import os
from pathlib import Path
import tokenize
+import sys
from typing import (
Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, TypeVar, Union
)
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
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):
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)
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:
# 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
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 commas > 1:
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 = ' '
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:
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:
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:
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)
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:
# 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
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.
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.append(leaf)
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)
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()
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]
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])
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])
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'
projects / etc / vim.git / blobdiff