Create app.py

app.py

@@ -0,0 +1,218 @@
+import gradio as gr
+import json
+import plotly.express as px
+import pandas as pd
+from groq import Groq
+from fpdf import FPDF
+from youtube_comment_downloader import YoutubeCommentDownloader
+import re
+import os
+import warnings
+
+warnings.filterwarnings("ignore")
+
+# ====================== CONFIG ======================
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+
+# ====================== SYSTEM PROMPT ======================
+SYSTEM_PROMPT = """
+You are an expert social media sentiment and poll analysis AI.
+Focus on Yes/No, Agree/Disagree, Support/Oppose, and sentiment.
+
+Handle English + Urdu well.
+Return ONLY valid JSON in this exact format:
+{
+  "main_poll": {
+    "question": "Suggested poll question",
+    "yes_count": int,
+    "no_count": int,
+    "agree_count": int,
+    "disagree_count": int,
+    "support_count": int,
+    "oppose_count": int,
+    "neutral_count": int
+  },
+  "sentiment": {
+    "positive": float,
+    "negative": float,
+    "neutral": float
+  },
+  "top_themes": ["theme1", "theme2", ...],
+  "summary": "Short professional summary",
+  "labeled_comments": [
+    {"comment": "...", "opinion": "Yes|No|Agree|Disagree|Positive|Negative|Neutral|Mixed"}
+  ]
+}
+"""
+
+# ====================== HELPER FUNCTIONS ======================
+def extract_youtube_id(url):
+    patterns = [
+        r'youtu\.be/([a-zA-Z0-9_-]+)',
+        r'v=([a-zA-Z0-9_-]+)',
+        r'/embed/([a-zA-Z0-9_-]+)',
+        r'/shorts/([a-zA-Z0-9_-]+)'
+    ]
+    for pattern in patterns:
+        match = re.search(pattern, url)
+        if match:
+            return match.group(1)
+    return None
+
+def clean_youtube_url(url):
+    video_id = extract_youtube_id(url)
+    if video_id:
+        return f"https://www.youtube.com/watch?v={video_id}"
+    return url
+
+def fetch_youtube_comments(url, limit=150):
+    try:
+        clean_url = clean_youtube_url(url)
+        downloader = YoutubeCommentDownloader()
+        comments = []
+        for comment in downloader.get_comments_from_url(clean_url, sort_by=0):
+            comments.append(comment['text'])
+            if len(comments) >= limit:
+                break
+        return comments
+    except Exception as e:
+        print(f"Fetch error: {e}")
+        return []
+
+def analyze_comments_with_groq(comments, post_context=""):
+    try:
+        client = Groq(api_key=GROQ_API_KEY)
+        comments_text = "\n\n".join([f"Comment {i+1}: {c[:250]}" for i, c in enumerate(comments[:130])])
+        
+        user_prompt = f"""
+Post Context: {post_context}
+Analyze these comments and return proper JSON:
+{comments_text}
+"""
+
+        response = client.chat.completions.create(
+            model="llama-3.3-70b-versatile",
+            messages=[
+                {"role": "system", "content": SYSTEM_PROMPT},
+                {"role": "user", "content": user_prompt}
+            ],
+            temperature=0.3,
+            max_tokens=4000,
+            response_format={"type": "json_object"}
+        )
+        return json.loads(response.choices[0].message.content)
+    except Exception as e:
+        print("Groq Error:", str(e))
+        return None
+
+def create_pdf_report(analysis_result, poll_question):
+    pdf = FPDF()
+    pdf.add_page()
+    pdf.set_font('Arial', 'B', 16)
+    pdf.cell(0, 10, 'CommentSurvey AI Report', 0, 1, 'C')
+    pdf.ln(10)
+
+    pdf.set_font('Arial', 'B', 12)
+    pdf.cell(0, 10, f'Poll Question: {poll_question}', 0, 1, 'L')
+    pdf.ln(5)
+
+    pdf.set_font('Arial', 'B', 12)
+    pdf.cell(0, 10, 'Summary:', 0, 1, 'L')
+    pdf.set_font('Arial', '', 11)
+    pdf.multi_cell(0, 6, analysis_result.get('summary', 'No summary available.'))
+    pdf.ln(10)
+
+    pdf.set_font('Arial', 'B', 12)
+    pdf.cell(0, 10, 'Top Themes:', 0, 1, 'L')
+    pdf.set_font('Arial', '', 11)
+    for theme in analysis_result.get('top_themes', [])[:8]:
+        pdf.cell(0, 6, f"• {theme}", 0, 1, 'L')
+
+    pdf.output("CommentSurvey_Report.pdf")
+    return "CommentSurvey_Report.pdf"
+
+# ====================== GRADIO INTERFACE ======================
+with gr.Blocks(title="CommentSurvey AI", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 📊 CommentSurvey AI\n**Turn YouTube Comments into Smart Surveys & Insights**")
+
+    with gr.Row():
+        with gr.Column(scale=4):
+            url_input = gr.Textbox(
+                label="🌐 Paste YouTube Video Link",
+                placeholder="https://youtu.be/xxxxxxxxxx",
+                lines=1
+            )
+            analyze_btn = gr.Button("🚀 Analyze Comments", variant="primary", size="large")
+
+        with gr.Column(scale=2):
+            status = gr.Markdown("**Status:** Ready")
+
+    with gr.Tabs():
+        with gr.Tab("📊 Poll Results"):
+            poll_plot = gr.Plot()
+            poll_md = gr.Markdown()
+
+        with gr.Tab("😊 Sentiment Analysis"):
+            sentiment_plot = gr.Plot()
+
+        with gr.Tab("🔍 Key Insights"):
+            insights_md = gr.Markdown()
+
+        with gr.Tab("📜 Raw Comments"):
+            raw_table = gr.Dataframe()
+
+    download_btn = gr.File(label="📥 Download PDF Report")
+
+    def analyze(url):
+        try:
+            if not GROQ_API_KEY:
+                return None, "**❌ Groq API Key is missing!**", None, None, None, None
+
+            comments = fetch_youtube_comments(url)
+            if len(comments) == 0:
+                return None, "**❌ Could not fetch comments.** Try a different public video.", None, None, None, None
+
+            result = analyze_comments_with_groq(comments)
+
+            if not result:
+                return None, "**❌ AI Analysis failed.**", None, None, None, None
+
+            main = result.get('main_poll', {})
+            poll_values = [
+                main.get('yes_count',0) + main.get('agree_count',0) + main.get('support_count',0),
+                main.get('no_count',0) + main.get('disagree_count',0) + main.get('oppose_count',0),
+                main.get('neutral_count',0)
+            ]
+
+            fig_poll = px.pie(
+                names=['Yes/Agree/Support', 'No/Disagree/Oppose', 'Neutral'],
+                values=poll_values,
+                title="Main Poll Results",
+                hole=0.4
+            )
+
+            sent = result.get('sentiment', {})
+            fig_sent = px.bar(
+                x=['Positive', 'Negative', 'Neutral'],
+                y=[sent.get('positive',0), sent.get('negative',0), sent.get('neutral',0)],
+                title="Sentiment Breakdown"
+            )
+
+            raw_df = pd.DataFrame(result.get('labeled_comments', []))
+            pdf_path = create_pdf_report(result, main.get('question', 'Social Media Poll'))
+
+            summary_text = f"**Poll Question:** {main.get('question','N/A')}\n\n**Summary:** {result.get('summary','')}"
+
+            return result, f"**✅ Success!** Analyzed {len(comments)} comments", fig_poll, fig_sent, summary_text, pdf_path
+
+        except Exception as e:
+            return None, f"**❌ Error:** {str(e)}", None, None, None, None
+
+    analyze_btn.click(
+        fn=analyze,
+        inputs=[url_input],
+        outputs=[gr.State(), status, poll_plot, sentiment_plot, insights_md, download_btn]
+    )
+
+if __name__ == "__main__":
+    demo.launch()