From 3740c89ed69dd3b9adb0b4bcb660ab4db6e856b3 Mon Sep 17 00:00:00 2001
From: Karma Riuk <riukkarma@gmail.com>
Date: Sat, 1 Mar 2025 18:29:04 +0100
Subject: [PATCH] added extraction of number of tests (both from output and
 grep)

---
 clone_repos.py |  1 +
 handlers.py    | 46 +++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 46 insertions(+), 1 deletion(-)

diff --git a/clone_repos.py b/clone_repos.py
index 97c74b3..3391b21 100644
--- a/clone_repos.py
+++ b/clone_repos.py
@@ -171,6 +171,7 @@ def save_df_with_updates(df, updates_list, verbose=False):
         n_tests_with_grep=None,
         n_tests_passed=None,
         n_tests_failed=None,
+        n_tests_errors=None,
         n_tests_skipped=None,
         good_repo_for_crab=None,
         error_msg=None,
diff --git a/handlers.py b/handlers.py
index ba03c57..e0eced1 100644
--- a/handlers.py
+++ b/handlers.py
@@ -1,5 +1,5 @@
 from abc import ABC, abstractmethod
-import os, re, docker
+import os, re, docker, subprocess
 
 USER_ID = os.getuid() # for container user
 GROUP_ID = os.getgid() 
@@ -80,6 +80,17 @@ class BuildHandler(ABC):
         self.updates["tested_successfully"] = True
         self.updates["error_msg"] = output
 
+        self.extract_test_numbers(output)
+
+        grep_cmd = f"grep -r --include='*.java' -E '@Test|@ParameterizedTest' {self.path} | wc -l" # NOTE: After inspection, this is an upper bound, since comments get matched 
+        try:
+            result = subprocess.run(grep_cmd, shell=True, capture_output=True, text=True, check=True)
+            test_count = result.stdout.strip()
+        except subprocess.CalledProcessError as e:
+            test_count = "-1"  # Default to 0 if grep command fails
+
+        self.updates["n_tests_with_grep"] = test_count
+
         return True
 
     def clean_repo(self) -> None:
@@ -94,6 +105,10 @@ class BuildHandler(ABC):
     def test_cmd(self) -> str:
         pass
 
+    @abstractmethod
+    def extract_test_numbers(self, output: str) -> None:
+        pass
+
     @abstractmethod
     def clean_cmd(self) -> str:
         pass
@@ -122,6 +137,28 @@ class MavenHandler(BuildHandler):
     def container_name(self) -> str:
         return "crab-maven"
 
+    def extract_test_numbers(self, output: str) -> None:
+        # NOTE: I'ma afraid this might be specific for junit and wouldn't work for other testing frameworks
+        pattern = r"\[INFO\] Results:\n\[INFO\]\s*\n\[INFO\] Tests run: (\d+), Failures: (\d+), Errors: (\d+), Skipped: (\d+)"
+
+        matches = re.findall(pattern, output)
+
+        self.updates["n_tests"] = 0
+        self.updates["n_tests_passed"] = 0  # Passed tests = Tests run - (Failures + Errors)
+        self.updates["n_tests_failed"] = 0
+        self.updates["n_tests_errors"] = 0
+        self.updates["n_tests_skipped"] = 0
+
+        for match in matches:
+            tests_run, failures, errors, skipped = map(int, match)
+            self.updates["n_tests"] += tests_run
+            self.updates["n_tests_failed"] += failures
+            self.updates["n_tests_errors"] += errors
+            self.updates["n_tests_skipped"] += skipped
+            self.updates["n_tests_passed"] += (tests_run - (failures + errors))  # Calculate passed tests
+
+        
+
 class GradleHandler(BuildHandler):
     def __init__(self, repo_path: str, build_file: str, updates: dict) -> None:
         super().__init__(repo_path, build_file, updates)
@@ -139,6 +176,13 @@ class GradleHandler(BuildHandler):
     def container_name(self) -> str:
         return "crab-gradle"
 
+    def extract_test_numbers(self, output: str) -> None:
+        self.updates["n_tests"] = -1
+        self.updates["n_tests_passed"] = -1
+        self.updates["n_tests_failed"] = -1
+        self.updates["n_tests_errors"] = -1
+        self.updates["n_tests_skipped"] = -1
+
 def merge_download_lines(lines: list) -> list:
     """
     Merges lines that are part of the same download block in Maven output.