Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.195
1.195 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.194 2017/03/03 13:41:28 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:
1.195 ! schwarze 171: static const struct mdoc_handler __mdocs[MDOC_MAX - MDOC_Dd] = {
1.184 schwarze 172: { NULL, 0, NODE_NOPRT }, /* Dd */
173: { NULL, 0, NODE_NOPRT }, /* Dt */
174: { NULL, 0, NODE_NOPRT }, /* Os */
175: { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
176: { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
177: { NULL, 0, 0 }, /* Pp */
178: { NULL, 0, 0 }, /* D1 */
179: { NULL, 0, 0 }, /* Dl */
180: { NULL, 0, 0 }, /* Bd */
181: { NULL, 0, 0 }, /* Ed */
182: { NULL, 0, 0 }, /* Bl */
183: { NULL, 0, 0 }, /* El */
184: { NULL, 0, 0 }, /* It */
185: { NULL, 0, 0 }, /* Ad */
186: { NULL, TYPE_An, 0 }, /* An */
1.195 ! schwarze 187: { NULL, 0, 0 }, /* Ap */
1.184 schwarze 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: };
1.195 ! schwarze 296: static const struct mdoc_handler *const mdocs = __mdocs - MDOC_Dd;
1.1 schwarze 297:
1.98 schwarze 298:
1.1 schwarze 299: int
1.3 schwarze 300: mandocdb(int argc, char *argv[])
1.1 schwarze 301: {
1.140 schwarze 302: struct manconf conf;
303: struct mparse *mp;
1.173 schwarze 304: struct dba *dba;
1.161 schwarze 305: const char *path_arg, *progname;
1.140 schwarze 306: size_t j, sz;
1.47 schwarze 307: int ch, i;
308:
1.158 millert 309: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.167 schwarze 310: warn("pledge");
1.155 schwarze 311: return (int)MANDOCLEVEL_SYSERR;
312: }
313:
1.140 schwarze 314: memset(&conf, 0, sizeof(conf));
1.47 schwarze 315:
316: /*
1.98 schwarze 317: * We accept a few different invocations.
1.47 schwarze 318: * The CHECKOP macro makes sure that invocation styles don't
319: * clobber each other.
320: */
321: #define CHECKOP(_op, _ch) do \
322: if (OP_DEFAULT != (_op)) { \
1.151 schwarze 323: warnx("-%c: Conflicting option", (_ch)); \
1.47 schwarze 324: goto usage; \
325: } while (/*CONSTCOND*/0)
1.10 schwarze 326:
1.47 schwarze 327: path_arg = NULL;
1.28 schwarze 328: op = OP_DEFAULT;
1.1 schwarze 329:
1.83 schwarze 330: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 331: switch (ch) {
1.98 schwarze 332: case 'a':
1.6 schwarze 333: use_all = 1;
334: break;
1.98 schwarze 335: case 'C':
1.47 schwarze 336: CHECKOP(op, ch);
337: path_arg = optarg;
1.28 schwarze 338: op = OP_CONFFILE;
1.25 schwarze 339: break;
1.98 schwarze 340: case 'D':
1.82 schwarze 341: debug++;
342: break;
1.98 schwarze 343: case 'd':
1.47 schwarze 344: CHECKOP(op, ch);
345: path_arg = optarg;
1.2 schwarze 346: op = OP_UPDATE;
1.1 schwarze 347: break;
1.98 schwarze 348: case 'n':
1.47 schwarze 349: nodb = 1;
350: break;
1.98 schwarze 351: case 'p':
1.83 schwarze 352: warnings = 1;
353: break;
1.98 schwarze 354: case 'Q':
1.73 schwarze 355: mparse_options |= MPARSE_QUICK;
1.59 schwarze 356: break;
1.98 schwarze 357: case 'T':
1.52 schwarze 358: if (strcmp(optarg, "utf8")) {
1.151 schwarze 359: warnx("-T%s: Unsupported output format",
360: optarg);
1.52 schwarze 361: goto usage;
362: }
363: write_utf8 = 1;
364: break;
1.98 schwarze 365: case 't':
1.47 schwarze 366: CHECKOP(op, ch);
1.28 schwarze 367: dup2(STDOUT_FILENO, STDERR_FILENO);
368: op = OP_TEST;
1.47 schwarze 369: nodb = warnings = 1;
1.28 schwarze 370: break;
1.98 schwarze 371: case 'u':
1.47 schwarze 372: CHECKOP(op, ch);
373: path_arg = optarg;
1.1 schwarze 374: op = OP_DELETE;
375: break;
1.98 schwarze 376: case 'v':
1.83 schwarze 377: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 378: break;
1.1 schwarze 379: default:
1.28 schwarze 380: goto usage;
1.1 schwarze 381: }
382:
383: argc -= optind;
384: argv += optind;
385:
1.163 schwarze 386: if (nodb) {
387: if (pledge("stdio rpath", NULL) == -1) {
1.167 schwarze 388: warn("pledge");
1.163 schwarze 389: return (int)MANDOCLEVEL_SYSERR;
390: }
1.155 schwarze 391: }
392:
1.28 schwarze 393: if (OP_CONFFILE == op && argc > 0) {
1.151 schwarze 394: warnx("-C: Too many arguments");
1.28 schwarze 395: goto usage;
396: }
397:
1.47 schwarze 398: exitcode = (int)MANDOCLEVEL_OK;
1.157 schwarze 399: mchars_alloc();
400: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.156 schwarze 401: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
402: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.2 schwarze 403:
1.47 schwarze 404: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 405:
1.47 schwarze 406: /*
1.109 schwarze 407: * Most of these deal with a specific directory.
1.97 schwarze 408: * Jump into that directory first.
1.47 schwarze 409: */
1.120 schwarze 410: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 411: goto out;
1.97 schwarze 412:
1.174 schwarze 413: dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
414: if (dba != NULL) {
1.97 schwarze 415: /*
416: * The existing database is usable. Process
417: * all files specified on the command-line.
418: */
1.163 schwarze 419: if (!nodb) {
420: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.167 schwarze 421: warn("pledge");
1.163 schwarze 422: exitcode = (int)MANDOCLEVEL_SYSERR;
423: goto out;
424: }
1.155 schwarze 425: }
1.97 schwarze 426: use_all = 1;
1.95 schwarze 427: for (i = 0; i < argc; i++)
428: filescan(argv[i]);
1.174 schwarze 429: if (nodb == 0)
1.173 schwarze 430: dbprune(dba);
1.95 schwarze 431: } else {
1.177 schwarze 432: /* Database missing or corrupt. */
1.179 schwarze 433: if (op != OP_UPDATE || errno != ENOENT)
434: say(MANDOC_DB, "%s: Automatically recreating"
435: " from scratch", strerror(errno));
1.97 schwarze 436: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 437: op = OP_DEFAULT;
438: if (0 == treescan())
439: goto out;
1.173 schwarze 440: dba = dba_new(128);
1.95 schwarze 441: }
1.47 schwarze 442: if (OP_DELETE != op)
1.173 schwarze 443: mpages_merge(dba, mp);
444: if (nodb == 0)
445: dbwrite(dba);
446: dba_free(dba);
1.47 schwarze 447: } else {
448: /*
449: * If we have arguments, use them as our manpaths.
1.181 schwarze 450: * If we don't, use man.conf(5).
1.47 schwarze 451: */
452: if (argc > 0) {
1.140 schwarze 453: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 454: argc, sizeof(char *));
1.140 schwarze 455: conf.manpath.sz = (size_t)argc;
1.47 schwarze 456: for (i = 0; i < argc; i++)
1.140 schwarze 457: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 458: } else
1.140 schwarze 459: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 460:
1.140 schwarze 461: if (conf.manpath.sz == 0) {
1.84 schwarze 462: exitcode = (int)MANDOCLEVEL_BADARG;
463: say("", "Empty manpath");
464: }
1.2 schwarze 465:
1.47 schwarze 466: /*
467: * First scan the tree rooted at a base directory, then
468: * build a new database and finally move it into place.
469: * Ignore zero-length directories and strip trailing
470: * slashes.
471: */
1.140 schwarze 472: for (j = 0; j < conf.manpath.sz; j++) {
473: sz = strlen(conf.manpath.paths[j]);
474: if (sz && conf.manpath.paths[j][sz - 1] == '/')
475: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 476: if (0 == sz)
477: continue;
1.2 schwarze 478:
1.47 schwarze 479: if (j) {
1.156 schwarze 480: mandoc_ohash_init(&mpages, 6,
481: offsetof(struct mpage, inodev));
482: mandoc_ohash_init(&mlinks, 6,
483: offsetof(struct mlink, file));
1.47 schwarze 484: }
1.1 schwarze 485:
1.140 schwarze 486: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 487: continue;
1.47 schwarze 488: if (0 == treescan())
1.120 schwarze 489: continue;
1.173 schwarze 490: dba = dba_new(128);
491: mpages_merge(dba, mp);
492: if (nodb == 0)
493: dbwrite(dba);
494: dba_free(dba);
1.28 schwarze 495:
1.140 schwarze 496: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 497: mpages_free();
498: ohash_delete(&mpages);
499: ohash_delete(&mlinks);
500: }
1.2 schwarze 501: }
1.47 schwarze 502: }
503: out:
1.140 schwarze 504: manconf_free(&conf);
1.123 schwarze 505: mparse_free(mp);
1.157 schwarze 506: mchars_free();
1.47 schwarze 507: mpages_free();
508: ohash_delete(&mpages);
509: ohash_delete(&mlinks);
1.150 schwarze 510: return exitcode;
1.47 schwarze 511: usage:
1.161 schwarze 512: progname = getprogname();
1.83 schwarze 513: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
514: " %s [-aDnpQ] [-Tutf8] dir ...\n"
515: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
516: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 517: " %s [-Q] -t file ...\n",
1.161 schwarze 518: progname, progname, progname, progname, progname);
1.1 schwarze 519:
1.150 schwarze 520: return (int)MANDOCLEVEL_BADARG;
1.47 schwarze 521: }
1.1 schwarze 522:
1.47 schwarze 523: /*
1.180 schwarze 524: * To get a singly linked list in alpha order while inserting entries
525: * at the beginning, process directory entries in reverse alpha order.
526: */
527: static int
528: fts_compare(const FTSENT **a, const FTSENT **b)
529: {
530: return -strcmp((*a)->fts_name, (*b)->fts_name);
531: }
532:
533: /*
1.47 schwarze 534: * Scan a directory tree rooted at "basedir" for manpages.
535: * We use fts(), scanning directory parts along the way for clues to our
536: * section and architecture.
537: *
538: * If use_all has been specified, grok all files.
539: * If not, sanitise paths to the following:
540: *
1.98 schwarze 541: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 542: * or
543: * [./]cat<section>[/<arch>]/<name>.0
544: *
1.169 krw 545: * TODO: accommodate for multi-language directories.
1.47 schwarze 546: */
547: static int
548: treescan(void)
549: {
1.96 schwarze 550: char buf[PATH_MAX];
1.47 schwarze 551: FTS *f;
552: FTSENT *ff;
553: struct mlink *mlink;
1.173 schwarze 554: int gzip;
555: enum form dform;
1.51 schwarze 556: char *dsec, *arch, *fsec, *cp;
557: const char *path;
1.47 schwarze 558: const char *argv[2];
1.1 schwarze 559:
1.47 schwarze 560: argv[0] = ".";
1.193 schwarze 561: argv[1] = NULL;
1.1 schwarze 562:
1.180 schwarze 563: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
564: fts_compare);
1.153 schwarze 565: if (f == NULL) {
1.47 schwarze 566: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 567: say("", "&fts_open");
1.150 schwarze 568: return 0;
1.47 schwarze 569: }
1.2 schwarze 570:
1.47 schwarze 571: dsec = arch = NULL;
572: dform = FORM_NONE;
1.2 schwarze 573:
1.153 schwarze 574: while ((ff = fts_read(f)) != NULL) {
1.47 schwarze 575: path = ff->fts_path + 2;
1.96 schwarze 576: switch (ff->fts_info) {
577:
578: /*
579: * Symbolic links require various sanity checks,
580: * then get handled just like regular files.
581: */
1.98 schwarze 582: case FTS_SL:
1.153 schwarze 583: if (realpath(path, buf) == NULL) {
1.96 schwarze 584: if (warnings)
585: say(path, "&realpath");
586: continue;
587: }
588: if (strstr(buf, basedir) != buf) {
589: if (warnings) say("",
590: "%s: outside base directory", buf);
591: continue;
592: }
593: /* Use logical inode to avoid mpages dupe. */
1.153 schwarze 594: if (stat(path, ff->fts_statp) == -1) {
1.96 schwarze 595: if (warnings)
596: say(path, "&stat");
597: continue;
598: }
599: /* FALLTHROUGH */
600:
1.15 schwarze 601: /*
1.47 schwarze 602: * If we're a regular file, add an mlink by using the
603: * stored directory data and handling the filename.
1.15 schwarze 604: */
1.98 schwarze 605: case FTS_F:
1.153 schwarze 606: if ( ! strcmp(path, MANDOC_DB))
1.47 schwarze 607: continue;
608: if ( ! use_all && ff->fts_level < 2) {
609: if (warnings)
610: say(path, "Extraneous file");
611: continue;
1.81 schwarze 612: }
613: gzip = 0;
614: fsec = NULL;
1.153 schwarze 615: while (fsec == NULL) {
1.81 schwarze 616: fsec = strrchr(ff->fts_name, '.');
1.153 schwarze 617: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.81 schwarze 618: break;
619: gzip = 1;
620: *fsec = '\0';
621: fsec = NULL;
622: }
1.153 schwarze 623: if (fsec == NULL) {
1.47 schwarze 624: if ( ! use_all) {
625: if (warnings)
626: say(path,
627: "No filename suffix");
628: continue;
629: }
1.153 schwarze 630: } else if ( ! strcmp(++fsec, "html")) {
1.47 schwarze 631: if (warnings)
632: say(path, "Skip html");
633: continue;
1.153 schwarze 634: } else if ( ! strcmp(fsec, "ps")) {
1.47 schwarze 635: if (warnings)
636: say(path, "Skip ps");
637: continue;
1.153 schwarze 638: } else if ( ! strcmp(fsec, "pdf")) {
1.47 schwarze 639: if (warnings)
640: say(path, "Skip pdf");
641: continue;
642: } else if ( ! use_all &&
1.153 schwarze 643: ((dform == FORM_SRC &&
1.138 schwarze 644: strncmp(fsec, dsec, strlen(dsec))) ||
1.153 schwarze 645: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.47 schwarze 646: if (warnings)
647: say(path, "Wrong filename suffix");
648: continue;
649: } else
650: fsec[-1] = '\0';
1.51 schwarze 651:
1.47 schwarze 652: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 653: if (strlcpy(mlink->file, path,
654: sizeof(mlink->file)) >=
655: sizeof(mlink->file)) {
656: say(path, "Filename too long");
657: free(mlink);
658: continue;
659: }
1.47 schwarze 660: mlink->dform = dform;
1.51 schwarze 661: mlink->dsec = dsec;
662: mlink->arch = arch;
663: mlink->name = ff->fts_name;
664: mlink->fsec = fsec;
1.81 schwarze 665: mlink->gzip = gzip;
1.47 schwarze 666: mlink_add(mlink, ff->fts_statp);
667: continue;
1.96 schwarze 668:
1.98 schwarze 669: case FTS_D:
670: case FTS_DP:
1.96 schwarze 671: break;
672:
673: default:
1.47 schwarze 674: if (warnings)
675: say(path, "Not a regular file");
676: continue;
677: }
678:
679: switch (ff->fts_level) {
1.98 schwarze 680: case 0:
1.47 schwarze 681: /* Ignore the root directory. */
682: break;
1.98 schwarze 683: case 1:
1.47 schwarze 684: /*
685: * This might contain manX/ or catX/.
686: * Try to infer this from the name.
687: * If we're not in use_all, enforce it.
688: */
689: cp = ff->fts_name;
1.153 schwarze 690: if (ff->fts_info == FTS_DP) {
691: dform = FORM_NONE;
692: dsec = NULL;
1.47 schwarze 693: break;
1.153 schwarze 694: }
1.15 schwarze 695:
1.153 schwarze 696: if ( ! strncmp(cp, "man", 3)) {
1.47 schwarze 697: dform = FORM_SRC;
698: dsec = cp + 3;
1.153 schwarze 699: } else if ( ! strncmp(cp, "cat", 3)) {
1.47 schwarze 700: dform = FORM_CAT;
701: dsec = cp + 3;
1.51 schwarze 702: } else {
703: dform = FORM_NONE;
704: dsec = NULL;
1.26 schwarze 705: }
1.47 schwarze 706:
1.153 schwarze 707: if (dsec != NULL || use_all)
1.47 schwarze 708: break;
709:
710: if (warnings)
711: say(path, "Unknown directory part");
712: fts_set(f, ff, FTS_SKIP);
713: break;
1.98 schwarze 714: case 2:
1.47 schwarze 715: /*
716: * Possibly our architecture.
717: * If we're descending, keep tabs on it.
718: */
1.153 schwarze 719: if (ff->fts_info != FTS_DP && dsec != NULL)
1.47 schwarze 720: arch = ff->fts_name;
1.51 schwarze 721: else
722: arch = NULL;
1.47 schwarze 723: break;
724: default:
1.153 schwarze 725: if (ff->fts_info == FTS_DP || use_all)
1.47 schwarze 726: break;
727: if (warnings)
728: say(path, "Extraneous directory part");
729: fts_set(f, ff, FTS_SKIP);
730: break;
1.14 schwarze 731: }
1.47 schwarze 732: }
733:
734: fts_close(f);
1.150 schwarze 735: return 1;
1.47 schwarze 736: }
1.1 schwarze 737:
1.47 schwarze 738: /*
739: * Add a file to the mlinks table.
740: * Do not verify that it's a "valid" looking manpage (we'll do that
741: * later).
742: *
743: * Try to infer the manual section, architecture, and page name from the
744: * path, assuming it looks like
745: *
1.98 schwarze 746: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 747: * or
748: * [./]cat<section>[/<arch>]/<name>.0
749: *
750: * See treescan() for the fts(3) version of this.
751: */
752: static void
753: filescan(const char *file)
754: {
755: char buf[PATH_MAX];
756: struct stat st;
757: struct mlink *mlink;
758: char *p, *start;
759:
760: assert(use_all);
761:
762: if (0 == strncmp(file, "./", 2))
763: file += 2;
764:
1.96 schwarze 765: /*
766: * We have to do lstat(2) before realpath(3) loses
767: * the information whether this is a symbolic link.
768: * We need to know that because for symbolic links,
769: * we want to use the orginal file name, while for
770: * regular files, we want to use the real path.
771: */
772: if (-1 == lstat(file, &st)) {
773: exitcode = (int)MANDOCLEVEL_BADARG;
774: say(file, "&lstat");
775: return;
776: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
777: exitcode = (int)MANDOCLEVEL_BADARG;
778: say(file, "Not a regular file");
779: return;
780: }
781:
782: /*
783: * We have to resolve the file name to the real path
784: * in any case for the base directory check.
785: */
1.47 schwarze 786: if (NULL == realpath(file, buf)) {
787: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 788: say(file, "&realpath");
1.47 schwarze 789: return;
1.63 schwarze 790: }
791:
1.109 schwarze 792: if (OP_TEST == op)
1.63 schwarze 793: start = buf;
1.109 schwarze 794: else if (strstr(buf, basedir) == buf)
795: start = buf + strlen(basedir);
1.63 schwarze 796: else {
1.47 schwarze 797: exitcode = (int)MANDOCLEVEL_BADARG;
798: say("", "%s: outside base directory", buf);
799: return;
1.63 schwarze 800: }
801:
1.96 schwarze 802: /*
803: * Now we are sure the file is inside our tree.
804: * If it is a symbolic link, ignore the real path
805: * and use the original name.
806: * This implies passing stuff like "cat1/../man1/foo.1"
807: * on the command line won't work. So don't do that.
808: * Note the stat(2) can still fail if the link target
809: * doesn't exist.
810: */
811: if (S_IFLNK & st.st_mode) {
812: if (-1 == stat(buf, &st)) {
813: exitcode = (int)MANDOCLEVEL_BADARG;
814: say(file, "&stat");
815: return;
816: }
1.100 schwarze 817: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
818: say(file, "Filename too long");
819: return;
820: }
1.109 schwarze 821: start = buf;
822: if (OP_TEST != op && strstr(buf, basedir) == buf)
823: start += strlen(basedir);
1.1 schwarze 824: }
1.63 schwarze 825:
1.47 schwarze 826: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 827: mlink->dform = FORM_NONE;
1.100 schwarze 828: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
829: sizeof(mlink->file)) {
830: say(start, "Filename too long");
1.134 schwarze 831: free(mlink);
1.100 schwarze 832: return;
1.188 schwarze 833: }
834:
835: /*
836: * In test mode or when the original name is absolute
837: * but outside our tree, guess the base directory.
838: */
839:
840: if (op == OP_TEST || (start == buf && *start == '/')) {
841: if (strncmp(buf, "man/", 4) == 0)
842: start = buf + 4;
843: else if ((start = strstr(buf, "/man/")) != NULL)
844: start += 5;
845: else
846: start = buf;
1.100 schwarze 847: }
1.1 schwarze 848:
1.10 schwarze 849: /*
1.47 schwarze 850: * First try to guess our directory structure.
851: * If we find a separator, try to look for man* or cat*.
852: * If we find one of these and what's underneath is a directory,
853: * assume it's an architecture.
1.10 schwarze 854: */
1.47 schwarze 855: if (NULL != (p = strchr(start, '/'))) {
856: *p++ = '\0';
857: if (0 == strncmp(start, "man", 3)) {
858: mlink->dform = FORM_SRC;
1.51 schwarze 859: mlink->dsec = start + 3;
1.47 schwarze 860: } else if (0 == strncmp(start, "cat", 3)) {
861: mlink->dform = FORM_CAT;
1.51 schwarze 862: mlink->dsec = start + 3;
1.47 schwarze 863: }
1.10 schwarze 864:
1.47 schwarze 865: start = p;
866: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
867: *p++ = '\0';
1.51 schwarze 868: mlink->arch = start;
1.47 schwarze 869: start = p;
1.41 deraadt 870: }
1.47 schwarze 871: }
1.7 schwarze 872:
1.47 schwarze 873: /*
874: * Now check the file suffix.
875: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
876: */
877: p = strrchr(start, '\0');
878: while (p-- > start && '/' != *p && '.' != *p)
879: /* Loop. */ ;
880:
881: if ('.' == *p) {
882: *p++ = '\0';
1.51 schwarze 883: mlink->fsec = p;
1.47 schwarze 884: }
1.41 deraadt 885:
1.47 schwarze 886: /*
887: * Now try to parse the name.
888: * Use the filename portion of the path.
889: */
890: mlink->name = start;
891: if (NULL != (p = strrchr(start, '/'))) {
892: mlink->name = p + 1;
893: *p = '\0';
894: }
895: mlink_add(mlink, &st);
896: }
1.2 schwarze 897:
1.47 schwarze 898: static void
899: mlink_add(struct mlink *mlink, const struct stat *st)
900: {
901: struct inodev inodev;
902: struct mpage *mpage;
903: unsigned int slot;
904:
905: assert(NULL != mlink->file);
906:
1.51 schwarze 907: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
908: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
909: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
910: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 911:
912: if ('0' == *mlink->fsec) {
913: free(mlink->fsec);
914: mlink->fsec = mandoc_strdup(mlink->dsec);
915: mlink->fform = FORM_CAT;
916: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
917: mlink->fform = FORM_SRC;
918: else
919: mlink->fform = FORM_NONE;
1.1 schwarze 920:
1.47 schwarze 921: slot = ohash_qlookup(&mlinks, mlink->file);
922: assert(NULL == ohash_find(&mlinks, slot));
923: ohash_insert(&mlinks, slot, mlink);
924:
1.139 schwarze 925: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 926: inodev.st_ino = st->st_ino;
927: inodev.st_dev = st->st_dev;
928: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
929: sizeof(struct inodev), inodev.st_ino);
930: mpage = ohash_find(&mpages, slot);
931: if (NULL == mpage) {
932: mpage = mandoc_calloc(1, sizeof(struct mpage));
933: mpage->inodev.st_ino = inodev.st_ino;
934: mpage->inodev.st_dev = inodev.st_dev;
1.173 schwarze 935: mpage->form = FORM_NONE;
1.180 schwarze 936: mpage->next = mpage_head;
937: mpage_head = mpage;
1.47 schwarze 938: ohash_insert(&mpages, slot, mpage);
939: } else
940: mlink->next = mpage->mlinks;
941: mpage->mlinks = mlink;
1.75 schwarze 942: mlink->mpage = mpage;
1.47 schwarze 943: }
1.1 schwarze 944:
1.47 schwarze 945: static void
946: mlink_free(struct mlink *mlink)
947: {
1.1 schwarze 948:
1.47 schwarze 949: free(mlink->dsec);
950: free(mlink->arch);
951: free(mlink->name);
952: free(mlink->fsec);
953: free(mlink);
954: }
1.1 schwarze 955:
1.47 schwarze 956: static void
957: mpages_free(void)
958: {
959: struct mpage *mpage;
960: struct mlink *mlink;
961:
1.180 schwarze 962: while ((mpage = mpage_head) != NULL) {
963: while ((mlink = mpage->mlinks) != NULL) {
1.47 schwarze 964: mpage->mlinks = mlink->next;
965: mlink_free(mlink);
966: }
1.180 schwarze 967: mpage_head = mpage->next;
1.47 schwarze 968: free(mpage->sec);
969: free(mpage->arch);
970: free(mpage->title);
971: free(mpage->desc);
972: free(mpage);
973: }
974: }
1.1 schwarze 975:
1.47 schwarze 976: /*
977: * For each mlink to the mpage, check whether the path looks like
978: * it is formatted, and if it does, check whether a source manual
979: * exists by the same name, ignoring the suffix.
980: * If both conditions hold, drop the mlink.
981: */
982: static void
983: mlinks_undupe(struct mpage *mpage)
984: {
985: char buf[PATH_MAX];
986: struct mlink **prev;
987: struct mlink *mlink;
988: char *bufp;
989:
990: mpage->form = FORM_CAT;
991: prev = &mpage->mlinks;
992: while (NULL != (mlink = *prev)) {
993: if (FORM_CAT != mlink->dform) {
994: mpage->form = FORM_NONE;
995: goto nextlink;
996: }
1.100 schwarze 997: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 998: bufp = strstr(buf, "cat");
999: assert(NULL != bufp);
1000: memcpy(bufp, "man", 3);
1001: if (NULL != (bufp = strrchr(buf, '.')))
1002: *++bufp = '\0';
1.100 schwarze 1003: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1004: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1005: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1006: goto nextlink;
1007: if (warnings)
1008: say(mlink->file, "Man source exists: %s", buf);
1009: if (use_all)
1010: goto nextlink;
1011: *prev = mlink->next;
1012: mlink_free(mlink);
1013: continue;
1014: nextlink:
1015: prev = &(*prev)->next;
1.1 schwarze 1016: }
1.47 schwarze 1017: }
1.1 schwarze 1018:
1.88 schwarze 1019: static void
1.50 schwarze 1020: mlink_check(struct mpage *mpage, struct mlink *mlink)
1021: {
1.88 schwarze 1022: struct str *str;
1023: unsigned int slot;
1.50 schwarze 1024:
1025: /*
1026: * Check whether the manual section given in a file
1027: * agrees with the directory where the file is located.
1028: * Some manuals have suffixes like (3p) on their
1029: * section number either inside the file or in the
1030: * directory name, some are linked into more than one
1031: * section, like encrypt(1) = makekey(8).
1032: */
1033:
1034: if (FORM_SRC == mpage->form &&
1.88 schwarze 1035: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1036: say(mlink->file, "Section \"%s\" manual in %s directory",
1037: mpage->sec, mlink->dsec);
1038:
1039: /*
1040: * Manual page directories exist for each kernel
1041: * architecture as returned by machine(1).
1042: * However, many manuals only depend on the
1043: * application architecture as returned by arch(1).
1044: * For example, some (2/ARM) manuals are shared
1045: * across the "armish" and "zaurus" kernel
1046: * architectures.
1047: * A few manuals are even shared across completely
1048: * different architectures, for example fdformat(1)
1.178 schwarze 1049: * on amd64, i386, and sparc64.
1.50 schwarze 1050: */
1051:
1.88 schwarze 1052: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1053: say(mlink->file, "Architecture \"%s\" manual in "
1054: "\"%s\" directory", mpage->arch, mlink->arch);
1055:
1.88 schwarze 1056: /*
1057: * XXX
1.90 schwarze 1058: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1059: */
1060:
1061: if (FORM_CAT == mpage->form)
1062: return;
1063:
1064: /*
1065: * Check whether this mlink
1066: * appears as a name in the NAME section.
1067: */
1.50 schwarze 1068:
1.90 schwarze 1069: slot = ohash_qlookup(&names, mlink->name);
1070: str = ohash_find(&names, slot);
1.88 schwarze 1071: assert(NULL != str);
1.90 schwarze 1072: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1073: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1074: }
1075:
1.47 schwarze 1076: /*
1077: * Run through the files in the global vector "mpages"
1078: * and add them to the database specified in "basedir".
1079: *
1080: * This handles the parsing scheme itself, using the cues of directory
1081: * and filename to determine whether the file is parsable or not.
1082: */
1083: static void
1.173 schwarze 1084: mpages_merge(struct dba *dba, struct mparse *mp)
1.47 schwarze 1085: {
1.180 schwarze 1086: struct mpage *mpage, *mpage_dest;
1.75 schwarze 1087: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1088: struct roff_man *man;
1.75 schwarze 1089: char *sodest;
1.69 schwarze 1090: char *cp;
1.117 schwarze 1091: int fd;
1.64 schwarze 1092:
1.180 schwarze 1093: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.47 schwarze 1094: mlinks_undupe(mpage);
1.180 schwarze 1095: if ((mlink = mpage->mlinks) == NULL)
1096: continue;
1.1 schwarze 1097:
1.90 schwarze 1098: name_mask = NAME_MASK;
1.156 schwarze 1099: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1100: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.47 schwarze 1101: mparse_reset(mp);
1102: man = NULL;
1.81 schwarze 1103: sodest = NULL;
1104:
1.166 schwarze 1105: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.136 schwarze 1106: say(mlink->file, "&open");
1.117 schwarze 1107: goto nextpage;
1.81 schwarze 1108: }
1.11 schwarze 1109:
1110: /*
1.137 schwarze 1111: * Interpret the file as mdoc(7) or man(7) source
1112: * code, unless it is known to be formatted.
1.11 schwarze 1113: */
1.136 schwarze 1114: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1115: mparse_readfd(mp, fd, mlink->file);
1.165 schwarze 1116: close(fd);
1.189 schwarze 1117: fd = -1;
1.145 schwarze 1118: mparse_result(mp, &man, &sodest);
1.47 schwarze 1119: }
1.11 schwarze 1120:
1.126 schwarze 1121: if (sodest != NULL) {
1.75 schwarze 1122: mlink_dest = ohash_find(&mlinks,
1123: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1124: if (mlink_dest == NULL) {
1125: mandoc_asprintf(&cp, "%s.gz", sodest);
1126: mlink_dest = ohash_find(&mlinks,
1127: ohash_qlookup(&mlinks, cp));
1128: free(cp);
1129: }
1130: if (mlink_dest != NULL) {
1.75 schwarze 1131:
1132: /* The .so target exists. */
1133:
1134: mpage_dest = mlink_dest->mpage;
1135: while (1) {
1136: mlink->mpage = mpage_dest;
1137:
1138: /*
1139: * If the target was already
1140: * processed, add the links
1141: * to the database now.
1142: * Otherwise, this will
1143: * happen when we come
1144: * to the target.
1145: */
1146:
1.173 schwarze 1147: if (mpage_dest->dba != NULL)
1148: dbadd_mlink(mlink);
1.75 schwarze 1149:
1.126 schwarze 1150: if (mlink->next == NULL)
1.75 schwarze 1151: break;
1152: mlink = mlink->next;
1153: }
1154:
1155: /* Move all links to the target. */
1156:
1157: mlink->next = mlink_dest->next;
1158: mlink_dest->next = mpage->mlinks;
1159: mpage->mlinks = NULL;
1160: }
1.81 schwarze 1161: goto nextpage;
1.145 schwarze 1162: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.159 schwarze 1163: mdoc_validate(man);
1.47 schwarze 1164: mpage->form = FORM_SRC;
1.146 schwarze 1165: mpage->sec = man->meta.msec;
1.112 schwarze 1166: mpage->sec = mandoc_strdup(
1.126 schwarze 1167: mpage->sec == NULL ? "" : mpage->sec);
1.146 schwarze 1168: mpage->arch = man->meta.arch;
1.47 schwarze 1169: mpage->arch = mandoc_strdup(
1.126 schwarze 1170: mpage->arch == NULL ? "" : mpage->arch);
1.146 schwarze 1171: mpage->title = mandoc_strdup(man->meta.title);
1.145 schwarze 1172: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.160 schwarze 1173: man_validate(man);
1.189 schwarze 1174: if (*man->meta.msec != '\0' ||
1.194 schwarze 1175: *man->meta.title != '\0') {
1.189 schwarze 1176: mpage->form = FORM_SRC;
1177: mpage->sec = mandoc_strdup(man->meta.msec);
1178: mpage->arch = mandoc_strdup(mlink->arch);
1179: mpage->title = mandoc_strdup(man->meta.title);
1180: } else
1181: man = NULL;
1182: }
1183:
1184: assert(mpage->desc == NULL);
1185: if (man == NULL) {
1.47 schwarze 1186: mpage->form = FORM_CAT;
1.136 schwarze 1187: mpage->sec = mandoc_strdup(mlink->dsec);
1188: mpage->arch = mandoc_strdup(mlink->arch);
1189: mpage->title = mandoc_strdup(mlink->name);
1.189 schwarze 1190: parse_cat(mpage, fd);
1191: } else if (man->macroset == MACROSET_MDOC)
1.146 schwarze 1192: parse_mdoc(mpage, &man->meta, man->first);
1.189 schwarze 1193: else
1.146 schwarze 1194: parse_man(mpage, &man->meta, man->first);
1.191 schwarze 1195: if (mpage->desc == NULL) {
1196: mpage->desc = mandoc_strdup(mlink->name);
1197: if (warnings)
1198: say(mlink->file, "No one-line description, "
1199: "using filename \"%s\"", mlink->name);
1200: }
1.88 schwarze 1201:
1.187 schwarze 1202: for (mlink = mpage->mlinks;
1203: mlink != NULL;
1204: mlink = mlink->next) {
1205: putkey(mpage, mlink->name, NAME_FILE);
1206: if (warnings && !use_all)
1.88 schwarze 1207: mlink_check(mpage, mlink);
1.187 schwarze 1208: }
1.6 schwarze 1209:
1.173 schwarze 1210: dbadd(dba, mpage);
1.81 schwarze 1211:
1212: nextpage:
1.47 schwarze 1213: ohash_delete(&strings);
1.90 schwarze 1214: ohash_delete(&names);
1.47 schwarze 1215: }
1216: }
1.6 schwarze 1217:
1.47 schwarze 1218: static void
1.81 schwarze 1219: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1220: {
1221: FILE *stream;
1.190 schwarze 1222: struct mlink *mlink;
1223: char *line, *p, *title, *sec;
1.162 schwarze 1224: size_t linesz, plen, titlesz;
1225: ssize_t len;
1226: int offs;
1.1 schwarze 1227:
1.190 schwarze 1228: mlink = mpage->mlinks;
1229: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1230: if (stream == NULL) {
1231: if (fd != -1)
1.114 doug 1232: close(fd);
1.47 schwarze 1233: if (warnings)
1.190 schwarze 1234: say(mlink->file, "&fopen");
1.47 schwarze 1235: return;
1236: }
1.1 schwarze 1237:
1.162 schwarze 1238: line = NULL;
1239: linesz = 0;
1240:
1.190 schwarze 1241: /* Parse the section number from the header line. */
1242:
1243: while (getline(&line, &linesz, stream) != -1) {
1244: if (*line == '\n')
1245: continue;
1246: if ((sec = strchr(line, '(')) == NULL)
1247: break;
1248: if ((p = strchr(++sec, ')')) == NULL)
1249: break;
1250: free(mpage->sec);
1251: mpage->sec = mandoc_strndup(sec, p - sec);
1252: if (warnings && *mlink->dsec != '\0' &&
1253: strcasecmp(mpage->sec, mlink->dsec))
1254: say(mlink->file,
1255: "Section \"%s\" manual in %s directory",
1256: mpage->sec, mlink->dsec);
1257: break;
1258: }
1259:
1.47 schwarze 1260: /* Skip to first blank line. */
1.1 schwarze 1261:
1.190 schwarze 1262: while (line == NULL || *line != '\n')
1263: if (getline(&line, &linesz, stream) == -1)
1.47 schwarze 1264: break;
1.1 schwarze 1265:
1.47 schwarze 1266: /*
1267: * Assume the first line that is not indented
1268: * is the first section header. Skip to it.
1269: */
1.1 schwarze 1270:
1.162 schwarze 1271: while (getline(&line, &linesz, stream) != -1)
1272: if (*line != '\n' && *line != ' ')
1.47 schwarze 1273: break;
1.98 schwarze 1274:
1.47 schwarze 1275: /*
1276: * Read up until the next section into a buffer.
1277: * Strip the leading and trailing newline from each read line,
1278: * appending a trailing space.
1279: * Ignore empty (whitespace-only) lines.
1280: */
1.28 schwarze 1281:
1.47 schwarze 1282: titlesz = 0;
1283: title = NULL;
1.38 schwarze 1284:
1.162 schwarze 1285: while ((len = getline(&line, &linesz, stream)) != -1) {
1286: if (*line != ' ')
1.47 schwarze 1287: break;
1.162 schwarze 1288: offs = 0;
1289: while (isspace((unsigned char)line[offs]))
1290: offs++;
1291: if (line[offs] == '\0')
1.47 schwarze 1292: continue;
1.162 schwarze 1293: title = mandoc_realloc(title, titlesz + len - offs);
1294: memcpy(title + titlesz, line + offs, len - offs);
1295: titlesz += len - offs;
1.47 schwarze 1296: title[titlesz - 1] = ' ';
1297: }
1.162 schwarze 1298: free(line);
1.28 schwarze 1299:
1.47 schwarze 1300: /*
1301: * If no page content can be found, or the input line
1302: * is already the next section header, or there is no
1303: * trailing newline, reuse the page title as the page
1304: * description.
1305: */
1.1 schwarze 1306:
1.47 schwarze 1307: if (NULL == title || '\0' == *title) {
1308: if (warnings)
1.190 schwarze 1309: say(mlink->file, "Cannot find NAME section");
1.47 schwarze 1310: fclose(stream);
1311: free(title);
1312: return;
1313: }
1.24 schwarze 1314:
1.162 schwarze 1315: title[titlesz - 1] = '\0';
1.24 schwarze 1316:
1.47 schwarze 1317: /*
1318: * Skip to the first dash.
1319: * Use the remaining line as the description (no more than 70
1320: * bytes).
1321: */
1.28 schwarze 1322:
1.47 schwarze 1323: if (NULL != (p = strstr(title, "- "))) {
1324: for (p += 2; ' ' == *p || '\b' == *p; p++)
1325: /* Skip to next word. */ ;
1326: } else {
1327: if (warnings)
1.191 schwarze 1328: say(mlink->file, "No dash in title line, "
1329: "reusing \"%s\" as one-line description", title);
1.47 schwarze 1330: p = title;
1331: }
1.1 schwarze 1332:
1.47 schwarze 1333: plen = strlen(p);
1.1 schwarze 1334:
1.47 schwarze 1335: /* Strip backspace-encoding from line. */
1.1 schwarze 1336:
1.47 schwarze 1337: while (NULL != (line = memchr(p, '\b', plen))) {
1338: len = line - p;
1339: if (0 == len) {
1340: memmove(line, line + 1, plen--);
1341: continue;
1.98 schwarze 1342: }
1.47 schwarze 1343: memmove(line - 1, line + 1, plen - len);
1344: plen -= 2;
1345: }
1.1 schwarze 1346:
1.47 schwarze 1347: mpage->desc = mandoc_strdup(p);
1348: fclose(stream);
1349: free(title);
1350: }
1.16 schwarze 1351:
1.47 schwarze 1352: /*
1353: * Put a type/word pair into the word database for this particular file.
1354: */
1355: static void
1.69 schwarze 1356: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1357: {
1358: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1359: }
1360:
1361: /*
1.47 schwarze 1362: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1363: */
1364: static void
1.47 schwarze 1365: putmdockey(const struct mpage *mpage,
1.184 schwarze 1366: const struct roff_node *n, uint64_t m, int taboo)
1.2 schwarze 1367: {
1.16 schwarze 1368:
1.47 schwarze 1369: for ( ; NULL != n; n = n->next) {
1.184 schwarze 1370: if (n->flags & taboo)
1371: continue;
1.47 schwarze 1372: if (NULL != n->child)
1.184 schwarze 1373: putmdockey(mpage, n->child, m, taboo);
1.141 schwarze 1374: if (n->type == ROFFT_TEXT)
1.47 schwarze 1375: putkey(mpage, n->string, m);
1376: }
1377: }
1.16 schwarze 1378:
1.47 schwarze 1379: static void
1.143 schwarze 1380: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1381: const struct roff_node *n)
1.47 schwarze 1382: {
1.142 schwarze 1383: const struct roff_node *head, *body;
1.78 schwarze 1384: char *start, *title;
1.47 schwarze 1385: char byte;
1.78 schwarze 1386: size_t sz;
1.16 schwarze 1387:
1.168 schwarze 1388: if (n == NULL)
1.47 schwarze 1389: return;
1.16 schwarze 1390:
1.47 schwarze 1391: /*
1392: * We're only searching for one thing: the first text child in
1393: * the BODY of a NAME section. Since we don't keep track of
1394: * sections in -man, run some hoops to find out whether we're in
1395: * the correct section or not.
1396: */
1.16 schwarze 1397:
1.141 schwarze 1398: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1399: body = n;
1.168 schwarze 1400: if ((head = body->parent->head) != NULL &&
1401: (head = head->child) != NULL &&
1402: head->next == NULL &&
1.141 schwarze 1403: head->type == ROFFT_TEXT &&
1.168 schwarze 1404: strcmp(head->string, "NAME") == 0 &&
1405: body->child != NULL) {
1.2 schwarze 1406:
1.47 schwarze 1407: /*
1408: * Suck the entire NAME section into memory.
1409: * Yes, we might run away.
1410: * But too many manuals have big, spread-out
1411: * NAME sections over many lines.
1412: */
1.2 schwarze 1413:
1.78 schwarze 1414: title = NULL;
1.147 schwarze 1415: deroff(&title, body);
1.47 schwarze 1416: if (NULL == title)
1417: return;
1.16 schwarze 1418:
1.98 schwarze 1419: /*
1.47 schwarze 1420: * Go through a special heuristic dance here.
1421: * Conventionally, one or more manual names are
1422: * comma-specified prior to a whitespace, then a
1423: * dash, then a description. Try to puzzle out
1424: * the name parts here.
1425: */
1.16 schwarze 1426:
1.78 schwarze 1427: start = title;
1.47 schwarze 1428: for ( ;; ) {
1429: sz = strcspn(start, " ,");
1430: if ('\0' == start[sz])
1431: break;
1.1 schwarze 1432:
1.47 schwarze 1433: byte = start[sz];
1434: start[sz] = '\0';
1.67 schwarze 1435:
1436: /*
1437: * Assume a stray trailing comma in the
1438: * name list if a name begins with a dash.
1439: */
1440:
1441: if ('-' == start[0] ||
1442: ('\\' == start[0] && '-' == start[1]))
1443: break;
1.1 schwarze 1444:
1.90 schwarze 1445: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1446: if ( ! (mpage->name_head_done ||
1447: strcasecmp(start, meta->title))) {
1448: putkey(mpage, start, NAME_HEAD);
1449: mpage->name_head_done = 1;
1450: }
1.1 schwarze 1451:
1.47 schwarze 1452: if (' ' == byte) {
1453: start += sz + 1;
1454: break;
1455: }
1.1 schwarze 1456:
1.47 schwarze 1457: assert(',' == byte);
1458: start += sz + 1;
1459: while (' ' == *start)
1460: start++;
1461: }
1.1 schwarze 1462:
1.78 schwarze 1463: if (start == title) {
1.90 schwarze 1464: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1465: if ( ! (mpage->name_head_done ||
1466: strcasecmp(start, meta->title))) {
1467: putkey(mpage, start, NAME_HEAD);
1468: mpage->name_head_done = 1;
1469: }
1.47 schwarze 1470: free(title);
1471: return;
1472: }
1.1 schwarze 1473:
1.47 schwarze 1474: while (isspace((unsigned char)*start))
1475: start++;
1.1 schwarze 1476:
1.47 schwarze 1477: if (0 == strncmp(start, "-", 1))
1478: start += 1;
1479: else if (0 == strncmp(start, "\\-\\-", 4))
1480: start += 4;
1481: else if (0 == strncmp(start, "\\-", 2))
1482: start += 2;
1483: else if (0 == strncmp(start, "\\(en", 4))
1484: start += 4;
1485: else if (0 == strncmp(start, "\\(em", 4))
1486: start += 4;
1.1 schwarze 1487:
1.47 schwarze 1488: while (' ' == *start)
1489: start++;
1.1 schwarze 1490:
1.47 schwarze 1491: mpage->desc = mandoc_strdup(start);
1492: free(title);
1493: return;
1494: }
1495: }
1.1 schwarze 1496:
1.47 schwarze 1497: for (n = n->child; n; n = n->next) {
1498: if (NULL != mpage->desc)
1499: break;
1.129 schwarze 1500: parse_man(mpage, meta, n);
1.1 schwarze 1501: }
1502: }
1503:
1504: static void
1.143 schwarze 1505: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1506: const struct roff_node *n)
1.1 schwarze 1507: {
1508:
1.195 ! schwarze 1509: for (n = n->child; n != NULL; n = n->next) {
! 1510: if (n->tok == TOKEN_NONE || n->flags & mdocs[n->tok].taboo)
1.184 schwarze 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:
1.195 ! schwarze 1518: if (mdocs[n->tok].fp != NULL &&
! 1519: (*mdocs[n->tok].fp)(mpage, meta, n) == 0)
! 1520: break;
1.47 schwarze 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:
1526: continue;
1527: }
1528: if (NULL != n->child)
1.127 schwarze 1529: parse_mdoc(mpage, meta, n);
1.1 schwarze 1530: }
1531: }
1532:
1.19 schwarze 1533: static int
1.143 schwarze 1534: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1535: const struct roff_node *n)
1.1 schwarze 1536: {
1.131 schwarze 1537: char *start, *end;
1.1 schwarze 1538: size_t sz;
1.19 schwarze 1539:
1.47 schwarze 1540: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1541: NULL == (n = n->child) ||
1.141 schwarze 1542: n->type != ROFFT_TEXT)
1.150 schwarze 1543: return 0;
1.1 schwarze 1544:
1545: /*
1546: * Only consider those `Fd' macro fields that begin with an
1547: * "inclusion" token (versus, e.g., #define).
1548: */
1.47 schwarze 1549:
1.1 schwarze 1550: if (strcmp("#include", n->string))
1.150 schwarze 1551: return 0;
1.1 schwarze 1552:
1.141 schwarze 1553: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.150 schwarze 1554: return 0;
1.1 schwarze 1555:
1556: /*
1557: * Strip away the enclosing angle brackets and make sure we're
1558: * not zero-length.
1559: */
1560:
1561: start = n->string;
1562: if ('<' == *start || '"' == *start)
1563: start++;
1564:
1565: if (0 == (sz = strlen(start)))
1.150 schwarze 1566: return 0;
1.1 schwarze 1567:
1568: end = &start[(int)sz - 1];
1569: if ('>' == *end || '"' == *end)
1570: end--;
1571:
1.47 schwarze 1572: if (end > start)
1573: putkeys(mpage, start, end - start + 1, TYPE_In);
1.150 schwarze 1574: return 0;
1.1 schwarze 1575: }
1576:
1.133 schwarze 1577: static void
1.142 schwarze 1578: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 schwarze 1579: {
1.69 schwarze 1580: char *cp;
1.133 schwarze 1581: size_t sz;
1.1 schwarze 1582:
1.141 schwarze 1583: if (n->type != ROFFT_TEXT)
1.133 schwarze 1584: return;
1.19 schwarze 1585:
1.133 schwarze 1586: /* Skip function pointer punctuation. */
1.1 schwarze 1587:
1.133 schwarze 1588: cp = n->string;
1589: while (*cp == '(' || *cp == '*')
1.1 schwarze 1590: cp++;
1.133 schwarze 1591: sz = strcspn(cp, "()");
1.1 schwarze 1592:
1.133 schwarze 1593: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1594: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1595: putkeys(mpage, cp, sz, NAME_SYN);
1596: }
1.19 schwarze 1597:
1.133 schwarze 1598: static int
1.143 schwarze 1599: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1600: const struct roff_node *n)
1.133 schwarze 1601: {
1602:
1603: if (n->child == NULL)
1.150 schwarze 1604: return 0;
1.133 schwarze 1605:
1606: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1607:
1.133 schwarze 1608: for (n = n->child->next; n != NULL; n = n->next)
1.141 schwarze 1609: if (n->type == ROFFT_TEXT)
1.47 schwarze 1610: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1611:
1.150 schwarze 1612: return 0;
1.128 schwarze 1613: }
1614:
1615: static int
1.143 schwarze 1616: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1617: const struct roff_node *n)
1.128 schwarze 1618: {
1.132 schwarze 1619:
1.141 schwarze 1620: if (n->type != ROFFT_HEAD)
1.150 schwarze 1621: return 1;
1.128 schwarze 1622:
1.133 schwarze 1623: if (n->child != NULL)
1624: parse_mdoc_fname(mpage, n->child);
1625:
1.150 schwarze 1626: return 0;
1.1 schwarze 1627: }
1628:
1.19 schwarze 1629: static int
1.164 schwarze 1630: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1631: const struct roff_node *n)
1632: {
1633: char *cp;
1634:
1635: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1636: return 0;
1637:
1.168 schwarze 1638: if (n->child != NULL &&
1639: n->child->next == NULL &&
1640: n->child->type == ROFFT_TEXT)
1.164 schwarze 1641: return 1;
1642:
1643: cp = NULL;
1644: deroff(&cp, n);
1645: if (cp != NULL) {
1646: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1647: n->type == ROFFT_BODY ? TYPE_Va : 0));
1648: free(cp);
1649: }
1650:
1651: return 0;
1652: }
1653:
1654: static int
1.143 schwarze 1655: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1656: const struct roff_node *n)
1.1 schwarze 1657: {
1.47 schwarze 1658: char *cp;
1.1 schwarze 1659:
1660: if (NULL == (n = n->child))
1.150 schwarze 1661: return 0;
1.1 schwarze 1662:
1.47 schwarze 1663: if (NULL == n->next) {
1664: putkey(mpage, n->string, TYPE_Xr);
1.150 schwarze 1665: return 0;
1.47 schwarze 1666: }
1.1 schwarze 1667:
1.77 schwarze 1668: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1669: putkey(mpage, cp, TYPE_Xr);
1670: free(cp);
1.150 schwarze 1671: return 0;
1.1 schwarze 1672: }
1673:
1.19 schwarze 1674: static int
1.143 schwarze 1675: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1676: const struct roff_node *n)
1.1 schwarze 1677: {
1678:
1.141 schwarze 1679: if (n->type == ROFFT_BODY)
1.147 schwarze 1680: deroff(&mpage->desc, n);
1.150 schwarze 1681: return 0;
1.1 schwarze 1682: }
1683:
1.19 schwarze 1684: static int
1.143 schwarze 1685: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1686: const struct roff_node *n)
1.1 schwarze 1687: {
1688:
1.86 schwarze 1689: if (SEC_NAME == n->sec)
1.184 schwarze 1690: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.141 schwarze 1691: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.127 schwarze 1692: if (n->child == NULL)
1693: putkey(mpage, meta->name, NAME_SYN);
1694: else
1.184 schwarze 1695: putmdockey(mpage, n->child, NAME_SYN, 0);
1.129 schwarze 1696: }
1697: if ( ! (mpage->name_head_done ||
1698: n->child == NULL || n->child->string == NULL ||
1699: strcasecmp(n->child->string, meta->title))) {
1.175 schwarze 1700: putkey(mpage, n->child->string, NAME_HEAD);
1.129 schwarze 1701: mpage->name_head_done = 1;
1.127 schwarze 1702: }
1.150 schwarze 1703: return 0;
1.1 schwarze 1704: }
1705:
1.19 schwarze 1706: static int
1.143 schwarze 1707: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1708: const struct roff_node *n)
1.1 schwarze 1709: {
1710:
1.150 schwarze 1711: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 schwarze 1712: }
1713:
1.47 schwarze 1714: static int
1.143 schwarze 1715: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1716: const struct roff_node *n)
1.1 schwarze 1717: {
1718:
1.150 schwarze 1719: return n->type == ROFFT_HEAD;
1.1 schwarze 1720: }
1721:
1.47 schwarze 1722: /*
1723: * Add a string to the hash table for the current manual.
1724: * Each string has a bitmask telling which macros it belongs to.
1725: * When we finish the manual, we'll dump the table.
1726: */
1.1 schwarze 1727: static void
1.131 schwarze 1728: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1729: {
1.90 schwarze 1730: struct ohash *htab;
1.47 schwarze 1731: struct str *s;
1.68 schwarze 1732: const char *end;
1.47 schwarze 1733: unsigned int slot;
1.131 schwarze 1734: int i, mustfree;
1.1 schwarze 1735:
1.47 schwarze 1736: if (0 == sz)
1737: return;
1.68 schwarze 1738:
1.131 schwarze 1739: mustfree = render_string(&cp, &sz);
1740:
1.90 schwarze 1741: if (TYPE_Nm & v) {
1742: htab = &names;
1743: v &= name_mask;
1.124 schwarze 1744: if (v & NAME_FIRST)
1745: name_mask &= ~NAME_FIRST;
1.90 schwarze 1746: if (debug > 1)
1747: say(mpage->mlinks->file,
1.183 schwarze 1748: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1749: (unsigned long long)v);
1.90 schwarze 1750: } else {
1751: htab = &strings;
1752: if (debug > 1)
1.173 schwarze 1753: for (i = 0; i < KEY_MAX; i++)
1.118 schwarze 1754: if ((uint64_t)1 << i & v)
1.90 schwarze 1755: say(mpage->mlinks->file,
1756: "Adding key %s=%*s",
1.172 schwarze 1757: mansearch_keynames[i], (int)sz, cp);
1.68 schwarze 1758: }
1.47 schwarze 1759:
1760: end = cp + sz;
1.90 schwarze 1761: slot = ohash_qlookupi(htab, cp, &end);
1762: s = ohash_find(htab, slot);
1.1 schwarze 1763:
1.47 schwarze 1764: if (NULL != s && mpage == s->mpage) {
1765: s->mask |= v;
1.1 schwarze 1766: return;
1.47 schwarze 1767: } else if (NULL == s) {
1.101 schwarze 1768: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1769: memcpy(s->key, cp, sz);
1.90 schwarze 1770: ohash_insert(htab, slot, s);
1.47 schwarze 1771: }
1772: s->mpage = mpage;
1773: s->mask = v;
1.131 schwarze 1774:
1775: if (mustfree)
1776: free(cp);
1.1 schwarze 1777: }
1778:
1779: /*
1.47 schwarze 1780: * Take a Unicode codepoint and produce its UTF-8 encoding.
1781: * This isn't the best way to do this, but it works.
1782: * The magic numbers are from the UTF-8 packaging.
1783: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1784: */
1.47 schwarze 1785: static size_t
1786: utf8(unsigned int cp, char out[7])
1.1 schwarze 1787: {
1.47 schwarze 1788: size_t rc;
1.1 schwarze 1789:
1.47 schwarze 1790: rc = 0;
1791: if (cp <= 0x0000007F) {
1792: rc = 1;
1793: out[0] = (char)cp;
1794: } else if (cp <= 0x000007FF) {
1795: rc = 2;
1796: out[0] = (cp >> 6 & 31) | 192;
1797: out[1] = (cp & 63) | 128;
1798: } else if (cp <= 0x0000FFFF) {
1799: rc = 3;
1800: out[0] = (cp >> 12 & 15) | 224;
1801: out[1] = (cp >> 6 & 63) | 128;
1802: out[2] = (cp & 63) | 128;
1803: } else if (cp <= 0x001FFFFF) {
1804: rc = 4;
1805: out[0] = (cp >> 18 & 7) | 240;
1806: out[1] = (cp >> 12 & 63) | 128;
1807: out[2] = (cp >> 6 & 63) | 128;
1808: out[3] = (cp & 63) | 128;
1809: } else if (cp <= 0x03FFFFFF) {
1810: rc = 5;
1811: out[0] = (cp >> 24 & 3) | 248;
1812: out[1] = (cp >> 18 & 63) | 128;
1813: out[2] = (cp >> 12 & 63) | 128;
1814: out[3] = (cp >> 6 & 63) | 128;
1815: out[4] = (cp & 63) | 128;
1816: } else if (cp <= 0x7FFFFFFF) {
1817: rc = 6;
1818: out[0] = (cp >> 30 & 1) | 252;
1819: out[1] = (cp >> 24 & 63) | 128;
1820: out[2] = (cp >> 18 & 63) | 128;
1821: out[3] = (cp >> 12 & 63) | 128;
1822: out[4] = (cp >> 6 & 63) | 128;
1823: out[5] = (cp & 63) | 128;
1824: } else
1.150 schwarze 1825: return 0;
1.19 schwarze 1826:
1.47 schwarze 1827: out[rc] = '\0';
1.150 schwarze 1828: return rc;
1.1 schwarze 1829: }
1830:
1.47 schwarze 1831: /*
1.131 schwarze 1832: * If the string contains escape sequences,
1833: * replace it with an allocated rendering and return 1,
1834: * such that the caller can free it after use.
1835: * Otherwise, do nothing and return 0.
1.47 schwarze 1836: */
1.131 schwarze 1837: static int
1838: render_string(char **public, size_t *psz)
1.1 schwarze 1839: {
1.131 schwarze 1840: const char *src, *scp, *addcp, *seq;
1841: char *dst;
1842: size_t ssz, dsz, addsz;
1.71 schwarze 1843: char utfbuf[7], res[6];
1.131 schwarze 1844: int seqlen, unicode;
1.47 schwarze 1845:
1846: res[0] = '\\';
1847: res[1] = '\t';
1848: res[2] = ASCII_NBRSP;
1849: res[3] = ASCII_HYPH;
1.71 schwarze 1850: res[4] = ASCII_BREAK;
1851: res[5] = '\0';
1.1 schwarze 1852:
1.131 schwarze 1853: src = scp = *public;
1854: ssz = *psz;
1855: dst = NULL;
1856: dsz = 0;
1857:
1858: while (scp < src + *psz) {
1859:
1860: /* Leave normal characters unchanged. */
1861:
1862: if (strchr(res, *scp) == NULL) {
1863: if (dst != NULL)
1864: dst[dsz++] = *scp;
1865: scp++;
1866: continue;
1867: }
1.1 schwarze 1868:
1.47 schwarze 1869: /*
1.131 schwarze 1870: * Found something that requires replacing,
1871: * make sure we have a destination buffer.
1.47 schwarze 1872: */
1.131 schwarze 1873:
1874: if (dst == NULL) {
1875: dst = mandoc_malloc(ssz + 1);
1876: dsz = scp - src;
1877: memcpy(dst, src, dsz);
1.47 schwarze 1878: }
1.39 schwarze 1879:
1.131 schwarze 1880: /* Handle single-char special characters. */
1881:
1882: switch (*scp) {
1883: case '\\':
1884: break;
1.98 schwarze 1885: case '\t':
1886: case ASCII_NBRSP:
1.131 schwarze 1887: dst[dsz++] = ' ';
1888: scp++;
1889: continue;
1890: case ASCII_HYPH:
1891: dst[dsz++] = '-';
1.71 schwarze 1892: /* FALLTHROUGH */
1.98 schwarze 1893: case ASCII_BREAK:
1.131 schwarze 1894: scp++;
1.47 schwarze 1895: continue;
1.71 schwarze 1896: default:
1.131 schwarze 1897: abort();
1.71 schwarze 1898: }
1.39 schwarze 1899:
1.47 schwarze 1900: /*
1.131 schwarze 1901: * Found an escape sequence.
1902: * Read past the slash, then parse it.
1903: * Ignore everything except characters.
1.47 schwarze 1904: */
1.52 schwarze 1905:
1.131 schwarze 1906: scp++;
1907: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1908: continue;
1.39 schwarze 1909:
1.47 schwarze 1910: /*
1.52 schwarze 1911: * Render the special character
1912: * as either UTF-8 or ASCII.
1.47 schwarze 1913: */
1.52 schwarze 1914:
1915: if (write_utf8) {
1.157 schwarze 1916: unicode = mchars_spec2cp(seq, seqlen);
1.131 schwarze 1917: if (unicode <= 0)
1.52 schwarze 1918: continue;
1.131 schwarze 1919: addsz = utf8(unicode, utfbuf);
1920: if (addsz == 0)
1.52 schwarze 1921: continue;
1.131 schwarze 1922: addcp = utfbuf;
1.52 schwarze 1923: } else {
1.157 schwarze 1924: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.131 schwarze 1925: if (addcp == NULL)
1.52 schwarze 1926: continue;
1.131 schwarze 1927: if (*addcp == ASCII_NBRSP) {
1928: addcp = " ";
1929: addsz = 1;
1.52 schwarze 1930: }
1931: }
1.1 schwarze 1932:
1.47 schwarze 1933: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1934:
1.131 schwarze 1935: ssz += addsz;
1936: dst = mandoc_realloc(dst, ssz + 1);
1937: memcpy(dst + dsz, addcp, addsz);
1938: dsz += addsz;
1939: }
1940: if (dst != NULL) {
1941: *public = dst;
1942: *psz = dsz;
1943: }
1944:
1945: /* Trim trailing whitespace and NUL-terminate. */
1946:
1947: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1948: --*psz;
1949: if (dst != NULL) {
1950: (*public)[*psz] = '\0';
1.150 schwarze 1951: return 1;
1.131 schwarze 1952: } else
1.150 schwarze 1953: return 0;
1.1 schwarze 1954: }
1955:
1.75 schwarze 1956: static void
1957: dbadd_mlink(const struct mlink *mlink)
1958: {
1.173 schwarze 1959: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
1960: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1961: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1962: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
1963: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.75 schwarze 1964: }
1965:
1.11 schwarze 1966: /*
1.47 schwarze 1967: * Flush the current page's terms (and their bits) into the database.
1.53 schwarze 1968: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1969: */
1970: static void
1.173 schwarze 1971: dbadd(struct dba *dba, struct mpage *mpage)
1.11 schwarze 1972: {
1.47 schwarze 1973: struct mlink *mlink;
1974: struct str *key;
1.131 schwarze 1975: char *cp;
1.173 schwarze 1976: uint64_t mask;
1.47 schwarze 1977: size_t i;
1978: unsigned int slot;
1.131 schwarze 1979: int mustfree;
1.47 schwarze 1980:
1.85 schwarze 1981: mlink = mpage->mlinks;
1.11 schwarze 1982:
1.85 schwarze 1983: if (nodb) {
1.104 schwarze 1984: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 1985: key = ohash_next(&names, &slot))
1.104 schwarze 1986: free(key);
1987: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 1988: key = ohash_next(&strings, &slot))
1.104 schwarze 1989: free(key);
1.102 schwarze 1990: if (0 == debug)
1991: return;
1.85 schwarze 1992: while (NULL != mlink) {
1993: fputs(mlink->name, stdout);
1994: if (NULL == mlink->next ||
1995: strcmp(mlink->dsec, mlink->next->dsec) ||
1996: strcmp(mlink->fsec, mlink->next->fsec) ||
1997: strcmp(mlink->arch, mlink->next->arch)) {
1998: putchar('(');
1999: if ('\0' == *mlink->dsec)
2000: fputs(mlink->fsec, stdout);
2001: else
2002: fputs(mlink->dsec, stdout);
2003: if ('\0' != *mlink->arch)
2004: printf("/%s", mlink->arch);
2005: putchar(')');
2006: }
2007: mlink = mlink->next;
2008: if (NULL != mlink)
2009: fputs(", ", stdout);
2010: }
1.89 schwarze 2011: printf(" - %s\n", mpage->desc);
1.11 schwarze 2012: return;
1.85 schwarze 2013: }
2014:
2015: if (debug)
2016: say(mlink->file, "Adding to database");
1.47 schwarze 2017:
1.131 schwarze 2018: cp = mpage->desc;
2019: i = strlen(cp);
2020: mustfree = render_string(&cp, &i);
1.176 schwarze 2021: mpage->dba = dba_page_new(dba->pages,
1.173 schwarze 2022: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2023: cp, mlink->file, mpage->form);
1.131 schwarze 2024: if (mustfree)
2025: free(cp);
1.176 schwarze 2026: dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
1.47 schwarze 2027:
1.173 schwarze 2028: while (mlink != NULL) {
1.75 schwarze 2029: dbadd_mlink(mlink);
1.85 schwarze 2030: mlink = mlink->next;
2031: }
1.47 schwarze 2032:
1.90 schwarze 2033: for (key = ohash_first(&names, &slot); NULL != key;
2034: key = ohash_next(&names, &slot)) {
2035: assert(key->mpage == mpage);
1.173 schwarze 2036: dba_page_alias(mpage->dba, key->key, key->mask);
1.90 schwarze 2037: free(key);
2038: }
1.47 schwarze 2039: for (key = ohash_first(&strings, &slot); NULL != key;
2040: key = ohash_next(&strings, &slot)) {
2041: assert(key->mpage == mpage);
1.173 schwarze 2042: i = 0;
2043: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2044: if (key->mask & mask)
2045: dba_macro_add(dba->macros, i,
2046: key->key, mpage->dba);
2047: i++;
2048: }
1.47 schwarze 2049: free(key);
1.33 schwarze 2050: }
1.47 schwarze 2051: }
1.41 deraadt 2052:
1.47 schwarze 2053: static void
1.173 schwarze 2054: dbprune(struct dba *dba)
1.47 schwarze 2055: {
1.173 schwarze 2056: struct dba_array *page, *files;
2057: char *file;
1.11 schwarze 2058:
1.173 schwarze 2059: dba_array_FOREACH(dba->pages, page) {
2060: files = dba_array_get(page, DBP_FILE);
2061: dba_array_FOREACH(files, file) {
2062: if (*file < ' ')
2063: file++;
2064: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2065: file)) != NULL) {
2066: if (debug)
2067: say(file, "Deleting from database");
2068: dba_array_del(dba->pages);
2069: break;
2070: }
1.63 schwarze 2071: }
1.11 schwarze 2072: }
1.47 schwarze 2073: }
1.22 schwarze 2074:
1.47 schwarze 2075: /*
1.173 schwarze 2076: * Write the database from memory to disk.
1.47 schwarze 2077: */
2078: static void
1.173 schwarze 2079: dbwrite(struct dba *dba)
1.47 schwarze 2080: {
1.173 schwarze 2081: char tfn[32];
1.72 schwarze 2082: int status;
2083: pid_t child;
1.11 schwarze 2084:
1.173 schwarze 2085: if (dba_write(MANDOC_DB "~", dba) != -1) {
2086: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2087: exitcode = (int)MANDOCLEVEL_SYSERR;
2088: say(MANDOC_DB, "&rename");
2089: unlink(MANDOC_DB "~");
2090: }
1.47 schwarze 2091: return;
1.28 schwarze 2092: }
1.22 schwarze 2093:
1.173 schwarze 2094: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2095: if (mkdtemp(tfn) == NULL) {
2096: exitcode = (int)MANDOCLEVEL_SYSERR;
2097: say("", "&%s", tfn);
1.47 schwarze 2098: return;
1.173 schwarze 2099: }
1.22 schwarze 2100:
1.173 schwarze 2101: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2102: if (dba_write(tfn, dba) == -1) {
2103: exitcode = (int)MANDOCLEVEL_SYSERR;
2104: say(tfn, "&dba_write");
2105: goto out;
1.72 schwarze 2106: }
2107:
2108: switch (child = fork()) {
1.98 schwarze 2109: case -1:
1.72 schwarze 2110: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2111: say("", "&fork cmp");
1.72 schwarze 2112: return;
1.98 schwarze 2113: case 0:
1.173 schwarze 2114: execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL);
1.80 schwarze 2115: say("", "&exec cmp");
1.72 schwarze 2116: exit(0);
2117: default:
2118: break;
2119: }
1.173 schwarze 2120: if (waitpid(child, &status, 0) == -1) {
1.72 schwarze 2121: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2122: say("", "&wait cmp");
1.72 schwarze 2123: } else if (WIFSIGNALED(status)) {
2124: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2125: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2126: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2127: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2128: say(MANDOC_DB,
2129: "Data changed, but cannot replace database");
2130: }
2131:
1.173 schwarze 2132: out:
2133: *strrchr(tfn, '/') = '\0';
1.72 schwarze 2134: switch (child = fork()) {
1.98 schwarze 2135: case -1:
1.72 schwarze 2136: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2137: say("", "&fork rm");
1.72 schwarze 2138: return;
1.98 schwarze 2139: case 0:
1.173 schwarze 2140: execlp("rm", "rm", "-rf", tfn, (char *)NULL);
1.80 schwarze 2141: say("", "&exec rm");
1.72 schwarze 2142: exit((int)MANDOCLEVEL_SYSERR);
2143: default:
2144: break;
2145: }
1.173 schwarze 2146: if (waitpid(child, &status, 0) == -1) {
1.72 schwarze 2147: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2148: say("", "&wait rm");
1.72 schwarze 2149: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2150: exitcode = (int)MANDOCLEVEL_SYSERR;
1.173 schwarze 2151: say("", "%s: Cannot remove temporary directory", tfn);
1.22 schwarze 2152: }
1.47 schwarze 2153: }
1.6 schwarze 2154:
1.47 schwarze 2155: static int
1.120 schwarze 2156: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2157: {
2158: static char startdir[PATH_MAX];
1.110 schwarze 2159: static int getcwd_status; /* 1 = ok, 2 = failure */
2160: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2161: char *cp;
1.6 schwarze 2162:
1.47 schwarze 2163: /*
1.110 schwarze 2164: * Remember the original working directory, if possible.
2165: * This will be needed if the second or a later directory
2166: * on the command line is given as a relative path.
2167: * Do not error out if the current directory is not
2168: * searchable: Maybe it won't be needed after all.
2169: */
2170: if (0 == getcwd_status) {
2171: if (NULL == getcwd(startdir, sizeof(startdir))) {
2172: getcwd_status = 2;
2173: (void)strlcpy(startdir, strerror(errno),
2174: sizeof(startdir));
2175: } else
2176: getcwd_status = 1;
2177: }
2178:
2179: /*
2180: * We are leaving the old base directory.
2181: * Do not use it any longer, not even for messages.
2182: */
2183: *basedir = '\0';
2184:
2185: /*
2186: * If and only if the directory was changed earlier and
2187: * the next directory to process is given as a relative path,
2188: * first go back, or bail out if that is impossible.
1.47 schwarze 2189: */
1.110 schwarze 2190: if (chdir_status && '/' != *targetdir) {
2191: if (2 == getcwd_status) {
1.47 schwarze 2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2193: say("", "getcwd: %s", startdir);
1.150 schwarze 2194: return 0;
1.47 schwarze 2195: }
1.110 schwarze 2196: if (-1 == chdir(startdir)) {
1.47 schwarze 2197: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2198: say("", "&chdir %s", startdir);
1.150 schwarze 2199: return 0;
1.2 schwarze 2200: }
2201: }
1.110 schwarze 2202:
2203: /*
2204: * Always resolve basedir to the canonicalized absolute
2205: * pathname and append a trailing slash, such that
2206: * we can reliably check whether files are inside.
2207: */
1.47 schwarze 2208: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2209: if (report_baddir || errno != ENOENT) {
2210: exitcode = (int)MANDOCLEVEL_BADARG;
2211: say("", "&%s: realpath", targetdir);
2212: }
1.150 schwarze 2213: return 0;
1.47 schwarze 2214: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2215: if (report_baddir || errno != ENOENT) {
2216: exitcode = (int)MANDOCLEVEL_BADARG;
2217: say("", "&chdir");
2218: }
1.150 schwarze 2219: return 0;
1.109 schwarze 2220: }
1.110 schwarze 2221: chdir_status = 1;
1.109 schwarze 2222: cp = strchr(basedir, '\0');
2223: if ('/' != cp[-1]) {
2224: if (cp - basedir >= PATH_MAX - 1) {
2225: exitcode = (int)MANDOCLEVEL_SYSERR;
2226: say("", "Filename too long");
1.150 schwarze 2227: return 0;
1.109 schwarze 2228: }
2229: *cp++ = '/';
2230: *cp = '\0';
1.47 schwarze 2231: }
1.150 schwarze 2232: return 1;
1.2 schwarze 2233: }
2234:
2235: static void
1.47 schwarze 2236: say(const char *file, const char *format, ...)
1.2 schwarze 2237: {
1.47 schwarze 2238: va_list ap;
1.80 schwarze 2239: int use_errno;
1.2 schwarze 2240:
1.47 schwarze 2241: if ('\0' != *basedir)
2242: fprintf(stderr, "%s", basedir);
2243: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2244: fputc('/', stderr);
1.47 schwarze 2245: if ('\0' != *file)
2246: fprintf(stderr, "%s", file);
1.31 schwarze 2247:
1.80 schwarze 2248: use_errno = 1;
2249: if (NULL != format) {
2250: switch (*format) {
1.98 schwarze 2251: case '&':
1.80 schwarze 2252: format++;
2253: break;
1.98 schwarze 2254: case '\0':
1.80 schwarze 2255: format = NULL;
2256: break;
2257: default:
2258: use_errno = 0;
2259: break;
2260: }
2261: }
2262: if (NULL != format) {
2263: if ('\0' != *basedir || '\0' != *file)
2264: fputs(": ", stderr);
2265: va_start(ap, format);
2266: vfprintf(stderr, format, ap);
2267: va_end(ap);
2268: }
2269: if (use_errno) {
2270: if ('\0' != *basedir || '\0' != *file || NULL != format)
2271: fputs(": ", stderr);
1.47 schwarze 2272: perror(NULL);
1.80 schwarze 2273: } else
2274: fputc('\n', stderr);
1.1 schwarze 2275: }