faster-boto3 0.1.0

Drop-in boto3 replacement powered by Zig — 1.2x faster S3/DynamoDB, 115x with TurboAPI

faster-boto3

Drop-in boto3 acceleration layer powered by Zig. Same API, faster internals.

Approach

NOT a boto3 rewrite. Instead, we replace the slow Python internals with Zig:

Component	boto3 (Python)	faster-boto3 (Zig)	Expected speedup
SigV4 signing	hmac + hashlib	Zig std.crypto.auth.hmac	5-10x
XML parsing (S3)	xml.etree.ElementTree	Zig XML parser	10-20x
JSON parsing (DynamoDB)	json.loads	Zig JSON parser	5-10x
HTTP client	urllib3 (pure Python)	Zig std.http.Client	3-5x
Request serialization	string formatting	Zig bufPrint	5-10x

Architecture

faster_boto3/
  __init__.py          # Drop-in: `from faster_boto3 import Session`
  session.py           # Wraps boto3.Session, injects Zig accelerators
  _accel.zig           # Zig native module (SigV4, parsers, HTTP)
  sigv4.zig            # HMAC-SHA256 signing
  parsers.zig          # XML + JSON response parsing

Feature Parity

boto3 has 551 unit tests and 905 functional tests. We use these as our compatibility gate:

# Run boto3's own tests against faster-boto3
python -m pytest tests/unit/ tests/functional/ -v

Phase 1: SigV4 Signing

The signing hot path runs on every single AWS API call. Moving it to Zig gives immediate speedup across all services (S3, DynamoDB, Lambda, etc).

Usage (target API)

# Option 1: Monkey-patch boto3
import faster_boto3
faster_boto3.patch()  # replaces slow internals with Zig

import boto3
s3 = boto3.client('s3')  # now uses Zig signing + parsing

# Option 2: Direct import
from faster_boto3 import Session
session = Session()
s3 = session.client('s3')