Added classes required for telemetry

src/databricks/sql/client.py

@@ -1,6 +1,5 @@
 import time
 from typing import Dict, Tuple, List, Optional, Any, Union, Sequence
-
 import pandas
 
 try:
@@ -234,6 +233,13 @@ def read(self) -> Optional[OAuthToken]:
             server_hostname, **kwargs
         )
 
+        self.server_telemetry_enabled = True
+        self.client_telemetry_enabled = kwargs.get("enable_telemetry", False)
+        self.telemetry_enabled = (
+            self.client_telemetry_enabled and self.server_telemetry_enabled
+        )
+        telemetry_batch_size = kwargs.get("telemetry_batch_size", 200)
+
         user_agent_entry = kwargs.get("user_agent_entry")
         if user_agent_entry is None:
             user_agent_entry = kwargs.get("_user_agent_entry")
@@ -425,6 +431,9 @@ def _close(self, close_cursors=True) -> None:
 
         self.open = False
 
+        if hasattr(self, "telemetry_client"):
+            self.telemetry_client.close()
+
     def commit(self):
         """No-op because Databricks does not support transactions"""
         pass

src/databricks/sql/telemetry/DriverConnectionParameters.py

@@ -0,0 +1,41 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.HostDetails import HostDetails
+from databricks.sql.telemetry.enums.AuthMech import AuthMech
+from databricks.sql.telemetry.enums.AuthFlow import AuthFlow
+from databricks.sql.telemetry.enums.DatabricksClientType import DatabricksClientType
+
+
+@dataclass
+class DriverConnectionParameters:
+    http_path: str
+    mode: DatabricksClientType
+    host_info: HostDetails
+    auth_mech: AuthMech
+    auth_flow: AuthFlow
+    auth_scope: str
+    discovery_url: str
+    allowed_volume_ingestion_paths: str
+    azure_tenant_id: str
+    socket_timeout: int
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Part of TelemetryEvent
+# DriverConnectionParameters connectionParams = new DriverConnectionParameters(
+#     httpPath = " /sql/1.0/endpoints/1234567890abcdef",
+#     driverMode = "THRIFT",
+#     hostDetails = new HostDetails(
+#         hostUrl = "https://my-workspace.cloud.databricks.com",
+#         port = 443
+#     ),
+#     authMech = "OAUTH",
+#     authFlow = "AZURE_MANAGED_IDENTITIES",
+#     authScope = "sql",
+#     discoveryUrl = "https://example-url",
+#     allowedVolumeIngestionPaths = "[]",
+#     azureTenantId = "1234567890abcdef",
+#     socketTimeout = 10000
+# )

src/databricks/sql/telemetry/DriverErrorInfo.py

@@ -0,0 +1,23 @@
+import json
+from dataclasses import dataclass, asdict
+
+
+@dataclass
+class DriverErrorInfo:
+    error_name: str
+    stack_trace: str
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Required for ErrorLogs
+# DriverErrorInfo errorInfo = new DriverErrorInfo(
+#     errorName="CONNECTION_ERROR",
+#     stackTrace="Connection failure while using the Databricks SQL Python connector. Failed to connect to server: https://my-workspace.cloud.databricks.com\n" +
+#               "databricks.sql.exc.OperationalError: Connection refused: connect\n" +
+#               "at databricks.sql.thrift_backend.ThriftBackend.make_request(ThriftBackend.py:329)\n" +
+#               "at databricks.sql.thrift_backend.ThriftBackend.attempt_request(ThriftBackend.py:366)\n" +
+#               "at databricks.sql.thrift_backend.ThriftBackend.open_session(ThriftBackend.py:575)\n" +
+#               "at databricks.sql.client.Connection.__init__(client.py:69)\n" +
+#               "at databricks.sql.client.connect(connection.py:123)")

src/databricks/sql/telemetry/DriverSystemConfiguration.py

@@ -0,0 +1,37 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql import __version__
+
+
+@dataclass
+class DriverSystemConfiguration:
+    driver_version: str
+    os_name: str
+    os_version: str
+    os_arch: str
+    runtime_name: str
+    runtime_version: str
+    runtime_vendor: str
+    client_app_name: str
+    locale_name: str
+    driver_name: str
+    char_set_encoding: str
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Part of TelemetryEvent
+# DriverSystemConfiguration systemConfig = new DriverSystemConfiguration(
+#     driver_version = "2.9.3",
+#     os_name = "Darwin",
+#     os_version = "24.4.0",
+#     os_arch = "arm64",
+#     runtime_name = "CPython",
+#     runtime_version = "3.13.3",
+#     runtime_vendor = "cpython",
+#     client_app_name = "databricks-sql-python",
+#     locale_name = "en_US",
+#     driver_name = "databricks-sql-python",
+#     char_set_encoding = "UTF-8"
+# )

src/databricks/sql/telemetry/DriverVolumeOperation.py

@@ -0,0 +1,21 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.enums.DriverVolumeOperationType import (
+    DriverVolumeOperationType,
+)
+
+
+@dataclass
+class DriverVolumeOperation:
+    volume_operation_type: DriverVolumeOperationType
+    volume_path: str
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Part of TelemetryEvent
+# DriverVolumeOperation volumeOperation = new DriverVolumeOperation(
+#     volumeOperationType = "LIST",
+#     volumePath = "/path/to/volume"
+# )

src/databricks/sql/telemetry/FrontendLogContext.py

@@ -0,0 +1,20 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.TelemetryClientContext import TelemetryClientContext
+
+
+@dataclass
+class FrontendLogContext:
+    client_context: TelemetryClientContext
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# used in TelemetryFrontendLog
+# FrontendLogContext frontendLogContext = new FrontendLogContext(
+#     clientContext = new TelemetryClientContext(
+#         timestampMillis = 1716489600000,
+#         userAgent = "databricks-sql-python-test"
+#     )
+# )

src/databricks/sql/telemetry/FrontendLogEntry.py

@@ -0,0 +1,11 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.TelemetryEvent import TelemetryEvent
+
+
+@dataclass
+class FrontendLogEntry:
+    sql_driver_log: TelemetryEvent
+
+    def to_json(self):
+        return json.dumps(asdict(self))

src/databricks/sql/telemetry/HostDetails.py

@@ -0,0 +1,18 @@
+import json
+from dataclasses import dataclass, asdict
+
+
+@dataclass
+class HostDetails:
+    host_url: str
+    port: int
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Part of DriverConnectionParameters
+# HostDetails hostDetails = new HostDetails(
+#     hostUrl = "https://my-workspace.cloud.databricks.com",
+#     port = 443
+# )

src/databricks/sql/telemetry/SqlExecutionEvent.py

@@ -0,0 +1,24 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.enums.StatementType import StatementType
+from databricks.sql.telemetry.enums.ExecutionResultFormat import ExecutionResultFormat
+
+
+@dataclass
+class SqlExecutionEvent:
+    statement_type: StatementType
+    is_compressed: bool
+    execution_result: ExecutionResultFormat
+    retry_count: int
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# Part of TelemetryEvent
+# SqlExecutionEvent sqlExecutionEvent = new SqlExecutionEvent(
+#     statementType = "QUERY",
+#     isCompressed = true,
+#     executionResult = "INLINE_ARROW",
+#     retryCount = 0
+# )

src/databricks/sql/telemetry/TelemetryClientContext.py

@@ -0,0 +1,18 @@
+from dataclasses import dataclass, asdict
+import json
+
+
+@dataclass
+class TelemetryClientContext:
+    timestamp_millis: int
+    user_agent: str
+
+    def to_json(self):
+        return json.dumps(asdict(self))
+
+
+# used in FrontendLogContext
+# TelemetryClientContext clientContext = new TelemetryClientContext(
+#     timestampMillis = 1716489600000,
+#     userAgent = "databricks-sql-python-test"
+# )

src/databricks/sql/telemetry/TelemetryEvent.py

@@ -0,0 +1,25 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.DriverSystemConfiguration import DriverSystemConfiguration
+from databricks.sql.telemetry.DriverConnectionParameters import (
+    DriverConnectionParameters,
+)
+from databricks.sql.telemetry.DriverVolumeOperation import DriverVolumeOperation
+from databricks.sql.telemetry.SqlExecutionEvent import SqlExecutionEvent
+from databricks.sql.telemetry.DriverErrorInfo import DriverErrorInfo
+
+
+@dataclass
+class TelemetryEvent:
+    session_id: str
+    sql_statement_id: str
+    system_configuration: DriverSystemConfiguration
+    driver_connection_params: DriverConnectionParameters
+    auth_type: str
+    vol_operation: DriverVolumeOperation
+    sql_operation: SqlExecutionEvent
+    error_info: DriverErrorInfo
+    operation_latency_ms: int
+
+    def to_json(self):
+        return json.dumps(asdict(self))

src/databricks/sql/telemetry/TelemetryFrontendLog.py

@@ -0,0 +1,15 @@
+import json
+from dataclasses import dataclass, asdict
+from databricks.sql.telemetry.FrontendLogContext import FrontendLogContext
+from databricks.sql.telemetry.FrontendLogEntry import FrontendLogEntry
+
+
+@dataclass
+class TelemetryFrontendLog:
+    workspace_id: int
+    frontend_log_event_id: str
+    context: FrontendLogContext
+    entry: FrontendLogEntry
+
+    def to_json(self):
+        return json.dumps(asdict(self))

src/databricks/sql/telemetry/TelemetryRequest.py

@@ -0,0 +1,13 @@
+import json
+from dataclasses import dataclass, asdict
+from typing import List, Optional
+
+
+@dataclass
+class TelemetryRequest:
+    uploadTime: int
+    items: List[str]
+    protoLogs: Optional[List[str]]
+
+    def to_json(self):
+        return json.dumps(asdict(self))

src/databricks/sql/telemetry/TelemetryResponse.py

@@ -0,0 +1,13 @@
+import json
+from dataclasses import dataclass, asdict
+from typing import List, Optional
+
+
+@dataclass
+class TelemetryResponse:
+    errors: List[str]
+    numSuccess: int
+    numProtoSuccess: int
+
+    def to_json(self):
+        return json.dumps(asdict(self))

src/databricks/sql/telemetry/enums/AuthFlow.py

@@ -0,0 +1,8 @@
+from enum import Enum
+
+
+class AuthFlow(Enum):
+    TOKEN_PASSTHROUGH = "token_passthrough"
+    CLIENT_CREDENTIALS = "client_credentials"
+    BROWSER_BASED_AUTHENTICATION = "browser_based_authentication"
+    AZURE_MANAGED_IDENTITIES = "azure_managed_identities"

src/databricks/sql/telemetry/enums/AuthMech.py

@@ -0,0 +1,7 @@
+from enum import Enum
+
+
+class AuthMech(Enum):
+    OTHER = "other"
+    PAT = "pat"
+    OAUTH = "oauth"

src/databricks/sql/telemetry/enums/DatabricksClientType.py

@@ -0,0 +1,6 @@
+from enum import Enum
+
+
+class DatabricksClientType(Enum):
+    SEA = "SEA"
+    THRIFT = "THRIFT"

src/databricks/sql/telemetry/enums/DriverVolumeOperationType.py

@@ -0,0 +1,10 @@
+from enum import Enum
+
+
+class DriverVolumeOperationType(Enum):
+    TYPE_UNSPECIFIED = "type_unspecified"
+    PUT = "put"
+    GET = "get"
+    DELETE = "delete"
+    LIST = "list"
+    QUERY = "query"

src/databricks/sql/telemetry/enums/ExecutionResultFormat.py

@@ -0,0 +1,8 @@
+from enum import Enum
+
+
+class ExecutionResultFormat(Enum):
+    FORMAT_UNSPECIFIED = "format_unspecified"
+    INLINE_ARROW = "inline_arrow"
+    EXTERNAL_LINKS = "external_links"
+    COLUMNAR_INLINE = "columnar_inline"

src/databricks/sql/telemetry/enums/StatementType.py

@@ -0,0 +1,9 @@
+from enum import Enum
+
+
+class StatementType(Enum):
+    NONE = "none"
+    QUERY = "query"
+    SQL = "sql"
+    UPDATE = "update"
+    METADATA = "metadata"