From 19ea8c91e7459756c763c16f70081fbab5de2ece Mon Sep 17 00:00:00 2001
From: Benedikt Schesch <37979523+benedikt-schesch@users.noreply.github.com>
Date: Fri, 7 Jun 2024 21:58:36 -0700
Subject: [PATCH] Make reproduction of results easier (#306)

---
 Makefile                      | 21 ++++++++++++---------
 run.sh                        | 24 ++++++++++++++++++++----
 src/python/test_repo_heads.py | 13 +------------
 3 files changed, 33 insertions(+), 25 deletions(-)

diff --git a/Makefile b/Makefile
index bdeb9571e4..5a407b2550 100644
--- a/Makefile
+++ b/Makefile
@@ -61,6 +61,12 @@ compress-cache:
 	if [ -f cache.tar.gz ]; then rm -f cache.tar.gz; fi
 	tar --exclude="lock" -czf cache.tar.gz cache
 
+# Compresses the cache without logs.
+compress-cache-without-logs:
+	if [ ! -d cache ]; then echo "cache does not exist"; exit 1; fi
+	if [ -f cache_without_logs.tar.gz ]; then rm -f cache_without_logs.tar.gz; fi
+	tar --exclude="lock" --exclude="logs" -czf cache_without_logs.tar.gz cache
+
 compress-small-cache:
 	if [ ! -d cache-small ]; then echo "cache-small does not exist"; exit 1; fi
 	if [ -f cache-small.tar ]; then rm -f cache-small.tar; fi
@@ -72,6 +78,12 @@ decompress-cache:
 	if [ -d cache ]; then echo "cache already exists"; exit 1; fi
 	tar -xzf cache.tar.gz
 
+# Decompresses the cache without logs.
+decompress-cache-without-logs:
+	if [ ! -f cache_without_logs.tar.gz ]; then echo "cache_without_logs.tar.gz does not exist"; exit 1; fi
+	if [ -d cache ]; then echo "cache already exists"; exit 1; fi
+	tar -xzf cache_without_logs.tar.gz
+
 decompress-small-cache:
 	if [ ! -f cache-small.tar ]; then echo "cache-small.tar does not exist"; exit 1; fi
 	if [ -d cache-small ]; then echo "cache-small already exists"; exit 1; fi
@@ -166,12 +178,3 @@ tags:
 
 run:
 	nice -n 5 sh run_full.sh | tee output.txt
-
-# Create a tarball of the artifacts for the paper.
-# Keep this target last in the file.
-create-artifacts:
-	rm -rf artifacts
-	git clone https://github.com/benedikt-schesch/AST-Merging-Evaluation.git artifacts
-	rm -rf artifacts/.git
-	sed -i '' 's/benedikt-schesch/anonymous-github-user/g' artifacts/README.md artifacts/Makefile
-	tar -czf artifacts.tar.gz artifacts
diff --git a/run.sh b/run.sh
index 82311ba3dd..6fcf56e7bc 100755
--- a/run.sh
+++ b/run.sh
@@ -97,10 +97,26 @@ git submodule update --init --recursive
 GIT_CONFIG_GLOBAL=$(pwd)/.gitconfig
 export GIT_CONFIG_GLOBAL
 
-# Check if cache.tar exists and cache is missing
-if [ -f cache.tar ] && [ ! -d cache ]; then
-    echo "Decompressing cache.tar"
-    # make decompress-cache
+# Check if cache.tar.gz exists and cache is missing
+if [ -f cache.tar.gz ] && [ ! -d cache ]; then
+    read -r -p "cache.tar.gz found and cache directory missing. Do you want to decompress? (y/n) " answer
+    if [ "$answer" = "y" ]; then
+        echo "Decompressing cache.tar.gz"
+        make decompress-cache
+    else
+        echo "Decompression aborted."
+    fi
+fi
+
+# Check if cache_without_logs.tar.gz exists and cache is missing
+if [ -f cache_without_logs.tar.gz ] && [ ! -d cache_without_logs ]; then
+    read -r -p "cache_without_logs.tar.gz found and cache_without_logs directory missing. Do you want to decompress? (y/n) " answer
+    if [ "$answer" = "y" ]; then
+        echo "Decompressing cache_without_logs.tar.gz"
+        make decompress-cache-without-logs
+    else
+        echo "Decompression aborted."
+    fi
 fi
 
 mvn -v | head -n 1 | cut -c 14-18 | grep -q 3.9. || { echo "Maven 3.9.* is required"; mvn -v; echo "PATH=$PATH"; exit 1; }
diff --git a/src/python/test_repo_heads.py b/src/python/test_repo_heads.py
index e750c05637..da31c897aa 100755
--- a/src/python/test_repo_heads.py
+++ b/src/python/test_repo_heads.py
@@ -89,18 +89,7 @@ def head_passes_tests(args: Tuple[pd.Series, Path]) -> pd.Series:
     test_state, _, repo_info["head tree fingerprint"] = repo.checkout_and_test(
         repo_info["head hash"], timeout=TIMEOUT_TESTING_PARENT, n_tests=3
     )
-    if test_state == TEST_STATE.Tests_passed:
-        # Make sure the repo is cloned and exists
-        repo.clone_repo()
-        if not repo.repo_path.exists():
-            logger.error(
-                f"head_passes_tests: Repo {repo_slug} does not exist after passing tests."
-            )
-            raise Exception(
-                f"Repo {repo_slug} does not exist after passing tests. \
-                    (Either clone the repo or remove the cache entry)"
-            )
-    else:
+    if test_state != TEST_STATE.Tests_passed:
         shutil.rmtree(repo.repo_path, ignore_errors=True)
 
     repo_info["head test result"] = test_state.name