Annotation of mandoc/mandocdb.c, Revision 1.265
1.265 ! schwarze 1: /* $Id: mandocdb.c,v 1.264 2020/01/25 22:59:22 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.264 schwarze 4: * Copyright (c) 2011-2020 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>
1.255 schwarze 22: #include <sys/mman.h>
1.50 kristaps 23: #include <sys/stat.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"
1.260 schwarze 55: #include "mandoc_parse.h"
1.187 schwarze 56: #include "manconf.h"
1.55 kristaps 57: #include "mansearch.h"
1.221 schwarze 58: #include "dba_array.h"
59: #include "dba.h"
1.1 kristaps 60:
1.111 schwarze 61: extern const char *const mansearch_keynames[];
62:
1.50 kristaps 63: enum op {
64: OP_DEFAULT = 0, /* new dbs from dir list or default config */
65: OP_CONFFILE, /* new databases from custom config file */
66: OP_UPDATE, /* delete/add entries in existing database */
67: OP_DELETE, /* delete entries from existing database */
68: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 69: };
70:
1.50 kristaps 71: struct str {
1.78 schwarze 72: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 73: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 74: char key[]; /* rendered text */
1.38 schwarze 75: };
76:
1.79 schwarze 77: struct inodev {
78: ino_t st_ino;
79: dev_t st_dev;
1.50 kristaps 80: };
1.5 kristaps 81:
1.78 schwarze 82: struct mpage {
1.79 schwarze 83: struct inodev inodev; /* used for hashing routine */
1.221 schwarze 84: struct dba_array *dba;
1.82 schwarze 85: char *sec; /* section from file content */
86: char *arch; /* architecture from file content */
87: char *title; /* title from file content */
88: char *desc; /* description from file content */
1.231 schwarze 89: struct mpage *next; /* singly linked list */
1.82 schwarze 90: struct mlink *mlinks; /* singly linked list */
1.174 schwarze 91: int name_head_done;
1.221 schwarze 92: enum form form; /* format from file content */
1.82 schwarze 93: };
94:
95: struct mlink {
1.58 schwarze 96: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 97: char *dsec; /* section from directory */
98: char *arch; /* architecture from directory */
99: char *name; /* name from file name (not empty) */
100: char *fsec; /* section from file name suffix */
1.85 schwarze 101: struct mlink *next; /* singly linked list */
1.118 schwarze 102: struct mpage *mpage; /* parent */
1.124 schwarze 103: int gzip; /* filename has a .gz suffix */
1.221 schwarze 104: enum form dform; /* format from directory */
105: enum form fform; /* format from file name suffix */
1.1 kristaps 106: };
107:
1.190 schwarze 108: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 109: const struct roff_node *);
1.1 kristaps 110:
1.50 kristaps 111: struct mdoc_handler {
112: mdoc_fp fp; /* optional handler */
113: uint64_t mask; /* set unless handler returns 0 */
1.235 schwarze 114: int taboo; /* node flags that must not be set */
1.1 kristaps 115: };
1.219 schwarze 116:
117:
118: int mandocdb(int, char *[]);
1.1 kristaps 119:
1.221 schwarze 120: static void dbadd(struct dba *, struct mpage *);
1.118 schwarze 121: static void dbadd_mlink(const struct mlink *mlink);
1.221 schwarze 122: static void dbprune(struct dba *);
123: static void dbwrite(struct dba *);
1.59 schwarze 124: static void filescan(const char *);
1.232 schwarze 125: #if HAVE_FTS_COMPARE_CONST
126: static int fts_compare(const FTSENT *const *, const FTSENT *const *);
127: #else
1.231 schwarze 128: static int fts_compare(const FTSENT **, const FTSENT **);
1.232 schwarze 129: #endif
1.84 schwarze 130: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 131: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 132: static void mlink_free(struct mlink *);
1.89 schwarze 133: static void mlinks_undupe(struct mpage *);
1.78 schwarze 134: static void mpages_free(void);
1.221 schwarze 135: static void mpages_merge(struct dba *, struct mparse *);
1.124 schwarze 136: static void parse_cat(struct mpage *, int);
1.190 schwarze 137: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 138: const struct roff_node *);
1.190 schwarze 139: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 140: const struct roff_node *);
1.190 schwarze 141: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 142: const struct roff_node *);
1.258 schwarze 143: static int parse_mdoc_Fa(struct mpage *, const struct roff_meta *,
144: const struct roff_node *);
1.190 schwarze 145: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 146: const struct roff_node *);
147: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 148: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 149: const struct roff_node *);
1.190 schwarze 150: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 151: const struct roff_node *);
1.190 schwarze 152: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 153: const struct roff_node *);
1.190 schwarze 154: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 155: const struct roff_node *);
1.190 schwarze 156: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 157: const struct roff_node *);
1.211 schwarze 158: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
159: const struct roff_node *);
1.190 schwarze 160: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 161: const struct roff_node *);
1.112 schwarze 162: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 163: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 164: static void putmdockey(const struct mpage *,
1.235 schwarze 165: const struct roff_node *, uint64_t, int);
1.176 schwarze 166: static int render_string(char **, size_t *);
1.220 schwarze 167: static void say(const char *, const char *, ...)
1.243 schwarze 168: __attribute__((__format__ (__printf__, 2, 3)));
1.165 schwarze 169: static int set_basedir(const char *, int);
1.59 schwarze 170: static int treescan(void);
1.50 kristaps 171: static size_t utf8(unsigned int, char [7]);
172:
1.102 schwarze 173: static int nodb; /* no database changes */
1.116 schwarze 174: static int mparse_options; /* abort the parse early */
1.141 schwarze 175: static int use_all; /* use all found files */
176: static int debug; /* print what we're doing */
177: static int warnings; /* warn about crap */
1.95 schwarze 178: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 179: static int exitcode; /* to be returned by main */
1.141 schwarze 180: static enum op op; /* operational mode */
1.59 schwarze 181: static char basedir[PATH_MAX]; /* current base directory */
1.264 schwarze 182: static size_t basedir_len; /* strlen(basedir) */
1.231 schwarze 183: static struct mpage *mpage_head; /* list of distinct manual pages */
1.78 schwarze 184: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 185: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 186: static struct ohash names; /* table of all names */
1.50 kristaps 187: static struct ohash strings; /* table of all strings */
1.133 schwarze 188: static uint64_t name_mask;
1.25 schwarze 189:
1.259 schwarze 190: static const struct mdoc_handler mdoc_handlers[MDOC_MAX - MDOC_Dd] = {
1.235 schwarze 191: { NULL, 0, NODE_NOPRT }, /* Dd */
192: { NULL, 0, NODE_NOPRT }, /* Dt */
193: { NULL, 0, NODE_NOPRT }, /* Os */
194: { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
195: { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
196: { NULL, 0, 0 }, /* Pp */
197: { NULL, 0, 0 }, /* D1 */
198: { NULL, 0, 0 }, /* Dl */
199: { NULL, 0, 0 }, /* Bd */
200: { NULL, 0, 0 }, /* Ed */
201: { NULL, 0, 0 }, /* Bl */
202: { NULL, 0, 0 }, /* El */
203: { NULL, 0, 0 }, /* It */
204: { NULL, 0, 0 }, /* Ad */
205: { NULL, TYPE_An, 0 }, /* An */
1.246 schwarze 206: { NULL, 0, 0 }, /* Ap */
1.235 schwarze 207: { NULL, TYPE_Ar, 0 }, /* Ar */
208: { NULL, TYPE_Cd, 0 }, /* Cd */
209: { NULL, TYPE_Cm, 0 }, /* Cm */
210: { NULL, TYPE_Dv, 0 }, /* Dv */
211: { NULL, TYPE_Er, 0 }, /* Er */
212: { NULL, TYPE_Ev, 0 }, /* Ev */
213: { NULL, 0, 0 }, /* Ex */
1.258 schwarze 214: { parse_mdoc_Fa, 0, 0 }, /* Fa */
1.235 schwarze 215: { parse_mdoc_Fd, 0, 0 }, /* Fd */
216: { NULL, TYPE_Fl, 0 }, /* Fl */
217: { parse_mdoc_Fn, 0, 0 }, /* Fn */
1.258 schwarze 218: { NULL, TYPE_Ft | TYPE_Vt, 0 }, /* Ft */
1.235 schwarze 219: { NULL, TYPE_Ic, 0 }, /* Ic */
220: { NULL, TYPE_In, 0 }, /* In */
221: { NULL, TYPE_Li, 0 }, /* Li */
222: { parse_mdoc_Nd, 0, 0 }, /* Nd */
223: { parse_mdoc_Nm, 0, 0 }, /* Nm */
224: { NULL, 0, 0 }, /* Op */
225: { NULL, 0, 0 }, /* Ot */
226: { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
227: { NULL, 0, 0 }, /* Rv */
228: { NULL, TYPE_St, 0 }, /* St */
229: { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
230: { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
231: { parse_mdoc_Xr, 0, 0 }, /* Xr */
232: { NULL, 0, 0 }, /* %A */
233: { NULL, 0, 0 }, /* %B */
234: { NULL, 0, 0 }, /* %D */
235: { NULL, 0, 0 }, /* %I */
236: { NULL, 0, 0 }, /* %J */
237: { NULL, 0, 0 }, /* %N */
238: { NULL, 0, 0 }, /* %O */
239: { NULL, 0, 0 }, /* %P */
240: { NULL, 0, 0 }, /* %R */
241: { NULL, 0, 0 }, /* %T */
242: { NULL, 0, 0 }, /* %V */
243: { NULL, 0, 0 }, /* Ac */
244: { NULL, 0, 0 }, /* Ao */
245: { NULL, 0, 0 }, /* Aq */
1.237 schwarze 246: { NULL, TYPE_At, 0 }, /* At */
1.235 schwarze 247: { NULL, 0, 0 }, /* Bc */
248: { NULL, 0, 0 }, /* Bf */
249: { NULL, 0, 0 }, /* Bo */
250: { NULL, 0, 0 }, /* Bq */
251: { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
1.236 schwarze 252: { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
1.235 schwarze 253: { NULL, 0, 0 }, /* Db */
254: { NULL, 0, 0 }, /* Dc */
255: { NULL, 0, 0 }, /* Do */
256: { NULL, 0, 0 }, /* Dq */
257: { NULL, 0, 0 }, /* Ec */
258: { NULL, 0, 0 }, /* Ef */
259: { NULL, TYPE_Em, 0 }, /* Em */
260: { NULL, 0, 0 }, /* Eo */
261: { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
262: { NULL, TYPE_Ms, 0 }, /* Ms */
263: { NULL, 0, 0 }, /* No */
264: { NULL, 0, 0 }, /* Ns */
265: { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
266: { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
267: { NULL, 0, 0 }, /* Pc */
268: { NULL, 0, 0 }, /* Pf */
269: { NULL, 0, 0 }, /* Po */
270: { NULL, 0, 0 }, /* Pq */
271: { NULL, 0, 0 }, /* Qc */
272: { NULL, 0, 0 }, /* Ql */
273: { NULL, 0, 0 }, /* Qo */
274: { NULL, 0, 0 }, /* Qq */
275: { NULL, 0, 0 }, /* Re */
276: { NULL, 0, 0 }, /* Rs */
277: { NULL, 0, 0 }, /* Sc */
278: { NULL, 0, 0 }, /* So */
279: { NULL, 0, 0 }, /* Sq */
280: { NULL, 0, 0 }, /* Sm */
281: { NULL, 0, 0 }, /* Sx */
282: { NULL, TYPE_Sy, 0 }, /* Sy */
283: { NULL, TYPE_Tn, 0 }, /* Tn */
284: { NULL, 0, NODE_NOSRC }, /* Ux */
285: { NULL, 0, 0 }, /* Xc */
286: { NULL, 0, 0 }, /* Xo */
287: { parse_mdoc_Fo, 0, 0 }, /* Fo */
288: { NULL, 0, 0 }, /* Fc */
289: { NULL, 0, 0 }, /* Oo */
290: { NULL, 0, 0 }, /* Oc */
291: { NULL, 0, 0 }, /* Bk */
292: { NULL, 0, 0 }, /* Ek */
293: { NULL, 0, 0 }, /* Bt */
294: { NULL, 0, 0 }, /* Hf */
295: { NULL, 0, 0 }, /* Fr */
296: { NULL, 0, 0 }, /* Ud */
1.237 schwarze 297: { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
1.235 schwarze 298: { NULL, 0, 0 }, /* Lp */
299: { NULL, TYPE_Lk, 0 }, /* Lk */
300: { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
301: { NULL, 0, 0 }, /* Brq */
302: { NULL, 0, 0 }, /* Bro */
303: { NULL, 0, 0 }, /* Brc */
304: { NULL, 0, 0 }, /* %C */
305: { NULL, 0, 0 }, /* Es */
306: { NULL, 0, 0 }, /* En */
307: { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
308: { NULL, 0, 0 }, /* %Q */
309: { NULL, 0, 0 }, /* %U */
310: { NULL, 0, 0 }, /* Ta */
1.1 kristaps 311: };
312:
1.141 schwarze 313:
1.1 kristaps 314: int
1.179 schwarze 315: mandocdb(int argc, char *argv[])
1.1 kristaps 316: {
1.187 schwarze 317: struct manconf conf;
318: struct mparse *mp;
1.221 schwarze 319: struct dba *dba;
1.206 schwarze 320: const char *path_arg, *progname;
1.187 schwarze 321: size_t j, sz;
1.59 schwarze 322: int ch, i;
1.50 kristaps 323:
1.207 schwarze 324: #if HAVE_PLEDGE
1.255 schwarze 325: if (pledge("stdio rpath wpath cpath", NULL) == -1) {
1.214 schwarze 326: warn("pledge");
1.218 kristaps 327: return (int)MANDOCLEVEL_SYSERR;
328: }
329: #endif
330:
331: #if HAVE_SANDBOX_INIT
332: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
333: warnx("sandbox_init");
1.207 schwarze 334: return (int)MANDOCLEVEL_SYSERR;
335: }
336: #endif
337:
1.187 schwarze 338: memset(&conf, 0, sizeof(conf));
1.50 kristaps 339:
340: /*
1.141 schwarze 341: * We accept a few different invocations.
1.50 kristaps 342: * The CHECKOP macro makes sure that invocation styles don't
343: * clobber each other.
344: */
345: #define CHECKOP(_op, _ch) do \
1.264 schwarze 346: if ((_op) != OP_DEFAULT) { \
1.198 schwarze 347: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 348: goto usage; \
349: } while (/*CONSTCOND*/0)
1.10 kristaps 350:
1.262 schwarze 351: mparse_options = MPARSE_VALIDATE;
1.59 schwarze 352: path_arg = NULL;
1.38 schwarze 353: op = OP_DEFAULT;
1.1 kristaps 354:
1.264 schwarze 355: while ((ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")) != -1)
1.1 kristaps 356: switch (ch) {
1.141 schwarze 357: case 'a':
1.12 schwarze 358: use_all = 1;
359: break;
1.141 schwarze 360: case 'C':
1.50 kristaps 361: CHECKOP(op, ch);
1.59 schwarze 362: path_arg = optarg;
1.38 schwarze 363: op = OP_CONFFILE;
1.34 schwarze 364: break;
1.141 schwarze 365: case 'D':
1.125 schwarze 366: debug++;
367: break;
1.141 schwarze 368: case 'd':
1.50 kristaps 369: CHECKOP(op, ch);
1.59 schwarze 370: path_arg = optarg;
1.5 kristaps 371: op = OP_UPDATE;
372: break;
1.141 schwarze 373: case 'n':
1.50 kristaps 374: nodb = 1;
375: break;
1.141 schwarze 376: case 'p':
1.126 schwarze 377: warnings = 1;
378: break;
1.141 schwarze 379: case 'Q':
1.116 schwarze 380: mparse_options |= MPARSE_QUICK;
1.102 schwarze 381: break;
1.141 schwarze 382: case 'T':
1.264 schwarze 383: if (strcmp(optarg, "utf8") != 0) {
1.198 schwarze 384: warnx("-T%s: Unsupported output format",
385: optarg);
1.95 schwarze 386: goto usage;
387: }
388: write_utf8 = 1;
389: break;
1.141 schwarze 390: case 't':
1.50 kristaps 391: CHECKOP(op, ch);
1.38 schwarze 392: dup2(STDOUT_FILENO, STDERR_FILENO);
393: op = OP_TEST;
1.50 kristaps 394: nodb = warnings = 1;
1.38 schwarze 395: break;
1.141 schwarze 396: case 'u':
1.50 kristaps 397: CHECKOP(op, ch);
1.59 schwarze 398: path_arg = optarg;
1.5 kristaps 399: op = OP_DELETE;
400: break;
1.141 schwarze 401: case 'v':
1.126 schwarze 402: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 403: break;
1.1 kristaps 404: default:
1.38 schwarze 405: goto usage;
1.1 kristaps 406: }
407:
408: argc -= optind;
409: argv += optind;
410:
1.207 schwarze 411: #if HAVE_PLEDGE
1.210 schwarze 412: if (nodb) {
413: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 414: warn("pledge");
1.210 schwarze 415: return (int)MANDOCLEVEL_SYSERR;
416: }
1.207 schwarze 417: }
418: #endif
419:
1.264 schwarze 420: if (op == OP_CONFFILE && argc > 0) {
1.198 schwarze 421: warnx("-C: Too many arguments");
1.38 schwarze 422: goto usage;
423: }
424:
1.59 schwarze 425: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 426: mchars_alloc();
1.261 schwarze 427: mp = mparse_alloc(mparse_options, MANDOC_OS_OTHER, NULL);
1.202 schwarze 428: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
429: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 430:
1.264 schwarze 431: if (op == OP_UPDATE || op == OP_DELETE || op == OP_TEST) {
1.59 schwarze 432:
1.50 kristaps 433: /*
1.150 schwarze 434: * Most of these deal with a specific directory.
1.140 schwarze 435: * Jump into that directory first.
1.50 kristaps 436: */
1.264 schwarze 437: if (op != OP_TEST && set_basedir(path_arg, 1) == 0)
1.50 kristaps 438: goto out;
1.140 schwarze 439:
1.225 schwarze 440: dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
441: if (dba != NULL) {
1.140 schwarze 442: /*
443: * The existing database is usable. Process
444: * all files specified on the command-line.
445: */
446: use_all = 1;
1.138 schwarze 447: for (i = 0; i < argc; i++)
448: filescan(argv[i]);
1.225 schwarze 449: if (nodb == 0)
1.221 schwarze 450: dbprune(dba);
1.138 schwarze 451: } else {
1.228 schwarze 452: /* Database missing or corrupt. */
1.230 schwarze 453: if (op != OP_UPDATE || errno != ENOENT)
454: say(MANDOC_DB, "%s: Automatically recreating"
455: " from scratch", strerror(errno));
1.140 schwarze 456: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 457: op = OP_DEFAULT;
1.264 schwarze 458: if (treescan() == 0)
1.138 schwarze 459: goto out;
1.221 schwarze 460: dba = dba_new(128);
1.138 schwarze 461: }
1.264 schwarze 462: if (op != OP_DELETE)
1.221 schwarze 463: mpages_merge(dba, mp);
464: if (nodb == 0)
465: dbwrite(dba);
466: dba_free(dba);
1.50 kristaps 467: } else {
468: /*
469: * If we have arguments, use them as our manpaths.
1.222 schwarze 470: * If we don't, use man.conf(5).
1.50 kristaps 471: */
472: if (argc > 0) {
1.187 schwarze 473: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 474: argc, sizeof(char *));
1.187 schwarze 475: conf.manpath.sz = (size_t)argc;
1.50 kristaps 476: for (i = 0; i < argc; i++)
1.187 schwarze 477: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 478: } else
1.187 schwarze 479: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 480:
1.187 schwarze 481: if (conf.manpath.sz == 0) {
1.127 schwarze 482: exitcode = (int)MANDOCLEVEL_BADARG;
483: say("", "Empty manpath");
484: }
1.50 kristaps 485:
486: /*
1.69 schwarze 487: * First scan the tree rooted at a base directory, then
488: * build a new database and finally move it into place.
1.50 kristaps 489: * Ignore zero-length directories and strip trailing
490: * slashes.
491: */
1.187 schwarze 492: for (j = 0; j < conf.manpath.sz; j++) {
493: sz = strlen(conf.manpath.paths[j]);
494: if (sz && conf.manpath.paths[j][sz - 1] == '/')
495: conf.manpath.paths[j][--sz] = '\0';
1.264 schwarze 496: if (sz == 0)
1.50 kristaps 497: continue;
1.66 schwarze 498:
499: if (j) {
1.202 schwarze 500: mandoc_ohash_init(&mpages, 6,
501: offsetof(struct mpage, inodev));
502: mandoc_ohash_init(&mlinks, 6,
503: offsetof(struct mlink, file));
1.66 schwarze 504: }
505:
1.264 schwarze 506: if (set_basedir(conf.manpath.paths[j], argc > 0) == 0)
1.165 schwarze 507: continue;
1.264 schwarze 508: if (treescan() == 0)
1.165 schwarze 509: continue;
1.221 schwarze 510: dba = dba_new(128);
511: mpages_merge(dba, mp);
512: if (nodb == 0)
513: dbwrite(dba);
514: dba_free(dba);
1.66 schwarze 515:
1.187 schwarze 516: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 517: mpages_free();
1.78 schwarze 518: ohash_delete(&mpages);
1.83 schwarze 519: ohash_delete(&mlinks);
1.66 schwarze 520: }
1.50 kristaps 521: }
522: }
523: out:
1.187 schwarze 524: manconf_free(&conf);
1.168 schwarze 525: mparse_free(mp);
1.203 schwarze 526: mchars_free();
1.80 schwarze 527: mpages_free();
1.78 schwarze 528: ohash_delete(&mpages);
1.83 schwarze 529: ohash_delete(&mlinks);
1.197 schwarze 530: return exitcode;
1.50 kristaps 531: usage:
1.206 schwarze 532: progname = getprogname();
1.126 schwarze 533: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
534: " %s [-aDnpQ] [-Tutf8] dir ...\n"
535: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
536: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 537: " %s [-Q] -t file ...\n",
1.206 schwarze 538: progname, progname, progname, progname, progname);
1.50 kristaps 539:
1.197 schwarze 540: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 541: }
542:
543: /*
1.231 schwarze 544: * To get a singly linked list in alpha order while inserting entries
545: * at the beginning, process directory entries in reverse alpha order.
546: */
547: static int
1.232 schwarze 548: #if HAVE_FTS_COMPARE_CONST
549: fts_compare(const FTSENT *const *a, const FTSENT *const *b)
550: #else
1.231 schwarze 551: fts_compare(const FTSENT **a, const FTSENT **b)
1.232 schwarze 552: #endif
1.231 schwarze 553: {
554: return -strcmp((*a)->fts_name, (*b)->fts_name);
555: }
556:
557: /*
1.59 schwarze 558: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 559: * We use fts(), scanning directory parts along the way for clues to our
560: * section and architecture.
561: *
562: * If use_all has been specified, grok all files.
563: * If not, sanitise paths to the following:
564: *
1.141 schwarze 565: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 566: * or
567: * [./]cat<section>[/<arch>]/<name>.0
568: *
1.216 schwarze 569: * TODO: accommodate for multi-language directories.
1.50 kristaps 570: */
571: static int
1.59 schwarze 572: treescan(void)
1.50 kristaps 573: {
1.139 schwarze 574: char buf[PATH_MAX];
1.50 kristaps 575: FTS *f;
576: FTSENT *ff;
1.84 schwarze 577: struct mlink *mlink;
1.221 schwarze 578: int gzip;
579: enum form dform;
1.94 schwarze 580: char *dsec, *arch, *fsec, *cp;
581: const char *path;
1.50 kristaps 582: const char *argv[2];
583:
584: argv[0] = ".";
1.244 schwarze 585: argv[1] = NULL;
1.50 kristaps 586:
1.231 schwarze 587: f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
588: fts_compare);
1.200 schwarze 589: if (f == NULL) {
1.59 schwarze 590: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 591: say("", "&fts_open");
1.197 schwarze 592: return 0;
1.50 kristaps 593: }
594:
595: dsec = arch = NULL;
596: dform = FORM_NONE;
597:
1.200 schwarze 598: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 599: path = ff->fts_path + 2;
1.139 schwarze 600: switch (ff->fts_info) {
601:
602: /*
603: * Symbolic links require various sanity checks,
604: * then get handled just like regular files.
605: */
1.141 schwarze 606: case FTS_SL:
1.200 schwarze 607: if (realpath(path, buf) == NULL) {
1.139 schwarze 608: if (warnings)
609: say(path, "&realpath");
610: continue;
611: }
1.264 schwarze 612: if (strncmp(buf, basedir, basedir_len) != 0
1.184 schwarze 613: #ifdef HOMEBREWDIR
1.264 schwarze 614: && strncmp(buf, HOMEBREWDIR, strlen(HOMEBREWDIR))
1.184 schwarze 615: #endif
616: ) {
1.139 schwarze 617: if (warnings) say("",
618: "%s: outside base directory", buf);
619: continue;
620: }
621: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 622: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 623: if (warnings)
624: say(path, "&stat");
625: continue;
626: }
627: /* FALLTHROUGH */
628:
1.50 kristaps 629: /*
1.83 schwarze 630: * If we're a regular file, add an mlink by using the
1.50 kristaps 631: * stored directory data and handling the filename.
632: */
1.141 schwarze 633: case FTS_F:
1.200 schwarze 634: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 635: continue;
1.50 kristaps 636: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 637: if (warnings)
1.59 schwarze 638: say(path, "Extraneous file");
1.50 kristaps 639: continue;
1.124 schwarze 640: }
641: gzip = 0;
642: fsec = NULL;
1.200 schwarze 643: while (fsec == NULL) {
1.124 schwarze 644: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 645: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 646: break;
647: gzip = 1;
648: *fsec = '\0';
649: fsec = NULL;
650: }
1.200 schwarze 651: if (fsec == NULL) {
1.60 schwarze 652: if ( ! use_all) {
1.56 schwarze 653: if (warnings)
1.60 schwarze 654: say(path,
655: "No filename suffix");
1.50 kristaps 656: continue;
657: }
1.200 schwarze 658: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 659: if (warnings)
660: say(path, "Skip html");
661: continue;
1.200 schwarze 662: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 663: if (warnings)
664: say(path, "Skip ps");
665: continue;
1.200 schwarze 666: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 667: if (warnings)
668: say(path, "Skip pdf");
669: continue;
670: } else if ( ! use_all &&
1.200 schwarze 671: ((dform == FORM_SRC &&
1.185 schwarze 672: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 673: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 674: if (warnings)
675: say(path, "Wrong filename suffix");
676: continue;
1.66 schwarze 677: } else
1.84 schwarze 678: fsec[-1] = '\0';
1.94 schwarze 679:
1.84 schwarze 680: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 681: if (strlcpy(mlink->file, path,
682: sizeof(mlink->file)) >=
683: sizeof(mlink->file)) {
684: say(path, "Filename too long");
685: free(mlink);
686: continue;
687: }
1.84 schwarze 688: mlink->dform = dform;
1.94 schwarze 689: mlink->dsec = dsec;
690: mlink->arch = arch;
691: mlink->name = ff->fts_name;
692: mlink->fsec = fsec;
1.124 schwarze 693: mlink->gzip = gzip;
1.84 schwarze 694: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 695: continue;
1.139 schwarze 696:
1.141 schwarze 697: case FTS_D:
698: case FTS_DP:
1.139 schwarze 699: break;
700:
701: default:
1.60 schwarze 702: if (warnings)
703: say(path, "Not a regular file");
1.50 kristaps 704: continue;
1.60 schwarze 705: }
1.1 kristaps 706:
1.50 kristaps 707: switch (ff->fts_level) {
1.141 schwarze 708: case 0:
1.50 kristaps 709: /* Ignore the root directory. */
710: break;
1.141 schwarze 711: case 1:
1.50 kristaps 712: /*
713: * This might contain manX/ or catX/.
714: * Try to infer this from the name.
715: * If we're not in use_all, enforce it.
716: */
717: cp = ff->fts_name;
1.200 schwarze 718: if (ff->fts_info == FTS_DP) {
719: dform = FORM_NONE;
720: dsec = NULL;
1.50 kristaps 721: break;
1.200 schwarze 722: }
1.1 kristaps 723:
1.200 schwarze 724: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 725: dform = FORM_SRC;
1.66 schwarze 726: dsec = cp + 3;
1.200 schwarze 727: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 728: dform = FORM_CAT;
1.66 schwarze 729: dsec = cp + 3;
1.94 schwarze 730: } else {
731: dform = FORM_NONE;
732: dsec = NULL;
1.50 kristaps 733: }
1.1 kristaps 734:
1.200 schwarze 735: if (dsec != NULL || use_all)
1.50 kristaps 736: break;
1.1 kristaps 737:
1.56 schwarze 738: if (warnings)
1.59 schwarze 739: say(path, "Unknown directory part");
1.50 kristaps 740: fts_set(f, ff, FTS_SKIP);
741: break;
1.141 schwarze 742: case 2:
1.50 kristaps 743: /*
744: * Possibly our architecture.
745: * If we're descending, keep tabs on it.
746: */
1.200 schwarze 747: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 748: arch = ff->fts_name;
1.94 schwarze 749: else
750: arch = NULL;
1.50 kristaps 751: break;
752: default:
1.200 schwarze 753: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 754: break;
1.56 schwarze 755: if (warnings)
1.59 schwarze 756: say(path, "Extraneous directory part");
1.50 kristaps 757: fts_set(f, ff, FTS_SKIP);
758: break;
1.5 kristaps 759: }
1.50 kristaps 760: }
761:
762: fts_close(f);
1.197 schwarze 763: return 1;
1.50 kristaps 764: }
1.5 kristaps 765:
1.50 kristaps 766: /*
1.89 schwarze 767: * Add a file to the mlinks table.
1.50 kristaps 768: * Do not verify that it's a "valid" looking manpage (we'll do that
769: * later).
770: *
771: * Try to infer the manual section, architecture, and page name from the
772: * path, assuming it looks like
773: *
1.141 schwarze 774: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 775: * or
776: * [./]cat<section>[/<arch>]/<name>.0
777: *
778: * See treescan() for the fts(3) version of this.
779: */
780: static void
1.59 schwarze 781: filescan(const char *file)
1.50 kristaps 782: {
1.59 schwarze 783: char buf[PATH_MAX];
1.84 schwarze 784: struct stat st;
785: struct mlink *mlink;
1.59 schwarze 786: char *p, *start;
1.5 kristaps 787:
1.50 kristaps 788: assert(use_all);
1.5 kristaps 789:
1.264 schwarze 790: if (strncmp(file, "./", 2) == 0)
1.50 kristaps 791: file += 2;
1.5 kristaps 792:
1.139 schwarze 793: /*
794: * We have to do lstat(2) before realpath(3) loses
795: * the information whether this is a symbolic link.
796: * We need to know that because for symbolic links,
797: * we want to use the orginal file name, while for
798: * regular files, we want to use the real path.
799: */
1.264 schwarze 800: if (lstat(file, &st) == -1) {
1.139 schwarze 801: exitcode = (int)MANDOCLEVEL_BADARG;
802: say(file, "&lstat");
803: return;
1.265 ! schwarze 804: } else if (S_ISREG(st.st_mode) == 0 && S_ISLNK(st.st_mode) == 0) {
1.139 schwarze 805: exitcode = (int)MANDOCLEVEL_BADARG;
806: say(file, "Not a regular file");
807: return;
808: }
809:
810: /*
811: * We have to resolve the file name to the real path
812: * in any case for the base directory check.
813: */
1.264 schwarze 814: if (realpath(file, buf) == NULL) {
1.59 schwarze 815: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 816: say(file, "&realpath");
1.59 schwarze 817: return;
1.106 schwarze 818: }
819:
1.264 schwarze 820: if (op == OP_TEST)
1.106 schwarze 821: start = buf;
1.264 schwarze 822: else if (strncmp(buf, basedir, basedir_len) == 0)
823: start = buf + basedir_len;
1.184 schwarze 824: #ifdef HOMEBREWDIR
1.264 schwarze 825: else if (strncmp(buf, HOMEBREWDIR, strlen(HOMEBREWDIR)) == 0)
1.184 schwarze 826: start = buf;
827: #endif
1.106 schwarze 828: else {
1.59 schwarze 829: exitcode = (int)MANDOCLEVEL_BADARG;
830: say("", "%s: outside base directory", buf);
831: return;
1.106 schwarze 832: }
833:
1.139 schwarze 834: /*
835: * Now we are sure the file is inside our tree.
836: * If it is a symbolic link, ignore the real path
837: * and use the original name.
838: * This implies passing stuff like "cat1/../man1/foo.1"
839: * on the command line won't work. So don't do that.
840: * Note the stat(2) can still fail if the link target
841: * doesn't exist.
842: */
1.265 ! schwarze 843: if (S_ISLNK(st.st_mode)) {
1.264 schwarze 844: if (stat(buf, &st) == -1) {
1.139 schwarze 845: exitcode = (int)MANDOCLEVEL_BADARG;
846: say(file, "&stat");
847: return;
848: }
1.143 schwarze 849: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
850: say(file, "Filename too long");
851: return;
852: }
1.150 schwarze 853: start = buf;
1.264 schwarze 854: if (op != OP_TEST && strncmp(buf, basedir, basedir_len) == 0)
855: start += basedir_len;
1.50 kristaps 856: }
1.106 schwarze 857:
1.84 schwarze 858: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 859: mlink->dform = FORM_NONE;
1.143 schwarze 860: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
861: sizeof(mlink->file)) {
862: say(start, "Filename too long");
1.180 schwarze 863: free(mlink);
1.143 schwarze 864: return;
1.239 schwarze 865: }
866:
867: /*
868: * In test mode or when the original name is absolute
869: * but outside our tree, guess the base directory.
870: */
871:
872: if (op == OP_TEST || (start == buf && *start == '/')) {
873: if (strncmp(buf, "man/", 4) == 0)
874: start = buf + 4;
875: else if ((start = strstr(buf, "/man/")) != NULL)
876: start += 5;
877: else
878: start = buf;
1.143 schwarze 879: }
1.17 schwarze 880:
1.50 kristaps 881: /*
882: * First try to guess our directory structure.
883: * If we find a separator, try to look for man* or cat*.
884: * If we find one of these and what's underneath is a directory,
885: * assume it's an architecture.
886: */
1.264 schwarze 887: if ((p = strchr(start, '/')) != NULL) {
1.50 kristaps 888: *p++ = '\0';
1.264 schwarze 889: if (strncmp(start, "man", 3) == 0) {
1.84 schwarze 890: mlink->dform = FORM_SRC;
1.94 schwarze 891: mlink->dsec = start + 3;
1.264 schwarze 892: } else if (strncmp(start, "cat", 3) == 0) {
1.84 schwarze 893: mlink->dform = FORM_CAT;
1.94 schwarze 894: mlink->dsec = start + 3;
1.17 schwarze 895: }
1.5 kristaps 896:
1.50 kristaps 897: start = p;
1.264 schwarze 898: if (mlink->dsec != NULL && (p = strchr(start, '/')) != NULL) {
1.50 kristaps 899: *p++ = '\0';
1.94 schwarze 900: mlink->arch = start;
1.50 kristaps 901: start = p;
1.84 schwarze 902: }
1.50 kristaps 903: }
904:
905: /*
906: * Now check the file suffix.
907: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
908: */
909: p = strrchr(start, '\0');
1.264 schwarze 910: while (p-- > start && *p != '/' && *p != '.')
911: continue;
1.50 kristaps 912:
1.264 schwarze 913: if (*p == '.') {
1.50 kristaps 914: *p++ = '\0';
1.94 schwarze 915: mlink->fsec = p;
1.5 kristaps 916: }
917:
1.10 kristaps 918: /*
1.50 kristaps 919: * Now try to parse the name.
920: * Use the filename portion of the path.
1.10 kristaps 921: */
1.84 schwarze 922: mlink->name = start;
1.264 schwarze 923: if ((p = strrchr(start, '/')) != NULL) {
1.84 schwarze 924: mlink->name = p + 1;
1.50 kristaps 925: *p = '\0';
1.84 schwarze 926: }
927: mlink_add(mlink, &st);
1.50 kristaps 928: }
1.5 kristaps 929:
1.50 kristaps 930: static void
1.84 schwarze 931: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 932: {
1.83 schwarze 933: struct inodev inodev;
1.78 schwarze 934: struct mpage *mpage;
1.83 schwarze 935: unsigned int slot;
1.50 kristaps 936:
1.84 schwarze 937: assert(NULL != mlink->file);
1.50 kristaps 938:
1.94 schwarze 939: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
940: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
941: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
942: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 943:
944: if ('0' == *mlink->fsec) {
945: free(mlink->fsec);
946: mlink->fsec = mandoc_strdup(mlink->dsec);
947: mlink->fform = FORM_CAT;
948: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
949: mlink->fform = FORM_SRC;
1.74 schwarze 950: else
1.84 schwarze 951: mlink->fform = FORM_NONE;
1.82 schwarze 952:
1.83 schwarze 953: slot = ohash_qlookup(&mlinks, mlink->file);
954: assert(NULL == ohash_find(&mlinks, slot));
955: ohash_insert(&mlinks, slot, mlink);
956:
1.186 schwarze 957: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 958: inodev.st_ino = st->st_ino;
959: inodev.st_dev = st->st_dev;
960: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
961: sizeof(struct inodev), inodev.st_ino);
962: mpage = ohash_find(&mpages, slot);
963: if (NULL == mpage) {
964: mpage = mandoc_calloc(1, sizeof(struct mpage));
965: mpage->inodev.st_ino = inodev.st_ino;
966: mpage->inodev.st_dev = inodev.st_dev;
1.221 schwarze 967: mpage->form = FORM_NONE;
1.231 schwarze 968: mpage->next = mpage_head;
969: mpage_head = mpage;
1.83 schwarze 970: ohash_insert(&mpages, slot, mpage);
971: } else
1.85 schwarze 972: mlink->next = mpage->mlinks;
1.82 schwarze 973: mpage->mlinks = mlink;
1.118 schwarze 974: mlink->mpage = mpage;
1.82 schwarze 975: }
976:
977: static void
978: mlink_free(struct mlink *mlink)
979: {
980:
981: free(mlink->dsec);
982: free(mlink->arch);
983: free(mlink->name);
984: free(mlink->fsec);
985: free(mlink);
1.50 kristaps 986: }
1.3 kristaps 987:
1.50 kristaps 988: static void
1.78 schwarze 989: mpages_free(void)
1.50 kristaps 990: {
1.78 schwarze 991: struct mpage *mpage;
1.82 schwarze 992: struct mlink *mlink;
1.3 kristaps 993:
1.231 schwarze 994: while ((mpage = mpage_head) != NULL) {
995: while ((mlink = mpage->mlinks) != NULL) {
1.85 schwarze 996: mpage->mlinks = mlink->next;
1.82 schwarze 997: mlink_free(mlink);
998: }
1.231 schwarze 999: mpage_head = mpage->next;
1.78 schwarze 1000: free(mpage->sec);
1001: free(mpage->arch);
1.82 schwarze 1002: free(mpage->title);
1003: free(mpage->desc);
1.78 schwarze 1004: free(mpage);
1.50 kristaps 1005: }
1006: }
1.38 schwarze 1007:
1.50 kristaps 1008: /*
1.89 schwarze 1009: * For each mlink to the mpage, check whether the path looks like
1010: * it is formatted, and if it does, check whether a source manual
1011: * exists by the same name, ignoring the suffix.
1012: * If both conditions hold, drop the mlink.
1013: */
1014: static void
1015: mlinks_undupe(struct mpage *mpage)
1016: {
1017: char buf[PATH_MAX];
1018: struct mlink **prev;
1019: struct mlink *mlink;
1020: char *bufp;
1021:
1022: mpage->form = FORM_CAT;
1.90 schwarze 1023: prev = &mpage->mlinks;
1024: while (NULL != (mlink = *prev)) {
1.89 schwarze 1025: if (FORM_CAT != mlink->dform) {
1026: mpage->form = FORM_NONE;
1.90 schwarze 1027: goto nextlink;
1.89 schwarze 1028: }
1.143 schwarze 1029: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1030: bufp = strstr(buf, "cat");
1031: assert(NULL != bufp);
1032: memcpy(bufp, "man", 3);
1033: if (NULL != (bufp = strrchr(buf, '.')))
1034: *++bufp = '\0';
1.143 schwarze 1035: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1036: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1037: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1038: goto nextlink;
1.89 schwarze 1039: if (warnings)
1040: say(mlink->file, "Man source exists: %s", buf);
1041: if (use_all)
1.90 schwarze 1042: goto nextlink;
1.89 schwarze 1043: *prev = mlink->next;
1044: mlink_free(mlink);
1.90 schwarze 1045: continue;
1046: nextlink:
1047: prev = &(*prev)->next;
1.89 schwarze 1048: }
1049: }
1050:
1.131 schwarze 1051: static void
1.93 schwarze 1052: mlink_check(struct mpage *mpage, struct mlink *mlink)
1053: {
1.131 schwarze 1054: struct str *str;
1055: unsigned int slot;
1.93 schwarze 1056:
1057: /*
1058: * Check whether the manual section given in a file
1059: * agrees with the directory where the file is located.
1060: * Some manuals have suffixes like (3p) on their
1061: * section number either inside the file or in the
1062: * directory name, some are linked into more than one
1063: * section, like encrypt(1) = makekey(8).
1064: */
1065:
1066: if (FORM_SRC == mpage->form &&
1.131 schwarze 1067: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1068: say(mlink->file, "Section \"%s\" manual in %s directory",
1069: mpage->sec, mlink->dsec);
1070:
1071: /*
1072: * Manual page directories exist for each kernel
1073: * architecture as returned by machine(1).
1074: * However, many manuals only depend on the
1075: * application architecture as returned by arch(1).
1076: * For example, some (2/ARM) manuals are shared
1077: * across the "armish" and "zaurus" kernel
1078: * architectures.
1079: * A few manuals are even shared across completely
1080: * different architectures, for example fdformat(1)
1.229 schwarze 1081: * on amd64, i386, and sparc64.
1.93 schwarze 1082: */
1083:
1.131 schwarze 1084: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1085: say(mlink->file, "Architecture \"%s\" manual in "
1086: "\"%s\" directory", mpage->arch, mlink->arch);
1087:
1.131 schwarze 1088: /*
1089: * XXX
1.133 schwarze 1090: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1091: */
1092:
1093: if (FORM_CAT == mpage->form)
1094: return;
1095:
1096: /*
1097: * Check whether this mlink
1098: * appears as a name in the NAME section.
1099: */
1.93 schwarze 1100:
1.133 schwarze 1101: slot = ohash_qlookup(&names, mlink->name);
1102: str = ohash_find(&names, slot);
1.131 schwarze 1103: assert(NULL != str);
1.133 schwarze 1104: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1105: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1106: }
1107:
1.89 schwarze 1108: /*
1.80 schwarze 1109: * Run through the files in the global vector "mpages"
1110: * and add them to the database specified in "basedir".
1.50 kristaps 1111: *
1112: * This handles the parsing scheme itself, using the cues of directory
1113: * and filename to determine whether the file is parsable or not.
1114: */
1.59 schwarze 1115: static void
1.221 schwarze 1116: mpages_merge(struct dba *dba, struct mparse *mp)
1.50 kristaps 1117: {
1.231 schwarze 1118: struct mpage *mpage, *mpage_dest;
1.118 schwarze 1119: struct mlink *mlink, *mlink_dest;
1.262 schwarze 1120: struct roff_meta *meta;
1.112 schwarze 1121: char *cp;
1.162 schwarze 1122: int fd;
1.69 schwarze 1123:
1.231 schwarze 1124: for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1.89 schwarze 1125: mlinks_undupe(mpage);
1.231 schwarze 1126: if ((mlink = mpage->mlinks) == NULL)
1127: continue;
1.50 kristaps 1128:
1.133 schwarze 1129: name_mask = NAME_MASK;
1.202 schwarze 1130: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1131: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1132: mparse_reset(mp);
1.262 schwarze 1133: meta = NULL;
1.124 schwarze 1134:
1.213 schwarze 1135: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1136: say(mlink->file, "&open");
1.162 schwarze 1137: goto nextpage;
1.124 schwarze 1138: }
1.14 schwarze 1139:
1140: /*
1.183 schwarze 1141: * Interpret the file as mdoc(7) or man(7) source
1142: * code, unless it is known to be formatted.
1.14 schwarze 1143: */
1.182 schwarze 1144: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1145: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1146: close(fd);
1.240 schwarze 1147: fd = -1;
1.262 schwarze 1148: meta = mparse_result(mp);
1.76 schwarze 1149: }
1.14 schwarze 1150:
1.262 schwarze 1151: if (meta != NULL && meta->sodest != NULL) {
1.118 schwarze 1152: mlink_dest = ohash_find(&mlinks,
1.262 schwarze 1153: ohash_qlookup(&mlinks, meta->sodest));
1.171 schwarze 1154: if (mlink_dest == NULL) {
1.262 schwarze 1155: mandoc_asprintf(&cp, "%s.gz", meta->sodest);
1.171 schwarze 1156: mlink_dest = ohash_find(&mlinks,
1157: ohash_qlookup(&mlinks, cp));
1158: free(cp);
1159: }
1160: if (mlink_dest != NULL) {
1.118 schwarze 1161:
1162: /* The .so target exists. */
1163:
1164: mpage_dest = mlink_dest->mpage;
1165: while (1) {
1166: mlink->mpage = mpage_dest;
1167:
1168: /*
1169: * If the target was already
1170: * processed, add the links
1171: * to the database now.
1172: * Otherwise, this will
1173: * happen when we come
1174: * to the target.
1175: */
1176:
1.221 schwarze 1177: if (mpage_dest->dba != NULL)
1178: dbadd_mlink(mlink);
1.118 schwarze 1179:
1.171 schwarze 1180: if (mlink->next == NULL)
1.118 schwarze 1181: break;
1182: mlink = mlink->next;
1183: }
1184:
1185: /* Move all links to the target. */
1186:
1187: mlink->next = mlink_dest->next;
1188: mlink_dest->next = mpage->mlinks;
1189: mpage->mlinks = NULL;
1.263 schwarze 1190: goto nextpage;
1.118 schwarze 1191: }
1.263 schwarze 1192: meta->macroset = MACROSET_NONE;
1193: }
1194: if (meta != NULL && meta->macroset == MACROSET_MDOC) {
1.82 schwarze 1195: mpage->form = FORM_SRC;
1.262 schwarze 1196: mpage->sec = meta->msec;
1.155 schwarze 1197: mpage->sec = mandoc_strdup(
1.171 schwarze 1198: mpage->sec == NULL ? "" : mpage->sec);
1.262 schwarze 1199: mpage->arch = meta->arch;
1.84 schwarze 1200: mpage->arch = mandoc_strdup(
1.171 schwarze 1201: mpage->arch == NULL ? "" : mpage->arch);
1.262 schwarze 1202: mpage->title = mandoc_strdup(meta->title);
1203: } else if (meta != NULL && meta->macroset == MACROSET_MAN) {
1204: if (*meta->msec != '\0' || *meta->title != '\0') {
1.240 schwarze 1205: mpage->form = FORM_SRC;
1.262 schwarze 1206: mpage->sec = mandoc_strdup(meta->msec);
1.240 schwarze 1207: mpage->arch = mandoc_strdup(mlink->arch);
1.262 schwarze 1208: mpage->title = mandoc_strdup(meta->title);
1.240 schwarze 1209: } else
1.262 schwarze 1210: meta = NULL;
1.240 schwarze 1211: }
1212:
1213: assert(mpage->desc == NULL);
1.263 schwarze 1214: if (meta == NULL || meta->sodest != NULL) {
1.182 schwarze 1215: mpage->sec = mandoc_strdup(mlink->dsec);
1216: mpage->arch = mandoc_strdup(mlink->arch);
1217: mpage->title = mandoc_strdup(mlink->name);
1.263 schwarze 1218: if (meta == NULL) {
1219: mpage->form = FORM_CAT;
1220: parse_cat(mpage, fd);
1221: } else
1222: mpage->form = FORM_SRC;
1.262 schwarze 1223: } else if (meta->macroset == MACROSET_MDOC)
1224: parse_mdoc(mpage, meta, meta->first);
1.240 schwarze 1225: else
1.262 schwarze 1226: parse_man(mpage, meta, meta->first);
1.242 schwarze 1227: if (mpage->desc == NULL) {
1228: mpage->desc = mandoc_strdup(mlink->name);
1229: if (warnings)
1230: say(mlink->file, "No one-line description, "
1231: "using filename \"%s\"", mlink->name);
1232: }
1.131 schwarze 1233:
1.238 schwarze 1234: for (mlink = mpage->mlinks;
1235: mlink != NULL;
1236: mlink = mlink->next) {
1237: putkey(mpage, mlink->name, NAME_FILE);
1238: if (warnings && !use_all)
1.131 schwarze 1239: mlink_check(mpage, mlink);
1.238 schwarze 1240: }
1.44 kristaps 1241:
1.221 schwarze 1242: dbadd(dba, mpage);
1.124 schwarze 1243:
1244: nextpage:
1.66 schwarze 1245: ohash_delete(&strings);
1.133 schwarze 1246: ohash_delete(&names);
1.50 kristaps 1247: }
1248: }
1.12 schwarze 1249:
1.50 kristaps 1250: static void
1.124 schwarze 1251: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1252: {
1253: FILE *stream;
1.241 schwarze 1254: struct mlink *mlink;
1255: char *line, *p, *title, *sec;
1.209 schwarze 1256: size_t linesz, plen, titlesz;
1257: ssize_t len;
1258: int offs;
1.12 schwarze 1259:
1.241 schwarze 1260: mlink = mpage->mlinks;
1261: stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1262: if (stream == NULL) {
1263: if (fd != -1)
1.159 schwarze 1264: close(fd);
1.56 schwarze 1265: if (warnings)
1.241 schwarze 1266: say(mlink->file, "&fopen");
1.50 kristaps 1267: return;
1268: }
1.1 kristaps 1269:
1.209 schwarze 1270: line = NULL;
1271: linesz = 0;
1272:
1.241 schwarze 1273: /* Parse the section number from the header line. */
1274:
1275: while (getline(&line, &linesz, stream) != -1) {
1276: if (*line == '\n')
1277: continue;
1278: if ((sec = strchr(line, '(')) == NULL)
1279: break;
1280: if ((p = strchr(++sec, ')')) == NULL)
1281: break;
1282: free(mpage->sec);
1283: mpage->sec = mandoc_strndup(sec, p - sec);
1284: if (warnings && *mlink->dsec != '\0' &&
1285: strcasecmp(mpage->sec, mlink->dsec))
1286: say(mlink->file,
1287: "Section \"%s\" manual in %s directory",
1288: mpage->sec, mlink->dsec);
1289: break;
1290: }
1291:
1.50 kristaps 1292: /* Skip to first blank line. */
1.1 kristaps 1293:
1.241 schwarze 1294: while (line == NULL || *line != '\n')
1295: if (getline(&line, &linesz, stream) == -1)
1.50 kristaps 1296: break;
1.1 kristaps 1297:
1.50 kristaps 1298: /*
1299: * Assume the first line that is not indented
1300: * is the first section header. Skip to it.
1301: */
1.1 kristaps 1302:
1.209 schwarze 1303: while (getline(&line, &linesz, stream) != -1)
1304: if (*line != '\n' && *line != ' ')
1.50 kristaps 1305: break;
1.141 schwarze 1306:
1.50 kristaps 1307: /*
1308: * Read up until the next section into a buffer.
1309: * Strip the leading and trailing newline from each read line,
1310: * appending a trailing space.
1311: * Ignore empty (whitespace-only) lines.
1312: */
1.1 kristaps 1313:
1.50 kristaps 1314: titlesz = 0;
1315: title = NULL;
1.38 schwarze 1316:
1.209 schwarze 1317: while ((len = getline(&line, &linesz, stream)) != -1) {
1318: if (*line != ' ')
1.50 kristaps 1319: break;
1.209 schwarze 1320: offs = 0;
1321: while (isspace((unsigned char)line[offs]))
1322: offs++;
1323: if (line[offs] == '\0')
1.38 schwarze 1324: continue;
1.209 schwarze 1325: title = mandoc_realloc(title, titlesz + len - offs);
1326: memcpy(title + titlesz, line + offs, len - offs);
1327: titlesz += len - offs;
1.50 kristaps 1328: title[titlesz - 1] = ' ';
1329: }
1.209 schwarze 1330: free(line);
1.38 schwarze 1331:
1.50 kristaps 1332: /*
1333: * If no page content can be found, or the input line
1334: * is already the next section header, or there is no
1335: * trailing newline, reuse the page title as the page
1336: * description.
1337: */
1.44 kristaps 1338:
1.50 kristaps 1339: if (NULL == title || '\0' == *title) {
1.56 schwarze 1340: if (warnings)
1.241 schwarze 1341: say(mlink->file, "Cannot find NAME section");
1.50 kristaps 1342: fclose(stream);
1343: free(title);
1344: return;
1345: }
1.1 kristaps 1346:
1.209 schwarze 1347: title[titlesz - 1] = '\0';
1.33 schwarze 1348:
1.50 kristaps 1349: /*
1350: * Skip to the first dash.
1351: * Use the remaining line as the description (no more than 70
1352: * bytes).
1353: */
1.33 schwarze 1354:
1.50 kristaps 1355: if (NULL != (p = strstr(title, "- "))) {
1356: for (p += 2; ' ' == *p || '\b' == *p; p++)
1357: /* Skip to next word. */ ;
1358: } else {
1.56 schwarze 1359: if (warnings)
1.242 schwarze 1360: say(mlink->file, "No dash in title line, "
1361: "reusing \"%s\" as one-line description", title);
1.50 kristaps 1362: p = title;
1363: }
1.38 schwarze 1364:
1.50 kristaps 1365: plen = strlen(p);
1.1 kristaps 1366:
1.50 kristaps 1367: /* Strip backspace-encoding from line. */
1.1 kristaps 1368:
1.50 kristaps 1369: while (NULL != (line = memchr(p, '\b', plen))) {
1370: len = line - p;
1371: if (0 == len) {
1372: memmove(line, line + 1, plen--);
1373: continue;
1.141 schwarze 1374: }
1.50 kristaps 1375: memmove(line - 1, line + 1, plen - len);
1376: plen -= 2;
1377: }
1.1 kristaps 1378:
1.257 schwarze 1379: /*
1380: * Cut off excessive one-line descriptions.
1381: * Bad pages are not worth better heuristics.
1382: */
1383:
1384: mpage->desc = mandoc_strndup(p, 150);
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:
1.257 schwarze 1528: /*
1529: * Cut off excessive one-line descriptions.
1530: * Bad pages are not worth better heuristics.
1531: */
1532:
1533: mpage->desc = mandoc_strndup(start, 150);
1.50 kristaps 1534: free(title);
1.61 schwarze 1535: return;
1.50 kristaps 1536: }
1537: }
1.1 kristaps 1538:
1.77 schwarze 1539: for (n = n->child; n; n = n->next) {
1.78 schwarze 1540: if (NULL != mpage->desc)
1.77 schwarze 1541: break;
1.174 schwarze 1542: parse_man(mpage, meta, n);
1.77 schwarze 1543: }
1.1 kristaps 1544: }
1545:
1546: static void
1.190 schwarze 1547: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1548: const struct roff_node *n)
1.1 kristaps 1549: {
1.259 schwarze 1550: const struct mdoc_handler *handler;
1.1 kristaps 1551:
1.246 schwarze 1552: for (n = n->child; n != NULL; n = n->next) {
1.259 schwarze 1553: if (n->tok == TOKEN_NONE || n->tok < ROFF_MAX)
1.235 schwarze 1554: continue;
1.247 schwarze 1555: assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1.259 schwarze 1556: handler = mdoc_handlers + (n->tok - MDOC_Dd);
1557: if (n->flags & handler->taboo)
1558: continue;
1559:
1.50 kristaps 1560: switch (n->type) {
1.188 schwarze 1561: case ROFFT_ELEM:
1562: case ROFFT_BLOCK:
1563: case ROFFT_HEAD:
1564: case ROFFT_BODY:
1565: case ROFFT_TAIL:
1.259 schwarze 1566: if (handler->fp != NULL &&
1567: (*handler->fp)(mpage, meta, n) == 0)
1.246 schwarze 1568: break;
1.259 schwarze 1569: if (handler->mask)
1.78 schwarze 1570: putmdockey(mpage, n->child,
1.259 schwarze 1571: handler->mask, handler->taboo);
1.50 kristaps 1572: break;
1573: default:
1574: continue;
1575: }
1576: if (NULL != n->child)
1.172 schwarze 1577: parse_mdoc(mpage, meta, n);
1.1 kristaps 1578: }
1579: }
1580:
1.25 schwarze 1581: static int
1.258 schwarze 1582: parse_mdoc_Fa(struct mpage *mpage, const struct roff_meta *meta,
1583: const struct roff_node *n)
1584: {
1585: uint64_t mask;
1586:
1587: mask = TYPE_Fa;
1588: if (n->sec == SEC_SYNOPSIS)
1589: mask |= TYPE_Vt;
1590:
1591: putmdockey(mpage, n->child, mask, 0);
1592: return 0;
1593: }
1594:
1595: static int
1.190 schwarze 1596: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1597: const struct roff_node *n)
1.1 kristaps 1598: {
1.176 schwarze 1599: char *start, *end;
1.1 kristaps 1600: size_t sz;
1.25 schwarze 1601:
1.50 kristaps 1602: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1603: NULL == (n = n->child) ||
1.188 schwarze 1604: n->type != ROFFT_TEXT)
1.197 schwarze 1605: return 0;
1.1 kristaps 1606:
1607: /*
1608: * Only consider those `Fd' macro fields that begin with an
1609: * "inclusion" token (versus, e.g., #define).
1610: */
1.50 kristaps 1611:
1.1 kristaps 1612: if (strcmp("#include", n->string))
1.197 schwarze 1613: return 0;
1.1 kristaps 1614:
1.188 schwarze 1615: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1616: return 0;
1.1 kristaps 1617:
1618: /*
1619: * Strip away the enclosing angle brackets and make sure we're
1620: * not zero-length.
1621: */
1622:
1623: start = n->string;
1624: if ('<' == *start || '"' == *start)
1625: start++;
1626:
1627: if (0 == (sz = strlen(start)))
1.197 schwarze 1628: return 0;
1.1 kristaps 1629:
1630: end = &start[(int)sz - 1];
1631: if ('>' == *end || '"' == *end)
1632: end--;
1633:
1.50 kristaps 1634: if (end > start)
1.78 schwarze 1635: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1636: return 0;
1.1 kristaps 1637: }
1638:
1.178 schwarze 1639: static void
1.189 schwarze 1640: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1641: {
1.112 schwarze 1642: char *cp;
1.178 schwarze 1643: size_t sz;
1.1 kristaps 1644:
1.188 schwarze 1645: if (n->type != ROFFT_TEXT)
1.178 schwarze 1646: return;
1.25 schwarze 1647:
1.178 schwarze 1648: /* Skip function pointer punctuation. */
1.1 kristaps 1649:
1.178 schwarze 1650: cp = n->string;
1651: while (*cp == '(' || *cp == '*')
1.1 kristaps 1652: cp++;
1.178 schwarze 1653: sz = strcspn(cp, "()");
1.1 kristaps 1654:
1.178 schwarze 1655: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1656: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1657: putkeys(mpage, cp, sz, NAME_SYN);
1658: }
1.25 schwarze 1659:
1.178 schwarze 1660: static int
1.190 schwarze 1661: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1662: const struct roff_node *n)
1.178 schwarze 1663: {
1.258 schwarze 1664: uint64_t mask;
1.178 schwarze 1665:
1666: if (n->child == NULL)
1.197 schwarze 1667: return 0;
1.178 schwarze 1668:
1669: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1670:
1.258 schwarze 1671: n = n->child->next;
1672: if (n != NULL && n->type == ROFFT_TEXT) {
1673: mask = TYPE_Fa;
1674: if (n->sec == SEC_SYNOPSIS)
1675: mask |= TYPE_Vt;
1676: putmdockey(mpage, n, mask, 0);
1677: }
1.25 schwarze 1678:
1.197 schwarze 1679: return 0;
1.173 schwarze 1680: }
1681:
1682: static int
1.190 schwarze 1683: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1684: const struct roff_node *n)
1.173 schwarze 1685: {
1.177 schwarze 1686:
1.188 schwarze 1687: if (n->type != ROFFT_HEAD)
1.197 schwarze 1688: return 1;
1.173 schwarze 1689:
1.178 schwarze 1690: if (n->child != NULL)
1691: parse_mdoc_fname(mpage, n->child);
1692:
1.197 schwarze 1693: return 0;
1.1 kristaps 1694: }
1695:
1.25 schwarze 1696: static int
1.211 schwarze 1697: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1698: const struct roff_node *n)
1699: {
1700: char *cp;
1701:
1702: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1703: return 0;
1704:
1.215 schwarze 1705: if (n->child != NULL &&
1706: n->child->next == NULL &&
1707: n->child->type == ROFFT_TEXT)
1.211 schwarze 1708: return 1;
1709:
1710: cp = NULL;
1711: deroff(&cp, n);
1712: if (cp != NULL) {
1713: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1714: n->type == ROFFT_BODY ? TYPE_Va : 0));
1715: free(cp);
1716: }
1717:
1718: return 0;
1719: }
1720:
1721: static int
1.190 schwarze 1722: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1723: const struct roff_node *n)
1.1 kristaps 1724: {
1.67 schwarze 1725: char *cp;
1.1 kristaps 1726:
1727: if (NULL == (n = n->child))
1.197 schwarze 1728: return 0;
1.1 kristaps 1729:
1.67 schwarze 1730: if (NULL == n->next) {
1.78 schwarze 1731: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1732: return 0;
1.67 schwarze 1733: }
1734:
1.120 schwarze 1735: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1736: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1737: free(cp);
1.197 schwarze 1738: return 0;
1.1 kristaps 1739: }
1740:
1.25 schwarze 1741: static int
1.190 schwarze 1742: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1743: const struct roff_node *n)
1.1 kristaps 1744: {
1745:
1.188 schwarze 1746: if (n->type == ROFFT_BODY)
1.194 schwarze 1747: deroff(&mpage->desc, n);
1.197 schwarze 1748: return 0;
1.1 kristaps 1749: }
1750:
1.25 schwarze 1751: static int
1.190 schwarze 1752: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1753: const struct roff_node *n)
1.1 kristaps 1754: {
1755:
1.129 schwarze 1756: if (SEC_NAME == n->sec)
1.235 schwarze 1757: putmdockey(mpage, n->child, NAME_TITLE, 0);
1.188 schwarze 1758: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1759: if (n->child == NULL)
1760: putkey(mpage, meta->name, NAME_SYN);
1761: else
1.235 schwarze 1762: putmdockey(mpage, n->child, NAME_SYN, 0);
1.174 schwarze 1763: }
1764: if ( ! (mpage->name_head_done ||
1765: n->child == NULL || n->child->string == NULL ||
1766: strcasecmp(n->child->string, meta->title))) {
1.226 schwarze 1767: putkey(mpage, n->child->string, NAME_HEAD);
1.174 schwarze 1768: mpage->name_head_done = 1;
1.172 schwarze 1769: }
1.197 schwarze 1770: return 0;
1.1 kristaps 1771: }
1772:
1.25 schwarze 1773: static int
1.190 schwarze 1774: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1775: const struct roff_node *n)
1.1 kristaps 1776: {
1777:
1.197 schwarze 1778: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1779: }
1780:
1.50 kristaps 1781: static int
1.190 schwarze 1782: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1783: const struct roff_node *n)
1.1 kristaps 1784: {
1785:
1.197 schwarze 1786: return n->type == ROFFT_HEAD;
1.1 kristaps 1787: }
1788:
1.50 kristaps 1789: /*
1.66 schwarze 1790: * Add a string to the hash table for the current manual.
1791: * Each string has a bitmask telling which macros it belongs to.
1792: * When we finish the manual, we'll dump the table.
1.50 kristaps 1793: */
1794: static void
1.176 schwarze 1795: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1796: {
1.133 schwarze 1797: struct ohash *htab;
1.50 kristaps 1798: struct str *s;
1.111 schwarze 1799: const char *end;
1.65 schwarze 1800: unsigned int slot;
1.176 schwarze 1801: int i, mustfree;
1.25 schwarze 1802:
1.50 kristaps 1803: if (0 == sz)
1804: return;
1.111 schwarze 1805:
1.176 schwarze 1806: mustfree = render_string(&cp, &sz);
1807:
1.133 schwarze 1808: if (TYPE_Nm & v) {
1809: htab = &names;
1810: v &= name_mask;
1.169 schwarze 1811: if (v & NAME_FIRST)
1812: name_mask &= ~NAME_FIRST;
1.133 schwarze 1813: if (debug > 1)
1814: say(mpage->mlinks->file,
1.234 schwarze 1815: "Adding name %*s, bits=0x%llx", (int)sz, cp,
1816: (unsigned long long)v);
1.133 schwarze 1817: } else {
1818: htab = &strings;
1819: if (debug > 1)
1.221 schwarze 1820: for (i = 0; i < KEY_MAX; i++)
1.163 schwarze 1821: if ((uint64_t)1 << i & v)
1.133 schwarze 1822: say(mpage->mlinks->file,
1823: "Adding key %s=%*s",
1.220 schwarze 1824: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1825: }
1.25 schwarze 1826:
1.65 schwarze 1827: end = cp + sz;
1.133 schwarze 1828: slot = ohash_qlookupi(htab, cp, &end);
1829: s = ohash_find(htab, slot);
1.25 schwarze 1830:
1.78 schwarze 1831: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1832: s->mask |= v;
1833: return;
1834: } else if (NULL == s) {
1.144 schwarze 1835: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1836: memcpy(s->key, cp, sz);
1.133 schwarze 1837: ohash_insert(htab, slot, s);
1.1 kristaps 1838: }
1.78 schwarze 1839: s->mpage = mpage;
1.50 kristaps 1840: s->mask = v;
1.176 schwarze 1841:
1842: if (mustfree)
1843: free(cp);
1.1 kristaps 1844: }
1845:
1.50 kristaps 1846: /*
1847: * Take a Unicode codepoint and produce its UTF-8 encoding.
1848: * This isn't the best way to do this, but it works.
1849: * The magic numbers are from the UTF-8 packaging.
1850: * They're not as scary as they seem: read the UTF-8 spec for details.
1851: */
1852: static size_t
1853: utf8(unsigned int cp, char out[7])
1.1 kristaps 1854: {
1.50 kristaps 1855: size_t rc;
1.1 kristaps 1856:
1.50 kristaps 1857: rc = 0;
1858: if (cp <= 0x0000007F) {
1859: rc = 1;
1860: out[0] = (char)cp;
1861: } else if (cp <= 0x000007FF) {
1862: rc = 2;
1863: out[0] = (cp >> 6 & 31) | 192;
1864: out[1] = (cp & 63) | 128;
1865: } else if (cp <= 0x0000FFFF) {
1866: rc = 3;
1867: out[0] = (cp >> 12 & 15) | 224;
1868: out[1] = (cp >> 6 & 63) | 128;
1869: out[2] = (cp & 63) | 128;
1870: } else if (cp <= 0x001FFFFF) {
1871: rc = 4;
1872: out[0] = (cp >> 18 & 7) | 240;
1873: out[1] = (cp >> 12 & 63) | 128;
1874: out[2] = (cp >> 6 & 63) | 128;
1875: out[3] = (cp & 63) | 128;
1876: } else if (cp <= 0x03FFFFFF) {
1877: rc = 5;
1878: out[0] = (cp >> 24 & 3) | 248;
1879: out[1] = (cp >> 18 & 63) | 128;
1880: out[2] = (cp >> 12 & 63) | 128;
1881: out[3] = (cp >> 6 & 63) | 128;
1882: out[4] = (cp & 63) | 128;
1883: } else if (cp <= 0x7FFFFFFF) {
1884: rc = 6;
1885: out[0] = (cp >> 30 & 1) | 252;
1886: out[1] = (cp >> 24 & 63) | 128;
1887: out[2] = (cp >> 18 & 63) | 128;
1888: out[3] = (cp >> 12 & 63) | 128;
1889: out[4] = (cp >> 6 & 63) | 128;
1890: out[5] = (cp & 63) | 128;
1891: } else
1.197 schwarze 1892: return 0;
1.1 kristaps 1893:
1.50 kristaps 1894: out[rc] = '\0';
1.197 schwarze 1895: return rc;
1.50 kristaps 1896: }
1.1 kristaps 1897:
1.50 kristaps 1898: /*
1.176 schwarze 1899: * If the string contains escape sequences,
1900: * replace it with an allocated rendering and return 1,
1901: * such that the caller can free it after use.
1902: * Otherwise, do nothing and return 0.
1.50 kristaps 1903: */
1.176 schwarze 1904: static int
1905: render_string(char **public, size_t *psz)
1.50 kristaps 1906: {
1.176 schwarze 1907: const char *src, *scp, *addcp, *seq;
1908: char *dst;
1909: size_t ssz, dsz, addsz;
1.114 schwarze 1910: char utfbuf[7], res[6];
1.176 schwarze 1911: int seqlen, unicode;
1.50 kristaps 1912:
1913: res[0] = '\\';
1914: res[1] = '\t';
1915: res[2] = ASCII_NBRSP;
1916: res[3] = ASCII_HYPH;
1.114 schwarze 1917: res[4] = ASCII_BREAK;
1918: res[5] = '\0';
1.1 kristaps 1919:
1.176 schwarze 1920: src = scp = *public;
1921: ssz = *psz;
1922: dst = NULL;
1923: dsz = 0;
1924:
1925: while (scp < src + *psz) {
1926:
1927: /* Leave normal characters unchanged. */
1928:
1929: if (strchr(res, *scp) == NULL) {
1930: if (dst != NULL)
1931: dst[dsz++] = *scp;
1932: scp++;
1933: continue;
1934: }
1.46 kristaps 1935:
1.50 kristaps 1936: /*
1.176 schwarze 1937: * Found something that requires replacing,
1938: * make sure we have a destination buffer.
1.50 kristaps 1939: */
1.176 schwarze 1940:
1941: if (dst == NULL) {
1942: dst = mandoc_malloc(ssz + 1);
1943: dsz = scp - src;
1944: memcpy(dst, src, dsz);
1.50 kristaps 1945: }
1.46 kristaps 1946:
1.176 schwarze 1947: /* Handle single-char special characters. */
1948:
1949: switch (*scp) {
1950: case '\\':
1951: break;
1.141 schwarze 1952: case '\t':
1953: case ASCII_NBRSP:
1.176 schwarze 1954: dst[dsz++] = ' ';
1955: scp++;
1956: continue;
1957: case ASCII_HYPH:
1958: dst[dsz++] = '-';
1.114 schwarze 1959: /* FALLTHROUGH */
1.141 schwarze 1960: case ASCII_BREAK:
1.176 schwarze 1961: scp++;
1.50 kristaps 1962: continue;
1.114 schwarze 1963: default:
1.176 schwarze 1964: abort();
1.114 schwarze 1965: }
1.46 kristaps 1966:
1.50 kristaps 1967: /*
1.176 schwarze 1968: * Found an escape sequence.
1969: * Read past the slash, then parse it.
1970: * Ignore everything except characters.
1.50 kristaps 1971: */
1.95 schwarze 1972:
1.176 schwarze 1973: scp++;
1974: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1975: continue;
1.1 kristaps 1976:
1.50 kristaps 1977: /*
1.95 schwarze 1978: * Render the special character
1979: * as either UTF-8 or ASCII.
1.50 kristaps 1980: */
1.95 schwarze 1981:
1982: if (write_utf8) {
1.203 schwarze 1983: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1984: if (unicode <= 0)
1.95 schwarze 1985: continue;
1.176 schwarze 1986: addsz = utf8(unicode, utfbuf);
1987: if (addsz == 0)
1.95 schwarze 1988: continue;
1.176 schwarze 1989: addcp = utfbuf;
1.95 schwarze 1990: } else {
1.203 schwarze 1991: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1992: if (addcp == NULL)
1.95 schwarze 1993: continue;
1.176 schwarze 1994: if (*addcp == ASCII_NBRSP) {
1995: addcp = " ";
1996: addsz = 1;
1.95 schwarze 1997: }
1998: }
1.1 kristaps 1999:
1.50 kristaps 2000: /* Copy the rendered glyph into the stream. */
1.1 kristaps 2001:
1.176 schwarze 2002: ssz += addsz;
2003: dst = mandoc_realloc(dst, ssz + 1);
2004: memcpy(dst + dsz, addcp, addsz);
2005: dsz += addsz;
2006: }
2007: if (dst != NULL) {
2008: *public = dst;
2009: *psz = dsz;
2010: }
2011:
2012: /* Trim trailing whitespace and NUL-terminate. */
2013:
2014: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2015: --*psz;
2016: if (dst != NULL) {
2017: (*public)[*psz] = '\0';
1.197 schwarze 2018: return 1;
1.176 schwarze 2019: } else
1.197 schwarze 2020: return 0;
1.1 kristaps 2021: }
2022:
1.118 schwarze 2023: static void
2024: dbadd_mlink(const struct mlink *mlink)
2025: {
1.221 schwarze 2026: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
2027: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1.223 schwarze 2028: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1.221 schwarze 2029: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
2030: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.118 schwarze 2031: }
2032:
1.14 schwarze 2033: /*
1.50 kristaps 2034: * Flush the current page's terms (and their bits) into the database.
1.96 schwarze 2035: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2036: */
2037: static void
1.221 schwarze 2038: dbadd(struct dba *dba, struct mpage *mpage)
1.14 schwarze 2039: {
1.87 schwarze 2040: struct mlink *mlink;
1.50 kristaps 2041: struct str *key;
1.176 schwarze 2042: char *cp;
1.221 schwarze 2043: uint64_t mask;
1.52 kristaps 2044: size_t i;
1.66 schwarze 2045: unsigned int slot;
1.176 schwarze 2046: int mustfree;
1.14 schwarze 2047:
1.128 schwarze 2048: mlink = mpage->mlinks;
1.43 kristaps 2049:
1.128 schwarze 2050: if (nodb) {
1.147 schwarze 2051: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2052: key = ohash_next(&names, &slot))
1.147 schwarze 2053: free(key);
2054: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2055: key = ohash_next(&strings, &slot))
1.147 schwarze 2056: free(key);
1.145 schwarze 2057: if (0 == debug)
2058: return;
1.128 schwarze 2059: while (NULL != mlink) {
2060: fputs(mlink->name, stdout);
2061: if (NULL == mlink->next ||
2062: strcmp(mlink->dsec, mlink->next->dsec) ||
2063: strcmp(mlink->fsec, mlink->next->fsec) ||
2064: strcmp(mlink->arch, mlink->next->arch)) {
2065: putchar('(');
2066: if ('\0' == *mlink->dsec)
2067: fputs(mlink->fsec, stdout);
2068: else
2069: fputs(mlink->dsec, stdout);
2070: if ('\0' != *mlink->arch)
2071: printf("/%s", mlink->arch);
2072: putchar(')');
2073: }
2074: mlink = mlink->next;
2075: if (NULL != mlink)
2076: fputs(", ", stdout);
2077: }
1.132 schwarze 2078: printf(" - %s\n", mpage->desc);
1.14 schwarze 2079: return;
1.128 schwarze 2080: }
2081:
2082: if (debug)
2083: say(mlink->file, "Adding to database");
1.28 kristaps 2084:
1.176 schwarze 2085: cp = mpage->desc;
2086: i = strlen(cp);
2087: mustfree = render_string(&cp, &i);
1.227 schwarze 2088: mpage->dba = dba_page_new(dba->pages,
1.223 schwarze 2089: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2090: cp, mlink->file, mpage->form);
1.176 schwarze 2091: if (mustfree)
2092: free(cp);
1.227 schwarze 2093: dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
1.81 schwarze 2094:
1.223 schwarze 2095: while (mlink != NULL) {
1.118 schwarze 2096: dbadd_mlink(mlink);
1.223 schwarze 2097: mlink = mlink->next;
2098: }
1.50 kristaps 2099:
1.133 schwarze 2100: for (key = ohash_first(&names, &slot); NULL != key;
2101: key = ohash_next(&names, &slot)) {
2102: assert(key->mpage == mpage);
1.221 schwarze 2103: dba_page_alias(mpage->dba, key->key, key->mask);
1.133 schwarze 2104: free(key);
2105: }
1.66 schwarze 2106: for (key = ohash_first(&strings, &slot); NULL != key;
2107: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2108: assert(key->mpage == mpage);
1.221 schwarze 2109: i = 0;
2110: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2111: if (key->mask & mask)
2112: dba_macro_add(dba->macros, i,
2113: key->key, mpage->dba);
2114: i++;
2115: }
1.66 schwarze 2116: free(key);
1.38 schwarze 2117: }
1.14 schwarze 2118: }
2119:
1.5 kristaps 2120: static void
1.221 schwarze 2121: dbprune(struct dba *dba)
1.5 kristaps 2122: {
1.221 schwarze 2123: struct dba_array *page, *files;
2124: char *file;
1.5 kristaps 2125:
1.221 schwarze 2126: dba_array_FOREACH(dba->pages, page) {
2127: files = dba_array_get(page, DBP_FILE);
2128: dba_array_FOREACH(files, file) {
2129: if (*file < ' ')
2130: file++;
2131: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2132: file)) != NULL) {
2133: if (debug)
2134: say(file, "Deleting from database");
2135: dba_array_del(dba->pages);
2136: break;
2137: }
1.106 schwarze 2138: }
1.5 kristaps 2139: }
2140: }
2141:
1.4 kristaps 2142: /*
1.221 schwarze 2143: * Write the database from memory to disk.
1.4 kristaps 2144: */
1.35 kristaps 2145: static void
1.221 schwarze 2146: dbwrite(struct dba *dba)
1.4 kristaps 2147: {
1.255 schwarze 2148: struct stat sb1, sb2;
2149: char tfn[33], *cp1, *cp2;
2150: off_t i;
2151: int fd1, fd2;
1.4 kristaps 2152:
1.250 schwarze 2153: /*
2154: * Do not write empty databases, and delete existing ones
2155: * when makewhatis -u causes them to become empty.
2156: */
2157:
2158: dba_array_start(dba->pages);
2159: if (dba_array_next(dba->pages) == NULL) {
1.252 schwarze 2160: if (unlink(MANDOC_DB) == -1 && errno != ENOENT)
1.250 schwarze 2161: say(MANDOC_DB, "&unlink");
2162: return;
2163: }
2164:
2165: /*
2166: * Build the database in a temporary file,
2167: * then atomically move it into place.
2168: */
2169:
1.221 schwarze 2170: if (dba_write(MANDOC_DB "~", dba) != -1) {
2171: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2172: exitcode = (int)MANDOCLEVEL_SYSERR;
2173: say(MANDOC_DB, "&rename");
2174: unlink(MANDOC_DB "~");
2175: }
1.38 schwarze 2176: return;
1.4 kristaps 2177: }
1.250 schwarze 2178:
2179: /*
2180: * We lack write permission and cannot replace the database
2181: * file, but let's at least check whether the data changed.
2182: */
1.4 kristaps 2183:
1.221 schwarze 2184: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2185: if (mkdtemp(tfn) == NULL) {
2186: exitcode = (int)MANDOCLEVEL_SYSERR;
2187: say("", "&%s", tfn);
1.50 kristaps 2188: return;
1.221 schwarze 2189: }
1.256 schwarze 2190: cp1 = cp2 = MAP_FAILED;
1.255 schwarze 2191: fd1 = fd2 = -1;
1.221 schwarze 2192: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2193: if (dba_write(tfn, dba) == -1) {
2194: say(tfn, "&dba_write");
1.255 schwarze 2195: goto err;
1.115 schwarze 2196: }
1.255 schwarze 2197: if ((fd1 = open(MANDOC_DB, O_RDONLY, 0)) == -1) {
2198: say(MANDOC_DB, "&open");
2199: goto err;
2200: }
2201: if ((fd2 = open(tfn, O_RDONLY, 0)) == -1) {
2202: say(tfn, "&open");
2203: goto err;
2204: }
2205: if (fstat(fd1, &sb1) == -1) {
2206: say(MANDOC_DB, "&fstat");
2207: goto err;
2208: }
2209: if (fstat(fd2, &sb2) == -1) {
2210: say(tfn, "&fstat");
2211: goto err;
2212: }
2213: if (sb1.st_size != sb2.st_size)
2214: goto err;
2215: if ((cp1 = mmap(NULL, sb1.st_size, PROT_READ, MAP_PRIVATE,
1.256 schwarze 2216: fd1, 0)) == MAP_FAILED) {
1.255 schwarze 2217: say(MANDOC_DB, "&mmap");
2218: goto err;
2219: }
2220: if ((cp2 = mmap(NULL, sb2.st_size, PROT_READ, MAP_PRIVATE,
1.256 schwarze 2221: fd2, 0)) == MAP_FAILED) {
1.255 schwarze 2222: say(tfn, "&mmap");
2223: goto err;
2224: }
2225: for (i = 0; i < sb1.st_size; i++)
2226: if (cp1[i] != cp2[i])
2227: goto err;
2228: goto out;
2229:
2230: err:
2231: exitcode = (int)MANDOCLEVEL_SYSERR;
2232: say(MANDOC_DB, "Data changed, but cannot replace database");
1.115 schwarze 2233:
1.221 schwarze 2234: out:
1.256 schwarze 2235: if (cp1 != MAP_FAILED)
1.255 schwarze 2236: munmap(cp1, sb1.st_size);
1.256 schwarze 2237: if (cp2 != MAP_FAILED)
1.255 schwarze 2238: munmap(cp2, sb2.st_size);
2239: if (fd1 != -1)
2240: close(fd1);
2241: if (fd2 != -1)
2242: close(fd2);
1.254 schwarze 2243: unlink(tfn);
1.221 schwarze 2244: *strrchr(tfn, '/') = '\0';
1.254 schwarze 2245: rmdir(tfn);
1.4 kristaps 2246: }
2247:
1.50 kristaps 2248: static int
1.165 schwarze 2249: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2250: {
1.59 schwarze 2251: static char startdir[PATH_MAX];
1.151 schwarze 2252: static int getcwd_status; /* 1 = ok, 2 = failure */
2253: static int chdir_status; /* 1 = changed directory */
1.4 kristaps 2254:
1.59 schwarze 2255: /*
1.151 schwarze 2256: * Remember the original working directory, if possible.
2257: * This will be needed if the second or a later directory
2258: * on the command line is given as a relative path.
2259: * Do not error out if the current directory is not
2260: * searchable: Maybe it won't be needed after all.
2261: */
1.264 schwarze 2262: if (getcwd_status == 0) {
2263: if (getcwd(startdir, sizeof(startdir)) == NULL) {
1.151 schwarze 2264: getcwd_status = 2;
2265: (void)strlcpy(startdir, strerror(errno),
2266: sizeof(startdir));
2267: } else
2268: getcwd_status = 1;
2269: }
2270:
2271: /*
2272: * We are leaving the old base directory.
2273: * Do not use it any longer, not even for messages.
2274: */
2275: *basedir = '\0';
1.264 schwarze 2276: basedir_len = 0;
1.151 schwarze 2277:
2278: /*
2279: * If and only if the directory was changed earlier and
2280: * the next directory to process is given as a relative path,
2281: * first go back, or bail out if that is impossible.
1.59 schwarze 2282: */
1.264 schwarze 2283: if (chdir_status && *targetdir != '/') {
2284: if (getcwd_status == 2) {
1.59 schwarze 2285: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2286: say("", "getcwd: %s", startdir);
1.197 schwarze 2287: return 0;
1.59 schwarze 2288: }
1.264 schwarze 2289: if (chdir(startdir) == -1) {
1.59 schwarze 2290: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2291: say("", "&chdir %s", startdir);
1.197 schwarze 2292: return 0;
1.59 schwarze 2293: }
2294: }
1.151 schwarze 2295:
2296: /*
2297: * Always resolve basedir to the canonicalized absolute
2298: * pathname and append a trailing slash, such that
2299: * we can reliably check whether files are inside.
2300: */
1.264 schwarze 2301: if (realpath(targetdir, basedir) == NULL) {
1.165 schwarze 2302: if (report_baddir || errno != ENOENT) {
2303: exitcode = (int)MANDOCLEVEL_BADARG;
2304: say("", "&%s: realpath", targetdir);
2305: }
1.264 schwarze 2306: *basedir = '\0';
1.197 schwarze 2307: return 0;
1.264 schwarze 2308: } else if (chdir(basedir) == -1) {
1.165 schwarze 2309: if (report_baddir || errno != ENOENT) {
2310: exitcode = (int)MANDOCLEVEL_BADARG;
2311: say("", "&chdir");
2312: }
1.264 schwarze 2313: *basedir = '\0';
1.197 schwarze 2314: return 0;
1.150 schwarze 2315: }
1.151 schwarze 2316: chdir_status = 1;
1.264 schwarze 2317: basedir_len = strlen(basedir);
2318: if (basedir[basedir_len - 1] != '/') {
2319: if (basedir_len >= PATH_MAX - 1) {
1.150 schwarze 2320: exitcode = (int)MANDOCLEVEL_SYSERR;
2321: say("", "Filename too long");
1.264 schwarze 2322: *basedir = '\0';
2323: basedir_len = 0;
1.197 schwarze 2324: return 0;
1.150 schwarze 2325: }
1.264 schwarze 2326: basedir[basedir_len++] = '/';
2327: basedir[basedir_len] = '\0';
1.4 kristaps 2328: }
1.197 schwarze 2329: return 1;
1.56 schwarze 2330: }
2331:
2332: static void
1.59 schwarze 2333: say(const char *file, const char *format, ...)
1.56 schwarze 2334: {
2335: va_list ap;
1.123 schwarze 2336: int use_errno;
1.56 schwarze 2337:
1.264 schwarze 2338: if (*basedir != '\0')
1.59 schwarze 2339: fprintf(stderr, "%s", basedir);
1.264 schwarze 2340: if (*basedir != '\0' && *file != '\0')
1.151 schwarze 2341: fputc('/', stderr);
1.264 schwarze 2342: if (*file != '\0')
1.59 schwarze 2343: fprintf(stderr, "%s", file);
2344:
1.123 schwarze 2345: use_errno = 1;
1.264 schwarze 2346: if (format != NULL) {
1.123 schwarze 2347: switch (*format) {
1.141 schwarze 2348: case '&':
1.123 schwarze 2349: format++;
2350: break;
1.141 schwarze 2351: case '\0':
1.123 schwarze 2352: format = NULL;
2353: break;
2354: default:
2355: use_errno = 0;
2356: break;
2357: }
2358: }
1.264 schwarze 2359: if (format != NULL) {
2360: if (*basedir != '\0' || *file != '\0')
1.123 schwarze 2361: fputs(": ", stderr);
2362: va_start(ap, format);
2363: vfprintf(stderr, format, ap);
2364: va_end(ap);
2365: }
2366: if (use_errno) {
1.264 schwarze 2367: if (*basedir != '\0' || *file != '\0' || format != NULL)
1.123 schwarze 2368: fputs(": ", stderr);
1.59 schwarze 2369: perror(NULL);
1.123 schwarze 2370: } else
2371: fputc('\n', stderr);
1.1 kristaps 2372: }
CVSweb