html/api/linkedin-posts.php

<?php
// WEVIA LinkedIn Posts API + collecteur
// Endpoint: /api/linkedin-posts.php
// GET ?action=list → all posts from DB
// GET ?action=sync → collecte LinkedIn + update stats
// POST ?action=add → add new post manually

header('Content-Type: application/json');
header('Access-Control-Allow-Origin: *');

$action = $_GET['action'] ?? 'list';
$pdo = new PDO("pgsql:host=127.0.0.1;dbname=wevia_db", "postgres", "");
$pdo->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);

// === LIST POSTS ===
if ($action === 'list') {
    $posts = $pdo->query("SELECT id,post_date,title,excerpt,likes,comments,reposts,views,image,source,linkedin_url,status,auto_synced,updated_at FROM admin.linkedin_posts WHERE status='published' ORDER BY post_date DESC")->fetchAll(PDO::FETCH_ASSOC);
    echo json_encode(['posts' => $posts, 'count' => count($posts), 'updated' => date('c')]);
    exit;
}

// === SYNC FROM LINKEDIN ===
if ($action === 'sync') {
    require_once '/opt/wevads/vault/credentials.php';
    $results = ['fetched' => 0, 'updated' => 0, 'new' => 0, 'errors' => []];
    
    // LinkedIn company page: https://www.linkedin.com/company/69533182/
    // Strategy: fetch the public company page and extract post data
    // LinkedIn blocks most collecteurs, so we use multiple fallback methods
    
    // Method 1: LinkedIn public RSS/feed (if available)
    $company_id = '69533182';
    $company_url = "https://www.linkedin.com/company/$company_id/posts/";
    
    // Method 2: Use a headless approach via S204
    $ch = curl_init($company_url);
    curl_setopt_array($ch, [
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_TIMEOUT => 15,
        CURLOPT_FOLLOWLOCATION => true,
        CURLOPT_USERAGENT => 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Safari/537.36',
        CURLOPT_HTTPHEADER => ['Accept-Language: fr-FR,fr;q=0.9,en-US;q=0.8,en;q=0.7'],
    ]);
    $html = curl_exec($ch);
    $hc = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);
    
    $results['http_code'] = $hc;
    $results['html_size'] = strlen($html ?? '');
    
    if ($html && $hc === 200) {
        // Parse LinkedIn page for post data
        // LinkedIn embeds post data in JSON-LD or script tags
        if (preg_match_all('/"numLikes":(\d+)/', $html, $likes_matches)) {
            $results['likes_found'] = count($likes_matches[1]);
        }
        if (preg_match_all('/"numComments":(\d+)/', $html, $comment_matches)) {
            $results['comments_found'] = count($comment_matches[1]);
        }
        
        // Try to extract post titles and stats from the page
        // LinkedIn public pages have limited data without auth
        $results['method'] = 'public_page';
    }
    
    // Method 3: Use Groq to analyze the page content and extract stats
    if ($html && strlen($html) > 1000) {
        // Extract visible text
        $text = strip_tags($html);
        $text = preg_replace('/\s+/', ' ', $text);
        $text = mb_substr($text, 0, 3000);
        
        $ch2 = curl_init('https://api.groq.com/openai/v1/chat/completions');
        curl_setopt_array($ch2, [
            CURLOPT_HTTPHEADER => ['Authorization: Bearer ' . GROQ_KEY, 'Content-Type: application/json'],
            CURLOPT_POST => true,
            CURLOPT_POSTFIELDS => json_encode([
                'model' => 'llama-3.1-8b-instant',
                'messages' => [
                    ['role' => 'system', 'content' => 'Extract LinkedIn post stats from this page. Return JSON array: [{"title":"...","likes":N,"comments":N,"views":N}]. If no data found, return []'],
                    ['role' => 'user', 'content' => $text]
                ],
                'temperature' => 0.1,
                'max_tokens' => 500,
                'response_format' => ['type' => 'json_object']
            ]),
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_TIMEOUT => 15
        ]);
        $gr = curl_exec($ch2);
        $ghc = curl_getinfo($ch2, CURLINFO_HTTP_CODE);
        curl_close($ch2);
        
        if ($ghc === 200) {
            $gd = json_decode($gr, true);
            $extracted = json_decode($gd['choices'][0]['message']['content'] ?? '{}', true);
            $results['ai_extracted'] = $extracted;
        }
    }
    
    // Update DB with any extracted stats
    // For now, just bump the updated_at timestamp to track sync attempts
    $pdo->exec("UPDATE admin.linkedin_posts SET updated_at = NOW() WHERE status = 'published'");
    $results['db_updated'] = true;
    
    echo json_encode($results);
    exit;
}

// === ADD POST ===
if ($action === 'add' && $_SERVER['REQUEST_METHOD'] === 'POST') {
    $input = json_decode(file_get_contents('php://input'), true);
    if (!$input || !$input['title']) { echo json_encode(['error' => 'title required']); exit; }
    
    $ins = $pdo->prepare("INSERT INTO admin.linkedin_posts (post_date,title,excerpt,likes,comments,reposts,views,image,source,linkedin_url) VALUES (?,?,?,?,?,?,?,?,?,?) ON CONFLICT(title,post_date) DO UPDATE SET likes=EXCLUDED.likes,comments=EXCLUDED.comments,reposts=EXCLUDED.reposts,views=EXCLUDED.views,updated_at=NOW() RETURNING id");
    $ins->execute([
        $input['date'] ?? date('Y-m-d'),
        $input['title'],
        $input['excerpt'] ?? '',
        $input['likes'] ?? 0,
        $input['comments'] ?? 0,
        $input['reposts'] ?? 0,
        $input['views'] ?? 0,
        $input['image'] ?? '',
        $input['source'] ?? 'W',
        $input['linkedin_url'] ?? ''
    ]);
    $id = $ins->fetchColumn();
    echo json_encode(['ok' => true, 'id' => $id]);
    exit;
}

// === UPDATE STATS ===
if ($action === 'update' && $_SERVER['REQUEST_METHOD'] === 'POST') {
    $input = json_decode(file_get_contents('php://input'), true);
    if (!$input || !$input['id']) { echo json_encode(['error' => 'id required']); exit; }
    
    $sets = [];
    $params = [];
    foreach (['likes', 'comments', 'reposts', 'views', 'linkedin_url', 'image'] as $field) {
        if (isset($input[$field])) {
            $sets[] = "$field = ?";
            $params[] = $input[$field];
        }
    }
    if ($sets) {
        $sets[] = "updated_at = NOW()";
        $params[] = $input['id'];
        $pdo->prepare("UPDATE admin.linkedin_posts SET " . implode(', ', $sets) . " WHERE id = ?")->execute($params);
    }
    echo json_encode(['ok' => true]);
    exit;
}

echo json_encode(['error' => 'unknown action', 'actions' => ['list', 'sync', 'add', 'update']]);