zm/tougao
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

框架本地无法运行，修复尝试

Browse Source
This commit is contained in:
wangjinlei
2026-06-04 13:33:13 +08:00
parent bbd690ca0f
commit 0ee7c38000
359 changed files with 2767 additions and 3114 deletions
Download Patch File Download Diff File Expand all files Collapse all files

705
application/common/BackgroundCheckService.php Normal file
View File

@@ -0,0 +1,705 @@
<?php
namespace app\common;
/**
* 背景调查公共服务
* 封装 OpenAlex / CrossRef / Retraction Watch 数据查询
*/
class BackgroundCheckService
{
private $openAlexBase = 'https://api.openalex.org';
private $crossRefBase = 'https://api.crossref.org';
private $mailto = 'publisher@tmrjournals.com';
// ===================== OpenAlex =====================
public function openAlexGet($path, $query = [])
{
$query['mailto'] = $this->mailto;
$url = $this->openAlexBase . $path . '?' . http_build_query($query);
$result = $this->httpGet($url, [
'Accept: application/json',
'User-Agent: TMRJournals-BackgroundCheck/1.0 (mailto:' . $this->mailto . ')',
]);
if (!$result['success']) {
return $result;
}
$data = json_decode($result['body'], true);
if (!is_array($data)) {
return ['success' => false, 'error' => 'OpenAlex返回数据格式异常'];
}
return ['success' => true, 'data' => $data];
}
public function resolveAuthor($params)
{
if (!empty($params['openalex_id'])) {
$id = preg_replace('/^https?:\/\/openalex\.org\//', '', $params['openalex_id']);
$res = $this->openAlexGet('/authors/' . urlencode($id));
if (!$res['success']) {
return ['success' => false, 'error' => $res['error']];
}
return ['success' => true, 'data' => $res['data']];
}
if (!empty($params['orcid'])) {
$orcid = $this->cleanOrcid($params['orcid']);
$res = $this->openAlexGet('/authors/https://orcid.org/' . $orcid);
if (!$res['success']) {
return ['success' => false, 'error' => '未在 OpenAlex 找到该 ORCID 对应学者'];
}
return ['success' => true, 'data' => $res['data']];
}
if (empty($params['name'])) {
return ['success' => false, 'error' => '请提供 openalex_id、orcid 或 name'];
}
$filter = 'display_name.search:' . $params['name'];
if (!empty($params['affiliation'])) {
$filter .= ',last_known_institutions.display_name.search:' . $params['affiliation'];
}
$res = $this->openAlexGet('/authors', [
'search' => $params['name'],
'filter' => $filter,
'sort' => 'cited_by_count:desc',
'per-page' => 1,
]);
if (!$res['success']) {
return ['success' => false, 'error' => $res['error']];
}
$results = $res['data']['results'] ?? [];
if (empty($results)) {
return ['success' => false, 'error' => '未找到匹配学者，请补充 affiliation 或使用 orcid'];
}
return ['success' => true, 'data' => $results[0]];
}
public function fetchRetractedWorksOpenAlex($openAlexId)
{
$res = $this->openAlexGet('/works', [
'filter' => 'authorships.author.id:' . $openAlexId . ',is_retracted:true',
'sort' => 'publication_date:desc',
'per-page' => 25,
]);
if (!$res['success']) {
return ['count' => 0, 'list' => [], 'error' => $res['error']];
}
$list = [];
foreach ($res['data']['results'] ?? [] as $work) {
$list[] = $this->formatOpenAlexWork($work);
}
return ['count' => count($list), 'list' => $list, 'source' => 'openalex'];
}
public function fetchRecentWorks($openAlexId, $limit = 5)
{
$res = $this->openAlexGet('/works', [
'filter' => 'authorships.author.id:' . $openAlexId,
'sort' => 'publication_date:desc',
'per-page' => $limit,
]);
if (!$res['success']) {
return [];
}
$list = [];
foreach ($res['data']['results'] ?? [] as $work) {
$item = $this->formatOpenAlexWork($work);
$item['is_retracted'] = !empty($work['is_retracted']);
$list[] = $item;
}
return $list;
}
/**
* 按领域/关键词批量搜索学者（OpenAlex）
*/
public function searchAuthorsByField($keyword, $options = [])
{
$minHIndex = intval($options['min_h_index'] ?? 5);
$limit = min(max(intval($options['limit'] ?? 10), 1), 30);
$page = max(intval($options['page'] ?? 1), 1);
$topicId = $this->resolveTopicId($keyword);
$filters = [];
if ($topicId !== '') {
$filters[] = 'topics.id:' . $topicId;
}
if ($minHIndex > 0) {
$filters[] = 'summary_stats.h_index:>' . $minHIndex;
}
$query = [
'sort' => 'cited_by_count:desc',
'per-page' => $limit,
'page' => $page,
];
if (!empty($filters)) {
$query['filter'] = implode(',', $filters);
$query['search'] = $keyword;
} else {
$query['search'] = $keyword;
}
$res = $this->openAlexGet('/authors', $query);
if (!$res['success']) {
return ['success' => false, 'error' => $res['error']];
}
$authors = [];
foreach ($res['data']['results'] ?? [] as $author) {
$authors[] = $this->formatAuthorBrief($author);
}
return [
'success' => true,
'data' => [
'keyword' => $keyword,
'topic_id' => $topicId,
'page' => $page,
'limit' => $limit,
'total' => $res['data']['meta']['count'] ?? count($authors),
'list' => $authors,
],
];
}
private function resolveTopicId($keyword)
{
$res = $this->openAlexGet('/topics', [
'search' => $keyword,
'sort' => 'works_count:desc',
'per-page' => 1,
]);
if (!$res['success']) {
return '';
}
$results = $res['data']['results'] ?? [];
if (empty($results)) {
return '';
}
return $this->extractOpenAlexId($results[0]['id'] ?? '');
}
// ===================== CrossRef =====================
public function cleanDoi($doi)
{
$doi = trim($doi);
$doi = preg_replace('/^https?:\/\/doi\.org\//', '', $doi);
$doi = preg_replace('/^doi:\s*/i', '', $doi);
return trim($doi);
}
public function fetchCrossRefWork($doi)
{
$doi = $this->cleanDoi($doi);
if ($doi === '') {
return ['success' => false, 'error' => 'DOI为空'];
}
$url = $this->crossRefBase . '/works/' . urlencode($doi);
$result = $this->httpGet($url, [
'Accept: application/json',
'User-Agent: TMRJournals-BackgroundCheck/1.0 (mailto:' . $this->mailto . ')',
]);
if (!$result['success']) {
return ['success' => false, 'error' => $result['error']];
}
if ($result['http_code'] == 404) {
return ['success' => false, 'error' => 'DOI在CrossRef中未找到'];
}
if ($result['http_code'] != 200) {
return ['success' => false, 'error' => 'CrossRef返回 HTTP ' . $result['http_code']];
}
$data = json_decode($result['body'], true);
if (!isset($data['message'])) {
return ['success' => false, 'error' => 'CrossRef返回数据格式异常'];
}
return ['success' => true, 'message' => $data['message']];
}
public function parseCrossRefRetractionDetail($doi, $message)
{
$retraction = $this->detectCrossRefRetraction($message);
return [
'doi' => $this->cleanDoi($doi),
'title' => isset($message['title'][0]) ? $message['title'][0] : '',
'is_retracted' => $retraction['is_retracted'],
'retraction_detail' => $retraction['retraction_detail'],
'journal' => isset($message['container-title'][0]) ? $message['container-title'][0] : '',
'publisher' => $message['publisher'] ?? '',
'published_date' => isset($message['published-print']) ? $this->parseDateParts($message['published-print']) : '',
'authors' => $this->parseCrossRefAuthors($message['author'] ?? []),
'url' => $message['URL'] ?? ('https://doi.org/' . $this->cleanDoi($doi)),
];
}
public function enrichRetractionsWithCrossRef($retractionList)
{
$enriched = [];
foreach ($retractionList as $item) {
$doi = $this->cleanDoi($item['doi'] ?? '');
if ($doi === '') {
$item['crossref'] = ['success' => false, 'error' => '无DOI'];
$enriched[] = $item;
continue;
}
$res = $this->fetchCrossRefWork($doi);
if (!$res['success']) {
$item['crossref'] = ['success' => false, 'error' => $res['error']];
} else {
$item['crossref'] = [
'success' => true,
'data' => $this->parseCrossRefRetractionDetail($doi, $res['message']),
];
}
$enriched[] = $item;
usleep(200000);
}
return $enriched;
}
private function detectCrossRefRetraction($message)
{
$isRetracted = false;
$retractionDetail = [
'sources' => [],
'retraction_notices' => [],
'record_ids' => [],
];
foreach (['updated-by', 'update-to'] as $field) {
if (!isset($message[$field]) || !is_array($message[$field])) {
continue;
}
foreach ($message[$field] as $update) {
$updateType = strtolower($update['type'] ?? '');
$updateLabel = strtolower($update['label'] ?? '');
if (strpos($updateType, 'retract') === false && strpos($updateLabel, 'retract') === false) {
continue;
}
$isRetracted = true;
$source = $update['source'] ?? 'publisher';
$retractionDetail['sources'][] = $source;
$notice = [
'type' => $update['type'] ?? '',
'label' => $update['label'] ?? '',
'source' => $source,
'notice_doi'=> $update['DOI'] ?? '',
'date' => isset($update['updated']) ? $this->parseDateParts($update['updated']) : '',
'record_id' => $update['record-id'] ?? '',
];
$retractionDetail['retraction_notices'][] = $notice;
if (!empty($notice['record_id'])) {
$retractionDetail['record_ids'][] = $notice['record_id'];
}
}
}
$type = strtolower($message['type'] ?? '');
$subtype = strtolower($message['subtype'] ?? '');
if (strpos($type, 'retract') !== false || strpos($subtype, 'retract') !== false) {
$isRetracted = true;
$retractionDetail['is_retraction_notice'] = true;
}
if (isset($message['relation']) && is_array($message['relation'])) {
foreach ($message['relation'] as $relType => $relations) {
if (strpos(strtolower($relType), 'retract') !== false) {
$isRetracted = true;
$retractionDetail['relation'] = [$relType => $relations];
break;
}
}
}
$retractionDetail['sources'] = array_values(array_unique($retractionDetail['sources']));
$retractionDetail['record_ids'] = array_values(array_unique($retractionDetail['record_ids']));
return ['is_retracted' => $isRetracted, 'retraction_detail' => $retractionDetail];
}
// ===================== Retraction Watch (via CrossRef) =====================
/**
* 通过 CrossRef 检索 Retraction Watch 来源的撤稿记录（按作者姓名）
*/
public function fetchRetractionWatchByAuthor($authorName)
{
$url = $this->crossRefBase . '/works?' . http_build_query([
'query.author' => $authorName,
'filter' => 'update-type:retraction',
'rows' => 25,
'mailto' => $this->mailto,
]);
$result = $this->httpGet($url, [
'Accept: application/json',
'User-Agent: TMRJournals-BackgroundCheck/1.0 (mailto:' . $this->mailto . ')',
]);
if (!$result['success']) {
return ['count' => 0, 'list' => [], 'error' => $result['error']];
}
if ($result['http_code'] != 200) {
return ['count' => 0, 'list' => [], 'error' => 'CrossRef返回 HTTP ' . $result['http_code']];
}
$data = json_decode($result['body'], true);
$items = $data['message']['items'] ?? [];
$list = [];
foreach ($items as $message) {
$parsed = $this->parseCrossRefRetractionDetail($message['DOI'] ?? '', $message);
if (!$parsed['is_retracted']) {
continue;
}
$rwSources = array_filter($parsed['retraction_detail']['sources'] ?? [], function ($s) {
return stripos($s, 'retraction-watch') !== false || stripos($s, 'retraction_watch') !== false;
});
$list[] = [
'title' => $parsed['title'],
'doi' => $parsed['doi'],
'journal' => $parsed['journal'],
'publisher' => $parsed['publisher'],
'published_date' => $parsed['published_date'],
'is_retracted' => true,
'retraction_detail' => $parsed['retraction_detail'],
'from_retraction_watch' => !empty($rwSources) || !empty($parsed['retraction_detail']['record_ids']),
'source' => 'retraction_watch',
];
}
return [
'count' => count($list),
'list' => $list,
'source' => 'retraction_watch',
];
}
/**
* 合并 OpenAlex + Retraction Watch 撤稿记录（按 DOI 去重）
*/
public function mergeRetractionRecords($openAlexRetractions, $rwRetractions, $withCrossRefDetail = false)
{
$merged = [];
$doiMap = [];
foreach ([$openAlexRetractions, $rwRetractions] as $sourceData) {
foreach ($sourceData['list'] ?? [] as $item) {
$doi = $this->cleanDoi($item['doi'] ?? '');
$key = $doi !== '' ? strtolower($doi) : md5(json_encode($item));
if (!isset($doiMap[$key])) {
$doiMap[$key] = [
'title' => $item['title'] ?? '',
'doi' => $doi,
'journal' => $item['journal'] ?? '',
'publication_date' => $item['publication_date'] ?? ($item['published_date'] ?? ''),
'sources' => [],
'retraction_detail'=> $item['retraction_detail'] ?? [],
'from_retraction_watch' => !empty($item['from_retraction_watch']),
];
}
$src = $item['source'] ?? 'unknown';
if (!in_array($src, $doiMap[$key]['sources'])) {
$doiMap[$key]['sources'][] = $src;
}
if (!empty($item['from_retraction_watch'])) {
$doiMap[$key]['from_retraction_watch'] = true;
}
if (!empty($item['retraction_detail']) && empty($doiMap[$key]['retraction_detail'])) {
$doiMap[$key]['retraction_detail'] = $item['retraction_detail'];
}
}
}
$merged = array_values($doiMap);
if ($withCrossRefDetail) {
$merged = $this->enrichRetractionsWithCrossRef($merged);
}
$rwOnlyCount = 0;
foreach ($merged as $row) {
if (!empty($row['from_retraction_watch']) && count($row['sources'] ?? []) <= 1) {
$rwOnlyCount++;
}
}
return [
'count' => count($merged),
'openalex_count' => intval($openAlexRetractions['count'] ?? 0),
'rw_count' => intval($rwRetractions['count'] ?? 0),
'rw_only_count' => $rwOnlyCount,
'list' => $merged,
];
}
// ===================== 格式化 =====================
public function formatAuthorBrief($author)
{
$institutions = [];
foreach ($author['last_known_institutions'] ?? [] as $inst) {
$institutions[] = [
'name' => $inst['display_name'] ?? '',
'country' => $inst['country_code'] ?? '',
];
}
return [
'openalex_id' => $this->extractOpenAlexId($author['id'] ?? ''),
'name' => $author['display_name'] ?? '',
'orcid' => $this->extractOrcid($author['orcid'] ?? ''),
'works_count' => intval($author['works_count'] ?? 0),
'cited_by_count' => intval($author['cited_by_count'] ?? 0),
'h_index' => intval($author['summary_stats']['h_index'] ?? 0),
'institutions' => $institutions,
'openalex_url' => $author['id'] ?? '',
];
}
public function parseAuthorMetrics($author)
{
$stats = $author['summary_stats'] ?? [];
return [
'works_count' => intval($author['works_count'] ?? 0),
'cited_by_count' => intval($author['cited_by_count'] ?? 0),
'h_index' => intval($stats['h_index'] ?? 0),
'i10_index' => intval($stats['i10_index'] ?? 0),
'two_year_mean_cited' => round(floatval($stats['2yr_mean_citedness'] ?? 0), 2),
'level_label' => $this->getAcademicLevelLabel($stats),
];
}
public function parseResearchTopics($author)
{
$topics = [];
foreach ($author['x_concepts'] ?? [] as $concept) {
if (empty($concept['display_name'])) {
continue;
}
$topics[] = [
'name' => $concept['display_name'],
'score' => round(floatval($concept['score'] ?? 0), 3),
];
}
if (empty($topics)) {
foreach ($author['topics'] ?? [] as $topic) {
if (empty($topic['display_name'])) {
continue;
}
$topics[] = [
'name' => $topic['display_name'],
'score' => round(floatval($topic['score'] ?? 0), 3),
];
}
}
return array_slice($topics, 0, 8);
}
public function assessRisk($metrics, $retractions)
{
$retractionCount = intval($retractions['count'] ?? 0);
$rwOnlyCount = intval($retractions['rw_only_count'] ?? 0);
$level = 'low';
$score = 0;
$reasons = [];
if ($retractionCount === 0) {
$level = 'low';
$score = 10;
$reasons[] = 'OpenAlex 与 Retraction Watch 均未发现撤稿记录';
} elseif ($retractionCount === 1) {
$level = 'medium';
$score = 50;
$reasons[] = '发现 1 篇撤稿论文，建议人工核实撤稿原因';
} else {
$level = 'high';
$score = 80 + min($retractionCount * 5, 20);
$reasons[] = '发现 ' . $retractionCount . ' 篇撤稿论文，存在较高学术风险';
}
if ($rwOnlyCount > 0) {
$reasons[] = 'Retraction Watch 额外发现 ' . $rwOnlyCount . ' 条 OpenAlex 未收录的撤稿记录';
if ($level === 'low') {
$level = 'medium';
$score = max($score, 45);
}
}
$worksCount = max(intval($metrics['works_count'] ?? 0), 1);
$retractionRate = round($retractionCount / $worksCount * 100, 2);
if ($retractionCount > 0 && $retractionRate >= 5) {
$reasons[] = '撤稿率 ' . $retractionRate . '%，比例偏高';
if ($level === 'medium') {
$level = 'high';
$score = max($score, 70);
}
}
return [
'level' => $level,
'level_label' => $this->getRiskLevelLabel($level),
'score' => min($score, 100),
'retraction_count' => $retractionCount,
'retraction_rate' => $retractionRate . '%',
'rw_only_count' => $rwOnlyCount,
'reasons' => $reasons,
];
}
// ===================== 内部工具 =====================
private function formatOpenAlexWork($work)
{
return [
'title' => $work['display_name'] ?? '',
'doi' => $this->extractDoi($work),
'publication_date' => $work['publication_date'] ?? '',
'journal' => $work['primary_location']['source']['display_name'] ?? '',
'cited_by_count' => intval($work['cited_by_count'] ?? 0),
'openalex_url' => $work['id'] ?? '',
'source' => 'openalex',
];
}
private function parseCrossRefAuthors($authorList)
{
if (empty($authorList) || !is_array($authorList)) {
return [];
}
$result = [];
foreach ($authorList as $a) {
$result[] = [
'given' => $a['given'] ?? '',
'family' => $a['family'] ?? '',
'name' => isset($a['name']) ? $a['name'] : trim(($a['given'] ?? '') . ' ' . ($a['family'] ?? '')),
'orcid' => $a['ORCID'] ?? '',
];
}
return $result;
}
private function parseDateParts($dateObj)
{
if (!isset($dateObj['date-parts'][0])) {
return '';
}
$parts = $dateObj['date-parts'][0];
$y = isset($parts[0]) ? $parts[0] : '';
$m = isset($parts[1]) ? sprintf('%02d', $parts[1]) : '';
$d = isset($parts[2]) ? sprintf('%02d', $parts[2]) : '';
if ($y && $m && $d) {
return "{$y}-{$m}-{$d}";
}
if ($y && $m) {
return "{$y}-{$m}";
}
return (string)$y;
}
private function getAcademicLevelLabel($stats)
{
$h = intval($stats['h_index'] ?? 0);
if ($h >= 50) return '国际顶尖学者';
if ($h >= 30) return '资深专家';
if ($h >= 15) return '活跃研究者';
if ($h >= 5) return '青年学者';
if ($h > 0) return '初入领域';
return '暂无足够公开数据';
}
private function getRiskLevelLabel($level)
{
$map = ['low' => '低风险', 'medium' => '中风险', 'high' => '高风险'];
return $map[$level] ?? '未知';
}
public function extractOpenAlexId($id)
{
return preg_replace('/^https?:\/\/openalex\.org\//', '', $id);
}
public function extractOrcid($orcid)
{
if ($orcid === '') return '';
return preg_replace('/^https?:\/\/orcid\.org\//', '', $orcid);
}
public function cleanOrcid($orcid)
{
$orcid = trim($orcid);
$orcid = preg_replace('/^https?:\/\/orcid\.org\//', '', $orcid);
return trim($orcid);
}
private function extractDoi($work)
{
$doi = $work['doi'] ?? '';
return preg_replace('/^https?:\/\/doi\.org\//', '', $doi);
}
private function httpGet($url, $headers = [])
{
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
curl_setopt($ch, CURLOPT_TIMEOUT, 30);
curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
$body = curl_exec($ch);
$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
if (curl_errno($ch)) {
$error = curl_error($ch);
curl_close($ch);
return ['success' => false, 'error' => 'HTTP请求失败: ' . $error];
}
curl_close($ch);
return ['success' => true, 'body' => $body, 'http_code' => $httpCode];
}
}