Merge pull request #89 from on2/strikethrough

Handle <del> and <ins> tags

Author: mtibben

src/Html2Text.php

@@ -61,6 +61,7 @@ class Html2Text
         '/<style\b[^>]*>.*?<\/style>/i',                  // <style>s -- which strip_tags supposedly has problems with
         '/<i\b[^>]*>(.*?)<\/i>/i',                        // <i>
         '/<em\b[^>]*>(.*?)<\/em>/i',                      // <em>
+        '/<ins\b[^>]*>(.*?)<\/ins>/i',                    // <ins>
         '/(<ul\b[^>]*>|<\/ul>)/i',                        // <ul> and </ul>
         '/(<ol\b[^>]*>|<\/ol>)/i',                        // <ol> and </ol>
         '/(<dl\b[^>]*>|<\/dl>)/i',                        // <dl> and </dl>
@@ -91,6 +92,7 @@ class Html2Text
         '',                              // <style>s -- which strip_tags supposedly has problems with
         '_\\1_',                         // <i>
         '_\\1_',                         // <em>
+        '_\\1_',                         // <ins>
         "\n\n",                          // <ul> and </ul>
         "\n\n",                          // <ol> and </ol>
         "\n\n",                          // <dl> and </dl>
@@ -148,6 +150,7 @@ class Html2Text
         '/<(br)[^>]*>[ ]*/i',                                    // <br> with leading whitespace after the newline.
         '/<(b)( [^>]*)?>(.*?)<\/b>/i',                           // <b>
         '/<(strong)( [^>]*)?>(.*?)<\/strong>/i',                 // <strong>
+        '/<(del)( [^>]*)?>(.*?)<\/del>/i',                       // <del>
         '/<(th)( [^>]*)?>(.*?)<\/th>/i',                         // <th> and </th>
         '/<(a) [^>]*href=("|\')([^"\']+)\2([^>]*)>(.*?)<\/a>/i'  // <a href="">
     );
@@ -226,6 +229,8 @@ class Html2Text
         'width' => 70,          //  Maximum width of the formatted text, in columns.
                                 //  Set this value to 0 (or less) to ignore word wrapping
                                 //  and not constrain text to a fixed-width column.
+
+        'strikethough_del' => false,  // use a combining character so that <del> appears struck-through.
     );
 
     private function legacyConstruct($html = '', $fromFile = false, array $options = array())
@@ -575,6 +580,8 @@ protected function pregCallback($matches)
             case 'b':
             case 'strong':
                 return $this->toupper($matches[3]);
+            case 'del':
+                return $this->tostrike($matches[3]);
             case 'th':
                 return $this->toupper("\t\t" . $matches[3] . "\n");
             case 'h':
@@ -640,4 +647,24 @@ protected function strtoupper($str)
 
         return $str;
     }
+
+    /**
+     * Helper function for DEL conversion.
+     *
+     * @param  string $text HTML content
+     * @return string Converted text
+     */
+    protected function tostrike($str)
+    {
+        if (self::ENCODING != 'UTF-8' || $this->options['strikethough_del'] !== true) {
+            return $str;
+        }
+        $rtn = '';
+        for ($i = 0; $i < mb_strlen($str); $i++) {
+            $chr = mb_substr($str, $i, 1);
+            $combiningChr = chr(0xC0 | 0x336 >> 6). chr(0x80 | 0x336 & 0x3F);
+            $rtn .= $chr . $combiningChr;
+        }
+        return $rtn;
+    }
 }

test/DelTest.php

@@ -0,0 +1,15 @@
+<?php
+
+namespace Html2Text;
+
+class DelTest extends \PHPUnit_Framework_TestCase
+{
+    public function testDel()
+    {
+        $html = 'My <del>Résumé</del> Curriculum Vitæ';
+        $expected = 'My R̶é̶s̶u̶m̶é̶ Curriculum Vitæ';
+
+        $html2text = new Html2Text($html, array('strikethough_del' => true));
+        $this->assertEquals($expected, $html2text->getText());
+    }
+}

test/InsTest.php

@@ -0,0 +1,15 @@
+<?php
+
+namespace Html2Text;
+
+class InsTest extends \PHPUnit_Framework_TestCase
+{
+    public function testIns()
+    {
+        $html = 'This is <ins>inserted</ins>';
+        $expected = 'This is _inserted_';
+
+        $html2text = new Html2Text($html);
+        $this->assertEquals($expected, $html2text->getText());
+    }
+}