def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(
context,
evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT')
or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
# Group parts by size and each group will be handled by one function execution.
total_num_of_parts, num_of_groups, num_of_parts_per_group = calc_groups(
evt["total_size"], evt["part_size"], evt["medium_file_limit"])
upload_id = dest_client.init_multipart_upload(evt["key"]).upload_id
return {
"upload_id": upload_id,
"total_num_of_parts": total_num_of_parts,
"groups": list(range(num_of_groups)),
"num_of_parts_per_group": num_of_parts_per_group
}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(
context,
evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT')
or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
# Download part of the file and upload as part
part_no = evt["part_no"]
part_size = evt["part_size"]
byte_range = ((part_no - 1) * part_size,
min(part_no * part_size, evt["total_size"]) - 1)
# src_client.get_object_to_file(evt["key"], "/tmp/testpart%d" % part_no, byte_range=byte_range)
object_stream = src_client.get_object(evt["key"], byte_range=byte_range)
res = dest_client.upload_part(evt["key"], evt["upload_id"], part_no,
object_stream)
return {"part_no": part_no, "etag": res.etag}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(context, evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT') or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
crcs = evt.get("crcs")
failed_crcs = []
start_time = time.time()
count = evt.get("count", 10)
base_size = evt.get("base_size", 1024)
def producer(queue):
for i in range(count):
queue.put(i)
def consumer(queue):
while queue.ok():
item = queue.get()
if item is None:
break
i = item
key = '%s/%d' % (evt["prefix"], i)
result = dest_client.head_object(key)
crc = result.headers["x-oss-hash-crc64ecma"]
if crcs != None:
if crc != str(crcs[i]):
logger.info("expected %s, actual %s", crcs[i], crc)
failed_crcs.append(i)
else:
result = src_client.head_object(key)
src_crc = result.headers["x-oss-hash-crc64ecma"]
if crc != str(src_crc):
logger.info("expected %s, actual %s", src_crc, crc)
failed_crcs.append(i)
task_q = TaskQueue(producer, [consumer] * 16)
task_q.run()
return {'failed_crcs': failed_crcs, 'success': len(failed_crcs) == 0}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(context, evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT') or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
copy(src_client, dest_client, evt["key"])
return {}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(
context,
evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT')
or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
parts = copy(gen_parts, src_client, dest_client, evt["key"],
evt["part_size"], evt["total_size"], evt["upload_id"],
evt["group_id"], evt["num_of_parts_per_group"],
evt["total_num_of_parts"])
return {"parts": parts}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
dest_bucket = evt["dest_bucket"]
dest_client = clients.get(dest_bucket)
if dest_client is None:
dest_client = get_oss_client(
context,
evt.get("dest_oss_endpoint") or os.environ.get('DEST_OSS_ENDPOINT')
or src_endpoint, dest_bucket, evt.get("dest_access_role"))
clients[dest_bucket] = dest_client
parts = []
for part in evt["parts"]:
parts.append(PartInfo(part["part_no"], part["etag"]))
dest_client.complete_multipart_upload(evt["key"], evt["upload_id"], parts)
return {}
def handler(event, context):
logger = logging.getLogger()
evt = json.loads(event)
logger.info("Handling event: %s", evt)
src_endpoint = 'https://oss-%s-internal.aliyuncs.com' % context.region
src_bucket = evt["src_bucket"]
src_client = clients.get(src_bucket)
if src_client is None:
src_client = get_oss_client(context, src_endpoint, src_bucket)
clients[src_bucket] = src_client
has_more = False
marker = evt["marker"]
group_threshold = evt.get("group_threshold", 50)
total_group_count = evt.get("total_group_count", 0)
medium_file_limit = evt.get("medium_file_limit")
small_file_limit = evt.get("small_file_limit")
small_group_total = 0
small_group = [
] # The total size of small files is less than medium_file_limit
small = [] # An array of small_group
medium = []
large = []
current_group_size = 0
leave_early = False
while True:
result = src_client.list_objects(prefix=evt["prefix"],
marker=marker,
delimiter=evt["delimiter"],
max_keys=500)
logger.info("Found %d objects", len(result.object_list))
marker = result.next_marker
has_more = result.is_truncated
# A function can process amount of files up to medium_file_limit
for i in range(0, len(result.object_list)):
obj = result.object_list[i]
logger.info("key: %s, size: %s, group size: %d", obj.key, obj.size,
current_group_size)
if (current_group_size * medium_file_limit + small_group_total +
obj.size + medium_file_limit -
1) // medium_file_limit > group_threshold:
# Leave early and override has_more and marker
has_more = True
leave_early = True
marker = result.object_list[i].key
break
# Group small files as many as possible but their total size should not exceed medium_file_limit
if obj.size <= small_file_limit:
if obj.size + small_group_total <= medium_file_limit:
small_group_total += obj.size
small_group.append(obj.key)
else:
small.append(small_group)
small_group_total = obj.size
small_group = []
small_group.append(obj.key)
current_group_size += 1
elif obj.size <= medium_file_limit:
medium.append([obj.key, obj.size])
current_group_size += 1
else:
large.append([obj.key, obj.size])
# The large file will be divided into small groups and each group size is up to medium_file_limit
current_group_size += (obj.size + medium_file_limit -
1) // medium_file_limit
if not has_more or leave_early:
break
if len(small_group) > 0:
small.append(small_group)
total_group_count += (current_group_size * medium_file_limit +
small_group_total + medium_file_limit -
1) // medium_file_limit
return {
"small":
small, # [["key1","key2","key3"],["key4","key5"]]
"medium":
medium, # [["key9",size],["key11",size]]
"large":
large, # [["key6",size],["key7",size]]
"has_more":
has_more,
"marker":
marker,
"total_group_count":
total_group_count,
"execution_name":
evt.get("execution_name", "") + "-" +
re.sub(r"[^a-zA-Z0-9-_]", "_", marker)
}
