# coding=utf-8
# Copyright 2026 HuggingFace Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import unittest

import numpy as np
import torch
from transformers import Qwen2Tokenizer, Qwen3Config, Qwen3Model

from diffusers import (
    AutoencoderKL,
    FlowMatchEulerDiscreteScheduler,
    OvisImagePipeline,
    OvisImageTransformer2DModel,
)

from ...testing_utils import torch_device
from ..pipeline_params import TEXT_TO_IMAGE_BATCH_PARAMS, TEXT_TO_IMAGE_IMAGE_PARAMS, TEXT_TO_IMAGE_PARAMS
from ..test_pipelines_common import PipelineTesterMixin


class OvisImagePipelineFastTests(PipelineTesterMixin, unittest.TestCase):
    pipeline_class = OvisImagePipeline
    params = TEXT_TO_IMAGE_PARAMS - {"cross_attention_kwargs"}
    batch_params = TEXT_TO_IMAGE_BATCH_PARAMS
    image_params = TEXT_TO_IMAGE_IMAGE_PARAMS
    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
    required_optional_params = frozenset(
        [
            "num_inference_steps",
            "generator",
            "latents",
            "return_dict",
            "callback_on_step_end",
            "callback_on_step_end_tensor_inputs",
        ]
    )
    test_xformers_attention = False
    test_layerwise_casting = True
    test_group_offloading = True

    def get_dummy_components(self):
        torch.manual_seed(0)
        transformer = OvisImageTransformer2DModel(
            patch_size=1,
            in_channels=4,
            out_channels=4,
            num_layers=1,
            num_single_layers=1,
            attention_head_dim=16,
            num_attention_heads=2,
            joint_attention_dim=32,
            axes_dims_rope=(4, 4, 8),
        )
        torch.manual_seed(0)
        vae = AutoencoderKL(
            sample_size=32,
            in_channels=3,
            out_channels=3,
            block_out_channels=(4,),
            layers_per_block=1,
            latent_channels=1,
            norm_num_groups=1,
            use_quant_conv=False,
            use_post_quant_conv=False,
            shift_factor=0.0609,
            scaling_factor=1.5035,
        )
        scheduler = FlowMatchEulerDiscreteScheduler()
        tokenizer = Qwen2Tokenizer.from_pretrained("hf-internal-testing/tiny-random-Qwen2VLForConditionalGeneration")
        torch.manual_seed(0)
        text_encoder = Qwen3Model(
            Qwen3Config(
                hidden_size=32,
                intermediate_size=64,
                num_hidden_layers=2,
                num_attention_heads=4,
                num_key_value_heads=2,
                head_dim=8,
                vocab_size=tokenizer.vocab_size + 4,
                max_position_embeddings=512,
            )
        )
        return {
            "scheduler": scheduler,
            "vae": vae,
            "text_encoder": text_encoder,
            "tokenizer": tokenizer,
            "transformer": transformer,
        }

    def get_dummy_inputs(self, device, seed=0):
        if str(device).startswith("mps"):
            generator = torch.manual_seed(seed)
        else:
            generator = torch.Generator(device=device).manual_seed(seed)

        return {
            "prompt": "a cat",
            "negative_prompt": "bad quality",
            "generator": generator,
            "num_inference_steps": 2,
            "guidance_scale": 2.0,
            "height": 16,
            "width": 16,
            "max_sequence_length": 32,
            "output_type": "np",
        }

    def test_inference(self):
        device = "cpu"
        components = self.get_dummy_components()
        pipe = self.pipeline_class(**components)
        pipe.to(device)
        pipe.set_progress_bar_config(disable=None)

        inputs = self.get_dummy_inputs(device)
        image = pipe(**inputs).images
        generated_image = image[0]
        self.assertEqual(generated_image.shape, (16, 16, 3))
        self.assertTrue(np.isfinite(image).all())

    def test_guidance_scale_is_set(self):
        # The `guidance_scale` property reads `self._guidance_scale`, which `__call__` must initialize.
        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
        inputs = self.get_dummy_inputs(torch_device)
        pipe(**inputs)
        assert pipe.guidance_scale == inputs["guidance_scale"]

    def test_max_sequence_length_is_used(self):
        # `max_sequence_length` should bound the encoded prompt length.
        pipe = self.pipeline_class(**self.get_dummy_components()).to(torch_device)
        embeds_16 = pipe.encode_prompt(
            "a cat", do_classifier_free_guidance=False, device=torch_device, max_sequence_length=16
        )[0]
        embeds_32 = pipe.encode_prompt(
            "a cat", do_classifier_free_guidance=False, device=torch_device, max_sequence_length=32
        )[0]
        assert embeds_16.shape[1] == 16
        assert embeds_32.shape[1] == 32