Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.140
1.140 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.139 2015/03/13 00:18:45 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.136 schwarze 4: * Copyright (c) 2011-2015 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 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: *
1.140 ! schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.140 ! schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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: */
1.117 schwarze 18: #include <sys/types.h>
1.47 schwarze 19: #include <sys/stat.h>
1.72 schwarze 20: #include <sys/wait.h>
1.1 schwarze 21:
22: #include <assert.h>
1.33 schwarze 23: #include <ctype.h>
1.34 schwarze 24: #include <errno.h>
1.1 schwarze 25: #include <fcntl.h>
1.47 schwarze 26: #include <fts.h>
1.1 schwarze 27: #include <getopt.h>
1.44 schwarze 28: #include <limits.h>
1.47 schwarze 29: #include <stddef.h>
1.1 schwarze 30: #include <stdio.h>
31: #include <stdint.h>
32: #include <stdlib.h>
33: #include <string.h>
1.14 schwarze 34: #include <unistd.h>
1.1 schwarze 35:
1.47 schwarze 36: #include <ohash.h>
37: #include <sqlite3.h>
38:
39: #include "mdoc.h"
1.1 schwarze 40: #include "man.h"
41: #include "mandoc.h"
1.76 schwarze 42: #include "mandoc_aux.h"
1.140 ! schwarze 43: #include "manconf.h"
1.47 schwarze 44: #include "mansearch.h"
1.1 schwarze 45:
1.68 schwarze 46: extern int mansearch_keymax;
47: extern const char *const mansearch_keynames[];
48:
1.47 schwarze 49: #define SQL_EXEC(_v) \
50: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 51: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 52: #define SQL_BIND_TEXT(_s, _i, _v) \
53: if (SQLITE_OK != sqlite3_bind_text \
54: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 55: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 56: #define SQL_BIND_INT(_s, _i, _v) \
57: if (SQLITE_OK != sqlite3_bind_int \
58: ((_s), (_i)++, (_v))) \
1.91 schwarze 59: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 60: #define SQL_BIND_INT64(_s, _i, _v) \
61: if (SQLITE_OK != sqlite3_bind_int64 \
62: ((_s), (_i)++, (_v))) \
1.91 schwarze 63: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 64: #define SQL_STEP(_s) \
65: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 66: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 67:
1.47 schwarze 68: enum op {
69: OP_DEFAULT = 0, /* new dbs from dir list or default config */
70: OP_CONFFILE, /* new databases from custom config file */
71: OP_UPDATE, /* delete/add entries in existing database */
72: OP_DELETE, /* delete entries from existing database */
73: OP_TEST /* change no databases, report potential problems */
74: };
1.11 schwarze 75:
1.47 schwarze 76: struct str {
77: const struct mpage *mpage; /* if set, the owning parse */
78: uint64_t mask; /* bitmask in sequence */
1.131 schwarze 79: char key[]; /* rendered text */
1.28 schwarze 80: };
81:
1.47 schwarze 82: struct inodev {
83: ino_t st_ino;
84: dev_t st_dev;
85: };
1.28 schwarze 86:
1.47 schwarze 87: struct mpage {
88: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 89: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 90: char *sec; /* section from file content */
91: char *arch; /* architecture from file content */
92: char *title; /* title from file content */
93: char *desc; /* description from file content */
94: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 95: int form; /* format from file content */
1.129 schwarze 96: int name_head_done;
1.28 schwarze 97: };
98:
1.47 schwarze 99: struct mlink {
100: char file[PATH_MAX]; /* filename rel. to manpath */
101: char *dsec; /* section from directory */
102: char *arch; /* architecture from directory */
103: char *name; /* name from file name (not empty) */
104: char *fsec; /* section from file name suffix */
105: struct mlink *next; /* singly linked list */
1.75 schwarze 106: struct mpage *mpage; /* parent */
1.116 schwarze 107: int dform; /* format from directory */
108: int fform; /* format from file name suffix */
1.81 schwarze 109: int gzip; /* filename has a .gz suffix */
1.2 schwarze 110: };
111:
1.47 schwarze 112: enum stmt {
113: STMT_DELETE_PAGE = 0, /* delete mpage */
114: STMT_INSERT_PAGE, /* insert mpage */
115: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 116: STMT_INSERT_NAME, /* insert name */
1.130 schwarze 117: STMT_SELECT_NAME, /* retrieve existing name flags */
1.47 schwarze 118: STMT_INSERT_KEY, /* insert parsed key */
119: STMT__MAX
1.1 schwarze 120: };
121:
1.127 schwarze 122: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_meta *,
123: const struct mdoc_node *);
1.1 schwarze 124:
1.19 schwarze 125: struct mdoc_handler {
1.47 schwarze 126: mdoc_fp fp; /* optional handler */
127: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 128: };
129:
1.47 schwarze 130: static void dbclose(int);
1.131 schwarze 131: static void dbadd(struct mpage *);
1.75 schwarze 132: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 133: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 134: static int dbopen(int);
135: static void dbprune(void);
136: static void filescan(const char *);
137: static void *hash_alloc(size_t, void *);
1.106 espie 138: static void hash_free(void *, void *);
139: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 140: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 141: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 142: static void mlink_free(struct mlink *);
143: static void mlinks_undupe(struct mpage *);
144: static void mpages_free(void);
1.131 schwarze 145: static void mpages_merge(struct mparse *);
1.87 schwarze 146: static void names_check(void);
1.81 schwarze 147: static void parse_cat(struct mpage *, int);
1.129 schwarze 148: static void parse_man(struct mpage *, const struct man_meta *,
149: const struct man_node *);
1.127 schwarze 150: static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
151: const struct mdoc_node *);
152: static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
153: const struct mdoc_node *);
154: static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
155: const struct mdoc_node *);
156: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
157: const struct mdoc_node *);
1.133 schwarze 158: static void parse_mdoc_fname(struct mpage *, const struct mdoc_node *);
1.127 schwarze 159: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
160: const struct mdoc_node *);
1.128 schwarze 161: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
162: const struct mdoc_node *);
1.127 schwarze 163: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
164: const struct mdoc_node *);
165: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
166: const struct mdoc_node *);
167: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
168: const struct mdoc_node *);
169: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
170: const struct mdoc_node *);
1.69 schwarze 171: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 172: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 173: static void putmdockey(const struct mpage *,
174: const struct mdoc_node *, uint64_t);
1.131 schwarze 175: static int render_string(char **, size_t *);
1.47 schwarze 176: static void say(const char *, const char *, ...);
1.120 schwarze 177: static int set_basedir(const char *, int);
1.47 schwarze 178: static int treescan(void);
179: static size_t utf8(unsigned int, char [7]);
180:
1.72 schwarze 181: static char tempfilename[32];
1.47 schwarze 182: static char *progname;
1.59 schwarze 183: static int nodb; /* no database changes */
1.73 schwarze 184: static int mparse_options; /* abort the parse early */
1.98 schwarze 185: static int use_all; /* use all found files */
186: static int debug; /* print what we're doing */
187: static int warnings; /* warn about crap */
1.52 schwarze 188: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 189: static int exitcode; /* to be returned by main */
1.98 schwarze 190: static enum op op; /* operational mode */
1.47 schwarze 191: static char basedir[PATH_MAX]; /* current base directory */
1.131 schwarze 192: static struct mchars *mchars; /* table of named characters */
1.47 schwarze 193: static struct ohash mpages; /* table of distinct manual pages */
194: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 195: static struct ohash names; /* table of all names */
1.47 schwarze 196: static struct ohash strings; /* table of all strings */
197: static sqlite3 *db = NULL; /* current database */
198: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 199: static uint64_t name_mask;
1.47 schwarze 200:
1.19 schwarze 201: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 202: { NULL, 0 }, /* Ap */
203: { NULL, 0 }, /* Dd */
204: { NULL, 0 }, /* Dt */
205: { NULL, 0 }, /* Os */
206: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
207: { parse_mdoc_head, TYPE_Ss }, /* Ss */
208: { NULL, 0 }, /* Pp */
209: { NULL, 0 }, /* D1 */
210: { NULL, 0 }, /* Dl */
211: { NULL, 0 }, /* Bd */
212: { NULL, 0 }, /* Ed */
213: { NULL, 0 }, /* Bl */
214: { NULL, 0 }, /* El */
215: { NULL, 0 }, /* It */
216: { NULL, 0 }, /* Ad */
217: { NULL, TYPE_An }, /* An */
218: { NULL, TYPE_Ar }, /* Ar */
219: { NULL, TYPE_Cd }, /* Cd */
220: { NULL, TYPE_Cm }, /* Cm */
221: { NULL, TYPE_Dv }, /* Dv */
222: { NULL, TYPE_Er }, /* Er */
223: { NULL, TYPE_Ev }, /* Ev */
224: { NULL, 0 }, /* Ex */
225: { NULL, TYPE_Fa }, /* Fa */
226: { parse_mdoc_Fd, 0 }, /* Fd */
227: { NULL, TYPE_Fl }, /* Fl */
228: { parse_mdoc_Fn, 0 }, /* Fn */
229: { NULL, TYPE_Ft }, /* Ft */
230: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 231: { NULL, TYPE_In }, /* In */
1.47 schwarze 232: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 233: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 234: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 235: { NULL, 0 }, /* Op */
236: { NULL, 0 }, /* Ot */
237: { NULL, TYPE_Pa }, /* Pa */
238: { NULL, 0 }, /* Rv */
1.49 schwarze 239: { NULL, TYPE_St }, /* St */
1.47 schwarze 240: { NULL, TYPE_Va }, /* Va */
241: { parse_mdoc_body, TYPE_Va }, /* Vt */
242: { parse_mdoc_Xr, 0 }, /* Xr */
243: { NULL, 0 }, /* %A */
244: { NULL, 0 }, /* %B */
245: { NULL, 0 }, /* %D */
246: { NULL, 0 }, /* %I */
247: { NULL, 0 }, /* %J */
248: { NULL, 0 }, /* %N */
249: { NULL, 0 }, /* %O */
250: { NULL, 0 }, /* %P */
251: { NULL, 0 }, /* %R */
252: { NULL, 0 }, /* %T */
253: { NULL, 0 }, /* %V */
254: { NULL, 0 }, /* Ac */
255: { NULL, 0 }, /* Ao */
256: { NULL, 0 }, /* Aq */
257: { NULL, TYPE_At }, /* At */
258: { NULL, 0 }, /* Bc */
259: { NULL, 0 }, /* Bf */
260: { NULL, 0 }, /* Bo */
261: { NULL, 0 }, /* Bq */
262: { NULL, TYPE_Bsx }, /* Bsx */
263: { NULL, TYPE_Bx }, /* Bx */
264: { NULL, 0 }, /* Db */
265: { NULL, 0 }, /* Dc */
266: { NULL, 0 }, /* Do */
267: { NULL, 0 }, /* Dq */
268: { NULL, 0 }, /* Ec */
269: { NULL, 0 }, /* Ef */
270: { NULL, TYPE_Em }, /* Em */
271: { NULL, 0 }, /* Eo */
272: { NULL, TYPE_Fx }, /* Fx */
273: { NULL, TYPE_Ms }, /* Ms */
274: { NULL, 0 }, /* No */
275: { NULL, 0 }, /* Ns */
276: { NULL, TYPE_Nx }, /* Nx */
277: { NULL, TYPE_Ox }, /* Ox */
278: { NULL, 0 }, /* Pc */
279: { NULL, 0 }, /* Pf */
280: { NULL, 0 }, /* Po */
281: { NULL, 0 }, /* Pq */
282: { NULL, 0 }, /* Qc */
283: { NULL, 0 }, /* Ql */
284: { NULL, 0 }, /* Qo */
285: { NULL, 0 }, /* Qq */
286: { NULL, 0 }, /* Re */
287: { NULL, 0 }, /* Rs */
288: { NULL, 0 }, /* Sc */
289: { NULL, 0 }, /* So */
290: { NULL, 0 }, /* Sq */
291: { NULL, 0 }, /* Sm */
292: { NULL, 0 }, /* Sx */
293: { NULL, TYPE_Sy }, /* Sy */
294: { NULL, TYPE_Tn }, /* Tn */
295: { NULL, 0 }, /* Ux */
296: { NULL, 0 }, /* Xc */
297: { NULL, 0 }, /* Xo */
1.128 schwarze 298: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 299: { NULL, 0 }, /* Fc */
300: { NULL, 0 }, /* Oo */
301: { NULL, 0 }, /* Oc */
302: { NULL, 0 }, /* Bk */
303: { NULL, 0 }, /* Ek */
304: { NULL, 0 }, /* Bt */
305: { NULL, 0 }, /* Hf */
306: { NULL, 0 }, /* Fr */
307: { NULL, 0 }, /* Ud */
308: { NULL, TYPE_Lb }, /* Lb */
309: { NULL, 0 }, /* Lp */
310: { NULL, TYPE_Lk }, /* Lk */
311: { NULL, TYPE_Mt }, /* Mt */
312: { NULL, 0 }, /* Brq */
313: { NULL, 0 }, /* Bro */
314: { NULL, 0 }, /* Brc */
315: { NULL, 0 }, /* %C */
316: { NULL, 0 }, /* Es */
317: { NULL, 0 }, /* En */
318: { NULL, TYPE_Dx }, /* Dx */
319: { NULL, 0 }, /* %Q */
320: { NULL, 0 }, /* br */
321: { NULL, 0 }, /* sp */
322: { NULL, 0 }, /* %U */
323: { NULL, 0 }, /* Ta */
1.121 schwarze 324: { NULL, 0 }, /* ll */
1.1 schwarze 325: };
326:
1.98 schwarze 327:
1.1 schwarze 328: int
1.3 schwarze 329: mandocdb(int argc, char *argv[])
1.1 schwarze 330: {
1.140 ! schwarze 331: struct manconf conf;
! 332: struct ohash_info mpages_info, mlinks_info;
! 333: struct mparse *mp;
! 334: const char *path_arg;
! 335: size_t j, sz;
1.47 schwarze 336: int ch, i;
337:
1.140 ! schwarze 338: memset(&conf, 0, sizeof(conf));
1.47 schwarze 339: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
340:
341: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 342: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
1.135 schwarze 343: mpages_info.free = mlinks_info.free = hash_free;
344: mpages_info.data = mlinks_info.data = NULL;
1.47 schwarze 345:
346: mpages_info.key_offset = offsetof(struct mpage, inodev);
347: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 348:
349: progname = strrchr(argv[0], '/');
350: if (progname == NULL)
351: progname = argv[0];
352: else
353: ++progname;
354:
1.47 schwarze 355: /*
1.98 schwarze 356: * We accept a few different invocations.
1.47 schwarze 357: * The CHECKOP macro makes sure that invocation styles don't
358: * clobber each other.
359: */
360: #define CHECKOP(_op, _ch) do \
361: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 362: fprintf(stderr, "%s: -%c: Conflicting option\n", \
363: progname, (_ch)); \
1.47 schwarze 364: goto usage; \
365: } while (/*CONSTCOND*/0)
1.10 schwarze 366:
1.47 schwarze 367: path_arg = NULL;
1.28 schwarze 368: op = OP_DEFAULT;
1.1 schwarze 369:
1.83 schwarze 370: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 371: switch (ch) {
1.98 schwarze 372: case 'a':
1.6 schwarze 373: use_all = 1;
374: break;
1.98 schwarze 375: case 'C':
1.47 schwarze 376: CHECKOP(op, ch);
377: path_arg = optarg;
1.28 schwarze 378: op = OP_CONFFILE;
1.25 schwarze 379: break;
1.98 schwarze 380: case 'D':
1.82 schwarze 381: debug++;
382: break;
1.98 schwarze 383: case 'd':
1.47 schwarze 384: CHECKOP(op, ch);
385: path_arg = optarg;
1.2 schwarze 386: op = OP_UPDATE;
1.1 schwarze 387: break;
1.98 schwarze 388: case 'n':
1.47 schwarze 389: nodb = 1;
390: break;
1.98 schwarze 391: case 'p':
1.83 schwarze 392: warnings = 1;
393: break;
1.98 schwarze 394: case 'Q':
1.73 schwarze 395: mparse_options |= MPARSE_QUICK;
1.59 schwarze 396: break;
1.98 schwarze 397: case 'T':
1.52 schwarze 398: if (strcmp(optarg, "utf8")) {
1.111 schwarze 399: fprintf(stderr, "%s: -T%s: "
400: "Unsupported output format\n",
401: progname, optarg);
1.52 schwarze 402: goto usage;
403: }
404: write_utf8 = 1;
405: break;
1.98 schwarze 406: case 't':
1.47 schwarze 407: CHECKOP(op, ch);
1.28 schwarze 408: dup2(STDOUT_FILENO, STDERR_FILENO);
409: op = OP_TEST;
1.47 schwarze 410: nodb = warnings = 1;
1.28 schwarze 411: break;
1.98 schwarze 412: case 'u':
1.47 schwarze 413: CHECKOP(op, ch);
414: path_arg = optarg;
1.1 schwarze 415: op = OP_DELETE;
416: break;
1.98 schwarze 417: case 'v':
1.83 schwarze 418: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 419: break;
1.1 schwarze 420: default:
1.28 schwarze 421: goto usage;
1.1 schwarze 422: }
423:
424: argc -= optind;
425: argv += optind;
426:
1.28 schwarze 427: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 428: fprintf(stderr, "%s: -C: Too many arguments\n",
429: progname);
1.28 schwarze 430: goto usage;
431: }
432:
1.47 schwarze 433: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 434: mchars = mchars_alloc();
1.137 schwarze 435: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.131 schwarze 436: mchars, NULL);
1.47 schwarze 437: ohash_init(&mpages, 6, &mpages_info);
438: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 439:
1.47 schwarze 440: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 441:
1.47 schwarze 442: /*
1.109 schwarze 443: * Most of these deal with a specific directory.
1.97 schwarze 444: * Jump into that directory first.
1.47 schwarze 445: */
1.120 schwarze 446: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 447: goto out;
1.97 schwarze 448:
1.95 schwarze 449: if (dbopen(1)) {
1.97 schwarze 450: /*
451: * The existing database is usable. Process
452: * all files specified on the command-line.
453: */
454: use_all = 1;
1.95 schwarze 455: for (i = 0; i < argc; i++)
456: filescan(argv[i]);
457: if (OP_TEST != op)
458: dbprune();
459: } else {
460: /*
461: * Database missing or corrupt.
462: * Recreate from scratch.
463: */
1.97 schwarze 464: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 465: op = OP_DEFAULT;
466: if (0 == treescan())
467: goto out;
468: if (0 == dbopen(0))
469: goto out;
470: }
1.47 schwarze 471: if (OP_DELETE != op)
1.131 schwarze 472: mpages_merge(mp);
1.95 schwarze 473: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 474: } else {
475: /*
476: * If we have arguments, use them as our manpaths.
477: * If we don't, grok from manpath(1) or however else
1.140 ! schwarze 478: * manconf_parse() wants to do it.
1.47 schwarze 479: */
480: if (argc > 0) {
1.140 ! schwarze 481: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 482: argc, sizeof(char *));
1.140 ! schwarze 483: conf.manpath.sz = (size_t)argc;
1.47 schwarze 484: for (i = 0; i < argc; i++)
1.140 ! schwarze 485: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 486: } else
1.140 ! schwarze 487: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 488:
1.140 ! schwarze 489: if (conf.manpath.sz == 0) {
1.84 schwarze 490: exitcode = (int)MANDOCLEVEL_BADARG;
491: say("", "Empty manpath");
492: }
1.2 schwarze 493:
1.47 schwarze 494: /*
495: * First scan the tree rooted at a base directory, then
496: * build a new database and finally move it into place.
497: * Ignore zero-length directories and strip trailing
498: * slashes.
499: */
1.140 ! schwarze 500: for (j = 0; j < conf.manpath.sz; j++) {
! 501: sz = strlen(conf.manpath.paths[j]);
! 502: if (sz && conf.manpath.paths[j][sz - 1] == '/')
! 503: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 504: if (0 == sz)
505: continue;
1.2 schwarze 506:
1.47 schwarze 507: if (j) {
508: ohash_init(&mpages, 6, &mpages_info);
509: ohash_init(&mlinks, 6, &mlinks_info);
510: }
1.1 schwarze 511:
1.140 ! schwarze 512: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 513: continue;
1.47 schwarze 514: if (0 == treescan())
1.120 schwarze 515: continue;
1.47 schwarze 516: if (0 == dbopen(0))
1.120 schwarze 517: continue;
1.41 deraadt 518:
1.131 schwarze 519: mpages_merge(mp);
1.104 schwarze 520: if (warnings && !nodb &&
1.87 schwarze 521: ! (MPARSE_QUICK & mparse_options))
522: names_check();
1.47 schwarze 523: dbclose(0);
1.28 schwarze 524:
1.140 ! schwarze 525: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 526: mpages_free();
527: ohash_delete(&mpages);
528: ohash_delete(&mlinks);
529: }
1.2 schwarze 530: }
1.47 schwarze 531: }
532: out:
1.140 ! schwarze 533: manconf_free(&conf);
1.123 schwarze 534: mparse_free(mp);
1.131 schwarze 535: mchars_free(mchars);
1.47 schwarze 536: mpages_free();
537: ohash_delete(&mpages);
538: ohash_delete(&mlinks);
539: return(exitcode);
540: usage:
1.83 schwarze 541: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
542: " %s [-aDnpQ] [-Tutf8] dir ...\n"
543: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
544: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 545: " %s [-Q] -t file ...\n",
1.98 schwarze 546: progname, progname, progname,
1.47 schwarze 547: progname, progname);
1.1 schwarze 548:
1.47 schwarze 549: return((int)MANDOCLEVEL_BADARG);
550: }
1.1 schwarze 551:
1.47 schwarze 552: /*
553: * Scan a directory tree rooted at "basedir" for manpages.
554: * We use fts(), scanning directory parts along the way for clues to our
555: * section and architecture.
556: *
557: * If use_all has been specified, grok all files.
558: * If not, sanitise paths to the following:
559: *
1.98 schwarze 560: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 561: * or
562: * [./]cat<section>[/<arch>]/<name>.0
563: *
564: * TODO: accomodate for multi-language directories.
565: */
566: static int
567: treescan(void)
568: {
1.96 schwarze 569: char buf[PATH_MAX];
1.47 schwarze 570: FTS *f;
571: FTSENT *ff;
572: struct mlink *mlink;
1.81 schwarze 573: int dform, gzip;
1.51 schwarze 574: char *dsec, *arch, *fsec, *cp;
575: const char *path;
1.47 schwarze 576: const char *argv[2];
1.1 schwarze 577:
1.47 schwarze 578: argv[0] = ".";
579: argv[1] = (char *)NULL;
1.1 schwarze 580:
1.96 schwarze 581: f = fts_open((char * const *)argv,
582: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 583: if (NULL == f) {
584: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 585: say("", "&fts_open");
1.47 schwarze 586: return(0);
587: }
1.2 schwarze 588:
1.47 schwarze 589: dsec = arch = NULL;
590: dform = FORM_NONE;
1.2 schwarze 591:
1.47 schwarze 592: while (NULL != (ff = fts_read(f))) {
593: path = ff->fts_path + 2;
1.96 schwarze 594: switch (ff->fts_info) {
595:
596: /*
597: * Symbolic links require various sanity checks,
598: * then get handled just like regular files.
599: */
1.98 schwarze 600: case FTS_SL:
1.96 schwarze 601: if (NULL == realpath(path, buf)) {
602: if (warnings)
603: say(path, "&realpath");
604: continue;
605: }
606: if (strstr(buf, basedir) != buf) {
607: if (warnings) say("",
608: "%s: outside base directory", buf);
609: continue;
610: }
611: /* Use logical inode to avoid mpages dupe. */
612: if (-1 == stat(path, ff->fts_statp)) {
613: if (warnings)
614: say(path, "&stat");
615: continue;
616: }
617: /* FALLTHROUGH */
618:
1.15 schwarze 619: /*
1.47 schwarze 620: * If we're a regular file, add an mlink by using the
621: * stored directory data and handling the filename.
1.15 schwarze 622: */
1.98 schwarze 623: case FTS_F:
1.47 schwarze 624: if (0 == strcmp(path, MANDOC_DB))
625: continue;
626: if ( ! use_all && ff->fts_level < 2) {
627: if (warnings)
628: say(path, "Extraneous file");
629: continue;
1.81 schwarze 630: }
631: gzip = 0;
632: fsec = NULL;
633: while (NULL == fsec) {
634: fsec = strrchr(ff->fts_name, '.');
635: if (NULL == fsec || strcmp(fsec+1, "gz"))
636: break;
637: gzip = 1;
638: *fsec = '\0';
639: fsec = NULL;
640: }
641: if (NULL == fsec) {
1.47 schwarze 642: if ( ! use_all) {
643: if (warnings)
644: say(path,
645: "No filename suffix");
646: continue;
647: }
648: } else if (0 == strcmp(++fsec, "html")) {
649: if (warnings)
650: say(path, "Skip html");
651: continue;
652: } else if (0 == strcmp(fsec, "ps")) {
653: if (warnings)
654: say(path, "Skip ps");
655: continue;
656: } else if (0 == strcmp(fsec, "pdf")) {
657: if (warnings)
658: say(path, "Skip pdf");
659: continue;
660: } else if ( ! use_all &&
1.138 schwarze 661: ((FORM_SRC == dform &&
662: strncmp(fsec, dsec, strlen(dsec))) ||
1.47 schwarze 663: (FORM_CAT == dform && strcmp(fsec, "0")))) {
664: if (warnings)
665: say(path, "Wrong filename suffix");
666: continue;
667: } else
668: fsec[-1] = '\0';
1.51 schwarze 669:
1.47 schwarze 670: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 671: if (strlcpy(mlink->file, path,
672: sizeof(mlink->file)) >=
673: sizeof(mlink->file)) {
674: say(path, "Filename too long");
675: free(mlink);
676: continue;
677: }
1.47 schwarze 678: mlink->dform = dform;
1.51 schwarze 679: mlink->dsec = dsec;
680: mlink->arch = arch;
681: mlink->name = ff->fts_name;
682: mlink->fsec = fsec;
1.81 schwarze 683: mlink->gzip = gzip;
1.47 schwarze 684: mlink_add(mlink, ff->fts_statp);
685: continue;
1.96 schwarze 686:
1.98 schwarze 687: case FTS_D:
1.96 schwarze 688: /* FALLTHROUGH */
1.98 schwarze 689: case FTS_DP:
1.96 schwarze 690: break;
691:
692: default:
1.47 schwarze 693: if (warnings)
694: say(path, "Not a regular file");
695: continue;
696: }
697:
698: switch (ff->fts_level) {
1.98 schwarze 699: case 0:
1.47 schwarze 700: /* Ignore the root directory. */
701: break;
1.98 schwarze 702: case 1:
1.47 schwarze 703: /*
704: * This might contain manX/ or catX/.
705: * Try to infer this from the name.
706: * If we're not in use_all, enforce it.
707: */
708: cp = ff->fts_name;
709: if (FTS_DP == ff->fts_info)
710: break;
1.15 schwarze 711:
1.47 schwarze 712: if (0 == strncmp(cp, "man", 3)) {
713: dform = FORM_SRC;
714: dsec = cp + 3;
715: } else if (0 == strncmp(cp, "cat", 3)) {
716: dform = FORM_CAT;
717: dsec = cp + 3;
1.51 schwarze 718: } else {
719: dform = FORM_NONE;
720: dsec = NULL;
1.26 schwarze 721: }
1.47 schwarze 722:
1.98 schwarze 723: if (NULL != dsec || use_all)
1.47 schwarze 724: break;
725:
726: if (warnings)
727: say(path, "Unknown directory part");
728: fts_set(f, ff, FTS_SKIP);
729: break;
1.98 schwarze 730: case 2:
1.47 schwarze 731: /*
732: * Possibly our architecture.
733: * If we're descending, keep tabs on it.
734: */
735: if (FTS_DP != ff->fts_info && NULL != dsec)
736: arch = ff->fts_name;
1.51 schwarze 737: else
738: arch = NULL;
1.47 schwarze 739: break;
740: default:
741: if (FTS_DP == ff->fts_info || use_all)
742: break;
743: if (warnings)
744: say(path, "Extraneous directory part");
745: fts_set(f, ff, FTS_SKIP);
746: break;
1.14 schwarze 747: }
1.47 schwarze 748: }
749:
750: fts_close(f);
751: return(1);
752: }
1.1 schwarze 753:
1.47 schwarze 754: /*
755: * Add a file to the mlinks table.
756: * Do not verify that it's a "valid" looking manpage (we'll do that
757: * later).
758: *
759: * Try to infer the manual section, architecture, and page name from the
760: * path, assuming it looks like
761: *
1.98 schwarze 762: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 763: * or
764: * [./]cat<section>[/<arch>]/<name>.0
765: *
766: * See treescan() for the fts(3) version of this.
767: */
768: static void
769: filescan(const char *file)
770: {
771: char buf[PATH_MAX];
772: struct stat st;
773: struct mlink *mlink;
774: char *p, *start;
775:
776: assert(use_all);
777:
778: if (0 == strncmp(file, "./", 2))
779: file += 2;
780:
1.96 schwarze 781: /*
782: * We have to do lstat(2) before realpath(3) loses
783: * the information whether this is a symbolic link.
784: * We need to know that because for symbolic links,
785: * we want to use the orginal file name, while for
786: * regular files, we want to use the real path.
787: */
788: if (-1 == lstat(file, &st)) {
789: exitcode = (int)MANDOCLEVEL_BADARG;
790: say(file, "&lstat");
791: return;
792: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
793: exitcode = (int)MANDOCLEVEL_BADARG;
794: say(file, "Not a regular file");
795: return;
796: }
797:
798: /*
799: * We have to resolve the file name to the real path
800: * in any case for the base directory check.
801: */
1.47 schwarze 802: if (NULL == realpath(file, buf)) {
803: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 804: say(file, "&realpath");
1.47 schwarze 805: return;
1.63 schwarze 806: }
807:
1.109 schwarze 808: if (OP_TEST == op)
1.63 schwarze 809: start = buf;
1.109 schwarze 810: else if (strstr(buf, basedir) == buf)
811: start = buf + strlen(basedir);
1.63 schwarze 812: else {
1.47 schwarze 813: exitcode = (int)MANDOCLEVEL_BADARG;
814: say("", "%s: outside base directory", buf);
815: return;
1.63 schwarze 816: }
817:
1.96 schwarze 818: /*
819: * Now we are sure the file is inside our tree.
820: * If it is a symbolic link, ignore the real path
821: * and use the original name.
822: * This implies passing stuff like "cat1/../man1/foo.1"
823: * on the command line won't work. So don't do that.
824: * Note the stat(2) can still fail if the link target
825: * doesn't exist.
826: */
827: if (S_IFLNK & st.st_mode) {
828: if (-1 == stat(buf, &st)) {
829: exitcode = (int)MANDOCLEVEL_BADARG;
830: say(file, "&stat");
831: return;
832: }
1.100 schwarze 833: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
834: say(file, "Filename too long");
835: return;
836: }
1.109 schwarze 837: start = buf;
838: if (OP_TEST != op && strstr(buf, basedir) == buf)
839: start += strlen(basedir);
1.1 schwarze 840: }
1.63 schwarze 841:
1.47 schwarze 842: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 843: mlink->dform = FORM_NONE;
1.100 schwarze 844: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
845: sizeof(mlink->file)) {
846: say(start, "Filename too long");
1.134 schwarze 847: free(mlink);
1.100 schwarze 848: return;
849: }
1.1 schwarze 850:
1.10 schwarze 851: /*
1.47 schwarze 852: * First try to guess our directory structure.
853: * If we find a separator, try to look for man* or cat*.
854: * If we find one of these and what's underneath is a directory,
855: * assume it's an architecture.
1.10 schwarze 856: */
1.47 schwarze 857: if (NULL != (p = strchr(start, '/'))) {
858: *p++ = '\0';
859: if (0 == strncmp(start, "man", 3)) {
860: mlink->dform = FORM_SRC;
1.51 schwarze 861: mlink->dsec = start + 3;
1.47 schwarze 862: } else if (0 == strncmp(start, "cat", 3)) {
863: mlink->dform = FORM_CAT;
1.51 schwarze 864: mlink->dsec = start + 3;
1.47 schwarze 865: }
1.10 schwarze 866:
1.47 schwarze 867: start = p;
868: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
869: *p++ = '\0';
1.51 schwarze 870: mlink->arch = start;
1.47 schwarze 871: start = p;
1.41 deraadt 872: }
1.47 schwarze 873: }
1.7 schwarze 874:
1.47 schwarze 875: /*
876: * Now check the file suffix.
877: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
878: */
879: p = strrchr(start, '\0');
880: while (p-- > start && '/' != *p && '.' != *p)
881: /* Loop. */ ;
882:
883: if ('.' == *p) {
884: *p++ = '\0';
1.51 schwarze 885: mlink->fsec = p;
1.47 schwarze 886: }
1.41 deraadt 887:
1.47 schwarze 888: /*
889: * Now try to parse the name.
890: * Use the filename portion of the path.
891: */
892: mlink->name = start;
893: if (NULL != (p = strrchr(start, '/'))) {
894: mlink->name = p + 1;
895: *p = '\0';
896: }
897: mlink_add(mlink, &st);
898: }
1.2 schwarze 899:
1.47 schwarze 900: static void
901: mlink_add(struct mlink *mlink, const struct stat *st)
902: {
903: struct inodev inodev;
904: struct mpage *mpage;
905: unsigned int slot;
906:
907: assert(NULL != mlink->file);
908:
1.51 schwarze 909: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
910: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
911: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
912: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 913:
914: if ('0' == *mlink->fsec) {
915: free(mlink->fsec);
916: mlink->fsec = mandoc_strdup(mlink->dsec);
917: mlink->fform = FORM_CAT;
918: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
919: mlink->fform = FORM_SRC;
920: else
921: mlink->fform = FORM_NONE;
1.1 schwarze 922:
1.47 schwarze 923: slot = ohash_qlookup(&mlinks, mlink->file);
924: assert(NULL == ohash_find(&mlinks, slot));
925: ohash_insert(&mlinks, slot, mlink);
926:
1.139 schwarze 927: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 928: inodev.st_ino = st->st_ino;
929: inodev.st_dev = st->st_dev;
930: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
931: sizeof(struct inodev), inodev.st_ino);
932: mpage = ohash_find(&mpages, slot);
933: if (NULL == mpage) {
934: mpage = mandoc_calloc(1, sizeof(struct mpage));
935: mpage->inodev.st_ino = inodev.st_ino;
936: mpage->inodev.st_dev = inodev.st_dev;
937: ohash_insert(&mpages, slot, mpage);
938: } else
939: mlink->next = mpage->mlinks;
940: mpage->mlinks = mlink;
1.75 schwarze 941: mlink->mpage = mpage;
1.47 schwarze 942: }
1.1 schwarze 943:
1.47 schwarze 944: static void
945: mlink_free(struct mlink *mlink)
946: {
1.1 schwarze 947:
1.47 schwarze 948: free(mlink->dsec);
949: free(mlink->arch);
950: free(mlink->name);
951: free(mlink->fsec);
952: free(mlink);
953: }
1.1 schwarze 954:
1.47 schwarze 955: static void
956: mpages_free(void)
957: {
958: struct mpage *mpage;
959: struct mlink *mlink;
960: unsigned int slot;
961:
962: mpage = ohash_first(&mpages, &slot);
963: while (NULL != mpage) {
964: while (NULL != (mlink = mpage->mlinks)) {
965: mpage->mlinks = mlink->next;
966: mlink_free(mlink);
967: }
968: free(mpage->sec);
969: free(mpage->arch);
970: free(mpage->title);
971: free(mpage->desc);
972: free(mpage);
973: mpage = ohash_next(&mpages, &slot);
974: }
975: }
1.1 schwarze 976:
1.47 schwarze 977: /*
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;
992: prev = &mpage->mlinks;
993: while (NULL != (mlink = *prev)) {
994: if (FORM_CAT != mlink->dform) {
995: mpage->form = FORM_NONE;
996: goto nextlink;
997: }
1.100 schwarze 998: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 999: bufp = strstr(buf, "cat");
1000: assert(NULL != bufp);
1001: memcpy(bufp, "man", 3);
1002: if (NULL != (bufp = strrchr(buf, '.')))
1003: *++bufp = '\0';
1.100 schwarze 1004: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1005: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1006: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1007: goto nextlink;
1008: if (warnings)
1009: say(mlink->file, "Man source exists: %s", buf);
1010: if (use_all)
1011: goto nextlink;
1012: *prev = mlink->next;
1013: mlink_free(mlink);
1014: continue;
1015: nextlink:
1016: prev = &(*prev)->next;
1.1 schwarze 1017: }
1.47 schwarze 1018: }
1.1 schwarze 1019:
1.88 schwarze 1020: static void
1.50 schwarze 1021: mlink_check(struct mpage *mpage, struct mlink *mlink)
1022: {
1.88 schwarze 1023: struct str *str;
1024: unsigned int slot;
1.50 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 &&
1.88 schwarze 1036: strcasecmp(mpage->sec, mlink->dsec))
1.50 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:
1.88 schwarze 1053: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1054: say(mlink->file, "Architecture \"%s\" manual in "
1055: "\"%s\" directory", mpage->arch, mlink->arch);
1056:
1.88 schwarze 1057: /*
1058: * XXX
1.90 schwarze 1059: * parse_cat() doesn't set NAME_TITLE yet.
1.88 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: */
1.50 schwarze 1069:
1.90 schwarze 1070: slot = ohash_qlookup(&names, mlink->name);
1071: str = ohash_find(&names, slot);
1.88 schwarze 1072: assert(NULL != str);
1.90 schwarze 1073: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1074: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1075: }
1076:
1.47 schwarze 1077: /*
1078: * Run through the files in the global vector "mpages"
1079: * and add them to the database specified in "basedir".
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: */
1084: static void
1.131 schwarze 1085: mpages_merge(struct mparse *mp)
1.47 schwarze 1086: {
1.70 schwarze 1087: char any[] = "any";
1.58 schwarze 1088: struct ohash_info str_info;
1.75 schwarze 1089: struct mpage *mpage, *mpage_dest;
1090: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1091: struct mdoc *mdoc;
1092: struct man *man;
1.75 schwarze 1093: char *sodest;
1.69 schwarze 1094: char *cp;
1.117 schwarze 1095: int fd;
1.58 schwarze 1096: unsigned int pslot;
1.47 schwarze 1097:
1098: str_info.alloc = hash_alloc;
1.106 espie 1099: str_info.calloc = hash_calloc;
1100: str_info.free = hash_free;
1.135 schwarze 1101: str_info.data = NULL;
1.47 schwarze 1102: str_info.key_offset = offsetof(struct str, key);
1103:
1.126 schwarze 1104: if ( ! nodb)
1.64 schwarze 1105: SQL_EXEC("BEGIN TRANSACTION");
1106:
1.47 schwarze 1107: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1108: while (mpage != NULL) {
1.47 schwarze 1109: mlinks_undupe(mpage);
1.136 schwarze 1110: if ((mlink = mpage->mlinks) == NULL) {
1.47 schwarze 1111: mpage = ohash_next(&mpages, &pslot);
1112: continue;
1113: }
1.1 schwarze 1114:
1.90 schwarze 1115: name_mask = NAME_MASK;
1116: ohash_init(&names, 4, &str_info);
1.47 schwarze 1117: ohash_init(&strings, 6, &str_info);
1118: mparse_reset(mp);
1119: mdoc = NULL;
1120: man = NULL;
1.81 schwarze 1121: sodest = NULL;
1122:
1.136 schwarze 1123: mparse_open(mp, &fd, mlink->file);
1.117 schwarze 1124: if (fd == -1) {
1.136 schwarze 1125: say(mlink->file, "&open");
1.117 schwarze 1126: goto nextpage;
1.81 schwarze 1127: }
1.11 schwarze 1128:
1129: /*
1.137 schwarze 1130: * Interpret the file as mdoc(7) or man(7) source
1131: * code, unless it is known to be formatted.
1.11 schwarze 1132: */
1.136 schwarze 1133: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1134: mparse_readfd(mp, fd, mlink->file);
1135: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1136: }
1.11 schwarze 1137:
1.126 schwarze 1138: if (sodest != NULL) {
1.75 schwarze 1139: mlink_dest = ohash_find(&mlinks,
1140: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1141: if (mlink_dest == NULL) {
1142: mandoc_asprintf(&cp, "%s.gz", sodest);
1143: mlink_dest = ohash_find(&mlinks,
1144: ohash_qlookup(&mlinks, cp));
1145: free(cp);
1146: }
1147: if (mlink_dest != NULL) {
1.75 schwarze 1148:
1149: /* The .so target exists. */
1150:
1151: mpage_dest = mlink_dest->mpage;
1152: while (1) {
1153: mlink->mpage = mpage_dest;
1154:
1155: /*
1156: * If the target was already
1157: * processed, add the links
1158: * to the database now.
1159: * Otherwise, this will
1160: * happen when we come
1161: * to the target.
1162: */
1163:
1.94 schwarze 1164: if (mpage_dest->pageid)
1.124 schwarze 1165: dbadd_mlink_name(mlink);
1.75 schwarze 1166:
1.126 schwarze 1167: if (mlink->next == NULL)
1.75 schwarze 1168: break;
1169: mlink = mlink->next;
1170: }
1171:
1172: /* Move all links to the target. */
1173:
1174: mlink->next = mlink_dest->next;
1175: mlink_dest->next = mpage->mlinks;
1176: mpage->mlinks = NULL;
1177: }
1.81 schwarze 1178: goto nextpage;
1.126 schwarze 1179: } else if (mdoc != NULL) {
1.47 schwarze 1180: mpage->form = FORM_SRC;
1.112 schwarze 1181: mpage->sec = mdoc_meta(mdoc)->msec;
1182: mpage->sec = mandoc_strdup(
1.126 schwarze 1183: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1184: mpage->arch = mdoc_meta(mdoc)->arch;
1185: mpage->arch = mandoc_strdup(
1.126 schwarze 1186: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1187: mpage->title =
1188: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1189: } else if (man != NULL) {
1.47 schwarze 1190: mpage->form = FORM_SRC;
1.136 schwarze 1191: mpage->sec = mandoc_strdup(man_meta(man)->msec);
1192: mpage->arch = mandoc_strdup(mlink->arch);
1193: mpage->title = mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1194: } else {
1.47 schwarze 1195: mpage->form = FORM_CAT;
1.136 schwarze 1196: mpage->sec = mandoc_strdup(mlink->dsec);
1197: mpage->arch = mandoc_strdup(mlink->arch);
1198: mpage->title = mandoc_strdup(mlink->name);
1.1 schwarze 1199: }
1.54 schwarze 1200: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1201: if (*mpage->arch != '\0')
1202: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1203:
1.136 schwarze 1204: for ( ; mlink != NULL; mlink = mlink->next) {
1.54 schwarze 1205: if ('\0' != *mlink->dsec)
1206: putkey(mpage, mlink->dsec, TYPE_sec);
1207: if ('\0' != *mlink->fsec)
1208: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1209: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1210: any : mlink->arch, TYPE_arch);
1.90 schwarze 1211: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1212: }
1.41 deraadt 1213:
1.129 schwarze 1214: assert(mpage->desc == NULL);
1215: if (mdoc != NULL)
1.127 schwarze 1216: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 schwarze 1217: else if (man != NULL)
1218: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1219: else
1.117 schwarze 1220: parse_cat(mpage, fd);
1.129 schwarze 1221: if (mpage->desc == NULL)
1.92 schwarze 1222: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1223:
1224: if (warnings && !use_all)
1225: for (mlink = mpage->mlinks; mlink;
1226: mlink = mlink->next)
1227: mlink_check(mpage, mlink);
1.6 schwarze 1228:
1.131 schwarze 1229: dbadd(mpage);
1.136 schwarze 1230: mlink = mpage->mlinks;
1.81 schwarze 1231:
1232: nextpage:
1.125 schwarze 1233: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1234: exitcode = (int)MANDOCLEVEL_SYSERR;
1.136 schwarze 1235: say(mlink->file, "&wait gunzip");
1.81 schwarze 1236: }
1.47 schwarze 1237: ohash_delete(&strings);
1.90 schwarze 1238: ohash_delete(&names);
1.47 schwarze 1239: mpage = ohash_next(&mpages, &pslot);
1240: }
1.64 schwarze 1241:
1242: if (0 == nodb)
1243: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1244: }
1245:
1246: static void
1247: names_check(void)
1248: {
1249: sqlite3_stmt *stmt;
1250: const char *name, *sec, *arch, *key;
1251: int irc;
1252:
1253: sqlite3_prepare_v2(db,
1254: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1255: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1256: "WHERE bits & ? AND NOT EXISTS ("
1257: "SELECT pageid FROM mlinks "
1.90 schwarze 1258: "WHERE mlinks.pageid == names.pageid "
1259: "AND mlinks.name == names.name"
1.87 schwarze 1260: ")"
1261: ") JOIN ("
1.99 schwarze 1262: "SELECT sec, arch, name, pageid FROM mlinks "
1263: "GROUP BY pageid"
1.87 schwarze 1264: ") USING (pageid);",
1265: -1, &stmt, NULL);
1266:
1.91 schwarze 1267: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1268: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1269:
1270: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1271: name = (const char *)sqlite3_column_text(stmt, 0);
1272: sec = (const char *)sqlite3_column_text(stmt, 1);
1273: arch = (const char *)sqlite3_column_text(stmt, 2);
1274: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1275: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1276: '\0' == *arch ? "" : "/",
1277: '\0' == *arch ? "" : arch, key);
1278: }
1279: sqlite3_finalize(stmt);
1.47 schwarze 1280: }
1.6 schwarze 1281:
1.47 schwarze 1282: static void
1.81 schwarze 1283: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1284: {
1285: FILE *stream;
1286: char *line, *p, *title;
1287: size_t len, plen, titlesz;
1.1 schwarze 1288:
1.81 schwarze 1289: stream = (-1 == fd) ?
1290: fopen(mpage->mlinks->file, "r") :
1291: fdopen(fd, "r");
1292: if (NULL == stream) {
1.114 doug 1293: if (-1 != fd)
1294: close(fd);
1.47 schwarze 1295: if (warnings)
1.80 schwarze 1296: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1297: return;
1298: }
1.1 schwarze 1299:
1.47 schwarze 1300: /* Skip to first blank line. */
1.1 schwarze 1301:
1.47 schwarze 1302: while (NULL != (line = fgetln(stream, &len)))
1303: if ('\n' == *line)
1304: break;
1.1 schwarze 1305:
1.47 schwarze 1306: /*
1307: * Assume the first line that is not indented
1308: * is the first section header. Skip to it.
1309: */
1.1 schwarze 1310:
1.47 schwarze 1311: while (NULL != (line = fgetln(stream, &len)))
1312: if ('\n' != *line && ' ' != *line)
1313: break;
1.98 schwarze 1314:
1.47 schwarze 1315: /*
1316: * Read up until the next section into a buffer.
1317: * Strip the leading and trailing newline from each read line,
1318: * appending a trailing space.
1319: * Ignore empty (whitespace-only) lines.
1320: */
1.28 schwarze 1321:
1.47 schwarze 1322: titlesz = 0;
1323: title = NULL;
1.38 schwarze 1324:
1.47 schwarze 1325: while (NULL != (line = fgetln(stream, &len))) {
1326: if (' ' != *line || '\n' != line[len - 1])
1327: break;
1328: while (len > 0 && isspace((unsigned char)*line)) {
1329: line++;
1330: len--;
1331: }
1332: if (1 == len)
1333: continue;
1334: title = mandoc_realloc(title, titlesz + len);
1335: memcpy(title + titlesz, line, len);
1336: titlesz += len;
1337: title[titlesz - 1] = ' ';
1338: }
1.28 schwarze 1339:
1.47 schwarze 1340: /*
1341: * If no page content can be found, or the input line
1342: * is already the next section header, or there is no
1343: * trailing newline, reuse the page title as the page
1344: * description.
1345: */
1.1 schwarze 1346:
1.47 schwarze 1347: if (NULL == title || '\0' == *title) {
1348: if (warnings)
1349: say(mpage->mlinks->file,
1350: "Cannot find NAME section");
1351: fclose(stream);
1352: free(title);
1353: return;
1354: }
1.24 schwarze 1355:
1.47 schwarze 1356: title = mandoc_realloc(title, titlesz + 1);
1357: title[titlesz] = '\0';
1.24 schwarze 1358:
1.47 schwarze 1359: /*
1360: * Skip to the first dash.
1361: * Use the remaining line as the description (no more than 70
1362: * bytes).
1363: */
1.28 schwarze 1364:
1.47 schwarze 1365: if (NULL != (p = strstr(title, "- "))) {
1366: for (p += 2; ' ' == *p || '\b' == *p; p++)
1367: /* Skip to next word. */ ;
1368: } else {
1369: if (warnings)
1370: say(mpage->mlinks->file,
1371: "No dash in title line");
1372: p = title;
1373: }
1.1 schwarze 1374:
1.47 schwarze 1375: plen = strlen(p);
1.1 schwarze 1376:
1.47 schwarze 1377: /* Strip backspace-encoding from line. */
1.1 schwarze 1378:
1.47 schwarze 1379: while (NULL != (line = memchr(p, '\b', plen))) {
1380: len = line - p;
1381: if (0 == len) {
1382: memmove(line, line + 1, plen--);
1383: continue;
1.98 schwarze 1384: }
1.47 schwarze 1385: memmove(line - 1, line + 1, plen - len);
1386: plen -= 2;
1387: }
1.1 schwarze 1388:
1.47 schwarze 1389: mpage->desc = mandoc_strdup(p);
1390: fclose(stream);
1391: free(title);
1392: }
1.16 schwarze 1393:
1.47 schwarze 1394: /*
1395: * Put a type/word pair into the word database for this particular file.
1396: */
1397: static void
1.69 schwarze 1398: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1399: {
1.69 schwarze 1400: char *cp;
1.37 schwarze 1401:
1.47 schwarze 1402: assert(NULL != value);
1.69 schwarze 1403: if (TYPE_arch == type)
1404: for (cp = value; *cp; cp++)
1405: if (isupper((unsigned char)*cp))
1406: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1407: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1408: }
1409:
1410: /*
1.47 schwarze 1411: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1412: */
1413: static void
1.47 schwarze 1414: putmdockey(const struct mpage *mpage,
1415: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1416: {
1.16 schwarze 1417:
1.47 schwarze 1418: for ( ; NULL != n; n = n->next) {
1419: if (NULL != n->child)
1420: putmdockey(mpage, n->child, m);
1421: if (MDOC_TEXT == n->type)
1422: putkey(mpage, n->string, m);
1423: }
1424: }
1.16 schwarze 1425:
1.47 schwarze 1426: static void
1.129 schwarze 1427: parse_man(struct mpage *mpage, const struct man_meta *meta,
1428: const struct man_node *n)
1.47 schwarze 1429: {
1430: const struct man_node *head, *body;
1.78 schwarze 1431: char *start, *title;
1.47 schwarze 1432: char byte;
1.78 schwarze 1433: size_t sz;
1.16 schwarze 1434:
1.47 schwarze 1435: if (NULL == n)
1436: return;
1.16 schwarze 1437:
1.47 schwarze 1438: /*
1439: * We're only searching for one thing: the first text child in
1440: * the BODY of a NAME section. Since we don't keep track of
1441: * sections in -man, run some hoops to find out whether we're in
1442: * the correct section or not.
1443: */
1.16 schwarze 1444:
1.47 schwarze 1445: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1446: body = n;
1447: assert(body->parent);
1448: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1449: 1 == head->nchild &&
1450: NULL != (head = (head->child)) &&
1451: MAN_TEXT == head->type &&
1452: 0 == strcmp(head->string, "NAME") &&
1453: NULL != body->child) {
1.2 schwarze 1454:
1.47 schwarze 1455: /*
1456: * Suck the entire NAME section into memory.
1457: * Yes, we might run away.
1458: * But too many manuals have big, spread-out
1459: * NAME sections over many lines.
1460: */
1.2 schwarze 1461:
1.78 schwarze 1462: title = NULL;
1463: man_deroff(&title, body);
1.47 schwarze 1464: if (NULL == title)
1465: return;
1.16 schwarze 1466:
1.98 schwarze 1467: /*
1.47 schwarze 1468: * Go through a special heuristic dance here.
1469: * Conventionally, one or more manual names are
1470: * comma-specified prior to a whitespace, then a
1471: * dash, then a description. Try to puzzle out
1472: * the name parts here.
1473: */
1.16 schwarze 1474:
1.78 schwarze 1475: start = title;
1.47 schwarze 1476: for ( ;; ) {
1477: sz = strcspn(start, " ,");
1478: if ('\0' == start[sz])
1479: break;
1.1 schwarze 1480:
1.47 schwarze 1481: byte = start[sz];
1482: start[sz] = '\0';
1.67 schwarze 1483:
1484: /*
1485: * Assume a stray trailing comma in the
1486: * name list if a name begins with a dash.
1487: */
1488:
1489: if ('-' == start[0] ||
1490: ('\\' == start[0] && '-' == start[1]))
1491: break;
1.1 schwarze 1492:
1.90 schwarze 1493: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1494: if ( ! (mpage->name_head_done ||
1495: strcasecmp(start, meta->title))) {
1496: putkey(mpage, start, NAME_HEAD);
1497: mpage->name_head_done = 1;
1498: }
1.1 schwarze 1499:
1.47 schwarze 1500: if (' ' == byte) {
1501: start += sz + 1;
1502: break;
1503: }
1.1 schwarze 1504:
1.47 schwarze 1505: assert(',' == byte);
1506: start += sz + 1;
1507: while (' ' == *start)
1508: start++;
1509: }
1.1 schwarze 1510:
1.78 schwarze 1511: if (start == title) {
1.90 schwarze 1512: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1513: if ( ! (mpage->name_head_done ||
1514: strcasecmp(start, meta->title))) {
1515: putkey(mpage, start, NAME_HEAD);
1516: mpage->name_head_done = 1;
1517: }
1.47 schwarze 1518: free(title);
1519: return;
1520: }
1.1 schwarze 1521:
1.47 schwarze 1522: while (isspace((unsigned char)*start))
1523: start++;
1.1 schwarze 1524:
1.47 schwarze 1525: if (0 == strncmp(start, "-", 1))
1526: start += 1;
1527: else if (0 == strncmp(start, "\\-\\-", 4))
1528: start += 4;
1529: else if (0 == strncmp(start, "\\-", 2))
1530: start += 2;
1531: else if (0 == strncmp(start, "\\(en", 4))
1532: start += 4;
1533: else if (0 == strncmp(start, "\\(em", 4))
1534: start += 4;
1.1 schwarze 1535:
1.47 schwarze 1536: while (' ' == *start)
1537: start++;
1.1 schwarze 1538:
1.47 schwarze 1539: mpage->desc = mandoc_strdup(start);
1540: free(title);
1541: return;
1542: }
1543: }
1.1 schwarze 1544:
1.47 schwarze 1545: for (n = n->child; n; n = n->next) {
1546: if (NULL != mpage->desc)
1547: break;
1.129 schwarze 1548: parse_man(mpage, meta, n);
1.1 schwarze 1549: }
1550: }
1551:
1552: static void
1.127 schwarze 1553: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1554: const struct mdoc_node *n)
1.1 schwarze 1555: {
1556:
1.47 schwarze 1557: assert(NULL != n);
1558: for (n = n->child; NULL != n; n = n->next) {
1559: switch (n->type) {
1.98 schwarze 1560: case MDOC_ELEM:
1.47 schwarze 1561: /* FALLTHROUGH */
1.98 schwarze 1562: case MDOC_BLOCK:
1.47 schwarze 1563: /* FALLTHROUGH */
1.98 schwarze 1564: case MDOC_HEAD:
1.47 schwarze 1565: /* FALLTHROUGH */
1.98 schwarze 1566: case MDOC_BODY:
1.47 schwarze 1567: /* FALLTHROUGH */
1.98 schwarze 1568: case MDOC_TAIL:
1.47 schwarze 1569: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1570: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1571: break;
1572: if (mdocs[n->tok].mask)
1573: putmdockey(mpage, n->child,
1574: mdocs[n->tok].mask);
1575: break;
1576: default:
1577: assert(MDOC_ROOT != n->type);
1578: continue;
1579: }
1580: if (NULL != n->child)
1.127 schwarze 1581: parse_mdoc(mpage, meta, n);
1.1 schwarze 1582: }
1583: }
1584:
1.19 schwarze 1585: static int
1.127 schwarze 1586: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1587: const struct mdoc_node *n)
1.1 schwarze 1588: {
1.131 schwarze 1589: char *start, *end;
1.1 schwarze 1590: size_t sz;
1.19 schwarze 1591:
1.47 schwarze 1592: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1593: NULL == (n = n->child) ||
1594: MDOC_TEXT != n->type)
1.19 schwarze 1595: return(0);
1.1 schwarze 1596:
1597: /*
1598: * Only consider those `Fd' macro fields that begin with an
1599: * "inclusion" token (versus, e.g., #define).
1600: */
1.47 schwarze 1601:
1.1 schwarze 1602: if (strcmp("#include", n->string))
1.19 schwarze 1603: return(0);
1.1 schwarze 1604:
1605: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1606: return(0);
1.1 schwarze 1607:
1608: /*
1609: * Strip away the enclosing angle brackets and make sure we're
1610: * not zero-length.
1611: */
1612:
1613: start = n->string;
1614: if ('<' == *start || '"' == *start)
1615: start++;
1616:
1617: if (0 == (sz = strlen(start)))
1.19 schwarze 1618: return(0);
1.1 schwarze 1619:
1620: end = &start[(int)sz - 1];
1621: if ('>' == *end || '"' == *end)
1622: end--;
1623:
1.47 schwarze 1624: if (end > start)
1625: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1626: return(0);
1.1 schwarze 1627: }
1628:
1.133 schwarze 1629: static void
1630: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1631: {
1.69 schwarze 1632: char *cp;
1.133 schwarze 1633: size_t sz;
1.1 schwarze 1634:
1.133 schwarze 1635: if (n->type != MDOC_TEXT)
1636: return;
1.19 schwarze 1637:
1.133 schwarze 1638: /* Skip function pointer punctuation. */
1.1 schwarze 1639:
1.133 schwarze 1640: cp = n->string;
1641: while (*cp == '(' || *cp == '*')
1.1 schwarze 1642: cp++;
1.133 schwarze 1643: sz = strcspn(cp, "()");
1.1 schwarze 1644:
1.133 schwarze 1645: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1646: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1647: putkeys(mpage, cp, sz, NAME_SYN);
1648: }
1.19 schwarze 1649:
1.133 schwarze 1650: static int
1651: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1652: const struct mdoc_node *n)
1653: {
1654:
1655: if (n->child == NULL)
1656: return(0);
1657:
1658: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1659:
1.133 schwarze 1660: for (n = n->child->next; n != NULL; n = n->next)
1661: if (n->type == MDOC_TEXT)
1.47 schwarze 1662: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1663:
1.128 schwarze 1664: return(0);
1665: }
1666:
1667: static int
1668: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1669: const struct mdoc_node *n)
1670: {
1.132 schwarze 1671:
1672: if (n->type != MDOC_HEAD)
1673: return(1);
1.128 schwarze 1674:
1.133 schwarze 1675: if (n->child != NULL)
1676: parse_mdoc_fname(mpage, n->child);
1677:
1.19 schwarze 1678: return(0);
1.1 schwarze 1679: }
1680:
1.19 schwarze 1681: static int
1.127 schwarze 1682: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1683: const struct mdoc_node *n)
1.1 schwarze 1684: {
1.47 schwarze 1685: char *cp;
1.1 schwarze 1686:
1687: if (NULL == (n = n->child))
1.19 schwarze 1688: return(0);
1.1 schwarze 1689:
1.47 schwarze 1690: if (NULL == n->next) {
1691: putkey(mpage, n->string, TYPE_Xr);
1692: return(0);
1693: }
1.1 schwarze 1694:
1.77 schwarze 1695: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1696: putkey(mpage, cp, TYPE_Xr);
1697: free(cp);
1698: return(0);
1.1 schwarze 1699: }
1700:
1.19 schwarze 1701: static int
1.127 schwarze 1702: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1703: const struct mdoc_node *n)
1.1 schwarze 1704: {
1705:
1.79 schwarze 1706: if (MDOC_BODY == n->type)
1707: mdoc_deroff(&mpage->desc, n);
1708: return(0);
1.1 schwarze 1709: }
1710:
1.19 schwarze 1711: static int
1.127 schwarze 1712: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1713: const struct mdoc_node *n)
1.1 schwarze 1714: {
1715:
1.86 schwarze 1716: if (SEC_NAME == n->sec)
1.90 schwarze 1717: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1718: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1719: if (n->child == NULL)
1720: putkey(mpage, meta->name, NAME_SYN);
1721: else
1722: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1723: }
1724: if ( ! (mpage->name_head_done ||
1725: n->child == NULL || n->child->string == NULL ||
1726: strcasecmp(n->child->string, meta->title))) {
1727: putkey(mpage, n->child->string, NAME_HEAD);
1728: mpage->name_head_done = 1;
1.127 schwarze 1729: }
1.86 schwarze 1730: return(0);
1.1 schwarze 1731: }
1732:
1.19 schwarze 1733: static int
1.127 schwarze 1734: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1735: const struct mdoc_node *n)
1.1 schwarze 1736: {
1737:
1.19 schwarze 1738: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1739: }
1740:
1.47 schwarze 1741: static int
1.127 schwarze 1742: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1743: const struct mdoc_node *n)
1.1 schwarze 1744: {
1745:
1.47 schwarze 1746: return(MDOC_HEAD == n->type);
1747: }
1.1 schwarze 1748:
1.47 schwarze 1749: static int
1.127 schwarze 1750: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1751: const struct mdoc_node *n)
1.47 schwarze 1752: {
1.1 schwarze 1753:
1.47 schwarze 1754: return(MDOC_BODY == n->type);
1.1 schwarze 1755: }
1756:
1.47 schwarze 1757: /*
1758: * Add a string to the hash table for the current manual.
1759: * Each string has a bitmask telling which macros it belongs to.
1760: * When we finish the manual, we'll dump the table.
1761: */
1.1 schwarze 1762: static void
1.131 schwarze 1763: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1764: {
1.90 schwarze 1765: struct ohash *htab;
1.47 schwarze 1766: struct str *s;
1.68 schwarze 1767: const char *end;
1.47 schwarze 1768: unsigned int slot;
1.131 schwarze 1769: int i, mustfree;
1.1 schwarze 1770:
1.47 schwarze 1771: if (0 == sz)
1772: return;
1.68 schwarze 1773:
1.131 schwarze 1774: mustfree = render_string(&cp, &sz);
1775:
1.90 schwarze 1776: if (TYPE_Nm & v) {
1777: htab = &names;
1778: v &= name_mask;
1.124 schwarze 1779: if (v & NAME_FIRST)
1780: name_mask &= ~NAME_FIRST;
1.90 schwarze 1781: if (debug > 1)
1782: say(mpage->mlinks->file,
1.130 schwarze 1783: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1784: } else {
1785: htab = &strings;
1786: if (debug > 1)
1787: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1788: if ((uint64_t)1 << i & v)
1.90 schwarze 1789: say(mpage->mlinks->file,
1790: "Adding key %s=%*s",
1791: mansearch_keynames[i], sz, cp);
1.68 schwarze 1792: }
1.47 schwarze 1793:
1794: end = cp + sz;
1.90 schwarze 1795: slot = ohash_qlookupi(htab, cp, &end);
1796: s = ohash_find(htab, slot);
1.1 schwarze 1797:
1.47 schwarze 1798: if (NULL != s && mpage == s->mpage) {
1799: s->mask |= v;
1.1 schwarze 1800: return;
1.47 schwarze 1801: } else if (NULL == s) {
1.101 schwarze 1802: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1803: memcpy(s->key, cp, sz);
1.90 schwarze 1804: ohash_insert(htab, slot, s);
1.47 schwarze 1805: }
1806: s->mpage = mpage;
1807: s->mask = v;
1.131 schwarze 1808:
1809: if (mustfree)
1810: free(cp);
1.1 schwarze 1811: }
1812:
1813: /*
1.47 schwarze 1814: * Take a Unicode codepoint and produce its UTF-8 encoding.
1815: * This isn't the best way to do this, but it works.
1816: * The magic numbers are from the UTF-8 packaging.
1817: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1818: */
1.47 schwarze 1819: static size_t
1820: utf8(unsigned int cp, char out[7])
1.1 schwarze 1821: {
1.47 schwarze 1822: size_t rc;
1.1 schwarze 1823:
1.47 schwarze 1824: rc = 0;
1825: if (cp <= 0x0000007F) {
1826: rc = 1;
1827: out[0] = (char)cp;
1828: } else if (cp <= 0x000007FF) {
1829: rc = 2;
1830: out[0] = (cp >> 6 & 31) | 192;
1831: out[1] = (cp & 63) | 128;
1832: } else if (cp <= 0x0000FFFF) {
1833: rc = 3;
1834: out[0] = (cp >> 12 & 15) | 224;
1835: out[1] = (cp >> 6 & 63) | 128;
1836: out[2] = (cp & 63) | 128;
1837: } else if (cp <= 0x001FFFFF) {
1838: rc = 4;
1839: out[0] = (cp >> 18 & 7) | 240;
1840: out[1] = (cp >> 12 & 63) | 128;
1841: out[2] = (cp >> 6 & 63) | 128;
1842: out[3] = (cp & 63) | 128;
1843: } else if (cp <= 0x03FFFFFF) {
1844: rc = 5;
1845: out[0] = (cp >> 24 & 3) | 248;
1846: out[1] = (cp >> 18 & 63) | 128;
1847: out[2] = (cp >> 12 & 63) | 128;
1848: out[3] = (cp >> 6 & 63) | 128;
1849: out[4] = (cp & 63) | 128;
1850: } else if (cp <= 0x7FFFFFFF) {
1851: rc = 6;
1852: out[0] = (cp >> 30 & 1) | 252;
1853: out[1] = (cp >> 24 & 63) | 128;
1854: out[2] = (cp >> 18 & 63) | 128;
1855: out[3] = (cp >> 12 & 63) | 128;
1856: out[4] = (cp >> 6 & 63) | 128;
1857: out[5] = (cp & 63) | 128;
1858: } else
1859: return(0);
1.19 schwarze 1860:
1.47 schwarze 1861: out[rc] = '\0';
1862: return(rc);
1.1 schwarze 1863: }
1864:
1.47 schwarze 1865: /*
1.131 schwarze 1866: * If the string contains escape sequences,
1867: * replace it with an allocated rendering and return 1,
1868: * such that the caller can free it after use.
1869: * Otherwise, do nothing and return 0.
1.47 schwarze 1870: */
1.131 schwarze 1871: static int
1872: render_string(char **public, size_t *psz)
1.1 schwarze 1873: {
1.131 schwarze 1874: const char *src, *scp, *addcp, *seq;
1875: char *dst;
1876: size_t ssz, dsz, addsz;
1.71 schwarze 1877: char utfbuf[7], res[6];
1.131 schwarze 1878: int seqlen, unicode;
1.47 schwarze 1879:
1880: res[0] = '\\';
1881: res[1] = '\t';
1882: res[2] = ASCII_NBRSP;
1883: res[3] = ASCII_HYPH;
1.71 schwarze 1884: res[4] = ASCII_BREAK;
1885: res[5] = '\0';
1.1 schwarze 1886:
1.131 schwarze 1887: src = scp = *public;
1888: ssz = *psz;
1889: dst = NULL;
1890: dsz = 0;
1891:
1892: while (scp < src + *psz) {
1893:
1894: /* Leave normal characters unchanged. */
1895:
1896: if (strchr(res, *scp) == NULL) {
1897: if (dst != NULL)
1898: dst[dsz++] = *scp;
1899: scp++;
1900: continue;
1901: }
1.1 schwarze 1902:
1.47 schwarze 1903: /*
1.131 schwarze 1904: * Found something that requires replacing,
1905: * make sure we have a destination buffer.
1.47 schwarze 1906: */
1.131 schwarze 1907:
1908: if (dst == NULL) {
1909: dst = mandoc_malloc(ssz + 1);
1910: dsz = scp - src;
1911: memcpy(dst, src, dsz);
1.47 schwarze 1912: }
1.39 schwarze 1913:
1.131 schwarze 1914: /* Handle single-char special characters. */
1915:
1916: switch (*scp) {
1917: case '\\':
1918: break;
1.98 schwarze 1919: case '\t':
1.71 schwarze 1920: /* FALLTHROUGH */
1.98 schwarze 1921: case ASCII_NBRSP:
1.131 schwarze 1922: dst[dsz++] = ' ';
1923: scp++;
1924: continue;
1925: case ASCII_HYPH:
1926: dst[dsz++] = '-';
1.71 schwarze 1927: /* FALLTHROUGH */
1.98 schwarze 1928: case ASCII_BREAK:
1.131 schwarze 1929: scp++;
1.47 schwarze 1930: continue;
1.71 schwarze 1931: default:
1.131 schwarze 1932: abort();
1.71 schwarze 1933: }
1.39 schwarze 1934:
1.47 schwarze 1935: /*
1.131 schwarze 1936: * Found an escape sequence.
1937: * Read past the slash, then parse it.
1938: * Ignore everything except characters.
1.47 schwarze 1939: */
1.52 schwarze 1940:
1.131 schwarze 1941: scp++;
1942: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1943: continue;
1.39 schwarze 1944:
1.47 schwarze 1945: /*
1.52 schwarze 1946: * Render the special character
1947: * as either UTF-8 or ASCII.
1.47 schwarze 1948: */
1.52 schwarze 1949:
1950: if (write_utf8) {
1.131 schwarze 1951: unicode = mchars_spec2cp(mchars, seq, seqlen);
1952: if (unicode <= 0)
1.52 schwarze 1953: continue;
1.131 schwarze 1954: addsz = utf8(unicode, utfbuf);
1955: if (addsz == 0)
1.52 schwarze 1956: continue;
1.131 schwarze 1957: addcp = utfbuf;
1.52 schwarze 1958: } else {
1.131 schwarze 1959: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1960: if (addcp == NULL)
1.52 schwarze 1961: continue;
1.131 schwarze 1962: if (*addcp == ASCII_NBRSP) {
1963: addcp = " ";
1964: addsz = 1;
1.52 schwarze 1965: }
1966: }
1.1 schwarze 1967:
1.47 schwarze 1968: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1969:
1.131 schwarze 1970: ssz += addsz;
1971: dst = mandoc_realloc(dst, ssz + 1);
1972: memcpy(dst + dsz, addcp, addsz);
1973: dsz += addsz;
1974: }
1975: if (dst != NULL) {
1976: *public = dst;
1977: *psz = dsz;
1978: }
1979:
1980: /* Trim trailing whitespace and NUL-terminate. */
1981:
1982: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1983: --*psz;
1984: if (dst != NULL) {
1985: (*public)[*psz] = '\0';
1986: return(1);
1987: } else
1988: return(0);
1.1 schwarze 1989: }
1990:
1.75 schwarze 1991: static void
1992: dbadd_mlink(const struct mlink *mlink)
1993: {
1994: size_t i;
1995:
1996: i = 1;
1997: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1998: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1999: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 2000: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 2001: SQL_STEP(stmts[STMT_INSERT_LINK]);
2002: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2003: }
2004:
2005: static void
2006: dbadd_mlink_name(const struct mlink *mlink)
2007: {
1.130 schwarze 2008: uint64_t bits;
1.124 schwarze 2009: size_t i;
2010:
2011: dbadd_mlink(mlink);
1.115 schwarze 2012:
2013: i = 1;
1.130 schwarze 2014: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2015: bits = NAME_FILE & NAME_MASK;
2016: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2017: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2018: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2019: }
2020:
2021: i = 1;
2022: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2023: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2024: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2025: SQL_STEP(stmts[STMT_INSERT_NAME]);
2026: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2027: }
2028:
1.11 schwarze 2029: /*
1.47 schwarze 2030: * Flush the current page's terms (and their bits) into the database.
2031: * Wrap the entire set of additions in a transaction to make sqlite be a
2032: * little faster.
1.53 schwarze 2033: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2034: */
2035: static void
1.131 schwarze 2036: dbadd(struct mpage *mpage)
1.11 schwarze 2037: {
1.47 schwarze 2038: struct mlink *mlink;
2039: struct str *key;
1.131 schwarze 2040: char *cp;
1.47 schwarze 2041: size_t i;
2042: unsigned int slot;
1.131 schwarze 2043: int mustfree;
1.47 schwarze 2044:
1.85 schwarze 2045: mlink = mpage->mlinks;
1.11 schwarze 2046:
1.85 schwarze 2047: if (nodb) {
1.104 schwarze 2048: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2049: key = ohash_next(&names, &slot))
1.104 schwarze 2050: free(key);
2051: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2052: key = ohash_next(&strings, &slot))
1.104 schwarze 2053: free(key);
1.102 schwarze 2054: if (0 == debug)
2055: return;
1.85 schwarze 2056: while (NULL != mlink) {
2057: fputs(mlink->name, stdout);
2058: if (NULL == mlink->next ||
2059: strcmp(mlink->dsec, mlink->next->dsec) ||
2060: strcmp(mlink->fsec, mlink->next->fsec) ||
2061: strcmp(mlink->arch, mlink->next->arch)) {
2062: putchar('(');
2063: if ('\0' == *mlink->dsec)
2064: fputs(mlink->fsec, stdout);
2065: else
2066: fputs(mlink->dsec, stdout);
2067: if ('\0' != *mlink->arch)
2068: printf("/%s", mlink->arch);
2069: putchar(')');
2070: }
2071: mlink = mlink->next;
2072: if (NULL != mlink)
2073: fputs(", ", stdout);
2074: }
1.89 schwarze 2075: printf(" - %s\n", mpage->desc);
1.11 schwarze 2076: return;
1.85 schwarze 2077: }
2078:
2079: if (debug)
2080: say(mlink->file, "Adding to database");
1.47 schwarze 2081:
1.131 schwarze 2082: cp = mpage->desc;
2083: i = strlen(cp);
2084: mustfree = render_string(&cp, &i);
1.47 schwarze 2085: i = 1;
1.131 schwarze 2086: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2087: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2088: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2089: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2090: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2091: if (mustfree)
2092: free(cp);
1.47 schwarze 2093:
1.85 schwarze 2094: while (NULL != mlink) {
1.75 schwarze 2095: dbadd_mlink(mlink);
1.85 schwarze 2096: mlink = mlink->next;
2097: }
1.91 schwarze 2098: mlink = mpage->mlinks;
1.47 schwarze 2099:
1.90 schwarze 2100: for (key = ohash_first(&names, &slot); NULL != key;
2101: key = ohash_next(&names, &slot)) {
2102: assert(key->mpage == mpage);
2103: i = 1;
2104: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2105: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2106: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2107: SQL_STEP(stmts[STMT_INSERT_NAME]);
2108: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2109: free(key);
2110: }
1.47 schwarze 2111: for (key = ohash_first(&strings, &slot); NULL != key;
2112: key = ohash_next(&strings, &slot)) {
2113: assert(key->mpage == mpage);
2114: i = 1;
2115: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2116: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2117: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2118: SQL_STEP(stmts[STMT_INSERT_KEY]);
2119: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2120: free(key);
1.33 schwarze 2121: }
1.47 schwarze 2122: }
1.41 deraadt 2123:
1.47 schwarze 2124: static void
2125: dbprune(void)
2126: {
2127: struct mpage *mpage;
2128: struct mlink *mlink;
2129: size_t i;
2130: unsigned int slot;
1.11 schwarze 2131:
1.63 schwarze 2132: if (0 == nodb)
2133: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2134:
1.63 schwarze 2135: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2136: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2137: mlink = mpage->mlinks;
1.82 schwarze 2138: if (debug)
1.63 schwarze 2139: say(mlink->file, "Deleting from database");
2140: if (nodb)
2141: continue;
2142: for ( ; NULL != mlink; mlink = mlink->next) {
2143: i = 1;
2144: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2145: i, mlink->dsec);
2146: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2147: i, mlink->arch);
2148: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2149: i, mlink->name);
2150: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2151: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2152: }
1.11 schwarze 2153: }
1.63 schwarze 2154:
2155: if (0 == nodb)
2156: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2157: }
1.22 schwarze 2158:
1.47 schwarze 2159: /*
2160: * Close an existing database and its prepared statements.
2161: * If "real" is not set, rename the temporary file into the real one.
2162: */
2163: static void
2164: dbclose(int real)
2165: {
2166: size_t i;
1.72 schwarze 2167: int status;
2168: pid_t child;
1.11 schwarze 2169:
1.47 schwarze 2170: if (nodb)
2171: return;
1.11 schwarze 2172:
1.47 schwarze 2173: for (i = 0; i < STMT__MAX; i++) {
2174: sqlite3_finalize(stmts[i]);
2175: stmts[i] = NULL;
1.28 schwarze 2176: }
1.22 schwarze 2177:
1.47 schwarze 2178: sqlite3_close(db);
2179: db = NULL;
1.11 schwarze 2180:
1.47 schwarze 2181: if (real)
2182: return;
1.22 schwarze 2183:
1.72 schwarze 2184: if ('\0' == *tempfilename) {
2185: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2186: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2187: say(MANDOC_DB, "&rename");
1.72 schwarze 2188: }
2189: return;
2190: }
2191:
2192: switch (child = fork()) {
1.98 schwarze 2193: case -1:
1.72 schwarze 2194: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2195: say("", "&fork cmp");
1.72 schwarze 2196: return;
1.98 schwarze 2197: case 0:
1.72 schwarze 2198: execlp("cmp", "cmp", "-s",
2199: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2200: say("", "&exec cmp");
1.72 schwarze 2201: exit(0);
2202: default:
2203: break;
2204: }
2205: if (-1 == waitpid(child, &status, 0)) {
2206: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2207: say("", "&wait cmp");
1.72 schwarze 2208: } else if (WIFSIGNALED(status)) {
2209: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2210: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2211: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2213: say(MANDOC_DB,
2214: "Data changed, but cannot replace database");
2215: }
2216:
2217: *strrchr(tempfilename, '/') = '\0';
2218: switch (child = fork()) {
1.98 schwarze 2219: case -1:
1.72 schwarze 2220: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2221: say("", "&fork rm");
1.72 schwarze 2222: return;
1.98 schwarze 2223: case 0:
1.72 schwarze 2224: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2225: say("", "&exec rm");
1.72 schwarze 2226: exit((int)MANDOCLEVEL_SYSERR);
2227: default:
2228: break;
2229: }
2230: if (-1 == waitpid(child, &status, 0)) {
2231: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2232: say("", "&wait rm");
1.72 schwarze 2233: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2234: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2235: say("", "%s: Cannot remove temporary directory",
2236: tempfilename);
1.22 schwarze 2237: }
1.11 schwarze 2238: }
2239:
1.47 schwarze 2240: /*
2241: * This is straightforward stuff.
2242: * Open a database connection to a "temporary" database, then open a set
2243: * of prepared statements we'll use over and over again.
2244: * If "real" is set, we use the existing database; if not, we truncate a
2245: * temporary one.
2246: * Must be matched by dbclose().
2247: */
2248: static int
2249: dbopen(int real)
1.2 schwarze 2250: {
1.72 schwarze 2251: const char *sql;
1.47 schwarze 2252: int rc, ofl;
1.6 schwarze 2253:
1.98 schwarze 2254: if (nodb)
1.47 schwarze 2255: return(1);
1.6 schwarze 2256:
1.72 schwarze 2257: *tempfilename = '\0';
1.47 schwarze 2258: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2259:
2260: if (real) {
2261: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2262: if (SQLITE_OK != rc) {
1.47 schwarze 2263: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2264: if (SQLITE_CANTOPEN != rc)
2265: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2266: return(0);
1.28 schwarze 2267: }
1.72 schwarze 2268: goto prepare_statements;
2269: }
2270:
2271: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2272:
1.72 schwarze 2273: remove(MANDOC_DB "~");
2274: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2275: if (SQLITE_OK == rc)
1.72 schwarze 2276: goto create_tables;
1.73 schwarze 2277: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2278: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2279: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2280: return(0);
2281: }
1.6 schwarze 2282:
1.100 schwarze 2283: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2284: sizeof(tempfilename));
1.72 schwarze 2285: if (NULL == mkdtemp(tempfilename)) {
2286: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2287: say("", "&%s", tempfilename);
1.72 schwarze 2288: return(0);
2289: }
1.100 schwarze 2290: (void)strlcat(tempfilename, "/" MANDOC_DB,
2291: sizeof(tempfilename));
1.72 schwarze 2292: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2293: if (SQLITE_OK != rc) {
1.47 schwarze 2294: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2295: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2296: return(0);
1.2 schwarze 2297: }
2298:
1.72 schwarze 2299: create_tables:
1.47 schwarze 2300: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2301: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2302: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2303: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2304: ");\n"
2305: "\n"
2306: "CREATE TABLE \"mlinks\" (\n"
2307: " \"sec\" TEXT NOT NULL,\n"
2308: " \"arch\" TEXT NOT NULL,\n"
2309: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2310: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2311: "ON DELETE CASCADE\n"
1.47 schwarze 2312: ");\n"
1.93 schwarze 2313: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2314: "\n"
1.90 schwarze 2315: "CREATE TABLE \"names\" (\n"
2316: " \"bits\" INTEGER NOT NULL,\n"
2317: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2318: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2319: "ON DELETE CASCADE,\n"
2320: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2321: ");\n"
2322: "\n"
1.47 schwarze 2323: "CREATE TABLE \"keys\" (\n"
2324: " \"bits\" INTEGER NOT NULL,\n"
2325: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2326: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2327: "ON DELETE CASCADE\n"
1.93 schwarze 2328: ");\n"
2329: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2330:
2331: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2332: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2333: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2334: sqlite3_close(db);
1.47 schwarze 2335: return(0);
1.2 schwarze 2336: }
2337:
1.47 schwarze 2338: prepare_statements:
1.103 schwarze 2339: if (SQLITE_OK != sqlite3_exec(db,
2340: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2341: exitcode = (int)MANDOCLEVEL_SYSERR;
2342: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2343: sqlite3_errmsg(db));
2344: sqlite3_close(db);
2345: return(0);
2346: }
2347:
1.94 schwarze 2348: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2349: "(SELECT pageid FROM mlinks WHERE "
2350: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2351: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2352: sql = "INSERT INTO mpages "
1.89 schwarze 2353: "(desc,form) VALUES (?,?)";
1.47 schwarze 2354: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2355: sql = "INSERT INTO mlinks "
1.61 schwarze 2356: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2357: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2358: sql = "SELECT bits FROM names where pageid = ?";
2359: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2360: sql = "INSERT INTO names "
2361: "(bits,name,pageid) VALUES (?,?,?)";
2362: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2363: sql = "INSERT INTO keys "
2364: "(bits,key,pageid) VALUES (?,?,?)";
2365: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2366:
1.47 schwarze 2367: /*
2368: * When opening a new database, we can turn off
2369: * synchronous mode for much better performance.
2370: */
1.6 schwarze 2371:
1.103 schwarze 2372: if (real && SQLITE_OK != sqlite3_exec(db,
2373: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2374: exitcode = (int)MANDOCLEVEL_SYSERR;
2375: say(MANDOC_DB, "PRAGMA synchronous: %s",
2376: sqlite3_errmsg(db));
2377: sqlite3_close(db);
2378: return(0);
2379: }
1.11 schwarze 2380:
1.47 schwarze 2381: return(1);
2382: }
1.6 schwarze 2383:
1.47 schwarze 2384: static void *
1.106 espie 2385: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2386: {
1.6 schwarze 2387:
1.106 espie 2388: return(mandoc_calloc(n, sz));
1.47 schwarze 2389: }
1.2 schwarze 2390:
1.47 schwarze 2391: static void *
2392: hash_alloc(size_t sz, void *arg)
2393: {
1.28 schwarze 2394:
1.47 schwarze 2395: return(mandoc_malloc(sz));
2396: }
1.6 schwarze 2397:
1.47 schwarze 2398: static void
1.106 espie 2399: hash_free(void *p, void *arg)
1.47 schwarze 2400: {
1.26 schwarze 2401:
1.47 schwarze 2402: free(p);
2403: }
1.6 schwarze 2404:
1.47 schwarze 2405: static int
1.120 schwarze 2406: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2407: {
2408: static char startdir[PATH_MAX];
1.110 schwarze 2409: static int getcwd_status; /* 1 = ok, 2 = failure */
2410: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2411: char *cp;
1.6 schwarze 2412:
1.47 schwarze 2413: /*
1.110 schwarze 2414: * Remember the original working directory, if possible.
2415: * This will be needed if the second or a later directory
2416: * on the command line is given as a relative path.
2417: * Do not error out if the current directory is not
2418: * searchable: Maybe it won't be needed after all.
2419: */
2420: if (0 == getcwd_status) {
2421: if (NULL == getcwd(startdir, sizeof(startdir))) {
2422: getcwd_status = 2;
2423: (void)strlcpy(startdir, strerror(errno),
2424: sizeof(startdir));
2425: } else
2426: getcwd_status = 1;
2427: }
2428:
2429: /*
2430: * We are leaving the old base directory.
2431: * Do not use it any longer, not even for messages.
2432: */
2433: *basedir = '\0';
2434:
2435: /*
2436: * If and only if the directory was changed earlier and
2437: * the next directory to process is given as a relative path,
2438: * first go back, or bail out if that is impossible.
1.47 schwarze 2439: */
1.110 schwarze 2440: if (chdir_status && '/' != *targetdir) {
2441: if (2 == getcwd_status) {
1.47 schwarze 2442: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2443: say("", "getcwd: %s", startdir);
1.47 schwarze 2444: return(0);
2445: }
1.110 schwarze 2446: if (-1 == chdir(startdir)) {
1.47 schwarze 2447: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2448: say("", "&chdir %s", startdir);
1.47 schwarze 2449: return(0);
1.2 schwarze 2450: }
2451: }
1.110 schwarze 2452:
2453: /*
2454: * Always resolve basedir to the canonicalized absolute
2455: * pathname and append a trailing slash, such that
2456: * we can reliably check whether files are inside.
2457: */
1.47 schwarze 2458: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2459: if (report_baddir || errno != ENOENT) {
2460: exitcode = (int)MANDOCLEVEL_BADARG;
2461: say("", "&%s: realpath", targetdir);
2462: }
1.47 schwarze 2463: return(0);
2464: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2465: if (report_baddir || errno != ENOENT) {
2466: exitcode = (int)MANDOCLEVEL_BADARG;
2467: say("", "&chdir");
2468: }
1.47 schwarze 2469: return(0);
1.109 schwarze 2470: }
1.110 schwarze 2471: chdir_status = 1;
1.109 schwarze 2472: cp = strchr(basedir, '\0');
2473: if ('/' != cp[-1]) {
2474: if (cp - basedir >= PATH_MAX - 1) {
2475: exitcode = (int)MANDOCLEVEL_SYSERR;
2476: say("", "Filename too long");
2477: return(0);
2478: }
2479: *cp++ = '/';
2480: *cp = '\0';
1.47 schwarze 2481: }
2482: return(1);
1.2 schwarze 2483: }
2484:
2485: static void
1.47 schwarze 2486: say(const char *file, const char *format, ...)
1.2 schwarze 2487: {
1.47 schwarze 2488: va_list ap;
1.80 schwarze 2489: int use_errno;
1.2 schwarze 2490:
1.47 schwarze 2491: if ('\0' != *basedir)
2492: fprintf(stderr, "%s", basedir);
2493: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2494: fputc('/', stderr);
1.47 schwarze 2495: if ('\0' != *file)
2496: fprintf(stderr, "%s", file);
1.31 schwarze 2497:
1.80 schwarze 2498: use_errno = 1;
2499: if (NULL != format) {
2500: switch (*format) {
1.98 schwarze 2501: case '&':
1.80 schwarze 2502: format++;
2503: break;
1.98 schwarze 2504: case '\0':
1.80 schwarze 2505: format = NULL;
2506: break;
2507: default:
2508: use_errno = 0;
2509: break;
2510: }
2511: }
2512: if (NULL != format) {
2513: if ('\0' != *basedir || '\0' != *file)
2514: fputs(": ", stderr);
2515: va_start(ap, format);
2516: vfprintf(stderr, format, ap);
2517: va_end(ap);
2518: }
2519: if (use_errno) {
2520: if ('\0' != *basedir || '\0' != *file || NULL != format)
2521: fputs(": ", stderr);
1.47 schwarze 2522: perror(NULL);
1.80 schwarze 2523: } else
2524: fputc('\n', stderr);
1.1 schwarze 2525: }