SUCCESS",
89: "RESERVED",
90: "WARNING",
91: "ERROR",
92: "FATAL",
93: "BADARG",
94: "SYSERR"
95: };
96:
97: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
98: MANDOCERR_OK,
99: MANDOCERR_WARNING,
100: MANDOCERR_WARNING,
101: MANDOCERR_ERROR,
102: MANDOCERR_FATAL,
103: MANDOCERR_MAX,
104: MANDOCERR_MAX
105: };
106:
1.33 schwarze 107: static const char * const mandocerrs[MANDOCERR_MAX] = {
108: "ok",
1.40 schwarze 109:
110: "generic warning",
111:
1.53 schwarze 112: /* related to the prologue */
113: "no title in document",
114: "document title should be all caps",
115: "unknown manual section",
1.73 schwarze 116: "date missing, using today's date",
117: "cannot parse date, using it verbatim",
1.53 schwarze 118: "prologue macros out of order",
119: "duplicate prologue macro",
120: "macro not allowed in prologue",
121: "macro not allowed in body",
122:
123: /* related to document structure */
1.54 schwarze 124: ".so is fragile, better use ln(1)",
1.53 schwarze 125: "NAME section must come first",
126: "bad NAME section contents",
127: "manual name not yet set",
1.34 schwarze 128: "sections out of conventional order",
1.53 schwarze 129: "duplicate section name",
130: "section not in conventional manual section",
131:
132: /* related to macros and nesting */
133: "skipping obsolete macro",
134: "skipping paragraph macro",
1.72 schwarze 135: "skipping no-space macro",
1.53 schwarze 136: "blocks badly nested",
137: "child violates parent syntax",
138: "nested displays are not portable",
139: "already in literal mode",
140:
141: /* related to missing macro arguments */
142: "skipping empty macro",
1.63 schwarze 143: "argument count wrong",
1.53 schwarze 144: "missing display type",
1.33 schwarze 145: "list type must come first",
1.53 schwarze 146: "tag lists require a width argument",
147: "missing font type",
1.68 schwarze 148: "skipping end of block that is not open",
1.53 schwarze 149:
150: /* related to bad macro arguments */
151: "skipping argument",
152: "duplicate argument",
153: "duplicate display type",
154: "duplicate list type",
155: "unknown AT&T UNIX version",
156: "bad Boolean value",
1.57 schwarze 157: "unknown font",
1.53 schwarze 158: "unknown standard specifier",
159: "bad width argument",
160:
161: /* related to plain text */
162: "blank line in non-literal context",
1.43 schwarze 163: "tab in non-literal context",
1.53 schwarze 164: "end of line whitespace",
165: "bad comment style",
166: "unknown escape sequence",
1.33 schwarze 167: "unterminated quoted string",
1.64 schwarze 168:
1.40 schwarze 169: "generic error",
170:
1.64 schwarze 171: /* related to tables */
172: "bad table syntax",
173: "bad table option",
174: "bad table layout",
175: "no table layout cells specified",
176: "no table data cells specified",
177: "ignore data in cell",
178: "data block still open",
1.67 schwarze 179: "ignoring extra data cells",
1.64 schwarze 180:
1.61 schwarze 181: "input stack limit exceeded, infinite loop?",
1.53 schwarze 182: "skipping bad character",
1.66 schwarze 183: "escaped character not allowed in a name",
1.53 schwarze 184: "skipping text before the first section header",
185: "skipping unknown macro",
1.69 schwarze 186: "NOT IMPLEMENTED, please use groff: skipping request",
1.33 schwarze 187: "line scope broken",
188: "argument count wrong",
1.53 schwarze 189: "skipping end of block that is not open",
1.50 schwarze 190: "missing end of block",
1.47 schwarze 191: "scope open on exit",
1.50 schwarze 192: "uname(3) system call failed",
1.33 schwarze 193: "macro requires line argument(s)",
194: "macro requires body argument(s)",
195: "macro requires argument(s)",
1.34 schwarze 196: "missing list type",
1.33 schwarze 197: "line argument(s) will be lost",
198: "body argument(s) will be lost",
1.40 schwarze 199:
200: "generic fatal error",
201:
1.34 schwarze 202: "column syntax is inconsistent",
1.56 schwarze 203: "NOT IMPLEMENTED: .Bd -file",
1.33 schwarze 204: "line scope broken, syntax violated",
205: "argument count wrong, violates syntax",
206: "child violates parent syntax",
207: "argument count wrong, violates syntax",
1.56 schwarze 208: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.33 schwarze 209: "no document body",
210: "no document prologue",
1.45 schwarze 211: "static buffer exhausted",
1.1 kristaps 212: };
213:
1.55 schwarze 214: static void parsebuf(struct curparse *, struct buf, int);
1.51 schwarze 215: static void pdesc(struct curparse *);
1.27 schwarze 216: static void fdesc(struct curparse *);
217: static void ffile(const char *, struct curparse *);
1.55 schwarze 218: static int pfile(const char *, struct curparse *);
1.1 kristaps 219: static int moptions(enum intt *, char *);
1.30 schwarze 220: static int mmsg(enum mandocerr, void *,
221: int, int, const char *);
1.59 schwarze 222: static void pset(const char *, int, struct curparse *);
1.27 schwarze 223: static int toptions(struct curparse *, char *);
224: static void usage(void) __attribute__((noreturn));
1.20 schwarze 225: static void version(void) __attribute__((noreturn));
1.45 schwarze 226: static int woptions(struct curparse *, char *);
1.1 kristaps 227:
1.19 schwarze 228: static const char *progname;
1.60 schwarze 229: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.45 schwarze 230: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 231:
232: int
233: main(int argc, char *argv[])
234: {
1.27 schwarze 235: int c;
1.1 kristaps 236: struct curparse curp;
237:
1.19 schwarze 238: progname = strrchr(argv[0], '/');
239: if (progname == NULL)
240: progname = argv[0];
241: else
242: ++progname;
243:
244: memset(&curp, 0, sizeof(struct curparse));
1.1 kristaps 245:
246: curp.inttype = INTT_AUTO;
247: curp.outtype = OUTT_ASCII;
1.45 schwarze 248: curp.wlevel = MANDOCLEVEL_FATAL;
1.1 kristaps 249:
250: /* LINTED */
1.45 schwarze 251: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 252: switch (c) {
253: case ('m'):
254: if ( ! moptions(&curp.inttype, optarg))
1.47 schwarze 255: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 256: break;
1.18 schwarze 257: case ('O'):
258: (void)strlcat(curp.outopts, optarg, BUFSIZ);
259: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.17 schwarze 260: break;
1.1 kristaps 261: case ('T'):
1.22 schwarze 262: if ( ! toptions(&curp, optarg))
1.47 schwarze 263: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 264: break;
265: case ('W'):
1.45 schwarze 266: if ( ! woptions(&curp, optarg))
1.47 schwarze 267: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 268: break;
1.3 schwarze 269: case ('V'):
270: version();
271: /* NOTREACHED */
1.1 kristaps 272: default:
273: usage();
274: /* NOTREACHED */
275: }
276:
277: argc -= optind;
278: argv += optind;
279:
1.7 schwarze 280: if (NULL == *argv) {
281: curp.file = "<stdin>";
282: curp.fd = STDIN_FILENO;
1.14 schwarze 283:
1.27 schwarze 284: fdesc(&curp);
285: }
286:
287: while (*argv) {
288: ffile(*argv, &curp);
1.45 schwarze 289: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.27 schwarze 290: break;
291: ++argv;
1.1 kristaps 292: }
293:
294: if (curp.outfree)
295: (*curp.outfree)(curp.outdata);
1.59 schwarze 296: if (curp.pmdoc)
297: mdoc_free(curp.pmdoc);
298: if (curp.pman)
299: man_free(curp.pman);
1.30 schwarze 300: if (curp.roff)
301: roff_free(curp.roff);
1.1 kristaps 302:
1.47 schwarze 303: return((int)exit_status);
1.1 kristaps 304: }
305:
306:
1.20 schwarze 307: static void
1.3 schwarze 308: version(void)
309: {
310:
1.19 schwarze 311: (void)printf("%s %s\n", progname, VERSION);
1.47 schwarze 312: exit((int)MANDOCLEVEL_OK);
1.3 schwarze 313: }
314:
315:
1.20 schwarze 316: static void
1.1 kristaps 317: usage(void)
318: {
319:
1.59 schwarze 320: (void)fprintf(stderr, "usage: %s "
321: "[-V] "
322: "[-foption] "
323: "[-mformat] "
324: "[-Ooption] "
325: "[-Toutput] "
326: "[-Werr] "
327: "[file...]\n",
328: progname);
329:
1.47 schwarze 330: exit((int)MANDOCLEVEL_BADARG);
1.1 kristaps 331: }
332:
1.27 schwarze 333: static void
334: ffile(const char *file, struct curparse *curp)
1.1 kristaps 335: {
336:
1.59 schwarze 337: /*
338: * Called once per input file. Get the file ready for reading,
339: * pass it through to the parser-driver, then close it out.
340: * XXX: don't do anything special as this is only called for
341: * files; stdin goes directly to fdesc().
342: */
343:
1.1 kristaps 344: curp->file = file;
1.59 schwarze 345:
1.1 kristaps 346: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.19 schwarze 347: perror(curp->file);
1.45 schwarze 348: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 349: return;
1.1 kristaps 350: }
351:
1.27 schwarze 352: fdesc(curp);
1.1 kristaps 353:
354: if (-1 == close(curp->fd))
1.19 schwarze 355: perror(curp->file);
1.27 schwarze 356: }
1.1 kristaps 357:
1.52 schwarze 358: static int
1.55 schwarze 359: pfile(const char *file, struct curparse *curp)
1.52 schwarze 360: {
361: const char *savefile;
362: int fd, savefd;
363:
364: if (-1 == (fd = open(file, O_RDONLY, 0))) {
365: perror(file);
1.60 schwarze 366: file_status = MANDOCLEVEL_SYSERR;
1.52 schwarze 367: return(0);
368: }
369:
370: savefile = curp->file;
371: savefd = curp->fd;
372:
373: curp->file = file;
374: curp->fd = fd;
375:
376: pdesc(curp);
377:
378: curp->file = savefile;
379: curp->fd = savefd;
380:
381: if (-1 == close(fd))
382: perror(file);
383:
1.60 schwarze 384: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.52 schwarze 385: }
386:
1.27 schwarze 387:
1.45 schwarze 388: static void
1.27 schwarze 389: resize_buf(struct buf *buf, size_t initial)
390: {
391:
1.62 schwarze 392: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
1.45 schwarze 393: buf->buf = realloc(buf->buf, buf->sz);
394: if (NULL == buf->buf) {
1.27 schwarze 395: perror(NULL);
1.47 schwarze 396: exit((int)MANDOCLEVEL_SYSERR);
1.27 schwarze 397: }
1.1 kristaps 398: }
399:
400:
401: static int
1.27 schwarze 402: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
1.1 kristaps 403: {
1.27 schwarze 404: struct stat st;
405: size_t off;
1.1 kristaps 406: ssize_t ssz;
1.27 schwarze 407:
408: if (-1 == fstat(curp->fd, &st)) {
409: perror(curp->file);
410: return(0);
411: }
412:
413: /*
414: * If we're a regular file, try just reading in the whole entry
415: * via mmap(). This is faster than reading it into blocks, and
416: * since each file is only a few bytes to begin with, I'm not
417: * concerned that this is going to tank any machines.
418: */
419:
420: if (S_ISREG(st.st_mode)) {
421: if (st.st_size >= (1U << 31)) {
422: fprintf(stderr, "%s: input too large\n",
423: curp->file);
424: return(0);
425: }
426: *with_mmap = 1;
1.30 schwarze 427: fb->sz = (size_t)st.st_size;
1.27 schwarze 428: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.35 schwarze 429: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.27 schwarze 430: if (fb->buf != MAP_FAILED)
431: return(1);
432: }
433:
434: /*
435: * If this isn't a regular file (like, say, stdin), then we must
436: * go the old way and just read things in bit by bit.
437: */
438:
439: *with_mmap = 0;
440: off = 0;
441: fb->sz = 0;
442: fb->buf = NULL;
443: for (;;) {
444: if (off == fb->sz) {
445: if (fb->sz == (1U << 31)) {
446: fprintf(stderr, "%s: input too large\n",
447: curp->file);
448: break;
449: }
1.45 schwarze 450: resize_buf(fb, 65536);
1.27 schwarze 451: }
1.30 schwarze 452: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.27 schwarze 453: if (ssz == 0) {
454: fb->sz = off;
455: return(1);
456: }
457: if (ssz == -1) {
458: perror(curp->file);
459: break;
460: }
1.30 schwarze 461: off += (size_t)ssz;
1.27 schwarze 462: }
463:
464: free(fb->buf);
465: fb->buf = NULL;
466: return(0);
467: }
468:
469:
470: static void
471: fdesc(struct curparse *curp)
472: {
1.59 schwarze 473:
474: /*
475: * Called once per file with an opened file descriptor. All
476: * pre-file-parse operations (whether stdin or a file) should go
477: * here.
478: *
479: * This calls down into the nested parser, which drills down and
480: * fully parses a file and all its dependences (i.e., `so'). It
481: * then runs the cleanup validators and pushes to output.
482: */
483:
484: /* Zero the parse type. */
485:
486: curp->mdoc = NULL;
487: curp->man = NULL;
1.60 schwarze 488: file_status = MANDOCLEVEL_OK;
1.59 schwarze 489:
490: /* Make sure the mandotory roff parser is initialised. */
491:
492: if (NULL == curp->roff) {
493: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
494: assert(curp->roff);
495: }
496:
497: /* Fully parse the file. */
1.51 schwarze 498:
499: pdesc(curp);
500:
1.60 schwarze 501: if (MANDOCLEVEL_FATAL <= file_status)
1.51 schwarze 502: goto cleanup;
503:
504: /* NOTE a parser may not have been assigned, yet. */
505:
1.59 schwarze 506: if ( ! (curp->man || curp->mdoc)) {
1.51 schwarze 507: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.60 schwarze 508: file_status = MANDOCLEVEL_FATAL;
1.51 schwarze 509: goto cleanup;
510: }
511:
512: /* Clean up the parse routine ASTs. */
513:
1.59 schwarze 514: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.60 schwarze 515: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 516: goto cleanup;
517: }
1.59 schwarze 518:
519: if (curp->man && ! man_endparse(curp->man)) {
1.60 schwarze 520: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 521: goto cleanup;
522: }
1.59 schwarze 523:
524: assert(curp->roff);
1.64 schwarze 525: roff_endparse(curp->roff);
1.51 schwarze 526:
527: /*
528: * With -Wstop and warnings or errors of at least
529: * the requested level, do not produce output.
530: */
531:
1.60 schwarze 532: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.51 schwarze 533: goto cleanup;
534:
535: /* If unset, allocate output dev now (if applicable). */
536:
537: if ( ! (curp->outman && curp->outmdoc)) {
538: switch (curp->outtype) {
539: case (OUTT_XHTML):
540: curp->outdata = xhtml_alloc(curp->outopts);
541: break;
542: case (OUTT_HTML):
543: curp->outdata = html_alloc(curp->outopts);
544: break;
545: case (OUTT_ASCII):
546: curp->outdata = ascii_alloc(curp->outopts);
547: curp->outfree = ascii_free;
548: break;
549: case (OUTT_PDF):
550: curp->outdata = pdf_alloc(curp->outopts);
551: curp->outfree = pspdf_free;
552: break;
553: case (OUTT_PS):
554: curp->outdata = ps_alloc(curp->outopts);
555: curp->outfree = pspdf_free;
556: break;
557: default:
558: break;
559: }
560:
561: switch (curp->outtype) {
562: case (OUTT_HTML):
563: /* FALLTHROUGH */
564: case (OUTT_XHTML):
565: curp->outman = html_man;
566: curp->outmdoc = html_mdoc;
567: curp->outfree = html_free;
568: break;
569: case (OUTT_TREE):
570: curp->outman = tree_man;
571: curp->outmdoc = tree_mdoc;
572: break;
573: case (OUTT_PDF):
574: /* FALLTHROUGH */
575: case (OUTT_ASCII):
576: /* FALLTHROUGH */
577: case (OUTT_PS):
578: curp->outman = terminal_man;
579: curp->outmdoc = terminal_mdoc;
580: break;
581: default:
582: break;
583: }
584: }
585:
586: /* Execute the out device, if it exists. */
587:
1.59 schwarze 588: if (curp->man && curp->outman)
589: (*curp->outman)(curp->outdata, curp->man);
590: if (curp->mdoc && curp->outmdoc)
591: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.51 schwarze 592:
593: cleanup:
1.59 schwarze 594:
1.51 schwarze 595: memset(&curp->regs, 0, sizeof(struct regset));
1.59 schwarze 596:
597: /* Reset the current-parse compilers. */
598:
599: if (curp->mdoc)
600: mdoc_reset(curp->mdoc);
601: if (curp->man)
602: man_reset(curp->man);
603:
604: assert(curp->roff);
605: roff_reset(curp->roff);
1.51 schwarze 606:
1.60 schwarze 607: if (exit_status < file_status)
608: exit_status = file_status;
609:
1.51 schwarze 610: return;
611: }
612:
613: static void
614: pdesc(struct curparse *curp)
615: {
1.55 schwarze 616: struct buf blk;
617: int with_mmap;
1.1 kristaps 618:
1.59 schwarze 619: /*
620: * Run for each opened file; may be called more than once for
621: * each full parse sequence if the opened file is nested (i.e.,
622: * from `so'). Simply sucks in the whole file and moves into
623: * the parse phase for the file.
624: */
625:
1.45 schwarze 626: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.60 schwarze 627: file_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 628: return;
1.45 schwarze 629: }
1.27 schwarze 630:
1.59 schwarze 631: /* Line number is per-file. */
1.55 schwarze 632:
633: curp->line = 1;
1.59 schwarze 634:
1.55 schwarze 635: parsebuf(curp, blk, 1);
636:
637: if (with_mmap)
638: munmap(blk.buf, blk.sz);
639: else
640: free(blk.buf);
641: }
642:
643: static void
644: parsebuf(struct curparse *curp, struct buf blk, int start)
645: {
1.71 schwarze 646: const struct tbl_span *span;
1.55 schwarze 647: struct buf ln;
1.59 schwarze 648: enum rofferr rr;
649: int i, of, rc;
650: int pos; /* byte number in the ln buffer */
651: int lnn; /* line number in the real file */
1.55 schwarze 652: unsigned char c;
1.59 schwarze 653:
654: /*
655: * Main parse routine for an opened file. This is called for
656: * each opened file and simply loops around the full input file,
657: * possibly nesting (i.e., with `so').
658: */
1.30 schwarze 659:
1.55 schwarze 660: memset(&ln, 0, sizeof(struct buf));
661:
1.59 schwarze 662: lnn = curp->line;
663: pos = 0;
1.55 schwarze 664:
1.59 schwarze 665: for (i = 0; i < (int)blk.sz; ) {
1.55 schwarze 666: if (0 == pos && '\0' == blk.buf[i])
667: break;
1.59 schwarze 668:
1.61 schwarze 669: if (start) {
1.55 schwarze 670: curp->line = lnn;
1.61 schwarze 671: curp->reparse_count = 0;
672: }
1.55 schwarze 673:
674: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 schwarze 675:
676: /*
677: * When finding an unescaped newline character,
678: * leave the character loop to process the line.
679: * Skip a preceding carriage return, if any.
680: */
681:
682: if ('\r' == blk.buf[i] && i + 1 < (int)blk.sz &&
683: '\n' == blk.buf[i + 1])
684: ++i;
1.29 schwarze 685: if ('\n' == blk.buf[i]) {
686: ++i;
687: ++lnn;
688: break;
689: }
1.43 schwarze 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:
1.44 schwarze 700: c = (unsigned char) blk.buf[i];
1.59 schwarze 701:
702: if ( ! (isascii(c) &&
703: (isgraph(c) || isblank(c)))) {
1.45 schwarze 704: mmsg(MANDOCERR_BADCHAR, curp,
1.55 schwarze 705: curp->line, pos, "ignoring byte");
1.43 schwarze 706: i++;
707: continue;
708: }
709:
1.59 schwarze 710: /* Trailing backslash = a plain char. */
711:
1.29 schwarze 712: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
713: if (pos >= (int)ln.sz)
1.45 schwarze 714: resize_buf(&ln, 256);
1.29 schwarze 715: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 716: continue;
1.29 schwarze 717: }
1.59 schwarze 718:
1.70 schwarze 719: /*
720: * Found escape and at least one other character.
721: * When it's a newline character, skip it.
722: * When there is a carriage return in between,
723: * skip that one as well.
724: */
1.59 schwarze 725:
1.70 schwarze 726: if ('\r' == blk.buf[i + 1] && i + 2 < (int)blk.sz &&
727: '\n' == blk.buf[i + 2])
728: ++i;
1.29 schwarze 729: if ('\n' == blk.buf[i + 1]) {
1.59 schwarze 730: i += 2;
1.29 schwarze 731: ++lnn;
1.27 schwarze 732: continue;
733: }
1.59 schwarze 734:
1.29 schwarze 735: if ('"' == blk.buf[i + 1]) {
736: i += 2;
737: /* Comment, skip to end of line */
738: for (; i < (int)blk.sz; ++i) {
739: if ('\n' == blk.buf[i]) {
740: ++i;
741: ++lnn;
742: break;
743: }
744: }
1.59 schwarze 745:
1.29 schwarze 746: /* Backout trailing whitespaces */
747: for (; pos > 0; --pos) {
748: if (ln.buf[pos - 1] != ' ')
749: break;
750: if (pos > 2 && ln.buf[pos - 2] == '\\')
751: break;
752: }
753: break;
754: }
1.59 schwarze 755:
756: /* Some other escape sequence, copy & cont. */
757:
1.29 schwarze 758: if (pos + 1 >= (int)ln.sz)
1.45 schwarze 759: resize_buf(&ln, 256);
1.1 kristaps 760:
1.29 schwarze 761: ln.buf[pos++] = blk.buf[i++];
762: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 763: }
1.1 kristaps 764:
1.29 schwarze 765: if (pos >= (int)ln.sz)
1.45 schwarze 766: resize_buf(&ln, 256);
1.59 schwarze 767:
1.30 schwarze 768: ln.buf[pos] = '\0';
769:
1.32 schwarze 770: /*
771: * A significant amount of complexity is contained by
772: * the roff preprocessor. It's line-oriented but can be
773: * expressed on one line, so we need at times to
774: * readjust our starting point and re-run it. The roff
775: * preprocessor can also readjust the buffers with new
776: * data, so we pass them in wholesale.
777: */
778:
779: of = 0;
1.59 schwarze 780:
1.55 schwarze 781: rerun:
1.59 schwarze 782: rr = roff_parseln
783: (curp->roff, curp->line,
784: &ln.buf, &ln.sz, of, &of);
785:
786: switch (rr) {
1.55 schwarze 787: case (ROFF_REPARSE):
1.61 schwarze 788: if (REPARSE_LIMIT >= ++curp->reparse_count)
789: parsebuf(curp, ln, 0);
790: else
791: mmsg(MANDOCERR_ROFFLOOP, curp,
792: curp->line, pos, NULL);
1.55 schwarze 793: pos = 0;
794: continue;
795: case (ROFF_APPEND):
796: pos = strlen(ln.buf);
797: continue;
798: case (ROFF_RERUN):
799: goto rerun;
800: case (ROFF_IGN):
801: pos = 0;
1.30 schwarze 802: continue;
1.55 schwarze 803: case (ROFF_ERR):
1.60 schwarze 804: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 805: break;
1.55 schwarze 806: case (ROFF_SO):
807: if (pfile(ln.buf + of, curp)) {
808: pos = 0;
1.52 schwarze 809: continue;
1.55 schwarze 810: } else
1.52 schwarze 811: break;
1.64 schwarze 812: default:
1.55 schwarze 813: break;
1.45 schwarze 814: }
1.65 schwarze 815:
816: /*
817: * If we encounter errors in the recursive parsebuf()
818: * call, make sure we don't continue parsing.
819: */
820:
821: if (MANDOCLEVEL_FATAL <= file_status)
822: break;
1.5 schwarze 823:
1.32 schwarze 824: /*
825: * If input parsers have not been allocated, do so now.
826: * We keep these instanced betwen parsers, but set them
827: * locally per parse routine since we can use different
828: * parsers with each one.
829: */
1.1 kristaps 830:
1.59 schwarze 831: if ( ! (curp->man || curp->mdoc))
832: pset(ln.buf + of, pos - of, curp);
1.5 schwarze 833:
1.59 schwarze 834: /*
835: * Lastly, push down into the parsers themselves. One
836: * of these will have already been set in the pset()
837: * routine.
1.64 schwarze 838: * If libroff returns ROFF_TBL, then add it to the
839: * currently open parse. Since we only get here if
840: * there does exist data (see tbl_data.c), we're
841: * guaranteed that something's been allocated.
1.74 ! schwarze 842: * Do the same for ROFF_EQN.
1.59 schwarze 843: */
844:
1.74 ! schwarze 845: rc = -1;
! 846:
! 847: if (ROFF_TBL == rr)
1.71 schwarze 848: while (NULL != (span = roff_span(curp->roff))) {
1.74 ! schwarze 849: rc = curp->man ?
! 850: man_addspan(curp->man, span) :
1.71 schwarze 851: mdoc_addspan(curp->mdoc, span);
1.74 ! schwarze 852: if (0 == rc)
! 853: break;
1.71 schwarze 854: }
1.74 ! schwarze 855: else if (ROFF_EQN == rr)
! 856: rc = curp->mdoc ?
! 857: mdoc_addeqn(curp->mdoc,
! 858: roff_eqn(curp->roff)) :
! 859: man_addeqn(curp->man,
! 860: roff_eqn(curp->roff));
! 861: else if (curp->man || curp->mdoc)
1.59 schwarze 862: rc = curp->man ?
863: man_parseln(curp->man,
864: curp->line, ln.buf, of) :
865: mdoc_parseln(curp->mdoc,
866: curp->line, ln.buf, of);
1.1 kristaps 867:
1.74 ! schwarze 868: if (0 == rc) {
! 869: assert(MANDOCLEVEL_FATAL <= file_status);
! 870: break;
1.1 kristaps 871: }
1.55 schwarze 872:
873: /* Temporary buffers typically are not full. */
1.59 schwarze 874:
1.55 schwarze 875: if (0 == start && '\0' == blk.buf[i])
876: break;
877:
878: /* Start the next input line. */
1.59 schwarze 879:
1.55 schwarze 880: pos = 0;
1.1 kristaps 881: }
882:
1.51 schwarze 883: free(ln.buf);
1.1 kristaps 884: }
885:
1.45 schwarze 886: static void
1.59 schwarze 887: pset(const char *buf, int pos, struct curparse *curp)
1.1 kristaps 888: {
1.5 schwarze 889: int i;
1.1 kristaps 890:
891: /*
892: * Try to intuit which kind of manual parser should be used. If
893: * passed in by command-line (-man, -mdoc), then use that
894: * explicitly. If passed as -mandoc, then try to guess from the
1.5 schwarze 895: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.1 kristaps 896: * default to -man, which is more lenient.
1.59 schwarze 897: *
898: * Separate out pmdoc/pman from mdoc/man: the first persists
899: * through all parsers, while the latter is used per-parse.
1.1 kristaps 900: */
901:
1.31 schwarze 902: if ('.' == buf[0] || '\'' == buf[0]) {
1.5 schwarze 903: for (i = 1; buf[i]; i++)
904: if (' ' != buf[i] && '\t' != buf[i])
905: break;
1.45 schwarze 906: if ('\0' == buf[i])
907: return;
1.5 schwarze 908: }
1.1 kristaps 909:
910: switch (curp->inttype) {
911: case (INTT_MDOC):
1.59 schwarze 912: if (NULL == curp->pmdoc)
913: curp->pmdoc = mdoc_alloc
914: (&curp->regs, curp, mmsg);
915: assert(curp->pmdoc);
916: curp->mdoc = curp->pmdoc;
1.45 schwarze 917: return;
1.1 kristaps 918: case (INTT_MAN):
1.59 schwarze 919: if (NULL == curp->pman)
920: curp->pman = man_alloc
921: (&curp->regs, curp, mmsg);
922: assert(curp->pman);
923: curp->man = curp->pman;
1.45 schwarze 924: return;
1.1 kristaps 925: default:
926: break;
927: }
928:
929: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.59 schwarze 930: if (NULL == curp->pmdoc)
931: curp->pmdoc = mdoc_alloc
932: (&curp->regs, curp, mmsg);
933: assert(curp->pmdoc);
934: curp->mdoc = curp->pmdoc;
1.45 schwarze 935: return;
1.1 kristaps 936: }
937:
1.59 schwarze 938: if (NULL == curp->pman)
939: curp->pman = man_alloc(&curp->regs, curp, mmsg);
940: assert(curp->pman);
941: curp->man = curp->pman;
1.1 kristaps 942: }
943:
944: static int
945: moptions(enum intt *tflags, char *arg)
946: {
947:
948: if (0 == strcmp(arg, "doc"))
949: *tflags = INTT_MDOC;
950: else if (0 == strcmp(arg, "andoc"))
951: *tflags = INTT_AUTO;
952: else if (0 == strcmp(arg, "an"))
953: *tflags = INTT_MAN;
954: else {
1.20 schwarze 955: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 956: return(0);
957: }
958:
959: return(1);
960: }
961:
962: static int
1.22 schwarze 963: toptions(struct curparse *curp, char *arg)
1.1 kristaps 964: {
965:
966: if (0 == strcmp(arg, "ascii"))
1.22 schwarze 967: curp->outtype = OUTT_ASCII;
968: else if (0 == strcmp(arg, "lint")) {
969: curp->outtype = OUTT_LINT;
1.45 schwarze 970: curp->wlevel = MANDOCLEVEL_WARNING;
1.22 schwarze 971: }
1.1 kristaps 972: else if (0 == strcmp(arg, "tree"))
1.22 schwarze 973: curp->outtype = OUTT_TREE;
1.17 schwarze 974: else if (0 == strcmp(arg, "html"))
1.22 schwarze 975: curp->outtype = OUTT_HTML;
1.21 schwarze 976: else if (0 == strcmp(arg, "xhtml"))
1.22 schwarze 977: curp->outtype = OUTT_XHTML;
1.36 schwarze 978: else if (0 == strcmp(arg, "ps"))
979: curp->outtype = OUTT_PS;
1.43 schwarze 980: else if (0 == strcmp(arg, "pdf"))
981: curp->outtype = OUTT_PDF;
1.1 kristaps 982: else {
1.20 schwarze 983: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 984: return(0);
985: }
986:
987: return(1);
988: }
989:
990: static int
1.45 schwarze 991: woptions(struct curparse *curp, char *arg)
1.1 kristaps 992: {
1.10 schwarze 993: char *v, *o;
1.45 schwarze 994: const char *toks[6];
1.1 kristaps 995:
1.45 schwarze 996: toks[0] = "stop";
997: toks[1] = "all";
998: toks[2] = "warning";
999: toks[3] = "error";
1000: toks[4] = "fatal";
1001: toks[5] = NULL;
1.1 kristaps 1002:
1.10 schwarze 1003: while (*arg) {
1004: o = arg;
1.17 schwarze 1005: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 1006: case (0):
1.45 schwarze 1007: curp->wstop = 1;
1.1 kristaps 1008: break;
1009: case (1):
1.45 schwarze 1010: /* FALLTHROUGH */
1.1 kristaps 1011: case (2):
1.45 schwarze 1012: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 1013: break;
1014: case (3):
1.45 schwarze 1015: curp->wlevel = MANDOCLEVEL_ERROR;
1.1 kristaps 1016: break;
1017: case (4):
1.45 schwarze 1018: curp->wlevel = MANDOCLEVEL_FATAL;
1.19 schwarze 1019: break;
1.1 kristaps 1020: default:
1.45 schwarze 1021: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 1022: return(0);
1023: }
1.10 schwarze 1024: }
1.1 kristaps 1025:
1026: return(1);
1027: }
1028:
1.30 schwarze 1029: static int
1030: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
1031: {
1032: struct curparse *cp;
1.45 schwarze 1033: enum mandoclevel level;
1034:
1035: level = MANDOCLEVEL_FATAL;
1036: while (t < mandoclimits[level])
1.47 schwarze 1037: /* LINTED */
1.45 schwarze 1038: level--;
1.30 schwarze 1039:
1040: cp = (struct curparse *)arg;
1.45 schwarze 1041: if (level < cp->wlevel)
1042: return(1);
1.30 schwarze 1043:
1.45 schwarze 1044: fprintf(stderr, "%s:%d:%d: %s: %s",
1045: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.30 schwarze 1046: if (msg)
1047: fprintf(stderr, ": %s", msg);
1048: fputc('\n', stderr);
1.33 schwarze 1049:
1.60 schwarze 1050: if (file_status < level)
1051: file_status = level;
1.45 schwarze 1052:
1053: return(level < MANDOCLEVEL_FATAL);
1.30 schwarze 1054: }