]> 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:

Use parentheses on method access on float and int literals (#2799)
[etc/vim.git] / src / black / linegen.py
index 2e16b6fde4457d24c06992a57d1324e63675d9e4..b572ed0b52fdbe6d546365ffbce2876466b568d0 100644 (file)
@@ -5,12 +5,11 @@ from functools import partial, wraps
 import sys
 from typing import Collection, Iterator, List, Optional, Set, Union
 
-from dataclasses import dataclass, field
-
-from black.nodes import WHITESPACE, STATEMENT, STANDALONE_COMMENT
+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 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_stub_suite, is_stub_body, is_atom_with_invisible_parens
 from black.nodes import wrap_in_parentheses
@@ -22,10 +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.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
@@ -40,7 +38,8 @@ class CannotSplit(CannotTransform):
     """A readable split that fits the allotted line length is impossible."""
 
 
-@dataclass
+# This isn't a dataclass because @dataclass + Generic breaks mypyc.
+# See also https://github.com/mypyc/mypyc/issues/827.
 class LineGenerator(Visitor[Line]):
     """Generates reformatted Line objects.  Empty lines are not emitted.
 
@@ -48,9 +47,10 @@ class LineGenerator(Visitor[Line]):
     in ways that will no longer stringify to valid Python code on the tree.
     """
 
-    mode: Mode
-    remove_u_prefix: bool = False
-    current_line: Line = field(init=False)
+    def __init__(self, mode: Mode) -> None:
+        self.mode = mode
+        self.current_line: Line
+        self.__post_init__()
 
     def line(self, indent: int = 0) -> Iterator[Line]:
         """Generate a line.
@@ -90,9 +90,7 @@ class LineGenerator(Visitor[Line]):
 
             normalize_prefix(node, inside_brackets=any_open_brackets)
             if self.mode.string_normalization and node.type == token.STRING:
-                node.value = normalize_string_prefix(
-                    node.value, remove_u_prefix=self.remove_u_prefix
-                )
+                node.value = normalize_string_prefix(node.value)
                 node.value = normalize_string_quotes(node.value)
             if node.type == token.NUMBER:
                 normalize_numeric_literal(node)
@@ -126,7 +124,7 @@ class LineGenerator(Visitor[Line]):
         """Visit a statement.
 
         This implementation is shared for `if`, `while`, `for`, `try`, `except`,
-        `def`, `with`, `class`, `assert` and assignments.
+        `def`, `with`, `class`, `assert`, and assignments.
 
         The relevant Python language `keywords` for a given statement will be
         NAME leaves within it. This methods puts those on a separate line.
@@ -136,11 +134,19 @@ class LineGenerator(Visitor[Line]):
         """
         normalize_invisible_parens(node, parens_after=parens)
         for child in node.children:
-            if child.type == token.NAME and child.value in keywords:  # type: ignore
+            if is_name_token(child) and child.value in keywords:
                 yield from self.line()
 
             yield from self.visit(child)
 
+    def visit_match_case(self, node: Node) -> Iterator[Line]:
+        """Visit either a match or case statement."""
+        normalize_invisible_parens(node, parens_after=set())
+
+        yield from self.line()
+        for child in node.children:
+            yield from self.visit(child)
+
     def visit_suite(self, node: Node) -> Iterator[Line]:
         """Visit a suite."""
         if self.mode.is_pyi and is_stub_suite(node):
@@ -191,6 +197,28 @@ class LineGenerator(Visitor[Line]):
             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()
@@ -226,8 +254,9 @@ class LineGenerator(Visitor[Line]):
         if is_docstring(leaf) and "\\\n" not in leaf.value:
             # We're ignoring docstrings with backslash newline escapes because changing
             # indentation of those changes the AST representation of the code.
-            prefix = get_string_prefix(leaf.value)
-            docstring = leaf.value[len(prefix) :]  # Remove the prefix
+            docstring = normalize_string_prefix(leaf.value)
+            prefix = get_string_prefix(docstring)
+            docstring = docstring[len(prefix) :]  # Remove the prefix
             quote_char = docstring[0]
             # A natural way to remove the outer quotes is to do:
             #   docstring = docstring.strip(quote_char)
@@ -236,6 +265,7 @@ class LineGenerator(Visitor[Line]):
             # characters but only if they are the same as the first.
             quote_len = 1 if docstring[1] != quote_char else 3
             docstring = docstring[quote_len:-quote_len]
+            docstring_started_empty = not docstring
 
             if is_multiline_string(leaf):
                 indent = " " * 4 * self.current_line.depth
@@ -255,8 +285,7 @@ class LineGenerator(Visitor[Line]):
                         # Odd number of tailing backslashes, add some padding to
                         # avoid escaping the closing string quote.
                         docstring += " "
-            else:
-                # Add some padding if the docstring is empty.
+            elif not docstring_started_empty:
                 docstring = " "
 
             # We could enforce triple quotes at this point.
@@ -291,6 +320,10 @@ class LineGenerator(Visitor[Line]):
         self.visit_async_funcdef = self.visit_async_stmt
         self.visit_decorated = self.visit_decorators
 
+        # PEP 634
+        self.visit_match_stmt = self.visit_match_case
+        self.visit_case_block = self.visit_match_case
+
 
 def transform_line(
     line: Line, mode: Mode, features: Collection[Feature] = ()
@@ -326,7 +359,7 @@ 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.
-        if mode.experimental_string_processing:
+        if Preview.string_processing in mode:
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
@@ -334,7 +367,9 @@ def transform_line(
         transformers = [left_hand_split]
     else:
 
-        def rhs(line: Line, features: Collection[Feature]) -> Iterator[Line]:
+        def _rhs(
+            self: object, line: Line, features: Collection[Feature]
+        ) -> Iterator[Line]:
             """Wraps calls to `right_hand_split`.
 
             The calls increasingly `omit` right-hand trailers (bracket pairs with
@@ -361,7 +396,13 @@ def transform_line(
                 line, line_length=mode.line_length, features=features
             )
 
-        if mode.experimental_string_processing:
+        # HACK: nested functions (like _rhs) compiled by mypyc don't retain their
+        # __name__ attribute which is needed in `run_transformer` further down.
+        # Unfortunately a nested class breaks mypyc too. So a class must be created
+        # via type ... https://github.com/mypyc/mypyc/issues/884
+        rhs = type("rhs", (), {"__call__": _rhs})()
+
+        if Preview.string_processing in mode:
             if line.inside_brackets:
                 transformers = [
                     string_merge,
@@ -385,6 +426,9 @@ def transform_line(
                 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
@@ -502,14 +546,14 @@ def right_hand_split(
             yield from right_hand_split(line, line_length, features=features, omit=omit)
             return
 
-        except CannotSplit:
+        except CannotSplit as e:
             if not (
                 can_be_split(body)
                 or is_line_short_enough(body, line_length=line_length)
             ):
                 raise CannotSplit(
                     "Splitting failed, body is still too long and can't be split."
-                )
+                ) from e
 
             elif head.contains_multiline_strings() or tail.contains_multiline_strings():
                 raise CannotSplit(
@@ -517,7 +561,7 @@ def right_hand_split(
                     " satisfy the splitting algorithm because the head or the tail"
                     " contains multiline strings which by definition never fit one"
                     " line."
-                )
+                ) from e
 
     ensure_visible(opening_bracket)
     ensure_visible(closing_bracket)
@@ -573,6 +617,20 @@ def bracket_split_build_line(
                 original.is_def
                 and opening_bracket.value == "("
                 and not any(leaf.type == token.COMMA for leaf in leaves)
+                # In particular, don't add one within a parenthesized return annotation.
+                # Unfortunately the indicator we're in a return annotation (RARROW) may
+                # be defined directly in the parent node, the parent of the parent ...
+                # and so on depending on how complex the return annotation is.
+                # This isn't perfect and there's some false negatives but they are in
+                # contexts were a comma is actually fine.
+                and not any(
+                    node.prev_sibling.type == RARROW
+                    for node in (
+                        leaves[0].parent,
+                        getattr(leaves[0].parent, "parent", None),
+                    )
+                    if isinstance(node, Node) and isinstance(node.prev_sibling, Leaf)
+                )
             )
 
             if original.is_import or no_commas:
@@ -620,13 +678,13 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
     try:
         last_leaf = line.leaves[-1]
     except IndexError:
-        raise CannotSplit("Line empty")
+        raise CannotSplit("Line empty") from None
 
     bt = line.bracket_tracker
     try:
         delimiter_priority = bt.max_delimiter_priority(exclude={id(last_leaf)})
     except ValueError:
-        raise CannotSplit("No delimiters found")
+        raise CannotSplit("No delimiters found") from None
 
     if delimiter_priority == DOT_PRIORITY:
         if bt.delimiter_count_with_priority(delimiter_priority) == 1:
@@ -777,10 +835,11 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
             elif node.type == syms.import_from:
                 # "import from" nodes store parentheses directly as part of
                 # the statement
-                if child.type == token.LPAR:
+                if is_lpar_token(child):
+                    assert is_rpar_token(node.children[-1])
                     # make parentheses invisible
-                    child.value = ""  # type: ignore
-                    node.children[-1].value = ""  # type: ignore
+                    child.value = ""
+                    node.children[-1].value = ""
                 elif child.type != token.STAR:
                     # insert invisible parentheses
                     node.insert_child(index, Leaf(token.LPAR, ""))
@@ -825,11 +884,11 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:
 
     first = node.children[0]
     last = node.children[-1]
-    if first.type == token.LPAR and last.type == token.RPAR:
+    if is_lpar_token(first) and is_rpar_token(last):
         middle = node.children[1]
         # make parentheses invisible
-        first.value = ""  # type: ignore
-        last.value = ""  # type: ignore
+        first.value = ""
+        last.value = ""
         maybe_make_parens_invisible_in_atom(middle, parent=parent)
 
         if is_atom_with_invisible_parens(middle):
@@ -905,6 +964,7 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
                 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
                     )
@@ -932,6 +992,7 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
             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.
@@ -960,14 +1021,19 @@ def run_transformer(
 
         result.extend(transform_line(transformed_line, mode=mode, features=features))
 
-    if not (
-        transform.__name__ == "rhs"
-        and line.bracket_tracker.invisible
-        and not any(bracket.value for bracket in line.bracket_tracker.invisible)
-        and not line.contains_multiline_strings()
-        and not result[0].contains_uncollapsable_type_comments()
-        and not result[0].contains_unsplittable_type_ignore()
-        and not is_line_short_enough(result[0], line_length=mode.line_length)
+    if (
+        transform.__class__.__name__ != "rhs"
+        or not line.bracket_tracker.invisible
+        or any(bracket.value for bracket in line.bracket_tracker.invisible)
+        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)
+        # 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
+        # call to `append_leaves()` to fail.
+        or any(leaf.parent is None for leaf in line.leaves)
     ):
         return result