Merge pull request coduo#61 from thunderer/string-shortcode-removal

Shortcode removal utilities

composer.json

@@ -18,7 +18,8 @@
         "symfony/intl": "~2.3",
         "symfony/config": "~2.3",
         "symfony/translation": "~2.3",
-        "symfony/yaml": "~2.3"
+        "symfony/yaml": "~2.3",
+        "thunderer/shortcode": "~0.5"
     },
     "require-dev": {
         "phpspec/phpspec": "2.0.*",

src/Coduo/PHPHumanizer/String.php

@@ -3,6 +3,7 @@
 namespace Coduo\PHPHumanizer;
 
 use Coduo\PHPHumanizer\String\Humanize;
+use Coduo\PHPHumanizer\String\ShortcodeProcessor;
 use Coduo\PHPHumanizer\String\TextTruncate;
 use Coduo\PHPHumanizer\String\HtmlTruncate;
 use Coduo\PHPHumanizer\String\WordBreakpoint;
@@ -14,7 +15,7 @@ class String
      * @param bool|true $capitalize
      * @param string $separator
      * @param array $forbiddenWords
-     * 
+     *
      * @return string
      */
     public static function humanize($text, $capitalize = true, $separator = '_', array $forbiddenWords = array())
@@ -26,13 +27,13 @@ public static function humanize($text, $capitalize = true, $separator = '_', arr
      * @param $text
      * @param $charactersCount
      * @param string $append
-     * 
+     *
      * @return string
      */
     public static function truncate($text, $charactersCount, $append = '')
     {
         $truncate = new TextTruncate(new WordBreakpoint(), $append);
-        
+
         return $truncate->truncate($text, $charactersCount);
     }
 
@@ -41,13 +42,27 @@ public static function truncate($text, $charactersCount, $append = '')
      * @param $charactersCount
      * @param string $allowedTags
      * @param string $append
-     * 
+     *
      * @return string
      */
     public static function truncateHtml($text, $charactersCount, $allowedTags = '', $append = '')
     {
         $truncate = new HtmlTruncate(new WordBreakpoint(), $allowedTags, $append);
-        
+
         return $truncate->truncate($text, $charactersCount);
     }
+
+    public static function removeShortcodes($text)
+    {
+        $processor = new ShortcodeProcessor();
+
+        return $processor->removeShortcodes($text);
+    }
+
+    public static function removeShortcodeTags($text)
+    {
+        $processor = new ShortcodeProcessor();
+
+        return $processor->removeShortcodeTags($text);
+    }
 }

src/Coduo/PHPHumanizer/String/ShortcodeProcessor.php

@@ -0,0 +1,51 @@
+<?php
+
+namespace Coduo\PHPHumanizer\String;
+
+use Thunder\Shortcode\HandlerContainer\HandlerContainer;
+use Thunder\Shortcode\Parser\RegularParser;
+use Thunder\Shortcode\Processor\Processor;
+use Thunder\Shortcode\Shortcode\ShortcodeInterface;
+
+class ShortcodeProcessor
+{
+    /**
+     * Removes all shortcodes from given text
+     *
+     * @param string $text
+     *
+     * @return string
+     */
+    public function removeShortcodes($text)
+    {
+        $nullHandler = function() {
+            return null;
+        };
+
+        return $this->createShortcodeProcessor($nullHandler)->process($text);
+    }
+
+    /**
+     * Removes only shortcode tags from given text (leaves their content as it is)
+     *
+     * @param string $text
+     *
+     * @return string
+     */
+    public function removeShortcodeTags($text)
+    {
+        $contentHandler = function(ShortcodeInterface $s) {
+            return $s->getContent();
+        };
+
+        return $this->createShortcodeProcessor($contentHandler)->process($text);
+    }
+
+    private function createShortcodeProcessor($defaultHandler)
+    {
+        $handlers = new HandlerContainer();
+        $handlers->setDefault($defaultHandler);
+
+        return new Processor(new RegularParser(), $handlers);
+    }
+}

tests/Coduo/PHPHumanizer/Tests/StringTest.php

@@ -49,7 +49,44 @@ function test_truncate_string_to_word_closest_to_a_certain_number_of_characters_
     }
 
     /**
+     * @dataProvider removeAllShortcodesProvider
      *
+     * @param $text
+     * @param $expected
+     */
+    function test_remove_all_shortcodes_from_text($text, $expected)
+    {
+        $this->assertEquals($expected, String::removeShortcodes($text));
+    }
+
+    /**
+     * @dataProvider removeShortcodeTagsProvider
+     *
+     * @param $text
+     * @param $expected
+     */
+    function test_remove_only_shortcode_tags_from_text($text, $expected)
+    {
+        $this->assertEquals($expected, String::removeShortcodeTags($text));
+    }
+
+    public function removeAllShortcodesProvider()
+    {
+        return array(
+            array('some [text] containing [shortcodes /] and [stuff]with[/stuff] content', 'some  containing  and  content'),
+            array('some [text] containing [shortcodes /] and [stuff]with[/stuff] content [/text]', 'some '),
+        );
+    }
+
+    public function removeShortcodeTagsProvider()
+    {
+        return array(
+            array('some [text] containing [shortcodes /] and [stuff]with[/stuff] content', 'some  containing  and with content'),
+            array('some [text] containing [shortcodes /] and [stuff]with[/stuff] content [/text]', 'some  containing  and with content '),
+        );
+    }
+
+    /**
      * @return array
      */
     public function humanizeStringProvider()
@@ -73,7 +110,7 @@ public function truncateStringProvider()
     {
         $longText = 'Lorem ipsum dolorem si amet, lorem ipsum. Dolorem sic et nunc.';
         $shortText = 'Short text';
-        
+
         return array(
             array($longText, 'Lorem', 2),
             array($longText, 'Lorem ipsum...', 10, '...'),
@@ -93,19 +130,19 @@ public function truncateStringProvider()
             array($shortText, "Short text", 10, '...')
         );
     }
-    
+
     /**
      *
      * @return array
      */
     public function truncateHtmlStringProvider()
     {
         $text = '<p><b>HyperText Markup Language</b>, commonly referred to as <b>HTML</b>, is the standard <a href="/wiki/Markup_language" title="Markup language">markup language</a> used to create <a href="/wiki/Web_page" title="Web page">web pages</a>.<sup id="cite_ref-1" class="reference"><a href="#cite_note-1"><span>[</span>1<span>]</span></a></sup> <a href="/wiki/Web_browser" title="Web browser">Web browsers</a> can read HTML files and render them into visible or audible web pages. HTML describes the structure of a <a href="/wiki/Website" title="Website">website</a> <a href="/wiki/Semantic" title="Semantic" class="mw-redirect">semantically</a> along with cues for presentation, making it a markup language, rather than a <a href="/wiki/Programming_language" title="Programming language">programming language</a>.</p>';
-        
+
         return array(
-            array($text, 3,  '<b><i><u><em><strong><a><span>',  "<b>HyperText</b>"), 
+            array($text, 3,  '<b><i><u><em><strong><a><span>',  "<b>HyperText</b>"),
             array($text, 12, '<b><i><u><em><strong><a><span>', "<b>HyperText Markup</b>"),
-            array($text, 30, '<b><i><u><em><strong><a><span>', "<b>HyperText Markup Language</b>, commonly"), 
+            array($text, 30, '<b><i><u><em><strong><a><span>', "<b>HyperText Markup Language</b>, commonly"),
             array($text, 50, '<b><i><u><em><strong><a><span>', "<b>HyperText Markup Language</b>, commonly referred to as"),
             array($text, 75, '<b><i><u><em><strong><a><span>', '<b>HyperText Markup Language</b>, commonly referred to as <b>HTML</b>, is the standard <a href="/wiki/Markup_language" title="Markup language">markup</a>'),
             array($text, 100,'<b><i><u><em><strong><a><span>', '<b>HyperText Markup Language</b>, commonly referred to as <b>HTML</b>, is the standard <a href="/wiki/Markup_language" title="Markup language">markup language</a> used to create'),