Skip to content

Latest commit

 

History

History
81 lines (66 loc) · 3.08 KB

README.md

File metadata and controls

81 lines (66 loc) · 3.08 KB
Raw

nosqlite.py

nosqlite.py is a pure python library for Python 3 that aims to provide a schemaless wrapper for interacting with sqlite databases. Much of the behavior follows how the API for pymongo works, so those familiar with that library should have a similar experience. Example::

import nosqlite

with nosqlite.Connection(':memory:') as conn:
    foo_collection = conn['foo_collection']
    foo_collection.insert({'foo': 'bar', 'baz': 'qux'})
    foo_collection.find({'foo': 'bar'})

Indexes

  • Index creation is supported for single key, compound key, and even nested key.

    foo_collection.insert({'a':1, 'b':2, 'c':10})
foo_collection.insert({'a':1, 'b':3, 'c':20})
foo_collection.insert({'x':{'y':100}, 'z':200})
foo_collection.create_index('a')        # single key
foo_collection.create_index(['a','b'])  # compound key
foo_collection.create_index('x.y')      # nested key

  • Index is actually being used in find operation. e.g. foo_collection.find({'a':1, 'b':3}) the library will put them in where clause as a subquery to search _id on index with a=1 and b=3 which result is a subset of _id on the collection, so searching on the collection will use its primary key instead of a whole table scan.

    • however, the other operations like $gt or $lt still use table scan

  • Unique index is supported and force contraint on insert, save (update).

    • You cannot create a unique index on collection with non-unique keys.
    • You cannot update a document to break uniqueness on a collection with unique index.
    • Any attempt above will raise an IntegrityError and the document will not be inserted or updated.

  • Reindex with only a subset of documents is supported.

  • Find with hint index is supported e.g. you don't have index on ('a', 'c') but already have index on 'a'.

    foo_collection.find({'a':1, 'c':20})  # use table scan
foo_collection.find({'a':1, 'c':20},
    hint='[foo_collection{a}]'        # use index to narrow down rows with a=1
)

Sorting

  • Result documents from find() can be sorted by any key and also nested key 
    foo_collection.find({'a':1}, sort={'c':nosqlite.DESCENDING})
foo_collection.find({'a':1}, sort={'b':nosqlite.ASCENDING,
                                   'c':nosqlite.DESCENDING})
foo_collection.find(sort={'x.y':nosqlite.DESCENDING})

TODOs

  • Support using index on other operations.
  • Support for embedded documents and queries on those (i.e. {'foo.bar': 5})

Contribution and License

This project was originally developed by Shaun Duncan <shaun[dot]duncan[at]gmail[dot]com> and is now maintained by Chaiwat Suttipongsakul <cwt[at]bashell[dot]com>. It is licensed under the MIT license, which means you are free to use, modify, and distribute the code for your own purposes.

Contributions to the project are highly encouraged and greatly appreciated. If you find a bug, have an enhancement in mind, or want to suggest a new feature, please feel free to contribute.