@cloudbitmaps/core 0.1.0-rc.0

package/dist/mongodb/index.cjs

@@ -0,0 +1,260 @@
+'use strict';
+
+var crypto = require('crypto');
+
+// src/drivers/mongodb/warm.ts
+
+// src/core/errors.ts
+var ERROR_BRAND = /* @__PURE__ */ Symbol.for("cloud-roaring.error");
+var TRANSIENT_BRAND = /* @__PURE__ */ Symbol.for("cloud-roaring.error.transient");
+var CloudRoaringError = class extends Error {
+  /** Cross-bundle brand — see the predicates ({@link isCloudRoaringError}, …). Non-enumerable-ish (symbol key ⇒ not in JSON). */
+  [ERROR_BRAND] = true;
+  constructor(message) {
+    super(message);
+    this.name = new.target.name;
+  }
+};
+var ValidationError = class extends CloudRoaringError {
+};
+var WriteConflictError = class extends CloudRoaringError {
+};
+var IntegrityError = class extends CloudRoaringError {
+};
+var TransientError = class extends CloudRoaringError {
+  /** A second brand so the whole transient subtree (incl. {@link TimeoutError}) is classifiable cross-bundle. */
+  [TRANSIENT_BRAND] = true;
+  constructor(message, options) {
+    super(message);
+    if (options && "cause" in options) this.cause = options.cause;
+  }
+};
+
+// src/core/ports.ts
+var NO_ROW = /* @__PURE__ */ Symbol.for("cloud-roaring.no-row");
+
+// src/core/validate.ts
+var CHUNK_KEY_MAX = 65535;
+var NAME = /^[A-Za-z0-9][A-Za-z0-9._-]{0,255}$/;
+function validatePart(value, field) {
+  if (typeof value !== "string" || !NAME.test(value) || value.includes("..")) {
+    throw new ValidationError(
+      `${field} must match ${String(NAME)} and contain no "..": got ${JSON.stringify(value)}`
+    );
+  }
+}
+function validateSegmentRef(ref) {
+  validatePart(ref.segment, "segment");
+  if (ref.namespace !== void 0) validatePart(ref.namespace, "namespace");
+}
+function validateChunkRef(ref) {
+  validateSegmentRef(ref);
+  if (!Number.isInteger(ref.chunkKey) || ref.chunkKey < 0 || ref.chunkKey > CHUNK_KEY_MAX) {
+    throw new ValidationError(
+      `chunkKey must be an integer in [0, ${CHUNK_KEY_MAX}]; got ${ref.chunkKey}`
+    );
+  }
+}
+
+// src/drivers/_shared/keys.ts
+var DEFAULT_NAMESPACE = "_default";
+function namespacePart(namespace) {
+  return namespace ?? DEFAULT_NAMESPACE;
+}
+
+// src/drivers/mongodb/keys.ts
+function normalizeKeyPrefix(prefix) {
+  if (prefix === void 0 || prefix === "") return "";
+  for (const ch of prefix) {
+    if (ch.charCodeAt(0) < 32 || ch === "|") {
+      throw new ValidationError('keyPrefix must not contain "|" or control characters');
+    }
+  }
+  return prefix;
+}
+function segmentScope(prefix, ref) {
+  validateSegmentRef(ref);
+  return { kp: prefix, ns: namespacePart(ref.namespace), seg: ref.segment };
+}
+function chunkDocId(prefix, ref) {
+  validateChunkRef(ref);
+  return `${prefix}|${namespacePart(ref.namespace)}|${ref.segment}|${ref.chunkKey}`;
+}
+
+// src/drivers/mongodb/mongodb-errors.ts
+function serverCode(err) {
+  const code = err?.code;
+  return typeof code === "number" ? code : void 0;
+}
+function networkCode(err) {
+  const code = err?.code;
+  return typeof code === "string" && /^E[A-Z_]+$/.test(code) ? code : void 0;
+}
+function isDuplicateKey(err) {
+  const code = serverCode(err);
+  return code === 11e3 || code === 11001;
+}
+var RETRYABLE_CODES = /* @__PURE__ */ new Set([
+  6,
+  // HostUnreachable
+  7,
+  // HostNotFound
+  89,
+  // NetworkTimeout
+  91,
+  // ShutdownInProgress
+  189,
+  // PrimarySteppedDown
+  262,
+  // ExceededTimeLimit
+  9001,
+  // SocketException
+  10107,
+  // NotWritablePrimary
+  11600,
+  // InterruptedAtShutdown
+  11602,
+  // InterruptedDueToReplStateChange
+  13435,
+  // NotPrimaryNoSecondaryOk
+  13436,
+  // NotPrimaryOrSecondary
+  63,
+  // StaleShardVersion
+  150
+  // StaleEpoch
+]);
+function isTransient(err) {
+  if (isDuplicateKey(err)) return false;
+  const name = err?.name;
+  if (name === "MongoNetworkError" || name === "MongoServerSelectionError") return true;
+  const hasLabel = err?.hasErrorLabel;
+  if (typeof hasLabel === "function") {
+    if (hasLabel.call(err, "RetryableWriteError") || hasLabel.call(err, "TransientTransactionError")) {
+      return true;
+    }
+  }
+  const code = serverCode(err);
+  if (code !== void 0 && RETRYABLE_CODES.has(code)) return true;
+  const net = networkCode(err);
+  return net === "ECONNRESET" || net === "ECONNABORTED" || net === "ETIMEDOUT" || net === "ESOCKETTIMEDOUT" || net === "ECONNREFUSED" || net === "EHOSTUNREACH" || net === "ENETUNREACH" || net === "EPIPE" || net === "EAI_AGAIN" || net === "ENOTFOUND";
+}
+
+// src/drivers/mongodb/warm.ts
+var DEFAULT_COLLECTION = "cloud_roaring_warm";
+var INDEX_NAME = "cr_warm_scope_ck";
+var DEFAULT_BATCH = 500;
+var SIMPLE_COLLATION = { locale: "simple" };
+async function ensureMongoWarmIndexes(db, collection = DEFAULT_COLLECTION) {
+  await db.collection(collection).createIndex(
+    { kp: 1, ns: 1, seg: 1, ck: 1 },
+    { name: INDEX_NAME, collation: SIMPLE_COLLATION }
+  );
+}
+var MongoWarmDriver = class {
+  col;
+  keyPrefix;
+  batch;
+  constructor(options) {
+    this.keyPrefix = normalizeKeyPrefix(options.keyPrefix);
+    const batch = options.listPageSize ?? DEFAULT_BATCH;
+    if (!Number.isSafeInteger(batch) || batch < 1) {
+      throw new ValidationError(`listPageSize must be a positive safe integer; got ${batch}`);
+    }
+    this.batch = batch;
+    this.col = options.db.collection(options.collection ?? DEFAULT_COLLECTION);
+  }
+  // Reads are strongly consistent when the Db targets the primary (see the `db` option), so the optional
+  // `WarmReadOptions` hint would be a no-op — the structurally-optional param is simply omitted, as in the
+  // LocalFs / Postgres / Redis drivers.
+  async get(ref) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    let doc;
+    try {
+      doc = await this.col.findOne({ _id }, { collation: SIMPLE_COLLATION });
+    } catch (err) {
+      throw this.mapError(err);
+    }
+    return doc === null ? null : this.rowFrom(doc, ref.chunkKey);
+  }
+  async putConditional(ref, bytes, expected) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    const token = crypto.randomUUID();
+    const payload = Buffer.from(bytes);
+    try {
+      if (expected === NO_ROW) {
+        const { kp, ns, seg } = segmentScope(this.keyPrefix, ref);
+        await this.col.insertOne({ _id, kp, ns, seg, ck: ref.chunkKey, token, payload });
+      } else {
+        const res = await this.col.updateOne(
+          { _id, token: expected },
+          { $set: { token, payload } },
+          { collation: SIMPLE_COLLATION }
+        );
+        if (res.matchedCount !== 1) {
+          throw new WriteConflictError(`OCC conflict on chunk ${ref.chunkKey}`);
+        }
+      }
+      return { token };
+    } catch (err) {
+      if (isDuplicateKey(err)) {
+        throw new WriteConflictError(`chunk ${ref.chunkKey} already exists (create-if-absent)`);
+      }
+      throw this.mapError(err);
+    }
+  }
+  async deleteConditional(ref, expected) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    let deletedCount;
+    try {
+      deletedCount = (await this.col.deleteOne({ _id, token: expected }, { collation: SIMPLE_COLLATION })).deletedCount;
+    } catch (err) {
+      throw this.mapError(err);
+    }
+    if (deletedCount !== 1) {
+      throw new WriteConflictError(`fenced-delete conflict on chunk ${ref.chunkKey}`);
+    }
+  }
+  async *listChunks(ref) {
+    const scope = segmentScope(this.keyPrefix, ref);
+    const cursor = this.col.find(scope, { collation: SIMPLE_COLLATION }).sort({ ck: 1 }).batchSize(this.batch);
+    try {
+      for await (const doc of cursor) {
+        if (typeof doc.ck !== "number") {
+          throw new IntegrityError("warm doc is missing its numeric chunk key (ck)");
+        }
+        yield { chunkKey: doc.ck, ...this.rowFrom(doc, doc.ck) };
+      }
+    } catch (err) {
+      throw this.mapError(err);
+    }
+  }
+  /** Build a {@link WarmRow} from a doc; reject one missing/!typed fields (corrupt/foreign). */
+  rowFrom(doc, chunkKey) {
+    if (typeof doc.token !== "string" || doc.token === "") {
+      throw new IntegrityError(`warm doc for chunk ${chunkKey} is missing its token`);
+    }
+    return { token: doc.token, bytes: decodePayload(doc.payload, chunkKey) };
+  }
+  /** Reclassify a transient Mongo fault as a retryable {@link TransientError}; else propagate unchanged. */
+  mapError(err) {
+    if (isTransient(err)) {
+      return new TransientError(
+        `transient MongoDB fault: ${err?.message ?? "unknown"}`,
+        { cause: err }
+      );
+    }
+    return err;
+  }
+};
+function decodePayload(p, chunkKey) {
+  if (p instanceof Uint8Array) return new Uint8Array(p);
+  const inner = p?.buffer;
+  if (inner instanceof Uint8Array) return new Uint8Array(inner);
+  throw new IntegrityError(`warm doc for chunk ${chunkKey} has no readable payload`);
+}
+
+exports.MongoWarmDriver = MongoWarmDriver;
+exports.ensureMongoWarmIndexes = ensureMongoWarmIndexes;
+//# sourceMappingURL=index.cjs.map
+//# sourceMappingURL=index.cjs.map

package/dist/mongodb/index.cjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/core/errors.ts","../../src/core/ports.ts","../../src/core/validate.ts","../../src/drivers/_shared/keys.ts","../../src/drivers/mongodb/keys.ts","../../src/drivers/mongodb/mongodb-errors.ts","../../src/drivers/mongodb/warm.ts"],"names":["randomUUID"],"mappings":";;;;;;;AAcA,IAAM,WAAA,mBAA6B,MAAA,CAAO,GAAA,CAAI,qBAAqB,CAAA;AACnE,IAAM,eAAA,mBAAiC,MAAA,CAAO,GAAA,CAAI,+BAA+B,CAAA;AAG1E,IAAM,iBAAA,GAAN,cAAgC,KAAA,CAAM;AAAA;AAAA,EAE3C,CAAU,WAAW,IAAI,IAAA;AAAA,EACzB,YAAY,OAAA,EAAiB;AAC3B,IAAA,KAAA,CAAM,OAAO,CAAA;AAGb,IAAA,IAAA,CAAK,OAAO,GAAA,CAAA,MAAA,CAAW,IAAA;AAAA,EACzB;AACF,CAAA;AAGO,IAAM,eAAA,GAAN,cAA8B,iBAAA,CAAkB;AAAC,CAAA;AAGjD,IAAM,kBAAA,GAAN,cAAiC,iBAAA,CAAkB;AAAC,CAAA;AAGpD,IAAM,cAAA,GAAN,cAA6B,iBAAA,CAAkB;AAAC,CAAA;AAwDhD,IAAM,cAAA,GAAN,cAA6B,iBAAA,CAAkB;AAAA;AAAA,EAEpD,CAAU,eAAe,IAAI,IAAA;AAAA,EAC7B,WAAA,CAAY,SAAiB,OAAA,EAA+B;AAC1D,IAAA,KAAA,CAAM,OAAO,CAAA;AACb,IAAA,IAAI,OAAA,IAAW,OAAA,IAAW,OAAA,EAAS,IAAA,CAAK,QAAQ,OAAA,CAAQ,KAAA;AAAA,EAC1D;AACF,CAAA;;;AC9EO,IAAM,MAAA,mBAAwB,MAAA,CAAO,GAAA,CAAI,sBAAsB,CAAA;;;ACbtE,IAAM,aAAA,GAAgB,KAAA;AAGtB,IAAM,IAAA,GAAO,oCAAA;AAEb,SAAS,YAAA,CAAa,OAAe,KAAA,EAAqB;AACxD,EAAA,IAAI,OAAO,KAAA,KAAU,QAAA,IAAY,CAAC,IAAA,CAAK,IAAA,CAAK,KAAK,CAAA,IAAK,KAAA,CAAM,QAAA,CAAS,IAAI,CAAA,EAAG;AAC1E,IAAA,MAAM,IAAI,eAAA;AAAA,MACR,CAAA,EAAG,KAAK,CAAA,YAAA,EAAe,MAAA,CAAO,IAAI,CAAC,CAAA,0BAAA,EAA6B,IAAA,CAAK,SAAA,CAAU,KAAK,CAAC,CAAA;AAAA,KACvF;AAAA,EACF;AACF;AAEO,SAAS,mBAAmB,GAAA,EAAuB;AACxD,EAAA,YAAA,CAAa,GAAA,CAAI,SAAS,SAAS,CAAA;AACnC,EAAA,IAAI,IAAI,SAAA,KAAc,MAAA,EAAW,YAAA,CAAa,GAAA,CAAI,WAAW,WAAW,CAAA;AAC1E;AAGO,SAAS,iBAAiB,GAAA,EAAqB;AACpD,EAAA,kBAAA,CAAmB,GAAG,CAAA;AACtB,EAAA,IAAI,CAAC,MAAA,CAAO,SAAA,CAAU,GAAA,CAAI,QAAQ,CAAA,IAAK,GAAA,CAAI,QAAA,GAAW,CAAA,IAAK,GAAA,CAAI,QAAA,GAAW,aAAA,EAAe;AACvF,IAAA,MAAM,IAAI,eAAA;AAAA,MACR,CAAA,mCAAA,EAAsC,aAAa,CAAA,OAAA,EAAU,GAAA,CAAI,QAAQ,CAAA;AAAA,KAC3E;AAAA,EACF;AACF;;;ACpBO,IAAM,iBAAA,GAAoB,UAAA;AAG1B,SAAS,cAAc,SAAA,EAAuC;AACnE,EAAA,OAAO,SAAA,IAAa,iBAAA;AACtB;;;ACQO,SAAS,mBAAmB,MAAA,EAAoC;AACrE,EAAA,IAAI,MAAA,KAAW,MAAA,IAAa,MAAA,KAAW,EAAA,EAAI,OAAO,EAAA;AAClD,EAAA,KAAA,MAAW,MAAM,MAAA,EAAQ;AACvB,IAAA,IAAI,GAAG,UAAA,CAAW,CAAC,CAAA,GAAI,EAAA,IAAQ,OAAO,GAAA,EAAK;AACzC,MAAA,MAAM,IAAI,gBAAgB,sDAAsD,CAAA;AAAA,IAClF;AAAA,EACF;AACA,EAAA,OAAO,MAAA;AACT;AAGO,SAAS,YAAA,CAAa,QAAgB,GAAA,EAA+B;AAC1E,EAAA,kBAAA,CAAmB,GAAG,CAAA;AACtB,EAAA,OAAO,EAAE,EAAA,EAAI,MAAA,EAAQ,EAAA,EAAI,aAAA,CAAc,IAAI,SAAS,CAAA,EAAG,GAAA,EAAK,GAAA,CAAI,OAAA,EAAQ;AAC1E;AAGO,SAAS,UAAA,CAAW,QAAgB,GAAA,EAAuB;AAChE,EAAA,gBAAA,CAAiB,GAAG,CAAA;AACpB,EAAA,OAAO,CAAA,EAAG,MAAM,CAAA,CAAA,EAAI,aAAA,CAAc,GAAA,CAAI,SAAS,CAAC,CAAA,CAAA,EAAI,GAAA,CAAI,OAAO,CAAA,CAAA,EAAI,GAAA,CAAI,QAAQ,CAAA,CAAA;AACjF;;;ACrCA,SAAS,WAAW,GAAA,EAAkC;AACpD,EAAA,MAAM,OAAQ,GAAA,EAAmC,IAAA;AACjD,EAAA,OAAO,OAAO,IAAA,KAAS,QAAA,GAAW,IAAA,GAAO,MAAA;AAC3C;AAGA,SAAS,YAAY,GAAA,EAAkC;AACrD,EAAA,MAAM,OAAQ,GAAA,EAAmC,IAAA;AACjD,EAAA,OAAO,OAAO,IAAA,KAAS,QAAA,IAAY,aAAa,IAAA,CAAK,IAAI,IAAI,IAAA,GAAO,MAAA;AACtE;AAGO,SAAS,eAAe,GAAA,EAAuB;AACpD,EAAA,MAAM,IAAA,GAAO,WAAW,GAAG,CAAA;AAC3B,EAAA,OAAO,IAAA,KAAS,QAAS,IAAA,KAAS,KAAA;AACpC;AAOA,IAAM,eAAA,uBAAsB,GAAA,CAAY;AAAA,EACtC,CAAA;AAAA;AAAA,EACA,CAAA;AAAA;AAAA,EACA,EAAA;AAAA;AAAA,EACA,EAAA;AAAA;AAAA,EACA,GAAA;AAAA;AAAA,EACA,GAAA;AAAA;AAAA,EACA,IAAA;AAAA;AAAA,EACA,KAAA;AAAA;AAAA,EACA,KAAA;AAAA;AAAA,EACA,KAAA;AAAA;AAAA,EACA,KAAA;AAAA;AAAA,EACA,KAAA;AAAA;AAAA,EACA,EAAA;AAAA;AAAA,EACA;AAAA;AACF,CAAC,CAAA;AAOM,SAAS,YAAY,GAAA,EAAuB;AACjD,EAAA,IAAI,cAAA,CAAe,GAAG,CAAA,EAAG,OAAO,KAAA;AAChC,EAAA,MAAM,OAAQ,GAAA,EAAmC,IAAA;AACjD,EAAA,IAAI,IAAA,KAAS,mBAAA,IAAuB,IAAA,KAAS,2BAAA,EAA6B,OAAO,IAAA;AACjF,EAAA,MAAM,WAAY,GAAA,EAA2D,aAAA;AAC7E,EAAA,IAAI,OAAO,aAAa,UAAA,EAAY;AAClC,IAAA,IACE,QAAA,CAAS,KAAK,GAAA,EAAK,qBAAqB,KACxC,QAAA,CAAS,IAAA,CAAK,GAAA,EAAK,2BAA2B,CAAA,EAC9C;AACA,MAAA,OAAO,IAAA;AAAA,IACT;AAAA,EACF;AACA,EAAA,MAAM,IAAA,GAAO,WAAW,GAAG,CAAA;AAC3B,EAAA,IAAI,SAAS,MAAA,IAAa,eAAA,CAAgB,GAAA,CAAI,IAAI,GAAG,OAAO,IAAA;AAC5D,EAAA,MAAM,GAAA,GAAM,YAAY,GAAG,CAAA;AAC3B,EAAA,OACE,QAAQ,YAAA,IACR,GAAA,KAAQ,kBACR,GAAA,KAAQ,WAAA,IACR,QAAQ,iBAAA,IACR,GAAA,KAAQ,cAAA,IACR,GAAA,KAAQ,kBACR,GAAA,KAAQ,aAAA,IACR,QAAQ,OAAA,IACR,GAAA,KAAQ,eACR,GAAA,KAAQ,WAAA;AAEZ;;;ACvDA,IAAM,kBAAA,GAAqB,oBAAA;AAC3B,IAAM,UAAA,GAAa,kBAAA;AAEnB,IAAM,aAAA,GAAgB,GAAA;AAatB,IAAM,gBAAA,GAAmB,EAAE,MAAA,EAAQ,QAAA,EAAS;AAmC5C,eAAsB,sBAAA,CACpB,EAAA,EACA,UAAA,GAAqB,kBAAA,EACN;AACf,EAAA,MAAM,EAAA,CACH,UAAA,CAAW,UAAU,CAAA,CACrB,WAAA;AAAA,IACC,EAAE,IAAI,CAAA,EAAG,EAAA,EAAI,GAAG,GAAA,EAAK,CAAA,EAAG,IAAI,CAAA,EAAE;AAAA,IAC9B,EAAE,IAAA,EAAM,UAAA,EAAY,SAAA,EAAW,gBAAA;AAAiB,GAClD;AACJ;AAEO,IAAM,kBAAN,MAA6C;AAAA,EACjC,GAAA;AAAA,EACA,SAAA;AAAA,EACA,KAAA;AAAA,EAEjB,YAAY,OAAA,EAAiC;AAC3C,IAAA,IAAA,CAAK,SAAA,GAAY,kBAAA,CAAmB,OAAA,CAAQ,SAAS,CAAA;AACrD,IAAA,MAAM,KAAA,GAAQ,QAAQ,YAAA,IAAgB,aAAA;AACtC,IAAA,IAAI,CAAC,MAAA,CAAO,aAAA,CAAc,KAAK,CAAA,IAAK,QAAQ,CAAA,EAAG;AAC7C,MAAA,MAAM,IAAI,eAAA,CAAgB,CAAA,kDAAA,EAAqD,KAAK,CAAA,CAAE,CAAA;AAAA,IACxF;AACA,IAAA,IAAA,CAAK,KAAA,GAAQ,KAAA;AACb,IAAA,IAAA,CAAK,MAAM,OAAA,CAAQ,EAAA,CAAG,UAAA,CAAoB,OAAA,CAAQ,cAAc,kBAAkB,CAAA;AAAA,EACpF;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,IAAI,GAAA,EAAwC;AAChD,IAAA,MAAM,GAAA,GAAM,UAAA,CAAW,IAAA,CAAK,SAAA,EAAW,GAAG,CAAA;AAC1C,IAAA,IAAI,GAAA;AACJ,IAAA,IAAI;AACF,MAAA,GAAA,GAAM,MAAM,IAAA,CAAK,GAAA,CAAI,OAAA,CAAQ,EAAE,KAAI,EAAG,EAAE,SAAA,EAAW,gBAAA,EAAkB,CAAA;AAAA,IACvE,SAAS,GAAA,EAAK;AACZ,MAAA,MAAM,IAAA,CAAK,SAAS,GAAG,CAAA;AAAA,IACzB;AACA,IAAA,OAAO,QAAQ,IAAA,GAAO,IAAA,GAAO,KAAK,OAAA,CAAQ,GAAA,EAAK,IAAI,QAAQ,CAAA;AAAA,EAC7D;AAAA,EAEA,MAAM,cAAA,CACJ,GAAA,EACA,KAAA,EACA,QAAA,EAC2B;AAC3B,IAAA,MAAM,GAAA,GAAM,UAAA,CAAW,IAAA,CAAK,SAAA,EAAW,GAAG,CAAA;AAC1C,IAAA,MAAM,QAAQA,iBAAA,EAAW;AACzB,IAAA,MAAM,OAAA,GAAU,MAAA,CAAO,IAAA,CAAK,KAAK,CAAA;AACjC,IAAA,IAAI;AACF,MAAA,IAAI,aAAa,MAAA,EAAQ;AACvB,QAAA,MAAM,EAAE,IAAI,EAAA,EAAI,GAAA,KAAQ,YAAA,CAAa,IAAA,CAAK,WAAW,GAAG,CAAA;AACxD,QAAA,MAAM,IAAA,CAAK,GAAA,CAAI,SAAA,CAAU,EAAE,GAAA,EAAK,EAAA,EAAI,EAAA,EAAI,GAAA,EAAK,EAAA,EAAI,GAAA,CAAI,QAAA,EAAU,KAAA,EAAO,SAAS,CAAA;AAAA,MACjF,CAAA,MAAO;AACL,QAAA,MAAM,GAAA,GAAM,MAAM,IAAA,CAAK,GAAA,CAAI,SAAA;AAAA,UACzB,EAAE,GAAA,EAAK,KAAA,EAAO,QAAA,EAAS;AAAA,UACvB,EAAE,IAAA,EAAM,EAAE,KAAA,EAAO,SAAQ,EAAE;AAAA,UAC3B,EAAE,WAAW,gBAAA;AAAiB,SAChC;AACA,QAAA,IAAI,GAAA,CAAI,iBAAiB,CAAA,EAAG;AAC1B,UAAA,MAAM,IAAI,kBAAA,CAAmB,CAAA,sBAAA,EAAyB,GAAA,CAAI,QAAQ,CAAA,CAAE,CAAA;AAAA,QACtE;AAAA,MACF;AACA,MAAA,OAAO,EAAE,KAAA,EAAM;AAAA,IACjB,SAAS,GAAA,EAAK;AAEZ,MAAA,IAAI,cAAA,CAAe,GAAG,CAAA,EAAG;AACvB,QAAA,MAAM,IAAI,kBAAA,CAAmB,CAAA,MAAA,EAAS,GAAA,CAAI,QAAQ,CAAA,kCAAA,CAAoC,CAAA;AAAA,MACxF;AACA,MAAA,MAAM,IAAA,CAAK,SAAS,GAAG,CAAA;AAAA,IACzB;AAAA,EACF;AAAA,EAEA,MAAM,iBAAA,CAAkB,GAAA,EAAe,QAAA,EAAgC;AACrE,IAAA,MAAM,GAAA,GAAM,UAAA,CAAW,IAAA,CAAK,SAAA,EAAW,GAAG,CAAA;AAC1C,IAAA,IAAI,YAAA;AACJ,IAAA,IAAI;AACF,MAAA,YAAA,GAAA,CACE,MAAM,IAAA,CAAK,GAAA,CAAI,SAAA,CAAU,EAAE,GAAA,EAAK,KAAA,EAAO,QAAA,EAAS,EAAG,EAAE,SAAA,EAAW,gBAAA,EAAkB,CAAA,EAClF,YAAA;AAAA,IACJ,SAAS,GAAA,EAAK;AACZ,MAAA,MAAM,IAAA,CAAK,SAAS,GAAG,CAAA;AAAA,IACzB;AACA,IAAA,IAAI,iBAAiB,CAAA,EAAG;AACtB,MAAA,MAAM,IAAI,kBAAA,CAAmB,CAAA,gCAAA,EAAmC,GAAA,CAAI,QAAQ,CAAA,CAAE,CAAA;AAAA,IAChF;AAAA,EACF;AAAA,EAEA,OAAO,WAAW,GAAA,EAAgE;AAChF,IAAA,MAAM,KAAA,GAAQ,YAAA,CAAa,IAAA,CAAK,SAAA,EAAW,GAAG,CAAA;AAK9C,IAAA,MAAM,SAAS,IAAA,CAAK,GAAA,CACjB,KAAK,KAAA,EAAO,EAAE,WAAW,gBAAA,EAAkB,CAAA,CAC3C,IAAA,CAAK,EAAE,EAAA,EAAI,CAAA,EAAG,CAAA,CACd,SAAA,CAAU,KAAK,KAAK,CAAA;AACvB,IAAA,IAAI;AACF,MAAA,WAAA,MAAiB,OAAO,MAAA,EAAQ;AAG9B,QAAA,IAAI,OAAO,GAAA,CAAI,EAAA,KAAO,QAAA,EAAU;AAC9B,UAAA,MAAM,IAAI,eAAe,gDAAgD,CAAA;AAAA,QAC3E;AACA,QAAA,MAAM,EAAE,QAAA,EAAU,GAAA,CAAI,EAAA,EAAI,GAAG,KAAK,OAAA,CAAQ,GAAA,EAAK,GAAA,CAAI,EAAE,CAAA,EAAE;AAAA,MACzD;AAAA,IACF,SAAS,GAAA,EAAK;AACZ,MAAA,MAAM,IAAA,CAAK,SAAS,GAAG,CAAA;AAAA,IACzB;AAAA,EACF;AAAA;AAAA,EAGQ,OAAA,CAAQ,KAAc,QAAA,EAA2B;AACvD,IAAA,IAAI,OAAO,GAAA,CAAI,KAAA,KAAU,QAAA,IAAY,GAAA,CAAI,UAAU,EAAA,EAAI;AACrD,MAAA,MAAM,IAAI,cAAA,CAAe,CAAA,mBAAA,EAAsB,QAAQ,CAAA,qBAAA,CAAuB,CAAA;AAAA,IAChF;AACA,IAAA,OAAO,EAAE,OAAO,GAAA,CAAI,KAAA,EAAO,OAAO,aAAA,CAAc,GAAA,CAAI,OAAA,EAAS,QAAQ,CAAA,EAAE;AAAA,EACzE;AAAA;AAAA,EAGQ,SAAS,GAAA,EAAuB;AACtC,IAAA,IAAI,WAAA,CAAY,GAAG,CAAA,EAAG;AACpB,MAAA,OAAO,IAAI,cAAA;AAAA,QACT,CAAA,yBAAA,EAA6B,GAAA,EAAsC,OAAA,IAAW,SAAS,CAAA,CAAA;AAAA,QACvF,EAAE,OAAO,GAAA;AAAI,OACf;AAAA,IACF;AACA,IAAA,OAAO,GAAA;AAAA,EACT;AACF;AAOA,SAAS,aAAA,CAAc,GAAY,QAAA,EAA8B;AAC/D,EAAA,IAAI,CAAA,YAAa,UAAA,EAAY,OAAO,IAAI,WAAW,CAAC,CAAA;AACpD,EAAA,MAAM,QAAS,CAAA,EAAmC,MAAA;AAClD,EAAA,IAAI,KAAA,YAAiB,UAAA,EAAY,OAAO,IAAI,WAAW,KAAK,CAAA;AAC5D,EAAA,MAAM,IAAI,cAAA,CAAe,CAAA,mBAAA,EAAsB,QAAQ,CAAA,wBAAA,CAA0B,CAAA;AACnF","file":"index.cjs","sourcesContent":["/**\n * Typed errors — callers learn *why* something failed, never by parsing strings. Retry is the driver\n * decorators' job, not the engine's: a driver classifies its backend's failures into this vocabulary, and\n * `withRetry` decides what is transient.\n */\n\n/**\n * Registry-symbol brands. The package ships as multiple bundles — the core entry and the `./s3` / `./dynamodb`\n * subpaths — and the builder inlines `core/errors` into each. A driver in a subpath bundle therefore throws a\n * *different* class object than the one the core engine/retry code would `instanceof`-check, so `instanceof`\n * silently returns false across that boundary in the published CJS package (defeating OCC/transient retry and\n * compaction race-handling). These `Symbol.for` brands are identity-stable across bundles/realms; classify\n * errors with the exported predicates below (never `instanceof`) anywhere an error may cross the boundary.\n */\nconst ERROR_BRAND: unique symbol = Symbol.for('cloud-roaring.error');\nconst TRANSIENT_BRAND: unique symbol = Symbol.for('cloud-roaring.error.transient');\n\n/** Base class for every error CloudRoaring throws. */\nexport class CloudRoaringError extends Error {\n  /** Cross-bundle brand — see the predicates ({@link isCloudRoaringError}, …). Non-enumerable-ish (symbol key ⇒ not in JSON). */\n  readonly [ERROR_BRAND] = true as const;\n  constructor(message: string) {\n    super(message);\n    // Subclass name (works under transpilation since we set it explicitly). Also the discriminator the\n    // predicates match on — a runtime string, so it survives bundling where the class identity does not.\n    this.name = new.target.name;\n  }\n}\n\n/** Invalid caller input (bad id, segment name, options). Raised before any storage call. */\nexport class ValidationError extends CloudRoaringError {}\n\n/** An OCC conditional write/delete lost the race — the row changed since it was read. */\nexport class WriteConflictError extends CloudRoaringError {}\n\n/** Bytes from a tier are corrupt, oversized, or fail a checksum/format check. */\nexport class IntegrityError extends CloudRoaringError {}\n\n/**\n * A requested object/row does not exist. Part of the driver error vocabulary; thrown by\n * persistent drivers from Phase 2 — the Phase-1 engine + in-memory drivers return `null` instead.\n */\nexport class NotFoundError extends CloudRoaringError {}\n\n/**\n * This build/configuration cannot perform the requested operation, though nothing is malformed. Two uses:\n * (1) **format** — the bytes are well-formed but unreadable here (an unknown `.crbm` major version, an\n * encrypted file before the crypto path exists) — distinct from `IntegrityError` (corruption); and (2)\n * **store configuration** — an operation this store's wiring doesn't support (e.g. a lifecycle helper like\n * `compact`/`eraseSubject` called on a store built without a raw cold driver + registry). Raised at\n * operation time, before any mutation.\n */\nexport class UnsupportedError extends CloudRoaringError {}\n\n/**\n * A driver cannot meet a capability the chosen topology requires (e.g. a Cold driver without range\n * reads). Raised fail-fast at wiring time, never mid-operation.\n */\nexport class CapabilityError extends CloudRoaringError {}\n\n/**\n * An operation would exceed its per-op **denial-of-wallet budget** — too many backend requests for a single\n * `count`/`iterate`/`intersect`/`subjectReport`/`eraseSubject` call — so it is refused **before** fanning out\n * (Decision #3 / invariant T3). Default-on but generous (normal ops never hit it); tune it\n * per store (`budget`) or per op, or disable with `budget: false`. Deterministic (never retried): the op is too\n * big by policy, not by luck. Each request's bytes are separately capped (the safe-deserialize ceiling), so\n * bounding the request count transitively bounds bytes. Carries the projected count + the limit, never data.\n */\nexport class BudgetExceededError extends CloudRoaringError {}\n\n/**\n * An encrypted segment's data key (DEK) cannot be unwrapped because the keystore holds none of the\n * key-encryption-keys (KEKs) its wrappings reference — the KEK was never configured, rotated away without\n * keeping the old key, or lost. Deterministic (never retried): without a KEK the ciphertext is unreadable by\n * design. The flip side of crypto-shred — when this is *intended* (a destroyed segment) the registry row is\n * already a `destroyed` tombstone; when it's *not*, restore the missing KEK (or its recovery KEK). Carries no\n * key material.\n */\nexport class KeyUnavailableError extends CloudRoaringError {}\n\n/**\n * A **transient** infrastructure fault that is safe to retry — throttling, a 5xx, a dropped connection,\n * a client-side request timeout. Drivers classify their backend's retryable faults and raise this (the\n * SDK-specific knowledge stays in the SDK-specific driver); the retry layer (`core/retry`) retries **only**\n * this class, never a deterministic error like {@link ValidationError}, {@link IntegrityError},\n * {@link NotFoundError}, or {@link WriteConflictError} (retrying those is pointless or wrong). The original\n * error is preserved in `cause` so callers can still inspect it.\n *\n * Note for logging hygiene (threat-model S12): `cause` is the **raw SDK error**, which may carry operational\n * metadata (endpoint host, request IDs, `$metadata`). The library's own `message` is identifier-only and safe\n * to log; if you serialize the whole error *chain*, be aware you're including that metadata.\n */\nexport class TransientError extends CloudRoaringError {\n  /** A second brand so the whole transient subtree (incl. {@link TimeoutError}) is classifiable cross-bundle. */\n  readonly [TRANSIENT_BRAND] = true as const;\n  constructor(message: string, options?: { cause?: unknown }) {\n    super(message);\n    if (options && 'cause' in options) this.cause = options.cause;\n  }\n}\n\n/**\n * A single attempt exceeded its time budget. Subclass of {@link TransientError} so the retry layer treats a\n * timeout as retryable by default — a stalled request often succeeds on a fresh connection. Raised by a\n * driver whose injected client reports a request timeout — setting one on your injected client is the\n * recommended way to bound a hang.\n */\nexport class TimeoutError extends TransientError {}\n\n/**\n * Bundle-safe error predicates — use these, not `instanceof`, wherever an error may cross the core↔driver\n * (`./s3` / `./dynamodb`) boundary (and prefer them in consumer `catch` blocks too, for the same reason). They\n * match the {@link ERROR_BRAND} registry brand + the runtime `name`, both of which survive separate bundling.\n */\nfunction hasBrand(err: unknown, brand: symbol): boolean {\n  return (\n    typeof err === 'object' && err !== null && (err as Record<symbol, unknown>)[brand] === true\n  );\n}\n\n/** Any error thrown by CloudRoaring (any tier, any bundle). */\nexport function isCloudRoaringError(err: unknown): err is CloudRoaringError {\n  return hasBrand(err, ERROR_BRAND);\n}\n\n/** An OCC conditional write/delete lost the race — retry the read-modify-write, don't fail. */\nexport function isWriteConflictError(err: unknown): err is WriteConflictError {\n  return isCloudRoaringError(err) && err.name === 'WriteConflictError';\n}\n\n/** A retryable transient infrastructure fault (incl. {@link TimeoutError}). The retry layer keys on this. */\nexport function isTransientError(err: unknown): err is TransientError {\n  return hasBrand(err, TRANSIENT_BRAND);\n}\n\n/** A requested object/row does not exist. */\nexport function isNotFoundError(err: unknown): err is NotFoundError {\n  return isCloudRoaringError(err) && err.name === 'NotFoundError';\n}\n\n/** Corrupt/oversized/failed-checksum bytes from a tier. */\nexport function isIntegrityError(err: unknown): err is IntegrityError {\n  return isCloudRoaringError(err) && err.name === 'IntegrityError';\n}\n\n/** Invalid caller input. */\nexport function isValidationError(err: unknown): err is ValidationError {\n  return isCloudRoaringError(err) && err.name === 'ValidationError';\n}\n","/**\n * Storage-driver contracts the engine depends on.\n *\n * Phase 1 uses the subset needed by the in-memory engine: the per-chunk Warm store (under OCC) and a\n * per-chunk read view of Cold. Drivers move opaque bytes + an OCC token; they never understand roaring,\n * `adds`/`removes`, or the `.crbm` layout. The full `IColdDriver` (blob/range) + `IRegistryDriver` arrive\n * with generations in Phase 2/3 — the `.crbm` reader will implement `ColdChunkSource`.\n */\n\nimport type { BlobSink } from './blob';\nimport type { WrappedDek } from './crypto';\n\n/** Opaque optimistic-concurrency token — unique per write, compared by equality only. */\nexport type Token = string;\n\n/** Sentinel for `putConditional` meaning \"the row must not exist yet\" (create). */\n// `Symbol.for` (the global registry), NOT `Symbol()`: the package ships multiple bundles (the core entry and\n// the `./dynamodb` / `./s3` subpaths), and tsup inlines `core/ports` into each. A plain `Symbol('no-row')`\n// would be a DISTINCT instance per bundle, so the engine's `NO_ROW` wouldn't `===` the one the DynamoDb warm\n// driver compares against — every create would misroute to the token-fenced path and fail. A registry symbol\n// is identity-stable across bundles/realms. (Invisible in tests, which share one source module graph.)\nexport const NO_ROW: unique symbol = Symbol.for('cloud-roaring.no-row');\nexport type NoRow = typeof NO_ROW;\n\nexport interface SegmentRef {\n  readonly namespace?: string;\n  readonly segment: string;\n}\n\nexport interface ChunkRef extends SegmentRef {\n  readonly chunkKey: number;\n}\n\n/** Identifies one immutable `.crbm` object — a single generation of a segment. */\nexport interface GenKey extends SegmentRef {\n  readonly generation: number;\n}\n\nexport interface WarmRow {\n  readonly token: Token;\n  /** Read-only: owned by the driver. Callers must NOT mutate the buffer (re-encode to write). */\n  readonly bytes: Uint8Array;\n}\n\n/**\n * Read consistency for a Warm fetch (gap #9). `consistent: true` (the default — omitted ⇒ true) is a\n * strong, read-your-writes read, as the OCC read-modify-write path requires. `consistent: false` requests an\n * **eventually-consistent** read — on DynamoDB that is ~½ the RCU cost; a driver that is always strongly\n * consistent (in-memory, LocalFs) ignores it. The engine uses it only on read paths (`has`/`count`/`iterate`/\n * `intersect`) when the store opts into `warmReadConsistency: 'eventual'`, trading read-after-write for cost.\n */\nexport interface WarmReadOptions {\n  readonly consistent?: boolean;\n}\n\n/** Per-chunk Warm store under optimistic concurrency. Returned byte buffers are read-only. */\nexport interface IWarmDriver {\n  get(ref: ChunkRef, opts?: WarmReadOptions): Promise<WarmRow | null>;\n  /** OCC write. `expected` = `NO_ROW` to create, else the token previously read. Throws `WriteConflictError` on mismatch. */\n  putConditional(\n    ref: ChunkRef,\n    bytes: Uint8Array,\n    expected: Token | NoRow,\n  ): Promise<{ token: Token }>;\n  /**\n   * Fenced delete — only if the stored token still equals `expected`; throws `WriteConflictError`\n   * otherwise. **Used by compaction (Phase 4); the Phase-1 engine never calls it** — drivers still\n   * implement it to satisfy the conformance suite.\n   */\n  deleteConditional(ref: ChunkRef, expected: Token): Promise<void>;\n  /** All dirty chunks of a segment, ascending `chunkKey`. Yielded `bytes` are read-only. */\n  listChunks(\n    ref: SegmentRef,\n    opts?: WarmReadOptions,\n  ): AsyncIterable<{ chunkKey: number } & WarmRow>;\n}\n\n/**\n * A segment's grounded on-disk footprint (Phase 5b) — the current generation's Cold object bytes, read\n * cheaply from the `.crbm` footer/index (no payload reads). Powers the grounded `costReport()`.\n */\nexport interface SegmentSize {\n  readonly sizeBytes: number;\n}\n\n/** Per-chunk read view of the immutable Cold tier (implemented by the `.crbm` reader from Phase 2). */\nexport interface ColdChunkSource {\n  /** Read-only bytes for the chunk, or `null` if absent. Callers must not mutate the buffer. */\n  getChunk(ref: ChunkRef): Promise<Uint8Array | null>;\n  listChunkKeys(ref: SegmentRef): Promise<number[]>;\n  /**\n   * Optional (Phase 5b): the current generation's grounded size, cheaply (from the already-parsed\n   * `.crbm` index — no payload reads), or `null` if the segment has no Cold generation. Powers the\n   * grounded `costReport()`.\n   */\n  sizeOf?(ref: SegmentRef): Promise<SegmentSize | null>;\n  /**\n   * Optional (Phase 5c): the current generation's **per-chunk cardinality** (`chunkKey → count`), read\n   * from the already-parsed `.crbm` index with **no payload reads**, or `null` if the segment has no Cold\n   * generation. Powers the cheap `count()` — a warm-delta-free chunk is counted straight from the index\n   * instead of fetching + deserializing it. A source with no index (e.g. the in-memory source) omits this,\n   * and `count()` falls back to fetching + merging every chunk.\n   */\n  cardinalities?(ref: SegmentRef): Promise<ReadonlyMap<number, number> | null>;\n  /**\n   * Optional (Phase B, gap #4): the segment's **current generation number** as this source resolves it right now\n   * (registry `currentGen`, or the highest cold generation), or `null` if the segment has no Cold generation. The\n   * engine keys its HOT chunk cache by this so a generation bump (a compaction commit) is observed — a new\n   * generation misses the cache instead of serving a stale decoded chunk, and an erased id can't resurrect from a\n   * cached pre-compaction chunk. Cheap: served from the source's own (short-TTL-refreshed) snapshot, **not** a\n   * fresh backend read per call. A source that pins one immutable generation for its whole lifetime and never\n   * refreshes may omit this — the engine then keys the cache without a generation, exactly as before.\n   */\n  currentGeneration?(ref: SegmentRef): Promise<number | null>;\n}\n\n/** Capabilities a Cold driver advertises; validated at wiring time, fail-fast. */\nexport interface ColdCaps {\n  /** REQUIRED — the format relies on byte-range reads. */\n  readonly rangeRead: true;\n  /** Largest single object the backend accepts (informs single-object-vs-shard, B7 — future). */\n  readonly maxObjectBytes: number;\n  /** Optional: enables the pure-object `LATEST`-pointer registry variant (S3 conditional put). */\n  readonly conditionalPut?: boolean;\n}\n\n/**\n * Immutable object storage for `.crbm` generations. A \"dumb byte mover\": it\n * understands neither roaring nor the `.crbm` layout, only opaque bytes addressed by a {@link GenKey}.\n * The core never reuses a key, so puts are write-once.\n */\nexport interface IColdDriver {\n  capabilities(): ColdCaps;\n  /**\n   * Stream a new immutable generation. The driver opens a destination, hands `write` a {@link BlobSink},\n   * then atomically commits (and computes the content hash). Throws if the key already exists (write-once).\n   */\n  putImmutable(\n    key: GenKey,\n    write: (sink: BlobSink) => Promise<void>,\n  ): Promise<{ size: number; sha256: string }>;\n  /** Range read; the caller bounds-checks. Out-of-range is rejected, never a short/adjacent read. */\n  getRange(key: GenKey, offset: number, length: number): Promise<Uint8Array>;\n  /** Speculative tail read: the last `min(maxBytes, size)` bytes + the total object size. */\n  getTail(key: GenKey, maxBytes: number): Promise<{ bytes: Uint8Array; size: number }>;\n  delete(key: GenKey): Promise<void>;\n  /** Enumerate the generations present for a segment (orphan sweep / latest-gen resolution). */\n  list(ref: SegmentRef): AsyncIterable<GenKey>;\n}\n\n/**\n * Lifecycle status of a segment. `active` is the steady state;\n * `compacting`/`erasing` are transient flags a daemon sets while it works (Phase 4d/4e); `destroyed` is the\n * post-crypto-shred tombstone (the row is kept for audit but the segment is logically gone). The 4c registry\n * stores and round-trips the field; the *transitions* are driven by their owning features.\n */\nexport type RegistryStatus = 'active' | 'compacting' | 'erasing' | 'destroyed';\n\n/** Free-form, JSON-serializable governance metadata (retention/residency policy); shape lands in Phase 6. */\nexport type GovernanceMeta = Record<string, unknown>;\n\n/**\n * One registry row — the authoritative per-segment record. Exactly one per segment.\n */\nexport interface RegistryRecord extends SegmentRef {\n  /** **The** authoritative LATEST pointer: which immutable Cold generation is current. */\n  readonly currentGen: number;\n  /**\n   * Per-segment data-key (DEK) wrappings for encryption-at-rest (Phase 4e): the DEK envelope-wrapped under one\n   * or more KEKs (active + optional recovery). Reading unwraps with any held KEK; **crypto-shred deletes this\n   * whole list**, making the segment's at-rest bytes permanently unrecoverable. Absent ⇒ the segment is\n   * cleartext. See [DECISIONS #19].\n   */\n  readonly wrappedDeks?: readonly WrappedDek[];\n  /**\n   * Optional **external** key reference (reserved) — e.g. a KMS key ARN for a future KMS keystore adapter that\n   * keeps wrapped material in the KMS rather than in-band {@link wrappedDeks}. Unused by the in-process BYOK\n   * keystore. Clearable on crypto-shred.\n   */\n  readonly keyId?: string;\n  /**\n   * Dirty-row hint written by `findCompactable`. Discovery still recomputes the live dirty count by scanning\n   * Warm each cycle (the count that decides candidacy), but it now **reads** this hint to gate a change-guarded\n   * CAS — the hint is only rewritten when the count actually moved, so an all-idle fleet issues no per-segment\n   * registry writes. It does not yet drive candidacy — turning the Warm scan itself into O(dirty) is a\n   * deferred cheap-enumeration fix.\n   */\n  readonly dirtyChunkCount: number;\n  readonly status: RegistryStatus;\n  /**\n   * Daemon health (Phase D, gap #2): epoch-ms of the last **successfully committed** compaction, and the\n   * count of **consecutive** compaction failures (reset to 0 by a successful commit). `lastCompactedAt`\n   * powers a dead-man's-switch / staleness alarm; `consecutiveFailures` drives **poison-segment quarantine** —\n   * discovery skips a segment past a failure threshold so one corrupt Warm row can't freeze the compaction of a\n   * segment **that has a committed generation** forever (with no alarm) while its Warm backlog grows unbounded.\n   * (A segment whose *very first* compaction keeps failing has no registry row to count against yet —\n   * bootstrap faults are surfaced as `error` results but not yet quarantined; gap #2.) Both\n   * **optional** (absent on rows written before Phase D, and on a segment never compacted) — read them as\n   * `lastCompactedAt ?? undefined`, `consecutiveFailures ?? 0`.\n   */\n  readonly lastCompactedAt?: number;\n  readonly consecutiveFailures?: number;\n  /**\n   * Compaction lease (Phase 4d): the opaque id of the worker currently compacting this segment, and the\n   * epoch-ms at which its lease expires. `status === 'compacting'` with `leaseExpiresAt` in the future means\n   * a live daemon owns it; once `leaseExpiresAt` is past, another worker may **steal** the lease (the prior\n   * holder crashed). Both absent in the steady (`active`) state. The lease is an efficiency guard only — the\n   * 2-phase commit is correct without it (OCC-fenced swap + write-once generations + version-fenced purge).\n   */\n  readonly leaseOwner?: string;\n  readonly leaseExpiresAt?: number;\n  /** Governance policy (Phase 6); stored + round-tripped now, semantics later. */\n  readonly retention?: GovernanceMeta;\n  readonly residency?: GovernanceMeta;\n  /** Epoch-ms of creation / last mutation (from the driver's injected clock). */\n  readonly createdAt: number;\n  readonly updatedAt: number;\n  /** Opaque OCC token — compare-by-equality, never reused (ABA-safe), exactly like {@link WarmRow.token}. */\n  readonly token: Token;\n}\n\n/** The caller-settable fields at {@link IRegistryDriver.create} (audit + token are driver-managed). */\nexport interface NewRegistryRecord {\n  readonly currentGen: number;\n  readonly wrappedDeks?: readonly WrappedDek[];\n  readonly keyId?: string;\n  /** Defaults to 0. */\n  readonly dirtyChunkCount?: number;\n  /** Defaults to `'active'`. */\n  readonly status?: RegistryStatus;\n  readonly retention?: GovernanceMeta;\n  readonly residency?: GovernanceMeta;\n}\n\n/** Fields a {@link IRegistryDriver.compareAndSwap} may mutate (identity + audit + token are off-limits). */\nexport type RegistryPatch = Partial<\n  Pick<\n    RegistryRecord,\n    | 'currentGen'\n    | 'wrappedDeks'\n    | 'keyId'\n    | 'dirtyChunkCount'\n    | 'status'\n    | 'leaseOwner'\n    | 'leaseExpiresAt'\n    | 'lastCompactedAt'\n    | 'consecutiveFailures'\n    | 'retention'\n    | 'residency'\n  >\n>;\n\n/** Capabilities a registry driver advertises; validated fail-fast at wiring time. */\nexport interface RegCaps {\n  /** REQUIRED — `currentGen` feeds read correctness + compaction CAS, so reads must be strongly consistent. */\n  readonly strongRead: true;\n}\n\n/**\n * Per-segment registry — the authoritative source of `currentGen`, the discovery index, and (Phase 4e) the\n * wrapped-DEK holder. One row per segment under OCC: a\n * never-reused, equality-compared {@link Token} (ABA-safe across delete→recreate, like the Warm tier).\n */\nexport interface IRegistryDriver {\n  capabilities(): RegCaps;\n  /** The segment's record, or `null` if it doesn't exist (or was deleted). */\n  get(ref: SegmentRef): Promise<RegistryRecord | null>;\n  /** Create the row; throws {@link WriteConflictError} if it already exists (use CAS to mutate). */\n  create(ref: SegmentRef, record: NewRegistryRecord): Promise<{ token: Token }>;\n  /** Server-side compare-and-set: apply `patch` iff the stored token equals `expected`, else `WriteConflictError`. */\n  compareAndSwap(ref: SegmentRef, expected: Token, patch: RegistryPatch): Promise<{ token: Token }>;\n  /** Discovery: every live record, optionally scoped to one namespace. Order is unspecified. */\n  list(namespace?: string): AsyncIterable<RegistryRecord>;\n  /** Remove the row (tombstoned for ABA-safety — a later `create` still gets a fresh, greater token). */\n  delete(ref: SegmentRef): Promise<void>;\n}\n","/**\n * Boundary validation for segment / namespace names (finding S2).\n * Names become object keys, partition keys, and file paths — so they're validated before they\n * ever reach storage: strict charset, bounded length, no path traversal.\n */\nimport { ValidationError } from './errors';\nimport type { ChunkRef, SegmentRef } from './ports';\n\nconst CHUNK_KEY_MAX = 0xffff;\n\n// The locked name grammar: 1 leading alphanumeric + up to 255 more (max 256 chars).\nconst NAME = /^[A-Za-z0-9][A-Za-z0-9._-]{0,255}$/;\n\nfunction validatePart(value: string, field: string): void {\n  if (typeof value !== 'string' || !NAME.test(value) || value.includes('..')) {\n    throw new ValidationError(\n      `${field} must match ${String(NAME)} and contain no \"..\": got ${JSON.stringify(value)}`,\n    );\n  }\n}\n\nexport function validateSegmentRef(ref: SegmentRef): void {\n  validatePart(ref.segment, 'segment');\n  if (ref.namespace !== undefined) validatePart(ref.namespace, 'namespace');\n}\n\n/** Validate a chunk ref: the segment/namespace grammar plus `chunkKey` ∈ `[0, 65535]` (a u16). */\nexport function validateChunkRef(ref: ChunkRef): void {\n  validateSegmentRef(ref);\n  if (!Number.isInteger(ref.chunkKey) || ref.chunkKey < 0 || ref.chunkKey > CHUNK_KEY_MAX) {\n    throw new ValidationError(\n      `chunkKey must be an integer in [0, ${CHUNK_KEY_MAX}]; got ${ref.chunkKey}`,\n    );\n  }\n}\n","/**\n * Shared key-grammar fragments used by every driver's logical-ref → physical-key mapping (Phase 4c).\n *\n * Extracted once the registry drivers became the 4th/5th consumer of the same `_default` namespace sentinel\n * (it lived copy-pasted in `s3/keys`, `dynamodb/keys`, `localfs/paths`). Pure string logic, no SDK, no I/O —\n * lives in the SDK-free `_shared` bundle so any driver may import it.\n */\n\n/**\n * The physical stand-in for an **absent** namespace. The name grammar forbids a leading underscore\n * so `_default` can never collide with a real\n * namespace — `segment(\"s\")` and `segment(\"s\", { namespace: \"_default\" })` would be a grammar error, not an\n * aliasing hazard.\n */\nexport const DEFAULT_NAMESPACE = '_default';\n\n/** Map an optional namespace to its physical part: the namespace itself, or {@link DEFAULT_NAMESPACE}. */\nexport function namespacePart(namespace: string | undefined): string {\n  return namespace ?? DEFAULT_NAMESPACE;\n}\n","/**\n * Logical-ref → MongoDB document-id mapping for {@link MongoWarmDriver} (Phase 7).\n *\n * Pure, SDK-free string logic — unit-testable without a live MongoDB. Each chunk is one document with a\n * **deterministic composite `_id`** `<prefix>|<ns>|<seg>|<chunkKey>`, so create-if-absent is a plain\n * `insertOne` (a duplicate `_id` → the unique-index conflict, no extra index needed) and `get` is a direct\n * `_id` lookup. The ref fields are ALSO stored as separate typed fields (`kp`/`ns`/`seg`/`ck`) so `listChunks`\n * can query + **numerically** sort by `ck` (an `_id`-string sort would order `10` before `9`). The name grammar\n * (validateSegmentRef) forbids `|` in ns/seg and {@link normalizeKeyPrefix} forbids it in the prefix, so the\n * `|`-joined id is unambiguous.\n */\nimport { ValidationError } from '@/core/errors';\nimport { validateChunkRef, validateSegmentRef } from '@/core/validate';\nimport type { ChunkRef, SegmentRef } from '@/core/ports';\nimport { namespacePart } from '../_shared/keys';\n\n/** The scope fields shared by every doc of a segment — the `listChunks` filter. */\nexport interface SegmentScope {\n  kp: string;\n  ns: string;\n  seg: string;\n}\n\n/**\n * Validate a caller-supplied key prefix: no control characters and no `|` (the composite-`_id` delimiter), so\n * it can neither corrupt an id nor alias another prefix. Empty/undefined ⇒ no prefix (stored as `''`).\n */\nexport function normalizeKeyPrefix(prefix: string | undefined): string {\n  if (prefix === undefined || prefix === '') return '';\n  for (const ch of prefix) {\n    if (ch.charCodeAt(0) < 0x20 || ch === '|') {\n      throw new ValidationError('keyPrefix must not contain \"|\" or control characters');\n    }\n  }\n  return prefix;\n}\n\n/** The scope filter for all of a segment's chunk docs. Validates the ref. */\nexport function segmentScope(prefix: string, ref: SegmentRef): SegmentScope {\n  validateSegmentRef(ref);\n  return { kp: prefix, ns: namespacePart(ref.namespace), seg: ref.segment };\n}\n\n/** The deterministic `_id` of one chunk's doc: `<prefix>|<ns>|<seg>|<chunkKey>`. Validates the full ref. */\nexport function chunkDocId(prefix: string, ref: ChunkRef): string {\n  validateChunkRef(ref);\n  return `${prefix}|${namespacePart(ref.namespace)}|${ref.segment}|${ref.chunkKey}`;\n}\n","/**\n * Pure helpers for classifying MongoDB (`mongodb`) errors (Phase 7; transient class mirrors the other drivers).\n *\n * SDK-free + side-effect-free — they read only structural shapes off the thrown value (`err.code`, `err.name`,\n * the driver's `hasErrorLabel`, or a Node socket code). The driver signals a create-if-absent conflict itself\n * (a duplicate-key error, {@link isDuplicateKey}) and a token-fenced conflict via a 0 matched/deleted count,\n * so there's no general OCC classifier here — only duplicate-key + transient-vs-not.\n */\n\n/** A MongoDB numeric server error code, if present (`err.code`). */\nfunction serverCode(err: unknown): number | undefined {\n  const code = (err as { code?: unknown } | null)?.code;\n  return typeof code === 'number' ? code : undefined;\n}\n\n/** A Node system/socket error `code` string (starts with `E`, uppercase letters + underscore, no digits). */\nfunction networkCode(err: unknown): string | undefined {\n  const code = (err as { code?: unknown } | null)?.code;\n  return typeof code === 'string' && /^E[A-Z_]+$/.test(code) ? code : undefined;\n}\n\n/** A create-if-absent lost the race — the `_id` already existed (duplicate key: 11000 / 11001). */\nexport function isDuplicateKey(err: unknown): boolean {\n  const code = serverCode(err);\n  return code === 11000 || code === 11001;\n}\n\n/**\n * Retryable MongoDB server error codes: failover / primary-stepdown / node-unreachable / shutdown / network\n * timeouts. (Names from the server error catalogue.) These are safe to retry; deterministic errors\n * (BadValue, duplicate key, etc.) are not listed, so they surface.\n */\nconst RETRYABLE_CODES = new Set<number>([\n  6, // HostUnreachable\n  7, // HostNotFound\n  89, // NetworkTimeout\n  91, // ShutdownInProgress\n  189, // PrimarySteppedDown\n  262, // ExceededTimeLimit\n  9001, // SocketException\n  10107, // NotWritablePrimary\n  11600, // InterruptedAtShutdown\n  11602, // InterruptedDueToReplStateChange\n  13435, // NotPrimaryNoSecondaryOk\n  13436, // NotPrimaryOrSecondary\n  63, // StaleShardVersion\n  150, // StaleEpoch\n]);\n\n/**\n * A transient MongoDB fault safe to retry: a network error, a server-selection timeout, a driver retryable\n * label (`RetryableWriteError` / `TransientTransactionError`), a retryable server code, or a dropped socket.\n * A duplicate-key error is NOT transient (it's the deterministic create-if-absent conflict).\n */\nexport function isTransient(err: unknown): boolean {\n  if (isDuplicateKey(err)) return false;\n  const name = (err as { name?: unknown } | null)?.name;\n  if (name === 'MongoNetworkError' || name === 'MongoServerSelectionError') return true;\n  const hasLabel = (err as { hasErrorLabel?: (l: string) => boolean } | null)?.hasErrorLabel;\n  if (typeof hasLabel === 'function') {\n    if (\n      hasLabel.call(err, 'RetryableWriteError') ||\n      hasLabel.call(err, 'TransientTransactionError')\n    ) {\n      return true;\n    }\n  }\n  const code = serverCode(err);\n  if (code !== undefined && RETRYABLE_CODES.has(code)) return true;\n  const net = networkCode(err);\n  return (\n    net === 'ECONNRESET' ||\n    net === 'ECONNABORTED' ||\n    net === 'ETIMEDOUT' ||\n    net === 'ESOCKETTIMEDOUT' ||\n    net === 'ECONNREFUSED' ||\n    net === 'EHOSTUNREACH' ||\n    net === 'ENETUNREACH' ||\n    net === 'EPIPE' ||\n    net === 'EAI_AGAIN' ||\n    net === 'ENOTFOUND'\n  );\n}\n","/**\n * `MongoWarmDriver` — an {@link IWarmDriver} over MongoDB / DocumentDB (Phase 7).\n *\n * Uses the official `mongodb` driver, an **optional peer dependency** — only consumers of\n * `cloud-roaring/mongodb` install it. A `Db` is **injected** (dependency injection): the driver owns no\n * connection/credential logic, so it's thin and reuses the caller's client.\n *\n * Each chunk is one document keyed by a deterministic composite `_id` (`<prefix>|<ns>|<seg>|<chunkKey>`) with\n * an opaque OCC **token** (a random UUID per write) and the delta `payload` (BSON binary). Optimistic\n * concurrency is per-document and server-side:\n * - **create-if-absent** = `insertOne` with that `_id`; a duplicate-key error ⇒ {@link WriteConflictError}.\n * - **token-fenced update / delete** = `updateOne` / `deleteOne` filtered on `{ _id, token: expected }`; a 0\n *   matched/deleted count ⇒ {@link WriteConflictError}.\n *\n * Each op is a single-document atomic operation, so no transaction is needed. Tokens are not reused across\n * delete→recreate (ABA-safe, D3 — probabilistic random UUID, like the Postgres/Redis drivers). Reads are\n * strongly consistent when the `Db` targets the primary (majority/primary read preference), which the OCC\n * read-modify-write requires — see the `db` option note. Drivers may use `node:crypto`.\n */\nimport { randomUUID } from 'node:crypto';\nimport type { Collection, Db } from 'mongodb';\nimport { IntegrityError, TransientError, ValidationError, WriteConflictError } from '@/core/errors';\nimport { NO_ROW } from '@/core/ports';\nimport type { ChunkRef, IWarmDriver, NoRow, SegmentRef, Token, WarmRow } from '@/core/ports';\nimport { chunkDocId, normalizeKeyPrefix, segmentScope } from './keys';\nimport { isDuplicateKey, isTransient } from './mongodb-errors';\n\nconst DEFAULT_COLLECTION = 'cloud_roaring_warm';\nconst INDEX_NAME = 'cr_warm_scope_ck';\n/** find() cursor batch — bounds resident docs on a very wide segment. */\nconst DEFAULT_BATCH = 500;\n/**\n * Force **binary (byte-exact) comparison** on every op + the index, overriding any case-/diacritic-insensitive\n * *default* collation the collection may have been created with. Grammar-legal segment/namespace names differ\n * only by case (`\"A\"` vs `\"a\"`) are distinct stores; under a case-insensitive default collation an unpinned\n * `findOne({_id})` / `updateOne` / `deleteOne` would match the *wrong* document → cross-segment read/leak and\n * wrong-doc OCC updates. This is the Mongo analogue of the MySQL `utf8mb4_bin` requirement. Operation-level\n * collation is binary in the common (no-default-collation) case, so this is a no-op there and a fix otherwise.\n * NOTE: `insertOne` uniqueness is governed by the `_id` index's collation, which cannot be overridden per-op —\n * so a collection created with a case-insensitive **default** collation can still collapse `\"a\"`/`\"A\"` on\n * *insert* (a spurious create-if-absent conflict, not a data leak). The operator guide states the warm\n * collection must use the simple default collation; this pin closes the get/update/delete/list matching hole.\n */\nconst SIMPLE_COLLATION = { locale: 'simple' } as const;\n\n/** One warm chunk document. */\ninterface WarmDoc {\n  _id: string;\n  kp: string;\n  ns: string;\n  seg: string;\n  ck: number;\n  token: string;\n  payload: unknown; // written as a Node Buffer; read back as a BSON Binary (or Buffer if promoteBuffers)\n}\n\nexport interface MongoWarmDriverOptions {\n  /**\n   * A `mongodb` `Db` (the driver calls `db.collection(...)`). The driver never connects/closes the client —\n   * the caller owns its lifecycle. Reads must be strongly consistent for OCC (read-your-writes): use a\n   * primary read preference (the default), not `secondary`/`secondaryPreferred`.\n   */\n  readonly db: Db;\n  /** Warm collection name (default `cloud_roaring_warm`). */\n  readonly collection?: string;\n  /** Optional key-prefix component of the composite `_id` so several logical stores can share one collection. */\n  readonly keyPrefix?: string;\n  /** Advanced: `listChunks` cursor batch size (default 500). Tunes peak memory on wide segments. */\n  readonly listPageSize?: number;\n}\n\n/**\n * Create the index `listChunks` needs — `{ kp, ns, seg, ck }` (scope filter + numeric ascending sort) —\n * idempotently. **Run this once at deploy time — it is effectively required, not merely an optimization:**\n * without it a `listChunks` over a segment whose dirty backlog exceeds MongoDB's 32 MB in-memory-sort ceiling\n * fails (server error 292, `QueryExceededMemoryLimitNoDiskUseAllowedException`) rather than merely running\n * slow. (create-if-absent needs no extra index — the composite `_id` is inherently unique.)\n */\nexport async function ensureMongoWarmIndexes(\n  db: Db,\n  collection: string = DEFAULT_COLLECTION,\n): Promise<void> {\n  await db\n    .collection(collection)\n    .createIndex(\n      { kp: 1, ns: 1, seg: 1, ck: 1 },\n      { name: INDEX_NAME, collation: SIMPLE_COLLATION },\n    );\n}\n\nexport class MongoWarmDriver implements IWarmDriver {\n  private readonly col: Collection<WarmDoc>;\n  private readonly keyPrefix: string;\n  private readonly batch: number;\n\n  constructor(options: MongoWarmDriverOptions) {\n    this.keyPrefix = normalizeKeyPrefix(options.keyPrefix);\n    const batch = options.listPageSize ?? DEFAULT_BATCH;\n    if (!Number.isSafeInteger(batch) || batch < 1) {\n      throw new ValidationError(`listPageSize must be a positive safe integer; got ${batch}`);\n    }\n    this.batch = batch;\n    this.col = options.db.collection<WarmDoc>(options.collection ?? DEFAULT_COLLECTION);\n  }\n\n  // Reads are strongly consistent when the Db targets the primary (see the `db` option), so the optional\n  // `WarmReadOptions` hint would be a no-op — the structurally-optional param is simply omitted, as in the\n  // LocalFs / Postgres / Redis drivers.\n  async get(ref: ChunkRef): Promise<WarmRow | null> {\n    const _id = chunkDocId(this.keyPrefix, ref); // validates ref\n    let doc: WarmDoc | null;\n    try {\n      doc = await this.col.findOne({ _id }, { collation: SIMPLE_COLLATION });\n    } catch (err) {\n      throw this.mapError(err);\n    }\n    return doc === null ? null : this.rowFrom(doc, ref.chunkKey);\n  }\n\n  async putConditional(\n    ref: ChunkRef,\n    bytes: Uint8Array,\n    expected: Token | NoRow,\n  ): Promise<{ token: Token }> {\n    const _id = chunkDocId(this.keyPrefix, ref); // validates ref\n    const token = randomUUID();\n    const payload = Buffer.from(bytes); // copy: severs any caller-owned/reused input buffer; stored as Binary\n    try {\n      if (expected === NO_ROW) {\n        const { kp, ns, seg } = segmentScope(this.keyPrefix, ref);\n        await this.col.insertOne({ _id, kp, ns, seg, ck: ref.chunkKey, token, payload });\n      } else {\n        const res = await this.col.updateOne(\n          { _id, token: expected },\n          { $set: { token, payload } },\n          { collation: SIMPLE_COLLATION },\n        );\n        if (res.matchedCount !== 1) {\n          throw new WriteConflictError(`OCC conflict on chunk ${ref.chunkKey}`);\n        }\n      }\n      return { token };\n    } catch (err) {\n      // A duplicate `_id` is the deterministic create-if-absent conflict — never a transient.\n      if (isDuplicateKey(err)) {\n        throw new WriteConflictError(`chunk ${ref.chunkKey} already exists (create-if-absent)`);\n      }\n      throw this.mapError(err); // WriteConflictError (from the update branch) passes straight through\n    }\n  }\n\n  async deleteConditional(ref: ChunkRef, expected: Token): Promise<void> {\n    const _id = chunkDocId(this.keyPrefix, ref); // validates ref\n    let deletedCount: number;\n    try {\n      deletedCount = (\n        await this.col.deleteOne({ _id, token: expected }, { collation: SIMPLE_COLLATION })\n      ).deletedCount;\n    } catch (err) {\n      throw this.mapError(err);\n    }\n    if (deletedCount !== 1) {\n      throw new WriteConflictError(`fenced-delete conflict on chunk ${ref.chunkKey}`);\n    }\n  }\n\n  async *listChunks(ref: SegmentRef): AsyncIterable<{ chunkKey: number } & WarmRow> {\n    const scope = segmentScope(this.keyPrefix, ref); // validates ref\n    // The driver's cursor streams in `batchSize` batches, so peak memory is bounded on a wide segment; the\n    // `{ kp, ns, seg, ck }` index (ensureMongoWarmIndexes) serves the filter + numeric ascending sort — and is\n    // effectively required: without it a wide segment's sort spills past MongoDB's 32 MB in-memory-sort limit\n    // and errors (292), not just runs slow.\n    const cursor = this.col\n      .find(scope, { collation: SIMPLE_COLLATION })\n      .sort({ ck: 1 })\n      .batchSize(this.batch);\n    try {\n      for await (const doc of cursor) {\n        // Validate the yielded chunkKey (untrusted-data posture, like the sibling drivers) — a foreign/corrupt\n        // doc matching the scope but lacking a numeric `ck` must not surface as `{ chunkKey: undefined }`.\n        if (typeof doc.ck !== 'number') {\n          throw new IntegrityError('warm doc is missing its numeric chunk key (ck)');\n        }\n        yield { chunkKey: doc.ck, ...this.rowFrom(doc, doc.ck) };\n      }\n    } catch (err) {\n      throw this.mapError(err);\n    }\n  }\n\n  /** Build a {@link WarmRow} from a doc; reject one missing/!typed fields (corrupt/foreign). */\n  private rowFrom(doc: WarmDoc, chunkKey: number): WarmRow {\n    if (typeof doc.token !== 'string' || doc.token === '') {\n      throw new IntegrityError(`warm doc for chunk ${chunkKey} is missing its token`);\n    }\n    return { token: doc.token, bytes: decodePayload(doc.payload, chunkKey) };\n  }\n\n  /** Reclassify a transient Mongo fault as a retryable {@link TransientError}; else propagate unchanged. */\n  private mapError(err: unknown): unknown {\n    if (isTransient(err)) {\n      return new TransientError(\n        `transient MongoDB fault: ${(err as { message?: unknown } | null)?.message ?? 'unknown'}`,\n        { cause: err },\n      );\n    }\n    return err;\n  }\n}\n\n/**\n * Decode a stored payload to owned bytes, duck-typed so the driver needs no runtime `mongodb`/`bson` import\n * (kept `import type`, so `cloud-roaring/mongodb`'s built bundle has no runtime SDK dependency). A Node\n * `Buffer` is a `Uint8Array` (the `promoteBuffers` case); a BSON `Binary` exposes its bytes on `.buffer`.\n */\nfunction decodePayload(p: unknown, chunkKey: number): Uint8Array {\n  if (p instanceof Uint8Array) return new Uint8Array(p); // Buffer / Uint8Array — copy\n  const inner = (p as { buffer?: unknown } | null)?.buffer; // BSON Binary → .buffer is a Buffer\n  if (inner instanceof Uint8Array) return new Uint8Array(inner);\n  throw new IntegrityError(`warm doc for chunk ${chunkKey} has no readable payload`);\n}\n"]}

package/dist/mongodb/index.d.cts

@@ -0,0 +1,45 @@
+import { Db } from 'mongodb';
+import { a as IWarmDriver, b as ChunkRef, c as WarmRow, T as Token, N as NoRow, S as SegmentRef } from '../ports-D3BrJ6ax.cjs';
+
+interface MongoWarmDriverOptions {
+    /**
+     * A `mongodb` `Db` (the driver calls `db.collection(...)`). The driver never connects/closes the client —
+     * the caller owns its lifecycle. Reads must be strongly consistent for OCC (read-your-writes): use a
+     * primary read preference (the default), not `secondary`/`secondaryPreferred`.
+     */
+    readonly db: Db;
+    /** Warm collection name (default `cloud_roaring_warm`). */
+    readonly collection?: string;
+    /** Optional key-prefix component of the composite `_id` so several logical stores can share one collection. */
+    readonly keyPrefix?: string;
+    /** Advanced: `listChunks` cursor batch size (default 500). Tunes peak memory on wide segments. */
+    readonly listPageSize?: number;
+}
+/**
+ * Create the index `listChunks` needs — `{ kp, ns, seg, ck }` (scope filter + numeric ascending sort) —
+ * idempotently. **Run this once at deploy time — it is effectively required, not merely an optimization:**
+ * without it a `listChunks` over a segment whose dirty backlog exceeds MongoDB's 32 MB in-memory-sort ceiling
+ * fails (server error 292, `QueryExceededMemoryLimitNoDiskUseAllowedException`) rather than merely running
+ * slow. (create-if-absent needs no extra index — the composite `_id` is inherently unique.)
+ */
+declare function ensureMongoWarmIndexes(db: Db, collection?: string): Promise<void>;
+declare class MongoWarmDriver implements IWarmDriver {
+    private readonly col;
+    private readonly keyPrefix;
+    private readonly batch;
+    constructor(options: MongoWarmDriverOptions);
+    get(ref: ChunkRef): Promise<WarmRow | null>;
+    putConditional(ref: ChunkRef, bytes: Uint8Array, expected: Token | NoRow): Promise<{
+        token: Token;
+    }>;
+    deleteConditional(ref: ChunkRef, expected: Token): Promise<void>;
+    listChunks(ref: SegmentRef): AsyncIterable<{
+        chunkKey: number;
+    } & WarmRow>;
+    /** Build a {@link WarmRow} from a doc; reject one missing/!typed fields (corrupt/foreign). */
+    private rowFrom;
+    /** Reclassify a transient Mongo fault as a retryable {@link TransientError}; else propagate unchanged. */
+    private mapError;
+}
+
+export { MongoWarmDriver, type MongoWarmDriverOptions, ensureMongoWarmIndexes };

package/dist/mongodb/index.d.ts

@@ -0,0 +1,45 @@
+import { Db } from 'mongodb';
+import { a as IWarmDriver, b as ChunkRef, c as WarmRow, T as Token, N as NoRow, S as SegmentRef } from '../ports-D3BrJ6ax.js';
+
+interface MongoWarmDriverOptions {
+    /**
+     * A `mongodb` `Db` (the driver calls `db.collection(...)`). The driver never connects/closes the client —
+     * the caller owns its lifecycle. Reads must be strongly consistent for OCC (read-your-writes): use a
+     * primary read preference (the default), not `secondary`/`secondaryPreferred`.
+     */
+    readonly db: Db;
+    /** Warm collection name (default `cloud_roaring_warm`). */
+    readonly collection?: string;
+    /** Optional key-prefix component of the composite `_id` so several logical stores can share one collection. */
+    readonly keyPrefix?: string;
+    /** Advanced: `listChunks` cursor batch size (default 500). Tunes peak memory on wide segments. */
+    readonly listPageSize?: number;
+}
+/**
+ * Create the index `listChunks` needs — `{ kp, ns, seg, ck }` (scope filter + numeric ascending sort) —
+ * idempotently. **Run this once at deploy time — it is effectively required, not merely an optimization:**
+ * without it a `listChunks` over a segment whose dirty backlog exceeds MongoDB's 32 MB in-memory-sort ceiling
+ * fails (server error 292, `QueryExceededMemoryLimitNoDiskUseAllowedException`) rather than merely running
+ * slow. (create-if-absent needs no extra index — the composite `_id` is inherently unique.)
+ */
+declare function ensureMongoWarmIndexes(db: Db, collection?: string): Promise<void>;
+declare class MongoWarmDriver implements IWarmDriver {
+    private readonly col;
+    private readonly keyPrefix;
+    private readonly batch;
+    constructor(options: MongoWarmDriverOptions);
+    get(ref: ChunkRef): Promise<WarmRow | null>;
+    putConditional(ref: ChunkRef, bytes: Uint8Array, expected: Token | NoRow): Promise<{
+        token: Token;
+    }>;
+    deleteConditional(ref: ChunkRef, expected: Token): Promise<void>;
+    listChunks(ref: SegmentRef): AsyncIterable<{
+        chunkKey: number;
+    } & WarmRow>;
+    /** Build a {@link WarmRow} from a doc; reject one missing/!typed fields (corrupt/foreign). */
+    private rowFrom;
+    /** Reclassify a transient Mongo fault as a retryable {@link TransientError}; else propagate unchanged. */
+    private mapError;
+}
+
+export { MongoWarmDriver, type MongoWarmDriverOptions, ensureMongoWarmIndexes };

package/dist/mongodb/index.js

@@ -0,0 +1,199 @@
+import { NO_ROW } from '../chunk-AS6ODRLT.js';
+import { ValidationError, WriteConflictError, IntegrityError, TransientError, validateChunkRef, namespacePart, validateSegmentRef } from '../chunk-NUIDEEFZ.js';
+import { randomUUID } from 'crypto';
+
+// src/drivers/mongodb/keys.ts
+function normalizeKeyPrefix(prefix) {
+  if (prefix === void 0 || prefix === "") return "";
+  for (const ch of prefix) {
+    if (ch.charCodeAt(0) < 32 || ch === "|") {
+      throw new ValidationError('keyPrefix must not contain "|" or control characters');
+    }
+  }
+  return prefix;
+}
+function segmentScope(prefix, ref) {
+  validateSegmentRef(ref);
+  return { kp: prefix, ns: namespacePart(ref.namespace), seg: ref.segment };
+}
+function chunkDocId(prefix, ref) {
+  validateChunkRef(ref);
+  return `${prefix}|${namespacePart(ref.namespace)}|${ref.segment}|${ref.chunkKey}`;
+}
+
+// src/drivers/mongodb/mongodb-errors.ts
+function serverCode(err) {
+  const code = err?.code;
+  return typeof code === "number" ? code : void 0;
+}
+function networkCode(err) {
+  const code = err?.code;
+  return typeof code === "string" && /^E[A-Z_]+$/.test(code) ? code : void 0;
+}
+function isDuplicateKey(err) {
+  const code = serverCode(err);
+  return code === 11e3 || code === 11001;
+}
+var RETRYABLE_CODES = /* @__PURE__ */ new Set([
+  6,
+  // HostUnreachable
+  7,
+  // HostNotFound
+  89,
+  // NetworkTimeout
+  91,
+  // ShutdownInProgress
+  189,
+  // PrimarySteppedDown
+  262,
+  // ExceededTimeLimit
+  9001,
+  // SocketException
+  10107,
+  // NotWritablePrimary
+  11600,
+  // InterruptedAtShutdown
+  11602,
+  // InterruptedDueToReplStateChange
+  13435,
+  // NotPrimaryNoSecondaryOk
+  13436,
+  // NotPrimaryOrSecondary
+  63,
+  // StaleShardVersion
+  150
+  // StaleEpoch
+]);
+function isTransient(err) {
+  if (isDuplicateKey(err)) return false;
+  const name = err?.name;
+  if (name === "MongoNetworkError" || name === "MongoServerSelectionError") return true;
+  const hasLabel = err?.hasErrorLabel;
+  if (typeof hasLabel === "function") {
+    if (hasLabel.call(err, "RetryableWriteError") || hasLabel.call(err, "TransientTransactionError")) {
+      return true;
+    }
+  }
+  const code = serverCode(err);
+  if (code !== void 0 && RETRYABLE_CODES.has(code)) return true;
+  const net = networkCode(err);
+  return net === "ECONNRESET" || net === "ECONNABORTED" || net === "ETIMEDOUT" || net === "ESOCKETTIMEDOUT" || net === "ECONNREFUSED" || net === "EHOSTUNREACH" || net === "ENETUNREACH" || net === "EPIPE" || net === "EAI_AGAIN" || net === "ENOTFOUND";
+}
+
+// src/drivers/mongodb/warm.ts
+var DEFAULT_COLLECTION = "cloud_roaring_warm";
+var INDEX_NAME = "cr_warm_scope_ck";
+var DEFAULT_BATCH = 500;
+var SIMPLE_COLLATION = { locale: "simple" };
+async function ensureMongoWarmIndexes(db, collection = DEFAULT_COLLECTION) {
+  await db.collection(collection).createIndex(
+    { kp: 1, ns: 1, seg: 1, ck: 1 },
+    { name: INDEX_NAME, collation: SIMPLE_COLLATION }
+  );
+}
+var MongoWarmDriver = class {
+  col;
+  keyPrefix;
+  batch;
+  constructor(options) {
+    this.keyPrefix = normalizeKeyPrefix(options.keyPrefix);
+    const batch = options.listPageSize ?? DEFAULT_BATCH;
+    if (!Number.isSafeInteger(batch) || batch < 1) {
+      throw new ValidationError(`listPageSize must be a positive safe integer; got ${batch}`);
+    }
+    this.batch = batch;
+    this.col = options.db.collection(options.collection ?? DEFAULT_COLLECTION);
+  }
+  // Reads are strongly consistent when the Db targets the primary (see the `db` option), so the optional
+  // `WarmReadOptions` hint would be a no-op — the structurally-optional param is simply omitted, as in the
+  // LocalFs / Postgres / Redis drivers.
+  async get(ref) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    let doc;
+    try {
+      doc = await this.col.findOne({ _id }, { collation: SIMPLE_COLLATION });
+    } catch (err) {
+      throw this.mapError(err);
+    }
+    return doc === null ? null : this.rowFrom(doc, ref.chunkKey);
+  }
+  async putConditional(ref, bytes, expected) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    const token = randomUUID();
+    const payload = Buffer.from(bytes);
+    try {
+      if (expected === NO_ROW) {
+        const { kp, ns, seg } = segmentScope(this.keyPrefix, ref);
+        await this.col.insertOne({ _id, kp, ns, seg, ck: ref.chunkKey, token, payload });
+      } else {
+        const res = await this.col.updateOne(
+          { _id, token: expected },
+          { $set: { token, payload } },
+          { collation: SIMPLE_COLLATION }
+        );
+        if (res.matchedCount !== 1) {
+          throw new WriteConflictError(`OCC conflict on chunk ${ref.chunkKey}`);
+        }
+      }
+      return { token };
+    } catch (err) {
+      if (isDuplicateKey(err)) {
+        throw new WriteConflictError(`chunk ${ref.chunkKey} already exists (create-if-absent)`);
+      }
+      throw this.mapError(err);
+    }
+  }
+  async deleteConditional(ref, expected) {
+    const _id = chunkDocId(this.keyPrefix, ref);
+    let deletedCount;
+    try {
+      deletedCount = (await this.col.deleteOne({ _id, token: expected }, { collation: SIMPLE_COLLATION })).deletedCount;
+    } catch (err) {
+      throw this.mapError(err);
+    }
+    if (deletedCount !== 1) {
+      throw new WriteConflictError(`fenced-delete conflict on chunk ${ref.chunkKey}`);
+    }
+  }
+  async *listChunks(ref) {
+    const scope = segmentScope(this.keyPrefix, ref);
+    const cursor = this.col.find(scope, { collation: SIMPLE_COLLATION }).sort({ ck: 1 }).batchSize(this.batch);
+    try {
+      for await (const doc of cursor) {
+        if (typeof doc.ck !== "number") {
+          throw new IntegrityError("warm doc is missing its numeric chunk key (ck)");
+        }
+        yield { chunkKey: doc.ck, ...this.rowFrom(doc, doc.ck) };
+      }
+    } catch (err) {
+      throw this.mapError(err);
+    }
+  }
+  /** Build a {@link WarmRow} from a doc; reject one missing/!typed fields (corrupt/foreign). */
+  rowFrom(doc, chunkKey) {
+    if (typeof doc.token !== "string" || doc.token === "") {
+      throw new IntegrityError(`warm doc for chunk ${chunkKey} is missing its token`);
+    }
+    return { token: doc.token, bytes: decodePayload(doc.payload, chunkKey) };
+  }
+  /** Reclassify a transient Mongo fault as a retryable {@link TransientError}; else propagate unchanged. */
+  mapError(err) {
+    if (isTransient(err)) {
+      return new TransientError(
+        `transient MongoDB fault: ${err?.message ?? "unknown"}`,
+        { cause: err }
+      );
+    }
+    return err;
+  }
+};
+function decodePayload(p, chunkKey) {
+  if (p instanceof Uint8Array) return new Uint8Array(p);
+  const inner = p?.buffer;
+  if (inner instanceof Uint8Array) return new Uint8Array(inner);
+  throw new IntegrityError(`warm doc for chunk ${chunkKey} has no readable payload`);
+}
+
+export { MongoWarmDriver, ensureMongoWarmIndexes };
+//# sourceMappingURL=index.js.map
+//# sourceMappingURL=index.js.map