]> git.madduck.net Git - etc/vim.git/blobdiff - black.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:

Require regex version 2019.8
[etc/vim.git] / black.py
index 29d56c923f6fa65c5e8b3d5bc83097e570137f2f..ff373c863d45b71849cb0ae0b1eb800af54c10a7 100644 (file)
--- a/black.py
+++ b/black.py
@@ -1,5 +1,7 @@
+import ast
 import asyncio
 from concurrent.futures import Executor, ProcessPoolExecutor
+from contextlib import contextmanager
 from datetime import datetime
 from enum import Enum
 from functools import lru_cache, partial, wraps
@@ -10,7 +12,7 @@ from multiprocessing import Manager, freeze_support
 import os
 from pathlib import Path
 import pickle
-import re
+import regex as re
 import signal
 import sys
 import tempfile
@@ -49,8 +51,8 @@ from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.grammar import Grammar
 from blib2to3.pgen2.parse import ParseError
 
+from _version import version as __version__
 
-__version__ = "19.3b0"
 DEFAULT_LINE_LENGTH = 88
 DEFAULT_EXCLUDES = (
     r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|_build|buck-out|build|dist)/"
@@ -140,6 +142,8 @@ class Feature(Enum):
     # set for every version of python.
     ASYNC_IDENTIFIERS = 6
     ASYNC_KEYWORDS = 7
+    ASSIGNMENT_EXPRESSIONS = 8
+    POS_ONLY_ARGUMENTS = 9
 
 
 VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
@@ -174,6 +178,8 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
         Feature.TRAILING_COMMA_IN_CALL,
         Feature.TRAILING_COMMA_IN_DEF,
         Feature.ASYNC_KEYWORDS,
+        Feature.ASSIGNMENT_EXPRESSIONS,
+        Feature.POS_ONLY_ARGUMENTS,
     },
 }
 
@@ -427,6 +433,7 @@ def main(
     report = Report(check=check, quiet=quiet, verbose=verbose)
     root = find_project_root(src)
     sources: Set[Path] = set()
+    path_empty(src, quiet, verbose, ctx)
     for s in src:
         p = Path(s)
         if p.is_dir():
@@ -440,7 +447,7 @@ def main(
             err(f"invalid path: {s}")
     if len(sources) == 0:
         if verbose or not quiet:
-            out("No paths given. Nothing to do 😴")
+            out("No Python files are present to be formatted. Nothing to do 😴")
         ctx.exit(0)
 
     if len(sources) == 1:
@@ -462,6 +469,16 @@ def main(
     ctx.exit(report.return_code)
 
 
+def path_empty(src: Tuple[str], quiet: bool, verbose: bool, ctx: click.Context) -> None:
+    """
+    Exit if there is no `src` provided for formatting
+    """
+    if not src:
+        if verbose or not quiet:
+            out("No Path provided. Nothing to do 😴")
+            ctx.exit(0)
+
+
 def reformat_one(
     src: Path, fast: bool, write_back: WriteBack, mode: FileMode, report: "Report"
 ) -> None:
@@ -523,6 +540,7 @@ def reformat_many(
         )
     finally:
         shutdown(loop)
+        executor.shutdown()
 
 
 async def schedule_formatting(
@@ -628,9 +646,8 @@ def format_file_in_place(
         src_name = f"{src}\t{then} +0000"
         dst_name = f"{src}\t{now} +0000"
         diff_contents = diff(src_contents, dst_contents, src_name, dst_name)
-        if lock:
-            lock.acquire()
-        try:
+
+        with lock or nullcontext():
             f = io.TextIOWrapper(
                 sys.stdout.buffer,
                 encoding=encoding,
@@ -639,9 +656,7 @@ def format_file_in_place(
             )
             f.write(diff_contents)
             f.detach()
-        finally:
-            if lock:
-                lock.release()
+
     return True
 
 
@@ -933,6 +948,7 @@ MATH_OPERATORS = {
     token.DOUBLESTAR,
 }
 STARS = {token.STAR, token.DOUBLESTAR}
+VARARGS_SPECIALS = STARS | {token.SLASH}
 VARARGS_PARENTS = {
     syms.arglist,
     syms.argument,  # double star in arglist
@@ -1276,7 +1292,7 @@ class Line:
                     return True
         return False
 
-    def contains_inner_type_comments(self) -> bool:
+    def contains_uncollapsable_type_comments(self) -> bool:
         ignored_ids = set()
         try:
             last_leaf = self.leaves[-1]
@@ -1293,13 +1309,54 @@ class Line:
         except IndexError:
             return False
 
+        # A type comment is uncollapsable if it is attached to a leaf
+        # that isn't at the end of the line (since that could cause it
+        # to get associated to a different argument) or if there are
+        # comments before it (since that could cause it to get hidden
+        # behind a comment.
+        comment_seen = False
         for leaf_id, comments in self.comments.items():
-            if leaf_id in ignored_ids:
-                continue
-
             for comment in comments:
                 if is_type_comment(comment):
-                    return True
+                    if leaf_id not in ignored_ids or comment_seen:
+                        return True
+
+                comment_seen = True
+
+        return False
+
+    def contains_unsplittable_type_ignore(self) -> bool:
+        if not self.leaves:
+            return False
+
+        # If a 'type: ignore' is attached to the end of a line, we
+        # can't split the line, because we can't know which of the
+        # subexpressions the ignore was meant to apply to.
+        #
+        # We only want this to apply to actual physical lines from the
+        # original source, though: we don't want the presence of a
+        # 'type: ignore' at the end of a multiline expression to
+        # justify pushing it all onto one line. Thus we
+        # (unfortunately) need to check the actual source lines and
+        # only report an unsplittable 'type: ignore' if this line was
+        # one line in the original code.
+
+        # Like in the type comment check above, we need to skip a black added
+        # trailing comma or invisible paren, since it will be the original leaf
+        # before it that has the original line number.
+        last_idx = -1
+        last_leaf = self.leaves[-1]
+        if len(self.leaves) > 2 and (
+            last_leaf.type == token.COMMA
+            or (last_leaf.type == token.RPAR and not last_leaf.value)
+        ):
+            last_idx = -2
+
+        if self.leaves[0].lineno == self.leaves[last_idx].lineno:
+            for node in self.leaves[last_idx:]:
+                for comment in self.comments.get(id(node), []):
+                    if is_type_comment(comment, " ignore"):
+                        return True
 
         return False
 
@@ -1475,7 +1532,13 @@ class EmptyLineTracker:
         lines (two on module-level).
         """
         before, after = self._maybe_empty_lines(current_line)
-        before -= self.previous_after
+        before = (
+            # Black should not insert empty lines at the beginning
+            # of the file
+            0
+            if self.previous_line is None
+            else before - self.previous_after
+        )
         self.previous_after = after
         self.previous_line = current_line
         return before, after
@@ -1845,7 +1908,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa: C901
                     # that, too.
                     return prevp.prefix
 
-        elif prevp.type in STARS:
+        elif prevp.type in VARARGS_SPECIALS:
             if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS):
                 return NO
 
@@ -1935,7 +1998,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa: C901
             if not prevp or prevp.type == token.LPAR:
                 return NO
 
-        elif prev.type in {token.EQUAL} | STARS:
+        elif prev.type in {token.EQUAL} | VARARGS_SPECIALS:
             return NO
 
     elif p.type == syms.decorator:
@@ -2302,9 +2365,12 @@ def split_line(
     line_str = str(line).strip("\n")
 
     if (
-        not line.contains_inner_type_comments()
+        not line.contains_uncollapsable_type_comments()
         and not line.should_explode
-        and is_line_short_enough(line, line_length=line_length, line_str=line_str)
+        and (
+            is_line_short_enough(line, line_length=line_length, line_str=line_str)
+            or line.contains_unsplittable_type_ignore()
+        )
     ):
         yield line
         return
@@ -2677,12 +2743,14 @@ def is_import(leaf: Leaf) -> bool:
     )
 
 
-def is_type_comment(leaf: Leaf) -> bool:
+def is_type_comment(leaf: Leaf, suffix: str = "") -> bool:
     """Return True if the given leaf is a special comment.
     Only returns true for type comments for now."""
     t = leaf.type
     v = leaf.value
-    return t in {token.COMMENT, t == STANDALONE_COMMENT} and v.startswith("# type:")
+    return t in {token.COMMENT, t == STANDALONE_COMMENT} and v.startswith(
+        "# type:" + suffix
+    )
 
 
 def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None:
@@ -2864,8 +2932,18 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
             check_lpar = True
 
         if check_lpar:
+            if is_walrus_assignment(child):
+                continue
             if child.type == syms.atom:
-                if maybe_make_parens_invisible_in_atom(child, parent=node):
+                # Determines if the underlying atom should be surrounded with
+                # invisible params - also makes parens invisible recursively
+                # within the atom and removes repeated invisible parens within
+                # the atom
+                should_surround_with_parens = maybe_make_parens_invisible_in_atom(
+                    child, parent=node
+                )
+
+                if should_surround_with_parens:
                     lpar = Leaf(token.LPAR, "")
                     rpar = Leaf(token.RPAR, "")
                     index = child.remove() or 0
@@ -2982,6 +3060,8 @@ def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]:
 
 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.
+    Additionally, remove repeated, adjacent invisible parens from the atom `node`
+    as they are redundant.
 
     Returns whether the node should itself be wrapped in invisible parentheses.
 
@@ -2998,16 +3078,40 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:
     first = node.children[0]
     last = node.children[-1]
     if first.type == token.LPAR and last.type == token.RPAR:
+        middle = node.children[1]
         # make parentheses invisible
         first.value = ""  # type: ignore
         last.value = ""  # type: ignore
-        if len(node.children) > 1:
-            maybe_make_parens_invisible_in_atom(node.children[1], parent=parent)
+        maybe_make_parens_invisible_in_atom(middle, parent=parent)
+
+        if is_atom_with_invisible_parens(middle):
+            # Strip the invisible parens from `middle` by replacing
+            # it with the child in-between the invisible parens
+            middle.replace(middle.children[1])
+
         return False
 
     return True
 
 
+def is_atom_with_invisible_parens(node: LN) -> bool:
+    """Given a `LN`, determines whether it's an atom `node` with invisible
+    parens. Useful in dedupe-ing and normalizing parens.
+    """
+    if isinstance(node, Leaf) or node.type != syms.atom:
+        return False
+
+    first, last = node.children[0], node.children[-1]
+    return (
+        isinstance(first, Leaf)
+        and first.type == token.LPAR
+        and first.value == ""
+        and isinstance(last, Leaf)
+        and last.type == token.RPAR
+        and last.value == ""
+    )
+
+
 def is_empty_tuple(node: LN) -> bool:
     """Return True if `node` holds an empty tuple."""
     return (
@@ -3018,18 +3122,24 @@ def is_empty_tuple(node: LN) -> bool:
     )
 
 
+def unwrap_singleton_parenthesis(node: LN) -> Optional[LN]:
+    """Returns `wrapped` if `node` is of the shape ( wrapped ).
+
+    Parenthesis can be optional. Returns None otherwise"""
+    if len(node.children) != 3:
+        return None
+    lpar, wrapped, rpar = node.children
+    if not (lpar.type == token.LPAR and rpar.type == token.RPAR):
+        return None
+
+    return wrapped
+
+
 def is_one_tuple(node: LN) -> bool:
     """Return True if `node` holds a tuple with one element, with or without parens."""
     if node.type == syms.atom:
-        if len(node.children) != 3:
-            return False
-
-        lpar, gexp, rpar = node.children
-        if not (
-            lpar.type == token.LPAR
-            and gexp.type == syms.testlist_gexp
-            and rpar.type == token.RPAR
-        ):
+        gexp = unwrap_singleton_parenthesis(node)
+        if gexp is None or gexp.type != syms.testlist_gexp:
             return False
 
         return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA
@@ -3041,6 +3151,12 @@ def is_one_tuple(node: LN) -> bool:
     )
 
 
+def is_walrus_assignment(node: LN) -> bool:
+    """Return True iff `node` is of the shape ( test := test )"""
+    inner = unwrap_singleton_parenthesis(node)
+    return inner is not None and inner.type == syms.namedexpr_test
+
+
 def is_yield(node: LN) -> bool:
     """Return True if `node` holds a `yield` or `yield from` expression."""
     if node.type == syms.yield_expr:
@@ -3070,7 +3186,7 @@ def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool:
     extended iterable unpacking (PEP 3132) and additional unpacking
     generalizations (PEP 448).
     """
-    if leaf.type not in STARS or not leaf.parent:
+    if leaf.type not in VARARGS_SPECIALS or not leaf.parent:
         return False
 
     p = leaf.parent
@@ -3185,8 +3301,9 @@ def get_features_used(node: Node) -> Set[Feature]:
 
     Currently looking for:
     - f-strings;
-    - underscores in numeric literals; and
-    - trailing commas after * or ** in function signatures and calls.
+    - underscores in numeric literals;
+    - trailing commas after * or ** in function signatures and calls;
+    - positional only arguments in function signatures and lambdas;
     """
     features: Set[Feature] = set()
     for n in node.pre_order():
@@ -3199,6 +3316,13 @@ def get_features_used(node: Node) -> Set[Feature]:
             if "_" in n.value:  # type: ignore
                 features.add(Feature.NUMERIC_UNDERSCORES)
 
+        elif n.type == token.SLASH:
+            if n.parent and n.parent.type in {syms.typedargslist, syms.arglist}:
+                features.add(Feature.POS_ONLY_ARGUMENTS)
+
+        elif n.type == token.COLONEQUAL:
+            features.add(Feature.ASSIGNMENT_EXPRESSIONS)
+
         elif (
             n.type in {syms.typedargslist, syms.arglist}
             and n.children
@@ -3480,32 +3604,58 @@ class 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
+def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]:
+    filename = "<unknown>"
+    if sys.version_info >= (3, 8):
+        # TODO: support Python 4+ ;)
+        for minor_version in range(sys.version_info[1], 4, -1):
+            try:
+                return ast.parse(src, filename, feature_version=(3, minor_version))
+            except SyntaxError:
+                continue
+    else:
+        for feature_version in (7, 6):
+            try:
+                return ast3.parse(src, filename, feature_version=feature_version)
+            except SyntaxError:
+                continue
 
     return ast27.parse(src)
 
 
+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."""
+    # casts are required until this is released:
+    # https://github.com/python/typeshed/pull/3142
+    if isinstance(node, (ast.Str, ast3.Str, ast27.Str, ast.Bytes, ast3.Bytes)):
+        return cast(ast.AST, ast.Constant(value=node.s))
+    elif isinstance(node, (ast.Num, ast3.Num, ast27.Num)):
+        return cast(ast.AST, ast.Constant(value=node.n))
+    elif isinstance(node, (ast.NameConstant, ast3.NameConstant)):
+        return cast(ast.AST, ast.Constant(value=node.value))
+    return node
+
+
 def assert_equivalent(src: str, dst: str) -> None:
     """Raise AssertionError if `src` and `dst` aren't equivalent."""
 
-    def _v(node: Union[ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]:
+    def _v(node: Union[ast.AST, ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]:
         """Simple visitor generating strings to compare ASTs by content."""
+
+        node = _fixup_ast_constants(node)
+
         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)):
+            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
 
-            # 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:
@@ -3519,15 +3669,15 @@ def assert_equivalent(src: str, dst: str) -> None:
                     # parentheses and they change the AST.
                     if (
                         field == "targets"
-                        and isinstance(node, (ast3.Delete, ast27.Delete))
-                        and isinstance(item, (ast3.Tuple, ast27.Tuple))
+                        and isinstance(node, (ast.Delete, ast3.Delete, ast27.Delete))
+                        and isinstance(item, (ast.Tuple, ast3.Tuple, ast27.Tuple))
                     ):
                         for item in item.elts:
                             yield from _v(item, depth + 2)
-                    elif isinstance(item, (ast3.AST, ast27.AST)):
+                    elif isinstance(item, (ast.AST, ast3.AST, ast27.AST)):
                         yield from _v(item, depth + 2)
 
-            elif isinstance(value, (ast3.AST, ast27.AST)):
+            elif isinstance(value, (ast.AST, ast3.AST, ast27.AST)):
                 yield from _v(value, depth + 2)
 
             else:
@@ -3549,7 +3699,7 @@ def assert_equivalent(src: str, dst: str) -> None:
         log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
         raise AssertionError(
             f"INTERNAL ERROR: Black produced invalid code: {exc}. "
-            f"Please report a bug on https://github.com/python/black/issues.  "
+            f"Please report a bug on https://github.com/psf/black/issues.  "
             f"This invalid output might be helpful: {log}"
         ) from None
 
@@ -3560,7 +3710,7 @@ def assert_equivalent(src: str, dst: str) -> None:
         raise AssertionError(
             f"INTERNAL ERROR: Black produced code that is not equivalent to "
             f"the source.  "
-            f"Please report a bug on https://github.com/python/black/issues.  "
+            f"Please report a bug on https://github.com/psf/black/issues.  "
             f"This diff might be helpful: {log}"
         ) from None
 
@@ -3576,7 +3726,7 @@ def assert_stable(src: str, dst: str, mode: FileMode) -> None:
         raise AssertionError(
             f"INTERNAL ERROR: Black produced different code on the second pass "
             f"of the formatter.  "
-            f"Please report a bug on https://github.com/python/black/issues.  "
+            f"Please report a bug on https://github.com/psf/black/issues.  "
             f"This diff might be helpful: {log}"
         ) from None
 
@@ -3593,6 +3743,13 @@ def dump_to_file(*output: str) -> str:
     return f.name
 
 
+@contextmanager
+def nullcontext() -> Iterator[None]:
+    """Return context manager that does nothing.
+    Similar to `nullcontext` from python 3.7"""
+    yield
+
+
 def diff(a: str, b: str, a_name: str, b_name: str) -> str:
     """Return a unified diff string between strings `a` and `b`."""
     import difflib
@@ -3653,7 +3810,8 @@ def re_compile_maybe_verbose(regex: str) -> Pattern[str]:
     """
     if "\n" in regex:
         regex = "(?x)" + regex
-    return re.compile(regex)
+    compiled: Pattern[str] = re.compile(regex)
+    return compiled
 
 
 def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]: