import re
import numpy as np
from fuzzywuzzy import fuzz
from shapely.geometry import Polygon
class Finder:
def __init__(self, pdf_info, ocr_results):
self.pdf_info = pdf_info
self.ocr_results = ocr_results
self.is_asp = False
self.item = {"words": None,
"position": None,
}
self.cn_re = re.compile(u'[\u4e00-\u9fa5]')
def gen_init_result(self, is_asp):
# 格式化算法输出
self.init_result = {"page_1": {"合同编号": self.item,
"所购车辆价格": self.item,
"车架号": self.item,
"贷款本金金额": {"大写": self.item,
"小写": self.item,
"车辆贷款本金金额": self.item,
"附加产品融资贷款本金总金额": self.item,
},
"贷款期限": self.item,
"附加产品融资贷款本金总金额明细": self.item,
"借款人签字及时间": self.item,
},
"page_2": {"合同编号": self.item,
"借款人及抵押人": {"name": self.item,
"id": self.item,
},
"共同借款人及共同抵押人": {"name": self.item,
"id": self.item,
},
"保证人1": {"name": self.item,
"id": self.item,
},
"保证人2": {"name": self.item,
"id": self.item,
},
"所购车辆价格": self.item,
"车架号": self.item,
"经销商": self.item,
"贷款本金金额": {"大写": self.item,
"小写": self.item,
"车辆贷款本金金额": self.item,
"附加产品融资贷款本金总金额": self.item,
},
"贷款期限": self.item,
"标准利率": self.item,
"借款人收款账户": {"账号": self.item,
"户名": self.item,
"开户行": self.item,
},
"还款账户": {"账号": self.item,
"户名": self.item,
"开户行": self.item,
},
},
"page_3": {"合同编号": self.item,
"还款计划表": self.item,
"车辆代理商": self.item,
},
"page_4": {"合同编号": self.item,
"附加产品融资贷款本金总金额明细": self.item,
},
"page_5": {"合同编号": self.item,
},
"page_6": {"合同编号": self.item,
},
}
if self.is_asp:
self.init_result["page_7"] = {"合同编号": self.item,
}
self.init_result["page_8"] = {"合同编号": self.item,
"主借人签字": {"签字": self.item,
"日期": self.item,
},
"共借人签字": {"签字": self.item,
"日期": self.item,
},
"保证人1签字": {"签字": self.item,
"日期": self.item,
},
"保证人2签字": {"签字": self.item,
"日期": self.item,
},
"见证人签字": {"签字": self.item,
"日期": self.item,
},
}
else:
self.init_result["page_7"] = {"合同编号": self.item,
"主借人签字": {"签字": self.item,
"日期": self.item,
},
"共借人签字": {"签字": self.item,
"日期": self.item,
},
"保证人1签字": {"签字": self.item,
"日期": self.item,
},
"保证人2签字": {"签字": self.item,
"日期": self.item,
},
"见证人签字": {"签字": self.item,
"日期": self.item,
},
}
def get_top_iou(self, poly, ocr_result):
"""传入一个多边形, 找到与之最匹配的多边形
Args:
poly (TYPE): Description
"""
iou_list = []
for key in ocr_result:
bbox, text = ocr_result[key]
g = Polygon(np.array(bbox).reshape((-1, 2)))
p = Polygon(np.array(poly).reshape((-1, 2)))
if not g.is_valid or not p.is_valid:
continue
inter = Polygon(g).intersection(Polygon(p)).area
union = g.area + p.area - inter
iou = inter/union
iou_list.append([iou, key])
if len(iou_list) == 0:
return -1, -1
top_iou = sorted(iou_list, key=lambda x: x[0])[-1]
return top_iou
def poly_to_rectangle(self, poly):
xmin, ymin, xmax, ymin, xmax, ymax, xmin, ymax = poly
bbox = [xmin, ymin, xmax, ymax]
return bbox
def get_contract_no(self, page_num):
"""传入页码,查看该页码右上角的编号
Args:
page_num (string):
Returns:
sting:
"""
contract_no = self.item.copy()
# contract_no['words'] = ''
# contract_no['position'] = [-1, -1, -1, -1]
# 只看第一页
for key in self.ocr_results[page_num]:
bbox, text = self.ocr_results[page_num][key]
if '合同编号:' in text:
words = text.split(':')[-1]
location = self.poly_to_rectangle(bbox)
contract_no['words'] = words
contract_no['position'] = location
return contract_no
def get_vehicle_price(self, page_num='0'):
vehicle_price = self.item.copy()
# vehicle_price['words'] = ''
# vehicle_price['position'] = [-1, -1, -1, -1]
for key in self.ocr_results[page_num]:
bbox, text = self.ocr_results[page_num][key]
if '所购车辆价格为人民币' in text:
words = text.split('币')[-1]
location = self.poly_to_rectangle(bbox)
vehicle_price['words'] = words
vehicle_price['position'] = location
return vehicle_price
def get_vin(self, page_num='0'):
vin = self.item.copy()
# vin['words'] = ''
# vin['position'] = [-1, -1, -1, -1]
for key in self.ocr_results[page_num]:
bbox, text = self.ocr_results[page_num][key]
if '车架号:' in text:
words = text.split(':')[-1]
location = self.poly_to_rectangle(bbox)
vin['words'] = words
vin['position'] = location
return vin
def cn_char_filter(self, src_str):
cn_chars = re.findall(self.cn_re, src_str)
cn_str = ''.join(cn_chars)
return cn_str
def get_loan_principal(self, page_num='0'):
chinese_keywords = ['壹', '贰', '叁', '肆', '伍', '陆', '柒', '捌', '玖', '拾',
'佰', '仟', '万', '亿', '元', '角', '分', '零', '整']
upper = self.item.copy()
lower = self.item.copy()
asp_1 = self.item.copy()
asp_2 = self.item.copy()
anchor_bbox = None
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if fuzz.ratio(''.join(chinese_keywords), self.cn_char_filter(text)) >= 10:
text = text.split('：')[-1].strip()
upper['position'] = bbox
upper['words'] = text
if '小写：¥' in text:
words = text.split('¥')[-1].strip()
lower['position'] = bbox
lower['words'] = words
if '附加产品融资贷款本金总金额' == text:
anchor_bbox = bbox
if anchor_bbox:
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if np.mean(bbox[1::2]) < np.mean(anchor_bbox[1::2]) and '人民币：小写：' in text:
words = re.findall(r'人民币：小写：\[(.*)\]', text)[0]
asp_1['position'] = bbox
asp_1['words'] = words
if np.mean(bbox[1::2]) > np.mean(anchor_bbox[1::2]) and '人民币：小写：' in text:
words = re.findall(r'人民币：小写：\[(.*)\]', text)[0]
asp_2['position'] = bbox
asp_2['words'] = words
return upper, lower, asp_1, asp_2
def get_loan_term(self, page_num='0'):
loan_term = self.item.copy()
all_text = ''
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
all_text += text
matchs = re.search(r'贷款期限(\d+)个月', all_text)
if matchs:
words = matchs.group(1)
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}个月' in text:
loan_term['position'] = bbox
loan_term['words'] = words
return loan_term
def get_standard_rate(self, page_num='0'):
standard_rate = self.item.copy()
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
matchs = re.search(r'本合同当期的标准利率为(\S+)%/年', text)
if matchs:
standard_rate['position'] = bbox
standard_rate['words'] = matchs.group(1)
return standard_rate
def mergelist(self, text_list):
pattern = re.compile("[^\u4e00-\u9fa5]") # 匹配不是中文的其他字符
mergeindex = -1
for index, i in enumerate(text_list):
if '所购' in i and len(pattern.sub('', pattern.sub('', text_list[index+1]))) != 0:
# if '所购' in i and '.00' not in text_list[index+1]:
mergeindex = index
if mergeindex == -1:
return text_list
else:
new_text_list = text_list[:mergeindex] + [text_list[mergeindex] + text_list[mergeindex+1]] + text_list[mergeindex+2:]
return self.mergelist(new_text_list)
def get_asp_details(self, page_num):
asp_details_table_term = self.item.copy()
asp_details_table = [['附加产品融资贷款本金总金额及贷款利率明细'], ['项目1', '用途总金额2', '贷款本金3']]
bbox_xm = None
bbox_ytzje = None
bbox_dkbj = None
bbox_total = None
for key in self.ocr_results[page_num]:
bbox, text = self.ocr_results[page_num][key]
if text == '项目1':
bbox_xm = bbox
if text == '用途总金额2':
bbox_ytzje = bbox
if text == '贷款本金3':
bbox_dkbj = bbox
if text in ['附加产品融资贷款本', '附加产品融资贷款本金', '附加产品融资贷']:
bbox_total = bbox
if bbox_xm:
for i in range(10):
rh = abs(bbox_xm[1]-bbox_xm[-1])
anchor = np.array(bbox_xm).reshape((-1 ,2))
anchor[:, 1] += int(rh*1.4)
_iou, _key = self.get_top_iou(poly=anchor, ocr_result=self.ocr_results[page_num])
if _iou > 0:
bbox, xm_text = self.ocr_results[page_num][_key]
bbox_xm = bbox
# 解决项目内容是两行的问题
if not '所购' in xm_text:
line = asp_details_table[-1]
line[0] += xm_text
asp_details_table[-1] = line
continue
# print(xm_text)
anchor_1 = [bbox_ytzje[0], bbox[1], bbox_ytzje[2], bbox[3],
bbox_ytzje[4], bbox[5], bbox_ytzje[6], bbox[7]]
_iou, _key = self.get_top_iou(poly=anchor_1, ocr_result=self.ocr_results[page_num])
bbox, ytzje_text = self.ocr_results[page_num][_key]
# print(ytzje_text)
anchor_2 = [bbox_dkbj[0], bbox[1], bbox_dkbj[2], bbox[3],
bbox_dkbj[4], bbox[5], bbox_dkbj[6], bbox[7]]
_iou, _key = self.get_top_iou(poly=anchor_2, ocr_result=self.ocr_results[page_num])
bbox, dkbj_text = self.ocr_results[page_num][_key]
# print(dkbj_text)
if xm_text == ytzje_text:
xm_text, ytzje_text = xm_text.split(' ')
line = [xm_text, ytzje_text, dkbj_text]
asp_details_table.append(line)
else:
break
if bbox_total:
anchor = [bbox_dkbj[0], bbox_total[1], bbox_dkbj[2], bbox_total[3],
bbox_dkbj[4], bbox_total[5], bbox_dkbj[6], bbox_total[7]]
_iou, _key = self.get_top_iou(poly=anchor, ocr_result=self.ocr_results[page_num])
bbox, total_text = self.ocr_results[page_num][_key]
asp_details_table.append(['附加产品融资贷款本金总金额:', '', total_text])
asp_details_table_term['words'] = asp_details_table
return asp_details_table_term
def get_signature(self):
signature = self.item.copy()
for block in self.pdf_info['0']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '签署日期' in text:
words = text
signature['words'] = words
signature['position'] = bbox
return signature
def get_somebody(self, top, bottom):
# 指定上下边界后,返回上下边界内的客户信息
_name = self.item.copy()
_id = self.item.copy()
# 只看第一页，先划定上下边界
y_top = 0
y_bottom = 0
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if top in text:
y_top = bbox[3]
if bottom in text:
y_bottom = bbox[3]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if y_top < bbox[3] < y_bottom:
# print(top, bottom, text)
if '姓名/名称' in text:
words = text.split('：')[-1]
_name['position'] = bbox
_name['words'] = words
if '自然人身份证件号码/法人执照号码' in text:
words = text.split('：')[-1]
_id['position'] = bbox
_id['words'] = words
return _name, _id
def get_seller(self):
seller = self.item.copy()
# 先找到 key
anchor_bbox = None
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text in ['经销商', '车辆销售方']:
anchor_bbox = bbox
# 当找到了 key, 则根据 key 去匹配 value
if anchor_bbox:
half_width = self.pdf_info['1']['width'] * 0.5
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if anchor_bbox[2]<np.mean(bbox[::2])<half_width and \
anchor_bbox[1]<np.mean(bbox[1::2])<anchor_bbox[3]:
seller['position'] = bbox
seller['words'] = text
return seller
def get_cldls(self):
seller = self.item.copy()
# 先找到 key
anchor_bbox = None
for block in self.pdf_info['2']['blocks']:
if anchor_bbox is not None:
break
if block['type'] != 0:
continue
for line in block['lines']:
if anchor_bbox is not None:
break
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text.strip() == '车辆代理商':
anchor_bbox = bbox
# print(anchor_bbox)
# 当找到了 key, 则根据 key 去匹配 value
if anchor_bbox:
half_width = self.pdf_info['2']['width'] * 0.5
for block in self.pdf_info['2']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if anchor_bbox[2]<np.mean(bbox[::2])<half_width and \
anchor_bbox[1]<np.mean(bbox[1::2])<anchor_bbox[3]:
seller['position'] = bbox
seller['words'] = text
return seller
return seller
def get_borrower_collection_account(self):
account = self.item.copy()
account_name = self.item.copy()
account_bank = self.item.copy()
all_text = ''
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
all_text += text
# 首先确定账户信息是哪种,我们只输出非另行通知的格式
if '借款人收款账户' in all_text:
all_text = all_text.replace(' ', '').replace(' ', '')
matchs_1 = re.findall(r'账号：(.*?)户名', all_text)
if matchs_1:
words = matchs_1[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}' in text:
account['position'] = bbox
account['words'] = words
matchs_2 = re.findall(r'户名：(.*?)开户行', all_text)
if matchs_2:
words = matchs_2[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}' in text:
account_name['position'] = bbox
account_name['words'] = words
matchs_3 = re.findall(r'开户行：(.*?)借款人', all_text)
if matchs_3:
words = matchs_3[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}' in text:
account_bank['position'] = bbox
account_bank['words'] = words
return account, account_name, account_bank
def get_payback_account(self):
account = self.item.copy()
account_name = self.item.copy()
account_bank = self.item.copy()
all_text = ''
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
all_text += text
# 首先确定账户信息是哪种,我们只输出非另行通知的格式
if '(13) 还款账户' in all_text:
all_text = all_text.split('(13) 还款账户')[-1]
all_text = all_text.replace(' ', '').replace(' ', '')
matchs_1 = re.findall(r'账号：(.*?)户名', all_text)
if matchs_1:
words = matchs_1[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}' in text:
account['position'] = bbox
account['words'] = words
matchs_2 = re.findall(r'户名：(.*?)开户行', all_text)
if matchs_2:
words = matchs_2[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'{words}' in text:
account_name['position'] = bbox
account_name['words'] = words
matchs_3 = re.findall(r'开户行：(.*?)；', all_text)
if matchs_3:
words = matchs_3[0]
for block in self.pdf_info['1']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if f'开户行：{words}；' in text.replace(' ', ''):
account_bank['position'] = bbox
account_bank['words'] = words
return account, account_name, account_bank
def get_repayment_schedule(self):
repayment_schedule = self.item.copy()
# 只看第二页
repayment_schedule_table = []
repayment_schedule_text_list = []
table = False
for block in self.pdf_info['2']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '序号' == text:
table = True
if '以上表格中所列的序号并非还款期数' in text:
table = False
if table == True:
repayment_schedule_text_list.append(text)
for i in range(len(repayment_schedule_text_list)//5):
line = []
# 5表示5列的意思
for j in range(5):
line.append(repayment_schedule_text_list[i*5+j])
if str(i+1) == line[1]:
break
repayment_schedule_table.append(line)
if len(repayment_schedule_table) > 0:
repayment_schedule['words'] = repayment_schedule_table
return repayment_schedule
def get_signature_role_1(self):
signature_role_1 = self.init_item.copy()
# 先定位签字区域
texts = []
boxes = []
page_num = None
position = None
words = None
region = False
for i in list(self.pdf_info.keys()):
for block in self.pdf_info[i]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '借款人(抵押人)' in text:
region = True
if '日期' in text:
region = False
if region == True:
page_num = i
texts.append(text)
boxes.append(bbox)
if len(texts) > 4:
words = '有'
else:
words = '无'
boxes = np.array(boxes).reshape((-1, 2))
position = [min(boxes[:,0]), min(boxes[:,1]), max(boxes[:,0]), max(boxes[:,1])]
signature_role_1['page_num'] = page_num
signature_role_1['position'] = position
signature_role_1['words'] = words
return signature_role_1
def get_signature_role_2(self):
signature_role_2 = self.init_item.copy()
# 先定位签字区域
texts = []
boxes = []
page_num = None
position = None
words = None
region = False
for i in list(self.pdf_info.keys()):
for block in self.pdf_info[i]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '共同借款人(共同抵押人)' in text:
region = True
if '日期' in text:
region = False
if region == True:
page_num = i
texts.append(text)
boxes.append(bbox)
if len(texts) > 4:
words = '有'
else:
words = '无'
boxes = np.array(boxes).reshape((-1, 2))
position = [min(boxes[:,0]), min(boxes[:,1]), max(boxes[:,0]), max(boxes[:,1])]
signature_role_2['page_num'] = page_num
signature_role_2['position'] = position
signature_role_2['words'] = words
return signature_role_2
def get_signature_role_3(self):
signature_role_3 = self.init_item.copy()
# 先定位签字区域
texts = []
boxes = []
page_num = None
position = None
words = None
region = False
for i in list(self.pdf_info.keys()):
for block in self.pdf_info[i]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '保证人1' in text and int(i) != 0:
region = True
if '日期' in text:
region = False
if region == True:
page_num = i
texts.append(text)
boxes.append(bbox)
if len(texts) > 4:
words = '有'
else:
words = '无'
boxes = np.array(boxes).reshape((-1, 2))
position = [min(boxes[:,0]), min(boxes[:,1]), max(boxes[:,0]), max(boxes[:,1])]
signature_role_3['page_num'] = page_num
signature_role_3['position'] = position
signature_role_3['words'] = words
return signature_role_3
def get_signature_role_4(self):
signature_role_4 = self.init_item.copy()
# 先定位签字区域
texts = []
boxes = []
page_num = None
position = None
words = None
region = False
for i in list(self.pdf_info.keys()):
for block in self.pdf_info[i]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '保证人2' in text and int(i) != 0:
region = True
if '日期' in text:
region = False
if region == True:
page_num = i
texts.append(text)
boxes.append(bbox)
if len(texts) > 4:
words = '有'
else:
words = '无'
boxes = np.array(boxes).reshape((-1, 2))
position = [min(boxes[:,0]), min(boxes[:,1]), max(boxes[:,0]), max(boxes[:,1])]
signature_role_4['page_num'] = page_num
signature_role_4['position'] = position
signature_role_4['words'] = words
return signature_role_4
def get_signature_role_5(self):
signature_role_5 = self.init_item.copy()
# 先定位签字区域
texts = []
boxes = []
page_num = None
position = None
words = None
region = False
for i in list(self.pdf_info.keys()):
for block in self.pdf_info[i]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '见证人签字' in text and int(i) != 0:
region = True
if '年' in text:
region = False
if region == True:
page_num = i
texts.append(text)
boxes.append(bbox)
print(texts)
if len(texts) > 4:
words = '有'
else:
words = '无'
boxes = np.array(boxes).reshape((-1, 2))
position = [min(boxes[:,0]), min(boxes[:,1]), max(boxes[:,0]), max(boxes[:,1])]
signature_role_5['page_num'] = page_num
signature_role_5['position'] = position
signature_role_5['words'] = words
return signature_role_5
def get_last_page_signature(self, page_num, top, bottom):
signature_name = self.item.copy()
signature_date = self.item.copy()
anchor_top = None
anchor_bottom = None
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if top in text:
anchor_top = bbox[1]
if bottom in text:
anchor_bottom = bbox[1]
# print(top, anchor_top, anchor_bottom)
if anchor_top is not None and anchor_bottom is not None:
for block in self.pdf_info[page_num]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '签署日期' in text and int(anchor_top)<np.mean(bbox[1::2])<int(anchor_bottom):
name = text.split(' ')[0]
date = text.split(':')[-1]
signature_name['words'] = name
signature_name['position'] = bbox
signature_date['words'] = date
signature_date['position'] = bbox
return signature_name, signature_date
def get_info(self):
"""
block['type'] == 0 : 表示该元素为图片
Returns:
dict: Description
"""
# 先判断是否为 ASP 产品
# 只看第一页，判断是否有 '附加产品融资贷款本金总金额' 这一句话，若有则为 ASP 产品
# print(self.pdf_info['0']['blocks'])
# for block in self.pdf_info['0']['blocks']:
# if block['type'] != 0:
# continue
# for line in block['lines']:
# for span in line['spans']:
# bbox, text = span['bbox'], span['text']
# if '附加产品融资贷款本金总金额' == text:
# self.is_asp = True
for key in self.ocr_results['0']:
bbox, text = self.ocr_results['0'][key]
if '附加产品融资贷款本金总金额' in text:
self.is_asp = True
self.gen_init_result(self.is_asp)
if len(list(self.ocr_results.keys())) <= 8: # 8.5 版本客户提供的样本出现串页的情况，暂时无法识别
# Page 1
# 找合同编号
contract_no = self.get_contract_no(page_num='0')
# print(contract_no)
self.init_result['page_1']['合同编号'] = contract_no
# 所购车辆价格
vehicle_price = self.get_vehicle_price()
# print(vehicle_price)
self.init_result['page_1']['所购车辆价格'] = vehicle_price
# 车架号
vin = self.get_vin()
# print(vin)
self.init_result['page_1']['车架号'] = vin
# 贷款本金金额(如果是 ASP产品)则'贷款本金金额'项目中包含'车辆贷款本金金额'和'附加产品融资贷款本金总金额'两个项目
upper, lower, asp_1, asp_2 = self.get_loan_principal()
# print(upper, lower, asp_1, asp_2)
self.init_result['page_1']['贷款本金金额']['大写'] = upper
self.init_result['page_1']['贷款本金金额']['小写'] = lower
self.init_result['page_1']['贷款本金金额']['车辆贷款本金金额'] = asp_1
self.init_result['page_1']['贷款本金金额']['附加产品融资贷款本金总金额'] = asp_2
# 贷款期限
loan_term = self.get_loan_term()
# print(loan_term)
self.init_result['page_1']['贷款期限'] = loan_term
# 附加产品融资贷款本金总金额明细（ASP-表格）
asp_details_table = self.get_asp_details(page_num='0')
# print(asp_details_table)
self.init_result['page_1']['附加产品融资贷款本金总金额明细'] = asp_details_table
# 借款人签字及时间
signature = self.get_signature()
# print(signature)
self.init_result['page_1']['借款人签字及时间'] = signature
#######################################
# Page 2
# 找合同编号
contract_no = self.get_contract_no(page_num='0')
# print(contract_no)
self.init_result['page_2']['合同编号'] = contract_no
# 找借款人及抵押人(地址字段原本有空格)
borrower_name, borrower_id = self.get_somebody(top='借款人及抵押人：', bottom='共同借款人：')
# 这是为了同时兼容 8.1 版本
if borrower_name['words'] == None:
borrower_name, borrower_id = self.get_somebody(top='借款人及抵押人：', bottom='共同借款人及共同抵押人：')
# 这是为了兼容车贷分离版本
if borrower_name['words'] == None:
borrower_name, borrower_id = self.get_somebody(top='借款人：', bottom='共同借款人及抵押人：')
# print(borrower_name, borrower_id)
self.init_result['page_2']['借款人及抵押人']['name'] = borrower_name
self.init_result['page_2']['借款人及抵押人']['id'] = borrower_id
# 找共同借款人及共同抵押人
co_borrower_name, co_borrower_id = self.get_somebody(top='共同借款人：', bottom='保证人1：')
# print(co_borrower_name, co_borrower_id)
self.init_result['page_2']['共同借款人及共同抵押人']['name'] = co_borrower_name
self.init_result['page_2']['共同借款人及共同抵押人']['id'] = co_borrower_id
# 保证人1
first_guarantor_name, first_guarantor_id = self.get_somebody(top='保证人1：', bottom='保证人2：')
self.init_result['page_2']['保证人1']['name'] = first_guarantor_name
self.init_result['page_2']['保证人1']['id'] = first_guarantor_id
# 保证人2
second_guarantor_name, second_guarantor_id = self.get_somebody(top='保证人2：', bottom='第一章')
self.init_result['page_2']['保证人2']['name'] = second_guarantor_name
self.init_result['page_2']['保证人2']['id'] = second_guarantor_id
# 所购车辆价格
vehicle_price = self.get_vehicle_price(page_num='1')
# print(vehicle_price)
self.init_result['page_2']['所购车辆价格'] = vehicle_price
# 车架号
vin = self.get_vin(page_num='1')
# print(vin)
self.init_result['page_2']['车架号'] = vin
# 经销商
seller = self.get_seller()
# print(seller)
self.init_result['page_2']['经销商'] = seller
# 贷款本金金额(如果是 ASP产品)则'贷款本金金额'项目中包含'车辆贷款本金金额'和'附加产品融资贷款本金总金额'两个项目
upper, lower, asp_1, asp_2 = self.get_loan_principal(page_num='1')
# print(upper, lower, asp_1, asp_2)
self.init_result['page_2']['贷款本金金额']['大写'] = upper
self.init_result['page_2']['贷款本金金额']['小写'] = lower
self.init_result['page_2']['贷款本金金额']['车辆贷款本金金额'] = asp_1
self.init_result['page_2']['贷款本金金额']['附加产品融资贷款本金总金额'] = asp_2
# 贷款期限
loan_term = self.get_loan_term(page_num='1')
# print(loan_term)
self.init_result['page_2']['贷款期限'] = loan_term
# 本合同当期的标准利率
standard_rate = self.get_standard_rate(page_num='1')
# print(standard_rate)
self.init_result['page_2']['标准利率'] = standard_rate
# 202212 release 新增借款人收款账户
account, account_name, account_bank = self.get_borrower_collection_account()
# print(account, account_name, account_bank)
self.init_result['page_2']['借款人收款账户']['账号'] = account
self.init_result['page_2']['借款人收款账户']['户名'] = account_name
self.init_result['page_2']['借款人收款账户']['开户行'] = account_bank
# 还款账户
account, account_name, account_bank = self.get_payback_account()
# print(account, account_name, account_bank)
self.init_result['page_2']['还款账户']['账号'] = account
self.init_result['page_2']['还款账户']['户名'] = account_name
self.init_result['page_2']['还款账户']['开户行'] = account_bank
#######################################
# Page 3
# 找合同编号
contract_no = self.get_contract_no(page_num='2')
self.init_result['page_3']['合同编号'] = contract_no
# 还款计划表（表格）
repayment_schedule_table = self.get_repayment_schedule()
# print(repayment_schedule_table)
self.init_result['page_3']['还款计划表'] = repayment_schedule_table
# 车辆代理商
cldls = self.get_cldls()
self.init_result['page_3']['车辆代理商'] = cldls
#######################################
# Page 4
# 找合同编号
contract_no = self.get_contract_no(page_num='3')
# print(contract_no)
self.init_result['page_4']['合同编号'] = contract_no
# 附加产品融资贷款本金总金额明细（ASP-表格）
asp_details_table = self.get_asp_details(page_num='3')
# print(asp_details_table)
self.init_result['page_4']['附加产品融资贷款本金总金额明细'] = asp_details_table
#######################################
# Page 5
# 找合同编号
contract_no = self.get_contract_no(page_num='4')
# print(contract_no)
self.init_result['page_5']['合同编号'] = contract_no
#######################################
# Page 6
# 找合同编号
contract_no = self.get_contract_no(page_num='5')
# print(contract_no)
self.init_result['page_6']['合同编号'] = contract_no
if self.is_asp:
# Page 7
# 找合同编号
contract_no = self.get_contract_no(page_num='6')
self.init_result['page_7']['合同编号'] = contract_no
# Page 8
# 找合同编号
contract_no = self.get_contract_no(page_num='7')
self.init_result['page_8']['合同编号'] = contract_no
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='合同编号', bottom='共同借款人')
if signature_name['words'] == None:
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='合同编号', bottom='共同借款人（抵押人）')
# print(signature_name, signature_date)
self.init_result['page_8']['主借人签字']['签字'] = signature_name
self.init_result['page_8']['主借人签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='共同借款人', bottom='保证人1')
if signature_name['words'] == None:
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='共同借款人（抵押人）', bottom='保证人1')
# print(signature_name, signature_date)
self.init_result['page_8']['共借人签字']['签字'] = signature_name
self.init_result['page_8']['共借人签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='保证人1', bottom='保证人2')
self.init_result['page_8']['保证人1签字']['签字'] = signature_name
self.init_result['page_8']['保证人1签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='保证人2', bottom='在本人面前亲笔签署本合同')
self.init_result['page_8']['保证人2签字']['签字'] = signature_name
self.init_result['page_8']['保证人2签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='7',
top='在本人面前亲笔签署本合同', bottom='以下无正文')
# print(signature_name, signature_date)
self.init_result['page_8']['见证人签字']['签字'] = signature_name
self.init_result['page_8']['见证人签字']['日期'] = signature_date
else:
# Page 7
# 找合同编号
contract_no = self.get_contract_no(page_num='6')
self.init_result['page_7']['合同编号'] = contract_no
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='合同编号', bottom='共同借款人')
if signature_name['words'] == None:
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='合同编号', bottom='共同借款人（抵押人）')
# print(signature_name, signature_date)
self.init_result['page_7']['主借人签字']['签字'] = signature_name
self.init_result['page_7']['主借人签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='共同借款人', bottom='保证人1')
if signature_name['words'] == None:
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='共同借款人（抵押人）', bottom='保证人1')
# print(signature_name, signature_date)
self.init_result['page_7']['共借人签字']['签字'] = signature_name
self.init_result['page_7']['共借人签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='保证人1', bottom='保证人2')
self.init_result['page_7']['保证人1签字']['签字'] = signature_name
self.init_result['page_7']['保证人1签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='保证人2', bottom='在本人面前亲笔签署本合同')
self.init_result['page_7']['保证人2签字']['签字'] = signature_name
self.init_result['page_7']['保证人2签字']['日期'] = signature_date
signature_name, signature_date = self.get_last_page_signature(page_num='6',
top='在本人面前亲笔签署本合同', bottom='以下无正文')
# print(signature_name, signature_date)
self.init_result['page_7']['见证人签字']['签字'] = signature_name
self.init_result['page_7']['见证人签字']['日期'] = signature_date
# 重新定制输出
new_results = {"is_asp": self.is_asp,
"page_info": self.init_result
}
return new_results