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