diff --git a/Crawler.php b/Crawler.php
index 3fd87bd7..0ab6b982 100644
--- a/Crawler.php
+++ b/Crawler.php
@@ -166,20 +166,43 @@ public function addHtmlContent($content, $charset = 'UTF-8')
$dom = new \DOMDocument('1.0', $charset);
$dom->validateOnParse = true;
- if (function_exists('mb_convert_encoding')) {
- $hasError = false;
- set_error_handler(function () use (&$hasError) {
- $hasError = true;
- });
- $tmpContent = @mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
-
- restore_error_handler();
-
- if (!$hasError) {
- $content = $tmpContent;
+ set_error_handler(function () {throw new \Exception();});
+
+ try {
+ // Convert charset to HTML-entities to work around bugs in DOMDocument::loadHTML()
+
+ if (function_exists('mb_convert_encoding')) {
+ $content = mb_convert_encoding($content, 'HTML-ENTITIES', $charset);
+ } elseif (function_exists('iconv')) {
+ $content = preg_replace_callback(
+ '/[\x80-\xFF]+/',
+ function ($m) {
+ $m = unpack('C*', $m[0]);
+ $i = 1;
+ $entities = '';
+
+ while (isset($m[$i])) {
+ if (0xF0 <= $m[$i]) {
+ $c = (($m[$i++] - 0xF0) << 18) + (($m[$i++] - 0x80) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++] - 0x80;
+ } elseif (0xE0 <= $m[$i]) {
+ $c = (($m[$i++] - 0xE0) << 12) + (($m[$i++] - 0x80) << 6) + $m[$i++] - 0x80;
+ } else {
+ $c = (($m[$i++] - 0xC0) << 6) + $m[$i++] - 0x80;
+ }
+
+ $entities .= '&#'.$c.';';
+ }
+
+ return $entities;
+ },
+ iconv($charset, 'UTF-8', $content)
+ );
}
+ } catch (\Exception $e) {
}
+ restore_error_handler();
+
if ('' !== trim($content)) {
@$dom->loadHTML($content);
}
diff --git a/Tests/CrawlerTest.php b/Tests/CrawlerTest.php
index 3c281eb4..9560d069 100755
--- a/Tests/CrawlerTest.php
+++ b/Tests/CrawlerTest.php
@@ -81,6 +81,7 @@ public function testAddHtmlContent()
/**
* @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+ * @requires extension mbstring
*/
public function testAddHtmlContentCharset()
{
@@ -115,6 +116,7 @@ public function testAddHtmlContentUnsupportedCharset()
/**
* @covers Symfony\Component\DomCrawler\Crawler::addHtmlContent
+ * @requires extension mbstring
*/
public function testAddHtmlContentCharsetGbk()
{
@@ -235,7 +237,7 @@ public function testAddContent()
$this->assertEquals('中文', $crawler->filterXPath('//span')->text(), '->addContent() guess wrong charset');
$crawler = new Crawler();
- $crawler->addContent(mb_convert_encoding('日本語', 'SJIS', 'UTF-8'));
+ $crawler->addContent(iconv('UTF-8', 'SJIS', '日本語'));
$this->assertEquals('日本語', $crawler->filterXPath('//body')->text(), '->addContent() can recognize "Shift_JIS" in html5 meta charset tag');
}