Ignore query string when testing image URLs

Allow image URLs to have any query string parameters, and split
the URL check messages into two so that it's clearer which is
being applied when a URL fails.

Bug: T422595

Author: samwilson

i18n/en.json

@@ -5,7 +5,8 @@
 	"form-heading": "Transcribe an image",
 	"image-url": "Image URL",
 	"image-url-help": "Insert an image URL hosted on a Wikimedia server such as: $1",
-	"image-url-error": "Image URL must begin with {{PLURAL:$1|the following domain name|one of the following domain names}} and end with a valid file extension: $2",
+	"image-url-error": "Image URL must be for {{PLURAL:$1|the following domain name|one of the following domain names}}: $2",
+	"image-url-error-extension": "Image URL have one of the following file extensions: $1",
 	"image-alt-text": "The original image",
 	"language-code": "Languages (optional)",
 	"engine": "OCR engine",

i18n/qqq.json

@@ -12,7 +12,8 @@
 	"form-heading": "Heading for the main form.",
 	"image-url": "Form label for the 'image' field.",
 	"image-url-help": "Help text for the 'image' field.",
-	"image-url-error": "Error message shown when the 'image' field is not valid.",
+	"image-url-error": "Error message shown when the 'image' field is not for a valid host.\n\nParameter:\n* $1 — Integer number of host names.\n* $2 — List of host names.",
+	"image-url-error-extension": "Error message shown when the 'image' field is not for a valid file extension.\n\nParameter:\n* $1 — List of permitted file extensions.",
 	"image-alt-text": "Alt text for the displayed image, next to the OCR text box.",
 	"language-code": "Form label for the 'languages' field.",
 	"engine": "Form label for the 'engine' field.\nOCR is a common abbreviation in English for \"Optical Characters Recognition\".",

src/Engine/EngineBase.php

@@ -137,14 +137,26 @@ public function getImageHosts(): array {
 	 * @throws OcrException
 	 */
 	public function checkImageUrl( string $imageUrl ): void {
-		$hostRegex = implode( '|', array_map( 'preg_quote', $this->getImageHosts() ) );
-		$formatRegex = implode( '|', self::ALLOWED_FORMATS );
-		$regex = "/^https?:\/\/($hostRegex)\/.+($formatRegex)$/";
-		$matches = preg_match( $regex, strtolower( $imageUrl ) );
-		if ( $matches !== 1 ) {
+		$parts = parse_url( $imageUrl );
+
+		// Check host.
+		if ( !isset( $parts['host'] ) || !in_array( $parts['host'], $this->getImageHosts(), true ) ) {
 			$params = [ count( $this->getImageHosts() ), $this->intuition->listToText( $this->getImageHosts() ) ];
 			throw new OcrException( 'image-url-error', $params );
 		}
+
+		// Check file extension.
+		$validExt = false;
+		if ( isset( $parts['path'] ) ) {
+			$formatRegex = implode( '|', self::ALLOWED_FORMATS );
+			$regex = "/.+($formatRegex)$/";
+			$matches = preg_match( $regex, strtolower( $parts['path'] ) );
+			$validExt = $matches === 1;
+		}
+		if ( !$validExt ) {
+			$params = [ $this->intuition->listToText( self::ALLOWED_FORMATS ) ];
+			throw new OcrException( 'image-url-error-extension', $params );
+		}
 	}
 
 	/**

tests/Engine/EngineBaseTest.php

@@ -56,6 +56,7 @@ public function provideCheckImageUrl(): array {
 			// Pass:
 			[ 'https://upload.wikimedia.org/wikipedia/commons/a/a9/Example.jpg', false ],
 			[ 'https://upload.wikimedia.org/wikipedia/commons/file.jpg', false ],
+			[ 'https://upload.wikimedia.org/wikipedia/commons/file.jpg?foo=bar', false ],
 			// Fail:
 			[ 'https://foo.example.com/wikipedia/commons/a/a9/Example.jpg', true ],
 			[ 'https://en.wikisource.org/file.mov', true ],