1 # Copyright 2019 Canonical
3 # Licensed under the Apache License, Version 2.0 (the "License"); you may
4 # not use this file except in compliance with the License. You may obtain
5 # a copy of the License at
7 # http://www.apache.org/licenses/LICENSE-2.0
9 # Unless required by applicable law or agreed to in writing, software
10 # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
11 # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
12 # License for the specific language governing permissions and limitations
15 # For those usages not covered by the Apache License, Version 2.0 please
16 # contact: eduardo.sousa@canonical.com
20 from http
import HTTPStatus
21 from io
import BytesIO
, StringIO
26 from gridfs
import GridFSBucket
, errors
27 from osm_common
.fsbase
import FsBase
, FsException
28 from pymongo
import MongoClient
31 __author__
= "Eduardo Sousa <eduardo.sousa@canonical.com>"
34 class GridByteStream(BytesIO
):
35 def __init__(self
, filename
, fs
, mode
):
36 BytesIO
.__init
__(self
)
38 self
.filename
= filename
41 self
.file_type
= "file" # Set "file" as default file_type
45 def __initialize__(self
):
48 cursor
= self
.fs
.find({"filename": self
.filename
})
50 for requested_file
in cursor
:
51 exception_file
= next(cursor
, None)
55 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
58 if requested_file
.metadata
["type"] in ("file", "sym"):
59 grid_file
= requested_file
60 self
.file_type
= requested_file
.metadata
["type"]
63 "Type isn't file", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
67 self
._id
= grid_file
._id
68 self
.fs
.download_to_stream(self
._id
, self
)
75 super(GridByteStream
, self
).close()
79 self
.fs
.delete(self
._id
)
81 cursor
= self
.fs
.find(
82 {"filename": self
.filename
.split("/")[0], "metadata": {"type": "dir"}}
85 parent_dir
= next(cursor
, None)
88 parent_dir_name
= self
.filename
.split("/")[0]
89 self
.filename
= self
.filename
.replace(
90 parent_dir_name
, parent_dir_name
[:-1], 1
95 self
.fs
.upload_from_stream_with_id(
96 self
._id
, self
.filename
, self
, metadata
={"type": self
.file_type
}
99 self
.fs
.upload_from_stream(
100 self
.filename
, self
, metadata
={"type": self
.file_type
}
102 super(GridByteStream
, self
).close()
107 def __exit__(self
, exc_type
, exc_val
, exc_tb
):
111 class GridStringStream(StringIO
):
112 def __init__(self
, filename
, fs
, mode
):
113 StringIO
.__init
__(self
)
115 self
.filename
= filename
118 self
.file_type
= "file" # Set "file" as default file_type
120 self
.__initialize
__()
122 def __initialize__(self
):
125 cursor
= self
.fs
.find({"filename": self
.filename
})
127 for requested_file
in cursor
:
128 exception_file
= next(cursor
, None)
132 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
135 if requested_file
.metadata
["type"] in ("file", "dir"):
136 grid_file
= requested_file
137 self
.file_type
= requested_file
.metadata
["type"]
140 "File type isn't file", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
145 self
._id
= grid_file
._id
146 self
.fs
.download_to_stream(self
._id
, stream
)
148 self
.write(stream
.read().decode("utf-8"))
156 super(GridStringStream
, self
).close()
160 self
.fs
.delete(self
._id
)
162 cursor
= self
.fs
.find(
163 {"filename": self
.filename
.split("/")[0], "metadata": {"type": "dir"}}
166 parent_dir
= next(cursor
, None)
169 parent_dir_name
= self
.filename
.split("/")[0]
170 self
.filename
= self
.filename
.replace(
171 parent_dir_name
, parent_dir_name
[:-1], 1
176 stream
.write(self
.read().encode("utf-8"))
179 self
.fs
.upload_from_stream_with_id(
180 self
._id
, self
.filename
, stream
, metadata
={"type": self
.file_type
}
183 self
.fs
.upload_from_stream(
184 self
.filename
, stream
, metadata
={"type": self
.file_type
}
187 super(GridStringStream
, self
).close()
192 def __exit__(self
, exc_type
, exc_val
, exc_tb
):
196 class FsMongo(FsBase
):
197 def __init__(self
, logger_name
="fs", lock
=False):
198 super().__init
__(logger_name
, lock
)
203 def __update_local_fs(self
, from_path
=None):
204 dir_cursor
= self
.fs
.find({"metadata.type": "dir"}, no_cursor_timeout
=True)
208 for directory
in dir_cursor
:
209 if from_path
and not directory
.filename
.startswith(from_path
):
211 os
.makedirs(self
.path
+ directory
.filename
, exist_ok
=True)
212 valid_paths
.append(self
.path
+ directory
.filename
)
214 file_cursor
= self
.fs
.find(
215 {"metadata.type": {"$in": ["file", "sym"]}}, no_cursor_timeout
=True
218 for writing_file
in file_cursor
:
219 if from_path
and not writing_file
.filename
.startswith(from_path
):
221 file_path
= self
.path
+ writing_file
.filename
223 if writing_file
.metadata
["type"] == "sym":
225 self
.fs
.download_to_stream(writing_file
._id
, b
)
227 link
= b
.read().decode("utf-8")
232 if e
.errno
!= errno
.ENOENT
:
233 # This is probably permission denied or worse
235 os
.symlink(link
, file_path
)
237 folder
= os
.path
.dirname(file_path
)
238 if folder
not in valid_paths
:
239 os
.makedirs(folder
, exist_ok
=True)
240 with
open(file_path
, "wb+") as file_stream
:
241 self
.fs
.download_to_stream(writing_file
._id
, file_stream
)
242 if "permissions" in writing_file
.metadata
:
243 os
.chmod(file_path
, writing_file
.metadata
["permissions"])
245 def get_params(self
):
246 return {"fs": "mongo", "path": self
.path
}
248 def fs_connect(self
, config
):
250 if "logger_name" in config
:
251 self
.logger
= logging
.getLogger(config
["logger_name"])
253 self
.path
= config
["path"]
255 raise FsException('Missing parameter "path"')
256 if not self
.path
.endswith("/"):
258 if not os
.path
.exists(self
.path
):
260 "Invalid configuration param at '[storage]': path '{}' does not exist".format(
264 elif not os
.access(self
.path
, os
.W_OK
):
266 "Invalid configuration param at '[storage]': path '{}' is not writable".format(
270 if all(key
in config
.keys() for key
in ["uri", "collection"]):
271 self
.client
= MongoClient(config
["uri"])
272 self
.fs
= GridFSBucket(self
.client
[config
["collection"]])
273 elif all(key
in config
.keys() for key
in ["host", "port", "collection"]):
274 self
.client
= MongoClient(config
["host"], config
["port"])
275 self
.fs
= GridFSBucket(self
.client
[config
["collection"]])
277 if "collection" not in config
.keys():
278 raise FsException('Missing parameter "collection"')
280 raise FsException('Missing parameters: "uri" or "host" + "port"')
283 except Exception as e
: # TODO refine
284 raise FsException(str(e
))
286 def fs_disconnect(self
):
289 def mkdir(self
, folder
):
291 Creates a folder or parent object location
293 :return: None or raises an exception
296 self
.fs
.upload_from_stream(folder
, BytesIO(), metadata
={"type": "dir"})
297 except errors
.FileExists
: # make it idempotent
299 except Exception as e
:
300 raise FsException(str(e
), http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
)
302 def dir_rename(self
, src
, dst
):
304 Rename one directory name. If dst exist, it replaces (deletes) existing directory
305 :param src: source directory
306 :param dst: destination directory
307 :return: None or raises and exception
310 dst_cursor
= self
.fs
.find(
311 {"filename": {"$regex": "^{}(/|$)".format(dst
)}}, no_cursor_timeout
=True
314 for dst_file
in dst_cursor
:
315 self
.fs
.delete(dst_file
._id
)
317 src_cursor
= self
.fs
.find(
318 {"filename": {"$regex": "^{}(/|$)".format(src
)}}, no_cursor_timeout
=True
321 for src_file
in src_cursor
:
322 self
.fs
.rename(src_file
._id
, src_file
.filename
.replace(src
, dst
, 1))
323 except Exception as e
:
324 raise FsException(str(e
), http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
)
326 def file_exists(self
, storage
, mode
=None):
328 Indicates if "storage" file exist
329 :param storage: can be a str or a str list
330 :param mode: can be 'file' exist as a regular file; 'dir' exists as a directory or; 'None' just exists
333 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
335 cursor
= self
.fs
.find({"filename": f
})
337 for requested_file
in cursor
:
338 exception_file
= next(cursor
, None)
342 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
345 # if no special mode is required just check it does exists
349 if requested_file
.metadata
["type"] == mode
:
352 if requested_file
.metadata
["type"] == "sym" and mode
== "file":
357 def file_size(self
, storage
):
360 :param storage: can be a str or a str list
363 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
365 cursor
= self
.fs
.find({"filename": f
})
367 for requested_file
in cursor
:
368 exception_file
= next(cursor
, None)
372 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
375 return requested_file
.length
377 def file_extract(self
, tar_object
, path
):
380 :param tar_object: object of type tar
381 :param path: can be a str or a str list, or a tar object where to extract the tar_object
384 f
= path
if isinstance(path
, str) else "/".join(path
)
386 for member
in tar_object
.getmembers():
388 stream
= tar_object
.extractfile(member
)
390 stream
= BytesIO(member
.linkname
.encode("utf-8"))
401 metadata
= {"type": file_type
, "permissions": member
.mode
}
403 self
.fs
.upload_from_stream(f
+ "/" + member
.name
, stream
, metadata
=metadata
)
407 def file_open(self
, storage
, mode
):
410 :param storage: can be a str or list of str
411 :param mode: file mode
415 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
418 return GridByteStream(f
, self
.fs
, mode
)
420 return GridStringStream(f
, self
.fs
, mode
)
421 except errors
.NoFile
:
423 "File {} does not exist".format(f
), http_code
=HTTPStatus
.NOT_FOUND
427 "File {} cannot be opened".format(f
), http_code
=HTTPStatus
.BAD_REQUEST
430 def dir_ls(self
, storage
):
432 return folder content
433 :param storage: can be a str or list of str
434 :return: folder content
437 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
440 dir_cursor
= self
.fs
.find({"filename": f
})
441 for requested_dir
in dir_cursor
:
442 exception_dir
= next(dir_cursor
, None)
446 "Multiple directories found",
447 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
450 if requested_dir
.metadata
["type"] != "dir":
452 "File {} does not exist".format(f
),
453 http_code
=HTTPStatus
.NOT_FOUND
,
456 files_cursor
= self
.fs
.find(
457 {"filename": {"$regex": "^{}/([^/])*".format(f
)}}
459 for children_file
in files_cursor
:
460 files
+= [children_file
.filename
.replace(f
+ "/", "", 1)]
465 "File {} cannot be opened".format(f
), http_code
=HTTPStatus
.BAD_REQUEST
468 def file_delete(self
, storage
, ignore_non_exist
=False):
470 Delete storage content recursively
471 :param storage: can be a str or list of str
472 :param ignore_non_exist: not raise exception if storage does not exist
476 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
478 file_cursor
= self
.fs
.find({"filename": f
})
480 for requested_file
in file_cursor
:
482 exception_file
= next(file_cursor
, None)
486 "Multiple files found",
487 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
490 if requested_file
.metadata
["type"] == "dir":
491 dir_cursor
= self
.fs
.find({"filename": {"$regex": "^{}".format(f
)}})
493 for tmp
in dir_cursor
:
494 self
.fs
.delete(tmp
._id
)
496 self
.fs
.delete(requested_file
._id
)
497 if not found
and not ignore_non_exist
:
499 "File {} does not exist".format(storage
),
500 http_code
=HTTPStatus
.NOT_FOUND
,
504 "File {} cannot be deleted: {}".format(f
, e
),
505 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
508 def sync(self
, from_path
=None):
510 Sync from FSMongo to local storage
511 :param from_path: if supplied, only copy content from this path, not all
515 if os
.path
.isabs(from_path
):
516 from_path
= os
.path
.relpath(from_path
, self
.path
)
517 self
.__update
_local
_fs
(from_path
=from_path
)
519 def _update_mongo_fs(self
, from_path
):
521 os_path
= self
.path
+ from_path
523 # Obtain list of files and dirs in filesystem
525 for root
, dirs
, files
in os
.walk(os_path
):
527 member
= {"filename": os
.path
.join(root
, folder
), "type": "dir"}
528 members
.append(member
)
530 filename
= os
.path
.join(root
, file)
531 if os
.path
.islink(filename
):
535 member
= {"filename": os
.path
.join(root
, file), "type": file_type
}
536 members
.append(member
)
538 # Obtain files in mongo dict
539 remote_files
= self
._get
_mongo
_files
(from_path
)
541 # Upload members if they do not exists or have been modified
542 # We will do this for performance (avoid updating unmodified files) and to avoid
543 # updating a file with an older one in case there are two sources for synchronization
544 # in high availability scenarios
545 for member
in members
:
547 mask
= int(oct(os
.stat(member
["filename"]).st_mode
)[-3:], 8)
549 # convert to relative path
550 rel_filename
= os
.path
.relpath(member
["filename"], self
.path
)
551 last_modified_date
= datetime
.datetime
.fromtimestamp(
552 os
.path
.getmtime(member
["filename"])
555 remote_file
= remote_files
.get(rel_filename
)
557 remote_file
[0].uploadDate
if remote_file
else datetime
.datetime
.min
559 # remove processed files from dict
560 remote_files
.pop(rel_filename
, None)
562 if last_modified_date
>= upload_date
:
567 file_type
= member
["type"]
568 if file_type
== "dir":
570 elif file_type
== "sym":
572 os
.readlink(member
["filename"]).encode("utf-8")
575 fh
= open(member
["filename"], "rb")
576 stream
= BytesIO(fh
.read())
578 metadata
= {"type": file_type
, "permissions": mask
}
580 self
.fs
.upload_from_stream(rel_filename
, stream
, metadata
=metadata
)
584 for file in remote_file
:
585 self
.fs
.delete(file._id
)
592 # delete files that are not any more in local fs
593 for remote_file
in remote_files
.values():
594 for file in remote_file
:
595 self
.fs
.delete(file._id
)
597 def _get_mongo_files(self
, from_path
=None):
600 file_cursor
= self
.fs
.find(no_cursor_timeout
=True, sort
=[("uploadDate", -1)])
601 for file in file_cursor
:
602 if from_path
and not file.filename
.startswith(from_path
):
604 if file.filename
in file_dict
:
605 file_dict
[file.filename
].append(file)
607 file_dict
[file.filename
] = [file]
610 def reverse_sync(self
, from_path
: str):
612 Sync from local storage to FSMongo
613 :param from_path: base directory to upload content to mongo fs
616 if os
.path
.isabs(from_path
):
617 from_path
= os
.path
.relpath(from_path
, self
.path
)
618 self
._update
_mongo
_fs
(from_path
=from_path
)