refactor: update snappy frame decompress (#7333)

* refactor: update snappy frame decompress

* chore: fix lint errors

* chore: separate code paths for ChunkTypes

packages/reqresp/src/encodingStrategies/sszSnappy/snappyFrames/common.ts

@@ -1,9 +1,38 @@
+import crc32c from "@chainsafe/fast-crc32c";
+
 export enum ChunkType {
   IDENTIFIER = 0xff,
   COMPRESSED = 0x00,
   UNCOMPRESSED = 0x01,
   PADDING = 0xfe,
+  SKIPPABLE = 0x80,
 }
 
 export const IDENTIFIER = Buffer.from([0x73, 0x4e, 0x61, 0x50, 0x70, 0x59]);
 export const IDENTIFIER_FRAME = Buffer.from([0xff, 0x06, 0x00, 0x00, 0x73, 0x4e, 0x61, 0x50, 0x70, 0x59]);
+
+/**
+ * As per the snappy framing format for streams, the size of any uncompressed chunk can be
+ * no longer than 65536 bytes.
+ *
+ * From: https://github.com/google/snappy/blob/main/framing_format.txt#L90:L92
+ */
+export const UNCOMPRESSED_CHUNK_SIZE = 65536;
+
+export function crc(value: Uint8Array): Buffer {
+  // this function doesn't actually need a buffer
+  // see https://github.com/napi-rs/node-rs/blob/main/packages/crc32/index.d.ts
+  const x = crc32c.calculate(value as Buffer);
+  const result = Buffer.allocUnsafe?.(4) ?? Buffer.alloc(4);
+
+  // As defined in section 3 of https://github.com/google/snappy/blob/master/framing_format.txt
+  // And other implementations for reference:
+  // Go: https://github.com/golang/snappy/blob/2e65f85255dbc3072edf28d6b5b8efc472979f5a/snappy.go#L97
+  // Python: https://github.com/andrix/python-snappy/blob/602e9c10d743f71bef0bac5e4c4dffa17340d7b3/snappy/snappy.py#L70
+  // Mask the right hand to (32 - 17) = 15 bits -> 0x7fff, to keep correct 32 bit values.
+  // Shift the left hand with >>> for correct 32 bit intermediate result.
+  // Then final >>> 0 for 32 bits output
+  result.writeUInt32LE((((x >>> 15) | ((x & 0x7fff) << 17)) + 0xa282ead8) >>> 0, 0);
+
+  return result;
+}

packages/reqresp/src/encodingStrategies/sszSnappy/snappyFrames/compress.ts

@@ -1,33 +1,8 @@
-import crc32c from "@chainsafe/fast-crc32c";
 import snappy from "snappy";
-import {ChunkType, IDENTIFIER_FRAME} from "./common.js";
+import {ChunkType, IDENTIFIER_FRAME, UNCOMPRESSED_CHUNK_SIZE, crc} from "./common.js";
 
 // The logic in this file is largely copied (in simplified form) from https://github.com/ChainSafe/node-snappy-stream/
 
-/**
- * As per the snappy framing format for streams, the size of any uncompressed chunk can be
- * no longer than 65536 bytes.
- *
- * From: https://github.com/google/snappy/blob/main/framing_format.txt#L90:L92
- */
-const UNCOMPRESSED_CHUNK_SIZE = 65536;
-
-function checksum(value: Buffer): Buffer {
-  const x = crc32c.calculate(value);
-  const result = Buffer.allocUnsafe?.(4) ?? Buffer.alloc(4);
-
-  // As defined in section 3 of https://github.com/google/snappy/blob/master/framing_format.txt
-  // And other implementations for reference:
-  // Go: https://github.com/golang/snappy/blob/2e65f85255dbc3072edf28d6b5b8efc472979f5a/snappy.go#L97
-  // Python: https://github.com/andrix/python-snappy/blob/602e9c10d743f71bef0bac5e4c4dffa17340d7b3/snappy/snappy.py#L70
-  // Mask the right hand to (32 - 17) = 15 bits -> 0x7fff, to keep correct 32 bit values.
-  // Shift the left hand with >>> for correct 32 bit intermediate result.
-  // Then final >>> 0 for 32 bits output
-  result.writeUInt32LE((((x >>> 15) | ((x & 0x7fff) << 17)) + 0xa282ead8) >>> 0, 0);
-
-  return result;
-}
-
 export async function* encodeSnappy(bytes: Buffer): AsyncGenerator<Buffer> {
   yield IDENTIFIER_FRAME;
 
@@ -36,17 +11,13 @@ export async function* encodeSnappy(bytes: Buffer): AsyncGenerator<Buffer> {
     const compressed = snappy.compressSync(chunk);
     if (compressed.length < chunk.length) {
       const size = compressed.length + 4;
-      yield Buffer.concat([
-        Buffer.from([ChunkType.COMPRESSED, size, size >> 8, size >> 16]),
-        checksum(chunk),
-        compressed,
-      ]);
+      yield Buffer.concat([Buffer.from([ChunkType.COMPRESSED, size, size >> 8, size >> 16]), crc(chunk), compressed]);
     } else {
       const size = chunk.length + 4;
       yield Buffer.concat([
         //
         Buffer.from([ChunkType.UNCOMPRESSED, size, size >> 8, size >> 16]),
-        checksum(chunk),
+        crc(chunk),
         chunk,
       ]);
     }

packages/reqresp/src/encodingStrategies/sszSnappy/snappyFrames/uncompress.ts

@@ -1,6 +1,6 @@
 import {uncompress} from "snappyjs";
 import {Uint8ArrayList} from "uint8arraylist";
-import {ChunkType, IDENTIFIER} from "./common.js";
+import {ChunkType, IDENTIFIER, UNCOMPRESSED_CHUNK_SIZE, crc} from "./common.js";
 
 export class SnappyFramesUncompress {
   private buffer = new Uint8ArrayList();
@@ -21,32 +21,55 @@ export class SnappyFramesUncompress {
       if (this.buffer.length < 4) break;
 
       const type = getChunkType(this.buffer.get(0));
+
+      if (!this.state.foundIdentifier && type !== ChunkType.IDENTIFIER) {
+        throw "malformed input: must begin with an identifier";
+      }
+
       const frameSize = getFrameSize(this.buffer, 1);
 
       if (this.buffer.length - 4 < frameSize) {
         break;
       }
 
-      const data = this.buffer.subarray(4, 4 + frameSize);
+      const frame = this.buffer.subarray(4, 4 + frameSize);
       this.buffer.consume(4 + frameSize);
 
-      if (!this.state.foundIdentifier && type !== ChunkType.IDENTIFIER) {
-        throw "malformed input: must begin with an identifier";
-      }
+      switch (type) {
+        case ChunkType.IDENTIFIER: {
+          if (!Buffer.prototype.equals.call(frame, IDENTIFIER)) {
+            throw "malformed input: bad identifier";
+          }
+          this.state.foundIdentifier = true;
+          continue;
+        }
+        case ChunkType.PADDING:
+        case ChunkType.SKIPPABLE:
+          continue;
+        case ChunkType.COMPRESSED: {
+          const checksum = frame.subarray(0, 4);
+          const data = frame.subarray(4);
 
-      if (type === ChunkType.IDENTIFIER) {
-        if (!Buffer.prototype.equals.call(data, IDENTIFIER)) {
-          throw "malformed input: bad identifier";
+          const uncompressed = uncompress(data, UNCOMPRESSED_CHUNK_SIZE);
+          if (crc(uncompressed).compare(checksum) !== 0) {
+            throw "malformed input: bad checksum";
+          }
+          result.append(uncompressed);
+          break;
         }
-        this.state.foundIdentifier = true;
-        continue;
-      }
+        case ChunkType.UNCOMPRESSED: {
+          const checksum = frame.subarray(0, 4);
+          const uncompressed = frame.subarray(4);
 
-      if (type === ChunkType.COMPRESSED) {
-        result.append(uncompress(data.subarray(4)));
-      }
-      if (type === ChunkType.UNCOMPRESSED) {
-        result.append(data.subarray(4));
+          if (uncompressed.length > UNCOMPRESSED_CHUNK_SIZE) {
+            throw "malformed input: too large";
+          }
+          if (crc(uncompressed).compare(checksum) !== 0) {
+            throw "malformed input: bad checksum";
+          }
+          result.append(uncompressed);
+          break;
+        }
       }
     }
     if (result.length === 0) {
@@ -82,6 +105,10 @@ function getChunkType(value: number): ChunkType {
     case ChunkType.PADDING:
       return ChunkType.PADDING;
     default:
+      // https://github.com/google/snappy/blob/main/framing_format.txt#L129
+      if (value >= 0x80 && value <= 0xfd) {
+        return ChunkType.SKIPPABLE;
+      }
       throw new Error("Unsupported snappy chunk type");
   }
 }

packages/reqresp/test/unit/encodingStrategies/sszSnappy/snappyFrames/uncompress.test.ts

@@ -1,6 +1,7 @@
 import {pipe} from "it-pipe";
 import {Uint8ArrayList} from "uint8arraylist";
 import {describe, expect, it} from "vitest";
+import {ChunkType, IDENTIFIER_FRAME, crc} from "../../../../../src/encodingStrategies/sszSnappy/snappyFrames/common.js";
 import {encodeSnappy} from "../../../../../src/encodingStrategies/sszSnappy/snappyFrames/compress.js";
 import {SnappyFramesUncompress} from "../../../../../src/encodingStrategies/sszSnappy/snappyFrames/uncompress.js";
 
@@ -56,4 +57,43 @@ describe("encodingStrategies / sszSnappy / snappy frames / uncompress", () => {
 
     expect(decompress.uncompress(new Uint8ArrayList(Buffer.alloc(3, 1)))).toBe(null);
   });
+
+  it("should detect invalid checksum", () => {
+    const chunks = new Uint8ArrayList();
+    chunks.append(IDENTIFIER_FRAME);
+
+    chunks.append(Uint8Array.from([ChunkType.UNCOMPRESSED, 0x80, 0x00, 0x00]));
+    // first 4 bytes are checksum
+    // 0xffffffff is clearly an invalid checksum
+    chunks.append(Uint8Array.from(Array.from({length: 0x80}, () => 0xff)));
+
+    const decompress = new SnappyFramesUncompress();
+    expect(() => decompress.uncompress(chunks)).toThrow(/checksum/);
+  });
+
+  it("should detect skippable frames", () => {
+    const chunks = new Uint8ArrayList();
+    chunks.append(IDENTIFIER_FRAME);
+
+    chunks.append(Uint8Array.from([ChunkType.SKIPPABLE, 0x80, 0x00, 0x00]));
+    chunks.append(Uint8Array.from(Array.from({length: 0x80}, () => 0xff)));
+
+    const decompress = new SnappyFramesUncompress();
+    expect(decompress.uncompress(chunks)).toBeNull();
+  });
+
+  it("should detect large data", () => {
+    const chunks = new Uint8ArrayList();
+    chunks.append(IDENTIFIER_FRAME);
+
+    // add a chunk of size 100000
+    chunks.append(Uint8Array.from([ChunkType.UNCOMPRESSED, 160, 134, 1]));
+    const data = Uint8Array.from(Array.from({length: 100000 - 4}, () => 0xff));
+    const checksum = crc(data);
+    chunks.append(checksum);
+    chunks.append(data);
+
+    const decompress = new SnappyFramesUncompress();
+    expect(() => decompress.uncompress(chunks)).toThrow(/large/);
+  });
 });