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

参考文献模型校对换阿里云百炼，完善各种校对细节

Browse Source
This commit is contained in:
wyn
2026-08-06 10:19:57 +08:00
parent bffac7755d 5d40b8358c
commit 16961465ac
7 changed files with 530 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files

169
application/common/PubmedService.php
View File

@@ -18,13 +18,43 @@ class PubmedService
private $timeout = 20;
private $tool = 'tmrjournals';
private $email = '';
// NCBI 限流：无 key 时 3 次/秒，配 key 后 10 次/秒。多 worker 并行处理参考文献时必须配。
private $apiKey = '';
private $maxRetry = 3;
public function __construct(array $config = [])
{
$this->email = (string)$this->envGet('pubmed_email', '');
$this->apiKey = (string)$this->envGet('pubmed_api_key', '');
$tool = trim((string)$this->envGet('pubmed_tool', ''));
if ($tool !== '') $this->tool = $tool;
if (isset($config['base'])) $this->base = rtrim((string)$config['base'], '/') . '/';
if (isset($config['timeout'])) $this->timeout = max(5, intval($config['timeout']));
if (isset($config['tool'])) $this->tool = (string)$config['tool'];
if (isset($config['email'])) $this->email = (string)$config['email'];
if (!empty($config['tool'])) $this->tool = (string)$config['tool'];
if (!empty($config['email'])) $this->email = (string)$config['email'];
if (!empty($config['api_key'])) $this->apiKey = (string)$config['api_key'];
if (isset($config['max_retry'])) $this->maxRetry = max(1, intval($config['max_retry']));
}
private function envGet($key, $default = '')
{
if (!class_exists('\think\Env')) {
return $default;
}
return \think\Env::get($key, $default);
}
/**
* tool/email/api_key 是 NCBI 要求的调用方标识，缺失会被更严格限流
*/
private function commonParams(): array
{
$params = ['tool' => $this->tool, 'email' => $this->email];
if ($this->apiKey !== '') {
$params['api_key'] = $this->apiKey;
}
return $params;
}
/**
@@ -60,18 +90,18 @@ class PubmedService
$pmid = trim($pmid);
if ($pmid === '') return null;
// v3：解析结果新增 language / journal_country / affiliations，换 key 避免命中旧缓存
// v3：解析结果新增 authors / volume / issue / pages / doilanguage / journal_country / affiliations换 key 避免命中旧缓存
$cacheKey = 'pmid_v3_' . $pmid;
$cached = $this->cacheGet($cacheKey, 30 * 86400);
if (is_array($cached)) return $cached;
$url = $this->base . 'efetch.fcgi?' . http_build_query([
$url = $this->base . 'efetch.fcgi?' . http_build_query(array_merge([
'db' => 'pubmed',
'id' => $pmid,
'retmode' => 'xml',
'tool' => $this->tool,
'email' => $this->email,
]);
], $this->commonParams()));
$xml = $this->httpGet($url);
if (!is_string($xml) || trim($xml) === '') return null;
@@ -283,6 +313,11 @@ class PubmedService
return null;
}
$doi = $this->xpText($xp, '//PubmedArticle//ArticleIdList/ArticleId[@IdType="doi"]');
if ($doi === '') {
$doi = $this->xpText($xp, '//PubmedArticle//ELocationID[@EIdType="doi"]');
}
return [
'title' => $title,
'abstract' => $abstract,
@@ -296,9 +331,133 @@ class PubmedService
'languages' => $languages,
'journal_country' => $journalCountry,
'affiliations' => $affiliations,
'authors' => $this->parseAuthors($xp),
'volume' => $this->xpText($xp, '//PubmedArticle//JournalIssue//Volume'),
'issue' => $this->xpText($xp, '//PubmedArticle//JournalIssue//Issue'),
'pages' => $this->parsePagination($xp),
'doi' => $doi,
];
}
/**
* 作者列表：LastName/ForeName/Initials，机构作者用 CollectiveName
*/
private function parseAuthors(\DOMXPath $xp): array
{
$out = [];
$nodes = $xp->query('//PubmedArticle//AuthorList/Author');
if (!$nodes) {
return $out;
}
foreach ($nodes as $n) {
$family = $given = $initials = $collective = '';
foreach ($n->childNodes as $c) {
switch ($c->nodeName) {
case 'LastName':
$family = trim($c->textContent);
break;
case 'ForeName':
$given = trim($c->textContent);
break;
case 'Initials':
$initials = trim($c->textContent);
break;
case 'CollectiveName':
$collective = trim($c->textContent);
break;
}
}
if ($family === '' && $given === '' && $collective === '') {
continue;
}
$out[] = [
'family' => $family,
'given' => $given,
'initials' => $initials,
'collective' => $collective,
];
}
return $out;
}
/**
* 页码：StartPage/EndPage → MedlinePgn → 电子刊文章号（ELocationID pii）
*
* MedlinePgn 用 NLM 缩写式尾页（210-8 表示 210–218），展开逻辑由调用方处理。
*/
private function parsePagination(\DOMXPath $xp): string
{
$start = $this->xpText($xp, '//PubmedArticle//Pagination//StartPage');
if ($start !== '') {
$end = $this->xpText($xp, '//PubmedArticle//Pagination//EndPage');
return $end !== '' ? $start . '-' . $end : $start;
}
$medlinePgn = $this->xpText($xp, '//PubmedArticle//Pagination//MedlinePgn');
if ($medlinePgn !== '') {
return $medlinePgn;
}
// 无连续页码的电子刊（PLOS 等）以文章号著录；pii 也可能是出版社流水号，故只认 e12345/12345
$pii = $this->xpText($xp, '//PubmedArticle//ELocationID[@EIdType="pii"]');
return preg_match('/^e?\d+$/i', $pii) === 1 ? $pii : '';
}
/**
* 引用格式作者串：姓全写 + 名首字母，超过 $maxAuthors 个取前 N 个 + et al
* 例：Smith JA, Jones B, Lee C, et al
*/
public function authorsCitation(array $authors, int $maxAuthors = 3): string
{
$list = [];
foreach ($authors as $a) {
$collective = trim((string)($a['collective'] ?? ''));
if ($collective !== '') {
$list[] = $collective;
continue;
}
$family = trim((string)($a['family'] ?? ''));
if ($family === '') {
continue;
}
$initials = trim((string)($a['initials'] ?? ''));
if ($initials === '') {
$initials = $this->givenToInitials((string)($a['given'] ?? ''));
}
$list[] = $initials !== '' ? $family . ' ' . $initials : $family;
}
if (empty($list)) {
return '';
}
$maxAuthors = max(1, $maxAuthors);
if (count($list) > $maxAuthors) {
return implode(', ', array_slice($list, 0, $maxAuthors)) . ', et al';
}
return implode(', ', $list);
}
private function givenToInitials($given): string
{
$given = trim((string)$given);
if ($given === '') {
return '';
}
$parts = preg_split('/[\s\-\.]+/u', $given, -1, PREG_SPLIT_NO_EMPTY);
$initials = '';
foreach ($parts as $p) {
$first = mb_substr($p, 0, 1);
if ($first !== '') {
$initials .= mb_strtoupper($first);
}
}
return $initials;
}
private function xpText(\DOMXPath $xp, string $query): string
{
$n = $xp->query($query);