podb 1.44

python object database

podb

(p)ython (o)bject (d)ata(b)ase

thread safe, file based, redis style database for python objects

reasons not to use this

• you want a database cluster
• you have multiple millions or billions of database entries and limited storage space
• you need relationships / multiple tables

reasons to use this

• pure python
• cross-platform (Tested: Windows & Linux)
• thread safe
• stores python objects
• filters, see below
• inserting 200 objects takes ~0.0003 seconds (i7-4702MQ)
  • size of db with 800 objects is ~600kB
• ~180 loc
• no extra package dependencies
• tested (Ryzen 7 2700X)
  • 8-8.3k inserts per second
  • 1000000 entry db is 672MB
    • mongodb 1 mil collection is 143MB
  • 1000000 entries in 2:04 minutes
    • mongodb inserts @ 2-2.3k per sec

notes

• not multiprocess safe (yet?)

functions

• find
• find_one
• find_after
• find_before
• find_by_uuid
• find_contains
• find_startswith
• find_endswith
• insert
• insert_many
• update
• update_many
• upsert
• upsert_many
• size
• contains
• columns
• drop
• remove / delete
• remove_by_uuid / delete_by_uuid
• delete_before

example

from podb import DB, DBEntry


class Company(DBEntry):
    def __init__(self, name: str):
        DBEntry.__init__(self)
        self.name = name

class Customer(DBEntry):
    def __init__(self, first_name: str, last_name: str, age: int,
                 height: float, companies: list[DBEntry]):
        DBEntry.__init__(self)
        self.first_name = first_name
        self.last_name = last_name
        self.age = age
        self.height = height
        self.companies = companies

db = DB("customers")

c0 = Customer("Jeff", "Bezoz", 42, 1.69,
              [Company("Whole Foods"), Company("Zappos"),
               Company("Ring"), Company("twitch")])
db.insert(c0)

c0 = db.find_one({
    "first_name": "Jeff",
    "last_name": "Bezoz"
})

c0.companies.append(Company("Audible"))

db.update(c0)

installation

pip3 install podb