Annotation of mandoc/mandocdb.c, Revision 1.148
1.148 ! schwarze 1: /* $Id: mandocdb.c,v 1.147 2014/05/07 15:10:36 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.95 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 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: */
18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.50 kristaps 22: #include <sys/stat.h>
1.115 schwarze 23: #include <sys/wait.h>
1.1 kristaps 24:
25: #include <assert.h>
1.43 kristaps 26: #include <ctype.h>
1.63 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.50 kristaps 29: #include <fts.h>
1.1 kristaps 30: #include <getopt.h>
1.58 schwarze 31: #include <limits.h>
1.50 kristaps 32: #include <stddef.h>
1.59 schwarze 33: #include <stdio.h>
1.1 kristaps 34: #include <stdint.h>
35: #include <stdlib.h>
36: #include <string.h>
1.17 schwarze 37: #include <unistd.h>
1.1 kristaps 38:
1.53 kristaps 39: #ifdef HAVE_OHASH
1.50 kristaps 40: #include <ohash.h>
1.53 kristaps 41: #else
42: #include "compat_ohash.h"
43: #endif
1.50 kristaps 44: #include <sqlite3.h>
1.1 kristaps 45:
1.50 kristaps 46: #include "mdoc.h"
1.1 kristaps 47: #include "man.h"
48: #include "mandoc.h"
1.119 schwarze 49: #include "mandoc_aux.h"
1.10 kristaps 50: #include "manpath.h"
1.55 kristaps 51: #include "mansearch.h"
1.1 kristaps 52:
1.111 schwarze 53: extern int mansearch_keymax;
54: extern const char *const mansearch_keynames[];
55:
1.52 kristaps 56: #define SQL_EXEC(_v) \
57: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 58: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 59: #define SQL_BIND_TEXT(_s, _i, _v) \
60: if (SQLITE_OK != sqlite3_bind_text \
61: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 62: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 63: #define SQL_BIND_INT(_s, _i, _v) \
64: if (SQLITE_OK != sqlite3_bind_int \
65: ((_s), (_i)++, (_v))) \
1.134 schwarze 66: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 67: #define SQL_BIND_INT64(_s, _i, _v) \
68: if (SQLITE_OK != sqlite3_bind_int64 \
69: ((_s), (_i)++, (_v))) \
1.134 schwarze 70: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 71: #define SQL_STEP(_s) \
72: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 73: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 74:
1.50 kristaps 75: enum op {
76: OP_DEFAULT = 0, /* new dbs from dir list or default config */
77: OP_CONFFILE, /* new databases from custom config file */
78: OP_UPDATE, /* delete/add entries in existing database */
79: OP_DELETE, /* delete entries from existing database */
80: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 81: };
82:
1.50 kristaps 83: enum form {
1.84 schwarze 84: FORM_NONE, /* format is unknown */
85: FORM_SRC, /* format is -man or -mdoc */
86: FORM_CAT /* format is cat */
1.50 kristaps 87: };
1.38 schwarze 88:
1.50 kristaps 89: struct str {
1.96 schwarze 90: char *rendered; /* key in UTF-8 or ASCII form */
1.78 schwarze 91: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 92: uint64_t mask; /* bitmask in sequence */
1.96 schwarze 93: char key[]; /* may contain escape sequences */
1.38 schwarze 94: };
95:
1.79 schwarze 96: struct inodev {
97: ino_t st_ino;
98: dev_t st_dev;
1.50 kristaps 99: };
1.5 kristaps 100:
1.78 schwarze 101: struct mpage {
1.79 schwarze 102: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 103: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 104: enum form form; /* format from file content */
105: char *sec; /* section from file content */
106: char *arch; /* architecture from file content */
107: char *title; /* title from file content */
108: char *desc; /* description from file content */
109: struct mlink *mlinks; /* singly linked list */
110: };
111:
112: struct mlink {
1.58 schwarze 113: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 114: enum form dform; /* format from directory */
1.84 schwarze 115: enum form fform; /* format from file name suffix */
1.82 schwarze 116: char *dsec; /* section from directory */
117: char *arch; /* architecture from directory */
118: char *name; /* name from file name (not empty) */
119: char *fsec; /* section from file name suffix */
1.85 schwarze 120: struct mlink *next; /* singly linked list */
1.118 schwarze 121: struct mpage *mpage; /* parent */
1.124 schwarze 122: int gzip; /* filename has a .gz suffix */
1.3 kristaps 123: };
124:
1.50 kristaps 125: enum stmt {
1.81 schwarze 126: STMT_DELETE_PAGE = 0, /* delete mpage */
127: STMT_INSERT_PAGE, /* insert mpage */
128: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 129: STMT_INSERT_NAME, /* insert name */
1.81 schwarze 130: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 131: STMT__MAX
1.1 kristaps 132: };
133:
1.78 schwarze 134: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 kristaps 135:
1.50 kristaps 136: struct mdoc_handler {
137: mdoc_fp fp; /* optional handler */
138: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 139: };
140:
1.59 schwarze 141: static void dbclose(int);
1.118 schwarze 142: static void dbadd(struct mpage *, struct mchars *);
143: static void dbadd_mlink(const struct mlink *mlink);
1.59 schwarze 144: static int dbopen(int);
145: static void dbprune(void);
146: static void filescan(const char *);
1.50 kristaps 147: static void *hash_alloc(size_t, void *);
148: static void hash_free(void *, size_t, void *);
149: static void *hash_halloc(size_t, void *);
1.84 schwarze 150: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 151: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 152: static void mlink_free(struct mlink *);
1.89 schwarze 153: static void mlinks_undupe(struct mpage *);
1.78 schwarze 154: static void mpages_free(void);
1.101 schwarze 155: static void mpages_merge(struct mchars *, struct mparse *);
1.130 schwarze 156: static void names_check(void);
1.124 schwarze 157: static void parse_cat(struct mpage *, int);
1.78 schwarze 158: static void parse_man(struct mpage *, const struct man_node *);
159: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
160: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
161: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
162: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
163: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
164: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
165: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
166: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
167: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
1.112 schwarze 168: static void putkey(const struct mpage *, char *, uint64_t);
1.78 schwarze 169: static void putkeys(const struct mpage *,
1.64 schwarze 170: const char *, size_t, uint64_t);
1.78 schwarze 171: static void putmdockey(const struct mpage *,
1.50 kristaps 172: const struct mdoc_node *, uint64_t);
1.96 schwarze 173: static void render_key(struct mchars *, struct str *);
1.59 schwarze 174: static void say(const char *, const char *, ...);
1.80 schwarze 175: static int set_basedir(const char *);
1.59 schwarze 176: static int treescan(void);
1.50 kristaps 177: static size_t utf8(unsigned int, char [7]);
178:
1.115 schwarze 179: static char tempfilename[32];
1.50 kristaps 180: static char *progname;
1.102 schwarze 181: static int nodb; /* no database changes */
1.116 schwarze 182: static int mparse_options; /* abort the parse early */
1.141 schwarze 183: static int use_all; /* use all found files */
184: static int debug; /* print what we're doing */
185: static int warnings; /* warn about crap */
1.95 schwarze 186: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 187: static int exitcode; /* to be returned by main */
1.141 schwarze 188: static enum op op; /* operational mode */
1.59 schwarze 189: static char basedir[PATH_MAX]; /* current base directory */
1.78 schwarze 190: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 191: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 192: static struct ohash names; /* table of all names */
1.50 kristaps 193: static struct ohash strings; /* table of all strings */
194: static sqlite3 *db = NULL; /* current database */
195: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 196: static uint64_t name_mask;
1.25 schwarze 197:
198: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 199: { NULL, 0 }, /* Ap */
200: { NULL, 0 }, /* Dd */
201: { NULL, 0 }, /* Dt */
202: { NULL, 0 }, /* Os */
203: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
204: { parse_mdoc_head, TYPE_Ss }, /* Ss */
205: { NULL, 0 }, /* Pp */
206: { NULL, 0 }, /* D1 */
207: { NULL, 0 }, /* Dl */
208: { NULL, 0 }, /* Bd */
209: { NULL, 0 }, /* Ed */
210: { NULL, 0 }, /* Bl */
211: { NULL, 0 }, /* El */
212: { NULL, 0 }, /* It */
213: { NULL, 0 }, /* Ad */
214: { NULL, TYPE_An }, /* An */
215: { NULL, TYPE_Ar }, /* Ar */
216: { NULL, TYPE_Cd }, /* Cd */
217: { NULL, TYPE_Cm }, /* Cm */
218: { NULL, TYPE_Dv }, /* Dv */
219: { NULL, TYPE_Er }, /* Er */
220: { NULL, TYPE_Ev }, /* Ev */
221: { NULL, 0 }, /* Ex */
222: { NULL, TYPE_Fa }, /* Fa */
223: { parse_mdoc_Fd, 0 }, /* Fd */
224: { NULL, TYPE_Fl }, /* Fl */
225: { parse_mdoc_Fn, 0 }, /* Fn */
226: { NULL, TYPE_Ft }, /* Ft */
227: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 228: { NULL, TYPE_In }, /* In */
1.68 schwarze 229: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 230: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 231: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 232: { NULL, 0 }, /* Op */
233: { NULL, 0 }, /* Ot */
234: { NULL, TYPE_Pa }, /* Pa */
235: { NULL, 0 }, /* Rv */
1.92 schwarze 236: { NULL, TYPE_St }, /* St */
1.68 schwarze 237: { NULL, TYPE_Va }, /* Va */
238: { parse_mdoc_body, TYPE_Va }, /* Vt */
239: { parse_mdoc_Xr, 0 }, /* Xr */
240: { NULL, 0 }, /* %A */
241: { NULL, 0 }, /* %B */
242: { NULL, 0 }, /* %D */
243: { NULL, 0 }, /* %I */
244: { NULL, 0 }, /* %J */
245: { NULL, 0 }, /* %N */
246: { NULL, 0 }, /* %O */
247: { NULL, 0 }, /* %P */
248: { NULL, 0 }, /* %R */
249: { NULL, 0 }, /* %T */
250: { NULL, 0 }, /* %V */
251: { NULL, 0 }, /* Ac */
252: { NULL, 0 }, /* Ao */
253: { NULL, 0 }, /* Aq */
254: { NULL, TYPE_At }, /* At */
255: { NULL, 0 }, /* Bc */
256: { NULL, 0 }, /* Bf */
257: { NULL, 0 }, /* Bo */
258: { NULL, 0 }, /* Bq */
259: { NULL, TYPE_Bsx }, /* Bsx */
260: { NULL, TYPE_Bx }, /* Bx */
261: { NULL, 0 }, /* Db */
262: { NULL, 0 }, /* Dc */
263: { NULL, 0 }, /* Do */
264: { NULL, 0 }, /* Dq */
265: { NULL, 0 }, /* Ec */
266: { NULL, 0 }, /* Ef */
267: { NULL, TYPE_Em }, /* Em */
268: { NULL, 0 }, /* Eo */
269: { NULL, TYPE_Fx }, /* Fx */
270: { NULL, TYPE_Ms }, /* Ms */
271: { NULL, 0 }, /* No */
272: { NULL, 0 }, /* Ns */
273: { NULL, TYPE_Nx }, /* Nx */
274: { NULL, TYPE_Ox }, /* Ox */
275: { NULL, 0 }, /* Pc */
276: { NULL, 0 }, /* Pf */
277: { NULL, 0 }, /* Po */
278: { NULL, 0 }, /* Pq */
279: { NULL, 0 }, /* Qc */
280: { NULL, 0 }, /* Ql */
281: { NULL, 0 }, /* Qo */
282: { NULL, 0 }, /* Qq */
283: { NULL, 0 }, /* Re */
284: { NULL, 0 }, /* Rs */
285: { NULL, 0 }, /* Sc */
286: { NULL, 0 }, /* So */
287: { NULL, 0 }, /* Sq */
288: { NULL, 0 }, /* Sm */
289: { NULL, 0 }, /* Sx */
290: { NULL, TYPE_Sy }, /* Sy */
291: { NULL, TYPE_Tn }, /* Tn */
292: { NULL, 0 }, /* Ux */
293: { NULL, 0 }, /* Xc */
294: { NULL, 0 }, /* Xo */
295: { parse_mdoc_head, 0 }, /* Fo */
296: { NULL, 0 }, /* Fc */
297: { NULL, 0 }, /* Oo */
298: { NULL, 0 }, /* Oc */
299: { NULL, 0 }, /* Bk */
300: { NULL, 0 }, /* Ek */
301: { NULL, 0 }, /* Bt */
302: { NULL, 0 }, /* Hf */
303: { NULL, 0 }, /* Fr */
304: { NULL, 0 }, /* Ud */
305: { NULL, TYPE_Lb }, /* Lb */
306: { NULL, 0 }, /* Lp */
307: { NULL, TYPE_Lk }, /* Lk */
308: { NULL, TYPE_Mt }, /* Mt */
309: { NULL, 0 }, /* Brq */
310: { NULL, 0 }, /* Bro */
311: { NULL, 0 }, /* Brc */
312: { NULL, 0 }, /* %C */
313: { NULL, 0 }, /* Es */
314: { NULL, 0 }, /* En */
315: { NULL, TYPE_Dx }, /* Dx */
316: { NULL, 0 }, /* %Q */
317: { NULL, 0 }, /* br */
318: { NULL, 0 }, /* sp */
319: { NULL, 0 }, /* %U */
320: { NULL, 0 }, /* Ta */
1.1 kristaps 321: };
322:
1.141 schwarze 323:
1.1 kristaps 324: int
325: main(int argc, char *argv[])
326: {
1.59 schwarze 327: int ch, i;
1.50 kristaps 328: size_t j, sz;
1.59 schwarze 329: const char *path_arg;
1.50 kristaps 330: struct mchars *mc;
331: struct manpaths dirs;
332: struct mparse *mp;
1.83 schwarze 333: struct ohash_info mpages_info, mlinks_info;
1.50 kristaps 334:
335: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
336: memset(&dirs, 0, sizeof(struct manpaths));
337:
1.83 schwarze 338: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
339: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
340: mpages_info.hfree = mlinks_info.hfree = hash_free;
1.50 kristaps 341:
1.79 schwarze 342: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 343: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 344:
345: progname = strrchr(argv[0], '/');
346: if (progname == NULL)
347: progname = argv[0];
348: else
349: ++progname;
350:
1.50 kristaps 351: /*
1.141 schwarze 352: * We accept a few different invocations.
1.50 kristaps 353: * The CHECKOP macro makes sure that invocation styles don't
354: * clobber each other.
355: */
356: #define CHECKOP(_op, _ch) do \
357: if (OP_DEFAULT != (_op)) { \
358: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
359: goto usage; \
360: } while (/*CONSTCOND*/0)
1.10 kristaps 361:
1.59 schwarze 362: path_arg = NULL;
1.38 schwarze 363: op = OP_DEFAULT;
1.1 kristaps 364:
1.126 schwarze 365: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 366: switch (ch) {
1.141 schwarze 367: case 'a':
1.12 schwarze 368: use_all = 1;
369: break;
1.141 schwarze 370: case 'C':
1.50 kristaps 371: CHECKOP(op, ch);
1.59 schwarze 372: path_arg = optarg;
1.38 schwarze 373: op = OP_CONFFILE;
1.34 schwarze 374: break;
1.141 schwarze 375: case 'D':
1.125 schwarze 376: debug++;
377: break;
1.141 schwarze 378: case 'd':
1.50 kristaps 379: CHECKOP(op, ch);
1.59 schwarze 380: path_arg = optarg;
1.5 kristaps 381: op = OP_UPDATE;
382: break;
1.141 schwarze 383: case 'n':
1.50 kristaps 384: nodb = 1;
385: break;
1.141 schwarze 386: case 'p':
1.126 schwarze 387: warnings = 1;
388: break;
1.141 schwarze 389: case 'Q':
1.116 schwarze 390: mparse_options |= MPARSE_QUICK;
1.102 schwarze 391: break;
1.141 schwarze 392: case 'T':
1.95 schwarze 393: if (strcmp(optarg, "utf8")) {
394: fprintf(stderr, "-T%s: Unsupported "
395: "output format\n", optarg);
396: goto usage;
397: }
398: write_utf8 = 1;
399: break;
1.141 schwarze 400: case 't':
1.50 kristaps 401: CHECKOP(op, ch);
1.38 schwarze 402: dup2(STDOUT_FILENO, STDERR_FILENO);
403: op = OP_TEST;
1.50 kristaps 404: nodb = warnings = 1;
1.38 schwarze 405: break;
1.141 schwarze 406: case 'u':
1.50 kristaps 407: CHECKOP(op, ch);
1.59 schwarze 408: path_arg = optarg;
1.5 kristaps 409: op = OP_DELETE;
410: break;
1.141 schwarze 411: case 'v':
1.126 schwarze 412: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 413: break;
1.1 kristaps 414: default:
1.38 schwarze 415: goto usage;
1.1 kristaps 416: }
417:
418: argc -= optind;
419: argv += optind;
420:
1.38 schwarze 421: if (OP_CONFFILE == op && argc > 0) {
1.50 kristaps 422: fprintf(stderr, "-C: Too many arguments\n");
1.38 schwarze 423: goto usage;
424: }
425:
1.59 schwarze 426: exitcode = (int)MANDOCLEVEL_OK;
1.116 schwarze 427: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.50 kristaps 428: mc = mchars_alloc();
429:
1.78 schwarze 430: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 431: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 432:
433: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 434:
1.50 kristaps 435: /*
436: * All of these deal with a specific directory.
1.140 schwarze 437: * Jump into that directory first.
1.50 kristaps 438: */
1.59 schwarze 439: if (0 == set_basedir(path_arg))
1.50 kristaps 440: goto out;
1.140 schwarze 441:
1.138 schwarze 442: if (dbopen(1)) {
1.140 schwarze 443: /*
444: * The existing database is usable. Process
445: * all files specified on the command-line.
446: */
447: use_all = 1;
1.138 schwarze 448: for (i = 0; i < argc; i++)
449: filescan(argv[i]);
450: if (OP_TEST != op)
451: dbprune();
452: } else {
453: /*
454: * Database missing or corrupt.
455: * Recreate from scratch.
456: */
1.140 schwarze 457: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 458: op = OP_DEFAULT;
459: if (0 == treescan())
460: goto out;
461: if (0 == dbopen(0))
462: goto out;
463: }
1.50 kristaps 464: if (OP_DELETE != op)
1.101 schwarze 465: mpages_merge(mc, mp);
1.138 schwarze 466: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 467: } else {
468: /*
469: * If we have arguments, use them as our manpaths.
470: * If we don't, grok from manpath(1) or however else
471: * manpath_parse() wants to do it.
472: */
473: if (argc > 0) {
1.144 schwarze 474: dirs.paths = mandoc_reallocarray(NULL,
475: argc, sizeof(char *));
1.50 kristaps 476: dirs.sz = (size_t)argc;
477: for (i = 0; i < argc; i++)
478: dirs.paths[i] = mandoc_strdup(argv[i]);
479: } else
1.59 schwarze 480: manpath_parse(&dirs, path_arg, NULL, NULL);
1.127 schwarze 481:
482: if (0 == dirs.sz) {
483: exitcode = (int)MANDOCLEVEL_BADARG;
484: say("", "Empty manpath");
485: }
1.50 kristaps 486:
487: /*
1.69 schwarze 488: * First scan the tree rooted at a base directory, then
489: * build a new database and finally move it into place.
1.50 kristaps 490: * Ignore zero-length directories and strip trailing
491: * slashes.
492: */
493: for (j = 0; j < dirs.sz; j++) {
494: sz = strlen(dirs.paths[j]);
495: if (sz && '/' == dirs.paths[j][sz - 1])
496: dirs.paths[j][--sz] = '\0';
497: if (0 == sz)
498: continue;
1.66 schwarze 499:
500: if (j) {
1.78 schwarze 501: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 502: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 503: }
504:
1.59 schwarze 505: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 506: goto out;
1.59 schwarze 507: if (0 == treescan())
1.50 kristaps 508: goto out;
1.59 schwarze 509: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 510: goto out;
1.59 schwarze 511: if (0 == dbopen(0))
1.50 kristaps 512: goto out;
1.52 kristaps 513:
1.101 schwarze 514: mpages_merge(mc, mp);
1.147 schwarze 515: if (warnings && !nodb &&
1.130 schwarze 516: ! (MPARSE_QUICK & mparse_options))
517: names_check();
1.59 schwarze 518: dbclose(0);
1.66 schwarze 519:
520: if (j + 1 < dirs.sz) {
1.80 schwarze 521: mpages_free();
1.78 schwarze 522: ohash_delete(&mpages);
1.83 schwarze 523: ohash_delete(&mlinks);
1.66 schwarze 524: }
1.50 kristaps 525: }
526: }
527: out:
1.59 schwarze 528: set_basedir(NULL);
1.50 kristaps 529: manpath_free(&dirs);
530: mchars_free(mc);
531: mparse_free(mp);
1.80 schwarze 532: mpages_free();
1.78 schwarze 533: ohash_delete(&mpages);
1.83 schwarze 534: ohash_delete(&mlinks);
1.59 schwarze 535: return(exitcode);
1.50 kristaps 536: usage:
1.126 schwarze 537: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
538: " %s [-aDnpQ] [-Tutf8] dir ...\n"
539: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
540: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 541: " %s [-Q] -t file ...\n",
1.141 schwarze 542: progname, progname, progname,
1.50 kristaps 543: progname, progname);
544:
1.59 schwarze 545: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 546: }
547:
548: /*
1.59 schwarze 549: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 550: * We use fts(), scanning directory parts along the way for clues to our
551: * section and architecture.
552: *
553: * If use_all has been specified, grok all files.
554: * If not, sanitise paths to the following:
555: *
1.141 schwarze 556: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 557: * or
558: * [./]cat<section>[/<arch>]/<name>.0
559: *
560: * TODO: accomodate for multi-language directories.
561: */
562: static int
1.59 schwarze 563: treescan(void)
1.50 kristaps 564: {
1.139 schwarze 565: char buf[PATH_MAX];
1.50 kristaps 566: FTS *f;
567: FTSENT *ff;
1.84 schwarze 568: struct mlink *mlink;
1.124 schwarze 569: int dform, gzip;
1.94 schwarze 570: char *dsec, *arch, *fsec, *cp;
571: const char *path;
1.50 kristaps 572: const char *argv[2];
573:
574: argv[0] = ".";
575: argv[1] = (char *)NULL;
576:
1.139 schwarze 577: f = fts_open((char * const *)argv,
578: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.50 kristaps 579: if (NULL == f) {
1.59 schwarze 580: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 581: say("", "&fts_open");
1.50 kristaps 582: return(0);
583: }
584:
585: dsec = arch = NULL;
586: dform = FORM_NONE;
587:
588: while (NULL != (ff = fts_read(f))) {
589: path = ff->fts_path + 2;
1.139 schwarze 590: switch (ff->fts_info) {
591:
592: /*
593: * Symbolic links require various sanity checks,
594: * then get handled just like regular files.
595: */
1.141 schwarze 596: case FTS_SL:
1.139 schwarze 597: if (NULL == realpath(path, buf)) {
598: if (warnings)
599: say(path, "&realpath");
600: continue;
601: }
602: if (strstr(buf, basedir) != buf) {
603: if (warnings) say("",
604: "%s: outside base directory", buf);
605: continue;
606: }
607: /* Use logical inode to avoid mpages dupe. */
608: if (-1 == stat(path, ff->fts_statp)) {
609: if (warnings)
610: say(path, "&stat");
611: continue;
612: }
613: /* FALLTHROUGH */
614:
1.50 kristaps 615: /*
1.83 schwarze 616: * If we're a regular file, add an mlink by using the
1.50 kristaps 617: * stored directory data and handling the filename.
618: */
1.141 schwarze 619: case FTS_F:
1.60 schwarze 620: if (0 == strcmp(path, MANDOC_DB))
621: continue;
1.50 kristaps 622: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 623: if (warnings)
1.59 schwarze 624: say(path, "Extraneous file");
1.50 kristaps 625: continue;
1.124 schwarze 626: }
627: gzip = 0;
628: fsec = NULL;
629: while (NULL == fsec) {
630: fsec = strrchr(ff->fts_name, '.');
631: if (NULL == fsec || strcmp(fsec+1, "gz"))
632: break;
633: gzip = 1;
634: *fsec = '\0';
635: fsec = NULL;
636: }
637: if (NULL == fsec) {
1.60 schwarze 638: if ( ! use_all) {
1.56 schwarze 639: if (warnings)
1.60 schwarze 640: say(path,
641: "No filename suffix");
1.50 kristaps 642: continue;
643: }
1.84 schwarze 644: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 645: if (warnings)
646: say(path, "Skip html");
647: continue;
1.84 schwarze 648: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 649: if (warnings)
650: say(path, "Skip ps");
651: continue;
1.84 schwarze 652: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 653: if (warnings)
654: say(path, "Skip pdf");
655: continue;
656: } else if ( ! use_all &&
1.84 schwarze 657: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
658: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 659: if (warnings)
660: say(path, "Wrong filename suffix");
661: continue;
1.66 schwarze 662: } else
1.84 schwarze 663: fsec[-1] = '\0';
1.94 schwarze 664:
1.84 schwarze 665: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 666: if (strlcpy(mlink->file, path,
667: sizeof(mlink->file)) >=
668: sizeof(mlink->file)) {
669: say(path, "Filename too long");
670: free(mlink);
671: continue;
672: }
1.84 schwarze 673: mlink->dform = dform;
1.94 schwarze 674: mlink->dsec = dsec;
675: mlink->arch = arch;
676: mlink->name = ff->fts_name;
677: mlink->fsec = fsec;
1.124 schwarze 678: mlink->gzip = gzip;
1.84 schwarze 679: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 680: continue;
1.139 schwarze 681:
1.141 schwarze 682: case FTS_D:
1.139 schwarze 683: /* FALLTHROUGH */
1.141 schwarze 684: case FTS_DP:
1.139 schwarze 685: break;
686:
687: default:
1.60 schwarze 688: if (warnings)
689: say(path, "Not a regular file");
1.50 kristaps 690: continue;
1.60 schwarze 691: }
1.1 kristaps 692:
1.50 kristaps 693: switch (ff->fts_level) {
1.141 schwarze 694: case 0:
1.50 kristaps 695: /* Ignore the root directory. */
696: break;
1.141 schwarze 697: case 1:
1.50 kristaps 698: /*
699: * This might contain manX/ or catX/.
700: * Try to infer this from the name.
701: * If we're not in use_all, enforce it.
702: */
703: cp = ff->fts_name;
704: if (FTS_DP == ff->fts_info)
705: break;
1.1 kristaps 706:
1.50 kristaps 707: if (0 == strncmp(cp, "man", 3)) {
708: dform = FORM_SRC;
1.66 schwarze 709: dsec = cp + 3;
1.50 kristaps 710: } else if (0 == strncmp(cp, "cat", 3)) {
711: dform = FORM_CAT;
1.66 schwarze 712: dsec = cp + 3;
1.94 schwarze 713: } else {
714: dform = FORM_NONE;
715: dsec = NULL;
1.50 kristaps 716: }
1.1 kristaps 717:
1.141 schwarze 718: if (NULL != dsec || use_all)
1.50 kristaps 719: break;
1.1 kristaps 720:
1.56 schwarze 721: if (warnings)
1.59 schwarze 722: say(path, "Unknown directory part");
1.50 kristaps 723: fts_set(f, ff, FTS_SKIP);
724: break;
1.141 schwarze 725: case 2:
1.50 kristaps 726: /*
727: * Possibly our architecture.
728: * If we're descending, keep tabs on it.
729: */
730: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 731: arch = ff->fts_name;
1.94 schwarze 732: else
733: arch = NULL;
1.50 kristaps 734: break;
735: default:
736: if (FTS_DP == ff->fts_info || use_all)
737: break;
1.56 schwarze 738: if (warnings)
1.59 schwarze 739: say(path, "Extraneous directory part");
1.50 kristaps 740: fts_set(f, ff, FTS_SKIP);
741: break;
1.5 kristaps 742: }
1.50 kristaps 743: }
744:
745: fts_close(f);
746: return(1);
747: }
1.5 kristaps 748:
1.50 kristaps 749: /*
1.89 schwarze 750: * Add a file to the mlinks table.
1.50 kristaps 751: * Do not verify that it's a "valid" looking manpage (we'll do that
752: * later).
753: *
754: * Try to infer the manual section, architecture, and page name from the
755: * path, assuming it looks like
756: *
1.141 schwarze 757: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 758: * or
759: * [./]cat<section>[/<arch>]/<name>.0
760: *
761: * See treescan() for the fts(3) version of this.
762: */
763: static void
1.59 schwarze 764: filescan(const char *file)
1.50 kristaps 765: {
1.59 schwarze 766: char buf[PATH_MAX];
1.84 schwarze 767: struct stat st;
768: struct mlink *mlink;
1.59 schwarze 769: char *p, *start;
1.5 kristaps 770:
1.50 kristaps 771: assert(use_all);
1.5 kristaps 772:
1.50 kristaps 773: if (0 == strncmp(file, "./", 2))
774: file += 2;
1.5 kristaps 775:
1.139 schwarze 776: /*
777: * We have to do lstat(2) before realpath(3) loses
778: * the information whether this is a symbolic link.
779: * We need to know that because for symbolic links,
780: * we want to use the orginal file name, while for
781: * regular files, we want to use the real path.
782: */
783: if (-1 == lstat(file, &st)) {
784: exitcode = (int)MANDOCLEVEL_BADARG;
785: say(file, "&lstat");
786: return;
787: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
788: exitcode = (int)MANDOCLEVEL_BADARG;
789: say(file, "Not a regular file");
790: return;
791: }
792:
793: /*
794: * We have to resolve the file name to the real path
795: * in any case for the base directory check.
796: */
1.59 schwarze 797: if (NULL == realpath(file, buf)) {
798: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 799: say(file, "&realpath");
1.59 schwarze 800: return;
1.106 schwarze 801: }
802:
803: if (strstr(buf, basedir) == buf)
804: start = buf + strlen(basedir) + 1;
805: else if (OP_TEST == op)
806: start = buf;
807: else {
1.59 schwarze 808: exitcode = (int)MANDOCLEVEL_BADARG;
809: say("", "%s: outside base directory", buf);
810: return;
1.106 schwarze 811: }
812:
1.139 schwarze 813: /*
814: * Now we are sure the file is inside our tree.
815: * If it is a symbolic link, ignore the real path
816: * and use the original name.
817: * This implies passing stuff like "cat1/../man1/foo.1"
818: * on the command line won't work. So don't do that.
819: * Note the stat(2) can still fail if the link target
820: * doesn't exist.
821: */
822: if (S_IFLNK & st.st_mode) {
823: if (-1 == stat(buf, &st)) {
824: exitcode = (int)MANDOCLEVEL_BADARG;
825: say(file, "&stat");
826: return;
827: }
1.143 schwarze 828: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
829: say(file, "Filename too long");
830: return;
831: }
1.139 schwarze 832: start = strstr(buf, basedir) == buf ?
833: buf + strlen(basedir) + 1 : buf;
1.50 kristaps 834: }
1.106 schwarze 835:
1.84 schwarze 836: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 837: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
838: sizeof(mlink->file)) {
839: say(start, "Filename too long");
840: return;
841: }
1.17 schwarze 842:
1.50 kristaps 843: /*
844: * First try to guess our directory structure.
845: * If we find a separator, try to look for man* or cat*.
846: * If we find one of these and what's underneath is a directory,
847: * assume it's an architecture.
848: */
849: if (NULL != (p = strchr(start, '/'))) {
850: *p++ = '\0';
851: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 852: mlink->dform = FORM_SRC;
1.94 schwarze 853: mlink->dsec = start + 3;
1.50 kristaps 854: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 855: mlink->dform = FORM_CAT;
1.94 schwarze 856: mlink->dsec = start + 3;
1.17 schwarze 857: }
1.5 kristaps 858:
1.50 kristaps 859: start = p;
1.84 schwarze 860: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 861: *p++ = '\0';
1.94 schwarze 862: mlink->arch = start;
1.50 kristaps 863: start = p;
1.84 schwarze 864: }
1.50 kristaps 865: }
866:
867: /*
868: * Now check the file suffix.
869: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
870: */
871: p = strrchr(start, '\0');
872: while (p-- > start && '/' != *p && '.' != *p)
873: /* Loop. */ ;
874:
875: if ('.' == *p) {
876: *p++ = '\0';
1.94 schwarze 877: mlink->fsec = p;
1.5 kristaps 878: }
879:
1.10 kristaps 880: /*
1.50 kristaps 881: * Now try to parse the name.
882: * Use the filename portion of the path.
1.10 kristaps 883: */
1.84 schwarze 884: mlink->name = start;
1.50 kristaps 885: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 886: mlink->name = p + 1;
1.50 kristaps 887: *p = '\0';
1.84 schwarze 888: }
889: mlink_add(mlink, &st);
1.50 kristaps 890: }
1.5 kristaps 891:
1.50 kristaps 892: static void
1.84 schwarze 893: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 894: {
1.83 schwarze 895: struct inodev inodev;
1.78 schwarze 896: struct mpage *mpage;
1.83 schwarze 897: unsigned int slot;
1.50 kristaps 898:
1.84 schwarze 899: assert(NULL != mlink->file);
1.50 kristaps 900:
1.94 schwarze 901: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
902: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
903: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
904: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 905:
906: if ('0' == *mlink->fsec) {
907: free(mlink->fsec);
908: mlink->fsec = mandoc_strdup(mlink->dsec);
909: mlink->fform = FORM_CAT;
910: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
911: mlink->fform = FORM_SRC;
1.74 schwarze 912: else
1.84 schwarze 913: mlink->fform = FORM_NONE;
1.82 schwarze 914:
1.83 schwarze 915: slot = ohash_qlookup(&mlinks, mlink->file);
916: assert(NULL == ohash_find(&mlinks, slot));
917: ohash_insert(&mlinks, slot, mlink);
918:
919: inodev.st_ino = st->st_ino;
920: inodev.st_dev = st->st_dev;
921: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
922: sizeof(struct inodev), inodev.st_ino);
923: mpage = ohash_find(&mpages, slot);
924: if (NULL == mpage) {
925: mpage = mandoc_calloc(1, sizeof(struct mpage));
926: mpage->inodev.st_ino = inodev.st_ino;
927: mpage->inodev.st_dev = inodev.st_dev;
928: ohash_insert(&mpages, slot, mpage);
929: } else
1.85 schwarze 930: mlink->next = mpage->mlinks;
1.82 schwarze 931: mpage->mlinks = mlink;
1.118 schwarze 932: mlink->mpage = mpage;
1.82 schwarze 933: }
934:
935: static void
936: mlink_free(struct mlink *mlink)
937: {
938:
939: free(mlink->dsec);
940: free(mlink->arch);
941: free(mlink->name);
942: free(mlink->fsec);
943: free(mlink);
1.50 kristaps 944: }
1.3 kristaps 945:
1.50 kristaps 946: static void
1.78 schwarze 947: mpages_free(void)
1.50 kristaps 948: {
1.78 schwarze 949: struct mpage *mpage;
1.82 schwarze 950: struct mlink *mlink;
1.80 schwarze 951: unsigned int slot;
1.3 kristaps 952:
1.80 schwarze 953: mpage = ohash_first(&mpages, &slot);
954: while (NULL != mpage) {
1.82 schwarze 955: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 956: mpage->mlinks = mlink->next;
1.82 schwarze 957: mlink_free(mlink);
958: }
1.78 schwarze 959: free(mpage->sec);
960: free(mpage->arch);
1.82 schwarze 961: free(mpage->title);
962: free(mpage->desc);
1.78 schwarze 963: free(mpage);
1.80 schwarze 964: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 965: }
966: }
1.38 schwarze 967:
1.50 kristaps 968: /*
1.89 schwarze 969: * For each mlink to the mpage, check whether the path looks like
970: * it is formatted, and if it does, check whether a source manual
971: * exists by the same name, ignoring the suffix.
972: * If both conditions hold, drop the mlink.
973: */
974: static void
975: mlinks_undupe(struct mpage *mpage)
976: {
977: char buf[PATH_MAX];
978: struct mlink **prev;
979: struct mlink *mlink;
980: char *bufp;
981:
982: mpage->form = FORM_CAT;
1.90 schwarze 983: prev = &mpage->mlinks;
984: while (NULL != (mlink = *prev)) {
1.89 schwarze 985: if (FORM_CAT != mlink->dform) {
986: mpage->form = FORM_NONE;
1.90 schwarze 987: goto nextlink;
1.89 schwarze 988: }
1.143 schwarze 989: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 990: bufp = strstr(buf, "cat");
991: assert(NULL != bufp);
992: memcpy(bufp, "man", 3);
993: if (NULL != (bufp = strrchr(buf, '.')))
994: *++bufp = '\0';
1.143 schwarze 995: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 996: if (NULL == ohash_find(&mlinks,
1.141 schwarze 997: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 998: goto nextlink;
1.89 schwarze 999: if (warnings)
1000: say(mlink->file, "Man source exists: %s", buf);
1001: if (use_all)
1.90 schwarze 1002: goto nextlink;
1.89 schwarze 1003: *prev = mlink->next;
1004: mlink_free(mlink);
1.90 schwarze 1005: continue;
1006: nextlink:
1007: prev = &(*prev)->next;
1.89 schwarze 1008: }
1009: }
1010:
1.131 schwarze 1011: static void
1.93 schwarze 1012: mlink_check(struct mpage *mpage, struct mlink *mlink)
1013: {
1.131 schwarze 1014: struct str *str;
1015: unsigned int slot;
1.93 schwarze 1016:
1017: /*
1018: * Check whether the manual section given in a file
1019: * agrees with the directory where the file is located.
1020: * Some manuals have suffixes like (3p) on their
1021: * section number either inside the file or in the
1022: * directory name, some are linked into more than one
1023: * section, like encrypt(1) = makekey(8).
1024: */
1025:
1026: if (FORM_SRC == mpage->form &&
1.131 schwarze 1027: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1028: say(mlink->file, "Section \"%s\" manual in %s directory",
1029: mpage->sec, mlink->dsec);
1030:
1031: /*
1032: * Manual page directories exist for each kernel
1033: * architecture as returned by machine(1).
1034: * However, many manuals only depend on the
1035: * application architecture as returned by arch(1).
1036: * For example, some (2/ARM) manuals are shared
1037: * across the "armish" and "zaurus" kernel
1038: * architectures.
1039: * A few manuals are even shared across completely
1040: * different architectures, for example fdformat(1)
1041: * on amd64, i386, sparc, and sparc64.
1042: */
1043:
1.131 schwarze 1044: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1045: say(mlink->file, "Architecture \"%s\" manual in "
1046: "\"%s\" directory", mpage->arch, mlink->arch);
1047:
1.131 schwarze 1048: /*
1049: * XXX
1.133 schwarze 1050: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1051: */
1052:
1053: if (FORM_CAT == mpage->form)
1054: return;
1055:
1056: /*
1057: * Check whether this mlink
1058: * appears as a name in the NAME section.
1059: */
1.93 schwarze 1060:
1.133 schwarze 1061: slot = ohash_qlookup(&names, mlink->name);
1062: str = ohash_find(&names, slot);
1.131 schwarze 1063: assert(NULL != str);
1.133 schwarze 1064: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1065: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1066: }
1067:
1.89 schwarze 1068: /*
1.80 schwarze 1069: * Run through the files in the global vector "mpages"
1070: * and add them to the database specified in "basedir".
1.50 kristaps 1071: *
1072: * This handles the parsing scheme itself, using the cues of directory
1073: * and filename to determine whether the file is parsable or not.
1074: */
1.59 schwarze 1075: static void
1.101 schwarze 1076: mpages_merge(struct mchars *mc, struct mparse *mp)
1.50 kristaps 1077: {
1.113 schwarze 1078: char any[] = "any";
1.101 schwarze 1079: struct ohash_info str_info;
1.124 schwarze 1080: int fd[2];
1.118 schwarze 1081: struct mpage *mpage, *mpage_dest;
1082: struct mlink *mlink, *mlink_dest;
1.69 schwarze 1083: struct mdoc *mdoc;
1084: struct man *man;
1.118 schwarze 1085: char *sodest;
1.112 schwarze 1086: char *cp;
1.124 schwarze 1087: pid_t child_pid;
1.131 schwarze 1088: int status;
1.101 schwarze 1089: unsigned int pslot;
1.69 schwarze 1090: enum mandoclevel lvl;
1091:
1.71 schwarze 1092: str_info.alloc = hash_alloc;
1093: str_info.halloc = hash_halloc;
1094: str_info.hfree = hash_free;
1095: str_info.key_offset = offsetof(struct str, key);
1096:
1.107 schwarze 1097: if (0 == nodb)
1098: SQL_EXEC("BEGIN TRANSACTION");
1099:
1.80 schwarze 1100: mpage = ohash_first(&mpages, &pslot);
1101: while (NULL != mpage) {
1.89 schwarze 1102: mlinks_undupe(mpage);
1103: if (NULL == mpage->mlinks) {
1104: mpage = ohash_next(&mpages, &pslot);
1105: continue;
1.50 kristaps 1106: }
1107:
1.133 schwarze 1108: name_mask = NAME_MASK;
1109: ohash_init(&names, 4, &str_info);
1.71 schwarze 1110: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1111: mparse_reset(mp);
1112: mdoc = NULL;
1113: man = NULL;
1.124 schwarze 1114: sodest = NULL;
1115: child_pid = 0;
1116: fd[0] = -1;
1117: fd[1] = -1;
1118:
1119: if (mpage->mlinks->gzip) {
1120: if (-1 == pipe(fd)) {
1121: exitcode = (int)MANDOCLEVEL_SYSERR;
1122: say(mpage->mlinks->file, "&pipe gunzip");
1123: goto nextpage;
1124: }
1125: switch (child_pid = fork()) {
1.141 schwarze 1126: case -1:
1.124 schwarze 1127: exitcode = (int)MANDOCLEVEL_SYSERR;
1128: say(mpage->mlinks->file, "&fork gunzip");
1129: child_pid = 0;
1130: close(fd[1]);
1131: close(fd[0]);
1132: goto nextpage;
1.141 schwarze 1133: case 0:
1.124 schwarze 1134: close(fd[0]);
1135: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1136: say(mpage->mlinks->file,
1137: "&dup gunzip");
1138: exit(1);
1139: }
1140: execlp("gunzip", "gunzip", "-c",
1141: mpage->mlinks->file, NULL);
1142: say(mpage->mlinks->file, "&exec gunzip");
1143: exit(1);
1144: default:
1145: close(fd[1]);
1146: break;
1147: }
1148: }
1.14 schwarze 1149:
1150: /*
1.33 schwarze 1151: * Try interpreting the file as mdoc(7) or man(7)
1152: * source code, unless it is already known to be
1153: * formatted. Fall back to formatted mode.
1.14 schwarze 1154: */
1.82 schwarze 1155: if (FORM_CAT != mpage->mlinks->dform ||
1.84 schwarze 1156: FORM_CAT != mpage->mlinks->fform) {
1.124 schwarze 1157: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.50 kristaps 1158: if (lvl < MANDOCLEVEL_FATAL)
1.118 schwarze 1159: mparse_result(mp, &mdoc, &man, &sodest);
1.76 schwarze 1160: }
1.14 schwarze 1161:
1.118 schwarze 1162: if (NULL != sodest) {
1163: mlink_dest = ohash_find(&mlinks,
1164: ohash_qlookup(&mlinks, sodest));
1165: if (NULL != mlink_dest) {
1166:
1167: /* The .so target exists. */
1168:
1169: mpage_dest = mlink_dest->mpage;
1170: mlink = mpage->mlinks;
1171: while (1) {
1172: mlink->mpage = mpage_dest;
1173:
1174: /*
1175: * If the target was already
1176: * processed, add the links
1177: * to the database now.
1178: * Otherwise, this will
1179: * happen when we come
1180: * to the target.
1181: */
1182:
1.137 schwarze 1183: if (mpage_dest->pageid)
1.118 schwarze 1184: dbadd_mlink(mlink);
1185:
1186: if (NULL == mlink->next)
1187: break;
1188: mlink = mlink->next;
1189: }
1190:
1191: /* Move all links to the target. */
1192:
1193: mlink->next = mlink_dest->next;
1194: mlink_dest->next = mpage->mlinks;
1195: mpage->mlinks = NULL;
1196: }
1.124 schwarze 1197: goto nextpage;
1.118 schwarze 1198: } else if (NULL != mdoc) {
1.82 schwarze 1199: mpage->form = FORM_SRC;
1200: mpage->sec =
1201: mandoc_strdup(mdoc_meta(mdoc)->msec);
1202: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 1203: mpage->arch = mandoc_strdup(
1204: NULL == mpage->arch ? "" : mpage->arch);
1.82 schwarze 1205: mpage->title =
1206: mandoc_strdup(mdoc_meta(mdoc)->title);
1.14 schwarze 1207: } else if (NULL != man) {
1.82 schwarze 1208: mpage->form = FORM_SRC;
1209: mpage->sec =
1210: mandoc_strdup(man_meta(man)->msec);
1211: mpage->arch =
1212: mandoc_strdup(mpage->mlinks->arch);
1213: mpage->title =
1214: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1215: } else {
1.82 schwarze 1216: mpage->form = FORM_CAT;
1217: mpage->sec =
1218: mandoc_strdup(mpage->mlinks->dsec);
1219: mpage->arch =
1220: mandoc_strdup(mpage->mlinks->arch);
1221: mpage->title =
1222: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1223: }
1.97 schwarze 1224: putkey(mpage, mpage->sec, TYPE_sec);
1.98 schwarze 1225: putkey(mpage, '\0' == *mpage->arch ?
1.113 schwarze 1226: any : mpage->arch, TYPE_arch);
1.50 kristaps 1227:
1.97 schwarze 1228: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1229: if ('\0' != *mlink->dsec)
1230: putkey(mpage, mlink->dsec, TYPE_sec);
1231: if ('\0' != *mlink->fsec)
1232: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1233: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1234: any : mlink->arch, TYPE_arch);
1.133 schwarze 1235: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1236: }
1.12 schwarze 1237:
1.135 schwarze 1238: assert(NULL == mpage->desc);
1.50 kristaps 1239: if (NULL != mdoc) {
1240: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.133 schwarze 1241: putkey(mpage, cp, NAME_HEAD);
1.78 schwarze 1242: parse_mdoc(mpage, mdoc_node(mdoc));
1.50 kristaps 1243: } else if (NULL != man)
1.78 schwarze 1244: parse_man(mpage, man_node(man));
1.50 kristaps 1245: else
1.124 schwarze 1246: parse_cat(mpage, fd[0]);
1.135 schwarze 1247: if (NULL == mpage->desc)
1248: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1249:
1250: if (warnings && !use_all)
1251: for (mlink = mpage->mlinks; mlink;
1252: mlink = mlink->next)
1253: mlink_check(mpage, mlink);
1.44 kristaps 1254:
1.105 schwarze 1255: dbadd(mpage, mc);
1.124 schwarze 1256:
1257: nextpage:
1258: if (child_pid) {
1259: if (-1 == waitpid(child_pid, &status, 0)) {
1260: exitcode = (int)MANDOCLEVEL_SYSERR;
1261: say(mpage->mlinks->file, "&wait gunzip");
1262: } else if (WIFSIGNALED(status)) {
1263: exitcode = (int)MANDOCLEVEL_SYSERR;
1264: say(mpage->mlinks->file,
1265: "gunzip died from signal %d",
1266: WTERMSIG(status));
1267: } else if (WEXITSTATUS(status)) {
1268: exitcode = (int)MANDOCLEVEL_SYSERR;
1269: say(mpage->mlinks->file,
1270: "gunzip failed with code %d",
1271: WEXITSTATUS(status));
1272: }
1273: }
1.66 schwarze 1274: ohash_delete(&strings);
1.133 schwarze 1275: ohash_delete(&names);
1.80 schwarze 1276: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1277: }
1.107 schwarze 1278:
1279: if (0 == nodb)
1280: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1281: }
1282:
1283: static void
1284: names_check(void)
1285: {
1286: sqlite3_stmt *stmt;
1287: const char *name, *sec, *arch, *key;
1288: int irc;
1289:
1290: sqlite3_prepare_v2(db,
1291: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1292: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1293: "WHERE bits & ? AND NOT EXISTS ("
1294: "SELECT pageid FROM mlinks "
1.133 schwarze 1295: "WHERE mlinks.pageid == names.pageid "
1296: "AND mlinks.name == names.name"
1.130 schwarze 1297: ")"
1298: ") JOIN ("
1.142 schwarze 1299: "SELECT sec, arch, name, pageid FROM mlinks "
1300: "GROUP BY pageid"
1.130 schwarze 1301: ") USING (pageid);",
1302: -1, &stmt, NULL);
1303:
1.134 schwarze 1304: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1305: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1306:
1307: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1308: name = sqlite3_column_text(stmt, 0);
1309: sec = sqlite3_column_text(stmt, 1);
1310: arch = sqlite3_column_text(stmt, 2);
1311: key = sqlite3_column_text(stmt, 3);
1312: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1313: '\0' == *arch ? "" : "/",
1314: '\0' == *arch ? "" : arch, key);
1315: }
1316: sqlite3_finalize(stmt);
1.50 kristaps 1317: }
1.12 schwarze 1318:
1.50 kristaps 1319: static void
1.124 schwarze 1320: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1321: {
1322: FILE *stream;
1323: char *line, *p, *title;
1324: size_t len, plen, titlesz;
1.12 schwarze 1325:
1.124 schwarze 1326: stream = (-1 == fd) ?
1327: fopen(mpage->mlinks->file, "r") :
1328: fdopen(fd, "r");
1329: if (NULL == stream) {
1.56 schwarze 1330: if (warnings)
1.123 schwarze 1331: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1332: return;
1333: }
1.1 kristaps 1334:
1.50 kristaps 1335: /* Skip to first blank line. */
1.1 kristaps 1336:
1.50 kristaps 1337: while (NULL != (line = fgetln(stream, &len)))
1338: if ('\n' == *line)
1339: break;
1.1 kristaps 1340:
1.50 kristaps 1341: /*
1342: * Assume the first line that is not indented
1343: * is the first section header. Skip to it.
1344: */
1.1 kristaps 1345:
1.50 kristaps 1346: while (NULL != (line = fgetln(stream, &len)))
1347: if ('\n' != *line && ' ' != *line)
1348: break;
1.141 schwarze 1349:
1.50 kristaps 1350: /*
1351: * Read up until the next section into a buffer.
1352: * Strip the leading and trailing newline from each read line,
1353: * appending a trailing space.
1354: * Ignore empty (whitespace-only) lines.
1355: */
1.1 kristaps 1356:
1.50 kristaps 1357: titlesz = 0;
1358: title = NULL;
1.38 schwarze 1359:
1.50 kristaps 1360: while (NULL != (line = fgetln(stream, &len))) {
1361: if (' ' != *line || '\n' != line[len - 1])
1362: break;
1363: while (len > 0 && isspace((unsigned char)*line)) {
1364: line++;
1365: len--;
1366: }
1367: if (1 == len)
1.38 schwarze 1368: continue;
1.50 kristaps 1369: title = mandoc_realloc(title, titlesz + len);
1370: memcpy(title + titlesz, line, len);
1371: titlesz += len;
1372: title[titlesz - 1] = ' ';
1373: }
1.38 schwarze 1374:
1.50 kristaps 1375: /*
1376: * If no page content can be found, or the input line
1377: * is already the next section header, or there is no
1378: * trailing newline, reuse the page title as the page
1379: * description.
1380: */
1.44 kristaps 1381:
1.50 kristaps 1382: if (NULL == title || '\0' == *title) {
1.56 schwarze 1383: if (warnings)
1.82 schwarze 1384: say(mpage->mlinks->file,
1385: "Cannot find NAME section");
1.50 kristaps 1386: fclose(stream);
1387: free(title);
1388: return;
1389: }
1.1 kristaps 1390:
1.50 kristaps 1391: title = mandoc_realloc(title, titlesz + 1);
1392: title[titlesz] = '\0';
1.33 schwarze 1393:
1.50 kristaps 1394: /*
1395: * Skip to the first dash.
1396: * Use the remaining line as the description (no more than 70
1397: * bytes).
1398: */
1.33 schwarze 1399:
1.50 kristaps 1400: if (NULL != (p = strstr(title, "- "))) {
1401: for (p += 2; ' ' == *p || '\b' == *p; p++)
1402: /* Skip to next word. */ ;
1403: } else {
1.56 schwarze 1404: if (warnings)
1.82 schwarze 1405: say(mpage->mlinks->file,
1406: "No dash in title line");
1.50 kristaps 1407: p = title;
1408: }
1.38 schwarze 1409:
1.50 kristaps 1410: plen = strlen(p);
1.1 kristaps 1411:
1.50 kristaps 1412: /* Strip backspace-encoding from line. */
1.1 kristaps 1413:
1.50 kristaps 1414: while (NULL != (line = memchr(p, '\b', plen))) {
1415: len = line - p;
1416: if (0 == len) {
1417: memmove(line, line + 1, plen--);
1418: continue;
1.141 schwarze 1419: }
1.50 kristaps 1420: memmove(line - 1, line + 1, plen - len);
1421: plen -= 2;
1422: }
1.1 kristaps 1423:
1.78 schwarze 1424: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1425: fclose(stream);
1426: free(title);
1427: }
1.1 kristaps 1428:
1.50 kristaps 1429: /*
1430: * Put a type/word pair into the word database for this particular file.
1431: */
1432: static void
1.112 schwarze 1433: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1434: {
1.112 schwarze 1435: char *cp;
1.18 kristaps 1436:
1.50 kristaps 1437: assert(NULL != value);
1.112 schwarze 1438: if (TYPE_arch == type)
1439: for (cp = value; *cp; cp++)
1440: if (isupper((unsigned char)*cp))
1441: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1442: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1443: }
1444:
1445: /*
1.50 kristaps 1446: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1447: */
1448: static void
1.78 schwarze 1449: putmdockey(const struct mpage *mpage,
1450: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1451: {
1.18 kristaps 1452:
1.50 kristaps 1453: for ( ; NULL != n; n = n->next) {
1454: if (NULL != n->child)
1.78 schwarze 1455: putmdockey(mpage, n->child, m);
1.50 kristaps 1456: if (MDOC_TEXT == n->type)
1.78 schwarze 1457: putkey(mpage, n->string, m);
1.50 kristaps 1458: }
1459: }
1.18 kristaps 1460:
1.61 schwarze 1461: static void
1.78 schwarze 1462: parse_man(struct mpage *mpage, const struct man_node *n)
1.50 kristaps 1463: {
1464: const struct man_node *head, *body;
1.121 schwarze 1465: char *start, *title;
1.50 kristaps 1466: char byte;
1.121 schwarze 1467: size_t sz;
1.18 kristaps 1468:
1.50 kristaps 1469: if (NULL == n)
1.61 schwarze 1470: return;
1.18 kristaps 1471:
1.50 kristaps 1472: /*
1473: * We're only searching for one thing: the first text child in
1474: * the BODY of a NAME section. Since we don't keep track of
1475: * sections in -man, run some hoops to find out whether we're in
1476: * the correct section or not.
1477: */
1.18 kristaps 1478:
1.50 kristaps 1479: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1480: body = n;
1481: assert(body->parent);
1482: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1483: 1 == head->nchild &&
1484: NULL != (head = (head->child)) &&
1485: MAN_TEXT == head->type &&
1486: 0 == strcmp(head->string, "NAME") &&
1487: NULL != body->child) {
1.3 kristaps 1488:
1.50 kristaps 1489: /*
1490: * Suck the entire NAME section into memory.
1491: * Yes, we might run away.
1492: * But too many manuals have big, spread-out
1493: * NAME sections over many lines.
1494: */
1.3 kristaps 1495:
1.121 schwarze 1496: title = NULL;
1497: man_deroff(&title, body);
1.50 kristaps 1498: if (NULL == title)
1.61 schwarze 1499: return;
1.18 kristaps 1500:
1.141 schwarze 1501: /*
1.50 kristaps 1502: * Go through a special heuristic dance here.
1503: * Conventionally, one or more manual names are
1504: * comma-specified prior to a whitespace, then a
1505: * dash, then a description. Try to puzzle out
1506: * the name parts here.
1507: */
1.18 kristaps 1508:
1.121 schwarze 1509: start = title;
1.50 kristaps 1510: for ( ;; ) {
1511: sz = strcspn(start, " ,");
1512: if ('\0' == start[sz])
1513: break;
1.1 kristaps 1514:
1.50 kristaps 1515: byte = start[sz];
1516: start[sz] = '\0';
1.110 schwarze 1517:
1518: /*
1519: * Assume a stray trailing comma in the
1520: * name list if a name begins with a dash.
1521: */
1522:
1523: if ('-' == start[0] ||
1524: ('\\' == start[0] && '-' == start[1]))
1525: break;
1.1 kristaps 1526:
1.133 schwarze 1527: putkey(mpage, start, NAME_TITLE);
1.1 kristaps 1528:
1.50 kristaps 1529: if (' ' == byte) {
1530: start += sz + 1;
1531: break;
1532: }
1.1 kristaps 1533:
1.50 kristaps 1534: assert(',' == byte);
1535: start += sz + 1;
1536: while (' ' == *start)
1537: start++;
1538: }
1.1 kristaps 1539:
1.121 schwarze 1540: if (start == title) {
1.133 schwarze 1541: putkey(mpage, start, NAME_TITLE);
1.50 kristaps 1542: free(title);
1.61 schwarze 1543: return;
1.50 kristaps 1544: }
1.1 kristaps 1545:
1.50 kristaps 1546: while (isspace((unsigned char)*start))
1547: start++;
1.1 kristaps 1548:
1.50 kristaps 1549: if (0 == strncmp(start, "-", 1))
1550: start += 1;
1551: else if (0 == strncmp(start, "\\-\\-", 4))
1552: start += 4;
1553: else if (0 == strncmp(start, "\\-", 2))
1554: start += 2;
1555: else if (0 == strncmp(start, "\\(en", 4))
1556: start += 4;
1557: else if (0 == strncmp(start, "\\(em", 4))
1558: start += 4;
1.1 kristaps 1559:
1.50 kristaps 1560: while (' ' == *start)
1561: start++;
1.1 kristaps 1562:
1.78 schwarze 1563: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1564: free(title);
1.61 schwarze 1565: return;
1.50 kristaps 1566: }
1567: }
1.1 kristaps 1568:
1.77 schwarze 1569: for (n = n->child; n; n = n->next) {
1.78 schwarze 1570: if (NULL != mpage->desc)
1.77 schwarze 1571: break;
1.78 schwarze 1572: parse_man(mpage, n);
1.77 schwarze 1573: }
1.1 kristaps 1574: }
1575:
1576: static void
1.78 schwarze 1577: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1578: {
1579:
1.50 kristaps 1580: assert(NULL != n);
1581: for (n = n->child; NULL != n; n = n->next) {
1582: switch (n->type) {
1.141 schwarze 1583: case MDOC_ELEM:
1.50 kristaps 1584: /* FALLTHROUGH */
1.141 schwarze 1585: case MDOC_BLOCK:
1.50 kristaps 1586: /* FALLTHROUGH */
1.141 schwarze 1587: case MDOC_HEAD:
1.50 kristaps 1588: /* FALLTHROUGH */
1.141 schwarze 1589: case MDOC_BODY:
1.50 kristaps 1590: /* FALLTHROUGH */
1.141 schwarze 1591: case MDOC_TAIL:
1.50 kristaps 1592: if (NULL != mdocs[n->tok].fp)
1.78 schwarze 1593: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1.50 kristaps 1594: break;
1.68 schwarze 1595: if (mdocs[n->tok].mask)
1.78 schwarze 1596: putmdockey(mpage, n->child,
1597: mdocs[n->tok].mask);
1.50 kristaps 1598: break;
1599: default:
1600: assert(MDOC_ROOT != n->type);
1601: continue;
1602: }
1603: if (NULL != n->child)
1.78 schwarze 1604: parse_mdoc(mpage, n);
1.1 kristaps 1605: }
1606: }
1607:
1.25 schwarze 1608: static int
1.78 schwarze 1609: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1610: {
1611: const char *start, *end;
1612: size_t sz;
1.25 schwarze 1613:
1.50 kristaps 1614: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1615: NULL == (n = n->child) ||
1616: MDOC_TEXT != n->type)
1.25 schwarze 1617: return(0);
1.1 kristaps 1618:
1619: /*
1620: * Only consider those `Fd' macro fields that begin with an
1621: * "inclusion" token (versus, e.g., #define).
1622: */
1.50 kristaps 1623:
1.1 kristaps 1624: if (strcmp("#include", n->string))
1.25 schwarze 1625: return(0);
1.1 kristaps 1626:
1627: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1628: return(0);
1.1 kristaps 1629:
1630: /*
1631: * Strip away the enclosing angle brackets and make sure we're
1632: * not zero-length.
1633: */
1634:
1635: start = n->string;
1636: if ('<' == *start || '"' == *start)
1637: start++;
1638:
1639: if (0 == (sz = strlen(start)))
1.25 schwarze 1640: return(0);
1.1 kristaps 1641:
1642: end = &start[(int)sz - 1];
1643: if ('>' == *end || '"' == *end)
1644: end--;
1645:
1.50 kristaps 1646: if (end > start)
1.78 schwarze 1647: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1648: return(0);
1.1 kristaps 1649: }
1650:
1.25 schwarze 1651: static int
1.78 schwarze 1652: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1653: {
1.112 schwarze 1654: char *cp;
1.1 kristaps 1655:
1.50 kristaps 1656: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1657: return(0);
1658:
1.141 schwarze 1659: /*
1.50 kristaps 1660: * Parse: .Fn "struct type *name" "char *arg".
1.141 schwarze 1661: * First strip away pointer symbol.
1.50 kristaps 1662: * Then store the function name, then type.
1.141 schwarze 1663: * Finally, store the arguments.
1.50 kristaps 1664: */
1.1 kristaps 1665:
1.50 kristaps 1666: if (NULL == (cp = strrchr(n->string, ' ')))
1667: cp = n->string;
1.1 kristaps 1668:
1669: while ('*' == *cp)
1670: cp++;
1671:
1.78 schwarze 1672: putkey(mpage, cp, TYPE_Fn);
1.25 schwarze 1673:
1.50 kristaps 1674: if (n->string < cp)
1.78 schwarze 1675: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.25 schwarze 1676:
1.50 kristaps 1677: for (n = n->next; NULL != n; n = n->next)
1678: if (MDOC_TEXT == n->type)
1.78 schwarze 1679: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1680:
1681: return(0);
1.1 kristaps 1682: }
1683:
1.25 schwarze 1684: static int
1.78 schwarze 1685: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1686: {
1.67 schwarze 1687: char *cp;
1.1 kristaps 1688:
1689: if (NULL == (n = n->child))
1.25 schwarze 1690: return(0);
1.1 kristaps 1691:
1.67 schwarze 1692: if (NULL == n->next) {
1.78 schwarze 1693: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1694: return(0);
1695: }
1696:
1.120 schwarze 1697: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1698: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1699: free(cp);
1700: return(0);
1.1 kristaps 1701: }
1702:
1.25 schwarze 1703: static int
1.78 schwarze 1704: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1705: {
1706:
1.122 schwarze 1707: if (MDOC_BODY == n->type)
1708: mdoc_deroff(&mpage->desc, n);
1709: return(0);
1.1 kristaps 1710: }
1711:
1.25 schwarze 1712: static int
1.78 schwarze 1713: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1714: {
1715:
1.129 schwarze 1716: if (SEC_NAME == n->sec)
1.133 schwarze 1717: putmdockey(mpage, n->child, NAME_TITLE);
1.129 schwarze 1718: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.133 schwarze 1719: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1720: return(0);
1.1 kristaps 1721: }
1722:
1.25 schwarze 1723: static int
1.78 schwarze 1724: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1725: {
1726:
1.25 schwarze 1727: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1728: }
1729:
1.50 kristaps 1730: static int
1.78 schwarze 1731: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1732: {
1733:
1.50 kristaps 1734: return(MDOC_HEAD == n->type);
1735: }
1.1 kristaps 1736:
1.50 kristaps 1737: static int
1.78 schwarze 1738: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1.50 kristaps 1739: {
1.1 kristaps 1740:
1.50 kristaps 1741: return(MDOC_BODY == n->type);
1.1 kristaps 1742: }
1743:
1.50 kristaps 1744: /*
1.66 schwarze 1745: * Add a string to the hash table for the current manual.
1746: * Each string has a bitmask telling which macros it belongs to.
1747: * When we finish the manual, we'll dump the table.
1.50 kristaps 1748: */
1749: static void
1.78 schwarze 1750: putkeys(const struct mpage *mpage,
1751: const char *cp, size_t sz, uint64_t v)
1.50 kristaps 1752: {
1.133 schwarze 1753: struct ohash *htab;
1.50 kristaps 1754: struct str *s;
1.111 schwarze 1755: const char *end;
1.65 schwarze 1756: unsigned int slot;
1.111 schwarze 1757: int i;
1.25 schwarze 1758:
1.50 kristaps 1759: if (0 == sz)
1760: return;
1.111 schwarze 1761:
1.133 schwarze 1762: if (TYPE_Nm & v) {
1763: htab = &names;
1764: v &= name_mask;
1765: name_mask &= ~NAME_FIRST;
1766: if (debug > 1)
1767: say(mpage->mlinks->file,
1768: "Adding name %*s", sz, cp);
1769: } else {
1770: htab = &strings;
1771: if (debug > 1)
1772: for (i = 0; i < mansearch_keymax; i++)
1773: if (1 << i & v)
1774: say(mpage->mlinks->file,
1775: "Adding key %s=%*s",
1776: mansearch_keynames[i], sz, cp);
1.111 schwarze 1777: }
1.25 schwarze 1778:
1.65 schwarze 1779: end = cp + sz;
1.133 schwarze 1780: slot = ohash_qlookupi(htab, cp, &end);
1781: s = ohash_find(htab, slot);
1.25 schwarze 1782:
1.78 schwarze 1783: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1784: s->mask |= v;
1785: return;
1786: } else if (NULL == s) {
1.144 schwarze 1787: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1788: memcpy(s->key, cp, sz);
1.133 schwarze 1789: ohash_insert(htab, slot, s);
1.1 kristaps 1790: }
1.78 schwarze 1791: s->mpage = mpage;
1.50 kristaps 1792: s->mask = v;
1.1 kristaps 1793: }
1794:
1.50 kristaps 1795: /*
1796: * Take a Unicode codepoint and produce its UTF-8 encoding.
1797: * This isn't the best way to do this, but it works.
1798: * The magic numbers are from the UTF-8 packaging.
1799: * They're not as scary as they seem: read the UTF-8 spec for details.
1800: */
1801: static size_t
1802: utf8(unsigned int cp, char out[7])
1.1 kristaps 1803: {
1.50 kristaps 1804: size_t rc;
1.1 kristaps 1805:
1.50 kristaps 1806: rc = 0;
1807: if (cp <= 0x0000007F) {
1808: rc = 1;
1809: out[0] = (char)cp;
1810: } else if (cp <= 0x000007FF) {
1811: rc = 2;
1812: out[0] = (cp >> 6 & 31) | 192;
1813: out[1] = (cp & 63) | 128;
1814: } else if (cp <= 0x0000FFFF) {
1815: rc = 3;
1816: out[0] = (cp >> 12 & 15) | 224;
1817: out[1] = (cp >> 6 & 63) | 128;
1818: out[2] = (cp & 63) | 128;
1819: } else if (cp <= 0x001FFFFF) {
1820: rc = 4;
1821: out[0] = (cp >> 18 & 7) | 240;
1822: out[1] = (cp >> 12 & 63) | 128;
1823: out[2] = (cp >> 6 & 63) | 128;
1824: out[3] = (cp & 63) | 128;
1825: } else if (cp <= 0x03FFFFFF) {
1826: rc = 5;
1827: out[0] = (cp >> 24 & 3) | 248;
1828: out[1] = (cp >> 18 & 63) | 128;
1829: out[2] = (cp >> 12 & 63) | 128;
1830: out[3] = (cp >> 6 & 63) | 128;
1831: out[4] = (cp & 63) | 128;
1832: } else if (cp <= 0x7FFFFFFF) {
1833: rc = 6;
1834: out[0] = (cp >> 30 & 1) | 252;
1835: out[1] = (cp >> 24 & 63) | 128;
1836: out[2] = (cp >> 18 & 63) | 128;
1837: out[3] = (cp >> 12 & 63) | 128;
1838: out[4] = (cp >> 6 & 63) | 128;
1839: out[5] = (cp & 63) | 128;
1840: } else
1.1 kristaps 1841: return(0);
1842:
1.50 kristaps 1843: out[rc] = '\0';
1844: return(rc);
1845: }
1.1 kristaps 1846:
1.50 kristaps 1847: /*
1.96 schwarze 1848: * Store the rendered version of a key, or alias the pointer
1849: * if the key contains no escape sequences.
1.50 kristaps 1850: */
1851: static void
1.96 schwarze 1852: render_key(struct mchars *mc, struct str *key)
1.50 kristaps 1853: {
1854: size_t sz, bsz, pos;
1.114 schwarze 1855: char utfbuf[7], res[6];
1.50 kristaps 1856: char *buf;
1857: const char *seq, *cpp, *val;
1858: int len, u;
1859: enum mandoc_esc esc;
1860:
1.96 schwarze 1861: assert(NULL == key->rendered);
1.50 kristaps 1862:
1863: res[0] = '\\';
1864: res[1] = '\t';
1865: res[2] = ASCII_NBRSP;
1866: res[3] = ASCII_HYPH;
1.114 schwarze 1867: res[4] = ASCII_BREAK;
1868: res[5] = '\0';
1.1 kristaps 1869:
1.50 kristaps 1870: val = key->key;
1871: bsz = strlen(val);
1.46 kristaps 1872:
1.50 kristaps 1873: /*
1874: * Pre-check: if we have no stop-characters, then set the
1875: * pointer as ourselvse and get out of here.
1876: */
1877: if (strcspn(val, res) == bsz) {
1.96 schwarze 1878: key->rendered = key->key;
1.50 kristaps 1879: return;
1.141 schwarze 1880: }
1.46 kristaps 1881:
1.50 kristaps 1882: /* Pre-allocate by the length of the input */
1.46 kristaps 1883:
1.50 kristaps 1884: buf = mandoc_malloc(++bsz);
1885: pos = 0;
1.46 kristaps 1886:
1.50 kristaps 1887: while ('\0' != *val) {
1888: /*
1889: * Halt on the first escape sequence.
1890: * This also halts on the end of string, in which case
1891: * we just copy, fallthrough, and exit the loop.
1892: */
1893: if ((sz = strcspn(val, res)) > 0) {
1894: memcpy(&buf[pos], val, sz);
1895: pos += sz;
1896: val += sz;
1897: }
1.46 kristaps 1898:
1.114 schwarze 1899: switch (*val) {
1.141 schwarze 1900: case ASCII_HYPH:
1.50 kristaps 1901: buf[pos++] = '-';
1902: val++;
1903: continue;
1.141 schwarze 1904: case '\t':
1.114 schwarze 1905: /* FALLTHROUGH */
1.141 schwarze 1906: case ASCII_NBRSP:
1.50 kristaps 1907: buf[pos++] = ' ';
1908: val++;
1.114 schwarze 1909: /* FALLTHROUGH */
1.141 schwarze 1910: case ASCII_BREAK:
1.50 kristaps 1911: continue;
1.114 schwarze 1912: default:
1913: break;
1914: }
1915: if ('\\' != *val)
1.50 kristaps 1916: break;
1.46 kristaps 1917:
1.50 kristaps 1918: /* Read past the slash. */
1.46 kristaps 1919:
1.50 kristaps 1920: val++;
1.46 kristaps 1921:
1.50 kristaps 1922: /*
1923: * Parse the escape sequence and see if it's a
1924: * predefined character or special character.
1925: */
1.95 schwarze 1926:
1.141 schwarze 1927: esc = mandoc_escape((const char **)&val,
1928: &seq, &len);
1.50 kristaps 1929: if (ESCAPE_ERROR == esc)
1930: break;
1931: if (ESCAPE_SPECIAL != esc)
1932: continue;
1.1 kristaps 1933:
1.50 kristaps 1934: /*
1.95 schwarze 1935: * Render the special character
1936: * as either UTF-8 or ASCII.
1.50 kristaps 1937: */
1.95 schwarze 1938:
1939: if (write_utf8) {
1940: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1941: continue;
1942: cpp = utfbuf;
1943: if (0 == (sz = utf8(u, utfbuf)))
1944: continue;
1945: sz = strlen(cpp);
1946: } else {
1947: cpp = mchars_spec2str(mc, seq, len, &sz);
1948: if (NULL == cpp)
1949: continue;
1950: if (ASCII_NBRSP == *cpp) {
1951: cpp = " ";
1952: sz = 1;
1953: }
1954: }
1.1 kristaps 1955:
1.50 kristaps 1956: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1957:
1.50 kristaps 1958: bsz += sz;
1959: buf = mandoc_realloc(buf, bsz);
1960: memcpy(&buf[pos], cpp, sz);
1961: pos += sz;
1.1 kristaps 1962: }
1963:
1.50 kristaps 1964: buf[pos] = '\0';
1.96 schwarze 1965: key->rendered = buf;
1.1 kristaps 1966: }
1967:
1.118 schwarze 1968: static void
1969: dbadd_mlink(const struct mlink *mlink)
1970: {
1971: size_t i;
1972:
1973: i = 1;
1974: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1975: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1976: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 1977: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 1978: SQL_STEP(stmts[STMT_INSERT_LINK]);
1979: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1980: }
1981:
1.14 schwarze 1982: /*
1.50 kristaps 1983: * Flush the current page's terms (and their bits) into the database.
1984: * Wrap the entire set of additions in a transaction to make sqlite be a
1985: * little faster.
1.96 schwarze 1986: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 1987: */
1988: static void
1.118 schwarze 1989: dbadd(struct mpage *mpage, struct mchars *mc)
1.14 schwarze 1990: {
1.87 schwarze 1991: struct mlink *mlink;
1.50 kristaps 1992: struct str *key;
1.52 kristaps 1993: size_t i;
1.66 schwarze 1994: unsigned int slot;
1.14 schwarze 1995:
1.128 schwarze 1996: mlink = mpage->mlinks;
1.43 kristaps 1997:
1.128 schwarze 1998: if (nodb) {
1.147 schwarze 1999: for (key = ohash_first(&names, &slot); NULL != key;
2000: key = ohash_next(&names, &slot)) {
2001: if (key->rendered != key->key)
2002: free(key->rendered);
2003: free(key);
2004: }
2005: for (key = ohash_first(&strings, &slot); NULL != key;
2006: key = ohash_next(&strings, &slot)) {
2007: if (key->rendered != key->key)
2008: free(key->rendered);
2009: free(key);
2010: }
1.145 schwarze 2011: if (0 == debug)
2012: return;
1.128 schwarze 2013: while (NULL != mlink) {
2014: fputs(mlink->name, stdout);
2015: if (NULL == mlink->next ||
2016: strcmp(mlink->dsec, mlink->next->dsec) ||
2017: strcmp(mlink->fsec, mlink->next->fsec) ||
2018: strcmp(mlink->arch, mlink->next->arch)) {
2019: putchar('(');
2020: if ('\0' == *mlink->dsec)
2021: fputs(mlink->fsec, stdout);
2022: else
2023: fputs(mlink->dsec, stdout);
2024: if ('\0' != *mlink->arch)
2025: printf("/%s", mlink->arch);
2026: putchar(')');
2027: }
2028: mlink = mlink->next;
2029: if (NULL != mlink)
2030: fputs(", ", stdout);
2031: }
1.132 schwarze 2032: printf(" - %s\n", mpage->desc);
1.14 schwarze 2033: return;
1.128 schwarze 2034: }
2035:
2036: if (debug)
2037: say(mlink->file, "Adding to database");
1.28 kristaps 2038:
1.148 ! schwarze 2039: i = strlen(mpage->desc) + 1;
! 2040: key = mandoc_calloc(1, sizeof(struct str) + i);
! 2041: memcpy(key->key, mpage->desc, i);
! 2042: render_key(mc, key);
! 2043:
1.52 kristaps 2044: i = 1;
1.148 ! schwarze 2045: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.82 schwarze 2046: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1.81 schwarze 2047: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2048: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2049: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.148 ! schwarze 2050:
! 2051: if (key->rendered != key->key)
! 2052: free(key->rendered);
! 2053: free(key);
1.81 schwarze 2054:
1.128 schwarze 2055: while (NULL != mlink) {
1.118 schwarze 2056: dbadd_mlink(mlink);
1.128 schwarze 2057: mlink = mlink->next;
2058: }
1.134 schwarze 2059: mlink = mpage->mlinks;
1.50 kristaps 2060:
1.133 schwarze 2061: for (key = ohash_first(&names, &slot); NULL != key;
2062: key = ohash_next(&names, &slot)) {
2063: assert(key->mpage == mpage);
2064: if (NULL == key->rendered)
2065: render_key(mc, key);
2066: i = 1;
2067: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2068: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.137 schwarze 2069: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2070: SQL_STEP(stmts[STMT_INSERT_NAME]);
2071: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2072: if (key->rendered != key->key)
2073: free(key->rendered);
2074: free(key);
2075: }
1.66 schwarze 2076: for (key = ohash_first(&strings, &slot); NULL != key;
2077: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2078: assert(key->mpage == mpage);
1.96 schwarze 2079: if (NULL == key->rendered)
2080: render_key(mc, key);
1.52 kristaps 2081: i = 1;
2082: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.96 schwarze 2083: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.137 schwarze 2084: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2085: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2086: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.96 schwarze 2087: if (key->rendered != key->key)
2088: free(key->rendered);
1.66 schwarze 2089: free(key);
1.38 schwarze 2090: }
1.14 schwarze 2091: }
2092:
1.5 kristaps 2093: static void
1.59 schwarze 2094: dbprune(void)
1.5 kristaps 2095: {
1.78 schwarze 2096: struct mpage *mpage;
1.82 schwarze 2097: struct mlink *mlink;
1.52 kristaps 2098: size_t i;
1.80 schwarze 2099: unsigned int slot;
1.5 kristaps 2100:
1.106 schwarze 2101: if (0 == nodb)
2102: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2103:
1.106 schwarze 2104: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2105: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2106: mlink = mpage->mlinks;
1.125 schwarze 2107: if (debug)
1.106 schwarze 2108: say(mlink->file, "Deleting from database");
2109: if (nodb)
2110: continue;
2111: for ( ; NULL != mlink; mlink = mlink->next) {
2112: i = 1;
2113: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2114: i, mlink->dsec);
2115: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2116: i, mlink->arch);
2117: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2118: i, mlink->name);
2119: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2120: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2121: }
1.5 kristaps 2122: }
1.106 schwarze 2123:
2124: if (0 == nodb)
2125: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2126: }
2127:
1.4 kristaps 2128: /*
1.50 kristaps 2129: * Close an existing database and its prepared statements.
2130: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2131: */
1.35 kristaps 2132: static void
1.59 schwarze 2133: dbclose(int real)
1.4 kristaps 2134: {
1.50 kristaps 2135: size_t i;
1.115 schwarze 2136: int status;
2137: pid_t child;
1.4 kristaps 2138:
1.50 kristaps 2139: if (nodb)
1.38 schwarze 2140: return;
1.50 kristaps 2141:
2142: for (i = 0; i < STMT__MAX; i++) {
2143: sqlite3_finalize(stmts[i]);
2144: stmts[i] = NULL;
1.4 kristaps 2145: }
2146:
1.50 kristaps 2147: sqlite3_close(db);
2148: db = NULL;
1.12 schwarze 2149:
1.50 kristaps 2150: if (real)
2151: return;
1.12 schwarze 2152:
1.115 schwarze 2153: if ('\0' == *tempfilename) {
2154: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2155: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2156: say(MANDOC_DB, "&rename");
1.115 schwarze 2157: }
2158: return;
2159: }
2160:
2161: switch (child = fork()) {
1.141 schwarze 2162: case -1:
1.115 schwarze 2163: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2164: say("", "&fork cmp");
1.115 schwarze 2165: return;
1.141 schwarze 2166: case 0:
1.115 schwarze 2167: execlp("cmp", "cmp", "-s",
2168: tempfilename, MANDOC_DB, NULL);
1.123 schwarze 2169: say("", "&exec cmp");
1.115 schwarze 2170: exit(0);
2171: default:
2172: break;
2173: }
2174: if (-1 == waitpid(child, &status, 0)) {
2175: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2176: say("", "&wait cmp");
1.115 schwarze 2177: } else if (WIFSIGNALED(status)) {
2178: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2179: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2180: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2181: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2182: say(MANDOC_DB,
2183: "Data changed, but cannot replace database");
2184: }
2185:
2186: *strrchr(tempfilename, '/') = '\0';
2187: switch (child = fork()) {
1.141 schwarze 2188: case -1:
1.115 schwarze 2189: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2190: say("", "&fork rm");
1.115 schwarze 2191: return;
1.141 schwarze 2192: case 0:
1.115 schwarze 2193: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.123 schwarze 2194: say("", "&exec rm");
1.115 schwarze 2195: exit((int)MANDOCLEVEL_SYSERR);
2196: default:
2197: break;
2198: }
2199: if (-1 == waitpid(child, &status, 0)) {
2200: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2201: say("", "&wait rm");
1.115 schwarze 2202: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2203: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2204: say("", "%s: Cannot remove temporary directory",
2205: tempfilename);
1.59 schwarze 2206: }
1.50 kristaps 2207: }
1.14 schwarze 2208:
1.50 kristaps 2209: /*
2210: * This is straightforward stuff.
2211: * Open a database connection to a "temporary" database, then open a set
2212: * of prepared statements we'll use over and over again.
2213: * If "real" is set, we use the existing database; if not, we truncate a
2214: * temporary one.
2215: * Must be matched by dbclose().
2216: */
2217: static int
1.59 schwarze 2218: dbopen(int real)
1.50 kristaps 2219: {
1.115 schwarze 2220: const char *sql;
1.50 kristaps 2221: int rc, ofl;
1.12 schwarze 2222:
1.141 schwarze 2223: if (nodb)
1.50 kristaps 2224: return(1);
1.12 schwarze 2225:
1.115 schwarze 2226: *tempfilename = '\0';
1.63 schwarze 2227: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2228:
2229: if (real) {
2230: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2231: if (SQLITE_OK != rc) {
1.63 schwarze 2232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2233: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.63 schwarze 2234: return(0);
2235: }
1.115 schwarze 2236: goto prepare_statements;
2237: }
2238:
2239: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2240:
1.115 schwarze 2241: remove(MANDOC_DB "~");
2242: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2243: if (SQLITE_OK == rc)
1.115 schwarze 2244: goto create_tables;
1.116 schwarze 2245: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2246: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2247: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.50 kristaps 2248: return(0);
2249: }
1.12 schwarze 2250:
1.143 schwarze 2251: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2252: sizeof(tempfilename));
1.115 schwarze 2253: if (NULL == mkdtemp(tempfilename)) {
2254: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2255: say("", "&%s", tempfilename);
1.115 schwarze 2256: return(0);
2257: }
1.143 schwarze 2258: (void)strlcat(tempfilename, "/" MANDOC_DB,
2259: sizeof(tempfilename));
1.115 schwarze 2260: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2261: if (SQLITE_OK != rc) {
1.59 schwarze 2262: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2263: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.50 kristaps 2264: return(0);
2265: }
1.12 schwarze 2266:
1.115 schwarze 2267: create_tables:
1.81 schwarze 2268: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2269: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2270: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2271: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2272: ");\n"
2273: "\n"
1.81 schwarze 2274: "CREATE TABLE \"mlinks\" (\n"
2275: " \"sec\" TEXT NOT NULL,\n"
2276: " \"arch\" TEXT NOT NULL,\n"
2277: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2278: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2279: "ON DELETE CASCADE\n"
1.81 schwarze 2280: ");\n"
1.136 schwarze 2281: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2282: "\n"
1.133 schwarze 2283: "CREATE TABLE \"names\" (\n"
2284: " \"bits\" INTEGER NOT NULL,\n"
2285: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2286: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.133 schwarze 2287: "ON DELETE CASCADE\n"
2288: ");\n"
2289: "\n"
1.50 kristaps 2290: "CREATE TABLE \"keys\" (\n"
2291: " \"bits\" INTEGER NOT NULL,\n"
2292: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2293: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2294: "ON DELETE CASCADE\n"
1.136 schwarze 2295: ");\n"
2296: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2297:
1.50 kristaps 2298: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2299: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2300: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2301: sqlite3_close(db);
1.50 kristaps 2302: return(0);
2303: }
1.4 kristaps 2304:
1.57 schwarze 2305: prepare_statements:
1.146 schwarze 2306: if (SQLITE_OK != sqlite3_exec(db,
2307: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2308: exitcode = (int)MANDOCLEVEL_SYSERR;
2309: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2310: sqlite3_errmsg(db));
2311: sqlite3_close(db);
2312: return(0);
2313: }
2314:
1.137 schwarze 2315: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2316: "(SELECT pageid FROM mlinks WHERE "
2317: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2318: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2319: sql = "INSERT INTO mpages "
1.132 schwarze 2320: "(desc,form) VALUES (?,?)";
1.81 schwarze 2321: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2322: sql = "INSERT INTO mlinks "
1.104 schwarze 2323: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2324: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.133 schwarze 2325: sql = "INSERT INTO names "
2326: "(bits,name,pageid) VALUES (?,?,?)";
2327: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2328: sql = "INSERT INTO keys "
1.81 schwarze 2329: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2330: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2331:
2332: #ifndef __APPLE__
2333: /*
2334: * When opening a new database, we can turn off
2335: * synchronous mode for much better performance.
2336: */
2337:
1.146 schwarze 2338: if (real && SQLITE_OK != sqlite3_exec(db,
2339: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2340: exitcode = (int)MANDOCLEVEL_SYSERR;
2341: say(MANDOC_DB, "PRAGMA synchronous: %s",
2342: sqlite3_errmsg(db));
2343: sqlite3_close(db);
2344: return(0);
2345: }
1.70 schwarze 2346: #endif
2347:
1.50 kristaps 2348: return(1);
2349: }
1.5 kristaps 2350:
1.50 kristaps 2351: static void *
2352: hash_halloc(size_t sz, void *arg)
2353: {
1.12 schwarze 2354:
1.144 schwarze 2355: return(mandoc_calloc(1, sz));
1.50 kristaps 2356: }
1.12 schwarze 2357:
1.50 kristaps 2358: static void *
2359: hash_alloc(size_t sz, void *arg)
2360: {
1.5 kristaps 2361:
1.50 kristaps 2362: return(mandoc_malloc(sz));
2363: }
1.41 kristaps 2364:
1.50 kristaps 2365: static void
2366: hash_free(void *p, size_t sz, void *arg)
2367: {
1.4 kristaps 2368:
1.50 kristaps 2369: free(p);
1.4 kristaps 2370: }
2371:
1.50 kristaps 2372: static int
1.59 schwarze 2373: set_basedir(const char *targetdir)
1.4 kristaps 2374: {
1.59 schwarze 2375: static char startdir[PATH_MAX];
2376: static int fd;
1.4 kristaps 2377:
1.59 schwarze 2378: /*
2379: * Remember where we started by keeping a fd open to the origin
2380: * path component: throughout this utility, we chdir() a lot to
2381: * handle relative paths, and by doing this, we can return to
2382: * the starting point.
2383: */
2384: if ('\0' == *startdir) {
2385: if (NULL == getcwd(startdir, PATH_MAX)) {
2386: exitcode = (int)MANDOCLEVEL_SYSERR;
2387: if (NULL != targetdir)
1.123 schwarze 2388: say("", "&getcwd");
1.59 schwarze 2389: return(0);
2390: }
2391: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2392: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2393: say("", "&open %s", startdir);
1.59 schwarze 2394: return(0);
2395: }
2396: if (NULL == targetdir)
2397: targetdir = startdir;
2398: } else {
2399: if (-1 == fd)
2400: return(0);
2401: if (-1 == fchdir(fd)) {
2402: close(fd);
2403: basedir[0] = '\0';
2404: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2405: say("", "&chdir %s", startdir);
1.59 schwarze 2406: return(0);
2407: }
2408: if (NULL == targetdir) {
2409: close(fd);
2410: return(1);
2411: }
2412: }
2413: if (NULL == realpath(targetdir, basedir)) {
2414: basedir[0] = '\0';
2415: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 2416: say("", "&%s: realpath", targetdir);
1.50 kristaps 2417: return(0);
1.59 schwarze 2418: } else if (-1 == chdir(basedir)) {
2419: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 2420: say("", "&chdir");
1.50 kristaps 2421: return(0);
1.4 kristaps 2422: }
1.50 kristaps 2423: return(1);
1.56 schwarze 2424: }
2425:
2426: static void
1.59 schwarze 2427: say(const char *file, const char *format, ...)
1.56 schwarze 2428: {
2429: va_list ap;
1.123 schwarze 2430: int use_errno;
1.56 schwarze 2431:
1.59 schwarze 2432: if ('\0' != *basedir)
2433: fprintf(stderr, "%s", basedir);
2434: if ('\0' != *basedir && '\0' != *file)
2435: fputs("//", stderr);
1.56 schwarze 2436: if ('\0' != *file)
1.59 schwarze 2437: fprintf(stderr, "%s", file);
2438:
1.123 schwarze 2439: use_errno = 1;
2440: if (NULL != format) {
2441: switch (*format) {
1.141 schwarze 2442: case '&':
1.123 schwarze 2443: format++;
2444: break;
1.141 schwarze 2445: case '\0':
1.123 schwarze 2446: format = NULL;
2447: break;
2448: default:
2449: use_errno = 0;
2450: break;
2451: }
2452: }
2453: if (NULL != format) {
2454: if ('\0' != *basedir || '\0' != *file)
2455: fputs(": ", stderr);
2456: va_start(ap, format);
2457: vfprintf(stderr, format, ap);
2458: va_end(ap);
2459: }
2460: if (use_errno) {
2461: if ('\0' != *basedir || '\0' != *file || NULL != format)
2462: fputs(": ", stderr);
1.59 schwarze 2463: perror(NULL);
1.123 schwarze 2464: } else
2465: fputc('\n', stderr);
1.1 kristaps 2466: }
CVSweb