deltacat 2.0.0b11__py3-none-any.whl → 2.0.0b12__py3-none-any.whl

deltacat/tests/compute/test_util_common.py

@@ -1,14 +1,22 @@
+"""
+Common utility functions for main storage compaction tests.
+
+These functions are shared between incremental and multiple rounds compaction tests.
+"""
 # Allow classes to use self-referencing Type hints in Python 3.7.
 from __future__ import annotations
 from enum import Enum
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Tuple
 import datetime as dt
-from boto3.resources.base import ServiceResource
 from datetime import timezone
 
-from deltacat.tests.compute.test_util_constant import (
-    TEST_S3_RCF_BUCKET_NAME,
-)
+import tempfile
+import os
+import shutil
+
+import pyarrow as pa
+
+
 from deltacat.tests.compute.test_util_constant import (
     BASE_TEST_SOURCE_NAMESPACE,
     BASE_TEST_SOURCE_TABLE_NAME,
@@ -26,11 +34,10 @@ from deltacat.compute.compactor import (
 from deltacat.compute.compactor.model.compaction_session_audit_info import (
     CompactionSessionAuditInfo,
 )
-
 from deltacat.storage.model.partition import (
     PartitionLocator,
     PartitionScheme,
-    PartitionKey as PartitionSchemeKey,
+    PartitionKey as StoragePartitionKey,
 )
 from deltacat.storage.model.stream import StreamLocator
 from deltacat.storage.model.table_version import TableVersionLocator
@@ -39,8 +46,22 @@ from deltacat.storage.model.namespace import NamespaceLocator
 from deltacat.storage.model.sort_key import (
     SortScheme,
 )
+from deltacat.storage.model.delta import (
+    Delta,
+    DeltaType,
+)
+from deltacat.storage.model.partition import (
+    Partition,
+    PartitionKeyList,
+)
+from deltacat.storage.model.stream import Stream
+from deltacat.storage.model.transform import IdentityTransform
+from deltacat.storage.model.schema import Schema
 from deltacat.compute.compactor.model.compactor_version import CompactorVersion
 
+from deltacat.storage import metastore
+from deltacat.catalog.model.properties import CatalogProperties
+
 
 class PartitionKeyType(str, Enum):
     INT = "int"
@@ -80,113 +101,741 @@ def get_test_partition_locator(partition_id):
     return partition_locator
 
 
-def _create_table(
+def create_main_deltacat_storage_kwargs() -> Dict[str, Any]:
+    """
+    Helper function to create main deltacat storage kwargs
+
+    Returns: kwargs to use for main deltacat storage, i.e. {"catalog": CatalogProperties(...)}
+    """
+    temp_dir = tempfile.mkdtemp()
+    catalog = CatalogProperties(root=temp_dir)
+    return {"catalog": catalog}
+
+
+def clean_up_main_deltacat_storage_kwargs(storage_kwargs: Dict[str, Any]):
+    """
+    Cleans up directory created by create_main_deltacat_storage_kwargs
+    """
+    catalog = storage_kwargs["catalog"]
+    if hasattr(catalog, "root") and os.path.exists(catalog.root):
+        shutil.rmtree(catalog.root)
+
+
+def _create_table_main(
     namespace: str,
     table_name: str,
     table_version: str,
     sort_keys: Optional[List[Any]],
     partition_keys: Optional[List[PartitionKey]],
+    input_deltas: Optional[pa.Table],
     ds_mock_kwargs: Optional[Dict[str, Any]],
 ):
-    import deltacat.tests.local_deltacat_storage as ds
-    from deltacat.types.media import ContentType
+    """
+    Main storage version of _create_table that works for both incremental and multiple rounds tests.
+
+    For incremental tests, input_deltas is provided to extract schema.
+    For multiple rounds tests, input_deltas can be None and we use a simpler approach.
+    """
+    # Create namespace first
+    metastore.create_namespace(namespace=namespace, **ds_mock_kwargs)
+
+    # Handle schema creation
+    if input_deltas is not None:
+        # Incremental test approach - extract schema from input deltas
+        schema = input_deltas.schema
+
+        # Add partition key fields to schema if they're not already present
+        if partition_keys:
+            for pk in partition_keys:
+                field_name = pk.key_name
+                if field_name not in schema.names:
+                    # Add partition key field with appropriate type
+                    if pk.key_type == PartitionKeyType.INT:
+                        field_type = pa.int32()
+                    elif pk.key_type == PartitionKeyType.STRING:
+                        field_type = pa.string()
+                    elif (
+                        pk.key_type == PartitionKeyType.TIMESTAMP
+                    ):  # Handle timestamp type properly
+                        field_type = pa.timestamp("us")
+                    else:
+                        field_type = pa.string()  # Default to string
+
+                    schema = schema.append(pa.field(field_name, field_type))
+
+        schema_obj = Schema.of(schema=schema)
+    else:
+        # Multiple rounds test approach - use None for schema (will be set later)
+        schema_obj = None
 
-    ds.create_namespace(namespace, {}, **ds_mock_kwargs)
-    partition_scheme = (
-        PartitionScheme.of(
-            [PartitionSchemeKey.of(key.key_name) for key in partition_keys]
-        )
-        if partition_keys
-        else None
-    )
     sort_scheme = SortScheme.of(sort_keys) if sort_keys else None
-    ds.create_table_version(
-        namespace,
-        table_name,
-        table_version,
-        sort_keys=sort_scheme,
+
+    # Convert test partition keys to storage partition keys
+    storage_partition_keys = []
+    if partition_keys:
+        for pk in partition_keys:
+            storage_partition_key = StoragePartitionKey.of(
+                key=[pk.key_name],
+                name=pk.key_name,
+                transform=IdentityTransform.of(),
+            )
+            storage_partition_keys.append(storage_partition_key)
+
+    # Create partition scheme
+    partition_scheme = None
+    if storage_partition_keys:
+        partition_scheme = PartitionScheme.of(
+            keys=PartitionKeyList.of(storage_partition_keys),
+            scheme_id="default_partition_scheme",
+        )
+
+    # Create table version (which creates table and stream automatically)
+    metastore.create_table_version(
+        namespace=namespace,
+        table_name=table_name,
+        table_version=table_version,
+        schema=schema_obj,
         partition_scheme=partition_scheme,
-        supported_content_types=[ContentType.PARQUET],
+        sort_keys=sort_scheme,
         **ds_mock_kwargs,
     )
+
     return namespace, table_name, table_version
 
 
-def create_src_table(
+def create_src_table_main(
     sort_keys: Optional[List[Any]],
     partition_keys: Optional[List[PartitionKey]],
+    input_deltas: Optional[pa.Table],
     ds_mock_kwargs: Optional[Dict[str, Any]],
 ):
+    """
+    Main storage version of create_src_table
+    """
     source_namespace: str = BASE_TEST_SOURCE_NAMESPACE
     source_table_name: str = BASE_TEST_SOURCE_TABLE_NAME
     source_table_version: str = BASE_TEST_SOURCE_TABLE_VERSION
-    return _create_table(
+    return _create_table_main(
         source_namespace,
         source_table_name,
         source_table_version,
         sort_keys,
         partition_keys,
+        input_deltas,
         ds_mock_kwargs,
     )
 
 
-def create_destination_table(
+def create_destination_table_main(
     sort_keys: Optional[List[Any]],
     partition_keys: Optional[List[PartitionKey]],
+    input_deltas: Optional[pa.Table],
     ds_mock_kwargs: Optional[Dict[str, Any]],
 ):
+    """
+    Main storage version of create_destination_table
+    """
     destination_namespace: str = BASE_TEST_DESTINATION_NAMESPACE
     destination_table_name: str = BASE_TEST_DESTINATION_TABLE_NAME
     destination_table_version: str = BASE_TEST_DESTINATION_TABLE_VERSION
-    return _create_table(
+    return _create_table_main(
         destination_namespace,
         destination_table_name,
         destination_table_version,
         sort_keys,
         partition_keys,
+        input_deltas,
         ds_mock_kwargs,
     )
 
 
-def create_rebase_table(
+def create_rebase_table_main(
     sort_keys: Optional[List[Any]],
     partition_keys: Optional[List[PartitionKey]],
+    input_deltas: Optional[pa.Table],
     ds_mock_kwargs: Optional[Dict[str, Any]],
 ):
+    """
+    Main storage version of create_rebase_table
+    """
     rebasing_namespace = REBASING_NAMESPACE
     rebasing_table_name = REBASING_TABLE_NAME
     rebasing_table_version = REBASING_TABLE_VERSION
-    return _create_table(
+    return _create_table_main(
         rebasing_namespace,
         rebasing_table_name,
         rebasing_table_version,
         sort_keys,
         partition_keys,
+        input_deltas,
         ds_mock_kwargs,
     )
 
 
-def get_rcf(s3_resource, rcf_file_s3_uri: str) -> RoundCompletionInfo:
-    from deltacat.tests.test_utils.utils import read_s3_contents
+def get_rci_from_partition(
+    partition_locator: PartitionLocator, deltacat_storage=None, **kwargs
+) -> RoundCompletionInfo:
+    """
+    Read RoundCompletionInfo from a partition metafile.
+
+    Args:
+        partition_locator: Locator of the partition containing the RoundCompletionInfo
+        deltacat_storage: Storage implementation (defaults to metastore)
+        **kwargs: Additional arguments to pass to deltacat_storage.get_partition (e.g., catalog)
+
+    Returns:
+        RoundCompletionInfo object from the partition, or None if not found
+    """
+    from deltacat.storage import metastore
+
+    if deltacat_storage is None:
+        deltacat_storage = metastore
+
+    partition = deltacat_storage.get_partition(
+        partition_locator.stream_locator, partition_locator.partition_values, **kwargs
+    )
+
+    if partition and partition.compaction_round_completion_info:
+        return partition.compaction_round_completion_info
+
+    return None
 
-    _, rcf_object_key = rcf_file_s3_uri.strip("s3://").split("/", 1)
-    rcf_file_output: Dict[str, Any] = read_s3_contents(
-        s3_resource, TEST_S3_RCF_BUCKET_NAME, rcf_object_key
+
+def _add_deltas_to_partition_main(
+    deltas_ingredients: List[Tuple[pa.Table, DeltaType, Optional[Dict[str, str]]]],
+    partition: Optional[Partition],
+    ds_mock_kwargs: Optional[Dict[str, Any]],
+) -> Tuple[Optional[Delta], int]:
+    """
+    Add deltas to a partition using main storage
+    """
+    all_deltas_length = 0
+    incremental_delta = None
+    for (delta_data, delta_type, delete_parameters) in deltas_ingredients:
+        staged_delta: Delta = metastore.stage_delta(
+            delta_data,
+            partition,
+            delta_type,
+            entry_params=delete_parameters,
+            **ds_mock_kwargs,
+        )
+        incremental_delta = metastore.commit_delta(
+            staged_delta,
+            **ds_mock_kwargs,
+        )
+        all_deltas_length += len(delta_data) if delta_data else 0
+    return incremental_delta, all_deltas_length
+
+
+def add_late_deltas_to_partition_main(
+    late_deltas: List[Tuple[pa.Table, DeltaType, Optional[Dict[str, str]]]],
+    source_partition: Optional[Partition],
+    ds_mock_kwargs: Optional[Dict[str, Any]],
+) -> Tuple[Optional[Delta], int]:
+    """
+    Add late deltas to a partition using main storage
+    """
+    return _add_deltas_to_partition_main(late_deltas, source_partition, ds_mock_kwargs)
+
+
+def multiple_rounds_create_src_w_deltas_destination_rebase_w_deltas_strategy_main(
+    sort_keys: Optional[List[Any]],
+    partition_keys: Optional[List[PartitionKey]],
+    input_deltas: List[pa.Table],
+    partition_values: Optional[List[Any]],
+    ds_mock_kwargs: Optional[Dict[str, Any]],
+) -> Tuple[Stream, Stream, Optional[Stream], bool]:
+    """
+    Main storage version of multiple_rounds_create_src_w_deltas_destination_rebase_w_deltas_strategy
+    """
+    # For multiple rounds, we need to extract the first delta to get schema
+    first_delta_table = input_deltas[0][0] if input_deltas else None
+    source_namespace, source_table_name, source_table_version = create_src_table_main(
+        sort_keys, partition_keys, first_delta_table, ds_mock_kwargs
     )
-    return RoundCompletionInfo(**rcf_file_output)
 
+    source_table_stream: Stream = metastore.get_stream(
+        namespace=source_namespace,
+        table_name=source_table_name,
+        table_version=source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Convert partition values to correct types
+    converted_partition_values = []
+    if partition_values and partition_keys:
+        for i, (value, pk) in enumerate(zip(partition_values, partition_keys)):
+            if pk.key_type == PartitionKeyType.INT:
+                converted_partition_values.append(int(value))
+            else:
+                converted_partition_values.append(value)
+    else:
+        converted_partition_values = partition_values
+
+    staged_partition: Partition = metastore.stage_partition(
+        source_table_stream,
+        converted_partition_values,
+        partition_scheme_id="default_partition_scheme" if partition_keys else None,
+        **ds_mock_kwargs,
+    )
+
+    is_delete = False
+    input_delta_length = 0
+    for (
+        input_delta,
+        input_delta_type,
+        input_delta_parameters,
+    ) in input_deltas:
+        if input_delta_type is DeltaType.DELETE:
+            is_delete = True
+        staged_delta = metastore.stage_delta(
+            input_delta,
+            staged_partition,
+            input_delta_type,
+            entry_params=input_delta_parameters,
+            **ds_mock_kwargs,
+        )
+        metastore.commit_delta(staged_delta, **ds_mock_kwargs)
+        input_delta_length += len(input_delta)
+    metastore.commit_partition(staged_partition, **ds_mock_kwargs)
+
+    (
+        destination_table_namespace,
+        destination_table_name,
+        destination_table_version,
+    ) = create_destination_table_main(
+        sort_keys, partition_keys, first_delta_table, ds_mock_kwargs
+    )
+    destination_table_stream: Stream = metastore.get_stream(
+        namespace=destination_table_namespace,
+        table_name=destination_table_name,
+        table_version=destination_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Always create rebase table for multiple rounds tests
+    (
+        rebasing_table_namespace,
+        rebasing_table_name,
+        rebasing_table_version,
+    ) = create_rebase_table_main(
+        sort_keys, partition_keys, first_delta_table, ds_mock_kwargs
+    )
+    rebasing_table_stream: Stream = metastore.get_stream(
+        namespace=rebasing_table_namespace,
+        table_name=rebasing_table_name,
+        table_version=rebasing_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Stage partition and add deltas to rebase table
+    rebased_staged_partition: Partition = metastore.stage_partition(
+        rebasing_table_stream,
+        converted_partition_values,
+        partition_scheme_id="default_partition_scheme" if partition_keys else None,
+        **ds_mock_kwargs,
+    )
+
+    for (
+        input_delta,
+        input_delta_type,
+        input_delta_parameters,
+    ) in input_deltas:
+        staged_delta = metastore.stage_delta(
+            input_delta,
+            rebased_staged_partition,
+            input_delta_type,
+            entry_params=input_delta_parameters,
+            **ds_mock_kwargs,
+        )
+        metastore.commit_delta(staged_delta, **ds_mock_kwargs)
+    metastore.commit_partition(rebased_staged_partition, **ds_mock_kwargs)
+
+    return (
+        source_table_stream,
+        destination_table_stream,
+        rebasing_table_stream,
+        is_delete,
+    )
+
+
+def create_src_w_deltas_destination_plus_destination_main(
+    sort_keys: Optional[List[Any]],
+    partition_keys: Optional[List[PartitionKey]],
+    input_deltas: pa.Table,
+    input_delta_type: DeltaType,
+    partition_values: Optional[List[Any]],
+    ds_mock_kwargs: Optional[Dict[str, Any]],
+    simulate_is_inplace: bool = False,
+) -> Tuple[Stream, Stream, Optional[Stream], str, str, str]:
+    """
+    Create source with deltas and destination tables for incremental compaction testing
+    """
+    source_namespace, source_table_name, source_table_version = create_src_table_main(
+        sort_keys, partition_keys, input_deltas, ds_mock_kwargs
+    )
 
-def get_compacted_delta_locator_from_rcf(
-    s3_resource: ServiceResource, rcf_file_s3_uri: str
+    source_table_stream: Stream = metastore.get_stream(
+        namespace=source_namespace,
+        table_name=source_table_name,
+        table_version=source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Convert partition values to correct types
+    converted_partition_values = []
+    if partition_values and partition_keys:
+        for i, (value, pk) in enumerate(zip(partition_values, partition_keys)):
+            if pk.key_type == PartitionKeyType.INT:
+                converted_partition_values.append(int(value))
+            else:
+                converted_partition_values.append(value)
+    else:
+        converted_partition_values = partition_values
+
+    staged_partition: Partition = metastore.stage_partition(
+        source_table_stream,
+        converted_partition_values,
+        partition_scheme_id="default_partition_scheme" if partition_keys else None,
+        **ds_mock_kwargs,
+    )
+    metastore.commit_delta(
+        metastore.stage_delta(
+            input_deltas, staged_partition, input_delta_type, **ds_mock_kwargs
+        ),
+        **ds_mock_kwargs,
+    )
+    metastore.commit_partition(staged_partition, **ds_mock_kwargs)
+    source_table_stream_after_committed: Stream = metastore.get_stream(
+        namespace=source_namespace,
+        table_name=source_table_name,
+        table_version=source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    destination_table_namespace: Optional[str] = None
+    destination_table_name: Optional[str] = None
+    destination_table_version: Optional[str] = None
+    if not simulate_is_inplace:
+        (
+            destination_table_namespace,
+            destination_table_name,
+            destination_table_version,
+        ) = create_destination_table_main(
+            sort_keys, partition_keys, input_deltas, ds_mock_kwargs
+        )
+    else:
+        destination_table_namespace = source_namespace
+        destination_table_name = source_table_name
+        destination_table_version = source_table_version
+
+    destination_table_stream: Stream = metastore.get_stream(
+        namespace=destination_table_namespace,
+        table_name=destination_table_name,
+        table_version=destination_table_version,
+        **ds_mock_kwargs,
+    )
+
+    return (
+        source_table_stream_after_committed,
+        destination_table_stream,
+        None,
+        source_namespace,
+        source_table_name,
+        source_table_version,
+    )
+
+
+def create_src_w_deltas_destination_rebase_w_deltas_strategy_main(
+    sort_keys: Optional[List[Any]],
+    partition_keys: Optional[List[PartitionKey]],
+    input_deltas: pa.Table,
+    input_delta_type: DeltaType,
+    partition_values: Optional[List[Any]],
+    ds_mock_kwargs: Optional[Dict[str, Any]],
+) -> Tuple[Stream, Stream, Optional[Stream]]:
+    """
+    Main storage version of create_src_w_deltas_destination_rebase_w_deltas_strategy
+
+    Creates source table with deltas, destination table, and rebase table for rebase testing.
+    This test scenario sets up different source and rebase partition locators to simulate
+    scenarios like hash bucket count changes.
+    """
+    from deltacat.utils.common import current_time_ms
+
+    last_stream_position = current_time_ms()
+    source_namespace, source_table_name, source_table_version = create_src_table_main(
+        sort_keys, partition_keys, input_deltas, ds_mock_kwargs
+    )
+
+    source_table_stream: Stream = metastore.get_stream(
+        namespace=source_namespace,
+        table_name=source_table_name,
+        table_version=source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Convert partition values to correct types, including timestamp handling
+    converted_partition_values = []
+    if partition_values and partition_keys:
+        for i, (value, pk) in enumerate(zip(partition_values, partition_keys)):
+            if pk.key_type == PartitionKeyType.INT:
+                converted_partition_values.append(int(value))
+            elif pk.key_type == PartitionKeyType.TIMESTAMP:
+                # Handle timestamp partition values
+                if isinstance(value, str) and "T" in value and value.endswith("Z"):
+                    import pandas as pd
+
+                    ts = pd.to_datetime(value)
+                    # Convert to microseconds since epoch for PyArrow timestamp[us]
+                    converted_partition_values.append(int(ts.timestamp() * 1_000_000))
+                else:
+                    converted_partition_values.append(value)
+            else:
+                converted_partition_values.append(value)
+    else:
+        converted_partition_values = partition_values
+
+    staged_partition: Partition = metastore.stage_partition(
+        source_table_stream,
+        converted_partition_values,
+        partition_scheme_id="default_partition_scheme" if partition_keys else None,
+        **ds_mock_kwargs,
+    )
+    staged_delta: Delta = metastore.stage_delta(
+        input_deltas, staged_partition, input_delta_type, **ds_mock_kwargs
+    )
+    staged_delta.locator.stream_position = last_stream_position
+    metastore.commit_delta(staged_delta, **ds_mock_kwargs)
+    metastore.commit_partition(staged_partition, **ds_mock_kwargs)
+
+    source_table_stream_after_committed: Stream = metastore.get_stream(
+        namespace=source_namespace,
+        table_name=source_table_name,
+        table_version=source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Create the destination table
+    (
+        destination_table_namespace,
+        destination_table_name,
+        destination_table_version,
+    ) = create_destination_table_main(
+        sort_keys, partition_keys, input_deltas, ds_mock_kwargs
+    )
+
+    # Create the rebase table
+    (
+        rebase_table_namespace,
+        rebase_table_name,
+        rebase_table_version,
+    ) = create_rebase_table_main(
+        sort_keys, partition_keys, input_deltas, ds_mock_kwargs
+    )
+
+    rebasing_table_stream: Stream = metastore.get_stream(
+        namespace=rebase_table_namespace,
+        table_name=rebase_table_name,
+        table_version=rebase_table_version,
+        **ds_mock_kwargs,
+    )
+
+    staged_partition: Partition = metastore.stage_partition(
+        rebasing_table_stream,
+        converted_partition_values,
+        partition_scheme_id="default_partition_scheme" if partition_keys else None,
+        **ds_mock_kwargs,
+    )
+    staged_delta: Delta = metastore.stage_delta(
+        input_deltas, staged_partition, **ds_mock_kwargs
+    )
+    staged_delta.locator.stream_position = last_stream_position
+    metastore.commit_delta(staged_delta, **ds_mock_kwargs)
+    metastore.commit_partition(staged_partition, **ds_mock_kwargs)
+
+    # Get destination stream
+    destination_table_stream: Stream = metastore.get_stream(
+        namespace=destination_table_namespace,
+        table_name=destination_table_name,
+        table_version=destination_table_version,
+        **ds_mock_kwargs,
+    )
+
+    rebased_stream_after_committed: Stream = metastore.get_stream(
+        namespace=rebase_table_namespace,
+        table_name=rebase_table_name,
+        table_version=rebase_table_version,
+        **ds_mock_kwargs,
+    )
+
+    return (
+        source_table_stream_after_committed,
+        destination_table_stream,
+        rebased_stream_after_committed,
+    )
+
+
+def create_incremental_deltas_on_source_table_main(
+    source_namespace: str,
+    source_table_name: str,
+    source_table_version: str,
+    source_table_stream: Stream,
+    partition_values_param,
+    incremental_deltas: List[Tuple[pa.Table, DeltaType, Optional[Dict[str, str]]]],
+    ds_mock_kwargs: Optional[Dict[str, Any]] = None,
+) -> Tuple[PartitionLocator, Delta, int, bool]:
+    """
+    Main storage version of create_incremental_deltas_on_source_table
+    """
+    total_records = 0
+    has_delete_deltas = False
+    new_delta = None
+
+    # Convert partition values for partition lookup (same as in other helper functions)
+    converted_partition_values_for_lookup = partition_values_param
+    if (
+        partition_values_param
+        and source_table_stream.partition_scheme
+        and source_table_stream.partition_scheme.keys
+    ):
+        converted_partition_values_for_lookup = []
+
+        # Get partition field names from the storage partition scheme
+        storage_partition_keys = source_table_stream.partition_scheme.keys
+        partition_field_names = []
+
+        for storage_key in storage_partition_keys:
+            # Each storage PartitionKey has a 'key' property that contains FieldLocators
+            # Extract the field name from the first FieldLocator
+            field_name = storage_key.key[0] if storage_key.key else None
+            partition_field_names.append(field_name)
+
+        for i, value in enumerate(partition_values_param):
+            # For timestamp fields like 'region_id', we need to convert the timestamp string
+            if i < len(partition_field_names):
+                field_name = partition_field_names[i]
+
+                # Check if this is likely a timestamp field based on the value format
+                if isinstance(value, str) and "T" in value and value.endswith("Z"):
+                    # This looks like a timestamp string - convert it
+                    import pandas as pd
+
+                    ts = pd.to_datetime(value)
+                    # Convert to microseconds since epoch for PyArrow timestamp[us]
+                    converted_partition_values_for_lookup.append(
+                        int(ts.timestamp() * 1_000_000)
+                    )
+                elif isinstance(value, str) and value.isdigit():
+                    # This looks like an integer string
+                    converted_partition_values_for_lookup.append(int(value))
+                else:
+                    # Keep as-is
+                    converted_partition_values_for_lookup.append(value)
+            else:
+                converted_partition_values_for_lookup.append(value)
+
+    # Get the current partition to stage deltas against
+    try:
+        source_partition: Partition = metastore.get_partition(
+            source_table_stream.locator,
+            converted_partition_values_for_lookup,
+            **ds_mock_kwargs,
+        )
+    except Exception:
+        # If we can't get the partition, it might not exist yet. Try to create it.
+        # Stage a new partition if it doesn't exist
+        staged_partition: Partition = metastore.stage_partition(
+            source_table_stream,
+            converted_partition_values_for_lookup,
+            partition_scheme_id="default_partition_scheme"
+            if source_table_stream.partition_scheme
+            else None,
+            **ds_mock_kwargs,
+        )
+        # Commit the empty partition first
+        metastore.commit_partition(staged_partition, **ds_mock_kwargs)
+
+        # Now try to get it again
+        source_partition: Partition = metastore.get_partition(
+            source_table_stream.locator,
+            converted_partition_values_for_lookup,
+            **ds_mock_kwargs,
+        )
+
+    if source_partition is None:
+        raise ValueError(
+            f"Could not create or retrieve partition for values: {converted_partition_values_for_lookup}"
+        )
+
+    for delta_table, delta_type, properties_dict in incremental_deltas:
+        # Skip None deltas (empty incremental deltas)
+        if delta_table is None:
+            continue
+
+        total_records += len(delta_table)
+
+        if delta_type == DeltaType.DELETE:
+            has_delete_deltas = True
+
+        # Stage and commit the delta
+        staged_delta: Delta = metastore.stage_delta(
+            delta_table,
+            source_partition,
+            delta_type,
+            entry_params=properties_dict,
+            **ds_mock_kwargs,
+        )
+        new_delta = metastore.commit_delta(staged_delta, **ds_mock_kwargs)
+
+    # If all deltas were None, return None for new_delta
+    if new_delta is None:
+        return None, None, total_records, has_delete_deltas
+
+    # Get updated stream after deltas were committed
+    source_table_stream_after_committed: Stream = metastore.get_stream(
+        source_namespace,
+        source_table_name,
+        source_table_version,
+        **ds_mock_kwargs,
+    )
+
+    # Get updated partition after deltas were committed
+    source_partition_after_committed: Partition = metastore.get_partition(
+        source_table_stream_after_committed.locator,
+        converted_partition_values_for_lookup,
+        **ds_mock_kwargs,
+    )
+
+    return (
+        source_partition_after_committed.locator,
+        new_delta,
+        total_records,
+        has_delete_deltas,
+    )
+
+
+def get_compacted_delta_locator_from_partition(
+    partition_locator: PartitionLocator, deltacat_storage=None, **kwargs
 ):
-    from deltacat.storage import DeltaLocator
+    """
+    Get compacted delta locator from partition RoundCompletionInfo.
 
-    round_completion_info: RoundCompletionInfo = get_rcf(s3_resource, rcf_file_s3_uri)
+    Args:
+        partition_locator: Locator of the partition containing the RoundCompletionInfo
+        deltacat_storage: Storage implementation (defaults to metastore)
+        **kwargs: Additional arguments to pass to get_rci_from_partition (e.g., catalog)
 
-    compacted_delta_locator: DeltaLocator = (
-        round_completion_info.compacted_delta_locator
+    Returns:
+        DeltaLocator of the compacted delta
+    """
+    round_completion_info: RoundCompletionInfo = get_rci_from_partition(
+        partition_locator, deltacat_storage, **kwargs
     )
-    return compacted_delta_locator
+
+    if round_completion_info:
+        return round_completion_info.compacted_delta_locator
+    return None
 
 
 def offer_iso8601_timestamp_list(
@@ -325,3 +974,27 @@ def assert_compaction_audit_no_hash_bucket(
         for entry in audit_entries:
             assert entry is not None
     return True
+
+
+def read_audit_file(audit_file_path: str, catalog_root: str) -> Dict[str, Any]:
+    """
+    Read audit file from any filesystem.
+
+    Args:
+        audit_file_path: Relative path to the audit file from catalog root
+        catalog_root: Absolute path to the catalog root directory
+
+    Returns:
+        Dictionary containing audit data
+    """
+    from deltacat.utils.filesystem import resolve_path_and_filesystem
+    import json
+    import posixpath
+
+    # Resolve absolute path from relative audit path
+    absolute_path = posixpath.join(catalog_root, audit_file_path)
+
+    path, filesystem = resolve_path_and_filesystem(absolute_path)
+    with filesystem.open_input_stream(path) as stream:
+        content = stream.read().decode("utf-8")
+        return json.loads(content)