Merge crates

Cargo.lock

@@ -2,12 +2,6 @@
 # It is not intended for manual editing.
 version = 4
 
-[[package]]
-name = "base64"
-version = "0.22.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "72b3254f16251a8381aa12e40e3c4d2f0199f8c6508fbecb9d91f575e0fbb8c6"
-
 [[package]]
 name = "bumpalo"
 version = "3.16.0"
@@ -206,11 +200,14 @@ dependencies = [
 name = "reconcile"
 version = "0.4.0"
 dependencies = [
+ "console_error_panic_hook",
  "insta",
  "pretty_assertions",
  "serde",
  "serde_yaml",
  "test-case",
+ "wasm-bindgen",
+ "wasm-bindgen-test",
 ]
 
 [[package]]
@@ -290,19 +287,6 @@ dependencies = [
  "unicode-ident",
 ]
 
-[[package]]
-name = "sync_lib"
-version = "0.4.0"
-dependencies = [
- "base64",
- "console_error_panic_hook",
- "insta",
- "reconcile",
- "thiserror",
- "wasm-bindgen",
- "wasm-bindgen-test",
-]
-
 [[package]]
 name = "test-case"
 version = "3.3.1"
@@ -336,26 +320,6 @@ dependencies = [
  "test-case-core",
 ]
 
-[[package]]
-name = "thiserror"
-version = "2.0.12"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "567b8a2dae586314f7be2a752ec7474332959c6460e02bde30d702a66d488708"
-dependencies = [
- "thiserror-impl",
-]
-
-[[package]]
-name = "thiserror-impl"
-version = "2.0.12"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f7cf42b4507d8ea322120659672cf1b9dbb93f8f2d4ecfd6e51350ff5b17a1d"
-dependencies = [
- "proc-macro2",
- "quote",
- "syn",
-]
-
 [[package]]
 name = "unicode-ident"
 version = "1.0.14"

Cargo.toml

@@ -1,21 +1,38 @@
-[workspace]
+[package]
-resolver = "2"
+name = "reconcile"
-members = [
+rust-version = "1.85"
-    "reconcile",
-    "sync_lib"
-]
-
-[workspace.package]
-rust-version = "1.83"
 authors = ["Andras Schmelczer <andras@schmelczer.dev>"]
 edition = "2024"
 license = "MIT"
-repository = "https://github.com/schmelczer/vault-link"
+repository = "https://github.com/schmelczer/reconcile"
 version = "0.4.0"
 
-[workspace.dependencies]
+[lib]
-serde = { version = "1.0.219", default-features = false, features = ["derive"] }
+crate-type = ["cdylib", "rlib"]
-thiserror = { version = "2.0.12", default-features = false  }
+
+[dependencies]
+serde = { version = "1.0.219", optional = true, features = ["derive"] }
+
+wasm-bindgen = { version = "0.2.99", optional = true }
+
+# The `console_error_panic_hook` crate provides better debugging of panics by
+# logging them with `console.error`. This is great for development, but requires
+# all the `std::fmt` and `std::panicking` infrastructure, so isn't great for
+# code size when deploying.
+console_error_panic_hook = { version = "0.1.7", optional = true }
+
+[features]
+default = ["serde", "wasm"]
+serde = [ "dep:serde" ]
+wasm = [ "dep:wasm-bindgen", "dep:console_error_panic_hook" ]
+
+[dev-dependencies]
+insta = "1.42.2"
+pretty_assertions = "1.4.1"
+serde = { version = "1.0.219", features = ["derive"] }
+serde_yaml ="0.9.34"
+test-case = "3.3.1"
+wasm-bindgen-test = "0.3.49"
 
 [profile.release]
 codegen-units = 1

reconcile/Cargo.toml

@@ -1,23 +0,0 @@
-[package]
-name = "reconcile"
-version.workspace = true
-edition.workspace = true
-authors.workspace = true
-license.workspace = true
-repository.workspace = true
-
-[dependencies]
-serde = { version = "1.0.219", optional = true, features = ["derive"] }
-
-[features]
-serde = [ "dep:serde" ]
-
-[dev-dependencies]
-insta = "1.42.2"
-pretty_assertions = "1.4.1"
-serde = { version = "1.0.219", features = ["derive"] }
-serde_yaml ="0.9.34"
-test-case = "3.3.1"
-
-[lints]
-workspace = true

src/diffs.rs

@@ -0,0 +1,2 @@
+pub mod myers;
+pub mod raw_operation;

src/diffs/myers.rs

@@ -0,0 +1,357 @@
+//! Taken from <https://github.com/mitsuhiko/similar/blob/7e15c44de11a1cd61e1149189929e189ef977fd8/src/algorithms/myers.rs>
+//!
+//! Myers' diff algorithm.
+//!
+//! * time: `O((N+M)D)`
+//! * space `O(N+M)`
+//!
+//! See [the original article by Eugene W. Myers](http://www.xmailserver.org/diff2.pdf)
+//! describing it.
+//!
+//! The implementation of this algorithm is based on the implementation by
+//! Brandon Williams.
+//!
+//! # Heuristics
+//!
+//! At present this implementation of Myers' does not implement any more
+//! advanced heuristics that would solve some pathological cases.  For instance
+//! passing two large and completely distinct sequences to the algorithm will
+//! make it spin without making reasonable progress.
+//! For potential improvements here see [similar#15](https://github.com/mitsuhiko/similar/issues/15).
+
+use std::{
+    ops::{Index, IndexMut, Range},
+    vec,
+};
+
+use super::raw_operation::RawOperation;
+use crate::{
+    tokenizer::token::Token,
+    utils::{common_prefix_len::common_prefix_len, common_suffix_len::common_suffix_len},
+};
+
+/// Myers' diff algorithm.
+///
+/// Diff `old`, between indices `old_range` and `new` between indices
+/// `new_range`.
+///
+/// The returned `RawOperations` all have a token count of 1.
+pub fn diff<T>(old: &[Token<T>], new: &[Token<T>]) -> Vec<RawOperation<T>>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    let max_d = (old.len() + new.len()).div_ceil(2) + 1;
+    let mut vb = V::new(max_d);
+    let mut vf = V::new(max_d);
+    let mut result: Vec<RawOperation<T>> = vec![];
+
+    conquer(
+        old,
+        0..old.len(),
+        new,
+        0..new.len(),
+        &mut vf,
+        &mut vb,
+        &mut result,
+    );
+
+    debug_assert!(
+        result.iter().all(|op| op.tokens().len() == 1),
+        "All operations should be of length 1"
+    );
+
+    result
+}
+
+// A D-path is a path which starts at (0,0) that has exactly D non-diagonal
+// edges. All D-paths consist of a (D - 1)-path followed by a non-diagonal edge
+// and then a possibly empty sequence of diagonal edges called a snake.
+
+/// `V` contains the endpoints of the furthest reaching `D-paths`. For each
+/// recorded endpoint `(x,y)` in diagonal `k`, we only need to retain `x`
+/// because `y` can be computed from `x - k`. In other words, `V` is an array of
+/// integers where `V[k]` contains the row index of the endpoint of the furthest
+/// reaching path in diagonal `k`.
+///
+/// We can't use a traditional Vec to represent `V` since we use `k` as an index
+/// and it can take on negative values. So instead `V` is represented as a
+/// light-weight wrapper around a Vec plus an `offset` which is the maximum
+/// value `k` can take on in order to map negative `k`'s back to a value >= 0.
+#[derive(Debug)]
+struct V {
+    offset: isize,
+    v: Vec<usize>, // Look into initializing this to -1 and storing isize
+}
+
+impl V {
+    fn new(max_d: usize) -> Self {
+        Self {
+            offset: max_d as isize,
+            v: vec![0; 2 * max_d],
+        }
+    }
+
+    fn len(&self) -> usize { self.v.len() }
+}
+
+impl Index<isize> for V {
+    type Output = usize;
+
+    fn index(&self, index: isize) -> &Self::Output { &self.v[(index + self.offset) as usize] }
+}
+
+impl IndexMut<isize> for V {
+    fn index_mut(&mut self, index: isize) -> &mut Self::Output {
+        &mut self.v[(index + self.offset) as usize]
+    }
+}
+
+fn split_at(range: Range<usize>, at: usize) -> (Range<usize>, Range<usize>) {
+    (range.start..at, at..range.end)
+}
+
+/// A `Snake` is a sequence of diagonal edges in the edit graph.  Normally
+/// a snake has a start end end point (and it is possible for a snake to have
+/// a length of zero, meaning the start and end points are the same) however
+/// we do not need the end point which is why it's not implemented here.
+///
+/// The divide part of a divide-and-conquer strategy. A D-path has D+1 snakes
+/// some of which may be empty. The divide step requires finding the ceil(D/2) +
+/// 1 or middle snake of an optimal D-path. The idea for doing so is to
+/// simultaneously run the basic algorithm in both the forward and reverse
+/// directions until furthest reaching forward and reverse paths starting at
+/// opposing corners 'overlap'.
+fn find_middle_snake<T>(
+    old: &[Token<T>],
+    old_range: Range<usize>,
+    new: &[Token<T>],
+    new_range: Range<usize>,
+    vf: &mut V,
+    vb: &mut V,
+) -> Option<(usize, usize)>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    let n = old_range.len();
+    let m = new_range.len();
+
+    // By Lemma 1 in the paper, the optimal edit script length is odd or even as
+    // `delta` is odd or even.
+    let delta = n as isize - m as isize;
+    let odd = delta & 1 == 1;
+
+    // The initial point at (0, -1)
+    vf[1] = 0;
+    // The initial point at (N, M+1)
+    vb[1] = 0;
+
+    let d_max = (n + m).div_ceil(2) + 1;
+    assert!(vf.len() >= d_max);
+    assert!(vb.len() >= d_max);
+
+    for d in 0..d_max as isize {
+        // Forward path
+        for k in (-d..=d).rev().step_by(2) {
+            let mut x = if k == -d || (k != d && vf[k - 1] < vf[k + 1]) {
+                vf[k + 1]
+            } else {
+                vf[k - 1] + 1
+            };
+            let y = (x as isize - k) as usize;
+
+            // The coordinate of the start of a snake
+            let (x0, y0) = (x, y);
+            //  While these sequences are identical, keep moving through the
+            //  graph with no cost
+            if x < old_range.len() && y < new_range.len() {
+                let advance = common_prefix_len(
+                    old,
+                    old_range.start + x..old_range.end,
+                    new,
+                    new_range.start + y..new_range.end,
+                );
+                x += advance;
+            }
+
+            // This is the new best x value
+            vf[k] = x;
+
+            // Only check for connections from the forward search when N - M is
+            // odd and when there is a reciprocal k line coming from the other
+            // direction.
+            if odd && (k - delta).abs() <= (d - 1) {
+                // TODO optimize this so we don't have to compare against n
+                if vf[k] + vb[-(k - delta)] >= n {
+                    // Return the snake
+                    return Some((x0 + old_range.start, y0 + new_range.start));
+                }
+            }
+        }
+
+        // Backward path
+        for k in (-d..=d).rev().step_by(2) {
+            let mut x = if k == -d || (k != d && vb[k - 1] < vb[k + 1]) {
+                vb[k + 1]
+            } else {
+                vb[k - 1] + 1
+            };
+            let mut y = (x as isize - k) as usize;
+
+            // The coordinate of the start of a snake
+            if x < n && y < m {
+                let advance = common_suffix_len(
+                    old,
+                    old_range.start..old_range.start + n - x,
+                    new,
+                    new_range.start..new_range.start + m - y,
+                );
+                x += advance;
+                y += advance;
+            }
+
+            // This is the new best x value
+            vb[k] = x;
+
+            if !odd && (k - delta).abs() <= d {
+                // TODO optimize this so we don't have to compare against n
+                if vb[k] + vf[-(k - delta)] >= n {
+                    // Return the snake
+                    return Some((n - x + old_range.start, m - y + new_range.start));
+                }
+            }
+        }
+
+        // TODO: Maybe there's an opportunity to optimize and bail early?
+    }
+
+    None
+}
+
+fn conquer<T>(
+    old: &[Token<T>],
+    mut old_range: Range<usize>,
+    new: &[Token<T>],
+    mut new_range: Range<usize>,
+    vf: &mut V,
+    vb: &mut V,
+    result: &mut Vec<RawOperation<T>>,
+) where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    // Check for common prefix
+    let common_prefix_len = common_prefix_len(old, old_range.clone(), new, new_range.clone());
+    if common_prefix_len > 0 {
+        result.extend(
+            old[old_range.start..old_range.start + common_prefix_len]
+                .iter()
+                .map(|token| RawOperation::Equal(vec![token.clone()])),
+        );
+    }
+    old_range.start += common_prefix_len;
+    new_range.start += common_prefix_len;
+
+    // Check for common suffix
+    let common_suffix_len = common_suffix_len(old, old_range.clone(), new, new_range.clone());
+    let common_suffix = (
+        old_range.end - common_suffix_len,
+        new_range.end - common_suffix_len,
+    );
+    old_range.end -= common_suffix_len;
+    new_range.end -= common_suffix_len;
+
+    if old_range.is_empty() && new_range.is_empty() {
+        // do nothing
+    } else if new_range.is_empty() {
+        result.extend(
+            old[old_range.start..old_range.start + old_range.len()]
+                .iter()
+                .map(|token| RawOperation::Delete(vec![token.clone()])),
+        );
+    } else if old_range.is_empty() {
+        result.extend(
+            new[new_range.start..new_range.start + new_range.len()]
+                .iter()
+                .map(|token| RawOperation::Insert(vec![token.clone()])),
+        );
+    } else if let Some((x_start, y_start)) =
+        find_middle_snake(old, old_range.clone(), new, new_range.clone(), vf, vb)
+    {
+        let (old_a, old_b) = split_at(old_range, x_start);
+        let (new_a, new_b) = split_at(new_range, y_start);
+        conquer(old, old_a, new, new_a, vf, vb, result);
+        conquer(old, old_b, new, new_b, vf, vb, result);
+    } else {
+        result.extend(
+            old[old_range.start..old_range.end]
+                .iter()
+                .map(|token| RawOperation::Delete(vec![token.clone()])),
+        );
+        result.extend(
+            new[new_range.start..new_range.end]
+                .iter()
+                .map(|token| RawOperation::Insert(vec![token.clone()])),
+        );
+    }
+
+    if common_suffix_len > 0 {
+        result.extend(
+            old[common_suffix.0..common_suffix.0 + common_suffix_len]
+                .iter()
+                .map(|token| RawOperation::Equal(vec![token.clone()])),
+        );
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use insta::assert_debug_snapshot;
+
+    use super::*;
+
+    #[test]
+    fn test_empty_diff() {
+        let old: Vec<Token<String>> = vec![];
+        let new: Vec<Token<String>> = vec![];
+        let result = diff(&old, &new);
+        assert_eq!(result.len(), 0);
+    }
+
+    #[test]
+    fn test_identical_content() {
+        let content = vec!["a".into(), "b".into(), "c".into()];
+        let result = diff(&content, &content);
+        assert_debug_snapshot!(result);
+    }
+
+    #[test]
+    fn test_insert_only() {
+        let old: Vec<Token<String>> = vec![];
+        let new: Vec<Token<String>> = vec!["a".into(), "b".into()];
+        let result = diff(&old, &new);
+        assert_debug_snapshot!(result);
+    }
+
+    #[test]
+    fn test_delete_only() {
+        let old = vec!["a".into(), "b".into()];
+        let new: Vec<Token<String>> = vec![];
+        let result = diff(&old, &new);
+        assert_debug_snapshot!(result);
+    }
+
+    #[test]
+    fn test_prefix_and_suffix() {
+        let old = vec!["a".into(), "b".into(), "c".into(), "d".into()];
+        let new = vec!["a".into(), "x".into(), "d".into()];
+        let result = diff(&old, &new);
+        assert_debug_snapshot!(result);
+    }
+
+    #[test]
+    fn test_complex_diff() {
+        let old = vec!["a".into(), "b".into(), "c".into(), "d".into()];
+        let new = vec!["a".into(), "x".into(), "c".into(), "y".into()];
+        let result = diff(&old, &new);
+        assert_debug_snapshot!(result);
+    }
+}

src/diffs/raw_operation.rs

@@ -0,0 +1,64 @@
+use crate::tokenizer::token::Token;
+
+#[derive(Debug, Clone, PartialEq)]
+pub enum RawOperation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    Insert(Vec<Token<T>>),
+    Delete(Vec<Token<T>>),
+    Equal(Vec<Token<T>>),
+}
+
+impl<T> RawOperation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    pub fn tokens(&self) -> &Vec<Token<T>> {
+        match self {
+            RawOperation::Insert(tokens)
+            | RawOperation::Delete(tokens)
+            | RawOperation::Equal(tokens) => tokens,
+        }
+    }
+
+    pub fn original_text_length(&self) -> usize {
+        self.tokens().iter().map(Token::get_original_length).sum()
+    }
+
+    pub fn get_original_text(self) -> String { self.tokens().iter().map(Token::original).collect() }
+
+    pub fn is_left_joinable(&self) -> bool {
+        let first_token = self.tokens().first();
+        first_token.is_none_or(|token| token.is_left_joinable)
+    }
+
+    pub fn is_right_joinable(&self) -> bool {
+        let last_token = self.tokens().last();
+        last_token.is_none_or(|token| token.is_right_joinable)
+    }
+
+    /// Extends the operation with another operation. Only operations of the
+    /// same type as self can be used to extend self, otherwise the function
+    /// will panic.
+    pub fn extend(self, other: RawOperation<T>) -> RawOperation<T> {
+        debug_assert!(
+            std::mem::discriminant(&self) == std::mem::discriminant(&other),
+            "Cannot extend operations of different types. This should have been handled before \
+             calling this function."
+        );
+
+        match (self, other) {
+            (RawOperation::Insert(self_tokens), RawOperation::Insert(other_tokens)) => {
+                RawOperation::Insert(self_tokens.into_iter().chain(other_tokens).collect())
+            }
+            (RawOperation::Delete(tokens1), RawOperation::Delete(tokens2)) => {
+                RawOperation::Delete(tokens1.into_iter().chain(tokens2).collect())
+            }
+            (RawOperation::Equal(tokens1), RawOperation::Equal(tokens2)) => {
+                RawOperation::Equal(tokens1.into_iter().chain(tokens2).collect())
+            }
+            _ => unreachable!("Only operations of the same type can be extended"),
+        }
+    }
+}

src/diffs/snapshots/reconcile__diffs__myers__tests__complex_diff.snap

@@ -0,0 +1,67 @@
+---
+source: reconcile/src/diffs/myers.rs
+expression: result
+snapshot_kind: text
+---
+[
+    Equal(
+        [
+            Token {
+                normalised: "a",
+                original: "a",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Insert(
+        [
+            Token {
+                normalised: "x",
+                original: "x",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Delete(
+        [
+            Token {
+                normalised: "b",
+                original: "b",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Equal(
+        [
+            Token {
+                normalised: "c",
+                original: "c",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Insert(
+        [
+            Token {
+                normalised: "y",
+                original: "y",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Delete(
+        [
+            Token {
+                normalised: "d",
+                original: "d",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+]

src/diffs/snapshots/reconcile__diffs__myers__tests__delete_only.snap

@@ -0,0 +1,27 @@
+---
+source: reconcile/src/diffs/myers.rs
+expression: result
+snapshot_kind: text
+---
+[
+    Delete(
+        [
+            Token {
+                normalised: "a",
+                original: "a",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Delete(
+        [
+            Token {
+                normalised: "b",
+                original: "b",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+]

src/diffs/snapshots/reconcile__diffs__myers__tests__identical_content.snap

@@ -0,0 +1,37 @@
+---
+source: reconcile/src/diffs/myers.rs
+expression: result
+snapshot_kind: text
+---
+[
+    Equal(
+        [
+            Token {
+                normalised: "a",
+                original: "a",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Equal(
+        [
+            Token {
+                normalised: "b",
+                original: "b",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Equal(
+        [
+            Token {
+                normalised: "c",
+                original: "c",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+]

src/diffs/snapshots/reconcile__diffs__myers__tests__insert_only.snap

@@ -0,0 +1,27 @@
+---
+source: reconcile/src/diffs/myers.rs
+expression: result
+snapshot_kind: text
+---
+[
+    Insert(
+        [
+            Token {
+                normalised: "a",
+                original: "a",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Insert(
+        [
+            Token {
+                normalised: "b",
+                original: "b",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+]

src/diffs/snapshots/reconcile__diffs__myers__tests__prefix_and_suffix.snap

@@ -0,0 +1,57 @@
+---
+source: reconcile/src/diffs/myers.rs
+expression: result
+snapshot_kind: text
+---
+[
+    Equal(
+        [
+            Token {
+                normalised: "a",
+                original: "a",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Delete(
+        [
+            Token {
+                normalised: "b",
+                original: "b",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Delete(
+        [
+            Token {
+                normalised: "c",
+                original: "c",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Insert(
+        [
+            Token {
+                normalised: "x",
+                original: "x",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+    Equal(
+        [
+            Token {
+                normalised: "d",
+                original: "d",
+                is_left_joinable: true,
+                is_right_joinable: true,
+            },
+        ],
+    ),
+]

src/lib.rs

@@ -0,0 +1,13 @@
+mod diffs;
+mod operation_transformation;
+mod tokenizer;
+mod utils;
+
+pub use operation_transformation::{
+    CursorPosition, EditedText, TextWithCursors, reconcile, reconcile_with_cursors,
+    reconcile_with_tokenizer,
+};
+pub use tokenizer::{Tokenizer, token::Token, word_tokenizer::word_tokenizer};
+
+#[cfg(feature = "wasm")]
+pub mod wasm;

src/operation_transformation.rs

@@ -0,0 +1,167 @@
+mod cursor;
+mod edited_text;
+mod merge_context;
+mod operation;
+mod ordered_operation;
+mod utils;
+
+pub use cursor::{CursorPosition, TextWithCursors};
+pub use edited_text::EditedText;
+pub use operation::Operation;
+
+use crate::Tokenizer;
+
+#[must_use]
+pub fn reconcile(original: &str, left: &str, right: &str) -> String {
+    reconcile_with_cursors(original, left.into(), right.into())
+        .text
+        .to_string()
+}
+
+#[must_use]
+pub fn reconcile_with_cursors<'a>(
+    original: &'a str,
+    left: TextWithCursors<'a>,
+    right: TextWithCursors<'a>,
+) -> TextWithCursors<'static> {
+    let left_operations = EditedText::from_strings(original, left);
+    let right_operations = EditedText::from_strings(original, right);
+
+    let merged_operations = left_operations.merge(right_operations);
+
+    TextWithCursors::new_owned(merged_operations.apply(), merged_operations.cursors)
+}
+
+#[must_use]
+pub fn reconcile_with_tokenizer<'a, F, T>(
+    original: &str,
+    left: TextWithCursors<'a>,
+    right: TextWithCursors<'a>,
+    tokenizer: &Tokenizer<T>,
+) -> TextWithCursors<'static>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    let left_operations = EditedText::from_strings_with_tokenizer(original, left, tokenizer);
+    let right_operations = EditedText::from_strings_with_tokenizer(original, right, tokenizer);
+
+    let merged_operations = left_operations.merge(right_operations);
+
+    TextWithCursors::new_owned(merged_operations.apply(), merged_operations.cursors)
+}
+
+#[cfg(test)]
+mod test {
+    use std::{fs, ops::Range, path::Path};
+
+    use pretty_assertions::assert_eq;
+    use test_case::test_matrix;
+
+    use super::*;
+    use crate::CursorPosition;
+
+    #[test]
+    fn test_cursor_complex() {
+        let original = "this is some complex text to test cursor positions";
+        let left = TextWithCursors::new(
+            "this is really complex text for testing cursor positions",
+            vec![
+                CursorPosition {
+                    id: 0,
+                    char_index: 8,
+                }, // after "this is "
+                CursorPosition {
+                    id: 1,
+                    char_index: 22,
+                }, // after "this is really complex text"
+            ],
+        );
+        let right = TextWithCursors::new(
+            "that was some complex sample to test cursor movements",
+            vec![
+                CursorPosition {
+                    id: 2,
+                    char_index: 5,
+                }, // after "that "
+                CursorPosition {
+                    id: 3,
+                    char_index: 29,
+                }, // after "some complex sample "
+            ],
+        );
+
+        let merged = reconcile_with_cursors(original, left, right);
+
+        assert_eq!(
+            merged,
+            TextWithCursors::new(
+                "that was really complex sample for testing cursor movements",
+                vec![
+                    CursorPosition {
+                        id: 2,
+                        char_index: 5
+                    }, // unchanged
+                    CursorPosition {
+                        id: 0,
+                        char_index: 9
+                    }, // before "really"
+                    CursorPosition {
+                        id: 1,
+                        char_index: 23
+                    }, // inside of "s|ample" because "text" got replaced by "sample"
+                    CursorPosition {
+                        id: 3,
+                        char_index: 43
+                    }, // before "cursor movements"
+                ]
+            )
+        );
+    }
+
+    #[ignore = "expensive to run, only run in CI"]
+    #[test_matrix( [
+        "pride_and_prejudice.txt",
+        "room_with_a_view.txt",
+        "kun_lu.txt",
+        "blns.txt"
+    ],  [
+        "pride_and_prejudice.txt",
+        "room_with_a_view.txt",
+        "kun_lu.txt",
+        "blns.txt"
+    ],  [
+        "pride_and_prejudice.txt",
+        "room_with_a_view.txt",
+        "kun_lu.txt",
+        "blns.txt"
+    ], [0..10000, 10000..20000], [0..10000, 10000..20000], [0..10000, 10000..20000])]
+    fn test_merge_files_without_panic(
+        file_name_1: &str,
+        file_name_2: &str,
+        file_name_3: &str,
+        range_1: Range<usize>,
+        range_2: Range<usize>,
+        range_3: Range<usize>,
+    ) {
+        let files = [file_name_1, file_name_2, file_name_3];
+        let permutations = [range_1, range_2, range_3];
+
+        let root = Path::new("tests/resources/");
+
+        let contents = files
+            .iter()
+            .zip(permutations.iter())
+            .map(|(file, range)| {
+                let path = root.join(file);
+                fs::read_to_string(&path)
+                    .unwrap()
+                    .chars()
+                    .skip(range.start)
+                    .take(range.end)
+                    .collect::<String>()
+            })
+            .collect::<Vec<_>>();
+
+        let _ = reconcile(&contents[0], &contents[1], &contents[2]);
+    }
+}

src/operation_transformation/cursor.rs

@@ -0,0 +1,57 @@
+use std::borrow::Cow;
+
+#[cfg(feature = "serde")]
+use serde::{Deserialize, Serialize};
+
+// CursorPosition represents the position of an identifiable cursor in a text
+// document based on its (UTF-8) character index.
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[derive(Debug, Clone, PartialEq, Default)]
+pub struct CursorPosition {
+    pub id: usize,
+    pub char_index: usize,
+}
+
+impl CursorPosition {
+    #[must_use]
+    pub fn with_index(&self, index: usize) -> Self {
+        CursorPosition {
+            id: self.id,
+            char_index: index,
+        }
+    }
+}
+
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[derive(Debug, Clone, PartialEq, Default)]
+pub struct TextWithCursors<'a> {
+    pub text: Cow<'a, str>,
+    pub cursors: Vec<CursorPosition>,
+}
+
+impl<'a> TextWithCursors<'a> {
+    #[must_use]
+    pub fn new(text: &'a str, cursors: Vec<CursorPosition>) -> Self {
+        Self {
+            text: text.into(),
+            cursors,
+        }
+    }
+
+    #[must_use]
+    pub fn new_owned(text: String, cursors: Vec<CursorPosition>) -> Self {
+        Self {
+            text: text.into(),
+            cursors,
+        }
+    }
+}
+
+impl<'a> From<&'a str> for TextWithCursors<'a> {
+    fn from(text: &'a str) -> Self {
+        Self {
+            text: text.into(),
+            cursors: Vec::new(),
+        }
+    }
+}

src/operation_transformation/edited_text.rs

@@ -0,0 +1,277 @@
+#[cfg(feature = "serde")]
+use serde::{Deserialize, Serialize};
+
+use super::{CursorPosition, Operation, TextWithCursors, ordered_operation::OrderedOperation};
+use crate::{
+    diffs::{myers::diff, raw_operation::RawOperation},
+    operation_transformation::{
+        merge_context::MergeContext,
+        utils::{cook_operations::cook_operations, elongate_operations::elongate_operations},
+    },
+    tokenizer::{Tokenizer, word_tokenizer::word_tokenizer},
+    utils::{side::Side, string_builder::StringBuilder},
+};
+
+/// A text document and a sequence of operations that can be applied to the text
+/// document. `EditedText` supports merging two sequences of operations using
+/// the principles of Operational Transformation.
+///
+/// It's mainly created through the `from_strings` method, then merged with
+/// another `EditedText` derived from the same original text and then applied to
+/// the original text to get the reconciled text of concurrent edits.
+///
+/// In addition to text and operations, it also keeps track of cursor positions
+/// in the original text. The cursor positions are updated when the operations
+/// are applied, so that the cursor positions can be used to restore the
+/// cursor positions in the updated text.
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[derive(Debug, Clone, PartialEq, Default)]
+pub struct EditedText<'a, T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    text: &'a str,
+    operations: Vec<OrderedOperation<T>>,
+    pub(crate) cursors: Vec<CursorPosition>,
+}
+
+impl<'a> EditedText<'a, String> {
+    /// Create an `EditedText` from the given original (old) and updated (new)
+    /// strings. The returned `EditedText` represents the changes from the
+    /// original to the updated text. When the return value is applied to
+    /// the original text, it will result in the updated text. The default
+    /// word tokenizer is used to tokenize the text which splits the text on
+    /// whitespaces.
+    #[must_use]
+    pub fn from_strings(original: &'a str, updated: TextWithCursors<'a>) -> Self {
+        Self::from_strings_with_tokenizer(original, updated, &word_tokenizer)
+    }
+}
+
+impl<'a, T> EditedText<'a, T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    /// Create an `EditedText` from the given original (old) and updated (new)
+    /// strings. The returned `EditedText` represents the changes from the
+    /// original to the updated text. When the return value is applied to
+    /// the original text, it will result in the updated text. The tokenizer
+    /// function is used to tokenize the text.
+    pub fn from_strings_with_tokenizer(
+        original: &'a str,
+        updated: TextWithCursors<'a>,
+        tokenizer: &Tokenizer<T>,
+    ) -> Self {
+        let original_tokens = (tokenizer)(original);
+        let updated_tokens = (tokenizer)(&updated.text);
+
+        let diff: Vec<RawOperation<T>> = diff(&original_tokens, &updated_tokens);
+
+        Self::new(
+            original,
+            cook_operations(elongate_operations(diff)).collect(),
+            updated.cursors,
+        )
+    }
+
+    /// Create a new `EditedText` with the given operations.
+    /// The operations must be in the order in which they are meant to be
+    /// applied. The operations must not overlap.
+    fn new(
+        text: &'a str,
+        operations: Vec<OrderedOperation<T>>,
+        mut cursors: Vec<CursorPosition>,
+    ) -> Self {
+        operations
+            .iter()
+            .zip(operations.iter().skip(1))
+            .for_each(|(previous, next)| {
+                debug_assert!(
+                    previous.operation.start_index() <= next.operation.start_index(),
+                    "{} must not come before {} yet it does",
+                    previous.operation,
+                    next.operation
+                );
+            });
+
+        cursors.sort_by_key(|cursor| cursor.char_index);
+
+        Self {
+            text,
+            operations,
+            cursors,
+        }
+    }
+
+    #[must_use]
+    pub fn merge(self, other: Self) -> Self {
+        debug_assert_eq!(
+            self.text, other.text,
+            "`EditedText`-s must be derived from the same text to be mergable"
+        );
+
+        let mut left_merge_context = MergeContext::default();
+        let mut right_merge_context = MergeContext::default();
+
+        let mut merged_cursors = Vec::with_capacity(self.cursors.len() + other.cursors.len());
+        let mut left_cursors = self.cursors.into_iter().peekable();
+        let mut right_cursors = other.cursors.into_iter().peekable();
+
+        let mut merged_operations: Vec<OrderedOperation<T>> =
+            Vec::with_capacity(self.operations.len() + other.operations.len());
+
+        let mut left_iter = self.operations.into_iter();
+        let mut right_iter = other.operations.into_iter();
+
+        let mut maybe_left_op = left_iter.next();
+        let mut maybe_right_op = right_iter.next();
+
+        loop {
+            let (side, OrderedOperation { operation, order }) =
+                match (maybe_left_op.clone(), maybe_right_op.clone()) {
+                    (Some(left_op), Some(right_op)) => {
+                        if left_op < right_op {
+                            (Side::Left, left_op)
+                        } else {
+                            (Side::Right, right_op)
+                        }
+                    }
+
+                    (Some(left_op), None) => (Side::Left, left_op),
+                    (None, Some(right_op)) => (Side::Right, right_op),
+                    (None, None) => break,
+                };
+
+            if side == Side::Left {
+                maybe_left_op = left_iter.next();
+            } else {
+                maybe_right_op = right_iter.next();
+            }
+
+            let original_start = operation.start_index() as i64;
+            let original_end = operation.end_index();
+            let original_length = operation.len() as i64;
+
+            let result = match side {
+                Side::Left => operation.merge_operations_with_context(
+                    &mut right_merge_context,
+                    &mut left_merge_context,
+                ),
+                Side::Right => operation.merge_operations_with_context(
+                    &mut left_merge_context,
+                    &mut right_merge_context,
+                ),
+            };
+
+            if let Some(ref op @ (Operation::Insert { .. } | Operation::Equal { .. })) = result {
+                let shift =
+                    op.start_index() as i64 - original_start + op.len() as i64 - original_length;
+                match side {
+                    Side::Left => {
+                        while let Some(cursor) =
+                            left_cursors.next_if(|cursor| cursor.char_index <= original_end + 1)
+                        {
+                            merged_cursors.push(cursor.with_index(
+                                (op.start_index() as i64).max(cursor.char_index as i64 + shift)
+                                    as usize,
+                            ));
+                        }
+                    }
+                    Side::Right => {
+                        while let Some(cursor) =
+                            right_cursors.next_if(|cursor| cursor.char_index <= original_end + 1)
+                        {
+                            merged_cursors.push(cursor.with_index(
+                                (op.start_index() as i64).max(cursor.char_index as i64 + shift)
+                                    as usize,
+                            ));
+                        }
+                    }
+                }
+            }
+
+            merged_operations.extend(result.into_iter().map(|op| OrderedOperation {
+                order,
+                operation: op,
+            }));
+        }
+
+        let last_index = merged_operations
+            .iter()
+            .filter(|operation| {
+                matches!(
+                    operation.operation,
+                    Operation::Insert { .. } | Operation::Equal { .. }
+                )
+            })
+            .next_back()
+            .map_or(0, |op| op.operation.end_index());
+
+        for cursor in left_cursors.chain(right_cursors) {
+            merged_cursors.push(cursor.with_index(last_index));
+        }
+
+        Self::new(self.text, merged_operations, merged_cursors)
+    }
+
+    /// Apply the operations to the text and return the resulting text.
+    #[must_use]
+    pub fn apply(&self) -> String {
+        let mut builder: StringBuilder<'_> = StringBuilder::new(self.text);
+
+        for OrderedOperation { operation, .. } in &self.operations {
+            builder = operation.apply(builder);
+        }
+
+        builder.build()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::env;
+
+    use insta::assert_debug_snapshot;
+    use pretty_assertions::assert_eq;
+
+    use super::*;
+
+    #[test]
+    fn test_calculate_operations() {
+        let left = "hello world! How are you?  Adam";
+        let right = "Hello, my friend! How are you doing? Albert";
+
+        let operations = EditedText::from_strings(left, right.into());
+
+        insta::assert_debug_snapshot!(operations);
+
+        let new_right = operations.apply();
+        assert_eq!(new_right.to_string(), right);
+    }
+
+    #[test]
+    fn test_calculate_operations_with_no_diff() {
+        let text = "hello world!";
+
+        let operations = EditedText::from_strings(text, text.into());
+
+        assert_debug_snapshot!(operations);
+
+        let new_right = operations.apply();
+        assert_eq!(new_right.to_string(), text);
+    }
+
+    #[test]
+    fn test_calculate_operations_with_insert() {
+        let original = "hello world! ...";
+        let left = "Hello world! I'm Andras.";
+        let right = "Hello world! How are you?";
+        let expected = "Hello world! How are you? I'm Andras.";
+
+        let operations_1 = EditedText::from_strings(original, left.into());
+        let operations_2 = EditedText::from_strings(original, right.into());
+
+        let operations = operations_1.merge(operations_2);
+        assert_eq!(operations.apply(), expected);
+    }
+}

src/operation_transformation/merge_context.rs

@@ -0,0 +1,73 @@
+use core::fmt::Debug;
+
+use crate::operation_transformation::Operation;
+
+#[derive(Clone, Debug)]
+pub struct MergeContext<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    last_operation: Option<Operation<T>>,
+    pub shift: i64,
+}
+
+impl<T> Default for MergeContext<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    fn default() -> Self {
+        MergeContext {
+            last_operation: None,
+            shift: 0,
+        }
+    }
+}
+
+impl<T> MergeContext<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    pub fn last_operation(&self) -> Option<&Operation<T>> { self.last_operation.as_ref() }
+
+    pub fn replace_last_operation(&mut self, operation: Option<Operation<T>>) {
+        self.last_operation = operation;
+    }
+
+    /// Replace the last delete operation (if there was one) with a new one
+    /// while applying it to the `shift` in case the last operation
+    /// was a delete.
+    pub fn consume_and_replace_last_operation(&mut self, operation: Option<Operation<T>>) {
+        if let Some(Operation::Delete {
+            deleted_character_count,
+            ..
+        }) = self.last_operation.take()
+        {
+            self.shift -= deleted_character_count as i64;
+        }
+
+        self.last_operation = operation;
+    }
+
+    /// Remove the last operation (if there was one) in case it is behind the
+    /// threshold operation. This updates the `shift` in case the last operation
+    /// was a delete.
+    pub fn consume_last_operation_if_it_is_too_behind(&mut self, threshold_index: i64) {
+        if let Some(last_operation) = self.last_operation.as_ref() {
+            if let Operation::Delete {
+                deleted_character_count,
+                ..
+            } = last_operation
+            {
+                if threshold_index + self.shift > last_operation.end_index() as i64 {
+                    self.shift -= *deleted_character_count as i64;
+                    self.last_operation = None;
+                }
+            } else if let Operation::Insert { .. } = last_operation
+                && threshold_index + self.shift - last_operation.len() as i64
+                    > last_operation.end_index() as i64
+            {
+                self.last_operation = None;
+            }
+        }
+    }
+}

src/operation_transformation/operation.rs

@@ -0,0 +1,513 @@
+use core::fmt::{Debug, Display};
+use std::ops::Range;
+
+#[cfg(feature = "serde")]
+use serde::{Deserialize, Serialize};
+
+use super::merge_context::MergeContext;
+use crate::{
+    Token,
+    utils::{
+        find_longest_prefix_contained_within::find_longest_prefix_contained_within,
+        string_builder::StringBuilder,
+    },
+};
+
+/// Represents a change that can be applied on a `StringBuilder`.
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[derive(Clone, PartialEq)]
+pub enum Operation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    Equal {
+        index: usize,
+        length: usize,
+
+        #[cfg(debug_assertions)]
+        text: Option<String>,
+    },
+
+    Insert {
+        index: usize,
+        text: Vec<Token<T>>,
+    },
+
+    Delete {
+        index: usize,
+        deleted_character_count: usize,
+
+        #[cfg(debug_assertions)]
+        deleted_text: Option<String>,
+    },
+}
+
+impl<T> Operation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    /// Creates an equal operation with the given index.
+    /// This operation is used to indicate that the text at the given index
+    /// is unchanged.
+    pub fn create_equal(index: usize, length: usize) -> Option<Self> {
+        if length == 0 {
+            return None;
+        }
+
+        Some(Operation::Equal {
+            index,
+            length,
+
+            #[cfg(debug_assertions)]
+            text: None,
+        })
+    }
+
+    pub fn create_equal_with_text(index: usize, text: String) -> Option<Self> {
+        if text.is_empty() {
+            return None;
+        }
+
+        Some(Operation::Equal {
+            index,
+            length: text.chars().count(),
+
+            #[cfg(debug_assertions)]
+            text: Some(text),
+        })
+    }
+
+    /// Creates an insert operation with the given index and text.
+    /// If the text is empty (meaning that the operation would be a no-op),
+    /// returns None.
+    pub fn create_insert(index: usize, text: Vec<Token<T>>) -> Option<Self> {
+        if text.is_empty() {
+            return None;
+        }
+
+        Some(Operation::Insert { index, text })
+    }
+
+    /// Creates a delete operation with the given index and number of
+    /// to-be-deleted characters. If the operation would delete 0 (meaning
+    /// that the operation would be a no-op), returns None.
+    pub fn create_delete(index: usize, deleted_character_count: usize) -> Option<Self> {
+        if deleted_character_count == 0 {
+            return None;
+        }
+
+        Some(Operation::Delete {
+            index,
+            deleted_character_count,
+
+            #[cfg(debug_assertions)]
+            deleted_text: None,
+        })
+    }
+
+    pub fn create_delete_with_text(index: usize, text: String) -> Option<Self> {
+        if text.is_empty() {
+            return None;
+        }
+
+        Some(Operation::Delete {
+            index,
+            deleted_character_count: text.chars().count(),
+
+            #[cfg(debug_assertions)]
+            deleted_text: Some(text),
+        })
+    }
+
+    /// Applies the operation to the given `StringBuilder`, returning the
+    /// modified `StringBuilder`.
+    ///
+    /// When compiled in debug mode, panics if a delete operation is attempted
+    /// on a range of text that does not match the text to be deleted.
+    pub fn apply<'a>(&self, mut builder: StringBuilder<'a>) -> StringBuilder<'a> {
+        match self {
+            Operation::Equal {
+                #[cfg(debug_assertions)]
+                text,
+                ..
+            } => {
+                #[cfg(debug_assertions)]
+                debug_assert!(
+                    text.as_ref()
+                        .is_none_or(|text| builder.get_slice(self.range()) == *text),
+                    "Text which is supposed to be equal does not match the text in the range"
+                );
+
+                return builder;
+            }
+            Operation::Insert { text, .. } => builder.insert(
+                self.start_index(),
+                &text.iter().map(Token::original).collect::<String>(),
+            ),
+            Operation::Delete {
+                #[cfg(debug_assertions)]
+                deleted_text,
+                ..
+            } => {
+                #[cfg(debug_assertions)]
+                debug_assert!(
+                    deleted_text
+                        .as_ref()
+                        .is_none_or(|text| builder.get_slice(self.range()) == *text),
+                    "Text to delete does not match the text in the range"
+                );
+
+                builder.delete(self.range());
+            }
+        }
+
+        builder
+    }
+
+    /// Returns the index of the first character that the operation affects.
+    pub fn start_index(&self) -> usize {
+        match self {
+            Operation::Equal { index, .. }
+            | Operation::Insert { index, .. }
+            | Operation::Delete { index, .. } => *index,
+        }
+    }
+
+    /// Returns the index of the last character that the operation affects.
+    pub fn end_index(&self) -> usize {
+        debug_assert!(
+            self.len() > 0,
+            " len() must be greater than 0 because operations must be non-empty"
+        );
+        self.start_index() + self.len() - 1
+    }
+
+    /// Returns the range of indices of characters that the operation affects.
+    #[allow(clippy::range_plus_one)]
+    pub fn range(&self) -> Range<usize> { self.start_index()..self.end_index() + 1 }
+
+    /// Returns the number of affected characters. It is always greater than 0
+    /// because empty operations cannot be created.
+    pub fn len(&self) -> usize {
+        match self {
+            Operation::Equal { length, .. } => *length,
+            Operation::Insert { text, .. } => text.iter().map(Token::get_original_length).sum(),
+            Operation::Delete {
+                deleted_character_count,
+                ..
+            } => *deleted_character_count,
+        }
+    }
+
+    /// Creates a new operation with the same type and text but with the given
+    /// index.
+    pub fn with_index(self, index: usize) -> Self {
+        match self {
+            Operation::Equal {
+                length,
+
+                #[cfg(debug_assertions)]
+                text,
+                ..
+            } => Operation::Equal {
+                index,
+                length,
+
+                #[cfg(debug_assertions)]
+                text,
+            },
+            Operation::Insert { text, .. } => Operation::Insert { index, text },
+            Operation::Delete {
+                deleted_character_count,
+
+                #[cfg(debug_assertions)]
+                deleted_text,
+                ..
+            } => Operation::Delete {
+                index,
+                deleted_character_count,
+
+                #[cfg(debug_assertions)]
+                deleted_text,
+            },
+        }
+    }
+
+    /// Creates a new operation with the same type and text but with the index
+    /// shifted by the given offset. The offset can be negative but the
+    /// resulting index must be non-negative.
+    ///
+    /// # Panics
+    ///
+    /// In debug mode, panics if the resulting index is negative.
+    pub fn with_shifted_index(self, offset: i64) -> Self {
+        let index = self.start_index() as i64 + offset;
+        debug_assert!(index >= 0, "Shifted index must be non-negative");
+
+        self.with_index(index as usize)
+    }
+
+    /// Merges the operation with the given context, producing a new operation
+    /// and updating the context. This implements a comples FSM that handles
+    /// the merging of operations in a way that is consistent with the text.
+    /// The contexts are updated in-place.
+    #[allow(clippy::too_many_lines)]
+    pub fn merge_operations_with_context(
+        self,
+        affecting_context: &mut MergeContext<T>,
+        produced_context: &mut MergeContext<T>,
+    ) -> Option<Operation<T>> {
+        affecting_context.consume_last_operation_if_it_is_too_behind(self.start_index() as i64);
+        let operation = self.with_shifted_index(affecting_context.shift);
+
+        match (operation, affecting_context.last_operation()) {
+            (operation @ Operation::Insert { .. }, None | Some(Operation::Equal { .. })) => {
+                produced_context.shift += operation.len() as i64;
+                produced_context.consume_and_replace_last_operation(Some(operation.clone()));
+                Some(operation)
+            }
+
+            (
+                Operation::Insert { text, index },
+                Some(Operation::Insert {
+                    text: previous_inserted_text,
+                    ..
+                }),
+            ) => {
+                // In case the current insert's prefix appears in the previously inserted text,
+                // we can trim the current insert to only include the non-overlapping part.
+                // This way, we don't end up duplicating text.
+                let offset_in_tokens =
+                    find_longest_prefix_contained_within(previous_inserted_text, &text);
+                let offset_in_length = text
+                    .iter()
+                    .take(offset_in_tokens)
+                    .map(Token::get_original_length)
+                    .sum::<usize>();
+                let trimmed_operation =
+                    Operation::create_insert(index, text[offset_in_tokens..].to_vec());
+
+                affecting_context.shift -= offset_in_length as i64;
+                produced_context.shift += trimmed_operation
+                    .as_ref()
+                    .map(Operation::len)
+                    .unwrap_or_default() as i64;
+                produced_context.consume_and_replace_last_operation(trimmed_operation.clone());
+
+                trimmed_operation
+            }
+
+            (
+                operation @ Operation::Delete { .. },
+                None | Some(Operation::Insert { .. } | Operation::Equal { .. }),
+            ) => {
+                produced_context.consume_and_replace_last_operation(Some(operation.clone()));
+                Some(operation)
+            }
+
+            (
+                operation @ Operation::Insert { .. },
+                Some(last_delete @ Operation::Delete { .. }),
+            ) => {
+                produced_context.shift += operation.len() as i64;
+
+                debug_assert!(
+                    last_delete.range().contains(&operation.start_index()),
+                    "There is a last delete ({last_delete}) but the operation ({operation}) is \
+                     not contained in it"
+                );
+
+                let difference = operation.start_index() as i64 - last_delete.start_index() as i64;
+
+                let moved_operation = operation.with_index(last_delete.start_index());
+
+                affecting_context.replace_last_operation(Operation::create_delete(
+                    moved_operation.end_index() + 1,
+                    (last_delete.len() as i64 - difference) as usize,
+                ));
+                affecting_context.shift -= difference;
+
+                produced_context.consume_and_replace_last_operation(Some(moved_operation.clone()));
+
+                Some(moved_operation)
+            }
+
+            (
+                operation @ Operation::Delete { .. },
+                Some(last_delete @ Operation::Delete { .. }),
+            ) => {
+                debug_assert!(
+                    last_delete.range().contains(&operation.start_index()),
+                    "There is a last delete ({last_delete}) but the operation ({operation}) is \
+                     not contained in it"
+                );
+
+                let difference = operation.start_index() as i64 - last_delete.start_index() as i64;
+
+                let updated_delete = Operation::create_delete(
+                    last_delete.start_index(),
+                    0.max(operation.end_index() as i64 - last_delete.end_index() as i64) as usize,
+                );
+
+                affecting_context.replace_last_operation(Operation::create_delete(
+                    last_delete.start_index(),
+                    0.max(last_delete.end_index() as i64 - operation.end_index() as i64) as usize,
+                ));
+                affecting_context.shift -= difference;
+
+                produced_context.consume_and_replace_last_operation(updated_delete.clone());
+
+                updated_delete
+            }
+            (
+                ref operation @ Operation::Equal {
+                    length,
+                    #[cfg(debug_assertions)]
+                    ref text,
+                    ..
+                },
+                Some(last_delete @ Operation::Delete { .. }),
+            ) => {
+                debug_assert!(
+                    last_delete.range().contains(&operation.start_index()),
+                    "There is a last delete ({last_delete}) but the operation ({operation}) is \
+                     not contained in it"
+                );
+
+                let overlap = (length as i64)
+                    .min(last_delete.end_index() as i64 - operation.start_index() as i64 + 1);
+
+                #[cfg(debug_assertions)]
+                let result = text.as_ref().map_or_else(
+                    || {
+                        Operation::create_equal(
+                            operation.end_index().min(last_delete.end_index()),
+                            (length as i64 - overlap) as usize,
+                        )
+                    },
+                    |text| {
+                        Operation::create_equal_with_text(
+                            operation.end_index().min(last_delete.end_index()),
+                            text.chars().skip(overlap as usize).collect::<String>(),
+                        )
+                    },
+                );
+
+                #[cfg(not(debug_assertions))]
+                let result = Operation::create_equal(
+                    operation.end_index().min(last_delete.end_index()),
+                    (length as i64 - overlap) as usize,
+                );
+
+                result
+            }
+            (operation @ Operation::Equal { .. }, _) => Some(operation),
+        }
+    }
+}
+
+impl<T> Display for Operation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
+        match self {
+            Operation::Equal {
+                index,
+                length,
+
+                #[cfg(debug_assertions)]
+                text,
+            } => {
+                #[cfg(debug_assertions)]
+                write!(
+                    f,
+                    "<equal {} from index {}>",
+                    text.as_ref()
+                        .map(|text| format!("'{text}'"))
+                        .unwrap_or(format!("{length} characters")),
+                    index
+                )?;
+
+                #[cfg(not(debug_assertions))]
+                write!(f, "<equal {length} from index {index}>")?;
+
+                Ok(())
+            }
+            Operation::Insert { index, text } => {
+                write!(
+                    f,
+                    "<insert '{}' from index {}>",
+                    text.iter().map(Token::original).collect::<String>(),
+                    index
+                )
+            }
+            Operation::Delete {
+                index,
+                deleted_character_count,
+
+                #[cfg(debug_assertions)]
+                deleted_text,
+            } => {
+                #[cfg(debug_assertions)]
+                write!(
+                    f,
+                    "<delete {} from index {}>",
+                    deleted_text
+                        .as_ref()
+                        .map(|text| format!("'{text}'"))
+                        .unwrap_or(format!("{deleted_character_count} characters")),
+                    index
+                )?;
+
+                #[cfg(not(debug_assertions))]
+                write!(
+                    f,
+                    "<delete {deleted_character_count} characters from index {index}>",
+                )?;
+
+                Ok(())
+            }
+        }
+    }
+}
+
+impl<T> Debug for Operation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result { write!(f, "{self}") }
+}
+
+#[cfg(test)]
+mod tests {
+    use pretty_assertions::assert_eq;
+
+    use super::*;
+
+    #[test]
+    #[should_panic(expected = "Shifted index must be non-negative")]
+    fn test_shifting_error() {
+        insta::assert_debug_snapshot!(
+            Operation::create_insert(1, vec!["hi".into()])
+                .unwrap()
+                .with_shifted_index(-2)
+        );
+    }
+
+    #[test]
+    fn test_apply_delete_with_create() {
+        let builder = StringBuilder::new("hello world");
+        let operation = Operation::<()>::create_delete_with_text(5, " world".to_owned()).unwrap();
+
+        assert_eq!(operation.apply(builder).build(), "hello");
+    }
+
+    #[test]
+    fn test_apply_insert() {
+        let builder = StringBuilder::new("hello");
+        let operation = Operation::create_insert(5, vec![" my friend".into()]).unwrap();
+
+        assert_eq!(operation.apply(builder).build(), "hello my friend");
+    }
+}

src/operation_transformation/ordered_operation.rs

@@ -0,0 +1,48 @@
+#[cfg(feature = "serde")]
+use serde::{Deserialize, Serialize};
+
+use crate::operation_transformation::Operation;
+
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[derive(Debug, Clone, PartialEq)]
+pub struct OrderedOperation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    pub order: usize,
+    pub operation: Operation<T>,
+}
+
+impl<T> OrderedOperation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    pub fn get_sort_key(&self) -> (usize, usize, String) {
+        (
+            self.order,
+            self.operation.start_index(),
+            // Make sure that the ordering is deterministic regardless of which text
+            // is left or right.
+            match &self.operation {
+                Operation::Equal { index, .. } => index.to_string(),
+                Operation::Insert { text, .. } => text
+                    .iter()
+                    .map(crate::tokenizer::token::Token::original)
+                    .collect::<String>(),
+                Operation::Delete {
+                    deleted_character_count,
+                    ..
+                } => deleted_character_count.to_string(),
+            },
+        )
+    }
+}
+
+impl<T> PartialOrd for OrderedOperation<T>
+where
+    T: PartialEq + Clone + std::fmt::Debug,
+{
+    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
+        self.get_sort_key().partial_cmp(&other.get_sort_key())
+    }
+}

src/operation_transformation/snapshots/reconcile__operation_transformation__edited_text__tests__calculate_operations.snap

@@ -0,0 +1,43 @@
+---
+source: reconcile/src/operation_transformation/edited_text.rs
+expression: operations
+snapshot_kind: text
+---
+EditedText {
+    text: "hello world! How are you?  Adam",
+    operations: [
+        OrderedOperation {
+            order: 0,
+            operation: <insert 'Hello, my friend!' from index 0>,
+        },
+        OrderedOperation {
+            order: 0,
+            operation: <delete 'hello world!' from index 17>,
+        },
+        OrderedOperation {
+            order: 12,
+            operation: <equal ' ' from index 17>,
+        },
+        OrderedOperation {
+            order: 13,
+            operation: <equal 'How' from index 18>,
+        },
+        OrderedOperation {
+            order: 16,
+            operation: <equal ' ' from index 21>,
+        },
+        OrderedOperation {
+            order: 17,
+            operation: <equal 'are' from index 22>,
+        },
+        OrderedOperation {
+            order: 20,
+            operation: <insert ' you doing? Albert' from index 25>,
+        },
+        OrderedOperation {
+            order: 20,
+            operation: <delete ' you?  Adam' from index 43>,
+        },
+    ],
+    cursors: [],
+}

src/operation_transformation/snapshots/reconcile__operation_transformation__edited_text__tests__calculate_operations_with_no_diff.snap

@@ -0,0 +1,23 @@
+---
+source: reconcile/src/operation_transformation/edited_text.rs
+expression: operations
+snapshot_kind: text
+---
+EditedText {
+    text: "hello world!",
+    operations: [
+        OrderedOperation {
+            order: 0,
+            operation: <equal 'hello' from index 0>,
+        },
+        OrderedOperation {
+            order: 5,
+            operation: <equal ' ' from index 5>,
+        },
+        OrderedOperation {
+            order: 6,
+            operation: <equal 'world!' from index 6>,
+        },
+    ],
+    cursors: [],
+}