Read Fernet Key Causes ValueError: Fernet key must be 32 url-safe base64-encoded bytes

Question

在此函数中，我尝试从文件中读取 Fernet 密钥，或者如果文件不包含密钥，则创建一个。

from cryptography.fernet import Fernet
import csv


with open("Keys.txt","rU") as csvfile:
    reader=csv.reader(csvfile)
    KeyFound=0
    print(KeyFound)
    for row in reader:
        try:
            print(row[0])
        except IndexError:
            continue
        if len(row[0])>4:
            print("KEY FOUND")
            KeyFound=1
            print(KeyFound)
            Key=row[0]
            print(Key)
            print(KeyFound)
        else:
            pass
if KeyFound==0:
    Key = Fernet.generate_key()
    print(Key)
    print("Created Key")
    csvfile.close()
#Writing Key to textfile
with open("Keys.txt", "w+") as csvfile:
    headers = ['key']
    writer=csv.DictWriter(csvfile, fieldnames=headers)
    writer.writeheader()
    writer.writerow({'key': Key})
    csvfile.close()
print(Key)
Ecy = Fernet(Key)

我在阅读文件时遇到了困难。当文件被读取时，密钥被读取为：

b'nNjpIl9Ax2LRtm-p6ryCRZ8lRsL0DtuY0f9JeAe2wG0='

但我收到此错误：

ValueError: Fernet key must be 32 url-safe base64-encoded bytes.

在这一行：

Ecy = Fernet(Key)

任何帮助将不胜感激。

Answer 1

这里的问题是密钥是如何写入文件的。

Fernet.generate_key() 返回一个bytes 实例：

>>> key = Fernet.generate_key()
>>> key
b'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='

密钥正按原样写入文件：

>>> with open('keys.csv', 'w+') as f:
...     headers = ['key']
...     writer = csv.DictWriter(f, fieldnames=headers)
...     writer.writeheader()
...     writer.writerow({'key': key})
... 
49
>>> 

如果我们查看文件，我们可以看到内容不是我们所期望的——b 表示已将 python 字节字符串写入文件：

$  cat keys.csv 
key
b'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='

csv.writer 在任何不是字符串的值上调用str。如果在bytes 实例上调用str，您将获得字节实例的字符串化repr，而不是bytes 实例的解码值，这就是您想要的^*.

>>> str(key)
"b'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='"  # <- note the extra quotes...
>>> key.decode('utf-8')
'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='

所以解决方法是在csv.writer之前调用bytes实例的decode方法 收到。

>>> with open('keys.csv', 'w+') as f:
...     headers = ['key']
...     writer = csv.DictWriter(f, fieldnames=headers)
...     writer.writeheader()
...     writer.writerow({'key': key.decode('utf-8')})
... 
46
>>>

这给了我们想要的文件内容：

$  cat keys.csv 
key
ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg=

其余代码按预期工作：

>>> with open('keys.csv') as f:
...     reader = csv.reader(f)
...     next(reader)      # <- skip the header row
...     for row in reader:
...         csv_key = row[0]
...         print(Fernet(csv_key))
... 
['key']                   # <- the headers are printed as a side effect of skipping
<cryptography.fernet.Fernet object at 0x7f3ad62fd4e0>

一个调试技巧。使用print() 调试代码时，有时最好打印对象的repr，而不是在对象上调用str 的结果（这是print() 所做的）。如果对象是字符串，情况尤其如此。例如：

>>> bad_key = str(key)
>>> print(bad_key)                                
b'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='      # <- Looks ok...
>>> print(repr(bad_key))
"b'ZmDfcTF7_60GrrY167zsiPd67pEvs0aGOv2oasOM1Pg='"    # <- See the problem
>>> 
>>> good_str = 'foo'
>>> bad_str = 'foo '
>>> print(bad_str)
foo                             # <- looks like good_str
>>> print(repr(bad_str))
'foo '                          # <- see the trailing space 

^* 如果您使用-b 标志调用Python - python -b myscript.py - 当您第一次尝试在bytes 实例上调用str 时，Python 将发出BytesWarning。如果使用-bb 标志，则会引发异常。