Annotation of mandoc/main.c, Revision 1.118
1.118 ! kristaps 1: /* $Id: main.c,v 1.117 2010/12/05 15:49:37 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",
147: "child violates parent syntax",
1.116 kristaps 148: "displays may not be nested",
1.79 kristaps 149: "bad AT&T symbol",
1.109 kristaps 150: "bad standard",
1.79 kristaps 151: "list type repeated",
152: "display type repeated",
153: "argument repeated",
1.108 kristaps 154: "ignoring argument",
1.79 kristaps 155: "manual name not yet set",
156: "obsolete macro ignored",
157: "empty macro ignored",
158: "macro not allowed in body",
159: "macro not allowed in prologue",
160: "bad character",
161: "bad NAME section contents",
162: "no blank lines",
163: "no text in this context",
164: "bad comment style",
165: "unknown macro will be lost",
1.110 kristaps 166: "NOT IMPLEMENTED: skipping request",
1.79 kristaps 167: "line scope broken",
168: "argument count wrong",
169: "request scope close w/none open",
170: "scope already open",
1.106 schwarze 171: "scope open on exit",
1.117 kristaps 172: "uname(3) system call failed",
1.79 kristaps 173: "macro requires line argument(s)",
174: "macro requires body argument(s)",
175: "macro requires argument(s)",
176: "no title in document",
1.82 kristaps 177: "missing list type",
1.87 kristaps 178: "missing display type",
1.96 kristaps 179: "missing font type",
1.79 kristaps 180: "line argument(s) will be lost",
181: "body argument(s) will be lost",
1.107 kristaps 182: "paragraph macro ignored",
1.94 schwarze 183:
184: "generic fatal error",
185:
1.80 kristaps 186: "column syntax is inconsistent",
1.87 kristaps 187: "unsupported display type",
1.95 schwarze 188: "blocks badly nested",
189: "no such block is open",
1.79 kristaps 190: "line scope broken, syntax violated",
191: "argument count wrong, violates syntax",
192: "child violates parent syntax",
193: "argument count wrong, violates syntax",
1.113 kristaps 194: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.79 kristaps 195: "no document body",
196: "no document prologue",
1.103 schwarze 197: "static buffer exhausted",
1.5 kristaps 198: };
1.1 kristaps 199:
1.111 kristaps 200: static void parsebuf(struct curparse *, struct buf, int);
201: static void pdesc(struct curparse *);
1.66 kristaps 202: static void fdesc(struct curparse *);
203: static void ffile(const char *, struct curparse *);
1.111 kristaps 204: static int pfile(const char *, struct curparse *);
1.10 kristaps 205: static int moptions(enum intt *, char *);
1.71 kristaps 206: static int mmsg(enum mandocerr, void *,
207: int, int, const char *);
1.112 kristaps 208: static void pset(const char *, int, struct curparse *);
1.66 kristaps 209: static int toptions(struct curparse *, char *);
210: static void usage(void) __attribute__((noreturn));
1.55 kristaps 211: static void version(void) __attribute__((noreturn));
1.103 schwarze 212: static int woptions(struct curparse *, char *);
1.1 kristaps 213:
1.54 kristaps 214: static const char *progname;
1.115 schwarze 215: static enum mandoclevel file_status = MANDOCLEVEL_OK;
1.103 schwarze 216: static enum mandoclevel exit_status = MANDOCLEVEL_OK;
1.1 kristaps 217:
218: int
219: main(int argc, char *argv[])
220: {
1.64 kristaps 221: int c;
1.5 kristaps 222: struct curparse curp;
1.1 kristaps 223:
1.54 kristaps 224: progname = strrchr(argv[0], '/');
225: if (progname == NULL)
226: progname = argv[0];
227: else
228: ++progname;
229:
1.51 kristaps 230: memset(&curp, 0, sizeof(struct curparse));
1.5 kristaps 231:
1.19 kristaps 232: curp.inttype = INTT_AUTO;
1.22 kristaps 233: curp.outtype = OUTT_ASCII;
1.103 schwarze 234: curp.wlevel = MANDOCLEVEL_FATAL;
1.19 kristaps 235:
1.1 kristaps 236: /* LINTED */
1.103 schwarze 237: while (-1 != (c = getopt(argc, argv, "m:O:T:VW:")))
1.1 kristaps 238: switch (c) {
1.10 kristaps 239: case ('m'):
1.19 kristaps 240: if ( ! moptions(&curp.inttype, optarg))
1.105 kristaps 241: return((int)MANDOCLEVEL_BADARG);
1.10 kristaps 242: break;
1.48 kristaps 243: case ('O'):
1.49 kristaps 244: (void)strlcat(curp.outopts, optarg, BUFSIZ);
245: (void)strlcat(curp.outopts, ",", BUFSIZ);
1.44 kristaps 246: break;
1.1 kristaps 247: case ('T'):
1.60 kristaps 248: if ( ! toptions(&curp, optarg))
1.105 kristaps 249: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 250: break;
251: case ('W'):
1.103 schwarze 252: if ( ! woptions(&curp, optarg))
1.105 kristaps 253: return((int)MANDOCLEVEL_BADARG);
1.1 kristaps 254: break;
255: case ('V'):
256: version();
257: /* NOTREACHED */
258: default:
259: usage();
260: /* NOTREACHED */
261: }
262:
263: argc -= optind;
264: argv += optind;
265:
1.30 kristaps 266: if (NULL == *argv) {
267: curp.file = "<stdin>";
268: curp.fd = STDIN_FILENO;
1.39 kristaps 269:
1.66 kristaps 270: fdesc(&curp);
1.64 kristaps 271: }
272:
273: while (*argv) {
1.66 kristaps 274: ffile(*argv, &curp);
1.103 schwarze 275: if (MANDOCLEVEL_OK != exit_status && curp.wstop)
1.64 kristaps 276: break;
277: ++argv;
1.1 kristaps 278: }
279:
1.22 kristaps 280: if (curp.outfree)
281: (*curp.outfree)(curp.outdata);
1.112 kristaps 282: if (curp.pmdoc)
283: mdoc_free(curp.pmdoc);
284: if (curp.pman)
285: man_free(curp.pman);
1.71 kristaps 286: if (curp.roff)
287: roff_free(curp.roff);
1.1 kristaps 288:
1.105 kristaps 289: return((int)exit_status);
1.1 kristaps 290: }
291:
292:
1.55 kristaps 293: static void
1.1 kristaps 294: version(void)
295: {
296:
1.54 kristaps 297: (void)printf("%s %s\n", progname, VERSION);
1.105 kristaps 298: exit((int)MANDOCLEVEL_OK);
1.1 kristaps 299: }
300:
301:
1.55 kristaps 302: static void
1.1 kristaps 303: usage(void)
304: {
305:
1.112 kristaps 306: (void)fprintf(stderr, "usage: %s "
307: "[-V] "
308: "[-foption] "
309: "[-mformat] "
310: "[-Ooption] "
311: "[-Toutput] "
312: "[-Werr] "
313: "[file...]\n",
314: progname);
315:
1.105 kristaps 316: exit((int)MANDOCLEVEL_BADARG);
1.19 kristaps 317: }
318:
1.64 kristaps 319: static void
1.66 kristaps 320: ffile(const char *file, struct curparse *curp)
1.1 kristaps 321: {
322:
1.112 kristaps 323: /*
324: * Called once per input file. Get the file ready for reading,
325: * pass it through to the parser-driver, then close it out.
326: * XXX: don't do anything special as this is only called for
327: * files; stdin goes directly to fdesc().
328: */
329:
1.19 kristaps 330: curp->file = file;
1.112 kristaps 331:
1.19 kristaps 332: if (-1 == (curp->fd = open(curp->file, O_RDONLY, 0))) {
1.53 kristaps 333: perror(curp->file);
1.103 schwarze 334: exit_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 335: return;
1.1 kristaps 336: }
337:
1.66 kristaps 338: fdesc(curp);
1.1 kristaps 339:
1.19 kristaps 340: if (-1 == close(curp->fd))
1.53 kristaps 341: perror(curp->file);
1.1 kristaps 342: }
343:
1.111 kristaps 344: static int
345: pfile(const char *file, struct curparse *curp)
346: {
347: const char *savefile;
348: int fd, savefd;
349:
350: if (-1 == (fd = open(file, O_RDONLY, 0))) {
351: perror(file);
1.115 schwarze 352: file_status = MANDOCLEVEL_SYSERR;
1.111 kristaps 353: return(0);
354: }
355:
356: savefile = curp->file;
357: savefd = curp->fd;
358:
359: curp->file = file;
360: curp->fd = fd;
361:
362: pdesc(curp);
363:
364: curp->file = savefile;
365: curp->fd = savefd;
366:
367: if (-1 == close(fd))
368: perror(file);
369:
1.115 schwarze 370: return(MANDOCLEVEL_FATAL > file_status ? 1 : 0);
1.111 kristaps 371: }
372:
1.1 kristaps 373:
1.103 schwarze 374: static void
1.68 joerg 375: resize_buf(struct buf *buf, size_t initial)
376: {
377:
1.103 schwarze 378: buf->sz = buf->sz ? 2 * buf->sz : initial;
379: buf->buf = realloc(buf->buf, buf->sz);
380: if (NULL == buf->buf) {
1.68 joerg 381: perror(NULL);
1.105 kristaps 382: exit((int)MANDOCLEVEL_SYSERR);
1.68 joerg 383: }
384: }
385:
386:
387: static int
1.66 kristaps 388: read_whole_file(struct curparse *curp, struct buf *fb, int *with_mmap)
389: {
390: struct stat st;
1.68 joerg 391: size_t off;
1.66 kristaps 392: ssize_t ssz;
393:
394: if (-1 == fstat(curp->fd, &st)) {
395: perror(curp->file);
396: return(0);
397: }
398:
399: /*
400: * If we're a regular file, try just reading in the whole entry
401: * via mmap(). This is faster than reading it into blocks, and
402: * since each file is only a few bytes to begin with, I'm not
403: * concerned that this is going to tank any machines.
404: */
405:
406: if (S_ISREG(st.st_mode)) {
407: if (st.st_size >= (1U << 31)) {
408: fprintf(stderr, "%s: input too large\n",
409: curp->file);
410: return(0);
411: }
412: *with_mmap = 1;
1.71 kristaps 413: fb->sz = (size_t)st.st_size;
1.66 kristaps 414: fb->buf = mmap(NULL, fb->sz, PROT_READ,
1.83 joerg 415: MAP_FILE|MAP_SHARED, curp->fd, 0);
1.66 kristaps 416: if (fb->buf != MAP_FAILED)
417: return(1);
418: }
419:
420: /*
421: * If this isn't a regular file (like, say, stdin), then we must
422: * go the old way and just read things in bit by bit.
423: */
424:
425: *with_mmap = 0;
426: off = 0;
427: fb->sz = 0;
428: fb->buf = NULL;
429: for (;;) {
430: if (off == fb->sz) {
431: if (fb->sz == (1U << 31)) {
432: fprintf(stderr, "%s: input too large\n",
433: curp->file);
434: break;
435: }
1.103 schwarze 436: resize_buf(fb, 65536);
1.66 kristaps 437: }
1.71 kristaps 438: ssz = read(curp->fd, fb->buf + (int)off, fb->sz - off);
1.66 kristaps 439: if (ssz == 0) {
440: fb->sz = off;
441: return(1);
442: }
443: if (ssz == -1) {
444: perror(curp->file);
445: break;
446: }
1.71 kristaps 447: off += (size_t)ssz;
1.66 kristaps 448: }
449:
450: free(fb->buf);
451: fb->buf = NULL;
452: return(0);
453: }
454:
455:
1.64 kristaps 456: static void
1.66 kristaps 457: fdesc(struct curparse *curp)
1.1 kristaps 458: {
1.112 kristaps 459:
460: /*
461: * Called once per file with an opened file descriptor. All
462: * pre-file-parse operations (whether stdin or a file) should go
463: * here.
464: *
465: * This calls down into the nested parser, which drills down and
466: * fully parses a file and all its dependences (i.e., `so'). It
467: * then runs the cleanup validators and pushes to output.
468: */
469:
470: /* Zero the parse type. */
471:
472: curp->mdoc = NULL;
473: curp->man = NULL;
1.115 schwarze 474: file_status = MANDOCLEVEL_OK;
1.112 kristaps 475:
476: /* Make sure the mandotory roff parser is initialised. */
477:
478: if (NULL == curp->roff) {
479: curp->roff = roff_alloc(&curp->regs, curp, mmsg);
480: assert(curp->roff);
481: }
482:
483: /* Fully parse the file. */
1.10 kristaps 484:
1.111 kristaps 485: pdesc(curp);
1.92 kristaps 486:
1.115 schwarze 487: if (MANDOCLEVEL_FATAL <= file_status)
1.111 kristaps 488: goto cleanup;
489:
490: /* NOTE a parser may not have been assigned, yet. */
491:
1.112 kristaps 492: if ( ! (curp->man || curp->mdoc)) {
1.111 kristaps 493: fprintf(stderr, "%s: Not a manual\n", curp->file);
1.115 schwarze 494: file_status = MANDOCLEVEL_FATAL;
1.111 kristaps 495: goto cleanup;
496: }
497:
498: /* Clean up the parse routine ASTs. */
499:
1.112 kristaps 500: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
1.115 schwarze 501: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 502: goto cleanup;
503: }
1.112 kristaps 504:
505: if (curp->man && ! man_endparse(curp->man)) {
1.115 schwarze 506: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 507: goto cleanup;
508: }
1.112 kristaps 509:
510: assert(curp->roff);
511: if ( ! roff_endparse(curp->roff)) {
1.115 schwarze 512: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 513: goto cleanup;
514: }
1.1 kristaps 515:
516: /*
1.111 kristaps 517: * With -Wstop and warnings or errors of at least
518: * the requested level, do not produce output.
1.1 kristaps 519: */
520:
1.115 schwarze 521: if (MANDOCLEVEL_OK != file_status && curp->wstop)
1.111 kristaps 522: goto cleanup;
523:
524: /* If unset, allocate output dev now (if applicable). */
525:
526: if ( ! (curp->outman && curp->outmdoc)) {
527: switch (curp->outtype) {
528: case (OUTT_XHTML):
529: curp->outdata = xhtml_alloc(curp->outopts);
530: break;
531: case (OUTT_HTML):
532: curp->outdata = html_alloc(curp->outopts);
533: break;
534: case (OUTT_ASCII):
535: curp->outdata = ascii_alloc(curp->outopts);
536: curp->outfree = ascii_free;
537: break;
538: case (OUTT_PDF):
539: curp->outdata = pdf_alloc(curp->outopts);
540: curp->outfree = pspdf_free;
541: break;
542: case (OUTT_PS):
543: curp->outdata = ps_alloc(curp->outopts);
544: curp->outfree = pspdf_free;
545: break;
546: default:
547: break;
548: }
549:
550: switch (curp->outtype) {
551: case (OUTT_HTML):
552: /* FALLTHROUGH */
553: case (OUTT_XHTML):
554: curp->outman = html_man;
555: curp->outmdoc = html_mdoc;
556: curp->outfree = html_free;
557: break;
558: case (OUTT_TREE):
559: curp->outman = tree_man;
560: curp->outmdoc = tree_mdoc;
561: break;
562: case (OUTT_PDF):
563: /* FALLTHROUGH */
564: case (OUTT_ASCII):
565: /* FALLTHROUGH */
566: case (OUTT_PS):
567: curp->outman = terminal_man;
568: curp->outmdoc = terminal_mdoc;
569: break;
570: default:
571: break;
572: }
573: }
574:
575: /* Execute the out device, if it exists. */
576:
1.112 kristaps 577: if (curp->man && curp->outman)
578: (*curp->outman)(curp->outdata, curp->man);
579: if (curp->mdoc && curp->outmdoc)
580: (*curp->outmdoc)(curp->outdata, curp->mdoc);
1.111 kristaps 581:
582: cleanup:
1.112 kristaps 583:
1.111 kristaps 584: memset(&curp->regs, 0, sizeof(struct regset));
1.112 kristaps 585:
586: /* Reset the current-parse compilers. */
587:
588: if (curp->mdoc)
589: mdoc_reset(curp->mdoc);
590: if (curp->man)
591: man_reset(curp->man);
592:
593: assert(curp->roff);
594: roff_reset(curp->roff);
1.111 kristaps 595:
1.115 schwarze 596: if (exit_status < file_status)
597: exit_status = file_status;
598:
1.111 kristaps 599: return;
600: }
601:
602: static void
603: pdesc(struct curparse *curp)
604: {
605: struct buf blk;
606: int with_mmap;
607:
1.112 kristaps 608: /*
609: * Run for each opened file; may be called more than once for
610: * each full parse sequence if the opened file is nested (i.e.,
611: * from `so'). Simply sucks in the whole file and moves into
612: * the parse phase for the file.
613: */
614:
1.103 schwarze 615: if ( ! read_whole_file(curp, &blk, &with_mmap)) {
1.115 schwarze 616: file_status = MANDOCLEVEL_SYSERR;
1.64 kristaps 617: return;
1.103 schwarze 618: }
1.1 kristaps 619:
1.112 kristaps 620: /* Line number is per-file. */
1.111 kristaps 621:
622: curp->line = 1;
1.112 kristaps 623:
1.111 kristaps 624: parsebuf(curp, blk, 1);
625:
626: if (with_mmap)
627: munmap(blk.buf, blk.sz);
628: else
629: free(blk.buf);
630: }
631:
632: static void
633: parsebuf(struct curparse *curp, struct buf blk, int start)
634: {
635: struct buf ln;
1.114 kristaps 636: enum rofferr rr;
1.112 kristaps 637: int i, of, rc;
638: int pos; /* byte number in the ln buffer */
639: int lnn; /* line number in the real file */
1.111 kristaps 640: unsigned char c;
1.112 kristaps 641:
642: /*
643: * Main parse routine for an opened file. This is called for
644: * each opened file and simply loops around the full input file,
645: * possibly nesting (i.e., with `so').
646: */
1.71 kristaps 647:
1.111 kristaps 648: memset(&ln, 0, sizeof(struct buf));
649:
1.112 kristaps 650: lnn = curp->line;
651: pos = 0;
1.111 kristaps 652:
1.112 kristaps 653: for (i = 0; i < (int)blk.sz; ) {
1.111 kristaps 654: if (0 == pos && '\0' == blk.buf[i])
655: break;
1.112 kristaps 656:
1.111 kristaps 657: if (start)
658: curp->line = lnn;
659:
660: while (i < (int)blk.sz && (start || '\0' != blk.buf[i])) {
1.70 joerg 661: if ('\n' == blk.buf[i]) {
662: ++i;
663: ++lnn;
664: break;
665: }
1.99 kristaps 666:
667: /*
668: * Warn about bogus characters. If you're using
669: * non-ASCII encoding, you're screwing your
670: * readers. Since I'd rather this not happen,
671: * I'll be helpful and drop these characters so
672: * we don't display gibberish. Note to manual
673: * writers: use special characters.
674: */
675:
1.102 schwarze 676: c = (unsigned char) blk.buf[i];
1.112 kristaps 677:
678: if ( ! (isascii(c) &&
679: (isgraph(c) || isblank(c)))) {
1.103 schwarze 680: mmsg(MANDOCERR_BADCHAR, curp,
1.111 kristaps 681: curp->line, pos, "ignoring byte");
1.99 kristaps 682: i++;
683: continue;
684: }
685:
1.112 kristaps 686: /* Trailing backslash = a plain char. */
687:
1.70 joerg 688: if ('\\' != blk.buf[i] || i + 1 == (int)blk.sz) {
689: if (pos >= (int)ln.sz)
1.103 schwarze 690: resize_buf(&ln, 256);
1.70 joerg 691: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 692: continue;
1.70 joerg 693: }
1.112 kristaps 694:
695: /* Found escape & at least one other char. */
696:
1.70 joerg 697: if ('\n' == blk.buf[i + 1]) {
1.112 kristaps 698: i += 2;
1.70 joerg 699: /* Escaped newlines are skipped over */
700: ++lnn;
1.67 joerg 701: continue;
702: }
1.112 kristaps 703:
1.70 joerg 704: if ('"' == blk.buf[i + 1]) {
705: i += 2;
706: /* Comment, skip to end of line */
707: for (; i < (int)blk.sz; ++i) {
708: if ('\n' == blk.buf[i]) {
709: ++i;
710: ++lnn;
711: break;
712: }
713: }
1.112 kristaps 714:
1.70 joerg 715: /* Backout trailing whitespaces */
716: for (; pos > 0; --pos) {
717: if (ln.buf[pos - 1] != ' ')
718: break;
719: if (pos > 2 && ln.buf[pos - 2] == '\\')
720: break;
721: }
722: break;
723: }
1.112 kristaps 724:
725: /* Some other escape sequence, copy & cont. */
726:
1.70 joerg 727: if (pos + 1 >= (int)ln.sz)
1.103 schwarze 728: resize_buf(&ln, 256);
1.1 kristaps 729:
1.70 joerg 730: ln.buf[pos++] = blk.buf[i++];
731: ln.buf[pos++] = blk.buf[i++];
1.67 joerg 732: }
1.1 kristaps 733:
1.70 joerg 734: if (pos >= (int)ln.sz)
1.103 schwarze 735: resize_buf(&ln, 256);
1.112 kristaps 736:
1.71 kristaps 737: ln.buf[pos] = '\0';
738:
1.76 kristaps 739: /*
740: * A significant amount of complexity is contained by
741: * the roff preprocessor. It's line-oriented but can be
742: * expressed on one line, so we need at times to
743: * readjust our starting point and re-run it. The roff
744: * preprocessor can also readjust the buffers with new
745: * data, so we pass them in wholesale.
746: */
747:
748: of = 0;
1.112 kristaps 749:
1.111 kristaps 750: rerun:
1.114 kristaps 751: rr = roff_parseln
1.112 kristaps 752: (curp->roff, curp->line,
753: &ln.buf, &ln.sz, of, &of);
754:
1.114 kristaps 755: switch (rr) {
1.111 kristaps 756: case (ROFF_REPARSE):
757: parsebuf(curp, ln, 0);
758: pos = 0;
759: continue;
760: case (ROFF_APPEND):
761: pos = strlen(ln.buf);
762: continue;
763: case (ROFF_RERUN):
764: goto rerun;
765: case (ROFF_IGN):
766: pos = 0;
1.71 kristaps 767: continue;
1.111 kristaps 768: case (ROFF_ERR):
1.115 schwarze 769: assert(MANDOCLEVEL_FATAL <= file_status);
1.111 kristaps 770: break;
771: case (ROFF_SO):
772: if (pfile(ln.buf + of, curp)) {
773: pos = 0;
774: continue;
775: } else
776: break;
777: case (ROFF_CONT):
778: break;
1.103 schwarze 779: }
1.29 kristaps 780:
1.76 kristaps 781: /*
782: * If input parsers have not been allocated, do so now.
783: * We keep these instanced betwen parsers, but set them
784: * locally per parse routine since we can use different
785: * parsers with each one.
786: */
1.19 kristaps 787:
1.112 kristaps 788: if ( ! (curp->man || curp->mdoc))
789: pset(ln.buf + of, pos - of, curp);
1.19 kristaps 790:
1.112 kristaps 791: /*
792: * Lastly, push down into the parsers themselves. One
793: * of these will have already been set in the pset()
794: * routine.
795: */
796:
797: if (curp->man || curp->mdoc) {
798: rc = curp->man ?
799: man_parseln(curp->man,
800: curp->line, ln.buf, of) :
801: mdoc_parseln(curp->mdoc,
802: curp->line, ln.buf, of);
1.30 kristaps 803:
1.112 kristaps 804: if ( ! rc) {
1.115 schwarze 805: assert(MANDOCLEVEL_FATAL <= file_status);
1.112 kristaps 806: break;
807: }
1.103 schwarze 808: }
1.19 kristaps 809:
1.111 kristaps 810: /* Temporary buffers typically are not full. */
1.112 kristaps 811:
1.111 kristaps 812: if (0 == start && '\0' == blk.buf[i])
1.59 kristaps 813: break;
1.85 kristaps 814:
1.111 kristaps 815: /* Start the next input line. */
1.112 kristaps 816:
1.111 kristaps 817: pos = 0;
1.22 kristaps 818: }
819:
1.111 kristaps 820: free(ln.buf);
1.19 kristaps 821: }
822:
1.103 schwarze 823: static void
1.112 kristaps 824: pset(const char *buf, int pos, struct curparse *curp)
1.19 kristaps 825: {
1.29 kristaps 826: int i;
1.19 kristaps 827:
828: /*
829: * Try to intuit which kind of manual parser should be used. If
830: * passed in by command-line (-man, -mdoc), then use that
831: * explicitly. If passed as -mandoc, then try to guess from the
1.30 kristaps 832: * line: either skip dot-lines, use -mdoc when finding `.Dt', or
1.19 kristaps 833: * default to -man, which is more lenient.
1.112 kristaps 834: *
835: * Separate out pmdoc/pman from mdoc/man: the first persists
836: * through all parsers, while the latter is used per-parse.
1.19 kristaps 837: */
838:
1.75 kristaps 839: if ('.' == buf[0] || '\'' == buf[0]) {
1.29 kristaps 840: for (i = 1; buf[i]; i++)
841: if (' ' != buf[i] && '\t' != buf[i])
842: break;
1.103 schwarze 843: if ('\0' == buf[i])
844: return;
1.29 kristaps 845: }
1.10 kristaps 846:
1.19 kristaps 847: switch (curp->inttype) {
848: case (INTT_MDOC):
1.112 kristaps 849: if (NULL == curp->pmdoc)
850: curp->pmdoc = mdoc_alloc
851: (&curp->regs, curp, mmsg);
852: assert(curp->pmdoc);
853: curp->mdoc = curp->pmdoc;
1.103 schwarze 854: return;
1.19 kristaps 855: case (INTT_MAN):
1.112 kristaps 856: if (NULL == curp->pman)
857: curp->pman = man_alloc
858: (&curp->regs, curp, mmsg);
859: assert(curp->pman);
860: curp->man = curp->pman;
1.103 schwarze 861: return;
1.19 kristaps 862: default:
863: break;
864: }
865:
866: if (pos >= 3 && 0 == memcmp(buf, ".Dd", 3)) {
1.112 kristaps 867: if (NULL == curp->pmdoc)
868: curp->pmdoc = mdoc_alloc
869: (&curp->regs, curp, mmsg);
870: assert(curp->pmdoc);
871: curp->mdoc = curp->pmdoc;
1.103 schwarze 872: return;
1.19 kristaps 873: }
874:
1.112 kristaps 875: if (NULL == curp->pman)
876: curp->pman = man_alloc(&curp->regs, curp, mmsg);
877: assert(curp->pman);
878: curp->man = curp->pman;
1.10 kristaps 879: }
880:
881: static int
882: moptions(enum intt *tflags, char *arg)
883: {
884:
1.17 kristaps 885: if (0 == strcmp(arg, "doc"))
1.10 kristaps 886: *tflags = INTT_MDOC;
1.19 kristaps 887: else if (0 == strcmp(arg, "andoc"))
888: *tflags = INTT_AUTO;
1.17 kristaps 889: else if (0 == strcmp(arg, "an"))
1.10 kristaps 890: *tflags = INTT_MAN;
891: else {
1.57 kristaps 892: fprintf(stderr, "%s: Bad argument\n", arg);
1.10 kristaps 893: return(0);
894: }
895:
896: return(1);
1.1 kristaps 897: }
898:
899: static int
1.60 kristaps 900: toptions(struct curparse *curp, char *arg)
1.1 kristaps 901: {
902:
903: if (0 == strcmp(arg, "ascii"))
1.60 kristaps 904: curp->outtype = OUTT_ASCII;
905: else if (0 == strcmp(arg, "lint")) {
906: curp->outtype = OUTT_LINT;
1.103 schwarze 907: curp->wlevel = MANDOCLEVEL_WARNING;
1.60 kristaps 908: }
1.1 kristaps 909: else if (0 == strcmp(arg, "tree"))
1.60 kristaps 910: curp->outtype = OUTT_TREE;
1.43 kristaps 911: else if (0 == strcmp(arg, "html"))
1.60 kristaps 912: curp->outtype = OUTT_HTML;
1.59 kristaps 913: else if (0 == strcmp(arg, "xhtml"))
1.60 kristaps 914: curp->outtype = OUTT_XHTML;
1.85 kristaps 915: else if (0 == strcmp(arg, "ps"))
916: curp->outtype = OUTT_PS;
1.100 kristaps 917: else if (0 == strcmp(arg, "pdf"))
918: curp->outtype = OUTT_PDF;
1.1 kristaps 919: else {
1.57 kristaps 920: fprintf(stderr, "%s: Bad argument\n", arg);
1.1 kristaps 921: return(0);
922: }
923:
924: return(1);
925: }
926:
927: static int
1.103 schwarze 928: woptions(struct curparse *curp, char *arg)
1.1 kristaps 929: {
1.34 kristaps 930: char *v, *o;
1.103 schwarze 931: const char *toks[6];
1.1 kristaps 932:
1.103 schwarze 933: toks[0] = "stop";
934: toks[1] = "all";
935: toks[2] = "warning";
936: toks[3] = "error";
937: toks[4] = "fatal";
938: toks[5] = NULL;
1.1 kristaps 939:
1.34 kristaps 940: while (*arg) {
941: o = arg;
1.45 kristaps 942: switch (getsubopt(&arg, UNCONST(toks), &v)) {
1.1 kristaps 943: case (0):
1.103 schwarze 944: curp->wstop = 1;
1.1 kristaps 945: break;
946: case (1):
1.103 schwarze 947: /* FALLTHROUGH */
1.1 kristaps 948: case (2):
1.103 schwarze 949: curp->wlevel = MANDOCLEVEL_WARNING;
1.1 kristaps 950: break;
1.20 kristaps 951: case (3):
1.103 schwarze 952: curp->wlevel = MANDOCLEVEL_ERROR;
1.24 kristaps 953: break;
954: case (4):
1.103 schwarze 955: curp->wlevel = MANDOCLEVEL_FATAL;
1.50 kristaps 956: break;
1.1 kristaps 957: default:
1.103 schwarze 958: fprintf(stderr, "-W%s: Bad argument\n", o);
1.1 kristaps 959: return(0);
960: }
1.34 kristaps 961: }
1.1 kristaps 962:
963: return(1);
964: }
965:
1.71 kristaps 966: static int
967: mmsg(enum mandocerr t, void *arg, int ln, int col, const char *msg)
968: {
969: struct curparse *cp;
1.103 schwarze 970: enum mandoclevel level;
971:
972: level = MANDOCLEVEL_FATAL;
973: while (t < mandoclimits[level])
1.105 kristaps 974: /* LINTED */
1.103 schwarze 975: level--;
1.71 kristaps 976:
977: cp = (struct curparse *)arg;
1.103 schwarze 978: if (level < cp->wlevel)
979: return(1);
1.73 kristaps 980:
1.103 schwarze 981: fprintf(stderr, "%s:%d:%d: %s: %s",
982: cp->file, ln, col + 1, mandoclevels[level], mandocerrs[t]);
1.73 kristaps 983: if (msg)
984: fprintf(stderr, ": %s", msg);
985: fputc('\n', stderr);
1.79 kristaps 986:
1.115 schwarze 987: if (file_status < level)
988: file_status = level;
1.103 schwarze 989:
990: return(level < MANDOCLEVEL_FATAL);
1.71 kristaps 991: }
CVSweb