igny8/igny8-wp-integration-plugin/data/link-graph.php

<?php
/**
 * Link Graph Collection
 *
 * Extracts WordPress internal link graph for IGNY8 Linker module
 *
 * @package Igny8Bridge
 */

// Prevent direct access
if (!defined('ABSPATH')) {
    exit;
}

/**
 * Extract internal links from post content
 *
 * @param int $post_id Post ID
 * @return array Array of link objects with source_url, target_url, anchor
 */
function igny8_extract_post_links($post_id) {
    $post = get_post($post_id);
    if (!$post) {
        return array();
    }
    
    $content = $post->post_content;
    $source_url = get_permalink($post_id);
    $site_url = get_site_url();
    $links = array();
    
    // Match all anchor tags with href attributes
    preg_match_all('/<a[^>]+href=["\']([^"\']+)["\'][^>]*>(.*?)<\/a>/is', $content, $matches, PREG_SET_ORDER);
    
    foreach ($matches as $match) {
        $href = $match[1];
        $anchor = strip_tags($match[2]);
        
        // Skip empty anchors
        if (empty(trim($anchor))) {
            continue;
        }
        
        // Only process internal links
        if (strpos($href, $site_url) === 0 || strpos($href, '/') === 0) {
            // Convert relative URLs to absolute
            if (strpos($href, '/') === 0 && strpos($href, '//') !== 0) {
                $href = $site_url . $href;
            }
            
            // Normalize URL (remove trailing slash, fragments, query params for matching)
            $target_url = rtrim($href, '/');
            
            // Skip if source and target are the same
            if ($source_url === $target_url) {
                continue;
            }
            
            $links[] = array(
                'source_url' => $source_url,
                'target_url' => $target_url,
                'anchor' => trim($anchor),
                'post_id' => $post_id
            );
        }
    }
    
    return $links;
}

/**
 * Extract link graph from all posts
 *
 * @param array $post_ids Optional array of post IDs to process. If empty, processes all enabled posts.
 * @return array Link graph array
 */
function igny8_extract_link_graph($post_ids = array()) {
    // Skip if connection is disabled
    if (!igny8_is_connection_enabled()) {
        return array();
    }
    
    if (function_exists('igny8_is_module_enabled') && !igny8_is_module_enabled('linker')) {
        return array();
    }
    
    $enabled_post_types = igny8_get_enabled_post_types();
    
    if (empty($post_ids)) {
        // Get all published posts of enabled types
        $query_args = array(
            'post_type' => $enabled_post_types,
            'post_status' => 'publish',
            'posts_per_page' => -1,
            'fields' => 'ids',
            'suppress_filters' => true
        );
        
        $post_ids = get_posts($query_args);
    }
    
    $link_graph = array();
    $processed = 0;
    
    foreach ($post_ids as $post_id) {
        $links = igny8_extract_post_links($post_id);
        
        if (!empty($links)) {
            $link_graph = array_merge($link_graph, $links);
        }
        
        $processed++;
        
        // Limit processing to prevent timeout (can be increased or made configurable)
        if ($processed >= 1000) {
            break;
        }
    }
    
    return $link_graph;
}

/**
 * Send link graph to IGNY8 Linker module
 *
 * @param int $site_id IGNY8 site ID
 * @param array $link_graph Link graph array (optional, will extract if not provided)
 * @return array|false Response data or false on failure
 */
function igny8_send_link_graph_to_igny8($site_id, $link_graph = null) {
    // Skip if connection is disabled
    if (!igny8_is_connection_enabled()) {
        return false;
    }
    
    if (function_exists('igny8_is_module_enabled') && !igny8_is_module_enabled('linker')) {
        return false;
    }
    
    $api = new Igny8API();
    
    if (!$api->is_authenticated()) {
        return false;
    }
    
    // Extract link graph if not provided
    if ($link_graph === null) {
        $link_graph = igny8_extract_link_graph();
    }
    
    if (empty($link_graph)) {
        return array('success' => true, 'message' => 'No links found', 'links_count' => 0);
    }
    
    // Send in batches (max 500 links per batch)
    $batch_size = 500;
    $batches = array_chunk($link_graph, $batch_size);
    $total_sent = 0;
    $errors = array();
    
    foreach ($batches as $batch) {
        $response = $api->post("/linker/link-map/", array(
            'site_id' => $site_id,
            'links' => $batch,
            'total_links' => count($link_graph),
            'batch_number' => count($batches) > 1 ? (count($batches) - count($batches) + array_search($batch, $batches) + 1) : 1,
            'total_batches' => count($batches)
        ));
        
        if ($response['success']) {
            $total_sent += count($batch);
        } else {
            $errors[] = $response['error'] ?? 'Unknown error';
        }
    }
    
    if ($total_sent > 0) {
        update_option('igny8_last_link_graph_sync', current_time('timestamp'));
        update_option('igny8_last_link_graph_count', $total_sent);
        
        return array(
            'success' => true,
            'links_sent' => $total_sent,
            'total_links' => count($link_graph),
            'batches' => count($batches),
            'errors' => $errors
        );
    }
    
    return false;
}