#!/usr/bin/env python
# coding: utf-8

# Copyright (c) Microsoft. All rights reserved.
# Licensed under the MIT license. See LICENSE.md file in the project root for full license information.

from time import sleep
import uuid
import os
import datetime
import logging
try:
    import customvoice
except ImportError:
    print('Pleae copy folder https://github.com/Azure-Samples/cognitive-services-speech-sdk/tree/master/'
          'samples/custom-voice/python/customvoice and keep the same folder structure as github.')
    quit()


def create_project():
    project = customvoice.Project.create(config, project_id, customvoice.ProjectKind.ProfessionalVoice)
    print('Project created. project id: %s' % project.id)


def upload_consent():
    # Custom Voice training need voice consent file with template below.
    # I [voice talent name] am aware that recordings of my voice will be used by [company name] to create
    # and use a synthetic version of my voice.
    # You can find sample consent file here
    # https://github.com/Azure-Samples/Cognitive-Speech-TTS/blob/master/CustomVoice/Sample%20Data/Individual%20utterances%20%2B%20matching%20script/VoiceTalentVerbalStatement.wav
    consent_file_path = r'D:\CNV_API\VoiceTalentVerbalStatement.wav'
    # Custom voice consent file path with voice actor details and consent information
    consent = customvoice.Consent.create(config, project_id, consent_id,
                                         voice_talent_name='Sample Voice Actor',
                                         company_name='Contoso',
                                         audio_file_path=consent_file_path,
                                         locale='en-us')
    print('Consent created. consent id: %s' % consent.id)


# Upload wave and script file in local_folder to Azure blob under https://blob_url/container_name/blob_prefix/
# If container_name doesnt exists, will create a new container.
def upload_training_set_to_blob(blob_url: str, blob_key: str, container_name: str, blob_prefix: str, local_folder: str):
    message = 'Uploading data from ' + local_folder + ' to https://' + blob_url + '/' + container_name + '/' + blob_prefix
    print(message)

    from azure.storage.blob import BlobServiceClient, generate_container_sas, ContainerSasPermissions
    from azure.core.exceptions import ResourceNotFoundError
    blob_service_client = BlobServiceClient(account_url=blob_url, credential=blob_key)
    container_client = blob_service_client.get_container_client(container_name)
    try:
        container_client.get_container_properties()
    except ResourceNotFoundError:
        container_client.create_container()

    for file_name in os.listdir(local_folder):
        file_path = os.path.join(local_folder, file_name)
        print('uploading ' + file_name)
        with open(file_path, "rb") as data:
            container_client.upload_blob(name=blob_prefix+file_name, data=data)

    sas_token = generate_container_sas(
                container_client.account_name,
                container_client.container_name,
                account_key=container_client.credential.account_key,
                permission=ContainerSasPermissions(read=True, list=True),
                expiry=datetime.datetime.utcnow() + datetime.timedelta(hours=1)
            )
    return sas_token


def upload_training_set():
    # Create training set
    training_set = customvoice.TrainingSet.create(config, project_id, training_set_id, locale='en-US')

    # Upload wave and script from local_folder to Azure blob.
    blob_url = 'contoso.blob.core.windows.net'
    blob_key = '<put your Azure Storage accout key here>'
    container_name = 'voicedata'
    blob_prefix = str(uuid.uuid4()) + '/'
    print('blob_prefix: %s' % blob_prefix)
    # You can find sample script and audio file here.
    # https://github.com/Azure-Samples/Cognitive-Speech-TTS/blob/master/CustomVoice/Sample%20Data/Individual%20utterances%20%2B%20matching%20script/SampleScript.txt
    # https://github.com/Azure-Samples/Cognitive-Speech-TTS/blob/master/CustomVoice/Sample%20Data/Individual%20utterances%20%2B%20matching%20script/SampleAudios.zip
    # Pleae unzip audio file. Put both audio and script file in foler below.
    local_folder = r'D:\CNV_API\SampleAudios'
    sas_token = upload_training_set_to_blob(blob_url, blob_key, container_name, blob_prefix, local_folder)
    print('sas_token: %s' % sas_token)

    # Upload data to training et
    containeUrlWithSAS = "https://" + blob_url + '/' + container_name + '?' + sas_token
    print('containeUrl: %s' % containeUrlWithSAS)
    audios = customvoice.AzureBlobContentSource(containeUrlWithSAS, blob_prefix, ['.wav'])
    scripts = customvoice.AzureBlobContentSource(containeUrlWithSAS, blob_prefix, ['.txt'])
    customvoice.TrainingSet.upload_data(config, training_set_id, kind=customvoice.DatasetKind.AudioAndScript,
                                        audios=audios, scripts=scripts)

    # Wait for training set ready
    print('Training set is processing in server. It takes around 5 minutes to 1 hour depend on data size.')
    training_set = customvoice.TrainingSet.get(config, training_set_id)
    while training_set.status != customvoice.Status.Succeeded and training_set.status != customvoice.Status.Failed:
        print('.', end='', flush=True)
        sleep(10)
        training_set = customvoice.TrainingSet.get(config, training_set_id)
    if training_set.status == customvoice.Status.Failed:
        print('Training set failed')
        raise Exception
    elif training_set.status == customvoice.Status.Succeeded:
        print('Training set succeeded')


def train_model():
    model = customvoice.Model.create(config, project_id, model_id,
                                     voice_name='SampleVoiceNeural',
                                     recipe_kind=customvoice.RecipeKind.Default.name,
                                     consent_id=consent_id,
                                     training_set_id=training_set_id)
    print('Started model training. model id: %s' % model.id)

    # Wait for model ready
    print('Model is training in server. It takes around 24 hours.')
    while model.status != customvoice.Status.Succeeded and model.status != customvoice.Status.Failed:
        print('.', end='', flush=True)
        sleep(300)
        model = customvoice.Model.get(config, model_id)
    if model.status == customvoice.Status.Failed:
        print('Model training failed. Failure reason: %s' % model.failure_reason)
        raise Exception
    elif model.status == customvoice.Status.Succeeded:
        print('Model training succeeded')


def deploy_model():
    endpoint = customvoice.Endpoint.create(config, project_id, endpoint_id, model_id)
    print('Start deploying model . endpoint id: %s' % endpoint.id)

    # Wait for model deploy
    print('Deploying model. It takes around 1 to 5 minutes.')
    while endpoint.status != customvoice.Status.Succeeded and endpoint.status != customvoice.Status.Failed:
        print('.', end='', flush=True)
        sleep(10)
        endpoint = customvoice.Endpoint.get(config, endpoint_id)
    if endpoint.status == customvoice.Status.Failed:
        print('Model deploy failed')
        raise Exception
    elif endpoint.status == customvoice.Status.Succeeded:
        print('Model deploy succeeded')


region = 'eastus'  # eastus, westeurope, southeastasia
key = '<put your speech accout key here>'

logging.basicConfig(filename="customvoice.log",
                    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
                    filemode='w')
logger = logging.getLogger()
logger.setLevel(logging.DEBUG)

config = customvoice.Config(key, region, logger)


project_id = 'professional-voice-project-1'
consent_id = 'professional-voice-consent-1'
training_set_id = 'professional-voice-trainingset-1'
model_id = 'professional-voice-model-1'
endpoint_id = str(uuid.uuid4())

# step 1: creat project
create_project()

# step 2: upload consent file
upload_consent()

# step 3: upload training set
upload_training_set()

# step 4: train model
train_model()

# step 5: deploy model
deploy_model()

# step 6: synthesis with endpoint
# You can find sample code here
# https://github.com/Azure-Samples/cognitive-services-speech-sdk/blob/d806061d8ab00e9729128839bf5051c4871ab35f/samples/python/console/speech_synthesis_sample.py#LL119C1-L119C43