Add Tesseract as optional OCR engine

CONTRIBUTING.md

@@ -14,6 +14,7 @@ If you need to make asset changes:
 * `npm install`
 * Add the missing values from `.env` to a `.env.local` file
   * Use `https://vision.googleapis.com/` as the `APP_GOOGLE_CLOUD_VISION_ENDPOINT`, with your own [Cloud Vision API](https://cloud.google.com/vision) key as the `APP_GOOGLE_CLOUD_VISION_KEY`. Google gives you 1,000 free lookups per month.
+* Install [Tesseract](https://tesseract-ocr.github.io) and make sure it's in your `$PATH`
 * `symfony serve` to start the application
 * `npm run dev-server` if you need to make JS/CSS changes.
   * Stop the dev-server and run `npm run build` before committing.

composer.json

@@ -1,6 +1,6 @@
 {
     "name": "wikimedia/wikimedia-ocr",
-    "description": "A simple wrapper around the Google Cloud Vision API, enabling Wikisources to submit images for OCR and retrieve the resultant text.",
+    "description": "A simple wrapper around multiple OCR engines, enabling Wikisources to submit images for OCR and retrieve the resultant text.",
     "type": "project",
     "license": "GPL-3.0-or-later",
     "require": {
@@ -16,6 +16,7 @@
         "symfony/twig-bundle": "5.2.*",
         "symfony/webpack-encore-bundle": "^1.11",
         "symfony/yaml": "5.2.*",
+        "thiagoalessio/tesseract_ocr": "^2.11",
         "twig/extra-bundle": "^2.12|^3.0",
         "twig/twig": "^2.12|^3.0",
         "wikimedia/toolforge-bundle": "^1.3",

config/services.yaml

@@ -26,12 +26,10 @@ services:
         resource: '../src/Controller/'
         tags: ['controller.service_arguments']
 
-    App\Controller\OcrController:
+    App\Engine\GoogleCloudVisionEngine:
         arguments:
-            - '@request_stack'
-            - '@Krinkle\Intuition\Intuition'
-            - '%env(APP_GOOGLE_CLOUD_VISION_ENDPOINT)%'
-            - '%env(APP_GOOGLE_CLOUD_VISION_KEY)%'
+            $endpoint: '%env(APP_GOOGLE_CLOUD_VISION_ENDPOINT)%'
+            $key: '%env(APP_GOOGLE_CLOUD_VISION_KEY)%'
 
     # please note that last definitions always *replace* previous ones
     # add more service definitions when explicit configuration is needed

i18n/en.json

@@ -1,14 +1,19 @@
 {
-    "title": "Wikisource Google OCR",
+    "title": "Wikimedia OCR",
     "image-url": "Image URL:",
     "image-url-help": "This must start with 'https://upload.wikimedia.org/' and be a full URL to an actual image file.",
     "image-url-error": "Image URL must begin with '$1'",
     "image-alt-text": "The original image",
     "language-code": "Two-letter language code (optional):",
     "language-code-help": "The ISO639 code of the language of the text in the image.",
+    "engine": "OCR engine:",
+    "engine-help": "Choose between the open-source Wikimedia-hosted Tesseract, and Google's Cloud Vision API.",
+    "engine-name-google": "Google",
+    "engine-name-tesseract": "Tesseract",
     "submit": "Go",
     "copy-to-clipboard": "Copy to clipboard",
     "copied-to-clipboard": "Copied!",
     "more-info": "For more information, see:",
-    "limit-exceeded": "Limit exceeded: $1"
+    "limit-exceeded": "Limit exceeded: $1",
+    "image-retrieval-failed": "Image retrieval failed: $1"
 }

src/Controller/OcrController.php

@@ -3,7 +3,9 @@
 
 namespace App\Controller;
 
-use App\Engine\GoogleCloudVisionEngine;
+use App\Engine\EngineBase;
+use App\Engine\EngineFactory;
+use App\Engine\TesseractEngine;
 use App\Exception\OcrException;
 use Krinkle\Intuition\Intuition;
 use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;
@@ -17,7 +19,7 @@ class OcrController extends AbstractController
     /** @var Intuition */
     protected $intuition;
 
-    /** @var GoogleCloudVisionEngine */
+    /** @var EngineBase */
     protected $engine;
 
     /** @var mixed[] Output params for the view or API response. */
@@ -33,16 +35,18 @@ class OcrController extends AbstractController
      * OcrController constructor.
      * @param RequestStack $requestStack
      * @param Intuition $intuition
-     * @param string $endpoint
-     * @param string $key
+     * @param EngineFactory $engineFactory
      */
-    public function __construct(RequestStack $requestStack, Intuition $intuition, string $endpoint, string $key)
+    public function __construct(RequestStack $requestStack, Intuition $intuition, EngineFactory $engineFactory)
     {
-        $request = $requestStack->getCurrentRequest();
-
         // Dependencies.
         $this->intuition = $intuition;
-        $this->engine = new GoogleCloudVisionEngine($endpoint, $key);
+
+        $request = $requestStack->getCurrentRequest();
+
+        // Engine.
+        $this->engine = $engineFactory->get($request->get('engine', 'google'));
+        $this->params['engine'] = $this->engine instanceof TesseractEngine ? 'tesseract' : 'google';
 
         // Parameters.
         $this->imageUrl = (string)$request->query->get('image');