feat：完成大部分数据文件上传解析功能

app/processor/nodes/screen_agent_node.py

@@ -7,6 +7,7 @@ import json
 
 from langchain_core.messages import HumanMessage
 
+from app.agent.data_tools import DataContext
 from app.agent.screen_agent import AgentReply, get_screen_agent
 from app.agent.tools import ScreenAgentContext
 from app.processor.base import BaseNode
@@ -53,7 +54,7 @@ class ScreenAgentNode(BaseNode):
             return updated_state
 
         context = state["input_context"]
-        agent = await get_screen_agent()
+        agent = await get_screen_agent(with_data=bool(context.data_files))
         result = await agent.ainvoke(
             {
                 # 会话历史作为此前的对话消息，本次请求是最后一条用户消息
@@ -64,7 +65,11 @@ class ScreenAgentNode(BaseNode):
                     ),
                 ]
             },
-            context=ScreenAgentContext(screen=context.screen, catalog=catalog),
+            context=ScreenAgentContext(
+                screen=context.screen,
+                catalog=catalog,
+                data=DataContext(tuple(context.data_files)),
+            ),
         )
         if result.get("structured_response") is None:
             last = result["messages"][-1]

app/processor/summary.py

@@ -4,11 +4,15 @@
 """
 
 from app.schemas.base import JsonObject
+from app.schemas.data_file import RequestDataFile
 from app.schemas.input_context import InputContext
 
+# 每张表在摘要里最多列出的列名，完整的列用 describe_data 查看
+SUMMARY_COLUMNS = 30
+
 
 def request_summary(context: InputContext) -> JsonObject:
-    """用户输入、澄清记录、编辑上下文、附件和大屏摘要"""
+    """用户输入、澄清记录、编辑上下文、附件、数据文件和大屏摘要"""
     nodes: list[JsonObject] = []
     for node in context.screen.nodes:
         # 图表标题帮助判断用户指的是哪个组件；没有 option 或标题格式不符时按无标题处理
@@ -45,8 +49,32 @@ def request_summary(context: InputContext) -> JsonObject:
                 for source in context.screen.dataSource
             ],
         },
+        # 数据文件的附件由 data_files 的 attached_this_turn 表示
         "attachments": [
             {key: item[key] for key in ("id", "kind", "name") if key in item}
             for item in context.attachments
+            if item.get("kind") != "data_file"
         ],
+        "data_files": [_data_file(item) for item in context.data_files],
+    }
+
+
+def _data_file(item: RequestDataFile) -> JsonObject:
+    """每个文件只给表名、label、行数和列名；列概况由 describe_data 按需查看"""
+    tables: list[JsonObject] = []
+    for table in item.tables:
+        summary: JsonObject = {
+            "table": table.table,
+            "label": table.label,
+            "rows": table.rows,
+            "columns": [column.name for column in table.columns[:SUMMARY_COLUMNS]],
+        }
+        if len(table.columns) > SUMMARY_COLUMNS:
+            summary["column_count"] = len(table.columns)
+        tables.append(summary)
+    return {
+        "file_id": item.file_id,
+        "name": item.name,
+        "attached_this_turn": item.attached_this_turn,
+        "tables": tables,
     }