From 247e9e68244054f8def6ca0b700b4f146f9ece39 Mon Sep 17 00:00:00 2001 From: Ivan Alejandro Date: Mon, 4 Nov 2013 16:41:38 -0300 Subject: Add document with good practices for encoding. --- docs/dev/encodings.rst | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++ docs/index.rst | 1 + 2 files changed, 65 insertions(+) create mode 100644 docs/dev/encodings.rst (limited to 'docs') diff --git a/docs/dev/encodings.rst b/docs/dev/encodings.rst new file mode 100644 index 00000000..a3fc4b70 --- /dev/null +++ b/docs/dev/encodings.rst @@ -0,0 +1,64 @@ +.. _encodings: + +Strings encoding problems +========================= + +This document is meant to avoid ``UnicodeError`` (``UnicodeEncodeError`` , ``UnicodeDecodeError``) and to set a base that allows the users to keep away headaches. + + +First approach +-------------- + +One of the problems with python 2 that makes hard to find out problems is the implicit conversion between ``str`` and ``unicode``. + +Look at this code:: + + >>> u'ä'.encode('utf-8') + '\xc3\xa4' + >>> + >>> u'ä'.decode('utf-8') + Traceback (most recent call last): + File "", line 1, in + File "/usr/lib/python2.7/encodings/utf_8.py", line 16, in decode + return codecs.utf_8_decode(input, errors, True) + UnicodeEncodeError: 'ascii' codec can't encode character u'\xe4' in position 0: ordinal not in range(128) + +A situation like this could happen if the user confuse one type for another. 'encode' is a method of ``unicode`` and 'decode' is a method of ``str``, since you call 'decode', python "knows" how to convert from ``unicode`` to ``str`` and then call the 'decode' method, *that* conversion is made with the safe default "ascii" which raises an exception. + + +We need to know which one we are using **every time**. A possible way to avoid mistakes is to use ``leap_assert_type`` at the beginning of each method that has a ``str``/``unicode`` parameter. +The best approach we need to use ``unicode`` internally and when we read/write/transmit data, encode it to bytes (``str``). + + +Examples of problems found +-------------------------- + +* **logging data**: ``logger.debug("some string {0}".format(some_data))`` may fail if we have an ``unicode`` parameter because of the conversion needed to output it. + We need to use ``repr(some_data)`` to avoid encoding problems when sending data to the stdout. An easy way to do it is: ``logger.debug("some string {0!r}".format(some_data))`` + +- **paths encoding**: we should return always ``unicode`` values from helpers and encode them when we need to use it. + The stdlib handles correctly ``unicode`` strings path parameters. + If we want to do something else with the paths, we need to convert them manually using the system encoding. + +Regarding the encoding, use a hardcoded encoding may be wrong. +Instead of encode/decode using for instance 'utf-8', we should use this ``sys.getfilesystemencoding()`` + +For the data stored in a db (or something that is some way isolated from the system) we may want to choose 'utf-8' explicitly. + +Steps to improve code +--------------------- + +#. From now on, keep in mind the difference between ``str`` and ``unicode`` and write code consequently. +#. For each method we can add a ``leap_assert_type(parameter_name, unicode)`` (or ``str``) to avoid type problems. +#. Each time that is possible move towards the unicode 'frontier' (``unicode`` inside, ``str`` (bytes) outside). +#. When is possible update the methods parameters in order to be certain of the data types that we are handling. + +Recommended info +---------------- + +* PyCon 2012 talk: https://www.youtube.com/watch?v=sgHbC6udIqc + * article and transcription: http://nedbatchelder.com/text/unipain.html +* PyConAr 2012 (Spanish): http://www.youtube.com/watch?v=pQJ0emlYv50 +* Overcoming frustrations: http://pythonhosted.org/kitchen/unicode-frustrations.html +* Python's Unicode howto: http://docs.python.org/2/howto/unicode.html +* An encoding primer: http://www.danielmiessler.com/study/encoding/ diff --git a/docs/index.rst b/docs/index.rst index f210be8c..acf97ff3 100644 --- a/docs/index.rst +++ b/docs/index.rst @@ -50,6 +50,7 @@ If you want to contribute to the project, we wrote this for you. dev/workflow dev/resources dev/internationalization + dev/encodings .. dev/internals dev/authors -- cgit v1.2.3 From ae09f76bc63b434e08a9ffc59bf78f054f814bac Mon Sep 17 00:00:00 2001 From: Ivan Alejandro Date: Mon, 11 Nov 2013 14:40:41 -0300 Subject: Make all by default. --- docs/dev/quickstart.rst | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'docs') diff --git a/docs/dev/quickstart.rst b/docs/dev/quickstart.rst index 8ef7dfb8..96dcaeb9 100644 --- a/docs/dev/quickstart.rst +++ b/docs/dev/quickstart.rst @@ -61,7 +61,7 @@ also install the needed dependencies:: Compile the resource files:: - (bitmask)$ make resources + (bitmask)$ make Copy necessary files into system folders, with root privileges:: -- cgit v1.2.3