]> 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:

18.5b0
[etc/vim.git] / black.py
index 06bff0815a23022a7d94da00eb034d2bff92c82b..57cd9a6b9da63eedabccd18d9f624108b2b3e3ee 100644 (file)
--- a/black.py
+++ b/black.py
@@ -30,6 +30,7 @@ from typing import (
     Type,
     TypeVar,
     Union,
+    cast,
 )
 
 from appdirs import user_cache_dir
@@ -43,7 +44,7 @@ from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.parse import ParseError
 
 
-__version__ = "18.4a6"
+__version__ = "18.5b0"
 DEFAULT_LINE_LENGTH = 88
 
 # types
@@ -626,21 +627,22 @@ LOGIC_PRIORITY = 14
 STRING_PRIORITY = 12
 COMPARATOR_PRIORITY = 10
 MATH_PRIORITIES = {
-    token.VBAR: 8,
-    token.CIRCUMFLEX: 7,
-    token.AMPER: 6,
-    token.LEFTSHIFT: 5,
-    token.RIGHTSHIFT: 5,
-    token.PLUS: 4,
-    token.MINUS: 4,
-    token.STAR: 3,
-    token.SLASH: 3,
-    token.DOUBLESLASH: 3,
-    token.PERCENT: 3,
-    token.AT: 3,
-    token.TILDE: 2,
-    token.DOUBLESTAR: 1,
+    token.VBAR: 9,
+    token.CIRCUMFLEX: 8,
+    token.AMPER: 7,
+    token.LEFTSHIFT: 6,
+    token.RIGHTSHIFT: 6,
+    token.PLUS: 5,
+    token.MINUS: 5,
+    token.STAR: 4,
+    token.SLASH: 4,
+    token.DOUBLESLASH: 4,
+    token.PERCENT: 4,
+    token.AT: 4,
+    token.TILDE: 3,
+    token.DOUBLESTAR: 2,
 }
+DOT_PRIORITY = 1
 
 
 @dataclass
@@ -775,6 +777,7 @@ class Line:
     comments: List[Tuple[Index, Leaf]] = Factory(list)
     bracket_tracker: BracketTracker = Factory(BracketTracker)
     inside_brackets: bool = False
+    should_explode: bool = False
 
     def append(self, leaf: Leaf, preformatted: bool = False) -> None:
         """Add a new `leaf` to the end of the line.
@@ -1314,7 +1317,7 @@ class LineGenerator(Visitor[Line]):
         The relevant Python language `keywords` for a given statement will be
         NAME leaves within it. This methods puts those on a separate line.
 
-        `parens` holds a set of string leaf values immeditely after which
+        `parens` holds a set of string leaf values immediately after which
         invisible parens should be put.
         """
         normalize_invisible_parens(node, parens_after=parens)
@@ -1472,7 +1475,9 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa C901
 
     assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}"
     if t == token.COLON and p.type not in {
-        syms.subscript, syms.subscriptlist, syms.sliceop
+        syms.subscript,
+        syms.subscriptlist,
+        syms.sliceop,
     }:
         return NO
 
@@ -1494,7 +1499,10 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa C901
         if prevp.type == token.EQUAL:
             if prevp.parent:
                 if prevp.parent.type in {
-                    syms.arglist, syms.argument, syms.parameters, syms.varargslist
+                    syms.arglist,
+                    syms.argument,
+                    syms.parameters,
+                    syms.varargslist,
                 }:
                     return NO
 
@@ -1648,7 +1656,8 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa C901
             prevp_parent = prevp.parent
             assert prevp_parent is not None
             if prevp.type == token.COLON and prevp_parent.type in {
-                syms.subscript, syms.sliceop
+                syms.subscript,
+                syms.sliceop,
             }:
                 return NO
 
@@ -1729,6 +1738,14 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
         # Don't treat them as a delimiter.
         return 0
 
+    if (
+        leaf.type == token.DOT
+        and leaf.parent
+        and leaf.parent.type not in {syms.import_from, syms.dotted_name}
+        and (previous is None or previous.type in CLOSING_BRACKETS)
+    ):
+        return DOT_PRIORITY
+
     if (
         leaf.type in MATH_OPERATORS
         and leaf.parent
@@ -1893,25 +1910,27 @@ def split_line(
         return
 
     line_str = str(line).strip("\n")
-    if is_line_short_enough(line, line_length=line_length, line_str=line_str):
+    if not line.should_explode and is_line_short_enough(
+        line, line_length=line_length, line_str=line_str
+    ):
         yield line
         return
 
     split_funcs: List[SplitFunc]
     if line.is_def:
         split_funcs = [left_hand_split]
-    elif line.is_import:
-        split_funcs = [explode_split]
     else:
 
         def rhs(line: Line, py36: bool = False) -> Iterator[Line]:
             for omit in generate_trailers_to_omit(line, line_length):
-                lines = list(right_hand_split(line, py36, omit=omit))
+                lines = list(right_hand_split(line, line_length, py36, omit=omit))
                 if is_line_short_enough(lines[0], line_length=line_length):
                     yield from lines
                     return
 
             # All splits failed, best effort split with no omits.
+            # This mostly happens to multiline strings that are by definition
+            # reported as not fitting a single line.
             yield from right_hand_split(line, py36)
 
         if line.inside_brackets:
@@ -1985,7 +2004,7 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]:
 
 
 def right_hand_split(
-    line: Line, py36: bool = False, omit: Collection[LeafID] = ()
+    line: Line, line_length: int, py36: bool = False, omit: Collection[LeafID] = ()
 ) -> Iterator[Line]:
     """Split line into many lines, starting with the last matching bracket pair.
 
@@ -2047,23 +2066,16 @@ def right_hand_split(
         and not line.is_import
     ):
         omit = {id(closing_bracket), *omit}
-        delimiter_count = body.bracket_tracker.delimiter_count_with_priority()
-        if (
-            delimiter_count == 0
-            or delimiter_count == 1
-            and (
-                body.leaves[0].type in OPENING_BRACKETS
-                or body.leaves[-1].type in CLOSING_BRACKETS
-            )
-        ):
+        if can_omit_invisible_parens(body, line_length):
             try:
-                yield from right_hand_split(line, py36=py36, omit=omit)
+                yield from right_hand_split(line, line_length, py36=py36, omit=omit)
                 return
             except CannotSplit:
                 pass
 
     ensure_visible(opening_bracket)
     ensure_visible(closing_bracket)
+    body.should_explode = should_explode(body, opening_bracket)
     for result in (head, body, tail):
         if result:
             yield result
@@ -2128,6 +2140,10 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]:
     except ValueError:
         raise CannotSplit("No delimiters found")
 
+    if delimiter_priority == DOT_PRIORITY:
+        if bt.delimiter_count_with_priority(delimiter_priority) == 1:
+            raise CannotSplit("Splitting a single attribute from its owner looks wrong")
+
     current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
     lowest_depth = sys.maxsize
     trailing_comma_safe = True
@@ -2199,26 +2215,6 @@ def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]:
         yield current_line
 
 
-def explode_split(
-    line: Line, py36: bool = False, omit: Collection[LeafID] = ()
-) -> Iterator[Line]:
-    """Split by rightmost bracket and immediately split contents by a delimiter."""
-    new_lines = list(right_hand_split(line, py36, omit))
-    if len(new_lines) != 3:
-        yield from new_lines
-        return
-
-    yield new_lines[0]
-
-    try:
-        yield from delimiter_split(new_lines[1], py36)
-
-    except CannotSplit:
-        yield new_lines[1]
-
-    yield new_lines[2]
-
-
 def is_import(leaf: Leaf) -> bool:
     """Return True if the given leaf starts an import statement."""
     p = leaf.parent
@@ -2348,7 +2344,7 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
                 rpar = Leaf(token.RPAR, ")")
                 index = child.remove() or 0
                 node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
-            else:
+            elif not (isinstance(child, Leaf) and is_multiline_string(child)):
                 # wrap child in invisible parentheses
                 lpar = Leaf(token.LPAR, "")
                 rpar = Leaf(token.RPAR, "")
@@ -2459,6 +2455,12 @@ def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool:
     return p.type in within
 
 
+def is_multiline_string(leaf: Leaf) -> bool:
+    """Return True if `leaf` is a multiline string that actually spans many lines."""
+    value = leaf.value.lstrip("furbFURB")
+    return value[:3] in {'"""', "'''"} and "\n" in value
+
+
 def is_stub_suite(node: Node) -> bool:
     """Return True if `node` is a suite with a stub body."""
     if (
@@ -2528,6 +2530,25 @@ def ensure_visible(leaf: Leaf) -> None:
         leaf.value = ")"
 
 
+def should_explode(line: Line, opening_bracket: Leaf) -> bool:
+    """Should `line` immediately be split with `delimiter_split()` after RHS?"""
+    if not (
+        opening_bracket.parent
+        and opening_bracket.parent.type in {syms.atom, syms.import_from}
+        and opening_bracket.value in "[{("
+    ):
+        return False
+
+    try:
+        last_leaf = line.leaves[-1]
+        exclude = {id(last_leaf)} if last_leaf.type == token.COMMA else set()
+        max_priority = line.bracket_tracker.max_delimiter_priority(exclude=exclude)
+    except (IndexError, ValueError):
+        return False
+
+    return max_priority == COMMA_PRIORITY
+
+
 def is_python36(node: Node) -> bool:
     """Return True if the current file is using Python 3.6+ features.
 
@@ -2576,22 +2597,11 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
     closing_bracket = None
     optional_brackets: Set[LeafID] = set()
     inner_brackets: Set[LeafID] = set()
-    for index, leaf in enumerate_reversed(line.leaves):
-        length += len(leaf.prefix) + len(leaf.value)
+    for index, leaf, leaf_length in enumerate_with_length(line, reversed=True):
+        length += leaf_length
         if length > line_length:
             break
 
-        comment: Optional[Leaf]
-        for comment in line.comments_after(leaf, index):
-            if "\n" in comment.prefix:
-                break  # Oops, standalone comment!
-
-            length += len(comment.value)
-        else:
-            comment = None
-        if comment is not None:
-            break  # There was a standalone comment, we can't continue.
-
         optional_brackets.discard(id(leaf))
         if opening_bracket:
             if leaf is opening_bracket:
@@ -2656,7 +2666,15 @@ def get_future_imports(node: Node) -> Set[str]:
 
 PYTHON_EXTENSIONS = {".py", ".pyi"}
 BLACKLISTED_DIRECTORIES = {
-    "build", "buck-out", "dist", "_build", ".git", ".hg", ".mypy_cache", ".tox", ".venv"
+    "build",
+    "buck-out",
+    "dist",
+    "_build",
+    ".git",
+    ".hg",
+    ".mypy_cache",
+    ".tox",
+    ".venv",
 }
 
 
@@ -2904,6 +2922,32 @@ def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]:
         index -= 1
 
 
+def enumerate_with_length(
+    line: Line, reversed: bool = False
+) -> Iterator[Tuple[Index, Leaf, int]]:
+    """Return an enumeration of leaves with their length.
+
+    Stops prematurely on multiline strings and standalone comments.
+    """
+    op = cast(
+        Callable[[Sequence[Leaf]], Iterator[Tuple[Index, Leaf]]],
+        enumerate_reversed if reversed else enumerate,
+    )
+    for index, leaf in op(line.leaves):
+        length = len(leaf.prefix) + len(leaf.value)
+        if "\n" in leaf.value:
+            return  # Multiline strings, we can't continue.
+
+        comment: Optional[Leaf]
+        for comment in line.comments_after(leaf, index):
+            if "\n" in comment.prefix:
+                return  # Oops, standalone comment!
+
+            length += len(comment.value)
+
+        yield index, leaf, length
+
+
 def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") -> bool:
     """Return True if `line` is no longer than `line_length`.
 
@@ -2918,6 +2962,95 @@ def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") ->
     )
 
 
+def can_omit_invisible_parens(line: Line, line_length: int) -> bool:
+    """Does `line` have a shape safe to reformat without optional parens around it?
+
+    Returns True for only a subset of potentially nice looking formattings but
+    the point is to not return false positives that end up producing lines that
+    are too long.
+    """
+    bt = line.bracket_tracker
+    if not bt.delimiters:
+        # Without delimiters the optional parentheses are useless.
+        return True
+
+    max_priority = bt.max_delimiter_priority()
+    if bt.delimiter_count_with_priority(max_priority) > 1:
+        # With more than one delimiter of a kind the optional parentheses read better.
+        return False
+
+    if max_priority == DOT_PRIORITY:
+        # A single stranded method call doesn't require optional parentheses.
+        return True
+
+    assert len(line.leaves) >= 2, "Stranded delimiter"
+
+    first = line.leaves[0]
+    second = line.leaves[1]
+    penultimate = line.leaves[-2]
+    last = line.leaves[-1]
+
+    # With a single delimiter, omit if the expression starts or ends with
+    # a bracket.
+    if first.type in OPENING_BRACKETS and second.type not in CLOSING_BRACKETS:
+        remainder = False
+        length = 4 * line.depth
+        for _index, leaf, leaf_length in enumerate_with_length(line):
+            if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first:
+                remainder = True
+            if remainder:
+                length += leaf_length
+                if length > line_length:
+                    break
+
+                if leaf.type in OPENING_BRACKETS:
+                    # There are brackets we can further split on.
+                    remainder = False
+
+        else:
+            # checked the entire string and line length wasn't exceeded
+            if len(line.leaves) == _index + 1:
+                return True
+
+        # Note: we are not returning False here because a line might have *both*
+        # a leading opening bracket and a trailing closing bracket.  If the
+        # opening bracket doesn't match our rule, maybe the closing will.
+
+    if (
+        last.type == token.RPAR
+        or last.type == token.RBRACE
+        or (
+            # don't use indexing for omitting optional parentheses;
+            # it looks weird
+            last.type == token.RSQB
+            and last.parent
+            and last.parent.type != syms.trailer
+        )
+    ):
+        if penultimate.type in OPENING_BRACKETS:
+            # Empty brackets don't help.
+            return False
+
+        if is_multiline_string(first):
+            # Additional wrapping of a multiline string in this situation is
+            # unnecessary.
+            return True
+
+        length = 4 * line.depth
+        seen_other_brackets = False
+        for _index, leaf, leaf_length in enumerate_with_length(line):
+            length += leaf_length
+            if leaf is last.opening_bracket:
+                if seen_other_brackets or length <= line_length:
+                    return True
+
+            elif leaf.type in OPENING_BRACKETS:
+                # There are brackets we can further split on.
+                seen_other_brackets = True
+
+    return False
+
+
 CACHE_DIR = Path(user_cache_dir("black", version=__version__))