自动查重

2026-05-18 18:34:48 +08:00
parent f99dbc6397
commit 53e6ddbd9e
7 changed files with 105 additions and 44 deletions
--- a/application/api/controller/EmailClient.php
+++ b/application/api/controller/EmailClient.php
@@ -2692,8 +2692,10 @@ class EmailClient extends Base
     * 根据 expert_type 分发选人逻辑
     *
     * - expert_type = 5：从 t_expert 库选人（按领域 / 国家 / 频次）
+     *   频次：e.ltime（成功发送后回写）+ t_promotion_email_log 中「待发送 state=0 的入队时间 ctime」
+     *   （避免「今日生成任务明日发送」时 ltime 未变导致连续两天选到同一拨人）
     * - expert_type ∈ {1,2,3,4}：从系统内部表选人（主编/编委/青年编委/作者），fields 与国家筛选忽略；
-     *   频次按 t_promotion_email_log 中相同 expert_type 维度的最近发送时间扣除
+     *   频次按 t_promotion_email_log：已发/退信用 send_time；待发送队列用 ctime（同上）
     *
     * 返回行 shape 已对齐：
     *   - type=5 行包含 e.* 全部字段（含 expert_id、country_id、ltime 等）
@@ -2741,9 +2743,25 @@ class EmailClient extends Base

        if ($noRepeatDays > 0) {
            $cutoff = time() - ($noRepeatDays * 86400);
+            // ltime：成功发出后回写；与 log 中 state=1 在「已送达」上部分重叠，但保留 ltime 可走索引、且退信 state=3 未必回写 ltime。
            $query->where(function ($q) use ($cutoff) {
                $q->where('e.ltime', 0)->whereOr('e.ltime', '<', $cutoff);
            });
+            // 一条 NOT EXISTS：待发(state=0 按 ctime) 或 已发/退信(按 send_time)，避免两段相同 join 的重复感
+            $query->where(function ($q) use ($cutoff) {
+                $q->table('t_promotion_email_log')->alias('pl')
+                    ->join('t_promotion_task pt', 'pt.task_id = pl.task_id', 'inner')
+                    ->where('pt.expert_type', 5)
+                    ->where('pl.expert_id', '>', 0)
+                    ->whereRaw('pl.expert_id = e.expert_id')
+                    ->where(function ($w) use ($cutoff) {
+                        $w->where(function ($a) use ($cutoff) {
+                            $a->where('pl.state', 0)->where('pl.ctime', '>', $cutoff);
+                        })->whereOr(function ($b) use ($cutoff) {
+                            $b->where('pl.state', 'in', [1, 3])->where('pl.send_time', '>', $cutoff);
+                        });
+                    });
+            }, 'not exists');
        }

        $countryIds = $this->resolveCountryIds($targetPartitions, $targetCountryIds);
@@ -2762,7 +2780,7 @@ class EmailClient extends Base
     * 系统内部受众选人（编委 / 主编 / 青年编委 / 作者）
     * 仅按 期刊 + 频次 过滤；领域 / 国家无关
     *
-     * 频次：扣除「同 expert_type 维度下，no_repeat_days 内已经发出 (state=1) 或退信 (state=3) 的人」
+     * 频次：扣除「同 expert_type 下，no_repeat_days 内 (1) 已发出或退信，或 (2) 仍在队列待发送（state=0，按 ctime）」的人
     *
     * @param int $expertType   1=主编 2=编委 3=青年编委 4=作者
     * @param int $journalId
@@ -2807,15 +2825,18 @@ class EmailClient extends Base
        if ($noRepeatDays > 0) {
            $cutoff         = intval(time() - ($noRepeatDays * 86400));
            $expertTypeSafe = intval($expertType);
-            // 关联子查询：相对于 NOT IN，避免把全部已发 user_id 拉到 PHP 再拼回 SQL；
-            // 配合 t_promotion_email_log(user_id, send_time) 复合索引做半连接探针，常量时间。
            $query->where(function ($q) use ($expertTypeSafe, $cutoff) {
                $q->table('t_promotion_email_log')->alias('l')
-                  ->join('t_promotion_task t', 't.task_id = l.task_id', 'inner')
-                  ->where('t.expert_type', $expertTypeSafe)
-                  ->where('l.state', 'in', [1, 3])
-                  ->where('l.send_time', '>', $cutoff)
-                  ->whereRaw('l.user_id = u.user_id');
+                    ->join('t_promotion_task t', 't.task_id = l.task_id', 'inner')
+                    ->where('t.expert_type', $expertTypeSafe)
+                    ->whereRaw('l.user_id = u.user_id')
+                    ->where(function ($w) use ($cutoff) {
+                        $w->where(function ($a) use ($cutoff) {
+                            $a->where('l.state', 0)->where('l.ctime', '>', $cutoff);
+                        })->whereOr(function ($b) use ($cutoff) {
+                            $b->where('l.state', 'in', [1, 3])->where('l.send_time', '>', $cutoff);
+                        });
+                    });
            }, 'not exists');
        }

--- a/application/api/controller/Plagiarism.php
+++ b/application/api/controller/Plagiarism.php
@@ -81,10 +81,49 @@ class Plagiarism extends Base
            return jsonError($rule->getError());
        }
        $tii = new TurnitinService();
-        $res = $tii->getSubmission($data['id']);
+        $res = $tii->parseSubmissionIngestState($data['id']);
        return jsonSuccess($res);
    }

+    public function testcconewait(){
+        $data = $this->request->post();
+        $rule = new Validate([
+            "checkId"=>"require"
+        ]);
+        if(!$rule->check($data)){
+            return jsonError($rule->getError());
+        }
+        $svc = new PlagiarismService();
+        $res = $svc->runIngestPollStep($data['checkId']);
+        return jsonSuccess($res);
+    }
+
+    public function testcconesimilar(){
+        $data = $this->request->post();
+        $rule = new Validate([
+            "checkId"=>"require"
+        ]);
+        if(!$rule->check($data)){
+            return jsonError($rule->getError());
+        }
+        $svc = new PlagiarismService();
+        $res = $svc->runTriggerSimilarityOnly($data['checkId']);
+        return jsonSuccess($res);
+    }
+
+    public function testcconelast(){
+        $data = $this->request->post();
+        $rule = new Validate([
+            "checkId"=>"require"
+        ]);
+        if(!$rule->check($data)){
+            return jsonError($rule->getError());
+        }
+        $svc = new PlagiarismService();
+        $re = $svc->runPollStatus($data['checkId']);
+        return jsonSuccess($re);
+    }
+
    /**
     * 重试 = 提交一次新查重（保留历史）
     */
--- a/application/api/job/PlagiarismPoll.php
+++ b/application/api/job/PlagiarismPoll.php
@@ -23,15 +23,16 @@ class PlagiarismPoll
    public function fire(Job $job, $data)
    {

-        $checkId = isset($data['check_id']) ? intval($data['check_id']) : 0;
-        $attempt = isset($data['attempt'])  ? intval($data['attempt'])  : 1;
-
-        if ($checkId <= 0) {
-            $job->delete();
-            return;
-        }
+//        $checkId = isset($data['check_id']) ? intval($data['check_id']) : 0;
+//        $attempt = isset($data['attempt'])  ? intval($data['attempt'])  : 1;
+//
+//        if ($checkId <= 0) {
+//            $job->delete();
+//            return;
+//        }
        $svc = new PlagiarismService();
-        $svc->runPollStatus($checkId, $attempt);
+        $svc->log("PlagiarismPoll job is running");
+//        $svc->runPollStatus($checkId, $attempt);
        $job->delete();
    }
 }
--- a/application/api/job/PlagiarismRun.php
+++ b/application/api/job/PlagiarismRun.php
@@ -29,7 +29,7 @@ class PlagiarismRun
            return;
        }
        $svc = new PlagiarismService();
-        $svc->log('PlagiarismRun job act check_id=' . $checkId);
+        $svc->log('PlagiarismRun job is running');
        try {
            $svc->runUploadOnly($checkId, $filePath);
        } catch (\Throwable $e) {
--- a/application/api/job/PlagiarismTriggerSimilarity.php
+++ b/application/api/job/PlagiarismTriggerSimilarity.php
@@ -23,6 +23,7 @@ class PlagiarismTriggerSimilarity
            return;
        }
        $svc = new PlagiarismService();
+        $svc->log("PlagiarismTriggerSimilarity job is running");
        try {
            $svc->runTriggerSimilarityOnly($checkId, $ingestAttempt);
        } catch (\Throwable $e) {
--- a/application/api/job/PlagiarismWaitIngest.php
+++ b/application/api/job/PlagiarismWaitIngest.php
@@ -23,6 +23,7 @@ class PlagiarismWaitIngest
            return;
        }
        $svc = new PlagiarismService();
+        $svc->log("PlagiarismWaitIngest job is running");
        try {
            $svc->runIngestPollStep($checkId, $attempt);
        } catch (\Throwable $e) {
--- a/application/common/PlagiarismService.php
+++ b/application/common/PlagiarismService.php
@@ -73,7 +73,7 @@ class PlagiarismService
            ->where('article_id', $articleId)
            ->value('journal_id');

-
+        $this->log("plagiarism submit is running");
        $now = time();
        $checkId = Db::name('plagiarism_check')->insertGetId([
            'article_id'       => $articleId,
@@ -86,7 +86,6 @@ class PlagiarismService
            'ctime'            => $now,
            'utime'            => $now,
        ]);
-        $this->log("submit service act");
        Queue::push(
            'app\\api\\job\\PlagiarismRun',
            ['check_id' => $checkId, 'file_path' => $filePath],
@@ -132,8 +131,6 @@ class PlagiarismService
        ]);

        $tii->uploadFile($submissionId, $filePath, basename($filePath));
-        $this->log('runUploadOnly uploaded submission_id=' . $submissionId);
-
        $firstDelay = $this->ingestChainFirstDelaySec();
        Queue::later(
            $firstDelay,
@@ -153,7 +150,7 @@ class PlagiarismService
            $this->markFailed($checkId, '[ingest] tii_submission_id empty');
            return;
        }
-
+        $this->log("runIngestPollStep is running");
        $maxAttempts = $this->ingestChainMaxAttempts();
        $interval    = $this->ingestChainPollIntervalSec();
        $tii         = new TurnitinService();
@@ -198,6 +195,7 @@ class PlagiarismService
            return;
        }

+        $this->log("runTriggerSimilarityOnly is running");
        $tii = new TurnitinService();
        $sid = $check['tii_submission_id'];

@@ -254,7 +252,7 @@ class PlagiarismService
            return;
        }

-        try {
+//        try {
            $tii = new TurnitinService();
            $statusResp = $tii->getSimilarityStatus($check['tii_submission_id']);
            $status = isset($statusResp['status']) ? strtoupper($statusResp['status']) : '';
@@ -301,24 +299,24 @@ class PlagiarismService
                ['check_id' => $checkId, 'attempt' => $attempt + 1],
                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,
-                    self::JOB_POLL,
-                    ['check_id' => $checkId, 'attempt' => $attempt + 1],
-                    self::QUEUE_CHAIN
-                );
-                $this->updateCheck($checkId, [
-                    'attempts'  => $attempt,
-                    'error_msg' => '[poll] transient: ' . $e->getMessage(),
-                ]);
-                return;
-            }
-            $this->markFailed($checkId, '[poll] exhausted: ' . $e->getMessage());
-            throw $e;
-        }
+//        } 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,
+//                    self::JOB_POLL,
+//                    ['check_id' => $checkId, 'attempt' => $attempt + 1],
+//                    self::QUEUE_CHAIN
+//                );
+//                $this->updateCheck($checkId, [
+//                    'attempts'  => $attempt,
+//                    'error_msg' => '[poll] transient: ' . $e->getMessage(),
+//                ]);
+//                return;
+//            }
+//            $this->markFailed($checkId, '[poll] exhausted: ' . $e->getMessage());
+//            throw $e;
+//        }
    }

    /**
@@ -437,7 +435,7 @@ class PlagiarismService

    private function ingestChainPollIntervalSec()
    {
-        return max(5, (int) Env::get('turnitin.ingest_chain_poll_interval', 15));
+        return max(60, (int) Env::get('turnitin.ingest_chain_poll_interval', 15));
    }

    private function ingestChainMaxAttempts()