Initial commit

2026-05-12 23:33:56 +02:00
commit 57719052f3
6 changed files with 1611 additions and 0 deletions
--- a/includes/class-hal-api.php
+++ b/includes/class-hal-api.php
@@ -0,0 +1,137 @@
+<?php
+/**
+ * HAL API Class - Handles communication with HAL API
+ */
+
+if (!defined('ABSPATH')) {
+    exit;
+}
+
+class Thalim_HAL_API {
+
+    // Fields to retrieve from HAL (expanded for import preview)
+    private const FIELDS = 'halId_s,title_s,docType_s,authFullName_s,authIdHal_s,publicationDate_s,producedDate_s,submittedDate_s,journalTitle_s,bookTitle_s,uri_s,fileMain_s,abstract_s,keyword_s,publisher_s,modifiedDate_s,doiId_s,citationFull_s,conferenceTitle_s,city_s,country_s,defenseDate_s';
+
+    /**
+     * Fetch publications from THALIM structure
+     *
+     * @param int    $rows      Number of results to fetch
+     * @param int    $start     Offset
+     * @param string $sort      Solr sort expression
+     * @param string $date_from Filter start date (YYYY-MM-DD), empty = no lower bound
+     * @param string $date_to   Filter end date (YYYY-MM-DD), empty = no upper bound
+     */
+    public function fetch_publications($rows = 100, $start = 0, $sort = 'producedDate_tdate desc', $date_from = '', $date_to = '', $author_hal_id = '') {
+        $url = $this->build_url($rows, $start, $sort, $date_from, $date_to, $author_hal_id);
+        return $this->request($url);
+    }
+
+    /**
+     * Fetch full HAL docs by a list of hal_ids (batched).
+     * Uses Solr fq=halId_s:(id1 OR id2 OR ...) syntax. No structId filter —
+     * fetch by halId exact, regardless of structure.
+     *
+     * @param string[] $hal_ids  HAL IDs to fetch.
+     * @param int      $batch    Batch size (default 100).
+     * @return array|WP_Error    Array keyed by halId_s, or WP_Error on failure.
+     */
+    public function fetch_by_hal_ids(array $hal_ids, int $batch = 100) {
+        $docs = [];
+        $chunks = array_chunk(array_values(array_unique($hal_ids)), $batch);
+        foreach ($chunks as $chunk) {
+            $filter = 'halId_s:(' . implode(' OR ', $chunk) . ')';
+            $params = [
+                'q='    . urlencode('*:*'),
+                'fq='   . urlencode($filter),
+                'rows=' . count($chunk),
+                'fl='   . urlencode(self::FIELDS),
+                'wt=json',
+            ];
+            $url  = THALIM_HAL_API_BASE . '?' . implode('&', $params);
+            $data = $this->request($url);
+            if (is_wp_error($data)) return $data;
+            foreach ($data['response']['docs'] ?? [] as $doc) {
+                if (!empty($doc['halId_s'])) {
+                    $docs[$doc['halId_s']] = $doc;
+                }
+            }
+            // Be polite with HAL if we have multiple chunks
+            if (count($chunks) > 1) usleep(250000);
+        }
+        return $docs;
+    }
+
+    /**
+     * Test API connection
+     */
+    public function test_connection() {
+        $result = $this->fetch_publications(5);
+        if (is_wp_error($result)) return $result;
+        return [
+            'success' => true,
+            'total'   => $result['response']['numFound'] ?? 0,
+            'sample'  => $result['response']['docs'] ?? []
+        ];
+    }
+
+    /**
+     * Build API URL with proper fq parameter handling
+     *
+     * @param int    $rows
+     * @param int    $start
+     * @param string $sort
+     * @param string $date_from YYYY-MM-DD or empty
+     * @param string $date_to   YYYY-MM-DD or empty
+     */
+    private function build_url($rows = 5, $start = 0, $sort = 'modifiedDate_tdate desc', $date_from = '', $date_to = '', $author_hal_id = '') {
+        $doc_types = implode(' OR ', THALIM_HAL_DOC_TYPES);
+
+        $from = $date_from ? $date_from . 'T00:00:00Z' : '*';
+        $to   = $date_to   ? $date_to   . 'T23:59:59Z' : '*';
+
+        $params = [
+            'q=' . urlencode('*:*'),
+            'fq=' . urlencode('structId_i:' . THALIM_HAL_STRUCT_ID),
+            'fq=' . urlencode('docType_s:(' . $doc_types . ')'),
+        ];
+
+        if ($from !== '*' || $to !== '*') {
+            $params[] = 'fq=' . urlencode('producedDate_tdate:[' . $from . ' TO ' . $to . ']');
+        }
+
+        if ($author_hal_id !== '') {
+            $params[] = 'fq=' . urlencode('authIdHal_s:' . $author_hal_id);
+        }
+
+        $params = array_merge($params, [
+            'rows=' . intval($rows),
+            'start=' . intval($start),
+            'sort=' . urlencode($sort),
+            'fl=' . urlencode(self::FIELDS),
+            'wt=json'
+        ]);
+
+        return THALIM_HAL_API_BASE . '?' . implode('&', $params);
+    }
+
+    /**
+     * Get API URL for debugging display
+     */
+    public function get_api_url($rows = 5) {
+        return $this->build_url($rows, 0);
+    }
+
+    /**
+     * Make HTTP request
+     */
+    private function request($url) {
+        $response = wp_remote_get($url, ['timeout' => 30, 'headers' => ['Accept' => 'application/json']]);
+        if (is_wp_error($response)) return $response;
+
+        $code = wp_remote_retrieve_response_code($response);
+        if ($code !== 200) return new WP_Error('api_error', "HTTP $code");
+
+        $data = json_decode(wp_remote_retrieve_body($response), true);
+        return json_last_error() === JSON_ERROR_NONE ? $data : new WP_Error('json_error', 'Invalid JSON');
+    }
+}