feat: git clone private repository inside Kaggle notebook with GH_TOKEN secret support

scripts/build_kaggle_deploy.py

@@ -27,10 +27,12 @@ def build_deploy_dir():
                 "metadata": {},
                 "outputs": [],
                 "source": [
-                    "# 1. System/GPU Information Check\n",
+                    "# 1. GPU & CUDA Check\n",
                     "import torch\n",
                     "import sys\n",
                     "import os\n",
+                    "import subprocess\n",
+                    "import shutil\n",
                     "\n",
                     "print(\"=== Kaggle GPU & System Check ===\")\n",
                     "print(f\"Python Version: {sys.version}\")\n",
@@ -41,30 +43,55 @@ def build_deploy_dir():
                     "else:\n",
                     "    print(\"WARNING: GPU is not active. Please enable GPU in the Kaggle settings for optimal performance.\")\n",
                     "\n",
-                    "# 2. Check Kaggle secrets\n",
+                    "# 2. Load Kaggle User Secrets\n",
                     "print(\"\\n=== Kaggle Secrets Validation ===\")\n",
+                    "gh_token = None\n",
                     "try:\n",
                     "    from kaggle_secrets import UserSecretsClient\n",
                     "    user_secrets = UserSecretsClient()\n",
                     "    db_url = user_secrets.get_secret(\"DATABASE_URL\")\n",
                     "    rabbitmq_url = user_secrets.get_secret(\"RABBITMQ_URL\")\n",
+                    "    gh_token = user_secrets.get_secret(\"GH_TOKEN\")\n",
                     "    print(f\"DATABASE_URL secret: {'FOUND' if db_url else 'MISSING'}\")\n",
                     "    print(f\"RABBITMQ_URL secret: {'FOUND' if rabbitmq_url else 'MISSING'}\")\n",
+                    "    print(f\"GH_TOKEN secret: {'FOUND' if gh_token else 'MISSING'}\")\n",
                     "except Exception as e:\n",
                     "    print(f\"Warning: Could not check secrets: {e}\")\n",
                     "\n",
-                    "# 3. Install Package dependencies\n",
+                    "# 3. Clone Repository (Since Kaggle CLI only uploads the notebook itself)\n",
+                    "print(\"\\n=== Cloning Repository ===\")\n",
+                    "repo_dir = \"/kaggle/working/ETL-Ingestion\"\n",
+                    "if os.path.exists(repo_dir):\n",
+                    "    print(f\"Removing existing repository directory at {repo_dir}...\")\n",
+                    "    shutil.rmtree(repo_dir)\n",
+                    "\n",
+                    "if gh_token:\n",
+                    "    print(\"Cloning private repository using GH_TOKEN...\")\n",
+                    "    clone_url = f\"https://x-access-token:{gh_token}@github.com/FP-KKA/ETL-Ingestion.git\"\n",
+                    "else:\n",
+                    "    print(\"WARNING: GH_TOKEN secret not found. Attempting public clone...\")\n",
+                    "    clone_url = \"https://github.com/FP-KKA/ETL-Ingestion.git\"\n",
+                    "\n",
+                    "try:\n",
+                    "    subprocess.check_call([\"git\", \"clone\", clone_url, repo_dir])\n",
+                    "    print(\"Repository cloned successfully.\")\n",
+                    "except Exception as e:\n",
+                    "    print(f\"Error cloning repository: {e}\")\n",
+                    "    sys.exit(1)\n",
+                    "\n",
+                    "# 4. Change Directory & Install Dependencies\n",
+                    "os.chdir(repo_dir)\n",
+                    "print(f\"Changed working directory to: {os.getcwd()}\")\n",
+                    "\n",
                     "print(\"\\n=== Installing Clean Architecture Package ===\")\n",
-                    "import subprocess\n",
                     "try:\n",
-                    "    # Installs the local code in editable mode with all optional dependencies\n",
                     "    subprocess.check_call([sys.executable, \"-m\", \"pip\", \"install\", \"-e\", \".[all]\"])\n",
                     "    print(\"Package & dependencies installed successfully.\")\n",
                     "except Exception as e:\n",
                     "    print(f\"Error installing package dependencies: {e}\")\n",
                     "    sys.exit(1)\n",
                     "\n",
-                    "# 4. Start the Message Consumer\n",
+                    "# 5. Launch Consumer Daemon\n",
                     "print(\"\\n=== Launching Consumer Daemon ===\")\n",
                     "try:\n",
                     "    subprocess.check_call([sys.executable, \"-m\", \"src.interface.consumer.run_consumer\"])\n",