implement rtf loader

composer.json

@@ -20,6 +20,7 @@
         "ext-zip": "*",
         "aws/aws-sdk-php": "^3.281",
         "illuminate/contracts": "^10.0",
+        "jstewmc/rtf": "^0.5.2",
         "league/flysystem-aws-s3-v3": "^3.16",
         "openai-php/laravel": "^0.7.0",
         "prinsfrank/standards": "^2.1",

src/TextLoader/Rtf.php

@@ -0,0 +1,18 @@
+<?php
+
+namespace HelgeSverre\ReceiptScanner\TextLoader;
+
+use HelgeSverre\ReceiptScanner\Contracts\TextLoader;
+use HelgeSverre\ReceiptScanner\TextContent;
+use Jstewmc\Rtf\Document;
+
+class Rtf implements TextLoader
+{
+    public function load(mixed $data): ?TextContent
+    {
+        $document = new Document($data);
+        $text = $document->getRoot()->toText();
+
+        return new TextContent($text);
+    }
+}

src/TextLoaderFactory.php

@@ -5,6 +5,7 @@
 use HelgeSverre\ReceiptScanner\Contracts\TextLoader;
 use HelgeSverre\ReceiptScanner\TextLoader\Html;
 use HelgeSverre\ReceiptScanner\TextLoader\Pdf;
+use HelgeSverre\ReceiptScanner\TextLoader\Rtf;
 use HelgeSverre\ReceiptScanner\TextLoader\Text;
 use HelgeSverre\ReceiptScanner\TextLoader\Textract;
 use HelgeSverre\ReceiptScanner\TextLoader\TextractUsingS3Upload;
@@ -28,6 +29,7 @@ public function create(string $type): TextLoader
             'html' => $this->container->make(Html::class),
             'pdf' => $this->container->make(Pdf::class),
             'text' => $this->container->make(Text::class),
+            'rtf' => $this->container->make(Rtf::class),
             'textract_s3' => $this->container->make(TextractUsingS3Upload::class),
             'textract' => $this->container->make(Textract::class),
             'web' => $this->container->make(Web::class),
@@ -52,6 +54,11 @@ public function text(mixed $data): ?TextContent
         return $this->create('text')->load($data);
     }
 
+    public function rtf(mixed $data): ?TextContent
+    {
+        return $this->create('rtf')->load($data);
+    }
+
     public function textractUsingS3Upload(mixed $data): ?TextContent
     {
         return $this->create('textract_s3')->load($data);

tests/TextLoaderTest.php

@@ -42,14 +42,23 @@
     );
 });
 
-it('Can load Word Documents', function () {
+it('Can load a Word documents i found on the internet', function () {
     $text = Text::word(file_get_contents(__DIR__.'/samples/word-document.doc'));
 
     expect($text)->toBeInstanceOf(TextContent::class)->and($text->toString())->toContain(
         'Mauris',
     );
 });
 
+it('Can load a Word document exported from google docs', function () {
+    $text = Text::word(file_get_contents(__DIR__.'/samples/contract.docx'));
+
+    expect($text)->toBeInstanceOf(TextContent::class)->and($text->toString())->toContain(
+        'Contract Agreement',
+        'Termination of the Agreement',
+    );
+});
+
 it('Can load text from website', function () {
     $text = Text::web('https://sparksuite.github.io/simple-html-invoice-template/');
 
@@ -67,3 +76,12 @@
         'NOK 1,246.25',
     );
 });
+
+it('Can load rtf files', function () {
+    $text = Text::rtf(file_get_contents(__DIR__.'/samples/contract.rtf'));
+
+    expect($text)->toBeInstanceOf(TextContent::class)->and($text->toString())->toContain(
+        'Contract Agreement',
+        'Termination of the Agreement',
+    );
+});