144 lines
4.6 KiB
Python
Executable File
144 lines
4.6 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
import javalang
|
|
import os
|
|
import pandas as pd
|
|
import glob
|
|
import warnings
|
|
import tabulate
|
|
|
|
DIR: str = os.path.dirname(os.path.realpath(__file__))
|
|
SOURCES: str = DIR + '/resources/xerces2-j-src'
|
|
OUT_DIR: str = DIR + '/feature_vectors'
|
|
IN_DIR: str = DIR + '/god_classes'
|
|
|
|
|
|
def clean_output():
|
|
filelist = glob.glob(OUT_DIR + '/*.csv')
|
|
for f in filelist:
|
|
os.remove(f)
|
|
|
|
|
|
def get_fields(java_class: javalang.tree.ClassDeclaration) -> set[str]:
|
|
names = set()
|
|
for f in java_class.fields:
|
|
names.add(f.declarators[0].name)
|
|
return names
|
|
|
|
|
|
def get_methods(java_class: javalang.tree.ClassDeclaration) -> set[str]:
|
|
names = set()
|
|
for f in java_class.methods:
|
|
names.add(f.name)
|
|
return names
|
|
|
|
|
|
def empty_qualifier(node: any) -> bool:
|
|
return node.qualifier is None or node.qualifier == '' or node.qualifier == 'this'
|
|
|
|
|
|
def get_fields_accessed_by_method(method: javalang.tree.MethodDeclaration, fields: set[str]) -> set[str]:
|
|
# Ignore formal parameters and local variable declarations
|
|
local_variables = set()
|
|
for _, node in method.filter(javalang.tree.LocalVariableDeclaration):
|
|
local_variables.add(node.declarators[0].name)
|
|
for _, node in method.filter(javalang.tree.FormalParameter):
|
|
local_variables.add(node.name)
|
|
|
|
nodes = set()
|
|
for _, node in method.filter(javalang.tree.MemberReference):
|
|
if empty_qualifier(node):
|
|
name = node.member
|
|
if name in fields:
|
|
nodes.add(name)
|
|
else:
|
|
# qualifier of 'a.b.c.x' is 'a.b.c', we only care about 'a'
|
|
name = node.qualifier.split('.', maxsplit=1)[0]
|
|
|
|
# cannot use field set here, since a class may be referenced (through a static property)
|
|
# hence we fetch the local variables (and parameters) in the method and we explicitly avoid them
|
|
if not name in local_variables:
|
|
|
|
# if a MemberReference includes a non empty qualifier (e.g., a.x),
|
|
# consider the qualifier (a), not the member (x)˝
|
|
nodes.add(name)
|
|
return nodes
|
|
|
|
|
|
def get_methods_accessed_by_method(method: javalang.tree.MethodDeclaration, methods: set[str]) -> set[str]:
|
|
nodes = set()
|
|
for _, node in method.filter(javalang.tree.MethodInvocation):
|
|
name: str = node.member
|
|
if empty_qualifier(node) and name in methods:
|
|
nodes.add(name)
|
|
return nodes
|
|
|
|
|
|
def parse(path: str, name: str, df_table):
|
|
# Get the AST of the file
|
|
with open(path) as file:
|
|
data = file.read()
|
|
tree = javalang.parse.parse(data)
|
|
|
|
# Fetch package name from package declaration
|
|
# if node is missing, assuming default package ('')
|
|
package_name = ''
|
|
for _, node in tree.filter(javalang.tree.PackageDeclaration):
|
|
package_name = node.name
|
|
break
|
|
|
|
for _, node in tree.filter(javalang.tree.ClassDeclaration):
|
|
|
|
# consider only the class matching the input file name, to skip inner classes
|
|
if path.endswith(node.name + '.java'):
|
|
fqdn = package_name + '.' + node.name
|
|
|
|
fields = get_fields(node)
|
|
methods = get_methods(node)
|
|
|
|
cols = sorted(fields.union(methods))
|
|
|
|
df = pd.DataFrame(columns=cols, dtype=int)
|
|
for m in node.methods:
|
|
# make sure method is included in csv file
|
|
df.loc[m.name, :] = 0
|
|
|
|
m_fields = get_fields_accessed_by_method(m, fields)
|
|
m_methods = get_methods_accessed_by_method(m, methods)
|
|
|
|
for member in m_fields.union(m_methods):
|
|
if member not in df:
|
|
df[member] = 0
|
|
df.loc[m.name, member] = 1
|
|
df = df.fillna(0)
|
|
|
|
for i in df.columns:
|
|
df[[i]] = df[[i]].astype(int)
|
|
|
|
df = df.loc[:, (df != 0).any(axis=0)]
|
|
|
|
df_table.loc[name, '# Feature Vectors'] = df.shape[0]
|
|
df_table.loc[name, '# Attributes'] = df.shape[1]
|
|
|
|
df.to_csv(OUT_DIR + '/' + fqdn + '.csv')
|
|
break
|
|
|
|
|
|
def main():
|
|
warnings.simplefilter(action='ignore', category=pd.errors.PerformanceWarning)
|
|
clean_output()
|
|
|
|
df = pd.read_csv(IN_DIR + '/god_classes.csv')
|
|
|
|
df_table = pd.DataFrame(columns=['# Feature Vectors', '# Attributes'])
|
|
|
|
for clazz in df['class_name'].to_list():
|
|
clazz_path = SOURCES + '/' + clazz.replace('.', '/') + '.java'
|
|
print(clazz_path)
|
|
parse(clazz_path, clazz, df_table)
|
|
|
|
df_table.index.name = "Class Name"
|
|
print(df_table.to_markdown())
|
|
|
|
if __name__ == '__main__':
|
|
main()
|