Пример #1
0
 def test_constructor(self):
     # component id not string nor HeronComponentSpec
     with self.assertRaises(TypeError):
         GlobalStreamId(componentId=123, streamId="default")
     # stream id not string
     with self.assertRaises(TypeError):
         GlobalStreamId(componentId="component", streamId=12345)
Пример #2
0
    def test_get_item(self):
        # HeronComponentSpec name set
        spec = HeronComponentSpec("spout", "class", True, 1)
        spec.outputs = [
            "string", "hello",
            Stream(fields=["abc", "def"], name="another_stream"),
            Stream(fields=["another", "default"], name="default")
        ]
        ret = spec['another_stream']
        self.assertEqual(ret, GlobalStreamId("spout", "another_stream"))

        # HeronComponentSpec name not set
        spec = HeronComponentSpec(None, "class", True, 1)
        spec.outputs = [
            "string", "hello",
            Stream(fields=["abc", "def"], name="another_stream"),
            Stream(fields=["another", "default"], name="default")
        ]
        ret = spec['default']
        self.assertEqual(ret, GlobalStreamId(spec, "default"))

        # stream id not registered
        spec = HeronComponentSpec(None, "class", True, 1)
        spec.outputs = [
            "string", "hello",
            Stream(fields=["abc", "def"], name="another_stream"),
            Stream(fields=["another", "default"], name="default")
        ]
        with self.assertRaises(ValueError):
            spec['non_existent_stream']
Пример #3
0
    def test_component_id_property(self):
        # component id is string
        gsi = GlobalStreamId(componentId="component", streamId="stream")
        self.assertEqual(gsi.component_id, "component")

        # component id is HeronComponentSpec with name
        spec = HeronComponentSpec("spout", "class", True, 1)
        gsi = GlobalStreamId(spec, "stream")
        self.assertEqual(gsi.component_id, "spout")

        # component id is HeronComponentSpec without name
        spec = HeronComponentSpec(None, "class", True, 1)
        gsi = GlobalStreamId(spec, "stream")
        # expecting "<No name available for HeronComponentSpec yet, uuid: %s>"
        self.assertIn(spec.uuid, gsi.component_id)
Пример #4
0
 def _calculate_inputs(self):
     return {
         GlobalStreamId(self._parents[0]._stage_name, self._parents[0]._output):
         Grouping.SHUFFLE
     }
Пример #5
0
 def _calculate_inputs(self):
   return {GlobalStreamId(self._parents[0]._stage_name, self._parents[0]._output) :
           Grouping.custom("heron.dsl.src.python.reducebykeyandwindowbolt.ReduceGrouping")}
Пример #6
0
 def _calculate_inputs(self):
     inputs = {}
     for parent in self._parents:
         inputs[GlobalStreamId(parent._stage_name, parent._output)] = \
                Grouping.custom("heron.dsl.src.python.joinbolt.JoinGrouping")
     return inputs
Пример #7
0
    def test_sanitize_inputs(self):
        # Note that _sanitize_inputs() should only be called after HeronComponentSpec's
        # name attribute is set

        # invalid inputs given as argument (valid ones are either dict, list, tuple or None)
        invalid_spec = HeronComponentSpec("name",
                                          "classpath",
                                          True,
                                          1,
                                          inputs="string")
        with self.assertRaises(TypeError):
            invalid_spec._sanitize_inputs()

        invalid_spec = HeronComponentSpec("name",
                                          "classpath",
                                          True,
                                          1,
                                          inputs=100)
        with self.assertRaises(TypeError):
            invalid_spec._sanitize_inputs()

        # dict <HeronComponentSpec -> Grouping>
        from_spec = HeronComponentSpec("spout", "sp_clspath", True, 1)
        to_spec = HeronComponentSpec("bolt",
                                     "bl_clspath",
                                     False,
                                     1,
                                     inputs={from_spec: Grouping.SHUFFLE})
        ret = to_spec._sanitize_inputs()
        self.assertEqual(
            ret, {GlobalStreamId("spout", "default"): Grouping.SHUFFLE})

        from_spec = HeronComponentSpec("spout", "sp_clspath", True, 1)
        from_spec.outputs = [Stream(name='another_stream')]
        to_spec = HeronComponentSpec(
            "bolt",
            "bl_clspath",
            False,
            1,
            inputs={from_spec['another_stream']: Grouping.ALL})
        ret = to_spec._sanitize_inputs()
        self.assertEqual(
            ret, {GlobalStreamId("spout", "another_stream"): Grouping.ALL})

        # HeronComponentSpec's name attribute not set
        from_spec = HeronComponentSpec(None, "sp_clspath", True, 1)
        to_spec = HeronComponentSpec("bolt",
                                     "bl_clspath",
                                     False,
                                     1,
                                     inputs={from_spec: Grouping.ALL})
        with self.assertRaises(RuntimeError):
            to_spec._sanitize_inputs()

        # dict <GlobalStreamId -> Grouping>
        inputs_dict = {
            GlobalStreamId("some_spout", "some_stream"):
            Grouping.NONE,
            GlobalStreamId("another_spout", "default"):
            Grouping.fields(['word', 'count'])
        }
        spec = HeronComponentSpec("bolt",
                                  "classpath",
                                  False,
                                  1,
                                  inputs=inputs_dict)
        ret = spec._sanitize_inputs()
        self.assertEqual(ret, inputs_dict)

        # list of HeronComponentSpec
        from_spec1 = HeronComponentSpec("spout1", "sp1_cls", True, 1)
        from_spec2 = HeronComponentSpec("spout2", "sp2_cls", True, 1)
        to_spec = HeronComponentSpec("bolt",
                                     "bl_cls",
                                     False,
                                     1,
                                     inputs=[from_spec1, from_spec2])
        ret = to_spec._sanitize_inputs()
        self.assertEqual(
            ret, {
                GlobalStreamId("spout1", "default"): Grouping.SHUFFLE,
                GlobalStreamId("spout2", "default"): Grouping.SHUFFLE
            })

        # HeronComponentSpec's name attribute not set
        from_spec = HeronComponentSpec(None, "sp_clspath", True, 1)
        to_spec = HeronComponentSpec("bolt",
                                     "bl_clspath",
                                     False,
                                     1,
                                     inputs=[from_spec])
        with self.assertRaises(RuntimeError):
            to_spec._sanitize_inputs()

        # list of GlobalStreamId
        inputs_list = [
            GlobalStreamId("spout1", "default"),
            GlobalStreamId("spout2", "some_stream")
        ]
        spec = HeronComponentSpec("bolt",
                                  "bl_cls",
                                  False,
                                  1,
                                  inputs=inputs_list)
        ret = spec._sanitize_inputs()
        self.assertEqual(ret, dict(zip(inputs_list, [Grouping.SHUFFLE] * 2)))

        # list of neither GlobalStreamId nor HeronComponentSpec
        inputs_list = [None, 123, "string", [GlobalStreamId("sp", "default")]]
        spec = HeronComponentSpec("bolt",
                                  "bl_cls",
                                  False,
                                  1,
                                  inputs=inputs_list)
        with self.assertRaises(ValueError):
            spec._sanitize_inputs()