def setUp(self):
self.testChunk = logChunk.logChunk("")
#Read in the full tests
self.chunk1 = logChunk.logChunk(self.readHelper("testChunk1.txt"))
self.chunk2 = logChunk.logChunk(self.readHelper("testChunk2.txt"))
self.chunk3 = logChunk.logChunk(self.readHelper("testChunk3.txt"))
self.chunk4 = logChunk.logChunk(self.readHelper("testChunk4.txt"))
self.chunk5 = logChunk.logChunk(self.readHelper("testChunk5.txt"))
self.chunk6 = logChunk.logChunk(self.readHelper("testChunk6.txt"))
self.chunk7 = logChunk.logChunk(self.readHelper("testChunk7.txt"))
self.chunk8 = logChunk.logChunk(self.readHelper("testChunk8.txt"))
self.chunk9 = logChunk.logChunk(self.readHelper("testChunk9.txt"))
def main():
print("Utility to BULK process github logs")
if len(sys.argv) < 2:
print("!!! Usage: python allRun.py top_project directory")
sys.exit()
if not os.path.isdir("../Results"):
os.mkdir("../Results")
fPtrChangeSummary = open("../Results/" + "ChangeSummary.csv", 'w')
fPtrChangeSummary.write(
"project,sha,author,author_email,commit_date,is_bug\n")
fPtrPatchSummary = open("../Results/" + "PatchSummary.csv", 'w')
fPtrMisMatchSummary = open("../Results/" + "MisMatchSummary.csv", 'w')
fPtrMisMatchSummary.write(
"project,Total,Match,MisMatch,Exception,matchException,misMatchException\n"
)
lst = []
listToDict = {}
mockChunk = logChunk("")
mockChunk.readKeywords(lst)
keywords = [sub_list[0] for sub_list in lst]
for keyword in keywords:
listToDict[str(keyword) + " Adds"] = 0
listToDict[str(keyword) + " Dels"] = 0
#fPtrPatchSummary.write("project, sha, language, file_name, is_test,bracket_diff,isExceptionPatch, method_name,total_add,total_del,uniqueExcepAdd,uniqueExcepDel,%s\n"%",".join(listToDict.keys()))
fPtrPatchSummary.write(
"project, sha, language, file_name, is_test, method_name,total_add,total_del,%s\n"
% ",".join(sorted(listToDict.keys())))
fPtrChangeSummary.close()
fPtrPatchSummary.close()
fPtrMisMatchSummary.close()
rootdir = str(sys.argv[1])
for dir in os.listdir(rootdir):
path = os.path.join(rootdir, dir)
print(path)
os.system('python ghProc.py %s' % path)
def main():
print "Utility to BULK process github logs"
if len(sys.argv) < 2:
print "!!! Usage: python allRun.py top_project directory"
sys.exit()
if not os.path.isdir("../Results"):
os.mkdir("../Results")
fPtrChangeSummary=open("../Results/"+"ChangeSummary.csv",'w')
fPtrChangeSummary.write("project,sha,author,commit_date,is_bug\n")
fPtrPatchSummary=open("../Results/"+"PatchSummary.csv",'w')
fPtrMisMatchSummary=open("../Results/"+"MisMatchSummary.csv",'w')
fPtrMisMatchSummary.write("project,Total,Match,MisMatch,Exception,matchException,misMatchException\n")
lst=[]
listToDict={}
mockChunk=logChunk("")
mockChunk.readKeywords(lst)
keywords= [sub_list[0] for sub_list in lst]
for keyword in keywords:
listToDict[str(keyword)+" Adds"]=0
listToDict[str(keyword)+" Dels"]=0
#fPtrPatchSummary.write("project, sha, language, file_name, is_test,bracket_diff,isExceptionPatch, method_name,total_add,total_del,uniqueExcepAdd,uniqueExcepDel,%s\n"%",".join(listToDict.keys()))
fPtrPatchSummary.write("project, sha, language, file_name, is_test, method_name,total_add,total_del,%s\n"%",".join(sorted(listToDict.keys())))
fPtrChangeSummary.close()
fPtrPatchSummary.close()
fPtrMisMatchSummary.close()
rootdir = str(sys.argv[1])
for dir in os.listdir(rootdir):
path= os.path.join(rootdir,dir)
print path
os.system('python ghProc.py %s'%path)
def processLog(self, config = ""):
if(config == ""):
config = self.config_info.CONFIG
signal.signal(signal.SIGALRM, timeout)
project1 = os.path.split(self.log_file)[0]
project1 = project1.rstrip(os.sep)
self.project_name = os.path.basename(project1)
print("---------- %s ------------\n" % (self.project_name))
if(self.config_info.DATABASE):
dl = dumpLogs(self.dbPass, self.config_info)
if(self.config_info.CSV):
if not os.path.isdir("../Results"):
os.mkdir("../Results")
inf1=open("../Results/"+str(self.project_name)+"ChangeSummary.csv",'w')
fPtrChangeSummary=open("../Results/"+"ChangeSummary.csv",'w')
inf1.write("project,sha,author,commit_date,is_bug\n")
inf2=open("../Results/"+str(self.project_name)+"PatchSummary.csv",'w')
fPtrPatchSummary=open("../Results/"+"PatchSummary.csv",'w')
lst=[]
listToDict={}
mockChunk=logChunk("", "C", self.config_info) #TODO: This is C specific, Why is this C specific?
lst = mockChunk.readKeywords(lst)
keywords= [k[0] for k in lst if k[1] == INCLUDED]
for keyword in keywords:
listToDict[str(keyword)+" Adds"]=0
listToDict[str(keyword)+" Dels"]=0
inf2.write("project, sha, language, file_name, is_test, method_name,total_add,total_del,%s\n"%",".join(sorted(listToDict.keys())))
inf = codecs.open(self.log_file, "r", "iso-8859-1")
shaObj = None
patchObj = None
is_diff = False
log_mssg = ""
is_no_prev_ver = False
is_no_next_ver = False
curLogChunk = logChunk("", "C", self.config_info)
linenum = 0
for l in inf:
try:
signal.alarm(0)
sha = self.isSha(l)
line = l
#if(self.config_info.DEBUGLITE):
# try:
# print(line)
# except:
# pass
if sha:
#Reverting back to version that outputs at the end...
#if(shaObj != None):
# if(self.config_info.DEBUGLITE):
# print("Writing Sha:" + sha)
# if(self.config_info.DATABASE):
# shaObj.dumpSha(dl)
# elif(self.config_info.CSV):
# shaObj.shaToCsv(inf1,inf2,fPtrChangeSummary,fPtrPatchSummary)
# else:
# shaObj.printSha()
shaObj = Sha(self.project_name, sha)
#if(self.config_info.DEBUGLITE): #Save for testing.
self.shas.append(shaObj) #This will become very memory intensive in large git logs.
is_diff = False
log_mssg = ""
continue
elif self.isAuthor(line,shaObj):
continue
elif self.isDate(line,shaObj):
continue
fullLine=line
line=line.rstrip()
if line.startswith('diff --git '):
shaObj.setLog(log_mssg)
is_diff = True
is_no_prev_ver = False
is_no_next_ver = False
continue
if patchObj != None:
shaObj.patches.append(patchObj)
elif is_diff == False:
if not line.strip():
continue
log_mssg += line + "\t"
if is_diff:
if line.startswith("--- a/"):
#Finish the changes to the old patch object
if(patchObj != None):
#If there is an existing chunk to parse, process it
if(curLogChunk.header != ""):
if(self.config_info.DEBUG):
print("New diff with previous version: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
#Reset the current chunk obj
if (self.config_info.DEBUG):
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang("." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatch(line)
shaObj.patches.append(patchObj)
#print patchObj
#print shaObj.patches
elif (line == '--- /dev/null'): #earlier file was empty
is_no_prev_ver = True
elif (line == '+++ /dev/null'): #next file version was empty
is_no_next_ver = True
continue
elif (is_no_prev_ver == True) and line.startswith("+++ b/"):
#Finish the changes to the old patch object
if(patchObj != None):
if(curLogChunk.header != ""): #If there is an existing chunk
if (self.config_info.DEBUG):
print("New diff with no previous version: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
if (self.config_info.DEBUG):
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang("." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatchWithNoPrevVersion(line)
shaObj.patches.append(patchObj)
else: #Then we reached a content line.
self.processPatch(fullLine, patchObj, curLogChunk)
except TimeExceededError.TimeExceededError:
print("Line Timed out, moving to next.")
continue
#Clear timeouts.
signal.alarm(0)
#Make sure to get the last patch in the file!
if(curLogChunk.header != ""): #If there is an existing chunk to parse
if(self.config_info.DEBUG):
print("Last Patch: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
#if shaObj != None:
# shaObj.patches.append(patchObj)
parseFinish = datetime.now()
if(self.shas != []): #If the log wasn't empty...
#Create the change summary table and the method change table now if necessary
if(self.config_info.DATABASE):
cfg = Config(self.config_info.CONFIG)
keywordFile = cfg.ConfigSectionMap("Keywords")
full_title = dumpLogs.getFullTitleString(curLogChunk.getEmptyKeywordDict())
dl.createSummaryTable()
if(full_title != ""): #Check if the changes table exists and create it if we have a title.
dl.createMethodChangesTable(full_title)
for s in self.shas:
#s.printSha()
if s != None:
if(self.config_info.DATABASE):
s.dumpSha(dl)
elif(self.config_info.CSV):
s.shaToCsv(inf1,inf2,fPtrChangeSummary,fPtrPatchSummary)
else:
s.printSha()
#Write out last sha.
#if(shaObj != None and self.config_info.DATABASE):
# if(self.config_info.DEBUGLITE):
# print("Writing to db.")
# shaObj.dumpSha(dl)
if(self.config_info.DATABASE):
print("Closing Time.")
dl.close()
if(self.config_info.CSV):
inf1.close()
inf2.close()
fPtrChangeSummary.close()
fPtrPatchSummary.close()
print("Sha's processed:")
print(len(self.shas))
return parseFinish
def setUp(self):
self.method1 = "def debugFunctions(self, funcList):"
self.method2 = "def _Funct1():"
self.method3 = " def Another_function(arg1, arg2 = 5): "
self.method4 = "def test( _args ): "
self.method5 = "def 1Bad():"
self.method6 = "def functionname([formal_args,] *var_args_tuple ):"
self.method7 = "def alsoBad() "
self.method8 = "noFunct():"
self.method9 = " def okay(args = 4):"
c_info = ConfigInfo("../util/pytest.ini")
self.testChunk = logChunk.logChunk("", "Python")
self.chunk1 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk1.txt"), "Python",
c_info)
self.chunk2 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk2.txt"), "Python",
c_info)
self.chunk3 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk3.txt"), "Python",
c_info)
self.chunk4 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk4.txt"), "Python",
c_info)
self.chunk5 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk5.txt"), "Python",
c_info)
self.chunk6 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk6.txt"), "Python",
c_info)
self.chunk7 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk7.txt"), "Python",
c_info)
self.chunk8 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk8.txt"), "Python",
c_info)
self.chunk9 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk9.txt"), "Python",
c_info)
self.chunk10 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk10.txt"), "Python",
c_info)
self.chunk11 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk11.txt"), "Python",
c_info)
self.chunk12 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk12.txt"), "Python",
c_info)
self.chunk13 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk13.txt"), "Python",
c_info)
self.chunk14 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk14.txt"), "Python",
c_info)
self.chunk15 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk15.txt"), "Python",
c_info)
self.chunk16 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk16.txt"), "Python",
c_info)
self.chunk17 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk17.txt"), "Python",
c_info)
self.chunk18 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk18.txt"), "Python",
c_info)
self.chunk19 = logChunk.logChunk(
self.readHelper("testfiles/Python/testChunk19.txt"), "Python",
c_info)
def processLog(self, config=""):
if (config == ""):
config = self.config_info.CONFIG
signal.signal(signal.SIGALRM, timeout)
project1 = os.path.split(self.log_file)[0]
project1 = project1.rstrip(os.sep)
self.project_name = os.path.basename(project1)
print(("---------- %s ------------\n" % (self.project_name)))
if (self.config_info.DATABASE):
dl = dumpLogs(self.dbPass, self.config_info)
if (self.config_info.CSV):
if not os.path.isdir("../Results"):
os.mkdir("../Results")
inf1 = open(
"../Results/" + str(self.project_name) + "ChangeSummary.csv",
'w')
fPtrChangeSummary = open("../Results/" + "ChangeSummary.csv", 'w')
inf1.write("project,sha,author,author_email,commit_date,is_bug\n")
inf2 = open(
"../Results/" + str(self.project_name) + "PatchSummary.csv",
'w')
fPtrPatchSummary = open("../Results/" + "PatchSummary.csv", 'w')
lst = []
listToDict = {}
mockChunk = logChunk(
"", "C", self.config_info
) #TODO: This is C specific, Why is this C specific?
lst = mockChunk.readKeywords(lst)
keywords = [k[0] for k in lst if k[1] == INCLUDED]
for keyword in keywords:
listToDict["\"" + str(keyword) + "\" adds"] = 0
listToDict["\"" + str(keyword) + "\" dels"] = 0
inf2.write(
"project, sha, language, file_name, is_test, method_name,total_add,total_del,%s\n"
% ",".join(sorted(listToDict.keys())))
inf = codecs.open(self.log_file, "r", "iso-8859-1")
shaObj = None
patchObj = None
is_diff = False
log_mssg = ""
is_no_prev_ver = False
is_no_next_ver = False
curLogChunk = logChunk("", "C", self.config_info)
linenum = 0
for l in inf:
try:
signal.alarm(0)
sha = self.isSha(l)
line = l
#if(self.config_info.DEBUGLITE):
# try:
# print(line)
# except:
# pass
if sha:
#Reverting back to version that outputs at the end...
#if(shaObj != None):
# if(self.config_info.DEBUGLITE):
# print("Writing Sha:" + sha)
# if(self.config_info.DATABASE):
# shaObj.dumpSha(dl)
# elif(self.config_info.CSV):
# shaObj.shaToCsv(inf1,inf2,fPtrChangeSummary,fPtrPatchSummary)
# else:
# shaObj.printSha()
shaObj = Sha(self.project_name, sha)
#if(self.config_info.DEBUGLITE): #Save for testing.
self.shas.append(
shaObj
) #This will become very memory intensive in large git logs.
is_diff = False
log_mssg = ""
continue
elif self.isAuthor(line, shaObj):
continue
elif self.isDate(line, shaObj):
continue
fullLine = line
line = line.rstrip()
if line.startswith('diff --git '):
shaObj.setLog(log_mssg)
is_diff = True
is_no_prev_ver = False
is_no_next_ver = False
continue
if patchObj != None:
shaObj.patches.append(patchObj)
elif is_diff == False:
if not line.strip():
continue
log_mssg += line + "\t"
if is_diff:
if line.startswith("--- a/"):
#Finish the changes to the old patch object
if (patchObj != None):
#If there is an existing chunk to parse, process it
if (curLogChunk.header != ""):
if (self.config_info.DEBUG):
print(("New diff with previous version: " +
line))
print(("HEADER: " + curLogChunk.header))
self.processLastChunk(patchObj, curLogChunk)
#Reset the current chunk obj
if (self.config_info.DEBUG):
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang(
"." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatch(line)
shaObj.patches.append(patchObj)
#print patchObj
#print shaObj.patches
elif (line == '--- /dev/null'): #earlier file was empty
is_no_prev_ver = True
elif (line == '+++ /dev/null'
): #next file version was empty
is_no_next_ver = True
continue
elif (is_no_prev_ver
== True) and line.startswith("+++ b/"):
#Finish the changes to the old patch object
if (patchObj != None):
if (curLogChunk.header !=
""): #If there is an existing chunk
if (self.config_info.DEBUG):
print((
"New diff with no previous version: " +
line))
print(("HEADER: " + curLogChunk.header))
self.processLastChunk(patchObj, curLogChunk)
if (self.config_info.DEBUG):
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang(
"." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatchWithNoPrevVersion(line)
shaObj.patches.append(patchObj)
else: #Then we reached a content line.
self.processPatch(fullLine, patchObj, curLogChunk)
except TimeExceededError.TimeExceededError:
print("Line Timed out, moving to next.")
continue
#Clear timeouts.
signal.alarm(0)
#Make sure to get the last patch in the file!
if (curLogChunk.header != ""): #If there is an existing chunk to parse
if (self.config_info.DEBUG):
print(("Last Patch: " + line))
print(("HEADER: " + curLogChunk.header))
self.processLastChunk(patchObj, curLogChunk)
#if shaObj != None:
# shaObj.patches.append(patchObj)
parseFinish = datetime.now()
if (self.shas != []): #If the log wasn't empty...
#Create the change summary table and the method change table now if necessary
if (self.config_info.DATABASE):
cfg = Config(self.config_info.CONFIG)
keywordFile = cfg.ConfigSectionMap("Keywords")
full_title = dumpLogs.getFullTitleString(
curLogChunk.getEmptyKeywordDict())
dl.createSummaryTable()
if (
full_title != ""
): #Check if the changes table exists and create it if we have a title.
dl.createMethodChangesTable(full_title)
for s in self.shas:
#s.printSha()
if s != None:
if (self.config_info.DATABASE):
s.dumpSha(dl)
elif (self.config_info.CSV):
s.shaToCsv(inf1, inf2, fPtrChangeSummary,
fPtrPatchSummary)
else:
s.printSha()
#Write out last sha.
#if(shaObj != None and self.config_info.DATABASE):
# if(self.config_info.DEBUGLITE):
# print("Writing to db.")
# shaObj.dumpSha(dl)
if (self.config_info.DATABASE):
print("Closing Time.")
dl.close()
if (self.config_info.CSV):
inf1.close()
inf2.close()
fPtrChangeSummary.close()
fPtrPatchSummary.close()
print("Sha's processed:")
print((len(self.shas)))
return parseFinish
def setUp(self):
self.javaMethod1 = "public static Intent createIntent(Context context, String username, String password) {"
self.javaMethod2 = " public <V> V post(final String uri, final Object params, final Type type) \n throws IOException {"
self.javaMethod3 = "public static Intent createIntent(final Collection<? extends Issue> issues,\n final Repository repository, final int position) {"
self.javaMethod4 = "@Override \n public List<User> run(Account account) throws Exception {"
self.javaMethod5 = "private JClass typeBoundsToJClass(GeneratedClassHolder holder, List<? extends TypeMirror> bounds, Map<String, TypeMirror> actualTypes) {"
self.javaMethod6 = " public JMethod implementMethod(GeneratedClassHolder holder, List<ExecutableElement> methods, String methodName, String returnType, String... parameterTypes) {"
self.javaMethod7 = "ProgrammerInterview pInstance = new ProgrammerInterview() {\npublic void read() {"
self.testChunk2 = logChunk.logChunk("", "Java")
#Read in the block tests
self.chunkb1 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk1.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb2 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk2.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb3 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk3.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb4 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk4.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb5 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk5.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb6 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk6.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb7 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk7.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb8 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk8.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb9 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk9.txt"), "Java", "../util/sample_conf2.ini")
self.chunkb10 = logChunk.logChunk(self.readHelper("testfiles/Block/testChunk10.txt"), "Java", "../util/sample_conf2.ini")
def setUp(self):
self.keyword1 = ["wtc/Assertions.h","excluded","single"]
self.keyword2 = ["ut_ad","included","single"]
self.keyword3 = ["try","Included","BLOCK"]
self.keyword4 = ["for","excludeD","block"]
self.keyword5 = ["ut_ad included single"]
self.keyword6 = ["printf(format","str)","included","single"]
self.keyword7 = ["printf(format, str)","included","single"]
self.keyword8 = ["assert","incuded","single"]
self.keyword9 = ["assert","included","lock"]
self.method1 = "static void blarg() {"
self.method2 = "int more(int stuff) {"
self.method3 = "ccv_string * getStuff (int[] stuffToGet) {"
self.method4 = "int add2to3(int (*functionPtr)(int, int)) {"
self.method5 = "public static void other(int one, int (*functionPtr)(int, int)) {"
self.method6 = "static void\n multiline(\n int arg1, string arg2\n) {"
self.method7 = "int lotsOfSpace (int stuff) {"
self.method8 = " .matrix = {"
self.method9 = "ccv_string* getStuff (int[] stuffToGet) {"
self.method10 = "ccv_string *getStuff (int[] stuffToGet) {"
self.method11 = "void NdbBlob::getBlobEvent(NdbEventImpl& be, const NdbEventImpl* e, const NdbColumnImpl* c) {"
self.method12 = "bool Repair_mrg_table_error_handler::handle_condition(THD *,uint sql_errno,const char*, MYSQL_ERROR::enum_warning_level level,const char*,MYSQL_ERROR ** cond_hdl) {"
self.method13 = "(_log2) += 1; static int CeilingLog2(unsigned int i) {"
#self.method14 = "(STATEMENT_CONSTRUCTOR_BASE_PARAMETERS, ExpressionPtr exp) : Statement(STATEMENT_CONSTRUCTOR_BASE_PARAMETER_VALUES), m_exp(exp) {"
self.method15 = "static JSC::UString& globalExceptionString(){"
self.method16 = "(jint) AWT_WINDOW_LOST_FOCUS, (jint) AWT_WINDOW_DEACTIVATED, static gboolean window_focus_in_cb (GtkWidget * widget, GdkEventFocus *event, jobject peer) {"
self.method17 = "LinuxPtraceDumper dumper(getpid()); } TEST(LinuxPtraceDumperTest, FindMappings) {"
self.method18 = "endif defined(HAVE_ALTIVEC) include elif defined(HAVE_SSE2) include endif ONLY64 inline static int idxof(int i) {"
self.method19 = "#define UNUSED_VARIABLE(x) (void)(x) const char *sfmt_get_idstring(sfmt_t * sfmt) {"
self.method20 = "static av_always_inline void hl_motion_420(H264Context *h, uint8_t *dest_y, uint8_t *dest_cb, uint8_t *dest_cr, qpel_mc_func (*qpix_put)[16], h264_chroma_mc_func (*chroma_put), qpel_mc_func (*qpix_avg)[16], h264_chroma_mc_func (*chroma_avg), h264_weight_func *weight_op, h264_biweight_func *weight_avg, int pixel_shift) {"
self.method21 = " for (j = 0; j < i; j++) if (!memcmp(h->pps.scaling_matrix8[j], h->pps.scaling_matrix8[i], 64 * sizeof(uint8_t))) {"
self.method22 = " c_type jl_unbox_##j_type(jl_value_t *v) {"
self.testChunk = logChunk.logChunk("", "C++")
self.testChunk_C = logChunk.logChunk("", "C")
self.javaMethod1 = "public static Intent createIntent(Context context, String username, String password) {"
self.javaMethod2 = " public <V> V post(final String uri, final Object params, final Type type) \n throws IOException {"
self.javaMethod3 = "public static Intent createIntent(final Collection<? extends Issue> issues,\n final Repository repository, final int position) {"
self.javaMethod4 = "@Override \n public List<User> run(Account account) throws Exception {"
self.javaMethod5 = "private JClass typeBoundsToJClass(GeneratedClassHolder holder, List<? extends TypeMirror> bounds, Map<String, TypeMirror> actualTypes) {"
self.javaMethod6 = " public JMethod implementMethod(GeneratedClassHolder holder, List<ExecutableElement> methods, String methodName, String returnType, String... parameterTypes) {"
#Read in the single tests
self.chunk1 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk1.txt"), "C++") #Check C++
self.chunk2 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk2.txt"), "C") #Check C
self.chunk3 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk3.txt"), "C++") #Check C++
#self.chunk4 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk4.txt")) #Nope
#self.chunk5 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk5.txt")) #Nope
self.chunk6 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk6.txt"), "C++") #Check C++
self.chunk7 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk7.txt"), "C++") #Check C++
self.chunk8 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk8.txt"), "C++") #Check C
self.chunk9 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk9.txt"), "C++") #Check C++
self.chunk10 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk10.txt"), "C++") #Check C++
self.chunk11 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk11.txt"), "C++") #Check C++
self.chunk12 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk12.txt"), "C++") #Check C++
self.chunk13 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk13.txt"), "C++") #Check C++
self.chunk14 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk14.txt"), "C++") #Check C++
self.chunk15 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk15.txt"), "C++") #Check C++
#self.chunk16 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk16.txt")) #Nope
#self.chunk17 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk17.txt")) #Nope
#self.chunk18 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk18.txt")) #Nope
#self.chunk19 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk19.txt")) #Nope
#self.chunk20 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk20.txt")) #Nope
self.chunk21 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk21.txt"), "C++") #Check C++
self.chunk22 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk22.txt"), "C++") #Check C++
self.chunk23 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk23.txt"), "C++") #Check C++
self.chunk24 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk24.txt"), "C++") #Check C++
self.chunk25 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk25.txt"), "C") #Check C
#self.chunk26 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk26.txt")) #Nope
self.chunk27 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk27.txt"), "C") #Check C
#self.chunk28 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk28.txt")) #Nope
self.chunk29 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk29.txt"), "C") #Check C
#self.chunk30 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk30.txt")) #Maybe?
self.chunk31 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk31.txt"), "C++") #Check C++
self.chunk32 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk32.txt"), "C++") #Check C++
self.chunk33 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk33.txt"), "C++") #Check C++
#self.chunk34 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk34.txt")) #Nope
self.chunk35 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk35.txt"), "C") #Check C
self.chunk36 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk36.txt"), "C++") #Check C++
self.chunk37 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk37.txt"), "C++") #Check C++
self.chunk38 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk38.txt"), "C++") #Check C++
self.chunk39 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk39.txt"), "C") #C
self.chunk40 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk40.txt"), "C") #Check C
self.chunk41 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk41.txt"), "C++") #Check C++
self.chunk42 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk42.txt"), "C++") # C++
self.chunk43 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk43.txt"), "C") #Check C
self.chunk44 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk44.txt"), "C++") # C++
self.chunk45 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk45.txt"), "C++") # C++
self.chunk46 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk46.txt"), "C++") # C++
self.chunk47 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk47.txt"), "C++") # C++
self.chunk48 = logChunk.logChunk(self.readHelper("testfiles/Single/testChunk48.txt"), "C++") # C++
import sys
import sys
def setUp(self):
self.method1 = "def debugFunctions(self, funcList):"
self.method2 = "def _Funct1():"
self.method3 = " def Another_function(arg1, arg2 = 5): "
self.method4 = "def test( _args ): "
self.method5 = "def 1Bad():"
self.method6 = "def functionname([formal_args,] *var_args_tuple ):"
self.method7 = "def alsoBad() "
self.method8 = "noFunct():"
self.method9 = " def okay(args = 4):"
c_info = ConfigInfo("../util/pytest.ini")
self.testChunk = logChunk.logChunk("", "Python")
self.chunk1 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk1.txt"), "Python", c_info)
self.chunk2 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk2.txt"), "Python", c_info)
self.chunk3 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk3.txt"), "Python", c_info)
self.chunk4 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk4.txt"), "Python", c_info)
self.chunk5 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk5.txt"), "Python", c_info)
self.chunk6 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk6.txt"), "Python", c_info)
self.chunk7 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk7.txt"), "Python", c_info)
self.chunk8 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk8.txt"), "Python", c_info)
self.chunk9 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk9.txt"), "Python", c_info)
self.chunk10 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk10.txt"), "Python", c_info)
self.chunk11 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk11.txt"), "Python", c_info)
self.chunk12 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk12.txt"), "Python", c_info)
self.chunk13 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk13.txt"), "Python", c_info)
self.chunk14 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk14.txt"), "Python", c_info)
self.chunk15 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk15.txt"), "Python", c_info)
self.chunk16 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk16.txt"), "Python", c_info)
self.chunk17 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk17.txt"), "Python", c_info)
self.chunk18 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk18.txt"), "Python", c_info)
self.chunk19 = logChunk.logChunk(self.readHelper("testfiles/Python/testChunk19.txt"), "Python", c_info)
def processLog(self, config = Util.CONFIG):
project1 = os.path.split(self.log_file)[0]
project1 = project1.rstrip(os.sep)
self.project_name = os.path.basename(project1)
print("---------- %s ------------\n" % (self.project_name))
if(Util.DATABASE == 1):
dl = dumpLogs()
if(Util.CSV==1):
if not os.path.isdir("../Results"):
os.mkdir("../Results")
inf1=open("../Results/"+str(self.project_name)+"ChangeSummary.csv",'w')
fPtrChangeSummary=open("../Results/"+"ChangeSummary.csv",'a')
inf1.write("project,sha,author,commit_date,is_bug\n")
inf2=open("../Results/"+str(self.project_name)+"PatchSummary.csv",'w')
fPtrPatchSummary=open("../Results/"+"PatchSummary.csv",'a')
lst=[]
listToDict={}
mockChunk=logChunk("", "C")
mockChunk.readKeywords(lst)
keywords= [sub_list[0] for sub_list in lst]
for keyword in keywords:
listToDict[str(keyword)+" Adds"]=0
listToDict[str(keyword)+" Dels"]=0
inf2.write("project, sha, language, file_name, is_test,isExceptionPatch, method_name,total_add,total_del,%s\n"%",".join(listToDict.keys()))
inf = codecs.open(self.log_file, "r", "iso-8859-1")
shaObj = None
patchObj = None
is_diff = False
log_mssg = ""
is_no_prev_ver = False
is_no_next_ver = False
curLogChunk = logChunk("", "C", config)
linenum = 0
for l in inf:
sha = self.isSha(l)
line = l
if sha:
if(shaObj != None):
if(Util.DATABASE):
shaObj.dumpSha(dl)
else:
shaObj.printSha()
if(Util.CSV):
shaObj.shaToCsv(inf1,inf2,fPtrChangeSummary,fPtrPatchSummary)
shaObj = Sha(self.project_name, sha)
if(Util.DEBUGLITE): #Save for testing.
self.shas.append(shaObj) #This will become very memory intensive in large git logs.
is_diff = False
log_mssg = ""
continue
elif self.isAuthor(line,shaObj):
continue
elif self.isDate(line,shaObj):
continue
fullLine=line
line=line.strip()
if line.startswith('diff --git '):
shaObj.setLog(log_mssg)
is_diff = True
is_no_prev_ver = False
is_no_next_ver = False
continue
if patchObj != None:
shaObj.patches.append(patchObj)
elif is_diff == False:
if not line.strip():
continue
log_mssg += line + "\t"
if is_diff:
if line.startswith("--- a/"):
#Finish the changes to the old patch object
if(patchObj != None):
#If there is an existing chunk to parse, process it
if(curLogChunk.header != ""):
if Util.DEBUG == 1:
print("New diff with previous version: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
#Reset the current chunk obj
if Util.DEBUG == 1:
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang("." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatch(line)
shaObj.patches.append(patchObj)
#print patchObj
#print shaObj.patches
elif (line == '--- /dev/null'): #earlier file was empty
is_no_prev_ver = True
elif (line == '+++ /dev/null'): #next file version was empty
is_no_next_ver = True
continue
elif (is_no_prev_ver == True) and line.startswith("+++ b/"):
#Finish the changes to the old patch object
if(patchObj != None):
if(curLogChunk.header != ""): #If there is an existing chunk
if Util.DEBUG == 1:
print("New diff with no previous version: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
if Util.DEBUG == 1:
print("Resetting.")
curLogChunk.reset()
curLogChunk.setLang("." + self.cur_lang) #DOUBLE CHECK ME!
patchObj = self.createPatchWithNoPrevVersion(line)
shaObj.patches.append(patchObj)
else: #Then we reached a content line.
self.processPatch(fullLine, patchObj, curLogChunk)
#if shaObj != None:
# shaObj.patches.append(patchObj)
#Make sure to get the last patch in the file!
if(curLogChunk.header != ""): #If there is an existing chunk to parse
if Util.DEBUG == 1:
print("Last Patch: " + line)
print("HEADER: " + curLogChunk.header)
self.processLastChunk(patchObj, curLogChunk)
#Write out last sha.
if(shaObj != None and Util.DATABASE):
if(Util.DEBUGLITE):
print("Writing to db.")
shaObj.dumpSha(dl)
if(Util.DATABASE == 1):
print("Closing Time.")
dl.close()
if(Util.CSV == 1):
shaObj.printSha();
shaObj.shaToCsv(inf1,inf2,fPtrChangeSummary,fPtrPatchSummary)
inf1.close()
inf2.close()
fPtrChangeSummary.close()
fPtrPatchSummary.close()