Upload 5 files

agents.py

@@ -0,0 +1,82 @@
+from langchain_openai import ChatOpenAI
+from langchain.schema import HumanMessage, SystemMessage
+
+def web_summarizer_company(scraped_data):
+
+    AI71_BASE_URL = "https://api.ai71.ai/v1/"
+    AI71_API_KEY = 'ai71-api-f65e97e6-af86-4de9-a667-dc61c1ef8c75'
+
+    llm = ChatOpenAI(
+        model="tiiuae/falcon-180b-chat",
+        api_key=AI71_API_KEY,
+        base_url=AI71_BASE_URL,
+        # streaming=True,
+    )
+    
+    summary = llm.invoke(
+    [
+        SystemMessage(content="You are a Web Data Summarizer. you are skilled at summarizing company webpages into short summaries. "),
+        HumanMessage(content=f"provide a neet summary of the company. Do not add up things. \n\n**Scraped Data:** \n\n{scraped_data}"),
+    ]
+    )
+    return summary
+
+def web_summarizer_person(scraped_data):
+
+    AI71_BASE_URL = "https://api.ai71.ai/v1/"
+    AI71_API_KEY = 'ai71-api-f65e97e6-af86-4de9-a667-dc61c1ef8c75'
+
+    llm = ChatOpenAI(
+        model="tiiuae/falcon-180b-chat",
+        api_key=AI71_API_KEY,
+        base_url=AI71_BASE_URL,
+        # streaming=True,
+    )
+    
+    summary = llm.invoke(
+    [
+        SystemMessage(content="You are a Web Data Summarizer. you are skilled at summarizing complex webpages into short summaries. "),
+        HumanMessage(content=f"provide a neet summary of the youself. Do not add up things. \n\n**Scraped Data from your portfolio:** \n\n{scraped_data}"),
+    ]
+    )
+    return summary
+
+def person_linkedin_agent(topic, summary, mood):
+
+    AI71_BASE_URL = "https://api.ai71.ai/v1/"
+    AI71_API_KEY = 'ai71-api-f65e97e6-af86-4de9-a667-dc61c1ef8c75'
+
+    llm = ChatOpenAI(
+        model="tiiuae/falcon-180b-chat",
+        api_key=AI71_API_KEY,
+        base_url=AI71_BASE_URL,
+        # streaming=True,
+    )
+
+    blog = llm.invoke(
+    [
+        SystemMessage(content=f"Your details are given below: \n\n{summary}"),
+        HumanMessage(content=f"Write a LinkedIn post on the topic: {topic}. Mention your details in the post. Add symbols and emojis to make the post attractive. "),
+    ]
+    )
+    return blog
+
+def company_linkedin_agent(topic, summary, mood):
+
+    AI71_BASE_URL = "https://api.ai71.ai/v1/"
+    AI71_API_KEY = 'ai71-api-f65e97e6-af86-4de9-a667-dc61c1ef8c75'
+
+    llm = ChatOpenAI(
+        model="tiiuae/falcon-180b-chat",
+        api_key=AI71_API_KEY,
+        base_url=AI71_BASE_URL,
+        # streaming=True,
+    )
+
+    blog = llm.invoke(
+    [
+        SystemMessage(content=f"You are a content Writer. You write content in {mood} way. You also add symbols, emojis to make the content attractive."),
+        HumanMessage(content=f"Write a LinkedIn post on the topic: {topic}.\n\nThe post is for the company mentioned below.\n\n{summary}"),
+    ]
+    )
+    return blog

requirements.txt

@@ -0,0 +1,7 @@
+crewai-tools
+langchain
+requests
+json
+Flask==2.2.2
+markupsafe==2.1.2
+langchain-openai

st_app.py

@@ -0,0 +1,64 @@
+import streamlit as st
+from agents import web_summarizer_company, web_summarizer_person, company_linkedin_agent, person_linkedin_agent
+from tools import scrape_website, post_on_linkedin, replace_i_with_you
+
+def main():
+    st.title('LinkedIn Post Generator')
+
+    with st.form(key='linkedin_form'):
+        topic = st.text_input('Topic')
+        url = st.text_input('Website URL')
+        mood = st.text_input('Mood')
+        post_company = st.checkbox('For Company')
+
+        generate_button = st.form_submit_button(label='Generate Post')
+
+        if generate_button:
+            if url:
+                
+                scraped_data = scrape_website(url)
+
+                if post_company:
+                    summary = web_summarizer_company(scraped_data)
+                else:
+                    summary = web_summarizer_person(scraped_data)
+                summary = replace_i_with_you(summary.content)
+                print(summary)
+
+                if post_company:
+                    post_content = company_linkedin_agent(topic, summary, mood)
+                else:
+                    post_content = person_linkedin_agent(topic, summary, mood)
+                
+                print(post_content.content)
+                post_content = post_content.content
+                
+                if post_content.endswith(':'):
+                    post_content = post_content[:-5]
+
+                st.session_state.post_content = post_content
+                st.session_state.post_generated = True
+
+                st.markdown(f"**Generated Post Content:**\n\n{st.session_state.post_content}")
+
+                # Text area for reviewing generated content
+                # st.text_area('Post Content (for review)', st.session_state.post_content, height=200)
+
+            else:
+                st.error('Please provide a URL.')
+
+    # Button to post on LinkedIn outside the form
+    if st.session_state.get('post_generated'):
+        post_linkedin = st.checkbox('Post on LinkedIn')
+        
+        if post_linkedin:
+            token = st.text_input('LinkedIn Token', type='password')
+            post_button = st.button('Post')
+            if post_button:
+                image_path = None
+                post_on_linkedin(token, 'linkedin post', st.session_state.post_content, image_path)
+                st.success('Post has been successfully published on LinkedIn!')
+                st.session_state.post_generated = False  # Reset state after posting
+
+if __name__ == "__main__":
+    main()

tools.py

@@ -0,0 +1,125 @@
+from crewai_tools import ScrapeWebsiteTool
+from langchain.tools import tool
+import requests
+import json
+
+def generate_image():
+    pass
+
+def replace_i_with_you(text):
+    # Replace different forms of 'I' with 'you'
+    text = text.replace(' I ', ' you ')
+    text = text.replace(' I.', ' you.')
+    text = text.replace(' I,', ' you,')
+    text = text.replace(' I\'m ', ' you\'re ')
+    text = text.replace(' I\'m', ' you\'re')
+    text = text.replace(' I am ', ' you are ')
+    text = text.replace(' I am', ' you are')
+    
+    return text
+
+def scrape_website(website_url):
+    """Scrapes all the information from the given website.       
+    Args:
+        website_url: A url of a company website.
+    Returns:
+        Scraped information from the given website.
+    """
+    scrapper = ScrapeWebsiteTool()
+    data = scrapper.run(website_url=website_url)
+    print(data)
+    if not data =='':
+        return data
+    return 'unable to scrape data'
+
+
+def escape_text(text):
+    chars = ["\\", "|", "{", "}", "@", "[", "]", "(", ")", "<", ">", "#", "*", "_", "~"]
+    for char in chars:
+        text = text.replace(char, "\\"+char)
+    return text
+
+def get_urn(token):
+    url = 'https://api.linkedin.com/v2/userinfo'
+    headers = {
+        'Authorization': f'Bearer {token}'
+    }
+    response = requests.get(url, headers=headers)
+    if response.status_code == 200:
+        user_info = response.json()
+        return user_info['sub']
+    else:
+        raise Exception(f'Failed to fetch user info: {response.status_code}, {response.text}')
+
+def post_on_linkedin(token, title, text_content, image_path=None):
+    """
+    Posts an article on LinkedIn with an optional image.
+
+    Args:
+    token: LinkedIn OAuth token.
+    title: LinkedIn post title.
+    text_content: LinkedIn post content.
+    image_path: file path of the image (optional).
+    """
+    text_content = escape_text(text_content)
+    owner = get_urn(token)
+
+    headers = {
+        "LinkedIn-Version": "202401",
+        "X-RestLi-Protocol-Version": "2.0.0",
+        "Content-Type": "application/json",
+        "Authorization": f"Bearer {token}",
+    }
+
+    image_urn = None
+    if image_path:
+        if image_path.startswith('sandbox'):
+            image_path = image_path.split(':')[1]
+        image_path = image_path.strip()
+        
+        # Initialize image upload
+        init_url = "https://api.linkedin.com/rest/images?action=initializeUpload"
+        init_data = json.dumps({"initializeUploadRequest": {"owner": f'urn:li:person:{owner}'}})
+        init_response = requests.post(init_url, headers=headers, data=init_data)
+        if init_response.status_code != 200:
+            raise Exception(f"Failed to initialize upload: {init_response.text}")
+
+        init_response_data = init_response.json()["value"]
+        upload_url = init_response_data["uploadUrl"]
+        image_urn = init_response_data["image"]
+
+        # Upload the file
+        with open(image_path, "rb") as f:
+            upload_response = requests.post(upload_url, files={"file": f})
+            if upload_response.status_code not in [200, 201]:
+                raise Exception(f"Failed to upload file: {upload_response.text}")
+
+    # Create the post
+    post_url = "https://api.linkedin.com/rest/posts"
+    post_data = {
+        "author": f'urn:li:person:{owner}',
+        "commentary": text_content,
+        "visibility": "PUBLIC",
+        "distribution": {
+            "feedDistribution": "MAIN_FEED",
+            "targetEntities": [],
+            "thirdPartyDistributionChannels": [],
+        },
+        "lifecycleState": "PUBLISHED",
+        "isReshareDisabledByAuthor": False,
+    }
+
+    if image_urn:
+        post_data["content"] = {
+            "media": {
+                "title": title,
+                "id": image_urn,
+            }
+        }
+
+    post_data_json = json.dumps(post_data)
+    post_response = requests.post(post_url, headers=headers, data=post_data_json)
+    if post_response.status_code in [200, 201]:
+        return "LinkedIn post generated and posted to user LinkedIn account successfully!"
+    else:
+        raise Exception(f"Failed to post article: {post_response.text}")

utils.py

@@ -0,0 +1,102 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+# import requests
+# from bs4 import BeautifulSoup
+# import os
+# import shutil
+# import re
+
+# import google.generativeai as genai
+
+# def download_image(img_url, folder):
+#     try:
+#         response = requests.get(img_url, stream=True)
+#         filename = os.path.join(folder, img_url.split("/")[-1])
+#         with open(filename, 'wb') as file:
+#             for chunk in response.iter_content(1024):
+#                 file.write(chunk)
+#         print(f"Downloaded {filename}")
+#     except Exception as e:
+#         print(f"Failed to download {img_url}. Reason: {e}")
+
+# def is_logo(img_url):
+#     logo_keywords = ['logo', 'brand', 'icon', 'favicon']
+#     return any(keyword in img_url.lower() for keyword in logo_keywords)
+
+# def extract_images(url, folder):
+#     if os.path.exists(folder):
+#         shutil.rmtree(folder)
+#     os.makedirs(folder)
+
+#     response = requests.get(url)
+#     soup = BeautifulSoup(response.text, 'html.parser')
+    
+#     images = soup.find_all('img')
+#     img_urls = [img['src'] for img in images if 'src' in img.attrs]
+    
+#     for img_url in img_urls:
+#         if img_url.startswith('http'):
+#             if not is_logo(img_url):
+#                 download_image(img_url, folder)
+#         else:
+#             img_url = requests.compat.urljoin(url, img_url)
+#             if not is_logo(img_url):
+#                 download_image(img_url, folder)
+
+
+# def process_script(script):
+#     """Used to process the script into dictionary format"""
+#     dict = {}
+#     title_matches = re.findall(r'<title>(.*?)</title>', script, re.DOTALL)
+#     description_matches = re.findall(r'<description>(.*?)</description>', script, re.DOTALL)
+#     dict['title'] = title_matches[0] if title_matches else "No title found"
+#     dict['description'] = description_matches[0] if description_matches else "No description found"
+#     return dict
+
+# def analyse_images(url):
+#     imgs_dicts = []
+
+#     genai.configure(api_key='AIzaSyBKo19PtvV9oSMRr4R1wJUueyWOL4n5e5c')
+
+#     model = genai.GenerativeModel(model_name="gemini-1.5-pro-latest")
+
+#     files = [ os.path.join('downloaded_images', file) for file in os.listdir('downloaded_images')]
+
+#     for img in files:
+#         sample_file = genai.upload_file(path=img)
+#         file = genai.get_file(name=sample_file.name)
+#         response = model.generate_content([sample_file, f"short description of the image from the website {url} and give a title for the image with title in <title> tag and description in <description> tag"])
+#         img_dict = process_script(response.text)
+#         img_dict['img_path'] = img
+#         imgs_dicts.append(img_dict)
+#         print(img_dict)
+
+#     return imgs_dicts