Merge pull request #2182 from zed-industries/faster-injections

Fix syntax-related performance problems on gigantic files
This commit is contained in:
Max Brunsfeld 2023-02-16 12:52:13 -08:00 committed by GitHub
commit 98b51634c4
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 51 additions and 14 deletions

2
Cargo.lock generated
View file

@ -7004,7 +7004,7 @@ dependencies = [
[[package]] [[package]]
name = "tree-sitter" name = "tree-sitter"
version = "0.20.9" version = "0.20.9"
source = "git+https://github.com/tree-sitter/tree-sitter?rev=36b5b6c89e55ad1a502f8b3234bb3e12ec83a5da#36b5b6c89e55ad1a502f8b3234bb3e12ec83a5da" source = "git+https://github.com/tree-sitter/tree-sitter?rev=c51896d32dcc11a38e41f36e3deb1a6a9c4f4b14#c51896d32dcc11a38e41f36e3deb1a6a9c4f4b14"
dependencies = [ dependencies = [
"cc", "cc",
"regex", "regex",

View file

@ -69,7 +69,7 @@ serde_json = { version = "1.0", features = ["preserve_order", "raw_value"] }
rand = { version = "0.8" } rand = { version = "0.8" }
[patch.crates-io] [patch.crates-io]
tree-sitter = { git = "https://github.com/tree-sitter/tree-sitter", rev = "36b5b6c89e55ad1a502f8b3234bb3e12ec83a5da" } tree-sitter = { git = "https://github.com/tree-sitter/tree-sitter", rev = "c51896d32dcc11a38e41f36e3deb1a6a9c4f4b14" }
async-task = { git = "https://github.com/zed-industries/async-task", rev = "341b57d6de98cdfd7b418567b8de2022ca993a6e" } async-task = { git = "https://github.com/zed-industries/async-task", rev = "341b57d6de98cdfd7b418567b8de2022ca993a6e" }
# TODO - Remove when a version is released with this PR: https://github.com/servo/core-foundation-rs/pull/457 # TODO - Remove when a version is released with this PR: https://github.com/servo/core-foundation-rs/pull/457

View file

@ -2247,7 +2247,6 @@ impl BufferSnapshot {
.map(|g| g.outline_config.as_ref().unwrap()) .map(|g| g.outline_config.as_ref().unwrap())
.collect::<Vec<_>>(); .collect::<Vec<_>>();
let mut chunks = self.chunks(0..self.len(), true);
let mut stack = Vec::<Range<usize>>::new(); let mut stack = Vec::<Range<usize>>::new();
let mut items = Vec::new(); let mut items = Vec::new();
while let Some(mat) = matches.peek() { while let Some(mat) = matches.peek() {
@ -2266,9 +2265,7 @@ impl BufferSnapshot {
continue; continue;
} }
let mut text = String::new(); let mut buffer_ranges = Vec::new();
let mut name_ranges = Vec::new();
let mut highlight_ranges = Vec::new();
for capture in mat.captures { for capture in mat.captures {
let node_is_name; let node_is_name;
if capture.index == config.name_capture_ix { if capture.index == config.name_capture_ix {
@ -2286,12 +2283,27 @@ impl BufferSnapshot {
range.start + self.line_len(start.row as u32) as usize - start.column; range.start + self.line_len(start.row as u32) as usize - start.column;
} }
buffer_ranges.push((range, node_is_name));
}
if buffer_ranges.is_empty() {
continue;
}
let mut text = String::new();
let mut highlight_ranges = Vec::new();
let mut name_ranges = Vec::new();
let mut chunks = self.chunks(
buffer_ranges.first().unwrap().0.start..buffer_ranges.last().unwrap().0.end,
true,
);
for (buffer_range, is_name) in buffer_ranges {
if !text.is_empty() { if !text.is_empty() {
text.push(' '); text.push(' ');
} }
if node_is_name { if is_name {
let mut start = text.len(); let mut start = text.len();
let end = start + range.len(); let end = start + buffer_range.len();
// When multiple names are captured, then the matcheable text // When multiple names are captured, then the matcheable text
// includes the whitespace in between the names. // includes the whitespace in between the names.
@ -2302,12 +2314,12 @@ impl BufferSnapshot {
name_ranges.push(start..end); name_ranges.push(start..end);
} }
let mut offset = range.start; let mut offset = buffer_range.start;
chunks.seek(offset); chunks.seek(offset);
for mut chunk in chunks.by_ref() { for mut chunk in chunks.by_ref() {
if chunk.text.len() > range.end - offset { if chunk.text.len() > buffer_range.end - offset {
chunk.text = &chunk.text[0..(range.end - offset)]; chunk.text = &chunk.text[0..(buffer_range.end - offset)];
offset = range.end; offset = buffer_range.end;
} else { } else {
offset += chunk.text.len(); offset += chunk.text.len();
} }
@ -2321,7 +2333,7 @@ impl BufferSnapshot {
highlight_ranges.push((start..end, style)); highlight_ranges.push((start..end, style));
} }
text.push_str(chunk.text); text.push_str(chunk.text);
if offset >= range.end { if offset >= buffer_range.end {
break; break;
} }
} }

View file

@ -608,6 +608,31 @@ impl SyntaxSnapshot {
self.layers = layers; self.layers = layers;
self.interpolated_version = text.version.clone(); self.interpolated_version = text.version.clone();
self.parsed_version = text.version.clone(); self.parsed_version = text.version.clone();
#[cfg(debug_assertions)]
self.check_invariants(text);
}
#[cfg(debug_assertions)]
fn check_invariants(&self, text: &BufferSnapshot) {
let mut max_depth = 0;
let mut prev_range: Option<Range<Anchor>> = None;
for layer in self.layers.iter() {
if layer.depth == max_depth {
if let Some(prev_range) = prev_range {
match layer.range.start.cmp(&prev_range.start, text) {
Ordering::Less => panic!("layers out of order"),
Ordering::Equal => {
assert!(layer.range.end.cmp(&prev_range.end, text).is_ge())
}
Ordering::Greater => {}
}
}
} else if layer.depth < max_depth {
panic!("layers out of order")
}
max_depth = layer.depth;
prev_range = Some(layer.range.clone());
}
} }
pub fn single_tree_captures<'a>( pub fn single_tree_captures<'a>(
@ -1419,7 +1444,7 @@ impl sum_tree::Summary for SyntaxLayerSummary {
self.max_depth = other.max_depth; self.max_depth = other.max_depth;
self.range = other.range.clone(); self.range = other.range.clone();
} else { } else {
if other.range.start.cmp(&self.range.start, buffer).is_lt() { if self.range == (Anchor::MAX..Anchor::MAX) {
self.range.start = other.range.start; self.range.start = other.range.start;
} }
if other.range.end.cmp(&self.range.end, buffer).is_gt() { if other.range.end.cmp(&self.range.end, buffer).is_gt() {