X-Git-Url: http://www.privoxy.org/gitweb/?a=blobdiff_plain;f=urlmatch.c;h=3dead33b03ba2c6c0a1af62a9434d51a1da61a72;hb=2ce1c56a208dfde0a2a91791a9bc12eb7c4353e2;hp=6f357d0ca606b9d4695381b725c000ab74dbfa84;hpb=e7f941972fe36b13a5fd727c8d2357f20e61e2da;p=privoxy.git

diff --git a/urlmatch.c b/urlmatch.c
index 6f357d0c..3dead33b 100644
--- a/urlmatch.c
+++ b/urlmatch.c
@@ -1,4 +1,4 @@
-const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.55 2009/06/03 16:43:16 fabiankeil Exp $";
+const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.62 2011/09/04 11:10:56 fabiankeil Exp $";
 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
@@ -58,7 +58,13 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.55 2009/06/03 16:43:16 fabianke
 
 const char urlmatch_h_rcs[] = URLMATCH_H_VERSION;
 
-enum regex_anchoring {NO_ANCHORING, LEFT_ANCHORED, RIGHT_ANCHORED};
+enum regex_anchoring
+{
+   NO_ANCHORING,
+   LEFT_ANCHORED,
+   RIGHT_ANCHORED,
+   RIGHT_ANCHORED_HOST
+};
 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern);
 
 /*********************************************************************
@@ -197,11 +203,11 @@ jb_err parse_http_url(const char *url, struct http_request *http, int require_pr
 
    /*
     * Check for * URI. If found, we're done.
-    */  
+    */
    if (*http->url == '*')
    {
       if  ( NULL == (http->path = strdup("*"))
-         || NULL == (http->hostport = strdup("")) ) 
+         || NULL == (http->hostport = strdup("")) )
       {
          return JB_ERR_MEMORY;
       }
@@ -405,7 +411,7 @@ jb_err parse_http_url(const char *url, struct http_request *http, int require_pr
  *********************************************************************/
 static int unknown_method(const char *method)
 {
-   static const char *known_http_methods[] = {
+   static const char * const known_http_methods[] = {
       /* Basic HTTP request type */
       "GET", "HEAD", "POST", "PUT", "DELETE", "OPTIONS", "TRACE", "CONNECT",
       /* webDAV extensions (RFC2518) */
@@ -414,14 +420,14 @@ static int unknown_method(const char *method)
        * Microsoft webDAV extension for Exchange 2000.  See:
        * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
        * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
-       */ 
+       */
       "BCOPY", "BMOVE", "BDELETE", "BPROPFIND", "BPROPPATCH",
       /*
        * Another Microsoft webDAV extension for Exchange 2000.  See:
        * http://systems.cs.colorado.edu/grunwald/MobileComputing/Papers/draft-cohen-gena-p-base-00.txt
        * http://lists.w3.org/Archives/Public/w3c-dist-auth/2002JanMar/0001.html
        * http://msdn.microsoft.com/library/en-us/wss/wss/_webdav_methods.asp
-       */ 
+       */
       "SUBSCRIBE", "UNSUBSCRIBE", "NOTIFY", "POLL",
       /*
        * Yet another WebDAV extension, this time for
@@ -546,9 +552,10 @@ jb_err parse_http_request(const char *req, struct http_request *http)
  *
  * Parameters  :
  *          1  :  pattern = The pattern to compile.
- *          2  :  anchoring = How the regex should be anchored.
- *                            Can be either one of NO_ANCHORING,
- *                            LEFT_ANCHORED or RIGHT_ANCHORED.
+ *          2  :  anchoring = How the regex should be modified
+ *                            before compilation. Can be either
+ *                            one of NO_ANCHORING, LEFT_ANCHORED,
+ *                            RIGHT_ANCHORED or RIGHT_ANCHORED_HOST.
  *          3  :  url     = In case of failures, the spec member is
  *                          logged and the structure freed.
  *          4  :  regex   = Where the compiled regex should be stored.
@@ -582,6 +589,9 @@ static jb_err compile_pattern(const char *pattern, enum regex_anchoring anchorin
       case RIGHT_ANCHORED:
          fmt = "%s$";
          break;
+      case RIGHT_ANCHORED_HOST:
+         fmt = "%s\\.?$";
+         break;
       case LEFT_ANCHORED:
          fmt = "^%s";
          break;
@@ -731,7 +741,7 @@ static jb_err compile_url_pattern(struct url_spec *url, char *buf)
  *********************************************************************/
 static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
 {
-   return compile_pattern(host_pattern, RIGHT_ANCHORED, url, &url->host_regex);
+   return compile_pattern(host_pattern, RIGHT_ANCHORED_HOST, url, &url->host_regex);
 }
 
 #else
@@ -769,7 +779,7 @@ static jb_err compile_host_pattern(struct url_spec *url, const char *host_patter
       url->unanchored |= ANCHOR_LEFT;
    }
 
-   /* 
+   /*
     * Split domain into components
     */
    url->dbuffer = strdup(host_pattern);
@@ -779,7 +789,7 @@ static jb_err compile_host_pattern(struct url_spec *url, const char *host_patter
       return JB_ERR_MEMORY;
    }
 
-   /* 
+   /*
     * Map to lower case
     */
    for (p = url->dbuffer; *p ; p++)
@@ -787,7 +797,7 @@ static jb_err compile_host_pattern(struct url_spec *url, const char *host_patter
       *p = (char)tolower((int)(unsigned char)*p);
    }
 
-   /* 
+   /*
     * Split the domain name into components
     */
    url->dcount = ssplit(url->dbuffer, ".", v, SZ(v), 1, 1);
@@ -799,11 +809,11 @@ static jb_err compile_host_pattern(struct url_spec *url, const char *host_patter
    }
    else if (url->dcount != 0)
    {
-      /* 
+      /*
        * Save a copy of the pointers in dvec
        */
       size = (size_t)url->dcount * sizeof(*url->dvec);
-      
+
       url->dvec = (char **)malloc(size);
       if (NULL == url->dvec)
       {
@@ -842,13 +852,13 @@ static int simplematch(const char *pattern, const char *text)
 {
    const unsigned char *pat = (const unsigned char *)pattern;
    const unsigned char *txt = (const unsigned char *)text;
-   const unsigned char *fallback = pat; 
+   const unsigned char *fallback = pat;
    int wildcard = 0;
-  
+
    unsigned char lastchar = 'a';
    unsigned i;
    unsigned char charmap[32];
-  
+
    while (*txt)
    {
 
@@ -866,15 +876,15 @@ static int simplematch(const char *pattern, const char *text)
       }
 
       /* '*' in the pattern?  */
-      if (*pat == '*') 
+      if (*pat == '*')
       {
-     
+
          /* The pattern ends afterwards? Speed up the return. */
          if (*++pat == '\0')
          {
             return 0;
          }
-     
+
          /* Else, set wildcard mode and remember position after '*' */
          wildcard = 1;
          fallback = pat;
@@ -888,7 +898,7 @@ static int simplematch(const char *pattern, const char *text)
          while (*++pat != ']')
          {
             if (!*pat)
-            { 
+            {
                return 1;
             }
             else if (*pat == '-')
@@ -900,7 +910,7 @@ static int simplematch(const char *pattern, const char *text)
                for (i = lastchar; i <= *pat; i++)
                {
                   charmap[i / 8] |= (unsigned char)(1 << (i % 8));
-               } 
+               }
             }
             else
             {
@@ -911,21 +921,21 @@ static int simplematch(const char *pattern, const char *text)
       } /* -END- if Character range specification */
 
 
-      /* 
-       * Char match, or char range match? 
+      /*
+       * Char match, or char range match?
        */
       if ( (*pat == *txt)
       ||   (*pat == '?')
       ||   ((*pat == ']') && (charmap[*txt / 8] & (1 << (*txt % 8)))) )
       {
-         /* 
-          * Sucess: Go ahead
+         /*
+          * Success: Go ahead
           */
          pat++;
       }
       else if (!wildcard)
       {
-         /* 
+         /*
           * No match && no wildcard: No luck
           */
          return 1;
@@ -1118,7 +1128,7 @@ jb_err create_url_spec(struct url_spec *url, char *buf)
    }
 
    /* Is it a tag pattern? */
-   if (0 == strncmpic("TAG:", url->spec, 4))
+   if (0 == strncmpic(url->spec, "TAG:", 4))
    {
       /* The pattern starts with the first character after "TAG:" */
       const char *tag_pattern = buf + 4;
@@ -1192,6 +1202,51 @@ static int port_matches(const int port, const char *port_list)
 }
 
 
+/*********************************************************************
+ *
+ * Function    :  host_matches
+ *
+ * Description :  Compares a host against a host pattern.
+ *
+ * Parameters  :
+ *          1  :  url = The URL to match
+ *          2  :  pattern = The URL pattern
+ *
+ * Returns     :  TRUE for yes, FALSE otherwise.
+ *
+ *********************************************************************/
+static int host_matches(const struct http_request *http,
+                        const struct url_spec *pattern)
+{
+#ifdef FEATURE_EXTENDED_HOST_PATTERNS
+   return ((NULL == pattern->host_regex)
+      || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)));
+#else
+   return ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)));
+#endif
+}
+
+
+/*********************************************************************
+ *
+ * Function    :  path_matches
+ *
+ * Description :  Compares a path against a path pattern.
+ *
+ * Parameters  :
+ *          1  :  path = The path to match
+ *          2  :  pattern = The URL pattern
+ *
+ * Returns     :  TRUE for yes, FALSE otherwise.
+ *
+ *********************************************************************/
+static int path_matches(const char *path, const struct url_spec *pattern)
+{
+   return ((NULL == pattern->preg)
+      || (0 == regexec(pattern->preg, path, 0, NULL, 0)));
+}
+
+
 /*********************************************************************
  *
  * Function    :  url_match
@@ -1208,21 +1263,14 @@ static int port_matches(const int port, const char *port_list)
 int url_match(const struct url_spec *pattern,
               const struct http_request *http)
 {
-   /* XXX: these should probably be functions. */
-#ifdef FEATURE_EXTENDED_HOST_PATTERNS
-#define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, http->host, 0, NULL, 0)))
-#else
-#define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, http)))
-#endif
-#define PATH_MATCHES ((NULL == pattern->preg) || (0 == regexec(pattern->preg, http->path, 0, NULL, 0)))
-
    if (pattern->tag_regex != NULL)
    {
       /* It's a tag pattern and shouldn't be matched against URLs */
       return 0;
-   } 
+   }
 
-   return (port_matches(http->port, pattern->port_list) && DOMAIN_MATCHES && PATH_MATCHES);
+   return (port_matches(http->port, pattern->port_list)
+      && host_matches(http, pattern) && path_matches(http->path, pattern));
 
 }