Use 1-based line numbers for RegexTransformer changes (#759)

* Use 1-based line numbers for RegexTransformer changes

* Add line number validation to CodeTF Change model

Author: drdavella

src/codemodder/codemods/regex_transformer.py

@@ -11,11 +11,13 @@
 
 
 class RegexTransformerPipeline(BaseTransformerPipeline):
-    pattern: Pattern
+    pattern: Pattern | str
     replacement: str
     change_description: str
 
-    def __init__(self, pattern: Pattern, replacement: str, change_description: str):
+    def __init__(
+        self, pattern: Pattern | str, replacement: str, change_description: str
+    ):
         super().__init__()
         self.pattern = pattern
         self.replacement = replacement
@@ -27,6 +29,7 @@ def apply(
         file_context: FileContext,
         results: list[Result] | None,
     ) -> ChangeSet | None:
+        del results
 
         changes = []
         updated_lines = []
@@ -35,12 +38,13 @@ def apply(
             original_lines = f.readlines()
 
         for lineno, line in enumerate(original_lines):
+            # TODO: use results to filter out which lines to change
             changed_line = re.sub(self.pattern, self.replacement, line)
             updated_lines.append(changed_line)
             if line != changed_line:
                 changes.append(
                     Change(
-                        lineNumber=lineno,
+                        lineNumber=lineno + 1,
                         description=self.change_description,
                         findings=file_context.get_findings_for_location(lineno),
                     )

src/codemodder/codetf.py

@@ -54,6 +54,12 @@ class Change(BaseModel):
     packageActions: Optional[list[PackageAction]] = None
     findings: Optional[list[Finding]] = None
 
+    @model_validator(mode="after")
+    def validate_lineNumber(self):
+        if self.lineNumber < 1:
+            raise ValueError("lineNumber must be greater than 0")
+        return self
+
 
 class AIMetadata(BaseModel):
     provider: Optional[str] = None

tests/test_codetf.py

@@ -58,6 +58,11 @@ def test_change_diffside(side):
     assert change.model_dump()["diffSide"] == side
 
 
+def test_change_invalid_line_number():
+    with pytest.raises(ValueError):
+        Change(lineNumber=0, description="Change 1 to 2")
+
+
 def test_write_codetf(tmpdir, mocker, codetf_schema):
     path = tmpdir / "test.codetf.json"
 

tests/test_regex_transformer.py

@@ -20,7 +20,7 @@ def test_transformer_no_change(mocker, caplog, tmp_path_factory):
         dry_run=True,
         verbose=False,
         registry=mocker.MagicMock(),
-        providers=None,
+        providers=mocker.MagicMock(),
         repo_manager=mocker.MagicMock(),
         path_include=[],
         path_exclude=[],
@@ -53,7 +53,7 @@ def test_transformer(mocker, tmp_path_factory):
         dry_run=False,
         verbose=False,
         registry=mocker.MagicMock(),
-        providers=None,
+        providers=mocker.MagicMock(),
         repo_manager=mocker.MagicMock(),
         path_include=[],
         path_exclude=[],
@@ -69,3 +69,4 @@ def test_transformer(mocker, tmp_path_factory):
     )
     assert changeset is not None
     assert code.read_text() == text.replace("hello", "bye")
+    assert changeset.changes[0].lineNumber == 1