class RepertoireModel:
def __init__(self):
self.processDirectory = True
self.num_operations = 0
self.operations_so_far = IntegerWrapper(0)
self.ccfx = CCFXEntryPoint('../ccFinder/ccfx',40,True,True)
self.flags = {"No":True, "Yes":False}
def setDiffPaths(self, path0 = None, path1 = None):
path0 = str(path0)
path1 = str(path1)
if (not os.path.isdir(path0) or
not os.path.isdir(path1)):
return False
self.paths = {'proj0':path0, 'proj1':path1}
return True
def setDiffPaths(self, path0 = None, path1 = None, isDirectory = True):
path0 = str(path0)
path1 = str(path1)
self.ccfx.isDirectory = self.processDirectory = isDirectory
if (isDirectory is True) and (not os.path.isdir(path0) or
not os.path.isdir(path1)):
return False
elif (isDirectory is False) and (not os.path.isfile(path0) or
not os.path.isfile(path1)):
return False
self.paths = {'proj0':path0, 'proj1':path1}
return True
def setTmpDirectory(self, path):
path = str(path)
if not os.path.isdir(path):
return False
# great, we have a scratch space, lets put our own directory there
# so we know we probably aren't going to fight someone else for names
uniq = 'repertoire_tmp_' + str(int(os.times()[4] * 100))
tmpPath = path + os.sep + uniq
os.mkdir(tmpPath)
self.tmpPath = tmpPath
return True
def setSuffixes(self, jSuff = '', cSuff = '', hSuff = ''):
jSuff = str(jSuff)
cSuff = str(cSuff)
hSuff = str(hSuff)
if jSuff.startswith('.'):
jSuff = jSuff[1:]
if cSuff.startswith('.'):
cSuff = cSuff[1:]
if hSuff.startswith('.'):
hSuff = hSuff[1:]
self.suffixes = {
'java':jSuff,
'cxx':cSuff,
'hxx':hSuff,
}
self.filters = {
'java' : DiffFilter(jSuff),
'cxx' : DiffFilter(cSuff),
'hxx' : DiffFilter(hSuff)
}
def setCcfxDirectory(self, path):
path = str(path)
if not os.path.isdir(path):
return False
ccfx_binary = path + "/ccfx"
if os.path.exists(ccfx_binary):
self.ccfx.ccfxPath = ccfx_binary
return True
return False
def setCcfxToken(self, token_size):
self.ccfx.tokenSize = token_size
print "setting ccFinder token size = " + token_size
return True
def setCcfxFileSeparator(self, flag):
self.ccfx.fileSep = self.flags[str(flag)]
print "setting ccFinder file separator flag to %d" % (self.ccfx.fileSep)
return True
def setCcfxGroupSeparator(self, flag):
self.ccfx.grpSep = self.flags[str(flag)]
print "setting ccFinder group separator flag to %d" % (self.ccfx.grpSep)
return True
def filterDiffProjs(self, interface):
# 3 different file formats, 2 operations each (filter/convert)
self.num_operations = len(os.listdir(self.paths['proj0'])) * 3 * 2
if self.paths['proj0'] != self.paths['proj1']:
self.num_operations += len(os.listdir(self.paths['proj1'])) * 3 * 2
self.num_operations += 2*6 #2 ccFinder call for all 6 output files
self.operations_so_far = IntegerWrapper(0)
for proj in ['proj0', 'proj1']:
for lang in ['java', 'cxx', 'hxx']:
the_filter = self.filters[lang]
for i, file_name in enumerate(os.listdir(self.paths[proj])):
if interface.cancelled():
return ('User cancelled processing', False)
interface.progress('Filtering {0} files'.format(lang),
self.operations_so_far.value / float(self.num_operations))
input_path = self.paths[proj] + os.sep + file_name
# out_path = (self.pb.getFilterOutputPath(proj, lang) +
# ('%04d' % i) + '.' + self.suffixes[lang])
out_path = (self.pb.getFilterOutputPath(proj, lang) +
file_name + '.' + self.suffixes[lang])
(ok, gotsome) = the_filter.filterDiff(input_path, out_path)
# this is actually tricky, if we got some output for java
# in one project but not the other, then we know that
# there can't be any clones
self.got_some[lang] = self.got_some[lang] and gotsome
if not ok:
return ('Error processing: ' + file_name, False)
self.operations_so_far.incr()
if self.paths['proj0'] == self.paths['proj1']:
print "filterDiffProjs: two paths same, breaking!!"
break
def filterDiffFiles(self, interface):
# 3 different file formats, 2 operations each (filter/convert)
self.num_operations = 3 * 2
if self.paths['proj0'] != self.paths['proj1']:
self.num_operations += 3 * 2
self.operations_so_far = IntegerWrapper(0)
input_file1 = self.paths['proj0']
input_file2 = self.paths['proj1']
lang1 = os.path.splitext(input_file1)[1] #extension
lang2 = os.path.splitext(input_file2)[1] #extension
if lang1 != lang2 :
print "!!the two files have different extension"
print "lang1 = " + lang1
print "lang2 = " + lang2
return False
for proj in ['proj0', 'proj1']:
for lang in ['java', 'cxx', 'hxx']:
the_filter = self.filters[lang]
if interface.cancelled():
return ('User cancelled processing', False)
interface.progress('Filtering {0} files'.format(lang),
self.operations_so_far.value / float(self.num_operations))
input_path = self.paths[proj]
out_path = (self.pb.getFilterOutputPath(proj, lang) +
os.path.basename(input_path) + '.' + self.suffixes[lang])
(ok, gotsome) = the_filter.filterDiff(input_path, out_path)
# this is actually tricky, if we got some output for java
# in one project but not the other, then we know that
# there can't be any clones
self.got_some[lang] = self.got_some[lang] and gotsome
if not ok:
return ('Error processing: ' + file_name, False)
self.operations_so_far.incr()
if self.paths['proj0'] == self.paths['proj1']:
print "filterDiffFiles: two paths same, breaking!!"
break
def filterDiffs(self, interface):
self.got_some = {'java':True, 'cxx':True, 'hxx':True}
# self.haveJava = haveC = haveH = False
self.pb = PathBuilder(self.tmpPath, force_clean = True)
# First, filter the input diffs by file type, so that all c diffs
# are in one set of files, and similarly for java/headers
if self.processDirectory is True:
self.filterDiffProjs(interface)
else:
self.filterDiffFiles(interface)
# Second, change each diff into ccFinder input format
converter = CCFXInputConverter()
callback = lambda: interface.progress(
'Converting to ccfx input format',
self.operations_so_far.incr() / float(self.num_operations))
converter.convert(self.pb, callback)
#new and old for 3 langs
self.num_operations = 3 * 2
self.operations_so_far = IntegerWrapper(0)
clone_path = self.pb.getCCFXOutputPath()
# Third, call ccfx for each directory
worked = True
for lang in ['java', 'cxx', 'hxx']:
if not self.got_some[lang]:
interface.progress('ccFinderX executing',
self.operations_so_far.incr() / float(self.num_operations))
continue
old_path0 = self.pb.getCCFXInputPath(PathBuilder.PROJ0, lang, False)
old_path1 = self.pb.getCCFXInputPath(PathBuilder.PROJ1, lang, False)
new_path0 = self.pb.getCCFXInputPath(PathBuilder.PROJ0, lang, True)
new_path1 = self.pb.getCCFXInputPath(PathBuilder.PROJ1, lang, True)
tmp_old_out = clone_path + self.pb.getCCFXOutputFileName(
lang, is_new = False, is_tmp = True)
tmp_new_out = clone_path + self.pb.getCCFXOutputFileName(
lang, is_new = True, is_tmp = True)
old_out = clone_path + self.pb.getCCFXOutputFileName(
lang, is_new = False, is_tmp = False)
new_out = clone_path + self.pb.getCCFXOutputFileName(
lang, is_new = True, is_tmp = False)
if self.paths['proj0'] == self.paths['proj1']:
old_path1 = old_path0
new_path1 = new_path0
worked = worked and self.ccfx.processPair(
old_path0, old_path1, tmp_old_out, old_out, lang)
interface.progress('ccFinderX executing',
self.operations_so_far.incr() / float(self.num_operations))
worked = worked and self.ccfx.processPair(
new_path0, new_path1, tmp_new_out, new_out, lang)
interface.progress('ccFinderX executing',
self.operations_so_far.incr() / float(self.num_operations))
if not worked:
return ('ccFinderX execution failed', False)
# Fourth, build up our database of clones
print "Repertoire filtering...."
#new and old for 3 langs
self.num_operations = 3 * 2
self.operations_so_far = IntegerWrapper(0)
for lang in ['java', 'cxx', 'hxx']:
if not self.got_some[lang]:
interface.progress('Repertoire filtering based on operation',
self.operations_so_far.incr() / float(self.num_operations))
continue
for is_new in [True, False]:
output = convert_ccfx_output(self.pb, lang, is_new)
rep_out_path = self.pb.getRepertoireOutputPath(lang, is_new)
suffix = '_old.txt'
if is_new:
suffix = '_new.txt'
output.writeToFile(rep_out_path + lang + suffix)
interface.progress('Repertoire filtering based on operation',
self.operations_so_far.incr() / float(self.num_operations))
print "Processing successful!!"
return ('Processing successful', True)
