From 4c7426d16b74249b5821e4c39b93b25c6139fda0 Mon Sep 17 00:00:00 2001
From: Michael Ernst <mernst@cs.washington.edu>
Date: Thu, 16 May 2024 11:38:42 -0700
Subject: [PATCH] Improvements to `replay_merge.py` and
 `select_from_results.py` (#293)

---
 src/python/replay_merge.py        | 58 +++++++++++++++++++++++++++++--
 src/python/repo.py                |  1 -
 src/python/select_from_results.py | 32 +++++++++++++----
 3 files changed, 80 insertions(+), 11 deletions(-)

diff --git a/src/python/replay_merge.py b/src/python/replay_merge.py
index 4fa2375fcc..adb6d24ebd 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
@@ -90,6 +91,57 @@ 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-input-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-input-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-input-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-input-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(
@@ -104,19 +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..176e4b67a4 100755
--- a/src/python/repo.py
+++ b/src/python/repo.py
@@ -242,7 +242,6 @@ 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]:
diff --git a/src/python/select_from_results.py b/src/python/select_from_results.py
index fe7ffbf60d..6c1f3b8733 100755
--- a/src/python/select_from_results.py
+++ b/src/python/select_from_results.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 """Output a subset of the results, to standard out.
-The arguments are a query and a list of columns.
+The arguments are a query and an optional list of columns.
 The query is executed (to select rows), then columns are output that include:
  * idx
  * all the columns that appear in the query
@@ -13,9 +13,14 @@
 Here are example invocations:
   select_from_results.py '(gitmerge_ort == "Merge_failed") and (spork != "Merge_failed")'
   select_from_results.py '(gitmerge_ort == "Merge_failed") != (spork == "Merge_failed")'
+
+The resulting .csv is useful for manual examination but cannot be passed to
+`replay_merge.py` because that requires a .csv file with all tools and all
+fingerprints.
 """
 
 import argparse
+import os
 from os import system
 import re
 import tempfile
@@ -25,9 +30,9 @@
 def columns_in_query(query):
     """Returns all the identifiers used in the query."""
     result = re.findall(r"""(?<!['"])\b[A-Za-z][A-Za-z_]*\b(?!['"])""", query)
-    if "and" in result:
+    while "and" in result:
         result.remove("and")
-    if "or" in result:
+    while "or" in result:
         result.remove("or")
     return result
 
@@ -44,8 +49,11 @@ def main():
         description="Outputs a subset of the results, to standard out",
     )
     parser.add_argument("query")
+    scriptdir = os.path.dirname(os.path.realpath(__file__))
     parser.add_argument(
-        "--input", action="store", default="../../results/combined/result.csv"
+        "--input",
+        action="store",
+        default=scriptdir + "/" + "../../results/combined/result.csv",
     )
     parser.add_argument("columns", nargs=argparse.REMAINDER)
     args = parser.parse_args()
@@ -57,16 +65,26 @@ def main():
 
     # Select some columns
     columns_to_select = (
-        ["idx", "repo-idx", "merge-idx", "branch_name", "merge", "left", "right"]
+        [
+            "idx",
+            "repo-idx",
+            "merge-idx",
+            "branch_name",
+            "merge",
+            "left",
+            "left_tree_fingerprint",
+            "right",
+            "right_tree_fingerprint",
+        ]
         + columns_in_query(args.query)
         + args.columns
+        + ["repository"]
     )
     df = df[columns_to_select]
 
     # Gross way to produce output to standard out
-    with tempfile.TemporaryFile() as tmpfile:
+    with tempfile.NamedTemporaryFile() as tmpfile:
         df.to_csv(tmpfile)
-        print(tmpfile.name)
         system("cat " + tmpfile.name)