Travis-CI Build Status Codacy Code Quality Status Codacy Code Coverage PyPI Package latest release PyPI Wheel Updates Join the chat at

A Python module that analyzes your architecture strength based on Design Structure Matrix (DSM) data.

Archan is a Python module that analyzes the strength of your project architecture according to some criteria described in “The Protection of Information in Computer Systems”, written by Jerome H. Saltzer and Michael D. Schroeder.


  • Usable directly on the command-line.
  • Support for plugins. See for example the Provider plugin in dependenpy. You can also take a look at django-meerkat, a Django app using Archan.
  • Configurable through command-line or configuration file (YAML format).
  • Read DSM data on standard input.


Just run pip install archan.


On ReadTheDocs

Archan defines three main classes: Analyzer, Provider and Checker. A provider is an object that will produce data and return it in the form of a DSM (Design Structure Matrix). The checker is an object that will analyze this DSM according to some criteria, and return a status code saying if the criteria are verified or not. An analyzer is just a combination of providers and checkers to run a analysis test suite.


On the command-line


archan -h


usage: archan [-c FILE] [-h] [-i FILE] [-l] [--no-color] [--no-config] [-v]

Analysis of your architecture strength based on DSM data

optional arguments:
  -c FILE, --config FILE  Configuration file to use.
  -h, --help              Show this help message and exit.
  -i FILE, --input FILE   Input file containing CSV data.
  -l, --list-plugins      Show the available plugins. Default: false.
  --no-color              Do not use colors. Default: false.
  --no-config             Do not load configuration from file. Default: false.
  -v, --version           Show the current version of the program and exit.

Other examples:

# Load configuration file and run archan
# See Configuration section to know how archan finds the config file

# No configuration, read CSV data from file
archan --no-config --input FILE.CSV

# No configuration, read CSV data from stdin
dependenpy archan --format=csv | archan --no-config

# Specify configuration file to load
archan --config my_config.yml

# Output the list of available plugins in the current environment
archan --list-plugins




Archan applies the following methods to find the configuration file folder:

  1. read the contents of the file .configconfig in the current directory to get the path to the configuration directory,
  2. use config folder in the current directory if it exists,
  3. use the current directory.

It then searches for a configuration file named:

  1. archan.yml
  2. archan.yaml
  3. .archan.yml
  4. .archan.yaml

Format of the configuration file is as follow:

analyzers: [list of strings and/or dict]
- identifier: [optional string]
  name: [string]
  description: [string]
  providers: [string or list]
  - provider.Name: [as string or dict]
      provider_arguments: as key value pairs
  checkers: [string or list]
  - checker.Name: [as string or dict]
      checker_arguments: as key value pairs

It means you can write:

# a first analyzer with one provider and several checker
- name: My first analyzer
  description: Optional description
  providers: just.UseThisProvider
  - and.ThisChecker
  - and.ThisOtherChecker:
      which: has
      some: arguments
# a second analyzer with several providers and one checker
- name: My second analyzer
  - use.ThisProvider
  checkers: and.ThisChecker
# a third analyzer, using its name directly
- some.Analyzer

Every checker support an ignore argument, set to True or False (default). If set to True, the check will not make the test suit fail.

You can reuse the same providers and checkers in different analyzers, they will be instantiated as different objects and won’t interfere between each other.

As an example, see Archan’s own configuration file.

To get the list of available plugins in your current environment, run archan --list-plugins or archan -l.

Writing a plugin

Plugin discovery

You can write three types of plugins: analyzers, providers and checkers. Your plugin does not need to be in an installable package. All it needs to be summoned is to be available in your current Python path. However, if you want it to be automatically discovered by Archan, you will have to make it installable, through pip or simply python install command or equivalent.

If you decide to write a Python package for your plugin, I recommend you to name it archan-your-plugin for consistency. If you plan to make it live along other code in an already existing package, just leave the name as it is.

To make your plugin discoverable by Archan, use the archan entry point in your

from setuptools import setup

    'entry_points': {
        'archan': [
            'mypackage.MyPlugin = mypackage.mymodule:MyPlugin',

The name of the entry point should by convention be composed of the name of your package in lower case, a dot, and the name of the Python class, though you can name it whatever you want. Remember that this name will be the one used in the configuration file.

Also a good thing is to make the plugin importable thanks to its name only:

import mypackage.MyPlugin

But again, this is just a convention.

Plugin class

You can write three types of plugins: analyzers, providers and checkers. For each of them, you have to inherit from its corresponding class:

from archan import Analyzer, Provider, Checker

class MyAnalyzer(Analyzer): ...
class MyProvider(Provider): ...
class MyChecker(Checker): ...

A provider or checker plugin must have the following class attributes:

  • identifier: the identifier of the plugin. It must be the same name as in your entry points, so that displaying its help tells how to summon it.
  • name: the verbose name of the plugin.
  • description: a description to explain what it does.
  • (optional) arguments: a tuple/list of Argument instances. This one is only used to display some help for the plugin. An argument is composed of a name, a type, a description and a default value.
from archan import Provider, Argument

class MyProvider(Provider):
    identifier = 'mypackage.MyProvider'
    name = 'This is my Provider'
    description = """
    Don't hesitate to use multi-line strings as the lines will be de-indented,
    concatenated again and wrapped to match the console width.

    Blank lines will be kept though, so the above line will not be removed.

    arguments = (
        Argument('my_arg', int, 'This argument is useful.', 42),
        # don't forget the ending comma if you have just one   ^   argument

Additionally, a checker plugin should have the hint class attribute (string). The hint describe what you should do if the check fails.

For now, the analyzers plugins just have the providers and checkers class attributes.

Plugin methods

A provider must implement the get_dsm(self, **kwargs) method. This method must return an instance of DSM. A DSM is composed of a two-dimensions array, the matrix, a list of strings, the keys or names for each line/column of the matrix, and optionally the categories for each key (a list of same size).

from archan import DSM, Provider

class MyProvider(Provider):
    name = 'mypackage.MyProvider'

    def get_dsm(self, my_arg=42, **kwargs):
        # this is where you compute your stuff
        matrix_data = [...]
        entities = [...]
        categories = [...] or None
        # and return a DSM instance
        return DSM(matrix_data, entities, categories)

A checker must implement the check(self, dsm, **kwargs) method.

from archan import DSM, Checker

class MyChecker(Checker):
    name = 'mypackage.MyChecker'

    def check(self, dsm, **kwargs):
        # this is where you check your stuff
        # with, dsm.entities, dsm.categories, dsm.size (rows, columns)
        # and return True, False, or a constant from Checker: PASSED or FAILED
        # with an optional message
        return Checker.FAILED, 'too much issues in module XXX'

Logging messages

Each plugin instance has a logger attribute available. Use it to log messages with self.logger.debug, info, warning, error or critical.

Available plugins

Here is the list of plugins available in other packages.


  • dependenpy.InternalDependencies: Provide matrix data about internal dependencies in a set of packages. Install it with pip install dependenpy.


Software licensed under ISC license.


To run all the tests: tox