X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/4c480bcac288883040e6fda3c75825a0da54eb66..28d1442d573abc9aeb567242f584f38eaec3f922:/black.py diff --git a/black.py b/black.py index e2c427c..8d4b095 100644 --- a/black.py +++ b/black.py @@ -7,11 +7,12 @@ import keyword import os from pathlib import Path import tokenize +import sys from typing import ( Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, TypeVar, Union ) -from attr import attrib, dataclass, Factory +from attr import dataclass, Factory import click # lib2to3 fork @@ -20,7 +21,7 @@ from blib2to3 import pygram, pytree from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.3a1" +__version__ = "18.3a2" DEFAULT_LINE_LENGTH = 88 # types syms = pygram.python_symbols @@ -192,6 +193,7 @@ def format_str(src_contents: str, line_length: int) -> FileContent: comments: List[Line] = [] lines = LineGenerator() elt = EmptyLineTracker() + py36 = is_python36(src_node) empty_line = Line() after = 0 for current_line in lines.visit(src_node): @@ -204,7 +206,7 @@ def format_str(src_contents: str, line_length: int) -> FileContent: for comment in comments: dst_contents += str(comment) comments = [] - for line in split_line(current_line, line_length=line_length): + for line in split_line(current_line, line_length=line_length, py36=py36): dst_contents += str(line) else: comments.append(current_line) @@ -263,7 +265,7 @@ class Visitor(Generic[T]): @dataclass class DebugVisitor(Visitor[T]): - tree_depth: int = attrib(default=0) + tree_depth: int = 0 def visit_default(self, node: LN) -> Iterator[T]: indent = ' ' * (2 * self.tree_depth) @@ -333,10 +335,10 @@ MATH_PRIORITY = 1 @dataclass class BracketTracker: - depth: int = attrib(default=0) - bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = attrib(default=Factory(dict)) - delimiters: Dict[LeafID, Priority] = attrib(default=Factory(dict)) - previous: Optional[Leaf] = attrib(default=None) + depth: int = 0 + bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = Factory(dict) + delimiters: Dict[LeafID, Priority] = Factory(dict) + previous: Optional[Leaf] = None def mark(self, leaf: Leaf) -> None: if leaf.type == token.COMMENT: @@ -345,8 +347,8 @@ class BracketTracker: if leaf.type in CLOSING_BRACKETS: self.depth -= 1 opening_bracket = self.bracket_match.pop((self.depth, leaf.type)) - leaf.opening_bracket = opening_bracket # type: ignore - leaf.bracket_depth = self.depth # type: ignore + leaf.opening_bracket = opening_bracket + leaf.bracket_depth = self.depth if self.depth == 0: delim = is_delimiter(leaf) if delim: @@ -387,11 +389,13 @@ class BracketTracker: @dataclass class Line: - depth: int = attrib(default=0) - leaves: List[Leaf] = attrib(default=Factory(list)) - comments: Dict[LeafID, Leaf] = attrib(default=Factory(dict)) - bracket_tracker: BracketTracker = attrib(default=Factory(BracketTracker)) - inside_brackets: bool = attrib(default=False) + depth: int = 0 + leaves: List[Leaf] = Factory(list) + comments: Dict[LeafID, Leaf] = Factory(dict) + bracket_tracker: BracketTracker = Factory(BracketTracker) + inside_brackets: bool = False + has_for: bool = False + _for_loop_variable: bool = False def append(self, leaf: Leaf, preformatted: bool = False) -> None: has_value = leaf.value.strip() @@ -403,8 +407,10 @@ class Line: # imports, for which we only preserve newlines. leaf.prefix += whitespace(leaf) if self.inside_brackets or not preformatted: + self.maybe_decrement_after_for_loop_variable(leaf) self.bracket_tracker.mark(leaf) self.maybe_remove_trailing_comma(leaf) + self.maybe_increment_for_loop_variable(leaf) if self.maybe_adapt_standalone_comment(leaf): return @@ -485,9 +491,9 @@ class Line: # For parens let's check if it's safe to remove the comma. If the # trailing one is the only one, we might mistakenly change a tuple # into a different type by removing the comma. - depth = closing.bracket_depth + 1 # type: ignore + depth = closing.bracket_depth + 1 commas = 0 - opening = closing.opening_bracket # type: ignore + opening = closing.opening_bracket for _opening_index, leaf in enumerate(self.leaves): if leaf is opening: break @@ -499,15 +505,42 @@ class Line: if leaf is closing: break - bracket_depth = leaf.bracket_depth # type: ignore + bracket_depth = leaf.bracket_depth if bracket_depth == depth and leaf.type == token.COMMA: commas += 1 + if leaf.parent and leaf.parent.type == syms.arglist: + commas += 1 + break + if commas > 1: self.leaves.pop() return True return False + def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: + """In a for loop, or comprehension, the variables are often unpacks. + + To avoid splitting on the comma in this situation, we will increase + the depth of tokens between `for` and `in`. + """ + if leaf.type == token.NAME and leaf.value == 'for': + self.has_for = True + self.bracket_tracker.depth += 1 + self._for_loop_variable = True + return True + + return False + + def maybe_decrement_after_for_loop_variable(self, leaf: Leaf) -> bool: + # See `maybe_increment_for_loop_variable` above for explanation. + if self._for_loop_variable and leaf.type == token.NAME and leaf.value == 'in': + self.bracket_tracker.depth -= 1 + self._for_loop_variable = False + return True + + return False + def maybe_adapt_standalone_comment(self, comment: Leaf) -> bool: """Hack a standalone comment to act as a trailing comment for line splitting. @@ -578,9 +611,9 @@ class EmptyLineTracker: Note: this tracker works on lines that haven't been split yet. """ - previous_line: Optional[Line] = attrib(default=None) - previous_after: int = attrib(default=0) - previous_defs: List[int] = attrib(default=Factory(list)) + previous_line: Optional[Line] = None + previous_after: int = 0 + previous_defs: List[int] = Factory(list) def maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]: """Returns the number of extra empty lines before and after the `current_line`. @@ -646,8 +679,8 @@ class LineGenerator(Visitor[Line]): Note: destroys the tree it's visiting by mutating prefixes of its leaves in ways that will no longer stringify to valid Python code on the tree. """ - current_line: Line = attrib(default=Factory(Line)) - standalone_comments: List[Leaf] = attrib(default=Factory(list)) + current_line: Line = Factory(Line) + standalone_comments: List[Leaf] = Factory(list) def line(self, indent: int = 0) -> Iterator[Line]: """Generate a line. @@ -775,9 +808,10 @@ BRACKET = {token.LPAR: token.RPAR, token.LSQB: token.RSQB, token.LBRACE: token.R OPENING_BRACKETS = set(BRACKET.keys()) CLOSING_BRACKETS = set(BRACKET.values()) BRACKETS = OPENING_BRACKETS | CLOSING_BRACKETS +ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, token.COLON, STANDALONE_COMMENT} -def whitespace(leaf: Leaf) -> str: +def whitespace(leaf: Leaf) -> str: # noqa C901 """Return whitespace prefix if needed for the given `leaf`.""" NO = '' SPACE = ' ' @@ -785,24 +819,12 @@ def whitespace(leaf: Leaf) -> str: t = leaf.type p = leaf.parent v = leaf.value - if t == token.COLON: - return NO - - if t == token.COMMA: - return NO - - if t == token.RPAR: + if t in ALWAYS_NO_SPACE: return NO if t == token.COMMENT: return DOUBLESPACE - if t == STANDALONE_COMMENT: - return NO - - if t in CLOSING_BRACKETS: - return NO - assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}" prev = leaf.prev_sibling if not prev: @@ -834,7 +856,7 @@ def whitespace(leaf: Leaf) -> str: if prevp.parent and prevp.parent.type == syms.subscript: return NO - elif prevp.parent and prevp.parent.type == syms.factor: + elif prevp.parent and prevp.parent.type in {syms.factor, syms.star_expr}: return NO elif prev.type in OPENING_BRACKETS: @@ -931,7 +953,14 @@ def whitespace(leaf: Leaf) -> str: elif p.type == syms.subscript: # indexing - if not prev or prev.type == token.COLON: + if not prev: + assert p.parent is not None, "subscripts are always parented" + if p.parent.type == syms.subscriptlist: + return SPACE + + return NO + + elif prev.type == token.COLON: return NO elif p.type == syms.atom: @@ -956,7 +985,7 @@ def whitespace(leaf: Leaf) -> str: if prev.type == token.DOUBLESTAR: return NO - elif p.type == syms.factor or p.type == syms.star_expr: + elif p.type in {syms.factor, syms.star_expr}: # unary ops if not prev: prevp = preceding_leaf(p) @@ -1081,13 +1110,18 @@ def generate_comments(leaf: Leaf) -> Iterator[Leaf]: yield Leaf(STANDALONE_COMMENT, line) -def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Line]: +def split_line( + line: Line, line_length: int, inner: bool = False, py36: bool = False +) -> Iterator[Line]: """Splits a `line` into potentially many lines. They should fit in the allotted `line_length` but might not be able to. `inner` signifies that there were a pair of brackets somewhere around the current `line`, possibly transitively. This means we can fallback to splitting by delimiters if the LHS/RHS don't yield any results. + + If `py36` is True, splitting may generate syntax that is only compatible + with Python 3.6 and later. """ line_str = str(line).strip('\n') if len(line_str) <= line_length and '\n' not in line_str: @@ -1110,11 +1144,13 @@ def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Li # split altogether. result: List[Line] = [] try: - for l in split_func(line): + for l in split_func(line, py36=py36): if str(l).strip('\n') == line_str: raise CannotSplit("Split function returned an unchanged result") - result.extend(split_line(l, line_length=line_length, inner=True)) + result.extend( + split_line(l, line_length=line_length, inner=True, py36=py36) + ) except CannotSplit as cs: continue @@ -1126,7 +1162,7 @@ def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Li yield line -def left_hand_split(line: Line) -> Iterator[Line]: +def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split line into many lines, starting with the first matching bracket pair. Note: this usually looks weird, only use this for function definitions. @@ -1144,9 +1180,9 @@ def left_hand_split(line: Line) -> Iterator[Line]: if ( current_leaves is body_leaves and leaf.type in CLOSING_BRACKETS and - leaf.opening_bracket is matching_bracket # type: ignore + leaf.opening_bracket is matching_bracket ): - current_leaves = tail_leaves + current_leaves = tail_leaves if body_leaves else head_leaves current_leaves.append(leaf) if current_leaves is head_leaves: if leaf.type in OPENING_BRACKETS: @@ -1164,24 +1200,13 @@ def left_hand_split(line: Line) -> Iterator[Line]: comment_after = line.comments.get(id(leaf)) if comment_after: result.append(comment_after, preformatted=True) - # Check if the split succeeded. - tail_len = len(str(tail)) - if not body: - if tail_len == 0: - raise CannotSplit("Splitting brackets produced the same line") - - elif tail_len < 3: - raise CannotSplit( - f"Splitting brackets on an empty body to save " - f"{tail_len} characters is not worth it" - ) - + split_succeeded_or_raise(head, body, tail) for result in (head, body, tail): if result: yield result -def right_hand_split(line: Line) -> Iterator[Line]: +def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split line into many lines, starting with the last matching bracket pair.""" head = Line(depth=line.depth) body = Line(depth=line.depth + 1, inside_brackets=True) @@ -1194,11 +1219,11 @@ def right_hand_split(line: Line) -> Iterator[Line]: for leaf in reversed(line.leaves): if current_leaves is body_leaves: if leaf is opening_bracket: - current_leaves = head_leaves + current_leaves = head_leaves if body_leaves else tail_leaves current_leaves.append(leaf) if current_leaves is tail_leaves: if leaf.type in CLOSING_BRACKETS: - opening_bracket = leaf.opening_bracket # type: ignore + opening_bracket = leaf.opening_bracket current_leaves = body_leaves tail_leaves.reverse() body_leaves.reverse() @@ -1215,8 +1240,14 @@ def right_hand_split(line: Line) -> Iterator[Line]: comment_after = line.comments.get(id(leaf)) if comment_after: result.append(comment_after, preformatted=True) - # Check if the split succeeded. - tail_len = len(str(tail).strip('\n')) + split_succeeded_or_raise(head, body, tail) + for result in (head, body, tail): + if result: + yield result + + +def split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None: + tail_len = len(str(tail).strip()) if not body: if tail_len == 0: raise CannotSplit("Splitting brackets produced the same line") @@ -1227,15 +1258,13 @@ def right_hand_split(line: Line) -> Iterator[Line]: f"{tail_len} characters is not worth it" ) - for result in (head, body, tail): - if result: - yield result - -def delimiter_split(line: Line) -> Iterator[Line]: +def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split according to delimiters of the highest priority. This kind of split doesn't increase indentation. + If `py36` is True, the split will add trailing commas also in function + signatures that contain * and **. """ try: last_leaf = line.leaves[-1] @@ -1249,11 +1278,20 @@ def delimiter_split(line: Line) -> Iterator[Line]: raise CannotSplit("No delimiters found") current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + lowest_depth = sys.maxsize + trailing_comma_safe = True for leaf in line.leaves: current_line.append(leaf, preformatted=True) comment_after = line.comments.get(id(leaf)) if comment_after: current_line.append(comment_after, preformatted=True) + lowest_depth = min(lowest_depth, leaf.bracket_depth) + if ( + leaf.bracket_depth == lowest_depth and + leaf.type == token.STAR or + leaf.type == token.DOUBLESTAR + ): + trailing_comma_safe = trailing_comma_safe and py36 leaf_priority = delimiters.get(id(leaf)) if leaf_priority == delimiter_priority: normalize_prefix(current_line.leaves[0]) @@ -1263,7 +1301,8 @@ def delimiter_split(line: Line) -> Iterator[Line]: if current_line: if ( delimiter_priority == COMMA_PRIORITY and - current_line.leaves[-1].type != token.COMMA + current_line.leaves[-1].type != token.COMMA and + trailing_comma_safe ): current_line.append(Leaf(token.COMMA, ',')) normalize_prefix(current_line.leaves[0]) @@ -1298,6 +1337,31 @@ def normalize_prefix(leaf: Leaf) -> None: leaf.prefix = '' +def is_python36(node: Node) -> bool: + """Returns True if the current file is using Python 3.6+ features. + + Currently looking for: + - f-strings; and + - trailing commas after * or ** in function signatures. + """ + for n in node.pre_order(): + if n.type == token.STRING: + value_head = n.value[:2] # type: ignore + if value_head in {'f"', 'F"', "f'", "F'", 'rf', 'fr', 'RF', 'FR'}: + return True + + elif ( + n.type == syms.typedargslist and + n.children and + n.children[-1].type == token.COMMA + ): + for ch in n.children: + if ch.type == token.STAR or ch.type == token.DOUBLESTAR: + return True + + return False + + PYTHON_EXTENSIONS = {'.py'} BLACKLISTED_DIRECTORIES = { 'build', 'buck-out', 'dist', '_build', '.git', '.hg', '.mypy_cache', '.tox', '.venv' @@ -1319,9 +1383,9 @@ def gen_python_files_in_dir(path: Path) -> Iterator[Path]: @dataclass class Report: """Provides a reformatting counter.""" - change_count: int = attrib(default=0) - same_count: int = attrib(default=0) - failure_count: int = attrib(default=0) + change_count: int = 0 + same_count: int = 0 + failure_count: int = 0 def done(self, src: Path, changed: bool) -> None: """Increment the counter for successful reformatting. Write out a message.""" @@ -1420,7 +1484,7 @@ def assert_equivalent(src: str, dst: str) -> None: raise AssertionError( f"INTERNAL ERROR: Black produced invalid code: {exc}. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This invalid output might be helpful: {log}", + f"This invalid output might be helpful: {log}" ) from None src_ast_str = '\n'.join(_v(src_ast)) @@ -1431,7 +1495,7 @@ def assert_equivalent(src: str, dst: str) -> None: f"INTERNAL ERROR: Black produced code that is not equivalent to " f"the source. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}", + f"This diff might be helpful: {log}" ) from None @@ -1450,7 +1514,7 @@ def assert_stable(src: str, dst: str, line_length: int) -> None: f"INTERNAL ERROR: Black produced different code on the second pass " f"of the formatter. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}", + f"This diff might be helpful: {log}" ) from None