api.py

''' A barebones Flask API for accessing Festival
'''
import os
import re
import contextlib
from functools import wraps
from time import gmtime, strftime

from sh import Command
from flask import abort, Flask, request, send_file
from flask_restful import Api, reqparse, Resource

FESTIVALDIR = os.getenv('FESTIVALDIR')
MODELPATH = os.getenv('MODELPATH')
VOICEPATH = os.getenv('VOICEPATH')
VOICENAME = os.getenv('VOICENAME')
AUDIODIR = os.getenv('AUDIODIR')

API_KEY = "ThisIsNotASecureKey"

festival = Command("{}/bin/festival".format(FESTIVALDIR))
application = Flask(__name__)
api = Api(application)

def require_appkey(view_function):
    @wraps(view_function)
    def decorated_function(*args, **kwargs):
        if API_KEY:
            if 'x-api-key' in request.headers and request.headers['x-api-key'] == API_KEY:
                return view_function(*args, **kwargs)
            else:
                abort(403)
        else:
            return view_function(*args, **kwargs)

    return decorated_function

@contextlib.contextmanager
def cd(path):
    old_path = os.getcwd()
    os.chdir(path)
    try:
        yield
    finally:
        os.chdir(old_path)

def digits_to_time(dig: str):
    ''' Turns the time requested into the proper format that 'saythistime' function expects in Festival
        
        >>> digits_to_time('111')
        '01:11'
    '''
    if not re.match(r'[0-9]{3,4}$', dig):
        abort(400)
    if len(dig) == 3:
        dig = '0' + dig
    return dig[:2] + ':' + dig[2:]

class TTSData(Resource):
    def __init__(self):
        self.parser = reqparse.RequestParser()
        self.parser.add_argument(
            'text', dest='text',
            type=str, location='args', action='store',
            required=True, help='The text to synthesize'
        )
        self.parser.add_argument(
            'attach', dest='attach',
            type=bool, location='args', default=False,
            required=False, help='Return audio as attachment'
        )

    # @require_appkey
    def get(self):
        args = self.parser.parse_args()
        text = args['text']
        if text == 'current-time':
            text = strftime("%H%M", gmtime())
        fname = "{}.wav".format(args['text'])
        audio_path = os.path.join(AUDIODIR, fname)
        if os.path.exists(audio_path):
            return send_file(audio_path,
                              as_attachment=args['attach'],
                             mimetype='audio/wav',
                             attachment_filename=fname)
        else:
            sub_args = [VOICEPATH,
                        '-b',
                        "(voice_{})".format(VOICENAME),
                        # I am using 'saythistime' because that is the function used
                        # by the Festival limited domain synthesizer for the talking clock in my model
                        # The function can be found here: model/eng_clock/festvox/nrc_time_ap.scm
                        # If you need your own function you can just swap this out.
                        "(utt.save.wave (saythistime \"{}\") \"{}\")".format(digits_to_time(text), audio_path)]
            with cd(MODELPATH):
                response = festival(sub_args)
                if isinstance(response, str) and 'Unknown' in response:
                    return 404
                else:
                    return send_file(audio_path,
                                      as_attachment=args['attach'],
                                     mimetype='audio/wav',
                                     attachment_filename=fname)

api.add_resource(TTSData, '/api/v1/tts')

if __name__ == '__main__':
    application.run()