Compress bytes using interleaved rANS.
(data: &[u8])
| 43 | |
| 44 | /// Compress bytes using interleaved rANS. |
| 45 | pub fn encode(data: &[u8]) -> Vec<u8> { |
| 46 | if data.is_empty() { |
| 47 | let out = vec![0u8; HEADER_SIZE]; |
| 48 | // uncompressed_size = 0, freq table = all zeros, compressed_size = 0 |
| 49 | return out; |
| 50 | } |
| 51 | |
| 52 | // Build frequency table. |
| 53 | let mut freqs = [0u32; 256]; |
| 54 | for &b in data { |
| 55 | freqs[b as usize] += 1; |
| 56 | } |
| 57 | |
| 58 | // Normalize frequencies to sum to PROB_SCALE. |
| 59 | let norm_freqs = normalize_frequencies(&freqs, data.len()); |
| 60 | |
| 61 | // Build cumulative frequency table. |
| 62 | let (cum_freqs, sym_freqs) = build_cum_table(&norm_freqs); |
| 63 | |
| 64 | // Encode using 4 interleaved streams. |
| 65 | // Each stream processes every 4th byte: stream 0 gets bytes 0,4,8,... |
| 66 | let mut streams: [Vec<u8>; NUM_STREAMS] = std::array::from_fn(|_| Vec::new()); |
| 67 | let mut states = [RANS_L; NUM_STREAMS]; |
| 68 | |
| 69 | // Encode in REVERSE order (rANS encodes backward, decodes forward). |
| 70 | for i in (0..data.len()).rev() { |
| 71 | let stream_idx = i % NUM_STREAMS; |
| 72 | let sym = data[i] as usize; |
| 73 | let freq = sym_freqs[sym]; |
| 74 | let start = cum_freqs[sym]; |
| 75 | |
| 76 | if freq == 0 { |
| 77 | continue; // Symbol with zero frequency — shouldn't happen after normalization. |
| 78 | } |
| 79 | |
| 80 | rans_encode_symbol( |
| 81 | &mut states[stream_idx], |
| 82 | &mut streams[stream_idx], |
| 83 | start, |
| 84 | freq, |
| 85 | ); |
| 86 | } |
| 87 | |
| 88 | // Flush final states. |
| 89 | for i in 0..NUM_STREAMS { |
| 90 | let s = states[i]; |
| 91 | streams[i].push((s & 0xFF) as u8); |
| 92 | streams[i].push(((s >> 8) & 0xFF) as u8); |
| 93 | streams[i].push(((s >> 16) & 0xFF) as u8); |
| 94 | streams[i].push(((s >> 24) & 0xFF) as u8); |
| 95 | } |
| 96 | |
| 97 | // Build output. |
| 98 | let total_compressed: usize = streams.iter().map(|s| s.len()).sum(); |
| 99 | let mut out = Vec::with_capacity(HEADER_SIZE + total_compressed + NUM_STREAMS * 4); |
| 100 | |
| 101 | // Header: uncompressed size. |
| 102 | out.extend_from_slice(&(data.len() as u32).to_le_bytes()); |