da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin/***********************************************************************
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* This software is part of the ast package *
3e14f97f673e8a630f076077de35afdd43dc1587Roger A. Faulkner* Copyright (c) 1985-2010 AT&T Intellectual Property *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* and is licensed under the *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* Common Public License, Version 1.0 *
7c2fbfb345896881c631598ee3852ce9ce33fb07April Chin* by AT&T Intellectual Property *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* A copy of the License is available at *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* (with md5 checksum 059e8cd6165cb4c31e351f2b69388fd9) *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* Information and Software Systems Research *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* AT&T Research *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* Florham Park NJ *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* Glenn Fowler <gsf@research.att.com> *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* David Korn <dgk@research.att.com> *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin* Phong Vo <kpv@research.att.com> *
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin***********************************************************************/
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * original code
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * James A. Woods, Informatics General Corporation,
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * NASA Ames Research Center, 6/81.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * Usenix ;login:, February/March, 1983, p. 8.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * Glenn Fowler
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * AT&T Research
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * modified from the original BSD source
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 'fastfind' scans a file list for the full pathname of a file
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * given only a piece of the name. The list is processed with
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * with "front-compression" and bigram coding. Front compression reduces
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * space by a factor of 4-5, bigram coding by a further 20-25%.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * there are 4 methods:
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_old original with 7 bit bigram encoding (no magic)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_gnu 8 bit clean front compression (FF_gnu_magic)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_dir FF_gnu with sfgetl/sfputl and trailing / on dirs (FF_dir_magic)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_typ FF_dir with (mime) types (FF_typ_magic)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * the bigram encoding steals the eighth bit (that's why its FF_old)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * maybe one day we'll limit it to readonly:
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 0-2*FF_OFF likeliest differential counts + offset to make nonnegative
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_ESC 4 byte big-endian out-of-range count+FF_OFF follows
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_MIN-FF_MAX ascii residue
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * >=FF_MAX bigram codes
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * a two-tiered string search technique is employed
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * a metacharacter-free subpattern and partial pathname is matched
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * backwards to avoid full expansion of the pathname list
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * then the actual shell glob-style regular expression (if in this form)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * is matched against the candidate pathnames using the slower regexec()
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * The original BSD code is covered by the BSD license:
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * Copyright (c) 1985, 1993, 1999
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * The Regents of the University of California. All rights reserved.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * Redistribution and use in source and binary forms, with or without
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * modification, are permitted provided that the following conditions
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * are met:
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 1. Redistributions of source code must retain the above copyright
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * notice, this list of conditions and the following disclaimer.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 2. Redistributions in binary form must reproduce the above copyright
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * notice, this list of conditions and the following disclaimer in the
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * documentation and/or other materials provided with the distribution.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 3. Neither the name of the University nor the names of its contributors
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * may be used to endorse or promote products derived from this software
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * without specific prior written permission.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * SUCH DAMAGE.
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinstatic const char id[] = "\n@(#)$Id: fastfind (AT&T Research) 2002-10-02 $\0\n";
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * this db could be anywhere
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * findcodes[] directories are checked for findnames[i]
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "locatedb",
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "locate.database",
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin "slocate.db",
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * convert t to lower case and drop leading x- and x- after /
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * converted value copied to b of size n
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int c;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* b = buf;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (c = *t++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if ((*b++ = c) == '/' && (*t == 'x' || *t == 'X') && *(t + 1) == '-')
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * return a fastfind stream handle for pattern
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinfindopen(const char* file, const char* pattern, const char* type, Finddisc_t* disc)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* p;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* s;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* b;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int i;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int j;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * NOTE: searching for FIND_CODES would be much simpler if we
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * just stuck with our own, but we also support GNU
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * locate codes and have to search for the one of a
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * bazillion possible names for that file
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(fp = (Find_t*)vmnewof(vm, 0, Find_t, 1, sizeof(Encode_t) - sizeof(Code_t))))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin j = (findcodes[0] = (char*)file) && *file == '/' ? 1 : elementsof(findcodes);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * look for the codes file, but since it may not exist yet,
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * also look for the containing directory if i<2 or if
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * it is sufficiently qualified (FIND_MATCH)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (i = 0; i < j; i++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->encode.file, sizeof(fp->encode.file), "%s/%s", path, findnames[k]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (strchr(findnames[k], '/') && (b = strrchr(fp->encode.file, '/')))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!stat(fp->encode.file, &st) && st.st_uid == uid && (st.st_mode & S_IWUSR))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->encode.file, sizeof(fp->encode.file), "%s", path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->encode.file, sizeof(fp->encode.file), "%s", path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!stat(fp->encode.file, &st) && st.st_uid == uid && (st.st_mode & S_IWUSR))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if (pathpath(fp->encode.file, path, "", PATH_REGULAR|PATH_READ|PATH_WRITE))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->encode.file, sizeof(fp->encode.file), "%-.*s", b - path, path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (pathpath(fp->encode.temp, fp->encode.file, "", PATH_EXECUTE|PATH_READ|PATH_WRITE) &&
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin !stat(fp->encode.temp, &st) && st.st_uid == uid && (st.st_mode & S_IWUSR))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->encode.file, sizeof(fp->encode.file), "%s%s", fp->encode.temp, b);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (i >= j)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: cannot locate codes", file ? file : findcodes[2]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * FF_old generates temp data that is read
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * in a second pass to generate the real codes
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "cannot create tmp file");
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * the rest generate into a temp file that
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * is simply renamed on completion
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!pathtemp(fp->encode.temp, sizeof(fp->encode.temp), p, "ff", &fd))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "%s: cannot create tmp file in this directory", p ? p : ".");
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(fp->fp = sfnew(NiL, NiL, (size_t)SF_UNBOUND, fd, SF_WRITE)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "%s: cannot open tmp file", fp->encode.temp);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(fp->encode.namedict = dtopen(&fp->encode.namedisc, Dttree)) || !(fp->encode.indexdict = dtopen(&fp->encode.indexdisc, Dttree)) || !(tp = newof(0, Type_t, 1, strlen(s) + 1)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "cannot allocate type table");
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * type index 1 is always system/dir
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin j = (findcodes[0] = (char*)file) && *file == '/' ? 1 : elementsof(findcodes);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (i = 0; i < j; i++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(fp->decode.path, sizeof(fp->decode.path), "%s/%s", path, findnames[k]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if ((path = pathpath(fp->decode.path, path, "", PATH_REGULAR|PATH_READ)) && (fp->fp = sfopen(NiL, path, "r")))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: cannot locate codes", file ? file : findcodes[2]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: cannot stat codes", path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (fp->secure = ((st.st_mode & (S_IRGRP|S_IROTH)) == S_IRGRP) && st.st_gid == getegid() && getegid() != getgid())
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(*s++ = fp->decode.bigram2[i] = j) && (i || fp->decode.bigram1[0] >= '0' && fp->decode.bigram1[0] <= '1'))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin type = (const char*)typefix(fp->decode.bigram2, sizeof(fp->decode.bigram2), type);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin memset(fp->decode.bigram1, 0, sizeof(fp->decode.bigram1));
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (j = 0, i = 1;; i++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (type && !j)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (j == EOF || fp->decode.count >= sizeof(fp->decode.path))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (i < 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * set up the physical dir table
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(fp->dirs = vmnewof(fp->vm, 0, char*, 2 * k + 1, 0)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * fill the dir list with logical and
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * physical names since we don't know
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * which way the db was encoded (it
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * could be *both* ways)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (i = q = 0; i < k; i++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if (!p && !(p = getcwd(fp->decode.path, sizeof(fp->decode.path))))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfsprintf(b, sizeof(fp->decode.temp) - 1, "%s/%s", p, s);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *(s + 1) = 0;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin *(s + 1) = 0;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (i = 0; i < q; i++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: %s code format does not support directory verification", path, fp->method == FF_gnu ? FF_gnu_magic : "OLD-BIGRAM");
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * extract last glob-free subpattern in name for fast pre-match
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * prepend 0 for backwards match
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (p = s = (char*)pattern)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin switch (*b++ = *p++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin case '\\':
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!*p++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (*p == ']')
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (i = regcomp(&fp->decode.re, pattern, REG_SHELL|REG_AUGMENTED|(fp->decode.ignorecase?REG_ICASE:0)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin regerror(i, &fp->decode.re, fp->decode.temp, sizeof(fp->decode.temp));
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: %s", pattern, fp->decode.temp);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (i = *s++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: invalid codes", path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * return the next fastfind path
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * 0 returned when list exhausted
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* p;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* q;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* s;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* b;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* e;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int c;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int n;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int m;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin unsigned char w[4];
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (c == 0x80)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = c << 8;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (n & 0x8000)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if ((n = c) & 0x80)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin } while (*p++ = c);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin c = (int32_t)((w[0] << 24) | (w[1] << 16) | (w[2] << 8) | w[3]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * the old format uses machine
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * byte order; this test uses
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * the smallest magnitude of
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * both byte orders on the
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * first encoded path motion
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * to determine the original
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * byte order
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (m < 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = (int32_t)((w[3] << 24) | (w[2] << 16) | (w[1] << 8) | w[0]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (n < 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (m < n)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin c = (int32_t)((w[3] << 24) | (w[2] << 16) | (w[1] << 8) | w[0]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin c = (int32_t)((w[3] << 24) | (w[2] << 16) | (w[1] << 8) | w[0]);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (p = fp->decode.path + fp->decode.count; (c = sfgetc(fp->fp)) > FF_ESC;)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * use the ordering and lengths to prune
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * comparison function calls
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * (*fp->dirs)[*fp->lens]=='/' if its
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * already been matched
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if (n == m)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(n = strcasecmp(*fp->dirs, fp->decode.path)) && (ignorecase || !strcmp(*fp->dirs, fp->decode.path)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (m > 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (n >= 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * n<0 skip this subtree
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * n==0 keep as is
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * n>0 read this dir now
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /* NOT IMPLEMENTED YET */
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (*(s = p) == '/')
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (; s >= b; s--)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (*s == *fp->decode.end || ignorecase && tolower(*s) == *fp->decode.end)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (e = fp->decode.end - 1, q = s - 1; *e && (*q == *e || tolower(*q) == *e); e--, q--);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (e = fp->decode.end - 1, q = s - 1; *e && *q == *e; e--, q--);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!fp->decode.match || strgrpmatch(fp->decode.path, fp->decode.pattern, NiL, 0, STR_MAXIMAL|STR_LEFT|STR_RIGHT|ignorecase))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (*p == '/')
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if (!fp->decode.match || !(n = regexec(&fp->decode.re, fp->decode.path, 0, NiL, 0)))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin else if (n != REG_NOMATCH)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin regerror(n, &fp->decode.re, fp->decode.temp, sizeof(fp->decode.temp));
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: %s", fp->decode.pattern, fp->decode.temp);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * add path to the code table
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * paths are assumed to be in sort order
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chinfindwrite(register Find_t* fp, const char* path, size_t len, const char* type)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char* s;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char* e;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned char* p;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int n;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int d;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register Type_t* x;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register unsigned long u;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin len = sfsprintf(fp->encode.mark, sizeof(fp->encode.mark), "%-.*s/", len, path);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin s = (unsigned char*)path;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin e = s + len++;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (s < e)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (*s != *p++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin n = s - (unsigned char*)path;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfprintf(fp->fp, "%ld", n - fp->encode.prefix + FF_OFF);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (s < e)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (s >= e)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (p < e)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin type = (const char*)typefix((char*)fp->encode.bigram, sizeof(fp->encode.bigram), type);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin /*FALLTHROUGH...*/
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * findsync() helper
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: write error [sfsync]", fp->encode.file);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: write error [sferror]", fp->encode.file);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: write error [sfclose]", fp->encode.file);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * finish the code table
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register char* s;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int n;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int m;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register int d;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin register Type_t* x;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * replace the real file with the temp file
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "%s: cannot rename from tmp file %s", fp->encode.file, fp->encode.temp);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * determine the top FF_MAX bigrams
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (n = 0; n < FF_MAX; n++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (m = 0; m < FF_MAX; m++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (n = USHRT_MAX; n >= 0; n--)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if ((m += d) > FF_MAX)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (--n >= 0)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * commit the real file
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "cannot rewind tmp file");
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * dump the bigrams
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin sfwrite(sp, fp->encode.bigram, sizeof(fp->encode.bigram));
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * encode the massaged paths
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin z = strtol(s, &t, 0);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin while (n = *s++)
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (!(m = *s++))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, ERROR_SYSTEM|2, "%s: cannot read tmp file", fp->encode.temp);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * commit the output file
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * write the header magic
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * write the type table in index order starting with 1
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin for (x = (Type_t*)dtfirst(fp->encode.indexdict); x; x = (Type_t*)dtnext(fp->encode.indexdict, x))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * append the front compressed strings
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin if (sfmove(fp->fp, sp, SF_UNBOUND, -1) < 0 || !sfeof(fp->fp))
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: cannot append codes", fp->encode.file);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin (*fp->disc->errorf)(fp, fp->disc, 2, "%s: cannot write codes", fp->encode.file);
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin * close an open fastfind stream
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin int n = 0;
da2e3ebdc1edfbc5028edf1354e7dd2fa69a7968chin return -1;