Merge pull request #18 from phergie/hotfix/url-title

svpernova09 · web-flow · commit 16badb1d4a70 · 2018-10-07T19:55:19.000-05:00
Make Mime\Html-&gt;extract() title parsing a bit better
diff --git a/src/Mime/Html.php b/src/Mime/Html.php
@@ -41,7 +41,7 @@ public function matches($mimeType)
      */
     public function extract(array $replacements, UrlInterface $url)
     {
-        if (preg_match('#<title[^>]*>(.*?)</title>#is', $url->getBody(), $match)) {
+        if (preg_match('#<title[^>]*>([^<]+)</title>#is', $url->getBody(), $match)) {
             $replacements['%composed-title%'] =
             $replacements['%title%'] =
             preg_replace('/[\s\v]+/', ' ', trim(html_entity_decode($match[1], ENT_QUOTES | ENT_HTML5)));
diff --git a/tests/Mime/HtmlTest.php b/tests/Mime/HtmlTest.php
@@ -60,6 +60,13 @@ public function testExtractProvider() {
                 ),
                 new Url('', '<html><title>foo&#39;s w&ouml;rk</title></html></html>', array(), 200, 1),
             ),
+            array(
+                array(
+                    '%title%' => 'Avatar: The Last Airbender Is Returning as a New Netflix Series',
+                    '%composed-title%' => 'Avatar: The Last Airbender Is Returning as a New Netflix Series',
+                ),
+                new Url('', '<title>pq-earther<\/title><path /><title>Avatar: The Last Airbender Is Returning as a New Netflix Series</title>', array(), 200, 1),
+            ),
             array(
                 array(),
                 new Url('', '', array(), 200, 1),

Original file line number	Diff line number	Diff line change
`@@ -41,7 +41,7 @@ public function matches($mimeType)`
`41`	`41`	`*/`
`42`	`42`	`public function extract(array $replacements, UrlInterface $url)`
`43`	`43`	`{`
`44`		`- if (preg_match('#<title[^>]>(.?)</title>#is', $url->getBody(), $match)) {`
	`44`	`+ if (preg_match('#<title[^>]*>([^<]+)</title>#is', $url->getBody(), $match)) {`
`45`	`45`	`$replacements['%composed-title%'] =`
`46`	`46`	`$replacements['%title%'] =`
`47`	`47`	`preg_replace('/[\s\v]+/', ' ', trim(html_entity_decode($match[1], ENT_QUOTES \| ENT_HTML5)));`