Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.97
1.97 ! schwarze 1: /* $Id: mandocdb.c,v 1.96 2014/04/19 02:29:12 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) {
1.2 schwarze 425:
1.47 schwarze 426: /*
427: * All of these deal with a specific directory.
1.97 ! schwarze 428: * Jump into that directory first.
1.47 schwarze 429: */
430: if (0 == set_basedir(path_arg))
431: goto out;
1.97 ! schwarze 432:
1.95 schwarze 433: if (dbopen(1)) {
1.97 ! schwarze 434: /*
! 435: * The existing database is usable. Process
! 436: * all files specified on the command-line.
! 437: */
! 438: use_all = 1;
1.95 schwarze 439: for (i = 0; i < argc; i++)
440: filescan(argv[i]);
441: if (OP_TEST != op)
442: dbprune();
443: } else {
444: /*
445: * Database missing or corrupt.
446: * Recreate from scratch.
447: */
1.97 ! schwarze 448: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 449: op = OP_DEFAULT;
450: if (0 == treescan())
451: goto out;
452: if (0 == dbopen(0))
453: goto out;
454: }
1.47 schwarze 455: if (OP_DELETE != op)
1.58 schwarze 456: mpages_merge(mc, mp);
1.95 schwarze 457: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 458: } else {
459: /*
460: * If we have arguments, use them as our manpaths.
461: * If we don't, grok from manpath(1) or however else
462: * manpath_parse() wants to do it.
463: */
464: if (argc > 0) {
465: dirs.paths = mandoc_calloc
466: (argc, sizeof(char *));
467: dirs.sz = (size_t)argc;
468: for (i = 0; i < argc; i++)
469: dirs.paths[i] = mandoc_strdup(argv[i]);
470: } else
471: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 472:
473: if (0 == dirs.sz) {
474: exitcode = (int)MANDOCLEVEL_BADARG;
475: say("", "Empty manpath");
476: }
1.2 schwarze 477:
1.47 schwarze 478: /*
479: * First scan the tree rooted at a base directory, then
480: * build a new database and finally move it into place.
481: * Ignore zero-length directories and strip trailing
482: * slashes.
483: */
484: for (j = 0; j < dirs.sz; j++) {
485: sz = strlen(dirs.paths[j]);
486: if (sz && '/' == dirs.paths[j][sz - 1])
487: dirs.paths[j][--sz] = '\0';
488: if (0 == sz)
489: continue;
1.2 schwarze 490:
1.47 schwarze 491: if (j) {
492: ohash_init(&mpages, 6, &mpages_info);
493: ohash_init(&mlinks, 6, &mlinks_info);
494: }
1.1 schwarze 495:
1.47 schwarze 496: if (0 == set_basedir(dirs.paths[j]))
497: goto out;
498: if (0 == treescan())
499: goto out;
500: if (0 == set_basedir(dirs.paths[j]))
501: goto out;
502: if (0 == dbopen(0))
503: goto out;
1.41 deraadt 504:
1.58 schwarze 505: mpages_merge(mc, mp);
1.87 schwarze 506: if (warnings &&
507: ! (MPARSE_QUICK & mparse_options))
508: names_check();
1.47 schwarze 509: dbclose(0);
1.28 schwarze 510:
1.47 schwarze 511: if (j + 1 < dirs.sz) {
512: mpages_free();
513: ohash_delete(&mpages);
514: ohash_delete(&mlinks);
515: }
1.2 schwarze 516: }
1.47 schwarze 517: }
518: out:
519: set_basedir(NULL);
520: manpath_free(&dirs);
521: mchars_free(mc);
522: mparse_free(mp);
523: mpages_free();
524: ohash_delete(&mpages);
525: ohash_delete(&mlinks);
526: return(exitcode);
527: usage:
1.83 schwarze 528: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
529: " %s [-aDnpQ] [-Tutf8] dir ...\n"
530: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
531: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 532: " %s [-Q] -t file ...\n",
1.47 schwarze 533: progname, progname, progname,
534: progname, progname);
1.1 schwarze 535:
1.47 schwarze 536: return((int)MANDOCLEVEL_BADARG);
537: }
1.1 schwarze 538:
1.47 schwarze 539: /*
540: * Scan a directory tree rooted at "basedir" for manpages.
541: * We use fts(), scanning directory parts along the way for clues to our
542: * section and architecture.
543: *
544: * If use_all has been specified, grok all files.
545: * If not, sanitise paths to the following:
546: *
547: * [./]man*[/<arch>]/<name>.<section>
548: * or
549: * [./]cat<section>[/<arch>]/<name>.0
550: *
551: * TODO: accomodate for multi-language directories.
552: */
553: static int
554: treescan(void)
555: {
1.96 schwarze 556: char buf[PATH_MAX];
1.47 schwarze 557: FTS *f;
558: FTSENT *ff;
559: struct mlink *mlink;
1.81 schwarze 560: int dform, gzip;
1.51 schwarze 561: char *dsec, *arch, *fsec, *cp;
562: const char *path;
1.47 schwarze 563: const char *argv[2];
1.1 schwarze 564:
1.47 schwarze 565: argv[0] = ".";
566: argv[1] = (char *)NULL;
1.1 schwarze 567:
1.96 schwarze 568: f = fts_open((char * const *)argv,
569: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 570: if (NULL == f) {
571: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 572: say("", "&fts_open");
1.47 schwarze 573: return(0);
574: }
1.2 schwarze 575:
1.47 schwarze 576: dsec = arch = NULL;
577: dform = FORM_NONE;
1.2 schwarze 578:
1.47 schwarze 579: while (NULL != (ff = fts_read(f))) {
580: path = ff->fts_path + 2;
1.96 schwarze 581: switch (ff->fts_info) {
582:
583: /*
584: * Symbolic links require various sanity checks,
585: * then get handled just like regular files.
586: */
587: case (FTS_SL):
588: if (NULL == realpath(path, buf)) {
589: if (warnings)
590: say(path, "&realpath");
591: continue;
592: }
593: if (strstr(buf, basedir) != buf) {
594: if (warnings) say("",
595: "%s: outside base directory", buf);
596: continue;
597: }
598: /* Use logical inode to avoid mpages dupe. */
599: if (-1 == stat(path, ff->fts_statp)) {
600: if (warnings)
601: say(path, "&stat");
602: continue;
603: }
604: /* FALLTHROUGH */
605:
1.15 schwarze 606: /*
1.47 schwarze 607: * If we're a regular file, add an mlink by using the
608: * stored directory data and handling the filename.
1.15 schwarze 609: */
1.96 schwarze 610: case (FTS_F):
1.47 schwarze 611: if (0 == strcmp(path, MANDOC_DB))
612: continue;
613: if ( ! use_all && ff->fts_level < 2) {
614: if (warnings)
615: say(path, "Extraneous file");
616: continue;
1.81 schwarze 617: }
618: gzip = 0;
619: fsec = NULL;
620: while (NULL == fsec) {
621: fsec = strrchr(ff->fts_name, '.');
622: if (NULL == fsec || strcmp(fsec+1, "gz"))
623: break;
624: gzip = 1;
625: *fsec = '\0';
626: fsec = NULL;
627: }
628: if (NULL == fsec) {
1.47 schwarze 629: if ( ! use_all) {
630: if (warnings)
631: say(path,
632: "No filename suffix");
633: continue;
634: }
635: } else if (0 == strcmp(++fsec, "html")) {
636: if (warnings)
637: say(path, "Skip html");
638: continue;
639: } else if (0 == strcmp(fsec, "ps")) {
640: if (warnings)
641: say(path, "Skip ps");
642: continue;
643: } else if (0 == strcmp(fsec, "pdf")) {
644: if (warnings)
645: say(path, "Skip pdf");
646: continue;
647: } else if ( ! use_all &&
648: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
649: (FORM_CAT == dform && strcmp(fsec, "0")))) {
650: if (warnings)
651: say(path, "Wrong filename suffix");
652: continue;
653: } else
654: fsec[-1] = '\0';
1.51 schwarze 655:
1.47 schwarze 656: mlink = mandoc_calloc(1, sizeof(struct mlink));
657: strlcpy(mlink->file, path, sizeof(mlink->file));
658: mlink->dform = dform;
1.51 schwarze 659: mlink->dsec = dsec;
660: mlink->arch = arch;
661: mlink->name = ff->fts_name;
662: mlink->fsec = fsec;
1.81 schwarze 663: mlink->gzip = gzip;
1.47 schwarze 664: mlink_add(mlink, ff->fts_statp);
665: continue;
1.96 schwarze 666:
667: case (FTS_D):
668: /* FALLTHROUGH */
669: case (FTS_DP):
670: break;
671:
672: default:
1.47 schwarze 673: if (warnings)
674: say(path, "Not a regular file");
675: continue;
676: }
677:
678: switch (ff->fts_level) {
679: case (0):
680: /* Ignore the root directory. */
681: break;
682: case (1):
683: /*
684: * This might contain manX/ or catX/.
685: * Try to infer this from the name.
686: * If we're not in use_all, enforce it.
687: */
688: cp = ff->fts_name;
689: if (FTS_DP == ff->fts_info)
690: break;
1.15 schwarze 691:
1.47 schwarze 692: if (0 == strncmp(cp, "man", 3)) {
693: dform = FORM_SRC;
694: dsec = cp + 3;
695: } else if (0 == strncmp(cp, "cat", 3)) {
696: dform = FORM_CAT;
697: dsec = cp + 3;
1.51 schwarze 698: } else {
699: dform = FORM_NONE;
700: dsec = NULL;
1.26 schwarze 701: }
1.47 schwarze 702:
703: if (NULL != dsec || use_all)
704: break;
705:
706: if (warnings)
707: say(path, "Unknown directory part");
708: fts_set(f, ff, FTS_SKIP);
709: break;
710: case (2):
711: /*
712: * Possibly our architecture.
713: * If we're descending, keep tabs on it.
714: */
715: if (FTS_DP != ff->fts_info && NULL != dsec)
716: arch = ff->fts_name;
1.51 schwarze 717: else
718: arch = NULL;
1.47 schwarze 719: break;
720: default:
721: if (FTS_DP == ff->fts_info || use_all)
722: break;
723: if (warnings)
724: say(path, "Extraneous directory part");
725: fts_set(f, ff, FTS_SKIP);
726: break;
1.14 schwarze 727: }
1.47 schwarze 728: }
729:
730: fts_close(f);
731: return(1);
732: }
1.1 schwarze 733:
1.47 schwarze 734: /*
735: * Add a file to the mlinks table.
736: * Do not verify that it's a "valid" looking manpage (we'll do that
737: * later).
738: *
739: * Try to infer the manual section, architecture, and page name from the
740: * path, assuming it looks like
741: *
742: * [./]man*[/<arch>]/<name>.<section>
743: * or
744: * [./]cat<section>[/<arch>]/<name>.0
745: *
746: * See treescan() for the fts(3) version of this.
747: */
748: static void
749: filescan(const char *file)
750: {
751: char buf[PATH_MAX];
752: struct stat st;
753: struct mlink *mlink;
754: char *p, *start;
755:
756: assert(use_all);
757:
758: if (0 == strncmp(file, "./", 2))
759: file += 2;
760:
1.96 schwarze 761: /*
762: * We have to do lstat(2) before realpath(3) loses
763: * the information whether this is a symbolic link.
764: * We need to know that because for symbolic links,
765: * we want to use the orginal file name, while for
766: * regular files, we want to use the real path.
767: */
768: if (-1 == lstat(file, &st)) {
769: exitcode = (int)MANDOCLEVEL_BADARG;
770: say(file, "&lstat");
771: return;
772: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
773: exitcode = (int)MANDOCLEVEL_BADARG;
774: say(file, "Not a regular file");
775: return;
776: }
777:
778: /*
779: * We have to resolve the file name to the real path
780: * in any case for the base directory check.
781: */
1.47 schwarze 782: if (NULL == realpath(file, buf)) {
783: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 784: say(file, "&realpath");
1.47 schwarze 785: return;
1.63 schwarze 786: }
787:
788: if (strstr(buf, basedir) == buf)
789: start = buf + strlen(basedir) + 1;
790: else if (OP_TEST == op)
791: start = buf;
792: else {
1.47 schwarze 793: exitcode = (int)MANDOCLEVEL_BADARG;
794: say("", "%s: outside base directory", buf);
795: return;
1.63 schwarze 796: }
797:
1.96 schwarze 798: /*
799: * Now we are sure the file is inside our tree.
800: * If it is a symbolic link, ignore the real path
801: * and use the original name.
802: * This implies passing stuff like "cat1/../man1/foo.1"
803: * on the command line won't work. So don't do that.
804: * Note the stat(2) can still fail if the link target
805: * doesn't exist.
806: */
807: if (S_IFLNK & st.st_mode) {
808: if (-1 == stat(buf, &st)) {
809: exitcode = (int)MANDOCLEVEL_BADARG;
810: say(file, "&stat");
811: return;
812: }
813: strlcpy(buf, file, sizeof(buf));
814: start = strstr(buf, basedir) == buf ?
815: buf + strlen(basedir) + 1 : buf;
1.1 schwarze 816: }
1.63 schwarze 817:
1.47 schwarze 818: mlink = mandoc_calloc(1, sizeof(struct mlink));
819: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 820:
1.10 schwarze 821: /*
1.47 schwarze 822: * First try to guess our directory structure.
823: * If we find a separator, try to look for man* or cat*.
824: * If we find one of these and what's underneath is a directory,
825: * assume it's an architecture.
1.10 schwarze 826: */
1.47 schwarze 827: if (NULL != (p = strchr(start, '/'))) {
828: *p++ = '\0';
829: if (0 == strncmp(start, "man", 3)) {
830: mlink->dform = FORM_SRC;
1.51 schwarze 831: mlink->dsec = start + 3;
1.47 schwarze 832: } else if (0 == strncmp(start, "cat", 3)) {
833: mlink->dform = FORM_CAT;
1.51 schwarze 834: mlink->dsec = start + 3;
1.47 schwarze 835: }
1.10 schwarze 836:
1.47 schwarze 837: start = p;
838: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
839: *p++ = '\0';
1.51 schwarze 840: mlink->arch = start;
1.47 schwarze 841: start = p;
1.41 deraadt 842: }
1.47 schwarze 843: }
1.7 schwarze 844:
1.47 schwarze 845: /*
846: * Now check the file suffix.
847: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
848: */
849: p = strrchr(start, '\0');
850: while (p-- > start && '/' != *p && '.' != *p)
851: /* Loop. */ ;
852:
853: if ('.' == *p) {
854: *p++ = '\0';
1.51 schwarze 855: mlink->fsec = p;
1.47 schwarze 856: }
1.41 deraadt 857:
1.47 schwarze 858: /*
859: * Now try to parse the name.
860: * Use the filename portion of the path.
861: */
862: mlink->name = start;
863: if (NULL != (p = strrchr(start, '/'))) {
864: mlink->name = p + 1;
865: *p = '\0';
866: }
867: mlink_add(mlink, &st);
868: }
1.2 schwarze 869:
1.47 schwarze 870: static void
871: mlink_add(struct mlink *mlink, const struct stat *st)
872: {
873: struct inodev inodev;
874: struct mpage *mpage;
875: unsigned int slot;
876:
877: assert(NULL != mlink->file);
878:
1.51 schwarze 879: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
880: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
881: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
882: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 883:
884: if ('0' == *mlink->fsec) {
885: free(mlink->fsec);
886: mlink->fsec = mandoc_strdup(mlink->dsec);
887: mlink->fform = FORM_CAT;
888: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
889: mlink->fform = FORM_SRC;
890: else
891: mlink->fform = FORM_NONE;
1.1 schwarze 892:
1.47 schwarze 893: slot = ohash_qlookup(&mlinks, mlink->file);
894: assert(NULL == ohash_find(&mlinks, slot));
895: ohash_insert(&mlinks, slot, mlink);
896:
897: inodev.st_ino = st->st_ino;
898: inodev.st_dev = st->st_dev;
899: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
900: sizeof(struct inodev), inodev.st_ino);
901: mpage = ohash_find(&mpages, slot);
902: if (NULL == mpage) {
903: mpage = mandoc_calloc(1, sizeof(struct mpage));
904: mpage->inodev.st_ino = inodev.st_ino;
905: mpage->inodev.st_dev = inodev.st_dev;
906: ohash_insert(&mpages, slot, mpage);
907: } else
908: mlink->next = mpage->mlinks;
909: mpage->mlinks = mlink;
1.75 schwarze 910: mlink->mpage = mpage;
1.47 schwarze 911: }
1.1 schwarze 912:
1.47 schwarze 913: static void
914: mlink_free(struct mlink *mlink)
915: {
1.1 schwarze 916:
1.47 schwarze 917: free(mlink->dsec);
918: free(mlink->arch);
919: free(mlink->name);
920: free(mlink->fsec);
921: free(mlink);
922: }
1.1 schwarze 923:
1.47 schwarze 924: static void
925: mpages_free(void)
926: {
927: struct mpage *mpage;
928: struct mlink *mlink;
929: unsigned int slot;
930:
931: mpage = ohash_first(&mpages, &slot);
932: while (NULL != mpage) {
933: while (NULL != (mlink = mpage->mlinks)) {
934: mpage->mlinks = mlink->next;
935: mlink_free(mlink);
936: }
937: free(mpage->sec);
938: free(mpage->arch);
939: free(mpage->title);
940: free(mpage->desc);
941: free(mpage);
942: mpage = ohash_next(&mpages, &slot);
943: }
944: }
1.1 schwarze 945:
1.47 schwarze 946: /*
947: * For each mlink to the mpage, check whether the path looks like
948: * it is formatted, and if it does, check whether a source manual
949: * exists by the same name, ignoring the suffix.
950: * If both conditions hold, drop the mlink.
951: */
952: static void
953: mlinks_undupe(struct mpage *mpage)
954: {
955: char buf[PATH_MAX];
956: struct mlink **prev;
957: struct mlink *mlink;
958: char *bufp;
959:
960: mpage->form = FORM_CAT;
961: prev = &mpage->mlinks;
962: while (NULL != (mlink = *prev)) {
963: if (FORM_CAT != mlink->dform) {
964: mpage->form = FORM_NONE;
965: goto nextlink;
966: }
967: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
968: if (warnings)
969: say(mlink->file, "Filename too long");
970: goto nextlink;
1.26 schwarze 971: }
1.47 schwarze 972: bufp = strstr(buf, "cat");
973: assert(NULL != bufp);
974: memcpy(bufp, "man", 3);
975: if (NULL != (bufp = strrchr(buf, '.')))
976: *++bufp = '\0';
977: strlcat(buf, mlink->dsec, PATH_MAX);
978: if (NULL == ohash_find(&mlinks,
979: ohash_qlookup(&mlinks, buf)))
980: goto nextlink;
981: if (warnings)
982: say(mlink->file, "Man source exists: %s", buf);
983: if (use_all)
984: goto nextlink;
985: *prev = mlink->next;
986: mlink_free(mlink);
987: continue;
988: nextlink:
989: prev = &(*prev)->next;
1.1 schwarze 990: }
1.47 schwarze 991: }
1.1 schwarze 992:
1.88 schwarze 993: static void
1.50 schwarze 994: mlink_check(struct mpage *mpage, struct mlink *mlink)
995: {
1.88 schwarze 996: struct str *str;
997: unsigned int slot;
1.50 schwarze 998:
999: /*
1000: * Check whether the manual section given in a file
1001: * agrees with the directory where the file is located.
1002: * Some manuals have suffixes like (3p) on their
1003: * section number either inside the file or in the
1004: * directory name, some are linked into more than one
1005: * section, like encrypt(1) = makekey(8).
1006: */
1007:
1008: if (FORM_SRC == mpage->form &&
1.88 schwarze 1009: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1010: say(mlink->file, "Section \"%s\" manual in %s directory",
1011: mpage->sec, mlink->dsec);
1012:
1013: /*
1014: * Manual page directories exist for each kernel
1015: * architecture as returned by machine(1).
1016: * However, many manuals only depend on the
1017: * application architecture as returned by arch(1).
1018: * For example, some (2/ARM) manuals are shared
1019: * across the "armish" and "zaurus" kernel
1020: * architectures.
1021: * A few manuals are even shared across completely
1022: * different architectures, for example fdformat(1)
1023: * on amd64, i386, sparc, and sparc64.
1024: */
1025:
1.88 schwarze 1026: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1027: say(mlink->file, "Architecture \"%s\" manual in "
1028: "\"%s\" directory", mpage->arch, mlink->arch);
1029:
1.88 schwarze 1030: /*
1031: * XXX
1.90 schwarze 1032: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1033: */
1034:
1035: if (FORM_CAT == mpage->form)
1036: return;
1037:
1038: /*
1039: * Check whether this mlink
1040: * appears as a name in the NAME section.
1041: */
1.50 schwarze 1042:
1.90 schwarze 1043: slot = ohash_qlookup(&names, mlink->name);
1044: str = ohash_find(&names, slot);
1.88 schwarze 1045: assert(NULL != str);
1.90 schwarze 1046: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1047: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1048: }
1049:
1.47 schwarze 1050: /*
1051: * Run through the files in the global vector "mpages"
1052: * and add them to the database specified in "basedir".
1053: *
1054: * This handles the parsing scheme itself, using the cues of directory
1055: * and filename to determine whether the file is parsable or not.
1056: */
1057: static void
1.58 schwarze 1058: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1059: {
1.70 schwarze 1060: char any[] = "any";
1.58 schwarze 1061: struct ohash_info str_info;
1.81 schwarze 1062: int fd[2];
1.75 schwarze 1063: struct mpage *mpage, *mpage_dest;
1064: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1065: struct mdoc *mdoc;
1066: struct man *man;
1.75 schwarze 1067: char *sodest;
1.69 schwarze 1068: char *cp;
1.81 schwarze 1069: pid_t child_pid;
1.88 schwarze 1070: int status;
1.58 schwarze 1071: unsigned int pslot;
1.47 schwarze 1072: enum mandoclevel lvl;
1073:
1074: str_info.alloc = hash_alloc;
1075: str_info.halloc = hash_halloc;
1076: str_info.hfree = hash_free;
1077: str_info.key_offset = offsetof(struct str, key);
1078:
1.64 schwarze 1079: if (0 == nodb)
1080: SQL_EXEC("BEGIN TRANSACTION");
1081:
1.47 schwarze 1082: mpage = ohash_first(&mpages, &pslot);
1083: while (NULL != mpage) {
1084: mlinks_undupe(mpage);
1085: if (NULL == mpage->mlinks) {
1086: mpage = ohash_next(&mpages, &pslot);
1087: continue;
1088: }
1.1 schwarze 1089:
1.90 schwarze 1090: name_mask = NAME_MASK;
1091: ohash_init(&names, 4, &str_info);
1.47 schwarze 1092: ohash_init(&strings, 6, &str_info);
1093: mparse_reset(mp);
1094: mdoc = NULL;
1095: man = NULL;
1.81 schwarze 1096: sodest = NULL;
1097: child_pid = 0;
1098: fd[0] = -1;
1099: fd[1] = -1;
1100:
1101: if (mpage->mlinks->gzip) {
1102: if (-1 == pipe(fd)) {
1103: exitcode = (int)MANDOCLEVEL_SYSERR;
1104: say(mpage->mlinks->file, "&pipe gunzip");
1105: goto nextpage;
1106: }
1107: switch (child_pid = fork()) {
1108: case (-1):
1109: exitcode = (int)MANDOCLEVEL_SYSERR;
1110: say(mpage->mlinks->file, "&fork gunzip");
1111: child_pid = 0;
1112: close(fd[1]);
1113: close(fd[0]);
1114: goto nextpage;
1115: case (0):
1116: close(fd[0]);
1117: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1118: say(mpage->mlinks->file,
1119: "&dup gunzip");
1120: exit(1);
1121: }
1122: execlp("gunzip", "gunzip", "-c",
1123: mpage->mlinks->file, NULL);
1124: say(mpage->mlinks->file, "&exec gunzip");
1125: exit(1);
1126: default:
1127: close(fd[1]);
1128: break;
1129: }
1130: }
1.11 schwarze 1131:
1132: /*
1.24 schwarze 1133: * Try interpreting the file as mdoc(7) or man(7)
1134: * source code, unless it is already known to be
1135: * formatted. Fall back to formatted mode.
1.11 schwarze 1136: */
1.47 schwarze 1137: if (FORM_CAT != mpage->mlinks->dform ||
1138: FORM_CAT != mpage->mlinks->fform) {
1.81 schwarze 1139: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.47 schwarze 1140: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1141: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1142: }
1.11 schwarze 1143:
1.75 schwarze 1144: if (NULL != sodest) {
1145: mlink_dest = ohash_find(&mlinks,
1146: ohash_qlookup(&mlinks, sodest));
1147: if (NULL != mlink_dest) {
1148:
1149: /* The .so target exists. */
1150:
1151: mpage_dest = mlink_dest->mpage;
1152: mlink = mpage->mlinks;
1153: while (1) {
1154: mlink->mpage = mpage_dest;
1155:
1156: /*
1157: * If the target was already
1158: * processed, add the links
1159: * to the database now.
1160: * Otherwise, this will
1161: * happen when we come
1162: * to the target.
1163: */
1164:
1.94 schwarze 1165: if (mpage_dest->pageid)
1.75 schwarze 1166: dbadd_mlink(mlink);
1167:
1168: if (NULL == mlink->next)
1169: break;
1170: mlink = mlink->next;
1171: }
1172:
1173: /* Move all links to the target. */
1174:
1175: mlink->next = mlink_dest->next;
1176: mlink_dest->next = mpage->mlinks;
1177: mpage->mlinks = NULL;
1178: }
1.81 schwarze 1179: goto nextpage;
1.75 schwarze 1180: } else if (NULL != mdoc) {
1.47 schwarze 1181: mpage->form = FORM_SRC;
1182: mpage->sec =
1183: mandoc_strdup(mdoc_meta(mdoc)->msec);
1184: mpage->arch = mdoc_meta(mdoc)->arch;
1185: mpage->arch = mandoc_strdup(
1186: NULL == mpage->arch ? "" : mpage->arch);
1187: mpage->title =
1188: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1189: } else if (NULL != man) {
1.47 schwarze 1190: mpage->form = FORM_SRC;
1191: mpage->sec =
1192: mandoc_strdup(man_meta(man)->msec);
1193: mpage->arch =
1194: mandoc_strdup(mpage->mlinks->arch);
1195: mpage->title =
1196: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1197: } else {
1.47 schwarze 1198: mpage->form = FORM_CAT;
1199: mpage->sec =
1200: mandoc_strdup(mpage->mlinks->dsec);
1201: mpage->arch =
1202: mandoc_strdup(mpage->mlinks->arch);
1203: mpage->title =
1204: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1205: }
1.54 schwarze 1206: putkey(mpage, mpage->sec, TYPE_sec);
1.55 schwarze 1207: putkey(mpage, '\0' == *mpage->arch ?
1.70 schwarze 1208: any : mpage->arch, TYPE_arch);
1.1 schwarze 1209:
1.54 schwarze 1210: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1211: if ('\0' != *mlink->dsec)
1212: putkey(mpage, mlink->dsec, TYPE_sec);
1213: if ('\0' != *mlink->fsec)
1214: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1215: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1216: any : mlink->arch, TYPE_arch);
1.90 schwarze 1217: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1218: }
1.41 deraadt 1219:
1.92 schwarze 1220: assert(NULL == mpage->desc);
1.47 schwarze 1221: if (NULL != mdoc) {
1222: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.90 schwarze 1223: putkey(mpage, cp, NAME_HEAD);
1.47 schwarze 1224: parse_mdoc(mpage, mdoc_node(mdoc));
1225: } else if (NULL != man)
1226: parse_man(mpage, man_node(man));
1227: else
1.81 schwarze 1228: parse_cat(mpage, fd[0]);
1.92 schwarze 1229: if (NULL == mpage->desc)
1230: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1231:
1232: if (warnings && !use_all)
1233: for (mlink = mpage->mlinks; mlink;
1234: mlink = mlink->next)
1235: mlink_check(mpage, mlink);
1.6 schwarze 1236:
1.62 schwarze 1237: dbadd(mpage, mc);
1.81 schwarze 1238:
1239: nextpage:
1240: if (child_pid) {
1241: if (-1 == waitpid(child_pid, &status, 0)) {
1242: exitcode = (int)MANDOCLEVEL_SYSERR;
1243: say(mpage->mlinks->file, "&wait gunzip");
1244: } else if (WIFSIGNALED(status)) {
1245: exitcode = (int)MANDOCLEVEL_SYSERR;
1246: say(mpage->mlinks->file,
1247: "gunzip died from signal %d",
1248: WTERMSIG(status));
1249: } else if (WEXITSTATUS(status)) {
1250: exitcode = (int)MANDOCLEVEL_SYSERR;
1251: say(mpage->mlinks->file,
1252: "gunzip failed with code %d",
1253: WEXITSTATUS(status));
1254: }
1255: }
1.47 schwarze 1256: ohash_delete(&strings);
1.90 schwarze 1257: ohash_delete(&names);
1.47 schwarze 1258: mpage = ohash_next(&mpages, &pslot);
1259: }
1.64 schwarze 1260:
1261: if (0 == nodb)
1262: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1263: }
1264:
1265: static void
1266: names_check(void)
1267: {
1268: sqlite3_stmt *stmt;
1269: const char *name, *sec, *arch, *key;
1270: int irc;
1271:
1272: sqlite3_prepare_v2(db,
1273: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1274: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1275: "WHERE bits & ? AND NOT EXISTS ("
1276: "SELECT pageid FROM mlinks "
1.90 schwarze 1277: "WHERE mlinks.pageid == names.pageid "
1278: "AND mlinks.name == names.name"
1.87 schwarze 1279: ")"
1280: ") JOIN ("
1281: "SELECT * FROM mlinks GROUP BY pageid"
1282: ") USING (pageid);",
1283: -1, &stmt, NULL);
1284:
1.91 schwarze 1285: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1286: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1287:
1288: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1289: name = sqlite3_column_text(stmt, 0);
1290: sec = sqlite3_column_text(stmt, 1);
1291: arch = sqlite3_column_text(stmt, 2);
1292: key = sqlite3_column_text(stmt, 3);
1293: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1294: '\0' == *arch ? "" : "/",
1295: '\0' == *arch ? "" : arch, key);
1296: }
1297: sqlite3_finalize(stmt);
1.47 schwarze 1298: }
1.6 schwarze 1299:
1.47 schwarze 1300: static void
1.81 schwarze 1301: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1302: {
1303: FILE *stream;
1304: char *line, *p, *title;
1305: size_t len, plen, titlesz;
1.1 schwarze 1306:
1.81 schwarze 1307: stream = (-1 == fd) ?
1308: fopen(mpage->mlinks->file, "r") :
1309: fdopen(fd, "r");
1310: if (NULL == stream) {
1.47 schwarze 1311: if (warnings)
1.80 schwarze 1312: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1313: return;
1314: }
1.1 schwarze 1315:
1.47 schwarze 1316: /* Skip to first blank line. */
1.1 schwarze 1317:
1.47 schwarze 1318: while (NULL != (line = fgetln(stream, &len)))
1319: if ('\n' == *line)
1320: break;
1.1 schwarze 1321:
1.47 schwarze 1322: /*
1323: * Assume the first line that is not indented
1324: * is the first section header. Skip to it.
1325: */
1.1 schwarze 1326:
1.47 schwarze 1327: while (NULL != (line = fgetln(stream, &len)))
1328: if ('\n' != *line && ' ' != *line)
1329: break;
1330:
1331: /*
1332: * Read up until the next section into a buffer.
1333: * Strip the leading and trailing newline from each read line,
1334: * appending a trailing space.
1335: * Ignore empty (whitespace-only) lines.
1336: */
1.28 schwarze 1337:
1.47 schwarze 1338: titlesz = 0;
1339: title = NULL;
1.38 schwarze 1340:
1.47 schwarze 1341: while (NULL != (line = fgetln(stream, &len))) {
1342: if (' ' != *line || '\n' != line[len - 1])
1343: break;
1344: while (len > 0 && isspace((unsigned char)*line)) {
1345: line++;
1346: len--;
1347: }
1348: if (1 == len)
1349: continue;
1350: title = mandoc_realloc(title, titlesz + len);
1351: memcpy(title + titlesz, line, len);
1352: titlesz += len;
1353: title[titlesz - 1] = ' ';
1354: }
1.28 schwarze 1355:
1.47 schwarze 1356: /*
1357: * If no page content can be found, or the input line
1358: * is already the next section header, or there is no
1359: * trailing newline, reuse the page title as the page
1360: * description.
1361: */
1.1 schwarze 1362:
1.47 schwarze 1363: if (NULL == title || '\0' == *title) {
1364: if (warnings)
1365: say(mpage->mlinks->file,
1366: "Cannot find NAME section");
1367: fclose(stream);
1368: free(title);
1369: return;
1370: }
1.24 schwarze 1371:
1.47 schwarze 1372: title = mandoc_realloc(title, titlesz + 1);
1373: title[titlesz] = '\0';
1.24 schwarze 1374:
1.47 schwarze 1375: /*
1376: * Skip to the first dash.
1377: * Use the remaining line as the description (no more than 70
1378: * bytes).
1379: */
1.28 schwarze 1380:
1.47 schwarze 1381: if (NULL != (p = strstr(title, "- "))) {
1382: for (p += 2; ' ' == *p || '\b' == *p; p++)
1383: /* Skip to next word. */ ;
1384: } else {
1385: if (warnings)
1386: say(mpage->mlinks->file,
1387: "No dash in title line");
1388: p = title;
1389: }
1.1 schwarze 1390:
1.47 schwarze 1391: plen = strlen(p);
1.1 schwarze 1392:
1.47 schwarze 1393: /* Strip backspace-encoding from line. */
1.1 schwarze 1394:
1.47 schwarze 1395: while (NULL != (line = memchr(p, '\b', plen))) {
1396: len = line - p;
1397: if (0 == len) {
1398: memmove(line, line + 1, plen--);
1399: continue;
1400: }
1401: memmove(line - 1, line + 1, plen - len);
1402: plen -= 2;
1403: }
1.1 schwarze 1404:
1.47 schwarze 1405: mpage->desc = mandoc_strdup(p);
1406: fclose(stream);
1407: free(title);
1408: }
1.16 schwarze 1409:
1.47 schwarze 1410: /*
1411: * Put a type/word pair into the word database for this particular file.
1412: */
1413: static void
1.69 schwarze 1414: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1415: {
1.69 schwarze 1416: char *cp;
1.37 schwarze 1417:
1.47 schwarze 1418: assert(NULL != value);
1.69 schwarze 1419: if (TYPE_arch == type)
1420: for (cp = value; *cp; cp++)
1421: if (isupper((unsigned char)*cp))
1422: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1423: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1424: }
1425:
1426: /*
1.47 schwarze 1427: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1428: */
1429: static void
1.47 schwarze 1430: putmdockey(const struct mpage *mpage,
1431: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1432: {
1.16 schwarze 1433:
1.47 schwarze 1434: for ( ; NULL != n; n = n->next) {
1435: if (NULL != n->child)
1436: putmdockey(mpage, n->child, m);
1437: if (MDOC_TEXT == n->type)
1438: putkey(mpage, n->string, m);
1439: }
1440: }
1.16 schwarze 1441:
1.47 schwarze 1442: static void
1443: parse_man(struct mpage *mpage, const struct man_node *n)
1444: {
1445: const struct man_node *head, *body;
1.78 schwarze 1446: char *start, *title;
1.47 schwarze 1447: char byte;
1.78 schwarze 1448: size_t sz;
1.16 schwarze 1449:
1.47 schwarze 1450: if (NULL == n)
1451: return;
1.16 schwarze 1452:
1.47 schwarze 1453: /*
1454: * We're only searching for one thing: the first text child in
1455: * the BODY of a NAME section. Since we don't keep track of
1456: * sections in -man, run some hoops to find out whether we're in
1457: * the correct section or not.
1458: */
1.16 schwarze 1459:
1.47 schwarze 1460: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1461: body = n;
1462: assert(body->parent);
1463: if (NULL != (head = body->parent->head) &&
1464: 1 == head->nchild &&
1465: NULL != (head = (head->child)) &&
1466: MAN_TEXT == head->type &&
1467: 0 == strcmp(head->string, "NAME") &&
1.78 schwarze 1468: NULL != body->child) {
1.2 schwarze 1469:
1.47 schwarze 1470: /*
1471: * Suck the entire NAME section into memory.
1472: * Yes, we might run away.
1473: * But too many manuals have big, spread-out
1474: * NAME sections over many lines.
1475: */
1.2 schwarze 1476:
1.78 schwarze 1477: title = NULL;
1478: man_deroff(&title, body);
1.47 schwarze 1479: if (NULL == title)
1480: return;
1.16 schwarze 1481:
1.47 schwarze 1482: /*
1483: * Go through a special heuristic dance here.
1484: * Conventionally, one or more manual names are
1485: * comma-specified prior to a whitespace, then a
1486: * dash, then a description. Try to puzzle out
1487: * the name parts here.
1488: */
1.16 schwarze 1489:
1.78 schwarze 1490: start = title;
1.47 schwarze 1491: for ( ;; ) {
1492: sz = strcspn(start, " ,");
1493: if ('\0' == start[sz])
1494: break;
1.1 schwarze 1495:
1.47 schwarze 1496: byte = start[sz];
1497: start[sz] = '\0';
1.67 schwarze 1498:
1499: /*
1500: * Assume a stray trailing comma in the
1501: * name list if a name begins with a dash.
1502: */
1503:
1504: if ('-' == start[0] ||
1505: ('\\' == start[0] && '-' == start[1]))
1506: break;
1.1 schwarze 1507:
1.90 schwarze 1508: putkey(mpage, start, NAME_TITLE);
1.1 schwarze 1509:
1.47 schwarze 1510: if (' ' == byte) {
1511: start += sz + 1;
1512: break;
1513: }
1.1 schwarze 1514:
1.47 schwarze 1515: assert(',' == byte);
1516: start += sz + 1;
1517: while (' ' == *start)
1518: start++;
1519: }
1.1 schwarze 1520:
1.78 schwarze 1521: if (start == title) {
1.90 schwarze 1522: putkey(mpage, start, NAME_TITLE);
1.47 schwarze 1523: free(title);
1524: return;
1525: }
1.1 schwarze 1526:
1.47 schwarze 1527: while (isspace((unsigned char)*start))
1528: start++;
1.1 schwarze 1529:
1.47 schwarze 1530: if (0 == strncmp(start, "-", 1))
1531: start += 1;
1532: else if (0 == strncmp(start, "\\-\\-", 4))
1533: start += 4;
1534: else if (0 == strncmp(start, "\\-", 2))
1535: start += 2;
1536: else if (0 == strncmp(start, "\\(en", 4))
1537: start += 4;
1538: else if (0 == strncmp(start, "\\(em", 4))
1539: start += 4;
1.1 schwarze 1540:
1.47 schwarze 1541: while (' ' == *start)
1542: start++;
1.1 schwarze 1543:
1.47 schwarze 1544: mpage->desc = mandoc_strdup(start);
1545: free(title);
1546: return;
1547: }
1548: }
1.1 schwarze 1549:
1.47 schwarze 1550: for (n = n->child; n; n = n->next) {
1551: if (NULL != mpage->desc)
1552: break;
1553: parse_man(mpage, n);
1.1 schwarze 1554: }
1555: }
1556:
1557: static void
1.47 schwarze 1558: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1559: {
1560:
1.47 schwarze 1561: assert(NULL != n);
1562: for (n = n->child; NULL != n; n = n->next) {
1563: switch (n->type) {
1564: case (MDOC_ELEM):
1565: /* FALLTHROUGH */
1566: case (MDOC_BLOCK):
1567: /* FALLTHROUGH */
1568: case (MDOC_HEAD):
1569: /* FALLTHROUGH */
1570: case (MDOC_BODY):
1571: /* FALLTHROUGH */
1572: case (MDOC_TAIL):
1573: if (NULL != mdocs[n->tok].fp)
1574: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1575: break;
1576: if (mdocs[n->tok].mask)
1577: putmdockey(mpage, n->child,
1578: mdocs[n->tok].mask);
1579: break;
1580: default:
1581: assert(MDOC_ROOT != n->type);
1582: continue;
1583: }
1584: if (NULL != n->child)
1585: parse_mdoc(mpage, n);
1.1 schwarze 1586: }
1587: }
1588:
1.19 schwarze 1589: static int
1.47 schwarze 1590: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1591: {
1592: const char *start, *end;
1593: size_t sz;
1.19 schwarze 1594:
1.47 schwarze 1595: if (SEC_SYNOPSIS != n->sec ||
1596: NULL == (n = n->child) ||
1597: MDOC_TEXT != n->type)
1.19 schwarze 1598: return(0);
1.1 schwarze 1599:
1600: /*
1601: * Only consider those `Fd' macro fields that begin with an
1602: * "inclusion" token (versus, e.g., #define).
1603: */
1.47 schwarze 1604:
1.1 schwarze 1605: if (strcmp("#include", n->string))
1.19 schwarze 1606: return(0);
1.1 schwarze 1607:
1608: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1609: return(0);
1.1 schwarze 1610:
1611: /*
1612: * Strip away the enclosing angle brackets and make sure we're
1613: * not zero-length.
1614: */
1615:
1616: start = n->string;
1617: if ('<' == *start || '"' == *start)
1618: start++;
1619:
1620: if (0 == (sz = strlen(start)))
1.19 schwarze 1621: return(0);
1.1 schwarze 1622:
1623: end = &start[(int)sz - 1];
1624: if ('>' == *end || '"' == *end)
1625: end--;
1626:
1.47 schwarze 1627: if (end > start)
1628: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1629: return(0);
1.1 schwarze 1630: }
1631:
1.19 schwarze 1632: static int
1.47 schwarze 1633: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1634: {
1.69 schwarze 1635: char *cp;
1.1 schwarze 1636:
1.47 schwarze 1637: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1638: return(0);
1639:
1.47 schwarze 1640: /*
1641: * Parse: .Fn "struct type *name" "char *arg".
1642: * First strip away pointer symbol.
1643: * Then store the function name, then type.
1644: * Finally, store the arguments.
1645: */
1.1 schwarze 1646:
1.47 schwarze 1647: if (NULL == (cp = strrchr(n->string, ' ')))
1648: cp = n->string;
1.1 schwarze 1649:
1650: while ('*' == *cp)
1651: cp++;
1652:
1.47 schwarze 1653: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1654:
1.47 schwarze 1655: if (n->string < cp)
1656: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1657:
1.47 schwarze 1658: for (n = n->next; NULL != n; n = n->next)
1659: if (MDOC_TEXT == n->type)
1660: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1661:
1662: return(0);
1.1 schwarze 1663: }
1664:
1.19 schwarze 1665: static int
1.47 schwarze 1666: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1667: {
1.47 schwarze 1668: char *cp;
1.1 schwarze 1669:
1670: if (NULL == (n = n->child))
1.19 schwarze 1671: return(0);
1.1 schwarze 1672:
1.47 schwarze 1673: if (NULL == n->next) {
1674: putkey(mpage, n->string, TYPE_Xr);
1675: return(0);
1676: }
1.1 schwarze 1677:
1.77 schwarze 1678: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1679: putkey(mpage, cp, TYPE_Xr);
1680: free(cp);
1681: return(0);
1.1 schwarze 1682: }
1683:
1.19 schwarze 1684: static int
1.47 schwarze 1685: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1686: {
1687:
1.79 schwarze 1688: if (MDOC_BODY == n->type)
1689: mdoc_deroff(&mpage->desc, n);
1690: return(0);
1.1 schwarze 1691: }
1692:
1.19 schwarze 1693: static int
1.47 schwarze 1694: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1695: {
1696:
1.86 schwarze 1697: if (SEC_NAME == n->sec)
1.90 schwarze 1698: putmdockey(mpage, n->child, NAME_TITLE);
1.86 schwarze 1699: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.90 schwarze 1700: putmdockey(mpage, n->child, NAME_SYN);
1.86 schwarze 1701: return(0);
1.1 schwarze 1702: }
1703:
1.19 schwarze 1704: static int
1.47 schwarze 1705: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1706: {
1707:
1.19 schwarze 1708: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1709: }
1710:
1.47 schwarze 1711: static int
1712: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1713: {
1714:
1.47 schwarze 1715: return(MDOC_HEAD == n->type);
1716: }
1.1 schwarze 1717:
1.47 schwarze 1718: static int
1719: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1720: {
1.1 schwarze 1721:
1.47 schwarze 1722: return(MDOC_BODY == n->type);
1.1 schwarze 1723: }
1724:
1.47 schwarze 1725: /*
1726: * Add a string to the hash table for the current manual.
1727: * Each string has a bitmask telling which macros it belongs to.
1728: * When we finish the manual, we'll dump the table.
1729: */
1.1 schwarze 1730: static void
1.47 schwarze 1731: putkeys(const struct mpage *mpage,
1732: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1733: {
1.90 schwarze 1734: struct ohash *htab;
1.47 schwarze 1735: struct str *s;
1.68 schwarze 1736: const char *end;
1.47 schwarze 1737: unsigned int slot;
1.68 schwarze 1738: int i;
1.1 schwarze 1739:
1.47 schwarze 1740: if (0 == sz)
1741: return;
1.68 schwarze 1742:
1.90 schwarze 1743: if (TYPE_Nm & v) {
1744: htab = &names;
1745: v &= name_mask;
1746: name_mask &= ~NAME_FIRST;
1747: if (debug > 1)
1748: say(mpage->mlinks->file,
1749: "Adding name %*s", sz, cp);
1750: } else {
1751: htab = &strings;
1752: if (debug > 1)
1753: for (i = 0; i < mansearch_keymax; i++)
1754: if (1 << i & v)
1755: say(mpage->mlinks->file,
1756: "Adding key %s=%*s",
1757: mansearch_keynames[i], sz, cp);
1.68 schwarze 1758: }
1.47 schwarze 1759:
1760: end = cp + sz;
1.90 schwarze 1761: slot = ohash_qlookupi(htab, cp, &end);
1762: s = ohash_find(htab, slot);
1.1 schwarze 1763:
1.47 schwarze 1764: if (NULL != s && mpage == s->mpage) {
1765: s->mask |= v;
1.1 schwarze 1766: return;
1.47 schwarze 1767: } else if (NULL == s) {
1768: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1769: memcpy(s->key, cp, sz);
1.90 schwarze 1770: ohash_insert(htab, slot, s);
1.47 schwarze 1771: }
1772: s->mpage = mpage;
1773: s->mask = v;
1.1 schwarze 1774: }
1775:
1776: /*
1.47 schwarze 1777: * Take a Unicode codepoint and produce its UTF-8 encoding.
1778: * This isn't the best way to do this, but it works.
1779: * The magic numbers are from the UTF-8 packaging.
1780: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1781: */
1.47 schwarze 1782: static size_t
1783: utf8(unsigned int cp, char out[7])
1.1 schwarze 1784: {
1.47 schwarze 1785: size_t rc;
1.1 schwarze 1786:
1.47 schwarze 1787: rc = 0;
1788: if (cp <= 0x0000007F) {
1789: rc = 1;
1790: out[0] = (char)cp;
1791: } else if (cp <= 0x000007FF) {
1792: rc = 2;
1793: out[0] = (cp >> 6 & 31) | 192;
1794: out[1] = (cp & 63) | 128;
1795: } else if (cp <= 0x0000FFFF) {
1796: rc = 3;
1797: out[0] = (cp >> 12 & 15) | 224;
1798: out[1] = (cp >> 6 & 63) | 128;
1799: out[2] = (cp & 63) | 128;
1800: } else if (cp <= 0x001FFFFF) {
1801: rc = 4;
1802: out[0] = (cp >> 18 & 7) | 240;
1803: out[1] = (cp >> 12 & 63) | 128;
1804: out[2] = (cp >> 6 & 63) | 128;
1805: out[3] = (cp & 63) | 128;
1806: } else if (cp <= 0x03FFFFFF) {
1807: rc = 5;
1808: out[0] = (cp >> 24 & 3) | 248;
1809: out[1] = (cp >> 18 & 63) | 128;
1810: out[2] = (cp >> 12 & 63) | 128;
1811: out[3] = (cp >> 6 & 63) | 128;
1812: out[4] = (cp & 63) | 128;
1813: } else if (cp <= 0x7FFFFFFF) {
1814: rc = 6;
1815: out[0] = (cp >> 30 & 1) | 252;
1816: out[1] = (cp >> 24 & 63) | 128;
1817: out[2] = (cp >> 18 & 63) | 128;
1818: out[3] = (cp >> 12 & 63) | 128;
1819: out[4] = (cp >> 6 & 63) | 128;
1820: out[5] = (cp & 63) | 128;
1821: } else
1822: return(0);
1.19 schwarze 1823:
1.47 schwarze 1824: out[rc] = '\0';
1825: return(rc);
1.1 schwarze 1826: }
1827:
1.47 schwarze 1828: /*
1.53 schwarze 1829: * Store the rendered version of a key, or alias the pointer
1830: * if the key contains no escape sequences.
1.47 schwarze 1831: */
1832: static void
1.53 schwarze 1833: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1834: {
1.47 schwarze 1835: size_t sz, bsz, pos;
1.71 schwarze 1836: char utfbuf[7], res[6];
1.47 schwarze 1837: char *buf;
1838: const char *seq, *cpp, *val;
1839: int len, u;
1840: enum mandoc_esc esc;
1841:
1.53 schwarze 1842: assert(NULL == key->rendered);
1.47 schwarze 1843:
1844: res[0] = '\\';
1845: res[1] = '\t';
1846: res[2] = ASCII_NBRSP;
1847: res[3] = ASCII_HYPH;
1.71 schwarze 1848: res[4] = ASCII_BREAK;
1849: res[5] = '\0';
1.1 schwarze 1850:
1.47 schwarze 1851: val = key->key;
1852: bsz = strlen(val);
1.1 schwarze 1853:
1854: /*
1.47 schwarze 1855: * Pre-check: if we have no stop-characters, then set the
1856: * pointer as ourselvse and get out of here.
1.1 schwarze 1857: */
1.47 schwarze 1858: if (strcspn(val, res) == bsz) {
1.53 schwarze 1859: key->rendered = key->key;
1.47 schwarze 1860: return;
1861: }
1.1 schwarze 1862:
1.47 schwarze 1863: /* Pre-allocate by the length of the input */
1.39 schwarze 1864:
1.47 schwarze 1865: buf = mandoc_malloc(++bsz);
1866: pos = 0;
1.39 schwarze 1867:
1.47 schwarze 1868: while ('\0' != *val) {
1869: /*
1870: * Halt on the first escape sequence.
1871: * This also halts on the end of string, in which case
1872: * we just copy, fallthrough, and exit the loop.
1873: */
1874: if ((sz = strcspn(val, res)) > 0) {
1875: memcpy(&buf[pos], val, sz);
1876: pos += sz;
1877: val += sz;
1878: }
1.39 schwarze 1879:
1.71 schwarze 1880: switch (*val) {
1881: case (ASCII_HYPH):
1.47 schwarze 1882: buf[pos++] = '-';
1883: val++;
1884: continue;
1.71 schwarze 1885: case ('\t'):
1886: /* FALLTHROUGH */
1887: case (ASCII_NBRSP):
1.47 schwarze 1888: buf[pos++] = ' ';
1889: val++;
1.71 schwarze 1890: /* FALLTHROUGH */
1891: case (ASCII_BREAK):
1.47 schwarze 1892: continue;
1.71 schwarze 1893: default:
1894: break;
1895: }
1896: if ('\\' != *val)
1.47 schwarze 1897: break;
1.39 schwarze 1898:
1.47 schwarze 1899: /* Read past the slash. */
1.39 schwarze 1900:
1.47 schwarze 1901: val++;
1.39 schwarze 1902:
1.47 schwarze 1903: /*
1904: * Parse the escape sequence and see if it's a
1905: * predefined character or special character.
1906: */
1.52 schwarze 1907:
1.47 schwarze 1908: esc = mandoc_escape
1909: ((const char **)&val, &seq, &len);
1910: if (ESCAPE_ERROR == esc)
1911: break;
1912: if (ESCAPE_SPECIAL != esc)
1913: continue;
1.39 schwarze 1914:
1.47 schwarze 1915: /*
1.52 schwarze 1916: * Render the special character
1917: * as either UTF-8 or ASCII.
1.47 schwarze 1918: */
1.52 schwarze 1919:
1920: if (write_utf8) {
1921: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1922: continue;
1923: cpp = utfbuf;
1924: if (0 == (sz = utf8(u, utfbuf)))
1925: continue;
1926: sz = strlen(cpp);
1927: } else {
1928: cpp = mchars_spec2str(mc, seq, len, &sz);
1929: if (NULL == cpp)
1930: continue;
1931: if (ASCII_NBRSP == *cpp) {
1932: cpp = " ";
1933: sz = 1;
1934: }
1935: }
1.1 schwarze 1936:
1.47 schwarze 1937: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1938:
1.47 schwarze 1939: bsz += sz;
1940: buf = mandoc_realloc(buf, bsz);
1941: memcpy(&buf[pos], cpp, sz);
1942: pos += sz;
1.1 schwarze 1943: }
1944:
1.47 schwarze 1945: buf[pos] = '\0';
1.53 schwarze 1946: key->rendered = buf;
1.1 schwarze 1947: }
1948:
1.75 schwarze 1949: static void
1950: dbadd_mlink(const struct mlink *mlink)
1951: {
1952: size_t i;
1953:
1954: i = 1;
1955: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1956: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1957: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1958: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1959: SQL_STEP(stmts[STMT_INSERT_LINK]);
1960: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1961: }
1962:
1.11 schwarze 1963: /*
1.47 schwarze 1964: * Flush the current page's terms (and their bits) into the database.
1965: * Wrap the entire set of additions in a transaction to make sqlite be a
1966: * little faster.
1.53 schwarze 1967: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1968: */
1969: static void
1.75 schwarze 1970: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1971: {
1.47 schwarze 1972: struct mlink *mlink;
1973: struct str *key;
1974: size_t i;
1975: unsigned int slot;
1976:
1.85 schwarze 1977: mlink = mpage->mlinks;
1.11 schwarze 1978:
1.85 schwarze 1979: if (nodb) {
1980: while (NULL != mlink) {
1981: fputs(mlink->name, stdout);
1982: if (NULL == mlink->next ||
1983: strcmp(mlink->dsec, mlink->next->dsec) ||
1984: strcmp(mlink->fsec, mlink->next->fsec) ||
1985: strcmp(mlink->arch, mlink->next->arch)) {
1986: putchar('(');
1987: if ('\0' == *mlink->dsec)
1988: fputs(mlink->fsec, stdout);
1989: else
1990: fputs(mlink->dsec, stdout);
1991: if ('\0' != *mlink->arch)
1992: printf("/%s", mlink->arch);
1993: putchar(')');
1994: }
1995: mlink = mlink->next;
1996: if (NULL != mlink)
1997: fputs(", ", stdout);
1998: }
1.89 schwarze 1999: printf(" - %s\n", mpage->desc);
1.11 schwarze 2000: return;
1.85 schwarze 2001: }
2002:
2003: if (debug)
2004: say(mlink->file, "Adding to database");
1.47 schwarze 2005:
2006: i = 1;
1.89 schwarze 2007: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc);
1.47 schwarze 2008: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
2009: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2010: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2011: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2012:
1.85 schwarze 2013: while (NULL != mlink) {
1.75 schwarze 2014: dbadd_mlink(mlink);
1.85 schwarze 2015: mlink = mlink->next;
2016: }
1.91 schwarze 2017: mlink = mpage->mlinks;
1.47 schwarze 2018:
1.90 schwarze 2019: for (key = ohash_first(&names, &slot); NULL != key;
2020: key = ohash_next(&names, &slot)) {
2021: assert(key->mpage == mpage);
2022: if (NULL == key->rendered)
2023: render_key(mc, key);
2024: i = 1;
2025: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2026: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2027: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2028: SQL_STEP(stmts[STMT_INSERT_NAME]);
2029: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2030: if (key->rendered != key->key)
2031: free(key->rendered);
2032: free(key);
2033: }
1.47 schwarze 2034: for (key = ohash_first(&strings, &slot); NULL != key;
2035: key = ohash_next(&strings, &slot)) {
2036: assert(key->mpage == mpage);
1.53 schwarze 2037: if (NULL == key->rendered)
2038: render_key(mc, key);
1.47 schwarze 2039: i = 1;
2040: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2041: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2042: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2043: SQL_STEP(stmts[STMT_INSERT_KEY]);
2044: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2045: if (key->rendered != key->key)
2046: free(key->rendered);
1.47 schwarze 2047: free(key);
1.33 schwarze 2048: }
1.47 schwarze 2049: }
1.41 deraadt 2050:
1.47 schwarze 2051: static void
2052: dbprune(void)
2053: {
2054: struct mpage *mpage;
2055: struct mlink *mlink;
2056: size_t i;
2057: unsigned int slot;
1.11 schwarze 2058:
1.63 schwarze 2059: if (0 == nodb)
2060: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2061:
1.63 schwarze 2062: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2063: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2064: mlink = mpage->mlinks;
1.82 schwarze 2065: if (debug)
1.63 schwarze 2066: say(mlink->file, "Deleting from database");
2067: if (nodb)
2068: continue;
2069: for ( ; NULL != mlink; mlink = mlink->next) {
2070: i = 1;
2071: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2072: i, mlink->dsec);
2073: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2074: i, mlink->arch);
2075: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2076: i, mlink->name);
2077: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2078: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2079: }
1.11 schwarze 2080: }
1.63 schwarze 2081:
2082: if (0 == nodb)
2083: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2084: }
1.22 schwarze 2085:
1.47 schwarze 2086: /*
2087: * Close an existing database and its prepared statements.
2088: * If "real" is not set, rename the temporary file into the real one.
2089: */
2090: static void
2091: dbclose(int real)
2092: {
2093: size_t i;
1.72 schwarze 2094: int status;
2095: pid_t child;
1.11 schwarze 2096:
1.47 schwarze 2097: if (nodb)
2098: return;
1.11 schwarze 2099:
1.47 schwarze 2100: for (i = 0; i < STMT__MAX; i++) {
2101: sqlite3_finalize(stmts[i]);
2102: stmts[i] = NULL;
1.28 schwarze 2103: }
1.22 schwarze 2104:
1.47 schwarze 2105: sqlite3_close(db);
2106: db = NULL;
1.11 schwarze 2107:
1.47 schwarze 2108: if (real)
2109: return;
1.22 schwarze 2110:
1.72 schwarze 2111: if ('\0' == *tempfilename) {
2112: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2113: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2114: say(MANDOC_DB, "&rename");
1.72 schwarze 2115: }
2116: return;
2117: }
2118:
2119: switch (child = fork()) {
2120: case (-1):
2121: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2122: say("", "&fork cmp");
1.72 schwarze 2123: return;
2124: case (0):
2125: execlp("cmp", "cmp", "-s",
2126: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2127: say("", "&exec cmp");
1.72 schwarze 2128: exit(0);
2129: default:
2130: break;
2131: }
2132: if (-1 == waitpid(child, &status, 0)) {
2133: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2134: say("", "&wait cmp");
1.72 schwarze 2135: } else if (WIFSIGNALED(status)) {
2136: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2137: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2138: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2139: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2140: say(MANDOC_DB,
2141: "Data changed, but cannot replace database");
2142: }
2143:
2144: *strrchr(tempfilename, '/') = '\0';
2145: switch (child = fork()) {
2146: case (-1):
2147: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2148: say("", "&fork rm");
1.72 schwarze 2149: return;
2150: case (0):
2151: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2152: say("", "&exec rm");
1.72 schwarze 2153: exit((int)MANDOCLEVEL_SYSERR);
2154: default:
2155: break;
2156: }
2157: if (-1 == waitpid(child, &status, 0)) {
2158: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2159: say("", "&wait rm");
1.72 schwarze 2160: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2161: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2162: say("", "%s: Cannot remove temporary directory",
2163: tempfilename);
1.22 schwarze 2164: }
1.11 schwarze 2165: }
2166:
1.47 schwarze 2167: /*
2168: * This is straightforward stuff.
2169: * Open a database connection to a "temporary" database, then open a set
2170: * of prepared statements we'll use over and over again.
2171: * If "real" is set, we use the existing database; if not, we truncate a
2172: * temporary one.
2173: * Must be matched by dbclose().
2174: */
2175: static int
2176: dbopen(int real)
1.2 schwarze 2177: {
1.72 schwarze 2178: const char *sql;
1.47 schwarze 2179: int rc, ofl;
1.6 schwarze 2180:
1.47 schwarze 2181: if (nodb)
2182: return(1);
1.6 schwarze 2183:
1.72 schwarze 2184: *tempfilename = '\0';
1.47 schwarze 2185: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2186:
2187: if (real) {
2188: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2189: if (SQLITE_OK != rc) {
1.47 schwarze 2190: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2191: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2192: return(0);
1.28 schwarze 2193: }
1.72 schwarze 2194: goto prepare_statements;
2195: }
2196:
2197: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2198:
1.72 schwarze 2199: remove(MANDOC_DB "~");
2200: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.47 schwarze 2201: if (SQLITE_OK == rc)
1.72 schwarze 2202: goto create_tables;
1.73 schwarze 2203: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2204: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2205: say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
1.47 schwarze 2206: return(0);
2207: }
1.6 schwarze 2208:
1.72 schwarze 2209: if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2210: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2212: say("", "/tmp/mandocdb.XXXXXX: Filename too long");
1.72 schwarze 2213: return(0);
2214: }
2215: if (NULL == mkdtemp(tempfilename)) {
2216: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2217: say("", "&%s", tempfilename);
1.72 schwarze 2218: return(0);
2219: }
2220: if (strlcat(tempfilename, "/" MANDOC_DB,
2221: sizeof(tempfilename)) >= sizeof(tempfilename)) {
2222: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2223: say("", "%s/" MANDOC_DB ": Filename too long",
2224: tempfilename);
1.72 schwarze 2225: return(0);
2226: }
2227: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2228: if (SQLITE_OK != rc) {
1.47 schwarze 2229: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2230: say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
1.47 schwarze 2231: return(0);
1.2 schwarze 2232: }
2233:
1.72 schwarze 2234: create_tables:
1.47 schwarze 2235: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2236: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2237: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2238: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2239: ");\n"
2240: "\n"
2241: "CREATE TABLE \"mlinks\" (\n"
2242: " \"sec\" TEXT NOT NULL,\n"
2243: " \"arch\" TEXT NOT NULL,\n"
2244: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2245: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2246: "ON DELETE CASCADE\n"
1.47 schwarze 2247: ");\n"
1.93 schwarze 2248: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2249: "\n"
1.90 schwarze 2250: "CREATE TABLE \"names\" (\n"
2251: " \"bits\" INTEGER NOT NULL,\n"
2252: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2253: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2254: "ON DELETE CASCADE\n"
2255: ");\n"
2256: "\n"
1.47 schwarze 2257: "CREATE TABLE \"keys\" (\n"
2258: " \"bits\" INTEGER NOT NULL,\n"
2259: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2260: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2261: "ON DELETE CASCADE\n"
1.93 schwarze 2262: ");\n"
2263: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2264:
2265: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2266: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2267: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.47 schwarze 2268: return(0);
1.2 schwarze 2269: }
2270:
1.47 schwarze 2271: prepare_statements:
2272: SQL_EXEC("PRAGMA foreign_keys = ON");
1.94 schwarze 2273: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2274: "(SELECT pageid FROM mlinks WHERE "
2275: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2276: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2277: sql = "INSERT INTO mpages "
1.89 schwarze 2278: "(desc,form) VALUES (?,?)";
1.47 schwarze 2279: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2280: sql = "INSERT INTO mlinks "
1.61 schwarze 2281: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2282: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2283: sql = "INSERT INTO names "
2284: "(bits,name,pageid) VALUES (?,?,?)";
2285: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2286: sql = "INSERT INTO keys "
2287: "(bits,key,pageid) VALUES (?,?,?)";
2288: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2289:
1.47 schwarze 2290: /*
2291: * When opening a new database, we can turn off
2292: * synchronous mode for much better performance.
2293: */
1.6 schwarze 2294:
1.47 schwarze 2295: if (real)
2296: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 2297:
1.47 schwarze 2298: return(1);
2299: }
1.6 schwarze 2300:
1.47 schwarze 2301: static void *
2302: hash_halloc(size_t sz, void *arg)
2303: {
1.6 schwarze 2304:
1.47 schwarze 2305: return(mandoc_calloc(sz, 1));
2306: }
1.2 schwarze 2307:
1.47 schwarze 2308: static void *
2309: hash_alloc(size_t sz, void *arg)
2310: {
1.28 schwarze 2311:
1.47 schwarze 2312: return(mandoc_malloc(sz));
2313: }
1.6 schwarze 2314:
1.47 schwarze 2315: static void
2316: hash_free(void *p, size_t sz, void *arg)
2317: {
1.26 schwarze 2318:
1.47 schwarze 2319: free(p);
2320: }
1.6 schwarze 2321:
1.47 schwarze 2322: static int
2323: set_basedir(const char *targetdir)
2324: {
2325: static char startdir[PATH_MAX];
2326: static int fd;
1.6 schwarze 2327:
1.47 schwarze 2328: /*
2329: * Remember where we started by keeping a fd open to the origin
2330: * path component: throughout this utility, we chdir() a lot to
2331: * handle relative paths, and by doing this, we can return to
2332: * the starting point.
2333: */
2334: if ('\0' == *startdir) {
2335: if (NULL == getcwd(startdir, PATH_MAX)) {
2336: exitcode = (int)MANDOCLEVEL_SYSERR;
2337: if (NULL != targetdir)
1.80 schwarze 2338: say("", "&getcwd");
1.47 schwarze 2339: return(0);
2340: }
2341: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2342: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2343: say("", "&open %s", startdir);
1.47 schwarze 2344: return(0);
1.11 schwarze 2345: }
1.47 schwarze 2346: if (NULL == targetdir)
2347: targetdir = startdir;
2348: } else {
2349: if (-1 == fd)
2350: return(0);
2351: if (-1 == fchdir(fd)) {
2352: close(fd);
2353: basedir[0] = '\0';
2354: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2355: say("", "&chdir %s", startdir);
1.47 schwarze 2356: return(0);
1.2 schwarze 2357: }
1.47 schwarze 2358: if (NULL == targetdir) {
2359: close(fd);
2360: return(1);
1.2 schwarze 2361: }
2362: }
1.47 schwarze 2363: if (NULL == realpath(targetdir, basedir)) {
2364: basedir[0] = '\0';
2365: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2366: say("", "&%s: realpath", targetdir);
1.47 schwarze 2367: return(0);
2368: } else if (-1 == chdir(basedir)) {
2369: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 2370: say("", "&chdir");
1.47 schwarze 2371: return(0);
2372: }
2373: return(1);
1.2 schwarze 2374: }
2375:
2376: static void
1.47 schwarze 2377: say(const char *file, const char *format, ...)
1.2 schwarze 2378: {
1.47 schwarze 2379: va_list ap;
1.80 schwarze 2380: int use_errno;
1.2 schwarze 2381:
1.47 schwarze 2382: if ('\0' != *basedir)
2383: fprintf(stderr, "%s", basedir);
2384: if ('\0' != *basedir && '\0' != *file)
2385: fputs("//", stderr);
2386: if ('\0' != *file)
2387: fprintf(stderr, "%s", file);
1.31 schwarze 2388:
1.80 schwarze 2389: use_errno = 1;
2390: if (NULL != format) {
2391: switch (*format) {
2392: case ('&'):
2393: format++;
2394: break;
2395: case ('\0'):
2396: format = NULL;
2397: break;
2398: default:
2399: use_errno = 0;
2400: break;
2401: }
2402: }
2403: if (NULL != format) {
2404: if ('\0' != *basedir || '\0' != *file)
2405: fputs(": ", stderr);
2406: va_start(ap, format);
2407: vfprintf(stderr, format, ap);
2408: va_end(ap);
2409: }
2410: if (use_errno) {
2411: if ('\0' != *basedir || '\0' != *file || NULL != format)
2412: fputs(": ", stderr);
1.47 schwarze 2413: perror(NULL);
1.80 schwarze 2414: } else
2415: fputc('\n', stderr);
1.1 schwarze 2416: }
![[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