import os
from pathlib import Path
import tokenize
+import sys
from typing import (
Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, TypeVar, Union
)
help='How many character per line to allow.',
show_default=True,
)
+@click.option(
+ '--check',
+ is_flag=True,
+ help=(
+ "Don't write back the files, just return the status. Return code 0 "
+ "means nothing changed. Return code 1 means some files were "
+ "reformatted. Return code 123 means there was an internal error."
+ ),
+)
@click.option(
'--fast/--safe',
is_flag=True,
type=click.Path(exists=True, file_okay=True, dir_okay=True, readable=True),
)
@click.pass_context
-def main(ctx: click.Context, line_length: int, fast: bool, src: List[str]) -> None:
+def main(
+ ctx: click.Context, line_length: int, check: bool, fast: bool, src: List[str]
+) -> None:
"""The uncompromising code formatter."""
sources: List[Path] = []
for s in src:
p = sources[0]
report = Report()
try:
- changed = format_file_in_place(p, line_length=line_length, fast=fast)
+ changed = format_file_in_place(
+ p, line_length=line_length, fast=fast, write_back=not check
+ )
report.done(p, changed)
except Exception as exc:
report.failed(p, str(exc))
return_code = 1
try:
return_code = loop.run_until_complete(
- schedule_formatting(sources, line_length, fast, loop, executor)
+ schedule_formatting(
+ sources, line_length, not check, fast, loop, executor
+ )
)
finally:
loop.close()
async def schedule_formatting(
sources: List[Path],
line_length: int,
+ write_back: bool,
fast: bool,
loop: BaseEventLoop,
executor: Executor,
) -> int:
tasks = {
src: loop.run_in_executor(
- executor, format_file_in_place, src, line_length, fast
+ executor, format_file_in_place, src, line_length, fast, write_back
)
for src in sources
}
return report.return_code
-def format_file_in_place(src: Path, line_length: int, fast: bool) -> bool:
+def format_file_in_place(
+ src: Path, line_length: int, fast: bool, write_back: bool = False
+) -> bool:
"""Format the file and rewrite if changed. Return True if changed."""
try:
contents, encoding = format_file(src, line_length=line_length, fast=fast)
except NothingChanged:
return False
- with open(src, "w", encoding=encoding) as f:
- f.write(contents)
+ if write_back:
+ with open(src, "w", encoding=encoding) as f:
+ f.write(contents)
return True
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)
comments: Dict[LeafID, Leaf] = attrib(default=Factory(dict))
bracket_tracker: BracketTracker = attrib(default=Factory(BracketTracker))
inside_brackets: bool = attrib(default=False)
+ has_for: bool = attrib(default=False)
+ _for_loop_variable: bool = attrib(default=False, init=False)
def append(self, leaf: Leaf, preformatted: bool = False) -> None:
has_value = leaf.value.strip()
# 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
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.
if prevp.parent and prevp.parent.type == syms.subscript:
return NO
+ elif prevp.parent and prevp.parent.type == syms.factor:
+ return NO
+
elif prev.type in OPENING_BRACKETS:
return NO
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.
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)
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 # type: ignore
+ 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:
+ assert isinstance(n, Leaf)
+ if n.value[:2] 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'
@property
def return_code(self) -> int:
"""Which return code should the app use considering the current state."""
- return 1 if self.failure_count else 0
+ # According to http://tldp.org/LDP/abs/html/exitcodes.html starting with
+ # 126 we have special returncodes reserved by the shell.
+ if self.failure_count:
+ return 123
+
+ elif self.change_count:
+ return 1
+
+ return 0
def __str__(self) -> str:
"""A color report of the current state.
projects / etc / vim.git / blobdiff