|
1
|
import boto
|
|
2
|
import boto.s3.connection
|
|
3
|
import sys
|
|
4
|
import pymongo
|
|
5
|
from pymongo import MongoClient
|
|
6
|
import os.path
|
|
7
|
import time
|
|
8
|
|
|
9
|
|
|
10
|
|
|
11
|
def file_exist(bucket, file_path):
|
|
12
|
c =bucket.get_key(file_path)
|
|
13
|
if c is not None and c.size > 0:
|
|
14
|
return True
|
|
15
|
return False
|
|
16
|
|
|
17
|
def exportItemForMongoCollection(obsId, db, bucket, log_file):
|
|
18
|
destination_collection =db[ 's3_'+obsId[:36]]
|
|
19
|
source_collection = db[ obsId[:36]]
|
|
20
|
print 'Start to exporting objectStore :%s '%obsId
|
|
21
|
i = 0
|
|
22
|
skip = 0
|
|
23
|
last_percentage = 0
|
|
24
|
total = source_collection.estimated_document_count()
|
|
25
|
for item in source_collection.find(no_cursor_timeout=True):
|
|
26
|
fs_path = item['fsPath']
|
|
27
|
objectId = item['id']
|
|
28
|
dest_item = destination_collection.find_one({'id':objectId})
|
|
29
|
if dest_item is None:
|
|
30
|
if os.path.isfile(fs_path):
|
|
31
|
i += 1
|
|
32
|
if not file_exist(bucket, '%s/%s'%(obsId,objectId)):
|
|
33
|
key = bucket.new_key('%s/%s'%(obsId,objectId))
|
|
34
|
try:
|
|
35
|
key.set_contents_from_filename(fs_path)
|
|
36
|
except Exception as e:
|
|
37
|
time.sleep(10)
|
|
38
|
key.set_contents_from_filename(fs_path)
|
|
39
|
item.pop('_id', None)
|
|
40
|
item.pop('fsPath')
|
|
41
|
item['uri'] = 's3://%s/%s'%(bucket.name, key.name)
|
|
42
|
destination_collection.insert_one(item)
|
|
43
|
if i % 100 == 0:
|
|
44
|
print "Exported %i/%i (skipped)%i "%(i, total, skip)
|
|
45
|
else:
|
|
46
|
log_file.writeline('Missing file for objectStoreid: %s ObjectId:%s path: %s'%(obsId, objectId, fs_path))
|
|
47
|
else:
|
|
48
|
skip += 1
|
|
49
|
|
|
50
|
|
|
51
|
def start_import(metadataCollection, bucket, log_file, done_file, skip_store):
|
|
52
|
client = MongoClient()
|
|
53
|
db = client['objectStore']
|
|
54
|
metadataCollection = db[metadataCollection]
|
|
55
|
for item in metadataCollection.find(no_cursor_timeout=True):
|
|
56
|
obsId = item['obsId']
|
|
57
|
if obsId not in skip_store:
|
|
58
|
exportItemForMongoCollection(obsId, db, bucket, log_file)
|
|
59
|
destination_collection =db[ 's3_'+obsId[:36]]
|
|
60
|
print "creating Index on ID"
|
|
61
|
destination_collection.create_index([('id',pymongo.ASCENDING)])
|
|
62
|
done_file.write('{}\n'.format(obsId))
|
|
63
|
|
|
64
|
|
|
65
|
|
|
66
|
|
|
67
|
|
|
68
|
if __name__=='__main__':
|
|
69
|
args = sys.argv
|
|
70
|
if not len(args) is not 3 :
|
|
71
|
print "Error on applying script usage: python scriptName.py s3cfgPath objectstoreBucket"
|
|
72
|
f = open(args[1])
|
|
73
|
props = {}
|
|
74
|
for line in f:
|
|
75
|
d =line.split('=')
|
|
76
|
if len(d) == 2:
|
|
77
|
props[d[0].strip()] = d[1].strip()
|
|
78
|
skip_store =[]
|
|
79
|
if os.path.isfile('store_done'):
|
|
80
|
f = open('store_done')
|
|
81
|
skip_store =[s.strip() for line in f if len(line) >0]
|
|
82
|
|
|
83
|
bname = args[2]
|
|
84
|
conn = boto.connect_s3(aws_access_key_id = props['access_key'], aws_secret_access_key = props['secret_key'], host = props['host_base'], calling_format = boto.s3.connection.OrdinaryCallingFormat())
|
|
85
|
bucket = conn.get_bucket(bname, validate=True)
|
|
86
|
log_file = open('s3_migration.log', 'w')
|
|
87
|
done_file = open('store_done', 'wb')
|
|
88
|
start_import('metadataObjectStore',bucket, log_file, done_file, skip_store)
|
|
89
|
log_file.close()
|