]> git.madduck.net Git - etc/vim.git/blobdiff - src/black/parsing.py

madduck's git repository

Every one of the projects in this repository is available at the canonical URL git://git.madduck.net/madduck/pub/<projectpath> — see each project's metadata for the exact URL.

All patches and comments are welcome. Please squash your changes to logical commits before using git-format-patch and git-send-email to patches@git.madduck.net. If you'd read over the Git project's submission guidelines and adhered to them, I'd be especially grateful.

SSH access, as well as push access can be individually arranged.

If you use my repositories frequently, consider adding the following snippet to ~/.gitconfig and using the third clone URL listed for each project:

[url "git://git.madduck.net/madduck/"]
  insteadOf = madduck:

Fix crash on type comment with trailing space (#3773)
[etc/vim.git] / src / black / parsing.py
index b673027022f44b4a2ebd5a9b4b3bd44d0b163b3b..e98e019cac6e9a1955f91cc35553a1b7393951dd 100644 (file)
@@ -2,45 +2,19 @@
 Parse Python code and perform AST validation.
 """
 import ast
-import platform
 import sys
-from typing import Any, Iterable, Iterator, List, Set, Tuple, Type, Union
+from typing import Final, Iterable, Iterator, List, Set, Tuple
 
-if sys.version_info < (3, 8):
-    from typing_extensions import Final
-else:
-    from typing import Final
-
-# lib2to3 fork
-from blib2to3.pytree import Node, Leaf
+from black.mode import VERSION_TO_FEATURES, Feature, TargetVersion, supports_feature
+from black.nodes import syms
 from blib2to3 import pygram
 from blib2to3.pgen2 import driver
 from blib2to3.pgen2.grammar import Grammar
 from blib2to3.pgen2.parse import ParseError
 from blib2to3.pgen2.tokenize import TokenError
+from blib2to3.pytree import Leaf, Node
 
-from black.mode import TargetVersion, Feature, supports_feature
-from black.nodes import syms
-
-ast3: Any
-ast27: Any
-
-_IS_PYPY = platform.python_implementation() == "PyPy"
-
-try:
-    from typed_ast import ast3, ast27
-except ImportError:
-    # Either our python version is too low, or we're on pypy
-    if sys.version_info < (3, 7) or (sys.version_info < (3, 8) and not _IS_PYPY):
-        print(
-            "The typed_ast package is required but not installed.\n"
-            "You can upgrade to Python 3.8+ or install typed_ast with\n"
-            "`python3 -m pip install typed-ast`.",
-            file=sys.stderr,
-        )
-        sys.exit(1)
-    else:
-        ast3 = ast27 = ast
+PY2_HINT: Final = "Python 2 support was removed in version 22.0."
 
 
 class InvalidInput(ValueError):
@@ -51,30 +25,15 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
     if not target_versions:
         # No target_version specified, so try all grammars.
         return [
-            # Python 3.7+
+            # Python 3.7-3.9
             pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords,
             # Python 3.0-3.6
             pygram.python_grammar_no_print_statement_no_exec_statement,
-            # Python 2.7 with future print_function import
-            pygram.python_grammar_no_print_statement,
-            # Python 2.7
-            pygram.python_grammar,
+            # Python 3.10+
+            pygram.python_grammar_soft_keywords,
         ]
 
-    if all(version.is_python2() for version in target_versions):
-        # Python 2-only code, so try Python 2 grammars.
-        return [
-            # Python 2.7 with future print_function import
-            pygram.python_grammar_no_print_statement,
-            # Python 2.7
-            pygram.python_grammar,
-        ]
-
-    # Python 3-compatible code, so only try Python 3 grammar.
     grammars = []
-    if supports_feature(target_versions, Feature.PATTERN_MATCHING):
-        # Python 3.10+
-        grammars.append(pygram.python_grammar_soft_keywords)
     # If we have to parse both, try to parse async as a keyword first
     if not supports_feature(
         target_versions, Feature.ASYNC_IDENTIFIERS
@@ -86,6 +45,10 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
     if not supports_feature(target_versions, Feature.ASYNC_KEYWORDS):
         # Python 3.0-3.6
         grammars.append(pygram.python_grammar_no_print_statement_no_exec_statement)
+    if any(Feature.PATTERN_MATCHING in VERSION_TO_FEATURES[v] for v in target_versions):
+        # Python 3.10+
+        grammars.append(pygram.python_grammar_soft_keywords)
+
     # At least one of the above branches must have been taken, because every Python
     # version has exactly one of the two 'ASYNC_*' flags
     return grammars
@@ -96,7 +59,9 @@ def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -
     if not src_txt.endswith("\n"):
         src_txt += "\n"
 
-    for grammar in get_grammars(set(target_versions)):
+    grammars = get_grammars(set(target_versions))
+    errors = {}
+    for grammar in grammars:
         drv = driver.Driver(grammar)
         try:
             result = drv.parse_string(src_txt, True)
@@ -109,14 +74,29 @@ def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -
                 faulty_line = lines[lineno - 1]
             except IndexError:
                 faulty_line = "<line number missing in source>"
-            exc = InvalidInput(f"Cannot parse: {lineno}:{column}: {faulty_line}")
+            errors[grammar.version] = InvalidInput(
+                f"Cannot parse: {lineno}:{column}: {faulty_line}"
+            )
 
         except TokenError as te:
             # In edge cases these are raised; and typically don't have a "faulty_line".
             lineno, column = te.args[1]
-            exc = InvalidInput(f"Cannot parse: {lineno}:{column}: {te.args[0]}")
+            errors[grammar.version] = InvalidInput(
+                f"Cannot parse: {lineno}:{column}: {te.args[0]}"
+            )
 
     else:
+        # Choose the latest version when raising the actual parsing error.
+        assert len(errors) >= 1
+        exc = errors[max(errors)]
+
+        if matches_grammar(src_txt, pygram.python_grammar) or matches_grammar(
+            src_txt, pygram.python_grammar_no_print_statement
+        ):
+            original_msg = exc.args[0]
+            msg = f"{original_msg}\n{PY2_HINT}"
+            raise InvalidInput(msg) from None
+
         raise exc from None
 
     if isinstance(result, Leaf):
@@ -124,6 +104,16 @@ def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -
     return result
 
 
+def matches_grammar(src_txt: str, grammar: Grammar) -> bool:
+    drv = driver.Driver(grammar)
+    try:
+        drv.parse_string(src_txt, True)
+    except (ParseError, TokenError, IndentationError):
+        return False
+    else:
+        return True
+
+
 def lib2to3_unparse(node: Node) -> str:
     """Given a lib2to3 node, return its string representation."""
     code = str(node)
@@ -131,66 +121,68 @@ def lib2to3_unparse(node: Node) -> str:
 
 
 def parse_single_version(
-    src: str, version: Tuple[int, int]
-) -> Union[ast.AST, ast3.AST, ast27.AST]:
+    src: str, version: Tuple[int, int], *, type_comments: bool
+) -> ast.AST:
     filename = "<unknown>"
-    # typed_ast is needed because of feature version limitations in the builtin ast
-    if sys.version_info >= (3, 8) and version >= (3,):
-        return ast.parse(src, filename, feature_version=version)
-    elif version >= (3,):
-        if _IS_PYPY:
-            return ast3.parse(src, filename)
-        else:
-            return ast3.parse(src, filename, feature_version=version[1])
-    elif version == (2, 7):
-        return ast27.parse(src)
-    raise AssertionError("INTERNAL ERROR: Tried parsing unsupported Python version!")
+    return ast.parse(
+        src, filename, feature_version=version, type_comments=type_comments
+    )
 
 
-def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]:
+def parse_ast(src: str) -> ast.AST:
     # TODO: support Python 4+ ;)
     versions = [(3, minor) for minor in range(3, sys.version_info[1] + 1)]
 
-    if ast27.__name__ != "ast":
-        versions.append((2, 7))
-
     first_error = ""
     for version in sorted(versions, reverse=True):
         try:
-            return parse_single_version(src, version)
+            return parse_single_version(src, version, type_comments=True)
         except SyntaxError as e:
             if not first_error:
                 first_error = str(e)
 
+    # Try to parse without type comments
+    for version in sorted(versions, reverse=True):
+        try:
+            return parse_single_version(src, version, type_comments=False)
+        except SyntaxError:
+            pass
+
     raise SyntaxError(first_error)
 
 
-ast3_AST: Final[Type[ast3.AST]] = ast3.AST
-ast27_AST: Final[Type[ast27.AST]] = ast27.AST
+def _normalize(lineend: str, value: str) -> str:
+    # To normalize, we strip any leading and trailing space from
+    # each line...
+    stripped: List[str] = [i.strip() for i in value.splitlines()]
+    normalized = lineend.join(stripped)
+    # ...and remove any blank lines at the beginning and end of
+    # the whole string
+    return normalized.strip()
 
 
-def stringify_ast(
-    node: Union[ast.AST, ast3.AST, ast27.AST], depth: int = 0
-) -> Iterator[str]:
+def stringify_ast(node: ast.AST, depth: int = 0) -> Iterator[str]:
     """Simple visitor generating strings to compare ASTs by content."""
 
-    node = fixup_ast_constants(node)
+    if (
+        isinstance(node, ast.Constant)
+        and isinstance(node.value, str)
+        and node.kind == "u"
+    ):
+        # It's a quirk of history that we strip the u prefix over here. We used to
+        # rewrite the AST nodes for Python version compatibility and we never copied
+        # over the kind
+        node.kind = None
 
     yield f"{'  ' * depth}{node.__class__.__name__}("
 
-    type_ignore_classes: Tuple[Type[Any], ...]
     for field in sorted(node._fields):  # noqa: F402
-        # TypeIgnore will not be present using pypy < 3.8, so need for this
-        if not (_IS_PYPY and sys.version_info < (3, 8)):
-            # TypeIgnore has only one field 'lineno' which breaks this comparison
-            type_ignore_classes = (ast3.TypeIgnore, ast27.TypeIgnore)
-            if sys.version_info >= (3, 8):
-                type_ignore_classes += (ast.TypeIgnore,)
-            if isinstance(node, type_ignore_classes):
-                break
+        # TypeIgnore has only one field 'lineno' which breaks this comparison
+        if isinstance(node, ast.TypeIgnore):
+            break
 
         try:
-            value = getattr(node, field)
+            value: object = getattr(node, field)
         except AttributeError:
             continue
 
@@ -202,62 +194,34 @@ def stringify_ast(
                 # parentheses and they change the AST.
                 if (
                     field == "targets"
-                    and isinstance(node, (ast.Delete, ast3.Delete, ast27.Delete))
-                    and isinstance(item, (ast.Tuple, ast3.Tuple, ast27.Tuple))
+                    and isinstance(node, ast.Delete)
+                    and isinstance(item, ast.Tuple)
                 ):
-                    for item in item.elts:
-                        yield from stringify_ast(item, depth + 2)
+                    for elt in item.elts:
+                        yield from stringify_ast(elt, depth + 2)
 
-                elif isinstance(item, (ast.AST, ast3.AST, ast27.AST)):
+                elif isinstance(item, ast.AST):
                     yield from stringify_ast(item, depth + 2)
 
-        # Note that we are referencing the typed-ast ASTs via global variables and not
-        # direct module attribute accesses because that breaks mypyc. It's probably
-        # something to do with the ast3 / ast27 variables being marked as Any leading
-        # mypy to think this branch is always taken, leaving the rest of the code
-        # unanalyzed. Tighting up the types for the typed-ast AST types avoids the
-        # mypyc crash.
-        elif isinstance(value, (ast.AST, ast3_AST, ast27_AST)):
+        elif isinstance(value, ast.AST):
             yield from stringify_ast(value, depth + 2)
 
         else:
-            # Constant strings may be indented across newlines, if they are
-            # docstrings; fold spaces after newlines when comparing. Similarly,
-            # trailing and leading space may be removed.
-            # Note that when formatting Python 2 code, at least with Windows
-            # line-endings, docstrings can end up here as bytes instead of
-            # str so make sure that we handle both cases.
+            normalized: object
             if (
                 isinstance(node, ast.Constant)
                 and field == "value"
-                and isinstance(value, (str, bytes))
+                and isinstance(value, str)
             ):
-                lineend = "\n" if isinstance(value, str) else b"\n"
-                # To normalize, we strip any leading and trailing space from
-                # each line...
-                stripped = [line.strip() for line in value.splitlines()]
-                normalized = lineend.join(stripped)  # type: ignore[attr-defined]
-                # ...and remove any blank lines at the beginning and end of
-                # the whole string
-                normalized = normalized.strip()
+                # Constant strings may be indented across newlines, if they are
+                # docstrings; fold spaces after newlines when comparing. Similarly,
+                # trailing and leading space may be removed.
+                normalized = _normalize("\n", value)
+            elif field == "type_comment" and isinstance(value, str):
+                # Trailing whitespace in type comments is removed.
+                normalized = value.rstrip()
             else:
                 normalized = value
             yield f"{'  ' * (depth+2)}{normalized!r},  # {value.__class__.__name__}"
 
     yield f"{'  ' * depth})  # /{node.__class__.__name__}"
-
-
-def fixup_ast_constants(
-    node: Union[ast.AST, ast3.AST, ast27.AST]
-) -> Union[ast.AST, ast3.AST, ast27.AST]:
-    """Map ast nodes deprecated in 3.8 to Constant."""
-    if isinstance(node, (ast.Str, ast3.Str, ast27.Str, ast.Bytes, ast3.Bytes)):
-        return ast.Constant(value=node.s)
-
-    if isinstance(node, (ast.Num, ast3.Num, ast27.Num)):
-        return ast.Constant(value=node.n)
-
-    if isinstance(node, (ast.NameConstant, ast3.NameConstant)):
-        return ast.Constant(value=node.value)
-
-    return node