cloudnet-api-client 0.12.6__tar.gz → 0.12.8__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/dataportal.env +3 -3
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/docker-compose.yml +1 -1
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/workflows/test.yml +4 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/CHANGELOG.md +10 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/PKG-INFO +9 -9
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/README.md +8 -8
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/cloudnet_api_client/client.py +2 -2
- cloudnet_api_client-0.12.8/cloudnet_api_client/dl.py +175 -0
- cloudnet_api_client-0.12.8/cloudnet_api_client/version.py +1 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/test_client.py +15 -2
- cloudnet_api_client-0.12.6/cloudnet_api_client/dl.py +0 -108
- cloudnet_api_client-0.12.6/cloudnet_api_client/version.py +0 -1
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/db.env +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/initdb.d/init-dbs.sh +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/ss.env +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.github/workflows/publish.yml +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.gitignore +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/.pre-commit-config.yaml +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/LICENSE +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/cloudnet_api_client/__init__.py +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/cloudnet_api_client/containers.py +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/cloudnet_api_client/py.typed +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/cloudnet_api_client/utils.py +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/pyproject.toml +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20140205_hyytiala_classification.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250801_Magurele_CHM170137_000.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250803_JOYCE_WST_01m.dat +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250808_Granada_CHM170119_0045_000.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250808_hyytiala_iwc-Z-T-method.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250814_bucharest_classification.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250821_limassol_parsivel_41582c49.nc +0 -0
- {cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250822_leipzig-lim_ecmwf-open.nc +0 -0
|
@@ -8,8 +8,8 @@ TYPEORM_PORT=5432
|
|
|
8
8
|
TYPEORM_SYNCHRONIZE=false
|
|
9
9
|
TYPEORM_MIGRATIONS_RUN=true
|
|
10
10
|
TYPEORM_LOGGING=false
|
|
11
|
-
TYPEORM_ENTITIES=build/entity/*.js
|
|
12
|
-
TYPEORM_MIGRATIONS=build/migration/*.js
|
|
11
|
+
TYPEORM_ENTITIES=build/app/src/entity/*.js
|
|
12
|
+
TYPEORM_MIGRATIONS=build/app/src/migration/*.js
|
|
13
13
|
DP_SS_URL=http://storage-service:5900
|
|
14
14
|
DP_SS_USER=test
|
|
15
15
|
DP_SS_PASSWORD=test
|
|
@@ -24,6 +24,6 @@ DATACITE_API_TIMEOUT_MS=2000
|
|
|
24
24
|
DATACITE_DOI_SERVER=http://handle.datacite.test
|
|
25
25
|
DATACITE_DOI_PREFIX=XXX
|
|
26
26
|
LABELLING_URL=http://localhost:5803
|
|
27
|
-
|
|
27
|
+
INSTRUMENTDB_URL=http://localhost:5805
|
|
28
28
|
DVAS_URL=https://dvas.test
|
|
29
29
|
DC_URL=https://dc.test
|
|
@@ -15,7 +15,7 @@ services:
|
|
|
15
15
|
[
|
|
16
16
|
"sh",
|
|
17
17
|
"-c",
|
|
18
|
-
"node build/fixtures.js /backend-fixtures/backend/fixtures TRUNCATE && node build/fixtures.js /dataportal-fixtures APPEND && npm run start",
|
|
18
|
+
"node build/app/src/fixtures.js /backend-fixtures/backend/fixtures TRUNCATE && node build/app/src/fixtures.js /dataportal-fixtures APPEND && npm run start",
|
|
19
19
|
]
|
|
20
20
|
db:
|
|
21
21
|
image: "postgres:16"
|
|
@@ -5,6 +5,16 @@ All notable changes to this project will be documented in this file.
|
|
|
5
5
|
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
|
|
6
6
|
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
|
|
7
7
|
|
|
8
|
+
## 0.12.8 – 2026-03-24
|
|
9
|
+
|
|
10
|
+
- Support downloading of single metadata objects
|
|
11
|
+
- Hide total progress bar when downloading a single file
|
|
12
|
+
|
|
13
|
+
## 0.12.7 – 2025-12-18
|
|
14
|
+
|
|
15
|
+
- Adjust progress bars
|
|
16
|
+
- Write partial file
|
|
17
|
+
|
|
8
18
|
## 0.12.6 – 2025-09-17
|
|
9
19
|
|
|
10
20
|
- Allow other iterables besides `list` as argument
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: cloudnet-api-client
|
|
3
|
-
Version: 0.12.
|
|
3
|
+
Version: 0.12.8
|
|
4
4
|
Summary: Cloudnet API client
|
|
5
5
|
Author-email: Simo Tukiainen <simo.tukiainen@fmi.fi>
|
|
6
6
|
License-File: LICENSE
|
|
@@ -51,7 +51,7 @@ sites = client.sites()
|
|
|
51
51
|
site = client.site("hyytiala")
|
|
52
52
|
|
|
53
53
|
products = client.products()
|
|
54
|
-
product = client.
|
|
54
|
+
product = client.product("classification")
|
|
55
55
|
|
|
56
56
|
models = client.models()
|
|
57
57
|
model = client.model("ecmwf-open")
|
|
@@ -239,13 +239,13 @@ Download files from the fetched metadata.
|
|
|
239
239
|
|
|
240
240
|
Parameters:
|
|
241
241
|
|
|
242
|
-
| name | type
|
|
243
|
-
| ----------------- |
|
|
244
|
-
| metadata | `
|
|
245
|
-
| output_directory | `PathLike` or `str`
|
|
246
|
-
| concurrency_limit | `int`
|
|
247
|
-
| progress | `bool` or `None`
|
|
248
|
-
| validate_checksum | `bool`
|
|
242
|
+
| name | type | default |
|
|
243
|
+
| ----------------- | ----------------------------------------------- | ----------------- |
|
|
244
|
+
| metadata | `RawMetadata`, `ProductMetadata` or `list[...]` | |
|
|
245
|
+
| output_directory | `PathLike` or `str` | Current directory |
|
|
246
|
+
| concurrency_limit | `int` | 5 |
|
|
247
|
+
| progress | `bool` or `None` | `None` |
|
|
248
|
+
| validate_checksum | `bool` | `False` |
|
|
249
249
|
|
|
250
250
|
There's also an asynchronous version of this function:
|
|
251
251
|
`cloudnet_api_client.adownload`. It's useful for usage inside Jupyter notebook.
|
|
@@ -22,7 +22,7 @@ sites = client.sites()
|
|
|
22
22
|
site = client.site("hyytiala")
|
|
23
23
|
|
|
24
24
|
products = client.products()
|
|
25
|
-
product = client.
|
|
25
|
+
product = client.product("classification")
|
|
26
26
|
|
|
27
27
|
models = client.models()
|
|
28
28
|
model = client.model("ecmwf-open")
|
|
@@ -210,13 +210,13 @@ Download files from the fetched metadata.
|
|
|
210
210
|
|
|
211
211
|
Parameters:
|
|
212
212
|
|
|
213
|
-
| name | type
|
|
214
|
-
| ----------------- |
|
|
215
|
-
| metadata | `
|
|
216
|
-
| output_directory | `PathLike` or `str`
|
|
217
|
-
| concurrency_limit | `int`
|
|
218
|
-
| progress | `bool` or `None`
|
|
219
|
-
| validate_checksum | `bool`
|
|
213
|
+
| name | type | default |
|
|
214
|
+
| ----------------- | ----------------------------------------------- | ----------------- |
|
|
215
|
+
| metadata | `RawMetadata`, `ProductMetadata` or `list[...]` | |
|
|
216
|
+
| output_directory | `PathLike` or `str` | Current directory |
|
|
217
|
+
| concurrency_limit | `int` | 5 |
|
|
218
|
+
| progress | `bool` or `None` | `None` |
|
|
219
|
+
| validate_checksum | `bool` | `False` |
|
|
220
220
|
|
|
221
221
|
There's also an asynchronous version of this function:
|
|
222
222
|
`cloudnet_api_client.adownload`. It's useful for usage inside Jupyter notebook.
|
|
@@ -322,7 +322,7 @@ class APIClient:
|
|
|
322
322
|
|
|
323
323
|
def download(
|
|
324
324
|
self,
|
|
325
|
-
metadata: MetadataList,
|
|
325
|
+
metadata: MetadataList | TMetadata,
|
|
326
326
|
output_directory: str | PathLike = ".",
|
|
327
327
|
concurrency_limit: int = 5,
|
|
328
328
|
progress: bool | None = None,
|
|
@@ -340,7 +340,7 @@ class APIClient:
|
|
|
340
340
|
|
|
341
341
|
async def adownload(
|
|
342
342
|
self,
|
|
343
|
-
metadata: MetadataList,
|
|
343
|
+
metadata: MetadataList | TMetadata,
|
|
344
344
|
output_directory: str | PathLike = ".",
|
|
345
345
|
concurrency_limit: int = 5,
|
|
346
346
|
progress: bool | None = None,
|
|
@@ -0,0 +1,175 @@
|
|
|
1
|
+
import asyncio
|
|
2
|
+
import logging
|
|
3
|
+
from collections.abc import Iterable
|
|
4
|
+
from dataclasses import dataclass
|
|
5
|
+
from pathlib import Path
|
|
6
|
+
|
|
7
|
+
import aiohttp
|
|
8
|
+
from tqdm import tqdm
|
|
9
|
+
|
|
10
|
+
from cloudnet_api_client import utils
|
|
11
|
+
from cloudnet_api_client.containers import (
|
|
12
|
+
Metadata,
|
|
13
|
+
ProductMetadata,
|
|
14
|
+
)
|
|
15
|
+
|
|
16
|
+
|
|
17
|
+
class BarConfig:
|
|
18
|
+
def __init__(
|
|
19
|
+
self,
|
|
20
|
+
disable: bool | None,
|
|
21
|
+
max_workers: int,
|
|
22
|
+
total_bytes: int,
|
|
23
|
+
n_files: int,
|
|
24
|
+
) -> None:
|
|
25
|
+
self.disable = disable
|
|
26
|
+
self.single_file = n_files <= 1
|
|
27
|
+
self.position_queue = self._init_position_queue(max_workers)
|
|
28
|
+
self.total_amount = tqdm(
|
|
29
|
+
total=total_bytes,
|
|
30
|
+
desc="Progress",
|
|
31
|
+
unit="iB",
|
|
32
|
+
unit_scale=True,
|
|
33
|
+
unit_divisor=1024,
|
|
34
|
+
disable=self.disable if not self.single_file else True,
|
|
35
|
+
position=0,
|
|
36
|
+
leave=False,
|
|
37
|
+
colour="green",
|
|
38
|
+
)
|
|
39
|
+
self.lock = asyncio.Lock()
|
|
40
|
+
|
|
41
|
+
def _init_position_queue(self, max_workers: int) -> asyncio.Queue:
|
|
42
|
+
queue: asyncio.Queue = asyncio.Queue()
|
|
43
|
+
start = 0 if self.single_file else 1
|
|
44
|
+
for i in range(start, start + max_workers):
|
|
45
|
+
queue.put_nowait(i)
|
|
46
|
+
return queue
|
|
47
|
+
|
|
48
|
+
|
|
49
|
+
@dataclass
|
|
50
|
+
class DlParams:
|
|
51
|
+
url: str
|
|
52
|
+
destination: Path
|
|
53
|
+
session: aiohttp.ClientSession
|
|
54
|
+
semaphore: asyncio.Semaphore
|
|
55
|
+
bar_config: BarConfig
|
|
56
|
+
disable: bool | None
|
|
57
|
+
|
|
58
|
+
|
|
59
|
+
async def download_files(
|
|
60
|
+
base_url: str,
|
|
61
|
+
metadata: Iterable[Metadata] | Metadata,
|
|
62
|
+
output_path: Path,
|
|
63
|
+
concurrency_limit: int,
|
|
64
|
+
disable_progress: bool | None,
|
|
65
|
+
validate_checksum: bool = False,
|
|
66
|
+
) -> list[Path]:
|
|
67
|
+
metas = list(metadata) if isinstance(metadata, Iterable) else [metadata]
|
|
68
|
+
file_exists = _checksum_matches if validate_checksum else _size_and_name_matches
|
|
69
|
+
semaphore = asyncio.Semaphore(concurrency_limit)
|
|
70
|
+
total_bytes = sum(meta.size for meta in metas)
|
|
71
|
+
bar_config = BarConfig(disable_progress, concurrency_limit, total_bytes, len(metas))
|
|
72
|
+
full_paths = []
|
|
73
|
+
async with aiohttp.ClientSession() as session:
|
|
74
|
+
tasks = []
|
|
75
|
+
for meta in metas:
|
|
76
|
+
download_url = f"{base_url}{meta.download_url.split('/api/')[-1]}"
|
|
77
|
+
destination = output_path / meta.download_url.split("/")[-1]
|
|
78
|
+
full_paths.append(destination)
|
|
79
|
+
if destination.exists() and file_exists(meta, destination):
|
|
80
|
+
logging.debug(f"Already downloaded: {destination}")
|
|
81
|
+
continue
|
|
82
|
+
dl_params = DlParams(
|
|
83
|
+
url=download_url,
|
|
84
|
+
destination=destination,
|
|
85
|
+
session=session,
|
|
86
|
+
semaphore=semaphore,
|
|
87
|
+
bar_config=bar_config,
|
|
88
|
+
disable=disable_progress,
|
|
89
|
+
)
|
|
90
|
+
task = asyncio.create_task(_download_file_with_retries(dl_params))
|
|
91
|
+
tasks.append(task)
|
|
92
|
+
if disable_progress is True:
|
|
93
|
+
print(f"Downloading {len(metas)} files...", end="", flush=True)
|
|
94
|
+
await asyncio.gather(*tasks)
|
|
95
|
+
bar_config.total_amount.close()
|
|
96
|
+
bar_config.total_amount.clear()
|
|
97
|
+
if disable_progress is True:
|
|
98
|
+
print(" done.", flush=True)
|
|
99
|
+
return full_paths
|
|
100
|
+
|
|
101
|
+
|
|
102
|
+
async def _download_file_with_retries(
|
|
103
|
+
params: DlParams,
|
|
104
|
+
max_retries: int = 3,
|
|
105
|
+
) -> None:
|
|
106
|
+
"""Attempt to download a file, retrying up to max_retries times if needed."""
|
|
107
|
+
position = await params.bar_config.position_queue.get()
|
|
108
|
+
try:
|
|
109
|
+
for attempt in range(1, max_retries + 1):
|
|
110
|
+
try:
|
|
111
|
+
await _download_file(params, position)
|
|
112
|
+
return
|
|
113
|
+
except aiohttp.ClientError as e:
|
|
114
|
+
logging.warning(f"Attempt {attempt} failed for {params.url}: {e}")
|
|
115
|
+
if attempt == max_retries:
|
|
116
|
+
logging.error(
|
|
117
|
+
f"Giving up on {params.url} after {max_retries} attempts."
|
|
118
|
+
)
|
|
119
|
+
raise e
|
|
120
|
+
else:
|
|
121
|
+
# Exponential backoff before retrying
|
|
122
|
+
await asyncio.sleep(2**attempt)
|
|
123
|
+
finally:
|
|
124
|
+
params.bar_config.position_queue.put_nowait(position)
|
|
125
|
+
raise RuntimeError("Unreachable code reached.")
|
|
126
|
+
|
|
127
|
+
|
|
128
|
+
async def _download_file(
|
|
129
|
+
params: DlParams,
|
|
130
|
+
position: int,
|
|
131
|
+
) -> None:
|
|
132
|
+
tmp_path = params.destination.with_suffix(f"{params.destination.suffix}.part")
|
|
133
|
+
async with params.semaphore, params.session.get(params.url) as response:
|
|
134
|
+
response.raise_for_status()
|
|
135
|
+
bar = tqdm(
|
|
136
|
+
desc=params.destination.name,
|
|
137
|
+
total=response.content_length,
|
|
138
|
+
unit="iB",
|
|
139
|
+
unit_scale=True,
|
|
140
|
+
unit_divisor=1024,
|
|
141
|
+
disable=params.bar_config.disable,
|
|
142
|
+
position=position,
|
|
143
|
+
leave=False,
|
|
144
|
+
colour="cyan",
|
|
145
|
+
)
|
|
146
|
+
try:
|
|
147
|
+
tmp_path.parent.mkdir(parents=True, exist_ok=True)
|
|
148
|
+
with tmp_path.open("wb") as f:
|
|
149
|
+
async for chunk in response.content.iter_chunked(8192):
|
|
150
|
+
f.write(chunk)
|
|
151
|
+
bar.update(len(chunk))
|
|
152
|
+
params.bar_config.total_amount.update(len(chunk))
|
|
153
|
+
tmp_path.replace(params.destination)
|
|
154
|
+
except Exception:
|
|
155
|
+
try:
|
|
156
|
+
if tmp_path.exists():
|
|
157
|
+
tmp_path.unlink()
|
|
158
|
+
except OSError:
|
|
159
|
+
pass
|
|
160
|
+
raise
|
|
161
|
+
finally:
|
|
162
|
+
bar.close()
|
|
163
|
+
bar.clear()
|
|
164
|
+
|
|
165
|
+
|
|
166
|
+
def _checksum_matches(meta: Metadata, destination: Path) -> bool:
|
|
167
|
+
fun = utils.sha256sum if isinstance(meta, ProductMetadata) else utils.md5sum
|
|
168
|
+
return fun(destination) == meta.checksum
|
|
169
|
+
|
|
170
|
+
|
|
171
|
+
def _size_and_name_matches(meta: Metadata, destination: Path) -> bool:
|
|
172
|
+
return (
|
|
173
|
+
destination.stat().st_size == meta.size
|
|
174
|
+
and destination.name == meta.download_url.split("/")[-1]
|
|
175
|
+
)
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
__version__ = "0.12.8"
|
|
@@ -382,7 +382,7 @@ class TestRawFiles:
|
|
|
382
382
|
|
|
383
383
|
def test_filter_by_filename_suffix(self, client: APIClient):
|
|
384
384
|
meta = client.raw_files(filename_suffix="000.nc")
|
|
385
|
-
assert len(meta) ==
|
|
385
|
+
assert len(meta) == 3
|
|
386
386
|
|
|
387
387
|
def test_filter_by_instrument_id(self, client: APIClient):
|
|
388
388
|
meta = client.raw_files(instrument_id="weather-station")
|
|
@@ -462,6 +462,19 @@ class TestDownloadingFunctionality:
|
|
|
462
462
|
assert paths1 == paths2
|
|
463
463
|
assert paths2[0].stat().st_size == original_size
|
|
464
464
|
|
|
465
|
+
def test_downloading_single_metadata(self, client: APIClient, tmp_path: Path):
|
|
466
|
+
uuid = "ab872770-9136-4e61-8958-31e62abdfb1b"
|
|
467
|
+
meta = client.file(uuid)
|
|
468
|
+
paths = client.download(meta, output_directory=tmp_path, progress=False)
|
|
469
|
+
assert len(paths) == 1
|
|
470
|
+
assert paths[0].exists()
|
|
471
|
+
|
|
472
|
+
def test_downloading_single_metadata_II(self, client: APIClient, tmp_path: Path):
|
|
473
|
+
meta = client.raw_files(date_from="2025-08-01")
|
|
474
|
+
assert len(meta) == 3
|
|
475
|
+
paths = client.download(meta[0], output_directory=tmp_path, progress=False)
|
|
476
|
+
assert len(paths) == 1
|
|
477
|
+
|
|
465
478
|
async def test_async_download(self, client: APIClient, tmp_path: Path):
|
|
466
479
|
meta = client.raw_files(date_from="2025-08-01")
|
|
467
480
|
assert len(meta) == 3
|
|
@@ -487,7 +500,7 @@ class TestFilterCombinations:
|
|
|
487
500
|
assert len(meta) == 0
|
|
488
501
|
|
|
489
502
|
def test_partial_filename_matches(self, client: APIClient):
|
|
490
|
-
meta = client.raw_files(filename_prefix="
|
|
503
|
+
meta = client.raw_files(filename_prefix="202508", filename_suffix=".nc")
|
|
491
504
|
assert len(meta) == 2
|
|
492
505
|
|
|
493
506
|
|
|
@@ -1,108 +0,0 @@
|
|
|
1
|
-
import asyncio
|
|
2
|
-
import logging
|
|
3
|
-
from collections.abc import Iterable
|
|
4
|
-
from pathlib import Path
|
|
5
|
-
|
|
6
|
-
import aiohttp
|
|
7
|
-
from tqdm import tqdm
|
|
8
|
-
from tqdm.asyncio import tqdm_asyncio
|
|
9
|
-
|
|
10
|
-
from cloudnet_api_client import utils
|
|
11
|
-
from cloudnet_api_client.containers import Metadata, ProductMetadata
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
async def download_files(
|
|
15
|
-
base_url: str,
|
|
16
|
-
metadata: Iterable[Metadata],
|
|
17
|
-
output_path: Path,
|
|
18
|
-
concurrency_limit: int,
|
|
19
|
-
disable_progress: bool | None,
|
|
20
|
-
validate_checksum: bool = False,
|
|
21
|
-
) -> list[Path]:
|
|
22
|
-
file_exists = _checksum_matches if validate_checksum else _size_and_name_matches
|
|
23
|
-
semaphore = asyncio.Semaphore(concurrency_limit)
|
|
24
|
-
full_paths = []
|
|
25
|
-
async with aiohttp.ClientSession() as session:
|
|
26
|
-
tasks = []
|
|
27
|
-
for meta in metadata:
|
|
28
|
-
download_url = f"{base_url}{meta.download_url.split('/api/')[-1]}"
|
|
29
|
-
destination = output_path / meta.download_url.split("/")[-1]
|
|
30
|
-
full_paths.append(destination)
|
|
31
|
-
if destination.exists() and file_exists(meta, destination):
|
|
32
|
-
logging.debug(f"Already downloaded: {destination}")
|
|
33
|
-
continue
|
|
34
|
-
task = asyncio.create_task(
|
|
35
|
-
_download_file_with_retries(
|
|
36
|
-
session, download_url, destination, semaphore, disable_progress
|
|
37
|
-
)
|
|
38
|
-
)
|
|
39
|
-
tasks.append(task)
|
|
40
|
-
await tqdm_asyncio.gather(
|
|
41
|
-
*tasks, desc="Completed files", disable=disable_progress
|
|
42
|
-
)
|
|
43
|
-
return full_paths
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
async def _download_file_with_retries(
|
|
47
|
-
session: aiohttp.ClientSession,
|
|
48
|
-
url: str,
|
|
49
|
-
destination: Path,
|
|
50
|
-
semaphore: asyncio.Semaphore,
|
|
51
|
-
disable_progress: bool | None,
|
|
52
|
-
max_retries: int = 3,
|
|
53
|
-
) -> None:
|
|
54
|
-
"""Attempt to download a file, retrying up to max_retries times if needed."""
|
|
55
|
-
for attempt in range(1, max_retries + 1):
|
|
56
|
-
try:
|
|
57
|
-
await _download_file(session, url, destination, semaphore, disable_progress)
|
|
58
|
-
return
|
|
59
|
-
except aiohttp.ClientError as e:
|
|
60
|
-
logging.warning(f"Attempt {attempt} failed for {url}: {e}")
|
|
61
|
-
if attempt == max_retries:
|
|
62
|
-
logging.error(f"Giving up on {url} after {max_retries} attempts.")
|
|
63
|
-
raise e
|
|
64
|
-
else:
|
|
65
|
-
# Exponential backoff before retrying
|
|
66
|
-
await asyncio.sleep(2**attempt)
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
async def _download_file(
|
|
70
|
-
session: aiohttp.ClientSession,
|
|
71
|
-
url: str,
|
|
72
|
-
destination: Path,
|
|
73
|
-
semaphore: asyncio.Semaphore,
|
|
74
|
-
disable_progress: bool | None,
|
|
75
|
-
) -> None:
|
|
76
|
-
async with semaphore:
|
|
77
|
-
async with session.get(url) as response:
|
|
78
|
-
response.raise_for_status()
|
|
79
|
-
with (
|
|
80
|
-
destination.open("wb") as file_out,
|
|
81
|
-
tqdm(
|
|
82
|
-
desc=destination.name,
|
|
83
|
-
total=response.content_length,
|
|
84
|
-
unit="iB",
|
|
85
|
-
unit_scale=True,
|
|
86
|
-
unit_divisor=1024,
|
|
87
|
-
disable=disable_progress,
|
|
88
|
-
) as bar,
|
|
89
|
-
):
|
|
90
|
-
while True:
|
|
91
|
-
chunk = await response.content.read(8192)
|
|
92
|
-
if not chunk:
|
|
93
|
-
break
|
|
94
|
-
file_out.write(chunk)
|
|
95
|
-
bar.update(len(chunk))
|
|
96
|
-
logging.debug(f"Downloaded: {destination}")
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
def _checksum_matches(meta: Metadata, destination: Path) -> bool:
|
|
100
|
-
fun = utils.sha256sum if isinstance(meta, ProductMetadata) else utils.md5sum
|
|
101
|
-
return fun(destination) == meta.checksum
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
def _size_and_name_matches(meta: Metadata, destination: Path) -> bool:
|
|
105
|
-
return (
|
|
106
|
-
destination.stat().st_size == meta.size
|
|
107
|
-
and destination.name == meta.download_url.split("/")[-1]
|
|
108
|
-
)
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
__version__ = "0.12.6"
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
{cloudnet_api_client-0.12.6 → cloudnet_api_client-0.12.8}/tests/data/20250803_JOYCE_WST_01m.dat
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|