textuml.py

"""
Fetches an umlgraph from a github wikipage and converts it into a 
png using graphviz and UMLGraph.

The script searches on a given page for a pre+code compbination with the class
"uml", which is then converted into a png file.

Requirements:
 * BeautifulSoup
 * graphviz
 * UMLGraph <http://www.umlgraph.org/>
 * java
"""

from __future__ import with_statement
import sys, shutil, urllib2, tempfile, subprocess, re, optparse, os
from os.path import dirname, join
from BeautifulSoup import BeautifulStoneSoup as BeautifulSoup

parser = optparse.OptionParser()
parser.usage = "%prog [OPTIONS] SRC TARGET"
parser.add_option('--umlgraph-path', dest='umlgraph', default=None, help="Path to UMLGraph's .jar file")
parser.add_option('--java-path', dest='java', default='java')
opts, args = parser.parse_args()

if opts.umlgraph is None:
    raise RuntimeError, "No UMLGraph path given"

url = args[0]
target = args[1]
tools_jar = join(os.environ['JAVA_HOME'], 'lib', 'tools.jar')
data = urllib2.urlopen(url).read()
data = data.replace('value="Search",', '') # Fix a minor markup issue on github
soup = BeautifulSoup(data)
uml = None
try:
    uml =  u''.join(soup.find('pre', attrs={'class': re.compile('uml')}).find('code').contents)
except AttributeError, e:
    print "Couldn't find any UML data on the given page"
    sys.exit(1)
        
dir_ = tempfile.mkdtemp()
try:
    uml_file = join(dir_, 'uml.java')
    dot_file = join(dir_, 'uml.dot')
    png_file = join(dir_, 'uml.png')
    with open(uml_file, 'w+') as fp:
        fp.write(uml)
    subprocess.call([opts.java, "-classpath", ':'.join([opts.umlgraph, tools_jar]), "org.umlgraph.doclet.UmlGraph", "-quiet", "-output", dot_file, "-package", "uml", uml_file])
    subprocess.call(['dot', '-Tpng', '-o'+png_file, dot_file])
    shutil.copy(png_file, target)
finally:
    shutil.rmtree(dir_)