Added backup of daily cboe

This commit is contained in:
Javier Rodríguez Chatruc
2019-08-21 18:38:25 -03:00
parent e6b3414cd1
commit 145797d9f2
4 changed files with 53 additions and 8 deletions
+3
View File
@@ -36,6 +36,9 @@ aggregate:
backup:
pipenv run python -m data_scraper -b
backup_daily:
pipenv run python -m data_scraper -d
bench:
pipenv run python backtester/test/run_benchmark.py
+5
View File
@@ -21,6 +21,9 @@ parser.add_argument(
help="Aggregate daily data files")
parser.add_argument(
"-b", "--backup", action="store_true", help="Backup files in S3 bucket")
parser.add_argument(
"-d","--backup_daily", action="store_true", help="Backup daily files in S3 bucket"
)
args = parser.parse_args()
module_dir = os.path.join(os.getcwd(), os.path.dirname(__file__))
@@ -36,6 +39,8 @@ if args.aggregate:
cboe.aggregate_monthly_data()
elif args.backup:
backup.backup_data()
elif args.backup_daily:
backup.backup_cboe_data_daily()
else:
if args.scraper == "tiingo":
scraper = tiingo
+44 -8
View File
@@ -14,13 +14,7 @@ def backup_data():
"""Uploads scraped files to S3 bucket.
Set bucket name in environment variable $S3_BUCKET
"""
try:
bucket_name = utils.get_environment_var("S3_BUCKET")
except EnvironmentError as e:
logger.error(str(e))
slack_notification("Backup failed. Set $S3_BUCKET env variable",
__name__)
raise e
bucket_name = get_bucket_name()
s3 = boto3.resource("s3")
bucket = s3.Bucket(bucket_name)
@@ -57,6 +51,37 @@ def backup_data():
msg = "Unable to backup symbols: " + ", ".join(done)
slack_notification(msg, __name__, status=Status.Warning)
def backup_cboe_data_daily():
"""Uploads daily cboe scraped files to S3 bucket.
Set bucket name in environment variable $S3_BUCKET
"""
bucket_name = get_bucket_name()
s3 = boto3.resource("s3")
bucket = s3.Bucket(bucket_name)
data_path = utils.get_save_data_path()
cboe_data = os.path.join(data_path, "cboe")
cboe_folders = []
if os.path.exists(cboe_data):
cboe_folders = [
os.path.join(cboe_data, folder) for folder in os.listdir(cboe_data)
if folder.endswith("daily")
]
done_cboe, fail_cboe = _upload_folders(
bucket, "cboe", cboe_folders, remove_files=False)
done = done_cboe
failed = fail_cboe
if len(done) > 0:
msg = "Successful backup of daily cboe symbols: " + ", ".join(done)
slack_notification(msg, __name__, status=Status.Success)
if len(failed) > 0:
msg = "Unable to backup daily symbols: " + ", ".join(done)
slack_notification(msg, __name__, status=Status.Warning)
def _upload_folders(bucket, scraper, folders, remove_files=False):
"""Uploads folders to S3 bucket and (optionally) removes old files"""
@@ -105,10 +130,21 @@ def _key_exists(bucket, key):
bucket.Object(key).load()
except ClientError as e:
return int(e.response["Error"]["Code"]) != 404
return False
return True
def _remove_old_files(bucket, prefix):
old_files = bucket.objects.filter(Prefix=prefix)
for file in old_files:
file.delete()
def get_bucket_name():
try:
bucket_name = utils.get_environment_var("S3_BUCKET")
except EnvironmentError as e:
logger.error(str(e))
slack_notification("Backup failed. Set $S3_BUCKET env variable",
__name__)
raise e
return bucket_name
+1
View File
@@ -1,4 +1,5 @@
0 19 * * 1-5 root cd /finance && run-task make scrape scraper=cboe
0 19 * * 1-5 root cd /finance && run-task make scrape scraper=tiingo
0 0 1 * * root cd /finance && run-task make aggregate; run-task make backup
0 0 * * 6 root cd /finance && run-task make backup_daily