pic-scale 0.7.8

/*
 * Copyright (c) Radzivon Bartoshyk 3/2026. All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification,
 * are permitted provided that the following conditions are met:
 *
 * 1.  Redistributions of source code must retain the above copyright notice, this
 * list of conditions and the following disclaimer.
 *
 * 2.  Redistributions in binary form must reproduce the above copyright notice,
 * this list of conditions and the following disclaimer in the documentation
 * and/or other materials provided with the distribution.
 *
 * 3.  Neither the name of the copyright holder nor the names of its
 * contributors may be used to endorse or promote products derived from
 * this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
#![forbid(unsafe_code)]

use crate::convolution::{
    ColumnFilter, ConvolutionOptions, HorizontalFilterPass, RowFilter, VerticalConvolutionPass,
};
use crate::factory::rgb_u8::vertical_strategy_u8;
use crate::filter_weights::*;
use crate::handler_provider::{handle_fixed_row_u8, handle_fixed_rows_4_u8};
#[cfg(all(target_arch = "aarch64", feature = "neon",))]
use crate::neon::*;
use crate::plan::HorizontalFiltering;
#[cfg(all(any(target_arch = "x86_64", target_arch = "x86"), feature = "sse"))]
use crate::sse::{convolve_horizontal_rgba_sse_rows_4, convolve_horizontal_rgba_sse_rows_one};
use crate::{ImageStore, ThreadingPolicy};
#[allow(dead_code)]
use num_traits::AsPrimitive;
use std::sync::Arc;

#[allow(dead_code)]
#[derive(Default)]
pub(crate) struct DefaultWeightsConverterQ7 {}

#[allow(dead_code)]
impl WeightsConverter<i8> for DefaultWeightsConverterQ7
where
    f64: AsPrimitive<i8>,
{
    fn prepare_weights(&self, weights: &FilterWeights<f32>) -> FilterWeights<i8> {
        weights.numerical_approximation_q0_7(0)
    }
}

impl HorizontalFilterPass<u8, f32, 4> for ImageStore<'_, u8, 4> {
    fn horizontal_plan(
        filter_weights: FilterWeights<f32>,
        threading_policy: ThreadingPolicy,
        _options: ConvolutionOptions,
    ) -> Arc<dyn RowFilter<u8, 4> + Send + Sync> {
        let _scale_factor = _options.src_size.width as f32 / _options.dst_size.width as f32;
        #[allow(clippy::type_complexity)]
        let mut _dispatcher_4_rows: Option<
            fn(&[u8], usize, &mut [u8], usize, &FilterWeights<i16>, u32),
        > = Some(handle_fixed_rows_4_u8::<4>);
        #[allow(clippy::type_complexity)]
        let mut _dispatcher_1_row: fn(&[u8], &mut [u8], &FilterWeights<i16>, u32) =
            handle_fixed_row_u8::<4>;
        #[cfg(all(target_arch = "aarch64", feature = "neon"))]
        {
            match _options.workload_strategy {
                crate::WorkloadStrategy::PreferQuality => {
                    _dispatcher_4_rows = Some(convolve_horizontal_rgba_neon_rows_4_u8);
                    _dispatcher_1_row = convolve_horizontal_rgba_neon_row;
                }
                crate::WorkloadStrategy::PreferSpeed => {
                    #[cfg(feature = "rdm")]
                    if _scale_factor < 8. && std::arch::is_aarch64_feature_detected!("rdm") {
                        _dispatcher_4_rows = Some(convolve_horizontal_rgba_neon_rows_4_u8_i16);
                        _dispatcher_1_row = convolve_horizontal_rgba_neon_row_i16;
                    } else {
                        _dispatcher_4_rows = Some(convolve_horizontal_rgba_neon_rows_4_u8);
                        _dispatcher_1_row = convolve_horizontal_rgba_neon_row;
                    }
                    #[cfg(not(feature = "rdm"))]
                    {
                        _dispatcher_4_rows = Some(convolve_horizontal_rgba_neon_rows_4_u8);
                        _dispatcher_1_row = convolve_horizontal_rgba_neon_row;
                    }
                    #[cfg(feature = "nightly_i8mm")]
                    if _scale_factor < 10. && std::arch::is_aarch64_feature_detected!("i8mm") {
                        let _dispatcher_4_rows: Option<
                            fn(&[u8], usize, &mut [u8], usize, &FilterWeights<i8>, u32),
                        > = Some(convolve_horizontal_rgba_neon_rows_4_u8_dot);
                        let _dispatcher_1_row = convolve_horizontal_rgba_neon_row_dot;
                        let i_weights = filter_weights.numerical_approximation_q0_7(0);
                        return Arc::new(HorizontalFiltering {
                            filter_weights: i_weights,
                            filter_4_rows: _dispatcher_4_rows,
                            filter_row: _dispatcher_1_row,
                            threading_policy,
                        });
                    }
                }
            }
        }
        #[cfg(all(any(target_arch = "x86_64", target_arch = "x86"), feature = "sse"))]
        {
            if std::arch::is_x86_feature_detected!("sse4.1") {
                _dispatcher_4_rows = Some(convolve_horizontal_rgba_sse_rows_4);
                _dispatcher_1_row = convolve_horizontal_rgba_sse_rows_one;
            }
        }
        #[cfg(all(target_arch = "x86_64", feature = "avx"))]
        {
            let has_avx = std::arch::is_x86_feature_detected!("avx2");
            if has_avx {
                use crate::avx2::{
                    convolve_horizontal_rgba_avx_row_1, convolve_horizontal_rgba_row_4,
                };
                _dispatcher_4_rows = Some(convolve_horizontal_rgba_row_4);
                _dispatcher_1_row = convolve_horizontal_rgba_avx_row_1;
            }
        }
        #[cfg(all(feature = "avx512", target_arch = "x86_64"))]
        {
            if std::arch::is_x86_feature_detected!("avxvnni")
                && _options.workload_strategy != crate::WorkloadStrategy::PreferSpeed
            {
                use crate::avx512::{
                    convolve_horizontal_rgba_vnni_row_1, convolve_horizontal_rgba_vnni_row_4,
                };
                _dispatcher_4_rows = Some(convolve_horizontal_rgba_vnni_row_4);
                _dispatcher_1_row = convolve_horizontal_rgba_vnni_row_1;
            }
        }
        #[cfg(all(target_arch = "wasm32", target_feature = "simd128"))]
        {
            use crate::wasm32::{
                convolve_horizontal_rgba_wasm_row, convolve_horizontal_rgba_wasm_rows_4_u8,
            };
            _dispatcher_4_rows = Some(convolve_horizontal_rgba_wasm_rows_4_u8);
            _dispatcher_1_row = convolve_horizontal_rgba_wasm_row;
        }
        use crate::support::PRECISION;
        let i_weights = filter_weights.numerical_approximation::<i16, PRECISION>(0);
        Arc::new(HorizontalFiltering {
            filter_weights: i_weights,
            filter_4_rows: _dispatcher_4_rows,
            filter_row: _dispatcher_1_row,
            threading_policy,
        })
    }
}

impl VerticalConvolutionPass<u8, f32, 4> for ImageStore<'_, u8, 4> {
    fn vertical_plan(
        filter_weights: FilterWeights<f32>,
        threading_policy: ThreadingPolicy,
        options: ConvolutionOptions,
    ) -> Arc<dyn ColumnFilter<u8, 4> + Send + Sync> {
        vertical_strategy_u8(filter_weights, threading_policy, options)
    }
}