From 310902d2212aab694743f5635cb6c63344d31ccb Mon Sep 17 00:00:00 2001
From: Isaac Na <isaac219@gmail.com>
Date: Tue, 21 Dec 2021 01:46:18 -0500
Subject: [PATCH] bugfix/use-consistent-datetime-in-hash-generation (#144)

* Link to test-deployment repo

* Link to test deployment UI

* Fix including print statements

* Refresh var

* Remove print statements

* Add unit tests

* Fix lint errors

* Remove extra line

* Remove redundant var

* Top level conditional branches

* Add datetime branch to top level
---
 cdp_backend/database/functions.py            | 20 ++++++++++++++------
 cdp_backend/tests/database/test_functions.py | 19 ++++++++++++++++++-
 2 files changed, 32 insertions(+), 7 deletions(-)

diff --git a/cdp_backend/database/functions.py b/cdp_backend/database/functions.py
index bdb6158f..15f655bb 100644
--- a/cdp_backend/database/functions.py
+++ b/cdp_backend/database/functions.py
@@ -43,22 +43,30 @@ def generate_and_attach_doc_hash_as_id(db_model: Model) -> Model:
     for pk in db_model._PRIMARY_KEYS:
         field = getattr(db_model, pk)
 
+        # Load the document for reference fields and add id to hasher
+        if isinstance(field, ReferenceDocLoader):
+            hasher.update(pickle.dumps(field.get().id, protocol=4))
+
         # Handle reference fields by using their doc path
-        if isinstance(field, Model):
+        elif isinstance(field, Model):
             # Ensure that the underlying model has an id
             # In place update to db_model for this field
             setattr(db_model, pk, generate_and_attach_doc_hash_as_id(field))
 
+            # Update variable after setattr
+            field = getattr(db_model, pk)
+
             # Now attach the generated hash document path
             hasher.update(pickle.dumps(field.id, protocol=4))
 
+        # If datetime, hash with epoch millis to avoid timezone issues
+        elif isinstance(field, datetime):
+            field = field.timestamp()
+            hasher.update(pickle.dumps(field, protocol=4))
+
         # Otherwise just simply add the primary key value
         else:
-            # Load the document for reference fields and add id to hasher
-            if isinstance(field, ReferenceDocLoader):
-                hasher.update(pickle.dumps(field.get().id, protocol=4))
-            else:
-                hasher.update(pickle.dumps(field, protocol=4))
+            hasher.update(pickle.dumps(field, protocol=4))
 
     # Set the id to the first twelve characters of hexdigest
     db_model.id = hasher.hexdigest()[:12]
diff --git a/cdp_backend/tests/database/test_functions.py b/cdp_backend/tests/database/test_functions.py
index a240aee9..cea92930 100644
--- a/cdp_backend/tests/database/test_functions.py
+++ b/cdp_backend/tests/database/test_functions.py
@@ -1,14 +1,17 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
+from datetime import datetime, timezone
+
 import pytest
+import pytz
 from fireo.models import Model
 
 from cdp_backend.database import functions as db_functions
 from cdp_backend.database import models as db_models
 
 ###############################################################################
-# Tests
+# Test constants
 
 body_a = db_models.Body()
 body_a.name = "Body A"
@@ -16,6 +19,17 @@
 body_b = db_models.Body()
 body_b.name = "Body B"
 
+a_datetime = datetime.fromisoformat("2021-08-16 07:42:10.318957+00:00")
+event_a = db_models.Event.Example()
+event_a.event_datetime = a_datetime
+
+event_b = db_models.Event.Example()
+local_tz = pytz.timezone("Europe/Moscow")
+modified_dt = event_a.event_datetime.replace(tzinfo=timezone.utc).astimezone(
+    tz=local_tz
+)
+event_b.event_datetime = local_tz.normalize(modified_dt)
+
 
 @pytest.mark.parametrize(
     "model, expected_id",
@@ -27,6 +41,9 @@
         # Testing models differ
         (body_a, "0a8a8e139258"),
         (body_b, "1535fef479ff"),
+        # Testing timezone difference
+        (event_a, "6291946d4094"),
+        (event_b, "6291946d4094"),
     ],
 )
 def test_generate_and_attach_doc_hash_as_id(model: Model, expected_id: str) -> None: