Update app.py

app.py

@@ -2,6 +2,7 @@ import streamlit as st
 import requests
 from bs4 import BeautifulSoup
 from datetime import datetime
+import re
 
 IPL_TEAMS = [
     "Chennai Super Kings", "Delhi Capitals", "Gujarat Titans", "Kolkata Knight Riders",
@@ -13,53 +14,97 @@ BASE_URL = "https://www.cricbuzz.com"
 
 # Dynamically get the current IPL year
 current_year = datetime.now().year
-print(f"📅 [INFO] Current year detected: {current_year}")
-
-# Detect team
-def detect_team(user_input):
-    print(f"\n🟢 [DETECT] Input: {user_input}")
-    for team in IPL_TEAMS:
-        if team.lower() in user_input.lower():
-            print(f"✅ [DETECT] Found team: {team}")
-            return team
-    print("❌ [DETECT] No IPL team found")
-    return None
-
-# Get live matches from cricbuzz homepage
+print(f"📅 Current year detected: {current_year}")
+
+
 def get_live_scores(team_name=None):
-    print("\n🌐 [LIVE] Fetching live scores page...")
+    print("🌐 [LIVE] Fetching live scores page...")
     url = "https://www.cricbuzz.com/cricket-match/live-scores"
     headers = {"User-Agent": "Mozilla/5.0"}
     response = requests.get(url, headers=headers)
-    print(f"✅ [LIVE] Response code: {response.status_code}")
+    print(f"✅ [LIVE] Response status code: {response.status_code}")
 
     soup = BeautifulSoup(response.text, "html.parser")
-    matches = soup.find_all("div", class_="cb-col cb-col-100 cb-ltst-wgt-hdr")
-    print(f"🔍 [LIVE] Found {len(matches)} match blocks")
+    pattern = re.compile(fr"/live-cricket-scores/.+-match-indian-premier-league-{current_year}")
+    matches = [a['href'] for a in soup.find_all('a', href=True) if pattern.search(a['href'])]
+    print(f"🔍 [LIVE] Matches found: {matches}")
+
+    pattern_4_score = re.compile(r'(\d+)(?:th|st|nd|rd)-match')
+    final_matches = sorted(
+        [(link, int(pattern_4_score.search(link).group(1))) for link in matches if pattern_4_score.search(link)],
+        key=lambda x: x[1],
+        reverse=True
+    )
+    print(f"🔍 [LIVE] Final matches sorted: {final_matches}")
+
+    if not final_matches:
+        print("❌ [LIVE] No match blocks found. Check HTML structure or class names.")
+        return []
 
     results = []
-    for block in matches:
-        text = block.get_text(separator=" ").strip()
-        if team_name is None or team_name.lower() in text.lower():
-            results.append("🟢 LIVE: " + text)
-            print(f"✅ [LIVE MATCH] {text[:100]}...")
+    seen_matches = set()
+
+    for match_url, _ in final_matches:
+        print(f"🌐 [LIVE] Fetching match details from: {BASE_URL + match_url}")
+        match_response = requests.get(BASE_URL + match_url, headers=headers)
+        soup = BeautifulSoup(match_response.text, "html.parser")
+
+        score_blocks = soup.select("div.cb-col.cb-col-100.cb-min-tm")
+        rr_score = score_blocks[0].text.strip() if len(score_blocks) > 0 else "N/A"
+        csk_score = score_blocks[1].text.strip() if len(score_blocks) > 1 else "N/A"
+        print(f"🔍 [LIVE] Scores: {rr_score}, {csk_score}")
+
+        result_block = soup.select_one("div.cb-min-stts")
+        result_text = result_block.text.strip() if result_block else "Result unavailable"
+        print(f"🔍 [LIVE] Result: {result_text}")
+
+        mom_block = soup.select_one("div.cb-mom-itm a.cb-link-undrln")
+        mom_text = mom_block.text.strip() if mom_block else "Unknown"
+        print(f"🔍 [LIVE] Player of the Match: {mom_text}")
+
+        commentary_blocks = soup.find_all("p", class_="cb-com-ln", limit=3)
+        commentary = [c.get_text(strip=True) for c in commentary_blocks if c.get_text(strip=True)]
+        commentary_text = " ".join(commentary)
+        print(f"🔍 [LIVE] Commentary: {commentary_text}")
+
+        # Skip match if no useful data
+        if rr_score == "N/A" and csk_score == "N/A" and result_text == "Result unavailable":
+            print("❌ [LIVE] Skipping match due to lack of useful data.")
+            continue
+
+        # Create a simple ID based on team scores and result
+        match_signature = f"{rr_score}|{csk_score}|{result_text}"
+        if match_signature in seen_matches:
+            print("❌ [LIVE] Skipping duplicate match.")
+            continue  # Skip duplicate
+        seen_matches.add(match_signature)
+
+        results.append({
+            "1st Team Score": rr_score,
+            "2nd Team Score": csk_score,
+            "Result": result_text,
+            "Player of the Match": mom_text,
+            "Latest Commentary": commentary_text[:100]
+        })
+        print(f"✅ [LIVE] Match added: {results[-1]}")
+
+    return results[:2] if results else []
 
-    return results[:5] if results else []
 
-# Get recent results if no live match
 def get_recent_results(team_name=None):
     try:
-        url = f"{BASE_URL}/cricket-series/7607/indian-premier-league-{current_year}/matches"
-        print(f"\n🌐 [RECENT] Fetching matches from: {url}")
+        url = f"{BASE_URL}/cricket-series/9237/Indian-Premier-League-{current_year}/matches"
+        print(f"🌐 [RECENT] Fetching matches from: {url}")
         headers = {"User-Agent": "Mozilla/5.0"}
         response = requests.get(url, headers=headers)
-        print(f"✅ [RECENT] Status code: {response.status_code}")
+        print(f"✅ [RECENT] Response status code: {response.status_code}")
 
         soup = BeautifulSoup(response.text, "html.parser")
         all_tags = soup.find_all("a", href=True)
-        print(f"🔍 [RECENT] Total <a> tags: {len(all_tags)}")
+        print(f"🔍 [RECENT] Total <a> tags found: {len(all_tags)}")
 
-        won, lost = [], []
+        results = []
+        seen = set()
 
         for i in range(len(all_tags) - 1):
             tag = all_tags[i]
@@ -68,57 +113,51 @@ def get_recent_results(team_name=None):
             href = tag['href']
             title = tag.get_text().strip()
             result = next_tag.get_text().strip()
+            print(f"🔍 [RECENT] Processing tag: {title} – {result}")
 
             if "cricket-scores" in href and any(t.lower() in title.lower() for t in IPL_TEAMS):
-                full_url = BASE_URL + href
-                full_text = f"{title} – {result}\n[View Match]({full_url})"
-
                 if team_name and team_name.lower() not in title.lower():
+                    print(f"❌ [RECENT] Skipping match for team: {team_name}")
                     continue
 
-                if "won by" in result.lower():
-                    print(f"✅ [WON] {title} => {result}")
-                    won.append(full_text)
-                elif "lost" in result.lower():
-                    print(f"🟥 [LOST] {title} => {result}")
-                    lost.append(full_text)
+                match_id = f"{title}-{result}"
+                if match_id in seen:
+                    print(f"❌ [RECENT] Skipping duplicate match: {match_id}")
+                    continue
+                seen.add(match_id)
 
-        top_results = won[:5] + lost[:5]
-        print(f"📊 [RECENT] Total displayed: {len(top_results)}")
-        return top_results if top_results else ["No recent match results found."]
+                full_url = BASE_URL + href
+                full_text = f"{title} – {result})"
+                print(f"✅ [RECENT] Match added: {full_text}")
+
+                results.append(full_text)
+
+        # Show up to 10 most recent results
+        print(f"📊 [RECENT] Total displayed results: {len(results)}")
+        return results[:10] if results else ["No recent match results found."]
 
     except Exception as e:
-        print(f"❌ [ERROR] Failed recent match fetch: {e}")
+        print(f"❌ [ERROR] Failed to fetch recent matches: {e}")
         return ["Error loading past matches."]
 
+
 # Streamlit UI
 st.set_page_config(page_title="IPL Chatbot", layout="centered")
 st.title("🏏 IPL Chatbot")
 
-user_input = st.text_input("Ask about IPL scores:", placeholder="e.g. Show me Mumbai Indians score")
-
-if user_input:
-    print(f"\n========== USER INPUT ==========")
-    print(user_input)
-    print("================================")
-
-    team = detect_team(user_input)
-
-    scores = get_live_scores(team)
-    if scores:
-        st.subheader("🟢 Live Match(es):")
-        for s in scores:
-            st.markdown(s)
-    else:
-        st.warning("❌ No live match found. Showing recent results:")
-        past = get_recent_results(team)
-        for p in past:
-            st.markdown(p)
-
-with st.expander("💡 Try examples"):
-    st.markdown("""
-- Show Mumbai Indians score  
-- Delhi Capitals recent results  
-- CSK won?  
-- RCB vs MI today?  
-    """)
+# Live Matches
+scores = get_live_scores()
+if scores:
+    st.subheader("🟢 Live Match:")
+    st.table(scores)
+else:
+    st.warning("❌ No live match found.")
+
+# Recent Results
+recent_results = get_recent_results()
+if recent_results:
+    st.subheader("📊 Recent Results:")
+    for result in recent_results:
+        st.markdown(result)
+else:
+    st.warning("❌ No recent results found.")