hjkl_buffer/undo.rs
1//! Undo/redo entry type for per-buffer undo history.
2//!
3//! Lives in `hjkl-buffer` so that [`crate::Buffer`] can own the undo stack
4//! directly, keeping per-buffer state co-located with the rope.
5
6use std::collections::BTreeMap;
7use std::sync::Arc;
8use std::time::{Duration, SystemTime, UNIX_EPOCH};
9
10use serde::{Deserialize, Serialize};
11
12/// A single entry in the undo or redo stack.
13///
14/// The `timestamp` records the wall-clock time at which the snapshot was
15/// taken (i.e. when `push_undo` was called), enabling the `:earlier` /
16/// `:later` time-travel ex commands to walk the stack by duration rather
17/// than by step count.
18///
19/// Stored as a `ropey::Rope` (O(1) Arc-clone) rather than a `String` so
20/// snapshot cost is negligible even on multi-MB buffers.
21#[derive(Debug, Clone)]
22pub struct UndoEntry {
23 pub rope: ropey::Rope,
24 pub cursor: (usize, usize),
25 pub timestamp: SystemTime,
26 /// Local marks / jumplist / changelist / this-buffer's-global-marks
27 /// snapshot, so undo/redo restore mark-ish positions alongside the
28 /// text instead of leaving them shifted by the edit being undone
29 /// (audit-r2 fix 2). `Default::default()` (all empty) for callers
30 /// that don't populate it — restoring an all-empty snapshot is a
31 /// no-op against a freshly-constructed buffer's own empty state, so
32 /// existing fixtures that only care about text/cursor stay valid.
33 pub marks: MarkSnapshot,
34}
35
36/// Buffer-scoped "edit coherence" state snapshotted alongside a
37/// [`UndoEntry`]'s rope so undo/redo can restore marks, not just text.
38///
39/// Positions are plain `(row, col)` (or `(row, col)` values keyed by
40/// mark char) — no buffer-id tagging needed here even for
41/// `global_marks`, because a `MarkSnapshot` always belongs to exactly
42/// one buffer's undo stack; the engine is responsible for reattaching
43/// its own `buffer_id` when writing entries back into the session-global
44/// marks map (see `Editor::restore_marks`).
45#[derive(Debug, Clone, Default, PartialEq, Eq, Serialize, Deserialize)]
46pub struct MarkSnapshot {
47 /// `ma`-`mz` local marks (`View::marks_cloned`).
48 pub local_marks: BTreeMap<char, (usize, usize)>,
49 /// Back-jumplist (`Ctrl-o` stack), newest at the back.
50 pub jump_back: Vec<(usize, usize)>,
51 /// Forward-jumplist (`Ctrl-i` stack), newest at the back.
52 pub jump_fwd: Vec<(usize, usize)>,
53 /// `` `. `` / `'.` — position of the most recent change.
54 pub change_last_edit: Option<(usize, usize)>,
55 /// Changelist ring (`g;` / `g,`).
56 pub change_list: Vec<(usize, usize)>,
57 /// Walk cursor into `change_list`; `None` outside a walk.
58 pub change_cursor: Option<usize>,
59 /// `mA`-`mZ` global marks that belong to THIS buffer (bare
60 /// `(row, col)` — the buffer-id is implicit, this buffer).
61 pub global_marks: BTreeMap<char, (usize, usize)>,
62}
63
64// ─── Reversible edge delta (Phase 3a) ──────────────────────────────────────────
65//
66// Phase 2b stored a FULL rope snapshot on every node. Phase 3a stores only a
67// reversible **delta** on each parent→child edge (the root keeps a full base
68// rope) plus a materialization cache, so the in-RAM hot path stays snapshot-fast
69// while a future undofile shrinks from hundreds of MB to KB. This slice changes
70// ONLY internal storage — every public signature, and every observable
71// behaviour, is byte-identical to Phase 2b.
72
73/// A reversible edit between two adjacent buffer states, expressed as a single
74/// spanning replacement in **char-offset space** on the rope.
75///
76/// The index space is ropey `char` offsets throughout — never bytes — so
77/// multi-byte UTF-8 round-trips (a byte offset could split a codepoint). In the
78/// PARENT state `chars[start .. start + old.chars().count()] == old`; replacing
79/// that region with `new` yields the CHILD state, and swapping the two inverts
80/// it. A whole undo group collapses to the one region spanning its edits
81/// (common-prefix / common-suffix diff); a `Vec<Delta>` for disjoint regions is
82/// an acceptable future generalization, but one spanning region is all Phase 3a
83/// needs.
84#[derive(Debug, Clone, PartialEq, Eq, Default, Serialize, Deserialize)]
85pub struct Delta {
86 /// Char offset of the first differing char (the common-prefix length).
87 pub start: usize,
88 /// Chars present in the PARENT but not the CHILD (removed going forward).
89 pub old: String,
90 /// Chars present in the CHILD but not the PARENT (inserted going forward).
91 pub new: String,
92}
93
94/// Is `byte_idx` a char boundary of `r`? (`str::is_char_boundary` for ropes.)
95///
96/// Ropey always splits chunks ON char boundaries, so the question is answerable
97/// inside the one chunk containing the byte — O(log N), no materialization.
98fn is_char_boundary(r: &ropey::Rope, byte_idx: usize) -> bool {
99 if byte_idx == 0 || byte_idx == r.len_bytes() {
100 return true;
101 }
102 let (chunk, chunk_start, _, _) = r.chunk_at_byte(byte_idx);
103 chunk.is_char_boundary(byte_idx - chunk_start)
104}
105
106/// Length of the longest common byte PREFIX of `a` and `b`, capped at `max`.
107///
108/// Walks both ropes' chunks in lockstep — never materializes either rope. Two
109/// fast paths keep the common editor case near-free: identical chunk pointers
110/// (ropey leaves are `Arc`-shared, so a clone-then-edit child shares every leaf
111/// outside the edit) are accepted without reading bytes, and otherwise whole
112/// overlapping runs are compared with one slice `==` (memcmp).
113fn common_prefix_bytes(a: &ropey::Rope, b: &ropey::Rope, max: usize) -> usize {
114 let mut a_chunks = a.chunks();
115 let mut b_chunks = b.chunks();
116 let mut at: &[u8] = &[];
117 let mut bt: &[u8] = &[];
118 let mut n = 0;
119 while n < max {
120 if at.is_empty() {
121 match a_chunks.next() {
122 Some(c) => at = c.as_bytes(),
123 None => break,
124 }
125 continue;
126 }
127 if bt.is_empty() {
128 match b_chunks.next() {
129 Some(c) => bt = c.as_bytes(),
130 None => break,
131 }
132 continue;
133 }
134 // Same shared leaf, wholly within the cap: equal without looking.
135 if at.as_ptr() == bt.as_ptr() && at.len() == bt.len() && at.len() <= max - n {
136 n += at.len();
137 at = &[];
138 bt = &[];
139 continue;
140 }
141 let m = at.len().min(bt.len()).min(max - n);
142 if at[..m] == bt[..m] {
143 n += m;
144 at = &at[m..];
145 bt = &bt[m..];
146 } else {
147 let mut i = 0;
148 while at[i] == bt[i] {
149 i += 1;
150 }
151 n += i;
152 break;
153 }
154 }
155 n
156}
157
158/// Length of the longest common byte SUFFIX of `a` and `b`, capped at `max`.
159///
160/// The mirror of [`common_prefix_bytes`], walking both ropes' chunk cursors
161/// backwards from the end via [`ropey::iter::Chunks::prev`].
162fn common_suffix_bytes(a: &ropey::Rope, b: &ropey::Rope, max: usize) -> usize {
163 let mut a_chunks = a.chunks_at_byte(a.len_bytes()).0;
164 let mut b_chunks = b.chunks_at_byte(b.len_bytes()).0;
165 let mut at: &[u8] = &[];
166 let mut bt: &[u8] = &[];
167 let mut n = 0;
168 while n < max {
169 if at.is_empty() {
170 match a_chunks.prev() {
171 Some(c) => at = c.as_bytes(),
172 None => break,
173 }
174 continue;
175 }
176 if bt.is_empty() {
177 match b_chunks.prev() {
178 Some(c) => bt = c.as_bytes(),
179 None => break,
180 }
181 continue;
182 }
183 if at.as_ptr() == bt.as_ptr() && at.len() == bt.len() && at.len() <= max - n {
184 n += at.len();
185 at = &[];
186 bt = &[];
187 continue;
188 }
189 let m = at.len().min(bt.len()).min(max - n);
190 let (a_tail, b_tail) = (&at[at.len() - m..], &bt[bt.len() - m..]);
191 if a_tail == b_tail {
192 n += m;
193 at = &at[..at.len() - m];
194 bt = &bt[..bt.len() - m];
195 } else {
196 let mut i = 0;
197 while a_tail[m - 1 - i] == b_tail[m - 1 - i] {
198 i += 1;
199 }
200 n += i;
201 break;
202 }
203 }
204 n
205}
206
207/// Common-prefix / common-suffix diff of two ropes → the minimal single spanning
208/// [`Delta`]. Guarantees `apply_forward(a, diff(a, b)) == b` and
209/// `apply_inverse(b, diff(a, b)) == a` for ALL `a`, `b` (see the property
210/// tests). Boundaries are found on bytes (fast) then snapped to char boundaries
211/// so `old`/`new` are always valid UTF-8 and `start` is a true char offset.
212///
213/// The scans walk the ropes' chunks directly and only the differing MIDDLE is
214/// materialized — a full `to_string()` of both sides used to dominate every edit
215/// on a multi-MB buffer (measured 1.55 ms + ~6.4 MB of allocation per push at
216/// 3.2 MB). `diff_reference` in the tests below is the old materializing
217/// implementation, kept as the differential-test oracle.
218fn diff(parent: &ropey::Rope, child: &ropey::Rope) -> Delta {
219 let a_len = parent.len_bytes();
220 let b_len = child.len_bytes();
221
222 // Longest common byte prefix, snapped DOWN to a char boundary.
223 let max_pre = a_len.min(b_len);
224 let mut pre = common_prefix_bytes(parent, child, max_pre);
225 while pre > 0 && !is_char_boundary(parent, pre) {
226 pre -= 1;
227 }
228
229 // Longest common byte suffix not overlapping the prefix. The cut points
230 // `a_end`/`b_end` sit at identical trailing bytes, so snapping `a_end` UP to
231 // a char boundary snaps `b_end` by the same byte delta simultaneously.
232 let suf = common_suffix_bytes(parent, child, max_pre - pre);
233 let mut a_end = a_len - suf;
234 while a_end < a_len && !is_char_boundary(parent, a_end) {
235 a_end += 1;
236 }
237 let b_end = b_len - (a_len - a_end);
238
239 Delta {
240 start: parent.byte_to_char(pre),
241 old: parent.byte_slice(pre..a_end).to_string(),
242 new: child.byte_slice(pre..b_end).to_string(),
243 }
244}
245
246/// Apply a forward delta (PARENT → CHILD) to `parent`, returning the child rope.
247fn apply_forward(parent: &ropey::Rope, d: &Delta) -> ropey::Rope {
248 let mut r = parent.clone();
249 let old_chars = d.old.chars().count();
250 r.remove(d.start..d.start + old_chars);
251 r.insert(d.start, &d.new);
252 r
253}
254
255/// Apply an inverse delta (CHILD → PARENT) to `child`, returning the parent rope.
256fn apply_inverse(child: &ropey::Rope, d: &Delta) -> ropey::Rope {
257 let mut r = child.clone();
258 let new_chars = d.new.chars().count();
259 r.remove(d.start..d.start + new_chars);
260 r.insert(d.start, &d.old);
261 r
262}
263
264// ─── Undo arena tree (Phase 2b + Phase 3a delta storage) ──────────────────────
265//
266// The undo history is a real arena TREE of buffer states (Phase 2a introduced
267// the arena; Phase 2b makes it branch; Phase 3a stores edges as deltas). An edit
268// after an undo FORKS a new child instead of truncating the forward branch, so
269// old branches stay reachable — matching nvim's undo tree. `seq` is
270// load-bearing: `g-`/`g+` and the `:earlier`/`:later` count forms walk ALL
271// states by global `seq` (see `seq_earlier_step`/`seq_later_step`), while
272// `u`/`<C-r>` stay branch-local (parent / `last_child`).
273//
274// The linear-history subset is unchanged: with no forks the tree is a single
275// root→current→leaf path and every operation degrades to the old two-stack
276// behaviour.
277//
278// - `current` points at the node representing the LIVE buffer state.
279// - The ancestors of `current` (parent, … up to `root`) are the reachable undo
280// line; `current.parent` is the `u` target.
281// - `current.last_child` is the `<C-r>` target. Landing on any node (undo,
282// redo, or a `g-`/`g+` jump) rewrites `last_child` down the root→node path so
283// a later `<C-r>` retraces the branch just taken.
284//
285// Storage (Phase 3a): each non-root node holds the reversible `delta` on its
286// edge from `parent`; the root holds a full `base` rope. A node's content is
287// reconstructed on demand (`materialize`) from the nearest cached ancestor (or
288// the root base) by replaying forward deltas, or — for the `u`/`<C-r>` hot path
289// — from the adjacent warm node by one delta apply. Recently materialized ropes
290// are kept in a bounded LRU (`warm`); `current` is always kept warm. A node's
291// `delta`/content is FINALIZED lazily on the way past it (whenever the live rope
292// is written into it), never read as a restore target until then — so the fresh
293// leaf `current` holds a placeholder edge that is corrected before it matters.
294//
295// Keyframes (issue #302): the warm LRU alone bounds nothing — a `g-` onto a node
296// far outside it replayed the WHOLE chain from the root, so one jump was O(depth)
297// and `:earlier 9999` was O(depth²) (measured 212 ms for a 1024-deep history).
298// Every node at a depth that is a multiple of `KEYFRAME_INTERVAL` therefore PINS
299// its materialized rope, capping any single replay at `KEYFRAME_INTERVAL - 1`
300// applies; and `materialize` caches every intermediate it replays, so the
301// step-by-step walk pays that replay once per interval rather than once per step.
302// Keyframes are a pure in-memory cache: they are recomputable from the root base
303// plus the deltas, so they are NOT part of the `SerTree` on-disk projection, and
304// dropping every one of them changes only speed, never content.
305
306/// Keyframe spacing, in nodes of depth. Every node whose depth from the root is
307/// a multiple of this pins its materialized rope, so `materialize` never replays
308/// more than `KEYFRAME_INTERVAL - 1` deltas from the nearest anchor.
309///
310/// **Why 16.** The cost of a keyframe is *not* a document copy. `ropey::Rope` is
311/// a persistent tree with `Arc`-shared leaves, so a snapshot taken between small
312/// edits shares every chunk outside the edited path with its neighbours and only
313/// retains the O(log N) interior nodes the edit rewrote. Measured marginal RSS of
314/// retaining one such snapshot (1024 small edits, keeping every 16th):
315///
316/// | document | bytes retained per keyframe |
317/// | --- | --- |
318/// | 119 KiB | ~3.0 KiB |
319/// | 11.9 MiB | ~7.2 KiB |
320/// | 11.9 MiB, 4 KiB edits | ~11.5 KiB |
321///
322/// i.e. essentially independent of document size — a 200 MB buffer does not pay
323/// 200 MB per keyframe. At one keyframe per 16 nodes that is well under a KiB of
324/// amortized overhead per undo state, next to the `Delta` (two `String`s of the
325/// changed span) every node already stores unconditionally. 16 also sits under
326/// [`WARM_CAP`], which is what lets a full walk stay linear (see there).
327///
328/// The one shape that would break the "cheap" argument — an edit that rewrites
329/// the entire document, so consecutive states share nothing — already costs two
330/// full-document `String`s in that node's own `Delta`, so the keyframe adds at
331/// most another 1/16 of a cost the tree was paying anyway.
332const KEYFRAME_INTERVAL: usize = 16;
333
334/// Hard ceiling on how many keyframes are pinned at once; beyond it the
335/// least-recently-touched keyframe is unpinned (it becomes an ordinary cold
336/// node, replayable as before — correctness is unaffected).
337///
338/// Deliberately a COUNT, not a byte budget: by the measurement on
339/// [`KEYFRAME_INTERVAL`] a keyframe's real retention is roughly document-size
340/// *independent*, so a byte budget computed from `len_bytes()` would be a wild
341/// over-estimate and would switch keyframes off precisely on the large documents
342/// that need them most. 512 keyframes covers 8192 undo states — past any sane
343/// `undolevels` — for a measured ceiling of a few MiB.
344const KEYFRAME_CAP: usize = 512;
345
346/// Index into [`UndoTree::nodes`]. Slots are reused via a free list, so an id is
347/// only valid while the node it names is live — the tree never hands ids out.
348pub type NodeId = usize;
349
350/// How many recently-materialized ORDINARY node ropes to keep warm (besides the
351/// root base, `current`, and the pinned keyframes, which are always available).
352///
353/// Kept above [`KEYFRAME_INTERVAL`] on purpose: `materialize` caches every
354/// intermediate it replays, so one keyframe interval's worth of intermediates has
355/// to survive here for a step-by-step history walk (`:earlier 9999`) to cost one
356/// replay per INTERVAL rather than one per step — the difference between an O(N)
357/// and an O(N·K) walk.
358const WARM_CAP: usize = 32;
359
360/// One node of the undo arena tree: a buffer state the user could land on, plus
361/// its links and the reversible edge to its parent. A node with `> 1` child is a
362/// branch point (Phase 2b); `last_child` records which child `<C-r>` follows.
363#[derive(Debug, Clone)]
364pub struct UndoNode {
365 pub parent: Option<NodeId>,
366 pub children: Vec<NodeId>,
367 pub last_child: Option<NodeId>,
368 /// Reversible edit from the parent's content to this node's content. `None`
369 /// only for the root (and any node promoted to root by pruning), which holds
370 /// `base` instead.
371 pub delta: Option<Delta>,
372 /// Full base rope. `Some` ONLY for the root — the anchor the delta chain
373 /// replays from. Non-root nodes leave this `None` and carry a `delta`.
374 pub base: Option<ropey::Rope>,
375 /// Materialized content, LRU-managed. Warm for `current`, recently visited
376 /// nodes, and keyframe-depth nodes (which are pinned rather than aged out);
377 /// `None` (cold) otherwise, reconstructable from deltas.
378 pub rope_cache: Option<ropey::Rope>,
379 /// Distance from the root, root == 0. Assigned once at creation and never
380 /// renumbered — root-side pruning shifts the whole numbering down uniformly,
381 /// which leaves keyframes exactly [`KEYFRAME_INTERVAL`] apart either way.
382 /// Purely a cache-placement input: a wrong depth costs speed, never content.
383 pub depth: usize,
384 /// Post-state cursor for this node (restored alongside the text).
385 pub cursor: (usize, usize),
386 /// Wall-clock time this state was created — drives `:earlier`/`:later`.
387 pub timestamp: SystemTime,
388 /// Marks / jumplist / changelist snapshot restored with the text.
389 ///
390 /// Shared (`Arc`) rather than owned: a `push` writes the SAME snapshot
391 /// into the node being left and into the fresh child, and a
392 /// `MarkSnapshot` is up to five collections. Nodes never mutate it in
393 /// place — it is only ever replaced wholesale — so sharing is invisible.
394 pub marks: Arc<MarkSnapshot>,
395 /// Global monotonic order across the whole tree — the change number that
396 /// `g-`/`g+`, `:earlier`/`:later`, and `:undolist` traverse and display.
397 pub seq: u64,
398 /// Is this node on the root→`current` path (inclusive of both ends)?
399 ///
400 /// A maintained INVARIANT, not a hint, and the whole reason
401 /// [`UndoTree::retarget_current`] no longer walks the chain: the set of
402 /// flagged nodes is exactly the chain, so the first flagged node found
403 /// walking up from a landing target is the fork point, and every ancestor
404 /// above it already names its on-path child. Every site that moves
405 /// `current` — `push`, `undo_step`, `redo_step`, `retarget_current`,
406 /// `pop_committed` — maintains it, and the bulk paths (`new`,
407 /// `clear_all`, `from_serializable`) establish it outright.
408 /// `path_flags_track_the_root_to_current_walk` pins it against a
409 /// brute-force parent walk after each of those.
410 ///
411 /// Runtime-only and derivable, so — like `depth` and `rope_cache` — it is
412 /// not part of the [`SerTree`] projection.
413 pub on_path: bool,
414}
415
416/// Arena tree of [`UndoNode`]s. Replaces the old `undo_stack`/`redo_stack`
417/// `Vec<UndoEntry>` pair on [`crate::Buffer`]; see the module comment for how
418/// `u`/`<C-r>` (branch-local) and `g-`/`g+` (seq-ordered) map onto it, and how
419/// Phase 3a stores edges as deltas behind a materialization cache.
420#[derive(Debug)]
421pub struct UndoTree {
422 /// Slab; `None` slots are free and recorded in `free`.
423 nodes: Vec<Option<UndoNode>>,
424 /// Reusable slot indices (frees push here, allocs pop here first).
425 free: Vec<NodeId>,
426 /// LRU of ORDINARY node ids with a warm `rope_cache` (root and keyframe-depth
427 /// nodes excluded — they live in `base` / `keyframes`), most-recently-touched
428 /// last. Bounded by [`WARM_CAP`]; `current` is never evicted.
429 warm: Vec<NodeId>,
430 /// Node ids at a keyframe depth whose `rope_cache` is PINNED — the replay
431 /// anchors that bound `materialize` at [`KEYFRAME_INTERVAL`] applies.
432 /// Most-recently-touched last, bounded by [`KEYFRAME_CAP`].
433 keyframes: Vec<NodeId>,
434 /// `seq` → node id for every LIVE node, ordered.
435 ///
436 /// `g-` / `g+` / `:earlier` / `:later` need the neighbouring node in seq
437 /// order tree-wide, which was a full arena scan per step — an O(N) floor
438 /// under every history step, so holding `g-` over a deep history was
439 /// O(N * depth) in the lookup alone, dwarfing the bounded delta replay
440 /// keyframes had already achieved. A `BTreeMap` answers both directions in
441 /// O(log N) via `range`.
442 ///
443 /// Every live node appears exactly once. `seq` is assigned at creation and
444 /// never mutated, so the only maintenance points are `alloc` and `free`
445 /// (plus the two bulk paths, `clear_all` and `from_serializable`, which
446 /// rebuild it). `seq_index_matches_the_arena` pins that invariant against a
447 /// brute-force scan.
448 by_seq: std::collections::BTreeMap<u64, NodeId>,
449 root: NodeId,
450 current: NodeId,
451 next_seq: u64,
452}
453
454/// Trim `list` (an LRU, oldest first) down to `cap`, dropping the evicted
455/// nodes' materialized ropes. `current` is never evicted — the live state must
456/// stay available without a replay.
457///
458/// A free function over the pieces rather than a method so it can hold `&mut`
459/// on one arena field and one list at the same time.
460fn evict_to(list: &mut Vec<NodeId>, nodes: &mut [Option<UndoNode>], current: NodeId, cap: usize) {
461 while list.len() > cap {
462 let Some(pos) = list.iter().position(|&n| n != current) else {
463 break;
464 };
465 let victim = list.remove(pos);
466 if let Some(node) = nodes[victim].as_mut() {
467 node.rope_cache = None;
468 }
469 }
470}
471
472impl UndoTree {
473 /// New tree with a single root == current node holding `rope` as its base
474 /// state (the buffer as opened / last saved). The root is always
475 /// materializable from this base.
476 pub(crate) fn new(rope: ropey::Rope) -> Self {
477 let root = UndoNode {
478 parent: None,
479 children: Vec::new(),
480 last_child: None,
481 delta: None,
482 base: Some(rope),
483 rope_cache: None,
484 depth: 0,
485 cursor: (0, 0),
486 timestamp: SystemTime::now(),
487 marks: Arc::default(),
488 seq: 0,
489 // The root is `current`, so it is the whole path.
490 on_path: true,
491 };
492 Self {
493 nodes: vec![Some(root)],
494 free: Vec::new(),
495 warm: Vec::new(),
496 keyframes: Vec::new(),
497 by_seq: std::iter::once((0, 0)).collect(),
498 root: 0,
499 current: 0,
500 next_seq: 1,
501 }
502 }
503
504 // ── slab helpers ─────────────────────────────────────────────────────────
505
506 fn get(&self, id: NodeId) -> &UndoNode {
507 self.nodes[id].as_ref().expect("live NodeId")
508 }
509
510 fn get_mut(&mut self, id: NodeId) -> &mut UndoNode {
511 self.nodes[id].as_mut().expect("live NodeId")
512 }
513
514 fn alloc(&mut self, node: UndoNode) -> NodeId {
515 let seq = node.seq;
516 let id = if let Some(id) = self.free.pop() {
517 self.nodes[id] = Some(node);
518 id
519 } else {
520 self.nodes.push(Some(node));
521 self.nodes.len() - 1
522 };
523 self.by_seq.insert(seq, id);
524 id
525 }
526
527 /// Free a single slot (does NOT recurse into children — callers detach
528 /// links first). Drops the node's delta + materialized cache and purges it
529 /// from both cache LRUs.
530 fn free(&mut self, id: NodeId) {
531 if let Some(n) = self.nodes[id].as_ref() {
532 self.by_seq.remove(&n.seq);
533 }
534 self.nodes[id] = None;
535 self.free.push(id);
536 self.warm.retain(|&n| n != id);
537 self.keyframes.retain(|&n| n != id);
538 }
539
540 // ── materialization (Phase 3a + keyframes) ───────────────────────────────
541
542 /// Is `id` at a keyframe depth, i.e. should its materialized rope be PINNED
543 /// as a replay anchor rather than aged out of the ordinary warm LRU?
544 ///
545 /// The root qualifies arithmetically (depth 0) but is excluded: it carries a
546 /// full `base` and is already an anchor.
547 fn is_keyframe(&self, id: NodeId) -> bool {
548 id != self.root && self.get(id).depth.is_multiple_of(KEYFRAME_INTERVAL)
549 }
550
551 /// Record `id` as freshly materialized. Keyframe-depth nodes go into the
552 /// pinned `keyframes` LRU (bounded by [`KEYFRAME_CAP`]), everything else into
553 /// the ordinary `warm` LRU (bounded by [`WARM_CAP`]). Neither ever evicts
554 /// `current`; the root is skipped entirely (it has no cache, it has `base`).
555 fn touch_warm(&mut self, id: NodeId) {
556 if id == self.root {
557 return;
558 }
559 let (list, cap) = if self.is_keyframe(id) {
560 (&mut self.keyframes, KEYFRAME_CAP)
561 } else {
562 (&mut self.warm, WARM_CAP)
563 };
564 list.retain(|&n| n != id);
565 list.push(id);
566 evict_to(list, &mut self.nodes, self.current, cap);
567 }
568
569 /// Materialize node `id`'s content, warming its cache. Uses the node's own
570 /// cache if present, else the root `base`, else replays forward deltas from
571 /// the nearest materialized ancestor — a warm node, a pinned keyframe, or the
572 /// root. Always terminates: the root carries a base.
573 ///
574 /// Every intermediate along the replay is cached too, not just the target:
575 /// they were computed anyway and a `ropey::Rope` clone is an `Arc` bump, so
576 /// caching them is free — and it is what makes a step-by-step history walk
577 /// (`g-` held down, `:earlier 9999`) pay ONE replay per keyframe interval
578 /// instead of one per step.
579 fn materialize(&mut self, id: NodeId) -> ropey::Rope {
580 if let Some(r) = &self.get(id).rope_cache {
581 return r.clone();
582 }
583 if let Some(base) = &self.get(id).base {
584 return base.clone();
585 }
586 // Walk up to the nearest ancestor that holds content (warm cache, pinned
587 // keyframe, or the root base), recording the path to replay forward.
588 // Bounded by the keyframe spacing whenever the ancestor chain has been
589 // materialized before.
590 let mut path = Vec::new();
591 let base_rope;
592 let mut anchor = id;
593 loop {
594 path.push(anchor);
595 let par = self
596 .get(anchor)
597 .parent
598 .expect("a non-root, non-based node always has a parent");
599 if let Some(r) = &self.get(par).rope_cache {
600 base_rope = r.clone();
601 break;
602 }
603 if let Some(b) = &self.get(par).base {
604 base_rope = b.clone();
605 break;
606 }
607 anchor = par;
608 }
609 let mut rope = base_rope;
610 // `path` is target-first, so replaying in reverse ends on `id` — which
611 // therefore lands last in its LRU and cannot be the eviction picked by
612 // its own `touch_warm`.
613 for &node in path.iter().rev() {
614 let d = self
615 .get(node)
616 .delta
617 .as_ref()
618 .expect("a non-root node always carries its edge delta");
619 rope = apply_forward(&rope, d);
620 self.get_mut(node).rope_cache = Some(rope.clone());
621 self.touch_warm(node);
622 }
623 rope
624 }
625
626 /// Reconstruct node `id`'s restorable [`UndoEntry`] — the byte-for-byte
627 /// equivalent of Phase 2b's `node.snapshot.clone()`.
628 fn entry_of(&mut self, id: NodeId) -> UndoEntry {
629 let rope = self.materialize(id);
630 let n = self.get(id);
631 UndoEntry {
632 rope,
633 cursor: n.cursor,
634 timestamp: n.timestamp,
635 marks: (*n.marks).clone(),
636 }
637 }
638
639 /// Finalize node `id` to hold `rope` as its content, recomputing its edge
640 /// delta (or the root base) and updating cursor/timestamp/marks. A no-op
641 /// diff is skipped when the content is unchanged (the common case on a
642 /// history walk, where only the fields move) — which also avoids
643 /// materializing the parent, keeping the walk cheap.
644 fn set_node_state(
645 &mut self,
646 id: NodeId,
647 rope: ropey::Rope,
648 cursor: (usize, usize),
649 timestamp: SystemTime,
650 marks: Arc<MarkSnapshot>,
651 ) {
652 let is_root = self.get(id).parent.is_none();
653 // `Rope`'s `PartialEq` walks both ropes, so this comparison alone was
654 // O(document) on every history step. `is_instance` is an `Arc::ptr_eq`
655 // on the shared root and answers the case that actually occurs here —
656 // the caller stashes back the very rope it was handed by the previous
657 // step, still a clone of what the node cached. It is only a sufficient
658 // test for equality, never a necessary one, so an unrelated-but-equal
659 // rope still falls through to the full compare and is found unchanged.
660 let same = |a: Option<&ropey::Rope>| a.is_some_and(|a| a.is_instance(&rope) || *a == rope);
661 let unchanged =
662 same(self.get(id).rope_cache.as_ref()) || (is_root && same(self.get(id).base.as_ref()));
663 {
664 let node = self.get_mut(id);
665 node.cursor = cursor;
666 node.timestamp = timestamp;
667 node.marks = marks;
668 }
669 if unchanged {
670 return;
671 }
672 if is_root {
673 self.get_mut(id).base = Some(rope);
674 // The root is materialized from `base`; keep no stale cache.
675 self.get_mut(id).rope_cache = None;
676 self.warm.retain(|&n| n != id);
677 self.keyframes.retain(|&n| n != id);
678 } else {
679 let par = self.get(id).parent.expect("non-root has a parent");
680 let par_rope = self.materialize(par);
681 let d = diff(&par_rope, &rope);
682 let node = self.get_mut(id);
683 node.delta = Some(d);
684 node.rope_cache = Some(rope);
685 self.touch_warm(id);
686 }
687 }
688
689 /// Free `id` and its whole subtree (iteratively, so a long redo chain can't
690 /// overflow the stack).
691 fn free_subtree(&mut self, id: NodeId) {
692 let mut stack = vec![id];
693 while let Some(n) = stack.pop() {
694 let kids = std::mem::take(&mut self.get_mut(n).children);
695 stack.extend(kids);
696 self.free(n);
697 }
698 }
699
700 // ── read-only queries (mirror the old stack accessors) ───────────────────
701
702 /// `undo_stack.is_empty()` ⇔ `current` has no parent (is the root).
703 pub(crate) fn is_at_root(&self) -> bool {
704 self.get(self.current).parent.is_none()
705 }
706
707 /// `!redo_stack.is_empty()` ⇔ `current` has a forward child.
708 pub(crate) fn has_redo(&self) -> bool {
709 self.get(self.current).last_child.is_some()
710 }
711
712 /// Number of ancestors of `current`, i.e. its distance from the root.
713 ///
714 /// Walked rather than read off `Node::depth` so it stays right after a
715 /// prune or a load has renumbered the tree. Backs `View::undo_stack_len`,
716 /// which is the `undo_stack.len()` of the pre-tree implementation.
717 pub(crate) fn depth_from_root(&self) -> usize {
718 let mut d = 0;
719 let mut n = self.get(self.current).parent;
720 while let Some(p) = n {
721 d += 1;
722 n = self.get(p).parent;
723 }
724 d
725 }
726
727 /// `undo_stack.last().timestamp` == `current.parent`'s timestamp.
728 pub(crate) fn parent_timestamp(&self) -> Option<SystemTime> {
729 self.get(self.current).parent.map(|p| self.get(p).timestamp)
730 }
731
732 /// `redo_stack.last().timestamp` == `current.last_child`'s timestamp.
733 pub(crate) fn child_timestamp(&self) -> Option<SystemTime> {
734 self.get(self.current)
735 .last_child
736 .map(|c| self.get(c).timestamp)
737 }
738
739 // ── mutations ────────────────────────────────────────────────────────────
740
741 /// Commit a new boundary from `current`, growing the tree (Phase 2b).
742 ///
743 /// `entry` is the pre-edit LIVE state. It is written into `current`'s
744 /// snapshot (making `current` a real, restorable state), then a fresh child
745 /// is APPENDED and becomes the new `current` for the edit about to happen.
746 ///
747 /// Unlike Phase 2a this does NOT drop `current`'s existing children: an edit
748 /// after an undo now forks a new branch and the old forward branch(es) stay
749 /// reachable via `g-`/`g+` and `:undolist`, matching nvim's undo tree. The
750 /// new child is made `last_child` so a subsequent `<C-r>` follows the branch
751 /// just created.
752 pub(crate) fn push(&mut self, entry: UndoEntry) {
753 let cur = self.current;
754 // ONE snapshot serves both the node being left and the fresh child —
755 // this runs per edit, and a deep `MarkSnapshot` copy is up to five
756 // collection allocations.
757 let marks = Arc::new(entry.marks);
758 // Finalize the node being left with the pre-edit live state, recomputing
759 // its edge delta from its parent (or the root base).
760 self.set_node_state(
761 cur,
762 entry.rope.clone(),
763 entry.cursor,
764 entry.timestamp,
765 Arc::clone(&marks),
766 );
767 let seq = self.next_seq;
768 self.next_seq += 1;
769 // Fresh child: identical to `cur` for now (empty edge delta + warm cache
770 // holding the pre-edit rope). Its true post-edit content is finalized on
771 // the way past it (next move) or by the next `push`, at which point the
772 // edge delta is recomputed against `cur`.
773 let child_depth = self.get(cur).depth + 1;
774 let child = self.alloc(UndoNode {
775 parent: Some(cur),
776 children: Vec::new(),
777 last_child: None,
778 delta: Some(Delta::default()),
779 base: None,
780 rope_cache: Some(entry.rope),
781 depth: child_depth,
782 cursor: entry.cursor,
783 timestamp: entry.timestamp,
784 marks,
785 seq,
786 // The fresh child becomes `current`, extending the path by one.
787 on_path: true,
788 });
789 let cur_node = self.get_mut(cur);
790 // Append (retain old branches); the freshest child is the redo target.
791 cur_node.children.push(child);
792 cur_node.last_child = Some(child);
793 self.current = child;
794 self.touch_warm(child);
795 }
796
797 /// The cursor stored in the current node — for a node with a forward
798 /// branch, where the change that `<C-r>` would redo started (vim's
799 /// `uh_cursor`).
800 pub(crate) fn current_cursor(&self) -> (usize, usize) {
801 self.get(self.current).cursor
802 }
803
804 /// The cursor to store in `node` as `u` / `<C-r>` leaves it. A node with a
805 /// child keeps the cursor `push` saved there before that change — vim's
806 /// `uh_cursor`, fixed per change, which a later redo returns to and an undo
807 /// back into the node restores — rather than wherever the cursor wandered
808 /// since. A tip, which no change has left yet, takes the live cursor.
809 fn departed_cursor(&self, node: NodeId, live: (usize, usize)) -> (usize, usize) {
810 let n = self.get(node);
811 if n.children.is_empty() {
812 live
813 } else {
814 n.cursor
815 }
816 }
817
818 /// One undo step. `live` is the current buffer state (the node being left);
819 /// it is written into that node but INHERITS the destination (parent)
820 /// timestamp — byte-parity with the old dance, where the pushed redo entry
821 /// took the popped undo entry's timestamp. Returns the parent snapshot to
822 /// restore, or `None` at the root.
823 pub(crate) fn undo_step(
824 &mut self,
825 rope: ropey::Rope,
826 cursor: (usize, usize),
827 marks: MarkSnapshot,
828 ) -> Option<UndoEntry> {
829 let cur = self.current;
830 let par = self.get(cur).parent?;
831 let dest_ts = self.get(par).timestamp;
832 let cursor = self.departed_cursor(cur, cursor);
833 self.set_node_state(cur, rope, cursor, dest_ts, Arc::new(marks));
834 // Redo from the parent must return to the node we just left.
835 self.get_mut(par).last_child = Some(cur);
836 // The path shortens by one: `cur` drops off its tip, `par` is the new
837 // tip and was already on it.
838 self.get_mut(cur).on_path = false;
839 self.current = par;
840 // Hot-path materialization: derive the (possibly cold) parent from the
841 // just-finalized child by one inverse delta apply, so `u` never walks the
842 // ancestor chain even far outside the warm window.
843 if self.get(par).rope_cache.is_none() && self.get(par).base.is_none() {
844 let child_rope = self.get(cur).rope_cache.clone();
845 let child_delta = self.get(cur).delta.clone();
846 if let (Some(cr), Some(d)) = (child_rope, child_delta) {
847 let par_rope = apply_inverse(&cr, &d);
848 self.get_mut(par).rope_cache = Some(par_rope);
849 self.touch_warm(par);
850 }
851 }
852 Some(self.entry_of(par))
853 }
854
855 /// One redo step. Symmetric to [`Self::undo_step`]: `live` is written into
856 /// the node being left (which becomes an undo ancestor) with the
857 /// destination (child) timestamp. Returns the child snapshot to restore, or
858 /// `None` when there is no forward branch.
859 pub(crate) fn redo_step(
860 &mut self,
861 rope: ropey::Rope,
862 cursor: (usize, usize),
863 marks: MarkSnapshot,
864 ) -> Option<UndoEntry> {
865 let cur = self.current;
866 let child = self.get(cur).last_child?;
867 let dest_ts = self.get(child).timestamp;
868 let cursor = self.departed_cursor(cur, cursor);
869 self.set_node_state(cur, rope, cursor, dest_ts, Arc::new(marks));
870 // The path extends by one onto `child`; `cur.last_child` already names
871 // it (it is what we followed), so the ancestor chain stays correct.
872 self.get_mut(child).on_path = true;
873 self.current = child;
874 // `cur` is now warm, so materializing the child is one forward apply.
875 Some(self.entry_of(child))
876 }
877
878 // ── seq-ordered tree walk (`g-` / `g+`, `:earlier`/`:later` — Phase 2b) ───
879 //
880 // `u`/`<C-r>` are branch-local (parent / `last_child`); `g-`/`g+` traverse
881 // ALL states by global `seq`, crossing branch boundaries. `g-` restores the
882 // node with the greatest `seq` strictly below `current`'s; `g+` the least
883 // `seq` strictly above. Confirmed against nvim v0.12.4 (`iA<Esc>uiB<Esc>`
884 // then `g-`/`g-g-`/`g-g+` walks empty↔A↔B by change number).
885
886 /// `seq` of the node the buffer currently shows.
887 fn current_seq(&self) -> u64 {
888 self.get(self.current).seq
889 }
890
891 /// Live node with the greatest `seq` strictly below `s` (the `g-` target).
892 ///
893 /// O(log N) through [`Self::by_seq`]; this used to scan the whole arena on
894 /// every history step.
895 fn node_below(&self, s: u64) -> Option<NodeId> {
896 self.by_seq.range(..s).next_back().map(|(_, &id)| id)
897 }
898
899 /// Live node with the least `seq` strictly above `s` (the `g+` target).
900 ///
901 /// O(log N) through [`Self::by_seq`]; see [`Self::node_below`].
902 fn node_above(&self, s: u64) -> Option<NodeId> {
903 use std::ops::Bound;
904 self.by_seq
905 .range((Bound::Excluded(s), Bound::Unbounded))
906 .next()
907 .map(|(_, &id)| id)
908 }
909
910 /// Point `current` at `target`, re-establishing the invariant that every
911 /// ancestor of `current` names the child on the root→`current` path as its
912 /// `last_child`, so a later `<C-r>` retraces the branch just landed on
913 /// (nvim parity: landing on a node updates its ancestors' redo direction).
914 ///
915 /// Costs O(tree distance from the old `current` to `target`) — 1 for a step
916 /// along a branch — not O(depth). It used to rewrite the whole root→target
917 /// chain on every landing, which was the remaining per-step floor under
918 /// `g-`/`g+` once keyframes had bounded the materialization.
919 ///
920 /// What makes that sound is [`UndoNode::on_path`] being a maintained
921 /// invariant rather than a local test: the flagged nodes are EXACTLY the
922 /// root→`current` chain, and every flagged non-tip node already names its
923 /// on-path child. So the first flagged node found walking up from `target`
924 /// is the fork of the two paths, everything above it is already correct by
925 /// that invariant, and only the segment below it can be wrong.
926 ///
927 /// Do NOT replace this with an early exit out of the old full-chain walk at
928 /// the first ancestor that already names the right child. That is unsound
929 /// and was tried: `last_child` is also written by [`Self::push`] (immediate
930 /// parent only) and by leaf removal, so an ancestor can point the right way
931 /// while ITS ancestors still point down an abandoned branch; the early exit
932 /// leaves those stale and a later `<C-r>` walks into the wrong subtree —
933 /// caught by `tree_matches_full_snapshot_reference_over_random_ops` as a
934 /// redo returning another branch's text. The fix has to come from the path
935 /// being known, which is what `on_path` supplies.
936 fn retarget_current(&mut self, target: NodeId) {
937 // Walk up from `target`, linking each node as its parent's `last_child`,
938 // until a node already on the path: the fork. The root is always on the
939 // path, so this terminates there at the latest.
940 let mut node = target;
941 while !self.get(node).on_path {
942 let p = self
943 .get(node)
944 .parent
945 .expect("the root is always on the path, so the walk stops there");
946 self.get_mut(p).last_child = Some(node);
947 self.get_mut(node).on_path = true;
948 node = p;
949 }
950 let fork = node;
951 // Everything from the old `current` up to (not including) the fork
952 // leaves the path. Their stored `last_child` is deliberately left alone:
953 // an off-path node remembers the branch last taken through it, which is
954 // what lets `<C-r>` retrace that branch if the user lands back on it.
955 let mut leaving = self.current;
956 while leaving != fork {
957 self.get_mut(leaving).on_path = false;
958 leaving = self
959 .get(leaving)
960 .parent
961 .expect("the fork is on the path, hence an ancestor of `current`");
962 }
963 self.current = target;
964 }
965
966 /// Stash the live buffer state into the node being left (it may be a fresh,
967 /// still-stale leaf), preserving that node's own timestamp, then move.
968 fn stash_and_move(
969 &mut self,
970 target: NodeId,
971 rope: ropey::Rope,
972 cursor: (usize, usize),
973 marks: MarkSnapshot,
974 ) {
975 let cur = self.current;
976 let ts = self.get(cur).timestamp;
977 self.set_node_state(cur, rope, cursor, ts, Arc::new(marks));
978 self.retarget_current(target);
979 }
980
981 /// One `g-` / `:earlier` step: move to the next-lower-`seq` node tree-wide.
982 /// Returns its snapshot to restore, or `None` at the lowest state.
983 pub(crate) fn seq_earlier_step(
984 &mut self,
985 rope: ropey::Rope,
986 cursor: (usize, usize),
987 marks: MarkSnapshot,
988 ) -> Option<UndoEntry> {
989 let target = self.node_below(self.current_seq())?;
990 self.stash_and_move(target, rope, cursor, marks);
991 Some(self.entry_of(target))
992 }
993
994 /// One `g+` / `:later` step: move to the next-higher-`seq` node tree-wide.
995 /// Returns its snapshot to restore, or `None` at the highest state.
996 pub(crate) fn seq_later_step(
997 &mut self,
998 rope: ropey::Rope,
999 cursor: (usize, usize),
1000 marks: MarkSnapshot,
1001 ) -> Option<UndoEntry> {
1002 let target = self.node_above(self.current_seq())?;
1003 self.stash_and_move(target, rope, cursor, marks);
1004 Some(self.entry_of(target))
1005 }
1006
1007 /// Timestamp of the next-lower-`seq` node (the `:earlier Ns` predicate walks
1008 /// the seq order tree-wide, stopping once this dips to/below the cutoff).
1009 pub(crate) fn seq_earlier_timestamp(&self) -> Option<SystemTime> {
1010 self.node_below(self.current_seq())
1011 .map(|id| self.get(id).timestamp)
1012 }
1013
1014 /// Timestamp of the next-higher-`seq` node (the `:later Ns` predicate).
1015 pub(crate) fn seq_later_timestamp(&self) -> Option<SystemTime> {
1016 self.node_above(self.current_seq())
1017 .map(|id| self.get(id).timestamp)
1018 }
1019
1020 /// Leaves of the tree (nodes with no children), each as
1021 /// `(seq, depth-from-root, timestamp, is_current)`, sorted by `seq`.
1022 /// Drives `:undolist`, which — like nvim — lists only branch leaves.
1023 pub(crate) fn leaves(&self) -> Vec<(u64, usize, SystemTime, bool)> {
1024 let mut out: Vec<(u64, usize, SystemTime, bool)> = Vec::new();
1025 for (id, slot) in self.nodes.iter().enumerate() {
1026 let Some(n) = slot else { continue };
1027 // The root is the base state (change number 0), never a listed
1028 // "change" — like nvim, an untouched buffer lists nothing.
1029 if id == self.root || !n.children.is_empty() {
1030 continue;
1031 }
1032 // Depth = number of ancestors (root leaf ⇒ 0).
1033 let mut depth = 0;
1034 let mut p = n.parent;
1035 while let Some(pid) = p {
1036 depth += 1;
1037 p = self.get(pid).parent;
1038 }
1039 out.push((n.seq, depth, n.timestamp, id == self.current));
1040 }
1041 out.sort_by_key(|&(seq, ..)| seq);
1042 out
1043 }
1044
1045 /// Number of live nodes (used by [`Self::cap`] as the state budget).
1046 fn live_count(&self) -> usize {
1047 self.nodes.iter().filter(|n| n.is_some()).count()
1048 }
1049
1050 /// `undo_stack.pop()` — discard the most-recent boundary WITHOUT moving the
1051 /// live state. Used by `:s` with zero replacements and by a no-op undo
1052 /// group; in both, `current` is the childless leaf the last [`Self::push`]
1053 /// created, so reverse that push: drop the leaf, step `current` back to its
1054 /// parent (its snapshot equals the unchanged buffer), and restore the
1055 /// parent's `last_child`. Retains any sibling branches the push appended to.
1056 /// Returns `false` at the root, or if `current` is not a childless leaf
1057 /// (nothing safe to pop).
1058 pub(crate) fn pop_committed(&mut self) -> bool {
1059 let cur = self.current;
1060 if !self.get(cur).children.is_empty() {
1061 return false;
1062 }
1063 let Some(par) = self.get(cur).parent else {
1064 return false;
1065 };
1066 let par_node = self.get_mut(par);
1067 par_node.children.retain(|&c| c != cur);
1068 // The freshest surviving sibling (if any) becomes the redo target again.
1069 par_node.last_child = par_node.children.last().copied();
1070 // The path shortens onto `par` (already on it); `cur` is about to be
1071 // freed, but clear its flag so the invariant holds at every point.
1072 self.get_mut(cur).on_path = false;
1073 self.current = par;
1074 // The popped leaf always holds the highest seq (push assigns it last),
1075 // so reclaim the seq to keep numbering gapless.
1076 if self.get(cur).seq + 1 == self.next_seq {
1077 self.next_seq -= 1;
1078 }
1079 self.free(cur);
1080 true
1081 }
1082
1083 /// Node budget (`undolevels`). While the number of undo states (live nodes
1084 /// minus the root) exceeds `cap`, prune — branch-aware (Phase 2b):
1085 ///
1086 /// 1. First drop the lowest-`seq` LEAF that is NOT on the root→`current`
1087 /// path — an abandoned branch tip. This never touches `current` or its
1088 /// ancestors, so the state you're on and its full undo line survive.
1089 /// 2. When only the main line remains (no off-path leaves left), fall back
1090 /// to promoting the root's on-path child to root and dropping the old
1091 /// root — the Phase 2a root-side prune, which matches nvim's linear
1092 /// `undolevels` trimming (oldest states drop first).
1093 ///
1094 /// `cap == 0` means unlimited (matches the old guard).
1095 pub(crate) fn cap(&mut self, cap: usize) {
1096 if cap == 0 {
1097 return;
1098 }
1099 // Guard against a pathological loop: at most one prune per live node.
1100 let mut budget_iters = self.live_count() + 1;
1101 while self.live_count().saturating_sub(1) > cap && budget_iters > 0 {
1102 budget_iters -= 1;
1103 if let Some(leaf) = self.lowest_offpath_leaf() {
1104 self.detach_leaf(leaf);
1105 } else if !self.prune_root_side() {
1106 break;
1107 }
1108 }
1109 }
1110
1111 /// Lowest-`seq` leaf that is not on the root→`current` path, if any.
1112 fn lowest_offpath_leaf(&self) -> Option<NodeId> {
1113 let mut best: Option<(u64, NodeId)> = None;
1114 for (id, slot) in self.nodes.iter().enumerate() {
1115 if let Some(n) = slot
1116 && n.children.is_empty()
1117 && !n.on_path
1118 && best.is_none_or(|(bs, _)| n.seq < bs)
1119 {
1120 best = Some((n.seq, id));
1121 }
1122 }
1123 best.map(|(_, id)| id)
1124 }
1125
1126 /// Unlink `leaf` from its parent and free it (leaf ⇒ no subtree to recurse).
1127 fn detach_leaf(&mut self, leaf: NodeId) {
1128 if let Some(par) = self.get(leaf).parent {
1129 let par_node = self.get_mut(par);
1130 par_node.children.retain(|&c| c != leaf);
1131 if par_node.last_child == Some(leaf) {
1132 par_node.last_child = par_node.children.last().copied();
1133 }
1134 }
1135 self.free(leaf);
1136 }
1137
1138 /// Promote the root's on-path child to the new root and free the old root.
1139 /// Returns `false` when the root is `current` (nothing left to trim).
1140 fn prune_root_side(&mut self) -> bool {
1141 let root = self.root;
1142 if root == self.current {
1143 return false;
1144 }
1145 // The child on the path to `current` (the root always has one here).
1146 let Some(&child) = self
1147 .get(root)
1148 .children
1149 .iter()
1150 .find(|c| self.get(**c).on_path)
1151 else {
1152 return false;
1153 };
1154 // Any OTHER root children are off-path branches; drop them with the root.
1155 let others: Vec<NodeId> = self
1156 .get(root)
1157 .children
1158 .iter()
1159 .copied()
1160 .filter(|&c| c != child)
1161 .collect();
1162 for c in others {
1163 self.free_subtree(c);
1164 }
1165 // The promoted child becomes the new root: materialize it (while the old
1166 // root still anchors the chain) into a full base rope, then drop its
1167 // now-meaningless parent edge. This keeps every delta below it valid.
1168 let base = self.materialize(child);
1169 {
1170 let node = self.get_mut(child);
1171 node.parent = None;
1172 node.base = Some(base);
1173 node.delta = None;
1174 node.rope_cache = None;
1175 }
1176 self.warm.retain(|&n| n != child);
1177 self.keyframes.retain(|&n| n != child);
1178 // `child` was already on the path (it is the on-path child) and is now
1179 // its root end; the old root drops off it as it is freed.
1180 self.get_mut(root).on_path = false;
1181 self.root = child;
1182 self.free(root);
1183 true
1184 }
1185
1186 /// `redo_stack.clear()` — drop `current`'s forward branch.
1187 pub(crate) fn clear_redo(&mut self) {
1188 let cur = self.current;
1189 let kids = std::mem::take(&mut self.get_mut(cur).children);
1190 self.get_mut(cur).last_child = None;
1191 for c in kids {
1192 self.free_subtree(c);
1193 }
1194 }
1195
1196 /// `undo_stack.clear(); redo_stack.clear()` — collapse to a single root ==
1197 /// current node, preserving the live state. Frees every other node.
1198 pub(crate) fn clear_all(&mut self) {
1199 let cur = self.current;
1200 // The survivor becomes a self-contained root: give it a full base rope
1201 // (materialized while the chain is still intact) so it needs no parent.
1202 let base = self.materialize(cur);
1203 for id in 0..self.nodes.len() {
1204 if id != cur && self.nodes[id].is_some() {
1205 self.nodes[id] = None;
1206 self.free.push(id);
1207 }
1208 }
1209 self.warm.clear();
1210 self.keyframes.clear();
1211 // Bulk free bypassed `free`, so rebuild the index around the survivor.
1212 self.by_seq.clear();
1213 self.by_seq.insert(self.get(cur).seq, cur);
1214 let node = self.get_mut(cur);
1215 node.parent = None;
1216 node.children.clear();
1217 node.last_child = None;
1218 node.delta = None;
1219 node.base = Some(base);
1220 node.rope_cache = None;
1221 // The survivor is the new root: restart the depth numbering under it so
1222 // its descendants land on the keyframe ladder from 0 again.
1223 node.depth = 0;
1224 // Sole survivor ⇒ root == current ⇒ it is the whole path.
1225 node.on_path = true;
1226 self.root = cur;
1227 }
1228}
1229
1230// ─── Serializable projection (Phase 3b) ───────────────────────────────────────
1231//
1232// The undofile persists the tree as a compact, self-consistent projection: the
1233// root's full base text (String) plus, per node, its edge `delta` and links.
1234// `rope_cache`/`warm` are runtime-only and dropped — every node reconstructs
1235// from the root base + deltas, so the round-trip reproduces identical content
1236// at every node. NodeIds are DENSE in the projection (the live-slab holes are
1237// compacted away and links remapped), so `from_serializable` rebuilds a fresh
1238// arena 1:1 with no free list.
1239
1240/// One node of the serialized undo tree. Mirrors [`UndoNode`] minus the
1241/// runtime-only materialization cache; ids are dense indices into
1242/// [`SerTree::nodes`].
1243#[derive(Debug, Clone, Serialize, Deserialize)]
1244pub struct SerNode {
1245 /// Parent index, `None` only for the root.
1246 pub parent: Option<u32>,
1247 /// Child indices (order preserved; `> 1` ⇒ branch point).
1248 pub children: Vec<u32>,
1249 /// `<C-r>` target child index.
1250 pub last_child: Option<u32>,
1251 /// Reversible edge delta from the parent, `None` only for the root.
1252 pub delta: Option<Delta>,
1253 /// Post-state cursor `(row, col)`.
1254 pub cursor: (u32, u32),
1255 /// Wall-clock creation time, ms since the UNIX epoch.
1256 pub timestamp_unix_ms: u64,
1257 /// Marks / jumplist / changelist snapshot.
1258 pub marks: MarkSnapshot,
1259 /// Global monotonic change number.
1260 pub seq: u64,
1261}
1262
1263/// Serializable projection of an [`UndoTree`] for the undofile and the swap
1264/// undo section. Postcard-encoded (non-self-describing, so a schema/version
1265/// drift surfaces as a parse `Err` that the reader discards). See
1266/// [`UndoTree::to_serializable`] / [`UndoTree::from_serializable`].
1267#[derive(Debug, Clone, Serialize, Deserialize)]
1268pub struct SerTree {
1269 /// Root base text (the anchor the delta chain replays from). Always
1270 /// `Some` on write; `from_serializable` rejects `None` — a tree without an
1271 /// anchor is structurally invalid outside the swap context. The swap
1272 /// writer drops it for single-node trees (the body IS the base) and the
1273 /// swap reader re-substitutes the body text before the tree is rebuilt.
1274 pub base: Option<String>,
1275 /// Dense node arena (no holes).
1276 pub nodes: Vec<SerNode>,
1277 /// Root index into `nodes`.
1278 pub root: u32,
1279 /// Current (live) index into `nodes`.
1280 pub current: u32,
1281 /// Next `seq` to assign.
1282 pub next_seq: u64,
1283}
1284
1285/// [`SystemTime`] → ms since the UNIX epoch (saturating, pre-epoch ⇒ 0).
1286fn system_time_to_unix_ms(t: SystemTime) -> u64 {
1287 t.duration_since(UNIX_EPOCH)
1288 .map_or(0, |d| d.as_millis() as u64)
1289}
1290
1291/// ms since the UNIX epoch → [`SystemTime`].
1292fn unix_ms_to_system_time(ms: u64) -> SystemTime {
1293 UNIX_EPOCH + Duration::from_millis(ms)
1294}
1295
1296impl UndoTree {
1297 /// `seq` of the current (live) node — the header's `current_seq` for the
1298 /// undofile (the just-saved content per the §6 invariant).
1299 pub(crate) fn current_node_seq(&self) -> u64 {
1300 self.get(self.current).seq
1301 }
1302
1303 /// Materialize the current (live) node's content. Used by the swap
1304 /// recovery consistency guard (docs §6c) to check a deserialized tree
1305 /// agrees with the freshly-recovered buffer text before it's installed.
1306 pub(crate) fn current_content(&mut self) -> ropey::Rope {
1307 let cur = self.current;
1308 self.materialize(cur)
1309 }
1310
1311 /// Stash `rope` into the current node as the live buffer state, preserving
1312 /// that node's own cursor/timestamp/marks. Called just before serializing so
1313 /// the on-disk tree's `current` edge is exact even when `current` is a fresh
1314 /// (still-stale) leaf — the in-session self-heal (first undo/edit stashes
1315 /// live) applied eagerly at save time.
1316 pub(crate) fn sync_current(&mut self, rope: ropey::Rope) {
1317 let cur = self.current;
1318 let (cursor, ts, marks) = {
1319 let n = self.get(cur);
1320 (n.cursor, n.timestamp, n.marks.clone())
1321 };
1322 self.set_node_state(cur, rope, cursor, ts, marks);
1323 }
1324
1325 /// Project the live tree into a serializable, dense form (holes compacted,
1326 /// links remapped). `rope_cache`/`warm` are dropped; the root's `base`
1327 /// carries the anchor text and every non-root node its edge `delta`.
1328 pub(crate) fn to_serializable(&self) -> SerTree {
1329 // Dense remap: old NodeId → new index, in slab order.
1330 let mut map: Vec<Option<u32>> = vec![None; self.nodes.len()];
1331 let mut order: Vec<NodeId> = Vec::new();
1332 for (id, slot) in self.nodes.iter().enumerate() {
1333 if slot.is_some() {
1334 map[id] = Some(order.len() as u32);
1335 order.push(id);
1336 }
1337 }
1338 let remap = |id: NodeId| map[id].expect("live link points at a live node");
1339 let nodes = order
1340 .iter()
1341 .map(|&id| {
1342 let n = self.get(id);
1343 SerNode {
1344 parent: n.parent.map(remap),
1345 children: n.children.iter().map(|&c| remap(c)).collect(),
1346 last_child: n.last_child.map(remap),
1347 delta: n.delta.clone(),
1348 cursor: (n.cursor.0 as u32, n.cursor.1 as u32),
1349 timestamp_unix_ms: system_time_to_unix_ms(n.timestamp),
1350 marks: (*n.marks).clone(),
1351 seq: n.seq,
1352 }
1353 })
1354 .collect();
1355 // Always `Some`: the undofile and the swap's multi-node trees anchor on
1356 // the root text; only the swap writer may drop it (single-node trees,
1357 // where the body IS the base) and it re-substitutes on read.
1358 let base = Some(
1359 self.get(self.root)
1360 .base
1361 .as_ref()
1362 .map(|r| r.to_string())
1363 .unwrap_or_default(),
1364 );
1365 SerTree {
1366 base,
1367 nodes,
1368 root: remap(self.root),
1369 current: remap(self.current),
1370 next_seq: self.next_seq,
1371 }
1372 }
1373
1374 /// Rebuild an arena tree from a projection. Returns `None` on any structural
1375 /// inconsistency (out-of-range link, a non-root node missing its delta, a
1376 /// root carrying one, `children` lists that do not partition the non-root
1377 /// nodes, a child disagreeing with the node that listed it, a node
1378 /// unreachable from the root, a repeated `seq`) so a corrupt-but-parseable
1379 /// file degrades to a fresh tree rather than a broken one. The root's
1380 /// content comes from `base`; the current node's is materialized on demand
1381 /// from base + deltas.
1382 ///
1383 /// The partition + reachability pair is what makes the parent links a tree,
1384 /// and that is load-bearing rather than tidiness: [`Self::materialize`] and
1385 /// [`Self::retarget_current`] follow `parent` in unbounded loops, so a
1386 /// parent-link cycle is a hang (or an unbounded `path`) rather than a
1387 /// degraded tree. Rejecting it here is the only place that can see it.
1388 pub(crate) fn from_serializable(s: &SerTree) -> Option<Self> {
1389 let len = s.nodes.len();
1390 if len == 0 || s.root as usize >= len || s.current as usize >= len {
1391 return None;
1392 }
1393 // A tree without an anchor is structurally invalid outside the swap
1394 // context, where the body supplies the base (the swap reader fills it
1395 // before this is ever reached). Reject rather than guess.
1396 let base_str = s.base.as_deref()?;
1397 // Validate links and the root/non-root delta discipline up front.
1398 let mut seqs = std::collections::BTreeSet::new();
1399 for (i, n) in s.nodes.iter().enumerate() {
1400 let is_root = i as u32 == s.root;
1401 match (is_root, &n.delta, &n.parent) {
1402 (true, None, None) => {}
1403 (false, Some(_), Some(_)) => {}
1404 _ => return None,
1405 }
1406 if let Some(p) = n.parent
1407 && p as usize >= len
1408 {
1409 return None;
1410 }
1411 if n.children.iter().any(|&c| c as usize >= len) {
1412 return None;
1413 }
1414 if let Some(c) = n.last_child
1415 && c as usize >= len
1416 {
1417 return None;
1418 }
1419 // `by_seq` is keyed by `seq`, so a repeat would drop a node from the
1420 // `g-` / `g+` index without dropping it from the arena.
1421 if !seqs.insert(n.seq) {
1422 return None;
1423 }
1424 }
1425 // The `children` lists must PARTITION the non-root nodes — each listed
1426 // once, by nobody for the root — and every child's own `parent` must
1427 // name the node that listed it. Mutual agreement alone is not enough:
1428 // two nodes can name each other as parent AND as child while both stay
1429 // reachable from the root through the lists they were originally in.
1430 // Requiring a single lister is what turns the two directions into one
1431 // tree. Runs after the range loop so the indices below are in bounds.
1432 let mut listed_by: Vec<Option<u32>> = vec![None; len];
1433 for (i, n) in s.nodes.iter().enumerate() {
1434 for &c in &n.children {
1435 if c == s.root || listed_by[c as usize].is_some() {
1436 return None;
1437 }
1438 listed_by[c as usize] = Some(i as u32);
1439 }
1440 }
1441 if s.nodes
1442 .iter()
1443 .zip(listed_by.iter())
1444 .any(|(n, &lister)| n.parent != lister)
1445 {
1446 return None;
1447 }
1448 let base = ropey::Rope::from_str(base_str);
1449 let (depths, reachable) = depths_from_root(s);
1450 // With the lists partitioning the nodes, reachability from the root is
1451 // what rules out a cycle: a node inside one is reachable from no root.
1452 if reachable.iter().any(|&r| !r) {
1453 return None;
1454 }
1455 let mut nodes: Vec<Option<UndoNode>> = s
1456 .nodes
1457 .iter()
1458 .enumerate()
1459 .map(|(i, n)| {
1460 let is_root = i as u32 == s.root;
1461 Some(UndoNode {
1462 parent: n.parent.map(|p| p as NodeId),
1463 children: n.children.iter().map(|&c| c as NodeId).collect(),
1464 last_child: n.last_child.map(|c| c as NodeId),
1465 delta: n.delta.clone(),
1466 base: if is_root { Some(base.clone()) } else { None },
1467 rope_cache: None,
1468 depth: depths[i],
1469 cursor: (n.cursor.0 as usize, n.cursor.1 as usize),
1470 timestamp: unix_ms_to_system_time(n.timestamp_unix_ms),
1471 marks: Arc::new(n.marks.clone()),
1472 seq: n.seq,
1473 // Set below, once the whole arena exists to walk.
1474 on_path: false,
1475 })
1476 })
1477 .collect();
1478 // Establish the root→`current` path invariant on the loaded tree: flag
1479 // the chain, and make each ancestor name its on-path child. A projection
1480 // written by `to_serializable` already agrees (it came from a tree
1481 // holding the invariant), so this is a no-op on any file we produced —
1482 // doing it unconditionally is what stops a hand-edited or truncated one
1483 // from loading into a tree whose `<C-r>` direction contradicts its own
1484 // `current`, which `retarget_current` would no longer repair on the way
1485 // past. The validation above already rules out a parent-link cycle, so
1486 // the `len` bound is belt-and-braces on the one walk that runs before
1487 // any invariant of the rebuilt tree holds.
1488 let mut node = s.current as NodeId;
1489 for _ in 0..len {
1490 let n = nodes[node].as_mut().expect("the projection is dense");
1491 n.on_path = true;
1492 let Some(p) = n.parent else { break };
1493 nodes[p]
1494 .as_mut()
1495 .expect("the projection is dense")
1496 .last_child = Some(node);
1497 node = p;
1498 }
1499 // A node's `last_child` is the `<C-r>` direction and must name one of
1500 // its OWN children. Range validation above is all a hand-edited file
1501 // trips otherwise — a `last_child` pointing at any in-bounds node
1502 // loads, and the first `<C-r>` from that node walks into the wrong
1503 // subtree. Repair rather than reject (consistent with the path-chain
1504 // repair above): clear a `last_child` that names a non-child. The
1505 // path loop just ran only ever sets an ancestor's `last_child` to its
1506 // on-path child, which IS in its children list, so nothing it did is
1507 // undone here.
1508 for slot in nodes.iter_mut().flatten() {
1509 if let Some(c) = slot.last_child
1510 && !slot.children.contains(&c)
1511 {
1512 slot.last_child = None;
1513 }
1514 }
1515 let by_seq = nodes
1516 .iter()
1517 .enumerate()
1518 .filter_map(|(id, slot)| slot.as_ref().map(|n| (n.seq, id)))
1519 .collect();
1520 Some(Self {
1521 nodes,
1522 free: Vec::new(),
1523 warm: Vec::new(),
1524 keyframes: Vec::new(),
1525 by_seq,
1526 root: s.root as NodeId,
1527 current: s.current as NodeId,
1528 next_seq: s.next_seq,
1529 })
1530 }
1531}
1532
1533/// Depth-from-root of every node in a projection, by BFS over `children`, plus
1534/// the reachable set the walk visited.
1535///
1536/// Depth is NOT part of the on-disk format — it is derivable, and the undofile
1537/// deliberately stores only what is not (issue #302: keyframes are an in-memory
1538/// cache, so nothing about them enters `SerTree`). The `seen` guard makes this
1539/// terminate on a malformed file whose links form a cycle; anything unreachable
1540/// from the root keeps depth 0. `from_serializable` rejects a projection with
1541/// any unreachable node, so the returned depths are only ever used on a tree
1542/// where every one of them was computed by the walk.
1543fn depths_from_root(s: &SerTree) -> (Vec<usize>, Vec<bool>) {
1544 let mut depths = vec![0usize; s.nodes.len()];
1545 let mut seen = vec![false; s.nodes.len()];
1546 let mut queue = std::collections::VecDeque::new();
1547 seen[s.root as usize] = true;
1548 queue.push_back(s.root as usize);
1549 while let Some(i) = queue.pop_front() {
1550 for &c in &s.nodes[i].children {
1551 let c = c as usize;
1552 if !seen[c] {
1553 seen[c] = true;
1554 depths[c] = depths[i] + 1;
1555 queue.push_back(c);
1556 }
1557 }
1558 }
1559 (depths, seen)
1560}
1561
1562#[cfg(test)]
1563impl UndoTree {
1564 /// Ids of every live node, for warm-vs-cold materialization checks.
1565 fn live_ids(&self) -> Vec<NodeId> {
1566 (0..self.nodes.len())
1567 .filter(|&i| self.nodes[i].is_some())
1568 .collect()
1569 }
1570
1571 /// Materialize `id` for a test (public wrapper over the private method).
1572 fn materialize_for_test(&mut self, id: NodeId) -> ropey::Rope {
1573 self.materialize(id)
1574 }
1575
1576 /// Evict every cache INCLUDING the pinned keyframes (root keeps its `base`),
1577 /// forcing the next materialization of any node to reconstruct purely from
1578 /// deltas off the root — the strongest cold path there is.
1579 fn drop_all_caches(&mut self) {
1580 for n in self.nodes.iter_mut().flatten() {
1581 n.rope_cache = None;
1582 }
1583 self.warm.clear();
1584 self.keyframes.clear();
1585 }
1586
1587 /// Evict only the ordinary warm LRU, leaving the pinned keyframes — the
1588 /// steady state a deep history walk actually runs in.
1589 fn drop_warm_caches(&mut self) {
1590 for id in std::mem::take(&mut self.warm) {
1591 if let Some(n) = self.nodes[id].as_mut() {
1592 n.rope_cache = None;
1593 }
1594 }
1595 }
1596
1597 /// How many forward delta applies `materialize(id)` would perform right now
1598 /// (0 when `id` already holds content). This is the cost keyframes exist to
1599 /// bound, made assertable.
1600 fn replay_distance(&self, id: NodeId) -> usize {
1601 let mut n = 0;
1602 let mut cur = id;
1603 loop {
1604 let node = self.get(cur);
1605 if node.rope_cache.is_some() || node.base.is_some() {
1606 return n;
1607 }
1608 n += 1;
1609 match node.parent {
1610 Some(p) => cur = p,
1611 None => return n,
1612 }
1613 }
1614 }
1615
1616 /// The root→`current` path, by brute force: walk `parent` links up from
1617 /// `current` and reverse. Deliberately consults NEITHER `on_path` nor
1618 /// `last_child`, so it is an independent oracle for both.
1619 fn brute_force_path(&self) -> Vec<NodeId> {
1620 let mut walk = Vec::new();
1621 let mut n = Some(self.current);
1622 while let Some(id) = n {
1623 walk.push(id);
1624 n = self.get(id).parent;
1625 }
1626 walk.reverse();
1627 walk
1628 }
1629
1630 /// The two halves of the path invariant `retarget_current` now relies on,
1631 /// checked against [`Self::brute_force_path`]:
1632 ///
1633 /// 1. `on_path` is set on EXACTLY the nodes of the root→`current` walk.
1634 /// 2. every node on that walk except the tip names its successor as
1635 /// `last_child` — the observable property (`<C-r>` retraces the branch
1636 /// landed on) that the old full-chain rewrite established directly.
1637 ///
1638 /// (1) is the maintenance; (2) is what the maintenance buys. Checking only
1639 /// (2) would pass on a tree whose flags had drifted but whose links happened
1640 /// to be right; checking only (1) would pass on a tree that had stopped
1641 /// linking. Both must hold after every operation that moves `current`,
1642 /// allocates, or frees.
1643 #[track_caller]
1644 fn assert_path_invariant(&self, when: &str) {
1645 let walk = self.brute_force_path();
1646 assert_eq!(
1647 walk.first().copied(),
1648 Some(self.root),
1649 "the root→current walk does not start at the root after {when}"
1650 );
1651 let mut want = walk.clone();
1652 want.sort_unstable();
1653 let flagged: Vec<NodeId> = (0..self.nodes.len())
1654 .filter(|&i| self.nodes[i].as_ref().is_some_and(|n| n.on_path))
1655 .collect();
1656 assert_eq!(
1657 flagged, want,
1658 "on_path flags name {flagged:?}, the root→current walk is {want:?}, after {when}"
1659 );
1660 for w in walk.windows(2) {
1661 assert_eq!(
1662 self.get(w[0]).last_child,
1663 Some(w[1]),
1664 "node {} last_child is {:?}, not its on-path child {}, after {when}",
1665 w[0],
1666 self.get(w[0]).last_child,
1667 w[1]
1668 );
1669 }
1670 }
1671
1672 /// Reconstruct `id`'s content the naive way: walk to the root and replay
1673 /// every forward delta off the root `base`, consulting NO cache and NO
1674 /// keyframe. The differential oracle for keyframe-accelerated
1675 /// [`Self::materialize`] — the two must agree exactly, always.
1676 fn materialize_naive(&self, id: NodeId) -> ropey::Rope {
1677 let mut path = vec![id];
1678 let mut cur = id;
1679 while let Some(p) = self.get(cur).parent {
1680 path.push(p);
1681 cur = p;
1682 }
1683 let mut rope = self
1684 .get(cur)
1685 .base
1686 .clone()
1687 .expect("the root always carries a base");
1688 // Skip the root itself (it has no edge delta); replay root-ward → target.
1689 for &node in path.iter().rev().skip(1) {
1690 let d = self
1691 .get(node)
1692 .delta
1693 .as_ref()
1694 .expect("a non-root node always carries its edge delta");
1695 rope = apply_forward(&rope, d);
1696 }
1697 rope
1698 }
1699}
1700
1701#[cfg(test)]
1702mod tree_tests {
1703 use super::*;
1704
1705 fn entry(text: &str) -> UndoEntry {
1706 UndoEntry {
1707 rope: ropey::Rope::from_str(text),
1708 cursor: (0, 0),
1709 timestamp: SystemTime::now(),
1710 marks: MarkSnapshot::default(),
1711 }
1712 }
1713
1714 fn live(text: &str) -> (ropey::Rope, (usize, usize), MarkSnapshot) {
1715 (ropey::Rope::from_str(text), (0, 0), MarkSnapshot::default())
1716 }
1717
1718 #[test]
1719 fn fresh_tree_is_root_current_empty() {
1720 let t = UndoTree::new(ropey::Rope::from_str("hello"));
1721 assert!(t.is_at_root());
1722 assert!(!t.has_redo());
1723 assert_eq!(t.depth_from_root(), 0);
1724 assert_eq!(t.root, t.current);
1725 }
1726
1727 #[test]
1728 fn push_links_child_and_advances_current() {
1729 let mut t = UndoTree::new(ropey::Rope::from_str("hello"));
1730 let root = t.current;
1731 t.push(entry("hello"));
1732 // root now parents current; current is a fresh leaf.
1733 assert_eq!(t.get(t.current).parent, Some(root));
1734 assert_eq!(t.get(root).last_child, Some(t.current));
1735 assert_eq!(t.get(root).children, vec![t.current]);
1736 assert_eq!(t.depth_from_root(), 1);
1737 assert!(!t.has_redo());
1738 assert!(!t.is_at_root());
1739 }
1740
1741 #[test]
1742 fn undo_then_redo_round_trips_links() {
1743 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
1744 t.push(entry("s0")); // commit s0, current = n1 (live s1)
1745 let n0 = t.root;
1746 let n1 = t.current;
1747 // undo: current -> n0, restores s0.
1748 let (r, c, m) = live("s1");
1749 let restored = t.undo_step(r, c, m).unwrap();
1750 assert_eq!(restored.rope.to_string(), "s0");
1751 assert_eq!(t.current, n0);
1752 assert!(t.has_redo());
1753 assert_eq!(t.get(n0).last_child, Some(n1));
1754 // redo: current -> n1, restores what we left (s1).
1755 let (r, c, m) = live("s0");
1756 let restored = t.redo_step(r, c, m).unwrap();
1757 assert_eq!(restored.rope.to_string(), "s1");
1758 assert_eq!(t.current, n1);
1759 assert!(!t.has_redo());
1760 }
1761
1762 #[test]
1763 fn undo_at_root_and_redo_at_leaf_are_noops() {
1764 let mut t = UndoTree::new(ropey::Rope::from_str("x"));
1765 let (r, c, m) = live("x");
1766 assert!(t.undo_step(r, c, m).is_none());
1767 let (r, c, m) = live("x");
1768 assert!(t.redo_step(r, c, m).is_none());
1769 assert_eq!(t.depth_from_root(), 0);
1770 }
1771
1772 #[test]
1773 fn push_retains_forward_branch() {
1774 // Phase 2b: an edit after an undo forks a new branch; the old forward
1775 // branch is NOT dropped and remains reachable by seq.
1776 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
1777 t.push(entry("A")); // root -> nA (seq1, "A")
1778 let root = t.root;
1779 let na = t.current;
1780 let (r, c, m) = live("A");
1781 t.undo_step(r, c, m); // back to root, nA is the redo child
1782 assert!(t.has_redo());
1783 // A new edit from the root forks a SECOND child (nB, seq2).
1784 t.push(entry("B"));
1785 let nb = t.current;
1786 assert_ne!(nb, na);
1787 // Both branches live: root now has two children.
1788 assert_eq!(t.get(root).children.len(), 2);
1789 assert!(t.get(root).children.contains(&na));
1790 assert!(t.get(root).children.contains(&nb));
1791 // `<C-r>` follows the freshest branch (nB).
1792 assert_eq!(t.get(root).last_child, Some(nb));
1793 // Four live nodes: root + nA + nB + (nB is current/leaf). No leak of nA.
1794 let live = t.nodes.iter().filter(|n| n.is_some()).count();
1795 assert_eq!(live, 3);
1796 }
1797
1798 #[test]
1799 fn seq_walk_crosses_branches() {
1800 // Mirror nvim `iA<Esc>uiB<Esc>` then g-/g+ (buffer starts empty "").
1801 // `push(entry)` writes `entry` into the node being LEFT (its true
1802 // pre-edit content); the fresh leaf holds the live post-edit state only
1803 // once it is stashed on the way past — exactly the engine's discipline.
1804 let mut t = UndoTree::new(ropey::Rope::from_str(""));
1805 t.push(entry("")); // leave root("") -> nA(seq1), live "A"
1806 let (r, c, m) = live("A");
1807 t.undo_step(r, c, m); // stash "A" into nA, back to root("")
1808 t.push(entry("")); // leave root("") -> nB(seq2), branch, live "B"
1809 let nb = t.current;
1810 // At B (seq2). g- -> greatest seq below 2 = seq1 = "A".
1811 let (r, c, m) = live("B");
1812 let a = t.seq_earlier_step(r, c, m).unwrap();
1813 assert_eq!(a.rope.to_string(), "A");
1814 // g- again -> root "".
1815 let (r, c, m) = live("A");
1816 let root_snap = t.seq_earlier_step(r, c, m).unwrap();
1817 assert_eq!(root_snap.rope.to_string(), "");
1818 // g+ -> back up to seq1 "A".
1819 let (r, c, m) = live("");
1820 let a2 = t.seq_later_step(r, c, m).unwrap();
1821 assert_eq!(a2.rope.to_string(), "A");
1822 // g+ -> seq2 "B" (crosses to the other branch).
1823 let (r, c, m) = live("A");
1824 let b = t.seq_later_step(r, c, m).unwrap();
1825 assert_eq!(b.rope.to_string(), "B");
1826 assert_eq!(t.current, nb);
1827 // At the tip: no higher seq.
1828 let (r, c, m) = live("B");
1829 assert!(t.seq_later_step(r, c, m).is_none());
1830 }
1831
1832 #[test]
1833 fn seq_walk_updates_retrace_path() {
1834 // Land on a deep leaf via g-, then u/u and <C-r>/<C-r> must retrace it
1835 // (nvim `iX<Esc>iY<Esc>uiZ<Esc>g-uu<C-r><C-r>`). State labels: root "R".
1836 let mut t = UndoTree::new(ropey::Rope::from_str("R"));
1837 t.push(entry("R")); // leave root("R") -> nX(seq1), live "X"
1838 t.push(entry("X")); // leave nX("X") -> nY(seq2), live "Y"
1839 let (r, c, m) = live("Y");
1840 t.undo_step(r, c, m); // stash "Y" into nY, back to nX("X")
1841 t.push(entry("X")); // leave nX("X") -> nZ(seq3), branch, live "Z"
1842 // g- from Z(seq3) -> nY(seq2) "Y".
1843 let (r, c, m) = live("Z");
1844 let y = t.seq_earlier_step(r, c, m).unwrap();
1845 assert_eq!(y.rope.to_string(), "Y");
1846 // u,u back to root.
1847 let (r, c, m) = live("Y");
1848 t.undo_step(r, c, m);
1849 let (r, c, m) = live("X");
1850 t.undo_step(r, c, m);
1851 assert!(t.is_at_root());
1852 // <C-r>,<C-r> retraces the branch we landed on: root->X->Y.
1853 let (r, c, m) = live("R");
1854 let x = t.redo_step(r, c, m).unwrap();
1855 assert_eq!(x.rope.to_string(), "X");
1856 let (r, c, m) = live("X");
1857 let y2 = t.redo_step(r, c, m).unwrap();
1858 assert_eq!(y2.rope.to_string(), "Y");
1859 }
1860
1861 #[test]
1862 fn leaves_lists_branch_tips_by_seq() {
1863 // root -> nX -> nY -> nW (leaf, seq3, depth3) and nX -> nZ (leaf, seq4,
1864 // depth2). Mirrors nvim `iX iY iW uu iZ`.
1865 let mut t = UndoTree::new(ropey::Rope::from_str(""));
1866 t.push(entry("X"));
1867 t.push(entry("Y"));
1868 t.push(entry("W"));
1869 let (r, c, m) = live("W");
1870 t.undo_step(r, c, m);
1871 let (r, c, m) = live("Y");
1872 t.undo_step(r, c, m); // back to nX
1873 t.push(entry("Z")); // nX -> nZ(seq4)
1874 let leaves = t.leaves();
1875 // Two leaves: W(seq3, depth3) and Z(seq4, depth2). Z is current.
1876 let dims: Vec<(u64, usize, bool)> =
1877 leaves.iter().map(|&(s, d, _, cur)| (s, d, cur)).collect();
1878 assert_eq!(dims, vec![(3, 3, false), (4, 2, true)]);
1879 }
1880
1881 #[test]
1882 fn cap_prunes_oldest_from_root_side() {
1883 let mut t = UndoTree::new(ropey::Rope::from_str("s"));
1884 for _ in 0..5 {
1885 t.push(entry("s"));
1886 }
1887 assert_eq!(t.depth_from_root(), 5);
1888 t.cap(3);
1889 assert_eq!(t.depth_from_root(), 3);
1890 // Redo side untouched (there is none), current unchanged.
1891 assert!(!t.has_redo());
1892 // Two oldest slots were reclaimed.
1893 assert_eq!(t.free.len(), 2);
1894 }
1895
1896 #[test]
1897 fn cap_drops_offpath_leaf_before_main_line() {
1898 // Fork two abandoned branches off the root, then extend the main line,
1899 // and cap: the lowest-seq OFF-PATH leaf must go first, and `current`
1900 // plus its ancestors must survive.
1901 let mut t = UndoTree::new(ropey::Rope::from_str(""));
1902 t.push(entry("A")); // root -> nA(seq1) [abandoned branch tip]
1903 let na = t.current;
1904 let (r, c, m) = live("A");
1905 t.undo_step(r, c, m);
1906 t.push(entry("B")); // root -> nB(seq2) [abandoned branch tip]
1907 let nb = t.current;
1908 let (r, c, m) = live("B");
1909 t.undo_step(r, c, m);
1910 t.push(entry("C")); // root -> nC(seq3), the live main line
1911 let nc = t.current;
1912 // 4 live nodes (root, nA, nB, nC) => 3 states. Cap to 2.
1913 assert_eq!(t.leaves().len(), 3);
1914 t.cap(2);
1915 // The lowest-seq off-path leaf (nA, seq1) was dropped; current (nC) and
1916 // its ancestor (root) survive, and the newer off-path leaf nB survives.
1917 assert!(t.nodes[na].is_none());
1918 assert!(t.nodes[nb].is_some());
1919 assert_eq!(t.current, nc);
1920 assert!(!t.is_at_root());
1921 assert!(t.get(t.root).children.contains(&nb));
1922 assert!(t.get(t.root).children.contains(&nc));
1923 }
1924
1925 #[test]
1926 fn pop_committed_reverses_last_push() {
1927 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
1928 t.push(entry("s0")); // depth 1, current = fresh leaf
1929 assert_eq!(t.depth_from_root(), 1);
1930 assert!(t.pop_committed());
1931 // The just-pushed leaf is gone; current stepped back to the root.
1932 assert_eq!(t.depth_from_root(), 0);
1933 assert!(t.is_at_root());
1934 assert_eq!(t.free.len(), 1);
1935 // Seq reclaimed so the next push is gapless.
1936 assert_eq!(t.next_seq, 1);
1937 }
1938
1939 #[test]
1940 fn pop_committed_retains_sibling_branches() {
1941 // Fork a branch, then a no-op push at the fork must pop cleanly without
1942 // orphaning the sibling branch.
1943 let mut t = UndoTree::new(ropey::Rope::from_str(""));
1944 t.push(entry("A")); // root -> nA(seq1)
1945 let na = t.current;
1946 let (r, c, m) = live("A");
1947 t.undo_step(r, c, m); // back to root
1948 t.push(entry("B")); // root -> nB(seq2); root children [nA, nB]
1949 let root = t.root;
1950 // A spurious no-op push at nB, then pop it.
1951 assert!(t.pop_committed());
1952 // nB is gone, current back at root; nA branch still intact & reachable.
1953 assert!(t.get(root).children.contains(&na));
1954 assert_eq!(t.get(root).children.len(), 1);
1955 assert_eq!(t.current, root);
1956 let live = t.nodes.iter().filter(|n| n.is_some()).count();
1957 assert_eq!(live, 2); // root + nA
1958 }
1959
1960 #[test]
1961 fn pop_committed_at_root_is_false() {
1962 let mut t = UndoTree::new(ropey::Rope::from_str("s"));
1963 assert!(!t.pop_committed());
1964 }
1965
1966 #[test]
1967 fn clear_redo_drops_forward_only() {
1968 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
1969 t.push(entry("s0"));
1970 let (r, c, m) = live("s1");
1971 t.undo_step(r, c, m);
1972 assert!(t.has_redo());
1973 assert_eq!(t.depth_from_root(), 0);
1974 t.clear_redo();
1975 assert!(!t.has_redo());
1976 assert_eq!(t.depth_from_root(), 0);
1977 }
1978
1979 #[test]
1980 fn clear_all_collapses_to_single_node() {
1981 let mut t = UndoTree::new(ropey::Rope::from_str("s"));
1982 for _ in 0..3 {
1983 t.push(entry("s"));
1984 }
1985 t.clear_all();
1986 assert!(t.is_at_root());
1987 assert!(!t.has_redo());
1988 assert_eq!(t.depth_from_root(), 0);
1989 assert_eq!(t.root, t.current);
1990 }
1991
1992 /// The depth measures where `current` sits, not how many nodes exist: an
1993 /// undo walks it back down while the branch it came from stays live, and a
1994 /// redo climbs the same steps again.
1995 #[test]
1996 fn depth_from_root_follows_current_not_tree_size() {
1997 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
1998 t.push(entry("s0"));
1999 t.push(entry("s1"));
2000 t.push(entry("s2"));
2001 assert_eq!(t.depth_from_root(), 3);
2002
2003 let (r, c, m) = live("s3");
2004 assert!(t.undo_step(r, c, m).is_some());
2005 let (r, c, m) = live("s2");
2006 assert!(t.undo_step(r, c, m).is_some());
2007 assert_eq!(t.depth_from_root(), 1);
2008 // Nothing was pruned — the three pushed nodes are still reachable.
2009 assert!(t.has_redo());
2010 assert_eq!(t.live_count(), 4);
2011
2012 let (r, c, m) = live("s1");
2013 assert!(t.redo_step(r, c, m).is_some());
2014 assert_eq!(t.depth_from_root(), 2);
2015 }
2016
2017 /// `on_path` and the ancestors' `last_child` chain must agree with a
2018 /// brute-force root→`current` walk after EVERY operation that moves
2019 /// `current`, allocates, or frees — that invariant is the only thing making
2020 /// `retarget_current`'s fork-point shortcut sound, and a drift in it is
2021 /// silent (a later `<C-r>` restores another branch's text, no error).
2022 ///
2023 /// One tree walked through every such site in turn: `new`, `push`,
2024 /// `undo_step`, `redo_step`, `seq_earlier_step`/`seq_later_step` (i.e.
2025 /// `retarget_current`, including a landing that crosses to a sibling
2026 /// branch), `clear_redo`, `cap` (both the off-path-leaf and the
2027 /// root-promotion prune), `pop_committed`, `from_serializable` and
2028 /// `clear_all`.
2029 #[test]
2030 fn path_flags_track_the_root_to_current_walk() {
2031 let mut t = UndoTree::new(ropey::Rope::from_str("R"));
2032 t.assert_path_invariant("new");
2033
2034 // Main line: R -> A -> B -> C.
2035 for s in ["R", "A", "B"] {
2036 t.push(entry(s));
2037 t.assert_path_invariant("push");
2038 }
2039 // Back down two, forking a sibling branch off the middle.
2040 let (r, c, m) = live("C");
2041 assert!(t.undo_step(r, c, m).is_some());
2042 t.assert_path_invariant("undo_step");
2043 let (r, c, m) = live("B");
2044 assert!(t.undo_step(r, c, m).is_some());
2045 t.assert_path_invariant("undo_step");
2046 let (r, c, m) = live("A");
2047 assert!(t.redo_step(r, c, m).is_some());
2048 t.assert_path_invariant("redo_step");
2049 let (r, c, m) = live("B");
2050 assert!(t.undo_step(r, c, m).is_some());
2051 t.push(entry("A")); // A -> X, a second branch under A
2052 t.assert_path_invariant("push forking a branch");
2053
2054 // `g-` / `g+` walk the whole tree by seq, so they land on nodes in the
2055 // OTHER branch — the case that actually exercises the fork point rather
2056 // than a parent/child step.
2057 let mut cur = String::from("X");
2058 for _ in 0..6 {
2059 if let Some(e) =
2060 t.seq_earlier_step(ropey::Rope::from_str(&cur), (0, 0), MarkSnapshot::default())
2061 {
2062 cur = e.rope.to_string();
2063 }
2064 t.assert_path_invariant("seq_earlier_step");
2065 }
2066 for _ in 0..6 {
2067 if let Some(e) =
2068 t.seq_later_step(ropey::Rope::from_str(&cur), (0, 0), MarkSnapshot::default())
2069 {
2070 cur = e.rope.to_string();
2071 }
2072 t.assert_path_invariant("seq_later_step");
2073 }
2074
2075 // A serialize round trip has to rebuild the flags from scratch.
2076 let round = UndoTree::from_serializable(&t.to_serializable()).expect("round trip");
2077 round.assert_path_invariant("from_serializable");
2078
2079 // Deepen, then prune: `cap` drops off-path leaves first and only then
2080 // promotes the root's on-path child, so both prune shapes run.
2081 for s in ["P", "Q", "S", "T"] {
2082 t.push(entry(s));
2083 }
2084 t.assert_path_invariant("pushes before cap");
2085 assert!(t.live_count() > 3, "cap has something to prune");
2086 t.cap(2);
2087 t.assert_path_invariant("cap");
2088
2089 assert!(t.pop_committed());
2090 t.assert_path_invariant("pop_committed");
2091
2092 t.push(entry("Z"));
2093 t.clear_redo();
2094 t.assert_path_invariant("clear_redo");
2095
2096 t.clear_all();
2097 t.assert_path_invariant("clear_all");
2098 }
2099}
2100
2101// ─── Phase 3a delta-storage tests ─────────────────────────────────────────────
2102//
2103// Correctness of the reversible delta and the warm/cold materialization is
2104// where text gets silently corrupted, so these lean hard on it: exact diff
2105// round-trips over random (incl. multi-byte) content, every node reconstructing
2106// identically warm and cold, and a random op stream cross-checked against a
2107// full-snapshot reference model kept alongside. All randomness is a deterministic
2108// xorshift seeded from a fixed constant — never `SystemTime`/entropy — so a
2109// failure reproduces exactly.
2110#[cfg(test)]
2111mod delta_tests {
2112 use super::*;
2113
2114 /// Iteration count for the randomized differential loops below, capped
2115 /// hard under miri.
2116 ///
2117 /// These loops are worth thousands of steps on a normal run: their value
2118 /// is statistical, shaking out logic bugs in the diff / keyframe code from
2119 /// a random op mix. That is a property of *executing* them, and miri
2120 /// interprets instead — six loops totalling ~22 300 iterations is the bulk
2121 /// of the weekly miri job's runtime, enough to push it past an hour.
2122 ///
2123 /// miri is there to catch UB, and UB shows up on the code *paths*, not on
2124 /// the thousandth repetition of one — so a short pass covers what miri can
2125 /// actually detect. Normal runs are untouched and keep the full count.
2126 fn stress_iters(n: usize) -> usize {
2127 if cfg!(miri) { n.min(50) } else { n }
2128 }
2129
2130 /// Deterministic xorshift64* PRNG, fixed-seeded so runs are reproducible.
2131 struct Rng(u64);
2132 impl Rng {
2133 fn new(seed: u64) -> Self {
2134 // xorshift needs a non-zero state.
2135 Self(if seed == 0 {
2136 0x9E37_79B9_7F4A_7C15
2137 } else {
2138 seed
2139 })
2140 }
2141 fn next_u64(&mut self) -> u64 {
2142 let mut x = self.0;
2143 x ^= x >> 12;
2144 x ^= x << 25;
2145 x ^= x >> 27;
2146 self.0 = x;
2147 x.wrapping_mul(0x2545_F491_4F6C_DD1D)
2148 }
2149 fn below(&mut self, n: usize) -> usize {
2150 (self.next_u64() % n as u64) as usize
2151 }
2152 }
2153
2154 /// A random char-granular mutation of `s`: insert, delete, or replace a
2155 /// span, drawing from an alphabet that mixes ASCII, accented, CJK, and
2156 /// emoji so multi-byte boundaries are exercised.
2157 fn mutate(s: &str, rng: &mut Rng) -> String {
2158 const ALPHABET: [char; 10] = ['a', 'b', '\n', 'é', '日', '本', '🎉', '語', 'x', 'z'];
2159 let chars: Vec<char> = s.chars().collect();
2160 let pick = |rng: &mut Rng| ALPHABET[rng.below(ALPHABET.len())];
2161 match rng.below(3) {
2162 0 => {
2163 let pos = rng.below(chars.len() + 1);
2164 let mut v = chars.clone();
2165 v.insert(pos, pick(rng));
2166 v.into_iter().collect()
2167 }
2168 1 if !chars.is_empty() => {
2169 let pos = rng.below(chars.len());
2170 let mut v = chars.clone();
2171 v.remove(pos);
2172 v.into_iter().collect()
2173 }
2174 _ => {
2175 if chars.is_empty() {
2176 return pick(rng).to_string();
2177 }
2178 let a = rng.below(chars.len());
2179 let b = (a + rng.below(chars.len() - a + 1)).min(chars.len());
2180 let mut v = chars[..a].to_vec();
2181 v.push(pick(rng));
2182 v.extend_from_slice(&chars[b..]);
2183 v.into_iter().collect()
2184 }
2185 }
2186 }
2187
2188 fn entry_str(s: &str) -> UndoEntry {
2189 UndoEntry {
2190 rope: ropey::Rope::from_str(s),
2191 cursor: (0, 0),
2192 timestamp: SystemTime::now(),
2193 marks: MarkSnapshot::default(),
2194 }
2195 }
2196
2197 // ── (0) differential oracle: the pre-chunk-walk `diff` ────────────────────
2198 //
2199 // The original implementation, verbatim, materializing BOTH ropes with
2200 // `to_string()` before scanning bytes. `diff` was rewritten to walk chunks
2201 // instead (no full materialization); this is the semantic pin — the two must
2202 // agree on the EXACT `Delta` for every input, not merely round-trip.
2203
2204 fn diff_reference(parent: &ropey::Rope, child: &ropey::Rope) -> Delta {
2205 let a = parent.to_string();
2206 let b = child.to_string();
2207 let ab = a.as_bytes();
2208 let bb = b.as_bytes();
2209
2210 let max_pre = ab.len().min(bb.len());
2211 let mut pre = 0;
2212 while pre < max_pre && ab[pre] == bb[pre] {
2213 pre += 1;
2214 }
2215 while pre > 0 && !a.is_char_boundary(pre) {
2216 pre -= 1;
2217 }
2218
2219 let max_suf = max_pre - pre;
2220 let mut suf = 0;
2221 while suf < max_suf && ab[ab.len() - 1 - suf] == bb[bb.len() - 1 - suf] {
2222 suf += 1;
2223 }
2224 let mut a_end = ab.len() - suf;
2225 while a_end < ab.len() && !a.is_char_boundary(a_end) {
2226 a_end += 1;
2227 }
2228 let b_end = bb.len() - (ab.len() - a_end);
2229
2230 Delta {
2231 start: a[..pre].chars().count(),
2232 old: a[pre..a_end].to_string(),
2233 new: b[pre..b_end].to_string(),
2234 }
2235 }
2236
2237 /// Assert the chunk-walking `diff` is byte-identical to `diff_reference`,
2238 /// over BOTH single-chunk ropes and multi-chunk ones (ropey only splits past
2239 /// its ~1 KiB leaf size, so short fixtures alone would never exercise the
2240 /// cross-chunk cursor logic).
2241 #[track_caller]
2242 fn assert_diff_matches_reference(sa: &str, sb: &str) {
2243 let a = ropey::Rope::from_str(sa);
2244 let b = ropey::Rope::from_str(sb);
2245 assert_eq!(
2246 diff(&a, &b),
2247 diff_reference(&a, &b),
2248 "diff != reference for {sa:?} -> {sb:?}"
2249 );
2250 // Same content, but built by insertion so the two ropes have DIFFERENT,
2251 // misaligned chunk layouts — the reference sees only bytes, the walker
2252 // sees chunk seams, and they must still agree.
2253 let mut a2 = ropey::Rope::new();
2254 a2.insert(0, sa);
2255 let mut b2 = ropey::Rope::new();
2256 for (i, c) in sb.chars().enumerate() {
2257 b2.insert_char(i, c);
2258 }
2259 assert_eq!(
2260 diff(&a2, &b2),
2261 diff_reference(&a2, &b2),
2262 "diff != reference (misaligned chunks) for {sa:?} -> {sb:?}"
2263 );
2264 }
2265
2266 #[test]
2267 // Deliberately NOT size-scaled for miri: the documents here are sized to span
2268 // several of ropey's ~1 KB leaf chunks, which is the entire property under
2269 // test, so shrinking them would quietly test something weaker. Running them
2270 // interpreted costs >10 min on its own. hjkl-buffer has no `unsafe`, so miri's
2271 // reach is UB in ropey/std on these code paths — already covered by the ~185
2272 // other tests in this crate that do run under it.
2273 #[cfg_attr(miri, ignore = "multi-chunk documents are too slow interpreted")]
2274 fn diff_matches_reference_on_edge_cases() {
2275 let cases: &[(&str, &str)] = &[
2276 // equal / empty
2277 ("", ""),
2278 ("", "a"),
2279 ("a", ""),
2280 ("abc", "abc"),
2281 ("café🎉", "café🎉"),
2282 // prefix-only / suffix-only change
2283 ("abcdef", "abcdefXY"),
2284 ("abcdefXY", "abcdef"),
2285 ("Xabcdef", "abcdef"),
2286 ("abcdef", "Xabcdef"),
2287 // change at position 0 and at the very end
2288 ("abcdef", "Zbcdef"),
2289 ("abcdef", "abcdeZ"),
2290 // overlapping repeats — prefix and suffix scans would collide
2291 ("abcabc", "abc"),
2292 ("abc", "abcabc"),
2293 ("aaaa", "aa"),
2294 ("aa", "aaaa"),
2295 ("abab", "ababab"),
2296 ("xyxyxy", "xyxy"),
2297 // multi-byte chars sitting exactly on the cut points
2298 ("café", "cafés"),
2299 ("cafés", "café"),
2300 ("café", "cafè"),
2301 ("日本語", "日語"),
2302 ("日本語", "日本本語"),
2303 ("🎉🎉🎉", "🎉🎉"),
2304 ("🎉🎉", "🎉🎉🎉"),
2305 ("🎉x🎉", "🎉y🎉"),
2306 ("a🎉b", "a🎊b"),
2307 ("é", "e"),
2308 ("e", "é"),
2309 ("🎉", ""),
2310 ("", "🎉"),
2311 // byte-level suffix match that is NOT a char boundary: the tails of
2312 // 'é' (0xC3 0xA9) and 'é' share no byte, but 日 (E6 97 A5) vs 旦
2313 // (E6 97 A6) share a two-byte prefix mid-codepoint.
2314 ("日", "旦"),
2315 ("x日y", "x旦y"),
2316 ("語", "誤"),
2317 // long enough to be multi-chunk in both ropes
2318 (
2319 &"the quick brown fox ".repeat(400),
2320 &"the quick brown fox ".repeat(400),
2321 ),
2322 ];
2323 for (sa, sb) in cases {
2324 assert_diff_matches_reference(sa, sb);
2325 }
2326
2327 // Multi-chunk with an edit in the middle / at each end.
2328 let big: String = "the quick brown fox jumps over the lazy dog\n".repeat(200);
2329 let mid = big.len() / 2;
2330 let mut edited = big.clone();
2331 edited.insert(mid, 'Z');
2332 assert_diff_matches_reference(&big, &edited);
2333 assert_diff_matches_reference(&edited, &big);
2334 assert_diff_matches_reference(&big, &format!("Z{big}"));
2335 assert_diff_matches_reference(&big, &format!("{big}Z"));
2336 assert_diff_matches_reference(&big, &big.repeat(2));
2337
2338 // Multi-chunk with multi-byte chars straddling likely leaf seams.
2339 let uni: String = "café 日本語 🎉 αβγ\n".repeat(200);
2340 let umid = uni.len() / 2;
2341 let umid = (0..=umid).rev().find(|i| uni.is_char_boundary(*i)).unwrap();
2342 let mut uedited = uni.clone();
2343 uedited.insert(umid, '🎊');
2344 assert_diff_matches_reference(&uni, &uedited);
2345 assert_diff_matches_reference(&uedited, &uni);
2346 }
2347
2348 #[test]
2349 fn diff_matches_reference_over_random_evolving_content() {
2350 let mut rng = Rng::new(0x0BAD_F00D_1234_5678);
2351 let mut s = String::from("seed café 日本語\n🎉");
2352 for _ in 0..stress_iters(4000) {
2353 let t = mutate(&s, &mut rng);
2354 let a = ropey::Rope::from_str(&s);
2355 let b = ropey::Rope::from_str(&t);
2356 assert_eq!(diff(&a, &b), diff_reference(&a, &b), "{s:?} -> {t:?}");
2357 assert_eq!(diff(&b, &a), diff_reference(&b, &a), "{t:?} -> {s:?}");
2358 s = t;
2359 }
2360 }
2361
2362 #[test]
2363 fn diff_matches_reference_on_shared_leaf_clones() {
2364 // The `Arc`-shared-leaf fast path: `child` is a CLONE of `parent` plus
2365 // one edit, so most chunks are pointer-identical. Exercised at several
2366 // edit positions across a multi-chunk rope, plus deletes and the
2367 // degenerate no-op clone.
2368 let base: String = "the quick brown fox jumps over the lazy dog\n".repeat(300);
2369 let parent = ropey::Rope::from_str(&base);
2370 assert_eq!(
2371 diff(&parent, &parent.clone()),
2372 diff_reference(&parent, &parent.clone())
2373 );
2374 let n = parent.len_chars();
2375 for at in [0, 1, n / 4, n / 2, n - 1, n] {
2376 let mut child = parent.clone();
2377 child.insert_char(at, '𝄞');
2378 assert_eq!(
2379 diff(&parent, &child),
2380 diff_reference(&parent, &child),
2381 "@{at}"
2382 );
2383 assert_eq!(
2384 diff(&child, &parent),
2385 diff_reference(&child, &parent),
2386 "@{at}"
2387 );
2388 }
2389 for at in [0, n / 3, n - 10] {
2390 let mut child = parent.clone();
2391 child.remove(at..at + 5);
2392 assert_eq!(
2393 diff(&parent, &child),
2394 diff_reference(&parent, &child),
2395 "-{at}"
2396 );
2397 assert_eq!(
2398 diff(&child, &parent),
2399 diff_reference(&child, &parent),
2400 "-{at}"
2401 );
2402 }
2403 }
2404
2405 #[test]
2406 // Same reasoning as `diff_matches_reference_on_edge_cases`: the 300-line base
2407 // document exists to force multi-chunk ropes, so it is not size-scaled and the
2408 // test is skipped under miri rather than weakened.
2409 #[cfg_attr(miri, ignore = "multi-chunk documents are too slow interpreted")]
2410 fn diff_matches_reference_over_random_multi_chunk_pairs() {
2411 // Random pairs built from a multi-chunk corpus, so chunk seams land in
2412 // arbitrary places relative to the common prefix/suffix.
2413 let mut rng = Rng::new(0xF00D_BEEF_0BAD_C0DE);
2414 let units = ["ab", "café ", "日本語", "🎉", "\n", "x", "語日", "é"];
2415 let build = |rng: &mut Rng| -> String {
2416 let mut s = String::new();
2417 for _ in 0..rng.below(400) {
2418 s.push_str(units[rng.below(units.len())]);
2419 }
2420 s
2421 };
2422 for _ in 0..stress_iters(300) {
2423 let sa = build(&mut rng);
2424 // Half the pairs share a long common prefix/suffix with `sa`.
2425 let sb = if rng.below(2) == 0 {
2426 build(&mut rng)
2427 } else {
2428 let mut t = sa.clone();
2429 if !t.is_empty() {
2430 let cut = rng.below(t.chars().count() + 1);
2431 let byte = t.char_indices().nth(cut).map_or(t.len(), |(i, _)| i);
2432 t.insert_str(byte, "🎊zz");
2433 }
2434 t
2435 };
2436 let a = ropey::Rope::from_str(&sa);
2437 let b = ropey::Rope::from_str(&sb);
2438 assert_eq!(diff(&a, &b), diff_reference(&a, &b));
2439 assert_eq!(diff(&b, &a), diff_reference(&b, &a));
2440 }
2441 }
2442
2443 // ── (i) delta round-trip: apply(diff(a,b))==b and apply_inverse==a ────────
2444
2445 #[test]
2446 fn diff_round_trips_over_random_evolving_content() {
2447 let mut rng = Rng::new(0x1234_5678_9ABC_DEF0);
2448 let mut s = String::from("seed café 日本語\n🎉");
2449 for _ in 0..stress_iters(4000) {
2450 let t = mutate(&s, &mut rng);
2451 let a = ropey::Rope::from_str(&s);
2452 let b = ropey::Rope::from_str(&t);
2453 let d = diff(&a, &b);
2454 assert_eq!(
2455 apply_forward(&a, &d).to_string(),
2456 t,
2457 "forward a->b failed (start={}, old={:?}, new={:?})",
2458 d.start,
2459 d.old,
2460 d.new
2461 );
2462 assert_eq!(
2463 apply_inverse(&b, &d).to_string(),
2464 s,
2465 "inverse b->a failed (start={}, old={:?}, new={:?})",
2466 d.start,
2467 d.old,
2468 d.new
2469 );
2470 s = t;
2471 }
2472 }
2473
2474 #[test]
2475 fn diff_round_trips_over_unrelated_pairs() {
2476 // Disjoint corpus pairs (not just single-edit neighbours) so the diff's
2477 // prefix/suffix logic is stressed on wholly different multi-byte text.
2478 let corpus = [
2479 "",
2480 "a",
2481 "café\n日本語\n",
2482 "🎉🎉🎉",
2483 "abcdef",
2484 "日本",
2485 "x\ny\nz\n",
2486 "aXb",
2487 "café",
2488 "語日本",
2489 "\n\n\n",
2490 "🎉x🎉y🎉",
2491 ];
2492 let mut rng = Rng::new(0xDEAD_BEEF_CAFE_1234);
2493 for _ in 0..stress_iters(3000) {
2494 let sa = corpus[rng.below(corpus.len())];
2495 let sb = corpus[rng.below(corpus.len())];
2496 let a = ropey::Rope::from_str(sa);
2497 let b = ropey::Rope::from_str(sb);
2498 let d = diff(&a, &b);
2499 assert_eq!(apply_forward(&a, &d).to_string(), sb);
2500 assert_eq!(apply_inverse(&b, &d).to_string(), sa);
2501 }
2502 }
2503
2504 // ── non-ASCII edit → undo → redo round-trip (multi-byte across a leave) ───
2505
2506 #[test]
2507 fn non_ascii_edit_undo_redo_round_trip() {
2508 // Edits land INSIDE multi-byte lines; undo/redo must round-trip the exact
2509 // bytes, proving the char-offset delta never splits a codepoint.
2510 let mut d = Driver::new("café\n日本語\n");
2511 d.edit("cafés\n日本語\n");
2512 d.edit("cafés\n日本語です\n");
2513 d.edit("cafés\n日本語です🎉\n");
2514 assert_eq!(d.undo().as_deref(), Some("cafés\n日本語です\n"));
2515 assert_eq!(d.undo().as_deref(), Some("cafés\n日本語\n"));
2516 assert_eq!(d.undo().as_deref(), Some("café\n日本語\n"));
2517 assert_eq!(d.redo().as_deref(), Some("cafés\n日本語\n"));
2518 assert_eq!(d.redo().as_deref(), Some("cafés\n日本語です\n"));
2519 assert_eq!(d.redo().as_deref(), Some("cafés\n日本語です🎉\n"));
2520 // Cold reconstruction of every node still matches (drop all caches).
2521 assert_warm_equals_cold(&mut d.t);
2522 }
2523
2524 // ── (ii) + (iii) random op stream vs a full-snapshot reference model ──────
2525
2526 #[test]
2527 fn tree_matches_full_snapshot_reference_over_random_ops() {
2528 let mut rng = Rng::new(0x9E37_79B9_7F4A_7C15);
2529 let start = "α\nβγ\n日本🎉\n";
2530 let mut real = UndoTree::new(ropey::Rope::from_str(start));
2531 let mut refr = RefTree::new(start);
2532 let mut live = start.to_string();
2533
2534 for step in 0..stress_iters(6000) {
2535 // Structural predicates stay in lockstep with the reference.
2536 assert_eq!(real.is_at_root(), refr.is_at_root(), "is_at_root @ {step}");
2537 assert_eq!(real.has_redo(), refr.has_redo(), "has_redo @ {step}");
2538 assert_eq!(real.depth_from_root(), refr.depth(), "depth @ {step}");
2539
2540 match rng.below(6) {
2541 0 | 1 => {
2542 // Edit: push the PRE-edit state (engine discipline), then
2543 // mutate the live buffer.
2544 let pre = live.clone();
2545 real.push(entry_str(&pre));
2546 refr.push(&pre);
2547 live = mutate(&live, &mut rng);
2548 }
2549 2 => {
2550 let got = real
2551 .undo_step(
2552 ropey::Rope::from_str(&live),
2553 (0, 0),
2554 MarkSnapshot::default(),
2555 )
2556 .map(|e| e.rope.to_string());
2557 let want = refr.undo_step(&live);
2558 assert_eq!(got, want, "undo @ {step}");
2559 if let Some(c) = got {
2560 live = c;
2561 }
2562 }
2563 3 => {
2564 let got = real
2565 .redo_step(
2566 ropey::Rope::from_str(&live),
2567 (0, 0),
2568 MarkSnapshot::default(),
2569 )
2570 .map(|e| e.rope.to_string());
2571 let want = refr.redo_step(&live);
2572 assert_eq!(got, want, "redo @ {step}");
2573 if let Some(c) = got {
2574 live = c;
2575 }
2576 }
2577 4 => {
2578 let got = real
2579 .seq_earlier_step(
2580 ropey::Rope::from_str(&live),
2581 (0, 0),
2582 MarkSnapshot::default(),
2583 )
2584 .map(|e| e.rope.to_string());
2585 let want = refr.seq_earlier_step(&live);
2586 assert_eq!(got, want, "g- @ {step}");
2587 if let Some(c) = got {
2588 live = c;
2589 }
2590 }
2591 _ => {
2592 let got = real
2593 .seq_later_step(
2594 ropey::Rope::from_str(&live),
2595 (0, 0),
2596 MarkSnapshot::default(),
2597 )
2598 .map(|e| e.rope.to_string());
2599 let want = refr.seq_later_step(&live);
2600 assert_eq!(got, want, "g+ @ {step}");
2601 if let Some(c) = got {
2602 live = c;
2603 }
2604 }
2605 }
2606
2607 // The reference model still rewrites the whole root→target chain on
2608 // every landing, so the two agreeing above already says the shortcut
2609 // reproduces it. Check the structure it relies on directly too: an
2610 // `on_path` drift is what would let the shortcut skip a stale
2611 // ancestor, and it is cheap to catch here at every step.
2612 real.assert_path_invariant(&format!("op @ {step}"));
2613
2614 // (ii) Every so often, assert warm and cold materialization agree
2615 // for every node — a cold-reconstructed node must equal the rope the
2616 // full-snapshot model would have held.
2617 if step % 200 == 0 {
2618 assert_warm_equals_cold(&mut real);
2619 }
2620 }
2621 assert_warm_equals_cold(&mut real);
2622 }
2623
2624 // ── (iv) keyframes: accelerated materialize vs the naive root replay ──────
2625 //
2626 // Keyframes (issue #302) pin a materialized rope every `KEYFRAME_INTERVAL`
2627 // nodes so a cold `g-` replays O(K) deltas instead of O(depth). They are a
2628 // CACHE: whatever they accelerate must be bit-identical to replaying every
2629 // delta from the root base with no cache at all. `materialize_naive` is that
2630 // oracle, in the same spirit as `diff_reference` above.
2631
2632 /// For every live node: the keyframe-accelerated `materialize` must equal the
2633 /// naive root-base replay exactly.
2634 #[track_caller]
2635 fn assert_materialize_matches_naive(t: &mut UndoTree) {
2636 for id in t.live_ids() {
2637 let naive = t.materialize_naive(id).to_string();
2638 let got = t.materialize_for_test(id).to_string();
2639 assert_eq!(got, naive, "accelerated != naive root replay for node {id}");
2640 }
2641 }
2642
2643 /// A linear history `n` states deep (so it crosses many keyframe intervals),
2644 /// plus the expected content of each state indexed by `seq`/depth. Every node
2645 /// is finalized, including the tip.
2646 fn deep_linear_history(n: usize) -> (UndoTree, Vec<String>) {
2647 // Under miri the document is shrunk 10x. `n` is deliberately NOT
2648 // touched: the keyframe ladder, the `n > 4 * KEYFRAME_INTERVAL`
2649 // assertion and every depth-related property stay exactly as they are
2650 // on a normal run — only the per-step rope volume drops. Without this
2651 // a single one of these tests ran for over 22 minutes under miri
2652 // (interpreted, not executed) and stalled the weekly job. The base
2653 // keeps its multi-line and multi-byte content, which is the part that
2654 // matters for the rope/delta paths.
2655 let reps = if cfg!(miri) { 2 } else { 20 };
2656 let base: String =
2657 "the quick brown fox\njumps over the lazy dog\ncafé 日本語 🎉\n".repeat(reps);
2658 let mut t = UndoTree::new(ropey::Rope::from_str(&base));
2659 let mut states = vec![base.clone()];
2660 let mut live = base;
2661 for i in 0..n {
2662 // Engine discipline: commit the PRE-edit state, then mutate.
2663 t.push(entry_str(&live));
2664 live = format!("e{i} {live}");
2665 states.push(live.clone());
2666 }
2667 // Stash the tip's live content so no node is left holding a stale edge.
2668 t.sync_current(ropey::Rope::from_str(&live));
2669 (t, states)
2670 }
2671
2672 #[test]
2673 fn deep_history_walks_back_and_forward_exactly() {
2674 // The `:earlier 9999` / `:later 9999` shape, deep enough that most jumps
2675 // land outside the warm window and go through a keyframe.
2676 // 65 under miri still satisfies the `> 4 * KEYFRAME_INTERVAL` floor
2677 // asserted below, so the walk still crosses four keyframes — the
2678 // property under test. See `deep_linear_history` for why.
2679 let n = if cfg!(miri) { 65 } else { 200 };
2680 assert!(n > 4 * KEYFRAME_INTERVAL);
2681 let (mut t, states) = deep_linear_history(n);
2682
2683 let mut live = states[n].clone();
2684 for want in (0..n).rev() {
2685 let got = t
2686 .seq_earlier_step(
2687 ropey::Rope::from_str(&live),
2688 (0, 0),
2689 MarkSnapshot::default(),
2690 )
2691 .expect("history is deeper than the walk");
2692 live = got.rope.to_string();
2693 assert_eq!(live, states[want], "g- onto seq {want}");
2694 }
2695 assert!(
2696 t.seq_earlier_step(
2697 ropey::Rope::from_str(&live),
2698 (0, 0),
2699 MarkSnapshot::default()
2700 )
2701 .is_none(),
2702 "walk ended at the oldest state"
2703 );
2704 for (seq, want) in states.iter().enumerate().skip(1) {
2705 let got = t
2706 .seq_later_step(
2707 ropey::Rope::from_str(&live),
2708 (0, 0),
2709 MarkSnapshot::default(),
2710 )
2711 .expect("history is deeper than the walk");
2712 live = got.rope.to_string();
2713 assert_eq!(&live, want, "g+ onto seq {seq}");
2714 }
2715 assert_materialize_matches_naive(&mut t);
2716 assert_warm_equals_cold(&mut t);
2717 }
2718
2719 /// `by_seq` must name exactly the live nodes, with the right ids. It is a
2720 /// second source of truth for `g-`/`g+` targets, so drift here silently
2721 /// sends history steps to the wrong state (or reports the end of history
2722 /// early) rather than failing loudly.
2723 ///
2724 /// Checked against a brute-force arena scan — the code `node_below` /
2725 /// `node_above` used before the index existed — after pushes, branch
2726 /// creation, undo/redo, pruning to a node budget, `clear_all`, and a
2727 /// serialize round trip, since those are the paths that allocate and free.
2728 #[test]
2729 fn seq_index_matches_the_arena() {
2730 fn brute(t: &UndoTree) -> std::collections::BTreeMap<u64, NodeId> {
2731 t.nodes
2732 .iter()
2733 .enumerate()
2734 .filter_map(|(id, slot)| slot.as_ref().map(|n| (n.seq, id)))
2735 .collect()
2736 }
2737 fn check(t: &UndoTree, when: &str) {
2738 assert_eq!(t.by_seq, brute(t), "seq index drifted after {when}");
2739 // Every step target agrees with a scan of the arena, which is what
2740 // the index replaced.
2741 for probe in 0..t.next_seq + 1 {
2742 let below = t
2743 .nodes
2744 .iter()
2745 .enumerate()
2746 .filter_map(|(id, s)| s.as_ref().map(|n| (n.seq, id)))
2747 .filter(|&(sq, _)| sq < probe)
2748 .max_by_key(|&(sq, _)| sq)
2749 .map(|(_, id)| id);
2750 let above = t
2751 .nodes
2752 .iter()
2753 .enumerate()
2754 .filter_map(|(id, s)| s.as_ref().map(|n| (n.seq, id)))
2755 .filter(|&(sq, _)| sq > probe)
2756 .min_by_key(|&(sq, _)| sq)
2757 .map(|(_, id)| id);
2758 assert_eq!(
2759 t.node_below(probe),
2760 below,
2761 "node_below({probe}) after {when}"
2762 );
2763 assert_eq!(
2764 t.node_above(probe),
2765 above,
2766 "node_above({probe}) after {when}"
2767 );
2768 }
2769 }
2770
2771 let mk = |text: &str| UndoEntry {
2772 rope: ropey::Rope::from_str(text),
2773 cursor: (0, 0),
2774 timestamp: SystemTime::now(),
2775 marks: MarkSnapshot::default(),
2776 };
2777
2778 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
2779 check(&t, "new");
2780
2781 for i in 1..=8 {
2782 t.push(mk(&format!("s{i}")));
2783 }
2784 check(&t, "pushes");
2785
2786 // Undo twice then push: forks a branch, frees nothing.
2787 t.undo_step(ropey::Rope::from_str("s8"), (0, 0), MarkSnapshot::default());
2788 t.undo_step(ropey::Rope::from_str("s7"), (0, 0), MarkSnapshot::default());
2789 t.push(mk("branch"));
2790 check(&t, "branch");
2791
2792 // Enforcing a node budget frees through `free` / `free_subtree`.
2793 t.cap(3);
2794 check(&t, "cap");
2795
2796 let round = UndoTree::from_serializable(&t.to_serializable()).expect("round trip");
2797 check(&round, "from_serializable");
2798
2799 t.clear_all();
2800 check(&t, "clear_all");
2801 }
2802
2803 #[test]
2804 fn keyframes_bound_the_cold_replay_distance() {
2805 // 65 still spans four keyframe intervals, which is what makes the
2806 // per-node replay-distance bound below meaningful. See
2807 // `deep_linear_history` for why miri gets a smaller history.
2808 let n = if cfg!(miri) { 65 } else { 200 };
2809 let (mut t, _) = deep_linear_history(n);
2810 // Steady state: the ordinary warm entries have aged out, the keyframes
2811 // are still pinned. Every node must be within one interval of an anchor.
2812 t.drop_warm_caches();
2813 for id in t.live_ids() {
2814 let d = t.replay_distance(id);
2815 assert!(
2816 d < KEYFRAME_INTERVAL,
2817 "node {id} (depth {}) replays {d} deltas, over the keyframe bound",
2818 t.get(id).depth
2819 );
2820 }
2821 // Drop the keyframes too and the bound is gone — proof that it is the
2822 // keyframes doing the bounding and not the warm LRU or the tree shape.
2823 let deepest = *t
2824 .live_ids()
2825 .iter()
2826 .max_by_key(|&&id| t.get(id).depth)
2827 .unwrap();
2828 t.drop_all_caches();
2829 assert!(t.replay_distance(deepest) > KEYFRAME_INTERVAL);
2830 // One materialize off the fully-cold tree re-pins the whole ladder.
2831 t.materialize_for_test(deepest);
2832 t.drop_warm_caches();
2833 for id in t.live_ids() {
2834 assert!(t.replay_distance(id) < KEYFRAME_INTERVAL, "node {id}");
2835 }
2836 }
2837
2838 #[test]
2839 fn keyframe_materialize_matches_naive_over_random_ops() {
2840 // Push-heavy op mix so the tree gets deep enough to cross many keyframe
2841 // intervals, with undo/redo/g-/g+ and periodic `cap` pruning mixed in —
2842 // pruning renumbers nothing but does free nodes and re-root the tree, so
2843 // it is where a stale keyframe would surface as corrupted text.
2844 let mut rng = Rng::new(0x0FF1_CE00_D15E_A5E5);
2845 let start = "α\nβγ\n日本🎉\nthe quick brown fox\n";
2846 let mut t = UndoTree::new(ropey::Rope::from_str(start));
2847 let mut live = start.to_string();
2848
2849 for step in 0..stress_iters(5000) {
2850 match rng.below(10) {
2851 0..=5 => {
2852 let pre = live.clone();
2853 t.push(entry_str(&pre));
2854 live = mutate(&live, &mut rng);
2855 }
2856 6 => {
2857 if let Some(e) = t.undo_step(
2858 ropey::Rope::from_str(&live),
2859 (0, 0),
2860 MarkSnapshot::default(),
2861 ) {
2862 live = e.rope.to_string();
2863 }
2864 }
2865 7 => {
2866 if let Some(e) = t.redo_step(
2867 ropey::Rope::from_str(&live),
2868 (0, 0),
2869 MarkSnapshot::default(),
2870 ) {
2871 live = e.rope.to_string();
2872 }
2873 }
2874 8 => {
2875 if let Some(e) = t.seq_earlier_step(
2876 ropey::Rope::from_str(&live),
2877 (0, 0),
2878 MarkSnapshot::default(),
2879 ) {
2880 live = e.rope.to_string();
2881 }
2882 }
2883 _ => {
2884 if let Some(e) = t.seq_later_step(
2885 ropey::Rope::from_str(&live),
2886 (0, 0),
2887 MarkSnapshot::default(),
2888 ) {
2889 live = e.rope.to_string();
2890 }
2891 }
2892 }
2893 // Whatever the tree hands back must be what the naive replay of the
2894 // node it landed on says — checked every step, cheaply.
2895 let cur = t.current;
2896 assert_eq!(
2897 t.materialize_for_test(cur).to_string(),
2898 t.materialize_naive(cur).to_string(),
2899 "current node diverged @ {step}"
2900 );
2901 t.assert_path_invariant(&format!("op @ {step}"));
2902 if step % 250 == 0 {
2903 assert_materialize_matches_naive(&mut t);
2904 }
2905 if step % 700 == 0 {
2906 t.cap(60);
2907 t.assert_path_invariant(&format!("cap @ {step}"));
2908 }
2909 }
2910 assert_materialize_matches_naive(&mut t);
2911 assert_warm_equals_cold(&mut t);
2912 }
2913
2914 #[test]
2915 fn deserialized_deep_tree_rebuilds_the_keyframe_ladder() {
2916 // Depth is NOT serialized (keyframes are a cache, the on-disk format is
2917 // untouched), so a loaded tree has to recompute it — otherwise every
2918 // cross-session `g-` would be a full replay again.
2919 // 65 still spans four keyframe intervals, so the reloaded tree must
2920 // still rebuild a real ladder rather than a trivial one. See
2921 // `deep_linear_history` for why miri gets a smaller history.
2922 let n = if cfg!(miri) { 65 } else { 100 };
2923 let (t, states) = deep_linear_history(n);
2924 let ser = t.to_serializable();
2925 let mut back = UndoTree::from_serializable(&ser).expect("valid projection");
2926
2927 let deepest = *back
2928 .live_ids()
2929 .iter()
2930 .max_by_key(|&&id| back.get(id).depth)
2931 .unwrap();
2932 assert_eq!(back.get(deepest).depth, n, "depths recomputed on load");
2933 assert_eq!(back.materialize_for_test(deepest).to_string(), states[n]);
2934 assert_materialize_matches_naive(&mut back);
2935
2936 back.drop_warm_caches();
2937 for id in back.live_ids() {
2938 assert!(back.replay_distance(id) < KEYFRAME_INTERVAL, "node {id}");
2939 }
2940 }
2941
2942 /// For every live node: materialize warm, drop all caches, materialize cold,
2943 /// assert identical. Restores nothing else (test-local).
2944 fn assert_warm_equals_cold(t: &mut UndoTree) {
2945 let ids = t.live_ids();
2946 let warm: Vec<String> = ids
2947 .iter()
2948 .map(|&id| t.materialize_for_test(id).to_string())
2949 .collect();
2950 t.drop_all_caches();
2951 for (i, &id) in ids.iter().enumerate() {
2952 let cold = t.materialize_for_test(id).to_string();
2953 assert_eq!(cold, warm[i], "warm != cold for node {id}");
2954 }
2955 }
2956
2957 /// Engine-faithful driver over the real (delta) [`UndoTree`]: mirrors how
2958 /// `editor.rs` pushes the PRE-edit state and restores returned content.
2959 struct Driver {
2960 t: UndoTree,
2961 live: String,
2962 }
2963 impl Driver {
2964 fn new(s: &str) -> Self {
2965 Self {
2966 t: UndoTree::new(ropey::Rope::from_str(s)),
2967 live: s.to_string(),
2968 }
2969 }
2970 fn edit(&mut self, new: &str) {
2971 self.t.push(entry_str(&self.live));
2972 self.live = new.to_string();
2973 }
2974 fn undo(&mut self) -> Option<String> {
2975 let e = self.t.undo_step(
2976 ropey::Rope::from_str(&self.live),
2977 (0, 0),
2978 MarkSnapshot::default(),
2979 )?;
2980 self.live = e.rope.to_string();
2981 Some(self.live.clone())
2982 }
2983 fn redo(&mut self) -> Option<String> {
2984 let e = self.t.redo_step(
2985 ropey::Rope::from_str(&self.live),
2986 (0, 0),
2987 MarkSnapshot::default(),
2988 )?;
2989 self.live = e.rope.to_string();
2990 Some(self.live.clone())
2991 }
2992 }
2993
2994 /// Full-snapshot reference tree — Phase 2b's model (a whole rope per node),
2995 /// the oracle the delta tree is cross-checked against. Content only (cursor /
2996 /// marks / timestamps are covered by the existing tree tests).
2997 struct RefNode {
2998 parent: Option<usize>,
2999 children: Vec<usize>,
3000 last_child: Option<usize>,
3001 content: String,
3002 seq: u64,
3003 }
3004 struct RefTree {
3005 nodes: Vec<Option<RefNode>>,
3006 current: usize,
3007 next_seq: u64,
3008 }
3009 impl RefTree {
3010 fn new(s: &str) -> Self {
3011 let root = RefNode {
3012 parent: None,
3013 children: Vec::new(),
3014 last_child: None,
3015 content: s.to_string(),
3016 seq: 0,
3017 };
3018 Self {
3019 nodes: vec![Some(root)],
3020 current: 0,
3021 next_seq: 1,
3022 }
3023 }
3024 fn get(&self, id: usize) -> &RefNode {
3025 self.nodes[id].as_ref().unwrap()
3026 }
3027 fn get_mut(&mut self, id: usize) -> &mut RefNode {
3028 self.nodes[id].as_mut().unwrap()
3029 }
3030 fn alloc(&mut self, n: RefNode) -> usize {
3031 self.nodes.push(Some(n));
3032 self.nodes.len() - 1
3033 }
3034 fn is_at_root(&self) -> bool {
3035 self.get(self.current).parent.is_none()
3036 }
3037 fn has_redo(&self) -> bool {
3038 self.get(self.current).last_child.is_some()
3039 }
3040 fn depth(&self) -> usize {
3041 let mut d = 0;
3042 let mut n = self.get(self.current).parent;
3043 while let Some(p) = n {
3044 d += 1;
3045 n = self.get(p).parent;
3046 }
3047 d
3048 }
3049 fn push(&mut self, pre: &str) {
3050 let cur = self.current;
3051 self.get_mut(cur).content = pre.to_string();
3052 let seq = self.next_seq;
3053 self.next_seq += 1;
3054 let child = self.alloc(RefNode {
3055 parent: Some(cur),
3056 children: Vec::new(),
3057 last_child: None,
3058 content: pre.to_string(),
3059 seq,
3060 });
3061 let c = self.get_mut(cur);
3062 c.children.push(child);
3063 c.last_child = Some(child);
3064 self.current = child;
3065 }
3066 fn undo_step(&mut self, live: &str) -> Option<String> {
3067 let cur = self.current;
3068 let par = self.get(cur).parent?;
3069 self.get_mut(cur).content = live.to_string();
3070 self.get_mut(par).last_child = Some(cur);
3071 self.current = par;
3072 Some(self.get(par).content.clone())
3073 }
3074 fn redo_step(&mut self, live: &str) -> Option<String> {
3075 let cur = self.current;
3076 let child = self.get(cur).last_child?;
3077 self.get_mut(cur).content = live.to_string();
3078 self.current = child;
3079 Some(self.get(child).content.clone())
3080 }
3081 fn current_seq(&self) -> u64 {
3082 self.get(self.current).seq
3083 }
3084 fn node_below(&self, s: u64) -> Option<usize> {
3085 let mut best: Option<(u64, usize)> = None;
3086 for (id, slot) in self.nodes.iter().enumerate() {
3087 if let Some(n) = slot
3088 && n.seq < s
3089 && best.is_none_or(|(bs, _)| n.seq > bs)
3090 {
3091 best = Some((n.seq, id));
3092 }
3093 }
3094 best.map(|(_, id)| id)
3095 }
3096 fn node_above(&self, s: u64) -> Option<usize> {
3097 let mut best: Option<(u64, usize)> = None;
3098 for (id, slot) in self.nodes.iter().enumerate() {
3099 if let Some(n) = slot
3100 && n.seq > s
3101 && best.is_none_or(|(bs, _)| n.seq < bs)
3102 {
3103 best = Some((n.seq, id));
3104 }
3105 }
3106 best.map(|(_, id)| id)
3107 }
3108 fn retarget(&mut self, target: usize) {
3109 self.current = target;
3110 let mut node = target;
3111 while let Some(p) = self.get(node).parent {
3112 self.get_mut(p).last_child = Some(node);
3113 node = p;
3114 }
3115 }
3116 fn stash_and_move(&mut self, target: usize, live: &str) {
3117 let cur = self.current;
3118 self.get_mut(cur).content = live.to_string();
3119 self.retarget(target);
3120 }
3121 fn seq_earlier_step(&mut self, live: &str) -> Option<String> {
3122 let target = self.node_below(self.current_seq())?;
3123 self.stash_and_move(target, live);
3124 Some(self.get(target).content.clone())
3125 }
3126 fn seq_later_step(&mut self, live: &str) -> Option<String> {
3127 let target = self.node_above(self.current_seq())?;
3128 self.stash_and_move(target, live);
3129 Some(self.get(target).content.clone())
3130 }
3131 }
3132}
3133
3134// ─── Phase 3b serialize/deserialize tests ─────────────────────────────────────
3135//
3136// The undofile is only as trustworthy as this round-trip: a projection that
3137// loses a branch, mislinks a parent, or reconstructs a node's content wrong
3138// would silently corrupt cross-session undo. These build the headline tree
3139// (5 edits, u, u), project it, rebuild, and assert BOTH the per-node content
3140// (keyed by the stable `seq`) and the live walk (`<C-r>` forward, `u` back)
3141// survive the trip.
3142#[cfg(test)]
3143mod serialize_tests {
3144 use super::*;
3145
3146 fn e(text: &str) -> UndoEntry {
3147 UndoEntry {
3148 rope: ropey::Rope::from_str(text),
3149 cursor: (0, 0),
3150 timestamp: SystemTime::now(),
3151 marks: MarkSnapshot::default(),
3152 }
3153 }
3154 fn l(text: &str) -> (ropey::Rope, (usize, usize), MarkSnapshot) {
3155 (ropey::Rope::from_str(text), (0, 0), MarkSnapshot::default())
3156 }
3157
3158 /// The headline tree: root "s0", five edits to live "s5", then `u` twice so
3159 /// `current` sits on "s3" with the forward branch (s4/s5) retained — exactly
3160 /// the state a `:wq` would persist.
3161 fn headline_tree() -> UndoTree {
3162 let mut t = UndoTree::new(ropey::Rope::from_str("s0"));
3163 for pre in ["s0", "s1", "s2", "s3", "s4"] {
3164 t.push(e(pre)); // engine discipline: push the PRE-edit live state
3165 }
3166 let (r, c, m) = l("s5");
3167 t.undo_step(r, c, m); // -> s4
3168 let (r, c, m) = l("s4");
3169 t.undo_step(r, c, m); // -> s3
3170 t.sync_current(ropey::Rope::from_str("s3")); // stash exact live, like save
3171 t
3172 }
3173
3174 /// Every node's content (keyed by `seq`), materialized cold-then-warm.
3175 fn content_by_seq(t: &mut UndoTree) -> std::collections::BTreeMap<u64, String> {
3176 t.live_ids()
3177 .into_iter()
3178 .map(|id| {
3179 let seq = t.get(id).seq;
3180 (seq, t.materialize_for_test(id).to_string())
3181 })
3182 .collect()
3183 }
3184
3185 #[test]
3186 fn round_trip_reproduces_structure_and_content() {
3187 let mut orig = headline_tree();
3188 let cur_seq = orig.current_node_seq();
3189 let ser = orig.to_serializable();
3190 let orig_content = content_by_seq(&mut orig);
3191
3192 let mut back = UndoTree::from_serializable(&ser).expect("valid projection");
3193 assert_eq!(back.current_node_seq(), cur_seq, "current preserved");
3194 assert_eq!(back.next_seq, orig.next_seq, "next_seq preserved");
3195 // Force cold reconstruction (fresh tree has no warm caches) and compare.
3196 assert_eq!(
3197 content_by_seq(&mut back),
3198 orig_content,
3199 "content at every node reproduced"
3200 );
3201 // Six states: s0..s5.
3202 assert_eq!(orig_content.len(), 6);
3203 assert_eq!(orig_content[&3], "s3");
3204 assert_eq!(orig_content[&5], "s5");
3205 }
3206
3207 #[test]
3208 fn deserialized_tree_walks_forward_and_back() {
3209 let ser = headline_tree().to_serializable();
3210 let mut t = UndoTree::from_serializable(&ser).unwrap();
3211 // `<C-r>` twice: s3 -> s4 -> s5 (the retained forward branch).
3212 let (r, c, m) = l("s3");
3213 assert_eq!(t.redo_step(r, c, m).unwrap().rope.to_string(), "s4");
3214 let (r, c, m) = l("s4");
3215 assert_eq!(t.redo_step(r, c, m).unwrap().rope.to_string(), "s5");
3216 // `u` all the way back to the root.
3217 let mut live = "s5".to_string();
3218 for want in ["s4", "s3", "s2", "s1", "s0"] {
3219 let (r, c, m) = l(&live);
3220 assert_eq!(t.undo_step(r, c, m).unwrap().rope.to_string(), want);
3221 live = want.to_string();
3222 }
3223 assert!(t.is_at_root());
3224 }
3225
3226 #[test]
3227 fn from_serializable_rejects_out_of_range_current() {
3228 let mut ser = headline_tree().to_serializable();
3229 ser.current = ser.nodes.len() as u32; // past the end
3230 assert!(UndoTree::from_serializable(&ser).is_none());
3231 }
3232
3233 /// A tree without a root base is structurally invalid outside the swap
3234 /// context (where the body supplies the base). `to_serializable` always
3235 /// emits `Some`; a `None` can only come from the swap writer's dedup, which
3236 /// the swap reader reverses before rebuilding.
3237 #[test]
3238 fn from_serializable_rejects_missing_base() {
3239 let mut ser = headline_tree().to_serializable();
3240 ser.base = None;
3241 assert!(UndoTree::from_serializable(&ser).is_none());
3242 }
3243
3244 /// A `last_child` naming an in-bounds node that is not one of the node's
3245 /// own children is a hand-edit a `to_serializable` file could never
3246 /// produce. Range validation passes it; without the repair the first
3247 /// `<C-r>` from that node would walk into the wrong subtree. The loader
3248 /// repairs rather than rejects (matching the path-chain repair).
3249 #[test]
3250 fn from_serializable_clears_last_child_that_names_a_non_child() {
3251 let mut ser = headline_tree().to_serializable();
3252 let cur = ser.current as usize;
3253 let non_child = (0..ser.nodes.len() as u32)
3254 .find(|&i| i != ser.root && !ser.nodes[cur].children.contains(&i))
3255 .expect("the headline tree has an off-path node");
3256 ser.nodes[cur].last_child = Some(non_child);
3257 let t = UndoTree::from_serializable(&ser).expect("repair, not rejection");
3258 assert_eq!(
3259 t.get(t.current).last_child,
3260 None,
3261 "a last_child that is not a child must be cleared on load"
3262 );
3263 // And the structural validity of every other link survives: the tree
3264 // still walks (a redo from current finds no forward branch).
3265 assert!(t.current_node_seq() > 0);
3266 }
3267
3268 #[test]
3269 fn from_serializable_rejects_non_root_missing_delta() {
3270 let mut ser = headline_tree().to_serializable();
3271 // Blank a non-root node's delta ⇒ structurally invalid ⇒ rejected.
3272 let victim = if ser.root == 0 { 1 } else { 0 };
3273 ser.nodes[victim].delta = None;
3274 assert!(UndoTree::from_serializable(&ser).is_none());
3275 }
3276
3277 /// A parent-link cycle must be rejected at load, not walked at use.
3278 /// `materialize` follows `parent` in an unbounded loop, so a cycle there
3279 /// hangs while growing `path` — this is the only place that can see it.
3280 #[test]
3281 fn from_serializable_rejects_a_parent_link_cycle() {
3282 let mut ser = headline_tree().to_serializable();
3283 assert!(ser.nodes.len() > 2, "fixture needs three nodes");
3284 // Point two non-root nodes at each other, both ways, so the links stay
3285 // mutually consistent and every node stays reachable from the root
3286 // through the list it was already in. Only the single-lister rule
3287 // rejects this.
3288 let (a, b) = match ser.root {
3289 0 => (1, 2),
3290 1 => (0, 2),
3291 _ => (0, 1),
3292 };
3293 ser.nodes[a].parent = Some(b as u32);
3294 ser.nodes[b].parent = Some(a as u32);
3295 ser.nodes[a].children.push(b as u32);
3296 ser.nodes[b].children.push(a as u32);
3297 assert!(UndoTree::from_serializable(&ser).is_none());
3298 }
3299
3300 /// A parent link that the named parent does not mirror as a child leaves
3301 /// the two walks (`children` forward, `parent` back) disagreeing.
3302 #[test]
3303 fn from_serializable_rejects_an_unmirrored_parent_link() {
3304 let mut ser = headline_tree().to_serializable();
3305 let victim = if ser.root == 0 { 1 } else { 0 };
3306 let parent = ser.nodes[victim].parent.expect("non-root") as usize;
3307 ser.nodes[parent].children.retain(|&c| c != victim as u32);
3308 assert!(UndoTree::from_serializable(&ser).is_none());
3309 }
3310
3311 /// A cycle that hangs off no root at all: nodes 1 and 2 name each other
3312 /// both ways, so the lists still partition — reachability from the root is
3313 /// the guard that catches this one. This is the shape that hung
3314 /// `install_recovered_undo_tree` before the loader checked for it.
3315 #[test]
3316 fn from_serializable_rejects_a_cycle_unreachable_from_the_root() {
3317 let d = || {
3318 Some(Delta {
3319 start: 0,
3320 old: String::new(),
3321 new: String::from("x"),
3322 })
3323 };
3324 let n = |parent, children, delta, seq| SerNode {
3325 parent,
3326 children,
3327 last_child: None,
3328 delta,
3329 cursor: (0, 0),
3330 timestamp_unix_ms: 0,
3331 marks: MarkSnapshot::default(),
3332 seq,
3333 };
3334 let ser = SerTree {
3335 base: Some("hello".into()),
3336 nodes: vec![
3337 n(None, vec![], None, 0),
3338 n(Some(2), vec![2], d(), 1),
3339 n(Some(1), vec![1], d(), 2),
3340 ],
3341 root: 0,
3342 current: 1,
3343 next_seq: 3,
3344 };
3345 assert!(UndoTree::from_serializable(&ser).is_none());
3346 }
3347
3348 /// `by_seq` is keyed by `seq`, so a repeat would silently drop a node from
3349 /// the `g-` / `g+` index while leaving it in the arena.
3350 #[test]
3351 fn from_serializable_rejects_a_repeated_seq() {
3352 let mut ser = headline_tree().to_serializable();
3353 let victim = if ser.root == 0 { 1 } else { 0 };
3354 let other = if victim == 0 { 1 } else { 0 };
3355 ser.nodes[victim].seq = ser.nodes[other].seq;
3356 assert!(UndoTree::from_serializable(&ser).is_none());
3357 }
3358
3359 #[test]
3360 fn multibyte_content_survives_round_trip() {
3361 let mut t = UndoTree::new(ropey::Rope::from_str("café\n日本語"));
3362 t.push(e("café\n日本語"));
3363 t.push(e("cafés\n日本語"));
3364 t.sync_current(ropey::Rope::from_str("cafés\n日本語です🎉"));
3365 let want = content_by_seq(&mut t);
3366 let ser = t.to_serializable();
3367 let mut back = UndoTree::from_serializable(&ser).unwrap();
3368 assert_eq!(content_by_seq(&mut back), want);
3369 }
3370}