weblib MDL-22664 html_to_text should not strip images, it should repl…

…ace them by their alt text. Also, an new optional argument to html_to_text to control word-wrapping.
moodle · Aug 4, 2010 · a194c21 · a194c21
1 parent 5be2373
commit a194c21
Show file tree

Hide file tree

Showing 4 changed files with 50 additions and 5 deletions.
diff --git a/lib/html2text.php b/lib/html2text.php
@@ -211,7 +211,7 @@ class html2text
         '-',
         '*',
         '£',
-        'EUR',                                  // Euro sign. � ?
+        'EUR',                                  // Euro sign. € ?
         ' '                                     // Runs of spaces, post-handling
     );
 
@@ -229,6 +229,7 @@ class html2text
         '/<(a) [^>]*href=("|\')([^"\']+)\2[^>]*>(.*?)<\/a>/i',
                                                    // <a href="">
         '/<(th)[^>]*>(.*?)<\/th>/i',               // <th> and </th>
+        '/<(img)[^>]*alt=\"([^>"]+)\"[^>]*>/i',    // <img> with alt
     );
 
    /**
@@ -565,6 +566,8 @@ function _preg_callback($matches)
             return $this->_strtoupper("\n\n". $matches[2] ."\n\n");
         case 'a':
             return $this->_build_link_list($matches[3], $matches[4]);
+        case 'img':
+            return '[' . $matches[2] . ']';
         }
     }
 

diff --git a/lib/html2text_readme.txt b/lib/html2text_readme.txt
@@ -3,3 +3,31 @@ html2text.php is an unmodified copy of a file shipped with the RoundCube project
   http://trac.roundcube.net/log/trunk/roundcubemail/program/lib/html2text.php
 
  -- Francois Marier <francois@catalyst.net.nz>  2009-05-22
+
+
+Modifications
+--------------
+
+1- Don't just strip images, replace them with their alt text.
+
+index b7e3e3e..96ef508 100644
+--- a/lib/html2text.php
++++ b/lib/html2text.php
+@@ -237,6 +237,7 @@ class html2text
+         '/<(a) [^>]*href=("|\')([^"\']+)\2[^>]*>(.*?)<\/a>/i',
+                                                    // <a href="">
+         '/<(th)[^>]*>(.*?)<\/th>/i',               // <th> and </th>
++        '/<(img)[^>]*alt=\"([^>"]+)\"[^>]*>/i',    // <img> with alt
+     );
+
+    /**
+@@ -574,6 +575,8 @@ class html2text
+             return $this->_strtoupper("\n\n". $matches[2] ."\n\n");
+         case 'a':
+             return $this->_build_link_list($matches[3], $matches[4]);
++        case 'img':
++            return '[' . $matches[2] . ']';
+         }
+     }
+
+ -- Tim Hunt 2010-08-04
diff --git a/lib/simpletest/testweblib.php b/lib/simpletest/testweblib.php
@@ -285,6 +285,19 @@ function test_convert_urls_into_links() {
 
         $this->assertEqual($fast_enough, true, 'Timing test: ' . $new_time . 'secs (new) < ' . $old_time . 'secs (old)');
     }
+
+    public function test_html_to_text_simple() {
+        $this->assertEqual("\n\n_Hello_ WORLD!", html_to_text('<p><i>Hello</i> <b>world</b>!</p>'));
+    }
+
+    public function test_html_to_text_image() {
+        $this->assertEqual('[edit]', html_to_text('<img src="edit.png" alt="edit" />'));
+    }
+
+    public function test_html_to_text_nowrap() {
+        $long = "Here is a long string, more than 75 characters long, since by default html_to_text wraps text at 75 chars.";
+        $this->assertEqual($long, html_to_text($long, 0));
+    }
 }
 
 
diff --git a/lib/weblib.php b/lib/weblib.php
@@ -1766,17 +1766,18 @@ function markdown_to_html($text) {
 /**
  * Given HTML text, make it into plain text using external function
  *
- * @global object
  * @param string $html The text to be converted.
- * @return string
+ * @param integer $width Width to wrap the text at. (optional, default 75 which
+ *      is a good value for email. 0 means do not limit line length.)
+ * @return string plain text equivalent of the HTML.
  */
-function html_to_text($html) {
+function html_to_text($html, $width = 75) {
 
     global $CFG;
 
     require_once($CFG->libdir .'/html2text.php');
 
-    $h2t = new html2text($html);
+    $h2t = new html2text($html, false, true, $width);
     $result = $h2t->get_text();
 
     return $result;