-const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.77 2013/11/24 14:24:18 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
* Purpose : Declares functions to match URLs against URL
* patterns.
*
- * Copyright : Written by and Copyright (C) 2001-2011
+ * Copyright : Written by and Copyright (C) 2001-2014
* the Privoxy team. http://www.privoxy.org/
*
* Based on the Internet Junkbuster originally written
#include "miscutil.h"
#include "errlog.h"
-const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
-
enum regex_anchoring
{
NO_ANCHORING,
*/
"VERSION-CONTROL", "REPORT", "CHECKOUT", "CHECKIN", "UNCHECKOUT",
"MKWORKSPACE", "UPDATE", "LABEL", "MERGE", "BASELINE-CONTROL", "MKACTIVITY",
+ /*
+ * The PATCH method is defined by RFC5789, the format of the
+ * actual patch in the body depends on the application, but from
+ * Privoxy's point of view it doesn't matter.
+ */
+ "PATCH",
};
int i;
}
+/*********************************************************************
+ *
+ * Function : normalize_http_version
+ *
+ * Description : Take a supported HTTP version string and remove
+ * leading zeroes etc., reject unsupported versions.
+ *
+ * This is an explicit RFC 2616 (3.1) MUST and
+ * RFC 7230 mandates that intermediaries send their
+ * own HTTP-version in forwarded messages.
+ *
+ * Parameters :
+ * 1 : http_version = HTTP version string
+ *
+ * Returns : JB_ERR_OK on success
+ * JB_ERR_PARSE if the HTTP version is unsupported
+ *
+ *********************************************************************/
+static jb_err normalize_http_version(char *http_version)
+{
+ unsigned int major_version;
+ unsigned int minor_version;
+
+ if (2 != sscanf(http_version, "HTTP/%u.%u", &major_version, &minor_version))
+ {
+ log_error(LOG_LEVEL_ERROR, "Unsupported HTTP version: %s", http_version);
+ return JB_ERR_PARSE;
+ }
+
+ if (major_version != 1 || (minor_version != 0 && minor_version != 1))
+ {
+ log_error(LOG_LEVEL_ERROR, "The only supported HTTP "
+ "versions are 1.0 and 1.1. This rules out: %s", http_version);
+ return JB_ERR_PARSE;
+ }
+
+ assert(strlen(http_version) >= 8);
+ snprintf(http_version, 9, "HTTP/%u.%u", major_version, minor_version);
+
+ return JB_ERR_OK;
+
+}
+
+
/*********************************************************************
*
* Function : parse_http_request
jb_err parse_http_request(const char *req, struct http_request *http)
{
char *buf;
- char *v[10]; /* XXX: Why 10? We should only need three. */
+ char *v[3];
int n;
jb_err err;
return JB_ERR_PARSE;
}
- if (strcmpic(v[2], "HTTP/1.1") && strcmpic(v[2], "HTTP/1.0"))
+ if (JB_ERR_OK != normalize_http_version(v[2]))
{
- log_error(LOG_LEVEL_ERROR, "The only supported HTTP "
- "versions are 1.0 and 1.1. This rules out: %s", v[2]);
freez(buf);
return JB_ERR_PARSE;
}
http->cmd = strdup_or_die(req);
http->gpc = strdup_or_die(v[0]);
http->ver = strdup_or_die(v[2]);
+ http->ocmd = strdup_or_die(http->cmd);
freez(buf);
* 4 : regex = Where the compiled regex should be stored.
*
* Returns : JB_ERR_OK - Success
- * JB_ERR_MEMORY - Out of memory
* JB_ERR_PARSE - Cannot parse regex
*
*********************************************************************/
struct pattern_spec *url, regex_t **regex)
{
int errcode;
- char rebuf[BUFFER_SIZE];
const char *fmt = NULL;
+ char *rebuf;
+ size_t rebuf_size;
assert(pattern);
- assert(strlen(pattern) < sizeof(rebuf) - 2);
if (pattern[0] == '\0')
{
log_error(LOG_LEVEL_FATAL,
"Invalid anchoring in compile_pattern %d", anchoring);
}
+ rebuf_size = strlen(pattern) + strlen(fmt);
+ rebuf = malloc_or_die(rebuf_size);
+ *regex = zalloc_or_die(sizeof(**regex));
- *regex = zalloc(sizeof(**regex));
- if (NULL == *regex)
- {
- free_pattern_spec(url);
- return JB_ERR_MEMORY;
- }
-
- snprintf(rebuf, sizeof(rebuf), fmt, pattern);
+ snprintf(rebuf, rebuf_size, fmt, pattern);
errcode = regcomp(*regex, rebuf, (REG_EXTENDED|REG_NOSUB|REG_ICASE));
if (errcode)
{
- size_t errlen = regerror(errcode, *regex, rebuf, sizeof(rebuf));
- if (errlen > (sizeof(rebuf) - (size_t)1))
+ size_t errlen = regerror(errcode, *regex, rebuf, rebuf_size);
+ if (errlen > (rebuf_size - (size_t)1))
{
- errlen = sizeof(rebuf) - (size_t)1;
+ errlen = rebuf_size - (size_t)1;
}
rebuf[errlen] = '\0';
log_error(LOG_LEVEL_ERROR, "error compiling %s from %s: %s",
pattern, url->spec, rebuf);
free_pattern_spec(url);
+ freez(rebuf);
return JB_ERR_PARSE;
}
+ freez(rebuf);
return JB_ERR_OK;
*/
if (host_pattern[strlen(host_pattern) - 1] == '.')
{
- url->unanchored |= ANCHOR_RIGHT;
+ url->pattern.url_spec.unanchored |= ANCHOR_RIGHT;
}
if (host_pattern[0] == '.')
{
- url->unanchored |= ANCHOR_LEFT;
+ url->pattern.url_spec.unanchored |= ANCHOR_LEFT;
}
/*
* Split domain into components
*/
- url->dbuffer = strdup_or_die(host_pattern);
+ url->pattern.url_spec.dbuffer = strdup_or_die(host_pattern);
/*
* Map to lower case
*/
- for (p = url->dbuffer; *p ; p++)
+ for (p = url->pattern.url_spec.dbuffer; *p ; p++)
{
*p = (char)privoxy_tolower(*p);
}
/*
* Split the domain name into components
*/
- url->dcount = ssplit(url->dbuffer, ".", v, SZ(v));
+ url->pattern.url_spec.dcount = ssplit(url->pattern.url_spec.dbuffer, ".", v, SZ(v));
- if (url->dcount < 0)
+ if (url->pattern.url_spec.dcount < 0)
{
free_pattern_spec(url);
return JB_ERR_PARSE;
}
- else if (url->dcount != 0)
+ else if (url->pattern.url_spec.dcount != 0)
{
/*
* Save a copy of the pointers in dvec
*/
- size = (size_t)url->dcount * sizeof(*url->dvec);
+ size = (size_t)url->pattern.url_spec.dcount * sizeof(*url->pattern.url_spec.dvec);
- url->dvec = malloc_or_die(size);
+ url->pattern.url_spec.dvec = malloc_or_die(size);
- memcpy(url->dvec, v, size);
+ memcpy(url->pattern.url_spec.dvec, v, size);
}
/*
* else dcount == 0 in which case we needn't do anything,
* Function : domain_match
*
* Description : Domain-wise Compare fqdn's. Governed by the bimap in
- * pattern->unachored, the comparison is un-, left-,
+ * p.pattern->unachored, the comparison is un-, left-,
* right-anchored, or both.
* The individual domain names are compared with
* simplematch().
*
* Parameters :
- * 1 : pattern = a domain that may contain a '*' as a wildcard.
+ * 1 : p = a domain that may contain a '*' as a wildcard.
* 2 : fqdn = domain name against which the patterns are compared.
*
* Returns : 0 => domains are equivalent, else no match.
*
*********************************************************************/
-static int domain_match(const struct pattern_spec *pattern, const struct http_request *fqdn)
+static int domain_match(const struct pattern_spec *p, const struct http_request *fqdn)
{
char **pv, **fv; /* vectors */
int plen, flen;
- int unanchored = pattern->unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
+ int unanchored = p->pattern.url_spec.unanchored & (ANCHOR_RIGHT | ANCHOR_LEFT);
- plen = pattern->dcount;
+ plen = p->pattern.url_spec.dcount;
flen = fqdn->dcount;
if (flen < plen)
return 1;
}
- pv = pattern->dvec;
+ pv = p->pattern.url_spec.dvec;
fv = fqdn->dvec;
if (unanchored == ANCHOR_LEFT)
*********************************************************************/
jb_err create_pattern_spec(struct pattern_spec *pattern, char *buf)
{
+ static const struct
+ {
+ /** The tag pattern prefix to match */
+ const char *prefix;
+
+ /** The length of the prefix to match */
+ const size_t prefix_length;
+
+ /** The pattern flag */
+ const unsigned flag;
+ } tag_pattern[] = {
+ { "TAG:", 4, PATTERN_SPEC_TAG_PATTERN},
+ #ifdef FEATURE_CLIENT_TAGS
+ { "CLIENT-TAG:", 11, PATTERN_SPEC_CLIENT_TAG_PATTERN},
+ #endif
+ { "NO-REQUEST-TAG:", 15, PATTERN_SPEC_NO_REQUEST_TAG_PATTERN},
+ { "NO-RESPONSE-TAG:", 16, PATTERN_SPEC_NO_RESPONSE_TAG_PATTERN}
+ };
+ int i;
+
assert(pattern);
assert(buf);
/* Remember the original specification for the CGI pages. */
pattern->spec = strdup_or_die(buf);
- /* Is it a positive tag pattern? */
- if (0 == strncmpic(pattern->spec, "TAG:", 4))
- {
- /* The pattern starts with the first character after "TAG:" */
- const char *tag_pattern = buf + 4;
- pattern->flags |= PATTERN_SPEC_TAG_PATTERN;
- return compile_pattern(tag_pattern, NO_ANCHORING, pattern, &pattern->pattern.tag_regex);
- }
- /* Is it a negative tag pattern? */
- if (0 == strncmpic(pattern->spec, "NO-REQUEST-TAG:", 15))
+ /* Check if it's a tag pattern */
+ for (i = 0; i < SZ(tag_pattern); i++)
{
- /* The pattern starts with the first character after "NO-REQUEST-TAG:" */
- const char *tag_pattern = buf + 15;
- pattern->flags |= PATTERN_SPEC_NO_REQUEST_TAG_PATTERN;
- return compile_pattern(tag_pattern, NO_ANCHORING, pattern, &pattern->pattern.tag_regex);
- }
- if (0 == strncmpic(pattern->spec, "NO-RESPONSE-TAG:", 16))
- {
- /* The pattern starts with the first character after "NO-RESPONSE-TAG:" */
- const char *tag_pattern = buf + 16;
- pattern->flags |= PATTERN_SPEC_NO_RESPONSE_TAG_PATTERN;
- return compile_pattern(tag_pattern, NO_ANCHORING, pattern, &pattern->pattern.tag_regex);
+ if (0 == strncmpic(pattern->spec, tag_pattern[i].prefix, tag_pattern[i].prefix_length))
+ {
+ /* The regex starts after the prefix */
+ const char *tag_regex = buf + tag_pattern[i].prefix_length;
+
+ pattern->flags |= tag_pattern[i].flag;
+
+ return compile_pattern(tag_regex, NO_ANCHORING, pattern,
+ &pattern->pattern.tag_regex);
+ }
}
+ /* If it isn't a tag pattern it must be an URL pattern. */
pattern->flags |= PATTERN_SPEC_URL_PATTERN;
- /* If it isn't a tag pattern it must be an URL pattern. */
return compile_url_pattern(pattern, buf);
+
}
static int host_matches(const struct http_request *http,
const struct pattern_spec *pattern)
{
+ assert(http->host != NULL);
#ifdef FEATURE_EXTENDED_HOST_PATTERNS
return ((NULL == pattern->pattern.url_spec.host_regex)
|| (0 == regexec(pattern->pattern.url_spec.host_regex, http->host, 0, NULL, 0)));