Sahibsingh12/linkedinpostcreator

1from langchain_community.document_loaders import WebBaseLoader2from langchain.prompts import ChatPromptTemplate3from langchain.output_parsers import ResponseSchema4from langchain.output_parsers import StructuredOutputParser5from langchain.prompts import PromptTemplate 6from langchain.chat_models import ChatOpenAI 7from langchain.chains import LLMChain8from dotenv import load_dotenv9import requests10import streamlit as st 11import re12import openai13 14 15load_dotenv()16 17def is_shortened_url(url):                                  # It is checking whether it is a shorten url or regular website url 18    try:19        response = requests.head(url, allow_redirects=True)20        final_url = response.url21        if final_url != url:22            return True23        return False24    except requests.exceptions.RequestException as e:25        print("Error:", e)26        return False27 28def expand_short_url(short_url):                      # It is converting shorten url to regular url 29    try:30        response = requests.head(short_url, allow_redirects=True)31        if response.status_code == 200:32            return response.url33        else:34            print("Error: Short URL couldn't be expanded.")35            return None36    except requests.exceptions.RequestException as e:37        print("Error:", e)38        return None39 40def get_original_url(url):41    if is_shortened_url(url):42        return expand_short_url(url)43    else:44        return url45 46 47 48# This is the complete code where we are extracting content from the url using WebBaseLoader , using LLM to extract blog content only and then paraphrasing it49def paraphrased_post(url): 50    loader=WebBaseLoader([url],encoding='utf-8')51    docs = loader.load()52 53    template="""You are a helpful LinkedIn webscrapper. You are provided with a data , extract the content of the post only.54            {docs}"""55 56 57    prompt=PromptTemplate(template=template,input_variables=['docs'])58    llm=ChatOpenAI(temperature=0)59    chain=LLMChain(llm=llm,prompt=prompt)60 61 62    result=chain.invoke({'docs':docs},return_only_outputs=True)63 64    data=result['text']65 66    template="""You are a helpful LinkedIn post paraphraser and plagiarism remover bot. You are provided with LinkedIn post content and your task is to paraphrase it and remove plagiarism .Return the output in the format with spaces or stickers if present.67                {data}"""68 69    prompt2=PromptTemplate(template=template,input_variables=['data'])70    llm=ChatOpenAI(temperature=0)71    chain2=LLMChain(llm=llm,prompt=prompt2)72 73    result2=chain2({'data':data},return_only_outputs=True)74    data2=extract_data(result2['text'])75    keywords=data2['Keywords'][:3]76    take_aways=data2['Take Aways'][:3]77    highlights=data2['Highlights'][:3]78    return result2['text'] ,keywords , take_aways, highlights79 80 81def extract_data(post_data):82    keywords = ResponseSchema(name="Keywords",83                        description="These are the keywords extracted from LinkedIn post",type="list")84 85    Take_aways = ResponseSchema(name="Take Aways",86                                description="These are the take aways extracted from LinkedIn post", type= "list")87    Highlights=ResponseSchema(name="Highlights",88                                description="These are the highlights extracted from LinkedIn post", type= "list")89 90    response_schema = [91        keywords,92        Take_aways,93        Highlights94 95    ]96    output_parser = StructuredOutputParser.from_response_schemas(response_schema)97    format_instructions = output_parser.get_format_instructions()98 99    template = """100        You are a helpful keywords , take aways and highlights extractor from the post of LinkedIn Bot. Your task is to extract relevant keywords , take aways and highlights in descending order of their scores in a list, means high relevant should  be on the top .101        From the following text message, extract the following information:102 103        text message: {content}104        {format_instructions}105        """106    107    prompt_template = ChatPromptTemplate.from_template(template)108    messages = prompt_template.format_messages(content=post_data, format_instructions=format_instructions)109    llm = ChatOpenAI(temperature=0)110    response = llm(messages)111    output_dict=  output_parser.parse(response.content)112    return  output_dict113 114 115 116 117 118def main():119    st.title("LinkedIn Post Creator")120    121    # Initialize SessionState dictionary122    session_state = st.session_state123    124    if 'paraphrase' not in session_state:125        session_state.paraphrase = ""126    if 'keywords' not in session_state:127        session_state.keywords = ""128    if 'take_aways' not in session_state:129        session_state.take_aways = ""130    if 'highlights' not in session_state:131        session_state.highlights = ""132    133    # User input for URL134    url = st.sidebar.text_input("Enter URL:", placeholder="Enter URL here...")135    136    # Button to submit URL137    if st.sidebar.button("Submit"):138        try:139            if url:140                original_url = get_original_url(url)141                match = re.match(r"https?://(?:www\.)?linkedin\.com/(posts|feed|pulse)/.*", original_url)  # checking domain and url page (means it should only be a post nothing else like login page or something else)142    143                if match:144                    session_state.paraphrase, session_state.keywords, session_state.take_aways, session_state.highlights = paraphrased_post(url)145                    146                else:147                    st.sidebar.error("Put a valid LinkedIn post URL only")148        except (openai.BadRequestError, TypeError) as e:149            st.sidebar.error("Put a valid LinkedIn post URL only")150 151 152 153    paraphrase_text=st.text_area("Paraphrase:", value=session_state.paraphrase, height=400)154    # import pyperclip155    # if st.button('Copy'):                         # For copying the content (Also install xclip (debian package) if error occured)156    #     pyperclip.copy(paraphrase_text)157    #     st.success('Text copied successfully!')158        159    if st.sidebar.button("Show Keywords") and session_state.keywords:160        st.write("Keywords:")161        for i, statement in enumerate(session_state.keywords, start=1):162            st.write(f"{i}. {statement}")163    164        165    if st.sidebar.button("Show Take Aways") and session_state.take_aways:166        st.write("Take Aways:")167        for i, statement in enumerate(session_state.take_aways, start=1):168            st.write(f"{i}. {statement}")169 170    if st.sidebar.button("Show Highlights") and session_state.highlights:171        st.write("Highlights:")172        for i, statement in enumerate(session_state.highlights, start=1):173            st.write(f"{i}. {statement}")174 175if __name__ == "__main__":176    main()177 178 179 180 181 182 183 184 185 186 187