Azure Image OCR

.github/workflows/lint.yml

@@ -31,6 +31,12 @@ jobs:
     steps:
     - name: Checkout
       uses: actions/checkout@v2
+    - name: Set PHP version
+      uses: shivammathur/setup-php@v2
+      with:
+        php-version: '7.2'
+        coverage: none
+        tools: composer:v1
     - name: composer install
       run: composer install
     - name: PHPCS check

includes/Classifai/Helpers.php

@@ -436,3 +436,50 @@ function get_largest_acceptable_image_url( $full_image, $full_url, $sizes, $max
 
 	return null;
 }
+
+/**
+ * Retrieves the URL of the largest image that matches filesize and dimensions.
+ *
+ * This will check that the filesize of an image matches our requirements and
+ * if so, will then check the dimensions match our requirements as well. If
+ * neither match, will move on to the next largest image size.
+ *
+ * @param string $full_image The path to the full-sized image source file.
+ * @param string $full_url   The URL of the full-sized image.
+ * @param array  $metadata   Attachment metadata, including intermediate sizes.
+ * @param array  $width      Array of minimimum and maximum width values. Default 0, 4200.
+ * @param array  $height     Array of minimimum and maximum height values. Default 0, 4200.
+ * @param int    $max_size   The maximum acceptable filesize. Default 1MB.
+ * @return string|null The image URL, or null if no acceptable image found.
+ */
+function get_largest_size_and_dimensions_image_url( $full_image, $full_url, $metadata, $width = [ 0, 4200 ], $height = [ 0, 4200 ], $max_size = MB_IN_BYTES ) {
+	// Check if the full size image meets our filesize and dimension requirements
+	$file_size = @filesize( $full_image ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
+	if (
+		( $file_size && $max_size >= $file_size )
+		&& ( $metadata['width'] >= $width[0] && $metadata['width'] <= $width[1] )
+		&& ( $metadata['height'] >= $height[0] && $metadata['height'] <= $height[1] )
+	) {
+		return $full_url;
+	}
+
+	// If the full size doesn't match, run the same checks on our resized images
+	if ( isset( $metadata['sizes'] ) && is_array( $metadata['sizes'] ) ) {
+		usort( $metadata['sizes'], __NAMESPACE__ . '\sort_images_by_size_cb' );
+
+		foreach ( $metadata['sizes'] as $size ) {
+			$sized_file = str_replace( basename( $full_image ), $size['file'], $full_image );
+			$file_size  = @filesize( $sized_file ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
+
+			if (
+				( $file_size && $max_size >= $file_size )
+				&& ( $size['width'] >= $width[0] && $size['width'] <= $width[1] )
+				&& ( $size['height'] >= $height[0] && $size['height'] <= $height[1] )
+			) {
+				return str_replace( basename( $full_url ), $size['file'], $full_url );
+			}
+		}
+	}
+
+	return null;
+}

includes/Classifai/Providers/Azure/ComputerVision.php

@@ -6,6 +6,7 @@
 namespace Classifai\Providers\Azure;
 
 use Classifai\Providers\Provider;
+use DOMDocument;
 use WP_Error;
 
 use function Classifai\computer_vision_max_filesize;
@@ -66,6 +67,106 @@ public function register() {
 		add_filter( 'wp_generate_attachment_metadata', [ $this, 'smart_crop_image' ], 8, 2 );
 		add_filter( 'wp_generate_attachment_metadata', [ $this, 'generate_image_alt_tags' ], 10, 2 );
 		add_filter( 'posts_clauses', [ $this, 'filter_attachment_query_keywords' ], 10, 1 );
+
+		$settings   = $this->get_settings();
+		$enable_ocr = isset( $settings['enable_ocr'] ) && '1' === $settings['enable_ocr'];
+
+		if ( $enable_ocr ) {
+			add_action( 'enqueue_block_editor_assets', [ $this, 'enqueue_editor_assets' ] );
+			add_filter( 'the_content', [ $this, 'add_ocr_aria_describedby' ] );
+			add_filter( 'rest_api_init', [ $this, 'add_ocr_data_to_api_response' ] );
+		}
+	}
+
+	/**
+	 * Include classifai_computer_vision_ocr in API response.
+	 */
+	public function add_ocr_data_to_api_response() {
+		register_rest_field(
+			'attachment',
+			'classifai_has_ocr',
+			[
+				'get_callback'    => function( $params ) {
+					return ! empty( get_post_meta( $params['id'], 'classifai_computer_vision_ocr', true ) );
+				},
+				'schema' => [
+					'type'    => 'boolean',
+					'context' => [ 'view' ],
+				],
+			]
+		);
+	}
+
+	/**
+	 * Enqueue the editor scripts.
+	 */
+	public function enqueue_editor_assets() {
+		wp_enqueue_script(
+			'editor-ocr',
+			CLASSIFAI_PLUGIN_URL . 'dist/js/editor-ocr.min.js',
+			[
+				'wp-block-editor',
+				'wp-blocks',
+			],
+			CLASSIFAI_PLUGIN_VERSION,
+			true
+		);
+	}
+
+	/**
+	 * Filter the post content to inject aria-describedby attribute.
+	 *
+	 * @param string $content Post content.
+	 *
+	 * @return string
+	 */
+	public function add_ocr_aria_describedby( $content ) {
+		$modified = false;
+
+		if ( ! is_singular() || empty( $content ) ) {
+			return $content;
+		}
+
+		$dom = new DOMDocument();
+
+		// Suppress warnings generated by loadHTML.
+		$errors = libxml_use_internal_errors( true );
+		$dom->loadHTML(
+			sprintf(
+				'<!DOCTYPE html><html><head><meta charset="%s"></head><body>%s</body></html>',
+				esc_attr( get_bloginfo( 'charset' ) ),
+				$content
+			)
+		);
+		libxml_use_internal_errors( $errors );
+
+		foreach ( $dom->getElementsByTagName( 'img' ) as $image ) {
+			foreach ( $image->attributes as $attribute ) {
+				if ( 'aria-describedby' === $attribute->name ) {
+					break;
+				}
+
+				if ( 'class' !== $attribute->name ) {
+					continue;
+				}
+
+				$image_id            = preg_match( '~wp-image-\K\d+~', $image->getAttribute( 'class' ), $out ) ? $out[0] : 0;
+				$ocr_scanned_text_id = "classifai-ocr-$image_id";
+				$ocr_scanned_text    = $dom->getElementById( $ocr_scanned_text_id );
+
+				if ( ! empty( $ocr_scanned_text ) ) {
+					$image->setAttribute( 'aria-describedby', $ocr_scanned_text_id );
+					$modified = true;
+				}
+			}
+		}
+
+		if ( $modified ) {
+			$body = $dom->getElementsByTagName( 'body' )->item( 0 );
+			return trim( $dom->saveHTML( $body ) );
+		}
+
+		return $content;
 	}
 
 	/**
@@ -91,6 +192,7 @@ public function attachment_data_meta_box( $post ) {
 		$settings   = get_option( 'classifai_computer_vision' );
 		$captions   = get_post_meta( $post->ID, '_wp_attachment_image_alt', true ) ? __( 'Rescan Alt Text', 'classifai' ) : __( 'Scan Alt Text', 'classifai' );
 		$tags       = ! empty( wp_get_object_terms( $post->ID, 'classifai-image-tags' ) ) ? __( 'Rescan Tags', 'classifai' ) : __( 'Generate Tags', 'classifai' );
+		$ocr        = get_post_meta( $post->ID, 'classifai_computer_vision_ocr', true ) ? __( 'Rescan Text', 'classifai' ) : __( 'Scan Text', 'classifai' );
 		$smart_crop = get_transient( 'classifai_azure_computer_vision_smart_cropping_latest_response' ) ? __( 'Regenerate Smart Thumbnail', 'classifai' ) : __( 'Generate Smart Thumbnail', 'classifai' );
 		?>
 		<div class="misc-publishing-actions">
@@ -106,14 +208,20 @@ public function attachment_data_meta_box( $post ) {
 					<?php echo esc_html( $tags ); ?>
 				</label>
 			</div>
-			<?php if ( $settings && isset( $settings['enable_smart_cropping'] ) && '1' === $settings['enable_smart_cropping'] ) : ?>
-				<div class="misc-pub-section">
-					<label for="rescan-smart-crop">
-						<input type="checkbox" value="yes" id="rescan-smart-crop" name="rescan-smart-crop"/>
-						<?php echo esc_html( $smart_crop ); ?>
-					</label>
-				</div>
-			<?php endif; ?>
+			<div class="misc-pub-section">
+				<label for="rescan-ocr">
+					<input type="checkbox" value="yes" id="rescan-ocr" name="rescan-ocr"/>
+					<?php echo esc_html( $ocr ); ?>
+				</label>
+			</div>
+		<?php if ( $settings && isset( $settings['enable_smart_cropping'] ) && '1' === $settings['enable_smart_cropping'] ) : ?>
+			<div class="misc-pub-section">
+				<label for="rescan-smart-crop">
+					<input type="checkbox" value="yes" id="rescan-smart-crop" name="rescan-smart-crop"/>
+					<?php echo esc_html( $smart_crop ); ?>
+				</label>
+			</div>
+		<?php endif; ?>
 		</div>
 		<?php
 	}
@@ -128,7 +236,8 @@ public function maybe_rescan_image( $attachment_id ) {
 		$image_url  = get_largest_acceptable_image_url(
 			get_attached_file( $attachment_id ),
 			wp_get_attachment_url( $attachment_id ),
-			$metadata['sizes']
+			$metadata['sizes'],
+			computer_vision_max_filesize()
 		);
 
 		if ( filter_input( INPUT_POST, 'rescan-captions' ) ) {
@@ -158,6 +267,11 @@ public function maybe_rescan_image( $attachment_id ) {
 				}
 			}
 		}
+
+		// Are we updating the OCR text?
+		if ( filter_input( INPUT_POST, 'rescan-ocr' ) ) {
+			$this->ocr_processing( wp_get_attachment_metadata( $attachment_id ), $attachment_id, true );
+		}
 	}
 
 	/**
@@ -220,7 +334,8 @@ public function smart_crop_image( $metadata, $attachment_id ) {
 	 */
 	public function generate_image_alt_tags( $metadata, $attachment_id ) {
 
-		$settings = $this->get_settings();
+		$image_scan = false;
+		$settings   = $this->get_settings();
 		if (
 			'no' !== $settings['enable_image_tagging'] ||
 			'no' !== $settings['enable_image_captions']
@@ -259,6 +374,57 @@ public function generate_image_alt_tags( $metadata, $attachment_id ) {
 			}
 		}
 
+		// OCR processing
+		$this->ocr_processing( $metadata, $attachment_id, false, is_wp_error( $image_scan ) ? false : $image_scan );
+
+		return $metadata;
+	}
+
+	/**
+	 * Runs text recognition on the attachment.
+	 *
+	 * @since 1.6.0
+	 *
+	 * @filter wp_generate_attachment_metadata
+	 *
+	 * @param array       $metadata Attachment metadata.
+	 * @param int         $attachment_id Attachment ID.
+	 * @param boolean     $force Whether to force processing or not. Default false.
+	 * @param bool|object $scan Previously run image scan. Default false.
+	 * @return array Filtered attachment metadata.
+	 */
+	public function ocr_processing( array $metadata = [], int $attachment_id = 0, bool $force = false, $scan = false ) {
+		$settings = $this->get_settings();
+
+		if ( ! is_array( $metadata ) || ! is_array( $settings ) ) {
+			return $metadata;
+		}
+
+		$should_ocr_scan = isset( $settings['enable_ocr'] ) && '1' === $settings['enable_ocr'];
+
+		/**
+		 * Filters whether to run OCR scanning on the current image.
+		 *
+		 * @since 1.6.0
+		 * @hook classifai_should_ocr_scan_image
+		 *
+		 * @param bool  $should_ocr_scan Whether to run OCR scanning. The default value is set in ComputerVision settings.
+		 * @param array $metadata        Image metadata.
+		 * @param int   $attachment_id   The attachment ID.
+		 *
+		 * @return bool Whether to run OCR scanning.
+		 */
+		if ( ! $force && ! apply_filters( 'classifai_should_ocr_scan_image', $should_ocr_scan, $metadata, $attachment_id ) ) {
+			return $metadata;
+		}
+
+		$ocr      = new OCR( $settings, $scan, $force );
+		$response = $ocr->generate_ocr_data( $metadata, $attachment_id );
+
+		if ( $force ) {
+			return $response;
+		}
+
 		return $metadata;
 	}
 
@@ -547,6 +713,23 @@ public function setup_fields_sections() {
 				),
 			]
 		);
+
+		add_settings_field(
+			'enable-ocr',
+			esc_html__( 'Enable OCR', 'classifai' ),
+			[ $this, 'render_input' ],
+			$this->get_option_name(),
+			$this->get_option_name(),
+			[
+				'label_for'     => 'enable_ocr',
+				'input_type'    => 'checkbox',
+				'default_value' => false,
+				'description'   => __(
+					'Detect text in an image and store that as post content',
+					'classifai'
+				),
+			]
+		);
 	}
 
 	/**
@@ -580,6 +763,7 @@ public function sanitize_settings( $settings ) {
 			'enable_image_captions',
 			'enable_image_tagging',
 			'enable_smart_cropping',
+			'enable_ocr',
 		];
 
 		foreach ( $checkbox_settings as $checkbox_setting ) {
@@ -680,6 +864,7 @@ public function get_provider_debug_information( $settings = null ) {
 			__( 'Caption threshold', 'classifai' )                => $settings['caption_threshold'] ?? null,
 			__( 'Latest response - Image Scan', 'classifai' )     => $this->get_formatted_latest_response( get_transient( 'classifai_azure_computer_vision_image_scan_latest_response' ) ),
 			__( 'Latest response - Smart Cropping', 'classifai' ) => $this->get_formatted_latest_response( get_transient( 'classifai_azure_computer_vision_smart_cropping_latest_response' ) ),
+			__( 'Latest response - OCR', 'classifai' )            => $this->get_formatted_latest_response( get_transient( 'classifai_azure_computer_vision_ocr_latest_response' ) ),
 		];
 	}
 
@@ -741,12 +926,23 @@ public function filter_attachment_query_keywords( $clauses ) {
 	 * @return array|string|WP_Error
 	 */
 	public function rest_endpoint_callback( $post_id, $route_to_call ) {
-		$metadata           = wp_get_attachment_metadata( $post_id );
-		$image_url          = get_largest_acceptable_image_url(
+		$metadata = wp_get_attachment_metadata( $post_id );
+
+		if ( 'ocr' === $route_to_call ) {
+			return $this->ocr_processing( $metadata, $post_id, true );
+		}
+
+		$image_url = get_largest_acceptable_image_url(
 			get_attached_file( $post_id ),
 			wp_get_attachment_url( $post_id ),
-			$metadata['sizes']
+			$metadata['sizes'],
+			computer_vision_max_filesize()
 		);
+
+		if ( empty( $image_url ) ) {
+			return new WP_Error( 'error', esc_html__( 'Valid image size not found. Make sure the image is less than 4MB.' ) );
+		}
+
 		$image_scan_results = $this->scan_image( $image_url, [ $route_to_call ] );
 
 		if ( is_wp_error( $image_scan_results ) ) {