From 548b1376c1aa65aa20bc203d83c21b28ddb1083c Mon Sep 17 00:00:00 2001
From: Andreas Motl <andreas.motl@panodata.org>
Date: Wed, 9 Oct 2024 07:33:20 +0200
Subject: [PATCH] MongoDB: Fixed BSON decoding of `{"$date": 1180690093000}`
 timestamps

---
 src/commons_codec/transform/mongodb.py         | 18 ++++++++++--------
 .../transform/mongodb/test_mongodb_convert.py  | 10 ++++++++--
 2 files changed, 18 insertions(+), 10 deletions(-)

diff --git a/src/commons_codec/transform/mongodb.py b/src/commons_codec/transform/mongodb.py
index 744ed54..0220033 100644
--- a/src/commons_codec/transform/mongodb.py
+++ b/src/commons_codec/transform/mongodb.py
@@ -121,15 +121,17 @@ def decode_extended_json(self, value: t.Dict[str, t.Any]) -> t.Any:
 
         # Special handling for datetime representation in NUMBERLONG format (emulated depth-first).
         type_ = next(iter(value))  # Get key of first item in dictionary.
-        is_date_numberlong = type_ == "$date" and "$numberLong" in value["$date"]
-        if is_date_numberlong:
-            try:
-                out = dt.datetime.fromtimestamp(int(value["$date"]["$numberLong"]) / 1000, tz=dt.timezone.utc)
-            except ValueError as ex:
-                logger.error(f"Decoding legacy timestamp failed: {ex}. value={value}")
-                out = 0
-        else:
+        if type_ == "$date" and isinstance(value["$date"], dict):
+            value = {"$date": int(value["$date"]["$numberLong"])}
+
+        # Invoke BSON decoder.
+        try:
             out = object_hook(value)
+        except bson.errors.InvalidBSON as ex:
+            logger.error(f"Decoding BSON value failed: {ex}. value={value}")
+            out = None
+            if "Python int too large to convert to C int" in str(ex):
+                out = 0
 
         is_bson = isinstance(out, all_bson_types())
 
diff --git a/tests/transform/mongodb/test_mongodb_convert.py b/tests/transform/mongodb/test_mongodb_convert.py
index a3e5afa..9e6b298 100644
--- a/tests/transform/mongodb/test_mongodb_convert.py
+++ b/tests/transform/mongodb/test_mongodb_convert.py
@@ -104,7 +104,7 @@ class DateConversionCase:
     DateConversionCase(
         converter=MongoDBCrateDBConverter(),
         data_in={"$date": {"$numberLong": "1655210544987"}},
-        data_out=dt.datetime(2022, 6, 14, 12, 42, 24, 987000, tzinfo=dt.timezone.utc),
+        data_out=dt.datetime(2022, 6, 14, 12, 42, 24, 987000),
     ),
     DateConversionCase(
         converter=MongoDBCrateDBConverter(timestamp_to_epoch=True, timestamp_use_milliseconds=True),
@@ -124,7 +124,12 @@ class DateConversionCase:
     DateConversionCase(
         converter=MongoDBCrateDBConverter(timestamp_to_iso8601=True),
         data_in={"$date": {"$numberLong": "1655210544987"}},
-        data_out="2022-06-14T12:42:24.987000+00:00",
+        data_out="2022-06-14T12:42:24.987000",
+    ),
+    DateConversionCase(
+        converter=MongoDBCrateDBConverter(timestamp_to_iso8601=True),
+        data_in={"$date": 1180690093000},
+        data_out="2007-06-01T09:28:13",
     ),
 ]
 
@@ -136,6 +141,7 @@ class DateConversionCase:
     "epochms-$date-legacy",
     "iso8601-$date-canonical",
     "iso8601-$date-legacy",
+    "iso8601-$date-ultra-legacy",
 ]