Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.154
1.154 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.153 2015/10/12 00:32:37 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>
1.151 schwarze 24: #include <err.h>
1.34 schwarze 25: #include <errno.h>
1.1 schwarze 26: #include <fcntl.h>
1.47 schwarze 27: #include <fts.h>
1.1 schwarze 28: #include <getopt.h>
1.44 schwarze 29: #include <limits.h>
1.47 schwarze 30: #include <stddef.h>
1.1 schwarze 31: #include <stdio.h>
32: #include <stdint.h>
33: #include <stdlib.h>
34: #include <string.h>
1.14 schwarze 35: #include <unistd.h>
1.1 schwarze 36:
1.47 schwarze 37: #include <ohash.h>
38: #include <sqlite3.h>
39:
1.141 schwarze 40: #include "mandoc_aux.h"
41: #include "mandoc.h"
42: #include "roff.h"
1.47 schwarze 43: #include "mdoc.h"
1.1 schwarze 44: #include "man.h"
1.140 schwarze 45: #include "manconf.h"
1.47 schwarze 46: #include "mansearch.h"
1.1 schwarze 47:
1.68 schwarze 48: extern int mansearch_keymax;
49: extern const char *const mansearch_keynames[];
50:
1.47 schwarze 51: #define SQL_EXEC(_v) \
52: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 53: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 54: #define SQL_BIND_TEXT(_s, _i, _v) \
55: if (SQLITE_OK != sqlite3_bind_text \
56: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 57: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 58: #define SQL_BIND_INT(_s, _i, _v) \
59: if (SQLITE_OK != sqlite3_bind_int \
60: ((_s), (_i)++, (_v))) \
1.91 schwarze 61: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 62: #define SQL_BIND_INT64(_s, _i, _v) \
63: if (SQLITE_OK != sqlite3_bind_int64 \
64: ((_s), (_i)++, (_v))) \
1.91 schwarze 65: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 66: #define SQL_STEP(_s) \
67: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 68: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 69:
1.47 schwarze 70: enum op {
71: OP_DEFAULT = 0, /* new dbs from dir list or default config */
72: OP_CONFFILE, /* new databases from custom config file */
73: OP_UPDATE, /* delete/add entries in existing database */
74: OP_DELETE, /* delete entries from existing database */
75: OP_TEST /* change no databases, report potential problems */
76: };
1.11 schwarze 77:
1.47 schwarze 78: struct str {
79: const struct mpage *mpage; /* if set, the owning parse */
80: uint64_t mask; /* bitmask in sequence */
1.131 schwarze 81: char key[]; /* rendered text */
1.28 schwarze 82: };
83:
1.47 schwarze 84: struct inodev {
85: ino_t st_ino;
86: dev_t st_dev;
87: };
1.28 schwarze 88:
1.47 schwarze 89: struct mpage {
90: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 91: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 92: char *sec; /* section from file content */
93: char *arch; /* architecture from file content */
94: char *title; /* title from file content */
95: char *desc; /* description from file content */
96: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 97: int form; /* format from file content */
1.129 schwarze 98: int name_head_done;
1.28 schwarze 99: };
100:
1.47 schwarze 101: struct mlink {
102: char file[PATH_MAX]; /* filename rel. to manpath */
103: char *dsec; /* section from directory */
104: char *arch; /* architecture from directory */
105: char *name; /* name from file name (not empty) */
106: char *fsec; /* section from file name suffix */
107: struct mlink *next; /* singly linked list */
1.75 schwarze 108: struct mpage *mpage; /* parent */
1.116 schwarze 109: int dform; /* format from directory */
110: int fform; /* format from file name suffix */
1.81 schwarze 111: int gzip; /* filename has a .gz suffix */
1.2 schwarze 112: };
113:
1.47 schwarze 114: enum stmt {
115: STMT_DELETE_PAGE = 0, /* delete mpage */
116: STMT_INSERT_PAGE, /* insert mpage */
117: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 118: STMT_INSERT_NAME, /* insert name */
1.130 schwarze 119: STMT_SELECT_NAME, /* retrieve existing name flags */
1.47 schwarze 120: STMT_INSERT_KEY, /* insert parsed key */
121: STMT__MAX
1.1 schwarze 122: };
123:
1.143 schwarze 124: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.142 schwarze 125: const struct roff_node *);
1.1 schwarze 126:
1.19 schwarze 127: struct mdoc_handler {
1.47 schwarze 128: mdoc_fp fp; /* optional handler */
129: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 130: };
131:
1.47 schwarze 132: static void dbclose(int);
1.131 schwarze 133: static void dbadd(struct mpage *);
1.75 schwarze 134: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 135: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 136: static int dbopen(int);
137: static void dbprune(void);
138: static void filescan(const char *);
139: static void *hash_alloc(size_t, void *);
1.106 espie 140: static void hash_free(void *, void *);
141: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 142: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 143: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 144: static void mlink_free(struct mlink *);
145: static void mlinks_undupe(struct mpage *);
146: static void mpages_free(void);
1.131 schwarze 147: static void mpages_merge(struct mparse *);
1.87 schwarze 148: static void names_check(void);
1.81 schwarze 149: static void parse_cat(struct mpage *, int);
1.143 schwarze 150: static void parse_man(struct mpage *, const struct roff_meta *,
1.142 schwarze 151: const struct roff_node *);
1.143 schwarze 152: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.142 schwarze 153: const struct roff_node *);
1.143 schwarze 154: static int parse_mdoc_body(struct mpage *, const struct roff_meta *,
1.142 schwarze 155: const struct roff_node *);
1.143 schwarze 156: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.142 schwarze 157: const struct roff_node *);
1.143 schwarze 158: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.142 schwarze 159: const struct roff_node *);
160: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.143 schwarze 161: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.142 schwarze 162: const struct roff_node *);
1.143 schwarze 163: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.142 schwarze 164: const struct roff_node *);
1.143 schwarze 165: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.142 schwarze 166: const struct roff_node *);
1.143 schwarze 167: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.142 schwarze 168: const struct roff_node *);
1.143 schwarze 169: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.142 schwarze 170: const struct roff_node *);
1.143 schwarze 171: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.142 schwarze 172: const struct roff_node *);
1.69 schwarze 173: static void putkey(const struct mpage *, char *, uint64_t);
1.131 schwarze 174: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.47 schwarze 175: static void putmdockey(const struct mpage *,
1.142 schwarze 176: const struct roff_node *, uint64_t);
1.131 schwarze 177: static int render_string(char **, size_t *);
1.47 schwarze 178: static void say(const char *, const char *, ...);
1.120 schwarze 179: static int set_basedir(const char *, int);
1.47 schwarze 180: static int treescan(void);
181: static size_t utf8(unsigned int, char [7]);
182:
1.151 schwarze 183: extern char *__progname;
184:
1.72 schwarze 185: static char tempfilename[32];
1.59 schwarze 186: static int nodb; /* no database changes */
1.73 schwarze 187: static int mparse_options; /* abort the parse early */
1.98 schwarze 188: static int use_all; /* use all found files */
189: static int debug; /* print what we're doing */
190: static int warnings; /* warn about crap */
1.52 schwarze 191: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 192: static int exitcode; /* to be returned by main */
1.98 schwarze 193: static enum op op; /* operational mode */
1.47 schwarze 194: static char basedir[PATH_MAX]; /* current base directory */
1.131 schwarze 195: static struct mchars *mchars; /* table of named characters */
1.47 schwarze 196: static struct ohash mpages; /* table of distinct manual pages */
197: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 198: static struct ohash names; /* table of all names */
1.47 schwarze 199: static struct ohash strings; /* table of all strings */
200: static sqlite3 *db = NULL; /* current database */
201: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 202: static uint64_t name_mask;
1.47 schwarze 203:
1.19 schwarze 204: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 205: { NULL, 0 }, /* Ap */
206: { NULL, 0 }, /* Dd */
207: { NULL, 0 }, /* Dt */
208: { NULL, 0 }, /* Os */
209: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
210: { parse_mdoc_head, TYPE_Ss }, /* Ss */
211: { NULL, 0 }, /* Pp */
212: { NULL, 0 }, /* D1 */
213: { NULL, 0 }, /* Dl */
214: { NULL, 0 }, /* Bd */
215: { NULL, 0 }, /* Ed */
216: { NULL, 0 }, /* Bl */
217: { NULL, 0 }, /* El */
218: { NULL, 0 }, /* It */
219: { NULL, 0 }, /* Ad */
220: { NULL, TYPE_An }, /* An */
221: { NULL, TYPE_Ar }, /* Ar */
222: { NULL, TYPE_Cd }, /* Cd */
223: { NULL, TYPE_Cm }, /* Cm */
224: { NULL, TYPE_Dv }, /* Dv */
225: { NULL, TYPE_Er }, /* Er */
226: { NULL, TYPE_Ev }, /* Ev */
227: { NULL, 0 }, /* Ex */
228: { NULL, TYPE_Fa }, /* Fa */
229: { parse_mdoc_Fd, 0 }, /* Fd */
230: { NULL, TYPE_Fl }, /* Fl */
231: { parse_mdoc_Fn, 0 }, /* Fn */
232: { NULL, TYPE_Ft }, /* Ft */
233: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 234: { NULL, TYPE_In }, /* In */
1.47 schwarze 235: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 236: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 237: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 238: { NULL, 0 }, /* Op */
239: { NULL, 0 }, /* Ot */
240: { NULL, TYPE_Pa }, /* Pa */
241: { NULL, 0 }, /* Rv */
1.49 schwarze 242: { NULL, TYPE_St }, /* St */
1.47 schwarze 243: { NULL, TYPE_Va }, /* Va */
244: { parse_mdoc_body, TYPE_Va }, /* Vt */
245: { parse_mdoc_Xr, 0 }, /* Xr */
246: { NULL, 0 }, /* %A */
247: { NULL, 0 }, /* %B */
248: { NULL, 0 }, /* %D */
249: { NULL, 0 }, /* %I */
250: { NULL, 0 }, /* %J */
251: { NULL, 0 }, /* %N */
252: { NULL, 0 }, /* %O */
253: { NULL, 0 }, /* %P */
254: { NULL, 0 }, /* %R */
255: { NULL, 0 }, /* %T */
256: { NULL, 0 }, /* %V */
257: { NULL, 0 }, /* Ac */
258: { NULL, 0 }, /* Ao */
259: { NULL, 0 }, /* Aq */
260: { NULL, TYPE_At }, /* At */
261: { NULL, 0 }, /* Bc */
262: { NULL, 0 }, /* Bf */
263: { NULL, 0 }, /* Bo */
264: { NULL, 0 }, /* Bq */
265: { NULL, TYPE_Bsx }, /* Bsx */
266: { NULL, TYPE_Bx }, /* Bx */
267: { NULL, 0 }, /* Db */
268: { NULL, 0 }, /* Dc */
269: { NULL, 0 }, /* Do */
270: { NULL, 0 }, /* Dq */
271: { NULL, 0 }, /* Ec */
272: { NULL, 0 }, /* Ef */
273: { NULL, TYPE_Em }, /* Em */
274: { NULL, 0 }, /* Eo */
275: { NULL, TYPE_Fx }, /* Fx */
276: { NULL, TYPE_Ms }, /* Ms */
277: { NULL, 0 }, /* No */
278: { NULL, 0 }, /* Ns */
279: { NULL, TYPE_Nx }, /* Nx */
280: { NULL, TYPE_Ox }, /* Ox */
281: { NULL, 0 }, /* Pc */
282: { NULL, 0 }, /* Pf */
283: { NULL, 0 }, /* Po */
284: { NULL, 0 }, /* Pq */
285: { NULL, 0 }, /* Qc */
286: { NULL, 0 }, /* Ql */
287: { NULL, 0 }, /* Qo */
288: { NULL, 0 }, /* Qq */
289: { NULL, 0 }, /* Re */
290: { NULL, 0 }, /* Rs */
291: { NULL, 0 }, /* Sc */
292: { NULL, 0 }, /* So */
293: { NULL, 0 }, /* Sq */
294: { NULL, 0 }, /* Sm */
295: { NULL, 0 }, /* Sx */
296: { NULL, TYPE_Sy }, /* Sy */
297: { NULL, TYPE_Tn }, /* Tn */
298: { NULL, 0 }, /* Ux */
299: { NULL, 0 }, /* Xc */
300: { NULL, 0 }, /* Xo */
1.128 schwarze 301: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 302: { NULL, 0 }, /* Fc */
303: { NULL, 0 }, /* Oo */
304: { NULL, 0 }, /* Oc */
305: { NULL, 0 }, /* Bk */
306: { NULL, 0 }, /* Ek */
307: { NULL, 0 }, /* Bt */
308: { NULL, 0 }, /* Hf */
309: { NULL, 0 }, /* Fr */
310: { NULL, 0 }, /* Ud */
311: { NULL, TYPE_Lb }, /* Lb */
312: { NULL, 0 }, /* Lp */
313: { NULL, TYPE_Lk }, /* Lk */
314: { NULL, TYPE_Mt }, /* Mt */
315: { NULL, 0 }, /* Brq */
316: { NULL, 0 }, /* Bro */
317: { NULL, 0 }, /* Brc */
318: { NULL, 0 }, /* %C */
319: { NULL, 0 }, /* Es */
320: { NULL, 0 }, /* En */
321: { NULL, TYPE_Dx }, /* Dx */
322: { NULL, 0 }, /* %Q */
323: { NULL, 0 }, /* br */
324: { NULL, 0 }, /* sp */
325: { NULL, 0 }, /* %U */
326: { NULL, 0 }, /* Ta */
1.121 schwarze 327: { NULL, 0 }, /* ll */
1.1 schwarze 328: };
329:
1.98 schwarze 330:
1.1 schwarze 331: int
1.3 schwarze 332: mandocdb(int argc, char *argv[])
1.1 schwarze 333: {
1.140 schwarze 334: struct manconf conf;
335: struct ohash_info mpages_info, mlinks_info;
336: struct mparse *mp;
337: const char *path_arg;
338: size_t j, sz;
1.47 schwarze 339: int ch, i;
340:
1.140 schwarze 341: memset(&conf, 0, sizeof(conf));
1.47 schwarze 342: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
343:
344: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 345: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
1.135 schwarze 346: mpages_info.free = mlinks_info.free = hash_free;
347: mpages_info.data = mlinks_info.data = NULL;
1.47 schwarze 348:
349: mpages_info.key_offset = offsetof(struct mpage, inodev);
350: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 351:
1.47 schwarze 352: /*
1.98 schwarze 353: * We accept a few different invocations.
1.47 schwarze 354: * The CHECKOP macro makes sure that invocation styles don't
355: * clobber each other.
356: */
357: #define CHECKOP(_op, _ch) do \
358: if (OP_DEFAULT != (_op)) { \
1.151 schwarze 359: warnx("-%c: Conflicting option", (_ch)); \
1.47 schwarze 360: goto usage; \
361: } while (/*CONSTCOND*/0)
1.10 schwarze 362:
1.47 schwarze 363: path_arg = NULL;
1.28 schwarze 364: op = OP_DEFAULT;
1.1 schwarze 365:
1.83 schwarze 366: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 367: switch (ch) {
1.98 schwarze 368: case 'a':
1.6 schwarze 369: use_all = 1;
370: break;
1.98 schwarze 371: case 'C':
1.47 schwarze 372: CHECKOP(op, ch);
373: path_arg = optarg;
1.28 schwarze 374: op = OP_CONFFILE;
1.25 schwarze 375: break;
1.98 schwarze 376: case 'D':
1.82 schwarze 377: debug++;
378: break;
1.98 schwarze 379: case 'd':
1.47 schwarze 380: CHECKOP(op, ch);
381: path_arg = optarg;
1.2 schwarze 382: op = OP_UPDATE;
1.1 schwarze 383: break;
1.98 schwarze 384: case 'n':
1.47 schwarze 385: nodb = 1;
386: break;
1.98 schwarze 387: case 'p':
1.83 schwarze 388: warnings = 1;
389: break;
1.98 schwarze 390: case 'Q':
1.73 schwarze 391: mparse_options |= MPARSE_QUICK;
1.59 schwarze 392: break;
1.98 schwarze 393: case 'T':
1.52 schwarze 394: if (strcmp(optarg, "utf8")) {
1.151 schwarze 395: warnx("-T%s: Unsupported output format",
396: optarg);
1.52 schwarze 397: goto usage;
398: }
399: write_utf8 = 1;
400: break;
1.98 schwarze 401: case 't':
1.47 schwarze 402: CHECKOP(op, ch);
1.28 schwarze 403: dup2(STDOUT_FILENO, STDERR_FILENO);
404: op = OP_TEST;
1.47 schwarze 405: nodb = warnings = 1;
1.28 schwarze 406: break;
1.98 schwarze 407: case 'u':
1.47 schwarze 408: CHECKOP(op, ch);
409: path_arg = optarg;
1.1 schwarze 410: op = OP_DELETE;
411: break;
1.98 schwarze 412: case 'v':
1.83 schwarze 413: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 414: break;
1.1 schwarze 415: default:
1.28 schwarze 416: goto usage;
1.1 schwarze 417: }
418:
419: argc -= optind;
420: argv += optind;
421:
1.28 schwarze 422: if (OP_CONFFILE == op && argc > 0) {
1.151 schwarze 423: warnx("-C: Too many arguments");
1.28 schwarze 424: goto usage;
425: }
426:
1.47 schwarze 427: exitcode = (int)MANDOCLEVEL_OK;
1.131 schwarze 428: mchars = mchars_alloc();
1.137 schwarze 429: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.131 schwarze 430: mchars, NULL);
1.47 schwarze 431: ohash_init(&mpages, 6, &mpages_info);
432: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 433:
1.47 schwarze 434: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 435:
1.47 schwarze 436: /*
1.109 schwarze 437: * Most of these deal with a specific directory.
1.97 schwarze 438: * Jump into that directory first.
1.47 schwarze 439: */
1.120 schwarze 440: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 441: goto out;
1.97 schwarze 442:
1.95 schwarze 443: if (dbopen(1)) {
1.97 schwarze 444: /*
445: * The existing database is usable. Process
446: * all files specified on the command-line.
447: */
448: use_all = 1;
1.95 schwarze 449: for (i = 0; i < argc; i++)
450: filescan(argv[i]);
451: if (OP_TEST != op)
452: dbprune();
453: } else {
454: /*
455: * Database missing or corrupt.
456: * Recreate from scratch.
457: */
1.97 schwarze 458: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 459: op = OP_DEFAULT;
460: if (0 == treescan())
461: goto out;
462: if (0 == dbopen(0))
463: goto out;
464: }
1.47 schwarze 465: if (OP_DELETE != op)
1.131 schwarze 466: mpages_merge(mp);
1.95 schwarze 467: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 468: } else {
469: /*
470: * If we have arguments, use them as our manpaths.
471: * If we don't, grok from manpath(1) or however else
1.140 schwarze 472: * manconf_parse() wants to do it.
1.47 schwarze 473: */
474: if (argc > 0) {
1.140 schwarze 475: conf.manpath.paths = mandoc_reallocarray(NULL,
1.101 schwarze 476: argc, sizeof(char *));
1.140 schwarze 477: conf.manpath.sz = (size_t)argc;
1.47 schwarze 478: for (i = 0; i < argc; i++)
1.140 schwarze 479: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.47 schwarze 480: } else
1.140 schwarze 481: manconf_parse(&conf, path_arg, NULL, NULL);
1.84 schwarze 482:
1.140 schwarze 483: if (conf.manpath.sz == 0) {
1.84 schwarze 484: exitcode = (int)MANDOCLEVEL_BADARG;
485: say("", "Empty manpath");
486: }
1.2 schwarze 487:
1.47 schwarze 488: /*
489: * First scan the tree rooted at a base directory, then
490: * build a new database and finally move it into place.
491: * Ignore zero-length directories and strip trailing
492: * slashes.
493: */
1.140 schwarze 494: for (j = 0; j < conf.manpath.sz; j++) {
495: sz = strlen(conf.manpath.paths[j]);
496: if (sz && conf.manpath.paths[j][sz - 1] == '/')
497: conf.manpath.paths[j][--sz] = '\0';
1.47 schwarze 498: if (0 == sz)
499: continue;
1.2 schwarze 500:
1.47 schwarze 501: if (j) {
502: ohash_init(&mpages, 6, &mpages_info);
503: ohash_init(&mlinks, 6, &mlinks_info);
504: }
1.1 schwarze 505:
1.140 schwarze 506: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.120 schwarze 507: continue;
1.47 schwarze 508: if (0 == treescan())
1.120 schwarze 509: continue;
1.47 schwarze 510: if (0 == dbopen(0))
1.120 schwarze 511: continue;
1.41 deraadt 512:
1.131 schwarze 513: mpages_merge(mp);
1.104 schwarze 514: if (warnings && !nodb &&
1.87 schwarze 515: ! (MPARSE_QUICK & mparse_options))
516: names_check();
1.47 schwarze 517: dbclose(0);
1.28 schwarze 518:
1.140 schwarze 519: if (j + 1 < conf.manpath.sz) {
1.47 schwarze 520: mpages_free();
521: ohash_delete(&mpages);
522: ohash_delete(&mlinks);
523: }
1.2 schwarze 524: }
1.47 schwarze 525: }
526: out:
1.140 schwarze 527: manconf_free(&conf);
1.123 schwarze 528: mparse_free(mp);
1.131 schwarze 529: mchars_free(mchars);
1.47 schwarze 530: mpages_free();
531: ohash_delete(&mpages);
532: ohash_delete(&mlinks);
1.150 schwarze 533: return exitcode;
1.47 schwarze 534: usage:
1.83 schwarze 535: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
536: " %s [-aDnpQ] [-Tutf8] dir ...\n"
537: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
538: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 539: " %s [-Q] -t file ...\n",
1.151 schwarze 540: __progname, __progname, __progname,
541: __progname, __progname);
1.1 schwarze 542:
1.150 schwarze 543: return (int)MANDOCLEVEL_BADARG;
1.47 schwarze 544: }
1.1 schwarze 545:
1.47 schwarze 546: /*
547: * Scan a directory tree rooted at "basedir" for manpages.
548: * We use fts(), scanning directory parts along the way for clues to our
549: * section and architecture.
550: *
551: * If use_all has been specified, grok all files.
552: * If not, sanitise paths to the following:
553: *
1.98 schwarze 554: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 555: * or
556: * [./]cat<section>[/<arch>]/<name>.0
557: *
558: * TODO: accomodate for multi-language directories.
559: */
560: static int
561: treescan(void)
562: {
1.96 schwarze 563: char buf[PATH_MAX];
1.47 schwarze 564: FTS *f;
565: FTSENT *ff;
566: struct mlink *mlink;
1.81 schwarze 567: int dform, gzip;
1.51 schwarze 568: char *dsec, *arch, *fsec, *cp;
569: const char *path;
1.47 schwarze 570: const char *argv[2];
1.1 schwarze 571:
1.47 schwarze 572: argv[0] = ".";
573: argv[1] = (char *)NULL;
1.1 schwarze 574:
1.96 schwarze 575: f = fts_open((char * const *)argv,
576: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.153 schwarze 577: if (f == NULL) {
1.47 schwarze 578: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 579: say("", "&fts_open");
1.150 schwarze 580: return 0;
1.47 schwarze 581: }
1.2 schwarze 582:
1.47 schwarze 583: dsec = arch = NULL;
584: dform = FORM_NONE;
1.2 schwarze 585:
1.153 schwarze 586: while ((ff = fts_read(f)) != NULL) {
1.47 schwarze 587: path = ff->fts_path + 2;
1.96 schwarze 588: switch (ff->fts_info) {
589:
590: /*
591: * Symbolic links require various sanity checks,
592: * then get handled just like regular files.
593: */
1.98 schwarze 594: case FTS_SL:
1.153 schwarze 595: if (realpath(path, buf) == NULL) {
1.96 schwarze 596: if (warnings)
597: say(path, "&realpath");
598: continue;
599: }
600: if (strstr(buf, basedir) != buf) {
601: if (warnings) say("",
602: "%s: outside base directory", buf);
603: continue;
604: }
605: /* Use logical inode to avoid mpages dupe. */
1.153 schwarze 606: if (stat(path, ff->fts_statp) == -1) {
1.96 schwarze 607: if (warnings)
608: say(path, "&stat");
609: continue;
610: }
611: /* FALLTHROUGH */
612:
1.15 schwarze 613: /*
1.47 schwarze 614: * If we're a regular file, add an mlink by using the
615: * stored directory data and handling the filename.
1.15 schwarze 616: */
1.98 schwarze 617: case FTS_F:
1.153 schwarze 618: if ( ! strcmp(path, MANDOC_DB))
1.47 schwarze 619: continue;
620: if ( ! use_all && ff->fts_level < 2) {
621: if (warnings)
622: say(path, "Extraneous file");
623: continue;
1.81 schwarze 624: }
625: gzip = 0;
626: fsec = NULL;
1.153 schwarze 627: while (fsec == NULL) {
1.81 schwarze 628: fsec = strrchr(ff->fts_name, '.');
1.153 schwarze 629: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.81 schwarze 630: break;
631: gzip = 1;
632: *fsec = '\0';
633: fsec = NULL;
634: }
1.153 schwarze 635: if (fsec == NULL) {
1.47 schwarze 636: if ( ! use_all) {
637: if (warnings)
638: say(path,
639: "No filename suffix");
640: continue;
641: }
1.153 schwarze 642: } else if ( ! strcmp(++fsec, "html")) {
1.47 schwarze 643: if (warnings)
644: say(path, "Skip html");
645: continue;
1.153 schwarze 646: } else if ( ! strcmp(fsec, "ps")) {
1.47 schwarze 647: if (warnings)
648: say(path, "Skip ps");
649: continue;
1.153 schwarze 650: } else if ( ! strcmp(fsec, "pdf")) {
1.47 schwarze 651: if (warnings)
652: say(path, "Skip pdf");
653: continue;
654: } else if ( ! use_all &&
1.153 schwarze 655: ((dform == FORM_SRC &&
1.138 schwarze 656: strncmp(fsec, dsec, strlen(dsec))) ||
1.153 schwarze 657: (dform == FORM_CAT && strcmp(fsec, "0")))) {
1.47 schwarze 658: if (warnings)
659: say(path, "Wrong filename suffix");
660: continue;
661: } else
662: fsec[-1] = '\0';
1.51 schwarze 663:
1.47 schwarze 664: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 665: if (strlcpy(mlink->file, path,
666: sizeof(mlink->file)) >=
667: sizeof(mlink->file)) {
668: say(path, "Filename too long");
669: free(mlink);
670: continue;
671: }
1.47 schwarze 672: mlink->dform = dform;
1.51 schwarze 673: mlink->dsec = dsec;
674: mlink->arch = arch;
675: mlink->name = ff->fts_name;
676: mlink->fsec = fsec;
1.81 schwarze 677: mlink->gzip = gzip;
1.47 schwarze 678: mlink_add(mlink, ff->fts_statp);
679: continue;
1.96 schwarze 680:
1.98 schwarze 681: case FTS_D:
682: case FTS_DP:
1.96 schwarze 683: break;
684:
685: default:
1.47 schwarze 686: if (warnings)
687: say(path, "Not a regular file");
688: continue;
689: }
690:
691: switch (ff->fts_level) {
1.98 schwarze 692: case 0:
1.47 schwarze 693: /* Ignore the root directory. */
694: break;
1.98 schwarze 695: case 1:
1.47 schwarze 696: /*
697: * This might contain manX/ or catX/.
698: * Try to infer this from the name.
699: * If we're not in use_all, enforce it.
700: */
701: cp = ff->fts_name;
1.153 schwarze 702: if (ff->fts_info == FTS_DP) {
703: dform = FORM_NONE;
704: dsec = NULL;
1.47 schwarze 705: break;
1.153 schwarze 706: }
1.15 schwarze 707:
1.153 schwarze 708: if ( ! strncmp(cp, "man", 3)) {
1.47 schwarze 709: dform = FORM_SRC;
710: dsec = cp + 3;
1.153 schwarze 711: } else if ( ! strncmp(cp, "cat", 3)) {
1.47 schwarze 712: dform = FORM_CAT;
713: dsec = cp + 3;
1.51 schwarze 714: } else {
715: dform = FORM_NONE;
716: dsec = NULL;
1.26 schwarze 717: }
1.47 schwarze 718:
1.153 schwarze 719: if (dsec != NULL || use_all)
1.47 schwarze 720: break;
721:
722: if (warnings)
723: say(path, "Unknown directory part");
724: fts_set(f, ff, FTS_SKIP);
725: break;
1.98 schwarze 726: case 2:
1.47 schwarze 727: /*
728: * Possibly our architecture.
729: * If we're descending, keep tabs on it.
730: */
1.153 schwarze 731: if (ff->fts_info != FTS_DP && dsec != NULL)
1.47 schwarze 732: arch = ff->fts_name;
1.51 schwarze 733: else
734: arch = NULL;
1.47 schwarze 735: break;
736: default:
1.153 schwarze 737: if (ff->fts_info == FTS_DP || use_all)
1.47 schwarze 738: break;
739: if (warnings)
740: say(path, "Extraneous directory part");
741: fts_set(f, ff, FTS_SKIP);
742: break;
1.14 schwarze 743: }
1.47 schwarze 744: }
745:
746: fts_close(f);
1.150 schwarze 747: return 1;
1.47 schwarze 748: }
1.1 schwarze 749:
1.47 schwarze 750: /*
751: * Add a file to the mlinks table.
752: * Do not verify that it's a "valid" looking manpage (we'll do that
753: * later).
754: *
755: * Try to infer the manual section, architecture, and page name from the
756: * path, assuming it looks like
757: *
1.98 schwarze 758: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 759: * or
760: * [./]cat<section>[/<arch>]/<name>.0
761: *
762: * See treescan() for the fts(3) version of this.
763: */
764: static void
765: filescan(const char *file)
766: {
767: char buf[PATH_MAX];
768: struct stat st;
769: struct mlink *mlink;
770: char *p, *start;
771:
772: assert(use_all);
773:
774: if (0 == strncmp(file, "./", 2))
775: file += 2;
776:
1.96 schwarze 777: /*
778: * We have to do lstat(2) before realpath(3) loses
779: * the information whether this is a symbolic link.
780: * We need to know that because for symbolic links,
781: * we want to use the orginal file name, while for
782: * regular files, we want to use the real path.
783: */
784: if (-1 == lstat(file, &st)) {
785: exitcode = (int)MANDOCLEVEL_BADARG;
786: say(file, "&lstat");
787: return;
788: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
789: exitcode = (int)MANDOCLEVEL_BADARG;
790: say(file, "Not a regular file");
791: return;
792: }
793:
794: /*
795: * We have to resolve the file name to the real path
796: * in any case for the base directory check.
797: */
1.47 schwarze 798: if (NULL == realpath(file, buf)) {
799: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 800: say(file, "&realpath");
1.47 schwarze 801: return;
1.63 schwarze 802: }
803:
1.109 schwarze 804: if (OP_TEST == op)
1.63 schwarze 805: start = buf;
1.109 schwarze 806: else if (strstr(buf, basedir) == buf)
807: start = buf + strlen(basedir);
1.63 schwarze 808: else {
1.47 schwarze 809: exitcode = (int)MANDOCLEVEL_BADARG;
810: say("", "%s: outside base directory", buf);
811: return;
1.63 schwarze 812: }
813:
1.96 schwarze 814: /*
815: * Now we are sure the file is inside our tree.
816: * If it is a symbolic link, ignore the real path
817: * and use the original name.
818: * This implies passing stuff like "cat1/../man1/foo.1"
819: * on the command line won't work. So don't do that.
820: * Note the stat(2) can still fail if the link target
821: * doesn't exist.
822: */
823: if (S_IFLNK & st.st_mode) {
824: if (-1 == stat(buf, &st)) {
825: exitcode = (int)MANDOCLEVEL_BADARG;
826: say(file, "&stat");
827: return;
828: }
1.100 schwarze 829: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
830: say(file, "Filename too long");
831: return;
832: }
1.109 schwarze 833: start = buf;
834: if (OP_TEST != op && strstr(buf, basedir) == buf)
835: start += strlen(basedir);
1.1 schwarze 836: }
1.63 schwarze 837:
1.47 schwarze 838: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 839: mlink->dform = FORM_NONE;
1.100 schwarze 840: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
841: sizeof(mlink->file)) {
842: say(start, "Filename too long");
1.134 schwarze 843: free(mlink);
1.100 schwarze 844: return;
845: }
1.1 schwarze 846:
1.10 schwarze 847: /*
1.47 schwarze 848: * First try to guess our directory structure.
849: * If we find a separator, try to look for man* or cat*.
850: * If we find one of these and what's underneath is a directory,
851: * assume it's an architecture.
1.10 schwarze 852: */
1.47 schwarze 853: if (NULL != (p = strchr(start, '/'))) {
854: *p++ = '\0';
855: if (0 == strncmp(start, "man", 3)) {
856: mlink->dform = FORM_SRC;
1.51 schwarze 857: mlink->dsec = start + 3;
1.47 schwarze 858: } else if (0 == strncmp(start, "cat", 3)) {
859: mlink->dform = FORM_CAT;
1.51 schwarze 860: mlink->dsec = start + 3;
1.47 schwarze 861: }
1.10 schwarze 862:
1.47 schwarze 863: start = p;
864: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
865: *p++ = '\0';
1.51 schwarze 866: mlink->arch = start;
1.47 schwarze 867: start = p;
1.41 deraadt 868: }
1.47 schwarze 869: }
1.7 schwarze 870:
1.47 schwarze 871: /*
872: * Now check the file suffix.
873: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
874: */
875: p = strrchr(start, '\0');
876: while (p-- > start && '/' != *p && '.' != *p)
877: /* Loop. */ ;
878:
879: if ('.' == *p) {
880: *p++ = '\0';
1.51 schwarze 881: mlink->fsec = p;
1.47 schwarze 882: }
1.41 deraadt 883:
1.47 schwarze 884: /*
885: * Now try to parse the name.
886: * Use the filename portion of the path.
887: */
888: mlink->name = start;
889: if (NULL != (p = strrchr(start, '/'))) {
890: mlink->name = p + 1;
891: *p = '\0';
892: }
893: mlink_add(mlink, &st);
894: }
1.2 schwarze 895:
1.47 schwarze 896: static void
897: mlink_add(struct mlink *mlink, const struct stat *st)
898: {
899: struct inodev inodev;
900: struct mpage *mpage;
901: unsigned int slot;
902:
903: assert(NULL != mlink->file);
904:
1.51 schwarze 905: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
906: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
907: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
908: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 909:
910: if ('0' == *mlink->fsec) {
911: free(mlink->fsec);
912: mlink->fsec = mandoc_strdup(mlink->dsec);
913: mlink->fform = FORM_CAT;
914: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
915: mlink->fform = FORM_SRC;
916: else
917: mlink->fform = FORM_NONE;
1.1 schwarze 918:
1.47 schwarze 919: slot = ohash_qlookup(&mlinks, mlink->file);
920: assert(NULL == ohash_find(&mlinks, slot));
921: ohash_insert(&mlinks, slot, mlink);
922:
1.139 schwarze 923: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.47 schwarze 924: inodev.st_ino = st->st_ino;
925: inodev.st_dev = st->st_dev;
926: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
927: sizeof(struct inodev), inodev.st_ino);
928: mpage = ohash_find(&mpages, slot);
929: if (NULL == mpage) {
930: mpage = mandoc_calloc(1, sizeof(struct mpage));
931: mpage->inodev.st_ino = inodev.st_ino;
932: mpage->inodev.st_dev = inodev.st_dev;
933: ohash_insert(&mpages, slot, mpage);
934: } else
935: mlink->next = mpage->mlinks;
936: mpage->mlinks = mlink;
1.75 schwarze 937: mlink->mpage = mpage;
1.47 schwarze 938: }
1.1 schwarze 939:
1.47 schwarze 940: static void
941: mlink_free(struct mlink *mlink)
942: {
1.1 schwarze 943:
1.47 schwarze 944: free(mlink->dsec);
945: free(mlink->arch);
946: free(mlink->name);
947: free(mlink->fsec);
948: free(mlink);
949: }
1.1 schwarze 950:
1.47 schwarze 951: static void
952: mpages_free(void)
953: {
954: struct mpage *mpage;
955: struct mlink *mlink;
956: unsigned int slot;
957:
958: mpage = ohash_first(&mpages, &slot);
959: while (NULL != mpage) {
960: while (NULL != (mlink = mpage->mlinks)) {
961: mpage->mlinks = mlink->next;
962: mlink_free(mlink);
963: }
964: free(mpage->sec);
965: free(mpage->arch);
966: free(mpage->title);
967: free(mpage->desc);
968: free(mpage);
969: mpage = ohash_next(&mpages, &slot);
970: }
971: }
1.1 schwarze 972:
1.47 schwarze 973: /*
974: * For each mlink to the mpage, check whether the path looks like
975: * it is formatted, and if it does, check whether a source manual
976: * exists by the same name, ignoring the suffix.
977: * If both conditions hold, drop the mlink.
978: */
979: static void
980: mlinks_undupe(struct mpage *mpage)
981: {
982: char buf[PATH_MAX];
983: struct mlink **prev;
984: struct mlink *mlink;
985: char *bufp;
986:
987: mpage->form = FORM_CAT;
988: prev = &mpage->mlinks;
989: while (NULL != (mlink = *prev)) {
990: if (FORM_CAT != mlink->dform) {
991: mpage->form = FORM_NONE;
992: goto nextlink;
993: }
1.100 schwarze 994: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 995: bufp = strstr(buf, "cat");
996: assert(NULL != bufp);
997: memcpy(bufp, "man", 3);
998: if (NULL != (bufp = strrchr(buf, '.')))
999: *++bufp = '\0';
1.100 schwarze 1000: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1001: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1002: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1003: goto nextlink;
1004: if (warnings)
1005: say(mlink->file, "Man source exists: %s", buf);
1006: if (use_all)
1007: goto nextlink;
1008: *prev = mlink->next;
1009: mlink_free(mlink);
1010: continue;
1011: nextlink:
1012: prev = &(*prev)->next;
1.1 schwarze 1013: }
1.47 schwarze 1014: }
1.1 schwarze 1015:
1.88 schwarze 1016: static void
1.50 schwarze 1017: mlink_check(struct mpage *mpage, struct mlink *mlink)
1018: {
1.88 schwarze 1019: struct str *str;
1020: unsigned int slot;
1.50 schwarze 1021:
1022: /*
1023: * Check whether the manual section given in a file
1024: * agrees with the directory where the file is located.
1025: * Some manuals have suffixes like (3p) on their
1026: * section number either inside the file or in the
1027: * directory name, some are linked into more than one
1028: * section, like encrypt(1) = makekey(8).
1029: */
1030:
1031: if (FORM_SRC == mpage->form &&
1.88 schwarze 1032: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1033: say(mlink->file, "Section \"%s\" manual in %s directory",
1034: mpage->sec, mlink->dsec);
1035:
1036: /*
1037: * Manual page directories exist for each kernel
1038: * architecture as returned by machine(1).
1039: * However, many manuals only depend on the
1040: * application architecture as returned by arch(1).
1041: * For example, some (2/ARM) manuals are shared
1042: * across the "armish" and "zaurus" kernel
1043: * architectures.
1044: * A few manuals are even shared across completely
1045: * different architectures, for example fdformat(1)
1046: * on amd64, i386, sparc, and sparc64.
1047: */
1048:
1.88 schwarze 1049: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1050: say(mlink->file, "Architecture \"%s\" manual in "
1051: "\"%s\" directory", mpage->arch, mlink->arch);
1052:
1.88 schwarze 1053: /*
1054: * XXX
1.90 schwarze 1055: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1056: */
1057:
1058: if (FORM_CAT == mpage->form)
1059: return;
1060:
1061: /*
1062: * Check whether this mlink
1063: * appears as a name in the NAME section.
1064: */
1.50 schwarze 1065:
1.90 schwarze 1066: slot = ohash_qlookup(&names, mlink->name);
1067: str = ohash_find(&names, slot);
1.88 schwarze 1068: assert(NULL != str);
1.90 schwarze 1069: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1070: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1071: }
1072:
1.47 schwarze 1073: /*
1074: * Run through the files in the global vector "mpages"
1075: * and add them to the database specified in "basedir".
1076: *
1077: * This handles the parsing scheme itself, using the cues of directory
1078: * and filename to determine whether the file is parsable or not.
1079: */
1080: static void
1.131 schwarze 1081: mpages_merge(struct mparse *mp)
1.47 schwarze 1082: {
1.70 schwarze 1083: char any[] = "any";
1.58 schwarze 1084: struct ohash_info str_info;
1.75 schwarze 1085: struct mpage *mpage, *mpage_dest;
1086: struct mlink *mlink, *mlink_dest;
1.144 schwarze 1087: struct roff_man *man;
1.75 schwarze 1088: char *sodest;
1.69 schwarze 1089: char *cp;
1.117 schwarze 1090: int fd;
1.58 schwarze 1091: unsigned int pslot;
1.47 schwarze 1092:
1093: str_info.alloc = hash_alloc;
1.106 espie 1094: str_info.calloc = hash_calloc;
1095: str_info.free = hash_free;
1.135 schwarze 1096: str_info.data = NULL;
1.47 schwarze 1097: str_info.key_offset = offsetof(struct str, key);
1098:
1.126 schwarze 1099: if ( ! nodb)
1.64 schwarze 1100: SQL_EXEC("BEGIN TRANSACTION");
1101:
1.47 schwarze 1102: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1103: while (mpage != NULL) {
1.47 schwarze 1104: mlinks_undupe(mpage);
1.136 schwarze 1105: if ((mlink = mpage->mlinks) == NULL) {
1.47 schwarze 1106: mpage = ohash_next(&mpages, &pslot);
1107: continue;
1108: }
1.1 schwarze 1109:
1.90 schwarze 1110: name_mask = NAME_MASK;
1111: ohash_init(&names, 4, &str_info);
1.47 schwarze 1112: ohash_init(&strings, 6, &str_info);
1113: mparse_reset(mp);
1114: man = NULL;
1.81 schwarze 1115: sodest = NULL;
1116:
1.136 schwarze 1117: mparse_open(mp, &fd, mlink->file);
1.117 schwarze 1118: if (fd == -1) {
1.136 schwarze 1119: say(mlink->file, "&open");
1.117 schwarze 1120: goto nextpage;
1.81 schwarze 1121: }
1.11 schwarze 1122:
1123: /*
1.137 schwarze 1124: * Interpret the file as mdoc(7) or man(7) source
1125: * code, unless it is known to be formatted.
1.11 schwarze 1126: */
1.136 schwarze 1127: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.137 schwarze 1128: mparse_readfd(mp, fd, mlink->file);
1.145 schwarze 1129: mparse_result(mp, &man, &sodest);
1.47 schwarze 1130: }
1.11 schwarze 1131:
1.126 schwarze 1132: if (sodest != NULL) {
1.75 schwarze 1133: mlink_dest = ohash_find(&mlinks,
1134: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1135: if (mlink_dest == NULL) {
1136: mandoc_asprintf(&cp, "%s.gz", sodest);
1137: mlink_dest = ohash_find(&mlinks,
1138: ohash_qlookup(&mlinks, cp));
1139: free(cp);
1140: }
1141: if (mlink_dest != NULL) {
1.75 schwarze 1142:
1143: /* The .so target exists. */
1144:
1145: mpage_dest = mlink_dest->mpage;
1146: while (1) {
1147: mlink->mpage = mpage_dest;
1148:
1149: /*
1150: * If the target was already
1151: * processed, add the links
1152: * to the database now.
1153: * Otherwise, this will
1154: * happen when we come
1155: * to the target.
1156: */
1157:
1.94 schwarze 1158: if (mpage_dest->pageid)
1.124 schwarze 1159: dbadd_mlink_name(mlink);
1.75 schwarze 1160:
1.126 schwarze 1161: if (mlink->next == NULL)
1.75 schwarze 1162: break;
1163: mlink = mlink->next;
1164: }
1165:
1166: /* Move all links to the target. */
1167:
1168: mlink->next = mlink_dest->next;
1169: mlink_dest->next = mpage->mlinks;
1170: mpage->mlinks = NULL;
1171: }
1.81 schwarze 1172: goto nextpage;
1.145 schwarze 1173: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.47 schwarze 1174: mpage->form = FORM_SRC;
1.146 schwarze 1175: mpage->sec = man->meta.msec;
1.112 schwarze 1176: mpage->sec = mandoc_strdup(
1.126 schwarze 1177: mpage->sec == NULL ? "" : mpage->sec);
1.146 schwarze 1178: mpage->arch = man->meta.arch;
1.47 schwarze 1179: mpage->arch = mandoc_strdup(
1.126 schwarze 1180: mpage->arch == NULL ? "" : mpage->arch);
1.146 schwarze 1181: mpage->title = mandoc_strdup(man->meta.title);
1.145 schwarze 1182: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.47 schwarze 1183: mpage->form = FORM_SRC;
1.146 schwarze 1184: mpage->sec = mandoc_strdup(man->meta.msec);
1.136 schwarze 1185: mpage->arch = mandoc_strdup(mlink->arch);
1.146 schwarze 1186: mpage->title = mandoc_strdup(man->meta.title);
1.11 schwarze 1187: } else {
1.47 schwarze 1188: mpage->form = FORM_CAT;
1.136 schwarze 1189: mpage->sec = mandoc_strdup(mlink->dsec);
1190: mpage->arch = mandoc_strdup(mlink->arch);
1191: mpage->title = mandoc_strdup(mlink->name);
1.1 schwarze 1192: }
1.54 schwarze 1193: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1194: if (*mpage->arch != '\0')
1195: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1196:
1.136 schwarze 1197: for ( ; mlink != NULL; mlink = mlink->next) {
1.54 schwarze 1198: if ('\0' != *mlink->dsec)
1199: putkey(mpage, mlink->dsec, TYPE_sec);
1200: if ('\0' != *mlink->fsec)
1201: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1202: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1203: any : mlink->arch, TYPE_arch);
1.90 schwarze 1204: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1205: }
1.41 deraadt 1206:
1.129 schwarze 1207: assert(mpage->desc == NULL);
1.145 schwarze 1208: if (man != NULL && man->macroset == MACROSET_MDOC)
1.146 schwarze 1209: parse_mdoc(mpage, &man->meta, man->first);
1.129 schwarze 1210: else if (man != NULL)
1.146 schwarze 1211: parse_man(mpage, &man->meta, man->first);
1.47 schwarze 1212: else
1.117 schwarze 1213: parse_cat(mpage, fd);
1.129 schwarze 1214: if (mpage->desc == NULL)
1.92 schwarze 1215: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1216:
1217: if (warnings && !use_all)
1218: for (mlink = mpage->mlinks; mlink;
1219: mlink = mlink->next)
1220: mlink_check(mpage, mlink);
1.6 schwarze 1221:
1.131 schwarze 1222: dbadd(mpage);
1.136 schwarze 1223: mlink = mpage->mlinks;
1.81 schwarze 1224:
1225: nextpage:
1.47 schwarze 1226: ohash_delete(&strings);
1.90 schwarze 1227: ohash_delete(&names);
1.47 schwarze 1228: mpage = ohash_next(&mpages, &pslot);
1229: }
1.64 schwarze 1230:
1231: if (0 == nodb)
1232: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1233: }
1234:
1235: static void
1236: names_check(void)
1237: {
1238: sqlite3_stmt *stmt;
1239: const char *name, *sec, *arch, *key;
1240:
1241: sqlite3_prepare_v2(db,
1242: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1243: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1244: "WHERE bits & ? AND NOT EXISTS ("
1245: "SELECT pageid FROM mlinks "
1.90 schwarze 1246: "WHERE mlinks.pageid == names.pageid "
1247: "AND mlinks.name == names.name"
1.87 schwarze 1248: ")"
1249: ") JOIN ("
1.99 schwarze 1250: "SELECT sec, arch, name, pageid FROM mlinks "
1251: "GROUP BY pageid"
1.87 schwarze 1252: ") USING (pageid);",
1253: -1, &stmt, NULL);
1254:
1.154 ! schwarze 1255: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.91 schwarze 1256: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1257:
1.154 ! schwarze 1258: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.113 schwarze 1259: name = (const char *)sqlite3_column_text(stmt, 0);
1260: sec = (const char *)sqlite3_column_text(stmt, 1);
1261: arch = (const char *)sqlite3_column_text(stmt, 2);
1262: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1263: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1264: '\0' == *arch ? "" : "/",
1265: '\0' == *arch ? "" : arch, key);
1266: }
1267: sqlite3_finalize(stmt);
1.47 schwarze 1268: }
1.6 schwarze 1269:
1.47 schwarze 1270: static void
1.81 schwarze 1271: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1272: {
1273: FILE *stream;
1274: char *line, *p, *title;
1275: size_t len, plen, titlesz;
1.1 schwarze 1276:
1.81 schwarze 1277: stream = (-1 == fd) ?
1278: fopen(mpage->mlinks->file, "r") :
1279: fdopen(fd, "r");
1280: if (NULL == stream) {
1.114 doug 1281: if (-1 != fd)
1282: close(fd);
1.47 schwarze 1283: if (warnings)
1.80 schwarze 1284: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1285: return;
1286: }
1.1 schwarze 1287:
1.47 schwarze 1288: /* Skip to first blank line. */
1.1 schwarze 1289:
1.47 schwarze 1290: while (NULL != (line = fgetln(stream, &len)))
1291: if ('\n' == *line)
1292: break;
1.1 schwarze 1293:
1.47 schwarze 1294: /*
1295: * Assume the first line that is not indented
1296: * is the first section header. Skip to it.
1297: */
1.1 schwarze 1298:
1.47 schwarze 1299: while (NULL != (line = fgetln(stream, &len)))
1300: if ('\n' != *line && ' ' != *line)
1301: break;
1.98 schwarze 1302:
1.47 schwarze 1303: /*
1304: * Read up until the next section into a buffer.
1305: * Strip the leading and trailing newline from each read line,
1306: * appending a trailing space.
1307: * Ignore empty (whitespace-only) lines.
1308: */
1.28 schwarze 1309:
1.47 schwarze 1310: titlesz = 0;
1311: title = NULL;
1.38 schwarze 1312:
1.47 schwarze 1313: while (NULL != (line = fgetln(stream, &len))) {
1314: if (' ' != *line || '\n' != line[len - 1])
1315: break;
1316: while (len > 0 && isspace((unsigned char)*line)) {
1317: line++;
1318: len--;
1319: }
1320: if (1 == len)
1321: continue;
1322: title = mandoc_realloc(title, titlesz + len);
1323: memcpy(title + titlesz, line, len);
1324: titlesz += len;
1325: title[titlesz - 1] = ' ';
1326: }
1.28 schwarze 1327:
1.47 schwarze 1328: /*
1329: * If no page content can be found, or the input line
1330: * is already the next section header, or there is no
1331: * trailing newline, reuse the page title as the page
1332: * description.
1333: */
1.1 schwarze 1334:
1.47 schwarze 1335: if (NULL == title || '\0' == *title) {
1336: if (warnings)
1337: say(mpage->mlinks->file,
1338: "Cannot find NAME section");
1339: fclose(stream);
1340: free(title);
1341: return;
1342: }
1.24 schwarze 1343:
1.47 schwarze 1344: title = mandoc_realloc(title, titlesz + 1);
1345: title[titlesz] = '\0';
1.24 schwarze 1346:
1.47 schwarze 1347: /*
1348: * Skip to the first dash.
1349: * Use the remaining line as the description (no more than 70
1350: * bytes).
1351: */
1.28 schwarze 1352:
1.47 schwarze 1353: if (NULL != (p = strstr(title, "- "))) {
1354: for (p += 2; ' ' == *p || '\b' == *p; p++)
1355: /* Skip to next word. */ ;
1356: } else {
1357: if (warnings)
1358: say(mpage->mlinks->file,
1359: "No dash in title line");
1360: p = title;
1361: }
1.1 schwarze 1362:
1.47 schwarze 1363: plen = strlen(p);
1.1 schwarze 1364:
1.47 schwarze 1365: /* Strip backspace-encoding from line. */
1.1 schwarze 1366:
1.47 schwarze 1367: while (NULL != (line = memchr(p, '\b', plen))) {
1368: len = line - p;
1369: if (0 == len) {
1370: memmove(line, line + 1, plen--);
1371: continue;
1.98 schwarze 1372: }
1.47 schwarze 1373: memmove(line - 1, line + 1, plen - len);
1374: plen -= 2;
1375: }
1.1 schwarze 1376:
1.47 schwarze 1377: mpage->desc = mandoc_strdup(p);
1378: fclose(stream);
1379: free(title);
1380: }
1.16 schwarze 1381:
1.47 schwarze 1382: /*
1383: * Put a type/word pair into the word database for this particular file.
1384: */
1385: static void
1.69 schwarze 1386: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1387: {
1.69 schwarze 1388: char *cp;
1.37 schwarze 1389:
1.47 schwarze 1390: assert(NULL != value);
1.69 schwarze 1391: if (TYPE_arch == type)
1392: for (cp = value; *cp; cp++)
1393: if (isupper((unsigned char)*cp))
1394: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1395: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1396: }
1397:
1398: /*
1.47 schwarze 1399: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1400: */
1401: static void
1.47 schwarze 1402: putmdockey(const struct mpage *mpage,
1.142 schwarze 1403: const struct roff_node *n, uint64_t m)
1.2 schwarze 1404: {
1.16 schwarze 1405:
1.47 schwarze 1406: for ( ; NULL != n; n = n->next) {
1407: if (NULL != n->child)
1408: putmdockey(mpage, n->child, m);
1.141 schwarze 1409: if (n->type == ROFFT_TEXT)
1.47 schwarze 1410: putkey(mpage, n->string, m);
1411: }
1412: }
1.16 schwarze 1413:
1.47 schwarze 1414: static void
1.143 schwarze 1415: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1416: const struct roff_node *n)
1.47 schwarze 1417: {
1.142 schwarze 1418: const struct roff_node *head, *body;
1.78 schwarze 1419: char *start, *title;
1.47 schwarze 1420: char byte;
1.78 schwarze 1421: size_t sz;
1.16 schwarze 1422:
1.47 schwarze 1423: if (NULL == n)
1424: return;
1.16 schwarze 1425:
1.47 schwarze 1426: /*
1427: * We're only searching for one thing: the first text child in
1428: * the BODY of a NAME section. Since we don't keep track of
1429: * sections in -man, run some hoops to find out whether we're in
1430: * the correct section or not.
1431: */
1.16 schwarze 1432:
1.141 schwarze 1433: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.47 schwarze 1434: body = n;
1435: assert(body->parent);
1436: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1437: 1 == head->nchild &&
1438: NULL != (head = (head->child)) &&
1.141 schwarze 1439: head->type == ROFFT_TEXT &&
1.98 schwarze 1440: 0 == strcmp(head->string, "NAME") &&
1441: NULL != body->child) {
1.2 schwarze 1442:
1.47 schwarze 1443: /*
1444: * Suck the entire NAME section into memory.
1445: * Yes, we might run away.
1446: * But too many manuals have big, spread-out
1447: * NAME sections over many lines.
1448: */
1.2 schwarze 1449:
1.78 schwarze 1450: title = NULL;
1.147 schwarze 1451: deroff(&title, body);
1.47 schwarze 1452: if (NULL == title)
1453: return;
1.16 schwarze 1454:
1.98 schwarze 1455: /*
1.47 schwarze 1456: * Go through a special heuristic dance here.
1457: * Conventionally, one or more manual names are
1458: * comma-specified prior to a whitespace, then a
1459: * dash, then a description. Try to puzzle out
1460: * the name parts here.
1461: */
1.16 schwarze 1462:
1.78 schwarze 1463: start = title;
1.47 schwarze 1464: for ( ;; ) {
1465: sz = strcspn(start, " ,");
1466: if ('\0' == start[sz])
1467: break;
1.1 schwarze 1468:
1.47 schwarze 1469: byte = start[sz];
1470: start[sz] = '\0';
1.67 schwarze 1471:
1472: /*
1473: * Assume a stray trailing comma in the
1474: * name list if a name begins with a dash.
1475: */
1476:
1477: if ('-' == start[0] ||
1478: ('\\' == start[0] && '-' == start[1]))
1479: break;
1.1 schwarze 1480:
1.90 schwarze 1481: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1482: if ( ! (mpage->name_head_done ||
1483: strcasecmp(start, meta->title))) {
1484: putkey(mpage, start, NAME_HEAD);
1485: mpage->name_head_done = 1;
1486: }
1.1 schwarze 1487:
1.47 schwarze 1488: if (' ' == byte) {
1489: start += sz + 1;
1490: break;
1491: }
1.1 schwarze 1492:
1.47 schwarze 1493: assert(',' == byte);
1494: start += sz + 1;
1495: while (' ' == *start)
1496: start++;
1497: }
1.1 schwarze 1498:
1.78 schwarze 1499: if (start == title) {
1.90 schwarze 1500: putkey(mpage, start, NAME_TITLE);
1.129 schwarze 1501: if ( ! (mpage->name_head_done ||
1502: strcasecmp(start, meta->title))) {
1503: putkey(mpage, start, NAME_HEAD);
1504: mpage->name_head_done = 1;
1505: }
1.47 schwarze 1506: free(title);
1507: return;
1508: }
1.1 schwarze 1509:
1.47 schwarze 1510: while (isspace((unsigned char)*start))
1511: start++;
1.1 schwarze 1512:
1.47 schwarze 1513: if (0 == strncmp(start, "-", 1))
1514: start += 1;
1515: else if (0 == strncmp(start, "\\-\\-", 4))
1516: start += 4;
1517: else if (0 == strncmp(start, "\\-", 2))
1518: start += 2;
1519: else if (0 == strncmp(start, "\\(en", 4))
1520: start += 4;
1521: else if (0 == strncmp(start, "\\(em", 4))
1522: start += 4;
1.1 schwarze 1523:
1.47 schwarze 1524: while (' ' == *start)
1525: start++;
1.1 schwarze 1526:
1.47 schwarze 1527: mpage->desc = mandoc_strdup(start);
1528: free(title);
1529: return;
1530: }
1531: }
1.1 schwarze 1532:
1.47 schwarze 1533: for (n = n->child; n; n = n->next) {
1534: if (NULL != mpage->desc)
1535: break;
1.129 schwarze 1536: parse_man(mpage, meta, n);
1.1 schwarze 1537: }
1538: }
1539:
1540: static void
1.143 schwarze 1541: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1542: const struct roff_node *n)
1.1 schwarze 1543: {
1544:
1.47 schwarze 1545: assert(NULL != n);
1546: for (n = n->child; NULL != n; n = n->next) {
1547: switch (n->type) {
1.141 schwarze 1548: case ROFFT_ELEM:
1549: case ROFFT_BLOCK:
1550: case ROFFT_HEAD:
1551: case ROFFT_BODY:
1552: case ROFFT_TAIL:
1.47 schwarze 1553: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1554: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1555: break;
1556: if (mdocs[n->tok].mask)
1557: putmdockey(mpage, n->child,
1558: mdocs[n->tok].mask);
1559: break;
1560: default:
1.141 schwarze 1561: assert(n->type != ROFFT_ROOT);
1.47 schwarze 1562: continue;
1563: }
1564: if (NULL != n->child)
1.127 schwarze 1565: parse_mdoc(mpage, meta, n);
1.1 schwarze 1566: }
1567: }
1568:
1.19 schwarze 1569: static int
1.143 schwarze 1570: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1571: const struct roff_node *n)
1.1 schwarze 1572: {
1.131 schwarze 1573: char *start, *end;
1.1 schwarze 1574: size_t sz;
1.19 schwarze 1575:
1.47 schwarze 1576: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1577: NULL == (n = n->child) ||
1.141 schwarze 1578: n->type != ROFFT_TEXT)
1.150 schwarze 1579: return 0;
1.1 schwarze 1580:
1581: /*
1582: * Only consider those `Fd' macro fields that begin with an
1583: * "inclusion" token (versus, e.g., #define).
1584: */
1.47 schwarze 1585:
1.1 schwarze 1586: if (strcmp("#include", n->string))
1.150 schwarze 1587: return 0;
1.1 schwarze 1588:
1.141 schwarze 1589: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.150 schwarze 1590: return 0;
1.1 schwarze 1591:
1592: /*
1593: * Strip away the enclosing angle brackets and make sure we're
1594: * not zero-length.
1595: */
1596:
1597: start = n->string;
1598: if ('<' == *start || '"' == *start)
1599: start++;
1600:
1601: if (0 == (sz = strlen(start)))
1.150 schwarze 1602: return 0;
1.1 schwarze 1603:
1604: end = &start[(int)sz - 1];
1605: if ('>' == *end || '"' == *end)
1606: end--;
1607:
1.47 schwarze 1608: if (end > start)
1609: putkeys(mpage, start, end - start + 1, TYPE_In);
1.150 schwarze 1610: return 0;
1.1 schwarze 1611: }
1612:
1.133 schwarze 1613: static void
1.142 schwarze 1614: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 schwarze 1615: {
1.69 schwarze 1616: char *cp;
1.133 schwarze 1617: size_t sz;
1.1 schwarze 1618:
1.141 schwarze 1619: if (n->type != ROFFT_TEXT)
1.133 schwarze 1620: return;
1.19 schwarze 1621:
1.133 schwarze 1622: /* Skip function pointer punctuation. */
1.1 schwarze 1623:
1.133 schwarze 1624: cp = n->string;
1625: while (*cp == '(' || *cp == '*')
1.1 schwarze 1626: cp++;
1.133 schwarze 1627: sz = strcspn(cp, "()");
1.1 schwarze 1628:
1.133 schwarze 1629: putkeys(mpage, cp, sz, TYPE_Fn);
1.128 schwarze 1630: if (n->sec == SEC_SYNOPSIS)
1.133 schwarze 1631: putkeys(mpage, cp, sz, NAME_SYN);
1632: }
1.19 schwarze 1633:
1.133 schwarze 1634: static int
1.143 schwarze 1635: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1636: const struct roff_node *n)
1.133 schwarze 1637: {
1638:
1639: if (n->child == NULL)
1.150 schwarze 1640: return 0;
1.133 schwarze 1641:
1642: parse_mdoc_fname(mpage, n->child);
1.19 schwarze 1643:
1.133 schwarze 1644: for (n = n->child->next; n != NULL; n = n->next)
1.141 schwarze 1645: if (n->type == ROFFT_TEXT)
1.47 schwarze 1646: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1647:
1.150 schwarze 1648: return 0;
1.128 schwarze 1649: }
1650:
1651: static int
1.143 schwarze 1652: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1653: const struct roff_node *n)
1.128 schwarze 1654: {
1.132 schwarze 1655:
1.141 schwarze 1656: if (n->type != ROFFT_HEAD)
1.150 schwarze 1657: return 1;
1.128 schwarze 1658:
1.133 schwarze 1659: if (n->child != NULL)
1660: parse_mdoc_fname(mpage, n->child);
1661:
1.150 schwarze 1662: return 0;
1.1 schwarze 1663: }
1664:
1.19 schwarze 1665: static int
1.143 schwarze 1666: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1667: const struct roff_node *n)
1.1 schwarze 1668: {
1.47 schwarze 1669: char *cp;
1.1 schwarze 1670:
1671: if (NULL == (n = n->child))
1.150 schwarze 1672: return 0;
1.1 schwarze 1673:
1.47 schwarze 1674: if (NULL == n->next) {
1675: putkey(mpage, n->string, TYPE_Xr);
1.150 schwarze 1676: return 0;
1.47 schwarze 1677: }
1.1 schwarze 1678:
1.77 schwarze 1679: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1680: putkey(mpage, cp, TYPE_Xr);
1681: free(cp);
1.150 schwarze 1682: return 0;
1.1 schwarze 1683: }
1684:
1.19 schwarze 1685: static int
1.143 schwarze 1686: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1687: const struct roff_node *n)
1.1 schwarze 1688: {
1689:
1.141 schwarze 1690: if (n->type == ROFFT_BODY)
1.147 schwarze 1691: deroff(&mpage->desc, n);
1.150 schwarze 1692: return 0;
1.1 schwarze 1693: }
1694:
1.19 schwarze 1695: static int
1.143 schwarze 1696: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1697: const struct roff_node *n)
1.1 schwarze 1698: {
1699:
1.86 schwarze 1700: if (SEC_NAME == n->sec)
1.90 schwarze 1701: putmdockey(mpage, n->child, NAME_TITLE);
1.141 schwarze 1702: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.127 schwarze 1703: if (n->child == NULL)
1704: putkey(mpage, meta->name, NAME_SYN);
1705: else
1706: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1707: }
1708: if ( ! (mpage->name_head_done ||
1709: n->child == NULL || n->child->string == NULL ||
1710: strcasecmp(n->child->string, meta->title))) {
1.141 schwarze 1711: putkey(mpage, n->child->string, ROFFT_HEAD);
1.129 schwarze 1712: mpage->name_head_done = 1;
1.127 schwarze 1713: }
1.150 schwarze 1714: return 0;
1.1 schwarze 1715: }
1716:
1.19 schwarze 1717: static int
1.143 schwarze 1718: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1719: const struct roff_node *n)
1.1 schwarze 1720: {
1721:
1.150 schwarze 1722: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 schwarze 1723: }
1724:
1.47 schwarze 1725: static int
1.143 schwarze 1726: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1727: const struct roff_node *n)
1.1 schwarze 1728: {
1729:
1.150 schwarze 1730: return n->type == ROFFT_HEAD;
1.47 schwarze 1731: }
1.1 schwarze 1732:
1.47 schwarze 1733: static int
1.143 schwarze 1734: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.142 schwarze 1735: const struct roff_node *n)
1.47 schwarze 1736: {
1.1 schwarze 1737:
1.150 schwarze 1738: return n->type == ROFFT_BODY;
1.1 schwarze 1739: }
1740:
1.47 schwarze 1741: /*
1742: * Add a string to the hash table for the current manual.
1743: * Each string has a bitmask telling which macros it belongs to.
1744: * When we finish the manual, we'll dump the table.
1745: */
1.1 schwarze 1746: static void
1.131 schwarze 1747: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.1 schwarze 1748: {
1.90 schwarze 1749: struct ohash *htab;
1.47 schwarze 1750: struct str *s;
1.68 schwarze 1751: const char *end;
1.47 schwarze 1752: unsigned int slot;
1.131 schwarze 1753: int i, mustfree;
1.1 schwarze 1754:
1.47 schwarze 1755: if (0 == sz)
1756: return;
1.68 schwarze 1757:
1.131 schwarze 1758: mustfree = render_string(&cp, &sz);
1759:
1.90 schwarze 1760: if (TYPE_Nm & v) {
1761: htab = &names;
1762: v &= name_mask;
1.124 schwarze 1763: if (v & NAME_FIRST)
1764: name_mask &= ~NAME_FIRST;
1.90 schwarze 1765: if (debug > 1)
1766: say(mpage->mlinks->file,
1.130 schwarze 1767: "Adding name %*s, bits=%d", sz, cp, v);
1.90 schwarze 1768: } else {
1769: htab = &strings;
1770: if (debug > 1)
1771: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1772: if ((uint64_t)1 << i & v)
1.90 schwarze 1773: say(mpage->mlinks->file,
1774: "Adding key %s=%*s",
1775: mansearch_keynames[i], sz, cp);
1.68 schwarze 1776: }
1.47 schwarze 1777:
1778: end = cp + sz;
1.90 schwarze 1779: slot = ohash_qlookupi(htab, cp, &end);
1780: s = ohash_find(htab, slot);
1.1 schwarze 1781:
1.47 schwarze 1782: if (NULL != s && mpage == s->mpage) {
1783: s->mask |= v;
1.1 schwarze 1784: return;
1.47 schwarze 1785: } else if (NULL == s) {
1.101 schwarze 1786: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1787: memcpy(s->key, cp, sz);
1.90 schwarze 1788: ohash_insert(htab, slot, s);
1.47 schwarze 1789: }
1790: s->mpage = mpage;
1791: s->mask = v;
1.131 schwarze 1792:
1793: if (mustfree)
1794: free(cp);
1.1 schwarze 1795: }
1796:
1797: /*
1.47 schwarze 1798: * Take a Unicode codepoint and produce its UTF-8 encoding.
1799: * This isn't the best way to do this, but it works.
1800: * The magic numbers are from the UTF-8 packaging.
1801: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1802: */
1.47 schwarze 1803: static size_t
1804: utf8(unsigned int cp, char out[7])
1.1 schwarze 1805: {
1.47 schwarze 1806: size_t rc;
1.1 schwarze 1807:
1.47 schwarze 1808: rc = 0;
1809: if (cp <= 0x0000007F) {
1810: rc = 1;
1811: out[0] = (char)cp;
1812: } else if (cp <= 0x000007FF) {
1813: rc = 2;
1814: out[0] = (cp >> 6 & 31) | 192;
1815: out[1] = (cp & 63) | 128;
1816: } else if (cp <= 0x0000FFFF) {
1817: rc = 3;
1818: out[0] = (cp >> 12 & 15) | 224;
1819: out[1] = (cp >> 6 & 63) | 128;
1820: out[2] = (cp & 63) | 128;
1821: } else if (cp <= 0x001FFFFF) {
1822: rc = 4;
1823: out[0] = (cp >> 18 & 7) | 240;
1824: out[1] = (cp >> 12 & 63) | 128;
1825: out[2] = (cp >> 6 & 63) | 128;
1826: out[3] = (cp & 63) | 128;
1827: } else if (cp <= 0x03FFFFFF) {
1828: rc = 5;
1829: out[0] = (cp >> 24 & 3) | 248;
1830: out[1] = (cp >> 18 & 63) | 128;
1831: out[2] = (cp >> 12 & 63) | 128;
1832: out[3] = (cp >> 6 & 63) | 128;
1833: out[4] = (cp & 63) | 128;
1834: } else if (cp <= 0x7FFFFFFF) {
1835: rc = 6;
1836: out[0] = (cp >> 30 & 1) | 252;
1837: out[1] = (cp >> 24 & 63) | 128;
1838: out[2] = (cp >> 18 & 63) | 128;
1839: out[3] = (cp >> 12 & 63) | 128;
1840: out[4] = (cp >> 6 & 63) | 128;
1841: out[5] = (cp & 63) | 128;
1842: } else
1.150 schwarze 1843: return 0;
1.19 schwarze 1844:
1.47 schwarze 1845: out[rc] = '\0';
1.150 schwarze 1846: return rc;
1.1 schwarze 1847: }
1848:
1.47 schwarze 1849: /*
1.131 schwarze 1850: * If the string contains escape sequences,
1851: * replace it with an allocated rendering and return 1,
1852: * such that the caller can free it after use.
1853: * Otherwise, do nothing and return 0.
1.47 schwarze 1854: */
1.131 schwarze 1855: static int
1856: render_string(char **public, size_t *psz)
1.1 schwarze 1857: {
1.131 schwarze 1858: const char *src, *scp, *addcp, *seq;
1859: char *dst;
1860: size_t ssz, dsz, addsz;
1.71 schwarze 1861: char utfbuf[7], res[6];
1.131 schwarze 1862: int seqlen, unicode;
1.47 schwarze 1863:
1864: res[0] = '\\';
1865: res[1] = '\t';
1866: res[2] = ASCII_NBRSP;
1867: res[3] = ASCII_HYPH;
1.71 schwarze 1868: res[4] = ASCII_BREAK;
1869: res[5] = '\0';
1.1 schwarze 1870:
1.131 schwarze 1871: src = scp = *public;
1872: ssz = *psz;
1873: dst = NULL;
1874: dsz = 0;
1875:
1876: while (scp < src + *psz) {
1877:
1878: /* Leave normal characters unchanged. */
1879:
1880: if (strchr(res, *scp) == NULL) {
1881: if (dst != NULL)
1882: dst[dsz++] = *scp;
1883: scp++;
1884: continue;
1885: }
1.1 schwarze 1886:
1.47 schwarze 1887: /*
1.131 schwarze 1888: * Found something that requires replacing,
1889: * make sure we have a destination buffer.
1.47 schwarze 1890: */
1.131 schwarze 1891:
1892: if (dst == NULL) {
1893: dst = mandoc_malloc(ssz + 1);
1894: dsz = scp - src;
1895: memcpy(dst, src, dsz);
1.47 schwarze 1896: }
1.39 schwarze 1897:
1.131 schwarze 1898: /* Handle single-char special characters. */
1899:
1900: switch (*scp) {
1901: case '\\':
1902: break;
1.98 schwarze 1903: case '\t':
1904: case ASCII_NBRSP:
1.131 schwarze 1905: dst[dsz++] = ' ';
1906: scp++;
1907: continue;
1908: case ASCII_HYPH:
1909: dst[dsz++] = '-';
1.71 schwarze 1910: /* FALLTHROUGH */
1.98 schwarze 1911: case ASCII_BREAK:
1.131 schwarze 1912: scp++;
1.47 schwarze 1913: continue;
1.71 schwarze 1914: default:
1.131 schwarze 1915: abort();
1.71 schwarze 1916: }
1.39 schwarze 1917:
1.47 schwarze 1918: /*
1.131 schwarze 1919: * Found an escape sequence.
1920: * Read past the slash, then parse it.
1921: * Ignore everything except characters.
1.47 schwarze 1922: */
1.52 schwarze 1923:
1.131 schwarze 1924: scp++;
1925: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.47 schwarze 1926: continue;
1.39 schwarze 1927:
1.47 schwarze 1928: /*
1.52 schwarze 1929: * Render the special character
1930: * as either UTF-8 or ASCII.
1.47 schwarze 1931: */
1.52 schwarze 1932:
1933: if (write_utf8) {
1.131 schwarze 1934: unicode = mchars_spec2cp(mchars, seq, seqlen);
1935: if (unicode <= 0)
1.52 schwarze 1936: continue;
1.131 schwarze 1937: addsz = utf8(unicode, utfbuf);
1938: if (addsz == 0)
1.52 schwarze 1939: continue;
1.131 schwarze 1940: addcp = utfbuf;
1.52 schwarze 1941: } else {
1.131 schwarze 1942: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1943: if (addcp == NULL)
1.52 schwarze 1944: continue;
1.131 schwarze 1945: if (*addcp == ASCII_NBRSP) {
1946: addcp = " ";
1947: addsz = 1;
1.52 schwarze 1948: }
1949: }
1.1 schwarze 1950:
1.47 schwarze 1951: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1952:
1.131 schwarze 1953: ssz += addsz;
1954: dst = mandoc_realloc(dst, ssz + 1);
1955: memcpy(dst + dsz, addcp, addsz);
1956: dsz += addsz;
1957: }
1958: if (dst != NULL) {
1959: *public = dst;
1960: *psz = dsz;
1961: }
1962:
1963: /* Trim trailing whitespace and NUL-terminate. */
1964:
1965: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1966: --*psz;
1967: if (dst != NULL) {
1968: (*public)[*psz] = '\0';
1.150 schwarze 1969: return 1;
1.131 schwarze 1970: } else
1.150 schwarze 1971: return 0;
1.1 schwarze 1972: }
1973:
1.75 schwarze 1974: static void
1975: dbadd_mlink(const struct mlink *mlink)
1976: {
1977: size_t i;
1978:
1979: i = 1;
1980: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1981: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1982: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1983: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1984: SQL_STEP(stmts[STMT_INSERT_LINK]);
1985: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 1986: }
1987:
1988: static void
1989: dbadd_mlink_name(const struct mlink *mlink)
1990: {
1.130 schwarze 1991: uint64_t bits;
1.124 schwarze 1992: size_t i;
1993:
1994: dbadd_mlink(mlink);
1.115 schwarze 1995:
1996: i = 1;
1.130 schwarze 1997: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
1998: bits = NAME_FILE & NAME_MASK;
1999: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2000: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2001: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2002: }
2003:
2004: i = 1;
2005: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.115 schwarze 2006: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2007: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2008: SQL_STEP(stmts[STMT_INSERT_NAME]);
2009: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2010: }
2011:
1.11 schwarze 2012: /*
1.47 schwarze 2013: * Flush the current page's terms (and their bits) into the database.
2014: * Wrap the entire set of additions in a transaction to make sqlite be a
2015: * little faster.
1.53 schwarze 2016: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2017: */
2018: static void
1.131 schwarze 2019: dbadd(struct mpage *mpage)
1.11 schwarze 2020: {
1.47 schwarze 2021: struct mlink *mlink;
2022: struct str *key;
1.131 schwarze 2023: char *cp;
1.47 schwarze 2024: size_t i;
2025: unsigned int slot;
1.131 schwarze 2026: int mustfree;
1.47 schwarze 2027:
1.85 schwarze 2028: mlink = mpage->mlinks;
1.11 schwarze 2029:
1.85 schwarze 2030: if (nodb) {
1.104 schwarze 2031: for (key = ohash_first(&names, &slot); NULL != key;
1.131 schwarze 2032: key = ohash_next(&names, &slot))
1.104 schwarze 2033: free(key);
2034: for (key = ohash_first(&strings, &slot); NULL != key;
1.131 schwarze 2035: key = ohash_next(&strings, &slot))
1.104 schwarze 2036: free(key);
1.102 schwarze 2037: if (0 == debug)
2038: return;
1.85 schwarze 2039: while (NULL != mlink) {
2040: fputs(mlink->name, stdout);
2041: if (NULL == mlink->next ||
2042: strcmp(mlink->dsec, mlink->next->dsec) ||
2043: strcmp(mlink->fsec, mlink->next->fsec) ||
2044: strcmp(mlink->arch, mlink->next->arch)) {
2045: putchar('(');
2046: if ('\0' == *mlink->dsec)
2047: fputs(mlink->fsec, stdout);
2048: else
2049: fputs(mlink->dsec, stdout);
2050: if ('\0' != *mlink->arch)
2051: printf("/%s", mlink->arch);
2052: putchar(')');
2053: }
2054: mlink = mlink->next;
2055: if (NULL != mlink)
2056: fputs(", ", stdout);
2057: }
1.89 schwarze 2058: printf(" - %s\n", mpage->desc);
1.11 schwarze 2059: return;
1.85 schwarze 2060: }
2061:
2062: if (debug)
2063: say(mlink->file, "Adding to database");
1.47 schwarze 2064:
1.131 schwarze 2065: cp = mpage->desc;
2066: i = strlen(cp);
2067: mustfree = render_string(&cp, &i);
1.47 schwarze 2068: i = 1;
1.131 schwarze 2069: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.116 schwarze 2070: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2071: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2072: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2073: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.131 schwarze 2074: if (mustfree)
2075: free(cp);
1.47 schwarze 2076:
1.85 schwarze 2077: while (NULL != mlink) {
1.75 schwarze 2078: dbadd_mlink(mlink);
1.85 schwarze 2079: mlink = mlink->next;
2080: }
1.91 schwarze 2081: mlink = mpage->mlinks;
1.47 schwarze 2082:
1.90 schwarze 2083: for (key = ohash_first(&names, &slot); NULL != key;
2084: key = ohash_next(&names, &slot)) {
2085: assert(key->mpage == mpage);
2086: i = 1;
2087: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.131 schwarze 2088: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.94 schwarze 2089: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2090: SQL_STEP(stmts[STMT_INSERT_NAME]);
2091: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2092: free(key);
2093: }
1.47 schwarze 2094: for (key = ohash_first(&strings, &slot); NULL != key;
2095: key = ohash_next(&strings, &slot)) {
2096: assert(key->mpage == mpage);
2097: i = 1;
2098: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.131 schwarze 2099: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.94 schwarze 2100: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2101: SQL_STEP(stmts[STMT_INSERT_KEY]);
2102: sqlite3_reset(stmts[STMT_INSERT_KEY]);
2103: free(key);
1.33 schwarze 2104: }
1.47 schwarze 2105: }
1.41 deraadt 2106:
1.47 schwarze 2107: static void
2108: dbprune(void)
2109: {
2110: struct mpage *mpage;
2111: struct mlink *mlink;
2112: size_t i;
2113: unsigned int slot;
1.11 schwarze 2114:
1.63 schwarze 2115: if (0 == nodb)
2116: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2117:
1.63 schwarze 2118: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2119: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2120: mlink = mpage->mlinks;
1.82 schwarze 2121: if (debug)
1.63 schwarze 2122: say(mlink->file, "Deleting from database");
2123: if (nodb)
2124: continue;
2125: for ( ; NULL != mlink; mlink = mlink->next) {
2126: i = 1;
2127: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2128: i, mlink->dsec);
2129: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2130: i, mlink->arch);
2131: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2132: i, mlink->name);
2133: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2134: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2135: }
1.11 schwarze 2136: }
1.63 schwarze 2137:
2138: if (0 == nodb)
2139: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2140: }
1.22 schwarze 2141:
1.47 schwarze 2142: /*
2143: * Close an existing database and its prepared statements.
2144: * If "real" is not set, rename the temporary file into the real one.
2145: */
2146: static void
2147: dbclose(int real)
2148: {
2149: size_t i;
1.72 schwarze 2150: int status;
2151: pid_t child;
1.11 schwarze 2152:
1.47 schwarze 2153: if (nodb)
2154: return;
1.11 schwarze 2155:
1.47 schwarze 2156: for (i = 0; i < STMT__MAX; i++) {
2157: sqlite3_finalize(stmts[i]);
2158: stmts[i] = NULL;
1.28 schwarze 2159: }
1.22 schwarze 2160:
1.47 schwarze 2161: sqlite3_close(db);
2162: db = NULL;
1.11 schwarze 2163:
1.47 schwarze 2164: if (real)
2165: return;
1.22 schwarze 2166:
1.72 schwarze 2167: if ('\0' == *tempfilename) {
2168: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2169: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2170: say(MANDOC_DB, "&rename");
1.72 schwarze 2171: }
2172: return;
2173: }
2174:
2175: switch (child = fork()) {
1.98 schwarze 2176: case -1:
1.72 schwarze 2177: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2178: say("", "&fork cmp");
1.72 schwarze 2179: return;
1.98 schwarze 2180: case 0:
1.72 schwarze 2181: execlp("cmp", "cmp", "-s",
1.149 schwarze 2182: tempfilename, MANDOC_DB, (char *)NULL);
1.80 schwarze 2183: say("", "&exec cmp");
1.72 schwarze 2184: exit(0);
2185: default:
2186: break;
2187: }
2188: if (-1 == waitpid(child, &status, 0)) {
2189: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2190: say("", "&wait cmp");
1.72 schwarze 2191: } else if (WIFSIGNALED(status)) {
2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2193: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2194: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2195: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2196: say(MANDOC_DB,
2197: "Data changed, but cannot replace database");
2198: }
2199:
2200: *strrchr(tempfilename, '/') = '\0';
2201: switch (child = fork()) {
1.98 schwarze 2202: case -1:
1.72 schwarze 2203: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2204: say("", "&fork rm");
1.72 schwarze 2205: return;
1.98 schwarze 2206: case 0:
1.149 schwarze 2207: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.80 schwarze 2208: say("", "&exec rm");
1.72 schwarze 2209: exit((int)MANDOCLEVEL_SYSERR);
2210: default:
2211: break;
2212: }
2213: if (-1 == waitpid(child, &status, 0)) {
2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2215: say("", "&wait rm");
1.72 schwarze 2216: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2217: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2218: say("", "%s: Cannot remove temporary directory",
2219: tempfilename);
1.22 schwarze 2220: }
1.11 schwarze 2221: }
2222:
1.47 schwarze 2223: /*
2224: * This is straightforward stuff.
2225: * Open a database connection to a "temporary" database, then open a set
2226: * of prepared statements we'll use over and over again.
2227: * If "real" is set, we use the existing database; if not, we truncate a
2228: * temporary one.
2229: * Must be matched by dbclose().
2230: */
2231: static int
2232: dbopen(int real)
1.2 schwarze 2233: {
1.72 schwarze 2234: const char *sql;
1.47 schwarze 2235: int rc, ofl;
1.6 schwarze 2236:
1.98 schwarze 2237: if (nodb)
1.150 schwarze 2238: return 1;
1.6 schwarze 2239:
1.72 schwarze 2240: *tempfilename = '\0';
1.47 schwarze 2241: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2242:
2243: if (real) {
2244: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2245: if (SQLITE_OK != rc) {
1.47 schwarze 2246: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2247: if (SQLITE_CANTOPEN != rc)
2248: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.150 schwarze 2249: return 0;
1.28 schwarze 2250: }
1.72 schwarze 2251: goto prepare_statements;
2252: }
2253:
2254: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2255:
1.72 schwarze 2256: remove(MANDOC_DB "~");
2257: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2258: if (SQLITE_OK == rc)
1.72 schwarze 2259: goto create_tables;
1.73 schwarze 2260: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2261: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2262: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.150 schwarze 2263: return 0;
1.47 schwarze 2264: }
1.6 schwarze 2265:
1.100 schwarze 2266: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2267: sizeof(tempfilename));
1.72 schwarze 2268: if (NULL == mkdtemp(tempfilename)) {
2269: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2270: say("", "&%s", tempfilename);
1.150 schwarze 2271: return 0;
1.72 schwarze 2272: }
1.100 schwarze 2273: (void)strlcat(tempfilename, "/" MANDOC_DB,
2274: sizeof(tempfilename));
1.72 schwarze 2275: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2276: if (SQLITE_OK != rc) {
1.47 schwarze 2277: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2278: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.150 schwarze 2279: return 0;
1.2 schwarze 2280: }
2281:
1.72 schwarze 2282: create_tables:
1.47 schwarze 2283: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2284: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2285: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2286: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2287: ");\n"
2288: "\n"
2289: "CREATE TABLE \"mlinks\" (\n"
2290: " \"sec\" TEXT NOT NULL,\n"
2291: " \"arch\" TEXT NOT NULL,\n"
2292: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2293: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2294: "ON DELETE CASCADE\n"
1.47 schwarze 2295: ");\n"
1.93 schwarze 2296: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2297: "\n"
1.90 schwarze 2298: "CREATE TABLE \"names\" (\n"
2299: " \"bits\" INTEGER NOT NULL,\n"
2300: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2301: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.130 schwarze 2302: "ON DELETE CASCADE,\n"
2303: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.90 schwarze 2304: ");\n"
2305: "\n"
1.47 schwarze 2306: "CREATE TABLE \"keys\" (\n"
2307: " \"bits\" INTEGER NOT NULL,\n"
2308: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2309: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2310: "ON DELETE CASCADE\n"
1.93 schwarze 2311: ");\n"
2312: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2313:
2314: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2315: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2316: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2317: sqlite3_close(db);
1.150 schwarze 2318: return 0;
1.2 schwarze 2319: }
2320:
1.47 schwarze 2321: prepare_statements:
1.103 schwarze 2322: if (SQLITE_OK != sqlite3_exec(db,
2323: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2324: exitcode = (int)MANDOCLEVEL_SYSERR;
2325: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2326: sqlite3_errmsg(db));
2327: sqlite3_close(db);
1.150 schwarze 2328: return 0;
1.103 schwarze 2329: }
2330:
1.94 schwarze 2331: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2332: "(SELECT pageid FROM mlinks WHERE "
2333: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2334: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2335: sql = "INSERT INTO mpages "
1.89 schwarze 2336: "(desc,form) VALUES (?,?)";
1.47 schwarze 2337: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2338: sql = "INSERT INTO mlinks "
1.61 schwarze 2339: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2340: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.130 schwarze 2341: sql = "SELECT bits FROM names where pageid = ?";
2342: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.90 schwarze 2343: sql = "INSERT INTO names "
2344: "(bits,name,pageid) VALUES (?,?,?)";
2345: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2346: sql = "INSERT INTO keys "
2347: "(bits,key,pageid) VALUES (?,?,?)";
2348: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2349:
1.47 schwarze 2350: /*
2351: * When opening a new database, we can turn off
2352: * synchronous mode for much better performance.
2353: */
1.6 schwarze 2354:
1.103 schwarze 2355: if (real && SQLITE_OK != sqlite3_exec(db,
2356: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2357: exitcode = (int)MANDOCLEVEL_SYSERR;
2358: say(MANDOC_DB, "PRAGMA synchronous: %s",
2359: sqlite3_errmsg(db));
2360: sqlite3_close(db);
1.150 schwarze 2361: return 0;
1.103 schwarze 2362: }
1.11 schwarze 2363:
1.150 schwarze 2364: return 1;
1.47 schwarze 2365: }
1.6 schwarze 2366:
1.47 schwarze 2367: static void *
1.106 espie 2368: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2369: {
1.6 schwarze 2370:
1.150 schwarze 2371: return mandoc_calloc(n, sz);
1.47 schwarze 2372: }
1.2 schwarze 2373:
1.47 schwarze 2374: static void *
2375: hash_alloc(size_t sz, void *arg)
2376: {
1.28 schwarze 2377:
1.150 schwarze 2378: return mandoc_malloc(sz);
1.47 schwarze 2379: }
1.6 schwarze 2380:
1.47 schwarze 2381: static void
1.106 espie 2382: hash_free(void *p, void *arg)
1.47 schwarze 2383: {
1.26 schwarze 2384:
1.47 schwarze 2385: free(p);
2386: }
1.6 schwarze 2387:
1.47 schwarze 2388: static int
1.120 schwarze 2389: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2390: {
2391: static char startdir[PATH_MAX];
1.110 schwarze 2392: static int getcwd_status; /* 1 = ok, 2 = failure */
2393: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2394: char *cp;
1.6 schwarze 2395:
1.47 schwarze 2396: /*
1.110 schwarze 2397: * Remember the original working directory, if possible.
2398: * This will be needed if the second or a later directory
2399: * on the command line is given as a relative path.
2400: * Do not error out if the current directory is not
2401: * searchable: Maybe it won't be needed after all.
2402: */
2403: if (0 == getcwd_status) {
2404: if (NULL == getcwd(startdir, sizeof(startdir))) {
2405: getcwd_status = 2;
2406: (void)strlcpy(startdir, strerror(errno),
2407: sizeof(startdir));
2408: } else
2409: getcwd_status = 1;
2410: }
2411:
2412: /*
2413: * We are leaving the old base directory.
2414: * Do not use it any longer, not even for messages.
2415: */
2416: *basedir = '\0';
2417:
2418: /*
2419: * If and only if the directory was changed earlier and
2420: * the next directory to process is given as a relative path,
2421: * first go back, or bail out if that is impossible.
1.47 schwarze 2422: */
1.110 schwarze 2423: if (chdir_status && '/' != *targetdir) {
2424: if (2 == getcwd_status) {
1.47 schwarze 2425: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2426: say("", "getcwd: %s", startdir);
1.150 schwarze 2427: return 0;
1.47 schwarze 2428: }
1.110 schwarze 2429: if (-1 == chdir(startdir)) {
1.47 schwarze 2430: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2431: say("", "&chdir %s", startdir);
1.150 schwarze 2432: return 0;
1.2 schwarze 2433: }
2434: }
1.110 schwarze 2435:
2436: /*
2437: * Always resolve basedir to the canonicalized absolute
2438: * pathname and append a trailing slash, such that
2439: * we can reliably check whether files are inside.
2440: */
1.47 schwarze 2441: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2442: if (report_baddir || errno != ENOENT) {
2443: exitcode = (int)MANDOCLEVEL_BADARG;
2444: say("", "&%s: realpath", targetdir);
2445: }
1.150 schwarze 2446: return 0;
1.47 schwarze 2447: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2448: if (report_baddir || errno != ENOENT) {
2449: exitcode = (int)MANDOCLEVEL_BADARG;
2450: say("", "&chdir");
2451: }
1.150 schwarze 2452: return 0;
1.109 schwarze 2453: }
1.110 schwarze 2454: chdir_status = 1;
1.109 schwarze 2455: cp = strchr(basedir, '\0');
2456: if ('/' != cp[-1]) {
2457: if (cp - basedir >= PATH_MAX - 1) {
2458: exitcode = (int)MANDOCLEVEL_SYSERR;
2459: say("", "Filename too long");
1.150 schwarze 2460: return 0;
1.109 schwarze 2461: }
2462: *cp++ = '/';
2463: *cp = '\0';
1.47 schwarze 2464: }
1.150 schwarze 2465: return 1;
1.2 schwarze 2466: }
2467:
2468: static void
1.47 schwarze 2469: say(const char *file, const char *format, ...)
1.2 schwarze 2470: {
1.47 schwarze 2471: va_list ap;
1.80 schwarze 2472: int use_errno;
1.2 schwarze 2473:
1.47 schwarze 2474: if ('\0' != *basedir)
2475: fprintf(stderr, "%s", basedir);
2476: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2477: fputc('/', stderr);
1.47 schwarze 2478: if ('\0' != *file)
2479: fprintf(stderr, "%s", file);
1.31 schwarze 2480:
1.80 schwarze 2481: use_errno = 1;
2482: if (NULL != format) {
2483: switch (*format) {
1.98 schwarze 2484: case '&':
1.80 schwarze 2485: format++;
2486: break;
1.98 schwarze 2487: case '\0':
1.80 schwarze 2488: format = NULL;
2489: break;
2490: default:
2491: use_errno = 0;
2492: break;
2493: }
2494: }
2495: if (NULL != format) {
2496: if ('\0' != *basedir || '\0' != *file)
2497: fputs(": ", stderr);
2498: va_start(ap, format);
2499: vfprintf(stderr, format, ap);
2500: va_end(ap);
2501: }
2502: if (use_errno) {
2503: if ('\0' != *basedir || '\0' != *file || NULL != format)
2504: fputs(": ", stderr);
1.47 schwarze 2505: perror(NULL);
1.80 schwarze 2506: } else
2507: fputc('\n', stderr);
1.1 schwarze 2508: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc