Script to update the metadata on Amazon S3

s3_meta.py

"""
To run the script
s3_meta.py --access-key=AWS_ACCESS_KEY_ID
           --secret-key=AWS_SECRET_ACCESS_KEY
           --bucket-name=BUCKET_NAME
           -d=METADATA
           
e.g.
python s3_meta.py --access_key=AKIAIOSFODNN7EXAMPLE --secret-key=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY --bucket-name=test-bucket -d '{"Cache-Control": "max-age=3600"}'
"""

# For Python2 Only

# General
import os, sys, json
import optparse
import mimetypes
from argparse import ArgumentParser

# Boto Imports
import boto
from boto.s3.connection import S3Connection

def get_input():
    """
    ----------
    Parameters
    ----------
        1. access_key
        2. secret_key
        3. bucket_name
        4. metadata
    """
    parser = optparse.OptionParser(
        usage=" %prog --access-key=AWS_ACCESS_KEY_ID "
        "--secret-key=AWS_SECRET_ACCESS_KEY "
        "--bucket-name=BUCKET_NAME "
        "--max-age=MAX_AGE "
        "-d=METADATA "
    )
    parser.add_option(
        '--access-key', dest='access_key', help="Access Key for AWS S3."
    )
    parser.add_option(
        '--secret-key', dest='secret_key', help="Secret Key for AWS S3."
    )
    parser.add_option(
        '--bucket-name', dest='bucket_name', help="AWS S3 Bucket Name."
    )
    parser.add_option(
        '-d', '--metadata', dest='metadata', default={},
        help="""Metadata to update. (Provide values in json format)"""
        """ e.g. '{"Cache-Control": "max-age=3600"}' """,
    )
    (options, args) = parser.parse_args()

    # Sanity check to ensure if all mandatory options are there.
    mandatories = ['access_key', 'secret_key', 'bucket_name', 'metadata']
    for item in mandatories:
        if not options.__dict__[item]:
            print("\n--{} option is mandatory.\n".format(item.replace("_", "-")))
            exit(-1)

    return options

def start_execution(options):
    """
    Connect to S3.
    Update the headers.
    """
    ALLOWED_CONTENT_TYPES = ('image/png', 'image/jpeg', )

    try:
        METADATA_TO_UPDATE = json.loads(options.metadata)
    except ValueError:
        print("\nInvalid data: {}\n".format(options.metadata))
        return

    try:
        connection = S3Connection(options.access_key, options.secret_key)
    except Exception as e:
        sys.stdout.write("Unable to connect to S3. An error occured: {}\n".format(e))
        sys.stdout.flush()
        return

    try:
        bucket = connection.get_bucket(options.bucket_name)
    except boto.exception.S3ResponseError as e:
        sys.stdout.write("An error occured: {}\n".format(e))
        sys.stdout.flush()
        return

    for key in bucket:
        content_type, encoding = mimetypes.guess_type(key.name)
        if content_type in ALLOWED_CONTENT_TYPES:
            sys.stdout.write("Updating... {}\n".format(key.name))
            key.set_remote_metadata(METADATA_TO_UPDATE, {}, True)

    sys.stdout.write("\nCaching completed.")



if __name__ == "__main__":
    options = get_input()
    start_execution(options)