Get diff against the base commit

packages/bot/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name    = "automa-bot"
-version = "0.3.0"
+version = "0.4.0"
 
 authors         = [{ name = "Sunkara, Inc.", email = "engineering@automa.app" }]
 description     = "Bot SDK for Automa"

packages/bot/src/automa/bot/resources/code.py

@@ -24,10 +24,15 @@
 
 
 # TODO: Use programmatic git instead of git command
-def get_diff(path: str) -> str:
+def get_diff(path: str, base_commit: str | None = None) -> str:
     """Get the diff for the given git repository path."""
+    args = ["git", "diff"]
+
+    if base_commit:
+        args.append(base_commit)
+
     result = subprocess.run(
-        ["git", "diff"],
+        args,
         cwd=path,
         capture_output=True,
         text=True,
@@ -78,12 +83,12 @@ def _read_token(self, folder: str) -> str | None:
 
         return token
 
-    def _read_base_commit(self, folder: str) -> str | None:
+    def _read_base_commit(self, folder: str, type: str) -> str | None:
         base_commit = None
 
         try:
             with open(
-                f"{folder}/.git/automa_proposal_base_commit", "r", encoding="utf8"
+                f"{folder}/.git/automa_{type}_base_commit", "r", encoding="utf8"
             ) as f:
                 base_commit = f.read().strip()
         except FileNotFoundError:
@@ -103,16 +108,8 @@ def _clone_code(self, folder: str, url: str) -> None:
             check=True,
         )
 
-        # Note down the base commit
-        result = subprocess.run(
-            ["git", "rev-parse", "HEAD"],
-            cwd=path,
-            capture_output=True,
-            text=True,
-            check=True,
-        )
-
-        self._write_base_commit(folder, result.stdout.strip())
+        # Note down the base commit to send in proposal
+        self._write_base_commit(folder, "proposal")
 
     def _extract_download(self, folder: str) -> None:
         rmtree(folder, ignore_errors=True)
@@ -121,17 +118,27 @@ def _extract_download(self, folder: str) -> None:
         with tarfile.open(f"{folder}.tar.gz", "r:gz") as tar:
             tar.extractall(path=folder)
 
+        # Note down the base commit to be used in generating diff
+        self._write_base_commit(folder, "diff")
+
     def _write_token(self, folder: str, token: str) -> None:
         # Save the proposal token for later use
         with open(f"{folder}/.git/automa_proposal_token", "w", encoding="utf8") as f:
             f.write(token)
 
-    def _write_base_commit(self, folder: str, base_commit: str) -> None:
-        # Save the base commit for later use
+    def _write_base_commit(self, folder: str, type: str) -> None:
+        result = subprocess.run(
+            ["git", "rev-parse", "HEAD"],
+            cwd=folder,
+            capture_output=True,
+            text=True,
+            check=True,
+        )
+
         with open(
-            f"{folder}/.git/automa_proposal_base_commit", "w", encoding="utf8"
+            f"{folder}/.git/automa_{type}_base_commit", "w", encoding="utf8"
         ) as f:
-            f.write(base_commit)
+            f.write(result.stdout.strip())
 
 
 class CodeResource(SyncAPIResource, BaseCodeResource):
@@ -202,12 +209,13 @@ def propose(
     ):
         path = self._path(body["task"])
         token = self._read_token(path)
-        base_commit = self._read_base_commit(path)
+        base_commit = self._read_base_commit(path, "proposal")
+        diff_base_commit = self._read_base_commit(path, "diff")
 
         if not token:
             raise ValueError("Failed to read the stored proposal token")
 
-        diff = get_diff(path)
+        diff = get_diff(path, base_commit or diff_base_commit)
 
         return self._client.post(
             "/bot/code/propose",
@@ -295,12 +303,13 @@ async def propose(
     ):
         path = self._path(body["task"])
         token = await to_thread(self._read_token, path)
-        base_commit = await to_thread(self._read_base_commit, path)
+        base_commit = await to_thread(self._read_base_commit, path, "proposal")
+        diff_base_commit = await to_thread(self._read_base_commit, path, "diff")
 
         if not token:
             raise ValueError("Failed to read the stored proposal token")
 
-        diff = await to_thread(get_diff, path)
+        diff = await to_thread(get_diff, path, base_commit or diff_base_commit)
 
         return await self._client.post(
             "/bot/code/propose",

packages/bot/tests/fixtures/download/_git/config

@@ -0,0 +1,5 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = false
+	logallrefupdates = true

packages/bot/tests/fixtures/download/_git/logs/HEAD

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 5575f40bc4ead411b690b6f2f09636e3468ef12e automa[bot] <60525818+automa[bot]@users.noreply.github.com> 1772718655 +0530	commit (initial): Downloaded code

packages/bot/tests/fixtures/download/_git/logs/refs/heads/master

@@ -0,0 +1 @@
+0000000000000000000000000000000000000000 5575f40bc4ead411b690b6f2f09636e3468ef12e automa[bot] <60525818+automa[bot]@users.noreply.github.com> 1772718655 +0530	commit (initial): Downloaded code

packages/bot/tests/fixtures/download/_git/refs/heads/master

@@ -0,0 +1 @@
+5575f40bc4ead411b690b6f2f09636e3468ef12e

packages/bot/tests/resources/test_code.py

@@ -12,6 +12,7 @@
 folder = "/tmp/automa/tasks/28"
 proposal_token_file = f"{folder}/.git/automa_proposal_token"
 proposal_base_commit_file = f"{folder}/.git/automa_proposal_base_commit"
+diff_base_commit_file = f"{folder}/.git/automa_diff_base_commit"
 
 
 @pytest.fixture
@@ -45,7 +46,7 @@ def async_code_resource():
 @pytest.fixture
 def fixture_git():
     tests_folder = Path(__file__).parent.parent
-    git_path = tests_folder / "fixtures" / "download_git"
+    git_path = tests_folder / "fixtures" / "download" / "_git"
 
     yield git_path
 
@@ -55,23 +56,20 @@ def fixture_tarfile():
     tests_folder = Path(__file__).parent.parent
     fixture = tests_folder / "fixtures" / "download"
     tarfile_path = tests_folder / "fixture.tar.gz"
+    git_path = fixture / ".git"
+    _git_path = fixture / "_git"
 
-    run(["git", "init"], cwd=fixture, capture_output=True)
-    run(["git", "add", "."], cwd=fixture, capture_output=True)
-    run(["git", "config", "user.name", "Tmp"], cwd=fixture, capture_output=True)
-    run(
-        ["git", "config", "user.email", "tmp@tmp.com"], cwd=fixture, capture_output=True
-    )
-    run(["git", "commit", "-m", "Initial commit"], cwd=fixture, capture_output=True)
+    _git_path.rename(git_path)
 
     with tarfile.open(tarfile_path, "w:gz") as tar:
         for subpath in listdir(fixture):
             tar.add(fixture / subpath, arcname=subpath)
 
+    git_path.rename(_git_path)
+
     yield tarfile_path
 
     remove(tarfile_path)
-    rmtree(fixture / ".git", ignore_errors=True)
 
 
 def test_cleanup_proxy(code_resource):
@@ -154,7 +152,7 @@ def test_download_invalid_token(code_resource):
 
 
 @pytest.mark.asyncio
-async def test_download_async_invalid_token(async_code_resource):
+async def test_download_invalid_token_async(async_code_resource):
     # Mock client response
     response_mock = MagicMock()
     response_mock.status_code = 403
@@ -299,13 +297,19 @@ def test_download_proxy(fixture_tarfile, code_resource):
     assert path.exists(folder)
     assert sorted(listdir(folder)) == [
         ".git",
+        "LICENSE",
         "README.md",
     ]
 
     # Saves proposal token
     with open(proposal_token_file, "r") as f:
         assert f.read() == "ghijkl"
 
+    # Saves diff base commit
+    assert not path.exists(proposal_base_commit_file)
+    with open(diff_base_commit_file, "r") as f:
+        assert f.read() == "5575f40bc4ead411b690b6f2f09636e3468ef12e"
+
     return created_folder
 
 
@@ -352,13 +356,19 @@ async def test_download_proxy_async(fixture_tarfile, async_code_resource):
     assert path.exists(folder)
     assert sorted(listdir(folder)) == [
         ".git",
+        "LICENSE",
         "README.md",
     ]
 
     # Saves proposal token
     with open(proposal_token_file, "r") as f:
         assert f.read() == "ghijkl"
 
+    # Saves diff base commit
+    assert not path.exists(proposal_base_commit_file)
+    with open(diff_base_commit_file, "r") as f:
+        assert f.read() == "5575f40bc4ead411b690b6f2f09636e3468ef12e"
+
 
 def test_propose_no_token(fixture_tarfile, code_resource):
     test_download_proxy(fixture_tarfile, code_resource)
@@ -373,7 +383,7 @@ def test_propose_no_token(fixture_tarfile, code_resource):
 
 
 @pytest.mark.asyncio
-async def test_propose_async_no_token(fixture_tarfile, async_code_resource):
+async def test_propose_no_token_async(fixture_tarfile, async_code_resource):
     await test_download_proxy_async(fixture_tarfile, async_code_resource)
 
     remove(proposal_token_file)
@@ -421,7 +431,7 @@ def test_propose_invalid_token(fixture_tarfile, code_resource):
 
 
 @pytest.mark.asyncio
-async def test_propose_async_invalid_token(fixture_tarfile, async_code_resource):
+async def test_propose_invalid_token_async(fixture_tarfile, async_code_resource):
     await test_download_proxy_async(fixture_tarfile, async_code_resource)
 
     with open(f"{folder}/README.md", "w") as f:
@@ -593,6 +603,58 @@ def test_propose_with_added_files_using_add_all(fixture_tarfile, code_resource):
     )
 
 
+def test_propose_with_intermediate_commits(fixture_tarfile, code_resource):
+    test_download_proxy(fixture_tarfile, code_resource)
+
+    with open(f"{folder}/LICENSE", "w") as f:
+        f.write("MIT\n")
+
+    run(["git", "add", "LICENSE"], cwd=folder, capture_output=True)
+    run(
+        [
+            "git",
+            "-c",
+            "user.name=Tmp",
+            "-c",
+            "user.email=tmp@tmp.com",
+            "commit",
+            "-m",
+            "Intermediate commit",
+        ],
+        cwd=folder,
+        capture_output=True,
+    )
+
+    with open(f"{folder}/README.md", "w") as f:
+        f.write("Content\n")
+
+    # Mock client response
+    response_mock = MagicMock()
+    response_mock.status_code = 204
+    response_mock.is_error = False
+
+    code_resource._client._client.request.return_value = response_mock
+
+    code_resource.propose({"task": {"id": 28, "token": "abcdef"}})
+
+    # Hits the API
+    code_resource._client._client.request.assert_called_once_with(
+        "post",
+        "/bot/code/propose",
+        json={
+            "task": {"id": 28, "token": "abcdef"},
+            "proposal": {
+                "token": "ghijkl",
+                "diff": "diff --git a/LICENSE b/LICENSE\nindex e69de29..a22a2da 100644\n--- a/LICENSE\n+++ b/LICENSE\n@@ -0,0 +1 @@\n+MIT\ndiff --git a/README.md b/README.md\nindex e69de29..39c9f36 100644\n--- a/README.md\n+++ b/README.md\n@@ -0,0 +1 @@\n+Content\n",
+            },
+        },
+        headers={
+            "Accept": "application/json",
+            "Content-Type": "application/json",
+        },
+    )
+
+
 def test_propose_with_proposal_properties(fixture_tarfile, code_resource):
     test_download_proxy(fixture_tarfile, code_resource)
 
@@ -710,6 +772,7 @@ def test_download_direct(fixture_git, code_resource):
     assert path.exists(folder)
     assert sorted(listdir(folder)) == [
         ".git",
+        "LICENSE",
         "README.md",
     ]
 
@@ -719,7 +782,8 @@ def test_download_direct(fixture_git, code_resource):
 
     # Saves the base commit
     with open(proposal_base_commit_file, "r") as f:
-        assert f.read() == "cc3f46ae7fdf71747b66b3e4272c0e5fe290d116"
+        assert f.read() == "5575f40bc4ead411b690b6f2f09636e3468ef12e"
+    assert not path.exists(diff_base_commit_file)
 
 
 @pytest.mark.asyncio
@@ -766,6 +830,7 @@ async def test_download_direct_async(fixture_git, async_code_resource):
     assert path.exists(folder)
     assert sorted(listdir(folder)) == [
         ".git",
+        "LICENSE",
         "README.md",
     ]
 
@@ -775,7 +840,8 @@ async def test_download_direct_async(fixture_git, async_code_resource):
 
     # Saves the base commit
     with open(proposal_base_commit_file, "r") as f:
-        assert f.read() == "cc3f46ae7fdf71747b66b3e4272c0e5fe290d116"
+        assert f.read() == "5575f40bc4ead411b690b6f2f09636e3468ef12e"
+    assert not path.exists(diff_base_commit_file)
 
 
 def test_propose_direct(fixture_git, code_resource):
@@ -802,7 +868,7 @@ def test_propose_direct(fixture_git, code_resource):
             "proposal": {
                 "token": "ghijkl",
                 "diff": "diff --git a/README.md b/README.md\nindex e69de29..39c9f36 100644\n--- a/README.md\n+++ b/README.md\n@@ -0,0 +1 @@\n+Content\n",
-                "base_commit": "cc3f46ae7fdf71747b66b3e4272c0e5fe290d116",
+                "base_commit": "5575f40bc4ead411b690b6f2f09636e3468ef12e",
             },
         },
         headers={
@@ -837,7 +903,7 @@ async def test_propose_direct_async(fixture_git, async_code_resource):
             "proposal": {
                 "token": "ghijkl",
                 "diff": "diff --git a/README.md b/README.md\nindex e69de29..39c9f36 100644\n--- a/README.md\n+++ b/README.md\n@@ -0,0 +1 @@\n+Content\n",
-                "base_commit": "cc3f46ae7fdf71747b66b3e4272c0e5fe290d116",
+                "base_commit": "5575f40bc4ead411b690b6f2f09636e3468ef12e",
             },
         },
         headers={