"""Distance computation package for TianXuan analysis pipeline.

Provides distance functions for various data types (IP, string, boolean,
bytes, timestamp), plus normalisation, SVD denoising, UMAP reduction,
and per-cluster SVD feature extraction.

Public API:
    ip_distance          — subnet-mask + Haversine geographic distance for IP pairs
    string_distance      — Levenshtein / enum-like distance for string columns
    bool_distance        — 0/1 match-to-mode distance for boolean columns
    bytes_distance       — Hamming (popcount) distance for hex byte columns
    timestamp_fft_distance — FFT phase distance for timestamp columns
    normalize_features   — StandardScaler normalisation (μ=0, σ=1)
    svd_denoise          — TruncatedSVD noise removal (returns noise_profile)
    umap_reduce          — UMAP dimensionality reduction (3D→2D fallback)
    cluster_svd_extract  — Per-cluster SVD distinguishing feature extraction
"""

from analysis.distance._geo import ip_distance
from analysis.distance._text import bool_distance, string_distance
from analysis.distance._numeric import bytes_distance, timestamp_fft_distance
from analysis.distance._normalize import normalize_features
from analysis.distance._svd import svd_denoise
from analysis.distance._umap import umap_reduce
from analysis.distance._cluster_svd import cluster_svd_extract

__all__ = [
    "ip_distance",
    "string_distance",
    "bool_distance",
    "bytes_distance",
    "timestamp_fft_distance",
    "normalize_features",
    "svd_denoise",
    "umap_reduce",
    "cluster_svd_extract",
]