Skip conversion from odt to docx.

README.md

@@ -15,7 +15,6 @@
 ### Ubuntu Packages
 ```bash
 # LibreOffice
-apt-get install python-software-properties
 apt-add-repository ppa:libreoffice/ppa
 apt-get update
 apt-get install libreoffice

app/Ingest/AbstractConvertor.php

@@ -7,12 +7,14 @@ abstract class AbstractConvertor
     protected $storage;
     protected $path;
     protected $directoryPath;
+    protected $type = null;
 
-    public function __construct($storage, $path)
+    public function __construct($storage, $path, $type = null)
     {
         $this->storage = $storage;
         $this->path = $path;
         $this->directoryPath = pathinfo($path, PATHINFO_DIRNAME);
+        $this->type = strtoupper($type);
     }
 
     abstract public function execute();

app/Ingest/Convertor.php

@@ -3,8 +3,6 @@
 namespace App\Ingest;
 
 use Illuminate\Support\Facades\Storage;
-use Symfony\Component\Process\Exception\ProcessFailedException;
-use Symfony\Component\Process\Process;
 
 class Convertor
 {
@@ -31,8 +29,8 @@ class Convertor
             $convertor = new TextConvertor($this->storage, $this->path);
         } else if ($this->type === 'pdf') {
             $convertor = new PDFConvertor($this->storage, $this->path);
-        } else if ($this->type === 'docx') {
-            $convertor = new DocxConvertor($this->storage, $this->path);
+        } else if ($this->type === 'docx' || $this->type === 'odt') {
+            $convertor = new DocxAndOdtConvertor($this->storage, $this->path, $this->type);
         } else {
             $convertor = new OtherConvertor($this->storage, $this->path);
         }
@@ -41,44 +39,4 @@ class Convertor
 
         //$this->convertToHtml();
     }
-
-    private function convertToHtml()
-    {
-        $office = new Office();
-
-        $success = $office->run(
-            'html:HTML:EmbedImages',
-            $this->storage->path($this->path),
-            $this->storage->path('contracts')
-        );
-
-        if (! $success) {
-            throw new \Exception('Something went wrong while tried converting to HTML for file: ' . $this->path);
-        }
-
-        $this->storage->delete($this->path);
-
-        $this->path = str_replace(".$this->type", '.html', $this->path);
-    }
-
-    private function convertToXML()
-    {
-        //Convert the file to xml using pdftohtml to xml and run a python scrypt to fix the paragraphs
-        $process = new Process([
-            'pdftohtml',
-            '-xml',
-            '-i',
-            $this->storage->path($this->path)
-        ]);
-
-        $process->run();
-
-        if (!$process->isSuccessful()) {
-            throw new ProcessFailedException($process);
-        }
-
-        $this->storage->delete($this->path);
-
-        $this->path = str_replace(".$this->type", '.xml', $this->path);
-    }
 }

app/Ingest/DocxConvertor.php → app/Ingest/DocxAndOdtConvertor.php

@@ -5,7 +5,7 @@ namespace App\Ingest;
 use Symfony\Component\Process\Exception\ProcessFailedException;
 use Symfony\Component\Process\Process;
 
-class DocxConvertor extends AbstractConvertor
+class DocxAndOdtConvertor extends AbstractConvertor
 {
     /**
      *
@@ -107,7 +107,7 @@ class DocxConvertor extends AbstractConvertor
         );
 
         if (! $success) {
-            throw new \Exception('Failed when converting from DOCX to ' . strtoupper($format) . ' for file: ' . $this->path);
+            throw new \Exception('Failed when converting from ' . $this->type . ' to ' . strtoupper($format) . ' for file: ' . $this->path);
         }
 
         $this->deleteOriginalDocument();

app/Ingest/OtherConvertor.php

@@ -2,16 +2,13 @@
 
 namespace App\Ingest;
 
-use Symfony\Component\Process\Exception\ProcessFailedException;
-use Symfony\Component\Process\Process;
-
 class OtherConvertor extends AbstractConvertor
 {
     public function execute()
     {
         $this->convertToDocx();
 
-        $convertor = new DocxConvertor($this->storage, "$this->directoryPath/document.docx");
+        $convertor = new DocxAndOdtConvertor($this->storage, "$this->directoryPath/document.docx");
 
         $convertor->execute();
     }

app/Jobs/RecreateDocument.php

@@ -2,7 +2,7 @@
 
 namespace App\Jobs;
 
-use App\Ingest\DocxConvertor;
+use App\Ingest\DocxAndOdtConvertor;
 use App\Ingest\DocxWriter;
 use Illuminate\Bus\Queueable;
 use Illuminate\Contracts\Queue\ShouldQueue;
@@ -122,7 +122,7 @@ class RecreateDocument implements ShouldQueue
             return;
         }
 
-        $convertor = new DocxConvertor($this->storage, $this->path);
+        $convertor = new DocxAndOdtConvertor($this->storage, $this->path);
 
         if ($this->data['document_format'] === 'pdf') {
             $convertor->convertToPdfWithLibreOffice();

app/Jobs/SendToCore.php

@@ -70,7 +70,7 @@ class SendToCore implements ShouldQueue
             $content = $this->getContent();
         }
 
-        $sent = $this->sendTheData($content);
+        $this->sendTheData($content);
 
 //        if ($this->directoryPath && $sent) {
         if ($this->directoryPath) {

composer.json

@@ -9,16 +9,15 @@
     "license": "MIT",
     "require": {
         "php": "^7.2",
+        "ext-json": "*",
         "cebe/markdown": "^1.2",
         "fideloper/proxy": "^4.0",
         "laravel/framework": "^6.2",
         "laravel/tinker": "^2.0",
-        "phpoffice/phpword": "^0.17.0",
         "predis/predis": "^1.1",
         "spatie/laravel-webhook-server": "^1.13",
         "spatie/pdf-to-text": "^1.3",
-        "thiagoalessio/tesseract_ocr": "^2.11",
-        "ext-json": "*"
+        "thiagoalessio/tesseract_ocr": "^2.11"
     },
     "require-dev": {
         "facade/ignition": "^1.4",