Cedric/mon-petit-cinema
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Actualiser api.php

Browse Source
This commit is contained in:
Cedric
2026-07-02 16:41:44 +02:00
parent 40c2bf927c
commit 2331efa4a1
1 changed files with 29 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
api.php
+29 -5
View File
@@ -116,7 +116,7 @@ function emptyPhysicalResult() {
]; ];
} }
// ── NOUVELLE FONCTION : SCRAPPING GO-UPC.COM (MODIFIÉE) ── // ── NOUVELLE FONCTION : SCRAPPING GO-UPC.COM (CORRIGÉE) ──
function fetchFromGoUpc($ean) { function fetchFromGoUpc($ean) {
$empty = emptyPhysicalResult(); $empty = emptyPhysicalResult();
$url = "https://go-upc.com/search?q=" . urlencode($ean); $url = "https://go-upc.com/search?q=" . urlencode($ean);
@@ -128,21 +128,45 @@ function fetchFromGoUpc($ean) {
$empty['title'] = trim(strip_tags($m[1])); $empty['title'] = trim(strip_tags($m[1]));
} elseif (preg_match('/<title[^>]*>([^<]+)<\/title>/i', $html, $m)) { } elseif (preg_match('/<title[^>]*>([^<]+)<\/title>/i', $html, $m)) {
$title = trim($m[1]); $title = trim($m[1]);
$title = preg_replace('/\s*[-–]\s*EAN\s+\d+\s*[-–]\s*Go-UPC.*$/i', '', $title);
$title = preg_replace('/\s*[-–]\s*Go-UPC.*$/i', '', $title); $title = preg_replace('/\s*[-–]\s*Go-UPC.*$/i', '', $title);
$empty['title'] = $title; $empty['title'] = $title;
} }
// ✅ RÉCUPÉRATION DE L'IMAGE DEPUIS GO-UPC // ✅ RÉCUPÉRATION DE L'IMAGE DEPUIS GO-UPC (VERSION ROBUSTE)
// Chercher dans les figures product-image // 1. Chercher directement l'URL Amazon S3 de go-upc (plus fiable)
if (preg_match('/<figure[^>]*class="product-image[^"]*"[^>]*>.*?<img[^>]*src="([^"]+\.(?:jpg|jpeg|png|webp))"/is', $html, $m)) { if (preg_match('/src="(https:\/\/go-upc\.s3\.amazonaws\.com\/images\/[^"]+)"/i', $html, $m)) {
$empty['poster'] = trim($m[1]); $empty['poster'] = trim($m[1]);
} elseif (preg_match('/<img[^>]*src="([^"]+go-upc\.s3\.amazonaws\.com[^"]+)"/i', $html, $m)) { }
// 2. Fallback : chercher dans les figures product-image
elseif (preg_match('/<figure[^>]*class="product-image[^"]*"[^>]*>.*?<img[^>]*src="([^"]+)"/is', $html, $m)) {
$empty['poster'] = trim($m[1]);
}
// 3. Fallback : chercher toute image de go-upc.s3.amazonaws.com
elseif (preg_match('/src="([^"]+go-upc\.s3\.amazonaws\.com[^"]+)"/i', $html, $m)) {
$empty['poster'] = trim($m[1]); $empty['poster'] = trim($m[1]);
} }
// Nettoyage du titre // Nettoyage du titre
if (!empty($empty['title'])) { if (!empty($empty['title'])) {
$empty['title'] = html_entity_decode($empty['title'], ENT_QUOTES | ENT_HTML5, 'UTF-8'); $empty['title'] = html_entity_decode($empty['title'], ENT_QUOTES | ENT_HTML5, 'UTF-8');
// ✅ APPLIQUER cleanTitle() pour supprimer les suffixes comme "- DVD", "- Édition Collector"
$empty['title'] = cleanTitle($empty['title']);
// ✅ Supprimer le nom du réalisateur au début si présent
// Pattern : "Prénom Nom Titre du film" → "Titre du film"
if (preg_match('/^([A-ZÀ-Ÿ][a-zà-ÿ]+\s+[A-ZÀ-Ÿ][a-zà-ÿ]+)\s+(.+)$/u', $empty['title'], $nameMatch)) {
$possibleDirector = $nameMatch[1];
$possibleTitle = $nameMatch[2];
// Vérifier si le "titre" restant semble être un vrai titre (pas juste un mot)
if (strlen($possibleTitle) > 3 && !preg_match('/^(DVD|Blu-ray|4K|VHS|CD|Vinyl)$/i', $possibleTitle)) {
$empty['director'] = $possibleDirector;
$empty['title'] = $possibleTitle;
}
}
$empty['title'] = trim($empty['title']); $empty['title'] = trim($empty['title']);
} }