python csv模块读取用逗号分隔的csv，但忽略双引号或单引号内的逗号

Question

我有一个 .csv 文件，其中列值包含一些逗号。以下是示例:

Header: ID     Value           Content                                            Date
        1      34             "market, business"                               12/20/2013
        2      15             "market, business", yesterday, metric            11/21/2014
        3      18             "market," business and yesterday                 10/20/2014
        4      19              yesterday, today,                               11/22/2014

这是 .csv 文件的格式，如果我在 Sublime Text 中打开该文件，它会显示为以下格式:

1, 34, "market, business", 12/20/2013
2, 15, "market, business", "yesterday, metric, 11/21/2014
3, 18, "market," business and yesterday, 10/20/2014
4, 19, yesterday, today, 11/22/2014

但是我想要的是python csv阅读器程序之后是:

[1, 34, "market, business", 12/20/2013]
[2, 15, "market, business" "yesterday metric, 11/21/2014]
[3, 18, "market," business and yesterday, 10/20/2014]
[4, 19, yesterday today, 11/22/2014]

这些只是我拥有的示例数据，“内容”列是这里令人头痛的原因，因为 csv 模块使用“，”作为分隔符，我使用了

reader = csv.reader(f, skipinitialspace=True)

如果所有字符串都在一个双引号内，则它适用于第一行。但如果引号外有逗号(单引号或双引号)，则不适用于第三行和第二行

如何解决这个问题？我现在只是在python中使用传统的csv模块，“panda”有能力解决这个问题吗？

谢谢。

我做了一些更新，我想我想要的是，在不同地方指定逗号的方法...... 现在我粘贴到这里似乎不合理，因为我无法在 csv 模块内部找到来区分字段内分隔符“，”和“，”的差异。即使是Excel也不能...

有什么想法吗？

Answer 1

如果我们可以假设

• 每行以两个用逗号分隔的整数开头，
• 每行以日期结尾，以逗号分隔
• 剩下的所有内容(中间)都属于第三列

那么你的数据可以这样解析:

data = list()
with open('data') as f:
    for line in f:
        parts = line.split(',', 2)
        parts[2:4] = parts[2].rsplit(',', 1)
        parts[:2] = map(int, parts[:2])
        parts[2:] = map(str.strip, parts[2:])
        data.append(parts)

for row in data:
    print(row)

产量

[1, 34, '"market, business"', '12/20/2013']
[2, 15, '"market, business", "yesterday, metric', '11/21/2014']
[3, 18, '"market," business and yesterday', '10/20/2014']
[4, 19, 'yesterday, today', '11/22/2014']

<小时/>

然后你可以创建一个像这样的 DataFrame:

import pandas as pd
df = pd.DataFrame(data, columns=['Id','Value','Content','Date'])
print(df)

产量

   Id  Value                                 Content        Date
0   1     34                      "market, business"  12/20/2013
1   2     15  "market, business", "yesterday, metric  11/21/2014
2   3     18        "market," business and yesterday  10/20/2014
3   4     19                        yesterday, today  11/22/2014