Annotation of mandoc/main.c, Revision 1.120
1.120 ! kristaps 1: /* $Id: main.c,v 1.119 2010/12/06 11:01:19 kristaps Exp $ */
1.1 kristaps 2: /*
1.97 schwarze 3: * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2010 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.25 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.25 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.58 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.66 kristaps 22: #include <sys/mman.h>
1.1 kristaps 23: #include <sys/stat.h>
24:
25: #include <assert.h>
1.99 kristaps 26: #include <ctype.h>
1.1 kristaps 27: #include <fcntl.h>
28: #include <stdio.h>
1.45 kristaps 29: #include <stdint.h>
1.1 kristaps 30: #include <stdlib.h>
31: #include <string.h>
32: #include <unistd.h>
33:
1.71 kristaps 34: #include "mandoc.h"
1.90 kristaps 35: #include "main.h"
1.1 kristaps 36: #include "mdoc.h"
1.10 kristaps 37: #include "man.h"
1.71 kristaps 38: #include "roff.h"
1.101 joerg 39:
40: #ifndef MAP_FILE
41: #define MAP_FILE 0
42: #endif
1.1 kristaps 43:
1.45 kristaps 44: #define UNCONST(a) ((void *)(uintptr_t)(const void *)(a))
45:
1.55 kristaps 46: /* FIXME: Intel's compiler? LLVM? pcc? */
47:
48: #if !defined(__GNUC__) || (__GNUC__ < 2)
1.56 kristaps 49: # if !defined(lint)
50: # define __attribute__(x)
51: # endif
1.55 kristaps 52: #endif /* !defined(__GNUC__) || (__GNUC__ < 2) */
1.16 kristaps 53:
1.42 kristaps 54: typedef void (*out_mdoc)(void *, const struct mdoc *);
55: typedef void (*out_man)(void *, const struct man *);
1.22 kristaps 56: typedef void (*out_free)(void *);
57:
1.5 kristaps 58: struct buf {
59: char *buf;
60: size_t sz;
61: };
62:
1.19 kristaps 63: enum intt {
64: INTT_AUTO,
65: INTT_MDOC,
66: INTT_MAN
67: };
68:
69: enum outt {
70: OUTT_ASCII = 0,
71: OUTT_TREE,
1.43 kristaps 72: OUTT_HTML,
1.59 kristaps 73: OUTT_XHTML,
1.85 kristaps 74: OUTT_LINT,
1.100 kristaps 75: OUTT_PS,
76: OUTT_PDF
1.19 kristaps 77: };
78:
1.5 kristaps 79: struct curparse {
1.22 kristaps 80: const char *file; /* Current parse. */
81: int fd; /* Current parse. */
1.111 kristaps 82: int line; /* Line number in the file. */
1.103 schwarze 83: enum mandoclevel wlevel; /* Ignore messages below this. */
84: int wstop; /* Stop after a file with a warning. */
1.79 kristaps 85: enum intt inttype; /* which parser to use */
1.112 kristaps 86: struct man *pman; /* persistent man parser */
87: struct mdoc *pmdoc; /* persistent mdoc parser */
1.79 kristaps 88: struct man *man; /* man parser */
89: struct mdoc *mdoc; /* mdoc parser */
90: struct roff *roff; /* roff parser (!NULL) */
1.92 kristaps 91: struct regset regs; /* roff registers */
1.79 kristaps 92: enum outt outtype; /* which output to use */
93: out_mdoc outmdoc; /* mdoc output ptr */
94: out_man outman; /* man output ptr */
95: out_free outfree; /* free output ptr */
96: void *outdata; /* data for output */
97: char outopts[BUFSIZ]; /* buf of output opts */
98: };
99:
1.103 schwarze 100: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
101: "SUCCESS",
102: "RESERVED",
103: "WARNING",
104: "ERROR",
105: "FATAL",
106: "BADARG",
107: "SYSERR"
108: };
109:
110: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
111: MANDOCERR_OK,
112: MANDOCERR_WARNING,
113: MANDOCERR_WARNING,
114: MANDOCERR_ERROR,
115: MANDOCERR_FATAL,
116: MANDOCERR_MAX,
117: MANDOCERR_MAX
118: };
119:
1.79 kristaps 120: static const char * const mandocerrs[MANDOCERR_MAX] = {
121: "ok",
1.94 schwarze 122:
123: "generic warning",
124:
1.113 kristaps 125: ".so is fragile, better use ln(1)",
1.79 kristaps 126: "text should be uppercase",
1.81 kristaps 127: "sections out of conventional order",
1.79 kristaps 128: "section name repeats",
129: "out of order prologue",
130: "repeated prologue entry",
131: "list type must come first",
1.99 kristaps 132: "tab in non-literal context",
1.79 kristaps 133: "bad escape sequence",
134: "unterminated quoted string",
135: "argument requires the width argument",
136: "bad date argument",
137: "bad width argument",
1.81 kristaps 138: "unknown manual section",
1.79 kristaps 139: "section not in conventional manual section",
140: "end of line whitespace",
1.95 schwarze 141: "blocks badly nested",
1.94 schwarze 142:
143: "generic error",
144:
1.79 kristaps 145: "NAME section must come first",
146: "bad Boolean value",
1.120 ! kristaps 147: "unknown font",
1.79 kristaps 148: "child violates parent syntax",
1.116 kristaps 149: "displays may not be nested",
1.79 kristaps 150: "bad AT&T symbol",
1.109 kristaps 151: "bad standard",
1.79 kristaps 152: "list type repeated",
153: "display type repeated",
154: "argument repeated",
1.108 kristaps 155: "ignoring argument",
1.79 kristaps 156: "manual name not yet set",
157: "obsolete macro ignored",
158: "empty macro ignored",
159: "macro not allowed in body",
160: "macro not allowed in prologue",
161: "bad character",
162: "bad NAME section contents",
163: "no blank lines",
164: "no text in this context",
165: "bad comment style",
166: "unknown macro will be lost",
1.110 kristaps 167: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 168: "line scope broken",
169: "argument count wrong",
170: "request scope close w/none open",
171: "scope already open",
1.106 schwarze 172: "scope open on exit",
1.117 kristaps 173: "uname(3) system call failed",
1.79 kristaps 174: "macro requires line argument(s)",
175: "macro requires body argument(s)",
176: "macro requires argument(s)",
177: "no title in document",
1.82 kristaps 178: "missing list type",
1.87 kristaps 179: "missing display type",
1.96 kristaps 180: "missing font type",
1.79 kristaps 181: "line argument(s) will be lost",
182: "body argument(s) will be lost",
1.107 kristaps 183: "paragraph macro ignored",
1.94 schwarze 184:
185: "generic fatal error",
186:
1.80 kristaps 187: "column syntax is inconsistent",
1.87 kristaps 188: "unsupported display type",
1.95 schwarze 189: "blocks badly nested",
190: "no such block is open",
1.119 kristaps 191: "missing end of block",
1.79 kristaps 192: "line scope broken, syntax violated",
193: "argument count wrong, violates syntax",
194: "child violates parent syntax",
195: "argument count wrong, violates syntax",
1.113 kristaps 196: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 197: "no document body",
198: "no document prologue",
1.103 schwarze 199: "static buffer exhausted",
1.5 kristaps 200: };
1.1 kristaps 201:
1.111 kristaps 202: static void parsebuf(struct curparse *, struct buf, int);
203: static void pdesc(struct curparse *);
1.66 kristaps 204: static void fdesc(struct curparse *);
205: static void ffile(const char *, struct curparse *);
1.111 kristaps 206: static int pfile(const char *, struct curparse *);
1.10 kristaps 207: static int moptions(enum intt *, char *);
1.71 kristaps 208: static int mmsg(enum mandocerr, void *,
209: int, int, const char *);
1.112 kristaps 210: static void pset(const char *, int, struct curparse *);
1.66 kristaps 211: static int toptions(struct curparse *, char *);
212: static void usage(void) __attribute__((noreturn));
1.55 kristaps 213: static void version(void) __attribute__((noreturn));
1.103 schwarze 214: static int woptions(struct curparse *, char *);
1.1 kristaps 215:
1.54 kristaps 216: static const char *progname;
1.115 schwarze 217: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 218: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 219:
220: int
221: main(int argc, char *argv[])
222: {
1.64 kristaps 223: int c;
1.5 kristaps 224: struct curparse curp;
1.1 kristaps 225:
1.54 kristaps 226: progname = strrchr(argv[0], '/');
227: if (progname == NULL)
228: progname = argv[0];
229: else
230: ++progname;
231:
1.51 kristaps 232: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 233:
1.19 kristaps 234: curp.inttype = INTT_AUTO;
1.22 kristaps 235: curp.outtype = OUTT_ASCII;
1.103 schwarze 236: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 237:
1.1 kristaps 238: /* LINTED */
1.103 schwarze 239: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 240: switch (c) {
1.10 kristaps 241: case ('m'):
1.19 kristaps 242: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 243: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 244: break;
1.48 kristaps 245: case ('O'):
1.49 kristaps 246: (void)strlcat(curp.outopts, optarg, BUFSIZ);
247: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 248: break;
1.1 kristaps 249: case ('T'):
1.60 kristaps 250: if ( ! toptions(&curp, optarg))
1.105 kristaps 251: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 252: break;
253: case ('W'):
1.103 schwarze 254: if ( ! woptions(&curp, optarg))
1.105 kristaps 255: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 256: break;
257: case ('V'):
258: version();
259: /* NOTREACHED */
260: default:
261: usage();
262: /* NOTREACHED */
263: }
264:
265: argc -= optind;
266: argv += optind;
267:
1.30 kristaps 268: if (NULL == *argv) {
269: curp.file = "<stdin>";
270: curp.fd = STDIN_FILENO;
1.39 kristaps 271:
1.66 kristaps 272: fdesc(&curp);
1.64 kristaps 273: }
274:
275: while (*argv) {
1.66 kristaps 276: ffile(*argv, &curp);
1.103 schwarze 277: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 278: break;
279: ++argv;
1.1 kristaps 280: }
281:
1.22 kristaps 282: if (curp.outfree)
283: (*curp.outfree)(curp.outdata);
1.112 kristaps 284: if (curp.pmdoc)
285: mdoc_free(curp.pmdoc);
286: if (curp.pman)
287: man_free(curp.pman);
1.71 kristaps 288: if (curp.roff)
289: roff_free(curp.roff);
1.1 kristaps 290:
1.105 kristaps 291: return((int)exit_status);
1.1 kristaps 292: }
293:
294:
1.55 kristaps 295: static void
1.1 kristaps 296: version(void)
297: {
298:
1.54 kristaps 299: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 300: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 301: }
302:
303:
1.55 kristaps 304: static void
1.1 kristaps 305: usage(void)
306: {
307:
1.112 kristaps 308: (void)fprintf(stderr, "usage: %s "
309: "[-V] "
310: "[-foption] "
311: "[-mformat] "
312: "[-Ooption] "
313: "[-Toutput] "
314: "[-Werr] "
315: "[file...]\n",
316: progname);
317:
1.105 kristaps 318: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 319: }
320:
1.64 kristaps 321: static void
1.66 kristaps 322: ffile(const char *file, struct curparse *curp)
1.1 kristaps 323: {
324:
1.112 kristaps 325: /*
326: * Called once per input file. Get the file ready for reading,
327: * pass it through to the parser-driver, then close it out.
328: * XXX: don't do anything special as this is only called for
329: * files; stdin goes directly to fdesc().
330: */
331:
1.19 kristaps 332: curp->file = file;
1.112 kristaps 333:
1.19 kristaps 334: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 335: perror(curp->file);
1.103 schwarze 336: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 337: return;
1.1 kristaps 338: }
339:
1.66 kristaps 340: fdesc(curp);
1.1 kristaps 341:
1.19 kristaps 342: if (-1 == close(curp->fd))
1.53 kristaps 343: perror(curp->file);
1.1 kristaps 344: }
345:
1.111 kristaps 346: static int
347: pfile(const char *file, struct curparse *curp)
348: {
349: const char *savefile;
350: int fd, savefd;
351:
352: if (-1 == (fd = open(file, O_RDONLY, 0))) {
353: perror(file);
1.115 schwarze 354: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 355: return(0);
356: }
357:
358: savefile = curp->file;
359: savefd = curp->fd;
360:
361: curp->file = file;
362: curp->fd = fd;
363:
364: pdesc(curp);
365:
366: curp->file = savefile;
367: curp->fd = savefd;
368:
369: if (-1 == close(fd))
370: perror(file);
371:
1.115 schwarze 372: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 373: }
374:
1.1 kristaps 375:
1.103 schwarze 376: static void
1.68 joerg 377: resize_buf(struct buf *buf, size_t initial)
378: {
379:
1.103 schwarze 380: buf->sz = buf->sz ? 2 * buf->sz : initial;
381: buf->buf = realloc(buf->buf, buf->sz);
382: if (NULL == buf->buf) {
1.68 joerg 383: perror(NULL);
1.105 kristaps 384: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 385: }
386: }
387:
388:
389: static int
1.66 kristaps 390: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
391: {
392: struct stat st;
1.68 joerg 393: size_t off;
1.66 kristaps 394: ssize_t ssz;
395:
396: if (-1 == fstat(curp->fd, &st)) {
397: perror(curp->file);
398: return(0);
399: }
400:
401: /*
402: * If we're a regular file, try just reading in the whole entry
403: * via mmap(). This is faster than reading it into blocks, and
404: * since each file is only a few bytes to begin with, I'm not
405: * concerned that this is going to tank any machines.
406: */
407:
408: if (S_ISREG(st.st_mode)) {
409: if (st.st_size >= (1U << 31)) {
410: fprintf(stderr, "%s: input too large\n",
411: curp->file);
412: return(0);
413: }
414: *with_mmap = 1;
1.71 kristaps 415: fb->sz = (size_t)st.st_size;
1.66 kristaps 416: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 417: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 418: if (fb->buf != MAP_FAILED)
419: return(1);
420: }
421:
422: /*
423: * If this isn't a regular file (like, say, stdin), then we must
424: * go the old way and just read things in bit by bit.
425: */
426:
427: *with_mmap = 0;
428: off = 0;
429: fb->sz = 0;
430: fb->buf = NULL;
431: for (;;) {
432: if (off == fb->sz) {
433: if (fb->sz == (1U << 31)) {
434: fprintf(stderr, "%s: input too large\n",
435: curp->file);
436: break;
437: }
1.103 schwarze 438: resize_buf(fb, 65536);
1.66 kristaps 439: }
1.71 kristaps 440: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 441: if (ssz == 0) {
442: fb->sz = off;
443: return(1);
444: }
445: if (ssz == -1) {
446: perror(curp->file);
447: break;
448: }
1.71 kristaps 449: off += (size_t)ssz;
1.66 kristaps 450: }
451:
452: free(fb->buf);
453: fb->buf = NULL;
454: return(0);
455: }
456:
457:
1.64 kristaps 458: static void
1.66 kristaps 459: fdesc(struct curparse *curp)
1.1 kristaps 460: {
1.112 kristaps 461:
462: /*
463: * Called once per file with an opened file descriptor. All
464: * pre-file-parse operations (whether stdin or a file) should go
465: * here.
466: *
467: * This calls down into the nested parser, which drills down and
468: * fully parses a file and all its dependences (i.e., `so'). It
469: * then runs the cleanup validators and pushes to output.
470: */
471:
472: /* Zero the parse type. */
473:
474: curp->mdoc = NULL;
475: curp->man = NULL;
1.115 schwarze 476: file_status = MANDOCLEVEL_OK;
1.112 kristaps 477:
478: /* Make sure the mandotory roff parser is initialised. */
479:
480: if (NULL == curp->roff) {
481: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
482: assert(curp->roff);
483: }
484:
485: /* Fully parse the file. */
1.10 kristaps 486:
1.111 kristaps 487: pdesc(curp);
1.92 kristaps 488:
1.115 schwarze 489: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 490: goto cleanup;
491:
492: /* NOTE a parser may not have been assigned, yet. */
493:
1.112 kristaps 494: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 495: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 496: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 497: goto cleanup;
498: }
499:
500: /* Clean up the parse routine ASTs. */
501:
1.112 kristaps 502: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 503: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 504: goto cleanup;
505: }
1.112 kristaps 506:
507: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 508: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 509: goto cleanup;
510: }
1.112 kristaps 511:
512: assert(curp->roff);
513: if ( ! roff_endparse(curp->roff)) {
1.115 schwarze 514: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 515: goto cleanup;
516: }
1.1 kristaps 517:
518: /*
1.111 kristaps 519: * With -Wstop and warnings or errors of at least
520: * the requested level, do not produce output.
1.1 kristaps 521: */
522:
1.115 schwarze 523: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 524: goto cleanup;
525:
526: /* If unset, allocate output dev now (if applicable). */
527:
528: if ( ! (curp->outman && curp->outmdoc)) {
529: switch (curp->outtype) {
530: case (OUTT_XHTML):
531: curp->outdata = xhtml_alloc(curp->outopts);
532: break;
533: case (OUTT_HTML):
534: curp->outdata = html_alloc(curp->outopts);
535: break;
536: case (OUTT_ASCII):
537: curp->outdata = ascii_alloc(curp->outopts);
538: curp->outfree = ascii_free;
539: break;
540: case (OUTT_PDF):
541: curp->outdata = pdf_alloc(curp->outopts);
542: curp->outfree = pspdf_free;
543: break;
544: case (OUTT_PS):
545: curp->outdata = ps_alloc(curp->outopts);
546: curp->outfree = pspdf_free;
547: break;
548: default:
549: break;
550: }
551:
552: switch (curp->outtype) {
553: case (OUTT_HTML):
554: /* FALLTHROUGH */
555: case (OUTT_XHTML):
556: curp->outman = html_man;
557: curp->outmdoc = html_mdoc;
558: curp->outfree = html_free;
559: break;
560: case (OUTT_TREE):
561: curp->outman = tree_man;
562: curp->outmdoc = tree_mdoc;
563: break;
564: case (OUTT_PDF):
565: /* FALLTHROUGH */
566: case (OUTT_ASCII):
567: /* FALLTHROUGH */
568: case (OUTT_PS):
569: curp->outman = terminal_man;
570: curp->outmdoc = terminal_mdoc;
571: break;
572: default:
573: break;
574: }
575: }
576:
577: /* Execute the out device, if it exists. */
578:
1.112 kristaps 579: if (curp->man && curp->outman)
580: (*curp->outman)(curp->outdata, curp->man);
581: if (curp->mdoc && curp->outmdoc)
582: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 583:
584: cleanup:
1.112 kristaps 585:
1.111 kristaps 586: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 587:
588: /* Reset the current-parse compilers. */
589:
590: if (curp->mdoc)
591: mdoc_reset(curp->mdoc);
592: if (curp->man)
593: man_reset(curp->man);
594:
595: assert(curp->roff);
596: roff_reset(curp->roff);
1.111 kristaps 597:
1.115 schwarze 598: if (exit_status < file_status)
599: exit_status = file_status;
600:
1.111 kristaps 601: return;
602: }
603:
604: static void
605: pdesc(struct curparse *curp)
606: {
607: struct buf blk;
608: int with_mmap;
609:
1.112 kristaps 610: /*
611: * Run for each opened file; may be called more than once for
612: * each full parse sequence if the opened file is nested (i.e.,
613: * from `so'). Simply sucks in the whole file and moves into
614: * the parse phase for the file.
615: */
616:
1.103 schwarze 617: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 618: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 619: return;
1.103 schwarze 620: }
1.1 kristaps 621:
1.112 kristaps 622: /* Line number is per-file. */
1.111 kristaps 623:
624: curp->line = 1;
1.112 kristaps 625:
1.111 kristaps 626: parsebuf(curp, blk, 1);
627:
628: if (with_mmap)
629: munmap(blk.buf, blk.sz);
630: else
631: free(blk.buf);
632: }
633:
634: static void
635: parsebuf(struct curparse *curp, struct buf blk, int start)
636: {
637: struct buf ln;
1.114 kristaps 638: enum rofferr rr;
1.112 kristaps 639: int i, of, rc;
640: int pos; /* byte number in the ln buffer */
641: int lnn; /* line number in the real file */
1.111 kristaps 642: unsigned char c;
1.112 kristaps 643:
644: /*
645: * Main parse routine for an opened file. This is called for
646: * each opened file and simply loops around the full input file,
647: * possibly nesting (i.e., with `so').
648: */
1.71 kristaps 649:
1.111 kristaps 650: memset(&ln, 0, sizeof(struct buf));
651:
1.112 kristaps 652: lnn = curp->line;
653: pos = 0;
1.111 kristaps 654:
1.112 kristaps 655: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 656: if (0 == pos && '\0' == blk.buf[i])
657: break;
1.112 kristaps 658:
1.111 kristaps 659: if (start)
660: curp->line = lnn;
661:
662: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 663: if ('\n' == blk.buf[i]) {
664: ++i;
665: ++lnn;
666: break;
667: }
1.99 kristaps 668:
669: /*
670: * Warn about bogus characters. If you're using
671: * non-ASCII encoding, you're screwing your
672: * readers. Since I'd rather this not happen,
673: * I'll be helpful and drop these characters so
674: * we don't display gibberish. Note to manual
675: * writers: use special characters.
676: */
677:
1.102 schwarze 678: c = (unsigned char) blk.buf[i];
1.112 kristaps 679:
680: if ( ! (isascii(c) &&
681: (isgraph(c) || isblank(c)))) {
1.103 schwarze 682: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 683: curp->line, pos, "ignoring byte");
1.99 kristaps 684: i++;
685: continue;
686: }
687:
1.112 kristaps 688: /* Trailing backslash = a plain char. */
689:
1.70 joerg 690: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
691: if (pos >= (int)ln.sz)
1.103 schwarze 692: resize_buf(&ln, 256);
1.70 joerg 693: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 694: continue;
1.70 joerg 695: }
1.112 kristaps 696:
697: /* Found escape & at least one other char. */
698:
1.70 joerg 699: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 700: i += 2;
1.70 joerg 701: /* Escaped newlines are skipped over */
702: ++lnn;
1.67 joerg 703: continue;
704: }
1.112 kristaps 705:
1.70 joerg 706: if ('"' == blk.buf[i + 1]) {
707: i += 2;
708: /* Comment, skip to end of line */
709: for (; i < (int)blk.sz; ++i) {
710: if ('\n' == blk.buf[i]) {
711: ++i;
712: ++lnn;
713: break;
714: }
715: }
1.112 kristaps 716:
1.70 joerg 717: /* Backout trailing whitespaces */
718: for (; pos > 0; --pos) {
719: if (ln.buf[pos - 1] != ' ')
720: break;
721: if (pos > 2 && ln.buf[pos - 2] == '\\')
722: break;
723: }
724: break;
725: }
1.112 kristaps 726:
727: /* Some other escape sequence, copy & cont. */
728:
1.70 joerg 729: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 730: resize_buf(&ln, 256);
1.1 kristaps 731:
1.70 joerg 732: ln.buf[pos++] = blk.buf[i++];
733: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 734: }
1.1 kristaps 735:
1.70 joerg 736: if (pos >= (int)ln.sz)
1.103 schwarze 737: resize_buf(&ln, 256);
1.112 kristaps 738:
1.71 kristaps 739: ln.buf[pos] = '\0';
740:
1.76 kristaps 741: /*
742: * A significant amount of complexity is contained by
743: * the roff preprocessor. It's line-oriented but can be
744: * expressed on one line, so we need at times to
745: * readjust our starting point and re-run it. The roff
746: * preprocessor can also readjust the buffers with new
747: * data, so we pass them in wholesale.
748: */
749:
750: of = 0;
1.112 kristaps 751:
1.111 kristaps 752: rerun:
1.114 kristaps 753: rr = roff_parseln
1.112 kristaps 754: (curp->roff, curp->line,
755: &ln.buf, &ln.sz, of, &of);
756:
1.114 kristaps 757: switch (rr) {
1.111 kristaps 758: case (ROFF_REPARSE):
759: parsebuf(curp, ln, 0);
760: pos = 0;
761: continue;
762: case (ROFF_APPEND):
763: pos = strlen(ln.buf);
764: continue;
765: case (ROFF_RERUN):
766: goto rerun;
767: case (ROFF_IGN):
768: pos = 0;
1.71 kristaps 769: continue;
1.111 kristaps 770: case (ROFF_ERR):
1.115 schwarze 771: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 772: break;
773: case (ROFF_SO):
774: if (pfile(ln.buf + of, curp)) {
775: pos = 0;
776: continue;
777: } else
778: break;
779: case (ROFF_CONT):
780: break;
1.103 schwarze 781: }
1.29 kristaps 782:
1.76 kristaps 783: /*
784: * If input parsers have not been allocated, do so now.
785: * We keep these instanced betwen parsers, but set them
786: * locally per parse routine since we can use different
787: * parsers with each one.
788: */
1.19 kristaps 789:
1.112 kristaps 790: if ( ! (curp->man || curp->mdoc))
791: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 792:
1.112 kristaps 793: /*
794: * Lastly, push down into the parsers themselves. One
795: * of these will have already been set in the pset()
796: * routine.
797: */
798:
799: if (curp->man || curp->mdoc) {
800: rc = curp->man ?
801: man_parseln(curp->man,
802: curp->line, ln.buf, of) :
803: mdoc_parseln(curp->mdoc,
804: curp->line, ln.buf, of);
1.30 kristaps 805:
1.112 kristaps 806: if ( ! rc) {
1.115 schwarze 807: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 808: break;
809: }
1.103 schwarze 810: }
1.19 kristaps 811:
1.111 kristaps 812: /* Temporary buffers typically are not full. */
1.112 kristaps 813:
1.111 kristaps 814: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 815: break;
1.85 kristaps 816:
1.111 kristaps 817: /* Start the next input line. */
1.112 kristaps 818:
1.111 kristaps 819: pos = 0;
1.22 kristaps 820: }
821:
1.111 kristaps 822: free(ln.buf);
1.19 kristaps 823: }
824:
1.103 schwarze 825: static void
1.112 kristaps 826: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 827: {
1.29 kristaps 828: int i;
1.19 kristaps 829:
830: /*
831: * Try to intuit which kind of manual parser should be used. If
832: * passed in by command-line (-man, -mdoc), then use that
833: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 834: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 835: * default to -man, which is more lenient.
1.112 kristaps 836: *
837: * Separate out pmdoc/pman from mdoc/man: the first persists
838: * through all parsers, while the latter is used per-parse.
1.19 kristaps 839: */
840:
1.75 kristaps 841: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 842: for (i = 1; buf[i]; i++)
843: if (' ' != buf[i] && '\t' != buf[i])
844: break;
1.103 schwarze 845: if ('\0' == buf[i])
846: return;
1.29 kristaps 847: }
1.10 kristaps 848:
1.19 kristaps 849: switch (curp->inttype) {
850: case (INTT_MDOC):
1.112 kristaps 851: if (NULL == curp->pmdoc)
852: curp->pmdoc = mdoc_alloc
853: (&curp->regs, curp, mmsg);
854: assert(curp->pmdoc);
855: curp->mdoc = curp->pmdoc;
1.103 schwarze 856: return;
1.19 kristaps 857: case (INTT_MAN):
1.112 kristaps 858: if (NULL == curp->pman)
859: curp->pman = man_alloc
860: (&curp->regs, curp, mmsg);
861: assert(curp->pman);
862: curp->man = curp->pman;
1.103 schwarze 863: return;
1.19 kristaps 864: default:
865: break;
866: }
867:
868: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 869: if (NULL == curp->pmdoc)
870: curp->pmdoc = mdoc_alloc
871: (&curp->regs, curp, mmsg);
872: assert(curp->pmdoc);
873: curp->mdoc = curp->pmdoc;
1.103 schwarze 874: return;
1.19 kristaps 875: }
876:
1.112 kristaps 877: if (NULL == curp->pman)
878: curp->pman = man_alloc(&curp->regs, curp, mmsg);
879: assert(curp->pman);
880: curp->man = curp->pman;
1.10 kristaps 881: }
882:
883: static int
884: moptions(enum intt *tflags, char *arg)
885: {
886:
1.17 kristaps 887: if (0 == strcmp(arg, "doc"))
1.10 kristaps 888: *tflags = INTT_MDOC;
1.19 kristaps 889: else if (0 == strcmp(arg, "andoc"))
890: *tflags = INTT_AUTO;
1.17 kristaps 891: else if (0 == strcmp(arg, "an"))
1.10 kristaps 892: *tflags = INTT_MAN;
893: else {
1.57 kristaps 894: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 895: return(0);
896: }
897:
898: return(1);
1.1 kristaps 899: }
900:
901: static int
1.60 kristaps 902: toptions(struct curparse *curp, char *arg)
1.1 kristaps 903: {
904:
905: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 906: curp->outtype = OUTT_ASCII;
907: else if (0 == strcmp(arg, "lint")) {
908: curp->outtype = OUTT_LINT;
1.103 schwarze 909: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 910: }
1.1 kristaps 911: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 912: curp->outtype = OUTT_TREE;
1.43 kristaps 913: else if (0 == strcmp(arg, "html"))
1.60 kristaps 914: curp->outtype = OUTT_HTML;
1.59 kristaps 915: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 916: curp->outtype = OUTT_XHTML;
1.85 kristaps 917: else if (0 == strcmp(arg, "ps"))
918: curp->outtype = OUTT_PS;
1.100 kristaps 919: else if (0 == strcmp(arg, "pdf"))
920: curp->outtype = OUTT_PDF;
1.1 kristaps 921: else {
1.57 kristaps 922: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 923: return(0);
924: }
925:
926: return(1);
927: }
928:
929: static int
1.103 schwarze 930: woptions(struct curparse *curp, char *arg)
1.1 kristaps 931: {
1.34 kristaps 932: char *v, *o;
1.103 schwarze 933: const char *toks[6];
1.1 kristaps 934:
1.103 schwarze 935: toks[0] = "stop";
936: toks[1] = "all";
937: toks[2] = "warning";
938: toks[3] = "error";
939: toks[4] = "fatal";
940: toks[5] = NULL;
1.1 kristaps 941:
1.34 kristaps 942: while (*arg) {
943: o = arg;
1.45 kristaps 944: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 945: case (0):
1.103 schwarze 946: curp->wstop = 1;
1.1 kristaps 947: break;
948: case (1):
1.103 schwarze 949: /* FALLTHROUGH */
1.1 kristaps 950: case (2):
1.103 schwarze 951: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 952: break;
1.20 kristaps 953: case (3):
1.103 schwarze 954: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 955: break;
956: case (4):
1.103 schwarze 957: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 958: break;
1.1 kristaps 959: default:
1.103 schwarze 960: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 961: return(0);
962: }
1.34 kristaps 963: }
1.1 kristaps 964:
965: return(1);
966: }
967:
1.71 kristaps 968: static int
969: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
970: {
971: struct curparse *cp;
1.103 schwarze 972: enum mandoclevel level;
973:
974: level = MANDOCLEVEL_FATAL;
975: while (t < mandoclimits[level])
1.105 kristaps 976: /* LINTED */
1.103 schwarze 977: level--;
1.71 kristaps 978:
979: cp = (struct curparse *)arg;
1.103 schwarze 980: if (level < cp->wlevel)
981: return(1);
1.73 kristaps 982:
1.103 schwarze 983: fprintf(stderr, "%s:%d:%d: %s: %s",
984: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 985: if (msg)
986: fprintf(stderr, ": %s", msg);
987: fputc('\n', stderr);
1.79 kristaps 988:
1.115 schwarze 989: if (file_status < level)
990: file_status = level;
1.103 schwarze 991:
992: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 993: }
CVSweb