From f24e1041c5018b1ec2caf89ef5bce29f127f3302 Mon Sep 17 00:00:00 2001
From: Tapas Kumar Senapati <tksenapa@ncsu.edu>
Date: Fri, 23 Sep 2022 14:44:55 +0530
Subject: [PATCH] Cml profile instrumentation (#45)

* Added cml instrumentation
---
 dbt/adapters/hive/cloudera_tracking.py | 21 +++++++++++++++++++++
 dbt/adapters/hive/connections.py       |  2 ++
 2 files changed, 23 insertions(+)

diff --git a/dbt/adapters/hive/cloudera_tracking.py b/dbt/adapters/hive/cloudera_tracking.py
index 1c06976..99391de 100644
--- a/dbt/adapters/hive/cloudera_tracking.py
+++ b/dbt/adapters/hive/cloudera_tracking.py
@@ -14,6 +14,7 @@
 
 import dbt.version
 import json
+import os
 import platform
 import requests
 import sys
@@ -41,6 +42,9 @@
 # Json object to store dbt profile(profile.yml) related information
 profile_info = {}
 
+# Json object to store cml environment variables
+cml_info = {}
+
 
 def populate_platform_info(cred: Credentials, ver):
     """
@@ -66,6 +70,22 @@ def populate_platform_info(cred: Credentials, ver):
     platform_info["dbt_adapter"] = f"{cred.type}-{ver.version}"
 
 
+def populate_cml_info():
+    """
+    populate cml environment variables if available to be passed on for tracking
+    """
+    default_value = ""  # if environment variables doesn't exist add empty string as default
+    cml_info["ml_runtime_edition"] = os.environ.get('ML_RUNTIME_EDITION', default_value)
+    cml_info["ml_runtime_git_hash"] = os.environ.get('ML_RUNTIME_GIT_HASH', default_value)
+    cml_info["ml_runtime_kernel"] = os.environ.get('ML_RUNTIME_KERNEL', default_value)
+    cml_info["ml_runtime_editor"] = os.environ.get('ML_RUNTIME_EDITOR', default_value)
+    cml_info["ml_runtime_gbn"] = os.environ.get('ML_RUNTIME_GBN', default_value)
+    cml_info["ml_runtime_full_version"] = os.environ.get('ML_RUNTIME_FULL_VERSION', default_value)
+    cml_info["ml_runtime_description"] = os.environ.get('ML_RUNTIME_DESCRIPTION', default_value)
+    cml_info["ml_runtime_maintenance_version"] = os.environ.get('ML_RUNTIME_MAINTENANCE_VERSION', default_value)
+    cml_info["ml_runtime_metadata_version"] = os.environ.get('ML_RUNTIME_METADATA_VERSION', default_value)
+
+
 def populate_unique_ids(cred: Credentials):
     host = str(cred.host).encode()
     user = str(cred.username).encode()
@@ -178,6 +198,7 @@ def track_usage(tracking_payload):
     # inject other static payload to tracking_payload
     tracking_payload = _merge_keys(unique_ids, tracking_payload)
     tracking_payload = _merge_keys(platform_info, tracking_payload)
+    tracking_payload = _merge_keys(cml_info, tracking_payload)
     tracking_payload = _merge_keys(profile_info, tracking_payload)
 
     # form the tracking data
diff --git a/dbt/adapters/hive/connections.py b/dbt/adapters/hive/connections.py
index e98c1c7..6997932 100644
--- a/dbt/adapters/hive/connections.py
+++ b/dbt/adapters/hive/connections.py
@@ -94,6 +94,8 @@ def __post_init__(self):
         tracker.usage_tracking = self.usage_tracking
         # get platform information for tracking
         tracker.populate_platform_info(self, ver)
+        # get cml information for tracking
+        tracker.populate_cml_info()
         # generate unique ids for tracking
         tracker.populate_unique_ids(self)