<?php
/**
 * A post as the video API's ordered content: its title and its `blocks` of text
 * and pictures.
 *
 * @package Ojelio
 */

namespace Ojelio;

use WP_Post;

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

final class Payload {

	/** The API's limits (apps/web/src/lib/video-options.ts and import/content.ts). */
	const TITLE_MAX  = 120;
	const MAX_IMAGES = 10;
	const ALT_MAX    = 300;

	/** Blocks whose text is read: their own HTML and their inner blocks'. */
	const TEXT_BLOCKS = array(
		'core/paragraph',
		'core/heading',
		'core/list',
		'core/list-item',
		'core/quote',
		'core/pullquote',
		'core/verse',
		'core/preformatted',
		'core/details',
	);

	/** Blocks that only hold others: read through. */
	const CONTAINERS = array( 'core/group', 'core/columns', 'core/column', 'core/row', 'core/stack', 'core/gallery' );

	/**
	 * The post's content as the API takes it.
	 *
	 * @param WP_Post $post The post.
	 * @return array {
	 *     @type string $title       Up to 120 characters.
	 *     @type array  $blocks      In order: array( 'type' => 'text', 'content' ) and array( 'type' => 'image', 'url', 'alt'? ); the featured image first.
	 *     @type string $source_url  The post's address once it is published; null before.
	 *     @type int    $words       The text's words.
	 *     @type array  $left_out    Pictures not sent, as array( 'url', 'why' ): "private" (Ojelio cannot fetch it) or "limit".
	 * }
	 */
	public static function from_post( WP_Post $post ) {
		$items = array();
		foreach ( parse_blocks( $post->post_content ) as $block ) {
			self::read_block( $block, $items );
		}
		// The featured image opens the video, unless the content already shows it.
		$featured = get_post_thumbnail_id( $post );
		$featured = $featured ? self::image_of( '', (int) $featured ) : null;
		if ( $featured && ! in_array( $featured['url'], wp_list_pluck( wp_list_filter( $items, array( 'type' => 'image' ) ), 'url' ), true ) ) {
			array_unshift( $items, $featured );
		}

		$blocks   = array();
		$left_out = array();
		$images   = 0;
		foreach ( $items as $item ) {
			if ( 'text' === $item['type'] ) {
				// Text that follows text joins it: the script reads on, and the list stays short.
				$last = count( $blocks ) - 1;
				if ( $last >= 0 && 'text' === $blocks[ $last ]['type'] ) {
					$blocks[ $last ]['content'] .= "\n\n" . $item['content'];
				} else {
					$blocks[] = $item;
				}
				continue;
			}
			if ( ! self::is_public_url( $item['url'] ) ) {
				$left_out[] = array( 'url' => $item['url'], 'why' => 'private' );
			} elseif ( $images >= self::MAX_IMAGES ) {
				$left_out[] = array( 'url' => $item['url'], 'why' => 'limit' );
			} else {
				++$images;
				$blocks[] = $item;
			}
		}

		$words = 0;
		foreach ( $blocks as $block ) {
			if ( 'text' === $block['type'] ) {
				$words += count( preg_split( '/\s+/u', $block['content'], -1, PREG_SPLIT_NO_EMPTY ) );
			}
		}

		return array(
			'title'      => self::title( $post ),
			'blocks'     => $blocks,
			'source_url' => 'publish' === $post->post_status ? get_permalink( $post ) : null,
			'words'      => $words,
			'left_out'   => $left_out,
		);
	}

	public static function title( WP_Post $post ) {
		$title = self::clean( get_the_title( $post ) );
		if ( '' === $title ) {
			$title = __( 'Untitled post', 'ojelio' );
		}
		return function_exists( 'mb_substr' ) ? mb_substr( $title, 0, self::TITLE_MAX ) : substr( $title, 0, self::TITLE_MAX );
	}

	/**
	 * Ojelio fetches the pictures itself, so an address on this computer or a
	 * private network would fail the whole video: such pictures are left out.
	 */
	public static function is_public_url( $url ) {
		$parts = wp_parse_url( $url );
		if ( ! $parts || empty( $parts['host'] ) || empty( $parts['scheme'] ) || ! in_array( strtolower( $parts['scheme'] ), array( 'http', 'https' ), true ) ) {
			return false;
		}
		$host = strtolower( trim( $parts['host'], '[]' ) );
		if ( filter_var( $host, FILTER_VALIDATE_IP ) ) {
			return (bool) filter_var( $host, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE );
		}
		if ( false === strpos( $host, '.' ) ) {
			return false;
		}
		return ! preg_match( '/\.(localhost|local|test|internal|lan|home|example|invalid)$/', $host );
	}

	// ---- Reading blocks ----------------------------------------------------------------

	private static function read_block( array $block, array &$items ) {
		$name = $block['blockName'];

		if ( null === $name ) {
			// Content from before the block editor (or between blocks): HTML as it is.
			self::read_html( $block['innerHTML'], $items );
			return;
		}
		if ( in_array( $name, self::TEXT_BLOCKS, true ) ) {
			self::add_text( self::text_of( $block ), $items );
			return;
		}
		if ( 'core/image' === $name ) {
			$image = self::image_of( $block['innerHTML'], isset( $block['attrs']['id'] ) ? (int) $block['attrs']['id'] : 0 );
			if ( $image ) {
				$items[] = $image;
			}
			return;
		}
		if ( 'core/media-text' === $name || 'core/cover' === $name ) {
			// The picture first, then the text over or beside it.
			$id    = isset( $block['attrs']['mediaId'] ) ? (int) $block['attrs']['mediaId'] : ( isset( $block['attrs']['id'] ) ? (int) $block['attrs']['id'] : 0 );
			$image = self::image_of( $block['innerHTML'], $id, isset( $block['attrs']['url'] ) ? $block['attrs']['url'] : null );
			if ( $image && ( 'core/media-text' !== $name || ! isset( $block['attrs']['mediaType'] ) || 'image' === $block['attrs']['mediaType'] ) ) {
				$items[] = $image;
			}
		} elseif ( ! in_array( $name, self::CONTAINERS, true ) && 'core/block' !== $name ) {
			// Embeds, code, tables, buttons, shortcodes, other plugins' blocks: left out.
			return;
		}
		if ( 'core/block' === $name && ! empty( $block['attrs']['ref'] ) ) {
			// A synced pattern: its own post's blocks.
			$pattern = get_post( (int) $block['attrs']['ref'] );
			if ( $pattern && 'wp_block' === $pattern->post_type && 'publish' === $pattern->post_status ) {
				foreach ( parse_blocks( $pattern->post_content ) as $inner ) {
					self::read_block( $inner, $items );
				}
			}
			return;
		}
		foreach ( $block['innerBlocks'] as $inner ) {
			self::read_block( $inner, $items );
		}
	}

	/** A text block's words: its own HTML and its inner blocks', in order. */
	private static function text_of( array $block ) {
		$html  = '';
		$inner = 0;
		foreach ( $block['innerContent'] as $piece ) {
			if ( null === $piece ) {
				$html .= "\n" . self::text_of( $block['innerBlocks'][ $inner ] ) . "\n";
				++$inner;
			} else {
				$html .= $piece;
			}
		}
		return self::html_to_text( $html );
	}

	/** Classic content: paragraphs and pictures in the order they come. */
	private static function read_html( $html, array &$items ) {
		$html  = wpautop( $html );
		$parts = preg_split( '/(<img\b[^>]*>)/i', $html, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY );
		foreach ( $parts as $part ) {
			if ( preg_match( '/^<img\b/i', $part ) ) {
				$image = self::image_of( $part, self::attachment_id_of( $part ) );
				if ( $image ) {
					$items[] = $image;
				}
			} else {
				self::add_text( self::html_to_text( $part ), $items );
			}
		}
	}

	private static function add_text( $text, array &$items ) {
		if ( '' !== $text ) {
			$items[] = array( 'type' => 'text', 'content' => $text );
		}
	}

	/**
	 * A picture: the full-size file of the attachment when there is one (the block's
	 * HTML usually shows a smaller size), else the image's own address.
	 */
	private static function image_of( $html, $attachment_id, $fallback_url = null ) {
		$url = $attachment_id ? wp_get_attachment_url( $attachment_id ) : null;
		if ( ! $url && preg_match( '/<img\b[^>]*\bsrc=(["\'])(.*?)\1/i', $html, $match ) ) {
			$url = html_entity_decode( $match[2], ENT_QUOTES, 'UTF-8' );
		}
		if ( ! $url ) {
			$url = $fallback_url;
		}
		if ( ! $url ) {
			return null;
		}
		$alt = '';
		if ( preg_match( '/<img\b[^>]*\balt=(["\'])(.*?)\1/i', $html, $match ) ) {
			$alt = self::clean( $match[2] );
		}
		if ( '' === $alt && $attachment_id ) {
			$alt = self::clean( (string) get_post_meta( $attachment_id, '_wp_attachment_image_alt', true ) );
		}
		$image = array( 'type' => 'image', 'url' => esc_url_raw( $url ) );
		if ( '' !== $alt ) {
			$image['alt'] = function_exists( 'mb_substr' ) ? mb_substr( $alt, 0, self::ALT_MAX ) : substr( $alt, 0, self::ALT_MAX );
		}
		return $image;
	}

	private static function attachment_id_of( $img ) {
		return preg_match( '/\bwp-image-(\d+)\b/', $img, $match ) ? (int) $match[1] : 0;
	}

	/** Visible text: line breaks kept between blocks of HTML, tags, shortcodes and entities gone. */
	private static function html_to_text( $html ) {
		$html = strip_shortcodes( $html );
		$html = preg_replace( '#<(script|style)\b[^>]*>.*?</\1>#is', '', $html );
		$html = preg_replace( '#<br\s*/?>|</(p|li|h[1-6]|blockquote|cite|summary|div|pre)>#i', "\n", $html );
		$text = html_entity_decode( wp_strip_all_tags( $html, false ), ENT_QUOTES, 'UTF-8' );
		$text = preg_replace( "/[ \t\x{00A0}]+/u", ' ', $text );
		$text = preg_replace( "/ *\n */", "\n", $text );
		$text = preg_replace( "/\n{2,}/", "\n", $text );
		return trim( $text );
	}

	private static function clean( $text ) {
		return trim( preg_replace( '/\s+/u', ' ', html_entity_decode( wp_strip_all_tags( $text ), ENT_QUOTES, 'UTF-8' ) ) );
	}
}