Merge pull request #10 from tahiat/issue9

honouring Specimin local path, error msg printing in log. PR for issue #9

TestMain.py

@@ -8,10 +8,17 @@ class TestMain(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        # cloning the specimin 
-        main.clone_repository('https://github.com/kelloggm/specimin.git', 'resources')
-        cls.json_data = main.read_json_from_file('resources/test_data.json')[0]
-        cls.specimin_dir = "resources/specimin"
+        #TODO: previously used cf-1291 (index 0) fails when absolute path of the target is used. Will investigate it later. 
+        #Currently using cf-6060 
+        cls.json_data = main.read_json_from_file('resources/test_data.json')[3]
+        sp_env_var = main.get_specimin_env_var()
+        if sp_env_var is not None and os.path.exists(sp_env_var) and os.path.isdir(sp_env_var):
+            print("Local Specimin copy is being used")
+            cls.specimin_dir = sp_env_var
+        else:
+            print("Local Specimin not found. Cloning a Specimin copy")
+            main.clone_repository('https://github.com/kelloggm/specimin.git', 'resources')
+            cls.specimin_dir = os.path.abspath("resources/specimin")
 
     @classmethod
     def tearDownClass(cls):
@@ -20,7 +27,7 @@ def tearDownClass(cls):
             shutil.rmtree('resources/specimin')
         except Exception as e:
             print(f"Error occurred {e}")
-        # removing any issue project cloned in resources
+        # removing any issue project cloned in resources   
         for root, dirs, files in os.walk('resources', topdown=False):
             for dir_name in dirs:
                 if 'cf-' in dir_name:
@@ -56,9 +63,9 @@ def test_build_specimin_command(self):
                     "file": "IndexMode.java",
                     "package": 'org.apache.cassandra.index.sasi.conf'
                    }]
-        specimin_dir = 'user/specimin'
-        target_dir = 'user/ISSUES/cf-6077'
-        command = main.build_specimin_command(proj_name, target_dir, specimin_dir, root, targets)
+
+        target_dir = '/user/ISSUES/cf-6077'
+        command = main.build_specimin_command(proj_name, target_dir, root, targets)
         target_command = ''
         with open('resources/specimin_command_cf-6077.txt','r') as file:
             target_command = file.read()
@@ -71,9 +78,9 @@ def test_build_specimin_command(self):
                     "file": "Avro2Confluent.java",
                     "package": 'com.fillmore_labs.kafka.sensors.serde.confluent.interop'
                    }]
-        specimin_dir = 'user/specimin'
-        target_dir = 'user/ISSUES/cf-6019'
-        command = main.build_specimin_command(proj_name, target_dir, specimin_dir, root, targets)
+
+        target_dir = '/user/ISSUES/cf-6019'
+        command = main.build_specimin_command(proj_name, target_dir, root, targets)
         with open('resources/specimin_command_cf-6019.txt','r') as file:
             target_command = file.read()
         self.assertEqual(command, target_command)
@@ -82,15 +89,16 @@ def test_build_specimin_command(self):
         # make 
         issue_name = self.json_data[JsonKeys.ISSUE_ID.value]
         main.create_issue_directory('resources', issue_name)
-        self.assertTrue(os.path.exists('resources/cf-1291/input'))
-        main.clone_repository(self.json_data[JsonKeys.URL.value], f"resources/{issue_name}/input")  
-
-        project_name = main.get_repository_name(self.json_data[JsonKeys.URL.value])
+        self.assertTrue(os.path.exists(f'resources/{issue_name}/input'))
+        main.clone_repository(self.json_data[JsonKeys.URL.value], f"resources/{issue_name}/input") 
 
+        project_name = main.get_repository_name(self.json_data[JsonKeys.URL.value])
         self.assertTrue(main.checkout_commit(self.json_data[JsonKeys.COMMIT_HASH.value],f"resources/{issue_name}/input/{project_name}"))
         self.assertTrue(main.is_git_directory(f"resources/{issue_name}/input/{project_name}")) 
+        main.change_branch(self.json_data[JsonKeys.BRANCH.value], f"resources/{issue_name}/input/{project_name}")
 
-        command = main.build_specimin_command(project_name, f"resources/{issue_name}", self.specimin_dir, self.json_data[JsonKeys.ROOT_DIR.value], self.json_data[JsonKeys.TARGETS.value])
+        #build_specimin_command accept absolute path of the root directory.
+        command = main.build_specimin_command(project_name, os.path.abspath(f"resources/{issue_name}"), self.json_data[JsonKeys.ROOT_DIR.value], self.json_data[JsonKeys.TARGETS.value])
         print(command)
         result = main.run_specimin(issue_name, command, self.specimin_dir)
         self.assertEqual(result.status, "PASS")
@@ -103,11 +111,10 @@ def test_run_specimin(self):
                     "file": "Simple.java",
                     "package": "com.example"
                    }]
-        specimin_dir = 'resources/specimin'
         target_dir = 'resources/onefilesimple'
 
-        command = main.build_specimin_command(proj_name, target_dir, specimin_dir, root, targets)
-        result = main.run_specimin(proj_name, command, 'resources/specimin')
+        command = main.build_specimin_command(proj_name, os.path.abspath(target_dir), root, targets)
+        result = main.run_specimin(proj_name, command, self.specimin_dir)
         self.assertEqual(result.status, "PASS")
 
 

main.py

@@ -12,6 +12,7 @@
 specimin_project_name = 'specimin'
 specimin_source_url = 'https://github.com/kelloggm/specimin.git'
 TIMEOUT_DURATION = 300
+specimin_env_var = "SPECIMIN"
 
 def read_json_from_file(file_path):
     '''
@@ -34,6 +35,14 @@ def read_json_from_file(file_path):
         print(f"File not found: {file_path}")
         return None
 
+def get_specimin_env_var():
+    '''
+    Check and returns the path of the Specimin program if defined
+    Retruns:
+        Path of the local Specimin program
+    '''
+    specimin_env_value = os.environ.get(specimin_env_var)
+    return specimin_env_value
 
 def get_repository_name(github_ssh: str):
     '''
@@ -60,8 +69,8 @@ def create_issue_directory(issue_container_dir, issue_id):
     |    |--- output 
 
     Parameters: 
-        issue_container_dir (str): The directory where new directory is created
-        issue_id (str): Name of the directory to be created
+        issue_container_dir (str): Absolute path of the container directory containing all program directory.
+        issue_id (str): Name of the directory to be created inside container directory.
 
     Returns:
         specimin_input_dir (str): A target directory of SPECIMIN. (issue_container_dir/issue_id/input) 
@@ -99,7 +108,7 @@ def clone_repository(url, directory):
         directory (str): directory to clone in
     '''
     project_name = get_repository_name(url)
-    if (os.path.exists(f"{directory}/{project_name}")):
+    if (os.path.exists(os.path.join(directory, project_name))):
         print(f"{project_name} repository already exists. Aborting cloning")
         return
     subprocess.run(["git", "clone", url], cwd=directory)
@@ -128,9 +137,11 @@ def checkout_commit(commit_hash, directory):
     if not is_git_directory(directory):
         raise ValueError(f"{directory} is not a valid git directory")
 
-    command = ["git", "checkout", commit_hash]
-    result = subprocess.run(command, cwd=directory, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-
+    if (commit_hash):
+        command = ["git", "checkout", commit_hash]
+        result = subprocess.run(command, cwd=directory, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    else: 
+        return True
     if result.returncode == 0:
         print(f"Successfully checked-out commit {commit_hash} in {directory}")
     else:
@@ -163,8 +174,7 @@ def clone_specimin(path_to_clone, url):
 
 
 def build_specimin_command(project_name: str,
-                           issue_input_dir: str,
-                           specimin_dir: str, 
+                           target_base_dir_path: str,
                            root_dir: str,  
                            targets: list):
     '''
@@ -180,19 +190,19 @@ def build_specimin_command(project_name: str,
     
     Parameters:
         project_name (str): Name of the target project. Example: daikon
-        issue_input_dir (str): path of the target project directory. Ex: ISSUES/cf-1291
-        specimin_dir (str): Specimin directory path
+        target_base_dir (str): path of the target project directory. Ex: ISSUES/cf-1291
         root_dir (str): A directory path relative to the project base directory where java package stored.
         targets ({'method': '', 'file': '', 'package': ''}) : target java file and method name data
     
     Retruns:
         command (str): The gradle command of SPECIMIN for the issue.
     '''
+
+    if not os.path.isabs(target_base_dir_path):
+        raise ValueError("Invalid argument: target_base_dir_path must be an absolute path")
 
-    relative_path_of_target_dir = os.path.relpath(issue_input_dir, specimin_dir)
-
-    output_dir = os.path.join(relative_path_of_target_dir, specimin_output)
-    root_dir = os.path.join(relative_path_of_target_dir, specimin_input, project_name, root_dir)
+    output_dir = os.path.join(target_base_dir_path, specimin_output)
+    root_dir = os.path.join(target_base_dir_path, specimin_input, project_name, root_dir)
     root_dir = root_dir.rstrip('/') + os.sep
 
     target_file_list = []
@@ -243,20 +253,30 @@ def run_specimin(issue_name, command, directory) -> Result:
         directory (str): The base directory of the specimin repository
     
     Returns: 
-        boolean: True/False based on successful execution of SPECIMIN
+        Result: execution result of Specimin
     '''
+    print(f"{issue_name} executing...")
     try:
         result = subprocess.run(command, cwd=directory, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True, timeout=TIMEOUT_DURATION)
+        print(f"{issue_name} execution ends.")
         if result.returncode == 0:
             return Result(issue_name, "PASS", "")
         else:
-            error_msg_file = os.path.join(issue_folder_dir, issue_name, f"{issue_name}_error.txt")
-            if os.path.exists(error_msg_file):
-                os.remove(error_msg_file)
-            with open(error_msg_file, 'w') as file:
-                file.write(result.stderr.decode("utf-8"))
+            error_msg_file = os.path.join(issue_folder_dir, issue_name, f"{issue_name}_error.txt") # not abs path. ISSUES/cf-1291/cf-1291_error.txt
+            try:
+                stderr_str = result.stderr.decode("utf-8") # this can fail.
+                stderr_lines = stderr_str.split('\n')[:5]
+                first_five_lines_stderr = '\n'.join(stderr_lines)
+                print(first_five_lines_stderr)
+                if os.path.exists(error_msg_file):
+                    os.remove(error_msg_file)
+                with open(error_msg_file, 'w') as file:
+                    file.write(stderr_str)
+            except UnicodeDecodeError as e:
+                 print("Error decoding stderr:", e)
             return Result(issue_name, "FAIL", f"{error_msg_file}")
     except subprocess.TimeoutExpired:
+        print(f"{issue_name} execution ends. TIMEOUT")
         return Result(issue_name, "FAIL", "Timeout")
     except Exception as e:
         return Result(issue_name, "FAIL", f"Unhandled exception occurred: {e}")
@@ -276,19 +296,27 @@ def performEvaluation(issue_data) -> Result:
     branch = issue_data[JsonKeys.BRANCH.value]
     commit_hash = issue_data[JsonKeys.COMMIT_HASH.value]
 
-    input_dir = create_issue_directory(issue_folder_dir, issue_id) # ../cf-12/input
-    clone_repository(url, input_dir)  # TODO: check if clonning is successful.
+    issue_folder_abs_dir = os.path.abspath(issue_folder_dir)
+    input_dir = create_issue_directory(issue_folder_abs_dir, issue_id)
+    clone_repository(url, input_dir) 
     repo_name = get_repository_name(url)
 
     if branch:
-        change_branch(branch, f"{input_dir}/{repo_name}")  
+        change_branch(branch, os.path.join(input_dir, repo_name))  
 
     if commit_hash:
-        checkout_commit(commit_hash, f"{input_dir}/{repo_name}")
-
-    specimin_command = build_specimin_command(repo_name, os.path.join(issue_folder_dir, issue_id), os.path.join(issue_folder_dir, specimin_project_name),issue_data[JsonKeys.ROOT_DIR.value], issue_data[JsonKeys.TARGETS.value])
-
-    result = run_specimin(issue_id ,specimin_command, os.path.join(issue_folder_dir, specimin_project_name))
+        checkout_commit(commit_hash, os.path.join(input_dir, repo_name))
+
+    specimin_command = ""
+    result: Result = None
+    specimin_path = get_specimin_env_var()
+    if specimin_path is not None and os.path.exists(specimin_path):
+        specimin_command = build_specimin_command(repo_name, os.path.join(issue_folder_abs_dir, issue_id), issue_data[JsonKeys.ROOT_DIR.value], issue_data[JsonKeys.TARGETS.value])
+        result = run_specimin(issue_id ,specimin_command, specimin_path)
+    else:
+        specimin_command = build_specimin_command(repo_name, os.path.join(issue_folder_abs_dir, issue_id),issue_data[JsonKeys.ROOT_DIR.value], issue_data[JsonKeys.TARGETS.value])
+        result = run_specimin(issue_id ,specimin_command, os.path.join(issue_folder_abs_dir, specimin_project_name))
+
     print(f"{result.name} - {result.status}")
     return result
 
@@ -298,16 +326,25 @@ def main():
     Main method of the script. It iterates over the json data and perform minimization for each cases.   
     '''
     os.makedirs(issue_folder_dir, exist_ok=True)   # create the issue holder directory
-    clone_specimin(issue_folder_dir, specimin_source_url)
+
+    specimin_path = get_specimin_env_var()
+    if specimin_path is not None and os.path.exists(specimin_path) and os.path.isdir(specimin_path):
+        print("Local Specimin copy is being used")
+    else:
+        print("Local Specimin not found. Cloning a Specimin copy")
+        clone_specimin(issue_folder_dir, specimin_source_url)
 
     json_file_path = 'resources/test_data.json'
     parsed_data = read_json_from_file(json_file_path)
 
     evaluation_results = []
     if parsed_data:
         for issue in parsed_data:
+            issue_id = issue["issue_id"]
+            print(f"{issue_id} execution starts =========>")
             result = performEvaluation(issue)
             evaluation_results.append(result)
+            print((f"{issue_id} <========= execution Ends."))
 
 
     report_generator = TableGenerator(evaluation_results)

resources/specimin_command_cf-6019.txt

@@ -1 +1 @@
-./gradlew run --args='--root "../ISSUES/cf-6019/input/kafka-sensors/src/main/java/" --outputDirectory "../ISSUES/cf-6019/output" --targetFile "com/fillmore_labs/kafka/sensors/serde/confluent/interop/Avro2Confluent.java" --targetMethod "com.fillmore_labs.kafka.sensors.serde.confluent.interop.Avro2Confluent#transform(String, byte[])"'
+./gradlew run --args='--root "/user/ISSUES/cf-6019/input/kafka-sensors/src/main/java/" --outputDirectory "/user/ISSUES/cf-6019/output" --targetFile "com/fillmore_labs/kafka/sensors/serde/confluent/interop/Avro2Confluent.java" --targetMethod "com.fillmore_labs.kafka.sensors.serde.confluent.interop.Avro2Confluent#transform(String, byte[])"'

resources/specimin_command_cf-6077.txt

@@ -1 +1 @@
-./gradlew run --args='--root "../ISSUES/cf-6077/input/cassandra/src/java/" --outputDirectory "../ISSUES/cf-6077/output" --targetFile "org/apache/cassandra/index/sasi/conf/IndexMode.java" --targetMethod "org.apache.cassandra.index.sasi.conf.IndexMode#getMode(ColumnMetadata, Map<String, String>)"'
+./gradlew run --args='--root "/user/ISSUES/cf-6077/input/cassandra/src/java/" --outputDirectory "/user/ISSUES/cf-6077/output" --targetFile "org/apache/cassandra/index/sasi/conf/IndexMode.java" --targetMethod "org.apache.cassandra.index.sasi.conf.IndexMode#getMode(ColumnMetadata, Map<String, String>)"'