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 create_list(self, table, indata_list):
 145         """
 146         Add several entries at once
 147         :param table: collection or table
 148         :param indata_list: list of elements to insert. Each element must be a dictionary.
 149             An '_id' key based on random uuid is added at each element if missing
 150         :return: list of inserted '_id's. Exception on error
 151         """
 152         raise DbException("Method 'create_list' not implemented")
 153
 154     def set_one(self, table, q_filter, update_dict, fail_on_empty=True, unset=None, pull=None, push=None):
 155         """
 156         Modifies an entry at database
 157         :param table: collection or table
 158         :param q_filter: Filter
 159         :param update_dict: Plain dictionary with the content to be updated. It is a dot separated keys and a value
 160         :param fail_on_empty: If nothing matches filter it returns None unless this flag is set tu True, in which case
 161         it raises a DbException
 162         :param unset: Plain dictionary with the content to be removed if exist. It is a dot separated keys, value is
 163                       ignored. If not exist, it is ignored
 164         :param pull: Plain dictionary with the content to be removed from an array. It is a dot separated keys and value
 165                      if exist in the array is removed. If not exist, it is ignored
 166         :param push: Plain dictionary with the content to be appended to an array. It is a dot separated keys and value
 167                      is appended to the end of the array
 168         :return: Dict with the number of entries modified. None if no matching is found.
 169         """
 170         raise DbException("Method 'set_one' not implemented")
 171
 172     def set_list(self, table, q_filter, update_dict, unset=None, pull=None, push=None):
 173         """
 174         Modifies al matching entries at database
 175         :param table: collection or table
 176         :param q_filter: Filter
 177         :param update_dict: Plain dictionary with the content to be updated. It is a dot separated keys and a value
 178         :param unset: Plain dictionary with the content to be removed if exist. It is a dot separated keys, value is
 179                       ignored. If not exist, it is ignored
 180         :param pull: Plain dictionary with the content to be removed from an array. It is a dot separated keys and value
 181                      if exist in the array is removed. If not exist, it is ignored
 182         :param push: Plain dictionary with the content to be appended to an array. It is a dot separated keys and value
 183                      is appended to the end of the array
 184         :return: Dict with the number of entries modified
 185         """
 186         raise DbException("Method 'set_list' not implemented")
 187
 188     def replace(self, table, _id, indata, fail_on_empty=True):
 189         """
 190         Replace the content of an entry
 191         :param table: collection or table
 192         :param _id: internal database id
 193         :param indata: content to replace
 194         :param fail_on_empty: If nothing matches filter it returns None unless this flag is set tu True, in which case
 195         it raises a DbException
 196         :return: Dict with the number of entries replaced
 197         """
 198         raise DbException("Method 'replace' not implemented")
 199
 200     def _join_secret_key(self, update_key):
 201         """
 202         Returns a xor byte combination of the internal secret_key and the provided update_key.
 203         It does not modify the internal secret_key. Used for adding salt, join keys, etc.
 204         :param update_key: Can be a string, byte or None. Recommended a long one (e.g. 32 byte length)
 205         :return: joined key in bytes with a 32 bytes length. Can be None if both internal secret_key and update_key
 206                  are None
 207         """
 208         if not update_key:
 209             return self.secret_key
 210         elif isinstance(update_key, str):
 211             update_key_bytes = update_key.encode()
 212         else:
 213             update_key_bytes = update_key
 214
 215         new_secret_key = bytearray(self.secret_key) if self.secret_key else bytearray(32)
 216         for i, b in enumerate(update_key_bytes):
 217             new_secret_key[i % 32] ^= b
 218         return bytes(new_secret_key)
 219
 220     def set_secret_key(self, new_secret_key, replace=False):
 221         """
 222         Updates internal secret_key used for encryption, with a byte xor
 223         :param new_secret_key: string or byte array. It is recommended a 32 byte length
 224         :param replace: if True, old value of internal secret_key is ignored and replaced. If false, a byte xor is used
 225         :return: None
 226         """
 227         if replace:
 228             self.secret_key = None
 229         self.secret_key = self._join_secret_key(new_secret_key)
 230
 231     def get_secret_key(self):
 232         """
 233         Get the database secret key in case it is not done when "connect" is called. It can happens when database is
 234         empty after an initial install. It should skip if secret is already obtained.
 235         """
 236         pass
 237
 238     def encrypt(self, value, schema_version=None, salt=None):
 239         """
 240         Encrypt a value
 241         :param value: value to be encrypted. It is string/unicode
 242         :param schema_version: used for version control. If None or '1.0' no encryption is done.
 243                If '1.1' symmetric AES encryption is done
 244         :param salt: optional salt to be used. Must be str
 245         :return: Encrypted content of value
 246         """
 247         self.get_secret_key()
 248         if not self.secret_key or not schema_version or schema_version == '1.0':
 249             return value
 250         else:
 251             secret_key = self._join_secret_key(salt)
 252             cipher = AES.new(secret_key)
 253             padded_private_msg = value + ('\0' * ((16-len(value)) % 16))
 254             encrypted_msg = cipher.encrypt(padded_private_msg)
 255             encoded_encrypted_msg = b64encode(encrypted_msg)
 256             return encoded_encrypted_msg.decode("ascii")
 257
 258     def decrypt(self, value, schema_version=None, salt=None):
 259         """
 260         Decrypt an encrypted value
 261         :param value: value to be decrypted. It is a base64 string
 262         :param schema_version: used for known encryption method used. If None or '1.0' no encryption has been done.
 263                If '1.1' symmetric AES encryption has been done
 264         :param salt: optional salt to be used
 265         :return: Plain content of value
 266         """
 267         self.get_secret_key()
 268         if not self.secret_key or not schema_version or schema_version == '1.0':
 269             return value
 270         else:
 271             secret_key = self._join_secret_key(salt)
 272             encrypted_msg = b64decode(value)
 273             cipher = AES.new(secret_key)
 274             decrypted_msg = cipher.decrypt(encrypted_msg)
 275             try:
 276                 unpadded_private_msg = decrypted_msg.decode().rstrip('\0')
 277             except UnicodeDecodeError:
 278                 raise DbException("Cannot decrypt information. Are you using same COMMONKEY in all OSM components?",
 279                                   http_code=HTTPStatus.INTERNAL_SERVER_ERROR)
 280             return unpadded_private_msg
 281
 282     def encrypt_decrypt_fields(self, item, action, fields=None, flags=re.I, schema_version=None, salt=None):
 283         if not fields:
 284             return
 285         self.get_secret_key()
 286         actions = ['encrypt', 'decrypt']
 287         if action.lower() not in actions:
 288             raise DbException("Unknown action ({}): Must be one of {}".format(action, actions),
 289                               http_code=HTTPStatus.INTERNAL_SERVER_ERROR)
 290         method = self.encrypt if action.lower() == 'encrypt' else self.decrypt
 291
 292         def process(item):
 293             if isinstance(item, list):
 294                 for elem in item:
 295                     process(elem)
 296             elif isinstance(item, dict):
 297                 for key, val in item.items():
 298                     if any(re.search(f, key, flags) for f in fields) and isinstance(val, str):
 299                         item[key] = method(val, schema_version, salt)
 300                     else:
 301                         process(val)
 302
 303         process(item)
 304
 305
 306 def deep_update_rfc7396(dict_to_change, dict_reference, key_list=None):
 307     """
 308     Modifies one dictionary with the information of the other following https://tools.ietf.org/html/rfc7396
 309     Basically is a recursive python 'dict_to_change.update(dict_reference)', but a value of None is used to delete.
 310     It implements an extra feature that allows modifying an array. RFC7396 only allows replacing the entire array.
 311     For that, dict_reference should contains a dict with keys starting by "$" with the following meaning:
 312         $[index]    <index> is an integer for targeting a concrete index from dict_to_change array. If the value is None
 313                     the element of the array is deleted, otherwise it is edited.
 314         $+[index]   The value is inserted at this <index>. A value of None has not sense and an exception is raised.
 315         $+          The value is appended at the end. A value of None has not sense and an exception is raised.
 316         $val        It looks for all the items in the array dict_to_change equal to <val>. <val> is evaluated as yaml,
 317                     that is, numbers are taken as type int, true/false as boolean, etc. Use quotes to force string.
 318                     Nothing happens if no match is found. If the value is None the matched elements are deleted.
 319         $key: val   In case a dictionary is passed in yaml format, if looks for all items in the array dict_to_change
 320                     that are dictionaries and contains this <key> equal to <val>. Several keys can be used by yaml
 321                     format '{key: val, key: val, ...}'; and all of them must match. Nothing happens if no match is
 322                     found. If value is None the matched items are deleted, otherwise they are edited.
 323         $+val       If no match if found (see '$val'), the value is appended to the array. If any match is found nothing
 324                     is changed. A value of None has not sense.
 325         $+key: val  If no match if found (see '$key: val'), the value is appended to the array. If any match is found
 326                     nothing is changed. A value of None has not sense.
 327     If there are several editions, insertions and deletions; editions and deletions are done first in reverse index
 328         order; then insertions also in reverse index order; and finally appends in any order. So indexes used at
 329         insertions must take into account the deleted items.
 330     :param dict_to_change:  Target dictionary to be changed.
 331     :param dict_reference: Dictionary that contains changes to be applied.
 332     :param key_list: This is used internally for recursive calls. Do not fill this parameter.
 333     :return: none or raises and exception only at array modification when there is a bad format or conflict.
 334     """
 335     def _deep_update_array(array_to_change, _dict_reference, _key_list):
 336         to_append = {}
 337         to_insert_at_index = {}
 338         values_to_edit_delete = {}
 339         indexes_to_edit_delete = []
 340         array_edition = None
 341         _key_list.append("")
 342         for k in _dict_reference:
 343             _key_list[-1] = str(k)
 344             if not isinstance(k, str) or not k.startswith("$"):
 345                 if array_edition is True:
 346                     raise DbException("Found array edition (keys starting with '$') and pure dictionary edition in the"
 347                                       " same dict at '{}'".format(":".join(_key_list[:-1])))
 348                 array_edition = False
 349                 continue
 350             else:
 351                 if array_edition is False:
 352                     raise DbException("Found array edition (keys starting with '$') and pure dictionary edition in the"
 353                                       " same dict at '{}'".format(":".join(_key_list[:-1])))
 354                 array_edition = True
 355             insert = False
 356             indexes = []  # indexes to edit or insert
 357             kitem = k[1:]
 358             if kitem.startswith('+'):
 359                 insert = True
 360                 kitem = kitem[1:]
 361                 if _dict_reference[k] is None:
 362                     raise DbException("A value of None has not sense for insertions at '{}'".format(
 363                         ":".join(_key_list)))
 364
 365             if kitem.startswith('[') and kitem.endswith(']'):
 366                 try:
 367                     index = int(kitem[1:-1])
 368                     if index < 0:
 369                         index += len(array_to_change)
 370                     if index < 0:
 371                         index = 0  # skip outside index edition
 372                     indexes.append(index)
 373                 except Exception:
 374                     raise DbException("Wrong format at '{}'. Expecting integer index inside quotes".format(
 375                         ":".join(_key_list)))
 376             elif kitem:
 377                 # match_found_skip = False
 378                 try:
 379                     filter_in = yaml.safe_load(kitem)
 380                 except Exception:
 381                     raise DbException("Wrong format at '{}'. Expecting '$<yaml-format>'".format(":".join(_key_list)))
 382                 if isinstance(filter_in, dict):
 383                     for index, item in enumerate(array_to_change):
 384                         for filter_k, filter_v in filter_in.items():
 385                             if not isinstance(item, dict) or filter_k not in item or item[filter_k] != filter_v:
 386                                 break
 387                         else:  # match found
 388                             if insert:
 389                                 # match_found_skip = True
 390                                 insert = False
 391                                 break
 392                             else:
 393                                 indexes.append(index)
 394                 else:
 395                     index = 0
 396                     try:
 397                         while True:  # if not match a ValueError exception will be raise
 398                             index = array_to_change.index(filter_in, index)
 399                             if insert:
 400                                 # match_found_skip = True
 401                                 insert = False
 402                                 break
 403                             indexes.append(index)
 404                             index += 1
 405                     except ValueError:
 406                         pass
 407
 408                 # if match_found_skip:
 409                 #     continue
 410             elif not insert:
 411                 raise DbException("Wrong format at '{}'. Expecting '$+', '$[<index]' or '$[<filter>]'".format(
 412                     ":".join(_key_list)))
 413             for index in indexes:
 414                 if insert:
 415                     if index in to_insert_at_index and to_insert_at_index[index] != _dict_reference[k]:
 416                         # Several different insertions on the same item of the array
 417                         raise DbException("Conflict at '{}'. Several insertions on same array index {}".format(
 418                             ":".join(_key_list), index))
 419                     to_insert_at_index[index] = _dict_reference[k]
 420                 else:
 421                     if index in indexes_to_edit_delete and values_to_edit_delete[index] != _dict_reference[k]:
 422                         # Several different editions on the same item of the array
 423                         raise DbException("Conflict at '{}'. Several editions on array index {}".format(
 424                             ":".join(_key_list), index))
 425                     indexes_to_edit_delete.append(index)
 426                     values_to_edit_delete[index] = _dict_reference[k]
 427             if not indexes:
 428                 if insert:
 429                     to_append[k] = _dict_reference[k]
 430                 # elif _dict_reference[k] is not None:
 431                 #     raise DbException("Not found any match to edit in the array, or wrong format at '{}'".format(
 432                 #         ":".join(_key_list)))
 433
 434         # edition/deletion is done before insertion
 435         indexes_to_edit_delete.sort(reverse=True)
 436         for index in indexes_to_edit_delete:
 437             _key_list[-1] = str(index)
 438             try:
 439                 if values_to_edit_delete[index] is None:  # None->Anything
 440                     try:
 441                         del (array_to_change[index])
 442                     except IndexError:
 443                         pass  # it is not consider an error if this index does not exist
 444                 elif not isinstance(values_to_edit_delete[index], dict):  # NotDict->Anything
 445                     array_to_change[index] = deepcopy(values_to_edit_delete[index])
 446                 elif isinstance(array_to_change[index], dict):  # Dict->Dict
 447                     deep_update_rfc7396(array_to_change[index], values_to_edit_delete[index], _key_list)
 448                 else:  # Dict->NotDict
 449                     if isinstance(array_to_change[index], list):  # Dict->List. Check extra array edition
 450                         if _deep_update_array(array_to_change[index], values_to_edit_delete[index], _key_list):
 451                             continue
 452                     array_to_change[index] = deepcopy(values_to_edit_delete[index])
 453                     # calling deep_update_rfc7396 to delete the None values
 454                     deep_update_rfc7396(array_to_change[index], values_to_edit_delete[index], _key_list)
 455             except IndexError:
 456                 raise DbException("Array edition index out of range at '{}'".format(":".join(_key_list)))
 457
 458         # insertion with indexes
 459         to_insert_indexes = list(to_insert_at_index.keys())
 460         to_insert_indexes.sort(reverse=True)
 461         for index in to_insert_indexes:
 462             array_to_change.insert(index, to_insert_at_index[index])
 463
 464         # append
 465         for k, insert_value in to_append.items():
 466             _key_list[-1] = str(k)
 467             insert_value_copy = deepcopy(insert_value)
 468             if isinstance(insert_value_copy, dict):
 469                 # calling deep_update_rfc7396 to delete the None values
 470                 deep_update_rfc7396(insert_value_copy, insert_value, _key_list)
 471             array_to_change.append(insert_value_copy)
 472
 473         _key_list.pop()
 474         if array_edition:
 475             return True
 476         return False
 477
 478     if key_list is None:
 479         key_list = []
 480     key_list.append("")
 481     for k in dict_reference:
 482         key_list[-1] = str(k)
 483         if dict_reference[k] is None:   # None->Anything
 484             if k in dict_to_change:
 485                 del dict_to_change[k]
 486         elif not isinstance(dict_reference[k], dict):  # NotDict->Anything
 487             dict_to_change[k] = deepcopy(dict_reference[k])
 488         elif k not in dict_to_change:  # Dict->Empty
 489             dict_to_change[k] = deepcopy(dict_reference[k])
 490             # calling deep_update_rfc7396 to delete the None values
 491             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 492         elif isinstance(dict_to_change[k], dict):  # Dict->Dict
 493             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 494         else:       # Dict->NotDict
 495             if isinstance(dict_to_change[k], list):  # Dict->List. Check extra array edition
 496                 if _deep_update_array(dict_to_change[k], dict_reference[k], key_list):
 497                     continue
 498             dict_to_change[k] = deepcopy(dict_reference[k])
 499             # calling deep_update_rfc7396 to delete the None values
 500             deep_update_rfc7396(dict_to_change[k], dict_reference[k], key_list)
 501     key_list.pop()
 502
 503
 504 def deep_update(dict_to_change, dict_reference):
 505     """ Maintained for backward compatibility. Use deep_update_rfc7396 instead"""
 506     return deep_update_rfc7396(dict_to_change, dict_reference)