feat：增加分析进度接口

app/datafiles/tables.py

@@ -8,7 +8,14 @@ import re
 from dataclasses import dataclass, field
 from typing import Any
 
-from app.datafiles.base import MAX_TABLE_ROWS, brief, rows_exceeded, where
+from app.datafiles.base import (
+    MAX_TABLE_ROWS,
+    Reporter,
+    brief,
+    no_report,
+    rows_exceeded,
+    where,
+)
 from app.datafiles.columns import as_text, classify
 from app.datafiles.readers import Grid
 from app.schemas.data_file import ParseWarning
@@ -60,13 +67,21 @@ class _Piece:
     warnings: list[ParseWarning]
 
 
-def split_tables(grid: Grid) -> tuple[list[RawTable], list[ParseWarning]]:
-    """返回识别出的表和不属于任何表的警告（被丢弃的块）"""
+def split_tables(
+    grid: Grid, report: Reporter = no_report
+) -> tuple[list[RawTable], list[ParseWarning]]:
+    """返回识别出的表和不属于任何表的警告（被丢弃的块）
+
+    几步都要扫一遍所有行，每步结束报告一次进度，比例来自大表上的实测耗时
+    """
     rows = _normalize(grid.rows)
+    report(0.25)
     sheet = grid.sheet
     pieces: list[_Piece] = []
     dropped: list[ParseWarning] = []
-    for block in _blocks(rows):
+    blocks = _blocks(rows)
+    report(0.3)
+    for number, block in enumerate(blocks, start=1):
         columns = [
             c for c in range(len(rows[0])) if any(rows[r][c] is not None for r in block)
         ]
@@ -88,10 +103,12 @@ def split_tables(grid: Grid) -> tuple[list[RawTable], list[ParseWarning]]:
             )
             continue
         pieces.append(_read_block(rows, block, columns, sheet))
+        report(0.3 + 0.25 * number / len(blocks))
 
     tables = []
-    for piece in pieces:
+    for number, piece in enumerate(pieces, start=1):
         table = _finish(rows, piece, sheet)
+        report(0.55 + 0.45 * number / len(pieces))
         if table is None:
             dropped.append(
                 ParseWarning(
@@ -101,6 +118,7 @@ def split_tables(grid: Grid) -> tuple[list[RawTable], list[ParseWarning]]:
             )
         else:
             tables.append(table)
+    report(1.0)
     if len(tables) > 1:
         for part, table in enumerate(tables, start=1):
             table.part = part