Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.121
1.121 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.120 2014/09/09 19:33:55 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: */
1.117 schwarze 18: #include <sys/types.h>
1.47 schwarze 19: #include <sys/stat.h>
1.72 schwarze 20: #include <sys/wait.h>
1.1 schwarze 21:
22: #include <assert.h>
1.33 schwarze 23: #include <ctype.h>
1.34 schwarze 24: #include <errno.h>
1.1 schwarze 25: #include <fcntl.h>
1.47 schwarze 26: #include <fts.h>
1.1 schwarze 27: #include <getopt.h>
1.44 schwarze 28: #include <limits.h>
1.47 schwarze 29: #include <stddef.h>
1.1 schwarze 30: #include <stdio.h>
31: #include <stdint.h>
32: #include <stdlib.h>
33: #include <string.h>
1.14 schwarze 34: #include <unistd.h>
1.1 schwarze 35:
1.47 schwarze 36: #include <ohash.h>
37: #include <sqlite3.h>
38:
39: #include "mdoc.h"
1.1 schwarze 40: #include "man.h"
41: #include "mandoc.h"
1.76 schwarze 42: #include "mandoc_aux.h"
1.10 schwarze 43: #include "manpath.h"
1.47 schwarze 44: #include "mansearch.h"
1.1 schwarze 45:
1.68 schwarze 46: extern int mansearch_keymax;
47: extern const char *const mansearch_keynames[];
48:
1.47 schwarze 49: #define SQL_EXEC(_v) \
50: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 51: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 52: #define SQL_BIND_TEXT(_s, _i, _v) \
53: if (SQLITE_OK != sqlite3_bind_text \
54: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 55: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 56: #define SQL_BIND_INT(_s, _i, _v) \
57: if (SQLITE_OK != sqlite3_bind_int \
58: ((_s), (_i)++, (_v))) \
1.91 schwarze 59: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 60: #define SQL_BIND_INT64(_s, _i, _v) \
61: if (SQLITE_OK != sqlite3_bind_int64 \
62: ((_s), (_i)++, (_v))) \
1.91 schwarze 63: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 64: #define SQL_STEP(_s) \
65: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 66: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 67:
1.47 schwarze 68: enum op {
69: OP_DEFAULT = 0, /* new dbs from dir list or default config */
70: OP_CONFFILE, /* new databases from custom config file */
71: OP_UPDATE, /* delete/add entries in existing database */
72: OP_DELETE, /* delete entries from existing database */
73: OP_TEST /* change no databases, report potential problems */
74: };
1.11 schwarze 75:
1.47 schwarze 76: struct str {
1.53 schwarze 77: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 78: const struct mpage *mpage; /* if set, the owning parse */
79: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 80: char key[]; /* may contain escape sequences */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct inodev {
84: ino_t st_ino;
85: dev_t st_dev;
86: };
1.28 schwarze 87:
1.47 schwarze 88: struct mpage {
89: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 90: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 91: char *sec; /* section from file content */
92: char *arch; /* architecture from file content */
93: char *title; /* title from file content */
94: char *desc; /* description from file content */
95: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 96: int form; /* format from file content */
1.28 schwarze 97: };
98:
1.47 schwarze 99: struct mlink {
100: char file[PATH_MAX]; /* filename rel. to manpath */
101: char *dsec; /* section from directory */
102: char *arch; /* architecture from directory */
103: char *name; /* name from file name (not empty) */
104: char *fsec; /* section from file name suffix */
105: struct mlink *next; /* singly linked list */
1.75 schwarze 106: struct mpage *mpage; /* parent */
1.116 schwarze 107: int dform; /* format from directory */
108: int fform; /* format from file name suffix */
1.81 schwarze 109: int gzip; /* filename has a .gz suffix */
1.2 schwarze 110: };
111:
1.47 schwarze 112: enum stmt {
113: STMT_DELETE_PAGE = 0, /* delete mpage */
114: STMT_INSERT_PAGE, /* insert mpage */
115: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 116: STMT_INSERT_NAME, /* insert name */
1.47 schwarze 117: STMT_INSERT_KEY, /* insert parsed key */
118: STMT__MAX
1.1 schwarze 119: };
120:
1.47 schwarze 121: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 122:
1.19 schwarze 123: struct mdoc_handler {
1.47 schwarze 124: mdoc_fp fp; /* optional handler */
125: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 126: };
127:
1.47 schwarze 128: static void dbclose(int);
1.75 schwarze 129: static void dbadd(struct mpage *, struct mchars *);
130: static void dbadd_mlink(const struct mlink *mlink);
1.47 schwarze 131: static int dbopen(int);
132: static void dbprune(void);
133: static void filescan(const char *);
134: static void *hash_alloc(size_t, void *);
1.106 espie 135: static void hash_free(void *, void *);
136: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 137: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 138: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 139: static void mlink_free(struct mlink *);
140: static void mlinks_undupe(struct mpage *);
141: static void mpages_free(void);
1.58 schwarze 142: static void mpages_merge(struct mchars *, struct mparse *);
1.87 schwarze 143: static void names_check(void);
1.81 schwarze 144: static void parse_cat(struct mpage *, int);
1.47 schwarze 145: static void parse_man(struct mpage *, const struct man_node *);
146: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
147: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
148: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
149: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
150: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
151: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
152: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
153: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
154: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.69 schwarze 155: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 156: static void putkeys(const struct mpage *,
157: const char *, size_t, uint64_t);
158: static void putmdockey(const struct mpage *,
159: const struct mdoc_node *, uint64_t);
1.53 schwarze 160: static void render_key(struct mchars *, struct str *);
1.47 schwarze 161: static void say(const char *, const char *, ...);
1.120 schwarze 162: static int set_basedir(const char *, int);
1.47 schwarze 163: static int treescan(void);
164: static size_t utf8(unsigned int, char [7]);
165:
1.72 schwarze 166: static char tempfilename[32];
1.47 schwarze 167: static char *progname;
1.59 schwarze 168: static int nodb; /* no database changes */
1.73 schwarze 169: static int mparse_options; /* abort the parse early */
1.98 schwarze 170: static int use_all; /* use all found files */
171: static int debug; /* print what we're doing */
172: static int warnings; /* warn about crap */
1.52 schwarze 173: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 174: static int exitcode; /* to be returned by main */
1.98 schwarze 175: static enum op op; /* operational mode */
1.47 schwarze 176: static char basedir[PATH_MAX]; /* current base directory */
177: static struct ohash mpages; /* table of distinct manual pages */
178: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 179: static struct ohash names; /* table of all names */
1.47 schwarze 180: static struct ohash strings; /* table of all strings */
181: static sqlite3 *db = NULL; /* current database */
182: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 183: static uint64_t name_mask;
1.47 schwarze 184:
1.19 schwarze 185: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 186: { NULL, 0 }, /* Ap */
187: { NULL, 0 }, /* Dd */
188: { NULL, 0 }, /* Dt */
189: { NULL, 0 }, /* Os */
190: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
191: { parse_mdoc_head, TYPE_Ss }, /* Ss */
192: { NULL, 0 }, /* Pp */
193: { NULL, 0 }, /* D1 */
194: { NULL, 0 }, /* Dl */
195: { NULL, 0 }, /* Bd */
196: { NULL, 0 }, /* Ed */
197: { NULL, 0 }, /* Bl */
198: { NULL, 0 }, /* El */
199: { NULL, 0 }, /* It */
200: { NULL, 0 }, /* Ad */
201: { NULL, TYPE_An }, /* An */
202: { NULL, TYPE_Ar }, /* Ar */
203: { NULL, TYPE_Cd }, /* Cd */
204: { NULL, TYPE_Cm }, /* Cm */
205: { NULL, TYPE_Dv }, /* Dv */
206: { NULL, TYPE_Er }, /* Er */
207: { NULL, TYPE_Ev }, /* Ev */
208: { NULL, 0 }, /* Ex */
209: { NULL, TYPE_Fa }, /* Fa */
210: { parse_mdoc_Fd, 0 }, /* Fd */
211: { NULL, TYPE_Fl }, /* Fl */
212: { parse_mdoc_Fn, 0 }, /* Fn */
213: { NULL, TYPE_Ft }, /* Ft */
214: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 215: { NULL, TYPE_In }, /* In */
1.47 schwarze 216: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 217: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 218: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 219: { NULL, 0 }, /* Op */
220: { NULL, 0 }, /* Ot */
221: { NULL, TYPE_Pa }, /* Pa */
222: { NULL, 0 }, /* Rv */
1.49 schwarze 223: { NULL, TYPE_St }, /* St */
1.47 schwarze 224: { NULL, TYPE_Va }, /* Va */
225: { parse_mdoc_body, TYPE_Va }, /* Vt */
226: { parse_mdoc_Xr, 0 }, /* Xr */
227: { NULL, 0 }, /* %A */
228: { NULL, 0 }, /* %B */
229: { NULL, 0 }, /* %D */
230: { NULL, 0 }, /* %I */
231: { NULL, 0 }, /* %J */
232: { NULL, 0 }, /* %N */
233: { NULL, 0 }, /* %O */
234: { NULL, 0 }, /* %P */
235: { NULL, 0 }, /* %R */
236: { NULL, 0 }, /* %T */
237: { NULL, 0 }, /* %V */
238: { NULL, 0 }, /* Ac */
239: { NULL, 0 }, /* Ao */
240: { NULL, 0 }, /* Aq */
241: { NULL, TYPE_At }, /* At */
242: { NULL, 0 }, /* Bc */
243: { NULL, 0 }, /* Bf */
244: { NULL, 0 }, /* Bo */
245: { NULL, 0 }, /* Bq */
246: { NULL, TYPE_Bsx }, /* Bsx */
247: { NULL, TYPE_Bx }, /* Bx */
248: { NULL, 0 }, /* Db */
249: { NULL, 0 }, /* Dc */
250: { NULL, 0 }, /* Do */
251: { NULL, 0 }, /* Dq */
252: { NULL, 0 }, /* Ec */
253: { NULL, 0 }, /* Ef */
254: { NULL, TYPE_Em }, /* Em */
255: { NULL, 0 }, /* Eo */
256: { NULL, TYPE_Fx }, /* Fx */
257: { NULL, TYPE_Ms }, /* Ms */
258: { NULL, 0 }, /* No */
259: { NULL, 0 }, /* Ns */
260: { NULL, TYPE_Nx }, /* Nx */
261: { NULL, TYPE_Ox }, /* Ox */
262: { NULL, 0 }, /* Pc */
263: { NULL, 0 }, /* Pf */
264: { NULL, 0 }, /* Po */
265: { NULL, 0 }, /* Pq */
266: { NULL, 0 }, /* Qc */
267: { NULL, 0 }, /* Ql */
268: { NULL, 0 }, /* Qo */
269: { NULL, 0 }, /* Qq */
270: { NULL, 0 }, /* Re */
271: { NULL, 0 }, /* Rs */
272: { NULL, 0 }, /* Sc */
273: { NULL, 0 }, /* So */
274: { NULL, 0 }, /* Sq */
275: { NULL, 0 }, /* Sm */
276: { NULL, 0 }, /* Sx */
277: { NULL, TYPE_Sy }, /* Sy */
278: { NULL, TYPE_Tn }, /* Tn */
279: { NULL, 0 }, /* Ux */
280: { NULL, 0 }, /* Xc */
281: { NULL, 0 }, /* Xo */
282: { parse_mdoc_head, 0 }, /* Fo */
283: { NULL, 0 }, /* Fc */
284: { NULL, 0 }, /* Oo */
285: { NULL, 0 }, /* Oc */
286: { NULL, 0 }, /* Bk */
287: { NULL, 0 }, /* Ek */
288: { NULL, 0 }, /* Bt */
289: { NULL, 0 }, /* Hf */
290: { NULL, 0 }, /* Fr */
291: { NULL, 0 }, /* Ud */
292: { NULL, TYPE_Lb }, /* Lb */
293: { NULL, 0 }, /* Lp */
294: { NULL, TYPE_Lk }, /* Lk */
295: { NULL, TYPE_Mt }, /* Mt */
296: { NULL, 0 }, /* Brq */
297: { NULL, 0 }, /* Bro */
298: { NULL, 0 }, /* Brc */
299: { NULL, 0 }, /* %C */
300: { NULL, 0 }, /* Es */
301: { NULL, 0 }, /* En */
302: { NULL, TYPE_Dx }, /* Dx */
303: { NULL, 0 }, /* %Q */
304: { NULL, 0 }, /* br */
305: { NULL, 0 }, /* sp */
306: { NULL, 0 }, /* %U */
307: { NULL, 0 }, /* Ta */
1.121 ! schwarze 308: { NULL, 0 }, /* ll */
1.1 schwarze 309: };
310:
1.98 schwarze 311:
1.1 schwarze 312: int
1.3 schwarze 313: mandocdb(int argc, char *argv[])
1.1 schwarze 314: {
1.47 schwarze 315: int ch, i;
316: size_t j, sz;
317: const char *path_arg;
318: struct mchars *mc;
319: struct manpaths dirs;
320: struct mparse *mp;
321: struct ohash_info mpages_info, mlinks_info;
322:
323: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
324: memset(&dirs, 0, sizeof(struct manpaths));
325:
326: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 327: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
328: mpages_info.free = mlinks_info.free = hash_free;
1.47 schwarze 329:
330: mpages_info.key_offset = offsetof(struct mpage, inodev);
331: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 332:
333: progname = strrchr(argv[0], '/');
334: if (progname == NULL)
335: progname = argv[0];
336: else
337: ++progname;
338:
1.47 schwarze 339: /*
1.98 schwarze 340: * We accept a few different invocations.
1.47 schwarze 341: * The CHECKOP macro makes sure that invocation styles don't
342: * clobber each other.
343: */
344: #define CHECKOP(_op, _ch) do \
345: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 346: fprintf(stderr, "%s: -%c: Conflicting option\n", \
347: progname, (_ch)); \
1.47 schwarze 348: goto usage; \
349: } while (/*CONSTCOND*/0)
1.10 schwarze 350:
1.47 schwarze 351: path_arg = NULL;
1.28 schwarze 352: op = OP_DEFAULT;
1.1 schwarze 353:
1.83 schwarze 354: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 355: switch (ch) {
1.98 schwarze 356: case 'a':
1.6 schwarze 357: use_all = 1;
358: break;
1.98 schwarze 359: case 'C':
1.47 schwarze 360: CHECKOP(op, ch);
361: path_arg = optarg;
1.28 schwarze 362: op = OP_CONFFILE;
1.25 schwarze 363: break;
1.98 schwarze 364: case 'D':
1.82 schwarze 365: debug++;
366: break;
1.98 schwarze 367: case 'd':
1.47 schwarze 368: CHECKOP(op, ch);
369: path_arg = optarg;
1.2 schwarze 370: op = OP_UPDATE;
1.1 schwarze 371: break;
1.98 schwarze 372: case 'n':
1.47 schwarze 373: nodb = 1;
374: break;
1.98 schwarze 375: case 'p':
1.83 schwarze 376: warnings = 1;
377: break;
1.98 schwarze 378: case 'Q':
1.73 schwarze 379: mparse_options |= MPARSE_QUICK;
1.59 schwarze 380: break;
1.98 schwarze 381: case 'T':
1.52 schwarze 382: if (strcmp(optarg, "utf8")) {
1.111 schwarze 383: fprintf(stderr, "%s: -T%s: "
384: "Unsupported output format\n",
385: progname, optarg);
1.52 schwarze 386: goto usage;
387: }
388: write_utf8 = 1;
389: break;
1.98 schwarze 390: case 't':
1.47 schwarze 391: CHECKOP(op, ch);
1.28 schwarze 392: dup2(STDOUT_FILENO, STDERR_FILENO);
393: op = OP_TEST;
1.47 schwarze 394: nodb = warnings = 1;
1.28 schwarze 395: break;
1.98 schwarze 396: case 'u':
1.47 schwarze 397: CHECKOP(op, ch);
398: path_arg = optarg;
1.1 schwarze 399: op = OP_DELETE;
400: break;
1.98 schwarze 401: case 'v':
1.83 schwarze 402: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 403: break;
1.1 schwarze 404: default:
1.28 schwarze 405: goto usage;
1.1 schwarze 406: }
407:
408: argc -= optind;
409: argv += optind;
410:
1.28 schwarze 411: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 412: fprintf(stderr, "%s: -C: Too many arguments\n",
413: progname);
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: /*
1.109 schwarze 427: * Most of these deal with a specific directory.
1.97 schwarze 428: * Jump into that directory first.
1.47 schwarze 429: */
1.120 schwarze 430: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 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) {
1.101 schwarze 465: dirs.paths = mandoc_reallocarray(NULL,
466: argc, sizeof(char *));
1.47 schwarze 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:
1.120 schwarze 496: if (0 == set_basedir(dirs.paths[j], argc > 0))
497: continue;
1.47 schwarze 498: if (0 == treescan())
1.120 schwarze 499: continue;
1.47 schwarze 500: if (0 == dbopen(0))
1.120 schwarze 501: continue;
1.41 deraadt 502:
1.58 schwarze 503: mpages_merge(mc, mp);
1.104 schwarze 504: if (warnings && !nodb &&
1.87 schwarze 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: manpath_free(&dirs);
518: mchars_free(mc);
519: mparse_free(mp);
520: mpages_free();
521: ohash_delete(&mpages);
522: ohash_delete(&mlinks);
523: return(exitcode);
524: usage:
1.83 schwarze 525: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
526: " %s [-aDnpQ] [-Tutf8] dir ...\n"
527: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
528: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 529: " %s [-Q] -t file ...\n",
1.98 schwarze 530: progname, progname, progname,
1.47 schwarze 531: progname, progname);
1.1 schwarze 532:
1.47 schwarze 533: return((int)MANDOCLEVEL_BADARG);
534: }
1.1 schwarze 535:
1.47 schwarze 536: /*
537: * Scan a directory tree rooted at "basedir" for manpages.
538: * We use fts(), scanning directory parts along the way for clues to our
539: * section and architecture.
540: *
541: * If use_all has been specified, grok all files.
542: * If not, sanitise paths to the following:
543: *
1.98 schwarze 544: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 545: * or
546: * [./]cat<section>[/<arch>]/<name>.0
547: *
548: * TODO: accomodate for multi-language directories.
549: */
550: static int
551: treescan(void)
552: {
1.96 schwarze 553: char buf[PATH_MAX];
1.47 schwarze 554: FTS *f;
555: FTSENT *ff;
556: struct mlink *mlink;
1.81 schwarze 557: int dform, gzip;
1.51 schwarze 558: char *dsec, *arch, *fsec, *cp;
559: const char *path;
1.47 schwarze 560: const char *argv[2];
1.1 schwarze 561:
1.47 schwarze 562: argv[0] = ".";
563: argv[1] = (char *)NULL;
1.1 schwarze 564:
1.96 schwarze 565: f = fts_open((char * const *)argv,
566: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 567: if (NULL == f) {
568: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 569: say("", "&fts_open");
1.47 schwarze 570: return(0);
571: }
1.2 schwarze 572:
1.47 schwarze 573: dsec = arch = NULL;
574: dform = FORM_NONE;
1.2 schwarze 575:
1.47 schwarze 576: while (NULL != (ff = fts_read(f))) {
577: path = ff->fts_path + 2;
1.96 schwarze 578: switch (ff->fts_info) {
579:
580: /*
581: * Symbolic links require various sanity checks,
582: * then get handled just like regular files.
583: */
1.98 schwarze 584: case FTS_SL:
1.96 schwarze 585: if (NULL == realpath(path, buf)) {
586: if (warnings)
587: say(path, "&realpath");
588: continue;
589: }
590: if (strstr(buf, basedir) != buf) {
591: if (warnings) say("",
592: "%s: outside base directory", buf);
593: continue;
594: }
595: /* Use logical inode to avoid mpages dupe. */
596: if (-1 == stat(path, ff->fts_statp)) {
597: if (warnings)
598: say(path, "&stat");
599: continue;
600: }
601: /* FALLTHROUGH */
602:
1.15 schwarze 603: /*
1.47 schwarze 604: * If we're a regular file, add an mlink by using the
605: * stored directory data and handling the filename.
1.15 schwarze 606: */
1.98 schwarze 607: case FTS_F:
1.47 schwarze 608: if (0 == strcmp(path, MANDOC_DB))
609: continue;
610: if ( ! use_all && ff->fts_level < 2) {
611: if (warnings)
612: say(path, "Extraneous file");
613: continue;
1.81 schwarze 614: }
615: gzip = 0;
616: fsec = NULL;
617: while (NULL == fsec) {
618: fsec = strrchr(ff->fts_name, '.');
619: if (NULL == fsec || strcmp(fsec+1, "gz"))
620: break;
621: gzip = 1;
622: *fsec = '\0';
623: fsec = NULL;
624: }
625: if (NULL == fsec) {
1.47 schwarze 626: if ( ! use_all) {
627: if (warnings)
628: say(path,
629: "No filename suffix");
630: continue;
631: }
632: } else if (0 == strcmp(++fsec, "html")) {
633: if (warnings)
634: say(path, "Skip html");
635: continue;
636: } else if (0 == strcmp(fsec, "ps")) {
637: if (warnings)
638: say(path, "Skip ps");
639: continue;
640: } else if (0 == strcmp(fsec, "pdf")) {
641: if (warnings)
642: say(path, "Skip pdf");
643: continue;
644: } else if ( ! use_all &&
645: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
646: (FORM_CAT == dform && strcmp(fsec, "0")))) {
647: if (warnings)
648: say(path, "Wrong filename suffix");
649: continue;
650: } else
651: fsec[-1] = '\0';
1.51 schwarze 652:
1.47 schwarze 653: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 654: if (strlcpy(mlink->file, path,
655: sizeof(mlink->file)) >=
656: sizeof(mlink->file)) {
657: say(path, "Filename too long");
658: free(mlink);
659: continue;
660: }
1.47 schwarze 661: mlink->dform = dform;
1.51 schwarze 662: mlink->dsec = dsec;
663: mlink->arch = arch;
664: mlink->name = ff->fts_name;
665: mlink->fsec = fsec;
1.81 schwarze 666: mlink->gzip = gzip;
1.47 schwarze 667: mlink_add(mlink, ff->fts_statp);
668: continue;
1.96 schwarze 669:
1.98 schwarze 670: case FTS_D:
1.96 schwarze 671: /* FALLTHROUGH */
1.98 schwarze 672: case FTS_DP:
1.96 schwarze 673: break;
674:
675: default:
1.47 schwarze 676: if (warnings)
677: say(path, "Not a regular file");
678: continue;
679: }
680:
681: switch (ff->fts_level) {
1.98 schwarze 682: case 0:
1.47 schwarze 683: /* Ignore the root directory. */
684: break;
1.98 schwarze 685: case 1:
1.47 schwarze 686: /*
687: * This might contain manX/ or catX/.
688: * Try to infer this from the name.
689: * If we're not in use_all, enforce it.
690: */
691: cp = ff->fts_name;
692: if (FTS_DP == ff->fts_info)
693: break;
1.15 schwarze 694:
1.47 schwarze 695: if (0 == strncmp(cp, "man", 3)) {
696: dform = FORM_SRC;
697: dsec = cp + 3;
698: } else if (0 == strncmp(cp, "cat", 3)) {
699: dform = FORM_CAT;
700: dsec = cp + 3;
1.51 schwarze 701: } else {
702: dform = FORM_NONE;
703: dsec = NULL;
1.26 schwarze 704: }
1.47 schwarze 705:
1.98 schwarze 706: if (NULL != dsec || use_all)
1.47 schwarze 707: break;
708:
709: if (warnings)
710: say(path, "Unknown directory part");
711: fts_set(f, ff, FTS_SKIP);
712: break;
1.98 schwarze 713: case 2:
1.47 schwarze 714: /*
715: * Possibly our architecture.
716: * If we're descending, keep tabs on it.
717: */
718: if (FTS_DP != ff->fts_info && NULL != dsec)
719: arch = ff->fts_name;
1.51 schwarze 720: else
721: arch = NULL;
1.47 schwarze 722: break;
723: default:
724: if (FTS_DP == ff->fts_info || use_all)
725: break;
726: if (warnings)
727: say(path, "Extraneous directory part");
728: fts_set(f, ff, FTS_SKIP);
729: break;
1.14 schwarze 730: }
1.47 schwarze 731: }
732:
733: fts_close(f);
734: return(1);
735: }
1.1 schwarze 736:
1.47 schwarze 737: /*
738: * Add a file to the mlinks table.
739: * Do not verify that it's a "valid" looking manpage (we'll do that
740: * later).
741: *
742: * Try to infer the manual section, architecture, and page name from the
743: * path, assuming it looks like
744: *
1.98 schwarze 745: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 746: * or
747: * [./]cat<section>[/<arch>]/<name>.0
748: *
749: * See treescan() for the fts(3) version of this.
750: */
751: static void
752: filescan(const char *file)
753: {
754: char buf[PATH_MAX];
755: struct stat st;
756: struct mlink *mlink;
757: char *p, *start;
758:
759: assert(use_all);
760:
761: if (0 == strncmp(file, "./", 2))
762: file += 2;
763:
1.96 schwarze 764: /*
765: * We have to do lstat(2) before realpath(3) loses
766: * the information whether this is a symbolic link.
767: * We need to know that because for symbolic links,
768: * we want to use the orginal file name, while for
769: * regular files, we want to use the real path.
770: */
771: if (-1 == lstat(file, &st)) {
772: exitcode = (int)MANDOCLEVEL_BADARG;
773: say(file, "&lstat");
774: return;
775: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
776: exitcode = (int)MANDOCLEVEL_BADARG;
777: say(file, "Not a regular file");
778: return;
779: }
780:
781: /*
782: * We have to resolve the file name to the real path
783: * in any case for the base directory check.
784: */
1.47 schwarze 785: if (NULL == realpath(file, buf)) {
786: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 787: say(file, "&realpath");
1.47 schwarze 788: return;
1.63 schwarze 789: }
790:
1.109 schwarze 791: if (OP_TEST == op)
1.63 schwarze 792: start = buf;
1.109 schwarze 793: else if (strstr(buf, basedir) == buf)
794: start = buf + strlen(basedir);
1.63 schwarze 795: else {
1.47 schwarze 796: exitcode = (int)MANDOCLEVEL_BADARG;
797: say("", "%s: outside base directory", buf);
798: return;
1.63 schwarze 799: }
800:
1.96 schwarze 801: /*
802: * Now we are sure the file is inside our tree.
803: * If it is a symbolic link, ignore the real path
804: * and use the original name.
805: * This implies passing stuff like "cat1/../man1/foo.1"
806: * on the command line won't work. So don't do that.
807: * Note the stat(2) can still fail if the link target
808: * doesn't exist.
809: */
810: if (S_IFLNK & st.st_mode) {
811: if (-1 == stat(buf, &st)) {
812: exitcode = (int)MANDOCLEVEL_BADARG;
813: say(file, "&stat");
814: return;
815: }
1.100 schwarze 816: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
817: say(file, "Filename too long");
818: return;
819: }
1.109 schwarze 820: start = buf;
821: if (OP_TEST != op && strstr(buf, basedir) == buf)
822: start += strlen(basedir);
1.1 schwarze 823: }
1.63 schwarze 824:
1.47 schwarze 825: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 826: mlink->dform = FORM_NONE;
1.100 schwarze 827: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
828: sizeof(mlink->file)) {
829: say(start, "Filename too long");
830: return;
831: }
1.1 schwarze 832:
1.10 schwarze 833: /*
1.47 schwarze 834: * First try to guess our directory structure.
835: * If we find a separator, try to look for man* or cat*.
836: * If we find one of these and what's underneath is a directory,
837: * assume it's an architecture.
1.10 schwarze 838: */
1.47 schwarze 839: if (NULL != (p = strchr(start, '/'))) {
840: *p++ = '\0';
841: if (0 == strncmp(start, "man", 3)) {
842: mlink->dform = FORM_SRC;
1.51 schwarze 843: mlink->dsec = start + 3;
1.47 schwarze 844: } else if (0 == strncmp(start, "cat", 3)) {
845: mlink->dform = FORM_CAT;
1.51 schwarze 846: mlink->dsec = start + 3;
1.47 schwarze 847: }
1.10 schwarze 848:
1.47 schwarze 849: start = p;
850: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
851: *p++ = '\0';
1.51 schwarze 852: mlink->arch = start;
1.47 schwarze 853: start = p;
1.41 deraadt 854: }
1.47 schwarze 855: }
1.7 schwarze 856:
1.47 schwarze 857: /*
858: * Now check the file suffix.
859: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
860: */
861: p = strrchr(start, '\0');
862: while (p-- > start && '/' != *p && '.' != *p)
863: /* Loop. */ ;
864:
865: if ('.' == *p) {
866: *p++ = '\0';
1.51 schwarze 867: mlink->fsec = p;
1.47 schwarze 868: }
1.41 deraadt 869:
1.47 schwarze 870: /*
871: * Now try to parse the name.
872: * Use the filename portion of the path.
873: */
874: mlink->name = start;
875: if (NULL != (p = strrchr(start, '/'))) {
876: mlink->name = p + 1;
877: *p = '\0';
878: }
879: mlink_add(mlink, &st);
880: }
1.2 schwarze 881:
1.47 schwarze 882: static void
883: mlink_add(struct mlink *mlink, const struct stat *st)
884: {
885: struct inodev inodev;
886: struct mpage *mpage;
887: unsigned int slot;
888:
889: assert(NULL != mlink->file);
890:
1.51 schwarze 891: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
892: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
893: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
894: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 895:
896: if ('0' == *mlink->fsec) {
897: free(mlink->fsec);
898: mlink->fsec = mandoc_strdup(mlink->dsec);
899: mlink->fform = FORM_CAT;
900: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
901: mlink->fform = FORM_SRC;
902: else
903: mlink->fform = FORM_NONE;
1.1 schwarze 904:
1.47 schwarze 905: slot = ohash_qlookup(&mlinks, mlink->file);
906: assert(NULL == ohash_find(&mlinks, slot));
907: ohash_insert(&mlinks, slot, mlink);
908:
909: inodev.st_ino = st->st_ino;
910: inodev.st_dev = st->st_dev;
911: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
912: sizeof(struct inodev), inodev.st_ino);
913: mpage = ohash_find(&mpages, slot);
914: if (NULL == mpage) {
915: mpage = mandoc_calloc(1, sizeof(struct mpage));
916: mpage->inodev.st_ino = inodev.st_ino;
917: mpage->inodev.st_dev = inodev.st_dev;
918: ohash_insert(&mpages, slot, mpage);
919: } else
920: mlink->next = mpage->mlinks;
921: mpage->mlinks = mlink;
1.75 schwarze 922: mlink->mpage = mpage;
1.47 schwarze 923: }
1.1 schwarze 924:
1.47 schwarze 925: static void
926: mlink_free(struct mlink *mlink)
927: {
1.1 schwarze 928:
1.47 schwarze 929: free(mlink->dsec);
930: free(mlink->arch);
931: free(mlink->name);
932: free(mlink->fsec);
933: free(mlink);
934: }
1.1 schwarze 935:
1.47 schwarze 936: static void
937: mpages_free(void)
938: {
939: struct mpage *mpage;
940: struct mlink *mlink;
941: unsigned int slot;
942:
943: mpage = ohash_first(&mpages, &slot);
944: while (NULL != mpage) {
945: while (NULL != (mlink = mpage->mlinks)) {
946: mpage->mlinks = mlink->next;
947: mlink_free(mlink);
948: }
949: free(mpage->sec);
950: free(mpage->arch);
951: free(mpage->title);
952: free(mpage->desc);
953: free(mpage);
954: mpage = ohash_next(&mpages, &slot);
955: }
956: }
1.1 schwarze 957:
1.47 schwarze 958: /*
959: * For each mlink to the mpage, check whether the path looks like
960: * it is formatted, and if it does, check whether a source manual
961: * exists by the same name, ignoring the suffix.
962: * If both conditions hold, drop the mlink.
963: */
964: static void
965: mlinks_undupe(struct mpage *mpage)
966: {
967: char buf[PATH_MAX];
968: struct mlink **prev;
969: struct mlink *mlink;
970: char *bufp;
971:
972: mpage->form = FORM_CAT;
973: prev = &mpage->mlinks;
974: while (NULL != (mlink = *prev)) {
975: if (FORM_CAT != mlink->dform) {
976: mpage->form = FORM_NONE;
977: goto nextlink;
978: }
1.100 schwarze 979: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 980: bufp = strstr(buf, "cat");
981: assert(NULL != bufp);
982: memcpy(bufp, "man", 3);
983: if (NULL != (bufp = strrchr(buf, '.')))
984: *++bufp = '\0';
1.100 schwarze 985: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 986: if (NULL == ohash_find(&mlinks,
1.98 schwarze 987: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 988: goto nextlink;
989: if (warnings)
990: say(mlink->file, "Man source exists: %s", buf);
991: if (use_all)
992: goto nextlink;
993: *prev = mlink->next;
994: mlink_free(mlink);
995: continue;
996: nextlink:
997: prev = &(*prev)->next;
1.1 schwarze 998: }
1.47 schwarze 999: }
1.1 schwarze 1000:
1.88 schwarze 1001: static void
1.50 schwarze 1002: mlink_check(struct mpage *mpage, struct mlink *mlink)
1003: {
1.88 schwarze 1004: struct str *str;
1005: unsigned int slot;
1.50 schwarze 1006:
1007: /*
1008: * Check whether the manual section given in a file
1009: * agrees with the directory where the file is located.
1010: * Some manuals have suffixes like (3p) on their
1011: * section number either inside the file or in the
1012: * directory name, some are linked into more than one
1013: * section, like encrypt(1) = makekey(8).
1014: */
1015:
1016: if (FORM_SRC == mpage->form &&
1.88 schwarze 1017: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1018: say(mlink->file, "Section \"%s\" manual in %s directory",
1019: mpage->sec, mlink->dsec);
1020:
1021: /*
1022: * Manual page directories exist for each kernel
1023: * architecture as returned by machine(1).
1024: * However, many manuals only depend on the
1025: * application architecture as returned by arch(1).
1026: * For example, some (2/ARM) manuals are shared
1027: * across the "armish" and "zaurus" kernel
1028: * architectures.
1029: * A few manuals are even shared across completely
1030: * different architectures, for example fdformat(1)
1031: * on amd64, i386, sparc, and sparc64.
1032: */
1033:
1.88 schwarze 1034: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1035: say(mlink->file, "Architecture \"%s\" manual in "
1036: "\"%s\" directory", mpage->arch, mlink->arch);
1037:
1.88 schwarze 1038: /*
1039: * XXX
1.90 schwarze 1040: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1041: */
1042:
1043: if (FORM_CAT == mpage->form)
1044: return;
1045:
1046: /*
1047: * Check whether this mlink
1048: * appears as a name in the NAME section.
1049: */
1.50 schwarze 1050:
1.90 schwarze 1051: slot = ohash_qlookup(&names, mlink->name);
1052: str = ohash_find(&names, slot);
1.88 schwarze 1053: assert(NULL != str);
1.90 schwarze 1054: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1055: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1056: }
1057:
1.47 schwarze 1058: /*
1059: * Run through the files in the global vector "mpages"
1060: * and add them to the database specified in "basedir".
1061: *
1062: * This handles the parsing scheme itself, using the cues of directory
1063: * and filename to determine whether the file is parsable or not.
1064: */
1065: static void
1.58 schwarze 1066: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1067: {
1.70 schwarze 1068: char any[] = "any";
1.58 schwarze 1069: struct ohash_info str_info;
1.75 schwarze 1070: struct mpage *mpage, *mpage_dest;
1071: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1072: struct mdoc *mdoc;
1073: struct man *man;
1.75 schwarze 1074: char *sodest;
1.69 schwarze 1075: char *cp;
1.81 schwarze 1076: pid_t child_pid;
1.117 schwarze 1077: int fd;
1.58 schwarze 1078: unsigned int pslot;
1.47 schwarze 1079: enum mandoclevel lvl;
1080:
1081: str_info.alloc = hash_alloc;
1.106 espie 1082: str_info.calloc = hash_calloc;
1083: str_info.free = hash_free;
1.47 schwarze 1084: str_info.key_offset = offsetof(struct str, key);
1085:
1.64 schwarze 1086: if (0 == nodb)
1087: SQL_EXEC("BEGIN TRANSACTION");
1088:
1.47 schwarze 1089: mpage = ohash_first(&mpages, &pslot);
1090: while (NULL != mpage) {
1091: mlinks_undupe(mpage);
1092: if (NULL == mpage->mlinks) {
1093: mpage = ohash_next(&mpages, &pslot);
1094: continue;
1095: }
1.1 schwarze 1096:
1.90 schwarze 1097: name_mask = NAME_MASK;
1098: ohash_init(&names, 4, &str_info);
1.47 schwarze 1099: ohash_init(&strings, 6, &str_info);
1100: mparse_reset(mp);
1101: mdoc = NULL;
1102: man = NULL;
1.81 schwarze 1103: sodest = NULL;
1104: child_pid = 0;
1105:
1.117 schwarze 1106: mparse_open(mp, &fd, mpage->mlinks->file, &child_pid);
1107: if (fd == -1) {
1108: say(mpage->mlinks->file, "&open");
1109: goto nextpage;
1.81 schwarze 1110: }
1.11 schwarze 1111:
1112: /*
1.24 schwarze 1113: * Try interpreting the file as mdoc(7) or man(7)
1114: * source code, unless it is already known to be
1115: * formatted. Fall back to formatted mode.
1.11 schwarze 1116: */
1.47 schwarze 1117: if (FORM_CAT != mpage->mlinks->dform ||
1118: FORM_CAT != mpage->mlinks->fform) {
1.117 schwarze 1119: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1120: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1121: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1122: }
1.11 schwarze 1123:
1.75 schwarze 1124: if (NULL != sodest) {
1125: mlink_dest = ohash_find(&mlinks,
1126: ohash_qlookup(&mlinks, sodest));
1127: if (NULL != mlink_dest) {
1128:
1129: /* The .so target exists. */
1130:
1131: mpage_dest = mlink_dest->mpage;
1132: mlink = mpage->mlinks;
1133: while (1) {
1134: mlink->mpage = mpage_dest;
1135:
1136: /*
1137: * If the target was already
1138: * processed, add the links
1139: * to the database now.
1140: * Otherwise, this will
1141: * happen when we come
1142: * to the target.
1143: */
1144:
1.94 schwarze 1145: if (mpage_dest->pageid)
1.75 schwarze 1146: dbadd_mlink(mlink);
1147:
1148: if (NULL == mlink->next)
1149: break;
1150: mlink = mlink->next;
1151: }
1152:
1153: /* Move all links to the target. */
1154:
1155: mlink->next = mlink_dest->next;
1156: mlink_dest->next = mpage->mlinks;
1157: mpage->mlinks = NULL;
1158: }
1.81 schwarze 1159: goto nextpage;
1.75 schwarze 1160: } else if (NULL != mdoc) {
1.47 schwarze 1161: mpage->form = FORM_SRC;
1.112 schwarze 1162: mpage->sec = mdoc_meta(mdoc)->msec;
1163: mpage->sec = mandoc_strdup(
1164: NULL == mpage->sec ? "" : mpage->sec);
1.47 schwarze 1165: mpage->arch = mdoc_meta(mdoc)->arch;
1166: mpage->arch = mandoc_strdup(
1167: NULL == mpage->arch ? "" : mpage->arch);
1168: mpage->title =
1169: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 1170: } else if (NULL != man) {
1.47 schwarze 1171: mpage->form = FORM_SRC;
1172: mpage->sec =
1173: mandoc_strdup(man_meta(man)->msec);
1174: mpage->arch =
1175: mandoc_strdup(mpage->mlinks->arch);
1176: mpage->title =
1177: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1178: } else {
1.47 schwarze 1179: mpage->form = FORM_CAT;
1180: mpage->sec =
1181: mandoc_strdup(mpage->mlinks->dsec);
1182: mpage->arch =
1183: mandoc_strdup(mpage->mlinks->arch);
1184: mpage->title =
1185: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1186: }
1.116 schwarze 1187: if (mpage->mlinks->gzip)
1188: mpage->form |= FORM_GZ;
1.54 schwarze 1189: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1190: if (*mpage->arch != '\0')
1191: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1192:
1.54 schwarze 1193: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1194: if ('\0' != *mlink->dsec)
1195: putkey(mpage, mlink->dsec, TYPE_sec);
1196: if ('\0' != *mlink->fsec)
1197: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1198: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1199: any : mlink->arch, TYPE_arch);
1.90 schwarze 1200: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1201: }
1.41 deraadt 1202:
1.92 schwarze 1203: assert(NULL == mpage->desc);
1.47 schwarze 1204: if (NULL != mdoc) {
1205: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.90 schwarze 1206: putkey(mpage, cp, NAME_HEAD);
1.47 schwarze 1207: parse_mdoc(mpage, mdoc_node(mdoc));
1208: } else if (NULL != man)
1209: parse_man(mpage, man_node(man));
1210: else
1.117 schwarze 1211: parse_cat(mpage, fd);
1.92 schwarze 1212: if (NULL == mpage->desc)
1213: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1214:
1215: if (warnings && !use_all)
1216: for (mlink = mpage->mlinks; mlink;
1217: mlink = mlink->next)
1218: mlink_check(mpage, mlink);
1.6 schwarze 1219:
1.62 schwarze 1220: dbadd(mpage, mc);
1.81 schwarze 1221:
1222: nextpage:
1.117 schwarze 1223: if (child_pid &&
1224: mparse_wait(mp, child_pid) != MANDOCLEVEL_OK) {
1225: exitcode = (int)MANDOCLEVEL_SYSERR;
1226: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1227: }
1.47 schwarze 1228: ohash_delete(&strings);
1.90 schwarze 1229: ohash_delete(&names);
1.47 schwarze 1230: mpage = ohash_next(&mpages, &pslot);
1231: }
1.64 schwarze 1232:
1233: if (0 == nodb)
1234: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1235: }
1236:
1237: static void
1238: names_check(void)
1239: {
1240: sqlite3_stmt *stmt;
1241: const char *name, *sec, *arch, *key;
1242: int irc;
1243:
1244: sqlite3_prepare_v2(db,
1245: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1246: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1247: "WHERE bits & ? AND NOT EXISTS ("
1248: "SELECT pageid FROM mlinks "
1.90 schwarze 1249: "WHERE mlinks.pageid == names.pageid "
1250: "AND mlinks.name == names.name"
1.87 schwarze 1251: ")"
1252: ") JOIN ("
1.99 schwarze 1253: "SELECT sec, arch, name, pageid FROM mlinks "
1254: "GROUP BY pageid"
1.87 schwarze 1255: ") USING (pageid);",
1256: -1, &stmt, NULL);
1257:
1.91 schwarze 1258: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1259: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1260:
1261: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1262: name = (const char *)sqlite3_column_text(stmt, 0);
1263: sec = (const char *)sqlite3_column_text(stmt, 1);
1264: arch = (const char *)sqlite3_column_text(stmt, 2);
1265: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1266: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1267: '\0' == *arch ? "" : "/",
1268: '\0' == *arch ? "" : arch, key);
1269: }
1270: sqlite3_finalize(stmt);
1.47 schwarze 1271: }
1.6 schwarze 1272:
1.47 schwarze 1273: static void
1.81 schwarze 1274: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1275: {
1276: FILE *stream;
1277: char *line, *p, *title;
1278: size_t len, plen, titlesz;
1.1 schwarze 1279:
1.81 schwarze 1280: stream = (-1 == fd) ?
1281: fopen(mpage->mlinks->file, "r") :
1282: fdopen(fd, "r");
1283: if (NULL == stream) {
1.114 doug 1284: if (-1 != fd)
1285: close(fd);
1.47 schwarze 1286: if (warnings)
1.80 schwarze 1287: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1288: return;
1289: }
1.1 schwarze 1290:
1.47 schwarze 1291: /* Skip to first blank line. */
1.1 schwarze 1292:
1.47 schwarze 1293: while (NULL != (line = fgetln(stream, &len)))
1294: if ('\n' == *line)
1295: break;
1.1 schwarze 1296:
1.47 schwarze 1297: /*
1298: * Assume the first line that is not indented
1299: * is the first section header. Skip to it.
1300: */
1.1 schwarze 1301:
1.47 schwarze 1302: while (NULL != (line = fgetln(stream, &len)))
1303: if ('\n' != *line && ' ' != *line)
1304: break;
1.98 schwarze 1305:
1.47 schwarze 1306: /*
1307: * Read up until the next section into a buffer.
1308: * Strip the leading and trailing newline from each read line,
1309: * appending a trailing space.
1310: * Ignore empty (whitespace-only) lines.
1311: */
1.28 schwarze 1312:
1.47 schwarze 1313: titlesz = 0;
1314: title = NULL;
1.38 schwarze 1315:
1.47 schwarze 1316: while (NULL != (line = fgetln(stream, &len))) {
1317: if (' ' != *line || '\n' != line[len - 1])
1318: break;
1319: while (len > 0 && isspace((unsigned char)*line)) {
1320: line++;
1321: len--;
1322: }
1323: if (1 == len)
1324: continue;
1325: title = mandoc_realloc(title, titlesz + len);
1326: memcpy(title + titlesz, line, len);
1327: titlesz += len;
1328: title[titlesz - 1] = ' ';
1329: }
1.28 schwarze 1330:
1.47 schwarze 1331: /*
1332: * If no page content can be found, or the input line
1333: * is already the next section header, or there is no
1334: * trailing newline, reuse the page title as the page
1335: * description.
1336: */
1.1 schwarze 1337:
1.47 schwarze 1338: if (NULL == title || '\0' == *title) {
1339: if (warnings)
1340: say(mpage->mlinks->file,
1341: "Cannot find NAME section");
1342: fclose(stream);
1343: free(title);
1344: return;
1345: }
1.24 schwarze 1346:
1.47 schwarze 1347: title = mandoc_realloc(title, titlesz + 1);
1348: title[titlesz] = '\0';
1.24 schwarze 1349:
1.47 schwarze 1350: /*
1351: * Skip to the first dash.
1352: * Use the remaining line as the description (no more than 70
1353: * bytes).
1354: */
1.28 schwarze 1355:
1.47 schwarze 1356: if (NULL != (p = strstr(title, "- "))) {
1357: for (p += 2; ' ' == *p || '\b' == *p; p++)
1358: /* Skip to next word. */ ;
1359: } else {
1360: if (warnings)
1361: say(mpage->mlinks->file,
1362: "No dash in title line");
1363: p = title;
1364: }
1.1 schwarze 1365:
1.47 schwarze 1366: plen = strlen(p);
1.1 schwarze 1367:
1.47 schwarze 1368: /* Strip backspace-encoding from line. */
1.1 schwarze 1369:
1.47 schwarze 1370: while (NULL != (line = memchr(p, '\b', plen))) {
1371: len = line - p;
1372: if (0 == len) {
1373: memmove(line, line + 1, plen--);
1374: continue;
1.98 schwarze 1375: }
1.47 schwarze 1376: memmove(line - 1, line + 1, plen - len);
1377: plen -= 2;
1378: }
1.1 schwarze 1379:
1.47 schwarze 1380: mpage->desc = mandoc_strdup(p);
1381: fclose(stream);
1382: free(title);
1383: }
1.16 schwarze 1384:
1.47 schwarze 1385: /*
1386: * Put a type/word pair into the word database for this particular file.
1387: */
1388: static void
1.69 schwarze 1389: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1390: {
1.69 schwarze 1391: char *cp;
1.37 schwarze 1392:
1.47 schwarze 1393: assert(NULL != value);
1.69 schwarze 1394: if (TYPE_arch == type)
1395: for (cp = value; *cp; cp++)
1396: if (isupper((unsigned char)*cp))
1397: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1398: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1399: }
1400:
1401: /*
1.47 schwarze 1402: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1403: */
1404: static void
1.47 schwarze 1405: putmdockey(const struct mpage *mpage,
1406: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1407: {
1.16 schwarze 1408:
1.47 schwarze 1409: for ( ; NULL != n; n = n->next) {
1410: if (NULL != n->child)
1411: putmdockey(mpage, n->child, m);
1412: if (MDOC_TEXT == n->type)
1413: putkey(mpage, n->string, m);
1414: }
1415: }
1.16 schwarze 1416:
1.47 schwarze 1417: static void
1418: parse_man(struct mpage *mpage, const struct man_node *n)
1419: {
1420: const struct man_node *head, *body;
1.78 schwarze 1421: char *start, *title;
1.47 schwarze 1422: char byte;
1.78 schwarze 1423: size_t sz;
1.16 schwarze 1424:
1.47 schwarze 1425: if (NULL == n)
1426: return;
1.16 schwarze 1427:
1.47 schwarze 1428: /*
1429: * We're only searching for one thing: the first text child in
1430: * the BODY of a NAME section. Since we don't keep track of
1431: * sections in -man, run some hoops to find out whether we're in
1432: * the correct section or not.
1433: */
1.16 schwarze 1434:
1.47 schwarze 1435: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1436: body = n;
1437: assert(body->parent);
1438: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1439: 1 == head->nchild &&
1440: NULL != (head = (head->child)) &&
1441: MAN_TEXT == head->type &&
1442: 0 == strcmp(head->string, "NAME") &&
1443: NULL != body->child) {
1.2 schwarze 1444:
1.47 schwarze 1445: /*
1446: * Suck the entire NAME section into memory.
1447: * Yes, we might run away.
1448: * But too many manuals have big, spread-out
1449: * NAME sections over many lines.
1450: */
1.2 schwarze 1451:
1.78 schwarze 1452: title = NULL;
1453: man_deroff(&title, body);
1.47 schwarze 1454: if (NULL == title)
1455: return;
1.16 schwarze 1456:
1.98 schwarze 1457: /*
1.47 schwarze 1458: * Go through a special heuristic dance here.
1459: * Conventionally, one or more manual names are
1460: * comma-specified prior to a whitespace, then a
1461: * dash, then a description. Try to puzzle out
1462: * the name parts here.
1463: */
1.16 schwarze 1464:
1.78 schwarze 1465: start = title;
1.47 schwarze 1466: for ( ;; ) {
1467: sz = strcspn(start, " ,");
1468: if ('\0' == start[sz])
1469: break;
1.1 schwarze 1470:
1.47 schwarze 1471: byte = start[sz];
1472: start[sz] = '\0';
1.67 schwarze 1473:
1474: /*
1475: * Assume a stray trailing comma in the
1476: * name list if a name begins with a dash.
1477: */
1478:
1479: if ('-' == start[0] ||
1480: ('\\' == start[0] && '-' == start[1]))
1481: break;
1.1 schwarze 1482:
1.90 schwarze 1483: putkey(mpage, start, NAME_TITLE);
1.1 schwarze 1484:
1.47 schwarze 1485: if (' ' == byte) {
1486: start += sz + 1;
1487: break;
1488: }
1.1 schwarze 1489:
1.47 schwarze 1490: assert(',' == byte);
1491: start += sz + 1;
1492: while (' ' == *start)
1493: start++;
1494: }
1.1 schwarze 1495:
1.78 schwarze 1496: if (start == title) {
1.90 schwarze 1497: putkey(mpage, start, NAME_TITLE);
1.47 schwarze 1498: free(title);
1499: return;
1500: }
1.1 schwarze 1501:
1.47 schwarze 1502: while (isspace((unsigned char)*start))
1503: start++;
1.1 schwarze 1504:
1.47 schwarze 1505: if (0 == strncmp(start, "-", 1))
1506: start += 1;
1507: else if (0 == strncmp(start, "\\-\\-", 4))
1508: start += 4;
1509: else if (0 == strncmp(start, "\\-", 2))
1510: start += 2;
1511: else if (0 == strncmp(start, "\\(en", 4))
1512: start += 4;
1513: else if (0 == strncmp(start, "\\(em", 4))
1514: start += 4;
1.1 schwarze 1515:
1.47 schwarze 1516: while (' ' == *start)
1517: start++;
1.1 schwarze 1518:
1.47 schwarze 1519: mpage->desc = mandoc_strdup(start);
1520: free(title);
1521: return;
1522: }
1523: }
1.1 schwarze 1524:
1.47 schwarze 1525: for (n = n->child; n; n = n->next) {
1526: if (NULL != mpage->desc)
1527: break;
1528: parse_man(mpage, n);
1.1 schwarze 1529: }
1530: }
1531:
1532: static void
1.47 schwarze 1533: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1534: {
1535:
1.47 schwarze 1536: assert(NULL != n);
1537: for (n = n->child; NULL != n; n = n->next) {
1538: switch (n->type) {
1.98 schwarze 1539: case MDOC_ELEM:
1.47 schwarze 1540: /* FALLTHROUGH */
1.98 schwarze 1541: case MDOC_BLOCK:
1.47 schwarze 1542: /* FALLTHROUGH */
1.98 schwarze 1543: case MDOC_HEAD:
1.47 schwarze 1544: /* FALLTHROUGH */
1.98 schwarze 1545: case MDOC_BODY:
1.47 schwarze 1546: /* FALLTHROUGH */
1.98 schwarze 1547: case MDOC_TAIL:
1.47 schwarze 1548: if (NULL != mdocs[n->tok].fp)
1549: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1550: break;
1551: if (mdocs[n->tok].mask)
1552: putmdockey(mpage, n->child,
1553: mdocs[n->tok].mask);
1554: break;
1555: default:
1556: assert(MDOC_ROOT != n->type);
1557: continue;
1558: }
1559: if (NULL != n->child)
1560: parse_mdoc(mpage, n);
1.1 schwarze 1561: }
1562: }
1563:
1.19 schwarze 1564: static int
1.47 schwarze 1565: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1566: {
1567: const char *start, *end;
1568: size_t sz;
1.19 schwarze 1569:
1.47 schwarze 1570: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1571: NULL == (n = n->child) ||
1572: MDOC_TEXT != n->type)
1.19 schwarze 1573: return(0);
1.1 schwarze 1574:
1575: /*
1576: * Only consider those `Fd' macro fields that begin with an
1577: * "inclusion" token (versus, e.g., #define).
1578: */
1.47 schwarze 1579:
1.1 schwarze 1580: if (strcmp("#include", n->string))
1.19 schwarze 1581: return(0);
1.1 schwarze 1582:
1583: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1584: return(0);
1.1 schwarze 1585:
1586: /*
1587: * Strip away the enclosing angle brackets and make sure we're
1588: * not zero-length.
1589: */
1590:
1591: start = n->string;
1592: if ('<' == *start || '"' == *start)
1593: start++;
1594:
1595: if (0 == (sz = strlen(start)))
1.19 schwarze 1596: return(0);
1.1 schwarze 1597:
1598: end = &start[(int)sz - 1];
1599: if ('>' == *end || '"' == *end)
1600: end--;
1601:
1.47 schwarze 1602: if (end > start)
1603: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1604: return(0);
1.1 schwarze 1605: }
1606:
1.19 schwarze 1607: static int
1.47 schwarze 1608: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1609: {
1.69 schwarze 1610: char *cp;
1.1 schwarze 1611:
1.47 schwarze 1612: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1613: return(0);
1614:
1.98 schwarze 1615: /*
1.47 schwarze 1616: * Parse: .Fn "struct type *name" "char *arg".
1.98 schwarze 1617: * First strip away pointer symbol.
1.47 schwarze 1618: * Then store the function name, then type.
1.98 schwarze 1619: * Finally, store the arguments.
1.47 schwarze 1620: */
1.1 schwarze 1621:
1.47 schwarze 1622: if (NULL == (cp = strrchr(n->string, ' ')))
1623: cp = n->string;
1.1 schwarze 1624:
1625: while ('*' == *cp)
1626: cp++;
1627:
1.47 schwarze 1628: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1629:
1.47 schwarze 1630: if (n->string < cp)
1631: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1632:
1.47 schwarze 1633: for (n = n->next; NULL != n; n = n->next)
1634: if (MDOC_TEXT == n->type)
1635: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1636:
1637: return(0);
1.1 schwarze 1638: }
1639:
1.19 schwarze 1640: static int
1.47 schwarze 1641: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1642: {
1.47 schwarze 1643: char *cp;
1.1 schwarze 1644:
1645: if (NULL == (n = n->child))
1.19 schwarze 1646: return(0);
1.1 schwarze 1647:
1.47 schwarze 1648: if (NULL == n->next) {
1649: putkey(mpage, n->string, TYPE_Xr);
1650: return(0);
1651: }
1.1 schwarze 1652:
1.77 schwarze 1653: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1654: putkey(mpage, cp, TYPE_Xr);
1655: free(cp);
1656: return(0);
1.1 schwarze 1657: }
1658:
1.19 schwarze 1659: static int
1.47 schwarze 1660: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1661: {
1662:
1.79 schwarze 1663: if (MDOC_BODY == n->type)
1664: mdoc_deroff(&mpage->desc, n);
1665: return(0);
1.1 schwarze 1666: }
1667:
1.19 schwarze 1668: static int
1.47 schwarze 1669: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1670: {
1671:
1.86 schwarze 1672: if (SEC_NAME == n->sec)
1.90 schwarze 1673: putmdockey(mpage, n->child, NAME_TITLE);
1.86 schwarze 1674: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.90 schwarze 1675: putmdockey(mpage, n->child, NAME_SYN);
1.86 schwarze 1676: return(0);
1.1 schwarze 1677: }
1678:
1.19 schwarze 1679: static int
1.47 schwarze 1680: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1681: {
1682:
1.19 schwarze 1683: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1684: }
1685:
1.47 schwarze 1686: static int
1687: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1688: {
1689:
1.47 schwarze 1690: return(MDOC_HEAD == n->type);
1691: }
1.1 schwarze 1692:
1.47 schwarze 1693: static int
1694: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1695: {
1.1 schwarze 1696:
1.47 schwarze 1697: return(MDOC_BODY == n->type);
1.1 schwarze 1698: }
1699:
1.47 schwarze 1700: /*
1701: * Add a string to the hash table for the current manual.
1702: * Each string has a bitmask telling which macros it belongs to.
1703: * When we finish the manual, we'll dump the table.
1704: */
1.1 schwarze 1705: static void
1.47 schwarze 1706: putkeys(const struct mpage *mpage,
1707: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1708: {
1.90 schwarze 1709: struct ohash *htab;
1.47 schwarze 1710: struct str *s;
1.68 schwarze 1711: const char *end;
1.47 schwarze 1712: unsigned int slot;
1.68 schwarze 1713: int i;
1.1 schwarze 1714:
1.47 schwarze 1715: if (0 == sz)
1716: return;
1.68 schwarze 1717:
1.90 schwarze 1718: if (TYPE_Nm & v) {
1719: htab = &names;
1720: v &= name_mask;
1721: name_mask &= ~NAME_FIRST;
1722: if (debug > 1)
1723: say(mpage->mlinks->file,
1724: "Adding name %*s", sz, cp);
1725: } else {
1726: htab = &strings;
1727: if (debug > 1)
1728: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1729: if ((uint64_t)1 << i & v)
1.90 schwarze 1730: say(mpage->mlinks->file,
1731: "Adding key %s=%*s",
1732: mansearch_keynames[i], sz, cp);
1.68 schwarze 1733: }
1.47 schwarze 1734:
1735: end = cp + sz;
1.90 schwarze 1736: slot = ohash_qlookupi(htab, cp, &end);
1737: s = ohash_find(htab, slot);
1.1 schwarze 1738:
1.47 schwarze 1739: if (NULL != s && mpage == s->mpage) {
1740: s->mask |= v;
1.1 schwarze 1741: return;
1.47 schwarze 1742: } else if (NULL == s) {
1.101 schwarze 1743: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1744: memcpy(s->key, cp, sz);
1.90 schwarze 1745: ohash_insert(htab, slot, s);
1.47 schwarze 1746: }
1747: s->mpage = mpage;
1748: s->mask = v;
1.1 schwarze 1749: }
1750:
1751: /*
1.47 schwarze 1752: * Take a Unicode codepoint and produce its UTF-8 encoding.
1753: * This isn't the best way to do this, but it works.
1754: * The magic numbers are from the UTF-8 packaging.
1755: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1756: */
1.47 schwarze 1757: static size_t
1758: utf8(unsigned int cp, char out[7])
1.1 schwarze 1759: {
1.47 schwarze 1760: size_t rc;
1.1 schwarze 1761:
1.47 schwarze 1762: rc = 0;
1763: if (cp <= 0x0000007F) {
1764: rc = 1;
1765: out[0] = (char)cp;
1766: } else if (cp <= 0x000007FF) {
1767: rc = 2;
1768: out[0] = (cp >> 6 & 31) | 192;
1769: out[1] = (cp & 63) | 128;
1770: } else if (cp <= 0x0000FFFF) {
1771: rc = 3;
1772: out[0] = (cp >> 12 & 15) | 224;
1773: out[1] = (cp >> 6 & 63) | 128;
1774: out[2] = (cp & 63) | 128;
1775: } else if (cp <= 0x001FFFFF) {
1776: rc = 4;
1777: out[0] = (cp >> 18 & 7) | 240;
1778: out[1] = (cp >> 12 & 63) | 128;
1779: out[2] = (cp >> 6 & 63) | 128;
1780: out[3] = (cp & 63) | 128;
1781: } else if (cp <= 0x03FFFFFF) {
1782: rc = 5;
1783: out[0] = (cp >> 24 & 3) | 248;
1784: out[1] = (cp >> 18 & 63) | 128;
1785: out[2] = (cp >> 12 & 63) | 128;
1786: out[3] = (cp >> 6 & 63) | 128;
1787: out[4] = (cp & 63) | 128;
1788: } else if (cp <= 0x7FFFFFFF) {
1789: rc = 6;
1790: out[0] = (cp >> 30 & 1) | 252;
1791: out[1] = (cp >> 24 & 63) | 128;
1792: out[2] = (cp >> 18 & 63) | 128;
1793: out[3] = (cp >> 12 & 63) | 128;
1794: out[4] = (cp >> 6 & 63) | 128;
1795: out[5] = (cp & 63) | 128;
1796: } else
1797: return(0);
1.19 schwarze 1798:
1.47 schwarze 1799: out[rc] = '\0';
1800: return(rc);
1.1 schwarze 1801: }
1802:
1.47 schwarze 1803: /*
1.53 schwarze 1804: * Store the rendered version of a key, or alias the pointer
1805: * if the key contains no escape sequences.
1.47 schwarze 1806: */
1807: static void
1.53 schwarze 1808: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1809: {
1.47 schwarze 1810: size_t sz, bsz, pos;
1.71 schwarze 1811: char utfbuf[7], res[6];
1.47 schwarze 1812: char *buf;
1813: const char *seq, *cpp, *val;
1814: int len, u;
1815: enum mandoc_esc esc;
1816:
1.53 schwarze 1817: assert(NULL == key->rendered);
1.47 schwarze 1818:
1819: res[0] = '\\';
1820: res[1] = '\t';
1821: res[2] = ASCII_NBRSP;
1822: res[3] = ASCII_HYPH;
1.71 schwarze 1823: res[4] = ASCII_BREAK;
1824: res[5] = '\0';
1.1 schwarze 1825:
1.47 schwarze 1826: val = key->key;
1827: bsz = strlen(val);
1.1 schwarze 1828:
1829: /*
1.47 schwarze 1830: * Pre-check: if we have no stop-characters, then set the
1831: * pointer as ourselvse and get out of here.
1.1 schwarze 1832: */
1.47 schwarze 1833: if (strcspn(val, res) == bsz) {
1.53 schwarze 1834: key->rendered = key->key;
1.47 schwarze 1835: return;
1.98 schwarze 1836: }
1.1 schwarze 1837:
1.47 schwarze 1838: /* Pre-allocate by the length of the input */
1.39 schwarze 1839:
1.47 schwarze 1840: buf = mandoc_malloc(++bsz);
1841: pos = 0;
1.39 schwarze 1842:
1.47 schwarze 1843: while ('\0' != *val) {
1844: /*
1845: * Halt on the first escape sequence.
1846: * This also halts on the end of string, in which case
1847: * we just copy, fallthrough, and exit the loop.
1848: */
1849: if ((sz = strcspn(val, res)) > 0) {
1850: memcpy(&buf[pos], val, sz);
1851: pos += sz;
1852: val += sz;
1853: }
1.39 schwarze 1854:
1.71 schwarze 1855: switch (*val) {
1.98 schwarze 1856: case ASCII_HYPH:
1.47 schwarze 1857: buf[pos++] = '-';
1858: val++;
1859: continue;
1.98 schwarze 1860: case '\t':
1.71 schwarze 1861: /* FALLTHROUGH */
1.98 schwarze 1862: case ASCII_NBRSP:
1.47 schwarze 1863: buf[pos++] = ' ';
1864: val++;
1.71 schwarze 1865: /* FALLTHROUGH */
1.98 schwarze 1866: case ASCII_BREAK:
1.47 schwarze 1867: continue;
1.71 schwarze 1868: default:
1869: break;
1870: }
1871: if ('\\' != *val)
1.47 schwarze 1872: break;
1.39 schwarze 1873:
1.47 schwarze 1874: /* Read past the slash. */
1.39 schwarze 1875:
1.47 schwarze 1876: val++;
1.39 schwarze 1877:
1.47 schwarze 1878: /*
1879: * Parse the escape sequence and see if it's a
1880: * predefined character or special character.
1881: */
1.52 schwarze 1882:
1.98 schwarze 1883: esc = mandoc_escape((const char **)&val,
1884: &seq, &len);
1.47 schwarze 1885: if (ESCAPE_ERROR == esc)
1886: break;
1887: if (ESCAPE_SPECIAL != esc)
1888: continue;
1.39 schwarze 1889:
1.47 schwarze 1890: /*
1.52 schwarze 1891: * Render the special character
1892: * as either UTF-8 or ASCII.
1.47 schwarze 1893: */
1.52 schwarze 1894:
1895: if (write_utf8) {
1896: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1897: continue;
1898: cpp = utfbuf;
1899: if (0 == (sz = utf8(u, utfbuf)))
1900: continue;
1901: sz = strlen(cpp);
1902: } else {
1903: cpp = mchars_spec2str(mc, seq, len, &sz);
1904: if (NULL == cpp)
1905: continue;
1906: if (ASCII_NBRSP == *cpp) {
1907: cpp = " ";
1908: sz = 1;
1909: }
1910: }
1.1 schwarze 1911:
1.47 schwarze 1912: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1913:
1.47 schwarze 1914: bsz += sz;
1915: buf = mandoc_realloc(buf, bsz);
1916: memcpy(&buf[pos], cpp, sz);
1917: pos += sz;
1.1 schwarze 1918: }
1919:
1.47 schwarze 1920: buf[pos] = '\0';
1.53 schwarze 1921: key->rendered = buf;
1.1 schwarze 1922: }
1923:
1.75 schwarze 1924: static void
1925: dbadd_mlink(const struct mlink *mlink)
1926: {
1927: size_t i;
1928:
1929: i = 1;
1930: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1931: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1932: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1933: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1934: SQL_STEP(stmts[STMT_INSERT_LINK]);
1935: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.115 schwarze 1936:
1937: i = 1;
1938: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, NAME_FILE);
1939: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
1940: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
1941: SQL_STEP(stmts[STMT_INSERT_NAME]);
1942: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 1943: }
1944:
1.11 schwarze 1945: /*
1.47 schwarze 1946: * Flush the current page's terms (and their bits) into the database.
1947: * Wrap the entire set of additions in a transaction to make sqlite be a
1948: * little faster.
1.53 schwarze 1949: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 1950: */
1951: static void
1.75 schwarze 1952: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1953: {
1.47 schwarze 1954: struct mlink *mlink;
1955: struct str *key;
1956: size_t i;
1957: unsigned int slot;
1958:
1.85 schwarze 1959: mlink = mpage->mlinks;
1.11 schwarze 1960:
1.85 schwarze 1961: if (nodb) {
1.104 schwarze 1962: for (key = ohash_first(&names, &slot); NULL != key;
1963: key = ohash_next(&names, &slot)) {
1964: if (key->rendered != key->key)
1965: free(key->rendered);
1966: free(key);
1967: }
1968: for (key = ohash_first(&strings, &slot); NULL != key;
1969: key = ohash_next(&strings, &slot)) {
1970: if (key->rendered != key->key)
1971: free(key->rendered);
1972: free(key);
1973: }
1.102 schwarze 1974: if (0 == debug)
1975: return;
1.85 schwarze 1976: while (NULL != mlink) {
1977: fputs(mlink->name, stdout);
1978: if (NULL == mlink->next ||
1979: strcmp(mlink->dsec, mlink->next->dsec) ||
1980: strcmp(mlink->fsec, mlink->next->fsec) ||
1981: strcmp(mlink->arch, mlink->next->arch)) {
1982: putchar('(');
1983: if ('\0' == *mlink->dsec)
1984: fputs(mlink->fsec, stdout);
1985: else
1986: fputs(mlink->dsec, stdout);
1987: if ('\0' != *mlink->arch)
1988: printf("/%s", mlink->arch);
1989: putchar(')');
1990: }
1991: mlink = mlink->next;
1992: if (NULL != mlink)
1993: fputs(", ", stdout);
1994: }
1.89 schwarze 1995: printf(" - %s\n", mpage->desc);
1.11 schwarze 1996: return;
1.85 schwarze 1997: }
1998:
1999: if (debug)
2000: say(mlink->file, "Adding to database");
1.47 schwarze 2001:
1.105 schwarze 2002: i = strlen(mpage->desc) + 1;
2003: key = mandoc_calloc(1, sizeof(struct str) + i);
2004: memcpy(key->key, mpage->desc, i);
2005: render_key(mc, key);
2006:
1.47 schwarze 2007: i = 1;
1.105 schwarze 2008: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.116 schwarze 2009: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2010: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2011: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2012: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.105 schwarze 2013:
2014: if (key->rendered != key->key)
2015: free(key->rendered);
2016: free(key);
1.47 schwarze 2017:
1.85 schwarze 2018: while (NULL != mlink) {
1.75 schwarze 2019: dbadd_mlink(mlink);
1.85 schwarze 2020: mlink = mlink->next;
2021: }
1.91 schwarze 2022: mlink = mpage->mlinks;
1.47 schwarze 2023:
1.90 schwarze 2024: for (key = ohash_first(&names, &slot); NULL != key;
2025: key = ohash_next(&names, &slot)) {
2026: assert(key->mpage == mpage);
2027: if (NULL == key->rendered)
2028: render_key(mc, key);
2029: i = 1;
2030: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2031: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2032: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2033: SQL_STEP(stmts[STMT_INSERT_NAME]);
2034: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2035: if (key->rendered != key->key)
2036: free(key->rendered);
2037: free(key);
2038: }
1.47 schwarze 2039: for (key = ohash_first(&strings, &slot); NULL != key;
2040: key = ohash_next(&strings, &slot)) {
2041: assert(key->mpage == mpage);
1.53 schwarze 2042: if (NULL == key->rendered)
2043: render_key(mc, key);
1.47 schwarze 2044: i = 1;
2045: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2046: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2047: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2048: SQL_STEP(stmts[STMT_INSERT_KEY]);
2049: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2050: if (key->rendered != key->key)
2051: free(key->rendered);
1.47 schwarze 2052: free(key);
1.33 schwarze 2053: }
1.47 schwarze 2054: }
1.41 deraadt 2055:
1.47 schwarze 2056: static void
2057: dbprune(void)
2058: {
2059: struct mpage *mpage;
2060: struct mlink *mlink;
2061: size_t i;
2062: unsigned int slot;
1.11 schwarze 2063:
1.63 schwarze 2064: if (0 == nodb)
2065: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2066:
1.63 schwarze 2067: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2068: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2069: mlink = mpage->mlinks;
1.82 schwarze 2070: if (debug)
1.63 schwarze 2071: say(mlink->file, "Deleting from database");
2072: if (nodb)
2073: continue;
2074: for ( ; NULL != mlink; mlink = mlink->next) {
2075: i = 1;
2076: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2077: i, mlink->dsec);
2078: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2079: i, mlink->arch);
2080: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2081: i, mlink->name);
2082: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2083: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2084: }
1.11 schwarze 2085: }
1.63 schwarze 2086:
2087: if (0 == nodb)
2088: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2089: }
1.22 schwarze 2090:
1.47 schwarze 2091: /*
2092: * Close an existing database and its prepared statements.
2093: * If "real" is not set, rename the temporary file into the real one.
2094: */
2095: static void
2096: dbclose(int real)
2097: {
2098: size_t i;
1.72 schwarze 2099: int status;
2100: pid_t child;
1.11 schwarze 2101:
1.47 schwarze 2102: if (nodb)
2103: return;
1.11 schwarze 2104:
1.47 schwarze 2105: for (i = 0; i < STMT__MAX; i++) {
2106: sqlite3_finalize(stmts[i]);
2107: stmts[i] = NULL;
1.28 schwarze 2108: }
1.22 schwarze 2109:
1.47 schwarze 2110: sqlite3_close(db);
2111: db = NULL;
1.11 schwarze 2112:
1.47 schwarze 2113: if (real)
2114: return;
1.22 schwarze 2115:
1.72 schwarze 2116: if ('\0' == *tempfilename) {
2117: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2118: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2119: say(MANDOC_DB, "&rename");
1.72 schwarze 2120: }
2121: return;
2122: }
2123:
2124: switch (child = fork()) {
1.98 schwarze 2125: case -1:
1.72 schwarze 2126: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2127: say("", "&fork cmp");
1.72 schwarze 2128: return;
1.98 schwarze 2129: case 0:
1.72 schwarze 2130: execlp("cmp", "cmp", "-s",
2131: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2132: say("", "&exec cmp");
1.72 schwarze 2133: exit(0);
2134: default:
2135: break;
2136: }
2137: if (-1 == waitpid(child, &status, 0)) {
2138: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2139: say("", "&wait cmp");
1.72 schwarze 2140: } else if (WIFSIGNALED(status)) {
2141: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2142: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2143: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2144: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2145: say(MANDOC_DB,
2146: "Data changed, but cannot replace database");
2147: }
2148:
2149: *strrchr(tempfilename, '/') = '\0';
2150: switch (child = fork()) {
1.98 schwarze 2151: case -1:
1.72 schwarze 2152: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2153: say("", "&fork rm");
1.72 schwarze 2154: return;
1.98 schwarze 2155: case 0:
1.72 schwarze 2156: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2157: say("", "&exec rm");
1.72 schwarze 2158: exit((int)MANDOCLEVEL_SYSERR);
2159: default:
2160: break;
2161: }
2162: if (-1 == waitpid(child, &status, 0)) {
2163: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2164: say("", "&wait rm");
1.72 schwarze 2165: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2166: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2167: say("", "%s: Cannot remove temporary directory",
2168: tempfilename);
1.22 schwarze 2169: }
1.11 schwarze 2170: }
2171:
1.47 schwarze 2172: /*
2173: * This is straightforward stuff.
2174: * Open a database connection to a "temporary" database, then open a set
2175: * of prepared statements we'll use over and over again.
2176: * If "real" is set, we use the existing database; if not, we truncate a
2177: * temporary one.
2178: * Must be matched by dbclose().
2179: */
2180: static int
2181: dbopen(int real)
1.2 schwarze 2182: {
1.72 schwarze 2183: const char *sql;
1.47 schwarze 2184: int rc, ofl;
1.6 schwarze 2185:
1.98 schwarze 2186: if (nodb)
1.47 schwarze 2187: return(1);
1.6 schwarze 2188:
1.72 schwarze 2189: *tempfilename = '\0';
1.47 schwarze 2190: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2191:
2192: if (real) {
2193: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2194: if (SQLITE_OK != rc) {
1.47 schwarze 2195: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2196: if (SQLITE_CANTOPEN != rc)
2197: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2198: return(0);
1.28 schwarze 2199: }
1.72 schwarze 2200: goto prepare_statements;
2201: }
2202:
2203: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2204:
1.72 schwarze 2205: remove(MANDOC_DB "~");
2206: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2207: if (SQLITE_OK == rc)
1.72 schwarze 2208: goto create_tables;
1.73 schwarze 2209: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2211: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2212: return(0);
2213: }
1.6 schwarze 2214:
1.100 schwarze 2215: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2216: sizeof(tempfilename));
1.72 schwarze 2217: if (NULL == mkdtemp(tempfilename)) {
2218: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2219: say("", "&%s", tempfilename);
1.72 schwarze 2220: return(0);
2221: }
1.100 schwarze 2222: (void)strlcat(tempfilename, "/" MANDOC_DB,
2223: sizeof(tempfilename));
1.72 schwarze 2224: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2225: if (SQLITE_OK != rc) {
1.47 schwarze 2226: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2227: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2228: return(0);
1.2 schwarze 2229: }
2230:
1.72 schwarze 2231: create_tables:
1.47 schwarze 2232: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2233: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2234: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2235: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2236: ");\n"
2237: "\n"
2238: "CREATE TABLE \"mlinks\" (\n"
2239: " \"sec\" TEXT NOT NULL,\n"
2240: " \"arch\" TEXT NOT NULL,\n"
2241: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2242: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2243: "ON DELETE CASCADE\n"
1.47 schwarze 2244: ");\n"
1.93 schwarze 2245: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2246: "\n"
1.90 schwarze 2247: "CREATE TABLE \"names\" (\n"
2248: " \"bits\" INTEGER NOT NULL,\n"
2249: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2250: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2251: "ON DELETE CASCADE\n"
2252: ");\n"
2253: "\n"
1.47 schwarze 2254: "CREATE TABLE \"keys\" (\n"
2255: " \"bits\" INTEGER NOT NULL,\n"
2256: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2257: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2258: "ON DELETE CASCADE\n"
1.93 schwarze 2259: ");\n"
2260: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2261:
2262: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2263: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2264: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2265: sqlite3_close(db);
1.47 schwarze 2266: return(0);
1.2 schwarze 2267: }
2268:
1.47 schwarze 2269: prepare_statements:
1.103 schwarze 2270: if (SQLITE_OK != sqlite3_exec(db,
2271: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2272: exitcode = (int)MANDOCLEVEL_SYSERR;
2273: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2274: sqlite3_errmsg(db));
2275: sqlite3_close(db);
2276: return(0);
2277: }
2278:
1.94 schwarze 2279: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2280: "(SELECT pageid FROM mlinks WHERE "
2281: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2282: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2283: sql = "INSERT INTO mpages "
1.89 schwarze 2284: "(desc,form) VALUES (?,?)";
1.47 schwarze 2285: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2286: sql = "INSERT INTO mlinks "
1.61 schwarze 2287: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2288: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2289: sql = "INSERT INTO names "
2290: "(bits,name,pageid) VALUES (?,?,?)";
2291: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2292: sql = "INSERT INTO keys "
2293: "(bits,key,pageid) VALUES (?,?,?)";
2294: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2295:
1.47 schwarze 2296: /*
2297: * When opening a new database, we can turn off
2298: * synchronous mode for much better performance.
2299: */
1.6 schwarze 2300:
1.103 schwarze 2301: if (real && SQLITE_OK != sqlite3_exec(db,
2302: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2303: exitcode = (int)MANDOCLEVEL_SYSERR;
2304: say(MANDOC_DB, "PRAGMA synchronous: %s",
2305: sqlite3_errmsg(db));
2306: sqlite3_close(db);
2307: return(0);
2308: }
1.11 schwarze 2309:
1.47 schwarze 2310: return(1);
2311: }
1.6 schwarze 2312:
1.47 schwarze 2313: static void *
1.106 espie 2314: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2315: {
1.6 schwarze 2316:
1.106 espie 2317: return(mandoc_calloc(n, sz));
1.47 schwarze 2318: }
1.2 schwarze 2319:
1.47 schwarze 2320: static void *
2321: hash_alloc(size_t sz, void *arg)
2322: {
1.28 schwarze 2323:
1.47 schwarze 2324: return(mandoc_malloc(sz));
2325: }
1.6 schwarze 2326:
1.47 schwarze 2327: static void
1.106 espie 2328: hash_free(void *p, void *arg)
1.47 schwarze 2329: {
1.26 schwarze 2330:
1.47 schwarze 2331: free(p);
2332: }
1.6 schwarze 2333:
1.47 schwarze 2334: static int
1.120 schwarze 2335: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2336: {
2337: static char startdir[PATH_MAX];
1.110 schwarze 2338: static int getcwd_status; /* 1 = ok, 2 = failure */
2339: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2340: char *cp;
1.6 schwarze 2341:
1.47 schwarze 2342: /*
1.110 schwarze 2343: * Remember the original working directory, if possible.
2344: * This will be needed if the second or a later directory
2345: * on the command line is given as a relative path.
2346: * Do not error out if the current directory is not
2347: * searchable: Maybe it won't be needed after all.
2348: */
2349: if (0 == getcwd_status) {
2350: if (NULL == getcwd(startdir, sizeof(startdir))) {
2351: getcwd_status = 2;
2352: (void)strlcpy(startdir, strerror(errno),
2353: sizeof(startdir));
2354: } else
2355: getcwd_status = 1;
2356: }
2357:
2358: /*
2359: * We are leaving the old base directory.
2360: * Do not use it any longer, not even for messages.
2361: */
2362: *basedir = '\0';
2363:
2364: /*
2365: * If and only if the directory was changed earlier and
2366: * the next directory to process is given as a relative path,
2367: * first go back, or bail out if that is impossible.
1.47 schwarze 2368: */
1.110 schwarze 2369: if (chdir_status && '/' != *targetdir) {
2370: if (2 == getcwd_status) {
1.47 schwarze 2371: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2372: say("", "getcwd: %s", startdir);
1.47 schwarze 2373: return(0);
2374: }
1.110 schwarze 2375: if (-1 == chdir(startdir)) {
1.47 schwarze 2376: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2377: say("", "&chdir %s", startdir);
1.47 schwarze 2378: return(0);
1.2 schwarze 2379: }
2380: }
1.110 schwarze 2381:
2382: /*
2383: * Always resolve basedir to the canonicalized absolute
2384: * pathname and append a trailing slash, such that
2385: * we can reliably check whether files are inside.
2386: */
1.47 schwarze 2387: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2388: if (report_baddir || errno != ENOENT) {
2389: exitcode = (int)MANDOCLEVEL_BADARG;
2390: say("", "&%s: realpath", targetdir);
2391: }
1.47 schwarze 2392: return(0);
2393: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2394: if (report_baddir || errno != ENOENT) {
2395: exitcode = (int)MANDOCLEVEL_BADARG;
2396: say("", "&chdir");
2397: }
1.47 schwarze 2398: return(0);
1.109 schwarze 2399: }
1.110 schwarze 2400: chdir_status = 1;
1.109 schwarze 2401: cp = strchr(basedir, '\0');
2402: if ('/' != cp[-1]) {
2403: if (cp - basedir >= PATH_MAX - 1) {
2404: exitcode = (int)MANDOCLEVEL_SYSERR;
2405: say("", "Filename too long");
2406: return(0);
2407: }
2408: *cp++ = '/';
2409: *cp = '\0';
1.47 schwarze 2410: }
2411: return(1);
1.2 schwarze 2412: }
2413:
2414: static void
1.47 schwarze 2415: say(const char *file, const char *format, ...)
1.2 schwarze 2416: {
1.47 schwarze 2417: va_list ap;
1.80 schwarze 2418: int use_errno;
1.2 schwarze 2419:
1.47 schwarze 2420: if ('\0' != *basedir)
2421: fprintf(stderr, "%s", basedir);
2422: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2423: fputc('/', stderr);
1.47 schwarze 2424: if ('\0' != *file)
2425: fprintf(stderr, "%s", file);
1.31 schwarze 2426:
1.80 schwarze 2427: use_errno = 1;
2428: if (NULL != format) {
2429: switch (*format) {
1.98 schwarze 2430: case '&':
1.80 schwarze 2431: format++;
2432: break;
1.98 schwarze 2433: case '\0':
1.80 schwarze 2434: format = NULL;
2435: break;
2436: default:
2437: use_errno = 0;
2438: break;
2439: }
2440: }
2441: if (NULL != format) {
2442: if ('\0' != *basedir || '\0' != *file)
2443: fputs(": ", stderr);
2444: va_start(ap, format);
2445: vfprintf(stderr, format, ap);
2446: va_end(ap);
2447: }
2448: if (use_errno) {
2449: if ('\0' != *basedir || '\0' != *file || NULL != format)
2450: fputs(": ", stderr);
1.47 schwarze 2451: perror(NULL);
1.80 schwarze 2452: } else
2453: fputc('\n', stderr);
1.1 schwarze 2454: }