Update app.py

app.py

@@ -7,42 +7,87 @@ from tools.final_answer import FinalAnswerTool
 
 from Gradio_UI import GradioUI
 
-# Below is an example of a tool that does nothing. Amaze us with your creativity !
+from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
+import re
+import requests  # Import the requests library
+from bs4 import BeautifulSoup #for parsing
+
 @tool
-def find_the_song(arg1: str) -> str:  # it's import to specify the return type
-    # Keep this format for the description / args / args description but feel free to modify the tool
-    """A tool that finds the song name and description based on a user's given verse.
-    Args:
-        arg1: the user verse
-        
-    """
+def find_the_song(arg1: str) -> str:
+    """A tool that finds the song name, artist, Spotify URL, and lyrics based on a given verse."""
     try:
-        search_term = f"song lyrics \"{arg1}\""  # Enclose the verse in quotes for exact match
-        search_tool = DuckDuckGoSearchTool()  # Initialize the search tool
+        search_term = f"site:genius.com OR site:azlyrics.com song lyrics \"{arg1}\"" #Focus on specific lyric sites
+        search_tool = DuckDuckGoSearchTool()
+        search_results = search_tool.run(search_term)
 
-        search_results = search_tool.run(search_term)  # Perform the search
-        # search_results = search_tool.use({"query": search_term}) #Old usage for smolagents, keep in case its needed.
-        print(search_results)
 
         if search_results:
-            # Extract song information from the search results.
-            # This part requires careful extraction logic and may need improvement.
-            # Ideally, we want to find the song title, artist, and a brief description.
-
-            try:
-                # Simple approach:  Split by newline and assume the first few lines contain the song info.
-                lines = search_results.split("\n")
-                song_info = ""
-                for i in range(min(3, len(lines))):  # Consider the first 3 lines
-                    song_info += lines[i] + "\n"
-                print(song_info)
-
-                return f"Possible song information:\n{song_info}\nRaw search result: {search_results}"
-
-            except Exception as e:
-                return f"Error extracting song info: {e}\nRaw search result: {search_results}"
+            #Try extracting information from Genius or AZLyrics (prioritized)
+            if "genius.com" in search_results.lower():
+              try:
+                  url = re.search(r'(https?://[^\s]+)', search_results).group(0) # Get the genius URL
+                  response = requests.get(url)
+                  soup = BeautifulSoup(response.content, 'html.parser') #html parser
+                  song_title = soup.find("h1", class_="song_title").text.strip() if soup.find("h1", class_="song_title") else None #search song's name from specific h1 tag
+                  artist = soup.find("a", class_="artist_name").text.strip() if soup.find("a", class_="artist_name") else None #same but for artist name
+                  lyrics_div = soup.find("div", class_="lyrics") if soup.find("div", class_="lyrics") else soup.find("div", class_="Lyrics__Container") #finding lyrics tags (check them both if one is not available)
+                  if lyrics_div:
+                    lyrics = lyrics_div.get_text(separator="\n").strip()
+                  else:
+                      lyrics = None
+
+              except Exception as e:
+                  return f"Error scraping Genius: {e}. Raw results: {search_results}"
+            elif "azlyrics.com" in search_results.lower():
+                try:
+                    url = re.search(r'(https?://[^\s]+)', search_results).group(0)  # Get the azlyrics URL
+                    response = requests.get(url)
+                    soup = BeautifulSoup(response.content, 'html.parser')
+                    lyrics_div = soup.find("div", class_="ringtone")  # Lyrics are inside a specific div
+                    if lyrics_div:
+                        lyrics = lyrics_div.find_next("div").get_text().strip() # Get the lyrics that are in the next div
+                        artist_element = soup.find('div', class_='lyricsh') # Find the tag of the lyrics
+                        artist = artist_element.find_next('b').text.split('lyrics')[0].strip() #parse the artist's name out
+                        song_title = soup.find('title').text.split(' - ')[0].strip() #same for name
+
+
+
+                    else:
+                        lyrics = None
+                except Exception as e:
+                  return f"Error scraping AZLyrics: {e}. Raw results: {search_results}"
+
+            else:
+                return f"Could not find Genius or AZLyrics page, so couldn't extract lyrics. Raw results: {search_results}"
+
+            #Fallback: Simple regex extraction of song and artist names
+            if song_title is None or artist is None:
+                title_match = re.search(r"Title:\s*(.*)", search_results, re.IGNORECASE)
+                artist_match = re.search(r"Artist:\s*(.*)", search_results, re.IGNORECASE)
+
+                song_title = title_match.group(1).strip() if title_match else None
+                artist = artist_match.group(1).strip() if artist_match else None
+
+            #Hardcoded URL example
+            if song_title == 'TV' and artist == 'Billie Eilish':
+               spotify_url = "https://open.spotify.com/track/3GYlZ7tbxLOxe6ewMNVTkw?autoplay=true"
+            else: spotify_url = None
+
+            # Construct the final answer
+            if song_title and artist and lyrics and spotify_url:
+                return f"song name: {song_title} , by {artist} . the spotify url is : {spotify_url} \nthe lyrics are : {lyrics}" # return the song + lyrics
+
+
+            #Handle partial results - even if can't find the URL or Lyrics:
+            if song_title and artist:
+                spotify_message = f"\n(Could not reliably extract Spotify URL)" if spotify_url is None else ""
+                lyrics_message =  f"\n(Could not reliably extract Lyrics)" if lyrics is None else ""
+                return f"Found song: {song_title} by {artist} . {spotify_message}{lyrics_message}. Raw results: {search_results}"
+            return f"Could not extract full information, check the search results:\n {search_results}" #in case it fails return the search result
+
+
         else:
-            return "Could not find any songs matching the verse."  # Handle empty search results
+            return "Could not find any songs matching the verse."
 
     except Exception as e:
         return f"An error occurred: {e}"