diff --git a/src/apps/doc/ocr/wb.py b/src/apps/doc/ocr/wb.py index c7f1b1d..6fc8e8b 100644 --- a/src/apps/doc/ocr/wb.py +++ b/src/apps/doc/ocr/wb.py @@ -79,18 +79,24 @@ class BSWorkbook(Workbook): # } # 第一行关键词 - find_count = 0 + header_col_list = [] for first_row in ws.iter_rows(max_row=1, min_row=1, values_only=True): sheet_header_info.setdefault(ws.title, {}).setdefault(consts.HEADER_KEY, first_row) for idx, header_value in enumerate(first_row): header_col = self.get_header_col(header_value, classify) if header_col is not None: - find_count += 1 - sheet_header_info.setdefault(ws.title, {}).setdefault(header_col, idx) - find_col_set = sheet_header_info.setdefault(ws.title, {}).setdefault(consts.FIND_COL_KEY, set()) - find_col_set.add(idx) - col_count = header_info.setdefault(header_col, {}).get(idx) - header_info.setdefault(header_col, {})[idx] = 1 if col_count is None else col_count+1 + header_col_list.append((idx, header_col)) + + find_count = len(header_col_list) + if find_count < 2: + find_count = 0 + else: + for idx, header_col in header_col_list: + sheet_header_info.setdefault(ws.title, {}).setdefault(header_col, idx) + find_col_set = sheet_header_info.setdefault(ws.title, {}).setdefault(consts.FIND_COL_KEY, set()) + find_col_set.add(idx) + col_count = header_info.setdefault(header_col, {}).get(idx) + header_info.setdefault(header_col, {})[idx] = 1 if col_count is None else col_count+1 sheet_header_info.setdefault(ws.title, {}).setdefault(consts.FIND_COUNT_KEY, find_count) min_row = 1 if find_count == 0 else 2