vdk-core: add vdk sql-query command

projects/vdk-core/requirements.txt

@@ -20,3 +20,4 @@ pytest-randomly
 pytest-sugar
 reorder-python-imports
 smtpdfix
+vdk-sqlite

projects/vdk-core/src/vdk/internal/builtin_plugins/builtin_hook_impl.py

@@ -22,6 +22,9 @@
 from vdk.internal.builtin_plugins.connection.connection_plugin import (
     QueryDecoratorPlugin,
 )
+from vdk.internal.builtin_plugins.connection.query_command_plugin import (
+    QueryCommandPlugin,
+)
 from vdk.internal.builtin_plugins.debug.debug import DebugPlugins
 from vdk.internal.builtin_plugins.ingestion.ingester_configuration_plugin import (
     IngesterConfigurationPlugin,
@@ -130,6 +133,7 @@ def vdk_start(plugin_registry: PluginRegistry, command_line_args: List) -> None:
     # connection plugins
     plugin_registry.add_hook_specs(ConnectionHookSpec)
     plugin_registry.load_plugin_with_hooks_impl(QueryDecoratorPlugin())
+    plugin_registry.load_plugin_with_hooks_impl(QueryCommandPlugin())
 
 
 @hookimpl(tryfirst=True)

projects/vdk-core/src/vdk/internal/builtin_plugins/connection/query_command_plugin.py

@@ -0,0 +1,83 @@
+# Copyright 2021-2023 VMware, Inc.
+# SPDX-License-Identifier: Apache-2.0
+import json
+import os
+from pathlib import Path
+
+import click
+from vdk.api.plugin.hook_markers import hookimpl
+from vdk.internal.builtin_plugins.connection.managed_connection_base import (
+    ManagedConnectionBase,
+)
+from vdk.internal.builtin_plugins.run.standalone_data_job import (
+    StandaloneDataJobFactory,
+)
+from vdk.internal.util.decorators import closing_noexcept_on_close
+
+OUTPUT_OPTIONS = ["json", "text"]
+
+# The command is hidden as it's still experimental.
+# TODOs:
+# Remove logs  by default only show output (and perhaps warning logs)
+# When logging omit "job details" logs
+# Make clear how it is authenticated.
+
+
+@click.command(
+    name="sql-query",
+    hidden=True,  # Hidden as it is experimental
+    help="executes SQL query against configured database. "
+    "The database used is configured with 'db_default_type' option. "
+    "See vdk config-help for more info. ",
+)
+@click.option(
+    "-q", "--query", type=click.STRING, required=True, help="The query to be executed"
+)
+@click.option(
+    "-o",
+    "--output",
+    type=click.Choice(OUTPUT_OPTIONS, False),
+    default="text",
+    help=f"Output format. It can be one of: {OUTPUT_OPTIONS}.",
+)
+@click.pass_context
+def sql_query(ctx: click.Context, query: str, output: str) -> None:
+    # TODO: os.getcwd() is passed to StandaloneDataJob, this means that if the current directory is same as data job
+    # it would authenticate using the data job credentials (if there are any like keytab)
+    # Is that a good idea?
+
+    with StandaloneDataJobFactory.create(
+        data_job_directory=Path(os.getcwd())
+    ) as job_input:
+        conn: ManagedConnectionBase = job_input.get_managed_connection()
+        with closing_noexcept_on_close(conn.cursor()) as cursor:
+            cursor.execute(query)
+            column_names = (
+                [column_info[0] for column_info in cursor.description]
+                if cursor.description
+                else ()  # same as the default value for the headers parameter of the tabulate function
+            )
+            # TODO: this is basically emulating
+            # https://github.com/vmware/versatile-data-kit/blob/main/projects/vdk-control-cli/src/vdk/internal/control/utils/output_printer.py
+            # It will be best printer functionality to be reusable (either moved to vdk-core or in a library/plugin)
+            result = cursor.fetchall()
+            if output.lower() == "json":
+                result = [
+                    {column_names[i]: v for i, v in enumerate(row)} for row in result
+                ]
+                click.echo(json.dumps(result))
+            elif output.lower() == "text":
+                from tabulate import tabulate
+
+                click.echo(tabulate(result, headers=column_names))
+            else:
+                raise ValueError(
+                    f"Unsupported output format. Choose between: {OUTPUT_OPTIONS}"
+                )
+
+
+class QueryCommandPlugin:
+    @staticmethod
+    @hookimpl
+    def vdk_command_line(root_command: click.Group) -> None:
+        root_command.add_command(sql_query)

projects/vdk-core/tests/functional/cli/test_query_command_plugin.py

@@ -0,0 +1,72 @@
+# Copyright 2021-2023 VMware, Inc.
+# SPDX-License-Identifier: Apache-2.0
+import json
+import os
+from unittest import mock
+
+import py
+from click.testing import CliRunner
+from click.testing import Result
+from vdk.plugin.test_utils.util_funcs import cli_assert_equal
+from vdk.plugin.test_utils.util_funcs import CliEntryBasedTestRunner
+
+VDK_DB_DEFAULT_TYPE = "VDK_DB_DEFAULT_TYPE"
+
+
+def test_execute_sql(tmpdir: py.path.local):
+    with mock.patch.dict(
+        os.environ,
+        {
+            VDK_DB_DEFAULT_TYPE: "sqlite",
+            "SQLITE_FILE": os.path.join(str(tmpdir), "sqlite-tmp"),
+        },
+    ):
+        runner = CliEntryBasedTestRunner()
+
+        result: Result = runner.invoke(
+            [
+                "sql-query",
+                "--query",
+                "CREATE TABLE cli_stocks (date text, symbol text, price real)",
+            ]
+        )
+        cli_assert_equal(0, result)
+
+        result: Result = runner.invoke(
+            [
+                "sql-query",
+                "--query",
+                """
+                INSERT INTO cli_stocks VALUES ('2020-01-01', 'GOOG', 123.0), ('2020-01-01', 'GOOG', 123.0)
+                """,
+            ]
+        )
+        cli_assert_equal(0, result)
+
+        result: Result = runner.invoke(
+            ["sql-query", "--query", "select * from cli_stocks"],
+            runner=CliRunner(
+                mix_stderr=False
+            ),  # TODO: replace when CliEntryBasedTestRunner add support for it
+        )
+        cli_assert_equal(0, result)
+        expected_stdout = (
+            "date        symbol      price\n"
+            "----------  --------  -------\n"
+            "2020-01-01  GOOG          123\n"
+            "2020-01-01  GOOG          123\n"
+        )
+        assert expected_stdout == result.stdout
+
+        result: Result = runner.invoke(
+            ["sql-query", "--output", "json", "--query", "select * from cli_stocks"],
+            runner=CliRunner(
+                mix_stderr=False
+            ),  # TODO: replace when CliEntryBasedTestRunner add support for it
+        )
+        cli_assert_equal(0, result)
+        expected_stdout_json = [
+            {"date": "2020-01-01", "symbol": "GOOG", "price": 123.0},
+            {"date": "2020-01-01", "symbol": "GOOG", "price": 123.0},
+        ]
+        assert expected_stdout_json == json.loads(result.stdout)