Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.205
1.205 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.204 2017/08/26 15:55:41 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>
1.204 schwarze 20: #include <sys/mman.h>
1.47 schwarze 21: #include <sys/stat.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 }, /* %U */
291: { NULL, 0, 0 }, /* Ta */
1.1 schwarze 292: };
1.195 schwarze 293: static const struct mdoc_handler *const mdocs = __mdocs - MDOC_Dd;
1.1 schwarze 294:
1.98 schwarze 295:
1.1 schwarze 296: int
1.3 schwarze 297: mandocdb(int argc, char *argv[])
1.1 schwarze 298: {
1.140 schwarze 299: struct manconf conf;
300: struct mparse *mp;
1.173 schwarze 301: struct dba *dba;
1.161 schwarze 302: const char *path_arg, *progname;
1.140 schwarze 303: size_t j, sz;
1.47 schwarze 304: int ch, i;
305:
1.204 schwarze 306: if (pledge("stdio rpath wpath cpath", NULL) == -1) {
1.167 schwarze 307: warn("pledge");
1.155 schwarze 308: return (int)MANDOCLEVEL_SYSERR;
309: }
310:
1.140 schwarze 311: memset(&conf, 0, sizeof(conf));
1.47 schwarze 312:
313: /*
1.98 schwarze 314: * We accept a few different invocations.
1.47 schwarze 315: * The CHECKOP macro makes sure that invocation styles don't
316: * clobber each other.
317: */
318: #define CHECKOP(_op, _ch) do \
319: if (OP_DEFAULT != (_op)) { \
1.151 schwarze 320: warnx("-%c: Conflicting option", (_ch)); \
1.47 schwarze 321: goto usage; \
322: } while (/*CONSTCOND*/0)
1.10 schwarze 323:
1.47 schwarze 324: path_arg = NULL;
1.28 schwarze 325: op = OP_DEFAULT;
1.1 schwarze 326:
1.83 schwarze 327: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 328: switch (ch) {
1.98 schwarze 329: case 'a':
1.6 schwarze 330: use_all = 1;
331: break;
1.98 schwarze 332: case 'C':
1.47 schwarze 333: CHECKOP(op, ch);
334: path_arg = optarg;
1.28 schwarze 335: op = OP_CONFFILE;
1.25 schwarze 336: break;
1.98 schwarze 337: case 'D':
1.82 schwarze 338: debug++;
339: break;
1.98 schwarze 340: case 'd':
1.47 schwarze 341: CHECKOP(op, ch);
342: path_arg = optarg;
1.2 schwarze 343: op = OP_UPDATE;
1.1 schwarze 344: break;
1.98 schwarze 345: case 'n':
1.47 schwarze 346: nodb = 1;
347: break;
1.98 schwarze 348: case 'p':
1.83 schwarze 349: warnings = 1;
350: break;
1.98 schwarze 351: case 'Q':
1.73 schwarze 352: mparse_options |= MPARSE_QUICK;
1.59 schwarze 353: break;
1.98 schwarze 354: case 'T':
1.52 schwarze 355: if (strcmp(optarg, "utf8")) {
1.151 schwarze 356: warnx("-T%s: Unsupported output format",
357: optarg);
1.52 schwarze 358: goto usage;
359: }
360: write_utf8 = 1;
361: break;
1.98 schwarze 362: case 't':
1.47 schwarze 363: CHECKOP(op, ch);
1.28 schwarze 364: dup2(STDOUT_FILENO, STDERR_FILENO);
365: op = OP_TEST;
1.47 schwarze 366: nodb = warnings = 1;
1.28 schwarze 367: break;
1.98 schwarze 368: case 'u':
1.47 schwarze 369: CHECKOP(op, ch);
370: path_arg = optarg;
1.1 schwarze 371: op = OP_DELETE;
372: break;
1.98 schwarze 373: case 'v':
1.83 schwarze 374: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 375: break;
1.1 schwarze 376: default:
1.28 schwarze 377: goto usage;
1.1 schwarze 378: }
379:
380: argc -= optind;
381: argv += optind;
382:
1.163 schwarze 383: if (nodb) {
384: if (pledge("stdio rpath", NULL) == -1) {
1.167 schwarze 385: warn("pledge");
1.163 schwarze 386: return (int)MANDOCLEVEL_SYSERR;
387: }
1.155 schwarze 388: }
389:
1.28 schwarze 390: if (OP_CONFFILE == op && argc > 0) {
1.151 schwarze 391: warnx("-C: Too many arguments");
1.28 schwarze 392: goto usage;
393: }
394:
1.47 schwarze 395: exitcode = (int)MANDOCLEVEL_OK;
1.157 schwarze 396: mchars_alloc();
1.202 schwarze 397: mp = mparse_alloc(mparse_options, MANDOCERR_MAX, NULL,
1.200 schwarze 398: MANDOC_OS_OTHER, 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: */
417: use_all = 1;
1.95 schwarze 418: for (i = 0; i < argc; i++)
419: filescan(argv[i]);
1.174 schwarze 420: if (nodb == 0)
1.173 schwarze 421: dbprune(dba);
1.95 schwarze 422: } else {
1.177 schwarze 423: /* Database missing or corrupt. */
1.179 schwarze 424: if (op != OP_UPDATE || errno != ENOENT)
425: say(MANDOC_DB, "%s: Automatically recreating"
426: " from scratch", strerror(errno));
1.97 schwarze 427: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 428: op = OP_DEFAULT;
429: if (0 == treescan())
430: goto out;
1.173 schwarze 431: dba = dba_new(128);
1.95 schwarze 432: }
1.47 schwarze 433: if (OP_DELETE != op)
1.173 schwarze 434: mpages_merge(dba, mp);
435: if (nodb == 0)
436: dbwrite(dba);
437: dba_free(dba);
1.47 schwarze 438: } else {
439: /*
440: * If we have arguments, use them as our manpaths.
1.181 schwarze 441: * If we don't, use man.conf(5).
1.47 schwarze 442: */
443: if (argc > 0) {
1.140 schwarze 444: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 445: argc, sizeof(char *));
1.140 schwarze 446: conf.manpath.sz = (size_t)argc;
1.47 schwarze 447: for (i = 0; i < argc; i++)
1.140 schwarze 448: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 449: } else
1.140 schwarze 450: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 451:
1.140 schwarze 452: if (conf.manpath.sz == 0) {
1.84 schwarze 453: exitcode = (int)MANDOCLEVEL_BADARG;
454: say("", "Empty manpath");
455: }
1.2 schwarze 456:
1.47 schwarze 457: /*
458: * First scan the tree rooted at a base directory, then
459: * build a new database and finally move it into place.
460: * Ignore zero-length directories and strip trailing
461: * slashes.
462: */
1.140 schwarze 463: for (j = 0; j < conf.manpath.sz; j++) {
464: sz = strlen(conf.manpath.paths[j]);
465: if (sz && conf.manpath.paths[j][sz - 1] == '/')
466: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 467: if (0 == sz)
468: continue;
1.2 schwarze 469:
1.47 schwarze 470: if (j) {
1.156 schwarze 471: mandoc_ohash_init(&mpages, 6,
472: offsetof(struct mpage, inodev));
473: mandoc_ohash_init(&mlinks, 6,
474: offsetof(struct mlink, file));
1.47 schwarze 475: }
1.1 schwarze 476:
1.140 schwarze 477: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 478: continue;
1.47 schwarze 479: if (0 == treescan())
1.120 schwarze 480: continue;
1.173 schwarze 481: dba = dba_new(128);
482: mpages_merge(dba, mp);
483: if (nodb == 0)
484: dbwrite(dba);
485: dba_free(dba);
1.28 schwarze 486:
1.140 schwarze 487: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 488: mpages_free();
489: ohash_delete(&mpages);
490: ohash_delete(&mlinks);
491: }
1.2 schwarze 492: }
1.47 schwarze 493: }
494: out:
1.140 schwarze 495: manconf_free(&conf);
1.123 schwarze 496: mparse_free(mp);
1.157 schwarze 497: mchars_free();
1.47 schwarze 498: mpages_free();
499: ohash_delete(&mpages);
500: ohash_delete(&mlinks);
1.150 schwarze 501: return exitcode;
1.47 schwarze 502: usage:
1.161 schwarze 503: progname = getprogname();
1.83 schwarze 504: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
505: " %s [-aDnpQ] [-Tutf8] dir ...\n"
506: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
507: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 508: " %s [-Q] -t file ...\n",
1.161 schwarze 509: progname, progname, progname, progname, progname);
1.1 schwarze 510:
1.150 schwarze 511: return (int)MANDOCLEVEL_BADARG;
1.47 schwarze 512: }
1.1 schwarze 513:
1.47 schwarze 514: /*
1.180 schwarze 515: * To get a singly linked list in alpha order while inserting entries
516: * at the beginning, process directory entries in reverse alpha order.
517: */
518: static int
519: fts_compare(const FTSENT **a, const FTSENT **b)
520: {
521: return -strcmp((*a)->fts_name, (*b)->fts_name);
522: }
523:
524: /*
1.47 schwarze 525: * Scan a directory tree rooted at "basedir" for manpages.
526: * We use fts(), scanning directory parts along the way for clues to our
527: * section and architecture.
528: *
529: * If use_all has been specified, grok all files.
530: * If not, sanitise paths to the following:
531: *
1.98 schwarze 532: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 533: * or
534: * [./]cat<section>[/<arch>]/<name>.0
535: *
1.169 krw 536: * TODO: accommodate for multi-language directories.
1.47 schwarze 537: */
538: static int
539: treescan(void)
540: {
1.96 schwarze 541: char buf[PATH_MAX];
1.47 schwarze 542: FTS *f;
543: FTSENT *ff;
544: struct mlink *mlink;
1.173 schwarze 545: int gzip;
546: enum form dform;
1.51 schwarze 547: char *dsec, *arch, *fsec, *cp;
548: const char *path;
1.47 schwarze 549: const char *argv[2];
1.1 schwarze 550:
1.47 schwarze 551: argv[0] = ".";
1.193 schwarze 552: argv[1] = NULL;
1.1 schwarze 553:
1.180 schwarze 554: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
555: fts_compare);
1.153 schwarze 556: if (f == NULL) {
1.47 schwarze 557: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 558: say("", "&fts_open");
1.150 schwarze 559: return 0;
1.47 schwarze 560: }
1.2 schwarze 561:
1.47 schwarze 562: dsec = arch = NULL;
563: dform = FORM_NONE;
1.2 schwarze 564:
1.153 schwarze 565: while ((ff = fts_read(f)) != NULL) {
1.47 schwarze 566: path = ff->fts_path + 2;
1.96 schwarze 567: switch (ff->fts_info) {
568:
569: /*
570: * Symbolic links require various sanity checks,
571: * then get handled just like regular files.
572: */
1.98 schwarze 573: case FTS_SL:
1.153 schwarze 574: if (realpath(path, buf) == NULL) {
1.96 schwarze 575: if (warnings)
576: say(path, "&realpath");
577: continue;
578: }
579: if (strstr(buf, basedir) != buf) {
580: if (warnings) say("",
581: "%s: outside base directory", buf);
582: continue;
583: }
584: /* Use logical inode to avoid mpages dupe. */
1.153 schwarze 585: if (stat(path, ff->fts_statp) == -1) {
1.96 schwarze 586: if (warnings)
587: say(path, "&stat");
588: continue;
589: }
590: /* FALLTHROUGH */
591:
1.15 schwarze 592: /*
1.47 schwarze 593: * If we're a regular file, add an mlink by using the
594: * stored directory data and handling the filename.
1.15 schwarze 595: */
1.98 schwarze 596: case FTS_F:
1.153 schwarze 597: if ( ! strcmp(path, MANDOC_DB))
1.47 schwarze 598: continue;
599: if ( ! use_all && ff->fts_level < 2) {
600: if (warnings)
601: say(path, "Extraneous file");
602: continue;
1.81 schwarze 603: }
604: gzip = 0;
605: fsec = NULL;
1.153 schwarze 606: while (fsec == NULL) {
1.81 schwarze 607: fsec = strrchr(ff->fts_name, '.');
1.153 schwarze 608: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.81 schwarze 609: break;
610: gzip = 1;
611: *fsec = '\0';
612: fsec = NULL;
613: }
1.153 schwarze 614: if (fsec == NULL) {
1.47 schwarze 615: if ( ! use_all) {
616: if (warnings)
617: say(path,
618: "No filename suffix");
619: continue;
620: }
1.153 schwarze 621: } else if ( ! strcmp(++fsec, "html")) {
1.47 schwarze 622: if (warnings)
623: say(path, "Skip html");
624: continue;
1.153 schwarze 625: } else if ( ! strcmp(fsec, "ps")) {
1.47 schwarze 626: if (warnings)
627: say(path, "Skip ps");
628: continue;
1.153 schwarze 629: } else if ( ! strcmp(fsec, "pdf")) {
1.47 schwarze 630: if (warnings)
631: say(path, "Skip pdf");
632: continue;
633: } else if ( ! use_all &&
1.153 schwarze 634: ((dform == FORM_SRC &&
1.138 schwarze 635: strncmp(fsec, dsec, strlen(dsec))) ||
1.153 schwarze 636: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.47 schwarze 637: if (warnings)
638: say(path, "Wrong filename suffix");
639: continue;
640: } else
641: fsec[-1] = '\0';
1.51 schwarze 642:
1.47 schwarze 643: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 644: if (strlcpy(mlink->file, path,
645: sizeof(mlink->file)) >=
646: sizeof(mlink->file)) {
647: say(path, "Filename too long");
648: free(mlink);
649: continue;
650: }
1.47 schwarze 651: mlink->dform = dform;
1.51 schwarze 652: mlink->dsec = dsec;
653: mlink->arch = arch;
654: mlink->name = ff->fts_name;
655: mlink->fsec = fsec;
1.81 schwarze 656: mlink->gzip = gzip;
1.47 schwarze 657: mlink_add(mlink, ff->fts_statp);
658: continue;
1.96 schwarze 659:
1.98 schwarze 660: case FTS_D:
661: case FTS_DP:
1.96 schwarze 662: break;
663:
664: default:
1.47 schwarze 665: if (warnings)
666: say(path, "Not a regular file");
667: continue;
668: }
669:
670: switch (ff->fts_level) {
1.98 schwarze 671: case 0:
1.47 schwarze 672: /* Ignore the root directory. */
673: break;
1.98 schwarze 674: case 1:
1.47 schwarze 675: /*
676: * This might contain manX/ or catX/.
677: * Try to infer this from the name.
678: * If we're not in use_all, enforce it.
679: */
680: cp = ff->fts_name;
1.153 schwarze 681: if (ff->fts_info == FTS_DP) {
682: dform = FORM_NONE;
683: dsec = NULL;
1.47 schwarze 684: break;
1.153 schwarze 685: }
1.15 schwarze 686:
1.153 schwarze 687: if ( ! strncmp(cp, "man", 3)) {
1.47 schwarze 688: dform = FORM_SRC;
689: dsec = cp + 3;
1.153 schwarze 690: } else if ( ! strncmp(cp, "cat", 3)) {
1.47 schwarze 691: dform = FORM_CAT;
692: dsec = cp + 3;
1.51 schwarze 693: } else {
694: dform = FORM_NONE;
695: dsec = NULL;
1.26 schwarze 696: }
1.47 schwarze 697:
1.153 schwarze 698: if (dsec != NULL || use_all)
1.47 schwarze 699: break;
700:
701: if (warnings)
702: say(path, "Unknown directory part");
703: fts_set(f, ff, FTS_SKIP);
704: break;
1.98 schwarze 705: case 2:
1.47 schwarze 706: /*
707: * Possibly our architecture.
708: * If we're descending, keep tabs on it.
709: */
1.153 schwarze 710: if (ff->fts_info != FTS_DP && dsec != NULL)
1.47 schwarze 711: arch = ff->fts_name;
1.51 schwarze 712: else
713: arch = NULL;
1.47 schwarze 714: break;
715: default:
1.153 schwarze 716: if (ff->fts_info == FTS_DP || use_all)
1.47 schwarze 717: break;
718: if (warnings)
719: say(path, "Extraneous directory part");
720: fts_set(f, ff, FTS_SKIP);
721: break;
1.14 schwarze 722: }
1.47 schwarze 723: }
724:
725: fts_close(f);
1.150 schwarze 726: return 1;
1.47 schwarze 727: }
1.1 schwarze 728:
1.47 schwarze 729: /*
730: * Add a file to the mlinks table.
731: * Do not verify that it's a "valid" looking manpage (we'll do that
732: * later).
733: *
734: * Try to infer the manual section, architecture, and page name from the
735: * path, assuming it looks like
736: *
1.98 schwarze 737: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 738: * or
739: * [./]cat<section>[/<arch>]/<name>.0
740: *
741: * See treescan() for the fts(3) version of this.
742: */
743: static void
744: filescan(const char *file)
745: {
746: char buf[PATH_MAX];
747: struct stat st;
748: struct mlink *mlink;
749: char *p, *start;
750:
751: assert(use_all);
752:
753: if (0 == strncmp(file, "./", 2))
754: file += 2;
755:
1.96 schwarze 756: /*
757: * We have to do lstat(2) before realpath(3) loses
758: * the information whether this is a symbolic link.
759: * We need to know that because for symbolic links,
760: * we want to use the orginal file name, while for
761: * regular files, we want to use the real path.
762: */
763: if (-1 == lstat(file, &st)) {
764: exitcode = (int)MANDOCLEVEL_BADARG;
765: say(file, "&lstat");
766: return;
767: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
768: exitcode = (int)MANDOCLEVEL_BADARG;
769: say(file, "Not a regular file");
770: return;
771: }
772:
773: /*
774: * We have to resolve the file name to the real path
775: * in any case for the base directory check.
776: */
1.47 schwarze 777: if (NULL == realpath(file, buf)) {
778: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 779: say(file, "&realpath");
1.47 schwarze 780: return;
1.63 schwarze 781: }
782:
1.109 schwarze 783: if (OP_TEST == op)
1.63 schwarze 784: start = buf;
1.109 schwarze 785: else if (strstr(buf, basedir) == buf)
786: start = buf + strlen(basedir);
1.63 schwarze 787: else {
1.47 schwarze 788: exitcode = (int)MANDOCLEVEL_BADARG;
789: say("", "%s: outside base directory", buf);
790: return;
1.63 schwarze 791: }
792:
1.96 schwarze 793: /*
794: * Now we are sure the file is inside our tree.
795: * If it is a symbolic link, ignore the real path
796: * and use the original name.
797: * This implies passing stuff like "cat1/../man1/foo.1"
798: * on the command line won't work. So don't do that.
799: * Note the stat(2) can still fail if the link target
800: * doesn't exist.
801: */
802: if (S_IFLNK & st.st_mode) {
803: if (-1 == stat(buf, &st)) {
804: exitcode = (int)MANDOCLEVEL_BADARG;
805: say(file, "&stat");
806: return;
807: }
1.100 schwarze 808: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
809: say(file, "Filename too long");
810: return;
811: }
1.109 schwarze 812: start = buf;
813: if (OP_TEST != op && strstr(buf, basedir) == buf)
814: start += strlen(basedir);
1.1 schwarze 815: }
1.63 schwarze 816:
1.47 schwarze 817: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 818: mlink->dform = FORM_NONE;
1.100 schwarze 819: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
820: sizeof(mlink->file)) {
821: say(start, "Filename too long");
1.134 schwarze 822: free(mlink);
1.100 schwarze 823: return;
1.188 schwarze 824: }
825:
826: /*
827: * In test mode or when the original name is absolute
828: * but outside our tree, guess the base directory.
829: */
830:
831: if (op == OP_TEST || (start == buf && *start == '/')) {
832: if (strncmp(buf, "man/", 4) == 0)
833: start = buf + 4;
834: else if ((start = strstr(buf, "/man/")) != NULL)
835: start += 5;
836: else
837: start = buf;
1.100 schwarze 838: }
1.1 schwarze 839:
1.10 schwarze 840: /*
1.47 schwarze 841: * First try to guess our directory structure.
842: * If we find a separator, try to look for man* or cat*.
843: * If we find one of these and what's underneath is a directory,
844: * assume it's an architecture.
1.10 schwarze 845: */
1.47 schwarze 846: if (NULL != (p = strchr(start, '/'))) {
847: *p++ = '\0';
848: if (0 == strncmp(start, "man", 3)) {
849: mlink->dform = FORM_SRC;
1.51 schwarze 850: mlink->dsec = start + 3;
1.47 schwarze 851: } else if (0 == strncmp(start, "cat", 3)) {
852: mlink->dform = FORM_CAT;
1.51 schwarze 853: mlink->dsec = start + 3;
1.47 schwarze 854: }
1.10 schwarze 855:
1.47 schwarze 856: start = p;
857: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
858: *p++ = '\0';
1.51 schwarze 859: mlink->arch = start;
1.47 schwarze 860: start = p;
1.41 deraadt 861: }
1.47 schwarze 862: }
1.7 schwarze 863:
1.47 schwarze 864: /*
865: * Now check the file suffix.
866: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
867: */
868: p = strrchr(start, '\0');
869: while (p-- > start && '/' != *p && '.' != *p)
870: /* Loop. */ ;
871:
872: if ('.' == *p) {
873: *p++ = '\0';
1.51 schwarze 874: mlink->fsec = p;
1.47 schwarze 875: }
1.41 deraadt 876:
1.47 schwarze 877: /*
878: * Now try to parse the name.
879: * Use the filename portion of the path.
880: */
881: mlink->name = start;
882: if (NULL != (p = strrchr(start, '/'))) {
883: mlink->name = p + 1;
884: *p = '\0';
885: }
886: mlink_add(mlink, &st);
887: }
1.2 schwarze 888:
1.47 schwarze 889: static void
890: mlink_add(struct mlink *mlink, const struct stat *st)
891: {
892: struct inodev inodev;
893: struct mpage *mpage;
894: unsigned int slot;
895:
896: assert(NULL != mlink->file);
897:
1.51 schwarze 898: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
899: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
900: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
901: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 902:
903: if ('0' == *mlink->fsec) {
904: free(mlink->fsec);
905: mlink->fsec = mandoc_strdup(mlink->dsec);
906: mlink->fform = FORM_CAT;
907: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
908: mlink->fform = FORM_SRC;
909: else
910: mlink->fform = FORM_NONE;
1.1 schwarze 911:
1.47 schwarze 912: slot = ohash_qlookup(&mlinks, mlink->file);
913: assert(NULL == ohash_find(&mlinks, slot));
914: ohash_insert(&mlinks, slot, mlink);
915:
1.139 schwarze 916: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 917: inodev.st_ino = st->st_ino;
918: inodev.st_dev = st->st_dev;
919: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
920: sizeof(struct inodev), inodev.st_ino);
921: mpage = ohash_find(&mpages, slot);
922: if (NULL == mpage) {
923: mpage = mandoc_calloc(1, sizeof(struct mpage));
924: mpage->inodev.st_ino = inodev.st_ino;
925: mpage->inodev.st_dev = inodev.st_dev;
1.173 schwarze 926: mpage->form = FORM_NONE;
1.180 schwarze 927: mpage->next = mpage_head;
928: mpage_head = mpage;
1.47 schwarze 929: ohash_insert(&mpages, slot, mpage);
930: } else
931: mlink->next = mpage->mlinks;
932: mpage->mlinks = mlink;
1.75 schwarze 933: mlink->mpage = mpage;
1.47 schwarze 934: }
1.1 schwarze 935:
1.47 schwarze 936: static void
937: mlink_free(struct mlink *mlink)
938: {
1.1 schwarze 939:
1.47 schwarze 940: free(mlink->dsec);
941: free(mlink->arch);
942: free(mlink->name);
943: free(mlink->fsec);
944: free(mlink);
945: }
1.1 schwarze 946:
1.47 schwarze 947: static void
948: mpages_free(void)
949: {
950: struct mpage *mpage;
951: struct mlink *mlink;
952:
1.180 schwarze 953: while ((mpage = mpage_head) != NULL) {
954: while ((mlink = mpage->mlinks) != NULL) {
1.47 schwarze 955: mpage->mlinks = mlink->next;
956: mlink_free(mlink);
957: }
1.180 schwarze 958: mpage_head = mpage->next;
1.47 schwarze 959: free(mpage->sec);
960: free(mpage->arch);
961: free(mpage->title);
962: free(mpage->desc);
963: free(mpage);
964: }
965: }
1.1 schwarze 966:
1.47 schwarze 967: /*
968: * For each mlink to the mpage, check whether the path looks like
969: * it is formatted, and if it does, check whether a source manual
970: * exists by the same name, ignoring the suffix.
971: * If both conditions hold, drop the mlink.
972: */
973: static void
974: mlinks_undupe(struct mpage *mpage)
975: {
976: char buf[PATH_MAX];
977: struct mlink **prev;
978: struct mlink *mlink;
979: char *bufp;
980:
981: mpage->form = FORM_CAT;
982: prev = &mpage->mlinks;
983: while (NULL != (mlink = *prev)) {
984: if (FORM_CAT != mlink->dform) {
985: mpage->form = FORM_NONE;
986: goto nextlink;
987: }
1.100 schwarze 988: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 989: bufp = strstr(buf, "cat");
990: assert(NULL != bufp);
991: memcpy(bufp, "man", 3);
992: if (NULL != (bufp = strrchr(buf, '.')))
993: *++bufp = '\0';
1.100 schwarze 994: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 995: if (NULL == ohash_find(&mlinks,
1.98 schwarze 996: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 997: goto nextlink;
998: if (warnings)
999: say(mlink->file, "Man source exists: %s", buf);
1000: if (use_all)
1001: goto nextlink;
1002: *prev = mlink->next;
1003: mlink_free(mlink);
1004: continue;
1005: nextlink:
1006: prev = &(*prev)->next;
1.1 schwarze 1007: }
1.47 schwarze 1008: }
1.1 schwarze 1009:
1.88 schwarze 1010: static void
1.50 schwarze 1011: mlink_check(struct mpage *mpage, struct mlink *mlink)
1012: {
1.88 schwarze 1013: struct str *str;
1014: unsigned int slot;
1.50 schwarze 1015:
1016: /*
1017: * Check whether the manual section given in a file
1018: * agrees with the directory where the file is located.
1019: * Some manuals have suffixes like (3p) on their
1020: * section number either inside the file or in the
1021: * directory name, some are linked into more than one
1022: * section, like encrypt(1) = makekey(8).
1023: */
1024:
1025: if (FORM_SRC == mpage->form &&
1.88 schwarze 1026: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1027: say(mlink->file, "Section \"%s\" manual in %s directory",
1028: mpage->sec, mlink->dsec);
1029:
1030: /*
1031: * Manual page directories exist for each kernel
1032: * architecture as returned by machine(1).
1033: * However, many manuals only depend on the
1034: * application architecture as returned by arch(1).
1035: * For example, some (2/ARM) manuals are shared
1036: * across the "armish" and "zaurus" kernel
1037: * architectures.
1038: * A few manuals are even shared across completely
1039: * different architectures, for example fdformat(1)
1.178 schwarze 1040: * on amd64, i386, and sparc64.
1.50 schwarze 1041: */
1042:
1.88 schwarze 1043: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1044: say(mlink->file, "Architecture \"%s\" manual in "
1045: "\"%s\" directory", mpage->arch, mlink->arch);
1046:
1.88 schwarze 1047: /*
1048: * XXX
1.90 schwarze 1049: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1050: */
1051:
1052: if (FORM_CAT == mpage->form)
1053: return;
1054:
1055: /*
1056: * Check whether this mlink
1057: * appears as a name in the NAME section.
1058: */
1.50 schwarze 1059:
1.90 schwarze 1060: slot = ohash_qlookup(&names, mlink->name);
1061: str = ohash_find(&names, slot);
1.88 schwarze 1062: assert(NULL != str);
1.90 schwarze 1063: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1064: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1065: }
1066:
1.47 schwarze 1067: /*
1068: * Run through the files in the global vector "mpages"
1069: * and add them to the database specified in "basedir".
1070: *
1071: * This handles the parsing scheme itself, using the cues of directory
1072: * and filename to determine whether the file is parsable or not.
1073: */
1074: static void
1.173 schwarze 1075: mpages_merge(struct dba *dba, struct mparse *mp)
1.47 schwarze 1076: {
1.180 schwarze 1077: struct mpage *mpage, *mpage_dest;
1.75 schwarze 1078: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1079: struct roff_man *man;
1.75 schwarze 1080: char *sodest;
1.69 schwarze 1081: char *cp;
1.117 schwarze 1082: int fd;
1.64 schwarze 1083:
1.180 schwarze 1084: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.47 schwarze 1085: mlinks_undupe(mpage);
1.180 schwarze 1086: if ((mlink = mpage->mlinks) == NULL)
1087: continue;
1.1 schwarze 1088:
1.90 schwarze 1089: name_mask = NAME_MASK;
1.156 schwarze 1090: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1091: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.47 schwarze 1092: mparse_reset(mp);
1093: man = NULL;
1.81 schwarze 1094: sodest = NULL;
1095:
1.166 schwarze 1096: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.136 schwarze 1097: say(mlink->file, "&open");
1.117 schwarze 1098: goto nextpage;
1.81 schwarze 1099: }
1.11 schwarze 1100:
1101: /*
1.137 schwarze 1102: * Interpret the file as mdoc(7) or man(7) source
1103: * code, unless it is known to be formatted.
1.11 schwarze 1104: */
1.136 schwarze 1105: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1106: mparse_readfd(mp, fd, mlink->file);
1.165 schwarze 1107: close(fd);
1.189 schwarze 1108: fd = -1;
1.145 schwarze 1109: mparse_result(mp, &man, &sodest);
1.47 schwarze 1110: }
1.11 schwarze 1111:
1.126 schwarze 1112: if (sodest != NULL) {
1.75 schwarze 1113: mlink_dest = ohash_find(&mlinks,
1114: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1115: if (mlink_dest == NULL) {
1116: mandoc_asprintf(&cp, "%s.gz", sodest);
1117: mlink_dest = ohash_find(&mlinks,
1118: ohash_qlookup(&mlinks, cp));
1119: free(cp);
1120: }
1121: if (mlink_dest != NULL) {
1.75 schwarze 1122:
1123: /* The .so target exists. */
1124:
1125: mpage_dest = mlink_dest->mpage;
1126: while (1) {
1127: mlink->mpage = mpage_dest;
1128:
1129: /*
1130: * If the target was already
1131: * processed, add the links
1132: * to the database now.
1133: * Otherwise, this will
1134: * happen when we come
1135: * to the target.
1136: */
1137:
1.173 schwarze 1138: if (mpage_dest->dba != NULL)
1139: dbadd_mlink(mlink);
1.75 schwarze 1140:
1.126 schwarze 1141: if (mlink->next == NULL)
1.75 schwarze 1142: break;
1143: mlink = mlink->next;
1144: }
1145:
1146: /* Move all links to the target. */
1147:
1148: mlink->next = mlink_dest->next;
1149: mlink_dest->next = mpage->mlinks;
1150: mpage->mlinks = NULL;
1151: }
1.81 schwarze 1152: goto nextpage;
1.145 schwarze 1153: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.159 schwarze 1154: mdoc_validate(man);
1.47 schwarze 1155: mpage->form = FORM_SRC;
1.146 schwarze 1156: mpage->sec = man->meta.msec;
1.112 schwarze 1157: mpage->sec = mandoc_strdup(
1.126 schwarze 1158: mpage->sec == NULL ? "" : mpage->sec);
1.146 schwarze 1159: mpage->arch = man->meta.arch;
1.47 schwarze 1160: mpage->arch = mandoc_strdup(
1.126 schwarze 1161: mpage->arch == NULL ? "" : mpage->arch);
1.146 schwarze 1162: mpage->title = mandoc_strdup(man->meta.title);
1.145 schwarze 1163: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.160 schwarze 1164: man_validate(man);
1.189 schwarze 1165: if (*man->meta.msec != '\0' ||
1.194 schwarze 1166: *man->meta.title != '\0') {
1.189 schwarze 1167: mpage->form = FORM_SRC;
1168: mpage->sec = mandoc_strdup(man->meta.msec);
1169: mpage->arch = mandoc_strdup(mlink->arch);
1170: mpage->title = mandoc_strdup(man->meta.title);
1171: } else
1172: man = NULL;
1173: }
1174:
1175: assert(mpage->desc == NULL);
1176: if (man == NULL) {
1.47 schwarze 1177: mpage->form = FORM_CAT;
1.136 schwarze 1178: mpage->sec = mandoc_strdup(mlink->dsec);
1179: mpage->arch = mandoc_strdup(mlink->arch);
1180: mpage->title = mandoc_strdup(mlink->name);
1.189 schwarze 1181: parse_cat(mpage, fd);
1182: } else if (man->macroset == MACROSET_MDOC)
1.146 schwarze 1183: parse_mdoc(mpage, &man->meta, man->first);
1.189 schwarze 1184: else
1.146 schwarze 1185: parse_man(mpage, &man->meta, man->first);
1.191 schwarze 1186: if (mpage->desc == NULL) {
1187: mpage->desc = mandoc_strdup(mlink->name);
1188: if (warnings)
1189: say(mlink->file, "No one-line description, "
1190: "using filename \"%s\"", mlink->name);
1191: }
1.88 schwarze 1192:
1.187 schwarze 1193: for (mlink = mpage->mlinks;
1194: mlink != NULL;
1195: mlink = mlink->next) {
1196: putkey(mpage, mlink->name, NAME_FILE);
1197: if (warnings && !use_all)
1.88 schwarze 1198: mlink_check(mpage, mlink);
1.187 schwarze 1199: }
1.6 schwarze 1200:
1.173 schwarze 1201: dbadd(dba, mpage);
1.81 schwarze 1202:
1203: nextpage:
1.47 schwarze 1204: ohash_delete(&strings);
1.90 schwarze 1205: ohash_delete(&names);
1.47 schwarze 1206: }
1207: }
1.6 schwarze 1208:
1.47 schwarze 1209: static void
1.81 schwarze 1210: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1211: {
1212: FILE *stream;
1.190 schwarze 1213: struct mlink *mlink;
1214: char *line, *p, *title, *sec;
1.162 schwarze 1215: size_t linesz, plen, titlesz;
1216: ssize_t len;
1217: int offs;
1.1 schwarze 1218:
1.190 schwarze 1219: mlink = mpage->mlinks;
1220: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1221: if (stream == NULL) {
1222: if (fd != -1)
1.114 doug 1223: close(fd);
1.47 schwarze 1224: if (warnings)
1.190 schwarze 1225: say(mlink->file, "&fopen");
1.47 schwarze 1226: return;
1227: }
1.1 schwarze 1228:
1.162 schwarze 1229: line = NULL;
1230: linesz = 0;
1231:
1.190 schwarze 1232: /* Parse the section number from the header line. */
1233:
1234: while (getline(&line, &linesz, stream) != -1) {
1235: if (*line == '\n')
1236: continue;
1237: if ((sec = strchr(line, '(')) == NULL)
1238: break;
1239: if ((p = strchr(++sec, ')')) == NULL)
1240: break;
1241: free(mpage->sec);
1242: mpage->sec = mandoc_strndup(sec, p - sec);
1243: if (warnings && *mlink->dsec != '\0' &&
1244: strcasecmp(mpage->sec, mlink->dsec))
1245: say(mlink->file,
1246: "Section \"%s\" manual in %s directory",
1247: mpage->sec, mlink->dsec);
1248: break;
1249: }
1250:
1.47 schwarze 1251: /* Skip to first blank line. */
1.1 schwarze 1252:
1.190 schwarze 1253: while (line == NULL || *line != '\n')
1254: if (getline(&line, &linesz, stream) == -1)
1.47 schwarze 1255: break;
1.1 schwarze 1256:
1.47 schwarze 1257: /*
1258: * Assume the first line that is not indented
1259: * is the first section header. Skip to it.
1260: */
1.1 schwarze 1261:
1.162 schwarze 1262: while (getline(&line, &linesz, stream) != -1)
1263: if (*line != '\n' && *line != ' ')
1.47 schwarze 1264: break;
1.98 schwarze 1265:
1.47 schwarze 1266: /*
1267: * Read up until the next section into a buffer.
1268: * Strip the leading and trailing newline from each read line,
1269: * appending a trailing space.
1270: * Ignore empty (whitespace-only) lines.
1271: */
1.28 schwarze 1272:
1.47 schwarze 1273: titlesz = 0;
1274: title = NULL;
1.38 schwarze 1275:
1.162 schwarze 1276: while ((len = getline(&line, &linesz, stream)) != -1) {
1277: if (*line != ' ')
1.47 schwarze 1278: break;
1.162 schwarze 1279: offs = 0;
1280: while (isspace((unsigned char)line[offs]))
1281: offs++;
1282: if (line[offs] == '\0')
1.47 schwarze 1283: continue;
1.162 schwarze 1284: title = mandoc_realloc(title, titlesz + len - offs);
1285: memcpy(title + titlesz, line + offs, len - offs);
1286: titlesz += len - offs;
1.47 schwarze 1287: title[titlesz - 1] = ' ';
1288: }
1.162 schwarze 1289: free(line);
1.28 schwarze 1290:
1.47 schwarze 1291: /*
1292: * If no page content can be found, or the input line
1293: * is already the next section header, or there is no
1294: * trailing newline, reuse the page title as the page
1295: * description.
1296: */
1.1 schwarze 1297:
1.47 schwarze 1298: if (NULL == title || '\0' == *title) {
1299: if (warnings)
1.190 schwarze 1300: say(mlink->file, "Cannot find NAME section");
1.47 schwarze 1301: fclose(stream);
1302: free(title);
1303: return;
1304: }
1.24 schwarze 1305:
1.162 schwarze 1306: title[titlesz - 1] = '\0';
1.24 schwarze 1307:
1.47 schwarze 1308: /*
1309: * Skip to the first dash.
1310: * Use the remaining line as the description (no more than 70
1311: * bytes).
1312: */
1.28 schwarze 1313:
1.47 schwarze 1314: if (NULL != (p = strstr(title, "- "))) {
1315: for (p += 2; ' ' == *p || '\b' == *p; p++)
1316: /* Skip to next word. */ ;
1317: } else {
1318: if (warnings)
1.191 schwarze 1319: say(mlink->file, "No dash in title line, "
1320: "reusing \"%s\" as one-line description", title);
1.47 schwarze 1321: p = title;
1322: }
1.1 schwarze 1323:
1.47 schwarze 1324: plen = strlen(p);
1.1 schwarze 1325:
1.47 schwarze 1326: /* Strip backspace-encoding from line. */
1.1 schwarze 1327:
1.47 schwarze 1328: while (NULL != (line = memchr(p, '\b', plen))) {
1329: len = line - p;
1330: if (0 == len) {
1331: memmove(line, line + 1, plen--);
1332: continue;
1.98 schwarze 1333: }
1.47 schwarze 1334: memmove(line - 1, line + 1, plen - len);
1335: plen -= 2;
1336: }
1.1 schwarze 1337:
1.47 schwarze 1338: mpage->desc = mandoc_strdup(p);
1339: fclose(stream);
1340: free(title);
1341: }
1.16 schwarze 1342:
1.47 schwarze 1343: /*
1344: * Put a type/word pair into the word database for this particular file.
1345: */
1346: static void
1.69 schwarze 1347: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1348: {
1349: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1350: }
1351:
1352: /*
1.47 schwarze 1353: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1354: */
1355: static void
1.47 schwarze 1356: putmdockey(const struct mpage *mpage,
1.184 schwarze 1357: const struct roff_node *n, uint64_t m, int taboo)
1.2 schwarze 1358: {
1.16 schwarze 1359:
1.47 schwarze 1360: for ( ; NULL != n; n = n->next) {
1.184 schwarze 1361: if (n->flags & taboo)
1362: continue;
1.47 schwarze 1363: if (NULL != n->child)
1.184 schwarze 1364: putmdockey(mpage, n->child, m, taboo);
1.141 schwarze 1365: if (n->type == ROFFT_TEXT)
1.47 schwarze 1366: putkey(mpage, n->string, m);
1367: }
1368: }
1.16 schwarze 1369:
1.47 schwarze 1370: static void
1.143 schwarze 1371: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1372: const struct roff_node *n)
1.47 schwarze 1373: {
1.142 schwarze 1374: const struct roff_node *head, *body;
1.78 schwarze 1375: char *start, *title;
1.47 schwarze 1376: char byte;
1.78 schwarze 1377: size_t sz;
1.16 schwarze 1378:
1.168 schwarze 1379: if (n == NULL)
1.47 schwarze 1380: return;
1.16 schwarze 1381:
1.47 schwarze 1382: /*
1383: * We're only searching for one thing: the first text child in
1384: * the BODY of a NAME section. Since we don't keep track of
1385: * sections in -man, run some hoops to find out whether we're in
1386: * the correct section or not.
1387: */
1.16 schwarze 1388:
1.141 schwarze 1389: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1390: body = n;
1.168 schwarze 1391: if ((head = body->parent->head) != NULL &&
1392: (head = head->child) != NULL &&
1393: head->next == NULL &&
1.141 schwarze 1394: head->type == ROFFT_TEXT &&
1.168 schwarze 1395: strcmp(head->string, "NAME") == 0 &&
1396: body->child != NULL) {
1.2 schwarze 1397:
1.47 schwarze 1398: /*
1399: * Suck the entire NAME section into memory.
1400: * Yes, we might run away.
1401: * But too many manuals have big, spread-out
1402: * NAME sections over many lines.
1403: */
1.2 schwarze 1404:
1.78 schwarze 1405: title = NULL;
1.147 schwarze 1406: deroff(&title, body);
1.47 schwarze 1407: if (NULL == title)
1408: return;
1.16 schwarze 1409:
1.98 schwarze 1410: /*
1.47 schwarze 1411: * Go through a special heuristic dance here.
1412: * Conventionally, one or more manual names are
1413: * comma-specified prior to a whitespace, then a
1414: * dash, then a description. Try to puzzle out
1415: * the name parts here.
1416: */
1.16 schwarze 1417:
1.78 schwarze 1418: start = title;
1.47 schwarze 1419: for ( ;; ) {
1420: sz = strcspn(start, " ,");
1421: if ('\0' == start[sz])
1422: break;
1.1 schwarze 1423:
1.47 schwarze 1424: byte = start[sz];
1425: start[sz] = '\0';
1.67 schwarze 1426:
1427: /*
1428: * Assume a stray trailing comma in the
1429: * name list if a name begins with a dash.
1430: */
1431:
1432: if ('-' == start[0] ||
1433: ('\\' == start[0] && '-' == start[1]))
1434: break;
1.1 schwarze 1435:
1.90 schwarze 1436: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1437: if ( ! (mpage->name_head_done ||
1438: strcasecmp(start, meta->title))) {
1439: putkey(mpage, start, NAME_HEAD);
1440: mpage->name_head_done = 1;
1441: }
1.1 schwarze 1442:
1.47 schwarze 1443: if (' ' == byte) {
1444: start += sz + 1;
1445: break;
1446: }
1.1 schwarze 1447:
1.47 schwarze 1448: assert(',' == byte);
1449: start += sz + 1;
1450: while (' ' == *start)
1451: start++;
1452: }
1.1 schwarze 1453:
1.78 schwarze 1454: if (start == title) {
1.90 schwarze 1455: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1456: if ( ! (mpage->name_head_done ||
1457: strcasecmp(start, meta->title))) {
1458: putkey(mpage, start, NAME_HEAD);
1459: mpage->name_head_done = 1;
1460: }
1.47 schwarze 1461: free(title);
1462: return;
1463: }
1.1 schwarze 1464:
1.47 schwarze 1465: while (isspace((unsigned char)*start))
1466: start++;
1.1 schwarze 1467:
1.47 schwarze 1468: if (0 == strncmp(start, "-", 1))
1469: start += 1;
1470: else if (0 == strncmp(start, "\\-\\-", 4))
1471: start += 4;
1472: else if (0 == strncmp(start, "\\-", 2))
1473: start += 2;
1474: else if (0 == strncmp(start, "\\(en", 4))
1475: start += 4;
1476: else if (0 == strncmp(start, "\\(em", 4))
1477: start += 4;
1.1 schwarze 1478:
1.47 schwarze 1479: while (' ' == *start)
1480: start++;
1.1 schwarze 1481:
1.47 schwarze 1482: mpage->desc = mandoc_strdup(start);
1483: free(title);
1484: return;
1485: }
1486: }
1.1 schwarze 1487:
1.47 schwarze 1488: for (n = n->child; n; n = n->next) {
1489: if (NULL != mpage->desc)
1490: break;
1.129 schwarze 1491: parse_man(mpage, meta, n);
1.1 schwarze 1492: }
1493: }
1494:
1495: static void
1.143 schwarze 1496: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1497: const struct roff_node *n)
1.1 schwarze 1498: {
1499:
1.195 schwarze 1500: for (n = n->child; n != NULL; n = n->next) {
1.196 schwarze 1501: if (n->tok == TOKEN_NONE ||
1502: n->tok < ROFF_MAX ||
1503: n->flags & mdocs[n->tok].taboo)
1.184 schwarze 1504: continue;
1.196 schwarze 1505: assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1.47 schwarze 1506: switch (n->type) {
1.141 schwarze 1507: case ROFFT_ELEM:
1508: case ROFFT_BLOCK:
1509: case ROFFT_HEAD:
1510: case ROFFT_BODY:
1511: case ROFFT_TAIL:
1.195 schwarze 1512: if (mdocs[n->tok].fp != NULL &&
1513: (*mdocs[n->tok].fp)(mpage, meta, n) == 0)
1514: break;
1.47 schwarze 1515: if (mdocs[n->tok].mask)
1516: putmdockey(mpage, n->child,
1.184 schwarze 1517: mdocs[n->tok].mask, mdocs[n->tok].taboo);
1.47 schwarze 1518: break;
1519: default:
1520: continue;
1521: }
1522: if (NULL != n->child)
1.127 schwarze 1523: parse_mdoc(mpage, meta, n);
1.1 schwarze 1524: }
1525: }
1526:
1.19 schwarze 1527: static int
1.143 schwarze 1528: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1529: const struct roff_node *n)
1.1 schwarze 1530: {
1.131 schwarze 1531: char *start, *end;
1.1 schwarze 1532: size_t sz;
1.19 schwarze 1533:
1.47 schwarze 1534: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1535: NULL == (n = n->child) ||
1.141 schwarze 1536: n->type != ROFFT_TEXT)
1.150 schwarze 1537: return 0;
1.1 schwarze 1538:
1539: /*
1540: * Only consider those `Fd' macro fields that begin with an
1541: * "inclusion" token (versus, e.g., #define).
1542: */
1.47 schwarze 1543:
1.1 schwarze 1544: if (strcmp("#include", n->string))
1.150 schwarze 1545: return 0;
1.1 schwarze 1546:
1.141 schwarze 1547: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.150 schwarze 1548: return 0;
1.1 schwarze 1549:
1550: /*
1551: * Strip away the enclosing angle brackets and make sure we're
1552: * not zero-length.
1553: */
1554:
1555: start = n->string;
1556: if ('<' == *start || '"' == *start)
1557: start++;
1558:
1559: if (0 == (sz = strlen(start)))
1.150 schwarze 1560: return 0;
1.1 schwarze 1561:
1562: end = &start[(int)sz - 1];
1563: if ('>' == *end || '"' == *end)
1564: end--;
1565:
1.47 schwarze 1566: if (end > start)
1567: putkeys(mpage, start, end - start + 1, TYPE_In);
1.150 schwarze 1568: return 0;
1.1 schwarze 1569: }
1570:
1.133 schwarze 1571: static void
1.142 schwarze 1572: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 schwarze 1573: {
1.69 schwarze 1574: char *cp;
1.133 schwarze 1575: size_t sz;
1.1 schwarze 1576:
1.141 schwarze 1577: if (n->type != ROFFT_TEXT)
1.133 schwarze 1578: return;
1.19 schwarze 1579:
1.133 schwarze 1580: /* Skip function pointer punctuation. */
1.1 schwarze 1581:
1.133 schwarze 1582: cp = n->string;
1583: while (*cp == '(' || *cp == '*')
1.1 schwarze 1584: cp++;
1.133 schwarze 1585: sz = strcspn(cp, "()");
1.1 schwarze 1586:
1.133 schwarze 1587: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1588: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1589: putkeys(mpage, cp, sz, NAME_SYN);
1590: }
1.19 schwarze 1591:
1.133 schwarze 1592: static int
1.143 schwarze 1593: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1594: const struct roff_node *n)
1.133 schwarze 1595: {
1596:
1597: if (n->child == NULL)
1.150 schwarze 1598: return 0;
1.133 schwarze 1599:
1600: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1601:
1.133 schwarze 1602: for (n = n->child->next; n != NULL; n = n->next)
1.141 schwarze 1603: if (n->type == ROFFT_TEXT)
1.47 schwarze 1604: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1605:
1.150 schwarze 1606: return 0;
1.128 schwarze 1607: }
1608:
1609: static int
1.143 schwarze 1610: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1611: const struct roff_node *n)
1.128 schwarze 1612: {
1.132 schwarze 1613:
1.141 schwarze 1614: if (n->type != ROFFT_HEAD)
1.150 schwarze 1615: return 1;
1.128 schwarze 1616:
1.133 schwarze 1617: if (n->child != NULL)
1618: parse_mdoc_fname(mpage, n->child);
1619:
1.150 schwarze 1620: return 0;
1.1 schwarze 1621: }
1622:
1.19 schwarze 1623: static int
1.164 schwarze 1624: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1625: const struct roff_node *n)
1626: {
1627: char *cp;
1628:
1629: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1630: return 0;
1631:
1.168 schwarze 1632: if (n->child != NULL &&
1633: n->child->next == NULL &&
1634: n->child->type == ROFFT_TEXT)
1.164 schwarze 1635: return 1;
1636:
1637: cp = NULL;
1638: deroff(&cp, n);
1639: if (cp != NULL) {
1640: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1641: n->type == ROFFT_BODY ? TYPE_Va : 0));
1642: free(cp);
1643: }
1644:
1645: return 0;
1646: }
1647:
1648: static int
1.143 schwarze 1649: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1650: const struct roff_node *n)
1.1 schwarze 1651: {
1.47 schwarze 1652: char *cp;
1.1 schwarze 1653:
1654: if (NULL == (n = n->child))
1.150 schwarze 1655: return 0;
1.1 schwarze 1656:
1.47 schwarze 1657: if (NULL == n->next) {
1658: putkey(mpage, n->string, TYPE_Xr);
1.150 schwarze 1659: return 0;
1.47 schwarze 1660: }
1.1 schwarze 1661:
1.77 schwarze 1662: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1663: putkey(mpage, cp, TYPE_Xr);
1664: free(cp);
1.150 schwarze 1665: return 0;
1.1 schwarze 1666: }
1667:
1.19 schwarze 1668: static int
1.143 schwarze 1669: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1670: const struct roff_node *n)
1.1 schwarze 1671: {
1672:
1.141 schwarze 1673: if (n->type == ROFFT_BODY)
1.147 schwarze 1674: deroff(&mpage->desc, n);
1.150 schwarze 1675: return 0;
1.1 schwarze 1676: }
1677:
1.19 schwarze 1678: static int
1.143 schwarze 1679: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1680: const struct roff_node *n)
1.1 schwarze 1681: {
1682:
1.86 schwarze 1683: if (SEC_NAME == n->sec)
1.184 schwarze 1684: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.141 schwarze 1685: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.127 schwarze 1686: if (n->child == NULL)
1687: putkey(mpage, meta->name, NAME_SYN);
1688: else
1.184 schwarze 1689: putmdockey(mpage, n->child, NAME_SYN, 0);
1.129 schwarze 1690: }
1691: if ( ! (mpage->name_head_done ||
1692: n->child == NULL || n->child->string == NULL ||
1693: strcasecmp(n->child->string, meta->title))) {
1.175 schwarze 1694: putkey(mpage, n->child->string, NAME_HEAD);
1.129 schwarze 1695: mpage->name_head_done = 1;
1.127 schwarze 1696: }
1.150 schwarze 1697: return 0;
1.1 schwarze 1698: }
1699:
1.19 schwarze 1700: static int
1.143 schwarze 1701: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1702: const struct roff_node *n)
1.1 schwarze 1703: {
1704:
1.150 schwarze 1705: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 schwarze 1706: }
1707:
1.47 schwarze 1708: static int
1.143 schwarze 1709: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1710: const struct roff_node *n)
1.1 schwarze 1711: {
1712:
1.150 schwarze 1713: return n->type == ROFFT_HEAD;
1.1 schwarze 1714: }
1715:
1.47 schwarze 1716: /*
1717: * Add a string to the hash table for the current manual.
1718: * Each string has a bitmask telling which macros it belongs to.
1719: * When we finish the manual, we'll dump the table.
1720: */
1.1 schwarze 1721: static void
1.131 schwarze 1722: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1723: {
1.90 schwarze 1724: struct ohash *htab;
1.47 schwarze 1725: struct str *s;
1.68 schwarze 1726: const char *end;
1.47 schwarze 1727: unsigned int slot;
1.131 schwarze 1728: int i, mustfree;
1.1 schwarze 1729:
1.47 schwarze 1730: if (0 == sz)
1731: return;
1.68 schwarze 1732:
1.131 schwarze 1733: mustfree = render_string(&cp, &sz);
1734:
1.90 schwarze 1735: if (TYPE_Nm & v) {
1736: htab = &names;
1737: v &= name_mask;
1.124 schwarze 1738: if (v & NAME_FIRST)
1739: name_mask &= ~NAME_FIRST;
1.90 schwarze 1740: if (debug > 1)
1741: say(mpage->mlinks->file,
1.183 schwarze 1742: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1743: (unsigned long long)v);
1.90 schwarze 1744: } else {
1745: htab = &strings;
1746: if (debug > 1)
1.173 schwarze 1747: for (i = 0; i < KEY_MAX; i++)
1.118 schwarze 1748: if ((uint64_t)1 << i & v)
1.90 schwarze 1749: say(mpage->mlinks->file,
1750: "Adding key %s=%*s",
1.172 schwarze 1751: mansearch_keynames[i], (int)sz, cp);
1.68 schwarze 1752: }
1.47 schwarze 1753:
1754: end = cp + sz;
1.90 schwarze 1755: slot = ohash_qlookupi(htab, cp, &end);
1756: s = ohash_find(htab, slot);
1.1 schwarze 1757:
1.47 schwarze 1758: if (NULL != s && mpage == s->mpage) {
1759: s->mask |= v;
1.1 schwarze 1760: return;
1.47 schwarze 1761: } else if (NULL == s) {
1.101 schwarze 1762: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1763: memcpy(s->key, cp, sz);
1.90 schwarze 1764: ohash_insert(htab, slot, s);
1.47 schwarze 1765: }
1766: s->mpage = mpage;
1767: s->mask = v;
1.131 schwarze 1768:
1769: if (mustfree)
1770: free(cp);
1.1 schwarze 1771: }
1772:
1773: /*
1.47 schwarze 1774: * Take a Unicode codepoint and produce its UTF-8 encoding.
1775: * This isn't the best way to do this, but it works.
1776: * The magic numbers are from the UTF-8 packaging.
1777: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1778: */
1.47 schwarze 1779: static size_t
1780: utf8(unsigned int cp, char out[7])
1.1 schwarze 1781: {
1.47 schwarze 1782: size_t rc;
1.1 schwarze 1783:
1.47 schwarze 1784: rc = 0;
1785: if (cp <= 0x0000007F) {
1786: rc = 1;
1787: out[0] = (char)cp;
1788: } else if (cp <= 0x000007FF) {
1789: rc = 2;
1790: out[0] = (cp >> 6 & 31) | 192;
1791: out[1] = (cp & 63) | 128;
1792: } else if (cp <= 0x0000FFFF) {
1793: rc = 3;
1794: out[0] = (cp >> 12 & 15) | 224;
1795: out[1] = (cp >> 6 & 63) | 128;
1796: out[2] = (cp & 63) | 128;
1797: } else if (cp <= 0x001FFFFF) {
1798: rc = 4;
1799: out[0] = (cp >> 18 & 7) | 240;
1800: out[1] = (cp >> 12 & 63) | 128;
1801: out[2] = (cp >> 6 & 63) | 128;
1802: out[3] = (cp & 63) | 128;
1803: } else if (cp <= 0x03FFFFFF) {
1804: rc = 5;
1805: out[0] = (cp >> 24 & 3) | 248;
1806: out[1] = (cp >> 18 & 63) | 128;
1807: out[2] = (cp >> 12 & 63) | 128;
1808: out[3] = (cp >> 6 & 63) | 128;
1809: out[4] = (cp & 63) | 128;
1810: } else if (cp <= 0x7FFFFFFF) {
1811: rc = 6;
1812: out[0] = (cp >> 30 & 1) | 252;
1813: out[1] = (cp >> 24 & 63) | 128;
1814: out[2] = (cp >> 18 & 63) | 128;
1815: out[3] = (cp >> 12 & 63) | 128;
1816: out[4] = (cp >> 6 & 63) | 128;
1817: out[5] = (cp & 63) | 128;
1818: } else
1.150 schwarze 1819: return 0;
1.19 schwarze 1820:
1.47 schwarze 1821: out[rc] = '\0';
1.150 schwarze 1822: return rc;
1.1 schwarze 1823: }
1824:
1.47 schwarze 1825: /*
1.131 schwarze 1826: * If the string contains escape sequences,
1827: * replace it with an allocated rendering and return 1,
1828: * such that the caller can free it after use.
1829: * Otherwise, do nothing and return 0.
1.47 schwarze 1830: */
1.131 schwarze 1831: static int
1832: render_string(char **public, size_t *psz)
1.1 schwarze 1833: {
1.131 schwarze 1834: const char *src, *scp, *addcp, *seq;
1835: char *dst;
1836: size_t ssz, dsz, addsz;
1.71 schwarze 1837: char utfbuf[7], res[6];
1.131 schwarze 1838: int seqlen, unicode;
1.47 schwarze 1839:
1840: res[0] = '\\';
1841: res[1] = '\t';
1842: res[2] = ASCII_NBRSP;
1843: res[3] = ASCII_HYPH;
1.71 schwarze 1844: res[4] = ASCII_BREAK;
1845: res[5] = '\0';
1.1 schwarze 1846:
1.131 schwarze 1847: src = scp = *public;
1848: ssz = *psz;
1849: dst = NULL;
1850: dsz = 0;
1851:
1852: while (scp < src + *psz) {
1853:
1854: /* Leave normal characters unchanged. */
1855:
1856: if (strchr(res, *scp) == NULL) {
1857: if (dst != NULL)
1858: dst[dsz++] = *scp;
1859: scp++;
1860: continue;
1861: }
1.1 schwarze 1862:
1.47 schwarze 1863: /*
1.131 schwarze 1864: * Found something that requires replacing,
1865: * make sure we have a destination buffer.
1.47 schwarze 1866: */
1.131 schwarze 1867:
1868: if (dst == NULL) {
1869: dst = mandoc_malloc(ssz + 1);
1870: dsz = scp - src;
1871: memcpy(dst, src, dsz);
1.47 schwarze 1872: }
1.39 schwarze 1873:
1.131 schwarze 1874: /* Handle single-char special characters. */
1875:
1876: switch (*scp) {
1877: case '\\':
1878: break;
1.98 schwarze 1879: case '\t':
1880: case ASCII_NBRSP:
1.131 schwarze 1881: dst[dsz++] = ' ';
1882: scp++;
1883: continue;
1884: case ASCII_HYPH:
1885: dst[dsz++] = '-';
1.71 schwarze 1886: /* FALLTHROUGH */
1.98 schwarze 1887: case ASCII_BREAK:
1.131 schwarze 1888: scp++;
1.47 schwarze 1889: continue;
1.71 schwarze 1890: default:
1.131 schwarze 1891: abort();
1.71 schwarze 1892: }
1.39 schwarze 1893:
1.47 schwarze 1894: /*
1.131 schwarze 1895: * Found an escape sequence.
1896: * Read past the slash, then parse it.
1897: * Ignore everything except characters.
1.47 schwarze 1898: */
1.52 schwarze 1899:
1.131 schwarze 1900: scp++;
1901: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1902: continue;
1.39 schwarze 1903:
1.47 schwarze 1904: /*
1.52 schwarze 1905: * Render the special character
1906: * as either UTF-8 or ASCII.
1.47 schwarze 1907: */
1.52 schwarze 1908:
1909: if (write_utf8) {
1.157 schwarze 1910: unicode = mchars_spec2cp(seq, seqlen);
1.131 schwarze 1911: if (unicode <= 0)
1.52 schwarze 1912: continue;
1.131 schwarze 1913: addsz = utf8(unicode, utfbuf);
1914: if (addsz == 0)
1.52 schwarze 1915: continue;
1.131 schwarze 1916: addcp = utfbuf;
1.52 schwarze 1917: } else {
1.157 schwarze 1918: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.131 schwarze 1919: if (addcp == NULL)
1.52 schwarze 1920: continue;
1.131 schwarze 1921: if (*addcp == ASCII_NBRSP) {
1922: addcp = " ";
1923: addsz = 1;
1.52 schwarze 1924: }
1925: }
1.1 schwarze 1926:
1.47 schwarze 1927: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1928:
1.131 schwarze 1929: ssz += addsz;
1930: dst = mandoc_realloc(dst, ssz + 1);
1931: memcpy(dst + dsz, addcp, addsz);
1932: dsz += addsz;
1933: }
1934: if (dst != NULL) {
1935: *public = dst;
1936: *psz = dsz;
1937: }
1938:
1939: /* Trim trailing whitespace and NUL-terminate. */
1940:
1941: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1942: --*psz;
1943: if (dst != NULL) {
1944: (*public)[*psz] = '\0';
1.150 schwarze 1945: return 1;
1.131 schwarze 1946: } else
1.150 schwarze 1947: return 0;
1.1 schwarze 1948: }
1949:
1.75 schwarze 1950: static void
1951: dbadd_mlink(const struct mlink *mlink)
1952: {
1.173 schwarze 1953: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
1954: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1955: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1956: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
1957: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.75 schwarze 1958: }
1959:
1.11 schwarze 1960: /*
1.47 schwarze 1961: * Flush the current page's terms (and their bits) into the database.
1.53 schwarze 1962: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1963: */
1964: static void
1.173 schwarze 1965: dbadd(struct dba *dba, struct mpage *mpage)
1.11 schwarze 1966: {
1.47 schwarze 1967: struct mlink *mlink;
1968: struct str *key;
1.131 schwarze 1969: char *cp;
1.173 schwarze 1970: uint64_t mask;
1.47 schwarze 1971: size_t i;
1972: unsigned int slot;
1.131 schwarze 1973: int mustfree;
1.47 schwarze 1974:
1.85 schwarze 1975: mlink = mpage->mlinks;
1.11 schwarze 1976:
1.85 schwarze 1977: if (nodb) {
1.104 schwarze 1978: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 1979: key = ohash_next(&names, &slot))
1.104 schwarze 1980: free(key);
1981: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 1982: key = ohash_next(&strings, &slot))
1.104 schwarze 1983: free(key);
1.102 schwarze 1984: if (0 == debug)
1985: return;
1.85 schwarze 1986: while (NULL != mlink) {
1987: fputs(mlink->name, stdout);
1988: if (NULL == mlink->next ||
1989: strcmp(mlink->dsec, mlink->next->dsec) ||
1990: strcmp(mlink->fsec, mlink->next->fsec) ||
1991: strcmp(mlink->arch, mlink->next->arch)) {
1992: putchar('(');
1993: if ('\0' == *mlink->dsec)
1994: fputs(mlink->fsec, stdout);
1995: else
1996: fputs(mlink->dsec, stdout);
1997: if ('\0' != *mlink->arch)
1998: printf("/%s", mlink->arch);
1999: putchar(')');
2000: }
2001: mlink = mlink->next;
2002: if (NULL != mlink)
2003: fputs(", ", stdout);
2004: }
1.89 schwarze 2005: printf(" - %s\n", mpage->desc);
1.11 schwarze 2006: return;
1.85 schwarze 2007: }
2008:
2009: if (debug)
2010: say(mlink->file, "Adding to database");
1.47 schwarze 2011:
1.131 schwarze 2012: cp = mpage->desc;
2013: i = strlen(cp);
2014: mustfree = render_string(&cp, &i);
1.176 schwarze 2015: mpage->dba = dba_page_new(dba->pages,
1.173 schwarze 2016: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2017: cp, mlink->file, mpage->form);
1.131 schwarze 2018: if (mustfree)
2019: free(cp);
1.176 schwarze 2020: dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
1.47 schwarze 2021:
1.173 schwarze 2022: while (mlink != NULL) {
1.75 schwarze 2023: dbadd_mlink(mlink);
1.85 schwarze 2024: mlink = mlink->next;
2025: }
1.47 schwarze 2026:
1.90 schwarze 2027: for (key = ohash_first(&names, &slot); NULL != key;
2028: key = ohash_next(&names, &slot)) {
2029: assert(key->mpage == mpage);
1.173 schwarze 2030: dba_page_alias(mpage->dba, key->key, key->mask);
1.90 schwarze 2031: free(key);
2032: }
1.47 schwarze 2033: for (key = ohash_first(&strings, &slot); NULL != key;
2034: key = ohash_next(&strings, &slot)) {
2035: assert(key->mpage == mpage);
1.173 schwarze 2036: i = 0;
2037: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2038: if (key->mask & mask)
2039: dba_macro_add(dba->macros, i,
2040: key->key, mpage->dba);
2041: i++;
2042: }
1.47 schwarze 2043: free(key);
1.33 schwarze 2044: }
1.47 schwarze 2045: }
1.41 deraadt 2046:
1.47 schwarze 2047: static void
1.173 schwarze 2048: dbprune(struct dba *dba)
1.47 schwarze 2049: {
1.173 schwarze 2050: struct dba_array *page, *files;
2051: char *file;
1.11 schwarze 2052:
1.173 schwarze 2053: dba_array_FOREACH(dba->pages, page) {
2054: files = dba_array_get(page, DBP_FILE);
2055: dba_array_FOREACH(files, file) {
2056: if (*file < ' ')
2057: file++;
2058: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2059: file)) != NULL) {
2060: if (debug)
2061: say(file, "Deleting from database");
2062: dba_array_del(dba->pages);
2063: break;
2064: }
1.63 schwarze 2065: }
1.11 schwarze 2066: }
1.47 schwarze 2067: }
1.22 schwarze 2068:
1.47 schwarze 2069: /*
1.173 schwarze 2070: * Write the database from memory to disk.
1.47 schwarze 2071: */
2072: static void
1.173 schwarze 2073: dbwrite(struct dba *dba)
1.47 schwarze 2074: {
1.204 schwarze 2075: struct stat sb1, sb2;
2076: char tfn[33], *cp1, *cp2;
2077: off_t i;
2078: int fd1, fd2;
1.11 schwarze 2079:
1.199 schwarze 2080: /*
2081: * Do not write empty databases, and delete existing ones
2082: * when makewhatis -u causes them to become empty.
2083: */
2084:
2085: dba_array_start(dba->pages);
2086: if (dba_array_next(dba->pages) == NULL) {
1.201 schwarze 2087: if (unlink(MANDOC_DB) == -1 && errno != ENOENT)
1.199 schwarze 2088: say(MANDOC_DB, "&unlink");
2089: return;
2090: }
2091:
2092: /*
2093: * Build the database in a temporary file,
2094: * then atomically move it into place.
2095: */
2096:
1.173 schwarze 2097: if (dba_write(MANDOC_DB "~", dba) != -1) {
2098: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2099: exitcode = (int)MANDOCLEVEL_SYSERR;
2100: say(MANDOC_DB, "&rename");
2101: unlink(MANDOC_DB "~");
2102: }
1.47 schwarze 2103: return;
1.28 schwarze 2104: }
1.199 schwarze 2105:
2106: /*
2107: * We lack write permission and cannot replace the database
2108: * file, but let's at least check whether the data changed.
2109: */
1.22 schwarze 2110:
1.173 schwarze 2111: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2112: if (mkdtemp(tfn) == NULL) {
2113: exitcode = (int)MANDOCLEVEL_SYSERR;
2114: say("", "&%s", tfn);
1.47 schwarze 2115: return;
1.173 schwarze 2116: }
1.205 ! schwarze 2117: cp1 = cp2 = MAP_FAILED;
1.204 schwarze 2118: fd1 = fd2 = -1;
1.173 schwarze 2119: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2120: if (dba_write(tfn, dba) == -1) {
2121: say(tfn, "&dba_write");
1.204 schwarze 2122: goto err;
1.72 schwarze 2123: }
1.204 schwarze 2124: if ((fd1 = open(MANDOC_DB, O_RDONLY, 0)) == -1) {
2125: say(MANDOC_DB, "&open");
2126: goto err;
2127: }
2128: if ((fd2 = open(tfn, O_RDONLY, 0)) == -1) {
2129: say(tfn, "&open");
2130: goto err;
2131: }
2132: if (fstat(fd1, &sb1) == -1) {
2133: say(MANDOC_DB, "&fstat");
2134: goto err;
2135: }
2136: if (fstat(fd2, &sb2) == -1) {
2137: say(tfn, "&fstat");
2138: goto err;
2139: }
2140: if (sb1.st_size != sb2.st_size)
2141: goto err;
2142: if ((cp1 = mmap(NULL, sb1.st_size, PROT_READ, MAP_PRIVATE,
1.205 ! schwarze 2143: fd1, 0)) == MAP_FAILED) {
1.204 schwarze 2144: say(MANDOC_DB, "&mmap");
2145: goto err;
2146: }
2147: if ((cp2 = mmap(NULL, sb2.st_size, PROT_READ, MAP_PRIVATE,
1.205 ! schwarze 2148: fd2, 0)) == MAP_FAILED) {
1.204 schwarze 2149: say(tfn, "&mmap");
2150: goto err;
2151: }
2152: for (i = 0; i < sb1.st_size; i++)
2153: if (cp1[i] != cp2[i])
2154: goto err;
2155: goto out;
2156:
2157: err:
2158: exitcode = (int)MANDOCLEVEL_SYSERR;
2159: say(MANDOC_DB, "Data changed, but cannot replace database");
1.72 schwarze 2160:
1.173 schwarze 2161: out:
1.205 ! schwarze 2162: if (cp1 != MAP_FAILED)
1.204 schwarze 2163: munmap(cp1, sb1.st_size);
1.205 ! schwarze 2164: if (cp2 != MAP_FAILED)
1.204 schwarze 2165: munmap(cp2, sb2.st_size);
2166: if (fd1 != -1)
2167: close(fd1);
2168: if (fd2 != -1)
2169: close(fd2);
1.203 schwarze 2170: unlink(tfn);
1.173 schwarze 2171: *strrchr(tfn, '/') = '\0';
1.203 schwarze 2172: rmdir(tfn);
1.47 schwarze 2173: }
1.6 schwarze 2174:
1.47 schwarze 2175: static int
1.120 schwarze 2176: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2177: {
2178: static char startdir[PATH_MAX];
1.110 schwarze 2179: static int getcwd_status; /* 1 = ok, 2 = failure */
2180: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2181: char *cp;
1.6 schwarze 2182:
1.47 schwarze 2183: /*
1.110 schwarze 2184: * Remember the original working directory, if possible.
2185: * This will be needed if the second or a later directory
2186: * on the command line is given as a relative path.
2187: * Do not error out if the current directory is not
2188: * searchable: Maybe it won't be needed after all.
2189: */
2190: if (0 == getcwd_status) {
2191: if (NULL == getcwd(startdir, sizeof(startdir))) {
2192: getcwd_status = 2;
2193: (void)strlcpy(startdir, strerror(errno),
2194: sizeof(startdir));
2195: } else
2196: getcwd_status = 1;
2197: }
2198:
2199: /*
2200: * We are leaving the old base directory.
2201: * Do not use it any longer, not even for messages.
2202: */
2203: *basedir = '\0';
2204:
2205: /*
2206: * If and only if the directory was changed earlier and
2207: * the next directory to process is given as a relative path,
2208: * first go back, or bail out if that is impossible.
1.47 schwarze 2209: */
1.110 schwarze 2210: if (chdir_status && '/' != *targetdir) {
2211: if (2 == getcwd_status) {
1.47 schwarze 2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2213: say("", "getcwd: %s", startdir);
1.150 schwarze 2214: return 0;
1.47 schwarze 2215: }
1.110 schwarze 2216: if (-1 == chdir(startdir)) {
1.47 schwarze 2217: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2218: say("", "&chdir %s", startdir);
1.150 schwarze 2219: return 0;
1.2 schwarze 2220: }
2221: }
1.110 schwarze 2222:
2223: /*
2224: * Always resolve basedir to the canonicalized absolute
2225: * pathname and append a trailing slash, such that
2226: * we can reliably check whether files are inside.
2227: */
1.47 schwarze 2228: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2229: if (report_baddir || errno != ENOENT) {
2230: exitcode = (int)MANDOCLEVEL_BADARG;
2231: say("", "&%s: realpath", targetdir);
2232: }
1.150 schwarze 2233: return 0;
1.47 schwarze 2234: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2235: if (report_baddir || errno != ENOENT) {
2236: exitcode = (int)MANDOCLEVEL_BADARG;
2237: say("", "&chdir");
2238: }
1.150 schwarze 2239: return 0;
1.109 schwarze 2240: }
1.110 schwarze 2241: chdir_status = 1;
1.109 schwarze 2242: cp = strchr(basedir, '\0');
2243: if ('/' != cp[-1]) {
2244: if (cp - basedir >= PATH_MAX - 1) {
2245: exitcode = (int)MANDOCLEVEL_SYSERR;
2246: say("", "Filename too long");
1.150 schwarze 2247: return 0;
1.109 schwarze 2248: }
2249: *cp++ = '/';
2250: *cp = '\0';
1.47 schwarze 2251: }
1.150 schwarze 2252: return 1;
1.2 schwarze 2253: }
2254:
2255: static void
1.47 schwarze 2256: say(const char *file, const char *format, ...)
1.2 schwarze 2257: {
1.47 schwarze 2258: va_list ap;
1.80 schwarze 2259: int use_errno;
1.2 schwarze 2260:
1.47 schwarze 2261: if ('\0' != *basedir)
2262: fprintf(stderr, "%s", basedir);
2263: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2264: fputc('/', stderr);
1.47 schwarze 2265: if ('\0' != *file)
2266: fprintf(stderr, "%s", file);
1.31 schwarze 2267:
1.80 schwarze 2268: use_errno = 1;
2269: if (NULL != format) {
2270: switch (*format) {
1.98 schwarze 2271: case '&':
1.80 schwarze 2272: format++;
2273: break;
1.98 schwarze 2274: case '\0':
1.80 schwarze 2275: format = NULL;
2276: break;
2277: default:
2278: use_errno = 0;
2279: break;
2280: }
2281: }
2282: if (NULL != format) {
2283: if ('\0' != *basedir || '\0' != *file)
2284: fputs(": ", stderr);
2285: va_start(ap, format);
2286: vfprintf(stderr, format, ap);
2287: va_end(ap);
2288: }
2289: if (use_errno) {
2290: if ('\0' != *basedir || '\0' != *file || NULL != format)
2291: fputs(": ", stderr);
1.47 schwarze 2292: perror(NULL);
1.80 schwarze 2293: } else
2294: fputc('\n', stderr);
1.1 schwarze 2295: }