Annotation of mandoc/read.c, Revision 1.95
1.95 ! schwarze 1: /* $Id: read.c,v 1.94 2014/10/28 17:36:19 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.40 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.42 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 kristaps 6: *
7: * Permission to use, copy, modify, and distribute this software for any
8: * purpose with or without fee is hereby granted, provided that the above
9: * copyright notice and this permission notice appear in all copies.
10: *
11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18: */
1.11 kristaps 19: #include "config.h"
20:
1.80 schwarze 21: #include <sys/types.h>
1.81 schwarze 22: #if HAVE_MMAP
1.82 schwarze 23: #include <sys/mman.h>
1.80 schwarze 24: #include <sys/stat.h>
1.15 kristaps 25: #endif
1.82 schwarze 26: #include <sys/wait.h>
1.1 kristaps 27:
28: #include <assert.h>
29: #include <ctype.h>
1.40 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.3 kristaps 32: #include <stdarg.h>
1.28 joerg 33: #include <stdint.h>
1.1 kristaps 34: #include <stdio.h>
35: #include <stdlib.h>
36: #include <string.h>
37: #include <unistd.h>
38:
39: #include "mandoc.h"
1.46 schwarze 40: #include "mandoc_aux.h"
1.3 kristaps 41: #include "libmandoc.h"
1.1 kristaps 42: #include "mdoc.h"
43: #include "man.h"
1.28 joerg 44: #include "main.h"
1.1 kristaps 45:
46: #define REPARSE_LIMIT 1000
47:
48: struct mparse {
49: struct man *pman; /* persistent man parser */
50: struct mdoc *pmdoc; /* persistent mdoc parser */
51: struct man *man; /* man parser */
52: struct mdoc *mdoc; /* mdoc parser */
53: struct roff *roff; /* roff parser (!NULL) */
1.94 schwarze 54: const struct mchars *mchars; /* character table */
1.45 schwarze 55: char *sodest; /* filename pointed to by .so */
1.83 schwarze 56: const char *file; /* filename of current input file */
57: struct buf *primary; /* buffer currently being parsed */
58: struct buf *secondary; /* preprocessed copy of input */
59: const char *defos; /* default operating system */
60: mandocmsg mmsg; /* warning/error message handler */
61: enum mandoclevel file_status; /* status of current parse */
62: enum mandoclevel wlevel; /* ignore messages below this */
63: int options; /* parser options */
1.93 schwarze 64: int filenc; /* encoding of the current file */
1.1 kristaps 65: int reparse_count; /* finite interp. stack */
1.83 schwarze 66: int line; /* line number in the file */
1.1 kristaps 67: };
68:
1.84 schwarze 69: static void choose_parser(struct mparse *);
1.1 kristaps 70: static void resize_buf(struct buf *, size_t);
1.95 ! schwarze 71: static void mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 72: static int read_whole_file(struct mparse *, const char *, int,
73: struct buf *, int *);
1.1 kristaps 74: static void mparse_end(struct mparse *);
1.37 schwarze 75: static void mparse_parse_buffer(struct mparse *, struct buf,
76: const char *);
1.1 kristaps 77:
1.3 kristaps 78: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
79: MANDOCERR_OK,
80: MANDOCERR_WARNING,
81: MANDOCERR_WARNING,
82: MANDOCERR_ERROR,
83: MANDOCERR_FATAL,
84: MANDOCERR_MAX,
85: MANDOCERR_MAX
86: };
87:
1.7 kristaps 88: static const char * const mandocerrs[MANDOCERR_MAX] = {
89: "ok",
90:
91: "generic warning",
92:
93: /* related to the prologue */
1.79 schwarze 94: "missing manual title, using UNTITLED",
95: "missing manual title, using \"\"",
1.54 schwarze 96: "lower case character in document title",
1.79 schwarze 97: "missing manual section, using \"\"",
1.7 kristaps 98: "unknown manual section",
1.32 schwarze 99: "unknown manual volume or arch",
1.54 schwarze 100: "missing date, using today's date",
1.7 kristaps 101: "cannot parse date, using it verbatim",
1.79 schwarze 102: "missing Os macro, using \"\"",
103: "duplicate prologue macro",
104: "late prologue macro",
105: "skipping late title macro",
1.7 kristaps 106: "prologue macros out of order",
107:
108: /* related to document structure */
109: ".so is fragile, better use ln(1)",
1.50 schwarze 110: "no document body",
1.54 schwarze 111: "content before first section header",
112: "first section is not \"NAME\"",
1.7 kristaps 113: "bad NAME section contents",
114: "sections out of conventional order",
1.54 schwarze 115: "duplicate section title",
116: "unexpected section",
1.87 schwarze 117: "unusual Xr order",
118: "unusual Xr punctuation",
1.86 schwarze 119: "AUTHORS section without An macro",
1.7 kristaps 120:
121: /* related to macros and nesting */
1.55 schwarze 122: "obsolete macro",
1.7 kristaps 123: "skipping paragraph macro",
1.31 schwarze 124: "moving paragraph macro out of list",
1.7 kristaps 125: "skipping no-space macro",
126: "blocks badly nested",
127: "nested displays are not portable",
1.57 schwarze 128: "moving content out of list",
129: ".Vt block has child macro",
1.78 schwarze 130: "fill mode already enabled, skipping",
131: "fill mode already disabled, skipping",
1.7 kristaps 132: "line scope broken",
133:
134: /* related to missing macro arguments */
1.58 schwarze 135: "skipping empty request",
136: "conditional request controls empty scope",
1.7 kristaps 137: "skipping empty macro",
1.62 schwarze 138: "empty argument, using 0n",
1.7 kristaps 139: "argument count wrong",
1.60 schwarze 140: "missing display type, using -ragged",
141: "list type is not the first argument",
142: "missing -width in -tag list, using 8n",
1.78 schwarze 143: "missing utility name, using \"\"",
1.60 schwarze 144: "empty head in list item",
145: "empty list item",
1.61 schwarze 146: "missing font type, using \\fR",
147: "unknown font type, using \\fR",
1.60 schwarze 148: "missing -std argument, adding it",
1.90 schwarze 149: "missing eqn box, using \"\"",
1.7 kristaps 150:
151: /* related to bad macro arguments */
1.64 schwarze 152: "unterminated quoted argument",
1.7 kristaps 153: "duplicate argument",
1.76 schwarze 154: "skipping duplicate argument",
1.63 schwarze 155: "skipping duplicate display type",
156: "skipping duplicate list type",
1.76 schwarze 157: "skipping -width argument",
1.7 kristaps 158: "unknown AT&T UNIX version",
1.88 schwarze 159: "comma in function argument",
1.89 schwarze 160: "parenthesis in function name",
1.67 schwarze 161: "invalid content in Rs block",
1.63 schwarze 162: "invalid Boolean argument",
163: "unknown font, skipping request",
1.7 kristaps 164:
165: /* related to plain text */
1.64 schwarze 166: "blank line in fill mode, using .sp",
167: "tab in filled text",
168: "whitespace at end of input line",
1.7 kristaps 169: "bad comment style",
1.64 schwarze 170: "invalid escape sequence",
171: "undefined string, using \"\"",
1.16 kristaps 172:
1.7 kristaps 173: "generic error",
1.17 kristaps 174:
175: /* related to equations */
1.20 kristaps 176: "unexpected equation scope closure",
177: "equation scope open on exit",
1.21 kristaps 178: "overlapping equation scopes",
179: "unexpected end of equation",
1.7 kristaps 180:
181: /* related to tables */
182: "bad table syntax",
183: "bad table option",
184: "bad table layout",
185: "no table layout cells specified",
186: "no table data cells specified",
187: "ignore data in cell",
188: "data block still open",
189: "ignoring extra data cells",
190:
1.68 schwarze 191: /* related to document structure and macros */
1.7 kristaps 192: "input stack limit exceeded, infinite loop?",
193: "skipping bad character",
1.68 schwarze 194: "skipping unknown macro",
1.73 schwarze 195: "skipping item outside list",
1.68 schwarze 196: "skipping column outside column list",
197: "skipping end of block that is not open",
198: "inserting missing end of block",
199: "appending missing end of block",
200:
201: /* related to request and macro arguments */
1.7 kristaps 202: "escaped character not allowed in a name",
203: "argument count wrong",
1.71 schwarze 204: "missing list type, using -item",
1.70 schwarze 205: "missing manual name, using \"\"",
1.71 schwarze 206: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 207: "unknown standard specifier",
1.71 schwarze 208: "skipping request without numeric argument",
1.61 schwarze 209: "skipping all arguments",
210: "skipping excess arguments",
1.92 kristaps 211: "divide by zero",
1.7 kristaps 212:
213: "generic fatal error",
214:
1.40 schwarze 215: "input too large",
1.78 schwarze 216: "NOT IMPLEMENTED: Bd -file",
1.7 kristaps 217: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.52 schwarze 218: ".so request failed",
1.40 schwarze 219:
220: /* system errors */
1.82 schwarze 221: "cannot dup file descriptor",
222: "cannot exec",
223: "gunzip failed with code",
224: "cannot fork",
1.51 schwarze 225: NULL,
1.82 schwarze 226: "cannot open pipe",
227: "cannot read file",
228: "gunzip died from signal",
1.40 schwarze 229: "cannot stat file",
1.82 schwarze 230: "wait failed",
1.7 kristaps 231: };
232:
233: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
234: "SUCCESS",
235: "RESERVED",
236: "WARNING",
237: "ERROR",
238: "FATAL",
239: "BADARG",
240: "SYSERR"
241: };
242:
1.47 schwarze 243:
1.1 kristaps 244: static void
245: resize_buf(struct buf *buf, size_t initial)
246: {
247:
248: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
249: buf->buf = mandoc_realloc(buf->buf, buf->sz);
250: }
251:
252: static void
1.84 schwarze 253: choose_parser(struct mparse *curp)
1.1 kristaps 254: {
1.83 schwarze 255: char *cp, *ep;
256: int format;
1.1 kristaps 257:
1.83 schwarze 258: /*
259: * If neither command line arguments -mdoc or -man select
260: * a parser nor the roff parser found a .Dd or .TH macro
261: * yet, look ahead in the main input buffer.
262: */
263:
264: if ((format = roff_getformat(curp->roff)) == 0) {
265: cp = curp->primary->buf;
266: ep = cp + curp->primary->sz;
267: while (cp < ep) {
1.85 schwarze 268: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 269: cp++;
270: if (cp[0] == 'D' && cp[1] == 'd') {
271: format = MPARSE_MDOC;
272: break;
273: }
274: if (cp[0] == 'T' && cp[1] == 'H') {
275: format = MPARSE_MAN;
276: break;
277: }
278: }
279: cp = memchr(cp, '\n', ep - cp);
280: if (cp == NULL)
281: break;
282: cp++;
283: }
1.1 kristaps 284: }
285:
1.83 schwarze 286: if (format == MPARSE_MDOC) {
1.47 schwarze 287: if (NULL == curp->pmdoc)
1.44 schwarze 288: curp->pmdoc = mdoc_alloc(
289: curp->roff, curp, curp->defos,
290: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 291: assert(curp->pmdoc);
292: curp->mdoc = curp->pmdoc;
293: return;
1.47 schwarze 294: }
1.1 kristaps 295:
1.83 schwarze 296: /* Fall back to man(7) as a last resort. */
297:
1.47 schwarze 298: if (NULL == curp->pman)
1.44 schwarze 299: curp->pman = man_alloc(curp->roff, curp,
300: MPARSE_QUICK & curp->options ? 1 : 0);
1.1 kristaps 301: assert(curp->pman);
302: curp->man = curp->pman;
303: }
304:
305: /*
1.95 ! schwarze 306: * Main parse routine for a buffer.
! 307: * It assumes encoding and line numbering are already set up.
! 308: * It can recurse directly (for invocations of user-defined
! 309: * macros, inline equations, and input line traps)
! 310: * and indirectly (for .so file inclusion).
1.1 kristaps 311: */
312: static void
1.95 ! schwarze 313: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 314: {
315: const struct tbl_span *span;
316: struct buf ln;
1.95 ! schwarze 317: size_t pos; /* byte number in the ln buffer */
1.1 kristaps 318: enum rofferr rr;
1.95 ! schwarze 319: int of, rc;
1.1 kristaps 320: int lnn; /* line number in the real file */
321: unsigned char c;
322:
1.95 ! schwarze 323: memset(&ln, 0, sizeof(ln));
1.1 kristaps 324:
1.47 schwarze 325: lnn = curp->line;
326: pos = 0;
1.1 kristaps 327:
1.95 ! schwarze 328: while (i < blk.sz) {
1.1 kristaps 329: if (0 == pos && '\0' == blk.buf[i])
330: break;
331:
332: if (start) {
333: curp->line = lnn;
334: curp->reparse_count = 0;
1.93 schwarze 335:
336: if (lnn < 3 &&
337: curp->filenc & MPARSE_UTF8 &&
1.95 ! schwarze 338: curp->filenc & MPARSE_LATIN1)
! 339: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 340: }
341:
1.95 ! schwarze 342: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 343:
344: /*
345: * When finding an unescaped newline character,
346: * leave the character loop to process the line.
347: * Skip a preceding carriage return, if any.
348: */
349:
1.95 ! schwarze 350: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 351: '\n' == blk.buf[i + 1])
352: ++i;
353: if ('\n' == blk.buf[i]) {
354: ++i;
355: ++lnn;
356: break;
357: }
358:
1.35 schwarze 359: /*
1.93 schwarze 360: * Make sure we have space for the worst
361: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 362: */
363:
1.95 ! schwarze 364: if (pos + 11 > ln.sz)
1.35 schwarze 365: resize_buf(&ln, 256);
366:
1.47 schwarze 367: /*
1.93 schwarze 368: * Encode 8-bit input.
1.1 kristaps 369: */
370:
1.93 schwarze 371: c = blk.buf[i];
372: if (c & 0x80) {
1.95 ! schwarze 373: if ( ! (curp->filenc && preconv_encode(
! 374: &blk, &i, &ln, &pos, &curp->filenc))) {
1.93 schwarze 375: mandoc_vmsg(MANDOCERR_BADCHAR,
376: curp, curp->line, pos,
377: "0x%x", c);
378: ln.buf[pos++] = '?';
379: i++;
380: }
381: continue;
382: }
383:
384: /*
385: * Exclude control characters.
386: */
1.1 kristaps 387:
1.93 schwarze 388: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.78 schwarze 389: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
390: curp->line, pos, "0x%x", c);
1.1 kristaps 391: i++;
1.27 joerg 392: ln.buf[pos++] = '?';
1.1 kristaps 393: continue;
394: }
395:
396: /* Trailing backslash = a plain char. */
397:
1.95 ! schwarze 398: if (blk.buf[i] != '\\' || i + 1 == blk.sz) {
1.1 kristaps 399: ln.buf[pos++] = blk.buf[i++];
400: continue;
401: }
402:
403: /*
404: * Found escape and at least one other character.
405: * When it's a newline character, skip it.
406: * When there is a carriage return in between,
407: * skip that one as well.
408: */
409:
1.95 ! schwarze 410: if ('\r' == blk.buf[i + 1] && i + 2 < blk.sz &&
1.1 kristaps 411: '\n' == blk.buf[i + 2])
412: ++i;
413: if ('\n' == blk.buf[i + 1]) {
414: i += 2;
415: ++lnn;
416: continue;
417: }
418:
1.13 kristaps 419: if ('"' == blk.buf[i + 1] || '#' == blk.buf[i + 1]) {
1.1 kristaps 420: i += 2;
421: /* Comment, skip to end of line */
1.95 ! schwarze 422: for (; i < blk.sz; ++i) {
1.1 kristaps 423: if ('\n' == blk.buf[i]) {
424: ++i;
425: ++lnn;
426: break;
427: }
428: }
429:
430: /* Backout trailing whitespaces */
431: for (; pos > 0; --pos) {
432: if (ln.buf[pos - 1] != ' ')
433: break;
434: if (pos > 2 && ln.buf[pos - 2] == '\\')
435: break;
436: }
437: break;
438: }
439:
1.35 schwarze 440: /* Catch escaped bogus characters. */
441:
442: c = (unsigned char) blk.buf[i+1];
443:
1.47 schwarze 444: if ( ! (isascii(c) &&
445: (isgraph(c) || isblank(c)))) {
1.78 schwarze 446: mandoc_vmsg(MANDOCERR_BADCHAR, curp,
447: curp->line, pos, "0x%x", c);
1.35 schwarze 448: i += 2;
449: ln.buf[pos++] = '?';
450: continue;
451: }
452:
1.1 kristaps 453: /* Some other escape sequence, copy & cont. */
454:
455: ln.buf[pos++] = blk.buf[i++];
456: ln.buf[pos++] = blk.buf[i++];
457: }
458:
1.95 ! schwarze 459: if (pos >= ln.sz)
1.1 kristaps 460: resize_buf(&ln, 256);
461:
462: ln.buf[pos] = '\0';
463:
464: /*
465: * A significant amount of complexity is contained by
466: * the roff preprocessor. It's line-oriented but can be
467: * expressed on one line, so we need at times to
468: * readjust our starting point and re-run it. The roff
469: * preprocessor can also readjust the buffers with new
470: * data, so we pass them in wholesale.
471: */
472:
473: of = 0;
474:
1.24 kristaps 475: /*
476: * Maintain a lookaside buffer of all parsed lines. We
477: * only do this if mparse_keep() has been invoked (the
478: * buffer may be accessed with mparse_getkeep()).
479: */
480:
481: if (curp->secondary) {
1.47 schwarze 482: curp->secondary->buf = mandoc_realloc(
483: curp->secondary->buf,
484: curp->secondary->sz + pos + 2);
485: memcpy(curp->secondary->buf +
486: curp->secondary->sz,
487: ln.buf, pos);
1.24 kristaps 488: curp->secondary->sz += pos;
489: curp->secondary->buf
490: [curp->secondary->sz] = '\n';
491: curp->secondary->sz++;
492: curp->secondary->buf
493: [curp->secondary->sz] = '\0';
494: }
1.1 kristaps 495: rerun:
1.47 schwarze 496: rr = roff_parseln(curp->roff, curp->line,
497: &ln.buf, &ln.sz, of, &of);
1.1 kristaps 498:
499: switch (rr) {
1.47 schwarze 500: case ROFF_REPARSE:
1.1 kristaps 501: if (REPARSE_LIMIT >= ++curp->reparse_count)
1.95 ! schwarze 502: mparse_buf_r(curp, ln, of, 0);
1.1 kristaps 503: else
1.3 kristaps 504: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 505: curp->line, pos, NULL);
1.1 kristaps 506: pos = 0;
507: continue;
1.47 schwarze 508: case ROFF_APPEND:
1.95 ! schwarze 509: pos = strlen(ln.buf);
1.1 kristaps 510: continue;
1.47 schwarze 511: case ROFF_RERUN:
1.1 kristaps 512: goto rerun;
1.47 schwarze 513: case ROFF_IGN:
1.1 kristaps 514: pos = 0;
515: continue;
1.47 schwarze 516: case ROFF_ERR:
1.1 kristaps 517: assert(MANDOCLEVEL_FATAL <= curp->file_status);
518: break;
1.47 schwarze 519: case ROFF_SO:
1.95 ! schwarze 520: if ( ! (curp->options & MPARSE_SO) &&
! 521: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 522: curp->sodest = mandoc_strdup(ln.buf + of);
523: free(ln.buf);
524: return;
525: }
1.24 kristaps 526: /*
527: * We remove `so' clauses from our lookaside
528: * buffer because we're going to descend into
529: * the file recursively.
530: */
1.47 schwarze 531: if (curp->secondary)
1.25 kristaps 532: curp->secondary->sz -= pos + 1;
1.36 schwarze 533: mparse_readfd(curp, -1, ln.buf + of);
1.52 schwarze 534: if (MANDOCLEVEL_FATAL <= curp->file_status) {
535: mandoc_vmsg(MANDOCERR_SO_FAIL,
536: curp, curp->line, pos,
537: ".so %s", ln.buf + of);
1.1 kristaps 538: break;
1.52 schwarze 539: }
1.1 kristaps 540: pos = 0;
541: continue;
542: default:
543: break;
544: }
545:
546: /*
547: * If we encounter errors in the recursive parse, make
548: * sure we don't continue parsing.
549: */
550:
551: if (MANDOCLEVEL_FATAL <= curp->file_status)
552: break;
553:
554: /*
555: * If input parsers have not been allocated, do so now.
1.14 kristaps 556: * We keep these instanced between parsers, but set them
1.1 kristaps 557: * locally per parse routine since we can use different
558: * parsers with each one.
559: */
560:
561: if ( ! (curp->man || curp->mdoc))
1.84 schwarze 562: choose_parser(curp);
1.1 kristaps 563:
1.47 schwarze 564: /*
1.84 schwarze 565: * Lastly, push down into the parsers themselves.
1.1 kristaps 566: * If libroff returns ROFF_TBL, then add it to the
567: * currently open parse. Since we only get here if
568: * there does exist data (see tbl_data.c), we're
569: * guaranteed that something's been allocated.
570: * Do the same for ROFF_EQN.
571: */
572:
573: rc = -1;
574:
575: if (ROFF_TBL == rr)
576: while (NULL != (span = roff_span(curp->roff))) {
577: rc = curp->man ?
1.47 schwarze 578: man_addspan(curp->man, span) :
579: mdoc_addspan(curp->mdoc, span);
1.1 kristaps 580: if (0 == rc)
581: break;
582: }
583: else if (ROFF_EQN == rr)
1.47 schwarze 584: rc = curp->mdoc ?
585: mdoc_addeqn(curp->mdoc,
586: roff_eqn(curp->roff)) :
587: man_addeqn(curp->man,
588: roff_eqn(curp->roff));
1.1 kristaps 589: else if (curp->man || curp->mdoc)
590: rc = curp->man ?
1.47 schwarze 591: man_parseln(curp->man,
592: curp->line, ln.buf, of) :
593: mdoc_parseln(curp->mdoc,
594: curp->line, ln.buf, of);
1.1 kristaps 595:
596: if (0 == rc) {
597: assert(MANDOCLEVEL_FATAL <= curp->file_status);
598: break;
1.41 schwarze 599: } else if (2 == rc)
600: break;
1.1 kristaps 601:
602: /* Temporary buffers typically are not full. */
603:
604: if (0 == start && '\0' == blk.buf[i])
605: break;
606:
607: /* Start the next input line. */
608:
609: pos = 0;
610: }
611:
612: free(ln.buf);
613: }
614:
615: static int
1.40 schwarze 616: read_whole_file(struct mparse *curp, const char *file, int fd,
617: struct buf *fb, int *with_mmap)
1.1 kristaps 618: {
619: size_t off;
620: ssize_t ssz;
621:
1.81 schwarze 622: #if HAVE_MMAP
1.15 kristaps 623: struct stat st;
1.1 kristaps 624: if (-1 == fstat(fd, &st)) {
1.40 schwarze 625: curp->file_status = MANDOCLEVEL_SYSERR;
626: if (curp->mmsg)
627: (*curp->mmsg)(MANDOCERR_SYSSTAT, curp->file_status,
628: file, 0, 0, strerror(errno));
1.1 kristaps 629: return(0);
630: }
631:
632: /*
633: * If we're a regular file, try just reading in the whole entry
634: * via mmap(). This is faster than reading it into blocks, and
635: * since each file is only a few bytes to begin with, I'm not
636: * concerned that this is going to tank any machines.
637: */
638:
639: if (S_ISREG(st.st_mode)) {
640: if (st.st_size >= (1U << 31)) {
1.40 schwarze 641: curp->file_status = MANDOCLEVEL_FATAL;
642: if (curp->mmsg)
643: (*curp->mmsg)(MANDOCERR_TOOLARGE,
644: curp->file_status, file, 0, 0, NULL);
1.1 kristaps 645: return(0);
646: }
647: *with_mmap = 1;
648: fb->sz = (size_t)st.st_size;
1.37 schwarze 649: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 650: if (fb->buf != MAP_FAILED)
651: return(1);
652: }
1.15 kristaps 653: #endif
1.1 kristaps 654:
655: /*
656: * If this isn't a regular file (like, say, stdin), then we must
657: * go the old way and just read things in bit by bit.
658: */
659:
660: *with_mmap = 0;
661: off = 0;
662: fb->sz = 0;
663: fb->buf = NULL;
664: for (;;) {
665: if (off == fb->sz) {
666: if (fb->sz == (1U << 31)) {
1.40 schwarze 667: curp->file_status = MANDOCLEVEL_FATAL;
668: if (curp->mmsg)
669: (*curp->mmsg)(MANDOCERR_TOOLARGE,
670: curp->file_status,
671: file, 0, 0, NULL);
1.1 kristaps 672: break;
673: }
674: resize_buf(fb, 65536);
675: }
676: ssz = read(fd, fb->buf + (int)off, fb->sz - off);
677: if (ssz == 0) {
678: fb->sz = off;
679: return(1);
680: }
681: if (ssz == -1) {
1.40 schwarze 682: curp->file_status = MANDOCLEVEL_SYSERR;
683: if (curp->mmsg)
684: (*curp->mmsg)(MANDOCERR_SYSREAD,
685: curp->file_status, file, 0, 0,
686: strerror(errno));
1.1 kristaps 687: break;
688: }
689: off += (size_t)ssz;
690: }
691:
692: free(fb->buf);
693: fb->buf = NULL;
694: return(0);
695: }
696:
697: static void
698: mparse_end(struct mparse *curp)
699: {
700:
701: if (MANDOCLEVEL_FATAL <= curp->file_status)
702: return;
703:
1.72 schwarze 704: if (curp->mdoc == NULL &&
705: curp->man == NULL &&
706: curp->sodest == NULL) {
707: if (curp->options & MPARSE_MDOC)
708: curp->mdoc = curp->pmdoc;
709: else {
710: if (curp->pman == NULL)
711: curp->pman = man_alloc(curp->roff, curp,
712: curp->options & MPARSE_QUICK ? 1 : 0);
713: curp->man = curp->pman;
714: }
715: }
716:
1.1 kristaps 717: if (curp->mdoc && ! mdoc_endparse(curp->mdoc)) {
718: assert(MANDOCLEVEL_FATAL <= curp->file_status);
719: return;
720: }
721:
722: if (curp->man && ! man_endparse(curp->man)) {
723: assert(MANDOCLEVEL_FATAL <= curp->file_status);
724: return;
725: }
726:
727: roff_endparse(curp->roff);
728: }
729:
730: static void
1.36 schwarze 731: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 732: {
1.85 schwarze 733: struct buf *svprimary;
1.28 joerg 734: const char *svfile;
1.95 ! schwarze 735: size_t offset;
1.36 schwarze 736: static int recursion_depth;
737:
738: if (64 < recursion_depth) {
739: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
740: return;
741: }
1.28 joerg 742:
743: /* Line number is per-file. */
744: svfile = curp->file;
745: curp->file = file;
1.85 schwarze 746: svprimary = curp->primary;
1.83 schwarze 747: curp->primary = &blk;
1.28 joerg 748: curp->line = 1;
1.36 schwarze 749: recursion_depth++;
1.28 joerg 750:
1.93 schwarze 751: /* Skip an UTF-8 byte order mark. */
752: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
753: (unsigned char)blk.buf[0] == 0xef &&
754: (unsigned char)blk.buf[1] == 0xbb &&
755: (unsigned char)blk.buf[2] == 0xbf) {
1.95 ! schwarze 756: offset = 3;
1.93 schwarze 757: curp->filenc &= ~MPARSE_LATIN1;
1.95 ! schwarze 758: } else
! 759: offset = 0;
1.93 schwarze 760:
1.95 ! schwarze 761: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 762:
1.36 schwarze 763: if (0 == --recursion_depth && MANDOCLEVEL_FATAL > curp->file_status)
1.28 joerg 764: mparse_end(curp);
765:
1.85 schwarze 766: curp->primary = svprimary;
1.28 joerg 767: curp->file = svfile;
768: }
769:
770: enum mandoclevel
771: mparse_readmem(struct mparse *curp, const void *buf, size_t len,
772: const char *file)
773: {
774: struct buf blk;
775:
776: blk.buf = UNCONST(buf);
777: blk.sz = len;
778:
1.36 schwarze 779: mparse_parse_buffer(curp, blk, file);
1.28 joerg 780: return(curp->file_status);
781: }
782:
1.36 schwarze 783: enum mandoclevel
784: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 785: {
1.28 joerg 786: struct buf blk;
787: int with_mmap;
1.93 schwarze 788: int save_filenc;
1.1 kristaps 789:
1.40 schwarze 790: if (-1 == fd && -1 == (fd = open(file, O_RDONLY, 0))) {
791: curp->file_status = MANDOCLEVEL_SYSERR;
792: if (curp->mmsg)
793: (*curp->mmsg)(MANDOCERR_SYSOPEN,
794: curp->file_status,
795: file, 0, 0, strerror(errno));
1.91 schwarze 796: return(curp->file_status);
1.40 schwarze 797: }
798:
1.28 joerg 799: /*
800: * Run for each opened file; may be called more than once for
801: * each full parse sequence if the opened file is nested (i.e.,
802: * from `so'). Simply sucks in the whole file and moves into
803: * the parse phase for the file.
804: */
1.1 kristaps 805:
1.91 schwarze 806: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 807: save_filenc = curp->filenc;
808: curp->filenc = curp->options &
809: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 810: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 811: curp->filenc = save_filenc;
1.81 schwarze 812: #if HAVE_MMAP
1.91 schwarze 813: if (with_mmap)
814: munmap(blk.buf, blk.sz);
815: else
1.28 joerg 816: #endif
1.91 schwarze 817: free(blk.buf);
818: }
1.1 kristaps 819:
820: if (STDIN_FILENO != fd && -1 == close(fd))
821: perror(file);
1.91 schwarze 822:
1.1 kristaps 823: return(curp->file_status);
1.82 schwarze 824: }
825:
826: enum mandoclevel
827: mparse_open(struct mparse *curp, int *fd, const char *file,
828: pid_t *child_pid)
829: {
830: int pfd[2];
831: char *cp;
832: enum mandocerr err;
833:
834: pfd[1] = -1;
835: curp->file = file;
836: if ((cp = strrchr(file, '.')) == NULL ||
837: strcmp(cp + 1, "gz")) {
838: *child_pid = 0;
839: if ((*fd = open(file, O_RDONLY)) == -1) {
840: err = MANDOCERR_SYSOPEN;
841: goto out;
842: }
843: return(MANDOCLEVEL_OK);
844: }
845:
846: if (pipe(pfd) == -1) {
847: err = MANDOCERR_SYSPIPE;
848: goto out;
849: }
850:
851: switch (*child_pid = fork()) {
852: case -1:
853: err = MANDOCERR_SYSFORK;
854: close(pfd[0]);
855: close(pfd[1]);
856: pfd[1] = -1;
857: break;
858: case 0:
859: close(pfd[0]);
860: if (dup2(pfd[1], STDOUT_FILENO) == -1) {
861: err = MANDOCERR_SYSDUP;
862: break;
863: }
864: execlp("gunzip", "gunzip", "-c", file, NULL);
865: err = MANDOCERR_SYSEXEC;
866: break;
867: default:
868: close(pfd[1]);
869: *fd = pfd[0];
870: return(MANDOCLEVEL_OK);
871: }
872:
873: out:
874: *fd = -1;
875: *child_pid = 0;
876: curp->file_status = MANDOCLEVEL_SYSERR;
877: if (curp->mmsg)
878: (*curp->mmsg)(err, curp->file_status, file,
879: 0, 0, strerror(errno));
880: if (pfd[1] != -1)
881: exit(1);
882: return(curp->file_status);
883: }
884:
885: enum mandoclevel
886: mparse_wait(struct mparse *curp, pid_t child_pid)
887: {
888: int status;
889:
890: if (waitpid(child_pid, &status, 0) == -1) {
891: mandoc_msg(MANDOCERR_SYSWAIT, curp, 0, 0,
892: strerror(errno));
893: curp->file_status = MANDOCLEVEL_SYSERR;
894: return(curp->file_status);
895: }
896: if (WIFSIGNALED(status)) {
897: mandoc_vmsg(MANDOCERR_SYSSIG, curp, 0, 0,
898: "%d", WTERMSIG(status));
899: curp->file_status = MANDOCLEVEL_SYSERR;
900: return(curp->file_status);
901: }
902: if (WEXITSTATUS(status)) {
903: mandoc_vmsg(MANDOCERR_SYSEXIT, curp, 0, 0,
904: "%d", WEXITSTATUS(status));
905: curp->file_status = MANDOCLEVEL_SYSERR;
906: return(curp->file_status);
907: }
908: return(MANDOCLEVEL_OK);
1.1 kristaps 909: }
910:
911: struct mparse *
1.94 schwarze 912: mparse_alloc(int options, enum mandoclevel wlevel, mandocmsg mmsg,
913: const struct mchars *mchars, const char *defos)
1.1 kristaps 914: {
915: struct mparse *curp;
1.10 kristaps 916:
917: assert(wlevel <= MANDOCLEVEL_FATAL);
1.1 kristaps 918:
919: curp = mandoc_calloc(1, sizeof(struct mparse));
920:
1.44 schwarze 921: curp->options = options;
1.3 kristaps 922: curp->wlevel = wlevel;
1.1 kristaps 923: curp->mmsg = mmsg;
1.29 schwarze 924: curp->defos = defos;
1.1 kristaps 925:
1.94 schwarze 926: curp->mchars = mchars;
927: curp->roff = roff_alloc(curp, curp->mchars, options);
1.72 schwarze 928: if (curp->options & MPARSE_MDOC)
929: curp->pmdoc = mdoc_alloc(
930: curp->roff, curp, curp->defos,
931: curp->options & MPARSE_QUICK ? 1 : 0);
932: if (curp->options & MPARSE_MAN)
933: curp->pman = man_alloc(curp->roff, curp,
934: curp->options & MPARSE_QUICK ? 1 : 0);
935:
1.1 kristaps 936: return(curp);
937: }
938:
939: void
940: mparse_reset(struct mparse *curp)
941: {
942:
943: roff_reset(curp->roff);
944:
945: if (curp->mdoc)
946: mdoc_reset(curp->mdoc);
947: if (curp->man)
948: man_reset(curp->man);
1.24 kristaps 949: if (curp->secondary)
950: curp->secondary->sz = 0;
1.1 kristaps 951:
952: curp->file_status = MANDOCLEVEL_OK;
953: curp->mdoc = NULL;
954: curp->man = NULL;
1.45 schwarze 955:
956: free(curp->sodest);
957: curp->sodest = NULL;
1.1 kristaps 958: }
959:
960: void
961: mparse_free(struct mparse *curp)
962: {
963:
964: if (curp->pmdoc)
965: mdoc_free(curp->pmdoc);
966: if (curp->pman)
967: man_free(curp->pman);
968: if (curp->roff)
969: roff_free(curp->roff);
1.24 kristaps 970: if (curp->secondary)
971: free(curp->secondary->buf);
1.1 kristaps 972:
1.24 kristaps 973: free(curp->secondary);
1.45 schwarze 974: free(curp->sodest);
1.1 kristaps 975: free(curp);
976: }
977:
978: void
1.45 schwarze 979: mparse_result(struct mparse *curp,
980: struct mdoc **mdoc, struct man **man, char **sodest)
1.1 kristaps 981: {
982:
1.45 schwarze 983: if (sodest && NULL != (*sodest = curp->sodest)) {
984: *mdoc = NULL;
985: *man = NULL;
986: return;
987: }
1.9 kristaps 988: if (mdoc)
989: *mdoc = curp->mdoc;
990: if (man)
991: *man = curp->man;
1.3 kristaps 992: }
993:
994: void
995: mandoc_vmsg(enum mandocerr t, struct mparse *m,
996: int ln, int pos, const char *fmt, ...)
997: {
998: char buf[256];
999: va_list ap;
1000:
1001: va_start(ap, fmt);
1.48 schwarze 1002: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 1003: va_end(ap);
1004:
1005: mandoc_msg(t, m, ln, pos, buf);
1006: }
1007:
1008: void
1.47 schwarze 1009: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 1010: int ln, int col, const char *msg)
1011: {
1012: enum mandoclevel level;
1013:
1014: level = MANDOCLEVEL_FATAL;
1015: while (er < mandoclimits[level])
1016: level--;
1017:
1018: if (level < m->wlevel)
1019: return;
1020:
1.8 kristaps 1021: if (m->mmsg)
1022: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 1023:
1024: if (m->file_status < level)
1025: m->file_status = level;
1.7 kristaps 1026: }
1027:
1028: const char *
1029: mparse_strerror(enum mandocerr er)
1030: {
1031:
1032: return(mandocerrs[er]);
1033: }
1034:
1035: const char *
1036: mparse_strlevel(enum mandoclevel lvl)
1037: {
1038: return(mandoclevels[lvl]);
1.24 kristaps 1039: }
1040:
1041: void
1042: mparse_keep(struct mparse *p)
1043: {
1044:
1045: assert(NULL == p->secondary);
1046: p->secondary = mandoc_calloc(1, sizeof(struct buf));
1047: }
1048:
1049: const char *
1050: mparse_getkeep(const struct mparse *p)
1051: {
1052:
1053: assert(p->secondary);
1054: return(p->secondary->sz ? p->secondary->buf : NULL);
1.1 kristaps 1055: }
CVSweb