]> git.madduck.net Git - etc/vim.git/blobdiff - src/black/linegen.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:

Try newer clang in diff-shades job (#3904)
[etc/vim.git] / src / black / linegen.py
index 7afb1733939d6d0472663b371c88f8a5cc435bda..9ddd4619f691275a8a1ce9b70d4bce7348596281 100644 (file)
@@ -2,7 +2,7 @@
 Generating lines of code.
 """
 import sys
-from dataclasses import dataclass
+from dataclasses import replace
 from enum import Enum, auto
 from functools import partial, wraps
 from typing import Collection, Iterator, List, Optional, Set, Union, cast
@@ -16,6 +16,7 @@ from black.brackets import (
 from black.comments import FMT_OFF, generate_comments, list_comments
 from black.lines import (
     Line,
+    RHSResult,
     append_leaves,
     can_be_split,
     can_omit_invisible_parens,
@@ -35,6 +36,7 @@ from black.nodes import (
     Visitor,
     ensure_visible,
     is_arith_like,
+    is_async_stmt_or_funcdef,
     is_atom_with_invisible_parens,
     is_docstring,
     is_empty_tuple,
@@ -47,6 +49,7 @@ from black.nodes import (
     is_stub_body,
     is_stub_suite,
     is_tuple_containing_walrus,
+    is_type_ignore_comment_string,
     is_vararg,
     is_walrus_assignment,
     is_yield,
@@ -109,6 +112,17 @@ class LineGenerator(Visitor[Line]):
             self.current_line.depth += indent
             return  # Line is empty, don't emit. Creating a new one unnecessary.
 
+        if (
+            Preview.improved_async_statements_handling in self.mode
+            and len(self.current_line.leaves) == 1
+            and is_async_stmt_or_funcdef(self.current_line.leaves[0])
+        ):
+            # Special case for async def/for/with statements. `visit_async_stmt`
+            # adds an `ASYNC` leaf then visits the child def/for/with statement
+            # nodes. Line yields from those nodes shouldn't treat the former
+            # `ASYNC` leaf as a complete line.
+            return
+
         complete_line = self.current_line
         self.current_line = Line(mode=self.mode, depth=complete_line.depth + indent)
         yield complete_line
@@ -202,6 +216,18 @@ class LineGenerator(Visitor[Line]):
 
             yield from self.visit(child)
 
+    def visit_typeparams(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[0].prefix = ""
+
+    def visit_typevartuple(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[1].prefix = ""
+
+    def visit_paramspec(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[1].prefix = ""
+
     def visit_dictsetmaker(self, node: Node) -> Iterator[Line]:
         if Preview.wrap_long_dict_values_in_parens in self.mode:
             for i, child in enumerate(node.children):
@@ -255,7 +281,9 @@ class LineGenerator(Visitor[Line]):
 
     def visit_suite(self, node: Node) -> Iterator[Line]:
         """Visit a suite."""
-        if self.mode.is_pyi and is_stub_suite(node):
+        if (
+            self.mode.is_pyi or Preview.dummy_implementations in self.mode
+        ) and is_stub_suite(node):
             yield from self.visit(node.children[2])
         else:
             yield from self.visit_default(node)
@@ -270,7 +298,9 @@ class LineGenerator(Visitor[Line]):
 
         is_suite_like = node.parent and node.parent.type in STATEMENT
         if is_suite_like:
-            if self.mode.is_pyi and is_stub_body(node):
+            if (
+                self.mode.is_pyi or Preview.dummy_implementations in self.mode
+            ) and is_stub_body(node):
                 yield from self.visit_default(node)
             else:
                 yield from self.line(+1)
@@ -279,7 +309,7 @@ class LineGenerator(Visitor[Line]):
 
         else:
             if (
-                not self.mode.is_pyi
+                not (self.mode.is_pyi or Preview.dummy_implementations in self.mode)
                 or not node.parent
                 or not is_stub_suite(node.parent)
             ):
@@ -300,8 +330,11 @@ class LineGenerator(Visitor[Line]):
                 break
 
         internal_stmt = next(children)
-        for child in internal_stmt.children:
-            yield from self.visit(child)
+        if Preview.improved_async_statements_handling in self.mode:
+            yield from self.visit(internal_stmt)
+        else:
+            for child in internal_stmt.children:
+                yield from self.visit(child)
 
     def visit_decorators(self, node: Node) -> Iterator[Line]:
         """Visit decorators."""
@@ -364,6 +397,24 @@ class LineGenerator(Visitor[Line]):
             node.insert_child(index, Node(syms.atom, [lpar, operand, rpar]))
         yield from self.visit_default(node)
 
+    def visit_tname(self, node: Node) -> Iterator[Line]:
+        """
+        Add potential parentheses around types in function parameter lists to be made
+        into real parentheses in case the type hint is too long to fit on a line
+        Examples:
+        def foo(a: int, b: float = 7): ...
+
+        ->
+
+        def foo(a: (int), b: (float) = 7): ...
+        """
+        if Preview.parenthesize_long_type_hints in self.mode:
+            assert len(node.children) == 3
+            if maybe_make_parens_invisible_in_atom(node.children[2], parent=node):
+                wrap_in_parentheses(node, node.children[2], visible=False)
+
+        yield from self.visit_default(node)
+
     def visit_STRING(self, leaf: Leaf) -> Iterator[Line]:
         if Preview.hex_codes_in_unicode_sequences in self.mode:
             normalize_unicode_escape_sequences(leaf)
@@ -465,7 +516,14 @@ class LineGenerator(Visitor[Line]):
         self.visit_except_clause = partial(v, keywords={"except"}, parens={"except"})
         self.visit_with_stmt = partial(v, keywords={"with"}, parens={"with"})
         self.visit_classdef = partial(v, keywords={"class"}, parens=Ø)
-        self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS)
+
+        # When this is moved out of preview, add ":" directly to ASSIGNMENTS in nodes.py
+        if Preview.parenthesize_long_type_hints in self.mode:
+            assignments = ASSIGNMENTS | {":"}
+        else:
+            assignments = ASSIGNMENTS
+        self.visit_expr_stmt = partial(v, keywords=Ø, parens=assignments)
+
         self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"})
         self.visit_import_from = partial(v, keywords=Ø, parens={"import"})
         self.visit_del_stmt = partial(v, keywords=Ø, parens={"del"})
@@ -505,7 +563,7 @@ def transform_line(
         and not line.should_split_rhs
         and not line.magic_trailing_comma
         and (
-            is_line_short_enough(line, line_length=mode.line_length, line_str=line_str)
+            is_line_short_enough(line, mode=mode, line_str=line_str)
             or line.contains_unsplittable_type_ignore()
         )
         and not (line.inside_brackets and line.contains_standalone_comments())
@@ -520,7 +578,7 @@ def transform_line(
     else:
 
         def _rhs(
-            self: object, line: Line, features: Collection[Feature]
+            self: object, line: Line, features: Collection[Feature], mode: Mode
         ) -> Iterator[Line]:
             """Wraps calls to `right_hand_split`.
 
@@ -529,14 +587,12 @@ def transform_line(
             bracket pair instead.
             """
             for omit in generate_trailers_to_omit(line, mode.line_length):
-                lines = list(
-                    right_hand_split(line, mode.line_length, features, omit=omit)
-                )
+                lines = list(right_hand_split(line, mode, features, omit=omit))
                 # Note: this check is only able to figure out if the first line of the
                 # *current* transformation fits in the line length.  This is true only
                 # for simple cases.  All others require running more transforms via
                 # `transform_line()`.  This check doesn't know if those would succeed.
-                if is_line_short_enough(lines[0], line_length=mode.line_length):
+                if is_line_short_enough(lines[0], mode=mode):
                     yield from lines
                     return
 
@@ -544,9 +600,7 @@ def transform_line(
             # This mostly happens to multiline strings that are by definition
             # reported as not fitting a single line, as well as lines that contain
             # trailing commas (those have to be exploded).
-            yield from right_hand_split(
-                line, line_length=mode.line_length, features=features
-            )
+            yield from right_hand_split(line, mode, features=features)
 
         # HACK: nested functions (like _rhs) compiled by mypyc don't retain their
         # __name__ attribute which is needed in `run_transformer` further down.
@@ -604,7 +658,9 @@ class _BracketSplitComponent(Enum):
     tail = auto()
 
 
-def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator[Line]:
+def left_hand_split(
+    line: Line, _features: Collection[Feature], mode: Mode
+) -> 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.
@@ -649,20 +705,9 @@ def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator
             yield result
 
 
-@dataclass
-class _RHSResult:
-    """Intermediate split result from a right hand split."""
-
-    head: Line
-    body: Line
-    tail: Line
-    opening_bracket: Leaf
-    closing_bracket: Leaf
-
-
 def right_hand_split(
     line: Line,
-    line_length: int,
+    mode: Mode,
     features: Collection[Feature] = (),
     omit: Collection[LeafID] = (),
 ) -> Iterator[Line]:
@@ -676,14 +721,14 @@ def right_hand_split(
     """
     rhs_result = _first_right_hand_split(line, omit=omit)
     yield from _maybe_split_omitting_optional_parens(
-        rhs_result, line, line_length, features=features, omit=omit
+        rhs_result, line, mode, features=features, omit=omit
     )
 
 
 def _first_right_hand_split(
     line: Line,
     omit: Collection[LeafID] = (),
-) -> _RHSResult:
+) -> RHSResult:
     """Split the line into head, body, tail starting with the last bracket pair.
 
     Note: this function should not have side effects. It's relied upon by
@@ -725,13 +770,13 @@ def _first_right_hand_split(
         tail_leaves, line, opening_bracket, component=_BracketSplitComponent.tail
     )
     bracket_split_succeeded_or_raise(head, body, tail)
-    return _RHSResult(head, body, tail, opening_bracket, closing_bracket)
+    return RHSResult(head, body, tail, opening_bracket, closing_bracket)
 
 
 def _maybe_split_omitting_optional_parens(
-    rhs: _RHSResult,
+    rhs: RHSResult,
     line: Line,
-    line_length: int,
+    mode: Mode,
     features: Collection[Feature] = (),
     omit: Collection[LeafID] = (),
 ) -> Iterator[Line]:
@@ -749,38 +794,39 @@ def _maybe_split_omitting_optional_parens(
         # there are no standalone comments in the body
         and not rhs.body.contains_standalone_comments(0)
         # and we can actually remove the parens
-        and can_omit_invisible_parens(rhs.body, line_length)
+        and can_omit_invisible_parens(rhs, mode.line_length)
     ):
         omit = {id(rhs.closing_bracket), *omit}
         try:
-            # The _RHSResult Omitting Optional Parens.
+            # The RHSResult Omitting Optional Parens.
             rhs_oop = _first_right_hand_split(line, omit=omit)
             if not (
                 Preview.prefer_splitting_right_hand_side_of_assignments in line.mode
                 # the split is right after `=`
                 and len(rhs.head.leaves) >= 2
                 and rhs.head.leaves[-2].type == token.EQUAL
-                # the left side of assignement contains brackets
+                # the left side of assignment contains brackets
                 and any(leaf.type in BRACKETS for leaf in rhs.head.leaves[:-1])
                 # the left side of assignment is short enough (the -1 is for the ending
                 # optional paren)
-                and is_line_short_enough(rhs.head, line_length=line_length - 1)
+                and is_line_short_enough(
+                    rhs.head, mode=replace(mode, line_length=mode.line_length - 1)
+                )
                 # the left side of assignment won't explode further because of magic
                 # trailing comma
                 and rhs.head.magic_trailing_comma is None
                 # the split by omitting optional parens isn't preferred by some other
                 # reason
-                and not _prefer_split_rhs_oop(rhs_oop, line_length=line_length)
+                and not _prefer_split_rhs_oop(rhs_oop, mode)
             ):
                 yield from _maybe_split_omitting_optional_parens(
-                    rhs_oop, line, line_length, features=features, omit=omit
+                    rhs_oop, line, mode, features=features, omit=omit
                 )
                 return
 
         except CannotSplit as e:
             if not (
-                can_be_split(rhs.body)
-                or is_line_short_enough(rhs.body, line_length=line_length)
+                can_be_split(rhs.body) or is_line_short_enough(rhs.body, mode=mode)
             ):
                 raise CannotSplit(
                     "Splitting failed, body is still too long and can't be split."
@@ -804,7 +850,7 @@ def _maybe_split_omitting_optional_parens(
             yield result
 
 
-def _prefer_split_rhs_oop(rhs_oop: _RHSResult, line_length: int) -> bool:
+def _prefer_split_rhs_oop(rhs_oop: RHSResult, mode: Mode) -> bool:
     """
     Returns whether we should prefer the result from a split omitting optional parens.
     """
@@ -824,7 +870,7 @@ def _prefer_split_rhs_oop(rhs_oop: _RHSResult, line_length: int) -> bool:
             # the first line still contains the `=`)
             any(leaf.type == token.EQUAL for leaf in rhs_oop.head.leaves)
             # the first line is short enough
-            and is_line_short_enough(rhs_oop.head, line_length=line_length)
+            and is_line_short_enough(rhs_oop.head, mode=mode)
         )
         # contains unsplittable type ignore
         or rhs_oop.head.contains_unsplittable_type_ignore()
@@ -902,6 +948,13 @@ def bracket_split_build_line(
                     )
                     if isinstance(node, Node) and isinstance(node.prev_sibling, Leaf)
                 )
+                # Except the false negatives above for PEP 604 unions where we
+                # can't add the comma.
+                and not (
+                    leaves[0].parent
+                    and leaves[0].parent.next_sibling
+                    and leaves[0].parent.next_sibling.type == token.VBAR
+                )
             )
 
             if original.is_import or no_commas:
@@ -940,16 +993,39 @@ def dont_increase_indentation(split_func: Transformer) -> Transformer:
     """
 
     @wraps(split_func)
-    def split_wrapper(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
-        for split_line in split_func(line, features):
+    def split_wrapper(
+        line: Line, features: Collection[Feature], mode: Mode
+    ) -> Iterator[Line]:
+        for split_line in split_func(line, features, mode):
             normalize_prefix(split_line.leaves[0], inside_brackets=True)
             yield split_line
 
     return split_wrapper
 
 
+def _get_last_non_comment_leaf(line: Line) -> Optional[int]:
+    for leaf_idx in range(len(line.leaves) - 1, 0, -1):
+        if line.leaves[leaf_idx].type != STANDALONE_COMMENT:
+            return leaf_idx
+    return None
+
+
+def _safe_add_trailing_comma(safe: bool, delimiter_priority: int, line: Line) -> Line:
+    if (
+        safe
+        and delimiter_priority == COMMA_PRIORITY
+        and line.leaves[-1].type != token.COMMA
+        and line.leaves[-1].type != STANDALONE_COMMENT
+    ):
+        new_comma = Leaf(token.COMMA, ",")
+        line.append(new_comma)
+    return line
+
+
 @dont_increase_indentation
-def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
+def delimiter_split(
+    line: Line, features: Collection[Feature], mode: Mode
+) -> Iterator[Line]:
     """Split according to delimiters of the highest priority.
 
     If the appropriate Features are given, the split will add trailing commas
@@ -989,7 +1065,8 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
             )
             current_line.append(leaf)
 
-    for leaf in line.leaves:
+    last_non_comment_leaf = _get_last_non_comment_leaf(line)
+    for leaf_idx, leaf in enumerate(line.leaves):
         yield from append_to_line(leaf)
 
         for comment_after in line.comments_after(leaf):
@@ -1006,6 +1083,15 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
                     trailing_comma_safe and Feature.TRAILING_COMMA_IN_CALL in features
                 )
 
+        if (
+            Preview.add_trailing_comma_consistently in mode
+            and last_leaf.type == STANDALONE_COMMENT
+            and leaf_idx == last_non_comment_leaf
+        ):
+            current_line = _safe_add_trailing_comma(
+                trailing_comma_safe, delimiter_priority, current_line
+            )
+
         leaf_priority = bt.delimiters.get(id(leaf))
         if leaf_priority == delimiter_priority:
             yield current_line
@@ -1014,20 +1100,15 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
                 mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets
             )
     if current_line:
-        if (
-            trailing_comma_safe
-            and delimiter_priority == COMMA_PRIORITY
-            and current_line.leaves[-1].type != token.COMMA
-            and current_line.leaves[-1].type != STANDALONE_COMMENT
-        ):
-            new_comma = Leaf(token.COMMA, ",")
-            current_line.append(new_comma)
+        current_line = _safe_add_trailing_comma(
+            trailing_comma_safe, delimiter_priority, current_line
+        )
         yield current_line
 
 
 @dont_increase_indentation
 def standalone_comment_split(
-    line: Line, features: Collection[Feature] = ()
+    line: Line, features: Collection[Feature], mode: Mode
 ) -> Iterator[Line]:
     """Split standalone comments from the rest of the line."""
     if not line.contains_standalone_comments(0):
@@ -1312,7 +1393,7 @@ def maybe_make_parens_invisible_in_atom(
     Returns whether the node should itself be wrapped in invisible parentheses.
     """
     if (
-        node.type != syms.atom
+        node.type not in (syms.atom, syms.expr)
         or is_empty_tuple(node)
         or is_one_tuple(node)
         or (is_yield(node) and parent.type != syms.expr_stmt)
@@ -1336,6 +1417,7 @@ def maybe_make_parens_invisible_in_atom(
             syms.except_clause,
             syms.funcdef,
             syms.with_stmt,
+            syms.tname,
             # these ones aren't useful to end users, but they do please fuzzers
             syms.for_stmt,
             syms.del_stmt,
@@ -1348,8 +1430,13 @@ def maybe_make_parens_invisible_in_atom(
     if is_lpar_token(first) and is_rpar_token(last):
         middle = node.children[1]
         # make parentheses invisible
-        first.value = ""
-        last.value = ""
+        if (
+            # If the prefix of `middle` includes a type comment with
+            # ignore annotation, then we do not remove the parentheses
+            not is_type_ignore_comment_string(middle.prefix.strip())
+        ):
+            first.value = ""
+            last.value = ""
         maybe_make_parens_invisible_in_atom(
             middle,
             parent=parent,
@@ -1480,7 +1567,7 @@ def run_transformer(
     if not line_str:
         line_str = line_to_string(line)
     result: List[Line] = []
-    for transformed_line in transform(line, features):
+    for transformed_line in transform(line, features, mode):
         if str(transformed_line).strip("\n") == line_str:
             raise CannotTransform("Line transformer returned an unchanged result")
 
@@ -1495,7 +1582,7 @@ def run_transformer(
         or line.contains_multiline_strings()
         or result[0].contains_uncollapsable_type_comments()
         or result[0].contains_unsplittable_type_ignore()
-        or is_line_short_enough(result[0], line_length=mode.line_length)
+        or is_line_short_enough(result[0], mode=mode)
         # If any leaves have no parents (which _can_ occur since
         # `transform(line)` potentially destroys the line's underlying node
         # structure), then we can't proceed. Doing so would cause the below
@@ -1510,8 +1597,6 @@ def run_transformer(
     second_opinion = run_transformer(
         line_copy, transform, mode, features_fop, line_str=line_str
     )
-    if all(
-        is_line_short_enough(ln, line_length=mode.line_length) for ln in second_opinion
-    ):
+    if all(is_line_short_enough(ln, mode=mode) for ln in second_opinion):
         result = second_opinion
     return result