Annotation of src/usr.bin/mandoc/main.c, Revision 1.60
1.60 ! schwarze 1: /* $Id: main.c,v 1.59 2010/12/01 23:38:11 schwarze Exp $ */
1.1 kristaps 2: /*
1.42 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
1.2 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:
37: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
1.1 kristaps 38:
1.16 schwarze 39: typedef void (*out_mdoc)(void *, const struct mdoc *);
40: typedef void (*out_man)(void *, const struct man *);
1.1 kristaps 41: typedef void (*out_free)(void *);
42:
43: struct buf {
44: char *buf;
45: size_t sz;
46: };
47:
48: enum intt {
49: INTT_AUTO,
50: INTT_MDOC,
51: INTT_MAN
52: };
53:
54: enum outt {
55: OUTT_ASCII = 0,
56: OUTT_TREE,
1.17 schwarze 57: OUTT_HTML,
1.21 schwarze 58: OUTT_XHTML,
1.36 schwarze 59: OUTT_LINT,
1.43 schwarze 60: OUTT_PS,
61: OUTT_PDF
1.1 kristaps 62: };
63:
64: struct curparse {
65: const char *file; /* Current parse. */
66: int fd; /* Current parse. */
1.55 schwarze 67: int line; /* Line number in the file. */
1.45 schwarze 68: enum mandoclevel wlevel; /* Ignore messages below this. */
69: int wstop; /* Stop after a file with a warning. */
1.33 schwarze 70: enum intt inttype; /* which parser to use */
1.59 schwarze 71: struct man *pman; /* persistent man parser */
72: struct mdoc *pmdoc; /* persistent mdoc parser */
1.33 schwarze 73: struct man *man; /* man parser */
74: struct mdoc *mdoc; /* mdoc parser */
75: struct roff *roff; /* roff parser (!NULL) */
1.38 schwarze 76: struct regset regs; /* roff registers */
1.33 schwarze 77: enum outt outtype; /* which output to use */
78: out_mdoc outmdoc; /* mdoc output ptr */
79: out_man outman; /* man output ptr */
80: out_free outfree; /* free output ptr */
81: void *outdata; /* data for output */
82: char outopts[BUFSIZ]; /* buf of output opts */
83: };
84:
1.45 schwarze 85: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
86: "SUCCESS",
87: "RESERVED",
88: "WARNING",
89: "ERROR",
90: "FATAL",
91: "BADARG",
92: "SYSERR"
93: };
94:
95: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
96: MANDOCERR_OK,
97: MANDOCERR_WARNING,
98: MANDOCERR_WARNING,
99: MANDOCERR_ERROR,
100: MANDOCERR_FATAL,
101: MANDOCERR_MAX,
102: MANDOCERR_MAX
103: };
104:
1.33 schwarze 105: static const char * const mandocerrs[MANDOCERR_MAX] = {
106: "ok",
1.40 schwarze 107:
108: "generic warning",
109:
1.53 schwarze 110: /* related to the prologue */
111: "no title in document",
112: "document title should be all caps",
113: "unknown manual section",
114: "cannot parse date argument",
115: "prologue macros out of order",
116: "duplicate prologue macro",
117: "macro not allowed in prologue",
118: "macro not allowed in body",
119:
120: /* related to document structure */
1.54 schwarze 121: ".so is fragile, better use ln(1)",
1.53 schwarze 122: "NAME section must come first",
123: "bad NAME section contents",
124: "manual name not yet set",
1.34 schwarze 125: "sections out of conventional order",
1.53 schwarze 126: "duplicate section name",
127: "section not in conventional manual section",
128:
129: /* related to macros and nesting */
130: "skipping obsolete macro",
131: "skipping paragraph macro",
132: "blocks badly nested",
133: "child violates parent syntax",
134: "nested displays are not portable",
135: "already in literal mode",
136:
137: /* related to missing macro arguments */
138: "skipping empty macro",
139: "missing display type",
1.33 schwarze 140: "list type must come first",
1.53 schwarze 141: "tag lists require a width argument",
142: "missing font type",
143:
144: /* related to bad macro arguments */
145: "skipping argument",
146: "duplicate argument",
147: "duplicate display type",
148: "duplicate list type",
149: "unknown AT&T UNIX version",
150: "bad Boolean value",
1.57 schwarze 151: "unknown font",
1.53 schwarze 152: "unknown standard specifier",
153: "bad width argument",
154:
155: /* related to plain text */
156: "blank line in non-literal context",
1.43 schwarze 157: "tab in non-literal context",
1.53 schwarze 158: "end of line whitespace",
159: "bad comment style",
160: "unknown escape sequence",
1.33 schwarze 161: "unterminated quoted string",
1.40 schwarze 162:
163: "generic error",
164:
1.53 schwarze 165: "skipping bad character",
166: "skipping text before the first section header",
167: "skipping unknown macro",
1.56 schwarze 168: "NOT IMPLEMENTED: skipping request",
1.33 schwarze 169: "line scope broken",
170: "argument count wrong",
1.53 schwarze 171: "skipping end of block that is not open",
1.50 schwarze 172: "missing end of block",
1.47 schwarze 173: "scope open on exit",
1.50 schwarze 174: "uname(3) system call failed",
1.33 schwarze 175: "macro requires line argument(s)",
176: "macro requires body argument(s)",
177: "macro requires argument(s)",
1.34 schwarze 178: "missing list type",
1.33 schwarze 179: "line argument(s) will be lost",
180: "body argument(s) will be lost",
1.49 schwarze 181: "tbl(1) error",
1.40 schwarze 182:
183: "generic fatal error",
184:
1.34 schwarze 185: "column syntax is inconsistent",
1.56 schwarze 186: "NOT IMPLEMENTED: .Bd -file",
1.33 schwarze 187: "line scope broken, syntax violated",
188: "argument count wrong, violates syntax",
189: "child violates parent syntax",
190: "argument count wrong, violates syntax",
1.56 schwarze 191: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.33 schwarze 192: "no document body",
193: "no document prologue",
1.45 schwarze 194: "static buffer exhausted",
1.1 kristaps 195: };
196:
1.55 schwarze 197: static void parsebuf(struct curparse *, struct buf, int);
1.51 schwarze 198: static void pdesc(struct curparse *);
1.27 schwarze 199: static void fdesc(struct curparse *);
200: static void ffile(const char *, struct curparse *);
1.55 schwarze 201: static int pfile(const char *, struct curparse *);
1.1 kristaps 202: static int moptions(enum intt *, char *);
1.30 schwarze 203: static int mmsg(enum mandocerr, void *,
204: int, int, const char *);
1.59 schwarze 205: static void pset(const char *, int, struct curparse *);
1.27 schwarze 206: static int toptions(struct curparse *, char *);
207: static void usage(void) __attribute__((noreturn));
1.20 schwarze 208: static void version(void) __attribute__((noreturn));
1.45 schwarze 209: static int woptions(struct curparse *, char *);
1.1 kristaps 210:
1.19 schwarze 211: static const char *progname;
1.60 ! schwarze 212: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.45 schwarze 213: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 214:
215: int
216: main(int argc, char *argv[])
217: {
1.27 schwarze 218: int c;
1.1 kristaps 219: struct curparse curp;
220:
1.19 schwarze 221: progname = strrchr(argv[0], '/');
222: if (progname == NULL)
223: progname = argv[0];
224: else
225: ++progname;
226:
227: memset(&curp, 0, sizeof(struct curparse));
1.1 kristaps 228:
229: curp.inttype = INTT_AUTO;
230: curp.outtype = OUTT_ASCII;
1.45 schwarze 231: curp.wlevel = MANDOCLEVEL_FATAL;
1.1 kristaps 232:
233: /* LINTED */
1.45 schwarze 234: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 235: switch (c) {
236: case ('m'):
237: if ( ! moptions(&curp.inttype, optarg))
1.47 schwarze 238: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 239: break;
1.18 schwarze 240: case ('O'):
241: (void)strlcat(curp.outopts, optarg, BUFSIZ);
242: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.17 schwarze 243: break;
1.1 kristaps 244: case ('T'):
1.22 schwarze 245: if ( ! toptions(&curp, optarg))
1.47 schwarze 246: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 247: break;
248: case ('W'):
1.45 schwarze 249: if ( ! woptions(&curp, optarg))
1.47 schwarze 250: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 251: break;
1.3 schwarze 252: case ('V'):
253: version();
254: /* NOTREACHED */
1.1 kristaps 255: default:
256: usage();
257: /* NOTREACHED */
258: }
259:
260: argc -= optind;
261: argv += optind;
262:
1.7 schwarze 263: if (NULL == *argv) {
264: curp.file = "<stdin>";
265: curp.fd = STDIN_FILENO;
1.14 schwarze 266:
1.27 schwarze 267: fdesc(&curp);
268: }
269:
270: while (*argv) {
271: ffile(*argv, &curp);
1.45 schwarze 272: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.27 schwarze 273: break;
274: ++argv;
1.1 kristaps 275: }
276:
277: if (curp.outfree)
278: (*curp.outfree)(curp.outdata);
1.59 schwarze 279: if (curp.pmdoc)
280: mdoc_free(curp.pmdoc);
281: if (curp.pman)
282: man_free(curp.pman);
1.30 schwarze 283: if (curp.roff)
284: roff_free(curp.roff);
1.1 kristaps 285:
1.47 schwarze 286: return((int)exit_status);
1.1 kristaps 287: }
288:
289:
1.20 schwarze 290: static void
1.3 schwarze 291: version(void)
292: {
293:
1.19 schwarze 294: (void)printf("%s %s\n", progname, VERSION);
1.47 schwarze 295: exit((int)MANDOCLEVEL_OK);
1.3 schwarze 296: }
297:
298:
1.20 schwarze 299: static void
1.1 kristaps 300: usage(void)
301: {
302:
1.59 schwarze 303: (void)fprintf(stderr, "usage: %s "
304: "[-V] "
305: "[-foption] "
306: "[-mformat] "
307: "[-Ooption] "
308: "[-Toutput] "
309: "[-Werr] "
310: "[file...]\n",
311: progname);
312:
1.47 schwarze 313: exit((int)MANDOCLEVEL_BADARG);
1.1 kristaps 314: }
315:
1.27 schwarze 316: static void
317: ffile(const char *file, struct curparse *curp)
1.1 kristaps 318: {
319:
1.59 schwarze 320: /*
321: * Called once per input file. Get the file ready for reading,
322: * pass it through to the parser-driver, then close it out.
323: * XXX: don't do anything special as this is only called for
324: * files; stdin goes directly to fdesc().
325: */
326:
1.1 kristaps 327: curp->file = file;
1.59 schwarze 328:
1.1 kristaps 329: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.19 schwarze 330: perror(curp->file);
1.45 schwarze 331: exit_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 332: return;
1.1 kristaps 333: }
334:
1.27 schwarze 335: fdesc(curp);
1.1 kristaps 336:
337: if (-1 == close(curp->fd))
1.19 schwarze 338: perror(curp->file);
1.27 schwarze 339: }
1.1 kristaps 340:
1.52 schwarze 341: static int
1.55 schwarze 342: pfile(const char *file, struct curparse *curp)
1.52 schwarze 343: {
344: const char *savefile;
345: int fd, savefd;
346:
347: if (-1 == (fd = open(file, O_RDONLY, 0))) {
348: perror(file);
1.60 ! schwarze 349: file_status = MANDOCLEVEL_SYSERR;
1.52 schwarze 350: return(0);
351: }
352:
353: savefile = curp->file;
354: savefd = curp->fd;
355:
356: curp->file = file;
357: curp->fd = fd;
358:
359: pdesc(curp);
360:
361: curp->file = savefile;
362: curp->fd = savefd;
363:
364: if (-1 == close(fd))
365: perror(file);
366:
1.60 ! schwarze 367: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.52 schwarze 368: }
369:
1.27 schwarze 370:
1.45 schwarze 371: static void
1.27 schwarze 372: resize_buf(struct buf *buf, size_t initial)
373: {
374:
1.45 schwarze 375: buf->sz = buf->sz ? 2 * buf->sz : initial;
376: buf->buf = realloc(buf->buf, buf->sz);
377: if (NULL == buf->buf) {
1.27 schwarze 378: perror(NULL);
1.47 schwarze 379: exit((int)MANDOCLEVEL_SYSERR);
1.27 schwarze 380: }
1.1 kristaps 381: }
382:
383:
384: static int
1.27 schwarze 385: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
1.1 kristaps 386: {
1.27 schwarze 387: struct stat st;
388: size_t off;
1.1 kristaps 389: ssize_t ssz;
1.27 schwarze 390:
391: if (-1 == fstat(curp->fd, &st)) {
392: perror(curp->file);
393: return(0);
394: }
395:
396: /*
397: * If we're a regular file, try just reading in the whole entry
398: * via mmap(). This is faster than reading it into blocks, and
399: * since each file is only a few bytes to begin with, I'm not
400: * concerned that this is going to tank any machines.
401: */
402:
403: if (S_ISREG(st.st_mode)) {
404: if (st.st_size >= (1U << 31)) {
405: fprintf(stderr, "%s: input too large\n",
406: curp->file);
407: return(0);
408: }
409: *with_mmap = 1;
1.30 schwarze 410: fb->sz = (size_t)st.st_size;
1.27 schwarze 411: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.35 schwarze 412: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.27 schwarze 413: if (fb->buf != MAP_FAILED)
414: return(1);
415: }
416:
417: /*
418: * If this isn't a regular file (like, say, stdin), then we must
419: * go the old way and just read things in bit by bit.
420: */
421:
422: *with_mmap = 0;
423: off = 0;
424: fb->sz = 0;
425: fb->buf = NULL;
426: for (;;) {
427: if (off == fb->sz) {
428: if (fb->sz == (1U << 31)) {
429: fprintf(stderr, "%s: input too large\n",
430: curp->file);
431: break;
432: }
1.45 schwarze 433: resize_buf(fb, 65536);
1.27 schwarze 434: }
1.30 schwarze 435: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.27 schwarze 436: if (ssz == 0) {
437: fb->sz = off;
438: return(1);
439: }
440: if (ssz == -1) {
441: perror(curp->file);
442: break;
443: }
1.30 schwarze 444: off += (size_t)ssz;
1.27 schwarze 445: }
446:
447: free(fb->buf);
448: fb->buf = NULL;
449: return(0);
450: }
451:
452:
453: static void
454: fdesc(struct curparse *curp)
455: {
1.59 schwarze 456:
457: /*
458: * Called once per file with an opened file descriptor. All
459: * pre-file-parse operations (whether stdin or a file) should go
460: * here.
461: *
462: * This calls down into the nested parser, which drills down and
463: * fully parses a file and all its dependences (i.e., `so'). It
464: * then runs the cleanup validators and pushes to output.
465: */
466:
467: /* Zero the parse type. */
468:
469: curp->mdoc = NULL;
470: curp->man = NULL;
1.60 ! schwarze 471: file_status = MANDOCLEVEL_OK;
1.59 schwarze 472:
473: /* Make sure the mandotory roff parser is initialised. */
474:
475: if (NULL == curp->roff) {
476: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
477: assert(curp->roff);
478: }
479:
480: /* Fully parse the file. */
1.51 schwarze 481:
482: pdesc(curp);
483:
1.60 ! schwarze 484: if (MANDOCLEVEL_FATAL <= file_status)
1.51 schwarze 485: goto cleanup;
486:
487: /* NOTE a parser may not have been assigned, yet. */
488:
1.59 schwarze 489: if ( ! (curp->man || curp->mdoc)) {
1.51 schwarze 490: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.60 ! schwarze 491: file_status = MANDOCLEVEL_FATAL;
1.51 schwarze 492: goto cleanup;
493: }
494:
495: /* Clean up the parse routine ASTs. */
496:
1.59 schwarze 497: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.60 ! schwarze 498: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 499: goto cleanup;
500: }
1.59 schwarze 501:
502: if (curp->man && ! man_endparse(curp->man)) {
1.60 ! schwarze 503: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 504: goto cleanup;
505: }
1.59 schwarze 506:
507: assert(curp->roff);
508: if ( ! roff_endparse(curp->roff)) {
1.60 ! schwarze 509: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 510: goto cleanup;
511: }
512:
513: /*
514: * With -Wstop and warnings or errors of at least
515: * the requested level, do not produce output.
516: */
517:
1.60 ! schwarze 518: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.51 schwarze 519: goto cleanup;
520:
521: /* If unset, allocate output dev now (if applicable). */
522:
523: if ( ! (curp->outman && curp->outmdoc)) {
524: switch (curp->outtype) {
525: case (OUTT_XHTML):
526: curp->outdata = xhtml_alloc(curp->outopts);
527: break;
528: case (OUTT_HTML):
529: curp->outdata = html_alloc(curp->outopts);
530: break;
531: case (OUTT_ASCII):
532: curp->outdata = ascii_alloc(curp->outopts);
533: curp->outfree = ascii_free;
534: break;
535: case (OUTT_PDF):
536: curp->outdata = pdf_alloc(curp->outopts);
537: curp->outfree = pspdf_free;
538: break;
539: case (OUTT_PS):
540: curp->outdata = ps_alloc(curp->outopts);
541: curp->outfree = pspdf_free;
542: break;
543: default:
544: break;
545: }
546:
547: switch (curp->outtype) {
548: case (OUTT_HTML):
549: /* FALLTHROUGH */
550: case (OUTT_XHTML):
551: curp->outman = html_man;
552: curp->outmdoc = html_mdoc;
553: curp->outfree = html_free;
554: break;
555: case (OUTT_TREE):
556: curp->outman = tree_man;
557: curp->outmdoc = tree_mdoc;
558: break;
559: case (OUTT_PDF):
560: /* FALLTHROUGH */
561: case (OUTT_ASCII):
562: /* FALLTHROUGH */
563: case (OUTT_PS):
564: curp->outman = terminal_man;
565: curp->outmdoc = terminal_mdoc;
566: break;
567: default:
568: break;
569: }
570: }
571:
572: /* Execute the out device, if it exists. */
573:
1.59 schwarze 574: if (curp->man && curp->outman)
575: (*curp->outman)(curp->outdata, curp->man);
576: if (curp->mdoc && curp->outmdoc)
577: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.51 schwarze 578:
579: cleanup:
1.59 schwarze 580:
1.51 schwarze 581: memset(&curp->regs, 0, sizeof(struct regset));
1.59 schwarze 582:
583: /* Reset the current-parse compilers. */
584:
585: if (curp->mdoc)
586: mdoc_reset(curp->mdoc);
587: if (curp->man)
588: man_reset(curp->man);
589:
590: assert(curp->roff);
591: roff_reset(curp->roff);
1.51 schwarze 592:
1.60 ! schwarze 593: if (exit_status < file_status)
! 594: exit_status = file_status;
! 595:
1.51 schwarze 596: return;
597: }
598:
599: static void
600: pdesc(struct curparse *curp)
601: {
1.55 schwarze 602: struct buf blk;
603: int with_mmap;
1.1 kristaps 604:
1.59 schwarze 605: /*
606: * Run for each opened file; may be called more than once for
607: * each full parse sequence if the opened file is nested (i.e.,
608: * from `so'). Simply sucks in the whole file and moves into
609: * the parse phase for the file.
610: */
611:
1.45 schwarze 612: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.60 ! schwarze 613: file_status = MANDOCLEVEL_SYSERR;
1.27 schwarze 614: return;
1.45 schwarze 615: }
1.27 schwarze 616:
1.59 schwarze 617: /* Line number is per-file. */
1.55 schwarze 618:
619: curp->line = 1;
1.59 schwarze 620:
1.55 schwarze 621: parsebuf(curp, blk, 1);
622:
623: if (with_mmap)
624: munmap(blk.buf, blk.sz);
625: else
626: free(blk.buf);
627: }
628:
629: static void
630: parsebuf(struct curparse *curp, struct buf blk, int start)
631: {
632: struct buf ln;
1.59 schwarze 633: enum rofferr rr;
634: int i, of, rc;
635: int pos; /* byte number in the ln buffer */
636: int lnn; /* line number in the real file */
1.55 schwarze 637: unsigned char c;
1.59 schwarze 638:
639: /*
640: * Main parse routine for an opened file. This is called for
641: * each opened file and simply loops around the full input file,
642: * possibly nesting (i.e., with `so').
643: */
1.30 schwarze 644:
1.55 schwarze 645: memset(&ln, 0, sizeof(struct buf));
646:
1.59 schwarze 647: lnn = curp->line;
648: pos = 0;
1.55 schwarze 649:
1.59 schwarze 650: for (i = 0; i < (int)blk.sz; ) {
1.55 schwarze 651: if (0 == pos && '\0' == blk.buf[i])
652: break;
1.59 schwarze 653:
1.55 schwarze 654: if (start)
655: curp->line = lnn;
656:
657: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.29 schwarze 658: if ('\n' == blk.buf[i]) {
659: ++i;
660: ++lnn;
661: break;
662: }
1.43 schwarze 663:
664: /*
665: * Warn about bogus characters. If you're using
666: * non-ASCII encoding, you're screwing your
667: * readers. Since I'd rather this not happen,
668: * I'll be helpful and drop these characters so
669: * we don't display gibberish. Note to manual
670: * writers: use special characters.
671: */
672:
1.44 schwarze 673: c = (unsigned char) blk.buf[i];
1.59 schwarze 674:
675: if ( ! (isascii(c) &&
676: (isgraph(c) || isblank(c)))) {
1.45 schwarze 677: mmsg(MANDOCERR_BADCHAR, curp,
1.55 schwarze 678: curp->line, pos, "ignoring byte");
1.43 schwarze 679: i++;
680: continue;
681: }
682:
1.59 schwarze 683: /* Trailing backslash = a plain char. */
684:
1.29 schwarze 685: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
686: if (pos >= (int)ln.sz)
1.45 schwarze 687: resize_buf(&ln, 256);
1.29 schwarze 688: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 689: continue;
1.29 schwarze 690: }
1.59 schwarze 691:
692: /* Found escape & at least one other char. */
693:
1.29 schwarze 694: if ('\n' == blk.buf[i + 1]) {
1.59 schwarze 695: i += 2;
1.29 schwarze 696: /* Escaped newlines are skipped over */
697: ++lnn;
1.27 schwarze 698: continue;
699: }
1.59 schwarze 700:
1.29 schwarze 701: if ('"' == blk.buf[i + 1]) {
702: i += 2;
703: /* Comment, skip to end of line */
704: for (; i < (int)blk.sz; ++i) {
705: if ('\n' == blk.buf[i]) {
706: ++i;
707: ++lnn;
708: break;
709: }
710: }
1.59 schwarze 711:
1.29 schwarze 712: /* Backout trailing whitespaces */
713: for (; pos > 0; --pos) {
714: if (ln.buf[pos - 1] != ' ')
715: break;
716: if (pos > 2 && ln.buf[pos - 2] == '\\')
717: break;
718: }
719: break;
720: }
1.59 schwarze 721:
722: /* Some other escape sequence, copy & cont. */
723:
1.29 schwarze 724: if (pos + 1 >= (int)ln.sz)
1.45 schwarze 725: resize_buf(&ln, 256);
1.1 kristaps 726:
1.29 schwarze 727: ln.buf[pos++] = blk.buf[i++];
728: ln.buf[pos++] = blk.buf[i++];
1.27 schwarze 729: }
1.1 kristaps 730:
1.29 schwarze 731: if (pos >= (int)ln.sz)
1.45 schwarze 732: resize_buf(&ln, 256);
1.59 schwarze 733:
1.30 schwarze 734: ln.buf[pos] = '\0';
735:
1.32 schwarze 736: /*
737: * A significant amount of complexity is contained by
738: * the roff preprocessor. It's line-oriented but can be
739: * expressed on one line, so we need at times to
740: * readjust our starting point and re-run it. The roff
741: * preprocessor can also readjust the buffers with new
742: * data, so we pass them in wholesale.
743: */
744:
745: of = 0;
1.59 schwarze 746:
1.55 schwarze 747: rerun:
1.59 schwarze 748: rr = roff_parseln
749: (curp->roff, curp->line,
750: &ln.buf, &ln.sz, of, &of);
751:
752: switch (rr) {
1.55 schwarze 753: case (ROFF_REPARSE):
754: parsebuf(curp, ln, 0);
755: pos = 0;
756: continue;
757: case (ROFF_APPEND):
758: pos = strlen(ln.buf);
759: continue;
760: case (ROFF_RERUN):
761: goto rerun;
762: case (ROFF_IGN):
763: pos = 0;
1.30 schwarze 764: continue;
1.55 schwarze 765: case (ROFF_ERR):
1.60 ! schwarze 766: assert(MANDOCLEVEL_FATAL <= file_status);
1.51 schwarze 767: break;
1.55 schwarze 768: case (ROFF_SO):
769: if (pfile(ln.buf + of, curp)) {
770: pos = 0;
1.52 schwarze 771: continue;
1.55 schwarze 772: } else
1.52 schwarze 773: break;
1.55 schwarze 774: case (ROFF_CONT):
775: break;
1.45 schwarze 776: }
1.5 schwarze 777:
1.32 schwarze 778: /*
779: * If input parsers have not been allocated, do so now.
780: * We keep these instanced betwen parsers, but set them
781: * locally per parse routine since we can use different
782: * parsers with each one.
783: */
1.1 kristaps 784:
1.59 schwarze 785: if ( ! (curp->man || curp->mdoc))
786: pset(ln.buf + of, pos - of, curp);
1.5 schwarze 787:
1.59 schwarze 788: /*
789: * Lastly, push down into the parsers themselves. One
790: * of these will have already been set in the pset()
791: * routine.
792: */
793:
794: if (curp->man || curp->mdoc) {
795: rc = curp->man ?
796: man_parseln(curp->man,
797: curp->line, ln.buf, of) :
798: mdoc_parseln(curp->mdoc,
799: curp->line, ln.buf, of);
1.1 kristaps 800:
1.59 schwarze 801: if ( ! rc) {
1.60 ! schwarze 802: assert(MANDOCLEVEL_FATAL <= file_status);
1.59 schwarze 803: break;
804: }
1.1 kristaps 805: }
1.55 schwarze 806:
807: /* Temporary buffers typically are not full. */
1.59 schwarze 808:
1.55 schwarze 809: if (0 == start && '\0' == blk.buf[i])
810: break;
811:
812: /* Start the next input line. */
1.59 schwarze 813:
1.55 schwarze 814: pos = 0;
1.1 kristaps 815: }
816:
1.51 schwarze 817: free(ln.buf);
1.1 kristaps 818: }
819:
1.45 schwarze 820: static void
1.59 schwarze 821: pset(const char *buf, int pos, struct curparse *curp)
1.1 kristaps 822: {
1.5 schwarze 823: int i;
1.1 kristaps 824:
825: /*
826: * Try to intuit which kind of manual parser should be used. If
827: * passed in by command-line (-man, -mdoc), then use that
828: * explicitly. If passed as -mandoc, then try to guess from the
1.5 schwarze 829: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.1 kristaps 830: * default to -man, which is more lenient.
1.59 schwarze 831: *
832: * Separate out pmdoc/pman from mdoc/man: the first persists
833: * through all parsers, while the latter is used per-parse.
1.1 kristaps 834: */
835:
1.31 schwarze 836: if ('.' == buf[0] || '\'' == buf[0]) {
1.5 schwarze 837: for (i = 1; buf[i]; i++)
838: if (' ' != buf[i] && '\t' != buf[i])
839: break;
1.45 schwarze 840: if ('\0' == buf[i])
841: return;
1.5 schwarze 842: }
1.1 kristaps 843:
844: switch (curp->inttype) {
845: case (INTT_MDOC):
1.59 schwarze 846: if (NULL == curp->pmdoc)
847: curp->pmdoc = mdoc_alloc
848: (&curp->regs, curp, mmsg);
849: assert(curp->pmdoc);
850: curp->mdoc = curp->pmdoc;
1.45 schwarze 851: return;
1.1 kristaps 852: case (INTT_MAN):
1.59 schwarze 853: if (NULL == curp->pman)
854: curp->pman = man_alloc
855: (&curp->regs, curp, mmsg);
856: assert(curp->pman);
857: curp->man = curp->pman;
1.45 schwarze 858: return;
1.1 kristaps 859: default:
860: break;
861: }
862:
863: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.59 schwarze 864: if (NULL == curp->pmdoc)
865: curp->pmdoc = mdoc_alloc
866: (&curp->regs, curp, mmsg);
867: assert(curp->pmdoc);
868: curp->mdoc = curp->pmdoc;
1.45 schwarze 869: return;
1.1 kristaps 870: }
871:
1.59 schwarze 872: if (NULL == curp->pman)
873: curp->pman = man_alloc(&curp->regs, curp, mmsg);
874: assert(curp->pman);
875: curp->man = curp->pman;
1.1 kristaps 876: }
877:
878: static int
879: moptions(enum intt *tflags, char *arg)
880: {
881:
882: if (0 == strcmp(arg, "doc"))
883: *tflags = INTT_MDOC;
884: else if (0 == strcmp(arg, "andoc"))
885: *tflags = INTT_AUTO;
886: else if (0 == strcmp(arg, "an"))
887: *tflags = INTT_MAN;
888: else {
1.20 schwarze 889: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 890: return(0);
891: }
892:
893: return(1);
894: }
895:
896: static int
1.22 schwarze 897: toptions(struct curparse *curp, char *arg)
1.1 kristaps 898: {
899:
900: if (0 == strcmp(arg, "ascii"))
1.22 schwarze 901: curp->outtype = OUTT_ASCII;
902: else if (0 == strcmp(arg, "lint")) {
903: curp->outtype = OUTT_LINT;
1.45 schwarze 904: curp->wlevel = MANDOCLEVEL_WARNING;
1.22 schwarze 905: }
1.1 kristaps 906: else if (0 == strcmp(arg, "tree"))
1.22 schwarze 907: curp->outtype = OUTT_TREE;
1.17 schwarze 908: else if (0 == strcmp(arg, "html"))
1.22 schwarze 909: curp->outtype = OUTT_HTML;
1.21 schwarze 910: else if (0 == strcmp(arg, "xhtml"))
1.22 schwarze 911: curp->outtype = OUTT_XHTML;
1.36 schwarze 912: else if (0 == strcmp(arg, "ps"))
913: curp->outtype = OUTT_PS;
1.43 schwarze 914: else if (0 == strcmp(arg, "pdf"))
915: curp->outtype = OUTT_PDF;
1.1 kristaps 916: else {
1.20 schwarze 917: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 918: return(0);
919: }
920:
921: return(1);
922: }
923:
924: static int
1.45 schwarze 925: woptions(struct curparse *curp, char *arg)
1.1 kristaps 926: {
1.10 schwarze 927: char *v, *o;
1.45 schwarze 928: const char *toks[6];
1.1 kristaps 929:
1.45 schwarze 930: toks[0] = "stop";
931: toks[1] = "all";
932: toks[2] = "warning";
933: toks[3] = "error";
934: toks[4] = "fatal";
935: toks[5] = NULL;
1.1 kristaps 936:
1.10 schwarze 937: while (*arg) {
938: o = arg;
1.17 schwarze 939: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 940: case (0):
1.45 schwarze 941: curp->wstop = 1;
1.1 kristaps 942: break;
943: case (1):
1.45 schwarze 944: /* FALLTHROUGH */
1.1 kristaps 945: case (2):
1.45 schwarze 946: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 947: break;
948: case (3):
1.45 schwarze 949: curp->wlevel = MANDOCLEVEL_ERROR;
1.1 kristaps 950: break;
951: case (4):
1.45 schwarze 952: curp->wlevel = MANDOCLEVEL_FATAL;
1.19 schwarze 953: break;
1.1 kristaps 954: default:
1.45 schwarze 955: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 956: return(0);
957: }
1.10 schwarze 958: }
1.1 kristaps 959:
960: return(1);
961: }
962:
1.30 schwarze 963: static int
964: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
965: {
966: struct curparse *cp;
1.45 schwarze 967: enum mandoclevel level;
968:
969: level = MANDOCLEVEL_FATAL;
970: while (t < mandoclimits[level])
1.47 schwarze 971: /* LINTED */
1.45 schwarze 972: level--;
1.30 schwarze 973:
974: cp = (struct curparse *)arg;
1.45 schwarze 975: if (level < cp->wlevel)
976: return(1);
1.30 schwarze 977:
1.45 schwarze 978: fprintf(stderr, "%s:%d:%d: %s: %s",
979: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.30 schwarze 980: if (msg)
981: fprintf(stderr, ": %s", msg);
982: fputc('\n', stderr);
1.33 schwarze 983:
1.60 ! schwarze 984: if (file_status < level)
! 985: file_status = level;
1.45 schwarze 986:
987: return(level < MANDOCLEVEL_FATAL);
1.30 schwarze 988: }