Annotation of mandoc/mandocdb.c, Revision 1.180
1.180 ! schwarze 1: /* $Id: mandocdb.c,v 1.179 2014/12/09 07:29:42 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: #include "config.h"
19:
1.162 schwarze 20: #include <sys/types.h>
1.50 kristaps 21: #include <sys/stat.h>
1.115 schwarze 22: #include <sys/wait.h>
1.1 kristaps 23:
24: #include <assert.h>
1.43 kristaps 25: #include <ctype.h>
1.63 schwarze 26: #include <errno.h>
1.1 kristaps 27: #include <fcntl.h>
1.158 schwarze 28: #if HAVE_FTS
1.50 kristaps 29: #include <fts.h>
1.157 schwarze 30: #else
31: #include "compat_fts.h"
32: #endif
1.1 kristaps 33: #include <getopt.h>
1.58 schwarze 34: #include <limits.h>
1.50 kristaps 35: #include <stddef.h>
1.59 schwarze 36: #include <stdio.h>
1.1 kristaps 37: #include <stdint.h>
38: #include <stdlib.h>
39: #include <string.h>
1.17 schwarze 40: #include <unistd.h>
1.1 kristaps 41:
1.158 schwarze 42: #if HAVE_OHASH
1.50 kristaps 43: #include <ohash.h>
1.53 kristaps 44: #else
45: #include "compat_ohash.h"
46: #endif
1.50 kristaps 47: #include <sqlite3.h>
1.1 kristaps 48:
1.50 kristaps 49: #include "mdoc.h"
1.1 kristaps 50: #include "man.h"
51: #include "mandoc.h"
1.119 schwarze 52: #include "mandoc_aux.h"
1.10 kristaps 53: #include "manpath.h"
1.55 kristaps 54: #include "mansearch.h"
1.1 kristaps 55:
1.111 schwarze 56: extern int mansearch_keymax;
57: extern const char *const mansearch_keynames[];
58:
1.52 kristaps 59: #define SQL_EXEC(_v) \
60: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 61: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 62: #define SQL_BIND_TEXT(_s, _i, _v) \
63: if (SQLITE_OK != sqlite3_bind_text \
64: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 65: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 66: #define SQL_BIND_INT(_s, _i, _v) \
67: if (SQLITE_OK != sqlite3_bind_int \
68: ((_s), (_i)++, (_v))) \
1.134 schwarze 69: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 70: #define SQL_BIND_INT64(_s, _i, _v) \
71: if (SQLITE_OK != sqlite3_bind_int64 \
72: ((_s), (_i)++, (_v))) \
1.134 schwarze 73: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 74: #define SQL_STEP(_s) \
75: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 76: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 77:
1.50 kristaps 78: enum op {
79: OP_DEFAULT = 0, /* new dbs from dir list or default config */
80: OP_CONFFILE, /* new databases from custom config file */
81: OP_UPDATE, /* delete/add entries in existing database */
82: OP_DELETE, /* delete entries from existing database */
83: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 84: };
85:
1.50 kristaps 86: struct str {
1.78 schwarze 87: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 88: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 89: char key[]; /* rendered text */
1.38 schwarze 90: };
91:
1.79 schwarze 92: struct inodev {
93: ino_t st_ino;
94: dev_t st_dev;
1.50 kristaps 95: };
1.5 kristaps 96:
1.78 schwarze 97: struct mpage {
1.79 schwarze 98: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 99: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 100: char *sec; /* section from file content */
101: char *arch; /* architecture from file content */
102: char *title; /* title from file content */
103: char *desc; /* description from file content */
104: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 105: int form; /* format from file content */
1.174 schwarze 106: int name_head_done;
1.82 schwarze 107: };
108:
109: struct mlink {
1.58 schwarze 110: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 111: char *dsec; /* section from directory */
112: char *arch; /* architecture from directory */
113: char *name; /* name from file name (not empty) */
114: char *fsec; /* section from file name suffix */
1.85 schwarze 115: struct mlink *next; /* singly linked list */
1.118 schwarze 116: struct mpage *mpage; /* parent */
1.161 schwarze 117: int dform; /* format from directory */
118: int fform; /* format from file name suffix */
1.124 schwarze 119: int gzip; /* filename has a .gz suffix */
1.3 kristaps 120: };
121:
1.50 kristaps 122: enum stmt {
1.81 schwarze 123: STMT_DELETE_PAGE = 0, /* delete mpage */
124: STMT_INSERT_PAGE, /* insert mpage */
125: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 126: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 127: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 128: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 129: STMT__MAX
1.1 kristaps 130: };
131:
1.172 schwarze 132: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_meta *,
133: const struct mdoc_node *);
1.1 kristaps 134:
1.50 kristaps 135: struct mdoc_handler {
136: mdoc_fp fp; /* optional handler */
137: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 138: };
139:
1.59 schwarze 140: static void dbclose(int);
1.176 schwarze 141: static void dbadd(struct mpage *);
1.118 schwarze 142: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 143: static void dbadd_mlink_name(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);
1.176 schwarze 155: static void mpages_merge(struct mparse *);
1.130 schwarze 156: static void names_check(void);
1.124 schwarze 157: static void parse_cat(struct mpage *, int);
1.174 schwarze 158: static void parse_man(struct mpage *, const struct man_meta *,
159: const struct man_node *);
1.172 schwarze 160: static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
161: const struct mdoc_node *);
162: static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
163: const struct mdoc_node *);
164: static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
165: const struct mdoc_node *);
166: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
167: const struct mdoc_node *);
1.178 schwarze 168: static void parse_mdoc_fname(struct mpage *, const struct mdoc_node *);
1.172 schwarze 169: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
170: const struct mdoc_node *);
1.173 schwarze 171: static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
172: const struct mdoc_node *);
1.172 schwarze 173: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
174: const struct mdoc_node *);
175: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
176: const struct mdoc_node *);
177: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
178: const struct mdoc_node *);
179: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
180: const struct mdoc_node *);
1.112 schwarze 181: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 182: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 183: static void putmdockey(const struct mpage *,
1.50 kristaps 184: const struct mdoc_node *, uint64_t);
1.176 schwarze 185: static int render_string(char **, size_t *);
1.59 schwarze 186: static void say(const char *, const char *, ...);
1.165 schwarze 187: static int set_basedir(const char *, int);
1.59 schwarze 188: static int treescan(void);
1.50 kristaps 189: static size_t utf8(unsigned int, char [7]);
190:
1.115 schwarze 191: static char tempfilename[32];
1.50 kristaps 192: static char *progname;
1.102 schwarze 193: static int nodb; /* no database changes */
1.116 schwarze 194: static int mparse_options; /* abort the parse early */
1.141 schwarze 195: static int use_all; /* use all found files */
196: static int debug; /* print what we're doing */
197: static int warnings; /* warn about crap */
1.95 schwarze 198: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 199: static int exitcode; /* to be returned by main */
1.141 schwarze 200: static enum op op; /* operational mode */
1.59 schwarze 201: static char basedir[PATH_MAX]; /* current base directory */
1.176 schwarze 202: static struct mchars *mchars; /* table of named characters */
1.78 schwarze 203: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 204: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 205: static struct ohash names; /* table of all names */
1.50 kristaps 206: static struct ohash strings; /* table of all strings */
207: static sqlite3 *db = NULL; /* current database */
208: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 209: static uint64_t name_mask;
1.25 schwarze 210:
211: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 212: { NULL, 0 }, /* Ap */
213: { NULL, 0 }, /* Dd */
214: { NULL, 0 }, /* Dt */
215: { NULL, 0 }, /* Os */
216: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
217: { parse_mdoc_head, TYPE_Ss }, /* Ss */
218: { NULL, 0 }, /* Pp */
219: { NULL, 0 }, /* D1 */
220: { NULL, 0 }, /* Dl */
221: { NULL, 0 }, /* Bd */
222: { NULL, 0 }, /* Ed */
223: { NULL, 0 }, /* Bl */
224: { NULL, 0 }, /* El */
225: { NULL, 0 }, /* It */
226: { NULL, 0 }, /* Ad */
227: { NULL, TYPE_An }, /* An */
228: { NULL, TYPE_Ar }, /* Ar */
229: { NULL, TYPE_Cd }, /* Cd */
230: { NULL, TYPE_Cm }, /* Cm */
231: { NULL, TYPE_Dv }, /* Dv */
232: { NULL, TYPE_Er }, /* Er */
233: { NULL, TYPE_Ev }, /* Ev */
234: { NULL, 0 }, /* Ex */
235: { NULL, TYPE_Fa }, /* Fa */
236: { parse_mdoc_Fd, 0 }, /* Fd */
237: { NULL, TYPE_Fl }, /* Fl */
238: { parse_mdoc_Fn, 0 }, /* Fn */
239: { NULL, TYPE_Ft }, /* Ft */
240: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 241: { NULL, TYPE_In }, /* In */
1.68 schwarze 242: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 243: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 244: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 245: { NULL, 0 }, /* Op */
246: { NULL, 0 }, /* Ot */
247: { NULL, TYPE_Pa }, /* Pa */
248: { NULL, 0 }, /* Rv */
1.92 schwarze 249: { NULL, TYPE_St }, /* St */
1.68 schwarze 250: { NULL, TYPE_Va }, /* Va */
251: { parse_mdoc_body, TYPE_Va }, /* Vt */
252: { parse_mdoc_Xr, 0 }, /* Xr */
253: { NULL, 0 }, /* %A */
254: { NULL, 0 }, /* %B */
255: { NULL, 0 }, /* %D */
256: { NULL, 0 }, /* %I */
257: { NULL, 0 }, /* %J */
258: { NULL, 0 }, /* %N */
259: { NULL, 0 }, /* %O */
260: { NULL, 0 }, /* %P */
261: { NULL, 0 }, /* %R */
262: { NULL, 0 }, /* %T */
263: { NULL, 0 }, /* %V */
264: { NULL, 0 }, /* Ac */
265: { NULL, 0 }, /* Ao */
266: { NULL, 0 }, /* Aq */
267: { NULL, TYPE_At }, /* At */
268: { NULL, 0 }, /* Bc */
269: { NULL, 0 }, /* Bf */
270: { NULL, 0 }, /* Bo */
271: { NULL, 0 }, /* Bq */
272: { NULL, TYPE_Bsx }, /* Bsx */
273: { NULL, TYPE_Bx }, /* Bx */
274: { NULL, 0 }, /* Db */
275: { NULL, 0 }, /* Dc */
276: { NULL, 0 }, /* Do */
277: { NULL, 0 }, /* Dq */
278: { NULL, 0 }, /* Ec */
279: { NULL, 0 }, /* Ef */
280: { NULL, TYPE_Em }, /* Em */
281: { NULL, 0 }, /* Eo */
282: { NULL, TYPE_Fx }, /* Fx */
283: { NULL, TYPE_Ms }, /* Ms */
284: { NULL, 0 }, /* No */
285: { NULL, 0 }, /* Ns */
286: { NULL, TYPE_Nx }, /* Nx */
287: { NULL, TYPE_Ox }, /* Ox */
288: { NULL, 0 }, /* Pc */
289: { NULL, 0 }, /* Pf */
290: { NULL, 0 }, /* Po */
291: { NULL, 0 }, /* Pq */
292: { NULL, 0 }, /* Qc */
293: { NULL, 0 }, /* Ql */
294: { NULL, 0 }, /* Qo */
295: { NULL, 0 }, /* Qq */
296: { NULL, 0 }, /* Re */
297: { NULL, 0 }, /* Rs */
298: { NULL, 0 }, /* Sc */
299: { NULL, 0 }, /* So */
300: { NULL, 0 }, /* Sq */
301: { NULL, 0 }, /* Sm */
302: { NULL, 0 }, /* Sx */
303: { NULL, TYPE_Sy }, /* Sy */
304: { NULL, TYPE_Tn }, /* Tn */
305: { NULL, 0 }, /* Ux */
306: { NULL, 0 }, /* Xc */
307: { NULL, 0 }, /* Xo */
1.173 schwarze 308: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 309: { NULL, 0 }, /* Fc */
310: { NULL, 0 }, /* Oo */
311: { NULL, 0 }, /* Oc */
312: { NULL, 0 }, /* Bk */
313: { NULL, 0 }, /* Ek */
314: { NULL, 0 }, /* Bt */
315: { NULL, 0 }, /* Hf */
316: { NULL, 0 }, /* Fr */
317: { NULL, 0 }, /* Ud */
318: { NULL, TYPE_Lb }, /* Lb */
319: { NULL, 0 }, /* Lp */
320: { NULL, TYPE_Lk }, /* Lk */
321: { NULL, TYPE_Mt }, /* Mt */
322: { NULL, 0 }, /* Brq */
323: { NULL, 0 }, /* Bro */
324: { NULL, 0 }, /* Brc */
325: { NULL, 0 }, /* %C */
326: { NULL, 0 }, /* Es */
327: { NULL, 0 }, /* En */
328: { NULL, TYPE_Dx }, /* Dx */
329: { NULL, 0 }, /* %Q */
330: { NULL, 0 }, /* br */
331: { NULL, 0 }, /* sp */
332: { NULL, 0 }, /* %U */
333: { NULL, 0 }, /* Ta */
1.166 schwarze 334: { NULL, 0 }, /* ll */
1.1 kristaps 335: };
336:
1.141 schwarze 337:
1.1 kristaps 338: int
1.179 schwarze 339: mandocdb(int argc, char *argv[])
1.1 kristaps 340: {
1.59 schwarze 341: int ch, i;
1.50 kristaps 342: size_t j, sz;
1.59 schwarze 343: const char *path_arg;
1.50 kristaps 344: struct manpaths dirs;
345: struct mparse *mp;
1.83 schwarze 346: struct ohash_info mpages_info, mlinks_info;
1.50 kristaps 347:
348: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
349: memset(&dirs, 0, sizeof(struct manpaths));
350:
1.83 schwarze 351: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.152 schwarze 352: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
353: mpages_info.free = mlinks_info.free = hash_free;
1.50 kristaps 354:
1.79 schwarze 355: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 356: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 357:
358: progname = strrchr(argv[0], '/');
359: if (progname == NULL)
360: progname = argv[0];
361: else
362: ++progname;
363:
1.50 kristaps 364: /*
1.141 schwarze 365: * We accept a few different invocations.
1.50 kristaps 366: * The CHECKOP macro makes sure that invocation styles don't
367: * clobber each other.
368: */
369: #define CHECKOP(_op, _ch) do \
370: if (OP_DEFAULT != (_op)) { \
1.153 schwarze 371: fprintf(stderr, "%s: -%c: Conflicting option\n", \
372: progname, (_ch)); \
1.50 kristaps 373: goto usage; \
374: } while (/*CONSTCOND*/0)
1.10 kristaps 375:
1.59 schwarze 376: path_arg = NULL;
1.38 schwarze 377: op = OP_DEFAULT;
1.1 kristaps 378:
1.126 schwarze 379: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 380: switch (ch) {
1.141 schwarze 381: case 'a':
1.12 schwarze 382: use_all = 1;
383: break;
1.141 schwarze 384: case 'C':
1.50 kristaps 385: CHECKOP(op, ch);
1.59 schwarze 386: path_arg = optarg;
1.38 schwarze 387: op = OP_CONFFILE;
1.34 schwarze 388: break;
1.141 schwarze 389: case 'D':
1.125 schwarze 390: debug++;
391: break;
1.141 schwarze 392: case 'd':
1.50 kristaps 393: CHECKOP(op, ch);
1.59 schwarze 394: path_arg = optarg;
1.5 kristaps 395: op = OP_UPDATE;
396: break;
1.141 schwarze 397: case 'n':
1.50 kristaps 398: nodb = 1;
399: break;
1.141 schwarze 400: case 'p':
1.126 schwarze 401: warnings = 1;
402: break;
1.141 schwarze 403: case 'Q':
1.116 schwarze 404: mparse_options |= MPARSE_QUICK;
1.102 schwarze 405: break;
1.141 schwarze 406: case 'T':
1.95 schwarze 407: if (strcmp(optarg, "utf8")) {
1.153 schwarze 408: fprintf(stderr, "%s: -T%s: "
409: "Unsupported output format\n",
410: progname, optarg);
1.95 schwarze 411: goto usage;
412: }
413: write_utf8 = 1;
414: break;
1.141 schwarze 415: case 't':
1.50 kristaps 416: CHECKOP(op, ch);
1.38 schwarze 417: dup2(STDOUT_FILENO, STDERR_FILENO);
418: op = OP_TEST;
1.50 kristaps 419: nodb = warnings = 1;
1.38 schwarze 420: break;
1.141 schwarze 421: case 'u':
1.50 kristaps 422: CHECKOP(op, ch);
1.59 schwarze 423: path_arg = optarg;
1.5 kristaps 424: op = OP_DELETE;
425: break;
1.141 schwarze 426: case 'v':
1.126 schwarze 427: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 428: break;
1.1 kristaps 429: default:
1.38 schwarze 430: goto usage;
1.1 kristaps 431: }
432:
433: argc -= optind;
434: argv += optind;
435:
1.38 schwarze 436: if (OP_CONFFILE == op && argc > 0) {
1.153 schwarze 437: fprintf(stderr, "%s: -C: Too many arguments\n",
438: progname);
1.38 schwarze 439: goto usage;
440: }
441:
1.59 schwarze 442: exitcode = (int)MANDOCLEVEL_OK;
1.176 schwarze 443: mchars = mchars_alloc();
1.168 schwarze 444: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
1.176 schwarze 445: mchars, NULL);
1.78 schwarze 446: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 447: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 448:
449: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 450:
1.50 kristaps 451: /*
1.150 schwarze 452: * Most of these deal with a specific directory.
1.140 schwarze 453: * Jump into that directory first.
1.50 kristaps 454: */
1.165 schwarze 455: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 456: goto out;
1.140 schwarze 457:
1.138 schwarze 458: if (dbopen(1)) {
1.140 schwarze 459: /*
460: * The existing database is usable. Process
461: * all files specified on the command-line.
462: */
463: use_all = 1;
1.138 schwarze 464: for (i = 0; i < argc; i++)
465: filescan(argv[i]);
466: if (OP_TEST != op)
467: dbprune();
468: } else {
469: /*
470: * Database missing or corrupt.
471: * Recreate from scratch.
472: */
1.140 schwarze 473: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 474: op = OP_DEFAULT;
475: if (0 == treescan())
476: goto out;
477: if (0 == dbopen(0))
478: goto out;
479: }
1.50 kristaps 480: if (OP_DELETE != op)
1.176 schwarze 481: mpages_merge(mp);
1.138 schwarze 482: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 483: } else {
484: /*
485: * If we have arguments, use them as our manpaths.
486: * If we don't, grok from manpath(1) or however else
487: * manpath_parse() wants to do it.
488: */
489: if (argc > 0) {
1.144 schwarze 490: dirs.paths = mandoc_reallocarray(NULL,
491: argc, sizeof(char *));
1.50 kristaps 492: dirs.sz = (size_t)argc;
493: for (i = 0; i < argc; i++)
494: dirs.paths[i] = mandoc_strdup(argv[i]);
495: } else
1.59 schwarze 496: manpath_parse(&dirs, path_arg, NULL, NULL);
1.127 schwarze 497:
498: if (0 == dirs.sz) {
499: exitcode = (int)MANDOCLEVEL_BADARG;
500: say("", "Empty manpath");
501: }
1.50 kristaps 502:
503: /*
1.69 schwarze 504: * First scan the tree rooted at a base directory, then
505: * build a new database and finally move it into place.
1.50 kristaps 506: * Ignore zero-length directories and strip trailing
507: * slashes.
508: */
509: for (j = 0; j < dirs.sz; j++) {
510: sz = strlen(dirs.paths[j]);
511: if (sz && '/' == dirs.paths[j][sz - 1])
512: dirs.paths[j][--sz] = '\0';
513: if (0 == sz)
514: continue;
1.66 schwarze 515:
516: if (j) {
1.78 schwarze 517: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 518: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 519: }
520:
1.165 schwarze 521: if (0 == set_basedir(dirs.paths[j], argc > 0))
522: continue;
1.59 schwarze 523: if (0 == treescan())
1.165 schwarze 524: continue;
1.59 schwarze 525: if (0 == dbopen(0))
1.165 schwarze 526: continue;
1.52 kristaps 527:
1.176 schwarze 528: mpages_merge(mp);
1.147 schwarze 529: if (warnings && !nodb &&
1.130 schwarze 530: ! (MPARSE_QUICK & mparse_options))
531: names_check();
1.59 schwarze 532: dbclose(0);
1.66 schwarze 533:
534: if (j + 1 < dirs.sz) {
1.80 schwarze 535: mpages_free();
1.78 schwarze 536: ohash_delete(&mpages);
1.83 schwarze 537: ohash_delete(&mlinks);
1.66 schwarze 538: }
1.50 kristaps 539: }
540: }
541: out:
542: manpath_free(&dirs);
1.168 schwarze 543: mparse_free(mp);
1.176 schwarze 544: mchars_free(mchars);
1.80 schwarze 545: mpages_free();
1.78 schwarze 546: ohash_delete(&mpages);
1.83 schwarze 547: ohash_delete(&mlinks);
1.59 schwarze 548: return(exitcode);
1.50 kristaps 549: usage:
1.126 schwarze 550: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
551: " %s [-aDnpQ] [-Tutf8] dir ...\n"
552: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
553: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 554: " %s [-Q] -t file ...\n",
1.141 schwarze 555: progname, progname, progname,
1.50 kristaps 556: progname, progname);
557:
1.59 schwarze 558: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 559: }
560:
561: /*
1.59 schwarze 562: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 563: * We use fts(), scanning directory parts along the way for clues to our
564: * section and architecture.
565: *
566: * If use_all has been specified, grok all files.
567: * If not, sanitise paths to the following:
568: *
1.141 schwarze 569: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 570: * or
571: * [./]cat<section>[/<arch>]/<name>.0
572: *
573: * TODO: accomodate for multi-language directories.
574: */
575: static int
1.59 schwarze 576: treescan(void)
1.50 kristaps 577: {
1.139 schwarze 578: char buf[PATH_MAX];
1.50 kristaps 579: FTS *f;
580: FTSENT *ff;
1.84 schwarze 581: struct mlink *mlink;
1.124 schwarze 582: int dform, gzip;
1.94 schwarze 583: char *dsec, *arch, *fsec, *cp;
584: const char *path;
1.50 kristaps 585: const char *argv[2];
586:
587: argv[0] = ".";
588: argv[1] = (char *)NULL;
589:
1.139 schwarze 590: f = fts_open((char * const *)argv,
591: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.50 kristaps 592: if (NULL == f) {
1.59 schwarze 593: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 594: say("", "&fts_open");
1.50 kristaps 595: return(0);
596: }
597:
598: dsec = arch = NULL;
599: dform = FORM_NONE;
600:
601: while (NULL != (ff = fts_read(f))) {
602: path = ff->fts_path + 2;
1.139 schwarze 603: switch (ff->fts_info) {
604:
605: /*
606: * Symbolic links require various sanity checks,
607: * then get handled just like regular files.
608: */
1.141 schwarze 609: case FTS_SL:
1.139 schwarze 610: if (NULL == realpath(path, buf)) {
611: if (warnings)
612: say(path, "&realpath");
613: continue;
614: }
615: if (strstr(buf, basedir) != buf) {
616: if (warnings) say("",
617: "%s: outside base directory", buf);
618: continue;
619: }
620: /* Use logical inode to avoid mpages dupe. */
621: if (-1 == stat(path, ff->fts_statp)) {
622: if (warnings)
623: say(path, "&stat");
624: continue;
625: }
626: /* FALLTHROUGH */
627:
1.50 kristaps 628: /*
1.83 schwarze 629: * If we're a regular file, add an mlink by using the
1.50 kristaps 630: * stored directory data and handling the filename.
631: */
1.141 schwarze 632: case FTS_F:
1.60 schwarze 633: if (0 == strcmp(path, MANDOC_DB))
634: continue;
1.50 kristaps 635: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 636: if (warnings)
1.59 schwarze 637: say(path, "Extraneous file");
1.50 kristaps 638: continue;
1.124 schwarze 639: }
640: gzip = 0;
641: fsec = NULL;
642: while (NULL == fsec) {
643: fsec = strrchr(ff->fts_name, '.');
644: if (NULL == fsec || strcmp(fsec+1, "gz"))
645: break;
646: gzip = 1;
647: *fsec = '\0';
648: fsec = NULL;
649: }
650: if (NULL == fsec) {
1.60 schwarze 651: if ( ! use_all) {
1.56 schwarze 652: if (warnings)
1.60 schwarze 653: say(path,
654: "No filename suffix");
1.50 kristaps 655: continue;
656: }
1.84 schwarze 657: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 658: if (warnings)
659: say(path, "Skip html");
660: continue;
1.84 schwarze 661: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 662: if (warnings)
663: say(path, "Skip ps");
664: continue;
1.84 schwarze 665: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 666: if (warnings)
667: say(path, "Skip pdf");
668: continue;
669: } else if ( ! use_all &&
1.84 schwarze 670: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
671: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 672: if (warnings)
673: say(path, "Wrong filename suffix");
674: continue;
1.66 schwarze 675: } else
1.84 schwarze 676: fsec[-1] = '\0';
1.94 schwarze 677:
1.84 schwarze 678: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 679: if (strlcpy(mlink->file, path,
680: sizeof(mlink->file)) >=
681: sizeof(mlink->file)) {
682: say(path, "Filename too long");
683: free(mlink);
684: continue;
685: }
1.84 schwarze 686: mlink->dform = dform;
1.94 schwarze 687: mlink->dsec = dsec;
688: mlink->arch = arch;
689: mlink->name = ff->fts_name;
690: mlink->fsec = fsec;
1.124 schwarze 691: mlink->gzip = gzip;
1.84 schwarze 692: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 693: continue;
1.139 schwarze 694:
1.141 schwarze 695: case FTS_D:
1.139 schwarze 696: /* FALLTHROUGH */
1.141 schwarze 697: case FTS_DP:
1.139 schwarze 698: break;
699:
700: default:
1.60 schwarze 701: if (warnings)
702: say(path, "Not a regular file");
1.50 kristaps 703: continue;
1.60 schwarze 704: }
1.1 kristaps 705:
1.50 kristaps 706: switch (ff->fts_level) {
1.141 schwarze 707: case 0:
1.50 kristaps 708: /* Ignore the root directory. */
709: break;
1.141 schwarze 710: case 1:
1.50 kristaps 711: /*
712: * This might contain manX/ or catX/.
713: * Try to infer this from the name.
714: * If we're not in use_all, enforce it.
715: */
716: cp = ff->fts_name;
717: if (FTS_DP == ff->fts_info)
718: break;
1.1 kristaps 719:
1.50 kristaps 720: if (0 == strncmp(cp, "man", 3)) {
721: dform = FORM_SRC;
1.66 schwarze 722: dsec = cp + 3;
1.50 kristaps 723: } else if (0 == strncmp(cp, "cat", 3)) {
724: dform = FORM_CAT;
1.66 schwarze 725: dsec = cp + 3;
1.94 schwarze 726: } else {
727: dform = FORM_NONE;
728: dsec = NULL;
1.50 kristaps 729: }
1.1 kristaps 730:
1.141 schwarze 731: if (NULL != dsec || use_all)
1.50 kristaps 732: break;
1.1 kristaps 733:
1.56 schwarze 734: if (warnings)
1.59 schwarze 735: say(path, "Unknown directory part");
1.50 kristaps 736: fts_set(f, ff, FTS_SKIP);
737: break;
1.141 schwarze 738: case 2:
1.50 kristaps 739: /*
740: * Possibly our architecture.
741: * If we're descending, keep tabs on it.
742: */
743: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 744: arch = ff->fts_name;
1.94 schwarze 745: else
746: arch = NULL;
1.50 kristaps 747: break;
748: default:
749: if (FTS_DP == ff->fts_info || use_all)
750: break;
1.56 schwarze 751: if (warnings)
1.59 schwarze 752: say(path, "Extraneous directory part");
1.50 kristaps 753: fts_set(f, ff, FTS_SKIP);
754: break;
1.5 kristaps 755: }
1.50 kristaps 756: }
757:
758: fts_close(f);
759: return(1);
760: }
1.5 kristaps 761:
1.50 kristaps 762: /*
1.89 schwarze 763: * Add a file to the mlinks table.
1.50 kristaps 764: * Do not verify that it's a "valid" looking manpage (we'll do that
765: * later).
766: *
767: * Try to infer the manual section, architecture, and page name from the
768: * path, assuming it looks like
769: *
1.141 schwarze 770: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 771: * or
772: * [./]cat<section>[/<arch>]/<name>.0
773: *
774: * See treescan() for the fts(3) version of this.
775: */
776: static void
1.59 schwarze 777: filescan(const char *file)
1.50 kristaps 778: {
1.59 schwarze 779: char buf[PATH_MAX];
1.84 schwarze 780: struct stat st;
781: struct mlink *mlink;
1.59 schwarze 782: char *p, *start;
1.5 kristaps 783:
1.50 kristaps 784: assert(use_all);
1.5 kristaps 785:
1.50 kristaps 786: if (0 == strncmp(file, "./", 2))
787: file += 2;
1.5 kristaps 788:
1.139 schwarze 789: /*
790: * We have to do lstat(2) before realpath(3) loses
791: * the information whether this is a symbolic link.
792: * We need to know that because for symbolic links,
793: * we want to use the orginal file name, while for
794: * regular files, we want to use the real path.
795: */
796: if (-1 == lstat(file, &st)) {
797: exitcode = (int)MANDOCLEVEL_BADARG;
798: say(file, "&lstat");
799: return;
800: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
801: exitcode = (int)MANDOCLEVEL_BADARG;
802: say(file, "Not a regular file");
803: return;
804: }
805:
806: /*
807: * We have to resolve the file name to the real path
808: * in any case for the base directory check.
809: */
1.59 schwarze 810: if (NULL == realpath(file, buf)) {
811: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 812: say(file, "&realpath");
1.59 schwarze 813: return;
1.106 schwarze 814: }
815:
1.150 schwarze 816: if (OP_TEST == op)
1.106 schwarze 817: start = buf;
1.150 schwarze 818: else if (strstr(buf, basedir) == buf)
819: start = buf + strlen(basedir);
1.106 schwarze 820: else {
1.59 schwarze 821: exitcode = (int)MANDOCLEVEL_BADARG;
822: say("", "%s: outside base directory", buf);
823: return;
1.106 schwarze 824: }
825:
1.139 schwarze 826: /*
827: * Now we are sure the file is inside our tree.
828: * If it is a symbolic link, ignore the real path
829: * and use the original name.
830: * This implies passing stuff like "cat1/../man1/foo.1"
831: * on the command line won't work. So don't do that.
832: * Note the stat(2) can still fail if the link target
833: * doesn't exist.
834: */
835: if (S_IFLNK & st.st_mode) {
836: if (-1 == stat(buf, &st)) {
837: exitcode = (int)MANDOCLEVEL_BADARG;
838: say(file, "&stat");
839: return;
840: }
1.143 schwarze 841: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
842: say(file, "Filename too long");
843: return;
844: }
1.150 schwarze 845: start = buf;
846: if (OP_TEST != op && strstr(buf, basedir) == buf)
847: start += strlen(basedir);
1.50 kristaps 848: }
1.106 schwarze 849:
1.84 schwarze 850: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 851: mlink->dform = FORM_NONE;
1.143 schwarze 852: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
853: sizeof(mlink->file)) {
854: say(start, "Filename too long");
1.180 ! schwarze 855: free(mlink);
1.143 schwarze 856: return;
857: }
1.17 schwarze 858:
1.50 kristaps 859: /*
860: * First try to guess our directory structure.
861: * If we find a separator, try to look for man* or cat*.
862: * If we find one of these and what's underneath is a directory,
863: * assume it's an architecture.
864: */
865: if (NULL != (p = strchr(start, '/'))) {
866: *p++ = '\0';
867: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 868: mlink->dform = FORM_SRC;
1.94 schwarze 869: mlink->dsec = start + 3;
1.50 kristaps 870: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 871: mlink->dform = FORM_CAT;
1.94 schwarze 872: mlink->dsec = start + 3;
1.17 schwarze 873: }
1.5 kristaps 874:
1.50 kristaps 875: start = p;
1.84 schwarze 876: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 877: *p++ = '\0';
1.94 schwarze 878: mlink->arch = start;
1.50 kristaps 879: start = p;
1.84 schwarze 880: }
1.50 kristaps 881: }
882:
883: /*
884: * Now check the file suffix.
885: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
886: */
887: p = strrchr(start, '\0');
888: while (p-- > start && '/' != *p && '.' != *p)
889: /* Loop. */ ;
890:
891: if ('.' == *p) {
892: *p++ = '\0';
1.94 schwarze 893: mlink->fsec = p;
1.5 kristaps 894: }
895:
1.10 kristaps 896: /*
1.50 kristaps 897: * Now try to parse the name.
898: * Use the filename portion of the path.
1.10 kristaps 899: */
1.84 schwarze 900: mlink->name = start;
1.50 kristaps 901: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 902: mlink->name = p + 1;
1.50 kristaps 903: *p = '\0';
1.84 schwarze 904: }
905: mlink_add(mlink, &st);
1.50 kristaps 906: }
1.5 kristaps 907:
1.50 kristaps 908: static void
1.84 schwarze 909: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 910: {
1.83 schwarze 911: struct inodev inodev;
1.78 schwarze 912: struct mpage *mpage;
1.83 schwarze 913: unsigned int slot;
1.50 kristaps 914:
1.84 schwarze 915: assert(NULL != mlink->file);
1.50 kristaps 916:
1.94 schwarze 917: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
918: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
919: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
920: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 921:
922: if ('0' == *mlink->fsec) {
923: free(mlink->fsec);
924: mlink->fsec = mandoc_strdup(mlink->dsec);
925: mlink->fform = FORM_CAT;
926: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
927: mlink->fform = FORM_SRC;
1.74 schwarze 928: else
1.84 schwarze 929: mlink->fform = FORM_NONE;
1.82 schwarze 930:
1.83 schwarze 931: slot = ohash_qlookup(&mlinks, mlink->file);
932: assert(NULL == ohash_find(&mlinks, slot));
933: ohash_insert(&mlinks, slot, mlink);
934:
935: inodev.st_ino = st->st_ino;
936: inodev.st_dev = st->st_dev;
937: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
938: sizeof(struct inodev), inodev.st_ino);
939: mpage = ohash_find(&mpages, slot);
940: if (NULL == mpage) {
941: mpage = mandoc_calloc(1, sizeof(struct mpage));
942: mpage->inodev.st_ino = inodev.st_ino;
943: mpage->inodev.st_dev = inodev.st_dev;
944: ohash_insert(&mpages, slot, mpage);
945: } else
1.85 schwarze 946: mlink->next = mpage->mlinks;
1.82 schwarze 947: mpage->mlinks = mlink;
1.118 schwarze 948: mlink->mpage = mpage;
1.82 schwarze 949: }
950:
951: static void
952: mlink_free(struct mlink *mlink)
953: {
954:
955: free(mlink->dsec);
956: free(mlink->arch);
957: free(mlink->name);
958: free(mlink->fsec);
959: free(mlink);
1.50 kristaps 960: }
1.3 kristaps 961:
1.50 kristaps 962: static void
1.78 schwarze 963: mpages_free(void)
1.50 kristaps 964: {
1.78 schwarze 965: struct mpage *mpage;
1.82 schwarze 966: struct mlink *mlink;
1.80 schwarze 967: unsigned int slot;
1.3 kristaps 968:
1.80 schwarze 969: mpage = ohash_first(&mpages, &slot);
970: while (NULL != mpage) {
1.82 schwarze 971: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 972: mpage->mlinks = mlink->next;
1.82 schwarze 973: mlink_free(mlink);
974: }
1.78 schwarze 975: free(mpage->sec);
976: free(mpage->arch);
1.82 schwarze 977: free(mpage->title);
978: free(mpage->desc);
1.78 schwarze 979: free(mpage);
1.80 schwarze 980: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 981: }
982: }
1.38 schwarze 983:
1.50 kristaps 984: /*
1.89 schwarze 985: * For each mlink to the mpage, check whether the path looks like
986: * it is formatted, and if it does, check whether a source manual
987: * exists by the same name, ignoring the suffix.
988: * If both conditions hold, drop the mlink.
989: */
990: static void
991: mlinks_undupe(struct mpage *mpage)
992: {
993: char buf[PATH_MAX];
994: struct mlink **prev;
995: struct mlink *mlink;
996: char *bufp;
997:
998: mpage->form = FORM_CAT;
1.90 schwarze 999: prev = &mpage->mlinks;
1000: while (NULL != (mlink = *prev)) {
1.89 schwarze 1001: if (FORM_CAT != mlink->dform) {
1002: mpage->form = FORM_NONE;
1.90 schwarze 1003: goto nextlink;
1.89 schwarze 1004: }
1.143 schwarze 1005: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1006: bufp = strstr(buf, "cat");
1007: assert(NULL != bufp);
1008: memcpy(bufp, "man", 3);
1009: if (NULL != (bufp = strrchr(buf, '.')))
1010: *++bufp = '\0';
1.143 schwarze 1011: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1012: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1013: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1014: goto nextlink;
1.89 schwarze 1015: if (warnings)
1016: say(mlink->file, "Man source exists: %s", buf);
1017: if (use_all)
1.90 schwarze 1018: goto nextlink;
1.89 schwarze 1019: *prev = mlink->next;
1020: mlink_free(mlink);
1.90 schwarze 1021: continue;
1022: nextlink:
1023: prev = &(*prev)->next;
1.89 schwarze 1024: }
1025: }
1026:
1.131 schwarze 1027: static void
1.93 schwarze 1028: mlink_check(struct mpage *mpage, struct mlink *mlink)
1029: {
1.131 schwarze 1030: struct str *str;
1031: unsigned int slot;
1.93 schwarze 1032:
1033: /*
1034: * Check whether the manual section given in a file
1035: * agrees with the directory where the file is located.
1036: * Some manuals have suffixes like (3p) on their
1037: * section number either inside the file or in the
1038: * directory name, some are linked into more than one
1039: * section, like encrypt(1) = makekey(8).
1040: */
1041:
1042: if (FORM_SRC == mpage->form &&
1.131 schwarze 1043: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1044: say(mlink->file, "Section \"%s\" manual in %s directory",
1045: mpage->sec, mlink->dsec);
1046:
1047: /*
1048: * Manual page directories exist for each kernel
1049: * architecture as returned by machine(1).
1050: * However, many manuals only depend on the
1051: * application architecture as returned by arch(1).
1052: * For example, some (2/ARM) manuals are shared
1053: * across the "armish" and "zaurus" kernel
1054: * architectures.
1055: * A few manuals are even shared across completely
1056: * different architectures, for example fdformat(1)
1057: * on amd64, i386, sparc, and sparc64.
1058: */
1059:
1.131 schwarze 1060: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1061: say(mlink->file, "Architecture \"%s\" manual in "
1062: "\"%s\" directory", mpage->arch, mlink->arch);
1063:
1.131 schwarze 1064: /*
1065: * XXX
1.133 schwarze 1066: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1067: */
1068:
1069: if (FORM_CAT == mpage->form)
1070: return;
1071:
1072: /*
1073: * Check whether this mlink
1074: * appears as a name in the NAME section.
1075: */
1.93 schwarze 1076:
1.133 schwarze 1077: slot = ohash_qlookup(&names, mlink->name);
1078: str = ohash_find(&names, slot);
1.131 schwarze 1079: assert(NULL != str);
1.133 schwarze 1080: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1081: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1082: }
1083:
1.89 schwarze 1084: /*
1.80 schwarze 1085: * Run through the files in the global vector "mpages"
1086: * and add them to the database specified in "basedir".
1.50 kristaps 1087: *
1088: * This handles the parsing scheme itself, using the cues of directory
1089: * and filename to determine whether the file is parsable or not.
1090: */
1.59 schwarze 1091: static void
1.176 schwarze 1092: mpages_merge(struct mparse *mp)
1.50 kristaps 1093: {
1.113 schwarze 1094: char any[] = "any";
1.101 schwarze 1095: struct ohash_info str_info;
1.118 schwarze 1096: struct mpage *mpage, *mpage_dest;
1097: struct mlink *mlink, *mlink_dest;
1.69 schwarze 1098: struct mdoc *mdoc;
1099: struct man *man;
1.118 schwarze 1100: char *sodest;
1.112 schwarze 1101: char *cp;
1.162 schwarze 1102: int fd;
1.101 schwarze 1103: unsigned int pslot;
1.69 schwarze 1104: enum mandoclevel lvl;
1105:
1.71 schwarze 1106: str_info.alloc = hash_alloc;
1.152 schwarze 1107: str_info.calloc = hash_calloc;
1108: str_info.free = hash_free;
1.71 schwarze 1109: str_info.key_offset = offsetof(struct str, key);
1110:
1.171 schwarze 1111: if ( ! nodb)
1.107 schwarze 1112: SQL_EXEC("BEGIN TRANSACTION");
1113:
1.80 schwarze 1114: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1115: while (mpage != NULL) {
1.89 schwarze 1116: mlinks_undupe(mpage);
1.171 schwarze 1117: if (mpage->mlinks == NULL) {
1.89 schwarze 1118: mpage = ohash_next(&mpages, &pslot);
1119: continue;
1.50 kristaps 1120: }
1121:
1.133 schwarze 1122: name_mask = NAME_MASK;
1123: ohash_init(&names, 4, &str_info);
1.71 schwarze 1124: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1125: mparse_reset(mp);
1126: mdoc = NULL;
1127: man = NULL;
1.124 schwarze 1128: sodest = NULL;
1129:
1.170 schwarze 1130: mparse_open(mp, &fd, mpage->mlinks->file);
1.162 schwarze 1131: if (fd == -1) {
1132: say(mpage->mlinks->file, "&open");
1133: goto nextpage;
1.124 schwarze 1134: }
1.14 schwarze 1135:
1136: /*
1.33 schwarze 1137: * Try interpreting the file as mdoc(7) or man(7)
1138: * source code, unless it is already known to be
1139: * formatted. Fall back to formatted mode.
1.14 schwarze 1140: */
1.171 schwarze 1141: if (mpage->mlinks->dform != FORM_CAT ||
1142: mpage->mlinks->fform != FORM_CAT) {
1.162 schwarze 1143: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.50 kristaps 1144: if (lvl < MANDOCLEVEL_FATAL)
1.118 schwarze 1145: mparse_result(mp, &mdoc, &man, &sodest);
1.76 schwarze 1146: }
1.14 schwarze 1147:
1.171 schwarze 1148: if (sodest != NULL) {
1.118 schwarze 1149: mlink_dest = ohash_find(&mlinks,
1150: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1151: if (mlink_dest == NULL) {
1152: mandoc_asprintf(&cp, "%s.gz", sodest);
1153: mlink_dest = ohash_find(&mlinks,
1154: ohash_qlookup(&mlinks, cp));
1155: free(cp);
1156: }
1157: if (mlink_dest != NULL) {
1.118 schwarze 1158:
1159: /* The .so target exists. */
1160:
1161: mpage_dest = mlink_dest->mpage;
1162: mlink = mpage->mlinks;
1163: while (1) {
1164: mlink->mpage = mpage_dest;
1165:
1166: /*
1167: * If the target was already
1168: * processed, add the links
1169: * to the database now.
1170: * Otherwise, this will
1171: * happen when we come
1172: * to the target.
1173: */
1174:
1.137 schwarze 1175: if (mpage_dest->pageid)
1.169 schwarze 1176: dbadd_mlink_name(mlink);
1.118 schwarze 1177:
1.171 schwarze 1178: if (mlink->next == NULL)
1.118 schwarze 1179: break;
1180: mlink = mlink->next;
1181: }
1182:
1183: /* Move all links to the target. */
1184:
1185: mlink->next = mlink_dest->next;
1186: mlink_dest->next = mpage->mlinks;
1187: mpage->mlinks = NULL;
1188: }
1.124 schwarze 1189: goto nextpage;
1.171 schwarze 1190: } else if (mdoc != NULL) {
1.82 schwarze 1191: mpage->form = FORM_SRC;
1.155 schwarze 1192: mpage->sec = mdoc_meta(mdoc)->msec;
1193: mpage->sec = mandoc_strdup(
1.171 schwarze 1194: mpage->sec == NULL ? "" : mpage->sec);
1.82 schwarze 1195: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 1196: mpage->arch = mandoc_strdup(
1.171 schwarze 1197: mpage->arch == NULL ? "" : mpage->arch);
1.82 schwarze 1198: mpage->title =
1199: mandoc_strdup(mdoc_meta(mdoc)->title);
1.171 schwarze 1200: } else if (man != NULL) {
1.82 schwarze 1201: mpage->form = FORM_SRC;
1202: mpage->sec =
1203: mandoc_strdup(man_meta(man)->msec);
1204: mpage->arch =
1205: mandoc_strdup(mpage->mlinks->arch);
1206: mpage->title =
1207: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1208: } else {
1.82 schwarze 1209: mpage->form = FORM_CAT;
1210: mpage->sec =
1211: mandoc_strdup(mpage->mlinks->dsec);
1212: mpage->arch =
1213: mandoc_strdup(mpage->mlinks->arch);
1214: mpage->title =
1215: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1216: }
1.97 schwarze 1217: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1218: if (*mpage->arch != '\0')
1219: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1220:
1.97 schwarze 1221: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1222: if ('\0' != *mlink->dsec)
1223: putkey(mpage, mlink->dsec, TYPE_sec);
1224: if ('\0' != *mlink->fsec)
1225: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1226: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1227: any : mlink->arch, TYPE_arch);
1.133 schwarze 1228: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1229: }
1.12 schwarze 1230:
1.174 schwarze 1231: assert(mpage->desc == NULL);
1232: if (mdoc != NULL)
1.172 schwarze 1233: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.174 schwarze 1234: else if (man != NULL)
1235: parse_man(mpage, man_meta(man), man_node(man));
1.50 kristaps 1236: else
1.162 schwarze 1237: parse_cat(mpage, fd);
1.174 schwarze 1238: if (mpage->desc == NULL)
1.135 schwarze 1239: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1240:
1241: if (warnings && !use_all)
1242: for (mlink = mpage->mlinks; mlink;
1243: mlink = mlink->next)
1244: mlink_check(mpage, mlink);
1.44 kristaps 1245:
1.176 schwarze 1246: dbadd(mpage);
1.124 schwarze 1247:
1248: nextpage:
1.170 schwarze 1249: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.162 schwarze 1250: exitcode = (int)MANDOCLEVEL_SYSERR;
1251: say(mpage->mlinks->file, "&wait gunzip");
1.124 schwarze 1252: }
1.66 schwarze 1253: ohash_delete(&strings);
1.133 schwarze 1254: ohash_delete(&names);
1.80 schwarze 1255: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1256: }
1.107 schwarze 1257:
1258: if (0 == nodb)
1259: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1260: }
1261:
1262: static void
1263: names_check(void)
1264: {
1265: sqlite3_stmt *stmt;
1266: const char *name, *sec, *arch, *key;
1267: int irc;
1268:
1269: sqlite3_prepare_v2(db,
1270: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1271: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1272: "WHERE bits & ? AND NOT EXISTS ("
1273: "SELECT pageid FROM mlinks "
1.133 schwarze 1274: "WHERE mlinks.pageid == names.pageid "
1275: "AND mlinks.name == names.name"
1.130 schwarze 1276: ")"
1277: ") JOIN ("
1.142 schwarze 1278: "SELECT sec, arch, name, pageid FROM mlinks "
1279: "GROUP BY pageid"
1.130 schwarze 1280: ") USING (pageid);",
1281: -1, &stmt, NULL);
1282:
1.134 schwarze 1283: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1284: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1285:
1286: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.154 schwarze 1287: name = (const char *)sqlite3_column_text(stmt, 0);
1288: sec = (const char *)sqlite3_column_text(stmt, 1);
1289: arch = (const char *)sqlite3_column_text(stmt, 2);
1290: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1291: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1292: '\0' == *arch ? "" : "/",
1293: '\0' == *arch ? "" : arch, key);
1294: }
1295: sqlite3_finalize(stmt);
1.50 kristaps 1296: }
1.12 schwarze 1297:
1.50 kristaps 1298: static void
1.124 schwarze 1299: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1300: {
1301: FILE *stream;
1302: char *line, *p, *title;
1303: size_t len, plen, titlesz;
1.12 schwarze 1304:
1.124 schwarze 1305: stream = (-1 == fd) ?
1306: fopen(mpage->mlinks->file, "r") :
1307: fdopen(fd, "r");
1308: if (NULL == stream) {
1.159 schwarze 1309: if (-1 != fd)
1310: close(fd);
1.56 schwarze 1311: if (warnings)
1.123 schwarze 1312: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1313: return;
1314: }
1.1 kristaps 1315:
1.50 kristaps 1316: /* Skip to first blank line. */
1.1 kristaps 1317:
1.50 kristaps 1318: while (NULL != (line = fgetln(stream, &len)))
1319: if ('\n' == *line)
1320: break;
1.1 kristaps 1321:
1.50 kristaps 1322: /*
1323: * Assume the first line that is not indented
1324: * is the first section header. Skip to it.
1325: */
1.1 kristaps 1326:
1.50 kristaps 1327: while (NULL != (line = fgetln(stream, &len)))
1328: if ('\n' != *line && ' ' != *line)
1329: break;
1.141 schwarze 1330:
1.50 kristaps 1331: /*
1332: * Read up until the next section into a buffer.
1333: * Strip the leading and trailing newline from each read line,
1334: * appending a trailing space.
1335: * Ignore empty (whitespace-only) lines.
1336: */
1.1 kristaps 1337:
1.50 kristaps 1338: titlesz = 0;
1339: title = NULL;
1.38 schwarze 1340:
1.50 kristaps 1341: while (NULL != (line = fgetln(stream, &len))) {
1342: if (' ' != *line || '\n' != line[len - 1])
1343: break;
1344: while (len > 0 && isspace((unsigned char)*line)) {
1345: line++;
1346: len--;
1347: }
1348: if (1 == len)
1.38 schwarze 1349: continue;
1.50 kristaps 1350: title = mandoc_realloc(title, titlesz + len);
1351: memcpy(title + titlesz, line, len);
1352: titlesz += len;
1353: title[titlesz - 1] = ' ';
1354: }
1.38 schwarze 1355:
1.50 kristaps 1356: /*
1357: * If no page content can be found, or the input line
1358: * is already the next section header, or there is no
1359: * trailing newline, reuse the page title as the page
1360: * description.
1361: */
1.44 kristaps 1362:
1.50 kristaps 1363: if (NULL == title || '\0' == *title) {
1.56 schwarze 1364: if (warnings)
1.82 schwarze 1365: say(mpage->mlinks->file,
1366: "Cannot find NAME section");
1.50 kristaps 1367: fclose(stream);
1368: free(title);
1369: return;
1370: }
1.1 kristaps 1371:
1.50 kristaps 1372: title = mandoc_realloc(title, titlesz + 1);
1373: title[titlesz] = '\0';
1.33 schwarze 1374:
1.50 kristaps 1375: /*
1376: * Skip to the first dash.
1377: * Use the remaining line as the description (no more than 70
1378: * bytes).
1379: */
1.33 schwarze 1380:
1.50 kristaps 1381: if (NULL != (p = strstr(title, "- "))) {
1382: for (p += 2; ' ' == *p || '\b' == *p; p++)
1383: /* Skip to next word. */ ;
1384: } else {
1.56 schwarze 1385: if (warnings)
1.82 schwarze 1386: say(mpage->mlinks->file,
1387: "No dash in title line");
1.50 kristaps 1388: p = title;
1389: }
1.38 schwarze 1390:
1.50 kristaps 1391: plen = strlen(p);
1.1 kristaps 1392:
1.50 kristaps 1393: /* Strip backspace-encoding from line. */
1.1 kristaps 1394:
1.50 kristaps 1395: while (NULL != (line = memchr(p, '\b', plen))) {
1396: len = line - p;
1397: if (0 == len) {
1398: memmove(line, line + 1, plen--);
1399: continue;
1.141 schwarze 1400: }
1.50 kristaps 1401: memmove(line - 1, line + 1, plen - len);
1402: plen -= 2;
1403: }
1.1 kristaps 1404:
1.78 schwarze 1405: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1406: fclose(stream);
1407: free(title);
1408: }
1.1 kristaps 1409:
1.50 kristaps 1410: /*
1411: * Put a type/word pair into the word database for this particular file.
1412: */
1413: static void
1.112 schwarze 1414: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1415: {
1.112 schwarze 1416: char *cp;
1.18 kristaps 1417:
1.50 kristaps 1418: assert(NULL != value);
1.112 schwarze 1419: if (TYPE_arch == type)
1420: for (cp = value; *cp; cp++)
1421: if (isupper((unsigned char)*cp))
1422: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1423: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1424: }
1425:
1426: /*
1.50 kristaps 1427: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1428: */
1429: static void
1.78 schwarze 1430: putmdockey(const struct mpage *mpage,
1431: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1432: {
1.18 kristaps 1433:
1.50 kristaps 1434: for ( ; NULL != n; n = n->next) {
1435: if (NULL != n->child)
1.78 schwarze 1436: putmdockey(mpage, n->child, m);
1.50 kristaps 1437: if (MDOC_TEXT == n->type)
1.78 schwarze 1438: putkey(mpage, n->string, m);
1.50 kristaps 1439: }
1440: }
1.18 kristaps 1441:
1.61 schwarze 1442: static void
1.174 schwarze 1443: parse_man(struct mpage *mpage, const struct man_meta *meta,
1444: const struct man_node *n)
1.50 kristaps 1445: {
1446: const struct man_node *head, *body;
1.121 schwarze 1447: char *start, *title;
1.50 kristaps 1448: char byte;
1.121 schwarze 1449: size_t sz;
1.18 kristaps 1450:
1.50 kristaps 1451: if (NULL == n)
1.61 schwarze 1452: return;
1.18 kristaps 1453:
1.50 kristaps 1454: /*
1455: * We're only searching for one thing: the first text child in
1456: * the BODY of a NAME section. Since we don't keep track of
1457: * sections in -man, run some hoops to find out whether we're in
1458: * the correct section or not.
1459: */
1.18 kristaps 1460:
1.50 kristaps 1461: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1462: body = n;
1463: assert(body->parent);
1464: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1465: 1 == head->nchild &&
1466: NULL != (head = (head->child)) &&
1467: MAN_TEXT == head->type &&
1468: 0 == strcmp(head->string, "NAME") &&
1469: NULL != body->child) {
1.3 kristaps 1470:
1.50 kristaps 1471: /*
1472: * Suck the entire NAME section into memory.
1473: * Yes, we might run away.
1474: * But too many manuals have big, spread-out
1475: * NAME sections over many lines.
1476: */
1.3 kristaps 1477:
1.121 schwarze 1478: title = NULL;
1479: man_deroff(&title, body);
1.50 kristaps 1480: if (NULL == title)
1.61 schwarze 1481: return;
1.18 kristaps 1482:
1.141 schwarze 1483: /*
1.50 kristaps 1484: * Go through a special heuristic dance here.
1485: * Conventionally, one or more manual names are
1486: * comma-specified prior to a whitespace, then a
1487: * dash, then a description. Try to puzzle out
1488: * the name parts here.
1489: */
1.18 kristaps 1490:
1.121 schwarze 1491: start = title;
1.50 kristaps 1492: for ( ;; ) {
1493: sz = strcspn(start, " ,");
1494: if ('\0' == start[sz])
1495: break;
1.1 kristaps 1496:
1.50 kristaps 1497: byte = start[sz];
1498: start[sz] = '\0';
1.110 schwarze 1499:
1500: /*
1501: * Assume a stray trailing comma in the
1502: * name list if a name begins with a dash.
1503: */
1504:
1505: if ('-' == start[0] ||
1506: ('\\' == start[0] && '-' == start[1]))
1507: break;
1.1 kristaps 1508:
1.133 schwarze 1509: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1510: if ( ! (mpage->name_head_done ||
1511: strcasecmp(start, meta->title))) {
1512: putkey(mpage, start, NAME_HEAD);
1513: mpage->name_head_done = 1;
1514: }
1.1 kristaps 1515:
1.50 kristaps 1516: if (' ' == byte) {
1517: start += sz + 1;
1518: break;
1519: }
1.1 kristaps 1520:
1.50 kristaps 1521: assert(',' == byte);
1522: start += sz + 1;
1523: while (' ' == *start)
1524: start++;
1525: }
1.1 kristaps 1526:
1.121 schwarze 1527: if (start == title) {
1.133 schwarze 1528: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1529: if ( ! (mpage->name_head_done ||
1530: strcasecmp(start, meta->title))) {
1531: putkey(mpage, start, NAME_HEAD);
1532: mpage->name_head_done = 1;
1533: }
1.50 kristaps 1534: free(title);
1.61 schwarze 1535: return;
1.50 kristaps 1536: }
1.1 kristaps 1537:
1.50 kristaps 1538: while (isspace((unsigned char)*start))
1539: start++;
1.1 kristaps 1540:
1.50 kristaps 1541: if (0 == strncmp(start, "-", 1))
1542: start += 1;
1543: else if (0 == strncmp(start, "\\-\\-", 4))
1544: start += 4;
1545: else if (0 == strncmp(start, "\\-", 2))
1546: start += 2;
1547: else if (0 == strncmp(start, "\\(en", 4))
1548: start += 4;
1549: else if (0 == strncmp(start, "\\(em", 4))
1550: start += 4;
1.1 kristaps 1551:
1.50 kristaps 1552: while (' ' == *start)
1553: start++;
1.1 kristaps 1554:
1.78 schwarze 1555: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1556: free(title);
1.61 schwarze 1557: return;
1.50 kristaps 1558: }
1559: }
1.1 kristaps 1560:
1.77 schwarze 1561: for (n = n->child; n; n = n->next) {
1.78 schwarze 1562: if (NULL != mpage->desc)
1.77 schwarze 1563: break;
1.174 schwarze 1564: parse_man(mpage, meta, n);
1.77 schwarze 1565: }
1.1 kristaps 1566: }
1567:
1568: static void
1.172 schwarze 1569: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1570: const struct mdoc_node *n)
1.1 kristaps 1571: {
1572:
1.50 kristaps 1573: assert(NULL != n);
1574: for (n = n->child; NULL != n; n = n->next) {
1575: switch (n->type) {
1.141 schwarze 1576: case MDOC_ELEM:
1.50 kristaps 1577: /* FALLTHROUGH */
1.141 schwarze 1578: case MDOC_BLOCK:
1.50 kristaps 1579: /* FALLTHROUGH */
1.141 schwarze 1580: case MDOC_HEAD:
1.50 kristaps 1581: /* FALLTHROUGH */
1.141 schwarze 1582: case MDOC_BODY:
1.50 kristaps 1583: /* FALLTHROUGH */
1.141 schwarze 1584: case MDOC_TAIL:
1.50 kristaps 1585: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1586: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1587: break;
1.68 schwarze 1588: if (mdocs[n->tok].mask)
1.78 schwarze 1589: putmdockey(mpage, n->child,
1590: mdocs[n->tok].mask);
1.50 kristaps 1591: break;
1592: default:
1593: assert(MDOC_ROOT != n->type);
1594: continue;
1595: }
1596: if (NULL != n->child)
1.172 schwarze 1597: parse_mdoc(mpage, meta, n);
1.1 kristaps 1598: }
1599: }
1600:
1.25 schwarze 1601: static int
1.172 schwarze 1602: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1603: const struct mdoc_node *n)
1.1 kristaps 1604: {
1.176 schwarze 1605: char *start, *end;
1.1 kristaps 1606: size_t sz;
1.25 schwarze 1607:
1.50 kristaps 1608: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1609: NULL == (n = n->child) ||
1610: MDOC_TEXT != n->type)
1.25 schwarze 1611: return(0);
1.1 kristaps 1612:
1613: /*
1614: * Only consider those `Fd' macro fields that begin with an
1615: * "inclusion" token (versus, e.g., #define).
1616: */
1.50 kristaps 1617:
1.1 kristaps 1618: if (strcmp("#include", n->string))
1.25 schwarze 1619: return(0);
1.1 kristaps 1620:
1621: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1622: return(0);
1.1 kristaps 1623:
1624: /*
1625: * Strip away the enclosing angle brackets and make sure we're
1626: * not zero-length.
1627: */
1628:
1629: start = n->string;
1630: if ('<' == *start || '"' == *start)
1631: start++;
1632:
1633: if (0 == (sz = strlen(start)))
1.25 schwarze 1634: return(0);
1.1 kristaps 1635:
1636: end = &start[(int)sz - 1];
1637: if ('>' == *end || '"' == *end)
1638: end--;
1639:
1.50 kristaps 1640: if (end > start)
1.78 schwarze 1641: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1642: return(0);
1.1 kristaps 1643: }
1644:
1.178 schwarze 1645: static void
1646: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1647: {
1.112 schwarze 1648: char *cp;
1.178 schwarze 1649: size_t sz;
1.1 kristaps 1650:
1.178 schwarze 1651: if (n->type != MDOC_TEXT)
1652: return;
1.25 schwarze 1653:
1.178 schwarze 1654: /* Skip function pointer punctuation. */
1.1 kristaps 1655:
1.178 schwarze 1656: cp = n->string;
1657: while (*cp == '(' || *cp == '*')
1.1 kristaps 1658: cp++;
1.178 schwarze 1659: sz = strcspn(cp, "()");
1.1 kristaps 1660:
1.178 schwarze 1661: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1662: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1663: putkeys(mpage, cp, sz, NAME_SYN);
1664: }
1.25 schwarze 1665:
1.178 schwarze 1666: static int
1667: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1668: const struct mdoc_node *n)
1669: {
1670:
1671: if (n->child == NULL)
1672: return(0);
1673:
1674: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1675:
1.178 schwarze 1676: for (n = n->child->next; n != NULL; n = n->next)
1677: if (n->type == MDOC_TEXT)
1.78 schwarze 1678: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1679:
1.173 schwarze 1680: return(0);
1681: }
1682:
1683: static int
1684: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1685: const struct mdoc_node *n)
1686: {
1.177 schwarze 1687:
1688: if (n->type != MDOC_HEAD)
1689: return(1);
1.173 schwarze 1690:
1.178 schwarze 1691: if (n->child != NULL)
1692: parse_mdoc_fname(mpage, n->child);
1693:
1.25 schwarze 1694: return(0);
1.1 kristaps 1695: }
1696:
1.25 schwarze 1697: static int
1.172 schwarze 1698: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1699: const struct mdoc_node *n)
1.1 kristaps 1700: {
1.67 schwarze 1701: char *cp;
1.1 kristaps 1702:
1703: if (NULL == (n = n->child))
1.25 schwarze 1704: return(0);
1.1 kristaps 1705:
1.67 schwarze 1706: if (NULL == n->next) {
1.78 schwarze 1707: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1708: return(0);
1709: }
1710:
1.120 schwarze 1711: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1712: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1713: free(cp);
1714: return(0);
1.1 kristaps 1715: }
1716:
1.25 schwarze 1717: static int
1.172 schwarze 1718: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1719: const struct mdoc_node *n)
1.1 kristaps 1720: {
1721:
1.122 schwarze 1722: if (MDOC_BODY == n->type)
1723: mdoc_deroff(&mpage->desc, n);
1724: return(0);
1.1 kristaps 1725: }
1726:
1.25 schwarze 1727: static int
1.172 schwarze 1728: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1729: const struct mdoc_node *n)
1.1 kristaps 1730: {
1731:
1.129 schwarze 1732: if (SEC_NAME == n->sec)
1.133 schwarze 1733: putmdockey(mpage, n->child, NAME_TITLE);
1.172 schwarze 1734: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1735: if (n->child == NULL)
1736: putkey(mpage, meta->name, NAME_SYN);
1737: else
1738: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1739: }
1740: if ( ! (mpage->name_head_done ||
1741: n->child == NULL || n->child->string == NULL ||
1742: strcasecmp(n->child->string, meta->title))) {
1743: putkey(mpage, n->child->string, NAME_HEAD);
1744: mpage->name_head_done = 1;
1.172 schwarze 1745: }
1.129 schwarze 1746: return(0);
1.1 kristaps 1747: }
1748:
1.25 schwarze 1749: static int
1.172 schwarze 1750: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1751: const struct mdoc_node *n)
1.1 kristaps 1752: {
1753:
1.25 schwarze 1754: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1755: }
1756:
1.50 kristaps 1757: static int
1.172 schwarze 1758: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1759: const struct mdoc_node *n)
1.1 kristaps 1760: {
1761:
1.50 kristaps 1762: return(MDOC_HEAD == n->type);
1763: }
1.1 kristaps 1764:
1.50 kristaps 1765: static int
1.172 schwarze 1766: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1767: const struct mdoc_node *n)
1.50 kristaps 1768: {
1.1 kristaps 1769:
1.50 kristaps 1770: return(MDOC_BODY == n->type);
1.1 kristaps 1771: }
1772:
1.50 kristaps 1773: /*
1.66 schwarze 1774: * Add a string to the hash table for the current manual.
1775: * Each string has a bitmask telling which macros it belongs to.
1776: * When we finish the manual, we'll dump the table.
1.50 kristaps 1777: */
1778: static void
1.176 schwarze 1779: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1780: {
1.133 schwarze 1781: struct ohash *htab;
1.50 kristaps 1782: struct str *s;
1.111 schwarze 1783: const char *end;
1.65 schwarze 1784: unsigned int slot;
1.176 schwarze 1785: int i, mustfree;
1.25 schwarze 1786:
1.50 kristaps 1787: if (0 == sz)
1788: return;
1.111 schwarze 1789:
1.176 schwarze 1790: mustfree = render_string(&cp, &sz);
1791:
1.133 schwarze 1792: if (TYPE_Nm & v) {
1793: htab = &names;
1794: v &= name_mask;
1.169 schwarze 1795: if (v & NAME_FIRST)
1796: name_mask &= ~NAME_FIRST;
1.133 schwarze 1797: if (debug > 1)
1798: say(mpage->mlinks->file,
1.175 schwarze 1799: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1800: } else {
1801: htab = &strings;
1802: if (debug > 1)
1803: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1804: if ((uint64_t)1 << i & v)
1.133 schwarze 1805: say(mpage->mlinks->file,
1806: "Adding key %s=%*s",
1807: mansearch_keynames[i], sz, cp);
1.111 schwarze 1808: }
1.25 schwarze 1809:
1.65 schwarze 1810: end = cp + sz;
1.133 schwarze 1811: slot = ohash_qlookupi(htab, cp, &end);
1812: s = ohash_find(htab, slot);
1.25 schwarze 1813:
1.78 schwarze 1814: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1815: s->mask |= v;
1816: return;
1817: } else if (NULL == s) {
1.144 schwarze 1818: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1819: memcpy(s->key, cp, sz);
1.133 schwarze 1820: ohash_insert(htab, slot, s);
1.1 kristaps 1821: }
1.78 schwarze 1822: s->mpage = mpage;
1.50 kristaps 1823: s->mask = v;
1.176 schwarze 1824:
1825: if (mustfree)
1826: free(cp);
1.1 kristaps 1827: }
1828:
1.50 kristaps 1829: /*
1830: * Take a Unicode codepoint and produce its UTF-8 encoding.
1831: * This isn't the best way to do this, but it works.
1832: * The magic numbers are from the UTF-8 packaging.
1833: * They're not as scary as they seem: read the UTF-8 spec for details.
1834: */
1835: static size_t
1836: utf8(unsigned int cp, char out[7])
1.1 kristaps 1837: {
1.50 kristaps 1838: size_t rc;
1.1 kristaps 1839:
1.50 kristaps 1840: rc = 0;
1841: if (cp <= 0x0000007F) {
1842: rc = 1;
1843: out[0] = (char)cp;
1844: } else if (cp <= 0x000007FF) {
1845: rc = 2;
1846: out[0] = (cp >> 6 & 31) | 192;
1847: out[1] = (cp & 63) | 128;
1848: } else if (cp <= 0x0000FFFF) {
1849: rc = 3;
1850: out[0] = (cp >> 12 & 15) | 224;
1851: out[1] = (cp >> 6 & 63) | 128;
1852: out[2] = (cp & 63) | 128;
1853: } else if (cp <= 0x001FFFFF) {
1854: rc = 4;
1855: out[0] = (cp >> 18 & 7) | 240;
1856: out[1] = (cp >> 12 & 63) | 128;
1857: out[2] = (cp >> 6 & 63) | 128;
1858: out[3] = (cp & 63) | 128;
1859: } else if (cp <= 0x03FFFFFF) {
1860: rc = 5;
1861: out[0] = (cp >> 24 & 3) | 248;
1862: out[1] = (cp >> 18 & 63) | 128;
1863: out[2] = (cp >> 12 & 63) | 128;
1864: out[3] = (cp >> 6 & 63) | 128;
1865: out[4] = (cp & 63) | 128;
1866: } else if (cp <= 0x7FFFFFFF) {
1867: rc = 6;
1868: out[0] = (cp >> 30 & 1) | 252;
1869: out[1] = (cp >> 24 & 63) | 128;
1870: out[2] = (cp >> 18 & 63) | 128;
1871: out[3] = (cp >> 12 & 63) | 128;
1872: out[4] = (cp >> 6 & 63) | 128;
1873: out[5] = (cp & 63) | 128;
1874: } else
1.1 kristaps 1875: return(0);
1876:
1.50 kristaps 1877: out[rc] = '\0';
1878: return(rc);
1879: }
1.1 kristaps 1880:
1.50 kristaps 1881: /*
1.176 schwarze 1882: * If the string contains escape sequences,
1883: * replace it with an allocated rendering and return 1,
1884: * such that the caller can free it after use.
1885: * Otherwise, do nothing and return 0.
1.50 kristaps 1886: */
1.176 schwarze 1887: static int
1888: render_string(char **public, size_t *psz)
1.50 kristaps 1889: {
1.176 schwarze 1890: const char *src, *scp, *addcp, *seq;
1891: char *dst;
1892: size_t ssz, dsz, addsz;
1.114 schwarze 1893: char utfbuf[7], res[6];
1.176 schwarze 1894: int seqlen, unicode;
1.50 kristaps 1895:
1896: res[0] = '\\';
1897: res[1] = '\t';
1898: res[2] = ASCII_NBRSP;
1899: res[3] = ASCII_HYPH;
1.114 schwarze 1900: res[4] = ASCII_BREAK;
1901: res[5] = '\0';
1.1 kristaps 1902:
1.176 schwarze 1903: src = scp = *public;
1904: ssz = *psz;
1905: dst = NULL;
1906: dsz = 0;
1907:
1908: while (scp < src + *psz) {
1909:
1910: /* Leave normal characters unchanged. */
1911:
1912: if (strchr(res, *scp) == NULL) {
1913: if (dst != NULL)
1914: dst[dsz++] = *scp;
1915: scp++;
1916: continue;
1917: }
1.46 kristaps 1918:
1.50 kristaps 1919: /*
1.176 schwarze 1920: * Found something that requires replacing,
1921: * make sure we have a destination buffer.
1.50 kristaps 1922: */
1.176 schwarze 1923:
1924: if (dst == NULL) {
1925: dst = mandoc_malloc(ssz + 1);
1926: dsz = scp - src;
1927: memcpy(dst, src, dsz);
1.50 kristaps 1928: }
1.46 kristaps 1929:
1.176 schwarze 1930: /* Handle single-char special characters. */
1931:
1932: switch (*scp) {
1933: case '\\':
1934: break;
1.141 schwarze 1935: case '\t':
1.114 schwarze 1936: /* FALLTHROUGH */
1.141 schwarze 1937: case ASCII_NBRSP:
1.176 schwarze 1938: dst[dsz++] = ' ';
1939: scp++;
1940: continue;
1941: case ASCII_HYPH:
1942: dst[dsz++] = '-';
1.114 schwarze 1943: /* FALLTHROUGH */
1.141 schwarze 1944: case ASCII_BREAK:
1.176 schwarze 1945: scp++;
1.50 kristaps 1946: continue;
1.114 schwarze 1947: default:
1.176 schwarze 1948: abort();
1.114 schwarze 1949: }
1.46 kristaps 1950:
1.50 kristaps 1951: /*
1.176 schwarze 1952: * Found an escape sequence.
1953: * Read past the slash, then parse it.
1954: * Ignore everything except characters.
1.50 kristaps 1955: */
1.95 schwarze 1956:
1.176 schwarze 1957: scp++;
1958: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1959: continue;
1.1 kristaps 1960:
1.50 kristaps 1961: /*
1.95 schwarze 1962: * Render the special character
1963: * as either UTF-8 or ASCII.
1.50 kristaps 1964: */
1.95 schwarze 1965:
1966: if (write_utf8) {
1.176 schwarze 1967: unicode = mchars_spec2cp(mchars, seq, seqlen);
1968: if (unicode <= 0)
1.95 schwarze 1969: continue;
1.176 schwarze 1970: addsz = utf8(unicode, utfbuf);
1971: if (addsz == 0)
1.95 schwarze 1972: continue;
1.176 schwarze 1973: addcp = utfbuf;
1.95 schwarze 1974: } else {
1.176 schwarze 1975: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1976: if (addcp == NULL)
1.95 schwarze 1977: continue;
1.176 schwarze 1978: if (*addcp == ASCII_NBRSP) {
1979: addcp = " ";
1980: addsz = 1;
1.95 schwarze 1981: }
1982: }
1.1 kristaps 1983:
1.50 kristaps 1984: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1985:
1.176 schwarze 1986: ssz += addsz;
1987: dst = mandoc_realloc(dst, ssz + 1);
1988: memcpy(dst + dsz, addcp, addsz);
1989: dsz += addsz;
1990: }
1991: if (dst != NULL) {
1992: *public = dst;
1993: *psz = dsz;
1994: }
1995:
1996: /* Trim trailing whitespace and NUL-terminate. */
1997:
1998: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1999: --*psz;
2000: if (dst != NULL) {
2001: (*public)[*psz] = '\0';
2002: return(1);
2003: } else
2004: return(0);
1.1 kristaps 2005: }
2006:
1.118 schwarze 2007: static void
2008: dbadd_mlink(const struct mlink *mlink)
2009: {
2010: size_t i;
2011:
2012: i = 1;
2013: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2014: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2015: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2016: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2017: SQL_STEP(stmts[STMT_INSERT_LINK]);
2018: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2019: }
2020:
2021: static void
2022: dbadd_mlink_name(const struct mlink *mlink)
2023: {
1.175 schwarze 2024: uint64_t bits;
1.169 schwarze 2025: size_t i;
2026:
2027: dbadd_mlink(mlink);
1.160 schwarze 2028:
2029: i = 1;
1.175 schwarze 2030: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2031: bits = NAME_FILE & NAME_MASK;
2032: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2033: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2034: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2035: }
2036:
2037: i = 1;
2038: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2039: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2040: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2041: SQL_STEP(stmts[STMT_INSERT_NAME]);
2042: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2043: }
2044:
1.14 schwarze 2045: /*
1.50 kristaps 2046: * Flush the current page's terms (and their bits) into the database.
2047: * Wrap the entire set of additions in a transaction to make sqlite be a
2048: * little faster.
1.96 schwarze 2049: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2050: */
2051: static void
1.176 schwarze 2052: dbadd(struct mpage *mpage)
1.14 schwarze 2053: {
1.87 schwarze 2054: struct mlink *mlink;
1.50 kristaps 2055: struct str *key;
1.176 schwarze 2056: char *cp;
1.52 kristaps 2057: size_t i;
1.66 schwarze 2058: unsigned int slot;
1.176 schwarze 2059: int mustfree;
1.14 schwarze 2060:
1.128 schwarze 2061: mlink = mpage->mlinks;
1.43 kristaps 2062:
1.128 schwarze 2063: if (nodb) {
1.147 schwarze 2064: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2065: key = ohash_next(&names, &slot))
1.147 schwarze 2066: free(key);
2067: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2068: key = ohash_next(&strings, &slot))
1.147 schwarze 2069: free(key);
1.145 schwarze 2070: if (0 == debug)
2071: return;
1.128 schwarze 2072: while (NULL != mlink) {
2073: fputs(mlink->name, stdout);
2074: if (NULL == mlink->next ||
2075: strcmp(mlink->dsec, mlink->next->dsec) ||
2076: strcmp(mlink->fsec, mlink->next->fsec) ||
2077: strcmp(mlink->arch, mlink->next->arch)) {
2078: putchar('(');
2079: if ('\0' == *mlink->dsec)
2080: fputs(mlink->fsec, stdout);
2081: else
2082: fputs(mlink->dsec, stdout);
2083: if ('\0' != *mlink->arch)
2084: printf("/%s", mlink->arch);
2085: putchar(')');
2086: }
2087: mlink = mlink->next;
2088: if (NULL != mlink)
2089: fputs(", ", stdout);
2090: }
1.132 schwarze 2091: printf(" - %s\n", mpage->desc);
1.14 schwarze 2092: return;
1.128 schwarze 2093: }
2094:
2095: if (debug)
2096: say(mlink->file, "Adding to database");
1.28 kristaps 2097:
1.176 schwarze 2098: cp = mpage->desc;
2099: i = strlen(cp);
2100: mustfree = render_string(&cp, &i);
1.52 kristaps 2101: i = 1;
1.176 schwarze 2102: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2103: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2104: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2105: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2106: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2107: if (mustfree)
2108: free(cp);
1.81 schwarze 2109:
1.128 schwarze 2110: while (NULL != mlink) {
1.118 schwarze 2111: dbadd_mlink(mlink);
1.128 schwarze 2112: mlink = mlink->next;
2113: }
1.134 schwarze 2114: mlink = mpage->mlinks;
1.50 kristaps 2115:
1.133 schwarze 2116: for (key = ohash_first(&names, &slot); NULL != key;
2117: key = ohash_next(&names, &slot)) {
2118: assert(key->mpage == mpage);
2119: i = 1;
2120: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2121: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2122: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2123: SQL_STEP(stmts[STMT_INSERT_NAME]);
2124: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2125: free(key);
2126: }
1.66 schwarze 2127: for (key = ohash_first(&strings, &slot); NULL != key;
2128: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2129: assert(key->mpage == mpage);
1.52 kristaps 2130: i = 1;
2131: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2132: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2133: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2134: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2135: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2136: free(key);
1.38 schwarze 2137: }
1.14 schwarze 2138: }
2139:
1.5 kristaps 2140: static void
1.59 schwarze 2141: dbprune(void)
1.5 kristaps 2142: {
1.78 schwarze 2143: struct mpage *mpage;
1.82 schwarze 2144: struct mlink *mlink;
1.52 kristaps 2145: size_t i;
1.80 schwarze 2146: unsigned int slot;
1.5 kristaps 2147:
1.106 schwarze 2148: if (0 == nodb)
2149: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2150:
1.106 schwarze 2151: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2152: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2153: mlink = mpage->mlinks;
1.125 schwarze 2154: if (debug)
1.106 schwarze 2155: say(mlink->file, "Deleting from database");
2156: if (nodb)
2157: continue;
2158: for ( ; NULL != mlink; mlink = mlink->next) {
2159: i = 1;
2160: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2161: i, mlink->dsec);
2162: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2163: i, mlink->arch);
2164: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2165: i, mlink->name);
2166: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2167: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2168: }
1.5 kristaps 2169: }
1.106 schwarze 2170:
2171: if (0 == nodb)
2172: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2173: }
2174:
1.4 kristaps 2175: /*
1.50 kristaps 2176: * Close an existing database and its prepared statements.
2177: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2178: */
1.35 kristaps 2179: static void
1.59 schwarze 2180: dbclose(int real)
1.4 kristaps 2181: {
1.50 kristaps 2182: size_t i;
1.115 schwarze 2183: int status;
2184: pid_t child;
1.4 kristaps 2185:
1.50 kristaps 2186: if (nodb)
1.38 schwarze 2187: return;
1.50 kristaps 2188:
2189: for (i = 0; i < STMT__MAX; i++) {
2190: sqlite3_finalize(stmts[i]);
2191: stmts[i] = NULL;
1.4 kristaps 2192: }
2193:
1.50 kristaps 2194: sqlite3_close(db);
2195: db = NULL;
1.12 schwarze 2196:
1.50 kristaps 2197: if (real)
2198: return;
1.12 schwarze 2199:
1.115 schwarze 2200: if ('\0' == *tempfilename) {
2201: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2202: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2203: say(MANDOC_DB, "&rename");
1.115 schwarze 2204: }
2205: return;
2206: }
2207:
2208: switch (child = fork()) {
1.141 schwarze 2209: case -1:
1.115 schwarze 2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2211: say("", "&fork cmp");
1.115 schwarze 2212: return;
1.141 schwarze 2213: case 0:
1.115 schwarze 2214: execlp("cmp", "cmp", "-s",
2215: tempfilename, MANDOC_DB, NULL);
1.123 schwarze 2216: say("", "&exec cmp");
1.115 schwarze 2217: exit(0);
2218: default:
2219: break;
2220: }
2221: if (-1 == waitpid(child, &status, 0)) {
2222: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2223: say("", "&wait cmp");
1.115 schwarze 2224: } else if (WIFSIGNALED(status)) {
2225: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2226: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2227: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2228: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2229: say(MANDOC_DB,
2230: "Data changed, but cannot replace database");
2231: }
2232:
2233: *strrchr(tempfilename, '/') = '\0';
2234: switch (child = fork()) {
1.141 schwarze 2235: case -1:
1.115 schwarze 2236: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2237: say("", "&fork rm");
1.115 schwarze 2238: return;
1.141 schwarze 2239: case 0:
1.115 schwarze 2240: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.123 schwarze 2241: say("", "&exec rm");
1.115 schwarze 2242: exit((int)MANDOCLEVEL_SYSERR);
2243: default:
2244: break;
2245: }
2246: if (-1 == waitpid(child, &status, 0)) {
2247: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2248: say("", "&wait rm");
1.115 schwarze 2249: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2250: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2251: say("", "%s: Cannot remove temporary directory",
2252: tempfilename);
1.59 schwarze 2253: }
1.50 kristaps 2254: }
1.14 schwarze 2255:
1.50 kristaps 2256: /*
2257: * This is straightforward stuff.
2258: * Open a database connection to a "temporary" database, then open a set
2259: * of prepared statements we'll use over and over again.
2260: * If "real" is set, we use the existing database; if not, we truncate a
2261: * temporary one.
2262: * Must be matched by dbclose().
2263: */
2264: static int
1.59 schwarze 2265: dbopen(int real)
1.50 kristaps 2266: {
1.115 schwarze 2267: const char *sql;
1.50 kristaps 2268: int rc, ofl;
1.12 schwarze 2269:
1.141 schwarze 2270: if (nodb)
1.50 kristaps 2271: return(1);
1.12 schwarze 2272:
1.115 schwarze 2273: *tempfilename = '\0';
1.63 schwarze 2274: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2275:
2276: if (real) {
2277: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2278: if (SQLITE_OK != rc) {
1.63 schwarze 2279: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2280: if (SQLITE_CANTOPEN != rc)
2281: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.63 schwarze 2282: return(0);
2283: }
1.115 schwarze 2284: goto prepare_statements;
2285: }
2286:
2287: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2288:
1.115 schwarze 2289: remove(MANDOC_DB "~");
2290: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2291: if (SQLITE_OK == rc)
1.115 schwarze 2292: goto create_tables;
1.116 schwarze 2293: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2294: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2295: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.50 kristaps 2296: return(0);
2297: }
1.12 schwarze 2298:
1.143 schwarze 2299: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2300: sizeof(tempfilename));
1.115 schwarze 2301: if (NULL == mkdtemp(tempfilename)) {
2302: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2303: say("", "&%s", tempfilename);
1.115 schwarze 2304: return(0);
2305: }
1.143 schwarze 2306: (void)strlcat(tempfilename, "/" MANDOC_DB,
2307: sizeof(tempfilename));
1.115 schwarze 2308: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2309: if (SQLITE_OK != rc) {
1.59 schwarze 2310: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2311: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.50 kristaps 2312: return(0);
2313: }
1.12 schwarze 2314:
1.115 schwarze 2315: create_tables:
1.81 schwarze 2316: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2317: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2318: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2319: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2320: ");\n"
2321: "\n"
1.81 schwarze 2322: "CREATE TABLE \"mlinks\" (\n"
2323: " \"sec\" TEXT NOT NULL,\n"
2324: " \"arch\" TEXT NOT NULL,\n"
2325: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2326: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2327: "ON DELETE CASCADE\n"
1.81 schwarze 2328: ");\n"
1.136 schwarze 2329: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2330: "\n"
1.133 schwarze 2331: "CREATE TABLE \"names\" (\n"
2332: " \"bits\" INTEGER NOT NULL,\n"
2333: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2334: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2335: "ON DELETE CASCADE,\n"
2336: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2337: ");\n"
2338: "\n"
1.50 kristaps 2339: "CREATE TABLE \"keys\" (\n"
2340: " \"bits\" INTEGER NOT NULL,\n"
2341: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2342: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2343: "ON DELETE CASCADE\n"
1.136 schwarze 2344: ");\n"
2345: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2346:
1.50 kristaps 2347: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2348: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2349: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2350: sqlite3_close(db);
1.50 kristaps 2351: return(0);
2352: }
1.4 kristaps 2353:
1.57 schwarze 2354: prepare_statements:
1.146 schwarze 2355: if (SQLITE_OK != sqlite3_exec(db,
2356: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2357: exitcode = (int)MANDOCLEVEL_SYSERR;
2358: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2359: sqlite3_errmsg(db));
2360: sqlite3_close(db);
2361: return(0);
2362: }
2363:
1.137 schwarze 2364: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2365: "(SELECT pageid FROM mlinks WHERE "
2366: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2367: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2368: sql = "INSERT INTO mpages "
1.132 schwarze 2369: "(desc,form) VALUES (?,?)";
1.81 schwarze 2370: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2371: sql = "INSERT INTO mlinks "
1.104 schwarze 2372: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2373: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2374: sql = "SELECT bits FROM names where pageid = ?";
2375: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2376: sql = "INSERT INTO names "
2377: "(bits,name,pageid) VALUES (?,?,?)";
2378: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2379: sql = "INSERT INTO keys "
1.81 schwarze 2380: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2381: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2382:
2383: #ifndef __APPLE__
2384: /*
2385: * When opening a new database, we can turn off
2386: * synchronous mode for much better performance.
2387: */
2388:
1.146 schwarze 2389: if (real && SQLITE_OK != sqlite3_exec(db,
2390: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2391: exitcode = (int)MANDOCLEVEL_SYSERR;
2392: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2393: sqlite3_errmsg(db));
1.146 schwarze 2394: sqlite3_close(db);
2395: return(0);
2396: }
1.70 schwarze 2397: #endif
2398:
1.50 kristaps 2399: return(1);
2400: }
1.5 kristaps 2401:
1.50 kristaps 2402: static void *
1.152 schwarze 2403: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2404: {
1.12 schwarze 2405:
1.152 schwarze 2406: return(mandoc_calloc(n, sz));
1.50 kristaps 2407: }
1.12 schwarze 2408:
1.50 kristaps 2409: static void *
2410: hash_alloc(size_t sz, void *arg)
2411: {
1.5 kristaps 2412:
1.50 kristaps 2413: return(mandoc_malloc(sz));
2414: }
1.41 kristaps 2415:
1.50 kristaps 2416: static void
1.152 schwarze 2417: hash_free(void *p, void *arg)
1.50 kristaps 2418: {
1.4 kristaps 2419:
1.50 kristaps 2420: free(p);
1.4 kristaps 2421: }
2422:
1.50 kristaps 2423: static int
1.165 schwarze 2424: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2425: {
1.59 schwarze 2426: static char startdir[PATH_MAX];
1.151 schwarze 2427: static int getcwd_status; /* 1 = ok, 2 = failure */
2428: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2429: char *cp;
1.4 kristaps 2430:
1.59 schwarze 2431: /*
1.151 schwarze 2432: * Remember the original working directory, if possible.
2433: * This will be needed if the second or a later directory
2434: * on the command line is given as a relative path.
2435: * Do not error out if the current directory is not
2436: * searchable: Maybe it won't be needed after all.
2437: */
2438: if (0 == getcwd_status) {
2439: if (NULL == getcwd(startdir, sizeof(startdir))) {
2440: getcwd_status = 2;
2441: (void)strlcpy(startdir, strerror(errno),
2442: sizeof(startdir));
2443: } else
2444: getcwd_status = 1;
2445: }
2446:
2447: /*
2448: * We are leaving the old base directory.
2449: * Do not use it any longer, not even for messages.
2450: */
2451: *basedir = '\0';
2452:
2453: /*
2454: * If and only if the directory was changed earlier and
2455: * the next directory to process is given as a relative path,
2456: * first go back, or bail out if that is impossible.
1.59 schwarze 2457: */
1.151 schwarze 2458: if (chdir_status && '/' != *targetdir) {
2459: if (2 == getcwd_status) {
1.59 schwarze 2460: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2461: say("", "getcwd: %s", startdir);
1.59 schwarze 2462: return(0);
2463: }
1.151 schwarze 2464: if (-1 == chdir(startdir)) {
1.59 schwarze 2465: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2466: say("", "&chdir %s", startdir);
1.59 schwarze 2467: return(0);
2468: }
2469: }
1.151 schwarze 2470:
2471: /*
2472: * Always resolve basedir to the canonicalized absolute
2473: * pathname and append a trailing slash, such that
2474: * we can reliably check whether files are inside.
2475: */
1.59 schwarze 2476: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2477: if (report_baddir || errno != ENOENT) {
2478: exitcode = (int)MANDOCLEVEL_BADARG;
2479: say("", "&%s: realpath", targetdir);
2480: }
1.50 kristaps 2481: return(0);
1.59 schwarze 2482: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2483: if (report_baddir || errno != ENOENT) {
2484: exitcode = (int)MANDOCLEVEL_BADARG;
2485: say("", "&chdir");
2486: }
1.50 kristaps 2487: return(0);
1.150 schwarze 2488: }
1.151 schwarze 2489: chdir_status = 1;
1.150 schwarze 2490: cp = strchr(basedir, '\0');
2491: if ('/' != cp[-1]) {
2492: if (cp - basedir >= PATH_MAX - 1) {
2493: exitcode = (int)MANDOCLEVEL_SYSERR;
2494: say("", "Filename too long");
2495: return(0);
2496: }
2497: *cp++ = '/';
2498: *cp = '\0';
1.4 kristaps 2499: }
1.50 kristaps 2500: return(1);
1.56 schwarze 2501: }
2502:
2503: static void
1.59 schwarze 2504: say(const char *file, const char *format, ...)
1.56 schwarze 2505: {
2506: va_list ap;
1.123 schwarze 2507: int use_errno;
1.56 schwarze 2508:
1.59 schwarze 2509: if ('\0' != *basedir)
2510: fprintf(stderr, "%s", basedir);
2511: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2512: fputc('/', stderr);
1.56 schwarze 2513: if ('\0' != *file)
1.59 schwarze 2514: fprintf(stderr, "%s", file);
2515:
1.123 schwarze 2516: use_errno = 1;
2517: if (NULL != format) {
2518: switch (*format) {
1.141 schwarze 2519: case '&':
1.123 schwarze 2520: format++;
2521: break;
1.141 schwarze 2522: case '\0':
1.123 schwarze 2523: format = NULL;
2524: break;
2525: default:
2526: use_errno = 0;
2527: break;
2528: }
2529: }
2530: if (NULL != format) {
2531: if ('\0' != *basedir || '\0' != *file)
2532: fputs(": ", stderr);
2533: va_start(ap, format);
2534: vfprintf(stderr, format, ap);
2535: va_end(ap);
2536: }
2537: if (use_errno) {
2538: if ('\0' != *basedir || '\0' != *file || NULL != format)
2539: fputs(": ", stderr);
1.59 schwarze 2540: perror(NULL);
1.123 schwarze 2541: } else
2542: fputc('\n', stderr);
1.1 kristaps 2543: }
CVSweb