Import unitaire, statut publié/pending et validation des idHAL

includes/class-hal-api.php

@@ -61,6 +61,49 @@ class Thalim_HAL_API {
         return $docs;
     }
 
+    /**
+     * Check which HAL IDs exist in HAL's author referential (ref/author).
+     * Returns map [normalized_hal_id => bool|null] — null when the API errored
+     * (treated as "unknown", not "invalid").
+     */
+    public function validate_hal_ids(array $hal_ids) {
+        $result = [];
+        $clean  = [];
+        foreach ($hal_ids as $id) {
+            $id = trim((string) $id);
+            if ($id !== '') $clean[strtolower($id)] = $id;
+        }
+        if (empty($clean)) return $result;
+
+        // Default all to false; flip to true when found
+        foreach ($clean as $norm => $_) $result[$norm] = false;
+
+        $endpoint = 'https://api.archives-ouvertes.fr/ref/author/';
+        $chunks   = array_chunk(array_values($clean), 100);
+        foreach ($chunks as $chunk) {
+            // Solr OR with quoted values to tolerate dashes/dots in slugs
+            $quoted = array_map(fn($id) => '"' . str_replace('"', '', $id) . '"', $chunk);
+            $params = [
+                'q='   . urlencode('idHal_s:(' . implode(' OR ', $quoted) . ')'),
+                'fl='  . urlencode('idHal_s'),
+                'rows=' . count($chunk),
+                'wt=json',
+            ];
+            $data = $this->request($endpoint . '?' . implode('&', $params));
+            if (is_wp_error($data)) {
+                foreach ($chunk as $id) $result[strtolower($id)] = null;
+                continue;
+            }
+            foreach ($data['response']['docs'] ?? [] as $doc) {
+                if (!empty($doc['idHal_s'])) {
+                    $result[strtolower($doc['idHal_s'])] = true;
+                }
+            }
+            if (count($chunks) > 1) usleep(250000);
+        }
+        return $result;
+    }
+
     /**
      * Test API connection
      */