--- spam_filter_url.module 2012-01-13 15:24:00.373197080 +0100
+++ spam_filter_url.NEU.module 2012-01-13 15:25:28.062071879 +0100
@@ -10,14 +10,19 @@
function spam_filter_url_spamapi($op, $type = NULL, $content = array(), $fields = array(), $extra = NULL) {
// Don't both with this hook unless the filter is actually enabled.
-
switch ($op) {
case 'filter':
- if (!module_invoke('spam', 'filter_enabled', 'spam_filter_url', $type, $content, $fields, $extra)) return;
- return spam_filter_url_spam_filter($content, $type, $fields, $extra);
+ if (!module_invoke('spam', 'filter_enabled', 'spam_filter_url', $type, $content, $fields, $extra)) {
+ return;
+ }
+ else {
+ return spam_filter_url_spam_filter($content, $type, $fields, $extra);
+ }
+ break;
case 'filter_module':
- return 'spam_filter_url';
+ return 'spam_filter_url';
+ break;
case 'filter_info':
return array(
@@ -41,93 +46,100 @@ function spam_filter_url_spamapi($op, $t
spam_log(SPAM_DEBUG, 'spam_filter_url_spamapi', t('@op', array('@op' => $op)), $type, $extra['id']);
$fields = spam_invoke_module($type, 'filter_fields', $extra['content']);
$spam_filter_urls = _spam_filter_url_extract($extra['content'], $type, $fields, $extra);
- spam_filter_url_update($spam_filter_urls, ($op == 'mark_as_spam' ? TRUE : FALSE), $type, $extra['id']);
+
+ // H.S. TODO: make this only if bayesian filter is ON
+ // spam_filter_url_update($spam_filter_urls, ($op == 'mark_as_spam' ? TRUE : FALSE), $type, $extra['id']);
break;
}
+
}
function _spam_filter_url_extract($content, $type, $fields, $extra = array()) {
+ // H.S.
static $spam_filter_urls = array();
+
$id = spam_invoke_module($type, 'content_id', $content, $extra);
if (is_object($content)) {
- $content = (array)$content;
+ $content = (array)$content;
+
+ if (!isset($spam_filter_urls[$id]["$type-$id"])) {
+ $string = '';
+ if (isset($fields['main']) && is_array($fields['main'])) {
+ foreach ($fields['main'] as $field) {
+ $string .= $content["$field"] .' ';
+ }
+ }
+ if (isset($fields['other']) && is_array($fields['other'])) {
+ foreach ($fields['other'] as $field) {
+ $string .= $content["$field"] .' ';
+ }
+ }
+
+ $URI = "(http://|https://|ftp://|mailto:|smb://|afp://|file://|gopher://|news://|ssl://|sslv2://|sslv3://|tls://|tcp://|udp://)";
+
+ preg_match_all('('.$URI.'([a-zA-Z0-9@:%_+*~#?&=.,/;-]*[a-zA-Z0-9@:%_+*~#&=/;-]))', $string, $matches_absolute);
+ preg_match_all('([^'.$URI.']www\.([a-zA-Z0-9@:%_+*~#?&=.,/;-]*[a-zA-Z0-9@:%_+*~#&=/;-]))', $string, $matches_www);
+
+ $matches = array_merge($matches_absolute[2], $matches_www[0]);
+
+ $u = array();
+ foreach ($matches as $spam_filter_url) {
+
+ $spam_filter_url = preg_replace("'$URI'", '', $spam_filter_url);
+ // get full domain (ie www.sample.com)
+ preg_match("/^()?([^\/\"\']+)/i", $spam_filter_url, $domain);
+ // get root domain (ie sample.com)
+ preg_match("/[^\.\/]+\.[^\.\/]+$/", $domain[2], $root);
+
+ $spam_filter_url = htmlspecialchars(drupal_strtolower($root[0]));
+ _spam_filter_url_count($spam_filter_url, $id);
+ $u[] = $spam_filter_url;
+ }
+ $spam_filter_urls[$id]["$type-$id"] = $u;
+ }
+
+ return $spam_filter_urls[$id]["$type-$id"];
}
-
- if (!isset($spam_filter_urls["$type-$id"])) {
- $string = '';
- if (isset($fields['main']) && is_array($fields['main'])) {
- foreach ($fields['main'] as $field) {
- $string .= $content["$field"] .' ';
- }
- }
- if (isset($fields['other']) && is_array($fields['other'])) {
- foreach ($fields['other'] as $field) {
- $string .= $content["$field"] .' ';
- }
- }
-
- $URI = "(http://|https://|ftp://|mailto:|smb://|afp://|file://|gopher://|news://|ssl://|sslv2://|sslv3://|tls://|tcp://|udp://)";
- // Find all urls in content. Code based on _filter_url().
- // Match absolute URLs.
- preg_match_all("`(|
||[ \n\r\t\(])($URI([a-zA-Z0-9@:%_+*~#?&=.,/;-]*[a-zA-Z0-9@:%_+*~#&=/;-]))([.,?!]*?)(?=(|||[ \n\r\t\)]))`i", $string, $matches_absolute);
- // Match e-mail addresses.
- preg_match_all("`(|
||[ \n\r\t\(])([A-Za-z0-9._-]+@[A-Za-z0-9._+-]+\.[A-Za-z]{2,4})([.,?!]*?)(?=(|||[ \n\r\t\)]))`i", $string, $matches_email);
- // Match www domains/addresses.
- preg_match_all("`(|
|[ \n\r\t\(])(www\.[a-zA-Z0-9@:%_+*~#?&=.,/;-]*[a-zA-Z0-9@:%_+~#\&=/;-])([.,?!]*?)(?=(|||[ \n\r\t\)]))`i", $string, $matches_www);
- $matches = array_merge($matches_absolute[2], $matches_email[2], $matches_www[2]);
- $u = array();
- foreach ($matches as $spam_filter_url) {
- $spam_filter_url = preg_replace("'$URI'", '', $spam_filter_url);
- // get full domain (ie www.sample.com)
- preg_match("/^()?([^\/\"\']+)/i", $spam_filter_url, $domain);
- // get root domain (ie sample.com)
- preg_match("/[^\.\/]+\.[^\.\/]+$/", $domain[2], $root);
- $spam_filter_url = htmlspecialchars(drupal_strtolower($root[0]));
- _spam_filter_url_count($spam_filter_url);
- $u[] = $spam_filter_url;
- }
- $spam_filter_urls["$type-$id"] = $u;
+ else {
+ return $spam_filter_urls;
}
-
- return $spam_filter_urls["$type-$id"];
}
/**
* Search for known spam urls in content.
*/
function spam_filter_url_spam_filter($content, $type, $fields, $extra = array(), $filter_test = FALSE) {
- $action = array();
+
+ $action = array();
$id = spam_invoke_module($type, 'content_id', $content, $extra);
$spam = FALSE;
-
$spam_filter_urls = _spam_filter_url_extract($content, $type, $fields, $extra);
-
$probability = variable_get('spam_filter_url_probability', 99);
if (is_array($spam_filter_urls) && !empty($spam_filter_urls)) {
+
$count = _spam_filter_url_count();
$limit = variable_get('spam_filter_url_limit_total', 10);
- if ($limit > -1 && $count['total'] > $limit) {
- spam_log(SPAM_VERBOSE, 'spam_filter_url_spam_filter', t('total urls(@total) > spam_filter_url_limit_total(@limit)', array('@total' => $count['total'], '@limit' => variable_get('spam_filter_url_limit_total', 10))), $type, $id);
+
+ if ($limit > -1 && $count[$id]['total'] > $limit) {
+ spam_log(SPAM_VERBOSE, 'spam_filter_url_spam_filter', t('total urls(@total) > spam_filter_url_limit_total(@limit)', array('@total' => $count[$id]['total'], '@limit' => variable_get('spam_filter_url_limit_total', 10))), $type, $id);
$action['spam_filter_url'][] = array(
'limit' => 'total',
- 'total' => $count['total'],
+ 'total' => $count[$id]['total'],
);
$action['total'] = $probability;
return $action;
}
+
$limit = variable_get('spam_filter_url_limit_repeat', 5);
if ($limit > -1) {
- // Sort urls from most repeated to least repeated.
- asort($count);
+ asort($count[$id]); // Sort urls from most repeated to least repeated.
+ array_pop($count[$id]); // skip count[$id]['total']
+ $max = array_pop($count[$id]);
- // skip count['total']
- array_pop($count);
-
- $max = array_pop($count);
if ($max > $limit) {
spam_log(SPAM_VERBOSE, 'spam_filter_url_spam_filter', t('repeated urls(@total) > spam_filter_url_limit_repeat(@limit)', array('@total' => $max, '@limit' => variable_get('spam_filter_url_limit_repeat', 5))), $type, $id);
$action['spam_filter_url'][] = array(
@@ -139,7 +151,7 @@ function spam_filter_url_spam_filter($co
}
}
- foreach ($spam_filter_urls as $spam_filter_url) {
+ foreach ($spam_filter_urls[$id] as $spam_filter_url) {
$p = db_fetch_object(db_query("SELECT probability FROM {spam_filter_bayesian_tokens} WHERE class = 'spam_filter_url' AND token = '%s'", $spam_filter_url));
$action['spam_filter_url'][] = array(
'spam_filter_url' => $spam_filter_url,
@@ -161,16 +173,22 @@ function spam_filter_url_spam_filter($co
$action['total'] = 0;
}
+
return $action;
}
+
/**
* Update url probabilities in database.
*/
+// H.S. TODO: make this only if bayesian filter is ON
function spam_filter_url_update($spam_filter_urls, $yes, $type, $id) {
module_invoke('spam_filter_bayesian', 'tokens_update', 'spam_filter_url', $spam_filter_urls, $yes, $type, $id);
}
+
+
+
/**
* Keep track of the total number of URLs found in the current content.
*
@@ -178,23 +196,26 @@ function spam_filter_url_update($spam_fi
* @return Array of URLs showing how many times each URL is present, and
* the total number of arrays.
*/
-function _spam_filter_url_count($spam_filter_url = NULL) {
+function _spam_filter_url_count($spam_filter_url = NULL, $id = 0) {
// build up an array of all URLs seen in current content
- static $spam_filter_urls = array('total' => 0);
+ static $spam_filter_urls;
if ($spam_filter_url != NULL) {
- if (empty($spam_filter_urls["$spam_filter_url"])) {
- $spam_filter_urls["$spam_filter_url"] = 1;
+ if (empty($spam_filter_urls[$id]["$spam_filter_url"])) {
+ $spam_filter_urls[$id]["$spam_filter_url"] = 1;
}
else {
- $spam_filter_urls["$spam_filter_url"]++;
+ $spam_filter_urls[$id]["$spam_filter_url"]++;
}
- $spam_filter_urls['total']++;
+ $spam_filter_urls[$id]['total']++;
}
return $spam_filter_urls;
}
+
+
+
function spam_filter_url_menu() {
$items = array();