如何使用熊猫导入多个 csv 文件并连接成一个数据帧

Question

我遇到问题< code >没有要连接的对象。我不能进口。csv文件，将它们连接成一个数据帧。我在用熊猫。旧答案没有帮助我，所以请不要标记为重复。

文件夹结构就是这样

main/*.csv
main/name1/name1/*.csv
main/name1/name2/*.csv
main/name2/name1/*.csv
main/name3/*.csv

import pandas as pd
import os
import glob

folder_selected = 'C:/Users/jacob/Documents/csv_files'

< li >不起作用

frame = pd.concat(map(pd.read_csv, glob.iglob(os.path.join(folder_selected, "/*.csv"))))

csv_paths = glob.glob('*.csv')
dfs = [pd.read_csv(folder_selected) for folder_selected in csv_paths]
df = pd.concat(dfs)

            all_files = []
            
            all_files = glob.glob (folder_selected + "/*.csv")
            
            file_path = []
            for file in all_files:
                df = pd.read_csv(file, index_col=None, header=0)
                file_path.append(df)
                    
        frame = pd.concat(file_path, axis=0, ignore_index=False)

Answer 1

蟒蛇的路径库是执行此类任务的工具

from pathlib import Path

FOLDER_SELECTED = 'C:/Users/jacob/Documents/csv_files'

path = Path(FOLDER_SELECTED) / Path("main")

# grab all csvs in main and subfolders
df = pd.concat(pd.read_csv(f.name) for f in path.rglob("*.csv"))

如果 CSV 需要预处理，您可以创建一个read_csv函数来处理问题，并将其放在pd.read_csv

Answer 2

检查Dask库如下，它读取许多文件到一个df

>>> import dask.dataframe as dd
>>> df = dd.read_csv('data*.csv')

阅读他们的文档https://examples.dask.org/dataframes/01-data-access.html#Read-CSV-files

Answer 3

您需要以递归方式搜索子目录。

folder = 'C:/Users/jacob/Documents/csv_files'
path = folder+"/**/*.csv"

< li >使用< code>glob.iglob

df = pd.concat(map(pd.read_csv, glob.iglob(path, recursive=True)))

csv_paths = glob.glob(path, recursive=True)
dfs = [pd.read_csv(csv_path) for csv_path in csv_paths]
df = pd.concat(dfs)

file_paths = []
for base, dirs, files in os.walk(folder):
    for file in fnmatch.filter(files, '*.csv'):
        file_paths.append(os.path.join(base, file))
df = pd.concat([pd.read_csv(file) for file in file_paths])

from pathlib import Path
files = Path(folder).rglob('*.csv')
df = pd.concat(map(pd.read_csv, files))