[Internal] Add CICD environment to the User Agent (#866)

## What changes are proposed in this pull request?

This PR adds CICD environment to the User Agent of each SDK outbound
request. The implementation is similar to the one used in the Go SDK.

## How is this tested?

Added unit tests.

Author: renaudhartert-db

databricks/sdk/useragent.py

@@ -148,4 +148,58 @@ def to_string(alternate_product_info: Optional[Tuple[str, str]] = None,
     base.extend(_extra)
     base.extend(_get_upstream_user_agent_info())
     base.extend(_get_runtime_info())
+    if cicd_provider() != "":
+        base.append((CICD_KEY, cicd_provider()))
     return " ".join(f"{k}/{v}" for k, v in base)
+
+
+# List of CI/CD providers and pairs of envvar/value that are used to detect them.
+_PROVIDERS = {
+    "github": [("GITHUB_ACTIONS", "true")],
+    "gitlab": [("GITLAB_CI", "true")],
+    "jenkins": [("JENKINS_URL", "")],
+    "azure-devops": [("TF_BUILD", "True")],
+    "circle": [("CIRCLECI", "true")],
+    "travis": [("TRAVIS", "true")],
+    "bitbucket": [("BITBUCKET_BUILD_NUMBER", "")],
+    "google-cloud-build": [("PROJECT_ID", ""), ("BUILD_ID", ""), ("PROJECT_NUMBER", ""), ("LOCATION", "")],
+    "aws-code-build": [("CODEBUILD_BUILD_ARN", "")],
+    "tf-cloud": [("TFC_RUN_ID", "")],
+}
+
+# Private variable to store the CI/CD provider. This value is computed at
+# the first invocation of cicd_providers() and is cached for subsequent calls.
+_cicd_provider = None
+
+
+def cicd_provider() -> str:
+    """Return the CI/CD provider if detected, or an empty string otherwise."""
+
+    # This function is safe because (i) assignation are atomic, and (ii)
+    # computating the CI/CD provider is idempotent.
+    global _cicd_provider
+    if _cicd_provider is not None:
+        return _cicd_provider
+
+    providers = []
+    for p in _PROVIDERS:
+        found = True
+        for envvar, value in _PROVIDERS[p]:
+            v = os.getenv(envvar)
+            if v is None or (value != "" and v != value):
+                found = False
+                break
+
+        if found:
+            providers.append(p)
+
+    if len(providers) == 0:
+        _cicd_provider = ""
+    else:
+        # TODO: reconsider what to do if multiple providers are detected.
+        # The current mechanism as the benefit of being deterministic and
+        # robust to ordering changes in _PROVIDERS.
+        providers.sort()
+        _cicd_provider = providers[0]
+
+    return _cicd_provider

tests/test_config.py

@@ -42,6 +42,11 @@ class MockUname:
         def system(self):
             return 'TestOS'
 
+    # Clear all environment variables and cached CICD provider.
+    for k in os.environ:
+        monkeypatch.delenv(k, raising=False)
+    useragent._cicd_provider = None
+
     monkeypatch.setattr(platform, 'python_version', lambda: '3.0.0')
     monkeypatch.setattr(platform, 'uname', MockUname)
     monkeypatch.setenv('DATABRICKS_SDK_UPSTREAM', "upstream-product")

tests/test_core.py

@@ -8,7 +8,7 @@
 
 import pytest
 
-from databricks.sdk import WorkspaceClient, errors
+from databricks.sdk import WorkspaceClient, errors, useragent
 from databricks.sdk.core import ApiClient, Config, DatabricksError
 from databricks.sdk.credentials_provider import (CliTokenSource,
                                                  CredentialsProvider,
@@ -178,6 +178,11 @@ class MockUname:
         def system(self):
             return 'TestOS'
 
+    # Clear all environment variables and cached CICD provider.
+    for k in os.environ:
+        monkeypatch.delenv(k, raising=False)
+    useragent._cicd_provider = None
+
     monkeypatch.setattr(platform, 'python_version', lambda: '3.0.0')
     monkeypatch.setattr(platform, 'uname', MockUname)
     monkeypatch.setenv('DATABRICKS_SDK_UPSTREAM', "upstream-product")

tests/test_user_agent.py

@@ -1,3 +1,5 @@
+import os
+
 import pytest
 
 from databricks.sdk.version import __version__
@@ -40,3 +42,45 @@ def test_user_agent_with_partner(user_agent):
     user_agent.with_partner('differenttest')
     assert 'partner/test' in user_agent.to_string()
     assert 'partner/differenttest' in user_agent.to_string()
+
+
+@pytest.fixture(scope="function")
+def clear_cicd():
+    # Save and clear env vars.
+    original_env = os.environ.copy()
+    os.environ.clear()
+
+    # Clear cached CICD provider.
+    from databricks.sdk import useragent
+    useragent._cicd_provider = None
+
+    yield
+
+    # Restore env vars.
+    os.environ = original_env
+
+
+def test_user_agent_cicd_no_provider(clear_cicd):
+    from databricks.sdk import useragent
+    user_agent = useragent.to_string()
+
+    assert 'cicd' not in user_agent
+
+
+def test_user_agent_cicd_one_provider(clear_cicd):
+    os.environ['GITHUB_ACTIONS'] = 'true'
+
+    from databricks.sdk import useragent
+    user_agent = useragent.to_string()
+
+    assert 'cicd/github' in user_agent
+
+
+def test_user_agent_cicd_two_provider(clear_cicd):
+    os.environ['GITHUB_ACTIONS'] = 'true'
+    os.environ['GITLAB_CI'] = 'true'
+
+    from databricks.sdk import useragent
+    user_agent = useragent.to_string()
+
+    assert 'cicd/github' in user_agent