1 /* Licensed to the Apache Software Foundation (ASF) under one or more
2 * contributor license agreements. See the NOTICE file distributed with
3 * this work for additional information regarding copyright ownership.
4 * The ASF licenses this file to You under the Apache License, Version 2.0
5 * (the "License"); you may not use this file except in compliance with
6 * the License. You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
18 * util.c: string utility things
21 * 1995-96 Many changes by the Apache Software Foundation
26 * #define DEBUG to trace all cfg_open*()/cfg_closefile() calls
27 * #define DEBUG_CFG_LINES to trace every line read from the config files
31 #include "apr_strings.h"
34 #define APR_WANT_STDIO
35 #define APR_WANT_STRFUNC
41 #if APR_HAVE_PROCESS_H
42 #include <process.h> /* for getpid() on Win32 */
45 #include <netdb.h> /* for gethostbyname() */
48 #include "ap_config.h"
49 #include "apr_base64.h"
51 #include "http_main.h"
53 #include "http_protocol.h"
54 #include "http_config.h"
55 #include "http_core.h"
56 #include "util_ebcdic.h"
57 #include "util_varbuf.h"
66 /* A bunch of functions in util.c scan strings looking for certain characters.
67 * To make that more efficient we encode a lookup table. The test_char_table
68 * is generated automatically by gen_test_char.c.
70 #include "test_char.h"
72 /* we assume the folks using this ensure 0 <= c < 256... which means
73 * you need a cast to (unsigned char) first, you can't just plug a
74 * char in here and get it to work, because if char is signed then it
75 * will first be sign extended.
77 #define TEST_CHAR(c, f) (test_char_table[(unsigned)(c)] & (f))
79 /* Win32/NetWare/OS2 need to check for both forward and back slashes
80 * in ap_getparents() and ap_escape_url.
82 #ifdef CASE_BLIND_FILESYSTEM
83 #define IS_SLASH(s) ((s == '/') || (s == '\\'))
86 #define IS_SLASH(s) (s == '/')
90 /* we know core's module_index is 0 */
91 #undef APLOG_MODULE_INDEX
92 #define APLOG_MODULE_INDEX AP_CORE_MODULE_INDEX
96 * Examine a field value (such as a media-/content-type) string and return
97 * it sans any parameters; e.g., strip off any ';charset=foo' and the like.
99 AP_DECLARE(char *) ap_field_noparam(apr_pool_t *p, const char *intype)
103 if (intype == NULL) return NULL;
105 semi = ap_strchr_c(intype, ';');
107 return apr_pstrdup(p, intype);
110 while ((semi > intype) && apr_isspace(semi[-1])) {
113 return apr_pstrndup(p, intype, semi - intype);
117 AP_DECLARE(char *) ap_ht_time(apr_pool_t *p, apr_time_t t, const char *fmt,
121 char ts[MAX_STRING_LEN];
122 char tf[MAX_STRING_LEN];
129 apr_time_exp_gmt(&xt, t);
130 /* Convert %Z to "GMT" and %z to "+0000";
131 * on hosts that do not have a time zone string in struct tm,
132 * strftime must assume its argument is local time.
134 for(strp = tf, f = fmt; strp < tf + sizeof(tf) - 6 && (*strp = *f)
136 if (*f != '%') continue;
147 case 'z': /* common extension */
161 apr_time_exp_lt(&xt, t);
164 /* check return code? */
165 apr_strftime(ts, &retcode, MAX_STRING_LEN, fmt, &xt);
166 ts[MAX_STRING_LEN - 1] = '\0';
167 return apr_pstrdup(p, ts);
170 /* Roy owes Rob beer. */
171 /* Rob owes Roy dinner. */
173 /* These legacy comments would make a lot more sense if Roy hadn't
174 * replaced the old later_than() routine with util_date.c.
176 * Well, okay, they still wouldn't make any sense.
179 /* Match = 0, NoMatch = 1, Abort = -1
180 * Based loosely on sections of wildmat.c by Rich Salz
181 * Hmmm... shouldn't this really go component by component?
183 AP_DECLARE(int) ap_strcmp_match(const char *str, const char *expected)
187 for (x = 0, y = 0; expected[y]; ++y, ++x) {
188 if ((!str[x]) && (expected[y] != '*'))
190 if (expected[y] == '*') {
191 while (expected[++y] == '*');
196 if ((ret = ap_strcmp_match(&str[x++], &expected[y])) != 1)
201 else if ((expected[y] != '?') && (str[x] != expected[y]))
204 return (str[x] != '\0');
207 AP_DECLARE(int) ap_strcasecmp_match(const char *str, const char *expected)
211 for (x = 0, y = 0; expected[y]; ++y, ++x) {
212 if (!str[x] && expected[y] != '*')
214 if (expected[y] == '*') {
215 while (expected[++y] == '*');
220 if ((ret = ap_strcasecmp_match(&str[x++], &expected[y])) != 1)
225 else if (expected[y] != '?'
226 && apr_tolower(str[x]) != apr_tolower(expected[y]))
229 return (str[x] != '\0');
232 /* We actually compare the canonical root to this root, (but we don't
233 * waste time checking the case), since every use of this function in
234 * httpd-2.1 tests if the path is 'proper', meaning we've already passed
235 * it through apr_filepath_merge, or we haven't.
237 AP_DECLARE(int) ap_os_is_path_absolute(apr_pool_t *p, const char *dir)
240 const char *ourdir = dir;
241 if (apr_filepath_root(&newpath, &dir, 0, p) != APR_SUCCESS
242 || strncmp(newpath, ourdir, strlen(newpath)) != 0) {
248 AP_DECLARE(int) ap_is_matchexp(const char *str)
252 for (x = 0; str[x]; x++)
253 if ((str[x] == '*') || (str[x] == '?'))
259 * Here's a pool-based interface to the POSIX-esque ap_regcomp().
260 * Note that we return ap_regex_t instead of being passed one.
261 * The reason is that if you use an already-used ap_regex_t structure,
262 * the memory that you've already allocated gets forgotten, and
263 * regfree() doesn't clear it. So we don't allow it.
266 static apr_status_t regex_cleanup(void *preg)
268 ap_regfree((ap_regex_t *) preg);
272 AP_DECLARE(ap_regex_t *) ap_pregcomp(apr_pool_t *p, const char *pattern,
275 ap_regex_t *preg = apr_palloc(p, sizeof *preg);
276 int err = ap_regcomp(preg, pattern, cflags);
278 if (err == AP_REG_ESPACE)
283 apr_pool_cleanup_register(p, (void *) preg, regex_cleanup,
284 apr_pool_cleanup_null);
289 AP_DECLARE(void) ap_pregfree(apr_pool_t *p, ap_regex_t *reg)
292 apr_pool_cleanup_kill(p, (void *) reg, regex_cleanup);
296 * Similar to standard strstr() but we ignore case in this version.
297 * Based on the strstr() implementation further below.
299 AP_DECLARE(char *) ap_strcasestr(const char *s1, const char *s2)
307 for ( ; (*s1 != '\0') && (apr_tolower(*s1) != apr_tolower(*s2)); s1++);
311 /* found first character of s2, see if the rest matches */
314 for (++p1, ++p2; apr_tolower(*p1) == apr_tolower(*p2); ++p1, ++p2) {
316 /* both strings ended together */
321 /* second string ended, a match */
324 /* didn't find a match here, try starting at next character in s1 */
331 * Returns an offsetted pointer in bigstring immediately after
332 * prefix. Returns bigstring if bigstring doesn't start with
333 * prefix or if prefix is longer than bigstring while still matching.
334 * NOTE: pointer returned is relative to bigstring, so we
335 * can use standard pointer comparisons in the calling function
336 * (eg: test if ap_stripprefix(a,b) == a)
338 AP_DECLARE(const char *) ap_stripprefix(const char *bigstring,
347 while (*p1 && *prefix) {
348 if (*p1++ != *prefix++)
354 /* hit the end of bigstring! */
358 /* This function substitutes for $0-$9, filling in regular expression
359 * submatches. Pass it the same nmatch and pmatch arguments that you
360 * passed ap_regexec(). pmatch should not be greater than the maximum number
361 * of subexpressions - i.e. one more than the re_nsub member of ap_regex_t.
363 * nmatch must be <=AP_MAX_REG_MATCH (10).
365 * input should be the string with the $-expressions, source should be the
366 * string that was matched against.
368 * It returns the substituted string, or NULL if a vbuf is used.
369 * On errors, returns the orig string.
371 * Parts of this code are based on Henry Spencer's regsub(), from his
372 * AT&T V8 regexp package.
375 static apr_status_t regsub_core(apr_pool_t *p, char **result,
376 struct ap_varbuf *vb, const char *input,
377 const char *source, apr_size_t nmatch,
378 ap_regmatch_t pmatch[], apr_size_t maxlen)
380 const char *src = input;
386 AP_DEBUG_ASSERT((result && p && !vb) || (vb && !p && !result));
387 if (!source || nmatch>AP_MAX_REG_MATCH)
391 if (maxlen > 0 && len >= maxlen)
394 *result = apr_pstrmemdup(p, src, len);
398 ap_varbuf_strmemcat(vb, src, len);
403 /* First pass, find the size */
404 while ((c = *src++) != '\0') {
405 if (c == '$' && apr_isdigit(*src))
408 no = AP_MAX_REG_MATCH;
410 if (no >= AP_MAX_REG_MATCH) { /* Ordinary character. */
411 if (c == '\\' && *src)
415 else if (no < nmatch && pmatch[no].rm_so < pmatch[no].rm_eo) {
416 if (APR_SIZE_MAX - len <= pmatch[no].rm_eo - pmatch[no].rm_so)
418 len += pmatch[no].rm_eo - pmatch[no].rm_so;
423 if (len >= maxlen && maxlen > 0)
427 *result = dst = apr_palloc(p, len + 1);
430 if (vb->strlen == AP_VARBUF_UNKNOWN)
431 vb->strlen = strlen(vb->buf);
432 ap_varbuf_grow(vb, vb->strlen + len);
433 dst = vb->buf + vb->strlen;
437 /* Now actually fill in the string */
441 while ((c = *src++) != '\0') {
442 if (c == '$' && apr_isdigit(*src))
445 no = AP_MAX_REG_MATCH;
447 if (no >= AP_MAX_REG_MATCH) { /* Ordinary character. */
448 if (c == '\\' && *src)
452 else if (no < nmatch && pmatch[no].rm_so < pmatch[no].rm_eo) {
453 len = pmatch[no].rm_eo - pmatch[no].rm_so;
454 memcpy(dst, source + pmatch[no].rm_so, len);
464 #ifndef AP_PREGSUB_MAXLEN
465 #define AP_PREGSUB_MAXLEN (HUGE_STRING_LEN * 8)
467 AP_DECLARE(char *) ap_pregsub(apr_pool_t *p, const char *input,
468 const char *source, apr_size_t nmatch,
469 ap_regmatch_t pmatch[])
472 apr_status_t rc = regsub_core(p, &result, NULL, input, source, nmatch,
473 pmatch, AP_PREGSUB_MAXLEN);
474 if (rc != APR_SUCCESS)
479 AP_DECLARE(apr_status_t) ap_pregsub_ex(apr_pool_t *p, char **result,
480 const char *input, const char *source,
481 apr_size_t nmatch, ap_regmatch_t pmatch[],
484 apr_status_t rc = regsub_core(p, result, NULL, input, source, nmatch,
486 if (rc != APR_SUCCESS)
492 * Parse .. so we don't compromise security
494 AP_DECLARE(void) ap_getparents(char *name)
499 /* Four paseses, as per RFC 1808 */
500 /* a) remove ./ path segments */
501 for (next = name; *next && (*next != '.'); next++) {
504 l = w = first_dot = next - name;
505 while (name[l] != '\0') {
506 if (name[l] == '.' && IS_SLASH(name[l + 1])
507 && (l == 0 || IS_SLASH(name[l - 1])))
510 name[w++] = name[l++];
513 /* b) remove trailing . path, segment */
514 if (w == 1 && name[0] == '.')
516 else if (w > 1 && name[w - 1] == '.' && IS_SLASH(name[w - 2]))
520 /* c) remove all xx/../ segments. (including leading ../ and /../) */
523 while (name[l] != '\0') {
524 if (name[l] == '.' && name[l + 1] == '.' && IS_SLASH(name[l + 2])
525 && (l == 0 || IS_SLASH(name[l - 1]))) {
526 register int m = l + 3, n;
530 while (l >= 0 && !IS_SLASH(name[l]))
537 while ((name[n] = name[m]))
544 /* d) remove trailing xx/.. segment. */
545 if (l == 2 && name[0] == '.' && name[1] == '.')
547 else if (l > 2 && name[l - 1] == '.' && name[l - 2] == '.'
548 && IS_SLASH(name[l - 3])) {
551 while (l >= 0 && !IS_SLASH(name[l]))
561 AP_DECLARE(void) ap_no2slash(char *name)
567 #ifdef HAVE_UNC_PATHS
568 /* Check for UNC names. Leave leading two slashes. */
569 if (s[0] == '/' && s[1] == '/')
574 if ((*d++ = *s) == '/') {
588 * copy at most n leading directories of s into d
589 * d should be at least as large as s plus 1 extra byte
591 * the return value is the ever useful pointer to the trailing \0 of d
593 * MODIFIED FOR HAVE_DRIVE_LETTERS and NETWARE environments,
594 * so that if n == 0, "/" is returned in d with n == 1
595 * and s == "e:/test.html", "e:/" is returned in d
596 * *** See also directory_walk in modules/http/http_request.c
599 * /a/b, 0 ==> / (true for all platforms)
608 * c:/a/b 3 ==> c:/a/b
609 * c:/a/b 4 ==> c:/a/b
611 AP_DECLARE(char *) ap_make_dirstr_prefix(char *d, const char *s, int n)
620 if (*s == '\0' || (*s == '/' && (--n) == 0)) {
632 * return the parent directory name including trailing / of the file s
634 AP_DECLARE(char *) ap_make_dirstr_parent(apr_pool_t *p, const char *s)
636 const char *last_slash = ap_strrchr_c(s, '/');
640 if (last_slash == NULL) {
641 return apr_pstrdup(p, "");
643 l = (last_slash - s) + 1;
644 d = apr_pstrmemdup(p, s, l);
650 AP_DECLARE(int) ap_count_dirs(const char *path)
654 for (x = 0, n = 0; path[x]; x++)
660 AP_DECLARE(char *) ap_getword_nc(apr_pool_t *atrans, char **line, char stop)
662 return ap_getword(atrans, (const char **) line, stop);
665 AP_DECLARE(char *) ap_getword(apr_pool_t *atrans, const char **line, char stop)
667 const char *pos = *line;
671 while ((*pos != stop) && *pos) {
676 res = apr_pstrmemdup(atrans, *line, len);
679 while (*pos == stop) {
688 AP_DECLARE(char *) ap_getword_white_nc(apr_pool_t *atrans, char **line)
690 return ap_getword_white(atrans, (const char **) line);
693 AP_DECLARE(char *) ap_getword_white(apr_pool_t *atrans, const char **line)
695 const char *pos = *line;
699 while (!apr_isspace(*pos) && *pos) {
704 res = apr_pstrmemdup(atrans, *line, len);
706 while (apr_isspace(*pos)) {
715 AP_DECLARE(char *) ap_getword_nulls_nc(apr_pool_t *atrans, char **line,
718 return ap_getword_nulls(atrans, (const char **) line, stop);
721 AP_DECLARE(char *) ap_getword_nulls(apr_pool_t *atrans, const char **line,
724 const char *pos = ap_strchr_c(*line, stop);
728 apr_size_t len = strlen(*line);
729 res = apr_pstrmemdup(atrans, *line, len);
734 res = apr_pstrndup(atrans, *line, pos - *line);
743 /* Get a word, (new) config-file style --- quoted strings and backslashes
747 static char *substring_conf(apr_pool_t *p, const char *start, int len,
750 char *result = apr_palloc(p, len + 2);
754 for (i = 0; i < len; ++i) {
755 if (start[i] == '\\' && (start[i + 1] == '\\'
756 || (quote && start[i + 1] == quote)))
757 *resp++ = start[++i];
763 #if RESOLVE_ENV_PER_TOKEN
764 return (char *)ap_resolve_env(p,result);
770 AP_DECLARE(char *) ap_getword_conf_nc(apr_pool_t *p, char **line)
772 return ap_getword_conf(p, (const char **) line);
775 AP_DECLARE(char *) ap_getword_conf(apr_pool_t *p, const char **line)
777 const char *str = *line, *strend;
781 while (*str && apr_isspace(*str))
789 if ((quote = *str) == '"' || quote == '\'') {
791 while (*strend && *strend != quote) {
792 if (*strend == '\\' && strend[1] &&
793 (strend[1] == quote || strend[1] == '\\')) {
800 res = substring_conf(p, str + 1, strend - str - 1, quote);
802 if (*strend == quote)
807 while (*strend && !apr_isspace(*strend))
810 res = substring_conf(p, str, strend - str, 0);
813 while (*strend && apr_isspace(*strend))
819 AP_DECLARE(int) ap_cfg_closefile(ap_configfile_t *cfp)
822 ap_log_error(APLOG_MARK, APLOG_DEBUG, 0, NULL, APLOGNO(00551)
823 "Done with config file %s", cfp->name);
825 return (cfp->close == NULL) ? 0 : cfp->close(cfp->param);
828 /* we can't use apr_file_* directly because of linking issues on Windows */
829 static apr_status_t cfg_close(void *param)
831 return apr_file_close(param);
834 static apr_status_t cfg_getch(char *ch, void *param)
836 return apr_file_getc(ch, param);
839 static apr_status_t cfg_getstr(void *buf, apr_size_t bufsiz, void *param)
841 return apr_file_gets(buf, bufsiz, param);
844 /* Open a ap_configfile_t as FILE, return open ap_configfile_t struct pointer */
845 AP_DECLARE(apr_status_t) ap_pcfg_openfile(ap_configfile_t **ret_cfg,
846 apr_pool_t *p, const char *name)
848 ap_configfile_t *new_cfg;
849 apr_file_t *file = NULL;
857 ap_log_error(APLOG_MARK, APLOG_ERR, 0, NULL, APLOGNO(00552)
858 "Internal error: pcfg_openfile() called with NULL filename");
862 status = apr_file_open(&file, name, APR_READ | APR_BUFFERED,
865 ap_log_error(APLOG_MARK, APLOG_DEBUG, 0, NULL, APLOGNO(00553)
866 "Opening config file %s (%s)",
867 name, (status != APR_SUCCESS) ?
868 apr_strerror(status, buf, sizeof(buf)) : "successful");
870 if (status != APR_SUCCESS)
873 status = apr_file_info_get(&finfo, APR_FINFO_TYPE, file);
874 if (status != APR_SUCCESS)
877 if (finfo.filetype != APR_REG &&
878 #if defined(WIN32) || defined(OS2) || defined(NETWARE)
879 strcasecmp(apr_filepath_name_get(name), "nul") != 0) {
881 strcmp(name, "/dev/null") != 0) {
882 #endif /* WIN32 || OS2 */
883 ap_log_error(APLOG_MARK, APLOG_ERR, 0, NULL, APLOGNO(00554)
884 "Access to file %s denied by server: not a regular file",
886 apr_file_close(file);
891 /* Some twisted character [no pun intended] at MS decided that a
892 * zero width joiner as the lead wide character would be ideal for
893 * describing Unicode text files. This was further convoluted to
894 * another MSism that the same character mapped into utf-8, EF BB BF
895 * would signify utf-8 text files.
897 * Since MS configuration files are all protecting utf-8 encoded
898 * Unicode path, file and resource names, we already have the correct
899 * WinNT encoding. But at least eat the stupid three bytes up front.
902 unsigned char buf[4];
904 status = apr_file_read(file, buf, &len);
905 if ((status != APR_SUCCESS) || (len < 3)
906 || memcmp(buf, "\xEF\xBB\xBF", 3) != 0) {
908 apr_file_seek(file, APR_SET, &zero);
913 new_cfg = apr_palloc(p, sizeof(*new_cfg));
914 new_cfg->param = file;
915 new_cfg->name = apr_pstrdup(p, name);
916 new_cfg->getch = cfg_getch;
917 new_cfg->getstr = cfg_getstr;
918 new_cfg->close = cfg_close;
919 new_cfg->line_number = 0;
925 /* Allocate a ap_configfile_t handle with user defined functions and params */
926 AP_DECLARE(ap_configfile_t *) ap_pcfg_open_custom(
927 apr_pool_t *p, const char *descr, void *param,
928 apr_status_t (*getc_func) (char *ch, void *param),
929 apr_status_t (*gets_func) (void *buf, apr_size_t bufsize, void *param),
930 apr_status_t (*close_func) (void *param))
932 ap_configfile_t *new_cfg = apr_palloc(p, sizeof(*new_cfg));
933 new_cfg->param = param;
934 new_cfg->name = descr;
935 new_cfg->getch = getc_func;
936 new_cfg->getstr = gets_func;
937 new_cfg->close = close_func;
938 new_cfg->line_number = 0;
942 /* Read one character from a configfile_t */
943 AP_DECLARE(apr_status_t) ap_cfg_getc(char *ch, ap_configfile_t *cfp)
945 apr_status_t rc = cfp->getch(ch, cfp->param);
946 if (rc == APR_SUCCESS && *ch == LF)
951 AP_DECLARE(const char *) ap_pcfg_strerror(apr_pool_t *p, ap_configfile_t *cfp,
954 char buf[MAX_STRING_LEN];
955 if (rc == APR_SUCCESS)
957 return apr_psprintf(p, "Error reading %s at line %d: %s",
958 cfp->name, cfp->line_number,
959 rc == APR_ENOSPC ? "Line too long"
960 : apr_strerror(rc, buf, sizeof(buf)));
963 /* Read one line from open ap_configfile_t, strip LF, increase line number */
964 /* If custom handler does not define a getstr() function, read char by char */
965 static apr_status_t ap_cfg_getline_core(char *buf, apr_size_t bufsize,
966 ap_configfile_t *cfp)
969 /* If a "get string" function is defined, use it */
970 if (cfp->getstr != NULL) {
973 apr_size_t cbufsize = bufsize;
977 rc = cfp->getstr(cbuf, cbufsize, cfp->param);
987 if (rc != APR_SUCCESS) {
992 * check for line continuation,
993 * i.e. match [^\\]\\[\r]\n only
997 if (cp > cbuf && cp[-1] == LF) {
999 if (cp > cbuf && cp[-1] == CR)
1001 if (cp > cbuf && cp[-1] == '\\') {
1004 * line continuation requested -
1005 * then remove backslash and continue
1007 cbufsize -= (cp-cbuf);
1012 else if (cp - buf >= bufsize - 1) {
1018 /* No "get string" function defined; read character by character */
1022 /* too small, assume caller is crazy */
1029 rc = cfp->getch(&c, cfp->param);
1030 if (rc == APR_EOF) {
1036 if (rc != APR_SUCCESS)
1040 /* check for line continuation */
1041 if (i > 0 && buf[i-1] == '\\') {
1049 else if (i >= bufsize - 2) {
1060 static int cfg_trim_line(char *buf)
1064 * Leading and trailing white space is eliminated completely
1067 while (apr_isspace(*start))
1069 /* blast trailing whitespace */
1070 end = &start[strlen(start)];
1071 while (--end >= start && apr_isspace(*end))
1073 /* Zap leading whitespace by shifting */
1075 memmove(buf, start, end - start + 2);
1076 #ifdef DEBUG_CFG_LINES
1077 ap_log_error(APLOG_MARK, APLOG_NOTICE, 0, NULL, APLOGNO(00555) "Read config: '%s'", buf);
1079 return end - start + 1;
1082 /* Read one line from open ap_configfile_t, strip LF, increase line number */
1083 /* If custom handler does not define a getstr() function, read char by char */
1084 AP_DECLARE(apr_status_t) ap_cfg_getline(char *buf, apr_size_t bufsize,
1085 ap_configfile_t *cfp)
1087 apr_status_t rc = ap_cfg_getline_core(buf, bufsize, cfp);
1088 if (rc == APR_SUCCESS)
1093 AP_DECLARE(apr_status_t) ap_varbuf_cfg_getline(struct ap_varbuf *vb,
1094 ap_configfile_t *cfp,
1102 if (vb->strlen == AP_VARBUF_UNKNOWN)
1103 vb->strlen = strlen(vb->buf);
1104 if (vb->avail - vb->strlen < 3) {
1105 new_len = vb->avail * 2;
1106 if (new_len > max_len)
1108 else if (new_len < 3)
1110 ap_varbuf_grow(vb, new_len);
1114 rc = ap_cfg_getline_core(vb->buf + vb->strlen, vb->avail - vb->strlen, cfp);
1115 if (rc == APR_ENOSPC || rc == APR_SUCCESS)
1116 vb->strlen += strlen(vb->buf + vb->strlen);
1117 if (rc != APR_ENOSPC)
1119 if (vb->avail >= max_len)
1121 new_len = vb->avail * 2;
1122 if (new_len > max_len)
1124 ap_varbuf_grow(vb, new_len);
1127 if (vb->strlen > max_len)
1129 if (rc == APR_SUCCESS)
1130 vb->strlen = cfg_trim_line(vb->buf);
1134 /* Size an HTTP header field list item, as separated by a comma.
1135 * The return value is a pointer to the beginning of the non-empty list item
1136 * within the original string (or NULL if there is none) and the address
1137 * of field is shifted to the next non-comma, non-whitespace character.
1138 * len is the length of the item excluding any beginning whitespace.
1140 AP_DECLARE(const char *) ap_size_list_item(const char **field, int *len)
1142 const unsigned char *ptr = (const unsigned char *)*field;
1143 const unsigned char *token;
1144 int in_qpair, in_qstr, in_com;
1146 /* Find first non-comma, non-whitespace byte */
1148 while (*ptr == ',' || apr_isspace(*ptr))
1153 /* Find the end of this item, skipping over dead bits */
1155 for (in_qpair = in_qstr = in_com = 0;
1156 *ptr && (in_qpair || in_qstr || in_com || *ptr != ',');
1164 case '\\': in_qpair = 1; /* quoted-pair */
1166 case '"' : if (!in_com) /* quoted string delim */
1169 case '(' : if (!in_qstr) /* comment (may nest) */
1172 case ')' : if (in_com) /* end comment */
1180 if ((*len = (ptr - token)) == 0) {
1181 *field = (const char *)ptr;
1185 /* Advance field pointer to the next non-comma, non-white byte */
1187 while (*ptr == ',' || apr_isspace(*ptr))
1190 *field = (const char *)ptr;
1191 return (const char *)token;
1194 /* Retrieve an HTTP header field list item, as separated by a comma,
1195 * while stripping insignificant whitespace and lowercasing anything not in
1196 * a quoted string or comment. The return value is a new string containing
1197 * the converted list item (or NULL if none) and the address pointed to by
1198 * field is shifted to the next non-comma, non-whitespace.
1200 AP_DECLARE(char *) ap_get_list_item(apr_pool_t *p, const char **field)
1202 const char *tok_start;
1203 const unsigned char *ptr;
1206 int addspace = 0, in_qpair = 0, in_qstr = 0, in_com = 0, tok_len = 0;
1208 /* Find the beginning and maximum length of the list item so that
1209 * we can allocate a buffer for the new string and reset the field.
1211 if ((tok_start = ap_size_list_item(field, &tok_len)) == NULL) {
1214 token = apr_palloc(p, tok_len + 1);
1216 /* Scan the token again, but this time copy only the good bytes.
1217 * We skip extra whitespace and any whitespace around a '=', '/',
1218 * or ';' and lowercase normal characters not within a comment,
1219 * quoted-string or quoted-pair.
1221 for (ptr = (const unsigned char *)tok_start, pos = (unsigned char *)token;
1222 *ptr && (in_qpair || in_qstr || in_com || *ptr != ',');
1231 case '\\': in_qpair = 1;
1237 case '"' : if (!in_com)
1244 case '(' : if (!in_qstr)
1251 case ')' : if (in_com)
1257 case '\t': if (addspace)
1259 if (in_com || in_qstr)
1266 case ';' : if (!(in_com || in_qstr))
1270 default : if (addspace == 1)
1272 *pos++ = (in_com || in_qstr) ? *ptr
1273 : apr_tolower(*ptr);
1284 /* Find an item in canonical form (lowercase, no extra spaces) within
1285 * an HTTP field value list. Returns 1 if found, 0 if not found.
1286 * This would be much more efficient if we stored header fields as
1287 * an array of list items as they are received instead of a plain string.
1289 AP_DECLARE(int) ap_find_list_item(apr_pool_t *p, const char *line,
1292 const unsigned char *pos;
1293 const unsigned char *ptr = (const unsigned char *)line;
1294 int good = 0, addspace = 0, in_qpair = 0, in_qstr = 0, in_com = 0;
1299 do { /* loop for each item in line's list */
1301 /* Find first non-comma, non-whitespace byte */
1303 while (*ptr == ',' || apr_isspace(*ptr))
1307 good = 1; /* until proven otherwise for this item */
1309 break; /* no items left and nothing good found */
1311 /* We skip extra whitespace and any whitespace around a '=', '/',
1312 * or ';' and lowercase normal characters not within a comment,
1313 * quoted-string or quoted-pair.
1315 for (pos = (const unsigned char *)tok;
1316 *ptr && (in_qpair || in_qstr || in_com || *ptr != ',');
1322 good = (*pos++ == *ptr);
1326 case '\\': in_qpair = 1;
1328 good = good && (*pos++ == ' ');
1329 good = good && (*pos++ == *ptr);
1332 case '"' : if (!in_com)
1335 good = good && (*pos++ == ' ');
1336 good = good && (*pos++ == *ptr);
1339 case '(' : if (!in_qstr)
1342 good = good && (*pos++ == ' ');
1343 good = good && (*pos++ == *ptr);
1346 case ')' : if (in_com)
1348 good = good && (*pos++ == *ptr);
1352 case '\t': if (addspace || !good)
1354 if (in_com || in_qstr)
1355 good = (*pos++ == *ptr);
1361 case ';' : if (!(in_com || in_qstr))
1363 good = good && (*pos++ == *ptr);
1365 default : if (!good)
1368 good = (*pos++ == ' ');
1369 if (in_com || in_qstr)
1370 good = good && (*pos++ == *ptr);
1372 good = good && (*pos++ == apr_tolower(*ptr));
1379 good = 0; /* not good if only a prefix was matched */
1381 } while (*ptr && !good);
1387 /* Retrieve a token, spacing over it and returning a pointer to
1388 * the first non-white byte afterwards. Note that these tokens
1389 * are delimited by semis and commas; and can also be delimited
1390 * by whitespace at the caller's option.
1393 AP_DECLARE(char *) ap_get_token(apr_pool_t *p, const char **accept_line,
1396 const char *ptr = *accept_line;
1397 const char *tok_start;
1401 /* Find first non-white byte */
1403 while (*ptr && apr_isspace(*ptr))
1408 /* find token end, skipping over quoted strings.
1409 * (comments are already gone).
1412 while (*ptr && (accept_white || !apr_isspace(*ptr))
1413 && *ptr != ';' && *ptr != ',') {
1420 tok_len = ptr - tok_start;
1421 token = apr_pstrndup(p, tok_start, tok_len);
1423 /* Advance accept_line pointer to the next non-white byte */
1425 while (*ptr && apr_isspace(*ptr))
1433 /* find http tokens, see the definition of token from RFC2068 */
1434 AP_DECLARE(int) ap_find_token(apr_pool_t *p, const char *line, const char *tok)
1436 const unsigned char *start_token;
1437 const unsigned char *s;
1442 s = (const unsigned char *)line;
1444 /* find start of token, skip all stop characters, note NUL
1445 * isn't a token stop, so we don't need to test for it
1447 while (TEST_CHAR(*s, T_HTTP_TOKEN_STOP)) {
1454 /* find end of the token */
1455 while (*s && !TEST_CHAR(*s, T_HTTP_TOKEN_STOP)) {
1458 if (!strncasecmp((const char *)start_token, (const char *)tok,
1469 AP_DECLARE(int) ap_find_last_token(apr_pool_t *p, const char *line,
1472 int llen, tlen, lidx;
1477 llen = strlen(line);
1482 (lidx > 0 && !(apr_isspace(line[lidx - 1]) || line[lidx - 1] == ',')))
1485 return (strncasecmp(&line[lidx], tok, tlen) == 0);
1488 AP_DECLARE(char *) ap_escape_shell_cmd(apr_pool_t *p, const char *str)
1492 const unsigned char *s;
1494 cmd = apr_palloc(p, 2 * strlen(str) + 1); /* Be safe */
1495 d = (unsigned char *)cmd;
1496 s = (const unsigned char *)str;
1499 #if defined(OS2) || defined(WIN32)
1501 * Newlines to Win32/OS2 CreateProcess() are ill advised.
1502 * Convert them to spaces since they are effectively white
1503 * space to most applications
1505 if (*s == '\r' || *s == '\n') {
1511 if (TEST_CHAR(*s, T_ESCAPE_SHELL_CMD)) {
1521 static char x2c(const char *what)
1523 register char digit;
1525 #if !APR_CHARSET_EBCDIC
1526 digit = ((what[0] >= 'A') ? ((what[0] & 0xdf) - 'A') + 10
1529 digit += (what[1] >= 'A' ? ((what[1] & 0xdf) - 'A') + 10
1531 #else /*APR_CHARSET_EBCDIC*/
1538 digit = apr_xlate_conv_byte(ap_hdrs_from_ascii,
1539 0xFF & strtol(xstr, NULL, 16));
1540 #endif /*APR_CHARSET_EBCDIC*/
1545 * Unescapes a URL, leaving reserved characters intact.
1546 * Returns 0 on success, non-zero on error
1548 * bad % escape returns HTTP_BAD_REQUEST
1550 * decoding %00 or a forbidden character returns HTTP_NOT_FOUND
1553 static int unescape_url(char *url, const char *forbid, const char *reserved)
1555 register int badesc, badpath;
1560 /* Initial scan for first '%'. Don't bother writing values before
1562 y = strchr(url, '%');
1566 for (x = y; *y; ++x, ++y) {
1571 if (!apr_isxdigit(*(y + 1)) || !apr_isxdigit(*(y + 2))) {
1577 decoded = x2c(y + 1);
1578 if ((decoded == '\0')
1579 || (forbid && ap_strchr_c(forbid, decoded))) {
1584 else if (reserved && ap_strchr_c(reserved, decoded)) {
1598 return HTTP_BAD_REQUEST;
1601 return HTTP_NOT_FOUND;
1607 AP_DECLARE(int) ap_unescape_url(char *url)
1610 return unescape_url(url, SLASHES, NULL);
1612 AP_DECLARE(int) ap_unescape_url_keep2f(char *url, int decode_slashes)
1614 /* AllowEncodedSlashes (corrected) */
1615 if (decode_slashes) {
1616 /* no chars reserved */
1617 return unescape_url(url, NULL, NULL);
1619 /* reserve (do not decode) encoded slashes */
1620 return unescape_url(url, NULL, SLASHES);
1624 /* IFDEF these out until they've been thought through.
1625 * Just a germ of an API extension for now
1627 AP_DECLARE(int) ap_unescape_url_proxy(char *url)
1629 /* leave RFC1738 reserved characters intact, * so proxied URLs
1630 * don't get mangled. Where does that leave encoded '&' ?
1632 return unescape_url(url, NULL, "/;?");
1634 AP_DECLARE(int) ap_unescape_url_reserved(char *url, const char *reserved)
1636 return unescape_url(url, NULL, reserved);
1640 AP_DECLARE(int) ap_unescape_urlencoded(char *query)
1644 /* replace plus with a space */
1646 for (slider = query; *slider; slider++) {
1647 if (*slider == '+') {
1653 /* unescape everything else */
1654 return unescape_url(query, NULL, NULL);
1657 AP_DECLARE(char *) ap_construct_server(apr_pool_t *p, const char *hostname,
1658 apr_port_t port, const request_rec *r)
1660 if (ap_is_default_port(port, r)) {
1661 return apr_pstrdup(p, hostname);
1664 return apr_psprintf(p, "%s:%u", hostname, port);
1668 AP_DECLARE(int) ap_unescape_all(char *url)
1670 return unescape_url(url, NULL, NULL);
1673 /* c2x takes an unsigned, and expects the caller has guaranteed that
1674 * 0 <= what < 256... which usually means that you have to cast to
1675 * unsigned char first, because (unsigned)(char)(x) first goes through
1676 * signed extension to an int before the unsigned cast.
1678 * The reason for this assumption is to assist gcc code generation --
1679 * the unsigned char -> unsigned extension is already done earlier in
1680 * both uses of this code, so there's no need to waste time doing it
1683 static const char c2x_table[] = "0123456789abcdef";
1685 static APR_INLINE unsigned char *c2x(unsigned what, unsigned char prefix,
1686 unsigned char *where)
1688 #if APR_CHARSET_EBCDIC
1689 what = apr_xlate_conv_byte(ap_hdrs_to_ascii, (unsigned char)what);
1690 #endif /*APR_CHARSET_EBCDIC*/
1692 *where++ = c2x_table[what >> 4];
1693 *where++ = c2x_table[what & 0xf];
1698 * escape_path_segment() escapes a path segment, as defined in RFC 1808. This
1699 * routine is (should be) OS independent.
1701 * os_escape_path() converts an OS path to a URL, in an OS dependent way. In all
1702 * cases if a ':' occurs before the first '/' in the URL, the URL should be
1703 * prefixed with "./" (or the ':' escaped). In the case of Unix, this means
1704 * leaving '/' alone, but otherwise doing what escape_path_segment() does. For
1705 * efficiency reasons, we don't use escape_path_segment(), which is provided for
1706 * reference. Again, RFC 1808 is where this stuff is defined.
1708 * If partial is set, os_escape_path() assumes that the path will be appended to
1709 * something with a '/' in it (and thus does not prefix "./").
1712 AP_DECLARE(char *) ap_escape_path_segment_buffer(char *copy, const char *segment)
1714 const unsigned char *s = (const unsigned char *)segment;
1715 unsigned char *d = (unsigned char *)copy;
1719 if (TEST_CHAR(c, T_ESCAPE_PATH_SEGMENT)) {
1731 AP_DECLARE(char *) ap_escape_path_segment(apr_pool_t *p, const char *segment)
1733 return ap_escape_path_segment_buffer(apr_palloc(p, 3 * strlen(segment) + 1), segment);
1736 AP_DECLARE(char *) ap_os_escape_path(apr_pool_t *p, const char *path, int partial)
1738 char *copy = apr_palloc(p, 3 * strlen(path) + 3);
1739 const unsigned char *s = (const unsigned char *)path;
1740 unsigned char *d = (unsigned char *)copy;
1744 const char *colon = ap_strchr_c(path, ':');
1745 const char *slash = ap_strchr_c(path, '/');
1747 if (colon && (!slash || colon < slash)) {
1753 if (TEST_CHAR(c, T_OS_ESCAPE_PATH)) {
1765 AP_DECLARE(char *) ap_escape_urlencoded_buffer(char *copy, const char *buffer)
1767 const unsigned char *s = (const unsigned char *)buffer;
1768 unsigned char *d = (unsigned char *)copy;
1772 if (TEST_CHAR(c, T_ESCAPE_URLENCODED)) {
1775 else if (c == ' ') {
1787 AP_DECLARE(char *) ap_escape_urlencoded(apr_pool_t *p, const char *buffer)
1789 return ap_escape_urlencoded_buffer(apr_palloc(p, 3 * strlen(buffer) + 1), buffer);
1792 /* ap_escape_uri is now a macro for os_escape_path */
1794 AP_DECLARE(char *) ap_escape_html2(apr_pool_t *p, const char *s, int toasc)
1799 /* first, count the number of extra characters */
1800 for (i = 0, j = 0; s[i] != '\0'; i++)
1801 if (s[i] == '<' || s[i] == '>')
1803 else if (s[i] == '&')
1805 else if (s[i] == '"')
1807 else if (toasc && !apr_isascii(s[i]))
1811 return apr_pstrmemdup(p, s, i);
1813 x = apr_palloc(p, i + j + 1);
1814 for (i = 0, j = 0; s[i] != '\0'; i++, j++)
1816 memcpy(&x[j], "<", 4);
1819 else if (s[i] == '>') {
1820 memcpy(&x[j], ">", 4);
1823 else if (s[i] == '&') {
1824 memcpy(&x[j], "&", 5);
1827 else if (s[i] == '"') {
1828 memcpy(&x[j], """, 6);
1831 else if (toasc && !apr_isascii(s[i])) {
1832 char *esc = apr_psprintf(p, "&#%3.3d;", (unsigned char)s[i]);
1833 memcpy(&x[j], esc, 6);
1842 AP_DECLARE(char *) ap_escape_logitem(apr_pool_t *p, const char *str)
1846 const unsigned char *s;
1852 ret = apr_palloc(p, 4 * strlen(str) + 1); /* Be safe */
1853 d = (unsigned char *)ret;
1854 s = (const unsigned char *)str;
1857 if (TEST_CHAR(*s, T_ESCAPE_LOGITEM)) {
1893 AP_DECLARE(apr_size_t) ap_escape_errorlog_item(char *dest, const char *source,
1896 unsigned char *d, *ep;
1897 const unsigned char *s;
1899 if (!source || !buflen) { /* be safe */
1903 d = (unsigned char *)dest;
1904 s = (const unsigned char *)source;
1905 ep = d + buflen - 1;
1907 for (; d < ep && *s; ++s) {
1909 if (TEST_CHAR(*s, T_ESCAPE_LOGITEM)) {
1935 case '"': /* no need for this in error log */
1940 ep = --d; /* break the for loop as well */
1953 return (d - (unsigned char *)dest);
1956 AP_DECLARE(int) ap_is_directory(apr_pool_t *p, const char *path)
1960 if (apr_stat(&finfo, path, APR_FINFO_TYPE, p) != APR_SUCCESS)
1961 return 0; /* in error condition, just return no */
1963 return (finfo.filetype == APR_DIR);
1966 AP_DECLARE(int) ap_is_rdirectory(apr_pool_t *p, const char *path)
1970 if (apr_stat(&finfo, path, APR_FINFO_LINK | APR_FINFO_TYPE, p) != APR_SUCCESS)
1971 return 0; /* in error condition, just return no */
1973 return (finfo.filetype == APR_DIR);
1976 AP_DECLARE(char *) ap_make_full_path(apr_pool_t *a, const char *src1,
1979 apr_size_t len1, len2;
1982 len1 = strlen(src1);
1983 len2 = strlen(src2);
1984 /* allocate +3 for '/' delimiter, trailing NULL and overallocate
1985 * one extra byte to allow the caller to add a trailing '/'
1987 path = (char *)apr_palloc(a, len1 + len2 + 3);
1990 memcpy(path + 1, src2, len2 + 1);
1994 memcpy(path, src1, len1);
1996 if (next[-1] != '/') {
1999 memcpy(next, src2, len2 + 1);
2005 * Check for an absoluteURI syntax (see section 3.2 in RFC2068).
2007 AP_DECLARE(int) ap_is_url(const char *u)
2011 for (x = 0; u[x] != ':'; x++) {
2013 ((!apr_isalpha(u[x])) && (!apr_isdigit(u[x])) &&
2014 (u[x] != '+') && (u[x] != '-') && (u[x] != '.'))) {
2019 return (x ? 1 : 0); /* If the first character is ':', it's broken, too */
2022 AP_DECLARE(int) ap_ind(const char *s, char c)
2024 const char *p = ap_strchr_c(s, c);
2031 AP_DECLARE(int) ap_rind(const char *s, char c)
2033 const char *p = ap_strrchr_c(s, c);
2040 AP_DECLARE(void) ap_str_tolower(char *str)
2043 *str = apr_tolower(*str);
2048 AP_DECLARE(void) ap_str_toupper(char *str)
2051 *str = apr_toupper(*str);
2057 * We must return a FQDN
2059 char *ap_get_local_host(apr_pool_t *a)
2061 #ifndef MAXHOSTNAMELEN
2062 #define MAXHOSTNAMELEN 256
2064 char str[MAXHOSTNAMELEN + 1];
2065 char *server_hostname = NULL;
2066 apr_sockaddr_t *sockaddr;
2069 if (apr_gethostname(str, sizeof(str) - 1, a) != APR_SUCCESS) {
2070 ap_log_perror(APLOG_MARK, APLOG_STARTUP | APLOG_WARNING, 0, a, APLOGNO(00556)
2071 "%s: apr_gethostname() failed to determine ServerName",
2074 str[sizeof(str) - 1] = '\0';
2075 if (apr_sockaddr_info_get(&sockaddr, str, APR_UNSPEC, 0, 0, a) == APR_SUCCESS) {
2076 if ( (apr_getnameinfo(&hostname, sockaddr, 0) == APR_SUCCESS) &&
2077 (ap_strchr_c(hostname, '.')) ) {
2078 server_hostname = apr_pstrdup(a, hostname);
2079 return server_hostname;
2080 } else if (ap_strchr_c(str, '.')) {
2081 server_hostname = apr_pstrdup(a, str);
2083 apr_sockaddr_ip_get(&hostname, sockaddr);
2084 server_hostname = apr_pstrdup(a, hostname);
2087 ap_log_perror(APLOG_MARK, APLOG_STARTUP | APLOG_WARNING, 0, a, APLOGNO(00557)
2088 "%s: apr_sockaddr_info_get() failed for %s",
2089 ap_server_argv0, str);
2093 if (!server_hostname)
2094 server_hostname = apr_pstrdup(a, "127.0.0.1");
2096 ap_log_perror(APLOG_MARK, APLOG_ALERT|APLOG_STARTUP, 0, a, APLOGNO(00558)
2097 "%s: Could not reliably determine the server's fully qualified "
2098 "domain name, using %s. Set the 'ServerName' directive globally "
2099 "to suppress this message",
2100 ap_server_argv0, server_hostname);
2102 return server_hostname;
2105 /* simple 'pool' alloc()ing glue to apr_base64.c
2107 AP_DECLARE(char *) ap_pbase64decode(apr_pool_t *p, const char *bufcoded)
2112 decoded = (char *) apr_palloc(p, 1 + apr_base64_decode_len(bufcoded));
2113 l = apr_base64_decode(decoded, bufcoded);
2114 decoded[l] = '\0'; /* make binary sequence into string */
2119 AP_DECLARE(char *) ap_pbase64encode(apr_pool_t *p, char *string)
2122 int l = strlen(string);
2124 encoded = (char *) apr_palloc(p, 1 + apr_base64_encode_len(l));
2125 l = apr_base64_encode(encoded, string, l);
2126 encoded[l] = '\0'; /* make binary sequence into string */
2131 /* we want to downcase the type/subtype for comparison purposes
2132 * but nothing else because ;parameter=foo values are case sensitive.
2133 * XXX: in truth we want to downcase parameter names... but really,
2134 * apache has never handled parameters and such correctly. You
2135 * also need to compress spaces and such to be able to compare
2138 AP_DECLARE(void) ap_content_type_tolower(char *str)
2142 semi = strchr(str, ';');
2147 ap_str_tolower(str);
2155 * Given a string, replace any bare " with \" .
2157 AP_DECLARE(char *) ap_escape_quotes(apr_pool_t *p, const char *instring)
2160 const char *inchr = instring;
2161 char *outchr, *outstring;
2164 * Look through the input string, jogging the length of the output
2165 * string up by an extra byte each time we find an unescaped ".
2167 while (*inchr != '\0') {
2169 if (*inchr == '"') {
2173 * If we find a slosh, and it's not the last byte in the string,
2174 * it's escaping something - advance past both bytes.
2176 if ((*inchr == '\\') && (inchr[1] != '\0')) {
2182 outstring = apr_palloc(p, newlen + 1);
2186 * Now copy the input string to the output string, inserting a slosh
2187 * in front of every " that doesn't already have one.
2189 while (*inchr != '\0') {
2190 if ((*inchr == '\\') && (inchr[1] != '\0')) {
2191 *outchr++ = *inchr++;
2192 *outchr++ = *inchr++;
2194 if (*inchr == '"') {
2197 if (*inchr != '\0') {
2198 *outchr++ = *inchr++;
2206 * Given a string, append the PID deliminated by delim.
2207 * Usually used to create a pid-appended filepath name
2208 * (eg: /a/b/foo -> /a/b/foo.6726). A function, and not
2209 * a macro, to avoid unistd.h dependency
2211 AP_DECLARE(char *) ap_append_pid(apr_pool_t *p, const char *string,
2214 return apr_psprintf(p, "%s%s%" APR_PID_T_FMT, string,
2220 * Parse a given timeout parameter string into an apr_interval_time_t value.
2221 * The unit of the time interval is given as postfix string to the numeric
2222 * string. Currently the following units are understood:
2229 * If no unit is contained in the given timeout parameter the default_time_unit
2230 * will be used instead.
2231 * @param timeout_parameter The string containing the timeout parameter.
2232 * @param timeout The timeout value to be returned.
2233 * @param default_time_unit The default time unit to use if none is specified
2234 * in timeout_parameter.
2235 * @return Status value indicating whether the parsing was successful or not.
2237 AP_DECLARE(apr_status_t) ap_timeout_parameter_parse(
2238 const char *timeout_parameter,
2239 apr_interval_time_t *timeout,
2240 const char *default_time_unit)
2243 const char *time_str;
2246 tout = apr_strtoi64(timeout_parameter, &endp, 10);
2250 if (!endp || !*endp) {
2251 time_str = default_time_unit;
2257 switch (*time_str) {
2258 /* Time is in seconds */
2260 *timeout = (apr_interval_time_t) apr_time_from_sec(tout);
2263 /* Time is in hours */
2264 *timeout = (apr_interval_time_t) apr_time_from_sec(tout * 3600);
2267 switch (*(++time_str)) {
2268 /* Time is in milliseconds */
2270 *timeout = (apr_interval_time_t) tout * 1000;
2272 /* Time is in minutes */
2274 *timeout = (apr_interval_time_t) apr_time_from_sec(tout * 60);
2277 return APR_EGENERAL;
2281 return APR_EGENERAL;
2287 * Determine if a request has a request body or not.
2289 * @param r the request_rec of the request
2290 * @return truth value
2292 AP_DECLARE(int) ap_request_has_body(request_rec *r)
2299 has_body = (!r->header_only
2301 || apr_table_get(r->headers_in, "Transfer-Encoding")
2302 || ( (cls = apr_table_get(r->headers_in, "Content-Length"))
2303 && (apr_strtoff(&cl, cls, &estr, 10) == APR_SUCCESS)
2311 AP_DECLARE_NONSTD(apr_status_t) ap_pool_cleanup_set_null(void *data_)
2313 void **ptr = (void **)data_;
2318 AP_DECLARE(apr_status_t) ap_str2_alnum(const char *src, char *dest) {
2320 for ( ; *src; src++, dest++)
2322 if (!apr_isprint(*src))
2324 else if (!apr_isalnum(*src))
2334 AP_DECLARE(apr_status_t) ap_pstr2_alnum(apr_pool_t *p, const char *src,
2337 char *new = apr_palloc(p, strlen(src)+1);
2341 return ap_str2_alnum(src, new);
2345 * Read the body and parse any form found, which must be of the
2346 * type application/x-www-form-urlencoded.
2348 * Name/value pairs are returned in an array, with the names as
2349 * strings with a maximum length of HUGE_STRING_LEN, and the
2350 * values as bucket brigades. This allows values to be arbitrarily
2353 * All url-encoding is removed from both the names and the values
2354 * on the fly. The names are interpreted as strings, while the
2355 * values are interpreted as blocks of binary data, that may
2356 * contain the 0 character.
2358 * In order to ensure that resource limits are not exceeded, a
2359 * maximum size must be provided. If the sum of the lengths of
2360 * the names and the values exceed this size, this function
2361 * will return HTTP_REQUEST_ENTITY_TOO_LARGE.
2363 * An optional number of parameters can be provided, if the number
2364 * of parameters provided exceeds this amount, this function will
2365 * return HTTP_REQUEST_ENTITY_TOO_LARGE. If this value is negative,
2366 * no limit is imposed, and the number of parameters is in turn
2367 * constrained by the size parameter above.
2369 * This function honours any kept_body configuration, and the
2370 * original raw request body will be saved to the kept_body brigade
2371 * if so configured, just as ap_discard_request_body does.
2373 * NOTE: File upload is not yet supported, but can be without change
2374 * to the function call.
2377 /* form parsing stuff */
2388 AP_DECLARE(int) ap_parse_form_data(request_rec *r, ap_filter_t *f,
2389 apr_array_header_t **ptr,
2390 apr_size_t num, apr_size_t usize)
2392 apr_bucket_brigade *bb = NULL;
2394 char buffer[HUGE_STRING_LEN + 1];
2396 apr_size_t offset = 0;
2398 ap_form_type_t state = FORM_NAME, percent = FORM_NORMAL;
2399 ap_form_pair_t *pair = NULL;
2400 apr_array_header_t *pairs = apr_array_make(r->pool, 4, sizeof(ap_form_pair_t));
2407 /* sanity check - we only support forms for now */
2408 ct = apr_table_get(r->headers_in, "Content-Type");
2409 if (!ct || strcmp("application/x-www-form-urlencoded", ct)) {
2410 return ap_discard_request_body(r);
2413 if (usize > APR_SIZE_MAX >> 1)
2414 size = APR_SIZE_MAX >> 1;
2419 f = r->input_filters;
2422 bb = apr_brigade_create(r->pool, r->connection->bucket_alloc);
2424 apr_bucket *bucket = NULL, *last = NULL;
2426 int rv = ap_get_brigade(f, bb, AP_MODE_READBYTES,
2427 APR_BLOCK_READ, HUGE_STRING_LEN);
2428 if (rv != APR_SUCCESS) {
2429 apr_brigade_destroy(bb);
2430 return (rv == AP_FILTER_ERROR) ? rv : HTTP_BAD_REQUEST;
2433 for (bucket = APR_BRIGADE_FIRST(bb);
2434 bucket != APR_BRIGADE_SENTINEL(bb);
2435 last = bucket, bucket = APR_BUCKET_NEXT(bucket)) {
2437 apr_size_t len, slide;
2440 apr_bucket_delete(last);
2442 if (APR_BUCKET_IS_EOS(bucket)) {
2446 if (bucket->length == 0) {
2450 rv = apr_bucket_read(bucket, &data, &len, APR_BLOCK_READ);
2451 if (rv != APR_SUCCESS) {
2452 apr_brigade_destroy(bb);
2453 return HTTP_BAD_REQUEST;
2457 while (state != FORM_ABORT && slide-- > 0 && size >= 0 && num != 0) {
2462 else if ('&' == c) {
2466 percent = FORM_PERCENTA;
2469 if (FORM_PERCENTA == percent) {
2473 else if (c >= 'A') {
2476 else if (c >= '0') {
2480 percent = FORM_PERCENTB;
2483 if (FORM_PERCENTB == percent) {
2487 else if (c >= 'A') {
2490 else if (c >= '0') {
2494 percent = FORM_NORMAL;
2499 const char *tmp = apr_pmemdup(r->pool, buffer, offset);
2500 apr_bucket *b = apr_bucket_pool_create(tmp, offset, r->pool, r->connection->bucket_alloc);
2501 APR_BRIGADE_INSERT_TAIL(pair->value, b);
2509 if (offset < HUGE_STRING_LEN) {
2513 pair = (ap_form_pair_t *) apr_array_push(pairs);
2514 pair->name = apr_pstrdup(r->pool, buffer);
2515 pair->value = apr_brigade_create(r->pool, r->connection->bucket_alloc);
2519 buffer[offset++] = c;
2528 if (offset >= HUGE_STRING_LEN) {
2529 const char *tmp = apr_pmemdup(r->pool, buffer, offset);
2530 apr_bucket *b = apr_bucket_pool_create(tmp, offset, r->pool, r->connection->bucket_alloc);
2531 APR_BRIGADE_INSERT_TAIL(pair->value, b);
2534 buffer[offset++] = c;
2544 apr_brigade_cleanup(bb);
2545 } while (!seen_eos);
2547 if (FORM_ABORT == state || size < 0 || num == 0) {
2548 return HTTP_REQUEST_ENTITY_TOO_LARGE;
2550 else if (FORM_VALUE == state && pair && offset > 0) {
2551 const char *tmp = apr_pmemdup(r->pool, buffer, offset);
2552 apr_bucket *b = apr_bucket_pool_create(tmp, offset, r->pool, r->connection->bucket_alloc);
2553 APR_BRIGADE_INSERT_TAIL(pair->value, b);
2560 #define VARBUF_SMALL_SIZE 2048
2561 #define VARBUF_MAX_SIZE (APR_SIZE_MAX - 1 - \
2562 APR_ALIGN_DEFAULT(sizeof(struct ap_varbuf_info)))
2564 struct ap_varbuf_info {
2565 struct apr_memnode_t *node;
2566 apr_allocator_t *allocator;
2569 static apr_status_t varbuf_cleanup(void *info_)
2571 struct ap_varbuf_info *info = info_;
2572 info->node->next = NULL;
2573 apr_allocator_free(info->allocator, info->node);
2577 const char nul = '\0';
2578 static char * const varbuf_empty = (char *)&nul;
2580 AP_DECLARE(void) ap_varbuf_init(apr_pool_t *p, struct ap_varbuf *vb,
2581 apr_size_t init_size)
2583 vb->buf = varbuf_empty;
2585 vb->strlen = AP_VARBUF_UNKNOWN;
2589 ap_varbuf_grow(vb, init_size);
2592 AP_DECLARE(void) ap_varbuf_grow(struct ap_varbuf *vb, apr_size_t new_len)
2594 apr_memnode_t *new_node = NULL;
2595 apr_allocator_t *allocator;
2596 struct ap_varbuf_info *new_info;
2599 AP_DEBUG_ASSERT(vb->strlen == AP_VARBUF_UNKNOWN || vb->avail >= vb->strlen);
2601 if (new_len <= vb->avail)
2604 if (new_len < 2 * vb->avail && vb->avail < VARBUF_MAX_SIZE/2) {
2605 /* at least double the size, to avoid repeated reallocations */
2606 new_len = 2 * vb->avail;
2608 else if (new_len > VARBUF_MAX_SIZE) {
2609 apr_abortfunc_t abort_fn = apr_pool_abort_get(vb->pool);
2610 ap_assert(abort_fn != NULL);
2611 abort_fn(APR_ENOMEM);
2615 new_len++; /* add space for trailing \0 */
2616 if (new_len <= VARBUF_SMALL_SIZE) {
2617 new_len = APR_ALIGN_DEFAULT(new_len);
2618 new = apr_palloc(vb->pool, new_len);
2619 if (vb->avail && vb->strlen != 0) {
2620 AP_DEBUG_ASSERT(vb->buf != NULL);
2621 AP_DEBUG_ASSERT(vb->buf != varbuf_empty);
2622 if (new == vb->buf + vb->avail + 1) {
2623 /* We are lucky: the new memory lies directly after our old
2624 * buffer, we can now use both.
2626 vb->avail += new_len;
2630 /* copy up to vb->strlen + 1 bytes */
2631 memcpy(new, vb->buf, vb->strlen == AP_VARBUF_UNKNOWN ?
2632 vb->avail + 1 : vb->strlen + 1);
2638 vb->avail = new_len - 1;
2643 /* The required block is rather larger. Use allocator directly so that
2644 * the memory can be freed independently from the pool. */
2645 allocator = apr_pool_allocator_get(vb->pool);
2646 if (new_len <= VARBUF_MAX_SIZE)
2647 new_node = apr_allocator_alloc(allocator,
2648 new_len + APR_ALIGN_DEFAULT(sizeof(*new_info)));
2650 apr_abortfunc_t abort_fn = apr_pool_abort_get(vb->pool);
2651 ap_assert(abort_fn != NULL);
2652 abort_fn(APR_ENOMEM);
2655 new_info = (struct ap_varbuf_info *)new_node->first_avail;
2656 new_node->first_avail += APR_ALIGN_DEFAULT(sizeof(*new_info));
2657 new_info->node = new_node;
2658 new_info->allocator = allocator;
2659 new = new_node->first_avail;
2660 AP_DEBUG_ASSERT(new_node->endp - new_node->first_avail >= new_len);
2661 new_len = new_node->endp - new_node->first_avail;
2663 if (vb->avail && vb->strlen != 0)
2664 memcpy(new, vb->buf, vb->strlen == AP_VARBUF_UNKNOWN ?
2665 vb->avail + 1 : vb->strlen + 1);
2669 apr_pool_cleanup_run(vb->pool, vb->info, varbuf_cleanup);
2670 apr_pool_cleanup_register(vb->pool, new_info, varbuf_cleanup,
2671 apr_pool_cleanup_null);
2672 vb->info = new_info;
2674 vb->avail = new_len - 1;
2677 AP_DECLARE(void) ap_varbuf_strmemcat(struct ap_varbuf *vb, const char *str,
2683 ap_varbuf_grow(vb, len);
2684 memcpy(vb->buf, str, len);
2685 vb->buf[len] = '\0';
2689 if (vb->strlen == AP_VARBUF_UNKNOWN)
2690 vb->strlen = strlen(vb->buf);
2691 ap_varbuf_grow(vb, vb->strlen + len);
2692 memcpy(vb->buf + vb->strlen, str, len);
2694 vb->buf[vb->strlen] = '\0';
2697 AP_DECLARE(void) ap_varbuf_free(struct ap_varbuf *vb)
2700 apr_pool_cleanup_run(vb->pool, vb->info, varbuf_cleanup);
2706 AP_DECLARE(char *) ap_varbuf_pdup(apr_pool_t *p, struct ap_varbuf *buf,
2707 const char *prepend, apr_size_t prepend_len,
2708 const char *append, apr_size_t append_len,
2709 apr_size_t *new_len)
2712 struct iovec vec[3];
2715 vec[i].iov_base = (void *)prepend;
2716 vec[i].iov_len = prepend_len;
2719 if (buf->avail && buf->strlen) {
2720 if (buf->strlen == AP_VARBUF_UNKNOWN)
2721 buf->strlen = strlen(buf->buf);
2722 vec[i].iov_base = (void *)buf->buf;
2723 vec[i].iov_len = buf->strlen;
2727 vec[i].iov_base = (void *)append;
2728 vec[i].iov_len = append_len;
2732 return apr_pstrcatv(p, vec, i, new_len);
2739 AP_DECLARE(apr_status_t) ap_varbuf_regsub(struct ap_varbuf *vb,
2743 ap_regmatch_t pmatch[],
2746 return regsub_core(NULL, NULL, vb, input, source, nmatch, pmatch, maxlen);
2749 static const char * const oom_message = "[crit] Memory allocation failed, "
2750 "aborting process." APR_EOL_STR;
2752 AP_DECLARE(void) ap_abort_on_oom()
2754 int written, count = strlen(oom_message);
2755 const char *buf = oom_message;
2757 written = write(STDERR_FILENO, buf, count);
2758 if (written == count)
2764 } while (written >= 0 || errno == EINTR);
2768 AP_DECLARE(void *) ap_malloc(size_t size)
2770 void *p = malloc(size);
2771 if (p == NULL && size != 0)
2776 AP_DECLARE(void *) ap_calloc(size_t nelem, size_t size)
2778 void *p = calloc(nelem, size);
2779 if (p == NULL && nelem != 0 && size != 0)
2784 AP_DECLARE(void *) ap_realloc(void *ptr, size_t size)
2786 void *p = realloc(ptr, size);
2787 if (p == NULL && size != 0)