Merge pull request oracc#80 from rillian/json

json serialization

pyoracc/atf/common/atffile.py

@@ -20,6 +20,8 @@
 import codecs
 import sys
 import logging
+import json
+from numbers import Number
 
 from pyoracc.atf.cdli.atflex import AtfCDLILexer
 from pyoracc.atf.cdli.atfyacc import AtfCDLIParser
@@ -72,6 +74,32 @@ def __str__(self):
     def serialize(self):
         return AtfFile.template.render_unicode(**vars(self))
 
+    def to_json(self, skip_empty=True, **kwargs):
+        '''Return a JSON representation of the parsed file.
+
+        The optional skip_empty argument determines whether keys
+        with empty values are included in the output. Set it to
+        False to see all possible object members.
+
+        Otherwise it accepts the same optional arguments as
+        json.dumps().'''
+        def _make_serializable(obj):
+            '''Construct a dict representation of an object.
+
+            This is necessary to handle our custom objects
+            which json.JSONEncoder doesn't know how to
+            serialize.'''
+
+            return {k: v
+                    for k, v in vars(obj).items()
+                    if not str(k).startswith('_') and not (
+                        skip_empty and not v and not isinstance(v, Number)
+                    )}
+
+        kwargs.setdefault('indent', 2)
+        kwargs.setdefault('default', _make_serializable)
+        return json.dumps(self.text, **kwargs)
+
 
 def check_atf(infile, atftype, verbose=False):
     content = codecs.open(infile,

pyoracc/test/atf/test_atffile.py

@@ -21,6 +21,7 @@
 from pyoracc.atf.common.atffile import AtfFile
 from ..fixtures import anzu, belsunu, sample_file
 import pytest
+import json
 
 
 def test_create():
@@ -108,3 +109,30 @@ def test_text_designation(name, code, description):
     afile = AtfFile(sample_file(name))
     assert afile.text.code == code
     assert afile.text.description == description
+
+
+# ATF filenames which fail the serialization tests.
+_xfail_texts = [
+    # Multilingual objects store the unmarked language
+    # under the `None` key in their `lines` dictionary,
+    # which is incompatible with `sort_keys=True`.
+    'bb_2_6',
+    ]
+
+
+@pytest.mark.parametrize('name', [
+    name if name not in _xfail_texts
+    else pytest.param(name, marks=[pytest.mark.xfail()])
+    for name in [text[0] for text in texts]])
+def test_json_serialization(name):
+    """
+    Parses ATF and verifies the to_json() method output.
+    """
+    afile = AtfFile(sample_file(name))
+    js = afile.to_json()
+    result = json.loads(js)
+    assert result
+    noskipjs = afile.to_json(skip_empty=False, sort_keys=True)
+    result = json.loads(noskipjs)
+    assert result
+    assert len(noskipjs) >= len(js)