webscout 8.2.7__py3-none-any.whl → 8.2.9__py3-none-any.whl

webscout/Provider/asksteve.py

@@ -0,0 +1,220 @@
+from typing import Any, Dict, Optional, Union
+from curl_cffi import CurlError
+from curl_cffi.requests import Session
+from webscout import exceptions
+from webscout.AIutel import Optimizers
+from webscout.AIutel import Conversation
+from webscout.AIutel import AwesomePrompts, sanitize_stream # Import sanitize_stream
+from webscout.AIbase import Provider 
+from webscout.litagent import LitAgent
+
+class AskSteve(Provider):
+    """
+    A class to interact with the AskSteve API.
+    """
+    AVAILABLE_MODELS = ["Gemini"]
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 600,
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+    ) -> None:
+        """Instantiates AskSteve
+
+        Args:
+            is_conversation (bool, optional): Flag for chatting conversationally. Defaults to True.
+            max_tokens (int, optional): Maximum number of tokens to be generated upon completion. Defaults to 600.
+            timeout (int, optional): Http request timeout. Defaults to 30.
+            intro (str, optional): Conversation introductory prompt. Defaults to None.
+            filepath (str, optional): Path to file containing conversation history. Defaults to None.
+            update_file (bool, optional): Add new prompts and responses to the file. Defaults to True.
+            proxies (dict, optional): Http request proxies. Defaults to {}.
+            history_offset (int, optional): Limit conversation history to this number of last texts. Defaults to 10250.
+            act (str|int, optional): Awesome prompt key or index. (Used as intro). Defaults to None.
+            system_prompt (str, optional): System prompt for AskSteve. Defaults to the provided string.
+        """
+        self.session = Session() # Use curl_cffi Session
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.api_endpoint = "https://quickstart.asksteve.to/quickStartRequest"
+        self.timeout = timeout
+        self.last_response = {}
+        self.headers = {
+            "accept": "*/*",
+            "accept-encoding": "gzip, deflate, br, zstd",
+            "accept-language": "en-US,en;q=0.9",
+            "content-type": "text/plain;charset=UTF-8",
+            "origin": "chrome-extension://gldebcpkoojijledacjeboaehblhfbjg",
+            "priority": "u=1, i",
+            "sec-fetch-dest": "empty",
+            "sec-fetch-mode": "cors",
+            "sec-fetch-site": "none",
+            "sec-fetch-storage-access": "active",
+            "user-agent": LitAgent().random(),
+        }
+
+        self.__available_optimizers = (
+            method
+            for method in dir(Optimizers)
+            if callable(getattr(Optimizers, method)) and not method.startswith("__")
+        )
+        self.session.headers.update(self.headers)
+        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_to_sample, filepath, update_file
+        )
+        self.conversation.history_offset = history_offset
+        self.session.proxies = proxies # Assign proxies directly
+    @staticmethod
+    def _asksteve_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from AskSteve JSON response."""
+        if isinstance(chunk, dict) and "candidates" in chunk and len(chunk["candidates"]) > 0:
+            parts = chunk["candidates"][0].get("content", {}).get("parts", [])
+            if parts and isinstance(parts[0].get("text"), str):
+                return parts[0]["text"]
+        return None
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = False,
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> dict:
+        """Chat with AI
+
+        Args:
+            prompt (str): Prompt to be send.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            raw (bool, optional): Stream back raw response as received. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name - `[code, shell_command]`. Defaults to None.
+            conversationally (bool, optional): Chat conversationally when using optimizer. Defaults to False.
+        Returns:
+           dict : {}
+        ```json
+        {
+           "text" : "How may I assist you today?"
+        }
+        ```
+        """
+        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 Exception(
+                    f"Optimizer is not one of {self.__available_optimizers}"
+                )
+
+        payload = {
+            "key": "asksteve",
+            "prompt": conversation_prompt
+        }
+
+
+        # This API doesn't stream, so we process the full response
+        try:
+            response = self.session.post(
+                self.api_endpoint,
+                headers=self.headers,
+                json=payload,
+                stream=False, # API doesn't stream
+                timeout=self.timeout,
+                impersonate="chrome120" # Add impersonate
+            )
+            response.raise_for_status()
+            response_text_raw = response.text # Get raw text
+
+            # Process the full JSON text using sanitize_stream
+            processed_stream = sanitize_stream(
+                data=response_text_raw,
+                to_json=True, # Parse the whole text as JSON
+                intro_value=None,
+                content_extractor=self._asksteve_extractor, # Use the specific extractor
+                yield_raw_on_error=False
+            )
+            # Extract the single result
+            text = next(processed_stream, None)
+            text = text if isinstance(text, str) else "" # Ensure it's a string
+
+            self.last_response.update(dict(text=text))
+            self.conversation.update_chat_history(
+                prompt, self.get_message(self.last_response)
+            )
+            # Always return a dict for consistency
+            return {"text": text} if raw else self.last_response
+
+        except CurlError as e:
+            raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {e}") from e
+        except Exception as e: # Catch other potential errors
+            raise exceptions.FailedToGenerateResponseError(f"Failed to get response ({type(e).__name__}): {e}") from e
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> str:
+        """Generate response `str`
+        Args:
+            prompt (str): Prompt to be send.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name - `[code, shell_command]`. Defaults to None.
+            conversationally (bool, optional): Chat conversationally when using optimizer. Defaults to False.
+        Returns:
+            str: Response generated
+        """
+
+        response_data = self.ask(
+            prompt,
+            stream=False, # Always False for this API
+            raw=False,    # Get the dict back
+            optimizer=optimizer,
+            conversationally=conversationally,
+        )
+        if stream:
+            def stream_wrapper():
+                yield self.get_message(response_data)
+            return stream_wrapper()
+        else:
+            return self.get_message(response_data)
+
+    def get_message(self, response) -> str:
+        """Retrieves message only from response
+
+        Args:
+            response (dict or str): Response generated by `self.ask` or a string
+
+        Returns:
+            str: Message extracted
+        """
+        if isinstance(response, dict):
+            return response.get("text", "") # Use .get for safety
+        elif isinstance(response, str):
+            return response
+        else:
+            raise TypeError(f"Unsupported response type: {type(response)}")
+
+
+if __name__ == "__main__":
+    from rich import print
+    ai = AskSteve()
+    response = ai.chat("write a short poem about AI", stream=True)
+    for chunk in response:
+        print(chunk, end="", flush=True) 

webscout/Provider/cerebras.py

@@ -0,0 +1,290 @@
+
+import re
+import curl_cffi
+from curl_cffi.requests import Session
+import json
+import os
+from typing import Any, Dict, Optional, Generator, List, Union
+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 as UserAgent
+
+class Cerebras(Provider):
+    """
+    A class to interact with the Cerebras API using a cookie for authentication.
+    """
+    
+    AVAILABLE_MODELS = [
+        "llama3.1-8b",
+        "llama-3.3-70b",
+        "deepseek-r1-distill-llama-70b",
+        "llama-4-scout-17b-16e-instruct",
+        "qwen-3-32b",
+
+
+    ]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 2049,
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+        cookie_path: str = "cookie.json",
+        model: str = "llama3.1-8b",
+        system_prompt: str = "You are a helpful assistant.",
+    ):
+        # Validate model choice
+        if model not in self.AVAILABLE_MODELS:
+            raise ValueError(
+                f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}"
+            )
+
+        # Initialize basic settings first
+        self.timeout = timeout
+        self.model = model
+        self.system_prompt = system_prompt
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.last_response = {}
+
+        self.session = Session() # Initialize curl_cffi session
+
+        # Get API key first
+        try:
+            self.api_key = self.get_demo_api_key(cookie_path)
+        except Exception as e:
+            raise exceptions.APIConnectionError(f"Failed to initialize Cerebras client: {e}")
+
+        # Initialize optimizers
+        self.__available_optimizers = (
+            method
+            for method in dir(Optimizers)
+            if callable(getattr(Optimizers, method)) and not method.startswith("__")
+        )
+
+        # Initialize conversation settings
+        Conversation.intro = (
+            AwesomePrompts().get_act(
+                act, raise_not_found=True, default=None, case_insensitive=True
+            )
+            if act
+            else None
+        )
+        self.conversation = Conversation(
+            is_conversation, self.max_tokens_to_sample, filepath, update_file
+        )
+        self.conversation.history_offset = history_offset
+        
+        # Apply proxies to the session
+        self.session.proxies = proxies
+
+    # Rest of the class implementation remains the same...
+    @staticmethod
+    def extract_query(text: str) -> str:
+        """Extracts the first code block from the given text."""
+        pattern = r"```(.*?)```"
+        matches = re.findall(pattern, text, re.DOTALL)
+        return matches[0].strip() if matches else text.strip()
+
+    @staticmethod
+    def refiner(text: str) -> str:
+        """Refines the input text by removing surrounding quotes."""
+        return text.strip('"')
+
+    @staticmethod
+    def _cerebras_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from Cerebras stream JSON objects."""
+        if isinstance(chunk, dict):
+            return chunk.get("choices", [{}])[0].get("delta", {}).get("content")
+        return None
+
+    def get_demo_api_key(self, cookie_path: str) -> str: # Keep this using requests or switch to curl_cffi
+        """Retrieves the demo API key using the provided cookie."""
+        try:
+            with open(cookie_path, "r") as file:
+                cookies = {item["name"]: item["value"] for item in json.load(file)}
+        except FileNotFoundError:
+            raise FileNotFoundError(f"Cookie file not found at path: {cookie_path}")
+        except json.JSONDecodeError:
+            raise json.JSONDecodeError("Invalid JSON format in the cookie file.", "", 0)
+
+        headers = {
+            "Accept": "*/*",
+            "Accept-Language": "en-US,en;q=0.9",
+            "Content-Type": "application/json",
+            "Origin": "https://inference.cerebras.ai",
+            "Referer": "https://inference.cerebras.ai/",
+            "user-agent": UserAgent().random(),
+        }
+
+        json_data = {
+            "operationName": "GetMyDemoApiKey",
+            "variables": {},
+            "query": "query GetMyDemoApiKey {\n  GetMyDemoApiKey\n}",
+        }
+
+        try:
+            # Use the initialized curl_cffi session
+            response = self.session.post(
+                "https://inference.cerebras.ai/api/graphql",
+                cookies=cookies,
+                headers=headers,
+                json=json_data,
+                timeout=self.timeout,
+                impersonate="chrome120" # Add impersonate
+            )
+            response.raise_for_status()
+            api_key = response.json().get("data", {}).get("GetMyDemoApiKey")
+            return api_key
+        except curl_cffi.CurlError as e:
+            raise exceptions.APIConnectionError(f"Failed to retrieve API key: {e}")
+        except KeyError:
+            raise exceptions.InvalidResponseError("API key not found in response.")
+
+    def _make_request(self, messages: List[Dict], stream: bool = False) -> Union[Dict, Generator]:
+        """Make a request to the Cerebras API."""
+        headers = {
+            "Authorization": f"Bearer {self.api_key}",
+            "Content-Type": "application/json",
+            "User-Agent": UserAgent().random(),
+        }
+
+        payload = {
+            "model": self.model,
+            "messages": messages,
+            "stream": stream
+        }
+
+        try:
+            # Use the initialized curl_cffi session
+            response = self.session.post(
+                "https://api.cerebras.ai/v1/chat/completions",
+                headers=headers,
+                json=payload,
+                stream=stream,
+                timeout=self.timeout,
+                impersonate="chrome120" # Add impersonate
+            )
+            response.raise_for_status()
+
+            if stream:
+                def generate_stream():
+                    # 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
+                        content_extractor=self._cerebras_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:
+                        if content_chunk and isinstance(content_chunk, str):
+                            yield content_chunk # Yield the extracted text chunk
+
+                return generate_stream()
+            else:
+                response_json = response.json()
+                # Extract content for non-streaming response
+                content = response_json.get("choices", [{}])[0].get("message", {}).get("content")
+                return content if content else "" # Return empty string if not found
+
+        except curl_cffi.CurlError as e:
+            raise exceptions.APIConnectionError(f"Request failed (CurlError): {e}") from e
+        except Exception as e: # Catch other potential errors
+            raise exceptions.APIConnectionError(f"Request failed: {e}")
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = False,
+        raw: bool = False, # Add raw parameter for consistency
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[Dict, Generator]:
+        """Send a prompt to the model and get a 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 Exception(f"Optimizer is not one of {self.__available_optimizers}")
+
+        messages = [
+            {"role": "system", "content": self.system_prompt},
+            {"role": "user", "content": conversation_prompt}
+        ]
+
+        try:
+            response = self._make_request(messages, stream)
+            
+            if stream:
+                # Wrap the generator to yield dicts or raw strings
+                def stream_wrapper():
+                    full_text = ""
+                    for chunk in response:
+                        full_text += chunk
+                        yield chunk if raw else {"text": chunk}
+                    # Update history after stream finishes
+                    self.last_response = {"text": full_text}
+                    self.conversation.update_chat_history(prompt, full_text)
+                return stream_wrapper()
+            else:
+                # Non-streaming response is already the full text string
+                self.last_response = {"text": response}
+                self.conversation.update_chat_history(prompt, response)
+                return self.last_response if not raw else response # Return dict or raw string
+
+        except Exception as e:
+            raise exceptions.FailedToGenerateResponseError(f"Error during request: {e}")
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[str, Generator]:
+        """Chat with the model."""
+        # Ask returns a generator for stream=True, dict/str for stream=False
+        response_gen_or_dict = self.ask(prompt, stream, raw=False, optimizer=optimizer, conversationally=conversationally)
+        
+        if stream:
+            # Wrap the generator from ask() to get message text
+            def stream_wrapper():
+                for chunk_dict in response_gen_or_dict:
+                    yield self.get_message(chunk_dict)
+            return stream_wrapper()
+        else:
+            # Non-streaming response is already a dict
+            return self.get_message(response_gen_or_dict)
+
+    def get_message(self, response: str) -> str:
+        """Retrieves message from response."""
+        # Updated to handle dict input from ask()
+        assert isinstance(response, dict), "Response should be of dict data-type only for get_message"
+        return response.get("text", "")
+
+
+if __name__ == "__main__":
+    from rich import print
+    
+    # Example usage
+    cerebras = Cerebras(
+        cookie_path=r'cookies.json',
+        model='llama3.1-8b',
+        system_prompt="You are a helpful AI assistant."
+    )
+    
+    # Test with streaming
+    response = cerebras.chat("Hello!", stream=True)
+    for chunk in response:
+        print(chunk, end="", flush=True)