Annotation of mandoc/mandocdb.c, Revision 1.198
1.198 ! schwarze 1: /* $Id: mandocdb.c,v 1.197 2015/10/06 18:32:19 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:
1.139 schwarze 696: /* FALLTHROUGH */
1.141 schwarze 697: case FTS_DP:
1.139 schwarze 698: break;
699:
700: default:
1.60 schwarze 701: if (warnings)
702: say(path, "Not a regular file");
1.50 kristaps 703: continue;
1.60 schwarze 704: }
1.1 kristaps 705:
1.50 kristaps 706: switch (ff->fts_level) {
1.141 schwarze 707: case 0:
1.50 kristaps 708: /* Ignore the root directory. */
709: break;
1.141 schwarze 710: case 1:
1.50 kristaps 711: /*
712: * This might contain manX/ or catX/.
713: * Try to infer this from the name.
714: * If we're not in use_all, enforce it.
715: */
716: cp = ff->fts_name;
717: if (FTS_DP == ff->fts_info)
718: break;
1.1 kristaps 719:
1.50 kristaps 720: if (0 == strncmp(cp, "man", 3)) {
721: dform = FORM_SRC;
1.66 schwarze 722: dsec = cp + 3;
1.50 kristaps 723: } else if (0 == strncmp(cp, "cat", 3)) {
724: dform = FORM_CAT;
1.66 schwarze 725: dsec = cp + 3;
1.94 schwarze 726: } else {
727: dform = FORM_NONE;
728: dsec = NULL;
1.50 kristaps 729: }
1.1 kristaps 730:
1.141 schwarze 731: if (NULL != dsec || use_all)
1.50 kristaps 732: break;
1.1 kristaps 733:
1.56 schwarze 734: if (warnings)
1.59 schwarze 735: say(path, "Unknown directory part");
1.50 kristaps 736: fts_set(f, ff, FTS_SKIP);
737: break;
1.141 schwarze 738: case 2:
1.50 kristaps 739: /*
740: * Possibly our architecture.
741: * If we're descending, keep tabs on it.
742: */
743: if (FTS_DP != ff->fts_info && NULL != dsec)
1.66 schwarze 744: arch = ff->fts_name;
1.94 schwarze 745: else
746: arch = NULL;
1.50 kristaps 747: break;
748: default:
749: if (FTS_DP == ff->fts_info || use_all)
750: break;
1.56 schwarze 751: if (warnings)
1.59 schwarze 752: say(path, "Extraneous directory part");
1.50 kristaps 753: fts_set(f, ff, FTS_SKIP);
754: break;
1.5 kristaps 755: }
1.50 kristaps 756: }
757:
758: fts_close(f);
1.197 schwarze 759: return 1;
1.50 kristaps 760: }
1.5 kristaps 761:
1.50 kristaps 762: /*
1.89 schwarze 763: * Add a file to the mlinks table.
1.50 kristaps 764: * Do not verify that it's a "valid" looking manpage (we'll do that
765: * later).
766: *
767: * Try to infer the manual section, architecture, and page name from the
768: * path, assuming it looks like
769: *
1.141 schwarze 770: * [./]man*[/<arch>]/<name>.<section>
1.50 kristaps 771: * or
772: * [./]cat<section>[/<arch>]/<name>.0
773: *
774: * See treescan() for the fts(3) version of this.
775: */
776: static void
1.59 schwarze 777: filescan(const char *file)
1.50 kristaps 778: {
1.59 schwarze 779: char buf[PATH_MAX];
1.84 schwarze 780: struct stat st;
781: struct mlink *mlink;
1.59 schwarze 782: char *p, *start;
1.5 kristaps 783:
1.50 kristaps 784: assert(use_all);
1.5 kristaps 785:
1.50 kristaps 786: if (0 == strncmp(file, "./", 2))
787: file += 2;
1.5 kristaps 788:
1.139 schwarze 789: /*
790: * We have to do lstat(2) before realpath(3) loses
791: * the information whether this is a symbolic link.
792: * We need to know that because for symbolic links,
793: * we want to use the orginal file name, while for
794: * regular files, we want to use the real path.
795: */
796: if (-1 == lstat(file, &st)) {
797: exitcode = (int)MANDOCLEVEL_BADARG;
798: say(file, "&lstat");
799: return;
800: } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
801: exitcode = (int)MANDOCLEVEL_BADARG;
802: say(file, "Not a regular file");
803: return;
804: }
805:
806: /*
807: * We have to resolve the file name to the real path
808: * in any case for the base directory check.
809: */
1.59 schwarze 810: if (NULL == realpath(file, buf)) {
811: exitcode = (int)MANDOCLEVEL_BADARG;
1.123 schwarze 812: say(file, "&realpath");
1.59 schwarze 813: return;
1.106 schwarze 814: }
815:
1.150 schwarze 816: if (OP_TEST == op)
1.106 schwarze 817: start = buf;
1.150 schwarze 818: else if (strstr(buf, basedir) == buf)
819: start = buf + strlen(basedir);
1.184 schwarze 820: #ifdef HOMEBREWDIR
821: else if (strstr(buf, HOMEBREWDIR) == buf)
822: start = buf;
823: #endif
1.106 schwarze 824: else {
1.59 schwarze 825: exitcode = (int)MANDOCLEVEL_BADARG;
826: say("", "%s: outside base directory", buf);
827: return;
1.106 schwarze 828: }
829:
1.139 schwarze 830: /*
831: * Now we are sure the file is inside our tree.
832: * If it is a symbolic link, ignore the real path
833: * and use the original name.
834: * This implies passing stuff like "cat1/../man1/foo.1"
835: * on the command line won't work. So don't do that.
836: * Note the stat(2) can still fail if the link target
837: * doesn't exist.
838: */
839: if (S_IFLNK & st.st_mode) {
840: if (-1 == stat(buf, &st)) {
841: exitcode = (int)MANDOCLEVEL_BADARG;
842: say(file, "&stat");
843: return;
844: }
1.143 schwarze 845: if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
846: say(file, "Filename too long");
847: return;
848: }
1.150 schwarze 849: start = buf;
850: if (OP_TEST != op && strstr(buf, basedir) == buf)
851: start += strlen(basedir);
1.50 kristaps 852: }
1.106 schwarze 853:
1.84 schwarze 854: mlink = mandoc_calloc(1, sizeof(struct mlink));
1.161 schwarze 855: mlink->dform = FORM_NONE;
1.143 schwarze 856: if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
857: sizeof(mlink->file)) {
858: say(start, "Filename too long");
1.180 schwarze 859: free(mlink);
1.143 schwarze 860: return;
861: }
1.17 schwarze 862:
1.50 kristaps 863: /*
864: * First try to guess our directory structure.
865: * If we find a separator, try to look for man* or cat*.
866: * If we find one of these and what's underneath is a directory,
867: * assume it's an architecture.
868: */
869: if (NULL != (p = strchr(start, '/'))) {
870: *p++ = '\0';
871: if (0 == strncmp(start, "man", 3)) {
1.84 schwarze 872: mlink->dform = FORM_SRC;
1.94 schwarze 873: mlink->dsec = start + 3;
1.50 kristaps 874: } else if (0 == strncmp(start, "cat", 3)) {
1.84 schwarze 875: mlink->dform = FORM_CAT;
1.94 schwarze 876: mlink->dsec = start + 3;
1.17 schwarze 877: }
1.5 kristaps 878:
1.50 kristaps 879: start = p;
1.84 schwarze 880: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
1.50 kristaps 881: *p++ = '\0';
1.94 schwarze 882: mlink->arch = start;
1.50 kristaps 883: start = p;
1.84 schwarze 884: }
1.50 kristaps 885: }
886:
887: /*
888: * Now check the file suffix.
889: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
890: */
891: p = strrchr(start, '\0');
892: while (p-- > start && '/' != *p && '.' != *p)
893: /* Loop. */ ;
894:
895: if ('.' == *p) {
896: *p++ = '\0';
1.94 schwarze 897: mlink->fsec = p;
1.5 kristaps 898: }
899:
1.10 kristaps 900: /*
1.50 kristaps 901: * Now try to parse the name.
902: * Use the filename portion of the path.
1.10 kristaps 903: */
1.84 schwarze 904: mlink->name = start;
1.50 kristaps 905: if (NULL != (p = strrchr(start, '/'))) {
1.84 schwarze 906: mlink->name = p + 1;
1.50 kristaps 907: *p = '\0';
1.84 schwarze 908: }
909: mlink_add(mlink, &st);
1.50 kristaps 910: }
1.5 kristaps 911:
1.50 kristaps 912: static void
1.84 schwarze 913: mlink_add(struct mlink *mlink, const struct stat *st)
1.50 kristaps 914: {
1.83 schwarze 915: struct inodev inodev;
1.78 schwarze 916: struct mpage *mpage;
1.83 schwarze 917: unsigned int slot;
1.50 kristaps 918:
1.84 schwarze 919: assert(NULL != mlink->file);
1.50 kristaps 920:
1.94 schwarze 921: mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
922: mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
923: mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
924: mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1.84 schwarze 925:
926: if ('0' == *mlink->fsec) {
927: free(mlink->fsec);
928: mlink->fsec = mandoc_strdup(mlink->dsec);
929: mlink->fform = FORM_CAT;
930: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
931: mlink->fform = FORM_SRC;
1.74 schwarze 932: else
1.84 schwarze 933: mlink->fform = FORM_NONE;
1.82 schwarze 934:
1.83 schwarze 935: slot = ohash_qlookup(&mlinks, mlink->file);
936: assert(NULL == ohash_find(&mlinks, slot));
937: ohash_insert(&mlinks, slot, mlink);
938:
1.186 schwarze 939: memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1.83 schwarze 940: inodev.st_ino = st->st_ino;
941: inodev.st_dev = st->st_dev;
942: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
943: sizeof(struct inodev), inodev.st_ino);
944: mpage = ohash_find(&mpages, slot);
945: if (NULL == mpage) {
946: mpage = mandoc_calloc(1, sizeof(struct mpage));
947: mpage->inodev.st_ino = inodev.st_ino;
948: mpage->inodev.st_dev = inodev.st_dev;
949: ohash_insert(&mpages, slot, mpage);
950: } else
1.85 schwarze 951: mlink->next = mpage->mlinks;
1.82 schwarze 952: mpage->mlinks = mlink;
1.118 schwarze 953: mlink->mpage = mpage;
1.82 schwarze 954: }
955:
956: static void
957: mlink_free(struct mlink *mlink)
958: {
959:
960: free(mlink->dsec);
961: free(mlink->arch);
962: free(mlink->name);
963: free(mlink->fsec);
964: free(mlink);
1.50 kristaps 965: }
1.3 kristaps 966:
1.50 kristaps 967: static void
1.78 schwarze 968: mpages_free(void)
1.50 kristaps 969: {
1.78 schwarze 970: struct mpage *mpage;
1.82 schwarze 971: struct mlink *mlink;
1.80 schwarze 972: unsigned int slot;
1.3 kristaps 973:
1.80 schwarze 974: mpage = ohash_first(&mpages, &slot);
975: while (NULL != mpage) {
1.82 schwarze 976: while (NULL != (mlink = mpage->mlinks)) {
1.85 schwarze 977: mpage->mlinks = mlink->next;
1.82 schwarze 978: mlink_free(mlink);
979: }
1.78 schwarze 980: free(mpage->sec);
981: free(mpage->arch);
1.82 schwarze 982: free(mpage->title);
983: free(mpage->desc);
1.78 schwarze 984: free(mpage);
1.80 schwarze 985: mpage = ohash_next(&mpages, &slot);
1.50 kristaps 986: }
987: }
1.38 schwarze 988:
1.50 kristaps 989: /*
1.89 schwarze 990: * For each mlink to the mpage, check whether the path looks like
991: * it is formatted, and if it does, check whether a source manual
992: * exists by the same name, ignoring the suffix.
993: * If both conditions hold, drop the mlink.
994: */
995: static void
996: mlinks_undupe(struct mpage *mpage)
997: {
998: char buf[PATH_MAX];
999: struct mlink **prev;
1000: struct mlink *mlink;
1001: char *bufp;
1002:
1003: mpage->form = FORM_CAT;
1.90 schwarze 1004: prev = &mpage->mlinks;
1005: while (NULL != (mlink = *prev)) {
1.89 schwarze 1006: if (FORM_CAT != mlink->dform) {
1007: mpage->form = FORM_NONE;
1.90 schwarze 1008: goto nextlink;
1.89 schwarze 1009: }
1.143 schwarze 1010: (void)strlcpy(buf, mlink->file, sizeof(buf));
1.89 schwarze 1011: bufp = strstr(buf, "cat");
1012: assert(NULL != bufp);
1013: memcpy(bufp, "man", 3);
1014: if (NULL != (bufp = strrchr(buf, '.')))
1015: *++bufp = '\0';
1.143 schwarze 1016: (void)strlcat(buf, mlink->dsec, sizeof(buf));
1.89 schwarze 1017: if (NULL == ohash_find(&mlinks,
1.141 schwarze 1018: ohash_qlookup(&mlinks, buf)))
1.90 schwarze 1019: goto nextlink;
1.89 schwarze 1020: if (warnings)
1021: say(mlink->file, "Man source exists: %s", buf);
1022: if (use_all)
1.90 schwarze 1023: goto nextlink;
1.89 schwarze 1024: *prev = mlink->next;
1025: mlink_free(mlink);
1.90 schwarze 1026: continue;
1027: nextlink:
1028: prev = &(*prev)->next;
1.89 schwarze 1029: }
1030: }
1031:
1.131 schwarze 1032: static void
1.93 schwarze 1033: mlink_check(struct mpage *mpage, struct mlink *mlink)
1034: {
1.131 schwarze 1035: struct str *str;
1036: unsigned int slot;
1.93 schwarze 1037:
1038: /*
1039: * Check whether the manual section given in a file
1040: * agrees with the directory where the file is located.
1041: * Some manuals have suffixes like (3p) on their
1042: * section number either inside the file or in the
1043: * directory name, some are linked into more than one
1044: * section, like encrypt(1) = makekey(8).
1045: */
1046:
1047: if (FORM_SRC == mpage->form &&
1.131 schwarze 1048: strcasecmp(mpage->sec, mlink->dsec))
1.93 schwarze 1049: say(mlink->file, "Section \"%s\" manual in %s directory",
1050: mpage->sec, mlink->dsec);
1051:
1052: /*
1053: * Manual page directories exist for each kernel
1054: * architecture as returned by machine(1).
1055: * However, many manuals only depend on the
1056: * application architecture as returned by arch(1).
1057: * For example, some (2/ARM) manuals are shared
1058: * across the "armish" and "zaurus" kernel
1059: * architectures.
1060: * A few manuals are even shared across completely
1061: * different architectures, for example fdformat(1)
1062: * on amd64, i386, sparc, and sparc64.
1063: */
1064:
1.131 schwarze 1065: if (strcasecmp(mpage->arch, mlink->arch))
1.93 schwarze 1066: say(mlink->file, "Architecture \"%s\" manual in "
1067: "\"%s\" directory", mpage->arch, mlink->arch);
1068:
1.131 schwarze 1069: /*
1070: * XXX
1.133 schwarze 1071: * parse_cat() doesn't set NAME_TITLE yet.
1.131 schwarze 1072: */
1073:
1074: if (FORM_CAT == mpage->form)
1075: return;
1076:
1077: /*
1078: * Check whether this mlink
1079: * appears as a name in the NAME section.
1080: */
1.93 schwarze 1081:
1.133 schwarze 1082: slot = ohash_qlookup(&names, mlink->name);
1083: str = ohash_find(&names, slot);
1.131 schwarze 1084: assert(NULL != str);
1.133 schwarze 1085: if ( ! (NAME_TITLE & str->mask))
1.131 schwarze 1086: say(mlink->file, "Name missing in NAME section");
1.93 schwarze 1087: }
1088:
1.89 schwarze 1089: /*
1.80 schwarze 1090: * Run through the files in the global vector "mpages"
1091: * and add them to the database specified in "basedir".
1.50 kristaps 1092: *
1093: * This handles the parsing scheme itself, using the cues of directory
1094: * and filename to determine whether the file is parsable or not.
1095: */
1.59 schwarze 1096: static void
1.176 schwarze 1097: mpages_merge(struct mparse *mp)
1.50 kristaps 1098: {
1.113 schwarze 1099: char any[] = "any";
1.101 schwarze 1100: struct ohash_info str_info;
1.118 schwarze 1101: struct mpage *mpage, *mpage_dest;
1102: struct mlink *mlink, *mlink_dest;
1.191 schwarze 1103: struct roff_man *man;
1.118 schwarze 1104: char *sodest;
1.112 schwarze 1105: char *cp;
1.162 schwarze 1106: int fd;
1.101 schwarze 1107: unsigned int pslot;
1.69 schwarze 1108:
1.71 schwarze 1109: str_info.alloc = hash_alloc;
1.152 schwarze 1110: str_info.calloc = hash_calloc;
1111: str_info.free = hash_free;
1.181 schwarze 1112: str_info.data = NULL;
1.71 schwarze 1113: str_info.key_offset = offsetof(struct str, key);
1114:
1.171 schwarze 1115: if ( ! nodb)
1.107 schwarze 1116: SQL_EXEC("BEGIN TRANSACTION");
1117:
1.80 schwarze 1118: mpage = ohash_first(&mpages, &pslot);
1.171 schwarze 1119: while (mpage != NULL) {
1.89 schwarze 1120: mlinks_undupe(mpage);
1.182 schwarze 1121: if ((mlink = mpage->mlinks) == NULL) {
1.89 schwarze 1122: mpage = ohash_next(&mpages, &pslot);
1123: continue;
1.50 kristaps 1124: }
1125:
1.133 schwarze 1126: name_mask = NAME_MASK;
1127: ohash_init(&names, 4, &str_info);
1.71 schwarze 1128: ohash_init(&strings, 6, &str_info);
1.50 kristaps 1129: mparse_reset(mp);
1130: man = NULL;
1.124 schwarze 1131: sodest = NULL;
1132:
1.182 schwarze 1133: mparse_open(mp, &fd, mlink->file);
1.162 schwarze 1134: if (fd == -1) {
1.182 schwarze 1135: say(mlink->file, "&open");
1.162 schwarze 1136: goto nextpage;
1.124 schwarze 1137: }
1.14 schwarze 1138:
1139: /*
1.183 schwarze 1140: * Interpret the file as mdoc(7) or man(7) source
1141: * code, unless it is known to be formatted.
1.14 schwarze 1142: */
1.182 schwarze 1143: if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1.183 schwarze 1144: mparse_readfd(mp, fd, mlink->file);
1.192 schwarze 1145: mparse_result(mp, &man, &sodest);
1.76 schwarze 1146: }
1.14 schwarze 1147:
1.171 schwarze 1148: if (sodest != NULL) {
1.118 schwarze 1149: mlink_dest = ohash_find(&mlinks,
1150: ohash_qlookup(&mlinks, sodest));
1.171 schwarze 1151: if (mlink_dest == NULL) {
1152: mandoc_asprintf(&cp, "%s.gz", sodest);
1153: mlink_dest = ohash_find(&mlinks,
1154: ohash_qlookup(&mlinks, cp));
1155: free(cp);
1156: }
1157: if (mlink_dest != NULL) {
1.118 schwarze 1158:
1159: /* The .so target exists. */
1160:
1161: mpage_dest = mlink_dest->mpage;
1162: while (1) {
1163: mlink->mpage = mpage_dest;
1164:
1165: /*
1166: * If the target was already
1167: * processed, add the links
1168: * to the database now.
1169: * Otherwise, this will
1170: * happen when we come
1171: * to the target.
1172: */
1173:
1.137 schwarze 1174: if (mpage_dest->pageid)
1.169 schwarze 1175: dbadd_mlink_name(mlink);
1.118 schwarze 1176:
1.171 schwarze 1177: if (mlink->next == NULL)
1.118 schwarze 1178: break;
1179: mlink = mlink->next;
1180: }
1181:
1182: /* Move all links to the target. */
1183:
1184: mlink->next = mlink_dest->next;
1185: mlink_dest->next = mpage->mlinks;
1186: mpage->mlinks = NULL;
1187: }
1.124 schwarze 1188: goto nextpage;
1.192 schwarze 1189: } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1.82 schwarze 1190: mpage->form = FORM_SRC;
1.193 schwarze 1191: mpage->sec = man->meta.msec;
1.155 schwarze 1192: mpage->sec = mandoc_strdup(
1.171 schwarze 1193: mpage->sec == NULL ? "" : mpage->sec);
1.193 schwarze 1194: mpage->arch = man->meta.arch;
1.84 schwarze 1195: mpage->arch = mandoc_strdup(
1.171 schwarze 1196: mpage->arch == NULL ? "" : mpage->arch);
1.193 schwarze 1197: mpage->title = mandoc_strdup(man->meta.title);
1.192 schwarze 1198: } else if (man != NULL && man->macroset == MACROSET_MAN) {
1.82 schwarze 1199: mpage->form = FORM_SRC;
1.193 schwarze 1200: mpage->sec = mandoc_strdup(man->meta.msec);
1.182 schwarze 1201: mpage->arch = mandoc_strdup(mlink->arch);
1.193 schwarze 1202: mpage->title = mandoc_strdup(man->meta.title);
1.76 schwarze 1203: } else {
1.82 schwarze 1204: mpage->form = FORM_CAT;
1.182 schwarze 1205: mpage->sec = mandoc_strdup(mlink->dsec);
1206: mpage->arch = mandoc_strdup(mlink->arch);
1207: mpage->title = mandoc_strdup(mlink->name);
1.76 schwarze 1208: }
1.97 schwarze 1209: putkey(mpage, mpage->sec, TYPE_sec);
1.164 schwarze 1210: if (*mpage->arch != '\0')
1211: putkey(mpage, mpage->arch, TYPE_arch);
1.50 kristaps 1212:
1.182 schwarze 1213: for ( ; mlink != NULL; mlink = mlink->next) {
1.97 schwarze 1214: if ('\0' != *mlink->dsec)
1215: putkey(mpage, mlink->dsec, TYPE_sec);
1216: if ('\0' != *mlink->fsec)
1217: putkey(mpage, mlink->fsec, TYPE_sec);
1.98 schwarze 1218: putkey(mpage, '\0' == *mlink->arch ?
1.113 schwarze 1219: any : mlink->arch, TYPE_arch);
1.133 schwarze 1220: putkey(mpage, mlink->name, NAME_FILE);
1.97 schwarze 1221: }
1.12 schwarze 1222:
1.174 schwarze 1223: assert(mpage->desc == NULL);
1.192 schwarze 1224: if (man != NULL && man->macroset == MACROSET_MDOC)
1.193 schwarze 1225: parse_mdoc(mpage, &man->meta, man->first);
1.174 schwarze 1226: else if (man != NULL)
1.193 schwarze 1227: parse_man(mpage, &man->meta, man->first);
1.50 kristaps 1228: else
1.162 schwarze 1229: parse_cat(mpage, fd);
1.174 schwarze 1230: if (mpage->desc == NULL)
1.135 schwarze 1231: mpage->desc = mandoc_strdup(mpage->mlinks->name);
1.131 schwarze 1232:
1233: if (warnings && !use_all)
1234: for (mlink = mpage->mlinks; mlink;
1235: mlink = mlink->next)
1236: mlink_check(mpage, mlink);
1.44 kristaps 1237:
1.176 schwarze 1238: dbadd(mpage);
1.182 schwarze 1239: mlink = mpage->mlinks;
1.124 schwarze 1240:
1241: nextpage:
1.66 schwarze 1242: ohash_delete(&strings);
1.133 schwarze 1243: ohash_delete(&names);
1.80 schwarze 1244: mpage = ohash_next(&mpages, &pslot);
1.50 kristaps 1245: }
1.107 schwarze 1246:
1247: if (0 == nodb)
1248: SQL_EXEC("END TRANSACTION");
1.130 schwarze 1249: }
1250:
1251: static void
1252: names_check(void)
1253: {
1254: sqlite3_stmt *stmt;
1255: const char *name, *sec, *arch, *key;
1256: int irc;
1257:
1258: sqlite3_prepare_v2(db,
1259: "SELECT name, sec, arch, key FROM ("
1.133 schwarze 1260: "SELECT name AS key, pageid FROM names "
1.130 schwarze 1261: "WHERE bits & ? AND NOT EXISTS ("
1262: "SELECT pageid FROM mlinks "
1.133 schwarze 1263: "WHERE mlinks.pageid == names.pageid "
1264: "AND mlinks.name == names.name"
1.130 schwarze 1265: ")"
1266: ") JOIN ("
1.142 schwarze 1267: "SELECT sec, arch, name, pageid FROM mlinks "
1268: "GROUP BY pageid"
1.130 schwarze 1269: ") USING (pageid);",
1270: -1, &stmt, NULL);
1271:
1.134 schwarze 1272: if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1273: say("", "%s", sqlite3_errmsg(db));
1.130 schwarze 1274:
1275: while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1.154 schwarze 1276: name = (const char *)sqlite3_column_text(stmt, 0);
1277: sec = (const char *)sqlite3_column_text(stmt, 1);
1278: arch = (const char *)sqlite3_column_text(stmt, 2);
1279: key = (const char *)sqlite3_column_text(stmt, 3);
1.130 schwarze 1280: say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1281: '\0' == *arch ? "" : "/",
1282: '\0' == *arch ? "" : arch, key);
1283: }
1284: sqlite3_finalize(stmt);
1.50 kristaps 1285: }
1.12 schwarze 1286:
1.50 kristaps 1287: static void
1.124 schwarze 1288: parse_cat(struct mpage *mpage, int fd)
1.50 kristaps 1289: {
1290: FILE *stream;
1291: char *line, *p, *title;
1292: size_t len, plen, titlesz;
1.12 schwarze 1293:
1.124 schwarze 1294: stream = (-1 == fd) ?
1295: fopen(mpage->mlinks->file, "r") :
1296: fdopen(fd, "r");
1297: if (NULL == stream) {
1.159 schwarze 1298: if (-1 != fd)
1299: close(fd);
1.56 schwarze 1300: if (warnings)
1.123 schwarze 1301: say(mpage->mlinks->file, "&fopen");
1.50 kristaps 1302: return;
1303: }
1.1 kristaps 1304:
1.50 kristaps 1305: /* Skip to first blank line. */
1.1 kristaps 1306:
1.50 kristaps 1307: while (NULL != (line = fgetln(stream, &len)))
1308: if ('\n' == *line)
1309: break;
1.1 kristaps 1310:
1.50 kristaps 1311: /*
1312: * Assume the first line that is not indented
1313: * is the first section header. Skip to it.
1314: */
1.1 kristaps 1315:
1.50 kristaps 1316: while (NULL != (line = fgetln(stream, &len)))
1317: if ('\n' != *line && ' ' != *line)
1318: break;
1.141 schwarze 1319:
1.50 kristaps 1320: /*
1321: * Read up until the next section into a buffer.
1322: * Strip the leading and trailing newline from each read line,
1323: * appending a trailing space.
1324: * Ignore empty (whitespace-only) lines.
1325: */
1.1 kristaps 1326:
1.50 kristaps 1327: titlesz = 0;
1328: title = NULL;
1.38 schwarze 1329:
1.50 kristaps 1330: while (NULL != (line = fgetln(stream, &len))) {
1331: if (' ' != *line || '\n' != line[len - 1])
1332: break;
1333: while (len > 0 && isspace((unsigned char)*line)) {
1334: line++;
1335: len--;
1336: }
1337: if (1 == len)
1.38 schwarze 1338: continue;
1.50 kristaps 1339: title = mandoc_realloc(title, titlesz + len);
1340: memcpy(title + titlesz, line, len);
1341: titlesz += len;
1342: title[titlesz - 1] = ' ';
1343: }
1.38 schwarze 1344:
1.50 kristaps 1345: /*
1346: * If no page content can be found, or the input line
1347: * is already the next section header, or there is no
1348: * trailing newline, reuse the page title as the page
1349: * description.
1350: */
1.44 kristaps 1351:
1.50 kristaps 1352: if (NULL == title || '\0' == *title) {
1.56 schwarze 1353: if (warnings)
1.82 schwarze 1354: say(mpage->mlinks->file,
1355: "Cannot find NAME section");
1.50 kristaps 1356: fclose(stream);
1357: free(title);
1358: return;
1359: }
1.1 kristaps 1360:
1.50 kristaps 1361: title = mandoc_realloc(title, titlesz + 1);
1362: title[titlesz] = '\0';
1.33 schwarze 1363:
1.50 kristaps 1364: /*
1365: * Skip to the first dash.
1366: * Use the remaining line as the description (no more than 70
1367: * bytes).
1368: */
1.33 schwarze 1369:
1.50 kristaps 1370: if (NULL != (p = strstr(title, "- "))) {
1371: for (p += 2; ' ' == *p || '\b' == *p; p++)
1372: /* Skip to next word. */ ;
1373: } else {
1.56 schwarze 1374: if (warnings)
1.82 schwarze 1375: say(mpage->mlinks->file,
1376: "No dash in title line");
1.50 kristaps 1377: p = title;
1378: }
1.38 schwarze 1379:
1.50 kristaps 1380: plen = strlen(p);
1.1 kristaps 1381:
1.50 kristaps 1382: /* Strip backspace-encoding from line. */
1.1 kristaps 1383:
1.50 kristaps 1384: while (NULL != (line = memchr(p, '\b', plen))) {
1385: len = line - p;
1386: if (0 == len) {
1387: memmove(line, line + 1, plen--);
1388: continue;
1.141 schwarze 1389: }
1.50 kristaps 1390: memmove(line - 1, line + 1, plen - len);
1391: plen -= 2;
1392: }
1.1 kristaps 1393:
1.78 schwarze 1394: mpage->desc = mandoc_strdup(p);
1.50 kristaps 1395: fclose(stream);
1396: free(title);
1397: }
1.1 kristaps 1398:
1.50 kristaps 1399: /*
1400: * Put a type/word pair into the word database for this particular file.
1401: */
1402: static void
1.112 schwarze 1403: putkey(const struct mpage *mpage, char *value, uint64_t type)
1.50 kristaps 1404: {
1.112 schwarze 1405: char *cp;
1.18 kristaps 1406:
1.50 kristaps 1407: assert(NULL != value);
1.112 schwarze 1408: if (TYPE_arch == type)
1409: for (cp = value; *cp; cp++)
1410: if (isupper((unsigned char)*cp))
1411: *cp = _tolower((unsigned char)*cp);
1.78 schwarze 1412: putkeys(mpage, value, strlen(value), type);
1.3 kristaps 1413: }
1414:
1415: /*
1.50 kristaps 1416: * Grok all nodes at or below a certain mdoc node into putkey().
1.3 kristaps 1417: */
1418: static void
1.78 schwarze 1419: putmdockey(const struct mpage *mpage,
1.189 schwarze 1420: const struct roff_node *n, uint64_t m)
1.3 kristaps 1421: {
1.18 kristaps 1422:
1.50 kristaps 1423: for ( ; NULL != n; n = n->next) {
1424: if (NULL != n->child)
1.78 schwarze 1425: putmdockey(mpage, n->child, m);
1.188 schwarze 1426: if (n->type == ROFFT_TEXT)
1.78 schwarze 1427: putkey(mpage, n->string, m);
1.50 kristaps 1428: }
1429: }
1.18 kristaps 1430:
1.61 schwarze 1431: static void
1.190 schwarze 1432: parse_man(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1433: const struct roff_node *n)
1.50 kristaps 1434: {
1.189 schwarze 1435: const struct roff_node *head, *body;
1.121 schwarze 1436: char *start, *title;
1.50 kristaps 1437: char byte;
1.121 schwarze 1438: size_t sz;
1.18 kristaps 1439:
1.50 kristaps 1440: if (NULL == n)
1.61 schwarze 1441: return;
1.18 kristaps 1442:
1.50 kristaps 1443: /*
1444: * We're only searching for one thing: the first text child in
1445: * the BODY of a NAME section. Since we don't keep track of
1446: * sections in -man, run some hoops to find out whether we're in
1447: * the correct section or not.
1448: */
1.18 kristaps 1449:
1.188 schwarze 1450: if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1.50 kristaps 1451: body = n;
1452: assert(body->parent);
1453: if (NULL != (head = body->parent->head) &&
1.141 schwarze 1454: 1 == head->nchild &&
1455: NULL != (head = (head->child)) &&
1.188 schwarze 1456: head->type == ROFFT_TEXT &&
1.141 schwarze 1457: 0 == strcmp(head->string, "NAME") &&
1458: NULL != body->child) {
1.3 kristaps 1459:
1.50 kristaps 1460: /*
1461: * Suck the entire NAME section into memory.
1462: * Yes, we might run away.
1463: * But too many manuals have big, spread-out
1464: * NAME sections over many lines.
1465: */
1.3 kristaps 1466:
1.121 schwarze 1467: title = NULL;
1.194 schwarze 1468: deroff(&title, body);
1.50 kristaps 1469: if (NULL == title)
1.61 schwarze 1470: return;
1.18 kristaps 1471:
1.141 schwarze 1472: /*
1.50 kristaps 1473: * Go through a special heuristic dance here.
1474: * Conventionally, one or more manual names are
1475: * comma-specified prior to a whitespace, then a
1476: * dash, then a description. Try to puzzle out
1477: * the name parts here.
1478: */
1.18 kristaps 1479:
1.121 schwarze 1480: start = title;
1.50 kristaps 1481: for ( ;; ) {
1482: sz = strcspn(start, " ,");
1483: if ('\0' == start[sz])
1484: break;
1.1 kristaps 1485:
1.50 kristaps 1486: byte = start[sz];
1487: start[sz] = '\0';
1.110 schwarze 1488:
1489: /*
1490: * Assume a stray trailing comma in the
1491: * name list if a name begins with a dash.
1492: */
1493:
1494: if ('-' == start[0] ||
1495: ('\\' == start[0] && '-' == start[1]))
1496: break;
1.1 kristaps 1497:
1.133 schwarze 1498: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1499: if ( ! (mpage->name_head_done ||
1500: strcasecmp(start, meta->title))) {
1501: putkey(mpage, start, NAME_HEAD);
1502: mpage->name_head_done = 1;
1503: }
1.1 kristaps 1504:
1.50 kristaps 1505: if (' ' == byte) {
1506: start += sz + 1;
1507: break;
1508: }
1.1 kristaps 1509:
1.50 kristaps 1510: assert(',' == byte);
1511: start += sz + 1;
1512: while (' ' == *start)
1513: start++;
1514: }
1.1 kristaps 1515:
1.121 schwarze 1516: if (start == title) {
1.133 schwarze 1517: putkey(mpage, start, NAME_TITLE);
1.174 schwarze 1518: if ( ! (mpage->name_head_done ||
1519: strcasecmp(start, meta->title))) {
1520: putkey(mpage, start, NAME_HEAD);
1521: mpage->name_head_done = 1;
1522: }
1.50 kristaps 1523: free(title);
1.61 schwarze 1524: return;
1.50 kristaps 1525: }
1.1 kristaps 1526:
1.50 kristaps 1527: while (isspace((unsigned char)*start))
1528: start++;
1.1 kristaps 1529:
1.50 kristaps 1530: if (0 == strncmp(start, "-", 1))
1531: start += 1;
1532: else if (0 == strncmp(start, "\\-\\-", 4))
1533: start += 4;
1534: else if (0 == strncmp(start, "\\-", 2))
1535: start += 2;
1536: else if (0 == strncmp(start, "\\(en", 4))
1537: start += 4;
1538: else if (0 == strncmp(start, "\\(em", 4))
1539: start += 4;
1.1 kristaps 1540:
1.50 kristaps 1541: while (' ' == *start)
1542: start++;
1.1 kristaps 1543:
1.78 schwarze 1544: mpage->desc = mandoc_strdup(start);
1.50 kristaps 1545: free(title);
1.61 schwarze 1546: return;
1.50 kristaps 1547: }
1548: }
1.1 kristaps 1549:
1.77 schwarze 1550: for (n = n->child; n; n = n->next) {
1.78 schwarze 1551: if (NULL != mpage->desc)
1.77 schwarze 1552: break;
1.174 schwarze 1553: parse_man(mpage, meta, n);
1.77 schwarze 1554: }
1.1 kristaps 1555: }
1556:
1557: static void
1.190 schwarze 1558: parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1559: const struct roff_node *n)
1.1 kristaps 1560: {
1561:
1.50 kristaps 1562: assert(NULL != n);
1563: for (n = n->child; NULL != n; n = n->next) {
1564: switch (n->type) {
1.188 schwarze 1565: case ROFFT_ELEM:
1.50 kristaps 1566: /* FALLTHROUGH */
1.188 schwarze 1567: case ROFFT_BLOCK:
1.50 kristaps 1568: /* FALLTHROUGH */
1.188 schwarze 1569: case ROFFT_HEAD:
1.50 kristaps 1570: /* FALLTHROUGH */
1.188 schwarze 1571: case ROFFT_BODY:
1.50 kristaps 1572: /* FALLTHROUGH */
1.188 schwarze 1573: case ROFFT_TAIL:
1.50 kristaps 1574: if (NULL != mdocs[n->tok].fp)
1.172 schwarze 1575: if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1.50 kristaps 1576: break;
1.68 schwarze 1577: if (mdocs[n->tok].mask)
1.78 schwarze 1578: putmdockey(mpage, n->child,
1579: mdocs[n->tok].mask);
1.50 kristaps 1580: break;
1581: default:
1.188 schwarze 1582: assert(n->type != ROFFT_ROOT);
1.50 kristaps 1583: continue;
1584: }
1585: if (NULL != n->child)
1.172 schwarze 1586: parse_mdoc(mpage, meta, n);
1.1 kristaps 1587: }
1588: }
1589:
1.25 schwarze 1590: static int
1.190 schwarze 1591: parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1592: const struct roff_node *n)
1.1 kristaps 1593: {
1.176 schwarze 1594: char *start, *end;
1.1 kristaps 1595: size_t sz;
1.25 schwarze 1596:
1.50 kristaps 1597: if (SEC_SYNOPSIS != n->sec ||
1.141 schwarze 1598: NULL == (n = n->child) ||
1.188 schwarze 1599: n->type != ROFFT_TEXT)
1.197 schwarze 1600: return 0;
1.1 kristaps 1601:
1602: /*
1603: * Only consider those `Fd' macro fields that begin with an
1604: * "inclusion" token (versus, e.g., #define).
1605: */
1.50 kristaps 1606:
1.1 kristaps 1607: if (strcmp("#include", n->string))
1.197 schwarze 1608: return 0;
1.1 kristaps 1609:
1.188 schwarze 1610: if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1.197 schwarze 1611: return 0;
1.1 kristaps 1612:
1613: /*
1614: * Strip away the enclosing angle brackets and make sure we're
1615: * not zero-length.
1616: */
1617:
1618: start = n->string;
1619: if ('<' == *start || '"' == *start)
1620: start++;
1621:
1622: if (0 == (sz = strlen(start)))
1.197 schwarze 1623: return 0;
1.1 kristaps 1624:
1625: end = &start[(int)sz - 1];
1626: if ('>' == *end || '"' == *end)
1627: end--;
1628:
1.50 kristaps 1629: if (end > start)
1.78 schwarze 1630: putkeys(mpage, start, end - start + 1, TYPE_In);
1.197 schwarze 1631: return 0;
1.1 kristaps 1632: }
1633:
1.178 schwarze 1634: static void
1.189 schwarze 1635: parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1.1 kristaps 1636: {
1.112 schwarze 1637: char *cp;
1.178 schwarze 1638: size_t sz;
1.1 kristaps 1639:
1.188 schwarze 1640: if (n->type != ROFFT_TEXT)
1.178 schwarze 1641: return;
1.25 schwarze 1642:
1.178 schwarze 1643: /* Skip function pointer punctuation. */
1.1 kristaps 1644:
1.178 schwarze 1645: cp = n->string;
1646: while (*cp == '(' || *cp == '*')
1.1 kristaps 1647: cp++;
1.178 schwarze 1648: sz = strcspn(cp, "()");
1.1 kristaps 1649:
1.178 schwarze 1650: putkeys(mpage, cp, sz, TYPE_Fn);
1.173 schwarze 1651: if (n->sec == SEC_SYNOPSIS)
1.178 schwarze 1652: putkeys(mpage, cp, sz, NAME_SYN);
1653: }
1.25 schwarze 1654:
1.178 schwarze 1655: static int
1.190 schwarze 1656: parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1657: const struct roff_node *n)
1.178 schwarze 1658: {
1659:
1660: if (n->child == NULL)
1.197 schwarze 1661: return 0;
1.178 schwarze 1662:
1663: parse_mdoc_fname(mpage, n->child);
1.25 schwarze 1664:
1.178 schwarze 1665: for (n = n->child->next; n != NULL; n = n->next)
1.188 schwarze 1666: if (n->type == ROFFT_TEXT)
1.78 schwarze 1667: putkey(mpage, n->string, TYPE_Fa);
1.25 schwarze 1668:
1.197 schwarze 1669: return 0;
1.173 schwarze 1670: }
1671:
1672: static int
1.190 schwarze 1673: parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1674: const struct roff_node *n)
1.173 schwarze 1675: {
1.177 schwarze 1676:
1.188 schwarze 1677: if (n->type != ROFFT_HEAD)
1.197 schwarze 1678: return 1;
1.173 schwarze 1679:
1.178 schwarze 1680: if (n->child != NULL)
1681: parse_mdoc_fname(mpage, n->child);
1682:
1.197 schwarze 1683: return 0;
1.1 kristaps 1684: }
1685:
1.25 schwarze 1686: static int
1.190 schwarze 1687: parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1688: const struct roff_node *n)
1.1 kristaps 1689: {
1.67 schwarze 1690: char *cp;
1.1 kristaps 1691:
1692: if (NULL == (n = n->child))
1.197 schwarze 1693: return 0;
1.1 kristaps 1694:
1.67 schwarze 1695: if (NULL == n->next) {
1.78 schwarze 1696: putkey(mpage, n->string, TYPE_Xr);
1.197 schwarze 1697: return 0;
1.67 schwarze 1698: }
1699:
1.120 schwarze 1700: mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1.78 schwarze 1701: putkey(mpage, cp, TYPE_Xr);
1.67 schwarze 1702: free(cp);
1.197 schwarze 1703: return 0;
1.1 kristaps 1704: }
1705:
1.25 schwarze 1706: static int
1.190 schwarze 1707: parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1708: const struct roff_node *n)
1.1 kristaps 1709: {
1710:
1.188 schwarze 1711: if (n->type == ROFFT_BODY)
1.194 schwarze 1712: deroff(&mpage->desc, n);
1.197 schwarze 1713: return 0;
1.1 kristaps 1714: }
1715:
1.25 schwarze 1716: static int
1.190 schwarze 1717: parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1718: const struct roff_node *n)
1.1 kristaps 1719: {
1720:
1.129 schwarze 1721: if (SEC_NAME == n->sec)
1.133 schwarze 1722: putmdockey(mpage, n->child, NAME_TITLE);
1.188 schwarze 1723: else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1.172 schwarze 1724: if (n->child == NULL)
1725: putkey(mpage, meta->name, NAME_SYN);
1726: else
1727: putmdockey(mpage, n->child, NAME_SYN);
1.174 schwarze 1728: }
1729: if ( ! (mpage->name_head_done ||
1730: n->child == NULL || n->child->string == NULL ||
1731: strcasecmp(n->child->string, meta->title))) {
1.188 schwarze 1732: putkey(mpage, n->child->string, ROFFT_HEAD);
1.174 schwarze 1733: mpage->name_head_done = 1;
1.172 schwarze 1734: }
1.197 schwarze 1735: return 0;
1.1 kristaps 1736: }
1737:
1.25 schwarze 1738: static int
1.190 schwarze 1739: parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1740: const struct roff_node *n)
1.1 kristaps 1741: {
1742:
1.197 schwarze 1743: return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1.1 kristaps 1744: }
1745:
1.50 kristaps 1746: static int
1.190 schwarze 1747: parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1748: const struct roff_node *n)
1.1 kristaps 1749: {
1750:
1.197 schwarze 1751: return n->type == ROFFT_HEAD;
1.50 kristaps 1752: }
1.1 kristaps 1753:
1.50 kristaps 1754: static int
1.190 schwarze 1755: parse_mdoc_body(struct mpage *mpage, const struct roff_meta *meta,
1.189 schwarze 1756: const struct roff_node *n)
1.50 kristaps 1757: {
1.1 kristaps 1758:
1.197 schwarze 1759: return n->type == ROFFT_BODY;
1.1 kristaps 1760: }
1761:
1.50 kristaps 1762: /*
1.66 schwarze 1763: * Add a string to the hash table for the current manual.
1764: * Each string has a bitmask telling which macros it belongs to.
1765: * When we finish the manual, we'll dump the table.
1.50 kristaps 1766: */
1767: static void
1.176 schwarze 1768: putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1.50 kristaps 1769: {
1.133 schwarze 1770: struct ohash *htab;
1.50 kristaps 1771: struct str *s;
1.111 schwarze 1772: const char *end;
1.65 schwarze 1773: unsigned int slot;
1.176 schwarze 1774: int i, mustfree;
1.25 schwarze 1775:
1.50 kristaps 1776: if (0 == sz)
1777: return;
1.111 schwarze 1778:
1.176 schwarze 1779: mustfree = render_string(&cp, &sz);
1780:
1.133 schwarze 1781: if (TYPE_Nm & v) {
1782: htab = &names;
1783: v &= name_mask;
1.169 schwarze 1784: if (v & NAME_FIRST)
1785: name_mask &= ~NAME_FIRST;
1.133 schwarze 1786: if (debug > 1)
1787: say(mpage->mlinks->file,
1.175 schwarze 1788: "Adding name %*s, bits=%d", sz, cp, v);
1.133 schwarze 1789: } else {
1790: htab = &strings;
1791: if (debug > 1)
1792: for (i = 0; i < mansearch_keymax; i++)
1.163 schwarze 1793: if ((uint64_t)1 << i & v)
1.133 schwarze 1794: say(mpage->mlinks->file,
1795: "Adding key %s=%*s",
1796: mansearch_keynames[i], sz, cp);
1.111 schwarze 1797: }
1.25 schwarze 1798:
1.65 schwarze 1799: end = cp + sz;
1.133 schwarze 1800: slot = ohash_qlookupi(htab, cp, &end);
1801: s = ohash_find(htab, slot);
1.25 schwarze 1802:
1.78 schwarze 1803: if (NULL != s && mpage == s->mpage) {
1.50 kristaps 1804: s->mask |= v;
1805: return;
1806: } else if (NULL == s) {
1.144 schwarze 1807: s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1.50 kristaps 1808: memcpy(s->key, cp, sz);
1.133 schwarze 1809: ohash_insert(htab, slot, s);
1.1 kristaps 1810: }
1.78 schwarze 1811: s->mpage = mpage;
1.50 kristaps 1812: s->mask = v;
1.176 schwarze 1813:
1814: if (mustfree)
1815: free(cp);
1.1 kristaps 1816: }
1817:
1.50 kristaps 1818: /*
1819: * Take a Unicode codepoint and produce its UTF-8 encoding.
1820: * This isn't the best way to do this, but it works.
1821: * The magic numbers are from the UTF-8 packaging.
1822: * They're not as scary as they seem: read the UTF-8 spec for details.
1823: */
1824: static size_t
1825: utf8(unsigned int cp, char out[7])
1.1 kristaps 1826: {
1.50 kristaps 1827: size_t rc;
1.1 kristaps 1828:
1.50 kristaps 1829: rc = 0;
1830: if (cp <= 0x0000007F) {
1831: rc = 1;
1832: out[0] = (char)cp;
1833: } else if (cp <= 0x000007FF) {
1834: rc = 2;
1835: out[0] = (cp >> 6 & 31) | 192;
1836: out[1] = (cp & 63) | 128;
1837: } else if (cp <= 0x0000FFFF) {
1838: rc = 3;
1839: out[0] = (cp >> 12 & 15) | 224;
1840: out[1] = (cp >> 6 & 63) | 128;
1841: out[2] = (cp & 63) | 128;
1842: } else if (cp <= 0x001FFFFF) {
1843: rc = 4;
1844: out[0] = (cp >> 18 & 7) | 240;
1845: out[1] = (cp >> 12 & 63) | 128;
1846: out[2] = (cp >> 6 & 63) | 128;
1847: out[3] = (cp & 63) | 128;
1848: } else if (cp <= 0x03FFFFFF) {
1849: rc = 5;
1850: out[0] = (cp >> 24 & 3) | 248;
1851: out[1] = (cp >> 18 & 63) | 128;
1852: out[2] = (cp >> 12 & 63) | 128;
1853: out[3] = (cp >> 6 & 63) | 128;
1854: out[4] = (cp & 63) | 128;
1855: } else if (cp <= 0x7FFFFFFF) {
1856: rc = 6;
1857: out[0] = (cp >> 30 & 1) | 252;
1858: out[1] = (cp >> 24 & 63) | 128;
1859: out[2] = (cp >> 18 & 63) | 128;
1860: out[3] = (cp >> 12 & 63) | 128;
1861: out[4] = (cp >> 6 & 63) | 128;
1862: out[5] = (cp & 63) | 128;
1863: } else
1.197 schwarze 1864: return 0;
1.1 kristaps 1865:
1.50 kristaps 1866: out[rc] = '\0';
1.197 schwarze 1867: return rc;
1.50 kristaps 1868: }
1.1 kristaps 1869:
1.50 kristaps 1870: /*
1.176 schwarze 1871: * If the string contains escape sequences,
1872: * replace it with an allocated rendering and return 1,
1873: * such that the caller can free it after use.
1874: * Otherwise, do nothing and return 0.
1.50 kristaps 1875: */
1.176 schwarze 1876: static int
1877: render_string(char **public, size_t *psz)
1.50 kristaps 1878: {
1.176 schwarze 1879: const char *src, *scp, *addcp, *seq;
1880: char *dst;
1881: size_t ssz, dsz, addsz;
1.114 schwarze 1882: char utfbuf[7], res[6];
1.176 schwarze 1883: int seqlen, unicode;
1.50 kristaps 1884:
1885: res[0] = '\\';
1886: res[1] = '\t';
1887: res[2] = ASCII_NBRSP;
1888: res[3] = ASCII_HYPH;
1.114 schwarze 1889: res[4] = ASCII_BREAK;
1890: res[5] = '\0';
1.1 kristaps 1891:
1.176 schwarze 1892: src = scp = *public;
1893: ssz = *psz;
1894: dst = NULL;
1895: dsz = 0;
1896:
1897: while (scp < src + *psz) {
1898:
1899: /* Leave normal characters unchanged. */
1900:
1901: if (strchr(res, *scp) == NULL) {
1902: if (dst != NULL)
1903: dst[dsz++] = *scp;
1904: scp++;
1905: continue;
1906: }
1.46 kristaps 1907:
1.50 kristaps 1908: /*
1.176 schwarze 1909: * Found something that requires replacing,
1910: * make sure we have a destination buffer.
1.50 kristaps 1911: */
1.176 schwarze 1912:
1913: if (dst == NULL) {
1914: dst = mandoc_malloc(ssz + 1);
1915: dsz = scp - src;
1916: memcpy(dst, src, dsz);
1.50 kristaps 1917: }
1.46 kristaps 1918:
1.176 schwarze 1919: /* Handle single-char special characters. */
1920:
1921: switch (*scp) {
1922: case '\\':
1923: break;
1.141 schwarze 1924: case '\t':
1.114 schwarze 1925: /* FALLTHROUGH */
1.141 schwarze 1926: case ASCII_NBRSP:
1.176 schwarze 1927: dst[dsz++] = ' ';
1928: scp++;
1929: continue;
1930: case ASCII_HYPH:
1931: dst[dsz++] = '-';
1.114 schwarze 1932: /* FALLTHROUGH */
1.141 schwarze 1933: case ASCII_BREAK:
1.176 schwarze 1934: scp++;
1.50 kristaps 1935: continue;
1.114 schwarze 1936: default:
1.176 schwarze 1937: abort();
1.114 schwarze 1938: }
1.46 kristaps 1939:
1.50 kristaps 1940: /*
1.176 schwarze 1941: * Found an escape sequence.
1942: * Read past the slash, then parse it.
1943: * Ignore everything except characters.
1.50 kristaps 1944: */
1.95 schwarze 1945:
1.176 schwarze 1946: scp++;
1947: if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1.50 kristaps 1948: continue;
1.1 kristaps 1949:
1.50 kristaps 1950: /*
1.95 schwarze 1951: * Render the special character
1952: * as either UTF-8 or ASCII.
1.50 kristaps 1953: */
1.95 schwarze 1954:
1955: if (write_utf8) {
1.176 schwarze 1956: unicode = mchars_spec2cp(mchars, seq, seqlen);
1957: if (unicode <= 0)
1.95 schwarze 1958: continue;
1.176 schwarze 1959: addsz = utf8(unicode, utfbuf);
1960: if (addsz == 0)
1.95 schwarze 1961: continue;
1.176 schwarze 1962: addcp = utfbuf;
1.95 schwarze 1963: } else {
1.176 schwarze 1964: addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1965: if (addcp == NULL)
1.95 schwarze 1966: continue;
1.176 schwarze 1967: if (*addcp == ASCII_NBRSP) {
1968: addcp = " ";
1969: addsz = 1;
1.95 schwarze 1970: }
1971: }
1.1 kristaps 1972:
1.50 kristaps 1973: /* Copy the rendered glyph into the stream. */
1.1 kristaps 1974:
1.176 schwarze 1975: ssz += addsz;
1976: dst = mandoc_realloc(dst, ssz + 1);
1977: memcpy(dst + dsz, addcp, addsz);
1978: dsz += addsz;
1979: }
1980: if (dst != NULL) {
1981: *public = dst;
1982: *psz = dsz;
1983: }
1984:
1985: /* Trim trailing whitespace and NUL-terminate. */
1986:
1987: while (*psz > 0 && (*public)[*psz - 1] == ' ')
1988: --*psz;
1989: if (dst != NULL) {
1990: (*public)[*psz] = '\0';
1.197 schwarze 1991: return 1;
1.176 schwarze 1992: } else
1.197 schwarze 1993: return 0;
1.1 kristaps 1994: }
1995:
1.118 schwarze 1996: static void
1997: dbadd_mlink(const struct mlink *mlink)
1998: {
1999: size_t i;
2000:
2001: i = 1;
2002: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2003: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2004: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1.137 schwarze 2005: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1.118 schwarze 2006: SQL_STEP(stmts[STMT_INSERT_LINK]);
2007: sqlite3_reset(stmts[STMT_INSERT_LINK]);
1.169 schwarze 2008: }
2009:
2010: static void
2011: dbadd_mlink_name(const struct mlink *mlink)
2012: {
1.175 schwarze 2013: uint64_t bits;
1.169 schwarze 2014: size_t i;
2015:
2016: dbadd_mlink(mlink);
1.160 schwarze 2017:
2018: i = 1;
1.175 schwarze 2019: SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2020: bits = NAME_FILE & NAME_MASK;
2021: if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2022: bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2023: sqlite3_reset(stmts[STMT_SELECT_NAME]);
2024: }
2025:
2026: i = 1;
2027: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
1.160 schwarze 2028: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2029: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2030: SQL_STEP(stmts[STMT_INSERT_NAME]);
2031: sqlite3_reset(stmts[STMT_INSERT_NAME]);
1.118 schwarze 2032: }
2033:
1.14 schwarze 2034: /*
1.50 kristaps 2035: * Flush the current page's terms (and their bits) into the database.
2036: * Wrap the entire set of additions in a transaction to make sqlite be a
2037: * little faster.
1.96 schwarze 2038: * Also, handle escape sequences at the last possible moment.
1.14 schwarze 2039: */
2040: static void
1.176 schwarze 2041: dbadd(struct mpage *mpage)
1.14 schwarze 2042: {
1.87 schwarze 2043: struct mlink *mlink;
1.50 kristaps 2044: struct str *key;
1.176 schwarze 2045: char *cp;
1.52 kristaps 2046: size_t i;
1.66 schwarze 2047: unsigned int slot;
1.176 schwarze 2048: int mustfree;
1.14 schwarze 2049:
1.128 schwarze 2050: mlink = mpage->mlinks;
1.43 kristaps 2051:
1.128 schwarze 2052: if (nodb) {
1.147 schwarze 2053: for (key = ohash_first(&names, &slot); NULL != key;
1.176 schwarze 2054: key = ohash_next(&names, &slot))
1.147 schwarze 2055: free(key);
2056: for (key = ohash_first(&strings, &slot); NULL != key;
1.176 schwarze 2057: key = ohash_next(&strings, &slot))
1.147 schwarze 2058: free(key);
1.145 schwarze 2059: if (0 == debug)
2060: return;
1.128 schwarze 2061: while (NULL != mlink) {
2062: fputs(mlink->name, stdout);
2063: if (NULL == mlink->next ||
2064: strcmp(mlink->dsec, mlink->next->dsec) ||
2065: strcmp(mlink->fsec, mlink->next->fsec) ||
2066: strcmp(mlink->arch, mlink->next->arch)) {
2067: putchar('(');
2068: if ('\0' == *mlink->dsec)
2069: fputs(mlink->fsec, stdout);
2070: else
2071: fputs(mlink->dsec, stdout);
2072: if ('\0' != *mlink->arch)
2073: printf("/%s", mlink->arch);
2074: putchar(')');
2075: }
2076: mlink = mlink->next;
2077: if (NULL != mlink)
2078: fputs(", ", stdout);
2079: }
1.132 schwarze 2080: printf(" - %s\n", mpage->desc);
1.14 schwarze 2081: return;
1.128 schwarze 2082: }
2083:
2084: if (debug)
2085: say(mlink->file, "Adding to database");
1.28 kristaps 2086:
1.176 schwarze 2087: cp = mpage->desc;
2088: i = strlen(cp);
2089: mustfree = render_string(&cp, &i);
1.52 kristaps 2090: i = 1;
1.176 schwarze 2091: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
1.161 schwarze 2092: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
1.81 schwarze 2093: SQL_STEP(stmts[STMT_INSERT_PAGE]);
1.137 schwarze 2094: mpage->pageid = sqlite3_last_insert_rowid(db);
1.81 schwarze 2095: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1.176 schwarze 2096: if (mustfree)
2097: free(cp);
1.81 schwarze 2098:
1.128 schwarze 2099: while (NULL != mlink) {
1.118 schwarze 2100: dbadd_mlink(mlink);
1.128 schwarze 2101: mlink = mlink->next;
2102: }
1.134 schwarze 2103: mlink = mpage->mlinks;
1.50 kristaps 2104:
1.133 schwarze 2105: for (key = ohash_first(&names, &slot); NULL != key;
2106: key = ohash_next(&names, &slot)) {
2107: assert(key->mpage == mpage);
2108: i = 1;
2109: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1.176 schwarze 2110: SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
1.137 schwarze 2111: SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
1.133 schwarze 2112: SQL_STEP(stmts[STMT_INSERT_NAME]);
2113: sqlite3_reset(stmts[STMT_INSERT_NAME]);
2114: free(key);
2115: }
1.66 schwarze 2116: for (key = ohash_first(&strings, &slot); NULL != key;
2117: key = ohash_next(&strings, &slot)) {
1.78 schwarze 2118: assert(key->mpage == mpage);
1.52 kristaps 2119: i = 1;
2120: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1.176 schwarze 2121: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
1.137 schwarze 2122: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
1.52 kristaps 2123: SQL_STEP(stmts[STMT_INSERT_KEY]);
1.50 kristaps 2124: sqlite3_reset(stmts[STMT_INSERT_KEY]);
1.66 schwarze 2125: free(key);
1.38 schwarze 2126: }
1.14 schwarze 2127: }
2128:
1.5 kristaps 2129: static void
1.59 schwarze 2130: dbprune(void)
1.5 kristaps 2131: {
1.78 schwarze 2132: struct mpage *mpage;
1.82 schwarze 2133: struct mlink *mlink;
1.52 kristaps 2134: size_t i;
1.80 schwarze 2135: unsigned int slot;
1.5 kristaps 2136:
1.106 schwarze 2137: if (0 == nodb)
2138: SQL_EXEC("BEGIN TRANSACTION");
1.12 schwarze 2139:
1.106 schwarze 2140: for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2141: mpage = ohash_next(&mpages, &slot)) {
1.82 schwarze 2142: mlink = mpage->mlinks;
1.125 schwarze 2143: if (debug)
1.106 schwarze 2144: say(mlink->file, "Deleting from database");
2145: if (nodb)
2146: continue;
2147: for ( ; NULL != mlink; mlink = mlink->next) {
2148: i = 1;
2149: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2150: i, mlink->dsec);
2151: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2152: i, mlink->arch);
2153: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2154: i, mlink->name);
2155: SQL_STEP(stmts[STMT_DELETE_PAGE]);
2156: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2157: }
1.5 kristaps 2158: }
1.106 schwarze 2159:
2160: if (0 == nodb)
2161: SQL_EXEC("END TRANSACTION");
1.5 kristaps 2162: }
2163:
1.4 kristaps 2164: /*
1.50 kristaps 2165: * Close an existing database and its prepared statements.
2166: * If "real" is not set, rename the temporary file into the real one.
1.4 kristaps 2167: */
1.35 kristaps 2168: static void
1.59 schwarze 2169: dbclose(int real)
1.4 kristaps 2170: {
1.50 kristaps 2171: size_t i;
1.115 schwarze 2172: int status;
2173: pid_t child;
1.4 kristaps 2174:
1.50 kristaps 2175: if (nodb)
1.38 schwarze 2176: return;
1.50 kristaps 2177:
2178: for (i = 0; i < STMT__MAX; i++) {
2179: sqlite3_finalize(stmts[i]);
2180: stmts[i] = NULL;
1.4 kristaps 2181: }
2182:
1.50 kristaps 2183: sqlite3_close(db);
2184: db = NULL;
1.12 schwarze 2185:
1.50 kristaps 2186: if (real)
2187: return;
1.12 schwarze 2188:
1.115 schwarze 2189: if ('\0' == *tempfilename) {
2190: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2191: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2192: say(MANDOC_DB, "&rename");
1.115 schwarze 2193: }
2194: return;
2195: }
2196:
2197: switch (child = fork()) {
1.141 schwarze 2198: case -1:
1.115 schwarze 2199: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2200: say("", "&fork cmp");
1.115 schwarze 2201: return;
1.141 schwarze 2202: case 0:
1.115 schwarze 2203: execlp("cmp", "cmp", "-s",
1.196 schwarze 2204: tempfilename, MANDOC_DB, (char *)NULL);
1.123 schwarze 2205: say("", "&exec cmp");
1.115 schwarze 2206: exit(0);
2207: default:
2208: break;
2209: }
2210: if (-1 == waitpid(child, &status, 0)) {
2211: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2212: say("", "&wait cmp");
1.115 schwarze 2213: } else if (WIFSIGNALED(status)) {
2214: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2215: say("", "cmp died from signal %d", WTERMSIG(status));
1.115 schwarze 2216: } else if (WEXITSTATUS(status)) {
1.59 schwarze 2217: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2218: say(MANDOC_DB,
2219: "Data changed, but cannot replace database");
2220: }
2221:
2222: *strrchr(tempfilename, '/') = '\0';
2223: switch (child = fork()) {
1.141 schwarze 2224: case -1:
1.115 schwarze 2225: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2226: say("", "&fork rm");
1.115 schwarze 2227: return;
1.141 schwarze 2228: case 0:
1.196 schwarze 2229: execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
1.123 schwarze 2230: say("", "&exec rm");
1.115 schwarze 2231: exit((int)MANDOCLEVEL_SYSERR);
2232: default:
2233: break;
2234: }
2235: if (-1 == waitpid(child, &status, 0)) {
2236: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2237: say("", "&wait rm");
1.115 schwarze 2238: } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2239: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2240: say("", "%s: Cannot remove temporary directory",
2241: tempfilename);
1.59 schwarze 2242: }
1.50 kristaps 2243: }
1.14 schwarze 2244:
1.50 kristaps 2245: /*
2246: * This is straightforward stuff.
2247: * Open a database connection to a "temporary" database, then open a set
2248: * of prepared statements we'll use over and over again.
2249: * If "real" is set, we use the existing database; if not, we truncate a
2250: * temporary one.
2251: * Must be matched by dbclose().
2252: */
2253: static int
1.59 schwarze 2254: dbopen(int real)
1.50 kristaps 2255: {
1.115 schwarze 2256: const char *sql;
1.50 kristaps 2257: int rc, ofl;
1.12 schwarze 2258:
1.141 schwarze 2259: if (nodb)
1.197 schwarze 2260: return 1;
1.12 schwarze 2261:
1.115 schwarze 2262: *tempfilename = '\0';
1.63 schwarze 2263: ofl = SQLITE_OPEN_READWRITE;
1.115 schwarze 2264:
2265: if (real) {
2266: rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2267: if (SQLITE_OK != rc) {
1.63 schwarze 2268: exitcode = (int)MANDOCLEVEL_SYSERR;
1.149 schwarze 2269: if (SQLITE_CANTOPEN != rc)
2270: say(MANDOC_DB, "%s", sqlite3_errstr(rc));
1.197 schwarze 2271: return 0;
1.63 schwarze 2272: }
1.115 schwarze 2273: goto prepare_statements;
2274: }
2275:
2276: ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
1.45 kristaps 2277:
1.115 schwarze 2278: remove(MANDOC_DB "~");
2279: rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
1.141 schwarze 2280: if (SQLITE_OK == rc)
1.115 schwarze 2281: goto create_tables;
1.116 schwarze 2282: if (MPARSE_QUICK & mparse_options) {
1.59 schwarze 2283: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2284: say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
1.197 schwarze 2285: return 0;
1.50 kristaps 2286: }
1.12 schwarze 2287:
1.143 schwarze 2288: (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2289: sizeof(tempfilename));
1.115 schwarze 2290: if (NULL == mkdtemp(tempfilename)) {
2291: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2292: say("", "&%s", tempfilename);
1.197 schwarze 2293: return 0;
1.115 schwarze 2294: }
1.143 schwarze 2295: (void)strlcat(tempfilename, "/" MANDOC_DB,
2296: sizeof(tempfilename));
1.115 schwarze 2297: rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2298: if (SQLITE_OK != rc) {
1.59 schwarze 2299: exitcode = (int)MANDOCLEVEL_SYSERR;
1.146 schwarze 2300: say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
1.197 schwarze 2301: return 0;
1.50 kristaps 2302: }
1.12 schwarze 2303:
1.115 schwarze 2304: create_tables:
1.81 schwarze 2305: sql = "CREATE TABLE \"mpages\" (\n"
1.132 schwarze 2306: " \"desc\" TEXT NOT NULL,\n"
1.50 kristaps 2307: " \"form\" INTEGER NOT NULL,\n"
1.137 schwarze 2308: " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
1.50 kristaps 2309: ");\n"
2310: "\n"
1.81 schwarze 2311: "CREATE TABLE \"mlinks\" (\n"
2312: " \"sec\" TEXT NOT NULL,\n"
2313: " \"arch\" TEXT NOT NULL,\n"
2314: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2315: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2316: "ON DELETE CASCADE\n"
1.81 schwarze 2317: ");\n"
1.136 schwarze 2318: "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
1.81 schwarze 2319: "\n"
1.133 schwarze 2320: "CREATE TABLE \"names\" (\n"
2321: " \"bits\" INTEGER NOT NULL,\n"
2322: " \"name\" TEXT NOT NULL,\n"
1.137 schwarze 2323: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.175 schwarze 2324: "ON DELETE CASCADE,\n"
2325: " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
1.133 schwarze 2326: ");\n"
2327: "\n"
1.50 kristaps 2328: "CREATE TABLE \"keys\" (\n"
2329: " \"bits\" INTEGER NOT NULL,\n"
2330: " \"key\" TEXT NOT NULL,\n"
1.137 schwarze 2331: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
1.109 schwarze 2332: "ON DELETE CASCADE\n"
1.136 schwarze 2333: ");\n"
2334: "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
1.14 schwarze 2335:
1.50 kristaps 2336: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
1.59 schwarze 2337: exitcode = (int)MANDOCLEVEL_SYSERR;
1.115 schwarze 2338: say(MANDOC_DB, "%s", sqlite3_errmsg(db));
1.146 schwarze 2339: sqlite3_close(db);
1.197 schwarze 2340: return 0;
1.50 kristaps 2341: }
1.4 kristaps 2342:
1.57 schwarze 2343: prepare_statements:
1.146 schwarze 2344: if (SQLITE_OK != sqlite3_exec(db,
2345: "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2346: exitcode = (int)MANDOCLEVEL_SYSERR;
2347: say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2348: sqlite3_errmsg(db));
2349: sqlite3_close(db);
1.197 schwarze 2350: return 0;
1.146 schwarze 2351: }
2352:
1.137 schwarze 2353: sql = "DELETE FROM mpages WHERE pageid IN "
1.106 schwarze 2354: "(SELECT pageid FROM mlinks WHERE "
2355: "sec=? AND arch=? AND name=?)";
1.81 schwarze 2356: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2357: sql = "INSERT INTO mpages "
1.132 schwarze 2358: "(desc,form) VALUES (?,?)";
1.81 schwarze 2359: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2360: sql = "INSERT INTO mlinks "
1.104 schwarze 2361: "(sec,arch,name,pageid) VALUES (?,?,?,?)";
1.81 schwarze 2362: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
1.175 schwarze 2363: sql = "SELECT bits FROM names where pageid = ?";
2364: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
1.133 schwarze 2365: sql = "INSERT INTO names "
2366: "(bits,name,pageid) VALUES (?,?,?)";
2367: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
1.50 kristaps 2368: sql = "INSERT INTO keys "
1.81 schwarze 2369: "(bits,key,pageid) VALUES (?,?,?)";
1.50 kristaps 2370: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.70 schwarze 2371:
2372: #ifndef __APPLE__
2373: /*
2374: * When opening a new database, we can turn off
2375: * synchronous mode for much better performance.
2376: */
2377:
1.146 schwarze 2378: if (real && SQLITE_OK != sqlite3_exec(db,
2379: "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2380: exitcode = (int)MANDOCLEVEL_SYSERR;
2381: say(MANDOC_DB, "PRAGMA synchronous: %s",
1.162 schwarze 2382: sqlite3_errmsg(db));
1.146 schwarze 2383: sqlite3_close(db);
1.197 schwarze 2384: return 0;
1.146 schwarze 2385: }
1.70 schwarze 2386: #endif
2387:
1.197 schwarze 2388: return 1;
1.50 kristaps 2389: }
1.5 kristaps 2390:
1.50 kristaps 2391: static void *
1.152 schwarze 2392: hash_calloc(size_t n, size_t sz, void *arg)
1.50 kristaps 2393: {
1.12 schwarze 2394:
1.197 schwarze 2395: return mandoc_calloc(n, sz);
1.50 kristaps 2396: }
1.12 schwarze 2397:
1.50 kristaps 2398: static void *
2399: hash_alloc(size_t sz, void *arg)
2400: {
1.5 kristaps 2401:
1.197 schwarze 2402: return mandoc_malloc(sz);
1.50 kristaps 2403: }
1.41 kristaps 2404:
1.50 kristaps 2405: static void
1.152 schwarze 2406: hash_free(void *p, void *arg)
1.50 kristaps 2407: {
1.4 kristaps 2408:
1.50 kristaps 2409: free(p);
1.4 kristaps 2410: }
2411:
1.50 kristaps 2412: static int
1.165 schwarze 2413: set_basedir(const char *targetdir, int report_baddir)
1.4 kristaps 2414: {
1.59 schwarze 2415: static char startdir[PATH_MAX];
1.151 schwarze 2416: static int getcwd_status; /* 1 = ok, 2 = failure */
2417: static int chdir_status; /* 1 = changed directory */
1.150 schwarze 2418: char *cp;
1.4 kristaps 2419:
1.59 schwarze 2420: /*
1.151 schwarze 2421: * Remember the original working directory, if possible.
2422: * This will be needed if the second or a later directory
2423: * on the command line is given as a relative path.
2424: * Do not error out if the current directory is not
2425: * searchable: Maybe it won't be needed after all.
2426: */
2427: if (0 == getcwd_status) {
2428: if (NULL == getcwd(startdir, sizeof(startdir))) {
2429: getcwd_status = 2;
2430: (void)strlcpy(startdir, strerror(errno),
2431: sizeof(startdir));
2432: } else
2433: getcwd_status = 1;
2434: }
2435:
2436: /*
2437: * We are leaving the old base directory.
2438: * Do not use it any longer, not even for messages.
2439: */
2440: *basedir = '\0';
2441:
2442: /*
2443: * If and only if the directory was changed earlier and
2444: * the next directory to process is given as a relative path,
2445: * first go back, or bail out if that is impossible.
1.59 schwarze 2446: */
1.151 schwarze 2447: if (chdir_status && '/' != *targetdir) {
2448: if (2 == getcwd_status) {
1.59 schwarze 2449: exitcode = (int)MANDOCLEVEL_SYSERR;
1.151 schwarze 2450: say("", "getcwd: %s", startdir);
1.197 schwarze 2451: return 0;
1.59 schwarze 2452: }
1.151 schwarze 2453: if (-1 == chdir(startdir)) {
1.59 schwarze 2454: exitcode = (int)MANDOCLEVEL_SYSERR;
1.123 schwarze 2455: say("", "&chdir %s", startdir);
1.197 schwarze 2456: return 0;
1.59 schwarze 2457: }
2458: }
1.151 schwarze 2459:
2460: /*
2461: * Always resolve basedir to the canonicalized absolute
2462: * pathname and append a trailing slash, such that
2463: * we can reliably check whether files are inside.
2464: */
1.59 schwarze 2465: if (NULL == realpath(targetdir, basedir)) {
1.165 schwarze 2466: if (report_baddir || errno != ENOENT) {
2467: exitcode = (int)MANDOCLEVEL_BADARG;
2468: say("", "&%s: realpath", targetdir);
2469: }
1.197 schwarze 2470: return 0;
1.59 schwarze 2471: } else if (-1 == chdir(basedir)) {
1.165 schwarze 2472: if (report_baddir || errno != ENOENT) {
2473: exitcode = (int)MANDOCLEVEL_BADARG;
2474: say("", "&chdir");
2475: }
1.197 schwarze 2476: return 0;
1.150 schwarze 2477: }
1.151 schwarze 2478: chdir_status = 1;
1.150 schwarze 2479: cp = strchr(basedir, '\0');
2480: if ('/' != cp[-1]) {
2481: if (cp - basedir >= PATH_MAX - 1) {
2482: exitcode = (int)MANDOCLEVEL_SYSERR;
2483: say("", "Filename too long");
1.197 schwarze 2484: return 0;
1.150 schwarze 2485: }
2486: *cp++ = '/';
2487: *cp = '\0';
1.4 kristaps 2488: }
1.197 schwarze 2489: return 1;
1.56 schwarze 2490: }
2491:
2492: static void
1.59 schwarze 2493: say(const char *file, const char *format, ...)
1.56 schwarze 2494: {
2495: va_list ap;
1.123 schwarze 2496: int use_errno;
1.56 schwarze 2497:
1.59 schwarze 2498: if ('\0' != *basedir)
2499: fprintf(stderr, "%s", basedir);
2500: if ('\0' != *basedir && '\0' != *file)
1.151 schwarze 2501: fputc('/', stderr);
1.56 schwarze 2502: if ('\0' != *file)
1.59 schwarze 2503: fprintf(stderr, "%s", file);
2504:
1.123 schwarze 2505: use_errno = 1;
2506: if (NULL != format) {
2507: switch (*format) {
1.141 schwarze 2508: case '&':
1.123 schwarze 2509: format++;
2510: break;
1.141 schwarze 2511: case '\0':
1.123 schwarze 2512: format = NULL;
2513: break;
2514: default:
2515: use_errno = 0;
2516: break;
2517: }
2518: }
2519: if (NULL != format) {
2520: if ('\0' != *basedir || '\0' != *file)
2521: fputs(": ", stderr);
2522: va_start(ap, format);
2523: vfprintf(stderr, format, ap);
2524: va_end(ap);
2525: }
2526: if (use_errno) {
2527: if ('\0' != *basedir || '\0' != *file || NULL != format)
2528: fputs(": ", stderr);
1.59 schwarze 2529: perror(NULL);
1.123 schwarze 2530: } else
2531: fputc('\n', stderr);
1.1 kristaps 2532: }
CVSweb