Skip to main content

A table dataformat based on json

Project description

Build and Test Status

This package is tested using pytest on travis-ci. The current build-status is:

The code is reviewed automatically on codacy:

What is ljson?

ljson is an attempt to create a database model suiting the needs of modern data processing. It is designed to work faster than usual json, but to keep the simple but yet elegant object representation.

ljson can be used instead of pure json to increase the performance when accessing a large set of data.

Why ljson?

There are a lot data storage formats out there: XML, JSON, CSV, SQL, NOSQL, binary packed, GNU-DB,…

Some of them are designed to store complete databases (SQL, NOSQL, …) and some are designed to store tables. And of course there are JSON and XML. They can be used to store more complex objects, are human-readable and data is stored in just one file.

But they suffer from one problem: If one wants to alter the data in the file he has to read the complete file and store all the data in his RAM. This is slow, maybe impossible (Big Data) and insecure. If the process cannot complete the operation properly this might corrupt all data.

ljson tries to bypass this by using a mixture of CSV (line based) and JSON (object based):

Every line is one object. If one wants to add another object he just opens the file in append mode and adds one line. If one line is corrupted the rest of the file is still valid.

Operating on large sets of objects is also possible by reading the file line by line.

Especially asynchronous operations can be performed easily, as the main part of the file stays untouched (unless you alter objects. Then the file has to be re-written).


ljson is designed to be stored in files, the definition of a ljson file is:

<ljson_file> = [<header>\n]<ljson_content>
<ljson_content> = <json_object>{\n<json_object>}

<json_object> can be any json object, as described on


If you use ljson you are restricted to the following python data types (and their ljson types):

  • int: "int"
  • str: "str"
  • bool: "bool"
  • float: "float"
  • bytes: "bytes"
  • dict: "json"
  • list: "json"

Because it is possible to convert all data types to one of these it is possible to store any kind of data.


Without a Python Module

ljson is designed to work without any third party python modules. One can read ljson data with the python built-in json module:

>>> import json
>>> ljson = '{"id": 1, "name": "foo"}\n{"id": 2, "name": "bar"}'
>>> for line in ljson.split("\n"):
...     print(json.loads(line))
{'name': 'foo', 'id': 1}
{'name': 'bar', 'id': 2}

And this should always be the preferred way to access ljson data, if all data is required.

If one wants to access specific fields it is better to use the ljson python module:

With the ljson Module

Using the ljson Module is simple and efficient if one wants to access just some fields, not the complete file.

There are two base implementations: ljson.base.mem that loads the file content into the RAM. This is way faster and supports files without a header and one is able to construct the Table without a file.

The second implementation is ljson.base.disk. This implementation does not load any data into RAM. If you are accessing huge sets you should use this implementation.

Creating a table is simple (at least for the memory tables):

>>> import ljson
>>> header = ljson.Header({"id": {"type": "int", "modifiers":["unique"]}, "name": {"type": "str", "modifiers": []}})
>>> table = ljson.Table(header, [{"id": 1, "name": "foo"}, {"id": 2, "name": "bar"}, {"id": 3, "name": "bar"}])

One can access items using python’s built-in __getitem__ and __setitem__:

>>> table[{"id": 1}]["name"]
>>> list(table[{"id": 1}])
[{'name': 'foo', 'id': 1}]

The table “index” must be a dict. This allows to access non-unique rows, like this:

>>> list(table[{"name":"bar"}])
[{'id': 2, 'name': 'bar'}, {'id': 3, 'name': 'bar'}]

Using ljson to store data

Using ljson to store data is pretty simple:

>>> from io import StringIO
>>> fout = StringIO()
'{"name": {"type": "str", "modifiers": []}, "__type__": "header", "id": {"type": "int", "modifiers": ["unique"]}}\n{"name": "foo", "id": 1}\n{"name": "bar", "id": 2}\n{"name": "bar", "id": 3}'
>>> table2 = ljson.Table.from_file(fout)
>>> list(table2)
[{'id': 1, 'name': 'foo'}, {'id': 2, 'name': 'bar'}, {'id': 3, 'name': 'bar'}]

Reading and writing csv files is pretty simple, too:

>>> from ljson.convert.csv import table2csv, csv2table
>>> fout = StringIO()
>>> table2csv(table, fout)
>>> table2 = csv2table(fout, types = {"id": "int", "name":"str"})
>>> list(table2)
[{'id': 1, 'name': 'foo'}, {'id': 2, 'name': 'bar'}, {'id': 3, 'name': 'bar'}]


  • store bytes as b64
  • fix the sql bytes representation

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for ljson, version 0.5.4
Filename, size File type Python version Upload date Hashes
Filename, size ljson-0.5.4-py3-none-any.whl (29.8 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size ljson-0.5.4.tar.gz (21.8 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page