salesforce-data-customcode 6.0.8.dev1__py3-none-any.whl → 6.1.0.dev1__py3-none-any.whl

datacustomcode/__init__.py

@@ -23,6 +23,7 @@ __all__ = [
     "QueryAPIDataCloudReader",
     "SparkEinsteinPredictions",
     "SparkLLMGateway",
+    "StreamingClient",
     "einstein_predict_col",
     "llm_gateway_generate_text_col",
 ]
@@ -34,6 +35,10 @@ def __getattr__(name: str):
         from datacustomcode.client import Client
 
         return Client
+    elif name == "StreamingClient":
+        from datacustomcode.client import StreamingClient
+
+        return StreamingClient
     elif name == "AuthType":
         from datacustomcode.credentials import AuthType
 

datacustomcode/cli.py

@@ -283,10 +283,19 @@ def deploy(
 )
 @click.option(
     "--use-in-feature",
-    default="SearchIndexChunking",
-    help="Feature where this function will be used (only applicable for function).",
+    "-u",
+    default=None,
+    help=(
+        "Invoke option for this package. For scripts: 'BatchTransform' "
+        "(default) or 'StreamingTransform'. For functions: 'SearchIndexChunking'."
+    ),
 )
 def init(directory: str, code_type: str, use_in_feature: Optional[str]):
+    from datacustomcode.constants import (
+        SCRIPT_USE_IN_FEATURE_BATCH,
+        SCRIPT_USE_IN_FEATURE_OPTIONS,
+        SCRIPT_USE_IN_FEATURE_STREAMING,
+    )
     from datacustomcode.scan import (
         dc_config_json_from_file,
         update_config,
@@ -294,9 +303,23 @@ def init(directory: str, code_type: str, use_in_feature: Optional[str]):
     )
     from datacustomcode.template import copy_function_template, copy_script_template
 
+    streaming = False
+    if code_type == "script":
+        use_in_feature = use_in_feature or SCRIPT_USE_IN_FEATURE_BATCH
+        if use_in_feature not in SCRIPT_USE_IN_FEATURE_OPTIONS:
+            click.secho(
+                f"Error: Invalid --use-in-feature '{use_in_feature}' for a "
+                f"script. Valid options: {', '.join(SCRIPT_USE_IN_FEATURE_OPTIONS)}.",
+                fg="red",
+            )
+            raise click.Abort()
+        streaming = use_in_feature == SCRIPT_USE_IN_FEATURE_STREAMING
+    else:
+        use_in_feature = use_in_feature or "SearchIndexChunking"
+
     click.echo("Copying template to " + click.style(directory, fg="blue", bold=True))
     if code_type == "script":
-        copy_script_template(directory)
+        copy_script_template(directory, streaming=streaming)
     elif code_type == "function":
         copy_function_template(directory, use_in_feature)
     entrypoint_path = os.path.join(directory, PAYLOAD_DIR, ENTRYPOINT_FILE)
@@ -306,7 +329,9 @@ def init(directory: str, code_type: str, use_in_feature: Optional[str]):
     sdk_config = {"type": code_type}
     write_sdk_config(directory, sdk_config)
 
-    config_json = dc_config_json_from_file(entrypoint_path, code_type)
+    config_json = dc_config_json_from_file(
+        entrypoint_path, code_type, streaming=streaming
+    )
     with open(config_location, "w") as f:
         json.dump(config_json, f, indent=2)
 

datacustomcode/client.py

@@ -21,7 +21,9 @@ from typing import (
     ClassVar,
     Dict,
     Optional,
+    TypeVar,
     Union,
+    cast,
 )
 
 from datacustomcode.config import config
@@ -34,7 +36,12 @@ from datacustomcode.spark.default import DefaultSparkSessionProvider
 if TYPE_CHECKING:
     from pathlib import Path
 
-    from pyspark.sql import Column, DataFrame as PySparkDataFrame
+    from pyspark.sql import (
+        Column,
+        DataFrame as PySparkDataFrame,
+        SparkSession,
+    )
+    from pyspark.sql.streaming import StreamingQuery
 
     from datacustomcode.einstein_predictions.spark_base import SparkEinsteinPredictions
     from datacustomcode.einstein_predictions.types import PredictionType
@@ -44,6 +51,40 @@ if TYPE_CHECKING:
     from datacustomcode.spark.base import BaseSparkSessionProvider
 
 
+def _streaming_source_name() -> str:
+    """Return the streaming transform's read-source name.
+
+    Resolved from ``config.streaming_source``, which ``run_entrypoint``
+    populates from config.json's ``streamingSource`` field.
+
+    Raises:
+        RuntimeError: If no ``streaming_source`` has been configured (e.g. the
+            transform's config.json has no ``streamingSource`` field).
+    """
+    source = config.streaming_source
+    if not source:
+        raise RuntimeError(
+            "No streaming source configured. A streaming transform must declare "
+            "its read source in config.json under 'streamingSource'."
+        )
+    return source
+
+
+def _active_client() -> "_BaseClient":
+    """Return the client backing the module-level Spark column helpers.
+
+    Prefers an already-initialized singleton so a streaming job reuses its
+    :class:`StreamingClient` (and a batch job its :class:`Client`) rather than
+    forcing an unrelated client into existence. Falls back to building the
+    batch :class:`Client` when neither has been created yet.
+    """
+    if Client._instance is not None:
+        return Client._instance
+    if StreamingClient._instance is not None:
+        return StreamingClient._instance
+    return Client()
+
+
 def _build_spark_llm_gateway() -> "SparkLLMGateway":
     """Instantiate the SDK-configured :class:`SparkLLMGateway`.
 
@@ -99,7 +140,7 @@ def llm_gateway_generate_text_col(
         the generated text; on failure, ``status == "ERROR"`` and the
         ``error_*`` fields carry diagnostic detail.
     """
-    gateway = Client()._get_spark_llm_gateway()
+    gateway = _active_client()._get_spark_llm_gateway()
     return gateway.llm_gateway_generate_text_col(template, values, model_id=model_id)
 
 
@@ -161,7 +202,7 @@ def einstein_predict_col(
         the JSON-serialized prediction payload; on failure, ``status ==
         "ERROR"`` and the ``error_*`` fields carry diagnostic detail.
     """
-    predictions = Client()._get_spark_einstein_predictions()
+    predictions = _active_client()._get_spark_einstein_predictions()
     return predictions.einstein_predict_col(
         model_api_name, prediction_type, features, settings=settings
     )
@@ -205,39 +246,39 @@ class DataCloudAccessLayerException(Exception):
         return msg
 
 
-class Client:
-    """Entrypoint for accessing DataCloud objects.
+_ClientT = TypeVar("_ClientT", bound="_BaseClient")
+
+
+class _BaseClient:
+    """Shared machinery for the Data Cloud client singletons.
 
-    This is the object used to access Data Cloud DLOs and DMOs. Accessing DLOs/DMOs
-    are tracked and will throw an exception if they are mixed. In other words, you
-    can read from DLOs and write to DLOs, read from DMOs and write to DMOs, but you
-    cannot read from DLOs and write to DMOs or read from DMOs and write to DLOs.
-    Furthermore you cannot mix during merging tables. This class is a singleton to
-    prevent accidental mixing of DLOs and DMOs.
+    Holds the wiring common to :class:`Client` (batch) and
+    :class:`StreamingClient`
 
-    You can provide custom readers and writers to the client for advanced use
-    cases, but this is not recommended for testing as they may result in unexpected
-    behavior once deployed to Data Cloud. By default, the client intercepts all
-    read/write operations and mocks access to Data Cloud. For example, during
-    writing, we print to the console instead of writing to Data Cloud.
+    This base class is not meant to be instantiated directly; use
+    :class:`Client` or :class:`StreamingClient`.
 
     Args:
-        finder: Find a file path
         reader: A custom reader to use for reading Data Cloud objects.
         writer: A custom writer to use for writing Data Cloud objects.
+        spark_provider: Optional custom :class:`BaseSparkSessionProvider`.
         spark_llm_gateway: Optional custom :class:`SparkLLMGateway`.
         spark_einstein_predictions: Optional custom
             :class:`SparkEinsteinPredictions`.
-
-    Example:
-    >>> client = Client()
-    >>> file_path = client.find_file_path("data.csv")
-    >>> dlo = client.read_dlo("my_dlo")
-    >>> client.write_to_dmo("my_dmo", dlo)
-    >>> answer = client.llm_gateway_generate_text("Generate a greeting message")
     """
 
-    _instance: ClassVar[Optional[Client]] = None
+    # Each concrete subclass gets its own ``_instance`` slot: reads fall through
+    # to this base default of ``None``, but ``cls._instance = ...`` in __new__
+    # always writes to the subclass, so ``Client`` and ``StreamingClient`` never
+    # share an instance.
+    _instance: ClassVar[Optional[_BaseClient]] = None
+    # Process-wide Spark session shared across BOTH client types. Unlike
+    # ``_instance``, this is written via ``_BaseClient._shared_spark`` (never
+    # ``cls._shared_spark``), so the slot lives on the base class and a
+    # ``Client`` and a ``StreamingClient`` in the same process reuse one session
+    # — and therefore one underlying connection — instead of opening two
+    # containing differing state
+    _shared_spark: ClassVar[Optional[SparkSession]] = None
     _reader: BaseDataCloudReader
     _writer: BaseDataCloudWriter
     _file: DefaultFindFilePath
@@ -247,37 +288,44 @@ class Client:
     _code_type: str
 
     def __new__(
-        cls,
+        cls: type[_ClientT],
         reader: Optional[BaseDataCloudReader] = None,
         writer: Optional[BaseDataCloudWriter] = None,
         spark_provider: Optional[BaseSparkSessionProvider] = None,
         spark_llm_gateway: Optional[SparkLLMGateway] = None,
         spark_einstein_predictions: Optional[SparkEinsteinPredictions] = None,
         code_type: str = "script",
-    ) -> Client:
+    ) -> _ClientT:
 
         if cls._instance is None:
-            cls._instance = super().__new__(cls)
-            cls._instance._spark_llm_gateway = spark_llm_gateway
-            cls._instance._spark_einstein_predictions = spark_einstein_predictions
+            instance = super().__new__(cls)
+            instance._spark_llm_gateway = spark_llm_gateway
+            instance._spark_einstein_predictions = spark_einstein_predictions
             # Initialize Readers and Writers from config
             # and/or provided reader and writer
             if reader is None or writer is None:
-                # We need a spark because we will initialize readers and writers
-                if config.spark_config is None:
-                    raise ValueError(
-                        "Spark config is required when reader/writer is not provided"
-                    )
-
-                provider: BaseSparkSessionProvider
-                if spark_provider is not None:
-                    provider = spark_provider
-                elif config.spark_provider_config is not None:
-                    provider = config.spark_provider_config.to_object()
+                # We need a spark because we will initialize readers and writers.
+                # Reuse the process-wide session if one client already built it,
+                # so a Client and a StreamingClient share a single connection.
+                if _BaseClient._shared_spark is not None:
+                    spark = _BaseClient._shared_spark
                 else:
-                    provider = DefaultSparkSessionProvider()
-
-                spark = provider.get_session(config.spark_config)
+                    if config.spark_config is None:
+                        raise ValueError(
+                            "Spark config is required when reader/writer is not "
+                            "provided"
+                        )
+
+                    provider: BaseSparkSessionProvider
+                    if spark_provider is not None:
+                        provider = spark_provider
+                    elif config.spark_provider_config is not None:
+                        provider = config.spark_provider_config.to_object()
+                    else:
+                        provider = DefaultSparkSessionProvider()
+
+                    spark = provider.get_session(config.spark_config)
+                    _BaseClient._shared_spark = spark
 
             if config.reader_config is None and reader is None:
                 raise ValueError(
@@ -300,66 +348,17 @@ class Client:
             else:
                 writer_init = writer
 
-            cls._instance._reader = reader_init
-            cls._instance._writer = writer_init
-            cls._instance._file = DefaultFindFilePath()
-            cls._instance._data_layer_history = {
+            instance._reader = reader_init
+            instance._writer = writer_init
+            instance._file = DefaultFindFilePath()
+            instance._data_layer_history = {
                 DataCloudObjectType.DLO: set(),
                 DataCloudObjectType.DMO: set(),
             }
-        elif (reader is not None or writer is not None) and cls._instance is not None:
+            cls._instance = instance
+        elif reader is not None or writer is not None:
             raise ValueError("Cannot set reader or writer after client is initialized")
-        return cls._instance
-
-    def read_dlo(self, name: str) -> PySparkDataFrame:
-        """Read a DLO from Data Cloud.
-
-        Args:
-            name: The name of the DLO to read.
-
-        Returns:
-            A PySpark DataFrame containing the DLO data.
-        """
-        self._record_dlo_access(name)
-        return self._reader.read_dlo(name)  # type: ignore[no-any-return]
-
-    def read_dmo(self, name: str) -> PySparkDataFrame:
-        """Read a DMO from Data Cloud.
-
-        Args:
-            name: The name of the DMO to read.
-
-        Returns:
-            A PySpark DataFrame containing the DMO data.
-        """
-        self._record_dmo_access(name)
-        return self._reader.read_dmo(name)  # type: ignore[no-any-return]
-
-    def write_to_dlo(
-        self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode, **kwargs
-    ) -> None:
-        """Write a PySpark DataFrame to a DLO in Data Cloud.
-
-        Args:
-            name: The name of the DLO to write to.
-            dataframe: The PySpark DataFrame to write.
-            write_mode: The write mode to use for writing to the DLO.
-        """
-        self._validate_data_layer_history_does_not_contain(DataCloudObjectType.DMO)
-        return self._writer.write_to_dlo(name, dataframe, write_mode, **kwargs)  # type: ignore[no-any-return]
-
-    def write_to_dmo(
-        self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode, **kwargs
-    ) -> None:
-        """Write a PySpark DataFrame to a DMO in Data Cloud.
-
-        Args:
-            name: The name of the DMO to write to.
-            dataframe: The PySpark DataFrame to write.
-            write_mode: The write mode to use for writing to the DMO.
-        """
-        self._validate_data_layer_history_does_not_contain(DataCloudObjectType.DLO)
-        return self._writer.write_to_dmo(name, dataframe, write_mode, **kwargs)  # type: ignore[no-any-return]
+        return cast(_ClientT, cls._instance)
 
     def find_file_path(self, file_name: str) -> Path:
         """Resolve a bundled file shipped in the package to an absolute path.
@@ -487,3 +486,115 @@ class Client:
 
     def _record_dmo_access(self, name: str) -> None:
         self._data_layer_history[DataCloudObjectType.DMO].add(name)
+
+
+class Client(_BaseClient):
+    """Entrypoint for batch access to Data Cloud objects.
+
+    This is the object used to read and write bounded snapshots of Data Cloud
+    DLOs and DMOs.
+    """
+
+    _instance: ClassVar[Optional[Client]] = None
+
+    def read_dlo(self, name: str) -> PySparkDataFrame:
+        """Read a DLO from Data Cloud.
+
+        Args:
+            name: The name of the DLO to read.
+
+        Returns:
+            A PySpark DataFrame containing the DLO data.
+        """
+        self._record_dlo_access(name)
+        return self._reader.read_dlo(name)  # type: ignore[no-any-return]
+
+    def read_dmo(self, name: str) -> PySparkDataFrame:
+        """Read a DMO from Data Cloud.
+
+        Args:
+            name: The name of the DMO to read.
+
+        Returns:
+            A PySpark DataFrame containing the DMO data.
+        """
+        self._record_dmo_access(name)
+        return self._reader.read_dmo(name)  # type: ignore[no-any-return]
+
+    def write_to_dlo(
+        self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode, **kwargs
+    ) -> None:
+        """Write a PySpark DataFrame to a DLO in Data Cloud.
+
+        Args:
+            name: The name of the DLO to write to.
+            dataframe: The PySpark DataFrame to write.
+            write_mode: The write mode to use for writing to the DLO.
+        """
+        self._validate_data_layer_history_does_not_contain(DataCloudObjectType.DMO)
+        return self._writer.write_to_dlo(name, dataframe, write_mode, **kwargs)  # type: ignore[no-any-return]
+
+    def write_to_dmo(
+        self, name: str, dataframe: PySparkDataFrame, write_mode: WriteMode, **kwargs
+    ) -> None:
+        """Write a PySpark DataFrame to a DMO in Data Cloud.
+
+        Args:
+            name: The name of the DMO to write to.
+            dataframe: The PySpark DataFrame to write.
+            write_mode: The write mode to use for writing to the DMO.
+        """
+        self._validate_data_layer_history_does_not_contain(DataCloudObjectType.DLO)
+        return self._writer.write_to_dmo(name, dataframe, write_mode, **kwargs)  # type: ignore[no-any-return]
+
+
+class StreamingClient(_BaseClient):
+    """Entrypoint for streaming (``DELTA_SYNC``) access to Data Cloud objects.
+
+    This is the streaming counterpart to :class:`Client`. Instead of reading and
+    writing bounded snapshots, it reads a DLO/DMO change feed as a streaming
+    DataFrame and writes the transformed stream back via a ``StreamingQuery``.
+    """
+
+    _instance: ClassVar[Optional[StreamingClient]] = None
+
+    def read_dlo_deltas(self) -> PySparkDataFrame:
+        """Read the streaming change feed (deltas) for a DLO from Data Cloud.
+
+        For use in a streaming (``DELTA_SYNC``) BYOC transform. Returns a
+        streaming DataFrame whose rows carry the change-feed metadata columns
+        (``_record_type``, ``_commit_*``) alongside the source columns.
+
+        Returns:
+            A streaming PySpark DataFrame over the DLO change feed.
+        """
+        self._record_dlo_access(_streaming_source_name())
+        return self._reader.read_dlo_deltas()  # type: ignore[no-any-return]
+
+    def read_dmo_deltas(self) -> PySparkDataFrame:
+        """Read the streaming change feed (deltas) for a DMO from Data Cloud.
+
+        Returns:
+            A streaming PySpark DataFrame over the DMO change feed.
+        """
+        self._record_dmo_access(_streaming_source_name())
+        return self._reader.read_dmo_deltas()  # type: ignore[no-any-return]
+
+    def write_dlo_deltas(
+        self, name: str, dataframe: PySparkDataFrame, **kwargs
+    ) -> StreamingQuery:
+        """Write a streaming DataFrame of deltas to a DLO in Data Cloud.
+
+        Starts a streaming query that writes each micro-batch to the
+        target DLO and returns the  ``StreamingQuery`` handle; the caller
+        typically calls ``query.awaitTermination()``.
+
+        Args:
+            name: The name of the DLO to write to.
+            dataframe: The streaming PySpark DataFrame to write.
+
+        Returns:
+            The started ``StreamingQuery``.
+        """
+        self._validate_data_layer_history_does_not_contain(DataCloudObjectType.DMO)
+        return self._writer.write_dlo_deltas(name, dataframe, **kwargs)  # type: ignore[no-any-return]

datacustomcode/config.py

@@ -89,6 +89,9 @@ class ClientConfig(BaseConfig):
     spark_provider_config: Union[
         SparkProviderConfig[BaseSparkSessionProvider], None
     ] = None
+    # Source object name for a streaming (DELTA_SYNC) transform, populated by
+    # ``run_entrypoint`` from config.json's ``streamingSource`` field
+    streaming_source: Union[str, None] = None
 
     def update(self, other: ClientConfig) -> ClientConfig:
         """Merge this ClientConfig with another, respecting force flags.
@@ -116,6 +119,8 @@ class ClientConfig(BaseConfig):
         self.spark_provider_config = merge(
             self.spark_provider_config, other.spark_provider_config
         )
+        if other.streaming_source is not None:
+            self.streaming_source = other.streaming_source
         return self
 
 

datacustomcode/constants.py

@@ -38,6 +38,14 @@ USE_IN_FEATURE_MAPPING_FOR_CONNECT_API = {
     "SearchIndexChunking": "UnstructuredChunking",
 }
 
+# Script (data transform) invoke options
+SCRIPT_USE_IN_FEATURE_BATCH = "BatchTransform"
+SCRIPT_USE_IN_FEATURE_STREAMING = "StreamingTransform"
+SCRIPT_USE_IN_FEATURE_OPTIONS = [
+    SCRIPT_USE_IN_FEATURE_BATCH,
+    SCRIPT_USE_IN_FEATURE_STREAMING,
+]
+
 # Pydantic request/response type names to feature names
 REQUEST_TYPE_TO_FEATURE = {
     "SearchIndexChunkingV1Request": "SearchIndexChunking",

datacustomcode/deploy.py

@@ -14,6 +14,7 @@
 # limitations under the License.
 from __future__ import annotations
 
+import copy
 from html import unescape
 import json
 import os
@@ -41,6 +42,7 @@ from datacustomcode.scan import find_base_directory, get_package_type
 
 DATA_CUSTOM_CODE_PATH = "services/data/v63.0/ssot/data-custom-code"
 DATA_TRANSFORMS_PATH = "services/data/v63.0/ssot/data-transforms"
+DATA_CUSTOM_CODE_INVOKE_OPTIONS_PATH = "services/data/v67.0/ssot/data-custom-code"
 WAIT_FOR_DEPLOYMENT_TIMEOUT = 3000
 
 # Available compute types for Data Cloud deployments.
@@ -108,6 +110,7 @@ class CodeExtensionMetadata(BaseModel):
     computeType: str
     codeType: str
     functionInvokeOptions: Union[list[str], None] = None
+    invokeOptions: Union[list[str], None] = None
 
     def __init__(self, **data):
         name = data.get("name", "")
@@ -200,7 +203,14 @@ def create_deployment(
     access_token: AccessTokenResponse, metadata: CodeExtensionMetadata
 ) -> CreateDeploymentResponse:
     """Create a custom code deployment in the DataCloud."""
-    url = _join_strip_url(access_token.instance_url, DATA_CUSTOM_CODE_PATH)
+    # invokeOptions only binds at v67.0; route there when it is set so the
+    # option isn't silently dropped. Everything else stays on v63.0.
+    code_custom_code_path = (
+        DATA_CUSTOM_CODE_INVOKE_OPTIONS_PATH
+        if metadata.invokeOptions
+        else DATA_CUSTOM_CODE_PATH
+    )
+    url = _join_strip_url(access_token.instance_url, code_custom_code_path)
     body = dict[str, Any](
         {
             "label": metadata.name,
@@ -213,6 +223,8 @@ def create_deployment(
     )
     if metadata.functionInvokeOptions:
         body["functionInvokeOptions"] = metadata.functionInvokeOptions
+    if metadata.invokeOptions:
+        body["invokeOptions"] = metadata.invokeOptions
     logger.debug(f"Creating deployment {metadata.name}...")
     try:
         response = _make_api_call(
@@ -388,6 +400,30 @@ class DataTransformConfig(BaseConfig):
     dataspace: str
     permissions: Permissions
     dataObjects: Optional[list[DataObject]] = None
+    streamingSource: Optional[StreamingSource] = None
+
+    @property
+    def is_streaming(self) -> bool:
+        return self.streamingSource is not None
+
+    @model_validator(mode="after")
+    def _validate_layers(self) -> "DataTransformConfig":
+        read_is_dlo = isinstance(self.permissions.read, DloPermission)
+        write_is_dlo = isinstance(self.permissions.write, DloPermission)
+        if self.is_streaming:
+            if not write_is_dlo:
+                raise ValueError(
+                    "A streaming transform must write to a DLO "
+                    "(permissions.write must be a 'dlo' entry)."
+                )
+        elif read_is_dlo != write_is_dlo:
+            raise ValueError(
+                "permissions.read and permissions.write must both reference "
+                "DLOs or both reference DMOs (got "
+                f"read={type(self.permissions.read).__name__}, "
+                f"write={type(self.permissions.write).__name__})"
+            )
+        return self
 
 
 class FunctionConfig(BaseConfig):
@@ -402,23 +438,15 @@ class DmoPermission(BaseModel):
     dmo: list[str]
 
 
+class StreamingSource(BaseModel):
+    type: str
+    name: str
+
+
 class Permissions(BaseModel):
     read: Union[DloPermission, DmoPermission]
     write: Union[DloPermission, DmoPermission]
 
-    @model_validator(mode="after")
-    def _no_mixed_layers(self) -> "Permissions":
-        read_is_dlo = isinstance(self.read, DloPermission)
-        write_is_dlo = isinstance(self.write, DloPermission)
-        if read_is_dlo != write_is_dlo:
-            raise ValueError(
-                "permissions.read and permissions.write must both reference "
-                "DLOs or both reference DMOs (got "
-                f"read={type(self.read).__name__}, "
-                f"write={type(self.write).__name__})"
-            )
-        return self
-
 
 def _permission_entries(perm: Union[DloPermission, DmoPermission]) -> list[str]:
     """Return the list of object names regardless of layer (DLO or DMO)."""
@@ -490,7 +518,9 @@ def create_data_transform(
 ) -> dict:
     """Create a data transform in the DataCloud."""
     script_name = metadata.name
-    request_hydrated = DATA_TRANSFORM_REQUEST_TEMPLATE.copy()
+    # Deep copy: the template's nested nodes/sources/macros dicts would
+    # otherwise be shared across calls and accumulate entries between deploys.
+    request_hydrated = copy.deepcopy(DATA_TRANSFORM_REQUEST_TEMPLATE)
 
     # Add nodes for each write entry (DLO or DMO)
     for i, name in enumerate(
@@ -533,7 +563,7 @@ def create_data_transform(
         "definition": definition,
         "label": f"{metadata.name}",
         "name": f"{metadata.name}",
-        "type": "BATCH",
+        "type": "STREAMING" if data_transform_config.is_streaming else "BATCH",
         "dataSpaceName": data_transform_config.dataspace,
     }
 
@@ -616,9 +646,18 @@ def deploy_full(
     callback=None,
 ) -> AccessTokenResponse:
     """Deploy a data transform in the DataCloud."""
+    from datacustomcode.constants import SCRIPT_USE_IN_FEATURE_STREAMING
+
     # prepare payload
     config = get_config(directory)
 
+    if (
+        isinstance(config, DataTransformConfig)
+        and config.is_streaming
+        and not metadata.invokeOptions
+    ):
+        metadata.invokeOptions = [SCRIPT_USE_IN_FEATURE_STREAMING]
+
     # create deployment and upload payload
     deployment = create_deployment(access_token, metadata)
     zip(directory, docker_network, metadata.codeType)