From 438deedabc602ed435822345d687530cf403185d Mon Sep 17 00:00:00 2001
From: FanYuliang <yf3@illinois.edu>
Date: Tue, 7 Dec 2021 17:37:56 -0600
Subject: [PATCH 1/2] squash commits

---
 auto-check-deleted/find_all_deleted.py | 69 ++++++++++++++++++++++++++
 auto-check-deleted/requirements.txt    |  2 +
 2 files changed, 71 insertions(+)
 create mode 100644 auto-check-deleted/find_all_deleted.py
 create mode 100644 auto-check-deleted/requirements.txt

diff --git a/auto-check-deleted/find_all_deleted.py b/auto-check-deleted/find_all_deleted.py
new file mode 100644
index 00000000..672062aa
--- /dev/null
+++ b/auto-check-deleted/find_all_deleted.py
@@ -0,0 +1,69 @@
+import pandas as pd
+import subprocess
+from glob import glob
+import os
+import shutil
+from tqdm import tqdm
+import json
+
+archieved = [
+	"https://github.com/gooddata/GoodData-CL",
+]
+
+df = pd.read_csv("../pr-data.csv")
+error_log = []
+test_file_deleted = []
+test_method_deleted = []
+base_dir = os.getcwd()
+
+df = df.groupby("Project URL").agg(lambda x: list(x))
+
+for url, row in tqdm(df.iterrows(), total=df.shape[0]):
+	if url in archieved:
+		continue
+	testname_list, status_list, note_list = row["Fully-Qualified Test Name (packageName.ClassName.methodName)"], row["Status"], row["Notes"]
+	print("url", url)
+
+	clone_cmd = "git clone --quiet {} tmp && cd tmp".format(url)
+	subprocess.run(clone_cmd, shell=True, stdout=open(os.devnull, 'wb'))
+	print("clone done!!!")
+	for full_testname, status, note in zip(testname_list, status_list, note_list):
+		testfile_name, test_method_name = full_testname.split(".")[-2], full_testname.split(".")[-1]
+		
+		if pd.isna(status) and pd.isna(note):
+			
+			if "[" in test_method_name or "\"" in test_method_name:
+				print("{}, {} is a PUT; skip".format(url, full_testname))
+				continue
+			if testfile_name[-4:] != "Test":
+				error_log.append("{}, {}".format(url, full_testname))
+				continue
+				
+			fnames = glob("**/{}.java".format(testfile_name), recursive=True)
+			if len(fnames) == 0:
+				# try to get the commit id
+				git_find_SHA = "cd tmp && git rev-list -n 1 HEAD -- **/{}.java".format(testfile_name)
+				commit_SHA = subprocess.check_output(git_find_SHA, shell=True).decode("utf-8")
+
+				if len(commit_SHA) > 0:
+					test_file_deleted.append("{}, {}, {}".format(url, full_testname, url+"/commit/"+commit_SHA))
+					print("[INFO]:detected Test File " + "{}, {} was deleted at SHA {}".format(url, full_testname, commit_SHA))
+				# test file not found; add to log
+				else:
+					test_file_deleted.append("{}, {}".format(url, full_testname))
+					print("[INFO]: detected deleted Test File: ", "{}, {}".format(url, full_testname))
+
+
+	# clean the repo and repeat
+	os.chdir(base_dir)
+	shutil.rmtree("tmp")
+
+
+with open("new_test_file_deleted.json", "w") as f:
+	json.dump(test_file_deleted, f)
+
+
+
+
+
+
diff --git a/auto-check-deleted/requirements.txt b/auto-check-deleted/requirements.txt
new file mode 100644
index 00000000..8c822a6e
--- /dev/null
+++ b/auto-check-deleted/requirements.txt
@@ -0,0 +1,2 @@
+pandas
+tqdm

From d0ae976483eaf4d24dff65697489afbb3ea838f7 Mon Sep 17 00:00:00 2001
From: FanYuliang <yf3@illinois.edu>
Date: Tue, 7 Dec 2021 19:42:20 -0600
Subject: [PATCH 2/2] update find_all_deleted.py

---
 auto-check-deleted/find_all_deleted.py | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/auto-check-deleted/find_all_deleted.py b/auto-check-deleted/find_all_deleted.py
index 672062aa..dd4dc0ee 100644
--- a/auto-check-deleted/find_all_deleted.py
+++ b/auto-check-deleted/find_all_deleted.py
@@ -7,9 +7,10 @@
 import json
 
 archieved = [
-	"https://github.com/gooddata/GoodData-CL",
+	"https://github.com/gooddata/GoodData-CL"
 ]
 
+
 df = pd.read_csv("../pr-data.csv")
 error_log = []
 test_file_deleted = []
@@ -19,10 +20,12 @@
 df = df.groupby("Project URL").agg(lambda x: list(x))
 
 for url, row in tqdm(df.iterrows(), total=df.shape[0]):
-	if url in archieved:
-		continue
+
 	testname_list, status_list, note_list = row["Fully-Qualified Test Name (packageName.ClassName.methodName)"], row["Status"], row["Notes"]
 	print("url", url)
+	if url in archieved:
+		print("this url has been archieved! skip")
+		continue
 
 	clone_cmd = "git clone --quiet {} tmp && cd tmp".format(url)
 	subprocess.run(clone_cmd, shell=True, stdout=open(os.devnull, 'wb'))
@@ -50,8 +53,7 @@
 					print("[INFO]:detected Test File " + "{}, {} was deleted at SHA {}".format(url, full_testname, commit_SHA))
 				# test file not found; add to log
 				else:
-					test_file_deleted.append("{}, {}".format(url, full_testname))
-					print("[INFO]: detected deleted Test File: ", "{}, {}".format(url, full_testname))
+					print("[INFO]: detected deleted Test File WITHOUT FINDING SHA: ", "{}, {}".format(url, full_testname))
 
 
 	# clean the repo and repeat