自动查重

This commit is contained in:
wangjinlei
2026-05-13 18:02:09 +08:00
parent fa878334cd
commit f99dbc6397
6 changed files with 411 additions and 91 deletions

View File

@@ -2,9 +2,11 @@
namespace app\api\controller;
use app\common\TurnitinService;
use think\Db;
use think\Response;
use app\common\PlagiarismService;
use think\Validate;
/**
* 论文查重Turnitin / Crossref Similarity Check控制器。
@@ -51,7 +53,6 @@ class Plagiarism extends Base
$localPath = $fileUrl !== ''
? $svc->resolveFileUrlToLocal($fileUrl)
: $svc->locateArticleManuscript($articleId);
echo $localPath;
$checkId = $svc->submit($articleId, $localPath, $editorId, 'manual');
return jsonSuccess(['check_id' => $checkId]);
} catch (\Throwable $e) {
@@ -60,12 +61,28 @@ class Plagiarism extends Base
}
public function testccone(){
/**
* 调试与线上一致走队列链upload → wait ingest → trigger → poll需 worker 消费 plagiarism 队列。
*/
public function testccone()
{
$svc = new PlagiarismService();
$checkId = 9;
$filePath = "/home/wwwroot/api.tmrjournals.com/public/manuscirpt/20260509/6832a56e8ace38fe99df390ab5221deb.docx";
$svc->runUploadAndTrigger($checkId,$filePath);
$filePath = '/home/wwwroot/api.tmrjournals.com/public/manuscirpt/20260509/6832a56e8ace38fe99df390ab5221deb.docx';
$svc->runUploadOnly($checkId, $filePath);
}
public function testcconegetstatus(){
$data = $this->request->post();
$rule = new Validate([
"id"=>"require"
]);
if(!$rule->check($data)){
return jsonError($rule->getError());
}
$tii = new TurnitinService();
$res = $tii->getSubmission($data['id']);
return jsonSuccess($res);
}
/**

View File

@@ -6,9 +6,9 @@ use think\queue\Job;
use app\common\PlagiarismService;
/**
* 队列任务:上传论文到 Turnitin + 触发 similarity 检测
* 队列任务:创建 Turnitin submission 并上传原稿ingest 轮询与触发 similarity 由后续 Job 完成
*
* 完成后会自动入队 PlagiarismPoll 进行后续轮询。
* PlagiarismRun → PlagiarismWaitIngest → PlagiarismTriggerSimilarity → PlagiarismPoll
*
* data:
* - check_id t_plagiarism_check.check_id
@@ -29,8 +29,12 @@ class PlagiarismRun
return;
}
$svc = new PlagiarismService();
$svc->log("PlagiarismRun job act!!");
$svc->runUploadAndTrigger($checkId, $filePath);
$svc->log('PlagiarismRun job act check_id=' . $checkId);
try {
$svc->runUploadOnly($checkId, $filePath);
} catch (\Throwable $e) {
$svc->markFailed($checkId, '[upload] ' . $e->getMessage());
}
$job->delete();
}
}

View File

@@ -0,0 +1,33 @@
<?php
namespace app\api\job;
use think\queue\Job;
use app\common\PlagiarismService;
/**
* 队列任务:在 ingest 就绪后调用 PUT /similarity并入队 PlagiarismPoll。
*
* data:
* - check_id t_plagiarism_check.check_id
* - ingest_attempt 来自 PlagiarismWaitIngest 的 attempt409 时用于继续轮询 ingest
*/
class PlagiarismTriggerSimilarity
{
public function fire(Job $job, $data)
{
$checkId = isset($data['check_id']) ? (int) $data['check_id'] : 0;
$ingestAttempt = isset($data['ingest_attempt']) ? (int) $data['ingest_attempt'] : 1;
if ($checkId <= 0) {
$job->delete();
return;
}
$svc = new PlagiarismService();
try {
$svc->runTriggerSimilarityOnly($checkId, $ingestAttempt);
} catch (\Throwable $e) {
$svc->markFailed($checkId, '[similarity] ' . $e->getMessage());
}
$job->delete();
}
}

View File

@@ -0,0 +1,33 @@
<?php
namespace app\api\job;
use think\queue\Job;
use app\common\PlagiarismService;
/**
* 队列任务:单次查询 Turnitin submission 是否解析完成ingest未完成则延迟再次入队。
*
* data:
* - check_id t_plagiarism_check.check_id
* - attempt 从 1 递增
*/
class PlagiarismWaitIngest
{
public function fire(Job $job, $data)
{
$checkId = isset($data['check_id']) ? (int) $data['check_id'] : 0;
$attempt = isset($data['attempt']) ? (int) $data['attempt'] : 1;
if ($checkId <= 0) {
$job->delete();
return;
}
$svc = new PlagiarismService();
try {
$svc->runIngestPollStep($checkId, $attempt);
} catch (\Throwable $e) {
$svc->markFailed($checkId, '[ingest] ' . $e->getMessage());
}
$job->delete();
}
}