From dc131b7751643079d2ea14e1652ef354b2f5a8dc Mon Sep 17 00:00:00 2001
From: Orzu Ionut <orzu.ionut@gmail.com>
Date: Wed, 8 Sep 2021 16:33:02 +0300
Subject: [PATCH] S&D on original document recreates document with the original
 format

---
 ...hAndDisplaceOriginalDocumentController.php |  8 ++-
 .../Ingest/HandleReceivedDocument.php         | 31 +++++++++---
 .../SearchAndDisplaceOriginalDocument.php     | 49 +++++++++++++++----
 3 files changed, 70 insertions(+), 18 deletions(-)

diff --git a/app/Http/Controllers/SearchAndDisplaceOriginalDocumentController.php b/app/Http/Controllers/SearchAndDisplaceOriginalDocumentController.php
index 975cbd1..bf8a1e7 100644
--- a/app/Http/Controllers/SearchAndDisplaceOriginalDocumentController.php
+++ b/app/Http/Controllers/SearchAndDisplaceOriginalDocumentController.php
@@ -14,7 +14,8 @@ class SearchAndDisplaceOriginalDocumentController extends Controller
                 'required',
                 'file',
                 'max:10000',
-                'mimes:doc,dot,docx,dotx,docm,dotm,odt,rtf,pdf,txt',
+//                'mimes:doc,dot,docx,dotx,docm,dotm,odt,rtf,pdf,txt',
+                'mimes:doc,docx,odt,rtf,pdf,txt',
             ],
 
             // Searchers is encoded.
@@ -36,7 +37,10 @@ class SearchAndDisplaceOriginalDocumentController extends Controller
         try {
             $handler = new SearchAndDisplaceOriginalDocument();
 
-            $id = $handler->start(request()->file('document'), json_decode(request()->get('searchers'), true));
+            $id = $handler->start(
+                request()->file('document'),
+                json_decode(request()->get('searchers'), true)
+            );
 
             return response()->json([
                 'status' => 'ok',
diff --git a/app/SearchDisplace/Ingest/HandleReceivedDocument.php b/app/SearchDisplace/Ingest/HandleReceivedDocument.php
index 36df47c..e475e61 100644
--- a/app/SearchDisplace/Ingest/HandleReceivedDocument.php
+++ b/app/SearchDisplace/Ingest/HandleReceivedDocument.php
@@ -13,6 +13,7 @@ class HandleReceivedDocument
     protected $id;
     protected $content;
     protected $fileResultType;
+    protected $documentFormat;
     protected $status;
 
     public function __construct($payload)
@@ -21,10 +22,14 @@ class HandleReceivedDocument
         $this->content = $payload['data']['content'];
         $this->fileResultType = $payload['data']['file_result_type'];
         $this->status = $payload['data']['status'];
+
+        if (isset($payload['data']['document_format'])) {
+            $this->documentFormat = $payload['data']['document_format'];
+        }
     }
 
     /**
-     * @throws \Exception
+     * @throws \Exception|\GuzzleHttp\Exception\GuzzleException
      */
     public function handle()
     {
@@ -78,27 +83,32 @@ class HandleReceivedDocument
         }
     }
 
+    /**
+     * @throws \Exception|\GuzzleHttp\Exception\GuzzleException
+     */
     protected function downloadDocumentFromIngest()
     {
         $storage = Storage::disk('local');
 
-
         if ($this->status === 'fail') {
             $storage->deleteDirectory("contracts/$this->id");
 
             return;
         }
 
-        $filePath = 'document.docx';
-        $fullPath = 'contracts/' . $this->id . '-' . $filePath;
+        if ( ! $this->documentFormat) {
+            throw new \Exception('No document format specified.');
+        }
 
-        $url = env('SD_INGEST_URL') . '/recreate-document/' . $this->id . '?file_path=' . $filePath;
+        $filePath = "document.$this->documentFormat";
+        $storeDocumentAtPath = 'contracts/' . $this->id . '-' . $filePath;
 
         $client = new Client();
+        $url = env('SD_INGEST_URL') . '/recreate-document/' . $this->id . '?file_path=' . $filePath;
 
         try {
             $client->request('GET', $url, [
-                'sink' => $storage->path($fullPath),
+                'sink' => $storage->path($storeDocumentAtPath),
             ]);
         } catch (ClientException $clientException) {
             $error = json_decode($clientException->getResponse()->getBody()->getContents(), true);
@@ -109,6 +119,9 @@ class HandleReceivedDocument
         }
     }
 
+    /**
+     * @throws \Exception|\GuzzleHttp\Exception\GuzzleException
+     */
     protected function handleDocumentJson()
     {
         $handler = new SearchAndDisplaceOriginalDocument();
@@ -119,6 +132,10 @@ class HandleReceivedDocument
             return;
         }
 
-        $handler->applySD($this->id, $this->content);
+        if ( ! $this->documentFormat) {
+            throw new \Exception('No document format specified.');
+        }
+
+        $handler->applySD($this->id, $this->content, $this->documentFormat);
     }
 }
diff --git a/app/SearchDisplace/SearchAndDisplaceOriginalDocument.php b/app/SearchDisplace/SearchAndDisplaceOriginalDocument.php
index b082d3b..8825fc3 100644
--- a/app/SearchDisplace/SearchAndDisplaceOriginalDocument.php
+++ b/app/SearchDisplace/SearchAndDisplaceOriginalDocument.php
@@ -25,9 +25,10 @@ class SearchAndDisplaceOriginalDocument
     /**
      * @param $id
      * @param $contents
+     * @param $documentFormat
      * @throws \GuzzleHttp\Exception\GuzzleException
      */
-    public function applySD($id, $contents)
+    public function applySD($id, $contents, $documentFormat)
     {
         $data = json_decode($contents['document'], true);
 
@@ -46,6 +47,8 @@ class SearchAndDisplaceOriginalDocument
             $x = $this->applyResultsOnIngestData($data['contents'], $result);
             $data['contents'] = $x;
 
+            $data['document_format'] = $documentFormat;
+
             $this->sendDataToIngestToRebuild($id, $data);
         } catch (\Exception $exception) {
             \Illuminate\Support\Facades\Log::info($exception->getMessage());
@@ -64,9 +67,16 @@ class SearchAndDisplaceOriginalDocument
     {
         $storage = Storage::disk('local');
         $basePath = "contracts/$id";
-        $filePath = $basePath . '-document.docx';
 
-        return ! $storage->exists($basePath) && ! $storage->exists($filePath);
+        if ($storage->exists($basePath)) {
+            return false;
+        }
+
+        if ($this->findDocumentById($id) !== null) {
+            return false;
+        }
+
+        return true;
     }
 
     public function isInProgress($id)
@@ -74,8 +84,7 @@ class SearchAndDisplaceOriginalDocument
         $storage = Storage::disk('local');
         $basePath = "contracts/$id";
 
-        // @TODO Set document extension.
-        return $storage->exists($basePath) && ! $storage->exists($basePath . '-document.docx');
+        return $storage->exists($basePath) && $this->findDocumentById($id) === null;
     }
 
     /**
@@ -87,12 +96,34 @@ class SearchAndDisplaceOriginalDocument
     {
         $storage = Storage::disk('local');
 
-        if ($this->hasFailed($id) || $this->isInProgress($id)) {
-            throw new \Exception('Document is not processed.');
+        // @TODO Improve this. Right now we are calling 'findDocumentById' multiple times.
+
+        if ($this->hasFailed($id)) {
+            throw new \Exception('Document has failed.');
+        }
+
+        if ($this->isInProgress($id)) {
+            throw new \Exception('Document is still processing.');
+        }
+
+        return $storage->path($this->findDocumentById($id));
+    }
+
+    protected function findDocumentById($id)
+    {
+        $storage = Storage::disk('local');
+
+        $documentPartialPath = "contracts/$id-document";
+
+        $contractFiles = $storage->files('contracts');
+
+        foreach ($contractFiles as $contractFile) {
+            if (substr($contractFile, 0, strlen($documentPartialPath)) === $documentPartialPath) {
+                return $contractFile;
+            }
         }
 
-        // @TODO Set document extension.
-        return $storage->path('contracts/' . $id . '-document.docx');
+        return null;
     }
 
     protected function applyResultsOnIngestData($ingestData, $sdResult)