Add PACS UPP OCR workflow

UPP列表处理/数据处理工作区/01_任务配置.json

@@ -0,0 +1,135 @@
+{
+  "project_name": "PACS UPP列表处理",
+  "record_name": "UPP列表记录",
+  "input_root": "待处理-UPP列表图片集群",
+  "processed_input_root": "已处理-UPP列表图片集群",
+  "result_root": "数据处理结果区",
+  "result_suffix": "-列表归档结果",
+  "unique_key": "",
+  "missing_unique_key_action": "keep_for_review",
+  "duplicate_unique_key_action": "keep_later",
+  "fields": [
+    {
+      "name": "姓名",
+      "required": true,
+      "type": "text",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "患者姓名，按截图显示保留拼音或符号。"
+    },
+    {
+      "name": "性别",
+      "required": true,
+      "type": "text",
+      "clean": ["strip", "upper"],
+      "pattern": "^[MF男女]$",
+      "description": "性别。"
+    },
+    {
+      "name": "年龄",
+      "required": false,
+      "type": "text",
+      "clean": ["remove_spaces", "upper"],
+      "pattern": "",
+      "description": "年龄，保留原始单位，例如 049Y。"
+    },
+    {
+      "name": "患者号",
+      "required": false,
+      "type": "text",
+      "clean": ["remove_spaces"],
+      "pattern": "",
+      "description": "患者号，截图中可能带省略号。"
+    },
+    {
+      "name": "检查号",
+      "required": true,
+      "type": "text",
+      "clean": ["remove_spaces", "upper"],
+      "pattern": "",
+      "description": "检查号。"
+    },
+    {
+      "name": "检查日期",
+      "required": false,
+      "type": "datetime",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "检查日期时间。"
+    },
+    {
+      "name": "任务创建时间",
+      "required": false,
+      "type": "datetime",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "UPP 任务创建时间。"
+    },
+    {
+      "name": "检查描述",
+      "required": false,
+      "type": "text",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "检查描述。"
+    },
+    {
+      "name": "检查设备",
+      "required": false,
+      "type": "text",
+      "clean": ["strip", "upper"],
+      "pattern": "",
+      "description": "检查设备。"
+    },
+    {
+      "name": "算法模型",
+      "required": false,
+      "type": "text",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "算法模型。"
+    },
+    {
+      "name": "状态",
+      "required": false,
+      "type": "text",
+      "clean": ["strip"],
+      "pattern": "",
+      "description": "处理状态。"
+    }
+  ],
+  "classification": {
+    "enabled": true,
+    "category_1_name": "业务分类1",
+    "category_2_name": "业务分类2",
+    "default": {
+      "业务分类1": "PACS",
+      "业务分类2": "UPP列表"
+    },
+    "folder_rules": [
+      {
+        "contains": "肝胆外科",
+        "业务分类1": "PACS",
+        "业务分类2": "肝胆外科"
+      }
+    ]
+  },
+  "ocr": {
+    "engine": "table-v3",
+    "region": "ap-shanghai",
+    "batch_size": 1,
+    "image_padding_y": 0,
+    "rows_per_image": 20,
+    "row_height_px": 0,
+    "skip_header_rows": 0,
+    "auto_skip_header": true,
+    "min_row_ratio": 0.85,
+    "timeout": 90,
+    "sleep": 0.2,
+    "max_retries": 1
+  },
+  "postgres": {
+    "table_name": "PACS_UPP_List_Records",
+    "env_prefix": "WORKFLOW_DB"
+  }
+}