Support search highlights when matching URL content

DefaultFormatter:
  - format 'a' tag content and not href attribute
  - format hashtags properly
Markdown(Extra)Formatter:
  - Extend Parsedown to format highlight properly: https://github.com/erusev/parsedown/wiki/Tutorial:-Create-Extensions

Fixes #1681

6 files changed, 89 insertions, 8 deletions

diff --git a/application/formatter/BookmarkDefaultFormatter.php b/application/formatter/BookmarkDefaultFormatter.php
index 7e0afafc..7e93bf71 100644
--- a/application/formatter/BookmarkDefaultFormatter.php
+++ b/application/formatter/BookmarkDefaultFormatter.php
@@ -12,8 +12,8 @@ namespace Shaarli\Formatter;
  */
 class BookmarkDefaultFormatter extends BookmarkFormatter
 {
-    protected const SEARCH_HIGHLIGHT_OPEN = '|@@HIGHLIGHT';
-    protected const SEARCH_HIGHLIGHT_CLOSE = 'HIGHLIGHT@@|';
+    public const SEARCH_HIGHLIGHT_OPEN = '||O_HIGHLIGHT';
+    public const SEARCH_HIGHLIGHT_CLOSE = '||C_HIGHLIGHT';
 
     /**
      * @inheritdoc
diff --git a/application/formatter/BookmarkMarkdownExtraFormatter.php b/application/formatter/BookmarkMarkdownExtraFormatter.php
index 0694b23f..da539bfd 100644
--- a/application/formatter/BookmarkMarkdownExtraFormatter.php
+++ b/application/formatter/BookmarkMarkdownExtraFormatter.php
@@ -3,6 +3,7 @@
 namespace Shaarli\Formatter;
 
 use Shaarli\Config\ConfigManager;
+use Shaarli\Formatter\Parsedown\ShaarliParsedownExtra;
 
 /**
  * Class BookmarkMarkdownExtraFormatter
@@ -18,7 +19,6 @@ class BookmarkMarkdownExtraFormatter extends BookmarkMarkdownFormatter
     public function __construct(ConfigManager $conf, bool $isLoggedIn)
     {
         parent::__construct($conf, $isLoggedIn);
-
-        $this->parsedown = new \ParsedownExtra();
+        $this->parsedown = new ShaarliParsedownExtra();
     }
 }
diff --git a/application/formatter/BookmarkMarkdownFormatter.php b/application/formatter/BookmarkMarkdownFormatter.php
index ee4e8dca..d4dccee6 100644
--- a/application/formatter/BookmarkMarkdownFormatter.php
+++ b/application/formatter/BookmarkMarkdownFormatter.php
@@ -3,6 +3,7 @@
 namespace Shaarli\Formatter;
 
 use Shaarli\Config\ConfigManager;
+use Shaarli\Formatter\Parsedown\ShaarliParsedown;
 
 /**
  * Class BookmarkMarkdownFormatter
@@ -42,7 +43,7 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter
     {
         parent::__construct($conf, $isLoggedIn);
 
-        $this->parsedown = new \Parsedown();
+        $this->parsedown = new ShaarliParsedown();
         $this->escape = $conf->get('security.markdown_escape', true);
         $this->allowedProtocols = $conf->get('security.allowed_protocols', []);
     }
@@ -128,6 +129,9 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter
     protected function formatHashTags($description)
     {
         $indexUrl = ! empty($this->contextData['index_url']) ? $this->contextData['index_url'] : '';
+        $tokens = '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN . ')' .
+            '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE . ')'
+        ;
 
         /*
          * To support unicode: http://stackoverflow.com/a/35498078/1484919
@@ -136,8 +140,15 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter
          * \p{L} - letter from any language
          * \p{Mn} - any non marking space (accents, umlauts, etc)
          */
-        $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}]+)/mui';
-        $replacement = '$1[#$2](' . $indexUrl . './add-tag/$2)';
+        $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}' . $tokens . ']+)/mui';
+        $replacement = function (array $match) use ($indexUrl): string {
+            $cleanMatch = str_replace(
+                BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN,
+                '',
+                str_replace(BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE, '', $match[2])
+            );
+            return $match[1] . '[#' . $match[2] . '](' . $indexUrl . './add-tag/' . $cleanMatch . ')';
+        };
 
         $descriptionLines = explode(PHP_EOL, $description);
         $descriptionOut = '';
@@ -156,7 +167,7 @@ class BookmarkMarkdownFormatter extends BookmarkDefaultFormatter
             }
 
             if (!$codeBlockOn && !$codeLineOn) {
-                $descriptionLine = preg_replace($regex, $replacement, $descriptionLine);
+                $descriptionLine = preg_replace_callback($regex, $replacement, $descriptionLine);
             }
 
             $descriptionOut .= $descriptionLine;
diff --git a/application/formatter/Parsedown/ShaarliParsedown.php b/application/formatter/Parsedown/ShaarliParsedown.php
new file mode 100644
index 00000000..d577bdfa
--- /dev/null
+++ b/application/formatter/Parsedown/ShaarliParsedown.php
@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Formatter\Parsedown;
+
+class ShaarliParsedown extends \Parsedown
+{
+    use ShaarliParsedownTrait;
+}
diff --git a/application/formatter/Parsedown/ShaarliParsedownExtra.php b/application/formatter/Parsedown/ShaarliParsedownExtra.php
new file mode 100644
index 00000000..92ad26ca
--- /dev/null
+++ b/application/formatter/Parsedown/ShaarliParsedownExtra.php
@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Formatter\Parsedown;
+
+class ShaarliParsedownExtra extends \ParsedownExtra
+{
+    use ShaarliParsedownTrait;
+}
diff --git a/application/formatter/Parsedown/ShaarliParsedownTrait.php b/application/formatter/Parsedown/ShaarliParsedownTrait.php
new file mode 100644
index 00000000..e6f4dabb
--- /dev/null
+++ b/application/formatter/Parsedown/ShaarliParsedownTrait.php
@@ -0,0 +1,50 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Formatter\Parsedown;
+
+use Shaarli\Formatter\BookmarkDefaultFormatter as Formatter;
+
+trait ShaarliParsedownTrait
+{
+    protected function inlineLink($excerpt)
+    {
+        return $this->shaarliFormatLink(parent::inlineLink($excerpt), true);
+    }
+
+    protected function inlineUrl($excerpt)
+    {
+        return $this->shaarliFormatLink(parent::inlineUrl($excerpt), false);
+    }
+
+    protected function shaarliFormatLink(?array $link, bool $fullWrap): ?array
+    {
+        if (
+            is_array($link)
+            && strpos($link['element']['attributes']['href'], Formatter::SEARCH_HIGHLIGHT_OPEN) !== false
+            && strpos($link['element']['attributes']['href'], Formatter::SEARCH_HIGHLIGHT_CLOSE) !== false
+        ) {
+            $link['element']['attributes']['href'] = $this->shaarliRemoveSearchTokens(
+                $link['element']['attributes']['href']
+            );
+
+            if ($fullWrap) {
+                $link['element']['text'] = Formatter::SEARCH_HIGHLIGHT_OPEN .
+                    $link['element']['text'] .
+                    Formatter::SEARCH_HIGHLIGHT_CLOSE
+                ;
+            }
+        }
+
+        return $link;
+    }
+
+    protected function shaarliRemoveSearchTokens(string $entry): string
+    {
+        $entry = str_replace(Formatter::SEARCH_HIGHLIGHT_OPEN, '', $entry);
+        $entry = str_replace(Formatter::SEARCH_HIGHLIGHT_CLOSE, '', $entry);
+
+        return $entry;
+    }
+}