<?php
/**
 * A byte-preserving heading parser and numbering helpers.
 *
 * @package Tocnest
 */

defined( 'ABSPATH' ) || exit;

/** Parses headings without reserializing a document. */
class Tocnest_Parser {
	/**
	 * Request cache.
	 *
	 * @var array
	 */
	private static $cache = array();

	/**
	 * Anchors claimed by different posts in this request.
	 *
	 * @var array
	 */
	private static $claimed = array();

	/**
	 * Claim a unique generated element ID against parsed content IDs.
	 *
	 * @param string $candidate Preferred ID.
	 * @return string
	 */
	public static function reserve_id( $candidate ) {
		$id     = $candidate;
		$suffix = 2;
		while ( isset( self::$claimed[ $id ] ) ) {
			$id = $candidate . '-' . $suffix;
			++$suffix;
		}
		self::$claimed[ $id ] = true;
		return $id;
	}

	/**
	 * Read a quoted or unquoted attribute from one start tag.
	 *
	 * @param string $tag Start tag.
	 * @param string $name Attribute name.
	 * @return string|null
	 */
	public static function attribute( $tag, $name ) {
		if ( class_exists( 'WP_HTML_Tag_Processor' ) ) {
			$processor = new WP_HTML_Tag_Processor( $tag );
			if ( $processor->next_tag() ) {
				$value = $processor->get_attribute( $name );
				return null === $value ? null : ( true === $value ? '' : (string) $value );
			}
			return null;
		}
		preg_match_all( '/([^\s=<>`"\x27\/]+)(?:\s*=\s*(?:"([^"]*)"|\x27([^\x27]*)\x27|([^\s>]+)))?/', $tag, $attributes, PREG_SET_ORDER );
		foreach ( $attributes as $attribute ) {
			if ( strtolower( $attribute[1] ) === strtolower( $name ) ) {
				$value = isset( $attribute[4] ) ? $attribute[4] : ( isset( $attribute[3] ) ? $attribute[3] : ( isset( $attribute[2] ) ? $attribute[2] : '' ) );
				return html_entity_decode( $value, ENT_QUOTES | ENT_HTML5, 'UTF-8' );
			}
		}
		return null;
	}

	/**
	 * Set an ID without duplicating an empty or valueless authored attribute.
	 *
	 * @param string $opening Heading start tag.
	 * @param string $id Generated ID.
	 * @return string
	 */
	private static function with_id( $opening, $id ) {
		if ( class_exists( 'WP_HTML_Tag_Processor' ) ) {
			$processor = new WP_HTML_Tag_Processor( $opening );
			$processor->next_tag();
			$processor->set_attribute( 'id', $id );
			return $processor->get_updated_html();
		}
		preg_match_all( '/([^\s=<>`"\x27\/]+)(?:\s*=\s*(?:"[^"]*"|\x27[^\x27]*\x27|[^\s>]+))?/', $opening, $attributes, PREG_SET_ORDER | PREG_OFFSET_CAPTURE );
		foreach ( $attributes as $attribute ) {
			if ( 'id' === strtolower( $attribute[1][0] ) ) {
				return substr_replace( $opening, 'id="' . esc_attr( $id ) . '"', $attribute[0][1], strlen( $attribute[0][0] ) );
			}
		}
		return substr_replace( $opening, ' id="' . esc_attr( $id ) . '"', -1, 0 );
	}

	/**
	 * Build an anchor from heading text.
	 *
	 * @param string $text Plain text.
	 * @param int    $index Heading counter.
	 * @param array  $settings Settings.
	 * @return string
	 */
	public static function anchor( $text, $index, $settings ) {
		if ( 'counter' === $settings['anchor_format'] ) {
			return 'toc-' . $index;
		}
		$text = 'text_raw' === $settings['anchor_format'] ? $text : remove_accents( $text );
		$text = preg_replace( '/[^\p{L}\p{N}_\-\s]+/u', '', $text );
		$text = preg_replace( '/\s+/u', '-', trim( $text ) );
		if ( $settings['dashes'] ) {
			$text = str_replace( '_', '-', $text );
		}
		if ( $settings['lowercase'] ) {
			$text = function_exists( 'mb_strtolower' ) ? mb_strtolower( $text, 'UTF-8' ) : strtolower( $text );
		}
		return '' === $text ? 'toc-' . $index : $text;
	}

	/**
	 * Parse once for each post, content and anchor configuration.
	 *
	 * The tokenizer skips comments and raw/code elements before it
	 * matches headings. Only missing heading ID attributes are inserted.
	 * Existing IDs anywhere in the fragment are reserved before generating IDs.
	 *
	 * @param string $content Original HTML.
	 * @param int    $post_id Post ID.
	 * @param array  $settings Anchor options.
	 * @return array HTML and heading records.
	 */
	public static function parse( $content, $post_id, $settings ) {
		$key = $post_id . ':' . md5( $content . wp_json_encode( array_intersect_key( $settings, array_flip( array( 'anchor_format', 'lowercase', 'dashes' ) ) ) ) );
		if ( isset( self::$cache[ $key ] ) ) {
			return self::$cache[ $key ];
		}
		$tag = '~<!--[\s\S]*?(?:-->|\z)|<![^>]*>|</?[a-zA-Z][a-zA-Z0-9:-]*(?:\s+(?:[^>"\x27]|"[^"]*"|\x27[^\x27]*\x27)*)?\s*/?>~';
		preg_match_all( $tag, $content, $tokens, PREG_OFFSET_CAPTURE );
		$used = self::$claimed;
		foreach ( $tokens[0] as $token ) {
			$id = self::attribute( $token[0], 'id' );
			if ( null !== $id ) {
				$used[ $id ]          = true;
				self::$claimed[ $id ] = true;
			}
		}
		$matches       = array();
		$current       = null;
		$blocked       = '';
		$blocked_depth = 0;
		foreach ( $tokens[0] as $token ) {
			if ( ! preg_match( '~^<(/?)([a-zA-Z][a-zA-Z0-9:-]*)\b~', $token[0], $name ) ) {
				continue;
			}
			$closing = '/' === $name[1];
			$name    = strtolower( $name[2] );
			if ( $blocked ) {
				if ( $name === $blocked ) {
					$blocked_depth += $closing ? -1 : ( in_array( $blocked, array( 'script', 'style', 'textarea' ), true ) ? 0 : 1 );
					if ( 0 === $blocked_depth ) {
						$blocked = '';
					}
				}
				continue;
			}
			if ( ! $closing && null === $current && in_array( $name, array( 'pre', 'code', 'script', 'style', 'textarea', 'template', 'nav' ), true ) ) {
				$blocked       = $name;
				$blocked_depth = 1;
				continue;
			}
			if ( ! preg_match( '/^h([1-6])$/', $name, $level ) ) {
				continue;
			}
			if ( ! $closing ) {
				$current = array(
					'opening' => $token,
					'level'   => (int) $level[1],
					'body'    => $token[1] + strlen( $token[0] ),
				);
			} elseif ( null !== $current && $current['level'] === (int) $level[1] ) {
				$current['text'] = substr( $content, $current['body'], $token[1] - $current['body'] );
				$matches[]       = $current;
				$current         = null;
			}
		}
		$headings = array();
		$edits    = array();
		foreach ( $matches as $match ) {
			$opening = $match['opening'][0];
			$text    = trim( preg_replace( '/\s+/u', ' ', html_entity_decode( wp_strip_all_tags( $match['text'] ), ENT_QUOTES | ENT_HTML5, 'UTF-8' ) ) );
			$id      = self::attribute( $opening, 'id' );
			if ( null === $id || '' === $id ) {
				$id     = self::anchor( $text, count( $headings ) + 1, $settings );
				$id     = (string) apply_filters(
					'tocnest_anchor',
					$id,
					$text,
					array(
						'level' => $match['level'],
						'text'  => $text,
					)
				);
				$id     = '' === $id ? 'toc-' . ( count( $headings ) + 1 ) : $id;
				$base   = $id;
				$suffix = 2;
				while ( isset( $used[ $id ] ) ) {
					$id = $base . '-' . $suffix;
					++$suffix;
				}
				$updated = self::with_id( $opening, $id );
				$edits[] = array( $match['opening'][1], strlen( $opening ), $updated );
				$opening = $updated;
			}
			$used[ $id ]          = true;
			self::$claimed[ $id ] = true;
			$headings[]           = array(
				'id'      => $id,
				'text'    => $text,
				'level'   => $match['level'],
				'class'   => (string) self::attribute( $opening, 'class' ),
				'opening' => $opening,
			);
		}
		foreach ( array_reverse( $edits ) as $edit ) {
			$content = substr_replace( $content, $edit[2], $edit[0], $edit[1] );
		}
		self::$cache[ $key ] = array(
			'html'     => $content,
			'headings' => $headings,
		);
		return self::$cache[ $key ];
	}

	/**
	 * Select headings and calculate structural nesting depth.
	 *
	 * @param array $headings Parsed records.
	 * @param array $settings Settings.
	 * @return array
	 */
	public static function select( $headings, $settings ) {
		$result       = array();
		$stack        = array();
		$skip_levels  = preg_split( '/[^1-6]+/', $settings['skip_levels'], -1, PREG_SPLIT_NO_EMPTY );
		$skip_classes = preg_split( '/\s+/', trim( 'tocnest-skip ' . $settings['skip_class'] ) );
		$patterns     = preg_split( '/\r\n|\r|\n/', $settings['skip_text'], -1, PREG_SPLIT_NO_EMPTY );
		foreach ( $headings as $heading ) {
			if ( '' === $heading['text'] || ! in_array( $heading['level'], $settings['levels'], true ) || in_array( (string) $heading['level'], $skip_levels, true ) || array_intersect( $skip_classes, preg_split( '/\s+/', $heading['class'] ) ) ) {
				continue;
			}
			foreach ( $patterns as $pattern ) {
				if ( preg_match( '/^' . str_replace( '\*', '.*', preg_quote( trim( $pattern ), '/' ) ) . '$/iu', $heading['text'] ) ) {
					continue 2;
				}
			}
			while ( $stack && end( $stack ) >= $heading['level'] ) {
				array_pop( $stack );
			}
			$stack[] = $heading['level'];
			if ( count( $stack ) > $settings['depth'] ) {
				continue;
			}
			$heading['depth'] = $settings['hierarchical'] ? count( $stack ) : 1;
			$result[]         = $heading;
		}
		return $result;
	}

	/**
	 * Format an integer in Roman notation.
	 *
	 * @param int $number Positive integer.
	 * @return string
	 */
	public static function roman( $number ) {
		$output = '';
		foreach ( array(
			1000 => 'M',
			900  => 'CM',
			500  => 'D',
			400  => 'CD',
			100  => 'C',
			90   => 'XC',
			50   => 'L',
			40   => 'XL',
			10   => 'X',
			9    => 'IX',
			5    => 'V',
			4    => 'IV',
			1    => 'I',
		) as $value => $symbol ) {
			while ( $number >= $value ) {
				$output .= $symbol;
				$number -= $value;
			}
		}
		return $output;
	}

	/**
	 * Number one row from its hierarchy counters.
	 *
	 * @param array  $counters Hierarchy counters.
	 * @param int    $index Flat index.
	 * @param string $style Numbering style.
	 * @param string $suffix Suffix enum.
	 * @return string
	 */
	public static function number( $counters, $index, $style, $suffix ) {
		if ( 'none' === $style ) {
			return '';
		}
		$parts = false !== strpos( $style, 'nested' ) ? $counters : array( $index );
		if ( 0 === strpos( $style, 'roman' ) ) {
			$parts = array_map( array( __CLASS__, 'roman' ), $parts );
		}
		return implode( '.', $parts ) . ( 'dot' === $suffix ? '.' : ( 'roundbracket' === $suffix ? ')' : '' ) );
	}
}