modify CA yhls sheet
Showing
2 changed files
with
28 additions
and
2 deletions
... | @@ -1145,6 +1145,7 @@ MS_ERROR_COL = (5, 6) | ... | @@ -1145,6 +1145,7 @@ MS_ERROR_COL = (5, 6) |
1145 | WECHART_CLASSIFY = 12 | 1145 | WECHART_CLASSIFY = 12 |
1146 | NEW_ZHIFUBAO_CLASSIFY = 48 | 1146 | NEW_ZHIFUBAO_CLASSIFY = 48 |
1147 | ALI_WECHART_CLASSIFY = {12, 13, 48} | 1147 | ALI_WECHART_CLASSIFY = {12, 13, 48} |
1148 | JSYH_CLASSIFY = {11, 27, 34} | ||
1148 | WECHART_ERROR_COL = (1, 2) | 1149 | WECHART_ERROR_COL = (1, 2) |
1149 | SPECIAL_HEADERS_MAPPING = copy.deepcopy(HEADERS_MAPPING) | 1150 | SPECIAL_HEADERS_MAPPING = copy.deepcopy(HEADERS_MAPPING) |
1150 | SPECIAL_HEADERS_MAPPING.update( | 1151 | SPECIAL_HEADERS_MAPPING.update( | ... | ... |
... | @@ -40,6 +40,18 @@ class BSWorkbook(Workbook): | ... | @@ -40,6 +40,18 @@ class BSWorkbook(Workbook): |
40 | self.need_follow = False | 40 | self.need_follow = False |
41 | 41 | ||
42 | @staticmethod | 42 | @staticmethod |
43 | def date_calibration(date_str): | ||
44 | result = True | ||
45 | try: | ||
46 | if date_str[-2] not in ['20', '21']: | ||
47 | result = False | ||
48 | if date_str[-5:-3] not in ['03', '06', '09', '12']: | ||
49 | result = False | ||
50 | except Exception as e: | ||
51 | result = False | ||
52 | return result | ||
53 | |||
54 | @staticmethod | ||
43 | def replace_newline(queryset_value): | 55 | def replace_newline(queryset_value): |
44 | new_set = set() | 56 | new_set = set() |
45 | for v in queryset_value: | 57 | for v in queryset_value: |
... | @@ -338,7 +350,7 @@ class BSWorkbook(Workbook): | ... | @@ -338,7 +350,7 @@ class BSWorkbook(Workbook): |
338 | metadata_rows.append(('Producer', producer)) | 350 | metadata_rows.append(('Producer', producer)) |
339 | if len(author) > 0: | 351 | if len(author) > 0: |
340 | metadata_highlight_row.append(6) | 352 | metadata_highlight_row.append(6) |
341 | if 'iText' not in producer and 'Qt' not in producer and 'Haru Free' not in producer: | 353 | if 'iText' not in producer and 'Qt' not in producer and 'Haru Free' not in producer and 'OpenPDF' not in producer: |
342 | metadata_highlight_row.append(7) | 354 | metadata_highlight_row.append(7) |
343 | metadata_rows.append(self.blank_row) | 355 | metadata_rows.append(self.blank_row) |
344 | 356 | ||
... | @@ -496,6 +508,7 @@ class BSWorkbook(Workbook): | ... | @@ -496,6 +508,7 @@ class BSWorkbook(Workbook): |
496 | 508 | ||
497 | tmp_ws = self.create_sheet('tmp_ws') | 509 | tmp_ws = self.create_sheet('tmp_ws') |
498 | tmp2_ws = self.create_sheet('tmp2_ws') | 510 | tmp2_ws = self.create_sheet('tmp2_ws') |
511 | tmp3_ws = self.create_sheet('tmp3_ws') | ||
499 | if classify in consts.ALI_WECHART_CLASSIFY: | 512 | if classify in consts.ALI_WECHART_CLASSIFY: |
500 | high_light_keyword = self.wechat_keyword | 513 | high_light_keyword = self.wechat_keyword |
501 | else: | 514 | else: |
... | @@ -629,7 +642,7 @@ class BSWorkbook(Workbook): | ... | @@ -629,7 +642,7 @@ class BSWorkbook(Workbook): |
629 | # 关键词1提取 | 642 | # 关键词1提取 |
630 | if summary_cell_value in self.interest_keyword: | 643 | if summary_cell_value in self.interest_keyword: |
631 | new_amount_cell_value = None if amount_cell is None else amount_cell.value | 644 | new_amount_cell_value = None if amount_cell is None else amount_cell.value |
632 | ms.append((summary_cell_value, date_cell_value, new_amount_cell_value)) | 645 | tmp3_ws.append((summary_cell_value, date_cell_value, new_amount_cell_value)) |
633 | # 关键词2提取至临时表 | 646 | # 关键词2提取至临时表 |
634 | elif summary_cell_value in self.salary_keyword: | 647 | elif summary_cell_value in self.salary_keyword: |
635 | new_amount_cell_value = None if amount_cell is None else amount_cell.value | 648 | new_amount_cell_value = None if amount_cell is None else amount_cell.value |
... | @@ -657,6 +670,18 @@ class BSWorkbook(Workbook): | ... | @@ -657,6 +670,18 @@ class BSWorkbook(Workbook): |
657 | # if summary_cell_idx is not None: | 670 | # if summary_cell_idx is not None: |
658 | # new_ws[row][summary_cell_idx].fill = self.amount_fill | 671 | # new_ws[row][summary_cell_idx].fill = self.amount_fill |
659 | 672 | ||
673 | # 关键词1信息提取:结息 | ||
674 | for row in tmp3_ws.iter_rows(values_only=True): | ||
675 | ms.append(row) | ||
676 | # 建设银行 | ||
677 | if classify in consts.JSYH_CLASSIFY: | ||
678 | if isinstance(row[1], str) and self.date_calibration(row[1]): | ||
679 | pass | ||
680 | else: | ||
681 | for cell in ms[ms.max_row]: | ||
682 | cell.fill = self.amount_fill | ||
683 | self.remove(tmp3_ws) | ||
684 | |||
660 | # 关键词2信息提取 | 685 | # 关键词2信息提取 |
661 | ms.append(self.blank_row) | 686 | ms.append(self.blank_row) |
662 | ms.append(self.salary_keyword_header) | 687 | ms.append(self.salary_keyword_header) | ... | ... |
-
Please register or sign in to post a comment