Adding Fix and Unit tests for UTF-8 Breakages

snowplow_tracker/__init__.py

@@ -1,6 +1,11 @@
 from snowplow_tracker._version import __version__
 from snowplow_tracker.subject import Subject
-from snowplow_tracker.emitters import logger, Emitter, AsyncEmitter, CeleryEmitter, RedisEmitter
+from snowplow_tracker.emitters import Emitter, AsyncEmitter, CeleryEmitter, RedisEmitter
 from snowplow_tracker.self_describing_json import SelfDescribingJson
 from snowplow_tracker.tracker import Tracker
 from contracts import disable_all as disable_contracts, enable_all as enable_contracts
+import logging
+
+# Set default logging handler to avoid "No handler found" warnings.
+logging.getLogger(__name__).addHandler(logging.NullHandler())
+

snowplow_tracker/emitters.py

@@ -39,7 +39,6 @@
 from snowplow_tracker.self_describing_json import SelfDescribingJson
 
 logger = logging.getLogger(__name__)
-logger.setLevel(logging.INFO)
 
 DEFAULT_MAX_LENGTH = 10
 PAYLOAD_DATA_SCHEMA = "iglu:com.snowplowanalytics.snowplow/payload_data/jsonschema/1-0-4"
@@ -52,16 +51,6 @@
 
 new_contract("redis", lambda x: isinstance(x, (redis.Redis, redis.StrictRedis)))
 
-try:
-    # Check whether a custom Celery configuration module named "snowplow_celery_config" exists
-    import snowplow_celery_config
-    app = Celery()
-    app.config_from_object(snowplow_celery_config)
-
-except ImportError:
-    # Otherwise configure Celery with default settings
-    app = Celery("Snowplow", broker="redis://guest@localhost//")
-
 
 class Emitter(object):
     """
@@ -172,7 +161,6 @@ def reached_limit(self):
         else:
             return self.bytes_queued >= self.byte_limit or len(self.buffer) >= self.buffer_size
 
-    @task(name="Flush")
     def flush(self):
         """
             Sends all events in the buffer to the collector.

snowplow_tracker/payload.py

@@ -78,15 +78,15 @@ def add_json(self, dict_, encode_base64, type_when_encoded, type_when_not_encode
         """
 
         if dict_ is not None and dict_ != {}:
-
             json_dict = json.dumps(dict_, ensure_ascii=False)
-
             if encode_base64:
-                encoded_dict = base64.urlsafe_b64encode(json_dict.encode("ascii"))
-                if not isinstance(encoded_dict, str):
+                try:
+                    encoded_dict = base64.urlsafe_b64encode(json_dict)
+                    if not isinstance(encoded_dict, str):
+                        encoded_dict = encoded_dict.decode("utf-8")
+                except UnicodeDecodeError:
                     encoded_dict = encoded_dict.decode("utf-8")
-                self.add(type_when_encoded, encoded_dict)                
-
+                self.add(type_when_encoded, encoded_dict)
             else:
                 self.add(type_when_not_encoded, json_dict)
 

snowplow_tracker/test/unit/test_payload.py

@@ -66,3 +66,21 @@ def test_add_dict(self):
         p.add_dict({"name4": 4, "name3": 3})            # Order doesn't matter
         output = {"n1": "v1", "n2": "v2", "name3": 3, "name4": 4}
         self.assertTrue(is_subset(output, p.nv_pairs))
+
+    def test_add_json_encode_base64(self):
+        p = payload.Payload()
+        p.add_json({"name5": 5, "name6": 6}, True, "Encoded_type", "Not_encoded_type")
+        output = {'Encoded_type': 'eyJuYW1lNiI6IDYsICJuYW1lNSI6IDV9'}
+        self.assertDictEqual(output, p.nv_pairs)
+
+    def test_add_json_not_encode_base64(self):
+        p = payload.Payload()
+        p.add_json({"name7": 7, "name8": 8}, False, "Encoded_type", "Not_encoded_type")
+        output = {'Not_encoded_type': '{"name7": 7, "name8": 8}'}
+        self.assertDictEqual(output, p.nv_pairs)
+
+    def test_add_json_with_emoji(self):
+        p = payload.Payload()
+        p.add_json({"name9": "\xF0\xe0\xe7\x82\xf1\U0001f44d", "name10": 10}, True, "Encoded_type", "Not_encoded_type")
+        output = {'Encoded_type': 'eyJuYW1lMTAiOiAxMCwgIm5hbWU5IjogIvDg54LxXFxVMDAwMWY0NGQifQ=='}
+        self.assertDictEqual(output, p.nv_pairs)

snowplow_tracker/tracker.py

@@ -53,7 +53,7 @@
 """
 
 
-class Tracker:
+class Tracker(object):
 
     new_contract("not_none", lambda s: s is not None)