Add nested Avro record schema inference test.

Also sort inferred fields by name and add Python 3.5 to Travis matrix.
mtth · Aug 27, 2016 · 4f08ae8 · 4f08ae8
1 parent 3a06dde
commit 4f08ae8
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 6 deletions.
diff --git a/.travis.yml b/.travis.yml
@@ -4,6 +4,7 @@ python:
   - 2.7
   - 3.3
   - 3.4
+  - 3.5
 install:
   - pip install . fastavro coverage mock
 before_script:

diff --git a/hdfs/ext/avro/__init__.py b/hdfs/ext/avro/__init__.py
@@ -83,7 +83,7 @@ def infer(self, obj):
         'type': 'record',
         'fields': [
           {'name': k, 'type': self.infer(v)}
-          for k, v in obj.items()
+          for k, v in sorted(obj.items()) # Sort fields by name.
         ]
       }
     raise ValueError('Cannot infer type from %s: %r' % (type(obj), obj))

diff --git a/test/test_ext_avro.py b/test/test_ext_avro.py
@@ -54,28 +54,55 @@ def test_buffered_read(self):
 
 class TestInferSchema(object):
 
+  def test_array(self):
+    eq_(
+      _SchemaInferrer().infer({'foo': 1, 'bar': ['hello']}),
+      {
+        'type': 'record',
+        'name': '__Record1',
+        'fields': [
+          {'type': {'type': 'array', 'items': 'string'}, 'name': 'bar'},
+          {'type': 'int', 'name': 'foo'},
+        ]
+      }
+    )
+
   def test_flat_record(self):
     eq_(
       _SchemaInferrer().infer({'foo': 1, 'bar': 'hello'}),
       {
         'type': 'record',
         'name': '__Record1',
         'fields': [
-          {'type': 'int', 'name': 'foo'},
           {'type': 'string', 'name': 'bar'},
+          {'type': 'int', 'name': 'foo'},
         ]
       }
     )
 
-  def test_array(self):
+  def test_nested_record(self):
     eq_(
-      _SchemaInferrer().infer({'foo': 1, 'bar': ['hello']}),
+      _SchemaInferrer().infer({'foo': {'bax': 2}, 'bar': {'baz': 3}}),
       {
         'type': 'record',
         'name': '__Record1',
         'fields': [
-          {'type': 'int', 'name': 'foo'},
-          {'type': {'type': 'array', 'items': 'string'}, 'name': 'bar'},
+          {
+            'type': {
+              'type': 'record',
+              'name': '__Record2',
+              'fields': [{'type': 'int', 'name': 'baz'}]
+            },
+            'name': 'bar',
+          },
+          {
+            'type': {
+              'type': 'record',
+              'name': '__Record3',
+              'fields': [{'type': 'int', 'name': 'bax'}]
+            },
+            'name': 'foo',
+          },
         ]
       }
     )