webscout 8.3.6__py3-none-any.whl → 8.3.7__py3-none-any.whl

webscout/Provider/AllenAI.py

@@ -1,440 +0,0 @@
-from curl_cffi.requests import Session
-from curl_cffi import CurlError
-import json
-import os
-from uuid import uuid4
-from typing import Any, Dict, Optional, Generator, Union
-
-from webscout.AIutel import Optimizers
-from webscout.AIutel import Conversation
-from webscout.AIutel import AwesomePrompts, sanitize_stream
-from webscout.AIbase import Provider, AsyncProvider
-from webscout import exceptions
-from webscout.litagent import LitAgent
-
-class AllenAI(Provider):
-    """
-    A class to interact with the AllenAI (Ai2 Playground) API.
-    """
-
-    AVAILABLE_MODELS = [
-        'olmo-2-0325-32b-instruct',
-        'tulu3-405b'
-    ]
-
-    # Default model options from JS implementation
-    DEFAULT_OPTIONS = {
-        "max_tokens": 2048, 
-        "temperature": 0.7,
-        "top_p": 1,
-        "n": 1,
-        "stop": None,
-        "logprobs": None
-    }
-    
-    # Host mapping for models - some models work best with specific hosts
-    MODEL_HOST_MAP = {
-        'olmo-2-0325-32b-instruct': 'modal',
-        'tulu3-405b': 'inferd'
-    }
-
-    def __init__(
-        self,
-        is_conversation: bool = True,
-        max_tokens: int = 2048,
-        timeout: int = 30,
-        intro: str = None,
-        filepath: str = None,
-        update_file: bool = True,
-        proxies: dict = {},
-        history_offset: int = 10250,
-        act: str = None,
-        model: str = "OLMo-2-1124-13B-Instruct",
-        host: str = None
-    ):
-        """Initializes the AllenAI API client."""
-        if model not in self.AVAILABLE_MODELS:
-            raise ValueError(f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}")
-            
-        self.url = "https://playground.allenai.org"
-        self.api_endpoint = "https://olmo-api.allen.ai/v3/message/stream"
-        self.whoami_endpoint = "https://olmo-api.allen.ai/v3/whoami"
-        
-        # Updated headers (remove those handled by impersonate)
-        self.headers = {
-            'Accept': '*/*',
-            'Accept-Language': 'id-ID,id;q=0.9',
-            'Origin': self.url,
-            'Referer': f"{self.url}/",
-            'Cache-Control': 'no-cache',
-            'Pragma': 'no-cache',
-            'Priority': 'u=1, i',
-            'Sec-Fetch-Dest': 'empty',
-            'Sec-Fetch-Mode': 'cors',
-            'Sec-Fetch-Site': 'cross-site',
-            'Content-Type': 'application/json'
-        }
-        
-        # Initialize curl_cffi Session
-        self.session = Session()
-        # Update curl_cffi session headers and proxies
-        self.session.headers.update(self.headers)
-        self.session.proxies = proxies
-        
-        self.model = model
-        
-        # Auto-detect host if not provided
-        if not host:
-            # Use the preferred host from the model-host map, or default to modal
-            self.host = self.MODEL_HOST_MAP.get(model, 'modal')
-        else:
-            self.host = host
-            
-        self.is_conversation = is_conversation
-        self.max_tokens_to_sample = max_tokens
-        self.timeout = timeout
-        self.last_response = {}       
-        # Generate user ID if needed
-        self.x_anonymous_user_id = None
-        self.parent = None
-        
-        # Default options
-        self.options = self.DEFAULT_OPTIONS.copy()
-        self.options["max_tokens"] = max_tokens
-
-        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_to_sample, filepath, update_file
-        )
-        self.conversation.history_offset = history_offset
-
-    def whoami(self):
-        """Gets or creates a user ID for authentication with Allen AI API"""
-        temp_id = str(uuid4())
-        request_headers = self.session.headers.copy() # Use session headers as base
-        request_headers.update({"x-anonymous-user-id": temp_id})
-        
-        try:
-            # Use curl_cffi session get with impersonate
-            response = self.session.get(
-                self.whoami_endpoint,
-                headers=request_headers, # Pass updated headers
-                timeout=self.timeout,
-                impersonate="chrome110" # Use a common impersonation profile
-            )
-            response.raise_for_status() # Check for HTTP errors
-            
-            data = response.json()
-            self.x_anonymous_user_id = data.get("client", temp_id)
-            return data
-                
-        except CurlError as e: # Catch CurlError
-            self.x_anonymous_user_id = temp_id
-            return {"client": temp_id, "error": f"CurlError: {e}"}
-        except Exception as e: # Catch other potential exceptions (like HTTPError, JSONDecodeError)
-            self.x_anonymous_user_id = temp_id
-            err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
-            return {"client": temp_id, "error": f"{type(e).__name__}: {e} - {err_text}"}
-    
-    @staticmethod
-    def _allenai_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
-        """Extracts content from AllenAI stream JSON objects."""
-        if isinstance(chunk, dict):
-            if chunk.get("message", "").startswith("msg_") and "content" in chunk:
-                return chunk.get("content")
-            elif "message" in chunk and chunk.get("content"): # Legacy handling
-                return chunk.get("content")
-        return None
-
-    def ask(
-        self,
-        prompt: str,
-        stream: bool = False, # API supports streaming
-        raw: bool = False,
-        optimizer: str = None,
-        conversationally: bool = False,
-        host: str = None,
-        private: bool = False,
-        top_p: float = None,
-        temperature: float = None,
-        options: dict = None,
-    ) -> Union[Dict[str, Any], Generator]:
-        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}")
-
-        # Ensure we have a user ID
-        if not self.x_anonymous_user_id:
-            self.whoami()
-            # Check if whoami failed and we still don't have an ID
-            if not self.x_anonymous_user_id:
-                 raise exceptions.AuthenticationError("Failed to obtain anonymous user ID.")
-        
-        # Prepare the API request headers for this specific request
-        request_headers = self.session.headers.copy()
-        request_headers.update({
-            "x-anonymous-user-id": self.x_anonymous_user_id,
-            "Content-Type": "application/json" # Ensure Content-Type is set
-        })
-        
-        # Create options dictionary
-        opts = self.options.copy()
-        if temperature is not None:
-            opts["temperature"] = temperature
-        if top_p is not None:
-            opts["top_p"] = top_p
-        if options:
-            opts.update(options)
-        
-        # Use the host param or the default host
-        use_host = host or self.host
-        
-        # List of hosts to try - start with provided host, then try alternative hosts
-        hosts_to_try = [use_host]
-        if use_host == 'modal':
-            hosts_to_try.append('inferd')
-        else:
-            hosts_to_try.append('modal')
-        
-        last_error = None
-        
-        # Try each host until one works
-        for current_host in hosts_to_try:
-            # Create the JSON payload as per the JS implementation
-            payload = {
-                "content": conversation_prompt,
-                "private": private,
-                "model": self.model,
-                "host": current_host,
-                "opts": opts
-            }
-            payload["host"] = current_host # Ensure host is updated in payload
-                
-            try:
-                if stream:
-                    # Pass request_headers to the stream method
-                    return self._stream_request(payload, prompt, request_headers, raw)
-                else:
-                    # Pass request_headers to the non-stream method
-                    return self._non_stream_request(payload, prompt, request_headers, raw)
-            except (exceptions.FailedToGenerateResponseError, CurlError, Exception) as e:
-                last_error = e
-                # Log the error but continue to try other hosts
-                print(f"Host '{current_host}' failed for model '{self.model}' ({type(e).__name__}), trying next host...")
-                continue
-        
-        # If we've tried all hosts and none worked, raise the last error
-        raise last_error or exceptions.FailedToGenerateResponseError("All hosts failed. Unable to complete request.")
-
-    def _stream_request(self, payload, prompt, request_headers, raw=False):
-        """Handle streaming requests with the given payload and headers"""
-        streaming_text = "" # Initialize outside try block
-        current_parent = None # Initialize outside try block
-        try:
-            # Use curl_cffi session post with impersonate
-            response = self.session.post(
-                self.api_endpoint,
-                headers=request_headers, # Use headers passed to this method
-                json=payload,
-                stream=True,
-                timeout=self.timeout,
-                impersonate="chrome110" # Use a common impersonation profile
-            )
-            response.raise_for_status() # Check for HTTP errors
-            
-            # 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._allenai_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 _allenai_extractor
-                if content_chunk and isinstance(content_chunk, str):
-                    streaming_text += content_chunk
-                    resp = dict(text=content_chunk)
-                    yield resp if not raw else content_chunk
-
-            # Try to extract parent ID from the *last* raw line (less reliable than before)
-            # This part is tricky as sanitize_stream consumes the raw lines.
-            # We might need to re-fetch or adjust if parent ID is critical per stream.
-            # For now, we'll rely on the non-stream request to update parent ID more reliably.
-            # Example placeholder logic (might not work reliably):
-            try:
-                last_line_data = json.loads(response.text.splitlines()[-1]) # Get last line if possible
-                if last_line_data.get("id"):
-                    current_parent = last_line_data.get("id")
-                elif last_line_data.get("children"):
-                    for child in last_line_data["children"]: # Use last_line_data here
-                        if child.get("role") == "assistant":
-                            current_parent = child.get("id")
-                            break
-                
-                # Handle completion
-                if last_line_data.get("final") or last_line_data.get("finish_reason") == "stop":
-                    if current_parent:
-                        self.parent = current_parent
-                    
-                    # Update conversation history only if not empty
-                    if streaming_text.strip():
-                        self.conversation.update_chat_history(prompt, streaming_text)
-                        self.last_response = {"text": streaming_text} # Update last response here
-                    return # End the generator
-            except Exception as e:
-                # Log the error but continue with the rest of the function
-                print(f"Error processing response data: {str(e)}")
-            
-            # If loop finishes without returning (e.g., no final message), update history
-            if current_parent:
-                self.parent = current_parent
-            self.conversation.update_chat_history(prompt, streaming_text)
-            self.last_response = {"text": streaming_text}
-
-        except CurlError as e: # Catch CurlError
-            raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {str(e)}") from e
-        except Exception as e: # Catch other potential exceptions (like HTTPError)
-            err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
-            raise exceptions.FailedToGenerateResponseError(f"Request failed ({type(e).__name__}): {str(e)} - {err_text}") from e
-
-
-    def _non_stream_request(self, payload, prompt, request_headers, raw=False):
-        """Handle non-streaming requests with the given payload and headers"""
-        try:
-            # Use curl_cffi session post with impersonate
-            response = self.session.post(
-                self.api_endpoint,
-                headers=request_headers, # Use headers passed to this method
-                json=payload,
-                stream=False, # Explicitly set stream to False
-                timeout=self.timeout,
-                impersonate="chrome110" # Use a common impersonation profile
-            )
-            response.raise_for_status() # Check for HTTP errors
-            
-            raw_response = response.text # Get raw text
-
-            # Process the full text using sanitize_stream line by line
-            processed_stream = sanitize_stream(
-                data=raw_response.splitlines(), # Split into lines
-                intro_value=None,
-                to_json=True,
-                content_extractor=self._allenai_extractor,
-                yield_raw_on_error=False
-            )
-            # Aggregate the results
-            parsed_response = "".join(list(processed_stream))
-
-            # Update parent ID from the full response if possible (might need adjustment based on actual non-stream response structure)
-            # This part is speculative as the non-stream structure isn't fully clear from the stream logic
-            try:
-                lines = raw_response.splitlines()
-                if lines:
-                    last_line_data = json.loads(lines[-1])
-                    if last_line_data.get("id"):
-                         self.parent = last_line_data.get("id")
-                    elif last_line_data.get("children"):
-                         for child in last_line_data["children"]:
-                             if child.get("role") == "assistant":
-                                 self.parent = child.get("id")
-                                 break
-            except (json.JSONDecodeError, IndexError):
-                pass # Ignore errors parsing parent ID from non-stream
-
-            self.conversation.update_chat_history(prompt, parsed_response)
-            self.last_response = {"text": parsed_response}
-            return self.last_response if not raw else parsed_response # Return dict or raw string
-            
-        except CurlError as e: # Catch CurlError
-            raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {str(e)}") from e
-        except Exception as e: # Catch other potential exceptions (like HTTPError, JSONDecodeError)
-            err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
-            raise exceptions.FailedToGenerateResponseError(f"Request failed ({type(e).__name__}): {str(e)} - {err_text}") from e
-
-
-    def chat(
-        self,
-        prompt: str,
-        stream: bool = False,
-        optimizer: str = None,
-        conversationally: bool = False,
-        host: str = None,
-        options: dict = None,
-    ) -> Union[str, Generator[str, None, None]]: # Corrected return type hint
-        def for_stream_chat(): # Renamed inner function
-            # ask() yields dicts or strings when streaming
-            gen = self.ask(
-                prompt, 
-                stream=True, 
-                raw=False, # Ensure ask yields dicts
-                optimizer=optimizer, 
-                conversationally=conversationally,
-                host=host,
-                options=options
-            )
-            for response_dict in gen:
-                yield self.get_message(response_dict) # get_message expects dict
-
-        def for_non_stream_chat(): # Renamed inner function
-            # ask() returns dict or str when not streaming
-            response_data = self.ask(
-                prompt, 
-                stream=False, 
-                raw=False, # Ensure ask returns dict
-                optimizer=optimizer, 
-                conversationally=conversationally,
-                host=host,
-                options=options
-            )
-            return self.get_message(response_data) # get_message expects dict
-
-        return for_stream_chat() if stream else for_non_stream_chat() # Use renamed functions
-
-    def get_message(self, response: dict) -> str:
-        assert isinstance(response, dict), "Response should be of dict data-type only"
-        return response["text"]
-
-
-
-if __name__ == "__main__":
-    # Ensure curl_cffi is installed
-    print("-" * 80)
-    print(f"{'Model':<50} {'Status':<10} {'Response'}")
-    print("-" * 80)
-
-    for model in AllenAI.AVAILABLE_MODELS:
-        try:
-            # Auto-detect host
-            test_ai = AllenAI(model=model, timeout=60)
-            # Pass the host explicitly to display accurate error messages
-            response = test_ai.chat("Say 'Hello' in one word")
-            response_text = response
-            
-            if response_text and len(response_text.strip()) > 0:
-                status = "✓"
-                # Truncate response if too long
-                display_text = response_text.strip()[:50] + "..." if len(response_text.strip()) > 50 else response_text.strip()
-                print(f"{model:<50} {status:<10} {display_text} (host: {test_ai.host})")
-            else:
-                status = "✗"
-                display_text = "Empty or invalid response"
-                print(f"{model:<50} {status:<10} {display_text}")
-        except Exception as e:
-            print(f"{model:<50} {'✗':<10} {str(e)}")