Annotation of src/usr.bin/mandoc/read.c, Revision 1.173
1.173 ! schwarze 1: /* $OpenBSD: read.c,v 1.172 2018/08/25 16:43:52 schwarze Exp $ */
1.1 schwarze 2: /*
3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.168 schwarze 4: * Copyright (c) 2010-2018 Ingo Schwarze <schwarze@openbsd.org>
1.20 schwarze 5: * Copyright (c) 2010, 2012 Joerg Sonnenberger <joerg@netbsd.org>
1.1 schwarze 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: *
1.108 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.108 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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: */
1.58 schwarze 19: #include <sys/types.h>
20: #include <sys/mman.h>
1.1 schwarze 21: #include <sys/stat.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.18 schwarze 25: #include <errno.h>
1.1 schwarze 26: #include <fcntl.h>
27: #include <stdarg.h>
28: #include <stdio.h>
29: #include <stdlib.h>
30: #include <string.h>
31: #include <unistd.h>
1.115 schwarze 32: #include <zlib.h>
1.1 schwarze 33:
1.108 schwarze 34: #include "mandoc_aux.h"
1.1 schwarze 35: #include "mandoc.h"
1.108 schwarze 36: #include "roff.h"
1.1 schwarze 37: #include "mdoc.h"
38: #include "man.h"
1.108 schwarze 39: #include "libmandoc.h"
1.173 ! schwarze 40: #include "roff_int.h"
1.1 schwarze 41:
42: #define REPARSE_LIMIT 1000
43:
44: struct mparse {
1.133 schwarze 45: struct roff *roff; /* roff parser (!NULL) */
1.109 schwarze 46: struct roff_man *man; /* man parser */
1.23 schwarze 47: char *sodest; /* filename pointed to by .so */
1.59 schwarze 48: const char *file; /* filename of current input file */
49: struct buf *primary; /* buffer currently being parsed */
1.170 schwarze 50: struct buf *secondary; /* copy of top level input */
1.171 schwarze 51: struct buf *loop; /* open .while request line */
1.151 schwarze 52: const char *os_s; /* default operating system */
1.59 schwarze 53: mandocmsg mmsg; /* warning/error message handler */
54: enum mandoclevel file_status; /* status of current parse */
1.151 schwarze 55: enum mandocerr mmin; /* ignore messages below this */
1.59 schwarze 56: int options; /* parser options */
1.115 schwarze 57: int gzip; /* current input file is gzipped */
1.70 schwarze 58: int filenc; /* encoding of the current file */
1.1 schwarze 59: int reparse_count; /* finite interp. stack */
1.59 schwarze 60: int line; /* line number in the file */
1.1 schwarze 61: };
62:
1.60 schwarze 63: static void choose_parser(struct mparse *);
1.170 schwarze 64: static void free_buf_list(struct buf *);
1.1 schwarze 65: static void resize_buf(struct buf *, size_t);
1.171 schwarze 66: static int mparse_buf_r(struct mparse *, struct buf, size_t, int);
1.18 schwarze 67: static int read_whole_file(struct mparse *, const char *, int,
68: struct buf *, int *);
1.1 schwarze 69: static void mparse_end(struct mparse *);
1.15 schwarze 70: static void mparse_parse_buffer(struct mparse *, struct buf,
71: const char *);
1.1 schwarze 72:
73: static const enum mandocerr mandoclimits[MANDOCLEVEL_MAX] = {
74: MANDOCERR_OK,
1.160 schwarze 75: MANDOCERR_OK,
1.1 schwarze 76: MANDOCERR_WARNING,
77: MANDOCERR_ERROR,
1.87 schwarze 78: MANDOCERR_UNSUPP,
1.1 schwarze 79: MANDOCERR_MAX,
80: MANDOCERR_MAX
81: };
82:
83: static const char * const mandocerrs[MANDOCERR_MAX] = {
84: "ok",
85:
1.151 schwarze 86: "base system convention",
1.138 schwarze 87:
1.148 schwarze 88: "Mdocdate found",
89: "Mdocdate missing",
1.153 schwarze 90: "unknown architecture",
1.152 schwarze 91: "operating system explicitly specified",
1.151 schwarze 92: "RCS id missing",
1.156 schwarze 93: "referenced manual not found",
1.151 schwarze 94:
95: "generic style suggestion",
96:
1.148 schwarze 97: "legacy man(7) date format",
1.168 schwarze 98: "normalizing date format to",
1.161 schwarze 99: "lower case character in document title",
1.150 schwarze 100: "duplicate RCS id",
1.165 jca 101: "possible typo in section name",
1.161 schwarze 102: "unterminated quoted argument",
1.138 schwarze 103: "useless macro",
1.139 schwarze 104: "consider using OS macro",
1.144 schwarze 105: "errnos out of order",
106: "duplicate errno",
1.159 schwarze 107: "trailing delimiter",
1.146 schwarze 108: "no blank before trailing delimiter",
1.161 schwarze 109: "fill mode already enabled, skipping",
110: "fill mode already disabled, skipping",
1.167 schwarze 111: "verbatim \"--\", maybe consider using \\(em",
1.147 schwarze 112: "function name without markup",
1.161 schwarze 113: "whitespace at end of input line",
114: "bad comment style",
1.138 schwarze 115:
1.1 schwarze 116: "generic warning",
117:
118: /* related to the prologue */
1.57 schwarze 119: "missing manual title, using UNTITLED",
120: "missing manual title, using \"\"",
121: "missing manual section, using \"\"",
1.1 schwarze 122: "unknown manual section",
1.32 schwarze 123: "missing date, using today's date",
1.1 schwarze 124: "cannot parse date, using it verbatim",
1.158 schwarze 125: "date in the future, using it anyway",
1.57 schwarze 126: "missing Os macro, using \"\"",
127: "late prologue macro",
1.1 schwarze 128: "prologue macros out of order",
129:
130: /* related to document structure */
131: ".so is fragile, better use ln(1)",
1.28 schwarze 132: "no document body",
1.32 schwarze 133: "content before first section header",
134: "first section is not \"NAME\"",
1.129 schwarze 135: "NAME section without Nm before Nd",
1.103 schwarze 136: "NAME section without description",
137: "description not at the end of NAME",
138: "bad NAME section content",
1.129 schwarze 139: "missing comma before name",
1.96 schwarze 140: "missing description line, using \"\"",
1.134 schwarze 141: "description line outside NAME section",
1.1 schwarze 142: "sections out of conventional order",
1.32 schwarze 143: "duplicate section title",
144: "unexpected section",
1.157 schwarze 145: "cross reference to self",
1.63 schwarze 146: "unusual Xr order",
147: "unusual Xr punctuation",
1.62 schwarze 148: "AUTHORS section without An macro",
1.1 schwarze 149:
150: /* related to macros and nesting */
1.33 schwarze 151: "obsolete macro",
1.78 schwarze 152: "macro neither callable nor escaped",
1.1 schwarze 153: "skipping paragraph macro",
1.10 schwarze 154: "moving paragraph macro out of list",
1.1 schwarze 155: "skipping no-space macro",
156: "blocks badly nested",
157: "nested displays are not portable",
1.35 schwarze 158: "moving content out of list",
1.155 schwarze 159: "first macro on line",
1.1 schwarze 160: "line scope broken",
1.141 schwarze 161: "skipping blank line in line scope",
1.1 schwarze 162:
163: /* related to missing macro arguments */
1.36 schwarze 164: "skipping empty request",
165: "conditional request controls empty scope",
1.1 schwarze 166: "skipping empty macro",
1.99 schwarze 167: "empty block",
1.40 schwarze 168: "empty argument, using 0n",
1.38 schwarze 169: "missing display type, using -ragged",
170: "list type is not the first argument",
1.125 schwarze 171: "missing -width in -tag list, using 6n",
1.56 schwarze 172: "missing utility name, using \"\"",
1.98 schwarze 173: "missing function name, using \"\"",
1.38 schwarze 174: "empty head in list item",
175: "empty list item",
1.155 schwarze 176: "missing argument, using next line",
1.39 schwarze 177: "missing font type, using \\fR",
178: "unknown font type, using \\fR",
1.79 schwarze 179: "nothing follows prefix",
1.97 schwarze 180: "empty reference block",
1.128 schwarze 181: "missing section argument",
1.38 schwarze 182: "missing -std argument, adding it",
1.100 schwarze 183: "missing option string, using \"\"",
184: "missing resource identifier, using \"\"",
1.66 schwarze 185: "missing eqn box, using \"\"",
1.1 schwarze 186:
187: /* related to bad macro arguments */
188: "duplicate argument",
1.54 schwarze 189: "skipping duplicate argument",
1.41 schwarze 190: "skipping duplicate display type",
191: "skipping duplicate list type",
1.54 schwarze 192: "skipping -width argument",
1.101 schwarze 193: "wrong number of cells",
1.1 schwarze 194: "unknown AT&T UNIX version",
1.64 schwarze 195: "comma in function argument",
1.65 schwarze 196: "parenthesis in function name",
1.145 schwarze 197: "unknown library name",
1.45 schwarze 198: "invalid content in Rs block",
1.41 schwarze 199: "invalid Boolean argument",
1.172 schwarze 200: "argument contains two font escapes",
1.41 schwarze 201: "unknown font, skipping request",
1.101 schwarze 202: "odd number of characters in request",
1.1 schwarze 203:
204: /* related to plain text */
1.42 schwarze 205: "blank line in fill mode, using .sp",
206: "tab in filled text",
1.131 schwarze 207: "new sentence, new line",
1.42 schwarze 208: "invalid escape sequence",
209: "undefined string, using \"\"",
1.3 schwarze 210:
1.93 schwarze 211: /* related to tables */
212: "tbl line starts with span",
213: "tbl column starts with span",
214: "skipping vertical bar in tbl layout",
215:
1.1 schwarze 216: "generic error",
1.3 schwarze 217:
1.1 schwarze 218: /* related to tables */
1.91 schwarze 219: "non-alphabetic character in tbl options",
220: "skipping unknown tbl option",
221: "missing tbl option argument",
222: "wrong tbl option argument size",
1.93 schwarze 223: "empty tbl layout",
224: "invalid character in tbl layout",
225: "unmatched parenthesis in tbl layout",
1.94 schwarze 226: "tbl without any data cells",
227: "ignoring data in spanned tbl cell",
228: "ignoring extra tbl data cells",
229: "data block open at end of tbl",
1.1 schwarze 230:
1.46 schwarze 231: /* related to document structure and macros */
1.83 schwarze 232: NULL,
1.161 schwarze 233: "duplicate prologue macro",
234: "skipping late title macro",
1.1 schwarze 235: "input stack limit exceeded, infinite loop?",
236: "skipping bad character",
1.46 schwarze 237: "skipping unknown macro",
1.169 schwarze 238: "ignoring request outside macro",
1.87 schwarze 239: "skipping insecure request",
1.51 schwarze 240: "skipping item outside list",
1.46 schwarze 241: "skipping column outside column list",
242: "skipping end of block that is not open",
1.90 schwarze 243: "fewer RS blocks open, skipping",
1.46 schwarze 244: "inserting missing end of block",
245: "appending missing end of block",
246:
247: /* related to request and macro arguments */
1.1 schwarze 248: "escaped character not allowed in a name",
1.169 schwarze 249: "using macro argument outside macro",
250: "argument number is not numeric",
1.75 schwarze 251: "NOT IMPLEMENTED: Bd -file",
1.120 schwarze 252: "skipping display without arguments",
1.49 schwarze 253: "missing list type, using -item",
1.143 schwarze 254: "argument is not numeric, using 1",
1.172 schwarze 255: "argument is not a character",
1.48 schwarze 256: "missing manual name, using \"\"",
1.49 schwarze 257: "uname(3) system call failed, using UNKNOWN",
1.41 schwarze 258: "unknown standard specifier",
1.49 schwarze 259: "skipping request without numeric argument",
1.169 schwarze 260: "excessive shift",
1.84 schwarze 261: "NOT IMPLEMENTED: .so with absolute path or \"..\"",
1.85 schwarze 262: ".so request failed",
1.39 schwarze 263: "skipping all arguments",
264: "skipping excess arguments",
1.68 schwarze 265: "divide by zero",
1.87 schwarze 266:
267: "unsupported feature",
268: "input too large",
1.89 schwarze 269: "unsupported control character",
1.87 schwarze 270: "unsupported roff request",
1.171 schwarze 271: "nested .while loops",
272: "end of scope with open .while loop",
273: "end of .while loop in inner scope",
274: "cannot continue this .while loop",
1.94 schwarze 275: "eqn delim option in tbl",
1.93 schwarze 276: "unsupported tbl layout modifier",
1.87 schwarze 277: "ignoring macro in table",
1.1 schwarze 278: };
279:
280: static const char * const mandoclevels[MANDOCLEVEL_MAX] = {
281: "SUCCESS",
1.137 schwarze 282: "STYLE",
1.1 schwarze 283: "WARNING",
284: "ERROR",
1.87 schwarze 285: "UNSUPP",
1.1 schwarze 286: "BADARG",
287: "SYSERR"
288: };
289:
1.25 schwarze 290:
1.1 schwarze 291: static void
292: resize_buf(struct buf *buf, size_t initial)
293: {
294:
295: buf->sz = buf->sz > initial/2 ? 2 * buf->sz : initial;
296: buf->buf = mandoc_realloc(buf->buf, buf->sz);
297: }
298:
299: static void
1.170 schwarze 300: free_buf_list(struct buf *buf)
301: {
302: struct buf *tmp;
303:
304: while (buf != NULL) {
305: tmp = buf;
306: buf = tmp->next;
307: free(tmp->buf);
308: free(tmp);
309: }
310: }
311:
312: static void
1.60 schwarze 313: choose_parser(struct mparse *curp)
1.1 schwarze 314: {
1.59 schwarze 315: char *cp, *ep;
316: int format;
1.1 schwarze 317:
1.59 schwarze 318: /*
319: * If neither command line arguments -mdoc or -man select
320: * a parser nor the roff parser found a .Dd or .TH macro
321: * yet, look ahead in the main input buffer.
322: */
323:
324: if ((format = roff_getformat(curp->roff)) == 0) {
325: cp = curp->primary->buf;
326: ep = cp + curp->primary->sz;
327: while (cp < ep) {
1.61 schwarze 328: if (*cp == '.' || *cp == '\'') {
1.59 schwarze 329: cp++;
330: if (cp[0] == 'D' && cp[1] == 'd') {
331: format = MPARSE_MDOC;
332: break;
333: }
334: if (cp[0] == 'T' && cp[1] == 'H') {
335: format = MPARSE_MAN;
336: break;
337: }
338: }
339: cp = memchr(cp, '\n', ep - cp);
340: if (cp == NULL)
341: break;
342: cp++;
343: }
1.1 schwarze 344: }
345:
1.59 schwarze 346: if (format == MPARSE_MDOC) {
1.112 schwarze 347: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 348: if (curp->man->mdocmac == NULL)
349: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 350: } else {
351: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 352: if (curp->man->manmac == NULL)
353: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.25 schwarze 354: }
1.136 schwarze 355: curp->man->first->tok = TOKEN_NONE;
1.1 schwarze 356: }
357:
358: /*
1.71 schwarze 359: * Main parse routine for a buffer.
360: * It assumes encoding and line numbering are already set up.
361: * It can recurse directly (for invocations of user-defined
362: * macros, inline equations, and input line traps)
363: * and indirectly (for .so file inclusion).
1.1 schwarze 364: */
1.171 schwarze 365: static int
1.71 schwarze 366: mparse_buf_r(struct mparse *curp, struct buf blk, size_t i, int start)
1.1 schwarze 367: {
368: struct buf ln;
1.171 schwarze 369: struct buf *firstln, *lastln, *thisln, *loop;
1.88 schwarze 370: const char *save_file;
1.85 schwarze 371: char *cp;
1.71 schwarze 372: size_t pos; /* byte number in the ln buffer */
1.171 schwarze 373: int line_result, result;
1.76 schwarze 374: int of;
1.1 schwarze 375: int lnn; /* line number in the real file */
1.85 schwarze 376: int fd;
1.171 schwarze 377: int inloop; /* Saw .while on this level. */
1.1 schwarze 378: unsigned char c;
379:
1.170 schwarze 380: ln.sz = 256;
381: ln.buf = mandoc_malloc(ln.sz);
382: ln.next = NULL;
1.171 schwarze 383: firstln = loop = NULL;
1.25 schwarze 384: lnn = curp->line;
385: pos = 0;
1.171 schwarze 386: inloop = 0;
1.170 schwarze 387: result = ROFF_CONT;
1.1 schwarze 388:
1.171 schwarze 389: while (i < blk.sz && (blk.buf[i] != '\0' || pos != 0)) {
1.1 schwarze 390: if (start) {
391: curp->line = lnn;
392: curp->reparse_count = 0;
1.70 schwarze 393:
394: if (lnn < 3 &&
395: curp->filenc & MPARSE_UTF8 &&
1.71 schwarze 396: curp->filenc & MPARSE_LATIN1)
397: curp->filenc = preconv_cue(&blk, i);
1.1 schwarze 398: }
399:
1.71 schwarze 400: while (i < blk.sz && (start || blk.buf[i] != '\0')) {
1.1 schwarze 401:
402: /*
403: * When finding an unescaped newline character,
404: * leave the character loop to process the line.
405: * Skip a preceding carriage return, if any.
406: */
407:
1.71 schwarze 408: if ('\r' == blk.buf[i] && i + 1 < blk.sz &&
1.1 schwarze 409: '\n' == blk.buf[i + 1])
410: ++i;
411: if ('\n' == blk.buf[i]) {
412: ++i;
413: ++lnn;
414: break;
415: }
416:
1.13 schwarze 417: /*
1.70 schwarze 418: * Make sure we have space for the worst
1.170 schwarze 419: * case of 12 bytes: "\\[u10ffff]\n\0"
1.13 schwarze 420: */
421:
1.170 schwarze 422: if (pos + 12 > ln.sz)
1.13 schwarze 423: resize_buf(&ln, 256);
424:
1.25 schwarze 425: /*
1.70 schwarze 426: * Encode 8-bit input.
1.1 schwarze 427: */
428:
1.70 schwarze 429: c = blk.buf[i];
430: if (c & 0x80) {
1.71 schwarze 431: if ( ! (curp->filenc && preconv_encode(
432: &blk, &i, &ln, &pos, &curp->filenc))) {
1.89 schwarze 433: mandoc_vmsg(MANDOCERR_CHAR_BAD, curp,
434: curp->line, pos, "0x%x", c);
1.70 schwarze 435: ln.buf[pos++] = '?';
436: i++;
437: }
438: continue;
439: }
440:
441: /*
442: * Exclude control characters.
443: */
1.1 schwarze 444:
1.70 schwarze 445: if (c == 0x7f || (c < 0x20 && c != 0x09)) {
1.89 schwarze 446: mandoc_vmsg(c == 0x00 || c == 0x04 ||
447: c > 0x0a ? MANDOCERR_CHAR_BAD :
448: MANDOCERR_CHAR_UNSUPP,
449: curp, curp->line, pos, "0x%x", c);
1.1 schwarze 450: i++;
1.102 schwarze 451: if (c != '\r')
452: ln.buf[pos++] = '?';
1.1 schwarze 453: continue;
454: }
455:
456: ln.buf[pos++] = blk.buf[i++];
457: }
1.170 schwarze 458: ln.buf[pos] = '\0';
459:
460: /*
461: * Maintain a lookaside buffer of all lines.
462: * parsed from this input source.
463: */
464:
465: thisln = mandoc_malloc(sizeof(*thisln));
466: thisln->buf = mandoc_strdup(ln.buf);
467: thisln->sz = strlen(ln.buf) + 1;
468: thisln->next = NULL;
469: if (firstln == NULL) {
470: firstln = lastln = thisln;
471: if (curp->secondary == NULL)
472: curp->secondary = firstln;
473: } else {
474: lastln->next = thisln;
475: lastln = thisln;
476: }
1.1 schwarze 477:
1.170 schwarze 478: /* XXX Ugly hack to mark the end of the input. */
1.1 schwarze 479:
1.170 schwarze 480: if (i == blk.sz || blk.buf[i] == '\0') {
1.142 schwarze 481: ln.buf[pos++] = '\n';
1.170 schwarze 482: ln.buf[pos] = '\0';
483: }
1.1 schwarze 484:
485: /*
486: * A significant amount of complexity is contained by
487: * the roff preprocessor. It's line-oriented but can be
488: * expressed on one line, so we need at times to
489: * readjust our starting point and re-run it. The roff
490: * preprocessor can also readjust the buffers with new
491: * data, so we pass them in wholesale.
492: */
493:
494: of = 0;
495: rerun:
1.169 schwarze 496: line_result = roff_parseln(curp->roff, curp->line, &ln, &of);
1.1 schwarze 497:
1.171 schwarze 498: /* Process options. */
499:
500: if (line_result & ROFF_APPEND)
501: assert(line_result == (ROFF_IGN | ROFF_APPEND));
502:
503: if (line_result & ROFF_USERCALL)
504: assert((line_result & ROFF_MASK) == ROFF_REPARSE);
505:
506: if (line_result & ROFF_USERRET) {
507: assert(line_result == (ROFF_IGN | ROFF_USERRET));
508: if (start == 0) {
509: /* Return from the current macro. */
510: result = ROFF_USERRET;
511: goto out;
512: }
513: }
514:
515: switch (line_result & ROFF_LOOPMASK) {
516: case ROFF_IGN:
517: break;
518: case ROFF_WHILE:
519: if (curp->loop != NULL) {
520: if (loop == curp->loop)
521: break;
522: mandoc_msg(MANDOCERR_WHILE_NEST,
523: curp, curp->line, pos, NULL);
524: }
525: curp->loop = thisln;
526: loop = NULL;
527: inloop = 1;
528: break;
529: case ROFF_LOOPCONT:
530: case ROFF_LOOPEXIT:
531: if (curp->loop == NULL) {
532: mandoc_msg(MANDOCERR_WHILE_FAIL,
533: curp, curp->line, pos, NULL);
534: break;
535: }
536: if (inloop == 0) {
537: mandoc_msg(MANDOCERR_WHILE_INTO,
538: curp, curp->line, pos, NULL);
539: curp->loop = loop = NULL;
540: break;
541: }
542: if (line_result & ROFF_LOOPCONT)
543: loop = curp->loop;
544: else {
545: curp->loop = loop = NULL;
546: inloop = 0;
547: }
548: break;
549: default:
550: abort();
551: }
552:
553: /* Process the main instruction from the roff parser. */
554:
555: switch (line_result & ROFF_MASK) {
556: case ROFF_IGN:
557: break;
558: case ROFF_CONT:
559: if (curp->man->macroset == MACROSET_NONE)
560: choose_parser(curp);
561: if ((curp->man->macroset == MACROSET_MDOC ?
562: mdoc_parseln(curp->man, curp->line, ln.buf, of) :
563: man_parseln(curp->man, curp->line, ln.buf, of)
564: ) == 2)
565: goto out;
566: break;
567: case ROFF_RERUN:
568: goto rerun;
1.25 schwarze 569: case ROFF_REPARSE:
1.169 schwarze 570: if (++curp->reparse_count > REPARSE_LIMIT) {
1.171 schwarze 571: /* Abort and return to the top level. */
1.170 schwarze 572: result = ROFF_IGN;
1.1 schwarze 573: mandoc_msg(MANDOCERR_ROFFLOOP, curp,
1.25 schwarze 574: curp->line, pos, NULL);
1.171 schwarze 575: goto out;
1.169 schwarze 576: }
1.171 schwarze 577: result = mparse_buf_r(curp, ln, of, 0);
578: if (line_result & ROFF_USERCALL) {
579: roff_userret(curp->roff);
580: /* Continue normally. */
581: if (result & ROFF_USERRET)
582: result = ROFF_CONT;
1.135 schwarze 583: }
1.171 schwarze 584: if (start == 0 && result != ROFF_CONT)
585: goto out;
586: break;
1.25 schwarze 587: case ROFF_SO:
1.71 schwarze 588: if ( ! (curp->options & MPARSE_SO) &&
589: (i >= blk.sz || blk.buf[i] == '\0')) {
1.23 schwarze 590: curp->sodest = mandoc_strdup(ln.buf + of);
1.170 schwarze 591: goto out;
1.23 schwarze 592: }
1.88 schwarze 593: save_file = curp->file;
1.122 schwarze 594: if ((fd = mparse_open(curp, ln.buf + of)) != -1) {
1.85 schwarze 595: mparse_readfd(curp, fd, ln.buf + of);
1.121 schwarze 596: close(fd);
1.88 schwarze 597: curp->file = save_file;
598: } else {
599: curp->file = save_file;
1.30 schwarze 600: mandoc_vmsg(MANDOCERR_SO_FAIL,
601: curp, curp->line, pos,
602: ".so %s", ln.buf + of);
1.85 schwarze 603: ln.sz = mandoc_asprintf(&cp,
604: ".sp\nSee the file %s.\n.sp",
605: ln.buf + of);
606: free(ln.buf);
607: ln.buf = cp;
608: of = 0;
609: mparse_buf_r(curp, ln, of, 0);
1.30 schwarze 610: }
1.171 schwarze 611: break;
1.1 schwarze 612: default:
1.171 schwarze 613: abort();
1.1 schwarze 614: }
615:
1.171 schwarze 616: /* Start the next input line. */
1.1 schwarze 617:
1.171 schwarze 618: if (loop != NULL &&
619: (line_result & ROFF_LOOPMASK) == ROFF_IGN)
620: loop = loop->next;
621:
622: if (loop != NULL) {
623: if ((line_result & ROFF_APPEND) == 0)
624: *ln.buf = '\0';
625: if (ln.sz < loop->sz)
626: resize_buf(&ln, loop->sz);
627: (void)strlcat(ln.buf, loop->buf, ln.sz);
628: of = 0;
629: goto rerun;
630: }
1.1 schwarze 631:
1.171 schwarze 632: pos = (line_result & ROFF_APPEND) ? strlen(ln.buf) : 0;
1.1 schwarze 633: }
1.170 schwarze 634: out:
1.171 schwarze 635: if (inloop) {
636: if (result != ROFF_USERRET)
637: mandoc_msg(MANDOCERR_WHILE_OUTOF, curp,
638: curp->line, pos, NULL);
639: curp->loop = NULL;
640: }
1.1 schwarze 641: free(ln.buf);
1.170 schwarze 642: if (firstln != curp->secondary)
643: free_buf_list(firstln);
644: return result;
1.1 schwarze 645: }
646:
647: static int
1.18 schwarze 648: read_whole_file(struct mparse *curp, const char *file, int fd,
649: struct buf *fb, int *with_mmap)
1.1 schwarze 650: {
651: struct stat st;
1.115 schwarze 652: gzFile gz;
1.1 schwarze 653: size_t off;
654: ssize_t ssz;
1.166 schwarze 655: int gzerrnum, retval;
1.1 schwarze 656:
1.164 schwarze 657: if (fstat(fd, &st) == -1) {
658: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
659: "fstat: %s", strerror(errno));
660: return 0;
661: }
1.1 schwarze 662:
663: /*
664: * If we're a regular file, try just reading in the whole entry
665: * via mmap(). This is faster than reading it into blocks, and
666: * since each file is only a few bytes to begin with, I'm not
667: * concerned that this is going to tank any machines.
668: */
669:
1.115 schwarze 670: if (curp->gzip == 0 && S_ISREG(st.st_mode)) {
1.106 schwarze 671: if (st.st_size > 0x7fffffff) {
1.86 schwarze 672: mandoc_msg(MANDOCERR_TOOLARGE, curp, 0, 0, NULL);
1.117 schwarze 673: return 0;
1.1 schwarze 674: }
675: *with_mmap = 1;
676: fb->sz = (size_t)st.st_size;
1.15 schwarze 677: fb->buf = mmap(NULL, fb->sz, PROT_READ, MAP_SHARED, fd, 0);
1.1 schwarze 678: if (fb->buf != MAP_FAILED)
1.117 schwarze 679: return 1;
1.1 schwarze 680: }
681:
1.115 schwarze 682: if (curp->gzip) {
1.166 schwarze 683: /*
684: * Duplicating the file descriptor is required
685: * because we will have to call gzclose(3)
686: * to free memory used internally by zlib,
687: * but that will also close the file descriptor,
688: * which this function must not do.
689: */
690: if ((fd = dup(fd)) == -1) {
691: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
692: "dup: %s", strerror(errno));
693: return 0;
694: }
1.164 schwarze 695: if ((gz = gzdopen(fd, "rb")) == NULL) {
696: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0,
697: "gzdopen: %s", strerror(errno));
1.166 schwarze 698: close(fd);
1.164 schwarze 699: return 0;
700: }
1.115 schwarze 701: } else
702: gz = NULL;
703:
1.1 schwarze 704: /*
705: * If this isn't a regular file (like, say, stdin), then we must
706: * go the old way and just read things in bit by bit.
707: */
708:
709: *with_mmap = 0;
710: off = 0;
1.166 schwarze 711: retval = 0;
1.1 schwarze 712: fb->sz = 0;
713: fb->buf = NULL;
714: for (;;) {
715: if (off == fb->sz) {
716: if (fb->sz == (1U << 31)) {
1.86 schwarze 717: mandoc_msg(MANDOCERR_TOOLARGE, curp,
718: 0, 0, NULL);
1.1 schwarze 719: break;
720: }
721: resize_buf(fb, 65536);
722: }
1.115 schwarze 723: ssz = curp->gzip ?
724: gzread(gz, fb->buf + (int)off, fb->sz - off) :
725: read(fd, fb->buf + (int)off, fb->sz - off);
1.1 schwarze 726: if (ssz == 0) {
727: fb->sz = off;
1.166 schwarze 728: retval = 1;
729: break;
1.1 schwarze 730: }
1.164 schwarze 731: if (ssz == -1) {
1.166 schwarze 732: if (curp->gzip)
733: (void)gzerror(gz, &gzerrnum);
734: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "read: %s",
735: curp->gzip && gzerrnum != Z_ERRNO ?
736: zError(gzerrnum) : strerror(errno));
1.164 schwarze 737: break;
738: }
1.1 schwarze 739: off += (size_t)ssz;
740: }
741:
1.166 schwarze 742: if (curp->gzip && (gzerrnum = gzclose(gz)) != Z_OK)
743: mandoc_vmsg(MANDOCERR_FILE, curp, 0, 0, "gzclose: %s",
744: gzerrnum == Z_ERRNO ? strerror(errno) :
745: zError(gzerrnum));
746: if (retval == 0) {
747: free(fb->buf);
748: fb->buf = NULL;
749: }
750: return retval;
1.1 schwarze 751: }
752:
753: static void
754: mparse_end(struct mparse *curp)
755: {
1.110 schwarze 756: if (curp->man->macroset == MACROSET_NONE)
757: curp->man->macroset = MACROSET_MAN;
758: if (curp->man->macroset == MACROSET_MDOC)
759: mdoc_endparse(curp->man);
760: else
1.86 schwarze 761: man_endparse(curp->man);
1.1 schwarze 762: roff_endparse(curp->roff);
763: }
764:
1.15 schwarze 765: static void
766: mparse_parse_buffer(struct mparse *curp, struct buf blk, const char *file)
1.1 schwarze 767: {
1.61 schwarze 768: struct buf *svprimary;
1.1 schwarze 769: const char *svfile;
1.71 schwarze 770: size_t offset;
1.14 schwarze 771: static int recursion_depth;
772:
773: if (64 < recursion_depth) {
774: mandoc_msg(MANDOCERR_ROFFLOOP, curp, curp->line, 0, NULL);
1.15 schwarze 775: return;
1.14 schwarze 776: }
1.1 schwarze 777:
1.15 schwarze 778: /* Line number is per-file. */
779: svfile = curp->file;
780: curp->file = file;
1.61 schwarze 781: svprimary = curp->primary;
1.59 schwarze 782: curp->primary = &blk;
1.15 schwarze 783: curp->line = 1;
784: recursion_depth++;
785:
1.70 schwarze 786: /* Skip an UTF-8 byte order mark. */
787: if (curp->filenc & MPARSE_UTF8 && blk.sz > 2 &&
788: (unsigned char)blk.buf[0] == 0xef &&
789: (unsigned char)blk.buf[1] == 0xbb &&
790: (unsigned char)blk.buf[2] == 0xbf) {
1.71 schwarze 791: offset = 3;
1.70 schwarze 792: curp->filenc &= ~MPARSE_LATIN1;
1.71 schwarze 793: } else
794: offset = 0;
1.70 schwarze 795:
1.71 schwarze 796: mparse_buf_r(curp, blk, offset, 1);
1.15 schwarze 797:
1.86 schwarze 798: if (--recursion_depth == 0)
1.15 schwarze 799: mparse_end(curp);
800:
1.61 schwarze 801: curp->primary = svprimary;
1.15 schwarze 802: curp->file = svfile;
803: }
804:
1.74 schwarze 805: /*
806: * Read the whole file into memory and call the parsers.
807: * Called recursively when an .so request is encountered.
808: */
1.15 schwarze 809: enum mandoclevel
810: mparse_readfd(struct mparse *curp, int fd, const char *file)
811: {
812: struct buf blk;
813: int with_mmap;
1.70 schwarze 814: int save_filenc;
1.1 schwarze 815:
1.67 schwarze 816: if (read_whole_file(curp, file, fd, &blk, &with_mmap)) {
1.70 schwarze 817: save_filenc = curp->filenc;
818: curp->filenc = curp->options &
819: (MPARSE_UTF8 | MPARSE_LATIN1);
1.67 schwarze 820: mparse_parse_buffer(curp, blk, file);
1.70 schwarze 821: curp->filenc = save_filenc;
1.67 schwarze 822: if (with_mmap)
823: munmap(blk.buf, blk.sz);
824: else
825: free(blk.buf);
826: }
1.117 schwarze 827: return curp->file_status;
1.58 schwarze 828: }
829:
1.122 schwarze 830: int
831: mparse_open(struct mparse *curp, const char *file)
1.58 schwarze 832: {
833: char *cp;
1.122 schwarze 834: int fd;
1.58 schwarze 835:
836: curp->file = file;
1.115 schwarze 837: cp = strrchr(file, '.');
838: curp->gzip = (cp != NULL && ! strcmp(cp + 1, "gz"));
1.74 schwarze 839:
1.115 schwarze 840: /* First try to use the filename as it is. */
1.74 schwarze 841:
1.122 schwarze 842: if ((fd = open(file, O_RDONLY)) != -1)
843: return fd;
1.74 schwarze 844:
1.115 schwarze 845: /*
846: * If that doesn't work and the filename doesn't
847: * already end in .gz, try appending .gz.
848: */
1.74 schwarze 849:
1.115 schwarze 850: if ( ! curp->gzip) {
1.74 schwarze 851: mandoc_asprintf(&cp, "%s.gz", file);
1.123 schwarze 852: fd = open(cp, O_RDONLY);
1.83 schwarze 853: free(cp);
1.122 schwarze 854: if (fd != -1) {
1.115 schwarze 855: curp->gzip = 1;
1.122 schwarze 856: return fd;
1.58 schwarze 857: }
858: }
859:
1.115 schwarze 860: /* Neither worked, give up. */
1.73 schwarze 861:
1.115 schwarze 862: mandoc_msg(MANDOCERR_FILE, curp, 0, 0, strerror(errno));
1.122 schwarze 863: return -1;
1.1 schwarze 864: }
865:
866: struct mparse *
1.151 schwarze 867: mparse_alloc(int options, enum mandocerr mmin, mandocmsg mmsg,
868: enum mandoc_os os_e, const char *os_s)
1.1 schwarze 869: {
870: struct mparse *curp;
871:
872: curp = mandoc_calloc(1, sizeof(struct mparse));
873:
1.22 schwarze 874: curp->options = options;
1.151 schwarze 875: curp->mmin = mmin;
1.1 schwarze 876: curp->mmsg = mmsg;
1.151 schwarze 877: curp->os_s = os_s;
1.1 schwarze 878:
1.119 schwarze 879: curp->roff = roff_alloc(curp, options);
1.151 schwarze 880: curp->man = roff_man_alloc(curp->roff, curp, curp->os_s,
1.112 schwarze 881: curp->options & MPARSE_QUICK ? 1 : 0);
1.111 schwarze 882: if (curp->options & MPARSE_MDOC) {
1.112 schwarze 883: curp->man->macroset = MACROSET_MDOC;
1.136 schwarze 884: if (curp->man->mdocmac == NULL)
885: curp->man->mdocmac = roffhash_alloc(MDOC_Dd, MDOC_MAX);
1.112 schwarze 886: } else if (curp->options & MPARSE_MAN) {
887: curp->man->macroset = MACROSET_MAN;
1.136 schwarze 888: if (curp->man->manmac == NULL)
889: curp->man->manmac = roffhash_alloc(MAN_TH, MAN_MAX);
1.111 schwarze 890: }
1.113 schwarze 891: curp->man->first->tok = TOKEN_NONE;
1.151 schwarze 892: curp->man->meta.os_e = os_e;
1.117 schwarze 893: return curp;
1.1 schwarze 894: }
895:
896: void
897: mparse_reset(struct mparse *curp)
898: {
899: roff_reset(curp->roff);
1.124 schwarze 900: roff_man_reset(curp->man);
1.170 schwarze 901: free_buf_list(curp->secondary);
902: curp->secondary = NULL;
1.133 schwarze 903:
904: free(curp->sodest);
905: curp->sodest = NULL;
906:
1.1 schwarze 907: curp->file_status = MANDOCLEVEL_OK;
1.132 schwarze 908: curp->gzip = 0;
1.1 schwarze 909: }
910:
911: void
912: mparse_free(struct mparse *curp)
913: {
1.136 schwarze 914: roffhash_free(curp->man->mdocmac);
915: roffhash_free(curp->man->manmac);
1.112 schwarze 916: roff_man_free(curp->man);
1.133 schwarze 917: roff_free(curp->roff);
1.170 schwarze 918: free_buf_list(curp->secondary);
1.23 schwarze 919: free(curp->sodest);
1.1 schwarze 920: free(curp);
921: }
922:
923: void
1.110 schwarze 924: mparse_result(struct mparse *curp, struct roff_man **man,
925: char **sodest)
1.1 schwarze 926: {
927:
1.23 schwarze 928: if (sodest && NULL != (*sodest = curp->sodest)) {
929: *man = NULL;
930: return;
931: }
1.1 schwarze 932: if (man)
933: *man = curp->man;
1.130 schwarze 934: }
935:
936: void
937: mparse_updaterc(struct mparse *curp, enum mandoclevel *rc)
938: {
939: if (curp->file_status > *rc)
940: *rc = curp->file_status;
1.1 schwarze 941: }
942:
943: void
944: mandoc_vmsg(enum mandocerr t, struct mparse *m,
945: int ln, int pos, const char *fmt, ...)
946: {
947: char buf[256];
948: va_list ap;
949:
950: va_start(ap, fmt);
1.26 schwarze 951: (void)vsnprintf(buf, sizeof(buf), fmt, ap);
1.1 schwarze 952: va_end(ap);
953:
954: mandoc_msg(t, m, ln, pos, buf);
955: }
956:
957: void
1.25 schwarze 958: mandoc_msg(enum mandocerr er, struct mparse *m,
1.1 schwarze 959: int ln, int col, const char *msg)
960: {
961: enum mandoclevel level;
962:
1.151 schwarze 963: if (er < m->mmin && er != MANDOCERR_FILE)
964: return;
965:
1.87 schwarze 966: level = MANDOCLEVEL_UNSUPP;
1.1 schwarze 967: while (er < mandoclimits[level])
968: level--;
969:
970: if (m->mmsg)
971: (*m->mmsg)(er, level, m->file, ln, col, msg);
972:
973: if (m->file_status < level)
974: m->file_status = level;
975: }
976:
977: const char *
978: mparse_strerror(enum mandocerr er)
979: {
980:
1.117 schwarze 981: return mandocerrs[er];
1.1 schwarze 982: }
983:
984: const char *
985: mparse_strlevel(enum mandoclevel lvl)
986: {
1.117 schwarze 987: return mandoclevels[lvl];
1.4 schwarze 988: }
989:
990: void
1.170 schwarze 991: mparse_copy(const struct mparse *p)
1.4 schwarze 992: {
1.170 schwarze 993: struct buf *buf;
1.4 schwarze 994:
1.170 schwarze 995: for (buf = p->secondary; buf != NULL; buf = buf->next)
996: puts(buf->buf);
1.1 schwarze 997: }