path: root/bin/pdf2mp3

#!/usr/bin/env python3

# first run this
# docker run -p 9977:9998 apache/tika:2.0.0
# export export TIKA_SERVER_ENDPOINT="http://127.0.0.1:9977"
# finally run the script
import argparse
from gtts import gTTS
from tika import parser


class Argparser():
    def __init__(self):
        parser = argparse.ArgumentParser()
        parser.add_argument(
            "--pdf",
            type=str, help="path to the pdf")
        parser.add_argument("--bool", action="store_true",
                            help="bool", default=False)
        self.args = parser.parse_args()


def main() -> None:
    argparser = Argparser()
    raw = parser.from_file(argparser.args.pdf)
    print(raw['content'])
    tts = gTTS(raw['content'])
    tts.save("out.mp3")


if __name__ == "__main__":
    main()