Add parsing of html image in base64 format (#1382)

* increased test coverage of new lines
* added exception control to file_get_contents error
* update changelog

Author: javier

CHANGELOG.md

@@ -23,6 +23,7 @@ v0.15.0 (?? ??? 2018)
 - Add support for table indent (tblInd) @Trainmaster #1343
 - Added parsing of internal links in HTML reader @lalop #1336
 - Several improvements to charts @JAEK-S #1332
+- Add parsing of html image in base64 format @jgpATs2w #1382
 
 ### Fixed
 - Fix reading of docx default style - @troosan #1238

samples/resources/Sample_30_ReadHTML.html

@@ -11,5 +11,15 @@ <h1>Adding element via HTML</h1>
 <ul><li>Item 1</li><li>Item 2</li><ul><li>Item 2.1</li><li>Item 2.1</li></ul></ul>
 <p>Ordered (numbered) list:</p>
 <ol><li>Item 1</li><li>Item 2</li></ol>
+
+<p style="line-height:2">Double height</p>
+
+<h2>Includes images</h2>
+<img src="https://phpword.readthedocs.io/en/latest/_images/phpword.png" alt=""/>
+
+<img src="https://localhost/gev/desarrollo/actividades/pruebas_14/5b064503587f7.jpeg" name="Imagen 12" align="bottom" width="208" height="183" border="0"/>
+<img src="http://localhost/gev/desarrollo/actividades/pruebas_14/5b064503589db.png" name="Imagen 13" align="bottom" width="143" height="202" border="0"/>
+<img src="http://localhost/gev/desarrollo/actividades/pruebas_14/5b0645035aac8.jpeg" name="Imagen 14" align="bottom" width="194" height="188" border="0"/>
+
 </body>
 </html>

samples/results/.gitignore

@@ -20,6 +20,7 @@
 use PhpOffice\PhpWord\Element\AbstractContainer;
 use PhpOffice\PhpWord\Element\Row;
 use PhpOffice\PhpWord\Element\Table;
+use PhpOffice\PhpWord\Settings;
 use PhpOffice\PhpWord\SimpleType\Jc;
 use PhpOffice\PhpWord\SimpleType\NumberFormat;
 
@@ -32,6 +33,7 @@ class Html
 {
     private static $listIndex = 0;
     private static $xpath;
+    private static $options;
 
     /**
      * Add HTML parts.
@@ -44,13 +46,17 @@ class Html
      * @param string $html The code to parse
      * @param bool $fullHTML If it's a full HTML, no need to add 'body' tag
      * @param bool $preserveWhiteSpace If false, the whitespaces between nodes will be removed
+     * @param array $options:
+     *                + IMG_SRC_SEARCH: optional to speed up images loading from remote url when files can be found locally
+     *                + IMG_SRC_REPLACE: optional to speed up images loading from remote url when files can be found locally
      */
-    public static function addHtml($element, $html, $fullHTML = false, $preserveWhiteSpace = true)
+    public static function addHtml($element, $html, $fullHTML = false, $preserveWhiteSpace = true, $options = null)
     {
         /*
          * @todo parse $stylesheet for default styles.  Should result in an array based on id, class and element,
          * which could be applied when such an element occurs in the parseNode function.
          */
+        self::$options = $options;
 
         // Preprocess: remove all line ends, decode HTML entity,
         // fix ampersand and angle brackets and add body tag for HTML fragments
@@ -141,6 +147,7 @@ protected static function parseNode($node, $element, $styles = array(), $data =
             'sup'       => array('Property',    null,   null,       $styles,    null,   'superScript',  true),
             'sub'       => array('Property',    null,   null,       $styles,    null,   'subScript',    true),
             'span'      => array('Span',        $node,  null,       $styles,    null,   null,           null),
+            'font'      => array('Span',        $node,  null,       $styles,    null,   null,           null),
             'table'     => array('Table',       $node,  $element,   $styles,    null,   null,           null),
             'tr'        => array('Row',         $node,  $element,   $styles,    null,   null,           null),
             'td'        => array('Cell',        $node,  $element,   $styles,    null,   null,           null),
@@ -648,7 +655,52 @@ private static function parseImage($node, $element)
                     break;
             }
         }
-        $newElement = $element->addImage($src, $style);
+        $originSrc = $src;
+        if (strpos($src, 'data:image') !== false) {
+            $tmpDir = Settings::getTempDir() . '/';
+
+            $match = array();
+            preg_match('/data:image\/(\w+);base64,(.+)/', $src, $match);
+
+            $src = $imgFile = $tmpDir . uniqid() . '.' . $match[1];
+
+            $ifp = fopen($imgFile, 'wb');
+
+            if ($ifp !== false) {
+                fwrite($ifp, base64_decode($match[2]));
+                fclose($ifp);
+            }
+        }
+        $src = urldecode($src);
+
+        if (!is_file($src)
+            && !is_null(self::$options)
+            && isset(self::$options['IMG_SRC_SEARCH'])
+            && isset(self::$options['IMG_SRC_REPLACE'])) {
+            $src = str_replace(self::$options['IMG_SRC_SEARCH'], self::$options['IMG_SRC_REPLACE'], $src);
+        }
+
+        if (!is_file($src)) {
+            if ($imgBlob = @file_get_contents($src)) {
+                $tmpDir = Settings::getTempDir() . '/';
+                $match = array();
+                preg_match('/.+\.(\w+)$/', $src, $match);
+                $src = $tmpDir . uniqid() . '.' . $match[1];
+
+                $ifp = fopen($src, 'wb');
+
+                if ($ifp !== false) {
+                    fwrite($ifp, $imgBlob);
+                    fclose($ifp);
+                }
+            }
+        }
+
+        if (is_file($src)) {
+            $newElement = $element->addImage($src, $style);
+        } else {
+            throw new \Exception("Could not load image $originSrc");
+        }
 
         return $newElement;
     }