Skip to content
/ braq Public

Customizable data format for config files, AI prompts, and more

License

Notifications You must be signed in to change notification settings

pyrustic/braq

Repository files navigation

License: MIT PyPI package version Downloads

Cover image

Braq document with 3 sections

Braq

Customizable data format for config files, AI prompts, and more

Table of contents

Overview

Braq (pronounced /ˈbɹæk/) is a human-readable customizable data format whose reference parser is an eponymous lightweight Python library available on PyPI.

Minimal format specification

A Braq document is made up of sections, each defined by a header surrounded by square brackets (hence the name Braq) and a body which is just lines of text.

Since a body is arbitrary text, it is possible to embed a complex dictionary data structure into a section by encoding it with the human-readable Paradict data format.

Its versatility and minimal format specification make Braq highly customizable and therefore allow it to be used in an eclectic set of use cases such as config files, AI prompts, and code documentation.

Intuitive programming interface

The Braq parser offers an intuitive programming interface to smoothly interact with a Braq document as well as a transparent integration with Paradict for embedding and loading complex dictionary data structures.

Braq provides functions for parsing documents line by line, creating document models, rendering documents, and performing file I/O operations, among other functionalities.

At a higher level, the Document class leverages the base functions to model documents and allow seamless interaction with them. This class also serves as the parent class for FileDoc that focuses specifically on documents with associated disk files.

Back to top

Data format specification

Here are the specs and recommended practices for a valid and neat Braq document:

  • A Braq document, when not empty, can be divided into sections.

  • A section is made up of two parts: a header and a body.

  • The header is defined on its own line, surrounded by two single square brackets (opening and closing brackets respectively).

  • The body is what is between two consecutive headers or between a header and the end of the document.

  • A section can be defined multiple times in the same document. In this case, the parser will concatenate its bodies from top to bottom.

  • A section with an empty header is called an unnamed section.

  • It is recommended to define no more than a single occurrence of a section in a document.

  • When a document contains a single unnamed section, it is recommended to place this section at the top of the document.

  • When an unnamed document starts a document, it is recommended to omit its header.

  • A dictionary data structure encoded with the human-readable Paradict data format can be safely embedded into a section. This section should then be referenced as a dict section.

  • It is recommended to put 1 empty line as spacing between two consecutive sections.

Example:

This is the unnamed section
that starts this document.

[section 1]
Lorem ipsum dolor sit amet, 
consectetur adipiscing elit.

[section 2]
# dictionary data structure encoded with Paradict
id = 42
user = "alex" 
books = (dict)
    sci-fi = (list)
        "book 1"
        "book 2"
    thriller = (set)
        "book 3"

[section 1]
it is not recommended to multiply the occurrences
of a section, however the parser will append this 
occurrence of 'section 1' to the previous one.

Back to top

Notable use cases

This section outlines three notable use cases for Braq, namely config files, AI prompts, and code documentation.

Config files

Being able to embed a dictionary data structure in a section makes Braq de facto suitable for config files.

Example of a Braq config file:

# This is the unnamed section of 'my-config.braq' file.
# This section will serve as HELP text.

[user]
id = 42
name = 'alex'

[gui]
theme = 'dark'
window-size = '1024x420'

Using functions to consume and create config files:

from braq import load_config, dump_config

config = load_config("my-config.braq")

# get the 'user' dict section
user = config["user"]

# test
assert user == {"id": 42, "name": "alex"}

# embed a 'server' dict section in the config file then dump the config 
config["server"] = {"ip-address": "127.0.0.1", "port": 80}
dump_config(config, "my-config.braq")  # persisted

Note that load_config and dump_config should be used only when the target filename is a config file that contains exclusively dict sections. For other files, use the FileDoc class

Using the FileDoc class to consume the config file:

from braq import FileDoc

confile = FileDoc("my-config.braq")

# build the 'user' dict section
user = confile.build("user")

# test
assert user == {"id": 42, "name": "alex"}

# retrieve the unnamed section
text = confile.get("")  # notice the empty header str

# retrieve the 'user' dict section as a text
text = confile.get("user")

# embed a 'server' dict section in the config file
server_conf = {"ip-address": "127.0.0.1", "port": 80}
confile.embed("server", server_conf)  # change persisted

A schema can be passed to a FileDoc instance to validate dict sections.

AI prompts

The capability to seamlessly interweave human-readable structured data with prose within a single document is a fundamental capability that a language designed to interact with AI must possess.

Additionally, the fact that Braq natively supports indentation removes the need for input sanitization, thereby eliminating an entire class of injection attacks.

Specs

Following are specs for building structured AI prompts with Braq:

  • A prompt document must start with the root instructions defined inside the top unnamed section.
  • The next section that the AI should actively care about, after the top unnamed section, should be explicitly referenced in the root instructions.
  • User input must be programmatically embedded as a text value of a dictionary key inside a section that is not the top unnamed section.

That's it ! The specification is deliberately short to avoid unnecessary complexity and also to leave room for creativity.

Example

You are an AI assistant, your name is Jarvis.

You will access the websites defined in the WEB section
to answer the question that will be submitted to you.
The question is stored in the 'input' key of the USER 
dict section.

Be kind and consider the conversation history stored
in the 'data' key of the HISTORY dict section.

[USER]
timestamp = 2024-12-25T16:20:59Z
input = (raw)
    Today, I want you to teach me prompt engineering.
    Please be concise.
    ---

[WEB]
https://github.com
https://www.xanadu.net
https://www.wikipedia.org
https://news.ycombinator.com

[HISTORY]
0 = (dict)
    timestamp = 2024-12-20T13:10:51Z
    input = (raw)
        What is the name of the planet
        closest to the sun ?
        ---
    output = (raw)
        Mercury is the planet closest
        to the sun !
        ---
1 = (dict)
    timestamp = 2024-12-22T14:15:54Z
    input = (raw)
        What is the largest planet in
        the solar system?
        ---
    output = (raw)
        Jupiter is the largest planet
        in the solar system !
        ---

Source code documentation

The flexibility of Braq gives the possibility to define custom data formats for specific use cases. Source code documentation is one of those use cases that need Braq with a custom format on top of it.

This is how Braq can be used to document a function:

def add(a, b):
    """
    This function adds together the values of 
    the provided arguments.
    
    [param]
    - a: first integer
    - b: second integer
    
    [return]
    Returns the sum of `a` and `b`
    """
    return a + b

MikeDoc is a docstring format for generating API references. The library uses Braq to parse docstrings.

Back to top

Classes for interacting with a document

The library exposes the Document and FileDoc classes for interacting with documents. In contrary to the Document class, FileDoc focuses specifically on documents with associated disk files such as config files.

Base document class

The Document class creates an editable model of a Braq document and also offers to validate it with a schema.

Usage example:

from braq import Document

INIT_TEXT = """
This document contains
configuration data

[user]
id = 42
name = 'alex'
"""

SCHEMA = {"user": {"id": "int", "name": "str"}}

document = Document(INIT_TEXT, schema=SCHEMA)

# get the body of the unnamed section as a text
text = document.get("")  # empty header string

# build the 'user' dict section
user = document.build("user")
# test
assert user == {"id": 42, "name": "alex"}

# set a section (here, we are editing the unnamed section)
document.set("", "line 1\nline 2")
assert document.get("") == "line 1\nline 2"

# embed a 'server' dict section
server_conf = {"ip-address": "127.0.0.1", "port": 80}
document.embed("server", server_conf)

# list headers
assert document.list_headers() == ("", "user", "server")

# validate specific dict sections
# (no args implies that the entire doc will be the target)
document.is_valid("user", "server")  # returns a bool
# beware, the 'validate' method may raise an exception
# for good reasons !

There is more to discover about the Document class, such as the clear, remove, and render methods, exposed properties, and more.

Check out the API reference for braq.Document here.

File document class

The FileDoc class is based on the Document class and focuses specifically on documents with associated disk files such as config files.

As with the Document class, a schema can be passed to a FileDoc instance to validate dict sections.

from braq import FileDoc

confile = FileDoc("config-file.braq")

# build the 'user' section
user = confile.build("user")

# test
assert user == {"id": 42, "name": "alex"}

# retrieve the unnamed section as a text
text = confile.get("")  # notice the empty header str

# retrieve the 'user' dict section as a text
text = confile.get("user")

# embed a 'server' dict section
server_conf = {"ip-address": "127.0.0.1", "port": 80}
confile.embed("server", server_conf)  # change persisted

There is more to discover about the FileDoc class, such as the load, save, and save_to methods, exposed properties, and more.

Check out the API reference for braq.FileDoc here.

Back to top

Section class

The Section class is an abstraction representing a Braq section. It exposes the header and body properties and renders itself when its __str__ method is called implicitly.

import braq

# create a Section object
header, body = "my header", ("line a", "line b")
section = braq.Section(header, body)

# test the properties
assert section.header == "my header"
assert section.body == "line a\nline b"

# test the rendering
assert str(section) == """\
[my header]
line a
line b"""

Back to top

Base functions

Base classes such as Document and FileDoc use several public functions under the hood that can be directly called by the programmer at the right time. These basic functions allow you to parse and render documents as well as read and write file documents.

Parse a document

The library exposes the parse function which takes as input the text stream to be parsed, then returns a dictionary whose keys and values are strings representing headers and bodies respectively.

Sections sharing the same header are concatenated ! The header of an unnamed section is an empty string.

import braq

text = """\
this is the unnamed section at
the top of this document...

[section 1]
this is section 1"""

d = braq.parse(text)

# check headers
assert tuple(d.keys()) == ("", "section 1")
# check the body of 'section 1'
assert d["section 1"] == "this is section 1"

Check out the API reference for braq.parse here.

Parse a document iteratively

A document can be parsed line by line as following:

import braq

text = """\
this is the unnamed section

[section 1]
this is section 1"""

for header, body in braq.parse_iter(text):
    if header:
        print("[" + header + "]")
    for line in body:
        print(line)

Output:

this is the unnamed section

[section 1]
this is section 1

Check out the API reference for braq.parse_iter here.

Read a file

The library exposes the read function which takes as input the path to a file to parse, then returns a dictionary whose keys and values are strings representing headers and bodies respectively.

Sections sharing the same header are concatenated !

import braq

path = "/home/alex/braqfile.txt"

r = braq.read(path)
assert tuple(r.keys()) == ("", "section 1")

Check out the API reference for braq.read here.

Read a file iteratively

A large text file can be parsed line by line as following:

import braq

path = "/home/alex/braqfile.txt"

for header, body in braq.read_iter(path):
    if header:
        print("[" + header + "]")
    for line in body:
        print(line)

Output:

this is the unnamed section

[section 1]
this is section 1

Check out the API reference for braq.read_iter here.

Render a document

Rendering a document involves transforming Python objects representing sections into Braq text which is a string that can be displayed on the screen or stored in a file.

The library exposes the render function which accepts as input a sequence of sections (either header-body tuples or Section objects) and returns a Braq document.

import braq

# sections
section_1 = braq.Section("section 1", "line a\nline b")
section_2 = "section 2", "line c\nline d"
section_3 = "section 3", ("line e", "line f")

# rendering
r = braq.render(section_1, section_2, section_3)

print(r)

Output:

[section 1]
line a
line b

[section 2]
line c
line d

[section 3]
line e
line f

The render function also accepts the spacing argument which defaults to 1 and represents the number of lines of spacing between two adjacent sections.

Check out the API reference for braq.render here.

Write to file

Following is a snippet for writting a Braq document to a file:

import braq

# sections
section_1 = braq.Section("", "welcome")
section_2 = braq.Section("section 2")
section_3 = "section 3", ("line a", "line b")

# path to file
path = "/home/alex/braqfile.txt"
# write to file
sections = (section_1, section_2, section_3)
r = braq.write(sections, path)

The contents of the Braq file:

welcome

[section 2]

[section 3]
line a
line b

Check out the API reference for braq.write here.

Back to top

Braq schema for data validation

Dict sections can be validated against a Braq schema. A Braq schema is a Python dictionary object that can be passed to a Document or a FileDoc. The keys of this dictionary are the headers of dict sections to validate and the values are Paradict schemas.

A Paradict schema is a dictionary containing specs for data validation.

A spec is either simply a string that represents an expected data type, or a Spec object that can contain a checking function for complex validation.

Supported spec strings are: dict, list, set, obj, bin, bin, bool, complex, date, datetime, float, grid, int, str, time

Example:

from paradict.validator import Spec
from braq import Document

# Braq text with 2 dict sections
TEXT = """
[user]
id = 42
name = 'alex'

[server]
ip-address = "127.0.0.1"
port = 80
"""

# Associated schema
SCHEMA = {"user": {"id": "int",
                   "name": "str"},
          "server": {"ip-address": "str",
                     "port": Spec("int", lambda x: 0 < x < 65535)}}

doc = Document(TEXT, schema=SCHEMA)
assert doc.is_valid()
# beware, the validate function returns a bool
# but it can also raises an exception when something is wrong 

Back to top

Misc functions

The check_header function accepts a line of text as input and then returns a boolean to indicate whether this line is a header or not.

import braq

line_1 = "[my header]"
line_2 = "[this isn't a header] at all"
assert braq.check_header(line_1) is True
assert braq.check_header(line_2) is False

The get_header function accepts a line of text as input and returns a string if the line is a header. Otherwise, None is returned.

import braq

line_1 = "[my header]"
line_2 = "[this isn't a header] at all"
assert braq.get_header(line_1) == "my header"
assert braq.get_header(line_2) is None

Check out the API reference for braq.check_header and braq.get_header here.

Back to top

Miscellaneous

Collection of miscellaneous notes.

Cover image

The beautiful cover image is generated with Carbon.

Back to top

Testing and contributing

Feel free to open an issue to report a bug, suggest some changes, show some useful code snippets, or discuss anything related to this project. You can also directly email me.

Setup your development environment

Following are instructions to setup your development environment

# create and activate a virtual environment
python -m venv venv
source venv/bin/activate

# clone the project then change into its directory
git clone https://github.com/pyrustic/braq.git
cd braq

# install the package locally (editable mode)
pip install -e .

# run tests
python -m unittest discover -f -s tests -t .

# deactivate the virtual environment
deactivate

Back to top

Installation

Braq is cross-platform. It is built on Ubuntu and should work on Python 3.5 or newer.

Create and activate a virtual environment

python -m venv venv
source venv/bin/activate

Install for the first time

pip install braq

Upgrade the package

pip install braq --upgrade --upgrade-strategy eager

Deactivate the virtual environment

deactivate

Back to top

About the author

Hello world, I'm Alex, a tech enthusiast ! Feel free to get in touch with me !




Back to top

About

Customizable data format for config files, AI prompts, and more

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages