<?php
/**
 * Parses attributes out of an existing <a> tag.
 *
 * @package Postlinkly
 * @since   1.0.0
 */

declare(strict_types=1);

namespace Postlinkly\Link;

/**
 * Parses attributes out of an existing `<a>` tag.
 *
 * Used during link removal and update operations to inspect what's
 * already in the content. The parser is intentionally tolerant — it
 * accepts both single- and double-quoted attribute values, ignores
 * malformed input, and never throws.
 *
 * @since 1.0.0
 */
final class LinkAttributeParser {

	/**
	 * Parses the attribute list out of an `<a>` opening tag.
	 *
	 * Input may be either a complete tag (`<a href="x">y</a>`) or just
	 * the opening tag (`<a href="x">`). Anything that isn't an `<a>`
	 * tag returns an empty array.
	 *
	 * @since 1.0.0
	 *
	 * @param string $tag The tag markup.
	 *
	 * @return array<string, string> Map of lower-cased attribute name to raw value.
	 */
	public function parse( string $tag ): array {
		// Capture the opening tag content between `<a` and the closing `>`.
		$matched = preg_match( '/<a\b([^>]*)>/i', $tag, $m );
		if ( 1 !== $matched || ! isset( $m[1] ) ) {
			return array();
		}
		$inside = $m[1];

		$out = array();
		// Match name="value" or name='value' attributes.
		if ( preg_match_all( '/([a-zA-Z][a-zA-Z0-9_:-]*)\s*=\s*("([^"]*)"|\'([^\']*)\')/', $inside, $matches, PREG_SET_ORDER ) ) {
			foreach ( $matches as $match ) {
				$name         = strtolower( $match[1] );
				$value        = isset( $match[3] ) && '' !== $match[3] ? $match[3] : ( $match[4] ?? '' );
				$out[ $name ] = $value;
			}
		}

		return $out;
	}

	/**
	 * Returns whether a tag was created or modified by Postlinkly.
	 *
	 * Detected via the {@see LinkAttributeBuilder::PLUGIN_MARKER_ATTRIBUTE}
	 * attribute. Useful when the link-removal flow needs to skip links the
	 * user added by hand.
	 *
	 * @since 1.0.0
	 *
	 * @param string $tag The tag markup.
	 *
	 * @return bool
	 */
	public function is_plugin_link( string $tag ): bool {
		$attrs = $this->parse( $tag );
		return isset( $attrs[ LinkAttributeBuilder::PLUGIN_MARKER_ATTRIBUTE ] );
	}

	/**
	 * Returns the href value from a tag, or empty string when missing.
	 *
	 * @since 1.0.0
	 *
	 * @param string $tag The tag markup.
	 *
	 * @return string
	 */
	public function get_href( string $tag ): string {
		$attrs = $this->parse( $tag );
		return $attrs['href'] ?? '';
	}

	/**
	 * Returns the inner text of an `<a>...</a>` tag, or empty string on failure.
	 *
	 * Does not decode HTML entities — callers can apply `html_entity_decode`
	 * themselves when they need plain text.
	 *
	 * @since 1.0.0
	 *
	 * @param string $tag Complete `<a>...</a>` markup.
	 *
	 * @return string
	 */
	public function get_inner_text( string $tag ): string {
		if ( 1 !== preg_match( '/<a\b[^>]*>(.*?)<\/a>/is', $tag, $m ) ) {
			return '';
		}
		return $m[1];
	}
}