<?php
/**
 * Accessibility-for-agents scanner.
 *
 * @package AgizoAI_Agentic
 */

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

/**
 * Accessibility scanner class.
 */
class AgizoAI_Agentic_Accessibility {

	/**
	 * Settings helper.
	 *
	 * @var AgizoAI_Agentic_Settings
	 */
	private $settings;

	/**
	 * Constructor.
	 *
	 * @param AgizoAI_Agentic_Settings $settings Settings helper.
	 */
	public function __construct( AgizoAI_Agentic_Settings $settings ) {
		$this->settings = $settings;
	}

	/**
	 * Run a lightweight public-content accessibility scan.
	 *
	 * @param int $limit Number of public items to scan.
	 * @return array<string,mixed>
	 */
	public function scan( $limit = 25 ) {
		$options = $this->settings->get_options();
		$limit   = max( 1, min( 50, absint( $limit ) ) );

		$post_types = isset( $options['include_post_types'] ) && is_array( $options['include_post_types'] ) ? $options['include_post_types'] : array( 'page', 'post' );
		$post_types = array_values( array_filter( array_map( 'sanitize_key', $post_types ) ) );

		if ( empty( $post_types ) ) {
			$post_types = array( 'page', 'post' );
		}

		$metrics = $this->empty_metrics();
		$issues  = array();
		$pages   = array();

		$query = new WP_Query(
			array(
				'post_type'           => $post_types,
				'post_status'         => 'publish',
				'posts_per_page'      => $limit,
				'orderby'             => 'date',
				'order'               => 'DESC',
				'ignore_sticky_posts' => true,
				'no_found_rows'       => true,
			)
		);

		if ( $query->have_posts() ) {
			foreach ( $query->posts as $post ) {
				$html = apply_filters( 'the_content', $post->post_content );
				$page = array(
					'id'    => absint( $post->ID ),
					'title' => get_the_title( $post ),
					'url'   => get_permalink( $post ),
					'type'  => get_post_type( $post ),
				);

				$page_result = $this->analyse_html( $html, $page );
				$pages[]     = array_merge(
					$page,
					array(
						'issues' => count( $page_result['issues'] ),
					)
				);

				foreach ( $page_result['metrics'] as $key => $value ) {
					if ( isset( $metrics[ $key ] ) ) {
						$metrics[ $key ] += absint( $value );
					}
				}

				$issues = array_merge( $issues, $page_result['issues'] );
			}
			wp_reset_postdata();
		}

		$metrics['pages_scanned']    = count( $pages );
		$metrics['language_defined'] = '' !== trim( (string) get_bloginfo( 'language' ) ) ? 1 : 0;

		if ( ! $metrics['language_defined'] ) {
			$issues[] = $this->issue(
				'language_missing',
				__( 'Page language is not defined', 'agizoai-agentic-browsing-optimizer' ),
				__( 'WordPress did not return a site language code.', 'agizoai-agentic-browsing-optimizer' ),
				__( 'Language metadata helps assistive technology and agents interpret page content correctly.', 'agizoai-agentic-browsing-optimizer' ),
				__( 'Set the site language in WordPress Settings > General.', 'agizoai-agentic-browsing-optimizer' ),
				'warning',
				array()
			);
		}

		$critical_count = count(
			array_filter(
				$issues,
				function ( $issue ) {
					return isset( $issue['severity'] ) && 'critical' === $issue['severity'];
				}
			)
		);

		$warning_count = count(
			array_filter(
				$issues,
				function ( $issue ) {
					return isset( $issue['severity'] ) && 'warning' === $issue['severity'];
				}
			)
		);

		return array(
			'enabled'        => ! empty( $options['accessibility_scanner_enabled'] ),
			'pages'          => $pages,
			'pages_scanned'  => absint( $metrics['pages_scanned'] ),
			'metrics'        => $metrics,
			'issues'         => $issues,
			'issue_count'    => count( $issues ),
			'critical_count' => $critical_count,
			'warning_count'  => $warning_count,
			'score'          => $this->score_from_metrics( $metrics, $issues ),
		);
	}

	/**
	 * Empty metrics array.
	 *
	 * @return array<string,int>
	 */
	private function empty_metrics() {
		return array(
			'pages_scanned'         => 0,
			'images_total'          => 0,
			'images_missing_alt'    => 0,
			'links_total'           => 0,
			'links_vague'           => 0,
			'links_empty'           => 0,
			'buttons_total'         => 0,
			'buttons_without_name'  => 0,
			'inputs_total'          => 0,
			'inputs_without_name'   => 0,
			'inputs_without_label'  => 0,
			'headings_total'        => 0,
			'heading_level_skips'   => 0,
			'h1_in_content'         => 0,
			'language_defined'      => 0,
		);
	}

	/**
	 * Analyse a rendered content HTML fragment.
	 *
	 * @param string              $html Rendered HTML.
	 * @param array<string,mixed> $page Page context.
	 * @return array<string,mixed>
	 */
	private function analyse_html( $html, array $page ) {
		if ( class_exists( 'DOMDocument' ) ) {
			return $this->analyse_with_dom( $html, $page );
		}

		return $this->analyse_with_regex( $html, $page );
	}

	/**
	 * Analyse HTML with DOMDocument.
	 *
	 * @param string              $html Rendered HTML.
	 * @param array<string,mixed> $page Page context.
	 * @return array<string,mixed>
	 */
	private function analyse_with_dom( $html, array $page ) {
		$metrics = $this->empty_metrics();
		$issues  = array();

		$dom = new DOMDocument();
		$previous = libxml_use_internal_errors( true );
		$encoded  = '<?xml encoding="utf-8" ?>' . (string) $html;
		$dom->loadHTML( $encoded, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD );
		libxml_clear_errors();
		libxml_use_internal_errors( $previous );
		$xpath = new DOMXPath( $dom );

		$images = $dom->getElementsByTagName( 'img' );
		$metrics['images_total'] = $images->length;
		foreach ( $images as $img ) {
			$alt = $img->hasAttribute( 'alt' ) ? trim( $img->getAttribute( 'alt' ) ) : '';
			if ( '' === $alt ) {
				$metrics['images_missing_alt']++;
				$issues[] = $this->issue(
					'image_alt_missing',
					__( 'Image is missing descriptive alt text', 'agizoai-agentic-browsing-optimizer' ),
					__( 'An image in public content has no useful alt text.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Agents and assistive technologies use alt text to understand visual content.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Add concise, descriptive alt text or mark decorative images intentionally.', 'agizoai-agentic-browsing-optimizer' ),
					'warning',
					$page
				);
			}
		}

		$links = $dom->getElementsByTagName( 'a' );
		$metrics['links_total'] = $links->length;
		foreach ( $links as $link ) {
			$name = $this->accessible_name_from_node( $link );
			$text = $this->normalise_text( $link->textContent );
			if ( '' === $name ) {
				$metrics['links_empty']++;
				$issues[] = $this->issue(
					'link_name_missing',
					__( 'Link has no accessible name', 'agizoai-agentic-browsing-optimizer' ),
					__( 'A link in public content has no readable text, title or aria-label.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Agents may not understand where the link goes or why it matters.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Add clear link text or an aria-label.', 'agizoai-agentic-browsing-optimizer' ),
					'critical',
					$page
				);
			} elseif ( $this->is_vague_link_text( $text ) ) {
				$metrics['links_vague']++;
				$issues[] = $this->issue(
					'link_text_vague',
					__( 'Link text is vague', 'agizoai-agentic-browsing-optimizer' ),
					__( 'A link uses generic text such as “click here” or “read more”.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Agents understand pages better when link text describes the destination.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Replace generic link text with descriptive anchor text.', 'agizoai-agentic-browsing-optimizer' ),
					'warning',
					$page
				);
			}
		}

		$buttons = $dom->getElementsByTagName( 'button' );
		$metrics['buttons_total'] += $buttons->length;
		foreach ( $buttons as $button ) {
			if ( '' === $this->accessible_name_from_node( $button ) ) {
				$metrics['buttons_without_name']++;
				$issues[] = $this->issue(
					'button_name_missing',
					__( 'Button has no accessible name', 'agizoai-agentic-browsing-optimizer' ),
					__( 'A button has no visible text, title or aria-label.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Agents use button names to decide which action to take.', 'agizoai-agentic-browsing-optimizer' ),
					__( 'Add visible button text or an aria-label.', 'agizoai-agentic-browsing-optimizer' ),
					'critical',
					$page
				);
			}
		}

		foreach ( array( 'input', 'select', 'textarea' ) as $tag ) {
			$nodes = $dom->getElementsByTagName( $tag );
			foreach ( $nodes as $node ) {
				$type = $node->hasAttribute( 'type' ) ? strtolower( $node->getAttribute( 'type' ) ) : '';
				if ( 'input' === $tag && in_array( $type, array( 'hidden', 'submit', 'button', 'reset', 'image' ), true ) ) {
					if ( in_array( $type, array( 'submit', 'button', 'reset', 'image' ), true ) ) {
						$metrics['buttons_total']++;
						if ( '' === $this->accessible_name_from_node( $node ) && ! $node->hasAttribute( 'value' ) ) {
							$metrics['buttons_without_name']++;
						}
					}
					continue;
				}

				$metrics['inputs_total']++;
				$name = $node->hasAttribute( 'name' ) ? trim( $node->getAttribute( 'name' ) ) : '';
				if ( '' === $name ) {
					$metrics['inputs_without_name']++;
					$issues[] = $this->issue(
						'input_name_missing',
						__( 'Form field is missing a name attribute', 'agizoai-agentic-browsing-optimizer' ),
						__( 'A public form field has no stable name attribute.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Agents and browsers need field names to map and submit form values correctly.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Set a field name in the form plugin or template markup.', 'agizoai-agentic-browsing-optimizer' ),
						'critical',
						$page
					);
				}

				if ( ! $this->field_has_label( $node, $xpath ) ) {
					$metrics['inputs_without_label']++;
					$issues[] = $this->issue(
						'input_label_missing',
						__( 'Form field has no accessible label', 'agizoai-agentic-browsing-optimizer' ),
						__( 'A public form field has no label, aria-label, title or useful placeholder.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Agents rely on field labels to understand what information is required.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Add a visible label or aria-label for this form field.', 'agizoai-agentic-browsing-optimizer' ),
						'critical',
						$page
					);
				}
			}
		}

		$heading_nodes = $xpath->query( '//*[self::h1 or self::h2 or self::h3 or self::h4 or self::h5 or self::h6]' );
		$previous_level = 0;
		if ( $heading_nodes ) {
			$metrics['headings_total'] = $heading_nodes->length;
			foreach ( $heading_nodes as $heading ) {
				$level = absint( substr( strtolower( $heading->nodeName ), 1 ) );
				if ( 1 === $level ) {
					$metrics['h1_in_content']++;
					$issues[] = $this->issue(
						'h1_in_content',
						__( 'Content contains an H1 heading', 'agizoai-agentic-browsing-optimizer' ),
						__( 'The rendered content includes an H1 heading inside the article body.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Most themes already use H1 for the page title; duplicate H1s can confuse hierarchy.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Use H2 for main article sections unless this template intentionally owns the page H1.', 'agizoai-agentic-browsing-optimizer' ),
						'warning',
						$page
					);
				}
				if ( $previous_level > 0 && $level > $previous_level + 1 ) {
					$metrics['heading_level_skips']++;
					$issues[] = $this->issue(
						'heading_level_skip',
						__( 'Heading level is skipped', 'agizoai-agentic-browsing-optimizer' ),
						__( 'A heading jumps by more than one level, for example H2 to H4.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'A clean heading outline helps agents understand page structure.', 'agizoai-agentic-browsing-optimizer' ),
						__( 'Use sequential heading levels where possible.', 'agizoai-agentic-browsing-optimizer' ),
						'warning',
						$page
					);
				}
				$previous_level = $level;
			}
		}

		return array(
			'metrics' => $metrics,
			'issues'  => $this->dedupe_and_limit_issues( $issues ),
		);
	}

	/**
	 * Basic regex fallback when DOMDocument is unavailable.
	 *
	 * @param string              $html Rendered HTML.
	 * @param array<string,mixed> $page Page context.
	 * @return array<string,mixed>
	 */
	private function analyse_with_regex( $html, array $page ) {
		$metrics = $this->empty_metrics();
		$issues  = array();
		$html    = (string) $html;

		if ( preg_match_all( '/<img\b[^>]*>/i', $html, $matches ) ) {
			$metrics['images_total'] = count( $matches[0] );
			foreach ( $matches[0] as $tag ) {
				if ( ! preg_match( '/\salt\s*=\s*(["\'])(.*?)\1/i', $tag, $alt ) || '' === trim( html_entity_decode( $alt[2] ) ) ) {
					$metrics['images_missing_alt']++;
					$issues[] = $this->issue( 'image_alt_missing', __( 'Image is missing descriptive alt text', 'agizoai-agentic-browsing-optimizer' ), __( 'An image in public content has no useful alt text.', 'agizoai-agentic-browsing-optimizer' ), __( 'Agents and assistive technologies use alt text to understand visual content.', 'agizoai-agentic-browsing-optimizer' ), __( 'Add concise, descriptive alt text or mark decorative images intentionally.', 'agizoai-agentic-browsing-optimizer' ), 'warning', $page );
				}
			}
		}

		if ( preg_match_all( '/<(input|select|textarea)\b[^>]*>/i', $html, $fields ) ) {
			foreach ( $fields[0] as $tag ) {
				if ( preg_match( '/\stype\s*=\s*(["\'])(hidden|submit|button|reset|image)\1/i', $tag ) ) {
					continue;
				}
				$metrics['inputs_total']++;
				if ( ! preg_match( '/\sname\s*=\s*(["\'])(.*?)\1/i', $tag, $name ) || '' === trim( $name[2] ) ) {
					$metrics['inputs_without_name']++;
					$issues[] = $this->issue( 'input_name_missing', __( 'Form field is missing a name attribute', 'agizoai-agentic-browsing-optimizer' ), __( 'A public form field has no stable name attribute.', 'agizoai-agentic-browsing-optimizer' ), __( 'Agents and browsers need field names to map and submit form values correctly.', 'agizoai-agentic-browsing-optimizer' ), __( 'Set a field name in the form plugin or template markup.', 'agizoai-agentic-browsing-optimizer' ), 'critical', $page );
				}
				if ( ! preg_match( '/\s(aria-label|title|placeholder)\s*=\s*(["\'])(.*?)\2/i', $tag ) ) {
					$metrics['inputs_without_label']++;
					$issues[] = $this->issue( 'input_label_missing', __( 'Form field has no accessible label', 'agizoai-agentic-browsing-optimizer' ), __( 'A public form field has no label, aria-label, title or useful placeholder.', 'agizoai-agentic-browsing-optimizer' ), __( 'Agents rely on field labels to understand what information is required.', 'agizoai-agentic-browsing-optimizer' ), __( 'Add a visible label or aria-label for this form field.', 'agizoai-agentic-browsing-optimizer' ), 'critical', $page );
				}
			}
		}

		return array(
			'metrics' => $metrics,
			'issues'  => $this->dedupe_and_limit_issues( $issues ),
		);
	}

	/**
	 * Get a practical accessible name from a node.
	 *
	 * @param DOMNode $node Node.
	 * @return string
	 */
	private function accessible_name_from_node( $node ) {
		if ( $node instanceof DOMElement ) {
			foreach ( array( 'aria-label', 'title', 'alt', 'value' ) as $attr ) {
				if ( $node->hasAttribute( $attr ) && '' !== trim( $node->getAttribute( $attr ) ) ) {
					return $this->normalise_text( $node->getAttribute( $attr ) );
				}
			}
		}

		$text = $this->normalise_text( $node->textContent );
		if ( '' !== $text ) {
			return $text;
		}

		if ( $node instanceof DOMElement ) {
			$imgs = $node->getElementsByTagName( 'img' );
			foreach ( $imgs as $img ) {
				if ( $img->hasAttribute( 'alt' ) && '' !== trim( $img->getAttribute( 'alt' ) ) ) {
					return $this->normalise_text( $img->getAttribute( 'alt' ) );
				}
			}
		}

		return '';
	}

	/**
	 * Check whether a form field has an accessible label.
	 *
	 * @param DOMElement $node Field node.
	 * @param DOMXPath   $xpath XPath helper.
	 * @return bool
	 */
	private function field_has_label( $node, $xpath ) {
		foreach ( array( 'aria-label', 'aria-labelledby', 'title', 'placeholder' ) as $attr ) {
			if ( $node->hasAttribute( $attr ) && '' !== trim( $node->getAttribute( $attr ) ) ) {
				return true;
			}
		}

		if ( $node->hasAttribute( 'id' ) && '' !== trim( $node->getAttribute( 'id' ) ) ) {
			$id     = $node->getAttribute( 'id' );
			$labels = $xpath->query( '//label[@for=' . $this->xpath_literal( $id ) . ']' );
			if ( $labels && $labels->length > 0 ) {
				return true;
			}
		}

		$parent = $node->parentNode;
		while ( $parent ) {
			if ( $parent instanceof DOMElement && 'label' === strtolower( $parent->tagName ) ) {
				return true;
			}
			$parent = $parent->parentNode;
		}

		return false;
	}

	/**
	 * Build an XPath string literal.
	 *
	 * @param string $value Value.
	 * @return string
	 */
	private function xpath_literal( $value ) {
		$value = (string) $value;
		if ( false === strpos( $value, '"' ) ) {
			return '"' . $value . '"';
		}
		if ( false === strpos( $value, "'" ) ) {
			return "'" . $value . "'";
		}

		$parts = explode( '"', $value );
		return 'concat("' . implode( '", \'"\', "', $parts ) . '")';
	}

	/**
	 * Return true when link text is too generic.
	 *
	 * @param string $text Link text.
	 * @return bool
	 */
	private function is_vague_link_text( $text ) {
		$text  = strtolower( $this->normalise_text( $text ) );
		$vague = array( 'click here', 'here', 'read more', 'more', 'learn more', 'view more', 'continue reading', 'details' );

		return in_array( $text, $vague, true );
	}

	/**
	 * Normalise whitespace in text.
	 *
	 * @param string $text Text.
	 * @return string
	 */
	private function normalise_text( $text ) {
		$text = wp_strip_all_tags( html_entity_decode( (string) $text, ENT_QUOTES, get_bloginfo( 'charset' ) ) );
		$text = preg_replace( '/\s+/', ' ', $text );

		return trim( (string) $text );
	}

	/**
	 * Dedupe and limit issues to keep dashboard readable.
	 *
	 * @param array<int,array<string,mixed>> $issues Issues.
	 * @return array<int,array<string,mixed>>
	 */
	private function dedupe_and_limit_issues( array $issues ) {
		$seen   = array();
		$result = array();

		foreach ( $issues as $issue ) {
			$key = $issue['type'] . '|' . ( isset( $issue['page_url'] ) ? $issue['page_url'] : '' );
			if ( isset( $seen[ $key ] ) && $seen[ $key ] >= 3 ) {
				continue;
			}
			if ( ! isset( $seen[ $key ] ) ) {
				$seen[ $key ] = 0;
			}
			$seen[ $key ]++;
			$result[] = $issue;
		}

		return array_slice( $result, 0, 80 );
	}

	/**
	 * Build one issue.
	 *
	 * @param string              $type Type.
	 * @param string              $title Title.
	 * @param string              $problem Problem.
	 * @param string              $why Why.
	 * @param string              $fix Fix.
	 * @param string              $severity Severity.
	 * @param array<string,mixed> $page Page context.
	 * @return array<string,mixed>
	 */
	private function issue( $type, $title, $problem, $why, $fix, $severity, array $page ) {
		return array(
			'type'       => sanitize_key( $type ),
			'title'      => $title,
			'problem'    => $problem,
			'why'        => $why,
			'fix'        => $fix,
			'severity'   => sanitize_key( $severity ),
			'page_id'    => isset( $page['id'] ) ? absint( $page['id'] ) : 0,
			'page_title' => isset( $page['title'] ) ? sanitize_text_field( $page['title'] ) : '',
			'page_url'   => isset( $page['url'] ) ? esc_url_raw( $page['url'] ) : '',
			'page_type'  => isset( $page['type'] ) ? sanitize_key( $page['type'] ) : '',
		);
	}

	/**
	 * Calculate a simple accessibility sub-score.
	 *
	 * @param array<string,int>              $metrics Metrics.
	 * @param array<int,array<string,mixed>> $issues Issues.
	 * @return int
	 */
	private function score_from_metrics( array $metrics, array $issues ) {
		if ( empty( $metrics['pages_scanned'] ) ) {
			return 0;
		}

		$score = 100;
		foreach ( $issues as $issue ) {
			$score -= isset( $issue['severity'] ) && 'critical' === $issue['severity'] ? 8 : 3;
		}

		return max( 0, min( 100, absint( $score ) ) );
	}
}