chalkruntime 3.43.75__tar.gz → 3.43.76__tar.gz

{chalkruntime-3.43.75 → chalkruntime-3.43.76}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: chalkruntime
-Version: 3.43.75
+Version: 3.43.76
 Summary: Runtime support library for Chalk AI
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown

{chalkruntime-3.43.75 → chalkruntime-3.43.76}/chalkruntime/invoker/vectorized_hasmany_sampler.py

@@ -689,6 +689,12 @@ class PolarsVectorizedHasManySampler(VectorizedHasManySampler):
         hm_lf = hm_df
         hm_mapping_lf = mapping_table
 
+        # See `HAS_MANY_INPUT_TS_COL_NAME`: present iff each has-many row is a version for one input time,
+        # in which case an input row only matches the rows computed for its own `__ts__`.
+        join_on_input_time = HAS_MANY_INPUT_TS_COL_NAME in hm_lf.columns
+        if join_on_input_time and TS_COL_NAME not in resolver_inputs_lf.columns:
+            raise ValueError(f"The resolver inputs must have a '{TS_COL_NAME}' column to match has-many data on it")
+
         # If we do not have the feature time column, append it with nulls
         if foreign_ts_feature.fqn not in hm_lf.columns:
             hm_lf = hm_lf.with_columns(pl.lit(None, pl.Datetime("us", "UTC")).alias(foreign_ts_feature.fqn))
@@ -703,6 +709,9 @@ class PolarsVectorizedHasManySampler(VectorizedHasManySampler):
             unique_on = [foreign_pkey_feature.fqn, foreign_ts_feature.fqn]
             if not allow_planner_postponed_has_many_sampling_planner_option:
                 unique_on.append(TS_COL_NAME)
+            if join_on_input_time:
+                # Versions computed for different input times are distinct rows, not duplicates.
+                unique_on.append(HAS_MANY_INPUT_TS_COL_NAME)
             # Add __index__ to prevent cross-transaction deduplication in bulk queries
             # This ensures that duplicate GL codes across different transactions in a bulk query
             # are treated as distinct results, not as duplicates to be removed
@@ -726,6 +735,11 @@ class PolarsVectorizedHasManySampler(VectorizedHasManySampler):
             over = [pl.col(foreign_pkey_feature.fqn)]
             if INDEX_COL_NAME in hm_lf.columns:
                 over.append(pl.col(INDEX_COL_NAME))
+            if join_on_input_time:
+                # A row is replaced only by a later version computed for the same input time; the versions for
+                # other input times are separate histories and would otherwise mark this one as replaced at its
+                # own feature time.
+                over.append(pl.col(HAS_MANY_INPUT_TS_COL_NAME))
             hm_lf = hm_lf.with_columns(
                 [
                     pl.col(foreign_ts_feature.fqn).alias(OBSERVED_AT_FEATURE_COL_NAME),
@@ -788,6 +802,9 @@ class PolarsVectorizedHasManySampler(VectorizedHasManySampler):
             for f in filters_from_feature
         ]
 
+        if join_on_input_time:
+            left_join_cols = [*left_join_cols, TS_COL_NAME]
+            right_join_cols = [*right_join_cols, HAS_MANY_INPUT_TS_COL_NAME]
         # Rename the left join column here since all the left table's column names had that prefix added.
         left_join_cols = [RESOLVER_INPUT_PREFIX + left_join_col for left_join_col in left_join_cols]
 
@@ -1472,6 +1489,8 @@ class PyArrowVectorizedHasManySampler(VectorizedHasManySampler):
             table_has_struct_or_list(resolver_inputs_table)
             or table_has_struct_or_list(has_many_table[0])
             or has_many_feature.underlying.get_nearest_neighbor_join_info()
+            # This sampler does not match per-input-time has-many rows on the input time.
+            or HAS_MANY_INPUT_TS_COL_NAME in has_many_table[0].column_names
         )
 
 

{chalkruntime-3.43.75 → chalkruntime-3.43.76}/chalkruntime.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: chalkruntime
-Version: 3.43.75
+Version: 3.43.76
 Summary: Runtime support library for Chalk AI
 Requires-Python: >=3.11
 Description-Content-Type: text/markdown

{chalkruntime-3.43.75 → chalkruntime-3.43.76}/pyproject.toml

@@ -45,7 +45,7 @@ name = "chalkruntime"
 description = "Runtime support library for Chalk AI"
 readme = "README.md"
 requires-python = ">=3.11"
-version = "3.43.75"
+version = "3.43.76"
 
 
 [tool.deptry]