feat: add format range functionality for partial code formatting by magic-akari · Pull Request #48 · g-plane/malva

magic-akari · 2026-01-02T15:42:42Z

Implement the ability to format specific ranges of CSS code instead of entire files, enabling editor integrations where users can select and format only portions of code.

Copilot

Pull request overview

This PR implements range formatting functionality for Malva, enabling users to format specific portions of CSS/SCSS/Sass/Less code rather than entire files. This is particularly valuable for editor integrations where users select and format only a portion of their code.

Key changes:

Implements format_range function that intelligently expands selections to complete syntactic units (statements or declarations)
Calculates and preserves base indentation levels for formatted ranges
Adds comprehensive test suite with snapshot testing across multiple CSS dialects

Reviewed changes

Copilot reviewed 19 out of 19 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
malva/src/range.rs	Core implementation of range formatting logic with node detection and indent calculation
malva/src/lib.rs	Exports public API for `format_range` function and `FormatRangeResult` type
malva/src/error.rs	Adds `RangeOutOfBounds` error variant for range validation
malva/tests/fmt_range.rs	Comprehensive test suite including unit tests and snapshot tests
malva/tests/fmt_range/*/.{css,scss,less}	Test fixtures for range formatting
malva/tests/fmt_range/*/.range	Range specification files for test cases
malva/tests/fmt_range/*/.snap	Expected output snapshots for test cases

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-01-02T15:46:57Z

+    // Wrap with base indent
+    let doc = doc.nest(base_indent);
+
+    // Print => doc


The comment "Print => doc" appears to be backwards or confusing. It should likely say "Print doc" or "Print the doc" since this is converting the doc to a string, not converting print to doc.

Suggested change

// Print => doc

// Print doc

Copilot · 2026-01-02T15:46:58Z

+
+        let output = format_range(&input, start..end, syntax, &options).unwrap();
+        let output = {
+            let mut result = input.clone();


The variable input is a string slice, so calling .clone() on line 119 is unnecessary and potentially misleading. Since input is an &str, you should use .to_string() or just construct the result directly without the intermediate clone.

Suggested change

let mut result = input.clone();

let mut result = input.to_string();

Copilot · 2026-01-02T15:46:58Z

+fn find_range_node<'a, 's>(
+    stylesheet: &'a Stylesheet<'s>,
+    range: &Range<usize>,
+    _source: &str,


The parameter _source is prefixed with an underscore indicating it's unused, but it's included in the function signature. If this parameter is genuinely not needed, it should be removed to avoid confusion. If it's kept for future use or consistency, a comment explaining why would be helpful.

Suggested change

_source: &str,

_source: &str, // kept for API compatibility / potential future use; intentionally unused

Copilot · 2026-01-02T15:46:58Z

+
+}
+@function grid-width($n) {
+    @return $very-very-very-very-very-very-vey-long-var * $very-very-very-very-very-very-vey-long-var + ($very-very-very-very-very-very-vey-long-var - 1) * $very-very-very-very-very-very-vey-long-var;


The word "vey" appears to be a typo and should be "very". This typo is repeated multiple times in this line.

Copilot · 2026-01-02T15:46:59Z

+
+}
+@function grid-width($n) {
+    @return $very-very-very-very-very-very-vey-long-var * $very-very-very-very-very-very-vey-long-var + ($very-very-very-very-very-very-vey-long-var - 1) * $very-very-very-very-very-very-vey-long-var;


The word "vey" appears to be a typo and should be "very". This typo is repeated multiple times in this line.

Copilot · 2026-01-02T15:46:59Z

+
+/// Represents a node or a list of sibling nodes that should be formatted together.
+enum RangeNode<'a, 's> {
+    /// No formatable node found in the range.


The word "formatable" should be spelled "formattable" with double 't'.

Suggested change

/// No formatable node found in the range.

/// No formattable node found in the range.

Copilot · 2026-01-02T15:46:59Z

+/// - A single node that completely contains the range, or
+/// - Multiple sibling nodes that together contain the range
+///
+/// For CSS, the minimum formatable unit is a "line-level" node:


The word "formatable" should be spelled "formattable" with double 't'.

Suggested change

/// For CSS, the minimum formatable unit is a "line-level" node:

/// For CSS, the minimum formattable unit is a "line-level" node:

g-plane

Declaration is also one kind of statement, so it can be merged in RangeNode.

Actually statements can be nested in any level depth, but it seems this PR only processes two level.

To be honest, due to the design of Raffia, range formatting is hard to to be implemented, and I don't consider this feature until it's highly requested. I don't think it's a rare use case but also I don't think developers can't live without this.

g-plane · 2026-01-08T02:00:04Z

+    options: &FormatOptions,
+) -> Result<FormatRangeResult, Error> {
+    // 1. Validate range
+    if range.start > source.len() || range.end > source.len() {


Suggested change

if range.start > source.len() || range.end > source.len() {

if !range.contains(source.len()) {

g-plane · 2026-01-08T02:01:25Z

+    /// The specified range is outside of the source file bounds.
+    RangeOutOfBounds {
+        range: std::ops::Range<usize>,
+        source_len: usize,


Suggested change

source_len: usize,

total: usize,

g-plane · 2026-01-08T02:04:12Z

+}
+
+impl RangeNode<'_, '_> {
+    /// Get the span of this range node.


Suggested change

/// Get the span of this range node.

g-plane · 2026-01-08T02:25:31Z

+    let statements = &stylesheet.statements;
+
+    // Check if the range covers the entire file
+    if range.start == 0 && range.end >= stylesheet.span().end {


Suggested change

if range.start == 0 && range.end >= stylesheet.span().end {

if range.start == 0 && range.end >= stylesheet.span.end {

g-plane · 2026-01-08T02:25:49Z

+    // Check if the range covers the entire file
+    if range.start == 0 && range.end >= stylesheet.span().end {
+        if let Some(first) = statements.first() {
+            return (RangeNode::MultipleStatements(statements.iter().collect()), first.span().clone());


Suggested change

return (RangeNode::MultipleStatements(statements.iter().collect()), first.span().clone());

return (RangeNode::MultipleStatements(statements.clone()), first.span().clone());

g-plane · 2026-01-08T02:43:33Z

+    let (range_node, ref_span) = find_range_node(&stylesheet, &range, source);
+
+    // 4. Calculate base indentation from reference span
+    let base_indent = calculate_base_indent(ref_span, source, options);


We don't need to do such complicated calculation. A simple solution: pick span of first statement, then get its column by using line_bounds.get_line_col. The column value will be the base indent.

g-plane · 2026-01-08T02:56:43Z

+            let mut docs = Vec::with_capacity(stmts.len() * 2);
+            for (i, stmt) in stmts.iter().enumerate() {
+                if i > 0 {
+                    docs.push(Doc::hard_line());


What about comments and consecutive line breaks?

magic-akari · 2026-01-09T02:01:33Z

To be honest, due to the design of Raffia, range formatting is hard to to be implemented, and I don't consider this feature until it's highly requested. I don't think it's a rare use case but also I don't think developers can't live without this.

I see your point. When implementing this, I looked at how ruff and biome handle it.

The main challenge, I believe, is tracking positions in the output, largely due to the design of tiny_pretty. Since CSS nesting levels are usually shallow, I tried implementing it by leveraging the text positions of input nodes along with some assumptions.

I'd like to hear your thoughts on this: would you prefer seeing this feature integrated into the lower-level infrastructure, or would you rather wait for more user demand before investing more effort into it? If you'd rather not pursue this further at the moment, I'm happy to put the PR on hold.

g-plane · 2026-01-09T08:32:52Z

largely due to the design of tiny_pretty

This shouldn't be related to tiny_pretty. It should be Raffia, the parser.

or would you rather wait for more user demand

I prefer waiting.

feat: add format range functionality for partial code formatting

1697949

Implement the ability to format specific ranges of CSS code instead of entire files, enabling editor integrations where users can select and format only portions of code.

magic-akari marked this pull request as ready for review January 2, 2026 15:44

Copilot AI review requested due to automatic review settings January 2, 2026 15:44

Copilot started reviewing on behalf of magic-akari January 2, 2026 15:44 View session

Copilot AI reviewed Jan 2, 2026

View reviewed changes

g-plane reviewed Jan 8, 2026

View reviewed changes

	let mut result = input.clone();
	let mut result = input.to_string();

	_source: &str,
	_source: &str, // kept for API compatibility / potential future use; intentionally unused

	/// No formatable node found in the range.
	/// No formattable node found in the range.

	/// For CSS, the minimum formatable unit is a "line-level" node:
	/// For CSS, the minimum formattable unit is a "line-level" node:

	if range.start > source.len() \|\| range.end > source.len() {
	if !range.contains(source.len()) {

	if range.start == 0 && range.end >= stylesheet.span().end {
	if range.start == 0 && range.end >= stylesheet.span.end {

	return (RangeNode::MultipleStatements(statements.iter().collect()), first.span().clone());
	return (RangeNode::MultipleStatements(statements.clone()), first.span().clone());

Uh oh!

Conversation

magic-akari commented Jan 2, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Jan 2, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

g-plane Jan 8, 2026

Choose a reason for hiding this comment

Uh oh!

magic-akari commented Jan 9, 2026

Uh oh!

g-plane commented Jan 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

g-plane left a comment •

edited

Loading