<?php
/**
 * The main link building engine (100% Hardcoded HTML Mode).
 * Phase 4: Now reads from wp_ailm_link_rules instead of _ailm_frozen_map.
 */

if (!defined('ABSPATH')) {
	exit;
}

class AILM_Link_Builder
{

	public function __construct()
	{
	}

	/**
	 * Physically injects links from active rules into the post_content.
	 *
	 * Reads injection targets from wp_ailm_link_rules WHERE donor_id = X AND status = 'active'.
	 * After successful injection, updates rule status to 'active' (re-confirmed).
	 * If injection failed (keyword not found), sets status to 'failed'.
	 *
	 * @param int   $post_id         The ID of the post to update.
	 * @param array $links_to_inject Optional. Legacy array of links. If empty, reads from rules table.
	 * @return bool True if content was modified and saved, false otherwise.
	 */
	public function inject_physical_links($post_id, $links_to_inject = [])
	{
		$post = get_post($post_id);
		if (!$post || empty($post->post_content)) {
			if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
				// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
				error_log('AILM inject_physical_links: abort - no post or empty content. post_id=' . $post_id);
			}
			return false;
		}

		// Phase 4: If no explicit links provided, read from the unified rules table.
		if (empty($links_to_inject)) {
			if (!class_exists('AILM_Link_Repository')) {
				require_once plugin_dir_path(__FILE__) . 'class-ailm-link-repository.php';
			}
			$rules = AILM_Link_Repository::get_rules_for_donor($post_id);
			if (empty($rules)) {
				return false;
			}

			$links_to_inject = array();
			foreach ($rules as $rule) {
				// Phase 2.1: Skip rules whose link is already present in the
				// donor's HTML. These were either injected previously, imported
				// from pre-existing manual links, or inserted by AI/Gutenberg.
				// Re-injecting them would create duplicates.
				//
				// Backward compat: on pre-2.1 DBs the `in_html` column does not
				// exist, so the property is null → (int)null === 0 → the rule
				// is injected exactly as before.
				if (isset($rule->in_html) && (int) $rule->in_html === 1) {
					continue;
				}

				$links_to_inject[] = array(
					'word' => $rule->anchor,
					'url' => $rule->target_url,
					'origin' => $rule->source,
					'occurrence' => (int) $rule->occurrence,
					'rule_id' => (int) $rule->id,
				);
			}
		} else {
			// Legacy: wrap raw arrays with a dummy rule_id for consistent tracking.
			foreach ($links_to_inject as $idx => $item) {
				if (!isset($item['rule_id'])) {
					$links_to_inject[$idx]['rule_id'] = 0;
				}
			}
		}

		if (empty($links_to_inject)) {
			if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
				// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
				error_log('AILM inject_physical_links: abort - no links to inject. post_id=' . $post_id);
			}
			return false;
		}

		if (!class_exists('AILM_DOM_Helper')) {
			require_once plugin_dir_path(__FILE__) . 'class-ailm-dom-helper.php';
		}

		$skip_chars = get_option('ailm_skip_chars', 100);
		$global_limit = get_option('ailm_global_limit', 5);
		$target_limit = get_option('ailm_target_limit', 5);
		$gap = get_option('ailm_min_gap', 200);

		$global_limit = $this->adjust_global_limit_for_manual_links($post, $global_limit);
		if ($global_limit <= 0) {
			if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
				// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
				error_log('AILM inject_physical_links: abort - global_limit <= 0 after manual link adjustment. post_id=' . $post_id);
			}
			return false;
		}

		$links_to_inject = array_slice($links_to_inject, 0, $global_limit);

		// Sort links by length (longest first) to prevent partial word overwriting.
		usort($links_to_inject, function ($a, $b) {
			return mb_strlen($b['word'], 'UTF-8') <=> mb_strlen($a['word'], 'UTF-8');
		});
		$links_to_inject = array_reverse($links_to_inject);

		$is_elementor = class_exists('AILM_Elementor_Handler') && AILM_Elementor_Handler::is_elementor_post($post_id);

		$links_inserted = 0;
		$occupied_ranges = array();

		if (!$is_elementor) {
			$dom_helper = new AILM_DOM_Helper($post->post_content);
			$manual_link_positions = $dom_helper->get_manual_links_positions();
			foreach ($manual_link_positions as $pos) {
				$occupied_ranges[] = array('start' => max(0, $pos - $gap), 'end' => $pos + $gap);
			}
		}

		$global_modified = false;
		$global_char_offset = 0;

		// Track which rules succeeded and which failed.
		$injected_rule_ids = array();
		$failed_rule_ids = array();

		$process_html = function ($html) use (&$global_modified, &$links_inserted, &$occupied_ranges, &$global_char_offset, &$injected_rule_ids, &$failed_rule_ids, $post_id, $global_limit, $target_limit, $skip_chars, $gap, $links_to_inject) {
			if ($links_inserted >= $global_limit) {
				return $html;
			}

			$dom_helper = new AILM_DOM_Helper($html);
			$virtual_text = $dom_helper->get_virtual_text();
			$widget_length = mb_strlen($virtual_text, 'UTF-8');
			if ($widget_length === 0) {
				return $html;
			}

			$local_modified = false;

			foreach ($links_to_inject as $item) {
				if ($links_inserted >= $global_limit) {
					break;
				}

				$kw = $item['word'];
				$url = $item['url'];
				$origin = isset($item['origin']) ? $item['origin'] : 'ki';
				$occurrence = isset($item['occurrence']) ? (int) $item['occurrence'] : 1;
				$rule_id = isset($item['rule_id']) ? (int) $item['rule_id'] : 0;

				$target_id = url_to_postid($url);
				if ($target_id && $target_id === $post_id) {
					if ($rule_id > 0) {
						$failed_rule_ids[] = $rule_id;
					}
					continue;
				}

				if ($target_id) {
					if (get_post_status($target_id) !== 'publish') {
						if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
							// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
							error_log('AILM inject_physical_links: target post ' . $target_id . ' is not published. rule_id=' . $rule_id);
						}
						if ($rule_id > 0) {
							$failed_rule_ids[] = $rule_id;
						}
						continue;
					}
					if (class_exists('AILM_Link_Repository')) {
						$inbound = AILM_Link_Repository::get_active_inbound_rule_count($target_id);
					} else {
						$inbound = (int) get_post_meta($target_id, '_ailm_inbound_count', true);
					}
					if ($inbound >= $target_limit) {
						if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
							// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
							error_log('AILM inject_physical_links: target post ' . $target_id . ' has reached inbound limit (' . $inbound . '/' . $target_limit . '). rule_id=' . $rule_id);
						}
						if ($rule_id > 0) {
							$failed_rule_ids[] = $rule_id;
						}
						continue;
					}
				}

				$escaped_parts = array_map(
					function ($p) {
						return str_replace("'", "['''`´]", preg_quote($p, '/'));
					},
					preg_split('/\s+/u', trim($kw), -1, PREG_SPLIT_NO_EMPTY)
				);

				$fuzzy_kw = implode('[\p{Z}\s\p{P}\p{S}]+', $escaped_parts);
				$pattern = '/(?<![\p{L}\p{N}])(' . $fuzzy_kw . ')(?![\p{L}\p{N}])/iu';

				if (preg_match_all($pattern, $virtual_text, $matches, PREG_OFFSET_CAPTURE)) {
					$match_injected = false;
					$current_match = 0;

					foreach ($matches[0] as $match) {
						if ($links_inserted >= $global_limit) {
							break;
						}

						$current_match++;
						if ($current_match !== $occurrence) {
							continue;
						}

						$found_word = $match[0];
						$byte_offset = $match[1];

						$char_offset = mb_strlen(substr($virtual_text, 0, $byte_offset), 'UTF-8');
						$absolute_offset = $global_char_offset + $char_offset;
						$char_length = mb_strlen($found_word, 'UTF-8');

						if ($absolute_offset < $skip_chars) {
							continue;
						}

						$overlap = false;
						foreach ($occupied_ranges as $range) {
							if ($absolute_offset < $range['end'] && ($absolute_offset + $char_length) > $range['start']) {
								$overlap = true;
								break;
							}
						}

						if (!$overlap) {
							$dom_helper->queue_link($char_offset, $char_length, $url, $found_word, $origin);

							$occupied_ranges[] = array(
								'start' => max(0, $absolute_offset - $gap),
								'end' => $absolute_offset + $char_length + $gap,
							);

							$links_inserted++;
							$local_modified = true;
							$global_modified = true;
							$match_injected = true;

							if ($target_id) {
								update_post_meta($target_id, '_ailm_inbound_count', $inbound + 1);
							}

							if ($rule_id > 0) {
								$injected_rule_ids[] = $rule_id;
							}

							break;
						}
					}

					if (!$match_injected && $rule_id > 0) {
						$failed_rule_ids[] = $rule_id;
					}
				} else {
					// Keyword not found at all.
					if ($rule_id > 0) {
						$failed_rule_ids[] = $rule_id;
					}
				}
			}

			$global_char_offset += $widget_length;

			if ($local_modified) {
				$dom_helper->apply_links();
				return $dom_helper->get_html();
			}

			return $html;
		};

		if ($is_elementor) {
			$elementor_ok = AILM_Elementor_Handler::process_content($post_id, $process_html);
			if (!$elementor_ok) {
				// Elementor path failed - safe to reset state and fallback to raw post_content.
				$global_char_offset = 0;
				$occupied_ranges = array();
				$links_inserted = 0;
				$new_content = $process_html($post->post_content);
				if ($global_modified) {
					wp_update_post(
						array(
							'ID' => $post_id,
							'post_content' => $new_content,
						)
					);
				}
			}
		} else {
			$new_content = $process_html($post->post_content);
			if ($global_modified) {
				wp_update_post(
					array(
						'ID' => $post_id,
						'post_content' => $new_content,
					)
				);
			}
		}

		// Phase 4: Update rule statuses after injection attempt.
		if (class_exists('AILM_Link_Repository')) {
			$injected_rule_ids = array_unique($injected_rule_ids);
			$failed_rule_ids = array_unique($failed_rule_ids);

			foreach ($injected_rule_ids as $rid) {
				AILM_Link_Repository::update_rule_status($rid, 'active');
				// Phase 2.1: Mark the rule as already present in HTML so the
				// next inject_physical_links() call skips it (no duplicates).
				// mark_in_html() is a no-op on pre-2.1 DBs.
				AILM_Link_Repository::mark_in_html($rid, true);
			}
			foreach ($failed_rule_ids as $rid) {
				AILM_Link_Repository::update_rule_status($rid, 'failed');
			}
		}

		if ($global_modified) {
			if (class_exists('AILM_Scout_Scanner')) {
				AILM_Scout_Scanner::process_post($post_id);
			}
			return true;
		}

		if (defined('WP_DEBUG') && WP_DEBUG && defined('WP_DEBUG_LOG') && WP_DEBUG_LOG) {
			// phpcs:ignore WordPress.PHP.DevelopmentFunctions.error_log_error_log
			error_log('AILM inject_physical_links: no links were injected. post_id=' . $post_id . ' rules_count=' . count($links_to_inject));
		}
		return false;
	}

	/**
	 * Checks if a phrase can be injected (limits, exclusions, etc).
	 *
	 * Phase 4: Reads active rules from wp_ailm_link_rules instead of _ailm_frozen_map.
	 *
	 * @param int    $post_id Post ID.
	 * @param string $phrase  Phrase to check.
	 * @return bool True if the phrase can be injected.
	 */
	public function can_inject_phrase($post_id, $phrase)
	{
		$post = get_post($post_id);
		if (!$post || empty($post->post_content)) {
			return false;
		}

		if (!class_exists('AILM_DOM_Helper')) {
			require_once plugin_dir_path(__FILE__) . 'class-ailm-dom-helper.php';
		}

		$skip_chars = get_option('ailm_skip_chars', 100);
		$gap = get_option('ailm_min_gap', 200);
		$global_limit = get_option('ailm_global_limit', 5);
		$global_limit = $this->adjust_global_limit_for_manual_links($post, $global_limit);
		if ($global_limit <= 0) {
			return false;
		}

		$dom_helper = new AILM_DOM_Helper($post->post_content);
		$virtual_text = $dom_helper->get_virtual_text();
		$manual_link_positions = $dom_helper->get_manual_links_positions();

		$occupied_ranges = array();
		foreach ($manual_link_positions as $pos) {
			$occupied_ranges[] = array('start' => max(0, $pos - $gap), 'end' => $pos + $gap);
		}

		// Phase 4: Check active rules from the unified rules table.
		if (class_exists('AILM_Link_Repository')) {
			$active_rules = AILM_Link_Repository::get_rules_for_donor($post_id);
			foreach ($active_rules as $rule) {
				$item_occ = (int) $rule->occurrence;

				$escaped_parts = array_map(
					function ($p) {
						return str_replace("'", "['''`´]", preg_quote($p, '/'));
					},
					preg_split('/\s+/u', trim($rule->anchor), -1, PREG_SPLIT_NO_EMPTY)
				);
				$fuzzy_kw = implode('[\p{Z}\s\p{P}\p{S}]+', $escaped_parts);
				$pattern = '/(?<![\p{L}\p{N}])(' . $fuzzy_kw . ')(?![\p{L}\p{N}])/iu';

				if (preg_match_all($pattern, $virtual_text, $matches, PREG_OFFSET_CAPTURE)) {
					$current_match_idx = 0;
					foreach ($matches[0] as $match) {
						$current_match_idx++;
						if ($current_match_idx === $item_occ) {
							$char_offset = mb_strlen(substr($virtual_text, 0, $match[1]), 'UTF-8');
							$char_length = mb_strlen($match[0], 'UTF-8');
							$occupied_ranges[] = array(
								'start' => max(0, $char_offset - $gap),
								'end' => $char_offset + $char_length + $gap,
							);
							break;
						}
					}
				}
			}
		}

		$escaped_parts = array_map(
			function ($p) {
				return str_replace("'", "['''`´]", preg_quote($p, '/'));
			},
			preg_split('/\s+/u', trim($phrase), -1, PREG_SPLIT_NO_EMPTY)
		);
		$fuzzy_kw = implode('[\p{Z}\s\p{P}\p{S}]+', $escaped_parts);
		$pattern = '/(?<![\p{L}\p{N}])(' . $fuzzy_kw . ')(?![\p{L}\p{N}])/iu';

		if (preg_match_all($pattern, $virtual_text, $matches, PREG_OFFSET_CAPTURE)) {
			foreach ($matches[0] as $match) {
				$char_offset = mb_strlen(substr($virtual_text, 0, $match[1]), 'UTF-8');
				$char_length = mb_strlen($match[0], 'UTF-8');

				if ($char_offset < $skip_chars) {
					continue;
				}

				$overlap = false;
				foreach ($occupied_ranges as $range) {
					if ($char_offset < $range['end'] && ($char_offset + $char_length) > $range['start']) {
						$overlap = true;
						break;
					}
				}

				if (!$overlap) {
					return true;
				}
			}
		}
		return false;
	}

	/**
	 * Adjusts the global link limit by subtracting manual links already present.
	 *
	 * @param WP_Post $post         Post object.
	 * @param int     $global_limit Base global limit.
	 * @return int Adjusted limit.
	 */
	private function adjust_global_limit_for_manual_links($post, $global_limit)
	{
		if (!get_option('ailm_count_manual_links', 0) || empty($post)) {
			return $global_limit;
		}
		$home_url = untrailingslashit(home_url());
		$manual_internal_count = 0;
		$raw_content = $post->post_content;

		if (preg_match_all('/<a\s+[^>]*href=[\'"]([^\'"\s]+)[\'"][^>]*>/i', $raw_content, $matches)) {
			foreach ($matches[1] as $href) {
				if (empty($href)) {
					continue;
				}
				if (strpos($href, $home_url) === 0 || (strpos($href, '/') === 0 && strpos($href, '//') !== 0)) {
					$manual_internal_count++;
				}
			}
		}
		return max(0, $global_limit - $manual_internal_count);
	}
}