Regex Transformer (#729)

* new regex transformer

* add typing

src/codemodder/codemods/regex_transformer.py

@@ -0,0 +1,63 @@
+import re
+from typing import Pattern
+
+from codemodder.codemods.base_transformer import BaseTransformerPipeline
+from codemodder.codetf import Change, ChangeSet
+from codemodder.context import CodemodExecutionContext
+from codemodder.diff import create_diff
+from codemodder.file_context import FileContext
+from codemodder.logging import logger
+from codemodder.result import Result
+
+
+class RegexTransformerPipeline(BaseTransformerPipeline):
+    pattern: Pattern
+    replacement: str
+    change_description: str
+
+    def __init__(self, pattern: Pattern, replacement: str, change_description: str):
+        super().__init__()
+        self.pattern = pattern
+        self.replacement = replacement
+        self.change_description = change_description
+
+    def apply(
+        self,
+        context: CodemodExecutionContext,
+        file_context: FileContext,
+        results: list[Result] | None,
+    ) -> ChangeSet | None:
+
+        changes = []
+        updated_lines = []
+
+        with open(file_context.file_path, "r", encoding="utf-8") as f:
+            original_lines = f.readlines()
+
+        for lineno, line in enumerate(original_lines):
+            changed_line = re.sub(self.pattern, self.replacement, line)
+            updated_lines.append(changed_line)
+            if line != changed_line:
+                changes.append(
+                    Change(
+                        lineNumber=lineno,
+                        description=self.change_description,
+                        findings=file_context.get_findings_for_location(lineno),
+                    )
+                )
+
+        if not changes:
+            logger.debug("No changes produced for %s", file_context.file_path)
+            return None
+
+        diff = create_diff(original_lines, updated_lines)
+
+        if not context.dry_run:
+            with open(file_context.file_path, "w+", encoding="utf-8") as original:
+                original.writelines(updated_lines)
+
+        return ChangeSet(
+            path=str(file_context.file_path.relative_to(context.directory)),
+            diff=diff,
+            changes=changes,
+        )

tests/test_regex_transformer.py

@@ -0,0 +1,71 @@
+import logging
+
+from codemodder.codemods.regex_transformer import RegexTransformerPipeline
+from codemodder.context import CodemodExecutionContext
+from codemodder.file_context import FileContext
+
+
+def test_transformer_no_change(mocker, caplog, tmp_path_factory):
+    caplog.set_level(logging.DEBUG)
+    base_dir = tmp_path_factory.mktemp("foo")
+    code = base_dir / "code.py"
+    code.write_text("# Something that won't match")
+
+    file_context = FileContext(
+        base_dir,
+        code,
+    )
+    execution_context = CodemodExecutionContext(
+        directory=base_dir,
+        dry_run=True,
+        verbose=False,
+        registry=mocker.MagicMock(),
+        providers=None,
+        repo_manager=mocker.MagicMock(),
+        path_include=[],
+        path_exclude=[],
+    )
+    pipeline = RegexTransformerPipeline(
+        pattern=r"hello", replacement="bye", change_description="testing"
+    )
+
+    changeset = pipeline.apply(
+        context=execution_context,
+        file_context=file_context,
+        results=None,
+    )
+    assert changeset is None
+    assert "No changes produced for" in caplog.text
+
+
+def test_transformer(mocker, tmp_path_factory):
+    base_dir = tmp_path_factory.mktemp("foo")
+    code = base_dir / "code.py"
+    text = "# Something that will match pattern hello"
+    code.write_text(text)
+
+    file_context = FileContext(
+        base_dir,
+        code,
+    )
+    execution_context = CodemodExecutionContext(
+        directory=base_dir,
+        dry_run=False,
+        verbose=False,
+        registry=mocker.MagicMock(),
+        providers=None,
+        repo_manager=mocker.MagicMock(),
+        path_include=[],
+        path_exclude=[],
+    )
+    pipeline = RegexTransformerPipeline(
+        pattern=r"hello", replacement="bye", change_description="testing"
+    )
+
+    changeset = pipeline.apply(
+        context=execution_context,
+        file_context=file_context,
+        results=None,
+    )
+    assert changeset is not None
+    assert code.read_text() == text.replace("hello", "bye")