Skip to content

Commit

Permalink
Add BestFittingMode (#5184)
Browse files Browse the repository at this point in the history
## Summary
Black supports for layouts when it comes to breaking binary expressions:

```rust
#[derive(Copy, Clone, Debug, Eq, PartialEq)]
enum BinaryLayout {
    /// Put each operand on their own line if either side expands
    Default,

    /// Try to expand the left to make it fit. Add parentheses if the left or right don't fit.
    ///
    ///```python
    /// [
    ///     a,
    ///     b
    /// ] & c
    ///```
    ExpandLeft,

    /// Try to expand the right to make it fix. Add parentheses if the left or right don't fit.
    ///
    /// ```python
    /// a & [
    ///     b,
    ///     c
    /// ]
    /// ```
    ExpandRight,

    /// Both the left and right side can be expanded. Try in the following order:
    /// * expand the right side
    /// * expand the left side
    /// * expand both sides
    ///
    /// to make the expression fit
    ///
    /// ```python
    /// [
    ///     a,
    ///     b
    /// ] & [
    ///     c,
    ///     d
    /// ]
    /// ```
    ExpandRightThenLeft,
}
```

Our current implementation only handles `ExpandRight` and `Default` correctly. `ExpandLeft` turns out to be surprisingly hard. This PR adds a new `BestFittingMode` parameter to `BestFitting` to support `ExpandLeft`.

There are 3 variants that `ExpandLeft` must support:

**Variant 1**: Everything fits on the line (easy)

```python
[a, b] + c
```

**Variant 2**: Left breaks, but right fits on the line. Doesn't need parentheses

```python
[
	a,
	b
] + c
```

**Variant 3**: The left breaks, but there's still not enough space for the right hand side. Parenthesize the whole expression:

```python
(
	[
		a, 
		b
	]
	+ ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc
)
```

Solving Variant 1 and 2 on their own is straightforward The printer gives us this behavior by nesting right inside of the group of left:

```
group(&format_args![
	if_group_breaks(&text("(")),
	soft_block_indent(&group(&format_args![
		left, 
		soft_line_break_or_space(), 
		op, 
		space(), 
		group(&right)
	])),
	if_group_breaks(&text(")"))
])
```

The fundamental problem is that the outer group, which adds the parentheses, always breaks if the left side breaks. That means, we end up with

```python
(
	[
		a,
		b
	] + c
)
```

which is not what we want (we only want parentheses if the right side doesn't fit). 

Okay, so nesting groups don't work because of the outer parentheses. Sequencing groups doesn't work because it results in a right-to-left breaking which is the opposite of what we want. 

Could we use best fitting? Almost! 

```
best_fitting![
	// All flat
	format_args![left, space(), op, space(), right],
	// Break left
	format_args!(group(&left).should_expand(true), space(), op, space(), right],
	// Break all
	format_args![
		text("("), 
		block_indent!(&format_args![
			left, 
			hard_line_break(), 
			op,
			space()
			right
		])
	]
]
```

I hope I managed to write this up correctly. The problem is that the printer never reaches the 3rd variant because the second variant always fits:

* The `group(&left).should_expand(true)` changes the group so that all `soft_line_breaks` are turned into hard line breaks. This is necessary because we want to test if the content fits if we break after the `[`. 
* Now, the whole idea of `best_fitting` is that you can pretend that some content fits on the line when it actually does not. The way this works is that the printer **only** tests if all the content of the variant **up to** the first line break fits on the line (we insert that line break by using `should_expand(true))`. The printer doesn't care whether the rest `a\n, b\n ] + c` all fits on (multiple?) lines. 

Why does breaking right work but not breaking the left? The difference is that we can make the decision whether to parenthesis the expression based on the left expression. We can't do this for breaking left because the decision whether to insert parentheses or not would depend on a lookahead: will the right side break. We simply don't know this yet when printing the parentheses (it would work for the right parentheses but not for the left and indent).

What we kind of want here is to tell the printer: Look, what comes here may or may not fit on a single line but we don't care. Simply test that what comes **after** fits on a line. 

This PR adds a new `BestFittingMode` that has a new `AllLines` option that gives us the desired behavior of testing all content and not just up to the first line break. 

## Test Plan

I added a new example to  `BestFitting::with_mode`
  • Loading branch information
MichaReiser authored Jun 20, 2023
1 parent 6929fcc commit d9e59b2
Show file tree
Hide file tree
Showing 8 changed files with 329 additions and 120 deletions.
126 changes: 119 additions & 7 deletions crates/ruff_formatter/src/builders.rs
Original file line number Diff line number Diff line change
Expand Up @@ -2131,11 +2131,12 @@ impl<'a, 'buf, Context> FillBuilder<'a, 'buf, Context> {
/// The first variant is the most flat, and the last is the most expanded variant.
/// See [`best_fitting!`] macro for a more in-detail documentation
#[derive(Copy, Clone)]
pub struct FormatBestFitting<'a, Context> {
pub struct BestFitting<'a, Context> {
variants: Arguments<'a, Context>,
mode: BestFittingMode,
}

impl<'a, Context> FormatBestFitting<'a, Context> {
impl<'a, Context> BestFitting<'a, Context> {
/// Creates a new best fitting IR with the given variants. The method itself isn't unsafe
/// but it is to discourage people from using it because the printer will panic if
/// the slice doesn't contain at least the least and most expanded variants.
Expand All @@ -2150,11 +2151,119 @@ impl<'a, Context> FormatBestFitting<'a, Context> {
"Requires at least the least expanded and most expanded variants"
);

Self { variants }
Self {
variants,
mode: BestFittingMode::default(),
}
}

/// Changes the mode used by this best fitting element to determine whether a variant fits.
///
/// ## Examples
///
/// ### All Lines
///
/// ```
/// use ruff_formatter::{Formatted, LineWidth, format, format_args, SimpleFormatOptions};
/// use ruff_formatter::prelude::*;
///
/// # fn main() -> FormatResult<()> {
/// let formatted = format!(
/// SimpleFormatContext::default(),
/// [
/// best_fitting!(
/// // Everything fits on a single line
/// format_args!(
/// group(&format_args![
/// text("["),
/// soft_block_indent(&format_args![
/// text("1,"),
/// soft_line_break_or_space(),
/// text("2,"),
/// soft_line_break_or_space(),
/// text("3"),
/// ]),
/// text("]")
/// ]),
/// space(),
/// text("+"),
/// space(),
/// text("aVeryLongIdentifier")
/// ),
///
/// // Breaks after `[` and prints each elements on a single line
/// // The group is necessary because the variant, by default is printed in flat mode and a
/// // hard line break indicates that the content doesn't fit.
/// format_args!(
/// text("["),
/// group(&block_indent(&format_args![text("1,"), hard_line_break(), text("2,"), hard_line_break(), text("3")])).should_expand(true),
/// text("]"),
/// space(),
/// text("+"),
/// space(),
/// text("aVeryLongIdentifier")
/// ),
///
/// // Adds parentheses and indents the body, breaks after the operator
/// format_args!(
/// text("("),
/// block_indent(&format_args![
/// text("["),
/// block_indent(&format_args![
/// text("1,"),
/// hard_line_break(),
/// text("2,"),
/// hard_line_break(),
/// text("3"),
/// ]),
/// text("]"),
/// hard_line_break(),
/// text("+"),
/// space(),
/// text("aVeryLongIdentifier")
/// ]),
/// text(")")
/// )
/// ).with_mode(BestFittingMode::AllLines)
/// ]
/// )?;
///
/// let document = formatted.into_document();
///
/// // Takes the first variant if everything fits on a single line
/// assert_eq!(
/// "[1, 2, 3] + aVeryLongIdentifier",
/// Formatted::new(document.clone(), SimpleFormatContext::default())
/// .print()?
/// .as_code()
/// );
///
/// // It takes the second if the first variant doesn't fit on a single line. The second variant
/// // has some additional line breaks to make sure inner groups don't break
/// assert_eq!(
/// "[\n\t1,\n\t2,\n\t3\n] + aVeryLongIdentifier",
/// Formatted::new(document.clone(), SimpleFormatContext::new(SimpleFormatOptions { line_width: 23.try_into().unwrap(), ..SimpleFormatOptions::default() }))
/// .print()?
/// .as_code()
/// );
///
/// // Prints the last option as last resort
/// assert_eq!(
/// "(\n\t[\n\t\t1,\n\t\t2,\n\t\t3\n\t]\n\t+ aVeryLongIdentifier\n)",
/// Formatted::new(document.clone(), SimpleFormatContext::new(SimpleFormatOptions { line_width: 22.try_into().unwrap(), ..SimpleFormatOptions::default() }))
/// .print()?
/// .as_code()
/// );
/// # Ok(())
/// # }
/// ```
pub fn with_mode(mut self, mode: BestFittingMode) -> Self {
self.mode = mode;
self
}
}

impl<Context> Format<Context> for FormatBestFitting<'_, Context> {
impl<Context> Format<Context> for BestFitting<'_, Context> {
fn fmt(&self, f: &mut Formatter<Context>) -> FormatResult<()> {
let mut buffer = VecBuffer::new(f.state_mut());
let variants = self.variants.items();
Expand All @@ -2172,9 +2281,12 @@ impl<Context> Format<Context> for FormatBestFitting<'_, Context> {
// SAFETY: The constructor guarantees that there are always at least two variants. It's, therefore,
// safe to call into the unsafe `from_vec_unchecked` function
let element = unsafe {
FormatElement::BestFitting(format_element::BestFitting::from_vec_unchecked(
formatted_variants,
))
FormatElement::BestFitting {
variants: format_element::BestFittingVariants::from_vec_unchecked(
formatted_variants,
),
mode: self.mode,
}
};

f.write_element(element)
Expand Down
99 changes: 71 additions & 28 deletions crates/ruff_formatter/src/format_element.rs
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@ use std::hash::{Hash, Hasher};
use std::ops::Deref;
use std::rc::Rc;

use crate::format_element::tag::{LabelId, Tag};
use crate::format_element::tag::{GroupMode, LabelId, Tag};
use crate::source_code::SourceCodeSlice;
use crate::TagKind;
use ruff_text_size::TextSize;
Expand Down Expand Up @@ -57,7 +57,10 @@ pub enum FormatElement {

/// A list of different variants representing the same content. The printer picks the best fitting content.
/// Line breaks inside of a best fitting don't propagate to parent groups.
BestFitting(BestFitting),
BestFitting {
variants: BestFittingVariants,
mode: BestFittingMode,
},

/// A [Tag] that marks the start/end of some content to which some special formatting is applied.
Tag(Tag),
Expand All @@ -84,9 +87,11 @@ impl std::fmt::Debug for FormatElement {
.field(contains_newlines)
.finish(),
FormatElement::LineSuffixBoundary => write!(fmt, "LineSuffixBoundary"),
FormatElement::BestFitting(best_fitting) => {
fmt.debug_tuple("BestFitting").field(&best_fitting).finish()
}
FormatElement::BestFitting { variants, mode } => fmt
.debug_struct("BestFitting")
.field("variants", variants)
.field("mode", &mode)
.finish(),
FormatElement::Interned(interned) => {
fmt.debug_list().entries(interned.deref()).finish()
}
Expand Down Expand Up @@ -134,6 +139,15 @@ impl PrintMode {
}
}

impl From<GroupMode> for PrintMode {
fn from(value: GroupMode) -> Self {
match value {
GroupMode::Flat => PrintMode::Flat,
GroupMode::Expand | GroupMode::Propagated => PrintMode::Expanded,
}
}
}

#[derive(Clone)]
pub struct Interned(Rc<[FormatElement]>);

Expand Down Expand Up @@ -256,7 +270,10 @@ impl FormatElements for FormatElement {
FormatElement::Interned(interned) => interned.will_break(),
// Traverse into the most flat version because the content is guaranteed to expand when even
// the most flat version contains some content that forces a break.
FormatElement::BestFitting(best_fitting) => best_fitting.most_flat().will_break(),
FormatElement::BestFitting {
variants: best_fitting,
..
} => best_fitting.most_flat().will_break(),
FormatElement::LineSuffixBoundary
| FormatElement::Space
| FormatElement::Tag(_)
Expand Down Expand Up @@ -284,19 +301,36 @@ impl FormatElements for FormatElement {
}
}

/// Provides the printer with different representations for the same element so that the printer
/// can pick the best fitting variant.
///
/// Best fitting is defined as the variant that takes the most horizontal space but fits on the line.
#[derive(Clone, Eq, PartialEq)]
pub struct BestFitting {
/// The different variants for this element.
/// The first element is the one that takes up the most space horizontally (the most flat),
/// The last element takes up the least space horizontally (but most horizontal space).
variants: Box<[Box<[FormatElement]>]>,
/// Mode used to determine if any variant (except the most expanded) fits for [`BestFittingVariants`].
#[repr(u8)]
#[derive(Copy, Clone, Debug, Eq, PartialEq, Default)]
pub enum BestFittingMode {
/// The variant fits if the content up to the first hard or a soft line break inside a [`Group`] with
/// [`PrintMode::Expanded`] fits on the line. The default mode.
///
/// [`Group`]: tag::Group
#[default]
FirstLine,

/// A variant fits if all lines fit into the configured print width. A line ends if by any
/// hard or a soft line break inside a [`Group`] with [`PrintMode::Expanded`].
/// The content doesn't fit if there's any hard line break outside a [`Group`] with [`PrintMode::Expanded`]
/// (a hard line break in content that should be considered in [`PrintMode::Flat`].
///
/// Use this mode with caution as it requires measuring all content of the variant which is more
/// expensive than using [`BestFittingMode::FirstLine`].
///
/// [`Group`]: tag::Group
AllLines,
}

impl BestFitting {
/// The different variants for this element.
/// The first element is the one that takes up the most space horizontally (the most flat),
/// The last element takes up the least space horizontally (but most horizontal space).
#[derive(Clone, Eq, PartialEq, Debug)]
pub struct BestFittingVariants(Box<[Box<[FormatElement]>]>);

impl BestFittingVariants {
/// Creates a new best fitting IR with the given variants. The method itself isn't unsafe
/// but it is to discourage people from using it because the printer will panic if
/// the slice doesn't contain at least the least and most expanded variants.
Expand All @@ -312,33 +346,42 @@ impl BestFitting {
"Requires at least the least expanded and most expanded variants"
);

Self {
variants: variants.into_boxed_slice(),
}
Self(variants.into_boxed_slice())
}

/// Returns the most expanded variant
pub fn most_expanded(&self) -> &[FormatElement] {
self.variants.last().expect(
self.0.last().expect(
"Most contain at least two elements, as guaranteed by the best fitting builder.",
)
}

pub fn variants(&self) -> &[Box<[FormatElement]>] {
&self.variants
pub fn as_slice(&self) -> &[Box<[FormatElement]>] {
&self.0
}

/// Returns the least expanded variant
pub fn most_flat(&self) -> &[FormatElement] {
self.variants.first().expect(
self.0.first().expect(
"Most contain at least two elements, as guaranteed by the best fitting builder.",
)
}
}

impl std::fmt::Debug for BestFitting {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
f.debug_list().entries(&*self.variants).finish()
impl Deref for BestFittingVariants {
type Target = [Box<[FormatElement]>];

fn deref(&self) -> &Self::Target {
self.as_slice()
}
}

impl<'a> IntoIterator for &'a BestFittingVariants {
type Item = &'a Box<[FormatElement]>;
type IntoIter = std::slice::Iter<'a, Box<[FormatElement]>>;

fn into_iter(self) -> Self::IntoIter {
self.as_slice().iter()
}
}

Expand Down Expand Up @@ -397,7 +440,7 @@ mod sizes {
assert_eq_size!(ruff_text_size::TextRange, [u8; 8]);
assert_eq_size!(crate::prelude::tag::VerbatimKind, [u8; 8]);
assert_eq_size!(crate::prelude::Interned, [u8; 16]);
assert_eq_size!(crate::format_element::BestFitting, [u8; 16]);
assert_eq_size!(crate::format_element::BestFittingVariants, [u8; 16]);

#[cfg(not(debug_assertions))]
assert_eq_size!(crate::SourceCodeSlice, [u8; 8]);
Expand Down
18 changes: 14 additions & 4 deletions crates/ruff_formatter/src/format_element/document.rs
Original file line number Diff line number Diff line change
Expand Up @@ -67,10 +67,10 @@ impl Document {
interned_expands
}
},
FormatElement::BestFitting(best_fitting) => {
FormatElement::BestFitting { variants, mode: _ } => {
enclosing.push(Enclosing::BestFitting);

for variant in best_fitting.variants() {
for variant in variants {
propagate_expands(variant, enclosing, checked_interned);
}

Expand Down Expand Up @@ -280,14 +280,14 @@ impl Format<IrFormatContext<'_>> for &[FormatElement] {
write!(f, [text("line_suffix_boundary")])?;
}

FormatElement::BestFitting(best_fitting) => {
FormatElement::BestFitting { variants, mode } => {
write!(f, [text("best_fitting([")])?;
f.write_elements([
FormatElement::Tag(StartIndent),
FormatElement::Line(LineMode::Hard),
])?;

for variant in best_fitting.variants() {
for variant in variants {
write!(f, [variant.deref(), hard_line_break()])?;
}

Expand All @@ -296,6 +296,16 @@ impl Format<IrFormatContext<'_>> for &[FormatElement] {
FormatElement::Line(LineMode::Hard),
])?;

if *mode != BestFittingMode::AllLines {
write!(
f,
[
dynamic_text(&std::format!("mode: {mode:?},"), None),
space()
]
)?;
}

write!(f, [text("])")])?;
}

Expand Down
Loading

0 comments on commit d9e59b2

Please sign in to comment.