343 lines
12 KiB
PHP
343 lines
12 KiB
PHP
<?php
|
||
|
||
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。
|
||
*
|
||
* 触发方式：纯手工（编辑后台点"查重"按钮）。
|
||
* 报告策略：PDF 在 poll 完成时落盘；在线 viewer URL 通过 getReportUrl 按需生成（临时签名）。
|
||
*
|
||
* 主要接口:
|
||
* POST submit 触发查重
|
||
* GET getStatus 轮询单条查重状态（前端 ajax）
|
||
* GET getList 列出某 article 的全部查重记录
|
||
* GET getReportUrl 获取/刷新在线查看 URL
|
||
* GET downloadReport 下载本地 PDF
|
||
* POST retry 重新触发（创建新行）
|
||
* GET features 探活（开发调试用）
|
||
*/
|
||
class Plagiarism extends Base
|
||
{
|
||
public function __construct(\think\Request $request = null)
|
||
{
|
||
parent::__construct($request);
|
||
}
|
||
|
||
/**
|
||
* 触发查重
|
||
*
|
||
* 入参:
|
||
* article_id 必填
|
||
* file_url 选填；不传则按 article_id 在 t_article_file 找 manuscirpt
|
||
* editor_id 选填；触发人 user_id（前端拿不到也可以传 0）
|
||
* type 选填；full（默认全文）| body_only（正文）| both（各提交一条）
|
||
*/
|
||
public function submit()
|
||
{
|
||
$articleId = intval($this->request->param('article_id', 0));
|
||
$fileUrl = trim($this->request->param('file_url', ''));
|
||
$editorId = intval($this->request->param('editor_id', 0));
|
||
$checkType = trim($this->request->param('type', 'full'));
|
||
|
||
if ($articleId <= 0) {
|
||
return jsonError('article_id required');
|
||
}
|
||
|
||
try {
|
||
$svc = new PlagiarismService();
|
||
$localPath = $fileUrl !== ''
|
||
? $svc->resolveFileUrlToLocal($fileUrl)
|
||
: $svc->locateArticleManuscript($articleId);
|
||
if (strtolower($checkType) === 'both') {
|
||
$ids = $svc->submitBoth($articleId, $localPath, $editorId, 'manual');
|
||
return jsonSuccess($ids);
|
||
}
|
||
$checkId = $svc->submit($articleId, $localPath, $editorId, 'manual', $checkType);
|
||
return jsonSuccess(['check_id' => $checkId, 'check_type' => strtolower($checkType) ?: 'full']);
|
||
} catch (\Throwable $e) {
|
||
return jsonError($e->getMessage());
|
||
}
|
||
}
|
||
|
||
|
||
/**
|
||
* 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';
|
||
$svc->runUploadOnly($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 = $tii->parseSubmissionIngestState($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）
|
||
*/
|
||
public function retry()
|
||
{
|
||
return $this->submit();
|
||
}
|
||
|
||
/**
|
||
* 取单条查重状态
|
||
*/
|
||
public function getStatus()
|
||
{
|
||
$checkId = intval($this->request->param('check_id', 0));
|
||
if ($checkId <= 0) {
|
||
return jsonError('check_id required');
|
||
}
|
||
$row = Db::name('plagiarism_check')->where('check_id', $checkId)->find();
|
||
if (!$row) {
|
||
return jsonError('not found');
|
||
}
|
||
$out = $this->formatRow($row);
|
||
if (!empty($row['raw_response'])) {
|
||
$raw = json_decode($row['raw_response'], true);
|
||
if (is_array($raw)) {
|
||
$out['similarity_meta'] = \app\common\TurnitinService::parseSimilarityReportMeta($raw);
|
||
}
|
||
}
|
||
$out['report_view_hint'] = 'PDF 多为 Match Overview 汇总样式；按来源库(Internet/Publication/Crossref)分类请用 getReportUrl 打开在线报告并切到 All Sources';
|
||
return jsonSuccess($out);
|
||
}
|
||
|
||
/**
|
||
* 列出某 article 的全部查重记录（按时间倒序）
|
||
*/
|
||
public function getList()
|
||
{
|
||
$articleId = intval($this->request->param('article_id', 0));
|
||
if ($articleId <= 0) {
|
||
return jsonError('article_id required');
|
||
}
|
||
$rows = Db::name('plagiarism_check')
|
||
->where('article_id', $articleId)
|
||
->order('check_id desc')
|
||
->select();
|
||
$out = [];
|
||
foreach ($rows as $r) {
|
||
$out[] = $this->formatRow($r);
|
||
}
|
||
return jsonSuccess(['list' => $out]);
|
||
}
|
||
|
||
/**
|
||
* 取在线查看 URL（Turnitin 一次性会话链接，关闭报告页后勿复用旧 URL）
|
||
*
|
||
* 入参:
|
||
* check_id 必填
|
||
* editor_id 选填，当前打开报告的编辑 user_id（与 viewer_user_id 对应，避免 session 认证失败）
|
||
* reuse 选填，1=在未过期时复用库内缓存；默认 0，每次调用重新向 Turnitin 申请
|
||
*/
|
||
public function getReportUrl()
|
||
{
|
||
$checkId = intval($this->request->param('check_id', 0));
|
||
$editorId = intval($this->request->param('editor_id', 0));
|
||
$reuse = intval($this->request->param('reuse', 0)) === 1;
|
||
if ($checkId <= 0) {
|
||
return jsonError('check_id required');
|
||
}
|
||
try {
|
||
$row = Db::name('plagiarism_check')->where('check_id', $checkId)->find();
|
||
if (!$row) {
|
||
return jsonError('not found');
|
||
}
|
||
if ($row['state'] != 3) {
|
||
return jsonError('check not completed yet, state=' . $row['state']);
|
||
}
|
||
$viewerContext = [];
|
||
if ($editorId > 0) {
|
||
$viewerContext['editor_id'] = $editorId;
|
||
}
|
||
$needRefresh = !$reuse
|
||
|| empty($row['view_only_url'])
|
||
|| intval($row['view_only_url_expire']) < time() + 60;
|
||
|
||
$usageHint = '每次打开请先调用本接口获取新链接；勿收藏或再次打开旧链接。请在新标签页打开，并允许 Turnitin 域名 Cookie。';
|
||
|
||
if ($needRefresh) {
|
||
$svc = new PlagiarismService();
|
||
$info = $svc->refreshViewerUrlFor($checkId, $viewerContext);
|
||
if ($info['url'] === '') {
|
||
return jsonError('Turnitin returned empty viewer_url');
|
||
}
|
||
return jsonSuccess([
|
||
'view_only_url' => $info['url'],
|
||
'expire' => $info['expire'],
|
||
'has_pdf' => !empty($info['local_pdf']),
|
||
'viewer_user_id' => $info['viewer_user_id'],
|
||
'refreshed' => true,
|
||
'usage_hint' => $usageHint,
|
||
]);
|
||
}
|
||
return jsonSuccess([
|
||
'view_only_url' => $row['view_only_url'],
|
||
'expire' => intval($row['view_only_url_expire']),
|
||
'has_pdf' => !empty($row['pdf_local_path']),
|
||
'refreshed' => false,
|
||
'usage_hint' => $usageHint,
|
||
]);
|
||
} catch (\Throwable $e) {
|
||
if (!empty($row['pdf_local_path'])) {
|
||
return jsonSuccess([
|
||
'view_only_url' => '',
|
||
'expire' => 0,
|
||
'has_pdf' => true,
|
||
'viewer_error' => $e->getMessage(),
|
||
'hint' => '在线报告暂不可用，请使用 downloadReport 下载 PDF',
|
||
]);
|
||
}
|
||
return jsonError($e->getMessage());
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 直接吐 PDF 二进制流给浏览器下载
|
||
*/
|
||
public function downloadReport()
|
||
{
|
||
$checkId = intval($this->request->param('check_id', 0));
|
||
if ($checkId <= 0) {
|
||
return jsonError('check_id required');
|
||
}
|
||
$row = Db::name('plagiarism_check')->where('check_id', $checkId)->find();
|
||
if (!$row || empty($row['pdf_local_path'])) {
|
||
return jsonError('report not available');
|
||
}
|
||
$rootDir = ROOT_PATH ?: dirname(dirname(dirname(__DIR__)));
|
||
$abs = rtrim($rootDir, '/\\') . DIRECTORY_SEPARATOR . str_replace(['/', '\\'], DIRECTORY_SEPARATOR, $row['pdf_local_path']);
|
||
if (!is_file($abs)) {
|
||
return jsonError('pdf file missing on disk: ' . $row['pdf_local_path']);
|
||
}
|
||
$filename = sprintf('plagiarism_check_%d_article_%d.pdf', $row['check_id'], $row['article_id']);
|
||
return Response::create(file_get_contents($abs), 'html', 200, [
|
||
'Content-Type' => 'application/pdf',
|
||
'Content-Disposition' => 'attachment; filename="' . $filename . '"',
|
||
'Content-Length' => (string)filesize($abs),
|
||
]);
|
||
}
|
||
|
||
/**
|
||
* Turnitin 探活（开发调试用）
|
||
*/
|
||
public function features()
|
||
{
|
||
try {
|
||
$tii = new \app\common\TurnitinService();
|
||
return jsonSuccess($tii->featuresEnabled());
|
||
} catch (\Throwable $e) {
|
||
return jsonError($e->getMessage());
|
||
}
|
||
}
|
||
|
||
// ---------- 内部 ----------
|
||
|
||
private function formatRow($r)
|
||
{
|
||
return [
|
||
'check_id' => intval($r['check_id']),
|
||
'article_id' => intval($r['article_id']),
|
||
'journal_id' => intval($r['journal_id']),
|
||
'state' => intval($r['state']),
|
||
'state_label' => $this->stateLabel($r['state']),
|
||
'similarity_score' => floatval($r['similarity_score']),
|
||
'tii_report_status' => (string)$r['tii_report_status'],
|
||
'has_pdf' => !empty($r['pdf_local_path']),
|
||
'local_pdf_url' => $r['pdf_local_path'],
|
||
'has_viewer_url' => !empty($r['view_only_url']) && intval($r['view_only_url_expire']) > time(),
|
||
'attempts' => intval($r['attempts']),
|
||
'error_msg' => (string)$r['error_msg'],
|
||
'source_file_name' => (string)$r['source_file_name'],
|
||
'check_type' => (string)($r['check_type'] ?? 'full'),
|
||
'check_type_label' => $this->checkTypeLabel($r['check_type'] ?? 'full'),
|
||
'derived_file_path'=> (string)($r['derived_file_path'] ?? ''),
|
||
'trigger_source' => (string)$r['trigger_source'],
|
||
'triggered_by' => intval($r['triggered_by']),
|
||
'ctime' => intval($r['ctime']),
|
||
'utime' => intval($r['utime']),
|
||
];
|
||
}
|
||
|
||
private function checkTypeLabel($checkType)
|
||
{
|
||
$t = strtolower(trim((string) $checkType));
|
||
if ($t === 'body_only' || $t === 'body') {
|
||
return '正文查重';
|
||
}
|
||
return '全文查重';
|
||
}
|
||
|
||
private function stateLabel($state)
|
||
{
|
||
$map = [
|
||
0 => '待上传',
|
||
1 => '上传中',
|
||
2 => '比对中',
|
||
3 => '完成',
|
||
4 => '失败',
|
||
];
|
||
return isset($map[$state]) ? $map[$state] : 'unknown';
|
||
}
|
||
}
|