自动查重

application/common/PlagiarismService.php

@@ -12,10 +12,14 @@ use think\Exception;
  * 并维护 t_plagiarism_check 状态机。
  *
  * 状态流：
- *   submit()                 → state=1（上传中），入队 PlagiarismRun
- *   PlagiarismRun.fire       → 上传 + 触发 similarity → state=2（比对中），入队 PlagiarismPoll
- *   PlagiarismPoll.fire      → 轮询 status，完成后下载 PDF → state=3（完成）
- *   任意环节抛异常           → state=4（失败），写 error_msg
+ *   submit()                    → state=1（上传中），入队 PlagiarismRun
+ *   PlagiarismRun               → 创建 submission + 上传文件 → 入队 PlagiarismWaitIngest
+ *   PlagiarismWaitIngest        → 单次 GET submission 状态；就绪则入队 PlagiarismTriggerSimilarity，否则延迟再入队
+ *   PlagiarismTriggerSimilarity → PUT similarity → state=2（比对中），入队 PlagiarismPoll
+ *   PlagiarismPoll              → 轮询 similarity，完成后下载 PDF → state=3（完成）
+ *   任意环节抛异常               → state=4（失败），写 error_msg
+ *
+ * Worker：请用 `queue:work` 消费队列 **plagiarism**（整条链与轮询均在此队列；若此前单独监听 PlagiarismRun / PlagiarismPoll，需改为 plagiarism）。
  */
 class PlagiarismService
 {
@@ -24,6 +28,13 @@ class PlagiarismService
      */
     const REPORT_DIR = 'public/plagiarism';
 
+    /** Run / WaitIngest / TriggerSimilarity / Poll 共用队列名 */
+    const QUEUE_CHAIN = 'plagiarism';
+
+    const JOB_WAIT_INGEST = 'app\\api\\job\\PlagiarismWaitIngest';
+    const JOB_TRIGGER_SIM = 'app\\api\\job\\PlagiarismTriggerSimilarity';
+    const JOB_POLL = 'app\\api\\job\\PlagiarismPoll';
+
     /**
      * 轮询间隔（秒）。Turnitin 一般 1-5 分钟出结果，30 秒一次比较合适
      */
@@ -76,78 +87,160 @@ class PlagiarismService
             'utime'            => $now,
         ]);
         $this->log("submit service act");
-        // 入队执行：上传 + 触发 similarity
         Queue::push(
             'app\\api\\job\\PlagiarismRun',
             ['check_id' => $checkId, 'file_path' => $filePath],
-            'PlagiarismRun'
+            self::QUEUE_CHAIN
         );
 
         return (int)$checkId;
     }
 
     /**
-     * Job 调用：上传文件到 Turnitin 并触发 similarity，然后入队 PlagiarismPoll
+     * Job 调用：仅创建 submission + 上传文件，随后由 PlagiarismWaitIngest 链式轮询 ingest，再 PlagiarismTriggerSimilarity。
+     */
+    public function runUploadOnly($checkId, $filePath)
+    {
+        $check = $this->mustGetCheck($checkId);
+        $this->log('runUploadOnly start check_id=' . $checkId);
+        $tii = new TurnitinService();
+
+        $articleTitle = (string) Db::name('article')
+            ->where('article_id', $check['article_id'])
+            ->value('title');
+        if ($articleTitle === '') {
+            $articleTitle = 'Article #' . $check['article_id'];
+        }
+
+        $createResp = $tii->createSubmission([
+            'title'     => mb_substr($articleTitle, 0, 250),
+            'owner'     => 'editor_' . $check['triggered_by'],
+            'submitter' => 'editor_' . $check['triggered_by'],
+            'metadata'  => [
+                'article_id' => (string) $check['article_id'],
+                'check_id'   => (string) $check['check_id'],
+            ],
+        ]);
+        $submissionId = isset($createResp['id']) ? $createResp['id'] : '';
+        if ($submissionId === '') {
+            throw new Exception('Turnitin createSubmission returned empty id: ' . json_encode($createResp));
+        }
+
+        $this->updateCheck($checkId, [
+            'tii_submission_id' => $submissionId,
+            'raw_response'      => json_encode($createResp, JSON_UNESCAPED_UNICODE),
+        ]);
+
+        $tii->uploadFile($submissionId, $filePath, basename($filePath));
+        $this->log('runUploadOnly uploaded submission_id=' . $submissionId);
+
+        $firstDelay = $this->ingestChainFirstDelaySec();
+        Queue::later(
+            $firstDelay,
+            self::JOB_WAIT_INGEST,
+            ['check_id' => $checkId, 'attempt' => 1],
+            self::QUEUE_CHAIN
+        );
+    }
+
+    /**
+     * 单次 ingest 检查（由 PlagiarismWaitIngest 调用）。不在本方法内 sleep 长循环。
+     */
+    public function runIngestPollStep($checkId, $attempt = 1)
+    {
+        $check = $this->mustGetCheck($checkId);
+        if (empty($check['tii_submission_id'])) {
+            $this->markFailed($checkId, '[ingest] tii_submission_id empty');
+            return;
+        }
+
+        $maxAttempts = $this->ingestChainMaxAttempts();
+        $interval    = $this->ingestChainPollIntervalSec();
+        $tii         = new TurnitinService();
+
+        try {
+            $parsed = $tii->parseSubmissionIngestState($check['tii_submission_id']);
+        } catch (\Throwable $e) {
+            if ($attempt >= $maxAttempts) {
+                $this->markFailed($checkId, '[ingest] request failed after ' . $attempt . ' tries: ' . $e->getMessage());
+                return;
+            }
+            Queue::later($interval, self::JOB_WAIT_INGEST, ['check_id' => $checkId, 'attempt' => $attempt + 1], self::QUEUE_CHAIN);
+            return;
+        }
+
+        if (!empty($parsed['failed'])) {
+            $this->markFailed($checkId, '[ingest] submission failed status=' . $parsed['status'] . ' ' . $parsed['snippet']);
+            return;
+        }
+        if (!empty($parsed['ready'])) {
+            Queue::push(self::JOB_TRIGGER_SIM, ['check_id' => $checkId, 'ingest_attempt' => $attempt], self::QUEUE_CHAIN);
+            return;
+        }
+        if ($attempt >= $maxAttempts) {
+            $this->markFailed($checkId, '[ingest] timeout last_status=' . ($parsed['status'] !== '' ? $parsed['status'] : '(empty)'));
+            return;
+        }
+        Queue::later($interval, self::JOB_WAIT_INGEST, ['check_id' => $checkId, 'attempt' => $attempt + 1], self::QUEUE_CHAIN);
+    }
+
+    /**
+     * 在 ingest 就绪后触发 similarity，并入队 PlagiarismPoll。
+     * 若仍返回 409，则重新入队 PlagiarismWaitIngest（不抛异常，避免误标失败）。
+     *
+     * @param int $ingestAttempt 来自 WaitIngest 的 attempt，供 409 时继续轮询
+     */
+    public function runTriggerSimilarityOnly($checkId, $ingestAttempt = 1)
+    {
+        $check = $this->mustGetCheck($checkId);
+        if (empty($check['tii_submission_id'])) {
+            $this->markFailed($checkId, '[similarity] tii_submission_id empty');
+            return;
+        }
+
+        $tii = new TurnitinService();
+        $sid = $check['tii_submission_id'];
+
+        try {
+            $simResp = $tii->triggerSimilarity($sid);
+        } catch (\Throwable $e) {
+            $msg = $e->getMessage();
+            $is409 = (stripos($msg, '409') !== false || stripos($msg, 'CONFLICT') !== false)
+                && (stripos($msg, 'not been completed') !== false || stripos($msg, 'completed yet') !== false);
+            if ($is409) {
+                $maxAttempts = $this->ingestChainMaxAttempts();
+                $next        = $ingestAttempt + 1;
+                if ($next > $maxAttempts) {
+                    $this->markFailed($checkId, '[similarity] still not ready after ingest attempts: ' . $msg);
+                    return;
+                }
+                $delay = max($this->ingestChainPollIntervalSec(), 20);
+                Queue::later($delay, self::JOB_WAIT_INGEST, ['check_id' => $checkId, 'attempt' => $next], self::QUEUE_CHAIN);
+                return;
+            }
+            throw $e;
+        }
+
+        $this->updateCheck($checkId, [
+            'state'             => 2,
+            'tii_report_status' => 'PROCESSING',
+            'raw_response'      => json_encode($simResp, JSON_UNESCAPED_UNICODE),
+        ]);
+
+        Queue::later(
+            self::POLL_INTERVAL,
+            self::JOB_POLL,
+            ['check_id' => $checkId, 'attempt' => 1],
+            self::QUEUE_CHAIN
+        );
+    }
+
+    /**
+     * @deprecated 与 runUploadOnly 等价；长耗时 ingest 已拆到队列 PlagiarismWaitIngest，勿在本方法内同步 wait。
      */
     public function runUploadAndTrigger($checkId, $filePath)
     {
-        $check = $this->mustGetCheck($checkId);
-        $this->log("runUploadAndTrigger is act0");
-        try {
-            $tii = new TurnitinService();
-
-            // 1. 创建 submission
-            $articleTitle = (string) Db::name('article')
-                ->where('article_id', $check['article_id'])
-                ->value('title');
-            if ($articleTitle === '') {
-                $articleTitle = 'Article #' . $check['article_id'];
-            }
-            $this->log("runUploadAndTrigger is act1");
-            $createResp = $tii->createSubmission([
-                'title'     => mb_substr($articleTitle, 0, 250),
-                'owner'     => 'editor_' . $check['triggered_by'],
-                'submitter' => 'editor_' . $check['triggered_by'],
-                'metadata'  => [
-                    'article_id' => (string)$check['article_id'],
-                    'check_id'   => (string)$check['check_id'],
-                ],
-            ]);
-            $submissionId = isset($createResp['id']) ? $createResp['id'] : '';
-            if ($submissionId === '') {
-                throw new Exception('Turnitin createSubmission returned empty id: ' . json_encode($createResp));
-            }
-
-            $this->updateCheck($checkId, [
-                'tii_submission_id' => $submissionId,
-                'raw_response'      => json_encode($createResp, JSON_UNESCAPED_UNICODE),
-            ]);
-            $this->log("runUploadAndTrigger is act2");
-            // 2. 上传文件
-            $tii->uploadFile($submissionId, $filePath, basename($filePath));
-
-            // 3. 触发 similarity
-            $simResp = $tii->triggerSimilarity($submissionId);
-
-            $this->updateCheck($checkId, [
-                'state'             => 2,                // 比对中
-                'tii_report_status' => 'PROCESSING',
-                'raw_response'      => json_encode($simResp, JSON_UNESCAPED_UNICODE),
-            ]);
-
-            $this->log("runUploadAndTrigger is act3");
-
-            // 4. 排队首次轮询（晚一点开始，让 Turnitin 先处理）
-            Queue::later(
-                self::POLL_INTERVAL,
-                'app\\api\\job\\PlagiarismPoll',
-                ['check_id' => $checkId, 'attempt' => 1],
-                'PlagiarismPoll'
-            );
-        } catch (\Throwable $e) {
-            $this->markFailed($checkId, '[upload] ' . $e->getMessage());
-            throw $e;
-        }
+        $this->runUploadOnly($checkId, $filePath);
     }
 
     /**
@@ -204,18 +297,18 @@ class PlagiarismService
             }
             Queue::later(
                 self::POLL_INTERVAL,
-                'app\\api\\job\\PlagiarismPoll',
+                self::JOB_POLL,
                 ['check_id' => $checkId, 'attempt' => $attempt + 1],
-                'plagiarism'
+                self::QUEUE_CHAIN
             );
         } catch (\Throwable $e) {
             // 网络抖动不要直接 fail，给一定容错次数
             if ($attempt < self::MAX_POLL_ATTEMPTS) {
                 Queue::later(
                     self::POLL_INTERVAL,
-                    'app\\api\\job\\PlagiarismPoll',
+                    self::JOB_POLL,
                     ['check_id' => $checkId, 'attempt' => $attempt + 1],
-                    'plagiarism'
+                    self::QUEUE_CHAIN
                 );
                 $this->updateCheck($checkId, [
                     'attempts'  => $attempt,
@@ -328,15 +421,30 @@ class PlagiarismService
         Db::name('plagiarism_check')->where('check_id', $checkId)->update($data);
     }
 
-    private function markFailed($checkId, $errMsg)
+    public function markFailed($checkId, $errMsg)
     {
-        $this->log("markFailed act");
+        $this->log('markFailed check_id=' . $checkId);
         $this->updateCheck($checkId, [
             'state'     => 4,
             'error_msg' => mb_substr($errMsg, 0, 1000),
         ]);
     }
 
+    private function ingestChainFirstDelaySec()
+    {
+        return max(3, (int) Env::get('turnitin.ingest_chain_first_delay', 10));
+    }
+
+    private function ingestChainPollIntervalSec()
+    {
+        return max(5, (int) Env::get('turnitin.ingest_chain_poll_interval', 15));
+    }
+
+    private function ingestChainMaxAttempts()
+    {
+        return max(10, (int) Env::get('turnitin.ingest_chain_max_attempts', 80));
+    }
+
     /**
      * 从 t_article_file 找到投稿主稿（manuscirpt）的本地绝对路径。
      * file_url 在系统里可能是 URL 或相对路径，调用方负责保证可读。