Skip to content

Fix format field parsing and lookup order #9438

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
youknowone merged 2 commits into from
Oct 29, 2026
Merged

Fix format field parsing and lookup order #9438

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
adc4d7e
Fix format field parsing and lookup order
codex Oct 28, 2026
f1a1d99
Retire passing logging formatter validation marker
codex Oct 28, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion Lib/test/test_logging.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4679,7 +4679,6 @@ def test_dollars(self):
f = logging.Formatter('${asctime}--', style='$')
self.assertTrue(f.usesTime())

@unittest.expectedFailure # TODO: RUSTPYTHON; ValueError: Unexpected error parsing format string
def test_format_validate(self):
# Check correct formatting
# Percentage style
Expand Down
1 change: 0 additions & 1 deletion Lib/test/test_str.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1092,7 +1092,6 @@ def test_issue18183(self):
'\U00100000'.ljust(3, '\U00010000')
'\U00100000'.rjust(3, '\U00010000')

@unittest.expectedFailure # TODO: RUSTPYTHON; '{0.}'.format() raises ValueError instead of IndexError
def test_format(self):
self.assertEqual(''.format(), '')
self.assertEqual('a'.format(), 'a')
Expand Down
308 changes: 151 additions & 157 deletions crates/common/src/format.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,6 @@
// spell-checker:ignore ddfe DTSF
use core::ops::Deref;
use core::{cmp, str::FromStr};
use itertools::{Itertools, PeekingNext};
use malachite_base::num::basic::floats::PrimitiveFloat;
use malachite_bigint::{BigInt, Sign};
use num_complex::Complex64;
Expand Down Expand Up @@ -1354,9 +1353,12 @@ pub enum FormatParseError {
MissingStartBracket,
UnescapedStartBracketInLiteral,
InvalidFormatSpecifier,
UnknownConversion,
MissingConversion,
MissingConversionColon,
UnexpectedStartBracket,
EmptyAttribute,
MissingRightBracket,
MissingRightSquareBracket,
InvalidCharacterAfterRightBracket,
TooManyDecimalDigits,
}
Expand All @@ -1375,43 +1377,56 @@ pub enum FieldNamePart {
StringIndex(Wtf8Buf),
}

impl FieldNamePart {
fn parse_part(
chars: &mut impl PeekingNext<Item = CodePoint>,
) -> Result<Option<Self>, FormatParseError> {
chars
.next()
.map(|ch| match ch.to_char_lossy() {
'.' => {
let mut attribute = Wtf8Buf::new();
for ch in chars.peeking_take_while(|ch| *ch != '.' && *ch != '[') {
attribute.push(ch);
}
if attribute.is_empty() {
Err(FormatParseError::EmptyAttribute)
} else {
Ok(Self::Attribute(attribute))
/// Lazily parse field suffixes so each lookup can happen before the next parse.
#[derive(Debug, PartialEq, Eq)]
pub struct FieldNameParts {
text: Wtf8Buf,
cursor: usize,
}

impl Iterator for FieldNameParts {
type Item = Result<FieldNamePart, FormatParseError>;

fn next(&mut self) -> Option<Self::Item> {
let ch = self.text[self.cursor..].code_points().next()?;
self.cursor += ch.len_wtf8();
let start = self.cursor;
Some(match ch.to_char_lossy() {
'.' => {
while let Some(ch) = self.text[self.cursor..].code_points().next() {
if ch == '.' || ch == '[' {
break;
}
self.cursor += ch.len_wtf8();
}
'[' => {
let mut index = Wtf8Buf::new();
for ch in chars {
if ch == ']' {
return if index.is_empty() {
Err(FormatParseError::EmptyAttribute)
} else if let Some(index) = parse_usize(&index) {
Ok(Self::Index(index))
} else {
Ok(Self::StringIndex(index))
};
}
index.push(ch);
if self.cursor == start {
Err(FormatParseError::EmptyAttribute)
} else {
Ok(FieldNamePart::Attribute(
self.text[start..self.cursor].to_owned(),
))
}
}
'[' => {
while let Some(ch) = self.text[self.cursor..].code_points().next() {
let end = self.cursor;
self.cursor += ch.len_wtf8();
if ch == ']' {
let index = &self.text[start..end];
return Some(if index.is_empty() {
Err(FormatParseError::EmptyAttribute)
} else {
parse_field_index(index).map(|number| match number {
Some(index) => FieldNamePart::Index(index),
None => FieldNamePart::StringIndex(index.to_owned()),
})
});
}
Err(FormatParseError::MissingRightBracket)
}
_ => Err(FormatParseError::InvalidCharacterAfterRightBracket),
})
.transpose()
Err(FormatParseError::MissingRightSquareBracket)
}
_ => Err(FormatParseError::InvalidCharacterAfterRightBracket),
})
}
}

Expand All @@ -1432,38 +1447,51 @@ fn parse_usize(s: &Wtf8) -> Option<usize> {
s.as_str().ok().and_then(|s| s.parse().ok())
}

impl FieldName {
pub fn parse(text: &Wtf8) -> Result<Self, FormatParseError> {
let mut chars = text.code_points().peekable();
let first: Wtf8Buf = chars
.peeking_take_while(|ch| *ch != '.' && *ch != '[')
.collect();
fn parse_field_index(text: &Wtf8) -> Result<Option<usize>, FormatParseError> {
if let Some(index) = parse_usize(text) {
if index > isize::MAX as usize {
return Err(FormatParseError::TooManyDecimalDigits);
}
Ok(Some(index))
} else if text
.as_str()
.is_ok_and(|s| s.bytes().all(|b| b.is_ascii_digit()))
{
Err(FormatParseError::TooManyDecimalDigits)
} else {
Ok(None)
}
}

impl FieldName {
pub fn split(text: &Wtf8) -> Result<(FieldType, FieldNameParts), FormatParseError> {
let split = text
.code_point_indices()
.find(|(_, ch)| *ch == '.' || *ch == '[')
.map_or(text.len(), |(index, _)| index);
let first = &text[..split];
let field_type = if first.is_empty() {
FieldType::Auto
} else if let Some(index) = parse_usize(&first) {
// Match CPython's get_integer: digits-only segment must fit in
// Py_ssize_t. Above that, raise ValueError at parse time.
if index > isize::MAX as usize {
return Err(FormatParseError::TooManyDecimalDigits);
}
} else if let Some(index) = parse_field_index(first)? {
FieldType::Index(index)
} else if first
.as_str()
.is_ok_and(|s| s.bytes().all(|b| b.is_ascii_digit()))
{
// All-digit segment whose value overflows usize itself.
return Err(FormatParseError::TooManyDecimalDigits);
} else {
FieldType::Keyword(first)
FieldType::Keyword(first.to_owned())
};
Ok((
field_type,
FieldNameParts {
text: text[split..].to_owned(),
cursor: 0,
},
))
}

let mut parts = Vec::new();
while let Some(part) = FieldNamePart::parse_part(&mut chars)? {
parts.push(part)
}

Ok(Self { field_type, parts })
pub fn parse(text: &Wtf8) -> Result<Self, FormatParseError> {
let (field_type, parts) = Self::split(text)?;
Ok(Self {
field_type,
parts: parts.collect::<Result<_, _>>()?,
})
}
}

Expand Down Expand Up @@ -1521,112 +1549,75 @@ impl FormatString {
Ok((FormatPart::Literal(result_string), "".as_ref()))
}

fn parse_part_in_brackets(text: &Wtf8) -> Result<FormatPart, FormatParseError> {
let mut chars = text.code_points().peekable();

let mut left = Wtf8Buf::new();
let mut right = Wtf8Buf::new();

let mut split = false;
let mut selected = &mut left;
let mut inside_brackets = false;

while let Some(char) = chars.next() {
if char == '[' {
inside_brackets = true;

selected.push(char);

while let Some(next_char) = chars.next() {
selected.push(next_char);
fn parse_spec(text: &Wtf8) -> Result<(FormatPart, &Wtf8), FormatParseError> {
let mut chars = text.code_point_indices();
if chars.next().is_none_or(|(_, ch)| ch != '{') {
return Err(FormatParseError::MissingStartBracket);
}
if text.len() == 1 {
return Err(FormatParseError::UnescapedStartBracketInLiteral);
}

if next_char == ']' {
inside_brackets = false;
break;
}
if chars.peek().is_none() {
// Braces, colons and conversion markers inside an item key are data.
let (field_end, mut delimiter) = loop {
let Some((index, ch)) = chars.next() else {
return Err(FormatParseError::UnmatchedBracket);
};
match ch.to_char_lossy() {
'[' => {
if !chars.any(|(_, ch)| ch == ']') {
return Err(FormatParseError::MissingRightBracket);
}
}
} else if char == ':' && !split && !inside_brackets {
split = true;
selected = &mut right;
} else {
selected.push(char);
'{' => return Err(FormatParseError::UnexpectedStartBracket),
'!' | ':' | '}' => break (index, ch),
_ => {}
}
};
let field_name = text[1..field_end].to_owned();
let mut delimiter_index = field_end;
let mut conversion_spec = None;
if delimiter == '!' {
conversion_spec = Some(chars.next().ok_or(FormatParseError::MissingConversion)?.1);
let Some((index, ch)) = chars.next() else {
return Err(FormatParseError::InvalidFormatSpecifier);
};
if ch != ':' && ch != '}' {
return Err(FormatParseError::MissingConversionColon);
}
delimiter_index = index;
delimiter = ch;
}

// before the comma is a keyword or arg index, after the comma is maybe a spec.
let arg_part: &Wtf8 = &left;

let format_spec = if split { right } else { Wtf8Buf::new() };

// left can still be the conversion (!r, !s, !a)
let parts: Vec<&Wtf8> = arg_part.splitn(2, "!".as_ref()).collect();
// before the bang is a keyword or arg index, after the comma is maybe a conversion spec.
let arg_part = parts[0];

let conversion_spec = parts
.get(1)
.map(|conversion| {
// conversions are only every one character
conversion
.code_points()
.exactly_one()
.map_err(|_| FormatParseError::UnknownConversion)
})
.transpose()?;

Ok(FormatPart::Field {
field_name: arg_part.to_owned(),
conversion_spec,
format_spec,
})
}

fn parse_spec(text: &Wtf8) -> Result<(FormatPart, &Wtf8), FormatParseError> {
let mut nested = false;
let mut end_bracket_pos = None;
let mut left = Wtf8Buf::new();

// There may be one layer nesting brackets in spec
for (idx, c) in text.code_point_indices() {
if idx == 0 {
if c != '{' {
return Err(FormatParseError::MissingStartBracket);
}
} else if c == '{' {
if nested {
let mut format_spec = Wtf8Buf::new();
if delimiter == ':' {
let spec_start = delimiter_index + 1;
let mut depth = 1usize;
let spec_end = loop {
let Some((index, ch)) = chars.next() else {
return Err(FormatParseError::InvalidFormatSpecifier);
};
if ch == '{' {
depth += 1;
} else if ch == '}' {
depth -= 1;
if depth == 0 {
break index;
}
}
nested = true;
left.push(c);
continue;
} else if c == '}' {
if nested {
nested = false;
left.push(c);
continue;
}
end_bracket_pos = Some(idx);
break;
} else {
left.push(c);
}
}
if let Some(pos) = end_bracket_pos {
let right = &text[pos..];
let format_part = Self::parse_part_in_brackets(&left)?;
Ok((format_part, &right[1..]))
} else if text.len() == 1 {
// Nothing follows the brace, so it is a stray one rather than a field
// that was left open: CPython separates "{" and "a{" from "{0" and
// "a{b" the same way. `{` is one byte, so a length of one is the brace
// on its own.
Err(FormatParseError::UnescapedStartBracketInLiteral)
} else {
Err(FormatParseError::UnmatchedBracket)
};
text[spec_start..spec_end].clone_into(&mut format_spec);
delimiter_index = spec_end;
}

Ok((
FormatPart::Field {
field_name,
conversion_spec,
format_spec,
},
&text[delimiter_index + 1..],
))
}
}

Expand Down Expand Up @@ -2478,7 +2469,10 @@ mod tests {
);
assert_eq!(parse("key.."), Err(FormatParseError::EmptyAttribute));
assert_eq!(parse("key[]"), Err(FormatParseError::EmptyAttribute));
assert_eq!(parse("key["), Err(FormatParseError::MissingRightBracket));
assert_eq!(
parse("key["),
Err(FormatParseError::MissingRightSquareBracket)
);
assert_eq!(
parse("key[0]after"),
Err(FormatParseError::InvalidCharacterAfterRightBracket)
Expand Down
Loading
Loading