Annotation of mandoc/mandocdb.c, Revision 1.250
1.250 ! schwarze 1: /* $Id: mandocdb.c,v 1.249 2017/05/05 15:17:32 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.235 schwarze 4: * Copyright (c) 2011-2017 Ingo Schwarze <schwarze@openbsd.org>
1.231 schwarze 5: * Copyright (c) 2016 Ed Maste <emaste@freebsd.org>
1.1 kristaps 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: *
1.187 schwarze 11: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 12: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 13: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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: */
19: #include "config.h"
20:
1.162 schwarze 21: #include <sys/types.h>
1.50 kristaps 22: #include <sys/stat.h>
1.115 schwarze 23: #include <sys/wait.h>
1.1 kristaps 24:
25: #include <assert.h>
1.43 kristaps 26: #include <ctype.h>
1.208 schwarze 27: #if HAVE_ERR
1.198 schwarze 28: #include <err.h>
1.208 schwarze 29: #endif
1.63 schwarze 30: #include <errno.h>
1.1 kristaps 31: #include <fcntl.h>
1.158 schwarze 32: #if HAVE_FTS
1.50 kristaps 33: #include <fts.h>
1.157 schwarze 34: #else
35: #include "compat_fts.h"
36: #endif
1.58 schwarze 37: #include <limits.h>
1.218 kristaps 38: #if HAVE_SANDBOX_INIT
39: #include <sandbox.h>
40: #endif
1.221 schwarze 41: #include <stdarg.h>
1.50 kristaps 42: #include <stddef.h>
1.59 schwarze 43: #include <stdio.h>
1.1 kristaps 44: #include <stdint.h>
45: #include <stdlib.h>
46: #include <string.h>
1.17 schwarze 47: #include <unistd.h>
1.1 kristaps 48:
1.188 schwarze 49: #include "mandoc_aux.h"
1.202 schwarze 50: #include "mandoc_ohash.h"
1.188 schwarze 51: #include "mandoc.h"
52: #include "roff.h"
1.50 kristaps 53: #include "mdoc.h"
1.1 kristaps 54: #include "man.h"
1.187 schwarze 55: #include "manconf.h"
1.55 kristaps 56: #include "mansearch.h"
1.221 schwarze 57: #include "dba_array.h"
58: #include "dba.h"
1.1 kristaps 59:
1.111 schwarze 60: extern const char *const mansearch_keynames[];
61:
1.50 kristaps 62: enum op {
63: OP_DEFAULT = 0, /* new dbs from dir list or default config */
64: OP_CONFFILE, /* new databases from custom config file */
65: OP_UPDATE, /* delete/add entries in existing database */
66: OP_DELETE, /* delete entries from existing database */
67: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 68: };
69:
1.50 kristaps 70: struct str {
1.78 schwarze 71: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 72: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 73: char key[]; /* rendered text */
1.38 schwarze 74: };
75:
1.79 schwarze 76: struct inodev {
77: ino_t st_ino;
78: dev_t st_dev;
1.50 kristaps 79: };
1.5 kristaps 80:
1.78 schwarze 81: struct mpage {
1.79 schwarze 82: struct inodev inodev; /* used for hashing routine */
1.221 schwarze 83: struct dba_array *dba;
1.82 schwarze 84: char *sec; /* section from file content */
85: char *arch; /* architecture from file content */
86: char *title; /* title from file content */
87: char *desc; /* description from file content */
1.231 schwarze 88: struct mpage *next; /* singly linked list */
1.82 schwarze 89: struct mlink *mlinks; /* singly linked list */
1.174 schwarze 90: int name_head_done;
1.221 schwarze 91: enum form form; /* format from file content */
1.82 schwarze 92: };
93:
94: struct mlink {
1.58 schwarze 95: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 96: char *dsec; /* section from directory */
97: char *arch; /* architecture from directory */
98: char *name; /* name from file name (not empty) */
99: char *fsec; /* section from file name suffix */
1.85 schwarze 100: struct mlink *next; /* singly linked list */
1.118 schwarze 101: struct mpage *mpage; /* parent */
1.124 schwarze 102: int gzip; /* filename has a .gz suffix */
1.221 schwarze 103: enum form dform; /* format from directory */
104: enum form fform; /* format from file name suffix */
1.1 kristaps 105: };
106:
1.190 schwarze 107: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 108: const struct roff_node *);
1.1 kristaps 109:
1.50 kristaps 110: struct mdoc_handler {
111: mdoc_fp fp; /* optional handler */
112: uint64_t mask; /* set unless handler returns 0 */
1.235 schwarze 113: int taboo; /* node flags that must not be set */
1.1 kristaps 114: };
1.219 schwarze 115:
116:
117: int mandocdb(int, char *[]);
1.1 kristaps 118:
1.221 schwarze 119: static void dbadd(struct dba *, struct mpage *);
1.118 schwarze 120: static void dbadd_mlink(const struct mlink *mlink);
1.221 schwarze 121: static void dbprune(struct dba *);
122: static void dbwrite(struct dba *);
1.59 schwarze 123: static void filescan(const char *);
1.232 schwarze 124: #if HAVE_FTS_COMPARE_CONST
125: static int fts_compare(const FTSENT *const *, const FTSENT *const *);
126: #else
1.231 schwarze 127: static int fts_compare(const FTSENT **, const FTSENT **);
1.232 schwarze 128: #endif
1.84 schwarze 129: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 130: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 131: static void mlink_free(struct mlink *);
1.89 schwarze 132: static void mlinks_undupe(struct mpage *);
1.78 schwarze 133: static void mpages_free(void);
1.221 schwarze 134: static void mpages_merge(struct dba *, struct mparse *);
1.124 schwarze 135: static void parse_cat(struct mpage *, int);
1.190 schwarze 136: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 137: const struct roff_node *);
1.190 schwarze 138: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 139: const struct roff_node *);
1.190 schwarze 140: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 141: const struct roff_node *);
1.190 schwarze 142: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 143: const struct roff_node *);
144: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 145: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 146: const struct roff_node *);
1.190 schwarze 147: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 148: const struct roff_node *);
1.190 schwarze 149: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 150: const struct roff_node *);
1.190 schwarze 151: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 152: const struct roff_node *);
1.190 schwarze 153: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 154: const struct roff_node *);
1.211 schwarze 155: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
156: const struct roff_node *);
1.190 schwarze 157: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 158: const struct roff_node *);
1.112 schwarze 159: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 160: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 161: static void putmdockey(const struct mpage *,
1.235 schwarze 162: const struct roff_node *, uint64_t, int);
1.176 schwarze 163: static int render_string(char **, size_t *);
1.220 schwarze 164: static void say(const char *, const char *, ...)
1.243 schwarze 165: __attribute__((__format__ (__printf__, 2, 3)));
1.165 schwarze 166: static int set_basedir(const char *, int);
1.59 schwarze 167: static int treescan(void);
1.50 kristaps 168: static size_t utf8(unsigned int, char [7]);
169:
1.102 schwarze 170: static int nodb; /* no database changes */
1.116 schwarze 171: static int mparse_options; /* abort the parse early */
1.141 schwarze 172: static int use_all; /* use all found files */
173: static int debug; /* print what we're doing */
174: static int warnings; /* warn about crap */
1.95 schwarze 175: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 176: static int exitcode; /* to be returned by main */
1.141 schwarze 177: static enum op op; /* operational mode */
1.59 schwarze 178: static char basedir[PATH_MAX]; /* current base directory */
1.231 schwarze 179: static struct mpage *mpage_head; /* list of distinct manual pages */
1.78 schwarze 180: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 181: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 182: static struct ohash names; /* table of all names */
1.50 kristaps 183: static struct ohash strings; /* table of all strings */
1.133 schwarze 184: static uint64_t name_mask;
1.25 schwarze 185:
1.246 schwarze 186: static const struct mdoc_handler __mdocs[MDOC_MAX - MDOC_Dd] = {
1.235 schwarze 187: { NULL, 0, NODE_NOPRT }, /* Dd */
188: { NULL, 0, NODE_NOPRT }, /* Dt */
189: { NULL, 0, NODE_NOPRT }, /* Os */
190: { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
191: { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
192: { NULL, 0, 0 }, /* Pp */
193: { NULL, 0, 0 }, /* D1 */
194: { NULL, 0, 0 }, /* Dl */
195: { NULL, 0, 0 }, /* Bd */
196: { NULL, 0, 0 }, /* Ed */
197: { NULL, 0, 0 }, /* Bl */
198: { NULL, 0, 0 }, /* El */
199: { NULL, 0, 0 }, /* It */
200: { NULL, 0, 0 }, /* Ad */
201: { NULL, TYPE_An, 0 }, /* An */
1.246 schwarze 202: { NULL, 0, 0 }, /* Ap */
1.235 schwarze 203: { NULL, TYPE_Ar, 0 }, /* Ar */
204: { NULL, TYPE_Cd, 0 }, /* Cd */
205: { NULL, TYPE_Cm, 0 }, /* Cm */
206: { NULL, TYPE_Dv, 0 }, /* Dv */
207: { NULL, TYPE_Er, 0 }, /* Er */
208: { NULL, TYPE_Ev, 0 }, /* Ev */
209: { NULL, 0, 0 }, /* Ex */
210: { NULL, TYPE_Fa, 0 }, /* Fa */
211: { parse_mdoc_Fd, 0, 0 }, /* Fd */
212: { NULL, TYPE_Fl, 0 }, /* Fl */
213: { parse_mdoc_Fn, 0, 0 }, /* Fn */
214: { NULL, TYPE_Ft, 0 }, /* Ft */
215: { NULL, TYPE_Ic, 0 }, /* Ic */
216: { NULL, TYPE_In, 0 }, /* In */
217: { NULL, TYPE_Li, 0 }, /* Li */
218: { parse_mdoc_Nd, 0, 0 }, /* Nd */
219: { parse_mdoc_Nm, 0, 0 }, /* Nm */
220: { NULL, 0, 0 }, /* Op */
221: { NULL, 0, 0 }, /* Ot */
222: { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
223: { NULL, 0, 0 }, /* Rv */
224: { NULL, TYPE_St, 0 }, /* St */
225: { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
226: { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
227: { parse_mdoc_Xr, 0, 0 }, /* Xr */
228: { NULL, 0, 0 }, /* %A */
229: { NULL, 0, 0 }, /* %B */
230: { NULL, 0, 0 }, /* %D */
231: { NULL, 0, 0 }, /* %I */
232: { NULL, 0, 0 }, /* %J */
233: { NULL, 0, 0 }, /* %N */
234: { NULL, 0, 0 }, /* %O */
235: { NULL, 0, 0 }, /* %P */
236: { NULL, 0, 0 }, /* %R */
237: { NULL, 0, 0 }, /* %T */
238: { NULL, 0, 0 }, /* %V */
239: { NULL, 0, 0 }, /* Ac */
240: { NULL, 0, 0 }, /* Ao */
241: { NULL, 0, 0 }, /* Aq */
1.237 schwarze 242: { NULL, TYPE_At, 0 }, /* At */
1.235 schwarze 243: { NULL, 0, 0 }, /* Bc */
244: { NULL, 0, 0 }, /* Bf */
245: { NULL, 0, 0 }, /* Bo */
246: { NULL, 0, 0 }, /* Bq */
247: { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
1.236 schwarze 248: { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
1.235 schwarze 249: { NULL, 0, 0 }, /* Db */
250: { NULL, 0, 0 }, /* Dc */
251: { NULL, 0, 0 }, /* Do */
252: { NULL, 0, 0 }, /* Dq */
253: { NULL, 0, 0 }, /* Ec */
254: { NULL, 0, 0 }, /* Ef */
255: { NULL, TYPE_Em, 0 }, /* Em */
256: { NULL, 0, 0 }, /* Eo */
257: { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
258: { NULL, TYPE_Ms, 0 }, /* Ms */
259: { NULL, 0, 0 }, /* No */
260: { NULL, 0, 0 }, /* Ns */
261: { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
262: { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
263: { NULL, 0, 0 }, /* Pc */
264: { NULL, 0, 0 }, /* Pf */
265: { NULL, 0, 0 }, /* Po */
266: { NULL, 0, 0 }, /* Pq */
267: { NULL, 0, 0 }, /* Qc */
268: { NULL, 0, 0 }, /* Ql */
269: { NULL, 0, 0 }, /* Qo */
270: { NULL, 0, 0 }, /* Qq */
271: { NULL, 0, 0 }, /* Re */
272: { NULL, 0, 0 }, /* Rs */
273: { NULL, 0, 0 }, /* Sc */
274: { NULL, 0, 0 }, /* So */
275: { NULL, 0, 0 }, /* Sq */
276: { NULL, 0, 0 }, /* Sm */
277: { NULL, 0, 0 }, /* Sx */
278: { NULL, TYPE_Sy, 0 }, /* Sy */
279: { NULL, TYPE_Tn, 0 }, /* Tn */
280: { NULL, 0, NODE_NOSRC }, /* Ux */
281: { NULL, 0, 0 }, /* Xc */
282: { NULL, 0, 0 }, /* Xo */
283: { parse_mdoc_Fo, 0, 0 }, /* Fo */
284: { NULL, 0, 0 }, /* Fc */
285: { NULL, 0, 0 }, /* Oo */
286: { NULL, 0, 0 }, /* Oc */
287: { NULL, 0, 0 }, /* Bk */
288: { NULL, 0, 0 }, /* Ek */
289: { NULL, 0, 0 }, /* Bt */
290: { NULL, 0, 0 }, /* Hf */
291: { NULL, 0, 0 }, /* Fr */
292: { NULL, 0, 0 }, /* Ud */
1.237 schwarze 293: { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
1.235 schwarze 294: { NULL, 0, 0 }, /* Lp */
295: { NULL, TYPE_Lk, 0 }, /* Lk */
296: { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
297: { NULL, 0, 0 }, /* Brq */
298: { NULL, 0, 0 }, /* Bro */
299: { NULL, 0, 0 }, /* Brc */
300: { NULL, 0, 0 }, /* %C */
301: { NULL, 0, 0 }, /* Es */
302: { NULL, 0, 0 }, /* En */
303: { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
304: { NULL, 0, 0 }, /* %Q */
305: { NULL, 0, 0 }, /* %U */
306: { NULL, 0, 0 }, /* Ta */
1.1 kristaps 307: };
1.246 schwarze 308: static const struct mdoc_handler *const mdocs = __mdocs - MDOC_Dd;
1.1 kristaps 309:
1.141 schwarze 310:
1.1 kristaps 311: int
1.179 schwarze 312: mandocdb(int argc, char *argv[])
1.1 kristaps 313: {
1.187 schwarze 314: struct manconf conf;
315: struct mparse *mp;
1.221 schwarze 316: struct dba *dba;
1.206 schwarze 317: const char *path_arg, *progname;
1.187 schwarze 318: size_t j, sz;
1.59 schwarze 319: int ch, i;
1.50 kristaps 320:
1.207 schwarze 321: #if HAVE_PLEDGE
322: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 323: warn("pledge");
1.218 kristaps 324: return (int)MANDOCLEVEL_SYSERR;
325: }
326: #endif
327:
328: #if HAVE_SANDBOX_INIT
329: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
330: warnx("sandbox_init");
1.207 schwarze 331: return (int)MANDOCLEVEL_SYSERR;
332: }
333: #endif
334:
1.187 schwarze 335: memset(&conf, 0, sizeof(conf));
1.50 kristaps 336:
337: /*
1.141 schwarze 338: * We accept a few different invocations.
1.50 kristaps 339: * The CHECKOP macro makes sure that invocation styles don't
340: * clobber each other.
341: */
342: #define CHECKOP(_op, _ch) do \
343: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 344: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 345: goto usage; \
346: } while (/*CONSTCOND*/0)
1.10 kristaps 347:
1.59 schwarze 348: path_arg = NULL;
1.38 schwarze 349: op = OP_DEFAULT;
1.1 kristaps 350:
1.126 schwarze 351: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 352: switch (ch) {
1.141 schwarze 353: case 'a':
1.12 schwarze 354: use_all = 1;
355: break;
1.141 schwarze 356: case 'C':
1.50 kristaps 357: CHECKOP(op, ch);
1.59 schwarze 358: path_arg = optarg;
1.38 schwarze 359: op = OP_CONFFILE;
1.34 schwarze 360: break;
1.141 schwarze 361: case 'D':
1.125 schwarze 362: debug++;
363: break;
1.141 schwarze 364: case 'd':
1.50 kristaps 365: CHECKOP(op, ch);
1.59 schwarze 366: path_arg = optarg;
1.5 kristaps 367: op = OP_UPDATE;
368: break;
1.141 schwarze 369: case 'n':
1.50 kristaps 370: nodb = 1;
371: break;
1.141 schwarze 372: case 'p':
1.126 schwarze 373: warnings = 1;
374: break;
1.141 schwarze 375: case 'Q':
1.116 schwarze 376: mparse_options |= MPARSE_QUICK;
1.102 schwarze 377: break;
1.141 schwarze 378: case 'T':
1.95 schwarze 379: if (strcmp(optarg, "utf8")) {
1.198 schwarze 380: warnx("-T%s: Unsupported output format",
381: optarg);
1.95 schwarze 382: goto usage;
383: }
384: write_utf8 = 1;
385: break;
1.141 schwarze 386: case 't':
1.50 kristaps 387: CHECKOP(op, ch);
1.38 schwarze 388: dup2(STDOUT_FILENO, STDERR_FILENO);
389: op = OP_TEST;
1.50 kristaps 390: nodb = warnings = 1;
1.38 schwarze 391: break;
1.141 schwarze 392: case 'u':
1.50 kristaps 393: CHECKOP(op, ch);
1.59 schwarze 394: path_arg = optarg;
1.5 kristaps 395: op = OP_DELETE;
396: break;
1.141 schwarze 397: case 'v':
1.126 schwarze 398: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 399: break;
1.1 kristaps 400: default:
1.38 schwarze 401: goto usage;
1.1 kristaps 402: }
403:
404: argc -= optind;
405: argv += optind;
406:
1.207 schwarze 407: #if HAVE_PLEDGE
1.210 schwarze 408: if (nodb) {
409: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 410: warn("pledge");
1.210 schwarze 411: return (int)MANDOCLEVEL_SYSERR;
412: }
1.207 schwarze 413: }
414: #endif
415:
1.38 schwarze 416: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 417: warnx("-C: Too many arguments");
1.38 schwarze 418: goto usage;
419: }
420:
1.59 schwarze 421: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 422: mchars_alloc();
423: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 424: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
425: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 426:
427: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 428:
1.50 kristaps 429: /*
1.150 schwarze 430: * Most of these deal with a specific directory.
1.140 schwarze 431: * Jump into that directory first.
1.50 kristaps 432: */
1.165 schwarze 433: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 434: goto out;
1.140 schwarze 435:
1.225 schwarze 436: dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
437: if (dba != NULL) {
1.140 schwarze 438: /*
439: * The existing database is usable. Process
440: * all files specified on the command-line.
441: */
1.207 schwarze 442: #if HAVE_PLEDGE
1.210 schwarze 443: if (!nodb) {
444: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 445: warn("pledge");
1.210 schwarze 446: exitcode = (int)MANDOCLEVEL_SYSERR;
447: goto out;
448: }
1.207 schwarze 449: }
450: #endif
1.140 schwarze 451: use_all = 1;
1.138 schwarze 452: for (i = 0; i < argc; i++)
453: filescan(argv[i]);
1.225 schwarze 454: if (nodb == 0)
1.221 schwarze 455: dbprune(dba);
1.138 schwarze 456: } else {
1.228 schwarze 457: /* Database missing or corrupt. */
1.230 schwarze 458: if (op != OP_UPDATE || errno != ENOENT)
459: say(MANDOC_DB, "%s: Automatically recreating"
460: " from scratch", strerror(errno));
1.140 schwarze 461: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 462: op = OP_DEFAULT;
463: if (0 == treescan())
464: goto out;
1.221 schwarze 465: dba = dba_new(128);
1.138 schwarze 466: }
1.50 kristaps 467: if (OP_DELETE != op)
1.221 schwarze 468: mpages_merge(dba, mp);
469: if (nodb == 0)
470: dbwrite(dba);
471: dba_free(dba);
1.50 kristaps 472: } else {
473: /*
474: * If we have arguments, use them as our manpaths.
1.222 schwarze 475: * If we don't, use man.conf(5).
1.50 kristaps 476: */
477: if (argc > 0) {
1.187 schwarze 478: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 479: argc, sizeof(char *));
1.187 schwarze 480: conf.manpath.sz = (size_t)argc;
1.50 kristaps 481: for (i = 0; i < argc; i++)
1.187 schwarze 482: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 483: } else
1.187 schwarze 484: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 485:
1.187 schwarze 486: if (conf.manpath.sz == 0) {
1.127 schwarze 487: exitcode = (int)MANDOCLEVEL_BADARG;
488: say("", "Empty manpath");
489: }
1.50 kristaps 490:
491: /*
1.69 schwarze 492: * First scan the tree rooted at a base directory, then
493: * build a new database and finally move it into place.
1.50 kristaps 494: * Ignore zero-length directories and strip trailing
495: * slashes.
496: */
1.187 schwarze 497: for (j = 0; j < conf.manpath.sz; j++) {
498: sz = strlen(conf.manpath.paths[j]);
499: if (sz && conf.manpath.paths[j][sz - 1] == '/')
500: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 501: if (0 == sz)
502: continue;
1.66 schwarze 503:
504: if (j) {
1.202 schwarze 505: mandoc_ohash_init(&mpages, 6,
506: offsetof(struct mpage, inodev));
507: mandoc_ohash_init(&mlinks, 6,
508: offsetof(struct mlink, file));
1.66 schwarze 509: }
510:
1.187 schwarze 511: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 512: continue;
1.59 schwarze 513: if (0 == treescan())
1.165 schwarze 514: continue;
1.221 schwarze 515: dba = dba_new(128);
516: mpages_merge(dba, mp);
517: if (nodb == 0)
518: dbwrite(dba);
519: dba_free(dba);
1.66 schwarze 520:
1.187 schwarze 521: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 522: mpages_free();
1.78 schwarze 523: ohash_delete(&mpages);
1.83 schwarze 524: ohash_delete(&mlinks);
1.66 schwarze 525: }
1.50 kristaps 526: }
527: }
528: out:
1.187 schwarze 529: manconf_free(&conf);
1.168 schwarze 530: mparse_free(mp);
1.203 schwarze 531: mchars_free();
1.80 schwarze 532: mpages_free();
1.78 schwarze 533: ohash_delete(&mpages);
1.83 schwarze 534: ohash_delete(&mlinks);
1.197 schwarze 535: return exitcode;
1.50 kristaps 536: usage:
1.206 schwarze 537: progname = getprogname();
1.126 schwarze 538: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
539: " %s [-aDnpQ] [-Tutf8] dir ...\n"
540: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
541: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 542: " %s [-Q] -t file ...\n",
1.206 schwarze 543: progname, progname, progname, progname, progname);
1.50 kristaps 544:
1.197 schwarze 545: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 546: }
547:
548: /*
1.231 schwarze 549: * To get a singly linked list in alpha order while inserting entries
550: * at the beginning, process directory entries in reverse alpha order.
551: */
552: static int
1.232 schwarze 553: #if HAVE_FTS_COMPARE_CONST
554: fts_compare(const FTSENT *const *a, const FTSENT *const *b)
555: #else
1.231 schwarze 556: fts_compare(const FTSENT **a, const FTSENT **b)
1.232 schwarze 557: #endif
1.231 schwarze 558: {
559: return -strcmp((*a)->fts_name, (*b)->fts_name);
560: }
561:
562: /*
1.59 schwarze 563: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 564: * We use fts(), scanning directory parts along the way for clues to our
565: * section and architecture.
566: *
567: * If use_all has been specified, grok all files.
568: * If not, sanitise paths to the following:
569: *
1.141 schwarze 570: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 571: * or
572: * [./]cat<section>[/<arch>]/<name>.0
573: *
1.216 schwarze 574: * TODO: accommodate for multi-language directories.
1.50 kristaps 575: */
576: static int
1.59 schwarze 577: treescan(void)
1.50 kristaps 578: {
1.139 schwarze 579: char buf[PATH_MAX];
1.50 kristaps 580: FTS *f;
581: FTSENT *ff;
1.84 schwarze 582: struct mlink *mlink;
1.221 schwarze 583: int gzip;
584: enum form dform;
1.94 schwarze 585: char *dsec, *arch, *fsec, *cp;
586: const char *path;
1.50 kristaps 587: const char *argv[2];
588:
589: argv[0] = ".";
1.244 schwarze 590: argv[1] = NULL;
1.50 kristaps 591:
1.231 schwarze 592: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
593: fts_compare);
1.200 schwarze 594: if (f == NULL) {
1.59 schwarze 595: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 596: say("", "&fts_open");
1.197 schwarze 597: return 0;
1.50 kristaps 598: }
599:
600: dsec = arch = NULL;
601: dform = FORM_NONE;
602:
1.200 schwarze 603: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 604: path = ff->fts_path + 2;
1.139 schwarze 605: switch (ff->fts_info) {
606:
607: /*
608: * Symbolic links require various sanity checks,
609: * then get handled just like regular files.
610: */
1.141 schwarze 611: case FTS_SL:
1.200 schwarze 612: if (realpath(path, buf) == NULL) {
1.139 schwarze 613: if (warnings)
614: say(path, "&realpath");
615: continue;
616: }
1.184 schwarze 617: if (strstr(buf, basedir) != buf
618: #ifdef HOMEBREWDIR
619: && strstr(buf, HOMEBREWDIR) != buf
620: #endif
621: ) {
1.139 schwarze 622: if (warnings) say("",
623: "%s: outside base directory", buf);
624: continue;
625: }
626: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 627: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 628: if (warnings)
629: say(path, "&stat");
630: continue;
631: }
632: /* FALLTHROUGH */
633:
1.50 kristaps 634: /*
1.83 schwarze 635: * If we're a regular file, add an mlink by using the
1.50 kristaps 636: * stored directory data and handling the filename.
637: */
1.141 schwarze 638: case FTS_F:
1.200 schwarze 639: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 640: continue;
1.50 kristaps 641: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 642: if (warnings)
1.59 schwarze 643: say(path, "Extraneous file");
1.50 kristaps 644: continue;
1.124 schwarze 645: }
646: gzip = 0;
647: fsec = NULL;
1.200 schwarze 648: while (fsec == NULL) {
1.124 schwarze 649: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 650: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 651: break;
652: gzip = 1;
653: *fsec = '\0';
654: fsec = NULL;
655: }
1.200 schwarze 656: if (fsec == NULL) {
1.60 schwarze 657: if ( ! use_all) {
1.56 schwarze 658: if (warnings)
1.60 schwarze 659: say(path,
660: "No filename suffix");
1.50 kristaps 661: continue;
662: }
1.200 schwarze 663: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 664: if (warnings)
665: say(path, "Skip html");
666: continue;
1.200 schwarze 667: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 668: if (warnings)
669: say(path, "Skip ps");
670: continue;
1.200 schwarze 671: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 672: if (warnings)
673: say(path, "Skip pdf");
674: continue;
675: } else if ( ! use_all &&
1.200 schwarze 676: ((dform == FORM_SRC &&
1.185 schwarze 677: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 678: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 679: if (warnings)
680: say(path, "Wrong filename suffix");
681: continue;
1.66 schwarze 682: } else
1.84 schwarze 683: fsec[-1] = '\0';
1.94 schwarze 684:
1.84 schwarze 685: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 686: if (strlcpy(mlink->file, path,
687: sizeof(mlink->file)) >=
688: sizeof(mlink->file)) {
689: say(path, "Filename too long");
690: free(mlink);
691: continue;
692: }
1.84 schwarze 693: mlink->dform = dform;
1.94 schwarze 694: mlink->dsec = dsec;
695: mlink->arch = arch;
696: mlink->name = ff->fts_name;
697: mlink->fsec = fsec;
1.124 schwarze 698: mlink->gzip = gzip;
1.84 schwarze 699: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 700: continue;
1.139 schwarze 701:
1.141 schwarze 702: case FTS_D:
703: case FTS_DP:
1.139 schwarze 704: break;
705:
706: default:
1.60 schwarze 707: if (warnings)
708: say(path, "Not a regular file");
1.50 kristaps 709: continue;
1.60 schwarze 710: }
1.1 kristaps 711:
1.50 kristaps 712: switch (ff->fts_level) {
1.141 schwarze 713: case 0:
1.50 kristaps 714: /* Ignore the root directory. */
715: break;
1.141 schwarze 716: case 1:
1.50 kristaps 717: /*
718: * This might contain manX/ or catX/.
719: * Try to infer this from the name.
720: * If we're not in use_all, enforce it.
721: */
722: cp = ff->fts_name;
1.200 schwarze 723: if (ff->fts_info == FTS_DP) {
724: dform = FORM_NONE;
725: dsec = NULL;
1.50 kristaps 726: break;
1.200 schwarze 727: }
1.1 kristaps 728:
1.200 schwarze 729: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 730: dform = FORM_SRC;
1.66 schwarze 731: dsec = cp + 3;
1.200 schwarze 732: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 733: dform = FORM_CAT;
1.66 schwarze 734: dsec = cp + 3;
1.94 schwarze 735: } else {
736: dform = FORM_NONE;
737: dsec = NULL;
1.50 kristaps 738: }
1.1 kristaps 739:
1.200 schwarze 740: if (dsec != NULL || use_all)
1.50 kristaps 741: break;
1.1 kristaps 742:
1.56 schwarze 743: if (warnings)
1.59 schwarze 744: say(path, "Unknown directory part");
1.50 kristaps 745: fts_set(f, ff, FTS_SKIP);
746: break;
1.141 schwarze 747: case 2:
1.50 kristaps 748: /*
749: * Possibly our architecture.
750: * If we're descending, keep tabs on it.
751: */
1.200 schwarze 752: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 753: arch = ff->fts_name;
1.94 schwarze 754: else
755: arch = NULL;
1.50 kristaps 756: break;
757: default:
1.200 schwarze 758: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 759: break;
1.56 schwarze 760: if (warnings)
1.59 schwarze 761: say(path, "Extraneous directory part");
1.50 kristaps 762: fts_set(f, ff, FTS_SKIP);
763: break;
1.5 kristaps 764: }
1.50 kristaps 765: }
766:
767: fts_close(f);
1.197 schwarze 768: return 1;
1.50 kristaps 769: }
1.5 kristaps 770:
1.50 kristaps 771: /*
1.89 schwarze 772: * Add a file to the mlinks table.
1.50 kristaps 773: * Do not verify that it's a "valid" looking manpage (we'll do that
774: * later).
775: *
776: * Try to infer the manual section, architecture, and page name from the
777: * path, assuming it looks like
778: *
1.141 schwarze 779: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 780: * or
781: * [./]cat<section>[/<arch>]/<name>.0
782: *
783: * See treescan() for the fts(3) version of this.
784: */
785: static void
1.59 schwarze 786: filescan(const char *file)
1.50 kristaps 787: {
1.59 schwarze 788: char buf[PATH_MAX];
1.84 schwarze 789: struct stat st;
790: struct mlink *mlink;
1.59 schwarze 791: char *p, *start;
1.5 kristaps 792:
1.50 kristaps 793: assert(use_all);
1.5 kristaps 794:
1.50 kristaps 795: if (0 == strncmp(file, "./", 2))
796: file += 2;
1.5 kristaps 797:
1.139 schwarze 798: /*
799: * We have to do lstat(2) before realpath(3) loses
800: * the information whether this is a symbolic link.
801: * We need to know that because for symbolic links,
802: * we want to use the orginal file name, while for
803: * regular files, we want to use the real path.
804: */
805: if (-1 == lstat(file, &st)) {
806: exitcode = (int)MANDOCLEVEL_BADARG;
807: say(file, "&lstat");
808: return;
809: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
810: exitcode = (int)MANDOCLEVEL_BADARG;
811: say(file, "Not a regular file");
812: return;
813: }
814:
815: /*
816: * We have to resolve the file name to the real path
817: * in any case for the base directory check.
818: */
1.59 schwarze 819: if (NULL == realpath(file, buf)) {
820: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 821: say(file, "&realpath");
1.59 schwarze 822: return;
1.106 schwarze 823: }
824:
1.150 schwarze 825: if (OP_TEST == op)
1.106 schwarze 826: start = buf;
1.150 schwarze 827: else if (strstr(buf, basedir) == buf)
828: start = buf + strlen(basedir);
1.184 schwarze 829: #ifdef HOMEBREWDIR
830: else if (strstr(buf, HOMEBREWDIR) == buf)
831: start = buf;
832: #endif
1.106 schwarze 833: else {
1.59 schwarze 834: exitcode = (int)MANDOCLEVEL_BADARG;
835: say("", "%s: outside base directory", buf);
836: return;
1.106 schwarze 837: }
838:
1.139 schwarze 839: /*
840: * Now we are sure the file is inside our tree.
841: * If it is a symbolic link, ignore the real path
842: * and use the original name.
843: * This implies passing stuff like "cat1/../man1/foo.1"
844: * on the command line won't work. So don't do that.
845: * Note the stat(2) can still fail if the link target
846: * doesn't exist.
847: */
848: if (S_IFLNK & st.st_mode) {
849: if (-1 == stat(buf, &st)) {
850: exitcode = (int)MANDOCLEVEL_BADARG;
851: say(file, "&stat");
852: return;
853: }
1.143 schwarze 854: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
855: say(file, "Filename too long");
856: return;
857: }
1.150 schwarze 858: start = buf;
859: if (OP_TEST != op && strstr(buf, basedir) == buf)
860: start += strlen(basedir);
1.50 kristaps 861: }
1.106 schwarze 862:
1.84 schwarze 863: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 864: mlink->dform = FORM_NONE;
1.143 schwarze 865: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
866: sizeof(mlink->file)) {
867: say(start, "Filename too long");
1.180 schwarze 868: free(mlink);
1.143 schwarze 869: return;
1.239 schwarze 870: }
871:
872: /*
873: * In test mode or when the original name is absolute
874: * but outside our tree, guess the base directory.
875: */
876:
877: if (op == OP_TEST || (start == buf && *start == '/')) {
878: if (strncmp(buf, "man/", 4) == 0)
879: start = buf + 4;
880: else if ((start = strstr(buf, "/man/")) != NULL)
881: start += 5;
882: else
883: start = buf;
1.143 schwarze 884: }
1.17 schwarze 885:
1.50 kristaps 886: /*
887: * First try to guess our directory structure.
888: * If we find a separator, try to look for man* or cat*.
889: * If we find one of these and what's underneath is a directory,
890: * assume it's an architecture.
891: */
892: if (NULL != (p = strchr(start, '/'))) {
893: *p++ = '\0';
894: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 895: mlink->dform = FORM_SRC;
1.94 schwarze 896: mlink->dsec = start + 3;
1.50 kristaps 897: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 898: mlink->dform = FORM_CAT;
1.94 schwarze 899: mlink->dsec = start + 3;
1.17 schwarze 900: }
1.5 kristaps 901:
1.50 kristaps 902: start = p;
1.84 schwarze 903: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 904: *p++ = '\0';
1.94 schwarze 905: mlink->arch = start;
1.50 kristaps 906: start = p;
1.84 schwarze 907: }
1.50 kristaps 908: }
909:
910: /*
911: * Now check the file suffix.
912: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
913: */
914: p = strrchr(start, '\0');
915: while (p-- > start && '/' != *p && '.' != *p)
916: /* Loop. */ ;
917:
918: if ('.' == *p) {
919: *p++ = '\0';
1.94 schwarze 920: mlink->fsec = p;
1.5 kristaps 921: }
922:
1.10 kristaps 923: /*
1.50 kristaps 924: * Now try to parse the name.
925: * Use the filename portion of the path.
1.10 kristaps 926: */
1.84 schwarze 927: mlink->name = start;
1.50 kristaps 928: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 929: mlink->name = p + 1;
1.50 kristaps 930: *p = '\0';
1.84 schwarze 931: }
932: mlink_add(mlink, &st);
1.50 kristaps 933: }
1.5 kristaps 934:
1.50 kristaps 935: static void
1.84 schwarze 936: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 937: {
1.83 schwarze 938: struct inodev inodev;
1.78 schwarze 939: struct mpage *mpage;
1.83 schwarze 940: unsigned int slot;
1.50 kristaps 941:
1.84 schwarze 942: assert(NULL != mlink->file);
1.50 kristaps 943:
1.94 schwarze 944: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
945: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
946: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
947: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 948:
949: if ('0' == *mlink->fsec) {
950: free(mlink->fsec);
951: mlink->fsec = mandoc_strdup(mlink->dsec);
952: mlink->fform = FORM_CAT;
953: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
954: mlink->fform = FORM_SRC;
1.74 schwarze 955: else
1.84 schwarze 956: mlink->fform = FORM_NONE;
1.82 schwarze 957:
1.83 schwarze 958: slot = ohash_qlookup(&mlinks, mlink->file);
959: assert(NULL == ohash_find(&mlinks, slot));
960: ohash_insert(&mlinks, slot, mlink);
961:
1.186 schwarze 962: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 963: inodev.st_ino = st->st_ino;
964: inodev.st_dev = st->st_dev;
965: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
966: sizeof(struct inodev), inodev.st_ino);
967: mpage = ohash_find(&mpages, slot);
968: if (NULL == mpage) {
969: mpage = mandoc_calloc(1, sizeof(struct mpage));
970: mpage->inodev.st_ino = inodev.st_ino;
971: mpage->inodev.st_dev = inodev.st_dev;
1.221 schwarze 972: mpage->form = FORM_NONE;
1.231 schwarze 973: mpage->next = mpage_head;
974: mpage_head = mpage;
1.83 schwarze 975: ohash_insert(&mpages, slot, mpage);
976: } else
1.85 schwarze 977: mlink->next = mpage->mlinks;
1.82 schwarze 978: mpage->mlinks = mlink;
1.118 schwarze 979: mlink->mpage = mpage;
1.82 schwarze 980: }
981:
982: static void
983: mlink_free(struct mlink *mlink)
984: {
985:
986: free(mlink->dsec);
987: free(mlink->arch);
988: free(mlink->name);
989: free(mlink->fsec);
990: free(mlink);
1.50 kristaps 991: }
1.3 kristaps 992:
1.50 kristaps 993: static void
1.78 schwarze 994: mpages_free(void)
1.50 kristaps 995: {
1.78 schwarze 996: struct mpage *mpage;
1.82 schwarze 997: struct mlink *mlink;
1.3 kristaps 998:
1.231 schwarze 999: while ((mpage = mpage_head) != NULL) {
1000: while ((mlink = mpage->mlinks) != NULL) {
1.85 schwarze 1001: mpage->mlinks = mlink->next;
1.82 schwarze 1002: mlink_free(mlink);
1003: }
1.231 schwarze 1004: mpage_head = mpage->next;
1.78 schwarze 1005: free(mpage->sec);
1006: free(mpage->arch);
1.82 schwarze 1007: free(mpage->title);
1008: free(mpage->desc);
1.78 schwarze 1009: free(mpage);
1.50 kristaps 1010: }
1011: }
1.38 schwarze 1012:
1.50 kristaps 1013: /*
1.89 schwarze 1014: * For each mlink to the mpage, check whether the path looks like
1015: * it is formatted, and if it does, check whether a source manual
1016: * exists by the same name, ignoring the suffix.
1017: * If both conditions hold, drop the mlink.
1018: */
1019: static void
1020: mlinks_undupe(struct mpage *mpage)
1021: {
1022: char buf[PATH_MAX];
1023: struct mlink **prev;
1024: struct mlink *mlink;
1025: char *bufp;
1026:
1027: mpage->form = FORM_CAT;
1.90 schwarze 1028: prev = &mpage->mlinks;
1029: while (NULL != (mlink = *prev)) {
1.89 schwarze 1030: if (FORM_CAT != mlink->dform) {
1031: mpage->form = FORM_NONE;
1.90 schwarze 1032: goto nextlink;
1.89 schwarze 1033: }
1.143 schwarze 1034: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1035: bufp = strstr(buf, "cat");
1036: assert(NULL != bufp);
1037: memcpy(bufp, "man", 3);
1038: if (NULL != (bufp = strrchr(buf, '.')))
1039: *++bufp = '\0';
1.143 schwarze 1040: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1041: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1042: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1043: goto nextlink;
1.89 schwarze 1044: if (warnings)
1045: say(mlink->file, "Man source exists: %s", buf);
1046: if (use_all)
1.90 schwarze 1047: goto nextlink;
1.89 schwarze 1048: *prev = mlink->next;
1049: mlink_free(mlink);
1.90 schwarze 1050: continue;
1051: nextlink:
1052: prev = &(*prev)->next;
1.89 schwarze 1053: }
1054: }
1055:
1.131 schwarze 1056: static void
1.93 schwarze 1057: mlink_check(struct mpage *mpage, struct mlink *mlink)
1058: {
1.131 schwarze 1059: struct str *str;
1060: unsigned int slot;
1.93 schwarze 1061:
1062: /*
1063: * Check whether the manual section given in a file
1064: * agrees with the directory where the file is located.
1065: * Some manuals have suffixes like (3p) on their
1066: * section number either inside the file or in the
1067: * directory name, some are linked into more than one
1068: * section, like encrypt(1) = makekey(8).
1069: */
1070:
1071: if (FORM_SRC == mpage->form &&
1.131 schwarze 1072: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1073: say(mlink->file, "Section \"%s\" manual in %s directory",
1074: mpage->sec, mlink->dsec);
1075:
1076: /*
1077: * Manual page directories exist for each kernel
1078: * architecture as returned by machine(1).
1079: * However, many manuals only depend on the
1080: * application architecture as returned by arch(1).
1081: * For example, some (2/ARM) manuals are shared
1082: * across the "armish" and "zaurus" kernel
1083: * architectures.
1084: * A few manuals are even shared across completely
1085: * different architectures, for example fdformat(1)
1.229 schwarze 1086: * on amd64, i386, and sparc64.
1.93 schwarze 1087: */
1088:
1.131 schwarze 1089: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1090: say(mlink->file, "Architecture \"%s\" manual in "
1091: "\"%s\" directory", mpage->arch, mlink->arch);
1092:
1.131 schwarze 1093: /*
1094: * XXX
1.133 schwarze 1095: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1096: */
1097:
1098: if (FORM_CAT == mpage->form)
1099: return;
1100:
1101: /*
1102: * Check whether this mlink
1103: * appears as a name in the NAME section.
1104: */
1.93 schwarze 1105:
1.133 schwarze 1106: slot = ohash_qlookup(&names, mlink->name);
1107: str = ohash_find(&names, slot);
1.131 schwarze 1108: assert(NULL != str);
1.133 schwarze 1109: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1110: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1111: }
1112:
1.89 schwarze 1113: /*
1.80 schwarze 1114: * Run through the files in the global vector "mpages"
1115: * and add them to the database specified in "basedir".
1.50 kristaps 1116: *
1117: * This handles the parsing scheme itself, using the cues of directory
1118: * and filename to determine whether the file is parsable or not.
1119: */
1.59 schwarze 1120: static void
1.221 schwarze 1121: mpages_merge(struct dba *dba, struct mparse *mp)
1.50 kristaps 1122: {
1.231 schwarze 1123: struct mpage *mpage, *mpage_dest;
1.118 schwarze 1124: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1125: struct roff_man *man;
1.118 schwarze 1126: char *sodest;
1.112 schwarze 1127: char *cp;
1.162 schwarze 1128: int fd;
1.69 schwarze 1129:
1.231 schwarze 1130: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.89 schwarze 1131: mlinks_undupe(mpage);
1.231 schwarze 1132: if ((mlink = mpage->mlinks) == NULL)
1133: continue;
1.50 kristaps 1134:
1.133 schwarze 1135: name_mask = NAME_MASK;
1.202 schwarze 1136: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1137: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1138: mparse_reset(mp);
1139: man = NULL;
1.124 schwarze 1140: sodest = NULL;
1141:
1.213 schwarze 1142: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1143: say(mlink->file, "&open");
1.162 schwarze 1144: goto nextpage;
1.124 schwarze 1145: }
1.14 schwarze 1146:
1147: /*
1.183 schwarze 1148: * Interpret the file as mdoc(7) or man(7) source
1149: * code, unless it is known to be formatted.
1.14 schwarze 1150: */
1.182 schwarze 1151: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1152: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1153: close(fd);
1.240 schwarze 1154: fd = -1;
1.192 schwarze 1155: mparse_result(mp, &man, &sodest);
1.76 schwarze 1156: }
1.14 schwarze 1157:
1.171 schwarze 1158: if (sodest != NULL) {
1.118 schwarze 1159: mlink_dest = ohash_find(&mlinks,
1160: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1161: if (mlink_dest == NULL) {
1162: mandoc_asprintf(&cp, "%s.gz", sodest);
1163: mlink_dest = ohash_find(&mlinks,
1164: ohash_qlookup(&mlinks, cp));
1165: free(cp);
1166: }
1167: if (mlink_dest != NULL) {
1.118 schwarze 1168:
1169: /* The .so target exists. */
1170:
1171: mpage_dest = mlink_dest->mpage;
1172: while (1) {
1173: mlink->mpage = mpage_dest;
1174:
1175: /*
1176: * If the target was already
1177: * processed, add the links
1178: * to the database now.
1179: * Otherwise, this will
1180: * happen when we come
1181: * to the target.
1182: */
1183:
1.221 schwarze 1184: if (mpage_dest->dba != NULL)
1185: dbadd_mlink(mlink);
1.118 schwarze 1186:
1.171 schwarze 1187: if (mlink->next == NULL)
1.118 schwarze 1188: break;
1189: mlink = mlink->next;
1190: }
1191:
1192: /* Move all links to the target. */
1193:
1194: mlink->next = mlink_dest->next;
1195: mlink_dest->next = mpage->mlinks;
1196: mpage->mlinks = NULL;
1197: }
1.124 schwarze 1198: goto nextpage;
1.192 schwarze 1199: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1200: mdoc_validate(man);
1.82 schwarze 1201: mpage->form = FORM_SRC;
1.193 schwarze 1202: mpage->sec = man->meta.msec;
1.155 schwarze 1203: mpage->sec = mandoc_strdup(
1.171 schwarze 1204: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1205: mpage->arch = man->meta.arch;
1.84 schwarze 1206: mpage->arch = mandoc_strdup(
1.171 schwarze 1207: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1208: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1209: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1210: man_validate(man);
1.240 schwarze 1211: if (*man->meta.msec != '\0' ||
1.245 schwarze 1212: *man->meta.title != '\0') {
1.240 schwarze 1213: mpage->form = FORM_SRC;
1214: mpage->sec = mandoc_strdup(man->meta.msec);
1215: mpage->arch = mandoc_strdup(mlink->arch);
1216: mpage->title = mandoc_strdup(man->meta.title);
1217: } else
1218: man = NULL;
1219: }
1220:
1221: assert(mpage->desc == NULL);
1222: if (man == NULL) {
1.82 schwarze 1223: mpage->form = FORM_CAT;
1.182 schwarze 1224: mpage->sec = mandoc_strdup(mlink->dsec);
1225: mpage->arch = mandoc_strdup(mlink->arch);
1226: mpage->title = mandoc_strdup(mlink->name);
1.240 schwarze 1227: parse_cat(mpage, fd);
1228: } else if (man->macroset == MACROSET_MDOC)
1.193 schwarze 1229: parse_mdoc(mpage, &man->meta, man->first);
1.240 schwarze 1230: else
1.193 schwarze 1231: parse_man(mpage, &man->meta, man->first);
1.242 schwarze 1232: if (mpage->desc == NULL) {
1233: mpage->desc = mandoc_strdup(mlink->name);
1234: if (warnings)
1235: say(mlink->file, "No one-line description, "
1236: "using filename \"%s\"", mlink->name);
1237: }
1.131 schwarze 1238:
1.238 schwarze 1239: for (mlink = mpage->mlinks;
1240: mlink != NULL;
1241: mlink = mlink->next) {
1242: putkey(mpage, mlink->name, NAME_FILE);
1243: if (warnings && !use_all)
1.131 schwarze 1244: mlink_check(mpage, mlink);
1.238 schwarze 1245: }
1.44 kristaps 1246:
1.221 schwarze 1247: dbadd(dba, mpage);
1.124 schwarze 1248:
1249: nextpage:
1.66 schwarze 1250: ohash_delete(&strings);
1.133 schwarze 1251: ohash_delete(&names);
1.50 kristaps 1252: }
1253: }
1.12 schwarze 1254:
1.50 kristaps 1255: static void
1.124 schwarze 1256: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1257: {
1258: FILE *stream;
1.241 schwarze 1259: struct mlink *mlink;
1260: char *line, *p, *title, *sec;
1.209 schwarze 1261: size_t linesz, plen, titlesz;
1262: ssize_t len;
1263: int offs;
1.12 schwarze 1264:
1.241 schwarze 1265: mlink = mpage->mlinks;
1266: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1267: if (stream == NULL) {
1268: if (fd != -1)
1.159 schwarze 1269: close(fd);
1.56 schwarze 1270: if (warnings)
1.241 schwarze 1271: say(mlink->file, "&fopen");
1.50 kristaps 1272: return;
1273: }
1.1 kristaps 1274:
1.209 schwarze 1275: line = NULL;
1276: linesz = 0;
1277:
1.241 schwarze 1278: /* Parse the section number from the header line. */
1279:
1280: while (getline(&line, &linesz, stream) != -1) {
1281: if (*line == '\n')
1282: continue;
1283: if ((sec = strchr(line, '(')) == NULL)
1284: break;
1285: if ((p = strchr(++sec, ')')) == NULL)
1286: break;
1287: free(mpage->sec);
1288: mpage->sec = mandoc_strndup(sec, p - sec);
1289: if (warnings && *mlink->dsec != '\0' &&
1290: strcasecmp(mpage->sec, mlink->dsec))
1291: say(mlink->file,
1292: "Section \"%s\" manual in %s directory",
1293: mpage->sec, mlink->dsec);
1294: break;
1295: }
1296:
1.50 kristaps 1297: /* Skip to first blank line. */
1.1 kristaps 1298:
1.241 schwarze 1299: while (line == NULL || *line != '\n')
1300: if (getline(&line, &linesz, stream) == -1)
1.50 kristaps 1301: break;
1.1 kristaps 1302:
1.50 kristaps 1303: /*
1304: * Assume the first line that is not indented
1305: * is the first section header. Skip to it.
1306: */
1.1 kristaps 1307:
1.209 schwarze 1308: while (getline(&line, &linesz, stream) != -1)
1309: if (*line != '\n' && *line != ' ')
1.50 kristaps 1310: break;
1.141 schwarze 1311:
1.50 kristaps 1312: /*
1313: * Read up until the next section into a buffer.
1314: * Strip the leading and trailing newline from each read line,
1315: * appending a trailing space.
1316: * Ignore empty (whitespace-only) lines.
1317: */
1.1 kristaps 1318:
1.50 kristaps 1319: titlesz = 0;
1320: title = NULL;
1.38 schwarze 1321:
1.209 schwarze 1322: while ((len = getline(&line, &linesz, stream)) != -1) {
1323: if (*line != ' ')
1.50 kristaps 1324: break;
1.209 schwarze 1325: offs = 0;
1326: while (isspace((unsigned char)line[offs]))
1327: offs++;
1328: if (line[offs] == '\0')
1.38 schwarze 1329: continue;
1.209 schwarze 1330: title = mandoc_realloc(title, titlesz + len - offs);
1331: memcpy(title + titlesz, line + offs, len - offs);
1332: titlesz += len - offs;
1.50 kristaps 1333: title[titlesz - 1] = ' ';
1334: }
1.209 schwarze 1335: free(line);
1.38 schwarze 1336:
1.50 kristaps 1337: /*
1338: * If no page content can be found, or the input line
1339: * is already the next section header, or there is no
1340: * trailing newline, reuse the page title as the page
1341: * description.
1342: */
1.44 kristaps 1343:
1.50 kristaps 1344: if (NULL == title || '\0' == *title) {
1.56 schwarze 1345: if (warnings)
1.241 schwarze 1346: say(mlink->file, "Cannot find NAME section");
1.50 kristaps 1347: fclose(stream);
1348: free(title);
1349: return;
1350: }
1.1 kristaps 1351:
1.209 schwarze 1352: title[titlesz - 1] = '\0';
1.33 schwarze 1353:
1.50 kristaps 1354: /*
1355: * Skip to the first dash.
1356: * Use the remaining line as the description (no more than 70
1357: * bytes).
1358: */
1.33 schwarze 1359:
1.50 kristaps 1360: if (NULL != (p = strstr(title, "- "))) {
1361: for (p += 2; ' ' == *p || '\b' == *p; p++)
1362: /* Skip to next word. */ ;
1363: } else {
1.56 schwarze 1364: if (warnings)
1.242 schwarze 1365: say(mlink->file, "No dash in title line, "
1366: "reusing \"%s\" as one-line description", title);
1.50 kristaps 1367: p = title;
1368: }
1.38 schwarze 1369:
1.50 kristaps 1370: plen = strlen(p);
1.1 kristaps 1371:
1.50 kristaps 1372: /* Strip backspace-encoding from line. */
1.1 kristaps 1373:
1.50 kristaps 1374: while (NULL != (line = memchr(p, '\b', plen))) {
1375: len = line - p;
1376: if (0 == len) {
1377: memmove(line, line + 1, plen--);
1378: continue;
1.141 schwarze 1379: }
1.50 kristaps 1380: memmove(line - 1, line + 1, plen - len);
1381: plen -= 2;
1382: }
1.1 kristaps 1383:
1.78 schwarze 1384: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1385: fclose(stream);
1386: free(title);
1387: }
1.1 kristaps 1388:
1.50 kristaps 1389: /*
1390: * Put a type/word pair into the word database for this particular file.
1391: */
1392: static void
1.112 schwarze 1393: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1394: {
1.78 schwarze 1395: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1396: }
1397:
1398: /*
1.50 kristaps 1399: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1400: */
1401: static void
1.78 schwarze 1402: putmdockey(const struct mpage *mpage,
1.235 schwarze 1403: const struct roff_node *n, uint64_t m, int taboo)
1.3 kristaps 1404: {
1.18 kristaps 1405:
1.50 kristaps 1406: for ( ; NULL != n; n = n->next) {
1.235 schwarze 1407: if (n->flags & taboo)
1408: continue;
1.50 kristaps 1409: if (NULL != n->child)
1.235 schwarze 1410: putmdockey(mpage, n->child, m, taboo);
1.188 schwarze 1411: if (n->type == ROFFT_TEXT)
1.78 schwarze 1412: putkey(mpage, n->string, m);
1.50 kristaps 1413: }
1414: }
1.18 kristaps 1415:
1.61 schwarze 1416: static void
1.190 schwarze 1417: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1418: const struct roff_node *n)
1.50 kristaps 1419: {
1.189 schwarze 1420: const struct roff_node *head, *body;
1.121 schwarze 1421: char *start, *title;
1.50 kristaps 1422: char byte;
1.121 schwarze 1423: size_t sz;
1.18 kristaps 1424:
1.215 schwarze 1425: if (n == NULL)
1.61 schwarze 1426: return;
1.18 kristaps 1427:
1.50 kristaps 1428: /*
1429: * We're only searching for one thing: the first text child in
1430: * the BODY of a NAME section. Since we don't keep track of
1431: * sections in -man, run some hoops to find out whether we're in
1432: * the correct section or not.
1433: */
1.18 kristaps 1434:
1.188 schwarze 1435: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1436: body = n;
1.215 schwarze 1437: if ((head = body->parent->head) != NULL &&
1438: (head = head->child) != NULL &&
1439: head->next == NULL &&
1.188 schwarze 1440: head->type == ROFFT_TEXT &&
1.215 schwarze 1441: strcmp(head->string, "NAME") == 0 &&
1442: body->child != NULL) {
1.3 kristaps 1443:
1.50 kristaps 1444: /*
1445: * Suck the entire NAME section into memory.
1446: * Yes, we might run away.
1447: * But too many manuals have big, spread-out
1448: * NAME sections over many lines.
1449: */
1.3 kristaps 1450:
1.121 schwarze 1451: title = NULL;
1.194 schwarze 1452: deroff(&title, body);
1.50 kristaps 1453: if (NULL == title)
1.61 schwarze 1454: return;
1.18 kristaps 1455:
1.141 schwarze 1456: /*
1.50 kristaps 1457: * Go through a special heuristic dance here.
1458: * Conventionally, one or more manual names are
1459: * comma-specified prior to a whitespace, then a
1460: * dash, then a description. Try to puzzle out
1461: * the name parts here.
1462: */
1.18 kristaps 1463:
1.121 schwarze 1464: start = title;
1.50 kristaps 1465: for ( ;; ) {
1466: sz = strcspn(start, " ,");
1467: if ('\0' == start[sz])
1468: break;
1.1 kristaps 1469:
1.50 kristaps 1470: byte = start[sz];
1471: start[sz] = '\0';
1.110 schwarze 1472:
1473: /*
1474: * Assume a stray trailing comma in the
1475: * name list if a name begins with a dash.
1476: */
1477:
1478: if ('-' == start[0] ||
1479: ('\\' == start[0] && '-' == start[1]))
1480: break;
1.1 kristaps 1481:
1.133 schwarze 1482: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1483: if ( ! (mpage->name_head_done ||
1484: strcasecmp(start, meta->title))) {
1485: putkey(mpage, start, NAME_HEAD);
1486: mpage->name_head_done = 1;
1487: }
1.1 kristaps 1488:
1.50 kristaps 1489: if (' ' == byte) {
1490: start += sz + 1;
1491: break;
1492: }
1.1 kristaps 1493:
1.50 kristaps 1494: assert(',' == byte);
1495: start += sz + 1;
1496: while (' ' == *start)
1497: start++;
1498: }
1.1 kristaps 1499:
1.121 schwarze 1500: if (start == title) {
1.133 schwarze 1501: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1502: if ( ! (mpage->name_head_done ||
1503: strcasecmp(start, meta->title))) {
1504: putkey(mpage, start, NAME_HEAD);
1505: mpage->name_head_done = 1;
1506: }
1.50 kristaps 1507: free(title);
1.61 schwarze 1508: return;
1.50 kristaps 1509: }
1.1 kristaps 1510:
1.50 kristaps 1511: while (isspace((unsigned char)*start))
1512: start++;
1.1 kristaps 1513:
1.50 kristaps 1514: if (0 == strncmp(start, "-", 1))
1515: start += 1;
1516: else if (0 == strncmp(start, "\\-\\-", 4))
1517: start += 4;
1518: else if (0 == strncmp(start, "\\-", 2))
1519: start += 2;
1520: else if (0 == strncmp(start, "\\(en", 4))
1521: start += 4;
1522: else if (0 == strncmp(start, "\\(em", 4))
1523: start += 4;
1.1 kristaps 1524:
1.50 kristaps 1525: while (' ' == *start)
1526: start++;
1.1 kristaps 1527:
1.78 schwarze 1528: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1529: free(title);
1.61 schwarze 1530: return;
1.50 kristaps 1531: }
1532: }
1.1 kristaps 1533:
1.77 schwarze 1534: for (n = n->child; n; n = n->next) {
1.78 schwarze 1535: if (NULL != mpage->desc)
1.77 schwarze 1536: break;
1.174 schwarze 1537: parse_man(mpage, meta, n);
1.77 schwarze 1538: }
1.1 kristaps 1539: }
1540:
1541: static void
1.190 schwarze 1542: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1543: const struct roff_node *n)
1.1 kristaps 1544: {
1545:
1.246 schwarze 1546: for (n = n->child; n != NULL; n = n->next) {
1.247 schwarze 1547: if (n->tok == TOKEN_NONE ||
1548: n->tok < ROFF_MAX ||
1549: n->flags & mdocs[n->tok].taboo)
1.235 schwarze 1550: continue;
1.247 schwarze 1551: assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1.50 kristaps 1552: switch (n->type) {
1.188 schwarze 1553: case ROFFT_ELEM:
1554: case ROFFT_BLOCK:
1555: case ROFFT_HEAD:
1556: case ROFFT_BODY:
1557: case ROFFT_TAIL:
1.246 schwarze 1558: if (mdocs[n->tok].fp != NULL &&
1559: (*mdocs[n->tok].fp)(mpage, meta, n) == 0)
1560: break;
1.68 schwarze 1561: if (mdocs[n->tok].mask)
1.78 schwarze 1562: putmdockey(mpage, n->child,
1.235 schwarze 1563: mdocs[n->tok].mask, mdocs[n->tok].taboo);
1.50 kristaps 1564: break;
1565: default:
1566: continue;
1567: }
1568: if (NULL != n->child)
1.172 schwarze 1569: parse_mdoc(mpage, meta, n);
1.1 kristaps 1570: }
1571: }
1572:
1.25 schwarze 1573: static int
1.190 schwarze 1574: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1575: const struct roff_node *n)
1.1 kristaps 1576: {
1.176 schwarze 1577: char *start, *end;
1.1 kristaps 1578: size_t sz;
1.25 schwarze 1579:
1.50 kristaps 1580: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1581: NULL == (n = n->child) ||
1.188 schwarze 1582: n->type != ROFFT_TEXT)
1.197 schwarze 1583: return 0;
1.1 kristaps 1584:
1585: /*
1586: * Only consider those `Fd' macro fields that begin with an
1587: * "inclusion" token (versus, e.g., #define).
1588: */
1.50 kristaps 1589:
1.1 kristaps 1590: if (strcmp("#include", n->string))
1.197 schwarze 1591: return 0;
1.1 kristaps 1592:
1.188 schwarze 1593: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1594: return 0;
1.1 kristaps 1595:
1596: /*
1597: * Strip away the enclosing angle brackets and make sure we're
1598: * not zero-length.
1599: */
1600:
1601: start = n->string;
1602: if ('<' == *start || '"' == *start)
1603: start++;
1604:
1605: if (0 == (sz = strlen(start)))
1.197 schwarze 1606: return 0;
1.1 kristaps 1607:
1608: end = &start[(int)sz - 1];
1609: if ('>' == *end || '"' == *end)
1610: end--;
1611:
1.50 kristaps 1612: if (end > start)
1.78 schwarze 1613: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1614: return 0;
1.1 kristaps 1615: }
1616:
1.178 schwarze 1617: static void
1.189 schwarze 1618: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1619: {
1.112 schwarze 1620: char *cp;
1.178 schwarze 1621: size_t sz;
1.1 kristaps 1622:
1.188 schwarze 1623: if (n->type != ROFFT_TEXT)
1.178 schwarze 1624: return;
1.25 schwarze 1625:
1.178 schwarze 1626: /* Skip function pointer punctuation. */
1.1 kristaps 1627:
1.178 schwarze 1628: cp = n->string;
1629: while (*cp == '(' || *cp == '*')
1.1 kristaps 1630: cp++;
1.178 schwarze 1631: sz = strcspn(cp, "()");
1.1 kristaps 1632:
1.178 schwarze 1633: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1634: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1635: putkeys(mpage, cp, sz, NAME_SYN);
1636: }
1.25 schwarze 1637:
1.178 schwarze 1638: static int
1.190 schwarze 1639: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1640: const struct roff_node *n)
1.178 schwarze 1641: {
1642:
1643: if (n->child == NULL)
1.197 schwarze 1644: return 0;
1.178 schwarze 1645:
1646: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1647:
1.178 schwarze 1648: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1649: if (n->type == ROFFT_TEXT)
1.78 schwarze 1650: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1651:
1.197 schwarze 1652: return 0;
1.173 schwarze 1653: }
1654:
1655: static int
1.190 schwarze 1656: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1657: const struct roff_node *n)
1.173 schwarze 1658: {
1.177 schwarze 1659:
1.188 schwarze 1660: if (n->type != ROFFT_HEAD)
1.197 schwarze 1661: return 1;
1.173 schwarze 1662:
1.178 schwarze 1663: if (n->child != NULL)
1664: parse_mdoc_fname(mpage, n->child);
1665:
1.197 schwarze 1666: return 0;
1.1 kristaps 1667: }
1668:
1.25 schwarze 1669: static int
1.211 schwarze 1670: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1671: const struct roff_node *n)
1672: {
1673: char *cp;
1674:
1675: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1676: return 0;
1677:
1.215 schwarze 1678: if (n->child != NULL &&
1679: n->child->next == NULL &&
1680: n->child->type == ROFFT_TEXT)
1.211 schwarze 1681: return 1;
1682:
1683: cp = NULL;
1684: deroff(&cp, n);
1685: if (cp != NULL) {
1686: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1687: n->type == ROFFT_BODY ? TYPE_Va : 0));
1688: free(cp);
1689: }
1690:
1691: return 0;
1692: }
1693:
1694: static int
1.190 schwarze 1695: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1696: const struct roff_node *n)
1.1 kristaps 1697: {
1.67 schwarze 1698: char *cp;
1.1 kristaps 1699:
1700: if (NULL == (n = n->child))
1.197 schwarze 1701: return 0;
1.1 kristaps 1702:
1.67 schwarze 1703: if (NULL == n->next) {
1.78 schwarze 1704: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1705: return 0;
1.67 schwarze 1706: }
1707:
1.120 schwarze 1708: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1709: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1710: free(cp);
1.197 schwarze 1711: return 0;
1.1 kristaps 1712: }
1713:
1.25 schwarze 1714: static int
1.190 schwarze 1715: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1716: const struct roff_node *n)
1.1 kristaps 1717: {
1718:
1.188 schwarze 1719: if (n->type == ROFFT_BODY)
1.194 schwarze 1720: deroff(&mpage->desc, n);
1.197 schwarze 1721: return 0;
1.1 kristaps 1722: }
1723:
1.25 schwarze 1724: static int
1.190 schwarze 1725: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1726: const struct roff_node *n)
1.1 kristaps 1727: {
1728:
1.129 schwarze 1729: if (SEC_NAME == n->sec)
1.235 schwarze 1730: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.188 schwarze 1731: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1732: if (n->child == NULL)
1733: putkey(mpage, meta->name, NAME_SYN);
1734: else
1.235 schwarze 1735: putmdockey(mpage, n->child, NAME_SYN, 0);
1.174 schwarze 1736: }
1737: if ( ! (mpage->name_head_done ||
1738: n->child == NULL || n->child->string == NULL ||
1739: strcasecmp(n->child->string, meta->title))) {
1.226 schwarze 1740: putkey(mpage, n->child->string, NAME_HEAD);
1.174 schwarze 1741: mpage->name_head_done = 1;
1.172 schwarze 1742: }
1.197 schwarze 1743: return 0;
1.1 kristaps 1744: }
1745:
1.25 schwarze 1746: static int
1.190 schwarze 1747: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1748: const struct roff_node *n)
1.1 kristaps 1749: {
1750:
1.197 schwarze 1751: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1752: }
1753:
1.50 kristaps 1754: static int
1.190 schwarze 1755: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1756: const struct roff_node *n)
1.1 kristaps 1757: {
1758:
1.197 schwarze 1759: return n->type == ROFFT_HEAD;
1.1 kristaps 1760: }
1761:
1.50 kristaps 1762: /*
1.66 schwarze 1763: * Add a string to the hash table for the current manual.
1764: * Each string has a bitmask telling which macros it belongs to.
1765: * When we finish the manual, we'll dump the table.
1.50 kristaps 1766: */
1767: static void
1.176 schwarze 1768: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1769: {
1.133 schwarze 1770: struct ohash *htab;
1.50 kristaps 1771: struct str *s;
1.111 schwarze 1772: const char *end;
1.65 schwarze 1773: unsigned int slot;
1.176 schwarze 1774: int i, mustfree;
1.25 schwarze 1775:
1.50 kristaps 1776: if (0 == sz)
1777: return;
1.111 schwarze 1778:
1.176 schwarze 1779: mustfree = render_string(&cp, &sz);
1780:
1.133 schwarze 1781: if (TYPE_Nm & v) {
1782: htab = &names;
1783: v &= name_mask;
1.169 schwarze 1784: if (v & NAME_FIRST)
1785: name_mask &= ~NAME_FIRST;
1.133 schwarze 1786: if (debug > 1)
1787: say(mpage->mlinks->file,
1.234 schwarze 1788: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1789: (unsigned long long)v);
1.133 schwarze 1790: } else {
1791: htab = &strings;
1792: if (debug > 1)
1.221 schwarze 1793: for (i = 0; i < KEY_MAX; i++)
1.163 schwarze 1794: if ((uint64_t)1 << i & v)
1.133 schwarze 1795: say(mpage->mlinks->file,
1796: "Adding key %s=%*s",
1.220 schwarze 1797: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1798: }
1.25 schwarze 1799:
1.65 schwarze 1800: end = cp + sz;
1.133 schwarze 1801: slot = ohash_qlookupi(htab, cp, &end);
1802: s = ohash_find(htab, slot);
1.25 schwarze 1803:
1.78 schwarze 1804: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1805: s->mask |= v;
1806: return;
1807: } else if (NULL == s) {
1.144 schwarze 1808: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1809: memcpy(s->key, cp, sz);
1.133 schwarze 1810: ohash_insert(htab, slot, s);
1.1 kristaps 1811: }
1.78 schwarze 1812: s->mpage = mpage;
1.50 kristaps 1813: s->mask = v;
1.176 schwarze 1814:
1815: if (mustfree)
1816: free(cp);
1.1 kristaps 1817: }
1818:
1.50 kristaps 1819: /*
1820: * Take a Unicode codepoint and produce its UTF-8 encoding.
1821: * This isn't the best way to do this, but it works.
1822: * The magic numbers are from the UTF-8 packaging.
1823: * They're not as scary as they seem: read the UTF-8 spec for details.
1824: */
1825: static size_t
1826: utf8(unsigned int cp, char out[7])
1.1 kristaps 1827: {
1.50 kristaps 1828: size_t rc;
1.1 kristaps 1829:
1.50 kristaps 1830: rc = 0;
1831: if (cp <= 0x0000007F) {
1832: rc = 1;
1833: out[0] = (char)cp;
1834: } else if (cp <= 0x000007FF) {
1835: rc = 2;
1836: out[0] = (cp >> 6 & 31) | 192;
1837: out[1] = (cp & 63) | 128;
1838: } else if (cp <= 0x0000FFFF) {
1839: rc = 3;
1840: out[0] = (cp >> 12 & 15) | 224;
1841: out[1] = (cp >> 6 & 63) | 128;
1842: out[2] = (cp & 63) | 128;
1843: } else if (cp <= 0x001FFFFF) {
1844: rc = 4;
1845: out[0] = (cp >> 18 & 7) | 240;
1846: out[1] = (cp >> 12 & 63) | 128;
1847: out[2] = (cp >> 6 & 63) | 128;
1848: out[3] = (cp & 63) | 128;
1849: } else if (cp <= 0x03FFFFFF) {
1850: rc = 5;
1851: out[0] = (cp >> 24 & 3) | 248;
1852: out[1] = (cp >> 18 & 63) | 128;
1853: out[2] = (cp >> 12 & 63) | 128;
1854: out[3] = (cp >> 6 & 63) | 128;
1855: out[4] = (cp & 63) | 128;
1856: } else if (cp <= 0x7FFFFFFF) {
1857: rc = 6;
1858: out[0] = (cp >> 30 & 1) | 252;
1859: out[1] = (cp >> 24 & 63) | 128;
1860: out[2] = (cp >> 18 & 63) | 128;
1861: out[3] = (cp >> 12 & 63) | 128;
1862: out[4] = (cp >> 6 & 63) | 128;
1863: out[5] = (cp & 63) | 128;
1864: } else
1.197 schwarze 1865: return 0;
1.1 kristaps 1866:
1.50 kristaps 1867: out[rc] = '\0';
1.197 schwarze 1868: return rc;
1.50 kristaps 1869: }
1.1 kristaps 1870:
1.50 kristaps 1871: /*
1.176 schwarze 1872: * If the string contains escape sequences,
1873: * replace it with an allocated rendering and return 1,
1874: * such that the caller can free it after use.
1875: * Otherwise, do nothing and return 0.
1.50 kristaps 1876: */
1.176 schwarze 1877: static int
1878: render_string(char **public, size_t *psz)
1.50 kristaps 1879: {
1.176 schwarze 1880: const char *src, *scp, *addcp, *seq;
1881: char *dst;
1882: size_t ssz, dsz, addsz;
1.114 schwarze 1883: char utfbuf[7], res[6];
1.176 schwarze 1884: int seqlen, unicode;
1.50 kristaps 1885:
1886: res[0] = '\\';
1887: res[1] = '\t';
1888: res[2] = ASCII_NBRSP;
1889: res[3] = ASCII_HYPH;
1.114 schwarze 1890: res[4] = ASCII_BREAK;
1891: res[5] = '\0';
1.1 kristaps 1892:
1.176 schwarze 1893: src = scp = *public;
1894: ssz = *psz;
1895: dst = NULL;
1896: dsz = 0;
1897:
1898: while (scp < src + *psz) {
1899:
1900: /* Leave normal characters unchanged. */
1901:
1902: if (strchr(res, *scp) == NULL) {
1903: if (dst != NULL)
1904: dst[dsz++] = *scp;
1905: scp++;
1906: continue;
1907: }
1.46 kristaps 1908:
1.50 kristaps 1909: /*
1.176 schwarze 1910: * Found something that requires replacing,
1911: * make sure we have a destination buffer.
1.50 kristaps 1912: */
1.176 schwarze 1913:
1914: if (dst == NULL) {
1915: dst = mandoc_malloc(ssz + 1);
1916: dsz = scp - src;
1917: memcpy(dst, src, dsz);
1.50 kristaps 1918: }
1.46 kristaps 1919:
1.176 schwarze 1920: /* Handle single-char special characters. */
1921:
1922: switch (*scp) {
1923: case '\\':
1924: break;
1.141 schwarze 1925: case '\t':
1926: case ASCII_NBRSP:
1.176 schwarze 1927: dst[dsz++] = ' ';
1928: scp++;
1929: continue;
1930: case ASCII_HYPH:
1931: dst[dsz++] = '-';
1.114 schwarze 1932: /* FALLTHROUGH */
1.141 schwarze 1933: case ASCII_BREAK:
1.176 schwarze 1934: scp++;
1.50 kristaps 1935: continue;
1.114 schwarze 1936: default:
1.176 schwarze 1937: abort();
1.114 schwarze 1938: }
1.46 kristaps 1939:
1.50 kristaps 1940: /*
1.176 schwarze 1941: * Found an escape sequence.
1942: * Read past the slash, then parse it.
1943: * Ignore everything except characters.
1.50 kristaps 1944: */
1.95 schwarze 1945:
1.176 schwarze 1946: scp++;
1947: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1948: continue;
1.1 kristaps 1949:
1.50 kristaps 1950: /*
1.95 schwarze 1951: * Render the special character
1952: * as either UTF-8 or ASCII.
1.50 kristaps 1953: */
1.95 schwarze 1954:
1955: if (write_utf8) {
1.203 schwarze 1956: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1957: if (unicode <= 0)
1.95 schwarze 1958: continue;
1.176 schwarze 1959: addsz = utf8(unicode, utfbuf);
1960: if (addsz == 0)
1.95 schwarze 1961: continue;
1.176 schwarze 1962: addcp = utfbuf;
1.95 schwarze 1963: } else {
1.203 schwarze 1964: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1965: if (addcp == NULL)
1.95 schwarze 1966: continue;
1.176 schwarze 1967: if (*addcp == ASCII_NBRSP) {
1968: addcp = " ";
1969: addsz = 1;
1.95 schwarze 1970: }
1971: }
1.1 kristaps 1972:
1.50 kristaps 1973: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1974:
1.176 schwarze 1975: ssz += addsz;
1976: dst = mandoc_realloc(dst, ssz + 1);
1977: memcpy(dst + dsz, addcp, addsz);
1978: dsz += addsz;
1979: }
1980: if (dst != NULL) {
1981: *public = dst;
1982: *psz = dsz;
1983: }
1984:
1985: /* Trim trailing whitespace and NUL-terminate. */
1986:
1987: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1988: --*psz;
1989: if (dst != NULL) {
1990: (*public)[*psz] = '\0';
1.197 schwarze 1991: return 1;
1.176 schwarze 1992: } else
1.197 schwarze 1993: return 0;
1.1 kristaps 1994: }
1995:
1.118 schwarze 1996: static void
1997: dbadd_mlink(const struct mlink *mlink)
1998: {
1.221 schwarze 1999: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
2000: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1.223 schwarze 2001: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1.221 schwarze 2002: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
2003: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.118 schwarze 2004: }
2005:
1.14 schwarze 2006: /*
1.50 kristaps 2007: * Flush the current page's terms (and their bits) into the database.
1.96 schwarze 2008: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2009: */
2010: static void
1.221 schwarze 2011: dbadd(struct dba *dba, struct mpage *mpage)
1.14 schwarze 2012: {
1.87 schwarze 2013: struct mlink *mlink;
1.50 kristaps 2014: struct str *key;
1.176 schwarze 2015: char *cp;
1.221 schwarze 2016: uint64_t mask;
1.52 kristaps 2017: size_t i;
1.66 schwarze 2018: unsigned int slot;
1.176 schwarze 2019: int mustfree;
1.14 schwarze 2020:
1.128 schwarze 2021: mlink = mpage->mlinks;
1.43 kristaps 2022:
1.128 schwarze 2023: if (nodb) {
1.147 schwarze 2024: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2025: key = ohash_next(&names, &slot))
1.147 schwarze 2026: free(key);
2027: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2028: key = ohash_next(&strings, &slot))
1.147 schwarze 2029: free(key);
1.145 schwarze 2030: if (0 == debug)
2031: return;
1.128 schwarze 2032: while (NULL != mlink) {
2033: fputs(mlink->name, stdout);
2034: if (NULL == mlink->next ||
2035: strcmp(mlink->dsec, mlink->next->dsec) ||
2036: strcmp(mlink->fsec, mlink->next->fsec) ||
2037: strcmp(mlink->arch, mlink->next->arch)) {
2038: putchar('(');
2039: if ('\0' == *mlink->dsec)
2040: fputs(mlink->fsec, stdout);
2041: else
2042: fputs(mlink->dsec, stdout);
2043: if ('\0' != *mlink->arch)
2044: printf("/%s", mlink->arch);
2045: putchar(')');
2046: }
2047: mlink = mlink->next;
2048: if (NULL != mlink)
2049: fputs(", ", stdout);
2050: }
1.132 schwarze 2051: printf(" - %s\n", mpage->desc);
1.14 schwarze 2052: return;
1.128 schwarze 2053: }
2054:
2055: if (debug)
2056: say(mlink->file, "Adding to database");
1.28 kristaps 2057:
1.176 schwarze 2058: cp = mpage->desc;
2059: i = strlen(cp);
2060: mustfree = render_string(&cp, &i);
1.227 schwarze 2061: mpage->dba = dba_page_new(dba->pages,
1.223 schwarze 2062: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2063: cp, mlink->file, mpage->form);
1.176 schwarze 2064: if (mustfree)
2065: free(cp);
1.227 schwarze 2066: dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
1.81 schwarze 2067:
1.223 schwarze 2068: while (mlink != NULL) {
1.118 schwarze 2069: dbadd_mlink(mlink);
1.223 schwarze 2070: mlink = mlink->next;
2071: }
1.50 kristaps 2072:
1.133 schwarze 2073: for (key = ohash_first(&names, &slot); NULL != key;
2074: key = ohash_next(&names, &slot)) {
2075: assert(key->mpage == mpage);
1.221 schwarze 2076: dba_page_alias(mpage->dba, key->key, key->mask);
1.133 schwarze 2077: free(key);
2078: }
1.66 schwarze 2079: for (key = ohash_first(&strings, &slot); NULL != key;
2080: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2081: assert(key->mpage == mpage);
1.221 schwarze 2082: i = 0;
2083: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2084: if (key->mask & mask)
2085: dba_macro_add(dba->macros, i,
2086: key->key, mpage->dba);
2087: i++;
2088: }
1.66 schwarze 2089: free(key);
1.38 schwarze 2090: }
1.14 schwarze 2091: }
2092:
1.5 kristaps 2093: static void
1.221 schwarze 2094: dbprune(struct dba *dba)
1.5 kristaps 2095: {
1.221 schwarze 2096: struct dba_array *page, *files;
2097: char *file;
1.5 kristaps 2098:
1.221 schwarze 2099: dba_array_FOREACH(dba->pages, page) {
2100: files = dba_array_get(page, DBP_FILE);
2101: dba_array_FOREACH(files, file) {
2102: if (*file < ' ')
2103: file++;
2104: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2105: file)) != NULL) {
2106: if (debug)
2107: say(file, "Deleting from database");
2108: dba_array_del(dba->pages);
2109: break;
2110: }
1.106 schwarze 2111: }
1.5 kristaps 2112: }
2113: }
2114:
1.4 kristaps 2115: /*
1.221 schwarze 2116: * Write the database from memory to disk.
1.4 kristaps 2117: */
1.35 kristaps 2118: static void
1.221 schwarze 2119: dbwrite(struct dba *dba)
1.4 kristaps 2120: {
1.221 schwarze 2121: char tfn[32];
1.115 schwarze 2122: int status;
2123: pid_t child;
1.4 kristaps 2124:
1.250 ! schwarze 2125: /*
! 2126: * Do not write empty databases, and delete existing ones
! 2127: * when makewhatis -u causes them to become empty.
! 2128: */
! 2129:
! 2130: dba_array_start(dba->pages);
! 2131: if (dba_array_next(dba->pages) == NULL) {
! 2132: if (unlink(MANDOC_DB) == -1)
! 2133: say(MANDOC_DB, "&unlink");
! 2134: return;
! 2135: }
! 2136:
! 2137: /*
! 2138: * Build the database in a temporary file,
! 2139: * then atomically move it into place.
! 2140: */
! 2141:
1.221 schwarze 2142: if (dba_write(MANDOC_DB "~", dba) != -1) {
2143: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2144: exitcode = (int)MANDOCLEVEL_SYSERR;
2145: say(MANDOC_DB, "&rename");
2146: unlink(MANDOC_DB "~");
2147: }
1.38 schwarze 2148: return;
1.4 kristaps 2149: }
1.250 ! schwarze 2150:
! 2151: /*
! 2152: * We lack write permission and cannot replace the database
! 2153: * file, but let's at least check whether the data changed.
! 2154: */
1.4 kristaps 2155:
1.221 schwarze 2156: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2157: if (mkdtemp(tfn) == NULL) {
2158: exitcode = (int)MANDOCLEVEL_SYSERR;
2159: say("", "&%s", tfn);
1.50 kristaps 2160: return;
1.221 schwarze 2161: }
1.12 schwarze 2162:
1.221 schwarze 2163: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2164: if (dba_write(tfn, dba) == -1) {
2165: exitcode = (int)MANDOCLEVEL_SYSERR;
2166: say(tfn, "&dba_write");
2167: goto out;
1.115 schwarze 2168: }
2169:
2170: switch (child = fork()) {
1.141 schwarze 2171: case -1:
1.115 schwarze 2172: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2173: say("", "&fork cmp");
1.115 schwarze 2174: return;
1.141 schwarze 2175: case 0:
1.221 schwarze 2176: execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL);
1.123 schwarze 2177: say("", "&exec cmp");
1.115 schwarze 2178: exit(0);
2179: default:
2180: break;
2181: }
1.221 schwarze 2182: if (waitpid(child, &status, 0) == -1) {
1.115 schwarze 2183: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2184: say("", "&wait cmp");
1.115 schwarze 2185: } else if (WIFSIGNALED(status)) {
2186: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2187: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2188: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2189: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2190: say(MANDOC_DB,
2191: "Data changed, but cannot replace database");
2192: }
2193:
1.221 schwarze 2194: out:
2195: *strrchr(tfn, '/') = '\0';
1.115 schwarze 2196: switch (child = fork()) {
1.141 schwarze 2197: case -1:
1.115 schwarze 2198: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2199: say("", "&fork rm");
1.115 schwarze 2200: return;
1.141 schwarze 2201: case 0:
1.221 schwarze 2202: execlp("rm", "rm", "-rf", tfn, (char *)NULL);
1.123 schwarze 2203: say("", "&exec rm");
1.115 schwarze 2204: exit((int)MANDOCLEVEL_SYSERR);
2205: default:
2206: break;
2207: }
1.221 schwarze 2208: if (waitpid(child, &status, 0) == -1) {
1.115 schwarze 2209: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2210: say("", "&wait rm");
1.115 schwarze 2211: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.221 schwarze 2213: say("", "%s: Cannot remove temporary directory", tfn);
1.146 schwarze 2214: }
1.4 kristaps 2215: }
2216:
1.50 kristaps 2217: static int
1.165 schwarze 2218: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2219: {
1.59 schwarze 2220: static char startdir[PATH_MAX];
1.151 schwarze 2221: static int getcwd_status; /* 1 = ok, 2 = failure */
2222: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2223: char *cp;
1.4 kristaps 2224:
1.59 schwarze 2225: /*
1.151 schwarze 2226: * Remember the original working directory, if possible.
2227: * This will be needed if the second or a later directory
2228: * on the command line is given as a relative path.
2229: * Do not error out if the current directory is not
2230: * searchable: Maybe it won't be needed after all.
2231: */
2232: if (0 == getcwd_status) {
2233: if (NULL == getcwd(startdir, sizeof(startdir))) {
2234: getcwd_status = 2;
2235: (void)strlcpy(startdir, strerror(errno),
2236: sizeof(startdir));
2237: } else
2238: getcwd_status = 1;
2239: }
2240:
2241: /*
2242: * We are leaving the old base directory.
2243: * Do not use it any longer, not even for messages.
2244: */
2245: *basedir = '\0';
2246:
2247: /*
2248: * If and only if the directory was changed earlier and
2249: * the next directory to process is given as a relative path,
2250: * first go back, or bail out if that is impossible.
1.59 schwarze 2251: */
1.151 schwarze 2252: if (chdir_status && '/' != *targetdir) {
2253: if (2 == getcwd_status) {
1.59 schwarze 2254: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2255: say("", "getcwd: %s", startdir);
1.197 schwarze 2256: return 0;
1.59 schwarze 2257: }
1.151 schwarze 2258: if (-1 == chdir(startdir)) {
1.59 schwarze 2259: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2260: say("", "&chdir %s", startdir);
1.197 schwarze 2261: return 0;
1.59 schwarze 2262: }
2263: }
1.151 schwarze 2264:
2265: /*
2266: * Always resolve basedir to the canonicalized absolute
2267: * pathname and append a trailing slash, such that
2268: * we can reliably check whether files are inside.
2269: */
1.59 schwarze 2270: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2271: if (report_baddir || errno != ENOENT) {
2272: exitcode = (int)MANDOCLEVEL_BADARG;
2273: say("", "&%s: realpath", targetdir);
2274: }
1.197 schwarze 2275: return 0;
1.59 schwarze 2276: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2277: if (report_baddir || errno != ENOENT) {
2278: exitcode = (int)MANDOCLEVEL_BADARG;
2279: say("", "&chdir");
2280: }
1.197 schwarze 2281: return 0;
1.150 schwarze 2282: }
1.151 schwarze 2283: chdir_status = 1;
1.150 schwarze 2284: cp = strchr(basedir, '\0');
2285: if ('/' != cp[-1]) {
2286: if (cp - basedir >= PATH_MAX - 1) {
2287: exitcode = (int)MANDOCLEVEL_SYSERR;
2288: say("", "Filename too long");
1.197 schwarze 2289: return 0;
1.150 schwarze 2290: }
2291: *cp++ = '/';
2292: *cp = '\0';
1.4 kristaps 2293: }
1.197 schwarze 2294: return 1;
1.56 schwarze 2295: }
2296:
2297: static void
1.59 schwarze 2298: say(const char *file, const char *format, ...)
1.56 schwarze 2299: {
2300: va_list ap;
1.123 schwarze 2301: int use_errno;
1.56 schwarze 2302:
1.59 schwarze 2303: if ('\0' != *basedir)
2304: fprintf(stderr, "%s", basedir);
2305: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2306: fputc('/', stderr);
1.56 schwarze 2307: if ('\0' != *file)
1.59 schwarze 2308: fprintf(stderr, "%s", file);
2309:
1.123 schwarze 2310: use_errno = 1;
2311: if (NULL != format) {
2312: switch (*format) {
1.141 schwarze 2313: case '&':
1.123 schwarze 2314: format++;
2315: break;
1.141 schwarze 2316: case '\0':
1.123 schwarze 2317: format = NULL;
2318: break;
2319: default:
2320: use_errno = 0;
2321: break;
2322: }
2323: }
2324: if (NULL != format) {
2325: if ('\0' != *basedir || '\0' != *file)
2326: fputs(": ", stderr);
2327: va_start(ap, format);
2328: vfprintf(stderr, format, ap);
2329: va_end(ap);
2330: }
2331: if (use_errno) {
2332: if ('\0' != *basedir || '\0' != *file || NULL != format)
2333: fputs(": ", stderr);
1.59 schwarze 2334: perror(NULL);
1.123 schwarze 2335: } else
2336: fputc('\n', stderr);
1.1 kristaps 2337: }
CVSweb