310 lines
11 KiB
PHP
310 lines
11 KiB
PHP
<?php
|
||
namespace app\api\controller;
|
||
use app\api\controller\Base;
|
||
use think\Db;
|
||
class Crossrefdoi extends Base{
|
||
|
||
public function __construct(\think\Request $request = null) {
|
||
parent::__construct($request);
|
||
}
|
||
// 配置项
|
||
private $mailto; // 邮箱（提升优先级）
|
||
private $timeout = 15; // 请求超时（秒）
|
||
private $maxRetry = 2; // 单个DOI最大重试次数
|
||
private $crossrefUrl = "https://api.crossref.org/works/"; //接口地址
|
||
/**
|
||
* 批量查询DOI信息（核心方法）
|
||
* @param array $dois DOI列表
|
||
* @return array 包含所有DOI的查询结果
|
||
*/
|
||
public function get($aParam = []){
|
||
|
||
//获取参数
|
||
$aParam = empty($aParam) ? $this->request->post() : $aParam;
|
||
|
||
$iPReferId = empty($aParam['p_refer_id']) ? 0 : $aParam['p_refer_id'];
|
||
if(empty($iPReferId)){
|
||
return json_encode(['status' => 2,'msg' => 'Please select the reference to be queried']);
|
||
}
|
||
$aWhere = ['p_refer_id' => $iPReferId,'state' => ['in', [0,2]]];
|
||
$aRefer = Db::name('production_article_refer')->field('title,joura,author,refer_doi,doilink,cs')->where($aWhere)->find();
|
||
if(empty($aRefer['refer_doi'])){
|
||
return json_encode(['status' => 3,'msg' => 'The doi of the reference is empty']);
|
||
}
|
||
|
||
$finalResult = [];
|
||
$sDoi = empty($aRefer['refer_doi']) ? '' : $aRefer['refer_doi'];
|
||
$sCheckDoi = $this->filterValidDoi($sDoi); // 过滤非法DOI
|
||
if(empty($sCheckDoi)){
|
||
return json_encode(['status' => 4,'msg' => 'Doi does not comply with the rules']);
|
||
}
|
||
|
||
// 调用单DOI查询（带重试）
|
||
$aDoiInfo = $this->fetchSingleDoiWithRetry($sCheckDoi);
|
||
if (!$aDoiInfo) {
|
||
$result['status'] = 'fail';
|
||
$result['fail_reason'] = "请求失败（重试{$this->maxRetry}次后仍失败）";
|
||
}
|
||
// 提取核心字段
|
||
$sTitle = $this->getTitle($aDoiInfo);
|
||
//期刊信息
|
||
$aPublisher = $this->getPublisher($aDoiInfo);
|
||
$sJoura = empty($aPublisher['title']) ? $aPublisher['short_title'] : $aPublisher['title'];
|
||
//作者信息
|
||
$aAuthor = $this->getAuthors($aDoiInfo);
|
||
$sAuthor = empty($aAuthor) ? '' : implode(',', $aAuthor);
|
||
$sDateno = $this->getVolumeIssuePages($aDoiInfo);
|
||
|
||
// 识别撤稿状态
|
||
$aRetractInfo = $this->checkRetracted($aDoiInfo);
|
||
$bIsRetracted = 2;
|
||
if($aRetractInfo['is_retracted'] == true){
|
||
$bIsRetracted = 1;
|
||
}
|
||
$sRetractReason = empty($aRetractInfo['reason']) ? '' : $aRetractInfo['reason'];
|
||
//获取dolink
|
||
$sDolink = $this->getDolink($aDoiInfo);
|
||
$sDolink = empty($sDolink) ? 'https://doi.org/' . $sCheckDoi : $sDolink;
|
||
|
||
//数据处理更新数据库
|
||
$aUpdate = [];
|
||
if(!empty($sTitle) && empty($aRefer['title'])){
|
||
$aUpdate['title'] = $sTitle;
|
||
}
|
||
if(!empty($sJoura) && empty($aRefer['joura'])){
|
||
$aUpdate['joura'] = $sJoura;
|
||
}
|
||
if(!empty($sAuthor) && empty($aRefer['author'])){
|
||
$aUpdate['author'] = $sAuthor;
|
||
}
|
||
if(!empty($sDateno)){
|
||
$aUpdate['dateno'] = $sDateno;
|
||
}
|
||
if($bIsRetracted == 1){
|
||
$aUpdate['is_retracted'] = 1;
|
||
}
|
||
if(!empty($sDolink) && empty($aRefer['doilink'])){
|
||
$aUpdate['doilink'] = $sDolink;
|
||
}
|
||
if(empty($aUpdate)){
|
||
return json_encode(['status' => 5,'msg' => 'No update information available']);
|
||
}
|
||
$aUpdate['update_time'] = time();
|
||
$aUpdate['cs'] = 1;
|
||
$aWhere = ['p_refer_id' => $iPReferId];
|
||
$result = Db::name('production_article_refer')->where($aWhere)->limit(1)->update($aUpdate);
|
||
if($result === false){
|
||
return json_encode(['status' => 6,'msg' => 'Update failed-Cs']);
|
||
}
|
||
return json_encode(['status' => 1,'msg' => 'Update successful']);
|
||
}
|
||
|
||
/**
|
||
* 过滤非法DOI（仅保留10.xxxx/xxx格式）
|
||
*/
|
||
private function filterValidDoi($doi = ''){
|
||
$doi = trim($doi);
|
||
if (empty($doi)) return '';
|
||
// 正则匹配：10. + 至少4位数字 + / + 任意字符
|
||
if (preg_match('/^10\.\d{4,}\/.+/', $doi)) {
|
||
return $doi;
|
||
}
|
||
return '';
|
||
}
|
||
|
||
/**
|
||
* 单DOI查询
|
||
*/
|
||
private function fetchSingleDoiWithRetry($doi){
|
||
$retryCount = 0;
|
||
while ($retryCount < $this->maxRetry) {
|
||
$url = $this->crossrefUrl. rawurlencode($doi) . "?mailto=" . rawurlencode($this->mailto);
|
||
|
||
$ch = curl_init();
|
||
curl_setopt($ch, CURLOPT_URL, $url);
|
||
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
|
||
curl_setopt($ch, CURLOPT_TIMEOUT, $this->timeout);
|
||
curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
|
||
curl_setopt($ch, CURLOPT_HTTPHEADER, [
|
||
"User-Agent: DOI-Fetcher/1.0 (mailto:{$this->mailto})"
|
||
]);
|
||
$response = curl_exec($ch);
|
||
$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
|
||
curl_close($ch);
|
||
|
||
// 成功返回
|
||
if ($httpCode == 200) {
|
||
$data = json_decode($response, true);
|
||
return $data['status'] == 'ok' ? $data['message'] : null;
|
||
}
|
||
|
||
// 429速率限制：延长等待后重试
|
||
if ($httpCode == 429) {
|
||
sleep(5);
|
||
$retryCount++;
|
||
continue;
|
||
}
|
||
|
||
$retryCount++;
|
||
sleep(1); // 普通失败，1秒后重试
|
||
}
|
||
return null;
|
||
}
|
||
|
||
/**
|
||
* 提取标题
|
||
*/
|
||
private function getTitle($aDoiInfo = []){
|
||
return $aDoiInfo['title'][0] ?? '';
|
||
}
|
||
|
||
/**
|
||
* 提取出版社名
|
||
*/
|
||
private function getPublisher($aDoiInfo = []){
|
||
$aJournal = [
|
||
'title' => isset($aDoiInfo['container-title'][0]) ? $aDoiInfo['container-title'][0] : '',
|
||
'short_title'=> isset($aDoiInfo['short-container-title'][0]) ? $aDoiInfo['short-container-title'][0] : '',
|
||
'ISSN' => $aDoiInfo['ISSN'] ?? [],
|
||
'publisher' => $aDoiInfo['publisher'] ?? '',
|
||
];
|
||
return $aJournal;
|
||
}
|
||
|
||
/**
|
||
* 提取作者
|
||
*/
|
||
private function getAuthors($aDoiInfo = []){
|
||
$authors = [];
|
||
if (!empty($aDoiInfo['author'])) {
|
||
foreach ($aDoiInfo['author'] as $author) {
|
||
$name = $author['family'] ?? '';
|
||
if (!empty($author['given'])) {
|
||
$name = $author['given'] . ' ' . $name;
|
||
}
|
||
if (!empty($name)) {
|
||
$authors[] = $name;
|
||
}
|
||
}
|
||
}
|
||
return $authors;
|
||
}
|
||
|
||
/**
|
||
* 提取发表年份
|
||
*/
|
||
private function getPublishYear($aDoiInfo = []){
|
||
if (!empty($aDoiInfo['issued']['date-parts'][0][0])) {
|
||
return (string)$aDoiInfo['issued']['date-parts'][0][0];
|
||
}
|
||
return '';
|
||
}
|
||
|
||
/**
|
||
* 提取卷(期):起始页-终止页（格式：2024;10(2):100-120）
|
||
*/
|
||
private function getVolumeIssuePages($aDoiInfo = []){
|
||
$parts = [];
|
||
// 年
|
||
$year = $this->getPublishYear($aDoiInfo);
|
||
if ($year) $parts[] = $year;
|
||
|
||
// 卷(期)
|
||
$volume = $aDoiInfo['volume'] ?? '';
|
||
$issue = $aDoiInfo['issue'] ?? '';
|
||
if ($volume) {
|
||
$volumeIssue = $volume . ($issue ? "({$issue})" : '');
|
||
$parts[] = $volumeIssue;
|
||
}
|
||
|
||
// 起始页-终止页
|
||
$pageStart = $aDoiInfo['page']['start'] ?? ($aDoiInfo['first-page'] ?? '');
|
||
$pageEnd = $aDoiInfo['page']['end'] ?? ($aDoiInfo['last-page'] ?? '');
|
||
$pages = '';
|
||
if ($pageStart) {
|
||
$pages = $pageStart . ($pageEnd ? "-{$pageEnd}" : '');
|
||
}else{
|
||
$pages = $aDoiInfo['page'] ?? '';
|
||
}
|
||
if ($pages) $parts[] = $pages;
|
||
|
||
return implode(':', $parts);
|
||
}
|
||
|
||
/**
|
||
* 识别撤稿文章
|
||
*/
|
||
private function checkRetracted($aDoiInfo = []){
|
||
$isRetracted = false;
|
||
$reason = "未撤稿";
|
||
|
||
// 1. 文章类型为撤稿声明type/subtype
|
||
$sType = strtolower($aDoiInfo['type'] ?? '');
|
||
$sSubtype = strtolower($aDoiInfo['subtype'] ?? '');
|
||
if (isset($sType) && in_array($sType, ['retraction', 'correction'])) {
|
||
$isRetracted = true;
|
||
$reason = "文章类型为{$sType}（撤稿/更正声明）";
|
||
}
|
||
|
||
if (isset($sSubtype) && in_array($sSubtype, ['retraction', 'correction'])) {
|
||
$isRetracted = true;
|
||
$reason = "文章类型为{$sSubtype}（撤稿/更正声明）";
|
||
}
|
||
// 2. update-type包含撤稿
|
||
if (isset($aDoiInfo['update-type']) && in_array('retraction', $aDoiInfo['update-type'])) {
|
||
$isRetracted = true;
|
||
$reason = "官方标记为撤稿（update-type: retraction）";
|
||
}
|
||
|
||
// 3. 关联撤稿文章
|
||
if (isset($aDoiInfo['relation']) && !empty($aDoiInfo['relation'])) {
|
||
foreach ($aDoiInfo['relation'] as $relType => $relItems) {
|
||
if (in_array($relType, ['is-retraction-of', 'corrects'])) {
|
||
$isRetracted = true;
|
||
$relatedDoi = $relItems[0]['id'] ?? '未知';
|
||
$reason = "关联撤稿文章{$relatedDoi}（关系：{$relType}）";
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
|
||
// 4. update-to 字段
|
||
if (isset($aDoiInfo['update-to']) && is_array($aDoiInfo['update-to'])) {
|
||
foreach ($aDoiInfo['update-to'] as $update) {
|
||
$updateType = strtolower($update['type'] ?? '');
|
||
$updateLabel = strtolower($update['label'] ?? '');
|
||
if (strpos($updateType, 'retract') !== false || strpos($updateLabel, 'retract') !== false) {
|
||
$isRetracted = true;
|
||
$retractionDetail['retraction_notice'] = [
|
||
'type' => $update['type'] ?? '',
|
||
'label' => $update['label'] ?? '',
|
||
'DOI' => $update['DOI'] ?? '',
|
||
'date' => isset($update['updated']) ? $this->parseDateParts($update['updated']) : '',
|
||
];
|
||
break;
|
||
}
|
||
}
|
||
}
|
||
//5.title 关键词
|
||
$aTitles = $aDoiInfo['title'] ?? [];
|
||
foreach ($aTitles as $value) {
|
||
$sTitleLower = strtolower($value);
|
||
if (strpos($sTitleLower, 'retraction') !== false || strpos($sTitleLower, 'retracted') !== false
|
||
|| strpos($sTitleLower, 'withdrawal') !== false || strpos($sTitleLower, 'withdrawn') !== false) {
|
||
$isRetracted = true;
|
||
$retractionDetail['title_keyword'] = $value;
|
||
break;
|
||
}
|
||
}
|
||
return [
|
||
'is_retracted' => $isRetracted,
|
||
'reason' => $reason
|
||
];
|
||
}
|
||
/**
|
||
* 识别doi链接
|
||
*/
|
||
private function getDolink($aDoiInfo = []){
|
||
return $aDoiInfo['URL'] ?? '';
|
||
}
|
||
} |