Markup migration

transfer-protocol/bin/rewrite-urls.php

@@ -0,0 +1,97 @@
+<?php
+
+use Rowbot\URL\URL;
+
+require_once __DIR__ . "/../bootstrap.php";
+
+if ( $argc < 2 ) {
+	echo "Usage: php script.php <command> --file <input-file> --from-site-url <current site url> --to-url <target url>\n";
+	echo "Commands:\n";
+	echo "  list_urls: List all the URLs found in the input file.\n";
+	echo "  migrate_urls: Migrate all the URLs found in the input file from the current site to the target site.\n";
+	exit( 1 );
+}
+
+$command = $argv[1];
+$options = [];
+
+for ( $i = 2; $i < $argc; $i ++ ) {
+	if ( str_starts_with( $argv[ $i ], '--' ) && isset( $argv[ $i + 1 ] ) ) {
+		$options[ substr( $argv[ $i ], 2 ) ] = $argv[ $i + 1 ];
+		$i ++;
+	}
+}
+
+if ( ! isset( $options['file'] ) ) {
+	echo "The file option is required.\n";
+	exit( 1 );
+}
+
+$inputFile    = $options['file'];
+$targetDomain = @$options['target-domain'];
+
+if ( ! file_exists( $inputFile ) ) {
+	echo "The file $inputFile does not exist.\n";
+	exit( 1 );
+}
+
+$block_markup = file_get_contents( $inputFile );
+
+// @TODO: Should a base URL be always required?
+$previous_url = $options['from-site-url'] ?? 'https://w.org';
+$p            = new WP_Block_Markup_Url_Processor( $block_markup, $previous_url );
+
+switch ( $command ) {
+	case 'list_urls':
+		echo "URLs found in the markup:\n\n";
+		while ( $p->next_url() ) {
+			// Skip empty relative URLs.
+			if ( ! trim( $p->get_url() ) ) {
+				continue;
+			}
+			echo '* ';
+			switch ( $p->get_token_type() ) {
+				case '#tag':
+					echo 'In <' . $p->get_tag() . '> tag attribute "' . $p->get_inspected_attribute_name() . '": ';
+					break;
+				case '#block-comment':
+					echo 'In a ' . $p->get_block_name() . ' block attribute "' . $p->get_block_attribute_key() . '": ';
+					break;
+				case '#text':
+					echo 'In #text: ';
+					break;
+			}
+			echo $p->get_url() . "\n";
+		}
+		echo "\n";
+		break;
+	case 'migrate_urls':
+		if ( ! isset( $options['from-site-url'] ) ) {
+			echo "The --from-site-url option is required for the migrate_urls command.\n";
+			exit( 1 );
+		}
+		if ( ! isset( $options['to-url'] ) ) {
+			echo "The --to-url option is required for the migrate_urls command.\n";
+			exit( 1 );
+		}
+		$parsed_prev_url = URL::parse( $options['from-site-url'] );
+		$next_url        = $options['to-url'];
+		$parsed_new_url  = URL::parse( $next_url );
+		echo "Replacing $previous_url with $next_url in the input.\n";
+		echo "Note this is not yet enough to migrate the site as both the previous and the new";
+		echo "site might be hosted on specific paths.\n\n";
+		while ( $p->next_url() ) {
+			$updated    = false;
+			$url        = $p->get_url();
+			$parsed_url = URL::parse( $url, $parsed_prev_url );
+			if ( $parsed_url->hostname === $parsed_prev_url->hostname ) {
+				$parsed_url->hostname = $parsed_new_url->hostname;
+				if ( str_starts_with( $parsed_url->pathname, $parsed_prev_url->pathname ) ) {
+					$parsed_url->pathname = $parsed_new_url->pathname . substr( $parsed_url->pathname, strlen( $parsed_prev_url->pathname ) );
+				}
+				$p->set_url( $parsed_url->toString() );
+			}
+		}
+		echo $p->get_updated_html();
+		break;
+}

transfer-protocol/bootstrap.php

@@ -36,3 +36,7 @@ function __($input) {
 function esc_attr($input) {
 	return htmlspecialchars($input);
 }
+
+function esc_html($input) {
+	return htmlspecialchars($input);
+}

transfer-protocol/married-short.html

@@ -0,0 +1,14 @@
+<!-- wp:image {"src": "https://mysite.com/wp-content/image.png", "meta": {"src": "https://mysite.com/wp-content/image.png"}} -->
+<img src="https://mysite.com/wp-content/image.png">
+<!-- /wp:image -->
+
+<!-- wp:paragraph -->
+<p>During the <a href="writeofpassage.school">Write of Passage</a>, I stubbornly tried to beat my writer’s block by writing until 3am multiple times. The burnout returned. I dropped everything and went to Greece for a week.</p>
+<!-- /wp:paragraph -->
+
+<!-- wp:paragraph -->
+<p>
+	Have you seen my blog, adamadam.blog? I told a story there of how I got my Bachelor's degree,
+	check it out: https://adamadam.blog/2021/09/16/how-i-got-bachelors-in-six-months/
+</p>
+<!-- /wp:paragraph -->

transfer-protocol/married.html

@@ -316,4 +316,4 @@ <h2 id="what-s-next">What’s next?</h2>
 
 <!-- wp:paragraph -->
 <p>I will make 2022 about doing&nbsp;<em>less</em>. My only map is my intuition.&nbsp;<a href="https://en.wikipedia.org/wiki/Amor_fati">Amor fati</a>, as the Stoics would say. I am open to whatever life will bring. Where is it going to take me? Who knows? We'll see next year!</p>
-<!-- /wp:paragraph -->
+<!-- /wp:paragraph -->

transfer-protocol/phpunit.xml

@@ -1,7 +1,11 @@
-<phpunit bootstrap="bootstrap.php">
+<phpunit
+    bootstrap="bootstrap.php"
+    colors="true"
+    verbose="true"
+>
     <testsuites>
         <testsuite name="Project Test Suite">
             <directory>./tests</directory>
         </testsuite>
     </testsuites>
-</phpunit>
+</phpunit>

transfer-protocol/run-tests.sh

@@ -1,6 +1,11 @@
 #!/bin/bash
 #COMMAND="phpunit tests/WP_Migration_*"
-#COMMAND="phpunit tests/WP_Block_Markup_P*"
-COMMAND="phpunit tests/WP_Block_Markup_Url*"
-$COMMAND
-fswatch -o ./**/*.php | xargs -n1 -I{} $COMMAND
+#COMMAND="phpunit tests/*.php"
+#COMMAND="phpunit tests/WP_Block_Markup_Url_Processor_Tests.php"
+#COMMAND="phpunit -c phpunit.xml"
+#$COMMAND
+#fswatch -o ./**/*.php | xargs -n1 -I{} $COMMAND
+
+for i in $(ls tests/*.php | grep -v URL_Parser); do
+    phpunit $i
+done