renamed utils.entity_utils.py to utils.java_utils.py + moved data fil…

…es related to hist, decor, incode and jdeodorant to their respective directories in approaches/

approaches/decor/core_metrics.py

@@ -6,7 +6,7 @@
 import os
 
 def getGCCoreMetrics(systemName):
-	decorMetricsFile = os.path.join(ROOT_DIR, 'data', 'metric_files', 'decor', systemName + '.csv')
+	decorMetricsFile = os.path.join(ROOT_DIR, 'approaches', 'decor', 'metric_files', systemName + '.csv')
 
 	dictionnary = {c:[0., 0., 0, 0] for c in data_utils.getClasses(systemName)}
 	with open(decorMetricsFile, 'rb') as csvfile:

approaches/hist/core_metrics.py

@@ -1,7 +1,7 @@
 from context import ROOT_DIR
 
 import utils.data_utils as data_utils
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 import numpy as np
 
 import csv
@@ -54,7 +54,7 @@ def getFECoreMetrics(systemName):
 
 	for i, m in enumerate(methods):
 		# Get the number of co-occurrences with methods of its own class (i.e., the embedding class)
-		idx_embedding_class = classToIndexMap[entity_utils.getEmbeddingClass(m)]
+		idx_embedding_class = classToIndexMap[java_utils.getEmbeddingClass(m)]
 		nb_co_occ_ec = 0.5 if co_occ_matrix[i, idx_embedding_class] == 0 else co_occ_matrix[i, idx_embedding_class]
 
 		co_occ_matrix[i,:] = co_occ_matrix[i,:]/nb_co_occ_ec
@@ -84,7 +84,7 @@ def getHistory(systemName, granularity):
 	'''
 
 	dirName = {"C": "class_changes", "M": "method_changes"}
-	historyFile = os.path.join(ROOT_DIR, 'data', 'history', dirName[granularity], systemName + '.csv')
+	historyFile = os.path.join(ROOT_DIR, 'approaches', 'hist', 'history', dirName[granularity], systemName + '.csv')
 	with open(historyFile, 'rb') as csvfile:
 		reader = csv.DictReader(csvfile, delimiter=';')
 		rawHistory = [{key: row[key] for key in row} for row in reader]
@@ -143,7 +143,7 @@ def getCoOccurrenceMatrix(systemName):
 				# Get the classes where these "other methods" are implemented
 				klasses = []
 				for m in coMethods:
-					embeddingClass = entity_utils.getEmbeddingClass(m)
+					embeddingClass = java_utils.getEmbeddingClass(m)
 					if embeddingClass in classes:
 						klasses.append(embeddingClass)
 

approaches/incode/core_metrics.py

@@ -1,14 +1,14 @@
 from context import ROOT_DIR
 
 import utils.data_utils as data_utils
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 
 import csv
 import os
 
 
 def getFECoreMetrics(systemName):
-	incodeMetricsFile = os.path.join(ROOT_DIR, 'data', 'metric_files', 'incode', systemName + '.csv')
+	incodeMetricsFile = os.path.join(ROOT_DIR, 'approaches', 'incode', 'metric_files', systemName + '.csv')
 
 	dictionnary = {e:[0., 0., 0.] for e in data_utils.getEntities('feature_envy', systemName)}
 
@@ -29,7 +29,7 @@ def getFECoreMetrics(systemName):
 
 			if (currentMethodName != methodName):
 				classToMetricMap = getClassToMetricMap(currentClassName, classAttributeMap)
-				normMethodName = entity_utils.normalizeMethodName(currentMethodName)
+				normMethodName = java_utils.normalizeMethodName(currentMethodName)
 				for klass in classToMetricMap:
 					entityName = normMethodName + ';' + klass
 					if entityName in dictionnary:
@@ -48,7 +48,7 @@ def getFECoreMetrics(systemName):
 				classAttributeMap[row['DeclaringClass']] = int(row['NbFields'])
 
 				classToMetricMap = getClassToMetricMap(currentClassName, classAttributeMap)
-				normMethodName = entity_utils.normalizeMethodName(currentMethodName)
+				normMethodName = java_utils.normalizeMethodName(currentMethodName)
 				for klass in classToMetricMap:
 					entityName = normMethodName + ';' + klass
 					if entityName in dictionnary:

approaches/jdeodorant/core_metrics.py

@@ -1,13 +1,13 @@
 from context import ROOT_DIR
 
 import utils.data_utils as data_utils
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 
 import csv
 import os
 
 def getGCCoreMetrics(systemName):
-	JDMetricsFile = os.path.join(ROOT_DIR, 'data', 'metric_files', 'jdeodorant', 'god_class_output', systemName + '.txt')
+	JDMetricsFile = os.path.join(ROOT_DIR, 'approaches', 'jdeodorant', 'metric_files', 'god_class_output', systemName + '.txt')
 
 	dictionnary = {c: [0] for c in data_utils.getClasses(systemName)}
 	with open(JDMetricsFile, 'r') as file:
@@ -21,7 +21,7 @@ def getGCCoreMetrics(systemName):
 
 
 def getFECoreMetrics(systemName):
-	JDMetricsFile = os.path.join(ROOT_DIR, 'data', 'metric_files', 'jdeodorant', 'feature_envy_metrics', systemName + '.csv')
+	JDMetricsFile = os.path.join(ROOT_DIR, 'approaches', 'jdeodorant', 'metric_files', 'feature_envy_metrics', systemName + '.csv')
 
 	dictionnary = {e:[0., 0., 0.] for e in data_utils.getEntities('feature_envy', systemName)}
 	with open(JDMetricsFile, 'rb') as metricfile:
@@ -34,7 +34,7 @@ def getFECoreMetrics(systemName):
 		method = data[0]['Method']
 		for i, line in enumerate(data):
 			if method != line['Method']:
-				normMethodName = entity_utils.normalizeMethodName(method)
+				normMethodName = java_utils.normalizeMethodName(method)
 				for tc in targetClasses:
 					entityName = normMethodName + ';' + tc['name']
 
@@ -53,14 +53,14 @@ def getFECoreMetrics(systemName):
 				distanceToEnclosingClass = 1.0
 				method = line['Method']
 
-			if entity_utils.getEmbeddingClass(entity_utils.normalizeMethodName(method)) == line['TargetClass']:
+			if java_utils.getEmbeddingClass(java_utils.normalizeMethodName(method)) == line['TargetClass']:
 				nbAccessToEnclosingClass = float(line['NbAccessedEntities'])
 				distanceToEnclosingClass = float(line['Distance_TC'])
 			else:
 				targetClasses.append({'name':line['TargetClass'], 'nbAccess':float(line['NbAccessedEntities']), 'distance':float(line['Distance_TC'])})
 
 			if i == len(data)-1:
-				normMethodName = entity_utils.normalizeMethodName(method)
+				normMethodName = java_utils.normalizeMethodName(method)
 				for tc in targetClasses:
 					entityName = normMethodName + ';' + tc['name']
 
@@ -74,14 +74,14 @@ def getFECoreMetrics(systemName):
 					if entityName in dictionnary:
 						dictionnary[entityName] = [nbAccessMetric, distanceMetric, 0.0]
 
-	JDOutputFile = os.path.join(ROOT_DIR, 'data', 'metric_files', 'jdeodorant', 'feature_envy_output', systemName + '.txt')
+	JDOutputFile = os.path.join(ROOT_DIR, 'approaches', 'jdeodorant', 'metric_files', 'feature_envy_output', systemName + '.txt')
 	with open(JDOutputFile, 'r') as file:
 		i = 0
 		for line in file:
 			if i > 0:
 				method = line.split('\t')[1].replace('::', '.')
 				method = method.split(':')[0]
-				method = entity_utils.normalizeMethodName(method)
+				method = java_utils.normalizeMethodName(method)
 				targetClass = line.split('\t')[2]
 				entityName = method + ';' + targetClass;
 

approaches/jdeodorant/detection_feature_envy.py

@@ -2,7 +2,7 @@
 
 import utils.data_utils as data_utils
 import utils.detection_utils as detection_utils
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 
 import os
 
@@ -18,7 +18,7 @@ def detect(systemName):
 			if i > 0:
 				source_entity = line.split('\t')[1].replace('::', '.')
 				source_entity = source_entity.split(':')[0]
-				source_entity = entity_utils.normalizeMethodName(source_entity)
+				source_entity = java_utils.normalizeMethodName(source_entity)
 				target_class = line.split('\t')[2]
 
 				if source_entity in methods:

data_construction/oracle_feature_envy/createOracle.py

@@ -1,6 +1,6 @@
 from context import ROOT_DIR
 
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 import numpy as np
 
 import csv
@@ -62,7 +62,7 @@ def getScore(answers):
 	return np.mean(np.array(floatValues))
 
 def getAnswers():
-	answerDir = os.path.join(ROOT_DIR, 'data_construction/oracle_feature_envy/forms_answers')
+	answerDir = os.path.join(ROOT_DIR, 'data_construction', 'oracle_feature_envy','forms_answers')
 
 	answers = []
 	for path, dirs, files in os.walk(answerDir):
@@ -74,14 +74,14 @@ def getAnswers():
 	return answers
 
 def getCandidates():
-	candidateDir = os.path.join(ROOT_DIR, 'data_construction/oracle_feature_envy/candidate_set/')
+	candidateDir = os.path.join(ROOT_DIR, 'data_construction', 'oracle_feature_envy' 'candidate_set')
 
 	candidates = []
 	for path, dirs, files in os.walk(candidateDir):
 		for file in fnmatch.filter(sorted(files), '*.csv'):
 			with open(os.path.join(path, file), 'r') as csvFile:
 				reader = csv.DictReader(csvFile)
-				candidates += [row['EMBED_CLASS'] + '.' + entity_utils.normalizeMethodName(row['METHOD']) + ';' + row['ENVIED_CLASS'] for row in reader]
+				candidates += [row['EMBED_CLASS'] + '.' + java_utils.normalizeMethodName(row['METHOD']) + ';' + row['ENVIED_CLASS'] for row in reader]
 
 	return candidates
 
@@ -95,7 +95,7 @@ def getCandidates():
 		smells = [candidates[i] for i in range(startIndex, endIndex) if getScore(answers[i]) >= 0.5]
 		startIndex = endIndex
 
-		labelFile = os.path.join(ROOT_DIR, 'data/antipatterns/feature_envy/' + system['name'] + '.txt')
+		labelFile = os.path.join(ROOT_DIR, 'data', 'antipatterns', 'feature_envy', system['name'] + '.txt')
 
 		with open(labelFile, 'w') as file:
 			for smell in smells:

data_construction/repository_miner/history_extractor.py

@@ -1,6 +1,6 @@
 from context import ROOT_DIR
 
-import utils.entity_utils as entity_utils
+import utils.java_utils as java_utils
 
 import javalang
 import os
@@ -61,21 +61,21 @@ def createHistoryFile(self, historyFilePath, granularity):
 
 
 	def __getClassChange(self, SHA, date, filePath, changeType):
-		directory = entity_utils.getDirectory(filePath)
+		directory = java_utils.getDirectory(filePath)
 		if directory in self.package_dirs_dictionary:
 			package = self.package_dirs_dictionary[directory]
 		else:
 			try:
 				if os.path.isfile(filePath):
-					package = entity_utils.getPackage(filePath)
+					package = java_utils.getPackage(filePath)
 				else:
 					if changeType == 'D':
 						self.__updateWorkingFile("../previousFile.java", filePath, SHA + "^")
-						package = entity_utils.getPackage("../previousFile.java")
+						package = java_utils.getPackage("../previousFile.java")
 
 					else:
 						self.__updateWorkingFile("../actualFile.java", filePath, SHA)
-						package = entity_utils.getPackage("../actualFile.java")
+						package = java_utils.getPackage("../actualFile.java")
 			except (javalang.tokenizer.LexerError, javalang.parser.JavaSyntaxError, AttributeError):
 				package = directory
 				self.exceptionDirs.append(directory)
@@ -85,12 +85,12 @@ def __getClassChange(self, SHA, date, filePath, changeType):
 				self.package_dirs_dictionary[directory] = package
 
 
-		line = SHA + ';' + date + ';' + package + '.' + entity_utils.getClassName(filePath) + ';' + changeType + '\n'
+		line = SHA + ';' + date + ';' + package + '.' + java_utils.getClassName(filePath) + ';' + changeType + '\n'
 		return line
 
 
 	def __getMethodeChange(self, SHA, date, filePath, changeType):
-		directory = entity_utils.getDirectory(filePath)
+		directory = java_utils.getDirectory(filePath)
 		if directory in self.package_dirs_dictionary:
 			package = self.package_dirs_dictionary[directory]
 		else:
@@ -101,7 +101,9 @@ def __getMethodeChange(self, SHA, date, filePath, changeType):
 			self.__updateWorkingFile("../actualFile.java", filePath, SHA)
 			self.__updateWorkingFile("../previousFile.java", filePath, SHA + "^")
 
-			diffjCommand = "java -jar " + ROOT_DIR + "/assets/jar/diffj-1.6.3.jar --brief ../previousFile.java ../actualFile.java"
+			diffj_jar = os.path.join(ROOT_DIR, 'assets', 'jar', 'diffj-1.6.3.jar')
+
+			diffjCommand = "java -jar " + diffj_jar + " --brief ../previousFile.java ../actualFile.java"
 			ps = subprocess.Popen(diffjCommand.split(), stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 			output, error = ps.communicate()
 
@@ -113,9 +115,9 @@ def __getMethodeChange(self, SHA, date, filePath, changeType):
 
 				if (ct == "ADDED") | (ct == "BODY_MODIFIED"):
 					if actual_method_class_dictionary is None:
-						actual_method_class_dictionary = entity_utils.getMethodClassDictionary("../actualFile.java")
+						actual_method_class_dictionary = java_utils.getMethodClassDictionary("../actualFile.java")
 
-					method = entity_utils.normalizeMethodName(method)
+					method = java_utils.normalizeMethodName(method)
 					if (method in actual_method_class_dictionary):
 						method = actual_method_class_dictionary[method] + '.' + method
 
@@ -125,9 +127,9 @@ def __getMethodeChange(self, SHA, date, filePath, changeType):
 
 				if ct == "DELETED":
 					if previous_method_class_dictionary is None:
-						previous_method_class_dictionary = entity_utils.getMethodClassDictionary("../previousFile.java")
+						previous_method_class_dictionary = java_utils.getMethodClassDictionary("../previousFile.java")
 
-					method = entity_utils.normalizeMethodName(method)
+					method = java_utils.normalizeMethodName(method)
 					if (method in previous_method_class_dictionary):
 						method = previous_method_class_dictionary[method] + '.' + method