From ed1d212ae2daea5e4bd043417610177093e99f19 Mon Sep 17 00:00:00 2001 From: Aleksander Machniak <alec@alec.pl> Date: Sat, 16 Jan 2016 03:03:51 -0500 Subject: [PATCH] Improved SVG cleanup code --- program/lib/Roundcube/rcube_string_replacer.php | 96 ++++++++++++++++++++++++++++++++++++++++------- 1 files changed, 81 insertions(+), 15 deletions(-) diff --git a/program/lib/Roundcube/rcube_string_replacer.php b/program/lib/Roundcube/rcube_string_replacer.php index 0fe982b..eb281ee 100644 --- a/program/lib/Roundcube/rcube_string_replacer.php +++ b/program/lib/Roundcube/rcube_string_replacer.php @@ -1,6 +1,6 @@ <?php -/* +/** +-----------------------------------------------------------------------+ | This file is part of the Roundcube Webmail client | | Copyright (C) 2009-2012, The Roundcube Dev Team | @@ -24,22 +24,31 @@ */ class rcube_string_replacer { - public static $pattern = '/##str_replacement\[([0-9]+)\]##/'; + public static $pattern = '/##str_replacement_(\d+)##/'; public $mailto_pattern; public $link_pattern; - private $values = array(); + public $linkref_index; + public $linkref_pattern; + + protected $values = array(); + protected $options = array(); + protected $linkrefs = array(); + protected $urls = array(); - function __construct() + function __construct($options = array()) { // Simplified domain expression for UTF8 characters handling // Support unicode/punycode in top-level domain part - $utf_domain = '[^?&@"\'\\/()\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; + $utf_domain = '[^?&@"\'\\/()<>\s\r\t\n]+\\.?([^\\x00-\\x2f\\x3b-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-zA-Z0-9]{2,})'; $url1 = '.:;,'; - $url2 = 'a-zA-Z0-9%=#$@+?!&\\/_~\\[\\]{}\*-'; + $url2 = 'a-zA-Z0-9%=#$@+?|!&\\/_~\\[\\]\\(\\){}\*\x80-\xFE-'; - $this->link_pattern = "/([\w]+:\/\/|\W[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)($utf_domain([$url1]?[$url2]+)*)/"; - $this->mailto_pattern = "/(" + $this->options = $options; + $this->linkref_index = '/\[([^\]#]+)\](:?\s*##str_replacement_(\d+)##)/'; + $this->linkref_pattern = '/\[([^\]#]+)\]/'; + $this->link_pattern = "/([\w]+:\/\/|\W[Ww][Ww][Ww]\.|^[Ww][Ww][Ww]\.)($utf_domain([$url1]*[$url2]+)*)/"; + $this->mailto_pattern = "/(" ."[-\w!\#\$%&\'*+~\/^`|{}=]+(?:\.[-\w!\#\$%&\'*+~\/^`|{}=]+)*" // local-part ."@$utf_domain" // domain-part ."(\?[$url1$url2]+)?" // e.g. ?subject=test... @@ -49,7 +58,8 @@ /** * Add a string to the internal list * - * @param string String value + * @param string String value + * * @return int Index of value for retrieval */ public function add($str) @@ -64,7 +74,7 @@ */ public function get_replacement($i) { - return '##str_replacement['.$i.']##'; + return '##str_replacement_' . $i . '##'; } /** @@ -89,21 +99,49 @@ if ($url) { $suffix = $this->parse_url_brackets($url); - $i = $this->add($prefix . html::a(array( - 'href' => $url_prefix . $url, - 'target' => '_blank' - ), rcube::Q($url)) . $suffix); + $attrib = (array)$this->options['link_attribs']; + $attrib['href'] = $url_prefix . $url; + + $i = $this->add(html::a($attrib, rcube::Q($url)) . $suffix); + $this->urls[$i] = $attrib['href']; } // Return valid link for recognized schemes, otherwise // return the unmodified string for unrecognized schemes. - return $i >= 0 ? $this->get_replacement($i) : $matches[0]; + return $i >= 0 ? $prefix . $this->get_replacement($i) : $matches[0]; + } + + /** + * Callback to add an entry to the link index + */ + public function linkref_addindex($matches) + { + $key = $matches[1]; + $this->linkrefs[$key] = $this->urls[$matches[3]]; + + return $this->get_replacement($this->add('['.$key.']')) . $matches[2]; + } + + /** + * Callback to replace link references with real links + */ + public function linkref_callback($matches) + { + $i = 0; + if ($url = $this->linkrefs[$matches[1]]) { + $attrib = (array)$this->options['link_attribs']; + $attrib['href'] = $url; + $i = $this->add(html::a($attrib, rcube::Q($matches[1]))); + } + + return $i > 0 ? '['.$this->get_replacement($i).']' : $matches[0]; } /** * Callback function used to build mailto: links around e-mail strings * * @param array Matches result from preg_replace_callback + * * @return int Index of saved string value */ public function mailto_callback($matches) @@ -139,6 +177,9 @@ // search for patterns like links and e-mail addresses $str = preg_replace_callback($this->link_pattern, array($this, 'link_callback'), $str); $str = preg_replace_callback($this->mailto_pattern, array($this, 'mailto_callback'), $str); + // resolve link references + $str = preg_replace_callback($this->linkref_index, array($this, 'linkref_addindex'), $str); + $str = preg_replace_callback($this->linkref_pattern, array($this, 'linkref_callback'), $str); return $str; } @@ -161,6 +202,9 @@ // "http://example.com/?a[b]=c". However we need to handle // properly situation when a bracket is placed at the end // of the link e.g. "[http://example.com]" + // Yes, this is not perfect handles correctly only paired characters + // but it should work for common cases + if (preg_match('/(\\[|\\])/', $url)) { $in = false; for ($i=0, $len=strlen($url); $i<$len; $i++) { @@ -182,6 +226,28 @@ } } + // Do the same for parentheses + if (preg_match('/(\\(|\\))/', $url)) { + $in = false; + for ($i=0, $len=strlen($url); $i<$len; $i++) { + if ($url[$i] == '(') { + if ($in) + break; + $in = true; + } + else if ($url[$i] == ')') { + if (!$in) + break; + $in = false; + } + } + + if ($i < $len) { + $suffix = substr($url, $i); + $url = substr($url, 0, $i); + } + } + return $suffix; } } -- Gitblit v1.9.1