qubit-codec-misc 0.8.3

// =============================================================================
//    Copyright (c) 2026 Haixing Hu.
//
//    SPDX-License-Identifier: Apache-2.0
//
//    Licensed under the Apache License, Version 2.0.
// =============================================================================
//! Representative codec throughput benchmarks.

use std::hint::black_box;

use criterion::BatchSize;
use criterion::Criterion;
use criterion::Throughput;
use criterion::criterion_group;
use criterion::criterion_main;
use qubit_codec::TranscodeStatus;
use qubit_codec::Transcoder;
use qubit_codec_misc::Base64Codec;
use qubit_codec_misc::Base64Decoder;
use qubit_codec_misc::Base64Encoder;
use qubit_codec_misc::CIntegerLiteralCodec;
use qubit_codec_misc::CStringLiteralCodec;
use qubit_codec_misc::FormUrlencodedCodec;
use qubit_codec_misc::HexCodec;
use qubit_codec_misc::PercentCodec;

const TEXT: &str = "field=value with spaces + punctuation / 中间文本";
const HEX_BYTES: &[u8] = b"benchmark bytes with a representative payload";
const INTEGER_LITERAL: &str = "0xBEEFC0DE";

fn benchmark_percent_and_form(c: &mut Criterion) {
    let percent = PercentCodec::new();
    let form = FormUrlencodedCodec::new();
    let mut group = c.benchmark_group("text-codecs");
    group.throughput(Throughput::Bytes(TEXT.len() as u64));
    group.bench_function("percent_encode", |bencher| {
        bencher.iter(|| black_box(percent.encode(black_box(TEXT))));
    });
    group.bench_function("form_encode", |bencher| {
        bencher.iter(|| black_box(form.encode(black_box(TEXT))));
    });
    let percent_encoded = percent.encode(TEXT);
    let form_encoded = form.encode(TEXT);
    group.bench_function("percent_decode", |bencher| {
        bencher.iter(|| black_box(percent.decode(black_box(&percent_encoded))));
    });
    group.bench_function("form_decode", |bencher| {
        bencher.iter(|| black_box(form.decode(black_box(&form_encoded))));
    });
    group.finish();
}

fn benchmark_hex(c: &mut Criterion) {
    let codec = HexCodec::upper()
        .with_byte_prefix("0x")
        .with_separator(" ")
        .with_ignored_ascii_whitespace(true);
    let encoded = codec.encode(HEX_BYTES);
    let mut group = c.benchmark_group("hex-codec");
    group.throughput(Throughput::Bytes(HEX_BYTES.len() as u64));
    group.bench_function("encode_prefixed_separated", |bencher| {
        bencher.iter(|| black_box(codec.encode(black_box(HEX_BYTES))));
    });
    group.bench_function("decode_prefixed_separated", |bencher| {
        bencher.iter(|| black_box(codec.decode(black_box(&encoded))));
    });
    group.finish();

    let mut group = c.benchmark_group("hex-codec-decode_into_reuse");
    for size in [64_usize, 64 * 1024] {
        let bytes = vec![0x5a; size];
        let plain_codec = HexCodec::new();
        let plain_encoded = plain_codec.encode(&bytes);
        let mut plain_output = Vec::with_capacity(size);
        group.throughput(Throughput::Bytes(size as u64));
        group.bench_function(format!("continuous_{size}B"), |bencher| {
            bencher.iter(|| {
                plain_output.clear();
                plain_codec
                    .decode_into(black_box(&plain_encoded), &mut plain_output)
                    .expect("valid continuous hex input");
                black_box(&plain_output);
            });
        });

        let separated_codec = HexCodec::new().with_byte_prefix("0x").with_separator(" ");
        let separated_encoded = separated_codec.encode(&bytes);
        let mut separated_output = Vec::with_capacity(size);
        group.bench_function(format!("prefixed_separated_{size}B"), |bencher| {
            bencher.iter(|| {
                separated_output.clear();
                separated_codec
                    .decode_into(black_box(&separated_encoded), &mut separated_output)
                    .expect("valid prefixed separated hex input");
                black_box(&separated_output);
            });
        });

        for (format, codec) in [
            ("continuous", HexCodec::new()),
            (
                "prefixed_separated",
                HexCodec::new().with_byte_prefix("0x").with_separator(" "),
            ),
        ] {
            let encoded = codec.encode(&bytes);
            for (allocation, initial_capacity) in [("empty", 0), ("presized", size)] {
                let mut observed = Vec::with_capacity(initial_capacity);
                let before = observed.capacity();
                codec.decode_into(&encoded, &mut observed).expect("valid hex input");
                let after = observed.capacity();
                eprintln!(
                    "hex cold capacity: input={} output={} mode={format}/{allocation} before={before} after={after} grew={}",
                    encoded.len(),
                    observed.len(),
                    after > before
                );

                group.bench_function(format!("{format}_{size}B_cold_{allocation}"), |bencher| {
                    bencher.iter_batched(
                        || Vec::with_capacity(initial_capacity),
                        |mut output| {
                            codec
                                .decode_into(black_box(&encoded), &mut output)
                                .expect("valid hex input");
                            black_box(output);
                        },
                        BatchSize::SmallInput,
                    );
                });
            }
        }
    }
    group.finish();
}

fn benchmark_base64_and_c_literals(c: &mut Criterion) {
    let base64 = Base64Codec::standard();
    let c_literals = CStringLiteralCodec::new();
    let integers = CIntegerLiteralCodec::new();
    let base64_encoded = base64.encode(HEX_BYTES);
    let c_literal_encoded = c_literals.encode(HEX_BYTES);
    let mut group = c.benchmark_group("misc-codecs");
    group.throughput(Throughput::Bytes(HEX_BYTES.len() as u64));
    group.bench_function("base64_encode", |bencher| {
        bencher.iter(|| black_box(base64.encode(black_box(HEX_BYTES))));
    });
    group.bench_function("base64_decode", |bencher| {
        bencher.iter(|| black_box(base64.decode(black_box(&base64_encoded))));
    });
    group.bench_function("c_string_encode", |bencher| {
        bencher.iter(|| black_box(c_literals.encode(black_box(HEX_BYTES))));
    });
    group.bench_function("c_string_decode", |bencher| {
        bencher.iter(|| black_box(c_literals.decode(black_box(&c_literal_encoded))));
    });
    group.throughput(Throughput::Bytes(INTEGER_LITERAL.len() as u64));
    group.bench_function("c_integer_decode", |bencher| {
        bencher.iter(|| black_box(integers.decode(black_box(INTEGER_LITERAL))));
    });
    group.finish();
}

/// Encodes the input incrementally using fixed-size chunks and finishes the
/// stream.
fn base64_encode_chunks(encoder: &mut Base64Encoder, input: &[u8], output: &mut [u8], chunk_size: usize) -> usize {
    encoder.reset(&mut [], 0).expect("reset");
    let mut written = 0;
    for chunk in input.chunks(chunk_size) {
        let p = encoder.transcode(chunk, 0, output, written).expect("encoding");
        written += p.written();
    }
    written + encoder.finish(output, written).expect("finish")
}

/// Decodes progressively visible input chunks and finishes after the final
/// chunk.
fn base64_decode_chunks(decoder: &mut Base64Decoder, input: &[u8], output: &mut [u8], chunk_size: usize) -> usize {
    decoder.reset(&mut [], 0).expect("reset");
    let mut cursor = 0;
    let mut written = 0;
    let mut visible_end = chunk_size.min(input.len());
    loop {
        let eof = visible_end == input.len();
        let p = if eof {
            decoder.transcode_eof(&input[..visible_end], cursor, output, written)
        } else {
            decoder.transcode(&input[..visible_end], cursor, output, written)
        }
        .expect("valid Base64");
        cursor += p.read();
        written += p.written();
        match p.status() {
            TranscodeStatus::Complete if eof => break,
            TranscodeStatus::Complete | TranscodeStatus::NeedInput { .. } => {
                if eof {
                    panic!("valid input must resolve at EOF");
                }
                visible_end = visible_end.saturating_add(chunk_size).min(input.len());
            }
            TranscodeStatus::NeedOutput { .. } => {
                panic!("complete bound must cover this stream")
            }
        }
    }
    written + decoder.finish(output, written).expect("finish")
}

fn benchmark_base64_transcoders(c: &mut Criterion) {
    let mut group = c.benchmark_group("base64-transcoders");
    for size in [64_usize, 64 * 1024] {
        let payload: Vec<u8> = (0..size).map(|i| (i as u8).wrapping_mul(37)).collect();
        let codec = Base64Codec::standard();
        let encoded = codec.encode(&payload);
        let mut encoder = codec.encoder();
        let mut decoder = codec.decoder();
        let mut text = vec![0; encoder.max_total_output_len(payload.len()).expect("bound")];
        let mut raw = vec![0; decoder.max_total_output_len(encoded.len()).expect("bound")];
        group.throughput(Throughput::Bytes(size as u64));
        group.bench_function(format!("encode_owned_{size}B"), |b| {
            b.iter(|| black_box(codec.encode(black_box(&payload))));
        });
        group.bench_function(format!("decode_owned_{size}B"), |b| {
            b.iter(|| black_box(codec.decode(black_box(&encoded)).expect("valid input")));
        });
        group.bench_function(format!("encode_buffer_{size}B"), |b| {
            b.iter(|| {
                black_box(
                    encoder
                        .transcode_complete_into(black_box(&payload), black_box(&mut text))
                        .expect("bound"),
                )
            });
        });
        group.bench_function(format!("decode_buffer_{size}B"), |b| {
            b.iter(|| {
                black_box(
                    decoder
                        .transcode_complete_into(black_box(encoded.as_bytes()), black_box(&mut raw))
                        .expect("bound"),
                )
            });
        });

        for chunk_size in [7, 4096] {
            let text_len = base64_encode_chunks(&mut encoder, &payload, &mut text, chunk_size);
            assert_eq!(&text[..text_len], encoded.as_bytes());
            assert_eq!(text_len, encoded.len());
            let raw_len = base64_decode_chunks(&mut decoder, encoded.as_bytes(), &mut raw, chunk_size);
            assert_eq!(&raw[..raw_len], payload.as_slice());
            assert_eq!(raw_len, payload.len());
            group.bench_function(format!("encode_stream_{size}B_chunk{chunk_size}"), |b| {
                b.iter(|| {
                    black_box(base64_encode_chunks(
                        &mut encoder,
                        black_box(&payload),
                        black_box(&mut text),
                        chunk_size,
                    ))
                });
            });
            group.bench_function(format!("decode_stream_{size}B_chunk{chunk_size}"), |b| {
                b.iter(|| {
                    black_box(base64_decode_chunks(
                        &mut decoder,
                        black_box(encoded.as_bytes()),
                        black_box(&mut raw),
                        chunk_size,
                    ))
                });
            });
        }
    }

    for (name, codec) in [
        ("standard_no_pad", Base64Codec::standard_no_pad()),
        ("url_safe", Base64Codec::url_safe()),
        ("url_safe_no_pad", Base64Codec::url_safe_no_pad()),
    ] {
        let payload = vec![0xfb; 64 * 1024];
        let encoded = codec.encode(&payload);
        let mut encoder = codec.encoder();
        let mut decoder = codec.decoder();
        let mut text = vec![0; encoder.max_total_output_len(payload.len()).expect("bound")];
        let mut raw = vec![0; decoder.max_total_output_len(encoded.len()).expect("bound")];
        group.throughput(Throughput::Bytes(payload.len() as u64));
        let text_len = base64_encode_chunks(&mut encoder, &payload, &mut text, 7);
        assert_eq!(&text[..text_len], encoded.as_bytes());
        assert_eq!(text_len, encoded.len());
        let raw_len = base64_decode_chunks(&mut decoder, encoded.as_bytes(), &mut raw, 7);
        assert_eq!(&raw[..raw_len], payload.as_slice());
        assert_eq!(raw_len, payload.len());
        group.bench_function(format!("encode_stream_{name}_64KiB_chunk7"), |b| {
            b.iter(|| {
                black_box(base64_encode_chunks(
                    &mut encoder,
                    black_box(&payload),
                    black_box(&mut text),
                    7,
                ))
            });
        });
        group.bench_function(format!("decode_stream_{name}_64KiB_chunk7"), |b| {
            b.iter(|| {
                black_box(base64_decode_chunks(
                    &mut decoder,
                    black_box(encoded.as_bytes()),
                    black_box(&mut raw),
                    7,
                ))
            });
        });
    }
    group.finish();
}

criterion_group!(
    benches,
    benchmark_percent_and_form,
    benchmark_hex,
    benchmark_base64_and_c_literals,
    benchmark_base64_transcoders
);
criterion_main!(benches);