airbyte-ci: Add basic segment tracking (#31354)

Co-authored-by: bnchrch <bnchrch@users.noreply.github.com>

airbyte-ci/connectors/pipelines/pipelines/commands/airbyte_ci.py

@@ -5,15 +5,14 @@
 """This module is the CLI entrypoint to the airbyte-ci commands."""
 
 import importlib
-import os
-import subprocess
 from typing import List
 
 import click
 from github import PullRequest
 from pipelines import github, main_logger
 from pipelines.bases import CIContext
 from pipelines.consts import LOCAL_PIPELINE_PACKAGE_PATH
+from pipelines.telemetry import track_command
 from pipelines.utils import (
     get_current_epoch_time,
     get_current_git_branch,
@@ -119,6 +118,7 @@ def get_modified_files(
 @click.option("--ci-job-key", envvar="CI_JOB_KEY", type=str)
 @click.option("--show-dagger-logs/--hide-dagger-logs", default=False, type=bool)
 @click.pass_context
+@track_command
 def airbyte_ci(
     ctx: click.Context,
     is_local: bool,

airbyte-ci/connectors/pipelines/pipelines/telemetry.py

@@ -0,0 +1,60 @@
+import getpass
+import hashlib
+import os
+import platform
+import sys
+
+import segment.analytics as analytics
+
+analytics.write_key = "G6G7whgro81g9xM00kN2buclGKvcOjFd"
+analytics.send = True
+analytics.debug = True
+
+
+def _is_airbyte_user():
+    """Returns True if the user is airbyter, False otherwise."""
+    return os.getenv("AIRBYTE_ROLE") == "airbyter"
+
+
+def _get_anonymous_system_id():
+    """Returns a unique anonymous hashid of the current system info."""
+    # Collect machine-specific information
+    machine_info = platform.node()
+    username = getpass.getuser()
+
+    unique_system_info = f"{machine_info}-{username}"
+
+    # Generate a unique hash
+    unique_id = hashlib.sha256(unique_system_info.encode()).hexdigest()
+
+    return unique_id
+
+
+def track_command(f):
+    """
+    Decorator to track CLI commands with segment.io
+    """
+
+    def wrapper(*args, **kwargs):
+        ctx = args[0]
+        top_level_command = ctx.command_path
+        full_cmd = " ".join(sys.argv)
+
+        # remove anything prior to the command name f.__name__
+        # to avoid logging inline secrets
+        santized_cmd = full_cmd[full_cmd.find(top_level_command) :]
+
+        sys_id = _get_anonymous_system_id()
+        sys_user_name = f"anonymous:{sys_id}"
+        airbyter = _is_airbyte_user()
+
+        is_local = kwargs.get("is_local", False)
+        user_id = "local-user" if is_local else "ci-user"
+        event = f"airbyte-ci:{f.__name__}"
+
+        # IMPORTANT! do not log kwargs as they may contain secrets
+        analytics.track(user_id, event, {"username": sys_user_name, "command": santized_cmd, "airbyter": airbyter})
+
+        return f(*args, **kwargs)
+
+    return wrapper

airbyte-ci/connectors/pipelines/pyproject.toml

@@ -24,6 +24,7 @@ airbyte-connectors-base-images = {path = "../base_images", develop = true}
 connector-ops = {path = "../connector_ops", develop = true}
 toml = "^0.10.2"
 sentry-sdk = "^1.28.1"
+segment-analytics-python = "^2.2.3"
 
 [tool.poetry.group.test.dependencies]
 pytest = "^6.2.5"