autorepr: makes civilized string representations

Now with Python 3 support!

Overview

Python makes classes easy, but __repr__ methods hard. Did you forget to reference self again? Probably. Did have you thought to yourself "eh, this class is real simple, it doesn't need a repr"? Without a doubt. Was production taken down three times last week because your __str__ returned unicode? ... no? Maybe that's just me.

autorepr makes it simple to build expressive, safe, and correct, __repr__, __str__, __unicode__, and __bytes__ methods in a single line each.

With autorepr, you get the repers you want, without worrying about the fiddly bits (like encoding and decoding), leaving you to focus on your project:

>>> from autorepr import autorepr, autotext
>>> class Person(object):
...     name = u"Alex ☃"
...     height = 123.456
...
...     __repr__ = autorepr(["name", "height:0.1f"])
...     __str__, __unicode__ = autotext("{self.name} ({self.height:0.0f} cm)")
...
>>> p = Person()
>>> repr(p)
"<__main__.Person name=u'Alex \\u2603' height=123.5 at 0x...>"
>>> unicode(p)
u'Alex \u2603 (123 cm)'
>>> str(p)
'Alex \xe2\x98\x83 (123 cm)'

Installation

$ pip install autorepr

Usage

autorepr exposes two main functions:

• autorepr, which builds a Python-esque __repr__ string by passing either a str.format-style string, or a list of attributes which should be included in a name=value list:

  autorepr(["name", "height:0.1f"]) -->
      "<pkg.Person name=u'Alex \u2603' height=123.5 at 0x...>"
  autorepr("{self.id} name={self.name!r}") -->
      "<pkg.Person 123 name=u'Alex \u2603' at 0x...>"
  

• autotext, which uses autostr and autounicode to create __str__ and __unicode__ methods in a Python 2 + 3 friendly way:

  __str__, __unicode__ = autotext("{self.name} ({self.height!d} cm)") -->
      str: 'Alex \xe2\x98\x83 (123cm)'
      unicode: u'Alex \u2603 (123cm)'
  

And three secondary functions - autostr, autounicode, and autobytes - which build __str__, __unicode__, and __bytes__ functions, respectively. The functions will do their best to avoid Unicode encoding / decoding errors, and will generally Do The Right Thing, even if the inputs aren't necessarily sensible.

Note: the examples shown here are Python 2, but everything works equally well under Python 3.

>>> from autorepr import autorepr, autotext, autostr, autounicode
>>> class Person(object):
...     name = u"Alex ☃"
...     height = 123.456
...
...     __repr__ = autorepr(["name", "height:0.1f"])
...     __str__, __unicode__ = autotext("{self.name} ({self.height:0.0f} cm)")
...
>>> p = Person()
>>> repr(p)
"<__main__.Person name=u'Alex \\u2603' height=123.5 at 0x...>"
>>> unicode(p)
u'Alex \u2603 (123 cm)'
>>> str(p)
'Alex \xe2\x98\x83 (123 cm)'

Notice that autostr and autorepr (as called here through autotext) are intelligent about converting to/from unicode (decoding/encoding as UTF-8) as necessary:

>>> p.name = u"unicode: ☃"
>>> unicode(p)
u'unicode: \u2603 (123 cm)'
>>> str(p)
'unicode: \xe2\x98\x83 (123 cm)'
>>> p.name = 'utf-8 bytes: \xe2\x98\x83'
>>> unicode(p)
u'utf-8 bytes: \u2603 (123 cm)'
>>> str(p)
'utf-8 bytes: \xe2\x98\x83 (123 cm)'

Note: autostr and autorepr won't crash on invalid UTF-8 (for example, if autounicode is asked to turn binary data into unicode), but the result is undefined and may not be desirable.

Additional properties can be passed in as kwargs, which will be called with the instance as a parameter:

>>> name_with_len = autostr("{self.name} length={len}",
...                         len=lambda self: len(self.name))
...
>>> p.name = 'Alex'
>>> name_with_len(p)
'Alex length=4'

This works with autorepr's list mode too:

>>> repr_with_len = autorepr(["name", "len"],
...                          len=lambda self: len(self.name))
...
>>> repr_with_len(p)
"<__main__.Person name='Alex' len=4 at 0x...>"

If a regular format string is passed to autorepr, it will use that instead of the generated string:

>>> repr_with_str = autorepr("{self.name!r}")
>>> repr_with_str(p)
"<__main__.Person 'Alex' at 0x...>"

And of course, if you don't want your __repr__ to be wrapped in <ClassName ...>, you can use autostr:

>>> repr_with_autostr = autostr("Person({self.name!r})")
>>> repr_with_autostr(p)
"Person('Alex')"

Format specifications can also be passed to autorepr if the default of !r is undesirable (for example, truncating floats):

>>> with_fmt_spec = autorepr(["duration:0.1f", "addr:x", "type!s"],
...                          duration=lambda x: 123.456,
...                          addr=lambda x: 0xabc123,
...                          type=lambda x: "foo")
>>> with_fmt_spec(None)
'<....NoneType duration=123.5 addr=abc123 type=foo at 0x...>'