valentin_le_moign/thalim-plugin-hal-importer
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Import unitaire, statut publié/pending et validation des idHAL

Browse Source
This commit is contained in:
Valentin Le Moign
2026-05-28 17:04:20 +02:00
parent a61619077d
commit d8053ac82e
3 changed files with 246 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

212
includes/class-admin-page.php
View File

@@ -81,6 +81,10 @@ class Thalim_HAL_Admin_Page {
$this->handle_import();
}
if ($action === 'import_single') {
$this->handle_import_single();
}
if (self::CSV_IMPORT_ENABLED) {
if ($action === 'csv_upload') $this->handle_csv_upload();
if ($action === 'csv_batch') $this->handle_csv_batch();
@@ -116,6 +120,7 @@ class Thalim_HAL_Admin_Page {
$imported = 0;
$skipped = 0;
$errors = [];
$cache_updates = []; // hal_id => ['id', 'status'] for cache mutation
foreach ($raw_docs as $doc) {
$hal_id = $doc['halId_s'] ?? '';
@@ -132,9 +137,12 @@ class Thalim_HAL_Admin_Page {
$errors[] = $hal_id . ': ' . $post_id->get_error_message();
} else {
$imported++;
$cache_updates[$hal_id] = ['id' => $post_id, 'status' => 'pending'];
}
}
$this->update_preview_cache_after_import($date_from, $date_to, $author_hal_id, $cache_updates);
$msg = sprintf('%d publication(s) importée(s) en statut "En attente".', $imported);
if ($skipped) $msg .= sprintf(' %d ignorée(s) (déjà importées ou sans membre THALIM correspondant).', $skipped);
if (!empty($errors)) $msg .= ' Erreurs : ' . implode('; ', $errors);
@@ -142,6 +150,61 @@ class Thalim_HAL_Admin_Page {
$this->message = [empty($errors) ? 'success' : 'warning', $msg];
}
/**
* Import a single ready publication. hal_id + filter values are POSTed
* from the per-row form so the cache lookup hits the right preview entry.
*/
private function handle_import_single() {
$hal_id = sanitize_text_field($_POST['hal_id'] ?? '');
$date_from = sanitize_text_field($_POST['hal_date_from'] ?? '');
$date_to = sanitize_text_field($_POST['hal_date_to'] ?? '');
$author_hal_id = sanitize_text_field($_POST['hal_author_id'] ?? '');
if (!$hal_id) {
$this->message = ['error', 'hal_id manquant.'];
return;
}
$preview = $this->get_preview_data($date_from, $date_to, $author_hal_id);
if (is_wp_error($preview)) {
$this->message = ['error', 'API Error: ' . $preview->get_error_message()];
return;
}
$doc = null;
foreach ($preview['raw_docs'] ?? [] as $d) {
if (($d['halId_s'] ?? '') === $hal_id) { $doc = $d; break; }
}
if (!$doc) {
$this->message = ['warning', "Publication $hal_id introuvable dans le tableau (cache expiré ?). Rafraîchir et réessayer."];
return;
}
$importer = new Thalim_HAL_Importer_Logic();
if ($importer->is_imported($hal_id)) {
$this->message = ['warning', "Publication $hal_id déjà importée."];
return;
}
$this->load_wp_users_hal_ids();
if (empty($this->match_authors_to_users($doc['authIdHal_s'] ?? []))) {
$this->message = ['error', "Aucun membre THALIM identifié pour $hal_id."];
return;
}
$post_id = $importer->import($doc, $this->wp_users_by_hal_id);
if (is_wp_error($post_id)) {
$this->message = ['error', "Erreur import : " . $post_id->get_error_message()];
return;
}
$this->update_preview_cache_after_import($date_from, $date_to, $author_hal_id, [
$hal_id => ['id' => $post_id, 'status' => 'pending'],
]);
$this->message = ['success', sprintf('Publication %s importée (post #%d, en attente).', $hal_id, $post_id)];
}
private function render_styles() {
?>
<style>
@@ -231,7 +294,11 @@ class Thalim_HAL_Admin_Page {
<?php else: ?>
<?php $this->render_wp_users_debug(); ?>
<?php $this->render_summary($preview['stats']); ?>
<?php $this->render_preview_table($preview['docs']); ?>
<?php $this->render_preview_table($preview['docs'], [
'date_from' => $date_from,
'date_to' => $date_to,
'author_hal_id' => $author_hal_id,
]); ?>
<?php $this->render_legend(); ?>
<?php endif; ?>
</div>
@@ -261,7 +328,7 @@ class Thalim_HAL_Admin_Page {
<?php
}
private function render_preview_table($docs) {
private function render_preview_table($docs, $filters = []) {
if (empty($docs)) {
echo '<p>No publications found.</p>';
return;
@@ -279,12 +346,13 @@ class Thalim_HAL_Admin_Page {
<th>Date</th>
<th>Membres THALIM</th>
<th>Lien HAL</th>
<th>Action</th>
</tr>
</thead>
<tbody>
<?php foreach ($docs as $doc): ?>
<tr class="<?php echo esc_attr($this->get_row_class($doc)); ?>">
<td><?php echo $this->get_status_icon($doc); ?></td>
<td><?php $this->render_status_cell($doc); ?></td>
<td><code style="font-size:11px"><?php echo esc_html($doc['hal_id']); ?></code></td>
<td>
<strong><?php echo esc_html($doc['title']); ?></strong>
@@ -325,6 +393,19 @@ class Thalim_HAL_Admin_Page {
<td>
<a href="<?php echo esc_url($doc['url']); ?>" target="_blank" class="button button-small">Voir sur HAL</a>
</td>
<td>
<?php if ($doc['has_match'] && !$doc['is_imported']): ?>
<form method="post" style="margin:0">
<?php wp_nonce_field('thalim_hal_action'); ?>
<input type="hidden" name="thalim_hal_action" value="import_single">
<input type="hidden" name="hal_id" value="<?php echo esc_attr($doc['hal_id']); ?>">
<input type="hidden" name="hal_date_from" value="<?php echo esc_attr($filters['date_from'] ?? ''); ?>">
<input type="hidden" name="hal_date_to" value="<?php echo esc_attr($filters['date_to'] ?? ''); ?>">
<input type="hidden" name="hal_author_id" value="<?php echo esc_attr($filters['author_hal_id'] ?? ''); ?>">
<button class="button button-primary button-small">Importer</button>
</form>
<?php endif; ?>
</td>
</tr>
<?php endforeach; ?>
</tbody>
@@ -343,24 +424,57 @@ class Thalim_HAL_Admin_Page {
<?php
}
/**
* Validate WP users' HAL IDs against HAL's author referential.
* Cached 24h to avoid hitting the API on every page load.
* Returns map [normalized_hal_id => true|false|null] (null = API error).
*/
private function get_hal_ids_validity(array $hal_ids) {
sort($hal_ids);
$cache_key = 'thalim_hal_ids_validity_' . md5(implode('|', $hal_ids));
$cached = get_transient($cache_key);
if ($cached !== false) return $cached;
$result = $this->api->validate_hal_ids($hal_ids);
set_transient($cache_key, $result, DAY_IN_SECONDS);
return $result;
}
private function render_wp_users_debug() {
$this->load_wp_users_hal_ids();
if (empty($this->wp_users_by_hal_id)) {
echo '<div class="notice notice-warning"><p>Aucun utilisateur WordPress n\'a le champ <code>identifiant_hal</code> renseigné.</p></div>';
return;
}
$validity = $this->get_hal_ids_validity(array_keys($this->wp_users_by_hal_id));
$invalid_count = count(array_filter($validity, fn($v) => $v === false));
?>
<details style="margin-bottom:15px;background:#f0f6fc;padding:10px;border-radius:4px;border-left:4px solid #0073aa">
<summary style="cursor:pointer;font-weight:bold">
Utilisateurs WordPress avec identifiant HAL (<?php echo count($this->wp_users_by_hal_id); ?> utilisateurs) — Cliquer pour déplier
Utilisateurs WordPress avec identifiant HAL (<?php echo count($this->wp_users_by_hal_id); ?> utilisateurs<?php
if ($invalid_count > 0) echo ', <span style="color:#dc3545">' . $invalid_count . ' invalide(s)</span>';
?>) — Cliquer pour déplier
</summary>
<table class="widefat" style="margin-top:10px;font-size:12px">
<thead><tr><th>Utilisateur</th><th>Identifiant HAL</th><th>Debug (brut)</th><th>Modifier</th></tr></thead>
<thead><tr><th>Utilisateur</th><th>Identifiant HAL</th><th>Validité HAL</th><th>Debug (brut)</th><th>Modifier</th></tr></thead>
<tbody>
<?php foreach ($this->wp_users_by_hal_id as $hal_id => $user): ?>
<tr>
<?php foreach ($this->wp_users_by_hal_id as $hal_id => $user):
$valid = $validity[$hal_id] ?? null;
$row_style = $valid === false ? 'background:#f8d7da' : '';
?>
<tr style="<?php echo $row_style; ?>">
<td><?php echo esc_html($user['name']); ?> <small style="color:#999">(ID : <?php echo esc_html($user['id']); ?>)</small></td>
<td><code><?php echo esc_html($hal_id); ?></code></td>
<td>
<?php if ($valid === true): ?>
<span style="color:#28a745" title="Trouvé dans le référentiel auteur HAL">✓ valide</span>
<?php elseif ($valid === false): ?>
<span style="color:#dc3545;font-weight:600" title="Aucun auteur HAL ne correspond à cet idHAL">✗ invalide</span>
<?php else: ?>
<span style="color:#999;font-style:italic" title="Erreur API ou validation indisponible">— inconnu</span>
<?php endif; ?>
</td>
<td><code style="background:#ffe;font-size:10px">"<?php echo esc_html($hal_id); ?>" (<?php echo strlen($hal_id); ?> car.)</code></td>
<td><a href="<?php echo esc_url(get_edit_user_link($user['id'])); ?>" target="_blank" class="button button-small">Modifier</a></td>
</tr>
@@ -371,8 +485,38 @@ class Thalim_HAL_Admin_Page {
<?php
}
private function get_preview_cache_key($date_from, $date_to, $author_hal_id) {
return 'thalim_hal_preview_' . md5($date_from . '|' . $date_to . '|' . $author_hal_id);
}
/**
* Mute the cached preview in place to reflect newly-imported posts —
* avoids re-hitting the HAL API just to refresh statuses after an import.
*
* @param array $imports Map hal_id => ['id' => int, 'status' => string]
*/
private function update_preview_cache_after_import($date_from, $date_to, $author_hal_id, array $imports) {
if (empty($imports)) return;
$cache_key = $this->get_preview_cache_key($date_from, $date_to, $author_hal_id);
$preview = get_transient($cache_key);
if (!$preview) return;
foreach ($preview['docs'] as &$d) {
if (isset($imports[$d['hal_id']]) && empty($d['is_imported'])) {
$d['is_imported'] = true;
$d['imported_post_id'] = $imports[$d['hal_id']]['id'];
$d['imported_post_status'] = $imports[$d['hal_id']]['status'];
$preview['stats']['ready']--;
$preview['stats']['imported']++;
}
}
unset($d);
set_transient($cache_key, $preview, 300);
}
private function get_preview_data($date_from = '', $date_to = '', $author_hal_id = '') {
$cache_key = 'thalim_hal_preview_' . md5($date_from . '|' . $date_to . '|' . $author_hal_id);
$cache_key = $this->get_preview_cache_key($date_from, $date_to, $author_hal_id);
$cached = get_transient($cache_key);
if ($cached !== false) return $cached;
@@ -396,7 +540,8 @@ class Thalim_HAL_Admin_Page {
foreach ($result['response']['docs'] ?? [] as $doc) {
$hal_id = $doc['halId_s'] ?? '';
$is_imported = $importer->is_imported($hal_id);
$imported_post = $importer->get_imported_post($hal_id);
$is_imported = $imported_post !== null;
$author_hal_ids = $doc['authIdHal_s'] ?? [];
$matched_users = $this->match_authors_to_users($author_hal_ids);
$has_match = !empty($matched_users);
@@ -411,18 +556,20 @@ class Thalim_HAL_Admin_Page {
}
$preview['docs'][] = [
'hal_id' => $hal_id,
'title' => $doc['title_s'][0] ?? 'N/A',
'type' => $doc['docType_s'] ?? '',
'authors' => $doc['authFullName_s'] ?? [],
'author_hal_ids' => $author_hal_ids,
'publication_date' => $doc['publicationDate_s'] ?? '',
'produced_date' => $doc['submittedDate_s'] ?? '',
'journal' => $doc['journalTitle_s'] ?? $doc['bookTitle_s'] ?? '',
'url' => $doc['uri_s'] ?? '',
'is_imported' => $is_imported,
'matched_users' => $matched_users,
'has_match' => $has_match,
'hal_id' => $hal_id,
'title' => $doc['title_s'][0] ?? 'N/A',
'type' => $doc['docType_s'] ?? '',
'authors' => $doc['authFullName_s'] ?? [],
'author_hal_ids' => $author_hal_ids,
'publication_date' => $doc['publicationDate_s'] ?? '',
'produced_date' => $doc['submittedDate_s'] ?? '',
'journal' => $doc['journalTitle_s'] ?? $doc['bookTitle_s'] ?? '',
'url' => $doc['uri_s'] ?? '',
'is_imported' => $is_imported,
'imported_post_id' => $imported_post['id'] ?? null,
'imported_post_status' => $imported_post['status'] ?? null,
'matched_users' => $matched_users,
'has_match' => $has_match,
];
$preview['raw_docs'][] = $doc; // Full HAL doc kept for import
}
@@ -478,9 +625,24 @@ class Thalim_HAL_Admin_Page {
return 'hal-status-blocked';
}
private function get_status_icon($doc) {
if ($doc['is_imported']) return '<span title="Already imported" style="color:#28a745;font-size:18px">✓</span>';
if ($doc['has_match']) return '<span title="Ready to import" style="color:#ffc107;font-size:18px">★</span>';
return '<span title="No matched user" style="color:#dc3545;font-size:18px">✗</span>';
private function render_status_cell($doc) {
if ($doc['is_imported']) {
$status = $doc['imported_post_status'] ?? '';
$is_pending = $status === 'pending';
$label = $is_pending ? 'En attente' : ($status === 'publish' ? 'Publié' : $status);
?>
<span title="Importé" style="color:#28a745;font-size:18px">✓</span>
<div><small><?php echo esc_html($label); ?></small></div>
<?php if ($is_pending && !empty($doc['imported_post_id'])): ?>
<a href="<?php echo esc_url(get_edit_post_link($doc['imported_post_id'])); ?>" class="button button-small" style="margin-top:3px">Modifier</a>
<?php endif; ?>
<?php
return;
}
if ($doc['has_match']) {
echo '<span title="Ready to import" style="color:#ffc107;font-size:18px">★</span>';
return;
}
echo '<span title="No matched user" style="color:#dc3545;font-size:18px">✗</span>';
}
}

43
includes/class-hal-api.php
View File

@@ -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
*/

16
includes/class-importer.php
View File

@@ -50,6 +50,22 @@ class Thalim_HAL_Importer_Logic {
)) > 0;
}
/**
* Return ['id' => int, 'status' => string] for the post matching this hal_id, or null.
*/
public function get_imported_post($hal_id) {
if (empty($hal_id)) return null;
global $wpdb;
$row = $wpdb->get_row($wpdb->prepare(
"SELECT p.ID, p.post_status FROM {$wpdb->posts} p
INNER JOIN {$wpdb->postmeta} pm ON pm.post_id = p.ID
WHERE pm.meta_key = 'hal_id' AND pm.meta_value = %s
LIMIT 1",
$hal_id
));
return $row ? ['id' => (int) $row->ID, 'status' => $row->post_status] : null;
}
/**
* Get category ID for HAL doc type
*/