Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.129
1.129 ! schwarze 1: /* $OpenBSD: mandocdb.c,v 1.128 2014/12/04 18:31:04 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.52 schwarze 4: * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
1.117 schwarze 18: #include <sys/types.h>
1.47 schwarze 19: #include <sys/stat.h>
1.72 schwarze 20: #include <sys/wait.h>
1.1 schwarze 21:
22: #include <assert.h>
1.33 schwarze 23: #include <ctype.h>
1.34 schwarze 24: #include <errno.h>
1.1 schwarze 25: #include <fcntl.h>
1.47 schwarze 26: #include <fts.h>
1.1 schwarze 27: #include <getopt.h>
1.44 schwarze 28: #include <limits.h>
1.47 schwarze 29: #include <stddef.h>
1.1 schwarze 30: #include <stdio.h>
31: #include <stdint.h>
32: #include <stdlib.h>
33: #include <string.h>
1.14 schwarze 34: #include <unistd.h>
1.1 schwarze 35:
1.47 schwarze 36: #include <ohash.h>
37: #include <sqlite3.h>
38:
39: #include "mdoc.h"
1.1 schwarze 40: #include "man.h"
41: #include "mandoc.h"
1.76 schwarze 42: #include "mandoc_aux.h"
1.10 schwarze 43: #include "manpath.h"
1.47 schwarze 44: #include "mansearch.h"
1.1 schwarze 45:
1.68 schwarze 46: extern int mansearch_keymax;
47: extern const char *const mansearch_keynames[];
48:
1.47 schwarze 49: #define SQL_EXEC(_v) \
50: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.91 schwarze 51: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.47 schwarze 52: #define SQL_BIND_TEXT(_s, _i, _v) \
53: if (SQLITE_OK != sqlite3_bind_text \
54: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.91 schwarze 55: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 56: #define SQL_BIND_INT(_s, _i, _v) \
57: if (SQLITE_OK != sqlite3_bind_int \
58: ((_s), (_i)++, (_v))) \
1.91 schwarze 59: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 60: #define SQL_BIND_INT64(_s, _i, _v) \
61: if (SQLITE_OK != sqlite3_bind_int64 \
62: ((_s), (_i)++, (_v))) \
1.91 schwarze 63: say(mlink->file, "%s", sqlite3_errmsg(db))
1.47 schwarze 64: #define SQL_STEP(_s) \
65: if (SQLITE_DONE != sqlite3_step((_s))) \
1.91 schwarze 66: say(mlink->file, "%s", sqlite3_errmsg(db))
1.1 schwarze 67:
1.47 schwarze 68: enum op {
69: OP_DEFAULT = 0, /* new dbs from dir list or default config */
70: OP_CONFFILE, /* new databases from custom config file */
71: OP_UPDATE, /* delete/add entries in existing database */
72: OP_DELETE, /* delete entries from existing database */
73: OP_TEST /* change no databases, report potential problems */
74: };
1.11 schwarze 75:
1.47 schwarze 76: struct str {
1.53 schwarze 77: char *rendered; /* key in UTF-8 or ASCII form */
1.47 schwarze 78: const struct mpage *mpage; /* if set, the owning parse */
79: uint64_t mask; /* bitmask in sequence */
1.53 schwarze 80: char key[]; /* may contain escape sequences */
1.28 schwarze 81: };
82:
1.47 schwarze 83: struct inodev {
84: ino_t st_ino;
85: dev_t st_dev;
86: };
1.28 schwarze 87:
1.47 schwarze 88: struct mpage {
89: struct inodev inodev; /* used for hashing routine */
1.94 schwarze 90: int64_t pageid; /* pageid in mpages SQL table */
1.47 schwarze 91: char *sec; /* section from file content */
92: char *arch; /* architecture from file content */
93: char *title; /* title from file content */
94: char *desc; /* description from file content */
95: struct mlink *mlinks; /* singly linked list */
1.116 schwarze 96: int form; /* format from file content */
1.129 ! schwarze 97: int name_head_done;
1.28 schwarze 98: };
99:
1.47 schwarze 100: struct mlink {
101: char file[PATH_MAX]; /* filename rel. to manpath */
102: char *dsec; /* section from directory */
103: char *arch; /* architecture from directory */
104: char *name; /* name from file name (not empty) */
105: char *fsec; /* section from file name suffix */
106: struct mlink *next; /* singly linked list */
1.75 schwarze 107: struct mpage *mpage; /* parent */
1.116 schwarze 108: int dform; /* format from directory */
109: int fform; /* format from file name suffix */
1.81 schwarze 110: int gzip; /* filename has a .gz suffix */
1.2 schwarze 111: };
112:
1.47 schwarze 113: enum stmt {
114: STMT_DELETE_PAGE = 0, /* delete mpage */
115: STMT_INSERT_PAGE, /* insert mpage */
116: STMT_INSERT_LINK, /* insert mlink */
1.90 schwarze 117: STMT_INSERT_NAME, /* insert name */
1.47 schwarze 118: STMT_INSERT_KEY, /* insert parsed key */
119: STMT__MAX
1.1 schwarze 120: };
121:
1.127 schwarze 122: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_meta *,
123: const struct mdoc_node *);
1.1 schwarze 124:
1.19 schwarze 125: struct mdoc_handler {
1.47 schwarze 126: mdoc_fp fp; /* optional handler */
127: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 128: };
129:
1.47 schwarze 130: static void dbclose(int);
1.75 schwarze 131: static void dbadd(struct mpage *, struct mchars *);
132: static void dbadd_mlink(const struct mlink *mlink);
1.124 schwarze 133: static void dbadd_mlink_name(const struct mlink *mlink);
1.47 schwarze 134: static int dbopen(int);
135: static void dbprune(void);
136: static void filescan(const char *);
137: static void *hash_alloc(size_t, void *);
1.106 espie 138: static void hash_free(void *, void *);
139: static void *hash_calloc(size_t, size_t, void *);
1.47 schwarze 140: static void mlink_add(struct mlink *, const struct stat *);
1.88 schwarze 141: static void mlink_check(struct mpage *, struct mlink *);
1.47 schwarze 142: static void mlink_free(struct mlink *);
143: static void mlinks_undupe(struct mpage *);
144: static void mpages_free(void);
1.58 schwarze 145: static void mpages_merge(struct mchars *, struct mparse *);
1.87 schwarze 146: static void names_check(void);
1.81 schwarze 147: static void parse_cat(struct mpage *, int);
1.129 ! schwarze 148: static void parse_man(struct mpage *, const struct man_meta *,
! 149: const struct man_node *);
1.127 schwarze 150: static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
151: const struct mdoc_node *);
152: static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
153: const struct mdoc_node *);
154: static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
155: const struct mdoc_node *);
156: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
157: const struct mdoc_node *);
158: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
159: const struct mdoc_node *);
1.128 schwarze 160: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
161: const struct mdoc_node *);
1.127 schwarze 162: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
163: const struct mdoc_node *);
164: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
165: const struct mdoc_node *);
166: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
167: const struct mdoc_node *);
168: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
169: const struct mdoc_node *);
1.69 schwarze 170: static void putkey(const struct mpage *, char *, uint64_t);
1.47 schwarze 171: static void putkeys(const struct mpage *,
172: const char *, size_t, uint64_t);
173: static void putmdockey(const struct mpage *,
174: const struct mdoc_node *, uint64_t);
1.53 schwarze 175: static void render_key(struct mchars *, struct str *);
1.47 schwarze 176: static void say(const char *, const char *, ...);
1.120 schwarze 177: static int set_basedir(const char *, int);
1.47 schwarze 178: static int treescan(void);
179: static size_t utf8(unsigned int, char [7]);
180:
1.72 schwarze 181: static char tempfilename[32];
1.47 schwarze 182: static char *progname;
1.59 schwarze 183: static int nodb; /* no database changes */
1.73 schwarze 184: static int mparse_options; /* abort the parse early */
1.98 schwarze 185: static int use_all; /* use all found files */
186: static int debug; /* print what we're doing */
187: static int warnings; /* warn about crap */
1.52 schwarze 188: static int write_utf8; /* write UTF-8 output; else ASCII */
1.47 schwarze 189: static int exitcode; /* to be returned by main */
1.98 schwarze 190: static enum op op; /* operational mode */
1.47 schwarze 191: static char basedir[PATH_MAX]; /* current base directory */
192: static struct ohash mpages; /* table of distinct manual pages */
193: static struct ohash mlinks; /* table of directory entries */
1.90 schwarze 194: static struct ohash names; /* table of all names */
1.47 schwarze 195: static struct ohash strings; /* table of all strings */
196: static sqlite3 *db = NULL; /* current database */
197: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.90 schwarze 198: static uint64_t name_mask;
1.47 schwarze 199:
1.19 schwarze 200: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 schwarze 201: { NULL, 0 }, /* Ap */
202: { NULL, 0 }, /* Dd */
203: { NULL, 0 }, /* Dt */
204: { NULL, 0 }, /* Os */
205: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
206: { parse_mdoc_head, TYPE_Ss }, /* Ss */
207: { NULL, 0 }, /* Pp */
208: { NULL, 0 }, /* D1 */
209: { NULL, 0 }, /* Dl */
210: { NULL, 0 }, /* Bd */
211: { NULL, 0 }, /* Ed */
212: { NULL, 0 }, /* Bl */
213: { NULL, 0 }, /* El */
214: { NULL, 0 }, /* It */
215: { NULL, 0 }, /* Ad */
216: { NULL, TYPE_An }, /* An */
217: { NULL, TYPE_Ar }, /* Ar */
218: { NULL, TYPE_Cd }, /* Cd */
219: { NULL, TYPE_Cm }, /* Cm */
220: { NULL, TYPE_Dv }, /* Dv */
221: { NULL, TYPE_Er }, /* Er */
222: { NULL, TYPE_Ev }, /* Ev */
223: { NULL, 0 }, /* Ex */
224: { NULL, TYPE_Fa }, /* Fa */
225: { parse_mdoc_Fd, 0 }, /* Fd */
226: { NULL, TYPE_Fl }, /* Fl */
227: { parse_mdoc_Fn, 0 }, /* Fn */
228: { NULL, TYPE_Ft }, /* Ft */
229: { NULL, TYPE_Ic }, /* Ic */
1.49 schwarze 230: { NULL, TYPE_In }, /* In */
1.47 schwarze 231: { NULL, TYPE_Li }, /* Li */
1.89 schwarze 232: { parse_mdoc_Nd, 0 }, /* Nd */
1.86 schwarze 233: { parse_mdoc_Nm, 0 }, /* Nm */
1.47 schwarze 234: { NULL, 0 }, /* Op */
235: { NULL, 0 }, /* Ot */
236: { NULL, TYPE_Pa }, /* Pa */
237: { NULL, 0 }, /* Rv */
1.49 schwarze 238: { NULL, TYPE_St }, /* St */
1.47 schwarze 239: { NULL, TYPE_Va }, /* Va */
240: { parse_mdoc_body, TYPE_Va }, /* Vt */
241: { parse_mdoc_Xr, 0 }, /* Xr */
242: { NULL, 0 }, /* %A */
243: { NULL, 0 }, /* %B */
244: { NULL, 0 }, /* %D */
245: { NULL, 0 }, /* %I */
246: { NULL, 0 }, /* %J */
247: { NULL, 0 }, /* %N */
248: { NULL, 0 }, /* %O */
249: { NULL, 0 }, /* %P */
250: { NULL, 0 }, /* %R */
251: { NULL, 0 }, /* %T */
252: { NULL, 0 }, /* %V */
253: { NULL, 0 }, /* Ac */
254: { NULL, 0 }, /* Ao */
255: { NULL, 0 }, /* Aq */
256: { NULL, TYPE_At }, /* At */
257: { NULL, 0 }, /* Bc */
258: { NULL, 0 }, /* Bf */
259: { NULL, 0 }, /* Bo */
260: { NULL, 0 }, /* Bq */
261: { NULL, TYPE_Bsx }, /* Bsx */
262: { NULL, TYPE_Bx }, /* Bx */
263: { NULL, 0 }, /* Db */
264: { NULL, 0 }, /* Dc */
265: { NULL, 0 }, /* Do */
266: { NULL, 0 }, /* Dq */
267: { NULL, 0 }, /* Ec */
268: { NULL, 0 }, /* Ef */
269: { NULL, TYPE_Em }, /* Em */
270: { NULL, 0 }, /* Eo */
271: { NULL, TYPE_Fx }, /* Fx */
272: { NULL, TYPE_Ms }, /* Ms */
273: { NULL, 0 }, /* No */
274: { NULL, 0 }, /* Ns */
275: { NULL, TYPE_Nx }, /* Nx */
276: { NULL, TYPE_Ox }, /* Ox */
277: { NULL, 0 }, /* Pc */
278: { NULL, 0 }, /* Pf */
279: { NULL, 0 }, /* Po */
280: { NULL, 0 }, /* Pq */
281: { NULL, 0 }, /* Qc */
282: { NULL, 0 }, /* Ql */
283: { NULL, 0 }, /* Qo */
284: { NULL, 0 }, /* Qq */
285: { NULL, 0 }, /* Re */
286: { NULL, 0 }, /* Rs */
287: { NULL, 0 }, /* Sc */
288: { NULL, 0 }, /* So */
289: { NULL, 0 }, /* Sq */
290: { NULL, 0 }, /* Sm */
291: { NULL, 0 }, /* Sx */
292: { NULL, TYPE_Sy }, /* Sy */
293: { NULL, TYPE_Tn }, /* Tn */
294: { NULL, 0 }, /* Ux */
295: { NULL, 0 }, /* Xc */
296: { NULL, 0 }, /* Xo */
1.128 schwarze 297: { parse_mdoc_Fo, 0 }, /* Fo */
1.47 schwarze 298: { NULL, 0 }, /* Fc */
299: { NULL, 0 }, /* Oo */
300: { NULL, 0 }, /* Oc */
301: { NULL, 0 }, /* Bk */
302: { NULL, 0 }, /* Ek */
303: { NULL, 0 }, /* Bt */
304: { NULL, 0 }, /* Hf */
305: { NULL, 0 }, /* Fr */
306: { NULL, 0 }, /* Ud */
307: { NULL, TYPE_Lb }, /* Lb */
308: { NULL, 0 }, /* Lp */
309: { NULL, TYPE_Lk }, /* Lk */
310: { NULL, TYPE_Mt }, /* Mt */
311: { NULL, 0 }, /* Brq */
312: { NULL, 0 }, /* Bro */
313: { NULL, 0 }, /* Brc */
314: { NULL, 0 }, /* %C */
315: { NULL, 0 }, /* Es */
316: { NULL, 0 }, /* En */
317: { NULL, TYPE_Dx }, /* Dx */
318: { NULL, 0 }, /* %Q */
319: { NULL, 0 }, /* br */
320: { NULL, 0 }, /* sp */
321: { NULL, 0 }, /* %U */
322: { NULL, 0 }, /* Ta */
1.121 schwarze 323: { NULL, 0 }, /* ll */
1.1 schwarze 324: };
325:
1.98 schwarze 326:
1.1 schwarze 327: int
1.3 schwarze 328: mandocdb(int argc, char *argv[])
1.1 schwarze 329: {
1.47 schwarze 330: int ch, i;
331: size_t j, sz;
332: const char *path_arg;
333: struct mchars *mc;
334: struct manpaths dirs;
335: struct mparse *mp;
336: struct ohash_info mpages_info, mlinks_info;
337:
338: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
339: memset(&dirs, 0, sizeof(struct manpaths));
340:
341: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.106 espie 342: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
343: mpages_info.free = mlinks_info.free = hash_free;
1.47 schwarze 344:
345: mpages_info.key_offset = offsetof(struct mpage, inodev);
346: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 347:
348: progname = strrchr(argv[0], '/');
349: if (progname == NULL)
350: progname = argv[0];
351: else
352: ++progname;
353:
1.47 schwarze 354: /*
1.98 schwarze 355: * We accept a few different invocations.
1.47 schwarze 356: * The CHECKOP macro makes sure that invocation styles don't
357: * clobber each other.
358: */
359: #define CHECKOP(_op, _ch) do \
360: if (OP_DEFAULT != (_op)) { \
1.111 schwarze 361: fprintf(stderr, "%s: -%c: Conflicting option\n", \
362: progname, (_ch)); \
1.47 schwarze 363: goto usage; \
364: } while (/*CONSTCOND*/0)
1.10 schwarze 365:
1.47 schwarze 366: path_arg = NULL;
1.28 schwarze 367: op = OP_DEFAULT;
1.1 schwarze 368:
1.83 schwarze 369: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 schwarze 370: switch (ch) {
1.98 schwarze 371: case 'a':
1.6 schwarze 372: use_all = 1;
373: break;
1.98 schwarze 374: case 'C':
1.47 schwarze 375: CHECKOP(op, ch);
376: path_arg = optarg;
1.28 schwarze 377: op = OP_CONFFILE;
1.25 schwarze 378: break;
1.98 schwarze 379: case 'D':
1.82 schwarze 380: debug++;
381: break;
1.98 schwarze 382: case 'd':
1.47 schwarze 383: CHECKOP(op, ch);
384: path_arg = optarg;
1.2 schwarze 385: op = OP_UPDATE;
1.1 schwarze 386: break;
1.98 schwarze 387: case 'n':
1.47 schwarze 388: nodb = 1;
389: break;
1.98 schwarze 390: case 'p':
1.83 schwarze 391: warnings = 1;
392: break;
1.98 schwarze 393: case 'Q':
1.73 schwarze 394: mparse_options |= MPARSE_QUICK;
1.59 schwarze 395: break;
1.98 schwarze 396: case 'T':
1.52 schwarze 397: if (strcmp(optarg, "utf8")) {
1.111 schwarze 398: fprintf(stderr, "%s: -T%s: "
399: "Unsupported output format\n",
400: progname, optarg);
1.52 schwarze 401: goto usage;
402: }
403: write_utf8 = 1;
404: break;
1.98 schwarze 405: case 't':
1.47 schwarze 406: CHECKOP(op, ch);
1.28 schwarze 407: dup2(STDOUT_FILENO, STDERR_FILENO);
408: op = OP_TEST;
1.47 schwarze 409: nodb = warnings = 1;
1.28 schwarze 410: break;
1.98 schwarze 411: case 'u':
1.47 schwarze 412: CHECKOP(op, ch);
413: path_arg = optarg;
1.1 schwarze 414: op = OP_DELETE;
415: break;
1.98 schwarze 416: case 'v':
1.83 schwarze 417: /* Compatibility with espie@'s makewhatis. */
1.28 schwarze 418: break;
1.1 schwarze 419: default:
1.28 schwarze 420: goto usage;
1.1 schwarze 421: }
422:
423: argc -= optind;
424: argv += optind;
425:
1.28 schwarze 426: if (OP_CONFFILE == op && argc > 0) {
1.111 schwarze 427: fprintf(stderr, "%s: -C: Too many arguments\n",
428: progname);
1.28 schwarze 429: goto usage;
430: }
431:
1.47 schwarze 432: exitcode = (int)MANDOCLEVEL_OK;
433: mc = mchars_alloc();
1.123 schwarze 434: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
435: mc, NULL);
1.47 schwarze 436: ohash_init(&mpages, 6, &mpages_info);
437: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 438:
1.47 schwarze 439: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.2 schwarze 440:
1.47 schwarze 441: /*
1.109 schwarze 442: * Most of these deal with a specific directory.
1.97 schwarze 443: * Jump into that directory first.
1.47 schwarze 444: */
1.120 schwarze 445: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.47 schwarze 446: goto out;
1.97 schwarze 447:
1.95 schwarze 448: if (dbopen(1)) {
1.97 schwarze 449: /*
450: * The existing database is usable. Process
451: * all files specified on the command-line.
452: */
453: use_all = 1;
1.95 schwarze 454: for (i = 0; i < argc; i++)
455: filescan(argv[i]);
456: if (OP_TEST != op)
457: dbprune();
458: } else {
459: /*
460: * Database missing or corrupt.
461: * Recreate from scratch.
462: */
1.97 schwarze 463: exitcode = (int)MANDOCLEVEL_OK;
1.95 schwarze 464: op = OP_DEFAULT;
465: if (0 == treescan())
466: goto out;
467: if (0 == dbopen(0))
468: goto out;
469: }
1.47 schwarze 470: if (OP_DELETE != op)
1.58 schwarze 471: mpages_merge(mc, mp);
1.95 schwarze 472: dbclose(OP_DEFAULT == op ? 0 : 1);
1.47 schwarze 473: } else {
474: /*
475: * If we have arguments, use them as our manpaths.
476: * If we don't, grok from manpath(1) or however else
477: * manpath_parse() wants to do it.
478: */
479: if (argc > 0) {
1.101 schwarze 480: dirs.paths = mandoc_reallocarray(NULL,
481: argc, sizeof(char *));
1.47 schwarze 482: dirs.sz = (size_t)argc;
483: for (i = 0; i < argc; i++)
484: dirs.paths[i] = mandoc_strdup(argv[i]);
485: } else
486: manpath_parse(&dirs, path_arg, NULL, NULL);
1.84 schwarze 487:
488: if (0 == dirs.sz) {
489: exitcode = (int)MANDOCLEVEL_BADARG;
490: say("", "Empty manpath");
491: }
1.2 schwarze 492:
1.47 schwarze 493: /*
494: * First scan the tree rooted at a base directory, then
495: * build a new database and finally move it into place.
496: * Ignore zero-length directories and strip trailing
497: * slashes.
498: */
499: for (j = 0; j < dirs.sz; j++) {
500: sz = strlen(dirs.paths[j]);
501: if (sz && '/' == dirs.paths[j][sz - 1])
502: dirs.paths[j][--sz] = '\0';
503: if (0 == sz)
504: continue;
1.2 schwarze 505:
1.47 schwarze 506: if (j) {
507: ohash_init(&mpages, 6, &mpages_info);
508: ohash_init(&mlinks, 6, &mlinks_info);
509: }
1.1 schwarze 510:
1.120 schwarze 511: if (0 == set_basedir(dirs.paths[j], argc > 0))
512: continue;
1.47 schwarze 513: if (0 == treescan())
1.120 schwarze 514: continue;
1.47 schwarze 515: if (0 == dbopen(0))
1.120 schwarze 516: continue;
1.41 deraadt 517:
1.58 schwarze 518: mpages_merge(mc, mp);
1.104 schwarze 519: if (warnings && !nodb &&
1.87 schwarze 520: ! (MPARSE_QUICK & mparse_options))
521: names_check();
1.47 schwarze 522: dbclose(0);
1.28 schwarze 523:
1.47 schwarze 524: if (j + 1 < dirs.sz) {
525: mpages_free();
526: ohash_delete(&mpages);
527: ohash_delete(&mlinks);
528: }
1.2 schwarze 529: }
1.47 schwarze 530: }
531: out:
532: manpath_free(&dirs);
1.123 schwarze 533: mparse_free(mp);
1.47 schwarze 534: mchars_free(mc);
535: mpages_free();
536: ohash_delete(&mpages);
537: ohash_delete(&mlinks);
538: return(exitcode);
539: usage:
1.83 schwarze 540: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
541: " %s [-aDnpQ] [-Tutf8] dir ...\n"
542: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
543: " %s [-Dnp] -u dir [file ...]\n"
1.59 schwarze 544: " %s [-Q] -t file ...\n",
1.98 schwarze 545: progname, progname, progname,
1.47 schwarze 546: progname, progname);
1.1 schwarze 547:
1.47 schwarze 548: return((int)MANDOCLEVEL_BADARG);
549: }
1.1 schwarze 550:
1.47 schwarze 551: /*
552: * Scan a directory tree rooted at "basedir" for manpages.
553: * We use fts(), scanning directory parts along the way for clues to our
554: * section and architecture.
555: *
556: * If use_all has been specified, grok all files.
557: * If not, sanitise paths to the following:
558: *
1.98 schwarze 559: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 560: * or
561: * [./]cat<section>[/<arch>]/<name>.0
562: *
563: * TODO: accomodate for multi-language directories.
564: */
565: static int
566: treescan(void)
567: {
1.96 schwarze 568: char buf[PATH_MAX];
1.47 schwarze 569: FTS *f;
570: FTSENT *ff;
571: struct mlink *mlink;
1.81 schwarze 572: int dform, gzip;
1.51 schwarze 573: char *dsec, *arch, *fsec, *cp;
574: const char *path;
1.47 schwarze 575: const char *argv[2];
1.1 schwarze 576:
1.47 schwarze 577: argv[0] = ".";
578: argv[1] = (char *)NULL;
1.1 schwarze 579:
1.96 schwarze 580: f = fts_open((char * const *)argv,
581: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.47 schwarze 582: if (NULL == f) {
583: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 584: say("", "&fts_open");
1.47 schwarze 585: return(0);
586: }
1.2 schwarze 587:
1.47 schwarze 588: dsec = arch = NULL;
589: dform = FORM_NONE;
1.2 schwarze 590:
1.47 schwarze 591: while (NULL != (ff = fts_read(f))) {
592: path = ff->fts_path + 2;
1.96 schwarze 593: switch (ff->fts_info) {
594:
595: /*
596: * Symbolic links require various sanity checks,
597: * then get handled just like regular files.
598: */
1.98 schwarze 599: case FTS_SL:
1.96 schwarze 600: if (NULL == realpath(path, buf)) {
601: if (warnings)
602: say(path, "&realpath");
603: continue;
604: }
605: if (strstr(buf, basedir) != buf) {
606: if (warnings) say("",
607: "%s: outside base directory", buf);
608: continue;
609: }
610: /* Use logical inode to avoid mpages dupe. */
611: if (-1 == stat(path, ff->fts_statp)) {
612: if (warnings)
613: say(path, "&stat");
614: continue;
615: }
616: /* FALLTHROUGH */
617:
1.15 schwarze 618: /*
1.47 schwarze 619: * If we're a regular file, add an mlink by using the
620: * stored directory data and handling the filename.
1.15 schwarze 621: */
1.98 schwarze 622: case FTS_F:
1.47 schwarze 623: if (0 == strcmp(path, MANDOC_DB))
624: continue;
625: if ( ! use_all && ff->fts_level < 2) {
626: if (warnings)
627: say(path, "Extraneous file");
628: continue;
1.81 schwarze 629: }
630: gzip = 0;
631: fsec = NULL;
632: while (NULL == fsec) {
633: fsec = strrchr(ff->fts_name, '.');
634: if (NULL == fsec || strcmp(fsec+1, "gz"))
635: break;
636: gzip = 1;
637: *fsec = '\0';
638: fsec = NULL;
639: }
640: if (NULL == fsec) {
1.47 schwarze 641: if ( ! use_all) {
642: if (warnings)
643: say(path,
644: "No filename suffix");
645: continue;
646: }
647: } else if (0 == strcmp(++fsec, "html")) {
648: if (warnings)
649: say(path, "Skip html");
650: continue;
651: } else if (0 == strcmp(fsec, "ps")) {
652: if (warnings)
653: say(path, "Skip ps");
654: continue;
655: } else if (0 == strcmp(fsec, "pdf")) {
656: if (warnings)
657: say(path, "Skip pdf");
658: continue;
659: } else if ( ! use_all &&
660: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
661: (FORM_CAT == dform && strcmp(fsec, "0")))) {
662: if (warnings)
663: say(path, "Wrong filename suffix");
664: continue;
665: } else
666: fsec[-1] = '\0';
1.51 schwarze 667:
1.47 schwarze 668: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.100 schwarze 669: if (strlcpy(mlink->file, path,
670: sizeof(mlink->file)) >=
671: sizeof(mlink->file)) {
672: say(path, "Filename too long");
673: free(mlink);
674: continue;
675: }
1.47 schwarze 676: mlink->dform = dform;
1.51 schwarze 677: mlink->dsec = dsec;
678: mlink->arch = arch;
679: mlink->name = ff->fts_name;
680: mlink->fsec = fsec;
1.81 schwarze 681: mlink->gzip = gzip;
1.47 schwarze 682: mlink_add(mlink, ff->fts_statp);
683: continue;
1.96 schwarze 684:
1.98 schwarze 685: case FTS_D:
1.96 schwarze 686: /* FALLTHROUGH */
1.98 schwarze 687: case FTS_DP:
1.96 schwarze 688: break;
689:
690: default:
1.47 schwarze 691: if (warnings)
692: say(path, "Not a regular file");
693: continue;
694: }
695:
696: switch (ff->fts_level) {
1.98 schwarze 697: case 0:
1.47 schwarze 698: /* Ignore the root directory. */
699: break;
1.98 schwarze 700: case 1:
1.47 schwarze 701: /*
702: * This might contain manX/ or catX/.
703: * Try to infer this from the name.
704: * If we're not in use_all, enforce it.
705: */
706: cp = ff->fts_name;
707: if (FTS_DP == ff->fts_info)
708: break;
1.15 schwarze 709:
1.47 schwarze 710: if (0 == strncmp(cp, "man", 3)) {
711: dform = FORM_SRC;
712: dsec = cp + 3;
713: } else if (0 == strncmp(cp, "cat", 3)) {
714: dform = FORM_CAT;
715: dsec = cp + 3;
1.51 schwarze 716: } else {
717: dform = FORM_NONE;
718: dsec = NULL;
1.26 schwarze 719: }
1.47 schwarze 720:
1.98 schwarze 721: if (NULL != dsec || use_all)
1.47 schwarze 722: break;
723:
724: if (warnings)
725: say(path, "Unknown directory part");
726: fts_set(f, ff, FTS_SKIP);
727: break;
1.98 schwarze 728: case 2:
1.47 schwarze 729: /*
730: * Possibly our architecture.
731: * If we're descending, keep tabs on it.
732: */
733: if (FTS_DP != ff->fts_info && NULL != dsec)
734: arch = ff->fts_name;
1.51 schwarze 735: else
736: arch = NULL;
1.47 schwarze 737: break;
738: default:
739: if (FTS_DP == ff->fts_info || use_all)
740: break;
741: if (warnings)
742: say(path, "Extraneous directory part");
743: fts_set(f, ff, FTS_SKIP);
744: break;
1.14 schwarze 745: }
1.47 schwarze 746: }
747:
748: fts_close(f);
749: return(1);
750: }
1.1 schwarze 751:
1.47 schwarze 752: /*
753: * Add a file to the mlinks table.
754: * Do not verify that it's a "valid" looking manpage (we'll do that
755: * later).
756: *
757: * Try to infer the manual section, architecture, and page name from the
758: * path, assuming it looks like
759: *
1.98 schwarze 760: * [./]man*[/<arch>]/<name>.<section>
1.47 schwarze 761: * or
762: * [./]cat<section>[/<arch>]/<name>.0
763: *
764: * See treescan() for the fts(3) version of this.
765: */
766: static void
767: filescan(const char *file)
768: {
769: char buf[PATH_MAX];
770: struct stat st;
771: struct mlink *mlink;
772: char *p, *start;
773:
774: assert(use_all);
775:
776: if (0 == strncmp(file, "./", 2))
777: file += 2;
778:
1.96 schwarze 779: /*
780: * We have to do lstat(2) before realpath(3) loses
781: * the information whether this is a symbolic link.
782: * We need to know that because for symbolic links,
783: * we want to use the orginal file name, while for
784: * regular files, we want to use the real path.
785: */
786: if (-1 == lstat(file, &st)) {
787: exitcode = (int)MANDOCLEVEL_BADARG;
788: say(file, "&lstat");
789: return;
790: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
791: exitcode = (int)MANDOCLEVEL_BADARG;
792: say(file, "Not a regular file");
793: return;
794: }
795:
796: /*
797: * We have to resolve the file name to the real path
798: * in any case for the base directory check.
799: */
1.47 schwarze 800: if (NULL == realpath(file, buf)) {
801: exitcode = (int)MANDOCLEVEL_BADARG;
1.80 schwarze 802: say(file, "&realpath");
1.47 schwarze 803: return;
1.63 schwarze 804: }
805:
1.109 schwarze 806: if (OP_TEST == op)
1.63 schwarze 807: start = buf;
1.109 schwarze 808: else if (strstr(buf, basedir) == buf)
809: start = buf + strlen(basedir);
1.63 schwarze 810: else {
1.47 schwarze 811: exitcode = (int)MANDOCLEVEL_BADARG;
812: say("", "%s: outside base directory", buf);
813: return;
1.63 schwarze 814: }
815:
1.96 schwarze 816: /*
817: * Now we are sure the file is inside our tree.
818: * If it is a symbolic link, ignore the real path
819: * and use the original name.
820: * This implies passing stuff like "cat1/../man1/foo.1"
821: * on the command line won't work. So don't do that.
822: * Note the stat(2) can still fail if the link target
823: * doesn't exist.
824: */
825: if (S_IFLNK & st.st_mode) {
826: if (-1 == stat(buf, &st)) {
827: exitcode = (int)MANDOCLEVEL_BADARG;
828: say(file, "&stat");
829: return;
830: }
1.100 schwarze 831: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
832: say(file, "Filename too long");
833: return;
834: }
1.109 schwarze 835: start = buf;
836: if (OP_TEST != op && strstr(buf, basedir) == buf)
837: start += strlen(basedir);
1.1 schwarze 838: }
1.63 schwarze 839:
1.47 schwarze 840: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.116 schwarze 841: mlink->dform = FORM_NONE;
1.100 schwarze 842: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
843: sizeof(mlink->file)) {
844: say(start, "Filename too long");
845: return;
846: }
1.1 schwarze 847:
1.10 schwarze 848: /*
1.47 schwarze 849: * First try to guess our directory structure.
850: * If we find a separator, try to look for man* or cat*.
851: * If we find one of these and what's underneath is a directory,
852: * assume it's an architecture.
1.10 schwarze 853: */
1.47 schwarze 854: if (NULL != (p = strchr(start, '/'))) {
855: *p++ = '\0';
856: if (0 == strncmp(start, "man", 3)) {
857: mlink->dform = FORM_SRC;
1.51 schwarze 858: mlink->dsec = start + 3;
1.47 schwarze 859: } else if (0 == strncmp(start, "cat", 3)) {
860: mlink->dform = FORM_CAT;
1.51 schwarze 861: mlink->dsec = start + 3;
1.47 schwarze 862: }
1.10 schwarze 863:
1.47 schwarze 864: start = p;
865: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
866: *p++ = '\0';
1.51 schwarze 867: mlink->arch = start;
1.47 schwarze 868: start = p;
1.41 deraadt 869: }
1.47 schwarze 870: }
1.7 schwarze 871:
1.47 schwarze 872: /*
873: * Now check the file suffix.
874: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
875: */
876: p = strrchr(start, '\0');
877: while (p-- > start && '/' != *p && '.' != *p)
878: /* Loop. */ ;
879:
880: if ('.' == *p) {
881: *p++ = '\0';
1.51 schwarze 882: mlink->fsec = p;
1.47 schwarze 883: }
1.41 deraadt 884:
1.47 schwarze 885: /*
886: * Now try to parse the name.
887: * Use the filename portion of the path.
888: */
889: mlink->name = start;
890: if (NULL != (p = strrchr(start, '/'))) {
891: mlink->name = p + 1;
892: *p = '\0';
893: }
894: mlink_add(mlink, &st);
895: }
1.2 schwarze 896:
1.47 schwarze 897: static void
898: mlink_add(struct mlink *mlink, const struct stat *st)
899: {
900: struct inodev inodev;
901: struct mpage *mpage;
902: unsigned int slot;
903:
904: assert(NULL != mlink->file);
905:
1.51 schwarze 906: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
907: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
908: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
909: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.47 schwarze 910:
911: if ('0' == *mlink->fsec) {
912: free(mlink->fsec);
913: mlink->fsec = mandoc_strdup(mlink->dsec);
914: mlink->fform = FORM_CAT;
915: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
916: mlink->fform = FORM_SRC;
917: else
918: mlink->fform = FORM_NONE;
1.1 schwarze 919:
1.47 schwarze 920: slot = ohash_qlookup(&mlinks, mlink->file);
921: assert(NULL == ohash_find(&mlinks, slot));
922: ohash_insert(&mlinks, slot, mlink);
923:
924: inodev.st_ino = st->st_ino;
925: inodev.st_dev = st->st_dev;
926: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
927: sizeof(struct inodev), inodev.st_ino);
928: mpage = ohash_find(&mpages, slot);
929: if (NULL == mpage) {
930: mpage = mandoc_calloc(1, sizeof(struct mpage));
931: mpage->inodev.st_ino = inodev.st_ino;
932: mpage->inodev.st_dev = inodev.st_dev;
933: ohash_insert(&mpages, slot, mpage);
934: } else
935: mlink->next = mpage->mlinks;
936: mpage->mlinks = mlink;
1.75 schwarze 937: mlink->mpage = mpage;
1.47 schwarze 938: }
1.1 schwarze 939:
1.47 schwarze 940: static void
941: mlink_free(struct mlink *mlink)
942: {
1.1 schwarze 943:
1.47 schwarze 944: free(mlink->dsec);
945: free(mlink->arch);
946: free(mlink->name);
947: free(mlink->fsec);
948: free(mlink);
949: }
1.1 schwarze 950:
1.47 schwarze 951: static void
952: mpages_free(void)
953: {
954: struct mpage *mpage;
955: struct mlink *mlink;
956: unsigned int slot;
957:
958: mpage = ohash_first(&mpages, &slot);
959: while (NULL != mpage) {
960: while (NULL != (mlink = mpage->mlinks)) {
961: mpage->mlinks = mlink->next;
962: mlink_free(mlink);
963: }
964: free(mpage->sec);
965: free(mpage->arch);
966: free(mpage->title);
967: free(mpage->desc);
968: free(mpage);
969: mpage = ohash_next(&mpages, &slot);
970: }
971: }
1.1 schwarze 972:
1.47 schwarze 973: /*
974: * For each mlink to the mpage, check whether the path looks like
975: * it is formatted, and if it does, check whether a source manual
976: * exists by the same name, ignoring the suffix.
977: * If both conditions hold, drop the mlink.
978: */
979: static void
980: mlinks_undupe(struct mpage *mpage)
981: {
982: char buf[PATH_MAX];
983: struct mlink **prev;
984: struct mlink *mlink;
985: char *bufp;
986:
987: mpage->form = FORM_CAT;
988: prev = &mpage->mlinks;
989: while (NULL != (mlink = *prev)) {
990: if (FORM_CAT != mlink->dform) {
991: mpage->form = FORM_NONE;
992: goto nextlink;
993: }
1.100 schwarze 994: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.47 schwarze 995: bufp = strstr(buf, "cat");
996: assert(NULL != bufp);
997: memcpy(bufp, "man", 3);
998: if (NULL != (bufp = strrchr(buf, '.')))
999: *++bufp = '\0';
1.100 schwarze 1000: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.47 schwarze 1001: if (NULL == ohash_find(&mlinks,
1.98 schwarze 1002: ohash_qlookup(&mlinks, buf)))
1.47 schwarze 1003: goto nextlink;
1004: if (warnings)
1005: say(mlink->file, "Man source exists: %s", buf);
1006: if (use_all)
1007: goto nextlink;
1008: *prev = mlink->next;
1009: mlink_free(mlink);
1010: continue;
1011: nextlink:
1012: prev = &(*prev)->next;
1.1 schwarze 1013: }
1.47 schwarze 1014: }
1.1 schwarze 1015:
1.88 schwarze 1016: static void
1.50 schwarze 1017: mlink_check(struct mpage *mpage, struct mlink *mlink)
1018: {
1.88 schwarze 1019: struct str *str;
1020: unsigned int slot;
1.50 schwarze 1021:
1022: /*
1023: * Check whether the manual section given in a file
1024: * agrees with the directory where the file is located.
1025: * Some manuals have suffixes like (3p) on their
1026: * section number either inside the file or in the
1027: * directory name, some are linked into more than one
1028: * section, like encrypt(1) = makekey(8).
1029: */
1030:
1031: if (FORM_SRC == mpage->form &&
1.88 schwarze 1032: strcasecmp(mpage->sec, mlink->dsec))
1.50 schwarze 1033: say(mlink->file, "Section \"%s\" manual in %s directory",
1034: mpage->sec, mlink->dsec);
1035:
1036: /*
1037: * Manual page directories exist for each kernel
1038: * architecture as returned by machine(1).
1039: * However, many manuals only depend on the
1040: * application architecture as returned by arch(1).
1041: * For example, some (2/ARM) manuals are shared
1042: * across the "armish" and "zaurus" kernel
1043: * architectures.
1044: * A few manuals are even shared across completely
1045: * different architectures, for example fdformat(1)
1046: * on amd64, i386, sparc, and sparc64.
1047: */
1048:
1.88 schwarze 1049: if (strcasecmp(mpage->arch, mlink->arch))
1.50 schwarze 1050: say(mlink->file, "Architecture \"%s\" manual in "
1051: "\"%s\" directory", mpage->arch, mlink->arch);
1052:
1.88 schwarze 1053: /*
1054: * XXX
1.90 schwarze 1055: * parse_cat() doesn't set NAME_TITLE yet.
1.88 schwarze 1056: */
1057:
1058: if (FORM_CAT == mpage->form)
1059: return;
1060:
1061: /*
1062: * Check whether this mlink
1063: * appears as a name in the NAME section.
1064: */
1.50 schwarze 1065:
1.90 schwarze 1066: slot = ohash_qlookup(&names, mlink->name);
1067: str = ohash_find(&names, slot);
1.88 schwarze 1068: assert(NULL != str);
1.90 schwarze 1069: if ( ! (NAME_TITLE & str->mask))
1.88 schwarze 1070: say(mlink->file, "Name missing in NAME section");
1.50 schwarze 1071: }
1072:
1.47 schwarze 1073: /*
1074: * Run through the files in the global vector "mpages"
1075: * and add them to the database specified in "basedir".
1076: *
1077: * This handles the parsing scheme itself, using the cues of directory
1078: * and filename to determine whether the file is parsable or not.
1079: */
1080: static void
1.58 schwarze 1081: mpages_merge(struct mchars *mc, struct mparse *mp)
1.47 schwarze 1082: {
1.70 schwarze 1083: char any[] = "any";
1.58 schwarze 1084: struct ohash_info str_info;
1.75 schwarze 1085: struct mpage *mpage, *mpage_dest;
1086: struct mlink *mlink, *mlink_dest;
1.47 schwarze 1087: struct mdoc *mdoc;
1088: struct man *man;
1.75 schwarze 1089: char *sodest;
1.69 schwarze 1090: char *cp;
1.117 schwarze 1091: int fd;
1.58 schwarze 1092: unsigned int pslot;
1.47 schwarze 1093: enum mandoclevel lvl;
1094:
1095: str_info.alloc = hash_alloc;
1.106 espie 1096: str_info.calloc = hash_calloc;
1097: str_info.free = hash_free;
1.47 schwarze 1098: str_info.key_offset = offsetof(struct str, key);
1099:
1.126 schwarze 1100: if ( ! nodb)
1.64 schwarze 1101: SQL_EXEC("BEGIN TRANSACTION");
1102:
1.47 schwarze 1103: mpage = ohash_first(&mpages, &pslot);
1.126 schwarze 1104: while (mpage != NULL) {
1.47 schwarze 1105: mlinks_undupe(mpage);
1.126 schwarze 1106: if (mpage->mlinks == NULL) {
1.47 schwarze 1107: mpage = ohash_next(&mpages, &pslot);
1108: continue;
1109: }
1.1 schwarze 1110:
1.90 schwarze 1111: name_mask = NAME_MASK;
1112: ohash_init(&names, 4, &str_info);
1.47 schwarze 1113: ohash_init(&strings, 6, &str_info);
1114: mparse_reset(mp);
1115: mdoc = NULL;
1116: man = NULL;
1.81 schwarze 1117: sodest = NULL;
1118:
1.125 schwarze 1119: mparse_open(mp, &fd, mpage->mlinks->file);
1.117 schwarze 1120: if (fd == -1) {
1121: say(mpage->mlinks->file, "&open");
1122: goto nextpage;
1.81 schwarze 1123: }
1.11 schwarze 1124:
1125: /*
1.24 schwarze 1126: * Try interpreting the file as mdoc(7) or man(7)
1127: * source code, unless it is already known to be
1128: * formatted. Fall back to formatted mode.
1.11 schwarze 1129: */
1.126 schwarze 1130: if (mpage->mlinks->dform != FORM_CAT ||
1131: mpage->mlinks->fform != FORM_CAT) {
1.117 schwarze 1132: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.47 schwarze 1133: if (lvl < MANDOCLEVEL_FATAL)
1.75 schwarze 1134: mparse_result(mp, &mdoc, &man, &sodest);
1.47 schwarze 1135: }
1.11 schwarze 1136:
1.126 schwarze 1137: if (sodest != NULL) {
1.75 schwarze 1138: mlink_dest = ohash_find(&mlinks,
1139: ohash_qlookup(&mlinks, sodest));
1.126 schwarze 1140: if (mlink_dest == NULL) {
1141: mandoc_asprintf(&cp, "%s.gz", sodest);
1142: mlink_dest = ohash_find(&mlinks,
1143: ohash_qlookup(&mlinks, cp));
1144: free(cp);
1145: }
1146: if (mlink_dest != NULL) {
1.75 schwarze 1147:
1148: /* The .so target exists. */
1149:
1150: mpage_dest = mlink_dest->mpage;
1151: mlink = mpage->mlinks;
1152: while (1) {
1153: mlink->mpage = mpage_dest;
1154:
1155: /*
1156: * If the target was already
1157: * processed, add the links
1158: * to the database now.
1159: * Otherwise, this will
1160: * happen when we come
1161: * to the target.
1162: */
1163:
1.94 schwarze 1164: if (mpage_dest->pageid)
1.124 schwarze 1165: dbadd_mlink_name(mlink);
1.75 schwarze 1166:
1.126 schwarze 1167: if (mlink->next == NULL)
1.75 schwarze 1168: break;
1169: mlink = mlink->next;
1170: }
1171:
1172: /* Move all links to the target. */
1173:
1174: mlink->next = mlink_dest->next;
1175: mlink_dest->next = mpage->mlinks;
1176: mpage->mlinks = NULL;
1177: }
1.81 schwarze 1178: goto nextpage;
1.126 schwarze 1179: } else if (mdoc != NULL) {
1.47 schwarze 1180: mpage->form = FORM_SRC;
1.112 schwarze 1181: mpage->sec = mdoc_meta(mdoc)->msec;
1182: mpage->sec = mandoc_strdup(
1.126 schwarze 1183: mpage->sec == NULL ? "" : mpage->sec);
1.47 schwarze 1184: mpage->arch = mdoc_meta(mdoc)->arch;
1185: mpage->arch = mandoc_strdup(
1.126 schwarze 1186: mpage->arch == NULL ? "" : mpage->arch);
1.47 schwarze 1187: mpage->title =
1188: mandoc_strdup(mdoc_meta(mdoc)->title);
1.126 schwarze 1189: } else if (man != NULL) {
1.47 schwarze 1190: mpage->form = FORM_SRC;
1191: mpage->sec =
1192: mandoc_strdup(man_meta(man)->msec);
1193: mpage->arch =
1194: mandoc_strdup(mpage->mlinks->arch);
1195: mpage->title =
1196: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 1197: } else {
1.47 schwarze 1198: mpage->form = FORM_CAT;
1199: mpage->sec =
1200: mandoc_strdup(mpage->mlinks->dsec);
1201: mpage->arch =
1202: mandoc_strdup(mpage->mlinks->arch);
1203: mpage->title =
1204: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 1205: }
1.54 schwarze 1206: putkey(mpage, mpage->sec, TYPE_sec);
1.119 schwarze 1207: if (*mpage->arch != '\0')
1208: putkey(mpage, mpage->arch, TYPE_arch);
1.1 schwarze 1209:
1.54 schwarze 1210: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1211: if ('\0' != *mlink->dsec)
1212: putkey(mpage, mlink->dsec, TYPE_sec);
1213: if ('\0' != *mlink->fsec)
1214: putkey(mpage, mlink->fsec, TYPE_sec);
1.55 schwarze 1215: putkey(mpage, '\0' == *mlink->arch ?
1.70 schwarze 1216: any : mlink->arch, TYPE_arch);
1.90 schwarze 1217: putkey(mpage, mlink->name, NAME_FILE);
1.54 schwarze 1218: }
1.41 deraadt 1219:
1.129 ! schwarze 1220: assert(mpage->desc == NULL);
! 1221: if (mdoc != NULL)
1.127 schwarze 1222: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.129 ! schwarze 1223: else if (man != NULL)
! 1224: parse_man(mpage, man_meta(man), man_node(man));
1.47 schwarze 1225: else
1.117 schwarze 1226: parse_cat(mpage, fd);
1.129 ! schwarze 1227: if (mpage->desc == NULL)
1.92 schwarze 1228: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.88 schwarze 1229:
1230: if (warnings && !use_all)
1231: for (mlink = mpage->mlinks; mlink;
1232: mlink = mlink->next)
1233: mlink_check(mpage, mlink);
1.6 schwarze 1234:
1.62 schwarze 1235: dbadd(mpage, mc);
1.81 schwarze 1236:
1237: nextpage:
1.125 schwarze 1238: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.117 schwarze 1239: exitcode = (int)MANDOCLEVEL_SYSERR;
1240: say(mpage->mlinks->file, "&wait gunzip");
1.81 schwarze 1241: }
1.47 schwarze 1242: ohash_delete(&strings);
1.90 schwarze 1243: ohash_delete(&names);
1.47 schwarze 1244: mpage = ohash_next(&mpages, &pslot);
1245: }
1.64 schwarze 1246:
1247: if (0 == nodb)
1248: SQL_EXEC("END TRANSACTION");
1.87 schwarze 1249: }
1250:
1251: static void
1252: names_check(void)
1253: {
1254: sqlite3_stmt *stmt;
1255: const char *name, *sec, *arch, *key;
1256: int irc;
1257:
1258: sqlite3_prepare_v2(db,
1259: "SELECT name, sec, arch, key FROM ("
1.90 schwarze 1260: "SELECT name AS key, pageid FROM names "
1.87 schwarze 1261: "WHERE bits & ? AND NOT EXISTS ("
1262: "SELECT pageid FROM mlinks "
1.90 schwarze 1263: "WHERE mlinks.pageid == names.pageid "
1264: "AND mlinks.name == names.name"
1.87 schwarze 1265: ")"
1266: ") JOIN ("
1.99 schwarze 1267: "SELECT sec, arch, name, pageid FROM mlinks "
1268: "GROUP BY pageid"
1.87 schwarze 1269: ") USING (pageid);",
1270: -1, &stmt, NULL);
1271:
1.91 schwarze 1272: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1273: say("", "%s", sqlite3_errmsg(db));
1.87 schwarze 1274:
1275: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.113 schwarze 1276: name = (const char *)sqlite3_column_text(stmt, 0);
1277: sec = (const char *)sqlite3_column_text(stmt, 1);
1278: arch = (const char *)sqlite3_column_text(stmt, 2);
1279: key = (const char *)sqlite3_column_text(stmt, 3);
1.87 schwarze 1280: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1281: '\0' == *arch ? "" : "/",
1282: '\0' == *arch ? "" : arch, key);
1283: }
1284: sqlite3_finalize(stmt);
1.47 schwarze 1285: }
1.6 schwarze 1286:
1.47 schwarze 1287: static void
1.81 schwarze 1288: parse_cat(struct mpage *mpage, int fd)
1.47 schwarze 1289: {
1290: FILE *stream;
1291: char *line, *p, *title;
1292: size_t len, plen, titlesz;
1.1 schwarze 1293:
1.81 schwarze 1294: stream = (-1 == fd) ?
1295: fopen(mpage->mlinks->file, "r") :
1296: fdopen(fd, "r");
1297: if (NULL == stream) {
1.114 doug 1298: if (-1 != fd)
1299: close(fd);
1.47 schwarze 1300: if (warnings)
1.80 schwarze 1301: say(mpage->mlinks->file, "&fopen");
1.47 schwarze 1302: return;
1303: }
1.1 schwarze 1304:
1.47 schwarze 1305: /* Skip to first blank line. */
1.1 schwarze 1306:
1.47 schwarze 1307: while (NULL != (line = fgetln(stream, &len)))
1308: if ('\n' == *line)
1309: break;
1.1 schwarze 1310:
1.47 schwarze 1311: /*
1312: * Assume the first line that is not indented
1313: * is the first section header. Skip to it.
1314: */
1.1 schwarze 1315:
1.47 schwarze 1316: while (NULL != (line = fgetln(stream, &len)))
1317: if ('\n' != *line && ' ' != *line)
1318: break;
1.98 schwarze 1319:
1.47 schwarze 1320: /*
1321: * Read up until the next section into a buffer.
1322: * Strip the leading and trailing newline from each read line,
1323: * appending a trailing space.
1324: * Ignore empty (whitespace-only) lines.
1325: */
1.28 schwarze 1326:
1.47 schwarze 1327: titlesz = 0;
1328: title = NULL;
1.38 schwarze 1329:
1.47 schwarze 1330: while (NULL != (line = fgetln(stream, &len))) {
1331: if (' ' != *line || '\n' != line[len - 1])
1332: break;
1333: while (len > 0 && isspace((unsigned char)*line)) {
1334: line++;
1335: len--;
1336: }
1337: if (1 == len)
1338: continue;
1339: title = mandoc_realloc(title, titlesz + len);
1340: memcpy(title + titlesz, line, len);
1341: titlesz += len;
1342: title[titlesz - 1] = ' ';
1343: }
1.28 schwarze 1344:
1.47 schwarze 1345: /*
1346: * If no page content can be found, or the input line
1347: * is already the next section header, or there is no
1348: * trailing newline, reuse the page title as the page
1349: * description.
1350: */
1.1 schwarze 1351:
1.47 schwarze 1352: if (NULL == title || '\0' == *title) {
1353: if (warnings)
1354: say(mpage->mlinks->file,
1355: "Cannot find NAME section");
1356: fclose(stream);
1357: free(title);
1358: return;
1359: }
1.24 schwarze 1360:
1.47 schwarze 1361: title = mandoc_realloc(title, titlesz + 1);
1362: title[titlesz] = '\0';
1.24 schwarze 1363:
1.47 schwarze 1364: /*
1365: * Skip to the first dash.
1366: * Use the remaining line as the description (no more than 70
1367: * bytes).
1368: */
1.28 schwarze 1369:
1.47 schwarze 1370: if (NULL != (p = strstr(title, "- "))) {
1371: for (p += 2; ' ' == *p || '\b' == *p; p++)
1372: /* Skip to next word. */ ;
1373: } else {
1374: if (warnings)
1375: say(mpage->mlinks->file,
1376: "No dash in title line");
1377: p = title;
1378: }
1.1 schwarze 1379:
1.47 schwarze 1380: plen = strlen(p);
1.1 schwarze 1381:
1.47 schwarze 1382: /* Strip backspace-encoding from line. */
1.1 schwarze 1383:
1.47 schwarze 1384: while (NULL != (line = memchr(p, '\b', plen))) {
1385: len = line - p;
1386: if (0 == len) {
1387: memmove(line, line + 1, plen--);
1388: continue;
1.98 schwarze 1389: }
1.47 schwarze 1390: memmove(line - 1, line + 1, plen - len);
1391: plen -= 2;
1392: }
1.1 schwarze 1393:
1.47 schwarze 1394: mpage->desc = mandoc_strdup(p);
1395: fclose(stream);
1396: free(title);
1397: }
1.16 schwarze 1398:
1.47 schwarze 1399: /*
1400: * Put a type/word pair into the word database for this particular file.
1401: */
1402: static void
1.69 schwarze 1403: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.47 schwarze 1404: {
1.69 schwarze 1405: char *cp;
1.37 schwarze 1406:
1.47 schwarze 1407: assert(NULL != value);
1.69 schwarze 1408: if (TYPE_arch == type)
1409: for (cp = value; *cp; cp++)
1410: if (isupper((unsigned char)*cp))
1411: *cp = _tolower((unsigned char)*cp);
1.47 schwarze 1412: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1413: }
1414:
1415: /*
1.47 schwarze 1416: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1417: */
1418: static void
1.47 schwarze 1419: putmdockey(const struct mpage *mpage,
1420: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1421: {
1.16 schwarze 1422:
1.47 schwarze 1423: for ( ; NULL != n; n = n->next) {
1424: if (NULL != n->child)
1425: putmdockey(mpage, n->child, m);
1426: if (MDOC_TEXT == n->type)
1427: putkey(mpage, n->string, m);
1428: }
1429: }
1.16 schwarze 1430:
1.47 schwarze 1431: static void
1.129 ! schwarze 1432: parse_man(struct mpage *mpage, const struct man_meta *meta,
! 1433: const struct man_node *n)
1.47 schwarze 1434: {
1435: const struct man_node *head, *body;
1.78 schwarze 1436: char *start, *title;
1.47 schwarze 1437: char byte;
1.78 schwarze 1438: size_t sz;
1.16 schwarze 1439:
1.47 schwarze 1440: if (NULL == n)
1441: return;
1.16 schwarze 1442:
1.47 schwarze 1443: /*
1444: * We're only searching for one thing: the first text child in
1445: * the BODY of a NAME section. Since we don't keep track of
1446: * sections in -man, run some hoops to find out whether we're in
1447: * the correct section or not.
1448: */
1.16 schwarze 1449:
1.47 schwarze 1450: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1451: body = n;
1452: assert(body->parent);
1453: if (NULL != (head = body->parent->head) &&
1.98 schwarze 1454: 1 == head->nchild &&
1455: NULL != (head = (head->child)) &&
1456: MAN_TEXT == head->type &&
1457: 0 == strcmp(head->string, "NAME") &&
1458: NULL != body->child) {
1.2 schwarze 1459:
1.47 schwarze 1460: /*
1461: * Suck the entire NAME section into memory.
1462: * Yes, we might run away.
1463: * But too many manuals have big, spread-out
1464: * NAME sections over many lines.
1465: */
1.2 schwarze 1466:
1.78 schwarze 1467: title = NULL;
1468: man_deroff(&title, body);
1.47 schwarze 1469: if (NULL == title)
1470: return;
1.16 schwarze 1471:
1.98 schwarze 1472: /*
1.47 schwarze 1473: * Go through a special heuristic dance here.
1474: * Conventionally, one or more manual names are
1475: * comma-specified prior to a whitespace, then a
1476: * dash, then a description. Try to puzzle out
1477: * the name parts here.
1478: */
1.16 schwarze 1479:
1.78 schwarze 1480: start = title;
1.47 schwarze 1481: for ( ;; ) {
1482: sz = strcspn(start, " ,");
1483: if ('\0' == start[sz])
1484: break;
1.1 schwarze 1485:
1.47 schwarze 1486: byte = start[sz];
1487: start[sz] = '\0';
1.67 schwarze 1488:
1489: /*
1490: * Assume a stray trailing comma in the
1491: * name list if a name begins with a dash.
1492: */
1493:
1494: if ('-' == start[0] ||
1495: ('\\' == start[0] && '-' == start[1]))
1496: break;
1.1 schwarze 1497:
1.90 schwarze 1498: putkey(mpage, start, NAME_TITLE);
1.129 ! schwarze 1499: if ( ! (mpage->name_head_done ||
! 1500: strcasecmp(start, meta->title))) {
! 1501: putkey(mpage, start, NAME_HEAD);
! 1502: mpage->name_head_done = 1;
! 1503: }
1.1 schwarze 1504:
1.47 schwarze 1505: if (' ' == byte) {
1506: start += sz + 1;
1507: break;
1508: }
1.1 schwarze 1509:
1.47 schwarze 1510: assert(',' == byte);
1511: start += sz + 1;
1512: while (' ' == *start)
1513: start++;
1514: }
1.1 schwarze 1515:
1.78 schwarze 1516: if (start == title) {
1.90 schwarze 1517: putkey(mpage, start, NAME_TITLE);
1.129 ! schwarze 1518: if ( ! (mpage->name_head_done ||
! 1519: strcasecmp(start, meta->title))) {
! 1520: putkey(mpage, start, NAME_HEAD);
! 1521: mpage->name_head_done = 1;
! 1522: }
1.47 schwarze 1523: free(title);
1524: return;
1525: }
1.1 schwarze 1526:
1.47 schwarze 1527: while (isspace((unsigned char)*start))
1528: start++;
1.1 schwarze 1529:
1.47 schwarze 1530: if (0 == strncmp(start, "-", 1))
1531: start += 1;
1532: else if (0 == strncmp(start, "\\-\\-", 4))
1533: start += 4;
1534: else if (0 == strncmp(start, "\\-", 2))
1535: start += 2;
1536: else if (0 == strncmp(start, "\\(en", 4))
1537: start += 4;
1538: else if (0 == strncmp(start, "\\(em", 4))
1539: start += 4;
1.1 schwarze 1540:
1.47 schwarze 1541: while (' ' == *start)
1542: start++;
1.1 schwarze 1543:
1.47 schwarze 1544: mpage->desc = mandoc_strdup(start);
1545: free(title);
1546: return;
1547: }
1548: }
1.1 schwarze 1549:
1.47 schwarze 1550: for (n = n->child; n; n = n->next) {
1551: if (NULL != mpage->desc)
1552: break;
1.129 ! schwarze 1553: parse_man(mpage, meta, n);
1.1 schwarze 1554: }
1555: }
1556:
1557: static void
1.127 schwarze 1558: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1559: const struct mdoc_node *n)
1.1 schwarze 1560: {
1561:
1.47 schwarze 1562: assert(NULL != n);
1563: for (n = n->child; NULL != n; n = n->next) {
1564: switch (n->type) {
1.98 schwarze 1565: case MDOC_ELEM:
1.47 schwarze 1566: /* FALLTHROUGH */
1.98 schwarze 1567: case MDOC_BLOCK:
1.47 schwarze 1568: /* FALLTHROUGH */
1.98 schwarze 1569: case MDOC_HEAD:
1.47 schwarze 1570: /* FALLTHROUGH */
1.98 schwarze 1571: case MDOC_BODY:
1.47 schwarze 1572: /* FALLTHROUGH */
1.98 schwarze 1573: case MDOC_TAIL:
1.47 schwarze 1574: if (NULL != mdocs[n->tok].fp)
1.127 schwarze 1575: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.47 schwarze 1576: break;
1577: if (mdocs[n->tok].mask)
1578: putmdockey(mpage, n->child,
1579: mdocs[n->tok].mask);
1580: break;
1581: default:
1582: assert(MDOC_ROOT != n->type);
1583: continue;
1584: }
1585: if (NULL != n->child)
1.127 schwarze 1586: parse_mdoc(mpage, meta, n);
1.1 schwarze 1587: }
1588: }
1589:
1.19 schwarze 1590: static int
1.127 schwarze 1591: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1592: const struct mdoc_node *n)
1.1 schwarze 1593: {
1594: const char *start, *end;
1595: size_t sz;
1.19 schwarze 1596:
1.47 schwarze 1597: if (SEC_SYNOPSIS != n->sec ||
1.98 schwarze 1598: NULL == (n = n->child) ||
1599: MDOC_TEXT != n->type)
1.19 schwarze 1600: return(0);
1.1 schwarze 1601:
1602: /*
1603: * Only consider those `Fd' macro fields that begin with an
1604: * "inclusion" token (versus, e.g., #define).
1605: */
1.47 schwarze 1606:
1.1 schwarze 1607: if (strcmp("#include", n->string))
1.19 schwarze 1608: return(0);
1.1 schwarze 1609:
1610: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1611: return(0);
1.1 schwarze 1612:
1613: /*
1614: * Strip away the enclosing angle brackets and make sure we're
1615: * not zero-length.
1616: */
1617:
1618: start = n->string;
1619: if ('<' == *start || '"' == *start)
1620: start++;
1621:
1622: if (0 == (sz = strlen(start)))
1.19 schwarze 1623: return(0);
1.1 schwarze 1624:
1625: end = &start[(int)sz - 1];
1626: if ('>' == *end || '"' == *end)
1627: end--;
1628:
1.47 schwarze 1629: if (end > start)
1630: putkeys(mpage, start, end - start + 1, TYPE_In);
1.49 schwarze 1631: return(0);
1.1 schwarze 1632: }
1633:
1.19 schwarze 1634: static int
1.127 schwarze 1635: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1636: const struct mdoc_node *n)
1.1 schwarze 1637: {
1.69 schwarze 1638: char *cp;
1.1 schwarze 1639:
1.47 schwarze 1640: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1641: return(0);
1642:
1.98 schwarze 1643: /*
1.47 schwarze 1644: * Parse: .Fn "struct type *name" "char *arg".
1.98 schwarze 1645: * First strip away pointer symbol.
1.47 schwarze 1646: * Then store the function name, then type.
1.98 schwarze 1647: * Finally, store the arguments.
1.47 schwarze 1648: */
1.1 schwarze 1649:
1.47 schwarze 1650: if (NULL == (cp = strrchr(n->string, ' ')))
1651: cp = n->string;
1.1 schwarze 1652:
1653: while ('*' == *cp)
1654: cp++;
1655:
1.47 schwarze 1656: putkey(mpage, cp, TYPE_Fn);
1.128 schwarze 1657: if (n->sec == SEC_SYNOPSIS)
1658: putkey(mpage, cp, NAME_SYN);
1.19 schwarze 1659:
1.47 schwarze 1660: if (n->string < cp)
1661: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1662:
1.47 schwarze 1663: for (n = n->next; NULL != n; n = n->next)
1664: if (MDOC_TEXT == n->type)
1665: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1666:
1.128 schwarze 1667: return(0);
1668: }
1669:
1670: static int
1671: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1672: const struct mdoc_node *n)
1673: {
1674:
1675: putmdockey(mpage, n->child, TYPE_Fn);
1676: if (n->sec == SEC_SYNOPSIS)
1677: putmdockey(mpage, n->child, NAME_SYN);
1.19 schwarze 1678: return(0);
1.1 schwarze 1679: }
1680:
1.19 schwarze 1681: static int
1.127 schwarze 1682: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1683: const struct mdoc_node *n)
1.1 schwarze 1684: {
1.47 schwarze 1685: char *cp;
1.1 schwarze 1686:
1687: if (NULL == (n = n->child))
1.19 schwarze 1688: return(0);
1.1 schwarze 1689:
1.47 schwarze 1690: if (NULL == n->next) {
1691: putkey(mpage, n->string, TYPE_Xr);
1692: return(0);
1693: }
1.1 schwarze 1694:
1.77 schwarze 1695: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.47 schwarze 1696: putkey(mpage, cp, TYPE_Xr);
1697: free(cp);
1698: return(0);
1.1 schwarze 1699: }
1700:
1.19 schwarze 1701: static int
1.127 schwarze 1702: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1703: const struct mdoc_node *n)
1.1 schwarze 1704: {
1705:
1.79 schwarze 1706: if (MDOC_BODY == n->type)
1707: mdoc_deroff(&mpage->desc, n);
1708: return(0);
1.1 schwarze 1709: }
1710:
1.19 schwarze 1711: static int
1.127 schwarze 1712: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1713: const struct mdoc_node *n)
1.1 schwarze 1714: {
1715:
1.86 schwarze 1716: if (SEC_NAME == n->sec)
1.90 schwarze 1717: putmdockey(mpage, n->child, NAME_TITLE);
1.127 schwarze 1718: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1719: if (n->child == NULL)
1720: putkey(mpage, meta->name, NAME_SYN);
1721: else
1722: putmdockey(mpage, n->child, NAME_SYN);
1.129 ! schwarze 1723: }
! 1724: if ( ! (mpage->name_head_done ||
! 1725: n->child == NULL || n->child->string == NULL ||
! 1726: strcasecmp(n->child->string, meta->title))) {
! 1727: putkey(mpage, n->child->string, NAME_HEAD);
! 1728: mpage->name_head_done = 1;
1.127 schwarze 1729: }
1.86 schwarze 1730: return(0);
1.1 schwarze 1731: }
1732:
1.19 schwarze 1733: static int
1.127 schwarze 1734: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1735: const struct mdoc_node *n)
1.1 schwarze 1736: {
1737:
1.19 schwarze 1738: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1739: }
1740:
1.47 schwarze 1741: static int
1.127 schwarze 1742: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1743: const struct mdoc_node *n)
1.1 schwarze 1744: {
1745:
1.47 schwarze 1746: return(MDOC_HEAD == n->type);
1747: }
1.1 schwarze 1748:
1.47 schwarze 1749: static int
1.127 schwarze 1750: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1751: const struct mdoc_node *n)
1.47 schwarze 1752: {
1.1 schwarze 1753:
1.47 schwarze 1754: return(MDOC_BODY == n->type);
1.1 schwarze 1755: }
1756:
1.47 schwarze 1757: /*
1758: * Add a string to the hash table for the current manual.
1759: * Each string has a bitmask telling which macros it belongs to.
1760: * When we finish the manual, we'll dump the table.
1761: */
1.1 schwarze 1762: static void
1.47 schwarze 1763: putkeys(const struct mpage *mpage,
1764: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1765: {
1.90 schwarze 1766: struct ohash *htab;
1.47 schwarze 1767: struct str *s;
1.68 schwarze 1768: const char *end;
1.47 schwarze 1769: unsigned int slot;
1.68 schwarze 1770: int i;
1.1 schwarze 1771:
1.47 schwarze 1772: if (0 == sz)
1773: return;
1.68 schwarze 1774:
1.90 schwarze 1775: if (TYPE_Nm & v) {
1776: htab = &names;
1777: v &= name_mask;
1.124 schwarze 1778: if (v & NAME_FIRST)
1779: name_mask &= ~NAME_FIRST;
1.90 schwarze 1780: if (debug > 1)
1781: say(mpage->mlinks->file,
1782: "Adding name %*s", sz, cp);
1783: } else {
1784: htab = &strings;
1785: if (debug > 1)
1786: for (i = 0; i < mansearch_keymax; i++)
1.118 schwarze 1787: if ((uint64_t)1 << i & v)
1.90 schwarze 1788: say(mpage->mlinks->file,
1789: "Adding key %s=%*s",
1790: mansearch_keynames[i], sz, cp);
1.68 schwarze 1791: }
1.47 schwarze 1792:
1793: end = cp + sz;
1.90 schwarze 1794: slot = ohash_qlookupi(htab, cp, &end);
1795: s = ohash_find(htab, slot);
1.1 schwarze 1796:
1.47 schwarze 1797: if (NULL != s && mpage == s->mpage) {
1798: s->mask |= v;
1.1 schwarze 1799: return;
1.47 schwarze 1800: } else if (NULL == s) {
1.101 schwarze 1801: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.47 schwarze 1802: memcpy(s->key, cp, sz);
1.90 schwarze 1803: ohash_insert(htab, slot, s);
1.47 schwarze 1804: }
1805: s->mpage = mpage;
1806: s->mask = v;
1.1 schwarze 1807: }
1808:
1809: /*
1.47 schwarze 1810: * Take a Unicode codepoint and produce its UTF-8 encoding.
1811: * This isn't the best way to do this, but it works.
1812: * The magic numbers are from the UTF-8 packaging.
1813: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1814: */
1.47 schwarze 1815: static size_t
1816: utf8(unsigned int cp, char out[7])
1.1 schwarze 1817: {
1.47 schwarze 1818: size_t rc;
1.1 schwarze 1819:
1.47 schwarze 1820: rc = 0;
1821: if (cp <= 0x0000007F) {
1822: rc = 1;
1823: out[0] = (char)cp;
1824: } else if (cp <= 0x000007FF) {
1825: rc = 2;
1826: out[0] = (cp >> 6 & 31) | 192;
1827: out[1] = (cp & 63) | 128;
1828: } else if (cp <= 0x0000FFFF) {
1829: rc = 3;
1830: out[0] = (cp >> 12 & 15) | 224;
1831: out[1] = (cp >> 6 & 63) | 128;
1832: out[2] = (cp & 63) | 128;
1833: } else if (cp <= 0x001FFFFF) {
1834: rc = 4;
1835: out[0] = (cp >> 18 & 7) | 240;
1836: out[1] = (cp >> 12 & 63) | 128;
1837: out[2] = (cp >> 6 & 63) | 128;
1838: out[3] = (cp & 63) | 128;
1839: } else if (cp <= 0x03FFFFFF) {
1840: rc = 5;
1841: out[0] = (cp >> 24 & 3) | 248;
1842: out[1] = (cp >> 18 & 63) | 128;
1843: out[2] = (cp >> 12 & 63) | 128;
1844: out[3] = (cp >> 6 & 63) | 128;
1845: out[4] = (cp & 63) | 128;
1846: } else if (cp <= 0x7FFFFFFF) {
1847: rc = 6;
1848: out[0] = (cp >> 30 & 1) | 252;
1849: out[1] = (cp >> 24 & 63) | 128;
1850: out[2] = (cp >> 18 & 63) | 128;
1851: out[3] = (cp >> 12 & 63) | 128;
1852: out[4] = (cp >> 6 & 63) | 128;
1853: out[5] = (cp & 63) | 128;
1854: } else
1855: return(0);
1.19 schwarze 1856:
1.47 schwarze 1857: out[rc] = '\0';
1858: return(rc);
1.1 schwarze 1859: }
1860:
1.47 schwarze 1861: /*
1.53 schwarze 1862: * Store the rendered version of a key, or alias the pointer
1863: * if the key contains no escape sequences.
1.47 schwarze 1864: */
1865: static void
1.53 schwarze 1866: render_key(struct mchars *mc, struct str *key)
1.1 schwarze 1867: {
1.47 schwarze 1868: size_t sz, bsz, pos;
1.71 schwarze 1869: char utfbuf[7], res[6];
1.47 schwarze 1870: char *buf;
1871: const char *seq, *cpp, *val;
1872: int len, u;
1873: enum mandoc_esc esc;
1874:
1.53 schwarze 1875: assert(NULL == key->rendered);
1.47 schwarze 1876:
1877: res[0] = '\\';
1878: res[1] = '\t';
1879: res[2] = ASCII_NBRSP;
1880: res[3] = ASCII_HYPH;
1.71 schwarze 1881: res[4] = ASCII_BREAK;
1882: res[5] = '\0';
1.1 schwarze 1883:
1.47 schwarze 1884: val = key->key;
1885: bsz = strlen(val);
1.1 schwarze 1886:
1887: /*
1.47 schwarze 1888: * Pre-check: if we have no stop-characters, then set the
1889: * pointer as ourselvse and get out of here.
1.1 schwarze 1890: */
1.47 schwarze 1891: if (strcspn(val, res) == bsz) {
1.53 schwarze 1892: key->rendered = key->key;
1.47 schwarze 1893: return;
1.98 schwarze 1894: }
1.1 schwarze 1895:
1.47 schwarze 1896: /* Pre-allocate by the length of the input */
1.39 schwarze 1897:
1.47 schwarze 1898: buf = mandoc_malloc(++bsz);
1899: pos = 0;
1.39 schwarze 1900:
1.47 schwarze 1901: while ('\0' != *val) {
1902: /*
1903: * Halt on the first escape sequence.
1904: * This also halts on the end of string, in which case
1905: * we just copy, fallthrough, and exit the loop.
1906: */
1907: if ((sz = strcspn(val, res)) > 0) {
1908: memcpy(&buf[pos], val, sz);
1909: pos += sz;
1910: val += sz;
1911: }
1.39 schwarze 1912:
1.71 schwarze 1913: switch (*val) {
1.98 schwarze 1914: case ASCII_HYPH:
1.47 schwarze 1915: buf[pos++] = '-';
1916: val++;
1917: continue;
1.98 schwarze 1918: case '\t':
1.71 schwarze 1919: /* FALLTHROUGH */
1.98 schwarze 1920: case ASCII_NBRSP:
1.47 schwarze 1921: buf[pos++] = ' ';
1922: val++;
1.71 schwarze 1923: /* FALLTHROUGH */
1.98 schwarze 1924: case ASCII_BREAK:
1.47 schwarze 1925: continue;
1.71 schwarze 1926: default:
1927: break;
1928: }
1929: if ('\\' != *val)
1.47 schwarze 1930: break;
1.39 schwarze 1931:
1.47 schwarze 1932: /* Read past the slash. */
1.39 schwarze 1933:
1.47 schwarze 1934: val++;
1.39 schwarze 1935:
1.47 schwarze 1936: /*
1937: * Parse the escape sequence and see if it's a
1938: * predefined character or special character.
1939: */
1.52 schwarze 1940:
1.98 schwarze 1941: esc = mandoc_escape((const char **)&val,
1942: &seq, &len);
1.47 schwarze 1943: if (ESCAPE_ERROR == esc)
1944: break;
1945: if (ESCAPE_SPECIAL != esc)
1946: continue;
1.39 schwarze 1947:
1.47 schwarze 1948: /*
1.52 schwarze 1949: * Render the special character
1950: * as either UTF-8 or ASCII.
1.47 schwarze 1951: */
1.52 schwarze 1952:
1953: if (write_utf8) {
1.122 schwarze 1954: if ((u = mchars_spec2cp(mc, seq, len)) <= 0)
1.52 schwarze 1955: continue;
1956: cpp = utfbuf;
1957: if (0 == (sz = utf8(u, utfbuf)))
1958: continue;
1959: sz = strlen(cpp);
1960: } else {
1961: cpp = mchars_spec2str(mc, seq, len, &sz);
1962: if (NULL == cpp)
1963: continue;
1964: if (ASCII_NBRSP == *cpp) {
1965: cpp = " ";
1966: sz = 1;
1967: }
1968: }
1.1 schwarze 1969:
1.47 schwarze 1970: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1971:
1.47 schwarze 1972: bsz += sz;
1973: buf = mandoc_realloc(buf, bsz);
1974: memcpy(&buf[pos], cpp, sz);
1975: pos += sz;
1.1 schwarze 1976: }
1977:
1.47 schwarze 1978: buf[pos] = '\0';
1.53 schwarze 1979: key->rendered = buf;
1.1 schwarze 1980: }
1981:
1.75 schwarze 1982: static void
1983: dbadd_mlink(const struct mlink *mlink)
1984: {
1985: size_t i;
1986:
1987: i = 1;
1988: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1989: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1990: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.94 schwarze 1991: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.75 schwarze 1992: SQL_STEP(stmts[STMT_INSERT_LINK]);
1993: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.124 schwarze 1994: }
1995:
1996: static void
1997: dbadd_mlink_name(const struct mlink *mlink)
1998: {
1999: size_t i;
2000:
2001: dbadd_mlink(mlink);
1.115 schwarze 2002:
2003: i = 1;
1.124 schwarze 2004: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, NAME_FILE & NAME_MASK);
1.115 schwarze 2005: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2006: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2007: SQL_STEP(stmts[STMT_INSERT_NAME]);
2008: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.75 schwarze 2009: }
2010:
1.11 schwarze 2011: /*
1.47 schwarze 2012: * Flush the current page's terms (and their bits) into the database.
2013: * Wrap the entire set of additions in a transaction to make sqlite be a
2014: * little faster.
1.53 schwarze 2015: * Also, handle escape sequences at the last possible moment.
1.11 schwarze 2016: */
2017: static void
1.75 schwarze 2018: dbadd(struct mpage *mpage, struct mchars *mc)
1.11 schwarze 2019: {
1.47 schwarze 2020: struct mlink *mlink;
2021: struct str *key;
2022: size_t i;
2023: unsigned int slot;
2024:
1.85 schwarze 2025: mlink = mpage->mlinks;
1.11 schwarze 2026:
1.85 schwarze 2027: if (nodb) {
1.104 schwarze 2028: for (key = ohash_first(&names, &slot); NULL != key;
2029: key = ohash_next(&names, &slot)) {
2030: if (key->rendered != key->key)
2031: free(key->rendered);
2032: free(key);
2033: }
2034: for (key = ohash_first(&strings, &slot); NULL != key;
2035: key = ohash_next(&strings, &slot)) {
2036: if (key->rendered != key->key)
2037: free(key->rendered);
2038: free(key);
2039: }
1.102 schwarze 2040: if (0 == debug)
2041: return;
1.85 schwarze 2042: while (NULL != mlink) {
2043: fputs(mlink->name, stdout);
2044: if (NULL == mlink->next ||
2045: strcmp(mlink->dsec, mlink->next->dsec) ||
2046: strcmp(mlink->fsec, mlink->next->fsec) ||
2047: strcmp(mlink->arch, mlink->next->arch)) {
2048: putchar('(');
2049: if ('\0' == *mlink->dsec)
2050: fputs(mlink->fsec, stdout);
2051: else
2052: fputs(mlink->dsec, stdout);
2053: if ('\0' != *mlink->arch)
2054: printf("/%s", mlink->arch);
2055: putchar(')');
2056: }
2057: mlink = mlink->next;
2058: if (NULL != mlink)
2059: fputs(", ", stdout);
2060: }
1.89 schwarze 2061: printf(" - %s\n", mpage->desc);
1.11 schwarze 2062: return;
1.85 schwarze 2063: }
2064:
2065: if (debug)
2066: say(mlink->file, "Adding to database");
1.47 schwarze 2067:
1.105 schwarze 2068: i = strlen(mpage->desc) + 1;
2069: key = mandoc_calloc(1, sizeof(struct str) + i);
2070: memcpy(key->key, mpage->desc, i);
2071: render_key(mc, key);
2072:
1.47 schwarze 2073: i = 1;
1.105 schwarze 2074: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
1.116 schwarze 2075: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.47 schwarze 2076: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.94 schwarze 2077: mpage->pageid = sqlite3_last_insert_rowid(db);
1.47 schwarze 2078: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.105 schwarze 2079:
2080: if (key->rendered != key->key)
2081: free(key->rendered);
2082: free(key);
1.47 schwarze 2083:
1.85 schwarze 2084: while (NULL != mlink) {
1.75 schwarze 2085: dbadd_mlink(mlink);
1.85 schwarze 2086: mlink = mlink->next;
2087: }
1.91 schwarze 2088: mlink = mpage->mlinks;
1.47 schwarze 2089:
1.90 schwarze 2090: for (key = ohash_first(&names, &slot); NULL != key;
2091: key = ohash_next(&names, &slot)) {
2092: assert(key->mpage == mpage);
2093: if (NULL == key->rendered)
2094: render_key(mc, key);
2095: i = 1;
2096: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2097: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1.94 schwarze 2098: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.90 schwarze 2099: SQL_STEP(stmts[STMT_INSERT_NAME]);
2100: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2101: if (key->rendered != key->key)
2102: free(key->rendered);
2103: free(key);
2104: }
1.47 schwarze 2105: for (key = ohash_first(&strings, &slot); NULL != key;
2106: key = ohash_next(&strings, &slot)) {
2107: assert(key->mpage == mpage);
1.53 schwarze 2108: if (NULL == key->rendered)
2109: render_key(mc, key);
1.47 schwarze 2110: i = 1;
2111: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.53 schwarze 2112: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1.94 schwarze 2113: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.47 schwarze 2114: SQL_STEP(stmts[STMT_INSERT_KEY]);
2115: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.53 schwarze 2116: if (key->rendered != key->key)
2117: free(key->rendered);
1.47 schwarze 2118: free(key);
1.33 schwarze 2119: }
1.47 schwarze 2120: }
1.41 deraadt 2121:
1.47 schwarze 2122: static void
2123: dbprune(void)
2124: {
2125: struct mpage *mpage;
2126: struct mlink *mlink;
2127: size_t i;
2128: unsigned int slot;
1.11 schwarze 2129:
1.63 schwarze 2130: if (0 == nodb)
2131: SQL_EXEC("BEGIN TRANSACTION");
1.47 schwarze 2132:
1.63 schwarze 2133: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2134: mpage = ohash_next(&mpages, &slot)) {
1.47 schwarze 2135: mlink = mpage->mlinks;
1.82 schwarze 2136: if (debug)
1.63 schwarze 2137: say(mlink->file, "Deleting from database");
2138: if (nodb)
2139: continue;
2140: for ( ; NULL != mlink; mlink = mlink->next) {
2141: i = 1;
2142: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2143: i, mlink->dsec);
2144: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2145: i, mlink->arch);
2146: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2147: i, mlink->name);
2148: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2149: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2150: }
1.11 schwarze 2151: }
1.63 schwarze 2152:
2153: if (0 == nodb)
2154: SQL_EXEC("END TRANSACTION");
1.47 schwarze 2155: }
1.22 schwarze 2156:
1.47 schwarze 2157: /*
2158: * Close an existing database and its prepared statements.
2159: * If "real" is not set, rename the temporary file into the real one.
2160: */
2161: static void
2162: dbclose(int real)
2163: {
2164: size_t i;
1.72 schwarze 2165: int status;
2166: pid_t child;
1.11 schwarze 2167:
1.47 schwarze 2168: if (nodb)
2169: return;
1.11 schwarze 2170:
1.47 schwarze 2171: for (i = 0; i < STMT__MAX; i++) {
2172: sqlite3_finalize(stmts[i]);
2173: stmts[i] = NULL;
1.28 schwarze 2174: }
1.22 schwarze 2175:
1.47 schwarze 2176: sqlite3_close(db);
2177: db = NULL;
1.11 schwarze 2178:
1.47 schwarze 2179: if (real)
2180: return;
1.22 schwarze 2181:
1.72 schwarze 2182: if ('\0' == *tempfilename) {
2183: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2184: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2185: say(MANDOC_DB, "&rename");
1.72 schwarze 2186: }
2187: return;
2188: }
2189:
2190: switch (child = fork()) {
1.98 schwarze 2191: case -1:
1.72 schwarze 2192: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2193: say("", "&fork cmp");
1.72 schwarze 2194: return;
1.98 schwarze 2195: case 0:
1.72 schwarze 2196: execlp("cmp", "cmp", "-s",
2197: tempfilename, MANDOC_DB, NULL);
1.80 schwarze 2198: say("", "&exec cmp");
1.72 schwarze 2199: exit(0);
2200: default:
2201: break;
2202: }
2203: if (-1 == waitpid(child, &status, 0)) {
2204: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2205: say("", "&wait cmp");
1.72 schwarze 2206: } else if (WIFSIGNALED(status)) {
2207: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2208: say("", "cmp died from signal %d", WTERMSIG(status));
1.72 schwarze 2209: } else if (WEXITSTATUS(status)) {
1.47 schwarze 2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2211: say(MANDOC_DB,
2212: "Data changed, but cannot replace database");
2213: }
2214:
2215: *strrchr(tempfilename, '/') = '\0';
2216: switch (child = fork()) {
1.98 schwarze 2217: case -1:
1.72 schwarze 2218: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2219: say("", "&fork rm");
1.72 schwarze 2220: return;
1.98 schwarze 2221: case 0:
1.72 schwarze 2222: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.80 schwarze 2223: say("", "&exec rm");
1.72 schwarze 2224: exit((int)MANDOCLEVEL_SYSERR);
2225: default:
2226: break;
2227: }
2228: if (-1 == waitpid(child, &status, 0)) {
2229: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2230: say("", "&wait rm");
1.72 schwarze 2231: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2233: say("", "%s: Cannot remove temporary directory",
2234: tempfilename);
1.22 schwarze 2235: }
1.11 schwarze 2236: }
2237:
1.47 schwarze 2238: /*
2239: * This is straightforward stuff.
2240: * Open a database connection to a "temporary" database, then open a set
2241: * of prepared statements we'll use over and over again.
2242: * If "real" is set, we use the existing database; if not, we truncate a
2243: * temporary one.
2244: * Must be matched by dbclose().
2245: */
2246: static int
2247: dbopen(int real)
1.2 schwarze 2248: {
1.72 schwarze 2249: const char *sql;
1.47 schwarze 2250: int rc, ofl;
1.6 schwarze 2251:
1.98 schwarze 2252: if (nodb)
1.47 schwarze 2253: return(1);
1.6 schwarze 2254:
1.72 schwarze 2255: *tempfilename = '\0';
1.47 schwarze 2256: ofl = SQLITE_OPEN_READWRITE;
1.72 schwarze 2257:
2258: if (real) {
2259: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2260: if (SQLITE_OK != rc) {
1.47 schwarze 2261: exitcode = (int)MANDOCLEVEL_SYSERR;
1.108 sthen 2262: if (SQLITE_CANTOPEN != rc)
2263: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.47 schwarze 2264: return(0);
1.28 schwarze 2265: }
1.72 schwarze 2266: goto prepare_statements;
2267: }
2268:
2269: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.6 schwarze 2270:
1.72 schwarze 2271: remove(MANDOC_DB "~");
2272: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.98 schwarze 2273: if (SQLITE_OK == rc)
1.72 schwarze 2274: goto create_tables;
1.73 schwarze 2275: if (MPARSE_QUICK & mparse_options) {
1.47 schwarze 2276: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2277: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.47 schwarze 2278: return(0);
2279: }
1.6 schwarze 2280:
1.100 schwarze 2281: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2282: sizeof(tempfilename));
1.72 schwarze 2283: if (NULL == mkdtemp(tempfilename)) {
2284: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2285: say("", "&%s", tempfilename);
1.72 schwarze 2286: return(0);
2287: }
1.100 schwarze 2288: (void)strlcat(tempfilename, "/" MANDOC_DB,
2289: sizeof(tempfilename));
1.72 schwarze 2290: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2291: if (SQLITE_OK != rc) {
1.47 schwarze 2292: exitcode = (int)MANDOCLEVEL_SYSERR;
1.103 schwarze 2293: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.47 schwarze 2294: return(0);
1.2 schwarze 2295: }
2296:
1.72 schwarze 2297: create_tables:
1.47 schwarze 2298: sql = "CREATE TABLE \"mpages\" (\n"
1.89 schwarze 2299: " \"desc\" TEXT NOT NULL,\n"
1.47 schwarze 2300: " \"form\" INTEGER NOT NULL,\n"
1.94 schwarze 2301: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.47 schwarze 2302: ");\n"
2303: "\n"
2304: "CREATE TABLE \"mlinks\" (\n"
2305: " \"sec\" TEXT NOT NULL,\n"
2306: " \"arch\" TEXT NOT NULL,\n"
2307: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2308: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2309: "ON DELETE CASCADE\n"
1.47 schwarze 2310: ");\n"
1.93 schwarze 2311: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.47 schwarze 2312: "\n"
1.90 schwarze 2313: "CREATE TABLE \"names\" (\n"
2314: " \"bits\" INTEGER NOT NULL,\n"
2315: " \"name\" TEXT NOT NULL,\n"
1.94 schwarze 2316: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.90 schwarze 2317: "ON DELETE CASCADE\n"
2318: ");\n"
2319: "\n"
1.47 schwarze 2320: "CREATE TABLE \"keys\" (\n"
2321: " \"bits\" INTEGER NOT NULL,\n"
2322: " \"key\" TEXT NOT NULL,\n"
1.94 schwarze 2323: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.66 schwarze 2324: "ON DELETE CASCADE\n"
1.93 schwarze 2325: ");\n"
2326: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.47 schwarze 2327:
2328: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2329: exitcode = (int)MANDOCLEVEL_SYSERR;
1.72 schwarze 2330: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.103 schwarze 2331: sqlite3_close(db);
1.47 schwarze 2332: return(0);
1.2 schwarze 2333: }
2334:
1.47 schwarze 2335: prepare_statements:
1.103 schwarze 2336: if (SQLITE_OK != sqlite3_exec(db,
2337: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2338: exitcode = (int)MANDOCLEVEL_SYSERR;
2339: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2340: sqlite3_errmsg(db));
2341: sqlite3_close(db);
2342: return(0);
2343: }
2344:
1.94 schwarze 2345: sql = "DELETE FROM mpages WHERE pageid IN "
1.63 schwarze 2346: "(SELECT pageid FROM mlinks WHERE "
2347: "sec=? AND arch=? AND name=?)";
1.47 schwarze 2348: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2349: sql = "INSERT INTO mpages "
1.89 schwarze 2350: "(desc,form) VALUES (?,?)";
1.47 schwarze 2351: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2352: sql = "INSERT INTO mlinks "
1.61 schwarze 2353: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.47 schwarze 2354: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.90 schwarze 2355: sql = "INSERT INTO names "
2356: "(bits,name,pageid) VALUES (?,?,?)";
2357: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.47 schwarze 2358: sql = "INSERT INTO keys "
2359: "(bits,key,pageid) VALUES (?,?,?)";
2360: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2361:
1.47 schwarze 2362: /*
2363: * When opening a new database, we can turn off
2364: * synchronous mode for much better performance.
2365: */
1.6 schwarze 2366:
1.103 schwarze 2367: if (real && SQLITE_OK != sqlite3_exec(db,
2368: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2369: exitcode = (int)MANDOCLEVEL_SYSERR;
2370: say(MANDOC_DB, "PRAGMA synchronous: %s",
2371: sqlite3_errmsg(db));
2372: sqlite3_close(db);
2373: return(0);
2374: }
1.11 schwarze 2375:
1.47 schwarze 2376: return(1);
2377: }
1.6 schwarze 2378:
1.47 schwarze 2379: static void *
1.106 espie 2380: hash_calloc(size_t n, size_t sz, void *arg)
1.47 schwarze 2381: {
1.6 schwarze 2382:
1.106 espie 2383: return(mandoc_calloc(n, sz));
1.47 schwarze 2384: }
1.2 schwarze 2385:
1.47 schwarze 2386: static void *
2387: hash_alloc(size_t sz, void *arg)
2388: {
1.28 schwarze 2389:
1.47 schwarze 2390: return(mandoc_malloc(sz));
2391: }
1.6 schwarze 2392:
1.47 schwarze 2393: static void
1.106 espie 2394: hash_free(void *p, void *arg)
1.47 schwarze 2395: {
1.26 schwarze 2396:
1.47 schwarze 2397: free(p);
2398: }
1.6 schwarze 2399:
1.47 schwarze 2400: static int
1.120 schwarze 2401: set_basedir(const char *targetdir, int report_baddir)
1.47 schwarze 2402: {
2403: static char startdir[PATH_MAX];
1.110 schwarze 2404: static int getcwd_status; /* 1 = ok, 2 = failure */
2405: static int chdir_status; /* 1 = changed directory */
1.109 schwarze 2406: char *cp;
1.6 schwarze 2407:
1.47 schwarze 2408: /*
1.110 schwarze 2409: * Remember the original working directory, if possible.
2410: * This will be needed if the second or a later directory
2411: * on the command line is given as a relative path.
2412: * Do not error out if the current directory is not
2413: * searchable: Maybe it won't be needed after all.
2414: */
2415: if (0 == getcwd_status) {
2416: if (NULL == getcwd(startdir, sizeof(startdir))) {
2417: getcwd_status = 2;
2418: (void)strlcpy(startdir, strerror(errno),
2419: sizeof(startdir));
2420: } else
2421: getcwd_status = 1;
2422: }
2423:
2424: /*
2425: * We are leaving the old base directory.
2426: * Do not use it any longer, not even for messages.
2427: */
2428: *basedir = '\0';
2429:
2430: /*
2431: * If and only if the directory was changed earlier and
2432: * the next directory to process is given as a relative path,
2433: * first go back, or bail out if that is impossible.
1.47 schwarze 2434: */
1.110 schwarze 2435: if (chdir_status && '/' != *targetdir) {
2436: if (2 == getcwd_status) {
1.47 schwarze 2437: exitcode = (int)MANDOCLEVEL_SYSERR;
1.110 schwarze 2438: say("", "getcwd: %s", startdir);
1.47 schwarze 2439: return(0);
2440: }
1.110 schwarze 2441: if (-1 == chdir(startdir)) {
1.47 schwarze 2442: exitcode = (int)MANDOCLEVEL_SYSERR;
1.80 schwarze 2443: say("", "&chdir %s", startdir);
1.47 schwarze 2444: return(0);
1.2 schwarze 2445: }
2446: }
1.110 schwarze 2447:
2448: /*
2449: * Always resolve basedir to the canonicalized absolute
2450: * pathname and append a trailing slash, such that
2451: * we can reliably check whether files are inside.
2452: */
1.47 schwarze 2453: if (NULL == realpath(targetdir, basedir)) {
1.120 schwarze 2454: if (report_baddir || errno != ENOENT) {
2455: exitcode = (int)MANDOCLEVEL_BADARG;
2456: say("", "&%s: realpath", targetdir);
2457: }
1.47 schwarze 2458: return(0);
2459: } else if (-1 == chdir(basedir)) {
1.120 schwarze 2460: if (report_baddir || errno != ENOENT) {
2461: exitcode = (int)MANDOCLEVEL_BADARG;
2462: say("", "&chdir");
2463: }
1.47 schwarze 2464: return(0);
1.109 schwarze 2465: }
1.110 schwarze 2466: chdir_status = 1;
1.109 schwarze 2467: cp = strchr(basedir, '\0');
2468: if ('/' != cp[-1]) {
2469: if (cp - basedir >= PATH_MAX - 1) {
2470: exitcode = (int)MANDOCLEVEL_SYSERR;
2471: say("", "Filename too long");
2472: return(0);
2473: }
2474: *cp++ = '/';
2475: *cp = '\0';
1.47 schwarze 2476: }
2477: return(1);
1.2 schwarze 2478: }
2479:
2480: static void
1.47 schwarze 2481: say(const char *file, const char *format, ...)
1.2 schwarze 2482: {
1.47 schwarze 2483: va_list ap;
1.80 schwarze 2484: int use_errno;
1.2 schwarze 2485:
1.47 schwarze 2486: if ('\0' != *basedir)
2487: fprintf(stderr, "%s", basedir);
2488: if ('\0' != *basedir && '\0' != *file)
1.110 schwarze 2489: fputc('/', stderr);
1.47 schwarze 2490: if ('\0' != *file)
2491: fprintf(stderr, "%s", file);
1.31 schwarze 2492:
1.80 schwarze 2493: use_errno = 1;
2494: if (NULL != format) {
2495: switch (*format) {
1.98 schwarze 2496: case '&':
1.80 schwarze 2497: format++;
2498: break;
1.98 schwarze 2499: case '\0':
1.80 schwarze 2500: format = NULL;
2501: break;
2502: default:
2503: use_errno = 0;
2504: break;
2505: }
2506: }
2507: if (NULL != format) {
2508: if ('\0' != *basedir || '\0' != *file)
2509: fputs(": ", stderr);
2510: va_start(ap, format);
2511: vfprintf(stderr, format, ap);
2512: va_end(ap);
2513: }
2514: if (use_errno) {
2515: if ('\0' != *basedir || '\0' != *file || NULL != format)
2516: fputs(": ", stderr);
1.47 schwarze 2517: perror(NULL);
1.80 schwarze 2518: } else
2519: fputc('\n', stderr);
1.1 schwarze 2520: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mandocdb.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc