X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=urlmatch.c;h=6949eedb82d2716f306ba8c0b73a2d4576813433;hp=c1fc5cbb3000b26d288cd90ba839650de8e3d4df;hb=d9527652c419834b2a38c0fa2985b66a9d2e0b3a;hpb=ba3a814851f5097de66ea12e3bbde560ca079446

diff --git a/urlmatch.c b/urlmatch.c
index c1fc5cbb..6949eedb 100644
--- a/urlmatch.c
+++ b/urlmatch.c
@@ -1,4 +1,3 @@
-const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.81 2014/05/26 10:48:07 fabiankeil Exp $";
 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
@@ -6,8 +5,8 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.81 2014/05/26 10:48:07 fabianke
  * Purpose     :  Declares functions to match URLs against URL
  *                patterns.
  *
- * Copyright   :  Written by and Copyright (C) 2001-2011
- *                the Privoxy team. http://www.privoxy.org/
+ * Copyright   :  Written by and Copyright (C) 2001-2014
+ *                the Privoxy team. https://www.privoxy.org/
  *
  *                Based on the Internet Junkbuster originally written
  *                by and Copyright (C) 1997 Anonymous Coders and
@@ -56,8 +55,6 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.81 2014/05/26 10:48:07 fabianke
 #include "miscutil.h"
 #include "errlog.h"
 
-const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
-
 enum regex_anchoring
 {
    NO_ANCHORING,
@@ -90,7 +87,7 @@ void free_http_request(struct http_request *http)
    freez(http->url);
    freez(http->hostport);
    freez(http->path);
-   freez(http->ver);
+   freez(http->version);
    freez(http->host_ip_addr_str);
 #ifndef FEATURE_EXTENDED_HOST_PATTERNS
    freez(http->dbuffer);
@@ -294,12 +291,17 @@ jb_err parse_http_url(const char *url, struct http_request *http, int require_pr
          /*
           * Got a path.
           *
-          * NOTE: The following line ignores the path for HTTPS URLS.
-          * This means that you get consistent behaviour if you type a
-          * https URL in and it's parsed by the function.  (When the
-          * URL is actually retrieved, SSL hides the path part).
+          * If FEATURE_HTTPS_INSPECTION isn't available, ignore the
+          * path for https URLs so that we get consistent behaviour
+          * if a https URL is parsed. When the URL is actually
+          * retrieved, https hides the path part.
           */
-         http->path = strdup_or_die(http->ssl ? "/" : url_path);
+         http->path = strdup_or_die(
+#ifndef FEATURE_HTTPS_INSPECTION
+            http->ssl ? "/" :
+#endif
+            url_path
+         );
          *url_path = '\0';
          http->hostport = strdup_or_die(url_noproto);
       }
@@ -454,6 +456,12 @@ static int unknown_method(const char *method)
        */
       "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;
 
@@ -470,6 +478,50 @@ static int unknown_method(const char *method)
 }
 
 
+/*********************************************************************
+ *
+ * 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
@@ -520,10 +572,8 @@ jb_err parse_http_request(const char *req, struct http_request *http)
       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;
    }
@@ -542,7 +592,7 @@ jb_err parse_http_request(const char *req, struct http_request *http)
     */
    http->cmd = strdup_or_die(req);
    http->gpc = strdup_or_die(v[0]);
-   http->ver = strdup_or_die(v[2]);
+   http->version = strdup_or_die(v[2]);
    http->ocmd = strdup_or_die(http->cmd);
 
    freez(buf);
@@ -569,7 +619,6 @@ jb_err parse_http_request(const char *req, struct http_request *http)
  *          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
  *
  *********************************************************************/
@@ -577,11 +626,11 @@ static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchorin
                               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')
    {
@@ -607,32 +656,30 @@ static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchorin
          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;
 
@@ -1119,6 +1166,9 @@ jb_err create_pattern_spec(struct pattern_spec *pattern, char *buf)
       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}
    };
@@ -1233,6 +1283,7 @@ static int port_matches(const int port, const char *port_list)
 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)));
@@ -1374,31 +1425,53 @@ int match_portlist(const char *portlist, int port)
  *
  * Function    :  parse_forwarder_address
  *
- * Description :  Parse out the host and port from a forwarder address.
+ * Description :  Parse out the username, password, host and port from
+ *                a forwarder address.
  *
  * Parameters  :
  *          1  :  address = The forwarder address to parse.
  *          2  :  hostname = Used to return the hostname. NULL on error.
  *          3  :  port = Used to return the port. Untouched if no port
  *                       is specified.
+ *          4  :  username = Used to return the username if any.
+ *          5  :  password = Used to return the password if any.
  *
  * Returns     :  JB_ERR_OK on success
  *                JB_ERR_MEMORY on out of memory
  *                JB_ERR_PARSE on malformed address.
  *
  *********************************************************************/
-jb_err parse_forwarder_address(char *address, char **hostname, int *port)
+jb_err parse_forwarder_address(char *address, char **hostname, int *port,
+                               char **username, char **password)
 {
-   char *p = address;
+   char *p;
+   char *tmp;
+
+   tmp = *hostname = strdup_or_die(address);
 
-   if ((*address == '[') && (NULL == strchr(address, ']')))
+   /* Parse username and password */
+   if (username && password && (NULL != (p = strchr(*hostname, '@'))))
+   {
+      *p++ = '\0';
+      *username = strdup_or_die(*hostname);
+      *hostname = strdup_or_die(p);
+
+      if (NULL != (p = strchr(*username, ':')))
+      {
+         *p++ = '\0';
+         *password = strdup_or_die(p);
+      }
+      freez(tmp);
+   }
+
+   /* Parse hostname and port */
+   p = *hostname;
+   if ((*p == '[') && (NULL == strchr(p, ']')))
    {
       /* XXX: Should do some more validity checks here. */
       return JB_ERR_PARSE;
    }
 
-   *hostname = strdup_or_die(address);
-
    if ((**hostname == '[') && (NULL != (p = strchr(*hostname, ']'))))
    {
       *p++ = '\0';