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

webscout/Provider/Hunyuan.py

@@ -0,0 +1,283 @@
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+import json
+import os
+from typing import Any, Dict, Optional, Generator, Union
+import time
+import uuid
+import re
+
+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 Hunyuan(Provider):
+    """
+    A class to interact with the Tencent Hunyuan API with LitAgent user-agent.
+    """
+
+    AVAILABLE_MODELS = [
+        "hunyuan-t1-latest",
+        # Add more models as they become available
+    ]
+
+    def __init__(
+        self,
+        is_conversation: bool = True,
+        max_tokens: int = 2048, # 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 = "hunyuan-t1-latest",
+        browser: str = "chrome", # Note: browser fingerprinting might be less effective with impersonate
+        api_key: str = None,
+        system_prompt: str = "You are a helpful assistant.",
+    ):
+
+        """Initializes the Hunyuan API client."""
+        if model not in self.AVAILABLE_MODELS:
+            raise ValueError(f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}")
+            
+        self.url = "https://llm.hunyuan.tencent.com/aide/api/v2/triton_image/demo_text_chat/"
+        
+        # Initialize LitAgent (keep if needed for other headers or logic)
+        self.agent = LitAgent()
+        # Fingerprint generation might be less relevant with impersonate
+        self.fingerprint = self.agent.generate_fingerprint(browser) 
+        
+        # Use the fingerprint for headers (keep relevant ones)
+        self.headers = {
+            "Accept": "*/*",
+            "Accept-Language": self.fingerprint["accept_language"], # Keep Accept-Language
+            "Content-Type": "application/json",
+            "DNT": "1", # Keep DNT
+            "Origin": "https://llm.hunyuan.tencent.com", # Keep Origin
+            "Referer": "https://llm.hunyuan.tencent.com/", # Keep Referer
+            "Sec-Fetch-Dest": "empty", # Keep Sec-Fetch-*
+            "Sec-Fetch-Mode": "cors",
+            "Sec-Fetch-Site": "same-origin",
+            "Sec-GPC": "1", # Keep Sec-GPC
+        }
+        
+        # Add authorization if API key is provided
+        if api_key:
+            self.headers["Authorization"] = f"Bearer {api_key}"
+        else:
+            # Default test key (may not work long-term)
+            self.headers["Authorization"] = "Bearer 7auGXNATFSKl7dF"
+        
+        # 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.system_message = system_prompt
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.timeout = timeout
+        self.last_response = {}
+        self.model = model
+
+        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 refresh_identity(self, browser: str = None):
+        """
+        Refreshes the browser identity fingerprint.
+        
+        Args:
+            browser: Specific browser to use for the new fingerprint
+        """
+        browser = browser or self.fingerprint.get("browser_type", "chrome")
+        self.fingerprint = self.agent.generate_fingerprint(browser)
+        
+        # Update headers with new fingerprint (only relevant ones)
+        self.headers.update({
+            "Accept-Language": self.fingerprint["accept_language"],
+        })
+        
+        # Update session headers
+        self.session.headers.update(self.headers) # Update only relevant headers
+        
+        return self.fingerprint
+
+    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}")
+
+        # Generate a unique query ID for each request
+        query_id = ''.join(re.findall(r'[a-z0-9]', str(uuid.uuid4())[:18]))
+    
+
+        # Payload construction
+        payload = {
+            "stream": True, # API seems to require stream=True based on response format
+            "model": self.model,
+            "query_id": query_id,
+            "messages": [
+                {"role": "system", "content": self.system_message},
+                {"role": "user", "content": "Always response in English\n\n" + conversation_prompt},
+            ],
+            "stream_moderation": True,
+            "enable_enhancement": False
+        }
+
+        def for_stream():
+            streaming_text = "" # Initialize outside try block
+            try:
+                # Use curl_cffi session post with impersonate
+                response = self.session.post(
+                    self.url, 
+                    data=json.dumps(payload), 
+                    stream=True, 
+                    timeout=self.timeout, 
+                    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 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:
+                                    choice = json_data['choices'][0]
+                                    if 'delta' in choice and 'content' in choice['delta']:
+                                        content = choice['delta']['content']
+                                        if content: # Ensure content is not None or empty
+                                            streaming_text += content
+                                            resp = dict(text=content)
+                                            # Yield dict or raw string chunk
+                                            yield resp if not raw else content
+                        except (json.JSONDecodeError, UnicodeDecodeError):
+                            continue # Ignore lines that are not valid JSON or cannot be decoded
+                    
+                # Update history after stream finishes
+                self.last_response = {"text": streaming_text}
+                self.conversation.update_chat_history(prompt, 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 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
+
+            # last_response and history are updated within for_stream
+            # Return the final aggregated response dict or raw string
+            return full_text if raw else self.last_response
+
+
+        # Since the API endpoint suggests streaming, always call the stream generator.
+        # The non-stream wrapper will handle aggregation if stream=False.
+        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]]:
+        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:
+        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 Hunyuan.AVAILABLE_MODELS:
+        try:
+            test_ai = Hunyuan(model=model, timeout=60)
+            response = test_ai.chat("Say 'Hello' in one word", stream=True)
+            response_text = ""
+            for chunk in response:
+                response_text += chunk
+            
+            if response_text and len(response_text.strip()) > 0:
+                status = "✓"
+                # Clean and truncate response
+                clean_text = response_text.strip().encode('utf-8', errors='ignore').decode('utf-8')
+                display_text = clean_text[:50] + "..." if len(clean_text) > 50 else clean_text
+            else:
+                status = "✗"
+                display_text = "Empty or invalid response"
+            print(f"\r{model:<50} {status:<10} {display_text}")
+        except Exception as e:
+            print(f"\r{model:<50} {'✗':<10} {str(e)}")

webscout/Provider/Jadve.py

@@ -0,0 +1,291 @@
+from curl_cffi.requests import Session
+from curl_cffi import CurlError
+import json
+import re
+from typing import Union, Any, Dict, Optional, Generator
+
+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
+
+class JadveOpenAI(Provider):
+    """
+    A class to interact with the OpenAI API through jadve.com using the streaming endpoint.
+    """
+
+    AVAILABLE_MODELS = ["gpt-4o-mini"]
+
+    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,
+        model: str = "gpt-4o-mini",
+        system_prompt: str = "You are a helpful AI assistant." # Note: system_prompt is not used by this API
+    ):
+        """
+        Initializes the JadveOpenAI client.
+
+        Args:
+            is_conversation (bool, optional): Enable conversational mode. Defaults to True.
+            max_tokens (int, optional): Maximum tokens for generation. Defaults to 600.
+            timeout (int, optional): HTTP request timeout in seconds. Defaults to 30.
+            intro (str, optional): Introductory prompt text. Defaults to None.
+            filepath (str, optional): Path to conversation history file. Defaults to None.
+            update_file (bool, optional): Whether to update the conversation history file. Defaults to True.
+            proxies (dict, optional): Proxies for HTTP requests. Defaults to {}.
+            history_offset (int, optional): Limit for conversation history. Defaults to 10250.
+            act (str|int, optional): Act key for AwesomePrompts. Defaults to None.
+            model (str, optional): AI model to be used. Defaults to "gpt-4o-mini".
+            system_prompt (str, optional): System prompt text. Defaults to "You are a helpful AI assistant."
+        """
+        if model not in self.AVAILABLE_MODELS:
+            raise ValueError(f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}")
+
+        # Initialize curl_cffi Session
+        self.session = Session()
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.api_endpoint = "https://openai.jadve.com/stream"
+        self.stream_chunk_size = 64
+        self.timeout = timeout
+        self.last_response = {}
+        self.model = model
+        self.system_prompt = system_prompt
+
+        # Headers for API requests
+        self.headers = {
+            "accept": "*/*",
+            "accept-language": "en-US,en;q=0.9,en-IN;q=0.8",
+            "content-type": "application/json",
+            "dnt": "1",
+            "origin": "https://jadve.com",
+            "priority": "u=1, i", # Keep priority header if needed
+            "referer": "https://jadve.com/",
+            "sec-fetch-dest": "empty",
+            "sec-fetch-mode": "cors",
+            "sec-fetch-site": "same-site",
+            "x-authorization": "Bearer" # Keep custom headers
+        }
+        
+        # Update curl_cffi session headers and proxies
+        self.session.headers.update(self.headers)
+        self.session.proxies = proxies # Assign proxies directly
+
+        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
+
+    @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,
+        stream: bool = False, # API supports streaming
+        raw: bool = False,
+        optimizer: str = None,
+        conversationally: bool = False,
+    ) -> Union[dict, Generator[dict, None, None]]:
+        """
+        Chat with AI.
+
+        Args:
+            prompt (str): Prompt to be sent.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            raw (bool, optional): Return raw content chunks. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name. Defaults to None.
+            conversationally (bool, optional): Flag for conversational optimization. Defaults to False.
+        Returns:
+            dict or generator: A dictionary with the generated text or a generator yielding text chunks.
+        """
+        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 {list(self.__available_optimizers)}")
+
+        payload = {
+            "messages": [
+                {"role": "system", "content": self.system_prompt},
+                {"role": "user", "content": [{"type": "text", "text": conversation_prompt}]}
+            ],
+            "model": self.model,
+            "botId": "",
+            "chatId": "",
+            "stream": True, # API endpoint suggests streaming is default/required
+            "temperature": 0.7,
+            "returnTokensUsage": True,
+            "useTools": False
+        }
+
+        def for_stream():
+            full_response_text = "" # Initialize outside try block
+            try:
+                # Use curl_cffi session post with impersonate
+                response = self.session.post(
+                    self.api_endpoint, 
+                    # headers are set on the session
+                    json=payload, 
+                    stream=True, 
+                    timeout=self.timeout,
+                    # proxies are set on the session
+                    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 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}
+                self.conversation.update_chat_history(prompt, full_response_text)
+
+            except CurlError as e: # Catch CurlError
+                raise exceptions.FailedToGenerateResponseError(f"Request failed (CurlError): {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"Failed to generate response ({type(e).__name__}): {e} - {err_text}") from e
+
+
+        def for_non_stream():
+            # Aggregate the stream using the updated for_stream logic
+            collected_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:
+                        collected_text += chunk_data["text"]
+                    # Handle raw string case if raw=True was passed
+                    elif raw and isinstance(chunk_data, str):
+                         collected_text += chunk_data
+            except Exception as e:
+                 # If aggregation fails but some text was received, use it. Otherwise, re-raise.
+                 if not collected_text:
+                     raise exceptions.FailedToGenerateResponseError(f"Failed to get non-stream response: {str(e)}") from e
+
+            # last_response and history are updated within for_stream
+            # Return the final aggregated response dict or raw string
+            return collected_text if raw else self.last_response
+
+
+        # Since the API endpoint suggests streaming, always call the stream generator.
+        # The non-stream wrapper will handle aggregation if stream=False.
+        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 a chat response (string). 
+
+        Args:
+            prompt (str): Prompt to be sent.
+            stream (bool, optional): Flag for streaming response. Defaults to False.
+            optimizer (str, optional): Prompt optimizer name. Defaults to None.
+            conversationally (bool, optional): Flag for conversational optimization. Defaults to False.
+        Returns:
+            str or generator: Generated response string or generator yielding response chunks.
+        """
+        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:
+        """
+        Retrieves message from the response.
+
+        Args:
+            response (dict): Response from the ask() method.
+        Returns:
+            str: Extracted text.
+        """
+        assert isinstance(response, dict), "Response should be of dict data-type only"
+        # Extractor handles formatting
+        return response.get("text", "")
+
+if __name__ == "__main__":
+    # Ensure curl_cffi is installed
+    print("-" * 80)
+    print(f"{'Model':<50} {'Status':<10} {'Response'}")
+    print("-" * 80)
+
+    for model in JadveOpenAI.AVAILABLE_MODELS:
+        try:
+            test_ai = JadveOpenAI(model=model, timeout=60)
+            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()
+            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)}")