387 lines
14 KiB
PHP
387 lines
14 KiB
PHP
<?php
|
||
|
||
namespace app\api\controller;
|
||
|
||
use think\Cache;
|
||
use app\common\BackgroundCheckService;
|
||
|
||
/**
|
||
* 背景调查控制器
|
||
*
|
||
* 功能：查询学者公开学术指标 + 撤稿/不端公开记录 + 按领域批量筛查
|
||
*
|
||
* 接口：
|
||
* api/background_check/demo - 快速体验
|
||
* api/background_check/searchAuthor - 按姓名搜索学者
|
||
* api/background_check/checkProfile - 完整背景调查报告
|
||
* api/background_check/checkRetractionByDoi - 单篇 DOI 撤稿详情（CrossRef + RW）
|
||
* api/background_check/batchScreenByField - 按领域批量筛查专家
|
||
*/
|
||
class BackgroundCheck extends Base
|
||
{
|
||
/** @var BackgroundCheckService */
|
||
private $service;
|
||
|
||
public function __construct(\think\Request $request = null)
|
||
{
|
||
parent::__construct($request);
|
||
$this->service = new BackgroundCheckService();
|
||
}
|
||
|
||
// ===================== 公开 API =====================
|
||
|
||
/**
|
||
* Demo：快速体验
|
||
*
|
||
* @param string orcid 可选，默认 0000-0002-2582-7480
|
||
*/
|
||
public function demo()
|
||
{
|
||
$orcid = $this->service->cleanOrcid($this->request->param('orcid', '0000-0002-2582-7480'));
|
||
|
||
$report = $this->buildProfileReport([
|
||
'orcid' => $orcid,
|
||
'with_crossref_detail'=> 1,
|
||
]);
|
||
if (!$report['success']) {
|
||
return jsonError($report['error']);
|
||
}
|
||
|
||
$report['data']['_demo_note'] = 'Demo 接口，正式使用请调用 checkProfile 或 batchScreenByField';
|
||
return jsonSuccess($report['data']);
|
||
}
|
||
|
||
/**
|
||
* 按姓名搜索学者
|
||
*
|
||
* @param string name 学者姓名（必填）
|
||
* @param string affiliation 单位关键词（可选）
|
||
* @param int limit 返回条数，默认5，最大20
|
||
*/
|
||
public function searchAuthor()
|
||
{
|
||
$name = trim($this->request->param('name', ''));
|
||
if ($name === '') {
|
||
return jsonError('name不能为空');
|
||
}
|
||
|
||
$affiliation = trim($this->request->param('affiliation', ''));
|
||
$limit = min(max(intval($this->request->param('limit', 5)), 1), 20);
|
||
|
||
$cacheKey = 'bg_search_' . md5($name . $affiliation . $limit);
|
||
$cached = Cache::get($cacheKey);
|
||
if ($cached) {
|
||
return jsonSuccess($cached);
|
||
}
|
||
|
||
$filter = 'display_name.search:' . $name;
|
||
if ($affiliation !== '') {
|
||
$filter .= ',last_known_institutions.display_name.search:' . $affiliation;
|
||
}
|
||
|
||
$res = $this->service->openAlexGet('/authors', [
|
||
'search' => $name,
|
||
'filter' => $filter,
|
||
'sort' => 'cited_by_count:desc',
|
||
'per-page' => $limit,
|
||
]);
|
||
|
||
if (!$res['success']) {
|
||
return jsonError($res['error']);
|
||
}
|
||
|
||
$list = [];
|
||
foreach ($res['data']['results'] ?? [] as $author) {
|
||
$list[] = $this->service->formatAuthorBrief($author);
|
||
}
|
||
|
||
$result = [
|
||
'query' => ['name' => $name, 'affiliation' => $affiliation],
|
||
'total' => $res['data']['meta']['count'] ?? count($list),
|
||
'list' => $list,
|
||
];
|
||
|
||
Cache::set($cacheKey, $result, 1800);
|
||
return jsonSuccess($result);
|
||
}
|
||
|
||
/**
|
||
* 完整背景调查报告
|
||
*
|
||
* @param string openalex_id OpenAlex 作者ID
|
||
* @param string orcid ORCID
|
||
* @param string name 姓名
|
||
* @param string affiliation 单位关键词
|
||
* @param int user_id 本地用户ID（可选）
|
||
* @param int with_crossref_detail 是否补充 CrossRef 撤稿详情，默认1
|
||
*/
|
||
public function checkProfile()
|
||
{
|
||
$params = [
|
||
'openalex_id' => trim($this->request->param('openalex_id', '')),
|
||
'orcid' => $this->service->cleanOrcid($this->request->param('orcid', '')),
|
||
'name' => trim($this->request->param('name', '')),
|
||
'affiliation' => trim($this->request->param('affiliation', '')),
|
||
'user_id' => intval($this->request->param('user_id', 0)),
|
||
'with_crossref_detail' => intval($this->request->param('with_crossref_detail', 1)),
|
||
];
|
||
|
||
if ($params['openalex_id'] === '' && $params['orcid'] === '' && $params['name'] === '') {
|
||
return jsonError('请至少提供 openalex_id、orcid 或 name 之一');
|
||
}
|
||
|
||
$report = $this->buildProfileReport($params);
|
||
if (!$report['success']) {
|
||
return jsonError($report['error']);
|
||
}
|
||
|
||
return jsonSuccess($report['data']);
|
||
}
|
||
|
||
/**
|
||
* 单篇 DOI 撤稿详情（CrossRef + Retraction Watch 来源标记）
|
||
*
|
||
* @param string doi 文章 DOI（必填）
|
||
*/
|
||
public function checkRetractionByDoi()
|
||
{
|
||
$doi = trim($this->request->param('doi', ''));
|
||
if ($doi === '') {
|
||
return jsonError('doi不能为空');
|
||
}
|
||
|
||
$res = $this->service->fetchCrossRefWork($doi);
|
||
if (!$res['success']) {
|
||
return jsonError($res['error']);
|
||
}
|
||
|
||
$detail = $this->service->parseCrossRefRetractionDetail($doi, $res['message']);
|
||
$sources = $detail['retraction_detail']['sources'] ?? [];
|
||
$fromRw = false;
|
||
foreach ($sources as $src) {
|
||
if (stripos($src, 'retraction-watch') !== false || stripos($src, 'retraction_watch') !== false) {
|
||
$fromRw = true;
|
||
break;
|
||
}
|
||
}
|
||
|
||
return jsonSuccess([
|
||
'doi' => $detail['doi'],
|
||
'title' => $detail['title'],
|
||
'journal' => $detail['journal'],
|
||
'publisher' => $detail['publisher'],
|
||
'authors' => $detail['authors'],
|
||
'is_retracted' => $detail['is_retracted'],
|
||
'from_retraction_watch' => $fromRw || !empty($detail['retraction_detail']['record_ids']),
|
||
'retraction_detail' => $detail['retraction_detail'],
|
||
'url' => $detail['url'],
|
||
'data_sources' => ['CrossRef', 'Retraction Watch'],
|
||
]);
|
||
}
|
||
|
||
/**
|
||
* 按领域批量筛查专家（含撤稿风险标签）
|
||
*
|
||
* @param string keyword 领域关键词（必填），如 immunotherapy、cancer
|
||
* @param int min_h_index 最低 h-index，默认5
|
||
* @param int limit 每页数量，默认10，最大30
|
||
* @param int page 页码，默认1
|
||
* @param int check_retraction 是否检查撤稿，默认1
|
||
*/
|
||
public function batchScreenByField()
|
||
{
|
||
$keyword = trim($this->request->param('keyword', ''));
|
||
if ($keyword === '') {
|
||
return jsonError('keyword不能为空');
|
||
}
|
||
|
||
$options = [
|
||
'min_h_index' => intval($this->request->param('min_h_index', 5)),
|
||
'limit' => min(max(intval($this->request->param('limit', 10)), 1), 30),
|
||
'page' => max(intval($this->request->param('page', 1)), 1),
|
||
];
|
||
$checkRetraction = intval($this->request->param('check_retraction', 1));
|
||
|
||
$cacheKey = 'bg_batch_' . md5(json_encode([$keyword, $options, $checkRetraction]));
|
||
$cached = Cache::get($cacheKey);
|
||
if ($cached) {
|
||
return jsonSuccess($cached);
|
||
}
|
||
|
||
$searchRes = $this->service->searchAuthorsByField($keyword, $options);
|
||
if (!$searchRes['success']) {
|
||
return jsonError($searchRes['error']);
|
||
}
|
||
|
||
$data = $searchRes['data'];
|
||
$screened = [];
|
||
$riskStats = ['low' => 0, 'medium' => 0, 'high' => 0];
|
||
|
||
foreach ($data['list'] as $author) {
|
||
$item = [
|
||
'profile' => $author,
|
||
'metrics' => [
|
||
'works_count' => $author['works_count'],
|
||
'cited_by_count' => $author['cited_by_count'],
|
||
'h_index' => $author['h_index'],
|
||
'level_label' => $this->service->parseAuthorMetrics([
|
||
'works_count' => $author['works_count'],
|
||
'cited_by_count' => $author['cited_by_count'],
|
||
'summary_stats' => ['h_index' => $author['h_index']],
|
||
])['level_label'],
|
||
],
|
||
];
|
||
|
||
if ($checkRetraction) {
|
||
$openAlexId = $author['openalex_id'];
|
||
$oaRetractions = $this->service->fetchRetractedWorksOpenAlex($openAlexId);
|
||
$rwRetractions = $this->service->fetchRetractionWatchByAuthor($author['name']);
|
||
$merged = $this->service->mergeRetractionRecords($oaRetractions, $rwRetractions, false);
|
||
|
||
$metrics = ['works_count' => $author['works_count']];
|
||
$risk = $this->service->assessRisk($metrics, $merged);
|
||
|
||
$item['retractions'] = [
|
||
'count' => $merged['count'],
|
||
'openalex_count'=> $merged['openalex_count'],
|
||
'rw_count' => $merged['rw_count'],
|
||
'rw_only_count' => $merged['rw_only_count'],
|
||
'list' => array_slice($merged['list'], 0, 3),
|
||
];
|
||
$item['risk_assessment'] = $risk;
|
||
$riskStats[$risk['level']]++;
|
||
} else {
|
||
$item['risk_assessment'] = [
|
||
'level' => 'unknown',
|
||
'level_label' => '未检测',
|
||
];
|
||
}
|
||
|
||
$screened[] = $item;
|
||
usleep(300000);
|
||
}
|
||
|
||
$result = [
|
||
'query' => [
|
||
'keyword' => $keyword,
|
||
'topic_id' => $data['topic_id'],
|
||
'min_h_index' => $options['min_h_index'],
|
||
'page' => $options['page'],
|
||
'limit' => $options['limit'],
|
||
'check_retraction' => $checkRetraction,
|
||
],
|
||
'total' => $data['total'],
|
||
'risk_stats' => $checkRetraction ? $riskStats : null,
|
||
'list' => $screened,
|
||
'disclaimer' => '批量筛查基于公开数据，同名作者可能存在误匹配，高风险条目需人工复核。',
|
||
];
|
||
|
||
Cache::set($cacheKey, $result, 900);
|
||
return jsonSuccess($result);
|
||
}
|
||
|
||
// ===================== 核心逻辑 =====================
|
||
|
||
/**
|
||
* 构建完整背景调查报告
|
||
*/
|
||
private function buildProfileReport($params)
|
||
{
|
||
$withCrossRef = !empty($params['with_crossref_detail']);
|
||
|
||
$author = $this->service->resolveAuthor($params);
|
||
if (!$author['success']) {
|
||
return $author;
|
||
}
|
||
|
||
$authorData = $author['data'];
|
||
$openAlexId = $this->service->extractOpenAlexId($authorData['id']);
|
||
$authorName = $authorData['display_name'] ?? '';
|
||
|
||
$metrics = $this->service->parseAuthorMetrics($authorData);
|
||
$topics = $this->service->parseResearchTopics($authorData);
|
||
$recentWorks = $this->service->fetchRecentWorks($openAlexId, 5);
|
||
|
||
// 扩展1: OpenAlex 撤稿
|
||
$oaRetractions = $this->service->fetchRetractedWorksOpenAlex($openAlexId);
|
||
|
||
// 扩展1: Retraction Watch 二次核对（CrossRef 来源）
|
||
$rwRetractions = $this->service->fetchRetractionWatchByAuthor($authorName);
|
||
|
||
// 扩展2: 合并并按需补充 CrossRef 撤稿详情
|
||
$retractions = $this->service->mergeRetractionRecords(
|
||
$oaRetractions,
|
||
$rwRetractions,
|
||
$withCrossRef
|
||
);
|
||
|
||
$risk = $this->service->assessRisk($metrics, $retractions);
|
||
|
||
$localInfo = null;
|
||
if (!empty($params['user_id'])) {
|
||
$localInfo = $this->fetchLocalUserInfo($params['user_id']);
|
||
}
|
||
|
||
$dataSources = ['OpenAlex', 'Retraction Watch (via CrossRef)'];
|
||
if ($withCrossRef) {
|
||
$dataSources[] = 'CrossRef';
|
||
}
|
||
|
||
return [
|
||
'success' => true,
|
||
'data' => [
|
||
'profile' => $this->service->formatAuthorBrief($authorData),
|
||
'metrics' => $metrics,
|
||
'research_topics' => $topics,
|
||
'recent_works' => $recentWorks,
|
||
'retractions' => $retractions,
|
||
'risk_assessment' => $risk,
|
||
'local_info' => $localInfo,
|
||
'data_sources' => $dataSources,
|
||
'disclaimer' => '本报告仅基于公开学术数据，不能替代正式背调或机构调查；未公开的不端行为无法检测。',
|
||
'generated_at' => date('Y-m-d H:i:s'),
|
||
],
|
||
];
|
||
}
|
||
|
||
/**
|
||
* 合并本地用户信息
|
||
*/
|
||
private function fetchLocalUserInfo($userId)
|
||
{
|
||
$user = $this->user_obj->where('user_id', $userId)->find();
|
||
if (!$user) {
|
||
return null;
|
||
}
|
||
|
||
$reviewer = $this->user_reviewer_info_obj
|
||
->where('reviewer_id', $userId)
|
||
->where('state', 0)
|
||
->find();
|
||
|
||
$majors = $this->major_to_user_obj
|
||
->where('user_id', $userId)
|
||
->where('state', 0)
|
||
->select();
|
||
|
||
$majorList = [];
|
||
foreach ($majors as $m) {
|
||
$majorList[] = [
|
||
'major_id' => $m['major_id'],
|
||
'path' => getMajorStr($m['major_id']),
|
||
];
|
||
}
|
||
|
||
return [
|
||
'user_id' => $userId,
|
||
'realname' => $user['realname'] ?? '',
|
||
'email' => $user['email'] ?? '',
|
||
'orcid' => $user['orcid'] ?? '',
|
||
'company' => $reviewer['company'] ?? '',
|
||
'title' => $reviewer['technical'] ?? '',
|
||
'field' => $reviewer['field'] ?? '',
|
||
'majors' => $majorList,
|
||
];
|
||
}
|
||
}
|