Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.193
1.193 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.192 2017/02/09 17:19:07 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.184 schwarze 4: * Copyright (c) 2011-2017 Ingo Schwarze <schwarze@openbsd.org>
1.180 schwarze 5: * Copyright (c) 2016 Ed Maste <emaste@freebsd.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: *
1.140 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
1.140 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: */
1.117 schwarze 19: #include <sys/types.h>
1.47 schwarze 20: #include <sys/stat.h>
1.72 schwarze 21: #include <sys/wait.h>
1.1 schwarze 22:
23: #include <assert.h>
1.33 schwarze 24: #include <ctype.h>
1.151 schwarze 25: #include <err.h>
1.34 schwarze 26: #include <errno.h>
1.1 schwarze 27: #include <fcntl.h>
1.47 schwarze 28: #include <fts.h>
1.44 schwarze 29: #include <limits.h>
1.173 schwarze 30: #include <stdarg.h>
1.47 schwarze 31: #include <stddef.h>
1.1 schwarze 32: #include <stdio.h>
33: #include <stdint.h>
34: #include <stdlib.h>
35: #include <string.h>
1.14 schwarze 36: #include <unistd.h>
1.1 schwarze 37:
1.141 schwarze 38: #include "mandoc_aux.h"
1.156 schwarze 39: #include "mandoc_ohash.h"
1.141 schwarze 40: #include "mandoc.h"
41: #include "roff.h"
1.47 schwarze 42: #include "mdoc.h"
1.1 schwarze 43: #include "man.h"
1.140 schwarze 44: #include "manconf.h"
1.47 schwarze 45: #include "mansearch.h"
1.173 schwarze 46: #include "dba_array.h"
47: #include "dba.h"
1.1 schwarze 48:
1.68 schwarze 49: extern const char *const mansearch_keynames[];
50:
1.47 schwarze 51: enum op {
52: OP_DEFAULT = 0, /* new dbs from dir list or default config */
53: OP_CONFFILE, /* new databases from custom config file */
54: OP_UPDATE, /* delete/add entries in existing database */
55: OP_DELETE, /* delete entries from existing database */
56: OP_TEST /* change no databases, report potential problems */
57: };
1.11 schwarze 58:
1.47 schwarze 59: struct str {
60: const struct mpage *mpage; /* if set, the owning parse */
61: uint64_t mask; /* bitmask in sequence */
1.131 schwarze 62: char key[]; /* rendered text */
1.28 schwarze 63: };
64:
1.47 schwarze 65: struct inodev {
66: ino_t st_ino;
67: dev_t st_dev;
68: };
1.28 schwarze 69:
1.47 schwarze 70: struct mpage {
71: struct inodev inodev; /* used for hashing routine */
1.173 schwarze 72: struct dba_array *dba;
1.47 schwarze 73: char *sec; /* section from file content */
74: char *arch; /* architecture from file content */
75: char *title; /* title from file content */
76: char *desc; /* description from file content */
1.180 schwarze 77: struct mpage *next; /* singly linked list */
1.47 schwarze 78: struct mlink *mlinks; /* singly linked list */
1.129 schwarze 79: int name_head_done;
1.173 schwarze 80: enum form form; /* format from file content */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct mlink {
84: char file[PATH_MAX]; /* filename rel. to manpath */
85: char *dsec; /* section from directory */
86: char *arch; /* architecture from directory */
87: char *name; /* name from file name (not empty) */
88: char *fsec; /* section from file name suffix */
89: struct mlink *next; /* singly linked list */
1.75 schwarze 90: struct mpage *mpage; /* parent */
1.81 schwarze 91: int gzip; /* filename has a .gz suffix */
1.173 schwarze 92: enum form dform; /* format from directory */
93: enum form fform; /* format from file name suffix */
1.1 schwarze 94: };
95:
1.143 schwarze 96: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.142 schwarze 97: const struct roff_node *);
1.1 schwarze 98:
1.19 schwarze 99: struct mdoc_handler {
1.47 schwarze 100: mdoc_fp fp; /* optional handler */
101: uint64_t mask; /* set unless handler returns 0 */
1.184 schwarze 102: int taboo; /* node flags that must not be set */
1.19 schwarze 103: };
1.171 schwarze 104:
105:
106: int mandocdb(int, char *[]);
1.19 schwarze 107:
1.173 schwarze 108: static void dbadd(struct dba *, struct mpage *);
1.75 schwarze 109: static void dbadd_mlink(const struct mlink *mlink);
1.173 schwarze 110: static void dbprune(struct dba *);
111: static void dbwrite(struct dba *);
1.47 schwarze 112: static void filescan(const char *);
1.180 schwarze 113: static int fts_compare(const FTSENT **, const FTSENT **);
1.47 schwarze 114: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 115: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 116: static void mlink_free(struct mlink *);
117: static void mlinks_undupe(struct mpage *);
118: static void mpages_free(void);
1.173 schwarze 119: static void mpages_merge(struct dba *, struct mparse *);
1.81 schwarze 120: static void parse_cat(struct mpage *, int);
1.143 schwarze 121: static void parse_man(struct mpage *, const struct roff_meta *,
1.142 schwarze 122: const struct roff_node *);
1.143 schwarze 123: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.142 schwarze 124: const struct roff_node *);
1.143 schwarze 125: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.142 schwarze 126: const struct roff_node *);
1.143 schwarze 127: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.142 schwarze 128: const struct roff_node *);
129: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.143 schwarze 130: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.142 schwarze 131: const struct roff_node *);
1.143 schwarze 132: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.142 schwarze 133: const struct roff_node *);
1.143 schwarze 134: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.142 schwarze 135: const struct roff_node *);
1.143 schwarze 136: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.142 schwarze 137: const struct roff_node *);
1.143 schwarze 138: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.142 schwarze 139: const struct roff_node *);
1.164 schwarze 140: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
141: const struct roff_node *);
1.143 schwarze 142: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.142 schwarze 143: const struct roff_node *);
1.69 schwarze 144: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 145: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 146: static void putmdockey(const struct mpage *,
1.184 schwarze 147: const struct roff_node *, uint64_t, int);
1.131 schwarze 148: static int render_string(char **, size_t *);
1.172 schwarze 149: static void say(const char *, const char *, ...)
1.192 schwarze 150: __attribute__((__format__ (__printf__, 2, 3)));
1.120 schwarze 151: static int set_basedir(const char *, int);
1.47 schwarze 152: static int treescan(void);
153: static size_t utf8(unsigned int, char [7]);
154:
1.59 schwarze 155: static int nodb; /* no database changes */
1.73 schwarze 156: static int mparse_options; /* abort the parse early */
1.98 schwarze 157: static int use_all; /* use all found files */
158: static int debug; /* print what we're doing */
159: static int warnings; /* warn about crap */
1.52 schwarze 160: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 161: static int exitcode; /* to be returned by main */
1.98 schwarze 162: static enum op op; /* operational mode */
1.47 schwarze 163: static char basedir[PATH_MAX]; /* current base directory */
1.180 schwarze 164: static struct mpage *mpage_head; /* list of distinct manual pages */
1.47 schwarze 165: static struct ohash mpages; /* table of distinct manual pages */
166: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 167: static struct ohash names; /* table of all names */
1.47 schwarze 168: static struct ohash strings; /* table of all strings */
1.90 schwarze 169: static uint64_t name_mask;
1.47 schwarze 170:
1.19 schwarze 171: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.184 schwarze 172: { NULL, 0, 0 }, /* Ap */
173: { NULL, 0, NODE_NOPRT }, /* Dd */
174: { NULL, 0, NODE_NOPRT }, /* Dt */
175: { NULL, 0, NODE_NOPRT }, /* Os */
176: { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
177: { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
178: { NULL, 0, 0 }, /* Pp */
179: { NULL, 0, 0 }, /* D1 */
180: { NULL, 0, 0 }, /* Dl */
181: { NULL, 0, 0 }, /* Bd */
182: { NULL, 0, 0 }, /* Ed */
183: { NULL, 0, 0 }, /* Bl */
184: { NULL, 0, 0 }, /* El */
185: { NULL, 0, 0 }, /* It */
186: { NULL, 0, 0 }, /* Ad */
187: { NULL, TYPE_An, 0 }, /* An */
188: { NULL, TYPE_Ar, 0 }, /* Ar */
189: { NULL, TYPE_Cd, 0 }, /* Cd */
190: { NULL, TYPE_Cm, 0 }, /* Cm */
191: { NULL, TYPE_Dv, 0 }, /* Dv */
192: { NULL, TYPE_Er, 0 }, /* Er */
193: { NULL, TYPE_Ev, 0 }, /* Ev */
194: { NULL, 0, 0 }, /* Ex */
195: { NULL, TYPE_Fa, 0 }, /* Fa */
196: { parse_mdoc_Fd, 0, 0 }, /* Fd */
197: { NULL, TYPE_Fl, 0 }, /* Fl */
198: { parse_mdoc_Fn, 0, 0 }, /* Fn */
199: { NULL, TYPE_Ft, 0 }, /* Ft */
200: { NULL, TYPE_Ic, 0 }, /* Ic */
201: { NULL, TYPE_In, 0 }, /* In */
202: { NULL, TYPE_Li, 0 }, /* Li */
203: { parse_mdoc_Nd, 0, 0 }, /* Nd */
204: { parse_mdoc_Nm, 0, 0 }, /* Nm */
205: { NULL, 0, 0 }, /* Op */
206: { NULL, 0, 0 }, /* Ot */
207: { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
208: { NULL, 0, 0 }, /* Rv */
209: { NULL, TYPE_St, 0 }, /* St */
210: { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
211: { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
212: { parse_mdoc_Xr, 0, 0 }, /* Xr */
213: { NULL, 0, 0 }, /* %A */
214: { NULL, 0, 0 }, /* %B */
215: { NULL, 0, 0 }, /* %D */
216: { NULL, 0, 0 }, /* %I */
217: { NULL, 0, 0 }, /* %J */
218: { NULL, 0, 0 }, /* %N */
219: { NULL, 0, 0 }, /* %O */
220: { NULL, 0, 0 }, /* %P */
221: { NULL, 0, 0 }, /* %R */
222: { NULL, 0, 0 }, /* %T */
223: { NULL, 0, 0 }, /* %V */
224: { NULL, 0, 0 }, /* Ac */
225: { NULL, 0, 0 }, /* Ao */
226: { NULL, 0, 0 }, /* Aq */
1.186 schwarze 227: { NULL, TYPE_At, 0 }, /* At */
1.184 schwarze 228: { NULL, 0, 0 }, /* Bc */
229: { NULL, 0, 0 }, /* Bf */
230: { NULL, 0, 0 }, /* Bo */
231: { NULL, 0, 0 }, /* Bq */
232: { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
1.185 schwarze 233: { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
1.184 schwarze 234: { NULL, 0, 0 }, /* Db */
235: { NULL, 0, 0 }, /* Dc */
236: { NULL, 0, 0 }, /* Do */
237: { NULL, 0, 0 }, /* Dq */
238: { NULL, 0, 0 }, /* Ec */
239: { NULL, 0, 0 }, /* Ef */
240: { NULL, TYPE_Em, 0 }, /* Em */
241: { NULL, 0, 0 }, /* Eo */
242: { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
243: { NULL, TYPE_Ms, 0 }, /* Ms */
244: { NULL, 0, 0 }, /* No */
245: { NULL, 0, 0 }, /* Ns */
246: { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
247: { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
248: { NULL, 0, 0 }, /* Pc */
249: { NULL, 0, 0 }, /* Pf */
250: { NULL, 0, 0 }, /* Po */
251: { NULL, 0, 0 }, /* Pq */
252: { NULL, 0, 0 }, /* Qc */
253: { NULL, 0, 0 }, /* Ql */
254: { NULL, 0, 0 }, /* Qo */
255: { NULL, 0, 0 }, /* Qq */
256: { NULL, 0, 0 }, /* Re */
257: { NULL, 0, 0 }, /* Rs */
258: { NULL, 0, 0 }, /* Sc */
259: { NULL, 0, 0 }, /* So */
260: { NULL, 0, 0 }, /* Sq */
261: { NULL, 0, 0 }, /* Sm */
262: { NULL, 0, 0 }, /* Sx */
263: { NULL, TYPE_Sy, 0 }, /* Sy */
264: { NULL, TYPE_Tn, 0 }, /* Tn */
265: { NULL, 0, NODE_NOSRC }, /* Ux */
266: { NULL, 0, 0 }, /* Xc */
267: { NULL, 0, 0 }, /* Xo */
268: { parse_mdoc_Fo, 0, 0 }, /* Fo */
269: { NULL, 0, 0 }, /* Fc */
270: { NULL, 0, 0 }, /* Oo */
271: { NULL, 0, 0 }, /* Oc */
272: { NULL, 0, 0 }, /* Bk */
273: { NULL, 0, 0 }, /* Ek */
274: { NULL, 0, 0 }, /* Bt */
275: { NULL, 0, 0 }, /* Hf */
276: { NULL, 0, 0 }, /* Fr */
277: { NULL, 0, 0 }, /* Ud */
1.186 schwarze 278: { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
1.184 schwarze 279: { NULL, 0, 0 }, /* Lp */
280: { NULL, TYPE_Lk, 0 }, /* Lk */
281: { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
282: { NULL, 0, 0 }, /* Brq */
283: { NULL, 0, 0 }, /* Bro */
284: { NULL, 0, 0 }, /* Brc */
285: { NULL, 0, 0 }, /* %C */
286: { NULL, 0, 0 }, /* Es */
287: { NULL, 0, 0 }, /* En */
288: { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
289: { NULL, 0, 0 }, /* %Q */
290: { NULL, 0, 0 }, /* br */
291: { NULL, 0, 0 }, /* sp */
292: { NULL, 0, 0 }, /* %U */
293: { NULL, 0, 0 }, /* Ta */
294: { NULL, 0, 0 }, /* ll */
1.1 schwarze 295: };
296:
1.98 schwarze 297:
1.1 schwarze 298: int
1.3 schwarze 299: mandocdb(int argc, char *argv[])
1.1 schwarze 300: {
1.140 schwarze 301: struct manconf conf;
302: struct mparse *mp;
1.173 schwarze 303: struct dba *dba;
1.161 schwarze 304: const char *path_arg, *progname;
1.140 schwarze 305: size_t j, sz;
1.47 schwarze 306: int ch, i;
307:
1.158 millert 308: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.167 schwarze 309: warn("pledge");
1.155 schwarze 310: return (int)MANDOCLEVEL_SYSERR;
311: }
312:
1.140 schwarze 313: memset(&conf, 0, sizeof(conf));
1.47 schwarze 314:
315: /*
1.98 schwarze 316: * We accept a few different invocations.
1.47 schwarze 317: * The CHECKOP macro makes sure that invocation styles don't
318: * clobber each other.
319: */
320: #define CHECKOP(_op, _ch) do \
321: if (OP_DEFAULT != (_op)) { \
1.151 schwarze 322: warnx("-%c: Conflicting option", (_ch)); \
1.47 schwarze 323: goto usage; \
324: } while (/*CONSTCOND*/0)
1.10 schwarze 325:
1.47 schwarze 326: path_arg = NULL;
1.28 schwarze 327: op = OP_DEFAULT;
1.1 schwarze 328:
1.83 schwarze 329: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 330: switch (ch) {
1.98 schwarze 331: case 'a':
1.6 schwarze 332: use_all = 1;
333: break;
1.98 schwarze 334: case 'C':
1.47 schwarze 335: CHECKOP(op, ch);
336: path_arg = optarg;
1.28 schwarze 337: op = OP_CONFFILE;
1.25 schwarze 338: break;
1.98 schwarze 339: case 'D':
1.82 schwarze 340: debug++;
341: break;
1.98 schwarze 342: case 'd':
1.47 schwarze 343: CHECKOP(op, ch);
344: path_arg = optarg;
1.2 schwarze 345: op = OP_UPDATE;
1.1 schwarze 346: break;
1.98 schwarze 347: case 'n':
1.47 schwarze 348: nodb = 1;
349: break;
1.98 schwarze 350: case 'p':
1.83 schwarze 351: warnings = 1;
352: break;
1.98 schwarze 353: case 'Q':
1.73 schwarze 354: mparse_options |= MPARSE_QUICK;
1.59 schwarze 355: break;
1.98 schwarze 356: case 'T':
1.52 schwarze 357: if (strcmp(optarg, "utf8")) {
1.151 schwarze 358: warnx("-T%s: Unsupported output format",
359: optarg);
1.52 schwarze 360: goto usage;
361: }
362: write_utf8 = 1;
363: break;
1.98 schwarze 364: case 't':
1.47 schwarze 365: CHECKOP(op, ch);
1.28 schwarze 366: dup2(STDOUT_FILENO, STDERR_FILENO);
367: op = OP_TEST;
1.47 schwarze 368: nodb = warnings = 1;
1.28 schwarze 369: break;
1.98 schwarze 370: case 'u':
1.47 schwarze 371: CHECKOP(op, ch);
372: path_arg = optarg;
1.1 schwarze 373: op = OP_DELETE;
374: break;
1.98 schwarze 375: case 'v':
1.83 schwarze 376: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 377: break;
1.1 schwarze 378: default:
1.28 schwarze 379: goto usage;
1.1 schwarze 380: }
381:
382: argc -= optind;
383: argv += optind;
384:
1.163 schwarze 385: if (nodb) {
386: if (pledge("stdio rpath", NULL) == -1) {
1.167 schwarze 387: warn("pledge");
1.163 schwarze 388: return (int)MANDOCLEVEL_SYSERR;
389: }
1.155 schwarze 390: }
391:
1.28 schwarze 392: if (OP_CONFFILE == op && argc > 0) {
1.151 schwarze 393: warnx("-C: Too many arguments");
1.28 schwarze 394: goto usage;
395: }
396:
1.47 schwarze 397: exitcode = (int)MANDOCLEVEL_OK;
1.157 schwarze 398: mchars_alloc();
399: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.156 schwarze 400: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
401: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.2 schwarze 402:
1.47 schwarze 403: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 404:
1.47 schwarze 405: /*
1.109 schwarze 406: * Most of these deal with a specific directory.
1.97 schwarze 407: * Jump into that directory first.
1.47 schwarze 408: */
1.120 schwarze 409: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 410: goto out;
1.97 schwarze 411:
1.174 schwarze 412: dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
413: if (dba != NULL) {
1.97 schwarze 414: /*
415: * The existing database is usable. Process
416: * all files specified on the command-line.
417: */
1.163 schwarze 418: if (!nodb) {
419: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.167 schwarze 420: warn("pledge");
1.163 schwarze 421: exitcode = (int)MANDOCLEVEL_SYSERR;
422: goto out;
423: }
1.155 schwarze 424: }
1.97 schwarze 425: use_all = 1;
1.95 schwarze 426: for (i = 0; i < argc; i++)
427: filescan(argv[i]);
1.174 schwarze 428: if (nodb == 0)
1.173 schwarze 429: dbprune(dba);
1.95 schwarze 430: } else {
1.177 schwarze 431: /* Database missing or corrupt. */
1.179 schwarze 432: if (op != OP_UPDATE || errno != ENOENT)
433: say(MANDOC_DB, "%s: Automatically recreating"
434: " from scratch", strerror(errno));
1.97 schwarze 435: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 436: op = OP_DEFAULT;
437: if (0 == treescan())
438: goto out;
1.173 schwarze 439: dba = dba_new(128);
1.95 schwarze 440: }
1.47 schwarze 441: if (OP_DELETE != op)
1.173 schwarze 442: mpages_merge(dba, mp);
443: if (nodb == 0)
444: dbwrite(dba);
445: dba_free(dba);
1.47 schwarze 446: } else {
447: /*
448: * If we have arguments, use them as our manpaths.
1.181 schwarze 449: * If we don't, use man.conf(5).
1.47 schwarze 450: */
451: if (argc > 0) {
1.140 schwarze 452: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 453: argc, sizeof(char *));
1.140 schwarze 454: conf.manpath.sz = (size_t)argc;
1.47 schwarze 455: for (i = 0; i < argc; i++)
1.140 schwarze 456: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 457: } else
1.140 schwarze 458: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 459:
1.140 schwarze 460: if (conf.manpath.sz == 0) {
1.84 schwarze 461: exitcode = (int)MANDOCLEVEL_BADARG;
462: say("", "Empty manpath");
463: }
1.2 schwarze 464:
1.47 schwarze 465: /*
466: * First scan the tree rooted at a base directory, then
467: * build a new database and finally move it into place.
468: * Ignore zero-length directories and strip trailing
469: * slashes.
470: */
1.140 schwarze 471: for (j = 0; j < conf.manpath.sz; j++) {
472: sz = strlen(conf.manpath.paths[j]);
473: if (sz && conf.manpath.paths[j][sz - 1] == '/')
474: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 475: if (0 == sz)
476: continue;
1.2 schwarze 477:
1.47 schwarze 478: if (j) {
1.156 schwarze 479: mandoc_ohash_init(&mpages, 6,
480: offsetof(struct mpage, inodev));
481: mandoc_ohash_init(&mlinks, 6,
482: offsetof(struct mlink, file));
1.47 schwarze 483: }
1.1 schwarze 484:
1.140 schwarze 485: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 486: continue;
1.47 schwarze 487: if (0 == treescan())
1.120 schwarze 488: continue;
1.173 schwarze 489: dba = dba_new(128);
490: mpages_merge(dba, mp);
491: if (nodb == 0)
492: dbwrite(dba);
493: dba_free(dba);
1.28 schwarze 494:
1.140 schwarze 495: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 496: mpages_free();
497: ohash_delete(&mpages);
498: ohash_delete(&mlinks);
499: }
1.2 schwarze 500: }
1.47 schwarze 501: }
502: out:
1.140 schwarze 503: manconf_free(&conf);
1.123 schwarze 504: mparse_free(mp);
1.157 schwarze 505: mchars_free();
1.47 schwarze 506: mpages_free();
507: ohash_delete(&mpages);
508: ohash_delete(&mlinks);
1.150 schwarze 509: return exitcode;
1.47 schwarze 510: usage:
1.161 schwarze 511: progname = getprogname();
1.83 schwarze 512: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
513: " %s [-aDnpQ] [-Tutf8] dir ...\n"
514: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
515: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 516: " %s [-Q] -t file ...\n",
1.161 schwarze 517: progname, progname, progname, progname, progname);
1.1 schwarze 518:
1.150 schwarze 519: return (int)MANDOCLEVEL_BADARG;
1.47 schwarze 520: }
1.1 schwarze 521:
1.47 schwarze 522: /*
1.180 schwarze 523: * To get a singly linked list in alpha order while inserting entries
524: * at the beginning, process directory entries in reverse alpha order.
525: */
526: static int
527: fts_compare(const FTSENT **a, const FTSENT **b)
528: {
529: return -strcmp((*a)->fts_name, (*b)->fts_name);
530: }
531:
532: /*
1.47 schwarze 533: * Scan a directory tree rooted at "basedir" for manpages.
534: * We use fts(), scanning directory parts along the way for clues to our
535: * section and architecture.
536: *
537: * If use_all has been specified, grok all files.
538: * If not, sanitise paths to the following:
539: *
1.98 schwarze 540: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 541: * or
542: * [./]cat<section>[/<arch>]/<name>.0
543: *
1.169 krw 544: * TODO: accommodate for multi-language directories.
1.47 schwarze 545: */
546: static int
547: treescan(void)
548: {
1.96 schwarze 549: char buf[PATH_MAX];
1.47 schwarze 550: FTS *f;
551: FTSENT *ff;
552: struct mlink *mlink;
1.173 schwarze 553: int gzip;
554: enum form dform;
1.51 schwarze 555: char *dsec, *arch, *fsec, *cp;
556: const char *path;
1.47 schwarze 557: const char *argv[2];
1.1 schwarze 558:
1.47 schwarze 559: argv[0] = ".";
1.193 ! schwarze 560: argv[1] = NULL;
1.1 schwarze 561:
1.180 schwarze 562: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
563: fts_compare);
1.153 schwarze 564: if (f == NULL) {
1.47 schwarze 565: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 566: say("", "&fts_open");
1.150 schwarze 567: return 0;
1.47 schwarze 568: }
1.2 schwarze 569:
1.47 schwarze 570: dsec = arch = NULL;
571: dform = FORM_NONE;
1.2 schwarze 572:
1.153 schwarze 573: while ((ff = fts_read(f)) != NULL) {
1.47 schwarze 574: path = ff->fts_path + 2;
1.96 schwarze 575: switch (ff->fts_info) {
576:
577: /*
578: * Symbolic links require various sanity checks,
579: * then get handled just like regular files.
580: */
1.98 schwarze 581: case FTS_SL:
1.153 schwarze 582: if (realpath(path, buf) == NULL) {
1.96 schwarze 583: if (warnings)
584: say(path, "&realpath");
585: continue;
586: }
587: if (strstr(buf, basedir) != buf) {
588: if (warnings) say("",
589: "%s: outside base directory", buf);
590: continue;
591: }
592: /* Use logical inode to avoid mpages dupe. */
1.153 schwarze 593: if (stat(path, ff->fts_statp) == -1) {
1.96 schwarze 594: if (warnings)
595: say(path, "&stat");
596: continue;
597: }
598: /* FALLTHROUGH */
599:
1.15 schwarze 600: /*
1.47 schwarze 601: * If we're a regular file, add an mlink by using the
602: * stored directory data and handling the filename.
1.15 schwarze 603: */
1.98 schwarze 604: case FTS_F:
1.153 schwarze 605: if ( ! strcmp(path, MANDOC_DB))
1.47 schwarze 606: continue;
607: if ( ! use_all && ff->fts_level < 2) {
608: if (warnings)
609: say(path, "Extraneous file");
610: continue;
1.81 schwarze 611: }
612: gzip = 0;
613: fsec = NULL;
1.153 schwarze 614: while (fsec == NULL) {
1.81 schwarze 615: fsec = strrchr(ff->fts_name, '.');
1.153 schwarze 616: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.81 schwarze 617: break;
618: gzip = 1;
619: *fsec = '\0';
620: fsec = NULL;
621: }
1.153 schwarze 622: if (fsec == NULL) {
1.47 schwarze 623: if ( ! use_all) {
624: if (warnings)
625: say(path,
626: "No filename suffix");
627: continue;
628: }
1.153 schwarze 629: } else if ( ! strcmp(++fsec, "html")) {
1.47 schwarze 630: if (warnings)
631: say(path, "Skip html");
632: continue;
1.153 schwarze 633: } else if ( ! strcmp(fsec, "ps")) {
1.47 schwarze 634: if (warnings)
635: say(path, "Skip ps");
636: continue;
1.153 schwarze 637: } else if ( ! strcmp(fsec, "pdf")) {
1.47 schwarze 638: if (warnings)
639: say(path, "Skip pdf");
640: continue;
641: } else if ( ! use_all &&
1.153 schwarze 642: ((dform == FORM_SRC &&
1.138 schwarze 643: strncmp(fsec, dsec, strlen(dsec))) ||
1.153 schwarze 644: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.47 schwarze 645: if (warnings)
646: say(path, "Wrong filename suffix");
647: continue;
648: } else
649: fsec[-1] = '\0';
1.51 schwarze 650:
1.47 schwarze 651: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 652: if (strlcpy(mlink->file, path,
653: sizeof(mlink->file)) >=
654: sizeof(mlink->file)) {
655: say(path, "Filename too long");
656: free(mlink);
657: continue;
658: }
1.47 schwarze 659: mlink->dform = dform;
1.51 schwarze 660: mlink->dsec = dsec;
661: mlink->arch = arch;
662: mlink->name = ff->fts_name;
663: mlink->fsec = fsec;
1.81 schwarze 664: mlink->gzip = gzip;
1.47 schwarze 665: mlink_add(mlink, ff->fts_statp);
666: continue;
1.96 schwarze 667:
1.98 schwarze 668: case FTS_D:
669: case FTS_DP:
1.96 schwarze 670: break;
671:
672: default:
1.47 schwarze 673: if (warnings)
674: say(path, "Not a regular file");
675: continue;
676: }
677:
678: switch (ff->fts_level) {
1.98 schwarze 679: case 0:
1.47 schwarze 680: /* Ignore the root directory. */
681: break;
1.98 schwarze 682: case 1:
1.47 schwarze 683: /*
684: * This might contain manX/ or catX/.
685: * Try to infer this from the name.
686: * If we're not in use_all, enforce it.
687: */
688: cp = ff->fts_name;
1.153 schwarze 689: if (ff->fts_info == FTS_DP) {
690: dform = FORM_NONE;
691: dsec = NULL;
1.47 schwarze 692: break;
1.153 schwarze 693: }
1.15 schwarze 694:
1.153 schwarze 695: if ( ! strncmp(cp, "man", 3)) {
1.47 schwarze 696: dform = FORM_SRC;
697: dsec = cp + 3;
1.153 schwarze 698: } else if ( ! strncmp(cp, "cat", 3)) {
1.47 schwarze 699: dform = FORM_CAT;
700: dsec = cp + 3;
1.51 schwarze 701: } else {
702: dform = FORM_NONE;
703: dsec = NULL;
1.26 schwarze 704: }
1.47 schwarze 705:
1.153 schwarze 706: if (dsec != NULL || use_all)
1.47 schwarze 707: break;
708:
709: if (warnings)
710: say(path, "Unknown directory part");
711: fts_set(f, ff, FTS_SKIP);
712: break;
1.98 schwarze 713: case 2:
1.47 schwarze 714: /*
715: * Possibly our architecture.
716: * If we're descending, keep tabs on it.
717: */
1.153 schwarze 718: if (ff->fts_info != FTS_DP && dsec != NULL)
1.47 schwarze 719: arch = ff->fts_name;
1.51 schwarze 720: else
721: arch = NULL;
1.47 schwarze 722: break;
723: default:
1.153 schwarze 724: if (ff->fts_info == FTS_DP || use_all)
1.47 schwarze 725: break;
726: if (warnings)
727: say(path, "Extraneous directory part");
728: fts_set(f, ff, FTS_SKIP);
729: break;
1.14 schwarze 730: }
1.47 schwarze 731: }
732:
733: fts_close(f);
1.150 schwarze 734: return 1;
1.47 schwarze 735: }
1.1 schwarze 736:
1.47 schwarze 737: /*
738: * Add a file to the mlinks table.
739: * Do not verify that it's a "valid" looking manpage (we'll do that
740: * later).
741: *
742: * Try to infer the manual section, architecture, and page name from the
743: * path, assuming it looks like
744: *
1.98 schwarze 745: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 746: * or
747: * [./]cat<section>[/<arch>]/<name>.0
748: *
749: * See treescan() for the fts(3) version of this.
750: */
751: static void
752: filescan(const char *file)
753: {
754: char buf[PATH_MAX];
755: struct stat st;
756: struct mlink *mlink;
757: char *p, *start;
758:
759: assert(use_all);
760:
761: if (0 == strncmp(file, "./", 2))
762: file += 2;
763:
1.96 schwarze 764: /*
765: * We have to do lstat(2) before realpath(3) loses
766: * the information whether this is a symbolic link.
767: * We need to know that because for symbolic links,
768: * we want to use the orginal file name, while for
769: * regular files, we want to use the real path.
770: */
771: if (-1 == lstat(file, &st)) {
772: exitcode = (int)MANDOCLEVEL_BADARG;
773: say(file, "&lstat");
774: return;
775: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
776: exitcode = (int)MANDOCLEVEL_BADARG;
777: say(file, "Not a regular file");
778: return;
779: }
780:
781: /*
782: * We have to resolve the file name to the real path
783: * in any case for the base directory check.
784: */
1.47 schwarze 785: if (NULL == realpath(file, buf)) {
786: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 787: say(file, "&realpath");
1.47 schwarze 788: return;
1.63 schwarze 789: }
790:
1.109 schwarze 791: if (OP_TEST == op)
1.63 schwarze 792: start = buf;
1.109 schwarze 793: else if (strstr(buf, basedir) == buf)
794: start = buf + strlen(basedir);
1.63 schwarze 795: else {
1.47 schwarze 796: exitcode = (int)MANDOCLEVEL_BADARG;
797: say("", "%s: outside base directory", buf);
798: return;
1.63 schwarze 799: }
800:
1.96 schwarze 801: /*
802: * Now we are sure the file is inside our tree.
803: * If it is a symbolic link, ignore the real path
804: * and use the original name.
805: * This implies passing stuff like "cat1/../man1/foo.1"
806: * on the command line won't work. So don't do that.
807: * Note the stat(2) can still fail if the link target
808: * doesn't exist.
809: */
810: if (S_IFLNK & st.st_mode) {
811: if (-1 == stat(buf, &st)) {
812: exitcode = (int)MANDOCLEVEL_BADARG;
813: say(file, "&stat");
814: return;
815: }
1.100 schwarze 816: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
817: say(file, "Filename too long");
818: return;
819: }
1.109 schwarze 820: start = buf;
821: if (OP_TEST != op && strstr(buf, basedir) == buf)
822: start += strlen(basedir);
1.1 schwarze 823: }
1.63 schwarze 824:
1.47 schwarze 825: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 826: mlink->dform = FORM_NONE;
1.100 schwarze 827: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
828: sizeof(mlink->file)) {
829: say(start, "Filename too long");
1.134 schwarze 830: free(mlink);
1.100 schwarze 831: return;
1.188 schwarze 832: }
833:
834: /*
835: * In test mode or when the original name is absolute
836: * but outside our tree, guess the base directory.
837: */
838:
839: if (op == OP_TEST || (start == buf && *start == '/')) {
840: if (strncmp(buf, "man/", 4) == 0)
841: start = buf + 4;
842: else if ((start = strstr(buf, "/man/")) != NULL)
843: start += 5;
844: else
845: start = buf;
1.100 schwarze 846: }
1.1 schwarze 847:
1.10 schwarze 848: /*
1.47 schwarze 849: * First try to guess our directory structure.
850: * If we find a separator, try to look for man* or cat*.
851: * If we find one of these and what's underneath is a directory,
852: * assume it's an architecture.
1.10 schwarze 853: */
1.47 schwarze 854: if (NULL != (p = strchr(start, '/'))) {
855: *p++ = '\0';
856: if (0 == strncmp(start, "man", 3)) {
857: mlink->dform = FORM_SRC;
1.51 schwarze 858: mlink->dsec = start + 3;
1.47 schwarze 859: } else if (0 == strncmp(start, "cat", 3)) {
860: mlink->dform = FORM_CAT;
1.51 schwarze 861: mlink->dsec = start + 3;
1.47 schwarze 862: }
1.10 schwarze 863:
1.47 schwarze 864: start = p;
865: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
866: *p++ = '\0';
1.51 schwarze 867: mlink->arch = start;
1.47 schwarze 868: start = p;
1.41 deraadt 869: }
1.47 schwarze 870: }
1.7 schwarze 871:
1.47 schwarze 872: /*
873: * Now check the file suffix.
874: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
875: */
876: p = strrchr(start, '\0');
877: while (p-- > start && '/' != *p && '.' != *p)
878: /* Loop. */ ;
879:
880: if ('.' == *p) {
881: *p++ = '\0';
1.51 schwarze 882: mlink->fsec = p;
1.47 schwarze 883: }
1.41 deraadt 884:
1.47 schwarze 885: /*
886: * Now try to parse the name.
887: * Use the filename portion of the path.
888: */
889: mlink->name = start;
890: if (NULL != (p = strrchr(start, '/'))) {
891: mlink->name = p + 1;
892: *p = '\0';
893: }
894: mlink_add(mlink, &st);
895: }
1.2 schwarze 896:
1.47 schwarze 897: static void
898: mlink_add(struct mlink *mlink, const struct stat *st)
899: {
900: struct inodev inodev;
901: struct mpage *mpage;
902: unsigned int slot;
903:
904: assert(NULL != mlink->file);
905:
1.51 schwarze 906: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
907: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
908: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
909: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 910:
911: if ('0' == *mlink->fsec) {
912: free(mlink->fsec);
913: mlink->fsec = mandoc_strdup(mlink->dsec);
914: mlink->fform = FORM_CAT;
915: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
916: mlink->fform = FORM_SRC;
917: else
918: mlink->fform = FORM_NONE;
1.1 schwarze 919:
1.47 schwarze 920: slot = ohash_qlookup(&mlinks, mlink->file);
921: assert(NULL == ohash_find(&mlinks, slot));
922: ohash_insert(&mlinks, slot, mlink);
923:
1.139 schwarze 924: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 925: inodev.st_ino = st->st_ino;
926: inodev.st_dev = st->st_dev;
927: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
928: sizeof(struct inodev), inodev.st_ino);
929: mpage = ohash_find(&mpages, slot);
930: if (NULL == mpage) {
931: mpage = mandoc_calloc(1, sizeof(struct mpage));
932: mpage->inodev.st_ino = inodev.st_ino;
933: mpage->inodev.st_dev = inodev.st_dev;
1.173 schwarze 934: mpage->form = FORM_NONE;
1.180 schwarze 935: mpage->next = mpage_head;
936: mpage_head = mpage;
1.47 schwarze 937: ohash_insert(&mpages, slot, mpage);
938: } else
939: mlink->next = mpage->mlinks;
940: mpage->mlinks = mlink;
1.75 schwarze 941: mlink->mpage = mpage;
1.47 schwarze 942: }
1.1 schwarze 943:
1.47 schwarze 944: static void
945: mlink_free(struct mlink *mlink)
946: {
1.1 schwarze 947:
1.47 schwarze 948: free(mlink->dsec);
949: free(mlink->arch);
950: free(mlink->name);
951: free(mlink->fsec);
952: free(mlink);
953: }
1.1 schwarze 954:
1.47 schwarze 955: static void
956: mpages_free(void)
957: {
958: struct mpage *mpage;
959: struct mlink *mlink;
960:
1.180 schwarze 961: while ((mpage = mpage_head) != NULL) {
962: while ((mlink = mpage->mlinks) != NULL) {
1.47 schwarze 963: mpage->mlinks = mlink->next;
964: mlink_free(mlink);
965: }
1.180 schwarze 966: mpage_head = mpage->next;
1.47 schwarze 967: free(mpage->sec);
968: free(mpage->arch);
969: free(mpage->title);
970: free(mpage->desc);
971: free(mpage);
972: }
973: }
1.1 schwarze 974:
1.47 schwarze 975: /*
976: * For each mlink to the mpage, check whether the path looks like
977: * it is formatted, and if it does, check whether a source manual
978: * exists by the same name, ignoring the suffix.
979: * If both conditions hold, drop the mlink.
980: */
981: static void
982: mlinks_undupe(struct mpage *mpage)
983: {
984: char buf[PATH_MAX];
985: struct mlink **prev;
986: struct mlink *mlink;
987: char *bufp;
988:
989: mpage->form = FORM_CAT;
990: prev = &mpage->mlinks;
991: while (NULL != (mlink = *prev)) {
992: if (FORM_CAT != mlink->dform) {
993: mpage->form = FORM_NONE;
994: goto nextlink;
995: }
1.100 schwarze 996: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 997: bufp = strstr(buf, "cat");
998: assert(NULL != bufp);
999: memcpy(bufp, "man", 3);
1000: if (NULL != (bufp = strrchr(buf, '.')))
1001: *++bufp = '\0';
1.100 schwarze 1002: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1003: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1004: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1005: goto nextlink;
1006: if (warnings)
1007: say(mlink->file, "Man source exists: %s", buf);
1008: if (use_all)
1009: goto nextlink;
1010: *prev = mlink->next;
1011: mlink_free(mlink);
1012: continue;
1013: nextlink:
1014: prev = &(*prev)->next;
1.1 schwarze 1015: }
1.47 schwarze 1016: }
1.1 schwarze 1017:
1.88 schwarze 1018: static void
1.50 schwarze 1019: mlink_check(struct mpage *mpage, struct mlink *mlink)
1020: {
1.88 schwarze 1021: struct str *str;
1022: unsigned int slot;
1.50 schwarze 1023:
1024: /*
1025: * Check whether the manual section given in a file
1026: * agrees with the directory where the file is located.
1027: * Some manuals have suffixes like (3p) on their
1028: * section number either inside the file or in the
1029: * directory name, some are linked into more than one
1030: * section, like encrypt(1) = makekey(8).
1031: */
1032:
1033: if (FORM_SRC == mpage->form &&
1.88 schwarze 1034: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1035: say(mlink->file, "Section \"%s\" manual in %s directory",
1036: mpage->sec, mlink->dsec);
1037:
1038: /*
1039: * Manual page directories exist for each kernel
1040: * architecture as returned by machine(1).
1041: * However, many manuals only depend on the
1042: * application architecture as returned by arch(1).
1043: * For example, some (2/ARM) manuals are shared
1044: * across the "armish" and "zaurus" kernel
1045: * architectures.
1046: * A few manuals are even shared across completely
1047: * different architectures, for example fdformat(1)
1.178 schwarze 1048: * on amd64, i386, and sparc64.
1.50 schwarze 1049: */
1050:
1.88 schwarze 1051: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1052: say(mlink->file, "Architecture \"%s\" manual in "
1053: "\"%s\" directory", mpage->arch, mlink->arch);
1054:
1.88 schwarze 1055: /*
1056: * XXX
1.90 schwarze 1057: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1058: */
1059:
1060: if (FORM_CAT == mpage->form)
1061: return;
1062:
1063: /*
1064: * Check whether this mlink
1065: * appears as a name in the NAME section.
1066: */
1.50 schwarze 1067:
1.90 schwarze 1068: slot = ohash_qlookup(&names, mlink->name);
1069: str = ohash_find(&names, slot);
1.88 schwarze 1070: assert(NULL != str);
1.90 schwarze 1071: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1072: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1073: }
1074:
1.47 schwarze 1075: /*
1076: * Run through the files in the global vector "mpages"
1077: * and add them to the database specified in "basedir".
1078: *
1079: * This handles the parsing scheme itself, using the cues of directory
1080: * and filename to determine whether the file is parsable or not.
1081: */
1082: static void
1.173 schwarze 1083: mpages_merge(struct dba *dba, struct mparse *mp)
1.47 schwarze 1084: {
1.180 schwarze 1085: struct mpage *mpage, *mpage_dest;
1.75 schwarze 1086: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1087: struct roff_man *man;
1.75 schwarze 1088: char *sodest;
1.69 schwarze 1089: char *cp;
1.117 schwarze 1090: int fd;
1.64 schwarze 1091:
1.180 schwarze 1092: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.47 schwarze 1093: mlinks_undupe(mpage);
1.180 schwarze 1094: if ((mlink = mpage->mlinks) == NULL)
1095: continue;
1.1 schwarze 1096:
1.90 schwarze 1097: name_mask = NAME_MASK;
1.156 schwarze 1098: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1099: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.47 schwarze 1100: mparse_reset(mp);
1101: man = NULL;
1.81 schwarze 1102: sodest = NULL;
1103:
1.166 schwarze 1104: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.136 schwarze 1105: say(mlink->file, "&open");
1.117 schwarze 1106: goto nextpage;
1.81 schwarze 1107: }
1.11 schwarze 1108:
1109: /*
1.137 schwarze 1110: * Interpret the file as mdoc(7) or man(7) source
1111: * code, unless it is known to be formatted.
1.11 schwarze 1112: */
1.136 schwarze 1113: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1114: mparse_readfd(mp, fd, mlink->file);
1.165 schwarze 1115: close(fd);
1.189 schwarze 1116: fd = -1;
1.145 schwarze 1117: mparse_result(mp, &man, &sodest);
1.47 schwarze 1118: }
1.11 schwarze 1119:
1.126 schwarze 1120: if (sodest != NULL) {
1.75 schwarze 1121: mlink_dest = ohash_find(&mlinks,
1122: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1123: if (mlink_dest == NULL) {
1124: mandoc_asprintf(&cp, "%s.gz", sodest);
1125: mlink_dest = ohash_find(&mlinks,
1126: ohash_qlookup(&mlinks, cp));
1127: free(cp);
1128: }
1129: if (mlink_dest != NULL) {
1.75 schwarze 1130:
1131: /* The .so target exists. */
1132:
1133: mpage_dest = mlink_dest->mpage;
1134: while (1) {
1135: mlink->mpage = mpage_dest;
1136:
1137: /*
1138: * If the target was already
1139: * processed, add the links
1140: * to the database now.
1141: * Otherwise, this will
1142: * happen when we come
1143: * to the target.
1144: */
1145:
1.173 schwarze 1146: if (mpage_dest->dba != NULL)
1147: dbadd_mlink(mlink);
1.75 schwarze 1148:
1.126 schwarze 1149: if (mlink->next == NULL)
1.75 schwarze 1150: break;
1151: mlink = mlink->next;
1152: }
1153:
1154: /* Move all links to the target. */
1155:
1156: mlink->next = mlink_dest->next;
1157: mlink_dest->next = mpage->mlinks;
1158: mpage->mlinks = NULL;
1159: }
1.81 schwarze 1160: goto nextpage;
1.145 schwarze 1161: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.159 schwarze 1162: mdoc_validate(man);
1.47 schwarze 1163: mpage->form = FORM_SRC;
1.146 schwarze 1164: mpage->sec = man->meta.msec;
1.112 schwarze 1165: mpage->sec = mandoc_strdup(
1.126 schwarze 1166: mpage->sec == NULL ? "" : mpage->sec);
1.146 schwarze 1167: mpage->arch = man->meta.arch;
1.47 schwarze 1168: mpage->arch = mandoc_strdup(
1.126 schwarze 1169: mpage->arch == NULL ? "" : mpage->arch);
1.146 schwarze 1170: mpage->title = mandoc_strdup(man->meta.title);
1.145 schwarze 1171: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.160 schwarze 1172: man_validate(man);
1.189 schwarze 1173: if (*man->meta.msec != '\0' ||
1174: *man->meta.msec != '\0') {
1175: mpage->form = FORM_SRC;
1176: mpage->sec = mandoc_strdup(man->meta.msec);
1177: mpage->arch = mandoc_strdup(mlink->arch);
1178: mpage->title = mandoc_strdup(man->meta.title);
1179: } else
1180: man = NULL;
1181: }
1182:
1183: assert(mpage->desc == NULL);
1184: if (man == NULL) {
1.47 schwarze 1185: mpage->form = FORM_CAT;
1.136 schwarze 1186: mpage->sec = mandoc_strdup(mlink->dsec);
1187: mpage->arch = mandoc_strdup(mlink->arch);
1188: mpage->title = mandoc_strdup(mlink->name);
1.189 schwarze 1189: parse_cat(mpage, fd);
1190: } else if (man->macroset == MACROSET_MDOC)
1.146 schwarze 1191: parse_mdoc(mpage, &man->meta, man->first);
1.189 schwarze 1192: else
1.146 schwarze 1193: parse_man(mpage, &man->meta, man->first);
1.191 schwarze 1194: if (mpage->desc == NULL) {
1195: mpage->desc = mandoc_strdup(mlink->name);
1196: if (warnings)
1197: say(mlink->file, "No one-line description, "
1198: "using filename \"%s\"", mlink->name);
1199: }
1.88 schwarze 1200:
1.187 schwarze 1201: for (mlink = mpage->mlinks;
1202: mlink != NULL;
1203: mlink = mlink->next) {
1204: putkey(mpage, mlink->name, NAME_FILE);
1205: if (warnings && !use_all)
1.88 schwarze 1206: mlink_check(mpage, mlink);
1.187 schwarze 1207: }
1.6 schwarze 1208:
1.173 schwarze 1209: dbadd(dba, mpage);
1.81 schwarze 1210:
1211: nextpage:
1.47 schwarze 1212: ohash_delete(&strings);
1.90 schwarze 1213: ohash_delete(&names);
1.47 schwarze 1214: }
1215: }
1.6 schwarze 1216:
1.47 schwarze 1217: static void
1.81 schwarze 1218: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1219: {
1220: FILE *stream;
1.190 schwarze 1221: struct mlink *mlink;
1222: char *line, *p, *title, *sec;
1.162 schwarze 1223: size_t linesz, plen, titlesz;
1224: ssize_t len;
1225: int offs;
1.1 schwarze 1226:
1.190 schwarze 1227: mlink = mpage->mlinks;
1228: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1229: if (stream == NULL) {
1230: if (fd != -1)
1.114 doug 1231: close(fd);
1.47 schwarze 1232: if (warnings)
1.190 schwarze 1233: say(mlink->file, "&fopen");
1.47 schwarze 1234: return;
1235: }
1.1 schwarze 1236:
1.162 schwarze 1237: line = NULL;
1238: linesz = 0;
1239:
1.190 schwarze 1240: /* Parse the section number from the header line. */
1241:
1242: while (getline(&line, &linesz, stream) != -1) {
1243: if (*line == '\n')
1244: continue;
1245: if ((sec = strchr(line, '(')) == NULL)
1246: break;
1247: if ((p = strchr(++sec, ')')) == NULL)
1248: break;
1249: free(mpage->sec);
1250: mpage->sec = mandoc_strndup(sec, p - sec);
1251: if (warnings && *mlink->dsec != '\0' &&
1252: strcasecmp(mpage->sec, mlink->dsec))
1253: say(mlink->file,
1254: "Section \"%s\" manual in %s directory",
1255: mpage->sec, mlink->dsec);
1256: break;
1257: }
1258:
1.47 schwarze 1259: /* Skip to first blank line. */
1.1 schwarze 1260:
1.190 schwarze 1261: while (line == NULL || *line != '\n')
1262: if (getline(&line, &linesz, stream) == -1)
1.47 schwarze 1263: break;
1.1 schwarze 1264:
1.47 schwarze 1265: /*
1266: * Assume the first line that is not indented
1267: * is the first section header. Skip to it.
1268: */
1.1 schwarze 1269:
1.162 schwarze 1270: while (getline(&line, &linesz, stream) != -1)
1271: if (*line != '\n' && *line != ' ')
1.47 schwarze 1272: break;
1.98 schwarze 1273:
1.47 schwarze 1274: /*
1275: * Read up until the next section into a buffer.
1276: * Strip the leading and trailing newline from each read line,
1277: * appending a trailing space.
1278: * Ignore empty (whitespace-only) lines.
1279: */
1.28 schwarze 1280:
1.47 schwarze 1281: titlesz = 0;
1282: title = NULL;
1.38 schwarze 1283:
1.162 schwarze 1284: while ((len = getline(&line, &linesz, stream)) != -1) {
1285: if (*line != ' ')
1.47 schwarze 1286: break;
1.162 schwarze 1287: offs = 0;
1288: while (isspace((unsigned char)line[offs]))
1289: offs++;
1290: if (line[offs] == '\0')
1.47 schwarze 1291: continue;
1.162 schwarze 1292: title = mandoc_realloc(title, titlesz + len - offs);
1293: memcpy(title + titlesz, line + offs, len - offs);
1294: titlesz += len - offs;
1.47 schwarze 1295: title[titlesz - 1] = ' ';
1296: }
1.162 schwarze 1297: free(line);
1.28 schwarze 1298:
1.47 schwarze 1299: /*
1300: * If no page content can be found, or the input line
1301: * is already the next section header, or there is no
1302: * trailing newline, reuse the page title as the page
1303: * description.
1304: */
1.1 schwarze 1305:
1.47 schwarze 1306: if (NULL == title || '\0' == *title) {
1307: if (warnings)
1.190 schwarze 1308: say(mlink->file, "Cannot find NAME section");
1.47 schwarze 1309: fclose(stream);
1310: free(title);
1311: return;
1312: }
1.24 schwarze 1313:
1.162 schwarze 1314: title[titlesz - 1] = '\0';
1.24 schwarze 1315:
1.47 schwarze 1316: /*
1317: * Skip to the first dash.
1318: * Use the remaining line as the description (no more than 70
1319: * bytes).
1320: */
1.28 schwarze 1321:
1.47 schwarze 1322: if (NULL != (p = strstr(title, "- "))) {
1323: for (p += 2; ' ' == *p || '\b' == *p; p++)
1324: /* Skip to next word. */ ;
1325: } else {
1326: if (warnings)
1.191 schwarze 1327: say(mlink->file, "No dash in title line, "
1328: "reusing \"%s\" as one-line description", title);
1.47 schwarze 1329: p = title;
1330: }
1.1 schwarze 1331:
1.47 schwarze 1332: plen = strlen(p);
1.1 schwarze 1333:
1.47 schwarze 1334: /* Strip backspace-encoding from line. */
1.1 schwarze 1335:
1.47 schwarze 1336: while (NULL != (line = memchr(p, '\b', plen))) {
1337: len = line - p;
1338: if (0 == len) {
1339: memmove(line, line + 1, plen--);
1340: continue;
1.98 schwarze 1341: }
1.47 schwarze 1342: memmove(line - 1, line + 1, plen - len);
1343: plen -= 2;
1344: }
1.1 schwarze 1345:
1.47 schwarze 1346: mpage->desc = mandoc_strdup(p);
1347: fclose(stream);
1348: free(title);
1349: }
1.16 schwarze 1350:
1.47 schwarze 1351: /*
1352: * Put a type/word pair into the word database for this particular file.
1353: */
1354: static void
1.69 schwarze 1355: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1356: {
1357: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1358: }
1359:
1360: /*
1.47 schwarze 1361: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1362: */
1363: static void
1.47 schwarze 1364: putmdockey(const struct mpage *mpage,
1.184 schwarze 1365: const struct roff_node *n, uint64_t m, int taboo)
1.2 schwarze 1366: {
1.16 schwarze 1367:
1.47 schwarze 1368: for ( ; NULL != n; n = n->next) {
1.184 schwarze 1369: if (n->flags & taboo)
1370: continue;
1.47 schwarze 1371: if (NULL != n->child)
1.184 schwarze 1372: putmdockey(mpage, n->child, m, taboo);
1.141 schwarze 1373: if (n->type == ROFFT_TEXT)
1.47 schwarze 1374: putkey(mpage, n->string, m);
1375: }
1376: }
1.16 schwarze 1377:
1.47 schwarze 1378: static void
1.143 schwarze 1379: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1380: const struct roff_node *n)
1.47 schwarze 1381: {
1.142 schwarze 1382: const struct roff_node *head, *body;
1.78 schwarze 1383: char *start, *title;
1.47 schwarze 1384: char byte;
1.78 schwarze 1385: size_t sz;
1.16 schwarze 1386:
1.168 schwarze 1387: if (n == NULL)
1.47 schwarze 1388: return;
1.16 schwarze 1389:
1.47 schwarze 1390: /*
1391: * We're only searching for one thing: the first text child in
1392: * the BODY of a NAME section. Since we don't keep track of
1393: * sections in -man, run some hoops to find out whether we're in
1394: * the correct section or not.
1395: */
1.16 schwarze 1396:
1.141 schwarze 1397: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1398: body = n;
1.168 schwarze 1399: if ((head = body->parent->head) != NULL &&
1400: (head = head->child) != NULL &&
1401: head->next == NULL &&
1.141 schwarze 1402: head->type == ROFFT_TEXT &&
1.168 schwarze 1403: strcmp(head->string, "NAME") == 0 &&
1404: body->child != NULL) {
1.2 schwarze 1405:
1.47 schwarze 1406: /*
1407: * Suck the entire NAME section into memory.
1408: * Yes, we might run away.
1409: * But too many manuals have big, spread-out
1410: * NAME sections over many lines.
1411: */
1.2 schwarze 1412:
1.78 schwarze 1413: title = NULL;
1.147 schwarze 1414: deroff(&title, body);
1.47 schwarze 1415: if (NULL == title)
1416: return;
1.16 schwarze 1417:
1.98 schwarze 1418: /*
1.47 schwarze 1419: * Go through a special heuristic dance here.
1420: * Conventionally, one or more manual names are
1421: * comma-specified prior to a whitespace, then a
1422: * dash, then a description. Try to puzzle out
1423: * the name parts here.
1424: */
1.16 schwarze 1425:
1.78 schwarze 1426: start = title;
1.47 schwarze 1427: for ( ;; ) {
1428: sz = strcspn(start, " ,");
1429: if ('\0' == start[sz])
1430: break;
1.1 schwarze 1431:
1.47 schwarze 1432: byte = start[sz];
1433: start[sz] = '\0';
1.67 schwarze 1434:
1435: /*
1436: * Assume a stray trailing comma in the
1437: * name list if a name begins with a dash.
1438: */
1439:
1440: if ('-' == start[0] ||
1441: ('\\' == start[0] && '-' == start[1]))
1442: break;
1.1 schwarze 1443:
1.90 schwarze 1444: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1445: if ( ! (mpage->name_head_done ||
1446: strcasecmp(start, meta->title))) {
1447: putkey(mpage, start, NAME_HEAD);
1448: mpage->name_head_done = 1;
1449: }
1.1 schwarze 1450:
1.47 schwarze 1451: if (' ' == byte) {
1452: start += sz + 1;
1453: break;
1454: }
1.1 schwarze 1455:
1.47 schwarze 1456: assert(',' == byte);
1457: start += sz + 1;
1458: while (' ' == *start)
1459: start++;
1460: }
1.1 schwarze 1461:
1.78 schwarze 1462: if (start == title) {
1.90 schwarze 1463: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1464: if ( ! (mpage->name_head_done ||
1465: strcasecmp(start, meta->title))) {
1466: putkey(mpage, start, NAME_HEAD);
1467: mpage->name_head_done = 1;
1468: }
1.47 schwarze 1469: free(title);
1470: return;
1471: }
1.1 schwarze 1472:
1.47 schwarze 1473: while (isspace((unsigned char)*start))
1474: start++;
1.1 schwarze 1475:
1.47 schwarze 1476: if (0 == strncmp(start, "-", 1))
1477: start += 1;
1478: else if (0 == strncmp(start, "\\-\\-", 4))
1479: start += 4;
1480: else if (0 == strncmp(start, "\\-", 2))
1481: start += 2;
1482: else if (0 == strncmp(start, "\\(en", 4))
1483: start += 4;
1484: else if (0 == strncmp(start, "\\(em", 4))
1485: start += 4;
1.1 schwarze 1486:
1.47 schwarze 1487: while (' ' == *start)
1488: start++;
1.1 schwarze 1489:
1.47 schwarze 1490: mpage->desc = mandoc_strdup(start);
1491: free(title);
1492: return;
1493: }
1494: }
1.1 schwarze 1495:
1.47 schwarze 1496: for (n = n->child; n; n = n->next) {
1497: if (NULL != mpage->desc)
1498: break;
1.129 schwarze 1499: parse_man(mpage, meta, n);
1.1 schwarze 1500: }
1501: }
1502:
1503: static void
1.143 schwarze 1504: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1505: const struct roff_node *n)
1.1 schwarze 1506: {
1507:
1.47 schwarze 1508: assert(NULL != n);
1509: for (n = n->child; NULL != n; n = n->next) {
1.184 schwarze 1510: if (n->flags & mdocs[n->tok].taboo)
1511: continue;
1.47 schwarze 1512: switch (n->type) {
1.141 schwarze 1513: case ROFFT_ELEM:
1514: case ROFFT_BLOCK:
1515: case ROFFT_HEAD:
1516: case ROFFT_BODY:
1517: case ROFFT_TAIL:
1.47 schwarze 1518: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1519: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1520: break;
1521: if (mdocs[n->tok].mask)
1522: putmdockey(mpage, n->child,
1.184 schwarze 1523: mdocs[n->tok].mask, mdocs[n->tok].taboo);
1.47 schwarze 1524: break;
1525: default:
1.141 schwarze 1526: assert(n->type != ROFFT_ROOT);
1.47 schwarze 1527: continue;
1528: }
1529: if (NULL != n->child)
1.127 schwarze 1530: parse_mdoc(mpage, meta, n);
1.1 schwarze 1531: }
1532: }
1533:
1.19 schwarze 1534: static int
1.143 schwarze 1535: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1536: const struct roff_node *n)
1.1 schwarze 1537: {
1.131 schwarze 1538: char *start, *end;
1.1 schwarze 1539: size_t sz;
1.19 schwarze 1540:
1.47 schwarze 1541: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1542: NULL == (n = n->child) ||
1.141 schwarze 1543: n->type != ROFFT_TEXT)
1.150 schwarze 1544: return 0;
1.1 schwarze 1545:
1546: /*
1547: * Only consider those `Fd' macro fields that begin with an
1548: * "inclusion" token (versus, e.g., #define).
1549: */
1.47 schwarze 1550:
1.1 schwarze 1551: if (strcmp("#include", n->string))
1.150 schwarze 1552: return 0;
1.1 schwarze 1553:
1.141 schwarze 1554: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.150 schwarze 1555: return 0;
1.1 schwarze 1556:
1557: /*
1558: * Strip away the enclosing angle brackets and make sure we're
1559: * not zero-length.
1560: */
1561:
1562: start = n->string;
1563: if ('<' == *start || '"' == *start)
1564: start++;
1565:
1566: if (0 == (sz = strlen(start)))
1.150 schwarze 1567: return 0;
1.1 schwarze 1568:
1569: end = &start[(int)sz - 1];
1570: if ('>' == *end || '"' == *end)
1571: end--;
1572:
1.47 schwarze 1573: if (end > start)
1574: putkeys(mpage, start, end - start + 1, TYPE_In);
1.150 schwarze 1575: return 0;
1.1 schwarze 1576: }
1577:
1.133 schwarze 1578: static void
1.142 schwarze 1579: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 schwarze 1580: {
1.69 schwarze 1581: char *cp;
1.133 schwarze 1582: size_t sz;
1.1 schwarze 1583:
1.141 schwarze 1584: if (n->type != ROFFT_TEXT)
1.133 schwarze 1585: return;
1.19 schwarze 1586:
1.133 schwarze 1587: /* Skip function pointer punctuation. */
1.1 schwarze 1588:
1.133 schwarze 1589: cp = n->string;
1590: while (*cp == '(' || *cp == '*')
1.1 schwarze 1591: cp++;
1.133 schwarze 1592: sz = strcspn(cp, "()");
1.1 schwarze 1593:
1.133 schwarze 1594: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1595: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1596: putkeys(mpage, cp, sz, NAME_SYN);
1597: }
1.19 schwarze 1598:
1.133 schwarze 1599: static int
1.143 schwarze 1600: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1601: const struct roff_node *n)
1.133 schwarze 1602: {
1603:
1604: if (n->child == NULL)
1.150 schwarze 1605: return 0;
1.133 schwarze 1606:
1607: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1608:
1.133 schwarze 1609: for (n = n->child->next; n != NULL; n = n->next)
1.141 schwarze 1610: if (n->type == ROFFT_TEXT)
1.47 schwarze 1611: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1612:
1.150 schwarze 1613: return 0;
1.128 schwarze 1614: }
1615:
1616: static int
1.143 schwarze 1617: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1618: const struct roff_node *n)
1.128 schwarze 1619: {
1.132 schwarze 1620:
1.141 schwarze 1621: if (n->type != ROFFT_HEAD)
1.150 schwarze 1622: return 1;
1.128 schwarze 1623:
1.133 schwarze 1624: if (n->child != NULL)
1625: parse_mdoc_fname(mpage, n->child);
1626:
1.150 schwarze 1627: return 0;
1.1 schwarze 1628: }
1629:
1.19 schwarze 1630: static int
1.164 schwarze 1631: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1632: const struct roff_node *n)
1633: {
1634: char *cp;
1635:
1636: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1637: return 0;
1638:
1.168 schwarze 1639: if (n->child != NULL &&
1640: n->child->next == NULL &&
1641: n->child->type == ROFFT_TEXT)
1.164 schwarze 1642: return 1;
1643:
1644: cp = NULL;
1645: deroff(&cp, n);
1646: if (cp != NULL) {
1647: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1648: n->type == ROFFT_BODY ? TYPE_Va : 0));
1649: free(cp);
1650: }
1651:
1652: return 0;
1653: }
1654:
1655: static int
1.143 schwarze 1656: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1657: const struct roff_node *n)
1.1 schwarze 1658: {
1.47 schwarze 1659: char *cp;
1.1 schwarze 1660:
1661: if (NULL == (n = n->child))
1.150 schwarze 1662: return 0;
1.1 schwarze 1663:
1.47 schwarze 1664: if (NULL == n->next) {
1665: putkey(mpage, n->string, TYPE_Xr);
1.150 schwarze 1666: return 0;
1.47 schwarze 1667: }
1.1 schwarze 1668:
1.77 schwarze 1669: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1670: putkey(mpage, cp, TYPE_Xr);
1671: free(cp);
1.150 schwarze 1672: return 0;
1.1 schwarze 1673: }
1674:
1.19 schwarze 1675: static int
1.143 schwarze 1676: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1677: const struct roff_node *n)
1.1 schwarze 1678: {
1679:
1.141 schwarze 1680: if (n->type == ROFFT_BODY)
1.147 schwarze 1681: deroff(&mpage->desc, n);
1.150 schwarze 1682: return 0;
1.1 schwarze 1683: }
1684:
1.19 schwarze 1685: static int
1.143 schwarze 1686: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1687: const struct roff_node *n)
1.1 schwarze 1688: {
1689:
1.86 schwarze 1690: if (SEC_NAME == n->sec)
1.184 schwarze 1691: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.141 schwarze 1692: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.127 schwarze 1693: if (n->child == NULL)
1694: putkey(mpage, meta->name, NAME_SYN);
1695: else
1.184 schwarze 1696: putmdockey(mpage, n->child, NAME_SYN, 0);
1.129 schwarze 1697: }
1698: if ( ! (mpage->name_head_done ||
1699: n->child == NULL || n->child->string == NULL ||
1700: strcasecmp(n->child->string, meta->title))) {
1.175 schwarze 1701: putkey(mpage, n->child->string, NAME_HEAD);
1.129 schwarze 1702: mpage->name_head_done = 1;
1.127 schwarze 1703: }
1.150 schwarze 1704: return 0;
1.1 schwarze 1705: }
1706:
1.19 schwarze 1707: static int
1.143 schwarze 1708: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1709: const struct roff_node *n)
1.1 schwarze 1710: {
1711:
1.150 schwarze 1712: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 schwarze 1713: }
1714:
1.47 schwarze 1715: static int
1.143 schwarze 1716: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1717: const struct roff_node *n)
1.1 schwarze 1718: {
1719:
1.150 schwarze 1720: return n->type == ROFFT_HEAD;
1.1 schwarze 1721: }
1722:
1.47 schwarze 1723: /*
1724: * Add a string to the hash table for the current manual.
1725: * Each string has a bitmask telling which macros it belongs to.
1726: * When we finish the manual, we'll dump the table.
1727: */
1.1 schwarze 1728: static void
1.131 schwarze 1729: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1730: {
1.90 schwarze 1731: struct ohash *htab;
1.47 schwarze 1732: struct str *s;
1.68 schwarze 1733: const char *end;
1.47 schwarze 1734: unsigned int slot;
1.131 schwarze 1735: int i, mustfree;
1.1 schwarze 1736:
1.47 schwarze 1737: if (0 == sz)
1738: return;
1.68 schwarze 1739:
1.131 schwarze 1740: mustfree = render_string(&cp, &sz);
1741:
1.90 schwarze 1742: if (TYPE_Nm & v) {
1743: htab = &names;
1744: v &= name_mask;
1.124 schwarze 1745: if (v & NAME_FIRST)
1746: name_mask &= ~NAME_FIRST;
1.90 schwarze 1747: if (debug > 1)
1748: say(mpage->mlinks->file,
1.183 schwarze 1749: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1750: (unsigned long long)v);
1.90 schwarze 1751: } else {
1752: htab = &strings;
1753: if (debug > 1)
1.173 schwarze 1754: for (i = 0; i < KEY_MAX; i++)
1.118 schwarze 1755: if ((uint64_t)1 << i & v)
1.90 schwarze 1756: say(mpage->mlinks->file,
1757: "Adding key %s=%*s",
1.172 schwarze 1758: mansearch_keynames[i], (int)sz, cp);
1.68 schwarze 1759: }
1.47 schwarze 1760:
1761: end = cp + sz;
1.90 schwarze 1762: slot = ohash_qlookupi(htab, cp, &end);
1763: s = ohash_find(htab, slot);
1.1 schwarze 1764:
1.47 schwarze 1765: if (NULL != s && mpage == s->mpage) {
1766: s->mask |= v;
1.1 schwarze 1767: return;
1.47 schwarze 1768: } else if (NULL == s) {
1.101 schwarze 1769: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1770: memcpy(s->key, cp, sz);
1.90 schwarze 1771: ohash_insert(htab, slot, s);
1.47 schwarze 1772: }
1773: s->mpage = mpage;
1774: s->mask = v;
1.131 schwarze 1775:
1776: if (mustfree)
1777: free(cp);
1.1 schwarze 1778: }
1779:
1780: /*
1.47 schwarze 1781: * Take a Unicode codepoint and produce its UTF-8 encoding.
1782: * This isn't the best way to do this, but it works.
1783: * The magic numbers are from the UTF-8 packaging.
1784: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1785: */
1.47 schwarze 1786: static size_t
1787: utf8(unsigned int cp, char out[7])
1.1 schwarze 1788: {
1.47 schwarze 1789: size_t rc;
1.1 schwarze 1790:
1.47 schwarze 1791: rc = 0;
1792: if (cp <= 0x0000007F) {
1793: rc = 1;
1794: out[0] = (char)cp;
1795: } else if (cp <= 0x000007FF) {
1796: rc = 2;
1797: out[0] = (cp >> 6 & 31) | 192;
1798: out[1] = (cp & 63) | 128;
1799: } else if (cp <= 0x0000FFFF) {
1800: rc = 3;
1801: out[0] = (cp >> 12 & 15) | 224;
1802: out[1] = (cp >> 6 & 63) | 128;
1803: out[2] = (cp & 63) | 128;
1804: } else if (cp <= 0x001FFFFF) {
1805: rc = 4;
1806: out[0] = (cp >> 18 & 7) | 240;
1807: out[1] = (cp >> 12 & 63) | 128;
1808: out[2] = (cp >> 6 & 63) | 128;
1809: out[3] = (cp & 63) | 128;
1810: } else if (cp <= 0x03FFFFFF) {
1811: rc = 5;
1812: out[0] = (cp >> 24 & 3) | 248;
1813: out[1] = (cp >> 18 & 63) | 128;
1814: out[2] = (cp >> 12 & 63) | 128;
1815: out[3] = (cp >> 6 & 63) | 128;
1816: out[4] = (cp & 63) | 128;
1817: } else if (cp <= 0x7FFFFFFF) {
1818: rc = 6;
1819: out[0] = (cp >> 30 & 1) | 252;
1820: out[1] = (cp >> 24 & 63) | 128;
1821: out[2] = (cp >> 18 & 63) | 128;
1822: out[3] = (cp >> 12 & 63) | 128;
1823: out[4] = (cp >> 6 & 63) | 128;
1824: out[5] = (cp & 63) | 128;
1825: } else
1.150 schwarze 1826: return 0;
1.19 schwarze 1827:
1.47 schwarze 1828: out[rc] = '\0';
1.150 schwarze 1829: return rc;
1.1 schwarze 1830: }
1831:
1.47 schwarze 1832: /*
1.131 schwarze 1833: * If the string contains escape sequences,
1834: * replace it with an allocated rendering and return 1,
1835: * such that the caller can free it after use.
1836: * Otherwise, do nothing and return 0.
1.47 schwarze 1837: */
1.131 schwarze 1838: static int
1839: render_string(char **public, size_t *psz)
1.1 schwarze 1840: {
1.131 schwarze 1841: const char *src, *scp, *addcp, *seq;
1842: char *dst;
1843: size_t ssz, dsz, addsz;
1.71 schwarze 1844: char utfbuf[7], res[6];
1.131 schwarze 1845: int seqlen, unicode;
1.47 schwarze 1846:
1847: res[0] = '\\';
1848: res[1] = '\t';
1849: res[2] = ASCII_NBRSP;
1850: res[3] = ASCII_HYPH;
1.71 schwarze 1851: res[4] = ASCII_BREAK;
1852: res[5] = '\0';
1.1 schwarze 1853:
1.131 schwarze 1854: src = scp = *public;
1855: ssz = *psz;
1856: dst = NULL;
1857: dsz = 0;
1858:
1859: while (scp < src + *psz) {
1860:
1861: /* Leave normal characters unchanged. */
1862:
1863: if (strchr(res, *scp) == NULL) {
1864: if (dst != NULL)
1865: dst[dsz++] = *scp;
1866: scp++;
1867: continue;
1868: }
1.1 schwarze 1869:
1.47 schwarze 1870: /*
1.131 schwarze 1871: * Found something that requires replacing,
1872: * make sure we have a destination buffer.
1.47 schwarze 1873: */
1.131 schwarze 1874:
1875: if (dst == NULL) {
1876: dst = mandoc_malloc(ssz + 1);
1877: dsz = scp - src;
1878: memcpy(dst, src, dsz);
1.47 schwarze 1879: }
1.39 schwarze 1880:
1.131 schwarze 1881: /* Handle single-char special characters. */
1882:
1883: switch (*scp) {
1884: case '\\':
1885: break;
1.98 schwarze 1886: case '\t':
1887: case ASCII_NBRSP:
1.131 schwarze 1888: dst[dsz++] = ' ';
1889: scp++;
1890: continue;
1891: case ASCII_HYPH:
1892: dst[dsz++] = '-';
1.71 schwarze 1893: /* FALLTHROUGH */
1.98 schwarze 1894: case ASCII_BREAK:
1.131 schwarze 1895: scp++;
1.47 schwarze 1896: continue;
1.71 schwarze 1897: default:
1.131 schwarze 1898: abort();
1.71 schwarze 1899: }
1.39 schwarze 1900:
1.47 schwarze 1901: /*
1.131 schwarze 1902: * Found an escape sequence.
1903: * Read past the slash, then parse it.
1904: * Ignore everything except characters.
1.47 schwarze 1905: */
1.52 schwarze 1906:
1.131 schwarze 1907: scp++;
1908: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1909: continue;
1.39 schwarze 1910:
1.47 schwarze 1911: /*
1.52 schwarze 1912: * Render the special character
1913: * as either UTF-8 or ASCII.
1.47 schwarze 1914: */
1.52 schwarze 1915:
1916: if (write_utf8) {
1.157 schwarze 1917: unicode = mchars_spec2cp(seq, seqlen);
1.131 schwarze 1918: if (unicode <= 0)
1.52 schwarze 1919: continue;
1.131 schwarze 1920: addsz = utf8(unicode, utfbuf);
1921: if (addsz == 0)
1.52 schwarze 1922: continue;
1.131 schwarze 1923: addcp = utfbuf;
1.52 schwarze 1924: } else {
1.157 schwarze 1925: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.131 schwarze 1926: if (addcp == NULL)
1.52 schwarze 1927: continue;
1.131 schwarze 1928: if (*addcp == ASCII_NBRSP) {
1929: addcp = " ";
1930: addsz = 1;
1.52 schwarze 1931: }
1932: }
1.1 schwarze 1933:
1.47 schwarze 1934: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1935:
1.131 schwarze 1936: ssz += addsz;
1937: dst = mandoc_realloc(dst, ssz + 1);
1938: memcpy(dst + dsz, addcp, addsz);
1939: dsz += addsz;
1940: }
1941: if (dst != NULL) {
1942: *public = dst;
1943: *psz = dsz;
1944: }
1945:
1946: /* Trim trailing whitespace and NUL-terminate. */
1947:
1948: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1949: --*psz;
1950: if (dst != NULL) {
1951: (*public)[*psz] = '\0';
1.150 schwarze 1952: return 1;
1.131 schwarze 1953: } else
1.150 schwarze 1954: return 0;
1.1 schwarze 1955: }
1956:
1.75 schwarze 1957: static void
1958: dbadd_mlink(const struct mlink *mlink)
1959: {
1.173 schwarze 1960: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
1961: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1962: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1963: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
1964: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.75 schwarze 1965: }
1966:
1.11 schwarze 1967: /*
1.47 schwarze 1968: * Flush the current page's terms (and their bits) into the database.
1.53 schwarze 1969: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1970: */
1971: static void
1.173 schwarze 1972: dbadd(struct dba *dba, struct mpage *mpage)
1.11 schwarze 1973: {
1.47 schwarze 1974: struct mlink *mlink;
1975: struct str *key;
1.131 schwarze 1976: char *cp;
1.173 schwarze 1977: uint64_t mask;
1.47 schwarze 1978: size_t i;
1979: unsigned int slot;
1.131 schwarze 1980: int mustfree;
1.47 schwarze 1981:
1.85 schwarze 1982: mlink = mpage->mlinks;
1.11 schwarze 1983:
1.85 schwarze 1984: if (nodb) {
1.104 schwarze 1985: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 1986: key = ohash_next(&names, &slot))
1.104 schwarze 1987: free(key);
1988: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 1989: key = ohash_next(&strings, &slot))
1.104 schwarze 1990: free(key);
1.102 schwarze 1991: if (0 == debug)
1992: return;
1.85 schwarze 1993: while (NULL != mlink) {
1994: fputs(mlink->name, stdout);
1995: if (NULL == mlink->next ||
1996: strcmp(mlink->dsec, mlink->next->dsec) ||
1997: strcmp(mlink->fsec, mlink->next->fsec) ||
1998: strcmp(mlink->arch, mlink->next->arch)) {
1999: putchar('(');
2000: if ('\0' == *mlink->dsec)
2001: fputs(mlink->fsec, stdout);
2002: else
2003: fputs(mlink->dsec, stdout);
2004: if ('\0' != *mlink->arch)
2005: printf("/%s", mlink->arch);
2006: putchar(')');
2007: }
2008: mlink = mlink->next;
2009: if (NULL != mlink)
2010: fputs(", ", stdout);
2011: }
1.89 schwarze 2012: printf(" - %s\n", mpage->desc);
1.11 schwarze 2013: return;
1.85 schwarze 2014: }
2015:
2016: if (debug)
2017: say(mlink->file, "Adding to database");
1.47 schwarze 2018:
1.131 schwarze 2019: cp = mpage->desc;
2020: i = strlen(cp);
2021: mustfree = render_string(&cp, &i);
1.176 schwarze 2022: mpage->dba = dba_page_new(dba->pages,
1.173 schwarze 2023: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2024: cp, mlink->file, mpage->form);
1.131 schwarze 2025: if (mustfree)
2026: free(cp);
1.176 schwarze 2027: dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
1.47 schwarze 2028:
1.173 schwarze 2029: while (mlink != NULL) {
1.75 schwarze 2030: dbadd_mlink(mlink);
1.85 schwarze 2031: mlink = mlink->next;
2032: }
1.47 schwarze 2033:
1.90 schwarze 2034: for (key = ohash_first(&names, &slot); NULL != key;
2035: key = ohash_next(&names, &slot)) {
2036: assert(key->mpage == mpage);
1.173 schwarze 2037: dba_page_alias(mpage->dba, key->key, key->mask);
1.90 schwarze 2038: free(key);
2039: }
1.47 schwarze 2040: for (key = ohash_first(&strings, &slot); NULL != key;
2041: key = ohash_next(&strings, &slot)) {
2042: assert(key->mpage == mpage);
1.173 schwarze 2043: i = 0;
2044: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2045: if (key->mask & mask)
2046: dba_macro_add(dba->macros, i,
2047: key->key, mpage->dba);
2048: i++;
2049: }
1.47 schwarze 2050: free(key);
1.33 schwarze 2051: }
1.47 schwarze 2052: }
1.41 deraadt 2053:
1.47 schwarze 2054: static void
1.173 schwarze 2055: dbprune(struct dba *dba)
1.47 schwarze 2056: {
1.173 schwarze 2057: struct dba_array *page, *files;
2058: char *file;
1.11 schwarze 2059:
1.173 schwarze 2060: dba_array_FOREACH(dba->pages, page) {
2061: files = dba_array_get(page, DBP_FILE);
2062: dba_array_FOREACH(files, file) {
2063: if (*file < ' ')
2064: file++;
2065: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2066: file)) != NULL) {
2067: if (debug)
2068: say(file, "Deleting from database");
2069: dba_array_del(dba->pages);
2070: break;
2071: }
1.63 schwarze 2072: }
1.11 schwarze 2073: }
1.47 schwarze 2074: }
1.22 schwarze 2075:
1.47 schwarze 2076: /*
1.173 schwarze 2077: * Write the database from memory to disk.
1.47 schwarze 2078: */
2079: static void
1.173 schwarze 2080: dbwrite(struct dba *dba)
1.47 schwarze 2081: {
1.173 schwarze 2082: char tfn[32];
1.72 schwarze 2083: int status;
2084: pid_t child;
1.11 schwarze 2085:
1.173 schwarze 2086: if (dba_write(MANDOC_DB "~", dba) != -1) {
2087: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2088: exitcode = (int)MANDOCLEVEL_SYSERR;
2089: say(MANDOC_DB, "&rename");
2090: unlink(MANDOC_DB "~");
2091: }
1.47 schwarze 2092: return;
1.28 schwarze 2093: }
1.22 schwarze 2094:
1.173 schwarze 2095: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2096: if (mkdtemp(tfn) == NULL) {
2097: exitcode = (int)MANDOCLEVEL_SYSERR;
2098: say("", "&%s", tfn);
1.47 schwarze 2099: return;
1.173 schwarze 2100: }
1.22 schwarze 2101:
1.173 schwarze 2102: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2103: if (dba_write(tfn, dba) == -1) {
2104: exitcode = (int)MANDOCLEVEL_SYSERR;
2105: say(tfn, "&dba_write");
2106: goto out;
1.72 schwarze 2107: }
2108:
2109: switch (child = fork()) {
1.98 schwarze 2110: case -1:
1.72 schwarze 2111: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2112: say("", "&fork cmp");
1.72 schwarze 2113: return;
1.98 schwarze 2114: case 0:
1.173 schwarze 2115: execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL);
1.80 schwarze 2116: say("", "&exec cmp");
1.72 schwarze 2117: exit(0);
2118: default:
2119: break;
2120: }
1.173 schwarze 2121: if (waitpid(child, &status, 0) == -1) {
1.72 schwarze 2122: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2123: say("", "&wait cmp");
1.72 schwarze 2124: } else if (WIFSIGNALED(status)) {
2125: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2126: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2127: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2128: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2129: say(MANDOC_DB,
2130: "Data changed, but cannot replace database");
2131: }
2132:
1.173 schwarze 2133: out:
2134: *strrchr(tfn, '/') = '\0';
1.72 schwarze 2135: switch (child = fork()) {
1.98 schwarze 2136: case -1:
1.72 schwarze 2137: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2138: say("", "&fork rm");
1.72 schwarze 2139: return;
1.98 schwarze 2140: case 0:
1.173 schwarze 2141: execlp("rm", "rm", "-rf", tfn, (char *)NULL);
1.80 schwarze 2142: say("", "&exec rm");
1.72 schwarze 2143: exit((int)MANDOCLEVEL_SYSERR);
2144: default:
2145: break;
2146: }
1.173 schwarze 2147: if (waitpid(child, &status, 0) == -1) {
1.72 schwarze 2148: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2149: say("", "&wait rm");
1.72 schwarze 2150: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2151: exitcode = (int)MANDOCLEVEL_SYSERR;
1.173 schwarze 2152: say("", "%s: Cannot remove temporary directory", tfn);
1.22 schwarze 2153: }
1.47 schwarze 2154: }
1.6 schwarze 2155:
1.47 schwarze 2156: static int
1.120 schwarze 2157: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2158: {
2159: static char startdir[PATH_MAX];
1.110 schwarze 2160: static int getcwd_status; /* 1 = ok, 2 = failure */
2161: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2162: char *cp;
1.6 schwarze 2163:
1.47 schwarze 2164: /*
1.110 schwarze 2165: * Remember the original working directory, if possible.
2166: * This will be needed if the second or a later directory
2167: * on the command line is given as a relative path.
2168: * Do not error out if the current directory is not
2169: * searchable: Maybe it won't be needed after all.
2170: */
2171: if (0 == getcwd_status) {
2172: if (NULL == getcwd(startdir, sizeof(startdir))) {
2173: getcwd_status = 2;
2174: (void)strlcpy(startdir, strerror(errno),
2175: sizeof(startdir));
2176: } else
2177: getcwd_status = 1;
2178: }
2179:
2180: /*
2181: * We are leaving the old base directory.
2182: * Do not use it any longer, not even for messages.
2183: */
2184: *basedir = '\0';
2185:
2186: /*
2187: * If and only if the directory was changed earlier and
2188: * the next directory to process is given as a relative path,
2189: * first go back, or bail out if that is impossible.
1.47 schwarze 2190: */
1.110 schwarze 2191: if (chdir_status && '/' != *targetdir) {
2192: if (2 == getcwd_status) {
1.47 schwarze 2193: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2194: say("", "getcwd: %s", startdir);
1.150 schwarze 2195: return 0;
1.47 schwarze 2196: }
1.110 schwarze 2197: if (-1 == chdir(startdir)) {
1.47 schwarze 2198: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2199: say("", "&chdir %s", startdir);
1.150 schwarze 2200: return 0;
1.2 schwarze 2201: }
2202: }
1.110 schwarze 2203:
2204: /*
2205: * Always resolve basedir to the canonicalized absolute
2206: * pathname and append a trailing slash, such that
2207: * we can reliably check whether files are inside.
2208: */
1.47 schwarze 2209: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2210: if (report_baddir || errno != ENOENT) {
2211: exitcode = (int)MANDOCLEVEL_BADARG;
2212: say("", "&%s: realpath", targetdir);
2213: }
1.150 schwarze 2214: return 0;
1.47 schwarze 2215: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2216: if (report_baddir || errno != ENOENT) {
2217: exitcode = (int)MANDOCLEVEL_BADARG;
2218: say("", "&chdir");
2219: }
1.150 schwarze 2220: return 0;
1.109 schwarze 2221: }
1.110 schwarze 2222: chdir_status = 1;
1.109 schwarze 2223: cp = strchr(basedir, '\0');
2224: if ('/' != cp[-1]) {
2225: if (cp - basedir >= PATH_MAX - 1) {
2226: exitcode = (int)MANDOCLEVEL_SYSERR;
2227: say("", "Filename too long");
1.150 schwarze 2228: return 0;
1.109 schwarze 2229: }
2230: *cp++ = '/';
2231: *cp = '\0';
1.47 schwarze 2232: }
1.150 schwarze 2233: return 1;
1.2 schwarze 2234: }
2235:
2236: static void
1.47 schwarze 2237: say(const char *file, const char *format, ...)
1.2 schwarze 2238: {
1.47 schwarze 2239: va_list ap;
1.80 schwarze 2240: int use_errno;
1.2 schwarze 2241:
1.47 schwarze 2242: if ('\0' != *basedir)
2243: fprintf(stderr, "%s", basedir);
2244: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2245: fputc('/', stderr);
1.47 schwarze 2246: if ('\0' != *file)
2247: fprintf(stderr, "%s", file);
1.31 schwarze 2248:
1.80 schwarze 2249: use_errno = 1;
2250: if (NULL != format) {
2251: switch (*format) {
1.98 schwarze 2252: case '&':
1.80 schwarze 2253: format++;
2254: break;
1.98 schwarze 2255: case '\0':
1.80 schwarze 2256: format = NULL;
2257: break;
2258: default:
2259: use_errno = 0;
2260: break;
2261: }
2262: }
2263: if (NULL != format) {
2264: if ('\0' != *basedir || '\0' != *file)
2265: fputs(": ", stderr);
2266: va_start(ap, format);
2267: vfprintf(stderr, format, ap);
2268: va_end(ap);
2269: }
2270: if (use_errno) {
2271: if ('\0' != *basedir || '\0' != *file || NULL != format)
2272: fputs(": ", stderr);
1.47 schwarze 2273: perror(NULL);
1.80 schwarze 2274: } else
2275: fputc('\n', stderr);
1.1 schwarze 2276: }