import streamlit as st
import langcodes
from allosaurus.app import read_recognizer
from pathlib import Path

  
def get_path_to_wav_format(uploaded_file):
#  st.write(dir(uploaded_file))
#  st.write(type(uploaded_file))
#  st.write(uploaded_file)
  uploaded_bytes = uploaded_file.getvalue()
  actual_file_path = Path(uploaded_file.name)
  actual_file_path.write_bytes(uploaded_bytes)
    

  if ".wav" in uploaded_file.name:
    return Path(uploaded_file.name)
  if ".mp3" in uploaded_file.name:
    new_desired_path = actual_file_path.with_suffix(".wav")
    waveform, sample_rate = torchaudio.load(actual_file_path)
    st.info(waveform, sample_rate)
    torchaudio.save(new_desired_path, waveform, sample_rate)
    return new_desired_path


def get_langcode_for_allosaurus(input_code, model):
  langcode = "ipa" # the default allosaurus recognizer
  description = "the default universal setting, not specific to any language"
  
  if not input_code:
    return langcode, description
    


  try:
    lang = langcodes.get(input_code)
    alpha3 = lang.to_alpha3()
    if model.is_available(alpha3 ): # https://github.com/xinjli/allosaurus/blob/f5aa296dc97a90e3035f6cc0fd281cfb64536228/allosaurus/app.py#L57    
      langcode = alpha3 
      description = lang.display_name()
      st.info(f"{langcode} is supported by Allosaurus}")
    else:
#      st.info(f"Could not find supported language for {input_code}")
      st.error(f"Allosaurus doesn't recognize the language code {langcode}}. Perhaps try looking for [alternate codes](https://huggingface.co/spaces/cdleong/langcode-search). Or possibly it's just that there's no support for that particular language code. Proceeding with default setting")
    
  except langcodes.LanguageTagError as e:
    st.error(f"langcodes library could not find a 3-letter ISO code for {input_code}. It may not be a valid code, try searching for [alternate codes](https://huggingface.co/spaces/cdleong/langcode-search). Proceeding with default setting"
    
  return langcode, description 
  
  
  


if __name__ == "__main__":
  input_code = st.text_input("(optional) 2 or 3-letter ISO code for input language. 2-letter codes will be converted to 3-letter codes", max_chars=3)



  
  model = read_recognizer()
  langcode, description = get_langcode_for_allosaurus(input_code)

  st.write(f"Instructing Allosaurus to recognize using language {langcode}. That is, {description}")

  
  uploaded_files = st.file_uploader("Choose a file", type=[
                                                          ".wav", 
#                                                          ".mp3", # TODO: convert .mp3 to .wav and save
                                                          ],
                                                    accept_multiple_files=True,
                                                          )
  
  for uploaded_file in uploaded_files:
    
    if uploaded_file is not None:
  
      
      
      st.audio(uploaded_file, format='audio/wav')
      
      wav_file = get_path_to_wav_format(uploaded_file)    
      st.write(wav_file)
      result = model.recognize(wav_file, langcode)
      st.write(result)