untangle files that have been accidentally overwritten in a versioning-enabled s3 bucket

untangle_s3.py

# this code will untangle files that have been accidentally overwritten in a versioning-enabled s3 bucket
# contact [email protected] for assistance

import boto.s3.connection, sys
from collections import defaultdict

accessKey="XXXXXXXX/XXXXXX.XXXX"
secretKey="XXXXXXX/XXXXXXXXXXX.XXXXXXXXX.XXXXXXXXXX"
srcKeyPrefix="path/to/key" #no leading or trailing slash
srcBucket="destination-bucket-name-here"
dstBucket="destination-bucket-name-here"
offline_list = defaultdict(list) #we have to save the s3 structure to a local var to deal with paging in the API

print "opening connection"
scon = boto.s3.connection.S3Connection(accessKey, secretKey)

print "opening bucket"
bucket = scon.get_bucket(srcBucket)

print "storing list of bucket contents locally"
bucket_list = bucket.list(prefix=srcKeyPrefix)
for i in bucket_list:
    for j in bucket.list_versions(prefix=i.name):
        offline_list[i.name].append(j.version_id)

print "copying all versioned files into a flat directory"
for q in offline_list:
    count=0
    while len(offline_list[q])>0:
        key_version = offline_list[q].pop()
        nameParts = q.rsplit('.', 1)
        key_to_copy = bucket.get_key(q, version_id=key_version)
        print "cp: " + key_to_copy.name + " v: " + key_to_copy.version_id + " to: " + nameParts[0] + chr(len(offline_list[q])+65) + '.' + nameParts[1]
        print "mod: " + key_to_copy.last_modified
        key_to_copy.copy(dstBucket, nameParts[0] + chr(len(offline_list[q])+65) + '.' + nameParts[1])
        count+=1