Update app.py

app.py

@@ -12,22 +12,6 @@ from huggingface_hub import hf_hub_download, InferenceClient
 import requests
 from bs4 import BeautifulSoup
 import urllib
-import random
-
-# List of user agents to choose from for requests
-_useragent_list = [
-    'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:66.0) Gecko/20100101 Firefox/66.0',
-    'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36',
-    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36',
-    'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/109.0.0.0 Safari/537.36',
-    'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36',
-    'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36 Edg/111.0.1661.62',
-    'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/111.0'
-]
-
-def get_useragent():
-    """Returns a random user agent from the list."""
-    return random.choice(_useragent_list)
 
 def extract_text_from_webpage(html_content):
     """Extracts visible text from HTML content using BeautifulSoup."""
@@ -39,54 +23,58 @@ def extract_text_from_webpage(html_content):
     visible_text = soup.get_text(strip=True)
     return visible_text
 
-# Perform a Google search and return the results
-def search(term, num_results=3, lang="en", advanced=True, timeout=5, safe="active", ssl_verify=None):
+def search(term, num_results=1, lang="en", advanced=True, sleep_interval=0, timeout=5, safe="active", ssl_verify=None):
     """Performs a Google search and returns the results."""
     escaped_term = urllib.parse.quote_plus(term)
     start = 0
     all_results = []
-    # Limit the number of characters from each webpage to stay under the token limit
-    max_chars_per_page = 3000  # Adjust this value based on your token limit and average webpage length
-
-    with requests.Session() as session:
-        while start < num_results:
-            resp = session.get(
-                url="https://www.google.com/search",
-                headers={"User-Agent":get_useragent()},
-                params={
-                    "q": term,
-                    "num": num_results - start,
-                    "hl": lang,
-                    "start": start,
-                    "safe": safe,
-                },
-                timeout=timeout,
-                verify=ssl_verify,
-            )
-            resp.raise_for_status()
-            soup = BeautifulSoup(resp.text, "html.parser")
-            result_block = soup.find_all("div", attrs={"class": "g"})
-            if not result_block:
-                start += 1
-                continue
-            for result in result_block:
-                link = result.find("a", href=True)
-                if link:
-                    link = link["href"]
-                    try:
-                        webpage = session.get(link, headers={"User-Agent": get_useragent()})
-                        webpage.raise_for_status()
-                        visible_text = extract_text_from_webpage(webpage.text)
-                        # Truncate text if it's too long
-                        if len(visible_text) > max_chars_per_page:
-                            visible_text = visible_text[:max_chars_per_page] + "..."
-                        all_results.append({"text": visible_text})
-                    except requests.exceptions.RequestException as e:
-                        print(f"Error fetching or processing {link}: {e}")
-                        all_results.append({"text": None})
-                else:
-                    all_results.append({"text": None})
-            start += len(result_block)
+
+    # Fetch results in batches
+    while start < num_results:
+        resp = requests.get(
+            url="https://www.google.com/search",
+            headers={"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36 Edg/111.0.1661.62"}, # Set random user agent
+            params={
+                "q": term,
+                "num": num_results - start, # Number of results to fetch in this batch
+                "hl": lang,
+                "start": start,
+                "safe": safe,
+            },
+            timeout=timeout,
+            verify=ssl_verify,
+        )
+        resp.raise_for_status() # Raise an exception if request fails
+
+        soup = BeautifulSoup(resp.text, "html.parser")
+        result_block = soup.find_all("div", attrs={"class": "g"})
+
+        # If no results, continue to the next batch
+        if not result_block:
+            start += 1
+            continue
+
+        # Extract link and text from each result
+        for result in result_block:
+            link = result.find("a", href=True)
+            if link:
+                link = link["href"]
+                try:
+                    # Fetch webpage content
+                    webpage = requests.get(link, headers={"User-Agent": get_useragent()})
+                    webpage.raise_for_status()
+                    # Extract visible text from webpage
+                    visible_text = extract_text_from_webpage(webpage.text)
+                    all_results.append({"link": link, "text": visible_text})
+                except requests.exceptions.RequestException as e:
+                    # Handle errors fetching or processing webpage
+                    print(f"Error fetching or processing {link}: {e}")
+                    all_results.append({"link": link, "text": None})
+            else:
+                all_results.append({"link": None, "text": None})
+
+        start += len(result_block) # Update starting index for next batch
+
     return all_results
 
 # Speech Recognition Model Configuration
@@ -100,7 +88,7 @@ tokenizer = spm.SentencePieceProcessor(hf_hub_download(model_name, "tokenizer.sp
 
 # Mistral Model Configuration
 client1 = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
-system_instructions1 = "<s>[SYSTEM] Answer as Real OpenGPT 4o, Made by 'KingNish', Keep conversation very short, clear, friendly and concise. The text provided is a request for a specific type of response from you, the virtual assistant. The request asks you to provide friendly responses. The expectation is that I will avoid introductions and start answering the query directly, elaborating on all aspects of the request. As an AI-powered assistant, your task is to generate responses that appear as if they are created by the user and not an AI-powered assistant. [USER]"
+system_instructions1 = "<s>[SYSTEM] Answer as Real Jarvis JARVIS, Made by 'Tony Stark', Keep conversation very short, clear, friendly and concise. The text provided is a request for a specific type of response from you, the virtual assistant. The request asks you to provide friendly responses as if You are the character Jarvis, made by 'Tony Stark.' The expectation is that I will avoid introductions and start answering the query directly, elaborating on all aspects of the request. As an AI-powered assistant, your task is to generate responses that appear as if they are created by the user and not an AI-powered assistant. [USER]"
 
 def resample(audio_fp32, sr):
     return soxr.resample(audio_fp32, sr, sample_rate)
@@ -132,12 +120,12 @@ def model(text, web_search):
     if web_search is True:
         """Performs a web search, feeds the results to a language model, and returns the answer."""
         web_results = search(text)
-        web2 = ' '.join([f"Text: {res['text']}\n\n" for res in web_results])
-        formatted_prompt = system_instructions1 + text + "[WEB]" + str(web2) + "[OpenGPT 4o]"
+        web2 = ' '.join([f"Link: {res['link']}\nText: {res['text']}\n\n" for res in web_results])
+        formatted_prompt = system_instructions1 + text + "[WEB]" + str(web2) + "[ANSWER]"
         stream = client1.text_generation(formatted_prompt, max_new_tokens=512, stream=True, details=True, return_full_text=False)
         return "".join([response.token.text for response in stream if response.token.text != "</s>"])
     else:
-        formatted_prompt = system_instructions1 + text + "[OpenGPT 4o]"
+        formatted_prompt = system_instructions1 + text + "[JARVIS]"
         stream = client1.text_generation(formatted_prompt, max_new_tokens=512, stream=True, details=True, return_full_text=False)
         return "".join([response.token.text for response in stream if response.token.text != "</s>"])
 
@@ -153,9 +141,9 @@ async def respond(audio, web_search):
 with gr.Blocks() as demo:    
     with gr.Row():
         web_search = gr.Checkbox(label="Web Search", value=False)
-        input = gr.Audio(label="Voice Chat", sources="microphone")
-        output = gr.Audio(label="AI",autoplay=True)
+        input = gr.Audio(label="Voice Chat (BETA)", sources="microphone", type="filepath", waveform_options=False)
+        output = gr.Audio(label="JARVIS", type="filepath", interactive=False, autoplay=True, elem_classes="audio")
         gr.Interface(fn=respond, inputs=[input, web_search], outputs=[output], live=True)
- 
+
 if __name__ == "__main__":
     demo.queue(max_size=200).launch()