from mypy_extensions import trait
-from black.brackets import BracketMatchError
from black.comments import contains_pragma_comment
from black.lines import Line, append_leaves
from black.mode import Feature
is_empty_lpar,
is_empty_par,
is_empty_rpar,
+ is_part_of_annotation,
parent_type,
replace_child,
syms,
Requirements:
(A) The line contains adjacent strings such that ALL of the validation checks
- listed in StringMerger.__validate_msg(...)'s docstring pass.
+ listed in StringMerger._validate_msg(...)'s docstring pass.
OR
(B) The line contains a string which uses line continuation backslashes.
and is_valid_index(i + 1)
and LL[i + 1].type == token.STRING
):
+ if is_part_of_annotation(leaf):
+ return TErr("String is part of type annotation.")
return Ok(i)
if leaf.type == token.STRING and "\\\n" in leaf.value:
Returns:
Ok(new_line), if ALL of the validation checks found in
- __validate_msg(...) pass.
+ _validate_msg(...) pass.
OR
Err(CannotTransform), otherwise.
"""
next_str_idx += 1
+ # Take a note on the index of the non-STRING leaf.
+ non_string_idx = next_str_idx
+
S_leaf = Leaf(token.STRING, S)
if self.normalize_strings:
S_leaf.value = normalize_string_quotes(S_leaf.value)
string_leaf = Leaf(token.STRING, S_leaf.value.replace(BREAK_MARK, ""))
if atom_node is not None:
- replace_child(atom_node, string_leaf)
+ # If not all children of the atom node are merged (this can happen
+ # when there is a standalone comment in the middle) ...
+ if non_string_idx - string_idx < len(atom_node.children):
+ # We need to replace the old STRING leaves with the new string leaf.
+ first_child_idx = LL[string_idx].remove()
+ for idx in range(string_idx + 1, non_string_idx):
+ LL[idx].remove()
+ if first_child_idx is not None:
+ atom_node.insert_child(first_child_idx, string_leaf)
+ else:
+ # Else replace the atom node with the new string leaf.
+ replace_child(atom_node, string_leaf)
# Build the final line ('new_line') that this method will later return.
new_line = line.clone()
def _validate_msg(line: Line, string_idx: int) -> TResult[None]:
"""Validate (M)erge (S)tring (G)roup
- Transform-time string validation logic for __merge_string_group(...).
+ Transform-time string validation logic for _merge_string_group(...).
Returns:
* Ok(None), if ALL validation checks (listed below) pass.
- The set of all string prefixes in the string group is of
length greater than one and is not equal to {"", "f"}.
- The string group consists of raw strings.
+ - The string group is stringified type annotations. We don't want to
+ process stringified type annotations since pyright doesn't support
+ them spanning multiple string values. (NOTE: mypy, pytype, pyre do
+ support them, so we can change if pyright also gains support in the
+ future. See https://github.com/microsoft/pyright/issues/4359.)
"""
# We first check for "inner" stand-alone comments (i.e. stand-alone
# comments that have a string leaf before them AND after them).
new_line = line.clone()
new_line.comments = line.comments.copy()
- try:
- append_leaves(new_line, line, LL[: string_idx - 1])
- except BracketMatchError:
- # HACK: I believe there is currently a bug somewhere in
- # right_hand_split() that is causing brackets to not be tracked
- # properly by a shared BracketTracker.
- append_leaves(new_line, line, LL[: string_idx - 1], preformatted=True)
+ append_leaves(new_line, line, LL[: string_idx - 1])
string_leaf = Leaf(token.STRING, LL[string_idx].value)
LL[string_idx - 1].remove()
max_string_length = self.line_length - offset
return max_string_length
+ @staticmethod
+ def _prefer_paren_wrap_match(LL: List[Leaf]) -> Optional[int]:
+ """
+ Returns:
+ string_idx such that @LL[string_idx] is equal to our target (i.e.
+ matched) string, if this line matches the "prefer paren wrap" statement
+ requirements listed in the 'Requirements' section of the StringParenWrapper
+ class's docstring.
+ OR
+ None, otherwise.
+ """
+ # The line must start with a string.
+ if LL[0].type != token.STRING:
+ return None
+
+ # If the string is surrounded by commas (or is the first/last child)...
+ prev_sibling = LL[0].prev_sibling
+ next_sibling = LL[0].next_sibling
+ if not prev_sibling and not next_sibling and parent_type(LL[0]) == syms.atom:
+ # If it's an atom string, we need to check the parent atom's siblings.
+ parent = LL[0].parent
+ assert parent is not None # For type checkers.
+ prev_sibling = parent.prev_sibling
+ next_sibling = parent.next_sibling
+ if (not prev_sibling or prev_sibling.type == token.COMMA) and (
+ not next_sibling or next_sibling.type == token.COMMA
+ ):
+ return 0
+
+ return None
+
def iter_fexpr_spans(s: str) -> Iterator[Tuple[int, int]]:
"""
def do_splitter_match(self, line: Line) -> TMatchResult:
LL = line.leaves
+ if self._prefer_paren_wrap_match(LL) is not None:
+ return TErr("Line needs to be wrapped in parens first.")
+
is_valid_index = is_valid_index_factory(LL)
idx = 0
string_op_leaves = self._get_string_operator_leaves(LL)
string_op_leaves_length = (
- sum([len(str(prefix_leaf)) for prefix_leaf in string_op_leaves]) + 1
+ sum(len(str(prefix_leaf)) for prefix_leaf in string_op_leaves) + 1
if string_op_leaves
else 0
)
# prefix, and the current custom split did NOT originally use a
# prefix...
if (
- next_value != self._normalize_f_string(next_value, prefix)
- and use_custom_breakpoints
+ use_custom_breakpoints
and not csplit.has_prefix
+ and (
+ # `next_value == prefix + QUOTE` happens when the custom
+ # split is an empty string.
+ next_value == prefix + QUOTE
+ or next_value != self._normalize_f_string(next_value, prefix)
+ )
):
# Then `csplit.break_idx` will be off by one after removing
# the 'f' prefix.
class StringParenWrapper(BaseStringSplitter, CustomSplitMapMixin):
"""
- StringTransformer that splits non-"atom" strings (i.e. strings that do not
- exist on lines by themselves).
+ StringTransformer that wraps strings in parens and then splits at the LPAR.
Requirements:
All of the requirements listed in BaseStringSplitter's docstring in
OR
* The line is a dictionary key assignment where some valid key is being
assigned the value of some string.
+ OR
+ * The line is an lambda expression and the value is a string.
+ OR
+ * The line starts with an "atom" string that prefers to be wrapped in
+ parens. It's preferred to be wrapped when the string is surrounded by
+ commas (or is the first/last child).
Transformations:
The chosen string is wrapped in parentheses and then split at the LPAR.
changed such that it no longer needs to be given its own line,
StringParenWrapper relies on StringParenStripper to clean up the
parentheses it created.
+
+ For "atom" strings that prefers to be wrapped in parens, it requires
+ StringSplitter to hold the split until the string is wrapped in parens.
"""
def do_splitter_match(self, line: Line) -> TMatchResult:
or self._else_match(LL)
or self._assert_match(LL)
or self._assign_match(LL)
- or self._dict_match(LL)
+ or self._dict_or_lambda_match(LL)
+ or self._prefer_paren_wrap_match(LL)
)
if string_idx is not None:
return None
@staticmethod
- def _dict_match(LL: List[Leaf]) -> Optional[int]:
+ def _dict_or_lambda_match(LL: List[Leaf]) -> Optional[int]:
"""
Returns:
string_idx such that @LL[string_idx] is equal to our target (i.e.
matched) string, if this line matches the dictionary key assignment
- statement requirements listed in the 'Requirements' section of this
- classes' docstring.
+ statement or lambda expression requirements listed in the
+ 'Requirements' section of this classes' docstring.
OR
None, otherwise.
"""
- # If this line is apart of a dictionary key assignment...
- if syms.dictsetmaker in [parent_type(LL[0]), parent_type(LL[0].parent)]:
+ # If this line is a part of a dictionary key assignment or lambda expression...
+ parent_types = [parent_type(LL[0]), parent_type(LL[0].parent)]
+ if syms.dictsetmaker in parent_types or syms.lambdef in parent_types:
is_valid_index = is_valid_index_factory(LL)
for i, leaf in enumerate(LL):
- # We MUST find a colon...
- if leaf.type == token.COLON:
+ # We MUST find a colon, it can either be dict's or lambda's colon...
+ if leaf.type == token.COLON and i < len(LL) - 1:
idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1
# That colon MUST be followed by a string...
f" (left_leaves={left_leaves}, right_leaves={right_leaves})"
)
old_rpar_leaf = right_leaves.pop()
+ elif right_leaves and right_leaves[-1].type == token.RPAR:
+ # Special case for lambda expressions as dict's value, e.g.:
+ # my_dict = {
+ # "key": lambda x: f"formatted: {x},
+ # }
+ # After wrapping the dict's value with parentheses, the string is
+ # followed by a RPAR but its opening bracket is lambda's, not
+ # the string's:
+ # "key": (lambda x: f"formatted: {x}),
+ opening_bracket = right_leaves[-1].opening_bracket
+ if opening_bracket is not None and opening_bracket in left_leaves:
+ index = left_leaves.index(opening_bracket)
+ if (
+ index > 0
+ and index < len(left_leaves) - 1
+ and left_leaves[index - 1].type == token.COLON
+ and left_leaves[index + 1].value == "lambda"
+ ):
+ right_leaves.pop()
append_leaves(string_line, line, right_leaves)