Delimit multiline expressions according to math operator priority

[etc/vim.git] / black.py

diff --git a/black.py b/black.py
index 5458198b17560621545fdcb0d15a4c32fcdb6522..36380e6ae847b2314b4ce0e8d2c816071cdc418b 100644 (file)
--- a/black.py
+++ b/black.py
@@ -1,5 +1,3 @@
-#!/usr/bin/env python3
-

 import asyncio
 import pickle
 from asyncio.base_events import BaseEventLoop
 import asyncio
 import pickle
 from asyncio.base_events import BaseEventLoop


@@ -43,7 +41,7 @@ from blib2to3 import pygram, pytree
 from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.parse import ParseError
 
 from blib2to3.pgen2 import driver, token
 from blib2to3.pgen2.parse import ParseError
 

-__version__ = "18.4a3"
+__version__ = "18.4a6"

 DEFAULT_LINE_LENGTH = 88
 
 # types
 DEFAULT_LINE_LENGTH = 88
 
 # types


@@ -89,11 +87,11 @@ class FormatError(Exception):
         self.consumed = consumed
 
     def trim_prefix(self, leaf: Leaf) -> None:
         self.consumed = consumed
 
     def trim_prefix(self, leaf: Leaf) -> None:

-        leaf.prefix = leaf.prefix[self.consumed:]
+        leaf.prefix = leaf.prefix[self.consumed :]

 
     def leaf_from_consumed(self, leaf: Leaf) -> Leaf:
         """Returns a new Leaf from the consumed part of the prefix."""
 
     def leaf_from_consumed(self, leaf: Leaf) -> Leaf:
         """Returns a new Leaf from the consumed part of the prefix."""

-        unformatted_prefix = leaf.prefix[:self.consumed]
+        unformatted_prefix = leaf.prefix[: self.consumed]

         return Leaf(token.NEWLINE, unformatted_prefix)
 
 
         return Leaf(token.NEWLINE, unformatted_prefix)
 
 


@@ -194,6 +192,7 @@ def main(
         write_back = WriteBack.YES
     report = Report(check=check, quiet=quiet)
     if len(sources) == 0:
         write_back = WriteBack.YES
     report = Report(check=check, quiet=quiet)
     if len(sources) == 0:

+        out("No paths given. Nothing to do 😴")

         ctx.exit(0)
         return
 
         ctx.exit(0)
         return
 


@@ -245,7 +244,7 @@ def reformat_one(
                 )
             ):
                 changed = Changed.YES
                 )
             ):
                 changed = Changed.YES

-            if write_back != WriteBack.DIFF and changed is not Changed.NO:
+            if write_back == WriteBack.YES and changed is not Changed.NO:

                 write_cache(cache, [src], line_length)
         report.done(src, changed)
     except Exception as exc:
                 write_cache(cache, [src], line_length)
         report.done(src, changed)
     except Exception as exc:


@@ -312,7 +311,7 @@ async def schedule_formatting(
 
     if cancelled:
         await asyncio.gather(*cancelled, loop=loop, return_exceptions=True)
 
     if cancelled:
         await asyncio.gather(*cancelled, loop=loop, return_exceptions=True)

-    if write_back != WriteBack.DIFF and formatted:
+    if write_back == WriteBack.YES and formatted:

         write_cache(cache, formatted, line_length)
 
 
         write_cache(cache, formatted, line_length)
 
 


@@ -554,19 +553,20 @@ COMPARATORS = {
     token.GREATEREQUAL,
 }
 MATH_OPERATORS = {
     token.GREATEREQUAL,
 }
 MATH_OPERATORS = {

+    token.VBAR,
+    token.CIRCUMFLEX,
+    token.AMPER,
+    token.LEFTSHIFT,
+    token.RIGHTSHIFT,

     token.PLUS,
     token.MINUS,
     token.STAR,
     token.SLASH,
     token.PLUS,
     token.MINUS,
     token.STAR,
     token.SLASH,

-    token.VBAR,
-    token.AMPER,
+    token.DOUBLESLASH,

     token.PERCENT,
     token.PERCENT,

-    token.CIRCUMFLEX,
+    token.AT,

     token.TILDE,
     token.TILDE,

-    token.LEFTSHIFT,
-    token.RIGHTSHIFT,

     token.DOUBLESTAR,
     token.DOUBLESTAR,

-    token.DOUBLESLASH,

 }
 STARS = {token.STAR, token.DOUBLESTAR}
 VARARGS_PARENTS = {
 }
 STARS = {token.STAR, token.DOUBLESTAR}
 VARARGS_PARENTS = {


@@ -582,13 +582,45 @@ UNPACKING_PARENTS = {
     syms.listmaker,
     syms.testlist_gexp,
 }
     syms.listmaker,
     syms.testlist_gexp,
 }

+TEST_DESCENDANTS = {
+    syms.test,
+    syms.lambdef,
+    syms.or_test,
+    syms.and_test,
+    syms.not_test,
+    syms.comparison,
+    syms.star_expr,
+    syms.expr,
+    syms.xor_expr,
+    syms.and_expr,
+    syms.shift_expr,
+    syms.arith_expr,
+    syms.trailer,
+    syms.term,
+    syms.power,
+}

 COMPREHENSION_PRIORITY = 20
 COMPREHENSION_PRIORITY = 20

-COMMA_PRIORITY = 10
-TERNARY_PRIORITY = 7
-LOGIC_PRIORITY = 5
-STRING_PRIORITY = 4
-COMPARATOR_PRIORITY = 3
-MATH_PRIORITY = 1
+COMMA_PRIORITY = 18
+TERNARY_PRIORITY = 16
+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,
+}

 
 
 @dataclass
 
 
 @dataclass


@@ -698,6 +730,10 @@ class BracketTracker:
 
         return False
 
 
         return False
 

+    def get_open_lsqb(self) -> Optional[Leaf]:
+        """Return the most recent opening square bracket (if any)."""
+        return self.bracket_match.get((self.depth - 1, token.RSQB))
+

 
 @dataclass
 class Line:
 
 @dataclass
 class Line:


@@ -723,14 +759,17 @@ class Line:
         if not has_value:
             return
 
         if not has_value:
             return
 

+        if token.COLON == leaf.type and self.is_class_paren_empty:
+            del self.leaves[-2:]

         if self.leaves and not preformatted:
             # Note: at this point leaf.prefix should be empty except for
             # imports, for which we only preserve newlines.
         if self.leaves and not preformatted:
             # Note: at this point leaf.prefix should be empty except for
             # imports, for which we only preserve newlines.

-            leaf.prefix += whitespace(leaf)
+            leaf.prefix += whitespace(
+                leaf, complex_subscript=self.is_complex_subscript(leaf)
+            )

         if self.inside_brackets or not preformatted:
             self.bracket_tracker.mark(leaf)
             self.maybe_remove_trailing_comma(leaf)
         if self.inside_brackets or not preformatted:
             self.bracket_tracker.mark(leaf)
             self.maybe_remove_trailing_comma(leaf)

-

         if not self.append_comment(leaf):
             self.leaves.append(leaf)
 
         if not self.append_comment(leaf):
             self.leaves.append(leaf)
 


@@ -818,6 +857,22 @@ class Line:
             and self.leaves[0].value == "yield"
         )
 
             and self.leaves[0].value == "yield"
         )
 

+    @property
+    def is_class_paren_empty(self) -> bool:
+        """Is this a class with no base classes but using parentheses?
+
+        Those are unnecessary and should be removed.
+        """
+        return (
+            bool(self)
+            and len(self.leaves) == 4
+            and self.is_class
+            and self.leaves[2].type == token.LPAR
+            and self.leaves[2].value == "("
+            and self.leaves[3].type == token.RPAR
+            and self.leaves[3].value == ")"
+        )
+

     def contains_standalone_comments(self, depth_limit: int = sys.maxsize) -> bool:
         """If so, needs to be split before emitting."""
         for leaf in self.leaves:
     def contains_standalone_comments(self, depth_limit: int = sys.maxsize) -> bool:
         """If so, needs to be split before emitting."""
         for leaf in self.leaves:


@@ -846,9 +901,14 @@ class Line:
                 self.remove_trailing_comma()
                 return True
 
                 self.remove_trailing_comma()
                 return True
 

-        # 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.
+        # For parens let's check if it's safe to remove the comma.
+        # Imports are always safe.
+        if self.is_import:
+            self.remove_trailing_comma()
+            return True
+
+        # Otheriwsse, 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
         commas = 0
         opening = closing.opening_bracket
         depth = closing.bracket_depth + 1
         commas = 0
         opening = closing.opening_bracket


@@ -859,7 +919,7 @@ class Line:
         else:
             return False
 
         else:
             return False
 

-        for leaf in self.leaves[_opening_index + 1:]:
+        for leaf in self.leaves[_opening_index + 1 :]:

             if leaf is closing:
                 break
 
             if leaf is closing:
                 break
 


@@ -920,6 +980,24 @@ class Line:
                 self.comments[i] = (comma_index - 1, comment)
         self.leaves.pop()
 
                 self.comments[i] = (comma_index - 1, comment)
         self.leaves.pop()
 

+    def is_complex_subscript(self, leaf: Leaf) -> bool:
+        """Return True iff `leaf` is part of a slice with non-trivial exprs."""
+        open_lsqb = (
+            leaf if leaf.type == token.LSQB else self.bracket_tracker.get_open_lsqb()
+        )
+        if open_lsqb is None:
+            return False
+
+        subscript_start = open_lsqb.next_sibling
+        if (
+            isinstance(subscript_start, Node)
+            and subscript_start.type == syms.subscriptlist
+        ):
+            subscript_start = child_towards(subscript_start, leaf)
+        return subscript_start is not None and any(
+            n.type in TEST_DESCENDANTS for n in subscript_start.pre_order()
+        )
+

     def __str__(self) -> str:
         """Render the line."""
         if not self:
     def __str__(self) -> str:
         """Render the line."""
         if not self:


@@ -1303,8 +1381,12 @@ BRACKETS = OPENING_BRACKETS | CLOSING_BRACKETS
 ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT}
 
 
 ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT}
 
 

-def whitespace(leaf: Leaf) -> str:  # noqa C901
-    """Return whitespace prefix if needed for the given `leaf`."""
+def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str:  # noqa C901
+    """Return whitespace prefix if needed for the given `leaf`.
+
+    `complex_subscript` signals whether the given leaf is part of a subscription
+    which has non-trivial arguments, like arithmetic expressions or function calls.
+    """

     NO = ""
     SPACE = " "
     DOUBLESPACE = "  "
     NO = ""
     SPACE = " "
     DOUBLESPACE = "  "


@@ -1318,7 +1400,10 @@ def whitespace(leaf: Leaf) -> str:  # noqa C901
         return DOUBLESPACE
 
     assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}"
         return DOUBLESPACE
 
     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}:
+    if (
+        t == token.COLON
+        and p.type not in {syms.subscript, syms.subscriptlist, syms.sliceop}
+    ):

         return NO
 
     prev = leaf.prev_sibling
         return NO
 
     prev = leaf.prev_sibling


@@ -1328,7 +1413,13 @@ def whitespace(leaf: Leaf) -> str:  # noqa C901
             return NO
 
         if t == token.COLON:
             return NO
 
         if t == token.COLON:

-            return SPACE if prevp.type == token.COMMA else NO
+            if prevp.type == token.COLON:
+                return NO
+
+            elif prevp.type != token.COMMA and not complex_subscript:
+                return NO
+
+            return SPACE

 
         if prevp.type == token.EQUAL:
             if prevp.parent:
 
         if prevp.type == token.EQUAL:
             if prevp.parent:


@@ -1349,7 +1440,7 @@ def whitespace(leaf: Leaf) -> str:  # noqa C901
 
         elif prevp.type == token.COLON:
             if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}:
 
         elif prevp.type == token.COLON:
             if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}:

-                return NO
+                return SPACE if complex_subscript else NO

 
         elif (
             prevp.parent
 
         elif (
             prevp.parent


@@ -1455,7 +1546,7 @@ def whitespace(leaf: Leaf) -> str:  # noqa C901
         if prev and prev.type == token.LPAR:
             return NO
 
         if prev and prev.type == token.LPAR:
             return NO
 

-    elif p.type == syms.subscript:
+    elif p.type in {syms.subscript, syms.sliceop}:

         # indexing
         if not prev:
             assert p.parent is not None, "subscripts are always parented"
         # indexing
         if not prev:
             assert p.parent is not None, "subscripts are always parented"


@@ -1464,7 +1555,7 @@ def whitespace(leaf: Leaf) -> str:  # noqa C901
 
             return NO
 
 
             return NO
 

-        else:
+        elif not complex_subscript:

             return NO
 
     elif p.type == syms.atom:
             return NO
 
     elif p.type == syms.atom:


@@ -1534,6 +1625,14 @@ def preceding_leaf(node: Optional[LN]) -> Optional[Leaf]:
     return None
 
 
     return None
 
 

+def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]:
+    """Return the child of `ancestor` that contains `descendant`."""
+    node: Optional[LN] = descendant
+    while node and node.parent != ancestor:
+        node = node.parent
+    return node
+
+

 def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
     """Return the priority of the `leaf` delimiter, given a line break after it.
 
 def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
     """Return the priority of the `leaf` delimiter, given a line break after it.
 


@@ -1566,7 +1665,7 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
         and leaf.parent
         and leaf.parent.type not in {syms.factor, syms.star_expr}
     ):
         and leaf.parent
         and leaf.parent.type not in {syms.factor, syms.star_expr}
     ):

-        return MATH_PRIORITY
+        return MATH_PRIORITIES[leaf.type]

 
     if leaf.type in COMPARATORS:
         return COMPARATOR_PRIORITY
 
     if leaf.type in COMPARATORS:
         return COMPARATOR_PRIORITY


@@ -1747,7 +1846,8 @@ 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.
     """Split line into many lines, starting with the first matching bracket pair.
 
     Note: this usually looks weird, only use this for function definitions.

-    Prefer RHS otherwise.
+    Prefer RHS otherwise.  This is why this function is not symmetrical with
+    :func:`right_hand_split` which also handles optional parentheses.

     """
     head = Line(depth=line.depth)
     body = Line(depth=line.depth + 1, inside_brackets=True)
     """
     head = Line(depth=line.depth)
     body = Line(depth=line.depth + 1, inside_brackets=True)


@@ -1787,7 +1887,10 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]:
 def right_hand_split(
     line: Line, py36: bool = False, omit: Collection[LeafID] = ()
 ) -> Iterator[Line]:
 def right_hand_split(
     line: Line, py36: bool = False, omit: Collection[LeafID] = ()
 ) -> Iterator[Line]:

-    """Split line into many lines, starting with the last matching bracket pair."""
+    """Split line into many lines, starting with the last matching bracket pair.
+
+    If the split was by optional parentheses, attempt splitting without them, too.
+    """

     head = Line(depth=line.depth)
     body = Line(depth=line.depth + 1, inside_brackets=True)
     tail = Line(depth=line.depth)
     head = Line(depth=line.depth)
     body = Line(depth=line.depth + 1, inside_brackets=True)
     tail = Line(depth=line.depth)


@@ -1826,20 +1929,25 @@ def right_hand_split(
     bracket_split_succeeded_or_raise(head, body, tail)
     assert opening_bracket and closing_bracket
     if (
     bracket_split_succeeded_or_raise(head, body, tail)
     assert opening_bracket and closing_bracket
     if (

+        # the opening bracket is an optional paren

         opening_bracket.type == token.LPAR
         and not opening_bracket.value
         opening_bracket.type == token.LPAR
         and not opening_bracket.value

+        # the closing bracket is an optional paren

         and closing_bracket.type == token.RPAR
         and not closing_bracket.value
         and closing_bracket.type == token.RPAR
         and not closing_bracket.value

+        # there are no delimiters or standalone comments in the body
+        and not body.bracket_tracker.delimiters
+        and not line.contains_standalone_comments(0)
+        # and it's not an import (optional parens are the only thing we can split
+        # on in this case; attempting a split without them is a waste of time)
+        and not line.is_import

     ):
     ):

-        # These parens were optional. If there aren't any delimiters or standalone
-        # comments in the body, they were unnecessary and another split without
-        # them should be attempted.
-        if not (
-            body.bracket_tracker.delimiters or line.contains_standalone_comments(0)
-        ):
-            omit = {id(closing_bracket), *omit}
+        omit = {id(closing_bracket), *omit}
+        try:

             yield from right_hand_split(line, py36=py36, omit=omit)
             return
             yield from right_hand_split(line, py36=py36, omit=omit)
             return

+        except CannotSplit:
+            pass

 
     ensure_visible(opening_bracket)
     ensure_visible(closing_bracket)
 
     ensure_visible(opening_bracket)
     ensure_visible(closing_bracket)


@@ -1994,6 +2102,7 @@ def explode_split(
 
     try:
         yield from delimiter_split(new_lines[1], py36)
 
     try:
         yield from delimiter_split(new_lines[1], py36)

+

     except CannotSplit:
         yield new_lines[1]
 
     except CannotSplit:
         yield new_lines[1]
 


@@ -2061,7 +2170,7 @@ def normalize_string_quotes(leaf: Leaf) -> None:
     unescaped_new_quote = re.compile(rf"(([^\\]|^)(\\\\)*){new_quote}")
     escaped_new_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{new_quote}")
     escaped_orig_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{orig_quote}")
     unescaped_new_quote = re.compile(rf"(([^\\]|^)(\\\\)*){new_quote}")
     escaped_new_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{new_quote}")
     escaped_orig_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{orig_quote}")

-    body = leaf.value[first_quote_pos + len(orig_quote):-len(orig_quote)]
+    body = leaf.value[first_quote_pos + len(orig_quote) : -len(orig_quote)]

     if "r" in prefix.casefold():
         if unescaped_new_quote.search(body):
             # There's at least one unescaped new_quote in this raw string
     if "r" in prefix.casefold():
         if unescaped_new_quote.search(body):
             # There's at least one unescaped new_quote in this raw string


@@ -2103,17 +2212,7 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
     for child in list(node.children):
         if check_lpar:
             if child.type == syms.atom:
     for child in list(node.children):
         if check_lpar:
             if child.type == syms.atom:

-                if not (
-                    is_empty_tuple(child)
-                    or is_one_tuple(child)
-                    or max_delimiter_priority_in_atom(child) >= COMMA_PRIORITY
-                ):
-                    first = child.children[0]
-                    last = child.children[-1]
-                    if first.type == token.LPAR and last.type == token.RPAR:
-                        # make parentheses invisible
-                        first.value = ""  # type: ignore
-                        last.value = ""  # type: ignore
+                maybe_make_parens_invisible_in_atom(child)

             elif is_one_tuple(child):
                 # wrap child in visible parentheses
                 lpar = Leaf(token.LPAR, "(")
             elif is_one_tuple(child):
                 # wrap child in visible parentheses
                 lpar = Leaf(token.LPAR, "(")


@@ -2130,6 +2229,29 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
         check_lpar = isinstance(child, Leaf) and child.value in parens_after
 
 
         check_lpar = isinstance(child, Leaf) and child.value in parens_after
 
 

+def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
+    """If it's safe, make the parens in the atom `node` invisible, recusively."""
+    if (
+        node.type != syms.atom
+        or is_empty_tuple(node)
+        or is_one_tuple(node)
+        or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
+    ):
+        return False
+
+    first = node.children[0]
+    last = node.children[-1]
+    if first.type == token.LPAR and last.type == token.RPAR:
+        # 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])
+        return True
+
+    return False
+
+

 def is_empty_tuple(node: LN) -> bool:
     """Return True if `node` holds an empty tuple."""
     return (
 def is_empty_tuple(node: LN) -> bool:
     """Return True if `node` holds an empty tuple."""
     return (


@@ -2231,7 +2353,7 @@ def is_python36(node: Node) -> bool:
 
     Currently looking for:
     - f-strings; and
 
     Currently looking for:
     - f-strings; and

-    - trailing commas after * or ** in function signatures.
+    - trailing commas after * or ** in function signatures and calls.

     """
     for n in node.pre_order():
         if n.type == token.STRING:
     """
     for n in node.pre_order():
         if n.type == token.STRING:


@@ -2240,7 +2362,7 @@ def is_python36(node: Node) -> bool:
                 return True
 
         elif (
                 return True
 
         elif (

-            n.type == syms.typedargslist
+            n.type in {syms.typedargslist, syms.arglist}

             and n.children
             and n.children[-1].type == token.COMMA
         ):
             and n.children
             and n.children[-1].type == token.COMMA
         ):


@@ -2248,6 +2370,11 @@ def is_python36(node: Node) -> bool:
                 if ch.type in STARS:
                     return True
 
                 if ch.type in STARS:
                     return True
 

+                if ch.type == syms.argument:
+                    for argch in ch.children:
+                        if argch.type in STARS:
+                            return True
+

     return False
 
 
     return False