]> 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:

Fix typo in file_collection_and_discovery.md (GH-2860)
[etc/vim.git] / src / black / linegen.py
index 6008c773f943f0c22e2fec52eb6f5bfbcea29ed8..4dc242a1dfe05b9f8ef7ff53b26333073469cf57 100644 (file)
@@ -7,7 +7,7 @@ from typing import Collection, Iterator, List, Optional, Set, Union
 
 from black.nodes import WHITESPACE, RARROW, STATEMENT, STANDALONE_COMMENT
 from black.nodes import ASSIGNMENTS, OPENING_BRACKETS, CLOSING_BRACKETS
 
 from black.nodes import WHITESPACE, RARROW, STATEMENT, STANDALONE_COMMENT
 from black.nodes import ASSIGNMENTS, OPENING_BRACKETS, CLOSING_BRACKETS
-from black.nodes import Visitor, syms, first_child_is_arith, ensure_visible
+from black.nodes import Visitor, syms, is_arith_like, ensure_visible
 from black.nodes import is_docstring, is_empty_tuple, is_one_tuple, is_one_tuple_between
 from black.nodes import is_name_token, is_lpar_token, is_rpar_token
 from black.nodes import is_walrus_assignment, is_yield, is_vararg, is_multiline_string
 from black.nodes import is_docstring, is_empty_tuple, is_one_tuple, is_one_tuple_between
 from black.nodes import is_name_token, is_lpar_token, is_rpar_token
 from black.nodes import is_walrus_assignment, is_yield, is_vararg, is_multiline_string
@@ -21,10 +21,9 @@ from black.comments import generate_comments, list_comments, FMT_OFF
 from black.numerics import normalize_numeric_literal
 from black.strings import get_string_prefix, fix_docstring
 from black.strings import normalize_string_prefix, normalize_string_quotes
 from black.numerics import normalize_numeric_literal
 from black.strings import get_string_prefix, fix_docstring
 from black.strings import normalize_string_prefix, normalize_string_quotes
-from black.trans import Transformer, CannotTransform, StringMerger
-from black.trans import StringSplitter, StringParenWrapper, StringParenStripper
-from black.mode import Mode
-from black.mode import Feature
+from black.trans import Transformer, CannotTransform, StringMerger, StringSplitter
+from black.trans import StringParenWrapper, StringParenStripper, hug_power_op
+from black.mode import Mode, Feature, Preview
 
 from blib2to3.pytree import Node, Leaf
 from blib2to3.pgen2 import token
 
 from blib2to3.pytree import Node, Leaf
 from blib2to3.pgen2 import token
@@ -157,8 +156,12 @@ class LineGenerator(Visitor[Line]):
 
     def visit_simple_stmt(self, node: Node) -> Iterator[Line]:
         """Visit a statement without nested statements."""
 
     def visit_simple_stmt(self, node: Node) -> Iterator[Line]:
         """Visit a statement without nested statements."""
-        if first_child_is_arith(node):
-            wrap_in_parentheses(node, node.children[0], visible=False)
+        prev_type: Optional[int] = None
+        for child in node.children:
+            if (prev_type is None or prev_type == token.SEMI) and is_arith_like(child):
+                wrap_in_parentheses(node, child, visible=False)
+            prev_type = child.type
+
         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):
         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):
@@ -198,6 +201,28 @@ class LineGenerator(Visitor[Line]):
             yield from self.line()
             yield from self.visit(child)
 
             yield from self.line()
             yield from self.visit(child)
 
+    def visit_power(self, node: Node) -> Iterator[Line]:
+        for idx, leaf in enumerate(node.children[:-1]):
+            next_leaf = node.children[idx + 1]
+
+            if not isinstance(leaf, Leaf):
+                continue
+
+            value = leaf.value.lower()
+            if (
+                leaf.type == token.NUMBER
+                and next_leaf.type == syms.trailer
+                # Ensure that we are in an attribute trailer
+                and next_leaf.children[0].type == token.DOT
+                # It shouldn't wrap hexadecimal, binary and octal literals
+                and not value.startswith(("0x", "0b", "0o"))
+                # It shouldn't wrap complex literals
+                and "j" not in value
+            ):
+                wrap_in_parentheses(node, leaf)
+
+        yield from self.visit_default(node)
+
     def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
         """Remove a semicolon and put the other statement on a separate line."""
         yield from self.line()
     def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
         """Remove a semicolon and put the other statement on a separate line."""
         yield from self.line()
@@ -338,7 +363,7 @@ def transform_line(
         and not (line.inside_brackets and line.contains_standalone_comments())
     ):
         # Only apply basic string preprocessing, since lines shouldn't be split here.
         and not (line.inside_brackets and line.contains_standalone_comments())
     ):
         # Only apply basic string preprocessing, since lines shouldn't be split here.
-        if mode.experimental_string_processing:
+        if Preview.string_processing in mode:
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
@@ -381,7 +406,7 @@ def transform_line(
         # via type ... https://github.com/mypyc/mypyc/issues/884
         rhs = type("rhs", (), {"__call__": _rhs})()
 
         # via type ... https://github.com/mypyc/mypyc/issues/884
         rhs = type("rhs", (), {"__call__": _rhs})()
 
-        if mode.experimental_string_processing:
+        if Preview.string_processing in mode:
             if line.inside_brackets:
                 transformers = [
                     string_merge,
             if line.inside_brackets:
                 transformers = [
                     string_merge,
@@ -405,6 +430,9 @@ def transform_line(
                 transformers = [delimiter_split, standalone_comment_split, rhs]
             else:
                 transformers = [rhs]
                 transformers = [delimiter_split, standalone_comment_split, rhs]
             else:
                 transformers = [rhs]
+    # It's always safe to attempt hugging of power operations and pretty much every line
+    # could match.
+    transformers.append(hug_power_op)
 
     for transform in transformers:
         # We are accumulating lines in `result` because we might want to abort
 
     for transform in transformers:
         # We are accumulating lines in `result` because we might want to abort
@@ -515,7 +543,7 @@ def right_hand_split(
         # there are no standalone comments in the body
         and not body.contains_standalone_comments(0)
         # and we can actually remove the parens
         # there are no standalone comments in the body
         and not body.contains_standalone_comments(0)
         # and we can actually remove the parens
-        and can_omit_invisible_parens(body, line_length, omit_on_explode=omit)
+        and can_omit_invisible_parens(body, line_length)
     ):
         omit = {id(closing_bracket), *omit}
         try:
     ):
         omit = {id(closing_bracket), *omit}
         try:
@@ -940,6 +968,7 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
                 if (
                     prev
                     and prev.type == token.COMMA
                 if (
                     prev
                     and prev.type == token.COMMA
+                    and leaf.opening_bracket is not None
                     and not is_one_tuple_between(
                         leaf.opening_bracket, leaf, line.leaves
                     )
                     and not is_one_tuple_between(
                         leaf.opening_bracket, leaf, line.leaves
                     )
@@ -967,6 +996,7 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
             if (
                 prev
                 and prev.type == token.COMMA
             if (
                 prev
                 and prev.type == token.COMMA
+                and leaf.opening_bracket is not None
                 and not is_one_tuple_between(leaf.opening_bracket, leaf, line.leaves)
             ):
                 # Never omit bracket pairs with trailing commas.
                 and not is_one_tuple_between(leaf.opening_bracket, leaf, line.leaves)
             ):
                 # Never omit bracket pairs with trailing commas.