<?php
namespace Elementor\Modules\Agents\Components\Readability\Extractors;
use Elementor\Modules\MarkdownRender\Html_To_Markdown;
use Elementor\Modules\MarkdownRender\Module;
if ( ! defined( 'ABSPATH' ) ) {
exit;
}
/**
* Extracts markdown from posts written with the WordPress block editor.
*
* Walks the parsed block tree recursively and converts each known core block
* to its markdown equivalent. Unrecognised or dynamic blocks (e.g. containers
* such as `core/group` or `core/columns`) are rendered via `render_block()`
* and then converted through `Html_To_Markdown`; their inner blocks are only
* walked separately when that rendered markdown is empty, to avoid duplicating
* content that `render_block()` already includes.
*
* Priority: 20 (runs after Elementor_Extractor).
*/
class Block_Extractor implements Extractor_Interface {
/**
* Return true when the post content contains at least one block.
*
* @param \WP_Post $post
* @return bool
*/
public function can_handle( \WP_Post $post ): bool {
return has_blocks( $post->post_content );
}
/**
* Parse the block tree and convert it to markdown.
*
* @param \WP_Post $post
* @return string Body markdown, or '' on failure.
*/
public function extract( \WP_Post $post ): string {
try {
return Module::execute_while_rendering_markdown( function () use ( $post ) {
$blocks = parse_blocks( $post->post_content );
$parts = [];
foreach ( $blocks as $block ) {
$md = $this->render_block_to_markdown( $block );
if ( '' !== trim( $md ) ) {
$parts[] = trim( $md );
}
}
return implode( "\n\n", $parts );
} );
} catch ( \Throwable $e ) {
return '';
}
}
/**
* @return string
*/
public function get_id(): string {
return 'block';
}
/**
* @return int
*/
public function get_priority(): int {
return 20;
}
// -------------------------------------------------------------------------
// Private helpers
// -------------------------------------------------------------------------
/**
* Convert a single parsed block (and its inner blocks) to markdown.
*
* @param array $block Parsed block array from `parse_blocks()`.
* @return string
*/
private function render_block_to_markdown( array $block ): string {
$name = $block['blockName'] ?? '';
$attrs = $block['attrs'] ?? [];
$inner = $block['innerHTML'] ?? '';
switch ( $name ) {
case 'core/heading':
$level = isset( $attrs['level'] ) ? (int) $attrs['level'] : 2;
$level = max( 1, min( 6, $level ) );
$text = wp_strip_all_tags( $inner );
$text = html_entity_decode( trim( $text ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
return str_repeat( '#', $level ) . ' ' . $text;
case 'core/paragraph':
case 'core/preformatted':
$text = wp_strip_all_tags( $inner );
$text = html_entity_decode( trim( $text ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
return $text;
case 'core/quote':
$text = wp_strip_all_tags( $inner );
$text = html_entity_decode( trim( $text ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
$lines = explode( "\n", $text );
return implode( "\n", array_map(
static fn( string $line ) => '> ' . $line,
$lines
) );
case 'core/list':
return $this->convert_list_html( $inner );
case 'core/image':
$src = $attrs['url'] ?? '';
$alt = $attrs['alt'] ?? '';
if ( '' === $src ) {
return '';
}
return '';
case 'core/code':
$code = isset( $attrs['content'] )
? html_entity_decode( $attrs['content'], ENT_QUOTES | ENT_HTML5, 'UTF-8' )
: wp_strip_all_tags( $inner );
$code = rtrim( $code, "\n" );
return "```\n" . $code . "\n```";
case 'core/separator':
return '---';
case 'core/table':
return $this->convert_table_html( $inner );
case null:
case '':
// Freeform / null block — recurse inner blocks, convert innerHTML.
$parts = $this->recurse_inner_blocks( $block );
if ( '' !== trim( $inner ) && empty( $block['innerBlocks'] ) ) {
$parts[] = Html_To_Markdown::convert( $inner );
}
return implode( "\n\n", array_filter( $parts, static fn( $p ) => '' !== trim( $p ) ) );
default:
// Unrecognised / dynamic block: render via WordPress then convert.
// `render_block()` already includes the markup of every inner
// block, so only fall back to walking `innerBlocks` when the
// rendered markdown is empty (e.g. an empty wrapper element) —
// otherwise the inner content would be duplicated.
$rendered_markdown = Html_To_Markdown::convert( render_block( $block ) );
if ( '' !== trim( $rendered_markdown ) ) {
return $rendered_markdown;
}
$parts = $this->recurse_inner_blocks( $block );
return implode( "\n\n", array_filter( $parts, static fn( $p ) => '' !== trim( $p ) ) );
}
}
/**
* Recursively convert inner blocks to markdown strings.
*
* @param array $block
* @return string[]
*/
private function recurse_inner_blocks( array $block ): array {
$parts = [];
foreach ( $block['innerBlocks'] ?? [] as $inner_block ) {
$md = $this->render_block_to_markdown( $inner_block );
if ( '' !== trim( $md ) ) {
$parts[] = trim( $md );
}
}
return $parts;
}
/**
* Convert a list block's inner HTML to a markdown list.
*
* Extracts `<li>` text values and prefixes each with `- `.
*
* @param string $html
* @return string
*/
private function convert_list_html( string $html ): string {
if ( '' === $html ) {
return '';
}
preg_match_all( '#<li[^>]*>(.*?)</li>#is', $html, $matches );
if ( empty( $matches[1] ) ) {
return Html_To_Markdown::convert( $html );
}
$lines = [];
foreach ( $matches[1] as $item ) {
$text = wp_strip_all_tags( $item );
$text = html_entity_decode( trim( $text ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
if ( '' !== $text ) {
$lines[] = '- ' . $text;
}
}
return implode( "\n", $lines );
}
/**
* Convert a table block's inner HTML to a best-effort plain-text table.
*
* @param string $html
* @return string
*/
private function convert_table_html( string $html ): string {
if ( '' === $html ) {
return '';
}
// Delegate to Html_To_Markdown which already handles <table> → markdown.
return Html_To_Markdown::convert( $html );
}
}