Annotation of mandoc/mandocdb.c, Revision 1.153
1.153 ! schwarze 1: /* $Id: mandocdb.c,v 1.152 2014/06/20 02:24:40 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 *);
1.152 schwarze 148: static void hash_free(void *, void *);
149: static void *hash_calloc(size_t, 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;
1.152 schwarze 339: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
340: mpages_info.free = mlinks_info.free = 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)) { \
1.153 ! schwarze 358: fprintf(stderr, "%s: -%c: Conflicting option\n", \
! 359: progname, (_ch)); \
1.50 kristaps 360: goto usage; \
361: } while (/*CONSTCOND*/0)
1.10 kristaps 362:
1.59 schwarze 363: path_arg = NULL;
1.38 schwarze 364: op = OP_DEFAULT;
1.1 kristaps 365:
1.126 schwarze 366: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 367: switch (ch) {
1.141 schwarze 368: case 'a':
1.12 schwarze 369: use_all = 1;
370: break;
1.141 schwarze 371: case 'C':
1.50 kristaps 372: CHECKOP(op, ch);
1.59 schwarze 373: path_arg = optarg;
1.38 schwarze 374: op = OP_CONFFILE;
1.34 schwarze 375: break;
1.141 schwarze 376: case 'D':
1.125 schwarze 377: debug++;
378: break;
1.141 schwarze 379: case 'd':
1.50 kristaps 380: CHECKOP(op, ch);
1.59 schwarze 381: path_arg = optarg;
1.5 kristaps 382: op = OP_UPDATE;
383: break;
1.141 schwarze 384: case 'n':
1.50 kristaps 385: nodb = 1;
386: break;
1.141 schwarze 387: case 'p':
1.126 schwarze 388: warnings = 1;
389: break;
1.141 schwarze 390: case 'Q':
1.116 schwarze 391: mparse_options |= MPARSE_QUICK;
1.102 schwarze 392: break;
1.141 schwarze 393: case 'T':
1.95 schwarze 394: if (strcmp(optarg, "utf8")) {
1.153 ! schwarze 395: fprintf(stderr, "%s: -T%s: "
! 396: "Unsupported output format\n",
! 397: progname, optarg);
1.95 schwarze 398: goto usage;
399: }
400: write_utf8 = 1;
401: break;
1.141 schwarze 402: case 't':
1.50 kristaps 403: CHECKOP(op, ch);
1.38 schwarze 404: dup2(STDOUT_FILENO, STDERR_FILENO);
405: op = OP_TEST;
1.50 kristaps 406: nodb = warnings = 1;
1.38 schwarze 407: break;
1.141 schwarze 408: case 'u':
1.50 kristaps 409: CHECKOP(op, ch);
1.59 schwarze 410: path_arg = optarg;
1.5 kristaps 411: op = OP_DELETE;
412: break;
1.141 schwarze 413: case 'v':
1.126 schwarze 414: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 415: break;
1.1 kristaps 416: default:
1.38 schwarze 417: goto usage;
1.1 kristaps 418: }
419:
420: argc -= optind;
421: argv += optind;
422:
1.38 schwarze 423: if (OP_CONFFILE == op && argc > 0) {
1.153 ! schwarze 424: fprintf(stderr, "%s: -C: Too many arguments\n",
! 425: progname);
1.38 schwarze 426: goto usage;
427: }
428:
1.59 schwarze 429: exitcode = (int)MANDOCLEVEL_OK;
1.116 schwarze 430: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
1.50 kristaps 431: mc = mchars_alloc();
432:
1.78 schwarze 433: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 434: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 435:
436: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 437:
1.50 kristaps 438: /*
1.150 schwarze 439: * Most of these deal with a specific directory.
1.140 schwarze 440: * Jump into that directory first.
1.50 kristaps 441: */
1.150 schwarze 442: if (OP_TEST != op && 0 == set_basedir(path_arg))
1.50 kristaps 443: goto out;
1.140 schwarze 444:
1.138 schwarze 445: if (dbopen(1)) {
1.140 schwarze 446: /*
447: * The existing database is usable. Process
448: * all files specified on the command-line.
449: */
450: use_all = 1;
1.138 schwarze 451: for (i = 0; i < argc; i++)
452: filescan(argv[i]);
453: if (OP_TEST != op)
454: dbprune();
455: } else {
456: /*
457: * Database missing or corrupt.
458: * Recreate from scratch.
459: */
1.140 schwarze 460: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 461: op = OP_DEFAULT;
462: if (0 == treescan())
463: goto out;
464: if (0 == dbopen(0))
465: goto out;
466: }
1.50 kristaps 467: if (OP_DELETE != op)
1.101 schwarze 468: mpages_merge(mc, mp);
1.138 schwarze 469: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 470: } else {
471: /*
472: * If we have arguments, use them as our manpaths.
473: * If we don't, grok from manpath(1) or however else
474: * manpath_parse() wants to do it.
475: */
476: if (argc > 0) {
1.144 schwarze 477: dirs.paths = mandoc_reallocarray(NULL,
478: argc, sizeof(char *));
1.50 kristaps 479: dirs.sz = (size_t)argc;
480: for (i = 0; i < argc; i++)
481: dirs.paths[i] = mandoc_strdup(argv[i]);
482: } else
1.59 schwarze 483: manpath_parse(&dirs, path_arg, NULL, NULL);
1.127 schwarze 484:
485: if (0 == dirs.sz) {
486: exitcode = (int)MANDOCLEVEL_BADARG;
487: say("", "Empty manpath");
488: }
1.50 kristaps 489:
490: /*
1.69 schwarze 491: * First scan the tree rooted at a base directory, then
492: * build a new database and finally move it into place.
1.50 kristaps 493: * Ignore zero-length directories and strip trailing
494: * slashes.
495: */
496: for (j = 0; j < dirs.sz; j++) {
497: sz = strlen(dirs.paths[j]);
498: if (sz && '/' == dirs.paths[j][sz - 1])
499: dirs.paths[j][--sz] = '\0';
500: if (0 == sz)
501: continue;
1.66 schwarze 502:
503: if (j) {
1.78 schwarze 504: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 505: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 506: }
507:
1.59 schwarze 508: if (0 == set_basedir(dirs.paths[j]))
1.50 kristaps 509: goto out;
1.59 schwarze 510: if (0 == treescan())
1.50 kristaps 511: goto out;
1.59 schwarze 512: if (0 == dbopen(0))
1.50 kristaps 513: goto out;
1.52 kristaps 514:
1.101 schwarze 515: mpages_merge(mc, mp);
1.147 schwarze 516: if (warnings && !nodb &&
1.130 schwarze 517: ! (MPARSE_QUICK & mparse_options))
518: names_check();
1.59 schwarze 519: dbclose(0);
1.66 schwarze 520:
521: if (j + 1 < dirs.sz) {
1.80 schwarze 522: mpages_free();
1.78 schwarze 523: ohash_delete(&mpages);
1.83 schwarze 524: ohash_delete(&mlinks);
1.66 schwarze 525: }
1.50 kristaps 526: }
527: }
528: out:
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:
1.150 schwarze 803: if (OP_TEST == op)
1.106 schwarze 804: start = buf;
1.150 schwarze 805: else if (strstr(buf, basedir) == buf)
806: start = buf + strlen(basedir);
1.106 schwarze 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: }
1.150 schwarze 832: start = buf;
833: if (OP_TEST != op && strstr(buf, basedir) == buf)
834: start += strlen(basedir);
1.50 kristaps 835: }
1.106 schwarze 836:
1.84 schwarze 837: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 838: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
839: sizeof(mlink->file)) {
840: say(start, "Filename too long");
841: return;
842: }
1.17 schwarze 843:
1.50 kristaps 844: /*
845: * First try to guess our directory structure.
846: * If we find a separator, try to look for man* or cat*.
847: * If we find one of these and what's underneath is a directory,
848: * assume it's an architecture.
849: */
850: if (NULL != (p = strchr(start, '/'))) {
851: *p++ = '\0';
852: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 853: mlink->dform = FORM_SRC;
1.94 schwarze 854: mlink->dsec = start + 3;
1.50 kristaps 855: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 856: mlink->dform = FORM_CAT;
1.94 schwarze 857: mlink->dsec = start + 3;
1.17 schwarze 858: }
1.5 kristaps 859:
1.50 kristaps 860: start = p;
1.84 schwarze 861: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 862: *p++ = '\0';
1.94 schwarze 863: mlink->arch = start;
1.50 kristaps 864: start = p;
1.84 schwarze 865: }
1.50 kristaps 866: }
867:
868: /*
869: * Now check the file suffix.
870: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
871: */
872: p = strrchr(start, '\0');
873: while (p-- > start && '/' != *p && '.' != *p)
874: /* Loop. */ ;
875:
876: if ('.' == *p) {
877: *p++ = '\0';
1.94 schwarze 878: mlink->fsec = p;
1.5 kristaps 879: }
880:
1.10 kristaps 881: /*
1.50 kristaps 882: * Now try to parse the name.
883: * Use the filename portion of the path.
1.10 kristaps 884: */
1.84 schwarze 885: mlink->name = start;
1.50 kristaps 886: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 887: mlink->name = p + 1;
1.50 kristaps 888: *p = '\0';
1.84 schwarze 889: }
890: mlink_add(mlink, &st);
1.50 kristaps 891: }
1.5 kristaps 892:
1.50 kristaps 893: static void
1.84 schwarze 894: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 895: {
1.83 schwarze 896: struct inodev inodev;
1.78 schwarze 897: struct mpage *mpage;
1.83 schwarze 898: unsigned int slot;
1.50 kristaps 899:
1.84 schwarze 900: assert(NULL != mlink->file);
1.50 kristaps 901:
1.94 schwarze 902: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
903: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
904: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
905: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 906:
907: if ('0' == *mlink->fsec) {
908: free(mlink->fsec);
909: mlink->fsec = mandoc_strdup(mlink->dsec);
910: mlink->fform = FORM_CAT;
911: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
912: mlink->fform = FORM_SRC;
1.74 schwarze 913: else
1.84 schwarze 914: mlink->fform = FORM_NONE;
1.82 schwarze 915:
1.83 schwarze 916: slot = ohash_qlookup(&mlinks, mlink->file);
917: assert(NULL == ohash_find(&mlinks, slot));
918: ohash_insert(&mlinks, slot, mlink);
919:
920: inodev.st_ino = st->st_ino;
921: inodev.st_dev = st->st_dev;
922: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
923: sizeof(struct inodev), inodev.st_ino);
924: mpage = ohash_find(&mpages, slot);
925: if (NULL == mpage) {
926: mpage = mandoc_calloc(1, sizeof(struct mpage));
927: mpage->inodev.st_ino = inodev.st_ino;
928: mpage->inodev.st_dev = inodev.st_dev;
929: ohash_insert(&mpages, slot, mpage);
930: } else
1.85 schwarze 931: mlink->next = mpage->mlinks;
1.82 schwarze 932: mpage->mlinks = mlink;
1.118 schwarze 933: mlink->mpage = mpage;
1.82 schwarze 934: }
935:
936: static void
937: mlink_free(struct mlink *mlink)
938: {
939:
940: free(mlink->dsec);
941: free(mlink->arch);
942: free(mlink->name);
943: free(mlink->fsec);
944: free(mlink);
1.50 kristaps 945: }
1.3 kristaps 946:
1.50 kristaps 947: static void
1.78 schwarze 948: mpages_free(void)
1.50 kristaps 949: {
1.78 schwarze 950: struct mpage *mpage;
1.82 schwarze 951: struct mlink *mlink;
1.80 schwarze 952: unsigned int slot;
1.3 kristaps 953:
1.80 schwarze 954: mpage = ohash_first(&mpages, &slot);
955: while (NULL != mpage) {
1.82 schwarze 956: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 957: mpage->mlinks = mlink->next;
1.82 schwarze 958: mlink_free(mlink);
959: }
1.78 schwarze 960: free(mpage->sec);
961: free(mpage->arch);
1.82 schwarze 962: free(mpage->title);
963: free(mpage->desc);
1.78 schwarze 964: free(mpage);
1.80 schwarze 965: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 966: }
967: }
1.38 schwarze 968:
1.50 kristaps 969: /*
1.89 schwarze 970: * For each mlink to the mpage, check whether the path looks like
971: * it is formatted, and if it does, check whether a source manual
972: * exists by the same name, ignoring the suffix.
973: * If both conditions hold, drop the mlink.
974: */
975: static void
976: mlinks_undupe(struct mpage *mpage)
977: {
978: char buf[PATH_MAX];
979: struct mlink **prev;
980: struct mlink *mlink;
981: char *bufp;
982:
983: mpage->form = FORM_CAT;
1.90 schwarze 984: prev = &mpage->mlinks;
985: while (NULL != (mlink = *prev)) {
1.89 schwarze 986: if (FORM_CAT != mlink->dform) {
987: mpage->form = FORM_NONE;
1.90 schwarze 988: goto nextlink;
1.89 schwarze 989: }
1.143 schwarze 990: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 991: bufp = strstr(buf, "cat");
992: assert(NULL != bufp);
993: memcpy(bufp, "man", 3);
994: if (NULL != (bufp = strrchr(buf, '.')))
995: *++bufp = '\0';
1.143 schwarze 996: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 997: if (NULL == ohash_find(&mlinks,
1.141 schwarze 998: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 999: goto nextlink;
1.89 schwarze 1000: if (warnings)
1001: say(mlink->file, "Man source exists: %s", buf);
1002: if (use_all)
1.90 schwarze 1003: goto nextlink;
1.89 schwarze 1004: *prev = mlink->next;
1005: mlink_free(mlink);
1.90 schwarze 1006: continue;
1007: nextlink:
1008: prev = &(*prev)->next;
1.89 schwarze 1009: }
1010: }
1011:
1.131 schwarze 1012: static void
1.93 schwarze 1013: mlink_check(struct mpage *mpage, struct mlink *mlink)
1014: {
1.131 schwarze 1015: struct str *str;
1016: unsigned int slot;
1.93 schwarze 1017:
1018: /*
1019: * Check whether the manual section given in a file
1020: * agrees with the directory where the file is located.
1021: * Some manuals have suffixes like (3p) on their
1022: * section number either inside the file or in the
1023: * directory name, some are linked into more than one
1024: * section, like encrypt(1) = makekey(8).
1025: */
1026:
1027: if (FORM_SRC == mpage->form &&
1.131 schwarze 1028: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1029: say(mlink->file, "Section \"%s\" manual in %s directory",
1030: mpage->sec, mlink->dsec);
1031:
1032: /*
1033: * Manual page directories exist for each kernel
1034: * architecture as returned by machine(1).
1035: * However, many manuals only depend on the
1036: * application architecture as returned by arch(1).
1037: * For example, some (2/ARM) manuals are shared
1038: * across the "armish" and "zaurus" kernel
1039: * architectures.
1040: * A few manuals are even shared across completely
1041: * different architectures, for example fdformat(1)
1042: * on amd64, i386, sparc, and sparc64.
1043: */
1044:
1.131 schwarze 1045: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1046: say(mlink->file, "Architecture \"%s\" manual in "
1047: "\"%s\" directory", mpage->arch, mlink->arch);
1048:
1.131 schwarze 1049: /*
1050: * XXX
1.133 schwarze 1051: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1052: */
1053:
1054: if (FORM_CAT == mpage->form)
1055: return;
1056:
1057: /*
1058: * Check whether this mlink
1059: * appears as a name in the NAME section.
1060: */
1.93 schwarze 1061:
1.133 schwarze 1062: slot = ohash_qlookup(&names, mlink->name);
1063: str = ohash_find(&names, slot);
1.131 schwarze 1064: assert(NULL != str);
1.133 schwarze 1065: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1066: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1067: }
1068:
1.89 schwarze 1069: /*
1.80 schwarze 1070: * Run through the files in the global vector "mpages"
1071: * and add them to the database specified in "basedir".
1.50 kristaps 1072: *
1073: * This handles the parsing scheme itself, using the cues of directory
1074: * and filename to determine whether the file is parsable or not.
1075: */
1.59 schwarze 1076: static void
1.101 schwarze 1077: mpages_merge(struct mchars *mc, struct mparse *mp)
1.50 kristaps 1078: {
1.113 schwarze 1079: char any[] = "any";
1.101 schwarze 1080: struct ohash_info str_info;
1.124 schwarze 1081: int fd[2];
1.118 schwarze 1082: struct mpage *mpage, *mpage_dest;
1083: struct mlink *mlink, *mlink_dest;
1.69 schwarze 1084: struct mdoc *mdoc;
1085: struct man *man;
1.118 schwarze 1086: char *sodest;
1.112 schwarze 1087: char *cp;
1.124 schwarze 1088: pid_t child_pid;
1.131 schwarze 1089: int status;
1.101 schwarze 1090: unsigned int pslot;
1.69 schwarze 1091: enum mandoclevel lvl;
1092:
1.71 schwarze 1093: str_info.alloc = hash_alloc;
1.152 schwarze 1094: str_info.calloc = hash_calloc;
1095: str_info.free = hash_free;
1.71 schwarze 1096: str_info.key_offset = offsetof(struct str, key);
1097:
1.107 schwarze 1098: if (0 == nodb)
1099: SQL_EXEC("BEGIN TRANSACTION");
1100:
1.80 schwarze 1101: mpage = ohash_first(&mpages, &pslot);
1102: while (NULL != mpage) {
1.89 schwarze 1103: mlinks_undupe(mpage);
1104: if (NULL == mpage->mlinks) {
1105: mpage = ohash_next(&mpages, &pslot);
1106: continue;
1.50 kristaps 1107: }
1108:
1.133 schwarze 1109: name_mask = NAME_MASK;
1110: ohash_init(&names, 4, &str_info);
1.71 schwarze 1111: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1112: mparse_reset(mp);
1113: mdoc = NULL;
1114: man = NULL;
1.124 schwarze 1115: sodest = NULL;
1116: child_pid = 0;
1117: fd[0] = -1;
1118: fd[1] = -1;
1119:
1120: if (mpage->mlinks->gzip) {
1121: if (-1 == pipe(fd)) {
1122: exitcode = (int)MANDOCLEVEL_SYSERR;
1123: say(mpage->mlinks->file, "&pipe gunzip");
1124: goto nextpage;
1125: }
1126: switch (child_pid = fork()) {
1.141 schwarze 1127: case -1:
1.124 schwarze 1128: exitcode = (int)MANDOCLEVEL_SYSERR;
1129: say(mpage->mlinks->file, "&fork gunzip");
1130: child_pid = 0;
1131: close(fd[1]);
1132: close(fd[0]);
1133: goto nextpage;
1.141 schwarze 1134: case 0:
1.124 schwarze 1135: close(fd[0]);
1136: if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1137: say(mpage->mlinks->file,
1138: "&dup gunzip");
1139: exit(1);
1140: }
1141: execlp("gunzip", "gunzip", "-c",
1142: mpage->mlinks->file, NULL);
1143: say(mpage->mlinks->file, "&exec gunzip");
1144: exit(1);
1145: default:
1146: close(fd[1]);
1147: break;
1148: }
1149: }
1.14 schwarze 1150:
1151: /*
1.33 schwarze 1152: * Try interpreting the file as mdoc(7) or man(7)
1153: * source code, unless it is already known to be
1154: * formatted. Fall back to formatted mode.
1.14 schwarze 1155: */
1.82 schwarze 1156: if (FORM_CAT != mpage->mlinks->dform ||
1.84 schwarze 1157: FORM_CAT != mpage->mlinks->fform) {
1.124 schwarze 1158: lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1.50 kristaps 1159: if (lvl < MANDOCLEVEL_FATAL)
1.118 schwarze 1160: mparse_result(mp, &mdoc, &man, &sodest);
1.76 schwarze 1161: }
1.14 schwarze 1162:
1.118 schwarze 1163: if (NULL != sodest) {
1164: mlink_dest = ohash_find(&mlinks,
1165: ohash_qlookup(&mlinks, sodest));
1166: if (NULL != mlink_dest) {
1167:
1168: /* The .so target exists. */
1169:
1170: mpage_dest = mlink_dest->mpage;
1171: mlink = mpage->mlinks;
1172: while (1) {
1173: mlink->mpage = mpage_dest;
1174:
1175: /*
1176: * If the target was already
1177: * processed, add the links
1178: * to the database now.
1179: * Otherwise, this will
1180: * happen when we come
1181: * to the target.
1182: */
1183:
1.137 schwarze 1184: if (mpage_dest->pageid)
1.118 schwarze 1185: dbadd_mlink(mlink);
1186:
1187: if (NULL == mlink->next)
1188: break;
1189: mlink = mlink->next;
1190: }
1191:
1192: /* Move all links to the target. */
1193:
1194: mlink->next = mlink_dest->next;
1195: mlink_dest->next = mpage->mlinks;
1196: mpage->mlinks = NULL;
1197: }
1.124 schwarze 1198: goto nextpage;
1.118 schwarze 1199: } else if (NULL != mdoc) {
1.82 schwarze 1200: mpage->form = FORM_SRC;
1201: mpage->sec =
1202: mandoc_strdup(mdoc_meta(mdoc)->msec);
1203: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 1204: mpage->arch = mandoc_strdup(
1205: NULL == mpage->arch ? "" : mpage->arch);
1.82 schwarze 1206: mpage->title =
1207: mandoc_strdup(mdoc_meta(mdoc)->title);
1.14 schwarze 1208: } else if (NULL != man) {
1.82 schwarze 1209: mpage->form = FORM_SRC;
1210: mpage->sec =
1211: mandoc_strdup(man_meta(man)->msec);
1212: mpage->arch =
1213: mandoc_strdup(mpage->mlinks->arch);
1214: mpage->title =
1215: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1216: } else {
1.82 schwarze 1217: mpage->form = FORM_CAT;
1218: mpage->sec =
1219: mandoc_strdup(mpage->mlinks->dsec);
1220: mpage->arch =
1221: mandoc_strdup(mpage->mlinks->arch);
1222: mpage->title =
1223: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1224: }
1.97 schwarze 1225: putkey(mpage, mpage->sec, TYPE_sec);
1.98 schwarze 1226: putkey(mpage, '\0' == *mpage->arch ?
1.113 schwarze 1227: any : mpage->arch, TYPE_arch);
1.50 kristaps 1228:
1.97 schwarze 1229: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1230: if ('\0' != *mlink->dsec)
1231: putkey(mpage, mlink->dsec, TYPE_sec);
1232: if ('\0' != *mlink->fsec)
1233: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1234: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1235: any : mlink->arch, TYPE_arch);
1.133 schwarze 1236: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1237: }
1.12 schwarze 1238:
1.135 schwarze 1239: assert(NULL == mpage->desc);
1.50 kristaps 1240: if (NULL != mdoc) {
1241: if (NULL != (cp = mdoc_meta(mdoc)->name))
1.133 schwarze 1242: putkey(mpage, cp, NAME_HEAD);
1.78 schwarze 1243: parse_mdoc(mpage, mdoc_node(mdoc));
1.50 kristaps 1244: } else if (NULL != man)
1.78 schwarze 1245: parse_man(mpage, man_node(man));
1.50 kristaps 1246: else
1.124 schwarze 1247: parse_cat(mpage, fd[0]);
1.135 schwarze 1248: if (NULL == mpage->desc)
1249: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1250:
1251: if (warnings && !use_all)
1252: for (mlink = mpage->mlinks; mlink;
1253: mlink = mlink->next)
1254: mlink_check(mpage, mlink);
1.44 kristaps 1255:
1.105 schwarze 1256: dbadd(mpage, mc);
1.124 schwarze 1257:
1258: nextpage:
1259: if (child_pid) {
1260: if (-1 == waitpid(child_pid, &status, 0)) {
1261: exitcode = (int)MANDOCLEVEL_SYSERR;
1262: say(mpage->mlinks->file, "&wait gunzip");
1263: } else if (WIFSIGNALED(status)) {
1264: exitcode = (int)MANDOCLEVEL_SYSERR;
1265: say(mpage->mlinks->file,
1266: "gunzip died from signal %d",
1267: WTERMSIG(status));
1268: } else if (WEXITSTATUS(status)) {
1269: exitcode = (int)MANDOCLEVEL_SYSERR;
1270: say(mpage->mlinks->file,
1271: "gunzip failed with code %d",
1272: WEXITSTATUS(status));
1273: }
1274: }
1.66 schwarze 1275: ohash_delete(&strings);
1.133 schwarze 1276: ohash_delete(&names);
1.80 schwarze 1277: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1278: }
1.107 schwarze 1279:
1280: if (0 == nodb)
1281: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1282: }
1283:
1284: static void
1285: names_check(void)
1286: {
1287: sqlite3_stmt *stmt;
1288: const char *name, *sec, *arch, *key;
1289: int irc;
1290:
1291: sqlite3_prepare_v2(db,
1292: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1293: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1294: "WHERE bits & ? AND NOT EXISTS ("
1295: "SELECT pageid FROM mlinks "
1.133 schwarze 1296: "WHERE mlinks.pageid == names.pageid "
1297: "AND mlinks.name == names.name"
1.130 schwarze 1298: ")"
1299: ") JOIN ("
1.142 schwarze 1300: "SELECT sec, arch, name, pageid FROM mlinks "
1301: "GROUP BY pageid"
1.130 schwarze 1302: ") USING (pageid);",
1303: -1, &stmt, NULL);
1304:
1.134 schwarze 1305: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1306: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1307:
1308: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1309: name = sqlite3_column_text(stmt, 0);
1310: sec = sqlite3_column_text(stmt, 1);
1311: arch = sqlite3_column_text(stmt, 2);
1312: key = sqlite3_column_text(stmt, 3);
1313: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1314: '\0' == *arch ? "" : "/",
1315: '\0' == *arch ? "" : arch, key);
1316: }
1317: sqlite3_finalize(stmt);
1.50 kristaps 1318: }
1.12 schwarze 1319:
1.50 kristaps 1320: static void
1.124 schwarze 1321: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1322: {
1323: FILE *stream;
1324: char *line, *p, *title;
1325: size_t len, plen, titlesz;
1.12 schwarze 1326:
1.124 schwarze 1327: stream = (-1 == fd) ?
1328: fopen(mpage->mlinks->file, "r") :
1329: fdopen(fd, "r");
1330: if (NULL == stream) {
1.56 schwarze 1331: if (warnings)
1.123 schwarze 1332: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1333: return;
1334: }
1.1 kristaps 1335:
1.50 kristaps 1336: /* Skip to first blank line. */
1.1 kristaps 1337:
1.50 kristaps 1338: while (NULL != (line = fgetln(stream, &len)))
1339: if ('\n' == *line)
1340: break;
1.1 kristaps 1341:
1.50 kristaps 1342: /*
1343: * Assume the first line that is not indented
1344: * is the first section header. Skip to it.
1345: */
1.1 kristaps 1346:
1.50 kristaps 1347: while (NULL != (line = fgetln(stream, &len)))
1348: if ('\n' != *line && ' ' != *line)
1349: break;
1.141 schwarze 1350:
1.50 kristaps 1351: /*
1352: * Read up until the next section into a buffer.
1353: * Strip the leading and trailing newline from each read line,
1354: * appending a trailing space.
1355: * Ignore empty (whitespace-only) lines.
1356: */
1.1 kristaps 1357:
1.50 kristaps 1358: titlesz = 0;
1359: title = NULL;
1.38 schwarze 1360:
1.50 kristaps 1361: while (NULL != (line = fgetln(stream, &len))) {
1362: if (' ' != *line || '\n' != line[len - 1])
1363: break;
1364: while (len > 0 && isspace((unsigned char)*line)) {
1365: line++;
1366: len--;
1367: }
1368: if (1 == len)
1.38 schwarze 1369: continue;
1.50 kristaps 1370: title = mandoc_realloc(title, titlesz + len);
1371: memcpy(title + titlesz, line, len);
1372: titlesz += len;
1373: title[titlesz - 1] = ' ';
1374: }
1.38 schwarze 1375:
1.50 kristaps 1376: /*
1377: * If no page content can be found, or the input line
1378: * is already the next section header, or there is no
1379: * trailing newline, reuse the page title as the page
1380: * description.
1381: */
1.44 kristaps 1382:
1.50 kristaps 1383: if (NULL == title || '\0' == *title) {
1.56 schwarze 1384: if (warnings)
1.82 schwarze 1385: say(mpage->mlinks->file,
1386: "Cannot find NAME section");
1.50 kristaps 1387: fclose(stream);
1388: free(title);
1389: return;
1390: }
1.1 kristaps 1391:
1.50 kristaps 1392: title = mandoc_realloc(title, titlesz + 1);
1393: title[titlesz] = '\0';
1.33 schwarze 1394:
1.50 kristaps 1395: /*
1396: * Skip to the first dash.
1397: * Use the remaining line as the description (no more than 70
1398: * bytes).
1399: */
1.33 schwarze 1400:
1.50 kristaps 1401: if (NULL != (p = strstr(title, "- "))) {
1402: for (p += 2; ' ' == *p || '\b' == *p; p++)
1403: /* Skip to next word. */ ;
1404: } else {
1.56 schwarze 1405: if (warnings)
1.82 schwarze 1406: say(mpage->mlinks->file,
1407: "No dash in title line");
1.50 kristaps 1408: p = title;
1409: }
1.38 schwarze 1410:
1.50 kristaps 1411: plen = strlen(p);
1.1 kristaps 1412:
1.50 kristaps 1413: /* Strip backspace-encoding from line. */
1.1 kristaps 1414:
1.50 kristaps 1415: while (NULL != (line = memchr(p, '\b', plen))) {
1416: len = line - p;
1417: if (0 == len) {
1418: memmove(line, line + 1, plen--);
1419: continue;
1.141 schwarze 1420: }
1.50 kristaps 1421: memmove(line - 1, line + 1, plen - len);
1422: plen -= 2;
1423: }
1.1 kristaps 1424:
1.78 schwarze 1425: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1426: fclose(stream);
1427: free(title);
1428: }
1.1 kristaps 1429:
1.50 kristaps 1430: /*
1431: * Put a type/word pair into the word database for this particular file.
1432: */
1433: static void
1.112 schwarze 1434: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1435: {
1.112 schwarze 1436: char *cp;
1.18 kristaps 1437:
1.50 kristaps 1438: assert(NULL != value);
1.112 schwarze 1439: if (TYPE_arch == type)
1440: for (cp = value; *cp; cp++)
1441: if (isupper((unsigned char)*cp))
1442: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1443: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1444: }
1445:
1446: /*
1.50 kristaps 1447: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1448: */
1449: static void
1.78 schwarze 1450: putmdockey(const struct mpage *mpage,
1451: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1452: {
1.18 kristaps 1453:
1.50 kristaps 1454: for ( ; NULL != n; n = n->next) {
1455: if (NULL != n->child)
1.78 schwarze 1456: putmdockey(mpage, n->child, m);
1.50 kristaps 1457: if (MDOC_TEXT == n->type)
1.78 schwarze 1458: putkey(mpage, n->string, m);
1.50 kristaps 1459: }
1460: }
1.18 kristaps 1461:
1.61 schwarze 1462: static void
1.78 schwarze 1463: parse_man(struct mpage *mpage, const struct man_node *n)
1.50 kristaps 1464: {
1465: const struct man_node *head, *body;
1.121 schwarze 1466: char *start, *title;
1.50 kristaps 1467: char byte;
1.121 schwarze 1468: size_t sz;
1.18 kristaps 1469:
1.50 kristaps 1470: if (NULL == n)
1.61 schwarze 1471: return;
1.18 kristaps 1472:
1.50 kristaps 1473: /*
1474: * We're only searching for one thing: the first text child in
1475: * the BODY of a NAME section. Since we don't keep track of
1476: * sections in -man, run some hoops to find out whether we're in
1477: * the correct section or not.
1478: */
1.18 kristaps 1479:
1.50 kristaps 1480: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1481: body = n;
1482: assert(body->parent);
1483: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1484: 1 == head->nchild &&
1485: NULL != (head = (head->child)) &&
1486: MAN_TEXT == head->type &&
1487: 0 == strcmp(head->string, "NAME") &&
1488: NULL != body->child) {
1.3 kristaps 1489:
1.50 kristaps 1490: /*
1491: * Suck the entire NAME section into memory.
1492: * Yes, we might run away.
1493: * But too many manuals have big, spread-out
1494: * NAME sections over many lines.
1495: */
1.3 kristaps 1496:
1.121 schwarze 1497: title = NULL;
1498: man_deroff(&title, body);
1.50 kristaps 1499: if (NULL == title)
1.61 schwarze 1500: return;
1.18 kristaps 1501:
1.141 schwarze 1502: /*
1.50 kristaps 1503: * Go through a special heuristic dance here.
1504: * Conventionally, one or more manual names are
1505: * comma-specified prior to a whitespace, then a
1506: * dash, then a description. Try to puzzle out
1507: * the name parts here.
1508: */
1.18 kristaps 1509:
1.121 schwarze 1510: start = title;
1.50 kristaps 1511: for ( ;; ) {
1512: sz = strcspn(start, " ,");
1513: if ('\0' == start[sz])
1514: break;
1.1 kristaps 1515:
1.50 kristaps 1516: byte = start[sz];
1517: start[sz] = '\0';
1.110 schwarze 1518:
1519: /*
1520: * Assume a stray trailing comma in the
1521: * name list if a name begins with a dash.
1522: */
1523:
1524: if ('-' == start[0] ||
1525: ('\\' == start[0] && '-' == start[1]))
1526: break;
1.1 kristaps 1527:
1.133 schwarze 1528: putkey(mpage, start, NAME_TITLE);
1.1 kristaps 1529:
1.50 kristaps 1530: if (' ' == byte) {
1531: start += sz + 1;
1532: break;
1533: }
1.1 kristaps 1534:
1.50 kristaps 1535: assert(',' == byte);
1536: start += sz + 1;
1537: while (' ' == *start)
1538: start++;
1539: }
1.1 kristaps 1540:
1.121 schwarze 1541: if (start == title) {
1.133 schwarze 1542: putkey(mpage, start, NAME_TITLE);
1.50 kristaps 1543: free(title);
1.61 schwarze 1544: return;
1.50 kristaps 1545: }
1.1 kristaps 1546:
1.50 kristaps 1547: while (isspace((unsigned char)*start))
1548: start++;
1.1 kristaps 1549:
1.50 kristaps 1550: if (0 == strncmp(start, "-", 1))
1551: start += 1;
1552: else if (0 == strncmp(start, "\\-\\-", 4))
1553: start += 4;
1554: else if (0 == strncmp(start, "\\-", 2))
1555: start += 2;
1556: else if (0 == strncmp(start, "\\(en", 4))
1557: start += 4;
1558: else if (0 == strncmp(start, "\\(em", 4))
1559: start += 4;
1.1 kristaps 1560:
1.50 kristaps 1561: while (' ' == *start)
1562: start++;
1.1 kristaps 1563:
1.78 schwarze 1564: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1565: free(title);
1.61 schwarze 1566: return;
1.50 kristaps 1567: }
1568: }
1.1 kristaps 1569:
1.77 schwarze 1570: for (n = n->child; n; n = n->next) {
1.78 schwarze 1571: if (NULL != mpage->desc)
1.77 schwarze 1572: break;
1.78 schwarze 1573: parse_man(mpage, n);
1.77 schwarze 1574: }
1.1 kristaps 1575: }
1576:
1577: static void
1.78 schwarze 1578: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1579: {
1580:
1.50 kristaps 1581: assert(NULL != n);
1582: for (n = n->child; NULL != n; n = n->next) {
1583: switch (n->type) {
1.141 schwarze 1584: case MDOC_ELEM:
1.50 kristaps 1585: /* FALLTHROUGH */
1.141 schwarze 1586: case MDOC_BLOCK:
1.50 kristaps 1587: /* FALLTHROUGH */
1.141 schwarze 1588: case MDOC_HEAD:
1.50 kristaps 1589: /* FALLTHROUGH */
1.141 schwarze 1590: case MDOC_BODY:
1.50 kristaps 1591: /* FALLTHROUGH */
1.141 schwarze 1592: case MDOC_TAIL:
1.50 kristaps 1593: if (NULL != mdocs[n->tok].fp)
1.78 schwarze 1594: if (0 == (*mdocs[n->tok].fp)(mpage, n))
1.50 kristaps 1595: break;
1.68 schwarze 1596: if (mdocs[n->tok].mask)
1.78 schwarze 1597: putmdockey(mpage, n->child,
1598: mdocs[n->tok].mask);
1.50 kristaps 1599: break;
1600: default:
1601: assert(MDOC_ROOT != n->type);
1602: continue;
1603: }
1604: if (NULL != n->child)
1.78 schwarze 1605: parse_mdoc(mpage, n);
1.1 kristaps 1606: }
1607: }
1608:
1.25 schwarze 1609: static int
1.78 schwarze 1610: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1611: {
1612: const char *start, *end;
1613: size_t sz;
1.25 schwarze 1614:
1.50 kristaps 1615: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1616: NULL == (n = n->child) ||
1617: MDOC_TEXT != n->type)
1.25 schwarze 1618: return(0);
1.1 kristaps 1619:
1620: /*
1621: * Only consider those `Fd' macro fields that begin with an
1622: * "inclusion" token (versus, e.g., #define).
1623: */
1.50 kristaps 1624:
1.1 kristaps 1625: if (strcmp("#include", n->string))
1.25 schwarze 1626: return(0);
1.1 kristaps 1627:
1628: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1629: return(0);
1.1 kristaps 1630:
1631: /*
1632: * Strip away the enclosing angle brackets and make sure we're
1633: * not zero-length.
1634: */
1635:
1636: start = n->string;
1637: if ('<' == *start || '"' == *start)
1638: start++;
1639:
1640: if (0 == (sz = strlen(start)))
1.25 schwarze 1641: return(0);
1.1 kristaps 1642:
1643: end = &start[(int)sz - 1];
1644: if ('>' == *end || '"' == *end)
1645: end--;
1646:
1.50 kristaps 1647: if (end > start)
1.78 schwarze 1648: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1649: return(0);
1.1 kristaps 1650: }
1651:
1.25 schwarze 1652: static int
1.78 schwarze 1653: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1654: {
1.112 schwarze 1655: char *cp;
1.1 kristaps 1656:
1.50 kristaps 1657: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.25 schwarze 1658: return(0);
1659:
1.141 schwarze 1660: /*
1.50 kristaps 1661: * Parse: .Fn "struct type *name" "char *arg".
1.141 schwarze 1662: * First strip away pointer symbol.
1.50 kristaps 1663: * Then store the function name, then type.
1.141 schwarze 1664: * Finally, store the arguments.
1.50 kristaps 1665: */
1.1 kristaps 1666:
1.50 kristaps 1667: if (NULL == (cp = strrchr(n->string, ' ')))
1668: cp = n->string;
1.1 kristaps 1669:
1670: while ('*' == *cp)
1671: cp++;
1672:
1.78 schwarze 1673: putkey(mpage, cp, TYPE_Fn);
1.25 schwarze 1674:
1.50 kristaps 1675: if (n->string < cp)
1.78 schwarze 1676: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.25 schwarze 1677:
1.50 kristaps 1678: for (n = n->next; NULL != n; n = n->next)
1679: if (MDOC_TEXT == n->type)
1.78 schwarze 1680: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1681:
1682: return(0);
1.1 kristaps 1683: }
1684:
1.25 schwarze 1685: static int
1.78 schwarze 1686: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1687: {
1.67 schwarze 1688: char *cp;
1.1 kristaps 1689:
1690: if (NULL == (n = n->child))
1.25 schwarze 1691: return(0);
1.1 kristaps 1692:
1.67 schwarze 1693: if (NULL == n->next) {
1.78 schwarze 1694: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1695: return(0);
1696: }
1697:
1.120 schwarze 1698: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1699: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1700: free(cp);
1701: return(0);
1.1 kristaps 1702: }
1703:
1.25 schwarze 1704: static int
1.78 schwarze 1705: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1706: {
1707:
1.122 schwarze 1708: if (MDOC_BODY == n->type)
1709: mdoc_deroff(&mpage->desc, n);
1710: return(0);
1.1 kristaps 1711: }
1712:
1.25 schwarze 1713: static int
1.78 schwarze 1714: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1715: {
1716:
1.129 schwarze 1717: if (SEC_NAME == n->sec)
1.133 schwarze 1718: putmdockey(mpage, n->child, NAME_TITLE);
1.129 schwarze 1719: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1.133 schwarze 1720: putmdockey(mpage, n->child, NAME_SYN);
1.129 schwarze 1721: return(0);
1.1 kristaps 1722: }
1723:
1.25 schwarze 1724: static int
1.78 schwarze 1725: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1726: {
1727:
1.25 schwarze 1728: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1729: }
1730:
1.50 kristaps 1731: static int
1.78 schwarze 1732: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1733: {
1734:
1.50 kristaps 1735: return(MDOC_HEAD == n->type);
1736: }
1.1 kristaps 1737:
1.50 kristaps 1738: static int
1.78 schwarze 1739: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1.50 kristaps 1740: {
1.1 kristaps 1741:
1.50 kristaps 1742: return(MDOC_BODY == n->type);
1.1 kristaps 1743: }
1744:
1.50 kristaps 1745: /*
1.66 schwarze 1746: * Add a string to the hash table for the current manual.
1747: * Each string has a bitmask telling which macros it belongs to.
1748: * When we finish the manual, we'll dump the table.
1.50 kristaps 1749: */
1750: static void
1.78 schwarze 1751: putkeys(const struct mpage *mpage,
1752: const char *cp, size_t sz, uint64_t v)
1.50 kristaps 1753: {
1.133 schwarze 1754: struct ohash *htab;
1.50 kristaps 1755: struct str *s;
1.111 schwarze 1756: const char *end;
1.65 schwarze 1757: unsigned int slot;
1.111 schwarze 1758: int i;
1.25 schwarze 1759:
1.50 kristaps 1760: if (0 == sz)
1761: return;
1.111 schwarze 1762:
1.133 schwarze 1763: if (TYPE_Nm & v) {
1764: htab = &names;
1765: v &= name_mask;
1766: name_mask &= ~NAME_FIRST;
1767: if (debug > 1)
1768: say(mpage->mlinks->file,
1769: "Adding name %*s", sz, cp);
1770: } else {
1771: htab = &strings;
1772: if (debug > 1)
1773: for (i = 0; i < mansearch_keymax; i++)
1774: if (1 << i & v)
1775: say(mpage->mlinks->file,
1776: "Adding key %s=%*s",
1777: mansearch_keynames[i], sz, cp);
1.111 schwarze 1778: }
1.25 schwarze 1779:
1.65 schwarze 1780: end = cp + sz;
1.133 schwarze 1781: slot = ohash_qlookupi(htab, cp, &end);
1782: s = ohash_find(htab, slot);
1.25 schwarze 1783:
1.78 schwarze 1784: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1785: s->mask |= v;
1786: return;
1787: } else if (NULL == s) {
1.144 schwarze 1788: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1789: memcpy(s->key, cp, sz);
1.133 schwarze 1790: ohash_insert(htab, slot, s);
1.1 kristaps 1791: }
1.78 schwarze 1792: s->mpage = mpage;
1.50 kristaps 1793: s->mask = v;
1.1 kristaps 1794: }
1795:
1.50 kristaps 1796: /*
1797: * Take a Unicode codepoint and produce its UTF-8 encoding.
1798: * This isn't the best way to do this, but it works.
1799: * The magic numbers are from the UTF-8 packaging.
1800: * They're not as scary as they seem: read the UTF-8 spec for details.
1801: */
1802: static size_t
1803: utf8(unsigned int cp, char out[7])
1.1 kristaps 1804: {
1.50 kristaps 1805: size_t rc;
1.1 kristaps 1806:
1.50 kristaps 1807: rc = 0;
1808: if (cp <= 0x0000007F) {
1809: rc = 1;
1810: out[0] = (char)cp;
1811: } else if (cp <= 0x000007FF) {
1812: rc = 2;
1813: out[0] = (cp >> 6 & 31) | 192;
1814: out[1] = (cp & 63) | 128;
1815: } else if (cp <= 0x0000FFFF) {
1816: rc = 3;
1817: out[0] = (cp >> 12 & 15) | 224;
1818: out[1] = (cp >> 6 & 63) | 128;
1819: out[2] = (cp & 63) | 128;
1820: } else if (cp <= 0x001FFFFF) {
1821: rc = 4;
1822: out[0] = (cp >> 18 & 7) | 240;
1823: out[1] = (cp >> 12 & 63) | 128;
1824: out[2] = (cp >> 6 & 63) | 128;
1825: out[3] = (cp & 63) | 128;
1826: } else if (cp <= 0x03FFFFFF) {
1827: rc = 5;
1828: out[0] = (cp >> 24 & 3) | 248;
1829: out[1] = (cp >> 18 & 63) | 128;
1830: out[2] = (cp >> 12 & 63) | 128;
1831: out[3] = (cp >> 6 & 63) | 128;
1832: out[4] = (cp & 63) | 128;
1833: } else if (cp <= 0x7FFFFFFF) {
1834: rc = 6;
1835: out[0] = (cp >> 30 & 1) | 252;
1836: out[1] = (cp >> 24 & 63) | 128;
1837: out[2] = (cp >> 18 & 63) | 128;
1838: out[3] = (cp >> 12 & 63) | 128;
1839: out[4] = (cp >> 6 & 63) | 128;
1840: out[5] = (cp & 63) | 128;
1841: } else
1.1 kristaps 1842: return(0);
1843:
1.50 kristaps 1844: out[rc] = '\0';
1845: return(rc);
1846: }
1.1 kristaps 1847:
1.50 kristaps 1848: /*
1.96 schwarze 1849: * Store the rendered version of a key, or alias the pointer
1850: * if the key contains no escape sequences.
1.50 kristaps 1851: */
1852: static void
1.96 schwarze 1853: render_key(struct mchars *mc, struct str *key)
1.50 kristaps 1854: {
1855: size_t sz, bsz, pos;
1.114 schwarze 1856: char utfbuf[7], res[6];
1.50 kristaps 1857: char *buf;
1858: const char *seq, *cpp, *val;
1859: int len, u;
1860: enum mandoc_esc esc;
1861:
1.96 schwarze 1862: assert(NULL == key->rendered);
1.50 kristaps 1863:
1864: res[0] = '\\';
1865: res[1] = '\t';
1866: res[2] = ASCII_NBRSP;
1867: res[3] = ASCII_HYPH;
1.114 schwarze 1868: res[4] = ASCII_BREAK;
1869: res[5] = '\0';
1.1 kristaps 1870:
1.50 kristaps 1871: val = key->key;
1872: bsz = strlen(val);
1.46 kristaps 1873:
1.50 kristaps 1874: /*
1875: * Pre-check: if we have no stop-characters, then set the
1876: * pointer as ourselvse and get out of here.
1877: */
1878: if (strcspn(val, res) == bsz) {
1.96 schwarze 1879: key->rendered = key->key;
1.50 kristaps 1880: return;
1.141 schwarze 1881: }
1.46 kristaps 1882:
1.50 kristaps 1883: /* Pre-allocate by the length of the input */
1.46 kristaps 1884:
1.50 kristaps 1885: buf = mandoc_malloc(++bsz);
1886: pos = 0;
1.46 kristaps 1887:
1.50 kristaps 1888: while ('\0' != *val) {
1889: /*
1890: * Halt on the first escape sequence.
1891: * This also halts on the end of string, in which case
1892: * we just copy, fallthrough, and exit the loop.
1893: */
1894: if ((sz = strcspn(val, res)) > 0) {
1895: memcpy(&buf[pos], val, sz);
1896: pos += sz;
1897: val += sz;
1898: }
1.46 kristaps 1899:
1.114 schwarze 1900: switch (*val) {
1.141 schwarze 1901: case ASCII_HYPH:
1.50 kristaps 1902: buf[pos++] = '-';
1903: val++;
1904: continue;
1.141 schwarze 1905: case '\t':
1.114 schwarze 1906: /* FALLTHROUGH */
1.141 schwarze 1907: case ASCII_NBRSP:
1.50 kristaps 1908: buf[pos++] = ' ';
1909: val++;
1.114 schwarze 1910: /* FALLTHROUGH */
1.141 schwarze 1911: case ASCII_BREAK:
1.50 kristaps 1912: continue;
1.114 schwarze 1913: default:
1914: break;
1915: }
1916: if ('\\' != *val)
1.50 kristaps 1917: break;
1.46 kristaps 1918:
1.50 kristaps 1919: /* Read past the slash. */
1.46 kristaps 1920:
1.50 kristaps 1921: val++;
1.46 kristaps 1922:
1.50 kristaps 1923: /*
1924: * Parse the escape sequence and see if it's a
1925: * predefined character or special character.
1926: */
1.95 schwarze 1927:
1.141 schwarze 1928: esc = mandoc_escape((const char **)&val,
1929: &seq, &len);
1.50 kristaps 1930: if (ESCAPE_ERROR == esc)
1931: break;
1932: if (ESCAPE_SPECIAL != esc)
1933: continue;
1.1 kristaps 1934:
1.50 kristaps 1935: /*
1.95 schwarze 1936: * Render the special character
1937: * as either UTF-8 or ASCII.
1.50 kristaps 1938: */
1.95 schwarze 1939:
1940: if (write_utf8) {
1941: if (0 == (u = mchars_spec2cp(mc, seq, len)))
1942: continue;
1943: cpp = utfbuf;
1944: if (0 == (sz = utf8(u, utfbuf)))
1945: continue;
1946: sz = strlen(cpp);
1947: } else {
1948: cpp = mchars_spec2str(mc, seq, len, &sz);
1949: if (NULL == cpp)
1950: continue;
1951: if (ASCII_NBRSP == *cpp) {
1952: cpp = " ";
1953: sz = 1;
1954: }
1955: }
1.1 kristaps 1956:
1.50 kristaps 1957: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1958:
1.50 kristaps 1959: bsz += sz;
1960: buf = mandoc_realloc(buf, bsz);
1961: memcpy(&buf[pos], cpp, sz);
1962: pos += sz;
1.1 kristaps 1963: }
1964:
1.50 kristaps 1965: buf[pos] = '\0';
1.96 schwarze 1966: key->rendered = buf;
1.1 kristaps 1967: }
1968:
1.118 schwarze 1969: static void
1970: dbadd_mlink(const struct mlink *mlink)
1971: {
1972: size_t i;
1973:
1974: i = 1;
1975: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1976: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1977: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 1978: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 1979: SQL_STEP(stmts[STMT_INSERT_LINK]);
1980: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1981: }
1982:
1.14 schwarze 1983: /*
1.50 kristaps 1984: * Flush the current page's terms (and their bits) into the database.
1985: * Wrap the entire set of additions in a transaction to make sqlite be a
1986: * little faster.
1.96 schwarze 1987: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 1988: */
1989: static void
1.118 schwarze 1990: dbadd(struct mpage *mpage, struct mchars *mc)
1.14 schwarze 1991: {
1.87 schwarze 1992: struct mlink *mlink;
1.50 kristaps 1993: struct str *key;
1.52 kristaps 1994: size_t i;
1.66 schwarze 1995: unsigned int slot;
1.14 schwarze 1996:
1.128 schwarze 1997: mlink = mpage->mlinks;
1.43 kristaps 1998:
1.128 schwarze 1999: if (nodb) {
1.147 schwarze 2000: for (key = ohash_first(&names, &slot); NULL != key;
2001: key = ohash_next(&names, &slot)) {
2002: if (key->rendered != key->key)
2003: free(key->rendered);
2004: free(key);
2005: }
2006: for (key = ohash_first(&strings, &slot); NULL != key;
2007: key = ohash_next(&strings, &slot)) {
2008: if (key->rendered != key->key)
2009: free(key->rendered);
2010: free(key);
2011: }
1.145 schwarze 2012: if (0 == debug)
2013: return;
1.128 schwarze 2014: while (NULL != mlink) {
2015: fputs(mlink->name, stdout);
2016: if (NULL == mlink->next ||
2017: strcmp(mlink->dsec, mlink->next->dsec) ||
2018: strcmp(mlink->fsec, mlink->next->fsec) ||
2019: strcmp(mlink->arch, mlink->next->arch)) {
2020: putchar('(');
2021: if ('\0' == *mlink->dsec)
2022: fputs(mlink->fsec, stdout);
2023: else
2024: fputs(mlink->dsec, stdout);
2025: if ('\0' != *mlink->arch)
2026: printf("/%s", mlink->arch);
2027: putchar(')');
2028: }
2029: mlink = mlink->next;
2030: if (NULL != mlink)
2031: fputs(", ", stdout);
2032: }
1.132 schwarze 2033: printf(" - %s\n", mpage->desc);
1.14 schwarze 2034: return;
1.128 schwarze 2035: }
2036:
2037: if (debug)
2038: say(mlink->file, "Adding to database");
1.28 kristaps 2039:
1.148 schwarze 2040: i = strlen(mpage->desc) + 1;
2041: key = mandoc_calloc(1, sizeof(struct str) + i);
2042: memcpy(key->key, mpage->desc, i);
2043: render_key(mc, key);
2044:
1.52 kristaps 2045: i = 1;
1.148 schwarze 2046: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.82 schwarze 2047: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1.81 schwarze 2048: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2049: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2050: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.148 schwarze 2051:
2052: if (key->rendered != key->key)
2053: free(key->rendered);
2054: free(key);
1.81 schwarze 2055:
1.128 schwarze 2056: while (NULL != mlink) {
1.118 schwarze 2057: dbadd_mlink(mlink);
1.128 schwarze 2058: mlink = mlink->next;
2059: }
1.134 schwarze 2060: mlink = mpage->mlinks;
1.50 kristaps 2061:
1.133 schwarze 2062: for (key = ohash_first(&names, &slot); NULL != key;
2063: key = ohash_next(&names, &slot)) {
2064: assert(key->mpage == mpage);
2065: if (NULL == key->rendered)
2066: render_key(mc, key);
2067: i = 1;
2068: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2069: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.137 schwarze 2070: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2071: SQL_STEP(stmts[STMT_INSERT_NAME]);
2072: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2073: if (key->rendered != key->key)
2074: free(key->rendered);
2075: free(key);
2076: }
1.66 schwarze 2077: for (key = ohash_first(&strings, &slot); NULL != key;
2078: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2079: assert(key->mpage == mpage);
1.96 schwarze 2080: if (NULL == key->rendered)
2081: render_key(mc, key);
1.52 kristaps 2082: i = 1;
2083: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.96 schwarze 2084: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.137 schwarze 2085: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2086: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2087: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.96 schwarze 2088: if (key->rendered != key->key)
2089: free(key->rendered);
1.66 schwarze 2090: free(key);
1.38 schwarze 2091: }
1.14 schwarze 2092: }
2093:
1.5 kristaps 2094: static void
1.59 schwarze 2095: dbprune(void)
1.5 kristaps 2096: {
1.78 schwarze 2097: struct mpage *mpage;
1.82 schwarze 2098: struct mlink *mlink;
1.52 kristaps 2099: size_t i;
1.80 schwarze 2100: unsigned int slot;
1.5 kristaps 2101:
1.106 schwarze 2102: if (0 == nodb)
2103: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2104:
1.106 schwarze 2105: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2106: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2107: mlink = mpage->mlinks;
1.125 schwarze 2108: if (debug)
1.106 schwarze 2109: say(mlink->file, "Deleting from database");
2110: if (nodb)
2111: continue;
2112: for ( ; NULL != mlink; mlink = mlink->next) {
2113: i = 1;
2114: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2115: i, mlink->dsec);
2116: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2117: i, mlink->arch);
2118: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2119: i, mlink->name);
2120: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2121: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2122: }
1.5 kristaps 2123: }
1.106 schwarze 2124:
2125: if (0 == nodb)
2126: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2127: }
2128:
1.4 kristaps 2129: /*
1.50 kristaps 2130: * Close an existing database and its prepared statements.
2131: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2132: */
1.35 kristaps 2133: static void
1.59 schwarze 2134: dbclose(int real)
1.4 kristaps 2135: {
1.50 kristaps 2136: size_t i;
1.115 schwarze 2137: int status;
2138: pid_t child;
1.4 kristaps 2139:
1.50 kristaps 2140: if (nodb)
1.38 schwarze 2141: return;
1.50 kristaps 2142:
2143: for (i = 0; i < STMT__MAX; i++) {
2144: sqlite3_finalize(stmts[i]);
2145: stmts[i] = NULL;
1.4 kristaps 2146: }
2147:
1.50 kristaps 2148: sqlite3_close(db);
2149: db = NULL;
1.12 schwarze 2150:
1.50 kristaps 2151: if (real)
2152: return;
1.12 schwarze 2153:
1.115 schwarze 2154: if ('\0' == *tempfilename) {
2155: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2156: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2157: say(MANDOC_DB, "&rename");
1.115 schwarze 2158: }
2159: return;
2160: }
2161:
2162: switch (child = fork()) {
1.141 schwarze 2163: case -1:
1.115 schwarze 2164: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2165: say("", "&fork cmp");
1.115 schwarze 2166: return;
1.141 schwarze 2167: case 0:
1.115 schwarze 2168: execlp("cmp", "cmp", "-s",
2169: tempfilename, MANDOC_DB, NULL);
1.123 schwarze 2170: say("", "&exec cmp");
1.115 schwarze 2171: exit(0);
2172: default:
2173: break;
2174: }
2175: if (-1 == waitpid(child, &status, 0)) {
2176: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2177: say("", "&wait cmp");
1.115 schwarze 2178: } else if (WIFSIGNALED(status)) {
2179: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2180: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2181: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2182: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2183: say(MANDOC_DB,
2184: "Data changed, but cannot replace database");
2185: }
2186:
2187: *strrchr(tempfilename, '/') = '\0';
2188: switch (child = fork()) {
1.141 schwarze 2189: case -1:
1.115 schwarze 2190: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2191: say("", "&fork rm");
1.115 schwarze 2192: return;
1.141 schwarze 2193: case 0:
1.115 schwarze 2194: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.123 schwarze 2195: say("", "&exec rm");
1.115 schwarze 2196: exit((int)MANDOCLEVEL_SYSERR);
2197: default:
2198: break;
2199: }
2200: if (-1 == waitpid(child, &status, 0)) {
2201: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2202: say("", "&wait rm");
1.115 schwarze 2203: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2204: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2205: say("", "%s: Cannot remove temporary directory",
2206: tempfilename);
1.59 schwarze 2207: }
1.50 kristaps 2208: }
1.14 schwarze 2209:
1.50 kristaps 2210: /*
2211: * This is straightforward stuff.
2212: * Open a database connection to a "temporary" database, then open a set
2213: * of prepared statements we'll use over and over again.
2214: * If "real" is set, we use the existing database; if not, we truncate a
2215: * temporary one.
2216: * Must be matched by dbclose().
2217: */
2218: static int
1.59 schwarze 2219: dbopen(int real)
1.50 kristaps 2220: {
1.115 schwarze 2221: const char *sql;
1.50 kristaps 2222: int rc, ofl;
1.12 schwarze 2223:
1.141 schwarze 2224: if (nodb)
1.50 kristaps 2225: return(1);
1.12 schwarze 2226:
1.115 schwarze 2227: *tempfilename = '\0';
1.63 schwarze 2228: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2229:
2230: if (real) {
2231: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2232: if (SQLITE_OK != rc) {
1.63 schwarze 2233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2234: if (SQLITE_CANTOPEN != rc)
2235: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.63 schwarze 2236: return(0);
2237: }
1.115 schwarze 2238: goto prepare_statements;
2239: }
2240:
2241: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2242:
1.115 schwarze 2243: remove(MANDOC_DB "~");
2244: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2245: if (SQLITE_OK == rc)
1.115 schwarze 2246: goto create_tables;
1.116 schwarze 2247: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2248: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2249: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.50 kristaps 2250: return(0);
2251: }
1.12 schwarze 2252:
1.143 schwarze 2253: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2254: sizeof(tempfilename));
1.115 schwarze 2255: if (NULL == mkdtemp(tempfilename)) {
2256: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2257: say("", "&%s", tempfilename);
1.115 schwarze 2258: return(0);
2259: }
1.143 schwarze 2260: (void)strlcat(tempfilename, "/" MANDOC_DB,
2261: sizeof(tempfilename));
1.115 schwarze 2262: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2263: if (SQLITE_OK != rc) {
1.59 schwarze 2264: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2265: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.50 kristaps 2266: return(0);
2267: }
1.12 schwarze 2268:
1.115 schwarze 2269: create_tables:
1.81 schwarze 2270: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2271: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2272: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2273: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2274: ");\n"
2275: "\n"
1.81 schwarze 2276: "CREATE TABLE \"mlinks\" (\n"
2277: " \"sec\" TEXT NOT NULL,\n"
2278: " \"arch\" TEXT NOT NULL,\n"
2279: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2280: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2281: "ON DELETE CASCADE\n"
1.81 schwarze 2282: ");\n"
1.136 schwarze 2283: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2284: "\n"
1.133 schwarze 2285: "CREATE TABLE \"names\" (\n"
2286: " \"bits\" INTEGER NOT NULL,\n"
2287: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2288: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.133 schwarze 2289: "ON DELETE CASCADE\n"
2290: ");\n"
2291: "\n"
1.50 kristaps 2292: "CREATE TABLE \"keys\" (\n"
2293: " \"bits\" INTEGER NOT NULL,\n"
2294: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2295: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2296: "ON DELETE CASCADE\n"
1.136 schwarze 2297: ");\n"
2298: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2299:
1.50 kristaps 2300: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2301: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2302: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2303: sqlite3_close(db);
1.50 kristaps 2304: return(0);
2305: }
1.4 kristaps 2306:
1.57 schwarze 2307: prepare_statements:
1.146 schwarze 2308: if (SQLITE_OK != sqlite3_exec(db,
2309: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2310: exitcode = (int)MANDOCLEVEL_SYSERR;
2311: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2312: sqlite3_errmsg(db));
2313: sqlite3_close(db);
2314: return(0);
2315: }
2316:
1.137 schwarze 2317: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2318: "(SELECT pageid FROM mlinks WHERE "
2319: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2320: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2321: sql = "INSERT INTO mpages "
1.132 schwarze 2322: "(desc,form) VALUES (?,?)";
1.81 schwarze 2323: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2324: sql = "INSERT INTO mlinks "
1.104 schwarze 2325: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2326: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.133 schwarze 2327: sql = "INSERT INTO names "
2328: "(bits,name,pageid) VALUES (?,?,?)";
2329: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2330: sql = "INSERT INTO keys "
1.81 schwarze 2331: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2332: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2333:
2334: #ifndef __APPLE__
2335: /*
2336: * When opening a new database, we can turn off
2337: * synchronous mode for much better performance.
2338: */
2339:
1.146 schwarze 2340: if (real && SQLITE_OK != sqlite3_exec(db,
2341: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2342: exitcode = (int)MANDOCLEVEL_SYSERR;
2343: say(MANDOC_DB, "PRAGMA synchronous: %s",
2344: sqlite3_errmsg(db));
2345: sqlite3_close(db);
2346: return(0);
2347: }
1.70 schwarze 2348: #endif
2349:
1.50 kristaps 2350: return(1);
2351: }
1.5 kristaps 2352:
1.50 kristaps 2353: static void *
1.152 schwarze 2354: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2355: {
1.12 schwarze 2356:
1.152 schwarze 2357: return(mandoc_calloc(n, sz));
1.50 kristaps 2358: }
1.12 schwarze 2359:
1.50 kristaps 2360: static void *
2361: hash_alloc(size_t sz, void *arg)
2362: {
1.5 kristaps 2363:
1.50 kristaps 2364: return(mandoc_malloc(sz));
2365: }
1.41 kristaps 2366:
1.50 kristaps 2367: static void
1.152 schwarze 2368: hash_free(void *p, void *arg)
1.50 kristaps 2369: {
1.4 kristaps 2370:
1.50 kristaps 2371: free(p);
1.4 kristaps 2372: }
2373:
1.50 kristaps 2374: static int
1.59 schwarze 2375: set_basedir(const char *targetdir)
1.4 kristaps 2376: {
1.59 schwarze 2377: static char startdir[PATH_MAX];
1.151 schwarze 2378: static int getcwd_status; /* 1 = ok, 2 = failure */
2379: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2380: char *cp;
1.4 kristaps 2381:
1.59 schwarze 2382: /*
1.151 schwarze 2383: * Remember the original working directory, if possible.
2384: * This will be needed if the second or a later directory
2385: * on the command line is given as a relative path.
2386: * Do not error out if the current directory is not
2387: * searchable: Maybe it won't be needed after all.
2388: */
2389: if (0 == getcwd_status) {
2390: if (NULL == getcwd(startdir, sizeof(startdir))) {
2391: getcwd_status = 2;
2392: (void)strlcpy(startdir, strerror(errno),
2393: sizeof(startdir));
2394: } else
2395: getcwd_status = 1;
2396: }
2397:
2398: /*
2399: * We are leaving the old base directory.
2400: * Do not use it any longer, not even for messages.
2401: */
2402: *basedir = '\0';
2403:
2404: /*
2405: * If and only if the directory was changed earlier and
2406: * the next directory to process is given as a relative path,
2407: * first go back, or bail out if that is impossible.
1.59 schwarze 2408: */
1.151 schwarze 2409: if (chdir_status && '/' != *targetdir) {
2410: if (2 == getcwd_status) {
1.59 schwarze 2411: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2412: say("", "getcwd: %s", startdir);
1.59 schwarze 2413: return(0);
2414: }
1.151 schwarze 2415: if (-1 == chdir(startdir)) {
1.59 schwarze 2416: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2417: say("", "&chdir %s", startdir);
1.59 schwarze 2418: return(0);
2419: }
2420: }
1.151 schwarze 2421:
2422: /*
2423: * Always resolve basedir to the canonicalized absolute
2424: * pathname and append a trailing slash, such that
2425: * we can reliably check whether files are inside.
2426: */
1.59 schwarze 2427: if (NULL == realpath(targetdir, basedir)) {
2428: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 2429: say("", "&%s: realpath", targetdir);
1.50 kristaps 2430: return(0);
1.59 schwarze 2431: } else if (-1 == chdir(basedir)) {
2432: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 2433: say("", "&chdir");
1.50 kristaps 2434: return(0);
1.150 schwarze 2435: }
1.151 schwarze 2436: chdir_status = 1;
1.150 schwarze 2437: cp = strchr(basedir, '\0');
2438: if ('/' != cp[-1]) {
2439: if (cp - basedir >= PATH_MAX - 1) {
2440: exitcode = (int)MANDOCLEVEL_SYSERR;
2441: say("", "Filename too long");
2442: return(0);
2443: }
2444: *cp++ = '/';
2445: *cp = '\0';
1.4 kristaps 2446: }
1.50 kristaps 2447: return(1);
1.56 schwarze 2448: }
2449:
2450: static void
1.59 schwarze 2451: say(const char *file, const char *format, ...)
1.56 schwarze 2452: {
2453: va_list ap;
1.123 schwarze 2454: int use_errno;
1.56 schwarze 2455:
1.59 schwarze 2456: if ('\0' != *basedir)
2457: fprintf(stderr, "%s", basedir);
2458: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2459: fputc('/', stderr);
1.56 schwarze 2460: if ('\0' != *file)
1.59 schwarze 2461: fprintf(stderr, "%s", file);
2462:
1.123 schwarze 2463: use_errno = 1;
2464: if (NULL != format) {
2465: switch (*format) {
1.141 schwarze 2466: case '&':
1.123 schwarze 2467: format++;
2468: break;
1.141 schwarze 2469: case '\0':
1.123 schwarze 2470: format = NULL;
2471: break;
2472: default:
2473: use_errno = 0;
2474: break;
2475: }
2476: }
2477: if (NULL != format) {
2478: if ('\0' != *basedir || '\0' != *file)
2479: fputs(": ", stderr);
2480: va_start(ap, format);
2481: vfprintf(stderr, format, ap);
2482: va_end(ap);
2483: }
2484: if (use_errno) {
2485: if ('\0' != *basedir || '\0' != *file || NULL != format)
2486: fputs(": ", stderr);
1.59 schwarze 2487: perror(NULL);
1.123 schwarze 2488: } else
2489: fputc('\n', stderr);
1.1 kristaps 2490: }
CVSweb