From b7ebd686df8a2b617632bc0080300af149aaf899 Mon Sep 17 00:00:00 2001
From: Radu Liviu Carjan <radu.carjan@gmail.com>
Date: Fri, 28 Oct 2022 13:18:40 +0300
Subject: [PATCH] Minor fix to OCR files

---
 app/SearchDisplace/Convertor/Convertor.php    | 12 ++++++++++
 .../Ingest/HandleReceivedDocument.php         | 14 +++++++++++-
 .../SearchAndDisplaceFromFiles.php            | 22 +++++++++++--------
 3 files changed, 38 insertions(+), 10 deletions(-)

diff --git a/app/SearchDisplace/Convertor/Convertor.php b/app/SearchDisplace/Convertor/Convertor.php
index 7284861..4be7bc9 100644
--- a/app/SearchDisplace/Convertor/Convertor.php
+++ b/app/SearchDisplace/Convertor/Convertor.php
@@ -4,6 +4,7 @@ namespace App\SearchDisplace\Convertor;
 
 use Symfony\Component\Process\Process;
 use Symfony\Component\Process\Exception\ProcessFailedException;
+use Illuminate\Support\Facades\Log;
 
 /**
  * Convert documents from formats supported by Libre Office
@@ -28,6 +29,17 @@ class Convertor {
             $folder = storage_path('app/tmp/');
         }
 
+        Log::info('Running `soffice` to convert "' . $original . '" to "' . $to . '". Output folder: "' . $folder . '"');
+        Log::info(
+            'COMMAND: ' .
+            'soffice ' .
+            '--convert-to ' .
+            $to . ' ' .
+            $original . ' ' .
+            '--outdir ' .
+            $folder
+        );
+
         $process = new Process(
             [
                 'soffice',
diff --git a/app/SearchDisplace/Ingest/HandleReceivedDocument.php b/app/SearchDisplace/Ingest/HandleReceivedDocument.php
index fe92cea..3ce8630 100644
--- a/app/SearchDisplace/Ingest/HandleReceivedDocument.php
+++ b/app/SearchDisplace/Ingest/HandleReceivedDocument.php
@@ -9,6 +9,8 @@ use GuzzleHttp\Client;
 use GuzzleHttp\Exception\ClientException;
 use Illuminate\Support\Facades\Log;
 use Illuminate\Support\Facades\Storage;
+use \Illuminate\Contracts\Filesystem\Filesystem;
+use App\SearchDisplace\Convertor\Convertor;
 
 class HandleReceivedDocument
 {
@@ -17,6 +19,7 @@ class HandleReceivedDocument
     protected $fileResultType;
     protected $documentFormat;
     protected $status;
+    protected Filesystem $storage;
 
     public function __construct($payload)
     {
@@ -28,6 +31,8 @@ class HandleReceivedDocument
         if (isset($payload['data']['document_format'])) {
             $this->documentFormat = $payload['data']['document_format'];
         }
+
+        $this->storage = Storage::disk('local');
     }
 
     /**
@@ -58,6 +63,7 @@ class HandleReceivedDocument
 
         // The .md extension signals the success status, the lack of signals the fail status.
         if ($this->status === 'success') {
+            $intermediateFileName = $fileName . '.odt';
             $fileName = $fileName . '.html';
         }
 
@@ -68,6 +74,11 @@ class HandleReceivedDocument
 
             $storage->put("$dir/$fileName", $this->content['document']);
 
+            // HTML document cannot be converted directly to xml. So we convert to ODT first, then to XML
+            Convertor::convert('odt', $this->storage->path("$dir/$fileName"));
+            Convertor::convert('xml', $this->storage->path("$dir/$intermediateFileName"));
+            $this->storage->delete("$dir/$intermediateFileName");
+
             foreach ($this->content['images'] as $image) {
                 $name = $image['name'];
                 $type = $image['type'];
@@ -81,7 +92,8 @@ class HandleReceivedDocument
             // Emit event so other sections of the app can work on it.
             IngestDocumentReceived::dispatch($this->id);
         } catch (\Exception $exception) {
-            \Illuminate\Support\Facades\Log::info('Exception: ' . $exception->getTraceAsString());
+            Log::error('Exception: ' . $exception->getMessage());
+            Log::error($exception->getTraceAsString());
         }
     }
 
diff --git a/app/SearchDisplace/SearchAndDisplaceFromFiles.php b/app/SearchDisplace/SearchAndDisplaceFromFiles.php
index ceb006d..34b8ee9 100644
--- a/app/SearchDisplace/SearchAndDisplaceFromFiles.php
+++ b/app/SearchDisplace/SearchAndDisplaceFromFiles.php
@@ -30,7 +30,7 @@ class SearchAndDisplaceFromFiles
         }
 
         try {
-            $documentContent = $this->storage->get("$this->directoryPath/document.md");
+            $documentContent = $this->storage->get("$this->directoryPath/document.html");
             $searchersContent = json_decode($this->storage->get($this->infoFilePath), true);
 
             $documentPath = $searchersContent['document_path'];
@@ -38,15 +38,19 @@ class SearchAndDisplaceFromFiles
 
             $this->storage->put($this->infoFilePath, json_encode($searchers[0]['content']));
 
-            $searchAndDisplace = new SearchAndDisplace($documentContent, [
-                'searchers' => [
-                    [
-                        'key' => $this->id,
-                        'type' => $searchers[0]['type'],
-                        'value' => $searchers[0]['value'],
-                    ]
+            $searchAndDisplace = new SearchAndDisplace(
+                $documentContent,
+                [
+                    'searchers' => [
+                        [
+                            'key' => $this->id,
+                            'type' => $searchers[0]['type'],
+                            'value' => $searchers[0]['value'],
+                        ]
+                    ],
                 ],
-            ]);
+                false
+            );
 
             $result = $searchAndDisplace->execute();