def download(model_endpoint):
"""
model_endpoint is instance of one of:
* pmworker.endpoint.DocumentUrl
* pmworker.endpoint.PageUrl.
Will download Document original file or Page associated .txt file
from remote S3 location (shared storage) to local MEDIA_ROOT.
This function makes sense only in case of settings.S3=True i.e.
only if S3 media storage is enabled.
"""
if model_endpoint.exists():
return True
remote_abspath = model_endpoint.url(ep=Endpoint.S3)
local_abspath = model_endpoint.url()
local_dirname = os.path.dirname(local_abspath)
if not os.path.exists(os.path.dirname(local_abspath)):
logger.debug(f"{local_dirname} does not exists. Creating.")
os.makedirs(local_dirname, exist_ok=True)
else:
logger.debug(f"{local_dirname} already exists.")
# if not s3_download:
# logger.debug("s3_download=False. Use test data dir.")
# logger.debug(f"Using file {test_local_alternative}")
# copy2doc_url(
# src_file_path=test_local_alternative,
# doc_url=model_endpoint.url()
# )
s3_client = boto3.client('s3')
bucketname = get_bucketname(model_endpoint.url(ep=Endpoint.S3))
keyname = get_keyname(model_endpoint.url(ep=Endpoint.S3))
if not model_endpoint.exists(ep=Endpoint.S3):
logger.info(f"Endpoint s3:/{bucketname}/{keyname} missing")
return False
try:
logger.debug((f"Downloading {remote_abspath} to {local_abspath}"))
logger.debug(f"Uploading to bucket={bucketname} "
f"Keyname={keyname} "
f"Local={local_abspath}.")
s3_client.download_file(bucketname, keyname, local_abspath)
except botocore.exceptions.ClientError:
logger.error(
f"Error while downloading "
f" {remote_abspath} to {local_abspath}"
f" bucketname={bucketname} keyname={keyname}",
exc_info=True)
return False
return True
def remove_file(doc_url):
"""
Will remove actually the directory associated with that user
"""
s3 = boto3.resource('s3')
bucketname = get_bucketname(doc_url)
keyname = get_keyname(doc_url)
bucket = s3.Bucket(bucketname)
bucket.Object(keyname).delete()
def upload_img(page_url):
s3_url = page_url.img_url(ep=Endpoint.S3)
img_url = page_url.img_url()
bucketname = get_bucketname(s3_url)
keyname = get_keyname(s3_url)
s3_client = boto3.client('s3')
logger.debug(f"Uploading to bucket={bucketname} "
f"Keyname={keyname} "
f"img_url={img_url} "
f"s3_url={s3_url}.")
s3_client.upload_file(img_url, bucketname, keyname)
def upload_document_to_s3(doc_ep):
s3_url = doc_ep.url(ep=Endpoint.S3)
local_url = doc_ep.url()
bucketname = get_bucketname(s3_url)
keyname = get_keyname(s3_url)
s3_client = boto3.client('s3')
if not os.path.exists(local_url):
raise ValueError(f"{local_url} path does not exits")
logger.debug(f"upload_document {local_url} to s3:/{bucketname}{keyname}")
s3_client.upload_file(local_url, bucketname, keyname)
def download_hocr(page_ep):
if page_ep.hocr_exists():
return True
remote_abspath = page_ep.hocr_url(ep=Endpoint.S3)
local_abspath = page_ep.hocr_url()
local_dirname = os.path.dirname(local_abspath)
if not os.path.exists(os.path.dirname(local_abspath)):
logger.debug(f"{local_dirname} does not exists. Creating.")
os.makedirs(local_dirname, exist_ok=True)
else:
logger.debug(f"{local_dirname} already exists.")
s3_client = boto3.client('s3')
bucketname = get_bucketname(page_ep.hocr_url(ep=Endpoint.S3))
keyname = get_keyname(page_ep.hocr_url(ep=Endpoint.S3))
if not s3_key_exists(remote_abspath):
logger.info(f"Endpoint s3:/{bucketname}/{keyname} missing")
return False
try:
logger.debug((f"Downloading {remote_abspath} to {local_abspath}"))
logger.debug(f"Uploading to bucket={bucketname} "
f"Keyname={keyname} "
f"Local={local_abspath}.")
s3_client.download_file(bucketname, keyname, local_abspath)
except botocore.exceptions.ClientError:
logger.error(
f"Error while downloading "
f" {remote_abspath} to {local_abspath}"
f" bucketname={bucketname} keyname={keyname}",
exc_info=True)
return False
return True
def test_getkeyname(self):
self.assertEqual(get_keyname("s3://my-bucket/some/path/to/x.pdf"),
"some/path/to/x.pdf")
self.assertEqual(get_keyname("s3:/my-bucket/some/path/to/x.pdf"),
"some/path/to/x.pdf")