webscout 8.2.4__py3-none-any.whl → 8.2.6__py3-none-any.whl

webscout/Provider/HeckAI.py

@@ -120,7 +120,7 @@ class HeckAI(Provider):
 
         def for_stream():
             streaming_text = "" # Initialize outside try block
-            in_answer = False # Initialize outside try block
+            # in_answer = False # No longer needed
             try:
                 # Use curl_cffi session post with impersonate
                 response = self.session.post(
@@ -132,41 +132,23 @@ class HeckAI(Provider):
                     impersonate="chrome110" # Use a common impersonation profile
                 )
                 response.raise_for_status() # Check for HTTP errors
-                    
-                # Iterate over bytes and decode manually
-                for line_bytes in response.iter_lines():
-                    if not line_bytes:
-                        continue
-                        
-                    try:
-                        line = line_bytes.decode('utf-8')
-                        # Remove "data: " prefix
-                        if line.startswith("data: "):
-                            data = line[6:]
-                        else:
-                            continue # Skip lines without the prefix
-                        
-                        # Check for control markers
-                        if data == "[ANSWER_START]":
-                            in_answer = True
-                            continue
-                            
-                        if data == "[ANSWER_DONE]":
-                            in_answer = False
-                            continue
-                            
-                        if data == "[RELATE_Q_START]" or data == "[RELATE_Q_DONE]":
-                            continue
-                            
-                        # Process content if we're in an answer section
-                        if in_answer:
-                            # Assuming 'data' is the text chunk here
-                            streaming_text += data
-                            resp = dict(text=data)
-                            # Yield dict or raw string chunk
-                            yield resp if not raw else data
-                    except UnicodeDecodeError:
-                        continue # Ignore decoding errors for specific lines
+
+                # Use sanitize_stream to process the stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value="data:",  # Prefix to remove *from lines between markers*
+                    to_json=False,        # Content is text
+                    start_marker="data: [ANSWER_START]", # Check against the raw line including prefix
+                    end_marker="data: [ANSWER_DONE]",     # Check against the raw line including prefix
+                    skip_markers=["[RELATE_Q_START]", "[RELATE_Q_DONE]"], # Skip these if they appear within answer block
+                    yield_raw_on_error=True
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the text between ANSWER_START and ANSWER_DONE
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk
+                        yield dict(text=content_chunk) if not raw else content_chunk
                 
                 # Update history and previous answer after stream finishes
                 self.previous_answer = streaming_text

webscout/Provider/Jadve.py

@@ -4,7 +4,7 @@ import json
 import re
 from typing import Union, Any, Dict, Optional, Generator
 
-from webscout.AIutel import Optimizers, Conversation, AwesomePrompts
+from webscout.AIutel import Optimizers, Conversation, AwesomePrompts, sanitize_stream # Import sanitize_stream
 from webscout.AIbase import Provider
 from webscout import exceptions
 from webscout.litagent import LitAgent
@@ -97,6 +97,17 @@ class JadveOpenAI(Provider):
         )
         self.conversation.history_offset = history_offset
 
+    @staticmethod
+    def _jadve_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from the Jadve stream format '0:"..."'."""
+        if isinstance(chunk, str):
+            match = re.search(r'0:"(.*?)"(?=,|$)', chunk) # Look for 0:"...", possibly followed by comma or end of string
+            if match:
+                # Decode potential unicode escapes like \u00e9 and handle escaped quotes/backslashes
+                content = match.group(1).encode().decode('unicode_escape')
+                return content.replace('\\\\', '\\').replace('\\"', '"')
+        return None
+
     def ask(
         self,
         prompt: str,
@@ -153,41 +164,22 @@ class JadveOpenAI(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
 
-                # Pattern to match the streaming chunks format: 0:"text"
-                pattern = r'0:"(.*?)"'
-                buffer = ""
-                
-                # Iterate over bytes and decode manually
-                for line_bytes in response.iter_lines():
-                    if not line_bytes:
-                        continue
-                    
-                    try:
-                        line = line_bytes.decode('utf-8')
-                        buffer += line
-                        
-                        # Try to match chunks in the current buffer
-                        matches = re.findall(pattern, buffer)
-                        if matches:
-                            for chunk in matches:
-                                # Handle potential escape sequences like \\n
-                                decoded_chunk = chunk.encode().decode('unicode_escape')
-                                full_response_text += decoded_chunk
-                                resp = {"text": decoded_chunk}
-                                # Yield dict or raw string chunk
-                                yield resp if not raw else decoded_chunk
-                            
-                            # Remove matched parts from the buffer
-                            # Be careful with buffer modification during iteration if issues arise
-                            matched_parts = [f'0:"{match}"' for match in matches]
-                            for part in matched_parts:
-                                buffer = buffer.replace(part, '', 1)
-                        
-                        # Check if we've reached the end of the response
-                        if 'e:' in line or 'd:' in line:
-                            break
-                    except UnicodeDecodeError:
-                        continue # Ignore decoding errors for specific lines
+                # Use sanitize_stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value=None, # No simple prefix
+                    to_json=False,    # Content is text after extraction
+                    content_extractor=self._jadve_extractor, # Use the specific extractor
+                    # end_marker="e:", # Add if 'e:' reliably marks the end
+                    yield_raw_on_error=True
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by _jadve_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        full_response_text += content_chunk
+                        resp = {"text": content_chunk}
+                        yield resp if not raw else content_chunk
 
                 # Update history after stream finishes
                 self.last_response = {"text": full_response_text}
@@ -272,7 +264,8 @@ class JadveOpenAI(Provider):
             str: Extracted text.
         """
         assert isinstance(response, dict), "Response should be of dict data-type only"
-        return response["text"]
+        # Extractor handles formatting
+        return response.get("text", "")
 
 if __name__ == "__main__":
     # Ensure curl_cffi is installed

webscout/Provider/LambdaChat.py

@@ -7,8 +7,8 @@ import re
 import uuid
 from typing import Any, Dict, List, Optional, Union, Generator
 
-from webscout.AIutel import Conversation
-from webscout.AIbase import Provider
+from webscout.AIutel import Conversation, sanitize_stream
+from webscout.AIbase import Provider # Import sanitize_stream
 from webscout import exceptions
 from webscout.litagent import LitAgent
 
@@ -182,62 +182,21 @@ class LambdaChat(Provider):
         boundary += "".join(random.choice(boundary_chars) for _ in range(16))
         return boundary
     
-    def process_response(self, response, prompt: str):
-        """Process streaming response and extract content."""
-        full_text = ""
-        sources = None
+    @staticmethod
+    def _lambdachat_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from LambdaChat stream JSON objects."""
+        if not isinstance(chunk, dict) or "type" not in chunk:
+            return None
+
         reasoning_text = ""
-        has_reasoning = False
-        
-        for line in response.iter_lines(decode_unicode=True):
-            if not line:
-                continue
-                
-            try:
-                # Parse each line as JSON
-                data = json.loads(line)
-                
-                # Handle different response types
-                if "type" not in data:
-                    continue
-                    
-                if data["type"] == "stream" and "token" in data:
-                    token = data["token"].replace("\u0000", "")
-                    full_text += token
-                    resp = {"text": token}
-                    yield resp
-                elif data["type"] == "finalAnswer":
-                    final_text = data.get("text", "")
-                    if final_text and not full_text:
-                        full_text = final_text
-                        resp = {"text": final_text}
-                        yield resp
-                elif data["type"] == "webSearch" and "sources" in data:
-                    sources = data["sources"]
-                elif data["type"] == "reasoning":
-                    has_reasoning = True
-                    if data.get("subtype") == "stream" and "token" in data:
-                        reasoning_text += data["token"]
-                    
-                    # If we have reasoning, prepend it to the next text output
-                    if reasoning_text and not full_text:
-                        resp = {"text": f"<think>\n{reasoning_text}\n</think>\n", "is_reasoning": True}
-                        yield resp
-                    
-            except json.JSONDecodeError:
-                continue
-        
-        # Update conversation history only for saving to file if needed
-        if full_text and self.conversation.file:
-            if has_reasoning:
-                full_text_with_reasoning = f"<think>\n{reasoning_text}\n</think>\n{full_text}"
-                self.last_response = {"text": full_text_with_reasoning}
-                self.conversation.update_chat_history(prompt, full_text_with_reasoning)
-            else:
-                self.last_response = {"text": full_text}
-                self.conversation.update_chat_history(prompt, full_text)
-        
-        return full_text
+        if chunk["type"] == "stream" and "token" in chunk:
+            return chunk["token"].replace("\u0000", "")
+        elif chunk["type"] == "finalAnswer":
+            return chunk.get("text")
+        elif chunk["type"] == "reasoning" and chunk.get("subtype") == "stream" and "token" in chunk:
+            # Prepend reasoning with <think> tags? Or handle separately? For now, just return token.
+            return chunk["token"] # Or potentially format as f"<think>{chunk['token']}</think>"
+        return None
     
     def ask(
         self,
@@ -296,6 +255,7 @@ class LambdaChat(Provider):
         multipart_headers["Content-Length"] = str(len(body))
         
         def for_stream():
+            streaming_text = "" # Initialize for history
             try:
                 # Try with multipart/form-data first
                 response = None
@@ -327,8 +287,21 @@ class LambdaChat(Provider):
                 
                 response.raise_for_status() # Check status after potential fallback
                 
-                # Process the streaming response
-                yield from self.process_response(response, prompt)
+                # Use sanitize_stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value=None, # No prefix
+                    to_json=True,     # Stream sends JSON lines
+                    content_extractor=self._lambdachat_extractor, # Use the specific extractor
+                    yield_raw_on_error=False # Skip non-JSON lines or lines where extractor fails
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by _lambdachat_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk # Aggregate text for history
+                        resp = {"text": content_chunk}
+                        yield resp if not raw else content_chunk
                 
             except (CurlError, exceptions.FailedToGenerateResponseError, Exception) as e: # Catch errors from both attempts
                 # Handle specific exceptions if needed
@@ -353,6 +326,10 @@ class LambdaChat(Provider):
                 # If we get here, all models failed
                 raise exceptions.FailedToGenerateResponseError(f"Request failed after trying fallback: {str(e)}") from e
 
+            # Update history after stream finishes
+            if streaming_text and self.conversation.file:
+                self.last_response = {"text": streaming_text}
+                self.conversation.update_chat_history(prompt, streaming_text)
 
         def for_non_stream():
             # Aggregate the stream using the updated for_stream logic

webscout/Provider/MCPCore.py

@@ -8,7 +8,7 @@ from curl_cffi import CurlError
 
 from webscout.AIutel import Optimizers
 from webscout.AIutel import Conversation
-from webscout.AIutel import AwesomePrompts
+from webscout.AIutel import AwesomePrompts, sanitize_stream 
 from webscout.AIbase import Provider
 from webscout import exceptions
 from webscout.litagent import LitAgent
@@ -146,11 +146,11 @@ class MCPCore(Provider):
             )
             return cookie_string, token
         except FileNotFoundError:
-            raise exceptions.InvalidAuthenticationError(
+            raise exceptions.FailedToGenerateResponseError(
                 f"Error: Cookies file not found at {self.cookies_path}!"
             )
         except json.JSONDecodeError:
-            raise exceptions.InvalidAuthenticationError(
+            raise exceptions.FailedToGenerateResponseError(
                 f"Error: Invalid JSON format in cookies file: {self.cookies_path}!"
             )
 
@@ -207,24 +207,21 @@ class MCPCore(Provider):
                 )
                 response.raise_for_status()
 
-                for line_bytes in response.iter_lines():
-                    if line_bytes:
-                        try:
-                            line = line_bytes.decode('utf-8').strip()
-                            if line.startswith("data: "):
-                                json_str = line[6:]
-                                if json_str == "[DONE]":
-                                    break
-                                json_data = json.loads(json_str)
-                                if 'choices' in json_data and len(json_data['choices']) > 0:
-                                    delta = json_data['choices'][0].get('delta', {})
-                                    content = delta.get('content')
-                                    if content:
-                                        streaming_text += content
-                                        resp = dict(text=content)
-                                        yield resp if not raw else content
-                        except (json.JSONDecodeError, UnicodeDecodeError):
-                            continue
+                # Use sanitize_stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value="data:",
+                    to_json=True,     # Stream sends JSON
+                    skip_markers=["[DONE]"],
+                    content_extractor=lambda chunk: chunk.get('choices', [{}])[0].get('delta', {}).get('content') if isinstance(chunk, dict) else None,
+                    yield_raw_on_error=False # Skip non-JSON or lines where extractor fails
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by the content_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk
+                        yield dict(text=content_chunk) if not raw else content_chunk
 
                 self.last_response = {"text": streaming_text}
                 self.conversation.update_chat_history(prompt, self.get_message(self.last_response))

webscout/Provider/Marcus.py

@@ -5,7 +5,7 @@ from typing import Union, Any, Dict, Optional, Generator
 
 from webscout.AIutel import Optimizers
 from webscout.AIutel import Conversation
-from webscout.AIutel import AwesomePrompts
+from webscout.AIutel import AwesomePrompts, sanitize_stream # Import sanitize_stream
 from webscout.AIbase import Provider
 from webscout import exceptions
 
@@ -103,18 +103,18 @@ class Marcus(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
                 
-                # Iterate over bytes and decode manually
-                for line_bytes in response.iter_lines():
-                    if line_bytes:
-                        try:
-                            decoded_line = line_bytes.decode('utf-8')
-                            streaming_text += decoded_line # Aggregate text
-                            resp = {"text": decoded_line}
-                            # Yield dict or raw string chunk
-                            yield resp if not raw else decoded_line
-                        except UnicodeDecodeError:
-                            continue # Ignore decoding errors
+                # Use sanitize_stream to decode bytes and yield text chunks
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value=None, # No prefix
+                    to_json=False,    # It's plain text
+                    yield_raw_on_error=True
+                )
 
+                for content_chunk in processed_stream:
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk # Aggregate text
+                        yield {"text": content_chunk} if not raw else content_chunk
                 # Update history after stream finishes
                 self.last_response = {"text": streaming_text} # Store aggregated text
                 self.conversation.update_chat_history(
@@ -140,8 +140,17 @@ class Marcus(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
                 
-                # Use response.text which is already decoded
-                full_response = response.text
+                response_text_raw = response.text # Get raw text
+
+                # Process the text using sanitize_stream (even though it's not streaming)
+                processed_stream = sanitize_stream(
+                    data=response_text_raw,
+                    intro_value=None, # No prefix
+                    to_json=False     # It's plain text
+                )
+                # Aggregate the single result
+                full_response = "".join(list(processed_stream))
+
                 self.last_response = {"text": full_response}
                 self.conversation.update_chat_history(prompt, full_response)
                 # Return dict or raw string

webscout/Provider/Nemotron.py

@@ -0,0 +1,218 @@
+import requests
+import json
+import random
+import datetime
+from typing import Any, Dict, Optional, Union, Generator
+from webscout.AIutel import Optimizers, Conversation, AwesomePrompts, sanitize_stream
+from webscout.AIbase import Provider
+from webscout import exceptions
+
+class NEMOTRON(Provider):
+    """NEMOTRON provider for interacting with the nemotron.one API."""
+    url = "https://nemotron.one/api/chat"
+
+    AVAILABLE_MODELS = [
+        "gpt4o",
+        "nemotron70b",
+    ]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 8000,
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+        model: str = "gpt4o"
+    ):
+        """Initialize NEMOTRON with configuration options."""
+        self.session = requests.Session()
+        self.max_tokens = max_tokens
+        self.is_conversation = is_conversation
+        self.timeout = timeout
+        self.last_response = {}
+        self.model = self.get_model(model)
+
+        self.headers = {
+            "authority": "nemotron.one",
+            "accept": "*/*",
+            "accept-language": "en-US,en;q=0.9",
+            "content-type": "application/json",
+            "origin": "https://nemotron.one",
+            "referer": f"https://nemotron.one/chat/{self.model}",
+            "sec-ch-ua": '"Chromium";v="136", "Not.A/Brand";v="99"',
+            "sec-ch-ua-mobile": "?0",
+            "sec-ch-ua-platform": '"Windows"',
+            "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/136.0.0.0 Safari/537.36"
+        }
+
+        self.__available_optimizers = (
+            method for method in dir(Optimizers)
+            if callable(getattr(Optimizers, method)) and not method.startswith("__")
+        )
+
+        Conversation.intro = (
+            AwesomePrompts().get_act(
+                act, raise_not_found=True, default=None, case_insensitive=True
+            )
+            if act
+            else intro or Conversation.intro
+        )
+        
+        self.conversation = Conversation(
+            is_conversation, self.max_tokens, filepath, update_file
+        )
+        self.conversation.history_offset = history_offset
+        self.session.proxies = proxies
+
+    @staticmethod
+    def _generate_random_email() -> str:
+        """Generate a random email address."""
+        random_letter = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789"
+        random_string = ''.join(random.choice(random_letter) for _ in range(10))
+        return f"{random_string}@gmail.com"
+
+    @staticmethod
+    def _generate_random_id() -> str:
+        """Generate a random user ID."""
+        timestamp = datetime.datetime.now().strftime("%Y%m%d%H%M%S%f")
+        random_letter = "abcdefghijklmnopqrstuvwxyz0123456789"
+        random_string = ''.join(random.choice(random_letter) for _ in range(8))
+        return f"cm{random_string}{timestamp[:10]}"
+
+    @classmethod
+    def get_model(cls, model: str) -> str:
+        """Resolve model name from alias."""
+        if model in cls.AVAILABLE_MODELS:
+            return model  # Simply return the model name if it's in the list
+        raise ValueError(f"Unknown model: {model}. Available models: {', '.join(cls.AVAILABLE_MODELS)}")
+
+    def _get_user_data(self) -> Dict[str, Any]:
+        """Generate user data for the request."""
+        current_time = datetime.datetime.now().isoformat()
+        return {
+            "name": "user",
+            "email": self._generate_random_email(),
+            "image": "https://lh3.googleusercontent.com/a/default-user=s96-c",
+            "id": self._generate_random_id(),
+            "password": None,
+            "emailVerified": None,
+            "credits": 100000000000,
+            "isPro": False,
+            "createdAt": current_time,
+            "updatedAt": current_time
+        }
+
+    def _make_request(
+        self,
+        message: str,
+        stream: bool = False
+    ) -> Generator[str, None, None]:
+        """Make request to NEMOTRON API."""
+        payload = {
+            "content": message,
+            "imageSrc": "",
+            "model": self.model,
+            "user": self._get_user_data(),
+            "conversationId": ""
+        }
+
+        try:
+            if stream:
+                with self.session.post(
+                    self.url,
+                    headers=self.headers,
+                    json=payload,
+                    stream=True,
+                    timeout=self.timeout
+                ) as response:
+                    response.raise_for_status()
+                    yield from sanitize_stream(
+                        response.iter_content(chunk_size=1024),
+                        to_json=False,
+                    )
+            else:
+                response = self.session.post(
+                    self.url,
+                    headers=self.headers,
+                    json=payload,
+                    timeout=self.timeout
+                )
+                response.raise_for_status()
+                yield response.text
+
+        except requests.exceptions.RequestException as e:
+            raise exceptions.ProviderConnectionError(f"Connection error: {str(e)}")
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = False,
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[Dict[str, str], Generator[Dict[str, str], None, None]]:
+        """Send a prompt to NEMOTRON API and return the response."""
+        conversation_prompt = self.conversation.gen_complete_prompt(prompt)
+        if optimizer:
+            if optimizer in self.__available_optimizers:
+                conversation_prompt = getattr(Optimizers, optimizer)(
+                    conversation_prompt if conversationally else prompt
+                )
+            else:
+                raise ValueError(f"Optimizer is not one of {self.__available_optimizers}")
+
+        def for_stream():
+            for text in self._make_request(conversation_prompt, stream=True):
+                yield {"text": text}
+
+        def for_non_stream():
+            response_text = next(self._make_request(conversation_prompt, stream=False))
+            self.last_response = {"text": response_text}
+            return self.last_response
+
+        return for_stream() if stream else for_non_stream()
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[str, Generator[str, None, None]]:
+        """Generate response as string."""
+        def for_stream():
+            for response in self.ask(
+                prompt,
+                stream=True,
+                optimizer=optimizer,
+                conversationally=conversationally
+            ):
+                yield self.get_message(response)
+
+        def for_non_stream():
+            return self.get_message(
+                self.ask(
+                    prompt,
+                    stream=False,
+                    optimizer=optimizer,
+                    conversationally=conversationally,
+                )
+            )
+
+        return for_stream() if stream else for_non_stream()
+
+    def get_message(self, response: Dict[str, Any]) -> str:
+        """Extract message from response dictionary."""
+        assert isinstance(response, dict), "Response should be of dict data-type only"
+        return response["text"]
+
+if __name__ == "__main__":
+    # Example usage
+    nemotron = NEMOTRON()
+    response = nemotron.chat("Hello, how are you?", stream=False)
+    print(response)