zed/crates/buffer_diff/src/buffer_diff.rs

use futures::channel::oneshot;
use git2::{DiffLineType as GitDiffLineType, DiffOptions as GitOptions, Patch as GitPatch};
use gpui::{App, AppContext as _, AsyncApp, Context, Entity, EventEmitter, Task, TaskLabel};
use language::{
    BufferRow, DiffOptions, File, Language, LanguageName, LanguageRegistry,
    language_settings::language_settings, word_diff_ranges,
};
use rope::Rope;
use std::{
    cmp::Ordering,
    future::Future,
    iter,
    ops::Range,
    sync::{Arc, LazyLock},
};
use sum_tree::SumTree;
use text::{Anchor, Bias, BufferId, OffsetRangeExt, Point, ToOffset as _, ToPoint as _};
use util::ResultExt;

pub static CALCULATE_DIFF_TASK: LazyLock<TaskLabel> = LazyLock::new(TaskLabel::new);
pub const MAX_WORD_DIFF_LINE_COUNT: usize = 5;

pub struct BufferDiff {
    pub buffer_id: BufferId,
    inner: BufferDiffInner,
    // diff of the index vs head
    secondary_diff: Option<Entity<BufferDiff>>,
}

#[derive(Clone, Debug)]
pub struct BufferDiffSnapshot {
    inner: BufferDiffInner,
    secondary_diff: Option<Box<BufferDiffSnapshot>>,
}

#[derive(Clone)]
struct BufferDiffInner {
    hunks: SumTree<InternalDiffHunk>,
    // Used for making staging mo
    pending_hunks: SumTree<PendingHunk>,
    base_text: language::BufferSnapshot,
    base_text_exists: bool,
}

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub struct DiffHunkStatus {
    pub kind: DiffHunkStatusKind,
    pub secondary: DiffHunkSecondaryStatus,
}

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
pub enum DiffHunkStatusKind {
    Added,
    Modified,
    Deleted,
}

#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
/// Diff of Working Copy vs Index
/// aka 'is this hunk staged or not'
pub enum DiffHunkSecondaryStatus {
    /// Unstaged
    HasSecondaryHunk,
    /// Partially staged
    OverlapsWithSecondaryHunk,
    /// Staged
    NoSecondaryHunk,
    /// We are unstaging
    SecondaryHunkAdditionPending,
    /// We are stagind
    SecondaryHunkRemovalPending,
}

/// A diff hunk resolved to rows in the buffer.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct DiffHunk {
    /// The buffer range as points.
    pub range: Range<Point>,
    /// The range in the buffer to which this hunk corresponds.
    pub buffer_range: Range<Anchor>,
    /// The range in the buffer's diff base text to which this hunk corresponds.
    pub diff_base_byte_range: Range<usize>,
    pub secondary_status: DiffHunkSecondaryStatus,
    // Anchors representing the word diff locations in the active buffer
    pub buffer_word_diffs: Vec<Range<Anchor>>,
    // Offsets relative to the start of the deleted diff that represent word diff locations
    pub base_word_diffs: Vec<Range<usize>>,
}

/// We store [`InternalDiffHunk`]s internally so we don't need to store the additional row range.
#[derive(Debug, Clone, PartialEq, Eq)]
struct InternalDiffHunk {
    buffer_range: Range<Anchor>,
    diff_base_byte_range: Range<usize>,
    base_word_diffs: Vec<Range<usize>>,
    buffer_word_diffs: Vec<Range<Anchor>>,
}

#[derive(Debug, Clone, PartialEq, Eq)]
struct PendingHunk {
    buffer_range: Range<Anchor>,
    diff_base_byte_range: Range<usize>,
    buffer_version: clock::Global,
    new_status: DiffHunkSecondaryStatus,
}

#[derive(Debug, Clone)]
pub struct DiffHunkSummary {
    buffer_range: Range<Anchor>,
    diff_base_byte_range: Range<usize>,
}

impl sum_tree::Item for InternalDiffHunk {
    type Summary = DiffHunkSummary;

    fn summary(&self, _cx: &text::BufferSnapshot) -> Self::Summary {
        DiffHunkSummary {
            buffer_range: self.buffer_range.clone(),
            diff_base_byte_range: self.diff_base_byte_range.clone(),
        }
    }
}

impl sum_tree::Item for PendingHunk {
    type Summary = DiffHunkSummary;

    fn summary(&self, _cx: &text::BufferSnapshot) -> Self::Summary {
        DiffHunkSummary {
            buffer_range: self.buffer_range.clone(),
            diff_base_byte_range: self.diff_base_byte_range.clone(),
        }
    }
}

impl sum_tree::Summary for DiffHunkSummary {
    type Context<'a> = &'a text::BufferSnapshot;

    fn zero(_cx: Self::Context<'_>) -> Self {
        DiffHunkSummary {
            buffer_range: Anchor::MIN..Anchor::MIN,
            diff_base_byte_range: 0..0,
        }
    }

    fn add_summary(&mut self, other: &Self, buffer: Self::Context<'_>) {
        self.buffer_range.start = *self
            .buffer_range
            .start
            .min(&other.buffer_range.start, buffer);
        self.buffer_range.end = *self.buffer_range.end.max(&other.buffer_range.end, buffer);

        self.diff_base_byte_range.start = self
            .diff_base_byte_range
            .start
            .min(other.diff_base_byte_range.start);
        self.diff_base_byte_range.end = self
            .diff_base_byte_range
            .end
            .max(other.diff_base_byte_range.end);
    }
}

impl sum_tree::SeekTarget<'_, DiffHunkSummary, DiffHunkSummary> for Anchor {
    fn cmp(&self, cursor_location: &DiffHunkSummary, buffer: &text::BufferSnapshot) -> Ordering {
        if self
            .cmp(&cursor_location.buffer_range.start, buffer)
            .is_lt()
        {
            Ordering::Less
        } else if self.cmp(&cursor_location.buffer_range.end, buffer).is_gt() {
            Ordering::Greater
        } else {
            Ordering::Equal
        }
    }
}

impl std::fmt::Debug for BufferDiffInner {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        f.debug_struct("BufferDiffSnapshot")
            .field("hunks", &self.hunks)
            .field("remote_id", &self.base_text.remote_id())
            .finish()
    }
}

impl BufferDiffSnapshot {
    pub fn buffer_diff_id(&self) -> BufferId {
        self.inner.base_text.remote_id()
    }

    fn empty(buffer: &text::BufferSnapshot, cx: &mut App) -> BufferDiffSnapshot {
        BufferDiffSnapshot {
            inner: BufferDiffInner {
                base_text: language::Buffer::build_empty_snapshot(cx),
                hunks: SumTree::new(buffer),
                pending_hunks: SumTree::new(buffer),
                base_text_exists: false,
            },
            secondary_diff: None,
        }
    }

    fn unchanged(
        buffer: &text::BufferSnapshot,
        base_text: language::BufferSnapshot,
    ) -> BufferDiffSnapshot {
        debug_assert_eq!(buffer.text(), base_text.text());
        BufferDiffSnapshot {
            inner: BufferDiffInner {
                base_text,
                hunks: SumTree::new(buffer),
                pending_hunks: SumTree::new(buffer),
                base_text_exists: false,
            },
            secondary_diff: None,
        }
    }

    fn new_with_base_text(
        buffer: text::BufferSnapshot,
        base_text: Option<Arc<String>>,
        language: Option<Arc<Language>>,
        language_registry: Option<Arc<LanguageRegistry>>,
        cx: &mut App,
    ) -> impl Future<Output = Self> + use<> {
        let base_text_pair;
        let base_text_exists;
        let base_text_snapshot;
        let diff_options = build_diff_options(
            None,
            language.as_ref().map(|l| l.name()),
            language.as_ref().map(|l| l.default_scope()),
            cx,
        );

        if let Some(text) = &base_text {
            let base_text_rope = Rope::from(text.as_str());
            base_text_pair = Some((text.clone(), base_text_rope.clone()));
            let snapshot =
                language::Buffer::build_snapshot(base_text_rope, language, language_registry, cx);
            base_text_snapshot = cx.background_spawn(snapshot);
            base_text_exists = true;
        } else {
            base_text_pair = None;
            base_text_snapshot = Task::ready(language::Buffer::build_empty_snapshot(cx));
            base_text_exists = false;
        };

        let hunks = cx
            .background_executor()
            .spawn_labeled(*CALCULATE_DIFF_TASK, {
                let buffer = buffer.clone();
                async move { compute_hunks(base_text_pair, buffer, diff_options) }
            });

        async move {
            let (base_text, hunks) = futures::join!(base_text_snapshot, hunks);
            Self {
                inner: BufferDiffInner {
                    base_text,
                    hunks,
                    base_text_exists,
                    pending_hunks: SumTree::new(&buffer),
                },
                secondary_diff: None,
            }
        }
    }

    pub fn new_with_base_buffer(
        buffer: text::BufferSnapshot,
        base_text: Option<Arc<String>>,
        base_text_snapshot: language::BufferSnapshot,
        cx: &App,
    ) -> impl Future<Output = Self> + use<> {
        let diff_options = build_diff_options(
            base_text_snapshot.file(),
            base_text_snapshot.language().map(|l| l.name()),
            base_text_snapshot.language().map(|l| l.default_scope()),
            cx,
        );
        let base_text_exists = base_text.is_some();
        let base_text_pair = base_text.map(|text| {
            debug_assert_eq!(&*text, &base_text_snapshot.text());
            (text, base_text_snapshot.as_rope().clone())
        });
        cx.background_executor()
            .spawn_labeled(*CALCULATE_DIFF_TASK, async move {
                Self {
                    inner: BufferDiffInner {
                        base_text: base_text_snapshot,
                        pending_hunks: SumTree::new(&buffer),
                        hunks: compute_hunks(base_text_pair, buffer, diff_options),
                        base_text_exists,
                    },
                    secondary_diff: None,
                }
            })
    }

    #[cfg(test)]
    fn new_sync(
        buffer: text::BufferSnapshot,
        diff_base: String,
        cx: &mut gpui::TestAppContext,
    ) -> BufferDiffSnapshot {
        cx.executor().block(cx.update(|cx| {
            Self::new_with_base_text(buffer, Some(Arc::new(diff_base)), None, None, cx)
        }))
    }

    pub fn is_empty(&self) -> bool {
        self.inner.hunks.is_empty()
    }

    pub fn secondary_diff(&self) -> Option<&BufferDiffSnapshot> {
        self.secondary_diff.as_deref()
    }

    pub fn hunks_intersecting_range<'a>(
        &'a self,
        range: Range<Anchor>,
        buffer: &'a text::BufferSnapshot,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        let unstaged_counterpart = self.secondary_diff.as_ref().map(|diff| &diff.inner);
        self.inner
            .hunks_intersecting_range(range, buffer, unstaged_counterpart)
    }

    pub fn hunks_intersecting_range_rev<'a>(
        &'a self,
        range: Range<Anchor>,
        buffer: &'a text::BufferSnapshot,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        self.inner.hunks_intersecting_range_rev(range, buffer)
    }

    pub fn base_text(&self) -> &language::BufferSnapshot {
        &self.inner.base_text
    }

    pub fn base_texts_eq(&self, other: &Self) -> bool {
        if self.inner.base_text_exists != other.inner.base_text_exists {
            return false;
        }
        let left = &self.inner.base_text;
        let right = &other.inner.base_text;
        let (old_id, old_empty) = (left.remote_id(), left.is_empty());
        let (new_id, new_empty) = (right.remote_id(), right.is_empty());
        new_id == old_id || (new_empty && old_empty)
    }

    pub fn row_to_base_text_row(&self, row: BufferRow, buffer: &text::BufferSnapshot) -> u32 {
        // TODO(split-diff) expose a parameter to reuse a cursor to avoid repeatedly seeking from the start

        // Find the last hunk that starts before this position.
        let mut cursor = self.inner.hunks.cursor::<DiffHunkSummary>(buffer);
        let position = buffer.anchor_before(Point::new(row, 0));
        cursor.seek(&position, Bias::Left);
        if cursor
            .item()
            .is_none_or(|hunk| hunk.buffer_range.start.cmp(&position, buffer).is_gt())
        {
            cursor.prev();
        }

        let unclipped_point = if let Some(hunk) = cursor.item()
            && hunk.buffer_range.start.cmp(&position, buffer).is_le()
        {
            let mut unclipped_point = cursor
                .end()
                .diff_base_byte_range
                .end
                .to_point(self.base_text());
            if position.cmp(&cursor.end().buffer_range.end, buffer).is_ge() {
                unclipped_point +=
                    Point::new(row, 0) - cursor.end().buffer_range.end.to_point(buffer);
            }
            // Move the cursor so that at the next step we can clip with the start of the next hunk.
            cursor.next();
            unclipped_point
        } else {
            // Position is before the added region for the first hunk.
            debug_assert!(self.inner.hunks.first().is_none_or(|first_hunk| {
                position.cmp(&first_hunk.buffer_range.start, buffer).is_le()
            }));
            Point::new(row, 0)
        };

        let max_point = if let Some(next_hunk) = cursor.item() {
            next_hunk
                .diff_base_byte_range
                .start
                .to_point(self.base_text())
        } else {
            self.base_text().max_point()
        };
        unclipped_point.min(max_point).row
    }
}

impl BufferDiffInner {
    /// Returns the new index text and new pending hunks.
    fn stage_or_unstage_hunks_impl(
        &mut self,
        unstaged_diff: &Self,
        stage: bool,
        hunks: &[DiffHunk],
        buffer: &text::BufferSnapshot,
        file_exists: bool,
    ) -> Option<Rope> {
        let head_text = self
            .base_text_exists
            .then(|| self.base_text.as_rope().clone());
        let index_text = unstaged_diff
            .base_text_exists
            .then(|| unstaged_diff.base_text.as_rope().clone());

        // If the file doesn't exist in either HEAD or the index, then the
        // entire file must be either created or deleted in the index.
        let (index_text, head_text) = match (index_text, head_text) {
            (Some(index_text), Some(head_text)) if file_exists || !stage => (index_text, head_text),
            (index_text, head_text) => {
                let (new_index_text, new_status) = if stage {
                    log::debug!("stage all");
                    (
                        file_exists.then(|| buffer.as_rope().clone()),
                        DiffHunkSecondaryStatus::SecondaryHunkRemovalPending,
                    )
                } else {
                    log::debug!("unstage all");
                    (
                        head_text,
                        DiffHunkSecondaryStatus::SecondaryHunkAdditionPending,
                    )
                };

                let hunk = PendingHunk {
                    buffer_range: Anchor::min_max_range_for_buffer(buffer.remote_id()),
                    diff_base_byte_range: 0..index_text.map_or(0, |rope| rope.len()),
                    buffer_version: buffer.version().clone(),
                    new_status,
                };
                self.pending_hunks = SumTree::from_item(hunk, buffer);
                return new_index_text;
            }
        };

        let mut pending_hunks = SumTree::new(buffer);
        let mut old_pending_hunks = self.pending_hunks.cursor::<DiffHunkSummary>(buffer);

        // first, merge new hunks into pending_hunks
        for DiffHunk {
            buffer_range,
            diff_base_byte_range,
            secondary_status,
            ..
        } in hunks.iter().cloned()
        {
            let preceding_pending_hunks = old_pending_hunks.slice(&buffer_range.start, Bias::Left);
            pending_hunks.append(preceding_pending_hunks, buffer);

            // Skip all overlapping or adjacent old pending hunks
            while old_pending_hunks.item().is_some_and(|old_hunk| {
                old_hunk
                    .buffer_range
                    .start
                    .cmp(&buffer_range.end, buffer)
                    .is_le()
            }) {
                old_pending_hunks.next();
            }

            if (stage && secondary_status == DiffHunkSecondaryStatus::NoSecondaryHunk)
                || (!stage && secondary_status == DiffHunkSecondaryStatus::HasSecondaryHunk)
            {
                continue;
            }

            pending_hunks.push(
                PendingHunk {
                    buffer_range,
                    diff_base_byte_range,
                    buffer_version: buffer.version().clone(),
                    new_status: if stage {
                        DiffHunkSecondaryStatus::SecondaryHunkRemovalPending
                    } else {
                        DiffHunkSecondaryStatus::SecondaryHunkAdditionPending
                    },
                },
                buffer,
            );
        }
        // append the remainder
        pending_hunks.append(old_pending_hunks.suffix(), buffer);

        let mut unstaged_hunk_cursor = unstaged_diff.hunks.cursor::<DiffHunkSummary>(buffer);
        unstaged_hunk_cursor.next();

        // then, iterate over all pending hunks (both new ones and the existing ones) and compute the edits
        let mut prev_unstaged_hunk_buffer_end = 0;
        let mut prev_unstaged_hunk_base_text_end = 0;
        let mut edits = Vec::<(Range<usize>, String)>::new();
        let mut pending_hunks_iter = pending_hunks.iter().cloned().peekable();
        while let Some(PendingHunk {
            buffer_range,
            diff_base_byte_range,
            new_status,
            ..
        }) = pending_hunks_iter.next()
        {
            // Advance unstaged_hunk_cursor to skip unstaged hunks before current hunk
            let skipped_unstaged = unstaged_hunk_cursor.slice(&buffer_range.start, Bias::Left);

            if let Some(unstaged_hunk) = skipped_unstaged.last() {
                prev_unstaged_hunk_base_text_end = unstaged_hunk.diff_base_byte_range.end;
                prev_unstaged_hunk_buffer_end = unstaged_hunk.buffer_range.end.to_offset(buffer);
            }

            // Find where this hunk is in the index if it doesn't overlap
            let mut buffer_offset_range = buffer_range.to_offset(buffer);
            let start_overshoot = buffer_offset_range.start - prev_unstaged_hunk_buffer_end;
            let mut index_start = prev_unstaged_hunk_base_text_end + start_overshoot;

            loop {
                // Merge this hunk with any overlapping unstaged hunks.
                if let Some(unstaged_hunk) = unstaged_hunk_cursor.item() {
                    let unstaged_hunk_offset_range = unstaged_hunk.buffer_range.to_offset(buffer);
                    if unstaged_hunk_offset_range.start <= buffer_offset_range.end {
                        prev_unstaged_hunk_base_text_end = unstaged_hunk.diff_base_byte_range.end;
                        prev_unstaged_hunk_buffer_end = unstaged_hunk_offset_range.end;

                        index_start = index_start.min(unstaged_hunk.diff_base_byte_range.start);
                        buffer_offset_range.start = buffer_offset_range
                            .start
                            .min(unstaged_hunk_offset_range.start);
                        buffer_offset_range.end =
                            buffer_offset_range.end.max(unstaged_hunk_offset_range.end);

                        unstaged_hunk_cursor.next();
                        continue;
                    }
                }

                // If any unstaged hunks were merged, then subsequent pending hunks may
                // now overlap this hunk. Merge them.
                if let Some(next_pending_hunk) = pending_hunks_iter.peek() {
                    let next_pending_hunk_offset_range =
                        next_pending_hunk.buffer_range.to_offset(buffer);
                    if next_pending_hunk_offset_range.start <= buffer_offset_range.end {
                        buffer_offset_range.end = next_pending_hunk_offset_range.end;
                        pending_hunks_iter.next();
                        continue;
                    }
                }

                break;
            }

            let end_overshoot = buffer_offset_range
                .end
                .saturating_sub(prev_unstaged_hunk_buffer_end);
            let index_end = prev_unstaged_hunk_base_text_end + end_overshoot;
            let index_byte_range = index_start..index_end;

            let replacement_text = match new_status {
                DiffHunkSecondaryStatus::SecondaryHunkRemovalPending => {
                    log::debug!("staging hunk {:?}", buffer_offset_range);
                    buffer
                        .text_for_range(buffer_offset_range)
                        .collect::<String>()
                }
                DiffHunkSecondaryStatus::SecondaryHunkAdditionPending => {
                    log::debug!("unstaging hunk {:?}", buffer_offset_range);
                    head_text
                        .chunks_in_range(diff_base_byte_range.clone())
                        .collect::<String>()
                }
                _ => {
                    debug_assert!(false);
                    continue;
                }
            };

            edits.push((index_byte_range, replacement_text));
        }
        drop(pending_hunks_iter);
        drop(old_pending_hunks);
        self.pending_hunks = pending_hunks;

        #[cfg(debug_assertions)] // invariants: non-overlapping and sorted
        {
            for window in edits.windows(2) {
                let (range_a, range_b) = (&window[0].0, &window[1].0);
                debug_assert!(range_a.end < range_b.start);
            }
        }

        let mut new_index_text = Rope::new();
        let mut index_cursor = index_text.cursor(0);

        for (old_range, replacement_text) in edits {
            new_index_text.append(index_cursor.slice(old_range.start));
            index_cursor.seek_forward(old_range.end);
            new_index_text.push(&replacement_text);
        }
        new_index_text.append(index_cursor.suffix());
        Some(new_index_text)
    }

    fn hunks_intersecting_range<'a>(
        &'a self,
        range: Range<Anchor>,
        buffer: &'a text::BufferSnapshot,
        secondary: Option<&'a Self>,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        let range = range.to_offset(buffer);

        let mut cursor = self
            .hunks
            .filter::<_, DiffHunkSummary>(buffer, move |summary| {
                let summary_range = summary.buffer_range.to_offset(buffer);
                let before_start = summary_range.end < range.start;
                let after_end = summary_range.start > range.end;
                !before_start && !after_end
            });

        let anchor_iter = iter::from_fn(move || {
            cursor.next();
            cursor.item()
        })
        .flat_map(move |hunk| {
            [
                (
                    &hunk.buffer_range.start,
                    (
                        hunk.buffer_range.start,
                        hunk.diff_base_byte_range.start,
                        hunk,
                    ),
                ),
                (
                    &hunk.buffer_range.end,
                    (hunk.buffer_range.end, hunk.diff_base_byte_range.end, hunk),
                ),
            ]
        });

        let mut pending_hunks_cursor = self.pending_hunks.cursor::<DiffHunkSummary>(buffer);
        pending_hunks_cursor.next();

        let mut secondary_cursor = None;
        if let Some(secondary) = secondary.as_ref() {
            let mut cursor = secondary.hunks.cursor::<DiffHunkSummary>(buffer);
            cursor.next();
            secondary_cursor = Some(cursor);
        }

        let max_point = buffer.max_point();
        let mut summaries = buffer.summaries_for_anchors_with_payload::<Point, _, _>(anchor_iter);
        iter::from_fn(move || {
            loop {
                let (start_point, (start_anchor, start_base, hunk)) = summaries.next()?;
                let (mut end_point, (mut end_anchor, end_base, _)) = summaries.next()?;

                let base_word_diffs = hunk.base_word_diffs.clone();
                let buffer_word_diffs = hunk.buffer_word_diffs.clone();

                if !start_anchor.is_valid(buffer) {
                    continue;
                }

                if end_point.column > 0 && end_point < max_point {
                    end_point.row += 1;
                    end_point.column = 0;
                    end_anchor = buffer.anchor_before(end_point);
                }

                let mut secondary_status = DiffHunkSecondaryStatus::NoSecondaryHunk;

                let mut has_pending = false;
                if start_anchor
                    .cmp(&pending_hunks_cursor.start().buffer_range.start, buffer)
                    .is_gt()
                {
                    pending_hunks_cursor.seek_forward(&start_anchor, Bias::Left);
                }

                if let Some(pending_hunk) = pending_hunks_cursor.item() {
                    let mut pending_range = pending_hunk.buffer_range.to_point(buffer);
                    if pending_range.end.column > 0 {
                        pending_range.end.row += 1;
                        pending_range.end.column = 0;
                    }

                    if pending_range == (start_point..end_point)
                        && !buffer.has_edits_since_in_range(
                            &pending_hunk.buffer_version,
                            start_anchor..end_anchor,
                        )
                    {
                        has_pending = true;
                        secondary_status = pending_hunk.new_status;
                    }
                }

                if let (Some(secondary_cursor), false) = (secondary_cursor.as_mut(), has_pending) {
                    if start_anchor
                        .cmp(&secondary_cursor.start().buffer_range.start, buffer)
                        .is_gt()
                    {
                        secondary_cursor.seek_forward(&start_anchor, Bias::Left);
                    }

                    if let Some(secondary_hunk) = secondary_cursor.item() {
                        let mut secondary_range = secondary_hunk.buffer_range.to_point(buffer);
                        if secondary_range.end.column > 0 {
                            secondary_range.end.row += 1;
                            secondary_range.end.column = 0;
                        }
                        if secondary_range.is_empty()
                            && secondary_hunk.diff_base_byte_range.is_empty()
                        {
                            // ignore
                        } else if secondary_range == (start_point..end_point) {
                            secondary_status = DiffHunkSecondaryStatus::HasSecondaryHunk;
                        } else if secondary_range.start <= end_point {
                            secondary_status = DiffHunkSecondaryStatus::OverlapsWithSecondaryHunk;
                        }
                    }
                }

                return Some(DiffHunk {
                    range: start_point..end_point,
                    diff_base_byte_range: start_base..end_base,
                    buffer_range: start_anchor..end_anchor,
                    base_word_diffs,
                    buffer_word_diffs,
                    secondary_status,
                });
            }
        })
    }

    fn hunks_intersecting_range_rev<'a>(
        &'a self,
        range: Range<Anchor>,
        buffer: &'a text::BufferSnapshot,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        let mut cursor = self
            .hunks
            .filter::<_, DiffHunkSummary>(buffer, move |summary| {
                let before_start = summary.buffer_range.end.cmp(&range.start, buffer).is_lt();
                let after_end = summary.buffer_range.start.cmp(&range.end, buffer).is_gt();
                !before_start && !after_end
            });

        iter::from_fn(move || {
            cursor.prev();

            let hunk = cursor.item()?;
            let range = hunk.buffer_range.to_point(buffer);

            Some(DiffHunk {
                range,
                diff_base_byte_range: hunk.diff_base_byte_range.clone(),
                buffer_range: hunk.buffer_range.clone(),
                // The secondary status is not used by callers of this method.
                secondary_status: DiffHunkSecondaryStatus::NoSecondaryHunk,
                base_word_diffs: hunk.base_word_diffs.clone(),
                buffer_word_diffs: hunk.buffer_word_diffs.clone(),
            })
        })
    }

    fn compare(&self, old: &Self, new_snapshot: &text::BufferSnapshot) -> Option<Range<Anchor>> {
        let mut new_cursor = self.hunks.cursor::<()>(new_snapshot);
        let mut old_cursor = old.hunks.cursor::<()>(new_snapshot);
        old_cursor.next();
        new_cursor.next();
        let mut start = None;
        let mut end = None;

        loop {
            match (new_cursor.item(), old_cursor.item()) {
                (Some(new_hunk), Some(old_hunk)) => {
                    match new_hunk
                        .buffer_range
                        .start
                        .cmp(&old_hunk.buffer_range.start, new_snapshot)
                    {
                        Ordering::Less => {
                            start.get_or_insert(new_hunk.buffer_range.start);
                            end.replace(new_hunk.buffer_range.end);
                            new_cursor.next();
                        }
                        Ordering::Equal => {
                            if new_hunk != old_hunk {
                                start.get_or_insert(new_hunk.buffer_range.start);
                                if old_hunk
                                    .buffer_range
                                    .end
                                    .cmp(&new_hunk.buffer_range.end, new_snapshot)
                                    .is_ge()
                                {
                                    end.replace(old_hunk.buffer_range.end);
                                } else {
                                    end.replace(new_hunk.buffer_range.end);
                                }
                            }

                            new_cursor.next();
                            old_cursor.next();
                        }
                        Ordering::Greater => {
                            start.get_or_insert(old_hunk.buffer_range.start);
                            end.replace(old_hunk.buffer_range.end);
                            old_cursor.next();
                        }
                    }
                }
                (Some(new_hunk), None) => {
                    start.get_or_insert(new_hunk.buffer_range.start);
                    end.replace(new_hunk.buffer_range.end);
                    new_cursor.next();
                }
                (None, Some(old_hunk)) => {
                    start.get_or_insert(old_hunk.buffer_range.start);
                    end.replace(old_hunk.buffer_range.end);
                    old_cursor.next();
                }
                (None, None) => break,
            }
        }

        start.zip(end).map(|(start, end)| start..end)
    }
}

fn build_diff_options(
    file: Option<&Arc<dyn File>>,
    language: Option<LanguageName>,
    language_scope: Option<language::LanguageScope>,
    cx: &App,
) -> Option<DiffOptions> {
    #[cfg(any(test, feature = "test-support"))]
    {
        if !cx.has_global::<settings::SettingsStore>() {
            return Some(DiffOptions {
                language_scope,
                max_word_diff_line_count: MAX_WORD_DIFF_LINE_COUNT,
                ..Default::default()
            });
        }
    }

    language_settings(language, file, cx)
        .word_diff_enabled
        .then_some(DiffOptions {
            language_scope,
            max_word_diff_line_count: MAX_WORD_DIFF_LINE_COUNT,
            ..Default::default()
        })
}

fn compute_hunks(
    diff_base: Option<(Arc<String>, Rope)>,
    buffer: text::BufferSnapshot,
    diff_options: Option<DiffOptions>,
) -> SumTree<InternalDiffHunk> {
    let mut tree = SumTree::new(&buffer);

    if let Some((diff_base, diff_base_rope)) = diff_base {
        let buffer_text = buffer.as_rope().to_string();

        let mut options = GitOptions::default();
        options.context_lines(0);
        let patch = GitPatch::from_buffers(
            diff_base.as_bytes(),
            None,
            buffer_text.as_bytes(),
            None,
            Some(&mut options),
        )
        .log_err();

        // A common case in Zed is that the empty buffer is represented as just a newline,
        // but if we just compute a naive diff you get a "preserved" line in the middle,
        // which is a bit odd.
        if buffer_text == "\n" && diff_base.ends_with("\n") && diff_base.len() > 1 {
            tree.push(
                InternalDiffHunk {
                    buffer_range: buffer.anchor_before(0)..buffer.anchor_before(0),
                    diff_base_byte_range: 0..diff_base.len() - 1,
                    base_word_diffs: Vec::default(),
                    buffer_word_diffs: Vec::default(),
                },
                &buffer,
            );
            return tree;
        }

        if let Some(patch) = patch {
            let mut divergence = 0;
            for hunk_index in 0..patch.num_hunks() {
                let hunk = process_patch_hunk(
                    &patch,
                    hunk_index,
                    &diff_base_rope,
                    &buffer,
                    &mut divergence,
                    diff_options.as_ref(),
                );
                tree.push(hunk, &buffer);
            }
        }
    } else {
        tree.push(
            InternalDiffHunk {
                buffer_range: Anchor::min_max_range_for_buffer(buffer.remote_id()),
                diff_base_byte_range: 0..0,
                base_word_diffs: Vec::default(),
                buffer_word_diffs: Vec::default(),
            },
            &buffer,
        );
    }

    tree
}

fn process_patch_hunk(
    patch: &GitPatch<'_>,
    hunk_index: usize,
    diff_base: &Rope,
    buffer: &text::BufferSnapshot,
    buffer_row_divergence: &mut i64,
    diff_options: Option<&DiffOptions>,
) -> InternalDiffHunk {
    let line_item_count = patch.num_lines_in_hunk(hunk_index).unwrap();
    assert!(line_item_count > 0);

    let mut first_deletion_buffer_row: Option<u32> = None;
    let mut buffer_row_range: Option<Range<u32>> = None;
    let mut diff_base_byte_range: Option<Range<usize>> = None;
    let mut first_addition_old_row: Option<u32> = None;

    for line_index in 0..line_item_count {
        let line = patch.line_in_hunk(hunk_index, line_index).unwrap();
        let kind = line.origin_value();
        let content_offset = line.content_offset() as isize;
        let content_len = line.content().len() as isize;
        match kind {
            GitDiffLineType::Addition => {
                if first_addition_old_row.is_none() {
                    first_addition_old_row = Some(
                        (line.new_lineno().unwrap() as i64 - *buffer_row_divergence - 1) as u32,
                    );
                }
                *buffer_row_divergence += 1;
                let row = line.new_lineno().unwrap().saturating_sub(1);

                match &mut buffer_row_range {
                    Some(Range { end, .. }) => *end = row + 1,
                    None => buffer_row_range = Some(row..row + 1),
                }
            }
            GitDiffLineType::Deletion => {
                let end = content_offset + content_len;

                match &mut diff_base_byte_range {
                    Some(head_byte_range) => head_byte_range.end = end as usize,
                    None => diff_base_byte_range = Some(content_offset as usize..end as usize),
                }

                if first_deletion_buffer_row.is_none() {
                    let old_row = line.old_lineno().unwrap().saturating_sub(1);
                    let row = old_row as i64 + *buffer_row_divergence;
                    first_deletion_buffer_row = Some(row as u32);
                }

                *buffer_row_divergence -= 1;
            }
            _ => {}
        }
    }

    let buffer_row_range = buffer_row_range.unwrap_or_else(|| {
        // Pure deletion hunk without addition.
        let row = first_deletion_buffer_row.unwrap();
        row..row
    });
    let diff_base_byte_range = diff_base_byte_range.unwrap_or_else(|| {
        // Pure addition hunk without deletion.
        let row = first_addition_old_row.unwrap();
        let offset = diff_base.point_to_offset(Point::new(row, 0));
        offset..offset
    });

    let start = Point::new(buffer_row_range.start, 0);
    let end = Point::new(buffer_row_range.end, 0);
    let buffer_range = buffer.anchor_before(start)..buffer.anchor_before(end);

    let base_line_count = line_item_count.saturating_sub(buffer_row_range.len());

    let (base_word_diffs, buffer_word_diffs) = if let Some(diff_options) = diff_options
        && !buffer_row_range.is_empty()
        && base_line_count == buffer_row_range.len()
        && diff_options.max_word_diff_line_count >= base_line_count
    {
        let base_text: String = diff_base
            .chunks_in_range(diff_base_byte_range.clone())
            .collect();

        let buffer_text: String = buffer.text_for_range(buffer_range.clone()).collect();

        let (base_word_diffs, buffer_word_diffs_relative) = word_diff_ranges(
            &base_text,
            &buffer_text,
            DiffOptions {
                language_scope: diff_options.language_scope.clone(),
                ..*diff_options
            },
        );

        let buffer_start_offset = buffer_range.start.to_offset(buffer);
        let buffer_word_diffs = buffer_word_diffs_relative
            .into_iter()
            .map(|range| {
                let start = buffer.anchor_after(buffer_start_offset + range.start);
                let end = buffer.anchor_after(buffer_start_offset + range.end);
                start..end
            })
            .collect();

        (base_word_diffs, buffer_word_diffs)
    } else {
        (Vec::default(), Vec::default())
    };

    InternalDiffHunk {
        buffer_range,
        diff_base_byte_range,
        base_word_diffs,
        buffer_word_diffs,
    }
}

impl std::fmt::Debug for BufferDiff {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        f.debug_struct("BufferChangeSet")
            .field("buffer_id", &self.buffer_id)
            .field("snapshot", &self.inner)
            .finish()
    }
}

#[derive(Clone, Debug)]
pub enum BufferDiffEvent {
    DiffChanged {
        changed_range: Option<Range<text::Anchor>>,
    },
    LanguageChanged,
    HunksStagedOrUnstaged(Option<Rope>),
}

impl EventEmitter<BufferDiffEvent> for BufferDiff {}

impl BufferDiff {
    pub fn new(buffer: &text::BufferSnapshot, cx: &mut App) -> Self {
        BufferDiff {
            buffer_id: buffer.remote_id(),
            inner: BufferDiffSnapshot::empty(buffer, cx).inner,
            secondary_diff: None,
        }
    }

    pub fn new_unchanged(
        buffer: &text::BufferSnapshot,
        base_text: language::BufferSnapshot,
    ) -> Self {
        debug_assert_eq!(buffer.text(), base_text.text());
        BufferDiff {
            buffer_id: buffer.remote_id(),
            inner: BufferDiffSnapshot::unchanged(buffer, base_text).inner,
            secondary_diff: None,
        }
    }

    #[cfg(any(test, feature = "test-support"))]
    pub fn new_with_base_text(
        base_text: &str,
        buffer: &Entity<language::Buffer>,
        cx: &mut App,
    ) -> Self {
        let mut base_text = base_text.to_owned();
        text::LineEnding::normalize(&mut base_text);
        let snapshot = BufferDiffSnapshot::new_with_base_text(
            buffer.read(cx).text_snapshot(),
            Some(base_text.into()),
            None,
            None,
            cx,
        );
        let snapshot = cx.background_executor().block(snapshot);
        Self {
            buffer_id: buffer.read(cx).remote_id(),
            inner: snapshot.inner,
            secondary_diff: None,
        }
    }

    pub fn set_secondary_diff(&mut self, diff: Entity<BufferDiff>) {
        self.secondary_diff = Some(diff);
    }

    pub fn secondary_diff(&self) -> Option<Entity<BufferDiff>> {
        self.secondary_diff.clone()
    }

    pub fn clear_pending_hunks(&mut self, cx: &mut Context<Self>) {
        if self.secondary_diff.is_some() {
            self.inner.pending_hunks = SumTree::from_summary(DiffHunkSummary {
                buffer_range: Anchor::min_min_range_for_buffer(self.buffer_id),
                diff_base_byte_range: 0..0,
            });
            cx.emit(BufferDiffEvent::DiffChanged {
                changed_range: Some(Anchor::min_max_range_for_buffer(self.buffer_id)),
            });
        }
    }

    pub fn stage_or_unstage_hunks(
        &mut self,
        stage: bool,
        hunks: &[DiffHunk],
        buffer: &text::BufferSnapshot,
        file_exists: bool,
        cx: &mut Context<Self>,
    ) -> Option<Rope> {
        let new_index_text = self.inner.stage_or_unstage_hunks_impl(
            &self.secondary_diff.as_ref()?.read(cx).inner,
            stage,
            hunks,
            buffer,
            file_exists,
        );

        cx.emit(BufferDiffEvent::HunksStagedOrUnstaged(
            new_index_text.clone(),
        ));
        if let Some((first, last)) = hunks.first().zip(hunks.last()) {
            let changed_range = first.buffer_range.start..last.buffer_range.end;
            cx.emit(BufferDiffEvent::DiffChanged {
                changed_range: Some(changed_range),
            });
        }
        new_index_text
    }

    pub fn range_to_hunk_range(
        &self,
        range: Range<Anchor>,
        buffer: &text::BufferSnapshot,
        cx: &App,
    ) -> Option<Range<Anchor>> {
        let start = self
            .hunks_intersecting_range(range.clone(), buffer, cx)
            .next()?
            .buffer_range
            .start;
        let end = self
            .hunks_intersecting_range_rev(range, buffer)
            .next()?
            .buffer_range
            .end;
        Some(start..end)
    }

    pub async fn update_diff(
        this: Entity<BufferDiff>,
        buffer: text::BufferSnapshot,
        base_text: Option<Arc<String>>,
        base_text_changed: bool,
        language_changed: bool,
        language: Option<Arc<Language>>,
        language_registry: Option<Arc<LanguageRegistry>>,
        cx: &mut AsyncApp,
    ) -> anyhow::Result<BufferDiffSnapshot> {
        Ok(if base_text_changed || language_changed {
            cx.update(|cx| {
                BufferDiffSnapshot::new_with_base_text(
                    buffer.clone(),
                    base_text,
                    language.clone(),
                    language_registry.clone(),
                    cx,
                )
            })?
            .await
        } else {
            this.read_with(cx, |this, cx| {
                BufferDiffSnapshot::new_with_base_buffer(
                    buffer.clone(),
                    base_text,
                    this.base_text().clone(),
                    cx,
                )
            })?
            .await
        })
    }

    pub fn language_changed(&mut self, cx: &mut Context<Self>) {
        cx.emit(BufferDiffEvent::LanguageChanged);
    }

    pub fn set_snapshot(
        &mut self,
        new_snapshot: BufferDiffSnapshot,
        buffer: &text::BufferSnapshot,
        cx: &mut Context<Self>,
    ) -> Option<Range<Anchor>> {
        self.set_snapshot_with_secondary(new_snapshot, buffer, None, false, cx)
    }

    pub fn set_snapshot_with_secondary(
        &mut self,
        new_snapshot: BufferDiffSnapshot,
        buffer: &text::BufferSnapshot,
        secondary_diff_change: Option<Range<Anchor>>,
        clear_pending_hunks: bool,
        cx: &mut Context<Self>,
    ) -> Option<Range<Anchor>> {
        log::debug!("set snapshot with secondary {secondary_diff_change:?}");

        let state = &mut self.inner;
        let new_state = new_snapshot.inner;
        let (base_text_changed, mut changed_range) =
            match (state.base_text_exists, new_state.base_text_exists) {
                (false, false) => (true, None),
                (true, true)
                    if state.base_text.remote_id() == new_state.base_text.remote_id()
                        && state.base_text.syntax_update_count()
                            == new_state.base_text.syntax_update_count() =>
                {
                    (false, new_state.compare(state, buffer))
                }
                _ => (
                    true,
                    Some(text::Anchor::min_max_range_for_buffer(self.buffer_id)),
                ),
            };

        if let Some(secondary_changed_range) = secondary_diff_change
            && let Some(secondary_hunk_range) =
                self.range_to_hunk_range(secondary_changed_range, buffer, cx)
        {
            if let Some(range) = &mut changed_range {
                range.start = *secondary_hunk_range.start.min(&range.start, buffer);
                range.end = *secondary_hunk_range.end.max(&range.end, buffer);
            } else {
                changed_range = Some(secondary_hunk_range);
            }
        }

        let state = &mut self.inner;
        state.base_text_exists = new_state.base_text_exists;
        state.base_text = new_state.base_text;
        state.hunks = new_state.hunks;
        if base_text_changed || clear_pending_hunks {
            if let Some((first, last)) = state.pending_hunks.first().zip(state.pending_hunks.last())
            {
                if let Some(range) = &mut changed_range {
                    range.start = *range.start.min(&first.buffer_range.start, buffer);
                    range.end = *range.end.max(&last.buffer_range.end, buffer);
                } else {
                    changed_range = Some(first.buffer_range.start..last.buffer_range.end);
                }
            }
            state.pending_hunks = SumTree::new(buffer);
        }

        cx.emit(BufferDiffEvent::DiffChanged {
            changed_range: changed_range.clone(),
        });
        changed_range
    }

    pub fn base_text(&self) -> &language::BufferSnapshot {
        &self.inner.base_text
    }

    pub fn base_text_exists(&self) -> bool {
        self.inner.base_text_exists
    }

    pub fn snapshot(&self, cx: &App) -> BufferDiffSnapshot {
        BufferDiffSnapshot {
            inner: self.inner.clone(),
            secondary_diff: self
                .secondary_diff
                .as_ref()
                .map(|diff| Box::new(diff.read(cx).snapshot(cx))),
        }
    }

    pub fn hunks<'a>(
        &'a self,
        buffer_snapshot: &'a text::BufferSnapshot,
        cx: &'a App,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        self.hunks_intersecting_range(
            Anchor::min_max_range_for_buffer(buffer_snapshot.remote_id()),
            buffer_snapshot,
            cx,
        )
    }

    pub fn hunks_intersecting_range<'a>(
        &'a self,
        range: Range<text::Anchor>,
        buffer_snapshot: &'a text::BufferSnapshot,
        cx: &'a App,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        let unstaged_counterpart = self
            .secondary_diff
            .as_ref()
            .map(|diff| &diff.read(cx).inner);
        self.inner
            .hunks_intersecting_range(range, buffer_snapshot, unstaged_counterpart)
    }

    pub fn hunks_intersecting_range_rev<'a>(
        &'a self,
        range: Range<text::Anchor>,
        buffer_snapshot: &'a text::BufferSnapshot,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        self.inner
            .hunks_intersecting_range_rev(range, buffer_snapshot)
    }

    pub fn hunks_in_row_range<'a>(
        &'a self,
        range: Range<u32>,
        buffer: &'a text::BufferSnapshot,
        cx: &'a App,
    ) -> impl 'a + Iterator<Item = DiffHunk> {
        let start = buffer.anchor_before(Point::new(range.start, 0));
        let end = buffer.anchor_after(Point::new(range.end, 0));
        self.hunks_intersecting_range(start..end, buffer, cx)
    }

    /// Used in cases where the change set isn't derived from git.
    pub fn set_base_text(
        &mut self,
        base_text: Option<Arc<String>>,
        language: Option<Arc<Language>>,
        language_registry: Option<Arc<LanguageRegistry>>,
        buffer: text::BufferSnapshot,
        cx: &mut Context<Self>,
    ) -> oneshot::Receiver<()> {
        let (tx, rx) = oneshot::channel();
        let this = cx.weak_entity();

        let snapshot = BufferDiffSnapshot::new_with_base_text(
            buffer.clone(),
            base_text,
            language,
            language_registry,
            cx,
        );
        let complete_on_drop = util::defer(|| {
            tx.send(()).ok();
        });
        cx.spawn(async move |_, cx| {
            let snapshot = snapshot.await;
            let Some(this) = this.upgrade() else {
                return;
            };
            this.update(cx, |this, cx| {
                this.set_snapshot(snapshot, &buffer, cx);
            })
            .log_err();
            drop(complete_on_drop)
        })
        .detach();
        rx
    }

    pub fn base_text_string(&self) -> Option<String> {
        self.inner
            .base_text_exists
            .then(|| self.inner.base_text.text())
    }

    #[cfg(any(test, feature = "test-support"))]
    pub fn recalculate_diff_sync(&mut self, buffer: text::BufferSnapshot, cx: &mut Context<Self>) {
        let base_text = self.base_text_string().map(Arc::new);
        let snapshot = BufferDiffSnapshot::new_with_base_buffer(
            buffer.clone(),
            base_text,
            self.inner.base_text.clone(),
            cx,
        );
        let snapshot = cx.background_executor().block(snapshot);
        self.set_snapshot(snapshot, &buffer, cx);
    }
}

impl DiffHunk {
    pub fn is_created_file(&self) -> bool {
        self.diff_base_byte_range == (0..0)
            && self.buffer_range.start.is_min()
            && self.buffer_range.end.is_max()
    }

    pub fn status(&self) -> DiffHunkStatus {
        let kind = if self.buffer_range.start == self.buffer_range.end {
            DiffHunkStatusKind::Deleted
        } else if self.diff_base_byte_range.is_empty() {
            DiffHunkStatusKind::Added
        } else {
            DiffHunkStatusKind::Modified
        };
        DiffHunkStatus {
            kind,
            secondary: self.secondary_status,
        }
    }
}

impl DiffHunkStatus {
    pub fn has_secondary_hunk(&self) -> bool {
        matches!(
            self.secondary,
            DiffHunkSecondaryStatus::HasSecondaryHunk
                | DiffHunkSecondaryStatus::SecondaryHunkAdditionPending
                | DiffHunkSecondaryStatus::OverlapsWithSecondaryHunk
        )
    }

    pub fn is_pending(&self) -> bool {
        matches!(
            self.secondary,
            DiffHunkSecondaryStatus::SecondaryHunkAdditionPending
                | DiffHunkSecondaryStatus::SecondaryHunkRemovalPending
        )
    }

    pub fn is_deleted(&self) -> bool {
        self.kind == DiffHunkStatusKind::Deleted
    }

    pub fn is_added(&self) -> bool {
        self.kind == DiffHunkStatusKind::Added
    }

    pub fn is_modified(&self) -> bool {
        self.kind == DiffHunkStatusKind::Modified
    }

    pub fn added(secondary: DiffHunkSecondaryStatus) -> Self {
        Self {
            kind: DiffHunkStatusKind::Added,
            secondary,
        }
    }

    pub fn modified(secondary: DiffHunkSecondaryStatus) -> Self {
        Self {
            kind: DiffHunkStatusKind::Modified,
            secondary,
        }
    }

    pub fn deleted(secondary: DiffHunkSecondaryStatus) -> Self {
        Self {
            kind: DiffHunkStatusKind::Deleted,
            secondary,
        }
    }

    pub fn deleted_none() -> Self {
        Self {
            kind: DiffHunkStatusKind::Deleted,
            secondary: DiffHunkSecondaryStatus::NoSecondaryHunk,
        }
    }

    pub fn added_none() -> Self {
        Self {
            kind: DiffHunkStatusKind::Added,
            secondary: DiffHunkSecondaryStatus::NoSecondaryHunk,
        }
    }

    pub fn modified_none() -> Self {
        Self {
            kind: DiffHunkStatusKind::Modified,
            secondary: DiffHunkSecondaryStatus::NoSecondaryHunk,
        }
    }
}

#[cfg(any(test, feature = "test-support"))]
#[track_caller]
pub fn assert_hunks<ExpectedText, HunkIter>(
    diff_hunks: HunkIter,
    buffer: &text::BufferSnapshot,
    diff_base: &str,
    // Line range, deleted, added, status
    expected_hunks: &[(Range<u32>, ExpectedText, ExpectedText, DiffHunkStatus)],
) where
    HunkIter: Iterator<Item = DiffHunk>,
    ExpectedText: AsRef<str>,
{
    let actual_hunks = diff_hunks
        .map(|hunk| {
            (
                hunk.range.clone(),
                &diff_base[hunk.diff_base_byte_range.clone()],
                buffer
                    .text_for_range(hunk.range.clone())
                    .collect::<String>(),
                hunk.status(),
            )
        })
        .collect::<Vec<_>>();

    let expected_hunks: Vec<_> = expected_hunks
        .iter()
        .map(|(line_range, deleted_text, added_text, status)| {
            (
                Point::new(line_range.start, 0)..Point::new(line_range.end, 0),
                deleted_text.as_ref(),
                added_text.as_ref().to_string(),
                *status,
            )
        })
        .collect();

    pretty_assertions::assert_eq!(actual_hunks, expected_hunks);
}

#[cfg(test)]
mod tests {
    use std::fmt::Write as _;

    use super::*;
    use gpui::TestAppContext;
    use pretty_assertions::{assert_eq, assert_ne};
    use rand::{Rng as _, rngs::StdRng};
    use text::{Buffer, BufferId, ReplicaId, Rope};
    use unindent::Unindent as _;
    use util::test::marked_text_ranges;

    #[ctor::ctor]
    fn init_logger() {
        zlog::init_test();
    }

    #[gpui::test]
    async fn test_buffer_diff_simple(cx: &mut gpui::TestAppContext) {
        let diff_base = "
            one
            two
            three
        "
        .unindent();

        let buffer_text = "
            one
            HELLO
            three
        "
        .unindent();

        let mut buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text);
        let mut diff = BufferDiffSnapshot::new_sync(buffer.clone(), diff_base.clone(), cx);
        assert_hunks(
            diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(buffer.remote_id()),
                &buffer,
            ),
            &buffer,
            &diff_base,
            &[(1..2, "two\n", "HELLO\n", DiffHunkStatus::modified_none())],
        );

        buffer.edit([(0..0, "point five\n")]);
        diff = BufferDiffSnapshot::new_sync(buffer.clone(), diff_base.clone(), cx);
        assert_hunks(
            diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(buffer.remote_id()),
                &buffer,
            ),
            &buffer,
            &diff_base,
            &[
                (0..1, "", "point five\n", DiffHunkStatus::added_none()),
                (2..3, "two\n", "HELLO\n", DiffHunkStatus::modified_none()),
            ],
        );

        diff = cx.update(|cx| BufferDiffSnapshot::empty(&buffer, cx));
        assert_hunks::<&str, _>(
            diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(buffer.remote_id()),
                &buffer,
            ),
            &buffer,
            &diff_base,
            &[],
        );
    }

    #[gpui::test]
    async fn test_buffer_diff_with_secondary(cx: &mut gpui::TestAppContext) {
        let head_text = "
            zero
            one
            two
            three
            four
            five
            six
            seven
            eight
            nine
        "
        .unindent();

        let index_text = "
            zero
            one
            TWO
            three
            FOUR
            five
            six
            seven
            eight
            NINE
        "
        .unindent();

        let buffer_text = "
            zero
            one
            TWO
            three
            FOUR
            FIVE
            six
            SEVEN
            eight
            nine
        "
        .unindent();

        let buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text);
        let unstaged_diff = BufferDiffSnapshot::new_sync(buffer.clone(), index_text, cx);
        let mut uncommitted_diff =
            BufferDiffSnapshot::new_sync(buffer.clone(), head_text.clone(), cx);
        uncommitted_diff.secondary_diff = Some(Box::new(unstaged_diff));

        let expected_hunks = vec![
            (2..3, "two\n", "TWO\n", DiffHunkStatus::modified_none()),
            (
                4..6,
                "four\nfive\n",
                "FOUR\nFIVE\n",
                DiffHunkStatus::modified(DiffHunkSecondaryStatus::OverlapsWithSecondaryHunk),
            ),
            (
                7..8,
                "seven\n",
                "SEVEN\n",
                DiffHunkStatus::modified(DiffHunkSecondaryStatus::HasSecondaryHunk),
            ),
        ];

        assert_hunks(
            uncommitted_diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(buffer.remote_id()),
                &buffer,
            ),
            &buffer,
            &head_text,
            &expected_hunks,
        );
    }

    #[gpui::test]
    async fn test_buffer_diff_range(cx: &mut TestAppContext) {
        let diff_base = Arc::new(
            "
            one
            two
            three
            four
            five
            six
            seven
            eight
            nine
            ten
        "
            .unindent(),
        );

        let buffer_text = "
            A
            one
            B
            two
            C
            three
            HELLO
            four
            five
            SIXTEEN
            seven
            eight
            WORLD
            nine

            ten

        "
        .unindent();

        let buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text);
        let diff = cx
            .update(|cx| {
                BufferDiffSnapshot::new_with_base_text(
                    buffer.snapshot(),
                    Some(diff_base.clone()),
                    None,
                    None,
                    cx,
                )
            })
            .await;
        assert_eq!(
            diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(buffer.remote_id()),
                &buffer
            )
            .count(),
            8
        );

        assert_hunks(
            diff.hunks_intersecting_range(
                buffer.anchor_before(Point::new(7, 0))..buffer.anchor_before(Point::new(12, 0)),
                &buffer,
            ),
            &buffer,
            &diff_base,
            &[
                (6..7, "", "HELLO\n", DiffHunkStatus::added_none()),
                (9..10, "six\n", "SIXTEEN\n", DiffHunkStatus::modified_none()),
                (12..13, "", "WORLD\n", DiffHunkStatus::added_none()),
            ],
        );
    }

    #[gpui::test]
    async fn test_stage_hunk(cx: &mut TestAppContext) {
        struct Example {
            name: &'static str,
            head_text: String,
            index_text: String,
            buffer_marked_text: String,
            final_index_text: String,
        }

        let table = [
            Example {
                name: "uncommitted hunk straddles end of unstaged hunk",
                head_text: "
                    one
                    two
                    three
                    four
                    five
                "
                .unindent(),
                index_text: "
                    one
                    TWO_HUNDRED
                    three
                    FOUR_HUNDRED
                    five
                "
                .unindent(),
                buffer_marked_text: "
                    ZERO
                    one
                    two
                    «THREE_HUNDRED
                    FOUR_HUNDRED»
                    five
                    SIX
                "
                .unindent(),
                final_index_text: "
                    one
                    two
                    THREE_HUNDRED
                    FOUR_HUNDRED
                    five
                "
                .unindent(),
            },
            Example {
                name: "uncommitted hunk straddles start of unstaged hunk",
                head_text: "
                    one
                    two
                    three
                    four
                    five
                "
                .unindent(),
                index_text: "
                    one
                    TWO_HUNDRED
                    three
                    FOUR_HUNDRED
                    five
                "
                .unindent(),
                buffer_marked_text: "
                    ZERO
                    one
                    «TWO_HUNDRED
                    THREE_HUNDRED»
                    four
                    five
                    SIX
                "
                .unindent(),
                final_index_text: "
                    one
                    TWO_HUNDRED
                    THREE_HUNDRED
                    four
                    five
                "
                .unindent(),
            },
            Example {
                name: "uncommitted hunk strictly contains unstaged hunks",
                head_text: "
                    one
                    two
                    three
                    four
                    five
                    six
                    seven
                "
                .unindent(),
                index_text: "
                    one
                    TWO
                    THREE
                    FOUR
                    FIVE
                    SIX
                    seven
                "
                .unindent(),
                buffer_marked_text: "
                    one
                    TWO
                    «THREE_HUNDRED
                    FOUR
                    FIVE_HUNDRED»
                    SIX
                    seven
                "
                .unindent(),
                final_index_text: "
                    one
                    TWO
                    THREE_HUNDRED
                    FOUR
                    FIVE_HUNDRED
                    SIX
                    seven
                "
                .unindent(),
            },
            Example {
                name: "uncommitted deletion hunk",
                head_text: "
                    one
                    two
                    three
                    four
                    five
                "
                .unindent(),
                index_text: "
                    one
                    two
                    three
                    four
                    five
                "
                .unindent(),
                buffer_marked_text: "
                    one
                    ˇfive
                "
                .unindent(),
                final_index_text: "
                    one
                    five
                "
                .unindent(),
            },
            Example {
                name: "one unstaged hunk that contains two uncommitted hunks",
                head_text: "
                    one
                    two

                    three
                    four
                "
                .unindent(),
                index_text: "
                    one
                    two
                    three
                    four
                "
                .unindent(),
                buffer_marked_text: "
                    «one

                    three // modified
                    four»
                "
                .unindent(),
                final_index_text: "
                    one

                    three // modified
                    four
                "
                .unindent(),
            },
            Example {
                name: "one uncommitted hunk that contains two unstaged hunks",
                head_text: "
                    one
                    two
                    three
                    four
                    five
                "
                .unindent(),
                index_text: "
                    ZERO
                    one
                    TWO
                    THREE
                    FOUR
                    five
                "
                .unindent(),
                buffer_marked_text: "
                    «one
                    TWO_HUNDRED
                    THREE
                    FOUR_HUNDRED
                    five»
                "
                .unindent(),
                final_index_text: "
                    ZERO
                    one
                    TWO_HUNDRED
                    THREE
                    FOUR_HUNDRED
                    five
                "
                .unindent(),
            },
        ];

        for example in table {
            let (buffer_text, ranges) = marked_text_ranges(&example.buffer_marked_text, false);
            let buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text);
            let hunk_range =
                buffer.anchor_before(ranges[0].start)..buffer.anchor_before(ranges[0].end);

            let unstaged =
                BufferDiffSnapshot::new_sync(buffer.clone(), example.index_text.clone(), cx);
            let uncommitted =
                BufferDiffSnapshot::new_sync(buffer.clone(), example.head_text.clone(), cx);

            let unstaged_diff = cx.new(|cx| {
                let mut diff = BufferDiff::new(&buffer, cx);
                diff.set_snapshot(unstaged, &buffer, cx);
                diff
            });

            let uncommitted_diff = cx.new(|cx| {
                let mut diff = BufferDiff::new(&buffer, cx);
                diff.set_snapshot(uncommitted, &buffer, cx);
                diff.set_secondary_diff(unstaged_diff);
                diff
            });

            uncommitted_diff.update(cx, |diff, cx| {
                let hunks = diff
                    .hunks_intersecting_range(hunk_range.clone(), &buffer, cx)
                    .collect::<Vec<_>>();
                for hunk in &hunks {
                    assert_ne!(
                        hunk.secondary_status,
                        DiffHunkSecondaryStatus::NoSecondaryHunk
                    )
                }

                let new_index_text = diff
                    .stage_or_unstage_hunks(true, &hunks, &buffer, true, cx)
                    .unwrap()
                    .to_string();

                let hunks = diff
                    .hunks_intersecting_range(hunk_range.clone(), &buffer, cx)
                    .collect::<Vec<_>>();
                for hunk in &hunks {
                    assert_eq!(
                        hunk.secondary_status,
                        DiffHunkSecondaryStatus::SecondaryHunkRemovalPending
                    )
                }

                pretty_assertions::assert_eq!(
                    new_index_text,
                    example.final_index_text,
                    "example: {}",
                    example.name
                );
            });
        }
    }

    #[gpui::test]
    async fn test_toggling_stage_and_unstage_same_hunk(cx: &mut TestAppContext) {
        let head_text = "
            one
            two
            three
        "
        .unindent();
        let index_text = head_text.clone();
        let buffer_text = "
            one
            three
        "
        .unindent();

        let buffer = Buffer::new(
            ReplicaId::LOCAL,
            BufferId::new(1).unwrap(),
            buffer_text.clone(),
        );
        let unstaged = BufferDiffSnapshot::new_sync(buffer.clone(), index_text, cx);
        let uncommitted = BufferDiffSnapshot::new_sync(buffer.clone(), head_text.clone(), cx);
        let unstaged_diff = cx.new(|cx| {
            let mut diff = BufferDiff::new(&buffer, cx);
            diff.set_snapshot(unstaged, &buffer, cx);
            diff
        });
        let uncommitted_diff = cx.new(|cx| {
            let mut diff = BufferDiff::new(&buffer, cx);
            diff.set_snapshot(uncommitted, &buffer, cx);
            diff.set_secondary_diff(unstaged_diff.clone());
            diff
        });

        uncommitted_diff.update(cx, |diff, cx| {
            let hunk = diff.hunks(&buffer, cx).next().unwrap();

            let new_index_text = diff
                .stage_or_unstage_hunks(true, std::slice::from_ref(&hunk), &buffer, true, cx)
                .unwrap()
                .to_string();
            assert_eq!(new_index_text, buffer_text);

            let hunk = diff.hunks(&buffer, cx).next().unwrap();
            assert_eq!(
                hunk.secondary_status,
                DiffHunkSecondaryStatus::SecondaryHunkRemovalPending
            );

            let index_text = diff
                .stage_or_unstage_hunks(false, &[hunk], &buffer, true, cx)
                .unwrap()
                .to_string();
            assert_eq!(index_text, head_text);

            let hunk = diff.hunks(&buffer, cx).next().unwrap();
            // optimistically unstaged (fine, could also be HasSecondaryHunk)
            assert_eq!(
                hunk.secondary_status,
                DiffHunkSecondaryStatus::SecondaryHunkAdditionPending
            );
        });
    }

    #[gpui::test]
    async fn test_buffer_diff_compare(cx: &mut TestAppContext) {
        let base_text = "
            zero
            one
            two
            three
            four
            five
            six
            seven
            eight
            nine
        "
        .unindent();

        let buffer_text_1 = "
            one
            three
            four
            five
            SIX
            seven
            eight
            NINE
        "
        .unindent();

        let mut buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text_1);

        let empty_diff = cx.update(|cx| BufferDiffSnapshot::empty(&buffer, cx));
        let diff_1 = BufferDiffSnapshot::new_sync(buffer.clone(), base_text.clone(), cx);
        let range = diff_1.inner.compare(&empty_diff.inner, &buffer).unwrap();
        assert_eq!(range.to_point(&buffer), Point::new(0, 0)..Point::new(8, 0));

        // Edit does not affect the diff.
        buffer.edit_via_marked_text(
            &"
                one
                three
                four
                five
                «SIX.5»
                seven
                eight
                NINE
            "
            .unindent(),
        );
        let diff_2 = BufferDiffSnapshot::new_sync(buffer.clone(), base_text.clone(), cx);
        assert_eq!(None, diff_2.inner.compare(&diff_1.inner, &buffer));

        // Edit turns a deletion hunk into a modification.
        buffer.edit_via_marked_text(
            &"
                one
                «THREE»
                four
                five
                SIX.5
                seven
                eight
                NINE
            "
            .unindent(),
        );
        let diff_3 = BufferDiffSnapshot::new_sync(buffer.clone(), base_text.clone(), cx);
        let range = diff_3.inner.compare(&diff_2.inner, &buffer).unwrap();
        assert_eq!(range.to_point(&buffer), Point::new(1, 0)..Point::new(2, 0));

        // Edit turns a modification hunk into a deletion.
        buffer.edit_via_marked_text(
            &"
                one
                THREE
                four
                five«»
                seven
                eight
                NINE
            "
            .unindent(),
        );
        let diff_4 = BufferDiffSnapshot::new_sync(buffer.clone(), base_text.clone(), cx);
        let range = diff_4.inner.compare(&diff_3.inner, &buffer).unwrap();
        assert_eq!(range.to_point(&buffer), Point::new(3, 4)..Point::new(4, 0));

        // Edit introduces a new insertion hunk.
        buffer.edit_via_marked_text(
            &"
                one
                THREE
                four«
                FOUR.5
                »five
                seven
                eight
                NINE
            "
            .unindent(),
        );
        let diff_5 = BufferDiffSnapshot::new_sync(buffer.snapshot(), base_text.clone(), cx);
        let range = diff_5.inner.compare(&diff_4.inner, &buffer).unwrap();
        assert_eq!(range.to_point(&buffer), Point::new(3, 0)..Point::new(4, 0));

        // Edit removes a hunk.
        buffer.edit_via_marked_text(
            &"
                one
                THREE
                four
                FOUR.5
                five
                seven
                eight
                «nine»
            "
            .unindent(),
        );
        let diff_6 = BufferDiffSnapshot::new_sync(buffer.snapshot(), base_text, cx);
        let range = diff_6.inner.compare(&diff_5.inner, &buffer).unwrap();
        assert_eq!(range.to_point(&buffer), Point::new(7, 0)..Point::new(8, 0));
    }

    #[gpui::test(iterations = 100)]
    async fn test_staging_and_unstaging_hunks(cx: &mut TestAppContext, mut rng: StdRng) {
        fn gen_line(rng: &mut StdRng) -> String {
            if rng.random_bool(0.2) {
                "\n".to_owned()
            } else {
                let c = rng.random_range('A'..='Z');
                format!("{c}{c}{c}\n")
            }
        }

        fn gen_working_copy(rng: &mut StdRng, head: &str) -> String {
            let mut old_lines = {
                let mut old_lines = Vec::new();
                let old_lines_iter = head.lines();
                for line in old_lines_iter {
                    assert!(!line.ends_with("\n"));
                    old_lines.push(line.to_owned());
                }
                if old_lines.last().is_some_and(|line| line.is_empty()) {
                    old_lines.pop();
                }
                old_lines.into_iter()
            };
            let mut result = String::new();
            let unchanged_count = rng.random_range(0..=old_lines.len());
            result +=
                &old_lines
                    .by_ref()
                    .take(unchanged_count)
                    .fold(String::new(), |mut s, line| {
                        writeln!(&mut s, "{line}").unwrap();
                        s
                    });
            while old_lines.len() > 0 {
                let deleted_count = rng.random_range(0..=old_lines.len());
                let _advance = old_lines
                    .by_ref()
                    .take(deleted_count)
                    .map(|line| line.len() + 1)
                    .sum::<usize>();
                let minimum_added = if deleted_count == 0 { 1 } else { 0 };
                let added_count = rng.random_range(minimum_added..=5);
                let addition = (0..added_count).map(|_| gen_line(rng)).collect::<String>();
                result += &addition;

                if old_lines.len() > 0 {
                    let blank_lines = old_lines.clone().take_while(|line| line.is_empty()).count();
                    if blank_lines == old_lines.len() {
                        break;
                    };
                    let unchanged_count =
                        rng.random_range((blank_lines + 1).max(1)..=old_lines.len());
                    result += &old_lines.by_ref().take(unchanged_count).fold(
                        String::new(),
                        |mut s, line| {
                            writeln!(&mut s, "{line}").unwrap();
                            s
                        },
                    );
                }
            }
            result
        }

        fn uncommitted_diff(
            working_copy: &language::BufferSnapshot,
            index_text: &Rope,
            head_text: String,
            cx: &mut TestAppContext,
        ) -> Entity<BufferDiff> {
            let inner =
                BufferDiffSnapshot::new_sync(working_copy.text.clone(), head_text, cx).inner;
            let secondary = BufferDiff {
                buffer_id: working_copy.remote_id(),
                inner: BufferDiffSnapshot::new_sync(
                    working_copy.text.clone(),
                    index_text.to_string(),
                    cx,
                )
                .inner,
                secondary_diff: None,
            };
            let secondary = cx.new(|_| secondary);
            cx.new(|_| BufferDiff {
                buffer_id: working_copy.remote_id(),
                inner,
                secondary_diff: Some(secondary),
            })
        }

        let operations = std::env::var("OPERATIONS")
            .map(|i| i.parse().expect("invalid `OPERATIONS` variable"))
            .unwrap_or(10);

        let rng = &mut rng;
        let head_text = ('a'..='z').fold(String::new(), |mut s, c| {
            writeln!(&mut s, "{c}{c}{c}").unwrap();
            s
        });
        let working_copy = gen_working_copy(rng, &head_text);
        let working_copy = cx.new(|cx| {
            language::Buffer::local_normalized(
                Rope::from(working_copy.as_str()),
                text::LineEnding::default(),
                cx,
            )
        });
        let working_copy = working_copy.read_with(cx, |working_copy, _| working_copy.snapshot());
        let mut index_text = if rng.random() {
            Rope::from(head_text.as_str())
        } else {
            working_copy.as_rope().clone()
        };

        let mut diff = uncommitted_diff(&working_copy, &index_text, head_text.clone(), cx);
        let mut hunks = diff.update(cx, |diff, cx| {
            diff.hunks_intersecting_range(
                Anchor::min_max_range_for_buffer(diff.buffer_id),
                &working_copy,
                cx,
            )
            .collect::<Vec<_>>()
        });
        if hunks.is_empty() {
            return;
        }

        for _ in 0..operations {
            let i = rng.random_range(0..hunks.len());
            let hunk = &mut hunks[i];
            let hunk_to_change = hunk.clone();
            let stage = match hunk.secondary_status {
                DiffHunkSecondaryStatus::HasSecondaryHunk => {
                    hunk.secondary_status = DiffHunkSecondaryStatus::NoSecondaryHunk;
                    true
                }
                DiffHunkSecondaryStatus::NoSecondaryHunk => {
                    hunk.secondary_status = DiffHunkSecondaryStatus::HasSecondaryHunk;
                    false
                }
                _ => unreachable!(),
            };

            index_text = diff.update(cx, |diff, cx| {
                diff.stage_or_unstage_hunks(stage, &[hunk_to_change], &working_copy, true, cx)
                    .unwrap()
            });

            diff = uncommitted_diff(&working_copy, &index_text, head_text.clone(), cx);
            let found_hunks = diff.update(cx, |diff, cx| {
                diff.hunks_intersecting_range(
                    Anchor::min_max_range_for_buffer(diff.buffer_id),
                    &working_copy,
                    cx,
                )
                .collect::<Vec<_>>()
            });
            assert_eq!(hunks.len(), found_hunks.len());

            for (expected_hunk, found_hunk) in hunks.iter().zip(&found_hunks) {
                assert_eq!(
                    expected_hunk.buffer_range.to_point(&working_copy),
                    found_hunk.buffer_range.to_point(&working_copy)
                );
                assert_eq!(
                    expected_hunk.diff_base_byte_range,
                    found_hunk.diff_base_byte_range
                );
                assert_eq!(expected_hunk.secondary_status, found_hunk.secondary_status);
            }
            hunks = found_hunks;
        }
    }

    #[gpui::test]
    async fn test_row_to_base_text_row(cx: &mut TestAppContext) {
        let base_text = "
            zero
            one
            two
            three
            four
            five
            six
            seven
            eight
        "
        .unindent();
        let buffer_text = "
            zero
            ONE
            two
            NINE
            five
            seven
        "
        .unindent();

        //   zero
        // - one
        // + ONE
        //   two
        // - three
        // - four
        // + NINE
        //   five
        // - six
        //   seven
        // + eight

        let buffer = Buffer::new(ReplicaId::LOCAL, BufferId::new(1).unwrap(), buffer_text);
        let buffer_snapshot = buffer.snapshot();
        let diff = BufferDiffSnapshot::new_sync(buffer_snapshot.clone(), base_text, cx);
        let expected_results = [
            // don't format me
            (0, 0),
            (1, 2),
            (2, 2),
            (3, 5),
            (4, 5),
            (5, 7),
            (6, 9),
        ];
        for (buffer_row, expected) in expected_results {
            assert_eq!(
                diff.row_to_base_text_row(buffer_row, &buffer_snapshot),
                expected,
                "{buffer_row}"
            );
        }
    }
}