danog/MadelineProto
2
2
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Fix HTML parser

Browse Source
This commit is contained in:
Daniil Gentili 2019-06-05 21:09:25 +02:00
parent 864172ce74
commit b59e62a725
1 changed files with 26 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

38
src/danog/MadelineProto/TL/Conversion/BotAPI.php
View File

@ -619,8 +619,8 @@ trait BotAPI
} }
$total = 0; $total = 0;
foreach ($multiple_args as $args) { foreach ($multiple_args as $args) {
if (count($args['entities']) > 100) { if (count($args['entities']) > $max_entity_length) {
$total += count($args['entities']) - 100; $total += count($args['entities']) - $max_entity_length;
} }
$c = 0; $c = 0;
foreach ($args['entities'] as $entity) { foreach ($args['entities'] as $entity) {
@ -628,7 +628,7 @@ trait BotAPI
$c += strlen($entity['url']); $c += strlen($entity['url']);
} }
} }
if ($c >= 8110) { if ($c >= $max_entity_size) {
$this->logger->logger('Entity size limit possibly exceeded, you may get an error indicating that the entities are too long. Reduce the number of entities and/or size of the URLs used.', Logger::FATAL_ERROR); $this->logger->logger('Entity size limit possibly exceeded, you may get an error indicating that the entities are too long. Reduce the number of entities and/or size of the URLs used.', Logger::FATAL_ERROR);
} }
} }
@ -657,25 +657,39 @@ trait BotAPI
public function html_fixtags($text) public function html_fixtags($text)
{ {
$diff = 0;
preg_match_all('#(.*?)(<(a|b|\bstrong\b|\bem\b|i|\bcode\b|\bpre\b)[^>]*>)(.*?)(<\s*/\s*\3>)#is', $text, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE); preg_match_all('#(.*?)(<(a|b|\bstrong\b|\bem\b|i|\bcode\b|\bpre\b)[^>]*>)(.*?)(<\s*/\s*\3>)#is', $text, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
if ($matches) { if ($matches) {
foreach ($matches as $match) { foreach ($matches as $match) {
if (trim($match[1][0]) != '') { if (trim($match[1][0]) != '') {
$temp = substr($text, 0, $match[1][1]); $mod = htmlentities($match[1][0]);
$temp .= htmlentities($match[1][0]);
$temp .= substr($text, $match[1][1] + strlen($match[1][0])); $temp = substr($text, 0, $match[1][1] + $diff);
$temp .= $mod;
$temp .= substr($text, $match[1][1] + $diff + strlen($match[1][0]));
$diff += strlen($mod) - strlen($match[1][0]);
$text = $temp; $text = $temp;
} }
$temp = substr($text, 0, $match[4][1]); $mod = htmlentities($match[4][0]);
$temp .= htmlentities($match[4][0]);
$temp .= substr($text, $match[4][1] + strlen($match[4][0])); $temp = substr($text, 0, $match[4][1] + $diff);
$temp .= $mod;
$temp .= substr($text, $match[4][1] + $diff + strlen($match[4][0]));
$diff += strlen($mod) - strlen($match[4][0]);
$text = $temp; $text = $temp;
} }
$diff = 0;
preg_match_all('#<a\s*href=("|\')(.+?)("|\')\s*>#is', $text, $matches, PREG_OFFSET_CAPTURE); preg_match_all('#<a\s*href=("|\')(.+?)("|\')\s*>#is', $text, $matches, PREG_OFFSET_CAPTURE);
foreach ($matches[2] as $match) { foreach ($matches[2] as $match) {
$temp = substr($text, 0, $match[1]); $mod = htmlentities($match[0]);
$temp .= htmlentities($match[0]); $temp = substr($text, 0, $match[1] + $diff);
$temp .= substr($text, $match[1] + strlen($match[0])); $temp .= $mod;
$temp .= substr($text, $match[1] + $diff + strlen($match[0]));
$diff += strlen($mod) - strlen($match[0]);
$text = $temp; $text = $temp;
} }