Use zip in backup (#268)

* Refactor auto backup.

* Backup to .zip file

* Backup to .zip file

* Backup to .zip file

* Look for any .json file in the .zip file

* Add github action to check if the backup file was created.

* Create github action to check backup.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Test github action.

* Add logs.

* Add backup flag validation.

* Import no_backup var.

* Import no_backup var from config file.

* Enable tmate for debugging.

* Force error.

* Force error.

* Refactor.

* Remove tmate.

* Remove no backup variable from config.

* Improve messages.

* Improve messages.

* Improve messages.

* Check backup file size.

* Update slack channel.

Author: everaldorodrigo

.github/scripts/check_backup.py

@@ -0,0 +1,96 @@
+"""
+This script checks if a backup file for the current date exists in a specified S3 bucket.
+If the backup file does not exist, a notification is sent to a Slack channel.
+
+Expected file format in the S3 bucket:
+- The file should be in the folder 'db_backup/' with the following naming pattern:
+  'smartapi_YYYYMMDD.zip', where YYYYMMDD corresponds to the current date.
+
+Required Environment Variables:
+- AWS_ACCESS_KEY_ID: The AWS access key ID to read the AWS s3 bucket.
+- AWS_SECRET_ACCESS_KEY: The AWS secret access key  to read the AWS s3 bucket.
+- BACKUP_BUCKET_NAME: The name of the AWS S3 bucket where backups are stored.
+- S3_FOLDER: The folder path within the S3 bucket where backups are stored (e.g., 'db_backup/').
+- AWS_REGION: The AWS region where the S3 bucket is located.
+- SLACK_CHANNEL: The Slack channel where notifications should be sent (e.g., '#observability-test').
+- SLACK_WEBHOOK_URL: The Slack Webhook URL used to send the notification.
+
+Functionality:
+1. The script uses the AWS SDK (boto3) to check for the existence of the backup file in the specified S3 bucket.
+2. If the file is found, it logs that no action is needed.
+3. If the file is not found, it sends a notification to the configured Slack channel.
+
+Dependencies:
+- boto3: For interacting with AWS S3.
+- requests: For sending HTTP POST requests to Slack.
+
+"""
+
+import boto3
+import botocore
+import os
+import requests
+
+from datetime import datetime
+
+
+def send_slack_notification(message):
+
+    print(f" └─ {message}")
+
+    # Create the payload for Slack
+    slack_data = {
+        "channel": os.getenv("SLACK_CHANNEL"),
+        "username": "SmartAPI",
+        "icon_emoji": ":thumbsdown:",
+        "text": message,
+    }
+
+    try:
+        print(" └─ Sending Slack notification.")
+        response = requests.post(os.getenv("SLACK_WEBHOOK_URL"), json=slack_data, timeout=10)
+        if response.status_code == 200:
+            print(" └─ Slack notification sent successfully.")
+        else:
+            print(f" └─ Failed to send message to Slack: {response.status_code}, {response.text}")
+    except requests.exceptions.Timeout as e:
+        print(" └─ Request timed out to Slack WebHook URL.")
+        raise e
+    except requests.exceptions.RequestException as e:
+        print(f" └─ Failed to send Slack notification. Error: {str(e)}")
+        raise e
+
+
+def check_backup_file():
+
+    # Create the expected file name
+    today_date = datetime.today().strftime("%Y%m%d")
+    expected_file = f"{os.getenv('S3_FOLDER')}smartapi_{today_date}.zip"
+
+    # Create the S3 client
+    s3_client = boto3.client("s3", region_name=os.getenv("AWS_REGION"))
+
+    # Try to fetch the file metadata
+    try:
+        response = s3_client.head_object(Bucket=os.getenv("BACKUP_BUCKET_NAME"), Key=expected_file)
+        print(f" └─ Backup file {expected_file} exists!")
+
+        # Get the file size in bytes
+        file_size = response['ContentLength']
+
+        # Check if the file is larger than 1MB
+        if file_size > 1048576:  # 1MB in bytes
+            print(f" └─ Backup file is larger than 1MB! Size: {file_size} bytes.")
+            print(" └─ Nothing to do!")
+        else:
+            message = f":alert: The backup file {expected_file} is smaller than 1MB!"
+            send_slack_notification(message)
+
+    except botocore.exceptions.ClientError as e:
+        print(e)
+        message = f":alert: The backup file {expected_file} was NOT created today!"
+        send_slack_notification(message)
+
+
+if __name__ == "__main__":
+    check_backup_file()

.github/workflows/check_backup.yml

@@ -0,0 +1,35 @@
+name: Check S3 Backup and Notify Slack
+
+on:
+  workflow_dispatch:  # Allows manual trigger from GitHub Actions UI
+  schedule:
+    - cron: '0 13 * * *' # 5:00 AM PST (UTC-8)
+
+jobs:
+  check-backup:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v2
+
+      - name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: '3.x'
+
+      - name: Install boto3 (AWS SDK for Python)
+        run: |
+          python -m pip install --upgrade pip
+          pip install boto3 requests
+
+      - name: Check if backup exists in S3
+        run: python .github/scripts/check_backup.py
+        env:
+          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+          AWS_REGION: ${{ secrets.AWS_REGION }}
+          BACKUP_BUCKET_NAME: "${{ secrets.BACKUP_BUCKET_NAME }}"
+          S3_FOLDER: "db_backup/"
+          SLACK_CHANNEL: "#ncats-translator"
+          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}

src/admin.py

@@ -26,6 +26,8 @@
 import logging
 import random
 import time
+import zipfile
+import io
 from datetime import datetime
 
 import boto3
@@ -37,20 +39,48 @@
 logging.basicConfig(level="INFO")
 
 
-def _default_filename():
-    return "smartapi_" + datetime.today().strftime("%Y%m%d") + ".json"
+def _default_filename(extension=".json"):
+    return "smartapi_" + datetime.today().strftime("%Y%m%d") + extension
 
 
-def save_to_file(mapping, filename=None):
-    filename = filename or _default_filename()
-    with open(filename, "w") as file:
-        json.dump(mapping, file, indent=2)
-
-
-def save_to_s3(mapping, filename=None, bucket="smartapi"):
-    filename = filename or _default_filename()
+def save_to_file(mapping, filename=None, format="zip"):
+    """
+    Save data to a file in either JSON or ZIP format.
+    :param mapping: Data to save
+    :param filename: File name
+    :param format: File format, either 'json' or 'zip'
+    """
+    if format == "zip":
+        filename = filename or _default_filename(".zip")
+        with zipfile.ZipFile(filename, 'w', zipfile.ZIP_DEFLATED) as zfile:
+            json_data = json.dumps(mapping, indent=2)
+            zfile.writestr(filename.replace(".zip", ".json"), json_data)
+    else:
+        filename = filename or _default_filename(".json")
+        with open(filename, "w") as file:
+            json.dump(mapping, file, indent=2)
+
+
+def save_to_s3(data, filename=None, bucket="smartapi", format="zip"):
+    """
+    Save data to S3 in either JSON or ZIP format.
+    :param data: Data to save
+    :param filename: File name
+    :param bucket: S3 bucket name
+    :param format: File format, either 'json' or 'zip'
+    """
+    filename = filename or _default_filename(f".{format}")
     s3 = boto3.resource("s3")
-    s3.Bucket(bucket).put_object(Key="db_backup/{}".format(filename), Body=json.dumps(mapping, indent=2))
+
+    if format == "zip":
+        with zipfile.ZipFile(filename, 'w', zipfile.ZIP_DEFLATED) as zfile:
+            json_data = json.dumps(data, indent=2)
+            zfile.writestr(filename.replace(".zip", ".json"), json_data)
+        logging.info(f"Uploading {filename} to AWS S3")
+        s3.Bucket(bucket).upload_file(Filename=filename, Key=f"db_backup/{filename}")
+    else:
+        logging.info(f"Uploading {filename} to AWS S3")
+        s3.Bucket(bucket).put_object(Key=f"db_backup/{filename}", Body=json.dumps(data, indent=2))
 
 
 def _backup():
@@ -69,14 +99,14 @@ def _backup():
     return smartapis
 
 
-def backup_to_file(filename=None):
+def backup_to_file(filename=None, format="zip"):
     smartapis = _backup()
-    save_to_file(smartapis, filename)
+    save_to_file(smartapis, filename, format)
 
 
-def backup_to_s3(filename=None, bucket="smartapi"):
+def backup_to_s3(filename=None, bucket="smartapi", format="zip"):
     smartapis = _backup()
-    save_to_s3(smartapis, filename, bucket)
+    save_to_s3(smartapis, filename, bucket, format)
 
 
 def _restore(smartapis):
@@ -99,7 +129,7 @@ def restore_from_s3(filename=None, bucket="smartapi"):
     s3 = boto3.client("s3")
 
     if not filename:
-        objects = s3.list_objects_v2(Bucket="smartapi", Prefix="db_backup")["Contents"]
+        objects = s3.list_objects_v2(Bucket=bucket, Prefix="db_backup")["Contents"]
         filename = max(objects, key=lambda x: x["LastModified"])["Key"]
 
     if not filename.startswith("db_backup/"):
@@ -108,14 +138,42 @@ def restore_from_s3(filename=None, bucket="smartapi"):
     logging.info("GET s3://%s/%s", bucket, filename)
 
     obj = s3.get_object(Bucket=bucket, Key=filename)
-    smartapis = json.loads(obj["Body"].read())
+
+    filename = filename.replace("db_backup/", "")
+
+    if filename.endswith(".zip"):
+        file_content = obj["Body"].read()
+        with zipfile.ZipFile(io.BytesIO(file_content)) as zfile:
+            # Search for a JSON file inside the ZIP
+            json_file = next((f for f in zfile.namelist() if f.endswith(".json")), None)
+            if not json_file:
+                raise ValueError("No JSON file found inside the ZIP archive.")
+            with zfile.open(json_file) as json_data:
+                smartapis = json.load(json_data)
+    elif filename.endswith(".json"):
+        smartapis = json.loads(obj["Body"].read())
+    else:
+        raise Exception("Unsupported backup file type!")
+
     _restore(smartapis)
 
 
 def restore_from_file(filename):
-    with open(filename) as file:
-        smartapis = json.load(file)
-        _restore(smartapis)
+    if filename.endswith(".zip"):
+        with zipfile.ZipFile(filename, 'r') as zfile:
+            # Search for a JSON file inside the ZIP
+            json_file = next((f for f in zfile.namelist() if f.endswith(".json")), None)
+            if not json_file:
+                raise ValueError("No JSON file found inside the ZIP archive.")
+            with zfile.open(json_file) as json_data:
+                smartapis = json.load(json_data)
+    elif filename.endswith(".json"):
+        with open(filename) as file:
+            smartapis = json.load(file)
+    else:
+        raise Exception("Unsupported backup file type!")
+
+    _restore(smartapis)
 
 
 def refresh_document():
@@ -226,7 +284,7 @@ def refresh_has_metakg():
 _lock = FileLock(".lock", timeout=0)
 
 
-def routine(no_backup=False):
+def routine(no_backup=False, format="zip"):
     logger = logging.getLogger("routine")
 
     # Add jitter: random delay between 100 and 500 milliseconds (adjust range as needed)
@@ -244,8 +302,8 @@ def routine(no_backup=False):
         if lock_acquired:
             logger.info("Schedule lock acquired successfully.")
             if not no_backup:
-                logger.info("backup_to_s3()")
-                backup_to_s3()
+                logger.info(f"backup_to_s3(format={format})")
+                backup_to_s3(format=format)
             logger.info("refresh_document()")
             refresh_document()
             logger.info("check_uptime()")
@@ -262,6 +320,7 @@ def routine(no_backup=False):
         logger.warning("Schedule lock acquired by another process. No need to run it in this process.")
     except Exception as e:
         logger.error(f"An error occurred during the routine: {e}")
+        logger.error("Stack trace:", exc_info=True)
     finally:
         if lock_acquired:
             _lock.release()