DB wrapper library


License
MIT
Install
pip install notanorm==3.9.2

Documentation

Overview

Simple library that makes working with databases more convenient in python.

notanorm can return objects instead of rows, protects you from injection, and has a very simply driver interface.

Decidedly not an ORM, since ORM's are typically mega libraries with often confusing semantics.

Example:

from notanorm import SqliteDb, open_db 
# from notanorm import MysqlDb 


fname = ":memory:"

# accepts all the same parameters as sqlite3.connect
db = SqliteDb(fname)

# this works too, params turn into kwargs
db = open_db("sqlite://:memory:")

db.uri      # get my uri
db.clone()  # make a copy of the db connection

# no special create support, just use a string
db.query("create table foo (bar text, iv integer primary key)")

# insert, select, update, and upsert are convenient and do the right thing
# preventing injection, normalizing across db types, etc.
db.insert("foo", bar="hi", iv=1)
db.insert("foo", bar="hi", iv=2)
ret = db.insert("foo", bar="ho")                   # sqlite autoincrement used
print(ret.lastrowid)                               # "3"

db.upsert("foo", bar="ho", iv=4)                   # upsert requires a primary key, updates or inserts
db.upsert("foo", bar="ho", _insert_only={"iv": 4}) # _insert_only values are never used to update rows

# can update with a where clause, or with an implied one
db.update("foo", iv=4, bar="up")                   # update, no where clause, inferred from primary key
ret = db.update("foo", {"bar": "up"}, bar="hop")   # update, primary key not needed
print(ret.rowcount, ret.lastrowid)                 # "1 4" <- 1 row updated, primary key was 4

# select_one (will raise if more than 1), select
db.select_one("foo", iv=1).bar                     # hi
db.select("foo", bar="hi")[0].bar                  # hi
db.select("foo", {"bar": "hi"})[0].iv              # 1

# just picks a random one, won't error if there's more than one
db.select_any_one("foo")

# use a generator, for speed
[row.iv for row in db.select_gen("foo", bar="hi")]                      # [1, 2]

# use order_by
[row.iv for row in db.select_gen("foo", bar="hi", order_by="iv desc")   # [2, 1]

db.count("foo", bar="hi")                          # 2


class Foo:
    def __init__(self, bar=None, iv=None):
        self.bar = bar
        self.iv = iv
    def iv_squared(self):
        return self.iv * self.iv

# create a class during select
db.register_class("foo", Foo)
obj = db.select_one("foo", bar="hop")

print(obj.iv_squared(), obj.bar)                   # 16 hop


# inside a transaction, updates and inserts are faster, and are atomic
with db.transaction():
    db.update("foo", iv=4, bar="up")
    db.update("foo", iv=3, bar="up")

# delete a row
db.delete("foo", iv=1)

# select greater than
db.select("foo", iv=notnaorm.Op(">", 3))

# wipe the table
db.delete_all("foo")

# select using a subquery
db.select("foo", bar=db.subq("other", ["foo_id"], 3))

# select using a join
db.select(db.join("tab1", "tab2", tab1id="tab2id"))


### cross database ddl management

# this requires: pip install sqlglot

model = notanorm.model_from_ddl("create table foo(bar integer)")

# create tables and indices from the generic model
db.create_model(model)

# capture sql statements instead of executing them
with db.capture_sql(execute=False) as cap:
   ... do some stuff

print(cap)

# print out the model of a db
print(db.model())

Database support

sqlite3 drivers come with python

To use mysql, pip install mysqlclient, or if that is not available, pip install pymysql

JsonDb driver

The "jsondb" driver is available if sqlglot is installed.

When using jsondb, be aware that there is no attempt at concurrency controls, and no attempt to monitor for on-disk changes. Likewise, joins and subqueries are not yet supported.

The caller is on the hook to refresh() and commit() or do changes in a transaction(). The driver was designed for the use case of porting existing sqlite code to use a json file instead.