Enrich telemetry metadata

src/xpk/core/config.py

@@ -84,7 +84,7 @@ def _save_configs(self, config_yaml: dict) -> None:
     with open(self._config, encoding='utf-8', mode='w') as stream:
       yaml.dump(config_yaml, stream)
 
-  def set(self, key: str, value: str) -> None:
+  def set(self, key: str, value: str | None) -> None:
     if key not in self._allowed_keys:
       xpk_print(f'Key {key} is not an allowed xpk config key.')
       return

src/xpk/core/telemetry.py

@@ -105,17 +105,6 @@ def _clearcut_flush(file_path: str) -> None:
     os.remove(file_path)
 
 
-def ensure_client_id() -> str:
-  """Generates Client ID and stores in configuration if not already present."""
-  current_client_id = xpk_config.get(CLIENT_ID_KEY)
-  if current_client_id is not None:
-    return current_client_id
-
-  new_client_id = str(uuid.uuid4())
-  xpk_config.set(CLIENT_ID_KEY, new_client_id)
-  return new_client_id
-
-
 class MetricsEventMetadataKey(Enum):
   SESSION_ID = "XPK_SESSION_ID"
   DRY_RUN = "XPK_DRY_RUN"
@@ -125,6 +114,8 @@ class MetricsEventMetadataKey(Enum):
   PROVISIONING_MODE = "XPK_PROVISIONING_MODE"
   COMMAND = "XPK_COMMAND"
   EXIT_CODE = "XPK_EXIT_CODE"
+  RUNNING_AS_PIP = "XPK_RUNNING_AS_PIP"
+  RUNNING_FROM_SOURCE = "XPK_RUNNING_FROM_SOURCE"
 
 
 @dataclass
@@ -222,16 +213,43 @@ def _get_base_event_metadata() -> dict[MetricsEventMetadataKey, str]:
       MetricsEventMetadataKey.SESSION_ID: _get_session_id(),
       MetricsEventMetadataKey.DRY_RUN: str(is_dry_run()).lower(),
       MetricsEventMetadataKey.PYTHON_VERSION: platform.python_version(),
+      MetricsEventMetadataKey.RUNNING_AS_PIP: str(_is_running_as_pip()).lower(),
+      MetricsEventMetadataKey.RUNNING_FROM_SOURCE: str(
+          _is_running_from_source()
+      ).lower(),
   }
 
 
 def _get_base_concord_event() -> dict[str, str]:
   return {
       "release_version": xpk_version,
       "console_type": "XPK",
-      "client_install_id": ensure_client_id(),
+      "client_install_id": _ensure_client_id(),
   }
 
 
+def _is_running_as_pip() -> bool:
+  return os.path.basename(sys.argv[0]) == "xpk"
+
+
+def _is_running_from_source() -> bool:
+  current_path = os.path.abspath(os.path.realpath(__file__))
+  return (
+      "site-packages" not in current_path
+      and "dist-packages" not in current_path
+  )
+
+
 def _get_session_id() -> str:
   return str(uuid.uuid4())
+
+
+def _ensure_client_id() -> str:
+  """Generates Client ID and stores in configuration if not already present."""
+  current_client_id = xpk_config.get(CLIENT_ID_KEY)
+  if current_client_id is not None:
+    return current_client_id
+
+  new_client_id = str(uuid.uuid4())
+  xpk_config.set(CLIENT_ID_KEY, new_client_id)
+  return new_client_id

src/xpk/core/telemetry_test.py

@@ -17,33 +17,24 @@
 import pytest
 import json
 from .config import xpk_config, CLIENT_ID_KEY
-from .telemetry import ensure_client_id, MetricsCollector, MetricsEventMetadataKey
+from .telemetry import MetricsCollector, MetricsEventMetadataKey
 from ..utils.execution_context import set_dry_run
+from pytest_mock import MockerFixture
 
 
 @pytest.fixture(autouse=True)
-def setup_mocks(mocker):
+def setup_mocks(mocker: MockerFixture):
   mocker.patch('xpk.core.telemetry._get_session_id', return_value='321231')
   mocker.patch('time.time', return_value=0)
   mocker.patch('platform.python_version', return_value='99.99.99')
+  mocker.patch('os.path.basename', return_value='xpk.py')
+  mocker.patch('os.path.abspath', return_value='/home/xpk_user')
+  set_dry_run(False)
   xpk_config.set(CLIENT_ID_KEY, 'client_id')
   yield
   xpk_config.set(CLIENT_ID_KEY, None)
 
 
-def test_ensure_client_id_generates_client_id_when_its_not_present():
-  xpk_config.set(CLIENT_ID_KEY, None)
-  ensure_client_id()
-  assert xpk_config.get(CLIENT_ID_KEY) is not None
-
-
-def test_ensure_client_id_does_not_regenerate_id_when_its_present():
-  client_id = '1337'
-  xpk_config.set(CLIENT_ID_KEY, client_id)
-  ensure_client_id()
-  assert xpk_config.get(CLIENT_ID_KEY) == client_id
-
-
 def test_metrics_collector_generates_client_id_if_not_present():
   xpk_config.set(CLIENT_ID_KEY, None)
   MetricsCollector.log_start(command='test')
@@ -54,7 +45,6 @@ def test_metrics_collector_generates_client_id_if_not_present():
 
 
 def test_metrics_collector_logs_start_event_correctly():
-  set_dry_run(False)
   MetricsCollector.log_start(command='test')
   payload = json.loads(MetricsCollector.flush())
   extension_json = json.loads(payload['log_event'][0]['source_extension_json'])
@@ -65,6 +55,8 @@ def test_metrics_collector_logs_start_event_correctly():
           {'key': 'XPK_SESSION_ID', 'value': '321231'},
           {'key': 'XPK_DRY_RUN', 'value': 'false'},
           {'key': 'XPK_PYTHON_VERSION', 'value': '99.99.99'},
+          {'key': 'XPK_RUNNING_AS_PIP', 'value': 'false'},
+          {'key': 'XPK_RUNNING_FROM_SOURCE', 'value': 'true'},
           {'key': 'XPK_COMMAND', 'value': 'test'},
       ],
       'event_name': 'start',
@@ -73,8 +65,16 @@ def test_metrics_collector_logs_start_event_correctly():
   }
 
 
+def test_metrics_collector_generates_client_id_when_not_present():
+  xpk_config.set(CLIENT_ID_KEY, None)
+  MetricsCollector.log_start(command='test')
+  payload = json.loads(MetricsCollector.flush())
+  extension_json = json.loads(payload['log_event'][0]['source_extension_json'])
+  assert extension_json['client_install_id'] is not None
+  assert len(extension_json['client_install_id']) > 0
+
+
 def test_metrics_collector_logs_complete_event_correctly():
-  set_dry_run(True)
   MetricsCollector.log_complete(exit_code=2)
   payload = json.loads(MetricsCollector.flush())
   extension_json = json.loads(payload['log_event'][0]['source_extension_json'])
@@ -83,8 +83,10 @@ def test_metrics_collector_logs_complete_event_correctly():
       'console_type': 'XPK',
       'event_metadata': [
           {'key': 'XPK_SESSION_ID', 'value': '321231'},
-          {'key': 'XPK_DRY_RUN', 'value': 'true'},
+          {'key': 'XPK_DRY_RUN', 'value': 'false'},
           {'key': 'XPK_PYTHON_VERSION', 'value': '99.99.99'},
+          {'key': 'XPK_RUNNING_AS_PIP', 'value': 'false'},
+          {'key': 'XPK_RUNNING_FROM_SOURCE', 'value': 'true'},
           {'key': 'XPK_EXIT_CODE', 'value': '2'},
       ],
       'event_name': 'complete',
@@ -94,7 +96,6 @@ def test_metrics_collector_logs_complete_event_correctly():
 
 
 def test_metrics_collector_logs_custom_event_correctly():
-  set_dry_run(False)
   MetricsCollector.log_custom(
       name='test', metadata={MetricsEventMetadataKey.PROVISIONING_MODE: 'flex'}
   )
@@ -107,6 +108,8 @@ def test_metrics_collector_logs_custom_event_correctly():
           {'key': 'XPK_SESSION_ID', 'value': '321231'},
           {'key': 'XPK_DRY_RUN', 'value': 'false'},
           {'key': 'XPK_PYTHON_VERSION', 'value': '99.99.99'},
+          {'key': 'XPK_RUNNING_AS_PIP', 'value': 'false'},
+          {'key': 'XPK_RUNNING_FROM_SOURCE', 'value': 'true'},
           {'key': 'XPK_PROVISIONING_MODE', 'value': 'flex'},
       ],
       'event_name': 'test',
@@ -134,3 +137,57 @@ def test_metrics_collector_does_not_flush_event_twice():
   MetricsCollector.log_start(command='version')
   payload = json.loads(MetricsCollector.flush())
   assert len(payload['log_event']) == 1
+
+
+@pytest.mark.parametrize(
+    argnames='dry_run,expected', argvalues=[(False, 'false'), (True, 'true')]
+)
+def test_metrics_collector_logs_correct_dry_run_value(
+    dry_run: bool, expected: str
+):
+  set_dry_run(dry_run)
+  MetricsCollector.log_start(command='test')
+  payload = MetricsCollector.flush()
+  assert _get_metadata_value(payload, 'XPK_DRY_RUN') == expected
+
+
+@pytest.mark.parametrize(
+    argnames='basename,expected',
+    argvalues=[
+        ('xpk', 'true'),
+        ('xpk.py', 'false'),
+    ],
+)
+def test_metrics_collectors_logs_correct_running_as_pip_value(
+    basename: str, expected: str, mocker: MockerFixture
+):
+  mocker.patch('os.path.basename', return_value=basename)
+  MetricsCollector.log_start(command='test')
+  payload = MetricsCollector.flush()
+  assert _get_metadata_value(payload, 'XPK_RUNNING_AS_PIP') == expected
+
+
+@pytest.mark.parametrize(
+    argnames='abspath,expected',
+    argvalues=[
+        ('/site-packages/', 'false'),
+        ('/dist-packages/', 'false'),
+        ('/home/xpk_user', 'true'),
+    ],
+)
+def test_metrics_collectors_logs_correct_running_from_source_value(
+    abspath: str, expected: str, mocker: MockerFixture
+):
+  mocker.patch('os.path.abspath', return_value=abspath)
+  MetricsCollector.log_start(command='test')
+  payload = MetricsCollector.flush()
+  assert _get_metadata_value(payload, 'XPK_RUNNING_FROM_SOURCE') == expected
+
+
+def _get_metadata_value(payload_str: str, key: str) -> str | None:
+  payload = json.loads(payload_str)
+  metadata = json.loads(payload['log_event'][0]['source_extension_json'])[
+      'event_metadata'
+  ]
+  matching = (item['value'] for item in metadata if item['key'] == key)
+  return next(matching, None)