Merge pull request #335 from DataRecce/feature/drc-477-add-a-runner-t…

…ype-in-the-recce-telemetry

[Chore] Track running environment and repository

recce/__init__.py

@@ -18,6 +18,22 @@ def is_ci_env():
     return False
 
 
+def get_runner():
+    # GitHub Action
+    if os.environ.get('GITHUB_ACTIONS', 'false') == 'true':
+        return 'github actions'
+
+    # GitHub Codespace
+    if os.environ.get('CODESPACES', 'false') == 'true':
+        return 'github codespaces'
+
+    # CircleCI
+    if os.environ.get('CIRCLECI', 'false') == 'true':
+        return 'circleci'
+
+    return None
+
+
 def get_version():
     version_file = os.path.normpath(os.path.join(os.path.dirname(__file__), 'VERSION'))
     with open(version_file) as fh:

recce/event/__init__.py

@@ -145,15 +145,6 @@ def log_event(prop, event_type, **kwargs):
         **prop,
     )
 
-    # only 'server' and 'run' commands carry the adapter info
-    params = kwargs.get('params')
-    adapter_type_sqlmesh = None
-    if params is not None:
-        adapter_type_sqlmesh = params.get('sqlmesh')
-
-    if adapter_type_sqlmesh is not None:
-        payload['adapter_type'] = 'SQLMesh' if adapter_type_sqlmesh else 'DBT'
-
     _collector.log_event(payload, event_type)
 
 

recce/event/track.py

@@ -3,13 +3,16 @@
 import time
 import traceback
 import typing as t
+from hashlib import sha256
 
 from click import Context
 from click.core import Command
 from rich.console import Console
 from rich.markup import escape
 
-from recce import event
+from recce import event, get_runner
+from recce.core import load_context
+from recce.git import hosting_repo
 
 console = Console()
 
@@ -77,6 +80,8 @@ def invoke(self, ctx: Context) -> t.Any:
         finally:
             end_time = time.time()
             duration = end_time - start_time
+            runner = get_runner()
+            repo = hosting_repo()
             command = ctx.command.name
             props = dict(
                 command=command,
@@ -85,5 +90,23 @@ def invoke(self, ctx: Context) -> t.Any:
                 duration=duration,
             )
 
+            if runner is not None:
+                props['runner_type'] = runner
+
+            if repo is not None:
+                props['repository'] = sha256(repo.encode()).hexdigest()
+
+            try:
+                recce_context = load_context()
+            except Exception:
+                # it's not a ready-for-use project
+                recce_context = None
+
+            if recce_context is not None:
+                if recce_context.adapter_type == "dbt":
+                    props['adapter_type'] = 'DBT'
+                elif recce_context.adapter_type == "sqlmesh":
+                    props['adapter_type'] = 'SQLMesh'
+
             event.log_event(props, command, params=ctx.params)
             event.flush_events()

recce/git.py

@@ -1,4 +1,6 @@
-from git import Repo
+import os
+
+from git import Repo, InvalidGitRepositoryError
 
 
 def current_branch():
@@ -12,3 +14,27 @@ def current_branch():
         return None
     except Exception:
         return None
+
+
+def hosting_repo():
+    try:
+        repo = Repo(search_parent_directories=True)
+        origin_url = repo.remote().url
+        remote_repo = None
+
+        if origin_url.startswith('git@'):
+            # Handle [email protected]:user/repo.git
+            remote_repo = origin_url.split(':')[1].replace('.git', '')
+
+        elif origin_url.startswith('https://') or origin_url.startswith('http://'):
+            # Handle https://github.com/user/repo.git or http://github.com/user/repo.git
+            remote_repo = '/'.join(origin_url.split('/')[-2:]).replace('.git', '')
+
+        return remote_repo
+    except ValueError:
+        repo = Repo(search_parent_directories=True)
+        toplevel_dir = repo.git.rev_parse("--show-toplevel")
+
+        return os.path.basename(toplevel_dir)
+    except InvalidGitRepositoryError:
+        return None