作者堆叠 专刊

This commit is contained in:
wyn
2026-07-29 17:05:42 +08:00
parent f583f67e24
commit 1ee795b6b9
10 changed files with 2833 additions and 141 deletions

View File

@@ -10,6 +10,7 @@ use PhpOffice\PhpWord\IOFactory;
use app\common\OpenAi; use app\common\OpenAi;
use app\common\CrossrefService; use app\common\CrossrefService;
use app\common\PubmedService; use app\common\PubmedService;
use app\common\ArticleParserService;
/** /**
* @title 文章接口 * @title 文章接口
@@ -1181,6 +1182,74 @@ class Article extends Base
return jsonSuccess($re); return jsonSuccess($re);
} }
/**
* @title 判断作者是否已按审稿意见修改(逐条核对最终版)
* @description 拆分审稿意见条目,结合修回稿与回复信,用 LLM 逐条判定 addressed/partial/not可缓存
* @param name:article_id type:int require:1 desc:文章id
* @param name:force type:int require:0 desc:1强制重跑忽略缓存
* @param name:mode type:string require:0 desc:full=逐条核对(默认) quick=仅时间门
*
* @url /api/Article/checkAuthorRevisedByReview
* @method POST
*/
public function checkAuthorRevisedByReview()
{
$data = $this->request->post();
$rule = new Validate([
'article_id' => 'require|number'
]);
if (!$rule->check($data)) {
return jsonError($rule->getError());
}
$articleId = intval($data['article_id']);
$mode = strtolower(trim((string)($data['mode'] ?? 'full')));
$force = !empty($data['force']) ? 1 : 0;
try {
$service = new \app\common\service\RevisionCommentMatchService();
if ($mode === 'quick') {
return jsonSuccess($service->quickCheck($articleId));
}
$result = $service->check($articleId, [
'force' => $force,
'use_cache' => $force ? 0 : 1,
]);
$result['mode'] = 'full';
return jsonSuccess($result);
} catch (\Exception $e) {
return jsonError($e->getMessage());
}
}
/**
* @title 获取最近一次逐条修回核对结果
* @param name:article_id type:int require:1 desc:文章id
* @url /api/Article/getRevisionCommentCheckResult
* @method POST
*/
public function getRevisionCommentCheckResult()
{
$data = $this->request->post();
$rule = new Validate([
'article_id' => 'require|number'
]);
if (!$rule->check($data)) {
return jsonError($rule->getError());
}
try {
$result = (new \app\common\service\RevisionCommentMatchService())
->getLatestResult(intval($data['article_id']));
if (empty($result)) {
return jsonError('暂无检测结果,请先调用 checkAuthorRevisedByReview');
}
return jsonSuccess($result);
} catch (\Exception $e) {
return jsonError($e->getMessage());
}
}
/**获取用户所投的文章 /**获取用户所投的文章
* @return \think\response\Json|void * @return \think\response\Json|void
@@ -3221,11 +3290,34 @@ class Article extends Base
} }
/** /**
* 获取文章作者信息
* 支持 article_id查库或 file_url解析稿件不读写库
* @return void * @return void
*/ */
public function getAuthors() public function getAuthors()
{ {
$data = $this->request->post(); $data = $this->request->post();
// 从稿件解析作者(对齐 Contribute 读取逻辑,不入库)
if (!empty($data['file_url'])) {
$sFileUrl = rtrim(ROOT_PATH, '/') . '/public/' . ltrim(ltrim($data['file_url'], '/'), 'public');
if (!file_exists($sFileUrl)) {
return jsonError('The uploaded file does not exist');
}
if (!is_readable($sFileUrl)) {
return jsonError('The uploaded file is unreadable');
}
$aDealData = json_decode(ArticleParserService::uploadAndParse($sFileUrl), true);
$iStatus = empty($aDealData['status']) ? 0 : $aDealData['status'];
if ($iStatus != 1) {
return jsonError(empty($aDealData['msg']) ? 'Content parsing failed' : $aDealData['msg']);
}
$aParseData = empty($aDealData['data']) ? [] : $aDealData['data'];
$re['authors'] = $this->buildAuthorsFromParseData($aParseData);
return jsonSuccess($re);
}
$rule = new Validate([ $rule = new Validate([
"article_id" => "require" "article_id" => "require"
]); ]);
@@ -3946,6 +4038,12 @@ class Article extends Base
if (isset($data['is_agree'])) { if (isset($data['is_agree'])) {
$update_l['is_agree'] = $data['is_agree']; $update_l['is_agree'] = $data['is_agree'];
} }
if (isset($data['special_num'])) {
$update_l['special_num'] = intval($data['special_num']);
}
if (isset($data['special_title'])) {
$update_l['special_title'] = trim((string)$data['special_title']);
}
if(!empty($sArticleSn)){ if(!empty($sArticleSn)){
$update_l['accept_sn'] = $sArticleSn; $update_l['accept_sn'] = $sArticleSn;
@@ -5737,7 +5835,9 @@ class Article extends Base
// if(empty($iJournalId)){ // if(empty($iJournalId)){
// return json_encode(['status' => 2,'msg' => 'Please select a journal']); // return json_encode(['status' => 2,'msg' => 'Please select a journal']);
// } // }
$aArticleInsert = ['journal_id' => $iJournalId,'title' => $sTitle,'state' => -1,'user_id' => $iUserId]; $aArticleInsert = ['journal_id' => $iJournalId,'title' => $sTitle,'abstrart' => '','use_ai_explain' => '','state' => -1,'user_id' => $iUserId];
if(isset($aParam['special_num']))$aArticleInsert['special_num'] = $aParam['special_num'];
if(isset($aParam['special_title']))$aArticleInsert['special_title'] = $aParam['special_title'];
$aArticleInsert['is_use_ai'] = 3; $aArticleInsert['is_use_ai'] = 3;
$aArticleInsert['is_figure_copyright'] = 3; $aArticleInsert['is_figure_copyright'] = 3;
// $aArticleInsert['is_transfer'] = 3; // $aArticleInsert['is_transfer'] = 3;
@@ -5785,7 +5885,7 @@ class Article extends Base
if($becomeRev == false){ if($becomeRev == false){
$aParam['is_become_reviewer'] = 2; $aParam['is_become_reviewer'] = 2;
} }
$aField = ['is_use_ai','use_ai_explain','is_figure_copyright','is_become_reviewer','approval','approval_file','approval_content','code','is_become_reviewer','is_agree','title','abstrart','keywords','topics','fund','type','journal_id'];//,'title','abstrart','keywords','topics','fund','type','is_transfer', $aField = ['is_use_ai','use_ai_explain','is_figure_copyright','is_become_reviewer','approval','approval_file','approval_content','code','is_become_reviewer','is_agree','title','abstrart','keywords','topics','fund','type','journal_id','special_num','special_title'];//,'title','abstrart','keywords','topics','fund','type','is_transfer',
$sMsg = ''; $sMsg = '';
$iIsUpdate = 1; $iIsUpdate = 1;
foreach ($aField as $key => $value) { foreach ($aField as $key => $value) {
@@ -6412,9 +6512,65 @@ class Article extends Base
// $update_result = Db::name('article')->where($aWhere)->limit(1)->update($aArticleUpdate); // $update_result = Db::name('article')->where($aWhere)->limit(1)->update($aArticleUpdate);
// } // }
//操作日志 //操作日志
$aLog = ['article_id' => $iArticleId,'user_id' => $iUserId,'type' => 7,'create_time' => time(),'content' => $sUserAccount . ':Operating the article copyright statement','is_view' => 1]; $aLog = ['article_id' => $iArticleId,'user_id' => $iUserId,'type' => 7,'create_time' => time(),'update_time' => time(),'content' => $sUserAccount . ':Operating the article copyright statement','is_view' => 1];
Db::name('user_act_log')->insert($aLog); Db::name('user_act_log')->insert($aLog);
Db::commit(); Db::commit();
return json_encode(['status' => 1,'msg' => 'success']); return json_encode(['status' => 1,'msg' => 'success']);
} }
/**
* 修改未发表文章对应专刊(仅更新 tougao 本地 t_article
* special_num = journal_special_idspecial_title 由前端一并传入
*/
public function changeArticleSpecialForSubmit()
{
$data = $this->request->post();
$rule = new Validate([
'article_id' => 'require',
/*'journal_special_id' => 'require',
'special_title' => 'require',*/
]);
if (!$rule->check($data)) {
return jsonError($rule->getError());
}
$article_info = $this->article_obj->where('article_id', $data['article_id'])->find();
if (!$article_info) {
return jsonError('article not found');
}
$journalSpecialRaw = isset($data['journal_special_id'])&&$data['journal_special_id'] ? trim((string)$data['journal_special_id']) : '';
// journal_special_id 为空:清空专刊信息
if ($journalSpecialRaw === '') {
$update = [
'special_num' => 0,
'special_title' => '',
];
$this->article_obj->where('article_id', $data['article_id'])->update($update);
return jsonSuccess([
'article_id' => intval($data['article_id']),
'special_num' => 0,
'special_title' => '',
]);
}
// journal_special_id 有值special_title 必填
$specialTitle = isset($data['special_title']) ? trim((string)$data['special_title']) : '';
if ($specialTitle === '') {
return jsonError('special_title is required');
}
$specialId = intval($journalSpecialRaw);
$update = [
'special_num' => $specialId,
'special_title' => $specialTitle,
];
$this->article_obj->where('article_id', $data['article_id'])->update($update);
return jsonSuccess([
'article_id' => intval($data['article_id']),
'special_num' => $specialId,
'special_title' => $update['special_title'],
]);
}
} }

View File

@@ -25,6 +25,21 @@ class References extends Base
public function __construct(\think\Request $request = null) { public function __construct(\think\Request $request = null) {
parent::__construct($request); parent::__construct($request);
} }
/**
* 参考文献堆叠统计可视化页
* 访问:/api/References/index?p_article_id=xxx
*/
public function index()
{
$pArticleId = intval($this->request->param('p_article_id', 0));
$this->assign([
'api_stats_js' => json_encode((string) url('api/References/referenceStackingStats'), JSON_UNESCAPED_SLASHES),
'init_p_article_id' => $pArticleId,
]);
return $this->fetch('references/index');
}
//OPENAI token //OPENAI token
private $sApiKey = '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'; private $sApiKey = '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';
//OPENAI URL //OPENAI URL
@@ -1847,6 +1862,8 @@ class References extends Base
/** /**
* 参考文献引用堆叠统计(同作者>15%、同刊>20%、自引>10%,实时计算) * 参考文献引用堆叠统计(同作者>15%、同刊>20%、自引>10%,实时计算)
* 仅读本地库,不实时请求 Crossref/OpenAlex。
* 额外返回 author_data_issues未获取到作者 / 作者信息不全的参考文献列表
* *
* POST/GET: p_article_id必填 * POST/GET: p_article_id必填
*/ */
@@ -1871,6 +1888,66 @@ class References extends Base
} }
} }
/**
* 参考文献堆叠摘要(展示用)
* 返回:同作者名称+比例、同刊名称+比例、自引比例(仅超阈值的同作者/同刊)
*
* POST/GET: p_article_id必填
*/
public function referenceStackingSummary()
{
$aParam = $this->request->post();
if (empty($aParam)) {
$aParam = $this->request->param();
}
$iPArticleId = empty($aParam['p_article_id']) ? 0 : intval($aParam['p_article_id']);
if ($iPArticleId <= 0) {
return json_encode(array('status' => 2, 'msg' => 'Please select an article'));
}
try {
$svc = new ReferenceStackingStatsService();
$result = $svc->getStackingSummaryByPArticleId($iPArticleId);
return jsonSuccess($result);
} catch (\Exception $e) {
return jsonError($e->getMessage());
}
}
/**
* 获取单条参考文献作者信息:先 Crossref/OpenAlex或兜底 refer.author同步入库再返回明细
*
* POST/GET: p_article_id必填, p_refer_id必填
*/
public function referenceReferAuthors()
{
$aParam = $this->request->post();
if (empty($aParam)) {
$aParam = $this->request->param();
}
$iPArticleId = empty($aParam['p_article_id']) ? 0 : intval($aParam['p_article_id']);
$iPReferId = empty($aParam['p_refer_id']) ? 0 : intval($aParam['p_refer_id']);
if ($iPArticleId <= 0) {
return json_encode(array('status' => 2, 'msg' => 'Please select an article'));
}
if ($iPReferId <= 0) {
return jsonError('p_refer_id is required');
}
try {
$svc = new ReferenceReferAuthorService();
$result = $svc->fetchAuthorsByPReferId($iPArticleId, $iPReferId);
if (empty($result['refer'])) {
return jsonError('Reference not found');
}
return jsonSuccess($result);
} catch (\Exception $e) {
return jsonError($e->getMessage());
}
}
/** /**
* 测试:同步参考文献作者明细到 t_production_article_refer_author * 测试:同步参考文献作者明细到 t_production_article_refer_author
* POST/GET: p_article_id必填, p_refer_id可选仅同步单条, include_authors可选默认 01 返回作者明细), sleep_ms可选默认 120 * POST/GET: p_article_id必填, p_refer_id可选仅同步单条, include_authors可选默认 01 返回作者明细), sleep_ms可选默认 120

File diff suppressed because it is too large Load Diff

View File

@@ -2081,7 +2081,149 @@ class User extends Base
$re['user'] = $user_info; $re['user'] = $user_info;
return jsonSuccess($re); return jsonSuccess($re);
} }
public function createUserForEditor(){
$data = $this->request->post();
$rule = new Validate([
"email"=>"require",
"address"=>"require",
"firstname"=>"require",
"lastname"=>"require",
"intro"=>"require"
]);
if(!$rule->check($data)){
return jsonError($rule->getError());
}
$account = isset($data['account'])?$data['account']:$data['email'];
$email = $data['email'];
$res_once = $this->user_obj->where("account='$account' or email = '$email'")->find();
if ($res_once != null) {
return json('existence');
}
Db::startTrans();
$inser_data['account'] = trim($account);
$inser_data['password'] = md5(isset($data['password'])?$data['password']:"123456qwe");
$inser_data['email'] = $email;
if(isset($data['phone']))$inser_data['phone'] = $data['phone'];
if(isset($data['orcid']))$inser_data['orcid'] = $data['orcid'];
$inser_data['realname'] = $data['firstname']." ".$data['lastname'];
$inser_data['icon'] = $data['icon'];
$inser_data['ctime'] = time();
$inser_data['openid'] = "";
$id = $this->user_obj->insertGetId($inser_data);
//存入个人额外信息
$insert_reviewer['reviewer_id'] = $id;
$insert_reviewer['test_from'] = "kzeditor";
$insert_reviewer['address'] = $data['address'];
$insert_reviewer['introduction'] = $data['intro'];
$insert_reviewer['firstname'] = $data['firstname'];
$insert_reviewer['lastname'] = $data['lastname'];
if(isset($data['website']))$insert_reviewer['website'] = $data['website'];
if(isset($data['interests']))$insert_reviewer['interests'] = $data['interests'];
$r_res = $this->user_reviewer_info_obj->insert($insert_reviewer);
if($id&&$r_res){
Db::commit();
return jsonSuccess($id);
}else{
Db::rollback();
return jsonError("system error");
}
}
/**
* 修改客座编辑本地资料
* t_user: orcid、phone、realname(firstname lastname)、icon
* t_user_reviewer_info: address、interests、firstname、lastname、introduction(intro)
*/
public function updateUserForEditor()
{
$data = $this->request->post();
$rule = new Validate([
'user_id' => 'require|number',
'firstname' => 'require',
'lastname' => 'require',
'intro' => 'require',
]);
if (!$rule->check($data)) {
return jsonError($rule->getError());
}
$userId = intval($data['user_id']);
$user = $this->user_obj->where('user_id', $userId)->where('state', 0)->find();
if (!$user) {
return jsonError('user not found');
}
Db::startTrans();
try {
$firstName = trim((string)$data['firstname']);
$lastName = trim((string)$data['lastname']);
$userUpdate = [
'realname' => trim($firstName . ' ' . $lastName),
];
if (isset($data['orcid'])) {
$userUpdate['orcid'] = trim((string)$data['orcid']);
}
if (isset($data['phone'])) {
$userUpdate['phone'] = trim((string)$data['phone']);
}
if (isset($data['icon'])) {
$userUpdate['icon'] = trim((string)$data['icon']);
}
$upUser = $this->user_obj->where('user_id', $userId)->update($userUpdate);
if ($upUser === false) {
throw new \Exception('update t_user failed');
}
$reviewerInfo = $this->user_reviewer_info_obj
->where('reviewer_id', $userId)
->where('state', 0)
->find();
$reviewerUpdate = [
'firstname' => $firstName,
'lastname' => $lastName,
];
if (isset($data['address'])) {
$reviewerUpdate['address'] = trim((string)$data['address']);
}
if (isset($data['website'])) {
$reviewerUpdate['website'] = trim((string)$data['website']);
}
if (isset($data['interests'])) {
$reviewerUpdate['interests'] = trim((string)$data['interests']);
}
$reviewerUpdate['introduction'] = trim((string)$data['intro']);
if ($reviewerInfo) {
$upReviewer = $this->user_reviewer_info_obj
->where('reviewer_id', $userId)
->where('state', 0)
->update($reviewerUpdate);
if ($upReviewer === false) {
throw new \Exception('update t_user_reviewer_info failed');
}
} else {
$reviewerUpdate['reviewer_id'] = $userId;
$reviewerUpdate['test_from'] = 'kzeditor';
$insReviewer = $this->user_reviewer_info_obj->insert($reviewerUpdate);
if (!$insReviewer) {
throw new \Exception('insert t_user_reviewer_info failed');
}
}
Db::commit();
return jsonSuccess([
'user_id' => $userId,
'updated' => 1,
]);
} catch (\Exception $e) {
Db::rollback();
return jsonError($e->getMessage());
}
}
/** /**
* 注册功能 * 注册功能
*/ */
@@ -3370,4 +3512,17 @@ class User extends Base
} }
return implode("\n", $lines); return implode("\n", $lines);
} }
public function up_editorIcon_file()
{
$file = request()->file('icon');
if ($file) {
$info = $file->move(ROOT_PATH . 'public' . DS . 'usericon');
if ($info) {
return json(['code' => 0, 'upurl' => str_replace("\\", "/", $info->getSaveName())]);
} else {
return json(['code' => 1, 'msg' => $file->getError()]);
}
}
}
} }

View File

@@ -0,0 +1,674 @@
<!DOCTYPE html>
<html lang="zh-CN">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<title>参考文献堆叠统计</title>
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link href="https://fonts.googleapis.com/css2?family=DM+Sans:ital,opsz,wght@0,9..40,400;0,9..40,500;0,9..40,600;0,9..40,700;1,9..40,400&family=Fraunces:opsz,wght@9..144,500;9..144,650&display=swap" rel="stylesheet" />
<style>
:root {
--ink: #14212b;
--muted: #5c6b76;
--line: #d7e0e6;
--accent: #0f766e;
--accent-soft: #d9f3ef;
--warn: #b45309;
--danger: #b91c1c;
--ok: #047857;
--font: "DM Sans", system-ui, sans-serif;
--display: "Fraunces", Georgia, serif;
}
* { box-sizing: border-box; }
html, body { margin: 0; height: 100%; }
body {
font-family: var(--font);
color: var(--ink);
line-height: 1.45;
background:
linear-gradient(160deg, rgba(15, 118, 110, 0.18), transparent 42%),
linear-gradient(340deg, rgba(20, 33, 43, 0.08), transparent 40%),
#cfd8df;
display: grid;
place-items: center;
padding: 16px;
overflow: hidden;
}
.window {
width: min(520px, 100%);
height: min(680px, calc(100vh - 32px));
background: #fff;
border-radius: 12px;
border: 1px solid rgba(20, 33, 43, 0.12);
box-shadow:
0 24px 48px rgba(20, 33, 43, 0.18),
0 2px 0 rgba(255, 255, 255, 0.5) inset;
display: flex;
flex-direction: column;
overflow: hidden;
}
.titlebar {
flex: 0 0 auto;
display: flex;
align-items: center;
justify-content: space-between;
gap: 10px;
padding: 10px 12px 10px 14px;
background: linear-gradient(180deg, #f7fafb, #eef3f6);
border-bottom: 1px solid var(--line);
cursor: default;
user-select: none;
}
.titlebar-left {
display: flex;
align-items: center;
gap: 10px;
min-width: 0;
}
.traffic {
display: flex;
gap: 6px;
}
.traffic i {
width: 10px;
height: 10px;
border-radius: 50%;
display: block;
}
.traffic .r { background: #ff5f57; }
.traffic .y { background: #febc2e; }
.traffic .g { background: #28c840; }
.titlebar h1 {
margin: 0;
font-family: var(--display);
font-size: 0.98rem;
font-weight: 650;
white-space: nowrap;
overflow: hidden;
text-overflow: ellipsis;
}
.titlebar .sub {
font-size: 0.72rem;
color: var(--muted);
white-space: nowrap;
}
.toolbar {
flex: 0 0 auto;
padding: 10px 12px;
border-bottom: 1px solid var(--line);
background: #fafcfd;
}
.start-grid {
display: grid;
grid-template-columns: 1fr auto;
gap: 8px;
align-items: end;
}
label {
display: block;
font-size: 0.72rem;
font-weight: 600;
color: var(--muted);
margin-bottom: 4px;
}
input[type="number"] {
width: 100%;
border: 1px solid var(--line);
border-radius: 8px;
padding: 8px 10px;
font: inherit;
font-size: 0.9rem;
color: var(--ink);
background: #fff;
outline: none;
}
input:focus {
border-color: #7bbbb4;
box-shadow: 0 0 0 3px rgba(15, 118, 110, 0.12);
}
.btn {
appearance: none;
border: 0;
border-radius: 8px;
padding: 8px 14px;
font: inherit;
font-size: 0.88rem;
font-weight: 600;
cursor: pointer;
height: 36px;
}
.btn:disabled { opacity: .55; cursor: not-allowed; }
.btn-primary { background: var(--accent); color: #fff; }
.error {
display: none;
margin-top: 8px;
padding: 8px 10px;
border-radius: 8px;
background: #fef2f2;
color: var(--danger);
border: 1px solid #fecaca;
font-size: 0.8rem;
}
.error.show { display: block; }
.body {
flex: 1 1 auto;
overflow: auto;
padding: 12px;
background: #fff;
}
.results { display: none; }
.results.show { display: block; }
.placeholder {
color: var(--muted);
font-size: 0.86rem;
text-align: center;
padding: 36px 12px;
}
.summary {
display: grid;
grid-template-columns: repeat(2, minmax(0, 1fr));
gap: 8px;
margin-bottom: 12px;
}
.stat {
border: 1px solid var(--line);
border-radius: 8px;
padding: 8px 10px;
background: #fafcfd;
}
.stat .k { font-size: 0.7rem; color: var(--muted); }
.stat .v {
font-family: var(--display);
font-size: 1.15rem;
font-weight: 650;
margin-top: 2px;
line-height: 1.2;
}
.stat .sub { font-size: 0.72rem; color: var(--muted); margin-top: 2px; }
.pill {
display: inline-flex;
align-items: center;
padding: 1px 7px;
border-radius: 999px;
font-size: 0.68rem;
font-weight: 650;
}
.pill.ok { background: #dcfce7; color: var(--ok); }
.pill.bad { background: #fee2e2; color: var(--danger); }
.pill.warn { background: #ffedd5; color: var(--warn); }
.section-title {
font-size: 0.88rem;
margin: 14px 0 8px;
font-weight: 700;
display: flex;
align-items: center;
gap: 6px;
flex-wrap: wrap;
}
.section-title .hint {
font-size: 0.72rem;
font-weight: 500;
color: var(--muted);
}
.card-list { display: grid; gap: 8px; }
.item-card {
border: 1px solid var(--line);
border-radius: 8px;
padding: 10px;
background: #fff;
}
.item-card h3 {
margin: 0 0 6px;
font-size: 0.88rem;
font-weight: 650;
}
.meta {
display: flex;
flex-wrap: wrap;
gap: 4px 10px;
color: var(--muted);
font-size: 0.76rem;
margin-bottom: 6px;
}
.meta strong { color: var(--ink); font-weight: 600; }
.ratio-bar {
height: 5px;
border-radius: 999px;
background: #e7eef2;
overflow: hidden;
margin: 6px 0 8px;
}
.ratio-bar > i {
display: block;
height: 100%;
background: linear-gradient(90deg, #0f766e, #14968c);
border-radius: inherit;
}
.ratio-bar.bad > i { background: linear-gradient(90deg, #b91c1c, #dc2626); }
.ref-chips { display: flex; flex-wrap: wrap; gap: 4px; }
.chip {
display: inline-flex;
padding: 1px 6px;
border-radius: 5px;
background: var(--accent-soft);
color: var(--accent);
font-size: 0.7rem;
font-weight: 600;
}
.chip.warn { background: #ffedd5; color: var(--warn); }
.chip.bad { background: #fee2e2; color: var(--danger); }
.ref-list {
margin: 8px 0 0;
padding: 0;
list-style: none;
display: grid;
gap: 4px;
}
.ref-list li {
font-size: 0.74rem;
color: var(--muted);
padding: 6px 8px;
border-radius: 6px;
background: #f7fafb;
border: 1px solid #e8eef2;
max-height: 3.2em;
overflow: hidden;
}
.ref-list .no {
color: var(--accent);
font-weight: 700;
margin-right: 4px;
}
.ref-list a {
color: inherit;
text-decoration: none;
}
.ref-list a:hover {
color: var(--accent);
text-decoration: underline;
}
.chip a {
color: inherit;
text-decoration: none;
}
.chip a:hover { text-decoration: underline; }
.chip.linkable { cursor: pointer; }
.empty {
padding: 12px;
text-align: center;
color: var(--muted);
border: 1px dashed var(--line);
border-radius: 8px;
background: #fafcfd;
font-size: 0.8rem;
}
.note {
margin-top: 6px;
font-size: 0.74rem;
color: var(--muted);
line-height: 1.45;
}
.foot-meta {
margin-top: 12px;
padding-top: 10px;
border-top: 1px solid var(--line);
font-size: 0.72rem;
color: var(--muted);
}
.statusbar {
flex: 0 0 auto;
padding: 6px 12px;
border-top: 1px solid var(--line);
background: #f7fafb;
font-size: 0.7rem;
color: var(--muted);
display: flex;
justify-content: space-between;
gap: 8px;
}
</style>
</head>
<body>
<div class="window" role="dialog" aria-label="参考文献堆叠统计">
<div class="titlebar">
<div class="titlebar-left">
<div class="traffic" aria-hidden="true"><i class="r"></i><i class="y"></i><i class="g"></i></div>
<div>
<h1>参考文献堆叠统计</h1>
<div class="sub">同作者 &gt;15% · 同刊 &gt;20% · 自引 &gt;10%</div>
</div>
</div>
</div>
<div class="toolbar">
<div class="start-grid">
<div>
<label for="pArticleId">p_article_id</label>
<input id="pArticleId" type="number" min="1" placeholder="例如 3649" value="{$init_p_article_id|default=0}" />
</div>
<button class="btn btn-primary" id="btnLoad" type="button">查询</button>
</div>
<div class="error" id="errorBox"></div>
</div>
<div class="body">
<div class="placeholder" id="placeholder">输入 p_article_id 后点击查询</div>
<div class="results" id="results">
<div class="summary" id="summary"></div>
<h2 class="section-title">
同作者堆叠
<span class="hint" id="authorThreshold"></span>
<span class="pill" id="authorPill"></span>
</h2>
<div id="authorSection"></div>
<h2 class="section-title">
同刊堆叠
<span class="hint" id="journalThreshold"></span>
<span class="pill" id="journalPill"></span>
</h2>
<div id="journalSection"></div>
<h2 class="section-title">
自引
<span class="hint" id="selfThreshold"></span>
<span class="pill" id="selfPill"></span>
</h2>
<div id="selfSection"></div>
<h2 class="section-title">
作者数据问题
<span class="hint">缺失 / 不全</span>
</h2>
<div id="issueSection"></div>
<p class="foot-meta" id="footMeta"></p>
</div>
</div>
<div class="statusbar">
<span>Stacking Stats</span>
<span id="statusText">就绪</span>
</div>
</div>
<script>
(function () {
var API_STATS = {$api_stats_js};
var input = document.getElementById('pArticleId');
var btn = document.getElementById('btnLoad');
var errorBox = document.getElementById('errorBox');
var results = document.getElementById('results');
var placeholder = document.getElementById('placeholder');
var statusText = document.getElementById('statusText');
if (Number(input.value) === 0) input.value = '';
function showError(msg) {
errorBox.textContent = msg || '请求失败';
errorBox.classList.add('show');
statusText.textContent = '出错';
}
function clearError() {
errorBox.textContent = '';
errorBox.classList.remove('show');
}
function pct(ratio) {
return (Number(ratio || 0) * 100).toFixed(1) + '%';
}
function escapeHtml(s) {
return String(s == null ? '' : s)
.replace(/&/g, '&amp;')
.replace(/</g, '&lt;')
.replace(/>/g, '&gt;')
.replace(/"/g, '&quot;');
}
function setPill(el, exceeded) {
el.className = 'pill ' + (exceeded ? 'bad' : 'ok');
el.textContent = exceeded ? '已超阈值' : '未超阈值';
}
function renderRefs(list) {
if (!list || !list.length) return '';
var html = '<ul class="ref-list">';
list.forEach(function (r) {
var text = escapeHtml(r.refer_text || ('p_refer_id=' + r.p_refer_id));
var no = '[' + escapeHtml(r.reference_no) + ']';
var url = String(r.url || '').trim();
if (url) {
html += '<li><a href="' + escapeHtml(url) + '" target="_blank" rel="noopener noreferrer" title="打开文献">'
+ '<span class="no">' + no + '</span>' + text + '</a></li>';
} else {
html += '<li><span class="no">' + no + '</span>' + text + '</li>';
}
});
html += '</ul>';
return html;
}
function renderChips(nos, cls, refs) {
if (!nos || !nos.length) return '<span class="chip">无</span>';
var urlMap = {};
(refs || []).forEach(function (r) {
if (r && r.reference_no != null && r.url) {
urlMap[String(r.reference_no)] = String(r.url);
}
});
return nos.map(function (n) {
var url = urlMap[String(n)] || '';
var label = '[' + escapeHtml(n) + ']';
if (url) {
return '<span class="chip linkable ' + (cls || '') + '"><a href="'
+ escapeHtml(url) + '" target="_blank" rel="noopener noreferrer" title="打开文献">'
+ label + '</a></span>';
}
return '<span class="chip ' + (cls || '') + '">' + label + '</span>';
}).join('');
}
function renderAuthor(data) {
var block = data.same_author_stacking || {};
document.getElementById('authorThreshold').textContent = '阈值 > ' + pct(block.threshold);
setPill(document.getElementById('authorPill'), !!block.exceeded);
var items = block.items || [];
var box = document.getElementById('authorSection');
if (!items.length) {
box.innerHTML = '<div class="empty">未发现超过阈值的同作者堆叠</div>';
return;
}
box.innerHTML = '<div class="card-list">' + items.map(function (it) {
return '<div class="item-card">'
+ '<h3>' + escapeHtml(it.author_name || '未知作者') + '</h3>'
+ '<div class="meta">'
+ '<span>引用 <strong>' + escapeHtml(it.cite_count) + '</strong> 条</span>'
+ '<span>占比 <strong>' + pct(it.cite_ratio) + '</strong></span>'
+ (it.orcid ? '<span>ORCID <strong>' + escapeHtml(it.orcid) + '</strong></span>' : '')
+ '</div>'
+ '<div class="ratio-bar bad"><i style="width:' + Math.min(100, Number(it.cite_ratio || 0) * 100) + '%"></i></div>'
+ '<div class="ref-chips">' + renderChips(it.reference_nos, '', it.references) + '</div>'
+ renderRefs(it.references)
+ '</div>';
}).join('') + '</div>';
}
function renderJournal(data) {
var block = data.same_journal_stacking || {};
document.getElementById('journalThreshold').textContent = '阈值 > ' + pct(block.threshold);
setPill(document.getElementById('journalPill'), !!block.exceeded);
var items = block.items || [];
var box = document.getElementById('journalSection');
if (!items.length) {
box.innerHTML = '<div class="empty">未发现超过阈值的同刊堆叠</div>';
return;
}
box.innerHTML = '<div class="card-list">' + items.map(function (it) {
return '<div class="item-card">'
+ '<h3>' + escapeHtml(it.journal_name || '未知期刊') + '</h3>'
+ '<div class="meta">'
+ '<span>引用 <strong>' + escapeHtml(it.cite_count) + '</strong> 条</span>'
+ '<span>占比 <strong>' + pct(it.cite_ratio) + '</strong></span>'
+ '</div>'
+ '<div class="ratio-bar bad"><i style="width:' + Math.min(100, Number(it.cite_ratio || 0) * 100) + '%"></i></div>'
+ '<div class="ref-chips">' + renderChips(it.reference_nos, '', it.references) + '</div>'
+ renderRefs(it.references)
+ '</div>';
}).join('') + '</div>';
}
function renderSelf(data) {
var block = data.self_citation || {};
document.getElementById('selfThreshold').textContent = '阈值 > ' + pct(block.threshold);
setPill(document.getElementById('selfPill'), !!block.exceeded);
var box = document.getElementById('selfSection');
var items = block.items || [];
var head = '<div class="item-card">'
+ '<div class="meta">'
+ '<span>自引条数 <strong>' + escapeHtml(block.cite_count || 0) + '</strong></span>'
+ '<span>占比 <strong>' + pct(block.cite_ratio) + '</strong></span>'
+ '</div>'
+ '<div class="ratio-bar ' + (block.exceeded ? 'bad' : '') + '"><i style="width:' + Math.min(100, Number(block.cite_ratio || 0) * 100) + '%"></i></div>'
+ '<div class="ref-chips">' + renderChips(block.reference_nos, block.exceeded ? 'bad' : '', items.map(function (it) { return it.reference; }).filter(Boolean)) + '</div>'
+ (block.note ? '<p class="note">' + escapeHtml(block.note) + '</p>' : '')
+ '</div>';
if (!items.length) {
box.innerHTML = head + '<div class="empty" style="margin-top:8px">无自引命中</div>';
return;
}
box.innerHTML = head + '<div class="card-list" style="margin-top:8px">' + items.map(function (it) {
return '<div class="item-card">'
+ '<h3>参考文献 [' + escapeHtml(it.reference_no) + ']</h3>'
+ '<div class="meta">'
+ '<span>本文作者 <strong>' + escapeHtml(it.manuscript_author || '-') + '</strong></span>'
+ '<span>匹配文献作者 <strong>' + escapeHtml(it.matched_refer_author || '-') + '</strong></span>'
+ (it.manuscript_orcid ? '<span>ORCID <strong>' + escapeHtml(it.manuscript_orcid) + '</strong></span>' : '')
+ '</div>'
+ (it.reference ? renderRefs([it.reference]) : '')
+ '</div>';
}).join('') + '</div>';
}
function renderIssues(data) {
var issues = data.author_data_issues || {};
var box = document.getElementById('issueSection');
var items = issues.items || [];
var refsForChips = items.map(function (it) { return it.reference; }).filter(Boolean);
var head = '<div class="item-card"><div class="meta">'
+ '<span>缺失 <strong>' + escapeHtml(issues.missing_count || 0) + '</strong></span>'
+ '<span>不全 <strong>' + escapeHtml(issues.incomplete_count || 0) + '</strong></span>'
+ '</div>'
+ '<div class="ref-chips" style="margin-top:6px">'
+ '<span style="font-size:0.7rem;color:var(--muted);margin-right:4px">缺失:</span>'
+ renderChips(issues.missing_reference_nos, 'bad', refsForChips)
+ '<span style="font-size:0.7rem;color:var(--muted);margin:0 4px 0 8px">不全:</span>'
+ renderChips(issues.incomplete_reference_nos, 'warn', refsForChips)
+ '</div></div>';
if (!items.length) {
box.innerHTML = head + '<div class="empty" style="margin-top:8px">作者数据无明显问题</div>';
return;
}
box.innerHTML = head + '<div class="card-list" style="margin-top:8px">' + items.map(function (it) {
var cls = it.issue_type === 'missing' ? 'bad' : 'warn';
return '<div class="item-card">'
+ '<h3>[' + escapeHtml(it.reference_no) + '] '
+ '<span class="pill ' + cls + '">' + escapeHtml(it.issue_type) + '</span></h3>'
+ '<div class="meta">'
+ '<span>原因 <strong>' + escapeHtml(it.reason || '-') + '</strong></span>'
+ '<span>来源 <strong>' + escapeHtml(it.author_source || '-') + '</strong></span>'
+ '<span>作者数 <strong>' + escapeHtml(it.author_count || 0) + '</strong></span>'
+ '</div>'
+ (it.raw_author ? '<p class="note">raw_author: ' + escapeHtml(it.raw_author) + '</p>' : '')
+ (it.reference ? renderRefs([it.reference]) : '')
+ '</div>';
}).join('') + '</div>';
}
function renderAll(data) {
var authorExceeded = !!(data.same_author_stacking && data.same_author_stacking.exceeded);
var journalExceeded = !!(data.same_journal_stacking && data.same_journal_stacking.exceeded);
var selfExceeded = !!(data.self_citation && data.self_citation.exceeded);
var issues = data.author_data_issues || {};
document.getElementById('summary').innerHTML =
'<div class="stat"><div class="k">参考文献总数</div><div class="v">' + escapeHtml(data.total_references || 0) + '</div></div>'
+ '<div class="stat"><div class="k">同作者堆叠</div><div class="v">' + ((data.same_author_stacking && data.same_author_stacking.items) || []).length + '</div><div class="sub">' + (authorExceeded ? '已超阈值' : '未超阈值') + '</div></div>'
+ '<div class="stat"><div class="k">同刊堆叠</div><div class="v">' + ((data.same_journal_stacking && data.same_journal_stacking.items) || []).length + '</div><div class="sub">' + (journalExceeded ? '已超阈值' : '未超阈值') + '</div></div>'
+ '<div class="stat"><div class="k">自引 / 作者问题</div><div class="v">' + escapeHtml((data.self_citation && data.self_citation.cite_count) || 0) + ' / ' + escapeHtml((issues.missing_count || 0) + (issues.incomplete_count || 0)) + '</div><div class="sub">' + (selfExceeded ? '自引已超阈值' : '自引未超阈值') + '</div></div>';
renderAuthor(data);
renderJournal(data);
renderSelf(data);
renderIssues(data);
document.getElementById('footMeta').textContent =
'p_article_id=' + (data.p_article_id || '-')
+ ' · article_id=' + (data.article_id || '-')
+ ' · ' + (data.computed_at || '-');
placeholder.style.display = 'none';
results.classList.add('show');
statusText.textContent = '共 ' + (data.total_references || 0) + ' 条 · ' + (data.computed_at || '');
}
function load() {
clearError();
var id = parseInt(input.value, 10) || 0;
if (id <= 0) {
showError('请填写有效的 p_article_id');
return;
}
btn.disabled = true;
btn.textContent = '…';
statusText.textContent = '查询中';
var url = API_STATS + (API_STATS.indexOf('?') >= 0 ? '&' : '?') + 'p_article_id=' + encodeURIComponent(id);
fetch(url, {
method: 'GET',
headers: { 'Accept': 'application/json' },
credentials: 'same-origin'
}).then(function (res) {
return res.json();
}).then(function (json) {
if (!json || Number(json.code) !== 0) {
throw new Error((json && json.msg) || '接口返回失败');
}
renderAll(json.data || {});
try {
var u = new URL(window.location.href);
u.searchParams.set('p_article_id', String(id));
window.history.replaceState({}, '', u.toString());
} catch (e) {}
}).catch(function (err) {
results.classList.remove('show');
placeholder.style.display = '';
showError(err.message || String(err));
}).finally(function () {
btn.disabled = false;
btn.textContent = '查询';
});
}
btn.addEventListener('click', load);
input.addEventListener('keydown', function (e) {
if (e.key === 'Enter') load();
});
if (parseInt(input.value, 10) > 0) load();
})();
</script>
</body>
</html>

View File

@@ -20,20 +20,28 @@ class ArticleParserService
if (!file_exists($filePath)) { if (!file_exists($filePath)) {
return json_encode(['status' => 5, 'msg' => '"文档不存在:{$filePath}"']); return json_encode(['status' => 5, 'msg' => '"文档不存在:{$filePath}"']);
} }
$processedFilePath = null;
try { try {
// 含 OMML 公式时先展平PhpWord 单独 saveXML(m:oMath) 会丢掉 xmlns:m触发 loadXML 告警
$loadPath = $filePath;
if ($this->docxContainsOfficeMath($filePath)) {
$processedFilePath = $this->removeEmfFromDocx($filePath);
$loadPath = $processedFilePath;
}
// 关键配置:关闭“仅读数据”,保留完整节结构 // 关键配置:关闭“仅读数据”,保留完整节结构
$reader = IOFactory::createReader(); $reader = IOFactory::createReader();
$reader->setReadDataOnly(false); $reader->setReadDataOnly(false);
Settings::setCompatibility(false); Settings::setCompatibility(false);
Settings::setOutputEscapingEnabled(true); // 避免XML转义冲突 Settings::setOutputEscapingEnabled(true); // 避免XML转义冲突
$doc = $reader->load($filePath); $this->phpWord = $reader->load($loadPath);
$sectionCount = count($doc->getSections());
// $this->log("✅ 文档直接加载成功,节数量:{$sectionCount}");
$this->phpWord = $reader->load($filePath);
$this->sections = $this->phpWord->getSections(); $this->sections = $this->phpWord->getSections();
} catch (\Throwable $e) { } catch (\Throwable $e) {
// 预处理:移除 EMF、表格内分页符等 PhpWord 不兼容内容后重试 // 预处理:移除 EMF、表格内分页符、OMML 公式等 PhpWord 不兼容内容后重试
if ($processedFilePath && is_file($processedFilePath)) {
@unlink($processedFilePath);
}
$processedFilePath = $this->removeEmfFromDocx($filePath); $processedFilePath = $this->removeEmfFromDocx($filePath);
$reader = IOFactory::createReader(); $reader = IOFactory::createReader();
$reader->setReadDataOnly(false); $reader->setReadDataOnly(false);
@@ -42,8 +50,8 @@ class ArticleParserService
$this->phpWord = $reader->load($processedFilePath); $this->phpWord = $reader->load($processedFilePath);
$this->sections = $this->phpWord->getSections(); $this->sections = $this->phpWord->getSections();
} finally {
if (is_file($processedFilePath)) { if ($processedFilePath && is_file($processedFilePath)) {
@unlink($processedFilePath); @unlink($processedFilePath);
} }
} }
@@ -77,7 +85,7 @@ class ArticleParserService
} }
} }
// 3.1 清理表格单元格内分页符PhpWord 无法解析,会抛 Cannot add PageBreak in Cell // 3.1 清理表格单元格内分页符、展平 OMML 公式(避免 PhpWord OfficeMathML 缺命名空间告警
$this->sanitizePageBreaksInDocxXmlFiles($tempDir); $this->sanitizePageBreaksInDocxXmlFiles($tempDir);
// 4. 重新打包为 DOCX // 4. 重新打包为 DOCX
@@ -98,7 +106,24 @@ class ArticleParserService
} }
/** /**
* 移除 word/*.xml 中表格单元格里的分页符,避免 PhpWord 读取失败 * 文档是否含 Word OMML 公式m:oMath / m:oMathPara
*/
private function docxContainsOfficeMath($docxPath): bool
{
$zip = new ZipArchive();
if ($zip->open($docxPath) !== true) {
return false;
}
$xml = $zip->getFromName('word/document.xml');
$zip->close();
if ($xml === false || $xml === '') {
return false;
}
return stripos($xml, 'oMath') !== false;
}
/**
* 清理 word/*.xml表格内分页符 + OMML 公式展平为纯文本
*/ */
private function sanitizePageBreaksInDocxXmlFiles($tempDir) private function sanitizePageBreaksInDocxXmlFiles($tempDir)
{ {
@@ -113,14 +138,14 @@ class ArticleParserService
} }
foreach ($xmlFiles as $xmlPath) { foreach ($xmlFiles as $xmlPath) {
$this->sanitizePageBreaksInWordXmlFile($xmlPath); $this->sanitizeWordXmlFile($xmlPath);
} }
} }
/** /**
* @param string $xmlPath * @param string $xmlPath
*/ */
private function sanitizePageBreaksInWordXmlFile($xmlPath) private function sanitizeWordXmlFile($xmlPath)
{ {
if (!is_file($xmlPath) || !is_readable($xmlPath)) { if (!is_file($xmlPath) || !is_readable($xmlPath)) {
return; return;
@@ -131,15 +156,72 @@ class ArticleParserService
return; return;
} }
// 缺 xmlns:m 时先补到根节点,便于 DOM 解析
if (stripos($xml, 'oMath') !== false && stripos($xml, 'xmlns:m=') === false) {
$xml = preg_replace(
'/<(w:document|w:hdr|w:ftr|w:footnotes|w:endnotes)\b([^>]*)>/',
'<$1$2 xmlns:m="http://schemas.openxmlformats.org/officeDocument/2006/math">',
$xml,
1
);
if (!is_string($xml)) {
return;
}
}
$prev = libxml_use_internal_errors(true);
$dom = new DOMDocument(); $dom = new DOMDocument();
$dom->preserveWhiteSpace = true; $dom->preserveWhiteSpace = true;
$dom->formatOutput = false; $dom->formatOutput = false;
if (@$dom->loadXML($xml) === false) { $ok = $dom->loadXML($xml);
libxml_clear_errors();
libxml_use_internal_errors($prev);
if ($ok === false) {
// DOM 失败时用正则展平公式,避免 PhpWord 再踩 oMath 命名空间问题
$flattened = $this->flattenOfficeMathByRegex($xml);
if ($flattened !== $xml) {
file_put_contents($xmlPath, $flattened);
}
return; return;
} }
$xpath = new DOMXPath($dom); $xpath = new DOMXPath($dom);
$xpath->registerNamespace('w', 'http://schemas.openxmlformats.org/wordprocessingml/2006/main'); $wNs = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
$mNs = 'http://schemas.openxmlformats.org/officeDocument/2006/math';
$xpath->registerNamespace('w', $wNs);
$xpath->registerNamespace('m', $mNs);
$changed = false;
// 先处理 oMathPara再处理剩余 oMath保留 m:t 可见文本
foreach (['//m:oMathPara', '//m:oMath'] as $query) {
$nodes = $xpath->query($query);
if (!$nodes || $nodes->length === 0) {
continue;
}
for ($i = $nodes->length - 1; $i >= 0; $i--) {
$mathNode = $nodes->item($i);
if (!$mathNode || !$mathNode->parentNode) {
continue;
}
$text = '';
$tNodes = $xpath->query('.//m:t', $mathNode);
if ($tNodes) {
foreach ($tNodes as $tNode) {
$text .= $tNode->textContent;
}
}
$run = $dom->createElementNS($wNs, 'w:r');
$tEl = $dom->createElementNS($wNs, 'w:t');
$tEl->appendChild($dom->createTextNode($text));
if ($text !== '' && preg_match('/^\s|\s$/u', $text)) {
$tEl->setAttributeNS('http://www.w3.org/XML/1998/namespace', 'xml:space', 'preserve');
}
$run->appendChild($tEl);
$mathNode->parentNode->replaceChild($run, $mathNode);
$changed = true;
}
}
$queries = [ $queries = [
'//w:tc//w:br[@w:type="page"]', '//w:tc//w:br[@w:type="page"]',
@@ -147,7 +229,6 @@ class ArticleParserService
'//w:tc//w:pPr/w:pageBreakBefore', '//w:tc//w:pPr/w:pageBreakBefore',
]; ];
$removed = false;
foreach ($queries as $query) { foreach ($queries as $query) {
$nodes = $xpath->query($query); $nodes = $xpath->query($query);
if (!$nodes || $nodes->length === 0) { if (!$nodes || $nodes->length === 0) {
@@ -157,16 +238,55 @@ class ArticleParserService
$node = $nodes->item($i); $node = $nodes->item($i);
if ($node && $node->parentNode) { if ($node && $node->parentNode) {
$node->parentNode->removeChild($node); $node->parentNode->removeChild($node);
$removed = true; $changed = true;
} }
} }
} }
if ($removed) { if ($changed) {
file_put_contents($xmlPath, $dom->saveXML()); file_put_contents($xmlPath, $dom->saveXML());
} }
} }
/**
* 正则展平 OMML保留 m:t 文本,去掉 oMath/oMathParaDOM 不可用时兜底)
*/
private function flattenOfficeMathByRegex(string $xml): string
{
if (stripos($xml, 'oMath') === false) {
return $xml;
}
$toRun = function (string $inner): string {
$text = '';
if (preg_match_all('/<m:t\b[^>]*>([\s\S]*?)<\/m:t>/i', $inner, $m)) {
foreach ($m[1] as $part) {
$text .= html_entity_decode(strip_tags($part), ENT_QUOTES | ENT_XML1, 'UTF-8');
}
}
$safe = htmlspecialchars($text, ENT_QUOTES | ENT_XML1, 'UTF-8');
$space = ($text !== '' && preg_match('/^\s|\s$/u', $text)) ? ' xml:space="preserve"' : '';
return '<w:r><w:t' . $space . '>' . $safe . '</w:t></w:r>';
};
$xml = preg_replace_callback(
'/<m:oMathPara\b[^>]*>([\s\S]*?)<\/m:oMathPara>/i',
function ($m) use ($toRun) {
return $toRun($m[1]);
},
$xml
);
$xml = preg_replace_callback(
'/<m:oMath\b[^>]*>([\s\S]*?)<\/m:oMath>/i',
function ($m) use ($toRun) {
return $toRun($m[1]);
},
is_string($xml) ? $xml : ''
);
return is_string($xml) ? $xml : '';
}
/** /**
* 递归添加目录文件到 ZipArchive * 递归添加目录文件到 ZipArchive
* @param string $dir 目录路径 * @param string $dir 目录路径

View File

@@ -246,6 +246,131 @@ class ReferenceReferAuthorService
return $result; return $result;
} }
/**
* 单条参考文献:外网拉取作者并入库,再返回明细
* 有 DOI → Crossref + OpenAlex失败或无 DOI → 解析 refer.author
*
* @return array{
* p_article_id:int,
* p_refer_id:int,
* reference_no:int,
* doi:string,
* synced:int,
* refer:array|null,
* author_count:int,
* authors:array
* }
*/
public function fetchAuthorsByPReferId($pArticleId, $pReferId)
{
$pArticleId = intval($pArticleId);
$pReferId = intval($pReferId);
if ($pArticleId <= 0 || $pReferId <= 0) {
throw new \InvalidArgumentException('p_article_id and p_refer_id are required');
}
$refer = Db::name('production_article_refer')
->where('p_refer_id', $pReferId)
->where('p_article_id', $pArticleId)
->where('state', 0)
->find();
if (empty($refer)) {
return [
'p_article_id' => $pArticleId,
'p_refer_id' => $pReferId,
'reference_no' => 0,
'doi' => '',
'synced' => 0,
'refer' => null,
'author_count' => 0,
'authors' => [],
];
}
$refUtil = new ReferenceCheckService();
$doi = $refUtil->extractDoiFromRefer($refer);
$syncedCount = $this->syncOneRefer($pReferId, $pArticleId, $refer);
$result = $this->getAuthorsByPReferId($pArticleId, $pReferId);
$result['doi'] = $doi;
$result['synced'] = $syncedCount > 0 ? 1 : 0;
return $result;
}
/**
* 按 p_article_id + p_refer_id 读取作者明细(只读本地库)
*
* @return array{p_article_id:int,p_refer_id:int,reference_no:int,refer:array|null,author_count:int,authors:array}
*/
public function getAuthorsByPReferId($pArticleId, $pReferId)
{
$pArticleId = intval($pArticleId);
$pReferId = intval($pReferId);
if ($pArticleId <= 0 || $pReferId <= 0) {
throw new \InvalidArgumentException('p_article_id and p_refer_id are required');
}
$refer = Db::name('production_article_refer')
->field('p_refer_id,p_article_id,index,author,title,joura,refer_doi,doilink,refer_type')
->where('p_refer_id', $pReferId)
->where('p_article_id', $pArticleId)
->where('state', 0)
->find();
if (empty($refer)) {
return [
'p_article_id' => $pArticleId,
'p_refer_id' => $pReferId,
'reference_no' => 0,
'refer' => null,
'author_count' => 0,
'authors' => [],
];
}
$rows = Db::name('production_article_refer_author')
->where('p_article_id', $pArticleId)
->where('p_refer_id', $pReferId)
->order('author_seq asc, id asc')
->field('author_seq,author_position,is_first_author,family,given,display_name,citation_name,orcid,openalex_id,identity_source')
->select();
$authors = [];
foreach ($rows as $row) {
$authors[] = [
'author_seq' => intval($row['author_seq'] ?? 0),
'author_position' => (string)($row['author_position'] ?? ''),
'is_first_author' => intval($row['is_first_author'] ?? 0),
'family' => (string)($row['family'] ?? ''),
'given' => (string)($row['given'] ?? ''),
'display_name' => (string)($row['display_name'] ?? ''),
'citation_name' => (string)($row['citation_name'] ?? ''),
'orcid' => (string)($row['orcid'] ?? ''),
'openalex_id' => (string)($row['openalex_id'] ?? ''),
'identity_source' => (string)($row['identity_source'] ?? ''),
];
}
return [
'p_article_id' => $pArticleId,
'p_refer_id' => $pReferId,
'reference_no' => intval($refer['index']) + 1,
'refer' => [
'p_refer_id' => $pReferId,
'reference_no' => intval($refer['index']) + 1,
'author' => (string)($refer['author'] ?? ''),
'title' => (string)($refer['title'] ?? ''),
'joura' => (string)($refer['joura'] ?? ''),
'refer_doi' => (string)($refer['refer_doi'] ?? ''),
'doilink' => (string)($refer['doilink'] ?? ''),
'refer_type' => (string)($refer['refer_type'] ?? ''),
],
'author_count' => count($authors),
'authors' => $authors,
];
}
/** /**
* 读取已入库的作者身份(与 ReferenceAuthorIdentityService 结构兼容) * 读取已入库的作者身份(与 ReferenceAuthorIdentityService 结构兼容)
* *

View File

@@ -6,33 +6,16 @@ use think\Db;
/** /**
* 参考文献引用堆叠统计:同刊、同作者、自引(实时计算,不入库) * 参考文献引用堆叠统计:同刊、同作者、自引(实时计算,不入库)
* 仅读本地库/字段,不请求 Crossref/OpenAlex避免网关超时。
*/ */
class ReferenceStackingStatsService class ReferenceStackingStatsService
{ {
const DETAIL_JOURNAL = 'journal'; const DETAIL_JOURNAL = 'journal';
const DETAIL_AUTHOR = 'author'; const DETAIL_AUTHOR = 'author';
const THRESHOLD_SAME_AUTHOR = 0.15; const THRESHOLD_SAME_AUTHOR = 0.15;//相同作者;
const THRESHOLD_SAME_JOURNAL = 0.20; const THRESHOLD_SAME_JOURNAL = 0.20;//同一期刊;
const THRESHOLD_SELF_CITATION = 0.10; const THRESHOLD_SELF_CITATION = 0.10;//作者自引;
/** @var ReferenceCheckService */
private $refUtil;
/** @var CrossrefService */
private $crossref;
/** @var ReferenceAuthorIdentityService */
private $identity;
public function __construct()
{
$this->refUtil = new ReferenceCheckService();
$this->crossref = new CrossrefService([
'mailto' => trim((string)\think\Env::get('crossref_mailto', '')),
]);
$this->identity = new ReferenceAuthorIdentityService();
}
/** /**
* 按阈值规则实时统计:同作者(>15%)、同刊(>20%)、自引(>10%) * 按阈值规则实时统计:同作者(>15%)、同刊(>20%)、自引(>10%)
@@ -50,6 +33,44 @@ class ReferenceStackingStatsService
return $this->formatThresholdStackingReport($this->compute($pArticleId)); return $this->formatThresholdStackingReport($this->compute($pArticleId));
} }
/**
* 堆叠摘要(展示用):同作者名称+比例、同刊名称+比例、自引比例
*
* @param int $pArticleId
* @return array
*/
public function getStackingSummaryByPArticleId($pArticleId)
{
$full = $this->getThresholdStackingByPArticleId($pArticleId);
$authors = [];
foreach ((array)(($full['same_author_stacking']['items'] ?? [])) as $item) {
$authors[] = [
'name' => (string)($item['author_name'] ?? ''),
'ratio' => round(floatval($item['cite_ratio'] ?? 0), 4),
];
}
$journals = [];
foreach ((array)(($full['same_journal_stacking']['items'] ?? [])) as $item) {
$journals[] = [
'name' => (string)($item['journal_name'] ?? ''),
'ratio' => round(floatval($item['cite_ratio'] ?? 0), 4),
];
}
$self = (array)($full['self_citation'] ?? []);
return [
'p_article_id' => intval($full['p_article_id'] ?? 0),
'total_references' => intval($full['total_references'] ?? 0),
'same_author' => $authors,
'same_journal' => $journals,
'self_citation_ratio' => round(floatval($self['cite_ratio'] ?? 0), 4),
'computed_at' => (string)($full['computed_at'] ?? ''),
];
}
/** /**
* @param array $full compute/analyze 或 getStored 的完整结果 * @param array $full compute/analyze 或 getStored 的完整结果
*/ */
@@ -118,6 +139,11 @@ class ReferenceStackingStatsService
]; ];
} }
$authorDataIssues = $this->formatAuthorDataIssues(
(array)($full['author_data_issues'] ?? []),
$referMap
);
return [ return [
'p_article_id' => intval($full['p_article_id'] ?? 0), 'p_article_id' => intval($full['p_article_id'] ?? 0),
'article_id' => intval($full['article_id'] ?? 0), 'article_id' => intval($full['article_id'] ?? 0),
@@ -141,6 +167,7 @@ class ReferenceStackingStatsService
'items' => $selfItems, 'items' => $selfItems,
'note' => (string)($full['author_identity_note'] ?? ''), 'note' => (string)($full['author_identity_note'] ?? ''),
], ],
'author_data_issues' => $authorDataIssues,
'computed_at' => (string)($full['computed_at'] ?? ''), 'computed_at' => (string)($full['computed_at'] ?? ''),
]; ];
} }
@@ -167,21 +194,22 @@ class ReferenceStackingStatsService
->order('index asc') ->order('index asc')
->select(); ->select();
$manuscriptAuthors = $this->identity->resolveManuscriptAuthors($pArticleId); $manuscriptAuthors = $this->loadManuscriptAuthorsLocal($pArticleId);
$ambiguousManuscriptNameKeys = $this->buildAmbiguousManuscriptNameKeys($manuscriptAuthors); $ambiguousManuscriptNameKeys = $this->buildAmbiguousManuscriptNameKeys($manuscriptAuthors);
$doiCache = []; $referAuthorRowsMap = $this->loadReferAuthorRowsByPArticleId($pArticleId);
$referMap = []; $referMap = [];
$journalBuckets = []; $journalBuckets = [];
$authorBuckets = []; $authorBuckets = [];
$selfCitationDetails = []; $selfCitationDetails = [];
$authorDataIssues = [];
foreach ($refers as $refer) { foreach ($refers as $refer) {
$refNo = intval($refer['index']) + 1; $refNo = intval($refer['index']) + 1;
$pReferId = intval($refer['p_refer_id']); $pReferId = intval($refer['p_refer_id']);
$referMap[$pReferId] = $refer; $referMap[$pReferId] = $refer;
$meta = $this->resolveReferMeta($refer, $doiCache); $meta = $this->resolveReferMetaLocal($refer);
$joura = (string)$meta['joura']; $joura = (string)$meta['joura'];
$journalKey = $this->normalizeJournalKey($joura); $journalKey = $this->normalizeJournalKey($joura);
if ($journalKey !== '') { if ($journalKey !== '') {
@@ -199,7 +227,14 @@ class ReferenceStackingStatsService
$journalBuckets[$journalKey]['p_refer_ids'][] = $pReferId; $journalBuckets[$journalKey]['p_refer_ids'][] = $pReferId;
} }
$referAuthors = $this->resolveReferAuthorsWithMeta($pReferId, $meta); $dbRows = isset($referAuthorRowsMap[$pReferId]) ? $referAuthorRowsMap[$pReferId] : [];
$resolved = $this->resolveReferAuthorsWithMetaDetailed($pReferId, $meta, $dbRows);
$referAuthors = $resolved['authors'];
$authorIssue = $this->buildAuthorDataIssue($refNo, $pReferId, $meta, $resolved);
if ($authorIssue !== null) {
$authorDataIssues[] = $authorIssue;
}
$this->accumulateAuthorBuckets($authorBuckets, $referAuthors, $refNo, $pReferId); $this->accumulateAuthorBuckets($authorBuckets, $referAuthors, $refNo, $pReferId);
$matchedManuscript = $this->matchManuscriptAuthorForSelfCitation( $matchedManuscript = $this->matchManuscriptAuthorForSelfCitation(
@@ -238,12 +273,61 @@ class ReferenceStackingStatsService
'journal_details' => $journalDetails, 'journal_details' => $journalDetails,
'author_details' => $authorDetails, 'author_details' => $authorDetails,
'self_citation_details' => $selfDetails, 'self_citation_details' => $selfDetails,
'author_data_issues' => $authorDataIssues,
'refer_map' => $referMap, 'refer_map' => $referMap,
'author_identity_note' => '同作者堆叠按 citation_name空则 display_name姓名精确匹配同名即同人本文出现重名作者如两位 Lin时跳过该姓名的自引判定。优先读 t_production_article_refer_author否则解析 refer.author。', 'author_identity_note' => '同作者堆叠按 citation_name空则 display_name姓名精确匹配同名即同人本文出现重名作者如两位 Lin时跳过该姓名的自引判定。优先读 t_production_article_refer_author否则解析 refer.author。统计仅用本地数据,不实时请求 Crossref/OpenAlex。',
'computed_at' => date('Y-m-d H:i:s'), 'computed_at' => date('Y-m-d H:i:s'),
]; ];
} }
/**
* 本文作者(仅本地库,不打 OpenAlex
*
* @return array<int, array{display_name:string,orcid:string}>
*/
private function loadManuscriptAuthorsLocal($pArticleId)
{
$rows = Db::name('production_article_author')
->field('first_name,last_name,author_name,orcid')
->where('p_article_id', intval($pArticleId))
->where('state', 0)
->select();
$list = [];
foreach ($rows as $row) {
$first = trim((string)($row['first_name'] ?? ''));
$last = trim((string)($row['last_name'] ?? ''));
$displayName = ($first !== '' && $last !== '')
? trim($first . ' ' . $last)
: trim((string)($row['author_name'] ?? ''));
$list[] = [
'display_name' => $displayName,
'orcid' => $this->cleanOrcid($row['orcid'] ?? ''),
];
}
return $list;
}
/**
* @return array<int, array>
*/
private function loadReferAuthorRowsByPArticleId($pArticleId)
{
$rows = Db::name('production_article_refer_author')
->where('p_article_id', intval($pArticleId))
->order('p_refer_id asc, author_seq asc, id asc')
->field('p_refer_id,display_name,citation_name,orcid')
->select();
$map = [];
foreach ($rows as $row) {
$map[intval($row['p_refer_id'])][] = $row;
}
return $map;
}
private function formatAuthorBucketDetails(array $buckets, $detailType) private function formatAuthorBucketDetails(array $buckets, $detailType)
{ {
$list = []; $list = [];
@@ -280,23 +364,47 @@ class ReferenceStackingStatsService
/** /**
* @return array<int, array{name:string,orcid:string}> * @return array<int, array{name:string,orcid:string}>
*/ */
private function resolveReferAuthorsWithMeta($pReferId, array $meta) private function resolveReferAuthorsWithMeta($pReferId, array $meta, array $dbRows = null)
{
return $this->resolveReferAuthorsWithMetaDetailed($pReferId, $meta, $dbRows)['authors'];
}
/**
* @param array|null $dbRows 预加载的 production_article_refer_author 行null 则按 p_refer_id 查询
* @return array{
* authors:array<int, array{name:string,orcid:string}>,
* source:string,
* db_row_count:int,
* db_usable_count:int,
* db_blank_name_count:int,
* has_et_al:bool
* }
*/
private function resolveReferAuthorsWithMetaDetailed($pReferId, array $meta, array $dbRows = null)
{ {
$pReferId = intval($pReferId); $pReferId = intval($pReferId);
$list = []; $list = [];
$dbRowCount = 0;
$dbBlankNameCount = 0;
$authorString = (string)($meta['author'] ?? '');
$hasEtAl = (bool)preg_match('/\bet\s+al\.?\b/iu', $authorString);
if ($pReferId > 0) { if ($pReferId > 0) {
$rows = Db::name('production_article_refer_author') if ($dbRows === null) {
->where('p_refer_id', $pReferId) $dbRows = Db::name('production_article_refer_author')
->order('author_seq asc, id asc') ->where('p_refer_id', $pReferId)
->field('display_name,citation_name,orcid') ->order('author_seq asc, id asc')
->select(); ->field('display_name,citation_name,orcid')
foreach ($rows as $row) { ->select();
}
$dbRowCount = count($dbRows);
foreach ($dbRows as $row) {
$name = trim((string)($row['citation_name'] ?? '')); $name = trim((string)($row['citation_name'] ?? ''));
if ($name === '') { if ($name === '') {
$name = trim((string)($row['display_name'] ?? '')); $name = trim((string)($row['display_name'] ?? ''));
} }
if ($name === '') { if ($name === '') {
$dbBlankNameCount++;
continue; continue;
} }
$list[] = [ $list[] = [
@@ -307,17 +415,148 @@ class ReferenceStackingStatsService
} }
if (!empty($list)) { if (!empty($list)) {
return $list; return [
'authors' => $list,
'source' => 'refer_author_table',
'db_row_count' => $dbRowCount,
'db_usable_count' => count($list),
'db_blank_name_count' => $dbBlankNameCount,
'has_et_al' => $hasEtAl,
];
} }
foreach ($this->parseAuthorStringParts((string)($meta['author'] ?? '')) as $name) { foreach ($this->parseAuthorStringParts($authorString) as $name) {
$list[] = [ $list[] = [
'name' => $name, 'name' => $name,
'orcid' => '', 'orcid' => '',
]; ];
} }
return $list; return [
'authors' => $list,
'source' => empty($list) ? 'none' : 'author_string',
'db_row_count' => $dbRowCount,
'db_usable_count' => 0,
'db_blank_name_count' => $dbBlankNameCount,
'has_et_al' => $hasEtAl,
];
}
/**
* @param array{
* authors:array,
* source:string,
* db_row_count:int,
* db_usable_count:int,
* db_blank_name_count:int,
* has_et_al:bool
* } $resolved
* @return array|null
*/
private function buildAuthorDataIssue($refNo, $pReferId, array $meta, array $resolved)
{
$authors = (array)($resolved['authors'] ?? []);
$authorCount = count($authors);
$source = (string)($resolved['source'] ?? 'none');
$dbRowCount = intval($resolved['db_row_count'] ?? 0);
$dbBlankNameCount = intval($resolved['db_blank_name_count'] ?? 0);
$hasEtAl = !empty($resolved['has_et_al']);
$rawAuthor = trim((string)($meta['author'] ?? ''));
if ($authorCount <= 0) {
$reason = '未获取到作者';
if ($dbRowCount > 0 && $dbBlankNameCount > 0) {
$reason = '作者明细表有记录但姓名均为空';
} elseif ($rawAuthor !== '') {
$reason = '作者字段无法解析出有效姓名';
}
return [
'issue_type' => 'missing',
'reason' => $reason,
'reference_no' => intval($refNo),
'p_refer_id' => intval($pReferId),
'author_source' => $source,
'author_count' => 0,
'raw_author' => $rawAuthor,
'meta_source' => (string)($meta['meta_source'] ?? ''),
];
}
$reasons = [];
if ($dbRowCount > 0 && $dbBlankNameCount > 0) {
$reasons[] = '部分作者姓名为空(空姓名 ' . $dbBlankNameCount . ' 条)';
}
// refer.author 常因引用格式截断带 et al.;明细表已有完整作者时不再判为不全
if ($hasEtAl && $source !== 'refer_author_table') {
$reasons[] = '作者列表含 et al.,可能不全';
}
if (empty($reasons)) {
return null;
}
return [
'issue_type' => 'incomplete',
'reason' => implode('', $reasons),
'reference_no' => intval($refNo),
'p_refer_id' => intval($pReferId),
'author_source' => $source,
'author_count' => $authorCount,
'raw_author' => $rawAuthor,
'meta_source' => (string)($meta['meta_source'] ?? ''),
];
}
/**
* @param array $issues
* @param array<int,array> $referMap
* @return array{
* missing_count:int,
* incomplete_count:int,
* missing_reference_nos:int[],
* incomplete_reference_nos:int[],
* items:array
* }
*/
private function formatAuthorDataIssues(array $issues, array $referMap)
{
$missing = [];
$incomplete = [];
$items = [];
usort($issues, function ($a, $b) {
return intval($a['reference_no'] ?? 0) <=> intval($b['reference_no'] ?? 0);
});
foreach ($issues as $issue) {
$pReferId = intval($issue['p_refer_id'] ?? 0);
$item = [
'issue_type' => (string)($issue['issue_type'] ?? ''),
'reason' => (string)($issue['reason'] ?? ''),
'reference_no' => intval($issue['reference_no'] ?? 0),
'p_refer_id' => $pReferId,
'author_source' => (string)($issue['author_source'] ?? ''),
'author_count' => intval($issue['author_count'] ?? 0),
'raw_author' => (string)($issue['raw_author'] ?? ''),
'meta_source' => (string)($issue['meta_source'] ?? ''),
'reference' => $this->buildReferBriefs([$pReferId], $referMap)[0] ?? null,
];
$items[] = $item;
if ($item['issue_type'] === 'missing') {
$missing[] = $item['reference_no'];
} elseif ($item['issue_type'] === 'incomplete') {
$incomplete[] = $item['reference_no'];
}
}
return [
'missing_count' => count($missing),
'incomplete_count' => count($incomplete),
'missing_reference_nos' => array_values($missing),
'incomplete_reference_nos' => array_values($incomplete),
'items' => $items,
];
} }
/** /**
@@ -462,7 +701,7 @@ class ReferenceStackingStatsService
/** /**
* @param int[] $pReferIds * @param int[] $pReferIds
* @param array<int,array> $referMap * @param array<int,array> $referMap
* @return array<int,array{p_refer_id:int,reference_no:int,refer_text:string}> * @return array<int,array{p_refer_id:int,reference_no:int,refer_text:string,doi:string,url:string}>
*/ */
private function buildReferBriefs(array $pReferIds, array $referMap) private function buildReferBriefs(array $pReferIds, array $referMap)
{ {
@@ -473,16 +712,62 @@ class ReferenceStackingStatsService
continue; continue;
} }
$refer = $referMap[$pReferId]; $refer = $referMap[$pReferId];
$doi = $this->extractReferDoi($refer);
$list[] = [ $list[] = [
'p_refer_id' => $pReferId, 'p_refer_id' => $pReferId,
'reference_no' => intval($refer['index'] ?? 0) + 1, 'reference_no' => intval($refer['index'] ?? 0) + 1,
'refer_text' => $this->referSnippet($refer), 'refer_text' => $this->referSnippet($refer),
'doi' => $doi,
'url' => $this->buildReferOpenUrl($refer, $doi),
]; ];
} }
return $list; return $list;
} }
/**
* 从参考文献行提取 DOI去前缀
*/
private function extractReferDoi(array $refer)
{
foreach (['refer_doi', 'doilink'] as $field) {
$raw = trim((string)($refer[$field] ?? ''));
if ($raw === '') {
continue;
}
$raw = preg_replace('#^https?://(dx\.)?doi\.org/#i', '', $raw);
$raw = trim($raw, " \t\n\r\0\x0B/");
if ($raw !== '' && stripos($raw, '10.') !== false) {
if (preg_match('#(10\.\d{4,9}/[^\s]+)#i', $raw, $m)) {
return rtrim($m[1], '.,;');
}
return $raw;
}
}
return '';
}
/**
* 可跳转打开的文献链接:优先 DOI其次 doilink 若已是 URL
*/
private function buildReferOpenUrl(array $refer, $doi = '')
{
$doi = trim((string)$doi);
if ($doi === '') {
$doi = $this->extractReferDoi($refer);
}
if ($doi !== '') {
return 'https://doi.org/' . $doi;
}
$doilink = trim((string)($refer['doilink'] ?? ''));
if ($doilink !== '' && preg_match('#^https?://#i', $doilink)) {
return $doilink;
}
return '';
}
private function loadReferMapByPArticleId($pArticleId) private function loadReferMapByPArticleId($pArticleId)
{ {
$pArticleId = intval($pArticleId); $pArticleId = intval($pArticleId);
@@ -540,7 +825,7 @@ class ReferenceStackingStatsService
} }
/** /**
* 合并 refer 行已有字段、Crossref、refer_frag/refer_content 解析结果 * 仅用本地字段 + refer_frag/refer_content 解析,不请求 Crossref
* *
* @return array{ * @return array{
* author:string, * author:string,
@@ -551,7 +836,7 @@ class ReferenceStackingStatsService
* resolved:bool * resolved:bool
* } * }
*/ */
private function resolveReferMeta(array $refer, array &$doiCache) private function resolveReferMetaLocal(array $refer)
{ {
$author = trim(trim((string)($refer['author'] ?? '')), '.'); $author = trim(trim((string)($refer['author'] ?? '')), '.');
$joura = trim(trim((string)($refer['joura'] ?? '')), '.'); $joura = trim(trim((string)($refer['joura'] ?? '')), '.');
@@ -563,38 +848,6 @@ class ReferenceStackingStatsService
$authorKeys = $this->extractAuthorKeysFromAuthorString($author); $authorKeys = $this->extractAuthorKeysFromAuthorString($author);
} }
$doi = $this->refUtil->extractDoiFromRefer($refer);
$summary = null;
if ($doi !== '') {
if (!array_key_exists($doi, $doiCache)) {
try {
$doiCache[$doi] = $this->crossref->fetchWorkSummary($doi);
} catch (\Throwable $e) {
$doiCache[$doi] = null;
}
}
$summary = $doiCache[$doi];
}
if (is_array($summary)) {
$sources[] = 'crossref';
if ($joura === '' && trim((string)($summary['joura'] ?? '')) !== '') {
$joura = trim((string)$summary['joura']);
}
$crossrefKeys = $this->authorKeysFromCrossrefMessage($summary['raw'] ?? []);
if ($author === '') {
$citationAuthor = $this->crossref->getAuthorsCitation($summary['raw'] ?? [], 3);
if ($citationAuthor !== '') {
$author = $citationAuthor;
}
$authorKeys = !empty($crossrefKeys)
? $crossrefKeys
: $this->extractAuthorKeysFromAuthorString($author);
} elseif (!empty($crossrefKeys)) {
$authorKeys = array_values(array_unique(array_merge($authorKeys, $crossrefKeys)));
}
}
if ($joura === '' || $author === '') { if ($joura === '' || $author === '') {
$fragParsed = $this->parseStructuredReferText($refer); $fragParsed = $this->parseStructuredReferText($refer);
if (is_array($fragParsed)) { if (is_array($fragParsed)) {
@@ -668,37 +921,6 @@ class ReferenceStackingStatsService
return null; return null;
} }
/**
* @return string[]
*/
private function authorKeysFromCrossrefMessage(array $message)
{
$keys = [];
if (empty($message['author']) || !is_array($message['author'])) {
return $keys;
}
foreach ($message['author'] as $author) {
if (!is_array($author)) {
continue;
}
$family = trim((string)($author['family'] ?? ''));
$given = trim((string)($author['given'] ?? ''));
if ($family === '' && $given === '') {
$org = trim((string)($author['name'] ?? ''));
if ($org !== '') {
$keys[] = $this->authorKeyFromCitationPart($org);
}
continue;
}
if ($family !== '') {
$keys[] = mb_strtoupper($family) . '|' . $this->givenToInitials($given);
}
}
return array_values(array_unique(array_filter($keys)));
}
private function referSnippet(array $refer) private function referSnippet(array $refer)
{ {
foreach (['refer_content', 'refer_frag'] as $field) { foreach (['refer_content', 'refer_frag'] as $field) {
@@ -805,23 +1027,6 @@ class ReferenceStackingStatsService
return $key; return $key;
} }
private function givenToInitials($given)
{
$given = trim((string)$given);
if ($given === '') {
return '';
}
$parts = preg_split('/[\s\-\.]+/u', $given, -1, PREG_SPLIT_NO_EMPTY);
$initials = '';
foreach ($parts as $part) {
$first = mb_substr($part, 0, 1);
if ($first !== '') {
$initials .= mb_strtoupper($first);
}
}
return $initials;
}
private function cleanOrcid($orcid) private function cleanOrcid($orcid)
{ {
$orcid = trim((string)$orcid); $orcid = trim((string)$orcid);

View File

@@ -26,6 +26,7 @@ class UserActLog
return ['status' => 2, 'msg' => '非法操作']; return ['status' => 2, 'msg' => '非法操作'];
} }
$aInsert['create_time'] = time(); $aInsert['create_time'] = time();
$aInsert['update_time'] = time();
$result = Db::name('user_act_log')->insertGetId($aInsert); $result = Db::name('user_act_log')->insertGetId($aInsert);
if(empty($result)){ if(empty($result)){
return ['status' => 3, 'msg' => '数据插入失败'.Db::getLastSql()."\n数据内容:",'data' => $aParam]; return ['status' => 3, 'msg' => '数据插入失败'.Db::getLastSql()."\n数据内容:",'data' => $aParam];

View File

@@ -65,6 +65,8 @@ class ReferenceCheckArticleWorker
} }
$this->svc->log('ReferenceCheckArticleWorker start p_article_id=' . $pArticleId . ' batch_id=' . $batchId); $this->svc->log('ReferenceCheckArticleWorker start p_article_id=' . $pArticleId . ' batch_id=' . $batchId);
// 快照本批待处理 id联合引用组长一次会整组落库循环计数会小于 total_count收尾按快照回填
$trackedIds = $this->listPendingCheckIds($pArticleId);
$done = 0; $done = 0;
$failed = 0; $failed = 0;
while (true) { while (true) {
@@ -84,12 +86,58 @@ class ReferenceCheckArticleWorker
} }
} }
if (!empty($trackedIds)) {
$stats = $this->summarizeTrackedCheckIds($trackedIds);
$done = intval($stats['done']);
$failed = intval($stats['failed']);
}
$this->finalizeBatch($batchId, $done, $failed); $this->finalizeBatch($batchId, $done, $failed);
$this->svc->log('ReferenceCheckArticleWorker done p_article_id=' . $pArticleId . ' batch_id=' . $batchId . ' done=' . $done . ' failed=' . $failed); $this->svc->log('ReferenceCheckArticleWorker done p_article_id=' . $pArticleId . ' batch_id=' . $batchId . ' done=' . $done . ' failed=' . $failed);
$this->publishNextWaitingBatch(); $this->publishNextWaitingBatch();
} }
private function listPendingCheckIds($pArticleId)
{
$rows = Db::name('article_reference_relevance_check_result')
->where('p_article_id', intval($pArticleId))
->where('queue_status', ReferenceRelevanceCheckService::QUEUE_PENDING)
->where('status', ReferenceRelevanceCheckService::RECORD_PENDING)
->field('id')
->select();
$ids = [];
foreach ($rows as $row) {
$id = intval(isset($row['id']) ? $row['id'] : 0);
if ($id > 0) {
$ids[] = $id;
}
}
return $ids;
}
private function summarizeTrackedCheckIds(array $checkIds)
{
$checkIds = array_values(array_filter(array_map('intval', $checkIds)));
if (empty($checkIds)) {
return ['done' => 0, 'failed' => 0];
}
$rows = Db::name('article_reference_relevance_check_result')
->whereIn('id', $checkIds)
->field('id,status')
->select();
$done = 0;
$failed = 0;
foreach ($rows as $row) {
$st = intval(isset($row['status']) ? $row['status'] : -1);
if ($st === ReferenceRelevanceCheckService::RECORD_COMPLETED) {
$done++;
} elseif ($st === ReferenceRelevanceCheckService::RECORD_FAILED) {
$failed++;
}
}
return ['done' => $done, 'failed' => $failed];
}
private function canStartArticleWork($batchId) private function canStartArticleWork($batchId)
{ {
$running = Db::name('article_reference_relevance_check_batch') $running = Db::name('article_reference_relevance_check_batch')
@@ -182,18 +230,25 @@ class ReferenceCheckArticleWorker
return; return;
} }
$total = intval($batch['total_count']); $total = intval($batch['total_count']);
$done = intval($done);
$failed = intval($failed);
// 快照回填后若实际终态条数多于入队 total抬升 total 保持一致
if (($done + $failed) > $total) {
$total = $done + $failed;
}
$status = self::BATCH_DONE; $status = self::BATCH_DONE;
if ($failed > 0) { if ($failed > 0) {
$status = self::BATCH_PARTIAL_FAILED; $status = self::BATCH_PARTIAL_FAILED;
} }
Db::name('article_reference_relevance_check_batch')->where('id', intval($batchId))->update([ Db::name('article_reference_relevance_check_batch')->where('id', intval($batchId))->update([
'batch_status' => $status, 'batch_status' => $status,
'done_count' => intval($done), 'total_count' => $total,
'failed_count' => intval($failed), 'done_count' => $done,
'failed_count' => $failed,
'updated_at' => date('Y-m-d H:i:s'), 'updated_at' => date('Y-m-d H:i:s'),
]); ]);
if ($total > 0 && ($done + $failed) < $total) { if ($total > 0 && ($done + $failed) < $total) {
$this->svc->log('ReferenceCheckArticleWorker batch_id=' . $batchId . ' incomplete total=' . $total); $this->svc->log('ReferenceCheckArticleWorker batch_id=' . $batchId . ' incomplete total=' . $total . ' done=' . $done . ' failed=' . $failed);
} }
} }