10 files changed, 184 insertions, 29 deletions
diff --git a/phpBB/includes/functions.php b/phpBB/includes/functions.php
index 4227a8a9a3..02f1553c43 100644
--- a/phpBB/includes/functions.php
+++ b/phpBB/includes/functions.php
@@ -2015,6 +2015,117 @@ function bump_topic_allowed($forum_id, $topic_bumped, $last_post_time, $topic_po
 }
 
 /**
+* Generates a text with approx. the specified length which contains the specified words and their context
+*
+* @param	string	$text	The full text from which context shall be extracted
+* @param	string	$words	An array of words which should be contained in the result, * is allowed as a wildcard
+* @param	int		$length	The desired length of the resulting text, however the result might be shorter or longer than this value
+*
+* @return	string			Context of the specified words seperated by "..."
+*/
+function get_context($text, $words, $length = 400)
+{
+	// first replace all whitespaces with single spaces
+	$text = preg_replace('/\s+/', ' ', $text);
+
+	$word_indizes = array();
+	if (sizeof($words))
+	{
+		$match = '';
+		// find the starting indizes of all words
+		foreach ($words as $word)
+		{
+			if (preg_match('#(?: |^)(' . str_replace('\*', '\w*?', preg_quote($word, '#')) . ')(?: |$)#i', $text, $match))
+			{
+				$pos = strpos($text, $match[1]);
+				if ($pos !== false)
+				{
+					$word_indizes[] = $pos;
+				}
+			}
+		}
+		unset($match);
+
+		if (sizeof($word_indizes))
+		{
+			$word_indizes = array_unique($word_indizes);
+			sort($word_indizes);
+
+			$wordnum = sizeof($word_indizes);
+			// number of characters on the right and left side of each word
+			$sequence_length = (int) ($length / (2 * $wordnum)) - 2;
+			$final_text = '';
+			$word = $j = 0;
+			$final_text_index = -1;
+
+			// cycle through every character in the original text
+			for ($i = $word_indizes[$word], $n = strlen($text); $i < $n; $i++)
+			{
+				// if the current position is the start of one of the words then append $sequence_length characters to the final text
+				if (isset($word_indizes[$word]) && ($i == $word_indizes[$word]))
+				{
+					if ($final_text_index < $i - $sequence_length - 1)
+					{
+						$final_text .= '... ' . preg_replace('#^([^ ]*)#', '', substr($text, $i - $sequence_length, $sequence_length));
+					}
+					else
+					{
+						// if the final text is already nearer to the current word than $sequence_length we only append the text
+						// from its current index on and distribute the unused length to all other sequenes
+						$sequence_length += (int) (($final_text_index - $i + $sequence_length + 1) / (2 * $wordnum));
+						$final_text .= substr($text, $final_text_index + 1, $i - $final_text_index - 1);
+					}
+					$final_text_index = $i - 1;
+
+					// add the following characters to the final text (see below)
+					$word++;
+					$j = 1;
+				}
+
+				if ($j > 0)
+				{
+					// add the character to the final text and increment the sequence counter
+					$final_text .= $text[$i];
+					$final_text_index++;
+					$j++;
+
+					// if this is a whitespace then check whether we are done with this sequence
+					if ($text[$i] == ' ')
+					{
+						// only check whether we have to exit the context generation completely if we haven't already reached the end anyway
+						if ($i + 4 < $n)
+						{
+							if (($j > $sequence_length && $word >= $wordnum) || strlen($final_text) > $length)
+							{
+								$final_text .= ' ...';
+								break;
+							}
+						}
+						else
+						{
+							// make sure the text really reaches the end
+							$j -= 4;
+						}
+
+						// stop context generation and wait for the next word
+						if ($j > $sequence_length)
+						{
+							$j = 0;
+						}
+					}
+				}
+			}
+			return $final_text;
+		}
+	}
+
+	if (!sizeof($words) || !sizeof($word_indizes))
+	{
+		return (strlen($text) >= $length + 3) ? substr($text, 0, $length) . '...' : $text;
+	}
+}
+
+/**
 * Decode text whereby text is coming from the db and expected to be pre-parsed content
 * We are placing this outside of the message parser because we are often in need of it...
 */
@@ -2053,6 +2164,33 @@ function decode_message(&$message, $bbcode_uid = '')
 }
 
 /**
+* Strips all bbcode from a text and returns the plain content
+*/
+function strip_bbcode(&$text, $uid = '')
+{
+	if (!$uid)
+	{
+		$uid = '[0-9a-z]{5,}';
+	}
+
+	$text = preg_replace("#\[\/?[a-z0-9\*\+\-]+(?:=.*?)?(?::[a-z])?(\:?$uid)\]#", ' ', $text);
+
+	$match = array(
+		'#<!\-\- e \-\-><a href="mailto:(.*?)">.*?</a><!\-\- e \-\->#',
+		'#<!\-\- m \-\-><a href="(.*?)" target="_blank">.*?</a><!\-\- m \-\->#',
+		'#<!\-\- w \-\-><a href="http:\/\/(.*?)" target="_blank">.*?</a><!\-\- w \-\->#',
+		'#<!\-\- l \-\-><a href="(.*?)">.*?</a><!\-\- l \-\->#',
+		'#<!\-\- s(.*?) \-\-><img src="\{SMILIES_PATH\}\/.*? \/><!\-\- s\1 \-\->#',
+		'#<!\-\- .*? \-\->#s',
+		'#<.*?>#s'
+	);
+	
+	$replace = array('\1', '\1', '\1', '\1', '\1', '', '');
+	
+	$text = preg_replace($match, $replace, $text);
+}
+
+/**
 * For display of custom parsed text on user-facing pages
 * Expects $text to be the value directly from the database (stored value)
 */
@@ -2065,6 +2203,8 @@ function generate_text_for_display($text, $uid, $bitfield, $flags)
 		return '';
 	}
 
+	$text = str_replace("\n", '<br />', censor_text($text));
+
 	// Parse bbcode if bbcode uid stored and bbcode enabled
 	if ($uid && ($flags & 1))
 	{
@@ -2074,7 +2214,7 @@ function generate_text_for_display($text, $uid, $bitfield, $flags)
 			include_once($phpbb_root_path . 'includes/bbcode.' . $phpEx);
 		}
 
-		if (empty($__bbcode))
+		if (empty($bbcode))
 		{
 			$bbcode = new bbcode($bitfield);
 		}
@@ -2087,7 +2227,6 @@ function generate_text_for_display($text, $uid, $bitfield, $flags)
 	}
 
 	$text = smiley_text($text, !($flags & 2));
-	$text = str_replace("\n", '<br />', censor_text($text));
 
 	return $text;
 }
diff --git a/phpBB/includes/functions_posting.php b/phpBB/includes/functions_posting.php
index 2ac732746e..a8c7078057 100644
--- a/phpBB/includes/functions_posting.php
+++ b/phpBB/includes/functions_posting.php
@@ -873,6 +873,8 @@ function topic_review($topic_id, $forum_id, $mode = 'topic_review', $cur_post_id
 
 		$post_subject = $row['post_subject'];
 		$message = $row['post_text'];
+		$message = censor_text($message);
+		$message = str_replace("\n", '<br />', $message);
 		$decoded_message = false;
 
 		if ($show_quote_button && $auth->acl_get('f_reply', $forum_id))
@@ -892,14 +894,13 @@ function topic_review($topic_id, $forum_id, $mode = 'topic_review', $cur_post_id
 		$message = smiley_text($message, !$row['enable_smilies']);
 
 		$post_subject = censor_text($post_subject);
-		$message = censor_text($message);
 
 		$template->assign_block_vars($mode . '_row', array(
 			'POSTER_NAME'		=> $poster,
 			'POST_SUBJECT'		=> $post_subject,
 			'MINI_POST_IMG'		=> $user->img('icon_post_target', $user->lang['POST']),
 			'POST_DATE'			=> $user->format_date($row['post_time']),
-			'MESSAGE'			=> str_replace("\n", '<br />', $message), 
+			'MESSAGE'			=> $message,
 			'DECODED_MESSAGE'	=> $decoded_message,
 
 			'U_POST_ID'			=> $row['post_id'],
diff --git a/phpBB/includes/mcp/mcp_post.php b/phpBB/includes/mcp/mcp_post.php
index 08bcc713f8..9cb3ec278b 100644
--- a/phpBB/includes/mcp/mcp_post.php
+++ b/phpBB/includes/mcp/mcp_post.php
@@ -95,6 +95,7 @@ function mcp_post_details($id, $mode, $action)
 
 	// Process message, leave it uncensored
 	$message = $post_info['post_text'];
+	$message = str_replace("\n", '<br />', $message);
 	if ($post_info['bbcode_bitfield'])
 	{
 		include_once($phpbb_root_path . 'includes/bbcode.' . $phpEx);
@@ -102,7 +103,6 @@ function mcp_post_details($id, $mode, $action)
 		$bbcode->bbcode_second_pass($message, $post_info['bbcode_uid'], $post_info['bbcode_bitfield']);
 	}
 	$message = smiley_text($message);
-	$message = str_replace("\n", '<br />', $message);
 
 	$template->assign_vars(array(
 		'U_MCP_ACTION'			=> "$url&amp;i=main&amp;quickmod=1", // Use this for mode paramaters
diff --git a/phpBB/includes/mcp/mcp_queue.php b/phpBB/includes/mcp/mcp_queue.php
index 95e89fa9dc..3deeff6d1f 100644
--- a/phpBB/includes/mcp/mcp_queue.php
+++ b/phpBB/includes/mcp/mcp_queue.php
@@ -109,6 +109,7 @@ class mcp_queue
 
 				// Process message, leave it uncensored
 				$message = $post_info['post_text'];
+				$message = str_replace("\n", '<br />', $message);
 				if ($post_info['bbcode_bitfield'])
 				{
 					include_once($phpbb_root_path . 'includes/bbcode.' . $phpEx);
diff --git a/phpBB/includes/mcp/mcp_reports.php b/phpBB/includes/mcp/mcp_reports.php
index a52bc02359..67af086292 100755
--- a/phpBB/includes/mcp/mcp_reports.php
+++ b/phpBB/includes/mcp/mcp_reports.php
@@ -117,6 +117,7 @@ class mcp_reports
 
 				// Process message, leave it uncensored
 				$message = $post_info['post_text'];
+				$message = str_replace("\n", '<br />', $message);
 				if ($post_info['bbcode_bitfield'])
 				{
 					include_once($phpbb_root_path . 'includes/bbcode.' . $phpEx);
diff --git a/phpBB/includes/mcp/mcp_topic.php b/phpBB/includes/mcp/mcp_topic.php
index 7f1bc2a0c4..466459415d 100644
--- a/phpBB/includes/mcp/mcp_topic.php
+++ b/phpBB/includes/mcp/mcp_topic.php
@@ -110,6 +110,7 @@ function mcp_topic_view($id, $mode, $action)
 
 		$message = $row['post_text'];
 		$post_subject = ($row['post_subject'] != '') ? $row['post_subject'] : $topic_info['topic_title'];
+		$message = str_replace("\n", '<br />', $message);
 
 		if ($row['bbcode_bitfield'])
 		{
@@ -117,7 +118,6 @@ function mcp_topic_view($id, $mode, $action)
 		}
 
 		$message = smiley_text($message);
-		$message = str_replace("\n", '<br />', $message);
 
 		if (!$row['post_approved'])
 		{
diff --git a/phpBB/includes/mcp/mcp_warn.php b/phpBB/includes/mcp/mcp_warn.php
index bc428570ab..9a6ae1ab8f 100755
--- a/phpBB/includes/mcp/mcp_warn.php
+++ b/phpBB/includes/mcp/mcp_warn.php
@@ -247,6 +247,7 @@ function mcp_warn_post_view($id, $mode, $action)
 	// We want to make the message available here as a reminder
 	// Parse the message and subject
 	$message = $userrow['post_text'];
+	$message = str_replace("\n", '<br />', censor_text($message));
 
 	// Second parse bbcode here
 	if ($userrow['bbcode_bitfield'])
@@ -260,9 +261,6 @@ function mcp_warn_post_view($id, $mode, $action)
 	// Always process smilies after parsing bbcodes
 	$message = smiley_text($message);
 
-	// Replace naughty words such as farty pants
-	$message = str_replace("\n", '<br />', censor_text($message));
-
 	// Generate the appropriate user information for the user we are looking at
 	$rank_title = $rank_img = '';
 //	get_user_rank($userrow['user_rank'], $userrow['user_posts'], $rank_title, $rank_img);
diff --git a/phpBB/includes/search/fulltext_native.php b/phpBB/includes/search/fulltext_native.php
index e4ff1f8fc2..90da8abcd1 100755
--- a/phpBB/includes/search/fulltext_native.php
+++ b/phpBB/includes/search/fulltext_native.php
@@ -119,6 +119,15 @@ class fulltext_native extends search_backend
 		// remove some useless bracket combinations which might be created by the previous regexps
 		$keywords = str_replace(array('()', ')|('), array('', '|'), $keywords);
 
+		$keywords = preg_replace_callback(
+			'#\((?:(?:[^)]*?) )*?[^)]*?\)#',
+			create_function(
+				'$matches',
+				'return str_replace(" ", "|", $matches[0]);'
+			),
+			$keywords
+		);
+
 		// $keywords input format: each word seperated by a space, words in a bracket are not seperated
 
 		// the user wants to search for any word, convert the search query
@@ -187,7 +196,7 @@ class fulltext_native extends search_backend
 				// a group of which at least one may not be in the resulting posts
 				if ($word[0] == '(')
 				{
-					$word = explode('|', substr($word, 1, -1));
+					$word = array_unique(explode('|', substr($word, 1, -1)));
 					$mode = 'must_exclude_one';
 				}
 				// one word which should not be in the resulting posts
@@ -209,7 +218,7 @@ class fulltext_native extends search_backend
 				// a group of words of which at least one word should be in every resulting post
 				if ($word[0] == '(')
 				{
-					$word = explode('|', substr($word, 1, -1));
+					$word = array_unique(explode('|', substr($word, 1, -1)));
 				}
 				$ignore_no_id = false;
 				$mode = 'must_contain';
@@ -880,7 +889,7 @@ class fulltext_native extends search_backend
 		// Do not index code
 		$match[] = '#\[code(?:=.*?)?(\:?[0-9a-z]{5,})\].*?\[\/code(\:?[0-9a-z]{5,})\]#is';
 		// BBcode
-		$match[] = '#\[\/?[a-z\*\+\-]+(?:=.*?)?(\:?[0-9a-z]{5,})\]#';
+		$match[] = '#\[\/?[a-z0-9\*\+\-]+(?:=.*?)?(?::[a-z])?(\:?[0-9a-z]{5,})\]#';
 
 		$min = $config['fulltext_native_min_chars'];
 		$max = $config['fulltext_native_max_chars'];
@@ -890,7 +899,7 @@ class fulltext_native extends search_backend
 		/**
 		* Clean up the string, remove HTML tags, remove BBCodes
 		*/
-		$word = strtok($this->cleanup(preg_replace($match, ' ', strip_tags($text)), '', $user->lang['ENCODING']), ' ');
+		$word = strtok($this->cleanup(preg_replace($match, ' ', strip_tags($text)), -1, $user->lang['ENCODING']), ' ');
 
 		while (isset($word[0]))
 		{
@@ -1146,14 +1155,14 @@ class fulltext_native extends search_backend
 
 		$destroy_cache_words = array();
 
-		// Remove common (> 60% of posts ) words
+		// Remove common (> 20% of posts ) words
 		if ($config['num_posts'] >= 100)
 		{
 			// First, get the IDs of common words
 			$sql = 'SELECT word_id
 				FROM ' . SEARCH_WORDMATCH_TABLE . '
 				GROUP BY word_id
-				HAVING COUNT(word_id) > ' . floor($config['num_posts'] * 0.6);
+				HAVING COUNT(word_id) > ' . floor($config['num_posts'] * 0.2);
 			$result = $db->sql_query($sql);
 
 			$sql_in = array();
diff --git a/phpBB/includes/session.php b/phpBB/includes/session.php
index 0e934edb8a..5441f7e556 100644
--- a/phpBB/includes/session.php
+++ b/phpBB/includes/session.php
@@ -1340,13 +1340,15 @@ class user extends session
 		static $imgs;
 		global $phpbb_root_path;
 
-		if (empty($imgs[$img . $suffix]) || $width !== false)
+		$img_data = $imgs[$img . $suffix];
+
+		if (empty($img_data) || $width !== false)
 		{
 			if (!isset($this->theme[$img]) || !$this->theme[$img])
 			{
 				// Do not fill the image to let designers decide what to do if the image is empty
-				$imgs[$img . $suffix] = '';
-				return $imgs[$img . $suffix];
+				$img_data = '';
+				return $img_data;
 			}
 
 			// Do not include dimensions?
@@ -1372,9 +1374,9 @@ class user extends session
 				$imgsrc = str_replace('{SUFFIX}', $suffix, $imgsrc);
 			}
 
-			$imgs[$img . $suffix]['src'] = $phpbb_root_path . 'styles/' . $this->theme['imageset_path'] . '/imageset/' . str_replace('{LANG}', $this->img_lang, $imgsrc);
-			$imgs[$img . $suffix]['width'] = $width;
-			$imgs[$img . $suffix]['height'] = $height;
+			$img_data['src'] = $phpbb_root_path . 'styles/' . $this->theme['imageset_path'] . '/imageset/' . str_replace('{LANG}', $this->img_lang, $imgsrc);
+			$img_data['width'] = $width;
+			$img_data['height'] = $height;
 		}
 
 		$alt = (!empty($this->lang[$alt])) ? $this->lang[$alt] : $alt;
@@ -1382,19 +1384,19 @@ class user extends session
 		switch ($type)
 		{
 			case 'src':
-				return $imgs[$img . $suffix]['src'];
+				return $img_data['src'];
 			break;
 			
 			case 'width':
-				return $imgs[$img . $suffix]['width'];
+				return $img_data['width'];
 			break;
 
 			case 'height':
-				return $imgs[$img . $suffix]['height'];
+				return $img_data['height'];
 			break;
 
 			default:
-				return '<img src="' . $imgs[$img . $suffix]['src'] . '"' . (($imgs[$img . $suffix]['width']) ? ' width="' . $imgs[$img . $suffix]['width'] . '"' : '') . (($imgs[$img . $suffix]['height']) ? ' height="' . $imgs[$img . $suffix]['height'] . '"' : '') . ' alt="' . $alt . '" title="' . $alt . '" />';
+				return '<img src="' . $img_data['src'] . '"' . (($img_data['width']) ? ' width="' . $img_data['width'] . '"' : '') . (($img_data['height']) ? ' height="' . $img_data['height'] . '"' : '') . ' alt="' . $alt . '" title="' . $alt . '" />';
 			break;
 		}
 	}
diff --git a/phpBB/includes/ucp/ucp_pm_viewmessage.php b/phpBB/includes/ucp/ucp_pm_viewmessage.php
index f0fa234944..3ffc19df55 100644
--- a/phpBB/includes/ucp/ucp_pm_viewmessage.php
+++ b/phpBB/includes/ucp/ucp_pm_viewmessage.php
@@ -54,6 +54,7 @@ function view_message($id, $mode, $folder_id, $msg_id, $folder, $message_row)
 
 	// Parse the message and subject
 	$message = $message_row['message_text'];
+	$message = str_replace("\n", '<br />', censor_text($message));
 
 	// Second parse bbcode here
 	if ($message_row['bbcode_bitfield'])
@@ -66,7 +67,6 @@ function view_message($id, $mode, $folder_id, $msg_id, $folder, $message_row)
 
 	// Replace naughty words such as farty pants
 	$message_row['message_subject'] = censor_text($message_row['message_subject']);
-	$message = str_replace("\n", '<br />', censor_text($message));
 
 	// Editing information
 	if ($message_row['message_edit_count'] && $config['display_last_edited'])
@@ -146,6 +146,9 @@ function view_message($id, $mode, $folder_id, $msg_id, $folder, $message_row)
 	// End signature parsing, only if needed
 	if ($signature)
 	{
+		$signature = censor_text($signature);
+		$signature = str_replace("\n", '<br />', censor_text($signature));
+
 		if ($user_info['user_sig_bbcode_bitfield'])
 		{
 			if ($bbcode === false)
@@ -158,7 +161,6 @@ function view_message($id, $mode, $folder_id, $msg_id, $folder, $message_row)
 		}
 
 		$signature = smiley_text($signature);
-		$signature = str_replace("\n", '<br />', censor_text($signature));
 	}
 
 	$url = append_sid("{$phpbb_root_path}ucp.$phpEx", 'i=pm');
@@ -318,6 +320,9 @@ function message_history($msg_id, $user_id, $message_row, $folder)
 		$subject	= $row['message_subject'];
 		$message	= $row['message_text'];
 
+		$message = censor_text($message);
+		$message = str_replace("\n", '<br />', $message)
+
 		if ($row['bbcode_bitfield'])
 		{
 			$bbcode->bbcode_second_pass($message, $row['bbcode_uid'], $row['bbcode_bitfield']);
@@ -326,7 +331,6 @@ function message_history($msg_id, $user_id, $message_row, $folder)
 		$message = smiley_text($message, !$row['enable_smilies']);
 
 		$subject = censor_text($subject);
-		$message = censor_text($message);
 
 		if ($id == $msg_id)
 		{
@@ -339,7 +343,7 @@ function message_history($msg_id, $user_id, $message_row, $folder)
 			'AUTHOR_NAME'	=> $author,
 			'SUBJECT'		=> $subject,
 			'SENT_DATE'		=> $user->format_date($row['message_time']),
-			'MESSAGE'		=> str_replace("\n", '<br />', $message),
+			'MESSAGE'		=> $message,
 			'FOLDER'		=> implode(', ', $row['folder']),
 
 			'S_CURRENT_MSG'	=> ($row['msg_id'] == $msg_id),