dataproc-spark-connect 1.0.0rc2__py2.py3-none-any.whl → 1.0.0rc3__py2.py3-none-any.whl

{dataproc_spark_connect-1.0.0rc2.dist-info → dataproc_spark_connect-1.0.0rc3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dataproc-spark-connect
-Version: 1.0.0rc2
+Version: 1.0.0rc3
 Summary: Dataproc client library for Spark Connect
 Home-page: https://github.com/GoogleCloudDataproc/dataproc-spark-connect-python
 Author: Google LLC

{dataproc_spark_connect-1.0.0rc2.dist-info → dataproc_spark_connect-1.0.0rc3.dist-info}/RECORD

@@ -1,13 +1,13 @@
-dataproc_spark_connect-1.0.0rc2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+dataproc_spark_connect-1.0.0rc3.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
 google/cloud/dataproc_spark_connect/__init__.py,sha256=dIqHNWVWWrSuRf26x11kX5e9yMKSHCtmI_GBj1-FDdE,1101
 google/cloud/dataproc_spark_connect/environment.py,sha256=l1wWiHMHtBQ9YonE-kHTpaZlN9vLE4fyJSTn7RZP6kA,2503
 google/cloud/dataproc_spark_connect/exceptions.py,sha256=WF-qdzgdofRwILCriIkjjsmjObZfF0P3Ecg4lv-Hmec,968
 google/cloud/dataproc_spark_connect/pypi_artifacts.py,sha256=gd-VMwiVP-EJuPp9Vf9Shx8pqps3oSKp0hBcSSZQS-A,1575
-google/cloud/dataproc_spark_connect/session.py,sha256=FdJI_F9k6EfIvlgC1-f-Qb_Uwg9SmkIyWhpNZlqGQhw,40405
+google/cloud/dataproc_spark_connect/session.py,sha256=u2QxRLn84EMdpFnv9pI6Y7F8OFQ2mqCllb6AsLd73eo,42402
 google/cloud/dataproc_spark_connect/client/__init__.py,sha256=6hCNSsgYlie6GuVpc5gjFsPnyeMTScTpXSPYqp1fplY,615
 google/cloud/dataproc_spark_connect/client/core.py,sha256=GRc4OCTBvIvdagjxOPoDO22vLtt8xDSerdREMRDeUBY,4659
 google/cloud/dataproc_spark_connect/client/proxy.py,sha256=qUZXvVY1yn934vE6nlO495XUZ53AUx9O74a9ozkGI9U,8976
-dataproc_spark_connect-1.0.0rc2.dist-info/METADATA,sha256=o2vfu5NRn2Pb0N7cavrBm2OLwP_LXQBVrclNjEtb9Do,3468
-dataproc_spark_connect-1.0.0rc2.dist-info/WHEEL,sha256=JNWh1Fm1UdwIQV075glCn4MVuCRs0sotJIq-J6rbxCU,109
-dataproc_spark_connect-1.0.0rc2.dist-info/top_level.txt,sha256=_1QvSJIhFAGfxb79D6DhB7SUw2X6T4rwnz_LLrbcD3c,7
-dataproc_spark_connect-1.0.0rc2.dist-info/RECORD,,
+dataproc_spark_connect-1.0.0rc3.dist-info/METADATA,sha256=yUMy-S__bGisjsxduDg_VmJzF3ZKqCybaNUwn5DMymo,3468
+dataproc_spark_connect-1.0.0rc3.dist-info/WHEEL,sha256=JNWh1Fm1UdwIQV075glCn4MVuCRs0sotJIq-J6rbxCU,109
+dataproc_spark_connect-1.0.0rc3.dist-info/top_level.txt,sha256=_1QvSJIhFAGfxb79D6DhB7SUw2X6T4rwnz_LLrbcD3c,7
+dataproc_spark_connect-1.0.0rc3.dist-info/RECORD,,

google/cloud/dataproc_spark_connect/session.py

@@ -24,6 +24,7 @@ import threading
 import time
 import uuid
 import tqdm
+from packaging import version
 from tqdm import tqdm as cli_tqdm
 from tqdm.notebook import tqdm as notebook_tqdm
 from types import MethodType
@@ -106,6 +107,7 @@ class DataprocSparkSession(SparkSession):
     """
 
     _DEFAULT_RUNTIME_VERSION = "3.0"
+    _MIN_RUNTIME_VERSION = "3.0"
 
     _active_s8s_session_uuid: ClassVar[Optional[str]] = None
     _project_id = None
@@ -254,6 +256,7 @@ class DataprocSparkSession(SparkSession):
             session._register_progress_execution_handler()
 
             DataprocSparkSession._set_default_and_active_session(session)
+
             return session
 
         def __create(self) -> "DataprocSparkSession":
@@ -268,6 +271,9 @@ class DataprocSparkSession(SparkSession):
 
                 dataproc_config: Session = self._get_dataproc_config()
 
+                # Check runtime version compatibility before creating session
+                self._check_runtime_compatibility(dataproc_config)
+
                 session_id = self.generate_dataproc_session_id()
                 dataproc_config.name = f"projects/{self._project_id}/locations/{self._region}/sessions/{session_id}"
                 logger.debug(
@@ -599,6 +605,43 @@ class DataprocSparkSession(SparkSession):
                         stacklevel=3,
                     )
 
+        def _check_runtime_compatibility(self, dataproc_config):
+            """Check if runtime version 3.0 client is compatible with older runtime versions.
+
+            Runtime version 3.0 clients do not support older runtime versions (pre-3.0).
+            There is no backward or forward compatibility between different runtime versions.
+
+            Args:
+                dataproc_config: The Session configuration containing runtime version
+
+            Raises:
+                DataprocSparkConnectException: If server is using pre-3.0 runtime version
+            """
+            runtime_version = dataproc_config.runtime_config.version
+
+            if not runtime_version:
+                return
+
+            logger.debug(f"Detected server runtime version: {runtime_version}")
+
+            # Parse runtime version to check if it's below minimum supported version
+            try:
+                server_version = version.parse(runtime_version)
+                min_version = version.parse(
+                    DataprocSparkSession._MIN_RUNTIME_VERSION
+                )
+
+                if server_version < min_version:
+                    raise DataprocSparkConnectException(
+                        f"Specified {runtime_version} Dataproc Runtime version is not supported, "
+                        f"use {DataprocSparkSession._MIN_RUNTIME_VERSION} version or higher."
+                    )
+            except version.InvalidVersion:
+                # If we can't parse the version, log a warning but continue
+                logger.warning(
+                    f"Could not parse runtime version: {runtime_version}"
+                )
+
         def _display_view_session_details_button(self, session_id):
             try:
                 session_url = f"https://console.cloud.google.com/dataproc/interactive/sessions/{session_id}/locations/{self._region}?project={self._project_id}"
@@ -789,6 +832,11 @@ class DataprocSparkSession(SparkSession):
         """
 
     def _display_operation_link(self, operation_id: str):
+        # Don't print per-operation Spark UI link for non-interactive (despite
+        # Ipython or non-IPython)
+        if not environment.is_interactive():
+            return
+
         assert all(
             [
                 operation_id is not None,
@@ -804,12 +852,13 @@ class DataprocSparkSession(SparkSession):
             f"associatedSqlOperationId={operation_id}?project={self._project_id}"
         )
 
+        if environment.is_interactive_terminal():
+            print(f"Spark Query: {url}")
+            return
+
         try:
             from IPython.display import display, HTML
-            from IPython.core.interactiveshell import InteractiveShell
 
-            if not InteractiveShell.initialized():
-                return
             html_element = f"""
               <div>
                   <p><a href="{url}">Spark Query</a> (Operation: {operation_id})</p>