| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 1 | # Copyright 2019 Canonical |
| 2 | # |
| 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 |
| 6 | # |
| 7 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | # |
| 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 |
| 13 | # under the License. |
| 14 | # |
| 15 | # For those usages not covered by the Apache License, Version 2.0 please |
| 16 | # contact: eduardo.sousa@canonical.com |
| 17 | ## |
| 18 | |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 19 | import errno |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 20 | from http import HTTPStatus |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 21 | from io import BytesIO, StringIO |
| 22 | import logging |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 23 | import os |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 24 | import datetime |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 25 | import tarfile |
| 26 | import zipfile |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 27 | |
| 28 | from gridfs import GridFSBucket, errors |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 29 | from osm_common.fsbase import FsBase, FsException |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 30 | from pymongo import MongoClient |
| 31 | |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 32 | |
| 33 | __author__ = "Eduardo Sousa <eduardo.sousa@canonical.com>" |
| 34 | |
| 35 | |
| 36 | class GridByteStream(BytesIO): |
| 37 | def __init__(self, filename, fs, mode): |
| 38 | BytesIO.__init__(self) |
| 39 | self._id = None |
| 40 | self.filename = filename |
| 41 | self.fs = fs |
| 42 | self.mode = mode |
| David Garcia | 7982b78 | 2020-05-20 12:09:37 +0200 | [diff] [blame] | 43 | self.file_type = "file" # Set "file" as default file_type |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 44 | |
| 45 | self.__initialize__() |
| 46 | |
| 47 | def __initialize__(self): |
| 48 | grid_file = None |
| 49 | |
| 50 | cursor = self.fs.find({"filename": self.filename}) |
| 51 | |
| 52 | for requested_file in cursor: |
| 53 | exception_file = next(cursor, None) |
| 54 | |
| 55 | if exception_file: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 56 | raise FsException( |
| 57 | "Multiple files found", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 58 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 59 | |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 60 | if requested_file.metadata["type"] in ("file", "sym"): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 61 | grid_file = requested_file |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 62 | self.file_type = requested_file.metadata["type"] |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 63 | else: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 64 | raise FsException( |
| 65 | "Type isn't file", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 66 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 67 | |
| 68 | if grid_file: |
| 69 | self._id = grid_file._id |
| 70 | self.fs.download_to_stream(self._id, self) |
| 71 | |
| 72 | if "r" in self.mode: |
| 73 | self.seek(0, 0) |
| 74 | |
| 75 | def close(self): |
| 76 | if "r" in self.mode: |
| 77 | super(GridByteStream, self).close() |
| 78 | return |
| 79 | |
| 80 | if self._id: |
| 81 | self.fs.delete(self._id) |
| 82 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 83 | cursor = self.fs.find( |
| 84 | {"filename": self.filename.split("/")[0], "metadata": {"type": "dir"}} |
| 85 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 86 | |
| 87 | parent_dir = next(cursor, None) |
| 88 | |
| 89 | if not parent_dir: |
| 90 | parent_dir_name = self.filename.split("/")[0] |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 91 | self.filename = self.filename.replace( |
| 92 | parent_dir_name, parent_dir_name[:-1], 1 |
| 93 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 94 | |
| 95 | self.seek(0, 0) |
| 96 | if self._id: |
| 97 | self.fs.upload_from_stream_with_id( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 98 | self._id, self.filename, self, metadata={"type": self.file_type} |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 99 | ) |
| 100 | else: |
| 101 | self.fs.upload_from_stream( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 102 | self.filename, self, metadata={"type": self.file_type} |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 103 | ) |
| 104 | super(GridByteStream, self).close() |
| 105 | |
| 106 | def __enter__(self): |
| 107 | return self |
| 108 | |
| 109 | def __exit__(self, exc_type, exc_val, exc_tb): |
| 110 | self.close() |
| 111 | |
| 112 | |
| 113 | class GridStringStream(StringIO): |
| 114 | def __init__(self, filename, fs, mode): |
| 115 | StringIO.__init__(self) |
| 116 | self._id = None |
| 117 | self.filename = filename |
| 118 | self.fs = fs |
| 119 | self.mode = mode |
| David Garcia | 7982b78 | 2020-05-20 12:09:37 +0200 | [diff] [blame] | 120 | self.file_type = "file" # Set "file" as default file_type |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 121 | |
| 122 | self.__initialize__() |
| 123 | |
| 124 | def __initialize__(self): |
| 125 | grid_file = None |
| 126 | |
| 127 | cursor = self.fs.find({"filename": self.filename}) |
| 128 | |
| 129 | for requested_file in cursor: |
| 130 | exception_file = next(cursor, None) |
| 131 | |
| 132 | if exception_file: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 133 | raise FsException( |
| 134 | "Multiple files found", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 135 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 136 | |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 137 | if requested_file.metadata["type"] in ("file", "dir"): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 138 | grid_file = requested_file |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 139 | self.file_type = requested_file.metadata["type"] |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 140 | else: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 141 | raise FsException( |
| 142 | "File type isn't file", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 143 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 144 | |
| 145 | if grid_file: |
| 146 | stream = BytesIO() |
| 147 | self._id = grid_file._id |
| 148 | self.fs.download_to_stream(self._id, stream) |
| 149 | stream.seek(0) |
| 150 | self.write(stream.read().decode("utf-8")) |
| 151 | stream.close() |
| 152 | |
| 153 | if "r" in self.mode: |
| 154 | self.seek(0, 0) |
| 155 | |
| 156 | def close(self): |
| 157 | if "r" in self.mode: |
| 158 | super(GridStringStream, self).close() |
| 159 | return |
| 160 | |
| 161 | if self._id: |
| 162 | self.fs.delete(self._id) |
| 163 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 164 | cursor = self.fs.find( |
| 165 | {"filename": self.filename.split("/")[0], "metadata": {"type": "dir"}} |
| 166 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 167 | |
| 168 | parent_dir = next(cursor, None) |
| 169 | |
| 170 | if not parent_dir: |
| 171 | parent_dir_name = self.filename.split("/")[0] |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 172 | self.filename = self.filename.replace( |
| 173 | parent_dir_name, parent_dir_name[:-1], 1 |
| 174 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 175 | |
| 176 | self.seek(0, 0) |
| 177 | stream = BytesIO() |
| 178 | stream.write(self.read().encode("utf-8")) |
| 179 | stream.seek(0, 0) |
| 180 | if self._id: |
| 181 | self.fs.upload_from_stream_with_id( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 182 | self._id, self.filename, stream, metadata={"type": self.file_type} |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 183 | ) |
| 184 | else: |
| 185 | self.fs.upload_from_stream( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 186 | self.filename, stream, metadata={"type": self.file_type} |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 187 | ) |
| 188 | stream.close() |
| 189 | super(GridStringStream, self).close() |
| 190 | |
| 191 | def __enter__(self): |
| 192 | return self |
| 193 | |
| 194 | def __exit__(self, exc_type, exc_val, exc_tb): |
| 195 | self.close() |
| 196 | |
| 197 | |
| 198 | class FsMongo(FsBase): |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 199 | def __init__(self, logger_name="fs", lock=False): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 200 | super().__init__(logger_name, lock) |
| 201 | self.path = None |
| 202 | self.client = None |
| 203 | self.fs = None |
| 204 | |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 205 | def __update_local_fs(self, from_path=None): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 206 | dir_cursor = self.fs.find({"metadata.type": "dir"}, no_cursor_timeout=True) |
| 207 | |
| bravof | f73a900 | 2021-11-23 10:34:43 -0300 | [diff] [blame] | 208 | valid_paths = [] |
| 209 | |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 210 | for directory in dir_cursor: |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 211 | if from_path and not directory.filename.startswith(from_path): |
| 212 | continue |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 213 | os.makedirs(self.path + directory.filename, exist_ok=True) |
| bravof | f73a900 | 2021-11-23 10:34:43 -0300 | [diff] [blame] | 214 | valid_paths.append(self.path + directory.filename) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 215 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 216 | file_cursor = self.fs.find( |
| 217 | {"metadata.type": {"$in": ["file", "sym"]}}, no_cursor_timeout=True |
| 218 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 219 | |
| 220 | for writing_file in file_cursor: |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 221 | if from_path and not writing_file.filename.startswith(from_path): |
| 222 | continue |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 223 | file_path = self.path + writing_file.filename |
| David Garcia | 8ab6cc6 | 2020-06-26 17:04:37 +0200 | [diff] [blame] | 224 | |
| 225 | if writing_file.metadata["type"] == "sym": |
| 226 | with BytesIO() as b: |
| 227 | self.fs.download_to_stream(writing_file._id, b) |
| 228 | b.seek(0) |
| 229 | link = b.read().decode("utf-8") |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 230 | |
| 231 | try: |
| 232 | os.remove(file_path) |
| 233 | except OSError as e: |
| 234 | if e.errno != errno.ENOENT: |
| 235 | # This is probably permission denied or worse |
| 236 | raise |
| David Garcia | 8ab6cc6 | 2020-06-26 17:04:37 +0200 | [diff] [blame] | 237 | os.symlink(link, file_path) |
| 238 | else: |
| bravof | f73a900 | 2021-11-23 10:34:43 -0300 | [diff] [blame] | 239 | folder = os.path.dirname(file_path) |
| 240 | if folder not in valid_paths: |
| 241 | os.makedirs(folder, exist_ok=True) |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 242 | with open(file_path, "wb+") as file_stream: |
| David Garcia | 8ab6cc6 | 2020-06-26 17:04:37 +0200 | [diff] [blame] | 243 | self.fs.download_to_stream(writing_file._id, file_stream) |
| 244 | if "permissions" in writing_file.metadata: |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 245 | os.chmod(file_path, writing_file.metadata["permissions"]) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 246 | |
| 247 | def get_params(self): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 248 | return {"fs": "mongo", "path": self.path} |
| 249 | |
| 250 | def fs_connect(self, config): |
| 251 | try: |
| 252 | if "logger_name" in config: |
| 253 | self.logger = logging.getLogger(config["logger_name"]) |
| 254 | if "path" in config: |
| 255 | self.path = config["path"] |
| 256 | else: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 257 | raise FsException('Missing parameter "path"') |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 258 | if not self.path.endswith("/"): |
| 259 | self.path += "/" |
| 260 | if not os.path.exists(self.path): |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 261 | raise FsException( |
| 262 | "Invalid configuration param at '[storage]': path '{}' does not exist".format( |
| 263 | config["path"] |
| 264 | ) |
| 265 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 266 | elif not os.access(self.path, os.W_OK): |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 267 | raise FsException( |
| 268 | "Invalid configuration param at '[storage]': path '{}' is not writable".format( |
| 269 | config["path"] |
| 270 | ) |
| 271 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 272 | if all(key in config.keys() for key in ["uri", "collection"]): |
| 273 | self.client = MongoClient(config["uri"]) |
| 274 | self.fs = GridFSBucket(self.client[config["collection"]]) |
| 275 | elif all(key in config.keys() for key in ["host", "port", "collection"]): |
| 276 | self.client = MongoClient(config["host"], config["port"]) |
| 277 | self.fs = GridFSBucket(self.client[config["collection"]]) |
| 278 | else: |
| 279 | if "collection" not in config.keys(): |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 280 | raise FsException('Missing parameter "collection"') |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 281 | else: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 282 | raise FsException('Missing parameters: "uri" or "host" + "port"') |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 283 | except FsException: |
| 284 | raise |
| 285 | except Exception as e: # TODO refine |
| 286 | raise FsException(str(e)) |
| 287 | |
| 288 | def fs_disconnect(self): |
| 289 | pass # TODO |
| 290 | |
| 291 | def mkdir(self, folder): |
| 292 | """ |
| 293 | Creates a folder or parent object location |
| 294 | :param folder: |
| 295 | :return: None or raises an exception |
| 296 | """ |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 297 | folder = folder.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 298 | try: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 299 | self.fs.upload_from_stream(folder, BytesIO(), metadata={"type": "dir"}) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 300 | except errors.FileExists: # make it idempotent |
| 301 | pass |
| 302 | except Exception as e: |
| 303 | raise FsException(str(e), http_code=HTTPStatus.INTERNAL_SERVER_ERROR) |
| 304 | |
| 305 | def dir_rename(self, src, dst): |
| 306 | """ |
| 307 | Rename one directory name. If dst exist, it replaces (deletes) existing directory |
| 308 | :param src: source directory |
| 309 | :param dst: destination directory |
| 310 | :return: None or raises and exception |
| 311 | """ |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 312 | dst = dst.rstrip("/") |
| 313 | src = src.rstrip("/") |
| 314 | |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 315 | try: |
| 316 | dst_cursor = self.fs.find( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 317 | {"filename": {"$regex": "^{}(/|$)".format(dst)}}, no_cursor_timeout=True |
| 318 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 319 | |
| 320 | for dst_file in dst_cursor: |
| 321 | self.fs.delete(dst_file._id) |
| 322 | |
| 323 | src_cursor = self.fs.find( |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 324 | {"filename": {"$regex": "^{}(/|$)".format(src)}}, no_cursor_timeout=True |
| 325 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 326 | |
| 327 | for src_file in src_cursor: |
| 328 | self.fs.rename(src_file._id, src_file.filename.replace(src, dst, 1)) |
| 329 | except Exception as e: |
| 330 | raise FsException(str(e), http_code=HTTPStatus.INTERNAL_SERVER_ERROR) |
| 331 | |
| 332 | def file_exists(self, storage, mode=None): |
| 333 | """ |
| 334 | Indicates if "storage" file exist |
| 335 | :param storage: can be a str or a str list |
| 336 | :param mode: can be 'file' exist as a regular file; 'dir' exists as a directory or; 'None' just exists |
| 337 | :return: True, False |
| 338 | """ |
| 339 | f = storage if isinstance(storage, str) else "/".join(storage) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 340 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 341 | |
| 342 | cursor = self.fs.find({"filename": f}) |
| 343 | |
| 344 | for requested_file in cursor: |
| 345 | exception_file = next(cursor, None) |
| 346 | |
| 347 | if exception_file: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 348 | raise FsException( |
| 349 | "Multiple files found", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 350 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 351 | |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 352 | self.logger.debug("Entry {} metadata {}".format(f, requested_file.metadata)) |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 353 | |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 354 | # if no special mode is required just check it does exists |
| 355 | if not mode: |
| 356 | return True |
| 357 | |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 358 | if requested_file.metadata["type"] == mode: |
| 359 | return True |
| beierlm | ff2e826 | 2020-07-08 16:32:50 -0400 | [diff] [blame] | 360 | |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 361 | if requested_file.metadata["type"] == "sym" and mode == "file": |
| 362 | return True |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 363 | |
| 364 | return False |
| 365 | |
| 366 | def file_size(self, storage): |
| 367 | """ |
| 368 | return file size |
| 369 | :param storage: can be a str or a str list |
| 370 | :return: file size |
| 371 | """ |
| 372 | f = storage if isinstance(storage, str) else "/".join(storage) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 373 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 374 | |
| 375 | cursor = self.fs.find({"filename": f}) |
| 376 | |
| 377 | for requested_file in cursor: |
| 378 | exception_file = next(cursor, None) |
| 379 | |
| 380 | if exception_file: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 381 | raise FsException( |
| 382 | "Multiple files found", http_code=HTTPStatus.INTERNAL_SERVER_ERROR |
| 383 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 384 | |
| 385 | return requested_file.length |
| 386 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 387 | def file_extract(self, compressed_object, path): |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 388 | """ |
| 389 | extract a tar file |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 390 | :param compressed_object: object of type tar or zip |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 391 | :param path: can be a str or a str list, or a tar object where to extract the tar_object |
| 392 | :return: None |
| 393 | """ |
| 394 | f = path if isinstance(path, str) else "/".join(path) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 395 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 396 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 397 | if type(compressed_object) is tarfile.TarFile: |
| 398 | for member in compressed_object.getmembers(): |
| 399 | if member.isfile(): |
| 400 | stream = compressed_object.extractfile(member) |
| 401 | elif member.issym(): |
| 402 | stream = BytesIO(member.linkname.encode("utf-8")) |
| 403 | else: |
| 404 | stream = BytesIO() |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 405 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 406 | if member.isfile(): |
| 407 | file_type = "file" |
| 408 | elif member.issym(): |
| 409 | file_type = "sym" |
| 410 | else: |
| 411 | file_type = "dir" |
| sousaedu | b95cca6 | 2020-03-12 11:12:25 +0000 | [diff] [blame] | 412 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 413 | metadata = {"type": file_type, "permissions": member.mode} |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 414 | member.name = member.name.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 415 | |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 416 | self.logger.debug("Uploading {}".format(member.name)) |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 417 | self.fs.upload_from_stream( |
| 418 | f + "/" + member.name, stream, metadata=metadata |
| 419 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 420 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 421 | stream.close() |
| 422 | elif type(compressed_object) is zipfile.ZipFile: |
| 423 | for member in compressed_object.infolist(): |
| 424 | if member.is_dir(): |
| 425 | stream = BytesIO() |
| 426 | else: |
| 427 | stream = compressed_object.read(member) |
| 428 | |
| 429 | if member.is_dir(): |
| 430 | file_type = "dir" |
| 431 | else: |
| 432 | file_type = "file" |
| 433 | |
| 434 | metadata = {"type": file_type} |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 435 | member.filename = member.filename.rstrip("/") |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 436 | |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 437 | self.logger.debug("Uploading {}".format(member.filename)) |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 438 | self.fs.upload_from_stream( |
| 439 | f + "/" + member.filename, stream, metadata=metadata |
| 440 | ) |
| 441 | |
| 442 | if member.is_dir(): |
| 443 | stream.close() |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 444 | |
| 445 | def file_open(self, storage, mode): |
| 446 | """ |
| 447 | Open a file |
| 448 | :param storage: can be a str or list of str |
| 449 | :param mode: file mode |
| 450 | :return: file object |
| 451 | """ |
| 452 | try: |
| 453 | f = storage if isinstance(storage, str) else "/".join(storage) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 454 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 455 | |
| 456 | if "b" in mode: |
| 457 | return GridByteStream(f, self.fs, mode) |
| 458 | else: |
| 459 | return GridStringStream(f, self.fs, mode) |
| 460 | except errors.NoFile: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 461 | raise FsException( |
| 462 | "File {} does not exist".format(f), http_code=HTTPStatus.NOT_FOUND |
| 463 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 464 | except IOError: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 465 | raise FsException( |
| 466 | "File {} cannot be opened".format(f), http_code=HTTPStatus.BAD_REQUEST |
| 467 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 468 | |
| 469 | def dir_ls(self, storage): |
| 470 | """ |
| 471 | return folder content |
| 472 | :param storage: can be a str or list of str |
| 473 | :return: folder content |
| 474 | """ |
| 475 | try: |
| 476 | f = storage if isinstance(storage, str) else "/".join(storage) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 477 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 478 | |
| 479 | files = [] |
| 480 | dir_cursor = self.fs.find({"filename": f}) |
| 481 | for requested_dir in dir_cursor: |
| 482 | exception_dir = next(dir_cursor, None) |
| 483 | |
| 484 | if exception_dir: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 485 | raise FsException( |
| 486 | "Multiple directories found", |
| 487 | http_code=HTTPStatus.INTERNAL_SERVER_ERROR, |
| 488 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 489 | |
| 490 | if requested_dir.metadata["type"] != "dir": |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 491 | raise FsException( |
| 492 | "File {} does not exist".format(f), |
| 493 | http_code=HTTPStatus.NOT_FOUND, |
| 494 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 495 | |
| bravof | 98fc8f0 | 2021-11-04 21:16:00 -0300 | [diff] [blame] | 496 | if f.endswith("/"): |
| 497 | f = f[:-1] |
| 498 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 499 | files_cursor = self.fs.find( |
| 500 | {"filename": {"$regex": "^{}/([^/])*".format(f)}} |
| 501 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 502 | for children_file in files_cursor: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 503 | files += [children_file.filename.replace(f + "/", "", 1)] |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 504 | |
| 505 | return files |
| 506 | except IOError: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 507 | raise FsException( |
| 508 | "File {} cannot be opened".format(f), http_code=HTTPStatus.BAD_REQUEST |
| 509 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 510 | |
| 511 | def file_delete(self, storage, ignore_non_exist=False): |
| 512 | """ |
| 513 | Delete storage content recursively |
| 514 | :param storage: can be a str or list of str |
| 515 | :param ignore_non_exist: not raise exception if storage does not exist |
| 516 | :return: None |
| 517 | """ |
| 518 | try: |
| 519 | f = storage if isinstance(storage, str) else "/".join(storage) |
| beierlm | 40f57c8 | 2022-04-14 14:45:07 -0400 | [diff] [blame^] | 520 | f = f.rstrip("/") |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 521 | |
| 522 | file_cursor = self.fs.find({"filename": f}) |
| 523 | found = False |
| 524 | for requested_file in file_cursor: |
| 525 | found = True |
| 526 | exception_file = next(file_cursor, None) |
| 527 | |
| 528 | if exception_file: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 529 | raise FsException( |
| 530 | "Multiple files found", |
| 531 | http_code=HTTPStatus.INTERNAL_SERVER_ERROR, |
| 532 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 533 | |
| 534 | if requested_file.metadata["type"] == "dir": |
| 535 | dir_cursor = self.fs.find({"filename": {"$regex": "^{}".format(f)}}) |
| 536 | |
| 537 | for tmp in dir_cursor: |
| 538 | self.fs.delete(tmp._id) |
| 539 | else: |
| 540 | self.fs.delete(requested_file._id) |
| 541 | if not found and not ignore_non_exist: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 542 | raise FsException( |
| 543 | "File {} does not exist".format(storage), |
| 544 | http_code=HTTPStatus.NOT_FOUND, |
| 545 | ) |
| Eduardo Sousa | 0593aba | 2019-06-04 12:55:43 +0100 | [diff] [blame] | 546 | except IOError as e: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 547 | raise FsException( |
| 548 | "File {} cannot be deleted: {}".format(f, e), |
| 549 | http_code=HTTPStatus.INTERNAL_SERVER_ERROR, |
| 550 | ) |
| David Garcia | 788b9d6 | 2020-01-20 13:21:06 +0100 | [diff] [blame] | 551 | |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 552 | def sync(self, from_path=None): |
| David Garcia | 788b9d6 | 2020-01-20 13:21:06 +0100 | [diff] [blame] | 553 | """ |
| 554 | Sync from FSMongo to local storage |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 555 | :param from_path: if supplied, only copy content from this path, not all |
| 556 | :return: None |
| David Garcia | 788b9d6 | 2020-01-20 13:21:06 +0100 | [diff] [blame] | 557 | """ |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 558 | if from_path: |
| 559 | if os.path.isabs(from_path): |
| 560 | from_path = os.path.relpath(from_path, self.path) |
| tierno | b07e4ef | 2020-05-06 14:22:48 +0000 | [diff] [blame] | 561 | self.__update_local_fs(from_path=from_path) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 562 | |
| 563 | def _update_mongo_fs(self, from_path): |
| 564 | |
| 565 | os_path = self.path + from_path |
| 566 | |
| 567 | # Obtain list of files and dirs in filesystem |
| 568 | members = [] |
| 569 | for root, dirs, files in os.walk(os_path): |
| 570 | for folder in dirs: |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 571 | member = {"filename": os.path.join(root, folder), "type": "dir"} |
| beierlm | 945fa22 | 2022-04-04 11:44:03 -0400 | [diff] [blame] | 572 | if os.path.islink(member["filename"]): |
| 573 | member["type"] = "sym" |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 574 | members.append(member) |
| 575 | for file in files: |
| 576 | filename = os.path.join(root, file) |
| 577 | if os.path.islink(filename): |
| 578 | file_type = "sym" |
| 579 | else: |
| 580 | file_type = "file" |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 581 | member = {"filename": os.path.join(root, file), "type": file_type} |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 582 | members.append(member) |
| 583 | |
| 584 | # Obtain files in mongo dict |
| 585 | remote_files = self._get_mongo_files(from_path) |
| 586 | |
| 587 | # Upload members if they do not exists or have been modified |
| 588 | # We will do this for performance (avoid updating unmodified files) and to avoid |
| 589 | # updating a file with an older one in case there are two sources for synchronization |
| 590 | # in high availability scenarios |
| 591 | for member in members: |
| 592 | # obtain permission |
| 593 | mask = int(oct(os.stat(member["filename"]).st_mode)[-3:], 8) |
| 594 | |
| 595 | # convert to relative path |
| 596 | rel_filename = os.path.relpath(member["filename"], self.path) |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 597 | last_modified_date = datetime.datetime.fromtimestamp( |
| 598 | os.path.getmtime(member["filename"]) |
| 599 | ) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 600 | |
| 601 | remote_file = remote_files.get(rel_filename) |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 602 | upload_date = ( |
| 603 | remote_file[0].uploadDate if remote_file else datetime.datetime.min |
| 604 | ) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 605 | # remove processed files from dict |
| 606 | remote_files.pop(rel_filename, None) |
| 607 | |
| 608 | if last_modified_date >= upload_date: |
| 609 | |
| 610 | stream = None |
| 611 | fh = None |
| 612 | try: |
| 613 | file_type = member["type"] |
| 614 | if file_type == "dir": |
| 615 | stream = BytesIO() |
| 616 | elif file_type == "sym": |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 617 | stream = BytesIO( |
| 618 | os.readlink(member["filename"]).encode("utf-8") |
| 619 | ) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 620 | else: |
| 621 | fh = open(member["filename"], "rb") |
| 622 | stream = BytesIO(fh.read()) |
| 623 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 624 | metadata = {"type": file_type, "permissions": mask} |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 625 | |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 626 | self.fs.upload_from_stream(rel_filename, stream, metadata=metadata) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 627 | |
| 628 | # delete old files |
| 629 | if remote_file: |
| 630 | for file in remote_file: |
| 631 | self.fs.delete(file._id) |
| 632 | finally: |
| 633 | if fh: |
| 634 | fh.close() |
| 635 | if stream: |
| 636 | stream.close() |
| 637 | |
| 638 | # delete files that are not any more in local fs |
| 639 | for remote_file in remote_files.values(): |
| 640 | for file in remote_file: |
| 641 | self.fs.delete(file._id) |
| 642 | |
| 643 | def _get_mongo_files(self, from_path=None): |
| 644 | |
| 645 | file_dict = {} |
| garciadeblas | 2644b76 | 2021-03-24 09:21:01 +0100 | [diff] [blame] | 646 | file_cursor = self.fs.find(no_cursor_timeout=True, sort=[("uploadDate", -1)]) |
| lloretgalleg | f296d2a | 2020-09-02 09:36:24 +0000 | [diff] [blame] | 647 | for file in file_cursor: |
| 648 | if from_path and not file.filename.startswith(from_path): |
| 649 | continue |
| 650 | if file.filename in file_dict: |
| 651 | file_dict[file.filename].append(file) |
| 652 | else: |
| 653 | file_dict[file.filename] = [file] |
| 654 | return file_dict |
| 655 | |
| 656 | def reverse_sync(self, from_path: str): |
| 657 | """ |
| 658 | Sync from local storage to FSMongo |
| 659 | :param from_path: base directory to upload content to mongo fs |
| 660 | :return: None |
| 661 | """ |
| 662 | if os.path.isabs(from_path): |
| 663 | from_path = os.path.relpath(from_path, self.path) |
| 664 | self._update_mongo_fs(from_path=from_path) |