"""Benchmark reshaping layers.

To run benchmarks, see the following command for an example, please change the
flag to your custom value:

```
python3 -m benchmarks.layer_benchmark.reshaping_benchmark \
    --benchmark_name=benchmark_cropping2d \
    --num_samples=2048 \
    --batch_size=256 \
    --jit_compile=True
```
"""

from absl import app
from absl import flags

from benchmarks.layer_benchmark.base_benchmark import LayerBenchmark

FLAGS = flags.FLAGS


def benchmark_cropping1d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "Cropping1D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[1024, 256],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_cropping2d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "Cropping2D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 256, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_cropping3d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "Cropping3D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[32, 32, 32, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_flatten(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "Flatten"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 256, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_permute(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "Permute"
    init_args = {
        "dims": (2, 1),
    }
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 256],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_up_sampling1d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "UpSampling1D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_up_sampling2d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "UpSampling2D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[128, 128, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_up_sampling3d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "UpSampling3D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[32, 16, 16, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_zero_padding1d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "ZeroPadding1D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_zero_padding2d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "ZeroPadding2D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[256, 256, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


def benchmark_zero_padding3d(
    num_samples,
    batch_size,
    jit_compile=True,
):
    layer_name = "ZeroPadding3D"
    init_args = {}
    benchmark = LayerBenchmark(
        layer_name,
        init_args,
        input_shape=[32, 32, 32, 3],
        jit_compile=jit_compile,
    )

    benchmark.benchmark_predict(
        num_samples=num_samples,
        batch_size=batch_size,
    )

    benchmark.benchmark_train(
        num_samples=num_samples,
        batch_size=batch_size,
    )


BENCHMARK_NAMES = {
    "benchmark_cropping1d": benchmark_cropping1d,
    "benchmark_cropping2d": benchmark_cropping2d,
    "benchmark_cropping3d": benchmark_cropping3d,
    "benchmark_flatten": benchmark_flatten,
    "benchmark_permute": benchmark_permute,
    "benchmark_up_sampling1d": benchmark_up_sampling1d,
    "benchmark_up_sampling2d": benchmark_up_sampling2d,
    "benchmark_up_sampling3d": benchmark_up_sampling3d,
    "benchmark_zero_padding1d": benchmark_zero_padding1d,
    "benchmark_zero_padding2d": benchmark_zero_padding2d,
    "benchmark_zero_padding3d": benchmark_zero_padding3d,
}


def main(_):
    benchmark_name = FLAGS.benchmark_name
    num_samples = FLAGS.num_samples
    batch_size = FLAGS.batch_size
    jit_compile = FLAGS.jit_compile

    if benchmark_name is None:
        for name, benchmark_fn in BENCHMARK_NAMES.items():
            benchmark_fn(num_samples, batch_size, jit_compile)
        return

    if benchmark_name not in BENCHMARK_NAMES:
        raise ValueError(
            f"Invalid benchmark name: {benchmark_name}, `benchmark_name` must "
            f"be one of {BENCHMARK_NAMES.keys()}"
        )
    benchmark_fn = BENCHMARK_NAMES[benchmark_name]
    benchmark_fn(num_samples, batch_size, jit_compile)


if __name__ == "__main__":
    app.run(main)