##// END OF EJS Templates
rust: fix unsound `OwningDirstateMap`...
rust: fix unsound `OwningDirstateMap` As per the previous patch, `OwningDirstateMap` is unsound. Self-referential structs are difficult to implement correctly in Rust since the compiler is free to move structs around as much as it wants to. They are also very rarely needed in practice, so the state-of-the-art on how they should be done within the Rust rules is still a bit new. The crate `ouroboros` is an attempt at providing a safe way (in the Rust sense) of declaring self-referential structs. It is getting a lot attention and was improved very quickly when soundness issues were found in the past: rather than relying on our own (limited) review circle, we might as well use the de-facto common crate to fix this problem. This will give us a much better chance of finding issues should any new ones be discovered as well as the benefit of fewer `unsafe` APIs of our own. I was starting to think about how I would present a safe API to the old struct but soon realized that the callback-based approach was already done in `ouroboros`, along with a lot more care towards refusing incorrect structs. In short: we don't return a mutable reference to the `DirstateMap` anymore, we expect users of its API to pass a `FnOnce` that takes the map as an argument. This allows our `OwningDirstateMap` to control the input and output lifetimes of the code that modifies it to prevent such issues. Changing to `ouroboros` meant changing every API with it, but it is relatively low churn in the end. It correctly identified the example buggy modification of `copy_map_insert` outlined in the previous patch as violating the borrow rules. Differential Revision: https://phab.mercurial-scm.org/D12429

File last commit:

r46174:b68b1910 default
r49857:c9f44fc9 stable
Show More
patch.rs
369 lines | 11.8 KiB | application/rls-services+xml | RustLexer
use byteorder::{BigEndian, ByteOrder};
/// A chunk of data to insert, delete or replace in a patch
///
/// A chunk is:
/// - an insertion when `!data.is_empty() && start == end`
/// - an deletion when `data.is_empty() && start < end`
/// - a replacement when `!data.is_empty() && start < end`
/// - not doing anything when `data.is_empty() && start == end`
#[derive(Debug, Clone)]
struct Chunk<'a> {
/// The start position of the chunk of data to replace
start: u32,
/// The end position of the chunk of data to replace (open end interval)
end: u32,
/// The data replacing the chunk
data: &'a [u8],
}
impl Chunk<'_> {
/// Adjusted start of the chunk to replace.
///
/// The offset, taking into account the growth/shrinkage of data
/// induced by previously applied chunks.
fn start_offset_by(&self, offset: i32) -> u32 {
let start = self.start as i32 + offset;
assert!(start >= 0, "negative chunk start should never happen");
start as u32
}
/// Adjusted end of the chunk to replace.
///
/// The offset, taking into account the growth/shrinkage of data
/// induced by previously applied chunks.
fn end_offset_by(&self, offset: i32) -> u32 {
self.start_offset_by(offset) + self.data.len() as u32
}
/// Length of the replaced chunk.
fn replaced_len(&self) -> u32 {
self.end - self.start
}
/// Length difference between the replacing data and the replaced data.
fn len_diff(&self) -> i32 {
self.data.len() as i32 - self.replaced_len() as i32
}
}
/// The delta between two revisions data.
#[derive(Debug, Clone)]
pub struct PatchList<'a> {
/// A collection of chunks to apply.
///
/// Those chunks are:
/// - ordered from the left-most replacement to the right-most replacement
/// - non-overlapping, meaning that two chucks can not change the same
/// chunk of the patched data
chunks: Vec<Chunk<'a>>,
}
impl<'a> PatchList<'a> {
/// Create a `PatchList` from bytes.
pub fn new(data: &'a [u8]) -> Self {
let mut chunks = vec![];
let mut data = data;
while !data.is_empty() {
let start = BigEndian::read_u32(&data[0..]);
let end = BigEndian::read_u32(&data[4..]);
let len = BigEndian::read_u32(&data[8..]);
assert!(start <= end);
chunks.push(Chunk {
start,
end,
data: &data[12..12 + (len as usize)],
});
data = &data[12 + (len as usize)..];
}
PatchList { chunks }
}
/// Return the final length of data after patching
/// given its initial length .
fn size(&self, initial_size: i32) -> i32 {
self.chunks
.iter()
.fold(initial_size, |acc, chunk| acc + chunk.len_diff())
}
/// Apply the patch to some data.
pub fn apply(&self, initial: &[u8]) -> Vec<u8> {
let mut last: usize = 0;
let mut vec =
Vec::with_capacity(self.size(initial.len() as i32) as usize);
for Chunk { start, end, data } in self.chunks.iter() {
vec.extend(&initial[last..(*start as usize)]);
vec.extend(data.iter());
last = *end as usize;
}
vec.extend(&initial[last..]);
vec
}
/// Combine two patch lists into a single patch list.
///
/// Applying consecutive patches can lead to waste of time and memory
/// as the changes introduced by one patch can be overridden by the next.
/// Combining patches optimizes the whole patching sequence.
fn combine(&mut self, other: &mut Self) -> Self {
let mut chunks = vec![];
// Keep track of each growth/shrinkage resulting from applying a chunk
// in order to adjust the start/end of subsequent chunks.
let mut offset = 0i32;
// Keep track of the chunk of self.chunks to process.
let mut pos = 0;
// For each chunk of `other`, chunks of `self` are processed
// until they start after the end of the current chunk.
for Chunk { start, end, data } in other.chunks.iter() {
// Add chunks of `self` that start before this chunk of `other`
// without overlap.
while pos < self.chunks.len()
&& self.chunks[pos].end_offset_by(offset) <= *start
{
let first = self.chunks[pos].clone();
offset += first.len_diff();
chunks.push(first);
pos += 1;
}
// The current chunk of `self` starts before this chunk of `other`
// with overlap.
// The left-most part of data is added as an insertion chunk.
// The right-most part data is kept in the chunk.
if pos < self.chunks.len()
&& self.chunks[pos].start_offset_by(offset) < *start
{
let first = &mut self.chunks[pos];
let (data_left, data_right) = first.data.split_at(
(*start - first.start_offset_by(offset)) as usize,
);
let left = Chunk {
start: first.start,
end: first.start,
data: data_left,
};
first.data = data_right;
offset += left.len_diff();
chunks.push(left);
// There is no index incrementation because the right-most part
// needs further examination.
}
// At this point remaining chunks of `self` starts after
// the current chunk of `other`.
// `start_offset` will be used to adjust the start of the current
// chunk of `other`.
// Offset tracking continues with `end_offset` to adjust the end
// of the current chunk of `other`.
let mut next_offset = offset;
// Discard the chunks of `self` that are totally overridden
// by the current chunk of `other`
while pos < self.chunks.len()
&& self.chunks[pos].end_offset_by(next_offset) <= *end
{
let first = &self.chunks[pos];
next_offset += first.len_diff();
pos += 1;
}
// Truncate the left-most part of chunk of `self` that overlaps
// the current chunk of `other`.
if pos < self.chunks.len()
&& self.chunks[pos].start_offset_by(next_offset) < *end
{
let first = &mut self.chunks[pos];
let how_much_to_discard =
*end - first.start_offset_by(next_offset);
first.data = &first.data[(how_much_to_discard as usize)..];
next_offset += how_much_to_discard as i32;
}
// Add the chunk of `other` with adjusted position.
chunks.push(Chunk {
start: (*start as i32 - offset) as u32,
end: (*end as i32 - next_offset) as u32,
data,
});
// Go back to normal offset tracking for the next `o` chunk
offset = next_offset;
}
// Add remaining chunks of `self`.
for elt in &self.chunks[pos..] {
chunks.push(elt.clone());
}
PatchList { chunks }
}
}
/// Combine a list of patch list into a single patch optimized patch list.
pub fn fold_patch_lists<'a>(lists: &[PatchList<'a>]) -> PatchList<'a> {
if lists.len() <= 1 {
if lists.is_empty() {
PatchList { chunks: vec![] }
} else {
lists[0].clone()
}
} else {
let (left, right) = lists.split_at(lists.len() / 2);
let mut left_res = fold_patch_lists(left);
let mut right_res = fold_patch_lists(right);
left_res.combine(&mut right_res)
}
}
#[cfg(test)]
mod tests {
use super::*;
struct PatchDataBuilder {
data: Vec<u8>,
}
impl PatchDataBuilder {
pub fn new() -> Self {
Self { data: vec![] }
}
pub fn replace(
&mut self,
start: usize,
end: usize,
data: &[u8],
) -> &mut Self {
assert!(start <= end);
self.data.extend(&(start as i32).to_be_bytes());
self.data.extend(&(end as i32).to_be_bytes());
self.data.extend(&(data.len() as i32).to_be_bytes());
self.data.extend(data.iter());
self
}
pub fn get(&mut self) -> &[u8] {
&self.data
}
}
#[test]
fn test_ends_before() {
let data = vec![0u8, 0u8, 0u8];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(0, 1, &[1, 2]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(2, 4, &[3, 4]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![1u8, 2, 3, 4]);
}
#[test]
fn test_starts_after() {
let data = vec![0u8, 0u8, 0u8];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(2, 3, &[3]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(1, 2, &[1, 2]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![0u8, 1, 2, 3]);
}
#[test]
fn test_overridden() {
let data = vec![0u8, 0, 0];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(1, 2, &[3, 4]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(1, 4, &[1, 2, 3]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![0u8, 1, 2, 3]);
}
#[test]
fn test_right_most_part_is_overridden() {
let data = vec![0u8, 0, 0];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(0, 1, &[1, 3]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(1, 4, &[2, 3, 4]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![1u8, 2, 3, 4]);
}
#[test]
fn test_left_most_part_is_overridden() {
let data = vec![0u8, 0, 0];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(1, 3, &[1, 3, 4]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(0, 2, &[1, 2]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![1u8, 2, 3, 4]);
}
#[test]
fn test_mid_is_overridden() {
let data = vec![0u8, 0, 0];
let mut patch1_data = PatchDataBuilder::new();
patch1_data.replace(0, 3, &[1, 3, 3, 4]);
let mut patch1 = PatchList::new(patch1_data.get());
let mut patch2_data = PatchDataBuilder::new();
patch2_data.replace(1, 3, &[2, 3]);
let mut patch2 = PatchList::new(patch2_data.get());
let patch = patch1.combine(&mut patch2);
let result = patch.apply(&data);
assert_eq!(result, vec![1u8, 2, 3, 4]);
}
}