Reorganize rust crates into a flat crates directory, simplify some CI steps (#4496)

* Move all rust crates (except lib) into crates dir, w/o nesting

* Remove stale path from .gitattributes

* Rename lib.rs files for easier navigation

* Rename mod.rs file for easier navigation

* Fix emscripten-version path

* Fix fixtures dir paths

* Use the default rustfmt settings

* Don't use nightly on CI
This commit is contained in:
Max Brunsfeld 2025-06-06 14:25:37 -07:00 committed by GitHub
parent a6e530b33d
commit 0fdf569571
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
163 changed files with 69 additions and 89 deletions

View file

@ -0,0 +1,122 @@
use std::{
collections::HashMap,
os::raw::c_void,
sync::{
atomic::{AtomicBool, AtomicUsize, Ordering::SeqCst},
Mutex,
},
};
#[ctor::ctor]
unsafe fn initialize_allocation_recording() {
tree_sitter::set_allocator(
Some(ts_record_malloc),
Some(ts_record_calloc),
Some(ts_record_realloc),
Some(ts_record_free),
);
}
#[derive(Debug, PartialEq, Eq, Hash)]
struct Allocation(*const c_void);
unsafe impl Send for Allocation {}
unsafe impl Sync for Allocation {}
#[derive(Default)]
struct AllocationRecorder {
enabled: AtomicBool,
allocation_count: AtomicUsize,
outstanding_allocations: Mutex<HashMap<Allocation, usize>>,
}
thread_local! {
static RECORDER: AllocationRecorder = AllocationRecorder::default();
}
extern "C" {
fn malloc(size: usize) -> *mut c_void;
fn calloc(count: usize, size: usize) -> *mut c_void;
fn realloc(ptr: *mut c_void, size: usize) -> *mut c_void;
fn free(ptr: *mut c_void);
}
pub fn record<T>(f: impl FnOnce() -> T) -> Result<T, String> {
RECORDER.with(|recorder| {
recorder.enabled.store(true, SeqCst);
recorder.allocation_count.store(0, SeqCst);
recorder.outstanding_allocations.lock().unwrap().clear();
});
let value = f();
let outstanding_allocation_indices = RECORDER.with(|recorder| {
recorder.enabled.store(false, SeqCst);
recorder.allocation_count.store(0, SeqCst);
recorder
.outstanding_allocations
.lock()
.unwrap()
.drain()
.map(|e| e.1)
.collect::<Vec<_>>()
});
if !outstanding_allocation_indices.is_empty() {
return Err(format!(
"Leaked allocation indices: {outstanding_allocation_indices:?}",
));
}
Ok(value)
}
fn record_alloc(ptr: *mut c_void) {
RECORDER.with(|recorder| {
if recorder.enabled.load(SeqCst) {
let count = recorder.allocation_count.fetch_add(1, SeqCst);
recorder
.outstanding_allocations
.lock()
.unwrap()
.insert(Allocation(ptr), count);
}
});
}
fn record_dealloc(ptr: *mut c_void) {
RECORDER.with(|recorder| {
if recorder.enabled.load(SeqCst) {
recorder
.outstanding_allocations
.lock()
.unwrap()
.remove(&Allocation(ptr));
}
});
}
unsafe extern "C" fn ts_record_malloc(size: usize) -> *mut c_void {
let result = malloc(size);
record_alloc(result);
result
}
unsafe extern "C" fn ts_record_calloc(count: usize, size: usize) -> *mut c_void {
let result = calloc(count, size);
record_alloc(result);
result
}
unsafe extern "C" fn ts_record_realloc(ptr: *mut c_void, size: usize) -> *mut c_void {
let result = realloc(ptr, size);
if ptr.is_null() {
record_alloc(result);
} else if !core::ptr::eq(ptr, result) {
record_dealloc(ptr);
record_alloc(result);
}
result
}
unsafe extern "C" fn ts_record_free(ptr: *mut c_void) {
record_dealloc(ptr);
free(ptr);
}

View file

@ -0,0 +1,147 @@
use tree_sitter::{LogType, Node, Parser, Point, Range, Tree};
use super::{scope_sequence::ScopeSequence, LOG_ENABLED, LOG_GRAPH_ENABLED};
use crate::util;
pub fn check_consistent_sizes(tree: &Tree, input: &[u8]) {
fn check(node: Node, line_offsets: &[usize]) {
let start_byte = node.start_byte();
let end_byte = node.end_byte();
let start_point = node.start_position();
let end_point = node.end_position();
assert!(start_byte <= end_byte);
assert!(start_point <= end_point);
assert_eq!(
start_byte,
line_offsets[start_point.row] + start_point.column
);
assert_eq!(end_byte, line_offsets[end_point.row] + end_point.column);
let mut last_child_end_byte = start_byte;
let mut last_child_end_point = start_point;
let mut some_child_has_changes = false;
let mut actual_named_child_count = 0;
for i in 0..node.child_count() {
let child = node.child(i).unwrap();
assert!(child.start_byte() >= last_child_end_byte);
assert!(child.start_position() >= last_child_end_point);
check(child, line_offsets);
if child.has_changes() {
some_child_has_changes = true;
}
if child.is_named() {
actual_named_child_count += 1;
}
last_child_end_byte = child.end_byte();
last_child_end_point = child.end_position();
}
assert_eq!(actual_named_child_count, node.named_child_count());
if node.child_count() > 0 {
assert!(end_byte >= last_child_end_byte);
assert!(end_point >= last_child_end_point);
}
if some_child_has_changes {
assert!(node.has_changes());
}
}
let mut line_offsets = vec![0];
for (i, c) in input.iter().enumerate() {
if *c == b'\n' {
line_offsets.push(i + 1);
}
}
check(tree.root_node(), &line_offsets);
}
pub fn check_changed_ranges(old_tree: &Tree, new_tree: &Tree, input: &[u8]) -> Result<(), String> {
let changed_ranges = old_tree.changed_ranges(new_tree).collect::<Vec<_>>();
let old_scope_sequence = ScopeSequence::new(old_tree);
let new_scope_sequence = ScopeSequence::new(new_tree);
let old_range = old_tree.root_node().range();
let new_range = new_tree.root_node().range();
let byte_range =
old_range.start_byte.min(new_range.start_byte)..old_range.end_byte.max(new_range.end_byte);
let point_range = old_range.start_point.min(new_range.start_point)
..old_range.end_point.max(new_range.end_point);
for range in &changed_ranges {
if range.end_byte > byte_range.end || range.end_point > point_range.end {
return Err(format!(
"changed range extends outside of the old and new trees {range:?}",
));
}
}
old_scope_sequence.check_changes(&new_scope_sequence, input, &changed_ranges)
}
pub fn set_included_ranges(parser: &mut Parser, input: &[u8], delimiters: Option<(&str, &str)>) {
if let Some((start, end)) = delimiters {
let mut ranges = Vec::new();
let mut ix = 0;
while ix < input.len() {
let Some(mut start_ix) = input[ix..]
.windows(2)
.position(|win| win == start.as_bytes())
else {
break;
};
start_ix += ix + start.len();
let end_ix = input[start_ix..]
.windows(2)
.position(|win| win == end.as_bytes())
.map_or(input.len(), |ix| start_ix + ix);
ix = end_ix;
ranges.push(Range {
start_byte: start_ix,
end_byte: end_ix,
start_point: point_for_offset(input, start_ix),
end_point: point_for_offset(input, end_ix),
});
}
parser.set_included_ranges(&ranges).unwrap();
} else {
parser.set_included_ranges(&[]).unwrap();
}
}
fn point_for_offset(text: &[u8], offset: usize) -> Point {
let mut point = Point::default();
for byte in &text[..offset] {
if *byte == b'\n' {
point.row += 1;
point.column = 0;
} else {
point.column += 1;
}
}
point
}
pub fn get_parser(session: &mut Option<util::LogSession>, log_filename: &str) -> Parser {
let mut parser = Parser::new();
if *LOG_ENABLED {
parser.set_logger(Some(Box::new(|log_type, msg| {
if log_type == LogType::Lex {
eprintln!(" {msg}");
} else {
eprintln!("{msg}");
}
})));
}
if *LOG_GRAPH_ENABLED {
*session = Some(util::log_graphs(&mut parser, log_filename, false).unwrap());
}
parser
}

View file

@ -0,0 +1,61 @@
use super::random::Rand;
#[derive(Debug)]
pub struct Edit {
pub position: usize,
pub deleted_length: usize,
pub inserted_text: Vec<u8>,
}
#[must_use]
pub fn invert_edit(input: &[u8], edit: &Edit) -> Edit {
let position = edit.position;
let removed_content = &input[position..(position + edit.deleted_length)];
Edit {
position,
deleted_length: edit.inserted_text.len(),
inserted_text: removed_content.to_vec(),
}
}
pub fn get_random_edit(rand: &mut Rand, input: &[u8]) -> Edit {
let choice = rand.unsigned(10);
if choice < 2 {
// Insert text at end
let inserted_text = rand.words(3);
Edit {
position: input.len(),
deleted_length: 0,
inserted_text,
}
} else if choice < 5 {
// Delete text from the end
let deleted_length = rand.unsigned(30).min(input.len());
Edit {
position: input.len() - deleted_length,
deleted_length,
inserted_text: vec![],
}
} else if choice < 8 {
// Insert at a random position
let position = rand.unsigned(input.len());
let word_count = 1 + rand.unsigned(3);
let inserted_text = rand.words(word_count);
Edit {
position,
deleted_length: 0,
inserted_text,
}
} else {
// Replace at random position
let position = rand.unsigned(input.len());
let deleted_length = rand.unsigned(input.len() - position);
let word_count = 1 + rand.unsigned(3);
let inserted_text = rand.words(word_count);
Edit {
position,
deleted_length,
inserted_text,
}
}
}

View file

@ -0,0 +1,44 @@
use rand::{
distributions::Alphanumeric,
prelude::{Rng, SeedableRng, StdRng},
};
const OPERATORS: &[char] = &[
'+', '-', '<', '>', '(', ')', '*', '/', '&', '|', '!', ',', '.', '%',
];
pub struct Rand(StdRng);
impl Rand {
#[must_use]
pub fn new(seed: usize) -> Self {
Self(StdRng::seed_from_u64(seed as u64))
}
pub fn unsigned(&mut self, max: usize) -> usize {
self.0.gen_range(0..=max)
}
pub fn words(&mut self, max_count: usize) -> Vec<u8> {
let mut result = Vec::new();
let word_count = self.unsigned(max_count);
for i in 0..word_count {
if i > 0 {
if self.unsigned(5) == 0 {
result.push(b'\n');
} else {
result.push(b' ');
}
}
if self.unsigned(3) == 0 {
let index = self.unsigned(OPERATORS.len() - 1);
result.push(OPERATORS[index] as u8);
} else {
for _ in 0..self.unsigned(8) {
result.push(self.0.sample(Alphanumeric));
}
}
}
result
}
}

View file

@ -0,0 +1,91 @@
use tree_sitter::{Point, Range, Tree};
#[derive(Debug)]
pub struct ScopeSequence(Vec<ScopeStack>);
type ScopeStack = Vec<&'static str>;
impl ScopeSequence {
#[must_use]
pub fn new(tree: &Tree) -> Self {
let mut result = Self(Vec::new());
let mut scope_stack = Vec::new();
let mut cursor = tree.walk();
let mut visited_children = false;
loop {
let node = cursor.node();
for _ in result.0.len()..node.start_byte() {
result.0.push(scope_stack.clone());
}
if visited_children {
for _ in result.0.len()..node.end_byte() {
result.0.push(scope_stack.clone());
}
scope_stack.pop();
if cursor.goto_next_sibling() {
visited_children = false;
} else if !cursor.goto_parent() {
break;
}
} else {
scope_stack.push(cursor.node().kind());
if !cursor.goto_first_child() {
visited_children = true;
}
}
}
result
}
pub fn check_changes(
&self,
other: &Self,
text: &[u8],
known_changed_ranges: &[Range],
) -> Result<(), String> {
let mut position = Point { row: 0, column: 0 };
for i in 0..(self.0.len().max(other.0.len())) {
let stack = &self.0.get(i);
let other_stack = &other.0.get(i);
if *stack != *other_stack && ![b'\r', b'\n'].contains(&text[i]) {
let containing_range = known_changed_ranges
.iter()
.find(|range| range.start_point <= position && position < range.end_point);
if containing_range.is_none() {
let line = &text[(i - position.column)..]
.split(|c| *c == b'\n')
.next()
.unwrap();
return Err(format!(
concat!(
"Position: {}\n",
"Byte offset: {}\n",
"Line: {}\n",
"{}^\n",
"Old scopes: {:?}\n",
"New scopes: {:?}\n",
"Invalidated ranges: {:?}",
),
position,
i,
String::from_utf8_lossy(line),
String::from(" ").repeat(position.column + "Line: ".len()),
stack,
other_stack,
known_changed_ranges,
));
}
}
if text[i] == b'\n' {
position.row += 1;
position.column = 0;
} else {
position.column += 1;
}
}
Ok(())
}
}