将多个 CSV 文件中的列数据合并到一个 CSV 文件中

Question

我是 Python 新手，尤其是数据处理。 这就是我正在努力实现的目标-

我在多台服务器上运行 CIS 测试并为每个服务器生成一个 CSV 文件（文件名与服务器名相同）。 将所有服务器的输出文件复制到中央服务器产生的输出如下所示（截断的输出）-

File1: dc1pp1v01.co.uk.csv
Description,Outcome,Result
1.1 Database Placement,/var/lib/mysql,PASSED
1.2 Use dedicated least privilaged account,mysql,PASSED
1.3 Diable MySQL history,Not Found,PASSED

File2: dc1pp2v01.co.uk.csv
Description,Outcome,Result
1.1 Database Placement,/var/lib/mysql,PASSED
1.2 Use dedicated least privilaged account,mysql,PASSED
1.3 Diable MySQL history,Not Found,PASSED

File..n: dc1pp1v02.co.uk.csv
Description,Outcome,Result
1.1 Database Placement,/var/lib/mysql,PASSED
1.2 Use dedicated least privilaged account,mysql,PASSED
1.3 Diable MySQL history,Found,FAILED

我想要的是输出应该看起来像-

Description  dc1pp1v01 dc1pp2v01 dc1pp1v02 
0  1.1 Database Placement PASSED   PASSED   PASSED
1  1.2 Use dedicated least privilaged account PASSED   PASSED   PASSED
2  1.3 Diable MySQL history PASSED   PASSED   FAILED

为了合并这些文件，我创建了另一个文件，其中只有描述字段和两列标题，如下所示 -

file: cis_report.csv
Description,Result
1.1 Database Placement,
1.2 Use dedicated least privilaged account,
1.3 Diable MySQL history,

我写了下面的代码来进行基于列的合并-

import glob
import os
import pandas as pd 

col_list = ["Description","Result"]
path = "/Users/Python/Data"
all_files = glob.glob(os.path.join(path, "dc*.csv"))

cis_df = pd.read_csv("/Users/Python/Data/cis_report.csv")

for fl in all_files:
   d = pd.read_csv(fl, usecols=col_list)
   f = cis_df.merge(d, on='Description')
   cis_df = f.copy()
   
print(cis_df.head())

我得到的输出是-

Description Result_x Result_y Result_x Result_y
0                      1.1 Database Placement      NaN   PASSED   PASSED   PASSED
1  1.2 Use dedicated least privilaged account      NaN   PASSED   PASSED   PASSED
2                    1.3 Diable MySQL history      NaN   PASSED   PASSED   FAILED

在我的代码中，我不确定如何将文件名作为结果的标题并摆脱 NaN。

另外，是否有更好的方法来实现我正在寻找的输出而不使用虚拟文件（cis_report.csv）？ 非常感谢您的帮助。

Answer 1

您需要DataFrme.pivot()函数。 下面的代码有很好的注释和一个完整的工作示例。 根据需要进行更改

import os
import pandas as pd

# Get all file names in a directory
# Use . to use current working directory or replace it with
# e.g. r'C:\Users\Dames\Desktop\csv_files'
file_names = os.listdir('.')

# Filter out all non .csv files
# You can skip this if you know that only .csv files will be in that folder
csv_file_names = [fn for fn in file_names if fn[-4:] == '.csv']

# This Loads a csv file into a dataframe and sets the Server column
def load_csv(file_name):
    df = pd.read_csv(file_name)
    df['Server'] = file_name.split('.')[0]
    return df

# Append all the csvfiles after being processed by load_csv
df = pd.DataFrame().append([load_csv(fn) for fn in csv_file_names])

# Turn DataFrame into Pivot Table
df = df.pivot('Description', 'Server', 'Result')

# Save DataFrame into CSV File
# If this script runs multiple times make sure that the final.csv is saved elsewhere!
# Or it will be read by the code above as an input file
df.to_csv('final.csv')

最终的 DataFrame 看起来像这样

Server                                     dc1pp1v01 dc1pp1v02 dc1pp2v01
Description
1.1 Database Placement                        PASSED    PASSED    PASSED
1.2 Use dedicated least privilaged account    PASSED    PASSED    PASSED
1.3 Diable MySQL history                      PASSED    FAILED    PASSED

和这样的 CSV 文件

Description,dc1pp1v01,dc1pp1v02,dc1pp2v01
1.1 Database Placement,PASSED,PASSED,PASSED
1.2 Use dedicated least privilaged account,PASSED,PASSED,PASSED
1.3 Diable MySQL history,PASSED,FAILED,PASSED

Answer 2

用 -

import glob
import os
import pandas as pd 

col_list = ["Description","Result"]
path = "/Users/Python/Data"
all_files = glob.glob(os.path.join(path, "dc*.csv"))

cis_df = pd.read_csv("/Users/Python/Data/cis_report.csv")
from functools import reduce
df_final = reduce(lambda left,right: pd.merge(left,right,on='Description'), [cis_df]+[pd.read_csv(i, usecols=col_list) for i in all_files])
df_final.drop([i for i in df_final.columns if 'Outcome' in i], axis=1).rename(columns={i:j for i,j in zip([i for i in df_final.columns if 'Result' in i], [i.replace('.co.uk.csv','') for i in all_files])})

输出

    Description dc1pp1v01   dc1pp2v01   dc1pp1v02
0   1.1 Database Placement  PASSED  PASSED  PASSED
1   1.2 Use dedicated least privilaged account  PASSED  PASSED  PASSED
2   1.3 Diable MySQL history    PASSED  PASSED  FAILED

Answer 3

最后，我设法自己做到了。 下面的解决方案对我有用，但我相信有更简洁的方法 -

import glob
import os
import pandas as pd 
from functools import reduce

col_list = ["Description","Result"]
path = "/Users/Python/Data"
all_files = glob.glob(os.path.join(path, "dc*.csv"))

final_cols = ['Description']
for j in all_files:
    final_cols.append(os.path.basename(j).split('.',1)[0]) 

cis_df = pd.read_csv("/Users/Python/Data/cis_report.csv")

df_final = reduce(lambda left,right: pd.merge(left,right,on='Description'), [cis_df]+[pd.read_csv(i, usecols=col_list) for i in all_files])
df_final.rename(columns=dict(zip(df_final.columns,final_cols)),inplace=True)

print(df_final.head())

我对描述保存文件做了一个小改动。 删除了每行末尾的结果字段和“,”

文件：cis_report.csv

Description
1.1 Database Placement
1.2 Use dedicated least privilaged account
1.3 Diable MySQL history

我得到的输出是-

Description dc1pp1v01 dc1pp2v01 dc2pp1v01
0                      1.1 Database Placement        PASSED        PASSED        PASSED
1  1.2 Use dedicated least privilaged account        PASSED        PASSED        PASSED
2                    1.3 Diable MySQL history        PASSED        PASSED        FAILED

Answer 4

你已经有一个赢家，但是：

import csv
from pathlib import Path

path = Path('/Users/Python/Data')

# Read the reports and store the results in a 2-dim list
results = []
for file in path.glob('dc*.co.uk.csv'):
    with open(file, 'r') as fin:
        results += [[file.name.split('.')[0]]
                    + [row[2] for row in csv.reader(fin)][1:]]

# Read the row labels
with open(path / 'cis_report.csv', 'r') as fin:
    labels = [row[0] for row in csv.reader(fin)]

# Prepare the output
output = [[label] + [result[i] for result in results]
          for i, label in enumerate(labels)]

# Write the output
with open(path / 'cis_reports_merged.csv', 'w') as fout:
    csv.writer(fout, delimiter='\t').writerows(output)