feat：LLM 分析文件并生成结构化建议

app/api/files.py

@@ -1,20 +1,29 @@
 """数据文件接口：上传限制、上传解析、列出、查看、删除，见 docs/screen-agent/data-files.md 第 5 节"""
 
-from typing import Annotated, get_args
+from typing import Annotated, Literal, get_args
 
 from fastapi import APIRouter, Depends, File, Form, Query, Request, UploadFile
 
 from app.common.response import ApiResponse, success_response
 from app.core.auth import get_current_user
-from app.datafiles.service import MAX_FILES_PER_CONVERSATION, DataFileService
+from app.datafiles.service import (
+    MAX_FILES_PER_CONVERSATION,
+    PARSE_CONCURRENCY,
+    DataFileService,
+)
 from app.schemas.base import NonBlankString
 from app.schemas.data_file import (
+    MAX_AGGREGATE_ROWS,
+    MAX_METRICS,
+    AggregateRequest,
+    AggregateResult,
     DataFileDetail,
     DataFileInfo,
     DataFileLimits,
     DataFileList,
     DataFileRecord,
     DataFormat,
+    DataTableRows,
 )
 from app.schemas.user import CurrentUser
 
@@ -23,6 +32,10 @@ limits_router = APIRouter(prefix="/files", tags=["DataFile"])
 router = APIRouter(prefix="/files", tags=["DataFile"])
 
 # 文件选择框的过滤条件；后端按文件内容识别格式，不看扩展名
+# 翻页取表数据：单页上限，和一张表最多的行数（100 万）
+MAX_PAGE_ROWS = 500
+MAX_OFFSET = 1_000_000
+
 ACCEPT = ".csv,.tsv,.txt,.xlsx,.xls,.xlsb,.json,.jsonl,.ndjson,.parquet"
 
 
@@ -43,6 +56,10 @@ async def get_limits(request: Request) -> ApiResponse[DataFileLimits]:
             max_files_per_conversation=MAX_FILES_PER_CONVERSATION,
             formats=list(get_args(DataFormat)),
             accept=ACCEPT,
+            parse_concurrency=PARSE_CONCURRENCY,
+            rows_max_page=MAX_PAGE_ROWS,
+            aggregate_max_metrics=MAX_METRICS,
+            aggregate_max_rows=MAX_AGGREGATE_ROWS,
         )
     )
 
@@ -85,6 +102,49 @@ async def get_file(
     return success_response(_detail(await _service(request).get_file(user.id, file_id)))
 
 
+@router.post("/{file_id}/insights")
+async def regenerate_insights(
+    file_id: str,
+    request: Request,
+    user: Annotated[CurrentUser, Depends(get_current_user)],
+) -> ApiResponse[DataFileDetail]:
+    """重新分析文件，上次分析失败后用；分析在后台进行，返回的 insights 为 pending"""
+    return success_response(
+        _detail(await _service(request).regenerate_insights(user.id, file_id))
+    )
+
+
+@router.get("/{file_id}/tables/{table}/rows")
+async def get_table_rows(
+    file_id: str,
+    table: str,
+    request: Request,
+    user: Annotated[CurrentUser, Depends(get_current_user)],
+    offset: Annotated[int, Query(ge=0, le=MAX_OFFSET)] = 0,
+    limit: Annotated[int, Query(ge=1, le=MAX_PAGE_ROWS)] = 100,
+    sort_by: Annotated[str | None, Query(min_length=1)] = None,
+    order: Annotated[Literal["asc", "desc"], Query()] = "asc",
+) -> ApiResponse[DataTableRows]:
+    return success_response(
+        await _service(request).get_rows(
+            user.id, file_id, table, offset, limit, sort_by, order == "desc"
+        )
+    )
+
+
+@router.post("/{file_id}/tables/{table}/aggregate")
+async def aggregate_table(
+    file_id: str,
+    table: str,
+    body: AggregateRequest,
+    request: Request,
+    user: Annotated[CurrentUser, Depends(get_current_user)],
+) -> ApiResponse[AggregateResult]:
+    return success_response(
+        await _service(request).aggregate(user.id, file_id, table, body)
+    )
+
+
 @router.delete("/{file_id}")
 async def delete_file(
     file_id: str,