Upload folder using huggingface_hub
Browse files- scripts/danbooru/create_tar.py +28 -0
- scripts/danbooru/danbooru-check.py +141 -0
- scripts/down_files.py +112 -0
- scripts/down_metadata.py +39 -0
scripts/danbooru/create_tar.py
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import tarfile
|
3 |
+
import shutil
|
4 |
+
from tqdm import tqdm
|
5 |
+
|
6 |
+
def create_tar_and_remove_dir(dirname, tar_name):
|
7 |
+
# Check if the directory exists
|
8 |
+
if not os.path.isdir(dirname):
|
9 |
+
print("The specified directory does not exist.")
|
10 |
+
return
|
11 |
+
|
12 |
+
# Create the tar file
|
13 |
+
with tarfile.open(tar_name, "w") as tar:
|
14 |
+
# Sort and add files to the tar file
|
15 |
+
for root, dirs, files in os.walk(dirname):
|
16 |
+
# Sort files for consistent ordering
|
17 |
+
for file in tqdm(sorted(files), desc="Creating tars..."):
|
18 |
+
full_path = os.path.join(root, file)
|
19 |
+
# Add the file to the tar archive
|
20 |
+
tar.add(full_path, arcname=os.path.relpath(full_path, dirname))
|
21 |
+
|
22 |
+
# Remove the original directory after archiving
|
23 |
+
# shutil.rmtree(dirname)
|
24 |
+
print(f"The directory {dirname} has been removed.")
|
25 |
+
|
26 |
+
# Example usage of the function
|
27 |
+
# Replace 'your_directory_name' with the actual directory name you want to tar and remove
|
28 |
+
create_tar_and_remove_dir("danbooru2/0202", f"danbooru-tars/0202.tar")
|
scripts/danbooru/danbooru-check.py
ADDED
@@ -0,0 +1,141 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import time, os, json
|
2 |
+
from tqdm import tqdm
|
3 |
+
from curl_cffi import requests
|
4 |
+
import concurrent.futures
|
5 |
+
from pathlib import Path
|
6 |
+
import tarfile
|
7 |
+
import shutil
|
8 |
+
from huggingface_hub import HfApi
|
9 |
+
|
10 |
+
os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
|
11 |
+
|
12 |
+
def main():
|
13 |
+
danbooru_path = Path("danbooru")
|
14 |
+
selected_range = set(['%03d' % i for i in [400]])
|
15 |
+
notexists, total = 0, 0
|
16 |
+
notexistsid = {}
|
17 |
+
with open("posts.json", "r") as f:
|
18 |
+
bar = tqdm(desc="Indexing files", ascii=True, leave=False)
|
19 |
+
cache = {}
|
20 |
+
data_size = {}
|
21 |
+
for d in selected_range:
|
22 |
+
data = []
|
23 |
+
fp = ('0' + d)
|
24 |
+
for file_path in (danbooru_path / fp).iterdir():
|
25 |
+
if file_path.is_file():
|
26 |
+
data.append(file_path.stem)
|
27 |
+
data_size[file_path.stem] = file_path.stat().st_size
|
28 |
+
bar.update(1)
|
29 |
+
cache[fp] = set(data)
|
30 |
+
|
31 |
+
bar = tqdm(desc="Checking files", total=6_857_737, ascii=True, leave=False)
|
32 |
+
for line in f:
|
33 |
+
post = json.loads(line)
|
34 |
+
file_id = post['id']
|
35 |
+
cutoff = str(file_id)[-3:]
|
36 |
+
if cutoff not in selected_range:
|
37 |
+
bar.update(1)
|
38 |
+
continue
|
39 |
+
cutoff = '0' + cutoff
|
40 |
+
assert (danbooru_path / cutoff).exists(), f"{(danbooru_path / cutoff)} not exixts"
|
41 |
+
exists = str(file_id) in cache[cutoff] and (data_size[str(file_id)] == post["file_size"] or int(file_id) < 5_020_995)
|
42 |
+
total += 1
|
43 |
+
if not exists and "file_url" in post:
|
44 |
+
notexists += 1
|
45 |
+
if cutoff not in notexistsid:
|
46 |
+
notexistsid[cutoff] = []
|
47 |
+
notexistsid[cutoff].append((
|
48 |
+
file_id, cutoff, post["file_url"],
|
49 |
+
))
|
50 |
+
# print(post["file_url"])
|
51 |
+
bar.update(1)
|
52 |
+
bar.set_postfix_str(f"not exists: {notexists}")
|
53 |
+
|
54 |
+
bar.close()
|
55 |
+
print(f"not exists: {notexists}, total: {total}")
|
56 |
+
|
57 |
+
|
58 |
+
with concurrent.futures.ThreadPoolExecutor(max_workers=4) as tar_executor:
|
59 |
+
for d in selected_range:
|
60 |
+
cut = '0' + d
|
61 |
+
if cut not in notexistsid:
|
62 |
+
tar_executor.submit(archive_and_upload, Path("danbooru") / cut, cut)
|
63 |
+
|
64 |
+
for key, group in notexistsid.items():
|
65 |
+
keybar = tqdm(desc=f"Downloading files in key={key}", total=len(group), position=1, ascii=True, leave=False)
|
66 |
+
ok = False
|
67 |
+
while not ok:
|
68 |
+
with concurrent.futures.ThreadPoolExecutor(max_workers=3) as executor:
|
69 |
+
for file_id, cutoff, file_url in group:
|
70 |
+
executor.submit(download, file_id, cutoff, file_url, keybar)
|
71 |
+
ok = True
|
72 |
+
for file_id, cutoff, file_url in group:
|
73 |
+
suffix = Path(file_url).suffix
|
74 |
+
if file_url != "" and not Path(f"danbooru/{cutoff}/{file_id}{suffix}").is_file():
|
75 |
+
ok = False
|
76 |
+
tar_executor.submit(archive_and_upload, Path(f"danbooru/{cutoff}"), cutoff)
|
77 |
+
print(f"Finished download group {cutoff}")
|
78 |
+
keybar.close()
|
79 |
+
|
80 |
+
def rm_tree(pth: Path):
|
81 |
+
for child in pth.iterdir():
|
82 |
+
if child.is_file():
|
83 |
+
child.unlink()
|
84 |
+
else:
|
85 |
+
rm_tree(child)
|
86 |
+
pth.rmdir()
|
87 |
+
|
88 |
+
def archive_and_upload(dirname, name):
|
89 |
+
tar_name = Path("danbooru-tars") / f"data-{name}.tar"
|
90 |
+
# Check if the directory exists
|
91 |
+
if not os.path.isdir(dirname):
|
92 |
+
print("The specified directory does not exist.")
|
93 |
+
return
|
94 |
+
|
95 |
+
# Create the tar file
|
96 |
+
print(f"Creating {tar_name}")
|
97 |
+
with tarfile.open(tar_name, "w") as tar:
|
98 |
+
# Sort and add files to the tar file
|
99 |
+
for root, dirs, files in os.walk(dirname):
|
100 |
+
# Sort files for consistent ordering
|
101 |
+
for file in tqdm(sorted(files), desc=f"Creating {tar_name}", ascii=True):
|
102 |
+
full_path = os.path.join(root, file)
|
103 |
+
# Add the file to the tar archive
|
104 |
+
tar.add(full_path, arcname=file)
|
105 |
+
|
106 |
+
# Remove the original directory after archiving
|
107 |
+
rm_tree(dirname)
|
108 |
+
print(f"The directory {dirname} has been removed.")
|
109 |
+
|
110 |
+
api = HfApi()
|
111 |
+
print(api.upload_file(
|
112 |
+
path_or_fileobj=tar_name,
|
113 |
+
path_in_repo=f"original/data-{name}.tar",
|
114 |
+
repo_id="nyanko7/danbooru2023",
|
115 |
+
repo_type="dataset",
|
116 |
+
))
|
117 |
+
Path(tar_name).unlink()
|
118 |
+
|
119 |
+
def download(idx, cutoff, file_url, bar):
|
120 |
+
suffix = Path(file_url).suffix
|
121 |
+
max_attempts = 3 # specify the maximum number of attempts
|
122 |
+
|
123 |
+
for attempt in range(max_attempts):
|
124 |
+
try:
|
125 |
+
r = requests.get(file_url, impersonate="chrome110", timeout=120)
|
126 |
+
if r.status_code == 200:
|
127 |
+
with open(f"danbooru/{cutoff}/{idx}{suffix}", "wb") as f:
|
128 |
+
f.write(r.content)
|
129 |
+
break # if the download is successful, break the loop
|
130 |
+
else:
|
131 |
+
print(f"Attempt {attempt+1} failed to download {file_url}: error {r.status_code}")
|
132 |
+
except Exception as e:
|
133 |
+
print(f"Attempt {attempt+1} failed to download {file_url}: error {e}")
|
134 |
+
|
135 |
+
time.sleep(1) # wait for 1 second before the next attempt
|
136 |
+
if attempt+1 == max_attempts:
|
137 |
+
print(f"Failed to download {file_url} after {max_attempts} attempts.")
|
138 |
+
bar.update(1)
|
139 |
+
|
140 |
+
if __name__ == "__main__":
|
141 |
+
main()
|
scripts/down_files.py
ADDED
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import time, os, json
|
2 |
+
from tqdm import tqdm
|
3 |
+
from curl_cffi import requests
|
4 |
+
import concurrent.futures
|
5 |
+
from pathlib import Path
|
6 |
+
import tarfile
|
7 |
+
from huggingface_hub import HfApi
|
8 |
+
|
9 |
+
os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
|
10 |
+
|
11 |
+
def main():
|
12 |
+
notexistsid = {}
|
13 |
+
yandere_data = json.loads(Path("posts.json").read_text())
|
14 |
+
for item in tqdm(yandere_data, desc="Processing yandere", ascii=True):
|
15 |
+
file_id = item["id"]
|
16 |
+
modula = int(file_id) % 1000
|
17 |
+
cutoff = str(modula).zfill(4)
|
18 |
+
if cutoff not in notexistsid:
|
19 |
+
notexistsid[cutoff] = []
|
20 |
+
notexistsid[cutoff].append((file_id, cutoff, item["file_url"]))
|
21 |
+
|
22 |
+
with concurrent.futures.ThreadPoolExecutor(max_workers=4) as tar_executor:
|
23 |
+
for key, group in notexistsid.items():
|
24 |
+
if key in [
|
25 |
+
'0862', '0863', '0865', '0867', '0869',
|
26 |
+
'0868', '0870', '0871', '0873',
|
27 |
+
'0874', '0876', '0877', '0878', '0879',
|
28 |
+
'0880', '0881', '0882', '0883', '0884', '0885',
|
29 |
+
'0886', '0887', '0889', '0890', '0891',
|
30 |
+
'0892', '0893', '0894', '0896', '0897', '0898',
|
31 |
+
'0899', '0900', '0901', '0902', '0903'
|
32 |
+
]:
|
33 |
+
continue
|
34 |
+
keybar = tqdm(desc=f"Downloading files in key={key}", total=len(group), position=1, ascii=True, leave=False)
|
35 |
+
os.makedirs(f"yandere/{key}/", exist_ok=True)
|
36 |
+
ok = False
|
37 |
+
while not ok:
|
38 |
+
with concurrent.futures.ThreadPoolExecutor(max_workers=12) as executor:
|
39 |
+
for file_id, cutoff, file_url in group:
|
40 |
+
executor.submit(download, file_id, cutoff, file_url, keybar)
|
41 |
+
ok = True
|
42 |
+
for file_id, cutoff, file_url in group:
|
43 |
+
suffix = Path(file_url).suffix
|
44 |
+
if file_url != "" and not Path(f"yandere/{cutoff}/{file_id}{suffix}").is_file():
|
45 |
+
ok = False
|
46 |
+
tar_executor.submit(archive_and_upload, Path(f"yandere/{cutoff}"), cutoff)
|
47 |
+
print(f"Finished download group {cutoff}")
|
48 |
+
keybar.close()
|
49 |
+
|
50 |
+
def rm_tree(pth: Path):
|
51 |
+
for child in pth.iterdir():
|
52 |
+
if child.is_file():
|
53 |
+
child.unlink()
|
54 |
+
else:
|
55 |
+
rm_tree(child)
|
56 |
+
pth.rmdir()
|
57 |
+
|
58 |
+
def archive_and_upload(dirname, name):
|
59 |
+
tar_name = Path("yandere-tars") / f"data-{name}.tar"
|
60 |
+
# Check if the directory exists
|
61 |
+
if not os.path.isdir(dirname):
|
62 |
+
print("The specified directory does not exist.")
|
63 |
+
return
|
64 |
+
|
65 |
+
# Create the tar file
|
66 |
+
print(f"Creating {tar_name}")
|
67 |
+
with tarfile.open(tar_name, "w") as tar:
|
68 |
+
# Sort and add files to the tar file
|
69 |
+
for root, dirs, files in os.walk(dirname):
|
70 |
+
# Sort files for consistent ordering
|
71 |
+
for file in tqdm(sorted(files), desc=f"Creating {tar_name}", ascii=True):
|
72 |
+
full_path = os.path.join(root, file)
|
73 |
+
# Add the file to the tar archive
|
74 |
+
tar.add(full_path, arcname=file)
|
75 |
+
|
76 |
+
# Remove the original directory after archiving
|
77 |
+
rm_tree(dirname)
|
78 |
+
print(f"The directory {dirname} has been removed.")
|
79 |
+
|
80 |
+
api = HfApi()
|
81 |
+
print(api.upload_file(
|
82 |
+
path_or_fileobj=tar_name,
|
83 |
+
path_in_repo=f"original/data-{name}.tar",
|
84 |
+
repo_id="nyanko7/yandere2023",
|
85 |
+
repo_type="dataset",
|
86 |
+
))
|
87 |
+
Path(tar_name).unlink()
|
88 |
+
|
89 |
+
def download(idx, cutoff, file_url, bar):
|
90 |
+
suffix = Path(file_url).suffix
|
91 |
+
max_attempts = 5 # specify the maximum number of attempts
|
92 |
+
|
93 |
+
for attempt in range(max_attempts):
|
94 |
+
try:
|
95 |
+
r = requests.get(file_url, impersonate="chrome110", timeout=120)
|
96 |
+
if r.status_code == 200:
|
97 |
+
with open(f"yandere/{cutoff}/{idx}{suffix}", "wb") as f:
|
98 |
+
f.write(r.content)
|
99 |
+
break # if the download is successful, break the loop
|
100 |
+
else:
|
101 |
+
print(f"Attempt {attempt+1} failed to download {file_url}: error {r.status_code}")
|
102 |
+
except Exception as e:
|
103 |
+
print(f"Attempt {attempt+1} failed to download {file_url}: error {e}")
|
104 |
+
|
105 |
+
time.sleep(1) # wait for 1 second before the next attempt
|
106 |
+
if attempt+1 == max_attempts:
|
107 |
+
print(f"Failed to download {file_url} after {max_attempts} attempts.")
|
108 |
+
bar.update(1)
|
109 |
+
|
110 |
+
|
111 |
+
if __name__ == "__main__":
|
112 |
+
main()
|
scripts/down_metadata.py
ADDED
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from curl_cffi import requests
|
2 |
+
import json
|
3 |
+
from concurrent.futures import ThreadPoolExecutor, as_completed
|
4 |
+
from tqdm.rich import tqdm
|
5 |
+
|
6 |
+
# 定义爬取函数
|
7 |
+
def fetch_url(page):
|
8 |
+
url = f"https://yande.re/post.json?limit=100&page={page}"
|
9 |
+
response = requests.get(url, impersonate="chrome110")
|
10 |
+
# 检查请求是否成功
|
11 |
+
if response.status_code == 200:
|
12 |
+
return response.json()
|
13 |
+
else:
|
14 |
+
print(f"Failed to retrieve data from {url}")
|
15 |
+
return None
|
16 |
+
|
17 |
+
# 定义保存为 JSON 的函数
|
18 |
+
def save_json(data, page,):
|
19 |
+
if data:
|
20 |
+
with open(f"jsonpart/page_{page}.json", 'w', encoding='utf-8') as f:
|
21 |
+
json.dump(data, f, ensure_ascii=False, indent=4)
|
22 |
+
|
23 |
+
# 使用 ThreadPoolExecutor 创建线程池来并发爬取
|
24 |
+
def main():
|
25 |
+
bar = tqdm(total=25000)
|
26 |
+
pages = range(25001) # 0 到 25000 页
|
27 |
+
# 设置线程池的最大线程数,例如此处设置为 20
|
28 |
+
with ThreadPoolExecutor(max_workers=20) as executor:
|
29 |
+
# 创建 future 到 URL 的映射
|
30 |
+
future_to_url = {executor.submit(fetch_url, page): page for page in pages}
|
31 |
+
# 等待线程完成
|
32 |
+
for future in as_completed(future_to_url):
|
33 |
+
page = future_to_url[future]
|
34 |
+
data = future.result()
|
35 |
+
save_json(data, page)
|
36 |
+
bar.update(1)
|
37 |
+
|
38 |
+
if __name__ == "__main__":
|
39 |
+
main()
|