@alteran/astro 0.3.8 → 0.5.2

package/src/pages/xrpc/com.atproto.sync.getRepo.ts

@@ -1,17 +1,62 @@
 import type { APIContext } from 'astro';
+import { errorMessage } from '../../lib/errors';
 import { buildRepoCar } from '../../services/car';
 
 export const prerender = false;
 
+/**
+ * com.atproto.sync.getRepo
+ * Returns a CAR snapshot of the repo for initial crawl/index.
+ */
 export async function GET({ locals, request }: APIContext) {
   const { env } = locals.runtime;
   const url = new URL(request.url);
-  const did = url.searchParams.get('did') ?? (env.PDS_DID ?? 'did:example:single-user');
-  const car = await buildRepoCar(env, did);
-  return new Response(car.bytes as any, {
+  const did = url.searchParams.get('did') ?? (env.PDS_DID as string);
+  // Phase 1: accept but ignore since param (diff to be implemented later)
+  const _since = url.searchParams.get('since');
+
+  try {
+    const { bytes } = await buildRepoCar(env, did);
+    const stream = new ReadableStream<Uint8Array>({
+      start(controller) {
+        controller.enqueue(bytes);
+        controller.close();
+      },
+    });
+    return new Response(stream as any, {
+      status: 200,
+      headers: {
+        // Official content type for CAR v1
+        'Content-Type': 'application/vnd.ipld.car',
+        'Cache-Control': 'no-store',
+      },
+    });
+  } catch (error) {
+    const msg = String(errorMessage(error) || error);
+    // Map to lexicon-specified errors
+    const known = ['RepoNotFound', 'RepoTakendown', 'RepoSuspended', 'RepoDeactivated'];
+    const name = known.find((n) => msg.includes(n)) || (msg.includes('HeadNotFound') ? 'RepoNotFound' : null);
+    if (name) {
+      return new Response(JSON.stringify({ error: name, message: msg }), {
+        status: 400,
+        headers: { 'Content-Type': 'application/json' },
+      });
+    }
+    console.error('getRepo (CAR) error:', error);
+    return new Response(JSON.stringify({ error: 'InternalServerError', message: msg }), {
+      status: 500,
+      headers: { 'Content-Type': 'application/json' },
+    });
+  }
+}
+
+// Avoid heavy CAR construction for HEAD. Respond with headers only.
+export async function HEAD() {
+  return new Response(null, {
+    status: 200,
     headers: {
-      'content-type': 'application/vnd.ipld.car; version=1',
-      'content-disposition': 'inline; filename="repo.car"',
+      'Content-Type': 'application/vnd.ipld.car',
+      'Cache-Control': 'no-store',
     },
   });
 }

package/src/pages/xrpc/com.atproto.sync.getRepoStatus.ts

@@ -0,0 +1,58 @@
+import type { APIContext } from 'astro';
+import { getRoot as getRepoRoot } from '../../db/repo';
+import { getAccountState } from '../../db/dal';
+import { toWireStatus } from '../../lib/account-state';
+
+export const prerender = false;
+
+/**
+ * com.atproto.sync.getRepoStatus
+ * Mirrors upstream PDS: returns did, active, optional status, and rev if active.
+ */
+export async function GET({ locals, request }: APIContext) {
+  const { env } = locals.runtime;
+  const url = new URL(request.url);
+  const configuredDid = typeof env.PDS_DID === 'string' ? env.PDS_DID : '';
+  const did = url.searchParams.get('did') ?? configuredDid;
+
+  try {
+    // Best-effort FSM lookup: an unmigrated row or a transient DB error both
+    // fall through to active=true so reads aren't blocked by an internal hiccup.
+    let active = true;
+    let status: string | undefined;
+    try {
+      const state = await getAccountState(env, did);
+      if (state) {
+        const wire = toWireStatus(state);
+        active = wire.active;
+        status = wire.status;
+      }
+    } catch (stateError) {
+      console.warn('getAccountState failed:', stateError);
+    }
+
+    let rev: string | undefined;
+    if (active) {
+      const head = await getRepoRoot(env);
+      if (head?.rev) rev = String(head.rev);
+    }
+
+    return new Response(
+      JSON.stringify({ did, active, ...(status ? { status } : {}), ...(rev ? { rev } : {}) }),
+      { status: 200, headers: { 'Content-Type': 'application/json' } },
+    );
+  } catch (error) {
+    const message = error instanceof Error ? error.message : String(error);
+    if (message.includes('RepoNotFound')) {
+      return new Response(JSON.stringify({ error: 'RepoNotFound', message }), {
+        status: 400,
+        headers: { 'Content-Type': 'application/json' },
+      });
+    }
+    return new Response(JSON.stringify({ error: 'InternalServerError', message }), {
+      status: 500,
+      headers: { 'Content-Type': 'application/json' },
+    });
+  }
+}
+

package/src/pages/xrpc/com.atproto.sync.listBlobs.ts

@@ -12,7 +12,7 @@ export const prerender = false;
 export async function GET({ locals, url }: APIContext) {
   const { env } = locals.runtime;
 
-  const did = url.searchParams.get('did') || env.PDS_DID || 'did:example:single-user';
+  const did = url.searchParams.get('did') || (env.PDS_DID as string);
   const since = url.searchParams.get('since') || '';
   const limit = parseInt(url.searchParams.get('limit') || '500', 10);
 

package/src/pages/xrpc/com.atproto.sync.listRepos.ts

@@ -1,4 +1,5 @@
 import type { APIContext } from 'astro';
+import { getRoot as getRepoRoot } from '../../db/repo';
 
 export const prerender = false;
 
@@ -11,14 +12,15 @@ export async function GET({ locals, url }: APIContext) {
 
   const did = env.PDS_DID || 'did:example:single-user';
   const handle = env.PDS_HANDLE || 'user.example.com';
+  const head = await getRepoRoot(env);
 
   return new Response(
     JSON.stringify({
       repos: [
         {
           did,
-          head: '', // TODO: Get from repo_root
-          rev: '', // TODO: Get from repo_root
+          head: head?.commitCid ?? null,
+          rev: head?.rev ?? null,
           active: true,
         },
       ],
@@ -28,4 +30,4 @@ export async function GET({ locals, url }: APIContext) {
       headers: { 'Content-Type': 'application/json' },
     }
   );
-}
+}

package/src/services/car.ts

@@ -7,6 +7,7 @@ import { CID } from 'multiformats/cid';
 import * as dagCbor from '@ipld/dag-cbor';
 import { sha256 } from 'multiformats/hashes/sha2';
 import { MST, Leaf, D1Blockstore } from '../lib/mst';
+import { NotFound } from '../lib/errors';
 
 export type CarSnapshot = {
   bytes: Uint8Array;
@@ -67,17 +68,22 @@ export async function buildRepoCar(env: Env, did: string): Promise<CarSnapshot>
         };
 
         const mstRoot = CID.parse(String(parsed.data));
-        // 1) Add all MST node blocks
-        await addMstBlocks(blockstore, mstRoot, seen, blocks);
+        // 1) Add all MST node blocks (batched, non-recursive) and collect leaf CIDs
+        const { mstBlocks, leafCids } = await collectMstBfs(blockstore, mstRoot);
+        for (const [cid, bytes] of mstBlocks) {
+          const k = cid.toString();
+          if (seen.has(k)) continue;
+          seen.add(k);
+          blocks.push({ cid, bytes });
+        }
 
-        // 2) Add all record leaf blocks by walking the MST
-        try {
-          const mst = MST.load(blockstore, mstRoot);
-          for await (const leaf of mst.walkLeavesFrom('')) {
-            await addBlock(leaf.value);
-          }
-        } catch (e) {
-          console.warn('Snapshot: failed traversing MST leaves:', e);
+        // 2) Add record leaf blocks by batched fetch
+        const leafFetched = await blockstore.getMany(leafCids);
+        for (const [cidStr, bytes] of leafFetched.blocks.entries()) {
+          const cid = CID.parse(cidStr);
+          if (seen.has(cidStr)) continue;
+          seen.add(cidStr);
+          blocks.push({ cid, bytes });
         }
 
         const bytes = encodeCar([commitCid], blocks);
@@ -88,20 +94,8 @@ export async function buildRepoCar(env: Env, did: string): Promise<CarSnapshot>
       console.warn('Failed to reconstruct signed commit from tip; falling back to snapshot:', e);
     }
   }
-
-  // Fallback: deterministic snapshot built from current records
-  const rows = await listRecords(env);
-  const blocks: { cid: CID; bytes: Uint8Array }[] = [];
-  for (const r of rows) {
-    if (!r.uri.startsWith(`at://${did}/`)) continue;
-    const value = JSON.parse(r.json);
-    const block = await encodeRecordBlock(value);
-    blocks.push(block);
-  }
-  const commitObj = { type: 'commit', did, records: blocks.map((b) => b.cid.toString()).sort() };
-  const commit = await encodeRecordBlock(commitObj);
-  const bytes = encodeCar([commit.cid], [...blocks, commit]);
-  return { bytes, root: commit.cid, blocks: [...blocks, commit] };
+  // No authoritative head to build from
+  throw new NotFound('RepoNotFound');
 }
 
 export async function buildRepoCarRange(env: Env, fromSeq: number, toSeq: number): Promise<CarSnapshot> {
@@ -179,6 +173,7 @@ export async function encodeBlocksForCommit(
   commitCid: CID,
   mstRoot: CID,
   ops: Array<{ path: string; cid: CID | null }>,
+  newMstBlocks?: Array<[CID, Uint8Array]>,
 ): Promise<Uint8Array> {
   const blockstore = new D1Blockstore(env);
   const blocks: { cid: CID; bytes: Uint8Array }[] = [];
@@ -189,18 +184,60 @@ export async function encodeBlocksForCommit(
     const cidStr = cid.toString();
     if (seen.has(cidStr)) return;
     seen.add(cidStr);
-
-    const bytes = await blockstore.get(cid);
-    if (bytes) {
-      blocks.push({ cid, bytes });
+    let bytes = await blockstore.get(cid);
+    if (!bytes) {
+      // Attempt to reconstruct commit block from commit_log if this is the commit cid
+      if (cidStr === commitCid.toString()) {
+        try {
+          const row = await (env.DB as any)
+            .prepare('SELECT data, sig FROM commit_log WHERE cid = ? LIMIT 1')
+            .bind(cidStr)
+            .first();
+          if (row && row.data && row.sig) {
+            const parsed = JSON.parse(String(row.data));
+            const sigBin = atob(String(row.sig));
+            const sig = new Uint8Array(sigBin.length);
+            for (let i = 0; i < sigBin.length; i++) sig[i] = sigBin.charCodeAt(i);
+            const signedCommit = {
+              did: String(parsed.did),
+              version: Number(parsed.version),
+              data: CID.parse(String(parsed.data)),
+              rev: String(parsed.rev),
+              prev: parsed.prev ? CID.parse(String(parsed.prev)) : null,
+              sig,
+            } as const;
+            bytes = dagCbor.encode(signedCommit);
+          }
+        } catch (e) {
+          console.warn('Failed to reconstruct commit block from commit_log:', e);
+        }
+      }
     }
+    if (bytes) blocks.push({ cid, bytes });
   };
 
   // 1. Add commit block
   await addBlock(commitCid);
 
-  // 2. Add MST nodes by traversing the tree
-  await addMstBlocks(blockstore, mstRoot, seen, blocks);
+  // 2. Add MST nodes
+  if (newMstBlocks && newMstBlocks.length > 0) {
+    // Prefer the exact set of MST nodes touched by this commit
+    for (const [cid, bytes] of newMstBlocks) {
+      const cidStr = cid.toString();
+      if (seen.has(cidStr)) continue;
+      seen.add(cidStr);
+      blocks.push({ cid, bytes });
+    }
+  } else {
+    // Fallback: add MST nodes by batched BFS
+    const { mstBlocks } = await collectMstBfs(blockstore, mstRoot);
+    for (const [cid, bytes] of mstBlocks) {
+      const k = cid.toString();
+      if (seen.has(k)) continue;
+      seen.add(k);
+      blocks.push({ cid, bytes });
+    }
+  }
 
   // 3. Add record blocks for all operations
   for (const op of ops) {
@@ -213,37 +250,152 @@ export async function encodeBlocksForCommit(
   return encodeCar([commitCid], blocks);
 }
 
+/**
+ * Build a CAR proving existence or non-existence of a record at collection/rkey
+ * Root is the latest signed commit block; includes MST path nodes and record block if present.
+ */
+export async function buildRecordProofCar(
+  env: Env,
+  did: string,
+  collection: string,
+  rkey: string,
+): Promise<{ bytes: Uint8Array }> {
+  const db = drizzle(env.DB);
+  const tip = await db.select().from(commit_log).orderBy(desc(commit_log.seq)).limit(1).get();
+  if (!tip) {
+    throw new NotFound('HeadNotFound');
+  }
+
+  // Reconstruct signed commit block and CID
+  const parsed = JSON.parse(tip.data as any);
+  const prevStr = parsed.prev ?? null;
+  const commitObj = {
+    did: String(parsed.did),
+    version: Number(parsed.version),
+    data: CID.parse(String(parsed.data)),
+    rev: String(parsed.rev),
+    prev: prevStr ? CID.parse(String(prevStr)) : null,
+    sig: (() => {
+      const bin = atob(String(tip.sig));
+      const u8 = new Uint8Array(bin.length);
+      for (let i = 0; i < bin.length; i++) u8[i] = bin.charCodeAt(i);
+      return u8;
+    })(),
+  } as const;
+  const commitBytes = dagCbor.encode(commitObj);
+  const hash = await sha256.digest(commitBytes);
+  const commitCid = CID.createV1(dagCbor.code, hash);
+
+  // Walk MST path to the target key
+  const blockstore = new D1Blockstore(env);
+  const mstRoot = CID.parse(String(parsed.data));
+  const key = `${collection}/${rkey}`;
+  const pathBlocks: Array<{ cid: CID; bytes: Uint8Array }> = [];
+  let cursor: CID | null = mstRoot;
+  let recordCid: CID | null = null;
+
+  while (cursor) {
+    const bytes = await blockstore.get(cursor);
+    if (!bytes) break;
+    pathBlocks.push({ cid: cursor, bytes });
+    try {
+      const node: any = dagCbor.decode(bytes);
+      // Reconstruct ordered entries: [l? subtree], then (leaf, subtree?)*
+      type Entry = { kind: 'tree'; cid: CID } | { kind: 'leaf'; key: string; value: CID };
+      const entries: Entry[] = [];
+      if (node?.l) entries.push({ kind: 'tree', cid: CID.asCID(node.l) ?? CID.parse(String(node.l)) });
+      let lastKey = '';
+      for (const e of (node?.e ?? [])) {
+        const keyStr = new TextDecoder('ascii').decode(e.k as Uint8Array);
+        const fullKey = lastKey.slice(0, e.p as number) + keyStr;
+        entries.push({ kind: 'leaf', key: fullKey, value: CID.asCID(e.v) ?? CID.parse(String(e.v)) });
+        lastKey = fullKey;
+        if (e.t) entries.push({ kind: 'tree', cid: CID.asCID(e.t) ?? CID.parse(String(e.t)) });
+      }
+      // Find first leaf >= key
+      let index = entries.findIndex((en) => en.kind === 'leaf' && (en as any).key >= key);
+      if (index < 0) index = entries.length;
+      const found = entries[index];
+      if (found && found.kind === 'leaf' && (found as any).key === key) {
+        recordCid = found.value;
+        break;
+      }
+      const prev = entries[index - 1];
+      if (prev && prev.kind === 'tree') {
+        cursor = prev.cid;
+        continue;
+      }
+      // Not found and no subtree to descend
+      break;
+    } catch {
+      break;
+    }
+  }
+
+  // Assemble CAR: commit as root; include path nodes; include record block if present
+  const blocks: { cid: CID; bytes: Uint8Array }[] = [{ cid: commitCid, bytes: commitBytes }];
+  const seen = new Set<string>([commitCid.toString()]);
+  for (const b of pathBlocks) {
+    const s = b.cid.toString();
+    if (!seen.has(s)) { seen.add(s); blocks.push(b); }
+  }
+  if (recordCid) {
+    const bytes = await blockstore.get(recordCid);
+    if (bytes) blocks.push({ cid: recordCid, bytes });
+  }
+  const bytes = encodeCar([commitCid], blocks);
+  return { bytes };
+}
+
 /**
  * Recursively add all MST node blocks
  */
-async function addMstBlocks(
+async function collectMstBfs(
   blockstore: D1Blockstore,
   rootCid: CID,
-  seen: Set<string>,
-  blocks: { cid: CID; bytes: Uint8Array }[],
-): Promise<void> {
-  const cidStr = rootCid.toString();
-  if (seen.has(cidStr)) return;
-  seen.add(cidStr);
-
-  // Add the MST node block itself
-  const bytes = await blockstore.get(rootCid);
-  if (!bytes) return;
-  blocks.push({ cid: rootCid, bytes });
-
-  // Load MST and traverse children
-  try {
-    const mst = MST.load(blockstore, rootCid);
-    const entries = await mst.getEntries();
-
-    for (const entry of entries) {
-      if (entry.isTree()) {
-        // Recursively add child MST blocks
-        const childCid = await entry.getPointer();
-        await addMstBlocks(blockstore, childCid, seen, blocks);
+): Promise<{ mstBlocks: Array<[CID, Uint8Array]>; leafCids: CID[] }> {
+  const mstBlocks: Array<[CID, Uint8Array]> = [];
+  const leafCids: CID[] = [];
+  const seen = new Set<string>();
+
+  let toFetch: CID[] = [rootCid];
+  const BATCH = 200;
+
+  while (toFetch.length > 0) {
+    const chunk = toFetch.slice(0, BATCH);
+    toFetch = toFetch.slice(BATCH);
+    const { blocks, missing } = await blockstore.getMany(chunk);
+    // Push node blocks we found
+    for (const [cidStr, bytes] of blocks.entries()) {
+      if (seen.has(cidStr)) continue;
+      seen.add(cidStr);
+      mstBlocks.push([CID.parse(cidStr), bytes]);
+    }
+    // Decode nodes to collect children and leaves
+    for (const [cidStr, bytes] of blocks.entries()) {
+      try {
+        const node: any = dagCbor.decode(bytes);
+        const l = node?.l ? (CID.asCID(node.l) ?? CID.parse(String(node.l))) : null;
+        if (l) {
+          const key = l.toString();
+          if (!seen.has(key)) toFetch.push(l);
+        }
+        const entries: any[] = Array.isArray(node?.e) ? node.e : [];
+        for (const e of entries) {
+          const v = CID.asCID(e?.v) ?? CID.parse(String(e?.v));
+          if (v) leafCids.push(v);
+          const t = e?.t ? (CID.asCID(e.t) ?? CID.parse(String(e.t))) : null;
+          if (t) {
+            const key = t.toString();
+            if (!seen.has(key)) toFetch.push(t);
+          }
+        }
+      } catch (error) {
+        console.warn('collectMstBfs: failed to decode node', cidStr, error);
       }
     }
-  } catch (error) {
-    console.error('Error traversing MST:', error);
+    // Ignore missing here; caller might not need full tree for snapshots
   }
+
+  return { mstBlocks, leafCids };
 }

package/src/services/r2-blob-store.ts

@@ -69,7 +69,7 @@ export class R2BlobStore {
     if (size > limit) throw new Error(`BlobTooLarge:${size}>${limit}`);
 
     const contentType = opts.contentType ?? 'application/octet-stream';
-    const sha = await crypto.subtle.digest('SHA-256', view);
+    const sha = await crypto.subtle.digest('SHA-256', R2BlobStore.toArrayBuffer(view));
     const shaB64 = R2BlobStore.b64url(sha);
     const key = R2BlobStore.cidKey(shaB64);
     const buffer = R2BlobStore.toArrayBuffer(view);

package/src/services/repo/blockstore-ops.ts

@@ -0,0 +1,29 @@
+import { CID } from 'multiformats/cid';
+import * as dagCbor from '@ipld/dag-cbor';
+import { cidForCbor } from '../../lib/mst/util';
+import type { D1Blockstore, MST, BlockMap } from '../../lib/mst';
+
+export async function storeRecord(
+  blockstore: D1Blockstore,
+  record: unknown,
+): Promise<CID> {
+  const bytes = dagCbor.encode(record);
+  const cid = await cidForCbor(record);
+  await blockstore.put(cid, bytes);
+  return cid;
+}
+
+export async function storeMstBlocks(
+  blockstore: D1Blockstore,
+  mst: MST,
+): Promise<BlockMap> {
+  const diff = await mst.getUnstoredBlocks();
+  for (const [cid, bytes] of diff.blocks) {
+    console.log(
+      `[RepoManager] Storing new MST block: ${cid.toString()}, size: ${bytes.length}`,
+    );
+    await blockstore.put(cid, bytes);
+  }
+  console.log(`[RepoManager] Stored ${diff.blocks.size} new MST blocks`);
+  return diff.blocks;
+}

package/src/services/repo/operations.ts

@@ -0,0 +1,133 @@
+import { CID } from 'multiformats/cid';
+import * as dagCbor from '@ipld/dag-cbor';
+import type { D1Blockstore, NodeData } from '../../lib/mst';
+import type { RepoOp } from '../../lib/firehose/frames';
+
+interface NodeEntry {
+  p: number;
+  k: Uint8Array;
+  v: unknown;
+  t?: unknown;
+}
+
+interface DecodedNode {
+  e?: NodeEntry[];
+  l?: unknown;
+}
+
+function coerceCid(value: unknown): CID {
+  const asCid = (CID as unknown as { asCID?: (v: unknown) => CID | null }).asCID?.(value);
+  if (asCid) return asCid;
+  return CID.parse(String(value));
+}
+
+export async function collectLeavesBatched(
+  blockstore: D1Blockstore,
+  root: CID,
+): Promise<Map<string, CID>> {
+  const result = new Map<string, CID>();
+  const visited = new Set<string>();
+  let toFetch: CID[] = [root];
+
+  // Limit per getMany() request; getMany chunks the IN() list further.
+  const batchSize = 200;
+
+  while (toFetch.length > 0) {
+    const chunk = toFetch.slice(0, batchSize);
+    toFetch = toFetch.slice(batchSize);
+
+    const { blocks, missing } = await blockstore.getMany(chunk);
+
+    if (missing.length > 0) {
+      // Fail fast: missing MST nodes mean an incomplete repo, and emitting
+      // ops without them would produce a wrong diff.
+      const missingStr = missing.map((c) => c.toString()).join(', ');
+      throw new Error(
+        `[RepoManager] collectLeavesBatched: missing MST nodes: ${missingStr}`,
+      );
+    }
+
+    for (const [cidStr, bytes] of blocks.entries()) {
+      if (visited.has(cidStr)) continue;
+      visited.add(cidStr);
+      try {
+        const node = dagCbor.decode(bytes) as DecodedNode;
+        let lastKey = '';
+        const entries = Array.isArray(node.e) ? node.e : [];
+        for (const entry of entries) {
+          const keyStr = new TextDecoder('ascii').decode(entry.k);
+          const fullKey = lastKey.slice(0, Number(entry.p)) + keyStr;
+          try {
+            const parts = fullKey.split('/');
+            if (parts.length === 2 && parts[0] && parts[1]) {
+              result.set(fullKey, coerceCid(entry.v));
+            }
+          } catch (decodeError) {
+            console.warn('[RepoManager] failed to decode leaf CID:', decodeError);
+          }
+          lastKey = fullKey;
+
+          if (entry.t) {
+            const subtree = coerceCid(entry.t);
+            if (!visited.has(subtree.toString())) toFetch.push(subtree);
+          }
+        }
+
+        if (node.l) {
+          const left = coerceCid(node.l);
+          if (!visited.has(left.toString())) toFetch.push(left);
+        }
+      } catch (error) {
+        console.warn('[RepoManager] collectLeavesBatched: failed to decode node', cidStr, error);
+      }
+    }
+  }
+
+  return result;
+}
+
+export async function extractOps(
+  blockstore: D1Blockstore,
+  prevRoot: CID | null,
+  newRoot: CID,
+): Promise<RepoOp[]> {
+  const ops: RepoOp[] = [];
+  const newMap = await collectLeavesBatched(blockstore, newRoot);
+  const prevMap = prevRoot
+    ? await collectLeavesBatched(blockstore, prevRoot)
+    : new Map<string, CID>();
+
+  for (const [path, cid] of Array.from(newMap.entries())) {
+    const prevCid = prevMap.get(path);
+    if (!prevCid) {
+      ops.push({ action: 'create', path, cid });
+    } else if (!prevCid.equals(cid)) {
+      ops.push({ action: 'update', path, cid, prev: prevCid });
+    }
+  }
+
+  for (const [path, prevCid] of Array.from(prevMap.entries())) {
+    if (!newMap.has(path)) {
+      ops.push({ action: 'delete', path, cid: null, prev: prevCid });
+    }
+  }
+
+  ops.sort((a, b) => a.path.localeCompare(b.path));
+  return ops;
+}
+
+// Used by the legacy in-memory diff path (kept for compatibility). The batched
+// version above is preferred because it avoids per-node round-trips to D1.
+export async function collectLeavesRecursive(
+  entries: ReadonlyArray<{ isLeaf(): boolean; key?: string; value?: CID; getEntries?: () => Promise<unknown[]> }>,
+  map: Map<string, CID>,
+): Promise<void> {
+  for (const entry of entries) {
+    if (entry.isLeaf() && entry.key && entry.value) {
+      map.set(entry.key, entry.value);
+    } else if (entry.getEntries) {
+      const subEntries = (await entry.getEntries()) as typeof entries;
+      await collectLeavesRecursive(subEntries, map);
+    }
+  }
+}