Annotation of mandoc/read.c, Revision 1.197
1.197 ! schwarze 1: /* $Id: read.c,v 1.196 2018/07/28 18:34:15 schwarze Exp $ */
1.1 kristaps 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.196 schwarze 4: * Copyright (c) 2010-2018 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: *
1.133 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.133 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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>
1.82 schwarze 22: #include <sys/mman.h>
1.80 schwarze 23: #include <sys/stat.h>
1.1 kristaps 24:
25: #include <assert.h>
26: #include <ctype.h>
1.40 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.3 kristaps 29: #include <stdarg.h>
1.1 kristaps 30: #include <stdio.h>
31: #include <stdlib.h>
32: #include <string.h>
33: #include <unistd.h>
1.140 schwarze 34: #include <zlib.h>
1.1 kristaps 35:
1.133 schwarze 36: #include "mandoc_aux.h"
1.1 kristaps 37: #include "mandoc.h"
1.133 schwarze 38: #include "roff.h"
1.1 kristaps 39: #include "mdoc.h"
40: #include "man.h"
1.133 schwarze 41: #include "libmandoc.h"
1.1 kristaps 42:
43: #define REPARSE_LIMIT 1000
44:
45: struct mparse {
1.160 schwarze 46: struct roff *roff; /* roff parser (!NULL) */
1.134 schwarze 47: struct roff_man *man; /* man parser */
1.45 schwarze 48: char *sodest; /* filename pointed to by .so */
1.83 schwarze 49: const char *file; /* filename of current input file */
50: struct buf *primary; /* buffer currently being parsed */
51: struct buf *secondary; /* preprocessed copy of input */
1.179 schwarze 52: const char *os_s; /* default operating system */
1.83 schwarze 53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
1.179 schwarze 55: enum mandocerr mmin; /* ignore messages below this */
1.83 schwarze 56: int options; /* parser options */
1.140 schwarze 57: int gzip; /* current input file is gzipped */
1.93 schwarze 58: int filenc; /* encoding of the current file */
1.1 kristaps 59: int reparse_count; /* finite interp. stack */
1.83 schwarze 60: int line; /* line number in the file */
1.1 kristaps 61: };
62:
1.84 schwarze 63: static void choose_parser(struct mparse *);
1.1 kristaps 64: static void resize_buf(struct buf *, size_t);
1.197 ! schwarze 65: static enum rofferr mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 66: static int read_whole_file(struct mparse *, const char *, int,
67: struct buf *, int *);
1.1 kristaps 68: static void mparse_end(struct mparse *);
1.37 schwarze 69: static void mparse_parse_buffer(struct mparse *, struct buf,
70: const char *);
1.1 kristaps 71:
1.3 kristaps 72: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
73: MANDOCERR_OK,
1.188 schwarze 74: MANDOCERR_OK,
1.3 kristaps 75: MANDOCERR_WARNING,
76: MANDOCERR_ERROR,
1.112 schwarze 77: MANDOCERR_UNSUPP,
1.3 kristaps 78: MANDOCERR_MAX,
79: MANDOCERR_MAX
80: };
81:
1.7 kristaps 82: static const char * const mandocerrs[MANDOCERR_MAX] = {
83: "ok",
84:
1.179 schwarze 85: "base system convention",
1.166 schwarze 86:
1.176 schwarze 87: "Mdocdate found",
88: "Mdocdate missing",
1.181 schwarze 89: "unknown architecture",
1.180 schwarze 90: "operating system explicitly specified",
1.179 schwarze 91: "RCS id missing",
1.184 schwarze 92: "referenced manual not found",
1.179 schwarze 93:
94: "generic style suggestion",
95:
1.176 schwarze 96: "legacy man(7) date format",
1.196 schwarze 97: "normalizing date format to",
1.189 schwarze 98: "lower case character in document title",
1.178 schwarze 99: "duplicate RCS id",
1.193 schwarze 100: "possible typo in section name",
1.189 schwarze 101: "unterminated quoted argument",
1.166 schwarze 102: "useless macro",
1.167 schwarze 103: "consider using OS macro",
1.172 schwarze 104: "errnos out of order",
105: "duplicate errno",
1.187 schwarze 106: "trailing delimiter",
1.174 schwarze 107: "no blank before trailing delimiter",
1.189 schwarze 108: "fill mode already enabled, skipping",
109: "fill mode already disabled, skipping",
1.195 schwarze 110: "verbatim \"--\", maybe consider using \\(em",
1.175 schwarze 111: "function name without markup",
1.189 schwarze 112: "whitespace at end of input line",
113: "bad comment style",
1.166 schwarze 114:
1.7 kristaps 115: "generic warning",
116:
117: /* related to the prologue */
1.79 schwarze 118: "missing manual title, using UNTITLED",
119: "missing manual title, using \"\"",
120: "missing manual section, using \"\"",
1.7 kristaps 121: "unknown manual section",
1.54 schwarze 122: "missing date, using today's date",
1.7 kristaps 123: "cannot parse date, using it verbatim",
1.186 schwarze 124: "date in the future, using it anyway",
1.79 schwarze 125: "missing Os macro, using \"\"",
126: "late prologue macro",
1.7 kristaps 127: "prologue macros out of order",
128:
129: /* related to document structure */
130: ".so is fragile, better use ln(1)",
1.50 schwarze 131: "no document body",
1.54 schwarze 132: "content before first section header",
133: "first section is not \"NAME\"",
1.156 schwarze 134: "NAME section without Nm before Nd",
1.128 schwarze 135: "NAME section without description",
136: "description not at the end of NAME",
137: "bad NAME section content",
1.156 schwarze 138: "missing comma before name",
1.121 schwarze 139: "missing description line, using \"\"",
1.162 schwarze 140: "description line outside NAME section",
1.7 kristaps 141: "sections out of conventional order",
1.54 schwarze 142: "duplicate section title",
143: "unexpected section",
1.185 schwarze 144: "cross reference to self",
1.87 schwarze 145: "unusual Xr order",
146: "unusual Xr punctuation",
1.86 schwarze 147: "AUTHORS section without An macro",
1.7 kristaps 148:
149: /* related to macros and nesting */
1.55 schwarze 150: "obsolete macro",
1.102 schwarze 151: "macro neither callable nor escaped",
1.7 kristaps 152: "skipping paragraph macro",
1.31 schwarze 153: "moving paragraph macro out of list",
1.7 kristaps 154: "skipping no-space macro",
155: "blocks badly nested",
156: "nested displays are not portable",
1.57 schwarze 157: "moving content out of list",
1.183 schwarze 158: "first macro on line",
1.7 kristaps 159: "line scope broken",
1.169 schwarze 160: "skipping blank line in line scope",
1.7 kristaps 161:
162: /* related to missing macro arguments */
1.58 schwarze 163: "skipping empty request",
164: "conditional request controls empty scope",
1.7 kristaps 165: "skipping empty macro",
1.124 schwarze 166: "empty block",
1.62 schwarze 167: "empty argument, using 0n",
1.60 schwarze 168: "missing display type, using -ragged",
169: "list type is not the first argument",
1.152 schwarze 170: "missing -width in -tag list, using 6n",
1.78 schwarze 171: "missing utility name, using \"\"",
1.123 schwarze 172: "missing function name, using \"\"",
1.60 schwarze 173: "empty head in list item",
174: "empty list item",
1.183 schwarze 175: "missing argument, using next line",
1.61 schwarze 176: "missing font type, using \\fR",
177: "unknown font type, using \\fR",
1.103 schwarze 178: "nothing follows prefix",
1.122 schwarze 179: "empty reference block",
1.155 schwarze 180: "missing section argument",
1.60 schwarze 181: "missing -std argument, adding it",
1.125 schwarze 182: "missing option string, using \"\"",
183: "missing resource identifier, using \"\"",
1.90 schwarze 184: "missing eqn box, using \"\"",
1.7 kristaps 185:
186: /* related to bad macro arguments */
187: "duplicate argument",
1.76 schwarze 188: "skipping duplicate argument",
1.63 schwarze 189: "skipping duplicate display type",
190: "skipping duplicate list type",
1.76 schwarze 191: "skipping -width argument",
1.126 schwarze 192: "wrong number of cells",
1.7 kristaps 193: "unknown AT&T UNIX version",
1.88 schwarze 194: "comma in function argument",
1.89 schwarze 195: "parenthesis in function name",
1.173 schwarze 196: "unknown library name",
1.67 schwarze 197: "invalid content in Rs block",
1.63 schwarze 198: "invalid Boolean argument",
199: "unknown font, skipping request",
1.126 schwarze 200: "odd number of characters in request",
1.7 kristaps 201:
202: /* related to plain text */
1.64 schwarze 203: "blank line in fill mode, using .sp",
204: "tab in filled text",
1.158 schwarze 205: "new sentence, new line",
1.64 schwarze 206: "invalid escape sequence",
207: "undefined string, using \"\"",
1.16 kristaps 208:
1.118 schwarze 209: /* related to tables */
210: "tbl line starts with span",
211: "tbl column starts with span",
212: "skipping vertical bar in tbl layout",
213:
1.7 kristaps 214: "generic error",
215:
216: /* related to tables */
1.116 schwarze 217: "non-alphabetic character in tbl options",
218: "skipping unknown tbl option",
219: "missing tbl option argument",
220: "wrong tbl option argument size",
1.118 schwarze 221: "empty tbl layout",
222: "invalid character in tbl layout",
223: "unmatched parenthesis in tbl layout",
1.119 schwarze 224: "tbl without any data cells",
225: "ignoring data in spanned tbl cell",
226: "ignoring extra tbl data cells",
227: "data block open at end of tbl",
1.7 kristaps 228:
1.68 schwarze 229: /* related to document structure and macros */
1.108 schwarze 230: NULL,
1.189 schwarze 231: "duplicate prologue macro",
232: "skipping late title macro",
1.7 kristaps 233: "input stack limit exceeded, infinite loop?",
234: "skipping bad character",
1.68 schwarze 235: "skipping unknown macro",
1.197 ! schwarze 236: "ignoring request outside macro",
1.112 schwarze 237: "skipping insecure request",
1.73 schwarze 238: "skipping item outside list",
1.68 schwarze 239: "skipping column outside column list",
240: "skipping end of block that is not open",
1.115 schwarze 241: "fewer RS blocks open, skipping",
1.68 schwarze 242: "inserting missing end of block",
243: "appending missing end of block",
244:
245: /* related to request and macro arguments */
1.7 kristaps 246: "escaped character not allowed in a name",
1.197 ! schwarze 247: "using macro argument outside macro",
! 248: "argument number is not numeric",
1.99 schwarze 249: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 250: "skipping display without arguments",
1.71 schwarze 251: "missing list type, using -item",
1.171 schwarze 252: "argument is not numeric, using 1",
1.70 schwarze 253: "missing manual name, using \"\"",
1.71 schwarze 254: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 255: "unknown standard specifier",
1.71 schwarze 256: "skipping request without numeric argument",
1.197 ! schwarze 257: "excessive shift",
1.109 schwarze 258: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 259: ".so request failed",
1.61 schwarze 260: "skipping all arguments",
261: "skipping excess arguments",
1.92 kristaps 262: "divide by zero",
1.112 schwarze 263:
264: "unsupported feature",
265: "input too large",
1.114 schwarze 266: "unsupported control character",
1.112 schwarze 267: "unsupported roff request",
1.119 schwarze 268: "eqn delim option in tbl",
1.118 schwarze 269: "unsupported tbl layout modifier",
1.112 schwarze 270: "ignoring macro in table",
1.7 kristaps 271: };
272:
273: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
274: "SUCCESS",
1.165 schwarze 275: "STYLE",
1.7 kristaps 276: "WARNING",
277: "ERROR",
1.112 schwarze 278: "UNSUPP",
1.7 kristaps 279: "BADARG",
280: "SYSERR"
281: };
282:
1.47 schwarze 283:
1.1 kristaps 284: static void
285: resize_buf(struct buf *buf, size_t initial)
286: {
287:
288: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
289: buf->buf = mandoc_realloc(buf->buf, buf->sz);
290: }
291:
292: static void
1.84 schwarze 293: choose_parser(struct mparse *curp)
1.1 kristaps 294: {
1.83 schwarze 295: char *cp, *ep;
296: int format;
1.1 kristaps 297:
1.83 schwarze 298: /*
299: * If neither command line arguments -mdoc or -man select
300: * a parser nor the roff parser found a .Dd or .TH macro
301: * yet, look ahead in the main input buffer.
302: */
303:
304: if ((format = roff_getformat(curp->roff)) == 0) {
305: cp = curp->primary->buf;
306: ep = cp + curp->primary->sz;
307: while (cp < ep) {
1.85 schwarze 308: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 309: cp++;
310: if (cp[0] == 'D' && cp[1] == 'd') {
311: format = MPARSE_MDOC;
312: break;
313: }
314: if (cp[0] == 'T' && cp[1] == 'H') {
315: format = MPARSE_MAN;
316: break;
317: }
318: }
319: cp = memchr(cp, '\n', ep - cp);
320: if (cp == NULL)
321: break;
322: cp++;
323: }
1.1 kristaps 324: }
325:
1.83 schwarze 326: if (format == MPARSE_MDOC) {
1.137 schwarze 327: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 328: if (curp->man->mdocmac == NULL)
329: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 330: } else {
331: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 332: if (curp->man->manmac == NULL)
333: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 334: }
1.164 schwarze 335: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 336: }
337:
338: /*
1.95 schwarze 339: * Main parse routine for a buffer.
340: * It assumes encoding and line numbering are already set up.
341: * It can recurse directly (for invocations of user-defined
342: * macros, inline equations, and input line traps)
343: * and indirectly (for .so file inclusion).
1.1 kristaps 344: */
1.197 ! schwarze 345: static enum rofferr
1.95 schwarze 346: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 347: {
348: struct buf ln;
1.113 schwarze 349: const char *save_file;
1.110 schwarze 350: char *cp;
1.95 schwarze 351: size_t pos; /* byte number in the ln buffer */
1.197 ! schwarze 352: enum rofferr line_result, sub_result;
1.100 schwarze 353: int of;
1.1 kristaps 354: int lnn; /* line number in the real file */
1.110 schwarze 355: int fd;
1.1 kristaps 356: unsigned char c;
357:
1.95 schwarze 358: memset(&ln, 0, sizeof(ln));
1.1 kristaps 359:
1.47 schwarze 360: lnn = curp->line;
361: pos = 0;
1.1 kristaps 362:
1.95 schwarze 363: while (i < blk.sz) {
1.1 kristaps 364: if (0 == pos && '\0' == blk.buf[i])
365: break;
366:
367: if (start) {
368: curp->line = lnn;
369: curp->reparse_count = 0;
1.93 schwarze 370:
371: if (lnn < 3 &&
372: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 373: curp->filenc & MPARSE_LATIN1)
374: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 375: }
376:
1.95 schwarze 377: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 378:
379: /*
380: * When finding an unescaped newline character,
381: * leave the character loop to process the line.
382: * Skip a preceding carriage return, if any.
383: */
384:
1.95 schwarze 385: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 386: '\n' == blk.buf[i + 1])
387: ++i;
388: if ('\n' == blk.buf[i]) {
389: ++i;
390: ++lnn;
391: break;
392: }
393:
1.35 schwarze 394: /*
1.93 schwarze 395: * Make sure we have space for the worst
396: * case of 11 bytes: "\\[u10ffff]\0"
1.35 schwarze 397: */
398:
1.95 schwarze 399: if (pos + 11 > ln.sz)
1.35 schwarze 400: resize_buf(&ln, 256);
401:
1.47 schwarze 402: /*
1.93 schwarze 403: * Encode 8-bit input.
1.1 kristaps 404: */
405:
1.93 schwarze 406: c = blk.buf[i];
407: if (c & 0x80) {
1.95 schwarze 408: if ( ! (curp->filenc && preconv_encode(
409: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 410: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
411: curp->line, pos, "0x%x", c);
1.93 schwarze 412: ln.buf[pos++] = '?';
413: i++;
414: }
415: continue;
416: }
417:
418: /*
419: * Exclude control characters.
420: */
1.1 kristaps 421:
1.93 schwarze 422: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 423: mandoc_vmsg(c == 0x00 || c == 0x04 ||
424: c > 0x0a ? MANDOCERR_CHAR_BAD :
425: MANDOCERR_CHAR_UNSUPP,
426: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 427: i++;
1.127 schwarze 428: if (c != '\r')
429: ln.buf[pos++] = '?';
1.1 kristaps 430: continue;
431: }
432:
433: ln.buf[pos++] = blk.buf[i++];
434: }
435:
1.170 schwarze 436: if (pos + 1 >= ln.sz)
1.1 kristaps 437: resize_buf(&ln, 256);
438:
1.170 schwarze 439: if (i == blk.sz || blk.buf[i] == '\0')
440: ln.buf[pos++] = '\n';
1.1 kristaps 441: ln.buf[pos] = '\0';
442:
443: /*
444: * A significant amount of complexity is contained by
445: * the roff preprocessor. It's line-oriented but can be
446: * expressed on one line, so we need at times to
447: * readjust our starting point and re-run it. The roff
448: * preprocessor can also readjust the buffers with new
449: * data, so we pass them in wholesale.
450: */
451:
452: of = 0;
453:
1.24 kristaps 454: /*
455: * Maintain a lookaside buffer of all parsed lines. We
456: * only do this if mparse_keep() has been invoked (the
457: * buffer may be accessed with mparse_getkeep()).
458: */
459:
460: if (curp->secondary) {
1.47 schwarze 461: curp->secondary->buf = mandoc_realloc(
462: curp->secondary->buf,
463: curp->secondary->sz + pos + 2);
464: memcpy(curp->secondary->buf +
465: curp->secondary->sz,
466: ln.buf, pos);
1.24 kristaps 467: curp->secondary->sz += pos;
468: curp->secondary->buf
469: [curp->secondary->sz] = '\n';
470: curp->secondary->sz++;
471: curp->secondary->buf
472: [curp->secondary->sz] = '\0';
473: }
1.1 kristaps 474: rerun:
1.197 ! schwarze 475: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 476:
1.197 ! schwarze 477: switch (line_result) {
1.47 schwarze 478: case ROFF_REPARSE:
1.197 ! schwarze 479: case ROFF_USERCALL:
! 480: if (++curp->reparse_count > REPARSE_LIMIT) {
! 481: sub_result = ROFF_IGN;
1.3 kristaps 482: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 483: curp->line, pos, NULL);
1.197 ! schwarze 484: } else {
! 485: sub_result = mparse_buf_r(curp, ln, of, 0);
! 486: if (line_result == ROFF_USERCALL) {
! 487: if (sub_result == ROFF_USERRET)
! 488: sub_result = ROFF_CONT;
! 489: roff_userret(curp->roff);
! 490: }
! 491: if (start || sub_result == ROFF_CONT) {
! 492: pos = 0;
! 493: continue;
! 494: }
! 495: }
! 496: free(ln.buf);
! 497: return sub_result;
! 498: case ROFF_USERRET:
! 499: if (start) {
1.163 schwarze 500: pos = 0;
501: continue;
502: }
503: free(ln.buf);
1.197 ! schwarze 504: return ROFF_USERRET;
1.47 schwarze 505: case ROFF_APPEND:
1.95 schwarze 506: pos = strlen(ln.buf);
1.1 kristaps 507: continue;
1.47 schwarze 508: case ROFF_RERUN:
1.1 kristaps 509: goto rerun;
1.47 schwarze 510: case ROFF_IGN:
1.1 kristaps 511: pos = 0;
512: continue;
1.47 schwarze 513: case ROFF_SO:
1.95 schwarze 514: if ( ! (curp->options & MPARSE_SO) &&
515: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 516: curp->sodest = mandoc_strdup(ln.buf + of);
517: free(ln.buf);
1.197 ! schwarze 518: return ROFF_CONT;
1.45 schwarze 519: }
1.24 kristaps 520: /*
521: * We remove `so' clauses from our lookaside
522: * buffer because we're going to descend into
523: * the file recursively.
524: */
1.47 schwarze 525: if (curp->secondary)
1.25 kristaps 526: curp->secondary->sz -= pos + 1;
1.113 schwarze 527: save_file = curp->file;
1.148 schwarze 528: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 529: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 530: close(fd);
1.113 schwarze 531: curp->file = save_file;
532: } else {
533: curp->file = save_file;
1.52 schwarze 534: mandoc_vmsg(MANDOCERR_SO_FAIL,
535: curp, curp->line, pos,
536: ".so %s", ln.buf + of);
1.110 schwarze 537: ln.sz = mandoc_asprintf(&cp,
538: ".sp\nSee the file %s.\n.sp",
539: ln.buf + of);
540: free(ln.buf);
541: ln.buf = cp;
542: of = 0;
543: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 544: }
1.1 kristaps 545: pos = 0;
546: continue;
547: default:
548: break;
549: }
550:
1.150 schwarze 551: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 552: choose_parser(curp);
1.1 kristaps 553:
1.191 schwarze 554: if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 555: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 556: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
557: break;
1.1 kristaps 558:
559: /* Temporary buffers typically are not full. */
560:
561: if (0 == start && '\0' == blk.buf[i])
562: break;
563:
564: /* Start the next input line. */
565:
566: pos = 0;
567: }
568:
569: free(ln.buf);
1.197 ! schwarze 570: return ROFF_CONT;
1.1 kristaps 571: }
572:
573: static int
1.40 schwarze 574: read_whole_file(struct mparse *curp, const char *file, int fd,
575: struct buf *fb, int *with_mmap)
1.1 kristaps 576: {
1.161 schwarze 577: struct stat st;
1.140 schwarze 578: gzFile gz;
1.1 kristaps 579: size_t off;
580: ssize_t ssz;
1.194 schwarze 581: int gzerrnum, retval;
1.143 schwarze 582:
1.192 schwarze 583: if (fstat(fd, &st) == -1) {
584: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
585: "fstat: %s", strerror(errno));
586: return 0;
587: }
1.1 kristaps 588:
589: /*
590: * If we're a regular file, try just reading in the whole entry
591: * via mmap(). This is faster than reading it into blocks, and
592: * since each file is only a few bytes to begin with, I'm not
593: * concerned that this is going to tank any machines.
594: */
595:
1.140 schwarze 596: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 597: if (st.st_size > 0x7fffffff) {
1.111 schwarze 598: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 599: return 0;
1.1 kristaps 600: }
601: *with_mmap = 1;
602: fb->sz = (size_t)st.st_size;
1.37 schwarze 603: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 604: if (fb->buf != MAP_FAILED)
1.142 schwarze 605: return 1;
1.1 kristaps 606: }
607:
1.140 schwarze 608: if (curp->gzip) {
1.194 schwarze 609: /*
610: * Duplicating the file descriptor is required
611: * because we will have to call gzclose(3)
612: * to free memory used internally by zlib,
613: * but that will also close the file descriptor,
614: * which this function must not do.
615: */
616: if ((fd = dup(fd)) == -1) {
617: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
618: "dup: %s", strerror(errno));
619: return 0;
620: }
1.192 schwarze 621: if ((gz = gzdopen(fd, "rb")) == NULL) {
622: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
623: "gzdopen: %s", strerror(errno));
1.194 schwarze 624: close(fd);
1.192 schwarze 625: return 0;
626: }
1.140 schwarze 627: } else
628: gz = NULL;
629:
1.1 kristaps 630: /*
631: * If this isn't a regular file (like, say, stdin), then we must
632: * go the old way and just read things in bit by bit.
633: */
634:
635: *with_mmap = 0;
636: off = 0;
1.194 schwarze 637: retval = 0;
1.1 kristaps 638: fb->sz = 0;
639: fb->buf = NULL;
640: for (;;) {
641: if (off == fb->sz) {
642: if (fb->sz == (1U << 31)) {
1.111 schwarze 643: mandoc_msg(MANDOCERR_TOOLARGE, curp,
644: 0, 0, NULL);
1.1 kristaps 645: break;
646: }
647: resize_buf(fb, 65536);
648: }
1.140 schwarze 649: ssz = curp->gzip ?
650: gzread(gz, fb->buf + (int)off, fb->sz - off) :
651: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 652: if (ssz == 0) {
653: fb->sz = off;
1.194 schwarze 654: retval = 1;
655: break;
1.1 kristaps 656: }
1.192 schwarze 657: if (ssz == -1) {
1.194 schwarze 658: if (curp->gzip)
659: (void)gzerror(gz, &gzerrnum);
660: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
661: curp->gzip && gzerrnum != Z_ERRNO ?
662: zError(gzerrnum) : strerror(errno));
1.192 schwarze 663: break;
664: }
1.1 kristaps 665: off += (size_t)ssz;
666: }
667:
1.194 schwarze 668: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
669: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
670: gzerrnum == Z_ERRNO ? strerror(errno) :
671: zError(gzerrnum));
672: if (retval == 0) {
673: free(fb->buf);
674: fb->buf = NULL;
675: }
676: return retval;
1.1 kristaps 677: }
678:
679: static void
680: mparse_end(struct mparse *curp)
681: {
1.135 schwarze 682: if (curp->man->macroset == MACROSET_NONE)
683: curp->man->macroset = MACROSET_MAN;
684: if (curp->man->macroset == MACROSET_MDOC)
685: mdoc_endparse(curp->man);
686: else
1.111 schwarze 687: man_endparse(curp->man);
1.1 kristaps 688: roff_endparse(curp->roff);
689: }
690:
691: static void
1.36 schwarze 692: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 693: {
1.85 schwarze 694: struct buf *svprimary;
1.28 joerg 695: const char *svfile;
1.95 schwarze 696: size_t offset;
1.36 schwarze 697: static int recursion_depth;
698:
699: if (64 < recursion_depth) {
700: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
701: return;
702: }
1.28 joerg 703:
704: /* Line number is per-file. */
705: svfile = curp->file;
706: curp->file = file;
1.85 schwarze 707: svprimary = curp->primary;
1.83 schwarze 708: curp->primary = &blk;
1.28 joerg 709: curp->line = 1;
1.36 schwarze 710: recursion_depth++;
1.28 joerg 711:
1.93 schwarze 712: /* Skip an UTF-8 byte order mark. */
713: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
714: (unsigned char)blk.buf[0] == 0xef &&
715: (unsigned char)blk.buf[1] == 0xbb &&
716: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 717: offset = 3;
1.93 schwarze 718: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 719: } else
720: offset = 0;
1.93 schwarze 721:
1.95 schwarze 722: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 723:
1.111 schwarze 724: if (--recursion_depth == 0)
1.28 joerg 725: mparse_end(curp);
726:
1.85 schwarze 727: curp->primary = svprimary;
1.28 joerg 728: curp->file = svfile;
729: }
730:
731: enum mandoclevel
1.104 schwarze 732: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 733: const char *file)
734: {
735: struct buf blk;
736:
1.104 schwarze 737: blk.buf = buf;
1.28 joerg 738: blk.sz = len;
739:
1.36 schwarze 740: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 741: return curp->file_status;
1.28 joerg 742: }
743:
1.98 schwarze 744: /*
745: * Read the whole file into memory and call the parsers.
746: * Called recursively when an .so request is encountered.
747: */
1.36 schwarze 748: enum mandoclevel
749: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 750: {
1.28 joerg 751: struct buf blk;
752: int with_mmap;
1.93 schwarze 753: int save_filenc;
1.1 kristaps 754:
1.91 schwarze 755: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 756: save_filenc = curp->filenc;
757: curp->filenc = curp->options &
758: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 759: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 760: curp->filenc = save_filenc;
1.91 schwarze 761: if (with_mmap)
762: munmap(blk.buf, blk.sz);
763: else
764: free(blk.buf);
765: }
1.142 schwarze 766: return curp->file_status;
1.82 schwarze 767: }
768:
1.148 schwarze 769: int
770: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 771: {
772: char *cp;
1.148 schwarze 773: int fd;
1.82 schwarze 774:
775: curp->file = file;
1.140 schwarze 776: cp = strrchr(file, '.');
777: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 778:
1.140 schwarze 779: /* First try to use the filename as it is. */
1.98 schwarze 780:
1.148 schwarze 781: if ((fd = open(file, O_RDONLY)) != -1)
782: return fd;
1.98 schwarze 783:
1.140 schwarze 784: /*
785: * If that doesn't work and the filename doesn't
786: * already end in .gz, try appending .gz.
787: */
1.98 schwarze 788:
1.140 schwarze 789: if ( ! curp->gzip) {
1.98 schwarze 790: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 791: fd = open(cp, O_RDONLY);
1.108 schwarze 792: free(cp);
1.148 schwarze 793: if (fd != -1) {
1.140 schwarze 794: curp->gzip = 1;
1.148 schwarze 795: return fd;
1.82 schwarze 796: }
797: }
798:
1.140 schwarze 799: /* Neither worked, give up. */
1.97 schwarze 800:
1.140 schwarze 801: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 802: return -1;
1.1 kristaps 803: }
804:
805: struct mparse *
1.179 schwarze 806: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
807: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 808: {
809: struct mparse *curp;
1.10 kristaps 810:
1.1 kristaps 811: curp = mandoc_calloc(1, sizeof(struct mparse));
812:
1.44 schwarze 813: curp->options = options;
1.179 schwarze 814: curp->mmin = mmin;
1.1 kristaps 815: curp->mmsg = mmsg;
1.179 schwarze 816: curp->os_s = os_s;
1.1 kristaps 817:
1.144 schwarze 818: curp->roff = roff_alloc(curp, options);
1.179 schwarze 819: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 820: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 821: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 822: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 823: if (curp->man->mdocmac == NULL)
824: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 825: } else if (curp->options & MPARSE_MAN) {
826: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 827: if (curp->man->manmac == NULL)
828: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 829: }
1.138 schwarze 830: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 831: curp->man->meta.os_e = os_e;
1.142 schwarze 832: return curp;
1.1 kristaps 833: }
834:
835: void
836: mparse_reset(struct mparse *curp)
837: {
838: roff_reset(curp->roff);
1.150 schwarze 839: roff_man_reset(curp->man);
1.160 schwarze 840:
841: free(curp->sodest);
842: curp->sodest = NULL;
843:
1.24 kristaps 844: if (curp->secondary)
845: curp->secondary->sz = 0;
1.1 kristaps 846:
847: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 848: curp->gzip = 0;
1.1 kristaps 849: }
850:
851: void
852: mparse_free(struct mparse *curp)
853: {
854:
1.164 schwarze 855: roffhash_free(curp->man->mdocmac);
856: roffhash_free(curp->man->manmac);
1.137 schwarze 857: roff_man_free(curp->man);
1.160 schwarze 858: roff_free(curp->roff);
1.24 kristaps 859: if (curp->secondary)
860: free(curp->secondary->buf);
1.1 kristaps 861:
1.24 kristaps 862: free(curp->secondary);
1.45 schwarze 863: free(curp->sodest);
1.1 kristaps 864: free(curp);
865: }
866:
867: void
1.135 schwarze 868: mparse_result(struct mparse *curp, struct roff_man **man,
869: char **sodest)
1.1 kristaps 870: {
871:
1.45 schwarze 872: if (sodest && NULL != (*sodest = curp->sodest)) {
873: *man = NULL;
874: return;
875: }
1.9 kristaps 876: if (man)
877: *man = curp->man;
1.157 schwarze 878: }
879:
880: void
881: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
882: {
883: if (curp->file_status > *rc)
884: *rc = curp->file_status;
1.3 kristaps 885: }
886:
887: void
888: mandoc_vmsg(enum mandocerr t, struct mparse *m,
889: int ln, int pos, const char *fmt, ...)
890: {
891: char buf[256];
892: va_list ap;
893:
894: va_start(ap, fmt);
1.48 schwarze 895: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 896: va_end(ap);
897:
898: mandoc_msg(t, m, ln, pos, buf);
899: }
900:
901: void
1.47 schwarze 902: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 903: int ln, int col, const char *msg)
904: {
905: enum mandoclevel level;
906:
1.179 schwarze 907: if (er < m->mmin && er != MANDOCERR_FILE)
908: return;
909:
1.112 schwarze 910: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 911: while (er < mandoclimits[level])
912: level--;
913:
1.8 kristaps 914: if (m->mmsg)
915: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 916:
917: if (m->file_status < level)
918: m->file_status = level;
1.7 kristaps 919: }
920:
921: const char *
922: mparse_strerror(enum mandocerr er)
923: {
924:
1.142 schwarze 925: return mandocerrs[er];
1.7 kristaps 926: }
927:
928: const char *
929: mparse_strlevel(enum mandoclevel lvl)
930: {
1.142 schwarze 931: return mandoclevels[lvl];
1.24 kristaps 932: }
933:
934: void
935: mparse_keep(struct mparse *p)
936: {
937:
938: assert(NULL == p->secondary);
939: p->secondary = mandoc_calloc(1, sizeof(struct buf));
940: }
941:
942: const char *
943: mparse_getkeep(const struct mparse *p)
944: {
945:
946: assert(p->secondary);
1.142 schwarze 947: return p->secondary->sz ? p->secondary->buf : NULL;
1.1 kristaps 948: }
CVSweb