MDL-17542 lib/html2text: update to latest upstream version (r2374)

from http://trac.roundcube.net/log/trunk/roundcubemail/program/lib/html2text.php
This commit is contained in:
fmarier 2009-05-22 02:10:22 +00:00
parent db26acd41d
commit ea2678d8b9
3 changed files with 51 additions and 50 deletions

View file

@ -176,7 +176,6 @@ class html2text
'/&(bull|#149|#8226);/i', // Bullet '/&(bull|#149|#8226);/i', // Bullet
'/&(pound|#163);/i', // Pound sign '/&(pound|#163);/i', // Pound sign
'/&(euro|#8364);/i', // Euro sign '/&(euro|#8364);/i', // Euro sign
'/&[^&;]+;/i', // Unknown/unhandled entities
'/[ ]{2,}/' // Runs of spaces, post-handling '/[ ]{2,}/' // Runs of spaces, post-handling
); );
@ -220,7 +219,6 @@ class html2text
'*', '*',
'£', '£',
'EUR', // Euro sign. € ? 'EUR', // Euro sign. € ?
'', // Unknown/unhandled entities
' ' // Runs of spaces, post-handling ' ' // Runs of spaces, post-handling
); );
@ -468,13 +466,16 @@ class html2text
// Convert <PRE> // Convert <PRE>
$this->_convert_pre($text); $this->_convert_pre($text);
// Replace known html entities
$text = html_entity_decode($text, ENT_COMPAT, 'UTF-8');
// Run our defined search-and-replace // Run our defined search-and-replace
$text = preg_replace($this->search, $this->replace, $text); $text = preg_replace($this->search, $this->replace, $text);
$text = preg_replace_callback($this->callback_search, array('html2text', '_preg_callback'), $text); $text = preg_replace_callback($this->callback_search, array('html2text', '_preg_callback'), $text);
// Replace known html entities
$text = html_entity_decode($text, ENT_COMPAT, 'UTF-8');
// Remove unknown/unhandled entities (this cannot be done in search-and-replace block)
$text = preg_replace('/&[^&;]+;/i', '', $text);
// Strip any other HTML tags // Strip any other HTML tags
$text = strip_tags($text, $this->allowed_tags); $text = strip_tags($text, $this->allowed_tags);
@ -546,8 +547,7 @@ class html2text
*/ */
function _convert_pre(&$text) function _convert_pre(&$text)
{ {
while(preg_match('/<pre[^>]*>(.*)<\/pre>/ismU', $text, $matches)) while(preg_match('/<pre[^>]*>(.*)<\/pre>/ismU', $text, $matches)) {
{
$result = preg_replace($this->pre_search, $this->pre_replace, $matches[1]); $result = preg_replace($this->pre_search, $this->pre_replace, $matches[1]);
$text = preg_replace('/<pre[^>]*>.*<\/pre>/ismU', '<div><br>' . $result . '<br></div>', $text, 1); $text = preg_replace('/<pre[^>]*>.*<\/pre>/ismU', '<div><br>' . $result . '<br></div>', $text, 1);
} }
@ -562,8 +562,7 @@ class html2text
*/ */
function _preg_callback($matches) function _preg_callback($matches)
{ {
switch($matches[1]) switch($matches[1]) {
{
case 'b': case 'b':
case 'strong': case 'strong':
return $this->_strtoupper($matches[2]); return $this->_strtoupper($matches[2]);

5
lib/html2text_readme.txt Normal file
View file

@ -0,0 +1,5 @@
html2text.php is an unmodified copy of a file shipped with the RoundCube project:
http://trac.roundcube.net/log/trunk/roundcubemail/program/lib/html2text.php
-- Francois Marier <francois@catalyst.net.nz> 2009-05-22

View file

@ -1986,9 +1986,6 @@ function html_to_text($html) {
$h2t = new html2text($html); $h2t = new html2text($html);
$result = $h2t->get_text(); $result = $h2t->get_text();
// html2text does not fix HTML entities so handle those here.
$result = trim(html_entity_decode($result, ENT_NOQUOTES, 'UTF-8'));
return $result; return $result;
} }