import ast
import platform
import sys
-from typing import Iterable, Iterator, List, Set, Union, Tuple
+from typing import Any, Iterable, Iterator, List, Set, Tuple, Type, Union
+
+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 blib2to3 import pygram, pytree
+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 black.mode import TargetVersion, Feature, supports_feature
from black.nodes import syms
+ast3: Any
+ast27: Any
+
_IS_PYPY = platform.python_implementation() == "PyPy"
try:
ast3 = ast27 = ast
+PY310_HINT: Final[
+ str
+] = "Consider using --target-version py310 to parse Python 3.10 code."
+
+
class InvalidInput(ValueError):
"""Raised when input source code fails all parse attempts."""
# 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):
- # Python 3.7+
+ if not supports_feature(
+ target_versions, Feature.ASYNC_IDENTIFIERS
+ ) and not supports_feature(target_versions, Feature.PATTERN_MATCHING):
+ # Python 3.7-3.9
grammars.append(
pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords
)
if not src_txt.endswith("\n"):
src_txt += "\n"
- for grammar in get_grammars(set(target_versions)):
- drv = driver.Driver(grammar, pytree.convert)
+ grammars = get_grammars(set(target_versions))
+ for grammar in grammars:
+ drv = driver.Driver(grammar)
try:
result = drv.parse_string(src_txt, True)
break
except IndexError:
faulty_line = "<line number missing in source>"
exc = 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]}")
+
else:
+ if pygram.python_grammar_soft_keywords not in grammars and matches_grammar(
+ src_txt, pygram.python_grammar_soft_keywords
+ ):
+ original_msg = exc.args[0]
+ msg = f"{original_msg}\n{PY310_HINT}"
+ raise InvalidInput(msg) from None
raise exc from None
if isinstance(result, Leaf):
return result
+def matches_grammar(src_txt: str, grammar: Grammar) -> bool:
+ drv = driver.Driver(grammar)
+ try:
+ drv.parse_string(src_txt, True)
+ except ParseError:
+ return False
+ else:
+ return True
+
+
def lib2to3_unparse(node: Node) -> str:
"""Given a lib2to3 node, return its string representation."""
code = str(node)
raise SyntaxError(first_error)
+ast3_AST: Final[Type[ast3.AST]] = ast3.AST
+ast27_AST: Final[Type[ast27.AST]] = ast27.AST
+
+
def stringify_ast(
node: Union[ast.AST, ast3.AST, ast27.AST], depth: int = 0
) -> Iterator[str]:
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)):
elif isinstance(item, (ast.AST, ast3.AST, ast27.AST)):
yield from stringify_ast(item, depth + 2)
- elif isinstance(value, (ast.AST, ast3.AST, ast27.AST)):
+ # 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)):
yield from stringify_ast(value, depth + 2)
else: