1ce59097b7110f39f13afee4f4c0a9c5e44bec54
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
27 from gridfs
import errors
, GridFSBucket
28 from osm_common
.fsbase
import FsBase
, FsException
29 from pymongo
import MongoClient
32 __author__
= "Eduardo Sousa <eduardo.sousa@canonical.com>"
35 class GridByteStream(BytesIO
):
36 def __init__(self
, filename
, fs
, mode
):
37 BytesIO
.__init
__(self
)
39 self
.filename
= filename
42 self
.file_type
= "file" # Set "file" as default file_type
46 def __initialize__(self
):
49 cursor
= self
.fs
.find({"filename": self
.filename
})
51 for requested_file
in cursor
:
52 exception_file
= next(cursor
, None)
56 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
59 if requested_file
.metadata
["type"] in ("file", "sym"):
60 grid_file
= requested_file
61 self
.file_type
= requested_file
.metadata
["type"]
64 "Type isn't file", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
68 self
._id
= grid_file
._id
69 self
.fs
.download_to_stream(self
._id
, self
)
76 super(GridByteStream
, self
).close()
80 self
.fs
.delete(self
._id
)
82 cursor
= self
.fs
.find(
83 {"filename": self
.filename
.split("/")[0], "metadata": {"type": "dir"}}
86 parent_dir
= next(cursor
, None)
89 parent_dir_name
= self
.filename
.split("/")[0]
90 self
.filename
= self
.filename
.replace(
91 parent_dir_name
, parent_dir_name
[:-1], 1
96 self
.fs
.upload_from_stream_with_id(
97 self
._id
, self
.filename
, self
, metadata
={"type": self
.file_type
}
100 self
.fs
.upload_from_stream(
101 self
.filename
, self
, metadata
={"type": self
.file_type
}
103 super(GridByteStream
, self
).close()
108 def __exit__(self
, exc_type
, exc_val
, exc_tb
):
112 class GridStringStream(StringIO
):
113 def __init__(self
, filename
, fs
, mode
):
114 StringIO
.__init
__(self
)
116 self
.filename
= filename
119 self
.file_type
= "file" # Set "file" as default file_type
121 self
.__initialize
__()
123 def __initialize__(self
):
126 cursor
= self
.fs
.find({"filename": self
.filename
})
128 for requested_file
in cursor
:
129 exception_file
= next(cursor
, None)
133 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
136 if requested_file
.metadata
["type"] in ("file", "dir"):
137 grid_file
= requested_file
138 self
.file_type
= requested_file
.metadata
["type"]
141 "File type isn't file", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
146 self
._id
= grid_file
._id
147 self
.fs
.download_to_stream(self
._id
, stream
)
149 self
.write(stream
.read().decode("utf-8"))
157 super(GridStringStream
, self
).close()
161 self
.fs
.delete(self
._id
)
163 cursor
= self
.fs
.find(
164 {"filename": self
.filename
.split("/")[0], "metadata": {"type": "dir"}}
167 parent_dir
= next(cursor
, None)
170 parent_dir_name
= self
.filename
.split("/")[0]
171 self
.filename
= self
.filename
.replace(
172 parent_dir_name
, parent_dir_name
[:-1], 1
177 stream
.write(self
.read().encode("utf-8"))
180 self
.fs
.upload_from_stream_with_id(
181 self
._id
, self
.filename
, stream
, metadata
={"type": self
.file_type
}
184 self
.fs
.upload_from_stream(
185 self
.filename
, stream
, metadata
={"type": self
.file_type
}
188 super(GridStringStream
, self
).close()
193 def __exit__(self
, exc_type
, exc_val
, exc_tb
):
197 class FsMongo(FsBase
):
198 def __init__(self
, logger_name
="fs", lock
=False):
199 super().__init
__(logger_name
, lock
)
204 def __update_local_fs(self
, from_path
=None):
205 dir_cursor
= self
.fs
.find({"metadata.type": "dir"}, no_cursor_timeout
=True)
209 for directory
in dir_cursor
:
210 if from_path
and not directory
.filename
.startswith(from_path
):
212 self
.logger
.debug("Making dir {}".format(self
.path
+ directory
.filename
))
213 os
.makedirs(self
.path
+ directory
.filename
, exist_ok
=True)
214 valid_paths
.append(self
.path
+ directory
.filename
)
216 file_cursor
= self
.fs
.find(
217 {"metadata.type": {"$in": ["file", "sym"]}}, no_cursor_timeout
=True
220 for writing_file
in file_cursor
:
221 if from_path
and not writing_file
.filename
.startswith(from_path
):
223 file_path
= self
.path
+ writing_file
.filename
225 if writing_file
.metadata
["type"] == "sym":
227 self
.fs
.download_to_stream(writing_file
._id
, b
)
229 link
= b
.read().decode("utf-8")
232 self
.logger
.debug("Sync removing {}".format(file_path
))
235 if e
.errno
!= errno
.ENOENT
:
236 # This is probably permission denied or worse
238 os
.symlink(link
, file_path
)
240 folder
= os
.path
.dirname(file_path
)
241 if folder
not in valid_paths
:
242 self
.logger
.debug("Sync local directory {}".format(file_path
))
243 os
.makedirs(folder
, exist_ok
=True)
244 with
open(file_path
, "wb+") as file_stream
:
245 self
.logger
.debug("Sync download {}".format(file_path
))
246 self
.fs
.download_to_stream(writing_file
._id
, file_stream
)
247 if "permissions" in writing_file
.metadata
:
248 os
.chmod(file_path
, writing_file
.metadata
["permissions"])
250 def get_params(self
):
251 return {"fs": "mongo", "path": self
.path
}
253 def fs_connect(self
, config
):
255 if "logger_name" in config
:
256 self
.logger
= logging
.getLogger(config
["logger_name"])
258 self
.path
= config
["path"]
260 raise FsException('Missing parameter "path"')
261 if not self
.path
.endswith("/"):
263 if not os
.path
.exists(self
.path
):
265 "Invalid configuration param at '[storage]': path '{}' does not exist".format(
269 elif not os
.access(self
.path
, os
.W_OK
):
271 "Invalid configuration param at '[storage]': path '{}' is not writable".format(
275 if all(key
in config
.keys() for key
in ["uri", "collection"]):
276 self
.client
= MongoClient(config
["uri"])
277 self
.fs
= GridFSBucket(self
.client
[config
["collection"]])
278 elif all(key
in config
.keys() for key
in ["host", "port", "collection"]):
279 self
.client
= MongoClient(config
["host"], config
["port"])
280 self
.fs
= GridFSBucket(self
.client
[config
["collection"]])
282 if "collection" not in config
.keys():
283 raise FsException('Missing parameter "collection"')
285 raise FsException('Missing parameters: "uri" or "host" + "port"')
288 except Exception as e
: # TODO refine
289 raise FsException(str(e
))
291 def fs_disconnect(self
):
294 def mkdir(self
, folder
):
296 Creates a folder or parent object location
298 :return: None or raises an exception
300 folder
= folder
.rstrip("/")
302 self
.fs
.upload_from_stream(folder
, BytesIO(), metadata
={"type": "dir"})
303 except errors
.FileExists
: # make it idempotent
305 except Exception as e
:
306 raise FsException(str(e
), http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
)
308 def dir_rename(self
, src
, dst
):
310 Rename one directory name. If dst exist, it replaces (deletes) existing directory
311 :param src: source directory
312 :param dst: destination directory
313 :return: None or raises and exception
315 dst
= dst
.rstrip("/")
316 src
= src
.rstrip("/")
319 dst_cursor
= self
.fs
.find(
320 {"filename": {"$regex": "^{}(/|$)".format(dst
)}}, no_cursor_timeout
=True
323 for dst_file
in dst_cursor
:
324 self
.fs
.delete(dst_file
._id
)
326 src_cursor
= self
.fs
.find(
327 {"filename": {"$regex": "^{}(/|$)".format(src
)}}, no_cursor_timeout
=True
330 for src_file
in src_cursor
:
331 self
.fs
.rename(src_file
._id
, src_file
.filename
.replace(src
, dst
, 1))
332 except Exception as e
:
333 raise FsException(str(e
), http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
)
335 def file_exists(self
, storage
, mode
=None):
337 Indicates if "storage" file exist
338 :param storage: can be a str or a str list
339 :param mode: can be 'file' exist as a regular file; 'dir' exists as a directory or; 'None' just exists
342 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
345 cursor
= self
.fs
.find({"filename": f
})
347 for requested_file
in cursor
:
348 exception_file
= next(cursor
, None)
352 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
355 self
.logger
.debug("Entry {} metadata {}".format(f
, requested_file
.metadata
))
357 # if no special mode is required just check it does exists
361 if requested_file
.metadata
["type"] == mode
:
364 if requested_file
.metadata
["type"] == "sym" and mode
== "file":
369 def file_size(self
, storage
):
372 :param storage: can be a str or a str list
375 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
378 cursor
= self
.fs
.find({"filename": f
})
380 for requested_file
in cursor
:
381 exception_file
= next(cursor
, None)
385 "Multiple files found", http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
388 return requested_file
.length
390 def file_extract(self
, compressed_object
, path
):
393 :param compressed_object: object of type tar or zip
394 :param path: can be a str or a str list, or a tar object where to extract the tar_object
397 f
= path
if isinstance(path
, str) else "/".join(path
)
400 if type(compressed_object
) is tarfile
.TarFile
:
401 for member
in compressed_object
.getmembers():
403 stream
= compressed_object
.extractfile(member
)
405 stream
= BytesIO(member
.linkname
.encode("utf-8"))
416 metadata
= {"type": file_type
, "permissions": member
.mode
}
417 member
.name
= member
.name
.rstrip("/")
419 self
.logger
.debug("Uploading {}/{}".format(f
, member
.name
))
420 self
.fs
.upload_from_stream(
421 f
+ "/" + member
.name
, stream
, metadata
=metadata
425 elif type(compressed_object
) is zipfile
.ZipFile
:
426 for member
in compressed_object
.infolist():
430 stream
= compressed_object
.read(member
)
437 metadata
= {"type": file_type
}
438 member
.filename
= member
.filename
.rstrip("/")
440 self
.logger
.debug("Uploading {}/{}".format(f
, member
.filename
))
441 self
.fs
.upload_from_stream(
442 f
+ "/" + member
.filename
, stream
, metadata
=metadata
448 def file_open(self
, storage
, mode
):
451 :param storage: can be a str or list of str
452 :param mode: file mode
456 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
460 return GridByteStream(f
, self
.fs
, mode
)
462 return GridStringStream(f
, self
.fs
, mode
)
463 except errors
.NoFile
:
465 "File {} does not exist".format(f
), http_code
=HTTPStatus
.NOT_FOUND
469 "File {} cannot be opened".format(f
), http_code
=HTTPStatus
.BAD_REQUEST
472 def dir_ls(self
, storage
):
474 return folder content
475 :param storage: can be a str or list of str
476 :return: folder content
479 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
483 dir_cursor
= self
.fs
.find({"filename": f
})
484 for requested_dir
in dir_cursor
:
485 exception_dir
= next(dir_cursor
, None)
489 "Multiple directories found",
490 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
493 if requested_dir
.metadata
["type"] != "dir":
495 "File {} does not exist".format(f
),
496 http_code
=HTTPStatus
.NOT_FOUND
,
502 files_cursor
= self
.fs
.find(
503 {"filename": {"$regex": "^{}/([^/])*".format(f
)}}
505 for children_file
in files_cursor
:
506 files
+= [children_file
.filename
.replace(f
+ "/", "", 1)]
511 "File {} cannot be opened".format(f
), http_code
=HTTPStatus
.BAD_REQUEST
514 def file_delete(self
, storage
, ignore_non_exist
=False):
516 Delete storage content recursively
517 :param storage: can be a str or list of str
518 :param ignore_non_exist: not raise exception if storage does not exist
522 f
= storage
if isinstance(storage
, str) else "/".join(storage
)
525 file_cursor
= self
.fs
.find({"filename": f
})
527 for requested_file
in file_cursor
:
529 exception_file
= next(file_cursor
, None)
533 "Cannot delete duplicate file: {} and {}".format(
534 requested_file
.filename
, exception_file
.filename
538 "Multiple files found",
539 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
542 if requested_file
.metadata
["type"] == "dir":
543 dir_cursor
= self
.fs
.find(
544 {"filename": {"$regex": "^{}/".format(f
)}}
547 for tmp
in dir_cursor
:
548 self
.logger
.debug("Deleting {}".format(tmp
.filename
))
549 self
.fs
.delete(tmp
._id
)
551 self
.logger
.debug("Deleting {}".format(requested_file
.filename
))
552 self
.fs
.delete(requested_file
._id
)
553 if not found
and not ignore_non_exist
:
555 "File {} does not exist".format(storage
),
556 http_code
=HTTPStatus
.NOT_FOUND
,
560 "File {} cannot be deleted: {}".format(f
, e
),
561 http_code
=HTTPStatus
.INTERNAL_SERVER_ERROR
,
564 def sync(self
, from_path
=None):
566 Sync from FSMongo to local storage
567 :param from_path: if supplied, only copy content from this path, not all
571 if os
.path
.isabs(from_path
):
572 from_path
= os
.path
.relpath(from_path
, self
.path
)
573 self
.__update
_local
_fs
(from_path
=from_path
)
575 def _update_mongo_fs(self
, from_path
):
577 os_path
= self
.path
+ from_path
579 # Obtain list of files and dirs in filesystem
581 for root
, dirs
, files
in os
.walk(os_path
):
583 member
= {"filename": os
.path
.join(root
, folder
), "type": "dir"}
584 if os
.path
.islink(member
["filename"]):
585 member
["type"] = "sym"
586 members
.append(member
)
588 filename
= os
.path
.join(root
, file)
589 if os
.path
.islink(filename
):
593 member
= {"filename": os
.path
.join(root
, file), "type": file_type
}
594 members
.append(member
)
596 # Obtain files in mongo dict
597 remote_files
= self
._get
_mongo
_files
(from_path
)
599 # Upload members if they do not exists or have been modified
600 # We will do this for performance (avoid updating unmodified files) and to avoid
601 # updating a file with an older one in case there are two sources for synchronization
602 # in high availability scenarios
603 for member
in members
:
605 mask
= int(oct(os
.stat(member
["filename"]).st_mode
)[-3:], 8)
607 # convert to relative path
608 rel_filename
= os
.path
.relpath(member
["filename"], self
.path
)
609 last_modified_date
= datetime
.datetime
.fromtimestamp(
610 os
.path
.getmtime(member
["filename"])
613 remote_file
= remote_files
.get(rel_filename
)
615 remote_file
[0].uploadDate
if remote_file
else datetime
.datetime
.min
617 # remove processed files from dict
618 remote_files
.pop(rel_filename
, None)
620 if last_modified_date
>= upload_date
:
625 file_type
= member
["type"]
626 if file_type
== "dir":
628 elif file_type
== "sym":
630 os
.readlink(member
["filename"]).encode("utf-8")
633 fh
= open(member
["filename"], "rb")
634 stream
= BytesIO(fh
.read())
636 metadata
= {"type": file_type
, "permissions": mask
}
638 self
.logger
.debug("Sync upload {}".format(rel_filename
))
639 self
.fs
.upload_from_stream(rel_filename
, stream
, metadata
=metadata
)
643 for file in remote_file
:
644 self
.logger
.debug("Sync deleting {}".format(file.filename
))
645 self
.fs
.delete(file._id
)
652 # delete files that are not any more in local fs
653 for remote_file
in remote_files
.values():
654 for file in remote_file
:
655 self
.fs
.delete(file._id
)
657 def _get_mongo_files(self
, from_path
=None):
660 file_cursor
= self
.fs
.find(no_cursor_timeout
=True, sort
=[("uploadDate", -1)])
661 for file in file_cursor
:
662 if from_path
and not file.filename
.startswith(from_path
):
664 if file.filename
in file_dict
:
665 file_dict
[file.filename
].append(file)
667 file_dict
[file.filename
] = [file]
670 def reverse_sync(self
, from_path
: str):
672 Sync from local storage to FSMongo
673 :param from_path: base directory to upload content to mongo fs
676 if os
.path
.isabs(from_path
):
677 from_path
= os
.path
.relpath(from_path
, self
.path
)
678 self
._update
_mongo
_fs
(from_path
=from_path
)