Annotation of mandoc/mandocdb.c, Revision 1.223
1.223 ! schwarze 1: /* $Id: mandocdb.c,v 1.222 2016/07/19 22:40:33 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.211 schwarze 4: * Copyright (c) 2011-2016 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.187 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18: #include "config.h"
19:
1.162 schwarze 20: #include <sys/types.h>
1.50 kristaps 21: #include <sys/stat.h>
1.115 schwarze 22: #include <sys/wait.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.208 schwarze 26: #if HAVE_ERR
1.198 schwarze 27: #include <err.h>
1.208 schwarze 28: #endif
1.63 schwarze 29: #include <errno.h>
1.1 kristaps 30: #include <fcntl.h>
1.158 schwarze 31: #if HAVE_FTS
1.50 kristaps 32: #include <fts.h>
1.157 schwarze 33: #else
34: #include "compat_fts.h"
35: #endif
1.58 schwarze 36: #include <limits.h>
1.218 kristaps 37: #if HAVE_SANDBOX_INIT
38: #include <sandbox.h>
39: #endif
1.221 schwarze 40: #include <stdarg.h>
1.50 kristaps 41: #include <stddef.h>
1.59 schwarze 42: #include <stdio.h>
1.1 kristaps 43: #include <stdint.h>
44: #include <stdlib.h>
45: #include <string.h>
1.17 schwarze 46: #include <unistd.h>
1.1 kristaps 47:
1.188 schwarze 48: #include "mandoc_aux.h"
1.202 schwarze 49: #include "mandoc_ohash.h"
1.188 schwarze 50: #include "mandoc.h"
51: #include "roff.h"
1.50 kristaps 52: #include "mdoc.h"
1.1 kristaps 53: #include "man.h"
1.187 schwarze 54: #include "manconf.h"
1.55 kristaps 55: #include "mansearch.h"
1.221 schwarze 56: #include "dba_array.h"
57: #include "dba.h"
1.1 kristaps 58:
1.111 schwarze 59: extern const char *const mansearch_keynames[];
60:
1.50 kristaps 61: enum op {
62: OP_DEFAULT = 0, /* new dbs from dir list or default config */
63: OP_CONFFILE, /* new databases from custom config file */
64: OP_UPDATE, /* delete/add entries in existing database */
65: OP_DELETE, /* delete entries from existing database */
66: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 67: };
68:
1.50 kristaps 69: struct str {
1.78 schwarze 70: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 71: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 72: char key[]; /* rendered text */
1.38 schwarze 73: };
74:
1.79 schwarze 75: struct inodev {
76: ino_t st_ino;
77: dev_t st_dev;
1.50 kristaps 78: };
1.5 kristaps 79:
1.78 schwarze 80: struct mpage {
1.79 schwarze 81: struct inodev inodev; /* used for hashing routine */
1.221 schwarze 82: struct dba_array *dba;
1.82 schwarze 83: char *sec; /* section from file content */
84: char *arch; /* architecture from file content */
85: char *title; /* title from file content */
86: char *desc; /* description from file content */
87: struct mlink *mlinks; /* singly linked list */
1.174 schwarze 88: int name_head_done;
1.221 schwarze 89: enum form form; /* format from file content */
1.82 schwarze 90: };
91:
92: struct mlink {
1.58 schwarze 93: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 94: char *dsec; /* section from directory */
95: char *arch; /* architecture from directory */
96: char *name; /* name from file name (not empty) */
97: char *fsec; /* section from file name suffix */
1.85 schwarze 98: struct mlink *next; /* singly linked list */
1.118 schwarze 99: struct mpage *mpage; /* parent */
1.124 schwarze 100: int gzip; /* filename has a .gz suffix */
1.221 schwarze 101: enum form dform; /* format from directory */
102: enum form fform; /* format from file name suffix */
1.1 kristaps 103: };
104:
1.190 schwarze 105: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 106: const struct roff_node *);
1.1 kristaps 107:
1.50 kristaps 108: struct mdoc_handler {
109: mdoc_fp fp; /* optional handler */
110: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 111: };
1.219 schwarze 112:
113:
114: int mandocdb(int, char *[]);
1.1 kristaps 115:
1.221 schwarze 116: static void dbadd(struct dba *, struct mpage *);
1.118 schwarze 117: static void dbadd_mlink(const struct mlink *mlink);
1.221 schwarze 118: static void dbprune(struct dba *);
119: static void dbwrite(struct dba *);
1.59 schwarze 120: static void filescan(const char *);
1.84 schwarze 121: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 122: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 123: static void mlink_free(struct mlink *);
1.89 schwarze 124: static void mlinks_undupe(struct mpage *);
1.78 schwarze 125: static void mpages_free(void);
1.221 schwarze 126: static void mpages_merge(struct dba *, struct mparse *);
1.124 schwarze 127: static void parse_cat(struct mpage *, int);
1.190 schwarze 128: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 129: const struct roff_node *);
1.190 schwarze 130: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 131: const struct roff_node *);
1.190 schwarze 132: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 133: const struct roff_node *);
1.190 schwarze 134: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 135: const struct roff_node *);
136: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 137: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 138: const struct roff_node *);
1.190 schwarze 139: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 140: const struct roff_node *);
1.190 schwarze 141: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 142: const struct roff_node *);
1.190 schwarze 143: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 144: const struct roff_node *);
1.190 schwarze 145: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 146: const struct roff_node *);
1.211 schwarze 147: static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
148: const struct roff_node *);
1.190 schwarze 149: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 150: const struct roff_node *);
1.112 schwarze 151: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 152: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 153: static void putmdockey(const struct mpage *,
1.189 schwarze 154: const struct roff_node *, uint64_t);
1.176 schwarze 155: static int render_string(char **, size_t *);
1.220 schwarze 156: static void say(const char *, const char *, ...)
157: __attribute__((__format__ (printf, 2, 3)));
1.165 schwarze 158: static int set_basedir(const char *, int);
1.59 schwarze 159: static int treescan(void);
1.50 kristaps 160: static size_t utf8(unsigned int, char [7]);
161:
1.102 schwarze 162: static int nodb; /* no database changes */
1.116 schwarze 163: static int mparse_options; /* abort the parse early */
1.141 schwarze 164: static int use_all; /* use all found files */
165: static int debug; /* print what we're doing */
166: static int warnings; /* warn about crap */
1.95 schwarze 167: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 168: static int exitcode; /* to be returned by main */
1.141 schwarze 169: static enum op op; /* operational mode */
1.59 schwarze 170: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 171: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 172: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 173: static struct ohash names; /* table of all names */
1.50 kristaps 174: static struct ohash strings; /* table of all strings */
1.133 schwarze 175: static uint64_t name_mask;
1.25 schwarze 176:
177: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 178: { NULL, 0 }, /* Ap */
179: { NULL, 0 }, /* Dd */
180: { NULL, 0 }, /* Dt */
181: { NULL, 0 }, /* Os */
182: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
183: { parse_mdoc_head, TYPE_Ss }, /* Ss */
184: { NULL, 0 }, /* Pp */
185: { NULL, 0 }, /* D1 */
186: { NULL, 0 }, /* Dl */
187: { NULL, 0 }, /* Bd */
188: { NULL, 0 }, /* Ed */
189: { NULL, 0 }, /* Bl */
190: { NULL, 0 }, /* El */
191: { NULL, 0 }, /* It */
192: { NULL, 0 }, /* Ad */
193: { NULL, TYPE_An }, /* An */
194: { NULL, TYPE_Ar }, /* Ar */
195: { NULL, TYPE_Cd }, /* Cd */
196: { NULL, TYPE_Cm }, /* Cm */
197: { NULL, TYPE_Dv }, /* Dv */
198: { NULL, TYPE_Er }, /* Er */
199: { NULL, TYPE_Ev }, /* Ev */
200: { NULL, 0 }, /* Ex */
201: { NULL, TYPE_Fa }, /* Fa */
202: { parse_mdoc_Fd, 0 }, /* Fd */
203: { NULL, TYPE_Fl }, /* Fl */
204: { parse_mdoc_Fn, 0 }, /* Fn */
205: { NULL, TYPE_Ft }, /* Ft */
206: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 207: { NULL, TYPE_In }, /* In */
1.68 schwarze 208: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 209: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 210: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 211: { NULL, 0 }, /* Op */
212: { NULL, 0 }, /* Ot */
213: { NULL, TYPE_Pa }, /* Pa */
214: { NULL, 0 }, /* Rv */
1.92 schwarze 215: { NULL, TYPE_St }, /* St */
1.211 schwarze 216: { parse_mdoc_Va, TYPE_Va }, /* Va */
217: { parse_mdoc_Va, TYPE_Vt }, /* Vt */
1.68 schwarze 218: { parse_mdoc_Xr, 0 }, /* Xr */
219: { NULL, 0 }, /* %A */
220: { NULL, 0 }, /* %B */
221: { NULL, 0 }, /* %D */
222: { NULL, 0 }, /* %I */
223: { NULL, 0 }, /* %J */
224: { NULL, 0 }, /* %N */
225: { NULL, 0 }, /* %O */
226: { NULL, 0 }, /* %P */
227: { NULL, 0 }, /* %R */
228: { NULL, 0 }, /* %T */
229: { NULL, 0 }, /* %V */
230: { NULL, 0 }, /* Ac */
231: { NULL, 0 }, /* Ao */
232: { NULL, 0 }, /* Aq */
233: { NULL, TYPE_At }, /* At */
234: { NULL, 0 }, /* Bc */
235: { NULL, 0 }, /* Bf */
236: { NULL, 0 }, /* Bo */
237: { NULL, 0 }, /* Bq */
238: { NULL, TYPE_Bsx }, /* Bsx */
239: { NULL, TYPE_Bx }, /* Bx */
240: { NULL, 0 }, /* Db */
241: { NULL, 0 }, /* Dc */
242: { NULL, 0 }, /* Do */
243: { NULL, 0 }, /* Dq */
244: { NULL, 0 }, /* Ec */
245: { NULL, 0 }, /* Ef */
246: { NULL, TYPE_Em }, /* Em */
247: { NULL, 0 }, /* Eo */
248: { NULL, TYPE_Fx }, /* Fx */
249: { NULL, TYPE_Ms }, /* Ms */
250: { NULL, 0 }, /* No */
251: { NULL, 0 }, /* Ns */
252: { NULL, TYPE_Nx }, /* Nx */
253: { NULL, TYPE_Ox }, /* Ox */
254: { NULL, 0 }, /* Pc */
255: { NULL, 0 }, /* Pf */
256: { NULL, 0 }, /* Po */
257: { NULL, 0 }, /* Pq */
258: { NULL, 0 }, /* Qc */
259: { NULL, 0 }, /* Ql */
260: { NULL, 0 }, /* Qo */
261: { NULL, 0 }, /* Qq */
262: { NULL, 0 }, /* Re */
263: { NULL, 0 }, /* Rs */
264: { NULL, 0 }, /* Sc */
265: { NULL, 0 }, /* So */
266: { NULL, 0 }, /* Sq */
267: { NULL, 0 }, /* Sm */
268: { NULL, 0 }, /* Sx */
269: { NULL, TYPE_Sy }, /* Sy */
270: { NULL, TYPE_Tn }, /* Tn */
271: { NULL, 0 }, /* Ux */
272: { NULL, 0 }, /* Xc */
273: { NULL, 0 }, /* Xo */
1.173 schwarze 274: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 275: { NULL, 0 }, /* Fc */
276: { NULL, 0 }, /* Oo */
277: { NULL, 0 }, /* Oc */
278: { NULL, 0 }, /* Bk */
279: { NULL, 0 }, /* Ek */
280: { NULL, 0 }, /* Bt */
281: { NULL, 0 }, /* Hf */
282: { NULL, 0 }, /* Fr */
283: { NULL, 0 }, /* Ud */
284: { NULL, TYPE_Lb }, /* Lb */
285: { NULL, 0 }, /* Lp */
286: { NULL, TYPE_Lk }, /* Lk */
287: { NULL, TYPE_Mt }, /* Mt */
288: { NULL, 0 }, /* Brq */
289: { NULL, 0 }, /* Bro */
290: { NULL, 0 }, /* Brc */
291: { NULL, 0 }, /* %C */
292: { NULL, 0 }, /* Es */
293: { NULL, 0 }, /* En */
294: { NULL, TYPE_Dx }, /* Dx */
295: { NULL, 0 }, /* %Q */
296: { NULL, 0 }, /* br */
297: { NULL, 0 }, /* sp */
298: { NULL, 0 }, /* %U */
299: { NULL, 0 }, /* Ta */
1.166 schwarze 300: { NULL, 0 }, /* ll */
1.1 kristaps 301: };
302:
1.141 schwarze 303:
1.1 kristaps 304: int
1.179 schwarze 305: mandocdb(int argc, char *argv[])
1.1 kristaps 306: {
1.187 schwarze 307: struct manconf conf;
308: struct mparse *mp;
1.221 schwarze 309: struct dba *dba;
1.206 schwarze 310: const char *path_arg, *progname;
1.187 schwarze 311: size_t j, sz;
1.59 schwarze 312: int ch, i;
1.50 kristaps 313:
1.207 schwarze 314: #if HAVE_PLEDGE
315: if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
1.214 schwarze 316: warn("pledge");
1.218 kristaps 317: return (int)MANDOCLEVEL_SYSERR;
318: }
319: #endif
320:
321: #if HAVE_SANDBOX_INIT
322: if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
323: warnx("sandbox_init");
1.207 schwarze 324: return (int)MANDOCLEVEL_SYSERR;
325: }
326: #endif
327:
1.187 schwarze 328: memset(&conf, 0, sizeof(conf));
1.50 kristaps 329:
330: /*
1.141 schwarze 331: * We accept a few different invocations.
1.50 kristaps 332: * The CHECKOP macro makes sure that invocation styles don't
333: * clobber each other.
334: */
335: #define CHECKOP(_op, _ch) do \
336: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 337: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 338: goto usage; \
339: } while (/*CONSTCOND*/0)
1.10 kristaps 340:
1.59 schwarze 341: path_arg = NULL;
1.38 schwarze 342: op = OP_DEFAULT;
1.1 kristaps 343:
1.126 schwarze 344: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 345: switch (ch) {
1.141 schwarze 346: case 'a':
1.12 schwarze 347: use_all = 1;
348: break;
1.141 schwarze 349: case 'C':
1.50 kristaps 350: CHECKOP(op, ch);
1.59 schwarze 351: path_arg = optarg;
1.38 schwarze 352: op = OP_CONFFILE;
1.34 schwarze 353: break;
1.141 schwarze 354: case 'D':
1.125 schwarze 355: debug++;
356: break;
1.141 schwarze 357: case 'd':
1.50 kristaps 358: CHECKOP(op, ch);
1.59 schwarze 359: path_arg = optarg;
1.5 kristaps 360: op = OP_UPDATE;
361: break;
1.141 schwarze 362: case 'n':
1.50 kristaps 363: nodb = 1;
364: break;
1.141 schwarze 365: case 'p':
1.126 schwarze 366: warnings = 1;
367: break;
1.141 schwarze 368: case 'Q':
1.116 schwarze 369: mparse_options |= MPARSE_QUICK;
1.102 schwarze 370: break;
1.141 schwarze 371: case 'T':
1.95 schwarze 372: if (strcmp(optarg, "utf8")) {
1.198 schwarze 373: warnx("-T%s: Unsupported output format",
374: optarg);
1.95 schwarze 375: goto usage;
376: }
377: write_utf8 = 1;
378: break;
1.141 schwarze 379: case 't':
1.50 kristaps 380: CHECKOP(op, ch);
1.38 schwarze 381: dup2(STDOUT_FILENO, STDERR_FILENO);
382: op = OP_TEST;
1.50 kristaps 383: nodb = warnings = 1;
1.38 schwarze 384: break;
1.141 schwarze 385: case 'u':
1.50 kristaps 386: CHECKOP(op, ch);
1.59 schwarze 387: path_arg = optarg;
1.5 kristaps 388: op = OP_DELETE;
389: break;
1.141 schwarze 390: case 'v':
1.126 schwarze 391: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 392: break;
1.1 kristaps 393: default:
1.38 schwarze 394: goto usage;
1.1 kristaps 395: }
396:
397: argc -= optind;
398: argv += optind;
399:
1.207 schwarze 400: #if HAVE_PLEDGE
1.210 schwarze 401: if (nodb) {
402: if (pledge("stdio rpath", NULL) == -1) {
1.214 schwarze 403: warn("pledge");
1.210 schwarze 404: return (int)MANDOCLEVEL_SYSERR;
405: }
1.207 schwarze 406: }
407: #endif
408:
1.38 schwarze 409: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 410: warnx("-C: Too many arguments");
1.38 schwarze 411: goto usage;
412: }
413:
1.59 schwarze 414: exitcode = (int)MANDOCLEVEL_OK;
1.203 schwarze 415: mchars_alloc();
416: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
1.202 schwarze 417: mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
418: mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
1.50 kristaps 419:
420: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 421:
1.50 kristaps 422: /*
1.150 schwarze 423: * Most of these deal with a specific directory.
1.140 schwarze 424: * Jump into that directory first.
1.50 kristaps 425: */
1.165 schwarze 426: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 427: goto out;
1.140 schwarze 428:
1.221 schwarze 429: if ((dba = dba_read(MANDOC_DB)) != NULL) {
1.140 schwarze 430: /*
431: * The existing database is usable. Process
432: * all files specified on the command-line.
433: */
1.207 schwarze 434: #if HAVE_PLEDGE
1.210 schwarze 435: if (!nodb) {
436: if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
1.214 schwarze 437: warn("pledge");
1.210 schwarze 438: exitcode = (int)MANDOCLEVEL_SYSERR;
439: goto out;
440: }
1.207 schwarze 441: }
442: #endif
1.140 schwarze 443: use_all = 1;
1.138 schwarze 444: for (i = 0; i < argc; i++)
445: filescan(argv[i]);
446: if (OP_TEST != op)
1.221 schwarze 447: dbprune(dba);
1.138 schwarze 448: } else {
449: /*
450: * Database missing or corrupt.
451: * Recreate from scratch.
452: */
1.140 schwarze 453: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 454: op = OP_DEFAULT;
455: if (0 == treescan())
456: goto out;
1.221 schwarze 457: dba = dba_new(128);
1.138 schwarze 458: }
1.50 kristaps 459: if (OP_DELETE != op)
1.221 schwarze 460: mpages_merge(dba, mp);
461: if (nodb == 0)
462: dbwrite(dba);
463: dba_free(dba);
1.50 kristaps 464: } else {
465: /*
466: * If we have arguments, use them as our manpaths.
1.222 schwarze 467: * If we don't, use man.conf(5).
1.50 kristaps 468: */
469: if (argc > 0) {
1.187 schwarze 470: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 471: argc, sizeof(char *));
1.187 schwarze 472: conf.manpath.sz = (size_t)argc;
1.50 kristaps 473: for (i = 0; i < argc; i++)
1.187 schwarze 474: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 475: } else
1.187 schwarze 476: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 477:
1.187 schwarze 478: if (conf.manpath.sz == 0) {
1.127 schwarze 479: exitcode = (int)MANDOCLEVEL_BADARG;
480: say("", "Empty manpath");
481: }
1.50 kristaps 482:
483: /*
1.69 schwarze 484: * First scan the tree rooted at a base directory, then
485: * build a new database and finally move it into place.
1.50 kristaps 486: * Ignore zero-length directories and strip trailing
487: * slashes.
488: */
1.187 schwarze 489: for (j = 0; j < conf.manpath.sz; j++) {
490: sz = strlen(conf.manpath.paths[j]);
491: if (sz && conf.manpath.paths[j][sz - 1] == '/')
492: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 493: if (0 == sz)
494: continue;
1.66 schwarze 495:
496: if (j) {
1.202 schwarze 497: mandoc_ohash_init(&mpages, 6,
498: offsetof(struct mpage, inodev));
499: mandoc_ohash_init(&mlinks, 6,
500: offsetof(struct mlink, file));
1.66 schwarze 501: }
502:
1.187 schwarze 503: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 504: continue;
1.59 schwarze 505: if (0 == treescan())
1.165 schwarze 506: continue;
1.221 schwarze 507: dba = dba_new(128);
508: mpages_merge(dba, mp);
509: if (nodb == 0)
510: dbwrite(dba);
511: dba_free(dba);
1.66 schwarze 512:
1.187 schwarze 513: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 514: mpages_free();
1.78 schwarze 515: ohash_delete(&mpages);
1.83 schwarze 516: ohash_delete(&mlinks);
1.66 schwarze 517: }
1.50 kristaps 518: }
519: }
520: out:
1.187 schwarze 521: manconf_free(&conf);
1.168 schwarze 522: mparse_free(mp);
1.203 schwarze 523: mchars_free();
1.80 schwarze 524: mpages_free();
1.78 schwarze 525: ohash_delete(&mpages);
1.83 schwarze 526: ohash_delete(&mlinks);
1.197 schwarze 527: return exitcode;
1.50 kristaps 528: usage:
1.206 schwarze 529: progname = getprogname();
1.126 schwarze 530: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
531: " %s [-aDnpQ] [-Tutf8] dir ...\n"
532: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
533: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 534: " %s [-Q] -t file ...\n",
1.206 schwarze 535: progname, progname, progname, progname, progname);
1.50 kristaps 536:
1.197 schwarze 537: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 538: }
539:
540: /*
1.59 schwarze 541: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 542: * We use fts(), scanning directory parts along the way for clues to our
543: * section and architecture.
544: *
545: * If use_all has been specified, grok all files.
546: * If not, sanitise paths to the following:
547: *
1.141 schwarze 548: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 549: * or
550: * [./]cat<section>[/<arch>]/<name>.0
551: *
1.216 schwarze 552: * TODO: accommodate for multi-language directories.
1.50 kristaps 553: */
554: static int
1.59 schwarze 555: treescan(void)
1.50 kristaps 556: {
1.139 schwarze 557: char buf[PATH_MAX];
1.50 kristaps 558: FTS *f;
559: FTSENT *ff;
1.84 schwarze 560: struct mlink *mlink;
1.221 schwarze 561: int gzip;
562: enum form dform;
1.94 schwarze 563: char *dsec, *arch, *fsec, *cp;
564: const char *path;
1.50 kristaps 565: const char *argv[2];
566:
567: argv[0] = ".";
568: argv[1] = (char *)NULL;
569:
1.139 schwarze 570: f = fts_open((char * const *)argv,
571: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 572: if (f == NULL) {
1.59 schwarze 573: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 574: say("", "&fts_open");
1.197 schwarze 575: return 0;
1.50 kristaps 576: }
577:
578: dsec = arch = NULL;
579: dform = FORM_NONE;
580:
1.200 schwarze 581: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 582: path = ff->fts_path + 2;
1.139 schwarze 583: switch (ff->fts_info) {
584:
585: /*
586: * Symbolic links require various sanity checks,
587: * then get handled just like regular files.
588: */
1.141 schwarze 589: case FTS_SL:
1.200 schwarze 590: if (realpath(path, buf) == NULL) {
1.139 schwarze 591: if (warnings)
592: say(path, "&realpath");
593: continue;
594: }
1.184 schwarze 595: if (strstr(buf, basedir) != buf
596: #ifdef HOMEBREWDIR
597: && strstr(buf, HOMEBREWDIR) != buf
598: #endif
599: ) {
1.139 schwarze 600: if (warnings) say("",
601: "%s: outside base directory", buf);
602: continue;
603: }
604: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 605: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 606: if (warnings)
607: say(path, "&stat");
608: continue;
609: }
610: /* FALLTHROUGH */
611:
1.50 kristaps 612: /*
1.83 schwarze 613: * If we're a regular file, add an mlink by using the
1.50 kristaps 614: * stored directory data and handling the filename.
615: */
1.141 schwarze 616: case FTS_F:
1.200 schwarze 617: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 618: continue;
1.50 kristaps 619: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 620: if (warnings)
1.59 schwarze 621: say(path, "Extraneous file");
1.50 kristaps 622: continue;
1.124 schwarze 623: }
624: gzip = 0;
625: fsec = NULL;
1.200 schwarze 626: while (fsec == NULL) {
1.124 schwarze 627: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 628: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 629: break;
630: gzip = 1;
631: *fsec = '\0';
632: fsec = NULL;
633: }
1.200 schwarze 634: if (fsec == NULL) {
1.60 schwarze 635: if ( ! use_all) {
1.56 schwarze 636: if (warnings)
1.60 schwarze 637: say(path,
638: "No filename suffix");
1.50 kristaps 639: continue;
640: }
1.200 schwarze 641: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 642: if (warnings)
643: say(path, "Skip html");
644: continue;
1.200 schwarze 645: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 646: if (warnings)
647: say(path, "Skip ps");
648: continue;
1.200 schwarze 649: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 650: if (warnings)
651: say(path, "Skip pdf");
652: continue;
653: } else if ( ! use_all &&
1.200 schwarze 654: ((dform == FORM_SRC &&
1.185 schwarze 655: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 656: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.60 schwarze 657: if (warnings)
658: say(path, "Wrong filename suffix");
659: continue;
1.66 schwarze 660: } else
1.84 schwarze 661: fsec[-1] = '\0';
1.94 schwarze 662:
1.84 schwarze 663: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 664: if (strlcpy(mlink->file, path,
665: sizeof(mlink->file)) >=
666: sizeof(mlink->file)) {
667: say(path, "Filename too long");
668: free(mlink);
669: continue;
670: }
1.84 schwarze 671: mlink->dform = dform;
1.94 schwarze 672: mlink->dsec = dsec;
673: mlink->arch = arch;
674: mlink->name = ff->fts_name;
675: mlink->fsec = fsec;
1.124 schwarze 676: mlink->gzip = gzip;
1.84 schwarze 677: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 678: continue;
1.139 schwarze 679:
1.141 schwarze 680: case FTS_D:
681: case FTS_DP:
1.139 schwarze 682: break;
683:
684: default:
1.60 schwarze 685: if (warnings)
686: say(path, "Not a regular file");
1.50 kristaps 687: continue;
1.60 schwarze 688: }
1.1 kristaps 689:
1.50 kristaps 690: switch (ff->fts_level) {
1.141 schwarze 691: case 0:
1.50 kristaps 692: /* Ignore the root directory. */
693: break;
1.141 schwarze 694: case 1:
1.50 kristaps 695: /*
696: * This might contain manX/ or catX/.
697: * Try to infer this from the name.
698: * If we're not in use_all, enforce it.
699: */
700: cp = ff->fts_name;
1.200 schwarze 701: if (ff->fts_info == FTS_DP) {
702: dform = FORM_NONE;
703: dsec = NULL;
1.50 kristaps 704: break;
1.200 schwarze 705: }
1.1 kristaps 706:
1.200 schwarze 707: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 708: dform = FORM_SRC;
1.66 schwarze 709: dsec = cp + 3;
1.200 schwarze 710: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 711: dform = FORM_CAT;
1.66 schwarze 712: dsec = cp + 3;
1.94 schwarze 713: } else {
714: dform = FORM_NONE;
715: dsec = NULL;
1.50 kristaps 716: }
1.1 kristaps 717:
1.200 schwarze 718: if (dsec != NULL || use_all)
1.50 kristaps 719: break;
1.1 kristaps 720:
1.56 schwarze 721: if (warnings)
1.59 schwarze 722: say(path, "Unknown directory part");
1.50 kristaps 723: fts_set(f, ff, FTS_SKIP);
724: break;
1.141 schwarze 725: case 2:
1.50 kristaps 726: /*
727: * Possibly our architecture.
728: * If we're descending, keep tabs on it.
729: */
1.200 schwarze 730: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 731: arch = ff->fts_name;
1.94 schwarze 732: else
733: arch = NULL;
1.50 kristaps 734: break;
735: default:
1.200 schwarze 736: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 737: break;
1.56 schwarze 738: if (warnings)
1.59 schwarze 739: say(path, "Extraneous directory part");
1.50 kristaps 740: fts_set(f, ff, FTS_SKIP);
741: break;
1.5 kristaps 742: }
1.50 kristaps 743: }
744:
745: fts_close(f);
1.197 schwarze 746: return 1;
1.50 kristaps 747: }
1.5 kristaps 748:
1.50 kristaps 749: /*
1.89 schwarze 750: * Add a file to the mlinks table.
1.50 kristaps 751: * Do not verify that it's a "valid" looking manpage (we'll do that
752: * later).
753: *
754: * Try to infer the manual section, architecture, and page name from the
755: * path, assuming it looks like
756: *
1.141 schwarze 757: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 758: * or
759: * [./]cat<section>[/<arch>]/<name>.0
760: *
761: * See treescan() for the fts(3) version of this.
762: */
763: static void
1.59 schwarze 764: filescan(const char *file)
1.50 kristaps 765: {
1.59 schwarze 766: char buf[PATH_MAX];
1.84 schwarze 767: struct stat st;
768: struct mlink *mlink;
1.59 schwarze 769: char *p, *start;
1.5 kristaps 770:
1.50 kristaps 771: assert(use_all);
1.5 kristaps 772:
1.50 kristaps 773: if (0 == strncmp(file, "./", 2))
774: file += 2;
1.5 kristaps 775:
1.139 schwarze 776: /*
777: * We have to do lstat(2) before realpath(3) loses
778: * the information whether this is a symbolic link.
779: * We need to know that because for symbolic links,
780: * we want to use the orginal file name, while for
781: * regular files, we want to use the real path.
782: */
783: if (-1 == lstat(file, &st)) {
784: exitcode = (int)MANDOCLEVEL_BADARG;
785: say(file, "&lstat");
786: return;
787: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
788: exitcode = (int)MANDOCLEVEL_BADARG;
789: say(file, "Not a regular file");
790: return;
791: }
792:
793: /*
794: * We have to resolve the file name to the real path
795: * in any case for the base directory check.
796: */
1.59 schwarze 797: if (NULL == realpath(file, buf)) {
798: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 799: say(file, "&realpath");
1.59 schwarze 800: return;
1.106 schwarze 801: }
802:
1.150 schwarze 803: if (OP_TEST == op)
1.106 schwarze 804: start = buf;
1.150 schwarze 805: else if (strstr(buf, basedir) == buf)
806: start = buf + strlen(basedir);
1.184 schwarze 807: #ifdef HOMEBREWDIR
808: else if (strstr(buf, HOMEBREWDIR) == buf)
809: start = buf;
810: #endif
1.106 schwarze 811: else {
1.59 schwarze 812: exitcode = (int)MANDOCLEVEL_BADARG;
813: say("", "%s: outside base directory", buf);
814: return;
1.106 schwarze 815: }
816:
1.139 schwarze 817: /*
818: * Now we are sure the file is inside our tree.
819: * If it is a symbolic link, ignore the real path
820: * and use the original name.
821: * This implies passing stuff like "cat1/../man1/foo.1"
822: * on the command line won't work. So don't do that.
823: * Note the stat(2) can still fail if the link target
824: * doesn't exist.
825: */
826: if (S_IFLNK & st.st_mode) {
827: if (-1 == stat(buf, &st)) {
828: exitcode = (int)MANDOCLEVEL_BADARG;
829: say(file, "&stat");
830: return;
831: }
1.143 schwarze 832: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
833: say(file, "Filename too long");
834: return;
835: }
1.150 schwarze 836: start = buf;
837: if (OP_TEST != op && strstr(buf, basedir) == buf)
838: start += strlen(basedir);
1.50 kristaps 839: }
1.106 schwarze 840:
1.84 schwarze 841: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 842: mlink->dform = FORM_NONE;
1.143 schwarze 843: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
844: sizeof(mlink->file)) {
845: say(start, "Filename too long");
1.180 schwarze 846: free(mlink);
1.143 schwarze 847: return;
848: }
1.17 schwarze 849:
1.50 kristaps 850: /*
851: * First try to guess our directory structure.
852: * If we find a separator, try to look for man* or cat*.
853: * If we find one of these and what's underneath is a directory,
854: * assume it's an architecture.
855: */
856: if (NULL != (p = strchr(start, '/'))) {
857: *p++ = '\0';
858: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 859: mlink->dform = FORM_SRC;
1.94 schwarze 860: mlink->dsec = start + 3;
1.50 kristaps 861: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 862: mlink->dform = FORM_CAT;
1.94 schwarze 863: mlink->dsec = start + 3;
1.17 schwarze 864: }
1.5 kristaps 865:
1.50 kristaps 866: start = p;
1.84 schwarze 867: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 868: *p++ = '\0';
1.94 schwarze 869: mlink->arch = start;
1.50 kristaps 870: start = p;
1.84 schwarze 871: }
1.50 kristaps 872: }
873:
874: /*
875: * Now check the file suffix.
876: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
877: */
878: p = strrchr(start, '\0');
879: while (p-- > start && '/' != *p && '.' != *p)
880: /* Loop. */ ;
881:
882: if ('.' == *p) {
883: *p++ = '\0';
1.94 schwarze 884: mlink->fsec = p;
1.5 kristaps 885: }
886:
1.10 kristaps 887: /*
1.50 kristaps 888: * Now try to parse the name.
889: * Use the filename portion of the path.
1.10 kristaps 890: */
1.84 schwarze 891: mlink->name = start;
1.50 kristaps 892: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 893: mlink->name = p + 1;
1.50 kristaps 894: *p = '\0';
1.84 schwarze 895: }
896: mlink_add(mlink, &st);
1.50 kristaps 897: }
1.5 kristaps 898:
1.50 kristaps 899: static void
1.84 schwarze 900: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 901: {
1.83 schwarze 902: struct inodev inodev;
1.78 schwarze 903: struct mpage *mpage;
1.83 schwarze 904: unsigned int slot;
1.50 kristaps 905:
1.84 schwarze 906: assert(NULL != mlink->file);
1.50 kristaps 907:
1.94 schwarze 908: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
909: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
910: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
911: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 912:
913: if ('0' == *mlink->fsec) {
914: free(mlink->fsec);
915: mlink->fsec = mandoc_strdup(mlink->dsec);
916: mlink->fform = FORM_CAT;
917: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
918: mlink->fform = FORM_SRC;
1.74 schwarze 919: else
1.84 schwarze 920: mlink->fform = FORM_NONE;
1.82 schwarze 921:
1.83 schwarze 922: slot = ohash_qlookup(&mlinks, mlink->file);
923: assert(NULL == ohash_find(&mlinks, slot));
924: ohash_insert(&mlinks, slot, mlink);
925:
1.186 schwarze 926: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 927: inodev.st_ino = st->st_ino;
928: inodev.st_dev = st->st_dev;
929: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
930: sizeof(struct inodev), inodev.st_ino);
931: mpage = ohash_find(&mpages, slot);
932: if (NULL == mpage) {
933: mpage = mandoc_calloc(1, sizeof(struct mpage));
934: mpage->inodev.st_ino = inodev.st_ino;
935: mpage->inodev.st_dev = inodev.st_dev;
1.221 schwarze 936: mpage->form = FORM_NONE;
1.83 schwarze 937: ohash_insert(&mpages, slot, mpage);
938: } else
1.85 schwarze 939: mlink->next = mpage->mlinks;
1.82 schwarze 940: mpage->mlinks = mlink;
1.118 schwarze 941: mlink->mpage = mpage;
1.82 schwarze 942: }
943:
944: static void
945: mlink_free(struct mlink *mlink)
946: {
947:
948: free(mlink->dsec);
949: free(mlink->arch);
950: free(mlink->name);
951: free(mlink->fsec);
952: free(mlink);
1.50 kristaps 953: }
1.3 kristaps 954:
1.50 kristaps 955: static void
1.78 schwarze 956: mpages_free(void)
1.50 kristaps 957: {
1.78 schwarze 958: struct mpage *mpage;
1.82 schwarze 959: struct mlink *mlink;
1.80 schwarze 960: unsigned int slot;
1.3 kristaps 961:
1.80 schwarze 962: mpage = ohash_first(&mpages, &slot);
963: while (NULL != mpage) {
1.82 schwarze 964: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 965: mpage->mlinks = mlink->next;
1.82 schwarze 966: mlink_free(mlink);
967: }
1.78 schwarze 968: free(mpage->sec);
969: free(mpage->arch);
1.82 schwarze 970: free(mpage->title);
971: free(mpage->desc);
1.78 schwarze 972: free(mpage);
1.80 schwarze 973: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 974: }
975: }
1.38 schwarze 976:
1.50 kristaps 977: /*
1.89 schwarze 978: * For each mlink to the mpage, check whether the path looks like
979: * it is formatted, and if it does, check whether a source manual
980: * exists by the same name, ignoring the suffix.
981: * If both conditions hold, drop the mlink.
982: */
983: static void
984: mlinks_undupe(struct mpage *mpage)
985: {
986: char buf[PATH_MAX];
987: struct mlink **prev;
988: struct mlink *mlink;
989: char *bufp;
990:
991: mpage->form = FORM_CAT;
1.90 schwarze 992: prev = &mpage->mlinks;
993: while (NULL != (mlink = *prev)) {
1.89 schwarze 994: if (FORM_CAT != mlink->dform) {
995: mpage->form = FORM_NONE;
1.90 schwarze 996: goto nextlink;
1.89 schwarze 997: }
1.143 schwarze 998: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 999: bufp = strstr(buf, "cat");
1000: assert(NULL != bufp);
1001: memcpy(bufp, "man", 3);
1002: if (NULL != (bufp = strrchr(buf, '.')))
1003: *++bufp = '\0';
1.143 schwarze 1004: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1005: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1006: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1007: goto nextlink;
1.89 schwarze 1008: if (warnings)
1009: say(mlink->file, "Man source exists: %s", buf);
1010: if (use_all)
1.90 schwarze 1011: goto nextlink;
1.89 schwarze 1012: *prev = mlink->next;
1013: mlink_free(mlink);
1.90 schwarze 1014: continue;
1015: nextlink:
1016: prev = &(*prev)->next;
1.89 schwarze 1017: }
1018: }
1019:
1.131 schwarze 1020: static void
1.93 schwarze 1021: mlink_check(struct mpage *mpage, struct mlink *mlink)
1022: {
1.131 schwarze 1023: struct str *str;
1024: unsigned int slot;
1.93 schwarze 1025:
1026: /*
1027: * Check whether the manual section given in a file
1028: * agrees with the directory where the file is located.
1029: * Some manuals have suffixes like (3p) on their
1030: * section number either inside the file or in the
1031: * directory name, some are linked into more than one
1032: * section, like encrypt(1) = makekey(8).
1033: */
1034:
1035: if (FORM_SRC == mpage->form &&
1.131 schwarze 1036: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1037: say(mlink->file, "Section \"%s\" manual in %s directory",
1038: mpage->sec, mlink->dsec);
1039:
1040: /*
1041: * Manual page directories exist for each kernel
1042: * architecture as returned by machine(1).
1043: * However, many manuals only depend on the
1044: * application architecture as returned by arch(1).
1045: * For example, some (2/ARM) manuals are shared
1046: * across the "armish" and "zaurus" kernel
1047: * architectures.
1048: * A few manuals are even shared across completely
1049: * different architectures, for example fdformat(1)
1050: * on amd64, i386, sparc, and sparc64.
1051: */
1052:
1.131 schwarze 1053: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1054: say(mlink->file, "Architecture \"%s\" manual in "
1055: "\"%s\" directory", mpage->arch, mlink->arch);
1056:
1.131 schwarze 1057: /*
1058: * XXX
1.133 schwarze 1059: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1060: */
1061:
1062: if (FORM_CAT == mpage->form)
1063: return;
1064:
1065: /*
1066: * Check whether this mlink
1067: * appears as a name in the NAME section.
1068: */
1.93 schwarze 1069:
1.133 schwarze 1070: slot = ohash_qlookup(&names, mlink->name);
1071: str = ohash_find(&names, slot);
1.131 schwarze 1072: assert(NULL != str);
1.133 schwarze 1073: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1074: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1075: }
1076:
1.89 schwarze 1077: /*
1.80 schwarze 1078: * Run through the files in the global vector "mpages"
1079: * and add them to the database specified in "basedir".
1.50 kristaps 1080: *
1081: * This handles the parsing scheme itself, using the cues of directory
1082: * and filename to determine whether the file is parsable or not.
1083: */
1.59 schwarze 1084: static void
1.221 schwarze 1085: mpages_merge(struct dba *dba, struct mparse *mp)
1.50 kristaps 1086: {
1.118 schwarze 1087: struct mpage *mpage, *mpage_dest;
1088: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1089: struct roff_man *man;
1.118 schwarze 1090: char *sodest;
1.112 schwarze 1091: char *cp;
1.162 schwarze 1092: int fd;
1.101 schwarze 1093: unsigned int pslot;
1.69 schwarze 1094:
1.80 schwarze 1095: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1096: while (mpage != NULL) {
1.89 schwarze 1097: mlinks_undupe(mpage);
1.182 schwarze 1098: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1099: mpage = ohash_next(&mpages, &pslot);
1100: continue;
1.50 kristaps 1101: }
1102:
1.133 schwarze 1103: name_mask = NAME_MASK;
1.202 schwarze 1104: mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1105: mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1.50 kristaps 1106: mparse_reset(mp);
1107: man = NULL;
1.124 schwarze 1108: sodest = NULL;
1109:
1.213 schwarze 1110: if ((fd = mparse_open(mp, mlink->file)) == -1) {
1.182 schwarze 1111: say(mlink->file, "&open");
1.162 schwarze 1112: goto nextpage;
1.124 schwarze 1113: }
1.14 schwarze 1114:
1115: /*
1.183 schwarze 1116: * Interpret the file as mdoc(7) or man(7) source
1117: * code, unless it is known to be formatted.
1.14 schwarze 1118: */
1.182 schwarze 1119: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1120: mparse_readfd(mp, fd, mlink->file);
1.212 schwarze 1121: close(fd);
1.192 schwarze 1122: mparse_result(mp, &man, &sodest);
1.76 schwarze 1123: }
1.14 schwarze 1124:
1.171 schwarze 1125: if (sodest != NULL) {
1.118 schwarze 1126: mlink_dest = ohash_find(&mlinks,
1127: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1128: if (mlink_dest == NULL) {
1129: mandoc_asprintf(&cp, "%s.gz", sodest);
1130: mlink_dest = ohash_find(&mlinks,
1131: ohash_qlookup(&mlinks, cp));
1132: free(cp);
1133: }
1134: if (mlink_dest != NULL) {
1.118 schwarze 1135:
1136: /* The .so target exists. */
1137:
1138: mpage_dest = mlink_dest->mpage;
1139: while (1) {
1140: mlink->mpage = mpage_dest;
1141:
1142: /*
1143: * If the target was already
1144: * processed, add the links
1145: * to the database now.
1146: * Otherwise, this will
1147: * happen when we come
1148: * to the target.
1149: */
1150:
1.221 schwarze 1151: if (mpage_dest->dba != NULL)
1152: dbadd_mlink(mlink);
1.118 schwarze 1153:
1.171 schwarze 1154: if (mlink->next == NULL)
1.118 schwarze 1155: break;
1156: mlink = mlink->next;
1157: }
1158:
1159: /* Move all links to the target. */
1160:
1161: mlink->next = mlink_dest->next;
1162: mlink_dest->next = mpage->mlinks;
1163: mpage->mlinks = NULL;
1164: }
1.124 schwarze 1165: goto nextpage;
1.192 schwarze 1166: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.204 schwarze 1167: mdoc_validate(man);
1.82 schwarze 1168: mpage->form = FORM_SRC;
1.193 schwarze 1169: mpage->sec = man->meta.msec;
1.155 schwarze 1170: mpage->sec = mandoc_strdup(
1.171 schwarze 1171: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1172: mpage->arch = man->meta.arch;
1.84 schwarze 1173: mpage->arch = mandoc_strdup(
1.171 schwarze 1174: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1175: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1176: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.205 schwarze 1177: man_validate(man);
1.82 schwarze 1178: mpage->form = FORM_SRC;
1.193 schwarze 1179: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1180: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1181: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1182: } else {
1.82 schwarze 1183: mpage->form = FORM_CAT;
1.182 schwarze 1184: mpage->sec = mandoc_strdup(mlink->dsec);
1185: mpage->arch = mandoc_strdup(mlink->arch);
1186: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1187: }
1.12 schwarze 1188:
1.174 schwarze 1189: assert(mpage->desc == NULL);
1.192 schwarze 1190: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1191: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1192: else if (man != NULL)
1.193 schwarze 1193: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1194: else
1.162 schwarze 1195: parse_cat(mpage, fd);
1.174 schwarze 1196: if (mpage->desc == NULL)
1.135 schwarze 1197: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1198:
1199: if (warnings && !use_all)
1200: for (mlink = mpage->mlinks; mlink;
1201: mlink = mlink->next)
1202: mlink_check(mpage, mlink);
1.44 kristaps 1203:
1.221 schwarze 1204: dbadd(dba, mpage);
1.182 schwarze 1205: mlink = mpage->mlinks;
1.124 schwarze 1206:
1207: nextpage:
1.66 schwarze 1208: ohash_delete(&strings);
1.133 schwarze 1209: ohash_delete(&names);
1.80 schwarze 1210: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1211: }
1212: }
1.12 schwarze 1213:
1.50 kristaps 1214: static void
1.124 schwarze 1215: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1216: {
1217: FILE *stream;
1218: char *line, *p, *title;
1.209 schwarze 1219: size_t linesz, plen, titlesz;
1220: ssize_t len;
1221: int offs;
1.12 schwarze 1222:
1.124 schwarze 1223: stream = (-1 == fd) ?
1224: fopen(mpage->mlinks->file, "r") :
1225: fdopen(fd, "r");
1226: if (NULL == stream) {
1.159 schwarze 1227: if (-1 != fd)
1228: close(fd);
1.56 schwarze 1229: if (warnings)
1.123 schwarze 1230: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1231: return;
1232: }
1.1 kristaps 1233:
1.209 schwarze 1234: line = NULL;
1235: linesz = 0;
1236:
1.50 kristaps 1237: /* Skip to first blank line. */
1.1 kristaps 1238:
1.209 schwarze 1239: while (getline(&line, &linesz, stream) != -1)
1240: if (*line == '\n')
1.50 kristaps 1241: break;
1.1 kristaps 1242:
1.50 kristaps 1243: /*
1244: * Assume the first line that is not indented
1245: * is the first section header. Skip to it.
1246: */
1.1 kristaps 1247:
1.209 schwarze 1248: while (getline(&line, &linesz, stream) != -1)
1249: if (*line != '\n' && *line != ' ')
1.50 kristaps 1250: break;
1.141 schwarze 1251:
1.50 kristaps 1252: /*
1253: * Read up until the next section into a buffer.
1254: * Strip the leading and trailing newline from each read line,
1255: * appending a trailing space.
1256: * Ignore empty (whitespace-only) lines.
1257: */
1.1 kristaps 1258:
1.50 kristaps 1259: titlesz = 0;
1260: title = NULL;
1.38 schwarze 1261:
1.209 schwarze 1262: while ((len = getline(&line, &linesz, stream)) != -1) {
1263: if (*line != ' ')
1.50 kristaps 1264: break;
1.209 schwarze 1265: offs = 0;
1266: while (isspace((unsigned char)line[offs]))
1267: offs++;
1268: if (line[offs] == '\0')
1.38 schwarze 1269: continue;
1.209 schwarze 1270: title = mandoc_realloc(title, titlesz + len - offs);
1271: memcpy(title + titlesz, line + offs, len - offs);
1272: titlesz += len - offs;
1.50 kristaps 1273: title[titlesz - 1] = ' ';
1274: }
1.209 schwarze 1275: free(line);
1.38 schwarze 1276:
1.50 kristaps 1277: /*
1278: * If no page content can be found, or the input line
1279: * is already the next section header, or there is no
1280: * trailing newline, reuse the page title as the page
1281: * description.
1282: */
1.44 kristaps 1283:
1.50 kristaps 1284: if (NULL == title || '\0' == *title) {
1.56 schwarze 1285: if (warnings)
1.82 schwarze 1286: say(mpage->mlinks->file,
1287: "Cannot find NAME section");
1.50 kristaps 1288: fclose(stream);
1289: free(title);
1290: return;
1291: }
1.1 kristaps 1292:
1.209 schwarze 1293: title[titlesz - 1] = '\0';
1.33 schwarze 1294:
1.50 kristaps 1295: /*
1296: * Skip to the first dash.
1297: * Use the remaining line as the description (no more than 70
1298: * bytes).
1299: */
1.33 schwarze 1300:
1.50 kristaps 1301: if (NULL != (p = strstr(title, "- "))) {
1302: for (p += 2; ' ' == *p || '\b' == *p; p++)
1303: /* Skip to next word. */ ;
1304: } else {
1.56 schwarze 1305: if (warnings)
1.82 schwarze 1306: say(mpage->mlinks->file,
1307: "No dash in title line");
1.50 kristaps 1308: p = title;
1309: }
1.38 schwarze 1310:
1.50 kristaps 1311: plen = strlen(p);
1.1 kristaps 1312:
1.50 kristaps 1313: /* Strip backspace-encoding from line. */
1.1 kristaps 1314:
1.50 kristaps 1315: while (NULL != (line = memchr(p, '\b', plen))) {
1316: len = line - p;
1317: if (0 == len) {
1318: memmove(line, line + 1, plen--);
1319: continue;
1.141 schwarze 1320: }
1.50 kristaps 1321: memmove(line - 1, line + 1, plen - len);
1322: plen -= 2;
1323: }
1.1 kristaps 1324:
1.78 schwarze 1325: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1326: fclose(stream);
1327: free(title);
1328: }
1.1 kristaps 1329:
1.50 kristaps 1330: /*
1331: * Put a type/word pair into the word database for this particular file.
1332: */
1333: static void
1.112 schwarze 1334: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1335: {
1.78 schwarze 1336: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1337: }
1338:
1339: /*
1.50 kristaps 1340: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1341: */
1342: static void
1.78 schwarze 1343: putmdockey(const struct mpage *mpage,
1.189 schwarze 1344: const struct roff_node *n, uint64_t m)
1.3 kristaps 1345: {
1.18 kristaps 1346:
1.50 kristaps 1347: for ( ; NULL != n; n = n->next) {
1348: if (NULL != n->child)
1.78 schwarze 1349: putmdockey(mpage, n->child, m);
1.188 schwarze 1350: if (n->type == ROFFT_TEXT)
1.78 schwarze 1351: putkey(mpage, n->string, m);
1.50 kristaps 1352: }
1353: }
1.18 kristaps 1354:
1.61 schwarze 1355: static void
1.190 schwarze 1356: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1357: const struct roff_node *n)
1.50 kristaps 1358: {
1.189 schwarze 1359: const struct roff_node *head, *body;
1.121 schwarze 1360: char *start, *title;
1.50 kristaps 1361: char byte;
1.121 schwarze 1362: size_t sz;
1.18 kristaps 1363:
1.215 schwarze 1364: if (n == NULL)
1.61 schwarze 1365: return;
1.18 kristaps 1366:
1.50 kristaps 1367: /*
1368: * We're only searching for one thing: the first text child in
1369: * the BODY of a NAME section. Since we don't keep track of
1370: * sections in -man, run some hoops to find out whether we're in
1371: * the correct section or not.
1372: */
1.18 kristaps 1373:
1.188 schwarze 1374: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1375: body = n;
1.215 schwarze 1376: if ((head = body->parent->head) != NULL &&
1377: (head = head->child) != NULL &&
1378: head->next == NULL &&
1.188 schwarze 1379: head->type == ROFFT_TEXT &&
1.215 schwarze 1380: strcmp(head->string, "NAME") == 0 &&
1381: body->child != NULL) {
1.3 kristaps 1382:
1.50 kristaps 1383: /*
1384: * Suck the entire NAME section into memory.
1385: * Yes, we might run away.
1386: * But too many manuals have big, spread-out
1387: * NAME sections over many lines.
1388: */
1.3 kristaps 1389:
1.121 schwarze 1390: title = NULL;
1.194 schwarze 1391: deroff(&title, body);
1.50 kristaps 1392: if (NULL == title)
1.61 schwarze 1393: return;
1.18 kristaps 1394:
1.141 schwarze 1395: /*
1.50 kristaps 1396: * Go through a special heuristic dance here.
1397: * Conventionally, one or more manual names are
1398: * comma-specified prior to a whitespace, then a
1399: * dash, then a description. Try to puzzle out
1400: * the name parts here.
1401: */
1.18 kristaps 1402:
1.121 schwarze 1403: start = title;
1.50 kristaps 1404: for ( ;; ) {
1405: sz = strcspn(start, " ,");
1406: if ('\0' == start[sz])
1407: break;
1.1 kristaps 1408:
1.50 kristaps 1409: byte = start[sz];
1410: start[sz] = '\0';
1.110 schwarze 1411:
1412: /*
1413: * Assume a stray trailing comma in the
1414: * name list if a name begins with a dash.
1415: */
1416:
1417: if ('-' == start[0] ||
1418: ('\\' == start[0] && '-' == start[1]))
1419: break;
1.1 kristaps 1420:
1.133 schwarze 1421: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1422: if ( ! (mpage->name_head_done ||
1423: strcasecmp(start, meta->title))) {
1424: putkey(mpage, start, NAME_HEAD);
1425: mpage->name_head_done = 1;
1426: }
1.1 kristaps 1427:
1.50 kristaps 1428: if (' ' == byte) {
1429: start += sz + 1;
1430: break;
1431: }
1.1 kristaps 1432:
1.50 kristaps 1433: assert(',' == byte);
1434: start += sz + 1;
1435: while (' ' == *start)
1436: start++;
1437: }
1.1 kristaps 1438:
1.121 schwarze 1439: if (start == title) {
1.133 schwarze 1440: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1441: if ( ! (mpage->name_head_done ||
1442: strcasecmp(start, meta->title))) {
1443: putkey(mpage, start, NAME_HEAD);
1444: mpage->name_head_done = 1;
1445: }
1.50 kristaps 1446: free(title);
1.61 schwarze 1447: return;
1.50 kristaps 1448: }
1.1 kristaps 1449:
1.50 kristaps 1450: while (isspace((unsigned char)*start))
1451: start++;
1.1 kristaps 1452:
1.50 kristaps 1453: if (0 == strncmp(start, "-", 1))
1454: start += 1;
1455: else if (0 == strncmp(start, "\\-\\-", 4))
1456: start += 4;
1457: else if (0 == strncmp(start, "\\-", 2))
1458: start += 2;
1459: else if (0 == strncmp(start, "\\(en", 4))
1460: start += 4;
1461: else if (0 == strncmp(start, "\\(em", 4))
1462: start += 4;
1.1 kristaps 1463:
1.50 kristaps 1464: while (' ' == *start)
1465: start++;
1.1 kristaps 1466:
1.78 schwarze 1467: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1468: free(title);
1.61 schwarze 1469: return;
1.50 kristaps 1470: }
1471: }
1.1 kristaps 1472:
1.77 schwarze 1473: for (n = n->child; n; n = n->next) {
1.78 schwarze 1474: if (NULL != mpage->desc)
1.77 schwarze 1475: break;
1.174 schwarze 1476: parse_man(mpage, meta, n);
1.77 schwarze 1477: }
1.1 kristaps 1478: }
1479:
1480: static void
1.190 schwarze 1481: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1482: const struct roff_node *n)
1.1 kristaps 1483: {
1484:
1.50 kristaps 1485: assert(NULL != n);
1486: for (n = n->child; NULL != n; n = n->next) {
1487: switch (n->type) {
1.188 schwarze 1488: case ROFFT_ELEM:
1489: case ROFFT_BLOCK:
1490: case ROFFT_HEAD:
1491: case ROFFT_BODY:
1492: case ROFFT_TAIL:
1.50 kristaps 1493: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1494: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1495: break;
1.68 schwarze 1496: if (mdocs[n->tok].mask)
1.78 schwarze 1497: putmdockey(mpage, n->child,
1498: mdocs[n->tok].mask);
1.50 kristaps 1499: break;
1500: default:
1.188 schwarze 1501: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1502: continue;
1503: }
1504: if (NULL != n->child)
1.172 schwarze 1505: parse_mdoc(mpage, meta, n);
1.1 kristaps 1506: }
1507: }
1508:
1.25 schwarze 1509: static int
1.190 schwarze 1510: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1511: const struct roff_node *n)
1.1 kristaps 1512: {
1.176 schwarze 1513: char *start, *end;
1.1 kristaps 1514: size_t sz;
1.25 schwarze 1515:
1.50 kristaps 1516: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1517: NULL == (n = n->child) ||
1.188 schwarze 1518: n->type != ROFFT_TEXT)
1.197 schwarze 1519: return 0;
1.1 kristaps 1520:
1521: /*
1522: * Only consider those `Fd' macro fields that begin with an
1523: * "inclusion" token (versus, e.g., #define).
1524: */
1.50 kristaps 1525:
1.1 kristaps 1526: if (strcmp("#include", n->string))
1.197 schwarze 1527: return 0;
1.1 kristaps 1528:
1.188 schwarze 1529: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1530: return 0;
1.1 kristaps 1531:
1532: /*
1533: * Strip away the enclosing angle brackets and make sure we're
1534: * not zero-length.
1535: */
1536:
1537: start = n->string;
1538: if ('<' == *start || '"' == *start)
1539: start++;
1540:
1541: if (0 == (sz = strlen(start)))
1.197 schwarze 1542: return 0;
1.1 kristaps 1543:
1544: end = &start[(int)sz - 1];
1545: if ('>' == *end || '"' == *end)
1546: end--;
1547:
1.50 kristaps 1548: if (end > start)
1.78 schwarze 1549: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1550: return 0;
1.1 kristaps 1551: }
1552:
1.178 schwarze 1553: static void
1.189 schwarze 1554: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1555: {
1.112 schwarze 1556: char *cp;
1.178 schwarze 1557: size_t sz;
1.1 kristaps 1558:
1.188 schwarze 1559: if (n->type != ROFFT_TEXT)
1.178 schwarze 1560: return;
1.25 schwarze 1561:
1.178 schwarze 1562: /* Skip function pointer punctuation. */
1.1 kristaps 1563:
1.178 schwarze 1564: cp = n->string;
1565: while (*cp == '(' || *cp == '*')
1.1 kristaps 1566: cp++;
1.178 schwarze 1567: sz = strcspn(cp, "()");
1.1 kristaps 1568:
1.178 schwarze 1569: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1570: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1571: putkeys(mpage, cp, sz, NAME_SYN);
1572: }
1.25 schwarze 1573:
1.178 schwarze 1574: static int
1.190 schwarze 1575: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1576: const struct roff_node *n)
1.178 schwarze 1577: {
1578:
1579: if (n->child == NULL)
1.197 schwarze 1580: return 0;
1.178 schwarze 1581:
1582: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1583:
1.178 schwarze 1584: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1585: if (n->type == ROFFT_TEXT)
1.78 schwarze 1586: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1587:
1.197 schwarze 1588: return 0;
1.173 schwarze 1589: }
1590:
1591: static int
1.190 schwarze 1592: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1593: const struct roff_node *n)
1.173 schwarze 1594: {
1.177 schwarze 1595:
1.188 schwarze 1596: if (n->type != ROFFT_HEAD)
1.197 schwarze 1597: return 1;
1.173 schwarze 1598:
1.178 schwarze 1599: if (n->child != NULL)
1600: parse_mdoc_fname(mpage, n->child);
1601:
1.197 schwarze 1602: return 0;
1.1 kristaps 1603: }
1604:
1.25 schwarze 1605: static int
1.211 schwarze 1606: parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1607: const struct roff_node *n)
1608: {
1609: char *cp;
1610:
1611: if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1612: return 0;
1613:
1.215 schwarze 1614: if (n->child != NULL &&
1615: n->child->next == NULL &&
1616: n->child->type == ROFFT_TEXT)
1.211 schwarze 1617: return 1;
1618:
1619: cp = NULL;
1620: deroff(&cp, n);
1621: if (cp != NULL) {
1622: putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1623: n->type == ROFFT_BODY ? TYPE_Va : 0));
1624: free(cp);
1625: }
1626:
1627: return 0;
1628: }
1629:
1630: static int
1.190 schwarze 1631: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1632: const struct roff_node *n)
1.1 kristaps 1633: {
1.67 schwarze 1634: char *cp;
1.1 kristaps 1635:
1636: if (NULL == (n = n->child))
1.197 schwarze 1637: return 0;
1.1 kristaps 1638:
1.67 schwarze 1639: if (NULL == n->next) {
1.78 schwarze 1640: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1641: return 0;
1.67 schwarze 1642: }
1643:
1.120 schwarze 1644: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1645: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1646: free(cp);
1.197 schwarze 1647: return 0;
1.1 kristaps 1648: }
1649:
1.25 schwarze 1650: static int
1.190 schwarze 1651: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1652: const struct roff_node *n)
1.1 kristaps 1653: {
1654:
1.188 schwarze 1655: if (n->type == ROFFT_BODY)
1.194 schwarze 1656: deroff(&mpage->desc, n);
1.197 schwarze 1657: return 0;
1.1 kristaps 1658: }
1659:
1.25 schwarze 1660: static int
1.190 schwarze 1661: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1662: const struct roff_node *n)
1.1 kristaps 1663: {
1664:
1.129 schwarze 1665: if (SEC_NAME == n->sec)
1.133 schwarze 1666: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1667: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1668: if (n->child == NULL)
1669: putkey(mpage, meta->name, NAME_SYN);
1670: else
1671: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1672: }
1673: if ( ! (mpage->name_head_done ||
1674: n->child == NULL || n->child->string == NULL ||
1675: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1676: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1677: mpage->name_head_done = 1;
1.172 schwarze 1678: }
1.197 schwarze 1679: return 0;
1.1 kristaps 1680: }
1681:
1.25 schwarze 1682: static int
1.190 schwarze 1683: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1684: const struct roff_node *n)
1.1 kristaps 1685: {
1686:
1.197 schwarze 1687: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1688: }
1689:
1.50 kristaps 1690: static int
1.190 schwarze 1691: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1692: const struct roff_node *n)
1.1 kristaps 1693: {
1694:
1.197 schwarze 1695: return n->type == ROFFT_HEAD;
1.1 kristaps 1696: }
1697:
1.50 kristaps 1698: /*
1.66 schwarze 1699: * Add a string to the hash table for the current manual.
1700: * Each string has a bitmask telling which macros it belongs to.
1701: * When we finish the manual, we'll dump the table.
1.50 kristaps 1702: */
1703: static void
1.176 schwarze 1704: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1705: {
1.133 schwarze 1706: struct ohash *htab;
1.50 kristaps 1707: struct str *s;
1.111 schwarze 1708: const char *end;
1.65 schwarze 1709: unsigned int slot;
1.176 schwarze 1710: int i, mustfree;
1.25 schwarze 1711:
1.50 kristaps 1712: if (0 == sz)
1713: return;
1.111 schwarze 1714:
1.176 schwarze 1715: mustfree = render_string(&cp, &sz);
1716:
1.133 schwarze 1717: if (TYPE_Nm & v) {
1718: htab = &names;
1719: v &= name_mask;
1.169 schwarze 1720: if (v & NAME_FIRST)
1721: name_mask &= ~NAME_FIRST;
1.133 schwarze 1722: if (debug > 1)
1723: say(mpage->mlinks->file,
1.220 schwarze 1724: "Adding name %*s, bits=0x%llu", (int)sz, cp, v);
1.133 schwarze 1725: } else {
1726: htab = &strings;
1727: if (debug > 1)
1.221 schwarze 1728: for (i = 0; i < KEY_MAX; i++)
1.163 schwarze 1729: if ((uint64_t)1 << i & v)
1.133 schwarze 1730: say(mpage->mlinks->file,
1731: "Adding key %s=%*s",
1.220 schwarze 1732: mansearch_keynames[i], (int)sz, cp);
1.111 schwarze 1733: }
1.25 schwarze 1734:
1.65 schwarze 1735: end = cp + sz;
1.133 schwarze 1736: slot = ohash_qlookupi(htab, cp, &end);
1737: s = ohash_find(htab, slot);
1.25 schwarze 1738:
1.78 schwarze 1739: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1740: s->mask |= v;
1741: return;
1742: } else if (NULL == s) {
1.144 schwarze 1743: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1744: memcpy(s->key, cp, sz);
1.133 schwarze 1745: ohash_insert(htab, slot, s);
1.1 kristaps 1746: }
1.78 schwarze 1747: s->mpage = mpage;
1.50 kristaps 1748: s->mask = v;
1.176 schwarze 1749:
1750: if (mustfree)
1751: free(cp);
1.1 kristaps 1752: }
1753:
1.50 kristaps 1754: /*
1755: * Take a Unicode codepoint and produce its UTF-8 encoding.
1756: * This isn't the best way to do this, but it works.
1757: * The magic numbers are from the UTF-8 packaging.
1758: * They're not as scary as they seem: read the UTF-8 spec for details.
1759: */
1760: static size_t
1761: utf8(unsigned int cp, char out[7])
1.1 kristaps 1762: {
1.50 kristaps 1763: size_t rc;
1.1 kristaps 1764:
1.50 kristaps 1765: rc = 0;
1766: if (cp <= 0x0000007F) {
1767: rc = 1;
1768: out[0] = (char)cp;
1769: } else if (cp <= 0x000007FF) {
1770: rc = 2;
1771: out[0] = (cp >> 6 & 31) | 192;
1772: out[1] = (cp & 63) | 128;
1773: } else if (cp <= 0x0000FFFF) {
1774: rc = 3;
1775: out[0] = (cp >> 12 & 15) | 224;
1776: out[1] = (cp >> 6 & 63) | 128;
1777: out[2] = (cp & 63) | 128;
1778: } else if (cp <= 0x001FFFFF) {
1779: rc = 4;
1780: out[0] = (cp >> 18 & 7) | 240;
1781: out[1] = (cp >> 12 & 63) | 128;
1782: out[2] = (cp >> 6 & 63) | 128;
1783: out[3] = (cp & 63) | 128;
1784: } else if (cp <= 0x03FFFFFF) {
1785: rc = 5;
1786: out[0] = (cp >> 24 & 3) | 248;
1787: out[1] = (cp >> 18 & 63) | 128;
1788: out[2] = (cp >> 12 & 63) | 128;
1789: out[3] = (cp >> 6 & 63) | 128;
1790: out[4] = (cp & 63) | 128;
1791: } else if (cp <= 0x7FFFFFFF) {
1792: rc = 6;
1793: out[0] = (cp >> 30 & 1) | 252;
1794: out[1] = (cp >> 24 & 63) | 128;
1795: out[2] = (cp >> 18 & 63) | 128;
1796: out[3] = (cp >> 12 & 63) | 128;
1797: out[4] = (cp >> 6 & 63) | 128;
1798: out[5] = (cp & 63) | 128;
1799: } else
1.197 schwarze 1800: return 0;
1.1 kristaps 1801:
1.50 kristaps 1802: out[rc] = '\0';
1.197 schwarze 1803: return rc;
1.50 kristaps 1804: }
1.1 kristaps 1805:
1.50 kristaps 1806: /*
1.176 schwarze 1807: * If the string contains escape sequences,
1808: * replace it with an allocated rendering and return 1,
1809: * such that the caller can free it after use.
1810: * Otherwise, do nothing and return 0.
1.50 kristaps 1811: */
1.176 schwarze 1812: static int
1813: render_string(char **public, size_t *psz)
1.50 kristaps 1814: {
1.176 schwarze 1815: const char *src, *scp, *addcp, *seq;
1816: char *dst;
1817: size_t ssz, dsz, addsz;
1.114 schwarze 1818: char utfbuf[7], res[6];
1.176 schwarze 1819: int seqlen, unicode;
1.50 kristaps 1820:
1821: res[0] = '\\';
1822: res[1] = '\t';
1823: res[2] = ASCII_NBRSP;
1824: res[3] = ASCII_HYPH;
1.114 schwarze 1825: res[4] = ASCII_BREAK;
1826: res[5] = '\0';
1.1 kristaps 1827:
1.176 schwarze 1828: src = scp = *public;
1829: ssz = *psz;
1830: dst = NULL;
1831: dsz = 0;
1832:
1833: while (scp < src + *psz) {
1834:
1835: /* Leave normal characters unchanged. */
1836:
1837: if (strchr(res, *scp) == NULL) {
1838: if (dst != NULL)
1839: dst[dsz++] = *scp;
1840: scp++;
1841: continue;
1842: }
1.46 kristaps 1843:
1.50 kristaps 1844: /*
1.176 schwarze 1845: * Found something that requires replacing,
1846: * make sure we have a destination buffer.
1.50 kristaps 1847: */
1.176 schwarze 1848:
1849: if (dst == NULL) {
1850: dst = mandoc_malloc(ssz + 1);
1851: dsz = scp - src;
1852: memcpy(dst, src, dsz);
1.50 kristaps 1853: }
1.46 kristaps 1854:
1.176 schwarze 1855: /* Handle single-char special characters. */
1856:
1857: switch (*scp) {
1858: case '\\':
1859: break;
1.141 schwarze 1860: case '\t':
1861: case ASCII_NBRSP:
1.176 schwarze 1862: dst[dsz++] = ' ';
1863: scp++;
1864: continue;
1865: case ASCII_HYPH:
1866: dst[dsz++] = '-';
1.114 schwarze 1867: /* FALLTHROUGH */
1.141 schwarze 1868: case ASCII_BREAK:
1.176 schwarze 1869: scp++;
1.50 kristaps 1870: continue;
1.114 schwarze 1871: default:
1.176 schwarze 1872: abort();
1.114 schwarze 1873: }
1.46 kristaps 1874:
1.50 kristaps 1875: /*
1.176 schwarze 1876: * Found an escape sequence.
1877: * Read past the slash, then parse it.
1878: * Ignore everything except characters.
1.50 kristaps 1879: */
1.95 schwarze 1880:
1.176 schwarze 1881: scp++;
1882: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1883: continue;
1.1 kristaps 1884:
1.50 kristaps 1885: /*
1.95 schwarze 1886: * Render the special character
1887: * as either UTF-8 or ASCII.
1.50 kristaps 1888: */
1.95 schwarze 1889:
1890: if (write_utf8) {
1.203 schwarze 1891: unicode = mchars_spec2cp(seq, seqlen);
1.176 schwarze 1892: if (unicode <= 0)
1.95 schwarze 1893: continue;
1.176 schwarze 1894: addsz = utf8(unicode, utfbuf);
1895: if (addsz == 0)
1.95 schwarze 1896: continue;
1.176 schwarze 1897: addcp = utfbuf;
1.95 schwarze 1898: } else {
1.203 schwarze 1899: addcp = mchars_spec2str(seq, seqlen, &addsz);
1.176 schwarze 1900: if (addcp == NULL)
1.95 schwarze 1901: continue;
1.176 schwarze 1902: if (*addcp == ASCII_NBRSP) {
1903: addcp = " ";
1904: addsz = 1;
1.95 schwarze 1905: }
1906: }
1.1 kristaps 1907:
1.50 kristaps 1908: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1909:
1.176 schwarze 1910: ssz += addsz;
1911: dst = mandoc_realloc(dst, ssz + 1);
1912: memcpy(dst + dsz, addcp, addsz);
1913: dsz += addsz;
1914: }
1915: if (dst != NULL) {
1916: *public = dst;
1917: *psz = dsz;
1918: }
1919:
1920: /* Trim trailing whitespace and NUL-terminate. */
1921:
1922: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1923: --*psz;
1924: if (dst != NULL) {
1925: (*public)[*psz] = '\0';
1.197 schwarze 1926: return 1;
1.176 schwarze 1927: } else
1.197 schwarze 1928: return 0;
1.1 kristaps 1929: }
1930:
1.118 schwarze 1931: static void
1932: dbadd_mlink(const struct mlink *mlink)
1933: {
1.221 schwarze 1934: dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
1935: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
1.223 ! schwarze 1936: dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
1.221 schwarze 1937: dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
1938: dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
1.118 schwarze 1939: }
1940:
1.14 schwarze 1941: /*
1.50 kristaps 1942: * Flush the current page's terms (and their bits) into the database.
1.96 schwarze 1943: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 1944: */
1945: static void
1.221 schwarze 1946: dbadd(struct dba *dba, struct mpage *mpage)
1.14 schwarze 1947: {
1.87 schwarze 1948: struct mlink *mlink;
1.50 kristaps 1949: struct str *key;
1.176 schwarze 1950: char *cp;
1.221 schwarze 1951: uint64_t mask;
1.52 kristaps 1952: size_t i;
1.66 schwarze 1953: unsigned int slot;
1.176 schwarze 1954: int mustfree;
1.14 schwarze 1955:
1.128 schwarze 1956: mlink = mpage->mlinks;
1.43 kristaps 1957:
1.128 schwarze 1958: if (nodb) {
1.147 schwarze 1959: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 1960: key = ohash_next(&names, &slot))
1.147 schwarze 1961: free(key);
1962: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 1963: key = ohash_next(&strings, &slot))
1.147 schwarze 1964: free(key);
1.145 schwarze 1965: if (0 == debug)
1966: return;
1.128 schwarze 1967: while (NULL != mlink) {
1968: fputs(mlink->name, stdout);
1969: if (NULL == mlink->next ||
1970: strcmp(mlink->dsec, mlink->next->dsec) ||
1971: strcmp(mlink->fsec, mlink->next->fsec) ||
1972: strcmp(mlink->arch, mlink->next->arch)) {
1973: putchar('(');
1974: if ('\0' == *mlink->dsec)
1975: fputs(mlink->fsec, stdout);
1976: else
1977: fputs(mlink->dsec, stdout);
1978: if ('\0' != *mlink->arch)
1979: printf("/%s", mlink->arch);
1980: putchar(')');
1981: }
1982: mlink = mlink->next;
1983: if (NULL != mlink)
1984: fputs(", ", stdout);
1985: }
1.132 schwarze 1986: printf(" - %s\n", mpage->desc);
1.14 schwarze 1987: return;
1.128 schwarze 1988: }
1989:
1990: if (debug)
1991: say(mlink->file, "Adding to database");
1.28 kristaps 1992:
1.176 schwarze 1993: cp = mpage->desc;
1994: i = strlen(cp);
1995: mustfree = render_string(&cp, &i);
1.223 ! schwarze 1996: mpage->dba = dba_page_new(dba->pages, mlink->name, mpage->sec,
! 1997: *mpage->arch == '\0' ? mlink->arch : mpage->arch,
! 1998: cp, mlink->file, mpage->form);
1.176 schwarze 1999: if (mustfree)
2000: free(cp);
1.81 schwarze 2001:
1.223 ! schwarze 2002: while (mlink != NULL) {
1.118 schwarze 2003: dbadd_mlink(mlink);
1.223 ! schwarze 2004: mlink = mlink->next;
! 2005: }
1.50 kristaps 2006:
1.133 schwarze 2007: for (key = ohash_first(&names, &slot); NULL != key;
2008: key = ohash_next(&names, &slot)) {
2009: assert(key->mpage == mpage);
1.221 schwarze 2010: dba_page_alias(mpage->dba, key->key, key->mask);
1.133 schwarze 2011: free(key);
2012: }
1.66 schwarze 2013: for (key = ohash_first(&strings, &slot); NULL != key;
2014: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2015: assert(key->mpage == mpage);
1.221 schwarze 2016: i = 0;
2017: for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2018: if (key->mask & mask)
2019: dba_macro_add(dba->macros, i,
2020: key->key, mpage->dba);
2021: i++;
2022: }
1.66 schwarze 2023: free(key);
1.38 schwarze 2024: }
1.14 schwarze 2025: }
2026:
1.5 kristaps 2027: static void
1.221 schwarze 2028: dbprune(struct dba *dba)
1.5 kristaps 2029: {
1.221 schwarze 2030: struct dba_array *page, *files;
2031: char *file;
1.5 kristaps 2032:
1.221 schwarze 2033: dba_array_FOREACH(dba->pages, page) {
2034: files = dba_array_get(page, DBP_FILE);
2035: dba_array_FOREACH(files, file) {
2036: if (*file < ' ')
2037: file++;
2038: if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2039: file)) != NULL) {
2040: if (debug)
2041: say(file, "Deleting from database");
2042: dba_array_del(dba->pages);
2043: break;
2044: }
1.106 schwarze 2045: }
1.5 kristaps 2046: }
2047: }
2048:
1.4 kristaps 2049: /*
1.221 schwarze 2050: * Write the database from memory to disk.
1.4 kristaps 2051: */
1.35 kristaps 2052: static void
1.221 schwarze 2053: dbwrite(struct dba *dba)
1.4 kristaps 2054: {
1.221 schwarze 2055: char tfn[32];
1.115 schwarze 2056: int status;
2057: pid_t child;
1.4 kristaps 2058:
1.221 schwarze 2059: if (dba_write(MANDOC_DB "~", dba) != -1) {
2060: if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2061: exitcode = (int)MANDOCLEVEL_SYSERR;
2062: say(MANDOC_DB, "&rename");
2063: unlink(MANDOC_DB "~");
2064: }
1.38 schwarze 2065: return;
1.4 kristaps 2066: }
2067:
1.221 schwarze 2068: (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2069: if (mkdtemp(tfn) == NULL) {
2070: exitcode = (int)MANDOCLEVEL_SYSERR;
2071: say("", "&%s", tfn);
1.50 kristaps 2072: return;
1.221 schwarze 2073: }
1.12 schwarze 2074:
1.221 schwarze 2075: (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2076: if (dba_write(tfn, dba) == -1) {
2077: exitcode = (int)MANDOCLEVEL_SYSERR;
2078: say(tfn, "&dba_write");
2079: goto out;
1.115 schwarze 2080: }
2081:
2082: switch (child = fork()) {
1.141 schwarze 2083: case -1:
1.115 schwarze 2084: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2085: say("", "&fork cmp");
1.115 schwarze 2086: return;
1.141 schwarze 2087: case 0:
1.221 schwarze 2088: execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL);
1.123 schwarze 2089: say("", "&exec cmp");
1.115 schwarze 2090: exit(0);
2091: default:
2092: break;
2093: }
1.221 schwarze 2094: if (waitpid(child, &status, 0) == -1) {
1.115 schwarze 2095: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2096: say("", "&wait cmp");
1.115 schwarze 2097: } else if (WIFSIGNALED(status)) {
2098: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2099: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2100: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2101: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2102: say(MANDOC_DB,
2103: "Data changed, but cannot replace database");
2104: }
2105:
1.221 schwarze 2106: out:
2107: *strrchr(tfn, '/') = '\0';
1.115 schwarze 2108: switch (child = fork()) {
1.141 schwarze 2109: case -1:
1.115 schwarze 2110: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2111: say("", "&fork rm");
1.115 schwarze 2112: return;
1.141 schwarze 2113: case 0:
1.221 schwarze 2114: execlp("rm", "rm", "-rf", tfn, (char *)NULL);
1.123 schwarze 2115: say("", "&exec rm");
1.115 schwarze 2116: exit((int)MANDOCLEVEL_SYSERR);
2117: default:
2118: break;
2119: }
1.221 schwarze 2120: if (waitpid(child, &status, 0) == -1) {
1.115 schwarze 2121: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2122: say("", "&wait rm");
1.115 schwarze 2123: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2124: exitcode = (int)MANDOCLEVEL_SYSERR;
1.221 schwarze 2125: say("", "%s: Cannot remove temporary directory", tfn);
1.146 schwarze 2126: }
1.4 kristaps 2127: }
2128:
1.50 kristaps 2129: static int
1.165 schwarze 2130: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2131: {
1.59 schwarze 2132: static char startdir[PATH_MAX];
1.151 schwarze 2133: static int getcwd_status; /* 1 = ok, 2 = failure */
2134: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2135: char *cp;
1.4 kristaps 2136:
1.59 schwarze 2137: /*
1.151 schwarze 2138: * Remember the original working directory, if possible.
2139: * This will be needed if the second or a later directory
2140: * on the command line is given as a relative path.
2141: * Do not error out if the current directory is not
2142: * searchable: Maybe it won't be needed after all.
2143: */
2144: if (0 == getcwd_status) {
2145: if (NULL == getcwd(startdir, sizeof(startdir))) {
2146: getcwd_status = 2;
2147: (void)strlcpy(startdir, strerror(errno),
2148: sizeof(startdir));
2149: } else
2150: getcwd_status = 1;
2151: }
2152:
2153: /*
2154: * We are leaving the old base directory.
2155: * Do not use it any longer, not even for messages.
2156: */
2157: *basedir = '\0';
2158:
2159: /*
2160: * If and only if the directory was changed earlier and
2161: * the next directory to process is given as a relative path,
2162: * first go back, or bail out if that is impossible.
1.59 schwarze 2163: */
1.151 schwarze 2164: if (chdir_status && '/' != *targetdir) {
2165: if (2 == getcwd_status) {
1.59 schwarze 2166: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2167: say("", "getcwd: %s", startdir);
1.197 schwarze 2168: return 0;
1.59 schwarze 2169: }
1.151 schwarze 2170: if (-1 == chdir(startdir)) {
1.59 schwarze 2171: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2172: say("", "&chdir %s", startdir);
1.197 schwarze 2173: return 0;
1.59 schwarze 2174: }
2175: }
1.151 schwarze 2176:
2177: /*
2178: * Always resolve basedir to the canonicalized absolute
2179: * pathname and append a trailing slash, such that
2180: * we can reliably check whether files are inside.
2181: */
1.59 schwarze 2182: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2183: if (report_baddir || errno != ENOENT) {
2184: exitcode = (int)MANDOCLEVEL_BADARG;
2185: say("", "&%s: realpath", targetdir);
2186: }
1.197 schwarze 2187: return 0;
1.59 schwarze 2188: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2189: if (report_baddir || errno != ENOENT) {
2190: exitcode = (int)MANDOCLEVEL_BADARG;
2191: say("", "&chdir");
2192: }
1.197 schwarze 2193: return 0;
1.150 schwarze 2194: }
1.151 schwarze 2195: chdir_status = 1;
1.150 schwarze 2196: cp = strchr(basedir, '\0');
2197: if ('/' != cp[-1]) {
2198: if (cp - basedir >= PATH_MAX - 1) {
2199: exitcode = (int)MANDOCLEVEL_SYSERR;
2200: say("", "Filename too long");
1.197 schwarze 2201: return 0;
1.150 schwarze 2202: }
2203: *cp++ = '/';
2204: *cp = '\0';
1.4 kristaps 2205: }
1.197 schwarze 2206: return 1;
1.56 schwarze 2207: }
2208:
2209: static void
1.59 schwarze 2210: say(const char *file, const char *format, ...)
1.56 schwarze 2211: {
2212: va_list ap;
1.123 schwarze 2213: int use_errno;
1.56 schwarze 2214:
1.59 schwarze 2215: if ('\0' != *basedir)
2216: fprintf(stderr, "%s", basedir);
2217: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2218: fputc('/', stderr);
1.56 schwarze 2219: if ('\0' != *file)
1.59 schwarze 2220: fprintf(stderr, "%s", file);
2221:
1.123 schwarze 2222: use_errno = 1;
2223: if (NULL != format) {
2224: switch (*format) {
1.141 schwarze 2225: case '&':
1.123 schwarze 2226: format++;
2227: break;
1.141 schwarze 2228: case '\0':
1.123 schwarze 2229: format = NULL;
2230: break;
2231: default:
2232: use_errno = 0;
2233: break;
2234: }
2235: }
2236: if (NULL != format) {
2237: if ('\0' != *basedir || '\0' != *file)
2238: fputs(": ", stderr);
2239: va_start(ap, format);
2240: vfprintf(stderr, format, ap);
2241: va_end(ap);
2242: }
2243: if (use_errno) {
2244: if ('\0' != *basedir || '\0' != *file || NULL != format)
2245: fputs(": ", stderr);
1.59 schwarze 2246: perror(NULL);
1.123 schwarze 2247: } else
2248: fputc('\n', stderr);
1.1 kristaps 2249: }
CVSweb