# -*- coding: utf-8 -*-
# @Author : lk
# @Email : 9428.al@gmail.com
# @Create Date : 2021-07-20 16:42:41
# @Last Modified : 2021-10-28 17:41:00
# @Description :
import re
import numpy as np
from fuzzywuzzy import fuzz
class Finder:
def __init__(self, pdf_info):
self.pdf_info = pdf_info
self.item = {"words": None,
"page": None,
"position": None,
}
# 格式化算法输出
self.init_result = {"合同编号": self.item,
"承租人-姓名": self.item,
"承租人-证件号码": self.item,
"承租人-法定代表人或授权代表": self.item,
"共同承租人-姓名": self.item,
"共同承租人-证件号码": self.item,
"共同承租人-法定代表人或授权代表": self.item,
"保证人1-姓名": self.item,
"保证人1-证件号码": self.item,
"保证人1-法定代表人或授权代表": self.item,
"保证人2-姓名": self.item,
"保证人2-证件号码": self.item,
"保证人2-法定代表人或授权代表": self.item,
"保证人3-姓名": self.item,
"保证人3-证件号码": self.item,
"保证人3-法定代表人或授权代表": self.item,
"合同编号（正文）": self.item,
"车辆识别代码": self.item,
"车辆卖方（经销商）": self.item,
"车辆原始销售价格（《机动车销售统一发票》所列金额）": self.item,
"车辆附加产品明细表": self.item,
"融资成本总额": self.item,
"租期": self.item,
"付款计划表": self.item,
"银行账户-户名": self.item,
"银行账户-银行账号": self.item,
"银行账户-开户行": self.item,
"签字页-承租人姓名": self.item,
"签字页-承租人签章": self.item,
"签字页-共同承租人姓名": self.item,
"签字页-共同承租人签章": self.item,
"签字页-保证人1姓名": self.item,
"签字页-保证人1签章": self.item,
"签字页-保证人2姓名": self.item,
"签字页-保证人2签章": self.item,
"签字页-保证人3姓名": self.item,
"签字页-保证人3签章": self.item,
}
# 格式化输出 车辆处置协议 要是别的字段
self.init_result_1 = {"合同编号": self.item,
"承租人-姓名": self.item,
"承租人-证件号码": self.item,
"销售经销商": self.item,
"合同编号（正文）": self.item,
"签字页-承租人姓名": self.item,
"签字页-承租人证件号码": self.item,
"签字页-承租人签章": self.item,
"签字页-销售经销商": self.item,
"签字页-销售经销商签章": self.item,
}
# 格式化输出 车辆租赁抵押合同
self.init_result_2 = {"合同编号": self.item,
"合同编号（正文）": self.item,
"抵押人姓名/名称": self.item,
"抵押人证件号码": self.item,
"抵押人配偶姓名/名称": self.item,
"抵押人配偶证件号码": self.item,
"车辆识别代码": self.item,
"租金总额": self.item,
"融资租赁期限": self.item,
"签字页-抵押人姓名": self.item,
"签字页-抵押人签章": self.item,
"签字页-抵押人配偶姓名": self.item,
"签字页-抵押人配偶签章": self.item,
}
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()
# 只看第一页
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 '合同编号：' in text:
words = text.split('：')[-1]
contract_no['position'] = bbox
contract_no['page'] = page_num
contract_no['words'] = words
if contract_no['words'] == '':
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 bbox[1] < contract_no['position'][3] and 'CH' in text:
contract_no['position'] = bbox
contract_no['page'] = page_num
contract_no['words'] = text
return contract_no
def get_vehicle_price(self, page_num='0'):
vehicle_price = 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']
if '所购车辆价格为人民币' in text:
words = text.split('币')[-1]
vehicle_price['position'] = bbox
vehicle_price['words'] = words
return vehicle_price
def get_contract_no_one(self):
# 查找正文中的合同编号,有可能存在换行的情况
contract_no = self.item.copy()
for pno in self.pdf_info:
all_text = ''
for block in self.pdf_info[pno]['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
all_text = all_text.replace(' ', '')
matchObj = re.search(r'（合同编号：\[(.*?)\]）', all_text)
if matchObj:
words = matchObj.group(1)
contract_no['position'] = None
contract_no['page'] = pno
# contract_no['words'] = words
contract_no['words'] = re.sub("\s", "", words).replace("）", "")
return contract_no
matchObj = re.search(r'编号为(.*?)的', all_text)
if matchObj:
words = matchObj.group(1).strip()
contract_no['position'] = None
contract_no['page'] = pno
# contract_no['words'] = words
contract_no['words'] = re.sub("\s", "", words).replace("）", "")
return contract_no
matchObj = re.search(r'编号为(.*?)）的', all_text)
if matchObj:
words = matchObj.group(1).strip()
contract_no['position'] = None
contract_no['page'] = pno
# contract_no['words'] = words
contract_no['words'] = re.sub("\s", "", words)
return contract_no
def get_key_value(self, key, page_num=None):
value = self.item.copy()
if page_num is not None:
pno = page_num
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key in text:
words = text.split('：')[-1].replace("。", "")
value['position'] = bbox
value['page'] = pno
# value['words'] = words
value['words'] = re.sub("\s", "", words)
else:
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key in text:
# print(self.pdf_info[pno])
words = text.split('：')[-1].replace("。", "")
value['position'] = bbox
value['page'] = pno
# value['words'] = words
value['words'] = re.sub("\s", "", words)
return value
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), text) > 15:
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_asp_details(self, page_num):
asp_details_table_term = self.item.copy()
asp_details_table = []
asp_details_text_list = []
table = False
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 '附加产品融资贷款本金总金额明细' == text:
table = True
if '第二条' in text or '征信管理' in text:
table = False
if table == True:
asp_details_text_list.append(text)
for i in range((len(asp_details_text_list) + 2) // 3):
line = []
if i == 0:
line = [asp_details_text_list[0]]
else:
for j in range(3):
line.append(asp_details_text_list[i * 3 - 2 + j])
asp_details_table.append(line)
if len(asp_details_table) > 0:
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:
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 '经销商' == text:
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_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 '☑账号' in all_text:
all_text = all_text.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()
repayment_schedule_text_list = []
table = False
page = None
left = 0
right = 0
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '剩余融资' in text:
right = bbox[2]
if '以上表格中所列序号' in text:
table = False
if table == True:
# 过滤汉字
if re.compile(r'[\u4e00-\u9fff]').search(text):
continue
# 过滤 1. - 61. 这些标题
if re.findall("\d+", text):
if len(re.findall("\d+", text)) == 1:
continue
if not left < bbox[0] < right:
continue
repayment_schedule_text_list.append(text)
if text.strip() == "61.":
page = pno
table = True
left = bbox[0]
# print("repayment_schedule_text_list = ", repayment_schedule_text_list)
# repayment_schedule_table = [['序号', '融资租赁成本', '融资租赁费用', '租金', '剩余融资租赁成本']]
repayment_schedule_table = [['序号', '租金']]
for i in range(len(repayment_schedule_text_list) // 4):
line = [f'{i + 1}.']
# 4表示4列的意思
for j in range(4):
line.append(repayment_schedule_text_list[i * 4 + j])
# 只保留序号和租金列
line = [line[0].replace('.', ''), line[3]]
repayment_schedule_table.append(line)
repayment_schedule['words'] = repayment_schedule_table
repayment_schedule['page'] = page
return repayment_schedule
def get_signature_role_1(self):
signature_role_1 = self.item.copy()
for pno in self.pdf_info:
for block in self.pdf_info[pno]['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:
signature_role_1['position'] = bbox
signature_role_1['page'] = pno
signature_role_1['words'] = text
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]
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_name['position'] = bbox
return signature_name, signature_date
def get_electronic_signature(self, top, bottom):
signature = self.item.copy()
anchor_top = None
anchor_bottom = None
for pno in self.pdf_info:
for block in self.pdf_info[pno]['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[3]
if anchor_top is not None and anchor_bottom is not None:
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
# ------------ #
# print("--text = ", text)
if '签署日期' in text and int(anchor_top) < np.mean(bbox[1::2]) < int(anchor_bottom):
words = text
signature['words'] = words
signature['page'] = pno
signature['position'] = bbox
return signature
def get_role_info(self, role_key, page_num='0'):
name = self.item.copy()
id_num = self.item.copy()
representative = self.item.copy()
# 以保证人3 的左上角为定位点
anchor = 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']
# 找到角色姓名
if re.match('保证人3', text) is not None:
anchor = [bbox[0], bbox[1]]
if anchor 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']
# 找到角色姓名
if re.match(role_key, text) is not None:
words = text.split('：')[-1]
name['words'] = words
name['page'] = page_num
name['position'] = bbox
if role_key == '承租人：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '保证人1：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '保证人2：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '保证人3：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
return name, id_num, representative
def get_table_add_product(self):
table_add_product = self.item.copy()
items = []
start = False
page = None
greater_equal_v35 = False
for pno in self.pdf_info:
condition = False
for block in self.pdf_info[f'{pno}']['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text == '租赁利率':
greater_equal_v35 = True
if '总计' in text:
start = True
if '注：出租人向承租人购买租赁车辆的对价' in text:
page = pno
start = False
if start == True:
items.append(text)
lines = [['项目', '购买价格', '实际融资金额']]
if greater_equal_v35:
for i in range(len(items) // 4):
line = [items[2 + i * 4 + 0], items[2 + i * 4 + 1], items[2 + i * 4 + 2]]
lines.append(line)
else:
for i in range(len(items) // 3):
line = [items[2 + i * 3 + 0], items[2 + i * 3 + 1], items[2 + i * 3 + 2]]
lines.append(line)
if len(items) > 0:
lines.append([items[0], '', items[1]])
table_add_product['words'] = lines
table_add_product['page'] = page
table_add_product['position'] = None
return table_add_product
def get_contract_no_dy(self):
# 查找抵押合同编号
contract_no = self.item.copy()
key_box = None
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if '抵押合同编号' in text:
key_box = bbox
if key_box is not None:
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] and 'CH-' in text:
contract_no['position'] = bbox
contract_no['page'] = pno
contract_no['words'] = text
return contract_no
def get_dyr_name_id(self):
name = self.item.copy()
_id = self.item.copy()
key_box = None
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text == '抵押人':
key_box = bbox
if key_box is not None:
rh = abs(key_box[1] - key_box[3])
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] + rh * 3 and '姓名' in text:
words = text.split('：')[-1]
name['position'] = bbox
name['page'] = pno
name['words'] = words
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] + rh * 3 and '证件号码' in text:
words = text.split('：')[-1]
_id['position'] = bbox
_id['page'] = pno
_id['words'] = words
return name, _id
def get_dyrpo_name_id(self):
name = self.item.copy()
_id = self.item.copy()
key_box = None
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text == '抵押人配偶(如适':
key_box = bbox
if key_box is not None:
rh = abs(key_box[1] - key_box[3])
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] + rh * 3 and '姓名' in text:
words = text.split('：')[-1]
name['position'] = bbox
name['page'] = pno
name['words'] = words
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] + rh * 3 and '证件号码' in text:
words = text.split('：')[-1]
_id['position'] = bbox
_id['page'] = pno
_id['words'] = words
return name, _id
def get_key_value_position(self, key):
value = self.item.copy()
key_box = None
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if text == key:
key_box = bbox
if key_box is not None:
rh = abs(key_box[1] - key_box[3])
for pno in self.pdf_info:
for block in self.pdf_info[pno]['blocks']:
if block['type'] != 0:
continue
for line in block['lines']:
for span in line['spans']:
bbox, text = span['bbox'], span['text']
if key_box[1] < np.mean(bbox[1::2]) < key_box[3] and key_box[0] < bbox[0] and abs(
key_box[2] - bbox[0]) < rh * 10:
words = text
value['position'] = bbox
value['page'] = pno
value['words'] = words
return value
def get_role_info_3_3(self, role_key, page_num='0'):
name = self.item.copy()
id_num = self.item.copy()
representative = self.item.copy()
# 以保证人2 的左上角为定位点
anchor = 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']
# 找到角色姓名
if re.match('保证人2', text) is not None:
anchor = [bbox[0], bbox[1]]
if anchor 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']
# 找到角色姓名
if re.match(role_key, text) is not None:
words = text.split('：')[-1]
name['words'] = words
name['page'] = page_num
name['position'] = bbox
if role_key == '承租人一：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '共同承租人：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) < anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '保证人1：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) < anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
if role_key == '保证人2：':
# 找到证件号码且确定位置
if re.match('证件号码：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
id_num['words'] = words
id_num['page'] = page_num
id_num['position'] = bbox
# 找到法人代表且确定位置
if re.match('法定代表人或授权代表：', text) is not None and np.mean(bbox[::2]) > anchor[0] and np.mean(
bbox[1::2]) > anchor[1]:
words = text.split('：')[-1]
representative['words'] = words
representative['page'] = page_num
representative['position'] = bbox
return name, id_num, representative
def get_info(self):
"""
block['type'] == 0 : 表示该元素为图片
Returns:
dict: Description
"""
if len(self.pdf_info) > 0:
# 取 Page 1 上的合同编号
contract_no = self.get_contract_no(page_num='0')
self.init_result['合同编号'] = contract_no
# 从第一页上取四个角色的姓名和证件号码
name, id_num, representative = self.get_role_info(role_key='承租人：', page_num='0')
if name["words"] == None:
name, id_num, representative = self.get_role_info_3_3(role_key='承租人一：', page_num='0')
self.init_result['承租人-姓名'] = name
self.init_result['承租人-证件号码'] = id_num
self.init_result['承租人-法定代表人或授权代表'] = representative
name, id_num, representative = self.get_role_info(role_key='保证人1：', page_num='0')
self.init_result['保证人1-姓名'] = name
self.init_result['保证人1-证件号码'] = id_num
self.init_result['保证人1-法定代表人或授权代表'] = representative
# if条件判别 对应3_3版本
if name["words"] == None:
name, id_num, representative = self.get_role_info_3_3(role_key='共同承租人：', page_num='0')
self.init_result['共同承租人-姓名'] = name
self.init_result['共同承租人-证件号码'] = id_num
self.init_result['共同承租人-法定代表人或授权代表'] = representative
name, id_num, representative = self.get_role_info(role_key='保证人2：', page_num='0')
self.init_result['保证人2-姓名'] = name
self.init_result['保证人2-证件号码'] = id_num
self.init_result['保证人2-法定代表人或授权代表'] = representative
# if条件判别 对应3_3版本
if name["words"] == None:
name, id_num, representative = self.get_role_info_3_3(role_key='保证人1：', page_num='0')
self.init_result['保证人2-姓名'] = name
self.init_result['保证人2-证件号码'] = id_num
self.init_result['保证人2-法定代表人或授权代表'] = representative
name, id_num, representative = self.get_role_info(role_key='保证人3：', page_num='0')
self.init_result['保证人3-姓名'] = name
self.init_result['保证人3-证件号码'] = id_num
self.init_result['保证人3-法定代表人或授权代表'] = representative
if name["words"] == None:
name, id_num, representative = self.get_role_info_3_3(role_key='保证人2：', page_num='0')
self.init_result['保证人3-姓名'] = name
self.init_result['保证人3-证件号码'] = id_num
self.init_result['保证人3-法定代表人或授权代表'] = representative
# 在所有页面中找正文中（第二部分 融资租赁主要条款及付款计划）的那个编号，因为存在换行的情况所以暂时不带位置输出
contract_no = self.get_contract_no_one()
self.init_result['合同编号（正文）'] = contract_no
# 找到车辆识别代码
vin = self.get_key_value(key='车辆识别代码：')
self.init_result['车辆识别代码'] = vin
# 找到经销商(车辆卖方(经销商))
seller = self.get_key_value(key='车辆卖方（经销商）：')
self.init_result['车辆卖方（经销商）'] = seller
# 找到 —— 车辆原始销售价格
vehicle_price = self.get_key_value(key='车辆原始销售价格（《机动车销售统一发票》所列金额）：')
self.init_result['车辆原始销售价格（《机动车销售统一发票》所列金额）'] = vehicle_price
# 找车辆附加产品明细（表）
table_add_product = self.get_table_add_product()
self.init_result['车辆附加产品明细表'] = table_add_product
# 找融资成本总额
financing_cost = self.get_key_value(key='融资成本总额：')
self.init_result['融资成本总额'] = financing_cost
# 找租期
lease_term = self.get_key_value(key='租期：')
self.init_result['租期'] = lease_term
# 找还款计划(表)
repayment_schedule = self.get_repayment_schedule()
self.init_result['付款计划表'] = repayment_schedule
# 找开户行户名、银行账号、银行
name = self.get_key_value(key='户名：')
self.init_result['银行账户-户名'] = name
account = self.get_key_value(key='银行账号：')
self.init_result['银行账户-银行账号'] = account
bank = self.get_key_value(key='开户银行：')
self.init_result['银行账户-开户行'] = bank
# 找签字页上的系列信息
# 承租人姓名、签章
name = self.get_key_value(key='承租人姓名：')
electronic_signature = self.get_electronic_signature(top='承租人姓名：', bottom='保证人1姓名：')
if name["words"] == None:
name = self.get_key_value(key='承租人一姓名：')
electronic_signature = self.get_electronic_signature(top='承租人一姓名：', bottom='共同承租人名称：')
self.init_result['签字页-承租人姓名'] = name
self.init_result['签字页-承租人签章'] = electronic_signature
# 保证人1姓名、签章
name = self.get_key_value(key='保证人1姓名：')
electronic_signature = self.get_electronic_signature(top='保证人1姓名：', bottom='保证人2姓名：')
self.init_result['签字页-保证人1姓名'] = name
self.init_result['签字页-保证人1签章'] = electronic_signature
# 这里用的是 name["words"] == ""
if name["words"] == "":
name = self.get_key_value(key='共同承租人名称：')
electronic_signature = self.get_electronic_signature(top='共同承租人名称：', bottom='保证人1姓名：')
self.init_result['签字页-共同承租人姓名'] = name
self.init_result['签字页-共同承租人签章'] = electronic_signature
# 保证人2姓名、签章
name = self.get_key_value(key='保证人2姓名：')
electronic_signature = self.get_electronic_signature(top='保证人2姓名：', bottom='保证人3姓名：')
self.init_result['签字页-保证人2姓名'] = name
self.init_result['签字页-保证人2签章'] = electronic_signature
# if判断条件对应3_3版本
if name["words"] == "":
name = self.get_key_value(key='保证人1姓名：')
electronic_signature = self.get_electronic_signature(top='保证人1姓名：', bottom='保证人2姓名：')
self.init_result['签字页-保证人1姓名'] = name
self.init_result['签字页-保证人1签章'] = electronic_signature
# 保证人3姓名、签章
name = self.get_key_value(key='保证人3姓名：')
electronic_signature = self.get_electronic_signature(top='保证人3姓名：', bottom='日期：')
self.init_result['签字页-保证人3姓名'] = name
self.init_result['签字页-保证人3签章'] = electronic_signature
# if判断条件对应3_3版本
if name["words"] == None:
name = self.get_key_value(key='保证人2姓名：')
electronic_signature = self.get_electronic_signature(top='保证人2姓名：', bottom='日期：')
self.init_result['签字页-保证人2姓名'] = name
self.init_result['签字页-保证人2签章'] = electronic_signature
return self.init_result
# results['is_shhz_contract'] = True
# results['pdf_info'] = self.init_result
# return results
def get_info_1(self):
if len(self.pdf_info) > 0:
contract_no = self.get_contract_no(page_num='0')
self.init_result_1['合同编号'] = contract_no
# 承租人姓名
name = self.get_key_value(key='承租人：', page_num='0')
self.init_result_1['承租人-姓名'] = name
# 承租人证件号码
_id = self.get_key_value(key='证件号码：', page_num='0')
self.init_result_1['承租人-证件号码'] = _id
# 销售经销商
seller = self.get_key_value(key='销售经销商：', page_num='0')
self.init_result_1['销售经销商'] = seller
# 合同编号（正文）
contract_no = self.get_contract_no_one()
self.init_result_1['合同编号（正文）'] = contract_no
# 签字页-承租人姓名
name = self.get_key_value(key='姓名/名称：')
self.init_result_1['签字页-承租人姓名'] = name
# 签字页-承租人证件号码
_id = self.get_key_value(key='自然人身份证件号码/法人执照号码：')
self.init_result_1['签字页-承租人证件号码'] = _id
# 签字页-承租人签章
signature_role_1 = self.get_signature_role_1()
self.init_result_1['签字页-承租人签章'] = signature_role_1
# 签字页-销售经销商
seller = self.get_key_value(key='销售经销商：')
self.init_result_1['签字页-销售经销商'] = seller
# 经销商签章
pass
return self.init_result_1
def get_info_2(self):
if len(self.pdf_info) > 0:
contract_no = self.get_contract_no_dy()
self.init_result_2['合同编号'] = contract_no
# 合同编号（正文）
contract_no = self.get_contract_no_one()
self.init_result_2['合同编号（正文）'] = contract_no
# 抵押人姓名/名称
name, _id = self.get_dyr_name_id()
self.init_result_2['抵押人姓名/名称'] = name
self.init_result_2['抵押人证件号码'] = _id
# 抵押人配偶信息
name, _id = self.get_dyrpo_name_id()
self.init_result_2['抵押人配偶姓名/名称'] = name
self.init_result_2['抵押人配偶证件号码'] = _id
# 车辆识别代码
vin = self.get_key_value(key='车辆识别代码：')
self.init_result_2['车辆识别代码'] = vin
# 租金总额
rent = self.get_key_value_position(key='租金总额')
self.init_result_2['租金总额'] = rent
# 融资租赁期限
lease_term = self.get_key_value_position(key='融资租赁期限')
self.init_result_2['融资租赁期限'] = lease_term
# 签字页抵押人姓名和签章
name = self.get_key_value(key='抵押人姓名：')
electronic_signature = self.get_electronic_signature(top='抵押权人盖章', bottom='抵押人配偶姓名：')
self.init_result_2['签字页-抵押人姓名'] = name
self.init_result_2['签字页-抵押人签章'] = electronic_signature
# 签字页抵押人配偶姓名和签章
name = self.get_key_value(key='抵押人配偶姓名：')
electronic_signature = self.get_electronic_signature(top='抵押人配偶姓名：', bottom='日期')
self.init_result_2['签字页-抵押人配偶姓名'] = name
self.init_result_2['签字页-抵押人配偶签章'] = electronic_signature
return self.init_result_2