osm_common/dbbase.py

   1 # -*- coding: utf-8 -*-
   2
   3 # Copyright 2018 Telefonica S.A.
   4 #
   5 # Licensed under the Apache License, Version 2.0 (the "License");
   6 # you may not use this file except in compliance with the License.
   7 # You may obtain a copy of the License at
   8 #
   9 #    http://www.apache.org/licenses/LICENSE-2.0
  10 #
  11 # Unless required by applicable law or agreed to in writing, software
  12 # distributed under the License is distributed on an "AS IS" BASIS,
  13 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
  14 # implied.
  15 # See the License for the specific language governing permissions and
  16 # limitations under the License.
  17
  18 import yaml
  19 import logging
  20 import re
  21 from http import HTTPStatus
  22 from copy import deepcopy
  23 from Crypto.Cipher import AES
  24 from base64 import b64decode, b64encode
  25 from osm_common.common_utils import FakeLock
  26 from threading import Lock
  27
  28 __author__ = "Alfonso Tierno <alfonso.tiernosepulveda@telefonica.com>"
  29
  30
  31 class DbException(Exception):
  32
  33     def __init__(self, message, http_code=HTTPStatus.NOT_FOUND):
  34         self.http_code = http_code
  35         Exception.__init__(self, "database exception " + str(message))
  36
  37
  38 class DbBase(object):
  39
  40     def __init__(self, logger_name='db', lock=False):
  41         """
  42         Constructor of dbBase
  43         :param logger_name: logging name
  44         :param lock: Used to protect simultaneous access to the same instance class by several threads:
  45             False, None: Do not protect, this object will only be accessed by one thread
  46             True: This object needs to be protected by several threads accessing.
  47             Lock object. Use thi Lock for the threads access protection
  48         """
  49         self.logger = logging.getLogger(logger_name)
  50         self.secret_key = None  # 32 bytes length array used for encrypt/decrypt
  51         if not lock:
  52             self.lock = FakeLock()
  53         elif lock is True:
  54             self.lock = Lock()
  55         elif isinstance(lock, Lock):
  56             self.lock = lock
  57         else:
  58             raise ValueError("lock parameter must be a Lock classclass or boolean")
  59
  60     def db_connect(self, config, target_version=None):
  61         """
  62         Connect to database
  63         :param config: Configuration of database. Contains among others:
  64             host:   database host (mandatory)
  65             port:   database port (mandatory)
  66             name:   database name (mandatory)
  67             user:   database username
  68             password:   database password
  69             commonkey: common OSM key used for sensible information encryption
  70             materpassword: same as commonkey, for backward compatibility. Deprecated, to be removed in the future
  71         :param target_version: if provided it checks if database contains required version, raising exception otherwise.
  72         :return: None or raises DbException on error
  73         """
  74         raise DbException("Method 'db_connect' not implemented")
  75
  76     def db_disconnect(self):
  77         """
  78         Disconnect from database
  79         :return: None
  80         """
  81         pass
  82
  83     def get_list(self, table, q_filter=None):
  84         """
  85         Obtain a list of entries matching q_filter
  86         :param table: collection or table
  87         :param q_filter: Filter
  88         :return: a list (can be empty) with the found entries. Raises DbException on error
  89         """
  90         raise DbException("Method 'get_list' not implemented")
  91
  92     def count(self, table, q_filter=None):
  93         """
  94         Count the number of entries matching q_filter
  95         :param table: collection or table
  96         :param q_filter: Filter
  97         :return: number of entries found (can be zero)
  98         :raise: DbException on error
  99         """
 100         raise DbException("Method 'count' not implemented")
 101
 102     def get_one(self, table, q_filter=None, fail_on_empty=True, fail_on_more=True):
 103         """
 104         Obtain one entry matching q_filter
 105         :param table: collection or table
 106         :param q_filter: Filter
 107         :param fail_on_empty: If nothing matches filter it returns None unless this flag is set tu True, in which case
 108         it raises a DbException
 109         :param fail_on_more: If more than one matches filter it returns one of then unless this flag is set tu True, so
 110         that it raises a DbException
 111         :return: The requested element, or None
 112         """
 113         raise DbException("Method 'get_one' not implemented")
 114
 115     def del_list(self, table, q_filter=None):
 116         """
 117         Deletes all entries that match q_filter
 118         :param table: collection or table
 119         :param q_filter: Filter
 120         :return: Dict with the number of entries deleted
 121         """
 122         raise DbException("Method 'del_list' not implemented")
 123
 124     def del_one(self, table, q_filter=None, fail_on_empty=True):
 125         """
 126         Deletes one entry that matches q_filter
 127         :param table: collection or table
 128         :param q_filter: Filter
 129         :param fail_on_empty: If nothing matches filter it returns '0' deleted unless this flag is set tu True, in
 130         which case it raises a DbException
 131         :return: Dict with the number of entries deleted
 132         """
 133         raise DbException("Method 'del_one' not implemented")
 134
 135     def create(self, table, indata):
 136         """
 137         Add a new entry at database
 138         :param table: collection or table
 139         :param indata: content to be added
 140         :return: database id of the inserted element. Raises a DbException on error
 141         """
 142         raise DbException("Method 'create' not implemented")
 143
 144     def set_one(self, table, q_filter, update_dict, fail_on_empty=True, unset=None, pull=None, push=None):
 145         """
 146         Modifies an entry at database
 147         :param table: collection or table
 148         :param q_filter: Filter
 149         :param update_dict: Plain dictionary with the content to be updated. It is a dot separated keys and a value
 150         :param fail_on_empty: If nothing matches filter it returns None unless this flag is set tu True, in which case
 151         it raises a DbException
 152         :param unset: Plain dictionary with the content to be removed if exist. It is a dot separated keys, value is
 153                       ignored. If not exist, it is ignored
 154         :param pull: Plain dictionary with the content to be removed from an array. It is a dot separated keys and value
 155                      if exist in the array is removed. If not exist, it is ignored
 156         :param push: Plain dictionary with the content to be appended to an array. It is a dot separated keys and value
 157                      is appended to the end of the array
 158         :return: Dict with the number of entries modified. None if no matching is found.
 159         """
 160         raise DbException("Method 'set_one' not implemented")
 161
 162     def set_list(self, table, q_filter, update_dict):
 163         """
 164         Modifies al matching entries at database
 165         :param table: collection or table
 166         :param q_filter: Filter
 167         :param update_dict: Plain dictionary with the content to be updated. It is a dot separated keys and a value
 168         :return: Dict with the number of entries modified
 169         """
 170         raise DbException("Method 'set_list' not implemented")
 171
 172     def replace(self, table, _id, indata, fail_on_empty=True):
 173         """
 174         Replace the content of an entry
 175         :param table: collection or table
 176         :param _id: internal database id
 177         :param indata: content to replace
 178         :param fail_on_empty: If nothing matches filter it returns None unless this flag is set tu True, in which case
 179         it raises a DbException
 180         :return: Dict with the number of entries replaced
 181         """
 182         raise DbException("Method 'replace' not implemented")
 183
 184     def _join_secret_key(self, update_key):
 185         """
 186         Returns a xor byte combination of the internal secret_key and the provided update_key.
 187         It does not modify the internal secret_key. Used for adding salt, join keys, etc.
 188         :param update_key: Can be a string, byte or None. Recommended a long one (e.g. 32 byte length)
 189         :return: joined key in bytes with a 32 bytes length. Can be None if both internal secret_key and update_key
 190                  are None
 191         """
 192         if not update_key:
 193             return self.secret_key
 194         elif isinstance(update_key, str):
 195             update_key_bytes = update_key.encode()
 196         else:
 197             update_key_bytes = update_key
 198
 199         new_secret_key = bytearray(self.secret_key) if self.secret_key else bytearray(32)
 200         for i, b in enumerate(update_key_bytes):
 201             new_secret_key[i % 32] ^= b
 202         return bytes(new_secret_key)
 203
 204     def set_secret_key(self, new_secret_key, replace=False):
 205         """
 206         Updates internal secret_key used for encryption, with a byte xor
 207         :param new_secret_key: string or byte array. It is recommended a 32 byte length
 208         :param replace: if True, old value of internal secret_key is ignored and replaced. If false, a byte xor is used
 209         :return: None
 210         """
 211         if replace:
 212             self.secret_key = None
 213         self.secret_key = self._join_secret_key(new_secret_key)
 214
 215     def get_secret_key(self):
 216         """
 217         Get the database secret key in case it is not done when "connect" is called. It can happens when database is
 218         empty after an initial install. It should skip if secret is already obtained.
 219         """
 220         pass
 221
 222     def encrypt(self, value, schema_version=None, salt=None):
 223         """
 224         Encrypt a value
 225         :param value: value to be encrypted. It is string/unicode
 226         :param schema_version: used for version control. If None or '1.0' no encryption is done.
 227                If '1.1' symmetric AES encryption is done
 228         :param salt: optional salt to be used. Must be str
 229         :return: Encrypted content of value
 230         """
 231         self.get_secret_key()
 232         if not self.secret_key or not schema_version or schema_version == '1.0':
 233             return value
 234         else:
 235             secret_key = self._join_secret_key(salt)
 236             cipher = AES.new(secret_key)
 237             padded_private_msg = value + ('\0' * ((16-len(value)) % 16))
 238             encrypted_msg = cipher.encrypt(padded_private_msg)
 239             encoded_encrypted_msg = b64encode(encrypted_msg)
 240             return encoded_encrypted_msg.decode("ascii")
 241
 242     def decrypt(self, value, schema_version=None, salt=None):
 243         """
 244         Decrypt an encrypted value
 245         :param value: value to be decrypted. It is a base64 string
 246         :param schema_version: used for known encryption method used. If None or '1.0' no encryption has been done.
 247                If '1.1' symmetric AES encryption has been done
 248         :param salt: optional salt to be used
 249         :return: Plain content of value
 250         """
 251         self.get_secret_key()
 252         if not self.secret_key or not schema_version or schema_version == '1.0':
 253             return value
 254         else:
 255             secret_key = self._join_secret_key(salt)
 256             encrypted_msg = b64decode(value)
 257             cipher = AES.new(secret_key)
 258             decrypted_msg = cipher.decrypt(encrypted_msg)
 259             try:
 260                 unpadded_private_msg = decrypted_msg.decode().rstrip('\0')
 261             except UnicodeDecodeError:
 262                 raise DbException("Cannot decrypt information. Are you using same COMMONKEY in all OSM components?",
 263                                   http_code=HTTPStatus.INTERNAL_SERVER_ERROR)
 264             return unpadded_private_msg
 265
 266     def encrypt_decrypt_fields(self, item, action, fields=None, flags=re.I, schema_version=None, salt=None):
 267         if not fields:
 268             return
 269         self.get_secret_key()
 270         actions = ['encrypt', 'decrypt']
 271         if action.lower() not in actions:
 272             raise DbException("Unknown action ({}): Must be one of {}".format(action, actions),
 273                               http_code=HTTPStatus.INTERNAL_SERVER_ERROR)
 274         method = self.encrypt if action.lower() == 'encrypt' else self.decrypt
 275
 276         def process(item):
 277             if isinstance(item, list):
 278                 for elem in item:
 279                     process(elem)
 280             elif isinstance(item, dict):
 281                 for key, val in item.items():
 282                     if any(re.search(f, key, flags) for f in fields) and isinstance(val, str):
 283                         item[key] = method(val, schema_version, salt)
 284                     else:
 285                         process(val)
 286
 287         process(item)
 288
 289
 290 def deep_update_rfc7396(dict_to_change, dict_reference, key_list=None):
 291     """
 292     Modifies one dictionary with the information of the other following https://tools.ietf.org/html/rfc7396
 293     Basically is a recursive python 'dict_to_change.update(dict_reference)', but a value of None is used to delete.
 294     It implements an extra feature that allows modifying an array. RFC7396 only allows replacing the entire array.
 295     For that, dict_reference should contains a dict with keys starting by "$" with the following meaning:
 296         $[index]    <index> is an integer for targeting a concrete index from dict_to_change array. If the value is None
 297                     the element of the array is deleted, otherwise it is edited.
 298         $+[index]   The value is inserted at this <index>. A value of None has not sense and an exception is raised.
 299         $+          The value is appended at the end. A value of None has not sense and an exception is raised.
 300         $val        It looks for all the items in the array dict_to_change equal to <val>. <val> is evaluated as yaml,
 301                     that is, numbers are taken as type int, true/false as boolean, etc. Use quotes to force string.
 302                     Nothing happens if no match is found. If the value is None the matched elements are deleted.
 303         $key: val   In case a dictionary is passed in yaml format, if looks for all items in the array dict_to_change
 304                     that are dictionaries and contains this <key> equal to <val>. Several keys can be used by yaml
 305                     format '{key: val, key: val, ...}'; and all of them must match. Nothing happens if no match is
 306                     found. If value is None the matched items are deleted, otherwise they are edited.
 307         $+val       If no match if found (see '$val'), the value is appended to the array. If any match is found nothing
 308                     is changed. A value of None has not sense.
 309         $+key: val  If no match if found (see '$key: val'), the value is appended to the array. If any match is found
 310                     nothing is changed. A value of None has not sense.
 311     If there are several editions, insertions and deletions; editions and deletions are done first in reverse index
 312         order; then insertions also in reverse index order; and finally appends in any order. So indexes used at
 313         insertions must take into account the deleted items.
 314     :param dict_to_change:  Target dictionary to be changed.
 315     :param dict_reference: Dictionary that contains changes to be applied.
 316     :param key_list: This is used internally for recursive calls. Do not fill this parameter.
 317     :return: none or raises and exception only at array modification when there is a bad format or conflict.
 318     """
 319     def _deep_update_array(array_to_change, _dict_reference, _key_list):
 320         to_append = {}
 321         to_insert_at_index = {}
 322         values_to_edit_delete = {}
 323         indexes_to_edit_delete = []
 324         array_edition = None
 325         _key_list.append("")
 326         for k in _dict_reference:
 327             _key_list[-1] = str(k)
 328             if not isinstance(k, str) or not k.startswith("$"):
 329                 if array_edition is True:
 330                     raise DbException("Found array edition (keys starting with '$') and pure dictionary edition in the"
 331                                       " same dict at '{}'".format(":".join(_key_list[:-1])))
 332                 array_edition = False
 333                 continue
 334             else:
 335                 if array_edition is False:
 336                     raise DbException("Found array edition (keys starting with '$') and pure dictionary edition in the"
 337                                       " same dict at '{}'".format(":".join(_key_list[:-1])))
 338                 array_edition = True
 339             insert = False
 340             indexes = []  # indexes to edit or insert
 341             kitem = k[1:]
 342             if kitem.startswith('+'):
 343                 insert = True
 344                 kitem = kitem[1:]
 345                 if _dict_reference[k] is None:
 346                     raise DbException("A value of None has not sense for insertions at '{}'".format(
 347                         ":".join(_key_list)))
 348
 349             if kitem.startswith('[') and kitem.endswith(']'):
 350                 try:
 351                     index = int(kitem[1:-1])
 352                     if index < 0:
 353                         index += len(array_to_change)
 354                     if index < 0:
 355                         index = 0  # skip outside index edition
 356                     indexes.append(index)
 357                 except Exception:
 358                     raise DbException("Wrong format at '{}'. Expecting integer index inside quotes".format(
 359                         ":".join(_key_list)))
 360             elif kitem:
 361                 # match_found_skip = False
 362                 try:
 363                     filter_in = yaml.safe_load(kitem)
 364                 except Exception:
 365                     raise DbException("Wrong format at '{}'. Expecting '$<yaml-format>'".format(":".join(_key_list)))
 366                 if isinstance(filter_in, dict):
 367                     for index, item in enumerate(array_to_change):
 368                         for filter_k, filter_v in filter_in.items():
 369                             if not isinstance(item, dict) or filter_k not in item or item[filter_k] != filter_v:
 370                                 break
 371                         else:  # match found
 372                             if insert:
 373                                 # match_found_skip = True
 374                                 insert = False
 375                                 break
 376                             else:
 377                                 indexes.append(index)
 378                 else:
 379                     index = 0
 380                     try:
 381                         while True:  # if not match a ValueError exception will be raise
 382                             index = array_to_change.index(filter_in, index)
 383                             if insert:
 384                                 # match_found_skip = True
 385                                 insert = False
 386                                 break
 387                             indexes.append(index)
 388                             index += 1
 389                     except ValueError:
 390                         pass
 391
 392                 # if match_found_skip:
 393                 #     continue
 394             elif not insert:
 395                 raise DbException("Wrong format at '{}'. Expecting '$+', '$[<index]' or '$[<filter>]'".format(
 396                     ":".join(_key_list)))
 397             for index in indexes:
 398                 if insert:
 399                     if index in to_insert_at_index and to_insert_at_index[index] != _dict_reference[k]:
 400                         # Several different insertions on the same item of the array
 401                         raise DbException("Conflict at '{}'. Several insertions on same array index {}".format(
 402                             ":".join(_key_list), index))
 403                     to_insert_at_index[index] = _dict_reference[k]
 404                 else:
 405                     if index in indexes_to_edit_delete and values_to_edit_delete[index] != _dict_reference[k]:
 406                         # Several different editions on the same item of the array
 407                         raise DbException("Conflict at '{}'. Several editions on array index {}".format(
 408                             ":".join(_key_list), index))
 409                     indexes_to_edit_delete.append(index)
 410                     values_to_edit_delete[index] = _dict_reference[k]
 411             if not indexes:
 412                 if insert:
 413                     to_append[k] = _dict_reference[k]
 414                 # elif _dict_reference[k] is not None:
 415                 #     raise DbException("Not found any match to edit in the array, or wrong format at '{}'".format(
 416                 #         ":".join(_key_list)))
 417
 418         # edition/deletion is done before insertion
 419         indexes_to_edit_delete.sort(reverse=True)
 420         for index in indexes_to_edit_delete:
 421             _key_list[-1] = str(index)
 422             try:
 423                 if values_to_edit_delete[index] is None:  # None->Anything
 424                     try:
 425                         del (array_to_change[index])
 426                     except IndexError:
 427                         pass  # it is not consider an error if this index does not exist
 428                 elif not isinstance(values_to_edit_delete[index], dict):  # NotDict->Anything
 429                     array_to_change[index] = deepcopy(values_to_edit_delete[index])
 430                 elif isinstance(array_to_change[index], dict):  # Dict->Dict
 431                     deep_update_rfc7396(array_to_change[index], values_to_edit_delete[index], _key_list)
 432                 else:  # Dict->NotDict
 433                     if isinstance(array_to_change[index], list):  # Dict->List. Check extra array edition
 434                         if _deep_update_array(array_to_change[index], values_to_edit_delete[index], _key_list):
 435                             continue
 436                     array_to_change[index] = deepcopy(values_to_edit_delete[index])
 437                     # calling deep_update_rfc7396 to delete the None values
 438                     deep_update_rfc7396(array_to_change[index], values_to_edit_delete[index], _key_list)
 439             except IndexError:
 440                 raise DbException("Array edition index out of range at '{}'".format(":".join(_key_list)))
 441
 442         # insertion with indexes
 443         to_insert_indexes = list(to_insert_at_index.keys())
 444         to_insert_indexes.sort(reverse=True)
 445         for index in to_insert_indexes:
 446             array_to_change.insert(index, to_insert_at_index[index])
 447
 448         # append
 449         for k, insert_value in to_append.items():
 450             _key_list[-1] = str(k)
 451             insert_value_copy = deepcopy(insert_value)
 452             if isinstance(insert_value_copy, dict):
 453                 # calling deep_update_rfc7396 to delete the None values
 454                 deep_update_rfc7396(insert_value_copy, insert_value, _key_list)
 455             array_to_change.append(insert_value_copy)
 456
 457         _key_list.pop()
 458         if array_edition:
 459             return True
 460         return False
 461
 462     if key_list is None:
 463         key_list = []
 464     key_list.append("")
 465     for k in dict_reference:
 466         key_list[-1] = str(k)
 467         if dict_reference[k] is None:   # None->Anything
 468             if k in dict_to_change:
 469                 del dict_to_change[k]
 470         elif not isinstance(dict_reference[k], dict):  # NotDict->Anything
 471             dict_to_change[k] = deepcopy(dict_reference[k])
 472         elif k not in dict_to_change:  # Dict->Empty
 473             dict_to_change[k] = deepcopy(dict_reference[k])
 474             # calling deep_update_rfc7396 to delete the None values
 475             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 476         elif isinstance(dict_to_change[k], dict):  # Dict->Dict
 477             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 478         else:       # Dict->NotDict
 479             if isinstance(dict_to_change[k], list):  # Dict->List. Check extra array edition
 480                 if _deep_update_array(dict_to_change[k], dict_reference[k], key_list):
 481                     continue
 482             dict_to_change[k] = deepcopy(dict_reference[k])
 483             # calling deep_update_rfc7396 to delete the None values
 484             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 485     key_list.pop()
 486
 487
 488 def deep_update(dict_to_change, dict_reference):
 489     """ Maintained for backward compatibility. Use deep_update_rfc7396 instead"""
 490     return deep_update_rfc7396(dict_to_change, dict_reference)