Merge branch 'feature/contract' into feature/0611

周伟奇
Showing 3 changed files with 65 additions and 6 deletions
src/apps/doc/consts.py
src/apps/doc/management/commands/ocr_process.py
src/apps/doc/ocr/wb.py
--- a/src/apps/doc/consts.py
View file @5fbb652
+++ b/src/apps/doc/consts.py
View file @5fbb652
@@ -1002,6 +1002,10 @@ JYPZ_FIELD_ORDER = (("type", "标题"),
                    ("stamp_signature_seller", "出卖方签字/盖章"),
                    ("stamp_signature_agent", "经销商签字/盖章"),)
+# 合同
+CONTRACT_CN_NAME = '合同'
+CONTRACT_CLASSIFY = 41
 SUCCESS_CODE_SET = {'0', 0}
 FIELD_ORDER_MAP = {
@@ -1628,4 +1632,4 @@ CA_COMPARE_FIELD = {
    DL_EN: (DL_OCR_FIELD, CA_DL_COMPARE_LOGIC, False),
 }
+CONTRACT_SHEET_NAME = '合同'
--- a/src/apps/doc/management/commands/ocr_process.py
View file @5fbb652
+++ b/src/apps/doc/management/commands/ocr_process.py
View file @5fbb652
@@ -207,6 +207,45 @@ class Command(BaseCommand, LoggerMixin):
        else:
            res_list.append((pno, ino, part_idx, consts.RES_SUCCESS_EMPTY))
+    def contract_process(self, ocr_data, contract_result, res_list, pno, ino, part_idx, img_path):
+        contract_dict = ocr_data.get('data')
+        if not contract_dict or 'page_num' not in contract_dict or 'page_info' not in contract_dict:
+            res_list.append((pno, ino, part_idx, consts.RES_SUCCESS_EMPTY))
+            return
+        res_list.append((pno, ino, part_idx, consts.RES_SUCCESS))
+        page_num = contract_dict.get('page_num')
+        rebuild_page_info = []
+        text_key = 'words'
+        for key, value in contract_dict.get('page_info', {}).items():
+            if value is None:
+                rebuild_page_info.append((key, ))
+            elif text_key in value:
+                if value[text_key] is None:
+                    rebuild_page_info.append((key,))
+                elif isinstance(value[text_key], str):
+                    rebuild_page_info.append((key, value[text_key]))
+                elif isinstance(value[text_key], list):
+                    rebuild_page_info.append((key,))
+                    for row_list in value[text_key]:
+                        rebuild_page_info.append(row_list)
+            else:
+                rebuild_page_info.append((key,))
+                for sub_key, sub_value in value.items():
+                    if sub_value is None:
+                        rebuild_page_info.append((sub_key,))
+                    elif text_key in sub_value:
+                        if sub_value[text_key] is None:
+                            rebuild_page_info.append((sub_key,))
+                        elif isinstance(sub_value[text_key], str):
+                            rebuild_page_info.append((sub_key, sub_value[text_key]))
+                        elif isinstance(sub_value[text_key], list):
+                            rebuild_page_info.append((sub_key,))
+                            for row_list in sub_value[text_key]:
+                                rebuild_page_info.append(row_list)
+        contract_result.setdefault(page_num, []).append(rebuild_page_info)
    def license1_process(self, ocr_data, license_summary, classify, res_list, pno, ino, part_idx, img_path, do_dda, dda_id_bc_mapping):
        # 类别：'0'身份证， '1'居住证
        license_data = ocr_data.get('data')
@@ -820,6 +859,7 @@ class Command(BaseCommand, LoggerMixin):
                        bs_summary = {}
                        unknown_summary = {}
                        license_summary = {}
+                        contract_result = {}
                        res_list = []
                        interest_keyword = Keywords.objects.filter(
                            type=KeywordsType.INTEREST.value, on_off=True).values_list('keyword', flat=True)
@@ -907,6 +947,9 @@ class Command(BaseCommand, LoggerMixin):
                                                res_list.append((pno, ino, part_idx, consts.RES_FAILED_2))
                                                self.online_log.warn(
                                                    '{0} [ocr_2 failed] [img_path={1}]'.format(self.log_base, img_path))
+                                        elif classify == consts.CONTRACT_CLASSIFY:
+                                            self.contract_process(ocr_data, contract_result, res_list, pno, ino,
+                                                                  part_idx, img_path)
                                        else:  # 流水处理
                                            bs_classify_set.add(classify)
                                            self.bs_process(wb, ocr_data, bs_summary, unknown_summary, classify, res_list, pno, ino, part_idx)
@@ -936,9 +979,9 @@ class Command(BaseCommand, LoggerMixin):
                        # self.bs_log.info('[task={0}] [bs_summary={1}]'.format(task_str, merged_bs_summary))
-                        self.online_log.info('{0} [task={1}] [merged_bs_summary={2}] [license_summary={3}] '
+                        self.online_log.info('{0} [task={1}] [merged_bs_summary={2}] [license_summary={3}] [contract={4}] '
-                                             '[res_list={4}]'.format(self.log_base, task_str, merged_bs_summary,
+                                             '[res_list={5}]'.format(self.log_base, task_str, merged_bs_summary,
-                                                                     license_summary, res_list))
+                                                                     contract_result, license_summary, res_list))
                    except Exception as e:
@@ -959,7 +1002,7 @@ class Command(BaseCommand, LoggerMixin):
                            # 重构Excel文件
                            # src_excel_path = os.path.join(doc_data_path, 'src.xlsx')
                            # wb.save(src_excel_path)
-                            count_list = wb.rebuild(merged_bs_summary, license_summary, res_list, doc.document_scheme)
+                            count_list = wb.rebuild(merged_bs_summary, license_summary, res_list, doc.document_scheme, contract_result)
                            wb.save(excel_path)
                        except Exception as e:
--- a/src/apps/doc/ocr/wb.py
View file @5fbb652
+++ b/src/apps/doc/ocr/wb.py
View file @5fbb652
@@ -702,6 +702,17 @@ class BSWorkbook(Workbook):
            if field_str is not None:
                count_list.append((field_str, count))
+    def contract_rebuild(self, contract_result):
+        if len(contract_result) == 0:
+            return
+        ws = self.create_sheet(consts.CONTRACT_SHEET_NAME)
+        for page_num, info_list in contract_result:
+            ws.append('page {0}'.format(page_num))
+            for info in info_list:
+                for row in info:
+                    ws.append(row)
+            ws.append((None, ))
    @staticmethod
    def remove_yuan(amount_key_set, key, src_str):
        if key in amount_key_set and isinstance(src_str, str):
@@ -801,12 +812,13 @@ class BSWorkbook(Workbook):
        if len(self.sheetnames) > 1:
            self.remove(self.get_sheet_by_name('Sheet'))
-    def rebuild(self, bs_summary, license_summary, res_list, document_scheme):
+    def rebuild(self, bs_summary, license_summary, res_list, document_scheme, contract_result):
        res_count_tuple = self.res_sheet(res_list)
        count_list = [(consts.MODEL_FIELD_BS, len(bs_summary))]
        if document_scheme == consts.DOC_SCHEME_LIST[1]:
            self.license_rebuild(license_summary, document_scheme, count_list)
+            self.contract_rebuild(contract_result)
            self.bs_rebuild(bs_summary, res_count_tuple)
        else:
            self.bs_rebuild(bs_summary, res_count_tuple)