Generating Exact-Byte Test Files: Binary Chunk Injection for QA & Boundary Testing

Why Precise Byte-Target Testing Matters: Proxies, WAFs & Cloud Gateways

In enterprise production systems, incoming files cross multiple architectural boundaries: Cloudflare WAF limits (100 MB or 500 MB), Nginx client_max_body_size thresholds, AWS API Gateway multipart limits (10 MB), and S3 presigned POST policy conditions (content-length-range).

Attempting to produce test files by arbitrarily adjusting image dimensions or JPEG quality sliders is imprecise. Due to Huffman entropy coding and DEFLATE compression, changing an image by 1 pixel or 1% quality can alter file size by several kilobytes unpredictably. To execute rigorous QA, engineers require byte-accurate synthetic test fixtures.

Anatomy of PNG Binary Chunks: IHDR, IDAT, and the Crucial IEND Marker

Every PNG begins with an immutable 8-byte magic header: 89 50 4E 47 0D 0A 1A 0A. Following the header, data is organized into structured chunks consisting of: Length (4 bytes big-endian), Chunk Type (4 bytes ASCII), Chunk Data (variable length), and CRC32 Checksum (4 bytes).

Crucially, the PNG specification allows ancillary chunks that image viewers safely ignore if unneeded. The standard text chunk (tEXt) contains a keyword, a null separator byte (0x00), and a string payload. By inserting a custom tEXt chunk immediately before the 12-byte IEND trailer (49 45 4E 44 AE 42 60 82), we can expand any base PNG to an exact mathematical byte target.

The Byte-Padding Injection Algorithm in TypeScript

To generate a PNG matching an exact target size (TargetBytes):

1. Render a lightweight base PNG in memory and measure its initial size (BaseSize).

2. Compute Delta = TargetBytes - BaseSize.

3. If Delta is less than 12 bytes (the minimum overhead of a chunk), recreate the base image at slightly different dimensions to widen the Delta.

4. Construct a tEXt chunk with Length = Delta - 12, Type = "tEXt", Keyword = "Padding", and payload of zero bytes (0x00).

5. Splice this chunk directly ahead of the terminal IEND chunk and recalculate CRC32.

// Byte injection engine snippet
export function injectPngPadding(baseBytes: Uint8Array, targetBytes: number): Uint8Array {
  const delta = targetBytes - baseBytes.byteLength;
  if (delta <= 0) return baseBytes;

  const payloadSize = delta - 12; // 4-len + 4-type + 4-crc
  const padded = new Uint8Array(targetBytes);
  const insertPos = baseBytes.byteLength - 12; // Just before 12-byte IEND

  padded.set(baseBytes.subarray(0, insertPos), 0);
  // Write length, 'tEXt' type, arbitrary payload, and CRC32
  // ...
  padded.set(baseBytes.subarray(insertPos), targetBytes - 12);
  return padded;
}

JPEG COM Markers & WebP RIFF Packaging Mechanics

For JPEG images, byte expansion utilizes the standard Comment Marker (0xFF 0xFE). The comment marker carries a 2-byte length field followed by raw bytes. Injecting this marker immediately prior to the 2-byte End-of-Image (EOI: 0xFF 0xD9) marker results in a fully compliant JPEG that renders normally in every browser.

For WebP files, which follow the Google RIFF container format, arbitrary padding bytes can be appended within custom chunks or appended to the container payload with an updated 4-byte RIFF file size header.

Negative Testing & Corrupted Upload Handling

Thorough QA pipelines test more than just valid files. Negative test cases are vital for testing exception handling:

• Zero-Byte Files: Validates that backend upload handlers do not crash with division-by-zero or empty-buffer exceptions.

• Truncated Bitstreams: Validates that image processing libraries (ImageMagick, Sharp, PIL) fail gracefully rather than leaking memory.

• Content-Type Mismatches: An executable file disguised with an image/png extension tests MIME-sniffing and magic-byte verification layers.