From 998a6e1625711b6954e729a7a6bbd193467793b3 Mon Sep 17 00:00:00 2001
From: Michael Ernst <mernst@cs.washington.edu>
Date: Thu, 16 May 2024 06:26:45 -0700
Subject: [PATCH] Get base, left, right, and programmer merge

---
 Makefile                   |  2 +-
 src/python/replay_merge.py | 64 ++++++++++++++++++++++++++++++++++++--
 src/python/repo.py         |  9 +++++-
 3 files changed, 70 insertions(+), 5 deletions(-)

diff --git a/Makefile b/Makefile
index b5f5159c2c..550074143b 100644
--- a/Makefile
+++ b/Makefile
@@ -134,7 +134,7 @@ clean-local:
 check-merges-reproducibility:
 	@echo "Running replay_merge for each idx in parallel using GNU Parallel..."
 	@set -e; \
-	tail -n +2 $(CSV_RESULTS) | awk -F, '{print $$1}' | parallel --halt now,fail=1 -j 50% python3 src/python/replay_merge.py --merges_csv $(CSV_RESULTS) -skip_build -delete_workdir --idx {}
+	tail -n +2 $(CSV_RESULTS) | awk -F, '{print $$1}' | parallel --halt now,fail=1 -j 50% python3 src/python/replay_merge.py -delete_workdir --idx {}
 
 protect-repos:
 	find repos -mindepth 1 -type d -exec chmod a-w {} +
diff --git a/src/python/replay_merge.py b/src/python/replay_merge.py
index 5722195b73..dcd2801086 100755
--- a/src/python/replay_merge.py
+++ b/src/python/replay_merge.py
@@ -12,6 +12,7 @@
 import tarfile
 from pathlib import Path
 import shutil
+import subprocess
 import pandas as pd
 from repo import Repository, MERGE_TOOL, TEST_STATE, MERGE_STATE
 from variables import TIMEOUT_TESTING_MERGE, N_TESTS, WORKDIR_DIRECTORY
@@ -84,6 +85,63 @@ def merge_replay(
             f"Replaying {repo_slug} {merge_data['left']} {merge_data['right']}",
             total=len(MERGE_TOOL),
         )
+
+        # Get base, left, right, and programmer merge.
+
+        workdir = Path(
+            f"{repo_slug}-merge-left"
+        )
+        if not (WORKDIR_DIRECTORY / workdir).exists():
+            repo = Repository(
+                repo_slug,
+                cache_directory=Path("no_cache/"),
+                workdir_id=workdir,
+                delete_workdir=False,
+                lazy_clone=False,
+            )
+            repo.checkout(merge_data['left'])
+
+        workdir = Path(
+            f"{repo_slug}-merge-right"
+        )
+        if not (WORKDIR_DIRECTORY / workdir).exists():
+            repo = Repository(
+                repo_slug,
+                cache_directory=Path("no_cache/"),
+                workdir_id=workdir,
+                delete_workdir=False,
+                lazy_clone=False,
+            )
+            repo.checkout(merge_data['right'])
+
+        workdir = Path(
+            f"{repo_slug}-merge-base"
+        )
+        if not (WORKDIR_DIRECTORY / workdir).exists():
+            repo = Repository(
+                repo_slug,
+                cache_directory=Path("no_cache/"),
+                workdir_id=workdir,
+                delete_workdir=False,
+                lazy_clone=False,
+            )
+            base_commit = subprocess.run(['git', 'merge-base', merge_data['left'], merge_data['right']], stdout=subprocess.PIPE).stdout.decode('utf-8')
+            repo.checkout(base_commit)
+
+        workdir = Path(
+            f"{repo_slug}-merge-programmer"
+        )
+        if not (WORKDIR_DIRECTORY / workdir).exists():
+            repo = Repository(
+                repo_slug,
+                cache_directory=Path("no_cache/"),
+                workdir_id=workdir,
+                delete_workdir=False,
+                lazy_clone=False,
+            )
+            repo.checkout(merge_data['merge'])
+
+
         for merge_tool in MERGE_TOOL:
             progress.update(task, advance=1)
             workdir = Path(
@@ -98,19 +156,19 @@ def merge_replay(
             if (WORKDIR_DIRECTORY / workdir).exists():
                 # Ask the user if they want to delete the workdir
                 logger.info(
-                    f"workdir {WORKDIR_DIRECTORY / workdir} already exists for idx: {merge_idx}"
+                    f"Workdir {WORKDIR_DIRECTORY / workdir} already exists for idx: {merge_idx}"
                 )
                 if delete_workdir:
                     answer = "y"
                 else:
                     answer = input(
-                        f"workdir {workdir} exists for idx: {merge_idx}. Delete it? (y/n)"
+                        f"Workdir {workdir} exists for idx: {merge_idx}. Delete it? (y/n)"
                     )
                 if answer == "y":
                     shutil.rmtree(WORKDIR_DIRECTORY / workdir)
                 else:
                     logger.info(
-                        f"workdir {WORKDIR_DIRECTORY/workdir} already exists. Skipping"
+                        f"Workdir {WORKDIR_DIRECTORY/workdir} already exists. Skipping."
                     )
                     continue
             try:
diff --git a/src/python/repo.py b/src/python/repo.py
index 3424647f23..b64c1867b4 100755
--- a/src/python/repo.py
+++ b/src/python/repo.py
@@ -215,6 +215,14 @@ def clone_repo(self) -> None:
         with lock:
             if self.repo_path.exists():
                 return
+            print(
+                "Cloning",
+                self.repo_slug,
+                "to",
+                self.repo_path,
+                "because:",
+                self.repo_path.exists(),
+            )
             try:
                 clone_repo(self.repo_slug, self.repo_path)
             except Exception as e:
@@ -242,7 +250,6 @@ def copy_repo(self) -> None:
             ignore_dangling_symlinks=True,
         )
         os.system("chmod -R 777 " + str(self.local_repo_path))
-
         self.repo = Repo(self.local_repo_path)
 
     def checkout(self, commit: str, use_cache: bool = True) -> Tuple[bool, str]: