Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.105
1.105 ! schwarze 1: /* $Id: mandocdb.c,v 1.104 2014/05/07 15:10:33 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 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: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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: */
1.47 schwarze 18: #include <sys/stat.h>
1.72 schwarze 19: #include <sys/wait.h>
1.1 schwarze 20:
21: #include <assert.h>
1.33 schwarze 22: #include <ctype.h>
1.34 schwarze 23: #include <errno.h>
1.1 schwarze 24: #include <fcntl.h>
1.47 schwarze 25: #include <fts.h>
1.1 schwarze 26: #include <getopt.h>
1.44 schwarze 27: #include <limits.h>
1.47 schwarze 28: #include <stddef.h>
1.1 schwarze 29: #include <stdio.h>
30: #include <stdint.h>
31: #include <stdlib.h>
32: #include <string.h>
1.14 schwarze 33: #include <unistd.h>
1.1 schwarze 34:
1.47 schwarze 35: #include <ohash.h>
36: #include <sqlite3.h>
37:
38: #include "mdoc.h"
1.1 schwarze 39: #include "man.h"
40: #include "mandoc.h"
1.76 schwarze 41: #include "mandoc_aux.h"
1.10 schwarze 42: #include "manpath.h"
1.47 schwarze 43: #include "mansearch.h"
1.1 schwarze 44:
1.68 schwarze 45: extern int mansearch_keymax;
46: extern const char *const mansearch_keynames[];
47:
1.47 schwarze 48: #define SQL_EXEC(_v) \
49: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 50: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 51: #define SQL_BIND_TEXT(_s, _i, _v) \
52: if (SQLITE_OK != sqlite3_bind_text \
53: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 54: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 55: #define SQL_BIND_INT(_s, _i, _v) \
56: if (SQLITE_OK != sqlite3_bind_int \
57: ((_s), (_i)++, (_v))) \
1.91 schwarze 58: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 59: #define SQL_BIND_INT64(_s, _i, _v) \
60: if (SQLITE_OK != sqlite3_bind_int64 \
61: ((_s), (_i)++, (_v))) \
1.91 schwarze 62: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 63: #define SQL_STEP(_s) \
64: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 65: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 66:
1.47 schwarze 67: enum op {
68: OP_DEFAULT = 0, /* new dbs from dir list or default config */
69: OP_CONFFILE, /* new databases from custom config file */
70: OP_UPDATE, /* delete/add entries in existing database */
71: OP_DELETE, /* delete entries from existing database */
72: OP_TEST /* change no databases, report potential problems */
73: };
1.11 schwarze 74:
1.47 schwarze 75: enum form {
76: FORM_NONE, /* format is unknown */
77: FORM_SRC, /* format is -man or -mdoc */
78: FORM_CAT /* format is cat */
79: };
1.28 schwarze 80:
1.47 schwarze 81: struct str {
1.53 schwarze 82: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 83: const struct mpage *mpage; /* if set, the owning parse */
84: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 85: char key[]; /* may contain escape sequences */
1.28 schwarze 86: };
87:
1.47 schwarze 88: struct inodev {
89: ino_t st_ino;
90: dev_t st_dev;
91: };
1.28 schwarze 92:
1.47 schwarze 93: struct mpage {
94: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 95: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 96: enum form form; /* format from file content */
97: char *sec; /* section from file content */
98: char *arch; /* architecture from file content */
99: char *title; /* title from file content */
100: char *desc; /* description from file content */
101: struct mlink *mlinks; /* singly linked list */
1.28 schwarze 102: };
103:
1.47 schwarze 104: struct mlink {
105: char file[PATH_MAX]; /* filename rel. to manpath */
106: enum form dform; /* format from directory */
107: enum form fform; /* format from file name suffix */
108: char *dsec; /* section from directory */
109: char *arch; /* architecture from directory */
110: char *name; /* name from file name (not empty) */
111: char *fsec; /* section from file name suffix */
112: struct mlink *next; /* singly linked list */
1.75 schwarze 113: struct mpage *mpage; /* parent */
1.81 schwarze 114: int gzip; /* filename has a .gz suffix */
1.2 schwarze 115: };
116:
1.47 schwarze 117: enum stmt {
118: STMT_DELETE_PAGE = 0, /* delete mpage */
119: STMT_INSERT_PAGE, /* insert mpage */
120: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 121: STMT_INSERT_NAME, /* insert name */
1.47 schwarze 122: STMT_INSERT_KEY, /* insert parsed key */
123: STMT__MAX
1.1 schwarze 124: };
125:
1.47 schwarze 126: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 127:
1.19 schwarze 128: struct mdoc_handler {
1.47 schwarze 129: mdoc_fp fp; /* optional handler */
130: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 131: };
132:
1.47 schwarze 133: static void dbclose(int);
1.75 schwarze 134: static void dbadd(struct mpage *, struct mchars *);
135: static void dbadd_mlink(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 *);
140: static void hash_free(void *, size_t, void *);
141: static void *hash_halloc(size_t, void *);
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);
1.58 schwarze 147: static void mpages_merge(struct mchars *, struct mparse *);
1.87 schwarze 148: static void names_check(void);
1.81 schwarze 149: static void parse_cat(struct mpage *, int);
1.47 schwarze 150: static void parse_man(struct mpage *, const struct man_node *);
151: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
155: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
156: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
157: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
158: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
159: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.69 schwarze 160: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 161: static void putkeys(const struct mpage *,
162: const char *, size_t, uint64_t);
163: static void putmdockey(const struct mpage *,
164: const struct mdoc_node *, uint64_t);
1.53 schwarze 165: static void render_key(struct mchars *, struct str *);
1.47 schwarze 166: static void say(const char *, const char *, ...);
167: static int set_basedir(const char *);
168: static int treescan(void);
169: static size_t utf8(unsigned int, char [7]);
170:
1.72 schwarze 171: static char tempfilename[32];
1.47 schwarze 172: static char *progname;
1.59 schwarze 173: static int nodb; /* no database changes */
1.73 schwarze 174: static int mparse_options; /* abort the parse early */
1.98 schwarze 175: static int use_all; /* use all found files */
176: static int debug; /* print what we're doing */
177: static int warnings; /* warn about crap */
1.52 schwarze 178: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 179: static int exitcode; /* to be returned by main */
1.98 schwarze 180: static enum op op; /* operational mode */
1.47 schwarze 181: static char basedir[PATH_MAX]; /* current base directory */
182: static struct ohash mpages; /* table of distinct manual pages */
183: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 184: static struct ohash names; /* table of all names */
1.47 schwarze 185: static struct ohash strings; /* table of all strings */
186: static sqlite3 *db = NULL; /* current database */
187: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 188: static uint64_t name_mask;
1.47 schwarze 189:
1.19 schwarze 190: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 191: { NULL, 0 }, /* Ap */
192: { NULL, 0 }, /* Dd */
193: { NULL, 0 }, /* Dt */
194: { NULL, 0 }, /* Os */
195: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
196: { parse_mdoc_head, TYPE_Ss }, /* Ss */
197: { NULL, 0 }, /* Pp */
198: { NULL, 0 }, /* D1 */
199: { NULL, 0 }, /* Dl */
200: { NULL, 0 }, /* Bd */
201: { NULL, 0 }, /* Ed */
202: { NULL, 0 }, /* Bl */
203: { NULL, 0 }, /* El */
204: { NULL, 0 }, /* It */
205: { NULL, 0 }, /* Ad */
206: { NULL, TYPE_An }, /* An */
207: { NULL, TYPE_Ar }, /* Ar */
208: { NULL, TYPE_Cd }, /* Cd */
209: { NULL, TYPE_Cm }, /* Cm */
210: { NULL, TYPE_Dv }, /* Dv */
211: { NULL, TYPE_Er }, /* Er */
212: { NULL, TYPE_Ev }, /* Ev */
213: { NULL, 0 }, /* Ex */
214: { NULL, TYPE_Fa }, /* Fa */
215: { parse_mdoc_Fd, 0 }, /* Fd */
216: { NULL, TYPE_Fl }, /* Fl */
217: { parse_mdoc_Fn, 0 }, /* Fn */
218: { NULL, TYPE_Ft }, /* Ft */
219: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 220: { NULL, TYPE_In }, /* In */
1.47 schwarze 221: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 222: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 223: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 224: { NULL, 0 }, /* Op */
225: { NULL, 0 }, /* Ot */
226: { NULL, TYPE_Pa }, /* Pa */
227: { NULL, 0 }, /* Rv */
1.49 schwarze 228: { NULL, TYPE_St }, /* St */
1.47 schwarze 229: { NULL, TYPE_Va }, /* Va */
230: { parse_mdoc_body, TYPE_Va }, /* Vt */
231: { parse_mdoc_Xr, 0 }, /* Xr */
232: { NULL, 0 }, /* %A */
233: { NULL, 0 }, /* %B */
234: { NULL, 0 }, /* %D */
235: { NULL, 0 }, /* %I */
236: { NULL, 0 }, /* %J */
237: { NULL, 0 }, /* %N */
238: { NULL, 0 }, /* %O */
239: { NULL, 0 }, /* %P */
240: { NULL, 0 }, /* %R */
241: { NULL, 0 }, /* %T */
242: { NULL, 0 }, /* %V */
243: { NULL, 0 }, /* Ac */
244: { NULL, 0 }, /* Ao */
245: { NULL, 0 }, /* Aq */
246: { NULL, TYPE_At }, /* At */
247: { NULL, 0 }, /* Bc */
248: { NULL, 0 }, /* Bf */
249: { NULL, 0 }, /* Bo */
250: { NULL, 0 }, /* Bq */
251: { NULL, TYPE_Bsx }, /* Bsx */
252: { NULL, TYPE_Bx }, /* Bx */
253: { NULL, 0 }, /* Db */
254: { NULL, 0 }, /* Dc */
255: { NULL, 0 }, /* Do */
256: { NULL, 0 }, /* Dq */
257: { NULL, 0 }, /* Ec */
258: { NULL, 0 }, /* Ef */
259: { NULL, TYPE_Em }, /* Em */
260: { NULL, 0 }, /* Eo */
261: { NULL, TYPE_Fx }, /* Fx */
262: { NULL, TYPE_Ms }, /* Ms */
263: { NULL, 0 }, /* No */
264: { NULL, 0 }, /* Ns */
265: { NULL, TYPE_Nx }, /* Nx */
266: { NULL, TYPE_Ox }, /* Ox */
267: { NULL, 0 }, /* Pc */
268: { NULL, 0 }, /* Pf */
269: { NULL, 0 }, /* Po */
270: { NULL, 0 }, /* Pq */
271: { NULL, 0 }, /* Qc */
272: { NULL, 0 }, /* Ql */
273: { NULL, 0 }, /* Qo */
274: { NULL, 0 }, /* Qq */
275: { NULL, 0 }, /* Re */
276: { NULL, 0 }, /* Rs */
277: { NULL, 0 }, /* Sc */
278: { NULL, 0 }, /* So */
279: { NULL, 0 }, /* Sq */
280: { NULL, 0 }, /* Sm */
281: { NULL, 0 }, /* Sx */
282: { NULL, TYPE_Sy }, /* Sy */
283: { NULL, TYPE_Tn }, /* Tn */
284: { NULL, 0 }, /* Ux */
285: { NULL, 0 }, /* Xc */
286: { NULL, 0 }, /* Xo */
287: { parse_mdoc_head, 0 }, /* Fo */
288: { NULL, 0 }, /* Fc */
289: { NULL, 0 }, /* Oo */
290: { NULL, 0 }, /* Oc */
291: { NULL, 0 }, /* Bk */
292: { NULL, 0 }, /* Ek */
293: { NULL, 0 }, /* Bt */
294: { NULL, 0 }, /* Hf */
295: { NULL, 0 }, /* Fr */
296: { NULL, 0 }, /* Ud */
297: { NULL, TYPE_Lb }, /* Lb */
298: { NULL, 0 }, /* Lp */
299: { NULL, TYPE_Lk }, /* Lk */
300: { NULL, TYPE_Mt }, /* Mt */
301: { NULL, 0 }, /* Brq */
302: { NULL, 0 }, /* Bro */
303: { NULL, 0 }, /* Brc */
304: { NULL, 0 }, /* %C */
305: { NULL, 0 }, /* Es */
306: { NULL, 0 }, /* En */
307: { NULL, TYPE_Dx }, /* Dx */
308: { NULL, 0 }, /* %Q */
309: { NULL, 0 }, /* br */
310: { NULL, 0 }, /* sp */
311: { NULL, 0 }, /* %U */
312: { NULL, 0 }, /* Ta */
1.1 schwarze 313: };
314:
1.98 schwarze 315:
1.1 schwarze 316: int
1.3 schwarze 317: mandocdb(int argc, char *argv[])
1.1 schwarze 318: {
1.47 schwarze 319: int ch, i;
320: size_t j, sz;
321: const char *path_arg;
322: struct mchars *mc;
323: struct manpaths dirs;
324: struct mparse *mp;
325: struct ohash_info mpages_info, mlinks_info;
326:
327: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
328: memset(&dirs, 0, sizeof(struct manpaths));
329:
330: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
331: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
332: mpages_info.hfree = mlinks_info.hfree = hash_free;
333:
334: mpages_info.key_offset = offsetof(struct mpage, inodev);
335: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 336:
337: progname = strrchr(argv[0], '/');
338: if (progname == NULL)
339: progname = argv[0];
340: else
341: ++progname;
342:
1.47 schwarze 343: /*
1.98 schwarze 344: * We accept a few different invocations.
1.47 schwarze 345: * The CHECKOP macro makes sure that invocation styles don't
346: * clobber each other.
347: */
348: #define CHECKOP(_op, _ch) do \
349: if (OP_DEFAULT != (_op)) { \
350: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
351: goto usage; \
352: } while (/*CONSTCOND*/0)
1.10 schwarze 353:
1.47 schwarze 354: path_arg = NULL;
1.28 schwarze 355: op = OP_DEFAULT;
1.1 schwarze 356:
1.83 schwarze 357: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 358: switch (ch) {
1.98 schwarze 359: case 'a':
1.6 schwarze 360: use_all = 1;
361: break;
1.98 schwarze 362: case 'C':
1.47 schwarze 363: CHECKOP(op, ch);
364: path_arg = optarg;
1.28 schwarze 365: op = OP_CONFFILE;
1.25 schwarze 366: break;
1.98 schwarze 367: case 'D':
1.82 schwarze 368: debug++;
369: break;
1.98 schwarze 370: case 'd':
1.47 schwarze 371: CHECKOP(op, ch);
372: path_arg = optarg;
1.2 schwarze 373: op = OP_UPDATE;
1.1 schwarze 374: break;
1.98 schwarze 375: case 'n':
1.47 schwarze 376: nodb = 1;
377: break;
1.98 schwarze 378: case 'p':
1.83 schwarze 379: warnings = 1;
380: break;
1.98 schwarze 381: case 'Q':
1.73 schwarze 382: mparse_options |= MPARSE_QUICK;
1.59 schwarze 383: break;
1.98 schwarze 384: case 'T':
1.52 schwarze 385: if (strcmp(optarg, "utf8")) {
386: fprintf(stderr, "-T%s: Unsupported "
387: "output format\n", optarg);
388: goto usage;
389: }
390: write_utf8 = 1;
391: break;
1.98 schwarze 392: case 't':
1.47 schwarze 393: CHECKOP(op, ch);
1.28 schwarze 394: dup2(STDOUT_FILENO, STDERR_FILENO);
395: op = OP_TEST;
1.47 schwarze 396: nodb = warnings = 1;
1.28 schwarze 397: break;
1.98 schwarze 398: case 'u':
1.47 schwarze 399: CHECKOP(op, ch);
400: path_arg = optarg;
1.1 schwarze 401: op = OP_DELETE;
402: break;
1.98 schwarze 403: case 'v':
1.83 schwarze 404: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 405: break;
1.1 schwarze 406: default:
1.28 schwarze 407: goto usage;
1.1 schwarze 408: }
409:
410: argc -= optind;
411: argv += optind;
412:
1.28 schwarze 413: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 414: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 415: goto usage;
416: }
417:
1.47 schwarze 418: exitcode = (int)MANDOCLEVEL_OK;
1.73 schwarze 419: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.47 schwarze 420: mc = mchars_alloc();
1.2 schwarze 421:
1.47 schwarze 422: ohash_init(&mpages, 6, &mpages_info);
423: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 424:
1.47 schwarze 425: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 426:
1.47 schwarze 427: /*
428: * All of these deal with a specific directory.
1.97 schwarze 429: * Jump into that directory first.
1.47 schwarze 430: */
431: if (0 == set_basedir(path_arg))
432: goto out;
1.97 schwarze 433:
1.95 schwarze 434: if (dbopen(1)) {
1.97 schwarze 435: /*
436: * The existing database is usable. Process
437: * all files specified on the command-line.
438: */
439: use_all = 1;
1.95 schwarze 440: for (i = 0; i < argc; i++)
441: filescan(argv[i]);
442: if (OP_TEST != op)
443: dbprune();
444: } else {
445: /*
446: * Database missing or corrupt.
447: * Recreate from scratch.
448: */
1.97 schwarze 449: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 450: op = OP_DEFAULT;
451: if (0 == treescan())
452: goto out;
453: if (0 == dbopen(0))
454: goto out;
455: }
1.47 schwarze 456: if (OP_DELETE != op)
1.58 schwarze 457: mpages_merge(mc, mp);
1.95 schwarze 458: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 459: } else {
460: /*
461: * If we have arguments, use them as our manpaths.
462: * If we don't, grok from manpath(1) or however else
463: * manpath_parse() wants to do it.
464: */
465: if (argc > 0) {
1.101 schwarze 466: dirs.paths = mandoc_reallocarray(NULL,
467: argc, sizeof(char *));
1.47 schwarze 468: dirs.sz = (size_t)argc;
469: for (i = 0; i < argc; i++)
470: dirs.paths[i] = mandoc_strdup(argv[i]);
471: } else
472: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 473:
474: if (0 == dirs.sz) {
475: exitcode = (int)MANDOCLEVEL_BADARG;
476: say("", "Empty manpath");
477: }
1.2 schwarze 478:
1.47 schwarze 479: /*
480: * First scan the tree rooted at a base directory, then
481: * build a new database and finally move it into place.
482: * Ignore zero-length directories and strip trailing
483: * slashes.
484: */
485: for (j = 0; j < dirs.sz; j++) {
486: sz = strlen(dirs.paths[j]);
487: if (sz && '/' == dirs.paths[j][sz - 1])
488: dirs.paths[j][--sz] = '\0';
489: if (0 == sz)
490: continue;
1.2 schwarze 491:
1.47 schwarze 492: if (j) {
493: ohash_init(&mpages, 6, &mpages_info);
494: ohash_init(&mlinks, 6, &mlinks_info);
495: }
1.1 schwarze 496:
1.47 schwarze 497: if (0 == set_basedir(dirs.paths[j]))
498: goto out;
499: if (0 == treescan())
500: goto out;
501: if (0 == set_basedir(dirs.paths[j]))
502: goto out;
503: if (0 == dbopen(0))
504: goto out;
1.41 deraadt 505:
1.58 schwarze 506: mpages_merge(mc, mp);
1.104 schwarze 507: if (warnings && !nodb &&
1.87 schwarze 508: ! (MPARSE_QUICK & mparse_options))
509: names_check();
1.47 schwarze 510: dbclose(0);
1.28 schwarze 511:
1.47 schwarze 512: if (j + 1 < dirs.sz) {
513: mpages_free();
514: ohash_delete(&mpages);
515: ohash_delete(&mlinks);
516: }
1.2 schwarze 517: }
1.47 schwarze 518: }
519: out:
520: set_basedir(NULL);
521: manpath_free(&dirs);
522: mchars_free(mc);
523: mparse_free(mp);
524: mpages_free();
525: ohash_delete(&mpages);
526: ohash_delete(&mlinks);
527: return(exitcode);
528: usage:
1.83 schwarze 529: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
530: " %s [-aDnpQ] [-Tutf8] dir ...\n"
531: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
532: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 533: " %s [-Q] -t file ...\n",
1.98 schwarze 534: progname, progname, progname,
1.47 schwarze 535: progname, progname);
1.1 schwarze 536:
1.47 schwarze 537: return((int)MANDOCLEVEL_BADARG);
538: }
1.1 schwarze 539:
1.47 schwarze 540: /*
541: * Scan a directory tree rooted at "basedir" for manpages.
542: * We use fts(), scanning directory parts along the way for clues to our
543: * section and architecture.
544: *
545: * If use_all has been specified, grok all files.
546: * If not, sanitise paths to the following:
547: *
1.98 schwarze 548: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 549: * or
550: * [./]cat<section>[/<arch>]/<name>.0
551: *
552: * TODO: accomodate for multi-language directories.
553: */
554: static int
555: treescan(void)
556: {
1.96 schwarze 557: char buf[PATH_MAX];
1.47 schwarze 558: FTS *f;
559: FTSENT *ff;
560: struct mlink *mlink;
1.81 schwarze 561: int dform, gzip;
1.51 schwarze 562: char *dsec, *arch, *fsec, *cp;
563: const char *path;
1.47 schwarze 564: const char *argv[2];
1.1 schwarze 565:
1.47 schwarze 566: argv[0] = ".";
567: argv[1] = (char *)NULL;
1.1 schwarze 568:
1.96 schwarze 569: f = fts_open((char * const *)argv,
570: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 571: if (NULL == f) {
572: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 573: say("", "&fts_open");
1.47 schwarze 574: return(0);
575: }
1.2 schwarze 576:
1.47 schwarze 577: dsec = arch = NULL;
578: dform = FORM_NONE;
1.2 schwarze 579:
1.47 schwarze 580: while (NULL != (ff = fts_read(f))) {
581: path = ff->fts_path + 2;
1.96 schwarze 582: switch (ff->fts_info) {
583:
584: /*
585: * Symbolic links require various sanity checks,
586: * then get handled just like regular files.
587: */
1.98 schwarze 588: case FTS_SL:
1.96 schwarze 589: if (NULL == realpath(path, buf)) {
590: if (warnings)
591: say(path, "&realpath");
592: continue;
593: }
594: if (strstr(buf, basedir) != buf) {
595: if (warnings) say("",
596: "%s: outside base directory", buf);
597: continue;
598: }
599: /* Use logical inode to avoid mpages dupe. */
600: if (-1 == stat(path, ff->fts_statp)) {
601: if (warnings)
602: say(path, "&stat");
603: continue;
604: }
605: /* FALLTHROUGH */
606:
1.15 schwarze 607: /*
1.47 schwarze 608: * If we're a regular file, add an mlink by using the
609: * stored directory data and handling the filename.
1.15 schwarze 610: */
1.98 schwarze 611: case FTS_F:
1.47 schwarze 612: if (0 == strcmp(path, MANDOC_DB))
613: continue;
614: if ( ! use_all && ff->fts_level < 2) {
615: if (warnings)
616: say(path, "Extraneous file");
617: continue;
1.81 schwarze 618: }
619: gzip = 0;
620: fsec = NULL;
621: while (NULL == fsec) {
622: fsec = strrchr(ff->fts_name, '.');
623: if (NULL == fsec || strcmp(fsec+1, "gz"))
624: break;
625: gzip = 1;
626: *fsec = '\0';
627: fsec = NULL;
628: }
629: if (NULL == fsec) {
1.47 schwarze 630: if ( ! use_all) {
631: if (warnings)
632: say(path,
633: "No filename suffix");
634: continue;
635: }
636: } else if (0 == strcmp(++fsec, "html")) {
637: if (warnings)
638: say(path, "Skip html");
639: continue;
640: } else if (0 == strcmp(fsec, "ps")) {
641: if (warnings)
642: say(path, "Skip ps");
643: continue;
644: } else if (0 == strcmp(fsec, "pdf")) {
645: if (warnings)
646: say(path, "Skip pdf");
647: continue;
648: } else if ( ! use_all &&
649: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
650: (FORM_CAT == dform && strcmp(fsec, "0")))) {
651: if (warnings)
652: say(path, "Wrong filename suffix");
653: continue;
654: } else
655: fsec[-1] = '\0';
1.51 schwarze 656:
1.47 schwarze 657: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 658: if (strlcpy(mlink->file, path,
659: sizeof(mlink->file)) >=
660: sizeof(mlink->file)) {
661: say(path, "Filename too long");
662: free(mlink);
663: continue;
664: }
1.47 schwarze 665: mlink->dform = dform;
1.51 schwarze 666: mlink->dsec = dsec;
667: mlink->arch = arch;
668: mlink->name = ff->fts_name;
669: mlink->fsec = fsec;
1.81 schwarze 670: mlink->gzip = gzip;
1.47 schwarze 671: mlink_add(mlink, ff->fts_statp);
672: continue;
1.96 schwarze 673:
1.98 schwarze 674: case FTS_D:
1.96 schwarze 675: /* FALLTHROUGH */
1.98 schwarze 676: case FTS_DP:
1.96 schwarze 677: break;
678:
679: default:
1.47 schwarze 680: if (warnings)
681: say(path, "Not a regular file");
682: continue;
683: }
684:
685: switch (ff->fts_level) {
1.98 schwarze 686: case 0:
1.47 schwarze 687: /* Ignore the root directory. */
688: break;
1.98 schwarze 689: case 1:
1.47 schwarze 690: /*
691: * This might contain manX/ or catX/.
692: * Try to infer this from the name.
693: * If we're not in use_all, enforce it.
694: */
695: cp = ff->fts_name;
696: if (FTS_DP == ff->fts_info)
697: break;
1.15 schwarze 698:
1.47 schwarze 699: if (0 == strncmp(cp, "man", 3)) {
700: dform = FORM_SRC;
701: dsec = cp + 3;
702: } else if (0 == strncmp(cp, "cat", 3)) {
703: dform = FORM_CAT;
704: dsec = cp + 3;
1.51 schwarze 705: } else {
706: dform = FORM_NONE;
707: dsec = NULL;
1.26 schwarze 708: }
1.47 schwarze 709:
1.98 schwarze 710: if (NULL != dsec || use_all)
1.47 schwarze 711: break;
712:
713: if (warnings)
714: say(path, "Unknown directory part");
715: fts_set(f, ff, FTS_SKIP);
716: break;
1.98 schwarze 717: case 2:
1.47 schwarze 718: /*
719: * Possibly our architecture.
720: * If we're descending, keep tabs on it.
721: */
722: if (FTS_DP != ff->fts_info && NULL != dsec)
723: arch = ff->fts_name;
1.51 schwarze 724: else
725: arch = NULL;
1.47 schwarze 726: break;
727: default:
728: if (FTS_DP == ff->fts_info || use_all)
729: break;
730: if (warnings)
731: say(path, "Extraneous directory part");
732: fts_set(f, ff, FTS_SKIP);
733: break;
1.14 schwarze 734: }
1.47 schwarze 735: }
736:
737: fts_close(f);
738: return(1);
739: }
1.1 schwarze 740:
1.47 schwarze 741: /*
742: * Add a file to the mlinks table.
743: * Do not verify that it's a "valid" looking manpage (we'll do that
744: * later).
745: *
746: * Try to infer the manual section, architecture, and page name from the
747: * path, assuming it looks like
748: *
1.98 schwarze 749: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 750: * or
751: * [./]cat<section>[/<arch>]/<name>.0
752: *
753: * See treescan() for the fts(3) version of this.
754: */
755: static void
756: filescan(const char *file)
757: {
758: char buf[PATH_MAX];
759: struct stat st;
760: struct mlink *mlink;
761: char *p, *start;
762:
763: assert(use_all);
764:
765: if (0 == strncmp(file, "./", 2))
766: file += 2;
767:
1.96 schwarze 768: /*
769: * We have to do lstat(2) before realpath(3) loses
770: * the information whether this is a symbolic link.
771: * We need to know that because for symbolic links,
772: * we want to use the orginal file name, while for
773: * regular files, we want to use the real path.
774: */
775: if (-1 == lstat(file, &st)) {
776: exitcode = (int)MANDOCLEVEL_BADARG;
777: say(file, "&lstat");
778: return;
779: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
780: exitcode = (int)MANDOCLEVEL_BADARG;
781: say(file, "Not a regular file");
782: return;
783: }
784:
785: /*
786: * We have to resolve the file name to the real path
787: * in any case for the base directory check.
788: */
1.47 schwarze 789: if (NULL == realpath(file, buf)) {
790: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 791: say(file, "&realpath");
1.47 schwarze 792: return;
1.63 schwarze 793: }
794:
795: if (strstr(buf, basedir) == buf)
796: start = buf + strlen(basedir) + 1;
797: else if (OP_TEST == op)
798: start = buf;
799: else {
1.47 schwarze 800: exitcode = (int)MANDOCLEVEL_BADARG;
801: say("", "%s: outside base directory", buf);
802: return;
1.63 schwarze 803: }
804:
1.96 schwarze 805: /*
806: * Now we are sure the file is inside our tree.
807: * If it is a symbolic link, ignore the real path
808: * and use the original name.
809: * This implies passing stuff like "cat1/../man1/foo.1"
810: * on the command line won't work. So don't do that.
811: * Note the stat(2) can still fail if the link target
812: * doesn't exist.
813: */
814: if (S_IFLNK & st.st_mode) {
815: if (-1 == stat(buf, &st)) {
816: exitcode = (int)MANDOCLEVEL_BADARG;
817: say(file, "&stat");
818: return;
819: }
1.100 schwarze 820: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
821: say(file, "Filename too long");
822: return;
823: }
1.96 schwarze 824: start = strstr(buf, basedir) == buf ?
825: buf + strlen(basedir) + 1 : buf;
1.1 schwarze 826: }
1.63 schwarze 827:
1.47 schwarze 828: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 829: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
830: sizeof(mlink->file)) {
831: say(start, "Filename too long");
832: return;
833: }
1.1 schwarze 834:
1.10 schwarze 835: /*
1.47 schwarze 836: * First try to guess our directory structure.
837: * If we find a separator, try to look for man* or cat*.
838: * If we find one of these and what's underneath is a directory,
839: * assume it's an architecture.
1.10 schwarze 840: */
1.47 schwarze 841: if (NULL != (p = strchr(start, '/'))) {
842: *p++ = '\0';
843: if (0 == strncmp(start, "man", 3)) {
844: mlink->dform = FORM_SRC;
1.51 schwarze 845: mlink->dsec = start + 3;
1.47 schwarze 846: } else if (0 == strncmp(start, "cat", 3)) {
847: mlink->dform = FORM_CAT;
1.51 schwarze 848: mlink->dsec = start + 3;
1.47 schwarze 849: }
1.10 schwarze 850:
1.47 schwarze 851: start = p;
852: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
853: *p++ = '\0';
1.51 schwarze 854: mlink->arch = start;
1.47 schwarze 855: start = p;
1.41 deraadt 856: }
1.47 schwarze 857: }
1.7 schwarze 858:
1.47 schwarze 859: /*
860: * Now check the file suffix.
861: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
862: */
863: p = strrchr(start, '\0');
864: while (p-- > start && '/' != *p && '.' != *p)
865: /* Loop. */ ;
866:
867: if ('.' == *p) {
868: *p++ = '\0';
1.51 schwarze 869: mlink->fsec = p;
1.47 schwarze 870: }
1.41 deraadt 871:
1.47 schwarze 872: /*
873: * Now try to parse the name.
874: * Use the filename portion of the path.
875: */
876: mlink->name = start;
877: if (NULL != (p = strrchr(start, '/'))) {
878: mlink->name = p + 1;
879: *p = '\0';
880: }
881: mlink_add(mlink, &st);
882: }
1.2 schwarze 883:
1.47 schwarze 884: static void
885: mlink_add(struct mlink *mlink, const struct stat *st)
886: {
887: struct inodev inodev;
888: struct mpage *mpage;
889: unsigned int slot;
890:
891: assert(NULL != mlink->file);
892:
1.51 schwarze 893: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
894: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
895: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
896: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 897:
898: if ('0' == *mlink->fsec) {
899: free(mlink->fsec);
900: mlink->fsec = mandoc_strdup(mlink->dsec);
901: mlink->fform = FORM_CAT;
902: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
903: mlink->fform = FORM_SRC;
904: else
905: mlink->fform = FORM_NONE;
1.1 schwarze 906:
1.47 schwarze 907: slot = ohash_qlookup(&mlinks, mlink->file);
908: assert(NULL == ohash_find(&mlinks, slot));
909: ohash_insert(&mlinks, slot, mlink);
910:
911: inodev.st_ino = st->st_ino;
912: inodev.st_dev = st->st_dev;
913: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
914: sizeof(struct inodev), inodev.st_ino);
915: mpage = ohash_find(&mpages, slot);
916: if (NULL == mpage) {
917: mpage = mandoc_calloc(1, sizeof(struct mpage));
918: mpage->inodev.st_ino = inodev.st_ino;
919: mpage->inodev.st_dev = inodev.st_dev;
920: ohash_insert(&mpages, slot, mpage);
921: } else
922: mlink->next = mpage->mlinks;
923: mpage->mlinks = mlink;
1.75 schwarze 924: mlink->mpage = mpage;
1.47 schwarze 925: }
1.1 schwarze 926:
1.47 schwarze 927: static void
928: mlink_free(struct mlink *mlink)
929: {
1.1 schwarze 930:
1.47 schwarze 931: free(mlink->dsec);
932: free(mlink->arch);
933: free(mlink->name);
934: free(mlink->fsec);
935: free(mlink);
936: }
1.1 schwarze 937:
1.47 schwarze 938: static void
939: mpages_free(void)
940: {
941: struct mpage *mpage;
942: struct mlink *mlink;
943: unsigned int slot;
944:
945: mpage = ohash_first(&mpages, &slot);
946: while (NULL != mpage) {
947: while (NULL != (mlink = mpage->mlinks)) {
948: mpage->mlinks = mlink->next;
949: mlink_free(mlink);
950: }
951: free(mpage->sec);
952: free(mpage->arch);
953: free(mpage->title);
954: free(mpage->desc);
955: free(mpage);
956: mpage = ohash_next(&mpages, &slot);
957: }
958: }
1.1 schwarze 959:
1.47 schwarze 960: /*
961: * For each mlink to the mpage, check whether the path looks like
962: * it is formatted, and if it does, check whether a source manual
963: * exists by the same name, ignoring the suffix.
964: * If both conditions hold, drop the mlink.
965: */
966: static void
967: mlinks_undupe(struct mpage *mpage)
968: {
969: char buf[PATH_MAX];
970: struct mlink **prev;
971: struct mlink *mlink;
972: char *bufp;
973:
974: mpage->form = FORM_CAT;
975: prev = &mpage->mlinks;
976: while (NULL != (mlink = *prev)) {
977: if (FORM_CAT != mlink->dform) {
978: mpage->form = FORM_NONE;
979: goto nextlink;
980: }
1.100 schwarze 981: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 982: bufp = strstr(buf, "cat");
983: assert(NULL != bufp);
984: memcpy(bufp, "man", 3);
985: if (NULL != (bufp = strrchr(buf, '.')))
986: *++bufp = '\0';
1.100 schwarze 987: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 988: if (NULL == ohash_find(&mlinks,
1.98 schwarze 989: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 990: goto nextlink;
991: if (warnings)
992: say(mlink->file, "Man source exists: %s", buf);
993: if (use_all)
994: goto nextlink;
995: *prev = mlink->next;
996: mlink_free(mlink);
997: continue;
998: nextlink:
999: prev = &(*prev)->next;
1.1 schwarze 1000: }
1.47 schwarze 1001: }
1.1 schwarze 1002:
1.88 schwarze 1003: static void
1.50 schwarze 1004: mlink_check(struct mpage *mpage, struct mlink *mlink)
1005: {
1.88 schwarze 1006: struct str *str;
1007: unsigned int slot;
1.50 schwarze 1008:
1009: /*
1010: * Check whether the manual section given in a file
1011: * agrees with the directory where the file is located.
1012: * Some manuals have suffixes like (3p) on their
1013: * section number either inside the file or in the
1014: * directory name, some are linked into more than one
1015: * section, like encrypt(1) = makekey(8).
1016: */
1017:
1018: if (FORM_SRC == mpage->form &&
1.88 schwarze 1019: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1020: say(mlink->file, "Section \"%s\" manual in %s directory",
1021: mpage->sec, mlink->dsec);
1022:
1023: /*
1024: * Manual page directories exist for each kernel
1025: * architecture as returned by machine(1).
1026: * However, many manuals only depend on the
1027: * application architecture as returned by arch(1).
1028: * For example, some (2/ARM) manuals are shared
1029: * across the "armish" and "zaurus" kernel
1030: * architectures.
1031: * A few manuals are even shared across completely
1032: * different architectures, for example fdformat(1)
1033: * on amd64, i386, sparc, and sparc64.
1034: */
1035:
1.88 schwarze 1036: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1037: say(mlink->file, "Architecture \"%s\" manual in "
1038: "\"%s\" directory", mpage->arch, mlink->arch);
1039:
1.88 schwarze 1040: /*
1041: * XXX
1.90 schwarze 1042: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1043: */
1044:
1045: if (FORM_CAT == mpage->form)
1046: return;
1047:
1048: /*
1049: * Check whether this mlink
1050: * appears as a name in the NAME section.
1051: */
1.50 schwarze 1052:
1.90 schwarze 1053: slot = ohash_qlookup(&names, mlink->name);
1054: str = ohash_find(&names, slot);
1.88 schwarze 1055: assert(NULL != str);
1.90 schwarze 1056: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1057: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1058: }
1059:
1.47 schwarze 1060: /*
1061: * Run through the files in the global vector "mpages"
1062: * and add them to the database specified in "basedir".
1063: *
1064: * This handles the parsing scheme itself, using the cues of directory
1065: * and filename to determine whether the file is parsable or not.
1066: */
1067: static void
1.58 schwarze 1068: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1069: {
1.70 schwarze 1070: char any[] = "any";
1.58 schwarze 1071: struct ohash_info str_info;
1.81 schwarze 1072: int fd[2];
1.75 schwarze 1073: struct mpage *mpage, *mpage_dest;
1074: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1075: struct mdoc *mdoc;
1076: struct man *man;
1.75 schwarze 1077: char *sodest;
1.69 schwarze 1078: char *cp;
1.81 schwarze 1079: pid_t child_pid;
1.88 schwarze 1080: int status;
1.58 schwarze 1081: unsigned int pslot;
1.47 schwarze 1082: enum mandoclevel lvl;
1083:
1084: str_info.alloc = hash_alloc;
1085: str_info.halloc = hash_halloc;
1086: str_info.hfree = hash_free;
1087: str_info.key_offset = offsetof(struct str, key);
1088:
1.64 schwarze 1089: if (0 == nodb)
1090: SQL_EXEC("BEGIN TRANSACTION");
1091:
1.47 schwarze 1092: mpage = ohash_first(&mpages, &pslot);
1093: while (NULL != mpage) {
1094: mlinks_undupe(mpage);
1095: if (NULL == mpage->mlinks) {
1096: mpage = ohash_next(&mpages, &pslot);
1097: continue;
1098: }
1.1 schwarze 1099:
1.90 schwarze 1100: name_mask = NAME_MASK;
1101: ohash_init(&names, 4, &str_info);
1.47 schwarze 1102: ohash_init(&strings, 6, &str_info);
1103: mparse_reset(mp);
1104: mdoc = NULL;
1105: man = NULL;
1.81 schwarze 1106: sodest = NULL;
1107: child_pid = 0;
1108: fd[0] = -1;
1109: fd[1] = -1;
1110:
1111: if (mpage->mlinks->gzip) {
1112: if (-1 == pipe(fd)) {
1113: exitcode = (int)MANDOCLEVEL_SYSERR;
1114: say(mpage->mlinks->file, "&pipe gunzip");
1115: goto nextpage;
1116: }
1117: switch (child_pid = fork()) {
1.98 schwarze 1118: case -1:
1.81 schwarze 1119: exitcode = (int)MANDOCLEVEL_SYSERR;
1120: say(mpage->mlinks->file, "&fork gunzip");
1121: child_pid = 0;
1122: close(fd[1]);
1123: close(fd[0]);
1124: goto nextpage;
1.98 schwarze 1125: case 0:
1.81 schwarze 1126: close(fd[0]);
1127: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1128: say(mpage->mlinks->file,
1129: "&dup gunzip");
1130: exit(1);
1131: }
1132: execlp("gunzip", "gunzip", "-c",
1133: mpage->mlinks->file, NULL);
1134: say(mpage->mlinks->file, "&exec gunzip");
1135: exit(1);
1136: default:
1137: close(fd[1]);
1138: break;
1139: }
1140: }
1.11 schwarze 1141:
1142: /*
1.24 schwarze 1143: * Try interpreting the file as mdoc(7) or man(7)
1144: * source code, unless it is already known to be
1145: * formatted. Fall back to formatted mode.
1.11 schwarze 1146: */
1.47 schwarze 1147: if (FORM_CAT != mpage->mlinks->dform ||
1148: FORM_CAT != mpage->mlinks->fform) {
1.81 schwarze 1149: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.47 schwarze 1150: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1151: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1152: }
1.11 schwarze 1153:
1.75 schwarze 1154: if (NULL != sodest) {
1155: mlink_dest = ohash_find(&mlinks,
1156: ohash_qlookup(&mlinks, sodest));
1157: if (NULL != mlink_dest) {
1158:
1159: /* The .so target exists. */
1160:
1161: mpage_dest = mlink_dest->mpage;
1162: mlink = mpage->mlinks;
1163: while (1) {
1164: mlink->mpage = mpage_dest;
1165:
1166: /*
1167: * If the target was already
1168: * processed, add the links
1169: * to the database now.
1170: * Otherwise, this will
1171: * happen when we come
1172: * to the target.
1173: */
1174:
1.94 schwarze 1175: if (mpage_dest->pageid)
1.75 schwarze 1176: dbadd_mlink(mlink);
1177:
1178: if (NULL == mlink->next)
1179: break;
1180: mlink = mlink->next;
1181: }
1182:
1183: /* Move all links to the target. */
1184:
1185: mlink->next = mlink_dest->next;
1186: mlink_dest->next = mpage->mlinks;
1187: mpage->mlinks = NULL;
1188: }
1.81 schwarze 1189: goto nextpage;
1.75 schwarze 1190: } else if (NULL != mdoc) {
1.47 schwarze 1191: mpage->form = FORM_SRC;
1192: mpage->sec =
1193: mandoc_strdup(mdoc_meta(mdoc)->msec);
1194: mpage->arch = mdoc_meta(mdoc)->arch;
1195: mpage->arch = mandoc_strdup(
1196: NULL == mpage->arch ? "" : mpage->arch);
1197: mpage->title =
1198: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1199: } else if (NULL != man) {
1.47 schwarze 1200: mpage->form = FORM_SRC;
1201: mpage->sec =
1202: mandoc_strdup(man_meta(man)->msec);
1203: mpage->arch =
1204: mandoc_strdup(mpage->mlinks->arch);
1205: mpage->title =
1206: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1207: } else {
1.47 schwarze 1208: mpage->form = FORM_CAT;
1209: mpage->sec =
1210: mandoc_strdup(mpage->mlinks->dsec);
1211: mpage->arch =
1212: mandoc_strdup(mpage->mlinks->arch);
1213: mpage->title =
1214: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1215: }
1.54 schwarze 1216: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1217: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1218: any : mpage->arch, TYPE_arch);
1.1 schwarze 1219:
1.54 schwarze 1220: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1221: if ('\0' != *mlink->dsec)
1222: putkey(mpage, mlink->dsec, TYPE_sec);
1223: if ('\0' != *mlink->fsec)
1224: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1225: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1226: any : mlink->arch, TYPE_arch);
1.90 schwarze 1227: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1228: }
1.41 deraadt 1229:
1.92 schwarze 1230: assert(NULL == mpage->desc);
1.47 schwarze 1231: if (NULL != mdoc) {
1232: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.90 schwarze 1233: putkey(mpage, cp, NAME_HEAD);
1.47 schwarze 1234: parse_mdoc(mpage, mdoc_node(mdoc));
1235: } else if (NULL != man)
1236: parse_man(mpage, man_node(man));
1237: else
1.81 schwarze 1238: parse_cat(mpage, fd[0]);
1.92 schwarze 1239: if (NULL == mpage->desc)
1240: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1241:
1242: if (warnings && !use_all)
1243: for (mlink = mpage->mlinks; mlink;
1244: mlink = mlink->next)
1245: mlink_check(mpage, mlink);
1.6 schwarze 1246:
1.62 schwarze 1247: dbadd(mpage, mc);
1.81 schwarze 1248:
1249: nextpage:
1250: if (child_pid) {
1251: if (-1 == waitpid(child_pid, &status, 0)) {
1252: exitcode = (int)MANDOCLEVEL_SYSERR;
1253: say(mpage->mlinks->file, "&wait gunzip");
1254: } else if (WIFSIGNALED(status)) {
1255: exitcode = (int)MANDOCLEVEL_SYSERR;
1256: say(mpage->mlinks->file,
1257: "gunzip died from signal %d",
1258: WTERMSIG(status));
1259: } else if (WEXITSTATUS(status)) {
1260: exitcode = (int)MANDOCLEVEL_SYSERR;
1261: say(mpage->mlinks->file,
1262: "gunzip failed with code %d",
1263: WEXITSTATUS(status));
1264: }
1265: }
1.47 schwarze 1266: ohash_delete(&strings);
1.90 schwarze 1267: ohash_delete(&names);
1.47 schwarze 1268: mpage = ohash_next(&mpages, &pslot);
1269: }
1.64 schwarze 1270:
1271: if (0 == nodb)
1272: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1273: }
1274:
1275: static void
1276: names_check(void)
1277: {
1278: sqlite3_stmt *stmt;
1279: const char *name, *sec, *arch, *key;
1280: int irc;
1281:
1282: sqlite3_prepare_v2(db,
1283: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1284: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1285: "WHERE bits & ? AND NOT EXISTS ("
1286: "SELECT pageid FROM mlinks "
1.90 schwarze 1287: "WHERE mlinks.pageid == names.pageid "
1288: "AND mlinks.name == names.name"
1.87 schwarze 1289: ")"
1290: ") JOIN ("
1.99 schwarze 1291: "SELECT sec, arch, name, pageid FROM mlinks "
1292: "GROUP BY pageid"
1.87 schwarze 1293: ") USING (pageid);",
1294: -1, &stmt, NULL);
1295:
1.91 schwarze 1296: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1297: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1298:
1299: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1300: name = sqlite3_column_text(stmt, 0);
1301: sec = sqlite3_column_text(stmt, 1);
1302: arch = sqlite3_column_text(stmt, 2);
1303: key = sqlite3_column_text(stmt, 3);
1304: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1305: '\0' == *arch ? "" : "/",
1306: '\0' == *arch ? "" : arch, key);
1307: }
1308: sqlite3_finalize(stmt);
1.47 schwarze 1309: }
1.6 schwarze 1310:
1.47 schwarze 1311: static void
1.81 schwarze 1312: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1313: {
1314: FILE *stream;
1315: char *line, *p, *title;
1316: size_t len, plen, titlesz;
1.1 schwarze 1317:
1.81 schwarze 1318: stream = (-1 == fd) ?
1319: fopen(mpage->mlinks->file, "r") :
1320: fdopen(fd, "r");
1321: if (NULL == stream) {
1.47 schwarze 1322: if (warnings)
1.80 schwarze 1323: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1324: return;
1325: }
1.1 schwarze 1326:
1.47 schwarze 1327: /* Skip to first blank line. */
1.1 schwarze 1328:
1.47 schwarze 1329: while (NULL != (line = fgetln(stream, &len)))
1330: if ('\n' == *line)
1331: break;
1.1 schwarze 1332:
1.47 schwarze 1333: /*
1334: * Assume the first line that is not indented
1335: * is the first section header. Skip to it.
1336: */
1.1 schwarze 1337:
1.47 schwarze 1338: while (NULL != (line = fgetln(stream, &len)))
1339: if ('\n' != *line && ' ' != *line)
1340: break;
1.98 schwarze 1341:
1.47 schwarze 1342: /*
1343: * Read up until the next section into a buffer.
1344: * Strip the leading and trailing newline from each read line,
1345: * appending a trailing space.
1346: * Ignore empty (whitespace-only) lines.
1347: */
1.28 schwarze 1348:
1.47 schwarze 1349: titlesz = 0;
1350: title = NULL;
1.38 schwarze 1351:
1.47 schwarze 1352: while (NULL != (line = fgetln(stream, &len))) {
1353: if (' ' != *line || '\n' != line[len - 1])
1354: break;
1355: while (len > 0 && isspace((unsigned char)*line)) {
1356: line++;
1357: len--;
1358: }
1359: if (1 == len)
1360: continue;
1361: title = mandoc_realloc(title, titlesz + len);
1362: memcpy(title + titlesz, line, len);
1363: titlesz += len;
1364: title[titlesz - 1] = ' ';
1365: }
1.28 schwarze 1366:
1.47 schwarze 1367: /*
1368: * If no page content can be found, or the input line
1369: * is already the next section header, or there is no
1370: * trailing newline, reuse the page title as the page
1371: * description.
1372: */
1.1 schwarze 1373:
1.47 schwarze 1374: if (NULL == title || '\0' == *title) {
1375: if (warnings)
1376: say(mpage->mlinks->file,
1377: "Cannot find NAME section");
1378: fclose(stream);
1379: free(title);
1380: return;
1381: }
1.24 schwarze 1382:
1.47 schwarze 1383: title = mandoc_realloc(title, titlesz + 1);
1384: title[titlesz] = '\0';
1.24 schwarze 1385:
1.47 schwarze 1386: /*
1387: * Skip to the first dash.
1388: * Use the remaining line as the description (no more than 70
1389: * bytes).
1390: */
1.28 schwarze 1391:
1.47 schwarze 1392: if (NULL != (p = strstr(title, "- "))) {
1393: for (p += 2; ' ' == *p || '\b' == *p; p++)
1394: /* Skip to next word. */ ;
1395: } else {
1396: if (warnings)
1397: say(mpage->mlinks->file,
1398: "No dash in title line");
1399: p = title;
1400: }
1.1 schwarze 1401:
1.47 schwarze 1402: plen = strlen(p);
1.1 schwarze 1403:
1.47 schwarze 1404: /* Strip backspace-encoding from line. */
1.1 schwarze 1405:
1.47 schwarze 1406: while (NULL != (line = memchr(p, '\b', plen))) {
1407: len = line - p;
1408: if (0 == len) {
1409: memmove(line, line + 1, plen--);
1410: continue;
1.98 schwarze 1411: }
1.47 schwarze 1412: memmove(line - 1, line + 1, plen - len);
1413: plen -= 2;
1414: }
1.1 schwarze 1415:
1.47 schwarze 1416: mpage->desc = mandoc_strdup(p);
1417: fclose(stream);
1418: free(title);
1419: }
1.16 schwarze 1420:
1.47 schwarze 1421: /*
1422: * Put a type/word pair into the word database for this particular file.
1423: */
1424: static void
1.69 schwarze 1425: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1426: {
1.69 schwarze 1427: char *cp;
1.37 schwarze 1428:
1.47 schwarze 1429: assert(NULL != value);
1.69 schwarze 1430: if (TYPE_arch == type)
1431: for (cp = value; *cp; cp++)
1432: if (isupper((unsigned char)*cp))
1433: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1434: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1435: }
1436:
1437: /*
1.47 schwarze 1438: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1439: */
1440: static void
1.47 schwarze 1441: putmdockey(const struct mpage *mpage,
1442: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1443: {
1.16 schwarze 1444:
1.47 schwarze 1445: for ( ; NULL != n; n = n->next) {
1446: if (NULL != n->child)
1447: putmdockey(mpage, n->child, m);
1448: if (MDOC_TEXT == n->type)
1449: putkey(mpage, n->string, m);
1450: }
1451: }
1.16 schwarze 1452:
1.47 schwarze 1453: static void
1454: parse_man(struct mpage *mpage, const struct man_node *n)
1455: {
1456: const struct man_node *head, *body;
1.78 schwarze 1457: char *start, *title;
1.47 schwarze 1458: char byte;
1.78 schwarze 1459: size_t sz;
1.16 schwarze 1460:
1.47 schwarze 1461: if (NULL == n)
1462: return;
1.16 schwarze 1463:
1.47 schwarze 1464: /*
1465: * We're only searching for one thing: the first text child in
1466: * the BODY of a NAME section. Since we don't keep track of
1467: * sections in -man, run some hoops to find out whether we're in
1468: * the correct section or not.
1469: */
1.16 schwarze 1470:
1.47 schwarze 1471: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1472: body = n;
1473: assert(body->parent);
1474: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1475: 1 == head->nchild &&
1476: NULL != (head = (head->child)) &&
1477: MAN_TEXT == head->type &&
1478: 0 == strcmp(head->string, "NAME") &&
1479: NULL != body->child) {
1.2 schwarze 1480:
1.47 schwarze 1481: /*
1482: * Suck the entire NAME section into memory.
1483: * Yes, we might run away.
1484: * But too many manuals have big, spread-out
1485: * NAME sections over many lines.
1486: */
1.2 schwarze 1487:
1.78 schwarze 1488: title = NULL;
1489: man_deroff(&title, body);
1.47 schwarze 1490: if (NULL == title)
1491: return;
1.16 schwarze 1492:
1.98 schwarze 1493: /*
1.47 schwarze 1494: * Go through a special heuristic dance here.
1495: * Conventionally, one or more manual names are
1496: * comma-specified prior to a whitespace, then a
1497: * dash, then a description. Try to puzzle out
1498: * the name parts here.
1499: */
1.16 schwarze 1500:
1.78 schwarze 1501: start = title;
1.47 schwarze 1502: for ( ;; ) {
1503: sz = strcspn(start, " ,");
1504: if ('\0' == start[sz])
1505: break;
1.1 schwarze 1506:
1.47 schwarze 1507: byte = start[sz];
1508: start[sz] = '\0';
1.67 schwarze 1509:
1510: /*
1511: * Assume a stray trailing comma in the
1512: * name list if a name begins with a dash.
1513: */
1514:
1515: if ('-' == start[0] ||
1516: ('\\' == start[0] && '-' == start[1]))
1517: break;
1.1 schwarze 1518:
1.90 schwarze 1519: putkey(mpage, start, NAME_TITLE);
1.1 schwarze 1520:
1.47 schwarze 1521: if (' ' == byte) {
1522: start += sz + 1;
1523: break;
1524: }
1.1 schwarze 1525:
1.47 schwarze 1526: assert(',' == byte);
1527: start += sz + 1;
1528: while (' ' == *start)
1529: start++;
1530: }
1.1 schwarze 1531:
1.78 schwarze 1532: if (start == title) {
1.90 schwarze 1533: putkey(mpage, start, NAME_TITLE);
1.47 schwarze 1534: free(title);
1535: return;
1536: }
1.1 schwarze 1537:
1.47 schwarze 1538: while (isspace((unsigned char)*start))
1539: start++;
1.1 schwarze 1540:
1.47 schwarze 1541: if (0 == strncmp(start, "-", 1))
1542: start += 1;
1543: else if (0 == strncmp(start, "\\-\\-", 4))
1544: start += 4;
1545: else if (0 == strncmp(start, "\\-", 2))
1546: start += 2;
1547: else if (0 == strncmp(start, "\\(en", 4))
1548: start += 4;
1549: else if (0 == strncmp(start, "\\(em", 4))
1550: start += 4;
1.1 schwarze 1551:
1.47 schwarze 1552: while (' ' == *start)
1553: start++;
1.1 schwarze 1554:
1.47 schwarze 1555: mpage->desc = mandoc_strdup(start);
1556: free(title);
1557: return;
1558: }
1559: }
1.1 schwarze 1560:
1.47 schwarze 1561: for (n = n->child; n; n = n->next) {
1562: if (NULL != mpage->desc)
1563: break;
1564: parse_man(mpage, n);
1.1 schwarze 1565: }
1566: }
1567:
1568: static void
1.47 schwarze 1569: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1570: {
1571:
1.47 schwarze 1572: assert(NULL != n);
1573: for (n = n->child; NULL != n; n = n->next) {
1574: switch (n->type) {
1.98 schwarze 1575: case MDOC_ELEM:
1.47 schwarze 1576: /* FALLTHROUGH */
1.98 schwarze 1577: case MDOC_BLOCK:
1.47 schwarze 1578: /* FALLTHROUGH */
1.98 schwarze 1579: case MDOC_HEAD:
1.47 schwarze 1580: /* FALLTHROUGH */
1.98 schwarze 1581: case MDOC_BODY:
1.47 schwarze 1582: /* FALLTHROUGH */
1.98 schwarze 1583: case MDOC_TAIL:
1.47 schwarze 1584: if (NULL != mdocs[n->tok].fp)
1585: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1586: break;
1587: if (mdocs[n->tok].mask)
1588: putmdockey(mpage, n->child,
1589: mdocs[n->tok].mask);
1590: break;
1591: default:
1592: assert(MDOC_ROOT != n->type);
1593: continue;
1594: }
1595: if (NULL != n->child)
1596: parse_mdoc(mpage, n);
1.1 schwarze 1597: }
1598: }
1599:
1.19 schwarze 1600: static int
1.47 schwarze 1601: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1602: {
1603: const char *start, *end;
1604: size_t sz;
1.19 schwarze 1605:
1.47 schwarze 1606: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1607: NULL == (n = n->child) ||
1608: MDOC_TEXT != n->type)
1.19 schwarze 1609: return(0);
1.1 schwarze 1610:
1611: /*
1612: * Only consider those `Fd' macro fields that begin with an
1613: * "inclusion" token (versus, e.g., #define).
1614: */
1.47 schwarze 1615:
1.1 schwarze 1616: if (strcmp("#include", n->string))
1.19 schwarze 1617: return(0);
1.1 schwarze 1618:
1619: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1620: return(0);
1.1 schwarze 1621:
1622: /*
1623: * Strip away the enclosing angle brackets and make sure we're
1624: * not zero-length.
1625: */
1626:
1627: start = n->string;
1628: if ('<' == *start || '"' == *start)
1629: start++;
1630:
1631: if (0 == (sz = strlen(start)))
1.19 schwarze 1632: return(0);
1.1 schwarze 1633:
1634: end = &start[(int)sz - 1];
1635: if ('>' == *end || '"' == *end)
1636: end--;
1637:
1.47 schwarze 1638: if (end > start)
1639: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1640: return(0);
1.1 schwarze 1641: }
1642:
1.19 schwarze 1643: static int
1.47 schwarze 1644: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1645: {
1.69 schwarze 1646: char *cp;
1.1 schwarze 1647:
1.47 schwarze 1648: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1649: return(0);
1650:
1.98 schwarze 1651: /*
1.47 schwarze 1652: * Parse: .Fn "struct type *name" "char *arg".
1.98 schwarze 1653: * First strip away pointer symbol.
1.47 schwarze 1654: * Then store the function name, then type.
1.98 schwarze 1655: * Finally, store the arguments.
1.47 schwarze 1656: */
1.1 schwarze 1657:
1.47 schwarze 1658: if (NULL == (cp = strrchr(n->string, ' ')))
1659: cp = n->string;
1.1 schwarze 1660:
1661: while ('*' == *cp)
1662: cp++;
1663:
1.47 schwarze 1664: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1665:
1.47 schwarze 1666: if (n->string < cp)
1667: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1668:
1.47 schwarze 1669: for (n = n->next; NULL != n; n = n->next)
1670: if (MDOC_TEXT == n->type)
1671: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1672:
1673: return(0);
1.1 schwarze 1674: }
1675:
1.19 schwarze 1676: static int
1.47 schwarze 1677: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1678: {
1.47 schwarze 1679: char *cp;
1.1 schwarze 1680:
1681: if (NULL == (n = n->child))
1.19 schwarze 1682: return(0);
1.1 schwarze 1683:
1.47 schwarze 1684: if (NULL == n->next) {
1685: putkey(mpage, n->string, TYPE_Xr);
1686: return(0);
1687: }
1.1 schwarze 1688:
1.77 schwarze 1689: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1690: putkey(mpage, cp, TYPE_Xr);
1691: free(cp);
1692: return(0);
1.1 schwarze 1693: }
1694:
1.19 schwarze 1695: static int
1.47 schwarze 1696: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1697: {
1698:
1.79 schwarze 1699: if (MDOC_BODY == n->type)
1700: mdoc_deroff(&mpage->desc, n);
1701: return(0);
1.1 schwarze 1702: }
1703:
1.19 schwarze 1704: static int
1.47 schwarze 1705: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1706: {
1707:
1.86 schwarze 1708: if (SEC_NAME == n->sec)
1.90 schwarze 1709: putmdockey(mpage, n->child, NAME_TITLE);
1.86 schwarze 1710: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.90 schwarze 1711: putmdockey(mpage, n->child, NAME_SYN);
1.86 schwarze 1712: return(0);
1.1 schwarze 1713: }
1714:
1.19 schwarze 1715: static int
1.47 schwarze 1716: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1717: {
1718:
1.19 schwarze 1719: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1720: }
1721:
1.47 schwarze 1722: static int
1723: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1724: {
1725:
1.47 schwarze 1726: return(MDOC_HEAD == n->type);
1727: }
1.1 schwarze 1728:
1.47 schwarze 1729: static int
1730: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1731: {
1.1 schwarze 1732:
1.47 schwarze 1733: return(MDOC_BODY == n->type);
1.1 schwarze 1734: }
1735:
1.47 schwarze 1736: /*
1737: * Add a string to the hash table for the current manual.
1738: * Each string has a bitmask telling which macros it belongs to.
1739: * When we finish the manual, we'll dump the table.
1740: */
1.1 schwarze 1741: static void
1.47 schwarze 1742: putkeys(const struct mpage *mpage,
1743: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1744: {
1.90 schwarze 1745: struct ohash *htab;
1.47 schwarze 1746: struct str *s;
1.68 schwarze 1747: const char *end;
1.47 schwarze 1748: unsigned int slot;
1.68 schwarze 1749: int i;
1.1 schwarze 1750:
1.47 schwarze 1751: if (0 == sz)
1752: return;
1.68 schwarze 1753:
1.90 schwarze 1754: if (TYPE_Nm & v) {
1755: htab = &names;
1756: v &= name_mask;
1757: name_mask &= ~NAME_FIRST;
1758: if (debug > 1)
1759: say(mpage->mlinks->file,
1760: "Adding name %*s", sz, cp);
1761: } else {
1762: htab = &strings;
1763: if (debug > 1)
1764: for (i = 0; i < mansearch_keymax; i++)
1765: if (1 << i & v)
1766: say(mpage->mlinks->file,
1767: "Adding key %s=%*s",
1768: mansearch_keynames[i], sz, cp);
1.68 schwarze 1769: }
1.47 schwarze 1770:
1771: end = cp + sz;
1.90 schwarze 1772: slot = ohash_qlookupi(htab, cp, &end);
1773: s = ohash_find(htab, slot);
1.1 schwarze 1774:
1.47 schwarze 1775: if (NULL != s && mpage == s->mpage) {
1776: s->mask |= v;
1.1 schwarze 1777: return;
1.47 schwarze 1778: } else if (NULL == s) {
1.101 schwarze 1779: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1780: memcpy(s->key, cp, sz);
1.90 schwarze 1781: ohash_insert(htab, slot, s);
1.47 schwarze 1782: }
1783: s->mpage = mpage;
1784: s->mask = v;
1.1 schwarze 1785: }
1786:
1787: /*
1.47 schwarze 1788: * Take a Unicode codepoint and produce its UTF-8 encoding.
1789: * This isn't the best way to do this, but it works.
1790: * The magic numbers are from the UTF-8 packaging.
1791: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1792: */
1.47 schwarze 1793: static size_t
1794: utf8(unsigned int cp, char out[7])
1.1 schwarze 1795: {
1.47 schwarze 1796: size_t rc;
1.1 schwarze 1797:
1.47 schwarze 1798: rc = 0;
1799: if (cp <= 0x0000007F) {
1800: rc = 1;
1801: out[0] = (char)cp;
1802: } else if (cp <= 0x000007FF) {
1803: rc = 2;
1804: out[0] = (cp >> 6 & 31) | 192;
1805: out[1] = (cp & 63) | 128;
1806: } else if (cp <= 0x0000FFFF) {
1807: rc = 3;
1808: out[0] = (cp >> 12 & 15) | 224;
1809: out[1] = (cp >> 6 & 63) | 128;
1810: out[2] = (cp & 63) | 128;
1811: } else if (cp <= 0x001FFFFF) {
1812: rc = 4;
1813: out[0] = (cp >> 18 & 7) | 240;
1814: out[1] = (cp >> 12 & 63) | 128;
1815: out[2] = (cp >> 6 & 63) | 128;
1816: out[3] = (cp & 63) | 128;
1817: } else if (cp <= 0x03FFFFFF) {
1818: rc = 5;
1819: out[0] = (cp >> 24 & 3) | 248;
1820: out[1] = (cp >> 18 & 63) | 128;
1821: out[2] = (cp >> 12 & 63) | 128;
1822: out[3] = (cp >> 6 & 63) | 128;
1823: out[4] = (cp & 63) | 128;
1824: } else if (cp <= 0x7FFFFFFF) {
1825: rc = 6;
1826: out[0] = (cp >> 30 & 1) | 252;
1827: out[1] = (cp >> 24 & 63) | 128;
1828: out[2] = (cp >> 18 & 63) | 128;
1829: out[3] = (cp >> 12 & 63) | 128;
1830: out[4] = (cp >> 6 & 63) | 128;
1831: out[5] = (cp & 63) | 128;
1832: } else
1833: return(0);
1.19 schwarze 1834:
1.47 schwarze 1835: out[rc] = '\0';
1836: return(rc);
1.1 schwarze 1837: }
1838:
1.47 schwarze 1839: /*
1.53 schwarze 1840: * Store the rendered version of a key, or alias the pointer
1841: * if the key contains no escape sequences.
1.47 schwarze 1842: */
1843: static void
1.53 schwarze 1844: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1845: {
1.47 schwarze 1846: size_t sz, bsz, pos;
1.71 schwarze 1847: char utfbuf[7], res[6];
1.47 schwarze 1848: char *buf;
1849: const char *seq, *cpp, *val;
1850: int len, u;
1851: enum mandoc_esc esc;
1852:
1.53 schwarze 1853: assert(NULL == key->rendered);
1.47 schwarze 1854:
1855: res[0] = '\\';
1856: res[1] = '\t';
1857: res[2] = ASCII_NBRSP;
1858: res[3] = ASCII_HYPH;
1.71 schwarze 1859: res[4] = ASCII_BREAK;
1860: res[5] = '\0';
1.1 schwarze 1861:
1.47 schwarze 1862: val = key->key;
1863: bsz = strlen(val);
1.1 schwarze 1864:
1865: /*
1.47 schwarze 1866: * Pre-check: if we have no stop-characters, then set the
1867: * pointer as ourselvse and get out of here.
1.1 schwarze 1868: */
1.47 schwarze 1869: if (strcspn(val, res) == bsz) {
1.53 schwarze 1870: key->rendered = key->key;
1.47 schwarze 1871: return;
1.98 schwarze 1872: }
1.1 schwarze 1873:
1.47 schwarze 1874: /* Pre-allocate by the length of the input */
1.39 schwarze 1875:
1.47 schwarze 1876: buf = mandoc_malloc(++bsz);
1877: pos = 0;
1.39 schwarze 1878:
1.47 schwarze 1879: while ('\0' != *val) {
1880: /*
1881: * Halt on the first escape sequence.
1882: * This also halts on the end of string, in which case
1883: * we just copy, fallthrough, and exit the loop.
1884: */
1885: if ((sz = strcspn(val, res)) > 0) {
1886: memcpy(&buf[pos], val, sz);
1887: pos += sz;
1888: val += sz;
1889: }
1.39 schwarze 1890:
1.71 schwarze 1891: switch (*val) {
1.98 schwarze 1892: case ASCII_HYPH:
1.47 schwarze 1893: buf[pos++] = '-';
1894: val++;
1895: continue;
1.98 schwarze 1896: case '\t':
1.71 schwarze 1897: /* FALLTHROUGH */
1.98 schwarze 1898: case ASCII_NBRSP:
1.47 schwarze 1899: buf[pos++] = ' ';
1900: val++;
1.71 schwarze 1901: /* FALLTHROUGH */
1.98 schwarze 1902: case ASCII_BREAK:
1.47 schwarze 1903: continue;
1.71 schwarze 1904: default:
1905: break;
1906: }
1907: if ('\\' != *val)
1.47 schwarze 1908: break;
1.39 schwarze 1909:
1.47 schwarze 1910: /* Read past the slash. */
1.39 schwarze 1911:
1.47 schwarze 1912: val++;
1.39 schwarze 1913:
1.47 schwarze 1914: /*
1915: * Parse the escape sequence and see if it's a
1916: * predefined character or special character.
1917: */
1.52 schwarze 1918:
1.98 schwarze 1919: esc = mandoc_escape((const char **)&val,
1920: &seq, &len);
1.47 schwarze 1921: if (ESCAPE_ERROR == esc)
1922: break;
1923: if (ESCAPE_SPECIAL != esc)
1924: continue;
1.39 schwarze 1925:
1.47 schwarze 1926: /*
1.52 schwarze 1927: * Render the special character
1928: * as either UTF-8 or ASCII.
1.47 schwarze 1929: */
1.52 schwarze 1930:
1931: if (write_utf8) {
1932: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1933: continue;
1934: cpp = utfbuf;
1935: if (0 == (sz = utf8(u, utfbuf)))
1936: continue;
1937: sz = strlen(cpp);
1938: } else {
1939: cpp = mchars_spec2str(mc, seq, len, &sz);
1940: if (NULL == cpp)
1941: continue;
1942: if (ASCII_NBRSP == *cpp) {
1943: cpp = " ";
1944: sz = 1;
1945: }
1946: }
1.1 schwarze 1947:
1.47 schwarze 1948: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1949:
1.47 schwarze 1950: bsz += sz;
1951: buf = mandoc_realloc(buf, bsz);
1952: memcpy(&buf[pos], cpp, sz);
1953: pos += sz;
1.1 schwarze 1954: }
1955:
1.47 schwarze 1956: buf[pos] = '\0';
1.53 schwarze 1957: key->rendered = buf;
1.1 schwarze 1958: }
1959:
1.75 schwarze 1960: static void
1961: dbadd_mlink(const struct mlink *mlink)
1962: {
1963: size_t i;
1964:
1965: i = 1;
1966: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1967: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1968: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1969: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1970: SQL_STEP(stmts[STMT_INSERT_LINK]);
1971: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1972: }
1973:
1.11 schwarze 1974: /*
1.47 schwarze 1975: * Flush the current page's terms (and their bits) into the database.
1976: * Wrap the entire set of additions in a transaction to make sqlite be a
1977: * little faster.
1.53 schwarze 1978: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1979: */
1980: static void
1.75 schwarze 1981: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1982: {
1.47 schwarze 1983: struct mlink *mlink;
1984: struct str *key;
1985: size_t i;
1986: unsigned int slot;
1987:
1.85 schwarze 1988: mlink = mpage->mlinks;
1.11 schwarze 1989:
1.85 schwarze 1990: if (nodb) {
1.104 schwarze 1991: for (key = ohash_first(&names, &slot); NULL != key;
1992: key = ohash_next(&names, &slot)) {
1993: if (key->rendered != key->key)
1994: free(key->rendered);
1995: free(key);
1996: }
1997: for (key = ohash_first(&strings, &slot); NULL != key;
1998: key = ohash_next(&strings, &slot)) {
1999: if (key->rendered != key->key)
2000: free(key->rendered);
2001: free(key);
2002: }
1.102 schwarze 2003: if (0 == debug)
2004: return;
1.85 schwarze 2005: while (NULL != mlink) {
2006: fputs(mlink->name, stdout);
2007: if (NULL == mlink->next ||
2008: strcmp(mlink->dsec, mlink->next->dsec) ||
2009: strcmp(mlink->fsec, mlink->next->fsec) ||
2010: strcmp(mlink->arch, mlink->next->arch)) {
2011: putchar('(');
2012: if ('\0' == *mlink->dsec)
2013: fputs(mlink->fsec, stdout);
2014: else
2015: fputs(mlink->dsec, stdout);
2016: if ('\0' != *mlink->arch)
2017: printf("/%s", mlink->arch);
2018: putchar(')');
2019: }
2020: mlink = mlink->next;
2021: if (NULL != mlink)
2022: fputs(", ", stdout);
2023: }
1.89 schwarze 2024: printf(" - %s\n", mpage->desc);
1.11 schwarze 2025: return;
1.85 schwarze 2026: }
2027:
2028: if (debug)
2029: say(mlink->file, "Adding to database");
1.47 schwarze 2030:
1.105 ! schwarze 2031: i = strlen(mpage->desc) + 1;
! 2032: key = mandoc_calloc(1, sizeof(struct str) + i);
! 2033: memcpy(key->key, mpage->desc, i);
! 2034: render_key(mc, key);
! 2035:
1.47 schwarze 2036: i = 1;
1.105 ! schwarze 2037: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.47 schwarze 2038: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
2039: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2040: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2041: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.105 ! schwarze 2042:
! 2043: if (key->rendered != key->key)
! 2044: free(key->rendered);
! 2045: free(key);
1.47 schwarze 2046:
1.85 schwarze 2047: while (NULL != mlink) {
1.75 schwarze 2048: dbadd_mlink(mlink);
1.85 schwarze 2049: mlink = mlink->next;
2050: }
1.91 schwarze 2051: mlink = mpage->mlinks;
1.47 schwarze 2052:
1.90 schwarze 2053: for (key = ohash_first(&names, &slot); NULL != key;
2054: key = ohash_next(&names, &slot)) {
2055: assert(key->mpage == mpage);
2056: if (NULL == key->rendered)
2057: render_key(mc, key);
2058: i = 1;
2059: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2060: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2061: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2062: SQL_STEP(stmts[STMT_INSERT_NAME]);
2063: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2064: if (key->rendered != key->key)
2065: free(key->rendered);
2066: free(key);
2067: }
1.47 schwarze 2068: for (key = ohash_first(&strings, &slot); NULL != key;
2069: key = ohash_next(&strings, &slot)) {
2070: assert(key->mpage == mpage);
1.53 schwarze 2071: if (NULL == key->rendered)
2072: render_key(mc, key);
1.47 schwarze 2073: i = 1;
2074: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2075: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2076: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2077: SQL_STEP(stmts[STMT_INSERT_KEY]);
2078: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2079: if (key->rendered != key->key)
2080: free(key->rendered);
1.47 schwarze 2081: free(key);
1.33 schwarze 2082: }
1.47 schwarze 2083: }
1.41 deraadt 2084:
1.47 schwarze 2085: static void
2086: dbprune(void)
2087: {
2088: struct mpage *mpage;
2089: struct mlink *mlink;
2090: size_t i;
2091: unsigned int slot;
1.11 schwarze 2092:
1.63 schwarze 2093: if (0 == nodb)
2094: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2095:
1.63 schwarze 2096: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2097: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2098: mlink = mpage->mlinks;
1.82 schwarze 2099: if (debug)
1.63 schwarze 2100: say(mlink->file, "Deleting from database");
2101: if (nodb)
2102: continue;
2103: for ( ; NULL != mlink; mlink = mlink->next) {
2104: i = 1;
2105: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2106: i, mlink->dsec);
2107: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2108: i, mlink->arch);
2109: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2110: i, mlink->name);
2111: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2112: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2113: }
1.11 schwarze 2114: }
1.63 schwarze 2115:
2116: if (0 == nodb)
2117: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2118: }
1.22 schwarze 2119:
1.47 schwarze 2120: /*
2121: * Close an existing database and its prepared statements.
2122: * If "real" is not set, rename the temporary file into the real one.
2123: */
2124: static void
2125: dbclose(int real)
2126: {
2127: size_t i;
1.72 schwarze 2128: int status;
2129: pid_t child;
1.11 schwarze 2130:
1.47 schwarze 2131: if (nodb)
2132: return;
1.11 schwarze 2133:
1.47 schwarze 2134: for (i = 0; i < STMT__MAX; i++) {
2135: sqlite3_finalize(stmts[i]);
2136: stmts[i] = NULL;
1.28 schwarze 2137: }
1.22 schwarze 2138:
1.47 schwarze 2139: sqlite3_close(db);
2140: db = NULL;
1.11 schwarze 2141:
1.47 schwarze 2142: if (real)
2143: return;
1.22 schwarze 2144:
1.72 schwarze 2145: if ('\0' == *tempfilename) {
2146: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2147: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2148: say(MANDOC_DB, "&rename");
1.72 schwarze 2149: }
2150: return;
2151: }
2152:
2153: switch (child = fork()) {
1.98 schwarze 2154: case -1:
1.72 schwarze 2155: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2156: say("", "&fork cmp");
1.72 schwarze 2157: return;
1.98 schwarze 2158: case 0:
1.72 schwarze 2159: execlp("cmp", "cmp", "-s",
2160: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2161: say("", "&exec cmp");
1.72 schwarze 2162: exit(0);
2163: default:
2164: break;
2165: }
2166: if (-1 == waitpid(child, &status, 0)) {
2167: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2168: say("", "&wait cmp");
1.72 schwarze 2169: } else if (WIFSIGNALED(status)) {
2170: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2171: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2172: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2173: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2174: say(MANDOC_DB,
2175: "Data changed, but cannot replace database");
2176: }
2177:
2178: *strrchr(tempfilename, '/') = '\0';
2179: switch (child = fork()) {
1.98 schwarze 2180: case -1:
1.72 schwarze 2181: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2182: say("", "&fork rm");
1.72 schwarze 2183: return;
1.98 schwarze 2184: case 0:
1.72 schwarze 2185: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2186: say("", "&exec rm");
1.72 schwarze 2187: exit((int)MANDOCLEVEL_SYSERR);
2188: default:
2189: break;
2190: }
2191: if (-1 == waitpid(child, &status, 0)) {
2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2193: say("", "&wait rm");
1.72 schwarze 2194: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2195: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2196: say("", "%s: Cannot remove temporary directory",
2197: tempfilename);
1.22 schwarze 2198: }
1.11 schwarze 2199: }
2200:
1.47 schwarze 2201: /*
2202: * This is straightforward stuff.
2203: * Open a database connection to a "temporary" database, then open a set
2204: * of prepared statements we'll use over and over again.
2205: * If "real" is set, we use the existing database; if not, we truncate a
2206: * temporary one.
2207: * Must be matched by dbclose().
2208: */
2209: static int
2210: dbopen(int real)
1.2 schwarze 2211: {
1.72 schwarze 2212: const char *sql;
1.47 schwarze 2213: int rc, ofl;
1.6 schwarze 2214:
1.98 schwarze 2215: if (nodb)
1.47 schwarze 2216: return(1);
1.6 schwarze 2217:
1.72 schwarze 2218: *tempfilename = '\0';
1.47 schwarze 2219: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2220:
2221: if (real) {
2222: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2223: if (SQLITE_OK != rc) {
1.47 schwarze 2224: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2225: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2226: return(0);
1.28 schwarze 2227: }
1.72 schwarze 2228: goto prepare_statements;
2229: }
2230:
2231: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2232:
1.72 schwarze 2233: remove(MANDOC_DB "~");
2234: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2235: if (SQLITE_OK == rc)
1.72 schwarze 2236: goto create_tables;
1.73 schwarze 2237: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2238: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2239: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2240: return(0);
2241: }
1.6 schwarze 2242:
1.100 schwarze 2243: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2244: sizeof(tempfilename));
1.72 schwarze 2245: if (NULL == mkdtemp(tempfilename)) {
2246: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2247: say("", "&%s", tempfilename);
1.72 schwarze 2248: return(0);
2249: }
1.100 schwarze 2250: (void)strlcat(tempfilename, "/" MANDOC_DB,
2251: sizeof(tempfilename));
1.72 schwarze 2252: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2253: if (SQLITE_OK != rc) {
1.47 schwarze 2254: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2255: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2256: return(0);
1.2 schwarze 2257: }
2258:
1.72 schwarze 2259: create_tables:
1.47 schwarze 2260: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2261: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2262: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2263: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2264: ");\n"
2265: "\n"
2266: "CREATE TABLE \"mlinks\" (\n"
2267: " \"sec\" TEXT NOT NULL,\n"
2268: " \"arch\" TEXT NOT NULL,\n"
2269: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2270: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2271: "ON DELETE CASCADE\n"
1.47 schwarze 2272: ");\n"
1.93 schwarze 2273: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2274: "\n"
1.90 schwarze 2275: "CREATE TABLE \"names\" (\n"
2276: " \"bits\" INTEGER NOT NULL,\n"
2277: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2278: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2279: "ON DELETE CASCADE\n"
2280: ");\n"
2281: "\n"
1.47 schwarze 2282: "CREATE TABLE \"keys\" (\n"
2283: " \"bits\" INTEGER NOT NULL,\n"
2284: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2285: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2286: "ON DELETE CASCADE\n"
1.93 schwarze 2287: ");\n"
2288: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2289:
2290: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2291: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2292: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2293: sqlite3_close(db);
1.47 schwarze 2294: return(0);
1.2 schwarze 2295: }
2296:
1.47 schwarze 2297: prepare_statements:
1.103 schwarze 2298: if (SQLITE_OK != sqlite3_exec(db,
2299: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2300: exitcode = (int)MANDOCLEVEL_SYSERR;
2301: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2302: sqlite3_errmsg(db));
2303: sqlite3_close(db);
2304: return(0);
2305: }
2306:
1.94 schwarze 2307: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2308: "(SELECT pageid FROM mlinks WHERE "
2309: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2310: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2311: sql = "INSERT INTO mpages "
1.89 schwarze 2312: "(desc,form) VALUES (?,?)";
1.47 schwarze 2313: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2314: sql = "INSERT INTO mlinks "
1.61 schwarze 2315: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2316: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2317: sql = "INSERT INTO names "
2318: "(bits,name,pageid) VALUES (?,?,?)";
2319: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2320: sql = "INSERT INTO keys "
2321: "(bits,key,pageid) VALUES (?,?,?)";
2322: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2323:
1.47 schwarze 2324: /*
2325: * When opening a new database, we can turn off
2326: * synchronous mode for much better performance.
2327: */
1.6 schwarze 2328:
1.103 schwarze 2329: if (real && SQLITE_OK != sqlite3_exec(db,
2330: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2331: exitcode = (int)MANDOCLEVEL_SYSERR;
2332: say(MANDOC_DB, "PRAGMA synchronous: %s",
2333: sqlite3_errmsg(db));
2334: sqlite3_close(db);
2335: return(0);
2336: }
1.11 schwarze 2337:
1.47 schwarze 2338: return(1);
2339: }
1.6 schwarze 2340:
1.47 schwarze 2341: static void *
2342: hash_halloc(size_t sz, void *arg)
2343: {
1.6 schwarze 2344:
1.101 schwarze 2345: return(mandoc_calloc(1, sz));
1.47 schwarze 2346: }
1.2 schwarze 2347:
1.47 schwarze 2348: static void *
2349: hash_alloc(size_t sz, void *arg)
2350: {
1.28 schwarze 2351:
1.47 schwarze 2352: return(mandoc_malloc(sz));
2353: }
1.6 schwarze 2354:
1.47 schwarze 2355: static void
2356: hash_free(void *p, size_t sz, void *arg)
2357: {
1.26 schwarze 2358:
1.47 schwarze 2359: free(p);
2360: }
1.6 schwarze 2361:
1.47 schwarze 2362: static int
2363: set_basedir(const char *targetdir)
2364: {
2365: static char startdir[PATH_MAX];
2366: static int fd;
1.6 schwarze 2367:
1.47 schwarze 2368: /*
2369: * Remember where we started by keeping a fd open to the origin
2370: * path component: throughout this utility, we chdir() a lot to
2371: * handle relative paths, and by doing this, we can return to
2372: * the starting point.
2373: */
2374: if ('\0' == *startdir) {
2375: if (NULL == getcwd(startdir, PATH_MAX)) {
2376: exitcode = (int)MANDOCLEVEL_SYSERR;
2377: if (NULL != targetdir)
1.80 schwarze 2378: say("", "&getcwd");
1.47 schwarze 2379: return(0);
2380: }
2381: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2382: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2383: say("", "&open %s", startdir);
1.47 schwarze 2384: return(0);
1.11 schwarze 2385: }
1.47 schwarze 2386: if (NULL == targetdir)
2387: targetdir = startdir;
2388: } else {
2389: if (-1 == fd)
2390: return(0);
2391: if (-1 == fchdir(fd)) {
2392: close(fd);
2393: basedir[0] = '\0';
2394: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2395: say("", "&chdir %s", startdir);
1.47 schwarze 2396: return(0);
1.2 schwarze 2397: }
1.47 schwarze 2398: if (NULL == targetdir) {
2399: close(fd);
2400: return(1);
1.2 schwarze 2401: }
2402: }
1.47 schwarze 2403: if (NULL == realpath(targetdir, basedir)) {
2404: basedir[0] = '\0';
2405: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2406: say("", "&%s: realpath", targetdir);
1.47 schwarze 2407: return(0);
2408: } else if (-1 == chdir(basedir)) {
2409: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2410: say("", "&chdir");
1.47 schwarze 2411: return(0);
2412: }
2413: return(1);
1.2 schwarze 2414: }
2415:
2416: static void
1.47 schwarze 2417: say(const char *file, const char *format, ...)
1.2 schwarze 2418: {
1.47 schwarze 2419: va_list ap;
1.80 schwarze 2420: int use_errno;
1.2 schwarze 2421:
1.47 schwarze 2422: if ('\0' != *basedir)
2423: fprintf(stderr, "%s", basedir);
2424: if ('\0' != *basedir && '\0' != *file)
2425: fputs("//", stderr);
2426: if ('\0' != *file)
2427: fprintf(stderr, "%s", file);
1.31 schwarze 2428:
1.80 schwarze 2429: use_errno = 1;
2430: if (NULL != format) {
2431: switch (*format) {
1.98 schwarze 2432: case '&':
1.80 schwarze 2433: format++;
2434: break;
1.98 schwarze 2435: case '\0':
1.80 schwarze 2436: format = NULL;
2437: break;
2438: default:
2439: use_errno = 0;
2440: break;
2441: }
2442: }
2443: if (NULL != format) {
2444: if ('\0' != *basedir || '\0' != *file)
2445: fputs(": ", stderr);
2446: va_start(ap, format);
2447: vfprintf(stderr, format, ap);
2448: va_end(ap);
2449: }
2450: if (use_errno) {
2451: if ('\0' != *basedir || '\0' != *file || NULL != format)
2452: fputs(": ", stderr);
1.47 schwarze 2453: perror(NULL);
1.80 schwarze 2454: } else
2455: fputc('\n', stderr);
1.1 schwarze 2456: }