Update hf_bucket_provider.py
Browse files- hf_bucket_provider.py +55 -89
hf_bucket_provider.py
CHANGED
|
@@ -1,76 +1,68 @@
|
|
| 1 |
import os
|
| 2 |
import tempfile
|
| 3 |
import time
|
| 4 |
-
from typing import List, Dict, Any
|
| 5 |
from huggingface_hub import (
|
| 6 |
batch_bucket_files,
|
| 7 |
download_bucket_files,
|
| 8 |
list_bucket_tree,
|
| 9 |
)
|
| 10 |
-
from wsgidav.dav_provider import DAVProvider,
|
| 11 |
from wsgidav.dav_error import DAVError, HTTP_FORBIDDEN, HTTP_NOT_FOUND
|
| 12 |
|
| 13 |
|
| 14 |
-
class
|
| 15 |
-
"""表示 Hugging Face Bucket 中的文件
|
| 16 |
|
| 17 |
-
def __init__(self, path, environ, provider):
|
| 18 |
super().__init__(path, environ)
|
| 19 |
self.provider = provider
|
| 20 |
-
self.
|
| 21 |
-
|
| 22 |
-
def _get_info(self):
|
| 23 |
-
if self._info is None:
|
| 24 |
-
self._info = self.provider._get_file_info(self.path)
|
| 25 |
-
return self._info
|
| 26 |
-
|
| 27 |
-
@property
|
| 28 |
-
def exists(self):
|
| 29 |
-
info = self._get_info()
|
| 30 |
-
return info.get('exists', False)
|
| 31 |
|
| 32 |
def get_content_length(self):
|
| 33 |
-
|
| 34 |
-
if not info or info.get('is_dir', False):
|
| 35 |
-
return None
|
| 36 |
-
return info.get('size', 0)
|
| 37 |
|
| 38 |
def get_content_type(self):
|
| 39 |
-
return None
|
| 40 |
|
| 41 |
def get_creation_date(self):
|
| 42 |
-
|
| 43 |
-
return info.get('last_modified', 0) if info else 0
|
| 44 |
|
| 45 |
def get_display_name(self):
|
| 46 |
return self.name
|
| 47 |
|
| 48 |
def get_etag(self):
|
| 49 |
-
|
| 50 |
-
if
|
| 51 |
-
return f'"{info["last_modified"]}"'
|
| 52 |
-
return None
|
| 53 |
|
| 54 |
def get_last_modified(self):
|
| 55 |
-
|
| 56 |
-
|
|
|
|
|
|
|
| 57 |
|
| 58 |
-
def
|
| 59 |
-
|
| 60 |
-
return info.get('is_dir', False) if info else False
|
| 61 |
|
| 62 |
-
def is_link(self):
|
| 63 |
-
return False
|
| 64 |
|
| 65 |
-
|
| 66 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
|
| 68 |
-
def
|
| 69 |
-
return
|
| 70 |
|
| 71 |
def get_member_names(self):
|
| 72 |
-
if not self.is_collection():
|
| 73 |
-
return []
|
| 74 |
return self.provider._list_directory(self.path)
|
| 75 |
|
| 76 |
def get_member(self, name):
|
|
@@ -84,54 +76,18 @@ class HfBucketResource(_DAVResource):
|
|
| 84 |
|
| 85 |
def create_empty_resource(self, name):
|
| 86 |
new_path = self.util.join_uri(self.path, name)
|
|
|
|
| 87 |
return self.provider.get_resource_inst(new_path, self.environ)
|
| 88 |
|
| 89 |
def delete(self):
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
member = self.get_member(member_name)
|
| 95 |
-
member.delete()
|
| 96 |
-
else:
|
| 97 |
-
self.provider._delete_file(self.path)
|
| 98 |
-
|
| 99 |
-
def move_recursive(self, dest_path):
|
| 100 |
-
self.copy_recursive(dest_path)
|
| 101 |
-
self.delete()
|
| 102 |
-
|
| 103 |
-
def copy_recursive(self, dest_path):
|
| 104 |
-
if not self.exists:
|
| 105 |
-
raise DAVError(HTTP_NOT_FOUND)
|
| 106 |
-
if self.is_collection():
|
| 107 |
-
dest_resource = self.provider.get_resource_inst(dest_path, self.environ)
|
| 108 |
-
if not dest_resource.exists:
|
| 109 |
-
self.provider._create_collection(dest_path)
|
| 110 |
-
dest_resource = self.provider.get_resource_inst(dest_path, self.environ)
|
| 111 |
-
for member_name in self.get_member_names():
|
| 112 |
-
member = self.get_member(member_name)
|
| 113 |
-
dest_member_path = self.util.join_uri(dest_path, member_name)
|
| 114 |
-
member.copy_recursive(dest_member_path)
|
| 115 |
-
else:
|
| 116 |
-
self.provider._copy_file(self.path, dest_path)
|
| 117 |
-
|
| 118 |
-
def get_content(self):
|
| 119 |
-
if not self.exists:
|
| 120 |
-
raise DAVError(HTTP_NOT_FOUND)
|
| 121 |
-
if self.is_collection():
|
| 122 |
-
raise DAVError(HTTP_FORBIDDEN)
|
| 123 |
-
return self.provider._get_file_content(self.path)
|
| 124 |
-
|
| 125 |
-
def begin_write(self, content_type=None):
|
| 126 |
-
if not self.exists:
|
| 127 |
-
# 新文件允许写入
|
| 128 |
-
pass
|
| 129 |
-
if self.is_collection():
|
| 130 |
-
raise DAVError(HTTP_FORBIDDEN)
|
| 131 |
-
return HfBucketWriteFile(self.provider, self.path)
|
| 132 |
|
| 133 |
|
| 134 |
class HfBucketWriteFile:
|
|
|
|
| 135 |
def __init__(self, provider, path):
|
| 136 |
self.provider = provider
|
| 137 |
self.path = path
|
|
@@ -147,6 +103,8 @@ class HfBucketWriteFile:
|
|
| 147 |
|
| 148 |
|
| 149 |
class HfBucketProvider(DAVProvider):
|
|
|
|
|
|
|
| 150 |
def __init__(self, bucket_id: str, token: str):
|
| 151 |
super().__init__()
|
| 152 |
self.bucket_id = bucket_id
|
|
@@ -159,8 +117,8 @@ class HfBucketProvider(DAVProvider):
|
|
| 159 |
path = path[:-1]
|
| 160 |
return path
|
| 161 |
|
| 162 |
-
def _get_file_info(self, path: str) -> Dict[str, Any]:
|
| 163 |
-
"""返回
|
| 164 |
path = self._normalize_path(path)
|
| 165 |
if path == '/':
|
| 166 |
return {'exists': True, 'is_dir': True, 'size': 0, 'last_modified': time.time()}
|
|
@@ -190,9 +148,9 @@ class HfBucketProvider(DAVProvider):
|
|
| 190 |
}
|
| 191 |
if items:
|
| 192 |
return {'exists': True, 'is_dir': True, 'size': 0, 'last_modified': time.time()}
|
| 193 |
-
return
|
| 194 |
except Exception:
|
| 195 |
-
return
|
| 196 |
|
| 197 |
def _list_directory(self, path: str) -> List[str]:
|
| 198 |
path = self._normalize_path(path)
|
|
@@ -279,8 +237,16 @@ class HfBucketProvider(DAVProvider):
|
|
| 279 |
os.unlink(tmp.name)
|
| 280 |
|
| 281 |
def get_resource_inst(self, path, environ):
|
| 282 |
-
|
| 283 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 284 |
|
| 285 |
def exists(self, path, environ):
|
| 286 |
-
return self.
|
|
|
|
| 1 |
import os
|
| 2 |
import tempfile
|
| 3 |
import time
|
| 4 |
+
from typing import List, Dict, Any, Optional
|
| 5 |
from huggingface_hub import (
|
| 6 |
batch_bucket_files,
|
| 7 |
download_bucket_files,
|
| 8 |
list_bucket_tree,
|
| 9 |
)
|
| 10 |
+
from wsgidav.dav_provider import DAVProvider, DAVNonCollection, DAVCollection
|
| 11 |
from wsgidav.dav_error import DAVError, HTTP_FORBIDDEN, HTTP_NOT_FOUND
|
| 12 |
|
| 13 |
|
| 14 |
+
class HfBucketFile(DAVNonCollection):
|
| 15 |
+
"""表示 Hugging Face Bucket 中的文件"""
|
| 16 |
|
| 17 |
+
def __init__(self, path, environ, provider, info):
|
| 18 |
super().__init__(path, environ)
|
| 19 |
self.provider = provider
|
| 20 |
+
self.info = info
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
def get_content_length(self):
|
| 23 |
+
return self.info.get('size', 0)
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
def get_content_type(self):
|
| 26 |
+
return None # 由 wsgidav 根据扩展名自动判断
|
| 27 |
|
| 28 |
def get_creation_date(self):
|
| 29 |
+
return self.info.get('last_modified', 0)
|
|
|
|
| 30 |
|
| 31 |
def get_display_name(self):
|
| 32 |
return self.name
|
| 33 |
|
| 34 |
def get_etag(self):
|
| 35 |
+
lm = self.info.get('last_modified', 0)
|
| 36 |
+
return f'"{lm}"' if lm else None
|
|
|
|
|
|
|
| 37 |
|
| 38 |
def get_last_modified(self):
|
| 39 |
+
return self.info.get('last_modified', 0)
|
| 40 |
+
|
| 41 |
+
def get_content(self):
|
| 42 |
+
return self.provider._get_file_content(self.path)
|
| 43 |
|
| 44 |
+
def begin_write(self, content_type=None):
|
| 45 |
+
return HfBucketWriteFile(self.provider, self.path)
|
|
|
|
| 46 |
|
|
|
|
|
|
|
| 47 |
|
| 48 |
+
class HfBucketFolder(DAVCollection):
|
| 49 |
+
"""表示 Hugging Face Bucket 中的目录"""
|
| 50 |
+
|
| 51 |
+
def __init__(self, path, environ, provider, info):
|
| 52 |
+
super().__init__(path, environ)
|
| 53 |
+
self.provider = provider
|
| 54 |
+
self.info = info
|
| 55 |
+
|
| 56 |
+
def get_creation_date(self):
|
| 57 |
+
return self.info.get('last_modified', 0)
|
| 58 |
+
|
| 59 |
+
def get_display_name(self):
|
| 60 |
+
return self.name
|
| 61 |
|
| 62 |
+
def get_last_modified(self):
|
| 63 |
+
return self.info.get('last_modified', 0)
|
| 64 |
|
| 65 |
def get_member_names(self):
|
|
|
|
|
|
|
| 66 |
return self.provider._list_directory(self.path)
|
| 67 |
|
| 68 |
def get_member(self, name):
|
|
|
|
| 76 |
|
| 77 |
def create_empty_resource(self, name):
|
| 78 |
new_path = self.util.join_uri(self.path, name)
|
| 79 |
+
# 空文件将在写入时创建,这里只需返回资源对象
|
| 80 |
return self.provider.get_resource_inst(new_path, self.environ)
|
| 81 |
|
| 82 |
def delete(self):
|
| 83 |
+
# 递归删除所有子项
|
| 84 |
+
for member_name in self.get_member_names():
|
| 85 |
+
member = self.get_member(member_name)
|
| 86 |
+
member.delete()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
|
| 88 |
|
| 89 |
class HfBucketWriteFile:
|
| 90 |
+
"""辅助类:处理文件写入"""
|
| 91 |
def __init__(self, provider, path):
|
| 92 |
self.provider = provider
|
| 93 |
self.path = path
|
|
|
|
| 103 |
|
| 104 |
|
| 105 |
class HfBucketProvider(DAVProvider):
|
| 106 |
+
"""将 Hugging Face Bucket 适配为 DAVProvider"""
|
| 107 |
+
|
| 108 |
def __init__(self, bucket_id: str, token: str):
|
| 109 |
super().__init__()
|
| 110 |
self.bucket_id = bucket_id
|
|
|
|
| 117 |
path = path[:-1]
|
| 118 |
return path
|
| 119 |
|
| 120 |
+
def _get_file_info(self, path: str) -> Optional[Dict[str, Any]]:
|
| 121 |
+
"""获取文件/目录信息,如果不存在则返回 None"""
|
| 122 |
path = self._normalize_path(path)
|
| 123 |
if path == '/':
|
| 124 |
return {'exists': True, 'is_dir': True, 'size': 0, 'last_modified': time.time()}
|
|
|
|
| 148 |
}
|
| 149 |
if items:
|
| 150 |
return {'exists': True, 'is_dir': True, 'size': 0, 'last_modified': time.time()}
|
| 151 |
+
return None
|
| 152 |
except Exception:
|
| 153 |
+
return None
|
| 154 |
|
| 155 |
def _list_directory(self, path: str) -> List[str]:
|
| 156 |
path = self._normalize_path(path)
|
|
|
|
| 237 |
os.unlink(tmp.name)
|
| 238 |
|
| 239 |
def get_resource_inst(self, path, environ):
|
| 240 |
+
"""返回资源实例(文件或目录)"""
|
| 241 |
+
info = self._get_file_info(path)
|
| 242 |
+
if info is None:
|
| 243 |
+
# 路径不存在,返回一个占位的文件资源(允许后续写入)
|
| 244 |
+
# 使用默认信息
|
| 245 |
+
info = {'exists': False, 'is_dir': False, 'size': 0, 'last_modified': 0}
|
| 246 |
+
if info['is_dir']:
|
| 247 |
+
return HfBucketFolder(path, environ, self, info)
|
| 248 |
+
else:
|
| 249 |
+
return HfBucketFile(path, environ, self, info)
|
| 250 |
|
| 251 |
def exists(self, path, environ):
|
| 252 |
+
return self._get_file_info(path) is not None
|