【python】csv

CSV生成器

from collections import namedtuple
def csv_reader(fd, delimiter='\t'):
    def gen():
        for i in fd:
            yield i.rstrip('\n').split(delimiter)
    return gen()

def read_tsv(input_file, quotechar=None):
    with open(input_file, 'r', encoding='utf8') as f:
        reader = csv_reader(f)
        headers = next(reader)
        Example = namedtuple('Example', headers) # header
        examples = []
        for line in reader:
            example = Example(*line)
            examples.append(example)
        return examples
print(read_tsv("49"))

输入

num	type
16111	definition
20381	digit-money

输出

[Example(num='16111', type='definition'), Example(num='20381', type='digit-money')]

读5G文件

一、读取5G的csv文件没问题，但是内存就用了11G，仍然没办法进行数据分析的操作。

只能读取的话，

import dask.dataframe as dd
import os
from tqdm import tqdm
TRAIN_PATH = '../input/train.csv'
# Set columns to most suitable type to optimize for memory usage
traintypes = {'fare_amount': 'float32',
              'pickup_datetime': 'str', 
              'pickup_longitude': 'float32',
              'pickup_latitude': 'float32',
              'dropoff_longitude': 'float32',
              'dropoff_latitude': 'float32',
              'passenger_count': 'uint8'
             }
cols = list(traintypes.keys())
chunksize = 10_000_000 # 5 million rows at one go. Or try 10 million

%%time
df_list = []
for df_chunk in tqdm(pd.read_csv(TRAIN_PATH, usecols=cols, dtype=traintypes, chunksize=chunksize)):
    df_list.append(df_chunk)
train_df = pd.concat(df_list)

del df_list
train_df.info()

二、dask

据说可以批处理，apply都能用，待尝试

--------------------------------------更新，已尝试
有一个比较致命的缺点，没办法重置索引。
但是我需要重置索引，所以目前失败，或许可以不重置
目前对于四个操作是秒完成的，train_df 包含5 000 000 + 条数据
（只是介绍一下操作，和dataframe很像，代码是无法运行的）

import pandas as pd
import numpy as np
import time
import datetime
import dask.dataframe as dd
train_df = dd.read_csv('../input/train.csv')
# drop
train_df = train_df.drop(['key','pickup_longitude','pickup_latitude','dropoff_longitude','dropoff_latitude'],axis=1)
# apply
train_df['price'] = train_df.apply(lambda x: func(x.distance, x.fare_amount), axis = 1)
# 强转
train_df['week'] = train_df['week'].astype(int)
# 筛选
train_df = train_df[train_df['price'] > 0]