├── README.md
├── address_sample_ 2.xlsx
├── input.xlsx
├── main.py
├── output.xlsx
└── start.bat


/README.md:
--------------------------------------------------------------------------------
1 | # Python-Address-Filter


--------------------------------------------------------------------------------
/address_sample_ 2.xlsx:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/AppleMac-DH-TECH/Python-Address-Filter/54bda9e6badca9bda05ce294c4c8d6cdd2a86fb8/address_sample_ 2.xlsx


--------------------------------------------------------------------------------
/input.xlsx:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/AppleMac-DH-TECH/Python-Address-Filter/54bda9e6badca9bda05ce294c4c8d6cdd2a86fb8/input.xlsx


--------------------------------------------------------------------------------
/main.py:
--------------------------------------------------------------------------------
 1 | import os
 2 | import pandas as pd
 3 | from pandas import DataFrame
 4 | 
 5 | cleaned = []
 6 | 
 7 | def train(src_xlsx_path: str):
 8 |     df = pd.read_excel(src_xlsx_path, sheet_name="Sheet1")
 9 |     
10 |     global cleaned
11 |     
12 |     for id in range(len(df)):
13 |         src_addr = df.at[id, 'EMP LOCATION']
14 |         dst_addr = df.at[id, 'CLEANED ADDRESS']
15 |         inx = 0
16 |         while src_addr[inx] == ' ':
17 |             inx += 1
18 |         src_addr = src_addr[inx:]
19 |         inx = 0
20 |         while dst_addr[inx] == ' ':
21 |             inx += 1
22 |         dst_addr = dst_addr[inx:]
23 |         # print(src_addr)
24 |         # print(dst_addr)
25 |         
26 |         diff_str = ''
27 |         j = 0
28 |         for i in src_addr:
29 |             if j >= len(dst_addr):
30 |                 break
31 |             if i == dst_addr[j]:
32 |                 if diff_str != '' and diff_str not in cleaned:
33 |                     cleaned.append(diff_str)
34 |                     print(diff_str)
35 |                     diff_str = ''
36 |                 j += 1
37 |             else:
38 |                 diff_str += i
39 |         
40 |                 
41 |     print(cleaned)
42 | 
43 | def filter(src_addr: str):
44 |     
45 |     global cleaned
46 |     
47 |     dst_addr = src_addr
48 |     
49 |     if not len(cleaned):
50 |         print("first training!")
51 |         return ''
52 |     
53 |     for item in cleaned:
54 |         dst_addr = dst_addr.replace(item, '')
55 |     
56 |     return dst_addr
57 |     
58 | 
59 | if __name__ == "__main__":
60 |     train_xlsx_path = os.getcwd() + "\\address_sample_ 2.xlsx"
61 |     src_addr_path = os.getcwd() + "\\input.xlsx"
62 |     train(train_xlsx_path)
63 |     
64 |     df = pd.read_excel(src_addr_path, sheet_name="Sheet1")
65 |     
66 |     dst_addr_list = []
67 |     
68 |     for id in range(len(df)):
69 |         src_addr = df.at[id, 'EMP LOCATION']
70 |         dst_addr = filter(src_addr)
71 |         if dst_addr != '':
72 |             dst_addr_list.append(dst_addr)
73 | 
74 |     df = DataFrame(dst_addr_list, columns=['CLEANED ADDRESS'])
75 |     df.to_excel("output.xlsx", sheet_name='Sheet1')


--------------------------------------------------------------------------------
/output.xlsx:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/AppleMac-DH-TECH/Python-Address-Filter/54bda9e6badca9bda05ce294c4c8d6cdd2a86fb8/output.xlsx


--------------------------------------------------------------------------------
/start.bat:
--------------------------------------------------------------------------------
1 | python main.py


--------------------------------------------------------------------------------