### Eclipse Workspace Patch 1.0
#P search_keywords
Index: search_keywords.module
===================================================================
RCS file: /cvs/drupal/contributions/modules/search_keywords/search_keywords.module,v
retrieving revision 1.19
diff -u -r1.19 search_keywords.module
--- search_keywords.module	27 Mar 2007 06:00:44 -0000	1.19
+++ search_keywords.module	12 Apr 2007 21:43:00 -0000
@@ -135,6 +135,14 @@
     '#options' => $period,
     '#description' => t('Older access log entries (including referrer search_keywords) will be automatically discarded.  Requires crontab.'));
 
+  $form['access']['search_keywords_search_keys'] = array(
+    '#type' => 'textfield',
+    '#title' => t('Search keys'),
+    '#default_value' => variable_get('search_keywords_search_keys', _search_keywords_default_search_keys()),
+    '#maxlength' => 255,
+    '#description' => t('Keys form part of the query string and mark the start of the search terms. Enter a list of terms, separated by spaces, that will be used to identify search terms in a referral.'),
+  );
+
   return system_settings_form($form);
 }
 
@@ -162,35 +170,49 @@
 }
 
 function _search_keywords_extract($url) {
-  $engines = array();
-  $engines[] = '/http:\/\/(.*?)\/search\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/base\/search\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/blogsearch\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/search\?.*?p=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/results.aspx\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/tags?\/(.*)/i';
-  $engines[] = '/http:\/\/(.*?)\/search\/(.*)/i';
-  $engines[] = '/http:\/\/(.*?)\/search\.php\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/\?.*?tag=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/cgi-bin\/search\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/aolcom\/search\?.*?query=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/jsp\/([^\.]+)\.jsp\?.*?searchfor=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/p\/search\?.*?qt=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/web\?.*?q=([^&]+)/i';
-  $engines[] = '/http:\/\/(.*?)\/searchScreen\?.*?q=([^&]+)/i'; //Seznam.cz
-  $engines[] = '/http:\/\/(.*?)\/searchScreen\?.*?w=([^&]+)/i'; //Seznam.cz
-  $engines[] = '/http:\/\/(.*?)\/index.php\?.*?q=([^&]+)/i'; //ZoooHoo.cz
-  $engines[] = '/http:\/\/(.*?)\/\?.*?q=([^&]+)/i'; //ZoooHoo.cz
-  foreach ($engines as $pattern) {
-    if (preg_match($pattern, $url, $matches)) {
-      $q = urldecode($matches[2]);
-      if (!_search_keywords_is_utf8($q)) {
-        $q = _search_keywords_to_utf8($q);
-      }
-      return array($matches[1],$q);
+
+  // Don't log if this is the host site (ignore non-clean url)
+  // Strip http(s)://www. if it exists, to handle sites that accept URLs both
+  // with and without the www prefix
+
+  // Get the base url
+  $base_url = $GLOBALS['base_url'];
+
+  // Construct the regex pattern to extract the common part
+  $pattern = '@^http(s)?\://(www\.)?(.*)$@i';
+  preg_match($pattern, $base_url, $results);
+  $base_url = $results[3];
+
+  // Get search engine query keys and prepare for regex
+  $engineterms = trim(variable_get('search_keywords_search_keys', _search_keywords_default_search_keys()));
+  $engineterms = str_replace(' ', '|', preg_quote($engineterms));
+
+  // Construct the regex pattern for checking the host and search terms
+  $pattern = '@^http(s)?\://(www\.)?('.preg_quote($base_url).')?.*[&?]('.$engineterms.')=([^&]+)@i';
+
+  // See if the regex found anything
+  if ( preg_match($pattern, $url, $matches) ) {
+
+    // If this was local referral then return nothing
+    if ( $base_url == $matches[3] ) {
+      return array(NULL, NULL);
     }
+
+    // Get the search terms
+    $terms = urldecode($matches[5]);
+
+	if (!_search_keywords_is_utf8($terms)) {
+      $terms = _search_keywords_to_utf8($terms);
+    }
+
+    // Get the host name
+    $host = parse_url($url);
+
+    return array($host[host], $terms);
   }
-  return array(NULL,NULL);
+
+  // If we get here then this was never a search
+  return array(NULL, NULL);
 }
 
 function _search_keywords_is_utf8($string) {
@@ -204,7 +226,7 @@
        | [\xF1-\xF3][\x80-\xBF]{3}          # planes 4-15
        |  \xF4[\x80-\x8F][\x80-\xBF]{2}    # plane 16
   )*$%xs', $string);
-  
+
 }
 
 function _search_keywords_to_utf8($string) {
@@ -218,7 +240,7 @@
 function _search_keywords_tis620_to_utf8($string) {
   if (!ereg("[\241-\377]", $string))
     return $string;
- 
+
     $tis620 = array(
 "\xa1" => "\xe0\xb8\x81",
 "\xa2" => "\xe0\xb8\x82",
@@ -308,7 +330,44 @@
 "\xfa" => "\xe0\xb9\x9a",
 "\xfb" => "\xe0\xb9\x9b"
   );
- 
+
   $string=strtr($string,$tis620);
   return $string;
 }
+
+function _search_keywords_default_search_keys() {
+
+  /* This list based on http://suda.co.uk/projects/SEHL/ */
+
+  $accepted_query_keys = array(
+	 'q',
+	 'p',
+	 'ask',
+	 'searchfor',
+	 'key',
+	 'query',
+	 'search',
+	 'keyword',
+	 'keywords',
+	 'qry',
+	 'searchitem',
+	 'kwd',
+	 'recherche',
+	 'search_text',
+	 'search_term',
+	 'term',
+	 'terms',
+	 'qq',
+	 'qry_str',
+	 'qu',
+	 's',
+	 'k',
+	 't',
+	 'va'
+	);
+
+  // Sort so it looks nice when displayed as the default list on settings page
+  asort($accepted_query_keys);
+  return implode(' ', $accepted_query_keys);
+
+}