Create app.py

app.py

@@ -1,6 +1,10 @@
 import streamlit as st
-from openai import OpenAI
+import openai
 import os
+import requests
+from bs4 import BeautifulSoup
+from urllib.parse import urljoin, urlparse
+import re
 from reportlab.lib.pagesizes import letter
 from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
 from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer
@@ -87,16 +91,184 @@ st.markdown("""
 </style>
 """, unsafe_allow_html=True)
 
-# Initialize OpenAI client
+# Initialize OpenAI API key
 @st.cache_resource
-def get_openai_client():
+def setup_openai():
     api_key = os.getenv("OPENAI_API_KEY")
     if not api_key:
         st.error("Please set your OPENAI_API_KEY environment variable")
-        return None
-    return OpenAI(api_key=api_key)
+        return False
+    openai.api_key = api_key
+    return True
+
+openai_ready = setup_openai()
 
-client = get_openai_client()
+# Website analysis functions
+def scrape_website(url, max_pages=3):
+    """Scrape website content for business information"""
+    if not url.startswith("http"):
+        url = f"https://{url}"
+    
+    visited = set()
+    to_visit = [url]
+    all_content = []
+    scrape_successful = False
+    
+    while to_visit and len(visited) < max_pages:
+        current_url = to_visit.pop(0)
+        if current_url in visited:
+            continue
+            
+        try:
+            response = requests.get(current_url, timeout=10, headers={
+                'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36'
+            })
+            response.raise_for_status()
+            soup = BeautifulSoup(response.content, "html.parser")
+            visited.add(current_url)
+            scrape_successful = True
+            
+            # Extract meaningful content
+            title = soup.find("title")
+            if title:
+                all_content.append(f"Title: {title.get_text(strip=True)}")
+            
+            meta_description = soup.find("meta", {"name": "description"})
+            if meta_description and meta_description.get("content"):
+                all_content.append(f"Description: {meta_description['content']}")
+            
+            # Extract headings
+            for heading in soup.find_all(["h1", "h2", "h3"])[:10]:
+                all_content.append(f"Heading: {heading.get_text(strip=True)}")
+            
+            # Extract paragraphs
+            paragraphs = soup.find_all("p")[:15]
+            for para in paragraphs:
+                text = para.get_text(strip=True)
+                if len(text) > 20:  # Only meaningful paragraphs
+                    all_content.append(text)
+            
+            # Look for about/services pages
+            links = soup.find_all("a", href=True)
+            for link in links[:10]:
+                href = link.get("href", "").lower()
+                if any(keyword in href for keyword in ["about", "service", "what-we-do", "our-story"]):
+                    full_url = urljoin(current_url, link["href"])
+                    if full_url not in visited and url in full_url:
+                        to_visit.append(full_url)
+                        
+        except Exception as e:
+            continue
+    
+    return " ".join(all_content[:2000]), scrape_successful  # Limit content length
+
+def search_business_info(url_or_name):
+    """Search for business information using web search"""
+    try:
+        if not openai_ready:
+            return "Unable to search - OpenAI not configured"
+            
+        domain = urlparse(url_or_name).netloc if url_or_name.startswith('http') else url_or_name
+        
+        response = openai.ChatCompletion.create(
+            model="gpt-4",
+            messages=[
+                {
+                    "role": "system", 
+                    "content": "You are a business research assistant. Based on a website URL or business name, provide information about what the business does, what problems they solve, their target audience, and their unique value proposition. If you don't have specific information, make reasonable inferences based on the domain name or business name."
+                },
+                {
+                    "role": "user", 
+                    "content": f"Research this business: {domain}. Provide information about: 1) What they do, 2) Problems they solve, 3) Target audience, 4) Services/products, 5) What makes them unique. Be specific and practical."
+                }
+            ],
+            temperature=0.3
+        )
+        
+        return response["choices"][0]["message"]["content"]
+    except Exception as e:
+        return f"Unable to research business information: {str(e)}"
+
+def extract_business_info(content):
+    """Extract structured business information from website content or research"""
+    try:
+        if not openai_ready:
+            return {}
+            
+        response = openai.ChatCompletion.create(
+            model="gpt-4",
+            messages=[
+                {
+                    "role": "system",
+                    "content": """You are a business analyst. Extract key business information from website content and return it in a structured format. Extract:
+                    1. Business description (what they do)
+                    2. Problems they solve for customers
+                    3. Main services/products (2-3 key offerings)
+                    4. Unique value proposition
+                    5. Business name
+                    6. Target audience (be specific)
+                    
+                    Return the information in this exact format:
+                    BUSINESS_DESCRIPTION: [description]
+                    PROBLEM_SOLVED: [problem]
+                    SERVICES: [services]
+                    UNIQUE_VALUE: [unique aspects]
+                    BUSINESS_NAME: [name]
+                    TARGET_AUDIENCE: [audience]
+                    
+                    If any information is unclear, make reasonable inferences based on available context."""
+                },
+                {
+                    "role": "user",
+                    "content": f"Extract business information from this content: {content[:1500]}"  # Limit content
+                }
+            ],
+            temperature=0.3
+        )
+        
+        result = response["choices"][0]["message"]["content"]
+        
+        # Parse the structured response
+        info = {}
+        lines = result.split('\n')
+        for line in lines:
+            if ':' in line:
+                key, value = line.split(':', 1)
+                key = key.strip()
+                value = value.strip()
+                
+                if key == "BUSINESS_DESCRIPTION":
+                    info['business'] = value
+                elif key == "PROBLEM_SOLVED":
+                    info['problem'] = value
+                elif key == "SERVICES":
+                    info['services'] = value
+                elif key == "UNIQUE_VALUE":
+                    info['unique'] = value
+                elif key == "BUSINESS_NAME":
+                    info['name'] = value
+                elif key == "TARGET_AUDIENCE":
+                    info['audience'] = value
+        
+        return info
+    except Exception as e:
+        return {}
+
+def analyze_website(url):
+    """Main function to analyze website and extract business information"""
+    if not url:
+        return {}
+    
+    # First, try to scrape the website
+    content, scrape_successful = scrape_website(url)
+    
+    if not scrape_successful or len(content) < 100:
+        # If scraping failed, try searching for business information
+        content = search_business_info(url)
+    
+    # Extract structured information
+    business_info = extract_business_info(content)
+    return business_info
 
 # Function to generate PDF
 def create_pdf(script_content, business_name, platform, video_type):
@@ -151,7 +323,7 @@ def create_pdf(script_content, business_name, platform, video_type):
 
 # Function to generate script with enhanced prompting
 def generate_video_script(prompt_data):
-    if not client:
+    if not openai_ready:
         return None
     
     try:
@@ -211,7 +383,7 @@ def generate_video_script(prompt_data):
         Make it authentic, conversational, and optimized for {prompt_data['platform']} consumption patterns.
         """
         
-        response = client.chat.completions.create(
+        response = openai.ChatCompletion.create(
             model="gpt-4",
             messages=[
                 {"role": "system", "content": system_message},
@@ -220,7 +392,7 @@ def generate_video_script(prompt_data):
             temperature=0.8
         )
         
-        return response.choices[0].message.content
+        return response["choices"][0]["message"]["content"]
     
     except Exception as e:
         st.error(f"Error generating script: {str(e)}")
@@ -270,6 +442,10 @@ if 'show_form' not in st.session_state:
     st.session_state.show_form = True
 if 'script_metadata' not in st.session_state:
     st.session_state.script_metadata = {}
+if 'website_analyzed' not in st.session_state:
+    st.session_state.website_analyzed = False
+if 'extracted_info' not in st.session_state:
+    st.session_state.extracted_info = {}
 
 # Header
 st.markdown('<h1 class="main-header">🎬 AI Video Script Generator Pro</h1>', unsafe_allow_html=True)
@@ -288,55 +464,117 @@ st.markdown("""
 """, unsafe_allow_html=True)
 
 if st.session_state.show_form:
-    # Input form in columns for better layout
-    col1, col2 = st.columns([2, 1])
+    # Website URL Analysis Section
+    st.markdown('<h2 class="section-header">🌐 Quick Business Analysis</h2>', unsafe_allow_html=True)
     
+    col1, col2 = st.columns([3, 1])
     with col1:
-        st.markdown('<h2 class="section-header">📝 Tell Us About Your Business</h2>', unsafe_allow_html=True)
-        
-        # Basic business info
-        business_description = st.text_area(
-            "What does your business do?",
-            placeholder="e.g., We're a local coffee shop that roasts our own beans and creates custom blends for coffee lovers.",
-            height=80,
-            help="Describe your business in simple terms that anyone can understand"
+        website_url = st.text_input(
+            "Enter your website URL (optional - we'll auto-fill details):",
+            placeholder="e.g., https://yourwebsite.com or yourwebsite.com",
+            help="We'll analyze your website to pre-fill business information"
         )
-        
+    
+    with col2:
+        st.markdown("<br>", unsafe_allow_html=True)  # Add spacing
+        if st.button("🔍 Analyze Website", use_container_width=True):
+            if website_url:
+                with st.spinner("🔍 Analyzing your website..."):
+                    extracted_info = analyze_website(website_url)
+                    st.session_state.extracted_info = extracted_info
+                    st.session_state.website_analyzed = True
+                    if extracted_info:
+                        st.success("✅ Website analyzed! Information has been pre-filled below.")
+                    else:
+                        st.warning("⚠️ Could not extract all information. Please fill in the details manually.")
+                    st.rerun()
+
+    # Input form - consolidated to 8 key inputs
+    st.markdown('<h2 class="section-header">📝 Video Details (8 Quick Steps)</h2>', unsafe_allow_html=True)
+    
+    # Get pre-filled values from website analysis
+    extracted = st.session_state.extracted_info
+    
+    # Step 1: Business Description
+    business_description = st.text_area(
+        "1️⃣ What does your business do?",
+        value=extracted.get('business', ''),
+        placeholder="e.g., We're a local coffee shop that roasts our own beans and creates custom blends for coffee lovers.",
+        height=80,
+        help="Describe your business in simple terms"
+    )
+    
+    # Step 2: Problem + Solution Combined
+    col1, col2 = st.columns(2)
+    with col1:
         problem_solved = st.text_input(
-            "What specific problem do you solve for customers?",
-            placeholder="e.g., We help busy professionals get amazing coffee without the wait",
-            help="Focus on the main pain point you address - be specific"
+            "2️⃣ What problem do you solve?",
+            value=extracted.get('problem', ''),
+            placeholder="e.g., Busy professionals need great coffee fast",
+            help="The main pain point you address"
         )
-        
+    
+    with col2:
         services_offered = st.text_input(
-            "What services/products should we highlight?",
-            placeholder="e.g., Custom coffee subscriptions, corporate catering, barista training",
-            help="List 2-3 key offerings you want to promote"
-        )
-        
-        unique_aspects = st.text_input(
-            "What makes you different from competitors?",
-            placeholder="e.g., Only coffee shop in town with beans roasted daily on-site",
-            help="Your unique selling proposition that sets you apart"
+            "3️⃣ Main services/products?",
+            value=extracted.get('services', ''),
+            placeholder="e.g., Custom blends, catering, subscriptions",
+            help="2-3 key offerings to highlight"
         )
-        
+    
+    # Step 3: Business Name + Unique Value
+    col1, col2 = st.columns(2)
+    with col1:
         name_to_include = st.text_input(
-            "Business name to include in the script:",
+            "4️⃣ Business name:",
+            value=extracted.get('name', ''),
             placeholder="e.g., Downtown Coffee Co.",
-            help="How you want your business referenced"
+            help="How to reference your business"
+        )
+    
+    with col2:
+        unique_aspects = st.text_input(
+            "5️⃣ What makes you unique?",
+            value=extracted.get('unique', ''),
+            placeholder="e.g., Only shop with daily on-site roasting",
+            help="Your key differentiator"
         )
+    
+    # Step 4: Target + Platform Combined
+    col1, col2 = st.columns(2)
+    with col1:
+        # Determine target audience from extracted info or provide options
+        audience_options = [
+            "Local Community",
+            "Young Professionals (25-35)",
+            "Small Business Owners",
+            "Families with Children",
+            "Seniors (55+)",
+            "Students",
+            "Health-Conscious Consumers",
+            "Industry Professionals",
+            "Budget-Conscious Shoppers",
+            "Luxury Market"
+        ]
+        
+        # Try to match extracted audience to options
+        default_audience = 0
+        extracted_audience = extracted.get('audience', '').lower()
+        for i, option in enumerate(audience_options):
+            if any(keyword in extracted_audience for keyword in option.lower().split()):
+                default_audience = i
+                break
         
-        call_to_action = st.text_input(
-            "What specific action should viewers take?",
-            placeholder="e.g., Visit us at 123 Main Street or order online at our website",
-            help="One clear, specific action you want viewers to take"
+        target_audience = st.selectbox(
+            "6️⃣ Target audience:",
+            audience_options,
+            index=default_audience,
+            help="Who are you trying to reach?"
         )
     
     with col2:
-        st.markdown('<h2 class="section-header">🎯 Platform & Style Optimization</h2>', unsafe_allow_html=True)
-        
         platform = st.selectbox(
-            "Primary Platform:",
+            "7️⃣ Primary platform:",
             [
                 "Instagram Reels",
                 "TikTok", 
@@ -346,162 +584,134 @@ if st.session_state.show_form:
                 "Twitter/X",
                 "Email Marketing"
             ],
-            help="Choose your main distribution platform for optimization"
-        )
-        
-        # Platform-specific tips
-        platform_tips = {
-            "Instagram Reels": "📱 Vertical format essential • 15-90 seconds optimal • Use trending audio • Captions required",
-            "TikTok": "🎵 31-60 seconds works best • Completion rate crucial • Native feel important • Trending sounds boost reach",
-            "YouTube Shorts": "📺 Up to 3 minutes allowed • Educational content performs well • Mobile-first viewing",
-            "LinkedIn": "💼 Professional tone essential • 30 seconds-2 minutes • Educational/business value focus",
-            "Facebook": "👥 Native uploads preferred • 74% watch without sound • Captions essential",
-            "Twitter/X": "⚡ 15 seconds optimal • Real-time relevance • Concise messaging",
-            "Email Marketing": "📧 1-2 minutes max • Clear thumbnail • Value-focused preview"
-        }
-        
-        st.markdown(f"""
-        <div class="platform-tips">
-        <strong>Platform Tips:</strong><br>
-        {platform_tips.get(platform, "General optimization tips")}
-        </div>
-        """, unsafe_allow_html=True)
-        
-        video_type = st.selectbox(
-            "Video Type:",
-            [
-                "Product/Service Showcase",
-                "Problem-Solution Focused", 
-                "Educational/How-To",
-                "Behind-the-Scenes",
-                "Customer Success Story",
-                "Brand Awareness",
-                "Promotional Offer",
-                "Comparison/Competition"
-            ],
-            help="Choose the primary purpose of your video"
-        )
-        
-        target_audience = st.selectbox(
-            "Target Audience:",
-            [
-                "Local Community",
-                "Young Professionals (25-35)",
-                "Small Business Owners",
-                "Families with Children",
-                "Seniors (55+)",
-                "Students",
-                "Health-Conscious Consumers",
-                "Luxury Market",
-                "Budget-Conscious Shoppers",
-                "Industry Professionals"
-            ],
-            help="Who are you trying to reach?"
+            help="Where will you post this video?"
         )
+    
+    # Step 5: Call to Action
+    call_to_action = st.text_input(
+        "8️⃣ What should viewers do next?",
+        placeholder="e.g., Visit us at 123 Main Street or order online",
+        help="One clear, specific action"
+    )
+    
+    # Advanced Options (Collapsible)
+    with st.expander("🎯 Advanced Options (Optional)"):
+        col1, col2, col3 = st.columns(3)
         
-        # Advanced options
-        st.markdown('<h3 style="color: #1f4e79; margin-top: 1.5rem;">🧠 Psychology & Storytelling</h3>', unsafe_allow_html=True)
-        
-        hook_type = st.selectbox(
-            "Hook Strategy:",
-            [
-                "Question Hook",
-                "Problem Statement",
-                "Curiosity Gap",
-                "Direct Call-out",
-                "Movement Hook",
-                "Negative Hook",
-                "Statistical Hook",
-                "Story Hook"
-            ],
-            help="How do you want to grab attention in the first 3-5 seconds?"
-        )
-        
-        storytelling_framework = st.selectbox(
-            "Storytelling Framework:",
-            [
-                "Problem-Solution",
-                "Before-After",
-                "Hero's Journey",
-                "Educational Value",
-                "Testimonial Story",
-                "Behind-the-Scenes",
-                "Pixar Framework"
-            ],
-            help="Choose your narrative structure"
-        )
-        
-        psychological_triggers = st.multiselect(
-            "Psychological Triggers:",
-            [
-                "Social Proof (testimonials, statistics)",
-                "Scarcity/FOMO (limited time, exclusive)",
-                "Authority (expert endorsement)",
-                "Reciprocity (free value first)",
-                "Curiosity (information gaps)",
-                "Empathy (shared experiences)",
-                "Urgency (immediate action needed)"
-            ],
-            default=["Social Proof (testimonials, statistics)", "Curiosity (information gaps)"],
-            help="Select psychological elements to include"
-        )
-        
-        tone_options = [
-            "Professional & Trustworthy",
-            "Friendly & Conversational", 
-            "Exciting & Energetic",
-            "Warm & Personal",
-            "Confident & Bold",
-            "Educational & Expert",
-            "Inspiring & Motivational",
-            "Urgent & Persuasive",
-            "Authentic & Real"
-        ]
-        
-        tone = st.selectbox("Script Tone:", tone_options, help="How do you want to sound?")
-        
-        script_length = st.slider(
-            "Script Length (words):",
-            min_value=50,
-            max_value=300,
-            value=150,
-            step=25,
-            help="Shorter = social media, Longer = detailed explanations"
-        )
+        with col1:
+            video_type = st.selectbox(
+                "Video Type:",
+                [
+                    "Product/Service Showcase",
+                    "Problem-Solution Focused", 
+                    "Educational/How-To",
+                    "Behind-the-Scenes",
+                    "Customer Success Story",
+                    "Brand Awareness",
+                    "Promotional Offer"
+                ]
+            )
+            
+            tone = st.selectbox(
+                "Tone:",
+                [
+                    "Friendly & Conversational", 
+                    "Professional & Trustworthy",
+                    "Exciting & Energetic",
+                    "Warm & Personal",
+                    "Educational & Expert",
+                    "Authentic & Real"
+                ]
+            )
         
-        # Length guide based on platform
-        length_guides = {
-            "Instagram Reels": "75-150 words (15-90 seconds)",
-            "TikTok": "75-125 words (31-60 seconds)", 
-            "YouTube Shorts": "100-200 words (up to 3 minutes)",
-            "LinkedIn": "100-175 words (30 seconds-2 minutes)",
-            "Facebook": "100-200 words (1-3 minutes)",
-            "Twitter/X": "50-75 words (15 seconds optimal)",
-            "Email Marketing": "150-300 words (1-2 minutes)"
-        }
+        with col2:
+            hook_type = st.selectbox(
+                "Hook Strategy:",
+                [
+                    "Question Hook",
+                    "Problem Statement",
+                    "Curiosity Gap",
+                    "Direct Call-out",
+                    "Statistical Hook",
+                    "Story Hook"
+                ]
+            )
+            
+            storytelling_framework = st.selectbox(
+                "Story Structure:",
+                [
+                    "Problem-Solution",
+                    "Before-After",
+                    "Educational Value",
+                    "Testimonial Story",
+                    "Behind-the-Scenes"
+                ]
+            )
         
-        st.markdown(f"""
-        <div class="feature-box">
-        <strong>📏 Recommended for {platform}:</strong><br>
-        {length_guides.get(platform, "100-200 words for general use")}
-        </div>
-        """, unsafe_allow_html=True)
+        with col3:
+            psychological_triggers = st.multiselect(
+                "Psychological Triggers:",
+                [
+                    "Social Proof",
+                    "Scarcity/FOMO",
+                    "Authority",
+                    "Curiosity",
+                    "Empathy"
+                ],
+                default=["Social Proof", "Curiosity"]
+            )
+            
+            script_length = st.slider(
+                "Length (words):",
+                min_value=50,
+                max_value=300,
+                value=150,
+                step=25
+            )
+    
+    # Platform-specific tips
+    platform_tips = {
+        "Instagram Reels": "📱 Vertical format • 15-90 seconds • Use trending audio • Captions required",
+        "TikTok": "🎵 31-60 seconds optimal • Completion rate crucial • Native feel important",
+        "YouTube Shorts": "📺 Up to 3 minutes • Educational content performs well • Mobile-first",
+        "LinkedIn": "💼 Professional tone • 30 seconds-2 minutes • Business value focus",
+        "Facebook": "👥 Native uploads preferred • 74% watch without sound • Captions essential",
+        "Twitter/X": "⚡ 15 seconds optimal • Real-time relevance • Concise messaging",
+        "Email Marketing": "📧 1-2 minutes max • Clear thumbnail • Value-focused preview"
+    }
+    
+    st.markdown(f"""
+    <div class="platform-tips">
+    <strong>📋 {platform} Tips:</strong> {platform_tips.get(platform, "General optimization")}
+    </div>
+    """, unsafe_allow_html=True)
 
     # Generate button
     st.markdown("<br>", unsafe_allow_html=True)
     col1, col2, col3 = st.columns([1, 2, 1])
     with col2:
         if st.button("🚀 Generate My Optimized Video Script", use_container_width=True):
-            # Validation
-            required_fields = [
-                business_description, problem_solved, services_offered,
-                unique_aspects, name_to_include, call_to_action
-            ]
+            # Validation - only require essential fields
+            required_fields = [business_description, problem_solved, services_offered, name_to_include, call_to_action]
             
             if not all(required_fields):
-                st.error("⚠️ Please fill in all required fields to generate your script.")
+                st.error("⚠️ Please fill in the required fields (1-5 and 8) to generate your script.")
             else:
                 with st.spinner("🎬 Crafting your conversion-optimized video script..."):
+                    # Set defaults for advanced options if not specified
+                    if 'video_type' not in locals():
+                        video_type = "Product/Service Showcase"
+                    if 'tone' not in locals():
+                        tone = "Friendly & Conversational"
+                    if 'hook_type' not in locals():
+                        hook_type = "Question Hook"
+                    if 'storytelling_framework' not in locals():
+                        storytelling_framework = "Problem-Solution"
+                    if 'psychological_triggers' not in locals():
+                        psychological_triggers = ["Social Proof", "Curiosity"]
+                    if 'script_length' not in locals():
+                        script_length = 150
+                    
                     prompt_data = {
                         'business': business_description,
                         'problem': problem_solved,
@@ -516,7 +726,7 @@ if st.session_state.show_form:
                         'audience': target_audience,
                         'hook_type': hook_type,
                         'storytelling': storytelling_framework,
-                        'psychological_triggers': ", ".join(psychological_triggers)
+                        'psychological_triggers': ", ".join(psychological_triggers) if isinstance(psychological_triggers, list) else psychological_triggers
                     }
                     
                     script = generate_video_script(prompt_data)