Annotation of mandoc/mandocdb.c, Revision 1.199
1.199 ! schwarze 1: /* $Id: mandocdb.c,v 1.198 2015/10/11 21:12:54 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);
1.50 kristaps 587: if (NULL == f) {
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:
596: while (NULL != (ff = fts_read(f))) {
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:
1.139 schwarze 605: if (NULL == realpath(path, buf)) {
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. */
620: if (-1 == stat(path, ff->fts_statp)) {
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:
1.60 schwarze 632: if (0 == strcmp(path, MANDOC_DB))
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;
641: while (NULL == fsec) {
642: fsec = strrchr(ff->fts_name, '.');
643: if (NULL == fsec || strcmp(fsec+1, "gz"))
644: break;
645: gzip = 1;
646: *fsec = '\0';
647: fsec = NULL;
648: }
649: if (NULL == fsec) {
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: }
1.84 schwarze 656: } else if (0 == strcmp(++fsec, "html")) {
1.60 schwarze 657: if (warnings)
658: say(path, "Skip html");
659: continue;
1.84 schwarze 660: } else if (0 == strcmp(fsec, "ps")) {
1.60 schwarze 661: if (warnings)
662: say(path, "Skip ps");
663: continue;
1.84 schwarze 664: } else if (0 == strcmp(fsec, "pdf")) {
1.60 schwarze 665: if (warnings)
666: say(path, "Skip pdf");
667: continue;
668: } else if ( ! use_all &&
1.185 schwarze 669: ((FORM_SRC == dform &&
670: strncmp(fsec, dsec, strlen(dsec))) ||
1.84 schwarze 671: (FORM_CAT == dform && strcmp(fsec, "0")))) {
1.60 schwarze 672: if (warnings)
673: say(path, "Wrong filename suffix");
674: continue;
1.66 schwarze 675: } else
1.84 schwarze 676: fsec[-1] = '\0';
1.94 schwarze 677:
1.84 schwarze 678: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.143 schwarze 679: if (strlcpy(mlink->file, path,
680: sizeof(mlink->file)) >=
681: sizeof(mlink->file)) {
682: say(path, "Filename too long");
683: free(mlink);
684: continue;
685: }
1.84 schwarze 686: mlink->dform = dform;
1.94 schwarze 687: mlink->dsec = dsec;
688: mlink->arch = arch;
689: mlink->name = ff->fts_name;
690: mlink->fsec = fsec;
1.124 schwarze 691: mlink->gzip = gzip;
1.84 schwarze 692: mlink_add(mlink, ff->fts_statp);
1.50 kristaps 693: continue;
1.139 schwarze 694:
1.141 schwarze 695: case FTS_D:
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;
716: if (FTS_DP == ff->fts_info)
717: break;
1.1 kristaps 718:
1.50 kristaps 719: if (0 == strncmp(cp, "man", 3)) {
720: dform = FORM_SRC;
1.66 schwarze 721: dsec = cp + 3;
1.50 kristaps 722: } else if (0 == strncmp(cp, "cat", 3)) {
723: dform = FORM_CAT;
1.66 schwarze 724: dsec = cp + 3;
1.94 schwarze 725: } else {
726: dform = FORM_NONE;
727: dsec = NULL;
1.50 kristaps 728: }
1.1 kristaps 729:
1.141 schwarze 730: if (NULL != dsec || use_all)
1.50 kristaps 731: break;
1.1 kristaps 732:
1.56 schwarze 733: if (warnings)
1.59 schwarze 734: say(path, "Unknown directory part");
1.50 kristaps 735: fts_set(f, ff, FTS_SKIP);
736: break;
1.141 schwarze 737: case 2:
1.50 kristaps 738: /*
739: * Possibly our architecture.
740: * If we're descending, keep tabs on it.
741: */
742: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 743: arch = ff->fts_name;
1.94 schwarze 744: else
745: arch = NULL;
1.50 kristaps 746: break;
747: default:
748: if (FTS_DP == ff->fts_info || use_all)
749: break;
1.56 schwarze 750: if (warnings)
1.59 schwarze 751: say(path, "Extraneous directory part");
1.50 kristaps 752: fts_set(f, ff, FTS_SKIP);
753: break;
1.5 kristaps 754: }
1.50 kristaps 755: }
756:
757: fts_close(f);
1.197 schwarze 758: return 1;
1.50 kristaps 759: }
1.5 kristaps 760:
1.50 kristaps 761: /*
1.89 schwarze 762: * Add a file to the mlinks table.
1.50 kristaps 763: * Do not verify that it's a "valid" looking manpage (we'll do that
764: * later).
765: *
766: * Try to infer the manual section, architecture, and page name from the
767: * path, assuming it looks like
768: *
1.141 schwarze 769: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 770: * or
771: * [./]cat<section>[/<arch>]/<name>.0
772: *
773: * See treescan() for the fts(3) version of this.
774: */
775: static void
1.59 schwarze 776: filescan(const char *file)
1.50 kristaps 777: {
1.59 schwarze 778: char buf[PATH_MAX];
1.84 schwarze 779: struct stat st;
780: struct mlink *mlink;
1.59 schwarze 781: char *p, *start;
1.5 kristaps 782:
1.50 kristaps 783: assert(use_all);
1.5 kristaps 784:
1.50 kristaps 785: if (0 == strncmp(file, "./", 2))
786: file += 2;
1.5 kristaps 787:
1.139 schwarze 788: /*
789: * We have to do lstat(2) before realpath(3) loses
790: * the information whether this is a symbolic link.
791: * We need to know that because for symbolic links,
792: * we want to use the orginal file name, while for
793: * regular files, we want to use the real path.
794: */
795: if (-1 == lstat(file, &st)) {
796: exitcode = (int)MANDOCLEVEL_BADARG;
797: say(file, "&lstat");
798: return;
799: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
800: exitcode = (int)MANDOCLEVEL_BADARG;
801: say(file, "Not a regular file");
802: return;
803: }
804:
805: /*
806: * We have to resolve the file name to the real path
807: * in any case for the base directory check.
808: */
1.59 schwarze 809: if (NULL == realpath(file, buf)) {
810: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 811: say(file, "&realpath");
1.59 schwarze 812: return;
1.106 schwarze 813: }
814:
1.150 schwarze 815: if (OP_TEST == op)
1.106 schwarze 816: start = buf;
1.150 schwarze 817: else if (strstr(buf, basedir) == buf)
818: start = buf + strlen(basedir);
1.184 schwarze 819: #ifdef HOMEBREWDIR
820: else if (strstr(buf, HOMEBREWDIR) == buf)
821: start = buf;
822: #endif
1.106 schwarze 823: else {
1.59 schwarze 824: exitcode = (int)MANDOCLEVEL_BADARG;
825: say("", "%s: outside base directory", buf);
826: return;
1.106 schwarze 827: }
828:
1.139 schwarze 829: /*
830: * Now we are sure the file is inside our tree.
831: * If it is a symbolic link, ignore the real path
832: * and use the original name.
833: * This implies passing stuff like "cat1/../man1/foo.1"
834: * on the command line won't work. So don't do that.
835: * Note the stat(2) can still fail if the link target
836: * doesn't exist.
837: */
838: if (S_IFLNK & st.st_mode) {
839: if (-1 == stat(buf, &st)) {
840: exitcode = (int)MANDOCLEVEL_BADARG;
841: say(file, "&stat");
842: return;
843: }
1.143 schwarze 844: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
845: say(file, "Filename too long");
846: return;
847: }
1.150 schwarze 848: start = buf;
849: if (OP_TEST != op && strstr(buf, basedir) == buf)
850: start += strlen(basedir);
1.50 kristaps 851: }
1.106 schwarze 852:
1.84 schwarze 853: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 854: mlink->dform = FORM_NONE;
1.143 schwarze 855: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
856: sizeof(mlink->file)) {
857: say(start, "Filename too long");
1.180 schwarze 858: free(mlink);
1.143 schwarze 859: return;
860: }
1.17 schwarze 861:
1.50 kristaps 862: /*
863: * First try to guess our directory structure.
864: * If we find a separator, try to look for man* or cat*.
865: * If we find one of these and what's underneath is a directory,
866: * assume it's an architecture.
867: */
868: if (NULL != (p = strchr(start, '/'))) {
869: *p++ = '\0';
870: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 871: mlink->dform = FORM_SRC;
1.94 schwarze 872: mlink->dsec = start + 3;
1.50 kristaps 873: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 874: mlink->dform = FORM_CAT;
1.94 schwarze 875: mlink->dsec = start + 3;
1.17 schwarze 876: }
1.5 kristaps 877:
1.50 kristaps 878: start = p;
1.84 schwarze 879: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 880: *p++ = '\0';
1.94 schwarze 881: mlink->arch = start;
1.50 kristaps 882: start = p;
1.84 schwarze 883: }
1.50 kristaps 884: }
885:
886: /*
887: * Now check the file suffix.
888: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
889: */
890: p = strrchr(start, '\0');
891: while (p-- > start && '/' != *p && '.' != *p)
892: /* Loop. */ ;
893:
894: if ('.' == *p) {
895: *p++ = '\0';
1.94 schwarze 896: mlink->fsec = p;
1.5 kristaps 897: }
898:
1.10 kristaps 899: /*
1.50 kristaps 900: * Now try to parse the name.
901: * Use the filename portion of the path.
1.10 kristaps 902: */
1.84 schwarze 903: mlink->name = start;
1.50 kristaps 904: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 905: mlink->name = p + 1;
1.50 kristaps 906: *p = '\0';
1.84 schwarze 907: }
908: mlink_add(mlink, &st);
1.50 kristaps 909: }
1.5 kristaps 910:
1.50 kristaps 911: static void
1.84 schwarze 912: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 913: {
1.83 schwarze 914: struct inodev inodev;
1.78 schwarze 915: struct mpage *mpage;
1.83 schwarze 916: unsigned int slot;
1.50 kristaps 917:
1.84 schwarze 918: assert(NULL != mlink->file);
1.50 kristaps 919:
1.94 schwarze 920: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
921: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
922: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
923: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 924:
925: if ('0' == *mlink->fsec) {
926: free(mlink->fsec);
927: mlink->fsec = mandoc_strdup(mlink->dsec);
928: mlink->fform = FORM_CAT;
929: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
930: mlink->fform = FORM_SRC;
1.74 schwarze 931: else
1.84 schwarze 932: mlink->fform = FORM_NONE;
1.82 schwarze 933:
1.83 schwarze 934: slot = ohash_qlookup(&mlinks, mlink->file);
935: assert(NULL == ohash_find(&mlinks, slot));
936: ohash_insert(&mlinks, slot, mlink);
937:
1.186 schwarze 938: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 939: inodev.st_ino = st->st_ino;
940: inodev.st_dev = st->st_dev;
941: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
942: sizeof(struct inodev), inodev.st_ino);
943: mpage = ohash_find(&mpages, slot);
944: if (NULL == mpage) {
945: mpage = mandoc_calloc(1, sizeof(struct mpage));
946: mpage->inodev.st_ino = inodev.st_ino;
947: mpage->inodev.st_dev = inodev.st_dev;
948: ohash_insert(&mpages, slot, mpage);
949: } else
1.85 schwarze 950: mlink->next = mpage->mlinks;
1.82 schwarze 951: mpage->mlinks = mlink;
1.118 schwarze 952: mlink->mpage = mpage;
1.82 schwarze 953: }
954:
955: static void
956: mlink_free(struct mlink *mlink)
957: {
958:
959: free(mlink->dsec);
960: free(mlink->arch);
961: free(mlink->name);
962: free(mlink->fsec);
963: free(mlink);
1.50 kristaps 964: }
1.3 kristaps 965:
1.50 kristaps 966: static void
1.78 schwarze 967: mpages_free(void)
1.50 kristaps 968: {
1.78 schwarze 969: struct mpage *mpage;
1.82 schwarze 970: struct mlink *mlink;
1.80 schwarze 971: unsigned int slot;
1.3 kristaps 972:
1.80 schwarze 973: mpage = ohash_first(&mpages, &slot);
974: while (NULL != mpage) {
1.82 schwarze 975: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 976: mpage->mlinks = mlink->next;
1.82 schwarze 977: mlink_free(mlink);
978: }
1.78 schwarze 979: free(mpage->sec);
980: free(mpage->arch);
1.82 schwarze 981: free(mpage->title);
982: free(mpage->desc);
1.78 schwarze 983: free(mpage);
1.80 schwarze 984: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 985: }
986: }
1.38 schwarze 987:
1.50 kristaps 988: /*
1.89 schwarze 989: * For each mlink to the mpage, check whether the path looks like
990: * it is formatted, and if it does, check whether a source manual
991: * exists by the same name, ignoring the suffix.
992: * If both conditions hold, drop the mlink.
993: */
994: static void
995: mlinks_undupe(struct mpage *mpage)
996: {
997: char buf[PATH_MAX];
998: struct mlink **prev;
999: struct mlink *mlink;
1000: char *bufp;
1001:
1002: mpage->form = FORM_CAT;
1.90 schwarze 1003: prev = &mpage->mlinks;
1004: while (NULL != (mlink = *prev)) {
1.89 schwarze 1005: if (FORM_CAT != mlink->dform) {
1006: mpage->form = FORM_NONE;
1.90 schwarze 1007: goto nextlink;
1.89 schwarze 1008: }
1.143 schwarze 1009: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1010: bufp = strstr(buf, "cat");
1011: assert(NULL != bufp);
1012: memcpy(bufp, "man", 3);
1013: if (NULL != (bufp = strrchr(buf, '.')))
1014: *++bufp = '\0';
1.143 schwarze 1015: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1016: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1017: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1018: goto nextlink;
1.89 schwarze 1019: if (warnings)
1020: say(mlink->file, "Man source exists: %s", buf);
1021: if (use_all)
1.90 schwarze 1022: goto nextlink;
1.89 schwarze 1023: *prev = mlink->next;
1024: mlink_free(mlink);
1.90 schwarze 1025: continue;
1026: nextlink:
1027: prev = &(*prev)->next;
1.89 schwarze 1028: }
1029: }
1030:
1.131 schwarze 1031: static void
1.93 schwarze 1032: mlink_check(struct mpage *mpage, struct mlink *mlink)
1033: {
1.131 schwarze 1034: struct str *str;
1035: unsigned int slot;
1.93 schwarze 1036:
1037: /*
1038: * Check whether the manual section given in a file
1039: * agrees with the directory where the file is located.
1040: * Some manuals have suffixes like (3p) on their
1041: * section number either inside the file or in the
1042: * directory name, some are linked into more than one
1043: * section, like encrypt(1) = makekey(8).
1044: */
1045:
1046: if (FORM_SRC == mpage->form &&
1.131 schwarze 1047: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1048: say(mlink->file, "Section \"%s\" manual in %s directory",
1049: mpage->sec, mlink->dsec);
1050:
1051: /*
1052: * Manual page directories exist for each kernel
1053: * architecture as returned by machine(1).
1054: * However, many manuals only depend on the
1055: * application architecture as returned by arch(1).
1056: * For example, some (2/ARM) manuals are shared
1057: * across the "armish" and "zaurus" kernel
1058: * architectures.
1059: * A few manuals are even shared across completely
1060: * different architectures, for example fdformat(1)
1061: * on amd64, i386, sparc, and sparc64.
1062: */
1063:
1.131 schwarze 1064: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1065: say(mlink->file, "Architecture \"%s\" manual in "
1066: "\"%s\" directory", mpage->arch, mlink->arch);
1067:
1.131 schwarze 1068: /*
1069: * XXX
1.133 schwarze 1070: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1071: */
1072:
1073: if (FORM_CAT == mpage->form)
1074: return;
1075:
1076: /*
1077: * Check whether this mlink
1078: * appears as a name in the NAME section.
1079: */
1.93 schwarze 1080:
1.133 schwarze 1081: slot = ohash_qlookup(&names, mlink->name);
1082: str = ohash_find(&names, slot);
1.131 schwarze 1083: assert(NULL != str);
1.133 schwarze 1084: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1085: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1086: }
1087:
1.89 schwarze 1088: /*
1.80 schwarze 1089: * Run through the files in the global vector "mpages"
1090: * and add them to the database specified in "basedir".
1.50 kristaps 1091: *
1092: * This handles the parsing scheme itself, using the cues of directory
1093: * and filename to determine whether the file is parsable or not.
1094: */
1.59 schwarze 1095: static void
1.176 schwarze 1096: mpages_merge(struct mparse *mp)
1.50 kristaps 1097: {
1.113 schwarze 1098: char any[] = "any";
1.101 schwarze 1099: struct ohash_info str_info;
1.118 schwarze 1100: struct mpage *mpage, *mpage_dest;
1101: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1102: struct roff_man *man;
1.118 schwarze 1103: char *sodest;
1.112 schwarze 1104: char *cp;
1.162 schwarze 1105: int fd;
1.101 schwarze 1106: unsigned int pslot;
1.69 schwarze 1107:
1.71 schwarze 1108: str_info.alloc = hash_alloc;
1.152 schwarze 1109: str_info.calloc = hash_calloc;
1110: str_info.free = hash_free;
1.181 schwarze 1111: str_info.data = NULL;
1.71 schwarze 1112: str_info.key_offset = offsetof(struct str, key);
1113:
1.171 schwarze 1114: if ( ! nodb)
1.107 schwarze 1115: SQL_EXEC("BEGIN TRANSACTION");
1116:
1.80 schwarze 1117: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1118: while (mpage != NULL) {
1.89 schwarze 1119: mlinks_undupe(mpage);
1.182 schwarze 1120: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1121: mpage = ohash_next(&mpages, &pslot);
1122: continue;
1.50 kristaps 1123: }
1124:
1.133 schwarze 1125: name_mask = NAME_MASK;
1126: ohash_init(&names, 4, &str_info);
1.71 schwarze 1127: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1128: mparse_reset(mp);
1129: man = NULL;
1.124 schwarze 1130: sodest = NULL;
1131:
1.182 schwarze 1132: mparse_open(mp, &fd, mlink->file);
1.162 schwarze 1133: if (fd == -1) {
1.182 schwarze 1134: say(mlink->file, "&open");
1.162 schwarze 1135: goto nextpage;
1.124 schwarze 1136: }
1.14 schwarze 1137:
1138: /*
1.183 schwarze 1139: * Interpret the file as mdoc(7) or man(7) source
1140: * code, unless it is known to be formatted.
1.14 schwarze 1141: */
1.182 schwarze 1142: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1143: mparse_readfd(mp, fd, mlink->file);
1.192 schwarze 1144: mparse_result(mp, &man, &sodest);
1.76 schwarze 1145: }
1.14 schwarze 1146:
1.171 schwarze 1147: if (sodest != NULL) {
1.118 schwarze 1148: mlink_dest = ohash_find(&mlinks,
1149: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1150: if (mlink_dest == NULL) {
1151: mandoc_asprintf(&cp, "%s.gz", sodest);
1152: mlink_dest = ohash_find(&mlinks,
1153: ohash_qlookup(&mlinks, cp));
1154: free(cp);
1155: }
1156: if (mlink_dest != NULL) {
1.118 schwarze 1157:
1158: /* The .so target exists. */
1159:
1160: mpage_dest = mlink_dest->mpage;
1161: while (1) {
1162: mlink->mpage = mpage_dest;
1163:
1164: /*
1165: * If the target was already
1166: * processed, add the links
1167: * to the database now.
1168: * Otherwise, this will
1169: * happen when we come
1170: * to the target.
1171: */
1172:
1.137 schwarze 1173: if (mpage_dest->pageid)
1.169 schwarze 1174: dbadd_mlink_name(mlink);
1.118 schwarze 1175:
1.171 schwarze 1176: if (mlink->next == NULL)
1.118 schwarze 1177: break;
1178: mlink = mlink->next;
1179: }
1180:
1181: /* Move all links to the target. */
1182:
1183: mlink->next = mlink_dest->next;
1184: mlink_dest->next = mpage->mlinks;
1185: mpage->mlinks = NULL;
1186: }
1.124 schwarze 1187: goto nextpage;
1.192 schwarze 1188: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.82 schwarze 1189: mpage->form = FORM_SRC;
1.193 schwarze 1190: mpage->sec = man->meta.msec;
1.155 schwarze 1191: mpage->sec = mandoc_strdup(
1.171 schwarze 1192: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1193: mpage->arch = man->meta.arch;
1.84 schwarze 1194: mpage->arch = mandoc_strdup(
1.171 schwarze 1195: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1196: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1197: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.82 schwarze 1198: mpage->form = FORM_SRC;
1.193 schwarze 1199: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1200: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1201: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1202: } else {
1.82 schwarze 1203: mpage->form = FORM_CAT;
1.182 schwarze 1204: mpage->sec = mandoc_strdup(mlink->dsec);
1205: mpage->arch = mandoc_strdup(mlink->arch);
1206: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1207: }
1.97 schwarze 1208: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1209: if (*mpage->arch != '\0')
1210: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1211:
1.182 schwarze 1212: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1213: if ('\0' != *mlink->dsec)
1214: putkey(mpage, mlink->dsec, TYPE_sec);
1215: if ('\0' != *mlink->fsec)
1216: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1217: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1218: any : mlink->arch, TYPE_arch);
1.133 schwarze 1219: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1220: }
1.12 schwarze 1221:
1.174 schwarze 1222: assert(mpage->desc == NULL);
1.192 schwarze 1223: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1224: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1225: else if (man != NULL)
1.193 schwarze 1226: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1227: else
1.162 schwarze 1228: parse_cat(mpage, fd);
1.174 schwarze 1229: if (mpage->desc == NULL)
1.135 schwarze 1230: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1231:
1232: if (warnings && !use_all)
1233: for (mlink = mpage->mlinks; mlink;
1234: mlink = mlink->next)
1235: mlink_check(mpage, mlink);
1.44 kristaps 1236:
1.176 schwarze 1237: dbadd(mpage);
1.182 schwarze 1238: mlink = mpage->mlinks;
1.124 schwarze 1239:
1240: nextpage:
1.66 schwarze 1241: ohash_delete(&strings);
1.133 schwarze 1242: ohash_delete(&names);
1.80 schwarze 1243: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1244: }
1.107 schwarze 1245:
1246: if (0 == nodb)
1247: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1248: }
1249:
1250: static void
1251: names_check(void)
1252: {
1253: sqlite3_stmt *stmt;
1254: const char *name, *sec, *arch, *key;
1255: int irc;
1256:
1257: sqlite3_prepare_v2(db,
1258: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1259: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1260: "WHERE bits & ? AND NOT EXISTS ("
1261: "SELECT pageid FROM mlinks "
1.133 schwarze 1262: "WHERE mlinks.pageid == names.pageid "
1263: "AND mlinks.name == names.name"
1.130 schwarze 1264: ")"
1265: ") JOIN ("
1.142 schwarze 1266: "SELECT sec, arch, name, pageid FROM mlinks "
1267: "GROUP BY pageid"
1.130 schwarze 1268: ") USING (pageid);",
1269: -1, &stmt, NULL);
1270:
1.134 schwarze 1271: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1272: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1273:
1274: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.154 schwarze 1275: name = (const char *)sqlite3_column_text(stmt, 0);
1276: sec = (const char *)sqlite3_column_text(stmt, 1);
1277: arch = (const char *)sqlite3_column_text(stmt, 2);
1278: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1279: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1280: '\0' == *arch ? "" : "/",
1281: '\0' == *arch ? "" : arch, key);
1282: }
1283: sqlite3_finalize(stmt);
1.50 kristaps 1284: }
1.12 schwarze 1285:
1.50 kristaps 1286: static void
1.124 schwarze 1287: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1288: {
1289: FILE *stream;
1290: char *line, *p, *title;
1291: size_t len, plen, titlesz;
1.12 schwarze 1292:
1.124 schwarze 1293: stream = (-1 == fd) ?
1294: fopen(mpage->mlinks->file, "r") :
1295: fdopen(fd, "r");
1296: if (NULL == stream) {
1.159 schwarze 1297: if (-1 != fd)
1298: close(fd);
1.56 schwarze 1299: if (warnings)
1.123 schwarze 1300: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1301: return;
1302: }
1.1 kristaps 1303:
1.50 kristaps 1304: /* Skip to first blank line. */
1.1 kristaps 1305:
1.50 kristaps 1306: while (NULL != (line = fgetln(stream, &len)))
1307: if ('\n' == *line)
1308: break;
1.1 kristaps 1309:
1.50 kristaps 1310: /*
1311: * Assume the first line that is not indented
1312: * is the first section header. Skip to it.
1313: */
1.1 kristaps 1314:
1.50 kristaps 1315: while (NULL != (line = fgetln(stream, &len)))
1316: if ('\n' != *line && ' ' != *line)
1317: break;
1.141 schwarze 1318:
1.50 kristaps 1319: /*
1320: * Read up until the next section into a buffer.
1321: * Strip the leading and trailing newline from each read line,
1322: * appending a trailing space.
1323: * Ignore empty (whitespace-only) lines.
1324: */
1.1 kristaps 1325:
1.50 kristaps 1326: titlesz = 0;
1327: title = NULL;
1.38 schwarze 1328:
1.50 kristaps 1329: while (NULL != (line = fgetln(stream, &len))) {
1330: if (' ' != *line || '\n' != line[len - 1])
1331: break;
1332: while (len > 0 && isspace((unsigned char)*line)) {
1333: line++;
1334: len--;
1335: }
1336: if (1 == len)
1.38 schwarze 1337: continue;
1.50 kristaps 1338: title = mandoc_realloc(title, titlesz + len);
1339: memcpy(title + titlesz, line, len);
1340: titlesz += len;
1341: title[titlesz - 1] = ' ';
1342: }
1.38 schwarze 1343:
1.50 kristaps 1344: /*
1345: * If no page content can be found, or the input line
1346: * is already the next section header, or there is no
1347: * trailing newline, reuse the page title as the page
1348: * description.
1349: */
1.44 kristaps 1350:
1.50 kristaps 1351: if (NULL == title || '\0' == *title) {
1.56 schwarze 1352: if (warnings)
1.82 schwarze 1353: say(mpage->mlinks->file,
1354: "Cannot find NAME section");
1.50 kristaps 1355: fclose(stream);
1356: free(title);
1357: return;
1358: }
1.1 kristaps 1359:
1.50 kristaps 1360: title = mandoc_realloc(title, titlesz + 1);
1361: title[titlesz] = '\0';
1.33 schwarze 1362:
1.50 kristaps 1363: /*
1364: * Skip to the first dash.
1365: * Use the remaining line as the description (no more than 70
1366: * bytes).
1367: */
1.33 schwarze 1368:
1.50 kristaps 1369: if (NULL != (p = strstr(title, "- "))) {
1370: for (p += 2; ' ' == *p || '\b' == *p; p++)
1371: /* Skip to next word. */ ;
1372: } else {
1.56 schwarze 1373: if (warnings)
1.82 schwarze 1374: say(mpage->mlinks->file,
1375: "No dash in title line");
1.50 kristaps 1376: p = title;
1377: }
1.38 schwarze 1378:
1.50 kristaps 1379: plen = strlen(p);
1.1 kristaps 1380:
1.50 kristaps 1381: /* Strip backspace-encoding from line. */
1.1 kristaps 1382:
1.50 kristaps 1383: while (NULL != (line = memchr(p, '\b', plen))) {
1384: len = line - p;
1385: if (0 == len) {
1386: memmove(line, line + 1, plen--);
1387: continue;
1.141 schwarze 1388: }
1.50 kristaps 1389: memmove(line - 1, line + 1, plen - len);
1390: plen -= 2;
1391: }
1.1 kristaps 1392:
1.78 schwarze 1393: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1394: fclose(stream);
1395: free(title);
1396: }
1.1 kristaps 1397:
1.50 kristaps 1398: /*
1399: * Put a type/word pair into the word database for this particular file.
1400: */
1401: static void
1.112 schwarze 1402: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1403: {
1.112 schwarze 1404: char *cp;
1.18 kristaps 1405:
1.50 kristaps 1406: assert(NULL != value);
1.112 schwarze 1407: if (TYPE_arch == type)
1408: for (cp = value; *cp; cp++)
1409: if (isupper((unsigned char)*cp))
1410: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1411: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1412: }
1413:
1414: /*
1.50 kristaps 1415: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1416: */
1417: static void
1.78 schwarze 1418: putmdockey(const struct mpage *mpage,
1.189 schwarze 1419: const struct roff_node *n, uint64_t m)
1.3 kristaps 1420: {
1.18 kristaps 1421:
1.50 kristaps 1422: for ( ; NULL != n; n = n->next) {
1423: if (NULL != n->child)
1.78 schwarze 1424: putmdockey(mpage, n->child, m);
1.188 schwarze 1425: if (n->type == ROFFT_TEXT)
1.78 schwarze 1426: putkey(mpage, n->string, m);
1.50 kristaps 1427: }
1428: }
1.18 kristaps 1429:
1.61 schwarze 1430: static void
1.190 schwarze 1431: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1432: const struct roff_node *n)
1.50 kristaps 1433: {
1.189 schwarze 1434: const struct roff_node *head, *body;
1.121 schwarze 1435: char *start, *title;
1.50 kristaps 1436: char byte;
1.121 schwarze 1437: size_t sz;
1.18 kristaps 1438:
1.50 kristaps 1439: if (NULL == n)
1.61 schwarze 1440: return;
1.18 kristaps 1441:
1.50 kristaps 1442: /*
1443: * We're only searching for one thing: the first text child in
1444: * the BODY of a NAME section. Since we don't keep track of
1445: * sections in -man, run some hoops to find out whether we're in
1446: * the correct section or not.
1447: */
1.18 kristaps 1448:
1.188 schwarze 1449: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1450: body = n;
1451: assert(body->parent);
1452: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1453: 1 == head->nchild &&
1454: NULL != (head = (head->child)) &&
1.188 schwarze 1455: head->type == ROFFT_TEXT &&
1.141 schwarze 1456: 0 == strcmp(head->string, "NAME") &&
1457: NULL != body->child) {
1.3 kristaps 1458:
1.50 kristaps 1459: /*
1460: * Suck the entire NAME section into memory.
1461: * Yes, we might run away.
1462: * But too many manuals have big, spread-out
1463: * NAME sections over many lines.
1464: */
1.3 kristaps 1465:
1.121 schwarze 1466: title = NULL;
1.194 schwarze 1467: deroff(&title, body);
1.50 kristaps 1468: if (NULL == title)
1.61 schwarze 1469: return;
1.18 kristaps 1470:
1.141 schwarze 1471: /*
1.50 kristaps 1472: * Go through a special heuristic dance here.
1473: * Conventionally, one or more manual names are
1474: * comma-specified prior to a whitespace, then a
1475: * dash, then a description. Try to puzzle out
1476: * the name parts here.
1477: */
1.18 kristaps 1478:
1.121 schwarze 1479: start = title;
1.50 kristaps 1480: for ( ;; ) {
1481: sz = strcspn(start, " ,");
1482: if ('\0' == start[sz])
1483: break;
1.1 kristaps 1484:
1.50 kristaps 1485: byte = start[sz];
1486: start[sz] = '\0';
1.110 schwarze 1487:
1488: /*
1489: * Assume a stray trailing comma in the
1490: * name list if a name begins with a dash.
1491: */
1492:
1493: if ('-' == start[0] ||
1494: ('\\' == start[0] && '-' == start[1]))
1495: break;
1.1 kristaps 1496:
1.133 schwarze 1497: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1498: if ( ! (mpage->name_head_done ||
1499: strcasecmp(start, meta->title))) {
1500: putkey(mpage, start, NAME_HEAD);
1501: mpage->name_head_done = 1;
1502: }
1.1 kristaps 1503:
1.50 kristaps 1504: if (' ' == byte) {
1505: start += sz + 1;
1506: break;
1507: }
1.1 kristaps 1508:
1.50 kristaps 1509: assert(',' == byte);
1510: start += sz + 1;
1511: while (' ' == *start)
1512: start++;
1513: }
1.1 kristaps 1514:
1.121 schwarze 1515: if (start == title) {
1.133 schwarze 1516: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1517: if ( ! (mpage->name_head_done ||
1518: strcasecmp(start, meta->title))) {
1519: putkey(mpage, start, NAME_HEAD);
1520: mpage->name_head_done = 1;
1521: }
1.50 kristaps 1522: free(title);
1.61 schwarze 1523: return;
1.50 kristaps 1524: }
1.1 kristaps 1525:
1.50 kristaps 1526: while (isspace((unsigned char)*start))
1527: start++;
1.1 kristaps 1528:
1.50 kristaps 1529: if (0 == strncmp(start, "-", 1))
1530: start += 1;
1531: else if (0 == strncmp(start, "\\-\\-", 4))
1532: start += 4;
1533: else if (0 == strncmp(start, "\\-", 2))
1534: start += 2;
1535: else if (0 == strncmp(start, "\\(en", 4))
1536: start += 4;
1537: else if (0 == strncmp(start, "\\(em", 4))
1538: start += 4;
1.1 kristaps 1539:
1.50 kristaps 1540: while (' ' == *start)
1541: start++;
1.1 kristaps 1542:
1.78 schwarze 1543: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1544: free(title);
1.61 schwarze 1545: return;
1.50 kristaps 1546: }
1547: }
1.1 kristaps 1548:
1.77 schwarze 1549: for (n = n->child; n; n = n->next) {
1.78 schwarze 1550: if (NULL != mpage->desc)
1.77 schwarze 1551: break;
1.174 schwarze 1552: parse_man(mpage, meta, n);
1.77 schwarze 1553: }
1.1 kristaps 1554: }
1555:
1556: static void
1.190 schwarze 1557: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1558: const struct roff_node *n)
1.1 kristaps 1559: {
1560:
1.50 kristaps 1561: assert(NULL != n);
1562: for (n = n->child; NULL != n; n = n->next) {
1563: switch (n->type) {
1.188 schwarze 1564: case ROFFT_ELEM:
1565: case ROFFT_BLOCK:
1566: case ROFFT_HEAD:
1567: case ROFFT_BODY:
1568: case ROFFT_TAIL:
1.50 kristaps 1569: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1570: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1571: break;
1.68 schwarze 1572: if (mdocs[n->tok].mask)
1.78 schwarze 1573: putmdockey(mpage, n->child,
1574: mdocs[n->tok].mask);
1.50 kristaps 1575: break;
1576: default:
1.188 schwarze 1577: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1578: continue;
1579: }
1580: if (NULL != n->child)
1.172 schwarze 1581: parse_mdoc(mpage, meta, n);
1.1 kristaps 1582: }
1583: }
1584:
1.25 schwarze 1585: static int
1.190 schwarze 1586: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1587: const struct roff_node *n)
1.1 kristaps 1588: {
1.176 schwarze 1589: char *start, *end;
1.1 kristaps 1590: size_t sz;
1.25 schwarze 1591:
1.50 kristaps 1592: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1593: NULL == (n = n->child) ||
1.188 schwarze 1594: n->type != ROFFT_TEXT)
1.197 schwarze 1595: return 0;
1.1 kristaps 1596:
1597: /*
1598: * Only consider those `Fd' macro fields that begin with an
1599: * "inclusion" token (versus, e.g., #define).
1600: */
1.50 kristaps 1601:
1.1 kristaps 1602: if (strcmp("#include", n->string))
1.197 schwarze 1603: return 0;
1.1 kristaps 1604:
1.188 schwarze 1605: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1606: return 0;
1.1 kristaps 1607:
1608: /*
1609: * Strip away the enclosing angle brackets and make sure we're
1610: * not zero-length.
1611: */
1612:
1613: start = n->string;
1614: if ('<' == *start || '"' == *start)
1615: start++;
1616:
1617: if (0 == (sz = strlen(start)))
1.197 schwarze 1618: return 0;
1.1 kristaps 1619:
1620: end = &start[(int)sz - 1];
1621: if ('>' == *end || '"' == *end)
1622: end--;
1623:
1.50 kristaps 1624: if (end > start)
1.78 schwarze 1625: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1626: return 0;
1.1 kristaps 1627: }
1628:
1.178 schwarze 1629: static void
1.189 schwarze 1630: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1631: {
1.112 schwarze 1632: char *cp;
1.178 schwarze 1633: size_t sz;
1.1 kristaps 1634:
1.188 schwarze 1635: if (n->type != ROFFT_TEXT)
1.178 schwarze 1636: return;
1.25 schwarze 1637:
1.178 schwarze 1638: /* Skip function pointer punctuation. */
1.1 kristaps 1639:
1.178 schwarze 1640: cp = n->string;
1641: while (*cp == '(' || *cp == '*')
1.1 kristaps 1642: cp++;
1.178 schwarze 1643: sz = strcspn(cp, "()");
1.1 kristaps 1644:
1.178 schwarze 1645: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1646: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1647: putkeys(mpage, cp, sz, NAME_SYN);
1648: }
1.25 schwarze 1649:
1.178 schwarze 1650: static int
1.190 schwarze 1651: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1652: const struct roff_node *n)
1.178 schwarze 1653: {
1654:
1655: if (n->child == NULL)
1.197 schwarze 1656: return 0;
1.178 schwarze 1657:
1658: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1659:
1.178 schwarze 1660: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1661: if (n->type == ROFFT_TEXT)
1.78 schwarze 1662: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1663:
1.197 schwarze 1664: return 0;
1.173 schwarze 1665: }
1666:
1667: static int
1.190 schwarze 1668: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1669: const struct roff_node *n)
1.173 schwarze 1670: {
1.177 schwarze 1671:
1.188 schwarze 1672: if (n->type != ROFFT_HEAD)
1.197 schwarze 1673: return 1;
1.173 schwarze 1674:
1.178 schwarze 1675: if (n->child != NULL)
1676: parse_mdoc_fname(mpage, n->child);
1677:
1.197 schwarze 1678: return 0;
1.1 kristaps 1679: }
1680:
1.25 schwarze 1681: static int
1.190 schwarze 1682: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1683: const struct roff_node *n)
1.1 kristaps 1684: {
1.67 schwarze 1685: char *cp;
1.1 kristaps 1686:
1687: if (NULL == (n = n->child))
1.197 schwarze 1688: return 0;
1.1 kristaps 1689:
1.67 schwarze 1690: if (NULL == n->next) {
1.78 schwarze 1691: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1692: return 0;
1.67 schwarze 1693: }
1694:
1.120 schwarze 1695: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1696: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1697: free(cp);
1.197 schwarze 1698: return 0;
1.1 kristaps 1699: }
1700:
1.25 schwarze 1701: static int
1.190 schwarze 1702: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1703: const struct roff_node *n)
1.1 kristaps 1704: {
1705:
1.188 schwarze 1706: if (n->type == ROFFT_BODY)
1.194 schwarze 1707: deroff(&mpage->desc, n);
1.197 schwarze 1708: return 0;
1.1 kristaps 1709: }
1710:
1.25 schwarze 1711: static int
1.190 schwarze 1712: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1713: const struct roff_node *n)
1.1 kristaps 1714: {
1715:
1.129 schwarze 1716: if (SEC_NAME == n->sec)
1.133 schwarze 1717: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1718: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1719: if (n->child == NULL)
1720: putkey(mpage, meta->name, NAME_SYN);
1721: else
1722: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1723: }
1724: if ( ! (mpage->name_head_done ||
1725: n->child == NULL || n->child->string == NULL ||
1726: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1727: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1728: mpage->name_head_done = 1;
1.172 schwarze 1729: }
1.197 schwarze 1730: return 0;
1.1 kristaps 1731: }
1732:
1.25 schwarze 1733: static int
1.190 schwarze 1734: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1735: const struct roff_node *n)
1.1 kristaps 1736: {
1737:
1.197 schwarze 1738: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1739: }
1740:
1.50 kristaps 1741: static int
1.190 schwarze 1742: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1743: const struct roff_node *n)
1.1 kristaps 1744: {
1745:
1.197 schwarze 1746: return n->type == ROFFT_HEAD;
1.50 kristaps 1747: }
1.1 kristaps 1748:
1.50 kristaps 1749: static int
1.190 schwarze 1750: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1751: const struct roff_node *n)
1.50 kristaps 1752: {
1.1 kristaps 1753:
1.197 schwarze 1754: return n->type == ROFFT_BODY;
1.1 kristaps 1755: }
1756:
1.50 kristaps 1757: /*
1.66 schwarze 1758: * Add a string to the hash table for the current manual.
1759: * Each string has a bitmask telling which macros it belongs to.
1760: * When we finish the manual, we'll dump the table.
1.50 kristaps 1761: */
1762: static void
1.176 schwarze 1763: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1764: {
1.133 schwarze 1765: struct ohash *htab;
1.50 kristaps 1766: struct str *s;
1.111 schwarze 1767: const char *end;
1.65 schwarze 1768: unsigned int slot;
1.176 schwarze 1769: int i, mustfree;
1.25 schwarze 1770:
1.50 kristaps 1771: if (0 == sz)
1772: return;
1.111 schwarze 1773:
1.176 schwarze 1774: mustfree = render_string(&cp, &sz);
1775:
1.133 schwarze 1776: if (TYPE_Nm & v) {
1777: htab = &names;
1778: v &= name_mask;
1.169 schwarze 1779: if (v & NAME_FIRST)
1780: name_mask &= ~NAME_FIRST;
1.133 schwarze 1781: if (debug > 1)
1782: say(mpage->mlinks->file,
1.175 schwarze 1783: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1784: } else {
1785: htab = &strings;
1786: if (debug > 1)
1787: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1788: if ((uint64_t)1 << i & v)
1.133 schwarze 1789: say(mpage->mlinks->file,
1790: "Adding key %s=%*s",
1791: mansearch_keynames[i], sz, cp);
1.111 schwarze 1792: }
1.25 schwarze 1793:
1.65 schwarze 1794: end = cp + sz;
1.133 schwarze 1795: slot = ohash_qlookupi(htab, cp, &end);
1796: s = ohash_find(htab, slot);
1.25 schwarze 1797:
1.78 schwarze 1798: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1799: s->mask |= v;
1800: return;
1801: } else if (NULL == s) {
1.144 schwarze 1802: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1803: memcpy(s->key, cp, sz);
1.133 schwarze 1804: ohash_insert(htab, slot, s);
1.1 kristaps 1805: }
1.78 schwarze 1806: s->mpage = mpage;
1.50 kristaps 1807: s->mask = v;
1.176 schwarze 1808:
1809: if (mustfree)
1810: free(cp);
1.1 kristaps 1811: }
1812:
1.50 kristaps 1813: /*
1814: * Take a Unicode codepoint and produce its UTF-8 encoding.
1815: * This isn't the best way to do this, but it works.
1816: * The magic numbers are from the UTF-8 packaging.
1817: * They're not as scary as they seem: read the UTF-8 spec for details.
1818: */
1819: static size_t
1820: utf8(unsigned int cp, char out[7])
1.1 kristaps 1821: {
1.50 kristaps 1822: size_t rc;
1.1 kristaps 1823:
1.50 kristaps 1824: rc = 0;
1825: if (cp <= 0x0000007F) {
1826: rc = 1;
1827: out[0] = (char)cp;
1828: } else if (cp <= 0x000007FF) {
1829: rc = 2;
1830: out[0] = (cp >> 6 & 31) | 192;
1831: out[1] = (cp & 63) | 128;
1832: } else if (cp <= 0x0000FFFF) {
1833: rc = 3;
1834: out[0] = (cp >> 12 & 15) | 224;
1835: out[1] = (cp >> 6 & 63) | 128;
1836: out[2] = (cp & 63) | 128;
1837: } else if (cp <= 0x001FFFFF) {
1838: rc = 4;
1839: out[0] = (cp >> 18 & 7) | 240;
1840: out[1] = (cp >> 12 & 63) | 128;
1841: out[2] = (cp >> 6 & 63) | 128;
1842: out[3] = (cp & 63) | 128;
1843: } else if (cp <= 0x03FFFFFF) {
1844: rc = 5;
1845: out[0] = (cp >> 24 & 3) | 248;
1846: out[1] = (cp >> 18 & 63) | 128;
1847: out[2] = (cp >> 12 & 63) | 128;
1848: out[3] = (cp >> 6 & 63) | 128;
1849: out[4] = (cp & 63) | 128;
1850: } else if (cp <= 0x7FFFFFFF) {
1851: rc = 6;
1852: out[0] = (cp >> 30 & 1) | 252;
1853: out[1] = (cp >> 24 & 63) | 128;
1854: out[2] = (cp >> 18 & 63) | 128;
1855: out[3] = (cp >> 12 & 63) | 128;
1856: out[4] = (cp >> 6 & 63) | 128;
1857: out[5] = (cp & 63) | 128;
1858: } else
1.197 schwarze 1859: return 0;
1.1 kristaps 1860:
1.50 kristaps 1861: out[rc] = '\0';
1.197 schwarze 1862: return rc;
1.50 kristaps 1863: }
1.1 kristaps 1864:
1.50 kristaps 1865: /*
1.176 schwarze 1866: * If the string contains escape sequences,
1867: * replace it with an allocated rendering and return 1,
1868: * such that the caller can free it after use.
1869: * Otherwise, do nothing and return 0.
1.50 kristaps 1870: */
1.176 schwarze 1871: static int
1872: render_string(char **public, size_t *psz)
1.50 kristaps 1873: {
1.176 schwarze 1874: const char *src, *scp, *addcp, *seq;
1875: char *dst;
1876: size_t ssz, dsz, addsz;
1.114 schwarze 1877: char utfbuf[7], res[6];
1.176 schwarze 1878: int seqlen, unicode;
1.50 kristaps 1879:
1880: res[0] = '\\';
1881: res[1] = '\t';
1882: res[2] = ASCII_NBRSP;
1883: res[3] = ASCII_HYPH;
1.114 schwarze 1884: res[4] = ASCII_BREAK;
1885: res[5] = '\0';
1.1 kristaps 1886:
1.176 schwarze 1887: src = scp = *public;
1888: ssz = *psz;
1889: dst = NULL;
1890: dsz = 0;
1891:
1892: while (scp < src + *psz) {
1893:
1894: /* Leave normal characters unchanged. */
1895:
1896: if (strchr(res, *scp) == NULL) {
1897: if (dst != NULL)
1898: dst[dsz++] = *scp;
1899: scp++;
1900: continue;
1901: }
1.46 kristaps 1902:
1.50 kristaps 1903: /*
1.176 schwarze 1904: * Found something that requires replacing,
1905: * make sure we have a destination buffer.
1.50 kristaps 1906: */
1.176 schwarze 1907:
1908: if (dst == NULL) {
1909: dst = mandoc_malloc(ssz + 1);
1910: dsz = scp - src;
1911: memcpy(dst, src, dsz);
1.50 kristaps 1912: }
1.46 kristaps 1913:
1.176 schwarze 1914: /* Handle single-char special characters. */
1915:
1916: switch (*scp) {
1917: case '\\':
1918: break;
1.141 schwarze 1919: case '\t':
1920: case ASCII_NBRSP:
1.176 schwarze 1921: dst[dsz++] = ' ';
1922: scp++;
1923: continue;
1924: case ASCII_HYPH:
1925: dst[dsz++] = '-';
1.114 schwarze 1926: /* FALLTHROUGH */
1.141 schwarze 1927: case ASCII_BREAK:
1.176 schwarze 1928: scp++;
1.50 kristaps 1929: continue;
1.114 schwarze 1930: default:
1.176 schwarze 1931: abort();
1.114 schwarze 1932: }
1.46 kristaps 1933:
1.50 kristaps 1934: /*
1.176 schwarze 1935: * Found an escape sequence.
1936: * Read past the slash, then parse it.
1937: * Ignore everything except characters.
1.50 kristaps 1938: */
1.95 schwarze 1939:
1.176 schwarze 1940: scp++;
1941: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1942: continue;
1.1 kristaps 1943:
1.50 kristaps 1944: /*
1.95 schwarze 1945: * Render the special character
1946: * as either UTF-8 or ASCII.
1.50 kristaps 1947: */
1.95 schwarze 1948:
1949: if (write_utf8) {
1.176 schwarze 1950: unicode = mchars_spec2cp(mchars, seq, seqlen);
1951: if (unicode <= 0)
1.95 schwarze 1952: continue;
1.176 schwarze 1953: addsz = utf8(unicode, utfbuf);
1954: if (addsz == 0)
1.95 schwarze 1955: continue;
1.176 schwarze 1956: addcp = utfbuf;
1.95 schwarze 1957: } else {
1.176 schwarze 1958: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1959: if (addcp == NULL)
1.95 schwarze 1960: continue;
1.176 schwarze 1961: if (*addcp == ASCII_NBRSP) {
1962: addcp = " ";
1963: addsz = 1;
1.95 schwarze 1964: }
1965: }
1.1 kristaps 1966:
1.50 kristaps 1967: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1968:
1.176 schwarze 1969: ssz += addsz;
1970: dst = mandoc_realloc(dst, ssz + 1);
1971: memcpy(dst + dsz, addcp, addsz);
1972: dsz += addsz;
1973: }
1974: if (dst != NULL) {
1975: *public = dst;
1976: *psz = dsz;
1977: }
1978:
1979: /* Trim trailing whitespace and NUL-terminate. */
1980:
1981: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1982: --*psz;
1983: if (dst != NULL) {
1984: (*public)[*psz] = '\0';
1.197 schwarze 1985: return 1;
1.176 schwarze 1986: } else
1.197 schwarze 1987: return 0;
1.1 kristaps 1988: }
1989:
1.118 schwarze 1990: static void
1991: dbadd_mlink(const struct mlink *mlink)
1992: {
1993: size_t i;
1994:
1995: i = 1;
1996: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1997: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1998: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 1999: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2000: SQL_STEP(stmts[STMT_INSERT_LINK]);
2001: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2002: }
2003:
2004: static void
2005: dbadd_mlink_name(const struct mlink *mlink)
2006: {
1.175 schwarze 2007: uint64_t bits;
1.169 schwarze 2008: size_t i;
2009:
2010: dbadd_mlink(mlink);
1.160 schwarze 2011:
2012: i = 1;
1.175 schwarze 2013: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2014: bits = NAME_FILE & NAME_MASK;
2015: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2016: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2017: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2018: }
2019:
2020: i = 1;
2021: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2022: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2023: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2024: SQL_STEP(stmts[STMT_INSERT_NAME]);
2025: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2026: }
2027:
1.14 schwarze 2028: /*
1.50 kristaps 2029: * Flush the current page's terms (and their bits) into the database.
2030: * Wrap the entire set of additions in a transaction to make sqlite be a
2031: * little faster.
1.96 schwarze 2032: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2033: */
2034: static void
1.176 schwarze 2035: dbadd(struct mpage *mpage)
1.14 schwarze 2036: {
1.87 schwarze 2037: struct mlink *mlink;
1.50 kristaps 2038: struct str *key;
1.176 schwarze 2039: char *cp;
1.52 kristaps 2040: size_t i;
1.66 schwarze 2041: unsigned int slot;
1.176 schwarze 2042: int mustfree;
1.14 schwarze 2043:
1.128 schwarze 2044: mlink = mpage->mlinks;
1.43 kristaps 2045:
1.128 schwarze 2046: if (nodb) {
1.147 schwarze 2047: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2048: key = ohash_next(&names, &slot))
1.147 schwarze 2049: free(key);
2050: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2051: key = ohash_next(&strings, &slot))
1.147 schwarze 2052: free(key);
1.145 schwarze 2053: if (0 == debug)
2054: return;
1.128 schwarze 2055: while (NULL != mlink) {
2056: fputs(mlink->name, stdout);
2057: if (NULL == mlink->next ||
2058: strcmp(mlink->dsec, mlink->next->dsec) ||
2059: strcmp(mlink->fsec, mlink->next->fsec) ||
2060: strcmp(mlink->arch, mlink->next->arch)) {
2061: putchar('(');
2062: if ('\0' == *mlink->dsec)
2063: fputs(mlink->fsec, stdout);
2064: else
2065: fputs(mlink->dsec, stdout);
2066: if ('\0' != *mlink->arch)
2067: printf("/%s", mlink->arch);
2068: putchar(')');
2069: }
2070: mlink = mlink->next;
2071: if (NULL != mlink)
2072: fputs(", ", stdout);
2073: }
1.132 schwarze 2074: printf(" - %s\n", mpage->desc);
1.14 schwarze 2075: return;
1.128 schwarze 2076: }
2077:
2078: if (debug)
2079: say(mlink->file, "Adding to database");
1.28 kristaps 2080:
1.176 schwarze 2081: cp = mpage->desc;
2082: i = strlen(cp);
2083: mustfree = render_string(&cp, &i);
1.52 kristaps 2084: i = 1;
1.176 schwarze 2085: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2086: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2087: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2088: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2089: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2090: if (mustfree)
2091: free(cp);
1.81 schwarze 2092:
1.128 schwarze 2093: while (NULL != mlink) {
1.118 schwarze 2094: dbadd_mlink(mlink);
1.128 schwarze 2095: mlink = mlink->next;
2096: }
1.134 schwarze 2097: mlink = mpage->mlinks;
1.50 kristaps 2098:
1.133 schwarze 2099: for (key = ohash_first(&names, &slot); NULL != key;
2100: key = ohash_next(&names, &slot)) {
2101: assert(key->mpage == mpage);
2102: i = 1;
2103: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2104: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2105: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2106: SQL_STEP(stmts[STMT_INSERT_NAME]);
2107: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2108: free(key);
2109: }
1.66 schwarze 2110: for (key = ohash_first(&strings, &slot); NULL != key;
2111: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2112: assert(key->mpage == mpage);
1.52 kristaps 2113: i = 1;
2114: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2115: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2116: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2117: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2118: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2119: free(key);
1.38 schwarze 2120: }
1.14 schwarze 2121: }
2122:
1.5 kristaps 2123: static void
1.59 schwarze 2124: dbprune(void)
1.5 kristaps 2125: {
1.78 schwarze 2126: struct mpage *mpage;
1.82 schwarze 2127: struct mlink *mlink;
1.52 kristaps 2128: size_t i;
1.80 schwarze 2129: unsigned int slot;
1.5 kristaps 2130:
1.106 schwarze 2131: if (0 == nodb)
2132: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2133:
1.106 schwarze 2134: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2135: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2136: mlink = mpage->mlinks;
1.125 schwarze 2137: if (debug)
1.106 schwarze 2138: say(mlink->file, "Deleting from database");
2139: if (nodb)
2140: continue;
2141: for ( ; NULL != mlink; mlink = mlink->next) {
2142: i = 1;
2143: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2144: i, mlink->dsec);
2145: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2146: i, mlink->arch);
2147: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2148: i, mlink->name);
2149: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2150: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2151: }
1.5 kristaps 2152: }
1.106 schwarze 2153:
2154: if (0 == nodb)
2155: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2156: }
2157:
1.4 kristaps 2158: /*
1.50 kristaps 2159: * Close an existing database and its prepared statements.
2160: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2161: */
1.35 kristaps 2162: static void
1.59 schwarze 2163: dbclose(int real)
1.4 kristaps 2164: {
1.50 kristaps 2165: size_t i;
1.115 schwarze 2166: int status;
2167: pid_t child;
1.4 kristaps 2168:
1.50 kristaps 2169: if (nodb)
1.38 schwarze 2170: return;
1.50 kristaps 2171:
2172: for (i = 0; i < STMT__MAX; i++) {
2173: sqlite3_finalize(stmts[i]);
2174: stmts[i] = NULL;
1.4 kristaps 2175: }
2176:
1.50 kristaps 2177: sqlite3_close(db);
2178: db = NULL;
1.12 schwarze 2179:
1.50 kristaps 2180: if (real)
2181: return;
1.12 schwarze 2182:
1.115 schwarze 2183: if ('\0' == *tempfilename) {
2184: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2185: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2186: say(MANDOC_DB, "&rename");
1.115 schwarze 2187: }
2188: return;
2189: }
2190:
2191: switch (child = fork()) {
1.141 schwarze 2192: case -1:
1.115 schwarze 2193: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2194: say("", "&fork cmp");
1.115 schwarze 2195: return;
1.141 schwarze 2196: case 0:
1.115 schwarze 2197: execlp("cmp", "cmp", "-s",
1.196 schwarze 2198: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2199: say("", "&exec cmp");
1.115 schwarze 2200: exit(0);
2201: default:
2202: break;
2203: }
2204: if (-1 == waitpid(child, &status, 0)) {
2205: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2206: say("", "&wait cmp");
1.115 schwarze 2207: } else if (WIFSIGNALED(status)) {
2208: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2209: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2210: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2212: say(MANDOC_DB,
2213: "Data changed, but cannot replace database");
2214: }
2215:
2216: *strrchr(tempfilename, '/') = '\0';
2217: switch (child = fork()) {
1.141 schwarze 2218: case -1:
1.115 schwarze 2219: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2220: say("", "&fork rm");
1.115 schwarze 2221: return;
1.141 schwarze 2222: case 0:
1.196 schwarze 2223: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2224: say("", "&exec rm");
1.115 schwarze 2225: exit((int)MANDOCLEVEL_SYSERR);
2226: default:
2227: break;
2228: }
2229: if (-1 == waitpid(child, &status, 0)) {
2230: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2231: say("", "&wait rm");
1.115 schwarze 2232: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2233: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2234: say("", "%s: Cannot remove temporary directory",
2235: tempfilename);
1.59 schwarze 2236: }
1.50 kristaps 2237: }
1.14 schwarze 2238:
1.50 kristaps 2239: /*
2240: * This is straightforward stuff.
2241: * Open a database connection to a "temporary" database, then open a set
2242: * of prepared statements we'll use over and over again.
2243: * If "real" is set, we use the existing database; if not, we truncate a
2244: * temporary one.
2245: * Must be matched by dbclose().
2246: */
2247: static int
1.59 schwarze 2248: dbopen(int real)
1.50 kristaps 2249: {
1.115 schwarze 2250: const char *sql;
1.50 kristaps 2251: int rc, ofl;
1.12 schwarze 2252:
1.141 schwarze 2253: if (nodb)
1.197 schwarze 2254: return 1;
1.12 schwarze 2255:
1.115 schwarze 2256: *tempfilename = '\0';
1.63 schwarze 2257: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2258:
2259: if (real) {
2260: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2261: if (SQLITE_OK != rc) {
1.63 schwarze 2262: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2263: if (SQLITE_CANTOPEN != rc)
2264: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2265: return 0;
1.63 schwarze 2266: }
1.115 schwarze 2267: goto prepare_statements;
2268: }
2269:
2270: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2271:
1.115 schwarze 2272: remove(MANDOC_DB "~");
2273: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2274: if (SQLITE_OK == rc)
1.115 schwarze 2275: goto create_tables;
1.116 schwarze 2276: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2277: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2278: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2279: return 0;
1.50 kristaps 2280: }
1.12 schwarze 2281:
1.143 schwarze 2282: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2283: sizeof(tempfilename));
1.115 schwarze 2284: if (NULL == mkdtemp(tempfilename)) {
2285: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2286: say("", "&%s", tempfilename);
1.197 schwarze 2287: return 0;
1.115 schwarze 2288: }
1.143 schwarze 2289: (void)strlcat(tempfilename, "/" MANDOC_DB,
2290: sizeof(tempfilename));
1.115 schwarze 2291: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2292: if (SQLITE_OK != rc) {
1.59 schwarze 2293: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2294: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2295: return 0;
1.50 kristaps 2296: }
1.12 schwarze 2297:
1.115 schwarze 2298: create_tables:
1.81 schwarze 2299: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2300: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2301: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2302: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2303: ");\n"
2304: "\n"
1.81 schwarze 2305: "CREATE TABLE \"mlinks\" (\n"
2306: " \"sec\" TEXT NOT NULL,\n"
2307: " \"arch\" TEXT NOT NULL,\n"
2308: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2309: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2310: "ON DELETE CASCADE\n"
1.81 schwarze 2311: ");\n"
1.136 schwarze 2312: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2313: "\n"
1.133 schwarze 2314: "CREATE TABLE \"names\" (\n"
2315: " \"bits\" INTEGER NOT NULL,\n"
2316: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2317: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2318: "ON DELETE CASCADE,\n"
2319: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2320: ");\n"
2321: "\n"
1.50 kristaps 2322: "CREATE TABLE \"keys\" (\n"
2323: " \"bits\" INTEGER NOT NULL,\n"
2324: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2325: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2326: "ON DELETE CASCADE\n"
1.136 schwarze 2327: ");\n"
2328: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2329:
1.50 kristaps 2330: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2331: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2332: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2333: sqlite3_close(db);
1.197 schwarze 2334: return 0;
1.50 kristaps 2335: }
1.4 kristaps 2336:
1.57 schwarze 2337: prepare_statements:
1.146 schwarze 2338: if (SQLITE_OK != sqlite3_exec(db,
2339: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2340: exitcode = (int)MANDOCLEVEL_SYSERR;
2341: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2342: sqlite3_errmsg(db));
2343: sqlite3_close(db);
1.197 schwarze 2344: return 0;
1.146 schwarze 2345: }
2346:
1.137 schwarze 2347: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2348: "(SELECT pageid FROM mlinks WHERE "
2349: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2350: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2351: sql = "INSERT INTO mpages "
1.132 schwarze 2352: "(desc,form) VALUES (?,?)";
1.81 schwarze 2353: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2354: sql = "INSERT INTO mlinks "
1.104 schwarze 2355: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2356: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2357: sql = "SELECT bits FROM names where pageid = ?";
2358: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2359: sql = "INSERT INTO names "
2360: "(bits,name,pageid) VALUES (?,?,?)";
2361: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2362: sql = "INSERT INTO keys "
1.81 schwarze 2363: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2364: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2365:
2366: #ifndef __APPLE__
2367: /*
2368: * When opening a new database, we can turn off
2369: * synchronous mode for much better performance.
2370: */
2371:
1.146 schwarze 2372: if (real && SQLITE_OK != sqlite3_exec(db,
2373: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2374: exitcode = (int)MANDOCLEVEL_SYSERR;
2375: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2376: sqlite3_errmsg(db));
1.146 schwarze 2377: sqlite3_close(db);
1.197 schwarze 2378: return 0;
1.146 schwarze 2379: }
1.70 schwarze 2380: #endif
2381:
1.197 schwarze 2382: return 1;
1.50 kristaps 2383: }
1.5 kristaps 2384:
1.50 kristaps 2385: static void *
1.152 schwarze 2386: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2387: {
1.12 schwarze 2388:
1.197 schwarze 2389: return mandoc_calloc(n, sz);
1.50 kristaps 2390: }
1.12 schwarze 2391:
1.50 kristaps 2392: static void *
2393: hash_alloc(size_t sz, void *arg)
2394: {
1.5 kristaps 2395:
1.197 schwarze 2396: return mandoc_malloc(sz);
1.50 kristaps 2397: }
1.41 kristaps 2398:
1.50 kristaps 2399: static void
1.152 schwarze 2400: hash_free(void *p, void *arg)
1.50 kristaps 2401: {
1.4 kristaps 2402:
1.50 kristaps 2403: free(p);
1.4 kristaps 2404: }
2405:
1.50 kristaps 2406: static int
1.165 schwarze 2407: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2408: {
1.59 schwarze 2409: static char startdir[PATH_MAX];
1.151 schwarze 2410: static int getcwd_status; /* 1 = ok, 2 = failure */
2411: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2412: char *cp;
1.4 kristaps 2413:
1.59 schwarze 2414: /*
1.151 schwarze 2415: * Remember the original working directory, if possible.
2416: * This will be needed if the second or a later directory
2417: * on the command line is given as a relative path.
2418: * Do not error out if the current directory is not
2419: * searchable: Maybe it won't be needed after all.
2420: */
2421: if (0 == getcwd_status) {
2422: if (NULL == getcwd(startdir, sizeof(startdir))) {
2423: getcwd_status = 2;
2424: (void)strlcpy(startdir, strerror(errno),
2425: sizeof(startdir));
2426: } else
2427: getcwd_status = 1;
2428: }
2429:
2430: /*
2431: * We are leaving the old base directory.
2432: * Do not use it any longer, not even for messages.
2433: */
2434: *basedir = '\0';
2435:
2436: /*
2437: * If and only if the directory was changed earlier and
2438: * the next directory to process is given as a relative path,
2439: * first go back, or bail out if that is impossible.
1.59 schwarze 2440: */
1.151 schwarze 2441: if (chdir_status && '/' != *targetdir) {
2442: if (2 == getcwd_status) {
1.59 schwarze 2443: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2444: say("", "getcwd: %s", startdir);
1.197 schwarze 2445: return 0;
1.59 schwarze 2446: }
1.151 schwarze 2447: if (-1 == chdir(startdir)) {
1.59 schwarze 2448: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2449: say("", "&chdir %s", startdir);
1.197 schwarze 2450: return 0;
1.59 schwarze 2451: }
2452: }
1.151 schwarze 2453:
2454: /*
2455: * Always resolve basedir to the canonicalized absolute
2456: * pathname and append a trailing slash, such that
2457: * we can reliably check whether files are inside.
2458: */
1.59 schwarze 2459: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2460: if (report_baddir || errno != ENOENT) {
2461: exitcode = (int)MANDOCLEVEL_BADARG;
2462: say("", "&%s: realpath", targetdir);
2463: }
1.197 schwarze 2464: return 0;
1.59 schwarze 2465: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2466: if (report_baddir || errno != ENOENT) {
2467: exitcode = (int)MANDOCLEVEL_BADARG;
2468: say("", "&chdir");
2469: }
1.197 schwarze 2470: return 0;
1.150 schwarze 2471: }
1.151 schwarze 2472: chdir_status = 1;
1.150 schwarze 2473: cp = strchr(basedir, '\0');
2474: if ('/' != cp[-1]) {
2475: if (cp - basedir >= PATH_MAX - 1) {
2476: exitcode = (int)MANDOCLEVEL_SYSERR;
2477: say("", "Filename too long");
1.197 schwarze 2478: return 0;
1.150 schwarze 2479: }
2480: *cp++ = '/';
2481: *cp = '\0';
1.4 kristaps 2482: }
1.197 schwarze 2483: return 1;
1.56 schwarze 2484: }
2485:
2486: static void
1.59 schwarze 2487: say(const char *file, const char *format, ...)
1.56 schwarze 2488: {
2489: va_list ap;
1.123 schwarze 2490: int use_errno;
1.56 schwarze 2491:
1.59 schwarze 2492: if ('\0' != *basedir)
2493: fprintf(stderr, "%s", basedir);
2494: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2495: fputc('/', stderr);
1.56 schwarze 2496: if ('\0' != *file)
1.59 schwarze 2497: fprintf(stderr, "%s", file);
2498:
1.123 schwarze 2499: use_errno = 1;
2500: if (NULL != format) {
2501: switch (*format) {
1.141 schwarze 2502: case '&':
1.123 schwarze 2503: format++;
2504: break;
1.141 schwarze 2505: case '\0':
1.123 schwarze 2506: format = NULL;
2507: break;
2508: default:
2509: use_errno = 0;
2510: break;
2511: }
2512: }
2513: if (NULL != format) {
2514: if ('\0' != *basedir || '\0' != *file)
2515: fputs(": ", stderr);
2516: va_start(ap, format);
2517: vfprintf(stderr, format, ap);
2518: va_end(ap);
2519: }
2520: if (use_errno) {
2521: if ('\0' != *basedir || '\0' != *file || NULL != format)
2522: fputs(": ", stderr);
1.59 schwarze 2523: perror(NULL);
1.123 schwarze 2524: } else
2525: fputc('\n', stderr);
1.1 kristaps 2526: }
CVSweb