[Core] Better link encryption

Pushword · Feb 9, 2022 · a1a71b9 · a1a71b9
1 parent 470e970
commit a1a71b9
Show file tree

Hide file tree

Showing 6 changed files with 36 additions and 9 deletions.
diff --git a/docs/contribute.xml b/docs/contribute.xml
@@ -19,7 +19,7 @@
                     <title><![CDATA[Roadmap]]></title>
                     <link>https://pushword.piedweb.com/roadmap</link>
                     <guid>https://pushword.piedweb.com/roadmap</guid>
-                    <pubDate>Fri, 04 Feb 2022 09:55:17 +0100</pubDate>
+                    <pubDate>Wed, 09 Feb 2022 09:36:24 +0100</pubDate>
                 </item>
 
     </channel>

diff --git a/docs/feed.xml b/docs/feed.xml
@@ -13,7 +13,7 @@
                     <title><![CDATA[Roadmap]]></title>
                     <link>https://pushword.piedweb.com/roadmap</link>
                     <guid>https://pushword.piedweb.com/roadmap</guid>
-                    <pubDate>Fri, 04 Feb 2022 09:55:17 +0100</pubDate>
+                    <pubDate>Wed, 09 Feb 2022 09:36:24 +0100</pubDate>
                 </item>
                                                                 <item>
                     <title><![CDATA[Upgrade Guide]]></title>

diff --git a/docs/roadmap.html b/docs/roadmap.html
diff --git a/docs/search.json b/docs/search.json
diff --git a/packages/core/src/Component/EntityFilter/Filter/HtmlEncryptedLink.php b/packages/core/src/Component/EntityFilter/Filter/HtmlEncryptedLink.php
@@ -2,12 +2,21 @@
 
 namespace Pushword\Core\Component\EntityFilter\Filter;
 
+use Exception;
+
 final class HtmlEncryptedLink extends EncryptedLink
 {
     /**
      * @var string
+     *             Attr Regex :     (?<=href=)((?P<hrefQuote>'|")(?P<href>.*?)(?P=hrefQuote)|(?P<href>[^"'>][^> \r\n\t\f\v]*))
+     *             Attr Regex :     (?<=rel=)((?P<relQuote>'|")encrypt(?P=relQuote)|encrypt)
+     *             Between Attr :   \s+(?:[^>]*\s{0,1})
+     *             or between Tag and attr
+     *             End Attr :       (?:[^>]*)?
+     *             /(<a\s+(?:[^>]*\s{0,1})(HREF-ESPACE-ENCRYPT|ENCRYPT-ESPACE-HREF)(?:[^>]*)?>(?P<anchor>((?!<\/a>).)*)<\/a>)/iJ
+     *             /(<a\s+(?:[^>]*\s{0,1})((?<=href=)((?P<hrefQuote>'|")(?P<href>.*?)(?P=hrefQuote)|(?P<href>[^"'>][^> \r\n\t\f\v]*))\s+(?:[^>]*\s{0,1})(?<=rel=)((?P<relQuote>'|")encrypt(?P=relQuote)|encrypt)|(?<=rel=)((?P<relQuote>'|")encrypt(?P=relQuote)|encrypt)\s+(?:[^>]*\s{0,1})(?<=href=)((?P<hrefQuote>'|")(?P<href>.*?)(?P=hrefQuote)|(?P<href>[^"'>][^> \r\n\t\f\v]*)))(?:[^>]*)?>(?P<anchor>((?!<\/a>).)*)<\/a>)/iJ
      */
-    public const HTML_REGEX = '/(<a\s+(?:[^>]*?\s+)?href=(["\'])(?P<href>((?!\2).)*)\2(?:[^>]*?\s+)?rel=(["\'])encrypt\5(?:[^>]*?\s+)?>(?P<anchor>((?!<\/a>).)*)<\/a>)/i';
+    public const HTML_REGEX = '/(<a\s+(?:[^>]*\s{0,1})((?<=href=)((?P<hrefQuote>\'|")(?P<href1>.*?)(?P=hrefQuote)|(?P<href2>[^"\'>][^> \r\n\t\f\v]*))\s+(?:[^>]*\s{0,1})(?<=rel=)((?P<relQuote>\'|")encrypt(?P=relQuote)|encrypt)|(?<=rel=)((?P<relQuote>\'|")encrypt(?P=relQuote)|encrypt)\s+(?:[^>]*\s{0,1})(?<=href=)((?P<hrefQuote>\'|")(?P<href3>.*?)(?P=hrefQuote)|(?P<href4>[^"\'>][^> \r\n\t\f\v]*)))(?:[^>]*)?>(?P<anchor>((?!<\/a>).)*)<\/a>)/iJ';
 
     /**
      * @var string
@@ -32,7 +41,7 @@ public function convertHtmlRelEncryptedLink(string $body): string
             return $body;
         }
 
-        return $this->replaceRelEncryptedLink($body, $matches, self::HTML_REGEX_HREF_KEY, self::HTML_REGEX_ANCHOR_KEY);
+        return $this->replaceRelEncryptedLink($body, $matches);
     }
 
     private function extractClass(string $openingTag): string
@@ -43,16 +52,34 @@ private function extractClass(string $openingTag): string
     /**
      * @param array<(string|int), array<int, string>> $matches
      */
-    private function replaceRelEncryptedLink(string $body, array $matches, string $hrefKey = '2', string $anchorKey = '1'): string
+    private function replaceRelEncryptedLink(string $body, array $matches): string
     {
         $nbrMatch = \count($matches[0]);
         for ($k = 0; $k < $nbrMatch; ++$k) {
             $attr = $this->extractClass($matches[1][$k]);
             $attr = '' !== $attr ? ['class' => $attr] : [];
-            $link = $this->renderLink($matches[$anchorKey][$k], $matches[$hrefKey][$k], $attr);
+            $link = $this->renderLink(
+                $matches[self::HTML_REGEX_ANCHOR_KEY][$k],
+                $this->getHrefValue($matches, $k),
+                $attr
+            );
             $body = str_replace($matches[0][$k],  $link, $body);
         }
 
         return $body;
     }
+
+    /**
+     * @param array<(string|int), array<int, string>> $matches
+     */
+    private function getHrefValue(array $matches, int $k): string
+    {
+        for ($i = 1; $i < 5; ++$i) {
+            if ('' !== $matches[self::HTML_REGEX_HREF_KEY.$i][$k]) {
+                return $matches[self::HTML_REGEX_HREF_KEY.$i][$k];
+            }
+        }
+
+        throw new Exception();
+    }
 }
diff --git a/packages/docs/content/roadmap.md b/packages/docs/content/roadmap.md
@@ -9,7 +9,7 @@ parent: contribute
 
 -   Implement ESlint
 
--   [Core] Rewrite MediaTrait/MediaSlugTrait and MediaListener (tag reference before starting refactoring 0.0.859)
+-   [Core] Rewrite MediaTrait/MediaSlugTrait and MediaListener (tag reference before starting refactoring 0.0.859)n
 
 -   [AdminBlockEditor] Add Layout https://github.com/hata6502/editorjs-layout#readme