如果是r模式，那么它读入内存时，需要先解码：
f= open('Capture.PNG','r', encoding='ascii',errors='replace')
到了内存里面，它是str

这个str你repr它，会看到最外层的引号，表示它是str字面量

然后这个str还可以eval成python的obj

…xb1\xb1\xe7\x94\xb2\xe5\x88\xb0\xe5\xba\x95\xe8\xaf\xb4\xe4\xba\x86\xe4\xbb\x80\xe4\xb9\x88\xef\xbc\x9f'

#当你上行那么写时，'\xe7'是四个ascii字符，写入py源码文件中。而本文第一行的四个中文是二进制 'e7'写入py文件
#当python解释器读到s=b''这个写法时，它解释器负责把\xe7这四个字符，理解为二进制的e7

characters U+0000 to U+00FF. latin-1 encoding is notable in that it will never produce
a decoding error when reading text of a possibly unknown encoding. Reading a file as
latin-1 might not produce a completely correct text decoding, but it still might be
enough to extract useful data out of it. Also, if you later write the data back out, the
original input data will be preserved.

…ws-command-line-how/388500#388500