Annotation of mandoc/read.c, Revision 1.198
1.198 ! schwarze 1: /* $Id: read.c,v 1.197 2018/08/23 14:29:38 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 */
1.198 ! schwarze 51: struct buf *secondary; /* copy of top level 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 *);
1.198 ! schwarze 64: static void free_buf_list(struct buf *);
1.1 kristaps 65: static void resize_buf(struct buf *, size_t);
1.197 schwarze 66: static enum rofferr mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.40 schwarze 67: static int read_whole_file(struct mparse *, const char *, int,
68: struct buf *, int *);
1.1 kristaps 69: static void mparse_end(struct mparse *);
1.37 schwarze 70: static void mparse_parse_buffer(struct mparse *, struct buf,
71: const char *);
1.1 kristaps 72:
1.3 kristaps 73: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
74: MANDOCERR_OK,
1.188 schwarze 75: MANDOCERR_OK,
1.3 kristaps 76: MANDOCERR_WARNING,
77: MANDOCERR_ERROR,
1.112 schwarze 78: MANDOCERR_UNSUPP,
1.3 kristaps 79: MANDOCERR_MAX,
80: MANDOCERR_MAX
81: };
82:
1.7 kristaps 83: static const char * const mandocerrs[MANDOCERR_MAX] = {
84: "ok",
85:
1.179 schwarze 86: "base system convention",
1.166 schwarze 87:
1.176 schwarze 88: "Mdocdate found",
89: "Mdocdate missing",
1.181 schwarze 90: "unknown architecture",
1.180 schwarze 91: "operating system explicitly specified",
1.179 schwarze 92: "RCS id missing",
1.184 schwarze 93: "referenced manual not found",
1.179 schwarze 94:
95: "generic style suggestion",
96:
1.176 schwarze 97: "legacy man(7) date format",
1.196 schwarze 98: "normalizing date format to",
1.189 schwarze 99: "lower case character in document title",
1.178 schwarze 100: "duplicate RCS id",
1.193 schwarze 101: "possible typo in section name",
1.189 schwarze 102: "unterminated quoted argument",
1.166 schwarze 103: "useless macro",
1.167 schwarze 104: "consider using OS macro",
1.172 schwarze 105: "errnos out of order",
106: "duplicate errno",
1.187 schwarze 107: "trailing delimiter",
1.174 schwarze 108: "no blank before trailing delimiter",
1.189 schwarze 109: "fill mode already enabled, skipping",
110: "fill mode already disabled, skipping",
1.195 schwarze 111: "verbatim \"--\", maybe consider using \\(em",
1.175 schwarze 112: "function name without markup",
1.189 schwarze 113: "whitespace at end of input line",
114: "bad comment style",
1.166 schwarze 115:
1.7 kristaps 116: "generic warning",
117:
118: /* related to the prologue */
1.79 schwarze 119: "missing manual title, using UNTITLED",
120: "missing manual title, using \"\"",
121: "missing manual section, using \"\"",
1.7 kristaps 122: "unknown manual section",
1.54 schwarze 123: "missing date, using today's date",
1.7 kristaps 124: "cannot parse date, using it verbatim",
1.186 schwarze 125: "date in the future, using it anyway",
1.79 schwarze 126: "missing Os macro, using \"\"",
127: "late prologue macro",
1.7 kristaps 128: "prologue macros out of order",
129:
130: /* related to document structure */
131: ".so is fragile, better use ln(1)",
1.50 schwarze 132: "no document body",
1.54 schwarze 133: "content before first section header",
134: "first section is not \"NAME\"",
1.156 schwarze 135: "NAME section without Nm before Nd",
1.128 schwarze 136: "NAME section without description",
137: "description not at the end of NAME",
138: "bad NAME section content",
1.156 schwarze 139: "missing comma before name",
1.121 schwarze 140: "missing description line, using \"\"",
1.162 schwarze 141: "description line outside NAME section",
1.7 kristaps 142: "sections out of conventional order",
1.54 schwarze 143: "duplicate section title",
144: "unexpected section",
1.185 schwarze 145: "cross reference to self",
1.87 schwarze 146: "unusual Xr order",
147: "unusual Xr punctuation",
1.86 schwarze 148: "AUTHORS section without An macro",
1.7 kristaps 149:
150: /* related to macros and nesting */
1.55 schwarze 151: "obsolete macro",
1.102 schwarze 152: "macro neither callable nor escaped",
1.7 kristaps 153: "skipping paragraph macro",
1.31 schwarze 154: "moving paragraph macro out of list",
1.7 kristaps 155: "skipping no-space macro",
156: "blocks badly nested",
157: "nested displays are not portable",
1.57 schwarze 158: "moving content out of list",
1.183 schwarze 159: "first macro on line",
1.7 kristaps 160: "line scope broken",
1.169 schwarze 161: "skipping blank line in line scope",
1.7 kristaps 162:
163: /* related to missing macro arguments */
1.58 schwarze 164: "skipping empty request",
165: "conditional request controls empty scope",
1.7 kristaps 166: "skipping empty macro",
1.124 schwarze 167: "empty block",
1.62 schwarze 168: "empty argument, using 0n",
1.60 schwarze 169: "missing display type, using -ragged",
170: "list type is not the first argument",
1.152 schwarze 171: "missing -width in -tag list, using 6n",
1.78 schwarze 172: "missing utility name, using \"\"",
1.123 schwarze 173: "missing function name, using \"\"",
1.60 schwarze 174: "empty head in list item",
175: "empty list item",
1.183 schwarze 176: "missing argument, using next line",
1.61 schwarze 177: "missing font type, using \\fR",
178: "unknown font type, using \\fR",
1.103 schwarze 179: "nothing follows prefix",
1.122 schwarze 180: "empty reference block",
1.155 schwarze 181: "missing section argument",
1.60 schwarze 182: "missing -std argument, adding it",
1.125 schwarze 183: "missing option string, using \"\"",
184: "missing resource identifier, using \"\"",
1.90 schwarze 185: "missing eqn box, using \"\"",
1.7 kristaps 186:
187: /* related to bad macro arguments */
188: "duplicate argument",
1.76 schwarze 189: "skipping duplicate argument",
1.63 schwarze 190: "skipping duplicate display type",
191: "skipping duplicate list type",
1.76 schwarze 192: "skipping -width argument",
1.126 schwarze 193: "wrong number of cells",
1.7 kristaps 194: "unknown AT&T UNIX version",
1.88 schwarze 195: "comma in function argument",
1.89 schwarze 196: "parenthesis in function name",
1.173 schwarze 197: "unknown library name",
1.67 schwarze 198: "invalid content in Rs block",
1.63 schwarze 199: "invalid Boolean argument",
200: "unknown font, skipping request",
1.126 schwarze 201: "odd number of characters in request",
1.7 kristaps 202:
203: /* related to plain text */
1.64 schwarze 204: "blank line in fill mode, using .sp",
205: "tab in filled text",
1.158 schwarze 206: "new sentence, new line",
1.64 schwarze 207: "invalid escape sequence",
208: "undefined string, using \"\"",
1.16 kristaps 209:
1.118 schwarze 210: /* related to tables */
211: "tbl line starts with span",
212: "tbl column starts with span",
213: "skipping vertical bar in tbl layout",
214:
1.7 kristaps 215: "generic error",
216:
217: /* related to tables */
1.116 schwarze 218: "non-alphabetic character in tbl options",
219: "skipping unknown tbl option",
220: "missing tbl option argument",
221: "wrong tbl option argument size",
1.118 schwarze 222: "empty tbl layout",
223: "invalid character in tbl layout",
224: "unmatched parenthesis in tbl layout",
1.119 schwarze 225: "tbl without any data cells",
226: "ignoring data in spanned tbl cell",
227: "ignoring extra tbl data cells",
228: "data block open at end of tbl",
1.7 kristaps 229:
1.68 schwarze 230: /* related to document structure and macros */
1.108 schwarze 231: NULL,
1.189 schwarze 232: "duplicate prologue macro",
233: "skipping late title macro",
1.7 kristaps 234: "input stack limit exceeded, infinite loop?",
235: "skipping bad character",
1.68 schwarze 236: "skipping unknown macro",
1.197 schwarze 237: "ignoring request outside macro",
1.112 schwarze 238: "skipping insecure request",
1.73 schwarze 239: "skipping item outside list",
1.68 schwarze 240: "skipping column outside column list",
241: "skipping end of block that is not open",
1.115 schwarze 242: "fewer RS blocks open, skipping",
1.68 schwarze 243: "inserting missing end of block",
244: "appending missing end of block",
245:
246: /* related to request and macro arguments */
1.7 kristaps 247: "escaped character not allowed in a name",
1.197 schwarze 248: "using macro argument outside macro",
249: "argument number is not numeric",
1.99 schwarze 250: "NOT IMPLEMENTED: Bd -file",
1.145 schwarze 251: "skipping display without arguments",
1.71 schwarze 252: "missing list type, using -item",
1.171 schwarze 253: "argument is not numeric, using 1",
1.70 schwarze 254: "missing manual name, using \"\"",
1.71 schwarze 255: "uname(3) system call failed, using UNKNOWN",
1.63 schwarze 256: "unknown standard specifier",
1.71 schwarze 257: "skipping request without numeric argument",
1.197 schwarze 258: "excessive shift",
1.109 schwarze 259: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.110 schwarze 260: ".so request failed",
1.61 schwarze 261: "skipping all arguments",
262: "skipping excess arguments",
1.92 kristaps 263: "divide by zero",
1.112 schwarze 264:
265: "unsupported feature",
266: "input too large",
1.114 schwarze 267: "unsupported control character",
1.112 schwarze 268: "unsupported roff request",
1.119 schwarze 269: "eqn delim option in tbl",
1.118 schwarze 270: "unsupported tbl layout modifier",
1.112 schwarze 271: "ignoring macro in table",
1.7 kristaps 272: };
273:
274: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
275: "SUCCESS",
1.165 schwarze 276: "STYLE",
1.7 kristaps 277: "WARNING",
278: "ERROR",
1.112 schwarze 279: "UNSUPP",
1.7 kristaps 280: "BADARG",
281: "SYSERR"
282: };
283:
1.47 schwarze 284:
1.1 kristaps 285: static void
286: resize_buf(struct buf *buf, size_t initial)
287: {
288:
289: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
290: buf->buf = mandoc_realloc(buf->buf, buf->sz);
291: }
292:
293: static void
1.198 ! schwarze 294: free_buf_list(struct buf *buf)
! 295: {
! 296: struct buf *tmp;
! 297:
! 298: while (buf != NULL) {
! 299: tmp = buf;
! 300: buf = tmp->next;
! 301: free(tmp->buf);
! 302: free(tmp);
! 303: }
! 304: }
! 305:
! 306: static void
1.84 schwarze 307: choose_parser(struct mparse *curp)
1.1 kristaps 308: {
1.83 schwarze 309: char *cp, *ep;
310: int format;
1.1 kristaps 311:
1.83 schwarze 312: /*
313: * If neither command line arguments -mdoc or -man select
314: * a parser nor the roff parser found a .Dd or .TH macro
315: * yet, look ahead in the main input buffer.
316: */
317:
318: if ((format = roff_getformat(curp->roff)) == 0) {
319: cp = curp->primary->buf;
320: ep = cp + curp->primary->sz;
321: while (cp < ep) {
1.85 schwarze 322: if (*cp == '.' || *cp == '\'') {
1.83 schwarze 323: cp++;
324: if (cp[0] == 'D' && cp[1] == 'd') {
325: format = MPARSE_MDOC;
326: break;
327: }
328: if (cp[0] == 'T' && cp[1] == 'H') {
329: format = MPARSE_MAN;
330: break;
331: }
332: }
333: cp = memchr(cp, '\n', ep - cp);
334: if (cp == NULL)
335: break;
336: cp++;
337: }
1.1 kristaps 338: }
339:
1.83 schwarze 340: if (format == MPARSE_MDOC) {
1.137 schwarze 341: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 342: if (curp->man->mdocmac == NULL)
343: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 344: } else {
345: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 346: if (curp->man->manmac == NULL)
347: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.47 schwarze 348: }
1.164 schwarze 349: curp->man->first->tok = TOKEN_NONE;
1.1 kristaps 350: }
351:
352: /*
1.95 schwarze 353: * Main parse routine for a buffer.
354: * It assumes encoding and line numbering are already set up.
355: * It can recurse directly (for invocations of user-defined
356: * macros, inline equations, and input line traps)
357: * and indirectly (for .so file inclusion).
1.1 kristaps 358: */
1.197 schwarze 359: static enum rofferr
1.95 schwarze 360: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 kristaps 361: {
362: struct buf ln;
1.198 ! schwarze 363: struct buf *firstln, *lastln, *thisln;
1.113 schwarze 364: const char *save_file;
1.110 schwarze 365: char *cp;
1.95 schwarze 366: size_t pos; /* byte number in the ln buffer */
1.198 ! schwarze 367: enum rofferr line_result, result;
1.100 schwarze 368: int of;
1.1 kristaps 369: int lnn; /* line number in the real file */
1.110 schwarze 370: int fd;
1.1 kristaps 371: unsigned char c;
372:
1.198 ! schwarze 373: ln.sz = 256;
! 374: ln.buf = mandoc_malloc(ln.sz);
! 375: ln.next = NULL;
! 376: firstln = NULL;
1.47 schwarze 377: lnn = curp->line;
378: pos = 0;
1.198 ! schwarze 379: result = ROFF_CONT;
1.1 kristaps 380:
1.95 schwarze 381: while (i < blk.sz) {
1.1 kristaps 382: if (0 == pos && '\0' == blk.buf[i])
383: break;
384:
385: if (start) {
386: curp->line = lnn;
387: curp->reparse_count = 0;
1.93 schwarze 388:
389: if (lnn < 3 &&
390: curp->filenc & MPARSE_UTF8 &&
1.95 schwarze 391: curp->filenc & MPARSE_LATIN1)
392: curp->filenc = preconv_cue(&blk, i);
1.1 kristaps 393: }
394:
1.95 schwarze 395: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 kristaps 396:
397: /*
398: * When finding an unescaped newline character,
399: * leave the character loop to process the line.
400: * Skip a preceding carriage return, if any.
401: */
402:
1.95 schwarze 403: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 kristaps 404: '\n' == blk.buf[i + 1])
405: ++i;
406: if ('\n' == blk.buf[i]) {
407: ++i;
408: ++lnn;
409: break;
410: }
411:
1.35 schwarze 412: /*
1.93 schwarze 413: * Make sure we have space for the worst
1.198 ! schwarze 414: * case of 12 bytes: "\\[u10ffff]\n\0"
1.35 schwarze 415: */
416:
1.198 ! schwarze 417: if (pos + 12 > ln.sz)
1.35 schwarze 418: resize_buf(&ln, 256);
419:
1.47 schwarze 420: /*
1.93 schwarze 421: * Encode 8-bit input.
1.1 kristaps 422: */
423:
1.93 schwarze 424: c = blk.buf[i];
425: if (c & 0x80) {
1.95 schwarze 426: if ( ! (curp->filenc && preconv_encode(
427: &blk, &i, &ln, &pos, &curp->filenc))) {
1.114 schwarze 428: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
429: curp->line, pos, "0x%x", c);
1.93 schwarze 430: ln.buf[pos++] = '?';
431: i++;
432: }
433: continue;
434: }
435:
436: /*
437: * Exclude control characters.
438: */
1.1 kristaps 439:
1.93 schwarze 440: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.114 schwarze 441: mandoc_vmsg(c == 0x00 || c == 0x04 ||
442: c > 0x0a ? MANDOCERR_CHAR_BAD :
443: MANDOCERR_CHAR_UNSUPP,
444: curp, curp->line, pos, "0x%x", c);
1.1 kristaps 445: i++;
1.127 schwarze 446: if (c != '\r')
447: ln.buf[pos++] = '?';
1.1 kristaps 448: continue;
449: }
450:
451: ln.buf[pos++] = blk.buf[i++];
452: }
1.198 ! schwarze 453: ln.buf[pos] = '\0';
! 454:
! 455: /*
! 456: * Maintain a lookaside buffer of all lines.
! 457: * parsed from this input source.
! 458: */
! 459:
! 460: thisln = mandoc_malloc(sizeof(*thisln));
! 461: thisln->buf = mandoc_strdup(ln.buf);
! 462: thisln->sz = strlen(ln.buf) + 1;
! 463: thisln->next = NULL;
! 464: if (firstln == NULL) {
! 465: firstln = lastln = thisln;
! 466: if (curp->secondary == NULL)
! 467: curp->secondary = firstln;
! 468: } else {
! 469: lastln->next = thisln;
! 470: lastln = thisln;
! 471: }
1.1 kristaps 472:
1.198 ! schwarze 473: /* XXX Ugly hack to mark the end of the input. */
1.1 kristaps 474:
1.198 ! schwarze 475: if (i == blk.sz || blk.buf[i] == '\0') {
1.170 schwarze 476: ln.buf[pos++] = '\n';
1.198 ! schwarze 477: ln.buf[pos] = '\0';
! 478: }
1.1 kristaps 479:
480: /*
481: * A significant amount of complexity is contained by
482: * the roff preprocessor. It's line-oriented but can be
483: * expressed on one line, so we need at times to
484: * readjust our starting point and re-run it. The roff
485: * preprocessor can also readjust the buffers with new
486: * data, so we pass them in wholesale.
487: */
488:
489: of = 0;
490: rerun:
1.197 schwarze 491: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 kristaps 492:
1.197 schwarze 493: switch (line_result) {
1.47 schwarze 494: case ROFF_REPARSE:
1.197 schwarze 495: case ROFF_USERCALL:
496: if (++curp->reparse_count > REPARSE_LIMIT) {
1.198 ! schwarze 497: result = ROFF_IGN;
1.3 kristaps 498: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.47 schwarze 499: curp->line, pos, NULL);
1.197 schwarze 500: } else {
1.198 ! schwarze 501: result = mparse_buf_r(curp, ln, of, 0);
1.197 schwarze 502: if (line_result == ROFF_USERCALL) {
1.198 ! schwarze 503: if (result == ROFF_USERRET)
! 504: result = ROFF_CONT;
1.197 schwarze 505: roff_userret(curp->roff);
506: }
1.198 ! schwarze 507: if (start || result == ROFF_CONT) {
1.197 schwarze 508: pos = 0;
509: continue;
510: }
511: }
1.198 ! schwarze 512: goto out;
1.197 schwarze 513: case ROFF_USERRET:
514: if (start) {
1.163 schwarze 515: pos = 0;
516: continue;
517: }
1.198 ! schwarze 518: result = ROFF_USERRET;
! 519: goto out;
1.47 schwarze 520: case ROFF_APPEND:
1.95 schwarze 521: pos = strlen(ln.buf);
1.1 kristaps 522: continue;
1.47 schwarze 523: case ROFF_RERUN:
1.1 kristaps 524: goto rerun;
1.47 schwarze 525: case ROFF_IGN:
1.1 kristaps 526: pos = 0;
527: continue;
1.47 schwarze 528: case ROFF_SO:
1.95 schwarze 529: if ( ! (curp->options & MPARSE_SO) &&
530: (i >= blk.sz || blk.buf[i] == '\0')) {
1.45 schwarze 531: curp->sodest = mandoc_strdup(ln.buf + of);
1.198 ! schwarze 532: goto out;
1.45 schwarze 533: }
1.113 schwarze 534: save_file = curp->file;
1.148 schwarze 535: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.110 schwarze 536: mparse_readfd(curp, fd, ln.buf + of);
1.147 schwarze 537: close(fd);
1.113 schwarze 538: curp->file = save_file;
539: } else {
540: curp->file = save_file;
1.52 schwarze 541: mandoc_vmsg(MANDOCERR_SO_FAIL,
542: curp, curp->line, pos,
543: ".so %s", ln.buf + of);
1.110 schwarze 544: ln.sz = mandoc_asprintf(&cp,
545: ".sp\nSee the file %s.\n.sp",
546: ln.buf + of);
547: free(ln.buf);
548: ln.buf = cp;
549: of = 0;
550: mparse_buf_r(curp, ln, of, 0);
1.52 schwarze 551: }
1.1 kristaps 552: pos = 0;
553: continue;
554: default:
555: break;
556: }
557:
1.150 schwarze 558: if (curp->man->macroset == MACROSET_NONE)
1.84 schwarze 559: choose_parser(curp);
1.1 kristaps 560:
1.191 schwarze 561: if ((curp->man->macroset == MACROSET_MDOC ?
1.135 schwarze 562: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
1.100 schwarze 563: man_parseln(curp->man, curp->line, ln.buf, of)) == 2)
564: break;
1.1 kristaps 565:
566: /* Temporary buffers typically are not full. */
567:
568: if (0 == start && '\0' == blk.buf[i])
569: break;
570:
571: /* Start the next input line. */
572:
573: pos = 0;
574: }
1.198 ! schwarze 575: out:
1.1 kristaps 576: free(ln.buf);
1.198 ! schwarze 577: if (firstln != curp->secondary)
! 578: free_buf_list(firstln);
! 579: return result;
1.1 kristaps 580: }
581:
582: static int
1.40 schwarze 583: read_whole_file(struct mparse *curp, const char *file, int fd,
584: struct buf *fb, int *with_mmap)
1.1 kristaps 585: {
1.161 schwarze 586: struct stat st;
1.140 schwarze 587: gzFile gz;
1.1 kristaps 588: size_t off;
589: ssize_t ssz;
1.194 schwarze 590: int gzerrnum, retval;
1.143 schwarze 591:
1.192 schwarze 592: if (fstat(fd, &st) == -1) {
593: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
594: "fstat: %s", strerror(errno));
595: return 0;
596: }
1.1 kristaps 597:
598: /*
599: * If we're a regular file, try just reading in the whole entry
600: * via mmap(). This is faster than reading it into blocks, and
601: * since each file is only a few bytes to begin with, I'm not
602: * concerned that this is going to tank any machines.
603: */
604:
1.140 schwarze 605: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.131 schwarze 606: if (st.st_size > 0x7fffffff) {
1.111 schwarze 607: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.142 schwarze 608: return 0;
1.1 kristaps 609: }
610: *with_mmap = 1;
611: fb->sz = (size_t)st.st_size;
1.37 schwarze 612: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 kristaps 613: if (fb->buf != MAP_FAILED)
1.142 schwarze 614: return 1;
1.1 kristaps 615: }
616:
1.140 schwarze 617: if (curp->gzip) {
1.194 schwarze 618: /*
619: * Duplicating the file descriptor is required
620: * because we will have to call gzclose(3)
621: * to free memory used internally by zlib,
622: * but that will also close the file descriptor,
623: * which this function must not do.
624: */
625: if ((fd = dup(fd)) == -1) {
626: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
627: "dup: %s", strerror(errno));
628: return 0;
629: }
1.192 schwarze 630: if ((gz = gzdopen(fd, "rb")) == NULL) {
631: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
632: "gzdopen: %s", strerror(errno));
1.194 schwarze 633: close(fd);
1.192 schwarze 634: return 0;
635: }
1.140 schwarze 636: } else
637: gz = NULL;
638:
1.1 kristaps 639: /*
640: * If this isn't a regular file (like, say, stdin), then we must
641: * go the old way and just read things in bit by bit.
642: */
643:
644: *with_mmap = 0;
645: off = 0;
1.194 schwarze 646: retval = 0;
1.1 kristaps 647: fb->sz = 0;
648: fb->buf = NULL;
649: for (;;) {
650: if (off == fb->sz) {
651: if (fb->sz == (1U << 31)) {
1.111 schwarze 652: mandoc_msg(MANDOCERR_TOOLARGE, curp,
653: 0, 0, NULL);
1.1 kristaps 654: break;
655: }
656: resize_buf(fb, 65536);
657: }
1.140 schwarze 658: ssz = curp->gzip ?
659: gzread(gz, fb->buf + (int)off, fb->sz - off) :
660: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 kristaps 661: if (ssz == 0) {
662: fb->sz = off;
1.194 schwarze 663: retval = 1;
664: break;
1.1 kristaps 665: }
1.192 schwarze 666: if (ssz == -1) {
1.194 schwarze 667: if (curp->gzip)
668: (void)gzerror(gz, &gzerrnum);
669: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
670: curp->gzip && gzerrnum != Z_ERRNO ?
671: zError(gzerrnum) : strerror(errno));
1.192 schwarze 672: break;
673: }
1.1 kristaps 674: off += (size_t)ssz;
675: }
676:
1.194 schwarze 677: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
678: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
679: gzerrnum == Z_ERRNO ? strerror(errno) :
680: zError(gzerrnum));
681: if (retval == 0) {
682: free(fb->buf);
683: fb->buf = NULL;
684: }
685: return retval;
1.1 kristaps 686: }
687:
688: static void
689: mparse_end(struct mparse *curp)
690: {
1.135 schwarze 691: if (curp->man->macroset == MACROSET_NONE)
692: curp->man->macroset = MACROSET_MAN;
693: if (curp->man->macroset == MACROSET_MDOC)
694: mdoc_endparse(curp->man);
695: else
1.111 schwarze 696: man_endparse(curp->man);
1.1 kristaps 697: roff_endparse(curp->roff);
698: }
699:
700: static void
1.36 schwarze 701: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.28 joerg 702: {
1.85 schwarze 703: struct buf *svprimary;
1.28 joerg 704: const char *svfile;
1.95 schwarze 705: size_t offset;
1.36 schwarze 706: static int recursion_depth;
707:
708: if (64 < recursion_depth) {
709: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
710: return;
711: }
1.28 joerg 712:
713: /* Line number is per-file. */
714: svfile = curp->file;
715: curp->file = file;
1.85 schwarze 716: svprimary = curp->primary;
1.83 schwarze 717: curp->primary = &blk;
1.28 joerg 718: curp->line = 1;
1.36 schwarze 719: recursion_depth++;
1.28 joerg 720:
1.93 schwarze 721: /* Skip an UTF-8 byte order mark. */
722: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
723: (unsigned char)blk.buf[0] == 0xef &&
724: (unsigned char)blk.buf[1] == 0xbb &&
725: (unsigned char)blk.buf[2] == 0xbf) {
1.95 schwarze 726: offset = 3;
1.93 schwarze 727: curp->filenc &= ~MPARSE_LATIN1;
1.95 schwarze 728: } else
729: offset = 0;
1.93 schwarze 730:
1.95 schwarze 731: mparse_buf_r(curp, blk, offset, 1);
1.28 joerg 732:
1.111 schwarze 733: if (--recursion_depth == 0)
1.28 joerg 734: mparse_end(curp);
735:
1.85 schwarze 736: curp->primary = svprimary;
1.28 joerg 737: curp->file = svfile;
738: }
739:
740: enum mandoclevel
1.104 schwarze 741: mparse_readmem(struct mparse *curp, void *buf, size_t len,
1.28 joerg 742: const char *file)
743: {
744: struct buf blk;
745:
1.104 schwarze 746: blk.buf = buf;
1.28 joerg 747: blk.sz = len;
748:
1.36 schwarze 749: mparse_parse_buffer(curp, blk, file);
1.142 schwarze 750: return curp->file_status;
1.28 joerg 751: }
752:
1.98 schwarze 753: /*
754: * Read the whole file into memory and call the parsers.
755: * Called recursively when an .so request is encountered.
756: */
1.36 schwarze 757: enum mandoclevel
758: mparse_readfd(struct mparse *curp, int fd, const char *file)
1.1 kristaps 759: {
1.28 joerg 760: struct buf blk;
761: int with_mmap;
1.93 schwarze 762: int save_filenc;
1.1 kristaps 763:
1.91 schwarze 764: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.93 schwarze 765: save_filenc = curp->filenc;
766: curp->filenc = curp->options &
767: (MPARSE_UTF8 | MPARSE_LATIN1);
1.91 schwarze 768: mparse_parse_buffer(curp, blk, file);
1.93 schwarze 769: curp->filenc = save_filenc;
1.91 schwarze 770: if (with_mmap)
771: munmap(blk.buf, blk.sz);
772: else
773: free(blk.buf);
774: }
1.142 schwarze 775: return curp->file_status;
1.82 schwarze 776: }
777:
1.148 schwarze 778: int
779: mparse_open(struct mparse *curp, const char *file)
1.82 schwarze 780: {
781: char *cp;
1.148 schwarze 782: int fd;
1.82 schwarze 783:
784: curp->file = file;
1.140 schwarze 785: cp = strrchr(file, '.');
786: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.98 schwarze 787:
1.140 schwarze 788: /* First try to use the filename as it is. */
1.98 schwarze 789:
1.148 schwarze 790: if ((fd = open(file, O_RDONLY)) != -1)
791: return fd;
1.98 schwarze 792:
1.140 schwarze 793: /*
794: * If that doesn't work and the filename doesn't
795: * already end in .gz, try appending .gz.
796: */
1.98 schwarze 797:
1.140 schwarze 798: if ( ! curp->gzip) {
1.98 schwarze 799: mandoc_asprintf(&cp, "%s.gz", file);
1.149 schwarze 800: fd = open(cp, O_RDONLY);
1.108 schwarze 801: free(cp);
1.148 schwarze 802: if (fd != -1) {
1.140 schwarze 803: curp->gzip = 1;
1.148 schwarze 804: return fd;
1.82 schwarze 805: }
806: }
807:
1.140 schwarze 808: /* Neither worked, give up. */
1.97 schwarze 809:
1.140 schwarze 810: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.148 schwarze 811: return -1;
1.1 kristaps 812: }
813:
814: struct mparse *
1.179 schwarze 815: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
816: enum mandoc_os os_e, const char *os_s)
1.1 kristaps 817: {
818: struct mparse *curp;
1.10 kristaps 819:
1.1 kristaps 820: curp = mandoc_calloc(1, sizeof(struct mparse));
821:
1.44 schwarze 822: curp->options = options;
1.179 schwarze 823: curp->mmin = mmin;
1.1 kristaps 824: curp->mmsg = mmsg;
1.179 schwarze 825: curp->os_s = os_s;
1.1 kristaps 826:
1.144 schwarze 827: curp->roff = roff_alloc(curp, options);
1.179 schwarze 828: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.137 schwarze 829: curp->options & MPARSE_QUICK ? 1 : 0);
1.136 schwarze 830: if (curp->options & MPARSE_MDOC) {
1.137 schwarze 831: curp->man->macroset = MACROSET_MDOC;
1.164 schwarze 832: if (curp->man->mdocmac == NULL)
833: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.137 schwarze 834: } else if (curp->options & MPARSE_MAN) {
835: curp->man->macroset = MACROSET_MAN;
1.164 schwarze 836: if (curp->man->manmac == NULL)
837: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.136 schwarze 838: }
1.138 schwarze 839: curp->man->first->tok = TOKEN_NONE;
1.179 schwarze 840: curp->man->meta.os_e = os_e;
1.142 schwarze 841: return curp;
1.1 kristaps 842: }
843:
844: void
845: mparse_reset(struct mparse *curp)
846: {
847: roff_reset(curp->roff);
1.150 schwarze 848: roff_man_reset(curp->man);
1.198 ! schwarze 849: free_buf_list(curp->secondary);
! 850: curp->secondary = NULL;
1.160 schwarze 851:
852: free(curp->sodest);
853: curp->sodest = NULL;
854:
1.1 kristaps 855: curp->file_status = MANDOCLEVEL_OK;
1.159 schwarze 856: curp->gzip = 0;
1.1 kristaps 857: }
858:
859: void
860: mparse_free(struct mparse *curp)
861: {
1.164 schwarze 862: roffhash_free(curp->man->mdocmac);
863: roffhash_free(curp->man->manmac);
1.137 schwarze 864: roff_man_free(curp->man);
1.160 schwarze 865: roff_free(curp->roff);
1.198 ! schwarze 866: free_buf_list(curp->secondary);
1.45 schwarze 867: free(curp->sodest);
1.1 kristaps 868: free(curp);
869: }
870:
871: void
1.135 schwarze 872: mparse_result(struct mparse *curp, struct roff_man **man,
873: char **sodest)
1.1 kristaps 874: {
875:
1.45 schwarze 876: if (sodest && NULL != (*sodest = curp->sodest)) {
877: *man = NULL;
878: return;
879: }
1.9 kristaps 880: if (man)
881: *man = curp->man;
1.157 schwarze 882: }
883:
884: void
885: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
886: {
887: if (curp->file_status > *rc)
888: *rc = curp->file_status;
1.3 kristaps 889: }
890:
891: void
892: mandoc_vmsg(enum mandocerr t, struct mparse *m,
893: int ln, int pos, const char *fmt, ...)
894: {
895: char buf[256];
896: va_list ap;
897:
898: va_start(ap, fmt);
1.48 schwarze 899: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.3 kristaps 900: va_end(ap);
901:
902: mandoc_msg(t, m, ln, pos, buf);
903: }
904:
905: void
1.47 schwarze 906: mandoc_msg(enum mandocerr er, struct mparse *m,
1.3 kristaps 907: int ln, int col, const char *msg)
908: {
909: enum mandoclevel level;
910:
1.179 schwarze 911: if (er < m->mmin && er != MANDOCERR_FILE)
912: return;
913:
1.112 schwarze 914: level = MANDOCLEVEL_UNSUPP;
1.3 kristaps 915: while (er < mandoclimits[level])
916: level--;
917:
1.8 kristaps 918: if (m->mmsg)
919: (*m->mmsg)(er, level, m->file, ln, col, msg);
1.3 kristaps 920:
921: if (m->file_status < level)
922: m->file_status = level;
1.7 kristaps 923: }
924:
925: const char *
926: mparse_strerror(enum mandocerr er)
927: {
928:
1.142 schwarze 929: return mandocerrs[er];
1.7 kristaps 930: }
931:
932: const char *
933: mparse_strlevel(enum mandoclevel lvl)
934: {
1.142 schwarze 935: return mandoclevels[lvl];
1.24 kristaps 936: }
937:
938: void
1.198 ! schwarze 939: mparse_copy(const struct mparse *p)
1.24 kristaps 940: {
1.198 ! schwarze 941: struct buf *buf;
1.24 kristaps 942:
1.198 ! schwarze 943: for (buf = p->secondary; buf != NULL; buf = buf->next)
! 944: puts(buf->buf);
1.1 kristaps 945: }
CVSweb