Memoize text width (#6552)

This commit is contained in:
Micha Reiser 2023-09-06 09:10:13 +02:00 committed by GitHub
parent fa6bff0078
commit 5f59101811
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
14 changed files with 213 additions and 184 deletions

View file

@ -1,7 +1,5 @@
use std::borrow::Cow;
use unicode_width::UnicodeWidthChar;
use ruff_formatter::{format_args, write, FormatError, FormatOptions, SourceCode};
use ruff_python_ast::node::{AnyNodeRef, AstNode};
use ruff_python_trivia::{lines_after, lines_after_ignoring_trivia, lines_before};
@ -377,16 +375,12 @@ impl Format<PyFormatContext<'_>> for FormatTrailingEndOfLineComment<'_> {
0
} else {
// Start with 2 because of the two leading spaces.
let mut width = 2;
// SAFETY: The formatted file is <= 4GB, and each comment should as well.
#[allow(clippy::cast_possible_truncation)]
for c in normalized_comment.chars() {
width += match c {
'\t' => f.options().tab_width().value(),
c => c.width().unwrap_or(0) as u32,
}
}
let width = 2u32.saturating_add(
TextWidth::from_text(&normalized_comment, f.options().tab_width())
.width()
.expect("Expected comment not to contain any newlines")
.value(),
);
width
};
@ -430,11 +424,9 @@ pub(crate) struct FormatNormalizedComment<'a> {
impl Format<PyFormatContext<'_>> for FormatNormalizedComment<'_> {
fn fmt(&self, f: &mut Formatter<PyFormatContext>) -> FormatResult<()> {
match self.comment {
Cow::Borrowed(borrowed) => source_text_slice(
TextRange::at(self.range.start(), borrowed.text_len()),
ContainsNewlines::No,
)
.fmt(f),
Cow::Borrowed(borrowed) => {
source_text_slice(TextRange::at(self.range.start(), borrowed.text_len())).fmt(f)
}
Cow::Owned(ref owned) => {
write!(

View file

@ -8,6 +8,6 @@ pub struct FormatExprIpyEscapeCommand;
impl FormatNodeRule<ExprIpyEscapeCommand> for FormatExprIpyEscapeCommand {
fn fmt_fields(&self, item: &ExprIpyEscapeCommand, f: &mut PyFormatter) -> FormatResult<()> {
source_text_slice(item.range(), ContainsNewlines::No).fmt(f)
source_text_slice(item.range()).fmt(f)
}
}

View file

@ -21,7 +21,7 @@ impl FormatNodeRule<ExprName> for FormatExprName {
.text(f.context().source_code())
);
write!(f, [source_text_slice(*range, ContainsNewlines::No)])
write!(f, [source_text_slice(*range)])
}
fn fmt_dangling_comments(

View file

@ -24,7 +24,7 @@ impl Format<PyFormatContext<'_>> for FormatInt<'_> {
let normalized = normalize_integer(content);
match normalized {
Cow::Borrowed(_) => source_text_slice(range, ContainsNewlines::No).fmt(f),
Cow::Borrowed(_) => source_text_slice(range).fmt(f),
Cow::Owned(normalized) => text(&normalized, Some(range.start())).fmt(f),
}
}
@ -49,7 +49,7 @@ impl Format<PyFormatContext<'_>> for FormatFloat<'_> {
let normalized = normalize_floating_number(content);
match normalized {
Cow::Borrowed(_) => source_text_slice(range, ContainsNewlines::No).fmt(f),
Cow::Borrowed(_) => source_text_slice(range).fmt(f),
Cow::Owned(normalized) => text(&normalized, Some(range.start())).fmt(f),
}
}
@ -75,7 +75,7 @@ impl Format<PyFormatContext<'_>> for FormatComplex<'_> {
match normalized {
Cow::Borrowed(_) => {
source_text_slice(range.sub_end(TextSize::from(1)), ContainsNewlines::No).fmt(f)?;
source_text_slice(range.sub_end(TextSize::from(1))).fmt(f)?;
}
Cow::Owned(normalized) => {
text(&normalized, Some(range.start())).fmt(f)?;

View file

@ -314,7 +314,7 @@ impl FormatStringPart {
impl Format<PyFormatContext<'_>> for FormatStringPart {
fn fmt(&self, f: &mut PyFormatter) -> FormatResult<()> {
let (normalized, contains_newlines) = normalize_string(
let normalized = normalize_string(
f.context().locator().slice(self.range),
self.preferred_quotes,
self.is_raw_string,
@ -323,7 +323,7 @@ impl Format<PyFormatContext<'_>> for FormatStringPart {
write!(f, [self.prefix, self.preferred_quotes])?;
match normalized {
Cow::Borrowed(_) => {
source_text_slice(self.range(), contains_newlines).fmt(f)?;
source_text_slice(self.range()).fmt(f)?;
}
Cow::Owned(normalized) => {
text(&normalized, Some(self.start())).fmt(f)?;
@ -604,11 +604,7 @@ impl Format<PyFormatContext<'_>> for StringQuotes {
/// with the provided `style`.
///
/// Returns the normalized string and whether it contains new lines.
fn normalize_string(
input: &str,
quotes: StringQuotes,
is_raw: bool,
) -> (Cow<str>, ContainsNewlines) {
fn normalize_string(input: &str, quotes: StringQuotes, is_raw: bool) -> Cow<str> {
// The normalized string if `input` is not yet normalized.
// `output` must remain empty if `input` is already normalized.
let mut output = String::new();
@ -616,8 +612,6 @@ fn normalize_string(
// If `last_index` is `0` at the end, then the input is already normalized and can be returned as is.
let mut last_index = 0;
let mut newlines = ContainsNewlines::No;
let style = quotes.style;
let preferred_quote = style.as_char();
let opposite_quote = style.invert().as_char();
@ -638,9 +632,6 @@ fn normalize_string(
}
last_index = index + '\r'.len_utf8();
newlines = ContainsNewlines::Yes;
} else if c == '\n' {
newlines = ContainsNewlines::Yes;
} else if !quotes.triple && !is_raw {
if c == '\\' {
if let Some(next) = input.as_bytes().get(index + 1).copied().map(char::from) {
@ -675,7 +666,7 @@ fn normalize_string(
Cow::Owned(output)
};
(normalized, newlines)
normalized
}
/// For docstring indentation, black counts spaces as 1 and tabs by increasing the indentation up
@ -792,7 +783,7 @@ fn format_docstring(string_part: &FormatStringPart, f: &mut PyFormatter) -> Form
return string_part.fmt(f);
}
let (normalized, _) = normalize_string(
let normalized = normalize_string(
locator.slice(string_part),
string_part.preferred_quotes,
string_part.is_raw_string,
@ -837,7 +828,7 @@ fn format_docstring(string_part: &FormatStringPart, f: &mut PyFormatter) -> Form
let trimmed_line_range =
TextRange::at(offset, trim_end.text_len()).add_start(leading_whitespace);
if already_normalized {
source_text_slice(trimmed_line_range, ContainsNewlines::No).fmt(f)?;
source_text_slice(trimmed_line_range).fmt(f)?;
} else {
text(trim_both, Some(trimmed_line_range.start())).fmt(f)?;
}
@ -954,7 +945,7 @@ fn format_docstring_line(
let trimmed_line_range =
TextRange::at(offset, trim_end.text_len()).add_start(stripped_indentation);
if already_normalized {
source_text_slice(trimmed_line_range, ContainsNewlines::No).fmt(f)?;
source_text_slice(trimmed_line_range).fmt(f)?;
} else {
// All indents are ascii spaces, so the slicing is correct
text(

View file

@ -8,7 +8,7 @@ pub struct FormatIdentifier;
impl FormatRule<Identifier, PyFormatContext<'_>> for FormatIdentifier {
fn fmt(&self, item: &Identifier, f: &mut PyFormatter) -> FormatResult<()> {
source_text_slice(item.range(), ContainsNewlines::No).fmt(f)
source_text_slice(item.range()).fmt(f)
}
}

View file

@ -9,7 +9,7 @@ pub struct FormatStmtIpyEscapeCommand;
impl FormatNodeRule<StmtIpyEscapeCommand> for FormatStmtIpyEscapeCommand {
fn fmt_fields(&self, item: &StmtIpyEscapeCommand, f: &mut PyFormatter) -> FormatResult<()> {
source_text_slice(item.range(), ContainsNewlines::No).fmt(f)
source_text_slice(item.range()).fmt(f)
}
fn is_suppressed(

View file

@ -709,7 +709,7 @@ impl Format<PyFormatContext<'_>> for FormatVerbatimStatementRange {
}
} else {
// Non empty line, write the text of the line
verbatim_text(trimmed_line_range, logical_line.contains_newlines).fmt(f)?;
verbatim_text(trimmed_line_range).fmt(f)?;
// Write the line separator that terminates the line, except if it is the last line (that isn't separated by a hard line break).
if logical_line.has_trailing_newline {
@ -760,7 +760,6 @@ where
fn next(&mut self) -> Option<Self::Item> {
let mut parens = 0u32;
let mut contains_newlines = ContainsNewlines::No;
let (content_end, full_end) = loop {
match self.lexer.next() {
@ -768,18 +767,12 @@ where
Tok::Newline => break (range.start(), range.end()),
// Ignore if inside an expression
Tok::NonLogicalNewline if parens == 0 => break (range.start(), range.end()),
Tok::NonLogicalNewline => {
contains_newlines = ContainsNewlines::Yes;
}
Tok::Lbrace | Tok::Lpar | Tok::Lsqb => {
parens = parens.saturating_add(1);
}
Tok::Rbrace | Tok::Rpar | Tok::Rsqb => {
parens = parens.saturating_sub(1);
}
Tok::String { value, .. } if value.contains(['\n', '\r']) => {
contains_newlines = ContainsNewlines::Yes;
}
_ => {}
},
None => {
@ -790,7 +783,6 @@ where
self.last_line_end = self.content_end;
Some(Ok(LogicalLine {
content_range: TextRange::new(content_start, self.content_end),
contains_newlines: ContainsNewlines::No,
has_trailing_newline: false,
}))
} else {
@ -810,7 +802,6 @@ where
Some(Ok(LogicalLine {
content_range: TextRange::new(line_start, content_end),
contains_newlines,
has_trailing_newline: true,
}))
}
@ -822,8 +813,6 @@ impl<I> FusedIterator for LogicalLinesIter<I> where I: Iterator<Item = LexResult
struct LogicalLine {
/// The range of this lines content (excluding the trailing newline)
content_range: TextRange,
/// Whether the content in `content_range` contains any newlines.
contains_newlines: ContainsNewlines,
/// Does this logical line have a trailing newline or does it just happen to be the last line.
has_trailing_newline: bool,
}
@ -836,16 +825,14 @@ impl Ranged for LogicalLine {
struct VerbatimText {
verbatim_range: TextRange,
contains_newlines: ContainsNewlines,
}
fn verbatim_text<T>(item: T, contains_newlines: ContainsNewlines) -> VerbatimText
fn verbatim_text<T>(item: T) -> VerbatimText
where
T: Ranged,
{
VerbatimText {
verbatim_range: item.range(),
contains_newlines,
}
}
@ -859,13 +846,7 @@ impl Format<PyFormatContext<'_>> for VerbatimText {
match normalize_newlines(f.context().locator().slice(self.verbatim_range), ['\r']) {
Cow::Borrowed(_) => {
write!(
f,
[source_text_slice(
self.verbatim_range,
self.contains_newlines
)]
)?;
write!(f, [source_text_slice(self.verbatim_range,)])?;
}
Cow::Owned(cleaned) => {
write!(
@ -924,7 +905,7 @@ impl Format<PyFormatContext<'_>> for FormatSuppressedNode<'_> {
f,
[
leading_comments(node_comments.leading),
verbatim_text(self.node, ContainsNewlines::Detect),
verbatim_text(self.node),
trailing_comments(node_comments.trailing)
]
)
@ -937,13 +918,7 @@ pub(crate) fn write_suppressed_clause_header(
f: &mut PyFormatter,
) -> FormatResult<()> {
// Write the outer comments and format the node as verbatim
write!(
f,
[verbatim_text(
header.range(f.context().source())?,
ContainsNewlines::Detect
)]
)?;
write!(f, [verbatim_text(header.range(f.context().source())?)])?;
let comments = f.context().comments();
header.visit(&mut |child| {