Parser for Hrbrt: Human-Readable BRanching Text format
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Hrbrt (Human-Readable BRanching Text) is a file format which expresses a directed graph of text nodes linked by choices, similar to a flow chart. Potential uses include decision trees, questionnaires, and videogame dialogue scripts.

Inspired by Markdown's philosophy, Hrbrt is machine-parsable yet perfectly human-readable. The lack of markup and other syntactic clutter in Hrbrt's syntax makes it easily understandable to non-technical users and convenient to author using any basic text editor.

hrbrt is a command line program (or python module) for parsing files in Hrbrt format. It can convert Hrbrt to other formats or "run" a file using command line or GUI-based prompts.

hrbrt Command


hrbrt requires Python 3.6 or above, and optionally the Tkinter python module to use the GUI option. Tkinter is part of Python's standard library and is likely to come installed with it, but some distributions do package it separately.


The easiest way to install hrbrt is via pip. Pip can download and install hrbrt directly from the Github project as follows:

pip install git+

Alternatively you can obtain the source code yourself from Github and run the Python script:





python -m hrbrt [OPTIONS] FILE

Positional Arguments


The Hrbrt file to read. Use - to read from standard input.


File to write output to. Use - to write to standard output. If input file is specified, output file defaults to <infile>.out.<ext> using the name of the input file and the appropriate file extension for the output format. If no input file is specified, defaults to standard output.


-h, --help : Show usage information and exit

-o FILE, --output FILE : Output the result to FILE. Use - to write to standard output. If --tofmt isn't also specified, the output format is inferred from the file extension, or Hrbrt by default.

-t FORMAT, --tofmt FORMAT : Output the result using the given format. One of hrbrt, json, xml or markdown. If --output isn't also specified, output is written to <FILE>.out.<EXT> where FILE is the input filename and EXT is the format's file extension or, if reading from standard input, output is written to standard output.

-r MODE, --run MODE : Run the document interactively in the specified mode. cli for command line mode, or gui for a basic graphical wizard mode. The gui option requires the TKinter python module.


Validate the file foobar.hb and report errors

$ hrbrt foobar.hb
Section "first" has no choice blocks and so cannot reach end of document

Convert questionnaire.hb to Markdown format

$ hrbrt -o questionnaire.hb

Convert dialogue.hb to XML and write to standard out

$ hrbrt -t xml -o - dialogue.hb
<?xml version="1.0" ?>
        <text>Nice hat!</text>

Read Hrbrt data from standard input, run using a GUI and output to foo.js in JSON format

$ hrbrt -r gui -o foo.js -
:: [] Yes
:  [] No

As a Python Package

The hrbrt Python package can be imported and used to parse Hrbrt in other Python code. The class reads Hrbrt data from a text stream and produces a hrbrt.parse.Document object representing the parse tree.

See the usage example below:

>>> from io import StringIO
>>> from import HrbrtIO
>>> # A minimal Hrbrt document
... data = ":: Test\n"
>>> # Create a simple string stream.
... # Could read from a file stream instead
... stream = StringIO(data)
>>> # Parse the data stream
... document =
>>> # Read data from the resulting parse tree
... print(document.sections[0].items[0].text)

Hrbrt Syntax

How to Read Hrbrt

A document in Hrbrt format can be parsed by the recipient's machine and presented to them interactively, or the recipient can edit and return the raw text document.

In either case, the recipient begins reading at the start of the document. When they encounter a set of choices, they mark their selection and read the response text beside it. If the response includes a GO TO statement, the reader jumps to that section and continues from there, otherwise the reader continues reading as normal. The reader continues to follow the flow of the document until they reach the end.

A recipient reading the raw Hrbrt text may mark their selection at each set of options by inserting character data in the corresponding box (typically an 'X' or '#'). They may also add feedback to the document by adding new lines of text or writing in the existing blank lines.


Below is an example of Hrbrt syntax:

:: Hi there. 

%% Please fill in my questionnaire!

:: What would you say
:  is your favourite animal?

:: [ ] Cat       -- GO TO cats
:  [ ] Dog       -- GO TO other
:  [ ] Turkey    -- GO TO other

=== Cats ===

:: What is your favourite breed of cat?

:: [ ] Burmese
:  [ ] Siamese
:  [ ] Persian
:  [ ] Other

:: What do you like most about cats?

:: [ ] Their ears    -- GO TO end
:  [ ] Their noses   -- GO TO end
:  [ ] Their paws    -- GO TO end
:  [ ] Their fur     -- GO TO end

=== Other ===

:: Are you sure you don't like 
:  CATS more?

:: [ ] Yes    -- GO TO end
:  [ ] No     -- Your finger slipped. I see.
:                GO TO cats

=== End ===

:: Thanks for your input!


Sections serve to split up the document and provide reference points to which the user can be directed as they traverse the document. (See explanation of "go-to" statements in the chapter "Choice Responses").

Each section contains one or more blocks of content. These may be any combination of text, instruction or choice blocks (see subsequent chapters for details).

The first section of the document is where the user begins reading, and has no heading. Each subsequent section is indicated by a heading with two or more equals signs on each side. The section heading defines the section's name, which is case-insensitive. Section names must only contain letters, numbers, underscores, hyphens or spaces. All content below this heading, down to the next heading, is contained within the section.


=== My Section ===

Text and Recipient's Feedback

Blocks of regular text can be added to the document by starting the first line of the block with a double colon, and each subsequent line with a single colon. A Hrbrt viewing tool will typically present separate text blocks to the user one at a time.

Lines without the preceding colons are assumed to be part of the recipient's feedback. It is assumed that if the recipient leaves feedback in the document, they will omit the colon from the start.


:: This is part of the 
:  document text that
:  the sender wrote

But this is the 
recipient's feedback

When the document is parsed, recipient feedback is included in the output. Feedback within a choice block (see "Choices" chapter) is grouped together and attached to the block, and other feedback is grouped together and attached to the section.


Instruction blocks are not presented to the user when parsed by a program such as a visual Hrbrt reader. They are intended for providing instructions to users reading the raw file text only. The first line of an instruction block starts with a double percent sign, and each subsequent line with a single percent sign.


%% Please fill in this document 
%  and send it back as soon as 
%  possible


Blocks of choices are used to present the user with options from which they can make a single selection. Users Reading the raw Hrbrt text indicate their selection by writing something inside the box beside it. This is typically an X or # character, but may contain any single-line content other than ].

Each choice of the block goes on its own line. The first choice starts with a double colon, and each subsequent choice starts with a single colon. Each option then consists of a pair of square brackets [] followed by the option description. The description may run onto multiple lines, each starting with a colon.


:: [] Animal
:  [] Some kind
:      of Mineral
:  [] Vegetable

A choice block may not immediately follow another choice block. They must be separated, using a text block for example.

Choice Responses

Each choice may optionally be followed by response text. This is separated from the choice description by a pair of hyphens --. The response text gives feedback and further instructions to the recipient on selection of that option.

A choice response may optionally be followed by a go-to statement. This consists of the words GO TO in uppercase, followed by a section name, and optional trailing punctuation. The section names are case-insensitive. The go-to statement instructs the recipient or viewing tool as to which section to jump to next.

The response and go-to statement may flow onto multiple lines, each starting with a colon.


:: [] Animal     -- Good choice! GO TO my section
:  [] Mineral    -- OK. GO TO some section.
:  [] Vegetable  -- Not bad, but I 
:                   think you could
:                   have chosen better.
:                   GO TO end


All lines in the document may optionally be prefixed with > markers, as would typically be added by an email client. The > markers themselves are ignored, but the following line content is still parsed. This allows recipients to reply directly to a document sent by email and it still be parsable.

Section Flow Rules

A valid Hrbrt document must allow the user to reach the end of the document's final section. Dead ends and infinite loops are not allowed.

Before reaching the end of a section (other than the final section) the user must be explicitly directed to a different section by a go-to statement. In other words, a document is not valid if the user can "fall through" to the end of a section.

For example, the following is not allowed:

:: [] Option A -- GO TO my section
:  [] Option B

== My Section ==

A choice block may not immediately follow another choice block (for readability reasons). They must be separated by another block, such as a text block.

Hrbrt Formal Definition

See syntax.bnf

Credits and Licence

The hrbrt tool is released under the MIT licence. For the full text of this licence, see LICENCE.txt