Import HAL : match des auteurs sur toutes les formes d'idHAL (résolution en fond via cron)

2026-07-07 13:44:08 +02:00
parent c38f8b4d7e
commit f083b36369
5 changed files with 262 additions and 5 deletions
--- a/includes/class-hal-forms-cache.php
+++ b/includes/class-hal-forms-cache.php
@@ -0,0 +1,115 @@
+<?php
+/**
+ * Background resolver + cache for alternate idHAL forms.
+ *
+ * Problem: a member's WP `identifiant_hal` stores ONE idHAL form, but HAL may
+ * tag their publications under other valid forms of the same person. Matching
+ * on the single stored form silently drops those links.
+ *
+ * Fix: resolve every member's full set of valid idHAL forms (via HAL ref/author)
+ * and cache them in usermeta `_hal_id_forms`. Matching then uses the cached
+ * forms — with ZERO network calls in the UI / import path.
+ *
+ * All resolution happens in the background via WP-Cron, in small batches, so the
+ * admin interface and the import flow stay as fast as before.
+ */
+
+if (!defined('ABSPATH')) {
+    exit;
+}
+
+class Thalim_HAL_Forms_Cache {
+
+    const META_FORMS   = '_hal_id_forms';      // usermeta: JSON array of forms
+    const META_SOURCE  = '_hal_id_forms_src';  // usermeta: the identifiant_hal the cache was built from
+    const CRON_HOOK    = 'thalim_hal_refresh_forms';
+    const BATCH_SIZE   = 15;                    // members resolved per cron tick
+
+    /** Wire the background hooks. Called once at plugins_loaded. */
+    public static function init() {
+        add_action(self::CRON_HOOK, [__CLASS__, 'run_batch']);
+
+        // Ensure the recurring event exists (hourly).
+        if (!wp_next_scheduled(self::CRON_HOOK)) {
+            wp_schedule_event(time() + 60, 'hourly', self::CRON_HOOK);
+        }
+
+        // When a profile's identifiant_hal changes, invalidate its cache so the
+        // next background tick re-resolves it (no synchronous API call here).
+        add_action('updated_user_meta', [__CLASS__, 'on_user_meta_change'], 10, 4);
+        add_action('added_user_meta',   [__CLASS__, 'on_user_meta_change'], 10, 4);
+    }
+
+    /** Remove the scheduled event (plugin deactivation). */
+    public static function clear_schedule() {
+        $ts = wp_next_scheduled(self::CRON_HOOK);
+        if ($ts) wp_unschedule_event($ts, self::CRON_HOOK);
+    }
+
+    /** Invalidate a user's cache when their identifiant_hal is edited. */
+    public static function on_user_meta_change($meta_id, $user_id, $meta_key, $_meta_value) {
+        if ($meta_key === 'identifiant_hal') {
+            delete_user_meta($user_id, self::META_SOURCE); // marks the cache stale
+        }
+    }
+
+    /**
+     * Read cached alternate forms for a user's identifiant_hal.
+     * Always returns at least the stored form itself (graceful degradation:
+     * behaves exactly like the old code until the background job has run).
+     *
+     * @return string[] lowercased idHAL forms.
+     */
+    public static function forms_for_user($user_id, $identifiant_hal) {
+        $base = strtolower(trim((string) $identifiant_hal));
+        if ($base === '') return [];
+
+        $src = get_user_meta($user_id, self::META_SOURCE, true);
+        if ($src === $base) {
+            $cached = json_decode((string) get_user_meta($user_id, self::META_FORMS, true), true);
+            if (is_array($cached) && $cached) {
+                $cached[] = $base;
+                return array_values(array_unique(array_map('strtolower', $cached)));
+            }
+        }
+        // Not resolved yet (or stale) -> fall back to the single stored form.
+        return [$base];
+    }
+
+    /**
+     * Cron worker: resolve forms for a batch of members whose cache is missing
+     * or stale. Runs in the background; safe to be interrupted (idempotent).
+     */
+    public static function run_batch() {
+        $users = get_users([
+            'meta_key'     => 'identifiant_hal',
+            'meta_compare' => 'EXISTS',
+            'fields'       => ['ID'],
+            'number'       => 500,
+        ]);
+
+        // Select those needing (re)resolution: stored form != cached source.
+        $todo = [];
+        foreach ($users as $u) {
+            $idhal = strtolower(trim((string) get_user_meta($u->ID, 'identifiant_hal', true)));
+            if ($idhal === '') continue;
+            $src = get_user_meta($u->ID, self::META_SOURCE, true);
+            if ($src !== $idhal) {
+                $todo[$idhal][] = (int) $u->ID; // group users sharing a slug
+            }
+            if (count($todo) >= self::BATCH_SIZE) break;
+        }
+        if (empty($todo)) return;
+
+        $api = new Thalim_HAL_API();
+        $expanded = $api->expand_hal_id_forms(array_keys($todo));
+
+        foreach ($todo as $idhal => $user_ids) {
+            $forms = $expanded[$idhal] ?? [$idhal];
+            foreach ($user_ids as $uid) {
+                update_user_meta($uid, self::META_FORMS, wp_json_encode(array_values($forms)));
+                update_user_meta($uid, self::META_SOURCE, $idhal);
+            }
+        }
+    }
+}