<?php
/**
 * HTML parsing for scans (S-SCAN).
 *
 * @package ForensixDigital
 */

declare(strict_types=1);

namespace ForensixDigital\Services\Scan;

defined( 'ABSPATH' ) || exit;

// phpcs:disable WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase -- DOM property names.

use ForensixDigital\Services\Sources\Source;

/**
 * Extracts links and resources from a page and attributes each to the innermost source marker
 * around it (D-21). Read-only; never changes the HTML.
 */
final class HtmlParser {

	private const TEXT_LENGTH = 120;

	/**
	 * Open markers, innermost last.
	 *
	 * @var Source[]
	 */
	private array $stack = array();

	/**
	 * Collected references.
	 *
	 * @var Reference[]
	 */
	private array $references = array();

	/**
	 * Collected headings.
	 *
	 * @var Heading[]
	 */
	private array $headings = array();

	/**
	 * Accessibility problems found while walking: kind, text, source type, source reference.
	 *
	 * @var array<int, array{0: string, 1: string, 2: string, 3: string}>
	 */
	private array $a11y = array();

	/**
	 * Content facts of the page being parsed: kind, text, source type, source reference.
	 *
	 * @var array<int, array{0: string, 1: string, 2: string, 3: string}>
	 */
	private array $content = array();

	/**
	 * Ids of fields that a <label for> points to.
	 *
	 * @var array<string, true>
	 */
	private array $label_for = array();

	/**
	 * How often each id appears, with the source of its first use: id => [count, type, ref].
	 *
	 * @var array<string, array{0: int, 1: string, 2: string}>
	 */
	private array $ids = array();

	/**
	 * Links seen so far: whether the first few include an in-page link, and the number of links.
	 *
	 * @var array{skip: bool, links: int}
	 */
	private array $links = array(
		'skip'  => false,
		'links' => 0,
	);

	/**
	 * Performance facts of the page: kind, text, source type, source reference.
	 *
	 * @var array<int, array{0: string, 1: string, 2: string, 3: string}>
	 */
	private array $perf = array();

	/**
	 * Images seen so far, in document order.
	 *
	 * @var int
	 */
	private int $image_count = 0;

	/**
	 * Base URL for relative values.
	 *
	 * @var string
	 */
	private string $base = '';

	/**
	 * Parses a page.
	 *
	 * @param string $html    HTML.
	 * @param string $url     Page URL, used to resolve relative URLs.
	 * @param array  $headers Response headers with lower-case names; X-Robots-Tag is read from them.
	 * @phpstan-param array<string, string> $headers
	 */
	public function parse( string $html, string $url, array $headers = array() ): ParsedPage {
		$this->stack      = array();
		$this->references = array();
		$this->base       = $url;

		$dom      = new \DOMDocument();
		$previous = libxml_use_internal_errors( true );
		$dom->loadHTML( '<?xml encoding="UTF-8">' . $html, LIBXML_NONET | LIBXML_COMPACT | LIBXML_NOERROR | LIBXML_NOWARNING );
		libxml_clear_errors();
		libxml_use_internal_errors( $previous );

		$title   = '';
		$noindex = false;
		$facts   = array(
			'title_count'  => 0,
			'descriptions' => array(),
			'canonicals'   => array(),
			'robots'       => array(),
			'open_graph'   => array(),
			'twitter'      => array(),
			'json_ld'      => array(),
			'hreflang'     => array(),
		);
		foreach ( $dom->getElementsByTagName( 'base' ) as $base ) {
			$resolved = Url::resolve( $base->getAttribute( 'href' ), $url );
			if ( null !== $resolved ) {
				$this->base = $resolved;
			}
			break;
		}
		foreach ( $dom->getElementsByTagName( 'title' ) as $node ) {
			if ( 0 === $facts['title_count'] ) {
				$title = trim( (string) preg_replace( '/\s+/', ' ', $node->textContent ) );
			}
			++$facts['title_count'];
		}
		foreach ( $dom->getElementsByTagName( 'meta' ) as $meta ) {
			$name    = strtolower( trim( $meta->getAttribute( 'name' ) ) );
			$prop    = strtolower( trim( $meta->getAttribute( 'property' ) ) );
			$content = trim( $meta->getAttribute( 'content' ) );
			if ( 'robots' === $name ) {
				$facts['robots'][] = strtolower( $content );
				if ( str_contains( strtolower( $content ), 'noindex' ) ) {
					$noindex = true;
				}
			} elseif ( 'description' === $name ) {
				$facts['descriptions'][] = trim( (string) preg_replace( '/\s+/', ' ', $content ) );
			} elseif ( str_starts_with( $prop, 'og:' ) ) {
				$facts['open_graph'][ $prop ][] = $content;
			} elseif ( str_starts_with( $name, 'twitter:' ) ) {
				$facts['twitter'][ $name ][] = $content;
			}
		}
		foreach ( $dom->getElementsByTagName( 'link' ) as $link ) {
			$rels = preg_split( '/\s+/', strtolower( trim( $link->getAttribute( 'rel' ) ) ) );
			if ( is_array( $rels ) && in_array( 'canonical', $rels, true ) ) {
				$href                  = trim( $link->getAttribute( 'href' ) );
				$facts['canonicals'][] = Url::resolve( $href, $this->base ) ?? $href;
			}
			if ( is_array( $rels ) && in_array( 'alternate', $rels, true ) && '' !== trim( $link->getAttribute( 'hreflang' ) ) ) {
				$href                = trim( $link->getAttribute( 'href' ) );
				$facts['hreflang'][] = array( trim( $link->getAttribute( 'hreflang' ) ), Url::resolve( $href, $this->base ) ?? $href );
			}
		}
		foreach ( $dom->getElementsByTagName( 'script' ) as $script ) {
			if ( 'application/ld+json' === strtolower( trim( $script->getAttribute( 'type' ) ) ) ) {
				$facts['json_ld'][] = trim( $script->textContent );
			}
		}

		$header_robots = strtolower( trim( $headers['x-robots-tag'] ?? '' ) );
		if ( str_contains( $header_robots, 'noindex' ) ) {
			$noindex = true;
		}

		$this->perf        = array( array( 'html_bytes', (string) strlen( $html ), Source::TEMPLATE, '' ) );
		$this->image_count = 0;

		$this->headings  = array();
		$this->a11y      = array();
		$this->content   = array();
		$this->ids       = array();
		$this->label_for = array();
		$this->links     = array(
			'skip'  => false,
			'links' => 0,
		);
		foreach ( $dom->getElementsByTagName( 'label' ) as $label ) {
			if ( '' !== trim( $label->getAttribute( 'for' ) ) ) {
				$this->label_for[ trim( $label->getAttribute( 'for' ) ) ] = true;
			}
		}
		$this->walk( $dom );
		$this->document_problems( $dom );
		$this->content_facts( $dom );
		foreach ( $facts['hreflang'] as $tag ) {
			$this->content[] = array( 'hreflang', $tag[0] . '|' . $tag[1], Source::TEMPLATE, '' );
		}
		$sketch = TextSketch::of( self::main_text( $dom ) );

		return new ParsedPage(
			$url,
			$title,
			$noindex,
			$this->references,
			new PageFacts(
				$facts['title_count'],
				$facts['descriptions'],
				$facts['canonicals'],
				$facts['robots'],
				$facts['open_graph'],
				$facts['twitter'],
				$this->headings,
				$facts['json_ld'],
				$header_robots,
				$sketch,
				$this->a11y,
				$this->content,
				$this->perf
			)
		);
	}

	/**
	 * Text of the main part of the page: the <main> element, or the body without navigation,
	 * header, footer, sidebars, forms and scripts. Changes the document, so it runs last.
	 *
	 * @param \DOMDocument $dom Document.
	 */
	private static function main_text( \DOMDocument $dom ): string {
		$root = $dom->getElementsByTagName( 'main' )->item( 0 ) ?? $dom->getElementsByTagName( 'body' )->item( 0 );
		if ( ! $root instanceof \DOMElement ) {
			return '';
		}
		foreach ( array( 'script', 'style', 'noscript', 'nav', 'header', 'footer', 'aside', 'form' ) as $tag ) {
			foreach ( iterator_to_array( $root->getElementsByTagName( $tag ) ) as $node ) {
				$node->parentNode?->removeChild( $node );
			}
		}
		return $root->textContent;
	}

	/**
	 * Visits nodes in document order.
	 *
	 * @param \DOMNode $node Parent node.
	 */
	private function walk( \DOMNode $node ): void {
		foreach ( $node->childNodes as $child ) {
			if ( $child instanceof \DOMComment ) {
				$this->marker( $child->data );
			} elseif ( $child instanceof \DOMElement ) {
				$this->element( $child );
				if ( $child->hasChildNodes() ) {
					$this->walk( $child );
				}
			}
		}
	}

	/**
	 * Opens or closes a source marker.
	 *
	 * @param string $comment Comment text.
	 */
	private function marker( string $comment ): void {
		$marker = Source::parse_marker( $comment );
		if ( null === $marker ) {
			return;
		}
		[ $kind, $source ] = $marker;
		if ( 's' === $kind ) {
			$this->stack[] = $source;
			return;
		}
		// Close the matching marker and anything left open inside it.
		for ( $i = count( $this->stack ) - 1; $i >= 0; $i-- ) {
			if ( $this->stack[ $i ]->type === $source->type && $this->stack[ $i ]->ref === $source->ref ) {
				$this->stack = array_slice( $this->stack, 0, $i );
				return;
			}
		}
	}

	/**
	 * Collects the URLs of one element.
	 *
	 * @param \DOMElement $el Element.
	 */
	private function element( \DOMElement $el ): void {
		$tag = strtolower( $el->tagName );

		if ( 1 === preg_match( '/^h([1-6])$/', $tag, $level ) ) {
			$text = trim( (string) preg_replace( '/\s+/', ' ', $el->textContent ) );
			if ( '' === $text ) {
				foreach ( $el->getElementsByTagName( 'img' ) as $img ) {
					$text = trim( $img->getAttribute( 'alt' ) );
					break;
				}
			}
			$this->headings[] = new Heading( (int) $level[1], mb_substr( $text, 0, self::TEXT_LENGTH ), array() === $this->stack ? Source::template() : $this->stack[ count( $this->stack ) - 1 ] );
		}

		$this->accessibility( $el, $tag );
		$this->performance( $el, $tag );

		switch ( $tag ) {
			case 'a':
			case 'area':
				$text = trim( (string) preg_replace( '/\s+/', ' ', $el->textContent ) );
				if ( '' === $text ) {
					foreach ( $el->getElementsByTagName( 'img' ) as $img ) {
						$text = trim( $img->getAttribute( 'alt' ) );
						break;
					}
				}
				$this->add( Reference::LINK, $el->getAttribute( 'href' ), $tag, $text );
				break;
			case 'img':
				$alt = $el->getAttribute( 'alt' );
				$this->add( Reference::IMAGE, $el->getAttribute( 'src' ), $tag, $alt );
				$this->add( Reference::IMAGE, $el->getAttribute( 'data-src' ), $tag, $alt );
				$this->add_srcset( $el->getAttribute( 'srcset' ), $tag, $alt );
				$this->add_srcset( $el->getAttribute( 'data-srcset' ), $tag, $alt );
				break;
			case 'source':
				$parent = $el->parentNode instanceof \DOMElement ? strtolower( $el->parentNode->tagName ) : '';
				if ( 'picture' === $parent ) {
					$this->add_srcset( $el->getAttribute( 'srcset' ), $tag, '' );
				} else {
					$this->add( Reference::MEDIA, $el->getAttribute( 'src' ), $tag, '' );
				}
				break;
			case 'video':
			case 'audio':
				$this->add( Reference::MEDIA, $el->getAttribute( 'src' ), $tag, '' );
				$this->add( Reference::IMAGE, $el->getAttribute( 'poster' ), $tag, '' );
				break;
			case 'script':
				$this->add( Reference::SCRIPT, $el->getAttribute( 'src' ), $tag, '' );
				break;
			case 'link':
				$rel = ' ' . strtolower( $el->getAttribute( 'rel' ) ) . ' ';
				if ( str_contains( $rel, ' stylesheet ' ) ) {
					$this->add( Reference::STYLE, $el->getAttribute( 'href' ), $tag, '' );
				}
				break;
			case 'iframe':
				$this->add( Reference::FRAME, $el->getAttribute( 'src' ), $tag, $el->getAttribute( 'title' ) );
				break;
		}

		$style = $el->getAttribute( 'style' );
		if ( '' !== $style && preg_match_all( '/url\(\s*([\'"]?)(.*?)\1\s*\)/i', $style, $m ) ) {
			foreach ( $m[2] as $value ) {
				$this->add( Reference::IMAGE, $value, $tag, '' );
			}
		}
	}

	/**
	 * Notes scripts that block the first paint and the lazy loading of images.
	 *
	 * @param \DOMElement $el  Element.
	 * @param string      $tag Lower-case element name.
	 */
	private function performance( \DOMElement $el, string $tag ): void {
		$source = array() === $this->stack ? Source::template() : $this->stack[ count( $this->stack ) - 1 ];
		if ( 'script' === $tag ) {
			$src  = trim( $el->getAttribute( 'src' ) );
			$type = strtolower( trim( $el->getAttribute( 'type' ) ) );
			$url  = '' === $src ? null : Url::resolve( $src, $this->base );
			if ( null !== $url && ! $el->hasAttribute( 'async' ) && ! $el->hasAttribute( 'defer' ) && 'module' !== $type && self::inside( $el, 'head' ) ) {
				$this->perf[] = array( 'blocking_script', $url, $source->type, $source->ref );
			}
			return;
		}
		if ( 'img' !== $tag || self::inside( $el, 'noscript' ) ) {
			return;
		}
		$src = trim( $el->getAttribute( 'src' ) );
		$url = '' === $src || str_starts_with( $src, 'data:' ) ? null : Url::resolve( $src, $this->base );
		if ( null === $url || ( '1' === $el->getAttribute( 'width' ) && '1' === $el->getAttribute( 'height' ) ) ) {
			return;
		}
		++$this->image_count;
		$lazy = 'lazy' === strtolower( trim( $el->getAttribute( 'loading' ) ) );
		if ( 1 === $this->image_count && $lazy ) {
			$this->perf[] = array( 'lazy_first', $url, $source->type, $source->ref );
		} elseif ( $this->image_count > 3 && ! $lazy && 'high' !== strtolower( $el->getAttribute( 'fetchpriority' ) ) ) {
			$this->perf[] = array( 'lazy_missing', $url, $source->type, $source->ref );
		}
	}

	/**
	 * Whether an element sits inside another element.
	 *
	 * @param \DOMElement $el  Element.
	 * @param string      $tag Lower-case name of the ancestor.
	 */
	private static function inside( \DOMElement $el, string $tag ): bool {
		for ( $node = $el->parentNode; $node instanceof \DOMElement; $node = $node->parentNode ) {
			if ( strtolower( $node->tagName ) === $tag ) {
				return true;
			}
		}
		return false;
	}

	/**
	 * Notes accessibility problems of one element, and the ids and links needed for page-wide ones.
	 *
	 * @param \DOMElement $el  Element.
	 * @param string      $tag Lower-case element name.
	 */
	private function accessibility( \DOMElement $el, string $tag ): void {
		$source = array() === $this->stack ? Source::template() : $this->stack[ count( $this->stack ) - 1 ];
		foreach ( A11yFacts::problems( $el, $this->label_for ) as $problem ) {
			$this->a11y[] = array( $problem[0], mb_substr( $problem[1], 0, self::TEXT_LENGTH ), $source->type, $source->ref );
		}
		$id = $el->getAttribute( 'id' );
		if ( '' !== $id ) {
			$seen             = $this->ids[ $id ] ?? array( 0, $source->type, $source->ref );
			$this->ids[ $id ] = array( $seen[0] + 1, $seen[1], $seen[2] );
		}
		if ( 'a' === $tag && '' !== trim( $el->getAttribute( 'href' ) ) ) {
			++$this->links['links'];
			if ( $this->links['links'] <= 8 && str_starts_with( trim( $el->getAttribute( 'href' ) ), '#' ) && strlen( trim( $el->getAttribute( 'href' ) ) ) > 1 ) {
				$this->links['skip'] = true;
			}
		}
	}

	/**
	 * Page-wide accessibility problems: no language, repeated ids, no skip link.
	 *
	 * @param \DOMDocument $dom Document.
	 */
	private function document_problems( \DOMDocument $dom ): void {
		$html = $dom->getElementsByTagName( 'html' )->item( 0 );
		$lang = $html instanceof \DOMElement ? trim( $html->getAttribute( 'lang' ) . $html->getAttribute( 'xml:lang' ) ) : '';
		if ( $html instanceof \DOMElement && '' === $lang ) {
			$this->a11y[] = array( 'lang_missing', '', Source::TEMPLATE, '' );
		}
		foreach ( $this->ids as $id => $seen ) {
			if ( $seen[0] > 1 ) {
				$this->a11y[] = array( 'duplicate_id', mb_substr( (string) $id, 0, self::TEXT_LENGTH ), $seen[1], $seen[2] );
			}
		}
		if ( $this->links['links'] >= 8 && ! $this->links['skip'] ) {
			$this->a11y[] = array( 'no_skip_link', '', Source::TEMPLATE, '' );
		}
	}

	/**
	 * Facts about the written content: the newest year in a copyright notice. The notice is
	 * normally in the footer of the theme, so it is reported for the whole site.
	 *
	 * @param \DOMDocument $dom Document.
	 */
	private function content_facts( \DOMDocument $dom ): void {
		$xpath = new \DOMXPath( $dom );
		$nodes = $xpath->query( '//text()[not(ancestor::script) and not(ancestor::style) and not(ancestor::noscript)]' );
		$year  = 0;
		foreach ( false === $nodes ? array() : $nodes as $node ) {
			if ( 1 === preg_match( '/(?:©|\(c\)|copyright)\s*(?:©|\(c\))?\s*((?:19|20)\d{2})(?:\s*[-–—]\s*((?:19|20)\d{2}))?/iu', $node->nodeValue ?? '', $found ) ) {
				$year = max( $year, (int) ( $found[2] ?? $found[1] ), (int) $found[1] );
			}
		}
		if ( $year > 0 ) {
			$this->content[] = array( 'copyright_year', (string) $year, Source::TEMPLATE, '' );
		}
	}

	/**
	 * Adds each candidate of a srcset.
	 *
	 * @param string $srcset Attribute value.
	 * @param string $tag    Element name.
	 * @param string $text   Alt text.
	 */
	private function add_srcset( string $srcset, string $tag, string $text ): void {
		foreach ( explode( ',', $srcset ) as $candidate ) {
			$parts = preg_split( '/\s+/', trim( $candidate ) );
			if ( is_array( $parts ) && '' !== $parts[0] ) {
				$this->add( Reference::IMAGE, $parts[0], $tag, $text );
			}
		}
	}

	/**
	 * Adds one reference if the value is a web address.
	 *
	 * @param string $kind  Reference kind.
	 * @param string $value Attribute value.
	 * @param string $tag   Element name.
	 * @param string $text  Text or alt.
	 */
	private function add( string $kind, string $value, string $tag, string $text ): void {
		if ( '' === trim( $value ) ) {
			return;
		}
		$url = Url::resolve( $value, $this->base );
		if ( null === $url ) {
			return;
		}
		$source             = array() === $this->stack ? Source::template() : $this->stack[ count( $this->stack ) - 1 ];
		$this->references[] = new Reference( $kind, $url, $value, $tag, mb_substr( trim( $text ), 0, self::TEXT_LENGTH ), $source );
	}
}