Merge pull request #1485 from Timanx/develop

Fix parsing of Office 365 documents

Author: troosan

CHANGELOG.md

@@ -11,6 +11,7 @@ v0.16.0 (xx xxx 2018)
 - Fix regex in `cloneBlock` function @nicoder #1269
 - HTML Title Writer loses text when Title contains a TextRun instead a string. @begnini #1436
 - RTF writer: Round getPageSizeW and getPageSizeH to avoid decimals @Patrick64 #1493
+- Fix parsing of Office 365 documents @Timanx #1485
 
 v0.15.0 (14 Jul 2018)
 ----------------------

src/PhpWord/TemplateProcessor.php

@@ -503,11 +503,19 @@ protected function getHeaderName($index)
     }
 
     /**
+     * Usually, the name of main part document will be 'document.xml'. However, some .docx files (possibly those from Office 365, experienced also on documents from Word Online created from blank templates) have file 'document22.xml' in their zip archive instead of 'document.xml'. This method searches content types file to correctly determine the file name.
+     *
      * @return string
      */
     protected function getMainPartName()
     {
-        return 'word/document.xml';
+        $contentTypes = $this->zipClass->getFromName('[Content_Types].xml');
+
+        $pattern = '~PartName="\/(word\/document.*?\.xml)" ContentType="application\/vnd\.openxmlformats-officedocument\.wordprocessingml\.document\.main\+xml"~';
+
+        preg_match($pattern, $contentTypes, $matches);
+
+        return array_key_exists(1, $matches) ? $matches[1] : 'word/document.xml';
     }
 
     /**

tests/PhpWord/TemplateProcessorTest.php

@@ -276,4 +276,13 @@ public function cloneBlockCanCloneABlockTwice()
             );
         }
     }
+
+    public function testMainPartNameDetection()
+    {
+        $templateProcessor = new TemplateProcessor(__DIR__ . '/_files/templates/document22-xml.docx');
+
+        $variables = array('test');
+
+        $this->assertEquals($variables, $templateProcessor->getVariables());
+    }
 }