perf(file-ops): eliminate redundant subprocess calls in write_file and V4A patch path

pyproject.toml

@@ -50,6 +50,11 @@ dependencies = [
   "rich==14.3.3",
   "tenacity==9.1.4",
   "pyyaml==6.0.3",
+  # tomli — TOML parser used as pre-3.11 fallback for stdlib tomllib
+  # (tools/file_operations.py: _lint_toml_inproc).  On 3.11+ this is a
+  # no-op install; the platform marker keeps it off the resolution for
+  # current Python so it doesn't add a dependency to every install.
+  "tomli==2.4.1; python_version < '3.11'",
   "ruamel.yaml==0.18.17",
   "requests==2.33.0",  # CVE-2026-25645
   "jinja2==3.1.6",

tests/tools/test_patch_parser.py

@@ -170,7 +170,7 @@ def read_file_raw(self, path):
                     error=None,
                 )
 
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 self.written = content
                 return SimpleNamespace(error=None)
 
@@ -216,7 +216,7 @@ def read_file_raw(self, path):
                     content="def main():\n    pass\n",
                     error=None,
                 )
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 self.written = content
                 return SimpleNamespace(error=None)
 
@@ -244,7 +244,7 @@ def read_file_raw(self, path):
                     content="existing = True\n",
                     error=None,
                 )
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 self.written = content
                 return SimpleNamespace(error=None)
 
@@ -281,7 +281,7 @@ class FakeFileOps:
             written = None
             def read_file_raw(self, path):
                 return SimpleNamespace(content=file_content, error=None)
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 self.written = content
                 return SimpleNamespace(error=None)
 
@@ -315,7 +315,7 @@ class FakeFileOps:
             written = None
             def read_file_raw(self, path):
                 return SimpleNamespace(content=file_content, error=None)
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 self.written = content
                 return SimpleNamespace(error=None)
 
@@ -358,7 +358,7 @@ def read_file_raw(self, path):
                     return SimpleNamespace(content=None, error=f"File not found: {path}")
                 return SimpleNamespace(content=content, error=None)
 
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 written[path] = content
                 return SimpleNamespace(error=None)
 
@@ -393,7 +393,7 @@ def read_file_raw(self, path):
                 }
                 return SimpleNamespace(content=files[path], error=None)
 
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 written[path] = content
                 return SimpleNamespace(error=None)
 
@@ -550,7 +550,7 @@ def test_lsp_diagnostics_propagated_from_write_file_on_add(self):
         )
 
         class FakeFileOps:
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 return SimpleNamespace(error=None, lsp_diagnostics=diag_block)
 
             def _check_lint(self, path):
@@ -584,7 +584,7 @@ class FakeFileOps:
             def read_file_raw(self, path):
                 return SimpleNamespace(content="ctx\nold\nctx\n", error=None)
 
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 return SimpleNamespace(error=None, lsp_diagnostics=diag_block)
 
             def _check_lint(self, path):
@@ -602,7 +602,7 @@ def test_lsp_diagnostics_none_when_no_blocks_emitted(self):
         ops = self._build_ops_writing("foo.py", "x = 1\n")
 
         class FakeFileOps:
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 # lsp_diagnostics omitted entirely (older WriteResult shape).
                 return SimpleNamespace(error=None)
 
@@ -635,7 +635,7 @@ def test_lsp_diagnostics_combined_across_multiple_files(self):
         }
 
         class FakeFileOps:
-            def write_file(self, path, content):
+            def write_file(self, path, content, pre_content=None):
                 return SimpleNamespace(error=None, lsp_diagnostics=per_file[path])
 
             def _check_lint(self, path):
@@ -647,3 +647,109 @@ def _check_lint(self, path):
         assert result.lsp_diagnostics is not None
         assert per_file["a.ts"] in result.lsp_diagnostics
         assert per_file["b.ts"] in result.lsp_diagnostics
+
+
+class TestV4ABomRoundTrip:
+    """V4A patches must not silently strip a UTF-8 BOM on UPDATE.
+
+    ``read_file_raw`` deliberately strips the BOM (the agent should
+    never see U+FEFF), but the underlying ``write_file`` must restore
+    it on rewrite — otherwise a V4A patch turns an existing BOM-bearing
+    file into a plain UTF-8 file.  Regression for teknium1 review on
+    PR #55661.
+    """
+
+    BOM = "\ufeff"
+
+    def _file_ops_for_update(self, file_path: str, original_bytes: bytes):
+        """Build a FakeFileOps whose ``write_file`` writes real bytes to
+        ``file_path``, simulating BOM-preserving behaviour like the real
+        ``FileOperations.write_file`` (which probes disk for the marker)."""
+        from pathlib import Path
+        from tools.file_operations import _has_bom, _UTF8_BOM
+
+        target = Path(file_path)
+        _bom = self.BOM  # capture for inner class
+
+        class FakeFileOps:
+            def read_file_raw(self, path):
+                # Simulate BOM-stripped read — same as the real
+                # read_file_raw which strips the marker before returning.
+                decoded = original_bytes.decode("utf-8")
+                if decoded.startswith(_bom):
+                    decoded = decoded[1:]
+                return SimpleNamespace(content=decoded, error=None)
+
+            def write_file(self, path, content, pre_content=None):
+                # Simulate real write_file: probe the target for a BOM
+                # (the real impl calls _file_has_bom → head -c 3) and
+                # prepend if the original had one.
+                had_bom = target.exists() and target.read_bytes().startswith(
+                    _bom.encode("utf-8")
+                )
+                if had_bom and not _has_bom(content):
+                    content = _UTF8_BOM + content
+                target.parent.mkdir(parents=True, exist_ok=True)
+                target.write_text(content, encoding="utf-8")
+                return SimpleNamespace(error=None)
+
+        return FakeFileOps()
+
+    def test_update_preserves_bom(self, tmp_path):
+        """A V4A UPDATE on a BOM-bearing file keeps the BOM."""
+        from tools.patch_parser import parse_v4a_patch, apply_v4a_operations
+
+        target = tmp_path / "bom_config.py"
+        original = self.BOM + "setting = 'old'\n"
+        target.write_text(original, encoding="utf-8")
+
+        patch = """\
+*** Begin Patch
+*** Update File: bom_config.py
+@@ setting @@
+-setting = 'old'
++setting = 'new'
+*** End Patch"""
+
+        ops, err = parse_v4a_patch(patch)
+        assert err is None
+
+        file_ops = self._file_ops_for_update(str(target), original.encode("utf-8"))
+        result = apply_v4a_operations(ops, file_ops)
+
+        assert result.success is True
+        raw = target.read_bytes()
+        assert raw.startswith(
+            self.BOM.encode("utf-8")
+        ), "BOM was stripped by V4A round-trip"
+        assert b"setting = 'new'" in raw
+        assert b"setting = 'old'" not in raw
+
+    def test_update_no_bom_when_original_had_none(self, tmp_path):
+        """A V4A UPDATE on a plain file must NOT inject a BOM."""
+        from tools.patch_parser import parse_v4a_patch, apply_v4a_operations
+
+        target = tmp_path / "plain.py"
+        original = "print('hello')\n"
+        target.write_text(original, encoding="utf-8")
+
+        patch = """\
+*** Begin Patch
+*** Update File: plain.py
+@@ print @@
+-print('hello')
++print('world')
+*** End Patch"""
+
+        ops, err = parse_v4a_patch(patch)
+        assert err is None
+
+        file_ops = self._file_ops_for_update(str(target), original.encode("utf-8"))
+        result = apply_v4a_operations(ops, file_ops)
+
+        assert result.success is True
+        raw = target.read_bytes()
+        assert not raw.startswith(
+            self.BOM.encode("utf-8")
+        ), "BOM was injected on a plain file"
+        assert b"print('world')" in raw