Merge pull request #1230 from codeflash-ai/fix/get-language-based-on-formatter

[FIX] Determine language based on common formatters

Author: misrasaurabh1

codeflash/code_utils/env_utils.py

@@ -13,10 +13,14 @@
 from codeflash.code_utils.code_utils import exit_with_message
 from codeflash.code_utils.formatter import format_code
 from codeflash.code_utils.shell_utils import read_api_key_from_shell_config, save_api_key_to_rc
+from codeflash.languages.base import Language
+from codeflash.languages.registry import get_language_support_by_common_formatters
 from codeflash.lsp.helpers import is_LSP_enabled
 
 
-def check_formatter_installed(formatter_cmds: list[str], exit_on_failure: bool = True) -> bool:
+def check_formatter_installed(
+    formatter_cmds: list[str], exit_on_failure: bool = True, language: str = "python"
+) -> bool:
     if not formatter_cmds or formatter_cmds[0] == "disabled":
         return True
     first_cmd = formatter_cmds[0]
@@ -35,10 +39,21 @@ def check_formatter_installed(formatter_cmds: list[str], exit_on_failure: bool =
         )
         return False
 
-    tmp_code = """print("hello world")"""
+    lang_support = get_language_support_by_common_formatters(formatter_cmds)
+    if not lang_support:
+        logger.debug(f"Could not determine language for formatter: {formatter_cmds}")
+        return True
+
+    if lang_support.language == Language.PYTHON:
+        tmp_code = """print("hello world")"""
+    elif lang_support.language in (Language.JAVASCRIPT, Language.TYPESCRIPT):
+        tmp_code = "console.log('hello world');"
+    else:
+        return True
+
     try:
         with tempfile.TemporaryDirectory() as tmpdir:
-            tmp_file = Path(tmpdir) / "test_codeflash_formatter.py"
+            tmp_file = Path(tmpdir) / ("test_codeflash_formatter" + lang_support.default_file_extension)
             tmp_file.write_text(tmp_code, encoding="utf-8")
             format_code(formatter_cmds, tmp_file, print_status=False, exit_on_failure=False)
             return True

codeflash/code_utils/formatter.py

@@ -13,6 +13,7 @@
 import isort
 
 from codeflash.cli_cmds.console import console, logger
+from codeflash.languages.registry import get_language_support
 from codeflash.lsp.helpers import is_LSP_enabled
 
 
@@ -47,8 +48,9 @@ def apply_formatter_cmds(
         raise FileNotFoundError(msg)
 
     file_path = path
+    lang_support = get_language_support(path)
     if test_dir_str:
-        file_path = Path(test_dir_str) / "temp.py"
+        file_path = Path(test_dir_str) / ("temp" + lang_support.default_file_extension)
         shutil.copy2(path, file_path)
 
     file_token = "$file"  # noqa: S105
@@ -87,13 +89,14 @@ def is_diff_line(line: str) -> bool:
     return len(diff_lines)
 
 
-def format_generated_code(generated_test_source: str, formatter_cmds: list[str]) -> str:
+def format_generated_code(generated_test_source: str, formatter_cmds: list[str], language: str = "python") -> str:
     formatter_name = formatter_cmds[0].lower() if formatter_cmds else "disabled"
     if formatter_name == "disabled":  # nothing to do if no formatter provided
         return re.sub(r"\n{2,}", "\n\n", generated_test_source)
     with tempfile.TemporaryDirectory() as test_dir_str:
         # try running formatter, if nothing changes (could be due to formatting failing or no actual formatting needed) return code with 2 or more newlines substituted with 2 newlines
-        original_temp = Path(test_dir_str) / "original_temp.py"
+        lang_support = get_language_support(language)
+        original_temp = Path(test_dir_str) / ("original_temp" + lang_support.default_file_extension)
         original_temp.write_text(generated_test_source, encoding="utf8")
         _, formatted_code, changed = apply_formatter_cmds(
             formatter_cmds, original_temp, test_dir_str, print_status=False, exit_on_failure=False
@@ -130,7 +133,8 @@ def format_code(
             # we don't count the formatting diff for the optimized function as it should be well-formatted
             original_code_without_opfunc = original_code.replace(optimized_code, "")
 
-            original_temp = Path(test_dir_str) / "original_temp.py"
+            lang_support = get_language_support(path)
+            original_temp = Path(test_dir_str) / ("original_temp" + lang_support.default_file_extension)
             original_temp.write_text(original_code_without_opfunc, encoding="utf8")
 
             formatted_temp, formatted_code, changed = apply_formatter_cmds(
@@ -160,6 +164,7 @@ def format_code(
         _, formatted_code, changed = apply_formatter_cmds(
             formatter_cmds, path, test_dir_str=None, print_status=print_status, exit_on_failure=exit_on_failure
         )
+
         if not changed:
             logger.warning(
                 f"No changes detected in {path} after formatting, are you sure you have valid formatter commands?"

codeflash/languages/base.py

@@ -278,6 +278,11 @@ def file_extensions(self) -> tuple[str, ...]:
         """
         ...
 
+    @property
+    def default_file_extension(self) -> str:
+        """Default file extension for this language."""
+        ...
+
     @property
     def test_framework(self) -> str:
         """Primary test framework name.

codeflash/languages/javascript/support.py

@@ -53,6 +53,11 @@ def file_extensions(self) -> tuple[str, ...]:
         """File extensions supported by JavaScript."""
         return (".js", ".jsx", ".mjs", ".cjs")
 
+    @property
+    def default_file_extension(self) -> str:
+        """Default file extension for JavaScript."""
+        return ".js"
+
     @property
     def test_framework(self) -> str:
         """Primary test framework for JavaScript."""

codeflash/languages/python/support.py

@@ -45,6 +45,11 @@ def file_extensions(self) -> tuple[str, ...]:
         """File extensions supported by Python."""
         return (".py", ".pyw")
 
+    @property
+    def default_file_extension(self) -> str:
+        """Default file extension for Python."""
+        return ".py"
+
     @property
     def test_framework(self) -> str:
         """Primary test framework for Python."""

codeflash/languages/registry.py

@@ -178,6 +178,41 @@ def get_language_support(identifier: Path | Language | str) -> LanguageSupport:
 _FRAMEWORK_CACHE: dict[str, LanguageSupport] = {}
 
 
+def get_language_support_by_common_formatters(formatter_cmd: str | list[str]) -> LanguageSupport | None:
+    language: Language | None = None
+    if isinstance(formatter_cmd, str):
+        formatter_cmd = [formatter_cmd]
+
+    if len(formatter_cmd) == 1:
+        formatter_cmd = formatter_cmd[0].split(" ")
+
+    # Try as extension first
+    ext = None
+
+    py_formatters = ["black", "isort", "ruff", "autopep8", "yapf", "pyfmt"]
+    js_ts_formatters = ["prettier", "eslint", "biome", "rome", "deno", "standard", "tslint"]
+
+    if any(cmd in py_formatters for cmd in formatter_cmd):
+        ext = ".py"
+    elif any(cmd in js_ts_formatters for cmd in formatter_cmd):
+        ext = ".js"
+
+    if ext is None:
+        # can't determine language
+        return None
+
+    cls = _EXTENSION_REGISTRY[ext]
+    language = cls().language
+
+    # Return cached instance or create new one
+    if language not in _SUPPORT_CACHE:
+        if language not in _LANGUAGE_REGISTRY:
+            raise UnsupportedLanguageError(str(language), get_supported_languages())
+        _SUPPORT_CACHE[language] = _LANGUAGE_REGISTRY[language]()
+
+    return _SUPPORT_CACHE[language]
+
+
 def get_language_support_by_framework(test_framework: str) -> LanguageSupport | None:
     """Get language support for a test framework.