# Copyright (c) 2022 Horizon Robotics. (authors: Binbin Zhang)
#               2022 Chengdong Liang (liangchengdong@mail.nwpu.edu.cn)
#
# 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 gradio as gr
import torch
from wenet.cli.model import load_model



def process_cat_embs(cat_embs):
    device = "cpu"
    cat_embs = torch.tensor(
        [float(c) for c in cat_embs.split(',')]).to(device)
    return cat_embs


def download_rev_models():
    from huggingface_hub import hf_hub_download
    import joblib

    REPO_ID = "Revai/reginald"

    files = ['10.jit.zip', 'tk.units.txt']
    downloaded_files = [hf_hub_download(repo_id=REPO_ID, filename=f) for f in files]
    model = load_model(downloaded_files[0], downloaded_files[1])
    return model

model = download_rev_models()
    

def recognition(audio, style=0):
    if audio is None:
        return "Input Error! Please enter one audio!"
    # NOTE: model supports 16k sample_rate

    cat_embs = ','.join([str(s) for s in (style, 1-style)])
    cat_embs = process_cat_embs(cat_embs)
    ans = model.transcribe(audio, cat_embs = cat_embs)

    if ans is None:
        return "ERROR! No text output! Please try again!"
    txt = ans['text']
    txt = txt.replace('▁', ' ')
    return txt


# input
inputs = [
    gr.inputs.Audio(source="microphone", type="filepath", label='Input audio'),
    gr.Slider(0, 1, value=0, label="Verbatimicity - from non-verbatim (0) to verbatim (1)", info="Choose a transcription style between non-verbatim and verbatim"),
]

examples = [
    ['examples/POD1000000012_S0000335.wav'],
    ['examples/POD1000000013_S0000062.wav'],
    ['examples/POD1000000032_S0000020.wav'], 
    ['examples/POD1000000032_S0000038.wav'],
    ['examples/POD1000000032_S0000050.wav'],
    ['examples/POD1000000032_S0000058.wav'],
]


output = gr.outputs.Textbox(label="Output Text")

text = "Reginald Transcription Styles Demo"

# description
description = (
    "Reginald supports verbatim and non-verbatim transcription. Try recording an audio with disfluencies (ex: \'uh\', \'um\') and testing both transcription styles."  # noqa
)

article = (
    "<p style='text-align: center'>"
    "<a href='https://rev.com' target='_blank'>Learn more about Rev</a>"  # noqa
    "</p>")

interface = gr.Interface(
    fn=recognition,
    inputs=inputs,
    outputs=output,
    title=text,
    description=description,
    article=article,
    examples=examples,
    theme='huggingface',
)

interface.launch(enable_queue=True)