Crate marqant 

Marqant - Quantum-Compressed Markdown Format

Marqant (.mq) is a revolutionary compression format designed specifically for AI consumption, achieving 90% token reduction while maintaining semantic integrity.

Features

• Token-based compression: Common markdown patterns become single tokens
• AI-optimized: Reduces token usage in LLM contexts by 70-90%
• Streaming support: Can process before full dictionary is loaded
• Multiple compression levels: From light tokenization to quantum compression
• DNS integration: Supports distributed token dictionaries via DNS

Usage

use marqant::Marqant;

let compressor = Marqant::default();
let compressed = compressor.compress("# Hello World\n\nThis is markdown content");
let decompressed = compressor.decompress(&compressed).unwrap();

Binary Format

The .mq format consists of:

• Header: Version, timestamp, sizes, flags
• Token dictionary with escaped patterns
• Compressed content using token substitution
• Optional metadata sections

Modules

angel_blessings

Angels & Demons: The Duality of Compression

bridge

Ayanese Bridge Model Prototype

data_bridge

Marqant Data Bridge (MQ-DBX) Prototype

digest_state

Digest State - Stateful Log Tracking

dns

log_summarizer

Smart Log Summarization

mem8_bridge

MEM|8 Bridge - UTL consciousness packets to wave memory storage

novelty

semantic

utl_enforced

Type-level enforced UTL pipeline

utl_phonetics

UTL Phonetics - Compact binary representation of spoken consciousness

utl_pipeline

Universal Language Pipeline Correct flow: Raw → UTL → Analysis/Storage → Human(lang)

Structs

Marqant

The main Marqant compressor

MqInfo

SmartTree

Constants

MQ2_UNI_DICT_ID

Functions

fnv1a64

mq2_uni_decode

mq2_uni_encode

read_mq_metadata

wordcloud_svg

Generate a deterministic SVG word cloud from input text.