Skip to content

sjednac/coffea

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Coffea

Coffea is a command line tool and Python library for analyzing static dependences in Java bytecode. Features:

  • Class processing handled entirely in Python (i.e. no JVM dependency and no class loader issues)
  • Recursive processing of directories (e.g. exploded deployments) and basic archive formats (jar, war and ear)
  • Package or class based dependency models
  • Node weight based on actual code size (i.e. bytecode size)
  • Node filters and mappers for basic noise reduction (eg. removing certain packages from the model or folding several packages into one node)
  • Basic graph visualisation using matplotlib
  • Exporting to common graph formats using standard networkx facilities (eg. dot, gml or graphml)

Usage

Run the command-line tool (use coffea -h for a complete list of options):

$ coffea -i <directory|jar|war|ear|class>

Example

Modelling JBoss AS 7.x internal dependency structure:

$ coffea -p -i /opt/jboss-7.2.0.GA/modules/ -Ip org.jboss. -Mrp org.jboss. -Mep 0 -El logging

Interactive mode equivalent:

>>> from coffea.builder import Builder
>>> from coffea.analyzer import Plotter
>>> from coffea.model import NodeIdFilter, NodeIdMapper
>>> b = Builder()
>>> b.model.node_filters.append(NodeIdFilter(lambda it: it.startswith('org.jboss.')))
>>> b.model.node_filters.append(NodeIdMapper(lambda it: it.replace('org.jboss.', '')))
>>> b.model.node_filters.append(NodeIdMapper(lambda it: it.split('.')[0]))
>>> b.model.node_filters.append(NodeIdFilter(lambda it: it not in ['logging']))
>>> b.append('/opt/jboss-7.2.0.GA/modules/')
No handlers could be found for logger "scanner"
>>> print len(b.model.nodes)
48
>>> p = Plotter(b.model)
>>> p.plot()

[Displays an interactive view of the dependency model]

>>> p.plot(filename='/tmp/jboss7_module_dependencies.png')

Output:

JBoss 7.x internal dependencies

Unit testing

You can run the test suite directly from the command line:

$ python -m unittest discover

Releases

No releases published

Packages

No packages published