getclasses.py

getclasses.py

import javalang
import sys


def addextends(out, etype, val):
  if isinstance(etype, list):
    for et in etype:
      addextends(out, et, val)
    return
  if not etype:
    ename = "Object"
  else:
    ename = etype.name
  print "Type ",val,"extends",ename
  out.setdefault(ename,[]).append(val)

def getextends(out, javafiles):
  for jfile in javafiles:
    if jfile.startswith('#'): continue
    print "opening ", jfile
    javastring = open(jfile, 'r').read()
    tree = javalang.parse.parse(javastring)
    for t in tree.types:
      # print t, t.attrs
      #addextends(out, t.extends, t)  #if we wanted to keep a mapping to the whole type
      extends = getattr(t,"extends",None)
      addextends(out, extends, t.name)
      # extends = getattr(t,"implements",None)
      # addextends(out, extends, t.name)


def getsubs(emap, name):
  ret = emap.get(name)
  if ret:
    #recurse
    sublist=[]
    for sub in ret:
      subsub = getsubs(emap, sub)
      if subsub: sublist += subsub
    ret += sublist
  else: 
    ret = []
  return ret


#emap = {}
#
#files = []
#for line in sys.stdin:
#  files += line.split()
#
#getextends(emap, files)

emap4 = {}
emap7 = {}

#getextends(emap4, open("small4.txt","r").read().split())
#getextends(emap7, open("small7.txt","r").read().split())

getextends(emap4, open("lusolr4files.txt","r").read().split())
getextends(emap7, open("lusolr7files.txt","r").read().split())


# test to make sure that we're getting whole hierarchy
# emap4['ApostropheFilterFactory']=['MyApostropheFilterFactory']

# find /opt/code/lusolr/4x -name \*.java | grep -v '/test' > lusolr4files.txt
# find /opt/code/lusolr/70 -name \*.java | grep -v '/test' > lusolr7files.txt
# issues parsing the following file, comment it out
# /opt/code/lusolr/70//lucene/highlighter/src/java/org/apache/lucene/search/uhighlight/UnifiedHighlighter.java

# extends Analyzer TokenStream FieldType


classes=["TokenizerFactory","TokenFilterFactory","Analyzer","FieldType","SimilarityFactory"]

for cls in classes:
  print "### Processing type", cls
  r4 = getsubs(emap4, cls)
  r7 = getsubs(emap7, cls)
  print set(r4) - set(r7)