DiffMatchPatch.zig

const std = @import("std");
const testing = std.testing;
const Allocator = std.mem.Allocator;

const DiffMatchPatch = @This();

/// DMP with default configuration options
pub fn initDefault(io: std.Io, allocator: Allocator) DiffMatchPatch {
    return .{
        .io = io,
        .allocator = allocator,
    };
}

pub const Diff = struct {
    pub const Operation = enum {
        insert,
        delete,
        equal,
    };

    operation: Operation,
    text: []const u8,

    pub fn format(value: Diff, writer: *std.Io.Writer) std.Io.Writer.Error!void {
        try writer.print("({s}, \"{s}\")", .{
            switch (value.operation) {
                .equal => "=",
                .insert => "+",
                .delete => "-",
            },
            value.text,
        });
    }

    pub fn eql(a: Diff, b: Diff) bool {
        return a.operation == b.operation and std.mem.eql(u8, a.text, b.text);
    }

    test eql {
        const equal_a: Diff = .{ .operation = .equal, .text = "a" };
        const insert_a: Diff = .{ .operation = .insert, .text = "a" };
        const equal_b: Diff = .{ .operation = .equal, .text = "b" };
        const delete_b: Diff = .{ .operation = .delete, .text = "b" };

        try testing.expect(equal_a.eql(equal_a));
        try testing.expect(!insert_a.eql(equal_a));
        try testing.expect(!equal_a.eql(equal_b));
        try testing.expect(!equal_a.eql(delete_b));
    }
};

io: std.Io,
allocator: Allocator,

/// Cost of an empty edit operation in terms of edit characters.
diff_edit_cost: u16 = 4,
/// Number of bytes in each string needed to trigger a line-based diff
diff_check_lines_over: u64 = 100,

/// At what point is no match declared (0.0 = perfection, 1.0 = very loose).
match_threshold: f32 = 0.5,
/// How far to search for a match (0 = exact location, 1000+ = broad match).
/// A match this many characters away from the expected location will add
/// 1.0 to the score (0.0 is a perfect match).
match_distance: u32 = 1000,
/// The number of bits in an int.
match_max_bits: u16 = 32,

/// When deleting a large block of text (over ~64 characters), how close
/// do the contents have to be to match the expected contents. (0.0 =
/// perfection, 1.0 = very loose).  Note that Match_Threshold controls
/// how closely the end points of a delete need to match.
patch_delete_threshold: f32 = 0.5,
/// Chunk size for context length.
patch_margin: u16 = 4,

pub const DiffError = error{OutOfMemory};

/// Find the differences between two texts.  The return value
/// must be freed with `deinitDiffList(allocator, &diffs)`.
/// @param before Old string to be diffed.
/// @param after New string to be diffed.
/// @param checklines Speedup flag.  If false, then don't run a
///     line-level diff first to identify the changed areas.
///     If true, then run a faster slightly less optimal diff.
/// @return List of Diff objects.
pub fn diff(
    dmp: *const DiffMatchPatch,
    before: []const u8,
    after: []const u8,
    /// If false, then don't run a line-level diff first
    /// to identify the changed areas. If true, then run
    /// a faster slightly less optimal diff.
    check_lines: bool,
    /// Duration to map a diff before falling back to a non-minimal diff. Use `std.Io.Duration.max` for disable timeouts.
    timeout: std.Io.Timeout,
) DiffError!DiffList {
    return dmp.diffInternal(
        before,
        after,
        check_lines,
        timeout.toTimestamp(dmp.io),
    );
}

const DiffList = std.ArrayList(Diff);

/// Deinit an `std.ArrayList(Diff)` and the allocated slices of
/// text in each `Diff`.
pub fn deinitDiffList(allocator: Allocator, diffs: *DiffList) void {
    defer diffs.deinit(allocator);
    for (diffs.items) |d| {
        allocator.free(d.text);
    }
}

fn freeRangeDiffList(
    allocator: Allocator,
    diffs: *DiffList,
    start: usize,
    len: usize,
) void {
    const range = diffs.items[start..][0..len];
    for (range) |d| {
        allocator.free(d.text);
    }
}

fn diffInternal(
    dmp: *const DiffMatchPatch,
    before: []const u8,
    after: []const u8,
    check_lines: bool,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!DiffList {
    const allocator = dmp.allocator;
    // Trim off common prefix (speedup).
    const common_prefix_length = std.mem.findDiff(u8, before, after) orelse {
        // equality
        var diffs: DiffList = .empty;
        errdefer deinitDiffList(allocator, &diffs);
        if (before.len != 0) {
            try diffs.ensureUnusedCapacity(allocator, 1);
            diffs.appendAssumeCapacity(.{
                .operation = .equal,
                .text = try allocator.dupe(u8, before),
            });
        }
        return diffs;
    };

    const common_prefix = before[0..common_prefix_length];
    var trimmed_before = before[common_prefix_length..];
    var trimmed_after = after[common_prefix_length..];

    // Trim off common suffix (speedup).
    const common_suffix_length = diffCommonSuffix(trimmed_before, trimmed_after);
    const common_suffix = trimmed_before[trimmed_before.len - common_suffix_length ..];
    trimmed_before = trimmed_before[0 .. trimmed_before.len - common_suffix_length];
    trimmed_after = trimmed_after[0 .. trimmed_after.len - common_suffix_length];

    // Compute the diff on the middle block.
    var diffs = try dmp.diffCompute(trimmed_before, trimmed_after, check_lines, deadline);
    errdefer deinitDiffList(allocator, &diffs);

    // Restore the prefix and suffix.

    if (common_prefix.len != 0) {
        try diffs.ensureUnusedCapacity(allocator, 1);
        diffs.insertAssumeCapacity(0, .{
            .operation = .equal,
            .text = try allocator.dupe(u8, common_prefix),
        });
    }
    if (common_suffix.len != 0) {
        try diffs.ensureUnusedCapacity(allocator, 1);
        diffs.appendAssumeCapacity(.{
            .operation = .equal,
            .text = try allocator.dupe(u8, common_suffix),
        });
    }

    try diffCleanupMerge(allocator, &diffs);
    return diffs;
}

fn indexOfDiff(comptime T: type, a: []const T, b: []const T) ?usize {
    const shortest = @min(a.len, b.len);
    for (a[0..shortest], b[0..shortest], 0..) |a_char, b_char, index| {
        if (a_char != b_char) return index;
    }
    return if (a.len == b.len) null else shortest;
}

fn diffCommonPrefix(before: []const u8, after: []const u8) usize {
    return indexOfDiff(u8, before, after) orelse @min(before.len, after.len);
}

fn diffCommonSuffix(before: []const u8, after: []const u8) usize {
    const n = @min(before.len, after.len);
    var i: usize = 1;

    while (i <= n) : (i += 1) {
        if (before[before.len - i] != after[after.len - i]) {
            return i - 1;
        }
    }

    return n;
}

/// Find the differences between two texts.  Assumes that the texts do not
/// have any common prefix or suffix.
/// @param before Old string to be diffed.
/// @param after New string to be diffed.
/// @param checklines Speedup flag.  If false, then don't run a
///     line-level diff first to identify the changed areas.
///     If true, then run a faster slightly less optimal diff.
/// @param timer Tracks time since diffing has started to detect a timeout.
/// @return List of Diff objects.
fn diffCompute(
    dmp: *const DiffMatchPatch,
    before: []const u8,
    after: []const u8,
    check_lines: bool,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!DiffList {
    const allocator = dmp.allocator;
    if (before.len == 0) {
        // Just add some text (speedup).
        var diffs: DiffList = .empty;
        errdefer deinitDiffList(allocator, &diffs);
        try diffs.ensureUnusedCapacity(allocator, 1);
        diffs.appendAssumeCapacity(.{
            .operation = .insert,
            .text = try allocator.dupe(u8, after),
        });
        return diffs;
    }

    if (after.len == 0) {
        // Just delete some text (speedup).
        var diffs: DiffList = .empty;
        errdefer deinitDiffList(allocator, &diffs);
        try diffs.ensureUnusedCapacity(allocator, 1);
        diffs.appendAssumeCapacity(.{
            .operation = .delete,
            .text = try allocator.dupe(u8, before),
        });
        return diffs;
    }

    const long_text = if (before.len > after.len) before else after;
    const short_text = if (before.len > after.len) after else before;

    if (std.mem.find(u8, long_text, short_text)) |index| {
        var diffs: DiffList = .empty;
        errdefer deinitDiffList(allocator, &diffs);
        // Shorter text is inside the longer text (speedup).
        const op: Diff.Operation = if (before.len > after.len)
            .delete
        else
            .insert;
        try diffs.ensureUnusedCapacity(allocator, 3);
        diffs.appendAssumeCapacity(.{
            .operation = op,
            .text = try allocator.dupe(u8, long_text[0..index]),
        });
        diffs.appendAssumeCapacity(.{
            .operation = .equal,
            .text = try allocator.dupe(u8, short_text),
        });
        diffs.appendAssumeCapacity(.{
            .operation = op,
            .text = try allocator.dupe(u8, long_text[index + short_text.len ..]),
        });
        return diffs;
    }

    if (short_text.len == 1) {
        // Single character string.
        // After the previous speedup, the character can't be an equality.
        var diffs: DiffList = .empty;
        errdefer deinitDiffList(allocator, &diffs);
        try diffs.ensureUnusedCapacity(allocator, 2);
        diffs.appendAssumeCapacity(.{
            .operation = .delete,
            .text = try allocator.dupe(u8, before),
        });
        diffs.appendAssumeCapacity(.{
            .operation = .insert,
            .text = try allocator.dupe(u8, after),
        });
        return diffs;
    }

    // Check to see if the problem can be split in two.
    if (try dmp.diffHalfMatch(before, after, deadline)) |half_match| {
        // A half-match was found, sort out the return data.
        defer half_match.deinit(allocator);
        // Send both pairs off for separate processing.
        var diffs = try dmp.diffInternal(
            half_match.prefix_before,
            half_match.prefix_after,
            check_lines,
            deadline,
        );
        errdefer deinitDiffList(allocator, &diffs);
        var diffs_b = try dmp.diffInternal(
            half_match.suffix_before,
            half_match.suffix_after,
            check_lines,
            deadline,
        );
        defer diffs_b.deinit(allocator);
        // we have to deinit regardless, so deinitDiffList would be
        // a double free:
        errdefer {
            for (diffs_b.items) |d| {
                allocator.free(d.text);
            }
        }

        // Merge the results.
        try diffs.ensureUnusedCapacity(allocator, 1);
        diffs.appendAssumeCapacity(.{
            .operation = .equal,
            .text = try allocator.dupe(u8, half_match.common_middle),
        });
        try diffs.appendSlice(allocator, diffs_b.items);
        return diffs;
    }
    if (check_lines and before.len > dmp.diff_check_lines_over and after.len > dmp.diff_check_lines_over) {
        return dmp.diffLineMode(before, after, deadline);
    }

    return dmp.diffBisect(before, after, deadline);
}

const HalfMatchResult = struct {
    prefix_before: []const u8,
    suffix_before: []const u8,
    prefix_after: []const u8,
    suffix_after: []const u8,
    common_middle: []const u8,

    pub fn deinit(hmr: HalfMatchResult, alloc: Allocator) void {
        alloc.free(hmr.prefix_before);
        alloc.free(hmr.suffix_before);
        alloc.free(hmr.prefix_after);
        alloc.free(hmr.suffix_after);
        alloc.free(hmr.common_middle);
    }
};

/// Do the two texts share a Substring which is at least half the length of
/// the longer text?
/// This speedup can produce non-minimal diffs.
/// @param before First string.
/// @param after Second string.
/// @return Five element String array, containing the prefix of text1, the
///     suffix of text1, the prefix of text2, the suffix of text2 and the
///     common middle.  Or null if there was no match.
fn diffHalfMatch(
    dmp: *const DiffMatchPatch,
    before: []const u8,
    after: []const u8,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!?HalfMatchResult {
    const allocator = dmp.allocator;
    if (deadline == null) {
        // Don't risk returning a non-optimal diff if we have unlimited time.
        return null;
    }
    const long_text = if (before.len > after.len) before else after;
    const short_text = if (before.len > after.len) after else before;

    if (long_text.len < 4 or short_text.len * 2 < long_text.len) {
        return null; // Pointless.
    }

    // First check if the second quarter is the seed for a half-match.
    const half_match_1 = try dmp.diffHalfMatchInternal(long_text, short_text, (long_text.len + 3) / 4);
    errdefer {
        if (half_match_1) |h_m| h_m.deinit(allocator);
    }
    // Check again based on the third quarter.
    const half_match_2 = try dmp.diffHalfMatchInternal(long_text, short_text, (long_text.len + 1) / 2);
    errdefer {
        if (half_match_2) |h_m| h_m.deinit(allocator);
    }

    var half_match: ?HalfMatchResult = null;
    if (half_match_1 == null and half_match_2 == null) {
        return null;
    } else if (half_match_2 == null) {
        half_match = half_match_1.?;
    } else if (half_match_1 == null) {
        half_match = half_match_2.?;
    } else {
        // Both matched. Select the longest.
        half_match = half: {
            if (half_match_1.?.common_middle.len > half_match_2.?.common_middle.len) {
                half_match_2.?.deinit(allocator);
                break :half half_match_1;
            } else {
                half_match_1.?.deinit(allocator);
                break :half half_match_2;
            }
        };
    }

    // A half-match was found, sort out the return data.
    if (before.len > after.len) {
        return half_match.?;
    } else {
        // Transfers ownership of all memory to new, permuted, half_match.
        const half_match_yes = half_match.?;
        return .{
            .prefix_before = half_match_yes.prefix_after,
            .suffix_before = half_match_yes.suffix_after,
            .prefix_after = half_match_yes.prefix_before,
            .suffix_after = half_match_yes.suffix_before,
            .common_middle = half_match_yes.common_middle,
        };
    }
}

/// Does a Substring of shorttext exist within longtext such that the
/// Substring is at least half the length of longtext?
/// @param longtext Longer string.
/// @param shorttext Shorter string.
/// @param i Start index of quarter length Substring within longtext.
/// @return Five element string array, containing the prefix of longtext, the
///     suffix of longtext, the prefix of shorttext, the suffix of shorttext
///     and the common middle.  Or null if there was no match.
fn diffHalfMatchInternal(
    dmp: *const DiffMatchPatch,
    long_text: []const u8,
    short_text: []const u8,
    i: usize,
) DiffError!?HalfMatchResult {
    const allocator = dmp.allocator;
    // Start with a 1/4 length Substring at position i as a seed.
    const seed = long_text[i .. i + long_text.len / 4];
    var j: isize = -1;

    var best_common: std.ArrayList(u8) = .empty;
    defer best_common.deinit(allocator);
    var best_long_text_a: []const u8 = "";
    var best_long_text_b: []const u8 = "";
    var best_short_text_a: []const u8 = "";
    var best_short_text_b: []const u8 = "";

    while (j < short_text.len and b: {
        j = @as(isize, @intCast(std.mem.find(u8, short_text[@intCast(j + 1)..], seed) orelse break :b false)) + j + 1;
        break :b true;
    }) {
        const prefix_length = diffCommonPrefix(long_text[i..], short_text[@intCast(j)..]);
        const suffix_length = diffCommonSuffix(long_text[0..i], short_text[0..@intCast(j)]);
        if (best_common.items.len < suffix_length + prefix_length) {
            best_common.clearRetainingCapacity();
            const a = short_text[@intCast(j - @as(isize, @intCast(suffix_length))) .. @as(usize, @intCast(j - @as(isize, @intCast(suffix_length)))) + suffix_length];
            try best_common.appendSlice(allocator, a);
            const b = short_text[@intCast(j) .. @as(usize, @intCast(j)) + prefix_length];
            try best_common.appendSlice(allocator, b);

            best_long_text_a = long_text[0 .. i - suffix_length];
            best_long_text_b = long_text[i + prefix_length ..];
            best_short_text_a = short_text[0..@intCast(j - @as(isize, @intCast(suffix_length)))];
            best_short_text_b = short_text[@intCast(j + @as(isize, @intCast(prefix_length)))..];
        }
    }
    if (best_common.items.len * 2 >= long_text.len) {
        const prefix_before = try allocator.dupe(u8, best_long_text_a);
        errdefer allocator.free(prefix_before);
        const suffix_before = try allocator.dupe(u8, best_long_text_b);
        errdefer allocator.free(suffix_before);
        const prefix_after = try allocator.dupe(u8, best_short_text_a);
        errdefer allocator.free(prefix_after);
        const suffix_after = try allocator.dupe(u8, best_short_text_b);
        errdefer allocator.free(suffix_after);
        const best_common_text = try best_common.toOwnedSlice(allocator);
        errdefer allocator.free(best_common_text);
        return .{
            .prefix_before = prefix_before,
            .suffix_before = suffix_before,
            .prefix_after = prefix_after,
            .suffix_after = suffix_after,
            .common_middle = best_common_text,
        };
    } else {
        return null;
    }
}

/// Find the 'middle snake' of a diff, split the problem in two
/// and return the recursively constructed diff.
/// See Myers 1986 paper: An O(ND) Difference Algorithm and Its Variations.
/// @param before Old string to be diffed.
/// @param after New string to be diffed.
/// @param timer Tracks time since diffing has started to detect a timeout.
/// @return List of Diff objects.
fn diffBisect(
    dmp: *const DiffMatchPatch,
    before: []const u8,
    after: []const u8,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!DiffList {
    const allocator = dmp.allocator;
    const before_length: isize = @intCast(before.len);
    const after_length: isize = @intCast(after.len);
    const max_d: isize = @intCast((before.len + after.len + 1) / 2);
    const v_offset = max_d;
    const v_length = 2 * max_d;

    var v1: std.ArrayList(isize) = try .initCapacity(allocator, @intCast(v_length));
    defer v1.deinit(allocator);
    v1.items.len = @intCast(v_length);
    var v2: std.ArrayList(isize) = try .initCapacity(allocator, @intCast(v_length));
    defer v2.deinit(allocator);
    v2.items.len = @intCast(v_length);

    var x: usize = 0;
    while (x < v_length) : (x += 1) {
        v1.items[x] = -1;
        v2.items[x] = -1;
    }
    v1.items[@intCast(v_offset + 1)] = 0;
    v2.items[@intCast(v_offset + 1)] = 0;
    const delta = before_length - after_length;
    // If the total number of characters is odd, then the front path will
    // collide with the reverse path.
    const front = (@mod(delta, 2) != 0);
    // Offsets for start and end of k loop.
    // Prevents mapping of space beyond the grid.
    var k1start: isize = 0;
    var k1end: isize = 0;
    var k2start: isize = 0;
    var k2end: isize = 0;

    var d: isize = 0;
    while (d < max_d) : (d += 1) {
        // Bail out if deadline is reached.
        if (deadline) |t| {
            const duration = t.untilNow(dmp.io);
            if (duration.raw.toNanoseconds() > 0) break; // Bail out if deadline is reached.
        }

        // Walk the front path one step.
        var k1 = -d + k1start;
        while (k1 <= d - k1end) : (k1 += 2) {
            const k1_offset = v_offset + k1;
            var x1: isize = 0;
            if (k1 == -d or (k1 != d and
                v1.items[@intCast(k1_offset - 1)] < v1.items[@intCast(k1_offset + 1)]))
            {
                x1 = v1.items[@intCast(k1_offset + 1)];
            } else {
                x1 = v1.items[@intCast(k1_offset - 1)] + 1;
            }
            var y1 = x1 - k1;
            while (x1 < before_length and
                y1 < after_length and before[@intCast(x1)] == after[@intCast(y1)])
            {
                x1 += 1;
                y1 += 1;
            }
            v1.items[@intCast(k1_offset)] = x1;
            if (x1 > before_length) {
                // Ran off the right of the graph.
                k1end += 2;
            } else if (y1 > after_length) {
                // Ran off the bottom of the graph.
                k1start += 2;
            } else if (front) {
                const k2_offset = v_offset + delta - k1;
                if (k2_offset >= 0 and k2_offset < v_length and v2.items[@intCast(k2_offset)] != -1) {
                    // Mirror x2 onto top-left coordinate system.
                    const x2 = before_length - v2.items[@intCast(k2_offset)];
                    if (x1 >= x2) {
                        // Overlap detected.
                        return dmp.diffBisectSplit(before, after, x1, y1, deadline);
                    }
                }
            }
        }

        // Walk the reverse path one step.
        var k2: isize = -d + k2start;
        while (k2 <= d - k2end) : (k2 += 2) {
            const k2_offset = v_offset + k2;
            var x2: isize = 0;
            if (k2 == -d or (k2 != d and
                v2.items[@intCast(k2_offset - 1)] < v2.items[@intCast(k2_offset + 1)]))
            {
                x2 = v2.items[@intCast(k2_offset + 1)];
            } else {
                x2 = v2.items[@intCast(k2_offset - 1)] + 1;
            }
            var y2: isize = x2 - k2;
            while (x2 < before_length and y2 < after_length and
                before[@intCast(before_length - x2 - 1)] ==
                    after[@intCast(after_length - y2 - 1)])
            {
                x2 += 1;
                y2 += 1;
            }
            v2.items[@intCast(k2_offset)] = x2;
            if (x2 > before_length) {
                // Ran off the left of the graph.
                k2end += 2;
            } else if (y2 > after_length) {
                // Ran off the top of the graph.
                k2start += 2;
            } else if (!front) {
                const k1_offset = v_offset + delta - k2;
                if (k1_offset >= 0 and k1_offset < v_length and v1.items[@intCast(k1_offset)] != -1) {
                    const x1 = v1.items[@intCast(k1_offset)];
                    const y1 = v_offset + x1 - k1_offset;
                    // Mirror x2 onto top-left coordinate system.
                    x2 = before_length - v2.items[@intCast(k2_offset)];
                    if (x1 >= x2) {
                        // Overlap detected.
                        return dmp.diffBisectSplit(before, after, x1, y1, deadline);
                    }
                }
            }
        }
    }
    // Diff took too long and hit the deadline or
    // number of diffs equals number of characters, no commonality at all.
    var diffs: DiffList = .empty;
    errdefer deinitDiffList(allocator, &diffs);
    try diffs.ensureUnusedCapacity(allocator, 2);
    diffs.appendAssumeCapacity(.{
        .operation = .delete,
        .text = try allocator.dupe(u8, before),
    });
    diffs.appendAssumeCapacity(.{
        .operation = .insert,
        .text = try allocator.dupe(u8, after),
    });
    return diffs;
}

/// Given the location of the 'middle snake', split the diff in two parts
/// and recurse.
/// @param text1 Old string to be diffed.
/// @param text2 New string to be diffed.
/// @param x Index of split point in text1.
/// @param y Index of split point in text2.
/// @param timer Tracks time since diffing has started to detect a timeout.
/// @return LinkedList of Diff objects.
fn diffBisectSplit(
    dmp: *const DiffMatchPatch,
    text1: []const u8,
    text2: []const u8,
    x: isize,
    y: isize,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!DiffList {
    const allocator = dmp.allocator;
    const text1a = text1[0..@intCast(x)];
    const text2a = text2[0..@intCast(y)];
    const text1b = text1[@intCast(x)..];
    const text2b = text2[@intCast(y)..];

    // Compute both diffs serially.
    var diffs = try dmp.diffInternal(text1a, text2a, false, deadline);
    errdefer deinitDiffList(allocator, &diffs);
    var diffs_b = try dmp.diffInternal(text1b, text2b, false, deadline);
    // Free the list, but not the contents:
    defer diffs_b.deinit(allocator);
    errdefer {
        for (diffs_b.items) |d| {
            allocator.free(d.text);
        }
    }
    try diffs.appendSlice(allocator, diffs_b.items);
    return diffs;
}

/// Do a quick line-level diff on both strings, then rediff the parts for
/// greater accuracy.
/// This speedup can produce non-minimal diffs.
/// @param text1 Old string to be diffed.
/// @param text2 New string to be diffed.
/// @param timer Tracks time since diffing has started to detect a timeout.
/// @return List of Diff objects.
fn diffLineMode(
    dmp: *const DiffMatchPatch,
    text1_in: []const u8,
    text2_in: []const u8,
    deadline: ?std.Io.Clock.Timestamp,
) DiffError!DiffList {
    const allocator = dmp.allocator;
    // Scan the text on a line-by-line basis first.
    var a = try diffLinesToChars(allocator, text1_in, text2_in);
    defer a.deinit(allocator);
    const text1 = a.chars_1;
    const text2 = a.chars_2;
    const line_array = a.line_array;
    var diffs: DiffList = undefined;
    {
        var char_diffs: DiffList = try dmp.diffInternal(text1, text2, false, deadline);
        defer deinitDiffList(allocator, &char_diffs);
        // Convert the diff back to original text.
        diffs = try diffCharsToLines(allocator, &char_diffs, line_array.items);
        // Eliminate freak matches (e.g. blank lines)
    }
    errdefer deinitDiffList(allocator, &diffs);
    try diffCleanupSemantic(allocator, &diffs);

    // Rediff any replacement blocks, this time character-by-character.
    // Add a dummy entry at the end.
    try diffs.append(allocator, .{ .operation = .equal, .text = "" });

    var pointer: usize = 0;
    var count_delete: usize = 0;
    var count_insert: usize = 0;
    var text_delete: std.ArrayList(u8) = .empty;
    var text_insert: std.ArrayList(u8) = .empty;
    defer {
        text_delete.deinit(allocator);
        text_insert.deinit(allocator);
    }

    while (pointer < diffs.items.len) {
        switch (diffs.items[pointer].operation) {
            .insert => {
                count_insert += 1;
                try text_insert.appendSlice(allocator, diffs.items[pointer].text);
            },
            .delete => {
                count_delete += 1;
                try text_delete.appendSlice(allocator, diffs.items[pointer].text);
            },
            .equal => {
                // Upon reaching an equality, check for prior redundancies.
                if (count_delete >= 1 and count_insert >= 1) {
                    // Delete the offending records and add the merged ones.
                    freeRangeDiffList(
                        allocator,
                        &diffs,
                        pointer - count_delete - count_insert,
                        count_delete + count_insert,
                    );
                    diffs.replaceRangeAssumeCapacity(
                        pointer - count_delete - count_insert,
                        count_delete + count_insert,
                        &.{},
                    );
                    pointer = pointer - count_delete - count_insert;
                    var sub_diff = try dmp.diffInternal(text_delete.items, text_insert.items, false, deadline);
                    {
                        errdefer deinitDiffList(allocator, &sub_diff);
                        try diffs.ensureUnusedCapacity(allocator, sub_diff.items.len);
                    }
                    defer sub_diff.deinit(allocator);
                    const new_diff = diffs.addManyAtAssumeCapacity(pointer, sub_diff.items.len);
                    @memcpy(new_diff, sub_diff.items);
                    pointer = pointer + sub_diff.items.len;
                }
                count_insert = 0;
                count_delete = 0;
                text_delete.clearRetainingCapacity();
                text_insert.clearRetainingCapacity();
            },
        }
        pointer += 1;
    }
    diffs.items.len -= 1; // Remove the dummy entry at the end.

    return diffs;
}

const LinesToCharsResult = struct {
    chars_1: []const u8,
    chars_2: []const u8,
    line_array: std.ArrayList([]const u8),

    pub fn deinit(self: *LinesToCharsResult, allocator: Allocator) void {
        allocator.free(self.chars_1);
        allocator.free(self.chars_2);
        self.line_array.deinit(allocator);
    }
};

/// Split two texts into a list of strings.  Reduce the texts to a string of
/// hashes where each Unicode character represents one line.
/// @param text1 First string.
/// @param text2 Second string.
/// @return Three element Object array, containing the encoded text1, the
///     encoded text2 and the List of unique strings.  The zeroth element
///     of the List of unique strings is intentionally blank.
fn diffLinesToChars(
    allocator: Allocator,
    text1: []const u8,
    text2: []const u8,
) DiffError!LinesToCharsResult {
    var line_array: std.ArrayList([]const u8) = .empty;
    errdefer line_array.deinit(allocator);
    var line_hash: std.StringHashMapUnmanaged(usize) = .empty;
    defer line_hash.deinit(allocator);
    // e.g. line_array[4] == "Hello\n"
    // e.g. line_hash.get("Hello\n") == 4

    // "\x00" is a valid character, but various debuggers don't like it.
    // So we'll insert a junk entry to avoid generating a null character.
    try line_array.append(allocator, "");

    // Allocate 2/3rds of the space for text1, the rest for text2.
    const chars1 = try diffLinesToCharsMunge(allocator, text1, &line_array, &line_hash, 170);
    errdefer allocator.free(chars1);
    const chars2 = try diffLinesToCharsMunge(allocator, text2, &line_array, &line_hash, 255);
    return .{ .chars_1 = chars1, .chars_2 = chars2, .line_array = line_array };
}

/// Split a text into a list of strings.  Reduce the texts to a string of
/// hashes where each Unicode character represents one line.
/// @param text String to encode.
/// @param lineArray List of unique strings.
/// @param lineHash Map of strings to indices.
/// @param maxLines Maximum length of lineArray.
/// @return Encoded string.
fn diffLinesToCharsMunge(
    allocator: Allocator,
    text: []const u8,
    line_array: *std.ArrayList([]const u8),
    line_hash: *std.StringHashMapUnmanaged(usize),
    max_lines: usize,
) DiffError![]const u8 {
    var line_start: isize = 0;
    var line_end: isize = -1;
    var chars: std.ArrayList(u8) = .empty;
    defer chars.deinit(allocator);
    // Walk the text, pulling out a Substring for each line.
    // TODO this can be handled with a Reader, avoiding all the manual splitting
    while (line_end < @as(isize, @intCast(text.len)) - 1) {
        line_end = b: {
            break :b @as(isize, @intCast(std.mem.find(u8, text[@intCast(line_start)..], "\n") orelse
                break :b @intCast(text.len - 1))) + line_start;
        };
        var line = text[@intCast(line_start) .. @as(usize, @intCast(line_start)) + @as(usize, @intCast(line_end + 1 - line_start))];

        if (line_hash.get(line)) |value| {
            try chars.append(allocator, @intCast(value));
        } else {
            if (line_array.items.len == max_lines) {
                // Bail out at 255 because char 256 == char 0.
                line = text[@intCast(line_start)..];
                line_end = @intCast(text.len);
            }
            try line_array.append(allocator, line);
            try line_hash.put(allocator, line, line_array.items.len - 1);
            try chars.append(allocator, @intCast(line_array.items.len - 1));
        }
        line_start = line_end + 1;
    }
    return try chars.toOwnedSlice(allocator);
}

/// Rehydrate the text in a diff from a string of line hashes to real lines
/// of text.
/// @param diffs List of Diff objects.
/// @param lineArray List of unique strings.
fn diffCharsToLines(
    allocator: Allocator,
    char_diffs: *DiffList,
    line_array: []const []const u8,
) DiffError!DiffList {
    var diffs: DiffList = .empty;
    errdefer deinitDiffList(allocator, &diffs);
    try diffs.ensureTotalCapacity(allocator, char_diffs.items.len);
    var text: std.ArrayList(u8) = .empty;
    defer text.deinit(allocator);

    for (char_diffs.items) |*d| {
        var j: usize = 0;
        while (j < d.text.len) : (j += 1) {
            try text.appendSlice(allocator, line_array[d.text[j]]);
        }
        diffs.appendAssumeCapacity(.{
            .operation = d.operation,
            .text = try text.toOwnedSlice(allocator),
        });
    }
    return diffs;
}

/// Reorder and merge like edit sections.  Merge equalities.
/// Any edit section can move as long as it doesn't cross an equality.
/// @param diffs List of Diff objects.
fn diffCleanupMerge(allocator: Allocator, diffs: *DiffList) DiffError!void {
    // Add a dummy entry at the end.
    try diffs.append(allocator, .{ .operation = .equal, .text = "" });
    var pointer: usize = 0;
    var count_delete: usize = 0;
    var count_insert: usize = 0;

    var text_delete: std.ArrayList(u8) = .empty;
    defer text_delete.deinit(allocator);

    var text_insert: std.ArrayList(u8) = .empty;
    defer text_insert.deinit(allocator);

    var common_length: usize = undefined;
    while (pointer < diffs.items.len) {
        switch (diffs.items[pointer].operation) {
            .insert => {
                count_insert += 1;
                try text_insert.appendSlice(allocator, diffs.items[pointer].text);
                pointer += 1;
            },
            .delete => {
                count_delete += 1;
                try text_delete.appendSlice(allocator, diffs.items[pointer].text);
                pointer += 1;
            },
            .equal => {
                // Upon reaching an equality, check for prior redundancies.
                if (count_delete + count_insert > 1) {
                    if (count_delete != 0 and count_insert != 0) {
                        // Factor out any common prefixies.
                        common_length = diffCommonPrefix(text_insert.items, text_delete.items);
                        if (common_length != 0) {
                            if ((pointer - count_delete - count_insert) > 0 and
                                diffs.items[pointer - count_delete - count_insert - 1].operation == .equal)
                            {
                                const ii = pointer - count_delete - count_insert - 1;
                                var nt = try allocator.alloc(u8, diffs.items[ii].text.len + common_length);
                                const ot = diffs.items[ii].text;
                                @memcpy(nt[0..ot.len], ot);
                                @memcpy(nt[ot.len..], text_insert.items[0..common_length]);
                                diffs.items[ii].text = nt;
                                allocator.free(ot);
                            } else {
                                try diffs.ensureUnusedCapacity(allocator, 1);
                                const text = try allocator.dupe(u8, text_insert.items[0..common_length]);
                                diffs.insertAssumeCapacity(0, .{ .operation = .equal, .text = text });
                                pointer += 1;
                            }
                            text_insert.replaceRangeAssumeCapacity(0, common_length, &.{});
                            text_delete.replaceRangeAssumeCapacity(0, common_length, &.{});
                        }
                        // Factor out any common suffixies.
                        // @ZigPort this seems very wrong
                        common_length = diffCommonSuffix(text_insert.items, text_delete.items);
                        if (common_length != 0) {
                            const old_text = diffs.items[pointer].text;
                            diffs.items[pointer].text = try std.mem.concat(allocator, u8, &.{
                                text_insert.items[text_insert.items.len - common_length ..],
                                old_text,
                            });
                            allocator.free(old_text);
                            text_insert.items.len -= common_length;
                            text_delete.items.len -= common_length;
                        }
                    }
                    // Delete the offending records and add the merged ones.
                    pointer -= count_delete + count_insert;
                    if (count_delete + count_insert > 0) {
                        freeRangeDiffList(allocator, diffs, pointer, count_delete + count_insert);
                        diffs.replaceRangeAssumeCapacity(pointer, count_delete + count_insert, &.{});
                    }

                    if (text_delete.items.len != 0) {
                        try diffs.ensureUnusedCapacity(allocator, 1);
                        diffs.insertAssumeCapacity(pointer, .{
                            .operation = .delete,
                            .text = try allocator.dupe(u8, text_delete.items),
                        });
                        pointer += 1;
                    }
                    if (text_insert.items.len != 0) {
                        try diffs.ensureUnusedCapacity(allocator, 1);
                        diffs.insertAssumeCapacity(pointer, .{
                            .operation = .insert,
                            .text = try allocator.dupe(u8, text_insert.items),
                        });
                        pointer += 1;
                    }
                    pointer += 1;
                } else if (pointer != 0 and diffs.items[pointer - 1].operation == .equal) {
                    // Merge this equality with the previous one.
                    // TODO: Fix using realloc or smth
                    // Note: can't use realloc because the text is const
                    var nt = try allocator.alloc(u8, diffs.items[pointer - 1].text.len + diffs.items[pointer].text.len);
                    const ot = diffs.items[pointer - 1].text;
                    defer (allocator.free(ot));
                    @memcpy(nt[0..ot.len], ot);
                    @memcpy(nt[ot.len..], diffs.items[pointer].text);
                    diffs.items[pointer - 1].text = nt;
                    const dead_diff = diffs.orderedRemove(pointer);
                    allocator.free(dead_diff.text);
                } else {
                    pointer += 1;
                }
                count_insert = 0;
                count_delete = 0;
                text_delete.clearRetainingCapacity();
                text_insert.clearRetainingCapacity();
            },
        }
    }
    if (diffs.items[diffs.items.len - 1].text.len == 0) {
        diffs.items.len -= 1;
    }

    // Second pass: look for single edits surrounded on both sides by
    // equalities which can be shifted sideways to eliminate an equality.
    // e.g: A<ins>BA</ins>C -> <ins>AB</ins>AC
    var changes = false;
    pointer = 1;
    // Intentionally ignore the first and last element (don't need checking).
    while (pointer < (diffs.items.len - 1)) {
        if (diffs.items[pointer - 1].operation == .equal and
            diffs.items[pointer + 1].operation == .equal)
        {
            // This is a single edit surrounded by equalities.
            if (std.mem.endsWith(u8, diffs.items[pointer].text, diffs.items[pointer - 1].text)) {
                const old_pt = diffs.items[pointer].text;
                const pt = try std.mem.concat(allocator, u8, &.{
                    diffs.items[pointer - 1].text,
                    diffs.items[pointer].text[0 .. diffs.items[pointer].text.len -
                        diffs.items[pointer - 1].text.len],
                });
                allocator.free(old_pt);
                diffs.items[pointer].text = pt;
                const old_pt1t = diffs.items[pointer + 1].text;
                const p1t = try std.mem.concat(allocator, u8, &.{
                    diffs.items[pointer - 1].text,
                    diffs.items[pointer + 1].text,
                });
                allocator.free(old_pt1t);
                diffs.items[pointer + 1].text = p1t;
                freeRangeDiffList(allocator, diffs, pointer - 1, 1);
                diffs.replaceRangeAssumeCapacity(pointer - 1, 1, &.{});
                changes = true;
            } else if (std.mem.startsWith(u8, diffs.items[pointer].text, diffs.items[pointer + 1].text)) {
                const old_ptm1 = diffs.items[pointer - 1].text;
                const pm1t = try std.mem.concat(allocator, u8, &.{
                    diffs.items[pointer - 1].text,
                    diffs.items[pointer + 1].text,
                });
                allocator.free(old_ptm1);
                diffs.items[pointer - 1].text = pm1t;
                const old_pt = diffs.items[pointer].text;
                const pt = try std.mem.concat(allocator, u8, &.{
                    diffs.items[pointer].text[diffs.items[pointer + 1].text.len..],
                    diffs.items[pointer + 1].text,
                });
                allocator.free(old_pt);
                diffs.items[pointer].text = pt;
                freeRangeDiffList(allocator, diffs, pointer + 1, 1);
                diffs.replaceRangeAssumeCapacity(pointer + 1, 1, &.{});
                changes = true;
            }
        }
        pointer += 1;
    }
    // If shifts were made, the diff needs reordering and another shift sweep.
    if (changes) {
        try diffCleanupMerge(allocator, diffs);
    }
}

/// Reduce the number of edits by eliminating semantically trivial
/// equalities.
/// @param diffs List of Diff objects.
pub fn diffCleanupSemantic(allocator: Allocator, diffs: *DiffList) DiffError!void {
    var changes = false;
    // Stack of indices where equalities are found.
    var equalities: std.ArrayList(isize) = .empty;
    defer equalities.deinit(allocator);
    // Always equal to equalities[equalitiesLength-1][1]
    var last_equality: ?[]const u8 = null;
    var pointer: isize = 0; // Index of current position.
    // Number of characters that changed prior to the equality.
    var length_insertions1: usize = 0;
    var length_deletions1: usize = 0;
    // Number of characters that changed after the equality.
    var length_insertions2: usize = 0;
    var length_deletions2: usize = 0;
    while (pointer < diffs.items.len) {
        if (diffs.items[@intCast(pointer)].operation == .equal) { // Equality found.
            try equalities.append(allocator, pointer);
            length_insertions1 = length_insertions2;
            length_deletions1 = length_deletions2;
            length_insertions2 = 0;
            length_deletions2 = 0;
            last_equality = diffs.items[@intCast(pointer)].text;
        } else { // an insertion or deletion
            if (diffs.items[@intCast(pointer)].operation == .insert) {
                length_insertions2 += diffs.items[@intCast(pointer)].text.len;
            } else {
                length_deletions2 += diffs.items[@intCast(pointer)].text.len;
            }
            // Eliminate an equality that is smaller or equal to the edits on both
            // sides of it.
            if (last_equality != null and
                (last_equality.?.len <= @max(length_insertions1, length_deletions1)) and
                (last_equality.?.len <= @max(length_insertions2, length_deletions2)))
            {
                // Duplicate record.
                try diffs.ensureUnusedCapacity(allocator, 1);
                diffs.insertAssumeCapacity(
                    @intCast(equalities.items[equalities.items.len - 1]),
                    .{
                        .operation = .delete,
                        .text = try allocator.dupe(u8, last_equality.?),
                    },
                );
                // Change second copy to insert.
                diffs.items[@intCast(equalities.items[equalities.items.len - 1] + 1)].operation = .insert;
                // Throw away the equality we just deleted.
                _ = equalities.pop();
                if (equalities.items.len > 0) {
                    _ = equalities.pop();
                }
                pointer = if (equalities.items.len > 0) equalities.items[equalities.items.len - 1] else -1;
                length_insertions1 = 0; // Reset the counters.
                length_deletions1 = 0;
                length_insertions2 = 0;
                length_deletions2 = 0;
                last_equality = null;
                changes = true;
            }
        }
        pointer += 1;
    }

    // Normalize the diff.
    if (changes) {
        try diffCleanupMerge(allocator, diffs);
    }
    try diffCleanupSemanticLossless(allocator, diffs);

    // Find any overlaps between deletions and insertions.
    // e.g: <del>abcxxx</del><ins>xxxdef</ins>
    //   -> <del>abc</del>xxx<ins>def</ins>
    // e.g: <del>xxxabc</del><ins>defxxx</ins>
    //   -> <ins>def</ins>xxx<del>abc</del>
    // Only extract an overlap if it is as big as the edit ahead or behind it.
    pointer = 1;
    while (pointer < diffs.items.len) {
        if (diffs.items[@intCast(pointer - 1)].operation == .delete and
            diffs.items[@intCast(pointer)].operation == .insert)
        {
            const deletion = diffs.items[@intCast(pointer - 1)].text;
            const insertion = diffs.items[@intCast(pointer)].text;
            const overlap_length1: usize = diffCommonOverlap(deletion, insertion);
            const overlap_length2: usize = diffCommonOverlap(insertion, deletion);
            if (overlap_length1 >= overlap_length2) {
                if (@as(f32, @floatFromInt(overlap_length1)) >= @as(f32, @floatFromInt(deletion.len)) / 2.0 or
                    @as(f32, @floatFromInt(overlap_length1)) >= @as(f32, @floatFromInt(insertion.len)) / 2.0)
                {
                    // Overlap found.
                    // Insert an equality and trim the surrounding edits.
                    try diffs.ensureUnusedCapacity(allocator, 1);
                    diffs.insertAssumeCapacity(@intCast(pointer), .{
                        .operation = .equal,
                        .text = try allocator.dupe(u8, insertion[0..overlap_length1]),
                    });
                    diffs.items[@intCast(pointer - 1)].text =
                        try allocator.dupe(u8, deletion[0 .. deletion.len - overlap_length1]);
                    allocator.free(deletion);
                    diffs.items[@intCast(pointer + 1)].text =
                        try allocator.dupe(u8, insertion[overlap_length1..]);
                    allocator.free(insertion);
                    pointer += 1;
                }
            } else {
                if (@as(f32, @floatFromInt(overlap_length2)) >= @as(f32, @floatFromInt(deletion.len)) / 2.0 or
                    @as(f32, @floatFromInt(overlap_length2)) >= @as(f32, @floatFromInt(insertion.len)) / 2.0)
                {
                    // Reverse overlap found.
                    // Insert an equality and swap and trim the surrounding edits.
                    try diffs.ensureUnusedCapacity(allocator, 1);
                    diffs.insertAssumeCapacity(@intCast(pointer), .{
                        .operation = .equal,
                        .text = try allocator.dupe(u8, deletion[0..overlap_length2]),
                    });
                    const new_minus = try allocator.dupe(u8, insertion[0 .. insertion.len - overlap_length2]);
                    errdefer allocator.free(new_minus); // necessary due to swap
                    const new_plus = try allocator.dupe(u8, deletion[overlap_length2..]);
                    allocator.free(deletion);
                    allocator.free(insertion);
                    diffs.items[@intCast(pointer - 1)].operation = .insert;
                    diffs.items[@intCast(pointer - 1)].text = new_minus;
                    diffs.items[@intCast(pointer + 1)].operation = .delete;
                    diffs.items[@intCast(pointer + 1)].text = new_plus;
                    pointer += 1;
                }
            }
            pointer += 1;
        }
        pointer += 1;
    }
}

/// Look for single edits surrounded on both sides by equalities
/// which can be shifted sideways to align the edit to a word boundary.
/// e.g: The c<ins>at c</ins>ame. -> The <ins>cat </ins>came.
pub fn diffCleanupSemanticLossless(
    allocator: Allocator,
    diffs: *DiffList,
) DiffError!void {
    var pointer: usize = 1;
    // Intentionally ignore the first and last element (don't need checking).
    while (pointer < @as(isize, @intCast(diffs.items.len)) - 1) {
        if (diffs.items[pointer - 1].operation == .equal and
            diffs.items[pointer + 1].operation == .equal)
        {
            // This is a single edit surrounded by equalities.
            var equality_1: std.ArrayList(u8) = .empty;
            defer equality_1.deinit(allocator);
            try equality_1.appendSlice(allocator, diffs.items[pointer - 1].text);

            var edit: std.ArrayList(u8) = .empty;
            defer edit.deinit(allocator);
            try edit.appendSlice(allocator, diffs.items[pointer].text);

            var equality_2: std.ArrayList(u8) = .empty;
            defer equality_2.deinit(allocator);
            try equality_2.appendSlice(allocator, diffs.items[pointer + 1].text);

            // First, shift the edit as far left as possible.
            const common_offset = diffCommonSuffix(equality_1.items, edit.items);
            if (common_offset > 0) {
                // TODO: Use buffer
                const common_string = try allocator.dupe(u8, edit.items[edit.items.len - common_offset ..]);
                defer allocator.free(common_string);

                equality_1.items.len = equality_1.items.len - common_offset;

                // edit.items.len = edit.items.len - common_offset;
                const not_common = try allocator.dupe(u8, edit.items[0 .. edit.items.len - common_offset]);
                defer allocator.free(not_common);

                edit.clearRetainingCapacity();
                try edit.appendSlice(allocator, common_string);
                try edit.appendSlice(allocator, not_common);

                try equality_2.insertSlice(allocator, 0, common_string);
            }

            // Second, step character by character right,
            // looking for the best fit.
            var best_equality_1: std.ArrayList(u8) = .empty;
            defer best_equality_1.deinit(allocator);
            try best_equality_1.appendSlice(allocator, equality_1.items);

            var best_edit: std.ArrayList(u8) = .empty;
            defer best_edit.deinit(allocator);
            try best_edit.appendSlice(allocator, edit.items);

            var best_equality_2: std.ArrayList(u8) = .empty;
            defer best_equality_2.deinit(allocator);
            try best_equality_2.appendSlice(allocator, equality_2.items);

            var best_score = diffCleanupSemanticScore(equality_1.items, edit.items) +
                diffCleanupSemanticScore(edit.items, equality_2.items);

            while (edit.items.len != 0 and equality_2.items.len != 0 and edit.items[0] == equality_2.items[0]) {
                try equality_1.append(allocator, edit.items[0]);

                _ = edit.orderedRemove(0);
                try edit.append(allocator, equality_2.items[0]);

                _ = equality_2.orderedRemove(0);

                const score = diffCleanupSemanticScore(equality_1.items, edit.items) +
                    diffCleanupSemanticScore(edit.items, equality_2.items);
                // The >= encourages trailing rather than leading whitespace on
                // edits.
                if (score >= best_score) {
                    best_score = score;

                    best_equality_1.clearRetainingCapacity();
                    try best_equality_1.appendSlice(allocator, equality_1.items);

                    best_edit.clearRetainingCapacity();
                    try best_edit.appendSlice(allocator, edit.items);

                    best_equality_2.clearRetainingCapacity();
                    try best_equality_2.appendSlice(allocator, equality_2.items);
                }
            }

            if (!std.mem.eql(u8, diffs.items[pointer - 1].text, best_equality_1.items)) {
                // We have an improvement, save it back to the diff.
                if (best_equality_1.items.len != 0) {
                    const old_text = diffs.items[pointer - 1].text;
                    diffs.items[pointer - 1].text = try allocator.dupe(u8, best_equality_1.items);
                    allocator.free(old_text);
                } else {
                    const old_diff = diffs.orderedRemove(pointer - 1);
                    allocator.free(old_diff.text);
                    pointer -= 1;
                }
                const old_text1 = diffs.items[pointer].text;
                diffs.items[pointer].text = try allocator.dupe(u8, best_edit.items);
                defer allocator.free(old_text1);
                if (best_equality_2.items.len != 0) {
                    const old_text2 = diffs.items[pointer + 1].text;
                    diffs.items[pointer + 1].text = try allocator.dupe(u8, best_equality_2.items);
                    allocator.free(old_text2);
                } else {
                    const old_diff = diffs.orderedRemove(pointer + 1);
                    allocator.free(old_diff.text);
                    pointer -= 1;
                }
            }
        }
        pointer += 1;
    }
}

/// Given two strings, compute a score representing whether the internal
/// boundary falls on logical boundaries.
/// Scores range from 6 (best) to 0 (worst).
/// @param one First string.
/// @param two Second string.
/// @return The score.
fn diffCleanupSemanticScore(one: []const u8, two: []const u8) usize {
    if (one.len == 0 or two.len == 0) {
        // Edges are the best.
        return 6;
    }

    // Each port of this function behaves slightly differently due to
    // subtle differences in each language's definition of things like
    // 'whitespace'.  Since this function's purpose is largely cosmetic,
    // the choice has been made to use each language's native features
    // rather than force total conformity.
    const char1 = one[one.len - 1];
    const char2 = two[0];
    const nonAlphaNumeric1 = !std.ascii.isAlphanumeric(char1);
    const nonAlphaNumeric2 = !std.ascii.isAlphanumeric(char2);
    const whitespace1 = nonAlphaNumeric1 and std.ascii.isWhitespace(char1);
    const whitespace2 = nonAlphaNumeric2 and std.ascii.isWhitespace(char2);
    const lineBreak1 = whitespace1 and std.ascii.isControl(char1);
    const lineBreak2 = whitespace2 and std.ascii.isControl(char2);
    const blankLine1 = lineBreak1 and
        // BLANKLINEEND.IsMatch(one);
        (std.mem.endsWith(u8, one, "\n\n") or std.mem.endsWith(u8, one, "\n\r\n"));
    const blankLine2 = lineBreak2 and
        // BLANKLINESTART.IsMatch(two);
        (std.mem.startsWith(u8, two, "\n\n") or
            std.mem.startsWith(u8, two, "\r\n\n") or
            std.mem.startsWith(u8, two, "\n\r\n") or
            std.mem.startsWith(u8, two, "\r\n\r\n"));

    if (blankLine1 or blankLine2) {
        // Five points for blank lines.
        return 5;
    } else if (lineBreak1 or lineBreak2) {
        // Four points for line breaks.
        return 4;
    } else if (nonAlphaNumeric1 and !whitespace1 and whitespace2) {
        // Three points for end of sentences.
        return 3;
    } else if (whitespace1 or whitespace2) {
        // Two points for whitespace.
        return 2;
    } else if (nonAlphaNumeric1 or nonAlphaNumeric2) {
        // One point for non-alphanumeric.
        return 1;
    }
    return 0;
}

/// Reduce the number of edits by eliminating operationally trivial
/// equalities.
pub fn diffCleanupEfficiency(
    dmp: *const DiffMatchPatch,
    diffs: *DiffList,
) DiffError!void {
    const allocator = dmp.allocator;
    var changes = false;
    // Stack of indices where equalities are found.
    var equalities: std.ArrayList(usize) = .empty;
    defer equalities.deinit(allocator);
    // Always equal to equalities[equalitiesLength-1][1]
    var last_equality: []const u8 = "";
    var ipointer: isize = 0; // Index of current position.
    // Is there an insertion operation before the last equality.
    var pre_ins = false;
    // Is there a deletion operation before the last equality.
    var pre_del = false;
    // Is there an insertion operation after the last equality.
    var post_ins = false;
    // Is there a deletion operation after the last equality.
    var post_del = false;
    while (ipointer < diffs.items.len) {
        const pointer: usize = @intCast(ipointer);
        if (diffs.items[pointer].operation == .equal) { // Equality found.
            if (diffs.items[pointer].text.len < dmp.diff_edit_cost and (post_ins or post_del)) {
                // Candidate found.
                try equalities.append(allocator, pointer);
                pre_ins = post_ins;
                pre_del = post_del;
                last_equality = diffs.items[pointer].text;
            } else {
                // Not a candidate, and can never become one.
                equalities.clearRetainingCapacity();
                last_equality = "";
            }
            post_ins = false;
            post_del = false;
        } else { // An insertion or deletion.
            if (diffs.items[pointer].operation == .delete) {
                post_del = true;
            } else {
                post_ins = true;
            }
            // Five types to be split:
            // <ins>A</ins><del>B</del>XY<ins>C</ins><del>D</del>
            // <ins>A</ins>X<ins>C</ins><del>D</del>
            // <ins>A</ins><del>B</del>X<ins>C</ins>
            // <ins>A</del>X<ins>C</ins><del>D</del>
            // <ins>A</ins><del>B</del>X<del>C</del>
            if ((last_equality.len != 0) and
                ((pre_ins and pre_del and post_ins and post_del) or
                    ((last_equality.len < dmp.diff_edit_cost / 2) and
                        (@as(u8, @intFromBool(pre_ins)) + @as(u8, @intFromBool(pre_del)) + @as(u8, @intFromBool(post_ins)) + @as(u8, @intFromBool(post_del)) == 3))))
            {
                // Duplicate record.
                try diffs.ensureUnusedCapacity(allocator, 1);
                diffs.insertAssumeCapacity(
                    equalities.items[equalities.items.len - 1],
                    .{
                        .operation = .delete,
                        .text = try allocator.dupe(u8, last_equality),
                    },
                );
                // Change second copy to insert.
                diffs.items[equalities.items[equalities.items.len - 1] + 1].operation = .insert;
                _ = equalities.pop(); // Throw away the equality we just deleted.
                last_equality = "";
                if (pre_ins and pre_del) {
                    // No changes made which could affect previous entry, keep going.
                    post_ins = true;
                    post_del = true;
                    equalities.clearRetainingCapacity();
                } else {
                    if (equalities.items.len > 0) {
                        _ = equalities.pop();
                    }

                    ipointer = if (equalities.items.len > 0) @intCast(equalities.items[equalities.items.len - 1]) else -1;
                    post_ins = false;
                    post_del = false;
                }
                changes = true;
            }
        }
        ipointer += 1;
    }

    if (changes) {
        try diffCleanupMerge(allocator, diffs);
    }
}

/// Determine if the suffix of one string is the prefix of another.
/// @param text1 First string.
/// @param text2 Second string.
/// @return The number of characters common to the end of the first
///     string and the start of the second string.
fn diffCommonOverlap(text1_in: []const u8, text2_in: []const u8) usize {
    var text1 = text1_in;
    var text2 = text2_in;

    // Cache the text lengths to prevent multiple calls.
    const text1_length = text1.len;
    const text2_length = text2.len;
    // Eliminate the null case.
    if (text1_length == 0 or text2_length == 0) {
        return 0;
    }
    // Truncate the longer string.
    if (text1_length > text2_length) {
        text1 = text1[text1_length - text2_length ..];
    } else if (text1_length < text2_length) {
        text2 = text2[0..text1_length];
    }
    const text_length = @min(text1_length, text2_length);
    // Quick check for the worst case.
    if (std.mem.eql(u8, text1, text2)) {
        return text_length;
    }

    // Start by looking for a single character match
    // and increase length until no match is found.
    // Performance analysis: https://neil.fraser.name/news/2010/11/04/
    var best: usize = 0;
    var length: usize = 1;
    while (true) {
        const pattern = text1[text_length - length ..];
        const found = std.mem.find(u8, text2, pattern) orelse
            return best;

        length += found;

        if (found == 0 or std.mem.eql(u8, text1[text_length - length ..], text2[0..length])) {
            best = length;
            length += 1;
        }
    }
}

// DONE [✅]: Allocate all text in diffs to
// not cause segfault while freeing

test diffCommonPrefix {
    // Detect any common suffix.
    try testing.expectEqual(@as(usize, 0), diffCommonPrefix("abc", "xyz")); // Null case
    try testing.expectEqual(@as(usize, 4), diffCommonPrefix("1234abcdef", "1234xyz")); // Non-null case
    try testing.expectEqual(@as(usize, 4), diffCommonPrefix("1234", "1234xyz")); // Whole case
}

test diffCommonSuffix {
    // Detect any common suffix.
    try testing.expectEqual(@as(usize, 0), diffCommonSuffix("abc", "xyz")); // Null case
    try testing.expectEqual(@as(usize, 4), diffCommonSuffix("abcdef1234", "xyz1234")); // Non-null case
    try testing.expectEqual(@as(usize, 4), diffCommonSuffix("1234", "xyz1234")); // Whole case
}

test diffCommonOverlap {
    // Detect any suffix/prefix overlap.
    try testing.expectEqual(@as(usize, 0), diffCommonOverlap("", "abcd")); // Null case
    try testing.expectEqual(@as(usize, 3), diffCommonOverlap("abc", "abcd")); // Whole case
    try testing.expectEqual(@as(usize, 0), diffCommonOverlap("123456", "abcd")); // No overlap
    try testing.expectEqual(@as(usize, 3), diffCommonOverlap("123456xxx", "xxxabcd")); // Overlap

    // Some overly clever languages (C#) may treat ligatures as equal to their
    // component letters.  E.g. U+FB01 == 'fi'
    try testing.expectEqual(@as(usize, 0), diffCommonOverlap("fi", "\u{fb01}")); // Unicode
}

fn testDiffHalfMatch(
    allocator: Allocator,
    params: struct {
        timeout: std.Io.Timeout,
        before: []const u8,
        after: []const u8,
        expected: ?HalfMatchResult,
    },
) !void {
    const dmp: DiffMatchPatch = .initDefault(testing.io, allocator);
    const maybe_result = try dmp.diffHalfMatch(
        params.before,
        params.after,
        params.timeout.toTimestamp(testing.io),
    );
    defer if (maybe_result) |result| result.deinit(allocator);
    try testing.expectEqualDeep(params.expected, maybe_result);
}

test diffHalfMatch {
    const no_timeout: std.Io.Timeout = .none;
    const with_timeout: std.Io.Timeout = .{ .deadline = .{ .clock = .awake, .raw = .fromNanoseconds(0) } };

    // No match #1
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "1234567890",
        .after = "abcdef",
        .expected = null,
    }});

    // No match #2
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "12345",
        .after = "23",
        .expected = null,
    }});

    // Single matches
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "1234567890",
        .after = "a345678z",
        .expected = .{
            .prefix_before = "12",
            .suffix_before = "90",
            .prefix_after = "a",
            .suffix_after = "z",
            .common_middle = "345678",
        },
    }});

    // Single Match #2
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "a345678z",
        .after = "1234567890",
        .expected = .{
            .prefix_before = "a",
            .suffix_before = "z",
            .prefix_after = "12",
            .suffix_after = "90",
            .common_middle = "345678",
        },
    }});

    // Single Match #3
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "abc56789z",
        .after = "1234567890",
        .expected = .{
            .prefix_before = "abc",
            .suffix_before = "z",
            .prefix_after = "1234",
            .suffix_after = "0",
            .common_middle = "56789",
        },
    }});

    // Single Match #4
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "a23456xyz",
        .after = "1234567890",
        .expected = .{
            .prefix_before = "a",
            .suffix_before = "xyz",
            .prefix_after = "1",
            .suffix_after = "7890",
            .common_middle = "23456",
        },
    }});

    // Multiple matches #1
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "121231234123451234123121",
        .after = "a1234123451234z",
        .expected = .{
            .prefix_before = "12123",
            .suffix_before = "123121",
            .prefix_after = "a",
            .suffix_after = "z",
            .common_middle = "1234123451234",
        },
    }});

    // Multiple Matches #2
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "x-=-=-=-=-=-=-=-=-=-=-=-=",
        .after = "xx-=-=-=-=-=-=-=",
        .expected = .{
            .prefix_before = "",
            .suffix_before = "-=-=-=-=-=",
            .prefix_after = "x",
            .suffix_after = "",
            .common_middle = "x-=-=-=-=-=-=-=",
        },
    }});

    // Multiple Matches #3
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "-=-=-=-=-=-=-=-=-=-=-=-=y",
        .after = "-=-=-=-=-=-=-=yy",
        .expected = .{
            .prefix_before = "-=-=-=-=-=",
            .suffix_before = "",
            .prefix_after = "",
            .suffix_after = "y",
            .common_middle = "-=-=-=-=-=-=-=y",
        },
    }});

    // Other cases

    // Optimal diff would be -q+x=H-i+e=lloHe+Hu=llo-Hew+y not -qHillo+x=HelloHe-w+Hulloy
    // Non-optimal halfmatch
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = with_timeout,
        .before = "qHilloHelloHew",
        .after = "xHelloHeHulloy",
        .expected = .{
            .prefix_before = "qHillo",
            .suffix_before = "w",
            .prefix_after = "x",
            .suffix_after = "Hulloy",
            .common_middle = "HelloHe",
        },
    }});

    // Non-optimal halfmatch
    try checkAllAllocationFailures(testing.allocator, testDiffHalfMatch, .{.{
        .timeout = no_timeout,
        .before = "qHilloHelloHew",
        .after = "xHelloHeHulloy",
        .expected = null,
    }});
}

test diffLinesToChars {
    const allocator = testing.allocator;
    // Convert lines down to characters.
    var tmp_array_list: std.ArrayList([]const u8) = .empty;
    defer tmp_array_list.deinit(allocator);
    try tmp_array_list.append(allocator, "");
    try tmp_array_list.append(allocator, "alpha\n");
    try tmp_array_list.append(allocator, "beta\n");

    var result = try diffLinesToChars(allocator, "alpha\nbeta\nalpha\n", "beta\nalpha\nbeta\n");
    try testing.expectEqualStrings("\u{0001}\u{0002}\u{0001}", result.chars_1); // Shared lines #1
    try testing.expectEqualStrings("\u{0002}\u{0001}\u{0002}", result.chars_2); // Shared lines #2
    try testing.expectEqualDeep(tmp_array_list.items, result.line_array.items); // Shared lines #3

    tmp_array_list.clearRetainingCapacity();
    try tmp_array_list.append(allocator, "");
    try tmp_array_list.append(allocator, "alpha\r\n");
    try tmp_array_list.append(allocator, "beta\r\n");
    try tmp_array_list.append(allocator, "\r\n");
    result.deinit(allocator);

    result = try diffLinesToChars(allocator, "", "alpha\r\nbeta\r\n\r\n\r\n");
    try testing.expectEqualStrings("", result.chars_1); // Empty string and blank lines #1
    try testing.expectEqualStrings("\u{0001}\u{0002}\u{0003}\u{0003}", result.chars_2); // Empty string and blank lines #2
    try testing.expectEqualDeep(tmp_array_list.items, result.line_array.items); // Empty string and blank lines #3

    tmp_array_list.clearRetainingCapacity();
    try tmp_array_list.append(allocator, "");
    try tmp_array_list.append(allocator, "a");
    try tmp_array_list.append(allocator, "b");
    result.deinit(allocator);

    result = try diffLinesToChars(allocator, "a", "b");
    try testing.expectEqualStrings("\u{0001}", result.chars_1); // No linebreaks #1.
    try testing.expectEqualStrings("\u{0002}", result.chars_2); // No linebreaks #2.
    try testing.expectEqualDeep(tmp_array_list.items, result.line_array.items); // No linebreaks #3.
    result.deinit(allocator);

    // TODO: More than 256 to reveal any 8-bit limitations but this requires
    // some unicode logic that I don't want to deal with
    //
    // Casting to Unicode is straightforward and should sort correctly, I'm
    // more concerned about the weird behavior when the 'char' is equal to a
    // newline.  Uncomment the EqualSlices below to see what I mean.
    // I think there's some cleanup logic in the actual linediff that should
    // take care of the problem, but I don't like it.

    const n: u8 = 255;
    tmp_array_list.clearRetainingCapacity();

    var line_list: std.ArrayList(u8) = .empty;
    defer line_list.deinit(allocator);
    var char_list: std.ArrayList(u8) = .empty;
    defer char_list.deinit(allocator);

    var i: u8 = 1;
    while (i < n) : (i += 1) {
        try tmp_array_list.append(allocator, &.{ i, '\n' });
        try line_list.appendSlice(allocator, &.{ i, '\n' });
        try char_list.append(allocator, i);
    }
    try testing.expectEqual(@as(usize, n - 1), tmp_array_list.items.len); // Test initialization fail #1
    try testing.expectEqual(@as(usize, n - 1), char_list.items.len); // Test initialization fail #2
    try tmp_array_list.insert(allocator, 0, "");
    result = try diffLinesToChars(allocator, line_list.items, "");
    defer result.deinit(allocator);
    // TODO: This isn't equal, should it be?
    // try testing.expectEqualSlices(u8, char_list.items, result.chars_1);
    try testing.expectEqualStrings("", result.chars_2);
    // TODO this is wrong because of the max_value I think?
    // try testing.expectEqualDeep(tmp_array_list.items, result.line_array.items);
}

fn testDiffCharsToLines(
    allocator: Allocator,
    params: struct {
        diffs: []const Diff,
        line_array: []const []const u8,
        expected: []const Diff,
    },
) !void {
    var char_diffs: DiffList = try .initCapacity(allocator, params.diffs.len);
    defer deinitDiffList(allocator, &char_diffs);

    for (params.diffs) |item| {
        char_diffs.appendAssumeCapacity(.{ .operation = item.operation, .text = try allocator.dupe(u8, item.text) });
    }

    var diffs = try diffCharsToLines(allocator, &char_diffs, params.line_array);
    defer deinitDiffList(allocator, &diffs);

    try testing.expectEqualDeep(params.expected, diffs.items);
}

test diffCharsToLines {
    // Convert chars up to lines.
    var diff_list: DiffList = .empty;
    defer deinitDiffList(testing.allocator, &diff_list);
    try diff_list.ensureTotalCapacity(testing.allocator, 2);
    diff_list.appendSliceAssumeCapacity(&.{
        .{ .operation = .equal, .text = try testing.allocator.dupe(u8, "\u{0001}\u{0002}\u{0001}") },
        .{ .operation = .insert, .text = try testing.allocator.dupe(u8, "\u{0002}\u{0001}\u{0002}") },
    });
    try checkAllAllocationFailures(testing.allocator, testDiffCharsToLines, .{.{
        .diffs = diff_list.items,
        .line_array = &[_][]const u8{
            "",
            "alpha\n",
            "beta\n",
        },
        .expected = &.{
            .{ .operation = .equal, .text = "alpha\nbeta\nalpha\n" },
            .{ .operation = .insert, .text = "beta\nalpha\nbeta\n" },
        },
    }});

    // TODO: Implement exhaustive tests
}

fn testDiffCleanupMerge(allocator: Allocator, params: struct {
    input: []const Diff,
    expected: []const Diff,
}) !void {
    var diffs: DiffList = try .initCapacity(allocator, params.input.len);
    defer deinitDiffList(allocator, &diffs);

    for (params.input) |item| {
        diffs.appendAssumeCapacity(.{ .operation = item.operation, .text = try allocator.dupe(u8, item.text) });
    }

    try diffCleanupMerge(allocator, &diffs);

    try testing.expectEqualDeep(params.expected, diffs.items);
}

test diffCleanupMerge {
    // Cleanup a messy diff.

    // No change case
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .insert, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .insert, .text = "c" },
        },
    }});

    // Merge equalities
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .equal, .text = "b" },
            .{ .operation = .equal, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "abc" },
        },
    }});

    // Merge deletions
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .delete, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abc" },
        },
    }});

    // Merge insertions
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .insert, .text = "a" },
            .{ .operation = .insert, .text = "b" },
            .{ .operation = .insert, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .insert, .text = "abc" },
        },
    }});

    // Merge interweave
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .insert, .text = "b" },
            .{ .operation = .delete, .text = "c" },
            .{ .operation = .insert, .text = "d" },
            .{ .operation = .equal, .text = "e" },
            .{ .operation = .equal, .text = "f" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "ac" },
            .{ .operation = .insert, .text = "bd" },
            .{ .operation = .equal, .text = "ef" },
        },
    }});

    // Prefix and suffix detection
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .insert, .text = "abc" },
            .{ .operation = .delete, .text = "dc" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "d" },
            .{ .operation = .insert, .text = "b" },
            .{ .operation = .equal, .text = "c" },
        },
    }});

    // Prefix and suffix detection with equalities
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "x" },
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .insert, .text = "abc" },
            .{ .operation = .delete, .text = "dc" },
            .{ .operation = .equal, .text = "y" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "xa" },
            .{ .operation = .delete, .text = "d" },
            .{ .operation = .insert, .text = "b" },
            .{ .operation = .equal, .text = "cy" },
        },
    }});

    // Slide edit left
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .insert, .text = "ba" },
            .{ .operation = .equal, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .insert, .text = "ab" },
            .{ .operation = .equal, .text = "ac" },
        },
    }});

    // Slide edit right
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "c" },
            .{ .operation = .insert, .text = "ab" },
            .{ .operation = .equal, .text = "a" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "ca" },
            .{ .operation = .insert, .text = "ba" },
        },
    }});

    // Slide edit left recursive
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .equal, .text = "c" },
            .{ .operation = .delete, .text = "ac" },
            .{ .operation = .equal, .text = "x" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abc" },
            .{ .operation = .equal, .text = "acx" },
        },
    }});

    // Slide edit right recursive
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "x" },
            .{ .operation = .delete, .text = "ca" },
            .{ .operation = .equal, .text = "c" },
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .equal, .text = "a" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "xca" },
            .{ .operation = .delete, .text = "cba" },
        },
    }});

    // Empty merge
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "b" },
            .{ .operation = .insert, .text = "ab" },
            .{ .operation = .equal, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .insert, .text = "a" },
            .{ .operation = .equal, .text = "bc" },
        },
    }});

    // Empty equality
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupMerge, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "" },
            .{ .operation = .insert, .text = "a" },
            .{ .operation = .equal, .text = "b" },
        },
        .expected = &.{
            .{ .operation = .insert, .text = "a" },
            .{ .operation = .equal, .text = "b" },
        },
    }});
}

fn testDiffCleanupSemanticLossless(
    allocator: Allocator,
    params: struct {
        input: []const Diff,
        expected: []const Diff,
    },
) !void {
    var diffs: DiffList = try .initCapacity(allocator, params.input.len);
    defer deinitDiffList(allocator, &diffs);

    for (params.input) |item| {
        diffs.appendAssumeCapacity(.{ .operation = item.operation, .text = try allocator.dupe(u8, item.text) });
    }

    try diffCleanupSemanticLossless(allocator, &diffs);

    try testing.expectEqualDeep(params.expected, diffs.items);
}

fn sliceToDiffList(allocator: Allocator, diff_slice: []const Diff) !DiffList {
    var diff_list: DiffList = .empty;
    errdefer deinitDiffList(allocator, &diff_list);
    try diff_list.ensureTotalCapacity(allocator, diff_slice.len);
    for (diff_slice) |d| {
        diff_list.appendAssumeCapacity(.{
            .operation = d.operation,
            .text = try allocator.dupe(u8, d.text),
        });
    }
    return diff_list;
}

test diffCleanupSemanticLossless {
    // Null case
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &[_]Diff{},
        .expected = &[_]Diff{},
    }});

    //defer deinitDiffList(allocator, &diffs);
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "AAA\r\n\r\nBBB" },
            .{ .operation = .insert, .text = "\r\nDDD\r\n\r\nBBB" },
            .{ .operation = .equal, .text = "\r\nEEE" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "AAA\r\n\r\n" },
            .{ .operation = .insert, .text = "BBB\r\nDDD\r\n\r\n" },
            .{ .operation = .equal, .text = "BBB\r\nEEE" },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "AAA\r\nBBB" },
            .{ .operation = .insert, .text = " DDD\r\nBBB" },
            .{ .operation = .equal, .text = " EEE" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "AAA\r\n" },
            .{ .operation = .insert, .text = "BBB DDD\r\n" },
            .{ .operation = .equal, .text = "BBB EEE" },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "The c" },
            .{ .operation = .insert, .text = "ow and the c" },
            .{ .operation = .equal, .text = "at." },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "The " },
            .{ .operation = .insert, .text = "cow and the " },
            .{ .operation = .equal, .text = "cat." },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "The-c" },
            .{ .operation = .insert, .text = "ow-and-the-c" },
            .{ .operation = .equal, .text = "at." },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "The-" },
            .{ .operation = .insert, .text = "cow-and-the-" },
            .{ .operation = .equal, .text = "cat." },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .equal, .text = "ax" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .equal, .text = "aax" },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "xa" },
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .equal, .text = "a" },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "xaa" },
            .{ .operation = .delete, .text = "a" },
        },
    }});

    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemanticLossless, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "The xxx. The " },
            .{ .operation = .insert, .text = "zzz. The " },
            .{ .operation = .equal, .text = "yyy." },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "The xxx." },
            .{ .operation = .insert, .text = " The zzz." },
            .{ .operation = .equal, .text = " The yyy." },
        },
    }});
}

fn rebuildtexts(allocator: Allocator, diffs: DiffList) ![2][]const u8 {
    var text: [2]std.ArrayList(u8) = .{ .empty, .empty };
    errdefer {
        text[0].deinit(allocator);
        text[1].deinit(allocator);
    }

    for (diffs.items) |myDiff| {
        if (myDiff.operation != .insert) {
            try text[0].appendSlice(allocator, myDiff.text);
        }
        if (myDiff.operation != .delete) {
            try text[1].appendSlice(allocator, myDiff.text);
        }
    }

    const first = try text[0].toOwnedSlice(allocator);
    errdefer allocator.free(first);

    const second = try text[1].toOwnedSlice(allocator);
    errdefer allocator.free(second);

    return .{ first, second };
}

fn testRebuildTexts(allocator: Allocator, diffs: DiffList, params: struct {
    before: []const u8,
    after: []const u8,
}) !void {
    const texts = try rebuildtexts(allocator, diffs);
    defer {
        allocator.free(texts[0]);
        allocator.free(texts[1]);
    }
    try testing.expectEqualStrings(params.before, texts[0]);
    try testing.expectEqualStrings(params.after, texts[1]);
}

test rebuildtexts {
    {
        var diffs = try sliceToDiffList(testing.allocator, &.{
            .{ .operation = .insert, .text = "abcabc" },
            .{ .operation = .equal, .text = "defdef" },
            .{ .operation = .delete, .text = "ghighi" },
        });
        defer deinitDiffList(testing.allocator, &diffs);
        try checkAllAllocationFailures(testing.allocator, testRebuildTexts, .{
            diffs,
            .{
                .before = "defdefghighi",
                .after = "abcabcdefdef",
            },
        });
    }
    {
        var diffs = try sliceToDiffList(testing.allocator, &.{
            .{ .operation = .insert, .text = "xxx" },
            .{ .operation = .delete, .text = "yyy" },
        });
        defer deinitDiffList(testing.allocator, &diffs);
        try checkAllAllocationFailures(testing.allocator, testRebuildTexts, .{
            diffs,
            .{
                .before = "yyy",
                .after = "xxx",
            },
        });
    }
    {
        var diffs = try sliceToDiffList(testing.allocator, &.{
            .{ .operation = .equal, .text = "xyz" },
            .{ .operation = .equal, .text = "pdq" },
        });
        defer deinitDiffList(testing.allocator, &diffs);
        try checkAllAllocationFailures(testing.allocator, testRebuildTexts, .{
            diffs,
            .{
                .before = "xyzpdq",
                .after = "xyzpdq",
            },
        });
    }
}

fn testDiffBisect(
    allocator: Allocator,
    params: struct {
        timeout: std.Io.Timeout,
        before: []const u8,
        after: []const u8,
        expected: []const Diff,
    },
) !void {
    const dmp: DiffMatchPatch = .initDefault(testing.io, allocator);
    var diffs = try dmp.diffBisect(
        params.before,
        params.after,
        params.timeout.toTimestamp(testing.io),
    );
    defer deinitDiffList(allocator, &diffs);
    try testing.expectEqualDeep(params.expected, diffs.items);
}

test diffBisect {
    const a = "cat";
    const b = "map";

    // Normal
    try checkAllAllocationFailures(testing.allocator, testDiffBisect, .{.{
        .timeout = .none,
        .before = a,
        .after = b,
        .expected = &.{
            .{ .operation = .delete, .text = "c" },
            .{ .operation = .insert, .text = "m" },
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "t" },
            .{ .operation = .insert, .text = "p" },
        },
    }});

    // Timeout
    try checkAllAllocationFailures(testing.allocator, testDiffBisect, .{.{
        .timeout = .{ .duration = .{ .clock = .awake, .raw = .fromNanoseconds(0) } },
        .before = a,
        .after = b,
        .expected = &.{
            .{ .operation = .delete, .text = "cat" },
            .{ .operation = .insert, .text = "map" },
        },
    }});
}

fn diffHalfMatchLeak(allocator: Allocator) !void {
    const dmp: DiffMatchPatch = .initDefault(testing.io, testing.allocator);
    const text1 = "The quick brown fox jumps over the lazy dog.";
    const text2 = "That quick brown fox jumped over a lazy dog.";
    var diffs = try dmp.diff(text2, text1, true, .none);
    deinitDiffList(allocator, &diffs);
}

test "diffHalfMatch leak regression test" {
    try checkAllAllocationFailures(testing.allocator, diffHalfMatchLeak, .{});
}

fn testDiff(
    allocator: Allocator,
    params: struct {
        before: []const u8,
        after: []const u8,
        check_lines: bool,
        expected: []const Diff,
    },
) !void {
    const dmp: DiffMatchPatch = .initDefault(testing.io, testing.allocator);
    var diffs = try dmp.diff(params.before, params.after, params.check_lines, .none);
    defer deinitDiffList(allocator, &diffs);
    try testing.expectEqualDeep(params.expected, diffs.items);
}

test diff {

    //  Null case.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "",
        .after = "",
        .check_lines = false,
        .expected = &[_]Diff{},
    }});

    //  Equality.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "abc",
        .after = "abc",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .equal, .text = "abc" },
        },
    }});

    // Simple insertion.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "abc",
        .after = "ab123c",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .equal, .text = "ab" },
            .{ .operation = .insert, .text = "123" },
            .{ .operation = .equal, .text = "c" },
        },
    }});

    // Simple deletion.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "a123bc",
        .after = "abc",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "123" },
            .{ .operation = .equal, .text = "bc" },
        },
    }});

    // Two insertions.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "abc",
        .after = "a123b456c",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .insert, .text = "123" },
            .{ .operation = .equal, .text = "b" },
            .{ .operation = .insert, .text = "456" },
            .{ .operation = .equal, .text = "c" },
        },
    }});

    // Two deletions.
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "a123b456c",
        .after = "abc",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "123" },
            .{ .operation = .equal, .text = "b" },
            .{ .operation = .delete, .text = "456" },
            .{ .operation = .equal, .text = "c" },
        },
    }});

    // Simple case #1
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "a",
        .after = "b",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .insert, .text = "b" },
        },
    }});

    // Simple case #2
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "Apples are a fruit.",
        .after = "Bananas are also fruit.",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .delete, .text = "Apple" },
            .{ .operation = .insert, .text = "Banana" },
            .{ .operation = .equal, .text = "s are a" },
            .{ .operation = .insert, .text = "lso" },
            .{ .operation = .equal, .text = " fruit." },
        },
    }});

    // Simple case #3
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "ax\t",
        .after = "\u{0680}x\x00",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .insert, .text = "\u{0680}" },
            .{ .operation = .equal, .text = "x" },
            .{ .operation = .delete, .text = "\t" },
            .{ .operation = .insert, .text = "\x00" },
        },
    }});

    // Overlap #1
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "1ayb2",
        .after = "abxab",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .delete, .text = "1" },
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "y" },
            .{ .operation = .equal, .text = "b" },
            .{ .operation = .delete, .text = "2" },
            .{ .operation = .insert, .text = "xab" },
        },
    }});

    // Overlap #2
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "abcy",
        .after = "xaxcxabc",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .insert, .text = "xaxcx" },
            .{ .operation = .equal, .text = "abc" },
            .{ .operation = .delete, .text = "y" },
        },
    }});

    // Overlap #3
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "ABCDa=bcd=efghijklmnopqrsEFGHIJKLMNOefg",
        .after = "a-bcd-efghijklmnopqrs",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .delete, .text = "ABCD" },
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .delete, .text = "=" },
            .{ .operation = .insert, .text = "-" },
            .{ .operation = .equal, .text = "bcd" },
            .{ .operation = .delete, .text = "=" },
            .{ .operation = .insert, .text = "-" },
            .{ .operation = .equal, .text = "efghijklmnopqrs" },
            .{ .operation = .delete, .text = "EFGHIJKLMNOefg" },
        },
    }});

    // Large equality
    try checkAllAllocationFailures(testing.allocator, testDiff, .{.{
        .before = "a [[Pennsylvania]] and [[New",
        .after = " and [[Pennsylvania]]",
        .check_lines = false,
        .expected = &.{
            .{ .operation = .insert, .text = " " },
            .{ .operation = .equal, .text = "a" },
            .{ .operation = .insert, .text = "nd" },
            .{ .operation = .equal, .text = " [[Pennsylvania]]" },
            .{ .operation = .delete, .text = " and [[New" },
        },
    }});

    const dmp: DiffMatchPatch = .initDefault(testing.io, testing.allocator);
    const allocator = testing.allocator;
    // TODO these tests should be checked for allocation failure

    // Increase the text lengths by 1024 times to ensure a timeout.
    {
        const a_text: *const [128]u8 = "`Twas brillig, and the slithy toves\nDid gyre and gimble in the wabe:\nAll mimsy were the borogoves,\nAnd the mome raths outgrabe.\n";
        const b_text: *const [209]u8 = "I am the very model of a modern major general,\nI've information vegetable, animal, and mineral,\nI know the kings of England, and I quote the fights historical,\nFrom Marathon to Waterloo, in order categorical.\n";
        const a: []const u8 = @ptrCast(&@as([1024][a_text.len]u8, @splat(a_text.*)));
        const b: []const u8 = @ptrCast(&@as([1024][b_text.len]u8, @splat(b_text.*)));

        const timeout: std.Io.Timeout = .{ .duration = .{ .clock = .awake, .raw = .fromMilliseconds(100) } };

        var timer: std.Io.Timestamp = .now(testing.io, .awake);
        {
            var time_diff = try dmp.diff(a, b, false, timeout);
            defer deinitDiffList(allocator, &time_diff);
        }
        const duration = timer.untilNow(testing.io, .awake);

        // Test that we took at least the timeout period.
        try testing.expect(timeout.duration.raw.toNanoseconds() <= duration.toNanoseconds()); // diff: Timeout min.
        // Test that we didn't take forever (be forgiving).
        // Theoretically this test could fail very occasionally if the
        // OS task swaps or locks up for a second at the wrong moment.
        try testing.expect((timeout.duration.raw.toNanoseconds()) * 10000 * 2 > duration.toNanoseconds()); // diff: Timeout max.
    }

    {
        // Test the linemode speedup.
        // Must be long to pass the 100 char cutoff.
        const a = "1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n";
        const b = "abcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\nabcdefghij\n";

        var diff_checked = try dmp.diff(a, b, true, .none);
        defer deinitDiffList(allocator, &diff_checked);

        var diff_unchecked = try dmp.diff(a, b, false, .none);
        defer deinitDiffList(allocator, &diff_unchecked);

        try testing.expectEqualDeep(diff_checked.items, diff_unchecked.items); // diff: Simple line-mode.
    }

    {
        const a = "1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890";
        const b = "abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij";

        var diff_checked = try dmp.diff(a, b, true, .none);
        defer deinitDiffList(allocator, &diff_checked);

        var diff_unchecked = try dmp.diff(a, b, false, .none);
        defer deinitDiffList(allocator, &diff_unchecked);

        try testing.expectEqualDeep(diff_checked.items, diff_unchecked.items); // diff: Single line-mode.
    }

    {
        // diff: Overlap line-mode.
        const a = "1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n1234567890\n";
        const b = "abcdefghij\n1234567890\n1234567890\n1234567890\nabcdefghij\n1234567890\n1234567890\n1234567890\nabcdefghij\n1234567890\n1234567890\n1234567890\nabcdefghij\n";

        var diffs_linemode = try dmp.diff(a, b, true, .none);
        defer deinitDiffList(allocator, &diffs_linemode);

        const texts_linemode = try rebuildtexts(allocator, diffs_linemode);
        defer {
            allocator.free(texts_linemode[0]);
            allocator.free(texts_linemode[1]);
        }

        var diffs_textmode = try dmp.diff(a, b, false, .none);
        defer deinitDiffList(allocator, &diffs_textmode);

        const texts_textmode = try rebuildtexts(allocator, diffs_textmode);
        defer {
            allocator.free(texts_textmode[0]);
            allocator.free(texts_textmode[1]);
        }

        try testing.expectEqualStrings(texts_textmode[0], texts_linemode[0]);
        try testing.expectEqualStrings(texts_textmode[1], texts_linemode[1]);
    }
}

fn testDiffLineMode(
    allocator: Allocator,
    before: []const u8,
    after: []const u8,
) !void {
    var dmp: DiffMatchPatch = .initDefault(testing.io, allocator);
    dmp.diff_check_lines_over = 20;

    var diff_checked = try dmp.diff(before, after, true, .none);
    defer deinitDiffList(allocator, &diff_checked);

    var diff_unchecked = try dmp.diff(before, after, false, .none);
    defer deinitDiffList(allocator, &diff_unchecked);

    try testing.expectEqualDeep(diff_checked.items, diff_unchecked.items); // diff: Simple line-mode.
    dmp.diff_check_lines_over = 100;
}

test "diffLineMode" {
    try checkAllAllocationFailures(
        testing.allocator,
        testDiffLineMode,
        .{
            "1234567890\n1234567890\n1234567890\n",
            "abcdefghij\nabcdefghij\nabcdefghij\n",
        },
    );
}

fn testDiffCleanupSemantic(
    allocator: Allocator,
    params: struct {
        input: []const Diff,
        expected: []const Diff,
    },
) !void {
    var diffs: DiffList = try .initCapacity(allocator, params.input.len);
    defer deinitDiffList(allocator, &diffs);

    for (params.input) |item| {
        diffs.appendAssumeCapacity(.{ .operation = item.operation, .text = try allocator.dupe(u8, item.text) });
    }

    try diffCleanupSemantic(allocator, &diffs);

    try testing.expectEqualDeep(params.expected, diffs.items);
}

test diffCleanupSemantic {
    // Null case.
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &[_]Diff{},
        .expected = &[_]Diff{},
    }});

    // No elimination #1
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "cd" },
            .{ .operation = .equal, .text = "12" },
            .{ .operation = .delete, .text = "e" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "cd" },
            .{ .operation = .equal, .text = "12" },
            .{ .operation = .delete, .text = "e" },
        },
    }});

    // No elimination #2
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "abc" },
            .{ .operation = .insert, .text = "ABC" },
            .{ .operation = .equal, .text = "1234" },
            .{ .operation = .delete, .text = "wxyz" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abc" },
            .{ .operation = .insert, .text = "ABC" },
            .{ .operation = .equal, .text = "1234" },
            .{ .operation = .delete, .text = "wxyz" },
        },
    }});

    // Simple elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "a" },
            .{ .operation = .equal, .text = "b" },
            .{ .operation = .delete, .text = "c" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abc" },
            .{ .operation = .insert, .text = "b" },
        },
    }});

    // Backpass elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .equal, .text = "cd" },
            .{ .operation = .delete, .text = "e" },
            .{ .operation = .equal, .text = "f" },
            .{ .operation = .insert, .text = "g" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abcdef" },
            .{ .operation = .insert, .text = "cdfg" },
        },
    }});

    // Multiple elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .insert, .text = "1" },
            .{ .operation = .equal, .text = "A" },
            .{ .operation = .delete, .text = "B" },
            .{ .operation = .insert, .text = "2" },
            .{ .operation = .equal, .text = "_" },
            .{ .operation = .insert, .text = "1" },
            .{ .operation = .equal, .text = "A" },
            .{ .operation = .delete, .text = "B" },
            .{ .operation = .insert, .text = "2" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "AB_AB" },
            .{ .operation = .insert, .text = "1A2_1A2" },
        },
    }});

    // Word boundaries
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .equal, .text = "The c" },
            .{ .operation = .delete, .text = "ow and the c" },
            .{ .operation = .equal, .text = "at." },
        },
        .expected = &.{
            .{ .operation = .equal, .text = "The " },
            .{ .operation = .delete, .text = "cow and the " },
            .{ .operation = .equal, .text = "cat." },
        },
    }});

    // No overlap elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "abcxx" },
            .{ .operation = .insert, .text = "xxdef" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abcxx" },
            .{ .operation = .insert, .text = "xxdef" },
        },
    }});

    // Overlap elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "abcxxx" },
            .{ .operation = .insert, .text = "xxxdef" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abc" },
            .{ .operation = .equal, .text = "xxx" },
            .{ .operation = .insert, .text = "def" },
        },
    }});

    // Reverse overlap elimination
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "xxxabc" },
            .{ .operation = .insert, .text = "defxxx" },
        },
        .expected = &.{
            .{ .operation = .insert, .text = "def" },
            .{ .operation = .equal, .text = "xxx" },
            .{ .operation = .delete, .text = "abc" },
        },
    }});

    // Two overlap eliminations
    try checkAllAllocationFailures(testing.allocator, testDiffCleanupSemantic, .{.{
        .input = &.{
            .{ .operation = .delete, .text = "abcd1212" },
            .{ .operation = .insert, .text = "1212efghi" },
            .{ .operation = .equal, .text = "----" },
            .{ .operation = .delete, .text = "A3" },
            .{ .operation = .insert, .text = "3BC" },
        },
        .expected = &.{
            .{ .operation = .delete, .text = "abcd" },
            .{ .operation = .equal, .text = "1212" },
            .{ .operation = .insert, .text = "efghi" },
            .{ .operation = .equal, .text = "----" },
            .{ .operation = .delete, .text = "A" },
            .{ .operation = .equal, .text = "3" },
            .{ .operation = .insert, .text = "BC" },
        },
    }});
}

fn testDiffCleanupEfficiency(
    allocator: Allocator,
    dmp: *const DiffMatchPatch,
    params: struct {
        input: []const Diff,
        expected: []const Diff,
    },
) !void {
    var diffs: DiffList = try .initCapacity(allocator, params.input.len);
    defer deinitDiffList(allocator, &diffs);
    for (params.input) |item| {
        diffs.appendAssumeCapacity(.{ .operation = item.operation, .text = try allocator.dupe(u8, item.text) });
    }
    try dmp.diffCleanupEfficiency(&diffs);

    try testing.expectEqualDeep(params.expected, diffs.items);
}

test "diffCleanupEfficiency" {
    const allocator = testing.allocator;
    var dmp: DiffMatchPatch = .initDefault(testing.io, allocator);
    dmp.diff_edit_cost = 4;
    { // Null case.
        var diffs: DiffList = .empty;
        try dmp.diffCleanupEfficiency(&diffs);
        try testing.expectEqualDeep(DiffList.empty, diffs);
    }
    { // No elimination.
        const dslice: []const Diff = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "12" },
            .{ .operation = .equal, .text = "wxyz" },
            .{ .operation = .delete, .text = "cd" },
            .{ .operation = .insert, .text = "34" },
        };
        try checkAllAllocationFailures(
            testing.allocator,
            testDiffCleanupEfficiency,
            .{
                &dmp,
                .{ .input = dslice, .expected = dslice },
            },
        );
    }
    { // Four-edit elimination.
        const dslice: []const Diff = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "12" },
            .{ .operation = .equal, .text = "xyz" },
            .{ .operation = .delete, .text = "cd" },
            .{ .operation = .insert, .text = "34" },
        };
        const d_after: []const Diff = &.{
            .{ .operation = .delete, .text = "abxyzcd" },
            .{ .operation = .insert, .text = "12xyz34" },
        };
        try checkAllAllocationFailures(
            testing.allocator,
            testDiffCleanupEfficiency,
            .{
                &dmp,
                .{ .input = dslice, .expected = d_after },
            },
        );
    }
    { // Three-edit elimination.
        const dslice: []const Diff = &.{
            .{ .operation = .insert, .text = "12" },
            .{ .operation = .equal, .text = "x" },
            .{ .operation = .delete, .text = "cd" },
            .{ .operation = .insert, .text = "34" },
        };
        const d_after: []const Diff = &.{
            .{ .operation = .delete, .text = "xcd" },
            .{ .operation = .insert, .text = "12x34" },
        };
        try checkAllAllocationFailures(
            testing.allocator,
            testDiffCleanupEfficiency,
            .{
                &dmp,
                .{ .input = dslice, .expected = d_after },
            },
        );
    }
    { // Backpass elimination.
        const dslice: []const Diff = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "12" },
            .{ .operation = .equal, .text = "xy" },
            .{ .operation = .insert, .text = "34" },
            .{ .operation = .equal, .text = "z" },
            .{ .operation = .delete, .text = "cd" },
            .{ .operation = .insert, .text = "56" },
        };
        const d_after: []const Diff = &.{
            .{ .operation = .delete, .text = "abxyzcd" },
            .{ .operation = .insert, .text = "12xy34z56" },
        };
        try checkAllAllocationFailures(
            testing.allocator,
            testDiffCleanupEfficiency,
            .{
                &dmp,
                .{ .input = dslice, .expected = d_after },
            },
        );
    }
    { // High cost elimination.
        dmp.diff_edit_cost = 5;
        const dslice: []const Diff = &.{
            .{ .operation = .delete, .text = "ab" },
            .{ .operation = .insert, .text = "12" },
            .{ .operation = .equal, .text = "wxyz" },
            .{ .operation = .delete, .text = "cd" },
            .{ .operation = .insert, .text = "34" },
        };
        const d_after: []const Diff = &.{
            .{ .operation = .delete, .text = "abwxyzcd" },
            .{ .operation = .insert, .text = "12wxyz34" },
        };
        try checkAllAllocationFailures(
            testing.allocator,
            testDiffCleanupEfficiency,
            .{
                &dmp,
                .{ .input = dslice, .expected = d_after },
            },
        );
        dmp.diff_edit_cost = 4;
    }
}

/// https://github.com/ziglang/zig/pull/23042/files
fn checkAllAllocationFailures(
    backing_allocator: Allocator,
    comptime test_fn: anytype,
    extra_args: CheckAllAllocationFailuresTuples(@TypeOf(test_fn)).ExtraArgsTuple,
) !void {
    return std.testing.checkAllAllocationFailures(backing_allocator, test_fn, extra_args);
}

fn CheckAllAllocationFailuresTuples(comptime TestFn: type) struct {
    /// `std.meta.ArgsTuple(TestFn)`
    ArgsTuple: type,
    /// `std.meta.ArgsTuple(TestFn)` without the first argument
    ExtraArgsTuple: type,
} {
    switch (@typeInfo(@typeInfo(TestFn).@"fn".return_type.?)) {
        .error_union => |info| {
            if (info.payload != void) {
                @compileError("Return type must be !void");
            }
        },
        else => @compileError("Return type must be !void"),
    }

    const ArgsTuple = std.meta.ArgsTuple(TestFn);

    const fn_args_fields = std.meta.fields(ArgsTuple);
    if (fn_args_fields.len == 0 or fn_args_fields[0].type != Allocator) {
        @compileError("The provided function must have an " ++ @typeName(Allocator) ++ " as its first argument");
    }

    // remove the first tuple field (`Allocator`)
    var extra_args_tuple_info = @typeInfo(ArgsTuple).@"struct";
    var field_types: [extra_args_tuple_info.fields.len - 1]type = undefined;
    for (&field_types, extra_args_tuple_info.fields[1..]) |*ty, field| {
        ty.* = field.type;
    }
    const ExtraArgsTuple = @Tuple(&field_types);

    return .{
        .ArgsTuple = ArgsTuple,
        .ExtraArgsTuple = ExtraArgsTuple,
    };
}