Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.197
1.197 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.196 2017/05/04 17:48:24 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 }, /* sp */
291: { NULL, 0, 0 }, /* %U */
292: { NULL, 0, 0 }, /* Ta */
1.1 schwarze 293: };
1.195 schwarze 294: static const struct mdoc_handler *const mdocs = __mdocs - MDOC_Dd;
1.1 schwarze 295:
1.98 schwarze 296:
1.1 schwarze 297: int
1.3 schwarze 298: mandocdb(int argc, char *argv[])
1.1 schwarze 299: {
1.140 schwarze 300: struct manconf conf;
301: struct mparse *mp;
1.173 schwarze 302: struct dba *dba;
1.161 schwarze 303: const char *path_arg, *progname;
1.140 schwarze 304: size_t j, sz;
1.47 schwarze 305: int ch, i;
306:
1.158 millert 307: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.167 schwarze 308: warn("pledge");
1.155 schwarze 309: return (int)MANDOCLEVEL_SYSERR;
310: }
311:
1.140 schwarze 312: memset(&conf, 0, sizeof(conf));
1.47 schwarze 313:
314: /*
1.98 schwarze 315: * We accept a few different invocations.
1.47 schwarze 316: * The CHECKOP macro makes sure that invocation styles don't
317: * clobber each other.
318: */
319: #define CHECKOP(_op, _ch) do \
320: if (OP_DEFAULT != (_op)) { \
1.151 schwarze 321: warnx("-%c: Conflicting option", (_ch)); \
1.47 schwarze 322: goto usage; \
323: } while (/*CONSTCOND*/0)
1.10 schwarze 324:
1.47 schwarze 325: path_arg = NULL;
1.28 schwarze 326: op = OP_DEFAULT;
1.1 schwarze 327:
1.83 schwarze 328: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 329: switch (ch) {
1.98 schwarze 330: case 'a':
1.6 schwarze 331: use_all = 1;
332: break;
1.98 schwarze 333: case 'C':
1.47 schwarze 334: CHECKOP(op, ch);
335: path_arg = optarg;
1.28 schwarze 336: op = OP_CONFFILE;
1.25 schwarze 337: break;
1.98 schwarze 338: case 'D':
1.82 schwarze 339: debug++;
340: break;
1.98 schwarze 341: case 'd':
1.47 schwarze 342: CHECKOP(op, ch);
343: path_arg = optarg;
1.2 schwarze 344: op = OP_UPDATE;
1.1 schwarze 345: break;
1.98 schwarze 346: case 'n':
1.47 schwarze 347: nodb = 1;
348: break;
1.98 schwarze 349: case 'p':
1.83 schwarze 350: warnings = 1;
351: break;
1.98 schwarze 352: case 'Q':
1.73 schwarze 353: mparse_options |= MPARSE_QUICK;
1.59 schwarze 354: break;
1.98 schwarze 355: case 'T':
1.52 schwarze 356: if (strcmp(optarg, "utf8")) {
1.151 schwarze 357: warnx("-T%s: Unsupported output format",
358: optarg);
1.52 schwarze 359: goto usage;
360: }
361: write_utf8 = 1;
362: break;
1.98 schwarze 363: case 't':
1.47 schwarze 364: CHECKOP(op, ch);
1.28 schwarze 365: dup2(STDOUT_FILENO, STDERR_FILENO);
366: op = OP_TEST;
1.47 schwarze 367: nodb = warnings = 1;
1.28 schwarze 368: break;
1.98 schwarze 369: case 'u':
1.47 schwarze 370: CHECKOP(op, ch);
371: path_arg = optarg;
1.1 schwarze 372: op = OP_DELETE;
373: break;
1.98 schwarze 374: case 'v':
1.83 schwarze 375: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 376: break;
1.1 schwarze 377: default:
1.28 schwarze 378: goto usage;
1.1 schwarze 379: }
380:
381: argc -= optind;
382: argv += optind;
383:
1.163 schwarze 384: if (nodb) {
385: if (pledge("stdio rpath", NULL) == -1) {
1.167 schwarze 386: warn("pledge");
1.163 schwarze 387: return (int)MANDOCLEVEL_SYSERR;
388: }
1.155 schwarze 389: }
390:
1.28 schwarze 391: if (OP_CONFFILE == op && argc > 0) {
1.151 schwarze 392: warnx("-C: Too many arguments");
1.28 schwarze 393: goto usage;
394: }
395:
1.47 schwarze 396: exitcode = (int)MANDOCLEVEL_OK;
1.157 schwarze 397: mchars_alloc();
398: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.156 schwarze 399: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
400: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.2 schwarze 401:
1.47 schwarze 402: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 403:
1.47 schwarze 404: /*
1.109 schwarze 405: * Most of these deal with a specific directory.
1.97 schwarze 406: * Jump into that directory first.
1.47 schwarze 407: */
1.120 schwarze 408: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 409: goto out;
1.97 schwarze 410:
1.174 schwarze 411: dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
412: if (dba != NULL) {
1.97 schwarze 413: /*
414: * The existing database is usable. Process
415: * all files specified on the command-line.
416: */
1.163 schwarze 417: if (!nodb) {
418: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.167 schwarze 419: warn("pledge");
1.163 schwarze 420: exitcode = (int)MANDOCLEVEL_SYSERR;
421: goto out;
422: }
1.155 schwarze 423: }
1.97 schwarze 424: use_all = 1;
1.95 schwarze 425: for (i = 0; i < argc; i++)
426: filescan(argv[i]);
1.174 schwarze 427: if (nodb == 0)
1.173 schwarze 428: dbprune(dba);
1.95 schwarze 429: } else {
1.177 schwarze 430: /* Database missing or corrupt. */
1.179 schwarze 431: if (op != OP_UPDATE || errno != ENOENT)
432: say(MANDOC_DB, "%s: Automatically recreating"
433: " from scratch", strerror(errno));
1.97 schwarze 434: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 435: op = OP_DEFAULT;
436: if (0 == treescan())
437: goto out;
1.173 schwarze 438: dba = dba_new(128);
1.95 schwarze 439: }
1.47 schwarze 440: if (OP_DELETE != op)
1.173 schwarze 441: mpages_merge(dba, mp);
442: if (nodb == 0)
443: dbwrite(dba);
444: dba_free(dba);
1.47 schwarze 445: } else {
446: /*
447: * If we have arguments, use them as our manpaths.
1.181 schwarze 448: * If we don't, use man.conf(5).
1.47 schwarze 449: */
450: if (argc > 0) {
1.140 schwarze 451: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 452: argc, sizeof(char *));
1.140 schwarze 453: conf.manpath.sz = (size_t)argc;
1.47 schwarze 454: for (i = 0; i < argc; i++)
1.140 schwarze 455: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 456: } else
1.140 schwarze 457: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 458:
1.140 schwarze 459: if (conf.manpath.sz == 0) {
1.84 schwarze 460: exitcode = (int)MANDOCLEVEL_BADARG;
461: say("", "Empty manpath");
462: }
1.2 schwarze 463:
1.47 schwarze 464: /*
465: * First scan the tree rooted at a base directory, then
466: * build a new database and finally move it into place.
467: * Ignore zero-length directories and strip trailing
468: * slashes.
469: */
1.140 schwarze 470: for (j = 0; j < conf.manpath.sz; j++) {
471: sz = strlen(conf.manpath.paths[j]);
472: if (sz && conf.manpath.paths[j][sz - 1] == '/')
473: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 474: if (0 == sz)
475: continue;
1.2 schwarze 476:
1.47 schwarze 477: if (j) {
1.156 schwarze 478: mandoc_ohash_init(&mpages, 6,
479: offsetof(struct mpage, inodev));
480: mandoc_ohash_init(&mlinks, 6,
481: offsetof(struct mlink, file));
1.47 schwarze 482: }
1.1 schwarze 483:
1.140 schwarze 484: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 485: continue;
1.47 schwarze 486: if (0 == treescan())
1.120 schwarze 487: continue;
1.173 schwarze 488: dba = dba_new(128);
489: mpages_merge(dba, mp);
490: if (nodb == 0)
491: dbwrite(dba);
492: dba_free(dba);
1.28 schwarze 493:
1.140 schwarze 494: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 495: mpages_free();
496: ohash_delete(&mpages);
497: ohash_delete(&mlinks);
498: }
1.2 schwarze 499: }
1.47 schwarze 500: }
501: out:
1.140 schwarze 502: manconf_free(&conf);
1.123 schwarze 503: mparse_free(mp);
1.157 schwarze 504: mchars_free();
1.47 schwarze 505: mpages_free();
506: ohash_delete(&mpages);
507: ohash_delete(&mlinks);
1.150 schwarze 508: return exitcode;
1.47 schwarze 509: usage:
1.161 schwarze 510: progname = getprogname();
1.83 schwarze 511: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
512: " %s [-aDnpQ] [-Tutf8] dir ...\n"
513: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
514: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 515: " %s [-Q] -t file ...\n",
1.161 schwarze 516: progname, progname, progname, progname, progname);
1.1 schwarze 517:
1.150 schwarze 518: return (int)MANDOCLEVEL_BADARG;
1.47 schwarze 519: }
1.1 schwarze 520:
1.47 schwarze 521: /*
1.180 schwarze 522: * To get a singly linked list in alpha order while inserting entries
523: * at the beginning, process directory entries in reverse alpha order.
524: */
525: static int
526: fts_compare(const FTSENT **a, const FTSENT **b)
527: {
528: return -strcmp((*a)->fts_name, (*b)->fts_name);
529: }
530:
531: /*
1.47 schwarze 532: * Scan a directory tree rooted at "basedir" for manpages.
533: * We use fts(), scanning directory parts along the way for clues to our
534: * section and architecture.
535: *
536: * If use_all has been specified, grok all files.
537: * If not, sanitise paths to the following:
538: *
1.98 schwarze 539: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 540: * or
541: * [./]cat<section>[/<arch>]/<name>.0
542: *
1.169 krw 543: * TODO: accommodate for multi-language directories.
1.47 schwarze 544: */
545: static int
546: treescan(void)
547: {
1.96 schwarze 548: char buf[PATH_MAX];
1.47 schwarze 549: FTS *f;
550: FTSENT *ff;
551: struct mlink *mlink;
1.173 schwarze 552: int gzip;
553: enum form dform;
1.51 schwarze 554: char *dsec, *arch, *fsec, *cp;
555: const char *path;
1.47 schwarze 556: const char *argv[2];
1.1 schwarze 557:
1.47 schwarze 558: argv[0] = ".";
1.193 schwarze 559: argv[1] = NULL;
1.1 schwarze 560:
1.180 schwarze 561: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
562: fts_compare);
1.153 schwarze 563: if (f == NULL) {
1.47 schwarze 564: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 565: say("", "&fts_open");
1.150 schwarze 566: return 0;
1.47 schwarze 567: }
1.2 schwarze 568:
1.47 schwarze 569: dsec = arch = NULL;
570: dform = FORM_NONE;
1.2 schwarze 571:
1.153 schwarze 572: while ((ff = fts_read(f)) != NULL) {
1.47 schwarze 573: path = ff->fts_path + 2;
1.96 schwarze 574: switch (ff->fts_info) {
575:
576: /*
577: * Symbolic links require various sanity checks,
578: * then get handled just like regular files.
579: */
1.98 schwarze 580: case FTS_SL:
1.153 schwarze 581: if (realpath(path, buf) == NULL) {
1.96 schwarze 582: if (warnings)
583: say(path, "&realpath");
584: continue;
585: }
586: if (strstr(buf, basedir) != buf) {
587: if (warnings) say("",
588: "%s: outside base directory", buf);
589: continue;
590: }
591: /* Use logical inode to avoid mpages dupe. */
1.153 schwarze 592: if (stat(path, ff->fts_statp) == -1) {
1.96 schwarze 593: if (warnings)
594: say(path, "&stat");
595: continue;
596: }
597: /* FALLTHROUGH */
598:
1.15 schwarze 599: /*
1.47 schwarze 600: * If we're a regular file, add an mlink by using the
601: * stored directory data and handling the filename.
1.15 schwarze 602: */
1.98 schwarze 603: case FTS_F:
1.153 schwarze 604: if ( ! strcmp(path, MANDOC_DB))
1.47 schwarze 605: continue;
606: if ( ! use_all && ff->fts_level < 2) {
607: if (warnings)
608: say(path, "Extraneous file");
609: continue;
1.81 schwarze 610: }
611: gzip = 0;
612: fsec = NULL;
1.153 schwarze 613: while (fsec == NULL) {
1.81 schwarze 614: fsec = strrchr(ff->fts_name, '.');
1.153 schwarze 615: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.81 schwarze 616: break;
617: gzip = 1;
618: *fsec = '\0';
619: fsec = NULL;
620: }
1.153 schwarze 621: if (fsec == NULL) {
1.47 schwarze 622: if ( ! use_all) {
623: if (warnings)
624: say(path,
625: "No filename suffix");
626: continue;
627: }
1.153 schwarze 628: } else if ( ! strcmp(++fsec, "html")) {
1.47 schwarze 629: if (warnings)
630: say(path, "Skip html");
631: continue;
1.153 schwarze 632: } else if ( ! strcmp(fsec, "ps")) {
1.47 schwarze 633: if (warnings)
634: say(path, "Skip ps");
635: continue;
1.153 schwarze 636: } else if ( ! strcmp(fsec, "pdf")) {
1.47 schwarze 637: if (warnings)
638: say(path, "Skip pdf");
639: continue;
640: } else if ( ! use_all &&
1.153 schwarze 641: ((dform == FORM_SRC &&
1.138 schwarze 642: strncmp(fsec, dsec, strlen(dsec))) ||
1.153 schwarze 643: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.47 schwarze 644: if (warnings)
645: say(path, "Wrong filename suffix");
646: continue;
647: } else
648: fsec[-1] = '\0';
1.51 schwarze 649:
1.47 schwarze 650: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 651: if (strlcpy(mlink->file, path,
652: sizeof(mlink->file)) >=
653: sizeof(mlink->file)) {
654: say(path, "Filename too long");
655: free(mlink);
656: continue;
657: }
1.47 schwarze 658: mlink->dform = dform;
1.51 schwarze 659: mlink->dsec = dsec;
660: mlink->arch = arch;
661: mlink->name = ff->fts_name;
662: mlink->fsec = fsec;
1.81 schwarze 663: mlink->gzip = gzip;
1.47 schwarze 664: mlink_add(mlink, ff->fts_statp);
665: continue;
1.96 schwarze 666:
1.98 schwarze 667: case FTS_D:
668: case FTS_DP:
1.96 schwarze 669: break;
670:
671: default:
1.47 schwarze 672: if (warnings)
673: say(path, "Not a regular file");
674: continue;
675: }
676:
677: switch (ff->fts_level) {
1.98 schwarze 678: case 0:
1.47 schwarze 679: /* Ignore the root directory. */
680: break;
1.98 schwarze 681: case 1:
1.47 schwarze 682: /*
683: * This might contain manX/ or catX/.
684: * Try to infer this from the name.
685: * If we're not in use_all, enforce it.
686: */
687: cp = ff->fts_name;
1.153 schwarze 688: if (ff->fts_info == FTS_DP) {
689: dform = FORM_NONE;
690: dsec = NULL;
1.47 schwarze 691: break;
1.153 schwarze 692: }
1.15 schwarze 693:
1.153 schwarze 694: if ( ! strncmp(cp, "man", 3)) {
1.47 schwarze 695: dform = FORM_SRC;
696: dsec = cp + 3;
1.153 schwarze 697: } else if ( ! strncmp(cp, "cat", 3)) {
1.47 schwarze 698: dform = FORM_CAT;
699: dsec = cp + 3;
1.51 schwarze 700: } else {
701: dform = FORM_NONE;
702: dsec = NULL;
1.26 schwarze 703: }
1.47 schwarze 704:
1.153 schwarze 705: if (dsec != NULL || use_all)
1.47 schwarze 706: break;
707:
708: if (warnings)
709: say(path, "Unknown directory part");
710: fts_set(f, ff, FTS_SKIP);
711: break;
1.98 schwarze 712: case 2:
1.47 schwarze 713: /*
714: * Possibly our architecture.
715: * If we're descending, keep tabs on it.
716: */
1.153 schwarze 717: if (ff->fts_info != FTS_DP && dsec != NULL)
1.47 schwarze 718: arch = ff->fts_name;
1.51 schwarze 719: else
720: arch = NULL;
1.47 schwarze 721: break;
722: default:
1.153 schwarze 723: if (ff->fts_info == FTS_DP || use_all)
1.47 schwarze 724: break;
725: if (warnings)
726: say(path, "Extraneous directory part");
727: fts_set(f, ff, FTS_SKIP);
728: break;
1.14 schwarze 729: }
1.47 schwarze 730: }
731:
732: fts_close(f);
1.150 schwarze 733: return 1;
1.47 schwarze 734: }
1.1 schwarze 735:
1.47 schwarze 736: /*
737: * Add a file to the mlinks table.
738: * Do not verify that it's a "valid" looking manpage (we'll do that
739: * later).
740: *
741: * Try to infer the manual section, architecture, and page name from the
742: * path, assuming it looks like
743: *
1.98 schwarze 744: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 745: * or
746: * [./]cat<section>[/<arch>]/<name>.0
747: *
748: * See treescan() for the fts(3) version of this.
749: */
750: static void
751: filescan(const char *file)
752: {
753: char buf[PATH_MAX];
754: struct stat st;
755: struct mlink *mlink;
756: char *p, *start;
757:
758: assert(use_all);
759:
760: if (0 == strncmp(file, "./", 2))
761: file += 2;
762:
1.96 schwarze 763: /*
764: * We have to do lstat(2) before realpath(3) loses
765: * the information whether this is a symbolic link.
766: * We need to know that because for symbolic links,
767: * we want to use the orginal file name, while for
768: * regular files, we want to use the real path.
769: */
770: if (-1 == lstat(file, &st)) {
771: exitcode = (int)MANDOCLEVEL_BADARG;
772: say(file, "&lstat");
773: return;
774: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
775: exitcode = (int)MANDOCLEVEL_BADARG;
776: say(file, "Not a regular file");
777: return;
778: }
779:
780: /*
781: * We have to resolve the file name to the real path
782: * in any case for the base directory check.
783: */
1.47 schwarze 784: if (NULL == realpath(file, buf)) {
785: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 786: say(file, "&realpath");
1.47 schwarze 787: return;
1.63 schwarze 788: }
789:
1.109 schwarze 790: if (OP_TEST == op)
1.63 schwarze 791: start = buf;
1.109 schwarze 792: else if (strstr(buf, basedir) == buf)
793: start = buf + strlen(basedir);
1.63 schwarze 794: else {
1.47 schwarze 795: exitcode = (int)MANDOCLEVEL_BADARG;
796: say("", "%s: outside base directory", buf);
797: return;
1.63 schwarze 798: }
799:
1.96 schwarze 800: /*
801: * Now we are sure the file is inside our tree.
802: * If it is a symbolic link, ignore the real path
803: * and use the original name.
804: * This implies passing stuff like "cat1/../man1/foo.1"
805: * on the command line won't work. So don't do that.
806: * Note the stat(2) can still fail if the link target
807: * doesn't exist.
808: */
809: if (S_IFLNK & st.st_mode) {
810: if (-1 == stat(buf, &st)) {
811: exitcode = (int)MANDOCLEVEL_BADARG;
812: say(file, "&stat");
813: return;
814: }
1.100 schwarze 815: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
816: say(file, "Filename too long");
817: return;
818: }
1.109 schwarze 819: start = buf;
820: if (OP_TEST != op && strstr(buf, basedir) == buf)
821: start += strlen(basedir);
1.1 schwarze 822: }
1.63 schwarze 823:
1.47 schwarze 824: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 825: mlink->dform = FORM_NONE;
1.100 schwarze 826: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
827: sizeof(mlink->file)) {
828: say(start, "Filename too long");
1.134 schwarze 829: free(mlink);
1.100 schwarze 830: return;
1.188 schwarze 831: }
832:
833: /*
834: * In test mode or when the original name is absolute
835: * but outside our tree, guess the base directory.
836: */
837:
838: if (op == OP_TEST || (start == buf && *start == '/')) {
839: if (strncmp(buf, "man/", 4) == 0)
840: start = buf + 4;
841: else if ((start = strstr(buf, "/man/")) != NULL)
842: start += 5;
843: else
844: start = buf;
1.100 schwarze 845: }
1.1 schwarze 846:
1.10 schwarze 847: /*
1.47 schwarze 848: * First try to guess our directory structure.
849: * If we find a separator, try to look for man* or cat*.
850: * If we find one of these and what's underneath is a directory,
851: * assume it's an architecture.
1.10 schwarze 852: */
1.47 schwarze 853: if (NULL != (p = strchr(start, '/'))) {
854: *p++ = '\0';
855: if (0 == strncmp(start, "man", 3)) {
856: mlink->dform = FORM_SRC;
1.51 schwarze 857: mlink->dsec = start + 3;
1.47 schwarze 858: } else if (0 == strncmp(start, "cat", 3)) {
859: mlink->dform = FORM_CAT;
1.51 schwarze 860: mlink->dsec = start + 3;
1.47 schwarze 861: }
1.10 schwarze 862:
1.47 schwarze 863: start = p;
864: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
865: *p++ = '\0';
1.51 schwarze 866: mlink->arch = start;
1.47 schwarze 867: start = p;
1.41 deraadt 868: }
1.47 schwarze 869: }
1.7 schwarze 870:
1.47 schwarze 871: /*
872: * Now check the file suffix.
873: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
874: */
875: p = strrchr(start, '\0');
876: while (p-- > start && '/' != *p && '.' != *p)
877: /* Loop. */ ;
878:
879: if ('.' == *p) {
880: *p++ = '\0';
1.51 schwarze 881: mlink->fsec = p;
1.47 schwarze 882: }
1.41 deraadt 883:
1.47 schwarze 884: /*
885: * Now try to parse the name.
886: * Use the filename portion of the path.
887: */
888: mlink->name = start;
889: if (NULL != (p = strrchr(start, '/'))) {
890: mlink->name = p + 1;
891: *p = '\0';
892: }
893: mlink_add(mlink, &st);
894: }
1.2 schwarze 895:
1.47 schwarze 896: static void
897: mlink_add(struct mlink *mlink, const struct stat *st)
898: {
899: struct inodev inodev;
900: struct mpage *mpage;
901: unsigned int slot;
902:
903: assert(NULL != mlink->file);
904:
1.51 schwarze 905: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
906: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
907: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
908: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 909:
910: if ('0' == *mlink->fsec) {
911: free(mlink->fsec);
912: mlink->fsec = mandoc_strdup(mlink->dsec);
913: mlink->fform = FORM_CAT;
914: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
915: mlink->fform = FORM_SRC;
916: else
917: mlink->fform = FORM_NONE;
1.1 schwarze 918:
1.47 schwarze 919: slot = ohash_qlookup(&mlinks, mlink->file);
920: assert(NULL == ohash_find(&mlinks, slot));
921: ohash_insert(&mlinks, slot, mlink);
922:
1.139 schwarze 923: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 924: inodev.st_ino = st->st_ino;
925: inodev.st_dev = st->st_dev;
926: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
927: sizeof(struct inodev), inodev.st_ino);
928: mpage = ohash_find(&mpages, slot);
929: if (NULL == mpage) {
930: mpage = mandoc_calloc(1, sizeof(struct mpage));
931: mpage->inodev.st_ino = inodev.st_ino;
932: mpage->inodev.st_dev = inodev.st_dev;
1.173 schwarze 933: mpage->form = FORM_NONE;
1.180 schwarze 934: mpage->next = mpage_head;
935: mpage_head = mpage;
1.47 schwarze 936: ohash_insert(&mpages, slot, mpage);
937: } else
938: mlink->next = mpage->mlinks;
939: mpage->mlinks = mlink;
1.75 schwarze 940: mlink->mpage = mpage;
1.47 schwarze 941: }
1.1 schwarze 942:
1.47 schwarze 943: static void
944: mlink_free(struct mlink *mlink)
945: {
1.1 schwarze 946:
1.47 schwarze 947: free(mlink->dsec);
948: free(mlink->arch);
949: free(mlink->name);
950: free(mlink->fsec);
951: free(mlink);
952: }
1.1 schwarze 953:
1.47 schwarze 954: static void
955: mpages_free(void)
956: {
957: struct mpage *mpage;
958: struct mlink *mlink;
959:
1.180 schwarze 960: while ((mpage = mpage_head) != NULL) {
961: while ((mlink = mpage->mlinks) != NULL) {
1.47 schwarze 962: mpage->mlinks = mlink->next;
963: mlink_free(mlink);
964: }
1.180 schwarze 965: mpage_head = mpage->next;
1.47 schwarze 966: free(mpage->sec);
967: free(mpage->arch);
968: free(mpage->title);
969: free(mpage->desc);
970: free(mpage);
971: }
972: }
1.1 schwarze 973:
1.47 schwarze 974: /*
975: * For each mlink to the mpage, check whether the path looks like
976: * it is formatted, and if it does, check whether a source manual
977: * exists by the same name, ignoring the suffix.
978: * If both conditions hold, drop the mlink.
979: */
980: static void
981: mlinks_undupe(struct mpage *mpage)
982: {
983: char buf[PATH_MAX];
984: struct mlink **prev;
985: struct mlink *mlink;
986: char *bufp;
987:
988: mpage->form = FORM_CAT;
989: prev = &mpage->mlinks;
990: while (NULL != (mlink = *prev)) {
991: if (FORM_CAT != mlink->dform) {
992: mpage->form = FORM_NONE;
993: goto nextlink;
994: }
1.100 schwarze 995: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 996: bufp = strstr(buf, "cat");
997: assert(NULL != bufp);
998: memcpy(bufp, "man", 3);
999: if (NULL != (bufp = strrchr(buf, '.')))
1000: *++bufp = '\0';
1.100 schwarze 1001: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1002: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1003: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1004: goto nextlink;
1005: if (warnings)
1006: say(mlink->file, "Man source exists: %s", buf);
1007: if (use_all)
1008: goto nextlink;
1009: *prev = mlink->next;
1010: mlink_free(mlink);
1011: continue;
1012: nextlink:
1013: prev = &(*prev)->next;
1.1 schwarze 1014: }
1.47 schwarze 1015: }
1.1 schwarze 1016:
1.88 schwarze 1017: static void
1.50 schwarze 1018: mlink_check(struct mpage *mpage, struct mlink *mlink)
1019: {
1.88 schwarze 1020: struct str *str;
1021: unsigned int slot;
1.50 schwarze 1022:
1023: /*
1024: * Check whether the manual section given in a file
1025: * agrees with the directory where the file is located.
1026: * Some manuals have suffixes like (3p) on their
1027: * section number either inside the file or in the
1028: * directory name, some are linked into more than one
1029: * section, like encrypt(1) = makekey(8).
1030: */
1031:
1032: if (FORM_SRC == mpage->form &&
1.88 schwarze 1033: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1034: say(mlink->file, "Section \"%s\" manual in %s directory",
1035: mpage->sec, mlink->dsec);
1036:
1037: /*
1038: * Manual page directories exist for each kernel
1039: * architecture as returned by machine(1).
1040: * However, many manuals only depend on the
1041: * application architecture as returned by arch(1).
1042: * For example, some (2/ARM) manuals are shared
1043: * across the "armish" and "zaurus" kernel
1044: * architectures.
1045: * A few manuals are even shared across completely
1046: * different architectures, for example fdformat(1)
1.178 schwarze 1047: * on amd64, i386, and sparc64.
1.50 schwarze 1048: */
1049:
1.88 schwarze 1050: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1051: say(mlink->file, "Architecture \"%s\" manual in "
1052: "\"%s\" directory", mpage->arch, mlink->arch);
1053:
1.88 schwarze 1054: /*
1055: * XXX
1.90 schwarze 1056: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1057: */
1058:
1059: if (FORM_CAT == mpage->form)
1060: return;
1061:
1062: /*
1063: * Check whether this mlink
1064: * appears as a name in the NAME section.
1065: */
1.50 schwarze 1066:
1.90 schwarze 1067: slot = ohash_qlookup(&names, mlink->name);
1068: str = ohash_find(&names, slot);
1.88 schwarze 1069: assert(NULL != str);
1.90 schwarze 1070: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1071: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1072: }
1073:
1.47 schwarze 1074: /*
1075: * Run through the files in the global vector "mpages"
1076: * and add them to the database specified in "basedir".
1077: *
1078: * This handles the parsing scheme itself, using the cues of directory
1079: * and filename to determine whether the file is parsable or not.
1080: */
1081: static void
1.173 schwarze 1082: mpages_merge(struct dba *dba, struct mparse *mp)
1.47 schwarze 1083: {
1.180 schwarze 1084: struct mpage *mpage, *mpage_dest;
1.75 schwarze 1085: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1086: struct roff_man *man;
1.75 schwarze 1087: char *sodest;
1.69 schwarze 1088: char *cp;
1.117 schwarze 1089: int fd;
1.64 schwarze 1090:
1.180 schwarze 1091: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.47 schwarze 1092: mlinks_undupe(mpage);
1.180 schwarze 1093: if ((mlink = mpage->mlinks) == NULL)
1094: continue;
1.1 schwarze 1095:
1.90 schwarze 1096: name_mask = NAME_MASK;
1.156 schwarze 1097: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1098: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.47 schwarze 1099: mparse_reset(mp);
1100: man = NULL;
1.81 schwarze 1101: sodest = NULL;
1102:
1.166 schwarze 1103: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.136 schwarze 1104: say(mlink->file, "&open");
1.117 schwarze 1105: goto nextpage;
1.81 schwarze 1106: }
1.11 schwarze 1107:
1108: /*
1.137 schwarze 1109: * Interpret the file as mdoc(7) or man(7) source
1110: * code, unless it is known to be formatted.
1.11 schwarze 1111: */
1.136 schwarze 1112: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1113: mparse_readfd(mp, fd, mlink->file);
1.165 schwarze 1114: close(fd);
1.189 schwarze 1115: fd = -1;
1.145 schwarze 1116: mparse_result(mp, &man, &sodest);
1.47 schwarze 1117: }
1.11 schwarze 1118:
1.126 schwarze 1119: if (sodest != NULL) {
1.75 schwarze 1120: mlink_dest = ohash_find(&mlinks,
1121: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1122: if (mlink_dest == NULL) {
1123: mandoc_asprintf(&cp, "%s.gz", sodest);
1124: mlink_dest = ohash_find(&mlinks,
1125: ohash_qlookup(&mlinks, cp));
1126: free(cp);
1127: }
1128: if (mlink_dest != NULL) {
1.75 schwarze 1129:
1130: /* The .so target exists. */
1131:
1132: mpage_dest = mlink_dest->mpage;
1133: while (1) {
1134: mlink->mpage = mpage_dest;
1135:
1136: /*
1137: * If the target was already
1138: * processed, add the links
1139: * to the database now.
1140: * Otherwise, this will
1141: * happen when we come
1142: * to the target.
1143: */
1144:
1.173 schwarze 1145: if (mpage_dest->dba != NULL)
1146: dbadd_mlink(mlink);
1.75 schwarze 1147:
1.126 schwarze 1148: if (mlink->next == NULL)
1.75 schwarze 1149: break;
1150: mlink = mlink->next;
1151: }
1152:
1153: /* Move all links to the target. */
1154:
1155: mlink->next = mlink_dest->next;
1156: mlink_dest->next = mpage->mlinks;
1157: mpage->mlinks = NULL;
1158: }
1.81 schwarze 1159: goto nextpage;
1.145 schwarze 1160: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.159 schwarze 1161: mdoc_validate(man);
1.47 schwarze 1162: mpage->form = FORM_SRC;
1.146 schwarze 1163: mpage->sec = man->meta.msec;
1.112 schwarze 1164: mpage->sec = mandoc_strdup(
1.126 schwarze 1165: mpage->sec == NULL ? "" : mpage->sec);
1.146 schwarze 1166: mpage->arch = man->meta.arch;
1.47 schwarze 1167: mpage->arch = mandoc_strdup(
1.126 schwarze 1168: mpage->arch == NULL ? "" : mpage->arch);
1.146 schwarze 1169: mpage->title = mandoc_strdup(man->meta.title);
1.145 schwarze 1170: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.160 schwarze 1171: man_validate(man);
1.189 schwarze 1172: if (*man->meta.msec != '\0' ||
1.194 schwarze 1173: *man->meta.title != '\0') {
1.189 schwarze 1174: mpage->form = FORM_SRC;
1175: mpage->sec = mandoc_strdup(man->meta.msec);
1176: mpage->arch = mandoc_strdup(mlink->arch);
1177: mpage->title = mandoc_strdup(man->meta.title);
1178: } else
1179: man = NULL;
1180: }
1181:
1182: assert(mpage->desc == NULL);
1183: if (man == NULL) {
1.47 schwarze 1184: mpage->form = FORM_CAT;
1.136 schwarze 1185: mpage->sec = mandoc_strdup(mlink->dsec);
1186: mpage->arch = mandoc_strdup(mlink->arch);
1187: mpage->title = mandoc_strdup(mlink->name);
1.189 schwarze 1188: parse_cat(mpage, fd);
1189: } else if (man->macroset == MACROSET_MDOC)
1.146 schwarze 1190: parse_mdoc(mpage, &man->meta, man->first);
1.189 schwarze 1191: else
1.146 schwarze 1192: parse_man(mpage, &man->meta, man->first);
1.191 schwarze 1193: if (mpage->desc == NULL) {
1194: mpage->desc = mandoc_strdup(mlink->name);
1195: if (warnings)
1196: say(mlink->file, "No one-line description, "
1197: "using filename \"%s\"", mlink->name);
1198: }
1.88 schwarze 1199:
1.187 schwarze 1200: for (mlink = mpage->mlinks;
1201: mlink != NULL;
1202: mlink = mlink->next) {
1203: putkey(mpage, mlink->name, NAME_FILE);
1204: if (warnings && !use_all)
1.88 schwarze 1205: mlink_check(mpage, mlink);
1.187 schwarze 1206: }
1.6 schwarze 1207:
1.173 schwarze 1208: dbadd(dba, mpage);
1.81 schwarze 1209:
1210: nextpage:
1.47 schwarze 1211: ohash_delete(&strings);
1.90 schwarze 1212: ohash_delete(&names);
1.47 schwarze 1213: }
1214: }
1.6 schwarze 1215:
1.47 schwarze 1216: static void
1.81 schwarze 1217: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1218: {
1219: FILE *stream;
1.190 schwarze 1220: struct mlink *mlink;
1221: char *line, *p, *title, *sec;
1.162 schwarze 1222: size_t linesz, plen, titlesz;
1223: ssize_t len;
1224: int offs;
1.1 schwarze 1225:
1.190 schwarze 1226: mlink = mpage->mlinks;
1227: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1228: if (stream == NULL) {
1229: if (fd != -1)
1.114 doug 1230: close(fd);
1.47 schwarze 1231: if (warnings)
1.190 schwarze 1232: say(mlink->file, "&fopen");
1.47 schwarze 1233: return;
1234: }
1.1 schwarze 1235:
1.162 schwarze 1236: line = NULL;
1237: linesz = 0;
1238:
1.190 schwarze 1239: /* Parse the section number from the header line. */
1240:
1241: while (getline(&line, &linesz, stream) != -1) {
1242: if (*line == '\n')
1243: continue;
1244: if ((sec = strchr(line, '(')) == NULL)
1245: break;
1246: if ((p = strchr(++sec, ')')) == NULL)
1247: break;
1248: free(mpage->sec);
1249: mpage->sec = mandoc_strndup(sec, p - sec);
1250: if (warnings && *mlink->dsec != '\0' &&
1251: strcasecmp(mpage->sec, mlink->dsec))
1252: say(mlink->file,
1253: "Section \"%s\" manual in %s directory",
1254: mpage->sec, mlink->dsec);
1255: break;
1256: }
1257:
1.47 schwarze 1258: /* Skip to first blank line. */
1.1 schwarze 1259:
1.190 schwarze 1260: while (line == NULL || *line != '\n')
1261: if (getline(&line, &linesz, stream) == -1)
1.47 schwarze 1262: break;
1.1 schwarze 1263:
1.47 schwarze 1264: /*
1265: * Assume the first line that is not indented
1266: * is the first section header. Skip to it.
1267: */
1.1 schwarze 1268:
1.162 schwarze 1269: while (getline(&line, &linesz, stream) != -1)
1270: if (*line != '\n' && *line != ' ')
1.47 schwarze 1271: break;
1.98 schwarze 1272:
1.47 schwarze 1273: /*
1274: * Read up until the next section into a buffer.
1275: * Strip the leading and trailing newline from each read line,
1276: * appending a trailing space.
1277: * Ignore empty (whitespace-only) lines.
1278: */
1.28 schwarze 1279:
1.47 schwarze 1280: titlesz = 0;
1281: title = NULL;
1.38 schwarze 1282:
1.162 schwarze 1283: while ((len = getline(&line, &linesz, stream)) != -1) {
1284: if (*line != ' ')
1.47 schwarze 1285: break;
1.162 schwarze 1286: offs = 0;
1287: while (isspace((unsigned char)line[offs]))
1288: offs++;
1289: if (line[offs] == '\0')
1.47 schwarze 1290: continue;
1.162 schwarze 1291: title = mandoc_realloc(title, titlesz + len - offs);
1292: memcpy(title + titlesz, line + offs, len - offs);
1293: titlesz += len - offs;
1.47 schwarze 1294: title[titlesz - 1] = ' ';
1295: }
1.162 schwarze 1296: free(line);
1.28 schwarze 1297:
1.47 schwarze 1298: /*
1299: * If no page content can be found, or the input line
1300: * is already the next section header, or there is no
1301: * trailing newline, reuse the page title as the page
1302: * description.
1303: */
1.1 schwarze 1304:
1.47 schwarze 1305: if (NULL == title || '\0' == *title) {
1306: if (warnings)
1.190 schwarze 1307: say(mlink->file, "Cannot find NAME section");
1.47 schwarze 1308: fclose(stream);
1309: free(title);
1310: return;
1311: }
1.24 schwarze 1312:
1.162 schwarze 1313: title[titlesz - 1] = '\0';
1.24 schwarze 1314:
1.47 schwarze 1315: /*
1316: * Skip to the first dash.
1317: * Use the remaining line as the description (no more than 70
1318: * bytes).
1319: */
1.28 schwarze 1320:
1.47 schwarze 1321: if (NULL != (p = strstr(title, "- "))) {
1322: for (p += 2; ' ' == *p || '\b' == *p; p++)
1323: /* Skip to next word. */ ;
1324: } else {
1325: if (warnings)
1.191 schwarze 1326: say(mlink->file, "No dash in title line, "
1327: "reusing \"%s\" as one-line description", title);
1.47 schwarze 1328: p = title;
1329: }
1.1 schwarze 1330:
1.47 schwarze 1331: plen = strlen(p);
1.1 schwarze 1332:
1.47 schwarze 1333: /* Strip backspace-encoding from line. */
1.1 schwarze 1334:
1.47 schwarze 1335: while (NULL != (line = memchr(p, '\b', plen))) {
1336: len = line - p;
1337: if (0 == len) {
1338: memmove(line, line + 1, plen--);
1339: continue;
1.98 schwarze 1340: }
1.47 schwarze 1341: memmove(line - 1, line + 1, plen - len);
1342: plen -= 2;
1343: }
1.1 schwarze 1344:
1.47 schwarze 1345: mpage->desc = mandoc_strdup(p);
1346: fclose(stream);
1347: free(title);
1348: }
1.16 schwarze 1349:
1.47 schwarze 1350: /*
1351: * Put a type/word pair into the word database for this particular file.
1352: */
1353: static void
1.69 schwarze 1354: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1355: {
1356: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1357: }
1358:
1359: /*
1.47 schwarze 1360: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1361: */
1362: static void
1.47 schwarze 1363: putmdockey(const struct mpage *mpage,
1.184 schwarze 1364: const struct roff_node *n, uint64_t m, int taboo)
1.2 schwarze 1365: {
1.16 schwarze 1366:
1.47 schwarze 1367: for ( ; NULL != n; n = n->next) {
1.184 schwarze 1368: if (n->flags & taboo)
1369: continue;
1.47 schwarze 1370: if (NULL != n->child)
1.184 schwarze 1371: putmdockey(mpage, n->child, m, taboo);
1.141 schwarze 1372: if (n->type == ROFFT_TEXT)
1.47 schwarze 1373: putkey(mpage, n->string, m);
1374: }
1375: }
1.16 schwarze 1376:
1.47 schwarze 1377: static void
1.143 schwarze 1378: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1379: const struct roff_node *n)
1.47 schwarze 1380: {
1.142 schwarze 1381: const struct roff_node *head, *body;
1.78 schwarze 1382: char *start, *title;
1.47 schwarze 1383: char byte;
1.78 schwarze 1384: size_t sz;
1.16 schwarze 1385:
1.168 schwarze 1386: if (n == NULL)
1.47 schwarze 1387: return;
1.16 schwarze 1388:
1.47 schwarze 1389: /*
1390: * We're only searching for one thing: the first text child in
1391: * the BODY of a NAME section. Since we don't keep track of
1392: * sections in -man, run some hoops to find out whether we're in
1393: * the correct section or not.
1394: */
1.16 schwarze 1395:
1.141 schwarze 1396: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1397: body = n;
1.168 schwarze 1398: if ((head = body->parent->head) != NULL &&
1399: (head = head->child) != NULL &&
1400: head->next == NULL &&
1.141 schwarze 1401: head->type == ROFFT_TEXT &&
1.168 schwarze 1402: strcmp(head->string, "NAME") == 0 &&
1403: body->child != NULL) {
1.2 schwarze 1404:
1.47 schwarze 1405: /*
1406: * Suck the entire NAME section into memory.
1407: * Yes, we might run away.
1408: * But too many manuals have big, spread-out
1409: * NAME sections over many lines.
1410: */
1.2 schwarze 1411:
1.78 schwarze 1412: title = NULL;
1.147 schwarze 1413: deroff(&title, body);
1.47 schwarze 1414: if (NULL == title)
1415: return;
1.16 schwarze 1416:
1.98 schwarze 1417: /*
1.47 schwarze 1418: * Go through a special heuristic dance here.
1419: * Conventionally, one or more manual names are
1420: * comma-specified prior to a whitespace, then a
1421: * dash, then a description. Try to puzzle out
1422: * the name parts here.
1423: */
1.16 schwarze 1424:
1.78 schwarze 1425: start = title;
1.47 schwarze 1426: for ( ;; ) {
1427: sz = strcspn(start, " ,");
1428: if ('\0' == start[sz])
1429: break;
1.1 schwarze 1430:
1.47 schwarze 1431: byte = start[sz];
1432: start[sz] = '\0';
1.67 schwarze 1433:
1434: /*
1435: * Assume a stray trailing comma in the
1436: * name list if a name begins with a dash.
1437: */
1438:
1439: if ('-' == start[0] ||
1440: ('\\' == start[0] && '-' == start[1]))
1441: break;
1.1 schwarze 1442:
1.90 schwarze 1443: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1444: if ( ! (mpage->name_head_done ||
1445: strcasecmp(start, meta->title))) {
1446: putkey(mpage, start, NAME_HEAD);
1447: mpage->name_head_done = 1;
1448: }
1.1 schwarze 1449:
1.47 schwarze 1450: if (' ' == byte) {
1451: start += sz + 1;
1452: break;
1453: }
1.1 schwarze 1454:
1.47 schwarze 1455: assert(',' == byte);
1456: start += sz + 1;
1457: while (' ' == *start)
1458: start++;
1459: }
1.1 schwarze 1460:
1.78 schwarze 1461: if (start == title) {
1.90 schwarze 1462: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1463: if ( ! (mpage->name_head_done ||
1464: strcasecmp(start, meta->title))) {
1465: putkey(mpage, start, NAME_HEAD);
1466: mpage->name_head_done = 1;
1467: }
1.47 schwarze 1468: free(title);
1469: return;
1470: }
1.1 schwarze 1471:
1.47 schwarze 1472: while (isspace((unsigned char)*start))
1473: start++;
1.1 schwarze 1474:
1.47 schwarze 1475: if (0 == strncmp(start, "-", 1))
1476: start += 1;
1477: else if (0 == strncmp(start, "\\-\\-", 4))
1478: start += 4;
1479: else if (0 == strncmp(start, "\\-", 2))
1480: start += 2;
1481: else if (0 == strncmp(start, "\\(en", 4))
1482: start += 4;
1483: else if (0 == strncmp(start, "\\(em", 4))
1484: start += 4;
1.1 schwarze 1485:
1.47 schwarze 1486: while (' ' == *start)
1487: start++;
1.1 schwarze 1488:
1.47 schwarze 1489: mpage->desc = mandoc_strdup(start);
1490: free(title);
1491: return;
1492: }
1493: }
1.1 schwarze 1494:
1.47 schwarze 1495: for (n = n->child; n; n = n->next) {
1496: if (NULL != mpage->desc)
1497: break;
1.129 schwarze 1498: parse_man(mpage, meta, n);
1.1 schwarze 1499: }
1500: }
1501:
1502: static void
1.143 schwarze 1503: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1504: const struct roff_node *n)
1.1 schwarze 1505: {
1506:
1.195 schwarze 1507: for (n = n->child; n != NULL; n = n->next) {
1.196 schwarze 1508: if (n->tok == TOKEN_NONE ||
1509: n->tok < ROFF_MAX ||
1510: n->flags & mdocs[n->tok].taboo)
1.184 schwarze 1511: continue;
1.196 schwarze 1512: assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1.47 schwarze 1513: switch (n->type) {
1.141 schwarze 1514: case ROFFT_ELEM:
1515: case ROFFT_BLOCK:
1516: case ROFFT_HEAD:
1517: case ROFFT_BODY:
1518: case ROFFT_TAIL:
1.195 schwarze 1519: if (mdocs[n->tok].fp != NULL &&
1520: (*mdocs[n->tok].fp)(mpage, meta, n) == 0)
1521: break;
1.47 schwarze 1522: if (mdocs[n->tok].mask)
1523: putmdockey(mpage, n->child,
1.184 schwarze 1524: mdocs[n->tok].mask, mdocs[n->tok].taboo);
1.47 schwarze 1525: break;
1526: default:
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: }