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

application/bookmark/LinkUtils.php

@@ -1,6 +1,7 @@
 <?php
 
 use Shaarli\Bookmark\Bookmark;
+use Shaarli\Formatter\BookmarkDefaultFormatter;
 
 /**
  * Extract title from an HTML document.
@@ -98,7 +99,18 @@ function html_extract_tag($tag, $html)
 function text2clickable($text)
 {
     $regex = '!(((?:https?|ftp|file)://|apt:|magnet:)\S+[a-z0-9\(\)]/?)!si';
-    return preg_replace($regex, '<a href="$1">$1</a>', $text);
+    $format = function (array $match): string {
+        return '<a href="' .
+            str_replace(
+                BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN,
+                '',
+                str_replace(BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE, '', $match[1])
+            ) .
+            '">' . $match[1] . '</a>'
+        ;
+    };
+
+    return preg_replace_callback($regex, $format, $text);
 }
 
 /**
@@ -111,6 +123,9 @@ function text2clickable($text)
  */
 function hashtag_autolink($description, $indexUrl = '')
 {
+    $tokens = '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN . ')' .
+              '(?:' . BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE . ')'
+    ;
     /*
      * To support unicode: http://stackoverflow.com/a/35498078/1484919
      * \p{Pc} - to match underscore
@@ -118,9 +133,20 @@ function hashtag_autolink($description, $indexUrl = '')
      * \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';
+    $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}' . $tokens . ']+)/mui';
-    $replacement = '$1<a href="' . $indexUrl . './add-tag/$2" title="Hashtag $2">#$2</a>';
+    $format = function (array $match) use ($indexUrl): string {
-    return preg_replace($regex, $replacement, $description);
+        $cleanMatch = str_replace(
+            BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_OPEN,
+            '',
+            str_replace(BookmarkDefaultFormatter::SEARCH_HIGHLIGHT_CLOSE, '', $match[2])
+        );
+        return $match[1] . '<a href="' . $indexUrl . './add-tag/' . $cleanMatch . '"' .
+            ' title="Hashtag ' . $cleanMatch . '">' .
+                '#' . $match[2] .
+        '</a>';
+    };
+
+    return preg_replace_callback($regex, $format, $description);
 }
 
 /**

application/formatter/BookmarkDefaultFormatter.php

@@ -12,8 +12,8 @@
  */
 class BookmarkDefaultFormatter extends BookmarkFormatter
 {
-    protected const SEARCH_HIGHLIGHT_OPEN = '|@@HIGHLIGHT';
+    public const SEARCH_HIGHLIGHT_OPEN = '||O_HIGHLIGHT';
-    protected const SEARCH_HIGHLIGHT_CLOSE = 'HIGHLIGHT@@|';
+    public const SEARCH_HIGHLIGHT_CLOSE = '||C_HIGHLIGHT';
 
     /**
      * @inheritdoc

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 ShaarliParsedownExtra();
-        $this->parsedown = new \ParsedownExtra();
     }
 }

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 @@ public function __construct(ConfigManager $conf, bool $isLoggedIn)
     {
         parent::__construct($conf, $isLoggedIn);
 
-        $this->parsedown = new \Parsedown();
+        $this->parsedown = new ShaarliParsedown();
         $this->escape = $conf->get('security.markdown_escape', true);
         $this->allowedProtocols = $conf->get('security.allowed_protocols', []);
     }
@@ -128,6 +129,9 @@ function ($match) use ($allowedProtocols, $indexUrl) {
     protected function formatHashTags($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 @@ protected function formatHashTags($description)
          * \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';
+        $regex = '/(^|\s)#([\p{Pc}\p{N}\p{L}\p{Mn}' . $tokens . ']+)/mui';
-        $replacement = '$1[#$2](' . $indexUrl . './add-tag/$2)';
+        $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);
         $descriptionOut = '';
@@ -156,7 +167,7 @@ protected function formatHashTags($description)
             }
 
             if (!$codeBlockOn && !$codeLineOn) {
-                $descriptionLine = preg_replace($regex, $replacement, $descriptionLine);
+                $descriptionLine = preg_replace_callback($regex, $replacement, $descriptionLine);
             }
 
             $descriptionOut .= $descriptionLine;

application/formatter/Parsedown/ShaarliParsedown.php

@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Formatter\Parsedown;
+
+class ShaarliParsedown extends \Parsedown
+{
+    use ShaarliParsedownTrait;
+}

application/formatter/Parsedown/ShaarliParsedownExtra.php

@@ -0,0 +1,10 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Shaarli\Formatter\Parsedown;
+
+class ShaarliParsedownExtra extends \ParsedownExtra
+{
+    use ShaarliParsedownTrait;
+}

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;
+    }
+}

application/front/controller/visitor/BookmarkListController.php

@@ -33,6 +33,7 @@ public function index(Request $request, Response $response): Response
 
         $formatter = $this->container->formatterFactory->getFormatter();
         $formatter->addContextData('base_path', $this->container->basePath);
+        $formatter->addContextData('index_url', index_url($this->container->environment));
 
         $searchTags = normalize_spaces($request->getParam('searchtags') ?? '');
         $searchTerm = escape(normalize_spaces($request->getParam('searchterm') ?? ''));
@@ -157,6 +158,7 @@ public function permalink(Request $request, Response $response, array $args): Re
 
         $formatter = $this->container->formatterFactory->getFormatter();
         $formatter->addContextData('base_path', $this->container->basePath);
+        $formatter->addContextData('index_url', index_url($this->container->environment));
 
         $data = array_merge(
             $this->initializeTemplateVars(),

tests/formatter/BookmarkDefaultFormatterTest.php

@@ -211,13 +211,17 @@ public function testFormatDescriptionWithSearchHighlight(): void
         $this->formatter = new BookmarkDefaultFormatter($this->conf, false);
 
         $bookmark = new Bookmark();
-        $bookmark->setDescription('This guide extends and expands on PSR-1, the basic coding standard.');
+        $bookmark->setDescription(
+            'This guide extends and expands on PSR-1, the basic coding standard.' . PHP_EOL .
+            'https://www.php-fig.org/psr/psr-1/'
+        );
         $bookmark->addAdditionalContentEntry(
             'search_highlight',
             ['description' => [
                 ['start' => 0, 'end' => 10], // "This guide"
                 ['start' => 45, 'end' => 50], // basic
                 ['start' => 58, 'end' => 67], // standard.
+                ['start' => 84, 'end' => 87], // fig
             ]]
         );
 
@@ -226,7 +230,10 @@ public function testFormatDescriptionWithSearchHighlight(): void
         $this->assertSame(
             '<span class="search-highlight">This guide</span> extends and expands on PSR-1, the ' .
             '<span class="search-highlight">basic</span> coding ' .
-            '<span class="search-highlight">standard.</span>',
+            '<span class="search-highlight">standard.</span><br />' . PHP_EOL .
+            '<a href="https://www.php-fig.org/psr/psr-1/">' .
+                'https://www.php-<span class="search-highlight">fig</span>.org/psr/psr-1/' .
+            '</a>',
             $link['description']
         );
     }

tests/formatter/BookmarkMarkdownFormatterTest.php

@@ -132,6 +132,49 @@ public function testFormatDescription()
         $this->assertEquals($description, $link['description']);
     }
 
+    /**
+     * Make sure that the description is properly formatted by the default formatter.
+     */
+    public function testFormatDescriptionWithSearchHighlight()
+    {
+        $description = 'This a <strong>description</strong>'. PHP_EOL;
+        $description .= 'text https://sub.domain.tld?query=here&for=real#hash more text'. PHP_EOL;
+        $description .= 'Also, there is an #hashtag added'. PHP_EOL;
+        $description .= '    A  N  D KEEP     SPACES    !   '. PHP_EOL;
+        $description .= 'And [yet another link](https://other.domain.tld)'. PHP_EOL;
+
+        $bookmark = new Bookmark();
+        $bookmark->setDescription($description);
+        $bookmark->addAdditionalContentEntry(
+            'search_highlight',
+            ['description' => [
+                ['start' => 18, 'end' => 26], // cription
+                ['start' => 49, 'end' => 52], // sub
+                ['start' => 84, 'end' => 88], // hash
+                ['start' => 118, 'end' => 123], // hasht
+                ['start' => 203, 'end' => 215], // other.domain
+            ]]
+        );
+
+        $link = $this->formatter->format($bookmark);
+
+        $description = '<div class="markdown"><p>';
+        $description .= 'This a &lt;strong&gt;des<span class="search-highlight">cription</span>&lt;/strong&gt;<br />' .
+            PHP_EOL;
+        $url = 'https://sub.domain.tld?query=here&amp;for=real#hash';
+        $highlighted = 'https://<span class="search-highlight">sub</span>.domain.tld';
+        $highlighted .= '?query=here&amp;for=real#<span class="search-highlight">hash</span>';
+        $description .= 'text <a href="'. $url .'">'. $highlighted .'</a> more text<br />'. PHP_EOL;
+        $description .= 'Also, there is an <a href="./add-tag/hashtag">#<span class="search-highlight">hasht</span>' .
+            'ag</a> added<br />'. PHP_EOL;
+        $description .= 'A  N  D KEEP     SPACES    !<br />' . PHP_EOL;
+        $description .= 'And <a href="https://other.domain.tld">' .
+            '<span class="search-highlight">yet another link</span></a>';
+        $description .= '</p></div>';
+
+        $this->assertEquals($description, $link['description']);
+    }
+
     /**
      * Test formatting URL with an index_url set
      * It should prepend relative links.