Annotation of mandoc/mandocdb.c, Revision 1.201
1.201 ! schwarze 1: /* $Id: mandocdb.c,v 1.200 2015/10/12 00:32:55 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.182 schwarze 4: * Copyright (c) 2011-2015 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: *
1.187 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.187 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 kristaps 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>
1.198 schwarze 26: #include <err.h>
1.63 schwarze 27: #include <errno.h>
1.1 kristaps 28: #include <fcntl.h>
1.158 schwarze 29: #if HAVE_FTS
1.50 kristaps 30: #include <fts.h>
1.157 schwarze 31: #else
32: #include "compat_fts.h"
33: #endif
1.1 kristaps 34: #include <getopt.h>
1.58 schwarze 35: #include <limits.h>
1.50 kristaps 36: #include <stddef.h>
1.59 schwarze 37: #include <stdio.h>
1.1 kristaps 38: #include <stdint.h>
39: #include <stdlib.h>
40: #include <string.h>
1.17 schwarze 41: #include <unistd.h>
1.1 kristaps 42:
1.158 schwarze 43: #if HAVE_OHASH
1.50 kristaps 44: #include <ohash.h>
1.53 kristaps 45: #else
46: #include "compat_ohash.h"
47: #endif
1.50 kristaps 48: #include <sqlite3.h>
1.1 kristaps 49:
1.188 schwarze 50: #include "mandoc_aux.h"
51: #include "mandoc.h"
52: #include "roff.h"
1.50 kristaps 53: #include "mdoc.h"
1.1 kristaps 54: #include "man.h"
1.187 schwarze 55: #include "manconf.h"
1.55 kristaps 56: #include "mansearch.h"
1.1 kristaps 57:
1.111 schwarze 58: extern int mansearch_keymax;
59: extern const char *const mansearch_keynames[];
60:
1.52 kristaps 61: #define SQL_EXEC(_v) \
62: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
1.134 schwarze 63: say("", "%s: %s", (_v), sqlite3_errmsg(db))
1.52 kristaps 64: #define SQL_BIND_TEXT(_s, _i, _v) \
65: if (SQLITE_OK != sqlite3_bind_text \
66: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
1.134 schwarze 67: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 68: #define SQL_BIND_INT(_s, _i, _v) \
69: if (SQLITE_OK != sqlite3_bind_int \
70: ((_s), (_i)++, (_v))) \
1.134 schwarze 71: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 72: #define SQL_BIND_INT64(_s, _i, _v) \
73: if (SQLITE_OK != sqlite3_bind_int64 \
74: ((_s), (_i)++, (_v))) \
1.134 schwarze 75: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 76: #define SQL_STEP(_s) \
77: if (SQLITE_DONE != sqlite3_step((_s))) \
1.134 schwarze 78: say(mlink->file, "%s", sqlite3_errmsg(db))
1.52 kristaps 79:
1.50 kristaps 80: enum op {
81: OP_DEFAULT = 0, /* new dbs from dir list or default config */
82: OP_CONFFILE, /* new databases from custom config file */
83: OP_UPDATE, /* delete/add entries in existing database */
84: OP_DELETE, /* delete entries from existing database */
85: OP_TEST /* change no databases, report potential problems */
1.38 schwarze 86: };
87:
1.50 kristaps 88: struct str {
1.78 schwarze 89: const struct mpage *mpage; /* if set, the owning parse */
1.50 kristaps 90: uint64_t mask; /* bitmask in sequence */
1.176 schwarze 91: char key[]; /* rendered text */
1.38 schwarze 92: };
93:
1.79 schwarze 94: struct inodev {
95: ino_t st_ino;
96: dev_t st_dev;
1.50 kristaps 97: };
1.5 kristaps 98:
1.78 schwarze 99: struct mpage {
1.79 schwarze 100: struct inodev inodev; /* used for hashing routine */
1.137 schwarze 101: int64_t pageid; /* pageid in mpages SQL table */
1.82 schwarze 102: char *sec; /* section from file content */
103: char *arch; /* architecture from file content */
104: char *title; /* title from file content */
105: char *desc; /* description from file content */
106: struct mlink *mlinks; /* singly linked list */
1.161 schwarze 107: int form; /* format from file content */
1.174 schwarze 108: int name_head_done;
1.82 schwarze 109: };
110:
111: struct mlink {
1.58 schwarze 112: char file[PATH_MAX]; /* filename rel. to manpath */
1.82 schwarze 113: char *dsec; /* section from directory */
114: char *arch; /* architecture from directory */
115: char *name; /* name from file name (not empty) */
116: char *fsec; /* section from file name suffix */
1.85 schwarze 117: struct mlink *next; /* singly linked list */
1.118 schwarze 118: struct mpage *mpage; /* parent */
1.161 schwarze 119: int dform; /* format from directory */
120: int fform; /* format from file name suffix */
1.124 schwarze 121: int gzip; /* filename has a .gz suffix */
1.3 kristaps 122: };
123:
1.50 kristaps 124: enum stmt {
1.81 schwarze 125: STMT_DELETE_PAGE = 0, /* delete mpage */
126: STMT_INSERT_PAGE, /* insert mpage */
127: STMT_INSERT_LINK, /* insert mlink */
1.133 schwarze 128: STMT_INSERT_NAME, /* insert name */
1.175 schwarze 129: STMT_SELECT_NAME, /* retrieve existing name flags */
1.81 schwarze 130: STMT_INSERT_KEY, /* insert parsed key */
1.50 kristaps 131: STMT__MAX
1.1 kristaps 132: };
133:
1.190 schwarze 134: typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
1.189 schwarze 135: const struct roff_node *);
1.1 kristaps 136:
1.50 kristaps 137: struct mdoc_handler {
138: mdoc_fp fp; /* optional handler */
139: uint64_t mask; /* set unless handler returns 0 */
1.1 kristaps 140: };
141:
1.59 schwarze 142: static void dbclose(int);
1.176 schwarze 143: static void dbadd(struct mpage *);
1.118 schwarze 144: static void dbadd_mlink(const struct mlink *mlink);
1.169 schwarze 145: static void dbadd_mlink_name(const struct mlink *mlink);
1.59 schwarze 146: static int dbopen(int);
147: static void dbprune(void);
148: static void filescan(const char *);
1.50 kristaps 149: static void *hash_alloc(size_t, void *);
1.152 schwarze 150: static void hash_free(void *, void *);
151: static void *hash_calloc(size_t, size_t, void *);
1.84 schwarze 152: static void mlink_add(struct mlink *, const struct stat *);
1.131 schwarze 153: static void mlink_check(struct mpage *, struct mlink *);
1.82 schwarze 154: static void mlink_free(struct mlink *);
1.89 schwarze 155: static void mlinks_undupe(struct mpage *);
1.78 schwarze 156: static void mpages_free(void);
1.176 schwarze 157: static void mpages_merge(struct mparse *);
1.130 schwarze 158: static void names_check(void);
1.124 schwarze 159: static void parse_cat(struct mpage *, int);
1.190 schwarze 160: static void parse_man(struct mpage *, const struct roff_meta *,
1.189 schwarze 161: const struct roff_node *);
1.190 schwarze 162: static void parse_mdoc(struct mpage *, const struct roff_meta *,
1.189 schwarze 163: const struct roff_node *);
1.190 schwarze 164: static int parse_mdoc_body(struct mpage *, const struct roff_meta *,
1.189 schwarze 165: const struct roff_node *);
1.190 schwarze 166: static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
1.189 schwarze 167: const struct roff_node *);
1.190 schwarze 168: static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
1.189 schwarze 169: const struct roff_node *);
170: static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
1.190 schwarze 171: static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
1.189 schwarze 172: const struct roff_node *);
1.190 schwarze 173: static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
1.189 schwarze 174: const struct roff_node *);
1.190 schwarze 175: static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
1.189 schwarze 176: const struct roff_node *);
1.190 schwarze 177: static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
1.189 schwarze 178: const struct roff_node *);
1.190 schwarze 179: static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
1.189 schwarze 180: const struct roff_node *);
1.190 schwarze 181: static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
1.189 schwarze 182: const struct roff_node *);
1.112 schwarze 183: static void putkey(const struct mpage *, char *, uint64_t);
1.176 schwarze 184: static void putkeys(const struct mpage *, char *, size_t, uint64_t);
1.78 schwarze 185: static void putmdockey(const struct mpage *,
1.189 schwarze 186: const struct roff_node *, uint64_t);
1.176 schwarze 187: static int render_string(char **, size_t *);
1.59 schwarze 188: static void say(const char *, const char *, ...);
1.165 schwarze 189: static int set_basedir(const char *, int);
1.59 schwarze 190: static int treescan(void);
1.50 kristaps 191: static size_t utf8(unsigned int, char [7]);
192:
1.198 schwarze 193: extern char *__progname;
194:
1.115 schwarze 195: static char tempfilename[32];
1.102 schwarze 196: static int nodb; /* no database changes */
1.116 schwarze 197: static int mparse_options; /* abort the parse early */
1.141 schwarze 198: static int use_all; /* use all found files */
199: static int debug; /* print what we're doing */
200: static int warnings; /* warn about crap */
1.95 schwarze 201: static int write_utf8; /* write UTF-8 output; else ASCII */
1.59 schwarze 202: static int exitcode; /* to be returned by main */
1.141 schwarze 203: static enum op op; /* operational mode */
1.59 schwarze 204: static char basedir[PATH_MAX]; /* current base directory */
1.176 schwarze 205: static struct mchars *mchars; /* table of named characters */
1.78 schwarze 206: static struct ohash mpages; /* table of distinct manual pages */
1.83 schwarze 207: static struct ohash mlinks; /* table of directory entries */
1.133 schwarze 208: static struct ohash names; /* table of all names */
1.50 kristaps 209: static struct ohash strings; /* table of all strings */
210: static sqlite3 *db = NULL; /* current database */
211: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
1.133 schwarze 212: static uint64_t name_mask;
1.25 schwarze 213:
214: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.68 schwarze 215: { NULL, 0 }, /* Ap */
216: { NULL, 0 }, /* Dd */
217: { NULL, 0 }, /* Dt */
218: { NULL, 0 }, /* Os */
219: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
220: { parse_mdoc_head, TYPE_Ss }, /* Ss */
221: { NULL, 0 }, /* Pp */
222: { NULL, 0 }, /* D1 */
223: { NULL, 0 }, /* Dl */
224: { NULL, 0 }, /* Bd */
225: { NULL, 0 }, /* Ed */
226: { NULL, 0 }, /* Bl */
227: { NULL, 0 }, /* El */
228: { NULL, 0 }, /* It */
229: { NULL, 0 }, /* Ad */
230: { NULL, TYPE_An }, /* An */
231: { NULL, TYPE_Ar }, /* Ar */
232: { NULL, TYPE_Cd }, /* Cd */
233: { NULL, TYPE_Cm }, /* Cm */
234: { NULL, TYPE_Dv }, /* Dv */
235: { NULL, TYPE_Er }, /* Er */
236: { NULL, TYPE_Ev }, /* Ev */
237: { NULL, 0 }, /* Ex */
238: { NULL, TYPE_Fa }, /* Fa */
239: { parse_mdoc_Fd, 0 }, /* Fd */
240: { NULL, TYPE_Fl }, /* Fl */
241: { parse_mdoc_Fn, 0 }, /* Fn */
242: { NULL, TYPE_Ft }, /* Ft */
243: { NULL, TYPE_Ic }, /* Ic */
1.92 schwarze 244: { NULL, TYPE_In }, /* In */
1.68 schwarze 245: { NULL, TYPE_Li }, /* Li */
1.132 schwarze 246: { parse_mdoc_Nd, 0 }, /* Nd */
1.129 schwarze 247: { parse_mdoc_Nm, 0 }, /* Nm */
1.68 schwarze 248: { NULL, 0 }, /* Op */
249: { NULL, 0 }, /* Ot */
250: { NULL, TYPE_Pa }, /* Pa */
251: { NULL, 0 }, /* Rv */
1.92 schwarze 252: { NULL, TYPE_St }, /* St */
1.68 schwarze 253: { NULL, TYPE_Va }, /* Va */
254: { parse_mdoc_body, TYPE_Va }, /* Vt */
255: { parse_mdoc_Xr, 0 }, /* Xr */
256: { NULL, 0 }, /* %A */
257: { NULL, 0 }, /* %B */
258: { NULL, 0 }, /* %D */
259: { NULL, 0 }, /* %I */
260: { NULL, 0 }, /* %J */
261: { NULL, 0 }, /* %N */
262: { NULL, 0 }, /* %O */
263: { NULL, 0 }, /* %P */
264: { NULL, 0 }, /* %R */
265: { NULL, 0 }, /* %T */
266: { NULL, 0 }, /* %V */
267: { NULL, 0 }, /* Ac */
268: { NULL, 0 }, /* Ao */
269: { NULL, 0 }, /* Aq */
270: { NULL, TYPE_At }, /* At */
271: { NULL, 0 }, /* Bc */
272: { NULL, 0 }, /* Bf */
273: { NULL, 0 }, /* Bo */
274: { NULL, 0 }, /* Bq */
275: { NULL, TYPE_Bsx }, /* Bsx */
276: { NULL, TYPE_Bx }, /* Bx */
277: { NULL, 0 }, /* Db */
278: { NULL, 0 }, /* Dc */
279: { NULL, 0 }, /* Do */
280: { NULL, 0 }, /* Dq */
281: { NULL, 0 }, /* Ec */
282: { NULL, 0 }, /* Ef */
283: { NULL, TYPE_Em }, /* Em */
284: { NULL, 0 }, /* Eo */
285: { NULL, TYPE_Fx }, /* Fx */
286: { NULL, TYPE_Ms }, /* Ms */
287: { NULL, 0 }, /* No */
288: { NULL, 0 }, /* Ns */
289: { NULL, TYPE_Nx }, /* Nx */
290: { NULL, TYPE_Ox }, /* Ox */
291: { NULL, 0 }, /* Pc */
292: { NULL, 0 }, /* Pf */
293: { NULL, 0 }, /* Po */
294: { NULL, 0 }, /* Pq */
295: { NULL, 0 }, /* Qc */
296: { NULL, 0 }, /* Ql */
297: { NULL, 0 }, /* Qo */
298: { NULL, 0 }, /* Qq */
299: { NULL, 0 }, /* Re */
300: { NULL, 0 }, /* Rs */
301: { NULL, 0 }, /* Sc */
302: { NULL, 0 }, /* So */
303: { NULL, 0 }, /* Sq */
304: { NULL, 0 }, /* Sm */
305: { NULL, 0 }, /* Sx */
306: { NULL, TYPE_Sy }, /* Sy */
307: { NULL, TYPE_Tn }, /* Tn */
308: { NULL, 0 }, /* Ux */
309: { NULL, 0 }, /* Xc */
310: { NULL, 0 }, /* Xo */
1.173 schwarze 311: { parse_mdoc_Fo, 0 }, /* Fo */
1.68 schwarze 312: { NULL, 0 }, /* Fc */
313: { NULL, 0 }, /* Oo */
314: { NULL, 0 }, /* Oc */
315: { NULL, 0 }, /* Bk */
316: { NULL, 0 }, /* Ek */
317: { NULL, 0 }, /* Bt */
318: { NULL, 0 }, /* Hf */
319: { NULL, 0 }, /* Fr */
320: { NULL, 0 }, /* Ud */
321: { NULL, TYPE_Lb }, /* Lb */
322: { NULL, 0 }, /* Lp */
323: { NULL, TYPE_Lk }, /* Lk */
324: { NULL, TYPE_Mt }, /* Mt */
325: { NULL, 0 }, /* Brq */
326: { NULL, 0 }, /* Bro */
327: { NULL, 0 }, /* Brc */
328: { NULL, 0 }, /* %C */
329: { NULL, 0 }, /* Es */
330: { NULL, 0 }, /* En */
331: { NULL, TYPE_Dx }, /* Dx */
332: { NULL, 0 }, /* %Q */
333: { NULL, 0 }, /* br */
334: { NULL, 0 }, /* sp */
335: { NULL, 0 }, /* %U */
336: { NULL, 0 }, /* Ta */
1.166 schwarze 337: { NULL, 0 }, /* ll */
1.1 kristaps 338: };
339:
1.141 schwarze 340:
1.1 kristaps 341: int
1.179 schwarze 342: mandocdb(int argc, char *argv[])
1.1 kristaps 343: {
1.187 schwarze 344: struct manconf conf;
345: struct ohash_info mpages_info, mlinks_info;
346: struct mparse *mp;
347: const char *path_arg;
348: size_t j, sz;
1.59 schwarze 349: int ch, i;
1.50 kristaps 350:
1.187 schwarze 351: memset(&conf, 0, sizeof(conf));
1.50 kristaps 352: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
353:
1.83 schwarze 354: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
1.152 schwarze 355: mpages_info.calloc = mlinks_info.calloc = hash_calloc;
1.181 schwarze 356: mpages_info.free = mlinks_info.free = hash_free;
357: mpages_info.data = mlinks_info.data = NULL;
1.50 kristaps 358:
1.79 schwarze 359: mpages_info.key_offset = offsetof(struct mpage, inodev);
1.83 schwarze 360: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 kristaps 361:
1.50 kristaps 362: /*
1.141 schwarze 363: * We accept a few different invocations.
1.50 kristaps 364: * The CHECKOP macro makes sure that invocation styles don't
365: * clobber each other.
366: */
367: #define CHECKOP(_op, _ch) do \
368: if (OP_DEFAULT != (_op)) { \
1.198 schwarze 369: warnx("-%c: Conflicting option", (_ch)); \
1.50 kristaps 370: goto usage; \
371: } while (/*CONSTCOND*/0)
1.10 kristaps 372:
1.59 schwarze 373: path_arg = NULL;
1.38 schwarze 374: op = OP_DEFAULT;
1.1 kristaps 375:
1.126 schwarze 376: while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
1.1 kristaps 377: switch (ch) {
1.141 schwarze 378: case 'a':
1.12 schwarze 379: use_all = 1;
380: break;
1.141 schwarze 381: case 'C':
1.50 kristaps 382: CHECKOP(op, ch);
1.59 schwarze 383: path_arg = optarg;
1.38 schwarze 384: op = OP_CONFFILE;
1.34 schwarze 385: break;
1.141 schwarze 386: case 'D':
1.125 schwarze 387: debug++;
388: break;
1.141 schwarze 389: case 'd':
1.50 kristaps 390: CHECKOP(op, ch);
1.59 schwarze 391: path_arg = optarg;
1.5 kristaps 392: op = OP_UPDATE;
393: break;
1.141 schwarze 394: case 'n':
1.50 kristaps 395: nodb = 1;
396: break;
1.141 schwarze 397: case 'p':
1.126 schwarze 398: warnings = 1;
399: break;
1.141 schwarze 400: case 'Q':
1.116 schwarze 401: mparse_options |= MPARSE_QUICK;
1.102 schwarze 402: break;
1.141 schwarze 403: case 'T':
1.95 schwarze 404: if (strcmp(optarg, "utf8")) {
1.198 schwarze 405: warnx("-T%s: Unsupported output format",
406: optarg);
1.95 schwarze 407: goto usage;
408: }
409: write_utf8 = 1;
410: break;
1.141 schwarze 411: case 't':
1.50 kristaps 412: CHECKOP(op, ch);
1.38 schwarze 413: dup2(STDOUT_FILENO, STDERR_FILENO);
414: op = OP_TEST;
1.50 kristaps 415: nodb = warnings = 1;
1.38 schwarze 416: break;
1.141 schwarze 417: case 'u':
1.50 kristaps 418: CHECKOP(op, ch);
1.59 schwarze 419: path_arg = optarg;
1.5 kristaps 420: op = OP_DELETE;
421: break;
1.141 schwarze 422: case 'v':
1.126 schwarze 423: /* Compatibility with espie@'s makewhatis. */
1.38 schwarze 424: break;
1.1 kristaps 425: default:
1.38 schwarze 426: goto usage;
1.1 kristaps 427: }
428:
429: argc -= optind;
430: argv += optind;
431:
1.38 schwarze 432: if (OP_CONFFILE == op && argc > 0) {
1.198 schwarze 433: warnx("-C: Too many arguments");
1.38 schwarze 434: goto usage;
435: }
436:
1.59 schwarze 437: exitcode = (int)MANDOCLEVEL_OK;
1.176 schwarze 438: mchars = mchars_alloc();
1.183 schwarze 439: mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
1.176 schwarze 440: mchars, NULL);
1.78 schwarze 441: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 442: ohash_init(&mlinks, 6, &mlinks_info);
1.50 kristaps 443:
444: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
1.59 schwarze 445:
1.50 kristaps 446: /*
1.150 schwarze 447: * Most of these deal with a specific directory.
1.140 schwarze 448: * Jump into that directory first.
1.50 kristaps 449: */
1.165 schwarze 450: if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
1.50 kristaps 451: goto out;
1.140 schwarze 452:
1.138 schwarze 453: if (dbopen(1)) {
1.140 schwarze 454: /*
455: * The existing database is usable. Process
456: * all files specified on the command-line.
457: */
458: use_all = 1;
1.138 schwarze 459: for (i = 0; i < argc; i++)
460: filescan(argv[i]);
461: if (OP_TEST != op)
462: dbprune();
463: } else {
464: /*
465: * Database missing or corrupt.
466: * Recreate from scratch.
467: */
1.140 schwarze 468: exitcode = (int)MANDOCLEVEL_OK;
1.138 schwarze 469: op = OP_DEFAULT;
470: if (0 == treescan())
471: goto out;
472: if (0 == dbopen(0))
473: goto out;
474: }
1.50 kristaps 475: if (OP_DELETE != op)
1.176 schwarze 476: mpages_merge(mp);
1.138 schwarze 477: dbclose(OP_DEFAULT == op ? 0 : 1);
1.50 kristaps 478: } else {
479: /*
480: * If we have arguments, use them as our manpaths.
481: * If we don't, grok from manpath(1) or however else
1.187 schwarze 482: * manconf_parse() wants to do it.
1.50 kristaps 483: */
484: if (argc > 0) {
1.187 schwarze 485: conf.manpath.paths = mandoc_reallocarray(NULL,
1.144 schwarze 486: argc, sizeof(char *));
1.187 schwarze 487: conf.manpath.sz = (size_t)argc;
1.50 kristaps 488: for (i = 0; i < argc; i++)
1.187 schwarze 489: conf.manpath.paths[i] = mandoc_strdup(argv[i]);
1.50 kristaps 490: } else
1.187 schwarze 491: manconf_parse(&conf, path_arg, NULL, NULL);
1.127 schwarze 492:
1.187 schwarze 493: if (conf.manpath.sz == 0) {
1.127 schwarze 494: exitcode = (int)MANDOCLEVEL_BADARG;
495: say("", "Empty manpath");
496: }
1.50 kristaps 497:
498: /*
1.69 schwarze 499: * First scan the tree rooted at a base directory, then
500: * build a new database and finally move it into place.
1.50 kristaps 501: * Ignore zero-length directories and strip trailing
502: * slashes.
503: */
1.187 schwarze 504: for (j = 0; j < conf.manpath.sz; j++) {
505: sz = strlen(conf.manpath.paths[j]);
506: if (sz && conf.manpath.paths[j][sz - 1] == '/')
507: conf.manpath.paths[j][--sz] = '\0';
1.50 kristaps 508: if (0 == sz)
509: continue;
1.66 schwarze 510:
511: if (j) {
1.78 schwarze 512: ohash_init(&mpages, 6, &mpages_info);
1.83 schwarze 513: ohash_init(&mlinks, 6, &mlinks_info);
1.66 schwarze 514: }
515:
1.187 schwarze 516: if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
1.165 schwarze 517: continue;
1.59 schwarze 518: if (0 == treescan())
1.165 schwarze 519: continue;
1.59 schwarze 520: if (0 == dbopen(0))
1.165 schwarze 521: continue;
1.52 kristaps 522:
1.176 schwarze 523: mpages_merge(mp);
1.147 schwarze 524: if (warnings && !nodb &&
1.130 schwarze 525: ! (MPARSE_QUICK & mparse_options))
526: names_check();
1.59 schwarze 527: dbclose(0);
1.66 schwarze 528:
1.187 schwarze 529: if (j + 1 < conf.manpath.sz) {
1.80 schwarze 530: mpages_free();
1.78 schwarze 531: ohash_delete(&mpages);
1.83 schwarze 532: ohash_delete(&mlinks);
1.66 schwarze 533: }
1.50 kristaps 534: }
535: }
536: out:
1.187 schwarze 537: manconf_free(&conf);
1.168 schwarze 538: mparse_free(mp);
1.176 schwarze 539: mchars_free(mchars);
1.80 schwarze 540: mpages_free();
1.78 schwarze 541: ohash_delete(&mpages);
1.83 schwarze 542: ohash_delete(&mlinks);
1.197 schwarze 543: return exitcode;
1.50 kristaps 544: usage:
1.126 schwarze 545: fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
546: " %s [-aDnpQ] [-Tutf8] dir ...\n"
547: " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
548: " %s [-Dnp] -u dir [file ...]\n"
1.102 schwarze 549: " %s [-Q] -t file ...\n",
1.198 schwarze 550: __progname, __progname, __progname,
551: __progname, __progname);
1.50 kristaps 552:
1.197 schwarze 553: return (int)MANDOCLEVEL_BADARG;
1.50 kristaps 554: }
555:
556: /*
1.59 schwarze 557: * Scan a directory tree rooted at "basedir" for manpages.
1.50 kristaps 558: * We use fts(), scanning directory parts along the way for clues to our
559: * section and architecture.
560: *
561: * If use_all has been specified, grok all files.
562: * If not, sanitise paths to the following:
563: *
1.141 schwarze 564: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 565: * or
566: * [./]cat<section>[/<arch>]/<name>.0
567: *
568: * TODO: accomodate for multi-language directories.
569: */
570: static int
1.59 schwarze 571: treescan(void)
1.50 kristaps 572: {
1.139 schwarze 573: char buf[PATH_MAX];
1.50 kristaps 574: FTS *f;
575: FTSENT *ff;
1.84 schwarze 576: struct mlink *mlink;
1.124 schwarze 577: int dform, gzip;
1.94 schwarze 578: char *dsec, *arch, *fsec, *cp;
579: const char *path;
1.50 kristaps 580: const char *argv[2];
581:
582: argv[0] = ".";
583: argv[1] = (char *)NULL;
584:
1.139 schwarze 585: f = fts_open((char * const *)argv,
586: FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1.200 schwarze 587: if (f == NULL) {
1.59 schwarze 588: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 589: say("", "&fts_open");
1.197 schwarze 590: return 0;
1.50 kristaps 591: }
592:
593: dsec = arch = NULL;
594: dform = FORM_NONE;
595:
1.200 schwarze 596: while ((ff = fts_read(f)) != NULL) {
1.50 kristaps 597: path = ff->fts_path + 2;
1.139 schwarze 598: switch (ff->fts_info) {
599:
600: /*
601: * Symbolic links require various sanity checks,
602: * then get handled just like regular files.
603: */
1.141 schwarze 604: case FTS_SL:
1.200 schwarze 605: if (realpath(path, buf) == NULL) {
1.139 schwarze 606: if (warnings)
607: say(path, "&realpath");
608: continue;
609: }
1.184 schwarze 610: if (strstr(buf, basedir) != buf
611: #ifdef HOMEBREWDIR
612: && strstr(buf, HOMEBREWDIR) != buf
613: #endif
614: ) {
1.139 schwarze 615: if (warnings) say("",
616: "%s: outside base directory", buf);
617: continue;
618: }
619: /* Use logical inode to avoid mpages dupe. */
1.200 schwarze 620: if (stat(path, ff->fts_statp) == -1) {
1.139 schwarze 621: if (warnings)
622: say(path, "&stat");
623: continue;
624: }
625: /* FALLTHROUGH */
626:
1.50 kristaps 627: /*
1.83 schwarze 628: * If we're a regular file, add an mlink by using the
1.50 kristaps 629: * stored directory data and handling the filename.
630: */
1.141 schwarze 631: case FTS_F:
1.200 schwarze 632: if ( ! strcmp(path, MANDOC_DB))
1.60 schwarze 633: continue;
1.50 kristaps 634: if ( ! use_all && ff->fts_level < 2) {
1.56 schwarze 635: if (warnings)
1.59 schwarze 636: say(path, "Extraneous file");
1.50 kristaps 637: continue;
1.124 schwarze 638: }
639: gzip = 0;
640: fsec = NULL;
1.200 schwarze 641: while (fsec == NULL) {
1.124 schwarze 642: fsec = strrchr(ff->fts_name, '.');
1.200 schwarze 643: if (fsec == NULL || strcmp(fsec+1, "gz"))
1.124 schwarze 644: break;
645: gzip = 1;
646: *fsec = '\0';
647: fsec = NULL;
648: }
1.200 schwarze 649: if (fsec == NULL) {
1.60 schwarze 650: if ( ! use_all) {
1.56 schwarze 651: if (warnings)
1.60 schwarze 652: say(path,
653: "No filename suffix");
1.50 kristaps 654: continue;
655: }
1.200 schwarze 656: } else if ( ! strcmp(++fsec, "html")) {
1.60 schwarze 657: if (warnings)
658: say(path, "Skip html");
659: continue;
1.200 schwarze 660: } else if ( ! strcmp(fsec, "ps")) {
1.60 schwarze 661: if (warnings)
662: say(path, "Skip ps");
663: continue;
1.200 schwarze 664: } else if ( ! strcmp(fsec, "pdf")) {
1.60 schwarze 665: if (warnings)
666: say(path, "Skip pdf");
667: continue;
668: } else if ( ! use_all &&
1.200 schwarze 669: ((dform == FORM_SRC &&
1.185 schwarze 670: strncmp(fsec, dsec, strlen(dsec))) ||
1.200 schwarze 671: (dform == FORM_CAT && 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:
696: case FTS_DP:
1.139 schwarze 697: break;
698:
699: default:
1.60 schwarze 700: if (warnings)
701: say(path, "Not a regular file");
1.50 kristaps 702: continue;
1.60 schwarze 703: }
1.1 kristaps 704:
1.50 kristaps 705: switch (ff->fts_level) {
1.141 schwarze 706: case 0:
1.50 kristaps 707: /* Ignore the root directory. */
708: break;
1.141 schwarze 709: case 1:
1.50 kristaps 710: /*
711: * This might contain manX/ or catX/.
712: * Try to infer this from the name.
713: * If we're not in use_all, enforce it.
714: */
715: cp = ff->fts_name;
1.200 schwarze 716: if (ff->fts_info == FTS_DP) {
717: dform = FORM_NONE;
718: dsec = NULL;
1.50 kristaps 719: break;
1.200 schwarze 720: }
1.1 kristaps 721:
1.200 schwarze 722: if ( ! strncmp(cp, "man", 3)) {
1.50 kristaps 723: dform = FORM_SRC;
1.66 schwarze 724: dsec = cp + 3;
1.200 schwarze 725: } else if ( ! strncmp(cp, "cat", 3)) {
1.50 kristaps 726: dform = FORM_CAT;
1.66 schwarze 727: dsec = cp + 3;
1.94 schwarze 728: } else {
729: dform = FORM_NONE;
730: dsec = NULL;
1.50 kristaps 731: }
1.1 kristaps 732:
1.200 schwarze 733: if (dsec != NULL || use_all)
1.50 kristaps 734: break;
1.1 kristaps 735:
1.56 schwarze 736: if (warnings)
1.59 schwarze 737: say(path, "Unknown directory part");
1.50 kristaps 738: fts_set(f, ff, FTS_SKIP);
739: break;
1.141 schwarze 740: case 2:
1.50 kristaps 741: /*
742: * Possibly our architecture.
743: * If we're descending, keep tabs on it.
744: */
1.200 schwarze 745: if (ff->fts_info != FTS_DP && dsec != NULL)
1.66 schwarze 746: arch = ff->fts_name;
1.94 schwarze 747: else
748: arch = NULL;
1.50 kristaps 749: break;
750: default:
1.200 schwarze 751: if (ff->fts_info == FTS_DP || use_all)
1.50 kristaps 752: break;
1.56 schwarze 753: if (warnings)
1.59 schwarze 754: say(path, "Extraneous directory part");
1.50 kristaps 755: fts_set(f, ff, FTS_SKIP);
756: break;
1.5 kristaps 757: }
1.50 kristaps 758: }
759:
760: fts_close(f);
1.197 schwarze 761: return 1;
1.50 kristaps 762: }
1.5 kristaps 763:
1.50 kristaps 764: /*
1.89 schwarze 765: * Add a file to the mlinks table.
1.50 kristaps 766: * Do not verify that it's a "valid" looking manpage (we'll do that
767: * later).
768: *
769: * Try to infer the manual section, architecture, and page name from the
770: * path, assuming it looks like
771: *
1.141 schwarze 772: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 773: * or
774: * [./]cat<section>[/<arch>]/<name>.0
775: *
776: * See treescan() for the fts(3) version of this.
777: */
778: static void
1.59 schwarze 779: filescan(const char *file)
1.50 kristaps 780: {
1.59 schwarze 781: char buf[PATH_MAX];
1.84 schwarze 782: struct stat st;
783: struct mlink *mlink;
1.59 schwarze 784: char *p, *start;
1.5 kristaps 785:
1.50 kristaps 786: assert(use_all);
1.5 kristaps 787:
1.50 kristaps 788: if (0 == strncmp(file, "./", 2))
789: file += 2;
1.5 kristaps 790:
1.139 schwarze 791: /*
792: * We have to do lstat(2) before realpath(3) loses
793: * the information whether this is a symbolic link.
794: * We need to know that because for symbolic links,
795: * we want to use the orginal file name, while for
796: * regular files, we want to use the real path.
797: */
798: if (-1 == lstat(file, &st)) {
799: exitcode = (int)MANDOCLEVEL_BADARG;
800: say(file, "&lstat");
801: return;
802: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
803: exitcode = (int)MANDOCLEVEL_BADARG;
804: say(file, "Not a regular file");
805: return;
806: }
807:
808: /*
809: * We have to resolve the file name to the real path
810: * in any case for the base directory check.
811: */
1.59 schwarze 812: if (NULL == realpath(file, buf)) {
813: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 814: say(file, "&realpath");
1.59 schwarze 815: return;
1.106 schwarze 816: }
817:
1.150 schwarze 818: if (OP_TEST == op)
1.106 schwarze 819: start = buf;
1.150 schwarze 820: else if (strstr(buf, basedir) == buf)
821: start = buf + strlen(basedir);
1.184 schwarze 822: #ifdef HOMEBREWDIR
823: else if (strstr(buf, HOMEBREWDIR) == buf)
824: start = buf;
825: #endif
1.106 schwarze 826: else {
1.59 schwarze 827: exitcode = (int)MANDOCLEVEL_BADARG;
828: say("", "%s: outside base directory", buf);
829: return;
1.106 schwarze 830: }
831:
1.139 schwarze 832: /*
833: * Now we are sure the file is inside our tree.
834: * If it is a symbolic link, ignore the real path
835: * and use the original name.
836: * This implies passing stuff like "cat1/../man1/foo.1"
837: * on the command line won't work. So don't do that.
838: * Note the stat(2) can still fail if the link target
839: * doesn't exist.
840: */
841: if (S_IFLNK & st.st_mode) {
842: if (-1 == stat(buf, &st)) {
843: exitcode = (int)MANDOCLEVEL_BADARG;
844: say(file, "&stat");
845: return;
846: }
1.143 schwarze 847: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
848: say(file, "Filename too long");
849: return;
850: }
1.150 schwarze 851: start = buf;
852: if (OP_TEST != op && strstr(buf, basedir) == buf)
853: start += strlen(basedir);
1.50 kristaps 854: }
1.106 schwarze 855:
1.84 schwarze 856: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 857: mlink->dform = FORM_NONE;
1.143 schwarze 858: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
859: sizeof(mlink->file)) {
860: say(start, "Filename too long");
1.180 schwarze 861: free(mlink);
1.143 schwarze 862: return;
863: }
1.17 schwarze 864:
1.50 kristaps 865: /*
866: * First try to guess our directory structure.
867: * If we find a separator, try to look for man* or cat*.
868: * If we find one of these and what's underneath is a directory,
869: * assume it's an architecture.
870: */
871: if (NULL != (p = strchr(start, '/'))) {
872: *p++ = '\0';
873: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 874: mlink->dform = FORM_SRC;
1.94 schwarze 875: mlink->dsec = start + 3;
1.50 kristaps 876: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 877: mlink->dform = FORM_CAT;
1.94 schwarze 878: mlink->dsec = start + 3;
1.17 schwarze 879: }
1.5 kristaps 880:
1.50 kristaps 881: start = p;
1.84 schwarze 882: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 883: *p++ = '\0';
1.94 schwarze 884: mlink->arch = start;
1.50 kristaps 885: start = p;
1.84 schwarze 886: }
1.50 kristaps 887: }
888:
889: /*
890: * Now check the file suffix.
891: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
892: */
893: p = strrchr(start, '\0');
894: while (p-- > start && '/' != *p && '.' != *p)
895: /* Loop. */ ;
896:
897: if ('.' == *p) {
898: *p++ = '\0';
1.94 schwarze 899: mlink->fsec = p;
1.5 kristaps 900: }
901:
1.10 kristaps 902: /*
1.50 kristaps 903: * Now try to parse the name.
904: * Use the filename portion of the path.
1.10 kristaps 905: */
1.84 schwarze 906: mlink->name = start;
1.50 kristaps 907: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 908: mlink->name = p + 1;
1.50 kristaps 909: *p = '\0';
1.84 schwarze 910: }
911: mlink_add(mlink, &st);
1.50 kristaps 912: }
1.5 kristaps 913:
1.50 kristaps 914: static void
1.84 schwarze 915: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 916: {
1.83 schwarze 917: struct inodev inodev;
1.78 schwarze 918: struct mpage *mpage;
1.83 schwarze 919: unsigned int slot;
1.50 kristaps 920:
1.84 schwarze 921: assert(NULL != mlink->file);
1.50 kristaps 922:
1.94 schwarze 923: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
924: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
925: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
926: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 927:
928: if ('0' == *mlink->fsec) {
929: free(mlink->fsec);
930: mlink->fsec = mandoc_strdup(mlink->dsec);
931: mlink->fform = FORM_CAT;
932: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
933: mlink->fform = FORM_SRC;
1.74 schwarze 934: else
1.84 schwarze 935: mlink->fform = FORM_NONE;
1.82 schwarze 936:
1.83 schwarze 937: slot = ohash_qlookup(&mlinks, mlink->file);
938: assert(NULL == ohash_find(&mlinks, slot));
939: ohash_insert(&mlinks, slot, mlink);
940:
1.186 schwarze 941: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 942: inodev.st_ino = st->st_ino;
943: inodev.st_dev = st->st_dev;
944: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
945: sizeof(struct inodev), inodev.st_ino);
946: mpage = ohash_find(&mpages, slot);
947: if (NULL == mpage) {
948: mpage = mandoc_calloc(1, sizeof(struct mpage));
949: mpage->inodev.st_ino = inodev.st_ino;
950: mpage->inodev.st_dev = inodev.st_dev;
951: ohash_insert(&mpages, slot, mpage);
952: } else
1.85 schwarze 953: mlink->next = mpage->mlinks;
1.82 schwarze 954: mpage->mlinks = mlink;
1.118 schwarze 955: mlink->mpage = mpage;
1.82 schwarze 956: }
957:
958: static void
959: mlink_free(struct mlink *mlink)
960: {
961:
962: free(mlink->dsec);
963: free(mlink->arch);
964: free(mlink->name);
965: free(mlink->fsec);
966: free(mlink);
1.50 kristaps 967: }
1.3 kristaps 968:
1.50 kristaps 969: static void
1.78 schwarze 970: mpages_free(void)
1.50 kristaps 971: {
1.78 schwarze 972: struct mpage *mpage;
1.82 schwarze 973: struct mlink *mlink;
1.80 schwarze 974: unsigned int slot;
1.3 kristaps 975:
1.80 schwarze 976: mpage = ohash_first(&mpages, &slot);
977: while (NULL != mpage) {
1.82 schwarze 978: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 979: mpage->mlinks = mlink->next;
1.82 schwarze 980: mlink_free(mlink);
981: }
1.78 schwarze 982: free(mpage->sec);
983: free(mpage->arch);
1.82 schwarze 984: free(mpage->title);
985: free(mpage->desc);
1.78 schwarze 986: free(mpage);
1.80 schwarze 987: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 988: }
989: }
1.38 schwarze 990:
1.50 kristaps 991: /*
1.89 schwarze 992: * For each mlink to the mpage, check whether the path looks like
993: * it is formatted, and if it does, check whether a source manual
994: * exists by the same name, ignoring the suffix.
995: * If both conditions hold, drop the mlink.
996: */
997: static void
998: mlinks_undupe(struct mpage *mpage)
999: {
1000: char buf[PATH_MAX];
1001: struct mlink **prev;
1002: struct mlink *mlink;
1003: char *bufp;
1004:
1005: mpage->form = FORM_CAT;
1.90 schwarze 1006: prev = &mpage->mlinks;
1007: while (NULL != (mlink = *prev)) {
1.89 schwarze 1008: if (FORM_CAT != mlink->dform) {
1009: mpage->form = FORM_NONE;
1.90 schwarze 1010: goto nextlink;
1.89 schwarze 1011: }
1.143 schwarze 1012: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1013: bufp = strstr(buf, "cat");
1014: assert(NULL != bufp);
1015: memcpy(bufp, "man", 3);
1016: if (NULL != (bufp = strrchr(buf, '.')))
1017: *++bufp = '\0';
1.143 schwarze 1018: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1019: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1020: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1021: goto nextlink;
1.89 schwarze 1022: if (warnings)
1023: say(mlink->file, "Man source exists: %s", buf);
1024: if (use_all)
1.90 schwarze 1025: goto nextlink;
1.89 schwarze 1026: *prev = mlink->next;
1027: mlink_free(mlink);
1.90 schwarze 1028: continue;
1029: nextlink:
1030: prev = &(*prev)->next;
1.89 schwarze 1031: }
1032: }
1033:
1.131 schwarze 1034: static void
1.93 schwarze 1035: mlink_check(struct mpage *mpage, struct mlink *mlink)
1036: {
1.131 schwarze 1037: struct str *str;
1038: unsigned int slot;
1.93 schwarze 1039:
1040: /*
1041: * Check whether the manual section given in a file
1042: * agrees with the directory where the file is located.
1043: * Some manuals have suffixes like (3p) on their
1044: * section number either inside the file or in the
1045: * directory name, some are linked into more than one
1046: * section, like encrypt(1) = makekey(8).
1047: */
1048:
1049: if (FORM_SRC == mpage->form &&
1.131 schwarze 1050: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1051: say(mlink->file, "Section \"%s\" manual in %s directory",
1052: mpage->sec, mlink->dsec);
1053:
1054: /*
1055: * Manual page directories exist for each kernel
1056: * architecture as returned by machine(1).
1057: * However, many manuals only depend on the
1058: * application architecture as returned by arch(1).
1059: * For example, some (2/ARM) manuals are shared
1060: * across the "armish" and "zaurus" kernel
1061: * architectures.
1062: * A few manuals are even shared across completely
1063: * different architectures, for example fdformat(1)
1064: * on amd64, i386, sparc, and sparc64.
1065: */
1066:
1.131 schwarze 1067: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1068: say(mlink->file, "Architecture \"%s\" manual in "
1069: "\"%s\" directory", mpage->arch, mlink->arch);
1070:
1.131 schwarze 1071: /*
1072: * XXX
1.133 schwarze 1073: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1074: */
1075:
1076: if (FORM_CAT == mpage->form)
1077: return;
1078:
1079: /*
1080: * Check whether this mlink
1081: * appears as a name in the NAME section.
1082: */
1.93 schwarze 1083:
1.133 schwarze 1084: slot = ohash_qlookup(&names, mlink->name);
1085: str = ohash_find(&names, slot);
1.131 schwarze 1086: assert(NULL != str);
1.133 schwarze 1087: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1088: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1089: }
1090:
1.89 schwarze 1091: /*
1.80 schwarze 1092: * Run through the files in the global vector "mpages"
1093: * and add them to the database specified in "basedir".
1.50 kristaps 1094: *
1095: * This handles the parsing scheme itself, using the cues of directory
1096: * and filename to determine whether the file is parsable or not.
1097: */
1.59 schwarze 1098: static void
1.176 schwarze 1099: mpages_merge(struct mparse *mp)
1.50 kristaps 1100: {
1.113 schwarze 1101: char any[] = "any";
1.101 schwarze 1102: struct ohash_info str_info;
1.118 schwarze 1103: struct mpage *mpage, *mpage_dest;
1104: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1105: struct roff_man *man;
1.118 schwarze 1106: char *sodest;
1.112 schwarze 1107: char *cp;
1.162 schwarze 1108: int fd;
1.101 schwarze 1109: unsigned int pslot;
1.69 schwarze 1110:
1.71 schwarze 1111: str_info.alloc = hash_alloc;
1.152 schwarze 1112: str_info.calloc = hash_calloc;
1113: str_info.free = hash_free;
1.181 schwarze 1114: str_info.data = NULL;
1.71 schwarze 1115: str_info.key_offset = offsetof(struct str, key);
1116:
1.171 schwarze 1117: if ( ! nodb)
1.107 schwarze 1118: SQL_EXEC("BEGIN TRANSACTION");
1119:
1.80 schwarze 1120: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1121: while (mpage != NULL) {
1.89 schwarze 1122: mlinks_undupe(mpage);
1.182 schwarze 1123: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1124: mpage = ohash_next(&mpages, &pslot);
1125: continue;
1.50 kristaps 1126: }
1127:
1.133 schwarze 1128: name_mask = NAME_MASK;
1129: ohash_init(&names, 4, &str_info);
1.71 schwarze 1130: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1131: mparse_reset(mp);
1132: man = NULL;
1.124 schwarze 1133: sodest = NULL;
1134:
1.182 schwarze 1135: mparse_open(mp, &fd, mlink->file);
1.162 schwarze 1136: if (fd == -1) {
1.182 schwarze 1137: say(mlink->file, "&open");
1.162 schwarze 1138: goto nextpage;
1.124 schwarze 1139: }
1.14 schwarze 1140:
1141: /*
1.183 schwarze 1142: * Interpret the file as mdoc(7) or man(7) source
1143: * code, unless it is known to be formatted.
1.14 schwarze 1144: */
1.182 schwarze 1145: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1146: mparse_readfd(mp, fd, mlink->file);
1.192 schwarze 1147: mparse_result(mp, &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: while (1) {
1165: mlink->mpage = mpage_dest;
1166:
1167: /*
1168: * If the target was already
1169: * processed, add the links
1170: * to the database now.
1171: * Otherwise, this will
1172: * happen when we come
1173: * to the target.
1174: */
1175:
1.137 schwarze 1176: if (mpage_dest->pageid)
1.169 schwarze 1177: dbadd_mlink_name(mlink);
1.118 schwarze 1178:
1.171 schwarze 1179: if (mlink->next == NULL)
1.118 schwarze 1180: break;
1181: mlink = mlink->next;
1182: }
1183:
1184: /* Move all links to the target. */
1185:
1186: mlink->next = mlink_dest->next;
1187: mlink_dest->next = mpage->mlinks;
1188: mpage->mlinks = NULL;
1189: }
1.124 schwarze 1190: goto nextpage;
1.192 schwarze 1191: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.82 schwarze 1192: mpage->form = FORM_SRC;
1.193 schwarze 1193: mpage->sec = man->meta.msec;
1.155 schwarze 1194: mpage->sec = mandoc_strdup(
1.171 schwarze 1195: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1196: mpage->arch = man->meta.arch;
1.84 schwarze 1197: mpage->arch = mandoc_strdup(
1.171 schwarze 1198: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1199: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1200: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.82 schwarze 1201: mpage->form = FORM_SRC;
1.193 schwarze 1202: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1203: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1204: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1205: } else {
1.82 schwarze 1206: mpage->form = FORM_CAT;
1.182 schwarze 1207: mpage->sec = mandoc_strdup(mlink->dsec);
1208: mpage->arch = mandoc_strdup(mlink->arch);
1209: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1210: }
1.97 schwarze 1211: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1212: if (*mpage->arch != '\0')
1213: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1214:
1.182 schwarze 1215: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1216: if ('\0' != *mlink->dsec)
1217: putkey(mpage, mlink->dsec, TYPE_sec);
1218: if ('\0' != *mlink->fsec)
1219: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1220: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1221: any : mlink->arch, TYPE_arch);
1.133 schwarze 1222: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1223: }
1.12 schwarze 1224:
1.174 schwarze 1225: assert(mpage->desc == NULL);
1.192 schwarze 1226: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1227: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1228: else if (man != NULL)
1.193 schwarze 1229: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1230: else
1.162 schwarze 1231: parse_cat(mpage, fd);
1.174 schwarze 1232: if (mpage->desc == NULL)
1.135 schwarze 1233: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1234:
1235: if (warnings && !use_all)
1236: for (mlink = mpage->mlinks; mlink;
1237: mlink = mlink->next)
1238: mlink_check(mpage, mlink);
1.44 kristaps 1239:
1.176 schwarze 1240: dbadd(mpage);
1.182 schwarze 1241: mlink = mpage->mlinks;
1.124 schwarze 1242:
1243: nextpage:
1.66 schwarze 1244: ohash_delete(&strings);
1.133 schwarze 1245: ohash_delete(&names);
1.80 schwarze 1246: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1247: }
1.107 schwarze 1248:
1249: if (0 == nodb)
1250: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1251: }
1252:
1253: static void
1254: names_check(void)
1255: {
1256: sqlite3_stmt *stmt;
1257: const char *name, *sec, *arch, *key;
1258:
1259: sqlite3_prepare_v2(db,
1260: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1261: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1262: "WHERE bits & ? AND NOT EXISTS ("
1263: "SELECT pageid FROM mlinks "
1.133 schwarze 1264: "WHERE mlinks.pageid == names.pageid "
1265: "AND mlinks.name == names.name"
1.130 schwarze 1266: ")"
1267: ") JOIN ("
1.142 schwarze 1268: "SELECT sec, arch, name, pageid FROM mlinks "
1269: "GROUP BY pageid"
1.130 schwarze 1270: ") USING (pageid);",
1271: -1, &stmt, NULL);
1272:
1.201 ! schwarze 1273: if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1.134 schwarze 1274: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1275:
1.201 ! schwarze 1276: while (sqlite3_step(stmt) == SQLITE_ROW) {
1.154 schwarze 1277: name = (const char *)sqlite3_column_text(stmt, 0);
1278: sec = (const char *)sqlite3_column_text(stmt, 1);
1279: arch = (const char *)sqlite3_column_text(stmt, 2);
1280: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1281: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1282: '\0' == *arch ? "" : "/",
1283: '\0' == *arch ? "" : arch, key);
1284: }
1285: sqlite3_finalize(stmt);
1.50 kristaps 1286: }
1.12 schwarze 1287:
1.50 kristaps 1288: static void
1.124 schwarze 1289: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1290: {
1291: FILE *stream;
1292: char *line, *p, *title;
1293: size_t len, plen, titlesz;
1.12 schwarze 1294:
1.124 schwarze 1295: stream = (-1 == fd) ?
1296: fopen(mpage->mlinks->file, "r") :
1297: fdopen(fd, "r");
1298: if (NULL == stream) {
1.159 schwarze 1299: if (-1 != fd)
1300: close(fd);
1.56 schwarze 1301: if (warnings)
1.123 schwarze 1302: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1303: return;
1304: }
1.1 kristaps 1305:
1.50 kristaps 1306: /* Skip to first blank line. */
1.1 kristaps 1307:
1.50 kristaps 1308: while (NULL != (line = fgetln(stream, &len)))
1309: if ('\n' == *line)
1310: break;
1.1 kristaps 1311:
1.50 kristaps 1312: /*
1313: * Assume the first line that is not indented
1314: * is the first section header. Skip to it.
1315: */
1.1 kristaps 1316:
1.50 kristaps 1317: while (NULL != (line = fgetln(stream, &len)))
1318: if ('\n' != *line && ' ' != *line)
1319: break;
1.141 schwarze 1320:
1.50 kristaps 1321: /*
1322: * Read up until the next section into a buffer.
1323: * Strip the leading and trailing newline from each read line,
1324: * appending a trailing space.
1325: * Ignore empty (whitespace-only) lines.
1326: */
1.1 kristaps 1327:
1.50 kristaps 1328: titlesz = 0;
1329: title = NULL;
1.38 schwarze 1330:
1.50 kristaps 1331: while (NULL != (line = fgetln(stream, &len))) {
1332: if (' ' != *line || '\n' != line[len - 1])
1333: break;
1334: while (len > 0 && isspace((unsigned char)*line)) {
1335: line++;
1336: len--;
1337: }
1338: if (1 == len)
1.38 schwarze 1339: continue;
1.50 kristaps 1340: title = mandoc_realloc(title, titlesz + len);
1341: memcpy(title + titlesz, line, len);
1342: titlesz += len;
1343: title[titlesz - 1] = ' ';
1344: }
1.38 schwarze 1345:
1.50 kristaps 1346: /*
1347: * If no page content can be found, or the input line
1348: * is already the next section header, or there is no
1349: * trailing newline, reuse the page title as the page
1350: * description.
1351: */
1.44 kristaps 1352:
1.50 kristaps 1353: if (NULL == title || '\0' == *title) {
1.56 schwarze 1354: if (warnings)
1.82 schwarze 1355: say(mpage->mlinks->file,
1356: "Cannot find NAME section");
1.50 kristaps 1357: fclose(stream);
1358: free(title);
1359: return;
1360: }
1.1 kristaps 1361:
1.50 kristaps 1362: title = mandoc_realloc(title, titlesz + 1);
1363: title[titlesz] = '\0';
1.33 schwarze 1364:
1.50 kristaps 1365: /*
1366: * Skip to the first dash.
1367: * Use the remaining line as the description (no more than 70
1368: * bytes).
1369: */
1.33 schwarze 1370:
1.50 kristaps 1371: if (NULL != (p = strstr(title, "- "))) {
1372: for (p += 2; ' ' == *p || '\b' == *p; p++)
1373: /* Skip to next word. */ ;
1374: } else {
1.56 schwarze 1375: if (warnings)
1.82 schwarze 1376: say(mpage->mlinks->file,
1377: "No dash in title line");
1.50 kristaps 1378: p = title;
1379: }
1.38 schwarze 1380:
1.50 kristaps 1381: plen = strlen(p);
1.1 kristaps 1382:
1.50 kristaps 1383: /* Strip backspace-encoding from line. */
1.1 kristaps 1384:
1.50 kristaps 1385: while (NULL != (line = memchr(p, '\b', plen))) {
1386: len = line - p;
1387: if (0 == len) {
1388: memmove(line, line + 1, plen--);
1389: continue;
1.141 schwarze 1390: }
1.50 kristaps 1391: memmove(line - 1, line + 1, plen - len);
1392: plen -= 2;
1393: }
1.1 kristaps 1394:
1.78 schwarze 1395: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1396: fclose(stream);
1397: free(title);
1398: }
1.1 kristaps 1399:
1.50 kristaps 1400: /*
1401: * Put a type/word pair into the word database for this particular file.
1402: */
1403: static void
1.112 schwarze 1404: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1405: {
1.112 schwarze 1406: char *cp;
1.18 kristaps 1407:
1.50 kristaps 1408: assert(NULL != value);
1.112 schwarze 1409: if (TYPE_arch == type)
1410: for (cp = value; *cp; cp++)
1411: if (isupper((unsigned char)*cp))
1412: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1413: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1414: }
1415:
1416: /*
1.50 kristaps 1417: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1418: */
1419: static void
1.78 schwarze 1420: putmdockey(const struct mpage *mpage,
1.189 schwarze 1421: const struct roff_node *n, uint64_t m)
1.3 kristaps 1422: {
1.18 kristaps 1423:
1.50 kristaps 1424: for ( ; NULL != n; n = n->next) {
1425: if (NULL != n->child)
1.78 schwarze 1426: putmdockey(mpage, n->child, m);
1.188 schwarze 1427: if (n->type == ROFFT_TEXT)
1.78 schwarze 1428: putkey(mpage, n->string, m);
1.50 kristaps 1429: }
1430: }
1.18 kristaps 1431:
1.61 schwarze 1432: static void
1.190 schwarze 1433: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1434: const struct roff_node *n)
1.50 kristaps 1435: {
1.189 schwarze 1436: const struct roff_node *head, *body;
1.121 schwarze 1437: char *start, *title;
1.50 kristaps 1438: char byte;
1.121 schwarze 1439: size_t sz;
1.18 kristaps 1440:
1.50 kristaps 1441: if (NULL == n)
1.61 schwarze 1442: return;
1.18 kristaps 1443:
1.50 kristaps 1444: /*
1445: * We're only searching for one thing: the first text child in
1446: * the BODY of a NAME section. Since we don't keep track of
1447: * sections in -man, run some hoops to find out whether we're in
1448: * the correct section or not.
1449: */
1.18 kristaps 1450:
1.188 schwarze 1451: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1452: body = n;
1453: assert(body->parent);
1454: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1455: 1 == head->nchild &&
1456: NULL != (head = (head->child)) &&
1.188 schwarze 1457: head->type == ROFFT_TEXT &&
1.141 schwarze 1458: 0 == strcmp(head->string, "NAME") &&
1459: NULL != body->child) {
1.3 kristaps 1460:
1.50 kristaps 1461: /*
1462: * Suck the entire NAME section into memory.
1463: * Yes, we might run away.
1464: * But too many manuals have big, spread-out
1465: * NAME sections over many lines.
1466: */
1.3 kristaps 1467:
1.121 schwarze 1468: title = NULL;
1.194 schwarze 1469: deroff(&title, body);
1.50 kristaps 1470: if (NULL == title)
1.61 schwarze 1471: return;
1.18 kristaps 1472:
1.141 schwarze 1473: /*
1.50 kristaps 1474: * Go through a special heuristic dance here.
1475: * Conventionally, one or more manual names are
1476: * comma-specified prior to a whitespace, then a
1477: * dash, then a description. Try to puzzle out
1478: * the name parts here.
1479: */
1.18 kristaps 1480:
1.121 schwarze 1481: start = title;
1.50 kristaps 1482: for ( ;; ) {
1483: sz = strcspn(start, " ,");
1484: if ('\0' == start[sz])
1485: break;
1.1 kristaps 1486:
1.50 kristaps 1487: byte = start[sz];
1488: start[sz] = '\0';
1.110 schwarze 1489:
1490: /*
1491: * Assume a stray trailing comma in the
1492: * name list if a name begins with a dash.
1493: */
1494:
1495: if ('-' == start[0] ||
1496: ('\\' == start[0] && '-' == start[1]))
1497: break;
1.1 kristaps 1498:
1.133 schwarze 1499: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1500: if ( ! (mpage->name_head_done ||
1501: strcasecmp(start, meta->title))) {
1502: putkey(mpage, start, NAME_HEAD);
1503: mpage->name_head_done = 1;
1504: }
1.1 kristaps 1505:
1.50 kristaps 1506: if (' ' == byte) {
1507: start += sz + 1;
1508: break;
1509: }
1.1 kristaps 1510:
1.50 kristaps 1511: assert(',' == byte);
1512: start += sz + 1;
1513: while (' ' == *start)
1514: start++;
1515: }
1.1 kristaps 1516:
1.121 schwarze 1517: if (start == title) {
1.133 schwarze 1518: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1519: if ( ! (mpage->name_head_done ||
1520: strcasecmp(start, meta->title))) {
1521: putkey(mpage, start, NAME_HEAD);
1522: mpage->name_head_done = 1;
1523: }
1.50 kristaps 1524: free(title);
1.61 schwarze 1525: return;
1.50 kristaps 1526: }
1.1 kristaps 1527:
1.50 kristaps 1528: while (isspace((unsigned char)*start))
1529: start++;
1.1 kristaps 1530:
1.50 kristaps 1531: if (0 == strncmp(start, "-", 1))
1532: start += 1;
1533: else if (0 == strncmp(start, "\\-\\-", 4))
1534: start += 4;
1535: else if (0 == strncmp(start, "\\-", 2))
1536: start += 2;
1537: else if (0 == strncmp(start, "\\(en", 4))
1538: start += 4;
1539: else if (0 == strncmp(start, "\\(em", 4))
1540: start += 4;
1.1 kristaps 1541:
1.50 kristaps 1542: while (' ' == *start)
1543: start++;
1.1 kristaps 1544:
1.78 schwarze 1545: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1546: free(title);
1.61 schwarze 1547: return;
1.50 kristaps 1548: }
1549: }
1.1 kristaps 1550:
1.77 schwarze 1551: for (n = n->child; n; n = n->next) {
1.78 schwarze 1552: if (NULL != mpage->desc)
1.77 schwarze 1553: break;
1.174 schwarze 1554: parse_man(mpage, meta, n);
1.77 schwarze 1555: }
1.1 kristaps 1556: }
1557:
1558: static void
1.190 schwarze 1559: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1560: const struct roff_node *n)
1.1 kristaps 1561: {
1562:
1.50 kristaps 1563: assert(NULL != n);
1564: for (n = n->child; NULL != n; n = n->next) {
1565: switch (n->type) {
1.188 schwarze 1566: case ROFFT_ELEM:
1567: case ROFFT_BLOCK:
1568: case ROFFT_HEAD:
1569: case ROFFT_BODY:
1570: case ROFFT_TAIL:
1.50 kristaps 1571: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1572: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1573: break;
1.68 schwarze 1574: if (mdocs[n->tok].mask)
1.78 schwarze 1575: putmdockey(mpage, n->child,
1576: mdocs[n->tok].mask);
1.50 kristaps 1577: break;
1578: default:
1.188 schwarze 1579: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1580: continue;
1581: }
1582: if (NULL != n->child)
1.172 schwarze 1583: parse_mdoc(mpage, meta, n);
1.1 kristaps 1584: }
1585: }
1586:
1.25 schwarze 1587: static int
1.190 schwarze 1588: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1589: const struct roff_node *n)
1.1 kristaps 1590: {
1.176 schwarze 1591: char *start, *end;
1.1 kristaps 1592: size_t sz;
1.25 schwarze 1593:
1.50 kristaps 1594: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1595: NULL == (n = n->child) ||
1.188 schwarze 1596: n->type != ROFFT_TEXT)
1.197 schwarze 1597: return 0;
1.1 kristaps 1598:
1599: /*
1600: * Only consider those `Fd' macro fields that begin with an
1601: * "inclusion" token (versus, e.g., #define).
1602: */
1.50 kristaps 1603:
1.1 kristaps 1604: if (strcmp("#include", n->string))
1.197 schwarze 1605: return 0;
1.1 kristaps 1606:
1.188 schwarze 1607: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1608: return 0;
1.1 kristaps 1609:
1610: /*
1611: * Strip away the enclosing angle brackets and make sure we're
1612: * not zero-length.
1613: */
1614:
1615: start = n->string;
1616: if ('<' == *start || '"' == *start)
1617: start++;
1618:
1619: if (0 == (sz = strlen(start)))
1.197 schwarze 1620: return 0;
1.1 kristaps 1621:
1622: end = &start[(int)sz - 1];
1623: if ('>' == *end || '"' == *end)
1624: end--;
1625:
1.50 kristaps 1626: if (end > start)
1.78 schwarze 1627: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1628: return 0;
1.1 kristaps 1629: }
1630:
1.178 schwarze 1631: static void
1.189 schwarze 1632: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1633: {
1.112 schwarze 1634: char *cp;
1.178 schwarze 1635: size_t sz;
1.1 kristaps 1636:
1.188 schwarze 1637: if (n->type != ROFFT_TEXT)
1.178 schwarze 1638: return;
1.25 schwarze 1639:
1.178 schwarze 1640: /* Skip function pointer punctuation. */
1.1 kristaps 1641:
1.178 schwarze 1642: cp = n->string;
1643: while (*cp == '(' || *cp == '*')
1.1 kristaps 1644: cp++;
1.178 schwarze 1645: sz = strcspn(cp, "()");
1.1 kristaps 1646:
1.178 schwarze 1647: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1648: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1649: putkeys(mpage, cp, sz, NAME_SYN);
1650: }
1.25 schwarze 1651:
1.178 schwarze 1652: static int
1.190 schwarze 1653: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1654: const struct roff_node *n)
1.178 schwarze 1655: {
1656:
1657: if (n->child == NULL)
1.197 schwarze 1658: return 0;
1.178 schwarze 1659:
1660: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1661:
1.178 schwarze 1662: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1663: if (n->type == ROFFT_TEXT)
1.78 schwarze 1664: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1665:
1.197 schwarze 1666: return 0;
1.173 schwarze 1667: }
1668:
1669: static int
1.190 schwarze 1670: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1671: const struct roff_node *n)
1.173 schwarze 1672: {
1.177 schwarze 1673:
1.188 schwarze 1674: if (n->type != ROFFT_HEAD)
1.197 schwarze 1675: return 1;
1.173 schwarze 1676:
1.178 schwarze 1677: if (n->child != NULL)
1678: parse_mdoc_fname(mpage, n->child);
1679:
1.197 schwarze 1680: return 0;
1.1 kristaps 1681: }
1682:
1.25 schwarze 1683: static int
1.190 schwarze 1684: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1685: const struct roff_node *n)
1.1 kristaps 1686: {
1.67 schwarze 1687: char *cp;
1.1 kristaps 1688:
1689: if (NULL == (n = n->child))
1.197 schwarze 1690: return 0;
1.1 kristaps 1691:
1.67 schwarze 1692: if (NULL == n->next) {
1.78 schwarze 1693: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1694: return 0;
1.67 schwarze 1695: }
1696:
1.120 schwarze 1697: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1698: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1699: free(cp);
1.197 schwarze 1700: return 0;
1.1 kristaps 1701: }
1702:
1.25 schwarze 1703: static int
1.190 schwarze 1704: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1705: const struct roff_node *n)
1.1 kristaps 1706: {
1707:
1.188 schwarze 1708: if (n->type == ROFFT_BODY)
1.194 schwarze 1709: deroff(&mpage->desc, n);
1.197 schwarze 1710: return 0;
1.1 kristaps 1711: }
1712:
1.25 schwarze 1713: static int
1.190 schwarze 1714: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1715: const struct roff_node *n)
1.1 kristaps 1716: {
1717:
1.129 schwarze 1718: if (SEC_NAME == n->sec)
1.133 schwarze 1719: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1720: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1721: if (n->child == NULL)
1722: putkey(mpage, meta->name, NAME_SYN);
1723: else
1724: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1725: }
1726: if ( ! (mpage->name_head_done ||
1727: n->child == NULL || n->child->string == NULL ||
1728: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1729: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1730: mpage->name_head_done = 1;
1.172 schwarze 1731: }
1.197 schwarze 1732: return 0;
1.1 kristaps 1733: }
1734:
1.25 schwarze 1735: static int
1.190 schwarze 1736: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1737: const struct roff_node *n)
1.1 kristaps 1738: {
1739:
1.197 schwarze 1740: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1741: }
1742:
1.50 kristaps 1743: static int
1.190 schwarze 1744: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1745: const struct roff_node *n)
1.1 kristaps 1746: {
1747:
1.197 schwarze 1748: return n->type == ROFFT_HEAD;
1.50 kristaps 1749: }
1.1 kristaps 1750:
1.50 kristaps 1751: static int
1.190 schwarze 1752: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1753: const struct roff_node *n)
1.50 kristaps 1754: {
1.1 kristaps 1755:
1.197 schwarze 1756: return n->type == ROFFT_BODY;
1.1 kristaps 1757: }
1758:
1.50 kristaps 1759: /*
1.66 schwarze 1760: * Add a string to the hash table for the current manual.
1761: * Each string has a bitmask telling which macros it belongs to.
1762: * When we finish the manual, we'll dump the table.
1.50 kristaps 1763: */
1764: static void
1.176 schwarze 1765: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1766: {
1.133 schwarze 1767: struct ohash *htab;
1.50 kristaps 1768: struct str *s;
1.111 schwarze 1769: const char *end;
1.65 schwarze 1770: unsigned int slot;
1.176 schwarze 1771: int i, mustfree;
1.25 schwarze 1772:
1.50 kristaps 1773: if (0 == sz)
1774: return;
1.111 schwarze 1775:
1.176 schwarze 1776: mustfree = render_string(&cp, &sz);
1777:
1.133 schwarze 1778: if (TYPE_Nm & v) {
1779: htab = &names;
1780: v &= name_mask;
1.169 schwarze 1781: if (v & NAME_FIRST)
1782: name_mask &= ~NAME_FIRST;
1.133 schwarze 1783: if (debug > 1)
1784: say(mpage->mlinks->file,
1.175 schwarze 1785: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1786: } else {
1787: htab = &strings;
1788: if (debug > 1)
1789: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1790: if ((uint64_t)1 << i & v)
1.133 schwarze 1791: say(mpage->mlinks->file,
1792: "Adding key %s=%*s",
1793: mansearch_keynames[i], sz, cp);
1.111 schwarze 1794: }
1.25 schwarze 1795:
1.65 schwarze 1796: end = cp + sz;
1.133 schwarze 1797: slot = ohash_qlookupi(htab, cp, &end);
1798: s = ohash_find(htab, slot);
1.25 schwarze 1799:
1.78 schwarze 1800: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1801: s->mask |= v;
1802: return;
1803: } else if (NULL == s) {
1.144 schwarze 1804: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1805: memcpy(s->key, cp, sz);
1.133 schwarze 1806: ohash_insert(htab, slot, s);
1.1 kristaps 1807: }
1.78 schwarze 1808: s->mpage = mpage;
1.50 kristaps 1809: s->mask = v;
1.176 schwarze 1810:
1811: if (mustfree)
1812: free(cp);
1.1 kristaps 1813: }
1814:
1.50 kristaps 1815: /*
1816: * Take a Unicode codepoint and produce its UTF-8 encoding.
1817: * This isn't the best way to do this, but it works.
1818: * The magic numbers are from the UTF-8 packaging.
1819: * They're not as scary as they seem: read the UTF-8 spec for details.
1820: */
1821: static size_t
1822: utf8(unsigned int cp, char out[7])
1.1 kristaps 1823: {
1.50 kristaps 1824: size_t rc;
1.1 kristaps 1825:
1.50 kristaps 1826: rc = 0;
1827: if (cp <= 0x0000007F) {
1828: rc = 1;
1829: out[0] = (char)cp;
1830: } else if (cp <= 0x000007FF) {
1831: rc = 2;
1832: out[0] = (cp >> 6 & 31) | 192;
1833: out[1] = (cp & 63) | 128;
1834: } else if (cp <= 0x0000FFFF) {
1835: rc = 3;
1836: out[0] = (cp >> 12 & 15) | 224;
1837: out[1] = (cp >> 6 & 63) | 128;
1838: out[2] = (cp & 63) | 128;
1839: } else if (cp <= 0x001FFFFF) {
1840: rc = 4;
1841: out[0] = (cp >> 18 & 7) | 240;
1842: out[1] = (cp >> 12 & 63) | 128;
1843: out[2] = (cp >> 6 & 63) | 128;
1844: out[3] = (cp & 63) | 128;
1845: } else if (cp <= 0x03FFFFFF) {
1846: rc = 5;
1847: out[0] = (cp >> 24 & 3) | 248;
1848: out[1] = (cp >> 18 & 63) | 128;
1849: out[2] = (cp >> 12 & 63) | 128;
1850: out[3] = (cp >> 6 & 63) | 128;
1851: out[4] = (cp & 63) | 128;
1852: } else if (cp <= 0x7FFFFFFF) {
1853: rc = 6;
1854: out[0] = (cp >> 30 & 1) | 252;
1855: out[1] = (cp >> 24 & 63) | 128;
1856: out[2] = (cp >> 18 & 63) | 128;
1857: out[3] = (cp >> 12 & 63) | 128;
1858: out[4] = (cp >> 6 & 63) | 128;
1859: out[5] = (cp & 63) | 128;
1860: } else
1.197 schwarze 1861: return 0;
1.1 kristaps 1862:
1.50 kristaps 1863: out[rc] = '\0';
1.197 schwarze 1864: return rc;
1.50 kristaps 1865: }
1.1 kristaps 1866:
1.50 kristaps 1867: /*
1.176 schwarze 1868: * If the string contains escape sequences,
1869: * replace it with an allocated rendering and return 1,
1870: * such that the caller can free it after use.
1871: * Otherwise, do nothing and return 0.
1.50 kristaps 1872: */
1.176 schwarze 1873: static int
1874: render_string(char **public, size_t *psz)
1.50 kristaps 1875: {
1.176 schwarze 1876: const char *src, *scp, *addcp, *seq;
1877: char *dst;
1878: size_t ssz, dsz, addsz;
1.114 schwarze 1879: char utfbuf[7], res[6];
1.176 schwarze 1880: int seqlen, unicode;
1.50 kristaps 1881:
1882: res[0] = '\\';
1883: res[1] = '\t';
1884: res[2] = ASCII_NBRSP;
1885: res[3] = ASCII_HYPH;
1.114 schwarze 1886: res[4] = ASCII_BREAK;
1887: res[5] = '\0';
1.1 kristaps 1888:
1.176 schwarze 1889: src = scp = *public;
1890: ssz = *psz;
1891: dst = NULL;
1892: dsz = 0;
1893:
1894: while (scp < src + *psz) {
1895:
1896: /* Leave normal characters unchanged. */
1897:
1898: if (strchr(res, *scp) == NULL) {
1899: if (dst != NULL)
1900: dst[dsz++] = *scp;
1901: scp++;
1902: continue;
1903: }
1.46 kristaps 1904:
1.50 kristaps 1905: /*
1.176 schwarze 1906: * Found something that requires replacing,
1907: * make sure we have a destination buffer.
1.50 kristaps 1908: */
1.176 schwarze 1909:
1910: if (dst == NULL) {
1911: dst = mandoc_malloc(ssz + 1);
1912: dsz = scp - src;
1913: memcpy(dst, src, dsz);
1.50 kristaps 1914: }
1.46 kristaps 1915:
1.176 schwarze 1916: /* Handle single-char special characters. */
1917:
1918: switch (*scp) {
1919: case '\\':
1920: break;
1.141 schwarze 1921: case '\t':
1922: case ASCII_NBRSP:
1.176 schwarze 1923: dst[dsz++] = ' ';
1924: scp++;
1925: continue;
1926: case ASCII_HYPH:
1927: dst[dsz++] = '-';
1.114 schwarze 1928: /* FALLTHROUGH */
1.141 schwarze 1929: case ASCII_BREAK:
1.176 schwarze 1930: scp++;
1.50 kristaps 1931: continue;
1.114 schwarze 1932: default:
1.176 schwarze 1933: abort();
1.114 schwarze 1934: }
1.46 kristaps 1935:
1.50 kristaps 1936: /*
1.176 schwarze 1937: * Found an escape sequence.
1938: * Read past the slash, then parse it.
1939: * Ignore everything except characters.
1.50 kristaps 1940: */
1.95 schwarze 1941:
1.176 schwarze 1942: scp++;
1943: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1944: continue;
1.1 kristaps 1945:
1.50 kristaps 1946: /*
1.95 schwarze 1947: * Render the special character
1948: * as either UTF-8 or ASCII.
1.50 kristaps 1949: */
1.95 schwarze 1950:
1951: if (write_utf8) {
1.176 schwarze 1952: unicode = mchars_spec2cp(mchars, seq, seqlen);
1953: if (unicode <= 0)
1.95 schwarze 1954: continue;
1.176 schwarze 1955: addsz = utf8(unicode, utfbuf);
1956: if (addsz == 0)
1.95 schwarze 1957: continue;
1.176 schwarze 1958: addcp = utfbuf;
1.95 schwarze 1959: } else {
1.176 schwarze 1960: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1961: if (addcp == NULL)
1.95 schwarze 1962: continue;
1.176 schwarze 1963: if (*addcp == ASCII_NBRSP) {
1964: addcp = " ";
1965: addsz = 1;
1.95 schwarze 1966: }
1967: }
1.1 kristaps 1968:
1.50 kristaps 1969: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1970:
1.176 schwarze 1971: ssz += addsz;
1972: dst = mandoc_realloc(dst, ssz + 1);
1973: memcpy(dst + dsz, addcp, addsz);
1974: dsz += addsz;
1975: }
1976: if (dst != NULL) {
1977: *public = dst;
1978: *psz = dsz;
1979: }
1980:
1981: /* Trim trailing whitespace and NUL-terminate. */
1982:
1983: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1984: --*psz;
1985: if (dst != NULL) {
1986: (*public)[*psz] = '\0';
1.197 schwarze 1987: return 1;
1.176 schwarze 1988: } else
1.197 schwarze 1989: return 0;
1.1 kristaps 1990: }
1991:
1.118 schwarze 1992: static void
1993: dbadd_mlink(const struct mlink *mlink)
1994: {
1995: size_t i;
1996:
1997: i = 1;
1998: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1999: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2000: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2001: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2002: SQL_STEP(stmts[STMT_INSERT_LINK]);
2003: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2004: }
2005:
2006: static void
2007: dbadd_mlink_name(const struct mlink *mlink)
2008: {
1.175 schwarze 2009: uint64_t bits;
1.169 schwarze 2010: size_t i;
2011:
2012: dbadd_mlink(mlink);
1.160 schwarze 2013:
2014: i = 1;
1.175 schwarze 2015: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2016: bits = NAME_FILE & NAME_MASK;
2017: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2018: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2019: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2020: }
2021:
2022: i = 1;
2023: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2024: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2025: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2026: SQL_STEP(stmts[STMT_INSERT_NAME]);
2027: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2028: }
2029:
1.14 schwarze 2030: /*
1.50 kristaps 2031: * Flush the current page's terms (and their bits) into the database.
2032: * Wrap the entire set of additions in a transaction to make sqlite be a
2033: * little faster.
1.96 schwarze 2034: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2035: */
2036: static void
1.176 schwarze 2037: dbadd(struct mpage *mpage)
1.14 schwarze 2038: {
1.87 schwarze 2039: struct mlink *mlink;
1.50 kristaps 2040: struct str *key;
1.176 schwarze 2041: char *cp;
1.52 kristaps 2042: size_t i;
1.66 schwarze 2043: unsigned int slot;
1.176 schwarze 2044: int mustfree;
1.14 schwarze 2045:
1.128 schwarze 2046: mlink = mpage->mlinks;
1.43 kristaps 2047:
1.128 schwarze 2048: if (nodb) {
1.147 schwarze 2049: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2050: key = ohash_next(&names, &slot))
1.147 schwarze 2051: free(key);
2052: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2053: key = ohash_next(&strings, &slot))
1.147 schwarze 2054: free(key);
1.145 schwarze 2055: if (0 == debug)
2056: return;
1.128 schwarze 2057: while (NULL != mlink) {
2058: fputs(mlink->name, stdout);
2059: if (NULL == mlink->next ||
2060: strcmp(mlink->dsec, mlink->next->dsec) ||
2061: strcmp(mlink->fsec, mlink->next->fsec) ||
2062: strcmp(mlink->arch, mlink->next->arch)) {
2063: putchar('(');
2064: if ('\0' == *mlink->dsec)
2065: fputs(mlink->fsec, stdout);
2066: else
2067: fputs(mlink->dsec, stdout);
2068: if ('\0' != *mlink->arch)
2069: printf("/%s", mlink->arch);
2070: putchar(')');
2071: }
2072: mlink = mlink->next;
2073: if (NULL != mlink)
2074: fputs(", ", stdout);
2075: }
1.132 schwarze 2076: printf(" - %s\n", mpage->desc);
1.14 schwarze 2077: return;
1.128 schwarze 2078: }
2079:
2080: if (debug)
2081: say(mlink->file, "Adding to database");
1.28 kristaps 2082:
1.176 schwarze 2083: cp = mpage->desc;
2084: i = strlen(cp);
2085: mustfree = render_string(&cp, &i);
1.52 kristaps 2086: i = 1;
1.176 schwarze 2087: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2088: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2089: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2090: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2091: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2092: if (mustfree)
2093: free(cp);
1.81 schwarze 2094:
1.128 schwarze 2095: while (NULL != mlink) {
1.118 schwarze 2096: dbadd_mlink(mlink);
1.128 schwarze 2097: mlink = mlink->next;
2098: }
1.134 schwarze 2099: mlink = mpage->mlinks;
1.50 kristaps 2100:
1.133 schwarze 2101: for (key = ohash_first(&names, &slot); NULL != key;
2102: key = ohash_next(&names, &slot)) {
2103: assert(key->mpage == mpage);
2104: i = 1;
2105: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2106: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2107: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2108: SQL_STEP(stmts[STMT_INSERT_NAME]);
2109: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2110: free(key);
2111: }
1.66 schwarze 2112: for (key = ohash_first(&strings, &slot); NULL != key;
2113: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2114: assert(key->mpage == mpage);
1.52 kristaps 2115: i = 1;
2116: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2117: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2118: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2119: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2120: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2121: free(key);
1.38 schwarze 2122: }
1.14 schwarze 2123: }
2124:
1.5 kristaps 2125: static void
1.59 schwarze 2126: dbprune(void)
1.5 kristaps 2127: {
1.78 schwarze 2128: struct mpage *mpage;
1.82 schwarze 2129: struct mlink *mlink;
1.52 kristaps 2130: size_t i;
1.80 schwarze 2131: unsigned int slot;
1.5 kristaps 2132:
1.106 schwarze 2133: if (0 == nodb)
2134: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2135:
1.106 schwarze 2136: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2137: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2138: mlink = mpage->mlinks;
1.125 schwarze 2139: if (debug)
1.106 schwarze 2140: say(mlink->file, "Deleting from database");
2141: if (nodb)
2142: continue;
2143: for ( ; NULL != mlink; mlink = mlink->next) {
2144: i = 1;
2145: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2146: i, mlink->dsec);
2147: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2148: i, mlink->arch);
2149: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2150: i, mlink->name);
2151: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2152: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2153: }
1.5 kristaps 2154: }
1.106 schwarze 2155:
2156: if (0 == nodb)
2157: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2158: }
2159:
1.4 kristaps 2160: /*
1.50 kristaps 2161: * Close an existing database and its prepared statements.
2162: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2163: */
1.35 kristaps 2164: static void
1.59 schwarze 2165: dbclose(int real)
1.4 kristaps 2166: {
1.50 kristaps 2167: size_t i;
1.115 schwarze 2168: int status;
2169: pid_t child;
1.4 kristaps 2170:
1.50 kristaps 2171: if (nodb)
1.38 schwarze 2172: return;
1.50 kristaps 2173:
2174: for (i = 0; i < STMT__MAX; i++) {
2175: sqlite3_finalize(stmts[i]);
2176: stmts[i] = NULL;
1.4 kristaps 2177: }
2178:
1.50 kristaps 2179: sqlite3_close(db);
2180: db = NULL;
1.12 schwarze 2181:
1.50 kristaps 2182: if (real)
2183: return;
1.12 schwarze 2184:
1.115 schwarze 2185: if ('\0' == *tempfilename) {
2186: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2187: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2188: say(MANDOC_DB, "&rename");
1.115 schwarze 2189: }
2190: return;
2191: }
2192:
2193: switch (child = fork()) {
1.141 schwarze 2194: case -1:
1.115 schwarze 2195: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2196: say("", "&fork cmp");
1.115 schwarze 2197: return;
1.141 schwarze 2198: case 0:
1.115 schwarze 2199: execlp("cmp", "cmp", "-s",
1.196 schwarze 2200: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2201: say("", "&exec cmp");
1.115 schwarze 2202: exit(0);
2203: default:
2204: break;
2205: }
2206: if (-1 == waitpid(child, &status, 0)) {
2207: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2208: say("", "&wait cmp");
1.115 schwarze 2209: } else if (WIFSIGNALED(status)) {
2210: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2211: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2212: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2213: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2214: say(MANDOC_DB,
2215: "Data changed, but cannot replace database");
2216: }
2217:
2218: *strrchr(tempfilename, '/') = '\0';
2219: switch (child = fork()) {
1.141 schwarze 2220: case -1:
1.115 schwarze 2221: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2222: say("", "&fork rm");
1.115 schwarze 2223: return;
1.141 schwarze 2224: case 0:
1.196 schwarze 2225: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2226: say("", "&exec rm");
1.115 schwarze 2227: exit((int)MANDOCLEVEL_SYSERR);
2228: default:
2229: break;
2230: }
2231: if (-1 == waitpid(child, &status, 0)) {
2232: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2233: say("", "&wait rm");
1.115 schwarze 2234: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2235: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2236: say("", "%s: Cannot remove temporary directory",
2237: tempfilename);
1.59 schwarze 2238: }
1.50 kristaps 2239: }
1.14 schwarze 2240:
1.50 kristaps 2241: /*
2242: * This is straightforward stuff.
2243: * Open a database connection to a "temporary" database, then open a set
2244: * of prepared statements we'll use over and over again.
2245: * If "real" is set, we use the existing database; if not, we truncate a
2246: * temporary one.
2247: * Must be matched by dbclose().
2248: */
2249: static int
1.59 schwarze 2250: dbopen(int real)
1.50 kristaps 2251: {
1.115 schwarze 2252: const char *sql;
1.50 kristaps 2253: int rc, ofl;
1.12 schwarze 2254:
1.141 schwarze 2255: if (nodb)
1.197 schwarze 2256: return 1;
1.12 schwarze 2257:
1.115 schwarze 2258: *tempfilename = '\0';
1.63 schwarze 2259: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2260:
2261: if (real) {
2262: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2263: if (SQLITE_OK != rc) {
1.63 schwarze 2264: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2265: if (SQLITE_CANTOPEN != rc)
2266: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2267: return 0;
1.63 schwarze 2268: }
1.115 schwarze 2269: goto prepare_statements;
2270: }
2271:
2272: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2273:
1.115 schwarze 2274: remove(MANDOC_DB "~");
2275: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2276: if (SQLITE_OK == rc)
1.115 schwarze 2277: goto create_tables;
1.116 schwarze 2278: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2279: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2280: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2281: return 0;
1.50 kristaps 2282: }
1.12 schwarze 2283:
1.143 schwarze 2284: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2285: sizeof(tempfilename));
1.115 schwarze 2286: if (NULL == mkdtemp(tempfilename)) {
2287: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2288: say("", "&%s", tempfilename);
1.197 schwarze 2289: return 0;
1.115 schwarze 2290: }
1.143 schwarze 2291: (void)strlcat(tempfilename, "/" MANDOC_DB,
2292: sizeof(tempfilename));
1.115 schwarze 2293: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2294: if (SQLITE_OK != rc) {
1.59 schwarze 2295: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2296: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2297: return 0;
1.50 kristaps 2298: }
1.12 schwarze 2299:
1.115 schwarze 2300: create_tables:
1.81 schwarze 2301: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2302: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2303: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2304: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2305: ");\n"
2306: "\n"
1.81 schwarze 2307: "CREATE TABLE \"mlinks\" (\n"
2308: " \"sec\" TEXT NOT NULL,\n"
2309: " \"arch\" TEXT NOT NULL,\n"
2310: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2311: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2312: "ON DELETE CASCADE\n"
1.81 schwarze 2313: ");\n"
1.136 schwarze 2314: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2315: "\n"
1.133 schwarze 2316: "CREATE TABLE \"names\" (\n"
2317: " \"bits\" INTEGER NOT NULL,\n"
2318: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2319: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2320: "ON DELETE CASCADE,\n"
2321: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2322: ");\n"
2323: "\n"
1.50 kristaps 2324: "CREATE TABLE \"keys\" (\n"
2325: " \"bits\" INTEGER NOT NULL,\n"
2326: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2327: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2328: "ON DELETE CASCADE\n"
1.136 schwarze 2329: ");\n"
2330: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2331:
1.50 kristaps 2332: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2333: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2334: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2335: sqlite3_close(db);
1.197 schwarze 2336: return 0;
1.50 kristaps 2337: }
1.4 kristaps 2338:
1.57 schwarze 2339: prepare_statements:
1.146 schwarze 2340: if (SQLITE_OK != sqlite3_exec(db,
2341: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2342: exitcode = (int)MANDOCLEVEL_SYSERR;
2343: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2344: sqlite3_errmsg(db));
2345: sqlite3_close(db);
1.197 schwarze 2346: return 0;
1.146 schwarze 2347: }
2348:
1.137 schwarze 2349: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2350: "(SELECT pageid FROM mlinks WHERE "
2351: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2352: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2353: sql = "INSERT INTO mpages "
1.132 schwarze 2354: "(desc,form) VALUES (?,?)";
1.81 schwarze 2355: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2356: sql = "INSERT INTO mlinks "
1.104 schwarze 2357: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2358: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2359: sql = "SELECT bits FROM names where pageid = ?";
2360: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2361: sql = "INSERT INTO names "
2362: "(bits,name,pageid) VALUES (?,?,?)";
2363: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2364: sql = "INSERT INTO keys "
1.81 schwarze 2365: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2366: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2367:
2368: #ifndef __APPLE__
2369: /*
2370: * When opening a new database, we can turn off
2371: * synchronous mode for much better performance.
2372: */
2373:
1.146 schwarze 2374: if (real && SQLITE_OK != sqlite3_exec(db,
2375: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2376: exitcode = (int)MANDOCLEVEL_SYSERR;
2377: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2378: sqlite3_errmsg(db));
1.146 schwarze 2379: sqlite3_close(db);
1.197 schwarze 2380: return 0;
1.146 schwarze 2381: }
1.70 schwarze 2382: #endif
2383:
1.197 schwarze 2384: return 1;
1.50 kristaps 2385: }
1.5 kristaps 2386:
1.50 kristaps 2387: static void *
1.152 schwarze 2388: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2389: {
1.12 schwarze 2390:
1.197 schwarze 2391: return mandoc_calloc(n, sz);
1.50 kristaps 2392: }
1.12 schwarze 2393:
1.50 kristaps 2394: static void *
2395: hash_alloc(size_t sz, void *arg)
2396: {
1.5 kristaps 2397:
1.197 schwarze 2398: return mandoc_malloc(sz);
1.50 kristaps 2399: }
1.41 kristaps 2400:
1.50 kristaps 2401: static void
1.152 schwarze 2402: hash_free(void *p, void *arg)
1.50 kristaps 2403: {
1.4 kristaps 2404:
1.50 kristaps 2405: free(p);
1.4 kristaps 2406: }
2407:
1.50 kristaps 2408: static int
1.165 schwarze 2409: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2410: {
1.59 schwarze 2411: static char startdir[PATH_MAX];
1.151 schwarze 2412: static int getcwd_status; /* 1 = ok, 2 = failure */
2413: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2414: char *cp;
1.4 kristaps 2415:
1.59 schwarze 2416: /*
1.151 schwarze 2417: * Remember the original working directory, if possible.
2418: * This will be needed if the second or a later directory
2419: * on the command line is given as a relative path.
2420: * Do not error out if the current directory is not
2421: * searchable: Maybe it won't be needed after all.
2422: */
2423: if (0 == getcwd_status) {
2424: if (NULL == getcwd(startdir, sizeof(startdir))) {
2425: getcwd_status = 2;
2426: (void)strlcpy(startdir, strerror(errno),
2427: sizeof(startdir));
2428: } else
2429: getcwd_status = 1;
2430: }
2431:
2432: /*
2433: * We are leaving the old base directory.
2434: * Do not use it any longer, not even for messages.
2435: */
2436: *basedir = '\0';
2437:
2438: /*
2439: * If and only if the directory was changed earlier and
2440: * the next directory to process is given as a relative path,
2441: * first go back, or bail out if that is impossible.
1.59 schwarze 2442: */
1.151 schwarze 2443: if (chdir_status && '/' != *targetdir) {
2444: if (2 == getcwd_status) {
1.59 schwarze 2445: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2446: say("", "getcwd: %s", startdir);
1.197 schwarze 2447: return 0;
1.59 schwarze 2448: }
1.151 schwarze 2449: if (-1 == chdir(startdir)) {
1.59 schwarze 2450: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2451: say("", "&chdir %s", startdir);
1.197 schwarze 2452: return 0;
1.59 schwarze 2453: }
2454: }
1.151 schwarze 2455:
2456: /*
2457: * Always resolve basedir to the canonicalized absolute
2458: * pathname and append a trailing slash, such that
2459: * we can reliably check whether files are inside.
2460: */
1.59 schwarze 2461: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2462: if (report_baddir || errno != ENOENT) {
2463: exitcode = (int)MANDOCLEVEL_BADARG;
2464: say("", "&%s: realpath", targetdir);
2465: }
1.197 schwarze 2466: return 0;
1.59 schwarze 2467: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2468: if (report_baddir || errno != ENOENT) {
2469: exitcode = (int)MANDOCLEVEL_BADARG;
2470: say("", "&chdir");
2471: }
1.197 schwarze 2472: return 0;
1.150 schwarze 2473: }
1.151 schwarze 2474: chdir_status = 1;
1.150 schwarze 2475: cp = strchr(basedir, '\0');
2476: if ('/' != cp[-1]) {
2477: if (cp - basedir >= PATH_MAX - 1) {
2478: exitcode = (int)MANDOCLEVEL_SYSERR;
2479: say("", "Filename too long");
1.197 schwarze 2480: return 0;
1.150 schwarze 2481: }
2482: *cp++ = '/';
2483: *cp = '\0';
1.4 kristaps 2484: }
1.197 schwarze 2485: return 1;
1.56 schwarze 2486: }
2487:
2488: static void
1.59 schwarze 2489: say(const char *file, const char *format, ...)
1.56 schwarze 2490: {
2491: va_list ap;
1.123 schwarze 2492: int use_errno;
1.56 schwarze 2493:
1.59 schwarze 2494: if ('\0' != *basedir)
2495: fprintf(stderr, "%s", basedir);
2496: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2497: fputc('/', stderr);
1.56 schwarze 2498: if ('\0' != *file)
1.59 schwarze 2499: fprintf(stderr, "%s", file);
2500:
1.123 schwarze 2501: use_errno = 1;
2502: if (NULL != format) {
2503: switch (*format) {
1.141 schwarze 2504: case '&':
1.123 schwarze 2505: format++;
2506: break;
1.141 schwarze 2507: case '\0':
1.123 schwarze 2508: format = NULL;
2509: break;
2510: default:
2511: use_errno = 0;
2512: break;
2513: }
2514: }
2515: if (NULL != format) {
2516: if ('\0' != *basedir || '\0' != *file)
2517: fputs(": ", stderr);
2518: va_start(ap, format);
2519: vfprintf(stderr, format, ap);
2520: va_end(ap);
2521: }
2522: if (use_errno) {
2523: if ('\0' != *basedir || '\0' != *file || NULL != format)
2524: fputs(": ", stderr);
1.59 schwarze 2525: perror(NULL);
1.123 schwarze 2526: } else
2527: fputc('\n', stderr);
1.1 kristaps 2528: }
CVSweb