Elara DB is an easy to use, lightweight key-value database written for python that can also be used as a fast in-memory cache for JSON-serializable data. Includes various methods and features to manipulate data structures in-memory, protect database files and export data.
$ pip install elara
Go through the release notes for details on upgrades as breaking changes might happen between version upgrades while Elara is in beta.
Elara DB has official support for python 3.6+
.
Key Features :
-
Offers three modes of execution - normal, cache and secure - secure mode generates a key file and encrypts the database for additional security.
-
Manipulate data structures such as strings, lists and dictionaries.
-
Fast and flexible in-memory LRU caching mechanism.
-
Supports keys of any type, not just strings!
-
Choose between manual commits after performing operations in-memory or automatically commit every change into the storage.
-
Includes methods to export certain keys from the database or the entire storage.
-
Incorporates checksums to verify database file integrity.
Table of Contents :
Installation
From pypi :
$ pip install elara
OR,
Clone the repository and install the dependencies :
$ pip install -r requirements.py
$ python -m unittest -v # Run tests
License
Copyright (c) 2021, Saurabh Pujari
All rights reserved.
This source code is licensed under the BSD-style license found in the LICENSE file in the root directory of this source tree.
Go back to the table of contents
Fundamentals
Basic usage :
import elara
db = elara.exe("new.db")
db.set("name", "Elara")
print(db.get("name"))
# Elara
You can choose between normally transacting data from the database or you can protect your database with a key.
import elara
# exe_secure() encrypts the db file
db = elara.exe_secure("new.db", True, "newdb.key")
#OR
# db = elara.exe_secure(path="path/new.db", commitdb=True, key_path="path/newdb.key")
db.set("name", "Elara")
print(db.get("name"))
# Elara
exe_secure(path, commitdb=False, key_path="edb.key")
- Loads the contents of the encrypted database (using the key file) into the program memory or generates a new key file (default -edb.key
) if it doesn't exist in the given path and it encrypts/decrypts the database file.
Using exe_secure()
without a key file or without the correct key file corresponding to the database will result in errors. Database files are verified with checksums to maintain integrity. Key files and DB files can be included inside the .gitignore
to ensure they're not pushed into an upstream repository.
commitdb
- this argument defaults toFalse
ie. you will have to manually call thecommit()
method to write the in-memory changes into the database. If set toTrue
, changes will be written into the file after every operation.
import elara
db = elara.exe_secure(path="path/new.db", key_path="path/edb.key") # commit=False
db.set("num", 20)
print(db.get("num"))
# 20
db.commit() # Writes in-memory changes into the file
exe(path, commitdb=False)
- Loads the contents of the database into the program memory or generates a new database file if it doesn't exist in the given path. The database file is NOT protected and can be accessed without a key.
import elara
db = elara.exe("new.db", True)
db.set("name", "Elara")
print(db.get("name"))
# Elara
All the following operations are methods that can be applied to the instance returned from exe()
or exe_secure()
(or exe_cache()
, as shown in the Cache section). These operations manipulate/analyse data in-memory after the Data is loaded from the file. Set the commit
argument to True
else manually use the commit()
method to sync in-memory data with the database file.
-
get(key)
- returns the corresponding value from the db orNone
-
set(key, value)
- returnsTrue
or an Exception. Thekey
can be any data type that is supported by python dictionaries (int, float, string etc.). -
rem(key)
- deletes the key-value pair if it exists. -
remkeys(keys=[])
- deletes all the key-value pairs from the list of keys given, if the key exists. -
incr(key, val=1)
- increments the value (has to be anint
or afloat
) present at the given key with theval
parameter (default1
,int
or afloat
). For float operations it rounds the result to 3 decimal points. -
decr(key, val=1)
- decrements the value (has to be anint
or afloat
) present at the given key with theval
parameter (default1
,int
or afloat
). For float operations it rounds the result to 3 decimal points. -
clear()
- clears the database data currently stored in-memory. -
exists(key)
- returnsTrue
if the key exists. -
commit()
- write in-memory changes into the database file.
-
getset(key, value)
- Sets the new value and returns the old value for that key or returnsFalse
. -
getkeys()
- returns the list of keys in the database with. The list is ordered with themost recently accessed
keys starting from index 0. -
numkeys()
- returns the number of keys in the database. -
getmatch(match)
- Takes thematch
argument and returns a Dictionary of key-value pairs of which the keys containmatch
as a sub string.
-
retkey()
- returns the Key used to encrypt/decrypt the db file; returnsNone
if the file is unprotected. -
retmem()
- returns all the in-memory db contents. -
retdb()
- returns all the db file contents.
Go back to the table of contents
import elara
db = elara.exe("new.db")
db.set("num1", 20)
# ("num1", 20) is written into the file db
db.commit()
db.set("num2", 30)
print(db.retmem())
# {'num1': 20, 'num2': 30}
print(db.retdb())
# {'num1': 20}
Note - retmem()
and retdb()
will return the same value if commit
is set to True
or if the commit()
method is used before calling retdb()
Go back to the table of contents
Elara adds some syntax sugar for get(), set() and rem() :
import elara
db = elara.exe("new.db")
db["key"] = "value"
print(db["key"])
# value
del db["key"]
print(db.retmem())
# {}
Go back to the table of contents
Cache:
Elara can also be used as a fast in-memory cache.
-
exe_cache(path, cache_param=None, commit=False)
- This function creates an instance with the settings defined incache_param
. Herecommit
defaults toFalse
to allow for in-memory manipulation.cache_param
- This argument is a dictionary that can define of 3optional
parameters.-
max_age
- This is the default amount of time inseconds
that any key stored (eg. usingset()
) into the cache will last for before being evicted. Defaults toNone
which indicates it will stay in memory for as long as the instance is running. -
max_size
- This is the maximum number of keys that will be stored in the cache. For every key addition request after themax_size
limit has been reached, an automaticcull()
is called to evict some keys based oncull_freq
. Defaults to positive infinity as limited by the device.
cull_freq
- This is the default amount of keys, in percentage, that will be evicted based on the LRU eviction strategy when the cache reaches itsmax_size
. 0 <=cull_freq
<=100. Defaults to20
ie. 20% of all keys will be deleted based on the LRU eviction strategy.
-
-
Some key points :
-
The LRU eviction searches for, and deletes, expired keys lazily after every function call.
-
In
exe_cache
, thepath
parameter is a required argument in case you need to commit your cache contents into the database. -
Once data is commited to the file, Elara
stops
keeping track of key-value expiry, cache max size etc. and resets to defaults. Hence, customcache_param
andmax_age
values only work while operatingin-memory
.
-
-
set(key, value, max_age=None)
- Theset()
function takes another argument,max_age
, that is set toNone
by default ie. the key-value pair will follow the defaultmax_age
set incache_param
OR they stay never get evicted ifcache_param
is not defined. Themax_age
param inset()
allows for more granular control over cache item expiry.max_age
should be an integer greater than 0.max_age = "i"
indicates the item will not be removed from memory (overrides defaultmax_age
ormax_age
defined incache_param
)
Similarly, lnew(key, max_age=None)
, hnew(key, max_age=None)
(read the API reference) and getset(key, value, max_age=None)
, all accept the optional max_age
argument.
import elara
import time
cache_param = {
"max_age": 900,
"max_size": 4,
"cull_freq": 25
}
cache = elara.exe_cache(path="new.db", cache_param=cache_param)
# OR
# cache = elara.exe_cache("new.db", cache_param)
cache.set("key1", "This one will be evicted in 900 seconds")
cache.set("key2", "This one will not be evicted", "i") # 'i' signifies it will never be evicted
cache.set("key3", "This one will be evicted in 50 seconds", 50)
print(cache.getkeys())
# ["key3", "key2", "key1"]
time.sleep(50)
print(cache.getkeys())
# ["key2", "key1"]
cache.set("key3", 5)
cache.set("key4", 10)
print(cache.getkeys())
# ["key4", "key3", "key2", "key1"]
cache.set("key1", 7, "i") # overwrite "key1" to never expire
print(cache.getkeys())
# ["key1", "key4", "key3", "key2"]
print(cache.get("key1"))
# 7
cache.set("key5", 20) # Automatic culling when max_size is reached
print(cache.getkeys())
# ["key5", "key1", "key4", "key3"]
Go back to the table of contents
Elara also allows for manual culling of cached items :
cull(percentage)
-percentage
(0 <= percentage <= 100) defines the percentage of Key-Value pairs to be deleted, with theLeast recently accessed
keys being deleted first. Elara maintains a simple LRU cache to track key access.
import elara
"""
Without the cache_param argument, all defauls will be set
Passing any one of the values is also valid as mentioned above
cache = elara.exe_cache("new.db", {"max_size": 100}))
"""
cache = elara.exe_cache("new.db")
cache.set("num1", 10)
cache.set("num2", 20)
cache.set("num3", 30)
cache.set("num4", 40)
if cache.exists("num1"):
print(cache.get("num1"))
# 10
print(cache.retmem())
# {'num1': 10, 'num2': 20, 'num3': 30, 'num4': 40}
# most recently accessed keys come first
print(cache.getkeys())
# ['num1', 'num4', 'num3', 'num2']
# delete 25% of the stale keys (follows LRU)
cache.cull(25)
# most recently accessed keys come first
print(cache.getkeys())
# ['num1', 'num4', 'num3']
Time to live :
-
ttl(key)
- returns the time to live of the key as adatetime.timedelta()
object or returnsNone
if it does not have an expiration value. ReturnsFalse
if the key is missing. -
ttls(key)
- returns the time to live of the key inseconds
. ReturnsFalse
if the key is missing. -
persist(key)
- sets the expiry value of the key toNone
, hence persisting it. ReturnsFalse
if the key is missing.
Go back to the table of contents
Serialization and Storage:
Elara supports basic python datatypes (int
, str
, dict
, list
etc.).
However, objects (simple and complex) can be stored and retrieved using get
, set
and other functions that apply to them as long as they are in-memory
and not persisted in the file
, as that would lead to serialization errors.
import elara
cache = elara.exe("new.db") # commit = False by default
class MyObj():
def __init__(self, num):
self.num = num
obj = MyObj(19)
cache.set("obj", obj)
print(cache.get("obj").num)
# 19
-
To persist a simple object as a dictionary, use the
__dict__
attribute. -
Elara uses checksums and a file version flag to verify database file integrity.
All database writes are atomic (uses the safer library). Database writes are done in a separate thread along with a thread lock.
Go back to the table of contents
API reference
Strings :
-
mget(keys)
- takes a list of keys as an argument and returns a list with all the corresponding values IF they exist; returns an empty list otherwise. -
mset(dict)
- takes a dictionary of key-value pairs as an argument and calls theset(key, value)
method for each pair. Keys have to be a String. -
setnx(key, value)
- Sets the key-value if the key does not exist and returnsTrue
; returnsFalse
otherwise. -
msetnx(dict)
- takes a dictionary of key-value pairs as an argument and calls thesetnx(key, value)
method for each pair. Keys have to be a string. -
slen(key)
- returns the length of the string value if the key exists; returns-1
otherwise. -
append(key, data)
- Append the data (String) to an existing string value; returnsFalse
if it fails.
Go back to the table of contents
Lists :
-
lnew(key)
- Initialises an empty list for the given key and returnsTrue
or an Exception; key can be any data type that is supported by python dictionaries (int, float, string etc.). -
lpush(key, value)
- Appends the given value to the list and returnsTrue
; returnsFalse
if the key does not exist. -
lpop(key)
- Pops and returns the last element of the list if it exists; returnsFalse
otherwise. Index of the element can be passed to delete a specific element usinglpop(key, pos)
.pos
defaults to-1
(last element of the list). -
lrem(key, value)
- remove a value from the list. ReturnsTrue
on success andFalse
otherwise. -
llen(key)
- returns length of the list if the key exists; returns-1
otherwise. -
lindex(key, index)
- takes the index as an argument and returns the value if the key and list exist; returnsFalse
otherwise. -
lrange(key, start, end)
- takesstart
andend
index as arguments and returns the list within the given range. Value atend
not included. Returns empty list if start/end are invalid. -
lextend(key, new_list)
- Extend the list withnew_list
if the key exists. ReturnsTrue
orFalse
if the key does not exist. -
lexists(key, value)
- returnsTrue
if the value is present in the list; returnsFalse
otherwise. -
lappend(key, pos, value)
- appendsvalue
to the existing data at indexpos
using the+
operator. ReturnsTrue
orFalse
.
import elara
db = elara.exe(path='new.db', commitdb=True)
db.lnew('newlist')
db.lpush('newlist', 3)
db.lpush('newlist', 4)
db.lpush('newlist', 5)
print(db.lpop('newlist'))
# 5
print(db.lindex('newlist', 0))
# 3
new_list = [6, 7, 8, 9]
db.lextend('newlist', new_list)
print(db.get('newlist'))
# [3, 4, 6, 7, 8, 9]
Go back to the table of contents
Hashtable/Dictionary :
-
hnew(key)
- Initialises an empty dictionary for the given key and returnsTrue
or an Exception; key can be any data type that is supported by python dictionaries (int, float, string etc.). -
hadd(key, dict_key, value)
- Assigns a value to a dictionary key and returnsTrue
; returnsFalse
if the dictionary doesn't exist. -
haddt(key, tuple)
- Add a new key-value tuple into the dictionary. ReturnsTrue
if the dictionary exists; returnsFalse
otherwise. -
hget(key, dict_key)
- Returns the value from the dictionary; returnsFalse
if the dictionary doesn't exist. -
hpop(key, dict_key)
- Deletes the given key-value pair from the dictionary and returns the deleted value; returnsFalse
if the dictionary doesn't exist. -
hkeys(key)
- returns all the Keys present in the dictionary. -
hvals(key)
- returns all the values present in the dictionary. -
hmerge(key, dict)
- updates (dict.update()) the dictionary pointed by the key with the new dictionarydict
passed as an argument.
Go back to the table of contents
Key operations :
-
ttl(key)
- returns the time to live of the key as adatetime.timedelta()
object or returnsNone
if it does not have an expiration value. ReturnsFalse
if the key is missing. -
ttls(key)
- returns the time to live of the key inseconds
or returnsNone
if it does not have an expiration value. ReturnsFalse
if the key is missing. -
persist(key)
- sets the expiry value of the key toNone
, hence persisting it. ReturnsFalse
if the key is missing. -
randomkey()
- returns a random key from the DB, else returnsNone
.
Go back to the table of contents
Update key and Secure DB :
updatekey(key_path)
- This method works for instances produced byexe_secure()
. It updates the key in the key file path and re-encyrpts the database with the new key. If the file doesn't exist, the method generates a new file with a key and uses that to encrypt the database file.
import elara
# exe_secure() encrypts the db file
db = elara.exe_secure("new.db", True, "newdb.key")
db.set("name", "Elara")
print(db.get("name"))
# Elara
db.updatekey('newkeypath.key')
# Regular program flow doesn't get affected by key update
print(db.get("name"))
# Elara
However, the next time you run the program, you have to pass the new updated key (newkeypath.key
in this case) to avoid errors.
securedb(key_path)
- Callsupdatekey(key_path)
for instances which are already protected with a key. For an unprotected instance ofexe()
, it generates a new key in the given key_path and encrypts the database file. This db file can henceforth only be used with theexe_secure()
function.
Go back to the table of contents
Export data :
-
exportdb(export_path, sort=True)
- Copies the entire content of the database file into the specified export file path usingjson.dump()
. To prevent sorting of Keys, useexportdb(export_path, False)
-
exportmem(export_path, sort=True)
- Copies the current database contents stored in-memory into the specified export file path usingjson.dump()
. To prevent sorting of Keys, useexportmem(export_path, False)
. -
exportkeys(export_path, keys = [], sort=True)
- Takes a list of keys as an argument and exports those specific keys from the in-memory data to the given export file path.
import elara
db = elara.exe('new.db', False)
db.set("one", 100)
db.set("two", 200)
db.commit()
db.set("three", 300)
db.exportdb('exportdb.txt')
db.exportmem('exportmem.txt')
db.exportkeys('exportkeys.txt', keys = ['one', 'three'])
"""
# exportdb.txt
{
"one": 100,
"two": 200
}
# exportmem.txt
{
"one": 100,
"three": 300,
"two": 200
}
# exportkeys.txt
{
"one": 100,
"three": 300
}
"""
Go back to the table of contents
Tests
Run this command inside the base directory to execute all tests inside the test
folder:
$ python -m unittest -v
Go back to the table of contents
Dependencies
cryptography
msgpack
safer
Go back to the table of contents
Release notes
- Latest -
v0.5.x
v0.5.4
- No breaking changesv0.5.3
v0.5.2
v0.5.1
v0.5.0
v0.5.x
comes with an internal re-architecture that allows for much better caching and granular control on item expiry. No breaking changes from v0.4.x
.
v0.4.x
moves away from the json-based (dump
, load
) storage approach used in earlier versions, instead storing it as bytes and has support for checksums and database file version flags for added security.
v0.3.x
uses utf-8
encoding while storing data.
v0.2.x
and earlier used a mix of ascii
and base64
encoding.
To safeguard data, its better to export all existing data from any existing database file before upgrading Elara. (use exportdb(export_path)
).
View Elara's detailed release history.
Go back to the table of contents
Contributors :
Original author and maintainer - Saurabh Pujari
Logo design - Jonah Eapen
Open source contributors :
Go back to the table of contents