File name
Commit message
Commit date
File name
Commit message
Commit date
import pandas as pd
import glob
def stripper(text):
text = text.strip().strip('"')
text = text.strip().strip('(')
text = text.strip().strip(')')
return text
files = glob.glob('국민연금/column좀 통일해라/*.csv')
for file in files:
df = pd.read_csv(file, sep=',', on_bad_lines='skip', low_memory=False)
# df.to_csv(file, encoding="utf-8")
col = df.columns
n_col = [None] * len(col)
output = ''
for i, row in enumerate(col):
out = list(map(stripper, row.split(' ')[0]))
out = ''.join([str(item) for item in out])
n_col[i] = out
df.columns = n_col
# this seems totally unreasonable,
# but this is necessary. trust me, they put newline char in their column names
print(file)
dfyc = df.query("(법정동주소광역시도코드 == 47) & (법정동주소광역시시군구코드 == 230)") # 경상북도 영천시
dfyc.to_csv(f"output/column좀 통일해라/{file.split('/')[-1].split('.')[0]}_영천.csv",)