app.py

import gradio as gr
import io 
import os
import uuid
import json
import requests
import subprocess
from pypdf import PdfReader
from pypipertts import PyPiper
from urllib.parse import urlparse, parse_qs
from huggingface_hub import HfApi
import datetime
token=os.environ.get("HF_TOKEN")
user_name="broadfield"
repo_name="a"
#save_data=f'https://huggingface.co/datasets/{user_name}/{repo_name}/raw/main/'

api = HfApi(token=token)

pp=PyPiper()
pp.load_mod()

def read_pdf(pdf_url):
    print(pdf_url)
    file_name=f"{uuid.uuid4()}.pdf"
    response = requests.get(pdf_url.replace('http:','https:'), stream=True)
    if response.status_code == 200:
        with open(file_name, "wb") as f:
            f.write(response.content)
    else:
        print(response.status_code)
    txt_out=""
    reader = PdfReader(file_name)
    number_of_pages = len(reader.pages)
    #file_name=str(pdf_path).split("\\")[-1]
    for i in range(number_of_pages):
        page = reader.pages[i]
        txt_out+=page.extract_text()
        #return txt_out
    return txt_out.replace("\n","")

def load_data(file):
    try:
        r = requests.get(f'{save_data}{file}.json') 
        lod = json.loads(r.text)
    except:
        lod=[]
        pass
    return(lod)
        
def save_data(t,m,l,n,w,p):
    uid=uuid.uuid4()
    with open('template.html','r') as file:
        file_cont=file.read()
    file.close()
    file_out=file_cont.replace('PDFURL',t).replace('MODEL',m).replace('LENGTH',l).replace('NOISE',n).replace('WIDTH',w).replace('PAUSE',p)
    buffer = io.StringIO()
    buffer.write(file_out)
    buffer.seek(0)
    
    
    api.upload_file(
        path_or_fileobj=buffer,
        path_in_repo=f"{uid}.html",
        repo_id=f'{user_name}/{repo_name}',
        token=token,
        repo_type="space",
    )



def load_html(url):
    html=f"""<iframe src="https://docs.google.com/viewer?url={url})&embedded=true" frameborder="0" height="1200px" width="100%"></iframe></div>"""
    return html
head = """
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width" />
<meta name="twitter:card" content="player"/>
<meta name="twitter:site" content=""/>
<meta name="twitter:player" content="https://broadfield-fast-voice-full.hf.space/?"/>
<meta name="twitter:player:stream" content="https://broadfield-fast-voice-full.hf.space/?"/>
<meta name="twitter:player:width" content="100%"/>
<meta name="twitter:player:height" content="100%"/>    
<meta property="og:title" content="PDF to Voice"/>
<meta property="og:description" content="Add PDF link to /?pdfurl= parameter"/>
<!---meta property="og:image" content=""/--->
<!---meta http-equiv="refresh" content="0; url=https://broadfield-fast-voice-full.hf.space/?pdfurl="--->
<script>


function changeMetaTag(name, content) {
  // Find the existing meta tag with the given name
  let metaTag = document.querySelector(`meta[name="${name}"]`);

  // If it doesn't exist, create a new meta tag
  if (!metaTag) {
    metaTag = document.createElement('meta');
    metaTag.name = name;
    document.head.appendChild(metaTag);
  }

  // Set the content of the meta tag
  metaTag.content = content;
}
const urlParams1 = new URLSearchParams(window.location.search);
var p_pdf = urlParams1.get('pdfurl')
console.log("params")
console.log(p_pdf)
/*changeMetaTag('twitter:player', 'https://broadfield-fast-voice-full.hf.space/?pdfurl=ZZZ'.replace('ZZZ',p_pdf))
changeMetaTag('twitter:player:stream', 'https://broadfield-fast-voice-full.hf.space/?pdfurl=ZZZ'.replace('ZZZ',p_pdf))*/
function run(url_in) {
    let url = String(url_in)
    console.log(url);
    
    var framediv = document.getElementById('cap');
    let child = framediv.lastElementChild;
    while (child) {
        framediv.removeChild(child);
        child=framediv.lastElementChild;
    }   
    const iframediv = document.createElement('iframe');
      iframediv.src = 'https://docs.google.com/viewer?url=' + url +'&embedded=true';
      iframediv.width = '100%'; 
      iframediv.height = '1000px';
      iframediv.frameborder="0";
      framediv.appendChild(iframediv);
}
</script>
"""
js="""
function () {
    const urlParams = new URLSearchParams(window.location.search);
    var p_pdf = urlParams.get('pdfurl')
    var p_mod = urlParams.get('mod')
    var p_len = urlParams.get('len')
    var p_nos = urlParams.get('nos')
    var p_wid = urlParams.get('wid')
    var p_pau = urlParams.get('pau')
    if (p_mod) {
        console.log(p_mod)
    } else {
        p_mod = "en_US-joe-medium";
    };
    if (p_len) {
        console.log(p_len)
    } else {
        p_len = "1";
    };
    if (p_nos) {
        console.log(p_nos)
    } else {
        p_nos = "0.5";
    };
    if (p_wid) {
        console.log(p_wid)
    } else {
        p_wid = "0.5";
    };
    if (p_pau) {
        console.log(p_pau)
    } else {
        p_pau = "1";
    };    

    run(p_pdf)
    return [p_pdf, p_mod, p_len, p_nos, p_wid, p_pau];
}

"""
css="""
#id {
height:500px;
width:100%;
}
"""
with gr.Blocks(head=head,css=css) as app:
    html=gr.HTML()
    a=gr.Audio(streaming=True,autoplay=True)
    with gr.Accordion("Voice Controls",open=False):
        with gr.Row():
            stp=gr.Button("Stop")
            upd=gr.Button("Update Voice")
        t=gr.Textbox(label="PDF URL",interactive=True,visible=True)
        m=gr.Dropdown(label="Voice", choices=pp.key_list,value="en_US-lessac-high",interactive=True)
        l=gr.Slider(label="Length", minimum=0.01, maximum=10.0, value=1,interactive=True)
        n=gr.Slider(label="Noise", minimum=0.01, maximum=3.0, value=0.5,interactive=True)
        w=gr.Slider(label="Noise Width", minimum=0.01, maximum=3.0, value=0.5,interactive=True)
        p=gr.Slider(label="Sentence Pause", minimum=0.1, maximum=10.0, value=1,interactive=True)
        save=gr.Button("Save Data")
    h=gr.HTML("""<div id='cap'></div>""")
    with gr.Column(visible=False):    
        bulk=gr.Textbox(label="bulk",interactive=False,visible=True)
    
    app.load(None,None,[t,m,l,n,w,p],js=js)
    tc=t.change(read_pdf,t,bulk)
    mc=m.change(pp.load_mod,m,None)
    bc=bulk.change(pp.stream_tts,[bulk,m,l,n,w,p],a)
    uc=upd.click(pp.stream_tts,[bulk,m,l,n,w,p],a)
    sb=save.click(save_data,[t,m,l,n,w,p],None)
    sc=stp.click(None,None,None, cancels=[tc,mc,bc,uc])
app.queue(default_concurrency_limit=20).launch(max_threads=40)