dreid / edn Goto Github PK

At the moment, there's no Character type, which means that characters parsed by edn cannot be reliably unparsed as characters, as there's no way to distinguish a single-element string from a character in Python.

I don't know if this is such a bad thing.

I think it would be fair if I were listed as a joint copyright holder.

No accompanying patch, because I don't want to publish a version which has incorrect LICENSE information.

We currently have a very basic formatter / printer / unparser (we really must settle on a term), but no pretty printer. Everyone loves a pretty printer. We should have one.

Not 100% sure this is a good idea, but here we go.

The natural way to use maps in Clojure is to use keywords for the keys. In Python, the natural way is to use strings for keys. When writing Python code that works with Clojure code via edn, it becomes tedious to make dicts with Keywords as keys.

So, perhaps it's worth adding a convenience class which is a dict where all of the keys are keywords, but can be accessed & updated using string syntax, but are stored internally as keywords.

These examples are meant to be illustrative, not prescriptive.

>>> x = keyworddict()
>>> x.with_pair('foo', 'bar')
>>> repr(x)
{Keyword(Symbol('foo')): 'bar'}
>>> x['foo']
'bar'
>>> dumps(x)
{:foo "bar"}
>>> x.with_pair(42, 'qux')
Traceback (most recent call last):
  ...
ValueError: Invalid keyword: 42

And also,

>>> x = keyworddict(prefix="myproject")
>>> x.with_pair('foo', 'bar')
>>> repr(x)
{Keyword(Symbol('foo', 'myproject')): 'bar'}
>>> x['foo']
'bar'
>>> dumps(x)
{:myproject/foo "bar"}

recording for posterity.

...............E.........................................................................................
======================================================================
ERROR: test_tag (edn.tests.test_ast.EDNTestCase)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Users/callen/code/edn/edn/tests/test_ast.py", line 179, in test_tag
    ('#foo {}', TaggedValue(Symbol('foo'), frozendict({})))
  File "/Users/callen/.virtualenv/sweden/lib/python2.7/site-packages/Parsley-1.2-py2.7.egg/terml/nodes.py", line 144, in mkterm
    tuple([coerceToTerm(a) for a in args]),
  File "/Users/callen/.virtualenv/sweden/lib/python2.7/site-packages/Parsley-1.2-py2.7.egg/terml/nodes.py", line 138, in coerceToTerm
    raise ValueError("Could not coerce %r to Term" % (val,))
ValueError: Could not coerce frozendict({}) to Term

----------------------------------------------------------------------
Ran 105 tests in 0.181s

FAILED (errors=1)

If I get a workable fix together, I'll PR it with my test(s).

Currently there's no support for loading edn data that contains multiple elements.

Relatedly, there's no support for taking a stream as input and parsing it incrementally.