Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.145
1.145 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.144 2015/04/18 16:04:40 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:
1.141 schwarze 39: #include "mandoc_aux.h"
40: #include "mandoc.h"
41: #include "roff.h"
1.47 schwarze 42: #include "mdoc.h"
1.1 schwarze 43: #include "man.h"
1.140 schwarze 44: #include "manconf.h"
1.47 schwarze 45: #include "mansearch.h"
1.1 schwarze 46:
1.68 schwarze 47: extern int mansearch_keymax;
48: extern const char *const mansearch_keynames[];
49:
1.47 schwarze 50: #define SQL_EXEC(_v) \
51: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 52: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 53: #define SQL_BIND_TEXT(_s, _i, _v) \
54: if (SQLITE_OK != sqlite3_bind_text \
55: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 56: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 57: #define SQL_BIND_INT(_s, _i, _v) \
58: if (SQLITE_OK != sqlite3_bind_int \
59: ((_s), (_i)++, (_v))) \
1.91 schwarze 60: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 61: #define SQL_BIND_INT64(_s, _i, _v) \
62: if (SQLITE_OK != sqlite3_bind_int64 \
63: ((_s), (_i)++, (_v))) \
1.91 schwarze 64: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 65: #define SQL_STEP(_s) \
66: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 67: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 68:
1.47 schwarze 69: enum op {
70: OP_DEFAULT = 0, /* new dbs from dir list or default config */
71: OP_CONFFILE, /* new databases from custom config file */
72: OP_UPDATE, /* delete/add entries in existing database */
73: OP_DELETE, /* delete entries from existing database */
74: OP_TEST /* change no databases, report potential problems */
75: };
1.11 schwarze 76:
1.47 schwarze 77: struct str {
78: const struct mpage *mpage; /* if set, the owning parse */
79: uint64_t mask; /* bitmask in sequence */
1.131 schwarze 80: char key[]; /* rendered text */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct inodev {
84: ino_t st_ino;
85: dev_t st_dev;
86: };
1.28 schwarze 87:
1.47 schwarze 88: struct mpage {
89: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 90: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 91: char *sec; /* section from file content */
92: char *arch; /* architecture from file content */
93: char *title; /* title from file content */
94: char *desc; /* description from file content */
95: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 96: int form; /* format from file content */
1.129 schwarze 97: int name_head_done;
1.28 schwarze 98: };
99:
1.47 schwarze 100: struct mlink {
101: char file[PATH_MAX]; /* filename rel. to manpath */
102: char *dsec; /* section from directory */
103: char *arch; /* architecture from directory */
104: char *name; /* name from file name (not empty) */
105: char *fsec; /* section from file name suffix */
106: struct mlink *next; /* singly linked list */
1.75 schwarze 107: struct mpage *mpage; /* parent */
1.116 schwarze 108: int dform; /* format from directory */
109: int fform; /* format from file name suffix */
1.81 schwarze 110: int gzip; /* filename has a .gz suffix */
1.2 schwarze 111: };
112:
1.47 schwarze 113: enum stmt {
114: STMT_DELETE_PAGE = 0, /* delete mpage */
115: STMT_INSERT_PAGE, /* insert mpage */
116: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 117: STMT_INSERT_NAME, /* insert name */
1.130 schwarze 118: STMT_SELECT_NAME, /* retrieve existing name flags */
1.47 schwarze 119: STMT_INSERT_KEY, /* insert parsed key */
120: STMT__MAX
1.1 schwarze 121: };
122:
1.143 schwarze 123: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.142 schwarze 124: const struct roff_node *);
1.1 schwarze 125:
1.19 schwarze 126: struct mdoc_handler {
1.47 schwarze 127: mdoc_fp fp; /* optional handler */
128: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 129: };
130:
1.47 schwarze 131: static void dbclose(int);
1.131 schwarze 132: static void dbadd(struct mpage *);
1.75 schwarze 133: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 134: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 135: static int dbopen(int);
136: static void dbprune(void);
137: static void filescan(const char *);
138: static void *hash_alloc(size_t, void *);
1.106 espie 139: static void hash_free(void *, void *);
140: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 141: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 142: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 143: static void mlink_free(struct mlink *);
144: static void mlinks_undupe(struct mpage *);
145: static void mpages_free(void);
1.131 schwarze 146: static void mpages_merge(struct mparse *);
1.87 schwarze 147: static void names_check(void);
1.81 schwarze 148: static void parse_cat(struct mpage *, int);
1.143 schwarze 149: static void parse_man(struct mpage *, const struct roff_meta *,
1.142 schwarze 150: const struct roff_node *);
1.143 schwarze 151: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.142 schwarze 152: const struct roff_node *);
1.143 schwarze 153: static int parse_mdoc_body(struct mpage *, const struct roff_meta *,
1.142 schwarze 154: const struct roff_node *);
1.143 schwarze 155: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.142 schwarze 156: const struct roff_node *);
1.143 schwarze 157: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.142 schwarze 158: const struct roff_node *);
159: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.143 schwarze 160: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.142 schwarze 161: const struct roff_node *);
1.143 schwarze 162: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.142 schwarze 163: const struct roff_node *);
1.143 schwarze 164: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.142 schwarze 165: const struct roff_node *);
1.143 schwarze 166: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.142 schwarze 167: const struct roff_node *);
1.143 schwarze 168: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.142 schwarze 169: const struct roff_node *);
1.143 schwarze 170: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.142 schwarze 171: const struct roff_node *);
1.69 schwarze 172: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 173: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 174: static void putmdockey(const struct mpage *,
1.142 schwarze 175: const struct roff_node *, uint64_t);
1.131 schwarze 176: static int render_string(char **, size_t *);
1.47 schwarze 177: static void say(const char *, const char *, ...);
1.120 schwarze 178: static int set_basedir(const char *, int);
1.47 schwarze 179: static int treescan(void);
180: static size_t utf8(unsigned int, char [7]);
181:
1.72 schwarze 182: static char tempfilename[32];
1.47 schwarze 183: static char *progname;
1.59 schwarze 184: static int nodb; /* no database changes */
1.73 schwarze 185: static int mparse_options; /* abort the parse early */
1.98 schwarze 186: static int use_all; /* use all found files */
187: static int debug; /* print what we're doing */
188: static int warnings; /* warn about crap */
1.52 schwarze 189: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 190: static int exitcode; /* to be returned by main */
1.98 schwarze 191: static enum op op; /* operational mode */
1.47 schwarze 192: static char basedir[PATH_MAX]; /* current base directory */
1.131 schwarze 193: static struct mchars *mchars; /* table of named characters */
1.47 schwarze 194: static struct ohash mpages; /* table of distinct manual pages */
195: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 196: static struct ohash names; /* table of all names */
1.47 schwarze 197: static struct ohash strings; /* table of all strings */
198: static sqlite3 *db = NULL; /* current database */
199: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 200: static uint64_t name_mask;
1.47 schwarze 201:
1.19 schwarze 202: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 203: { NULL, 0 }, /* Ap */
204: { NULL, 0 }, /* Dd */
205: { NULL, 0 }, /* Dt */
206: { NULL, 0 }, /* Os */
207: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
208: { parse_mdoc_head, TYPE_Ss }, /* Ss */
209: { NULL, 0 }, /* Pp */
210: { NULL, 0 }, /* D1 */
211: { NULL, 0 }, /* Dl */
212: { NULL, 0 }, /* Bd */
213: { NULL, 0 }, /* Ed */
214: { NULL, 0 }, /* Bl */
215: { NULL, 0 }, /* El */
216: { NULL, 0 }, /* It */
217: { NULL, 0 }, /* Ad */
218: { NULL, TYPE_An }, /* An */
219: { NULL, TYPE_Ar }, /* Ar */
220: { NULL, TYPE_Cd }, /* Cd */
221: { NULL, TYPE_Cm }, /* Cm */
222: { NULL, TYPE_Dv }, /* Dv */
223: { NULL, TYPE_Er }, /* Er */
224: { NULL, TYPE_Ev }, /* Ev */
225: { NULL, 0 }, /* Ex */
226: { NULL, TYPE_Fa }, /* Fa */
227: { parse_mdoc_Fd, 0 }, /* Fd */
228: { NULL, TYPE_Fl }, /* Fl */
229: { parse_mdoc_Fn, 0 }, /* Fn */
230: { NULL, TYPE_Ft }, /* Ft */
231: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 232: { NULL, TYPE_In }, /* In */
1.47 schwarze 233: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 234: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 235: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 236: { NULL, 0 }, /* Op */
237: { NULL, 0 }, /* Ot */
238: { NULL, TYPE_Pa }, /* Pa */
239: { NULL, 0 }, /* Rv */
1.49 schwarze 240: { NULL, TYPE_St }, /* St */
1.47 schwarze 241: { NULL, TYPE_Va }, /* Va */
242: { parse_mdoc_body, TYPE_Va }, /* Vt */
243: { parse_mdoc_Xr, 0 }, /* Xr */
244: { NULL, 0 }, /* %A */
245: { NULL, 0 }, /* %B */
246: { NULL, 0 }, /* %D */
247: { NULL, 0 }, /* %I */
248: { NULL, 0 }, /* %J */
249: { NULL, 0 }, /* %N */
250: { NULL, 0 }, /* %O */
251: { NULL, 0 }, /* %P */
252: { NULL, 0 }, /* %R */
253: { NULL, 0 }, /* %T */
254: { NULL, 0 }, /* %V */
255: { NULL, 0 }, /* Ac */
256: { NULL, 0 }, /* Ao */
257: { NULL, 0 }, /* Aq */
258: { NULL, TYPE_At }, /* At */
259: { NULL, 0 }, /* Bc */
260: { NULL, 0 }, /* Bf */
261: { NULL, 0 }, /* Bo */
262: { NULL, 0 }, /* Bq */
263: { NULL, TYPE_Bsx }, /* Bsx */
264: { NULL, TYPE_Bx }, /* Bx */
265: { NULL, 0 }, /* Db */
266: { NULL, 0 }, /* Dc */
267: { NULL, 0 }, /* Do */
268: { NULL, 0 }, /* Dq */
269: { NULL, 0 }, /* Ec */
270: { NULL, 0 }, /* Ef */
271: { NULL, TYPE_Em }, /* Em */
272: { NULL, 0 }, /* Eo */
273: { NULL, TYPE_Fx }, /* Fx */
274: { NULL, TYPE_Ms }, /* Ms */
275: { NULL, 0 }, /* No */
276: { NULL, 0 }, /* Ns */
277: { NULL, TYPE_Nx }, /* Nx */
278: { NULL, TYPE_Ox }, /* Ox */
279: { NULL, 0 }, /* Pc */
280: { NULL, 0 }, /* Pf */
281: { NULL, 0 }, /* Po */
282: { NULL, 0 }, /* Pq */
283: { NULL, 0 }, /* Qc */
284: { NULL, 0 }, /* Ql */
285: { NULL, 0 }, /* Qo */
286: { NULL, 0 }, /* Qq */
287: { NULL, 0 }, /* Re */
288: { NULL, 0 }, /* Rs */
289: { NULL, 0 }, /* Sc */
290: { NULL, 0 }, /* So */
291: { NULL, 0 }, /* Sq */
292: { NULL, 0 }, /* Sm */
293: { NULL, 0 }, /* Sx */
294: { NULL, TYPE_Sy }, /* Sy */
295: { NULL, TYPE_Tn }, /* Tn */
296: { NULL, 0 }, /* Ux */
297: { NULL, 0 }, /* Xc */
298: { NULL, 0 }, /* Xo */
1.128 schwarze 299: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 300: { NULL, 0 }, /* Fc */
301: { NULL, 0 }, /* Oo */
302: { NULL, 0 }, /* Oc */
303: { NULL, 0 }, /* Bk */
304: { NULL, 0 }, /* Ek */
305: { NULL, 0 }, /* Bt */
306: { NULL, 0 }, /* Hf */
307: { NULL, 0 }, /* Fr */
308: { NULL, 0 }, /* Ud */
309: { NULL, TYPE_Lb }, /* Lb */
310: { NULL, 0 }, /* Lp */
311: { NULL, TYPE_Lk }, /* Lk */
312: { NULL, TYPE_Mt }, /* Mt */
313: { NULL, 0 }, /* Brq */
314: { NULL, 0 }, /* Bro */
315: { NULL, 0 }, /* Brc */
316: { NULL, 0 }, /* %C */
317: { NULL, 0 }, /* Es */
318: { NULL, 0 }, /* En */
319: { NULL, TYPE_Dx }, /* Dx */
320: { NULL, 0 }, /* %Q */
321: { NULL, 0 }, /* br */
322: { NULL, 0 }, /* sp */
323: { NULL, 0 }, /* %U */
324: { NULL, 0 }, /* Ta */
1.121 schwarze 325: { NULL, 0 }, /* ll */
1.1 schwarze 326: };
327:
1.98 schwarze 328:
1.1 schwarze 329: int
1.3 schwarze 330: mandocdb(int argc, char *argv[])
1.1 schwarze 331: {
1.140 schwarze 332: struct manconf conf;
333: struct ohash_info mpages_info, mlinks_info;
334: struct mparse *mp;
335: const char *path_arg;
336: size_t j, sz;
1.47 schwarze 337: int ch, i;
338:
1.140 schwarze 339: memset(&conf, 0, sizeof(conf));
1.47 schwarze 340: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
341:
342: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 343: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
1.135 schwarze 344: mpages_info.free = mlinks_info.free = hash_free;
345: mpages_info.data = mlinks_info.data = NULL;
1.47 schwarze 346:
347: mpages_info.key_offset = offsetof(struct mpage, inodev);
348: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 349:
350: progname = strrchr(argv[0], '/');
351: if (progname == NULL)
352: progname = argv[0];
353: else
354: ++progname;
355:
1.47 schwarze 356: /*
1.98 schwarze 357: * We accept a few different invocations.
1.47 schwarze 358: * The CHECKOP macro makes sure that invocation styles don't
359: * clobber each other.
360: */
361: #define CHECKOP(_op, _ch) do \
362: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 363: fprintf(stderr, "%s: -%c: Conflicting option\n", \
364: progname, (_ch)); \
1.47 schwarze 365: goto usage; \
366: } while (/*CONSTCOND*/0)
1.10 schwarze 367:
1.47 schwarze 368: path_arg = NULL;
1.28 schwarze 369: op = OP_DEFAULT;
1.1 schwarze 370:
1.83 schwarze 371: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 372: switch (ch) {
1.98 schwarze 373: case 'a':
1.6 schwarze 374: use_all = 1;
375: break;
1.98 schwarze 376: case 'C':
1.47 schwarze 377: CHECKOP(op, ch);
378: path_arg = optarg;
1.28 schwarze 379: op = OP_CONFFILE;
1.25 schwarze 380: break;
1.98 schwarze 381: case 'D':
1.82 schwarze 382: debug++;
383: break;
1.98 schwarze 384: case 'd':
1.47 schwarze 385: CHECKOP(op, ch);
386: path_arg = optarg;
1.2 schwarze 387: op = OP_UPDATE;
1.1 schwarze 388: break;
1.98 schwarze 389: case 'n':
1.47 schwarze 390: nodb = 1;
391: break;
1.98 schwarze 392: case 'p':
1.83 schwarze 393: warnings = 1;
394: break;
1.98 schwarze 395: case 'Q':
1.73 schwarze 396: mparse_options |= MPARSE_QUICK;
1.59 schwarze 397: break;
1.98 schwarze 398: case 'T':
1.52 schwarze 399: if (strcmp(optarg, "utf8")) {
1.111 schwarze 400: fprintf(stderr, "%s: -T%s: "
401: "Unsupported output format\n",
402: progname, optarg);
1.52 schwarze 403: goto usage;
404: }
405: write_utf8 = 1;
406: break;
1.98 schwarze 407: case 't':
1.47 schwarze 408: CHECKOP(op, ch);
1.28 schwarze 409: dup2(STDOUT_FILENO, STDERR_FILENO);
410: op = OP_TEST;
1.47 schwarze 411: nodb = warnings = 1;
1.28 schwarze 412: break;
1.98 schwarze 413: case 'u':
1.47 schwarze 414: CHECKOP(op, ch);
415: path_arg = optarg;
1.1 schwarze 416: op = OP_DELETE;
417: break;
1.98 schwarze 418: case 'v':
1.83 schwarze 419: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 420: break;
1.1 schwarze 421: default:
1.28 schwarze 422: goto usage;
1.1 schwarze 423: }
424:
425: argc -= optind;
426: argv += optind;
427:
1.28 schwarze 428: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 429: fprintf(stderr, "%s: -C: Too many arguments\n",
430: progname);
1.28 schwarze 431: goto usage;
432: }
433:
1.47 schwarze 434: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 435: mchars = mchars_alloc();
1.137 schwarze 436: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.131 schwarze 437: mchars, NULL);
1.47 schwarze 438: ohash_init(&mpages, 6, &mpages_info);
439: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 440:
1.47 schwarze 441: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 442:
1.47 schwarze 443: /*
1.109 schwarze 444: * Most of these deal with a specific directory.
1.97 schwarze 445: * Jump into that directory first.
1.47 schwarze 446: */
1.120 schwarze 447: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 448: goto out;
1.97 schwarze 449:
1.95 schwarze 450: if (dbopen(1)) {
1.97 schwarze 451: /*
452: * The existing database is usable. Process
453: * all files specified on the command-line.
454: */
455: use_all = 1;
1.95 schwarze 456: for (i = 0; i < argc; i++)
457: filescan(argv[i]);
458: if (OP_TEST != op)
459: dbprune();
460: } else {
461: /*
462: * Database missing or corrupt.
463: * Recreate from scratch.
464: */
1.97 schwarze 465: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 466: op = OP_DEFAULT;
467: if (0 == treescan())
468: goto out;
469: if (0 == dbopen(0))
470: goto out;
471: }
1.47 schwarze 472: if (OP_DELETE != op)
1.131 schwarze 473: mpages_merge(mp);
1.95 schwarze 474: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 475: } else {
476: /*
477: * If we have arguments, use them as our manpaths.
478: * If we don't, grok from manpath(1) or however else
1.140 schwarze 479: * manconf_parse() wants to do it.
1.47 schwarze 480: */
481: if (argc > 0) {
1.140 schwarze 482: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 483: argc, sizeof(char *));
1.140 schwarze 484: conf.manpath.sz = (size_t)argc;
1.47 schwarze 485: for (i = 0; i < argc; i++)
1.140 schwarze 486: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 487: } else
1.140 schwarze 488: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 489:
1.140 schwarze 490: if (conf.manpath.sz == 0) {
1.84 schwarze 491: exitcode = (int)MANDOCLEVEL_BADARG;
492: say("", "Empty manpath");
493: }
1.2 schwarze 494:
1.47 schwarze 495: /*
496: * First scan the tree rooted at a base directory, then
497: * build a new database and finally move it into place.
498: * Ignore zero-length directories and strip trailing
499: * slashes.
500: */
1.140 schwarze 501: for (j = 0; j < conf.manpath.sz; j++) {
502: sz = strlen(conf.manpath.paths[j]);
503: if (sz && conf.manpath.paths[j][sz - 1] == '/')
504: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 505: if (0 == sz)
506: continue;
1.2 schwarze 507:
1.47 schwarze 508: if (j) {
509: ohash_init(&mpages, 6, &mpages_info);
510: ohash_init(&mlinks, 6, &mlinks_info);
511: }
1.1 schwarze 512:
1.140 schwarze 513: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 514: continue;
1.47 schwarze 515: if (0 == treescan())
1.120 schwarze 516: continue;
1.47 schwarze 517: if (0 == dbopen(0))
1.120 schwarze 518: continue;
1.41 deraadt 519:
1.131 schwarze 520: mpages_merge(mp);
1.104 schwarze 521: if (warnings && !nodb &&
1.87 schwarze 522: ! (MPARSE_QUICK & mparse_options))
523: names_check();
1.47 schwarze 524: dbclose(0);
1.28 schwarze 525:
1.140 schwarze 526: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 527: mpages_free();
528: ohash_delete(&mpages);
529: ohash_delete(&mlinks);
530: }
1.2 schwarze 531: }
1.47 schwarze 532: }
533: out:
1.140 schwarze 534: manconf_free(&conf);
1.123 schwarze 535: mparse_free(mp);
1.131 schwarze 536: mchars_free(mchars);
1.47 schwarze 537: mpages_free();
538: ohash_delete(&mpages);
539: ohash_delete(&mlinks);
540: return(exitcode);
541: usage:
1.83 schwarze 542: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
543: " %s [-aDnpQ] [-Tutf8] dir ...\n"
544: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
545: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 546: " %s [-Q] -t file ...\n",
1.98 schwarze 547: progname, progname, progname,
1.47 schwarze 548: progname, progname);
1.1 schwarze 549:
1.47 schwarze 550: return((int)MANDOCLEVEL_BADARG);
551: }
1.1 schwarze 552:
1.47 schwarze 553: /*
554: * Scan a directory tree rooted at "basedir" for manpages.
555: * We use fts(), scanning directory parts along the way for clues to our
556: * section and architecture.
557: *
558: * If use_all has been specified, grok all files.
559: * If not, sanitise paths to the following:
560: *
1.98 schwarze 561: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 562: * or
563: * [./]cat<section>[/<arch>]/<name>.0
564: *
565: * TODO: accomodate for multi-language directories.
566: */
567: static int
568: treescan(void)
569: {
1.96 schwarze 570: char buf[PATH_MAX];
1.47 schwarze 571: FTS *f;
572: FTSENT *ff;
573: struct mlink *mlink;
1.81 schwarze 574: int dform, gzip;
1.51 schwarze 575: char *dsec, *arch, *fsec, *cp;
576: const char *path;
1.47 schwarze 577: const char *argv[2];
1.1 schwarze 578:
1.47 schwarze 579: argv[0] = ".";
580: argv[1] = (char *)NULL;
1.1 schwarze 581:
1.96 schwarze 582: f = fts_open((char * const *)argv,
583: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 584: if (NULL == f) {
585: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 586: say("", "&fts_open");
1.47 schwarze 587: return(0);
588: }
1.2 schwarze 589:
1.47 schwarze 590: dsec = arch = NULL;
591: dform = FORM_NONE;
1.2 schwarze 592:
1.47 schwarze 593: while (NULL != (ff = fts_read(f))) {
594: path = ff->fts_path + 2;
1.96 schwarze 595: switch (ff->fts_info) {
596:
597: /*
598: * Symbolic links require various sanity checks,
599: * then get handled just like regular files.
600: */
1.98 schwarze 601: case FTS_SL:
1.96 schwarze 602: if (NULL == realpath(path, buf)) {
603: if (warnings)
604: say(path, "&realpath");
605: continue;
606: }
607: if (strstr(buf, basedir) != buf) {
608: if (warnings) say("",
609: "%s: outside base directory", buf);
610: continue;
611: }
612: /* Use logical inode to avoid mpages dupe. */
613: if (-1 == stat(path, ff->fts_statp)) {
614: if (warnings)
615: say(path, "&stat");
616: continue;
617: }
618: /* FALLTHROUGH */
619:
1.15 schwarze 620: /*
1.47 schwarze 621: * If we're a regular file, add an mlink by using the
622: * stored directory data and handling the filename.
1.15 schwarze 623: */
1.98 schwarze 624: case FTS_F:
1.47 schwarze 625: if (0 == strcmp(path, MANDOC_DB))
626: continue;
627: if ( ! use_all && ff->fts_level < 2) {
628: if (warnings)
629: say(path, "Extraneous file");
630: continue;
1.81 schwarze 631: }
632: gzip = 0;
633: fsec = NULL;
634: while (NULL == fsec) {
635: fsec = strrchr(ff->fts_name, '.');
636: if (NULL == fsec || strcmp(fsec+1, "gz"))
637: break;
638: gzip = 1;
639: *fsec = '\0';
640: fsec = NULL;
641: }
642: if (NULL == fsec) {
1.47 schwarze 643: if ( ! use_all) {
644: if (warnings)
645: say(path,
646: "No filename suffix");
647: continue;
648: }
649: } else if (0 == strcmp(++fsec, "html")) {
650: if (warnings)
651: say(path, "Skip html");
652: continue;
653: } else if (0 == strcmp(fsec, "ps")) {
654: if (warnings)
655: say(path, "Skip ps");
656: continue;
657: } else if (0 == strcmp(fsec, "pdf")) {
658: if (warnings)
659: say(path, "Skip pdf");
660: continue;
661: } else if ( ! use_all &&
1.138 schwarze 662: ((FORM_SRC == dform &&
663: strncmp(fsec, dsec, strlen(dsec))) ||
1.47 schwarze 664: (FORM_CAT == dform && strcmp(fsec, "0")))) {
665: if (warnings)
666: say(path, "Wrong filename suffix");
667: continue;
668: } else
669: fsec[-1] = '\0';
1.51 schwarze 670:
1.47 schwarze 671: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 672: if (strlcpy(mlink->file, path,
673: sizeof(mlink->file)) >=
674: sizeof(mlink->file)) {
675: say(path, "Filename too long");
676: free(mlink);
677: continue;
678: }
1.47 schwarze 679: mlink->dform = dform;
1.51 schwarze 680: mlink->dsec = dsec;
681: mlink->arch = arch;
682: mlink->name = ff->fts_name;
683: mlink->fsec = fsec;
1.81 schwarze 684: mlink->gzip = gzip;
1.47 schwarze 685: mlink_add(mlink, ff->fts_statp);
686: continue;
1.96 schwarze 687:
1.98 schwarze 688: case FTS_D:
1.96 schwarze 689: /* FALLTHROUGH */
1.98 schwarze 690: case FTS_DP:
1.96 schwarze 691: break;
692:
693: default:
1.47 schwarze 694: if (warnings)
695: say(path, "Not a regular file");
696: continue;
697: }
698:
699: switch (ff->fts_level) {
1.98 schwarze 700: case 0:
1.47 schwarze 701: /* Ignore the root directory. */
702: break;
1.98 schwarze 703: case 1:
1.47 schwarze 704: /*
705: * This might contain manX/ or catX/.
706: * Try to infer this from the name.
707: * If we're not in use_all, enforce it.
708: */
709: cp = ff->fts_name;
710: if (FTS_DP == ff->fts_info)
711: break;
1.15 schwarze 712:
1.47 schwarze 713: if (0 == strncmp(cp, "man", 3)) {
714: dform = FORM_SRC;
715: dsec = cp + 3;
716: } else if (0 == strncmp(cp, "cat", 3)) {
717: dform = FORM_CAT;
718: dsec = cp + 3;
1.51 schwarze 719: } else {
720: dform = FORM_NONE;
721: dsec = NULL;
1.26 schwarze 722: }
1.47 schwarze 723:
1.98 schwarze 724: if (NULL != dsec || use_all)
1.47 schwarze 725: break;
726:
727: if (warnings)
728: say(path, "Unknown directory part");
729: fts_set(f, ff, FTS_SKIP);
730: break;
1.98 schwarze 731: case 2:
1.47 schwarze 732: /*
733: * Possibly our architecture.
734: * If we're descending, keep tabs on it.
735: */
736: if (FTS_DP != ff->fts_info && NULL != dsec)
737: arch = ff->fts_name;
1.51 schwarze 738: else
739: arch = NULL;
1.47 schwarze 740: break;
741: default:
742: if (FTS_DP == ff->fts_info || use_all)
743: break;
744: if (warnings)
745: say(path, "Extraneous directory part");
746: fts_set(f, ff, FTS_SKIP);
747: break;
1.14 schwarze 748: }
1.47 schwarze 749: }
750:
751: fts_close(f);
752: return(1);
753: }
1.1 schwarze 754:
1.47 schwarze 755: /*
756: * Add a file to the mlinks table.
757: * Do not verify that it's a "valid" looking manpage (we'll do that
758: * later).
759: *
760: * Try to infer the manual section, architecture, and page name from the
761: * path, assuming it looks like
762: *
1.98 schwarze 763: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 764: * or
765: * [./]cat<section>[/<arch>]/<name>.0
766: *
767: * See treescan() for the fts(3) version of this.
768: */
769: static void
770: filescan(const char *file)
771: {
772: char buf[PATH_MAX];
773: struct stat st;
774: struct mlink *mlink;
775: char *p, *start;
776:
777: assert(use_all);
778:
779: if (0 == strncmp(file, "./", 2))
780: file += 2;
781:
1.96 schwarze 782: /*
783: * We have to do lstat(2) before realpath(3) loses
784: * the information whether this is a symbolic link.
785: * We need to know that because for symbolic links,
786: * we want to use the orginal file name, while for
787: * regular files, we want to use the real path.
788: */
789: if (-1 == lstat(file, &st)) {
790: exitcode = (int)MANDOCLEVEL_BADARG;
791: say(file, "&lstat");
792: return;
793: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
794: exitcode = (int)MANDOCLEVEL_BADARG;
795: say(file, "Not a regular file");
796: return;
797: }
798:
799: /*
800: * We have to resolve the file name to the real path
801: * in any case for the base directory check.
802: */
1.47 schwarze 803: if (NULL == realpath(file, buf)) {
804: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 805: say(file, "&realpath");
1.47 schwarze 806: return;
1.63 schwarze 807: }
808:
1.109 schwarze 809: if (OP_TEST == op)
1.63 schwarze 810: start = buf;
1.109 schwarze 811: else if (strstr(buf, basedir) == buf)
812: start = buf + strlen(basedir);
1.63 schwarze 813: else {
1.47 schwarze 814: exitcode = (int)MANDOCLEVEL_BADARG;
815: say("", "%s: outside base directory", buf);
816: return;
1.63 schwarze 817: }
818:
1.96 schwarze 819: /*
820: * Now we are sure the file is inside our tree.
821: * If it is a symbolic link, ignore the real path
822: * and use the original name.
823: * This implies passing stuff like "cat1/../man1/foo.1"
824: * on the command line won't work. So don't do that.
825: * Note the stat(2) can still fail if the link target
826: * doesn't exist.
827: */
828: if (S_IFLNK & st.st_mode) {
829: if (-1 == stat(buf, &st)) {
830: exitcode = (int)MANDOCLEVEL_BADARG;
831: say(file, "&stat");
832: return;
833: }
1.100 schwarze 834: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
835: say(file, "Filename too long");
836: return;
837: }
1.109 schwarze 838: start = buf;
839: if (OP_TEST != op && strstr(buf, basedir) == buf)
840: start += strlen(basedir);
1.1 schwarze 841: }
1.63 schwarze 842:
1.47 schwarze 843: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 844: mlink->dform = FORM_NONE;
1.100 schwarze 845: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
846: sizeof(mlink->file)) {
847: say(start, "Filename too long");
1.134 schwarze 848: free(mlink);
1.100 schwarze 849: return;
850: }
1.1 schwarze 851:
1.10 schwarze 852: /*
1.47 schwarze 853: * First try to guess our directory structure.
854: * If we find a separator, try to look for man* or cat*.
855: * If we find one of these and what's underneath is a directory,
856: * assume it's an architecture.
1.10 schwarze 857: */
1.47 schwarze 858: if (NULL != (p = strchr(start, '/'))) {
859: *p++ = '\0';
860: if (0 == strncmp(start, "man", 3)) {
861: mlink->dform = FORM_SRC;
1.51 schwarze 862: mlink->dsec = start + 3;
1.47 schwarze 863: } else if (0 == strncmp(start, "cat", 3)) {
864: mlink->dform = FORM_CAT;
1.51 schwarze 865: mlink->dsec = start + 3;
1.47 schwarze 866: }
1.10 schwarze 867:
1.47 schwarze 868: start = p;
869: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
870: *p++ = '\0';
1.51 schwarze 871: mlink->arch = start;
1.47 schwarze 872: start = p;
1.41 deraadt 873: }
1.47 schwarze 874: }
1.7 schwarze 875:
1.47 schwarze 876: /*
877: * Now check the file suffix.
878: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
879: */
880: p = strrchr(start, '\0');
881: while (p-- > start && '/' != *p && '.' != *p)
882: /* Loop. */ ;
883:
884: if ('.' == *p) {
885: *p++ = '\0';
1.51 schwarze 886: mlink->fsec = p;
1.47 schwarze 887: }
1.41 deraadt 888:
1.47 schwarze 889: /*
890: * Now try to parse the name.
891: * Use the filename portion of the path.
892: */
893: mlink->name = start;
894: if (NULL != (p = strrchr(start, '/'))) {
895: mlink->name = p + 1;
896: *p = '\0';
897: }
898: mlink_add(mlink, &st);
899: }
1.2 schwarze 900:
1.47 schwarze 901: static void
902: mlink_add(struct mlink *mlink, const struct stat *st)
903: {
904: struct inodev inodev;
905: struct mpage *mpage;
906: unsigned int slot;
907:
908: assert(NULL != mlink->file);
909:
1.51 schwarze 910: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
911: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
912: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
913: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 914:
915: if ('0' == *mlink->fsec) {
916: free(mlink->fsec);
917: mlink->fsec = mandoc_strdup(mlink->dsec);
918: mlink->fform = FORM_CAT;
919: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
920: mlink->fform = FORM_SRC;
921: else
922: mlink->fform = FORM_NONE;
1.1 schwarze 923:
1.47 schwarze 924: slot = ohash_qlookup(&mlinks, mlink->file);
925: assert(NULL == ohash_find(&mlinks, slot));
926: ohash_insert(&mlinks, slot, mlink);
927:
1.139 schwarze 928: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 929: inodev.st_ino = st->st_ino;
930: inodev.st_dev = st->st_dev;
931: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
932: sizeof(struct inodev), inodev.st_ino);
933: mpage = ohash_find(&mpages, slot);
934: if (NULL == mpage) {
935: mpage = mandoc_calloc(1, sizeof(struct mpage));
936: mpage->inodev.st_ino = inodev.st_ino;
937: mpage->inodev.st_dev = inodev.st_dev;
938: ohash_insert(&mpages, slot, mpage);
939: } else
940: mlink->next = mpage->mlinks;
941: mpage->mlinks = mlink;
1.75 schwarze 942: mlink->mpage = mpage;
1.47 schwarze 943: }
1.1 schwarze 944:
1.47 schwarze 945: static void
946: mlink_free(struct mlink *mlink)
947: {
1.1 schwarze 948:
1.47 schwarze 949: free(mlink->dsec);
950: free(mlink->arch);
951: free(mlink->name);
952: free(mlink->fsec);
953: free(mlink);
954: }
1.1 schwarze 955:
1.47 schwarze 956: static void
957: mpages_free(void)
958: {
959: struct mpage *mpage;
960: struct mlink *mlink;
961: unsigned int slot;
962:
963: mpage = ohash_first(&mpages, &slot);
964: while (NULL != mpage) {
965: while (NULL != (mlink = mpage->mlinks)) {
966: mpage->mlinks = mlink->next;
967: mlink_free(mlink);
968: }
969: free(mpage->sec);
970: free(mpage->arch);
971: free(mpage->title);
972: free(mpage->desc);
973: free(mpage);
974: mpage = ohash_next(&mpages, &slot);
975: }
976: }
1.1 schwarze 977:
1.47 schwarze 978: /*
979: * For each mlink to the mpage, check whether the path looks like
980: * it is formatted, and if it does, check whether a source manual
981: * exists by the same name, ignoring the suffix.
982: * If both conditions hold, drop the mlink.
983: */
984: static void
985: mlinks_undupe(struct mpage *mpage)
986: {
987: char buf[PATH_MAX];
988: struct mlink **prev;
989: struct mlink *mlink;
990: char *bufp;
991:
992: mpage->form = FORM_CAT;
993: prev = &mpage->mlinks;
994: while (NULL != (mlink = *prev)) {
995: if (FORM_CAT != mlink->dform) {
996: mpage->form = FORM_NONE;
997: goto nextlink;
998: }
1.100 schwarze 999: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 1000: bufp = strstr(buf, "cat");
1001: assert(NULL != bufp);
1002: memcpy(bufp, "man", 3);
1003: if (NULL != (bufp = strrchr(buf, '.')))
1004: *++bufp = '\0';
1.100 schwarze 1005: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1006: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1007: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1008: goto nextlink;
1009: if (warnings)
1010: say(mlink->file, "Man source exists: %s", buf);
1011: if (use_all)
1012: goto nextlink;
1013: *prev = mlink->next;
1014: mlink_free(mlink);
1015: continue;
1016: nextlink:
1017: prev = &(*prev)->next;
1.1 schwarze 1018: }
1.47 schwarze 1019: }
1.1 schwarze 1020:
1.88 schwarze 1021: static void
1.50 schwarze 1022: mlink_check(struct mpage *mpage, struct mlink *mlink)
1023: {
1.88 schwarze 1024: struct str *str;
1025: unsigned int slot;
1.50 schwarze 1026:
1027: /*
1028: * Check whether the manual section given in a file
1029: * agrees with the directory where the file is located.
1030: * Some manuals have suffixes like (3p) on their
1031: * section number either inside the file or in the
1032: * directory name, some are linked into more than one
1033: * section, like encrypt(1) = makekey(8).
1034: */
1035:
1036: if (FORM_SRC == mpage->form &&
1.88 schwarze 1037: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1038: say(mlink->file, "Section \"%s\" manual in %s directory",
1039: mpage->sec, mlink->dsec);
1040:
1041: /*
1042: * Manual page directories exist for each kernel
1043: * architecture as returned by machine(1).
1044: * However, many manuals only depend on the
1045: * application architecture as returned by arch(1).
1046: * For example, some (2/ARM) manuals are shared
1047: * across the "armish" and "zaurus" kernel
1048: * architectures.
1049: * A few manuals are even shared across completely
1050: * different architectures, for example fdformat(1)
1051: * on amd64, i386, sparc, and sparc64.
1052: */
1053:
1.88 schwarze 1054: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1055: say(mlink->file, "Architecture \"%s\" manual in "
1056: "\"%s\" directory", mpage->arch, mlink->arch);
1057:
1.88 schwarze 1058: /*
1059: * XXX
1.90 schwarze 1060: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1061: */
1062:
1063: if (FORM_CAT == mpage->form)
1064: return;
1065:
1066: /*
1067: * Check whether this mlink
1068: * appears as a name in the NAME section.
1069: */
1.50 schwarze 1070:
1.90 schwarze 1071: slot = ohash_qlookup(&names, mlink->name);
1072: str = ohash_find(&names, slot);
1.88 schwarze 1073: assert(NULL != str);
1.90 schwarze 1074: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1075: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1076: }
1077:
1.47 schwarze 1078: /*
1079: * Run through the files in the global vector "mpages"
1080: * and add them to the database specified in "basedir".
1081: *
1082: * This handles the parsing scheme itself, using the cues of directory
1083: * and filename to determine whether the file is parsable or not.
1084: */
1085: static void
1.131 schwarze 1086: mpages_merge(struct mparse *mp)
1.47 schwarze 1087: {
1.70 schwarze 1088: char any[] = "any";
1.58 schwarze 1089: struct ohash_info str_info;
1.75 schwarze 1090: struct mpage *mpage, *mpage_dest;
1091: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1092: struct roff_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: man = NULL;
1.81 schwarze 1120: sodest = NULL;
1121:
1.136 schwarze 1122: mparse_open(mp, &fd, mlink->file);
1.117 schwarze 1123: if (fd == -1) {
1.136 schwarze 1124: say(mlink->file, "&open");
1.117 schwarze 1125: goto nextpage;
1.81 schwarze 1126: }
1.11 schwarze 1127:
1128: /*
1.137 schwarze 1129: * Interpret the file as mdoc(7) or man(7) source
1130: * code, unless it is known to be formatted.
1.11 schwarze 1131: */
1.136 schwarze 1132: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1133: mparse_readfd(mp, fd, mlink->file);
1.145 ! schwarze 1134: mparse_result(mp, &man, &sodest);
1.47 schwarze 1135: }
1.11 schwarze 1136:
1.126 schwarze 1137: if (sodest != NULL) {
1.75 schwarze 1138: mlink_dest = ohash_find(&mlinks,
1139: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1140: if (mlink_dest == NULL) {
1141: mandoc_asprintf(&cp, "%s.gz", sodest);
1142: mlink_dest = ohash_find(&mlinks,
1143: ohash_qlookup(&mlinks, cp));
1144: free(cp);
1145: }
1146: if (mlink_dest != NULL) {
1.75 schwarze 1147:
1148: /* The .so target exists. */
1149:
1150: mpage_dest = mlink_dest->mpage;
1151: while (1) {
1152: mlink->mpage = mpage_dest;
1153:
1154: /*
1155: * If the target was already
1156: * processed, add the links
1157: * to the database now.
1158: * Otherwise, this will
1159: * happen when we come
1160: * to the target.
1161: */
1162:
1.94 schwarze 1163: if (mpage_dest->pageid)
1.124 schwarze 1164: dbadd_mlink_name(mlink);
1.75 schwarze 1165:
1.126 schwarze 1166: if (mlink->next == NULL)
1.75 schwarze 1167: break;
1168: mlink = mlink->next;
1169: }
1170:
1171: /* Move all links to the target. */
1172:
1173: mlink->next = mlink_dest->next;
1174: mlink_dest->next = mpage->mlinks;
1175: mpage->mlinks = NULL;
1176: }
1.81 schwarze 1177: goto nextpage;
1.145 ! schwarze 1178: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.47 schwarze 1179: mpage->form = FORM_SRC;
1.145 ! schwarze 1180: mpage->sec = mdoc_meta(man)->msec;
1.112 schwarze 1181: mpage->sec = mandoc_strdup(
1.126 schwarze 1182: mpage->sec == NULL ? "" : mpage->sec);
1.145 ! schwarze 1183: mpage->arch = mdoc_meta(man)->arch;
1.47 schwarze 1184: mpage->arch = mandoc_strdup(
1.126 schwarze 1185: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1186: mpage->title =
1.145 ! schwarze 1187: mandoc_strdup(mdoc_meta(man)->title);
! 1188: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.47 schwarze 1189: mpage->form = FORM_SRC;
1.136 schwarze 1190: mpage->sec = mandoc_strdup(man_meta(man)->msec);
1191: mpage->arch = mandoc_strdup(mlink->arch);
1192: mpage->title = mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1193: } else {
1.47 schwarze 1194: mpage->form = FORM_CAT;
1.136 schwarze 1195: mpage->sec = mandoc_strdup(mlink->dsec);
1196: mpage->arch = mandoc_strdup(mlink->arch);
1197: mpage->title = mandoc_strdup(mlink->name);
1.1 schwarze 1198: }
1.54 schwarze 1199: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1200: if (*mpage->arch != '\0')
1201: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1202:
1.136 schwarze 1203: for ( ; mlink != NULL; mlink = mlink->next) {
1.54 schwarze 1204: if ('\0' != *mlink->dsec)
1205: putkey(mpage, mlink->dsec, TYPE_sec);
1206: if ('\0' != *mlink->fsec)
1207: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1208: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1209: any : mlink->arch, TYPE_arch);
1.90 schwarze 1210: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1211: }
1.41 deraadt 1212:
1.129 schwarze 1213: assert(mpage->desc == NULL);
1.145 ! schwarze 1214: if (man != NULL && man->macroset == MACROSET_MDOC)
! 1215: parse_mdoc(mpage, mdoc_meta(man), mdoc_node(man));
1.129 schwarze 1216: else if (man != NULL)
1217: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1218: else
1.117 schwarze 1219: parse_cat(mpage, fd);
1.129 schwarze 1220: if (mpage->desc == NULL)
1.92 schwarze 1221: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1222:
1223: if (warnings && !use_all)
1224: for (mlink = mpage->mlinks; mlink;
1225: mlink = mlink->next)
1226: mlink_check(mpage, mlink);
1.6 schwarze 1227:
1.131 schwarze 1228: dbadd(mpage);
1.136 schwarze 1229: mlink = mpage->mlinks;
1.81 schwarze 1230:
1231: nextpage:
1.125 schwarze 1232: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.136 schwarze 1234: say(mlink->file, "&wait gunzip");
1.81 schwarze 1235: }
1.47 schwarze 1236: ohash_delete(&strings);
1.90 schwarze 1237: ohash_delete(&names);
1.47 schwarze 1238: mpage = ohash_next(&mpages, &pslot);
1239: }
1.64 schwarze 1240:
1241: if (0 == nodb)
1242: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1243: }
1244:
1245: static void
1246: names_check(void)
1247: {
1248: sqlite3_stmt *stmt;
1249: const char *name, *sec, *arch, *key;
1250: int irc;
1251:
1252: sqlite3_prepare_v2(db,
1253: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1254: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1255: "WHERE bits & ? AND NOT EXISTS ("
1256: "SELECT pageid FROM mlinks "
1.90 schwarze 1257: "WHERE mlinks.pageid == names.pageid "
1258: "AND mlinks.name == names.name"
1.87 schwarze 1259: ")"
1260: ") JOIN ("
1.99 schwarze 1261: "SELECT sec, arch, name, pageid FROM mlinks "
1262: "GROUP BY pageid"
1.87 schwarze 1263: ") USING (pageid);",
1264: -1, &stmt, NULL);
1265:
1.91 schwarze 1266: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1267: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1268:
1269: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1270: name = (const char *)sqlite3_column_text(stmt, 0);
1271: sec = (const char *)sqlite3_column_text(stmt, 1);
1272: arch = (const char *)sqlite3_column_text(stmt, 2);
1273: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1274: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1275: '\0' == *arch ? "" : "/",
1276: '\0' == *arch ? "" : arch, key);
1277: }
1278: sqlite3_finalize(stmt);
1.47 schwarze 1279: }
1.6 schwarze 1280:
1.47 schwarze 1281: static void
1.81 schwarze 1282: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1283: {
1284: FILE *stream;
1285: char *line, *p, *title;
1286: size_t len, plen, titlesz;
1.1 schwarze 1287:
1.81 schwarze 1288: stream = (-1 == fd) ?
1289: fopen(mpage->mlinks->file, "r") :
1290: fdopen(fd, "r");
1291: if (NULL == stream) {
1.114 doug 1292: if (-1 != fd)
1293: close(fd);
1.47 schwarze 1294: if (warnings)
1.80 schwarze 1295: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1296: return;
1297: }
1.1 schwarze 1298:
1.47 schwarze 1299: /* Skip to first blank line. */
1.1 schwarze 1300:
1.47 schwarze 1301: while (NULL != (line = fgetln(stream, &len)))
1302: if ('\n' == *line)
1303: break;
1.1 schwarze 1304:
1.47 schwarze 1305: /*
1306: * Assume the first line that is not indented
1307: * is the first section header. Skip to it.
1308: */
1.1 schwarze 1309:
1.47 schwarze 1310: while (NULL != (line = fgetln(stream, &len)))
1311: if ('\n' != *line && ' ' != *line)
1312: break;
1.98 schwarze 1313:
1.47 schwarze 1314: /*
1315: * Read up until the next section into a buffer.
1316: * Strip the leading and trailing newline from each read line,
1317: * appending a trailing space.
1318: * Ignore empty (whitespace-only) lines.
1319: */
1.28 schwarze 1320:
1.47 schwarze 1321: titlesz = 0;
1322: title = NULL;
1.38 schwarze 1323:
1.47 schwarze 1324: while (NULL != (line = fgetln(stream, &len))) {
1325: if (' ' != *line || '\n' != line[len - 1])
1326: break;
1327: while (len > 0 && isspace((unsigned char)*line)) {
1328: line++;
1329: len--;
1330: }
1331: if (1 == len)
1332: continue;
1333: title = mandoc_realloc(title, titlesz + len);
1334: memcpy(title + titlesz, line, len);
1335: titlesz += len;
1336: title[titlesz - 1] = ' ';
1337: }
1.28 schwarze 1338:
1.47 schwarze 1339: /*
1340: * If no page content can be found, or the input line
1341: * is already the next section header, or there is no
1342: * trailing newline, reuse the page title as the page
1343: * description.
1344: */
1.1 schwarze 1345:
1.47 schwarze 1346: if (NULL == title || '\0' == *title) {
1347: if (warnings)
1348: say(mpage->mlinks->file,
1349: "Cannot find NAME section");
1350: fclose(stream);
1351: free(title);
1352: return;
1353: }
1.24 schwarze 1354:
1.47 schwarze 1355: title = mandoc_realloc(title, titlesz + 1);
1356: title[titlesz] = '\0';
1.24 schwarze 1357:
1.47 schwarze 1358: /*
1359: * Skip to the first dash.
1360: * Use the remaining line as the description (no more than 70
1361: * bytes).
1362: */
1.28 schwarze 1363:
1.47 schwarze 1364: if (NULL != (p = strstr(title, "- "))) {
1365: for (p += 2; ' ' == *p || '\b' == *p; p++)
1366: /* Skip to next word. */ ;
1367: } else {
1368: if (warnings)
1369: say(mpage->mlinks->file,
1370: "No dash in title line");
1371: p = title;
1372: }
1.1 schwarze 1373:
1.47 schwarze 1374: plen = strlen(p);
1.1 schwarze 1375:
1.47 schwarze 1376: /* Strip backspace-encoding from line. */
1.1 schwarze 1377:
1.47 schwarze 1378: while (NULL != (line = memchr(p, '\b', plen))) {
1379: len = line - p;
1380: if (0 == len) {
1381: memmove(line, line + 1, plen--);
1382: continue;
1.98 schwarze 1383: }
1.47 schwarze 1384: memmove(line - 1, line + 1, plen - len);
1385: plen -= 2;
1386: }
1.1 schwarze 1387:
1.47 schwarze 1388: mpage->desc = mandoc_strdup(p);
1389: fclose(stream);
1390: free(title);
1391: }
1.16 schwarze 1392:
1.47 schwarze 1393: /*
1394: * Put a type/word pair into the word database for this particular file.
1395: */
1396: static void
1.69 schwarze 1397: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1398: {
1.69 schwarze 1399: char *cp;
1.37 schwarze 1400:
1.47 schwarze 1401: assert(NULL != value);
1.69 schwarze 1402: if (TYPE_arch == type)
1403: for (cp = value; *cp; cp++)
1404: if (isupper((unsigned char)*cp))
1405: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1406: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1407: }
1408:
1409: /*
1.47 schwarze 1410: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1411: */
1412: static void
1.47 schwarze 1413: putmdockey(const struct mpage *mpage,
1.142 schwarze 1414: const struct roff_node *n, uint64_t m)
1.2 schwarze 1415: {
1.16 schwarze 1416:
1.47 schwarze 1417: for ( ; NULL != n; n = n->next) {
1418: if (NULL != n->child)
1419: putmdockey(mpage, n->child, m);
1.141 schwarze 1420: if (n->type == ROFFT_TEXT)
1.47 schwarze 1421: putkey(mpage, n->string, m);
1422: }
1423: }
1.16 schwarze 1424:
1.47 schwarze 1425: static void
1.143 schwarze 1426: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1427: const struct roff_node *n)
1.47 schwarze 1428: {
1.142 schwarze 1429: const struct roff_node *head, *body;
1.78 schwarze 1430: char *start, *title;
1.47 schwarze 1431: char byte;
1.78 schwarze 1432: size_t sz;
1.16 schwarze 1433:
1.47 schwarze 1434: if (NULL == n)
1435: return;
1.16 schwarze 1436:
1.47 schwarze 1437: /*
1438: * We're only searching for one thing: the first text child in
1439: * the BODY of a NAME section. Since we don't keep track of
1440: * sections in -man, run some hoops to find out whether we're in
1441: * the correct section or not.
1442: */
1.16 schwarze 1443:
1.141 schwarze 1444: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1445: body = n;
1446: assert(body->parent);
1447: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1448: 1 == head->nchild &&
1449: NULL != (head = (head->child)) &&
1.141 schwarze 1450: head->type == ROFFT_TEXT &&
1.98 schwarze 1451: 0 == strcmp(head->string, "NAME") &&
1452: NULL != body->child) {
1.2 schwarze 1453:
1.47 schwarze 1454: /*
1455: * Suck the entire NAME section into memory.
1456: * Yes, we might run away.
1457: * But too many manuals have big, spread-out
1458: * NAME sections over many lines.
1459: */
1.2 schwarze 1460:
1.78 schwarze 1461: title = NULL;
1462: man_deroff(&title, body);
1.47 schwarze 1463: if (NULL == title)
1464: return;
1.16 schwarze 1465:
1.98 schwarze 1466: /*
1.47 schwarze 1467: * Go through a special heuristic dance here.
1468: * Conventionally, one or more manual names are
1469: * comma-specified prior to a whitespace, then a
1470: * dash, then a description. Try to puzzle out
1471: * the name parts here.
1472: */
1.16 schwarze 1473:
1.78 schwarze 1474: start = title;
1.47 schwarze 1475: for ( ;; ) {
1476: sz = strcspn(start, " ,");
1477: if ('\0' == start[sz])
1478: break;
1.1 schwarze 1479:
1.47 schwarze 1480: byte = start[sz];
1481: start[sz] = '\0';
1.67 schwarze 1482:
1483: /*
1484: * Assume a stray trailing comma in the
1485: * name list if a name begins with a dash.
1486: */
1487:
1488: if ('-' == start[0] ||
1489: ('\\' == start[0] && '-' == start[1]))
1490: break;
1.1 schwarze 1491:
1.90 schwarze 1492: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1493: if ( ! (mpage->name_head_done ||
1494: strcasecmp(start, meta->title))) {
1495: putkey(mpage, start, NAME_HEAD);
1496: mpage->name_head_done = 1;
1497: }
1.1 schwarze 1498:
1.47 schwarze 1499: if (' ' == byte) {
1500: start += sz + 1;
1501: break;
1502: }
1.1 schwarze 1503:
1.47 schwarze 1504: assert(',' == byte);
1505: start += sz + 1;
1506: while (' ' == *start)
1507: start++;
1508: }
1.1 schwarze 1509:
1.78 schwarze 1510: if (start == title) {
1.90 schwarze 1511: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1512: if ( ! (mpage->name_head_done ||
1513: strcasecmp(start, meta->title))) {
1514: putkey(mpage, start, NAME_HEAD);
1515: mpage->name_head_done = 1;
1516: }
1.47 schwarze 1517: free(title);
1518: return;
1519: }
1.1 schwarze 1520:
1.47 schwarze 1521: while (isspace((unsigned char)*start))
1522: start++;
1.1 schwarze 1523:
1.47 schwarze 1524: if (0 == strncmp(start, "-", 1))
1525: start += 1;
1526: else if (0 == strncmp(start, "\\-\\-", 4))
1527: start += 4;
1528: else if (0 == strncmp(start, "\\-", 2))
1529: start += 2;
1530: else if (0 == strncmp(start, "\\(en", 4))
1531: start += 4;
1532: else if (0 == strncmp(start, "\\(em", 4))
1533: start += 4;
1.1 schwarze 1534:
1.47 schwarze 1535: while (' ' == *start)
1536: start++;
1.1 schwarze 1537:
1.47 schwarze 1538: mpage->desc = mandoc_strdup(start);
1539: free(title);
1540: return;
1541: }
1542: }
1.1 schwarze 1543:
1.47 schwarze 1544: for (n = n->child; n; n = n->next) {
1545: if (NULL != mpage->desc)
1546: break;
1.129 schwarze 1547: parse_man(mpage, meta, n);
1.1 schwarze 1548: }
1549: }
1550:
1551: static void
1.143 schwarze 1552: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1553: const struct roff_node *n)
1.1 schwarze 1554: {
1555:
1.47 schwarze 1556: assert(NULL != n);
1557: for (n = n->child; NULL != n; n = n->next) {
1558: switch (n->type) {
1.141 schwarze 1559: case ROFFT_ELEM:
1.47 schwarze 1560: /* FALLTHROUGH */
1.141 schwarze 1561: case ROFFT_BLOCK:
1.47 schwarze 1562: /* FALLTHROUGH */
1.141 schwarze 1563: case ROFFT_HEAD:
1.47 schwarze 1564: /* FALLTHROUGH */
1.141 schwarze 1565: case ROFFT_BODY:
1.47 schwarze 1566: /* FALLTHROUGH */
1.141 schwarze 1567: case ROFFT_TAIL:
1.47 schwarze 1568: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1569: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1570: break;
1571: if (mdocs[n->tok].mask)
1572: putmdockey(mpage, n->child,
1573: mdocs[n->tok].mask);
1574: break;
1575: default:
1.141 schwarze 1576: assert(n->type != ROFFT_ROOT);
1.47 schwarze 1577: continue;
1578: }
1579: if (NULL != n->child)
1.127 schwarze 1580: parse_mdoc(mpage, meta, n);
1.1 schwarze 1581: }
1582: }
1583:
1.19 schwarze 1584: static int
1.143 schwarze 1585: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1586: const struct roff_node *n)
1.1 schwarze 1587: {
1.131 schwarze 1588: char *start, *end;
1.1 schwarze 1589: size_t sz;
1.19 schwarze 1590:
1.47 schwarze 1591: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1592: NULL == (n = n->child) ||
1.141 schwarze 1593: n->type != ROFFT_TEXT)
1.19 schwarze 1594: return(0);
1.1 schwarze 1595:
1596: /*
1597: * Only consider those `Fd' macro fields that begin with an
1598: * "inclusion" token (versus, e.g., #define).
1599: */
1.47 schwarze 1600:
1.1 schwarze 1601: if (strcmp("#include", n->string))
1.19 schwarze 1602: return(0);
1.1 schwarze 1603:
1.141 schwarze 1604: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.19 schwarze 1605: return(0);
1.1 schwarze 1606:
1607: /*
1608: * Strip away the enclosing angle brackets and make sure we're
1609: * not zero-length.
1610: */
1611:
1612: start = n->string;
1613: if ('<' == *start || '"' == *start)
1614: start++;
1615:
1616: if (0 == (sz = strlen(start)))
1.19 schwarze 1617: return(0);
1.1 schwarze 1618:
1619: end = &start[(int)sz - 1];
1620: if ('>' == *end || '"' == *end)
1621: end--;
1622:
1.47 schwarze 1623: if (end > start)
1624: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1625: return(0);
1.1 schwarze 1626: }
1627:
1.133 schwarze 1628: static void
1.142 schwarze 1629: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 schwarze 1630: {
1.69 schwarze 1631: char *cp;
1.133 schwarze 1632: size_t sz;
1.1 schwarze 1633:
1.141 schwarze 1634: if (n->type != ROFFT_TEXT)
1.133 schwarze 1635: return;
1.19 schwarze 1636:
1.133 schwarze 1637: /* Skip function pointer punctuation. */
1.1 schwarze 1638:
1.133 schwarze 1639: cp = n->string;
1640: while (*cp == '(' || *cp == '*')
1.1 schwarze 1641: cp++;
1.133 schwarze 1642: sz = strcspn(cp, "()");
1.1 schwarze 1643:
1.133 schwarze 1644: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1645: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1646: putkeys(mpage, cp, sz, NAME_SYN);
1647: }
1.19 schwarze 1648:
1.133 schwarze 1649: static int
1.143 schwarze 1650: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1651: const struct roff_node *n)
1.133 schwarze 1652: {
1653:
1654: if (n->child == NULL)
1655: return(0);
1656:
1657: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1658:
1.133 schwarze 1659: for (n = n->child->next; n != NULL; n = n->next)
1.141 schwarze 1660: if (n->type == ROFFT_TEXT)
1.47 schwarze 1661: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1662:
1.128 schwarze 1663: return(0);
1664: }
1665:
1666: static int
1.143 schwarze 1667: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1668: const struct roff_node *n)
1.128 schwarze 1669: {
1.132 schwarze 1670:
1.141 schwarze 1671: if (n->type != ROFFT_HEAD)
1.132 schwarze 1672: return(1);
1.128 schwarze 1673:
1.133 schwarze 1674: if (n->child != NULL)
1675: parse_mdoc_fname(mpage, n->child);
1676:
1.19 schwarze 1677: return(0);
1.1 schwarze 1678: }
1679:
1.19 schwarze 1680: static int
1.143 schwarze 1681: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1682: const struct roff_node *n)
1.1 schwarze 1683: {
1.47 schwarze 1684: char *cp;
1.1 schwarze 1685:
1686: if (NULL == (n = n->child))
1.19 schwarze 1687: return(0);
1.1 schwarze 1688:
1.47 schwarze 1689: if (NULL == n->next) {
1690: putkey(mpage, n->string, TYPE_Xr);
1691: return(0);
1692: }
1.1 schwarze 1693:
1.77 schwarze 1694: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1695: putkey(mpage, cp, TYPE_Xr);
1696: free(cp);
1697: return(0);
1.1 schwarze 1698: }
1699:
1.19 schwarze 1700: static int
1.143 schwarze 1701: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1702: const struct roff_node *n)
1.1 schwarze 1703: {
1704:
1.141 schwarze 1705: if (n->type == ROFFT_BODY)
1.79 schwarze 1706: mdoc_deroff(&mpage->desc, n);
1707: return(0);
1.1 schwarze 1708: }
1709:
1.19 schwarze 1710: static int
1.143 schwarze 1711: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1712: const struct roff_node *n)
1.1 schwarze 1713: {
1714:
1.86 schwarze 1715: if (SEC_NAME == n->sec)
1.90 schwarze 1716: putmdockey(mpage, n->child, NAME_TITLE);
1.141 schwarze 1717: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.127 schwarze 1718: if (n->child == NULL)
1719: putkey(mpage, meta->name, NAME_SYN);
1720: else
1721: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1722: }
1723: if ( ! (mpage->name_head_done ||
1724: n->child == NULL || n->child->string == NULL ||
1725: strcasecmp(n->child->string, meta->title))) {
1.141 schwarze 1726: putkey(mpage, n->child->string, ROFFT_HEAD);
1.129 schwarze 1727: mpage->name_head_done = 1;
1.127 schwarze 1728: }
1.86 schwarze 1729: return(0);
1.1 schwarze 1730: }
1731:
1.19 schwarze 1732: static int
1.143 schwarze 1733: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1734: const struct roff_node *n)
1.1 schwarze 1735: {
1736:
1.141 schwarze 1737: return(n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD);
1.1 schwarze 1738: }
1739:
1.47 schwarze 1740: static int
1.143 schwarze 1741: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1742: const struct roff_node *n)
1.1 schwarze 1743: {
1744:
1.141 schwarze 1745: return(n->type == ROFFT_HEAD);
1.47 schwarze 1746: }
1.1 schwarze 1747:
1.47 schwarze 1748: static int
1.143 schwarze 1749: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1750: const struct roff_node *n)
1.47 schwarze 1751: {
1.1 schwarze 1752:
1.141 schwarze 1753: return(n->type == ROFFT_BODY);
1.1 schwarze 1754: }
1755:
1.47 schwarze 1756: /*
1757: * Add a string to the hash table for the current manual.
1758: * Each string has a bitmask telling which macros it belongs to.
1759: * When we finish the manual, we'll dump the table.
1760: */
1.1 schwarze 1761: static void
1.131 schwarze 1762: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1763: {
1.90 schwarze 1764: struct ohash *htab;
1.47 schwarze 1765: struct str *s;
1.68 schwarze 1766: const char *end;
1.47 schwarze 1767: unsigned int slot;
1.131 schwarze 1768: int i, mustfree;
1.1 schwarze 1769:
1.47 schwarze 1770: if (0 == sz)
1771: return;
1.68 schwarze 1772:
1.131 schwarze 1773: mustfree = render_string(&cp, &sz);
1774:
1.90 schwarze 1775: if (TYPE_Nm & v) {
1776: htab = &names;
1777: v &= name_mask;
1.124 schwarze 1778: if (v & NAME_FIRST)
1779: name_mask &= ~NAME_FIRST;
1.90 schwarze 1780: if (debug > 1)
1781: say(mpage->mlinks->file,
1.130 schwarze 1782: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1783: } else {
1784: htab = &strings;
1785: if (debug > 1)
1786: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1787: if ((uint64_t)1 << i & v)
1.90 schwarze 1788: say(mpage->mlinks->file,
1789: "Adding key %s=%*s",
1790: mansearch_keynames[i], sz, cp);
1.68 schwarze 1791: }
1.47 schwarze 1792:
1793: end = cp + sz;
1.90 schwarze 1794: slot = ohash_qlookupi(htab, cp, &end);
1795: s = ohash_find(htab, slot);
1.1 schwarze 1796:
1.47 schwarze 1797: if (NULL != s && mpage == s->mpage) {
1798: s->mask |= v;
1.1 schwarze 1799: return;
1.47 schwarze 1800: } else if (NULL == s) {
1.101 schwarze 1801: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1802: memcpy(s->key, cp, sz);
1.90 schwarze 1803: ohash_insert(htab, slot, s);
1.47 schwarze 1804: }
1805: s->mpage = mpage;
1806: s->mask = v;
1.131 schwarze 1807:
1808: if (mustfree)
1809: free(cp);
1.1 schwarze 1810: }
1811:
1812: /*
1.47 schwarze 1813: * Take a Unicode codepoint and produce its UTF-8 encoding.
1814: * This isn't the best way to do this, but it works.
1815: * The magic numbers are from the UTF-8 packaging.
1816: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1817: */
1.47 schwarze 1818: static size_t
1819: utf8(unsigned int cp, char out[7])
1.1 schwarze 1820: {
1.47 schwarze 1821: size_t rc;
1.1 schwarze 1822:
1.47 schwarze 1823: rc = 0;
1824: if (cp <= 0x0000007F) {
1825: rc = 1;
1826: out[0] = (char)cp;
1827: } else if (cp <= 0x000007FF) {
1828: rc = 2;
1829: out[0] = (cp >> 6 & 31) | 192;
1830: out[1] = (cp & 63) | 128;
1831: } else if (cp <= 0x0000FFFF) {
1832: rc = 3;
1833: out[0] = (cp >> 12 & 15) | 224;
1834: out[1] = (cp >> 6 & 63) | 128;
1835: out[2] = (cp & 63) | 128;
1836: } else if (cp <= 0x001FFFFF) {
1837: rc = 4;
1838: out[0] = (cp >> 18 & 7) | 240;
1839: out[1] = (cp >> 12 & 63) | 128;
1840: out[2] = (cp >> 6 & 63) | 128;
1841: out[3] = (cp & 63) | 128;
1842: } else if (cp <= 0x03FFFFFF) {
1843: rc = 5;
1844: out[0] = (cp >> 24 & 3) | 248;
1845: out[1] = (cp >> 18 & 63) | 128;
1846: out[2] = (cp >> 12 & 63) | 128;
1847: out[3] = (cp >> 6 & 63) | 128;
1848: out[4] = (cp & 63) | 128;
1849: } else if (cp <= 0x7FFFFFFF) {
1850: rc = 6;
1851: out[0] = (cp >> 30 & 1) | 252;
1852: out[1] = (cp >> 24 & 63) | 128;
1853: out[2] = (cp >> 18 & 63) | 128;
1854: out[3] = (cp >> 12 & 63) | 128;
1855: out[4] = (cp >> 6 & 63) | 128;
1856: out[5] = (cp & 63) | 128;
1857: } else
1858: return(0);
1.19 schwarze 1859:
1.47 schwarze 1860: out[rc] = '\0';
1861: return(rc);
1.1 schwarze 1862: }
1863:
1.47 schwarze 1864: /*
1.131 schwarze 1865: * If the string contains escape sequences,
1866: * replace it with an allocated rendering and return 1,
1867: * such that the caller can free it after use.
1868: * Otherwise, do nothing and return 0.
1.47 schwarze 1869: */
1.131 schwarze 1870: static int
1871: render_string(char **public, size_t *psz)
1.1 schwarze 1872: {
1.131 schwarze 1873: const char *src, *scp, *addcp, *seq;
1874: char *dst;
1875: size_t ssz, dsz, addsz;
1.71 schwarze 1876: char utfbuf[7], res[6];
1.131 schwarze 1877: int seqlen, unicode;
1.47 schwarze 1878:
1879: res[0] = '\\';
1880: res[1] = '\t';
1881: res[2] = ASCII_NBRSP;
1882: res[3] = ASCII_HYPH;
1.71 schwarze 1883: res[4] = ASCII_BREAK;
1884: res[5] = '\0';
1.1 schwarze 1885:
1.131 schwarze 1886: src = scp = *public;
1887: ssz = *psz;
1888: dst = NULL;
1889: dsz = 0;
1890:
1891: while (scp < src + *psz) {
1892:
1893: /* Leave normal characters unchanged. */
1894:
1895: if (strchr(res, *scp) == NULL) {
1896: if (dst != NULL)
1897: dst[dsz++] = *scp;
1898: scp++;
1899: continue;
1900: }
1.1 schwarze 1901:
1.47 schwarze 1902: /*
1.131 schwarze 1903: * Found something that requires replacing,
1904: * make sure we have a destination buffer.
1.47 schwarze 1905: */
1.131 schwarze 1906:
1907: if (dst == NULL) {
1908: dst = mandoc_malloc(ssz + 1);
1909: dsz = scp - src;
1910: memcpy(dst, src, dsz);
1.47 schwarze 1911: }
1.39 schwarze 1912:
1.131 schwarze 1913: /* Handle single-char special characters. */
1914:
1915: switch (*scp) {
1916: case '\\':
1917: break;
1.98 schwarze 1918: case '\t':
1.71 schwarze 1919: /* FALLTHROUGH */
1.98 schwarze 1920: case ASCII_NBRSP:
1.131 schwarze 1921: dst[dsz++] = ' ';
1922: scp++;
1923: continue;
1924: case ASCII_HYPH:
1925: dst[dsz++] = '-';
1.71 schwarze 1926: /* FALLTHROUGH */
1.98 schwarze 1927: case ASCII_BREAK:
1.131 schwarze 1928: scp++;
1.47 schwarze 1929: continue;
1.71 schwarze 1930: default:
1.131 schwarze 1931: abort();
1.71 schwarze 1932: }
1.39 schwarze 1933:
1.47 schwarze 1934: /*
1.131 schwarze 1935: * Found an escape sequence.
1936: * Read past the slash, then parse it.
1937: * Ignore everything except characters.
1.47 schwarze 1938: */
1.52 schwarze 1939:
1.131 schwarze 1940: scp++;
1941: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1942: continue;
1.39 schwarze 1943:
1.47 schwarze 1944: /*
1.52 schwarze 1945: * Render the special character
1946: * as either UTF-8 or ASCII.
1.47 schwarze 1947: */
1.52 schwarze 1948:
1949: if (write_utf8) {
1.131 schwarze 1950: unicode = mchars_spec2cp(mchars, seq, seqlen);
1951: if (unicode <= 0)
1.52 schwarze 1952: continue;
1.131 schwarze 1953: addsz = utf8(unicode, utfbuf);
1954: if (addsz == 0)
1.52 schwarze 1955: continue;
1.131 schwarze 1956: addcp = utfbuf;
1.52 schwarze 1957: } else {
1.131 schwarze 1958: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1959: if (addcp == NULL)
1.52 schwarze 1960: continue;
1.131 schwarze 1961: if (*addcp == ASCII_NBRSP) {
1962: addcp = " ";
1963: addsz = 1;
1.52 schwarze 1964: }
1965: }
1.1 schwarze 1966:
1.47 schwarze 1967: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1968:
1.131 schwarze 1969: ssz += addsz;
1970: dst = mandoc_realloc(dst, ssz + 1);
1971: memcpy(dst + dsz, addcp, addsz);
1972: dsz += addsz;
1973: }
1974: if (dst != NULL) {
1975: *public = dst;
1976: *psz = dsz;
1977: }
1978:
1979: /* Trim trailing whitespace and NUL-terminate. */
1980:
1981: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1982: --*psz;
1983: if (dst != NULL) {
1984: (*public)[*psz] = '\0';
1985: return(1);
1986: } else
1987: return(0);
1.1 schwarze 1988: }
1989:
1.75 schwarze 1990: static void
1991: dbadd_mlink(const struct mlink *mlink)
1992: {
1993: size_t i;
1994:
1995: i = 1;
1996: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1997: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1998: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1999: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 2000: SQL_STEP(stmts[STMT_INSERT_LINK]);
2001: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 2002: }
2003:
2004: static void
2005: dbadd_mlink_name(const struct mlink *mlink)
2006: {
1.130 schwarze 2007: uint64_t bits;
1.124 schwarze 2008: size_t i;
2009:
2010: dbadd_mlink(mlink);
1.115 schwarze 2011:
2012: i = 1;
1.130 schwarze 2013: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2014: bits = NAME_FILE & NAME_MASK;
2015: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2016: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2017: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2018: }
2019:
2020: i = 1;
2021: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2022: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2023: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2024: SQL_STEP(stmts[STMT_INSERT_NAME]);
2025: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2026: }
2027:
1.11 schwarze 2028: /*
1.47 schwarze 2029: * Flush the current page's terms (and their bits) into the database.
2030: * Wrap the entire set of additions in a transaction to make sqlite be a
2031: * little faster.
1.53 schwarze 2032: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2033: */
2034: static void
1.131 schwarze 2035: dbadd(struct mpage *mpage)
1.11 schwarze 2036: {
1.47 schwarze 2037: struct mlink *mlink;
2038: struct str *key;
1.131 schwarze 2039: char *cp;
1.47 schwarze 2040: size_t i;
2041: unsigned int slot;
1.131 schwarze 2042: int mustfree;
1.47 schwarze 2043:
1.85 schwarze 2044: mlink = mpage->mlinks;
1.11 schwarze 2045:
1.85 schwarze 2046: if (nodb) {
1.104 schwarze 2047: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2048: key = ohash_next(&names, &slot))
1.104 schwarze 2049: free(key);
2050: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2051: key = ohash_next(&strings, &slot))
1.104 schwarze 2052: free(key);
1.102 schwarze 2053: if (0 == debug)
2054: return;
1.85 schwarze 2055: while (NULL != mlink) {
2056: fputs(mlink->name, stdout);
2057: if (NULL == mlink->next ||
2058: strcmp(mlink->dsec, mlink->next->dsec) ||
2059: strcmp(mlink->fsec, mlink->next->fsec) ||
2060: strcmp(mlink->arch, mlink->next->arch)) {
2061: putchar('(');
2062: if ('\0' == *mlink->dsec)
2063: fputs(mlink->fsec, stdout);
2064: else
2065: fputs(mlink->dsec, stdout);
2066: if ('\0' != *mlink->arch)
2067: printf("/%s", mlink->arch);
2068: putchar(')');
2069: }
2070: mlink = mlink->next;
2071: if (NULL != mlink)
2072: fputs(", ", stdout);
2073: }
1.89 schwarze 2074: printf(" - %s\n", mpage->desc);
1.11 schwarze 2075: return;
1.85 schwarze 2076: }
2077:
2078: if (debug)
2079: say(mlink->file, "Adding to database");
1.47 schwarze 2080:
1.131 schwarze 2081: cp = mpage->desc;
2082: i = strlen(cp);
2083: mustfree = render_string(&cp, &i);
1.47 schwarze 2084: i = 1;
1.131 schwarze 2085: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2086: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2087: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2088: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2089: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2090: if (mustfree)
2091: free(cp);
1.47 schwarze 2092:
1.85 schwarze 2093: while (NULL != mlink) {
1.75 schwarze 2094: dbadd_mlink(mlink);
1.85 schwarze 2095: mlink = mlink->next;
2096: }
1.91 schwarze 2097: mlink = mpage->mlinks;
1.47 schwarze 2098:
1.90 schwarze 2099: for (key = ohash_first(&names, &slot); NULL != key;
2100: key = ohash_next(&names, &slot)) {
2101: assert(key->mpage == mpage);
2102: i = 1;
2103: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2104: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2105: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2106: SQL_STEP(stmts[STMT_INSERT_NAME]);
2107: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2108: free(key);
2109: }
1.47 schwarze 2110: for (key = ohash_first(&strings, &slot); NULL != key;
2111: key = ohash_next(&strings, &slot)) {
2112: assert(key->mpage == mpage);
2113: i = 1;
2114: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2115: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2116: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2117: SQL_STEP(stmts[STMT_INSERT_KEY]);
2118: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2119: free(key);
1.33 schwarze 2120: }
1.47 schwarze 2121: }
1.41 deraadt 2122:
1.47 schwarze 2123: static void
2124: dbprune(void)
2125: {
2126: struct mpage *mpage;
2127: struct mlink *mlink;
2128: size_t i;
2129: unsigned int slot;
1.11 schwarze 2130:
1.63 schwarze 2131: if (0 == nodb)
2132: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2133:
1.63 schwarze 2134: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2135: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2136: mlink = mpage->mlinks;
1.82 schwarze 2137: if (debug)
1.63 schwarze 2138: say(mlink->file, "Deleting from database");
2139: if (nodb)
2140: continue;
2141: for ( ; NULL != mlink; mlink = mlink->next) {
2142: i = 1;
2143: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2144: i, mlink->dsec);
2145: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2146: i, mlink->arch);
2147: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2148: i, mlink->name);
2149: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2150: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2151: }
1.11 schwarze 2152: }
1.63 schwarze 2153:
2154: if (0 == nodb)
2155: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2156: }
1.22 schwarze 2157:
1.47 schwarze 2158: /*
2159: * Close an existing database and its prepared statements.
2160: * If "real" is not set, rename the temporary file into the real one.
2161: */
2162: static void
2163: dbclose(int real)
2164: {
2165: size_t i;
1.72 schwarze 2166: int status;
2167: pid_t child;
1.11 schwarze 2168:
1.47 schwarze 2169: if (nodb)
2170: return;
1.11 schwarze 2171:
1.47 schwarze 2172: for (i = 0; i < STMT__MAX; i++) {
2173: sqlite3_finalize(stmts[i]);
2174: stmts[i] = NULL;
1.28 schwarze 2175: }
1.22 schwarze 2176:
1.47 schwarze 2177: sqlite3_close(db);
2178: db = NULL;
1.11 schwarze 2179:
1.47 schwarze 2180: if (real)
2181: return;
1.22 schwarze 2182:
1.72 schwarze 2183: if ('\0' == *tempfilename) {
2184: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2185: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2186: say(MANDOC_DB, "&rename");
1.72 schwarze 2187: }
2188: return;
2189: }
2190:
2191: switch (child = fork()) {
1.98 schwarze 2192: case -1:
1.72 schwarze 2193: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2194: say("", "&fork cmp");
1.72 schwarze 2195: return;
1.98 schwarze 2196: case 0:
1.72 schwarze 2197: execlp("cmp", "cmp", "-s",
2198: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2199: say("", "&exec cmp");
1.72 schwarze 2200: exit(0);
2201: default:
2202: break;
2203: }
2204: if (-1 == waitpid(child, &status, 0)) {
2205: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2206: say("", "&wait cmp");
1.72 schwarze 2207: } else if (WIFSIGNALED(status)) {
2208: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2209: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2210: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2212: say(MANDOC_DB,
2213: "Data changed, but cannot replace database");
2214: }
2215:
2216: *strrchr(tempfilename, '/') = '\0';
2217: switch (child = fork()) {
1.98 schwarze 2218: case -1:
1.72 schwarze 2219: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2220: say("", "&fork rm");
1.72 schwarze 2221: return;
1.98 schwarze 2222: case 0:
1.72 schwarze 2223: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2224: say("", "&exec rm");
1.72 schwarze 2225: exit((int)MANDOCLEVEL_SYSERR);
2226: default:
2227: break;
2228: }
2229: if (-1 == waitpid(child, &status, 0)) {
2230: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2231: say("", "&wait rm");
1.72 schwarze 2232: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2234: say("", "%s: Cannot remove temporary directory",
2235: tempfilename);
1.22 schwarze 2236: }
1.11 schwarze 2237: }
2238:
1.47 schwarze 2239: /*
2240: * This is straightforward stuff.
2241: * Open a database connection to a "temporary" database, then open a set
2242: * of prepared statements we'll use over and over again.
2243: * If "real" is set, we use the existing database; if not, we truncate a
2244: * temporary one.
2245: * Must be matched by dbclose().
2246: */
2247: static int
2248: dbopen(int real)
1.2 schwarze 2249: {
1.72 schwarze 2250: const char *sql;
1.47 schwarze 2251: int rc, ofl;
1.6 schwarze 2252:
1.98 schwarze 2253: if (nodb)
1.47 schwarze 2254: return(1);
1.6 schwarze 2255:
1.72 schwarze 2256: *tempfilename = '\0';
1.47 schwarze 2257: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2258:
2259: if (real) {
2260: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2261: if (SQLITE_OK != rc) {
1.47 schwarze 2262: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2263: if (SQLITE_CANTOPEN != rc)
2264: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2265: return(0);
1.28 schwarze 2266: }
1.72 schwarze 2267: goto prepare_statements;
2268: }
2269:
2270: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2271:
1.72 schwarze 2272: remove(MANDOC_DB "~");
2273: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2274: if (SQLITE_OK == rc)
1.72 schwarze 2275: goto create_tables;
1.73 schwarze 2276: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2277: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2278: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2279: return(0);
2280: }
1.6 schwarze 2281:
1.100 schwarze 2282: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2283: sizeof(tempfilename));
1.72 schwarze 2284: if (NULL == mkdtemp(tempfilename)) {
2285: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2286: say("", "&%s", tempfilename);
1.72 schwarze 2287: return(0);
2288: }
1.100 schwarze 2289: (void)strlcat(tempfilename, "/" MANDOC_DB,
2290: sizeof(tempfilename));
1.72 schwarze 2291: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2292: if (SQLITE_OK != rc) {
1.47 schwarze 2293: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2294: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2295: return(0);
1.2 schwarze 2296: }
2297:
1.72 schwarze 2298: create_tables:
1.47 schwarze 2299: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2300: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2301: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2302: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2303: ");\n"
2304: "\n"
2305: "CREATE TABLE \"mlinks\" (\n"
2306: " \"sec\" TEXT NOT NULL,\n"
2307: " \"arch\" TEXT NOT NULL,\n"
2308: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2309: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2310: "ON DELETE CASCADE\n"
1.47 schwarze 2311: ");\n"
1.93 schwarze 2312: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2313: "\n"
1.90 schwarze 2314: "CREATE TABLE \"names\" (\n"
2315: " \"bits\" INTEGER NOT NULL,\n"
2316: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2317: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2318: "ON DELETE CASCADE,\n"
2319: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2320: ");\n"
2321: "\n"
1.47 schwarze 2322: "CREATE TABLE \"keys\" (\n"
2323: " \"bits\" INTEGER NOT NULL,\n"
2324: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2325: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2326: "ON DELETE CASCADE\n"
1.93 schwarze 2327: ");\n"
2328: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2329:
2330: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2331: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2332: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2333: sqlite3_close(db);
1.47 schwarze 2334: return(0);
1.2 schwarze 2335: }
2336:
1.47 schwarze 2337: prepare_statements:
1.103 schwarze 2338: if (SQLITE_OK != sqlite3_exec(db,
2339: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2340: exitcode = (int)MANDOCLEVEL_SYSERR;
2341: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2342: sqlite3_errmsg(db));
2343: sqlite3_close(db);
2344: return(0);
2345: }
2346:
1.94 schwarze 2347: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2348: "(SELECT pageid FROM mlinks WHERE "
2349: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2350: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2351: sql = "INSERT INTO mpages "
1.89 schwarze 2352: "(desc,form) VALUES (?,?)";
1.47 schwarze 2353: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2354: sql = "INSERT INTO mlinks "
1.61 schwarze 2355: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2356: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2357: sql = "SELECT bits FROM names where pageid = ?";
2358: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2359: sql = "INSERT INTO names "
2360: "(bits,name,pageid) VALUES (?,?,?)";
2361: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2362: sql = "INSERT INTO keys "
2363: "(bits,key,pageid) VALUES (?,?,?)";
2364: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2365:
1.47 schwarze 2366: /*
2367: * When opening a new database, we can turn off
2368: * synchronous mode for much better performance.
2369: */
1.6 schwarze 2370:
1.103 schwarze 2371: if (real && SQLITE_OK != sqlite3_exec(db,
2372: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2373: exitcode = (int)MANDOCLEVEL_SYSERR;
2374: say(MANDOC_DB, "PRAGMA synchronous: %s",
2375: sqlite3_errmsg(db));
2376: sqlite3_close(db);
2377: return(0);
2378: }
1.11 schwarze 2379:
1.47 schwarze 2380: return(1);
2381: }
1.6 schwarze 2382:
1.47 schwarze 2383: static void *
1.106 espie 2384: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2385: {
1.6 schwarze 2386:
1.106 espie 2387: return(mandoc_calloc(n, sz));
1.47 schwarze 2388: }
1.2 schwarze 2389:
1.47 schwarze 2390: static void *
2391: hash_alloc(size_t sz, void *arg)
2392: {
1.28 schwarze 2393:
1.47 schwarze 2394: return(mandoc_malloc(sz));
2395: }
1.6 schwarze 2396:
1.47 schwarze 2397: static void
1.106 espie 2398: hash_free(void *p, void *arg)
1.47 schwarze 2399: {
1.26 schwarze 2400:
1.47 schwarze 2401: free(p);
2402: }
1.6 schwarze 2403:
1.47 schwarze 2404: static int
1.120 schwarze 2405: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2406: {
2407: static char startdir[PATH_MAX];
1.110 schwarze 2408: static int getcwd_status; /* 1 = ok, 2 = failure */
2409: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2410: char *cp;
1.6 schwarze 2411:
1.47 schwarze 2412: /*
1.110 schwarze 2413: * Remember the original working directory, if possible.
2414: * This will be needed if the second or a later directory
2415: * on the command line is given as a relative path.
2416: * Do not error out if the current directory is not
2417: * searchable: Maybe it won't be needed after all.
2418: */
2419: if (0 == getcwd_status) {
2420: if (NULL == getcwd(startdir, sizeof(startdir))) {
2421: getcwd_status = 2;
2422: (void)strlcpy(startdir, strerror(errno),
2423: sizeof(startdir));
2424: } else
2425: getcwd_status = 1;
2426: }
2427:
2428: /*
2429: * We are leaving the old base directory.
2430: * Do not use it any longer, not even for messages.
2431: */
2432: *basedir = '\0';
2433:
2434: /*
2435: * If and only if the directory was changed earlier and
2436: * the next directory to process is given as a relative path,
2437: * first go back, or bail out if that is impossible.
1.47 schwarze 2438: */
1.110 schwarze 2439: if (chdir_status && '/' != *targetdir) {
2440: if (2 == getcwd_status) {
1.47 schwarze 2441: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2442: say("", "getcwd: %s", startdir);
1.47 schwarze 2443: return(0);
2444: }
1.110 schwarze 2445: if (-1 == chdir(startdir)) {
1.47 schwarze 2446: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2447: say("", "&chdir %s", startdir);
1.47 schwarze 2448: return(0);
1.2 schwarze 2449: }
2450: }
1.110 schwarze 2451:
2452: /*
2453: * Always resolve basedir to the canonicalized absolute
2454: * pathname and append a trailing slash, such that
2455: * we can reliably check whether files are inside.
2456: */
1.47 schwarze 2457: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2458: if (report_baddir || errno != ENOENT) {
2459: exitcode = (int)MANDOCLEVEL_BADARG;
2460: say("", "&%s: realpath", targetdir);
2461: }
1.47 schwarze 2462: return(0);
2463: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2464: if (report_baddir || errno != ENOENT) {
2465: exitcode = (int)MANDOCLEVEL_BADARG;
2466: say("", "&chdir");
2467: }
1.47 schwarze 2468: return(0);
1.109 schwarze 2469: }
1.110 schwarze 2470: chdir_status = 1;
1.109 schwarze 2471: cp = strchr(basedir, '\0');
2472: if ('/' != cp[-1]) {
2473: if (cp - basedir >= PATH_MAX - 1) {
2474: exitcode = (int)MANDOCLEVEL_SYSERR;
2475: say("", "Filename too long");
2476: return(0);
2477: }
2478: *cp++ = '/';
2479: *cp = '\0';
1.47 schwarze 2480: }
2481: return(1);
1.2 schwarze 2482: }
2483:
2484: static void
1.47 schwarze 2485: say(const char *file, const char *format, ...)
1.2 schwarze 2486: {
1.47 schwarze 2487: va_list ap;
1.80 schwarze 2488: int use_errno;
1.2 schwarze 2489:
1.47 schwarze 2490: if ('\0' != *basedir)
2491: fprintf(stderr, "%s", basedir);
2492: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2493: fputc('/', stderr);
1.47 schwarze 2494: if ('\0' != *file)
2495: fprintf(stderr, "%s", file);
1.31 schwarze 2496:
1.80 schwarze 2497: use_errno = 1;
2498: if (NULL != format) {
2499: switch (*format) {
1.98 schwarze 2500: case '&':
1.80 schwarze 2501: format++;
2502: break;
1.98 schwarze 2503: case '\0':
1.80 schwarze 2504: format = NULL;
2505: break;
2506: default:
2507: use_errno = 0;
2508: break;
2509: }
2510: }
2511: if (NULL != format) {
2512: if ('\0' != *basedir || '\0' != *file)
2513: fputs(": ", stderr);
2514: va_start(ap, format);
2515: vfprintf(stderr, format, ap);
2516: va_end(ap);
2517: }
2518: if (use_errno) {
2519: if ('\0' != *basedir || '\0' != *file || NULL != format)
2520: fputs(": ", stderr);
1.47 schwarze 2521: perror(NULL);
1.80 schwarze 2522: } else
2523: fputc('\n', stderr);
1.1 schwarze 2524: }