haystack-ml-stack 0.4.8__tar.gz → 0.4.10__tar.gz

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haystack-ml-stack
-Version: 0.4.8
+Version: 0.4.10
 Summary: Functions related to Haystack ML
 Author-email: Oscar Vega <oscar@haystack.tv>
 License: MIT

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "haystack-ml-stack"
-version = "0.4.8"
+version = "0.4.10"
 description = "Functions related to Haystack ML"
 readme = "README.md"
 authors = [{ name = "Oscar Vega", email = "oscar@haystack.tv" }]

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/_kafka.py

@@ -72,7 +72,7 @@ async def send_channels_to_kafka(
         "userid": user.get("userid"),
         "client_os": user.get("clientOs"),
         "model_input": {"user": user, "channels": channels},
-        "model_output": model_output,
+        "model_output": {c: v for c, v in model_output.items() if c != "__meta__"},
         "model_name": state["model_name"].replace(".pkl", "")
         if state["model_name"]
         else None,
@@ -83,6 +83,7 @@ async def send_channels_to_kafka(
             "user_features": state.get("user_features", []),
             "global_features": state.get("global_features", []),
             "query_params": query_params,
+            **model_output.get("__meta__", {})
         },
         "processed_at": processed_at.isoformat(),
     }

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/_serializers.py

@@ -275,6 +275,8 @@ class UserPlaylistStatsSerializerV1(SimpleSerializer):
             category_msg.active_days = int(stats["active_days"])
             category_msg.total_watched = float(stats["total_watched"])
             category_msg.capped_total_watched = float(stats["capped_total_watched"])
+            category_msg.capped_total_log_watched = float(stats.get("capped_total_log_watched", 0.0))
+            category_msg.first_active_date = str(stats.get("first_active_date", ""))
         return root_msg
 
 
@@ -295,6 +297,7 @@ class GlobalChannelsSerializerV1(SimpleSerializer):
             channel_msg.name = str(channel["name"])
             channel_msg.category_group = str(channel["category_group"])
             channel_msg.start_date = int(channel["start_date"])
+            channel_msg.required_features.extend(str(f) for f in (channel.get("required_features") or []))
         return root_msg
 
 

haystack_ml_stack-0.4.10/src/haystack_ml_stack/_version.py

@@ -0,0 +1 @@
+__version__ = "0.4.10"

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/app.py

@@ -516,6 +516,7 @@ def create_channel_app(
             )
 
         model = state["model"]
+        exploration_mode = query_params.get("exploration") in ["True", "true", "1"]
         try:
             preprocess_start = time.perf_counter_ns()
             model["params"]["query_params"] = query_params
@@ -526,6 +527,16 @@ def create_channel_app(
             )
             predict_start = time.perf_counter_ns()
             model_output = model["predict"](model_input, model["params"])
+            if exploration_mode:
+                if "exploration" not in model:
+                    raise ValueError("Model does not support exploration mode")
+                model_output = model["exploration"](
+                    user,
+                    channels,
+                    model_output,
+                    model["params"],
+                )
+
             predict_end = time.perf_counter_ns()
         except Exception as e:
             logger.error("Model prediction failed: \n%s", traceback.format_exc())
@@ -547,7 +558,7 @@ def create_channel_app(
             "retrieval_ms": retrieval_meta.retrieval_ms,
             "preprocess_ms": (predict_start - preprocess_start) * 1e-6,
             "predict_ms": (predict_end - predict_start) * 1e-6,
-            "total_channels": len(model_output),
+            "total_channels": len(model_output) - ("__meta__" in model_output),
         }
         newrelic.agent.record_custom_event(
             "ChannelInference",

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/dynamo.py

@@ -499,31 +499,33 @@ def _process_channels(
     if playlist_stats_global:
         for stat in playlist_stats_global.values():
             all_channels.update(stat.data.keys())
-    if playlist_stats_user:
-        for stat in playlist_stats_user.values():
-            all_channels.update(stat.data.keys())
 
     all_channels = all_channels | set(DEFAULT_CHANNELS)
 
-    # My headlines will stay on top so not part of the candidates
+    # My headlines and local news will stay on top so not part of the candidates
+    # watch history and liked videos are always part of the bottom also
     # The group labels are also ignored, not real channels for UI
-    ignore_channels = set(["my headlines", "national_favorite", "local_favorite"])
+    ignore_channels = {"my headlines", "local news", "national_favorite", "local_favorite", "watch history", "liked videos", "nn_join_us"}
 
-    # Rule for weather
-    if user.get("geo", {}).get("country") != "US":
-        ignore_channels.add("weather")
-    # Rule for live_es
-    if "es" not in user.get("languages", []):
-        ignore_channels.add("live_es")
+    # Rules to filter candidates
+    channel_features = user.get("channelFeatures", {})
+    channel_eligibility = channel_features.get("eligibility", {})
+    channel_feature_predicates = channel_features.get("features", {})
 
     inserted_channel_names = set()
     if channel_candidates is not None:
         preferred = set(user.get("preferredChannels", []) or [])
         for ch in channel_candidates.data:
+            # Eligibility rules
+            if not channel_eligibility.get(ch.name, True):
+                ignore_channels.add(ch.name)
+            # Feature predicate rules for special channels
+            if ch.required_features and not all(channel_feature_predicates.get(f, False) for f in ch.required_features):
+                ignore_channels.add(ch.name)
             # Rule for authors
             if (
                 ch.category_group in ("national_favorite", "local_favorite")
-                and ch.name not in preferred
+                and (ch.name not in preferred or not channel_eligibility.get(ch.category_group, True))
             ):
                 ignore_channels.add(ch.name)
             # Rule for shows

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/generated/v1/features_pb2.py

@@ -24,7 +24,7 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x0e\x66\x65\x61tures.proto\x12\x1ahaystack_ml_stack.features\"7\n\x12\x45ntryContextCounts\x12\x10\n\x08\x61ttempts\x18\x01 \x01(\x05\x12\x0f\n\x07watched\x18\x02 \x01(\x05\"_\n\x0cSelectCounts\x12\x15\n\rtotal_selects\x18\x01 \x01(\x05\x12!\n\x19total_selects_and_watched\x18\x02 \x01(\x05\x12\x15\n\rtotal_browsed\x18\x03 \x01(\x05\"\xf3\x02\n\x14\x45ntryContextPWatched\x12@\n\x08\x61utoplay\x18\x01 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12\x41\n\tsel_thumb\x18\x02 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12\x43\n\x0b\x63hoose_next\x18\x03 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12@\n\x08\x63h_swtch\x18\x04 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12O\n\x17launch_first_in_session\x18\x05 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\"\x85\x02\n\x0fPositionPSelect\x12;\n\tfirst_pos\x18\x01 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12<\n\nsecond_pos\x18\x02 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12;\n\tthird_pos\x18\x03 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12:\n\x08rest_pos\x18\x04 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\"\xa9\x01\n\x1f\x42rowsedDebiasedPositionPSelects\x12\x44\n\x0fup_to_4_browsed\x18\x01 \x01(\x0b\x32+.haystack_ml_stack.features.PositionPSelect\x12@\n\x0b\x61ll_browsed\x18\x02 \x01(\x0b\x32+.haystack_ml_stack.features.PositionPSelect\"\xb8\x01\n\x16PlaylistStatsForGlobal\x12\x15\n\rwatched_count\x18\x01 \x01(\x05\x12\x19\n\x11not_watched_count\x18\x02 \x01(\x05\x12\x1b\n\x13\x63\x61pped_watched_secs\x18\x03 \x01(\x02\x12\x1f\n\x17\x63\x61pped_not_watched_secs\x18\x04 \x01(\x02\x12\x14\n\x0cwatched_secs\x18\x05 \x01(\x02\x12\x18\n\x10not_watched_secs\x18\x06 \x01(\x02\"\x88\x01\n\x14PlaylistStatsForUser\x12\x12\n\ntotal_days\x18\x01 \x01(\x05\x12\x12\n\nstart_days\x18\x02 \x01(\x05\x12\x13\n\x0b\x61\x63tive_days\x18\x03 \x01(\x05\x12\x15\n\rtotal_watched\x18\x04 \x01(\x02\x12\x1c\n\x14\x63\x61pped_total_watched\x18\x05 \x01(\x02\"C\n\x07\x43hannel\x12\x0c\n\x04name\x18\x01 \x01(\t\x12\x16\n\x0e\x63\x61tegory_group\x18\x02 \x01(\t\x12\x12\n\nstart_date\x18\x03 \x01(\x05\"k\n\rStreamPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12I\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects\"a\n\x0eStreamPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12>\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched\"_\n\x0cUserPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12>\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched\"\xda\x01\n\x19UserPersonalizingPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12M\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32?.haystack_ml_stack.features.UserPersonalizingPWatched.DataEntry\x1a]\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12?\n\x05value\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched:\x02\x38\x01\"i\n\x0bUserPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12I\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects\"\xe3\x01\n\x18UserPersonalizingPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12L\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32>.haystack_ml_stack.features.UserPersonalizingPSelect.DataEntry\x1ah\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12J\n\x05value\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects:\x02\x38\x01\"\xa2\x01\n\x16StreamSimilarityScores\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12J\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32<.haystack_ml_stack.features.StreamSimilarityScores.DataEntry\x1a+\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x01:\x02\x38\x01\"\xd0\x01\n\x13GlobalPlaylistStats\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12G\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32\x39.haystack_ml_stack.features.GlobalPlaylistStats.DataEntry\x1a_\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\x41\n\x05value\x18\x02 \x01(\x0b\x32\x32.haystack_ml_stack.features.PlaylistStatsForGlobal:\x02\x38\x01\"\xca\x01\n\x11UserPlaylistStats\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12\x45\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32\x37.haystack_ml_stack.features.UserPlaylistStats.DataEntry\x1a]\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12?\n\x05value\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.PlaylistStatsForUser:\x02\x38\x01\"T\n\x0eGlobalChannels\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12\x31\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32#.haystack_ml_stack.features.Channelb\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x0e\x66\x65\x61tures.proto\x12\x1ahaystack_ml_stack.features\"7\n\x12\x45ntryContextCounts\x12\x10\n\x08\x61ttempts\x18\x01 \x01(\x05\x12\x0f\n\x07watched\x18\x02 \x01(\x05\"_\n\x0cSelectCounts\x12\x15\n\rtotal_selects\x18\x01 \x01(\x05\x12!\n\x19total_selects_and_watched\x18\x02 \x01(\x05\x12\x15\n\rtotal_browsed\x18\x03 \x01(\x05\"\xf3\x02\n\x14\x45ntryContextPWatched\x12@\n\x08\x61utoplay\x18\x01 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12\x41\n\tsel_thumb\x18\x02 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12\x43\n\x0b\x63hoose_next\x18\x03 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12@\n\x08\x63h_swtch\x18\x04 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\x12O\n\x17launch_first_in_session\x18\x05 \x01(\x0b\x32..haystack_ml_stack.features.EntryContextCounts\"\x85\x02\n\x0fPositionPSelect\x12;\n\tfirst_pos\x18\x01 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12<\n\nsecond_pos\x18\x02 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12;\n\tthird_pos\x18\x03 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\x12:\n\x08rest_pos\x18\x04 \x01(\x0b\x32(.haystack_ml_stack.features.SelectCounts\"\xa9\x01\n\x1f\x42rowsedDebiasedPositionPSelects\x12\x44\n\x0fup_to_4_browsed\x18\x01 \x01(\x0b\x32+.haystack_ml_stack.features.PositionPSelect\x12@\n\x0b\x61ll_browsed\x18\x02 \x01(\x0b\x32+.haystack_ml_stack.features.PositionPSelect\"\xb8\x01\n\x16PlaylistStatsForGlobal\x12\x15\n\rwatched_count\x18\x01 \x01(\x05\x12\x19\n\x11not_watched_count\x18\x02 \x01(\x05\x12\x1b\n\x13\x63\x61pped_watched_secs\x18\x03 \x01(\x02\x12\x1f\n\x17\x63\x61pped_not_watched_secs\x18\x04 \x01(\x02\x12\x14\n\x0cwatched_secs\x18\x05 \x01(\x02\x12\x18\n\x10not_watched_secs\x18\x06 \x01(\x02\"\xc5\x01\n\x14PlaylistStatsForUser\x12\x12\n\ntotal_days\x18\x01 \x01(\x05\x12\x12\n\nstart_days\x18\x02 \x01(\x05\x12\x13\n\x0b\x61\x63tive_days\x18\x03 \x01(\x05\x12\x15\n\rtotal_watched\x18\x04 \x01(\x02\x12\x1c\n\x14\x63\x61pped_total_watched\x18\x05 \x01(\x02\x12 \n\x18\x63\x61pped_total_log_watched\x18\x06 \x01(\x02\x12\x19\n\x11\x66irst_active_date\x18\x07 \x01(\t\"^\n\x07\x43hannel\x12\x0c\n\x04name\x18\x01 \x01(\t\x12\x16\n\x0e\x63\x61tegory_group\x18\x02 \x01(\t\x12\x12\n\nstart_date\x18\x03 \x01(\x05\x12\x19\n\x11required_features\x18\x04 \x03(\t\"k\n\rStreamPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12I\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects\"a\n\x0eStreamPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12>\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched\"_\n\x0cUserPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12>\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched\"\xda\x01\n\x19UserPersonalizingPWatched\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12M\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32?.haystack_ml_stack.features.UserPersonalizingPWatched.DataEntry\x1a]\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12?\n\x05value\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.EntryContextPWatched:\x02\x38\x01\"i\n\x0bUserPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12I\n\x04\x64\x61ta\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects\"\xe3\x01\n\x18UserPersonalizingPSelect\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12L\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32>.haystack_ml_stack.features.UserPersonalizingPSelect.DataEntry\x1ah\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12J\n\x05value\x18\x02 \x01(\x0b\x32;.haystack_ml_stack.features.BrowsedDebiasedPositionPSelects:\x02\x38\x01\"\xa2\x01\n\x16StreamSimilarityScores\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12J\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32<.haystack_ml_stack.features.StreamSimilarityScores.DataEntry\x1a+\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x01:\x02\x38\x01\"\xd0\x01\n\x13GlobalPlaylistStats\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12G\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32\x39.haystack_ml_stack.features.GlobalPlaylistStats.DataEntry\x1a_\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\x41\n\x05value\x18\x02 \x01(\x0b\x32\x32.haystack_ml_stack.features.PlaylistStatsForGlobal:\x02\x38\x01\"\xca\x01\n\x11UserPlaylistStats\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12\x45\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32\x37.haystack_ml_stack.features.UserPlaylistStats.DataEntry\x1a]\n\tDataEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12?\n\x05value\x18\x02 \x01(\x0b\x32\x30.haystack_ml_stack.features.PlaylistStatsForUser:\x02\x38\x01\"T\n\x0eGlobalChannels\x12\x0f\n\x07version\x18\x01 \x01(\x05\x12\x31\n\x04\x64\x61ta\x18\x02 \x03(\x0b\x32#.haystack_ml_stack.features.Channelb\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
@@ -54,37 +54,37 @@ if not _descriptor._USE_C_DESCRIPTORS:
   _globals['_PLAYLISTSTATSFORGLOBAL']._serialized_start=1011
   _globals['_PLAYLISTSTATSFORGLOBAL']._serialized_end=1195
   _globals['_PLAYLISTSTATSFORUSER']._serialized_start=1198
-  _globals['_PLAYLISTSTATSFORUSER']._serialized_end=1334
-  _globals['_CHANNEL']._serialized_start=1336
-  _globals['_CHANNEL']._serialized_end=1403
-  _globals['_STREAMPSELECT']._serialized_start=1405
-  _globals['_STREAMPSELECT']._serialized_end=1512
-  _globals['_STREAMPWATCHED']._serialized_start=1514
-  _globals['_STREAMPWATCHED']._serialized_end=1611
-  _globals['_USERPWATCHED']._serialized_start=1613
-  _globals['_USERPWATCHED']._serialized_end=1708
-  _globals['_USERPERSONALIZINGPWATCHED']._serialized_start=1711
-  _globals['_USERPERSONALIZINGPWATCHED']._serialized_end=1929
-  _globals['_USERPERSONALIZINGPWATCHED_DATAENTRY']._serialized_start=1836
-  _globals['_USERPERSONALIZINGPWATCHED_DATAENTRY']._serialized_end=1929
-  _globals['_USERPSELECT']._serialized_start=1931
-  _globals['_USERPSELECT']._serialized_end=2036
-  _globals['_USERPERSONALIZINGPSELECT']._serialized_start=2039
-  _globals['_USERPERSONALIZINGPSELECT']._serialized_end=2266
-  _globals['_USERPERSONALIZINGPSELECT_DATAENTRY']._serialized_start=2162
-  _globals['_USERPERSONALIZINGPSELECT_DATAENTRY']._serialized_end=2266
-  _globals['_STREAMSIMILARITYSCORES']._serialized_start=2269
-  _globals['_STREAMSIMILARITYSCORES']._serialized_end=2431
-  _globals['_STREAMSIMILARITYSCORES_DATAENTRY']._serialized_start=2388
-  _globals['_STREAMSIMILARITYSCORES_DATAENTRY']._serialized_end=2431
-  _globals['_GLOBALPLAYLISTSTATS']._serialized_start=2434
-  _globals['_GLOBALPLAYLISTSTATS']._serialized_end=2642
-  _globals['_GLOBALPLAYLISTSTATS_DATAENTRY']._serialized_start=2547
-  _globals['_GLOBALPLAYLISTSTATS_DATAENTRY']._serialized_end=2642
-  _globals['_USERPLAYLISTSTATS']._serialized_start=2645
-  _globals['_USERPLAYLISTSTATS']._serialized_end=2847
-  _globals['_USERPLAYLISTSTATS_DATAENTRY']._serialized_start=2754
-  _globals['_USERPLAYLISTSTATS_DATAENTRY']._serialized_end=2847
-  _globals['_GLOBALCHANNELS']._serialized_start=2849
-  _globals['_GLOBALCHANNELS']._serialized_end=2933
+  _globals['_PLAYLISTSTATSFORUSER']._serialized_end=1395
+  _globals['_CHANNEL']._serialized_start=1397
+  _globals['_CHANNEL']._serialized_end=1491
+  _globals['_STREAMPSELECT']._serialized_start=1493
+  _globals['_STREAMPSELECT']._serialized_end=1600
+  _globals['_STREAMPWATCHED']._serialized_start=1602
+  _globals['_STREAMPWATCHED']._serialized_end=1699
+  _globals['_USERPWATCHED']._serialized_start=1701
+  _globals['_USERPWATCHED']._serialized_end=1796
+  _globals['_USERPERSONALIZINGPWATCHED']._serialized_start=1799
+  _globals['_USERPERSONALIZINGPWATCHED']._serialized_end=2017
+  _globals['_USERPERSONALIZINGPWATCHED_DATAENTRY']._serialized_start=1924
+  _globals['_USERPERSONALIZINGPWATCHED_DATAENTRY']._serialized_end=2017
+  _globals['_USERPSELECT']._serialized_start=2019
+  _globals['_USERPSELECT']._serialized_end=2124
+  _globals['_USERPERSONALIZINGPSELECT']._serialized_start=2127
+  _globals['_USERPERSONALIZINGPSELECT']._serialized_end=2354
+  _globals['_USERPERSONALIZINGPSELECT_DATAENTRY']._serialized_start=2250
+  _globals['_USERPERSONALIZINGPSELECT_DATAENTRY']._serialized_end=2354
+  _globals['_STREAMSIMILARITYSCORES']._serialized_start=2357
+  _globals['_STREAMSIMILARITYSCORES']._serialized_end=2519
+  _globals['_STREAMSIMILARITYSCORES_DATAENTRY']._serialized_start=2476
+  _globals['_STREAMSIMILARITYSCORES_DATAENTRY']._serialized_end=2519
+  _globals['_GLOBALPLAYLISTSTATS']._serialized_start=2522
+  _globals['_GLOBALPLAYLISTSTATS']._serialized_end=2730
+  _globals['_GLOBALPLAYLISTSTATS_DATAENTRY']._serialized_start=2635
+  _globals['_GLOBALPLAYLISTSTATS_DATAENTRY']._serialized_end=2730
+  _globals['_USERPLAYLISTSTATS']._serialized_start=2733
+  _globals['_USERPLAYLISTSTATS']._serialized_end=2935
+  _globals['_USERPLAYLISTSTATS_DATAENTRY']._serialized_start=2842
+  _globals['_USERPLAYLISTSTATS_DATAENTRY']._serialized_end=2935
+  _globals['_GLOBALCHANNELS']._serialized_start=2937
+  _globals['_GLOBALCHANNELS']._serialized_end=3021
 # @@protoc_insertion_point(module_scope)

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/generated/v1/features_pb2.pyi

@@ -81,22 +81,28 @@ class PlaylistStatsForUser(_message.Message):
     ACTIVE_DAYS_FIELD_NUMBER: _ClassVar[int]
     TOTAL_WATCHED_FIELD_NUMBER: _ClassVar[int]
     CAPPED_TOTAL_WATCHED_FIELD_NUMBER: _ClassVar[int]
+    CAPPED_TOTAL_LOG_WATCHED_FIELD_NUMBER: _ClassVar[int]
+    FIRST_ACTIVE_DATE_FIELD_NUMBER: _ClassVar[int]
     total_days: int
     start_days: int
     active_days: int
     total_watched: float
     capped_total_watched: float
-    def __init__(self, total_days: _Optional[int] = ..., start_days: _Optional[int] = ..., active_days: _Optional[int] = ..., total_watched: _Optional[float] = ..., capped_total_watched: _Optional[float] = ...) -> None: ...
+    capped_total_log_watched: float
+    first_active_date: str
+    def __init__(self, total_days: _Optional[int] = ..., start_days: _Optional[int] = ..., active_days: _Optional[int] = ..., total_watched: _Optional[float] = ..., capped_total_watched: _Optional[float] = ..., capped_total_log_watched: _Optional[float] = ..., first_active_date: _Optional[str] = ...) -> None: ...
 
 class Channel(_message.Message):
     __slots__ = ()
     NAME_FIELD_NUMBER: _ClassVar[int]
     CATEGORY_GROUP_FIELD_NUMBER: _ClassVar[int]
     START_DATE_FIELD_NUMBER: _ClassVar[int]
+    REQUIRED_FEATURES_FIELD_NUMBER: _ClassVar[int]
     name: str
     category_group: str
     start_date: int
-    def __init__(self, name: _Optional[str] = ..., category_group: _Optional[str] = ..., start_date: _Optional[int] = ...) -> None: ...
+    required_features: _containers.RepeatedScalarFieldContainer[str]
+    def __init__(self, name: _Optional[str] = ..., category_group: _Optional[str] = ..., start_date: _Optional[int] = ..., required_features: _Optional[_Iterable[str]] = ...) -> None: ...
 
 class StreamPSelect(_message.Message):
     __slots__ = ()

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack/utils.py

@@ -715,6 +715,8 @@ def _complete_features_for_channels(
         "active_days",
         "total_watched",
         "capped_total_watched",
+        "capped_total_log_watched",
+        "first_active_date",
     ]
 
     for ch in channels:

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/src/haystack_ml_stack.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haystack-ml-stack
-Version: 0.4.8
+Version: 0.4.10
 Summary: Functions related to Haystack ML
 Author-email: Oscar Vega <oscar@haystack.tv>
 License: MIT

{haystack_ml_stack-0.4.8 → haystack_ml_stack-0.4.10}/tests/test_serializers.py

@@ -150,3 +150,64 @@ def test_user_bias_serializers():
         "User pselect serialization does not match expected value."
     )
     return
+
+
+def test_user_playlist_stats_serializer_backward_compatibility():
+    serializer = SerializerRegistry[("USER", "PLAYLIST_CATEGORY_STATS#3M", "v1")]
+
+    # Case 1: true old binary — built without fields 6 and 7, identical to what
+    # the old serializer would have written before the proto change
+    old_msg = features_pb2_v1.UserPlaylistStats()
+    old_msg.version = 1
+    entry = old_msg.data["sports"]
+    entry.total_days = 30
+    entry.start_days = 5
+    entry.active_days = 10
+    entry.total_watched = 120.5
+    entry.capped_total_watched = 100.0
+    old_binary = old_msg.SerializeToString()
+
+    result = serializer.deserialize(old_binary)
+    assert result.data["sports"].total_days == 30
+    assert result.data["sports"].capped_total_watched == 100.0
+    assert result.data["sports"].capped_total_log_watched == 0.0
+    assert result.data["sports"].first_active_date == ""
+
+    # Case 2: new data with new fields — full round-trip
+    new_data = {
+        "version": 1,
+        "data": {
+            "sports": {
+                "total_days": 30,
+                "start_days": 5,
+                "active_days": 10,
+                "total_watched": 120.5,
+                "capped_total_watched": 100.0,
+                "capped_total_log_watched": 4.8,
+                "first_active_date": "2024-01-15",
+            }
+        },
+    }
+    binary = serializer.serialize(new_data)
+    result = serializer.deserialize(binary)
+    assert abs(result.data["sports"].capped_total_log_watched - 4.8) < 1e-6
+    assert result.data["sports"].first_active_date == "2024-01-15"
+
+    # Case 3: old-format dict missing the new fields — .get() fallback
+    old_format_data = {
+        "version": 1,
+        "data": {
+            "sports": {
+                "total_days": 30,
+                "start_days": 5,
+                "active_days": 10,
+                "total_watched": 120.5,
+                "capped_total_watched": 100.0,
+            }
+        },
+    }
+    binary = serializer.serialize(old_format_data)
+    result = serializer.deserialize(binary)
+    assert result.data["sports"].total_days == 30
+    assert result.data["sports"].capped_total_log_watched == 0.0
+    assert result.data["sports"].first_active_date == ""

haystack_ml_stack-0.4.8/src/haystack_ml_stack/_version.py

@@ -1 +0,0 @@
-__version__ = "0.4.8"