webscout 8.2.7__py3-none-any.whl → 8.2.8__py3-none-any.whl

webscout/Provider/HF_space/qwen_qwen2.py

@@ -0,0 +1,206 @@
+from dataclasses import dataclass
+from enum import Enum
+import requests
+import json
+import re
+import uuid
+from typing import Union, List, Dict, Generator, Optional, Any, TypedDict, Final
+
+# Type definitions
+class Role(Enum):
+    SYSTEM = "system"
+    USER = "user"
+    ASSISTANT = "assistant"
+
+class Message(TypedDict):
+    role: str
+    content: str
+
+class APIResponse(TypedDict):
+    event_id: str
+    fn_index: int
+    data: List[Any]
+
+class StreamData(TypedDict):
+    msg: str
+    output: Dict[str, Any]
+
+@dataclass
+class APIConfig:
+    url: Final[str] = "https://qwen-qwen2-72b-instruct.hf.space"
+    api_endpoint: Final[str] = "https://qwen-qwen2-72b-instruct.hf.space/queue/join?"
+
+@dataclass
+class RequestHeaders:
+    join: Dict[str, str]
+    data: Dict[str, str]
+
+    @classmethod
+    def create_default(cls, base_url: str) -> 'RequestHeaders':
+        common_headers = {
+            'accept-language': 'en-US,en;q=0.9',
+            'user-agent': 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36'
+        }
+        
+        return cls(
+            join={
+                **common_headers,
+                'accept': '*/*',
+                'content-type': 'application/json',
+                'origin': base_url,
+                'referer': f'{base_url}/',
+            },
+            data={
+                **common_headers,
+                'accept': 'text/event-stream',
+                'referer': f'{base_url}/',
+            }
+        )
+
+class QwenAPI:
+    def __init__(self, config: APIConfig = APIConfig()):
+        self.config = config
+        self.headers = RequestHeaders.create_default(config.url)
+
+    @staticmethod
+    def generate_session_hash() -> str:
+        """Generate a unique session hash."""
+        return str(uuid.uuid4()).replace('-', '')[:12]
+
+    @staticmethod
+    def format_prompt(messages: List[Message]) -> str:
+        """
+        Formats a list of messages into a single prompt string.
+
+        Args:
+            messages: A list of message dictionaries with "role" and "content" keys.
+
+        Returns:
+            str: The formatted prompt.
+        """
+        return "\n".join(f"{message['role']}: {message['content']}" for message in messages)
+
+    def create_sync_generator(
+        self,
+        model: str,
+        messages: List[Message],
+        proxy: Optional[str] = None,
+        **kwargs: Any
+    ) -> Generator[str, None, None]:
+        """
+        Synchronously streams responses from the Qwen_Qwen2_72B_Instruct API.
+
+        Args:
+            model: The model to use for the request.
+            messages: A list of message dictionaries with "role" and "content" keys.
+            proxy: Optional proxy URL for the request.
+            **kwargs: Additional keyword arguments.
+
+        Yields:
+            str: Text chunks from the API response.
+
+        Raises:
+            requests.exceptions.RequestException: If the API request fails.
+            json.JSONDecodeError: If the response cannot be parsed as JSON.
+        """
+        session_hash: str = self.generate_session_hash()
+
+        # Prepare the prompt
+        system_messages: List[str] = [
+            message["content"] 
+            for message in messages 
+            if message["role"] == Role.SYSTEM.value
+        ]
+        system_prompt: str = "\n".join(system_messages)
+        
+        user_messages: List[Message] = [
+            message 
+            for message in messages 
+            if message["role"] != Role.SYSTEM.value
+        ]
+        prompt: str = self.format_prompt(user_messages)
+
+        payload_join: Dict[str, Any] = {
+            "data": [prompt, [], system_prompt],
+            "event_data": None,
+            "fn_index": 0,
+            "trigger_id": 11,
+            "session_hash": session_hash
+        }
+
+        with requests.Session() as session:
+            # Send join request
+            response = session.post(
+                self.config.api_endpoint,
+                headers=self.headers.join,
+                json=payload_join
+            )
+            response.raise_for_status()
+            event_data: APIResponse = response.json()
+
+            # Prepare data stream request
+            url_data: str = f'{self.config.url}/queue/data'
+            params_data: Dict[str, str] = {'session_hash': session_hash}
+
+            # Send data stream request
+            full_response: str = ""
+            final_full_response: str = ""
+
+            with session.get(
+                url_data,
+                headers=self.headers.data,
+                params=params_data,
+                stream=True
+            ) as response:
+                response.raise_for_status()
+                
+                for line in response.iter_lines():
+                    if line:
+                        decoded_line: str = line.decode('utf-8')
+                        if decoded_line.startswith('data: '):
+                            try:
+                                json_data: StreamData = json.loads(decoded_line[6:])
+
+                                if json_data.get('msg') == 'process_generating':
+                                    if 'output' in json_data and 'data' in json_data['output']:
+                                        output_data: List[Any] = json_data['output']['data']
+                                        if len(output_data) > 1 and len(output_data[1]) > 0:
+                                            for item in output_data[1]:
+                                                if isinstance(item, list) and len(item) > 1:
+                                                    fragment: str = str(item[1])
+                                                    if not re.match(r'^\[.*\]$', fragment) and not full_response.endswith(fragment):
+                                                        full_response += fragment
+                                                        yield fragment
+
+                                if json_data.get('msg') == 'process_completed':
+                                    if 'output' in json_data and 'data' in json_data['output']:
+                                        output_data = json_data['output']['data']
+                                        if len(output_data) > 1 and len(output_data[1]) > 0:
+                                            final_full_response = output_data[1][0][1]
+                                            
+                                            if final_full_response.startswith(full_response):
+                                                final_full_response = final_full_response[len(full_response):]
+                                            
+                                            if final_full_response:
+                                                yield final_full_response
+                                        break
+
+                            except json.JSONDecodeError as e:
+                                print(f"Could not parse JSON: {decoded_line}")
+                                raise e
+
+
+def main() -> None:
+    messages: List[Message] = [
+        {"role": Role.SYSTEM.value, "content": "You are a helpful assistant."},
+        {"role": Role.USER.value, "content": "LOL"}
+    ]
+
+    api = QwenAPI()
+    for text in api.create_sync_generator("qwen-qwen2-72b-instruct", messages):
+        print(text, end="", flush=True)
+    print("\n---\n")
+
+
+if __name__ == "__main__":
+    main()

webscout/Provider/HeckAI.py

@@ -0,0 +1,285 @@
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+import json
+import uuid
+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 HeckAI(Provider):
+    """
+    A class to interact with the HeckAI API with LitAgent user-agent.
+    """
+
+    AVAILABLE_MODELS = [
+        "google/gemini-2.0-flash-001",
+        "deepseek/deepseek-chat",
+        "deepseek/deepseek-r1",
+        "openai/gpt-4o-mini",
+        "openai/gpt-4.1-mini",
+        "x-ai/grok-3-mini-beta",
+        "meta-llama/llama-4-scout"
+    ]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 2049, # Note: max_tokens is not used by this API
+        timeout: int = 30,
+        intro: str = None,
+        filepath: str = None,
+        update_file: bool = True,
+        proxies: dict = {},
+        history_offset: int = 10250,
+        act: str = None,
+        model: str = "google/gemini-2.0-flash-001",
+        language: str = "English"
+    ):
+        """Initializes the HeckAI API client."""
+        if model not in self.AVAILABLE_MODELS:
+            raise ValueError(f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}")
+            
+        self.url = "https://api.heckai.weight-wave.com/api/ha/v1/chat"
+        self.session_id = str(uuid.uuid4())
+        self.language = language
+        
+        # Use LitAgent (keep if needed for other headers or logic)
+        self.headers = {
+            'Content-Type': 'application/json',
+            'Origin': 'https://heck.ai', # Keep Origin
+            'Referer': 'https://heck.ai/', # Keep Referer
+        }
+        
+        # Initialize curl_cffi Session
+        self.session = Session()
+        # Update curl_cffi session headers and proxies
+        self.session.headers.update(self.headers)
+        self.session.proxies = proxies # Assign proxies directly
+
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.timeout = timeout
+        self.last_response = {}
+        self.model = model
+        self.previous_question = None
+        self.previous_answer = None
+
+        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 ask(
+        self,
+        prompt: str,
+        stream: bool = False, # API supports streaming
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> 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}")
+
+        # Payload construction
+        payload = {
+            "model": self.model,
+            "question": conversation_prompt,
+            "language": self.language,
+            "sessionId": self.session_id,
+            "previousQuestion": self.previous_question,
+            "previousAnswer": self.previous_answer,
+            "imgUrls": [],
+            "superSmartMode": False  # Added based on API request data
+        }
+        
+        # Store this message as previous for next request
+        self.previous_question = conversation_prompt
+
+        def for_stream():
+            streaming_text = "" # Initialize outside try block
+            try:
+                # Use curl_cffi session post with impersonate
+                response = self.session.post(
+                    self.url, 
+                    # headers are set on the session
+                    data=json.dumps(payload), 
+                    stream=True, 
+                    timeout=self.timeout,
+                    impersonate="chrome110" # Use a common impersonation profile
+                )
+                response.raise_for_status() # Check for HTTP errors
+
+                # Use sanitize_stream to process the stream
+                processed_stream = sanitize_stream(
+                data=response.iter_content(chunk_size=1024), # Pass byte iterator
+                intro_value="data: ",  # Prefix to remove (note the space)
+                to_json=False,        # Content is text
+                start_marker="data: [ANSWER_START]",
+                end_marker="data: [ANSWER_DONE]",
+                skip_markers=["data: [RELATE_Q_START]", "data: [RELATE_Q_DONE]", "data: [REASON_START]", "data: [REASON_DONE]"],
+                yield_raw_on_error=True,
+                strip_chars=" \n\r\t"  # Strip whitespace characters from chunks
+                )
+
+                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
+                
+                # Only update history if we received a valid response
+                if streaming_text:
+                    # Update history and previous answer after stream finishes
+                    self.previous_answer = streaming_text
+                    # Convert to simple text before updating conversation
+                    try:
+                        # Ensure content is valid before updating conversation
+                        if streaming_text and isinstance(streaming_text, str):
+                            # Sanitize the content to ensure it's valid
+                            sanitized_text = streaming_text.strip()
+                            if sanitized_text:  # Only update if we have non-empty content
+                                self.conversation.update_chat_history(prompt, sanitized_text)
+                    except Exception as e:
+                        # If conversation update fails, log but don't crash
+                        print(f"Warning: Failed to update conversation history: {str(e)}")
+                    
+            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 for_non_stream():
+            # Aggregate the stream using the updated for_stream logic
+            full_text = ""
+            try:
+                # Ensure raw=False so for_stream yields dicts
+                for chunk_data in for_stream():
+                    if isinstance(chunk_data, dict) and "text" in chunk_data:
+                        full_text += chunk_data["text"]
+                    # Handle raw string case if raw=True was passed
+                    elif raw and isinstance(chunk_data, str):
+                         full_text += chunk_data
+            except Exception as e:
+                 # If aggregation fails but some text was received, use it. Otherwise, re-raise.
+                 if not full_text:
+                     raise exceptions.FailedToGenerateResponseError(f"Failed to get non-stream response: {str(e)}") from e
+
+            # Return the final aggregated response dict or raw string
+            self.last_response = {"text": full_text} # Update last_response here
+            return full_text if raw else self.last_response
+
+
+        return for_stream() if stream else for_non_stream()
+
+    @staticmethod
+    def fix_encoding(text):
+        if isinstance(text, dict) and "text" in text:
+            try:
+                text["text"] = text["text"].encode("latin1").decode("utf-8")
+                return text
+            except (UnicodeError, AttributeError) as e:
+                return text
+        elif isinstance(text, str):
+            try:
+                return text.encode("latin1").decode("utf-8")
+            except (UnicodeError, AttributeError) as e:
+                return text
+        return text
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[str, Generator[str, None, None]]: # Corrected return type hint
+        def for_stream_chat():
+            # ask() yields dicts or strings when streaming
+            gen = self.ask(
+                prompt, stream=True, raw=False, # Ensure ask yields dicts
+                optimizer=optimizer, conversationally=conversationally
+            )
+            for response_dict in gen:
+                yield self.get_message(response_dict) # get_message expects dict
+                
+        def for_non_stream_chat():
+            # 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
+            )
+            return self.get_message(response_data) # get_message expects dict
+            
+        return for_stream_chat() if stream else for_non_stream_chat()
+
+    def get_message(self, response: dict) -> str:
+        # Validate response format
+        if not isinstance(response, dict):
+            raise TypeError(f"Expected dict response, got {type(response).__name__}")
+        
+        # Handle missing text key gracefully
+        if "text" not in response:
+            return ""
+        
+        # Ensure text is a string
+        text = response["text"]
+        if not isinstance(text, str):
+            return str(text)
+            
+        return text
+
+if __name__ == "__main__":
+    # Ensure curl_cffi is installed
+    print("-" * 80)
+    print(f"{'Model':<50} {'Status':<10} {'Response'}")
+    print("-" * 80)
+
+    for model in HeckAI.AVAILABLE_MODELS:
+        try:
+            test_ai = HeckAI(model=model, timeout=60)
+            # Use non-streaming mode first to avoid potential streaming issues
+            try:
+                response_text = test_ai.chat("Say 'Hello' in one word", stream=False)
+                print(f"\r{model:<50} {'✓':<10} {response_text.strip()[:50]}")
+            except Exception as e1:
+                # Fall back to streaming if non-streaming fails
+                print(f"\r{model:<50} {'Testing stream...':<10}", end="", flush=True)
+                response = test_ai.chat("Say 'Hello' in one word", stream=True)
+                response_text = ""
+                for chunk in response:
+                    if chunk and isinstance(chunk, str):
+                        response_text += chunk
+                
+                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"\r{model:<50} {status:<10} {display_text}")
+                else:
+                    raise ValueError("Empty or invalid response")
+        except Exception as e:
+            print(f"\r{model:<50} {'✗':<10} {str(e)}")