At a first glance, opening files in Python is easy. All you have to do is call the built-in function
open() and then you start reading from the file. However, often the content to be read is text, not just a binary stream. In this case encodings become relevant. In modern Python this is not much of an issue, as the
open() has a parameter
encoding that allows to specify just that. Even if unspecified, it will use the value from
locale.getpreferredencoding(False) which will usually make everything behave just as expected. Sadly, legacy Python is still around and it does not offer this parameter.
So how can encodings of a file be handled properly? The trivial approach, of course, would be to open the files in binary mode and manually utilize the
decode() methods to take care of the encoding. This, however, is tedious and breaks at the moment you have to pass the file pointer to another library that expects that it can directly read from or write to it. Curiously, legacy Python’s file object does have the encoding attribute and conversion logic. It is just not possible to set it when using the standard built-in
open() function. Luckily, there is a function providing this functionality:
[codecs.open()](https://docs.python.org/3.6/library/codecs.html#codecs.open). This function is available across Python versions and should thus be the first choice when handling text files.