From b275f37db6697cf296dc328b234c3d4b7b485986 Mon Sep 17 00:00:00 2001 From: Chayim Refael Friedman Date: Mon, 23 Sep 2024 01:54:40 +0300 Subject: [PATCH] Fix a bug in span map merge, and add explanations of how span maps are stored Because it took me hours to figure out that contrary to common sense, the offset stored is the *end* of the node, and we search by the *start*. Which is why we need a convoluted `partition_point()` instead of a simple `binary_search()`. And this was not documented at all. Which made me make mistakes with my implementation of `SpanMap::merge()`. The other bug fixed about span map merging is correctly keeping track of the current offset in presence of multiple sibling macro invocations. Unrelated, but because of the previous issue it took me hours to debug, so I figured out I'll put them together for posterity. --- crates/ide/src/expand_macro.rs | 12 +++++++++--- crates/span/src/map.rs | 26 ++++++++++++++++++++++++-- 2 files changed, 33 insertions(+), 5 deletions(-) diff --git a/crates/ide/src/expand_macro.rs b/crates/ide/src/expand_macro.rs index f55082de67..79fdf75b7f 100644 --- a/crates/ide/src/expand_macro.rs +++ b/crates/ide/src/expand_macro.rs @@ -149,14 +149,14 @@ fn expand_macro_recur( expanded.text_range().len(), &expansion_span_map, ); - Some(expand(sema, expanded, result_span_map, offset_in_original_node)) + Some(expand(sema, expanded, result_span_map, u32::from(offset_in_original_node) as i32)) } fn expand( sema: &Semantics<'_, RootDatabase>, expanded: SyntaxNode, result_span_map: &mut SpanMap, - offset_in_original_node: TextSize, + mut offset_in_original_node: i32, ) -> SyntaxNode { let children = expanded.descendants().filter_map(ast::Item::cast); let mut replacements = Vec::new(); @@ -166,8 +166,14 @@ fn expand( sema, &child, result_span_map, - offset_in_original_node + child.syntax().text_range().start(), + TextSize::new( + (offset_in_original_node + (u32::from(child.syntax().text_range().start()) as i32)) + as u32, + ), ) { + offset_in_original_node = offset_in_original_node + + (u32::from(new_node.text_range().len()) as i32) + - (u32::from(child.syntax().text_range().len()) as i32); // check if the whole original syntax is replaced if expanded == *child.syntax() { return new_node; diff --git a/crates/span/src/map.rs b/crates/span/src/map.rs index 2680e5036c..f80de05ec6 100644 --- a/crates/span/src/map.rs +++ b/crates/span/src/map.rs @@ -13,6 +13,7 @@ use crate::{ /// Maps absolute text ranges for the corresponding file to the relevant span data. #[derive(Debug, PartialEq, Eq, Clone, Hash)] pub struct SpanMap { + /// The offset stored here is the *end* of the node. spans: Vec<(TextSize, SpanData)>, /// Index of the matched macro arm on successful expansion for declarative macros. // FIXME: Does it make sense to have this here? @@ -109,11 +110,32 @@ where /// /// The length of the replacement node needs to be `other_size`. pub fn merge(&mut self, other_range: TextRange, other_size: TextSize, other: &SpanMap) { + // I find the following diagram helpful to illustrate the bounds and why we use `<` or `<=`: + // -------------------------------------------------------------------- + // 1 3 5 6 7 10 11 <-- offsets we store + // 0-1 1-3 3-5 5-6 6-7 7-10 10-11 <-- ranges these offsets refer to + // 3 .. 7 <-- other_range + // 3-5 5-6 6-7 <-- ranges we replace (len = 7-3 = 4) + // ^^^^^^^^^^^ ^^^^^^^^^^ + // remove shift + // 2 3 5 9 <-- offsets we insert + // 0-2 2-3 3-5 5-9 <-- ranges we insert (other_size = 9-0 = 9) + // ------------------------------------ + // 1 3 + // 0-1 1-3 <-- these remain intact + // 5 6 8 12 + // 3-5 5-6 6-8 8-12 <-- we shift these by other_range.start() and insert them + // 15 16 + // 12-15 15-16 <-- we shift these by other_size-other_range.len() = 9-4 = 5 + // ------------------------------------ + // 1 3 5 6 8 12 15 16 <-- final offsets we store + // 0-1 1-3 3-5 5-6 6-8 8-12 12-15 15-16 <-- final ranges + self.spans.retain_mut(|(offset, _)| { - if other_range.contains(*offset) { + if other_range.start() < *offset && *offset <= other_range.end() { false } else { - if *offset >= other_range.end() { + if *offset > other_range.end() { *offset += other_size; *offset -= other_range.len(); }