Formatting

tests/test_formatting.py

@@ -0,0 +1,165 @@
+import ast
+import contextlib
+import io
+import json
+import os
+import pathlib
+import subprocess
+import tempfile
+import sys
+import unittest
+from unittest.mock import patch
+
+import modulus.cli
+import modulus.config
+import modulus.export
+import modulus.format_layout
+import modulus.formatting
+
+class formatting_test(unittest.TestCase):
+  def formatted(self, source, suffix):
+    result = modulus.format_layout.format_source(source, suffix)
+    self.assertEqual(modulus.format_layout.format_source(result, suffix), result)
+    return result
+  def test_cpp_units_namespaces_and_include_groups(self):
+    source = '#include <vector>\n#include <string>\n\n#include "local.hpp"\n\nnamespace sample {\nint one() {\n\nint x=1;\n\nx++;\nreturn x;\n}\nint two() {return 2;}\n}\n'
+    result = self.formatted(source, ".cpp")
+    self.assertTrue(result.startswith(source.split("namespace")[0]))
+    self.assertIn("namespace sample {\n\nint one()", result)
+    self.assertIn("int x = 1;\n  x++;\n  return x;", result)
+    self.assertIn("}\n\nint two()", result)
+    self.assertTrue(result.endswith("}\n\n}\n"))
+  def test_cpp_no_column_limit_and_simple_lists(self):
+    name = "argument_" + "a" * 180
+    result = self.formatted(f"void function(\nint {name},\nint b) {{\n  call(\n{name},\nb);\n}}\n", ".cpp")
+    self.assertIn(f"void function(int {name}, int b)", result)
+    self.assertIn(f"call({name}, b);", result)
+  def test_cpp_class_and_struct_members_have_no_blank_lines(self):
+    for kind in ("class", "struct"):
+      with self.subTest(kind=kind):
+        source = f'{kind} source_location {{\npublic:\n\n  source_location() = default;\n\n  source_location(source_location const&) = delete;\n\n  int line() const;\n\n  int value() const {{\n\n    return line_;\n\n  }}\n\n  struct nested {{\n\n    int value;\n\n  }};\n\nprivate:\n\n  int line_;\n\n}};\n\nstruct next {{ int value; }};\n'
+        result = self.formatted(source, ".hpp")
+        body = result.split(f"{kind} source_location {{\n", 1)[1].split("\n};", 1)[0]
+        self.assertNotIn("\n\n", body)
+        self.assertIn("};\n\nstruct next", result)
+  def test_cpp_class_keeps_literal_and_comment_blank_lines(self):
+    literal = 'R"raw(first\n\nsecond)raw"'
+    comment = '/* first\n\nsecond */'
+    source = f'struct example {{\n\n  const char* text = {literal};\n\n  {comment}\n\n  int value;\n}};\n'
+    result = self.formatted(source, ".hpp")
+    self.assertIn(literal, result)
+    self.assertIn(comment, result)
+    self.assertNotIn("\n\n", result.replace(literal, "literal").replace(comment, "comment"))
+  def test_cpp_literals_macros_templates_and_lambdas(self):
+    literal = 'R"raw(first\n\n  second\n)raw"'
+    source = f'template<class T>\nvoid test(T v) {{\n  const char* s = {literal};\n\n  visit(v, [](auto a) {{\n\nreturn a;\n}});\n}}\nBOOST_AUTO_TEST_CASE(example) {{\n\nCHECK(true);\n\n}}\n'
+    result = self.formatted(source, ".cpp")
+    self.assertIn(literal, result)
+    self.assertIn("BOOST_AUTO_TEST_CASE(example) {\n  CHECK(true);\n}", result)
+    self.assertIn("[](auto a) {\n", result)
+  def test_python_ast_literals_decorators_and_nested_units(self):
+    source = 'import os\n\nimport pathlib\n\n@decorator\ndef outer(\n    first,\n    second,\n):\n    text = """first\n\n    second\n"""\n\n    first += 1\n    def nested():\n        return 1\n    return call(\n        first,\n        second,\n    )\n'
+    result = self.formatted(source, ".py")
+    self.assertEqual(ast.dump(ast.parse(source)), ast.dump(ast.parse(result)))
+    self.assertIn('"""first\n\n    second\n"""', result)
+    self.assertIn("@decorator\ndef outer(first, second):", result)
+    self.assertIn("first += 1\n\n  def nested()", result)
+    self.assertIn("return 1\n\n  return call(first, second)", result)
+    self.assertTrue(result.startswith("import os\n\nimport pathlib\n"))
+  def test_python_no_column_limit_and_structured_literals(self):
+    name = "a" * 250
+    source = f"def test({name},\n other):\n return call({name},\nother)\n"
+    result = self.formatted(source, ".py")
+    self.assertIn(f"def test({name}, other):", result)
+    data = 'def test():\n  return call({\n    "a": 1,\n    "b": 2,\n  })\n'
+    self.assertIn('{\n', self.formatted(data, ".py"))
+  def test_python_classes_keep_members_together(self):
+    source = 'class Example:\n\n  """first\n\nsecond"""\n\n  field = 1\n\n  @staticmethod\n  def value():\n\n    return 1\n\n  class Nested:\n\n    field = 2\n\n  async def other(self):\n\n    return 2\n\nclass Next:\n\n  field = 3\n'
+    result = self.formatted(source, ".py")
+    self.assertEqual(ast.dump(ast.parse(source)), ast.dump(ast.parse(result)))
+    self.assertIn('"""first\n\nsecond"""', result)
+    self.assertNotIn("\n\n", result.split("\n\nclass Next", 1)[0].replace('"""first\n\nsecond"""', 'docstring'))
+    self.assertIn("return 2\n\nclass Next:\n  field = 3", result)
+  def test_python_tuple_continuations_use_two_spaces(self):
+    source = 'runtime_ignore_patterns = (\n    ".git",\n    ".tmp",\n    "__pycache__",\n)\n'
+    expected = 'runtime_ignore_patterns = (\n  ".git",\n  ".tmp",\n  "__pycache__",\n)\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+  def test_python_nested_collections_use_two_spaces_per_opening_line(self):
+    source = 'def example():\n    result = build({\n        "items": [\n            (\n                1,\n                2,\n            ),\n        ],\n        "names": {\n            "a",\n            "b",\n        },\n    })\n    return result\n'
+    expected = 'def example():\n  result = build({\n    "items": [\n      (\n        1,\n        2,\n      ),\n    ],\n    "names": {\n      "a",\n      "b",\n    },\n  })\n  return result\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+    self.assertEqual(self.formatted("values = [[\n    1,\n    2,\n]]\n", ".py"), "values = [[\n  1,\n  2,\n]]\n")
+  def test_python_continued_imports_headers_and_comments(self):
+    source = 'from somewhere import (\n    first,\n    second,\n)\n\n@decorator(\n    first,  # comment\n    second,\n)\ndef example(\n    first,  # comment\n    second,\n):\n    if (\n        first  # condition\n        and second\n    ):\n        return first\n'
+    expected = 'from somewhere import (\n  first,\n  second,\n)\n\n@decorator(\n  first,  # comment\n  second,\n)\ndef example(\n  first,  # comment\n  second,\n):\n  if (\n    first  # condition\n    and second\n  ):\n    return first\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+  def test_python_explicit_continuations_use_two_spaces(self):
+    source = 'def example():\n    with first() as one, \\\n            second() as two:\n        return one + \\\n            two\n'
+    expected = 'def example():\n  with first() as one, \\\n    second() as two:\n    return one + \\\n      two\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+  def test_python_continuations_preserve_multiline_literal_contents(self):
+    source = 'def example():\n    return [\n        # first value\n        f"{call({1: 2})}",\n        """first\n    second\n\n        third""",\n        f"""start {name}\n    {value}\n\n        end""",\n        "next",\n    ]\n'
+    expected = 'def example():\n  return [\n    # first value\n    f"{call({1: 2})}",\n    """first\n    second\n\n        third""",\n    f"""start {name}\n    {value}\n\n        end""",\n    "next",\n  ]\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+  def test_python_adjacent_literals_are_continuations_not_literal_contents(self):
+    source = 'def example():\n    text = (\n        "first"\n        "second"\n        f"{value}"\n        f"{other}"\n    )\n    return text\n'
+    expected = 'def example():\n  text = (\n    "first"\n    "second"\n    f"{value}"\n    f"{other}"\n  )\n  return text\n'
+    self.assertEqual(self.formatted(source, ".py"), expected)
+  def test_aliases_and_portability_exception(self):
+    cpp = "namespace std { using bool_t = decltype(true); }\nusing integer = int;\nnamespace fs = std::filesystem;\ntypedef int count;\n"
+    self.assertEqual(len(list(modulus.format_layout.alias_diagnostics(cpp, ".hpp"))), 4)
+    self.assertEqual(len(list(modulus.format_layout.alias_diagnostics(cpp, ".hpp", True))), 3)
+    python = "import pathlib as p\nfrom typing import Literal\nRole = Literal['a', 'b']\n"
+    self.assertEqual(len(list(modulus.format_layout.alias_diagnostics(python, ".py"))), 2)
+  def test_stdin_failure_has_no_output_and_never_builds_image(self):
+    with tempfile.TemporaryDirectory() as directory:
+      root = pathlib.Path(directory)
+      context = modulus.cli.build_context(root, modulus.config.normalize({"package": {"name": "sample"}}))
+      context.docker_marker = root / "not-in-container"
+      output, errors = io.StringIO(), io.StringIO()
+      with patch.dict(os.environ, {"MODULUS_DIRECT": "1"}), patch("sys.stdin", io.StringIO("def broken(")), contextlib.redirect_stdout(output), contextlib.redirect_stderr(errors):
+        self.assertEqual(modulus.formatting.launch(context, ["--stdin-filename", "test.py"]), 1)
+      self.assertEqual(output.getvalue(), "")
+      with patch.dict(os.environ, {"MODULUS_DIRECT": "0"}), patch.object(modulus.clangd, "current_image", return_value=None), patch.object(modulus.cli, "ensure_build_image") as build, contextlib.redirect_stderr(errors):
+        self.assertEqual(modulus.formatting.launch(context, ["--check"]), 1)
+      build.assert_not_called()
+  def test_discovery_excludes_generated_and_symlinks(self):
+    with tempfile.TemporaryDirectory() as directory:
+      root = pathlib.Path(directory)
+      subprocess.run(["git", "init", "-q", str(root)], check=True)
+      for name in ["keep.cpp", "skip.cpp", "build/generated.cpp", "vendor/library.cpp"]:
+        path = root / name
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.touch()
+      (root / "link.cpp").symlink_to(root / "keep.cpp")
+      self.assertEqual(modulus.formatting.source_files(root, [], ["skip.cpp"]), [root / "keep.cpp"])
+  def test_nested_lists_join_in_one_pass_without_changing_string_spaces(self):
+    cpp = 'void example() { outer(\n inner(\n 1,\n 2),\n "a  b"); }\n'
+    self.assertIn('outer(inner(1, 2), "a  b");', self.formatted(cpp, ".cpp"))
+    python = 'def example():\n  return outer(\n    inner(\n      1,\n      2),\n    "a  b")\n'
+    self.assertIn('outer(inner(1, 2), "a  b")', self.formatted(python, ".py"))
+  def test_multiline_fstrings_preserve_blank_lines_on_new_python_tokenizers(self):
+    source = 'def example(name):\n  return f"""first {name}\n\nsecond\n\nthird"""\n'
+    self.assertEqual(self.formatted(source, ".py"), source)
+  def test_alias_check_distinguishes_module_aliases_from_saved_callbacks(self):
+    source = 'import modulus.cli\nsaved = modulus.cli.docker_run\nconfig = modulus.cli\nItems = list[int]\n'
+    self.assertEqual([line for line, _ in modulus.format_layout.alias_diagnostics(source, ".py")], [3, 4])
+  def test_exported_stdin_formatter_needs_no_compile_database(self):
+    with tempfile.TemporaryDirectory() as directory:
+      root = pathlib.Path(directory)
+      (root / "modulus.json").write_text(json.dumps({"package": {"name": "sample"}}))
+      modulus.export.sync_project(root, pathlib.Path(__file__).resolve().parent.parent)
+      result = subprocess.run([sys.executable, str(root / "build/build_tools/format.py"), "--stdin-filename", "sample.py"], cwd=root, env=dict(os.environ, MODULUS_DIRECT="1", PYTHONDONTWRITEBYTECODE="1"), input="def sample():\n    values = (\n        1,\n        2,\n    )\n    return values\n", text=True, capture_output=True)
+      self.assertEqual(result.returncode, 0, result.stderr)
+      self.assertEqual(result.stdout, "def sample():\n  values = (\n    1,\n    2,\n  )\n  return values\n")
+      self.assertEqual(result.stderr, "")
+  @unittest.skipUnless(os.environ.get("MODULUS_FORMAT_REPO") and os.environ.get("MODULUS_CONFORM_PATH"), "set MODULUS_FORMAT_REPO and MODULUS_CONFORM_PATH for the Neovim integration test")
+  def test_neovim_container_formatting(self):
+    root = pathlib.Path(__file__).resolve().parent.parent
+    env = dict(os.environ, MODULUS_FORMAT_HELPER=str(root / "tools/nvim/modulus.lua"))
+    result = subprocess.run(["nvim", "--headless", "-u", "NONE", "-l", str(root / "tests/nvim_format.lua")], env=env, text=True, capture_output=True, timeout=60)
+    self.assertEqual(result.returncode, 0, result.stdout + result.stderr)
+    self.assertIn("MODULUS_FORMAT_OK", result.stdout + result.stderr)
+
+if __name__ == "__main__":
+  unittest.main()