Annotation of src/usr.bin/mandoc/mandocdb.c, Revision 1.47
1.47 ! schwarze 1: /* $Id: mandocdb.c,v 1.46 2013/11/21 01:47:10 schwarze Exp $ */
1.1 schwarze 2: /*
1.47 ! schwarze 3: * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
! 4: * Copyright (c) 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
1.47 ! schwarze 18: #include <sys/stat.h>
1.1 schwarze 19:
20: #include <assert.h>
1.33 schwarze 21: #include <ctype.h>
1.34 schwarze 22: #include <errno.h>
1.1 schwarze 23: #include <fcntl.h>
1.47 ! schwarze 24: #include <fts.h>
1.1 schwarze 25: #include <getopt.h>
1.44 schwarze 26: #include <limits.h>
1.47 ! schwarze 27: #include <stddef.h>
1.1 schwarze 28: #include <stdio.h>
29: #include <stdint.h>
30: #include <stdlib.h>
31: #include <string.h>
1.14 schwarze 32: #include <unistd.h>
1.1 schwarze 33:
1.47 ! schwarze 34: #include <ohash.h>
! 35: #include <sqlite3.h>
! 36:
! 37: #include "mdoc.h"
1.1 schwarze 38: #include "man.h"
39: #include "mandoc.h"
1.10 schwarze 40: #include "manpath.h"
1.47 ! schwarze 41: #include "mansearch.h"
1.1 schwarze 42:
1.47 ! schwarze 43: #define SQL_EXEC(_v) \
! 44: if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
! 45: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
! 46: #define SQL_BIND_TEXT(_s, _i, _v) \
! 47: if (SQLITE_OK != sqlite3_bind_text \
! 48: ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
! 49: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
! 50: #define SQL_BIND_INT(_s, _i, _v) \
! 51: if (SQLITE_OK != sqlite3_bind_int \
! 52: ((_s), (_i)++, (_v))) \
! 53: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
! 54: #define SQL_BIND_INT64(_s, _i, _v) \
! 55: if (SQLITE_OK != sqlite3_bind_int64 \
! 56: ((_s), (_i)++, (_v))) \
! 57: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
! 58: #define SQL_STEP(_s) \
! 59: if (SQLITE_DONE != sqlite3_step((_s))) \
! 60: fprintf(stderr, "%s\n", sqlite3_errmsg(db))
1.1 schwarze 61:
1.47 ! schwarze 62: enum op {
! 63: OP_DEFAULT = 0, /* new dbs from dir list or default config */
! 64: OP_CONFFILE, /* new databases from custom config file */
! 65: OP_UPDATE, /* delete/add entries in existing database */
! 66: OP_DELETE, /* delete entries from existing database */
! 67: OP_TEST /* change no databases, report potential problems */
! 68: };
1.11 schwarze 69:
1.47 ! schwarze 70: enum form {
! 71: FORM_NONE, /* format is unknown */
! 72: FORM_SRC, /* format is -man or -mdoc */
! 73: FORM_CAT /* format is cat */
! 74: };
1.28 schwarze 75:
1.47 ! schwarze 76: struct str {
! 77: char *utf8; /* key in UTF-8 form */
! 78: const struct mpage *mpage; /* if set, the owning parse */
! 79: uint64_t mask; /* bitmask in sequence */
! 80: char key[]; /* the string itself */
1.28 schwarze 81: };
82:
1.47 ! schwarze 83: struct inodev {
! 84: ino_t st_ino;
! 85: dev_t st_dev;
! 86: };
1.28 schwarze 87:
1.47 ! schwarze 88: struct mpage {
! 89: struct inodev inodev; /* used for hashing routine */
! 90: enum form form; /* format from file content */
! 91: char *sec; /* section from file content */
! 92: char *arch; /* architecture from file content */
! 93: char *title; /* title from file content */
! 94: char *desc; /* description from file content */
! 95: struct mlink *mlinks; /* singly linked list */
1.28 schwarze 96: };
97:
1.47 ! schwarze 98: struct mlink {
! 99: char file[PATH_MAX]; /* filename rel. to manpath */
! 100: enum form dform; /* format from directory */
! 101: enum form fform; /* format from file name suffix */
! 102: char *dsec; /* section from directory */
! 103: char *arch; /* architecture from directory */
! 104: char *name; /* name from file name (not empty) */
! 105: char *fsec; /* section from file name suffix */
! 106: struct mlink *next; /* singly linked list */
1.2 schwarze 107: };
108:
1.47 ! schwarze 109: struct title {
! 110: char *title; /* name(sec/arch) given inside the file */
! 111: char *file; /* file name in case of mismatch */
1.1 schwarze 112: };
113:
1.47 ! schwarze 114: enum stmt {
! 115: STMT_DELETE_PAGE = 0, /* delete mpage */
! 116: STMT_INSERT_PAGE, /* insert mpage */
! 117: STMT_INSERT_LINK, /* insert mlink */
! 118: STMT_INSERT_KEY, /* insert parsed key */
! 119: STMT__MAX
1.1 schwarze 120: };
121:
1.47 ! schwarze 122: typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
1.1 schwarze 123:
1.19 schwarze 124: struct mdoc_handler {
1.47 ! schwarze 125: mdoc_fp fp; /* optional handler */
! 126: uint64_t mask; /* set unless handler returns 0 */
1.19 schwarze 127: };
128:
1.47 ! schwarze 129: static void dbclose(int);
! 130: static void dbindex(const struct mpage *, struct mchars *);
! 131: static int dbopen(int);
! 132: static void dbprune(void);
! 133: static void filescan(const char *);
! 134: static void *hash_alloc(size_t, void *);
! 135: static void hash_free(void *, size_t, void *);
! 136: static void *hash_halloc(size_t, void *);
! 137: static void mlink_add(struct mlink *, const struct stat *);
! 138: static void mlink_free(struct mlink *);
! 139: static void mlinks_undupe(struct mpage *);
! 140: static void mpages_free(void);
! 141: static void mpages_merge(struct mchars *, struct mparse *, int);
! 142: static void parse_cat(struct mpage *);
! 143: static void parse_man(struct mpage *, const struct man_node *);
! 144: static void parse_mdoc(struct mpage *, const struct mdoc_node *);
! 145: static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
! 146: static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
! 147: static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
! 148: static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
! 149: static int parse_mdoc_In(struct mpage *, const struct mdoc_node *);
! 150: static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
! 151: static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
! 152: static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
! 153: static int parse_mdoc_St(struct mpage *, const struct mdoc_node *);
! 154: static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
! 155: static void putkey(const struct mpage *,
! 156: const char *, uint64_t);
! 157: static void putkeys(const struct mpage *,
! 158: const char *, size_t, uint64_t);
! 159: static void putmdockey(const struct mpage *,
! 160: const struct mdoc_node *, uint64_t);
! 161: static void say(const char *, const char *, ...);
! 162: static int set_basedir(const char *);
! 163: static int treescan(void);
! 164: static size_t utf8(unsigned int, char [7]);
! 165: static void utf8key(struct mchars *, struct str *);
! 166:
! 167: static char *progname;
! 168: static int use_all; /* use all found files */
! 169: static int nodb; /* no database changes */
! 170: static int verb; /* print what we're doing */
! 171: static int warnings; /* warn about crap */
! 172: static int exitcode; /* to be returned by main */
! 173: static enum op op; /* operational mode */
! 174: static char basedir[PATH_MAX]; /* current base directory */
! 175: static struct ohash mpages; /* table of distinct manual pages */
! 176: static struct ohash mlinks; /* table of directory entries */
! 177: static struct ohash strings; /* table of all strings */
! 178: static sqlite3 *db = NULL; /* current database */
! 179: static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
! 180:
1.19 schwarze 181: static const struct mdoc_handler mdocs[MDOC_MAX] = {
1.47 ! schwarze 182: { NULL, 0 }, /* Ap */
! 183: { NULL, 0 }, /* Dd */
! 184: { NULL, 0 }, /* Dt */
! 185: { NULL, 0 }, /* Os */
! 186: { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
! 187: { parse_mdoc_head, TYPE_Ss }, /* Ss */
! 188: { NULL, 0 }, /* Pp */
! 189: { NULL, 0 }, /* D1 */
! 190: { NULL, 0 }, /* Dl */
! 191: { NULL, 0 }, /* Bd */
! 192: { NULL, 0 }, /* Ed */
! 193: { NULL, 0 }, /* Bl */
! 194: { NULL, 0 }, /* El */
! 195: { NULL, 0 }, /* It */
! 196: { NULL, 0 }, /* Ad */
! 197: { NULL, TYPE_An }, /* An */
! 198: { NULL, TYPE_Ar }, /* Ar */
! 199: { NULL, TYPE_Cd }, /* Cd */
! 200: { NULL, TYPE_Cm }, /* Cm */
! 201: { NULL, TYPE_Dv }, /* Dv */
! 202: { NULL, TYPE_Er }, /* Er */
! 203: { NULL, TYPE_Ev }, /* Ev */
! 204: { NULL, 0 }, /* Ex */
! 205: { NULL, TYPE_Fa }, /* Fa */
! 206: { parse_mdoc_Fd, 0 }, /* Fd */
! 207: { NULL, TYPE_Fl }, /* Fl */
! 208: { parse_mdoc_Fn, 0 }, /* Fn */
! 209: { NULL, TYPE_Ft }, /* Ft */
! 210: { NULL, TYPE_Ic }, /* Ic */
! 211: { parse_mdoc_In, TYPE_In }, /* In */
! 212: { NULL, TYPE_Li }, /* Li */
! 213: { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
! 214: { parse_mdoc_Nm, TYPE_Nm }, /* Nm */
! 215: { NULL, 0 }, /* Op */
! 216: { NULL, 0 }, /* Ot */
! 217: { NULL, TYPE_Pa }, /* Pa */
! 218: { NULL, 0 }, /* Rv */
! 219: { parse_mdoc_St, 0 }, /* St */
! 220: { NULL, TYPE_Va }, /* Va */
! 221: { parse_mdoc_body, TYPE_Va }, /* Vt */
! 222: { parse_mdoc_Xr, 0 }, /* Xr */
! 223: { NULL, 0 }, /* %A */
! 224: { NULL, 0 }, /* %B */
! 225: { NULL, 0 }, /* %D */
! 226: { NULL, 0 }, /* %I */
! 227: { NULL, 0 }, /* %J */
! 228: { NULL, 0 }, /* %N */
! 229: { NULL, 0 }, /* %O */
! 230: { NULL, 0 }, /* %P */
! 231: { NULL, 0 }, /* %R */
! 232: { NULL, 0 }, /* %T */
! 233: { NULL, 0 }, /* %V */
! 234: { NULL, 0 }, /* Ac */
! 235: { NULL, 0 }, /* Ao */
! 236: { NULL, 0 }, /* Aq */
! 237: { NULL, TYPE_At }, /* At */
! 238: { NULL, 0 }, /* Bc */
! 239: { NULL, 0 }, /* Bf */
! 240: { NULL, 0 }, /* Bo */
! 241: { NULL, 0 }, /* Bq */
! 242: { NULL, TYPE_Bsx }, /* Bsx */
! 243: { NULL, TYPE_Bx }, /* Bx */
! 244: { NULL, 0 }, /* Db */
! 245: { NULL, 0 }, /* Dc */
! 246: { NULL, 0 }, /* Do */
! 247: { NULL, 0 }, /* Dq */
! 248: { NULL, 0 }, /* Ec */
! 249: { NULL, 0 }, /* Ef */
! 250: { NULL, TYPE_Em }, /* Em */
! 251: { NULL, 0 }, /* Eo */
! 252: { NULL, TYPE_Fx }, /* Fx */
! 253: { NULL, TYPE_Ms }, /* Ms */
! 254: { NULL, 0 }, /* No */
! 255: { NULL, 0 }, /* Ns */
! 256: { NULL, TYPE_Nx }, /* Nx */
! 257: { NULL, TYPE_Ox }, /* Ox */
! 258: { NULL, 0 }, /* Pc */
! 259: { NULL, 0 }, /* Pf */
! 260: { NULL, 0 }, /* Po */
! 261: { NULL, 0 }, /* Pq */
! 262: { NULL, 0 }, /* Qc */
! 263: { NULL, 0 }, /* Ql */
! 264: { NULL, 0 }, /* Qo */
! 265: { NULL, 0 }, /* Qq */
! 266: { NULL, 0 }, /* Re */
! 267: { NULL, 0 }, /* Rs */
! 268: { NULL, 0 }, /* Sc */
! 269: { NULL, 0 }, /* So */
! 270: { NULL, 0 }, /* Sq */
! 271: { NULL, 0 }, /* Sm */
! 272: { NULL, 0 }, /* Sx */
! 273: { NULL, TYPE_Sy }, /* Sy */
! 274: { NULL, TYPE_Tn }, /* Tn */
! 275: { NULL, 0 }, /* Ux */
! 276: { NULL, 0 }, /* Xc */
! 277: { NULL, 0 }, /* Xo */
! 278: { parse_mdoc_head, 0 }, /* Fo */
! 279: { NULL, 0 }, /* Fc */
! 280: { NULL, 0 }, /* Oo */
! 281: { NULL, 0 }, /* Oc */
! 282: { NULL, 0 }, /* Bk */
! 283: { NULL, 0 }, /* Ek */
! 284: { NULL, 0 }, /* Bt */
! 285: { NULL, 0 }, /* Hf */
! 286: { NULL, 0 }, /* Fr */
! 287: { NULL, 0 }, /* Ud */
! 288: { NULL, TYPE_Lb }, /* Lb */
! 289: { NULL, 0 }, /* Lp */
! 290: { NULL, TYPE_Lk }, /* Lk */
! 291: { NULL, TYPE_Mt }, /* Mt */
! 292: { NULL, 0 }, /* Brq */
! 293: { NULL, 0 }, /* Bro */
! 294: { NULL, 0 }, /* Brc */
! 295: { NULL, 0 }, /* %C */
! 296: { NULL, 0 }, /* Es */
! 297: { NULL, 0 }, /* En */
! 298: { NULL, TYPE_Dx }, /* Dx */
! 299: { NULL, 0 }, /* %Q */
! 300: { NULL, 0 }, /* br */
! 301: { NULL, 0 }, /* sp */
! 302: { NULL, 0 }, /* %U */
! 303: { NULL, 0 }, /* Ta */
1.1 schwarze 304: };
305:
306: int
1.3 schwarze 307: mandocdb(int argc, char *argv[])
1.1 schwarze 308: {
1.47 ! schwarze 309: int ch, i;
! 310: size_t j, sz;
! 311: const char *path_arg;
! 312: struct mchars *mc;
! 313: struct manpaths dirs;
! 314: struct mparse *mp;
! 315: struct ohash_info mpages_info, mlinks_info;
! 316:
! 317: memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
! 318: memset(&dirs, 0, sizeof(struct manpaths));
! 319:
! 320: mpages_info.alloc = mlinks_info.alloc = hash_alloc;
! 321: mpages_info.halloc = mlinks_info.halloc = hash_halloc;
! 322: mpages_info.hfree = mlinks_info.hfree = hash_free;
! 323:
! 324: mpages_info.key_offset = offsetof(struct mpage, inodev);
! 325: mlinks_info.key_offset = offsetof(struct mlink, file);
1.1 schwarze 326:
327: progname = strrchr(argv[0], '/');
328: if (progname == NULL)
329: progname = argv[0];
330: else
331: ++progname;
332:
1.47 ! schwarze 333: /*
! 334: * We accept a few different invocations.
! 335: * The CHECKOP macro makes sure that invocation styles don't
! 336: * clobber each other.
! 337: */
! 338: #define CHECKOP(_op, _ch) do \
! 339: if (OP_DEFAULT != (_op)) { \
! 340: fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
! 341: goto usage; \
! 342: } while (/*CONSTCOND*/0)
1.10 schwarze 343:
1.47 ! schwarze 344: path_arg = NULL;
1.28 schwarze 345: op = OP_DEFAULT;
1.1 schwarze 346:
1.47 ! schwarze 347: while (-1 != (ch = getopt(argc, argv, "aC:d:ntu:vW")))
1.1 schwarze 348: switch (ch) {
1.6 schwarze 349: case ('a'):
350: use_all = 1;
351: break;
1.25 schwarze 352: case ('C'):
1.47 ! schwarze 353: CHECKOP(op, ch);
! 354: path_arg = optarg;
1.28 schwarze 355: op = OP_CONFFILE;
1.25 schwarze 356: break;
1.1 schwarze 357: case ('d'):
1.47 ! schwarze 358: CHECKOP(op, ch);
! 359: path_arg = optarg;
1.2 schwarze 360: op = OP_UPDATE;
1.1 schwarze 361: break;
1.47 ! schwarze 362: case ('n'):
! 363: nodb = 1;
! 364: break;
1.28 schwarze 365: case ('t'):
1.47 ! schwarze 366: CHECKOP(op, ch);
1.28 schwarze 367: dup2(STDOUT_FILENO, STDERR_FILENO);
368: op = OP_TEST;
1.47 ! schwarze 369: nodb = warnings = 1;
1.28 schwarze 370: break;
1.2 schwarze 371: case ('u'):
1.47 ! schwarze 372: CHECKOP(op, ch);
! 373: path_arg = optarg;
1.1 schwarze 374: op = OP_DELETE;
375: break;
376: case ('v'):
377: verb++;
378: break;
1.28 schwarze 379: case ('W'):
380: warnings = 1;
381: break;
1.1 schwarze 382: default:
1.28 schwarze 383: goto usage;
1.1 schwarze 384: }
385:
386: argc -= optind;
387: argv += optind;
388:
1.28 schwarze 389: if (OP_CONFFILE == op && argc > 0) {
1.47 ! schwarze 390: fprintf(stderr, "-C: Too many arguments\n");
1.28 schwarze 391: goto usage;
392: }
393:
1.47 ! schwarze 394: exitcode = (int)MANDOCLEVEL_OK;
! 395: mp = mparse_alloc(MPARSE_AUTO,
! 396: MANDOCLEVEL_FATAL, NULL, NULL, NULL);
! 397: mc = mchars_alloc();
1.2 schwarze 398:
1.47 ! schwarze 399: ohash_init(&mpages, 6, &mpages_info);
! 400: ohash_init(&mlinks, 6, &mlinks_info);
1.2 schwarze 401:
1.47 ! schwarze 402: if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
! 403: /*
! 404: * Force processing all files.
! 405: */
! 406: use_all = 1;
1.2 schwarze 407:
1.47 ! schwarze 408: /*
! 409: * All of these deal with a specific directory.
! 410: * Jump into that directory then collect files specified
! 411: * on the command-line.
! 412: */
! 413: if (0 == set_basedir(path_arg))
! 414: goto out;
! 415: for (i = 0; i < argc; i++)
! 416: filescan(argv[i]);
! 417: if (0 == dbopen(1))
! 418: goto out;
! 419: if (OP_TEST != op)
! 420: dbprune();
! 421: if (OP_DELETE != op)
! 422: mpages_merge(mc, mp, 0);
! 423: dbclose(1);
! 424: } else {
! 425: /*
! 426: * If we have arguments, use them as our manpaths.
! 427: * If we don't, grok from manpath(1) or however else
! 428: * manpath_parse() wants to do it.
! 429: */
! 430: if (argc > 0) {
! 431: dirs.paths = mandoc_calloc
! 432: (argc, sizeof(char *));
! 433: dirs.sz = (size_t)argc;
! 434: for (i = 0; i < argc; i++)
! 435: dirs.paths[i] = mandoc_strdup(argv[i]);
! 436: } else
! 437: manpath_parse(&dirs, path_arg, NULL, NULL);
1.2 schwarze 438:
1.47 ! schwarze 439: /*
! 440: * First scan the tree rooted at a base directory, then
! 441: * build a new database and finally move it into place.
! 442: * Ignore zero-length directories and strip trailing
! 443: * slashes.
! 444: */
! 445: for (j = 0; j < dirs.sz; j++) {
! 446: sz = strlen(dirs.paths[j]);
! 447: if (sz && '/' == dirs.paths[j][sz - 1])
! 448: dirs.paths[j][--sz] = '\0';
! 449: if (0 == sz)
! 450: continue;
1.2 schwarze 451:
1.47 ! schwarze 452: if (j) {
! 453: ohash_init(&mpages, 6, &mpages_info);
! 454: ohash_init(&mlinks, 6, &mlinks_info);
! 455: }
1.1 schwarze 456:
1.47 ! schwarze 457: if (0 == set_basedir(dirs.paths[j]))
! 458: goto out;
! 459: if (0 == treescan())
! 460: goto out;
! 461: if (0 == set_basedir(dirs.paths[j]))
! 462: goto out;
! 463: if (0 == dbopen(0))
! 464: goto out;
1.41 deraadt 465:
1.47 ! schwarze 466: mpages_merge(mc, mp, warnings && !use_all);
! 467: dbclose(0);
1.28 schwarze 468:
1.47 ! schwarze 469: if (j + 1 < dirs.sz) {
! 470: mpages_free();
! 471: ohash_delete(&mpages);
! 472: ohash_delete(&mlinks);
! 473: }
1.2 schwarze 474: }
1.47 ! schwarze 475: }
! 476: out:
! 477: set_basedir(NULL);
! 478: manpath_free(&dirs);
! 479: mchars_free(mc);
! 480: mparse_free(mp);
! 481: mpages_free();
! 482: ohash_delete(&mpages);
! 483: ohash_delete(&mlinks);
! 484: return(exitcode);
! 485: usage:
! 486: fprintf(stderr, "usage: %s [-anvW] [-C file]\n"
! 487: " %s [-anvW] dir ...\n"
! 488: " %s [-nvW] -d dir [file ...]\n"
! 489: " %s [-nvW] -u dir [file ...]\n"
! 490: " %s -t file ...\n",
! 491: progname, progname, progname,
! 492: progname, progname);
1.1 schwarze 493:
1.47 ! schwarze 494: return((int)MANDOCLEVEL_BADARG);
! 495: }
1.1 schwarze 496:
1.47 ! schwarze 497: /*
! 498: * Scan a directory tree rooted at "basedir" for manpages.
! 499: * We use fts(), scanning directory parts along the way for clues to our
! 500: * section and architecture.
! 501: *
! 502: * If use_all has been specified, grok all files.
! 503: * If not, sanitise paths to the following:
! 504: *
! 505: * [./]man*[/<arch>]/<name>.<section>
! 506: * or
! 507: * [./]cat<section>[/<arch>]/<name>.0
! 508: *
! 509: * TODO: accomodate for multi-language directories.
! 510: */
! 511: static int
! 512: treescan(void)
! 513: {
! 514: FTS *f;
! 515: FTSENT *ff;
! 516: struct mlink *mlink;
! 517: int dform;
! 518: char *fsec;
! 519: const char *dsec, *arch, *cp, *path;
! 520: const char *argv[2];
1.1 schwarze 521:
1.47 ! schwarze 522: argv[0] = ".";
! 523: argv[1] = (char *)NULL;
1.1 schwarze 524:
1.47 ! schwarze 525: /*
! 526: * Walk through all components under the directory, using the
! 527: * logical descent of files.
! 528: */
! 529: f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
! 530: if (NULL == f) {
! 531: exitcode = (int)MANDOCLEVEL_SYSERR;
! 532: say("", NULL);
! 533: return(0);
! 534: }
1.2 schwarze 535:
1.47 ! schwarze 536: dsec = arch = NULL;
! 537: dform = FORM_NONE;
1.2 schwarze 538:
1.47 ! schwarze 539: while (NULL != (ff = fts_read(f))) {
! 540: path = ff->fts_path + 2;
1.15 schwarze 541: /*
1.47 ! schwarze 542: * If we're a regular file, add an mlink by using the
! 543: * stored directory data and handling the filename.
1.15 schwarze 544: */
1.47 ! schwarze 545: if (FTS_F == ff->fts_info) {
! 546: if (0 == strcmp(path, MANDOC_DB))
! 547: continue;
! 548: if ( ! use_all && ff->fts_level < 2) {
! 549: if (warnings)
! 550: say(path, "Extraneous file");
! 551: continue;
! 552: } else if (NULL == (fsec =
! 553: strrchr(ff->fts_name, '.'))) {
! 554: if ( ! use_all) {
! 555: if (warnings)
! 556: say(path,
! 557: "No filename suffix");
! 558: continue;
! 559: }
! 560: } else if (0 == strcmp(++fsec, "html")) {
! 561: if (warnings)
! 562: say(path, "Skip html");
! 563: continue;
! 564: } else if (0 == strcmp(fsec, "gz")) {
! 565: if (warnings)
! 566: say(path, "Skip gz");
! 567: continue;
! 568: } else if (0 == strcmp(fsec, "ps")) {
! 569: if (warnings)
! 570: say(path, "Skip ps");
! 571: continue;
! 572: } else if (0 == strcmp(fsec, "pdf")) {
! 573: if (warnings)
! 574: say(path, "Skip pdf");
! 575: continue;
! 576: } else if ( ! use_all &&
! 577: ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
! 578: (FORM_CAT == dform && strcmp(fsec, "0")))) {
! 579: if (warnings)
! 580: say(path, "Wrong filename suffix");
! 581: continue;
! 582: } else
! 583: fsec[-1] = '\0';
! 584: mlink = mandoc_calloc(1, sizeof(struct mlink));
! 585: strlcpy(mlink->file, path, sizeof(mlink->file));
! 586: mlink->dform = dform;
! 587: if (NULL != dsec)
! 588: mlink->dsec = mandoc_strdup(dsec);
! 589: if (NULL != arch)
! 590: mlink->arch = mandoc_strdup(arch);
! 591: mlink->name = mandoc_strdup(ff->fts_name);
! 592: if (NULL != fsec)
! 593: mlink->fsec = mandoc_strdup(fsec);
! 594: mlink_add(mlink, ff->fts_statp);
! 595: continue;
! 596: } else if (FTS_D != ff->fts_info &&
! 597: FTS_DP != ff->fts_info) {
! 598: if (warnings)
! 599: say(path, "Not a regular file");
! 600: continue;
! 601: }
! 602:
! 603: switch (ff->fts_level) {
! 604: case (0):
! 605: /* Ignore the root directory. */
! 606: break;
! 607: case (1):
! 608: /*
! 609: * This might contain manX/ or catX/.
! 610: * Try to infer this from the name.
! 611: * If we're not in use_all, enforce it.
! 612: */
! 613: dsec = NULL;
! 614: dform = FORM_NONE;
! 615: cp = ff->fts_name;
! 616: if (FTS_DP == ff->fts_info)
! 617: break;
1.15 schwarze 618:
1.47 ! schwarze 619: if (0 == strncmp(cp, "man", 3)) {
! 620: dform = FORM_SRC;
! 621: dsec = cp + 3;
! 622: } else if (0 == strncmp(cp, "cat", 3)) {
! 623: dform = FORM_CAT;
! 624: dsec = cp + 3;
1.26 schwarze 625: }
1.47 ! schwarze 626:
! 627: if (NULL != dsec || use_all)
! 628: break;
! 629:
! 630: if (warnings)
! 631: say(path, "Unknown directory part");
! 632: fts_set(f, ff, FTS_SKIP);
! 633: break;
! 634: case (2):
! 635: /*
! 636: * Possibly our architecture.
! 637: * If we're descending, keep tabs on it.
! 638: */
! 639: arch = NULL;
! 640: if (FTS_DP != ff->fts_info && NULL != dsec)
! 641: arch = ff->fts_name;
! 642: break;
! 643: default:
! 644: if (FTS_DP == ff->fts_info || use_all)
! 645: break;
! 646: if (warnings)
! 647: say(path, "Extraneous directory part");
! 648: fts_set(f, ff, FTS_SKIP);
! 649: break;
1.14 schwarze 650: }
1.47 ! schwarze 651: }
! 652:
! 653: fts_close(f);
! 654: return(1);
! 655: }
1.1 schwarze 656:
1.47 ! schwarze 657: /*
! 658: * Add a file to the mlinks table.
! 659: * Do not verify that it's a "valid" looking manpage (we'll do that
! 660: * later).
! 661: *
! 662: * Try to infer the manual section, architecture, and page name from the
! 663: * path, assuming it looks like
! 664: *
! 665: * [./]man*[/<arch>]/<name>.<section>
! 666: * or
! 667: * [./]cat<section>[/<arch>]/<name>.0
! 668: *
! 669: * See treescan() for the fts(3) version of this.
! 670: */
! 671: static void
! 672: filescan(const char *file)
! 673: {
! 674: char buf[PATH_MAX];
! 675: struct stat st;
! 676: struct mlink *mlink;
! 677: char *p, *start;
! 678:
! 679: assert(use_all);
! 680:
! 681: if (0 == strncmp(file, "./", 2))
! 682: file += 2;
! 683:
! 684: if (NULL == realpath(file, buf)) {
! 685: exitcode = (int)MANDOCLEVEL_BADARG;
! 686: say(file, NULL);
! 687: return;
! 688: } else if (OP_TEST != op && strstr(buf, basedir) != buf) {
! 689: exitcode = (int)MANDOCLEVEL_BADARG;
! 690: say("", "%s: outside base directory", buf);
! 691: return;
! 692: } else if (-1 == stat(buf, &st)) {
! 693: exitcode = (int)MANDOCLEVEL_BADARG;
! 694: say(file, NULL);
! 695: return;
! 696: } else if ( ! (S_IFREG & st.st_mode)) {
! 697: exitcode = (int)MANDOCLEVEL_BADARG;
! 698: say(file, "Not a regular file");
! 699: return;
1.1 schwarze 700: }
1.47 ! schwarze 701: start = buf + strlen(basedir);
! 702: mlink = mandoc_calloc(1, sizeof(struct mlink));
! 703: strlcpy(mlink->file, start, sizeof(mlink->file));
1.1 schwarze 704:
1.10 schwarze 705: /*
1.47 ! schwarze 706: * First try to guess our directory structure.
! 707: * If we find a separator, try to look for man* or cat*.
! 708: * If we find one of these and what's underneath is a directory,
! 709: * assume it's an architecture.
1.10 schwarze 710: */
1.47 ! schwarze 711: if (NULL != (p = strchr(start, '/'))) {
! 712: *p++ = '\0';
! 713: if (0 == strncmp(start, "man", 3)) {
! 714: mlink->dform = FORM_SRC;
! 715: mlink->dsec = mandoc_strdup(start + 3);
! 716: } else if (0 == strncmp(start, "cat", 3)) {
! 717: mlink->dform = FORM_CAT;
! 718: mlink->dsec = mandoc_strdup(start + 3);
! 719: }
1.10 schwarze 720:
1.47 ! schwarze 721: start = p;
! 722: if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
! 723: *p++ = '\0';
! 724: mlink->arch = mandoc_strdup(start);
! 725: start = p;
1.41 deraadt 726: }
1.47 ! schwarze 727: }
1.7 schwarze 728:
1.47 ! schwarze 729: /*
! 730: * Now check the file suffix.
! 731: * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
! 732: */
! 733: p = strrchr(start, '\0');
! 734: while (p-- > start && '/' != *p && '.' != *p)
! 735: /* Loop. */ ;
! 736:
! 737: if ('.' == *p) {
! 738: *p++ = '\0';
! 739: mlink->fsec = mandoc_strdup(p);
! 740: }
1.41 deraadt 741:
1.47 ! schwarze 742: /*
! 743: * Now try to parse the name.
! 744: * Use the filename portion of the path.
! 745: */
! 746: mlink->name = start;
! 747: if (NULL != (p = strrchr(start, '/'))) {
! 748: mlink->name = p + 1;
! 749: *p = '\0';
! 750: }
! 751: mlink->name = mandoc_strdup(mlink->name);
1.28 schwarze 752:
1.47 ! schwarze 753: mlink_add(mlink, &st);
! 754: }
1.2 schwarze 755:
1.47 ! schwarze 756: static void
! 757: mlink_add(struct mlink *mlink, const struct stat *st)
! 758: {
! 759: struct inodev inodev;
! 760: struct mpage *mpage;
! 761: unsigned int slot;
! 762:
! 763: assert(NULL != mlink->file);
! 764:
! 765: if (NULL == mlink->dsec)
! 766: mlink->dsec = mandoc_strdup("");
! 767: if (NULL == mlink->arch)
! 768: mlink->arch = mandoc_strdup("");
! 769: if (NULL == mlink->name)
! 770: mlink->name = mandoc_strdup("");
! 771: if (NULL == mlink->fsec)
! 772: mlink->fsec = mandoc_strdup("");
! 773:
! 774: if ('0' == *mlink->fsec) {
! 775: free(mlink->fsec);
! 776: mlink->fsec = mandoc_strdup(mlink->dsec);
! 777: mlink->fform = FORM_CAT;
! 778: } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
! 779: mlink->fform = FORM_SRC;
! 780: else
! 781: mlink->fform = FORM_NONE;
1.1 schwarze 782:
1.47 ! schwarze 783: slot = ohash_qlookup(&mlinks, mlink->file);
! 784: assert(NULL == ohash_find(&mlinks, slot));
! 785: ohash_insert(&mlinks, slot, mlink);
! 786:
! 787: inodev.st_ino = st->st_ino;
! 788: inodev.st_dev = st->st_dev;
! 789: slot = ohash_lookup_memory(&mpages, (char *)&inodev,
! 790: sizeof(struct inodev), inodev.st_ino);
! 791: mpage = ohash_find(&mpages, slot);
! 792: if (NULL == mpage) {
! 793: mpage = mandoc_calloc(1, sizeof(struct mpage));
! 794: mpage->inodev.st_ino = inodev.st_ino;
! 795: mpage->inodev.st_dev = inodev.st_dev;
! 796: ohash_insert(&mpages, slot, mpage);
! 797: } else
! 798: mlink->next = mpage->mlinks;
! 799: mpage->mlinks = mlink;
! 800: }
1.1 schwarze 801:
1.47 ! schwarze 802: static void
! 803: mlink_free(struct mlink *mlink)
! 804: {
1.1 schwarze 805:
1.47 ! schwarze 806: free(mlink->dsec);
! 807: free(mlink->arch);
! 808: free(mlink->name);
! 809: free(mlink->fsec);
! 810: free(mlink);
! 811: }
1.1 schwarze 812:
1.47 ! schwarze 813: static void
! 814: mpages_free(void)
! 815: {
! 816: struct mpage *mpage;
! 817: struct mlink *mlink;
! 818: unsigned int slot;
! 819:
! 820: mpage = ohash_first(&mpages, &slot);
! 821: while (NULL != mpage) {
! 822: while (NULL != (mlink = mpage->mlinks)) {
! 823: mpage->mlinks = mlink->next;
! 824: mlink_free(mlink);
! 825: }
! 826: free(mpage->sec);
! 827: free(mpage->arch);
! 828: free(mpage->title);
! 829: free(mpage->desc);
! 830: free(mpage);
! 831: mpage = ohash_next(&mpages, &slot);
! 832: }
! 833: }
1.1 schwarze 834:
1.47 ! schwarze 835: /*
! 836: * For each mlink to the mpage, check whether the path looks like
! 837: * it is formatted, and if it does, check whether a source manual
! 838: * exists by the same name, ignoring the suffix.
! 839: * If both conditions hold, drop the mlink.
! 840: */
! 841: static void
! 842: mlinks_undupe(struct mpage *mpage)
! 843: {
! 844: char buf[PATH_MAX];
! 845: struct mlink **prev;
! 846: struct mlink *mlink;
! 847: char *bufp;
! 848:
! 849: mpage->form = FORM_CAT;
! 850: prev = &mpage->mlinks;
! 851: while (NULL != (mlink = *prev)) {
! 852: if (FORM_CAT != mlink->dform) {
! 853: mpage->form = FORM_NONE;
! 854: goto nextlink;
! 855: }
! 856: if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
! 857: if (warnings)
! 858: say(mlink->file, "Filename too long");
! 859: goto nextlink;
1.26 schwarze 860: }
1.47 ! schwarze 861: bufp = strstr(buf, "cat");
! 862: assert(NULL != bufp);
! 863: memcpy(bufp, "man", 3);
! 864: if (NULL != (bufp = strrchr(buf, '.')))
! 865: *++bufp = '\0';
! 866: strlcat(buf, mlink->dsec, PATH_MAX);
! 867: if (NULL == ohash_find(&mlinks,
! 868: ohash_qlookup(&mlinks, buf)))
! 869: goto nextlink;
! 870: if (warnings)
! 871: say(mlink->file, "Man source exists: %s", buf);
! 872: if (use_all)
! 873: goto nextlink;
! 874: *prev = mlink->next;
! 875: mlink_free(mlink);
! 876: continue;
! 877: nextlink:
! 878: prev = &(*prev)->next;
1.1 schwarze 879: }
1.47 ! schwarze 880: }
1.1 schwarze 881:
1.47 ! schwarze 882: /*
! 883: * Run through the files in the global vector "mpages"
! 884: * and add them to the database specified in "basedir".
! 885: *
! 886: * This handles the parsing scheme itself, using the cues of directory
! 887: * and filename to determine whether the file is parsable or not.
! 888: */
! 889: static void
! 890: mpages_merge(struct mchars *mc, struct mparse *mp, int check_reachable)
! 891: {
! 892: struct ohash title_table;
! 893: struct ohash_info title_info, str_info;
! 894: struct mpage *mpage;
! 895: struct mdoc *mdoc;
! 896: struct man *man;
! 897: struct title *title_entry;
! 898: char *title_str;
! 899: const char *cp;
! 900: int match;
! 901: unsigned int pslot, tslot;
! 902: enum mandoclevel lvl;
! 903:
! 904: str_info.alloc = hash_alloc;
! 905: str_info.halloc = hash_halloc;
! 906: str_info.hfree = hash_free;
! 907: str_info.key_offset = offsetof(struct str, key);
! 908:
! 909: if (check_reachable) {
! 910: title_info.alloc = hash_alloc;
! 911: title_info.halloc = hash_halloc;
! 912: title_info.hfree = hash_free;
! 913: title_info.key_offset = offsetof(struct title, title);
! 914: ohash_init(&title_table, 6, &title_info);
! 915: }
! 916:
! 917: mpage = ohash_first(&mpages, &pslot);
! 918: while (NULL != mpage) {
! 919: mlinks_undupe(mpage);
! 920: if (NULL == mpage->mlinks) {
! 921: mpage = ohash_next(&mpages, &pslot);
! 922: continue;
! 923: }
1.1 schwarze 924:
1.47 ! schwarze 925: ohash_init(&strings, 6, &str_info);
! 926: mparse_reset(mp);
! 927: mdoc = NULL;
! 928: man = NULL;
! 929: match = 1;
1.11 schwarze 930:
931: /*
1.24 schwarze 932: * Try interpreting the file as mdoc(7) or man(7)
933: * source code, unless it is already known to be
934: * formatted. Fall back to formatted mode.
1.11 schwarze 935: */
1.47 ! schwarze 936: if (FORM_CAT != mpage->mlinks->dform ||
! 937: FORM_CAT != mpage->mlinks->fform) {
! 938: lvl = mparse_readfd(mp, -1, mpage->mlinks->file);
! 939: if (lvl < MANDOCLEVEL_FATAL)
! 940: mparse_result(mp, &mdoc, &man);
! 941: }
1.11 schwarze 942:
943: if (NULL != mdoc) {
1.47 ! schwarze 944: mpage->form = FORM_SRC;
! 945: mpage->sec =
! 946: mandoc_strdup(mdoc_meta(mdoc)->msec);
! 947: mpage->arch = mdoc_meta(mdoc)->arch;
! 948: mpage->arch = mandoc_strdup(
! 949: NULL == mpage->arch ? "" : mpage->arch);
! 950: mpage->title =
! 951: mandoc_strdup(mdoc_meta(mdoc)->title);
1.11 schwarze 952: } else if (NULL != man) {
1.47 ! schwarze 953: mpage->form = FORM_SRC;
! 954: mpage->sec =
! 955: mandoc_strdup(man_meta(man)->msec);
! 956: mpage->arch =
! 957: mandoc_strdup(mpage->mlinks->arch);
! 958: mpage->title =
! 959: mandoc_strdup(man_meta(man)->title);
1.11 schwarze 960: } else {
1.47 ! schwarze 961: mpage->form = FORM_CAT;
! 962: mpage->sec =
! 963: mandoc_strdup(mpage->mlinks->dsec);
! 964: mpage->arch =
! 965: mandoc_strdup(mpage->mlinks->arch);
! 966: mpage->title =
! 967: mandoc_strdup(mpage->mlinks->name);
1.1 schwarze 968: }
969:
1.6 schwarze 970: /*
1.38 schwarze 971: * Check whether the manual section given in a file
972: * agrees with the directory where the file is located.
973: * Some manuals have suffixes like (3p) on their
974: * section number either inside the file or in the
975: * directory name, some are linked into more than one
976: * section, like encrypt(1) = makekey(8). Do not skip
977: * manuals for such reasons.
1.6 schwarze 978: */
1.47 ! schwarze 979: if (warnings && !use_all && FORM_SRC == mpage->form &&
! 980: strcasecmp(mpage->sec, mpage->mlinks->dsec)) {
! 981: match = 0;
! 982: say(mpage->mlinks->file, "Section \"%s\" "
! 983: "manual in %s directory",
! 984: mpage->sec, mpage->mlinks->dsec);
! 985: }
1.41 deraadt 986:
1.32 schwarze 987: /*
988: * Manual page directories exist for each kernel
989: * architecture as returned by machine(1).
990: * However, many manuals only depend on the
991: * application architecture as returned by arch(1).
992: * For example, some (2/ARM) manuals are shared
993: * across the "armish" and "zaurus" kernel
994: * architectures.
995: * A few manuals are even shared across completely
996: * different architectures, for example fdformat(1)
997: * on amd64, i386, sparc, and sparc64.
998: * Thus, warn about architecture mismatches,
999: * but don't skip manuals for this reason.
1000: */
1.47 ! schwarze 1001: if (warnings && !use_all &&
! 1002: strcasecmp(mpage->arch, mpage->mlinks->arch)) {
! 1003: match = 0;
! 1004: say(mpage->mlinks->file, "Architecture \"%s\" "
! 1005: "manual in \"%s\" directory",
! 1006: mpage->arch, mpage->mlinks->arch);
! 1007: }
! 1008: if (warnings && !use_all &&
! 1009: strcasecmp(mpage->title, mpage->mlinks->name))
! 1010: match = 0;
1.32 schwarze 1011:
1.47 ! schwarze 1012: putkey(mpage, mpage->mlinks->name, TYPE_Nm);
1.6 schwarze 1013:
1.47 ! schwarze 1014: if (NULL != mdoc) {
! 1015: if (NULL != (cp = mdoc_meta(mdoc)->name))
! 1016: putkey(mpage, cp, TYPE_Nm);
! 1017: assert(NULL == mpage->desc);
! 1018: parse_mdoc(mpage, mdoc_node(mdoc));
! 1019: putkey(mpage, NULL != mpage->desc ?
! 1020: mpage->desc : mpage->mlinks->name, TYPE_Nd);
! 1021: } else if (NULL != man)
! 1022: parse_man(mpage, man_node(man));
! 1023: else
! 1024: parse_cat(mpage);
1.6 schwarze 1025:
1.37 schwarze 1026: /*
1027: * Build a title string for the file. If it matches
1028: * the location of the file, remember the title as
1029: * found; else, remember it as missing.
1030: */
1031:
1.47 ! schwarze 1032: if (check_reachable) {
! 1033: if (-1 == asprintf(&title_str, "%s(%s%s%s)",
! 1034: mpage->title, mpage->sec,
! 1035: '\0' == *mpage->arch ? "" : "/",
! 1036: mpage->arch)) {
! 1037: perror(NULL);
! 1038: exit((int)MANDOCLEVEL_SYSERR);
1.37 schwarze 1039: }
1.47 ! schwarze 1040: tslot = ohash_qlookup(&title_table, title_str);
! 1041: title_entry = ohash_find(&title_table, tslot);
! 1042: if (NULL == title_entry) {
! 1043: title_entry = mandoc_malloc(
! 1044: sizeof(struct title));
! 1045: title_entry->title = title_str;
! 1046: title_entry->file = mandoc_strdup(
! 1047: match ? "" : mpage->mlinks->file);
! 1048: ohash_insert(&title_table, tslot,
! 1049: title_entry);
! 1050: } else {
! 1051: if (match)
! 1052: *title_entry->file = '\0';
! 1053: free(title_str);
1.37 schwarze 1054: }
1055: }
1056:
1.47 ! schwarze 1057: dbindex(mpage, mc);
! 1058: ohash_delete(&strings);
! 1059: mpage = ohash_next(&mpages, &pslot);
! 1060: }
! 1061:
! 1062: if (check_reachable) {
! 1063: title_entry = ohash_first(&title_table, &tslot);
! 1064: while (NULL != title_entry) {
! 1065: if ('\0' != *title_entry->file)
! 1066: say(title_entry->file,
! 1067: "Probably unreachable, title is %s",
! 1068: title_entry->title);
! 1069: free(title_entry->title);
! 1070: free(title_entry->file);
! 1071: free(title_entry);
! 1072: title_entry = ohash_next(&title_table, &tslot);
! 1073: }
! 1074: ohash_delete(&title_table);
! 1075: }
! 1076: }
1.6 schwarze 1077:
1.47 ! schwarze 1078: static void
! 1079: parse_cat(struct mpage *mpage)
! 1080: {
! 1081: FILE *stream;
! 1082: char *line, *p, *title;
! 1083: size_t len, plen, titlesz;
1.1 schwarze 1084:
1.47 ! schwarze 1085: if (NULL == (stream = fopen(mpage->mlinks->file, "r"))) {
! 1086: if (warnings)
! 1087: say(mpage->mlinks->file, NULL);
! 1088: return;
! 1089: }
1.1 schwarze 1090:
1.47 ! schwarze 1091: /* Skip to first blank line. */
1.1 schwarze 1092:
1.47 ! schwarze 1093: while (NULL != (line = fgetln(stream, &len)))
! 1094: if ('\n' == *line)
! 1095: break;
1.1 schwarze 1096:
1.47 ! schwarze 1097: /*
! 1098: * Assume the first line that is not indented
! 1099: * is the first section header. Skip to it.
! 1100: */
1.1 schwarze 1101:
1.47 ! schwarze 1102: while (NULL != (line = fgetln(stream, &len)))
! 1103: if ('\n' != *line && ' ' != *line)
! 1104: break;
! 1105:
! 1106: /*
! 1107: * Read up until the next section into a buffer.
! 1108: * Strip the leading and trailing newline from each read line,
! 1109: * appending a trailing space.
! 1110: * Ignore empty (whitespace-only) lines.
! 1111: */
1.28 schwarze 1112:
1.47 ! schwarze 1113: titlesz = 0;
! 1114: title = NULL;
1.38 schwarze 1115:
1.47 ! schwarze 1116: while (NULL != (line = fgetln(stream, &len))) {
! 1117: if (' ' != *line || '\n' != line[len - 1])
! 1118: break;
! 1119: while (len > 0 && isspace((unsigned char)*line)) {
! 1120: line++;
! 1121: len--;
! 1122: }
! 1123: if (1 == len)
! 1124: continue;
! 1125: title = mandoc_realloc(title, titlesz + len);
! 1126: memcpy(title + titlesz, line, len);
! 1127: titlesz += len;
! 1128: title[titlesz - 1] = ' ';
! 1129: }
1.28 schwarze 1130:
1.47 ! schwarze 1131: /*
! 1132: * If no page content can be found, or the input line
! 1133: * is already the next section header, or there is no
! 1134: * trailing newline, reuse the page title as the page
! 1135: * description.
! 1136: */
1.1 schwarze 1137:
1.47 ! schwarze 1138: if (NULL == title || '\0' == *title) {
! 1139: if (warnings)
! 1140: say(mpage->mlinks->file,
! 1141: "Cannot find NAME section");
! 1142: assert(NULL == mpage->desc);
! 1143: mpage->desc = mandoc_strdup(mpage->mlinks->name);
! 1144: putkey(mpage, mpage->mlinks->name, TYPE_Nd);
! 1145: fclose(stream);
! 1146: free(title);
! 1147: return;
! 1148: }
1.24 schwarze 1149:
1.47 ! schwarze 1150: title = mandoc_realloc(title, titlesz + 1);
! 1151: title[titlesz] = '\0';
1.24 schwarze 1152:
1.47 ! schwarze 1153: /*
! 1154: * Skip to the first dash.
! 1155: * Use the remaining line as the description (no more than 70
! 1156: * bytes).
! 1157: */
1.28 schwarze 1158:
1.47 ! schwarze 1159: if (NULL != (p = strstr(title, "- "))) {
! 1160: for (p += 2; ' ' == *p || '\b' == *p; p++)
! 1161: /* Skip to next word. */ ;
! 1162: } else {
! 1163: if (warnings)
! 1164: say(mpage->mlinks->file,
! 1165: "No dash in title line");
! 1166: p = title;
! 1167: }
1.1 schwarze 1168:
1.47 ! schwarze 1169: plen = strlen(p);
1.1 schwarze 1170:
1.47 ! schwarze 1171: /* Strip backspace-encoding from line. */
1.1 schwarze 1172:
1.47 ! schwarze 1173: while (NULL != (line = memchr(p, '\b', plen))) {
! 1174: len = line - p;
! 1175: if (0 == len) {
! 1176: memmove(line, line + 1, plen--);
! 1177: continue;
! 1178: }
! 1179: memmove(line - 1, line + 1, plen - len);
! 1180: plen -= 2;
! 1181: }
1.1 schwarze 1182:
1.47 ! schwarze 1183: assert(NULL == mpage->desc);
! 1184: mpage->desc = mandoc_strdup(p);
! 1185: putkey(mpage, mpage->desc, TYPE_Nd);
! 1186: fclose(stream);
! 1187: free(title);
! 1188: }
1.16 schwarze 1189:
1.47 ! schwarze 1190: /*
! 1191: * Put a type/word pair into the word database for this particular file.
! 1192: */
! 1193: static void
! 1194: putkey(const struct mpage *mpage, const char *value, uint64_t type)
! 1195: {
1.37 schwarze 1196:
1.47 ! schwarze 1197: assert(NULL != value);
! 1198: putkeys(mpage, value, strlen(value), type);
1.2 schwarze 1199: }
1200:
1201: /*
1.47 ! schwarze 1202: * Grok all nodes at or below a certain mdoc node into putkey().
1.2 schwarze 1203: */
1204: static void
1.47 ! schwarze 1205: putmdockey(const struct mpage *mpage,
! 1206: const struct mdoc_node *n, uint64_t m)
1.2 schwarze 1207: {
1.16 schwarze 1208:
1.47 ! schwarze 1209: for ( ; NULL != n; n = n->next) {
! 1210: if (NULL != n->child)
! 1211: putmdockey(mpage, n->child, m);
! 1212: if (MDOC_TEXT == n->type)
! 1213: putkey(mpage, n->string, m);
! 1214: }
! 1215: }
1.16 schwarze 1216:
1.47 ! schwarze 1217: static void
! 1218: parse_man(struct mpage *mpage, const struct man_node *n)
! 1219: {
! 1220: const struct man_node *head, *body;
! 1221: char *start, *sv, *title;
! 1222: char byte;
! 1223: size_t sz, titlesz;
1.16 schwarze 1224:
1.47 ! schwarze 1225: if (NULL == n)
! 1226: return;
1.16 schwarze 1227:
1.47 ! schwarze 1228: /*
! 1229: * We're only searching for one thing: the first text child in
! 1230: * the BODY of a NAME section. Since we don't keep track of
! 1231: * sections in -man, run some hoops to find out whether we're in
! 1232: * the correct section or not.
! 1233: */
1.16 schwarze 1234:
1.47 ! schwarze 1235: if (MAN_BODY == n->type && MAN_SH == n->tok) {
! 1236: body = n;
! 1237: assert(body->parent);
! 1238: if (NULL != (head = body->parent->head) &&
! 1239: 1 == head->nchild &&
! 1240: NULL != (head = (head->child)) &&
! 1241: MAN_TEXT == head->type &&
! 1242: 0 == strcmp(head->string, "NAME") &&
! 1243: NULL != (body = body->child) &&
! 1244: MAN_TEXT == body->type) {
1.2 schwarze 1245:
1.47 ! schwarze 1246: title = NULL;
! 1247: titlesz = 0;
1.2 schwarze 1248:
1.47 ! schwarze 1249: /*
! 1250: * Suck the entire NAME section into memory.
! 1251: * Yes, we might run away.
! 1252: * But too many manuals have big, spread-out
! 1253: * NAME sections over many lines.
! 1254: */
1.2 schwarze 1255:
1.47 ! schwarze 1256: for ( ; NULL != body; body = body->next) {
! 1257: if (MAN_TEXT != body->type)
! 1258: break;
! 1259: if (0 == (sz = strlen(body->string)))
! 1260: continue;
! 1261: title = mandoc_realloc
! 1262: (title, titlesz + sz + 1);
! 1263: memcpy(title + titlesz, body->string, sz);
! 1264: titlesz += sz + 1;
! 1265: title[titlesz - 1] = ' ';
! 1266: }
! 1267: if (NULL == title)
! 1268: return;
1.16 schwarze 1269:
1.47 ! schwarze 1270: title = mandoc_realloc(title, titlesz + 1);
! 1271: title[titlesz] = '\0';
1.16 schwarze 1272:
1.47 ! schwarze 1273: /* Skip leading space. */
1.16 schwarze 1274:
1.47 ! schwarze 1275: sv = title;
! 1276: while (isspace((unsigned char)*sv))
! 1277: sv++;
1.16 schwarze 1278:
1.47 ! schwarze 1279: if (0 == (sz = strlen(sv))) {
! 1280: free(title);
! 1281: return;
! 1282: }
1.1 schwarze 1283:
1.47 ! schwarze 1284: /* Erase trailing space. */
1.1 schwarze 1285:
1.47 ! schwarze 1286: start = &sv[sz - 1];
! 1287: while (start > sv && isspace((unsigned char)*start))
! 1288: *start-- = '\0';
1.1 schwarze 1289:
1.47 ! schwarze 1290: if (start == sv) {
! 1291: free(title);
! 1292: return;
! 1293: }
1.1 schwarze 1294:
1.47 ! schwarze 1295: start = sv;
1.16 schwarze 1296:
1.47 ! schwarze 1297: /*
! 1298: * Go through a special heuristic dance here.
! 1299: * Conventionally, one or more manual names are
! 1300: * comma-specified prior to a whitespace, then a
! 1301: * dash, then a description. Try to puzzle out
! 1302: * the name parts here.
! 1303: */
1.16 schwarze 1304:
1.47 ! schwarze 1305: for ( ;; ) {
! 1306: sz = strcspn(start, " ,");
! 1307: if ('\0' == start[sz])
! 1308: break;
1.1 schwarze 1309:
1.47 ! schwarze 1310: byte = start[sz];
! 1311: start[sz] = '\0';
1.1 schwarze 1312:
1.47 ! schwarze 1313: putkey(mpage, start, TYPE_Nm);
1.1 schwarze 1314:
1.47 ! schwarze 1315: if (' ' == byte) {
! 1316: start += sz + 1;
! 1317: break;
! 1318: }
1.1 schwarze 1319:
1.47 ! schwarze 1320: assert(',' == byte);
! 1321: start += sz + 1;
! 1322: while (' ' == *start)
! 1323: start++;
! 1324: }
1.1 schwarze 1325:
1.47 ! schwarze 1326: if (sv == start) {
! 1327: putkey(mpage, start, TYPE_Nm);
! 1328: free(title);
! 1329: return;
! 1330: }
1.1 schwarze 1331:
1.47 ! schwarze 1332: while (isspace((unsigned char)*start))
! 1333: start++;
1.1 schwarze 1334:
1.47 ! schwarze 1335: if (0 == strncmp(start, "-", 1))
! 1336: start += 1;
! 1337: else if (0 == strncmp(start, "\\-\\-", 4))
! 1338: start += 4;
! 1339: else if (0 == strncmp(start, "\\-", 2))
! 1340: start += 2;
! 1341: else if (0 == strncmp(start, "\\(en", 4))
! 1342: start += 4;
! 1343: else if (0 == strncmp(start, "\\(em", 4))
! 1344: start += 4;
1.1 schwarze 1345:
1.47 ! schwarze 1346: while (' ' == *start)
! 1347: start++;
1.1 schwarze 1348:
1.47 ! schwarze 1349: assert(NULL == mpage->desc);
! 1350: mpage->desc = mandoc_strdup(start);
! 1351: putkey(mpage, mpage->desc, TYPE_Nd);
! 1352: free(title);
! 1353: return;
! 1354: }
! 1355: }
1.1 schwarze 1356:
1.47 ! schwarze 1357: for (n = n->child; n; n = n->next) {
! 1358: if (NULL != mpage->desc)
! 1359: break;
! 1360: parse_man(mpage, n);
1.1 schwarze 1361: }
1362: }
1363:
1364: static void
1.47 ! schwarze 1365: parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1366: {
1367:
1.47 ! schwarze 1368: assert(NULL != n);
! 1369: for (n = n->child; NULL != n; n = n->next) {
! 1370: switch (n->type) {
! 1371: case (MDOC_ELEM):
! 1372: /* FALLTHROUGH */
! 1373: case (MDOC_BLOCK):
! 1374: /* FALLTHROUGH */
! 1375: case (MDOC_HEAD):
! 1376: /* FALLTHROUGH */
! 1377: case (MDOC_BODY):
! 1378: /* FALLTHROUGH */
! 1379: case (MDOC_TAIL):
! 1380: if (NULL != mdocs[n->tok].fp)
! 1381: if (0 == (*mdocs[n->tok].fp)(mpage, n))
! 1382: break;
! 1383: if (mdocs[n->tok].mask)
! 1384: putmdockey(mpage, n->child,
! 1385: mdocs[n->tok].mask);
! 1386: break;
! 1387: default:
! 1388: assert(MDOC_ROOT != n->type);
! 1389: continue;
! 1390: }
! 1391: if (NULL != n->child)
! 1392: parse_mdoc(mpage, n);
1.1 schwarze 1393: }
1394: }
1395:
1.19 schwarze 1396: static int
1.47 ! schwarze 1397: parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1398: {
1399: const char *start, *end;
1400: size_t sz;
1.19 schwarze 1401:
1.47 ! schwarze 1402: if (SEC_SYNOPSIS != n->sec ||
! 1403: NULL == (n = n->child) ||
! 1404: MDOC_TEXT != n->type)
1.19 schwarze 1405: return(0);
1.1 schwarze 1406:
1407: /*
1408: * Only consider those `Fd' macro fields that begin with an
1409: * "inclusion" token (versus, e.g., #define).
1410: */
1.47 ! schwarze 1411:
1.1 schwarze 1412: if (strcmp("#include", n->string))
1.19 schwarze 1413: return(0);
1.1 schwarze 1414:
1415: if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1.19 schwarze 1416: return(0);
1.1 schwarze 1417:
1418: /*
1419: * Strip away the enclosing angle brackets and make sure we're
1420: * not zero-length.
1421: */
1422:
1423: start = n->string;
1424: if ('<' == *start || '"' == *start)
1425: start++;
1426:
1427: if (0 == (sz = strlen(start)))
1.19 schwarze 1428: return(0);
1.1 schwarze 1429:
1430: end = &start[(int)sz - 1];
1431: if ('>' == *end || '"' == *end)
1432: end--;
1433:
1.47 ! schwarze 1434: if (end > start)
! 1435: putkeys(mpage, start, end - start + 1, TYPE_In);
1.19 schwarze 1436: return(1);
1.1 schwarze 1437: }
1438:
1.19 schwarze 1439: static int
1.47 ! schwarze 1440: parse_mdoc_In(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1441: {
1442:
1.47 ! schwarze 1443: if (NULL != n->child && MDOC_TEXT == n->child->type)
1.19 schwarze 1444: return(0);
1.1 schwarze 1445:
1.47 ! schwarze 1446: putkey(mpage, n->child->string, TYPE_In);
1.19 schwarze 1447: return(1);
1.1 schwarze 1448: }
1449:
1.19 schwarze 1450: static int
1.47 ! schwarze 1451: parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1452: {
1453: const char *cp;
1454:
1.47 ! schwarze 1455: if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1.19 schwarze 1456: return(0);
1457:
1.47 ! schwarze 1458: /*
! 1459: * Parse: .Fn "struct type *name" "char *arg".
! 1460: * First strip away pointer symbol.
! 1461: * Then store the function name, then type.
! 1462: * Finally, store the arguments.
! 1463: */
1.1 schwarze 1464:
1.47 ! schwarze 1465: if (NULL == (cp = strrchr(n->string, ' ')))
! 1466: cp = n->string;
1.1 schwarze 1467:
1468: while ('*' == *cp)
1469: cp++;
1470:
1.47 ! schwarze 1471: putkey(mpage, cp, TYPE_Fn);
1.19 schwarze 1472:
1.47 ! schwarze 1473: if (n->string < cp)
! 1474: putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1.19 schwarze 1475:
1.47 ! schwarze 1476: for (n = n->next; NULL != n; n = n->next)
! 1477: if (MDOC_TEXT == n->type)
! 1478: putkey(mpage, n->string, TYPE_Fa);
1.19 schwarze 1479:
1480: return(0);
1.1 schwarze 1481: }
1482:
1.19 schwarze 1483: static int
1.47 ! schwarze 1484: parse_mdoc_St(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1485: {
1.19 schwarze 1486:
1.1 schwarze 1487: if (NULL == n->child || MDOC_TEXT != n->child->type)
1.19 schwarze 1488: return(0);
1.1 schwarze 1489:
1.47 ! schwarze 1490: putkey(mpage, n->child->string, TYPE_St);
1.19 schwarze 1491: return(1);
1.1 schwarze 1492: }
1493:
1.19 schwarze 1494: static int
1.47 ! schwarze 1495: parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1496: {
1.47 ! schwarze 1497: char *cp;
1.1 schwarze 1498:
1499: if (NULL == (n = n->child))
1.19 schwarze 1500: return(0);
1.1 schwarze 1501:
1.47 ! schwarze 1502: if (NULL == n->next) {
! 1503: putkey(mpage, n->string, TYPE_Xr);
! 1504: return(0);
! 1505: }
1.1 schwarze 1506:
1.47 ! schwarze 1507: if (-1 == asprintf(&cp, "%s(%s)", n->string, n->next->string)) {
! 1508: perror(NULL);
! 1509: exit((int)MANDOCLEVEL_SYSERR);
! 1510: }
! 1511: putkey(mpage, cp, TYPE_Xr);
! 1512: free(cp);
! 1513: return(0);
1.1 schwarze 1514: }
1515:
1.19 schwarze 1516: static int
1.47 ! schwarze 1517: parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1518: {
1.47 ! schwarze 1519: size_t sz;
1.1 schwarze 1520:
1521: if (MDOC_BODY != n->type)
1.19 schwarze 1522: return(0);
1.1 schwarze 1523:
1.47 ! schwarze 1524: /*
! 1525: * Special-case the `Nd' because we need to put the description
! 1526: * into the document table.
! 1527: */
! 1528:
! 1529: for (n = n->child; NULL != n; n = n->next) {
! 1530: if (MDOC_TEXT == n->type) {
! 1531: if (NULL != mpage->desc) {
! 1532: sz = strlen(mpage->desc) +
! 1533: strlen(n->string) + 2;
! 1534: mpage->desc = mandoc_realloc(
! 1535: mpage->desc, sz);
! 1536: strlcat(mpage->desc, " ", sz);
! 1537: strlcat(mpage->desc, n->string, sz);
! 1538: } else
! 1539: mpage->desc = mandoc_strdup(n->string);
! 1540: }
! 1541: if (NULL != n->child)
! 1542: parse_mdoc_Nd(mpage, n);
! 1543: }
1.19 schwarze 1544: return(1);
1.1 schwarze 1545: }
1546:
1.19 schwarze 1547: static int
1.47 ! schwarze 1548: parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1549: {
1550:
1.19 schwarze 1551: if (SEC_NAME == n->sec)
1552: return(1);
1553: else if (SEC_SYNOPSIS != n->sec || MDOC_HEAD != n->type)
1554: return(0);
1.1 schwarze 1555:
1.19 schwarze 1556: return(1);
1.1 schwarze 1557: }
1558:
1.19 schwarze 1559: static int
1.47 ! schwarze 1560: parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1561: {
1562:
1.19 schwarze 1563: return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1.1 schwarze 1564: }
1565:
1.47 ! schwarze 1566: static int
! 1567: parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1.1 schwarze 1568: {
1569:
1.47 ! schwarze 1570: return(MDOC_HEAD == n->type);
! 1571: }
1.1 schwarze 1572:
1.47 ! schwarze 1573: static int
! 1574: parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
! 1575: {
1.1 schwarze 1576:
1.47 ! schwarze 1577: return(MDOC_BODY == n->type);
1.1 schwarze 1578: }
1579:
1.47 ! schwarze 1580: /*
! 1581: * Add a string to the hash table for the current manual.
! 1582: * Each string has a bitmask telling which macros it belongs to.
! 1583: * When we finish the manual, we'll dump the table.
! 1584: */
1.1 schwarze 1585: static void
1.47 ! schwarze 1586: putkeys(const struct mpage *mpage,
! 1587: const char *cp, size_t sz, uint64_t v)
1.1 schwarze 1588: {
1.47 ! schwarze 1589: struct str *s;
! 1590: unsigned int slot;
! 1591: const char *end;
1.1 schwarze 1592:
1.47 ! schwarze 1593: if (0 == sz)
! 1594: return;
! 1595:
! 1596: end = cp + sz;
! 1597: slot = ohash_qlookupi(&strings, cp, &end);
! 1598: s = ohash_find(&strings, slot);
1.1 schwarze 1599:
1.47 ! schwarze 1600: if (NULL != s && mpage == s->mpage) {
! 1601: s->mask |= v;
1.1 schwarze 1602: return;
1.47 ! schwarze 1603: } else if (NULL == s) {
! 1604: s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
! 1605: memcpy(s->key, cp, sz);
! 1606: ohash_insert(&strings, slot, s);
! 1607: }
! 1608: s->mpage = mpage;
! 1609: s->mask = v;
1.1 schwarze 1610: }
1611:
1612: /*
1.47 ! schwarze 1613: * Take a Unicode codepoint and produce its UTF-8 encoding.
! 1614: * This isn't the best way to do this, but it works.
! 1615: * The magic numbers are from the UTF-8 packaging.
! 1616: * They're not as scary as they seem: read the UTF-8 spec for details.
1.1 schwarze 1617: */
1.47 ! schwarze 1618: static size_t
! 1619: utf8(unsigned int cp, char out[7])
1.1 schwarze 1620: {
1.47 ! schwarze 1621: size_t rc;
1.1 schwarze 1622:
1.47 ! schwarze 1623: rc = 0;
! 1624: if (cp <= 0x0000007F) {
! 1625: rc = 1;
! 1626: out[0] = (char)cp;
! 1627: } else if (cp <= 0x000007FF) {
! 1628: rc = 2;
! 1629: out[0] = (cp >> 6 & 31) | 192;
! 1630: out[1] = (cp & 63) | 128;
! 1631: } else if (cp <= 0x0000FFFF) {
! 1632: rc = 3;
! 1633: out[0] = (cp >> 12 & 15) | 224;
! 1634: out[1] = (cp >> 6 & 63) | 128;
! 1635: out[2] = (cp & 63) | 128;
! 1636: } else if (cp <= 0x001FFFFF) {
! 1637: rc = 4;
! 1638: out[0] = (cp >> 18 & 7) | 240;
! 1639: out[1] = (cp >> 12 & 63) | 128;
! 1640: out[2] = (cp >> 6 & 63) | 128;
! 1641: out[3] = (cp & 63) | 128;
! 1642: } else if (cp <= 0x03FFFFFF) {
! 1643: rc = 5;
! 1644: out[0] = (cp >> 24 & 3) | 248;
! 1645: out[1] = (cp >> 18 & 63) | 128;
! 1646: out[2] = (cp >> 12 & 63) | 128;
! 1647: out[3] = (cp >> 6 & 63) | 128;
! 1648: out[4] = (cp & 63) | 128;
! 1649: } else if (cp <= 0x7FFFFFFF) {
! 1650: rc = 6;
! 1651: out[0] = (cp >> 30 & 1) | 252;
! 1652: out[1] = (cp >> 24 & 63) | 128;
! 1653: out[2] = (cp >> 18 & 63) | 128;
! 1654: out[3] = (cp >> 12 & 63) | 128;
! 1655: out[4] = (cp >> 6 & 63) | 128;
! 1656: out[5] = (cp & 63) | 128;
! 1657: } else
! 1658: return(0);
1.19 schwarze 1659:
1.47 ! schwarze 1660: out[rc] = '\0';
! 1661: return(rc);
1.1 schwarze 1662: }
1663:
1.47 ! schwarze 1664: /*
! 1665: * Store the UTF-8 version of a key, or alias the pointer if the key has
! 1666: * no UTF-8 transcription marks in it.
! 1667: */
! 1668: static void
! 1669: utf8key(struct mchars *mc, struct str *key)
1.1 schwarze 1670: {
1.47 ! schwarze 1671: size_t sz, bsz, pos;
! 1672: char utfbuf[7], res[5];
! 1673: char *buf;
! 1674: const char *seq, *cpp, *val;
! 1675: int len, u;
! 1676: enum mandoc_esc esc;
! 1677:
! 1678: assert(NULL == key->utf8);
! 1679:
! 1680: res[0] = '\\';
! 1681: res[1] = '\t';
! 1682: res[2] = ASCII_NBRSP;
! 1683: res[3] = ASCII_HYPH;
! 1684: res[4] = '\0';
1.1 schwarze 1685:
1.47 ! schwarze 1686: val = key->key;
! 1687: bsz = strlen(val);
1.1 schwarze 1688:
1689: /*
1.47 ! schwarze 1690: * Pre-check: if we have no stop-characters, then set the
! 1691: * pointer as ourselvse and get out of here.
1.1 schwarze 1692: */
1.47 ! schwarze 1693: if (strcspn(val, res) == bsz) {
! 1694: key->utf8 = key->key;
! 1695: return;
! 1696: }
1.1 schwarze 1697:
1.47 ! schwarze 1698: /* Pre-allocate by the length of the input */
1.39 schwarze 1699:
1.47 ! schwarze 1700: buf = mandoc_malloc(++bsz);
! 1701: pos = 0;
1.39 schwarze 1702:
1.47 ! schwarze 1703: while ('\0' != *val) {
! 1704: /*
! 1705: * Halt on the first escape sequence.
! 1706: * This also halts on the end of string, in which case
! 1707: * we just copy, fallthrough, and exit the loop.
! 1708: */
! 1709: if ((sz = strcspn(val, res)) > 0) {
! 1710: memcpy(&buf[pos], val, sz);
! 1711: pos += sz;
! 1712: val += sz;
! 1713: }
1.39 schwarze 1714:
1.47 ! schwarze 1715: if (ASCII_HYPH == *val) {
! 1716: buf[pos++] = '-';
! 1717: val++;
! 1718: continue;
! 1719: } else if ('\t' == *val || ASCII_NBRSP == *val) {
! 1720: buf[pos++] = ' ';
! 1721: val++;
! 1722: continue;
! 1723: } else if ('\\' != *val)
! 1724: break;
1.39 schwarze 1725:
1.47 ! schwarze 1726: /* Read past the slash. */
1.39 schwarze 1727:
1.47 ! schwarze 1728: val++;
! 1729: u = 0;
1.39 schwarze 1730:
1.47 ! schwarze 1731: /*
! 1732: * Parse the escape sequence and see if it's a
! 1733: * predefined character or special character.
! 1734: */
! 1735: esc = mandoc_escape
! 1736: ((const char **)&val, &seq, &len);
! 1737: if (ESCAPE_ERROR == esc)
! 1738: break;
1.39 schwarze 1739:
1.47 ! schwarze 1740: if (ESCAPE_SPECIAL != esc)
! 1741: continue;
! 1742: if (0 == (u = mchars_spec2cp(mc, seq, len)))
! 1743: continue;
1.39 schwarze 1744:
1.47 ! schwarze 1745: /*
! 1746: * If we have a Unicode codepoint, try to convert that
! 1747: * to a UTF-8 byte string.
! 1748: */
! 1749: cpp = utfbuf;
! 1750: if (0 == (sz = utf8(u, utfbuf)))
! 1751: continue;
1.1 schwarze 1752:
1.47 ! schwarze 1753: /* Copy the rendered glyph into the stream. */
1.1 schwarze 1754:
1.47 ! schwarze 1755: sz = strlen(cpp);
! 1756: bsz += sz;
1.1 schwarze 1757:
1.47 ! schwarze 1758: buf = mandoc_realloc(buf, bsz);
1.1 schwarze 1759:
1.47 ! schwarze 1760: memcpy(&buf[pos], cpp, sz);
! 1761: pos += sz;
1.1 schwarze 1762: }
1763:
1.47 ! schwarze 1764: buf[pos] = '\0';
! 1765: key->utf8 = buf;
1.1 schwarze 1766: }
1767:
1.11 schwarze 1768: /*
1.47 ! schwarze 1769: * Flush the current page's terms (and their bits) into the database.
! 1770: * Wrap the entire set of additions in a transaction to make sqlite be a
! 1771: * little faster.
! 1772: * Also, UTF-8-encode the description at the last possible moment.
1.11 schwarze 1773: */
1774: static void
1.47 ! schwarze 1775: dbindex(const struct mpage *mpage, struct mchars *mc)
1.11 schwarze 1776: {
1.47 ! schwarze 1777: struct mlink *mlink;
! 1778: struct str *key;
! 1779: const char *desc;
! 1780: int64_t recno;
! 1781: size_t i;
! 1782: unsigned int slot;
! 1783:
! 1784: if (verb)
! 1785: say(mpage->mlinks->file, "Adding to index");
1.11 schwarze 1786:
1.47 ! schwarze 1787: if (nodb)
1.11 schwarze 1788: return;
1.47 ! schwarze 1789:
! 1790: desc = "";
! 1791: if (NULL != mpage->desc && '\0' != *mpage->desc) {
! 1792: key = ohash_find(&strings,
! 1793: ohash_qlookup(&strings, mpage->desc));
! 1794: assert(NULL != key);
! 1795: if (NULL == key->utf8)
! 1796: utf8key(mc, key);
! 1797: desc = key->utf8;
1.11 schwarze 1798: }
1799:
1.47 ! schwarze 1800: SQL_EXEC("BEGIN TRANSACTION");
1.22 schwarze 1801:
1.47 ! schwarze 1802: i = 1;
1.22 schwarze 1803: /*
1.47 ! schwarze 1804: * XXX The following three lines are obsolete
! 1805: * and only kept for backward compatibility
! 1806: * until apropos(1) and friends have caught up.
1.22 schwarze 1807: */
1.47 ! schwarze 1808: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->file);
! 1809: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->dsec);
! 1810: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->mlinks->arch);
! 1811: SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, desc);
! 1812: SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
! 1813: SQL_STEP(stmts[STMT_INSERT_PAGE]);
! 1814: recno = sqlite3_last_insert_rowid(db);
! 1815: sqlite3_reset(stmts[STMT_INSERT_PAGE]);
! 1816:
! 1817: for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
! 1818: i = 1;
! 1819: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->file);
! 1820: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
! 1821: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
! 1822: SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
! 1823: SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, recno);
! 1824: SQL_STEP(stmts[STMT_INSERT_LINK]);
! 1825: sqlite3_reset(stmts[STMT_INSERT_LINK]);
! 1826: }
! 1827:
! 1828: for (key = ohash_first(&strings, &slot); NULL != key;
! 1829: key = ohash_next(&strings, &slot)) {
! 1830: assert(key->mpage == mpage);
! 1831: if (NULL == key->utf8)
! 1832: utf8key(mc, key);
! 1833: i = 1;
! 1834: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
! 1835: SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->utf8);
! 1836: SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, recno);
! 1837: SQL_STEP(stmts[STMT_INSERT_KEY]);
! 1838: sqlite3_reset(stmts[STMT_INSERT_KEY]);
! 1839: if (key->utf8 != key->key)
! 1840: free(key->utf8);
! 1841: free(key);
1.33 schwarze 1842: }
1843:
1.47 ! schwarze 1844: SQL_EXEC("END TRANSACTION");
! 1845: }
1.41 deraadt 1846:
1.47 ! schwarze 1847: static void
! 1848: dbprune(void)
! 1849: {
! 1850: struct mpage *mpage;
! 1851: struct mlink *mlink;
! 1852: size_t i;
! 1853: unsigned int slot;
1.11 schwarze 1854:
1.47 ! schwarze 1855: if (nodb)
1.11 schwarze 1856: return;
1.47 ! schwarze 1857:
! 1858: mpage = ohash_first(&mpages, &slot);
! 1859: while (NULL != mpage) {
! 1860: mlink = mpage->mlinks;
! 1861: i = 1;
! 1862: SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], i, mlink->file);
! 1863: SQL_STEP(stmts[STMT_DELETE_PAGE]);
! 1864: sqlite3_reset(stmts[STMT_DELETE_PAGE]);
! 1865: if (verb)
! 1866: say(mlink->file, "Deleted from index");
! 1867: mpage = ohash_next(&mpages, &slot);
1.11 schwarze 1868: }
1.47 ! schwarze 1869: }
1.22 schwarze 1870:
1.47 ! schwarze 1871: /*
! 1872: * Close an existing database and its prepared statements.
! 1873: * If "real" is not set, rename the temporary file into the real one.
! 1874: */
! 1875: static void
! 1876: dbclose(int real)
! 1877: {
! 1878: size_t i;
1.11 schwarze 1879:
1.47 ! schwarze 1880: if (nodb)
! 1881: return;
1.11 schwarze 1882:
1.47 ! schwarze 1883: for (i = 0; i < STMT__MAX; i++) {
! 1884: sqlite3_finalize(stmts[i]);
! 1885: stmts[i] = NULL;
1.28 schwarze 1886: }
1.22 schwarze 1887:
1.47 ! schwarze 1888: sqlite3_close(db);
! 1889: db = NULL;
1.11 schwarze 1890:
1.47 ! schwarze 1891: if (real)
! 1892: return;
1.22 schwarze 1893:
1.47 ! schwarze 1894: if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
! 1895: exitcode = (int)MANDOCLEVEL_SYSERR;
! 1896: say(MANDOC_DB, NULL);
1.22 schwarze 1897: }
1.11 schwarze 1898: }
1899:
1.47 ! schwarze 1900: /*
! 1901: * This is straightforward stuff.
! 1902: * Open a database connection to a "temporary" database, then open a set
! 1903: * of prepared statements we'll use over and over again.
! 1904: * If "real" is set, we use the existing database; if not, we truncate a
! 1905: * temporary one.
! 1906: * Must be matched by dbclose().
! 1907: */
! 1908: static int
! 1909: dbopen(int real)
1.2 schwarze 1910: {
1.47 ! schwarze 1911: const char *file, *sql;
! 1912: int rc, ofl;
1.6 schwarze 1913:
1.47 ! schwarze 1914: if (nodb)
! 1915: return(1);
1.6 schwarze 1916:
1.47 ! schwarze 1917: ofl = SQLITE_OPEN_READWRITE;
! 1918: if (0 == real) {
! 1919: file = MANDOC_DB "~";
! 1920: if (-1 == remove(file) && ENOENT != errno) {
! 1921: exitcode = (int)MANDOCLEVEL_SYSERR;
! 1922: say(file, NULL);
! 1923: return(0);
1.28 schwarze 1924: }
1.47 ! schwarze 1925: ofl |= SQLITE_OPEN_EXCLUSIVE;
! 1926: } else
! 1927: file = MANDOC_DB;
1.6 schwarze 1928:
1.47 ! schwarze 1929: rc = sqlite3_open_v2(file, &db, ofl, NULL);
! 1930: if (SQLITE_OK == rc)
! 1931: goto prepare_statements;
! 1932: if (SQLITE_CANTOPEN != rc) {
! 1933: exitcode = (int)MANDOCLEVEL_SYSERR;
! 1934: say(file, NULL);
! 1935: return(0);
! 1936: }
1.6 schwarze 1937:
1.47 ! schwarze 1938: sqlite3_close(db);
! 1939: db = NULL;
1.6 schwarze 1940:
1.47 ! schwarze 1941: if (SQLITE_OK != (rc = sqlite3_open(file, &db))) {
! 1942: exitcode = (int)MANDOCLEVEL_SYSERR;
! 1943: say(file, NULL);
! 1944: return(0);
1.2 schwarze 1945: }
1946:
1.47 ! schwarze 1947: /*
! 1948: * XXX The first three columns in table mpages are obsolete
! 1949: * and only kept for backward compatibility
! 1950: * until apropos(1) and friends have caught up.
! 1951: */
! 1952: sql = "CREATE TABLE \"mpages\" (\n"
! 1953: " \"file\" TEXT NOT NULL,\n"
! 1954: " \"sec\" TEXT NOT NULL,\n"
! 1955: " \"arch\" TEXT NOT NULL,\n"
! 1956: " \"desc\" TEXT NOT NULL,\n"
! 1957: " \"form\" INTEGER NOT NULL,\n"
! 1958: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
! 1959: ");\n"
! 1960: "\n"
! 1961: "CREATE TABLE \"mlinks\" (\n"
! 1962: " \"file\" TEXT NOT NULL,\n"
! 1963: " \"sec\" TEXT NOT NULL,\n"
! 1964: " \"arch\" TEXT NOT NULL,\n"
! 1965: " \"name\" TEXT NOT NULL,\n"
! 1966: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
! 1967: "ON DELETE CASCADE,\n"
! 1968: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
! 1969: ");\n"
! 1970: "\n"
! 1971: "CREATE TABLE \"keys\" (\n"
! 1972: " \"bits\" INTEGER NOT NULL,\n"
! 1973: " \"key\" TEXT NOT NULL,\n"
! 1974: " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
! 1975: "ON DELETE CASCADE,\n"
! 1976: " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
! 1977: ");\n"
! 1978: "\n"
! 1979: "CREATE INDEX \"key_index\" ON keys (key);\n";
! 1980:
! 1981: if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
! 1982: exitcode = (int)MANDOCLEVEL_SYSERR;
! 1983: say(file, "%s", sqlite3_errmsg(db));
! 1984: return(0);
1.2 schwarze 1985: }
1986:
1.47 ! schwarze 1987: prepare_statements:
! 1988: SQL_EXEC("PRAGMA foreign_keys = ON");
! 1989: sql = "DELETE FROM mpages where file=?";
! 1990: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
! 1991: sql = "INSERT INTO mpages "
! 1992: "(file,sec,arch,desc,form) VALUES (?,?,?,?,?)";
! 1993: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
! 1994: sql = "INSERT INTO mlinks "
! 1995: "(file,sec,arch,name,pageid) VALUES (?,?,?,?,?)";
! 1996: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
! 1997: sql = "INSERT INTO keys "
! 1998: "(bits,key,pageid) VALUES (?,?,?)";
! 1999: sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
1.6 schwarze 2000:
1.47 ! schwarze 2001: /*
! 2002: * When opening a new database, we can turn off
! 2003: * synchronous mode for much better performance.
! 2004: */
1.6 schwarze 2005:
1.47 ! schwarze 2006: if (real)
! 2007: SQL_EXEC("PRAGMA synchronous = OFF");
1.11 schwarze 2008:
1.47 ! schwarze 2009: return(1);
! 2010: }
1.6 schwarze 2011:
1.47 ! schwarze 2012: static void *
! 2013: hash_halloc(size_t sz, void *arg)
! 2014: {
1.6 schwarze 2015:
1.47 ! schwarze 2016: return(mandoc_calloc(sz, 1));
! 2017: }
1.2 schwarze 2018:
1.47 ! schwarze 2019: static void *
! 2020: hash_alloc(size_t sz, void *arg)
! 2021: {
1.28 schwarze 2022:
1.47 ! schwarze 2023: return(mandoc_malloc(sz));
! 2024: }
1.6 schwarze 2025:
1.47 ! schwarze 2026: static void
! 2027: hash_free(void *p, size_t sz, void *arg)
! 2028: {
1.26 schwarze 2029:
1.47 ! schwarze 2030: free(p);
! 2031: }
1.6 schwarze 2032:
1.47 ! schwarze 2033: static int
! 2034: set_basedir(const char *targetdir)
! 2035: {
! 2036: static char startdir[PATH_MAX];
! 2037: static int fd;
1.6 schwarze 2038:
1.47 ! schwarze 2039: /*
! 2040: * Remember where we started by keeping a fd open to the origin
! 2041: * path component: throughout this utility, we chdir() a lot to
! 2042: * handle relative paths, and by doing this, we can return to
! 2043: * the starting point.
! 2044: */
! 2045: if ('\0' == *startdir) {
! 2046: if (NULL == getcwd(startdir, PATH_MAX)) {
! 2047: exitcode = (int)MANDOCLEVEL_SYSERR;
! 2048: if (NULL != targetdir)
! 2049: say(".", NULL);
! 2050: return(0);
! 2051: }
! 2052: if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
! 2053: exitcode = (int)MANDOCLEVEL_SYSERR;
! 2054: say(startdir, NULL);
! 2055: return(0);
1.11 schwarze 2056: }
1.47 ! schwarze 2057: if (NULL == targetdir)
! 2058: targetdir = startdir;
! 2059: } else {
! 2060: if (-1 == fd)
! 2061: return(0);
! 2062: if (-1 == fchdir(fd)) {
! 2063: close(fd);
! 2064: basedir[0] = '\0';
! 2065: exitcode = (int)MANDOCLEVEL_SYSERR;
! 2066: say(startdir, NULL);
! 2067: return(0);
1.2 schwarze 2068: }
1.47 ! schwarze 2069: if (NULL == targetdir) {
! 2070: close(fd);
! 2071: return(1);
1.2 schwarze 2072: }
2073: }
1.47 ! schwarze 2074: if (NULL == realpath(targetdir, basedir)) {
! 2075: basedir[0] = '\0';
! 2076: exitcode = (int)MANDOCLEVEL_BADARG;
! 2077: say(targetdir, NULL);
! 2078: return(0);
! 2079: } else if (-1 == chdir(basedir)) {
! 2080: exitcode = (int)MANDOCLEVEL_BADARG;
! 2081: say("", NULL);
! 2082: return(0);
! 2083: }
! 2084: return(1);
1.2 schwarze 2085: }
2086:
2087: static void
1.47 ! schwarze 2088: say(const char *file, const char *format, ...)
1.2 schwarze 2089: {
1.47 ! schwarze 2090: va_list ap;
1.2 schwarze 2091:
1.47 ! schwarze 2092: if ('\0' != *basedir)
! 2093: fprintf(stderr, "%s", basedir);
! 2094: if ('\0' != *basedir && '\0' != *file)
! 2095: fputs("//", stderr);
! 2096: if ('\0' != *file)
! 2097: fprintf(stderr, "%s", file);
! 2098: fputs(": ", stderr);
1.31 schwarze 2099:
1.47 ! schwarze 2100: if (NULL == format) {
! 2101: perror(NULL);
! 2102: return;
1.2 schwarze 2103: }
1.47 ! schwarze 2104:
! 2105: va_start(ap, format);
! 2106: vfprintf(stderr, format, ap);
! 2107: va_end(ap);
! 2108:
! 2109: fputc('\n', stderr);
1.1 schwarze 2110: }