Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.96
1.96 ! schwarze 1: /* $Id: mandocdb.c,v 1.95 2014/04/18 21:54:48 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 */
1.47 schwarze 175: static int use_all; /* use all found files */
1.82 schwarze 176: static int debug; /* print what we're doing */
1.47 schwarze 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 */
180: static enum op op; /* operational mode */
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:
315: int
1.3 schwarze 316: mandocdb(int argc, char *argv[])
1.1 schwarze 317: {
1.47 schwarze 318: int ch, i;
319: size_t j, sz;
320: const char *path_arg;
321: struct mchars *mc;
322: struct manpaths dirs;
323: struct mparse *mp;
324: struct ohash_info mpages_info, mlinks_info;
325:
326: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
327: memset(&dirs, 0, sizeof(struct manpaths));
328:
329: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
330: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
331: mpages_info.hfree = mlinks_info.hfree = hash_free;
332:
333: mpages_info.key_offset = offsetof(struct mpage, inodev);
334: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 335:
336: progname = strrchr(argv[0], '/');
337: if (progname == NULL)
338: progname = argv[0];
339: else
340: ++progname;
341:
1.47 schwarze 342: /*
343: * We accept a few different invocations.
344: * The CHECKOP macro makes sure that invocation styles don't
345: * clobber each other.
346: */
347: #define CHECKOP(_op, _ch) do \
348: if (OP_DEFAULT != (_op)) { \
349: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
350: goto usage; \
351: } while (/*CONSTCOND*/0)
1.10 schwarze 352:
1.47 schwarze 353: path_arg = NULL;
1.28 schwarze 354: op = OP_DEFAULT;
1.1 schwarze 355:
1.83 schwarze 356: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 357: switch (ch) {
1.6 schwarze 358: case ('a'):
359: use_all = 1;
360: break;
1.25 schwarze 361: case ('C'):
1.47 schwarze 362: CHECKOP(op, ch);
363: path_arg = optarg;
1.28 schwarze 364: op = OP_CONFFILE;
1.25 schwarze 365: break;
1.82 schwarze 366: case ('D'):
367: debug++;
368: break;
1.1 schwarze 369: case ('d'):
1.47 schwarze 370: CHECKOP(op, ch);
371: path_arg = optarg;
1.2 schwarze 372: op = OP_UPDATE;
1.1 schwarze 373: break;
1.47 schwarze 374: case ('n'):
375: nodb = 1;
376: break;
1.83 schwarze 377: case ('p'):
378: warnings = 1;
379: break;
1.59 schwarze 380: case ('Q'):
1.73 schwarze 381: mparse_options |= MPARSE_QUICK;
1.59 schwarze 382: break;
1.52 schwarze 383: case ('T'):
384: if (strcmp(optarg, "utf8")) {
385: fprintf(stderr, "-T%s: Unsupported "
386: "output format\n", optarg);
387: goto usage;
388: }
389: write_utf8 = 1;
390: break;
1.28 schwarze 391: case ('t'):
1.47 schwarze 392: CHECKOP(op, ch);
1.28 schwarze 393: dup2(STDOUT_FILENO, STDERR_FILENO);
394: op = OP_TEST;
1.47 schwarze 395: nodb = warnings = 1;
1.28 schwarze 396: break;
1.2 schwarze 397: case ('u'):
1.47 schwarze 398: CHECKOP(op, ch);
399: path_arg = optarg;
1.1 schwarze 400: op = OP_DELETE;
401: break;
1.83 schwarze 402: case ('v'):
403: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 404: break;
1.1 schwarze 405: default:
1.28 schwarze 406: goto usage;
1.1 schwarze 407: }
408:
409: argc -= optind;
410: argv += optind;
411:
1.28 schwarze 412: if (OP_CONFFILE == op && argc > 0) {
1.47 schwarze 413: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 414: goto usage;
415: }
416:
1.47 schwarze 417: exitcode = (int)MANDOCLEVEL_OK;
1.73 schwarze 418: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.47 schwarze 419: mc = mchars_alloc();
1.2 schwarze 420:
1.47 schwarze 421: ohash_init(&mpages, 6, &mpages_info);
422: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 423:
1.47 schwarze 424: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
425: /*
426: * Force processing all files.
427: */
428: use_all = 1;
1.2 schwarze 429:
1.47 schwarze 430: /*
431: * All of these deal with a specific directory.
432: * Jump into that directory then collect files specified
433: * on the command-line.
434: */
435: if (0 == set_basedir(path_arg))
436: goto out;
1.95 schwarze 437: if (dbopen(1)) {
438: for (i = 0; i < argc; i++)
439: filescan(argv[i]);
440: if (OP_TEST != op)
441: dbprune();
442: } else {
443: /*
444: * Database missing or corrupt.
445: * Recreate from scratch.
446: */
447: op = OP_DEFAULT;
448: if (0 == treescan())
449: goto out;
450: if (0 == dbopen(0))
451: goto out;
452: }
1.47 schwarze 453: if (OP_DELETE != op)
1.58 schwarze 454: mpages_merge(mc, mp);
1.95 schwarze 455: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 456: } else {
457: /*
458: * If we have arguments, use them as our manpaths.
459: * If we don't, grok from manpath(1) or however else
460: * manpath_parse() wants to do it.
461: */
462: if (argc > 0) {
463: dirs.paths = mandoc_calloc
464: (argc, sizeof(char *));
465: dirs.sz = (size_t)argc;
466: for (i = 0; i < argc; i++)
467: dirs.paths[i] = mandoc_strdup(argv[i]);
468: } else
469: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 470:
471: if (0 == dirs.sz) {
472: exitcode = (int)MANDOCLEVEL_BADARG;
473: say("", "Empty manpath");
474: }
1.2 schwarze 475:
1.47 schwarze 476: /*
477: * First scan the tree rooted at a base directory, then
478: * build a new database and finally move it into place.
479: * Ignore zero-length directories and strip trailing
480: * slashes.
481: */
482: for (j = 0; j < dirs.sz; j++) {
483: sz = strlen(dirs.paths[j]);
484: if (sz && '/' == dirs.paths[j][sz - 1])
485: dirs.paths[j][--sz] = '\0';
486: if (0 == sz)
487: continue;
1.2 schwarze 488:
1.47 schwarze 489: if (j) {
490: ohash_init(&mpages, 6, &mpages_info);
491: ohash_init(&mlinks, 6, &mlinks_info);
492: }
1.1 schwarze 493:
1.47 schwarze 494: if (0 == set_basedir(dirs.paths[j]))
495: goto out;
496: if (0 == treescan())
497: goto out;
498: if (0 == set_basedir(dirs.paths[j]))
499: goto out;
500: if (0 == dbopen(0))
501: goto out;
1.41 deraadt 502:
1.58 schwarze 503: mpages_merge(mc, mp);
1.87 schwarze 504: if (warnings &&
505: ! (MPARSE_QUICK & mparse_options))
506: names_check();
1.47 schwarze 507: dbclose(0);
1.28 schwarze 508:
1.47 schwarze 509: if (j + 1 < dirs.sz) {
510: mpages_free();
511: ohash_delete(&mpages);
512: ohash_delete(&mlinks);
513: }
1.2 schwarze 514: }
1.47 schwarze 515: }
516: out:
517: set_basedir(NULL);
518: manpath_free(&dirs);
519: mchars_free(mc);
520: mparse_free(mp);
521: mpages_free();
522: ohash_delete(&mpages);
523: ohash_delete(&mlinks);
524: return(exitcode);
525: usage:
1.83 schwarze 526: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
527: " %s [-aDnpQ] [-Tutf8] dir ...\n"
528: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
529: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 530: " %s [-Q] -t file ...\n",
1.47 schwarze 531: progname, progname, progname,
532: progname, progname);
1.1 schwarze 533:
1.47 schwarze 534: return((int)MANDOCLEVEL_BADARG);
535: }
1.1 schwarze 536:
1.47 schwarze 537: /*
538: * Scan a directory tree rooted at "basedir" for manpages.
539: * We use fts(), scanning directory parts along the way for clues to our
540: * section and architecture.
541: *
542: * If use_all has been specified, grok all files.
543: * If not, sanitise paths to the following:
544: *
545: * [./]man*[/<arch>]/<name>.<section>
546: * or
547: * [./]cat<section>[/<arch>]/<name>.0
548: *
549: * TODO: accomodate for multi-language directories.
550: */
551: static int
552: treescan(void)
553: {
1.96 ! schwarze 554: char buf[PATH_MAX];
1.47 schwarze 555: FTS *f;
556: FTSENT *ff;
557: struct mlink *mlink;
1.81 schwarze 558: int dform, gzip;
1.51 schwarze 559: char *dsec, *arch, *fsec, *cp;
560: const char *path;
1.47 schwarze 561: const char *argv[2];
1.1 schwarze 562:
1.47 schwarze 563: argv[0] = ".";
564: argv[1] = (char *)NULL;
1.1 schwarze 565:
1.96 ! schwarze 566: f = fts_open((char * const *)argv,
! 567: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 568: if (NULL == f) {
569: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 570: say("", "&fts_open");
1.47 schwarze 571: return(0);
572: }
1.2 schwarze 573:
1.47 schwarze 574: dsec = arch = NULL;
575: dform = FORM_NONE;
1.2 schwarze 576:
1.47 schwarze 577: while (NULL != (ff = fts_read(f))) {
578: path = ff->fts_path + 2;
1.96 ! schwarze 579: switch (ff->fts_info) {
! 580:
! 581: /*
! 582: * Symbolic links require various sanity checks,
! 583: * then get handled just like regular files.
! 584: */
! 585: case (FTS_SL):
! 586: if (NULL == realpath(path, buf)) {
! 587: if (warnings)
! 588: say(path, "&realpath");
! 589: continue;
! 590: }
! 591: if (strstr(buf, basedir) != buf) {
! 592: if (warnings) say("",
! 593: "%s: outside base directory", buf);
! 594: continue;
! 595: }
! 596: /* Use logical inode to avoid mpages dupe. */
! 597: if (-1 == stat(path, ff->fts_statp)) {
! 598: if (warnings)
! 599: say(path, "&stat");
! 600: continue;
! 601: }
! 602: /* FALLTHROUGH */
! 603:
1.15 schwarze 604: /*
1.47 schwarze 605: * If we're a regular file, add an mlink by using the
606: * stored directory data and handling the filename.
1.15 schwarze 607: */
1.96 ! schwarze 608: case (FTS_F):
1.47 schwarze 609: if (0 == strcmp(path, MANDOC_DB))
610: continue;
611: if ( ! use_all && ff->fts_level < 2) {
612: if (warnings)
613: say(path, "Extraneous file");
614: continue;
1.81 schwarze 615: }
616: gzip = 0;
617: fsec = NULL;
618: while (NULL == fsec) {
619: fsec = strrchr(ff->fts_name, '.');
620: if (NULL == fsec || strcmp(fsec+1, "gz"))
621: break;
622: gzip = 1;
623: *fsec = '\0';
624: fsec = NULL;
625: }
626: if (NULL == fsec) {
1.47 schwarze 627: if ( ! use_all) {
628: if (warnings)
629: say(path,
630: "No filename suffix");
631: continue;
632: }
633: } else if (0 == strcmp(++fsec, "html")) {
634: if (warnings)
635: say(path, "Skip html");
636: continue;
637: } else if (0 == strcmp(fsec, "ps")) {
638: if (warnings)
639: say(path, "Skip ps");
640: continue;
641: } else if (0 == strcmp(fsec, "pdf")) {
642: if (warnings)
643: say(path, "Skip pdf");
644: continue;
645: } else if ( ! use_all &&
646: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
647: (FORM_CAT == dform && strcmp(fsec, "0")))) {
648: if (warnings)
649: say(path, "Wrong filename suffix");
650: continue;
651: } else
652: fsec[-1] = '\0';
1.51 schwarze 653:
1.47 schwarze 654: mlink = mandoc_calloc(1, sizeof(struct mlink));
655: strlcpy(mlink->file, path, sizeof(mlink->file));
656: mlink->dform = dform;
1.51 schwarze 657: mlink->dsec = dsec;
658: mlink->arch = arch;
659: mlink->name = ff->fts_name;
660: mlink->fsec = fsec;
1.81 schwarze 661: mlink->gzip = gzip;
1.47 schwarze 662: mlink_add(mlink, ff->fts_statp);
663: continue;
1.96 ! schwarze 664:
! 665: case (FTS_D):
! 666: /* FALLTHROUGH */
! 667: case (FTS_DP):
! 668: break;
! 669:
! 670: default:
1.47 schwarze 671: if (warnings)
672: say(path, "Not a regular file");
673: continue;
674: }
675:
676: switch (ff->fts_level) {
677: case (0):
678: /* Ignore the root directory. */
679: break;
680: case (1):
681: /*
682: * This might contain manX/ or catX/.
683: * Try to infer this from the name.
684: * If we're not in use_all, enforce it.
685: */
686: cp = ff->fts_name;
687: if (FTS_DP == ff->fts_info)
688: break;
1.15 schwarze 689:
1.47 schwarze 690: if (0 == strncmp(cp, "man", 3)) {
691: dform = FORM_SRC;
692: dsec = cp + 3;
693: } else if (0 == strncmp(cp, "cat", 3)) {
694: dform = FORM_CAT;
695: dsec = cp + 3;
1.51 schwarze 696: } else {
697: dform = FORM_NONE;
698: dsec = NULL;
1.26 schwarze 699: }
1.47 schwarze 700:
701: if (NULL != dsec || use_all)
702: break;
703:
704: if (warnings)
705: say(path, "Unknown directory part");
706: fts_set(f, ff, FTS_SKIP);
707: break;
708: case (2):
709: /*
710: * Possibly our architecture.
711: * If we're descending, keep tabs on it.
712: */
713: if (FTS_DP != ff->fts_info && NULL != dsec)
714: arch = ff->fts_name;
1.51 schwarze 715: else
716: arch = NULL;
1.47 schwarze 717: break;
718: default:
719: if (FTS_DP == ff->fts_info || use_all)
720: break;
721: if (warnings)
722: say(path, "Extraneous directory part");
723: fts_set(f, ff, FTS_SKIP);
724: break;
1.14 schwarze 725: }
1.47 schwarze 726: }
727:
728: fts_close(f);
729: return(1);
730: }
1.1 schwarze 731:
1.47 schwarze 732: /*
733: * Add a file to the mlinks table.
734: * Do not verify that it's a "valid" looking manpage (we'll do that
735: * later).
736: *
737: * Try to infer the manual section, architecture, and page name from the
738: * path, assuming it looks like
739: *
740: * [./]man*[/<arch>]/<name>.<section>
741: * or
742: * [./]cat<section>[/<arch>]/<name>.0
743: *
744: * See treescan() for the fts(3) version of this.
745: */
746: static void
747: filescan(const char *file)
748: {
749: char buf[PATH_MAX];
750: struct stat st;
751: struct mlink *mlink;
752: char *p, *start;
753:
754: assert(use_all);
755:
756: if (0 == strncmp(file, "./", 2))
757: file += 2;
758:
1.96 ! schwarze 759: /*
! 760: * We have to do lstat(2) before realpath(3) loses
! 761: * the information whether this is a symbolic link.
! 762: * We need to know that because for symbolic links,
! 763: * we want to use the orginal file name, while for
! 764: * regular files, we want to use the real path.
! 765: */
! 766: if (-1 == lstat(file, &st)) {
! 767: exitcode = (int)MANDOCLEVEL_BADARG;
! 768: say(file, "&lstat");
! 769: return;
! 770: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
! 771: exitcode = (int)MANDOCLEVEL_BADARG;
! 772: say(file, "Not a regular file");
! 773: return;
! 774: }
! 775:
! 776: /*
! 777: * We have to resolve the file name to the real path
! 778: * in any case for the base directory check.
! 779: */
1.47 schwarze 780: if (NULL == realpath(file, buf)) {
781: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 782: say(file, "&realpath");
1.47 schwarze 783: return;
1.63 schwarze 784: }
785:
786: if (strstr(buf, basedir) == buf)
787: start = buf + strlen(basedir) + 1;
788: else if (OP_TEST == op)
789: start = buf;
790: else {
1.47 schwarze 791: exitcode = (int)MANDOCLEVEL_BADARG;
792: say("", "%s: outside base directory", buf);
793: return;
1.63 schwarze 794: }
795:
1.96 ! schwarze 796: /*
! 797: * Now we are sure the file is inside our tree.
! 798: * If it is a symbolic link, ignore the real path
! 799: * and use the original name.
! 800: * This implies passing stuff like "cat1/../man1/foo.1"
! 801: * on the command line won't work. So don't do that.
! 802: * Note the stat(2) can still fail if the link target
! 803: * doesn't exist.
! 804: */
! 805: if (S_IFLNK & st.st_mode) {
! 806: if (-1 == stat(buf, &st)) {
! 807: exitcode = (int)MANDOCLEVEL_BADARG;
! 808: say(file, "&stat");
! 809: return;
! 810: }
! 811: strlcpy(buf, file, sizeof(buf));
! 812: start = strstr(buf, basedir) == buf ?
! 813: buf + strlen(basedir) + 1 : buf;
1.1 schwarze 814: }
1.63 schwarze 815:
1.47 schwarze 816: mlink = mandoc_calloc(1, sizeof(struct mlink));
817: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 818:
1.10 schwarze 819: /*
1.47 schwarze 820: * First try to guess our directory structure.
821: * If we find a separator, try to look for man* or cat*.
822: * If we find one of these and what's underneath is a directory,
823: * assume it's an architecture.
1.10 schwarze 824: */
1.47 schwarze 825: if (NULL != (p = strchr(start, '/'))) {
826: *p++ = '\0';
827: if (0 == strncmp(start, "man", 3)) {
828: mlink->dform = FORM_SRC;
1.51 schwarze 829: mlink->dsec = start + 3;
1.47 schwarze 830: } else if (0 == strncmp(start, "cat", 3)) {
831: mlink->dform = FORM_CAT;
1.51 schwarze 832: mlink->dsec = start + 3;
1.47 schwarze 833: }
1.10 schwarze 834:
1.47 schwarze 835: start = p;
836: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
837: *p++ = '\0';
1.51 schwarze 838: mlink->arch = start;
1.47 schwarze 839: start = p;
1.41 deraadt 840: }
1.47 schwarze 841: }
1.7 schwarze 842:
1.47 schwarze 843: /*
844: * Now check the file suffix.
845: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
846: */
847: p = strrchr(start, '\0');
848: while (p-- > start && '/' != *p && '.' != *p)
849: /* Loop. */ ;
850:
851: if ('.' == *p) {
852: *p++ = '\0';
1.51 schwarze 853: mlink->fsec = p;
1.47 schwarze 854: }
1.41 deraadt 855:
1.47 schwarze 856: /*
857: * Now try to parse the name.
858: * Use the filename portion of the path.
859: */
860: mlink->name = start;
861: if (NULL != (p = strrchr(start, '/'))) {
862: mlink->name = p + 1;
863: *p = '\0';
864: }
865: mlink_add(mlink, &st);
866: }
1.2 schwarze 867:
1.47 schwarze 868: static void
869: mlink_add(struct mlink *mlink, const struct stat *st)
870: {
871: struct inodev inodev;
872: struct mpage *mpage;
873: unsigned int slot;
874:
875: assert(NULL != mlink->file);
876:
1.51 schwarze 877: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
878: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
879: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
880: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 881:
882: if ('0' == *mlink->fsec) {
883: free(mlink->fsec);
884: mlink->fsec = mandoc_strdup(mlink->dsec);
885: mlink->fform = FORM_CAT;
886: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
887: mlink->fform = FORM_SRC;
888: else
889: mlink->fform = FORM_NONE;
1.1 schwarze 890:
1.47 schwarze 891: slot = ohash_qlookup(&mlinks, mlink->file);
892: assert(NULL == ohash_find(&mlinks, slot));
893: ohash_insert(&mlinks, slot, mlink);
894:
895: inodev.st_ino = st->st_ino;
896: inodev.st_dev = st->st_dev;
897: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
898: sizeof(struct inodev), inodev.st_ino);
899: mpage = ohash_find(&mpages, slot);
900: if (NULL == mpage) {
901: mpage = mandoc_calloc(1, sizeof(struct mpage));
902: mpage->inodev.st_ino = inodev.st_ino;
903: mpage->inodev.st_dev = inodev.st_dev;
904: ohash_insert(&mpages, slot, mpage);
905: } else
906: mlink->next = mpage->mlinks;
907: mpage->mlinks = mlink;
1.75 schwarze 908: mlink->mpage = mpage;
1.47 schwarze 909: }
1.1 schwarze 910:
1.47 schwarze 911: static void
912: mlink_free(struct mlink *mlink)
913: {
1.1 schwarze 914:
1.47 schwarze 915: free(mlink->dsec);
916: free(mlink->arch);
917: free(mlink->name);
918: free(mlink->fsec);
919: free(mlink);
920: }
1.1 schwarze 921:
1.47 schwarze 922: static void
923: mpages_free(void)
924: {
925: struct mpage *mpage;
926: struct mlink *mlink;
927: unsigned int slot;
928:
929: mpage = ohash_first(&mpages, &slot);
930: while (NULL != mpage) {
931: while (NULL != (mlink = mpage->mlinks)) {
932: mpage->mlinks = mlink->next;
933: mlink_free(mlink);
934: }
935: free(mpage->sec);
936: free(mpage->arch);
937: free(mpage->title);
938: free(mpage->desc);
939: free(mpage);
940: mpage = ohash_next(&mpages, &slot);
941: }
942: }
1.1 schwarze 943:
1.47 schwarze 944: /*
945: * For each mlink to the mpage, check whether the path looks like
946: * it is formatted, and if it does, check whether a source manual
947: * exists by the same name, ignoring the suffix.
948: * If both conditions hold, drop the mlink.
949: */
950: static void
951: mlinks_undupe(struct mpage *mpage)
952: {
953: char buf[PATH_MAX];
954: struct mlink **prev;
955: struct mlink *mlink;
956: char *bufp;
957:
958: mpage->form = FORM_CAT;
959: prev = &mpage->mlinks;
960: while (NULL != (mlink = *prev)) {
961: if (FORM_CAT != mlink->dform) {
962: mpage->form = FORM_NONE;
963: goto nextlink;
964: }
965: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
966: if (warnings)
967: say(mlink->file, "Filename too long");
968: goto nextlink;
1.26 schwarze 969: }
1.47 schwarze 970: bufp = strstr(buf, "cat");
971: assert(NULL != bufp);
972: memcpy(bufp, "man", 3);
973: if (NULL != (bufp = strrchr(buf, '.')))
974: *++bufp = '\0';
975: strlcat(buf, mlink->dsec, PATH_MAX);
976: if (NULL == ohash_find(&mlinks,
977: ohash_qlookup(&mlinks, buf)))
978: goto nextlink;
979: if (warnings)
980: say(mlink->file, "Man source exists: %s", buf);
981: if (use_all)
982: goto nextlink;
983: *prev = mlink->next;
984: mlink_free(mlink);
985: continue;
986: nextlink:
987: prev = &(*prev)->next;
1.1 schwarze 988: }
1.47 schwarze 989: }
1.1 schwarze 990:
1.88 schwarze 991: static void
1.50 schwarze 992: mlink_check(struct mpage *mpage, struct mlink *mlink)
993: {
1.88 schwarze 994: struct str *str;
995: unsigned int slot;
1.50 schwarze 996:
997: /*
998: * Check whether the manual section given in a file
999: * agrees with the directory where the file is located.
1000: * Some manuals have suffixes like (3p) on their
1001: * section number either inside the file or in the
1002: * directory name, some are linked into more than one
1003: * section, like encrypt(1) = makekey(8).
1004: */
1005:
1006: if (FORM_SRC == mpage->form &&
1.88 schwarze 1007: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1008: say(mlink->file, "Section \"%s\" manual in %s directory",
1009: mpage->sec, mlink->dsec);
1010:
1011: /*
1012: * Manual page directories exist for each kernel
1013: * architecture as returned by machine(1).
1014: * However, many manuals only depend on the
1015: * application architecture as returned by arch(1).
1016: * For example, some (2/ARM) manuals are shared
1017: * across the "armish" and "zaurus" kernel
1018: * architectures.
1019: * A few manuals are even shared across completely
1020: * different architectures, for example fdformat(1)
1021: * on amd64, i386, sparc, and sparc64.
1022: */
1023:
1.88 schwarze 1024: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1025: say(mlink->file, "Architecture \"%s\" manual in "
1026: "\"%s\" directory", mpage->arch, mlink->arch);
1027:
1.88 schwarze 1028: /*
1029: * XXX
1.90 schwarze 1030: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1031: */
1032:
1033: if (FORM_CAT == mpage->form)
1034: return;
1035:
1036: /*
1037: * Check whether this mlink
1038: * appears as a name in the NAME section.
1039: */
1.50 schwarze 1040:
1.90 schwarze 1041: slot = ohash_qlookup(&names, mlink->name);
1042: str = ohash_find(&names, slot);
1.88 schwarze 1043: assert(NULL != str);
1.90 schwarze 1044: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1045: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1046: }
1047:
1.47 schwarze 1048: /*
1049: * Run through the files in the global vector "mpages"
1050: * and add them to the database specified in "basedir".
1051: *
1052: * This handles the parsing scheme itself, using the cues of directory
1053: * and filename to determine whether the file is parsable or not.
1054: */
1055: static void
1.58 schwarze 1056: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1057: {
1.70 schwarze 1058: char any[] = "any";
1.58 schwarze 1059: struct ohash_info str_info;
1.81 schwarze 1060: int fd[2];
1.75 schwarze 1061: struct mpage *mpage, *mpage_dest;
1062: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1063: struct mdoc *mdoc;
1064: struct man *man;
1.75 schwarze 1065: char *sodest;
1.69 schwarze 1066: char *cp;
1.81 schwarze 1067: pid_t child_pid;
1.88 schwarze 1068: int status;
1.58 schwarze 1069: unsigned int pslot;
1.47 schwarze 1070: enum mandoclevel lvl;
1071:
1072: str_info.alloc = hash_alloc;
1073: str_info.halloc = hash_halloc;
1074: str_info.hfree = hash_free;
1075: str_info.key_offset = offsetof(struct str, key);
1076:
1.64 schwarze 1077: if (0 == nodb)
1078: SQL_EXEC("BEGIN TRANSACTION");
1079:
1.47 schwarze 1080: mpage = ohash_first(&mpages, &pslot);
1081: while (NULL != mpage) {
1082: mlinks_undupe(mpage);
1083: if (NULL == mpage->mlinks) {
1084: mpage = ohash_next(&mpages, &pslot);
1085: continue;
1086: }
1.1 schwarze 1087:
1.90 schwarze 1088: name_mask = NAME_MASK;
1089: ohash_init(&names, 4, &str_info);
1.47 schwarze 1090: ohash_init(&strings, 6, &str_info);
1091: mparse_reset(mp);
1092: mdoc = NULL;
1093: man = NULL;
1.81 schwarze 1094: sodest = NULL;
1095: child_pid = 0;
1096: fd[0] = -1;
1097: fd[1] = -1;
1098:
1099: if (mpage->mlinks->gzip) {
1100: if (-1 == pipe(fd)) {
1101: exitcode = (int)MANDOCLEVEL_SYSERR;
1102: say(mpage->mlinks->file, "&pipe gunzip");
1103: goto nextpage;
1104: }
1105: switch (child_pid = fork()) {
1106: case (-1):
1107: exitcode = (int)MANDOCLEVEL_SYSERR;
1108: say(mpage->mlinks->file, "&fork gunzip");
1109: child_pid = 0;
1110: close(fd[1]);
1111: close(fd[0]);
1112: goto nextpage;
1113: case (0):
1114: close(fd[0]);
1115: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1116: say(mpage->mlinks->file,
1117: "&dup gunzip");
1118: exit(1);
1119: }
1120: execlp("gunzip", "gunzip", "-c",
1121: mpage->mlinks->file, NULL);
1122: say(mpage->mlinks->file, "&exec gunzip");
1123: exit(1);
1124: default:
1125: close(fd[1]);
1126: break;
1127: }
1128: }
1.11 schwarze 1129:
1130: /*
1.24 schwarze 1131: * Try interpreting the file as mdoc(7) or man(7)
1132: * source code, unless it is already known to be
1133: * formatted. Fall back to formatted mode.
1.11 schwarze 1134: */
1.47 schwarze 1135: if (FORM_CAT != mpage->mlinks->dform ||
1136: FORM_CAT != mpage->mlinks->fform) {
1.81 schwarze 1137: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.47 schwarze 1138: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1139: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1140: }
1.11 schwarze 1141:
1.75 schwarze 1142: if (NULL != sodest) {
1143: mlink_dest = ohash_find(&mlinks,
1144: ohash_qlookup(&mlinks, sodest));
1145: if (NULL != mlink_dest) {
1146:
1147: /* The .so target exists. */
1148:
1149: mpage_dest = mlink_dest->mpage;
1150: mlink = mpage->mlinks;
1151: while (1) {
1152: mlink->mpage = mpage_dest;
1153:
1154: /*
1155: * If the target was already
1156: * processed, add the links
1157: * to the database now.
1158: * Otherwise, this will
1159: * happen when we come
1160: * to the target.
1161: */
1162:
1.94 schwarze 1163: if (mpage_dest->pageid)
1.75 schwarze 1164: dbadd_mlink(mlink);
1165:
1166: if (NULL == mlink->next)
1167: break;
1168: mlink = mlink->next;
1169: }
1170:
1171: /* Move all links to the target. */
1172:
1173: mlink->next = mlink_dest->next;
1174: mlink_dest->next = mpage->mlinks;
1175: mpage->mlinks = NULL;
1176: }
1.81 schwarze 1177: goto nextpage;
1.75 schwarze 1178: } else if (NULL != mdoc) {
1.47 schwarze 1179: mpage->form = FORM_SRC;
1180: mpage->sec =
1181: mandoc_strdup(mdoc_meta(mdoc)->msec);
1182: mpage->arch = mdoc_meta(mdoc)->arch;
1183: mpage->arch = mandoc_strdup(
1184: NULL == mpage->arch ? "" : mpage->arch);
1185: mpage->title =
1186: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1187: } else if (NULL != man) {
1.47 schwarze 1188: mpage->form = FORM_SRC;
1189: mpage->sec =
1190: mandoc_strdup(man_meta(man)->msec);
1191: mpage->arch =
1192: mandoc_strdup(mpage->mlinks->arch);
1193: mpage->title =
1194: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1195: } else {
1.47 schwarze 1196: mpage->form = FORM_CAT;
1197: mpage->sec =
1198: mandoc_strdup(mpage->mlinks->dsec);
1199: mpage->arch =
1200: mandoc_strdup(mpage->mlinks->arch);
1201: mpage->title =
1202: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1203: }
1.54 schwarze 1204: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1205: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1206: any : mpage->arch, TYPE_arch);
1.1 schwarze 1207:
1.54 schwarze 1208: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1209: if ('\0' != *mlink->dsec)
1210: putkey(mpage, mlink->dsec, TYPE_sec);
1211: if ('\0' != *mlink->fsec)
1212: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1213: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1214: any : mlink->arch, TYPE_arch);
1.90 schwarze 1215: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1216: }
1.41 deraadt 1217:
1.92 schwarze 1218: assert(NULL == mpage->desc);
1.47 schwarze 1219: if (NULL != mdoc) {
1220: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.90 schwarze 1221: putkey(mpage, cp, NAME_HEAD);
1.47 schwarze 1222: parse_mdoc(mpage, mdoc_node(mdoc));
1223: } else if (NULL != man)
1224: parse_man(mpage, man_node(man));
1225: else
1.81 schwarze 1226: parse_cat(mpage, fd[0]);
1.92 schwarze 1227: if (NULL == mpage->desc)
1228: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1229:
1230: if (warnings && !use_all)
1231: for (mlink = mpage->mlinks; mlink;
1232: mlink = mlink->next)
1233: mlink_check(mpage, mlink);
1.6 schwarze 1234:
1.62 schwarze 1235: dbadd(mpage, mc);
1.81 schwarze 1236:
1237: nextpage:
1238: if (child_pid) {
1239: if (-1 == waitpid(child_pid, &status, 0)) {
1240: exitcode = (int)MANDOCLEVEL_SYSERR;
1241: say(mpage->mlinks->file, "&wait gunzip");
1242: } else if (WIFSIGNALED(status)) {
1243: exitcode = (int)MANDOCLEVEL_SYSERR;
1244: say(mpage->mlinks->file,
1245: "gunzip died from signal %d",
1246: WTERMSIG(status));
1247: } else if (WEXITSTATUS(status)) {
1248: exitcode = (int)MANDOCLEVEL_SYSERR;
1249: say(mpage->mlinks->file,
1250: "gunzip failed with code %d",
1251: WEXITSTATUS(status));
1252: }
1253: }
1.47 schwarze 1254: ohash_delete(&strings);
1.90 schwarze 1255: ohash_delete(&names);
1.47 schwarze 1256: mpage = ohash_next(&mpages, &pslot);
1257: }
1.64 schwarze 1258:
1259: if (0 == nodb)
1260: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1261: }
1262:
1263: static void
1264: names_check(void)
1265: {
1266: sqlite3_stmt *stmt;
1267: const char *name, *sec, *arch, *key;
1268: int irc;
1269:
1270: sqlite3_prepare_v2(db,
1271: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1272: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1273: "WHERE bits & ? AND NOT EXISTS ("
1274: "SELECT pageid FROM mlinks "
1.90 schwarze 1275: "WHERE mlinks.pageid == names.pageid "
1276: "AND mlinks.name == names.name"
1.87 schwarze 1277: ")"
1278: ") JOIN ("
1279: "SELECT * FROM mlinks GROUP BY pageid"
1280: ") USING (pageid);",
1281: -1, &stmt, NULL);
1282:
1.91 schwarze 1283: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1284: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1285:
1286: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1287: name = sqlite3_column_text(stmt, 0);
1288: sec = sqlite3_column_text(stmt, 1);
1289: arch = sqlite3_column_text(stmt, 2);
1290: key = sqlite3_column_text(stmt, 3);
1291: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1292: '\0' == *arch ? "" : "/",
1293: '\0' == *arch ? "" : arch, key);
1294: }
1295: sqlite3_finalize(stmt);
1.47 schwarze 1296: }
1.6 schwarze 1297:
1.47 schwarze 1298: static void
1.81 schwarze 1299: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1300: {
1301: FILE *stream;
1302: char *line, *p, *title;
1303: size_t len, plen, titlesz;
1.1 schwarze 1304:
1.81 schwarze 1305: stream = (-1 == fd) ?
1306: fopen(mpage->mlinks->file, "r") :
1307: fdopen(fd, "r");
1308: if (NULL == stream) {
1.47 schwarze 1309: if (warnings)
1.80 schwarze 1310: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1311: return;
1312: }
1.1 schwarze 1313:
1.47 schwarze 1314: /* Skip to first blank line. */
1.1 schwarze 1315:
1.47 schwarze 1316: while (NULL != (line = fgetln(stream, &len)))
1317: if ('\n' == *line)
1318: break;
1.1 schwarze 1319:
1.47 schwarze 1320: /*
1321: * Assume the first line that is not indented
1322: * is the first section header. Skip to it.
1323: */
1.1 schwarze 1324:
1.47 schwarze 1325: while (NULL != (line = fgetln(stream, &len)))
1326: if ('\n' != *line && ' ' != *line)
1327: break;
1328:
1329: /*
1330: * Read up until the next section into a buffer.
1331: * Strip the leading and trailing newline from each read line,
1332: * appending a trailing space.
1333: * Ignore empty (whitespace-only) lines.
1334: */
1.28 schwarze 1335:
1.47 schwarze 1336: titlesz = 0;
1337: title = NULL;
1.38 schwarze 1338:
1.47 schwarze 1339: while (NULL != (line = fgetln(stream, &len))) {
1340: if (' ' != *line || '\n' != line[len - 1])
1341: break;
1342: while (len > 0 && isspace((unsigned char)*line)) {
1343: line++;
1344: len--;
1345: }
1346: if (1 == len)
1347: continue;
1348: title = mandoc_realloc(title, titlesz + len);
1349: memcpy(title + titlesz, line, len);
1350: titlesz += len;
1351: title[titlesz - 1] = ' ';
1352: }
1.28 schwarze 1353:
1.47 schwarze 1354: /*
1355: * If no page content can be found, or the input line
1356: * is already the next section header, or there is no
1357: * trailing newline, reuse the page title as the page
1358: * description.
1359: */
1.1 schwarze 1360:
1.47 schwarze 1361: if (NULL == title || '\0' == *title) {
1362: if (warnings)
1363: say(mpage->mlinks->file,
1364: "Cannot find NAME section");
1365: fclose(stream);
1366: free(title);
1367: return;
1368: }
1.24 schwarze 1369:
1.47 schwarze 1370: title = mandoc_realloc(title, titlesz + 1);
1371: title[titlesz] = '\0';
1.24 schwarze 1372:
1.47 schwarze 1373: /*
1374: * Skip to the first dash.
1375: * Use the remaining line as the description (no more than 70
1376: * bytes).
1377: */
1.28 schwarze 1378:
1.47 schwarze 1379: if (NULL != (p = strstr(title, "- "))) {
1380: for (p += 2; ' ' == *p || '\b' == *p; p++)
1381: /* Skip to next word. */ ;
1382: } else {
1383: if (warnings)
1384: say(mpage->mlinks->file,
1385: "No dash in title line");
1386: p = title;
1387: }
1.1 schwarze 1388:
1.47 schwarze 1389: plen = strlen(p);
1.1 schwarze 1390:
1.47 schwarze 1391: /* Strip backspace-encoding from line. */
1.1 schwarze 1392:
1.47 schwarze 1393: while (NULL != (line = memchr(p, '\b', plen))) {
1394: len = line - p;
1395: if (0 == len) {
1396: memmove(line, line + 1, plen--);
1397: continue;
1398: }
1399: memmove(line - 1, line + 1, plen - len);
1400: plen -= 2;
1401: }
1.1 schwarze 1402:
1.47 schwarze 1403: mpage->desc = mandoc_strdup(p);
1404: fclose(stream);
1405: free(title);
1406: }
1.16 schwarze 1407:
1.47 schwarze 1408: /*
1409: * Put a type/word pair into the word database for this particular file.
1410: */
1411: static void
1.69 schwarze 1412: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1413: {
1.69 schwarze 1414: char *cp;
1.37 schwarze 1415:
1.47 schwarze 1416: assert(NULL != value);
1.69 schwarze 1417: if (TYPE_arch == type)
1418: for (cp = value; *cp; cp++)
1419: if (isupper((unsigned char)*cp))
1420: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1421: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1422: }
1423:
1424: /*
1.47 schwarze 1425: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1426: */
1427: static void
1.47 schwarze 1428: putmdockey(const struct mpage *mpage,
1429: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1430: {
1.16 schwarze 1431:
1.47 schwarze 1432: for ( ; NULL != n; n = n->next) {
1433: if (NULL != n->child)
1434: putmdockey(mpage, n->child, m);
1435: if (MDOC_TEXT == n->type)
1436: putkey(mpage, n->string, m);
1437: }
1438: }
1.16 schwarze 1439:
1.47 schwarze 1440: static void
1441: parse_man(struct mpage *mpage, const struct man_node *n)
1442: {
1443: const struct man_node *head, *body;
1.78 schwarze 1444: char *start, *title;
1.47 schwarze 1445: char byte;
1.78 schwarze 1446: size_t sz;
1.16 schwarze 1447:
1.47 schwarze 1448: if (NULL == n)
1449: return;
1.16 schwarze 1450:
1.47 schwarze 1451: /*
1452: * We're only searching for one thing: the first text child in
1453: * the BODY of a NAME section. Since we don't keep track of
1454: * sections in -man, run some hoops to find out whether we're in
1455: * the correct section or not.
1456: */
1.16 schwarze 1457:
1.47 schwarze 1458: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1459: body = n;
1460: assert(body->parent);
1461: if (NULL != (head = body->parent->head) &&
1462: 1 == head->nchild &&
1463: NULL != (head = (head->child)) &&
1464: MAN_TEXT == head->type &&
1465: 0 == strcmp(head->string, "NAME") &&
1.78 schwarze 1466: NULL != body->child) {
1.2 schwarze 1467:
1.47 schwarze 1468: /*
1469: * Suck the entire NAME section into memory.
1470: * Yes, we might run away.
1471: * But too many manuals have big, spread-out
1472: * NAME sections over many lines.
1473: */
1.2 schwarze 1474:
1.78 schwarze 1475: title = NULL;
1476: man_deroff(&title, body);
1.47 schwarze 1477: if (NULL == title)
1478: return;
1.16 schwarze 1479:
1.47 schwarze 1480: /*
1481: * Go through a special heuristic dance here.
1482: * Conventionally, one or more manual names are
1483: * comma-specified prior to a whitespace, then a
1484: * dash, then a description. Try to puzzle out
1485: * the name parts here.
1486: */
1.16 schwarze 1487:
1.78 schwarze 1488: start = title;
1.47 schwarze 1489: for ( ;; ) {
1490: sz = strcspn(start, " ,");
1491: if ('\0' == start[sz])
1492: break;
1.1 schwarze 1493:
1.47 schwarze 1494: byte = start[sz];
1495: start[sz] = '\0';
1.67 schwarze 1496:
1497: /*
1498: * Assume a stray trailing comma in the
1499: * name list if a name begins with a dash.
1500: */
1501:
1502: if ('-' == start[0] ||
1503: ('\\' == start[0] && '-' == start[1]))
1504: break;
1.1 schwarze 1505:
1.90 schwarze 1506: putkey(mpage, start, NAME_TITLE);
1.1 schwarze 1507:
1.47 schwarze 1508: if (' ' == byte) {
1509: start += sz + 1;
1510: break;
1511: }
1.1 schwarze 1512:
1.47 schwarze 1513: assert(',' == byte);
1514: start += sz + 1;
1515: while (' ' == *start)
1516: start++;
1517: }
1.1 schwarze 1518:
1.78 schwarze 1519: if (start == title) {
1.90 schwarze 1520: putkey(mpage, start, NAME_TITLE);
1.47 schwarze 1521: free(title);
1522: return;
1523: }
1.1 schwarze 1524:
1.47 schwarze 1525: while (isspace((unsigned char)*start))
1526: start++;
1.1 schwarze 1527:
1.47 schwarze 1528: if (0 == strncmp(start, "-", 1))
1529: start += 1;
1530: else if (0 == strncmp(start, "\\-\\-", 4))
1531: start += 4;
1532: else if (0 == strncmp(start, "\\-", 2))
1533: start += 2;
1534: else if (0 == strncmp(start, "\\(en", 4))
1535: start += 4;
1536: else if (0 == strncmp(start, "\\(em", 4))
1537: start += 4;
1.1 schwarze 1538:
1.47 schwarze 1539: while (' ' == *start)
1540: start++;
1.1 schwarze 1541:
1.47 schwarze 1542: mpage->desc = mandoc_strdup(start);
1543: free(title);
1544: return;
1545: }
1546: }
1.1 schwarze 1547:
1.47 schwarze 1548: for (n = n->child; n; n = n->next) {
1549: if (NULL != mpage->desc)
1550: break;
1551: parse_man(mpage, n);
1.1 schwarze 1552: }
1553: }
1554:
1555: static void
1.47 schwarze 1556: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1557: {
1558:
1.47 schwarze 1559: assert(NULL != n);
1560: for (n = n->child; NULL != n; n = n->next) {
1561: switch (n->type) {
1562: case (MDOC_ELEM):
1563: /* FALLTHROUGH */
1564: case (MDOC_BLOCK):
1565: /* FALLTHROUGH */
1566: case (MDOC_HEAD):
1567: /* FALLTHROUGH */
1568: case (MDOC_BODY):
1569: /* FALLTHROUGH */
1570: case (MDOC_TAIL):
1571: if (NULL != mdocs[n->tok].fp)
1572: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1573: break;
1574: if (mdocs[n->tok].mask)
1575: putmdockey(mpage, n->child,
1576: mdocs[n->tok].mask);
1577: break;
1578: default:
1579: assert(MDOC_ROOT != n->type);
1580: continue;
1581: }
1582: if (NULL != n->child)
1583: parse_mdoc(mpage, n);
1.1 schwarze 1584: }
1585: }
1586:
1.19 schwarze 1587: static int
1.47 schwarze 1588: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1589: {
1590: const char *start, *end;
1591: size_t sz;
1.19 schwarze 1592:
1.47 schwarze 1593: if (SEC_SYNOPSIS != n->sec ||
1594: NULL == (n = n->child) ||
1595: MDOC_TEXT != n->type)
1.19 schwarze 1596: return(0);
1.1 schwarze 1597:
1598: /*
1599: * Only consider those `Fd' macro fields that begin with an
1600: * "inclusion" token (versus, e.g., #define).
1601: */
1.47 schwarze 1602:
1.1 schwarze 1603: if (strcmp("#include", n->string))
1.19 schwarze 1604: return(0);
1.1 schwarze 1605:
1606: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1607: return(0);
1.1 schwarze 1608:
1609: /*
1610: * Strip away the enclosing angle brackets and make sure we're
1611: * not zero-length.
1612: */
1613:
1614: start = n->string;
1615: if ('<' == *start || '"' == *start)
1616: start++;
1617:
1618: if (0 == (sz = strlen(start)))
1.19 schwarze 1619: return(0);
1.1 schwarze 1620:
1621: end = &start[(int)sz - 1];
1622: if ('>' == *end || '"' == *end)
1623: end--;
1624:
1.47 schwarze 1625: if (end > start)
1626: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1627: return(0);
1.1 schwarze 1628: }
1629:
1.19 schwarze 1630: static int
1.47 schwarze 1631: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1632: {
1.69 schwarze 1633: char *cp;
1.1 schwarze 1634:
1.47 schwarze 1635: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1636: return(0);
1637:
1.47 schwarze 1638: /*
1639: * Parse: .Fn "struct type *name" "char *arg".
1640: * First strip away pointer symbol.
1641: * Then store the function name, then type.
1642: * Finally, store the arguments.
1643: */
1.1 schwarze 1644:
1.47 schwarze 1645: if (NULL == (cp = strrchr(n->string, ' ')))
1646: cp = n->string;
1.1 schwarze 1647:
1648: while ('*' == *cp)
1649: cp++;
1650:
1.47 schwarze 1651: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1652:
1.47 schwarze 1653: if (n->string < cp)
1654: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1655:
1.47 schwarze 1656: for (n = n->next; NULL != n; n = n->next)
1657: if (MDOC_TEXT == n->type)
1658: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1659:
1660: return(0);
1.1 schwarze 1661: }
1662:
1.19 schwarze 1663: static int
1.47 schwarze 1664: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1665: {
1.47 schwarze 1666: char *cp;
1.1 schwarze 1667:
1668: if (NULL == (n = n->child))
1.19 schwarze 1669: return(0);
1.1 schwarze 1670:
1.47 schwarze 1671: if (NULL == n->next) {
1672: putkey(mpage, n->string, TYPE_Xr);
1673: return(0);
1674: }
1.1 schwarze 1675:
1.77 schwarze 1676: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1677: putkey(mpage, cp, TYPE_Xr);
1678: free(cp);
1679: return(0);
1.1 schwarze 1680: }
1681:
1.19 schwarze 1682: static int
1.47 schwarze 1683: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1684: {
1685:
1.79 schwarze 1686: if (MDOC_BODY == n->type)
1687: mdoc_deroff(&mpage->desc, n);
1688: return(0);
1.1 schwarze 1689: }
1690:
1.19 schwarze 1691: static int
1.47 schwarze 1692: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1693: {
1694:
1.86 schwarze 1695: if (SEC_NAME == n->sec)
1.90 schwarze 1696: putmdockey(mpage, n->child, NAME_TITLE);
1.86 schwarze 1697: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.90 schwarze 1698: putmdockey(mpage, n->child, NAME_SYN);
1.86 schwarze 1699: return(0);
1.1 schwarze 1700: }
1701:
1.19 schwarze 1702: static int
1.47 schwarze 1703: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1704: {
1705:
1.19 schwarze 1706: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1707: }
1708:
1.47 schwarze 1709: static int
1710: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1711: {
1712:
1.47 schwarze 1713: return(MDOC_HEAD == n->type);
1714: }
1.1 schwarze 1715:
1.47 schwarze 1716: static int
1717: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1718: {
1.1 schwarze 1719:
1.47 schwarze 1720: return(MDOC_BODY == n->type);
1.1 schwarze 1721: }
1722:
1.47 schwarze 1723: /*
1724: * Add a string to the hash table for the current manual.
1725: * Each string has a bitmask telling which macros it belongs to.
1726: * When we finish the manual, we'll dump the table.
1727: */
1.1 schwarze 1728: static void
1.47 schwarze 1729: putkeys(const struct mpage *mpage,
1730: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1731: {
1.90 schwarze 1732: struct ohash *htab;
1.47 schwarze 1733: struct str *s;
1.68 schwarze 1734: const char *end;
1.47 schwarze 1735: unsigned int slot;
1.68 schwarze 1736: int i;
1.1 schwarze 1737:
1.47 schwarze 1738: if (0 == sz)
1739: return;
1.68 schwarze 1740:
1.90 schwarze 1741: if (TYPE_Nm & v) {
1742: htab = &names;
1743: v &= name_mask;
1744: name_mask &= ~NAME_FIRST;
1745: if (debug > 1)
1746: say(mpage->mlinks->file,
1747: "Adding name %*s", sz, cp);
1748: } else {
1749: htab = &strings;
1750: if (debug > 1)
1751: for (i = 0; i < mansearch_keymax; i++)
1752: if (1 << i & v)
1753: say(mpage->mlinks->file,
1754: "Adding key %s=%*s",
1755: mansearch_keynames[i], sz, cp);
1.68 schwarze 1756: }
1.47 schwarze 1757:
1758: end = cp + sz;
1.90 schwarze 1759: slot = ohash_qlookupi(htab, cp, &end);
1760: s = ohash_find(htab, slot);
1.1 schwarze 1761:
1.47 schwarze 1762: if (NULL != s && mpage == s->mpage) {
1763: s->mask |= v;
1.1 schwarze 1764: return;
1.47 schwarze 1765: } else if (NULL == s) {
1766: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1767: memcpy(s->key, cp, sz);
1.90 schwarze 1768: ohash_insert(htab, slot, s);
1.47 schwarze 1769: }
1770: s->mpage = mpage;
1771: s->mask = v;
1.1 schwarze 1772: }
1773:
1774: /*
1.47 schwarze 1775: * Take a Unicode codepoint and produce its UTF-8 encoding.
1776: * This isn't the best way to do this, but it works.
1777: * The magic numbers are from the UTF-8 packaging.
1778: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1779: */
1.47 schwarze 1780: static size_t
1781: utf8(unsigned int cp, char out[7])
1.1 schwarze 1782: {
1.47 schwarze 1783: size_t rc;
1.1 schwarze 1784:
1.47 schwarze 1785: rc = 0;
1786: if (cp <= 0x0000007F) {
1787: rc = 1;
1788: out[0] = (char)cp;
1789: } else if (cp <= 0x000007FF) {
1790: rc = 2;
1791: out[0] = (cp >> 6 & 31) | 192;
1792: out[1] = (cp & 63) | 128;
1793: } else if (cp <= 0x0000FFFF) {
1794: rc = 3;
1795: out[0] = (cp >> 12 & 15) | 224;
1796: out[1] = (cp >> 6 & 63) | 128;
1797: out[2] = (cp & 63) | 128;
1798: } else if (cp <= 0x001FFFFF) {
1799: rc = 4;
1800: out[0] = (cp >> 18 & 7) | 240;
1801: out[1] = (cp >> 12 & 63) | 128;
1802: out[2] = (cp >> 6 & 63) | 128;
1803: out[3] = (cp & 63) | 128;
1804: } else if (cp <= 0x03FFFFFF) {
1805: rc = 5;
1806: out[0] = (cp >> 24 & 3) | 248;
1807: out[1] = (cp >> 18 & 63) | 128;
1808: out[2] = (cp >> 12 & 63) | 128;
1809: out[3] = (cp >> 6 & 63) | 128;
1810: out[4] = (cp & 63) | 128;
1811: } else if (cp <= 0x7FFFFFFF) {
1812: rc = 6;
1813: out[0] = (cp >> 30 & 1) | 252;
1814: out[1] = (cp >> 24 & 63) | 128;
1815: out[2] = (cp >> 18 & 63) | 128;
1816: out[3] = (cp >> 12 & 63) | 128;
1817: out[4] = (cp >> 6 & 63) | 128;
1818: out[5] = (cp & 63) | 128;
1819: } else
1820: return(0);
1.19 schwarze 1821:
1.47 schwarze 1822: out[rc] = '\0';
1823: return(rc);
1.1 schwarze 1824: }
1825:
1.47 schwarze 1826: /*
1.53 schwarze 1827: * Store the rendered version of a key, or alias the pointer
1828: * if the key contains no escape sequences.
1.47 schwarze 1829: */
1830: static void
1.53 schwarze 1831: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1832: {
1.47 schwarze 1833: size_t sz, bsz, pos;
1.71 schwarze 1834: char utfbuf[7], res[6];
1.47 schwarze 1835: char *buf;
1836: const char *seq, *cpp, *val;
1837: int len, u;
1838: enum mandoc_esc esc;
1839:
1.53 schwarze 1840: assert(NULL == key->rendered);
1.47 schwarze 1841:
1842: res[0] = '\\';
1843: res[1] = '\t';
1844: res[2] = ASCII_NBRSP;
1845: res[3] = ASCII_HYPH;
1.71 schwarze 1846: res[4] = ASCII_BREAK;
1847: res[5] = '\0';
1.1 schwarze 1848:
1.47 schwarze 1849: val = key->key;
1850: bsz = strlen(val);
1.1 schwarze 1851:
1852: /*
1.47 schwarze 1853: * Pre-check: if we have no stop-characters, then set the
1854: * pointer as ourselvse and get out of here.
1.1 schwarze 1855: */
1.47 schwarze 1856: if (strcspn(val, res) == bsz) {
1.53 schwarze 1857: key->rendered = key->key;
1.47 schwarze 1858: return;
1859: }
1.1 schwarze 1860:
1.47 schwarze 1861: /* Pre-allocate by the length of the input */
1.39 schwarze 1862:
1.47 schwarze 1863: buf = mandoc_malloc(++bsz);
1864: pos = 0;
1.39 schwarze 1865:
1.47 schwarze 1866: while ('\0' != *val) {
1867: /*
1868: * Halt on the first escape sequence.
1869: * This also halts on the end of string, in which case
1870: * we just copy, fallthrough, and exit the loop.
1871: */
1872: if ((sz = strcspn(val, res)) > 0) {
1873: memcpy(&buf[pos], val, sz);
1874: pos += sz;
1875: val += sz;
1876: }
1.39 schwarze 1877:
1.71 schwarze 1878: switch (*val) {
1879: case (ASCII_HYPH):
1.47 schwarze 1880: buf[pos++] = '-';
1881: val++;
1882: continue;
1.71 schwarze 1883: case ('\t'):
1884: /* FALLTHROUGH */
1885: case (ASCII_NBRSP):
1.47 schwarze 1886: buf[pos++] = ' ';
1887: val++;
1.71 schwarze 1888: /* FALLTHROUGH */
1889: case (ASCII_BREAK):
1.47 schwarze 1890: continue;
1.71 schwarze 1891: default:
1892: break;
1893: }
1894: if ('\\' != *val)
1.47 schwarze 1895: break;
1.39 schwarze 1896:
1.47 schwarze 1897: /* Read past the slash. */
1.39 schwarze 1898:
1.47 schwarze 1899: val++;
1.39 schwarze 1900:
1.47 schwarze 1901: /*
1902: * Parse the escape sequence and see if it's a
1903: * predefined character or special character.
1904: */
1.52 schwarze 1905:
1.47 schwarze 1906: esc = mandoc_escape
1907: ((const char **)&val, &seq, &len);
1908: if (ESCAPE_ERROR == esc)
1909: break;
1910: if (ESCAPE_SPECIAL != esc)
1911: continue;
1.39 schwarze 1912:
1.47 schwarze 1913: /*
1.52 schwarze 1914: * Render the special character
1915: * as either UTF-8 or ASCII.
1.47 schwarze 1916: */
1.52 schwarze 1917:
1918: if (write_utf8) {
1919: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1920: continue;
1921: cpp = utfbuf;
1922: if (0 == (sz = utf8(u, utfbuf)))
1923: continue;
1924: sz = strlen(cpp);
1925: } else {
1926: cpp = mchars_spec2str(mc, seq, len, &sz);
1927: if (NULL == cpp)
1928: continue;
1929: if (ASCII_NBRSP == *cpp) {
1930: cpp = " ";
1931: sz = 1;
1932: }
1933: }
1.1 schwarze 1934:
1.47 schwarze 1935: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1936:
1.47 schwarze 1937: bsz += sz;
1938: buf = mandoc_realloc(buf, bsz);
1939: memcpy(&buf[pos], cpp, sz);
1940: pos += sz;
1.1 schwarze 1941: }
1942:
1.47 schwarze 1943: buf[pos] = '\0';
1.53 schwarze 1944: key->rendered = buf;
1.1 schwarze 1945: }
1946:
1.75 schwarze 1947: static void
1948: dbadd_mlink(const struct mlink *mlink)
1949: {
1950: size_t i;
1951:
1952: i = 1;
1953: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1954: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1955: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1956: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1957: SQL_STEP(stmts[STMT_INSERT_LINK]);
1958: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1959: }
1960:
1.11 schwarze 1961: /*
1.47 schwarze 1962: * Flush the current page's terms (and their bits) into the database.
1963: * Wrap the entire set of additions in a transaction to make sqlite be a
1964: * little faster.
1.53 schwarze 1965: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1966: */
1967: static void
1.75 schwarze 1968: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1969: {
1.47 schwarze 1970: struct mlink *mlink;
1971: struct str *key;
1972: size_t i;
1973: unsigned int slot;
1974:
1.85 schwarze 1975: mlink = mpage->mlinks;
1.11 schwarze 1976:
1.85 schwarze 1977: if (nodb) {
1978: while (NULL != mlink) {
1979: fputs(mlink->name, stdout);
1980: if (NULL == mlink->next ||
1981: strcmp(mlink->dsec, mlink->next->dsec) ||
1982: strcmp(mlink->fsec, mlink->next->fsec) ||
1983: strcmp(mlink->arch, mlink->next->arch)) {
1984: putchar('(');
1985: if ('\0' == *mlink->dsec)
1986: fputs(mlink->fsec, stdout);
1987: else
1988: fputs(mlink->dsec, stdout);
1989: if ('\0' != *mlink->arch)
1990: printf("/%s", mlink->arch);
1991: putchar(')');
1992: }
1993: mlink = mlink->next;
1994: if (NULL != mlink)
1995: fputs(", ", stdout);
1996: }
1.89 schwarze 1997: printf(" - %s\n", mpage->desc);
1.11 schwarze 1998: return;
1.85 schwarze 1999: }
2000:
2001: if (debug)
2002: say(mlink->file, "Adding to database");
1.47 schwarze 2003:
2004: i = 1;
1.89 schwarze 2005: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc);
1.47 schwarze 2006: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
2007: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2008: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2009: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2010:
1.85 schwarze 2011: while (NULL != mlink) {
1.75 schwarze 2012: dbadd_mlink(mlink);
1.85 schwarze 2013: mlink = mlink->next;
2014: }
1.91 schwarze 2015: mlink = mpage->mlinks;
1.47 schwarze 2016:
1.90 schwarze 2017: for (key = ohash_first(&names, &slot); NULL != key;
2018: key = ohash_next(&names, &slot)) {
2019: assert(key->mpage == mpage);
2020: if (NULL == key->rendered)
2021: render_key(mc, key);
2022: i = 1;
2023: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2024: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2025: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2026: SQL_STEP(stmts[STMT_INSERT_NAME]);
2027: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2028: if (key->rendered != key->key)
2029: free(key->rendered);
2030: free(key);
2031: }
1.47 schwarze 2032: for (key = ohash_first(&strings, &slot); NULL != key;
2033: key = ohash_next(&strings, &slot)) {
2034: assert(key->mpage == mpage);
1.53 schwarze 2035: if (NULL == key->rendered)
2036: render_key(mc, key);
1.47 schwarze 2037: i = 1;
2038: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2039: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2040: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2041: SQL_STEP(stmts[STMT_INSERT_KEY]);
2042: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2043: if (key->rendered != key->key)
2044: free(key->rendered);
1.47 schwarze 2045: free(key);
1.33 schwarze 2046: }
1.47 schwarze 2047: }
1.41 deraadt 2048:
1.47 schwarze 2049: static void
2050: dbprune(void)
2051: {
2052: struct mpage *mpage;
2053: struct mlink *mlink;
2054: size_t i;
2055: unsigned int slot;
1.11 schwarze 2056:
1.63 schwarze 2057: if (0 == nodb)
2058: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2059:
1.63 schwarze 2060: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2061: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2062: mlink = mpage->mlinks;
1.82 schwarze 2063: if (debug)
1.63 schwarze 2064: say(mlink->file, "Deleting from database");
2065: if (nodb)
2066: continue;
2067: for ( ; NULL != mlink; mlink = mlink->next) {
2068: i = 1;
2069: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2070: i, mlink->dsec);
2071: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2072: i, mlink->arch);
2073: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2074: i, mlink->name);
2075: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2076: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2077: }
1.11 schwarze 2078: }
1.63 schwarze 2079:
2080: if (0 == nodb)
2081: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2082: }
1.22 schwarze 2083:
1.47 schwarze 2084: /*
2085: * Close an existing database and its prepared statements.
2086: * If "real" is not set, rename the temporary file into the real one.
2087: */
2088: static void
2089: dbclose(int real)
2090: {
2091: size_t i;
1.72 schwarze 2092: int status;
2093: pid_t child;
1.11 schwarze 2094:
1.47 schwarze 2095: if (nodb)
2096: return;
1.11 schwarze 2097:
1.47 schwarze 2098: for (i = 0; i < STMT__MAX; i++) {
2099: sqlite3_finalize(stmts[i]);
2100: stmts[i] = NULL;
1.28 schwarze 2101: }
1.22 schwarze 2102:
1.47 schwarze 2103: sqlite3_close(db);
2104: db = NULL;
1.11 schwarze 2105:
1.47 schwarze 2106: if (real)
2107: return;
1.22 schwarze 2108:
1.72 schwarze 2109: if ('\0' == *tempfilename) {
2110: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2111: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2112: say(MANDOC_DB, "&rename");
1.72 schwarze 2113: }
2114: return;
2115: }
2116:
2117: switch (child = fork()) {
2118: case (-1):
2119: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2120: say("", "&fork cmp");
1.72 schwarze 2121: return;
2122: case (0):
2123: execlp("cmp", "cmp", "-s",
2124: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2125: say("", "&exec cmp");
1.72 schwarze 2126: exit(0);
2127: default:
2128: break;
2129: }
2130: if (-1 == waitpid(child, &status, 0)) {
2131: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2132: say("", "&wait cmp");
1.72 schwarze 2133: } else if (WIFSIGNALED(status)) {
2134: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2135: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2136: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2137: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2138: say(MANDOC_DB,
2139: "Data changed, but cannot replace database");
2140: }
2141:
2142: *strrchr(tempfilename, '/') = '\0';
2143: switch (child = fork()) {
2144: case (-1):
2145: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2146: say("", "&fork rm");
1.72 schwarze 2147: return;
2148: case (0):
2149: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2150: say("", "&exec rm");
1.72 schwarze 2151: exit((int)MANDOCLEVEL_SYSERR);
2152: default:
2153: break;
2154: }
2155: if (-1 == waitpid(child, &status, 0)) {
2156: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2157: say("", "&wait rm");
1.72 schwarze 2158: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2159: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2160: say("", "%s: Cannot remove temporary directory",
2161: tempfilename);
1.22 schwarze 2162: }
1.11 schwarze 2163: }
2164:
1.47 schwarze 2165: /*
2166: * This is straightforward stuff.
2167: * Open a database connection to a "temporary" database, then open a set
2168: * of prepared statements we'll use over and over again.
2169: * If "real" is set, we use the existing database; if not, we truncate a
2170: * temporary one.
2171: * Must be matched by dbclose().
2172: */
2173: static int
2174: dbopen(int real)
1.2 schwarze 2175: {
1.72 schwarze 2176: const char *sql;
1.47 schwarze 2177: int rc, ofl;
1.6 schwarze 2178:
1.47 schwarze 2179: if (nodb)
2180: return(1);
1.6 schwarze 2181:
1.72 schwarze 2182: *tempfilename = '\0';
1.47 schwarze 2183: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2184:
2185: if (real) {
2186: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2187: if (SQLITE_OK != rc) {
1.47 schwarze 2188: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2189: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2190: return(0);
1.28 schwarze 2191: }
1.72 schwarze 2192: goto prepare_statements;
2193: }
2194:
2195: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2196:
1.72 schwarze 2197: remove(MANDOC_DB "~");
2198: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.47 schwarze 2199: if (SQLITE_OK == rc)
1.72 schwarze 2200: goto create_tables;
1.73 schwarze 2201: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2202: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2203: say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
1.47 schwarze 2204: return(0);
2205: }
1.6 schwarze 2206:
1.72 schwarze 2207: if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2208: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2209: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2210: say("", "/tmp/mandocdb.XXXXXX: Filename too long");
1.72 schwarze 2211: return(0);
2212: }
2213: if (NULL == mkdtemp(tempfilename)) {
2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2215: say("", "&%s", tempfilename);
1.72 schwarze 2216: return(0);
2217: }
2218: if (strlcat(tempfilename, "/" MANDOC_DB,
2219: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2220: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2221: say("", "%s/" MANDOC_DB ": Filename too long",
2222: tempfilename);
1.72 schwarze 2223: return(0);
2224: }
2225: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2226: if (SQLITE_OK != rc) {
1.47 schwarze 2227: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2228: say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
1.47 schwarze 2229: return(0);
1.2 schwarze 2230: }
2231:
1.72 schwarze 2232: create_tables:
1.47 schwarze 2233: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2234: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2235: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2236: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2237: ");\n"
2238: "\n"
2239: "CREATE TABLE \"mlinks\" (\n"
2240: " \"sec\" TEXT NOT NULL,\n"
2241: " \"arch\" TEXT NOT NULL,\n"
2242: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2243: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2244: "ON DELETE CASCADE\n"
1.47 schwarze 2245: ");\n"
1.93 schwarze 2246: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2247: "\n"
1.90 schwarze 2248: "CREATE TABLE \"names\" (\n"
2249: " \"bits\" INTEGER NOT NULL,\n"
2250: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2251: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2252: "ON DELETE CASCADE\n"
2253: ");\n"
2254: "\n"
1.47 schwarze 2255: "CREATE TABLE \"keys\" (\n"
2256: " \"bits\" INTEGER NOT NULL,\n"
2257: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2258: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2259: "ON DELETE CASCADE\n"
1.93 schwarze 2260: ");\n"
2261: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2262:
2263: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2264: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2265: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2266: return(0);
1.2 schwarze 2267: }
2268:
1.47 schwarze 2269: prepare_statements:
2270: SQL_EXEC("PRAGMA foreign_keys = ON");
1.94 schwarze 2271: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2272: "(SELECT pageid FROM mlinks WHERE "
2273: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2274: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2275: sql = "INSERT INTO mpages "
1.89 schwarze 2276: "(desc,form) VALUES (?,?)";
1.47 schwarze 2277: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2278: sql = "INSERT INTO mlinks "
1.61 schwarze 2279: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2280: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2281: sql = "INSERT INTO names "
2282: "(bits,name,pageid) VALUES (?,?,?)";
2283: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2284: sql = "INSERT INTO keys "
2285: "(bits,key,pageid) VALUES (?,?,?)";
2286: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2287:
1.47 schwarze 2288: /*
2289: * When opening a new database, we can turn off
2290: * synchronous mode for much better performance.
2291: */
1.6 schwarze 2292:
1.47 schwarze 2293: if (real)
2294: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 2295:
1.47 schwarze 2296: return(1);
2297: }
1.6 schwarze 2298:
1.47 schwarze 2299: static void *
2300: hash_halloc(size_t sz, void *arg)
2301: {
1.6 schwarze 2302:
1.47 schwarze 2303: return(mandoc_calloc(sz, 1));
2304: }
1.2 schwarze 2305:
1.47 schwarze 2306: static void *
2307: hash_alloc(size_t sz, void *arg)
2308: {
1.28 schwarze 2309:
1.47 schwarze 2310: return(mandoc_malloc(sz));
2311: }
1.6 schwarze 2312:
1.47 schwarze 2313: static void
2314: hash_free(void *p, size_t sz, void *arg)
2315: {
1.26 schwarze 2316:
1.47 schwarze 2317: free(p);
2318: }
1.6 schwarze 2319:
1.47 schwarze 2320: static int
2321: set_basedir(const char *targetdir)
2322: {
2323: static char startdir[PATH_MAX];
2324: static int fd;
1.6 schwarze 2325:
1.47 schwarze 2326: /*
2327: * Remember where we started by keeping a fd open to the origin
2328: * path component: throughout this utility, we chdir() a lot to
2329: * handle relative paths, and by doing this, we can return to
2330: * the starting point.
2331: */
2332: if ('\0' == *startdir) {
2333: if (NULL == getcwd(startdir, PATH_MAX)) {
2334: exitcode = (int)MANDOCLEVEL_SYSERR;
2335: if (NULL != targetdir)
1.80 schwarze 2336: say("", "&getcwd");
1.47 schwarze 2337: return(0);
2338: }
2339: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2340: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2341: say("", "&open %s", startdir);
1.47 schwarze 2342: return(0);
1.11 schwarze 2343: }
1.47 schwarze 2344: if (NULL == targetdir)
2345: targetdir = startdir;
2346: } else {
2347: if (-1 == fd)
2348: return(0);
2349: if (-1 == fchdir(fd)) {
2350: close(fd);
2351: basedir[0] = '\0';
2352: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2353: say("", "&chdir %s", startdir);
1.47 schwarze 2354: return(0);
1.2 schwarze 2355: }
1.47 schwarze 2356: if (NULL == targetdir) {
2357: close(fd);
2358: return(1);
1.2 schwarze 2359: }
2360: }
1.47 schwarze 2361: if (NULL == realpath(targetdir, basedir)) {
2362: basedir[0] = '\0';
2363: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2364: say("", "&%s: realpath", targetdir);
1.47 schwarze 2365: return(0);
2366: } else if (-1 == chdir(basedir)) {
2367: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2368: say("", "&chdir");
1.47 schwarze 2369: return(0);
2370: }
2371: return(1);
1.2 schwarze 2372: }
2373:
2374: static void
1.47 schwarze 2375: say(const char *file, const char *format, ...)
1.2 schwarze 2376: {
1.47 schwarze 2377: va_list ap;
1.80 schwarze 2378: int use_errno;
1.2 schwarze 2379:
1.47 schwarze 2380: if ('\0' != *basedir)
2381: fprintf(stderr, "%s", basedir);
2382: if ('\0' != *basedir && '\0' != *file)
2383: fputs("//", stderr);
2384: if ('\0' != *file)
2385: fprintf(stderr, "%s", file);
1.31 schwarze 2386:
1.80 schwarze 2387: use_errno = 1;
2388: if (NULL != format) {
2389: switch (*format) {
2390: case ('&'):
2391: format++;
2392: break;
2393: case ('\0'):
2394: format = NULL;
2395: break;
2396: default:
2397: use_errno = 0;
2398: break;
2399: }
2400: }
2401: if (NULL != format) {
2402: if ('\0' != *basedir || '\0' != *file)
2403: fputs(": ", stderr);
2404: va_start(ap, format);
2405: vfprintf(stderr, format, ap);
2406: va_end(ap);
2407: }
2408: if (use_errno) {
2409: if ('\0' != *basedir || '\0' != *file || NULL != format)
2410: fputs(": ", stderr);
1.47 schwarze 2411: perror(NULL);
1.80 schwarze 2412: } else
2413: fputc('\n', stderr);
1.1 schwarze 2414: }
![[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