main.py

import chromadb
import requests
import chromadb.utils.embedding_functions as embedding_functions
import bs4
import json
import gradio as gr
import os
embeddingfunc = embedding_functions.HuggingFaceEmbeddingFunction(api_key=os.environ["hf_token"],model_name="BAAI/bge-small-en-v1.5")
# client = chromadb.PersistentClient(path="booksofjainism")
client = chromadb.HttpClient("https://shethjenil-chromadb-server.hf.space/",port=443)
elibbookAI = client.get_or_create_collection("jainebooks")
allbookdata = json.load(open("jainbooks.json","r"))
allsearch = [i['search'] for i in allbookdata]
class jainnlp:
    @classmethod
    def books(cls)->list[str]:
      return list(set(elibbookAI.get(include=[ "documents" ])["documents"]))
    @classmethod
    def loaddata(cls,search:str,progress = gr.Progress(),lang:str="gu")->None:
      for bookdata in allbookdata:
       if bookdata['search'] == search:
        bookname = bookdata['title_english']
        id = bookdata['sr_no']
        pages = int(bookdata["pages"])
        if id not in cls.books():
            for page,content in enumerate(["\n".join(i.split("\n")[3:]) for i in [i for i in bs4.BeautifulSoup(requests.get(f'https://jainqq.org/booktext/{bookname.replace(" ","_")}/{id}').content, 'html.parser').find('div').stripped_strings][::2]]):
                try:
                    contenteng = requests.post("https://translate-pa.googleapis.com/v1/translateHtml", headers={"Content-Type": "application/json+protobuf","User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36","X-Goog-Api-Key": "AIzaSyATBXajvzQLTDHEQbcpq0Ihe0vWDHmO520"}, json=[[content,lang,"en"],"wt_lib"]).json()[0][0]
                    elibbookAI.add(embeddings=embeddingfunc(contenteng),metadatas={"bookname":bookdata['search'],"page":page,"bookid":id,"originalcontent":content,"contenteng":contenteng,"contentimg":bs4.BeautifulSoup(requests.get(f"https://jainqq.org/explore/{id}/{page}").content, "html.parser").find("img",class_="img-fluid").get("src")},ids=f"{id}-{page}",documents=id)            
                    progress(page/pages)
                except:
                    pass
        return "done"
    @classmethod
    def qna(thisclass,query:str,booklist:list[str] = None,notbooklist:list[str] = None,limit:int=1,lang:str="gu")->list:
      if booklist:
        booklist = {"bookid": {"$in": booklist}}
      if notbooklist:
        notbooklist = {"bookid": {"$nin": notbooklist}}
      return [i["contentimg"] for i in elibbookAI.query(embeddingfunc(requests.get(f"https://translate.googleapis.com/translate_a/single?client=gtx&sl={lang}&tl=en&dt=t&q={query}").json()[0][0][0]),n_results=limit,where=booklist)["metadatas"][0]]
    @classmethod
    def reset():
       client.reset()
       return "done"
upload = gr.Interface(jainnlp.loaddata, gr.Dropdown(allsearch),gr.Textbox())
chatref = gr.Interface(jainnlp.qna,gr.Textbox(),gr.Gallery())
if __name__ == "__main__":
    gr.TabbedInterface([upload,chatref],["Upload","Chat"]).launch()