支持文件元数据透传 (#414)

* 支持界面展示报错

* 支持文件元数据透传

Author: hhhhsc701

frontend/src/i18n/locales/en/common.json

@@ -1432,6 +1432,8 @@
         "processedFileType": "Processed File Type",
         "beforeSize": "Before Size",
         "afterSize": "After Size",
+        "result": "Result",
+        "resultDetail": "Result Detail",
         "status": "Status",
         "actions": "Actions",
         "searchFileName": "Search file name",

frontend/src/i18n/locales/zh/common.json

@@ -1432,6 +1432,8 @@
         "processedFileType": "处理后文件类型",
         "beforeSize": "处理前大小",
         "afterSize": "处理后大小",
+        "result": "处理结果",
+        "resultDetail": "处理结果详情",
         "status": "状态",
         "actions": "操作",
         "searchFileName": "搜索文件名",

frontend/src/pages/DataCleansing/Detail/components/FileTable.tsx

@@ -262,6 +262,53 @@ export default function FileTable({result, fetchTaskResult}) {
         <span className="font-mono text-sm">{formatFileSize(number)}</span>
       ),
     },
+    {
+      title: t("dataCleansing.detail.fileTable.result"),
+      dataIndex: "result",
+      key: "result",
+      width: 200,
+      render: (text: string) => {
+        if (!text) return <span className="text-gray-400">-</span>;
+
+        try {
+          const parsed = JSON.parse(text);
+          const jsonString = JSON.stringify(parsed, null, 2);
+          const displayText = typeof parsed === 'object'
+            ? (Array.isArray(parsed) ? `[${parsed.length} items]` : '{...}')
+            : String(parsed);
+
+          return (
+            <Popover
+              content={
+                <pre className="max-w-md max-h-64 overflow-auto text-xs bg-gray-50 p-2 rounded">
+                  {jsonString}
+                </pre>
+              }
+              title={t("dataCleansing.detail.fileTable.resultDetail")}
+              trigger="click"
+            >
+              <span className="font-mono text-sm text-blue-600 cursor-pointer hover:text-blue-800">
+                {displayText}
+              </span>
+            </Popover>
+          );
+        } catch {
+          const displayText = text.length > 30 ? text.substring(0, 30) + '...' : text;
+          return (
+            <Popover
+              content={<div className="max-w-md text-sm">{text}</div>}
+              title={t("dataCleansing.detail.fileTable.resultDetail")}
+              trigger="click"
+              disabled={text.length <= 30}
+            >
+              <span className="font-mono text-sm cursor-default">
+                {displayText}
+              </span>
+            </Popover>
+          );
+        }
+      },
+    },
     {
       title: t("dataCleansing.detail.fileTable.status"),
       dataIndex: "status",

runtime/datamate-python/app/module/cleaning/service/cleaning_task_service.py

@@ -326,7 +326,7 @@ async def scan_dataset(
         target_file_path.parent.mkdir(parents=True, exist_ok=True)
 
         query = text("""
-            SELECT id, file_name, file_path, file_type, file_size
+            SELECT id, file_name, file_path, file_type, file_size, metadata
             FROM t_dm_dataset_files
             WHERE dataset_id = :dataset_id
             ORDER BY created_at
@@ -340,12 +340,22 @@ async def scan_dataset(
                 if succeed_files and file.id in succeed_files:
                     continue
 
+                metadata_dict = {}
+                if file.metadata:
+                    try:
+                        parsed = json.loads(file.metadata)
+                        if isinstance(parsed, dict):
+                            metadata_dict = parsed
+                    except (json.JSONDecodeError, TypeError):
+                        pass
+
                 file_info = {
                     "fileId": file.id,
                     "fileName": file.file_name,
                     "filePath": file.file_path,
                     "fileType": file.file_type,
                     "fileSize": file.file_size,
+                    "metadata": metadata_dict,
                 }
                 f.write(json.dumps(file_info, ensure_ascii=False) + "\n")
 

runtime/python-executor/datamate/wrappers/executor.py

@@ -43,6 +43,8 @@ def load_meta(self, line):
             meta["sourceFileType"] = meta.get("fileType")
         if meta.get("fileSize"):
             meta["sourceFileSize"] = meta.get("fileSize")
+        else:
+            meta["sourceFileSize"] = 0
         if not meta.get("totalPageNum"):
             meta["totalPageNum"] = 0
         if not meta.get("extraFilePath"):