Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
b9e0884e by 冯轩
Options

Merge branch 'feature/fsm-contract' into feature/uat-tmp

2 parents 3474c20c 4c67581e
Expand all
Inline Side-by-side
Showing 3 changed files with 97 additions and 1 deletions
src/common/electronic_afc_contract/get_char_fsm.py
...@@ -62,6 +62,7 @@ class Finder: ...@@ -62,6 +62,7 @@ class Finder:
62 }, 62 },
63 "page_3": {"合同编号": self.item, 63 "page_3": {"合同编号": self.item,
64 "还款计划表": self.item, 64 "还款计划表": self.item,
65 "车辆代理商": self.item,
65 }, 66 },
66 "page_4": {"合同编号": self.item, 67 "page_4": {"合同编号": self.item,
67 "附加产品融资贷款本金总金额明细": self.item, 68 "附加产品融资贷款本金总金额明细": self.item,
...@@ -71,6 +72,7 @@ class Finder: ...@@ -71,6 +72,7 @@ class Finder:
71 "page_6": {"合同编号": self.item, 72 "page_6": {"合同编号": self.item,
72 }, 73 },
73 } 74 }
75 if self.is_asp:
74 self.init_result["page_7"] = {"合同编号": self.item, 76 self.init_result["page_7"] = {"合同编号": self.item,
75 } 77 }
76 self.init_result["page_8"] = {"合同编号": self.item, 78 self.init_result["page_8"] = {"合同编号": self.item,
...@@ -90,6 +92,25 @@ class Finder: ...@@ -90,6 +92,25 @@ class Finder:
90 "日期": self.item, 92 "日期": self.item,
91 }, 93 },
92 } 94 }
95 else:
96 self.init_result["page_7"] = {"合同编号": self.item,
97 "主借人签字": {"签字": self.item,
98 "日期": self.item,
99 },
100 "共借人签字": {"签字": self.item,
101 "日期": self.item,
102 },
103 "保证人1签字": {"签字": self.item,
104 "日期": self.item,
105 },
106 "保证人2签字": {"签字": self.item,
107 "日期": self.item,
108 },
109 "见证人签字": {"签字": self.item,
110 "日期": self.item,
111 },
112 }
113
93 114
94 def get_top_iou(self, poly, ocr_result): 115 def get_top_iou(self, poly, ocr_result):
95 """传入一个多边形, 找到与之最匹配的多边形 116 """传入一个多边形, 找到与之最匹配的多边形
...@@ -398,6 +419,39 @@ class Finder: ...@@ -398,6 +419,39 @@ class Finder:
398 seller['words'] = text 419 seller['words'] = text
399 return seller 420 return seller
400 421
422 def get_cldls(self):
423 seller = self.item.copy()
424 # 先找到 key
425 anchor_bbox = None
426 for block in self.pdf_info['2']['blocks']:
427 if anchor_bbox is not None:
428 break
429 if block['type'] != 0:
430 continue
431 for line in block['lines']:
432 if anchor_bbox is not None:
433 break
434 for span in line['spans']:
435 bbox, text = span['bbox'], span['text']
436 if text.strip() == '车辆代理商':
437 anchor_bbox = bbox
438 # print(anchor_bbox)
439 # 当找到了 key, 则根据 key 去匹配 value
440 if anchor_bbox:
441 half_width = self.pdf_info['2']['width'] * 0.5
442 for block in self.pdf_info['2']['blocks']:
443 if block['type'] != 0:
444 continue
445 for line in block['lines']:
446 for span in line['spans']:
447 bbox, text = span['bbox'], span['text']
448 if anchor_bbox[2]<np.mean(bbox[::2])<half_width and \
449 anchor_bbox[1]<np.mean(bbox[1::2])<anchor_bbox[3]:
450 seller['position'] = bbox
451 seller['words'] = text
452 return seller
453 return seller
454
401 def get_borrower_collection_account(self): 455 def get_borrower_collection_account(self):
402 account = self.item.copy() 456 account = self.item.copy()
403 account_name = self.item.copy() 457 account_name = self.item.copy()
...@@ -885,6 +939,9 @@ class Finder: ...@@ -885,6 +939,9 @@ class Finder:
885 repayment_schedule_table = self.get_repayment_schedule() 939 repayment_schedule_table = self.get_repayment_schedule()
886 # print(repayment_schedule_table) 940 # print(repayment_schedule_table)
887 self.init_result['page_3']['还款计划表'] = repayment_schedule_table 941 self.init_result['page_3']['还款计划表'] = repayment_schedule_table
942 # 车辆代理商
943 cldls = self.get_cldls()
944 self.init_result['page_3']['车辆代理商'] = cldls
888 ####################################### 945 #######################################
889 # Page 4 946 # Page 4
890 # 找合同编号 947 # 找合同编号
...@@ -907,6 +964,8 @@ class Finder: ...@@ -907,6 +964,8 @@ class Finder:
907 contract_no = self.get_contract_no(page_num='5') 964 contract_no = self.get_contract_no(page_num='5')
908 # print(contract_no) 965 # print(contract_no)
909 self.init_result['page_6']['合同编号'] = contract_no 966 self.init_result['page_6']['合同编号'] = contract_no
967
968 if self.is_asp:
910 # Page 7 969 # Page 7
911 # 找合同编号 970 # 找合同编号
912 contract_no = self.get_contract_no(page_num='6') 971 contract_no = self.get_contract_no(page_num='6')
...@@ -944,6 +1003,41 @@ class Finder: ...@@ -944,6 +1003,41 @@ class Finder:
944 # print(signature_name, signature_date) 1003 # print(signature_name, signature_date)
945 self.init_result['page_8']['见证人签字']['签字'] = signature_name 1004 self.init_result['page_8']['见证人签字']['签字'] = signature_name
946 self.init_result['page_8']['见证人签字']['日期'] = signature_date 1005 self.init_result['page_8']['见证人签字']['日期'] = signature_date
1006 else:
1007 # Page 7
1008 # 找合同编号
1009 contract_no = self.get_contract_no(page_num='6')
1010 self.init_result['page_7']['合同编号'] = contract_no
1011 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1012 top='合同编号', bottom='共同借款人')
1013 if signature_name['words'] == None:
1014 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1015 top='合同编号', bottom='共同借款人（抵押人）')
1016 # print(signature_name, signature_date)
1017 self.init_result['page_7']['主借人签字']['签字'] = signature_name
1018 self.init_result['page_7']['主借人签字']['日期'] = signature_date
1019 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1020 top='共同借款人', bottom='保证人1')
1021 if signature_name['words'] == None:
1022 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1023 top='共同借款人（抵押人）', bottom='保证人1')
1024 # print(signature_name, signature_date)
1025 self.init_result['page_7']['共借人签字']['签字'] = signature_name
1026 self.init_result['page_7']['共借人签字']['日期'] = signature_date
1027 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1028 top='保证人1', bottom='保证人2')
1029 self.init_result['page_7']['保证人1签字']['签字'] = signature_name
1030 self.init_result['page_7']['保证人1签字']['日期'] = signature_date
1031 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1032 top='保证人2', bottom='在本人面前亲笔签署本合同')
1033 self.init_result['page_7']['保证人2签字']['签字'] = signature_name
1034 self.init_result['page_7']['保证人2签字']['日期'] = signature_date
1035 signature_name, signature_date = self.get_last_page_signature(page_num='6',
1036 top='在本人面前亲笔签署本合同', bottom='以下无正文')
1037 # print(signature_name, signature_date)
1038 self.init_result['page_7']['见证人签字']['签字'] = signature_name
1039 self.init_result['page_7']['见证人签字']['日期'] = signature_date
1040
947 1041
948 # 重新定制输出 1042 # 重新定制输出
949 new_results = {"is_asp": self.is_asp, 1043 new_results = {"is_asp": self.is_asp,
......
src/common/fsm_econtract/retriever.py
...@@ -6,6 +6,7 @@ class HMHRetriever: ...@@ -6,6 +6,7 @@ class HMHRetriever:
6 def __init__(self): 6 def __init__(self):
7 self.words_str = 'words' 7 self.words_str = 'words'
8 self.position_str = 'location' 8 self.position_str = 'location'
9 self.fix_hava_str = '有'
9 self.default_position = [0, 0, 0, 0] 10 self.default_position = [0, 0, 0, 0]
10 self.search_fields_list = [ 11 self.search_fields_list = [
11 ('借款/承租人姓名', ''), 12 ('借款/承租人姓名', ''),
...@@ -51,7 +52,8 @@ class HMHRetriever: ...@@ -51,7 +52,8 @@ class HMHRetriever:
51 for name_date_tuple in name_date_list: 52 for name_date_tuple in name_date_list:
52 if len(name_date_tuple) == 2: 53 if len(name_date_tuple) == 2:
53 result[self.search_fields_list[4][0]] = { 54 result[self.search_fields_list[4][0]] = {
54 self.words_str: '{0} {1}'.format(name_date_tuple[0].replace('\u3000', '').strip(), name_date_tuple[1]), 55 # self.words_str: '{0} {1}'.format(name_date_tuple[0].replace('\u3000', '').strip(), name_date_tuple[1]),
56 self.words_str: self.fix_hava_str,
55 self.position_str: bbox 57 self.position_str: bbox
56 } 58 }
57 is_find_name_date = True 59 is_find_name_date = True
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!