fast-multipart ⚡

Lightning-fast multipart parsing for Python - An enhanced successor to pymulter

🚀 Performance

fast-multipart significantly outperforms the standard multipart library across all file sizes:

File Size	Library	Avg Time	Avg Throughput	Performance Gain
Small (413 B)	fast-multipart	0.0006s	623.1 KB/s	4.1x faster
	multipart	0.0026s	152.6 KB/s
Medium (8 MB)	fast-multipart	0.0055s	1.4 GB/s	3.0x faster
	multipart	0.0165s	483.6 MB/s
Large (80 MB)	fast-multipart	0.0433s	1.8 GB/s	2.4x faster
	multipart	0.1039s	770.2 MB/s
XL (200 MB)	fast-multipart	0.0956s	2.0 GB/s	2.6x faster
	multipart	0.2463s	812.0 MB/s
2XL (1000 MB)	fast-multipart	0.4324s	2.3 GB/s	2.9x faster
	multipart	1.2743s	784.8 MB/s

✨ Key Features

• Up to 4.1x Faster: Significant performance gains over pure Python parsers
• Streaming Parser: Process large files chunk-by-chunk
• Callback-Based: Handle fields and data as they arrive
• Type-Safe: Full type hints support for better development experience

📦 Installation

# Recommended: using uv
uv add fast-multipart

# Or with pip
pip install fast-multipart

🎯 Quick Start

from fast_multipart import MultipartParser, FieldPart

def on_field(field: FieldPart) -> None:
    print(f"Field: {field.name}")
    if field.filename:
        print(f"Filename: {field.filename}")
    print(f"Content-Type: {field.content_type}")

def on_field_data(data: bytes) -> None:
    # Process chunk of field data
    print(f"Received {len(data)} bytes")

def on_field_end() -> None:
    print("Field completed")

# Create parser with boundary from Content-Type header
boundary = "----WebKitFormBoundary7MA4YWxkTrZu0gW"
parser = MultipartParser(
    boundary=boundary,
    on_field=on_field,
    on_field_data=on_field_data,
    on_field_end=on_field_end
)

# Feed data to parser
with open('multipart_data.bin', 'rb') as f:
    while chunk := f.read(8192):
        parser.feed(chunk)

parser.close()

For more info, feel free to check out the tests file.

🔧 API Reference

MultipartParser

The core streaming parser for multipart data.

MultipartParser(
    boundary: str,                              # Boundary string from Content-Type
    on_field: Callable[[FieldPart], Any],      # Called when new field starts
    on_field_data: Callable[[bytes], Any],     # Called for each data chunk
    on_field_end: Callable[[], Any],           # Called when field ends
    buffer_cap: Optional[int] = 8912           # Internal buffer size
)

FieldPart

Contains metadata for each multipart field.

class FieldPart:
    name: str                           # Field name
    filename: Optional[str]             # Original filename (if file upload)
    content_type: Optional[str]         # MIME type
    headers: dict[str, str]            # All field headers