Merge branch 'main' into triple-backticks

aider/__init__.py

@@ -1 +1 @@
-__version__ = "0.7.0"
+__version__ = "0.7.1"

aider/coders/wholefile_coder.py

@@ -44,7 +44,7 @@ class WholeFileCoder(Coder):
                     full_path = (Path(self.root) / fname).absolute()
 
                     if mode == "diff" and full_path.exists():
-                        orig_lines = full_path.read_text().splitlines()
+                        orig_lines = full_path.read_text().splitlines(keepends=True)
 
                         show_diff = diffs.diff_partial_update(
                             orig_lines,
@@ -99,7 +99,7 @@ class WholeFileCoder(Coder):
                 full_path = (Path(self.root) / fname).absolute()
 
                 if mode == "diff" and full_path.exists():
-                    orig_lines = full_path.read_text().splitlines()
+                    orig_lines = full_path.read_text().splitlines(keepends=True)
 
                     show_diff = diffs.diff_partial_update(
                         orig_lines,

aider/coders/wholefile_prompts.py

@@ -14,7 +14,8 @@ Once you understand the request you MUST:
 3. If changes are needed, output a copy of each file that needs changes.
 """
 
-    system_reminder = """To return code you MUST use this *file listing* format:
+    system_reminder = """To suggest changes to a file you MUST return the entire content of the updated file.
+You MUST use this *file listing* format:
 
 path/to/filename.js
 {fence}javascript

aider/diffs.py

@@ -33,6 +33,13 @@ def create_progress_bar(percentage):
     return bar
 
 
+def assert_newlines(lines):
+    if not lines:
+        return
+    for line in lines:
+        assert line and line[-1] == "\n", line
+
+
 def diff_partial_update(lines_orig, lines_updated, final=False, fname=None):
     """
     Given only the first part of an updated file, show the diff while
@@ -43,6 +50,9 @@ def diff_partial_update(lines_orig, lines_updated, final=False, fname=None):
     # dump(lines_orig)
     # dump(lines_updated)
 
+    assert_newlines(lines_orig)
+    assert_newlines(lines_orig)
+
     num_orig_lines = len(lines_orig)
 
     if final:

scripts/benchmark.py

@@ -10,6 +10,7 @@ from collections import defaultdict
 from json.decoder import JSONDecodeError
 from pathlib import Path
 
+import git
 import lox
 from rich.console import Console
 
@@ -26,6 +27,11 @@ assert ORIGINAL_DNAME.exists() and ORIGINAL_DNAME.is_dir()
 
 
 def main():
+    repo = git.Repo(search_parent_directories=True)
+    commit_hash = repo.head.object.hexsha[:7]
+    if repo.is_dirty():
+        commit_hash += "-dirty"
+
     parser = argparse.ArgumentParser(description="Aider Benchmark")
     parser.add_argument("dirname", type=str, help="Directory name")
     parser.add_argument("--model", "-m", type=str, help="Model name", default="gpt-3.5-turbo")
@@ -117,6 +123,7 @@ def main():
                 args.no_test,
                 args.verbose,
                 args.stats_only,
+                commit_hash,
             )
 
             all_results.append(results)
@@ -133,6 +140,7 @@ def main():
                 args.no_test,
                 args.verbose,
                 args.stats_only,
+                commit_hash,
             )
         all_results = run_test_threaded.gather(tqdm=True)
 
@@ -172,11 +180,10 @@ def summarize_results(dirname, all_results, total_tests=None):
         total_cost += results["cost"]
         duration += results["duration"]
 
-        for key in "model edit_format".split():
-            if key in results:
-                variants[key].add(results[key])
+        for key in "model edit_format commit_hash".split():
+            val = results.get(key)
+            variants[key].add(val)
 
-    dump(completed_tests)
     if not completed_tests:
         return
 
@@ -189,7 +196,7 @@ def summarize_results(dirname, all_results, total_tests=None):
             style = "red"
         else:
             style = None
-        val = ", ".join(val)
+        val = ", ".join(map(str, val))
         console.print(f"{key}: {val}", style=style)
 
     console.print()
@@ -213,7 +220,7 @@ def summarize_results(dirname, all_results, total_tests=None):
     console.rule()
 
 
-def run_test(testdir, model_name, edit_format, retries, no_test, verbose, stats_only):
+def run_test(testdir, model_name, edit_format, retries, no_test, verbose, stats_only, commit_hash):
     if not stats_only:
         dump(testdir)
 
@@ -248,12 +255,18 @@ def run_test(testdir, model_name, edit_format, retries, no_test, verbose, stats_
             shutil.copy(original_fname, fname)
 
     file_list = " ".join(fname.name for fname in fnames)
-    instructions = (testdir / ".docs/instructions.md").read_text()
-    instructions += (
-        "\n\n=====\n\nModify these files according to the above instructions. Only use standard"
-        " python libraries, don't suggest installing any packages.\n"
-    )
-    instructions += file_list
+    intro = testdir / ".docs/introduction.md"
+    if intro.exists():
+        instructions = intro.read_text() + "\n\n"
+    else:
+        instructions = ""
+    instructions += (testdir / ".docs/instructions.md").read_text()
+    instructions += f"""
+=====
+Use the above instructions to modify the supplied files: {file_list}
+Keep and implement the existing function or class stubs, they will be called from unit tests.
+Only use standard python libraries, don't suggest installing any packages.
+"""
 
     io = InputOutput(
         pretty=True,
@@ -319,6 +332,7 @@ def run_test(testdir, model_name, edit_format, retries, no_test, verbose, stats_
         tests_outcomes=test_outcomes,
         cost=coder.total_cost,
         duration=dur,
+        commit_hash=commit_hash,
     )
     dump(results)