NEW phpoffice/phpword support

silverstripe · Sep 19, 2023 · 8e4eb0b · 8e4eb0b
1 parent 5a6cd32
commit 8e4eb0b
Show file tree

Hide file tree

Showing 6 changed files with 141 additions and 17 deletions.
diff --git a/README.md b/README.md
@@ -19,12 +19,29 @@ Install with [composer](https://getcomposer.org/) by running `composer require s
 
 ## Configuration
 
-You will need to set the following three environment variables:
+**Note:** Using of of docvert is primarily designed for Common Web Platform (CWP) clients
+
+If you are using docver then you will need to set the following three environment variables:
 - `DOCVERT_USERNAME`
 - `DOCVERT_PASSWORD`
 - `DOCVERT_URL`
 
-**Note:** This module is primarily designed for Common Web Platform (CWP) clients. There will be additional setup required to use this module as intended, if you are not using the CWP government edition.
+If do not have the cwp/cwp-core module installed then enable document converter with the following configuration - note will be automatically applied if you also have the cwp/cwp-core module installed and the `DOCVERT_USERNAME` environment variable set.
+
+```yaml
+Page:
+  extensions:
+    - SilverStripe\DocumentConverter\PageExtension
+```
+
+By default this module will use docvert, though it's highly recommend you instead use the phpoffice/phpword module instead. Enable this with the following configuration:
+
+```yaml
+SilverStripe\DocumentConverter\ImportField:
+  importer_class: SilverStripe\DocumentConverter\PHPWordImporter
+```
+
+docvert support is now deprecated and will be removed in the next major version
 
 ## User Guide
 

diff --git a/composer.json b/composer.json
@@ -10,7 +10,8 @@
     "require": {
         "php": "^7.4 || ^8.0",
         "silverstripe/cms": "^4",
-        "silverstripe/asset-admin": "^1"
+        "silverstripe/asset-admin": "^1",
+        "phpoffice/phpword": "^1.1"
     },
     "require-dev": {
         "ext-curl": "*",

diff --git a/src/ImportField.php b/src/ImportField.php
@@ -13,7 +13,7 @@
 use SilverStripe\Assets\Folder;
 use SilverStripe\Assets\Image;
 use SilverStripe\Assets\Upload;
-use SilverStripe\Core\Config\Config;
+use SilverStripe\Control\Controller;
 use SilverStripe\Core\Convert;
 use SilverStripe\Core\Injector\Injector;
 use SilverStripe\Control\Director;
@@ -25,6 +25,7 @@
 use SilverStripe\Versioned\Versioned;
 use SilverStripe\View\Parsers\HTMLValue;
 use Tidy;
+use SilverStripe\AssetAdmin\Controller\AssetAdmin;
 
 /**
  * DocumentImporterField is built on top of UploadField to access a document
@@ -58,10 +59,12 @@ class ImportField extends UploadField
 
     private static $importer_class = ServiceConnector::class;
 
+    protected $attachEnabled = false;
+
     /**
      * Process the document immediately upon upload.
      */
-    public function upload(HTTPRequest $request)
+    public function Xupload(HTTPRequest $request)
     {
         if ($this->isDisabled() || $this->isReadonly()) {
             return $this->httpError(403);
@@ -94,11 +97,12 @@ public function upload(HTTPRequest $request)
 
         if (!$return['error']) {
             // Get options for this import.
-            $splitHeader = (int)$request->postVar('SplitHeader');
-            $keepSource = (bool)$request->postVar('KeepSource');
-            $chosenFolderID = (int)$request->postVar('ChosenFolderID');
-            $publishPages = (bool)$request->postVar('PublishPages');
-            $includeTOC = (bool)$request->postVar('IncludeTOC');
+            $splitHeader = (int) $request->postVar('SplitHeader');
+            $keepSource = (bool) $request->postVar('KeepSource');
+            $chosenFolderID = (int) $request->postVar('ChosenFolderID');
+            $publishPages = (bool) $request->postVar('PublishPages');
+            $includeTOC = (bool) $request->postVar('IncludeTOC');
+            // ^^^ These do nothing, they're aren't sent on upload
 
             // Process the document and write the page.
             $preservedDocument = null;
@@ -114,6 +118,31 @@ public function upload(HTTPRequest $request)
             }
         }
 
+        if (($return['error'] ?? 1) == 0) {
+            // asset-admin UploadField.js considers any error including 0 to be an error
+            // so simply unset the key if there is no error
+            unset($return['error']);
+        }
+
+        // generate the same result as UploadField
+        // note we don't need to do this if there is an actual error because the JSON that's
+        // returned is good enough to display an error message
+        if (!isset($return['error'])) {
+            if ($preservedDocument) {
+                $file = $preservedDocument;
+            } else {
+                // create a temporary File object to return to the client
+                $upload = Upload::create();
+                $file = File::create();
+                $upload->loadIntoFile($tmpfile, $file, $chosenFolderID);
+            }
+            $result = AssetAdmin::singleton()->getObjectFromData($file);
+            $return = array_merge($result, $return);
+            if (!$keepSource) {
+                // $file->delete();
+            }
+        }
+
         $response = HTTPResponse::create(json_encode([$return]));
         $response->addHeader('Content-Type', 'application/json');
         return $response;
@@ -277,7 +306,6 @@ protected function writeContent($subtitle, $subdoc, $subnode, $sort = null, $pub
      */
     public function importFromPOST($tmpFile, $splitHeader = false, $publishPages = false, $chosenFolderID = null)
     {
-
         $fileDescriptor = [
             'name' => $tmpFile['name'],
             'path' => $tmpFile['tmp_name'],
@@ -286,6 +314,7 @@ public function importFromPOST($tmpFile, $splitHeader = false, $publishPages = f
 
         $sourcePage = $this->form->getRecord();
         $importerClass = $this->config()->get('importer_class');
+        /** @var Importer $importer */
         $importer = Injector::inst()->create($importerClass, $fileDescriptor, $chosenFolderID);
         $content = $importer->import();
 
@@ -316,27 +345,39 @@ public function importFromPOST($tmpFile, $splitHeader = false, $publishPages = f
         // make sure any images are added as Image records with a relative link to assets
         $chosenFolder = ($this->chosenFolderID) ? DataObject::get_by_id(Folder::class, $this->chosenFolderID) : null;
         $folderName = ($chosenFolder) ? '/' . $chosenFolder->Name : '';
+        $dir = Controller::join_links(ASSETS_DIR, $folderName);
         $imgs = $xpath->query('//img');
         for ($i = 0; $i < $imgs->length; $i++) {
             $img = $imgs->item($i);
-            $originalPath = 'assets/' . $folderName . '/' . $img->getAttribute('src');
+            $originalPath = Controller::join_links($dir, $img->getAttribute('src'));
+            // ignore base64 encoded images which show up when importing using PHPOffice/PHPWord Word2007
+            // counter-intuitively it seems that we can simply ignore these and images
+            // are still imported correctly
+            if (preg_match("#data:image/.+?;base64,#", $originalPath)) {
+                continue;
+            }
+            // base64 inline image - note intentionally not starting regex ^ as it may start with ASSETS_DIR
+            // if (preg_match("#data:image/(.+?);base64,(.+)$#", $originalPath, $matches)) {
+            //     $ext = $matches[1];
+            //     $contents = $matches[2];
+            //     $filepath = tempnam($dir, 'image') . '.' . $ext;
+            //     file_put_contents($filepath, $contents);
+            //     $originalPath = $filepath;
+            // }
             $name = FileNameFilter::create()->filter(basename($originalPath ?? ''));
-
             $image = Image::get()->filter([
                 'Name' => $name,
-                'ParentID' => (int)$chosenFolderID
+                'ParentID' => (int) $chosenFolderID
             ])->first();
             if (!($image && $image->exists())) {
                 $image = Image::create();
-                $image->ParentID = (int)$chosenFolderID;
+                $image->ParentID = (int) $chosenFolderID;
                 $image->Name = $name;
                 $image->write();
             }
-
             // make sure it's put in place correctly so Image record knows where it is.
             // e.g. in the case of underscores being renamed to dashes.
             @rename(Director::getAbsFile($originalPath) ?? '', Director::getAbsFile($image->getFilename()) ?? '');
-
             $img->setAttribute('src', $image->getFilename());
         }
 

diff --git a/src/Importer.php b/src/Importer.php
@@ -0,0 +1,10 @@
+<?php
+
+namespace SilverStripe\DocumentConverter;
+
+interface Importer
+{
+    public function __construct($fileDescriptor, $chosenFolderID = null);
+
+    public function import();
+}
diff --git a/src/PHPWordImporter.php b/src/PHPWordImporter.php
@@ -0,0 +1,53 @@
+<?php
+
+namespace SilverStripe\DocumentConverter;
+
+use GraphQL\Exception\InvalidArgument;
+use SilverStripe\ORM\DataObject;
+
+class PHPWordImporter implements Importer
+{
+    private array $fileDescriptor;
+    private ?int $chosenFolderID;
+
+    /**
+     * @param array $fileDescriptor
+     * @param int|null $chosenFolderID
+     */
+    public function __construct($fileDescriptor, $chosenFolderID = null)
+    {
+        if (!is_array($fileDescriptor)) {
+            throw new InvalidArgument('fileDescriptor must be an array');
+        }
+        if (!is_int($chosenFolderID) && !is_null($chosenFolderID)) {
+            throw new InvalidArgument('chosenFolderID must be an int or null');
+        }
+        $this->fileDescriptor = $fileDescriptor;
+        $this->chosenFolderID = $chosenFolderID;
+    }
+
+    /**
+     * @return string
+     */
+    public function import()
+    {
+        // read word doc
+        $source = $this->fileDescriptor['path'];
+        $ext = pathinfo($this->fileDescriptor['name'], PATHINFO_EXTENSION);
+        $readerName = 'Word2007';
+        if ($ext === 'doc') {
+            // Word 1997
+            $readerName = 'MsDoc';
+        }
+        $phpWord = \PhpOffice\PhpWord\IOFactory::load($source, $readerName);
+        // write it out as HTML
+        $chosenFolder = ($this->chosenFolderID) ? DataObject::get_by_id(Folder::class, $this->chosenFolderID) : null;
+        $folderName = ($chosenFolder) ? '/' . $chosenFolder->Name : '';
+        $filepath = tempnam(ASSETS_PATH . $folderName, 'converted');
+        $objWriter = \PhpOffice\PhpWord\IOFactory::createWriter($phpWord, 'HTML');
+        $objWriter->save($filepath);
+        $content = file_get_contents($filepath);
+        unlink($filepath);
+        return $content;
+    }
+}
diff --git a/src/ServiceConnector.php b/src/ServiceConnector.php
@@ -11,6 +11,8 @@
 use ZipArchive;
 
 /**
+ * This class uses the legacy docvert service
+ *
  * Utility class hiding the specifics of the document conversion process.
  */
 class ServiceConnector