more test and comments

inc/migrate-post.php

@@ -45,11 +45,32 @@ public function __construct( $post ) {
 
 		$this->front_matter_src = $this->extract_front_matter( $this->post );
 		$this->front_matter = $this->transform_front_matter( $this->front_matter_src );
-		$this->content = $this->transform_post_content( $this->post->post_content );
+		$this->markdown = $this->transform_post_content( $this->post->post_content );
+
+		// Check content for excessive HTML tags, flag for manual inspection
+		$this->num_tags = $this->count_tags( $this->markdown );
+		if ( 10 < $this->num_tags ) {
+			return array( 'error' => sprintf( 'Markdown for post %s contains %s HTML tags; manual inspection required.', $this->post->ID, $this->num_tags ) );
+		}
+
+		// Write output to file
 
 		return array( 'success' => 'Migrated post ' . $this->post->ID );
 	}
 
+	/**
+	 * Count the number of HTML tags in the string. Used to flag Markdown for manual review
+	 *
+	 * @param string $input
+	 * @param int
+	 */
+	public function count_tags( $input ) {
+		// split string by html opening tags, doesn't need to be exact
+		$count = count( preg_split( '/<([A-Z][A-Z0-9]*)\b[^>]*>/i', $input ) );
+
+		return $count - 1;
+	}
+
 	/**
 	 * Extract data that will be converted to YAML front matter
 	 *
@@ -122,6 +143,7 @@ public function transform_post_content( $content ) {
 
 	/**
 	 * Handle some known gotchas in WP -> Markdown
+	 * @todo unit testing for this method
 	 */
 	public function filter_markdown( $markdown ) {
 		// Update links
@@ -209,6 +231,9 @@ public function convert_link_embeds( $content ) {
 		return $content;
 	}
 
+	/**
+	 * Convert Jetpack media embeds and other shortcodes
+	 */
 	public function convert_shortcodes( $content ) {
 		// Override Jetpack shortcodes
 		add_shortcode( 'youtube', function( $atts ) {
@@ -220,10 +245,21 @@ public function convert_shortcodes( $content ) {
 			return $this->hugo_shortcode( 'vimeo', $id );
 		} );
 
+		/**
+		 * @todo handle quote, caption
+		 */
+
 		$content = do_shortcode( $content );
 		return $content;
 	}
 
+	/**
+	 * Generate Hugo shortcode, e.g. `{{< tweet 12341234 >}}`
+	 *
+	 * @param string $provider
+	 * @param string|int $id Can be id or link
+	 * @param bool $from_link If true, parse id from URL depending on provider
+	 */
 	public function hugo_shortcode( $provider, $id, $from_link = false ) {
 		// Convert URL to ID if needed
 		if ( $from_link) {

tests/data/test/count_tags.html

@@ -0,0 +1,7 @@
+<ul>
+	<li>Dublin is having a merry <a href="https://www.meetup.com/hacks-hackers-dublin/events/236094987/">Christmas gathering</a></li>
+	<li>San Francisco is <a href="https://www.eventbrite.com/e/holiday-party-for-bay-area-journalists-tickets-27677632544">meeting</a> Solutions Journalism Network's Bay Area chapter for a drink</li>
+	<li><a href="http://www.meetup.com/Hacks-Hackers-Miami/">Miami</a> is holding its regular OpenHack Miami</li>
+	<li><a href="http://www.meetup.com/hackshackersIRE/">IRE</a> in Missouri is holding its weekly open lab</li>
+	<li>Nairobi is <a href="https://docs.google.com/forms/d/13iJf6EHjGSSPJWCCgv5SSrSRn43yoMP-uuITZ4sgwvI/viewform?edit_requested=true">discussing how</a> governments can be held more accountable</li>
+</ul>

tests/test-migrate-post.php

@@ -40,6 +40,17 @@ public function setUp() {
 		$this->migrator = new HH_Hugo\Migrate_Post('test');
 	}
 
+	/**
+	 * Get contents of test data file
+	 *
+	 * @param string $type 'test' or 'expect'
+	 * @param string $filename
+	 * @return string Contents of file
+	 */
+	private function _get_test_data( $type, $filename ) {
+		return file_get_contents( HH_HUGO_COMMAND_DIR . '/tests/data/' . $type . '/' . $filename );
+	}
+
 	public function test_class_exists() {
 		$this->assertTrue( class_exists( '\\HH_Hugo\\Migrate_Post' ) );
 	}
@@ -59,7 +70,7 @@ public function test_extract_front_matter() {
 		// Delete migration data since we can't test the ID and timestamp
 		unset( $front_matter['_migration'] );
 		$this->assertEquals(
-			file_get_contents( HH_HUGO_COMMAND_DIR . '/tests/data/expect/front-matter.yml' ),
+			$this->_get_test_data( 'expect', 'front-matter.yml' ),
 			$this->migrator->transform_front_matter( $front_matter )
 		);
 
@@ -121,13 +132,22 @@ public function test_transform_post_content() {
 	}
 
 	private function _test_transform_post_content( $filename, $dump = false ) {
-		$input = file_get_contents( HH_HUGO_COMMAND_DIR . '/tests/data/test/' . $filename . '.html' );
-		$output = file_get_contents( HH_HUGO_COMMAND_DIR . '/tests/data/expect/' . $filename . '.md' );
+		$input = $this->_get_test_data( 'test', $filename . '.html' );
+		$output = $this->_get_test_data( 'expect', $filename . '.md' );
 
 		if ( $dump ) {
 			die( var_dump( $this->migrator->transform_post_content( $input ) ) );
 		}
 
 		$this->assertEquals( $output, $this->migrator->transform_post_content( $input ) );
 	}
+
+	public function test_count_tags() {
+		$input = $this->_get_test_data( 'test', 'count_tags.html' );
+		$this->assertEquals( 11, $this->migrator->count_tags( $input ) );
+		$this->assertEquals( 1, $this->migrator->count_tags( '<h1>title</h1>' ) );
+		$this->assertEquals( 1, $this->migrator->count_tags( '<img src="http://foo.bar/img.png" />' ) );
+		$this->assertEquals( 2, $this->migrator->count_tags( '<p><span>hi</span></p>' ) );
+		$this->assertEquals( 0, $this->migrator->count_tags( 'hello world' ) );
+	}
 }