Annotation of mandoc/mandocdb.c, Revision 1.181
1.181 ! schwarze 1: /* $Id: mandocdb.c,v 1.180 2014/12/30 20:41:00 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;
1.181 ! schwarze 353: mpages_info.free = mlinks_info.free = hash_free;
! 354: mpages_info.data = mlinks_info.data = NULL;
1.50 kristaps 355:
1.79 schwarze 356: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 357: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 358:
359: progname = strrchr(argv[0], '/');
360: if (progname == NULL)
361: progname = argv[0];
362: else
363: ++progname;
364:
1.50 kristaps 365: /*
1.141 schwarze 366: * We accept a few different invocations.
1.50 kristaps 367: * The CHECKOP macro makes sure that invocation styles don't
368: * clobber each other.
369: */
370: #define CHECKOP(_op, _ch) do \
371: if (OP_DEFAULT != (_op)) { \
1.153 schwarze 372: fprintf(stderr, "%s: -%c: Conflicting option\n", \
373: progname, (_ch)); \
1.50 kristaps 374: goto usage; \
375: } while (/*CONSTCOND*/0)
1.10 kristaps 376:
1.59 schwarze 377: path_arg = NULL;
1.38 schwarze 378: op = OP_DEFAULT;
1.1 kristaps 379:
1.126 schwarze 380: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 381: switch (ch) {
1.141 schwarze 382: case 'a':
1.12 schwarze 383: use_all = 1;
384: break;
1.141 schwarze 385: case 'C':
1.50 kristaps 386: CHECKOP(op, ch);
1.59 schwarze 387: path_arg = optarg;
1.38 schwarze 388: op = OP_CONFFILE;
1.34 schwarze 389: break;
1.141 schwarze 390: case 'D':
1.125 schwarze 391: debug++;
392: break;
1.141 schwarze 393: case 'd':
1.50 kristaps 394: CHECKOP(op, ch);
1.59 schwarze 395: path_arg = optarg;
1.5 kristaps 396: op = OP_UPDATE;
397: break;
1.141 schwarze 398: case 'n':
1.50 kristaps 399: nodb = 1;
400: break;
1.141 schwarze 401: case 'p':
1.126 schwarze 402: warnings = 1;
403: break;
1.141 schwarze 404: case 'Q':
1.116 schwarze 405: mparse_options |= MPARSE_QUICK;
1.102 schwarze 406: break;
1.141 schwarze 407: case 'T':
1.95 schwarze 408: if (strcmp(optarg, "utf8")) {
1.153 schwarze 409: fprintf(stderr, "%s: -T%s: "
410: "Unsupported output format\n",
411: progname, optarg);
1.95 schwarze 412: goto usage;
413: }
414: write_utf8 = 1;
415: break;
1.141 schwarze 416: case 't':
1.50 kristaps 417: CHECKOP(op, ch);
1.38 schwarze 418: dup2(STDOUT_FILENO, STDERR_FILENO);
419: op = OP_TEST;
1.50 kristaps 420: nodb = warnings = 1;
1.38 schwarze 421: break;
1.141 schwarze 422: case 'u':
1.50 kristaps 423: CHECKOP(op, ch);
1.59 schwarze 424: path_arg = optarg;
1.5 kristaps 425: op = OP_DELETE;
426: break;
1.141 schwarze 427: case 'v':
1.126 schwarze 428: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 429: break;
1.1 kristaps 430: default:
1.38 schwarze 431: goto usage;
1.1 kristaps 432: }
433:
434: argc -= optind;
435: argv += optind;
436:
1.38 schwarze 437: if (OP_CONFFILE == op && argc > 0) {
1.153 schwarze 438: fprintf(stderr, "%s: -C: Too many arguments\n",
439: progname);
1.38 schwarze 440: goto usage;
441: }
442:
1.59 schwarze 443: exitcode = (int)MANDOCLEVEL_OK;
1.176 schwarze 444: mchars = mchars_alloc();
1.168 schwarze 445: mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
1.176 schwarze 446: mchars, NULL);
1.78 schwarze 447: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 448: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 449:
450: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 451:
1.50 kristaps 452: /*
1.150 schwarze 453: * Most of these deal with a specific directory.
1.140 schwarze 454: * Jump into that directory first.
1.50 kristaps 455: */
1.165 schwarze 456: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 457: goto out;
1.140 schwarze 458:
1.138 schwarze 459: if (dbopen(1)) {
1.140 schwarze 460: /*
461: * The existing database is usable. Process
462: * all files specified on the command-line.
463: */
464: use_all = 1;
1.138 schwarze 465: for (i = 0; i < argc; i++)
466: filescan(argv[i]);
467: if (OP_TEST != op)
468: dbprune();
469: } else {
470: /*
471: * Database missing or corrupt.
472: * Recreate from scratch.
473: */
1.140 schwarze 474: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 475: op = OP_DEFAULT;
476: if (0 == treescan())
477: goto out;
478: if (0 == dbopen(0))
479: goto out;
480: }
1.50 kristaps 481: if (OP_DELETE != op)
1.176 schwarze 482: mpages_merge(mp);
1.138 schwarze 483: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 484: } else {
485: /*
486: * If we have arguments, use them as our manpaths.
487: * If we don't, grok from manpath(1) or however else
488: * manpath_parse() wants to do it.
489: */
490: if (argc > 0) {
1.144 schwarze 491: dirs.paths = mandoc_reallocarray(NULL,
492: argc, sizeof(char *));
1.50 kristaps 493: dirs.sz = (size_t)argc;
494: for (i = 0; i < argc; i++)
495: dirs.paths[i] = mandoc_strdup(argv[i]);
496: } else
1.59 schwarze 497: manpath_parse(&dirs, path_arg, NULL, NULL);
1.127 schwarze 498:
499: if (0 == dirs.sz) {
500: exitcode = (int)MANDOCLEVEL_BADARG;
501: say("", "Empty manpath");
502: }
1.50 kristaps 503:
504: /*
1.69 schwarze 505: * First scan the tree rooted at a base directory, then
506: * build a new database and finally move it into place.
1.50 kristaps 507: * Ignore zero-length directories and strip trailing
508: * slashes.
509: */
510: for (j = 0; j < dirs.sz; j++) {
511: sz = strlen(dirs.paths[j]);
512: if (sz && '/' == dirs.paths[j][sz - 1])
513: dirs.paths[j][--sz] = '\0';
514: if (0 == sz)
515: continue;
1.66 schwarze 516:
517: if (j) {
1.78 schwarze 518: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 519: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 520: }
521:
1.165 schwarze 522: if (0 == set_basedir(dirs.paths[j], argc > 0))
523: continue;
1.59 schwarze 524: if (0 == treescan())
1.165 schwarze 525: continue;
1.59 schwarze 526: if (0 == dbopen(0))
1.165 schwarze 527: continue;
1.52 kristaps 528:
1.176 schwarze 529: mpages_merge(mp);
1.147 schwarze 530: if (warnings && !nodb &&
1.130 schwarze 531: ! (MPARSE_QUICK & mparse_options))
532: names_check();
1.59 schwarze 533: dbclose(0);
1.66 schwarze 534:
535: if (j + 1 < dirs.sz) {
1.80 schwarze 536: mpages_free();
1.78 schwarze 537: ohash_delete(&mpages);
1.83 schwarze 538: ohash_delete(&mlinks);
1.66 schwarze 539: }
1.50 kristaps 540: }
541: }
542: out:
543: manpath_free(&dirs);
1.168 schwarze 544: mparse_free(mp);
1.176 schwarze 545: mchars_free(mchars);
1.80 schwarze 546: mpages_free();
1.78 schwarze 547: ohash_delete(&mpages);
1.83 schwarze 548: ohash_delete(&mlinks);
1.59 schwarze 549: return(exitcode);
1.50 kristaps 550: usage:
1.126 schwarze 551: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
552: " %s [-aDnpQ] [-Tutf8] dir ...\n"
553: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
554: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 555: " %s [-Q] -t file ...\n",
1.141 schwarze 556: progname, progname, progname,
1.50 kristaps 557: progname, progname);
558:
1.59 schwarze 559: return((int)MANDOCLEVEL_BADARG);
1.50 kristaps 560: }
561:
562: /*
1.59 schwarze 563: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 564: * We use fts(), scanning directory parts along the way for clues to our
565: * section and architecture.
566: *
567: * If use_all has been specified, grok all files.
568: * If not, sanitise paths to the following:
569: *
1.141 schwarze 570: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 571: * or
572: * [./]cat<section>[/<arch>]/<name>.0
573: *
574: * TODO: accomodate for multi-language directories.
575: */
576: static int
1.59 schwarze 577: treescan(void)
1.50 kristaps 578: {
1.139 schwarze 579: char buf[PATH_MAX];
1.50 kristaps 580: FTS *f;
581: FTSENT *ff;
1.84 schwarze 582: struct mlink *mlink;
1.124 schwarze 583: int dform, gzip;
1.94 schwarze 584: char *dsec, *arch, *fsec, *cp;
585: const char *path;
1.50 kristaps 586: const char *argv[2];
587:
588: argv[0] = ".";
589: argv[1] = (char *)NULL;
590:
1.139 schwarze 591: f = fts_open((char * const *)argv,
592: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.50 kristaps 593: if (NULL == f) {
1.59 schwarze 594: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 595: say("", "&fts_open");
1.50 kristaps 596: return(0);
597: }
598:
599: dsec = arch = NULL;
600: dform = FORM_NONE;
601:
602: while (NULL != (ff = fts_read(f))) {
603: path = ff->fts_path + 2;
1.139 schwarze 604: switch (ff->fts_info) {
605:
606: /*
607: * Symbolic links require various sanity checks,
608: * then get handled just like regular files.
609: */
1.141 schwarze 610: case FTS_SL:
1.139 schwarze 611: if (NULL == realpath(path, buf)) {
612: if (warnings)
613: say(path, "&realpath");
614: continue;
615: }
616: if (strstr(buf, basedir) != buf) {
617: if (warnings) say("",
618: "%s: outside base directory", buf);
619: continue;
620: }
621: /* Use logical inode to avoid mpages dupe. */
622: if (-1 == stat(path, ff->fts_statp)) {
623: if (warnings)
624: say(path, "&stat");
625: continue;
626: }
627: /* FALLTHROUGH */
628:
1.50 kristaps 629: /*
1.83 schwarze 630: * If we're a regular file, add an mlink by using the
1.50 kristaps 631: * stored directory data and handling the filename.
632: */
1.141 schwarze 633: case FTS_F:
1.60 schwarze 634: if (0 == strcmp(path, MANDOC_DB))
635: continue;
1.50 kristaps 636: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 637: if (warnings)
1.59 schwarze 638: say(path, "Extraneous file");
1.50 kristaps 639: continue;
1.124 schwarze 640: }
641: gzip = 0;
642: fsec = NULL;
643: while (NULL == fsec) {
644: fsec = strrchr(ff->fts_name, '.');
645: if (NULL == fsec || strcmp(fsec+1, "gz"))
646: break;
647: gzip = 1;
648: *fsec = '\0';
649: fsec = NULL;
650: }
651: if (NULL == fsec) {
1.60 schwarze 652: if ( ! use_all) {
1.56 schwarze 653: if (warnings)
1.60 schwarze 654: say(path,
655: "No filename suffix");
1.50 kristaps 656: continue;
657: }
1.84 schwarze 658: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 659: if (warnings)
660: say(path, "Skip html");
661: continue;
1.84 schwarze 662: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 663: if (warnings)
664: say(path, "Skip ps");
665: continue;
1.84 schwarze 666: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 667: if (warnings)
668: say(path, "Skip pdf");
669: continue;
670: } else if ( ! use_all &&
1.84 schwarze 671: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
672: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 673: if (warnings)
674: say(path, "Wrong filename suffix");
675: continue;
1.66 schwarze 676: } else
1.84 schwarze 677: fsec[-1] = '\0';
1.94 schwarze 678:
1.84 schwarze 679: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 680: if (strlcpy(mlink->file, path,
681: sizeof(mlink->file)) >=
682: sizeof(mlink->file)) {
683: say(path, "Filename too long");
684: free(mlink);
685: continue;
686: }
1.84 schwarze 687: mlink->dform = dform;
1.94 schwarze 688: mlink->dsec = dsec;
689: mlink->arch = arch;
690: mlink->name = ff->fts_name;
691: mlink->fsec = fsec;
1.124 schwarze 692: mlink->gzip = gzip;
1.84 schwarze 693: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 694: continue;
1.139 schwarze 695:
1.141 schwarze 696: case FTS_D:
1.139 schwarze 697: /* FALLTHROUGH */
1.141 schwarze 698: case FTS_DP:
1.139 schwarze 699: break;
700:
701: default:
1.60 schwarze 702: if (warnings)
703: say(path, "Not a regular file");
1.50 kristaps 704: continue;
1.60 schwarze 705: }
1.1 kristaps 706:
1.50 kristaps 707: switch (ff->fts_level) {
1.141 schwarze 708: case 0:
1.50 kristaps 709: /* Ignore the root directory. */
710: break;
1.141 schwarze 711: case 1:
1.50 kristaps 712: /*
713: * This might contain manX/ or catX/.
714: * Try to infer this from the name.
715: * If we're not in use_all, enforce it.
716: */
717: cp = ff->fts_name;
718: if (FTS_DP == ff->fts_info)
719: break;
1.1 kristaps 720:
1.50 kristaps 721: if (0 == strncmp(cp, "man", 3)) {
722: dform = FORM_SRC;
1.66 schwarze 723: dsec = cp + 3;
1.50 kristaps 724: } else if (0 == strncmp(cp, "cat", 3)) {
725: dform = FORM_CAT;
1.66 schwarze 726: dsec = cp + 3;
1.94 schwarze 727: } else {
728: dform = FORM_NONE;
729: dsec = NULL;
1.50 kristaps 730: }
1.1 kristaps 731:
1.141 schwarze 732: if (NULL != dsec || use_all)
1.50 kristaps 733: break;
1.1 kristaps 734:
1.56 schwarze 735: if (warnings)
1.59 schwarze 736: say(path, "Unknown directory part");
1.50 kristaps 737: fts_set(f, ff, FTS_SKIP);
738: break;
1.141 schwarze 739: case 2:
1.50 kristaps 740: /*
741: * Possibly our architecture.
742: * If we're descending, keep tabs on it.
743: */
744: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 745: arch = ff->fts_name;
1.94 schwarze 746: else
747: arch = NULL;
1.50 kristaps 748: break;
749: default:
750: if (FTS_DP == ff->fts_info || use_all)
751: break;
1.56 schwarze 752: if (warnings)
1.59 schwarze 753: say(path, "Extraneous directory part");
1.50 kristaps 754: fts_set(f, ff, FTS_SKIP);
755: break;
1.5 kristaps 756: }
1.50 kristaps 757: }
758:
759: fts_close(f);
760: return(1);
761: }
1.5 kristaps 762:
1.50 kristaps 763: /*
1.89 schwarze 764: * Add a file to the mlinks table.
1.50 kristaps 765: * Do not verify that it's a "valid" looking manpage (we'll do that
766: * later).
767: *
768: * Try to infer the manual section, architecture, and page name from the
769: * path, assuming it looks like
770: *
1.141 schwarze 771: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 772: * or
773: * [./]cat<section>[/<arch>]/<name>.0
774: *
775: * See treescan() for the fts(3) version of this.
776: */
777: static void
1.59 schwarze 778: filescan(const char *file)
1.50 kristaps 779: {
1.59 schwarze 780: char buf[PATH_MAX];
1.84 schwarze 781: struct stat st;
782: struct mlink *mlink;
1.59 schwarze 783: char *p, *start;
1.5 kristaps 784:
1.50 kristaps 785: assert(use_all);
1.5 kristaps 786:
1.50 kristaps 787: if (0 == strncmp(file, "./", 2))
788: file += 2;
1.5 kristaps 789:
1.139 schwarze 790: /*
791: * We have to do lstat(2) before realpath(3) loses
792: * the information whether this is a symbolic link.
793: * We need to know that because for symbolic links,
794: * we want to use the orginal file name, while for
795: * regular files, we want to use the real path.
796: */
797: if (-1 == lstat(file, &st)) {
798: exitcode = (int)MANDOCLEVEL_BADARG;
799: say(file, "&lstat");
800: return;
801: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
802: exitcode = (int)MANDOCLEVEL_BADARG;
803: say(file, "Not a regular file");
804: return;
805: }
806:
807: /*
808: * We have to resolve the file name to the real path
809: * in any case for the base directory check.
810: */
1.59 schwarze 811: if (NULL == realpath(file, buf)) {
812: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 813: say(file, "&realpath");
1.59 schwarze 814: return;
1.106 schwarze 815: }
816:
1.150 schwarze 817: if (OP_TEST == op)
1.106 schwarze 818: start = buf;
1.150 schwarze 819: else if (strstr(buf, basedir) == buf)
820: start = buf + strlen(basedir);
1.106 schwarze 821: else {
1.59 schwarze 822: exitcode = (int)MANDOCLEVEL_BADARG;
823: say("", "%s: outside base directory", buf);
824: return;
1.106 schwarze 825: }
826:
1.139 schwarze 827: /*
828: * Now we are sure the file is inside our tree.
829: * If it is a symbolic link, ignore the real path
830: * and use the original name.
831: * This implies passing stuff like "cat1/../man1/foo.1"
832: * on the command line won't work. So don't do that.
833: * Note the stat(2) can still fail if the link target
834: * doesn't exist.
835: */
836: if (S_IFLNK & st.st_mode) {
837: if (-1 == stat(buf, &st)) {
838: exitcode = (int)MANDOCLEVEL_BADARG;
839: say(file, "&stat");
840: return;
841: }
1.143 schwarze 842: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
843: say(file, "Filename too long");
844: return;
845: }
1.150 schwarze 846: start = buf;
847: if (OP_TEST != op && strstr(buf, basedir) == buf)
848: start += strlen(basedir);
1.50 kristaps 849: }
1.106 schwarze 850:
1.84 schwarze 851: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 852: mlink->dform = FORM_NONE;
1.143 schwarze 853: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
854: sizeof(mlink->file)) {
855: say(start, "Filename too long");
1.180 schwarze 856: free(mlink);
1.143 schwarze 857: return;
858: }
1.17 schwarze 859:
1.50 kristaps 860: /*
861: * First try to guess our directory structure.
862: * If we find a separator, try to look for man* or cat*.
863: * If we find one of these and what's underneath is a directory,
864: * assume it's an architecture.
865: */
866: if (NULL != (p = strchr(start, '/'))) {
867: *p++ = '\0';
868: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 869: mlink->dform = FORM_SRC;
1.94 schwarze 870: mlink->dsec = start + 3;
1.50 kristaps 871: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 872: mlink->dform = FORM_CAT;
1.94 schwarze 873: mlink->dsec = start + 3;
1.17 schwarze 874: }
1.5 kristaps 875:
1.50 kristaps 876: start = p;
1.84 schwarze 877: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 878: *p++ = '\0';
1.94 schwarze 879: mlink->arch = start;
1.50 kristaps 880: start = p;
1.84 schwarze 881: }
1.50 kristaps 882: }
883:
884: /*
885: * Now check the file suffix.
886: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
887: */
888: p = strrchr(start, '\0');
889: while (p-- > start && '/' != *p && '.' != *p)
890: /* Loop. */ ;
891:
892: if ('.' == *p) {
893: *p++ = '\0';
1.94 schwarze 894: mlink->fsec = p;
1.5 kristaps 895: }
896:
1.10 kristaps 897: /*
1.50 kristaps 898: * Now try to parse the name.
899: * Use the filename portion of the path.
1.10 kristaps 900: */
1.84 schwarze 901: mlink->name = start;
1.50 kristaps 902: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 903: mlink->name = p + 1;
1.50 kristaps 904: *p = '\0';
1.84 schwarze 905: }
906: mlink_add(mlink, &st);
1.50 kristaps 907: }
1.5 kristaps 908:
1.50 kristaps 909: static void
1.84 schwarze 910: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 911: {
1.83 schwarze 912: struct inodev inodev;
1.78 schwarze 913: struct mpage *mpage;
1.83 schwarze 914: unsigned int slot;
1.50 kristaps 915:
1.84 schwarze 916: assert(NULL != mlink->file);
1.50 kristaps 917:
1.94 schwarze 918: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
919: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
920: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
921: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 922:
923: if ('0' == *mlink->fsec) {
924: free(mlink->fsec);
925: mlink->fsec = mandoc_strdup(mlink->dsec);
926: mlink->fform = FORM_CAT;
927: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
928: mlink->fform = FORM_SRC;
1.74 schwarze 929: else
1.84 schwarze 930: mlink->fform = FORM_NONE;
1.82 schwarze 931:
1.83 schwarze 932: slot = ohash_qlookup(&mlinks, mlink->file);
933: assert(NULL == ohash_find(&mlinks, slot));
934: ohash_insert(&mlinks, slot, mlink);
935:
936: inodev.st_ino = st->st_ino;
937: inodev.st_dev = st->st_dev;
938: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
939: sizeof(struct inodev), inodev.st_ino);
940: mpage = ohash_find(&mpages, slot);
941: if (NULL == mpage) {
942: mpage = mandoc_calloc(1, sizeof(struct mpage));
943: mpage->inodev.st_ino = inodev.st_ino;
944: mpage->inodev.st_dev = inodev.st_dev;
945: ohash_insert(&mpages, slot, mpage);
946: } else
1.85 schwarze 947: mlink->next = mpage->mlinks;
1.82 schwarze 948: mpage->mlinks = mlink;
1.118 schwarze 949: mlink->mpage = mpage;
1.82 schwarze 950: }
951:
952: static void
953: mlink_free(struct mlink *mlink)
954: {
955:
956: free(mlink->dsec);
957: free(mlink->arch);
958: free(mlink->name);
959: free(mlink->fsec);
960: free(mlink);
1.50 kristaps 961: }
1.3 kristaps 962:
1.50 kristaps 963: static void
1.78 schwarze 964: mpages_free(void)
1.50 kristaps 965: {
1.78 schwarze 966: struct mpage *mpage;
1.82 schwarze 967: struct mlink *mlink;
1.80 schwarze 968: unsigned int slot;
1.3 kristaps 969:
1.80 schwarze 970: mpage = ohash_first(&mpages, &slot);
971: while (NULL != mpage) {
1.82 schwarze 972: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 973: mpage->mlinks = mlink->next;
1.82 schwarze 974: mlink_free(mlink);
975: }
1.78 schwarze 976: free(mpage->sec);
977: free(mpage->arch);
1.82 schwarze 978: free(mpage->title);
979: free(mpage->desc);
1.78 schwarze 980: free(mpage);
1.80 schwarze 981: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 982: }
983: }
1.38 schwarze 984:
1.50 kristaps 985: /*
1.89 schwarze 986: * For each mlink to the mpage, check whether the path looks like
987: * it is formatted, and if it does, check whether a source manual
988: * exists by the same name, ignoring the suffix.
989: * If both conditions hold, drop the mlink.
990: */
991: static void
992: mlinks_undupe(struct mpage *mpage)
993: {
994: char buf[PATH_MAX];
995: struct mlink **prev;
996: struct mlink *mlink;
997: char *bufp;
998:
999: mpage->form = FORM_CAT;
1.90 schwarze 1000: prev = &mpage->mlinks;
1001: while (NULL != (mlink = *prev)) {
1.89 schwarze 1002: if (FORM_CAT != mlink->dform) {
1003: mpage->form = FORM_NONE;
1.90 schwarze 1004: goto nextlink;
1.89 schwarze 1005: }
1.143 schwarze 1006: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1007: bufp = strstr(buf, "cat");
1008: assert(NULL != bufp);
1009: memcpy(bufp, "man", 3);
1010: if (NULL != (bufp = strrchr(buf, '.')))
1011: *++bufp = '\0';
1.143 schwarze 1012: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1013: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1014: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1015: goto nextlink;
1.89 schwarze 1016: if (warnings)
1017: say(mlink->file, "Man source exists: %s", buf);
1018: if (use_all)
1.90 schwarze 1019: goto nextlink;
1.89 schwarze 1020: *prev = mlink->next;
1021: mlink_free(mlink);
1.90 schwarze 1022: continue;
1023: nextlink:
1024: prev = &(*prev)->next;
1.89 schwarze 1025: }
1026: }
1027:
1.131 schwarze 1028: static void
1.93 schwarze 1029: mlink_check(struct mpage *mpage, struct mlink *mlink)
1030: {
1.131 schwarze 1031: struct str *str;
1032: unsigned int slot;
1.93 schwarze 1033:
1034: /*
1035: * Check whether the manual section given in a file
1036: * agrees with the directory where the file is located.
1037: * Some manuals have suffixes like (3p) on their
1038: * section number either inside the file or in the
1039: * directory name, some are linked into more than one
1040: * section, like encrypt(1) = makekey(8).
1041: */
1042:
1043: if (FORM_SRC == mpage->form &&
1.131 schwarze 1044: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1045: say(mlink->file, "Section \"%s\" manual in %s directory",
1046: mpage->sec, mlink->dsec);
1047:
1048: /*
1049: * Manual page directories exist for each kernel
1050: * architecture as returned by machine(1).
1051: * However, many manuals only depend on the
1052: * application architecture as returned by arch(1).
1053: * For example, some (2/ARM) manuals are shared
1054: * across the "armish" and "zaurus" kernel
1055: * architectures.
1056: * A few manuals are even shared across completely
1057: * different architectures, for example fdformat(1)
1058: * on amd64, i386, sparc, and sparc64.
1059: */
1060:
1.131 schwarze 1061: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1062: say(mlink->file, "Architecture \"%s\" manual in "
1063: "\"%s\" directory", mpage->arch, mlink->arch);
1064:
1.131 schwarze 1065: /*
1066: * XXX
1.133 schwarze 1067: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1068: */
1069:
1070: if (FORM_CAT == mpage->form)
1071: return;
1072:
1073: /*
1074: * Check whether this mlink
1075: * appears as a name in the NAME section.
1076: */
1.93 schwarze 1077:
1.133 schwarze 1078: slot = ohash_qlookup(&names, mlink->name);
1079: str = ohash_find(&names, slot);
1.131 schwarze 1080: assert(NULL != str);
1.133 schwarze 1081: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1082: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1083: }
1084:
1.89 schwarze 1085: /*
1.80 schwarze 1086: * Run through the files in the global vector "mpages"
1087: * and add them to the database specified in "basedir".
1.50 kristaps 1088: *
1089: * This handles the parsing scheme itself, using the cues of directory
1090: * and filename to determine whether the file is parsable or not.
1091: */
1.59 schwarze 1092: static void
1.176 schwarze 1093: mpages_merge(struct mparse *mp)
1.50 kristaps 1094: {
1.113 schwarze 1095: char any[] = "any";
1.101 schwarze 1096: struct ohash_info str_info;
1.118 schwarze 1097: struct mpage *mpage, *mpage_dest;
1098: struct mlink *mlink, *mlink_dest;
1.69 schwarze 1099: struct mdoc *mdoc;
1100: struct man *man;
1.118 schwarze 1101: char *sodest;
1.112 schwarze 1102: char *cp;
1.162 schwarze 1103: int fd;
1.101 schwarze 1104: unsigned int pslot;
1.69 schwarze 1105: enum mandoclevel lvl;
1106:
1.71 schwarze 1107: str_info.alloc = hash_alloc;
1.152 schwarze 1108: str_info.calloc = hash_calloc;
1109: str_info.free = hash_free;
1.181 ! schwarze 1110: str_info.data = NULL;
1.71 schwarze 1111: str_info.key_offset = offsetof(struct str, key);
1112:
1.171 schwarze 1113: if ( ! nodb)
1.107 schwarze 1114: SQL_EXEC("BEGIN TRANSACTION");
1115:
1.80 schwarze 1116: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1117: while (mpage != NULL) {
1.89 schwarze 1118: mlinks_undupe(mpage);
1.171 schwarze 1119: if (mpage->mlinks == NULL) {
1.89 schwarze 1120: mpage = ohash_next(&mpages, &pslot);
1121: continue;
1.50 kristaps 1122: }
1123:
1.133 schwarze 1124: name_mask = NAME_MASK;
1125: ohash_init(&names, 4, &str_info);
1.71 schwarze 1126: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1127: mparse_reset(mp);
1128: mdoc = NULL;
1129: man = NULL;
1.124 schwarze 1130: sodest = NULL;
1131:
1.170 schwarze 1132: mparse_open(mp, &fd, mpage->mlinks->file);
1.162 schwarze 1133: if (fd == -1) {
1134: say(mpage->mlinks->file, "&open");
1135: goto nextpage;
1.124 schwarze 1136: }
1.14 schwarze 1137:
1138: /*
1.33 schwarze 1139: * Try interpreting the file as mdoc(7) or man(7)
1140: * source code, unless it is already known to be
1141: * formatted. Fall back to formatted mode.
1.14 schwarze 1142: */
1.171 schwarze 1143: if (mpage->mlinks->dform != FORM_CAT ||
1144: mpage->mlinks->fform != FORM_CAT) {
1.162 schwarze 1145: lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1.50 kristaps 1146: if (lvl < MANDOCLEVEL_FATAL)
1.118 schwarze 1147: mparse_result(mp, &mdoc, &man, &sodest);
1.76 schwarze 1148: }
1.14 schwarze 1149:
1.171 schwarze 1150: if (sodest != NULL) {
1.118 schwarze 1151: mlink_dest = ohash_find(&mlinks,
1152: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1153: if (mlink_dest == NULL) {
1154: mandoc_asprintf(&cp, "%s.gz", sodest);
1155: mlink_dest = ohash_find(&mlinks,
1156: ohash_qlookup(&mlinks, cp));
1157: free(cp);
1158: }
1159: if (mlink_dest != NULL) {
1.118 schwarze 1160:
1161: /* The .so target exists. */
1162:
1163: mpage_dest = mlink_dest->mpage;
1164: mlink = mpage->mlinks;
1165: while (1) {
1166: mlink->mpage = mpage_dest;
1167:
1168: /*
1169: * If the target was already
1170: * processed, add the links
1171: * to the database now.
1172: * Otherwise, this will
1173: * happen when we come
1174: * to the target.
1175: */
1176:
1.137 schwarze 1177: if (mpage_dest->pageid)
1.169 schwarze 1178: dbadd_mlink_name(mlink);
1.118 schwarze 1179:
1.171 schwarze 1180: if (mlink->next == NULL)
1.118 schwarze 1181: break;
1182: mlink = mlink->next;
1183: }
1184:
1185: /* Move all links to the target. */
1186:
1187: mlink->next = mlink_dest->next;
1188: mlink_dest->next = mpage->mlinks;
1189: mpage->mlinks = NULL;
1190: }
1.124 schwarze 1191: goto nextpage;
1.171 schwarze 1192: } else if (mdoc != NULL) {
1.82 schwarze 1193: mpage->form = FORM_SRC;
1.155 schwarze 1194: mpage->sec = mdoc_meta(mdoc)->msec;
1195: mpage->sec = mandoc_strdup(
1.171 schwarze 1196: mpage->sec == NULL ? "" : mpage->sec);
1.82 schwarze 1197: mpage->arch = mdoc_meta(mdoc)->arch;
1.84 schwarze 1198: mpage->arch = mandoc_strdup(
1.171 schwarze 1199: mpage->arch == NULL ? "" : mpage->arch);
1.82 schwarze 1200: mpage->title =
1201: mandoc_strdup(mdoc_meta(mdoc)->title);
1.171 schwarze 1202: } else if (man != NULL) {
1.82 schwarze 1203: mpage->form = FORM_SRC;
1204: mpage->sec =
1205: mandoc_strdup(man_meta(man)->msec);
1206: mpage->arch =
1207: mandoc_strdup(mpage->mlinks->arch);
1208: mpage->title =
1209: mandoc_strdup(man_meta(man)->title);
1.76 schwarze 1210: } else {
1.82 schwarze 1211: mpage->form = FORM_CAT;
1212: mpage->sec =
1213: mandoc_strdup(mpage->mlinks->dsec);
1214: mpage->arch =
1215: mandoc_strdup(mpage->mlinks->arch);
1216: mpage->title =
1217: mandoc_strdup(mpage->mlinks->name);
1.76 schwarze 1218: }
1.97 schwarze 1219: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1220: if (*mpage->arch != '\0')
1221: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1222:
1.97 schwarze 1223: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1224: if ('\0' != *mlink->dsec)
1225: putkey(mpage, mlink->dsec, TYPE_sec);
1226: if ('\0' != *mlink->fsec)
1227: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1228: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1229: any : mlink->arch, TYPE_arch);
1.133 schwarze 1230: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1231: }
1.12 schwarze 1232:
1.174 schwarze 1233: assert(mpage->desc == NULL);
1234: if (mdoc != NULL)
1.172 schwarze 1235: parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1.174 schwarze 1236: else if (man != NULL)
1237: parse_man(mpage, man_meta(man), man_node(man));
1.50 kristaps 1238: else
1.162 schwarze 1239: parse_cat(mpage, fd);
1.174 schwarze 1240: if (mpage->desc == NULL)
1.135 schwarze 1241: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1242:
1243: if (warnings && !use_all)
1244: for (mlink = mpage->mlinks; mlink;
1245: mlink = mlink->next)
1246: mlink_check(mpage, mlink);
1.44 kristaps 1247:
1.176 schwarze 1248: dbadd(mpage);
1.124 schwarze 1249:
1250: nextpage:
1.170 schwarze 1251: if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1.162 schwarze 1252: exitcode = (int)MANDOCLEVEL_SYSERR;
1253: say(mpage->mlinks->file, "&wait gunzip");
1.124 schwarze 1254: }
1.66 schwarze 1255: ohash_delete(&strings);
1.133 schwarze 1256: ohash_delete(&names);
1.80 schwarze 1257: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1258: }
1.107 schwarze 1259:
1260: if (0 == nodb)
1261: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1262: }
1263:
1264: static void
1265: names_check(void)
1266: {
1267: sqlite3_stmt *stmt;
1268: const char *name, *sec, *arch, *key;
1269: int irc;
1270:
1271: sqlite3_prepare_v2(db,
1272: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1273: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1274: "WHERE bits & ? AND NOT EXISTS ("
1275: "SELECT pageid FROM mlinks "
1.133 schwarze 1276: "WHERE mlinks.pageid == names.pageid "
1277: "AND mlinks.name == names.name"
1.130 schwarze 1278: ")"
1279: ") JOIN ("
1.142 schwarze 1280: "SELECT sec, arch, name, pageid FROM mlinks "
1281: "GROUP BY pageid"
1.130 schwarze 1282: ") USING (pageid);",
1283: -1, &stmt, NULL);
1284:
1.134 schwarze 1285: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1286: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1287:
1288: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.154 schwarze 1289: name = (const char *)sqlite3_column_text(stmt, 0);
1290: sec = (const char *)sqlite3_column_text(stmt, 1);
1291: arch = (const char *)sqlite3_column_text(stmt, 2);
1292: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1293: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1294: '\0' == *arch ? "" : "/",
1295: '\0' == *arch ? "" : arch, key);
1296: }
1297: sqlite3_finalize(stmt);
1.50 kristaps 1298: }
1.12 schwarze 1299:
1.50 kristaps 1300: static void
1.124 schwarze 1301: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1302: {
1303: FILE *stream;
1304: char *line, *p, *title;
1305: size_t len, plen, titlesz;
1.12 schwarze 1306:
1.124 schwarze 1307: stream = (-1 == fd) ?
1308: fopen(mpage->mlinks->file, "r") :
1309: fdopen(fd, "r");
1310: if (NULL == stream) {
1.159 schwarze 1311: if (-1 != fd)
1312: close(fd);
1.56 schwarze 1313: if (warnings)
1.123 schwarze 1314: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1315: return;
1316: }
1.1 kristaps 1317:
1.50 kristaps 1318: /* Skip to first blank line. */
1.1 kristaps 1319:
1.50 kristaps 1320: while (NULL != (line = fgetln(stream, &len)))
1321: if ('\n' == *line)
1322: break;
1.1 kristaps 1323:
1.50 kristaps 1324: /*
1325: * Assume the first line that is not indented
1326: * is the first section header. Skip to it.
1327: */
1.1 kristaps 1328:
1.50 kristaps 1329: while (NULL != (line = fgetln(stream, &len)))
1330: if ('\n' != *line && ' ' != *line)
1331: break;
1.141 schwarze 1332:
1.50 kristaps 1333: /*
1334: * Read up until the next section into a buffer.
1335: * Strip the leading and trailing newline from each read line,
1336: * appending a trailing space.
1337: * Ignore empty (whitespace-only) lines.
1338: */
1.1 kristaps 1339:
1.50 kristaps 1340: titlesz = 0;
1341: title = NULL;
1.38 schwarze 1342:
1.50 kristaps 1343: while (NULL != (line = fgetln(stream, &len))) {
1344: if (' ' != *line || '\n' != line[len - 1])
1345: break;
1346: while (len > 0 && isspace((unsigned char)*line)) {
1347: line++;
1348: len--;
1349: }
1350: if (1 == len)
1.38 schwarze 1351: continue;
1.50 kristaps 1352: title = mandoc_realloc(title, titlesz + len);
1353: memcpy(title + titlesz, line, len);
1354: titlesz += len;
1355: title[titlesz - 1] = ' ';
1356: }
1.38 schwarze 1357:
1.50 kristaps 1358: /*
1359: * If no page content can be found, or the input line
1360: * is already the next section header, or there is no
1361: * trailing newline, reuse the page title as the page
1362: * description.
1363: */
1.44 kristaps 1364:
1.50 kristaps 1365: if (NULL == title || '\0' == *title) {
1.56 schwarze 1366: if (warnings)
1.82 schwarze 1367: say(mpage->mlinks->file,
1368: "Cannot find NAME section");
1.50 kristaps 1369: fclose(stream);
1370: free(title);
1371: return;
1372: }
1.1 kristaps 1373:
1.50 kristaps 1374: title = mandoc_realloc(title, titlesz + 1);
1375: title[titlesz] = '\0';
1.33 schwarze 1376:
1.50 kristaps 1377: /*
1378: * Skip to the first dash.
1379: * Use the remaining line as the description (no more than 70
1380: * bytes).
1381: */
1.33 schwarze 1382:
1.50 kristaps 1383: if (NULL != (p = strstr(title, "- "))) {
1384: for (p += 2; ' ' == *p || '\b' == *p; p++)
1385: /* Skip to next word. */ ;
1386: } else {
1.56 schwarze 1387: if (warnings)
1.82 schwarze 1388: say(mpage->mlinks->file,
1389: "No dash in title line");
1.50 kristaps 1390: p = title;
1391: }
1.38 schwarze 1392:
1.50 kristaps 1393: plen = strlen(p);
1.1 kristaps 1394:
1.50 kristaps 1395: /* Strip backspace-encoding from line. */
1.1 kristaps 1396:
1.50 kristaps 1397: while (NULL != (line = memchr(p, '\b', plen))) {
1398: len = line - p;
1399: if (0 == len) {
1400: memmove(line, line + 1, plen--);
1401: continue;
1.141 schwarze 1402: }
1.50 kristaps 1403: memmove(line - 1, line + 1, plen - len);
1404: plen -= 2;
1405: }
1.1 kristaps 1406:
1.78 schwarze 1407: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1408: fclose(stream);
1409: free(title);
1410: }
1.1 kristaps 1411:
1.50 kristaps 1412: /*
1413: * Put a type/word pair into the word database for this particular file.
1414: */
1415: static void
1.112 schwarze 1416: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1417: {
1.112 schwarze 1418: char *cp;
1.18 kristaps 1419:
1.50 kristaps 1420: assert(NULL != value);
1.112 schwarze 1421: if (TYPE_arch == type)
1422: for (cp = value; *cp; cp++)
1423: if (isupper((unsigned char)*cp))
1424: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1425: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1426: }
1427:
1428: /*
1.50 kristaps 1429: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1430: */
1431: static void
1.78 schwarze 1432: putmdockey(const struct mpage *mpage,
1433: const struct mdoc_node *n, uint64_t m)
1.3 kristaps 1434: {
1.18 kristaps 1435:
1.50 kristaps 1436: for ( ; NULL != n; n = n->next) {
1437: if (NULL != n->child)
1.78 schwarze 1438: putmdockey(mpage, n->child, m);
1.50 kristaps 1439: if (MDOC_TEXT == n->type)
1.78 schwarze 1440: putkey(mpage, n->string, m);
1.50 kristaps 1441: }
1442: }
1.18 kristaps 1443:
1.61 schwarze 1444: static void
1.174 schwarze 1445: parse_man(struct mpage *mpage, const struct man_meta *meta,
1446: const struct man_node *n)
1.50 kristaps 1447: {
1448: const struct man_node *head, *body;
1.121 schwarze 1449: char *start, *title;
1.50 kristaps 1450: char byte;
1.121 schwarze 1451: size_t sz;
1.18 kristaps 1452:
1.50 kristaps 1453: if (NULL == n)
1.61 schwarze 1454: return;
1.18 kristaps 1455:
1.50 kristaps 1456: /*
1457: * We're only searching for one thing: the first text child in
1458: * the BODY of a NAME section. Since we don't keep track of
1459: * sections in -man, run some hoops to find out whether we're in
1460: * the correct section or not.
1461: */
1.18 kristaps 1462:
1.50 kristaps 1463: if (MAN_BODY == n->type && MAN_SH == n->tok) {
1464: body = n;
1465: assert(body->parent);
1466: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1467: 1 == head->nchild &&
1468: NULL != (head = (head->child)) &&
1469: MAN_TEXT == head->type &&
1470: 0 == strcmp(head->string, "NAME") &&
1471: NULL != body->child) {
1.3 kristaps 1472:
1.50 kristaps 1473: /*
1474: * Suck the entire NAME section into memory.
1475: * Yes, we might run away.
1476: * But too many manuals have big, spread-out
1477: * NAME sections over many lines.
1478: */
1.3 kristaps 1479:
1.121 schwarze 1480: title = NULL;
1481: man_deroff(&title, body);
1.50 kristaps 1482: if (NULL == title)
1.61 schwarze 1483: return;
1.18 kristaps 1484:
1.141 schwarze 1485: /*
1.50 kristaps 1486: * Go through a special heuristic dance here.
1487: * Conventionally, one or more manual names are
1488: * comma-specified prior to a whitespace, then a
1489: * dash, then a description. Try to puzzle out
1490: * the name parts here.
1491: */
1.18 kristaps 1492:
1.121 schwarze 1493: start = title;
1.50 kristaps 1494: for ( ;; ) {
1495: sz = strcspn(start, " ,");
1496: if ('\0' == start[sz])
1497: break;
1.1 kristaps 1498:
1.50 kristaps 1499: byte = start[sz];
1500: start[sz] = '\0';
1.110 schwarze 1501:
1502: /*
1503: * Assume a stray trailing comma in the
1504: * name list if a name begins with a dash.
1505: */
1506:
1507: if ('-' == start[0] ||
1508: ('\\' == start[0] && '-' == start[1]))
1509: break;
1.1 kristaps 1510:
1.133 schwarze 1511: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1512: if ( ! (mpage->name_head_done ||
1513: strcasecmp(start, meta->title))) {
1514: putkey(mpage, start, NAME_HEAD);
1515: mpage->name_head_done = 1;
1516: }
1.1 kristaps 1517:
1.50 kristaps 1518: if (' ' == byte) {
1519: start += sz + 1;
1520: break;
1521: }
1.1 kristaps 1522:
1.50 kristaps 1523: assert(',' == byte);
1524: start += sz + 1;
1525: while (' ' == *start)
1526: start++;
1527: }
1.1 kristaps 1528:
1.121 schwarze 1529: if (start == title) {
1.133 schwarze 1530: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1531: if ( ! (mpage->name_head_done ||
1532: strcasecmp(start, meta->title))) {
1533: putkey(mpage, start, NAME_HEAD);
1534: mpage->name_head_done = 1;
1535: }
1.50 kristaps 1536: free(title);
1.61 schwarze 1537: return;
1.50 kristaps 1538: }
1.1 kristaps 1539:
1.50 kristaps 1540: while (isspace((unsigned char)*start))
1541: start++;
1.1 kristaps 1542:
1.50 kristaps 1543: if (0 == strncmp(start, "-", 1))
1544: start += 1;
1545: else if (0 == strncmp(start, "\\-\\-", 4))
1546: start += 4;
1547: else if (0 == strncmp(start, "\\-", 2))
1548: start += 2;
1549: else if (0 == strncmp(start, "\\(en", 4))
1550: start += 4;
1551: else if (0 == strncmp(start, "\\(em", 4))
1552: start += 4;
1.1 kristaps 1553:
1.50 kristaps 1554: while (' ' == *start)
1555: start++;
1.1 kristaps 1556:
1.78 schwarze 1557: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1558: free(title);
1.61 schwarze 1559: return;
1.50 kristaps 1560: }
1561: }
1.1 kristaps 1562:
1.77 schwarze 1563: for (n = n->child; n; n = n->next) {
1.78 schwarze 1564: if (NULL != mpage->desc)
1.77 schwarze 1565: break;
1.174 schwarze 1566: parse_man(mpage, meta, n);
1.77 schwarze 1567: }
1.1 kristaps 1568: }
1569:
1570: static void
1.172 schwarze 1571: parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1572: const struct mdoc_node *n)
1.1 kristaps 1573: {
1574:
1.50 kristaps 1575: assert(NULL != n);
1576: for (n = n->child; NULL != n; n = n->next) {
1577: switch (n->type) {
1.141 schwarze 1578: case MDOC_ELEM:
1.50 kristaps 1579: /* FALLTHROUGH */
1.141 schwarze 1580: case MDOC_BLOCK:
1.50 kristaps 1581: /* FALLTHROUGH */
1.141 schwarze 1582: case MDOC_HEAD:
1.50 kristaps 1583: /* FALLTHROUGH */
1.141 schwarze 1584: case MDOC_BODY:
1.50 kristaps 1585: /* FALLTHROUGH */
1.141 schwarze 1586: case MDOC_TAIL:
1.50 kristaps 1587: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1588: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1589: break;
1.68 schwarze 1590: if (mdocs[n->tok].mask)
1.78 schwarze 1591: putmdockey(mpage, n->child,
1592: mdocs[n->tok].mask);
1.50 kristaps 1593: break;
1594: default:
1595: assert(MDOC_ROOT != n->type);
1596: continue;
1597: }
1598: if (NULL != n->child)
1.172 schwarze 1599: parse_mdoc(mpage, meta, n);
1.1 kristaps 1600: }
1601: }
1602:
1.25 schwarze 1603: static int
1.172 schwarze 1604: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1605: const struct mdoc_node *n)
1.1 kristaps 1606: {
1.176 schwarze 1607: char *start, *end;
1.1 kristaps 1608: size_t sz;
1.25 schwarze 1609:
1.50 kristaps 1610: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1611: NULL == (n = n->child) ||
1612: MDOC_TEXT != n->type)
1.25 schwarze 1613: return(0);
1.1 kristaps 1614:
1615: /*
1616: * Only consider those `Fd' macro fields that begin with an
1617: * "inclusion" token (versus, e.g., #define).
1618: */
1.50 kristaps 1619:
1.1 kristaps 1620: if (strcmp("#include", n->string))
1.25 schwarze 1621: return(0);
1.1 kristaps 1622:
1623: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.25 schwarze 1624: return(0);
1.1 kristaps 1625:
1626: /*
1627: * Strip away the enclosing angle brackets and make sure we're
1628: * not zero-length.
1629: */
1630:
1631: start = n->string;
1632: if ('<' == *start || '"' == *start)
1633: start++;
1634:
1635: if (0 == (sz = strlen(start)))
1.25 schwarze 1636: return(0);
1.1 kristaps 1637:
1638: end = &start[(int)sz - 1];
1639: if ('>' == *end || '"' == *end)
1640: end--;
1641:
1.50 kristaps 1642: if (end > start)
1.78 schwarze 1643: putkeys(mpage, start, end - start + 1, TYPE_In);
1.92 schwarze 1644: return(0);
1.1 kristaps 1645: }
1646:
1.178 schwarze 1647: static void
1648: parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1.1 kristaps 1649: {
1.112 schwarze 1650: char *cp;
1.178 schwarze 1651: size_t sz;
1.1 kristaps 1652:
1.178 schwarze 1653: if (n->type != MDOC_TEXT)
1654: return;
1.25 schwarze 1655:
1.178 schwarze 1656: /* Skip function pointer punctuation. */
1.1 kristaps 1657:
1.178 schwarze 1658: cp = n->string;
1659: while (*cp == '(' || *cp == '*')
1.1 kristaps 1660: cp++;
1.178 schwarze 1661: sz = strcspn(cp, "()");
1.1 kristaps 1662:
1.178 schwarze 1663: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1664: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1665: putkeys(mpage, cp, sz, NAME_SYN);
1666: }
1.25 schwarze 1667:
1.178 schwarze 1668: static int
1669: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1670: const struct mdoc_node *n)
1671: {
1672:
1673: if (n->child == NULL)
1674: return(0);
1675:
1676: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1677:
1.178 schwarze 1678: for (n = n->child->next; n != NULL; n = n->next)
1679: if (n->type == MDOC_TEXT)
1.78 schwarze 1680: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1681:
1.173 schwarze 1682: return(0);
1683: }
1684:
1685: static int
1686: parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1687: const struct mdoc_node *n)
1688: {
1.177 schwarze 1689:
1690: if (n->type != MDOC_HEAD)
1691: return(1);
1.173 schwarze 1692:
1.178 schwarze 1693: if (n->child != NULL)
1694: parse_mdoc_fname(mpage, n->child);
1695:
1.25 schwarze 1696: return(0);
1.1 kristaps 1697: }
1698:
1.25 schwarze 1699: static int
1.172 schwarze 1700: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1701: const struct mdoc_node *n)
1.1 kristaps 1702: {
1.67 schwarze 1703: char *cp;
1.1 kristaps 1704:
1705: if (NULL == (n = n->child))
1.25 schwarze 1706: return(0);
1.1 kristaps 1707:
1.67 schwarze 1708: if (NULL == n->next) {
1.78 schwarze 1709: putkey(mpage, n->string, TYPE_Xr);
1.67 schwarze 1710: return(0);
1711: }
1712:
1.120 schwarze 1713: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1714: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1715: free(cp);
1716: return(0);
1.1 kristaps 1717: }
1718:
1.25 schwarze 1719: static int
1.172 schwarze 1720: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1721: const struct mdoc_node *n)
1.1 kristaps 1722: {
1723:
1.122 schwarze 1724: if (MDOC_BODY == n->type)
1725: mdoc_deroff(&mpage->desc, n);
1726: return(0);
1.1 kristaps 1727: }
1728:
1.25 schwarze 1729: static int
1.172 schwarze 1730: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1731: const struct mdoc_node *n)
1.1 kristaps 1732: {
1733:
1.129 schwarze 1734: if (SEC_NAME == n->sec)
1.133 schwarze 1735: putmdockey(mpage, n->child, NAME_TITLE);
1.172 schwarze 1736: else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1737: if (n->child == NULL)
1738: putkey(mpage, meta->name, NAME_SYN);
1739: else
1740: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1741: }
1742: if ( ! (mpage->name_head_done ||
1743: n->child == NULL || n->child->string == NULL ||
1744: strcasecmp(n->child->string, meta->title))) {
1745: putkey(mpage, n->child->string, NAME_HEAD);
1746: mpage->name_head_done = 1;
1.172 schwarze 1747: }
1.129 schwarze 1748: return(0);
1.1 kristaps 1749: }
1750:
1.25 schwarze 1751: static int
1.172 schwarze 1752: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1753: const struct mdoc_node *n)
1.1 kristaps 1754: {
1755:
1.25 schwarze 1756: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 kristaps 1757: }
1758:
1.50 kristaps 1759: static int
1.172 schwarze 1760: parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1761: const struct mdoc_node *n)
1.1 kristaps 1762: {
1763:
1.50 kristaps 1764: return(MDOC_HEAD == n->type);
1765: }
1.1 kristaps 1766:
1.50 kristaps 1767: static int
1.172 schwarze 1768: parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1769: const struct mdoc_node *n)
1.50 kristaps 1770: {
1.1 kristaps 1771:
1.50 kristaps 1772: return(MDOC_BODY == n->type);
1.1 kristaps 1773: }
1774:
1.50 kristaps 1775: /*
1.66 schwarze 1776: * Add a string to the hash table for the current manual.
1777: * Each string has a bitmask telling which macros it belongs to.
1778: * When we finish the manual, we'll dump the table.
1.50 kristaps 1779: */
1780: static void
1.176 schwarze 1781: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1782: {
1.133 schwarze 1783: struct ohash *htab;
1.50 kristaps 1784: struct str *s;
1.111 schwarze 1785: const char *end;
1.65 schwarze 1786: unsigned int slot;
1.176 schwarze 1787: int i, mustfree;
1.25 schwarze 1788:
1.50 kristaps 1789: if (0 == sz)
1790: return;
1.111 schwarze 1791:
1.176 schwarze 1792: mustfree = render_string(&cp, &sz);
1793:
1.133 schwarze 1794: if (TYPE_Nm & v) {
1795: htab = &names;
1796: v &= name_mask;
1.169 schwarze 1797: if (v & NAME_FIRST)
1798: name_mask &= ~NAME_FIRST;
1.133 schwarze 1799: if (debug > 1)
1800: say(mpage->mlinks->file,
1.175 schwarze 1801: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1802: } else {
1803: htab = &strings;
1804: if (debug > 1)
1805: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1806: if ((uint64_t)1 << i & v)
1.133 schwarze 1807: say(mpage->mlinks->file,
1808: "Adding key %s=%*s",
1809: mansearch_keynames[i], sz, cp);
1.111 schwarze 1810: }
1.25 schwarze 1811:
1.65 schwarze 1812: end = cp + sz;
1.133 schwarze 1813: slot = ohash_qlookupi(htab, cp, &end);
1814: s = ohash_find(htab, slot);
1.25 schwarze 1815:
1.78 schwarze 1816: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1817: s->mask |= v;
1818: return;
1819: } else if (NULL == s) {
1.144 schwarze 1820: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1821: memcpy(s->key, cp, sz);
1.133 schwarze 1822: ohash_insert(htab, slot, s);
1.1 kristaps 1823: }
1.78 schwarze 1824: s->mpage = mpage;
1.50 kristaps 1825: s->mask = v;
1.176 schwarze 1826:
1827: if (mustfree)
1828: free(cp);
1.1 kristaps 1829: }
1830:
1.50 kristaps 1831: /*
1832: * Take a Unicode codepoint and produce its UTF-8 encoding.
1833: * This isn't the best way to do this, but it works.
1834: * The magic numbers are from the UTF-8 packaging.
1835: * They're not as scary as they seem: read the UTF-8 spec for details.
1836: */
1837: static size_t
1838: utf8(unsigned int cp, char out[7])
1.1 kristaps 1839: {
1.50 kristaps 1840: size_t rc;
1.1 kristaps 1841:
1.50 kristaps 1842: rc = 0;
1843: if (cp <= 0x0000007F) {
1844: rc = 1;
1845: out[0] = (char)cp;
1846: } else if (cp <= 0x000007FF) {
1847: rc = 2;
1848: out[0] = (cp >> 6 & 31) | 192;
1849: out[1] = (cp & 63) | 128;
1850: } else if (cp <= 0x0000FFFF) {
1851: rc = 3;
1852: out[0] = (cp >> 12 & 15) | 224;
1853: out[1] = (cp >> 6 & 63) | 128;
1854: out[2] = (cp & 63) | 128;
1855: } else if (cp <= 0x001FFFFF) {
1856: rc = 4;
1857: out[0] = (cp >> 18 & 7) | 240;
1858: out[1] = (cp >> 12 & 63) | 128;
1859: out[2] = (cp >> 6 & 63) | 128;
1860: out[3] = (cp & 63) | 128;
1861: } else if (cp <= 0x03FFFFFF) {
1862: rc = 5;
1863: out[0] = (cp >> 24 & 3) | 248;
1864: out[1] = (cp >> 18 & 63) | 128;
1865: out[2] = (cp >> 12 & 63) | 128;
1866: out[3] = (cp >> 6 & 63) | 128;
1867: out[4] = (cp & 63) | 128;
1868: } else if (cp <= 0x7FFFFFFF) {
1869: rc = 6;
1870: out[0] = (cp >> 30 & 1) | 252;
1871: out[1] = (cp >> 24 & 63) | 128;
1872: out[2] = (cp >> 18 & 63) | 128;
1873: out[3] = (cp >> 12 & 63) | 128;
1874: out[4] = (cp >> 6 & 63) | 128;
1875: out[5] = (cp & 63) | 128;
1876: } else
1.1 kristaps 1877: return(0);
1878:
1.50 kristaps 1879: out[rc] = '\0';
1880: return(rc);
1881: }
1.1 kristaps 1882:
1.50 kristaps 1883: /*
1.176 schwarze 1884: * If the string contains escape sequences,
1885: * replace it with an allocated rendering and return 1,
1886: * such that the caller can free it after use.
1887: * Otherwise, do nothing and return 0.
1.50 kristaps 1888: */
1.176 schwarze 1889: static int
1890: render_string(char **public, size_t *psz)
1.50 kristaps 1891: {
1.176 schwarze 1892: const char *src, *scp, *addcp, *seq;
1893: char *dst;
1894: size_t ssz, dsz, addsz;
1.114 schwarze 1895: char utfbuf[7], res[6];
1.176 schwarze 1896: int seqlen, unicode;
1.50 kristaps 1897:
1898: res[0] = '\\';
1899: res[1] = '\t';
1900: res[2] = ASCII_NBRSP;
1901: res[3] = ASCII_HYPH;
1.114 schwarze 1902: res[4] = ASCII_BREAK;
1903: res[5] = '\0';
1.1 kristaps 1904:
1.176 schwarze 1905: src = scp = *public;
1906: ssz = *psz;
1907: dst = NULL;
1908: dsz = 0;
1909:
1910: while (scp < src + *psz) {
1911:
1912: /* Leave normal characters unchanged. */
1913:
1914: if (strchr(res, *scp) == NULL) {
1915: if (dst != NULL)
1916: dst[dsz++] = *scp;
1917: scp++;
1918: continue;
1919: }
1.46 kristaps 1920:
1.50 kristaps 1921: /*
1.176 schwarze 1922: * Found something that requires replacing,
1923: * make sure we have a destination buffer.
1.50 kristaps 1924: */
1.176 schwarze 1925:
1926: if (dst == NULL) {
1927: dst = mandoc_malloc(ssz + 1);
1928: dsz = scp - src;
1929: memcpy(dst, src, dsz);
1.50 kristaps 1930: }
1.46 kristaps 1931:
1.176 schwarze 1932: /* Handle single-char special characters. */
1933:
1934: switch (*scp) {
1935: case '\\':
1936: break;
1.141 schwarze 1937: case '\t':
1.114 schwarze 1938: /* FALLTHROUGH */
1.141 schwarze 1939: case ASCII_NBRSP:
1.176 schwarze 1940: dst[dsz++] = ' ';
1941: scp++;
1942: continue;
1943: case ASCII_HYPH:
1944: dst[dsz++] = '-';
1.114 schwarze 1945: /* FALLTHROUGH */
1.141 schwarze 1946: case ASCII_BREAK:
1.176 schwarze 1947: scp++;
1.50 kristaps 1948: continue;
1.114 schwarze 1949: default:
1.176 schwarze 1950: abort();
1.114 schwarze 1951: }
1.46 kristaps 1952:
1.50 kristaps 1953: /*
1.176 schwarze 1954: * Found an escape sequence.
1955: * Read past the slash, then parse it.
1956: * Ignore everything except characters.
1.50 kristaps 1957: */
1.95 schwarze 1958:
1.176 schwarze 1959: scp++;
1960: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1961: continue;
1.1 kristaps 1962:
1.50 kristaps 1963: /*
1.95 schwarze 1964: * Render the special character
1965: * as either UTF-8 or ASCII.
1.50 kristaps 1966: */
1.95 schwarze 1967:
1968: if (write_utf8) {
1.176 schwarze 1969: unicode = mchars_spec2cp(mchars, seq, seqlen);
1970: if (unicode <= 0)
1.95 schwarze 1971: continue;
1.176 schwarze 1972: addsz = utf8(unicode, utfbuf);
1973: if (addsz == 0)
1.95 schwarze 1974: continue;
1.176 schwarze 1975: addcp = utfbuf;
1.95 schwarze 1976: } else {
1.176 schwarze 1977: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1978: if (addcp == NULL)
1.95 schwarze 1979: continue;
1.176 schwarze 1980: if (*addcp == ASCII_NBRSP) {
1981: addcp = " ";
1982: addsz = 1;
1.95 schwarze 1983: }
1984: }
1.1 kristaps 1985:
1.50 kristaps 1986: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1987:
1.176 schwarze 1988: ssz += addsz;
1989: dst = mandoc_realloc(dst, ssz + 1);
1990: memcpy(dst + dsz, addcp, addsz);
1991: dsz += addsz;
1992: }
1993: if (dst != NULL) {
1994: *public = dst;
1995: *psz = dsz;
1996: }
1997:
1998: /* Trim trailing whitespace and NUL-terminate. */
1999:
2000: while (*psz > 0 && (*public)[*psz - 1] == ' ')
2001: --*psz;
2002: if (dst != NULL) {
2003: (*public)[*psz] = '\0';
2004: return(1);
2005: } else
2006: return(0);
1.1 kristaps 2007: }
2008:
1.118 schwarze 2009: static void
2010: dbadd_mlink(const struct mlink *mlink)
2011: {
2012: size_t i;
2013:
2014: i = 1;
2015: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2016: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2017: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2018: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2019: SQL_STEP(stmts[STMT_INSERT_LINK]);
2020: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2021: }
2022:
2023: static void
2024: dbadd_mlink_name(const struct mlink *mlink)
2025: {
1.175 schwarze 2026: uint64_t bits;
1.169 schwarze 2027: size_t i;
2028:
2029: dbadd_mlink(mlink);
1.160 schwarze 2030:
2031: i = 1;
1.175 schwarze 2032: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2033: bits = NAME_FILE & NAME_MASK;
2034: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2035: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2036: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2037: }
2038:
2039: i = 1;
2040: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2041: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2042: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2043: SQL_STEP(stmts[STMT_INSERT_NAME]);
2044: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2045: }
2046:
1.14 schwarze 2047: /*
1.50 kristaps 2048: * Flush the current page's terms (and their bits) into the database.
2049: * Wrap the entire set of additions in a transaction to make sqlite be a
2050: * little faster.
1.96 schwarze 2051: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2052: */
2053: static void
1.176 schwarze 2054: dbadd(struct mpage *mpage)
1.14 schwarze 2055: {
1.87 schwarze 2056: struct mlink *mlink;
1.50 kristaps 2057: struct str *key;
1.176 schwarze 2058: char *cp;
1.52 kristaps 2059: size_t i;
1.66 schwarze 2060: unsigned int slot;
1.176 schwarze 2061: int mustfree;
1.14 schwarze 2062:
1.128 schwarze 2063: mlink = mpage->mlinks;
1.43 kristaps 2064:
1.128 schwarze 2065: if (nodb) {
1.147 schwarze 2066: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2067: key = ohash_next(&names, &slot))
1.147 schwarze 2068: free(key);
2069: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2070: key = ohash_next(&strings, &slot))
1.147 schwarze 2071: free(key);
1.145 schwarze 2072: if (0 == debug)
2073: return;
1.128 schwarze 2074: while (NULL != mlink) {
2075: fputs(mlink->name, stdout);
2076: if (NULL == mlink->next ||
2077: strcmp(mlink->dsec, mlink->next->dsec) ||
2078: strcmp(mlink->fsec, mlink->next->fsec) ||
2079: strcmp(mlink->arch, mlink->next->arch)) {
2080: putchar('(');
2081: if ('\0' == *mlink->dsec)
2082: fputs(mlink->fsec, stdout);
2083: else
2084: fputs(mlink->dsec, stdout);
2085: if ('\0' != *mlink->arch)
2086: printf("/%s", mlink->arch);
2087: putchar(')');
2088: }
2089: mlink = mlink->next;
2090: if (NULL != mlink)
2091: fputs(", ", stdout);
2092: }
1.132 schwarze 2093: printf(" - %s\n", mpage->desc);
1.14 schwarze 2094: return;
1.128 schwarze 2095: }
2096:
2097: if (debug)
2098: say(mlink->file, "Adding to database");
1.28 kristaps 2099:
1.176 schwarze 2100: cp = mpage->desc;
2101: i = strlen(cp);
2102: mustfree = render_string(&cp, &i);
1.52 kristaps 2103: i = 1;
1.176 schwarze 2104: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2105: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2106: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2107: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2108: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2109: if (mustfree)
2110: free(cp);
1.81 schwarze 2111:
1.128 schwarze 2112: while (NULL != mlink) {
1.118 schwarze 2113: dbadd_mlink(mlink);
1.128 schwarze 2114: mlink = mlink->next;
2115: }
1.134 schwarze 2116: mlink = mpage->mlinks;
1.50 kristaps 2117:
1.133 schwarze 2118: for (key = ohash_first(&names, &slot); NULL != key;
2119: key = ohash_next(&names, &slot)) {
2120: assert(key->mpage == mpage);
2121: i = 1;
2122: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2123: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2124: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2125: SQL_STEP(stmts[STMT_INSERT_NAME]);
2126: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2127: free(key);
2128: }
1.66 schwarze 2129: for (key = ohash_first(&strings, &slot); NULL != key;
2130: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2131: assert(key->mpage == mpage);
1.52 kristaps 2132: i = 1;
2133: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2134: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2135: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2136: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2137: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2138: free(key);
1.38 schwarze 2139: }
1.14 schwarze 2140: }
2141:
1.5 kristaps 2142: static void
1.59 schwarze 2143: dbprune(void)
1.5 kristaps 2144: {
1.78 schwarze 2145: struct mpage *mpage;
1.82 schwarze 2146: struct mlink *mlink;
1.52 kristaps 2147: size_t i;
1.80 schwarze 2148: unsigned int slot;
1.5 kristaps 2149:
1.106 schwarze 2150: if (0 == nodb)
2151: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2152:
1.106 schwarze 2153: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2154: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2155: mlink = mpage->mlinks;
1.125 schwarze 2156: if (debug)
1.106 schwarze 2157: say(mlink->file, "Deleting from database");
2158: if (nodb)
2159: continue;
2160: for ( ; NULL != mlink; mlink = mlink->next) {
2161: i = 1;
2162: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2163: i, mlink->dsec);
2164: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2165: i, mlink->arch);
2166: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2167: i, mlink->name);
2168: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2169: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2170: }
1.5 kristaps 2171: }
1.106 schwarze 2172:
2173: if (0 == nodb)
2174: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2175: }
2176:
1.4 kristaps 2177: /*
1.50 kristaps 2178: * Close an existing database and its prepared statements.
2179: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2180: */
1.35 kristaps 2181: static void
1.59 schwarze 2182: dbclose(int real)
1.4 kristaps 2183: {
1.50 kristaps 2184: size_t i;
1.115 schwarze 2185: int status;
2186: pid_t child;
1.4 kristaps 2187:
1.50 kristaps 2188: if (nodb)
1.38 schwarze 2189: return;
1.50 kristaps 2190:
2191: for (i = 0; i < STMT__MAX; i++) {
2192: sqlite3_finalize(stmts[i]);
2193: stmts[i] = NULL;
1.4 kristaps 2194: }
2195:
1.50 kristaps 2196: sqlite3_close(db);
2197: db = NULL;
1.12 schwarze 2198:
1.50 kristaps 2199: if (real)
2200: return;
1.12 schwarze 2201:
1.115 schwarze 2202: if ('\0' == *tempfilename) {
2203: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2204: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2205: say(MANDOC_DB, "&rename");
1.115 schwarze 2206: }
2207: return;
2208: }
2209:
2210: switch (child = fork()) {
1.141 schwarze 2211: case -1:
1.115 schwarze 2212: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2213: say("", "&fork cmp");
1.115 schwarze 2214: return;
1.141 schwarze 2215: case 0:
1.115 schwarze 2216: execlp("cmp", "cmp", "-s",
2217: tempfilename, MANDOC_DB, NULL);
1.123 schwarze 2218: say("", "&exec cmp");
1.115 schwarze 2219: exit(0);
2220: default:
2221: break;
2222: }
2223: if (-1 == waitpid(child, &status, 0)) {
2224: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2225: say("", "&wait cmp");
1.115 schwarze 2226: } else if (WIFSIGNALED(status)) {
2227: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2228: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2229: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2230: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2231: say(MANDOC_DB,
2232: "Data changed, but cannot replace database");
2233: }
2234:
2235: *strrchr(tempfilename, '/') = '\0';
2236: switch (child = fork()) {
1.141 schwarze 2237: case -1:
1.115 schwarze 2238: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2239: say("", "&fork rm");
1.115 schwarze 2240: return;
1.141 schwarze 2241: case 0:
1.115 schwarze 2242: execlp("rm", "rm", "-rf", tempfilename, NULL);
1.123 schwarze 2243: say("", "&exec rm");
1.115 schwarze 2244: exit((int)MANDOCLEVEL_SYSERR);
2245: default:
2246: break;
2247: }
2248: if (-1 == waitpid(child, &status, 0)) {
2249: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2250: say("", "&wait rm");
1.115 schwarze 2251: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2252: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2253: say("", "%s: Cannot remove temporary directory",
2254: tempfilename);
1.59 schwarze 2255: }
1.50 kristaps 2256: }
1.14 schwarze 2257:
1.50 kristaps 2258: /*
2259: * This is straightforward stuff.
2260: * Open a database connection to a "temporary" database, then open a set
2261: * of prepared statements we'll use over and over again.
2262: * If "real" is set, we use the existing database; if not, we truncate a
2263: * temporary one.
2264: * Must be matched by dbclose().
2265: */
2266: static int
1.59 schwarze 2267: dbopen(int real)
1.50 kristaps 2268: {
1.115 schwarze 2269: const char *sql;
1.50 kristaps 2270: int rc, ofl;
1.12 schwarze 2271:
1.141 schwarze 2272: if (nodb)
1.50 kristaps 2273: return(1);
1.12 schwarze 2274:
1.115 schwarze 2275: *tempfilename = '\0';
1.63 schwarze 2276: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2277:
2278: if (real) {
2279: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2280: if (SQLITE_OK != rc) {
1.63 schwarze 2281: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2282: if (SQLITE_CANTOPEN != rc)
2283: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.63 schwarze 2284: return(0);
2285: }
1.115 schwarze 2286: goto prepare_statements;
2287: }
2288:
2289: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2290:
1.115 schwarze 2291: remove(MANDOC_DB "~");
2292: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2293: if (SQLITE_OK == rc)
1.115 schwarze 2294: goto create_tables;
1.116 schwarze 2295: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2296: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2297: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.50 kristaps 2298: return(0);
2299: }
1.12 schwarze 2300:
1.143 schwarze 2301: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2302: sizeof(tempfilename));
1.115 schwarze 2303: if (NULL == mkdtemp(tempfilename)) {
2304: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2305: say("", "&%s", tempfilename);
1.115 schwarze 2306: return(0);
2307: }
1.143 schwarze 2308: (void)strlcat(tempfilename, "/" MANDOC_DB,
2309: sizeof(tempfilename));
1.115 schwarze 2310: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2311: if (SQLITE_OK != rc) {
1.59 schwarze 2312: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2313: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.50 kristaps 2314: return(0);
2315: }
1.12 schwarze 2316:
1.115 schwarze 2317: create_tables:
1.81 schwarze 2318: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2319: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2320: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2321: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2322: ");\n"
2323: "\n"
1.81 schwarze 2324: "CREATE TABLE \"mlinks\" (\n"
2325: " \"sec\" TEXT NOT NULL,\n"
2326: " \"arch\" TEXT NOT NULL,\n"
2327: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2328: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2329: "ON DELETE CASCADE\n"
1.81 schwarze 2330: ");\n"
1.136 schwarze 2331: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2332: "\n"
1.133 schwarze 2333: "CREATE TABLE \"names\" (\n"
2334: " \"bits\" INTEGER NOT NULL,\n"
2335: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2336: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2337: "ON DELETE CASCADE,\n"
2338: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2339: ");\n"
2340: "\n"
1.50 kristaps 2341: "CREATE TABLE \"keys\" (\n"
2342: " \"bits\" INTEGER NOT NULL,\n"
2343: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2344: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2345: "ON DELETE CASCADE\n"
1.136 schwarze 2346: ");\n"
2347: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2348:
1.50 kristaps 2349: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2350: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2351: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2352: sqlite3_close(db);
1.50 kristaps 2353: return(0);
2354: }
1.4 kristaps 2355:
1.57 schwarze 2356: prepare_statements:
1.146 schwarze 2357: if (SQLITE_OK != sqlite3_exec(db,
2358: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2359: exitcode = (int)MANDOCLEVEL_SYSERR;
2360: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2361: sqlite3_errmsg(db));
2362: sqlite3_close(db);
2363: return(0);
2364: }
2365:
1.137 schwarze 2366: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2367: "(SELECT pageid FROM mlinks WHERE "
2368: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2369: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2370: sql = "INSERT INTO mpages "
1.132 schwarze 2371: "(desc,form) VALUES (?,?)";
1.81 schwarze 2372: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2373: sql = "INSERT INTO mlinks "
1.104 schwarze 2374: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2375: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2376: sql = "SELECT bits FROM names where pageid = ?";
2377: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2378: sql = "INSERT INTO names "
2379: "(bits,name,pageid) VALUES (?,?,?)";
2380: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2381: sql = "INSERT INTO keys "
1.81 schwarze 2382: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2383: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2384:
2385: #ifndef __APPLE__
2386: /*
2387: * When opening a new database, we can turn off
2388: * synchronous mode for much better performance.
2389: */
2390:
1.146 schwarze 2391: if (real && SQLITE_OK != sqlite3_exec(db,
2392: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2393: exitcode = (int)MANDOCLEVEL_SYSERR;
2394: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2395: sqlite3_errmsg(db));
1.146 schwarze 2396: sqlite3_close(db);
2397: return(0);
2398: }
1.70 schwarze 2399: #endif
2400:
1.50 kristaps 2401: return(1);
2402: }
1.5 kristaps 2403:
1.50 kristaps 2404: static void *
1.152 schwarze 2405: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2406: {
1.12 schwarze 2407:
1.152 schwarze 2408: return(mandoc_calloc(n, sz));
1.50 kristaps 2409: }
1.12 schwarze 2410:
1.50 kristaps 2411: static void *
2412: hash_alloc(size_t sz, void *arg)
2413: {
1.5 kristaps 2414:
1.50 kristaps 2415: return(mandoc_malloc(sz));
2416: }
1.41 kristaps 2417:
1.50 kristaps 2418: static void
1.152 schwarze 2419: hash_free(void *p, void *arg)
1.50 kristaps 2420: {
1.4 kristaps 2421:
1.50 kristaps 2422: free(p);
1.4 kristaps 2423: }
2424:
1.50 kristaps 2425: static int
1.165 schwarze 2426: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2427: {
1.59 schwarze 2428: static char startdir[PATH_MAX];
1.151 schwarze 2429: static int getcwd_status; /* 1 = ok, 2 = failure */
2430: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2431: char *cp;
1.4 kristaps 2432:
1.59 schwarze 2433: /*
1.151 schwarze 2434: * Remember the original working directory, if possible.
2435: * This will be needed if the second or a later directory
2436: * on the command line is given as a relative path.
2437: * Do not error out if the current directory is not
2438: * searchable: Maybe it won't be needed after all.
2439: */
2440: if (0 == getcwd_status) {
2441: if (NULL == getcwd(startdir, sizeof(startdir))) {
2442: getcwd_status = 2;
2443: (void)strlcpy(startdir, strerror(errno),
2444: sizeof(startdir));
2445: } else
2446: getcwd_status = 1;
2447: }
2448:
2449: /*
2450: * We are leaving the old base directory.
2451: * Do not use it any longer, not even for messages.
2452: */
2453: *basedir = '\0';
2454:
2455: /*
2456: * If and only if the directory was changed earlier and
2457: * the next directory to process is given as a relative path,
2458: * first go back, or bail out if that is impossible.
1.59 schwarze 2459: */
1.151 schwarze 2460: if (chdir_status && '/' != *targetdir) {
2461: if (2 == getcwd_status) {
1.59 schwarze 2462: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2463: say("", "getcwd: %s", startdir);
1.59 schwarze 2464: return(0);
2465: }
1.151 schwarze 2466: if (-1 == chdir(startdir)) {
1.59 schwarze 2467: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2468: say("", "&chdir %s", startdir);
1.59 schwarze 2469: return(0);
2470: }
2471: }
1.151 schwarze 2472:
2473: /*
2474: * Always resolve basedir to the canonicalized absolute
2475: * pathname and append a trailing slash, such that
2476: * we can reliably check whether files are inside.
2477: */
1.59 schwarze 2478: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2479: if (report_baddir || errno != ENOENT) {
2480: exitcode = (int)MANDOCLEVEL_BADARG;
2481: say("", "&%s: realpath", targetdir);
2482: }
1.50 kristaps 2483: return(0);
1.59 schwarze 2484: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2485: if (report_baddir || errno != ENOENT) {
2486: exitcode = (int)MANDOCLEVEL_BADARG;
2487: say("", "&chdir");
2488: }
1.50 kristaps 2489: return(0);
1.150 schwarze 2490: }
1.151 schwarze 2491: chdir_status = 1;
1.150 schwarze 2492: cp = strchr(basedir, '\0');
2493: if ('/' != cp[-1]) {
2494: if (cp - basedir >= PATH_MAX - 1) {
2495: exitcode = (int)MANDOCLEVEL_SYSERR;
2496: say("", "Filename too long");
2497: return(0);
2498: }
2499: *cp++ = '/';
2500: *cp = '\0';
1.4 kristaps 2501: }
1.50 kristaps 2502: return(1);
1.56 schwarze 2503: }
2504:
2505: static void
1.59 schwarze 2506: say(const char *file, const char *format, ...)
1.56 schwarze 2507: {
2508: va_list ap;
1.123 schwarze 2509: int use_errno;
1.56 schwarze 2510:
1.59 schwarze 2511: if ('\0' != *basedir)
2512: fprintf(stderr, "%s", basedir);
2513: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2514: fputc('/', stderr);
1.56 schwarze 2515: if ('\0' != *file)
1.59 schwarze 2516: fprintf(stderr, "%s", file);
2517:
1.123 schwarze 2518: use_errno = 1;
2519: if (NULL != format) {
2520: switch (*format) {
1.141 schwarze 2521: case '&':
1.123 schwarze 2522: format++;
2523: break;
1.141 schwarze 2524: case '\0':
1.123 schwarze 2525: format = NULL;
2526: break;
2527: default:
2528: use_errno = 0;
2529: break;
2530: }
2531: }
2532: if (NULL != format) {
2533: if ('\0' != *basedir || '\0' != *file)
2534: fputs(": ", stderr);
2535: va_start(ap, format);
2536: vfprintf(stderr, format, ap);
2537: va_end(ap);
2538: }
2539: if (use_errno) {
2540: if ('\0' != *basedir || '\0' != *file || NULL != format)
2541: fputs(": ", stderr);
1.59 schwarze 2542: perror(NULL);
1.123 schwarze 2543: } else
2544: fputc('\n', stderr);
1.1 kristaps 2545: }
CVSweb