@aztec/stdlib 5.0.0-nightly.20260324 → 5.0.0-nightly.20260331

package/src/proofs/chonk_proof.ts

@@ -1,21 +1,41 @@
+import { BarretenbergSync, flattenChonkProofFields } from '@aztec/bb.js';
 import { CHONK_PROOF_LENGTH } from '@aztec/constants';
 import { times } from '@aztec/foundation/collection';
 import { randomBytes } from '@aztec/foundation/crypto/random';
 import { Fr } from '@aztec/foundation/curves/bn254';
 import { bufferSchemaFor } from '@aztec/foundation/schemas';
-import { BufferReader, serializeToBuffer } from '@aztec/foundation/serialize';
+import { BufferReader, numToUInt32BE, serializeToBuffer } from '@aztec/foundation/serialize';
+
+/**
+ * Serialization format detection for ChonkProof is size-based:
+ *   - UNCOMPRESSED (legacy): [field_count=1632: uint32] [fields...]  → total ≈ 52KB (>= 40KB)
+ *   - COMPRESSED:            [byte_count: uint32] [compressed_bytes] → total ≈ 35KB (< 40KB)
+ *
+ * Detection: if the first uint32 equals CHONK_PROOF_LENGTH (1632), it's legacy format
+ * (field count). Otherwise, it's compressed format (byte count). The old uncompressed
+ * format is never smaller than 40KB; compressed proofs are always smaller than 40KB.
+ */
 
 // CHONK: "Client Honk" - An UltraHonk variant with incremental folding and delayed non-native arithmetic.
 export class ChonkProof {
+  /**
+   * Optional compressed proof bytes from chonk compression (point compression + u256 encoding).
+   * When set, toBuffer() will serialize in compressed format (~1.7x smaller).
+   * When reading from compressed format, this is populated and fields are decompressed on demand.
+   */
+  public compressedProof?: Buffer;
+
   constructor(
     // The proof fields.
     // For native verification, attach public inputs via `attachPublicInputs(publicInputs)`.
     // Not using Tuple here due to the length being too high.
     public fields: Fr[],
+    compressedProof?: Buffer,
   ) {
     if (fields.length !== CHONK_PROOF_LENGTH) {
       throw new Error(`Invalid ChonkProof length: ${fields.length}`);
     }
+    this.compressedProof = compressedProof;
   }
 
   public attachPublicInputs(publicInputs: Fr[]) {
@@ -53,19 +73,84 @@ export class ChonkProof {
     return this.toBuffer();
   }
 
+  /**
+   * Deserialize a ChonkProof from a buffer.
+   * Supports both legacy (field elements) and compressed (chonk compression) formats.
+   *
+   * Size-based format detection:
+   *   - First uint32 == CHONK_PROOF_LENGTH (1632): legacy format, read field elements
+   *     Total proof data ≈ 52KB (always >= 40KB)
+   *   - Otherwise: compressed format, first uint32 is byte count of compressed data
+   *     Total proof data ≈ 35KB (always < 40KB)
+   */
   static fromBuffer(buffer: Buffer | BufferReader): ChonkProof {
     const reader = BufferReader.asReader(buffer);
-    const proofLength = reader.readNumber();
-    const proof = reader.readArray(proofLength, Fr);
-    return new ChonkProof(proof);
+    const firstUint32 = reader.readNumber();
+
+    if (firstUint32 === CHONK_PROOF_LENGTH) {
+      // Legacy format: firstUint32 is the field count (1632)
+      // Widen to `number` to prevent TS from narrowing to literal 1632,
+      // which would cause Tuple<Fr, 1632> to exceed the recursion limit.
+      const fieldCount: number = firstUint32;
+      const proof = reader.readArray(fieldCount, Fr);
+      return new ChonkProof(proof);
+    }
+
+    // Compressed format: firstUint32 is the compressed byte count
+    const compressedBytes = reader.readBytes(firstUint32);
+    return ChonkProof.fromCompressedBytes(Buffer.from(compressedBytes));
+  }
+
+  /**
+   * Create a ChonkProof from compressed bytes by decompressing via the BarretenbergSync API.
+   * The compressed format uses point compression and u256 encoding (from PR #20645).
+   *
+   * @param compressed - Compressed proof bytes from chonk compression
+   * @returns ChonkProof with both fields and compressed bytes populated
+   */
+  static fromCompressedBytes(compressed: Buffer): ChonkProof {
+    const api = BarretenbergSync.getSingleton();
+    const result = api.chonkDecompressProof({ compressedProof: new Uint8Array(compressed) });
+
+    // Flatten the structured bb.js ChonkProof into flat Fr[] field elements
+    const flatFields = flattenChonkProofFields(result.proof);
+    const fields = flatFields.map(f => Fr.fromBuffer(Buffer.from(f)));
+
+    // The decompressed proof includes public inputs in hidingOinkProof.
+    // Since ChonkProof stores fields WITHOUT public inputs, strip them.
+    // The number of public inputs = total fields - CHONK_PROOF_LENGTH
+    if (fields.length > CHONK_PROOF_LENGTH) {
+      const numPubInputs = fields.length - CHONK_PROOF_LENGTH;
+      const proofFields = fields.slice(numPubInputs);
+      return new ChonkProof(proofFields, compressed);
+    }
+
+    return new ChonkProof(fields, compressed);
   }
 
+  /**
+   * Serialize the proof to a buffer.
+   * If compressed bytes are available, uses the compressed format (~1.7x smaller).
+   * Otherwise falls back to legacy field element format.
+   */
   public toBuffer() {
+    if (this.compressedProof) {
+      // Compressed format: [compressed_byte_count: uint32] [compressed_bytes]
+      return Buffer.concat([numToUInt32BE(this.compressedProof.length), this.compressedProof]);
+    }
+    // Legacy format: [field_count=1632: uint32] [fields...]
     return serializeToBuffer(this.fields.length, this.fields);
   }
 }
 
 export class ChonkProofWithPublicInputs {
+  /**
+   * Optional compressed proof bytes (covers the full proof WITH public inputs).
+   * Set by the prover when using chonk compression. Flows through to ChonkProof
+   * via removePublicInputs() so the Tx can serialize in compressed format.
+   */
+  public compressedProof?: Buffer;
+
   constructor(
     // The proof fields with public inputs.
     // For recursive verification, use without public inputs via `removePublicInputs()`.
@@ -83,7 +168,8 @@ export class ChonkProofWithPublicInputs {
 
   public removePublicInputs() {
     const numPublicInputs = this.fieldsWithPublicInputs.length - CHONK_PROOF_LENGTH;
-    return new ChonkProof(this.fieldsWithPublicInputs.slice(numPublicInputs));
+    // Flow compressed proof bytes through so the ChonkProof can serialize efficiently
+    return new ChonkProof(this.fieldsWithPublicInputs.slice(numPublicInputs), this.compressedProof);
   }
 
   public isEmpty() {

package/src/tests/factories.ts

@@ -1715,6 +1715,10 @@ export function makeL2Tips(
       block: { number: bn, hash },
       checkpoint: { number: cpn, hash: cph },
     },
+    proposedCheckpoint: {
+      block: { number: bn, hash },
+      checkpoint: { number: cpn, hash: cph },
+    },
     proven: {
       block: { number: bn, hash },
       checkpoint: { number: cpn, hash: cph },

package/src/tx/capsule.ts

@@ -19,6 +19,8 @@ export class Capsule {
     public readonly storageSlot: Fr,
     /** Data passed to the contract  */
     public readonly data: Fr[],
+    /** Optional namespace for the capsule contents */
+    public readonly scope?: AztecAddress,
   ) {}
 
   static get schema() {
@@ -30,12 +32,18 @@ export class Capsule {
   }
 
   toBuffer() {
-    return serializeToBuffer(this.contractAddress, this.storageSlot, new Vector(this.data));
+    return this.scope
+      ? serializeToBuffer(this.contractAddress, this.storageSlot, new Vector(this.data), true, this.scope)
+      : serializeToBuffer(this.contractAddress, this.storageSlot, new Vector(this.data), false);
   }
 
   static fromBuffer(buffer: Buffer | BufferReader): Capsule {
     const reader = BufferReader.asReader(buffer);
-    return new Capsule(AztecAddress.fromBuffer(reader), Fr.fromBuffer(reader), reader.readVector(Fr));
+    const contractAddress = AztecAddress.fromBuffer(reader);
+    const storageSlot = Fr.fromBuffer(reader);
+    const data = reader.readVector(Fr);
+    const hasScope = reader.readBoolean();
+    return new Capsule(contractAddress, storageSlot, data, hasScope ? AztecAddress.fromBuffer(reader) : undefined);
   }
 
   toString() {

package/src/tx/global_variable_builder.ts

@@ -1,3 +1,5 @@
+import type { FeeHeader } from '@aztec/ethereum/contracts';
+import type { CheckpointNumber } from '@aztec/foundation/branded-types';
 import type { EthAddress } from '@aztec/foundation/eth-address';
 import type { SlotNumber } from '@aztec/foundation/schemas';
 
@@ -6,6 +8,18 @@ import type { GasFees } from '../gas/gas_fees.js';
 import type { UInt32 } from '../types/index.js';
 import type { CheckpointGlobalVariables, GlobalVariables } from './global_variables.js';
 
+/** Fee header fields needed for pipelining overrides. */
+export type ForceProposedFeeHeader = {
+  checkpointNumber: CheckpointNumber;
+  feeHeader: FeeHeader;
+};
+
+/** Options for building checkpoint global variables during pipelining. */
+export type BuildCheckpointGlobalVariablesOpts = {
+  forcePendingCheckpointNumber?: CheckpointNumber;
+  forceProposedFeeHeader?: ForceProposedFeeHeader;
+};
+
 /**
  * Interface for building global variables for Aztec blocks.
  */
@@ -32,5 +46,6 @@ export interface GlobalVariableBuilder {
     coinbase: EthAddress,
     feeRecipient: AztecAddress,
     slotNumber: SlotNumber,
+    opts?: BuildCheckpointGlobalVariablesOpts,
   ): Promise<CheckpointGlobalVariables>;
 }