Add support for always tokenizing async/await as keywords

[etc/vim.git] / black.py

diff --git a/black.py b/black.py
index 9ecfbe18fe8443a6bb9c7640de036523127cad35..c8aa30b812b49795c797e0e262159ebda024f67a 100644 (file)
--- a/black.py
+++ b/black.py
@@ -40,6 +40,7 @@ from appdirs import user_cache_dir
 from attr import dataclass, evolve, Factory
 import click
 import toml
 from attr import dataclass, evolve, Factory
 import click
 import toml

+from typed_ast import ast3, ast27

 
 # lib2to3 fork
 from blib2to3.pytree import Node, Leaf, type_repr
 
 # lib2to3 fork
 from blib2to3.pytree import Node, Leaf, type_repr


@@ -47,6 +48,7 @@ from blib2to3 import pygram, pytree
 from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.grammar import Grammar
 from blib2to3.pgen2.parse import ParseError
 from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.grammar import Grammar
 from blib2to3.pgen2.parse import ParseError

+from blib2to3.pgen2.tokenize import TokenizerConfig

 
 
 __version__ = "19.3b0"
 
 
 __version__ = "19.3b0"


@@ -135,19 +137,28 @@ class Feature(Enum):
     NUMERIC_UNDERSCORES = 3
     TRAILING_COMMA_IN_CALL = 4
     TRAILING_COMMA_IN_DEF = 5
     NUMERIC_UNDERSCORES = 3
     TRAILING_COMMA_IN_CALL = 4
     TRAILING_COMMA_IN_DEF = 5

+    # The following two feature-flags are mutually exclusive, and exactly one should be
+    # set for every version of python.
+    ASYNC_IS_VALID_IDENTIFIER = 6
+    ASYNC_IS_RESERVED_KEYWORD = 7

 
 
 VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
 
 
 VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {

-    TargetVersion.PY27: set(),
-    TargetVersion.PY33: {Feature.UNICODE_LITERALS},
-    TargetVersion.PY34: {Feature.UNICODE_LITERALS},
-    TargetVersion.PY35: {Feature.UNICODE_LITERALS, Feature.TRAILING_COMMA_IN_CALL},
+    TargetVersion.PY27: {Feature.ASYNC_IS_VALID_IDENTIFIER},
+    TargetVersion.PY33: {Feature.UNICODE_LITERALS, Feature.ASYNC_IS_VALID_IDENTIFIER},
+    TargetVersion.PY34: {Feature.UNICODE_LITERALS, Feature.ASYNC_IS_VALID_IDENTIFIER},
+    TargetVersion.PY35: {
+        Feature.UNICODE_LITERALS,
+        Feature.TRAILING_COMMA_IN_CALL,
+        Feature.ASYNC_IS_VALID_IDENTIFIER,
+    },

     TargetVersion.PY36: {
         Feature.UNICODE_LITERALS,
         Feature.F_STRINGS,
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,
     TargetVersion.PY36: {
         Feature.UNICODE_LITERALS,
         Feature.F_STRINGS,
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,

+        Feature.ASYNC_IS_VALID_IDENTIFIER,

     },
     TargetVersion.PY37: {
         Feature.UNICODE_LITERALS,
     },
     TargetVersion.PY37: {
         Feature.UNICODE_LITERALS,


@@ -155,6 +166,7 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,

+        Feature.ASYNC_IS_RESERVED_KEYWORD,

     },
     TargetVersion.PY38: {
         Feature.UNICODE_LITERALS,
     },
     TargetVersion.PY38: {
         Feature.UNICODE_LITERALS,


@@ -162,6 +174,7 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,
         Feature.NUMERIC_UNDERSCORES,
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,

+        Feature.ASYNC_IS_RESERVED_KEYWORD,

     },
 }
 
     },
 }
 


@@ -231,6 +244,7 @@ def read_pyproject_toml(
 
 
 @click.command(context_settings=dict(help_option_names=["-h", "--help"]))
 
 
 @click.command(context_settings=dict(help_option_names=["-h", "--help"]))

+@click.option("-c", "--code", type=str, help="Format the code passed in as a string.")

 @click.option(
     "-l",
     "--line-length",
 @click.option(
     "-l",
     "--line-length",


@@ -357,6 +371,7 @@ def read_pyproject_toml(
 @click.pass_context
 def main(
     ctx: click.Context,
 @click.pass_context
 def main(
     ctx: click.Context,

+    code: Optional[str],

     line_length: int,
     target_version: List[TargetVersion],
     check: bool,
     line_length: int,
     target_version: List[TargetVersion],
     check: bool,


@@ -397,6 +412,9 @@ def main(
     )
     if config and verbose:
         out(f"Using configuration from {config}.", bold=False, fg="blue")
     )
     if config and verbose:
         out(f"Using configuration from {config}.", bold=False, fg="blue")

+    if code is not None:
+        print(format_str(code, mode=mode))
+        ctx.exit(0)

     try:
         include_regex = re_compile_maybe_verbose(include)
     except re.error:
     try:
         include_regex = re_compile_maybe_verbose(include)
     except re.error:


@@ -435,22 +453,10 @@ def main(
             report=report,
         )
     else:
             report=report,
         )
     else:

-        loop = asyncio.get_event_loop()
-        executor = ProcessPoolExecutor(max_workers=os.cpu_count())
-        try:
-            loop.run_until_complete(
-                schedule_formatting(
-                    sources=sources,
-                    fast=fast,
-                    write_back=write_back,
-                    mode=mode,
-                    report=report,
-                    loop=loop,
-                    executor=executor,
-                )
-            )
-        finally:
-            shutdown(loop)
+        reformat_many(
+            sources=sources, fast=fast, write_back=write_back, mode=mode, report=report
+        )
+

     if verbose or not quiet:
         bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨"
         out(f"All done! {bang}")
     if verbose or not quiet:
         bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨"
         out(f"All done! {bang}")


@@ -492,6 +498,36 @@ def reformat_one(
         report.failed(src, str(exc))
 
 
         report.failed(src, str(exc))
 
 

+def reformat_many(
+    sources: Set[Path],
+    fast: bool,
+    write_back: WriteBack,
+    mode: FileMode,
+    report: "Report",
+) -> None:
+    """Reformat multiple files using a ProcessPoolExecutor."""
+    loop = asyncio.get_event_loop()
+    worker_count = os.cpu_count()
+    if sys.platform == "win32":
+        # Work around https://bugs.python.org/issue26903
+        worker_count = min(worker_count, 61)
+    executor = ProcessPoolExecutor(max_workers=worker_count)
+    try:
+        loop.run_until_complete(
+            schedule_formatting(
+                sources=sources,
+                fast=fast,
+                write_back=write_back,
+                mode=mode,
+                report=report,
+                loop=loop,
+                executor=executor,
+            )
+        )
+    finally:
+        shutdown(loop)
+
+

 async def schedule_formatting(
     sources: Set[Path],
     fast: bool,
 async def schedule_formatting(
     sources: Set[Path],
     fast: bool,


@@ -724,20 +760,62 @@ def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]:
         return tiow.read(), encoding, newline
 
 
         return tiow.read(), encoding, newline
 
 

-def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
+@dataclass(frozen=True)
+class ParserConfig:
+    grammar: Grammar
+    tokenizer_config: TokenizerConfig = TokenizerConfig()
+
+
+def get_parser_configs(target_versions: Set[TargetVersion]) -> List[ParserConfig]:

     if not target_versions:
         # No target_version specified, so try all grammars.
         return [
     if not target_versions:
         # No target_version specified, so try all grammars.
         return [

-            pygram.python_grammar_no_print_statement_no_exec_statement,
-            pygram.python_grammar_no_print_statement,
-            pygram.python_grammar,
+            # Python 3.7+
+            ParserConfig(
+                pygram.python_grammar_no_print_statement_no_exec_statement,
+                TokenizerConfig(async_is_reserved_keyword=True),
+            ),
+            # Python 3.0-3.6
+            ParserConfig(
+                pygram.python_grammar_no_print_statement_no_exec_statement,
+                TokenizerConfig(async_is_reserved_keyword=False),
+            ),
+            # Python 2.7 with future print_function import
+            ParserConfig(pygram.python_grammar_no_print_statement),
+            # Python 2.7
+            ParserConfig(pygram.python_grammar),

         ]
     elif all(version.is_python2() for version in target_versions):
         # Python 2-only code, so try Python 2 grammars.
         ]
     elif all(version.is_python2() for version in target_versions):
         # Python 2-only code, so try Python 2 grammars.

-        return [pygram.python_grammar_no_print_statement, pygram.python_grammar]
+        return [
+            # Python 2.7 with future print_function import
+            ParserConfig(pygram.python_grammar_no_print_statement),
+            # Python 2.7
+            ParserConfig(pygram.python_grammar),
+        ]

     else:
         # Python 3-compatible code, so only try Python 3 grammar.
     else:
         # Python 3-compatible code, so only try Python 3 grammar.

-        return [pygram.python_grammar_no_print_statement_no_exec_statement]
+        configs = []
+        # If we have to parse both, try to parse async as a keyword first
+        if not supports_feature(target_versions, Feature.ASYNC_IS_VALID_IDENTIFIER):
+            # Python 3.7+
+            configs.append(
+                ParserConfig(
+                    pygram.python_grammar_no_print_statement_no_exec_statement,
+                    TokenizerConfig(async_is_reserved_keyword=True),
+                )
+            )
+        if not supports_feature(target_versions, Feature.ASYNC_IS_RESERVED_KEYWORD):
+            # Python 3.0-3.6
+            configs.append(
+                ParserConfig(
+                    pygram.python_grammar_no_print_statement_no_exec_statement,
+                    TokenizerConfig(async_is_reserved_keyword=False),
+                )
+            )
+        # At least one of the above branches must have been taken, because every Python
+        # version has exactly one of the two 'ASYNC_IS_*' flags
+        return configs

 
 
 def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node:
 
 
 def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node:


@@ -745,8 +823,12 @@ def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -
     if src_txt[-1:] != "\n":
         src_txt += "\n"
 
     if src_txt[-1:] != "\n":
         src_txt += "\n"
 

-    for grammar in get_grammars(set(target_versions)):
-        drv = driver.Driver(grammar, pytree.convert)
+    for parser_config in get_parser_configs(set(target_versions)):
+        drv = driver.Driver(
+            parser_config.grammar,
+            pytree.convert,
+            tokenizer_config=parser_config.tokenizer_config,
+        )

         try:
             result = drv.parse_string(src_txt, True)
             break
         try:
             result = drv.parse_string(src_txt, True)
             break


@@ -2122,15 +2204,21 @@ def list_comments(prefix: str, *, is_endmarker: bool) -> List[ProtoComment]:
 
     consumed = 0
     nlines = 0
 
     consumed = 0
     nlines = 0

+    ignored_lines = 0

     for index, line in enumerate(prefix.split("\n")):
         consumed += len(line) + 1  # adding the length of the split '\n'
         line = line.lstrip()
         if not line:
             nlines += 1
         if not line.startswith("#"):
     for index, line in enumerate(prefix.split("\n")):
         consumed += len(line) + 1  # adding the length of the split '\n'
         line = line.lstrip()
         if not line:
             nlines += 1
         if not line.startswith("#"):

+            # Escaped newlines outside of a comment are not really newlines at
+            # all. We treat a single-line comment following an escaped newline
+            # as a simple trailing comment.
+            if line.endswith("\\"):
+                ignored_lines += 1

             continue
 
             continue
 

-        if index == 0 and not is_endmarker:
+        if index == ignored_lines and not is_endmarker:

             comment_type = token.COMMENT  # simple trailing comment
         else:
             comment_type = STANDALONE_COMMENT
             comment_type = token.COMMENT  # simple trailing comment
         else:
             comment_type = STANDALONE_COMMENT


@@ -2726,9 +2814,17 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
 
     check_lpar = False
     for index, child in enumerate(list(node.children)):
 
     check_lpar = False
     for index, child in enumerate(list(node.children)):

+        # Add parentheses around long tuple unpacking in assignments.
+        if (
+            index == 0
+            and isinstance(child, Node)
+            and child.type == syms.testlist_star_expr
+        ):
+            check_lpar = True
+

         if check_lpar:
             if child.type == syms.atom:
         if check_lpar:
             if child.type == syms.atom:

-                if maybe_make_parens_invisible_in_atom(child):
+                if maybe_make_parens_invisible_in_atom(child, parent=node):

                     lpar = Leaf(token.LPAR, "")
                     rpar = Leaf(token.RPAR, "")
                     index = child.remove() or 0
                     lpar = Leaf(token.LPAR, "")
                     rpar = Leaf(token.RPAR, "")
                     index = child.remove() or 0


@@ -2757,7 +2853,11 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
                 lpar = Leaf(token.LPAR, "")
                 rpar = Leaf(token.RPAR, "")
                 index = child.remove() or 0
                 lpar = Leaf(token.LPAR, "")
                 rpar = Leaf(token.RPAR, "")
                 index = child.remove() or 0

-                node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
+                prefix = child.prefix
+                child.prefix = ""
+                new_child = Node(syms.atom, [lpar, child, rpar])
+                new_child.prefix = prefix
+                node.insert_child(index, new_child)

 
         check_lpar = isinstance(child, Leaf) and child.value in parens_after
 
 
         check_lpar = isinstance(child, Leaf) and child.value in parens_after
 


@@ -2839,7 +2939,7 @@ def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]:
         container = container.next_sibling
 
 
         container = container.next_sibling
 
 

-def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
+def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:

     """If it's safe, make the parens in the atom `node` invisible, recursively.
 
     Returns whether the node should itself be wrapped in invisible parentheses.
     """If it's safe, make the parens in the atom `node` invisible, recursively.
 
     Returns whether the node should itself be wrapped in invisible parentheses.


@@ -2849,7 +2949,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
         node.type != syms.atom
         or is_empty_tuple(node)
         or is_one_tuple(node)
         node.type != syms.atom
         or is_empty_tuple(node)
         or is_one_tuple(node)

-        or is_yield(node)
+        or (is_yield(node) and parent.type != syms.expr_stmt)

         or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
     ):
         return False
         or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
     ):
         return False


@@ -2861,7 +2961,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
         first.value = ""  # type: ignore
         last.value = ""  # type: ignore
         if len(node.children) > 1:
         first.value = ""  # type: ignore
         last.value = ""  # type: ignore
         if len(node.children) > 1:

-            maybe_make_parens_invisible_in_atom(node.children[1])
+            maybe_make_parens_invisible_in_atom(node.children[1], parent=parent)

         return False
 
     return True
         return False
 
     return True


@@ -3339,17 +3439,34 @@ class Report:
         return ", ".join(report) + "."
 
 
         return ", ".join(report) + "."
 
 

+def parse_ast(src: str) -> Union[ast3.AST, ast27.AST]:
+    for feature_version in (7, 6):
+        try:
+            return ast3.parse(src, feature_version=feature_version)
+        except SyntaxError:
+            continue
+
+    return ast27.parse(src)
+
+

 def assert_equivalent(src: str, dst: str) -> None:
     """Raise AssertionError if `src` and `dst` aren't equivalent."""
 
 def assert_equivalent(src: str, dst: str) -> None:
     """Raise AssertionError if `src` and `dst` aren't equivalent."""
 

-    import ast

     import traceback
 
     import traceback
 

-    def _v(node: ast.AST, depth: int = 0) -> Iterator[str]:
+    def _v(node: Union[ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]:

         """Simple visitor generating strings to compare ASTs by content."""
         yield f"{'  ' * depth}{node.__class__.__name__}("
 
         for field in sorted(node._fields):
         """Simple visitor generating strings to compare ASTs by content."""
         yield f"{'  ' * depth}{node.__class__.__name__}("
 
         for field in sorted(node._fields):

+            # TypeIgnore has only one field 'lineno' which breaks this comparison
+            if isinstance(node, (ast3.TypeIgnore, ast27.TypeIgnore)):
+                break
+
+            # Ignore str kind which is case sensitive / and ignores unicode_literals
+            if isinstance(node, (ast3.Str, ast27.Str, ast3.Bytes)) and field == "kind":
+                continue
+

             try:
                 value = getattr(node, field)
             except AttributeError:
             try:
                 value = getattr(node, field)
             except AttributeError:


@@ -3363,15 +3480,15 @@ def assert_equivalent(src: str, dst: str) -> None:
                     # parentheses and they change the AST.
                     if (
                         field == "targets"
                     # parentheses and they change the AST.
                     if (
                         field == "targets"

-                        and isinstance(node, ast.Delete)
-                        and isinstance(item, ast.Tuple)
+                        and isinstance(node, (ast3.Delete, ast27.Delete))
+                        and isinstance(item, (ast3.Tuple, ast27.Tuple))

                     ):
                         for item in item.elts:
                             yield from _v(item, depth + 2)
                     ):
                         for item in item.elts:
                             yield from _v(item, depth + 2)

-                    elif isinstance(item, ast.AST):
+                    elif isinstance(item, (ast3.AST, ast27.AST)):

                         yield from _v(item, depth + 2)
 
                         yield from _v(item, depth + 2)
 

-            elif isinstance(value, ast.AST):
+            elif isinstance(value, (ast3.AST, ast27.AST)):

                 yield from _v(value, depth + 2)
 
             else:
                 yield from _v(value, depth + 2)
 
             else:


@@ -3380,17 +3497,15 @@ def assert_equivalent(src: str, dst: str) -> None:
         yield f"{'  ' * depth})  # /{node.__class__.__name__}"
 
     try:
         yield f"{'  ' * depth})  # /{node.__class__.__name__}"
 
     try:

-        src_ast = ast.parse(src)
+        src_ast = parse_ast(src)

     except Exception as exc:
     except Exception as exc:

-        major, minor = sys.version_info[:2]

         raise AssertionError(
         raise AssertionError(

-            f"cannot use --safe with this file; failed to parse source file "
-            f"with Python {major}.{minor}'s builtin AST. Re-run with --fast "
-            f"or stop using deprecated Python 2 syntax. AST error message: {exc}"
+            f"cannot use --safe with this file; failed to parse source file.  "
+            f"AST error message: {exc}"

         )
 
     try:
         )
 
     try:

-        dst_ast = ast.parse(dst)
+        dst_ast = parse_ast(dst)

     except Exception as exc:
         log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
         raise AssertionError(
     except Exception as exc:
         log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
         raise AssertionError(