[BlockMarkupURLProcessor] Support URLs in CSS

components/DataLiberation/BlockMarkup/class-blockmarkupurlprocessor.php

@@ -4,10 +4,8 @@
 
 use Rowbot\URL\URL;
 use WordPress\DataLiberation\URL\URLInTextProcessor;
+use WordPress\DataLiberation\URL\CSSURLProcessor;
 use WordPress\DataLiberation\URL\WPURL;
-use WordPress\DataLiberation\URL\ConvertedUrl;
-
-use function WordPress\DataLiberation\URL\urldecode_n;
 
 /**
  * Reports all the URLs in the imported post and enables rewriting them.
@@ -23,6 +21,8 @@ class BlockMarkupUrlProcessor extends BlockMarkupProcessor {
 	private $base_url_object;
 	private $url_in_text_processor;
 	private $url_in_text_node_updated;
+	private $css_url_processor;
+	private $css_url_processor_updated;
 
 	/**
 	 * The list of names of URL-related HTML attributes that may be available on
@@ -52,6 +52,14 @@ public function get_updated_html(): string {
 			$this->url_in_text_node_updated = false;
 		}
 
+		if ( $this->css_url_processor_updated ) {
+			if ( null !== $this->css_url_processor ) {
+				$updated_css = $this->css_url_processor->get_updated_css();
+				$this->set_attribute( 'style', $updated_css );
+			}
+			$this->css_url_processor_updated = false;
+		}
+
 		return parent::get_updated_html();
 	}
 
@@ -70,8 +78,11 @@ public function next_token(): bool {
 		$this->parsed_url                 = null;
 		$this->inspecting_html_attributes = null;
 		$this->url_in_text_processor      = null;
-		// Do not reset url_in_text_node_updated – it's reset in get_updated_html() which
-		// is called in parent::next_token().
+		$this->css_url_processor          = null;
+		/*
+		 * Do not reset url_in_text_node_updated or css_url_processor_updated – they're reset
+		 * in get_updated_html() which is called in parent::next_token().
+		 */
 
 		return parent::next_token();
 	}
@@ -111,7 +122,7 @@ private function next_url_in_text_node() {
 			 * way to recognize a substring "WordPress.org" as a URL. We might
 			 * get some false positives this way, e.g. in this string:
 			 *
-			 * > And that's how you build a theme.Now let's take a look at..."
+			 * > And that's how you build a theme. Now let's take a look at..."
 			 *
 			 * `theme.Now` would be recognized as a URL. It's up to the API consumer
 			 * to filter out such false positives e.g. by checking the domain against
@@ -130,20 +141,75 @@ private function next_url_in_text_node() {
 		return false;
 	}
 
+	/**
+	 * Advances to the next CSS URL in the `style` attribute of the current tag token.
+	 *
+	 * @return bool Whether a CSS URL was found.
+	 */
+	private function next_url_in_css() {
+		if ( '#tag' !== $this->get_token_type() ) {
+			return false;
+		}
+
+		if ( null === $this->css_url_processor ) {
+			$css_value = $this->get_attribute( 'style' );
+			if ( ! is_string( $css_value ) ) {
+				return false;
+			}
+
+			$this->css_url_processor = new CSSURLProcessor( $css_value );
+		}
+
+		while ( $this->css_url_processor->next_url() ) {
+			/**
+			 * Skip data URIs. They may be really large and they don't
+			 * have a hostname to migrate.
+			 */
+			if ( $this->css_url_processor->is_data_uri() ) {
+				continue;
+			}
+			$this->raw_url    = $this->css_url_processor->get_raw_url();
+			$this->parsed_url = WPURL::parse( $this->raw_url, $this->base_url_string );
+			if ( false === $this->parsed_url ) {
+				continue;
+			}
+
+			return true;
+		}
+
+		return false;
+	}
+
 	private function next_url_attribute() {
 		$tag = $this->get_tag();
 
-		if ( ! array_key_exists( $tag, self::HTML_ATTRIBUTES_TO_ACCEPT_RELATIVE_URLS_FROM ) ) {
-			return false;
+		// Check if we have a style attribute with CSS URLs to process.
+		if ( null !== $this->css_url_processor ) {
+			if ( $this->next_url_in_css() ) {
+				return true;
+			}
+			// Done with CSS URLs in this attribute, apply any pending updates and move on.
+			$this->get_updated_html();
+			$this->css_url_processor = null;
 		}
 
 		if ( null === $this->inspecting_html_attributes ) {
-			/**
-			 * Initialize the list on the first call to next_url_attribute()
-			 * for the current token. The last element is the attribute we'll
-			 * inspect in the while() loop below.
-			 */
-			$this->inspecting_html_attributes = self::HTML_ATTRIBUTES_TO_ACCEPT_RELATIVE_URLS_FROM[ $tag ];
+			if ( array_key_exists( $tag, self::HTML_ATTRIBUTES_TO_ACCEPT_RELATIVE_URLS_FROM ) ) {
+				/**
+				 * Initialize the list on the first call to next_url_attribute()
+				 * for the current token. The last element is the attribute we'll
+				 * inspect in the while() loop below.
+				 */
+				$this->inspecting_html_attributes = self::HTML_ATTRIBUTES_TO_ACCEPT_RELATIVE_URLS_FROM[ $tag ];
+				// Add style attribute to the list if it exists.
+				if ( null !== $this->get_attribute( 'style' ) ) {
+					$this->inspecting_html_attributes[] = 'style';
+				}
+			} elseif ( null !== $this->get_attribute( 'style' ) ) {
+				$this->inspecting_html_attributes = array( 'style' );
+			} else {
+				return false;
+			}
 		} else {
 			/**
 			 * Forget the attribute we've inspected on the previous call to
@@ -160,6 +226,18 @@ private function next_url_attribute() {
 				continue;
 			}
 
+			// Rewrite any CSS `url()` declarations in the `style` attribute.
+			if ( 'style' === $attr ) {
+				$this->css_url_processor = new CSSURLProcessor( $url_maybe );
+				if ( $this->next_url_in_css() ) {
+					return true;
+				}
+				// No CSS URLs found, move to next attribute.
+				$this->css_url_processor = null;
+				array_pop( $this->inspecting_html_attributes );
+				continue;
+			}
+
 			/*
 			 * Use base URL to resolve known URI attributes as we are certain we're
 			 * dealing with URI values.
@@ -277,6 +355,12 @@ public function set_url( $raw_url, $parsed_url ) {
 		$this->parsed_url = $parsed_url;
 		switch ( parent::get_token_type() ) {
 			case '#tag':
+				// Check if we're processing a CSS URL.
+				if ( null !== $this->css_url_processor ) {
+					$this->css_url_processor_updated = true;
+					return $this->css_url_processor->set_raw_url( $raw_url );
+				}
+
 				$attr = $this->get_inspected_attribute_name();
 				if ( false === $attr ) {
 					return false;

components/DataLiberation/URL/class-cssprocessor.php → components/DataLiberation/CSS/class-cssprocessor.php

@@ -1,6 +1,6 @@
 <?php
 
-namespace WordPress\DataLiberation\URL;
+namespace WordPress\DataLiberation\CSS;
 
 use function WordPress\Encoding\codepoint_to_utf8_bytes;
 use function WordPress\Encoding\compat\_wp_scan_utf8;
@@ -742,6 +742,32 @@ public function get_token_value() {
 		return $this->token_value;
 	}
 
+	/**
+	 * Determines whether the current token is a data URI.
+	 *
+	 * Only meaningful for URL and STRING tokens. Returns false for all other token types.
+	 *
+	 * @return bool Whether the current token value starts with "data:" (case-insensitive).
+	 */
+	public function is_data_uri(): bool {
+		if ( null === $this->token_value_starts_at || null === $this->token_value_length ) {
+			return false;
+		}
+
+		if ( $this->token_value_length < 5 ) {
+			return false;
+		}
+
+		$offset = $this->token_value_starts_at;
+		return (
+			( 'd' === $this->css[ $offset ] || 'D' === $this->css[ $offset ] ) &&
+			( 'a' === $this->css[ $offset + 1 ] || 'A' === $this->css[ $offset + 1 ] ) &&
+			( 't' === $this->css[ $offset + 2 ] || 'T' === $this->css[ $offset + 2 ] ) &&
+			( 'a' === $this->css[ $offset + 3 ] || 'A' === $this->css[ $offset + 3 ] ) &&
+			':' === $this->css[ $offset + 4 ]
+		);
+	}
+
 	/**
 	 * Gets the token start at.
 	 *
@@ -812,27 +838,26 @@ public function get_token_value_length(): ?int {
 	 * @return bool Whether the value was successfully updated.
 	 */
 	public function set_token_value( string $new_value ): bool {
-		// Only URL tokens are currently supported.
-		if ( self::TOKEN_URL !== $this->token_type ) {
-			return false;
-		}
-
-		// Ensure we have valid token value boundaries.
-		if ( null === $this->token_value_starts_at || null === $this->token_value_length ) {
-			return false;
+		// Only URL and string tokens are currently supported.
+		switch ( $this->token_type ) {
+			case self::TOKEN_URL:
+				$this->lexical_updates[] = array(
+					'start'  => $this->token_value_starts_at,
+					'length' => $this->token_value_length,
+					'text'   => $this->escape_url_value( $new_value ),
+				);
+				return true;
+			case self::TOKEN_STRING:
+				$this->lexical_updates[] = array(
+					'start'  => $this->token_starts_at,
+					'length' => $this->token_length,
+					'text'   => $this->escape_url_value( $new_value ),
+				);
+				return true;
+			default:
+				_doing_it_wrong( __METHOD__, 'set_token_value() only supports URL and string tokens. Got token type: ' . $this->token_type, '1.0.0' );
+				return false;
 		}
-
-		// Escape the URL value for unquoted URL syntax.
-		$escaped_value = $this->escape_url_value( $new_value );
-
-		// Queue the lexical update.
-		$this->lexical_updates[] = array(
-			'start'  => $this->token_value_starts_at,
-			'length' => $this->token_value_length,
-			'text'   => $escaped_value,
-		);
-
-		return true;
 	}
 
 	/**