Format
Signed-off-by: András Schmelczer <andras@schmelczer.dev>
This commit is contained in:
parent
889e79174b
commit
60cd55c0cd
13 changed files with 168 additions and 116 deletions
|
|
@ -1,2 +1,2 @@
|
|||
from .progress_bar import DownloadProgressBar, UploadProgressBar
|
||||
from .human_readable_to_byte import human_readable_to_byte
|
||||
from .progress_bar import DownloadProgressBar, UploadProgressBar
|
||||
|
|
|
|||
|
|
@ -27,5 +27,5 @@ def human_readable_to_byte(size: str) -> int:
|
|||
|
||||
scalar = float(results["scalar"])
|
||||
idx = possible_units.index(results["unit"].upper())
|
||||
factor = 1024 ** idx
|
||||
factor = 1024**idx
|
||||
return round(scalar * factor)
|
||||
|
|
|
|||
|
|
@ -1,40 +1,35 @@
|
|||
import os
|
||||
import threading
|
||||
from logging import Logger
|
||||
from typing import IO, Any, Optional
|
||||
|
||||
from tqdm.auto import tqdm
|
||||
from pathlib import Path
|
||||
|
||||
import os
|
||||
import sys
|
||||
import threading
|
||||
|
||||
|
||||
class ProgressBar:
|
||||
def __init__(self, file_size: int, logger: Logger, prefix: str):
|
||||
self._file_size = file_size
|
||||
self._logger = logger
|
||||
self._prefix=prefix
|
||||
self._prefix = prefix
|
||||
|
||||
self._seen_so_far = 0
|
||||
self._lock = threading.Lock()
|
||||
|
||||
def __call__(self, bytes_amount: int):
|
||||
def __call__(self, bytes_amount: int) -> None:
|
||||
with self._lock:
|
||||
self._seen_so_far += bytes_amount
|
||||
percentage = (self._seen_so_far / float(self._file_size)) * 100
|
||||
size_length = len(str(self._file_size))
|
||||
progress = str(self._seen_so_far).rjust(size_length)
|
||||
self._logger.info(f"{self._prefix} {progress}/{self._file_size} bytes ({percentage:.1f}%)")
|
||||
self._logger.info(
|
||||
f"{self._prefix} {progress}/{self._file_size} bytes ({percentage:.1f}%)"
|
||||
)
|
||||
|
||||
|
||||
class DownloadProgressBar(ProgressBar):
|
||||
def __init__(self, name: str, size: int, logger: Logger):
|
||||
super().__init__(file_size=size, logger=logger, prefix=f'Downloading {name}')
|
||||
super().__init__(file_size=size, logger=logger, prefix=f"Downloading {name}")
|
||||
|
||||
|
||||
class UploadProgressBar(ProgressBar):
|
||||
def __init__(self, path: Path, logger: Logger):
|
||||
size = os.path.getsize(path)
|
||||
super().__init__(file_size=size, logger=logger, prefix=f'Uploading {path.name}')
|
||||
super().__init__(file_size=size, logger=logger, prefix=f"Uploading {path.name}")
|
||||
|
|
|
|||
|
|
@ -6,9 +6,8 @@ import tempfile
|
|||
from pathlib import Path
|
||||
from types import TracebackType
|
||||
from typing import IO, Any, Dict, List, Optional, Type, Union
|
||||
import boto3
|
||||
from botocore.exceptions import ClientError
|
||||
|
||||
import boto3
|
||||
from helper import DownloadProgressBar, UploadProgressBar, human_readable_to_byte
|
||||
|
||||
logger = logging.getLogger("open_s3")
|
||||
|
|
@ -62,7 +61,7 @@ class LargeFile:
|
|||
newline: Optional[str] = None,
|
||||
version: Optional[int] = None,
|
||||
keep_last_n: Optional[int] = None,
|
||||
offline_mode: bool = False
|
||||
offline_mode: bool = False,
|
||||
):
|
||||
self._name: str = name
|
||||
self._version = version
|
||||
|
|
@ -79,7 +78,7 @@ class LargeFile:
|
|||
|
||||
self._find_versions()
|
||||
self._check_mode_and_set_version()
|
||||
|
||||
|
||||
@classmethod
|
||||
def configure_credentials(
|
||||
cls,
|
||||
|
|
@ -154,7 +153,7 @@ class LargeFile:
|
|||
logger.exception("Could not finish operation.")
|
||||
|
||||
return True
|
||||
|
||||
|
||||
@property
|
||||
def version_ids(self) -> List[int]:
|
||||
return [self._get_version_from_key(key) for key in self._versions]
|
||||
|
|
@ -175,8 +174,17 @@ class LargeFile:
|
|||
with tempfile.TemporaryDirectory() as tmp:
|
||||
tmp_file_archive = Path(tmp) / f"{self._local_name}.tar.gz"
|
||||
|
||||
size = self._client.head_object(Bucket=self.bucket_name, Key=key)['ContentLength']
|
||||
self._client.download_file(Bucket=self.bucket_name, Key=key, Filename=str(tmp_file_archive), Callback=None if hide_progress else DownloadProgressBar(size=size, name=key, logger=logger))
|
||||
size = self._client.head_object(Bucket=self.bucket_name, Key=key)[
|
||||
"ContentLength"
|
||||
]
|
||||
self._client.download_file(
|
||||
Bucket=self.bucket_name,
|
||||
Key=key,
|
||||
Filename=str(tmp_file_archive),
|
||||
Callback=None
|
||||
if hide_progress
|
||||
else DownloadProgressBar(size=size, name=key, logger=logger),
|
||||
)
|
||||
logger.info(f"Decompressing {self._local_name}")
|
||||
shutil.unpack_archive(str(tmp_file_archive), tmp, "gztar")
|
||||
tmp_file = Path(tmp) / self._local_name
|
||||
|
|
@ -193,7 +201,7 @@ class LargeFile:
|
|||
with tempfile.TemporaryDirectory() as tmp:
|
||||
if path.is_file():
|
||||
logger.info(f"Copying file for {self._local_name}")
|
||||
copy = shutil.copy
|
||||
copy: Any = shutil.copy
|
||||
else:
|
||||
logger.info(f"Copying directory for {self._local_name}")
|
||||
copy = shutil.copytree
|
||||
|
|
@ -216,7 +224,14 @@ class LargeFile:
|
|||
logger.info(f"Uploading {self._local_name} to S3 from {path}")
|
||||
|
||||
file_to_be_uploaded = Path(tmp) / f"{self._local_name}.tar.gz"
|
||||
self._client.upload_file(Filename=str(file_to_be_uploaded), Bucket=self.bucket_name, Key=self._s3_name, Callback=None if hide_progress else UploadProgressBar(file_to_be_uploaded, logger=logger))
|
||||
self._client.upload_file(
|
||||
Filename=str(file_to_be_uploaded),
|
||||
Bucket=self.bucket_name,
|
||||
Key=self._s3_name,
|
||||
Callback=None
|
||||
if hide_progress
|
||||
else UploadProgressBar(file_to_be_uploaded, logger=logger),
|
||||
)
|
||||
|
||||
self.clean_up()
|
||||
|
||||
|
|
@ -239,15 +254,21 @@ class LargeFile:
|
|||
"Please configure the S3 access options by calling LargeFile.configure_credentials or set offline_mode=True in the constructor."
|
||||
)
|
||||
|
||||
self._client = boto3.client('s3', aws_access_key_id=self.access_key_id, aws_secret_access_key=self.secret_access_key, region_name=self.region_name, endpoint_url=self.endpoint_url)
|
||||
|
||||
self._client = boto3.client(
|
||||
"s3",
|
||||
aws_access_key_id=self.access_key_id,
|
||||
aws_secret_access_key=self.secret_access_key,
|
||||
region_name=self.region_name,
|
||||
endpoint_url=self.endpoint_url,
|
||||
)
|
||||
|
||||
def _find_versions(self) -> None:
|
||||
if self._offline_mode:
|
||||
self._fetch_versions_from_cache()
|
||||
else:
|
||||
self._create_client()
|
||||
self._fetch_versions_from_s3()
|
||||
|
||||
|
||||
if self._versions:
|
||||
logger.info(f"Found versions: {self.version_ids}")
|
||||
else:
|
||||
|
|
@ -257,8 +278,7 @@ class LargeFile:
|
|||
logger.info(f"Fetching offline versions of {self._name}")
|
||||
|
||||
self._versions = [
|
||||
path
|
||||
for path in self.cache_path.glob(f'{self._local_name}-*')
|
||||
path for path in self.cache_path.glob(f"{self._local_name}-*")
|
||||
]
|
||||
|
||||
def _fetch_versions_from_s3(self) -> None:
|
||||
|
|
@ -267,10 +287,7 @@ class LargeFile:
|
|||
Bucket=self.bucket_name, Prefix=self._name
|
||||
)
|
||||
self._versions = (
|
||||
sorted(
|
||||
o["Key"]
|
||||
for o in found_objects["Contents"]
|
||||
)
|
||||
sorted(o["Key"] for o in found_objects["Contents"])
|
||||
if "Contents" in found_objects
|
||||
else []
|
||||
)
|
||||
|
|
@ -309,22 +326,25 @@ class LargeFile:
|
|||
@property
|
||||
def _s3_name(self) -> str:
|
||||
return f"{self._name}/{self._version}"
|
||||
|
||||
|
||||
@staticmethod
|
||||
def _get_version_from_key(key: Union[str, Path]) -> int:
|
||||
if isinstance(key, Path):
|
||||
return int(key.name.split("-")[-1])
|
||||
return int(key.split("/")[-1])
|
||||
|
||||
|
||||
def _delete_old_versions_from_s3(self) -> None:
|
||||
if self._keep_last_n is not None:
|
||||
for key in (self._versions[: -self._keep_last_n] if self._keep_last_n > 0 else self._versions):
|
||||
for key in (
|
||||
self._versions[: -self._keep_last_n]
|
||||
if self._keep_last_n > 0
|
||||
else self._versions
|
||||
):
|
||||
logger.info(
|
||||
f"Removing old version (keep_last_n={self._keep_last_n}): {key}"
|
||||
)
|
||||
self._client.delete_object(Bucket=self.bucket_name, Key=key)
|
||||
|
||||
|
||||
def _delete_old_versions_from_disk(self) -> None:
|
||||
self.cache_path.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
|
|
@ -343,4 +363,4 @@ class LargeFile:
|
|||
logger.info(
|
||||
f"Deleting file from cache to meet quota (max_cache_size={self.max_cache_size}): {file}"
|
||||
)
|
||||
os.unlink(file)
|
||||
os.unlink(file)
|
||||
|
|
|
|||
|
|
@ -4,22 +4,22 @@ from src.open_s3.helper import human_readable_to_byte
|
|||
|
||||
|
||||
class TestHumanReadableToByte(unittest.TestCase):
|
||||
def test_simple_cases(self):
|
||||
def test_simple_cases(self) -> None:
|
||||
self.assertEqual(human_readable_to_byte("1KB"), 1024)
|
||||
self.assertEqual(human_readable_to_byte("2KB"), 2048)
|
||||
|
||||
def test_fractions(self):
|
||||
def test_fractions(self) -> None:
|
||||
self.assertEqual(human_readable_to_byte("0.5KB"), 512)
|
||||
self.assertEqual(human_readable_to_byte("20.5KB"), 1024 * 20 + 512)
|
||||
|
||||
def test_formating(self):
|
||||
def test_formating(self) -> None:
|
||||
self.assertEqual(human_readable_to_byte(" 1MB"), 1024 * 1024)
|
||||
self.assertEqual(human_readable_to_byte(" 2 MB"), 1024 * 1024 * 2)
|
||||
self.assertEqual(human_readable_to_byte(" 4 MB "), 1024 * 1024 * 4)
|
||||
self.assertEqual(human_readable_to_byte("8MB "), 1024 * 1024 * 8)
|
||||
self.assertEqual(human_readable_to_byte(" 1.5 MB "), 1024 * 1024 * 1.5)
|
||||
|
||||
def test_casing(self):
|
||||
def test_casing(self) -> None:
|
||||
self.assertEqual(human_readable_to_byte("0.5GB"), 0.5 * 1024 * 1024 * 1024)
|
||||
self.assertEqual(human_readable_to_byte("0.5gB"), 0.5 * 1024 * 1024 * 1024)
|
||||
self.assertEqual(human_readable_to_byte("0.5Gb"), 0.5 * 1024 * 1024 * 1024)
|
||||
|
|
|
|||
|
|
@ -1,5 +1,6 @@
|
|||
from pathlib import Path
|
||||
import unittest
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
from unittest.mock import Mock, create_autospec, patch
|
||||
|
||||
import botocore.session
|
||||
|
|
@ -14,23 +15,22 @@ credentials = {
|
|||
"aws_access_key_id": "YOUR_ACCESS_KEY_ID",
|
||||
"aws_secret_access_key": "YOUR_VERY_SECRET_ACCESS_KEY",
|
||||
"large_files_bucket_name": "create_a_bucket_and_put_its_name_here",
|
||||
"other_key": 23,
|
||||
"endpoint_url": "this is optional, for backblaze, use this: https://s3.us-west-002.backblazeb2.com",
|
||||
}
|
||||
|
||||
|
||||
class TestLargeFile(unittest.TestCase):
|
||||
def test_uninitialized(self):
|
||||
def test_uninitialized(self) -> None:
|
||||
self.assertRaises(ValueError, LargeFile, "test-file")
|
||||
|
||||
def test_bad_file_modes(self):
|
||||
def test_bad_file_modes(self) -> None:
|
||||
self.assertRaises(ValueError, LargeFile, "test-file", "w", version=3)
|
||||
self.assertRaises(ValueError, LargeFile, "test-file", "wb", version=3)
|
||||
self.assertRaises(ValueError, LargeFile, "test-file", "w+r")
|
||||
self.assertRaises(ValueError, LargeFile, "test-file", "test")
|
||||
|
||||
@patch("botocore.session")
|
||||
def test_initialized_with_dict(self, session):
|
||||
def test_initialized_with_dict(self, session) -> None:
|
||||
session_mock = Mock()
|
||||
session.get_session = create_autospec(
|
||||
botocore.session.get_session, return_value=session_mock
|
||||
|
|
@ -79,7 +79,7 @@ class TestLargeFile(unittest.TestCase):
|
|||
self.assertEqual(lf._s3_name, "test-file/2")
|
||||
|
||||
@patch("botocore.session")
|
||||
def test_initialized_with_file(self, session):
|
||||
def test_initialized_with_file(self, session: Any) -> None:
|
||||
session_mock = Mock()
|
||||
session.get_session = create_autospec(
|
||||
botocore.session.get_session, return_value=session_mock
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue