Hide keyboard shortcuts

Hot-keys on this page

r m x p   toggle line displays

j k   next/prev highlighted chunk

0   (zero) top of page

1   (one) first highlighted chunk

1# This file is part of daf_butler. 

2# 

3# Developed for the LSST Data Management System. 

4# This product includes software developed by the LSST Project 

5# (http://www.lsst.org). 

6# See the COPYRIGHT file at the top-level directory of this distribution 

7# for details of code ownership. 

8# 

9# This program is free software: you can redistribute it and/or modify 

10# it under the terms of the GNU General Public License as published by 

11# the Free Software Foundation, either version 3 of the License, or 

12# (at your option) any later version. 

13# 

14# This program is distributed in the hope that it will be useful, 

15# but WITHOUT ANY WARRANTY; without even the implied warranty of 

16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 

17# GNU General Public License for more details. 

18# 

19# You should have received a copy of the GNU General Public License 

20# along with this program. If not, see <http://www.gnu.org/licenses/>. 

21 

22from __future__ import annotations 

23 

24import logging 

25import re 

26import tempfile 

27 

28__all__ = ('ButlerS3URI',) 

29 

30from typing import ( 

31 TYPE_CHECKING, 

32 Optional, 

33 Any, 

34 Callable, 

35 Iterator, 

36 List, 

37 Tuple, 

38 Union, 

39) 

40 

41from .utils import NoTransaction 

42from ._butlerUri import ButlerURI 

43from .s3utils import getS3Client, s3CheckFileExists, bucketExists 

44 

45from botocore.exceptions import ClientError 

46from http.client import ImproperConnectionState, HTTPException 

47from urllib3.exceptions import RequestError, HTTPError 

48 

49if TYPE_CHECKING: 49 ↛ 50line 49 didn't jump to line 50, because the condition on line 49 was never true

50 try: 

51 import boto3 

52 except ImportError: 

53 pass 

54 from ..datastore import DatastoreTransaction 

55 

56# https://pypi.org/project/backoff/ 

57try: 

58 import backoff 

59except ImportError: 

60 class Backoff(): 

61 @staticmethod 

62 def expo(func: Callable, *args: Any, **kwargs: Any) -> Callable: 

63 return func 

64 

65 @staticmethod 

66 def on_exception(func: Callable, *args: Any, **kwargs: Any) -> Callable: 

67 return func 

68 

69 backoff = Backoff 

70 

71# settings for "backoff" retry decorators. these retries are belt-and- 

72# suspenders along with the retries built into Boto3, to account for 

73# semantic differences in errors between S3-like providers. 

74retryable_io_errors = ( 

75 # http.client 

76 ImproperConnectionState, HTTPException, 

77 # urllib3.exceptions 

78 RequestError, HTTPError, 

79 # built-ins 

80 TimeoutError, ConnectionError) 

81retryable_client_errors = ( 

82 # botocore.exceptions 

83 ClientError, 

84 # built-ins 

85 PermissionError) 

86all_retryable_errors = retryable_client_errors + retryable_io_errors 

87max_retry_time = 60 

88 

89 

90log = logging.getLogger(__name__) 

91 

92 

93class ButlerS3URI(ButlerURI): 

94 """S3 URI implementation class.""" 

95 

96 @property 

97 def client(self) -> boto3.client: 

98 """Client object to address remote resource.""" 

99 # Defer import for circular dependencies 

100 return getS3Client() 

101 

102 @backoff.on_exception(backoff.expo, retryable_client_errors, max_time=max_retry_time) 

103 def exists(self) -> bool: 

104 """Check that the S3 resource exists.""" 

105 if self.is_root: 105 ↛ 107line 105 didn't jump to line 107, because the condition on line 105 was never true

106 # Only check for the bucket since the path is irrelevant 

107 return bucketExists(self.netloc) 

108 exists, _ = s3CheckFileExists(self, client=self.client) 

109 return exists 

110 

111 @backoff.on_exception(backoff.expo, retryable_client_errors, max_time=max_retry_time) 

112 def size(self) -> int: 

113 """Return the size of the resource in bytes.""" 

114 if self.dirLike: 114 ↛ 115line 114 didn't jump to line 115, because the condition on line 114 was never true

115 return 0 

116 _, sz = s3CheckFileExists(self, client=self.client) 

117 return sz 

118 

119 @backoff.on_exception(backoff.expo, retryable_client_errors, max_time=max_retry_time) 

120 def remove(self) -> None: 

121 """Remove the resource.""" 

122 # https://github.com/boto/boto3/issues/507 - there is no 

123 # way of knowing if the file was actually deleted except 

124 # for checking all the keys again, reponse is HTTP 204 OK 

125 # response all the time 

126 self.client.delete_object(Bucket=self.netloc, Key=self.relativeToPathRoot) 

127 

128 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

129 def read(self, size: int = -1) -> bytes: 

130 """Read the contents of the resource.""" 

131 args = {} 

132 if size > 0: 

133 args["Range"] = f"bytes=0-{size-1}" 

134 try: 

135 response = self.client.get_object(Bucket=self.netloc, 

136 Key=self.relativeToPathRoot, 

137 **args) 

138 except (self.client.exceptions.NoSuchKey, self.client.exceptions.NoSuchBucket) as err: 

139 raise FileNotFoundError(f"No such resource: {self}") from err 

140 body = response["Body"].read() 

141 response["Body"].close() 

142 return body 

143 

144 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

145 def write(self, data: bytes, overwrite: bool = True) -> None: 

146 """Write the supplied data to the resource.""" 

147 if not overwrite: 

148 if self.exists(): 148 ↛ 149line 148 didn't jump to line 149, because the condition on line 148 was never true

149 raise FileExistsError(f"Remote resource {self} exists and overwrite has been disabled") 

150 self.client.put_object(Bucket=self.netloc, Key=self.relativeToPathRoot, 

151 Body=data) 

152 

153 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

154 def mkdir(self) -> None: 

155 """Write a directory key to S3.""" 

156 if not bucketExists(self.netloc): 156 ↛ 157line 156 didn't jump to line 157, because the condition on line 156 was never true

157 raise ValueError(f"Bucket {self.netloc} does not exist for {self}!") 

158 

159 if not self.dirLike: 159 ↛ 160line 159 didn't jump to line 160, because the condition on line 159 was never true

160 raise ValueError(f"Can not create a 'directory' for file-like URI {self}") 

161 

162 # don't create S3 key when root is at the top-level of an Bucket 

163 if not self.path == "/": 163 ↛ exitline 163 didn't return from function 'mkdir', because the condition on line 163 was never false

164 self.client.put_object(Bucket=self.netloc, Key=self.relativeToPathRoot) 

165 

166 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

167 def _as_local(self) -> Tuple[str, bool]: 

168 """Download object from S3 and place in temporary directory. 

169 

170 Returns 

171 ------- 

172 path : `str` 

173 Path to local temporary file. 

174 temporary : `bool` 

175 Always returns `True`. This is always a temporary file. 

176 """ 

177 with tempfile.NamedTemporaryFile(suffix=self.getExtension(), delete=False) as tmpFile: 

178 self.client.download_fileobj(self.netloc, self.relativeToPathRoot, tmpFile) 

179 return tmpFile.name, True 

180 

181 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

182 def transfer_from(self, src: ButlerURI, transfer: str = "copy", 

183 overwrite: bool = False, 

184 transaction: Optional[Union[DatastoreTransaction, NoTransaction]] = None) -> None: 

185 """Transfer the current resource to an S3 bucket. 

186 

187 Parameters 

188 ---------- 

189 src : `ButlerURI` 

190 Source URI. 

191 transfer : `str` 

192 Mode to use for transferring the resource. Supports the following 

193 options: copy. 

194 overwrite : `bool`, optional 

195 Allow an existing file to be overwritten. Defaults to `False`. 

196 transaction : `DatastoreTransaction`, optional 

197 Currently unused. 

198 """ 

199 # Fail early to prevent delays if remote resources are requested 

200 if transfer not in self.transferModes: 

201 raise ValueError(f"Transfer mode '{transfer}' not supported by URI scheme {self.scheme}") 

202 

203 log.debug(f"Transferring {src} [exists: {src.exists()}] -> " 

204 f"{self} [exists: {self.exists()}] (transfer={transfer})") 

205 

206 if not overwrite and self.exists(): 

207 raise FileExistsError(f"Destination path '{self}' already exists.") 

208 

209 if transfer == "auto": 209 ↛ 210line 209 didn't jump to line 210, because the condition on line 209 was never true

210 transfer = self.transferDefault 

211 

212 if isinstance(src, type(self)): 

213 # Looks like an S3 remote uri so we can use direct copy 

214 # note that boto3.resource.meta.copy is cleverer than the low 

215 # level copy_object 

216 copy_source = { 

217 "Bucket": src.netloc, 

218 "Key": src.relativeToPathRoot, 

219 } 

220 self.client.copy_object(CopySource=copy_source, Bucket=self.netloc, Key=self.relativeToPathRoot) 

221 else: 

222 # Use local file and upload it 

223 with src.as_local() as local_uri: 

224 

225 # resource.meta.upload_file seems like the right thing 

226 # but we have a low level client 

227 with open(local_uri.ospath, "rb") as fh: 

228 self.client.put_object(Bucket=self.netloc, 

229 Key=self.relativeToPathRoot, Body=fh) 

230 

231 # This was an explicit move requested from a remote resource 

232 # try to remove that resource 

233 if transfer == "move": 233 ↛ 235line 233 didn't jump to line 235, because the condition on line 233 was never true

234 # Transactions do not work here 

235 src.remove() 

236 

237 @backoff.on_exception(backoff.expo, all_retryable_errors, max_time=max_retry_time) 

238 def walk(self, file_filter: Optional[Union[str, re.Pattern]] = None) -> Iterator[Union[List, 

239 Tuple[ButlerURI, 

240 List[str], 

241 List[str]]]]: 

242 """Walk the directory tree returning matching files and directories. 

243 

244 Parameters 

245 ---------- 

246 file_filter : `str` or `re.Pattern`, optional 

247 Regex to filter out files from the list before it is returned. 

248 

249 Yields 

250 ------ 

251 dirpath : `ButlerURI` 

252 Current directory being examined. 

253 dirnames : `list` of `str` 

254 Names of subdirectories within dirpath. 

255 filenames : `list` of `str` 

256 Names of all the files within dirpath. 

257 """ 

258 # We pretend that S3 uses directories and files and not simply keys 

259 if not (self.isdir() or self.is_root): 259 ↛ 260line 259 didn't jump to line 260, because the condition on line 259 was never true

260 raise ValueError(f"Can not walk a non-directory URI: {self}") 

261 

262 if isinstance(file_filter, str): 262 ↛ 263line 262 didn't jump to line 263, because the condition on line 262 was never true

263 file_filter = re.compile(file_filter) 

264 

265 s3_paginator = self.client.get_paginator('list_objects_v2') 

266 

267 # Limit each query to a single "directory" to match os.walk 

268 # We could download all keys at once with no delimiter and work 

269 # it out locally but this could potentially lead to large memory 

270 # usage for millions of keys. It will also make the initial call 

271 # to this method potentially very slow. If making this method look 

272 # like os.walk was not required, we could query all keys with 

273 # pagination and return them in groups of 1000, but that would 

274 # be a different interface since we can't guarantee we would get 

275 # them all grouped properly across the 1000 limit boundary. 

276 prefix = self.relativeToPathRoot if not self.is_root else "" 

277 prefix_len = len(prefix) 

278 dirnames = [] 

279 filenames = [] 

280 files_there = False 

281 

282 for page in s3_paginator.paginate(Bucket=self.netloc, Prefix=prefix, Delimiter="/"): 

283 # All results are returned as full key names and we must 

284 # convert them back to the root form. The prefix is fixed 

285 # and delimited so that is a simple trim 

286 

287 # Directories are reported in the CommonPrefixes result 

288 # which reports the entire key and must be stripped. 

289 found_dirs = [dir["Prefix"][prefix_len:] for dir in page.get("CommonPrefixes", ())] 

290 dirnames.extend(found_dirs) 

291 

292 found_files = [file["Key"][prefix_len:] for file in page.get("Contents", ())] 

293 if found_files: 

294 files_there = True 

295 if file_filter is not None: 

296 found_files = [f for f in found_files if file_filter.search(f)] 

297 

298 filenames.extend(found_files) 

299 

300 # Directories do not exist so we can't test for them. If no files 

301 # or directories were found though, this means that it effectively 

302 # does not exist and we should match os.walk() behavior and return 

303 # []. 

304 if not dirnames and not files_there: 304 ↛ 305line 304 didn't jump to line 305, because the condition on line 304 was never true

305 yield [] 

306 else: 

307 yield self, dirnames, filenames 

308 

309 for dir in dirnames: 

310 new_uri = self.join(dir) 

311 yield from new_uri.walk(file_filter)