Google

Edit File: designing-unicode-apis.html

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">

<html xmlns="http://www.w3.org/1999/xhtml">
 <head>
 <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
 
 <title>Designing Unicode Aware APIs &mdash; kitchen 1.1.1 documentation</title>
 
 <link rel="stylesheet" href="_static/default.css" type="text/css" />
 <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
 
 <script type="text/javascript">
 var DOCUMENTATION_OPTIONS = {
 URL_ROOT: '',
 VERSION: '1.1.1',
 COLLAPSE_INDEX: false,
 FILE_SUFFIX: '.html',
 HAS_SOURCE: true
 };
 </script>
 <script type="text/javascript" src="_static/jquery.js"></script>
 <script type="text/javascript" src="_static/underscore.js"></script>
 <script type="text/javascript" src="_static/doctools.js"></script>
 <link rel="search" type="application/opensearchdescription+xml"
 title="Search within kitchen 1.1.1 documentation"
 href="_static/opensearch.xml"/>
 <link rel="top" title="kitchen 1.1.1 documentation" href="index.html" />
 <link rel="up" title="Using kitchen to write good code" href="tutorial.html" />
 <link rel="next" title="Kitchen API" href="api-overview.html" />
 <link rel="prev" title="Overcoming frustration: Correctly using unicode in python2" href="unicode-frustrations.html" /> 
 </head>
 <body>
 <div class="related">
 <h3>Navigation</h3>
 <ul>
 <li class="right" style="margin-right: 10px">
 <a href="genindex.html" title="General Index"
 accesskey="I">index</a></li>
 <li class="right" >
 <a href="py-modindex.html" title="Python Module Index"
 >modules</a> |</li>
 <li class="right" >
 <a href="api-overview.html" title="Kitchen API"
 accesskey="N">next</a> |</li>
 <li class="right" >
 <a href="unicode-frustrations.html" title="Overcoming frustration: Correctly using unicode in python2"
 accesskey="P">previous</a> |</li>
 <li><a href="index.html">kitchen 1.1.1 documentation</a> &raquo;</li>
 <li><a href="tutorial.html" accesskey="U">Using kitchen to write good code</a> &raquo;</li> 
 </ul>
 </div>

<div class="document">
 <div class="documentwrapper">
 <div class="bodywrapper">
 <div class="body">
 
 <div class="section" id="designing-unicode-aware-apis">
<h1><a class="toc-backref" href="#id1">Designing Unicode Aware APIs</a><a class="headerlink" href="#designing-unicode-aware-apis" title="Permalink to this headline">¶</a></h1>
APIs that deal with byte <tt class="xref py py-class docutils literal">str</tt> and <tt class="xref py py-class docutils literal">unicode</tt> strings are
difficult to get right. Here are a few strategies with pros and cons of each.
<div class="contents topic" id="contents">
Contents
<ul class="simple">
<li><a class="reference internal" href="#designing-unicode-aware-apis" id="id1">Designing Unicode Aware APIs</a><ul>
<li><a class="reference internal" href="#take-either-bytes-or-unicode-output-only-unicode" id="id2">Take either bytes or unicode, output only unicode</a></li>
<li><a class="reference internal" href="#take-either-bytes-or-unicode-output-the-same-type" id="id3">Take either bytes or unicode, output the same type</a></li>
<li><a class="reference internal" href="#separate-functions" id="id4">Separate functions</a></li>
<li><a class="reference internal" href="#deciding-whether-to-take-str-or-unicode-when-no-value-is-returned" id="id5">Deciding whether to take str or unicode when no value is returned</a><ul>
<li><a class="reference internal" href="#writing-to-external-data" id="id6">Writing to external data</a></li>
<li><a class="reference internal" href="#updating-data-structures" id="id7">Updating data structures</a></li>
</ul>
</li>
<li><a class="reference internal" href="#apis-to-avoid" id="id8">APIs to Avoid</a><ul>
<li><a class="reference internal" href="#returning-unicode-unless-a-conversion-fails" id="id9">Returning unicode unless a conversion fails</a></li>
<li><a class="reference internal" href="#ignoring-values-with-no-chance-of-recovery" id="id10">Ignoring values with no chance of recovery</a></li>
<li><a class="reference internal" href="#raising-a-unicodeexception-with-no-chance-of-recovery" id="id11">Raising a UnicodeException with no chance of recovery</a></li>
</ul>
</li>
<li><a class="reference internal" href="#knowing-your-data" id="id12">Knowing your data</a><ul>
<li><a class="reference internal" href="#do-you-need-to-operate-on-both-bytes-and-unicode" id="id13">Do you need to operate on both bytes and unicode?</a></li>
<li><a class="reference internal" href="#can-you-restrict-the-encodings" id="id14">Can you restrict the encodings?</a><ul>
<li><a class="reference internal" href="#single-byte-encodings" id="id15">Single byte encodings</a></li>
<li><a class="reference internal" href="#multibyte-encodings" id="id16">Multibyte encodings</a><ul>
<li><a class="reference internal" href="#fixed-width" id="id17">Fixed width</a></li>
<li><a class="reference internal" href="#variable-width" id="id18">Variable Width</a><ul>
<li><a class="reference internal" href="#ascii-compatible" id="id19">ASCII compatible</a></li>
<li><a class="reference internal" href="#escaped" id="id20">Escaped</a></li>
<li><a class="reference internal" href="#other" id="id21">Other</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
<div class="section" id="take-either-bytes-or-unicode-output-only-unicode">
<h2><a class="toc-backref" href="#id2">Take either bytes or unicode, output only unicode</a><a class="headerlink" href="#take-either-bytes-or-unicode-output-only-unicode" title="Permalink to this headline">¶</a></h2>
In this strategy, you allow the user to enter either <tt class="xref py py-class docutils literal">unicode</tt> strings
or byte <tt class="xref py py-class docutils literal">str</tt> but what you give back is always <tt class="xref py py-class docutils literal">unicode</tt>. This
strategy is easy for novice endusers to start using immediately as they will
be able to feed either type of string into the function and get back a string
that they can use in other places.
However, it does lead to the novice writing code that functions correctly when
testing it with <a class="reference internal" href="glossary.html#term-ascii">ASCII</a>-only data but fails when given data that contains
non-<a class="reference internal" href="glossary.html#term-ascii">ASCII</a> characters. Worse, if your API is not designed to be
flexible, the consumer of your code won&#8217;t be able to easily correct those
problems once they find them.
Here&#8217;s a good API that uses this strategy:
<div class="highlight-python"><div class="highlight"><pre>from kitchen.text.converters import to_unicode

def truncate(msg, max_length, encoding=&#39;utf8&#39;, errors=&#39;replace&#39;):
 msg = to_unicode(msg, encoding, errors)
 return msg[:max_length]
</pre></div>
</div>
The call to <tt class="xref py py-func docutils literal">truncate()</tt> starts with the essential parameters for
performing the task. It ends with two optional keyword arguments that define
the encoding to use to transform from a byte <tt class="xref py py-class docutils literal">str</tt> to <tt class="xref py py-class docutils literal">unicode</tt>
and the strategy to use if undecodable bytes are encountered. The defaults
may vary depending on the use cases you have in mind. When the output is
generally going to be printed for the user to see, <tt class="docutils literal">errors='replace'</tt> is
a good default. If you are constructing keys to a database, raisng an
exception (with <tt class="docutils literal">errors='strict'</tt>) may be a better default. In either case,
having both parameters allows the person using your API to choose how they
want to handle any problems. Having the values is also a clue to them that
a conversion from byte <tt class="xref py py-class docutils literal">str</tt> to <tt class="xref py py-class docutils literal">unicode</tt> string is going to
occur.
<div class="admonition note">
Note
If you&#8217;re targeting python-3.1 and above, <tt class="docutils literal">errors='surrogateescape'</tt> may
be a better default than <tt class="docutils literal">errors='strict'</tt>. You need to be mindful of
a few things when using <tt class="docutils literal">surrogateescape</tt> though:
<ul class="simple">
<li><tt class="docutils literal">surrogateescape</tt> will cause issues if a non-<a class="reference internal" href="glossary.html#term-ascii">ASCII</a> compatible
encoding is used (for instance, UTF-16 and UTF-32.) That makes it
unhelpful in situations where a true general purpose method of encoding
must be found. <a class="pep reference external" href="http://www.python.org/dev/peps/pep-0383">PEP 383</a> mentions that <tt class="docutils literal">surrogateescape</tt> was
specifically designed with the limitations of translating using system
locales (where <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> compatibility is generally seen as
inescapable) so you should keep that in mind.</li>
<li>If you use <tt class="docutils literal">surrogateescape</tt> to decode from <tt class="xref py py-class docutils literal">bytes</tt>
to <tt class="xref py py-class docutils literal">unicode</tt> you will need to use an error handler other than
<tt class="docutils literal">strict</tt> to encode as the lone surrogate that this error handler
creates makes for invalid unicode that must be handled when encoding.
In Python-3.1.2 or less, a bug in the encoder error handlers mean that
you can only use <tt class="docutils literal">surrogateescape</tt> to encode; anything else will throw
an error.</li>
</ul>
Evaluate your usages of the variables in question to see what makes sense.
</div>
Here&#8217;s a bad example of using this strategy:
<div class="highlight-python"><div class="highlight"><pre>from kitchen.text.converters import to_unicode

msg = to_unicode(msg, encoding=&#39;euc_jp&#39;, errors=&#39;ignore&#39;)
new_msg = truncate(msg, 5)
</pre></div>
</div>
</div>
<div class="section" id="take-either-bytes-or-unicode-output-the-same-type">
<h2><a class="toc-backref" href="#id3">Take either bytes or unicode, output the same type</a><a class="headerlink" href="#take-either-bytes-or-unicode-output-the-same-type" title="Permalink to this headline">¶</a></h2>
This strategy is sometimes called polymorphic because the type of data that is
returned is dependent on the type of data that is received. The concept is
that when you are given a byte <tt class="xref py py-class docutils literal">str</tt> to process, you return a byte
<tt class="xref py py-class docutils literal">str</tt> in your output. When you are given <tt class="xref py py-class docutils literal">unicode</tt> strings to
process, you return <tt class="xref py py-class docutils literal">unicode</tt> strings in your output.
This can work well for end users as the ones that know about the difference
between the two string types will already have transformed the strings to
their desired type before giving it to this function. The ones that don&#8217;t can
remain blissfully ignorant (at least, as far as your function is concerned) as
the function does not change the type.
In cases where the encoding of the byte <tt class="xref py py-class docutils literal">str</tt> is known or can be
discovered based on the input data this works well. If you can&#8217;t figure out
the input encoding, however, this strategy can fail in any of the following
cases:
<ol class="arabic simple">
<li>It needs to do an internal conversion between byte <tt class="xref py py-class docutils literal">str</tt> and
<tt class="xref py py-class docutils literal">unicode</tt> string.</li>
<li>It cannot return the same data as either a <tt class="xref py py-class docutils literal">unicode</tt> string or byte
<tt class="xref py py-class docutils literal">str</tt>.</li>
<li>You may need to deal with byte strings that are not byte-compatible with
<a class="reference internal" href="glossary.html#term-ascii">ASCII</a></li>
</ol>
First, a couple examples of using this strategy in a good way:
<div class="highlight-python"><div class="highlight"><pre>def translate(msg, table):
 replacements = table.keys()
 new_msg = []
 for index, char in enumerate(msg):
 if char in replacements:
 new_msg.append(table[char])
 else:
 new_msg.append(char)

return &#39;&#39;.join(new_msg)
</pre></div>
</div>
In this example, all of the strings that we use (except the empty string which
is okay because it doesn&#8217;t have any characters to encode) come from outside of
the function. Due to that, the user is responsible for making sure that the
<tt class="xref py py-attr docutils literal">msg</tt>, and the keys and values in <tt class="xref py py-attr docutils literal">table</tt> all match in terms of
type (<tt class="xref py py-class docutils literal">unicode</tt> vs <tt class="xref py py-class docutils literal">str</tt>) and encoding (You can do some error
checking to make sure the user gave all the same type but you can&#8217;t do the
same for the user giving different encodings). You do not need to make
changes to the string that require you to know the encoding or type of the
string; everything is a simple replacement of one element in the array of
characters in message with the character in table.
<div class="highlight-python"><pre>import json
from kitchen.text.converters import to_unicode, to_bytes

def first_field_from_json_data(json_string):
    '''Return the first field in a json data structure.

The format of the json data is a simple list of strings.
    '["one", "two", "three"]'
    '''
    if isinstance(json_string, unicode):
        # On all python versions, json.loads() returns unicode if given
        # a unicode string
        return json.loads(json_string)[0]

# Byte str: figure out which encoding we're dealing with
    if '\x00' not in json_data[:2]
        encoding = 'utf8'
    elif '\x00\x00\x00' == json_data[:3]:
        encoding = 'utf-32-be'
    elif '\x00\x00\x00' == json_data[1:4]:
        encoding = 'utf-32-le'
    elif '\x00' == json_data[0] and '\x00' == json_data[2]:
        encoding = 'utf-16-be'
    else:
        encoding = 'utf-16-le'

data = json.loads(unicode(json_string, encoding))
 return data[0].encode(encoding)</pre>
</div>
In this example the function takes either a byte <tt class="xref py py-class docutils literal">str</tt> type or
a <tt class="xref py py-class docutils literal">unicode</tt> string that has a list in json format and returns the first
field from it as the type of the input string. The first section of code is
very straightforward; we receive a <tt class="xref py py-class docutils literal">unicode</tt> string, parse it with
a function, and then return the first field from our parsed data (which our
function returned to us as json data).
The second portion that deals with byte <tt class="xref py py-class docutils literal">str</tt> is not so
straightforward. Before we can parse the string we have to determine what
characters the bytes in the string map to. If we didn&#8217;t do that, we wouldn&#8217;t
be able to properly find which characters are present in the string. In order
to do that we have to figure out the encoding of the byte <tt class="xref py py-class docutils literal">str</tt>.
Luckily, the json specification states that all strings are unicode and
encoded with one of UTF32be, UTF32le, UTF16be, UTF16le, or <a class="reference internal" href="glossary.html#term-utf-8">UTF-8</a>. It further
defines the format such that the first two characters are always
<a class="reference internal" href="glossary.html#term-ascii">ASCII</a>. Each of these has a different sequence of NULLs when they
encode an <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> character. We can use that to detect which encoding
was used to create the byte <tt class="xref py py-class docutils literal">str</tt>.
Finally, we return the byte <tt class="xref py py-class docutils literal">str</tt> by encoding the <tt class="xref py py-class docutils literal">unicode</tt> back
to a byte <tt class="xref py py-class docutils literal">str</tt>.
As you can see, in this example we have to convert from byte <tt class="xref py py-class docutils literal">str</tt> to
<tt class="xref py py-class docutils literal">unicode</tt> and back. But we know from the json specification that byte
<tt class="xref py py-class docutils literal">str</tt> has to be one of a limited number of encodings that we are able
to detect. That ability makes this strategy work.
Now for some examples of using this strategy in ways that fail:
<div class="highlight-python"><div class="highlight"><pre>import unicodedata
def first_char(msg):
 &#39;&#39;&#39;Return the first character in a string&#39;&#39;&#39;
 if not isinstance(msg, unicode):
 try:
 msg = unicode(msg, &#39;utf8&#39;)
 except UnicodeError:
 msg = unicode(msg, &#39;latin1&#39;)
 msg = unicodedata.normalize(&#39;NFC&#39;, msg)
 return msg[0]
</pre></div>
</div>
If you look at that code and think that there&#8217;s something fragile and prone to
breaking in the <tt class="docutils literal">try: except:</tt> block you are correct in being suspicious.
This code will fail on multi-byte character sets that aren&#8217;t <a class="reference internal" href="glossary.html#term-utf-8">UTF-8</a>. It
can also fail on data where the sequence of bytes is valid <a class="reference internal" href="glossary.html#term-utf-8">UTF-8</a> but
the bytes are actually of a different encoding. The reasons this code fails
is that we don&#8217;t know what encoding the bytes are in and the code must convert
from a byte <tt class="xref py py-class docutils literal">str</tt> to a <tt class="xref py py-class docutils literal">unicode</tt> string in order to function.
In order to make this code robust we must know the encoding of <tt class="xref py py-attr docutils literal">msg</tt>.
The only way to know that is to ask the user so the API must do that:
<div class="highlight-python"><div class="highlight"><pre>import unicodedata
def number_of_chars(msg, encoding=&#39;utf8&#39;, errors=&#39;strict&#39;):
 if not isinstance(msg, unicode):
 msg = unicode(msg, encoding, errors)
 msg = unicodedata.normalize(&#39;NFC&#39;, msg)
 return len(msg)
</pre></div>
</div>
Another example of failure:
<div class="highlight-python"><div class="highlight"><pre>import os
def listdir(directory):
 files = os.listdir(directory)
 if isinstance(directory, str):
 return files
 # files could contain both bytes and unicode
 new_files = []
 for filename in files:
 if not isinstance(filename, unicode):
 # What to do here?
 continue
 new_files.appen(filename)
 return new_files
</pre></div>
</div>
This function illustrates the second failure mode. Here, not all of the
possible values can be represented as <tt class="xref py py-class docutils literal">unicode</tt> without knowing more
about the encoding of each of the filenames involved. Since each filename
could have a different encoding there&#8217;s a few different options to pursue. We
could make this function always return byte <tt class="xref py py-class docutils literal">str</tt> since that can
accurately represent anything that could be returned. If we want to return
<tt class="xref py py-class docutils literal">unicode</tt> we need to at least allow the user to specify what to do in
case of an error decoding the bytes to <tt class="xref py py-class docutils literal">unicode</tt>. We can also let the
user specify the encoding to use for doing the decoding but that won&#8217;t help in
all cases since not all files will be in the same encoding (or even
necessarily in any encoding):
<div class="highlight-python"><div class="highlight"><pre>import locale
import os
def listdir(directory, encoding=locale.getpreferredencoding(), errors=&#39;strict&#39;):
 # Note: In python-3.1+, surrogateescape may be a better default
 files = os.listdir(directory)
 if isinstance(directory, str):
 return files
 new_files = []
 for filename in files:
 if not isinstance(filename, unicode):
 filename = unicode(filename, encoding=encoding, errors=errors)
 new_files.append(filename)
 return new_files
</pre></div>
</div>
Note that although we use <tt class="xref py py-attr docutils literal">errors</tt> in this example as what to pass to
the codec that decodes to <tt class="xref py py-class docutils literal">unicode</tt> we could also have an
<tt class="xref py py-attr docutils literal">errors</tt> argument that decides other things to do like skip a filename
entirely, return a placeholder (<tt class="docutils literal">Nondisplayable filename</tt>), or raise an
exception.
This leaves us with one last failure to describe:
<div class="highlight-python"><div class="highlight"><pre>def first_field(csv_string):
 &#39;&#39;&#39;Return the first field in a comma separated values string.&#39;&#39;&#39;
 try:
 return csv_string[:csv_string.index(&#39;,&#39;)]
 except ValueError:
 return csv_string
</pre></div>
</div>
This code looks simple enough. The hidden error here is that we are searching
for a comma character in a byte <tt class="xref py py-class docutils literal">str</tt> but not all encodings will use
the same sequence of bytes to represent the comma. If you use an encoding
that&#8217;s not <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> compatible on the byte level, then the literal comma
<tt class="docutils literal">','</tt> in the above code will match inappropriate bytes. Some examples of
how it can fail:
<ul class="simple">
<li>Will find the byte representing an <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> comma in another character</li>
<li>Will find the comma but leave trailing garbage bytes on the end of the
string</li>
<li>Will not match the character that represents the comma in this encoding</li>
</ul>
There are two ways to solve this. You can either take the encoding value from
the user or you can take the separator value from the user. Of the two,
taking the encoding is the better option for two reasons:
<ol class="arabic">
<li>Taking a separator argument doesn&#8217;t clearly document for the API user that
the reason they must give it is to properly match the encoding of the
<tt class="xref py py-attr docutils literal">csv_string</tt>. They&#8217;re just as likely to think that it&#8217;s simply a way
to specify an alternate character (like &#8221;:&#8221; or &#8220;|&#8221;) for the separator.
</li>
<li>It&#8217;s possible for a variable width encoding to reuse the same byte sequence
for different characters in multiple sequences.
<div class="admonition note">
Note
<a class="reference internal" href="glossary.html#term-utf-8">UTF-8</a> is resistant to this as any character&#8217;s sequence of
bytes will never be a subset of another character&#8217;s sequence of bytes.
</div>
</li>
</ol>
With that in mind, here&#8217;s how to improve the API:
<div class="highlight-python"><div class="highlight"><pre>def first_field(csv_string, encoding=&#39;utf-8&#39;, errors=&#39;replace&#39;):
 if not isinstance(csv_string, unicode):
 u_string = unicode(csv_string, encoding, errors)
 is_unicode = False
 else:
 u_string = csv_string

try:
 field = u_string[:U_string.index(u&#39;,&#39;)]
 except ValueError:
 return csv_string

if not is_unicode:
 field = field.encode(encoding, errors)
 return field
</pre></div>
</div>
<div class="admonition note">
Note
If you decide you&#8217;ll never encounter a variable width encoding that reuses
byte sequences you can use this code instead:
<div class="last highlight-python"><div class="highlight"><pre>def first_field(csv_string, encoding=&#39;utf-8&#39;):
 try:
 return csv_string[:csv_string.index(&#39;,&#39;.encode(encoding))]
 except ValueError:
 return csv_string
</pre></div>
</div>
</div>
</div>
<div class="section" id="separate-functions">
<h2><a class="toc-backref" href="#id4">Separate functions</a><a class="headerlink" href="#separate-functions" title="Permalink to this headline">¶</a></h2>
Sometimes you want to be able to take either byte <tt class="xref py py-class docutils literal">str</tt> or
<tt class="xref py py-class docutils literal">unicode</tt> strings, perform similar operations on either one and then
return data in the same format as was given. Probably the easiest way to do
that is to have separate functions for each and adopt a naming convention to
show that one is for working with byte <tt class="xref py py-class docutils literal">str</tt> and the other is for
working with <tt class="xref py py-class docutils literal">unicode</tt> strings:
<div class="highlight-python"><div class="highlight"><pre>def translate_b(msg, table):
 &#39;&#39;&#39;Replace values in str with other byte values like unicode.translate&#39;&#39;&#39;
 if not isinstance(msg, str):
 raise TypeError(&#39;msg must be of type str&#39;)
 str_table = [chr(s) for s in xrange(0,256)]
 delete_chars = []
 for chr_val in (k for k in table.keys() if isinstance(k, int)):
 if chr_val &gt; 255:
 raise ValueError(&#39;Keys in table must not exceed 255)&#39;)
 if table[chr_val] == None:
 delete_chars.append(chr(chr_val))
 elif isinstance(table[chr_val], int):
 if table[chr_val] &gt; 255:
 raise TypeError(&#39;table values cannot be more than 255 or less than 0&#39;)
 str_table[chr_val] = chr(table[chr_val])
 else:
 if not isinstance(table[chr_val], str):
 raise TypeError(&#39;character mapping must return integer, None or str&#39;)
 str_table[chr_val] = table[chr_val]
 str_table = &#39;&#39;.join(str_table)
 delete_chars = &#39;&#39;.join(delete_chars)
 return msg.translate(str_table, delete_chars)

def translate(msg, table):
 &#39;&#39;&#39;Replace values in a unicode string with other values&#39;&#39;&#39;
 if not isinstance(msg, unicode):
 raise TypeError(&#39;msg must be of type unicode&#39;)
 return msg.translate(table)
</pre></div>
</div>
There&#8217;s several things that we have to do in this API:
<ul class="simple">
<li>Because the function names might not be enough of a clue to the user of the
functions of the value types that are expected, we have to check that the
types are correct.</li>
<li>We keep the behaviour of the two functions as close to the same as possible,
just with byte <tt class="xref py py-class docutils literal">str</tt> and <tt class="xref py py-class docutils literal">unicode</tt> strings substituted for
each other.</li>
</ul>
</div>
<div class="section" id="deciding-whether-to-take-str-or-unicode-when-no-value-is-returned">
<h2><a class="toc-backref" href="#id5">Deciding whether to take str or unicode when no value is returned</a><a class="headerlink" href="#deciding-whether-to-take-str-or-unicode-when-no-value-is-returned" title="Permalink to this headline">¶</a></h2>
Not all functions have a return value. Sometimes a function is there to
interact with something external to python, for instance, writing a file out
to disk or a method exists to update the internal state of a data structure.
One of the main questions with these APIs is whether to take byte
<tt class="xref py py-class docutils literal">str</tt>, <tt class="xref py py-class docutils literal">unicode</tt> string, or both. The answer depends on your
use case but I&#8217;ll give some examples here.
<div class="section" id="writing-to-external-data">
<h3><a class="toc-backref" href="#id6">Writing to external data</a><a class="headerlink" href="#writing-to-external-data" title="Permalink to this headline">¶</a></h3>
When your information is going to an external data source like writing to
a file you need to decide whether to take in <tt class="xref py py-class docutils literal">unicode</tt> strings or byte
<tt class="xref py py-class docutils literal">str</tt>. Remember that most external data sources are not going to be
dealing with unicode directly. Instead, they&#8217;re going to be dealing with
a sequence of bytes that may be interpreted as unicode. With that in mind,
you either need to have the user give you a byte <tt class="xref py py-class docutils literal">str</tt> or convert to
a byte <tt class="xref py py-class docutils literal">str</tt> inside the function.
Next you need to think about the type of data that you&#8217;re receiving. If it&#8217;s
textual data, (for instance, this is a chat client and the user is typing
messages that they expect to be read by another person) it probably makes sense to
take in <tt class="xref py py-class docutils literal">unicode</tt> strings and do the conversion inside your function.
On the other hand, if this is a lower level function that&#8217;s passing data into
a network socket, it probably should be taking byte <tt class="xref py py-class docutils literal">str</tt> instead.
Just as noted in the API notes above, you should specify an <tt class="xref py py-attr docutils literal">encoding</tt>
and <tt class="xref py py-attr docutils literal">errors</tt> argument if you need to transform from <tt class="xref py py-class docutils literal">unicode</tt>
string to byte <tt class="xref py py-class docutils literal">str</tt> and you are unable to guess the encoding from the
data itself.
</div>
<div class="section" id="updating-data-structures">
<h3><a class="toc-backref" href="#id7">Updating data structures</a><a class="headerlink" href="#updating-data-structures" title="Permalink to this headline">¶</a></h3>
Sometimes your API is just going to update a data structure and not
immediately output that data anywhere. Just as when writing external data,
you should think about both what your function is going to do with the data
eventually and what the caller of your function is thinking that they&#8217;re
giving you. Most of the time, you&#8217;ll want to take <tt class="xref py py-class docutils literal">unicode</tt> strings
and enter them into the data structure as <tt class="xref py py-class docutils literal">unicode</tt> when the data is
textual in nature. You&#8217;ll want to take byte <tt class="xref py py-class docutils literal">str</tt> and enter them into
the data structure as byte <tt class="xref py py-class docutils literal">str</tt> when the data is not text. Use
a naming convention so the user knows what&#8217;s expected.
</div>
</div>
<div class="section" id="apis-to-avoid">
<h2><a class="toc-backref" href="#id8">APIs to Avoid</a><a class="headerlink" href="#apis-to-avoid" title="Permalink to this headline">¶</a></h2>
There are a few APIs that are just wrong. If you catch yourself making an API
that does one of these things, change it before anyone sees your code.
<div class="section" id="returning-unicode-unless-a-conversion-fails">
<h3><a class="toc-backref" href="#id9">Returning unicode unless a conversion fails</a><a class="headerlink" href="#returning-unicode-unless-a-conversion-fails" title="Permalink to this headline">¶</a></h3>
This type of API usually deals with byte <tt class="xref py py-class docutils literal">str</tt> at some point and
converts it to <tt class="xref py py-class docutils literal">unicode</tt> because it&#8217;s usually thought to be text.
However, there are times when the bytes fail to convert to a <tt class="xref py py-class docutils literal">unicode</tt>
string. When that happens, this API returns the raw byte <tt class="xref py py-class docutils literal">str</tt> instead
of a <tt class="xref py py-class docutils literal">unicode</tt> string. One example of this is present in the <a class="reference external" href="http://docs.python.org/library">python standard library</a>:
python2&#8217;s <a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a>:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; import os
&gt;&gt;&gt; import locale
&gt;&gt;&gt; locale.getpreferredencoding()
&#39;UTF-8&#39;
&gt;&gt;&gt; os.mkdir(&#39;/tmp/mine&#39;)
&gt;&gt;&gt; os.chdir(&#39;/tmp/mine&#39;)
&gt;&gt;&gt; open(&#39;nonsense_char_\xff&#39;, &#39;w&#39;).close()
&gt;&gt;&gt; open(&#39;all_ascii&#39;, &#39;w&#39;).close()
&gt;&gt;&gt; os.listdir(u&#39;.&#39;)
[u&#39;all_ascii&#39;, &#39;nonsense_char_\xff&#39;]
</pre></div>
</div>
The problem with APIs like this is that they cause failures that are hard to
debug because they don&#8217;t happen where the variables are set. For instance,
let&#8217;s say you take the filenames from <a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a> and give it to this
function:
<div class="highlight-python"><div class="highlight"><pre>def normalize_filename(filename):
 &#39;&#39;&#39;Change spaces and dashes into underscores&#39;&#39;&#39;
 return filename.translate({ord(u&#39; &#39;):u&#39;_&#39;, ord(u&#39; &#39;):u&#39;_&#39;})
</pre></div>
</div>
When you test this, you use filenames that all are decodable in your preferred
encoding and everything seems to work. But when this code is run on a machine
that has filenames in multiple encodings the filenames returned by
<a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a> suddenly include byte <tt class="xref py py-class docutils literal">str</tt>. And byte <tt class="xref py py-class docutils literal">str</tt>
has a different <a class="reference external" href="http://docs.python.org/library/string.html#string.translate" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">string.translate()</tt></a> function that takes different values.
So the code raises an exception where it&#8217;s not immediately obvious that
<a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a> is at fault.
</div>
<div class="section" id="ignoring-values-with-no-chance-of-recovery">
<h3><a class="toc-backref" href="#id10">Ignoring values with no chance of recovery</a><a class="headerlink" href="#ignoring-values-with-no-chance-of-recovery" title="Permalink to this headline">¶</a></h3>
An early version of python3 attempted to fix the <a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a> problem
pointed out in the last section by returning all values that were decodable to
<tt class="xref py py-class docutils literal">unicode</tt> and omitting the filenames that were not. This lead to the
following output:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; import os
&gt;&gt;&gt; import locale
&gt;&gt;&gt; locale.getpreferredencoding()
&#39;UTF-8&#39;
&gt;&gt;&gt; os.mkdir(&#39;/tmp/mine&#39;)
&gt;&gt;&gt; os.chdir(&#39;/tmp/mine&#39;)
&gt;&gt;&gt; open(b&#39;nonsense_char_\xff&#39;, &#39;w&#39;).close()
&gt;&gt;&gt; open(&#39;all_ascii&#39;, &#39;w&#39;).close()
&gt;&gt;&gt; os.listdir(&#39;.&#39;)
[&#39;all_ascii&#39;]
</pre></div>
</div>
The issue with this type of code is that it is silently doing something
surprising. The caller expects to get a full list of files back from
<a class="reference external" href="http://docs.python.org/library/os.html#os.listdir" title="(in Python v2.7)"><tt class="xref py py-func docutils literal">os.listdir()</tt></a>. Instead, it silently ignores some of the files, returning
only a subset. This leads to code that doesn&#8217;t do what is expected that may
go unnoticed until the code is in production and someone notices that
something important is being missed.
</div>
<div class="section" id="raising-a-unicodeexception-with-no-chance-of-recovery">
<h3><a class="toc-backref" href="#id11">Raising a UnicodeException with no chance of recovery</a><a class="headerlink" href="#raising-a-unicodeexception-with-no-chance-of-recovery" title="Permalink to this headline">¶</a></h3>
Believe it or not, a few libraries exist that make it impossible to deal
with unicode text without raising a <tt class="xref py py-exc docutils literal">UnicodeError</tt>. What seems to occur
in these libraries is that the library has functions that expect to receive
a <tt class="xref py py-class docutils literal">unicode</tt> string. However, internally, those functions call other
functions that expect to receive a byte <tt class="xref py py-class docutils literal">str</tt>. The programmer of the
API was smart enough to convert from a <tt class="xref py py-class docutils literal">unicode</tt> string to a byte
<tt class="xref py py-class docutils literal">str</tt> but they did not give the user the chance to specify the
encodings to use or how to deal with errors. This results in exceptions when
the user passes in a byte <tt class="xref py py-class docutils literal">str</tt> because the initial function wants
a <tt class="xref py py-class docutils literal">unicode</tt> string and exceptions when the user passes in
a <tt class="xref py py-class docutils literal">unicode</tt> string because the function can&#8217;t convert the string to
bytes in the encoding that it&#8217;s selected.
Do not put the user in the position of not being able to use your API without
raising a <tt class="xref py py-exc docutils literal">UnicodeError</tt> with certain values. If you can only safely
take <tt class="xref py py-class docutils literal">unicode</tt> strings, document that byte <tt class="xref py py-class docutils literal">str</tt> is not allowed
and vice versa. If you have to convert internally, make sure to give the
caller of your function parameters to control the encoding and how to treat
errors that may occur during the encoding/decoding process. If your code will
raise a <tt class="xref py py-exc docutils literal">UnicodeError</tt> with non-<a class="reference internal" href="glossary.html#term-ascii">ASCII</a> values no matter what, you
should probably rethink your API.
</div>
</div>
<div class="section" id="knowing-your-data">
<h2><a class="toc-backref" href="#id12">Knowing your data</a><a class="headerlink" href="#knowing-your-data" title="Permalink to this headline">¶</a></h2>
If you&#8217;ve read all the way down to this section without skipping you&#8217;ve seen
several admonitions about the type of data you are processing affecting the
viability of the various API choices.
Here&#8217;s a few things to consider in your data:
<div class="section" id="do-you-need-to-operate-on-both-bytes-and-unicode">
<h3><a class="toc-backref" href="#id13">Do you need to operate on both bytes and unicode?</a><a class="headerlink" href="#do-you-need-to-operate-on-both-bytes-and-unicode" title="Permalink to this headline">¶</a></h3>
Much of the data in libraries, programs, and the general environment outside
of python is written where strings are sequences of bytes. So when we
interact with data that comes from outside of python or data that is about to
leave python it may make sense to only operate on the data as a byte
<tt class="xref py py-class docutils literal">str</tt>. There&#8217;s two times when this may make sense:
<ol class="arabic simple">
<li>The user is intended to hand the data to the function and then the function
takes care of sending the data outside of python (to the filesystem, over
the network, etc).</li>
<li>The data is not representable as text. For instance, writing a binary
file format.</li>
</ol>
Even when your code is operating in this area you still need to think a little
more about your data. For instance, it might make sense for the person using
your API to pass in <tt class="xref py py-class docutils literal">unicode</tt> strings and let the function convert that
into the byte <tt class="xref py py-class docutils literal">str</tt> that it then sends over the wire.
There are also times when it might make sense to operate only on
<tt class="xref py py-class docutils literal">unicode</tt> strings. <tt class="xref py py-class docutils literal">unicode</tt> represents text so anytime that
you are working on textual data that isn&#8217;t going to leave python it has the
potential to be a <tt class="xref py py-class docutils literal">unicode</tt>-only API. However, there&#8217;s two things that
you should consider when designing a <tt class="xref py py-class docutils literal">unicode</tt>-only API:
<ol class="arabic simple">
<li>As your API gains popularity, people are going to use your API in places
that you may not have thought of. Corner cases in these other places may
mean that processing bytes is desirable.</li>
<li>In python2, byte <tt class="xref py py-class docutils literal">str</tt> and <tt class="xref py py-class docutils literal">unicode</tt> are often used
interchangably with each other. That means that people programming against
your API may have received <tt class="xref py py-class docutils literal">str</tt> from some other API and it would be
most convenient for their code if your API accepted it.</li>
</ol>
<div class="admonition note">
Note
In python3, the separation between the text type and the byte type
are more clear. So in python3, there&#8217;s less need to have all APIs take
both unicode and bytes.
</div>
</div>
<div class="section" id="can-you-restrict-the-encodings">
<h3><a class="toc-backref" href="#id14">Can you restrict the encodings?</a><a class="headerlink" href="#can-you-restrict-the-encodings" title="Permalink to this headline">¶</a></h3>
If you determine that you have to deal with byte <tt class="xref py py-class docutils literal">str</tt> you should
realize that not all encodings are created equal. Each has different
properties that may make it possible to provide a simpler API provided that
you can reasonably tell the users of your API that they cannot use certain
classes of encodings.
As one example, if you are required to find a comma (<tt class="docutils literal">,</tt>) in a byte
<tt class="xref py py-class docutils literal">str</tt> you have different choices based on what encodings are allowed.
If you can reasonably restrict your API users to only giving <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII
compatible</a> encodings you can do this simply by searching for the literal
comma character because that character will be represented by the same byte
sequence in all <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII compatible</a> encodings.
The following are some classes of encodings to be aware of as you decide how
generic your code needs to be.
<div class="section" id="single-byte-encodings">
<h4><a class="toc-backref" href="#id15">Single byte encodings</a><a class="headerlink" href="#single-byte-encodings" title="Permalink to this headline">¶</a></h4>
Single byte encodings can only represent 256 total characters. They encode
the <a class="reference internal" href="glossary.html#term-code-points">code points</a> for a character to the equivalent number in a single
byte.
Most single byte encodings are <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII compatible</a>. <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII
compatible</a> encodings are the most likely to be usable without changes to code
so this is good news. A notable exception to this is the <a class="reference external" href="http://en.wikipedia.org/wiki/Extended_Binary_Coded_Decimal_Interchange_Code">EBDIC</a>
family of encodings.
</div>
<div class="section" id="multibyte-encodings">
<h4><a class="toc-backref" href="#id16">Multibyte encodings</a><a class="headerlink" href="#multibyte-encodings" title="Permalink to this headline">¶</a></h4>
Multibyte encodings use more than one byte to encode some characters.
<div class="section" id="fixed-width">
<h5><a class="toc-backref" href="#id17">Fixed width</a><a class="headerlink" href="#fixed-width" title="Permalink to this headline">¶</a></h5>
Fixed width encodings have a set number of bytes to represent all of the
characters in the character set. <tt class="docutils literal">UTF-32</tt> is an example of a fixed width
encoding that uses four bytes per character and can express every unicode
characters. There are a number of problems with writing APIs that need to
operate on fixed width, multibyte characters. To go back to our earlier
example of finding a comma in a string, we have to realize that even in
<tt class="docutils literal">UTF-32</tt> where the <a class="reference internal" href="glossary.html#term-code-point">code point</a> for <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> characters is the
same as in <a class="reference internal" href="glossary.html#term-ascii">ASCII</a>, the byte sequence for them is different. So you
cannot search for the literal byte character as it may pick up false
positives and may break a byte sequence in an odd place.
</div>
<div class="section" id="variable-width">
<h5><a class="toc-backref" href="#id18">Variable Width</a><a class="headerlink" href="#variable-width" title="Permalink to this headline">¶</a></h5>
<div class="section" id="ascii-compatible">
<h6><a class="toc-backref" href="#id19">ASCII compatible</a><a class="headerlink" href="#ascii-compatible" title="Permalink to this headline">¶</a></h6>
<a class="reference internal" href="glossary.html#term-utf-8">UTF-8</a> and the <a class="reference external" href="http://en.wikipedia.org/wiki/Extended_Unix_Code">EUC</a>
family of encodings are examples of <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII compatible</a> multi-byte
encodings. They achieve this by adhering to two principles:
<ul class="simple">
<li>All of the <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> characters are represented by the byte that they
are in the <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> encoding.</li>
<li>None of the <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> byte sequences are reused in any other byte
sequence for a different character.</li>
</ul>
</div>
<div class="section" id="escaped">
<h6><a class="toc-backref" href="#id20">Escaped</a><a class="headerlink" href="#escaped" title="Permalink to this headline">¶</a></h6>
Some multibyte encodings work by using only bytes from the <a class="reference internal" href="glossary.html#term-ascii">ASCII</a>
encoding but when a particular sequence of those byes is found, they are
interpreted as meaning something other than their <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> values.
<tt class="docutils literal">UTF-7</tt> is one such encoding that can encode all of the unicode
<a class="reference internal" href="glossary.html#term-code-points">code points</a>. For instance, here&#8217;s a some Japanese characters encoded as
<tt class="docutils literal">UTF-7</tt>:
<div class="highlight-python"><div class="highlight"><pre>&gt;&gt;&gt; a = u&#39;\u304f\u3089\u3068\u307f&#39;
&gt;&gt;&gt; print a
くらとみ
&gt;&gt;&gt; print a.encode(&#39;utf-7&#39;)
+ME8wiTBoMH8-
</pre></div>
</div>
These encodings can be used when you need to encode unicode data that may
contain non-<a class="reference internal" href="glossary.html#term-ascii">ASCII</a> characters for inclusion in an <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> only
transport medium or file.
However, they are not <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII compatible</a> in the sense that we used
earlier as the bytes that represent a <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> character are being reused
as part of other characters. If you were to search for a literal plus sign in
this encoded string, you would run across many false positives, for instance.
</div>
<div class="section" id="other">
<h6><a class="toc-backref" href="#id21">Other</a><a class="headerlink" href="#other" title="Permalink to this headline">¶</a></h6>
There are many other popular variable width encodings, for instance <tt class="docutils literal">UTF-16</tt>
and <tt class="docutils literal">shift-JIS</tt>. Many of these are not <a class="reference internal" href="glossary.html#term-ascii-compatible">ASCII compatible</a> so you
cannot search for a literal <a class="reference internal" href="glossary.html#term-ascii">ASCII</a> character without danger of false
positives or false negatives.
</div>
</div>
</div>
</div>
</div>
</div>

</div>
 </div>
 </div>
 <div class="sphinxsidebar">
 <div class="sphinxsidebarwrapper">
 <h3><a href="index.html">Table Of Contents</a></h3>
 <ul>
<li><a class="reference internal" href="#">Designing Unicode Aware APIs</a><ul>
<li><a class="reference internal" href="#take-either-bytes-or-unicode-output-only-unicode">Take either bytes or unicode, output only unicode</a></li>
<li><a class="reference internal" href="#take-either-bytes-or-unicode-output-the-same-type">Take either bytes or unicode, output the same type</a></li>
<li><a class="reference internal" href="#separate-functions">Separate functions</a></li>
<li><a class="reference internal" href="#deciding-whether-to-take-str-or-unicode-when-no-value-is-returned">Deciding whether to take str or unicode when no value is returned</a><ul>
<li><a class="reference internal" href="#writing-to-external-data">Writing to external data</a></li>
<li><a class="reference internal" href="#updating-data-structures">Updating data structures</a></li>
</ul>
</li>
<li><a class="reference internal" href="#apis-to-avoid">APIs to Avoid</a><ul>
<li><a class="reference internal" href="#returning-unicode-unless-a-conversion-fails">Returning unicode unless a conversion fails</a></li>
<li><a class="reference internal" href="#ignoring-values-with-no-chance-of-recovery">Ignoring values with no chance of recovery</a></li>
<li><a class="reference internal" href="#raising-a-unicodeexception-with-no-chance-of-recovery">Raising a UnicodeException with no chance of recovery</a></li>
</ul>
</li>
<li><a class="reference internal" href="#knowing-your-data">Knowing your data</a><ul>
<li><a class="reference internal" href="#do-you-need-to-operate-on-both-bytes-and-unicode">Do you need to operate on both bytes and unicode?</a></li>
<li><a class="reference internal" href="#can-you-restrict-the-encodings">Can you restrict the encodings?</a><ul>
<li><a class="reference internal" href="#single-byte-encodings">Single byte encodings</a></li>
<li><a class="reference internal" href="#multibyte-encodings">Multibyte encodings</a><ul>
<li><a class="reference internal" href="#fixed-width">Fixed width</a></li>
<li><a class="reference internal" href="#variable-width">Variable Width</a><ul>
<li><a class="reference internal" href="#ascii-compatible">ASCII compatible</a></li>
<li><a class="reference internal" href="#escaped">Escaped</a></li>
<li><a class="reference internal" href="#other">Other</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>

<h4>Previous topic</h4>
 <a href="unicode-frustrations.html"
 title="previous chapter">Overcoming frustration: Correctly using unicode in python2</a>
 <h4>Next topic</h4>
 <a href="api-overview.html"
 title="next chapter">Kitchen API</a>
 <h3>This Page</h3>
 <ul class="this-page-menu">
 <li><a href="_sources/designing-unicode-apis.txt"
 rel="nofollow">Show Source</a></li>
 </ul>
<div id="searchbox" style="display: none">
 <h3>Quick search</h3>
 <form class="search" action="search.html" method="get">
 <input type="text" name="q" />
 <input type="submit" value="Go" />
 <input type="hidden" name="check_keywords" value="yes" />
 <input type="hidden" name="area" value="default" />
 </form>
 
 Enter search terms or a module, class or function name.
 
</div>
<script type="text/javascript">$('#searchbox').show(0);</script>
 </div>
 </div>
 <div class="clearer"></div>
 </div>
 <div class="related">
 <h3>Navigation</h3>
 <ul>
 <li class="right" style="margin-right: 10px">
 <a href="genindex.html" title="General Index"
 >index</a></li>
 <li class="right" >
 <a href="py-modindex.html" title="Python Module Index"
 >modules</a> |</li>
 <li class="right" >
 <a href="api-overview.html" title="Kitchen API"
 >next</a> |</li>
 <li class="right" >
 <a href="unicode-frustrations.html" title="Overcoming frustration: Correctly using unicode in python2"
 >previous</a> |</li>
 <li><a href="index.html">kitchen 1.1.1 documentation</a> &raquo;</li>
 <li><a href="tutorial.html" >Using kitchen to write good code</a> &raquo;</li> 
 </ul>
 </div>
 <div class="footer">
 &copy; Copyright 2011 Red Hat, Inc. and others.
 Created using <a href="http://sphinx.pocoo.org/">Sphinx</a> 1.1.3.
 </div>
 </body>
</html>