A Python script to count the number of classes and methods in a .jar file using javap.

jarstats.py

#!/usr/bin/env python

import sys, os, re, zipfile, subprocess

# Rudimentary parsing of javap output
regex_type = r'[a-zA-Z0-9.<>, ?$[\]]+'
regex_class = r'(?:(public|private|protected) )?((?:(?:static|abstract|final) ?)*)(class|interface) (' + regex_type + ') (?:extends ((?:' + regex_type +'),?)+ )?(?:implements ((?:[a-zA-Z0-9\\.<>\\?\\$])+,?)+ )?{([^}]+)}'
regex_method = r'(?:(public|private|protected) )?((?:static|abstract|final) ?)*(?:(' + regex_type + ') )?([a-zA-Z]+)\\(([^\\)]*)\\)'
regex_field = r'(?:(public|private|protected) )?((?:(?:static|abstract|final) ?)*)(' + regex_type + ') ([a-zA-Z0-9]+)'
RE_TYPE = re.compile(regex_type)
RE_CLASS = re.compile(regex_class)
RE_METHOD = re.compile(regex_method)
RE_FIELD = re.compile(regex_field)


def classnamesfromjar(jarfilepath):
    '''Return list of fully-qualified classnames.'''
    with zipfile.ZipFile(jarfilepath, 'r') as zf:
        for zi in zf.infolist():
            fn = zi.filename
            if fn.endswith('.class'):
                dotted = fn.replace('/', '.')
                assert dotted.endswith('.class')
                yield dotted[:len(dotted)-len('.class')]

def runjavap(jarfilepath):
    '''Run javap on this jar for all the classes in the jar. Return the output of the javap command.'''
    classnames = classnamesfromjar(jarfilepath)
    return runjavap_aux(classpath=jarfilepath, classnames=classnames)

def runjavap_aux(classpath, classnames):
    '''Run javap on the given classpath for the given classnames. Return the output of the javap command.'''
    cmdarglist = ['javap', '-classpath', classpath] + list(classnames)
    p = subprocess.Popen(cmdarglist, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    tup_output = p.communicate()
    stdout = tup_output[0]
    stderr = tup_output[1]
    return stdout

def countclassesandmethodsinjarfile(jarfilepath):
    '''Return a tuple (nc, nm) with the number of classes and number of methods in the jarfile.'''
    javap_output = runjavap(jarfilepath)
    nc = len(RE_CLASS.findall(javap_output))
    nm = len(RE_METHOD.findall(javap_output))
    return nc, nm

if __name__ == '__main__':
    for jarpathname in sys.argv[1:]:
        nc, nm = countclassesandmethodsinjarfile(jarpathname)
        print("%s\t%d\t%d" % (jarpathname, nc, nm))