encoding isn't a valid parameter to smart_open

[jayantj]

This isn't a bug, but it can be slightly confusing -

Steps to reproduce

print(open('cp852.tsv.txt', 'r', encoding='cp852').read())
# tímto	budeš
# budem	byli

print(smart_open('cp852.tsv.txt', 'r', encoding='cp852').read())
# UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa1 in position 1: invalid start byte

cp852.tsv.txt

The encoding parameter isn't mentioned in the documentation anywhere, so it isn't reasonable for the above to work. It would be helpful though to log a warning, or raise an exception specifying it is an invalid argument, or to have the parameter supported in the first place (if possible).

[piskvorky]

Definitely a bug -- smart_open must remain a drop-in replacement for open, for local filesystem usage.

Thanks for reporting.

@menshikh-iv we should be passing all extra (unknown) parameters straight to the underlying storage. I believe that's how smart_open always worked, so not sure what the current bug is due to.

[jayantj]

Might have something to do with the fact that encoding isn't a valid argument to open for python2.

[mpenkov]

As @jayantj pointed out, open doesn't take keyword arguments in Py2, so this bug is only reproducible under Py3.

(venv)bash-3.2$ source venv/bin/activate
(venv)bash-3.2$ python bug.py
Traceback (most recent call last):
  File "bug.py", line 3, in <module>
    print(open('cp852.tsv.txt', 'r', encoding='cp852').read())
TypeError: 'encoding' is an invalid keyword argument for this function
(venv)bash-3.2$ source venv3/bin/activate
(venv3)bash-3.2$ python bug.py
tímto   budeš
budem   byli
Traceback (most recent call last):
  File "bug.py", line 7, in <module>
    print(smart_open('cp852.tsv.txt', 'r', encoding='cp852').read())
  File "/Users/misha/git/smart_open/venv3/bin/../lib/python3.5/codecs.py", line 321, in decode
    (result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xa1 in position 1: invalid start byte

The reason encoding isn't being picked up is here. The encoding, which is a keyword argument, isn't being passed to the file_smart_open function.

I'll look into fixing this.