webscout 8.3.1__py3-none-any.whl → 8.3.3__py3-none-any.whl

webscout/Provider/MiniMax.py

@@ -0,0 +1,207 @@
+import os
+import json
+import requests
+from typing import Any, Dict, Optional, Union, Generator
+from webscout.AIutel import sanitize_stream, Optimizers, Conversation, AwesomePrompts
+from webscout.AIbase import Provider
+from webscout import exceptions
+
+class MiniMax(Provider):
+    """
+    Provider for MiniMax-Reasoning-01 API, following the standard provider interface.
+    """
+    AVAILABLE_MODELS = ["MiniMax-Reasoning-01"]
+    API_URL = "https://api.minimaxi.chat/v1/text/chatcompletion_v2"
+    # TODO: Move API_KEY to env/config for security
+    API_KEY = os.environ.get("MINIMAX_API_KEY") or """eyJhbGciOiJSUzI1NiIsInR5cCI6IkpXVCJ9.eyJHcm91cE5hbWUiOiJtbyBuaSIsIlVzZXJOYW1lIjoibW8gbmkiLCJBY2NvdW50IjoiIiwiU3ViamVjdElEIjoiMTg3NjIwMDY0ODA2NDYzNTI0MiIsIlBob25lIjoiIiwiR3JvdXBJRCI6IjE4NzYyMDA2NDgwNjA0NDA5MzgiLCJQYWdlTmFtZSI6IiIsIk1haWwiOiJuaW1vQHN1YnN1cC52aXAiLCJDcmVhdGVUaW1lIjoiMjAyNS0wMS0wNyAxMToyNzowNyIsIlRva2VuVHlwZSI6MSwiaXNzIjoibWluaW1heCJ9.Ge1ZnpFPUfXVdMini0P_qXbP_9VYwzXiffG9DsNQck4GtYEOs33LDeAiwrVsrrLZfvJ2icQZ4sRZS54wmPuWua_Dav6pYJty8ZtahmUX1IuhlUX5YErhhCRAIy3J1xB8FkLHLyylChuBHpkNz6O6BQLmPqmoa-cOYK9Qrc6IDeu8SX1iMzO9-MSkcWNvkvpCF2Pf9tekBVWNKMDK6IZoMEPbtkaPXdDyP6l0M0e2AlL_E0oM9exg3V-ohAi8OTPFyqM6dcd4TwF-b9DULxfIsRFw401mvIxcTDWa42u2LULewdATVRD2BthU65tuRqEiWeFWMvFlPj2soMze_QIiUA"""
+    MODEL_CONTROL_DEFAULTS = {"tokens_to_generate": 40000, "temperature": 1, "top_p": 0.95}
+
+    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,
+        model: str = "MiniMax-Reasoning-01",
+        system_prompt: str = "You are a helpful assistant, always respond in english",
+    ):
+        if model not in self.AVAILABLE_MODELS:
+            raise ValueError(f"Invalid model: {model}. Choose from: {self.AVAILABLE_MODELS}")
+        self.model = model
+        self.api_url = self.API_URL
+        self.api_key = self.API_KEY
+        self.timeout = timeout
+        self.is_conversation = is_conversation
+        self.max_tokens_to_sample = max_tokens
+        self.last_response = {}
+        self.system_prompt = system_prompt
+        self.proxies = proxies
+        self.__available_optimizers = tuple(
+            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 _extract_content(chunk: Any) -> Optional[dict]:
+        if not isinstance(chunk, dict):
+            return None
+        choice = chunk.get('choices', [{}])[0]
+        delta = choice.get('delta', {})
+        content = delta.get('content')
+        reasoning = delta.get('reasoning_content')
+        result = {}
+        if content:
+            result['content'] = content
+        if reasoning:
+            result['reasoning_content'] = reasoning
+        return result if result else None
+
+    def ask(
+        self,
+        prompt: str,
+        stream: bool = True,
+        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}")
+        messages = [
+            {'role': 'system', 'content': self.system_prompt},
+            {'role': 'user', 'content': conversation_prompt}
+        ]
+        data = {
+            'model': self.model,
+            'messages': messages,
+            'stream': True,
+            'max_tokens': self.MODEL_CONTROL_DEFAULTS.get('tokens_to_generate', 512),
+            'temperature': self.MODEL_CONTROL_DEFAULTS.get('temperature', 1.0),
+            'top_p': self.MODEL_CONTROL_DEFAULTS.get('top_p', 1.0),
+        }
+        headers = {
+            'Content-Type': 'application/json',
+            'Authorization': f'Bearer {self.api_key}',
+        }
+        def for_stream():
+            try:
+                response = requests.post(
+                    self.api_url,
+                    headers=headers,
+                    data=json.dumps(data),
+                    stream=True,
+                    timeout=self.timeout,
+                    proxies=self.proxies if self.proxies else None
+                )
+                if not response.ok:
+                    raise exceptions.FailedToGenerateResponseError(
+                        f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
+                    )
+                streaming_response = ""
+                last_content = ""
+                last_reasoning = ""
+                in_think = False
+                processed_stream = sanitize_stream(
+                    response.iter_lines(),
+                    intro_value="data:",
+                    to_json=True,
+                    content_extractor=self._extract_content,
+                    raw=False # always process as dict for logic
+                )
+                for chunk in processed_stream:
+                    if not chunk:
+                        continue
+                    content = chunk.get('content') if isinstance(chunk, dict) else None
+                    reasoning = chunk.get('reasoning_content') if isinstance(chunk, dict) else None
+                    # Handle reasoning_content with <think> tags
+                    if reasoning and reasoning != last_reasoning:
+                        if not in_think:
+                            yield "<think>\n\n"
+                            in_think = True
+                        yield reasoning
+                        last_reasoning = reasoning
+                    # If we were in <think> and now have new content, close <think>
+                    if in_think and content and content != last_content:
+                        yield "</think>\n\n"
+                        in_think = False
+                    # Handle normal content
+                    if content and content != last_content:
+                        yield content
+                        streaming_response += content
+                        last_content = content
+                if not raw:
+                    self.last_response = {"text": streaming_response}
+                    self.conversation.update_chat_history(prompt, streaming_response)
+            except Exception as e:
+                raise exceptions.FailedToGenerateResponseError(f"Request failed: {e}")
+        def for_non_stream():
+            full_response = ""
+            for chunk in for_stream():
+                if isinstance(chunk, dict) and "text" in chunk:
+                    full_response += chunk["text"]
+                elif isinstance(chunk, str):
+                    full_response += chunk
+            if not raw:
+                self.last_response = {"text": full_response}
+                self.conversation.update_chat_history(prompt, full_response)
+                return {"text": full_response}
+            else:
+                return full_response
+        return for_stream() if stream else for_non_stream()
+
+    def chat(
+        self,
+        prompt: str,
+        stream: bool = True,
+        optimizer: str = None,
+        conversationally: bool = False,
+        raw: bool = False,
+    ) -> Union[str, Generator[str, None, None]]:
+        def for_stream():
+            for response in self.ask(
+                prompt, stream=True, raw=raw, optimizer=optimizer, conversationally=conversationally
+            ):
+                if raw:
+                    yield response
+                else:
+                    yield response
+        def for_non_stream():
+            result = self.ask(
+                prompt, stream=False, raw=raw, optimizer=optimizer, conversationally=conversationally
+            )
+            if raw:
+                return result
+            else:
+                return self.get_message(result)
+        return for_stream() if stream else for_non_stream()
+
+    def get_message(self, response: dict) -> str:
+        assert isinstance(response, dict), "Response should be of dict data-type only"
+        return response.get("text", "")
+
+if __name__ == "__main__":
+    ai = MiniMax()
+    resp = ai.chat("What is the capital of France?", stream=True, raw=False)
+    for chunk in resp:
+        print(chunk, end="", flush=True)

webscout/Provider/Nemotron.py

@@ -110,7 +110,8 @@ class NEMOTRON(Provider):
     def _make_request(
         self,
         message: str,
-        stream: bool = False
+        stream: bool = False,
+        raw: bool = False
     ) -> Generator[str, None, None]:
         """Make request to NEMOTRON API."""
         payload = {
@@ -131,10 +132,26 @@ class NEMOTRON(Provider):
                     timeout=self.timeout
                 ) as response:
                     response.raise_for_status()
-                    yield from sanitize_stream(
-                        response.iter_content(chunk_size=1024),
-                        to_json=False,
-                    )
+                    buffer = ""
+                    chunk_size = 32
+                    for chunk in response.iter_content(chunk_size=chunk_size):
+                        if not chunk:
+                            continue
+                        text = chunk.decode(errors="ignore")
+                        buffer += text
+                        while len(buffer) >= chunk_size:
+                            out = buffer[:chunk_size]
+                            buffer = buffer[chunk_size:]
+                            if out.strip():
+                                if raw:
+                                    yield out
+                                else:
+                                    yield out
+                    if buffer.strip():
+                        if raw:
+                            yield buffer
+                        else:
+                            yield buffer
             else:
                 response = self.session.post(
                     self.url,
@@ -143,7 +160,10 @@ class NEMOTRON(Provider):
                     timeout=self.timeout
                 )
                 response.raise_for_status()
-                yield response.text
+                if raw:
+                    yield response.text
+                else:
+                    yield response.text
 
         except requests.exceptions.RequestException as e:
             raise exceptions.ProviderConnectionError(f"Connection error: {str(e)}")
@@ -167,13 +187,20 @@ class NEMOTRON(Provider):
                 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}
+            for text in self._make_request(conversation_prompt, stream=True, raw=raw):
+                if raw:
+                    yield text
+                else:
+                    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
+            response_text = next(self._make_request(conversation_prompt, stream=False, raw=raw))
+            if raw:
+                self.last_response = response_text
+                return response_text
+            else:
+                self.last_response = {"text": response_text}
+                return self.last_response
 
         return for_stream() if stream else for_non_stream()
 
@@ -214,5 +241,6 @@ class NEMOTRON(Provider):
 if __name__ == "__main__":
     # Example usage
     nemotron = NEMOTRON()
-    response = nemotron.chat("Hello, how are you?", stream=False)
-    print(response)
+    response = nemotron.chat("write me about humans in points", stream=True)
+    for part in response:
+        print(part, end="", flush=True)

webscout/Provider/Netwrck.py

@@ -127,75 +127,58 @@ class Netwrck(Provider):
 
         def for_stream():
             try:
-                # Use curl_cffi session post with impersonate
                 response = self.session.post(
                     "https://netwrck.com/api/chatpred_or",
                     json=payload,
-                    # headers are set on the session
-                    # proxies are set on the session
                     timeout=self.timeout,
                     stream=True,
-                    impersonate="chrome110" # Use a common impersonation profile
+                    impersonate="chrome110"
                 )
-                response.raise_for_status() # Check for HTTP errors
-
-                streaming_text = ""
-                # Use sanitize_stream
-                processed_stream = sanitize_stream(
-                    data=response.iter_content(chunk_size=None), # Pass byte iterator
-                    intro_value=None, # No prefix
-                    to_json=False,    # It's text
-                    content_extractor=self._netwrck_extractor, # Use the quote stripper
-                    yield_raw_on_error=True
-                )
-                for content_chunk in processed_stream:
-                    if content_chunk and isinstance(content_chunk, str):
-                        streaming_text += content_chunk
-                        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(payload["query"], streaming_text)
-
-            except CurlError as e: # Catch CurlError
+                response.raise_for_status()
+                buffer = ""
+                chunk_size = 32
+                for chunk in response.iter_content(chunk_size=chunk_size):
+                    if not chunk:
+                        continue
+                    text = chunk.decode(errors="ignore")
+                    buffer += text
+                    while len(buffer) >= chunk_size:
+                        out = buffer[:chunk_size]
+                        buffer = buffer[chunk_size:]
+                        if out.strip():
+                            if raw:
+                                yield out
+                            else:
+                                yield {"text": out}
+                if buffer.strip():
+                    if raw:
+                        yield buffer
+                    else:
+                        yield {"text": buffer}
+                self.last_response = {"text": buffer}
+                self.conversation.update_chat_history(payload["query"], buffer)
+            except CurlError as e:
                 raise exceptions.ProviderConnectionError(f"Network error (CurlError): {str(e)}") from e
-            except Exception as e: # Catch other potential exceptions (like HTTPError)
+            except Exception as e:
                 err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
                 raise exceptions.ProviderConnectionError(f"Unexpected error ({type(e).__name__}): {str(e)} - {err_text}") from e
 
         def for_non_stream():
             try:
-                # Use curl_cffi session post with impersonate
                 response = self.session.post(
                     "https://netwrck.com/api/chatpred_or",
                     json=payload,
-                    # headers are set on the session
-                    # proxies are set on the session
                     timeout=self.timeout,
-                    impersonate="chrome110" # Use a common impersonation profile
-                )
-                response.raise_for_status() # Check for HTTP errors
-                
-                response_text_raw = response.text # Get raw text
-
-                # Process the text using sanitize_stream
-                processed_stream = sanitize_stream(
-                    data=response_text_raw,
-                    intro_value=None,
-                    to_json=False,
-                    content_extractor=self._netwrck_extractor
+                    impersonate="chrome110"
                 )
-                # Aggregate the single result
-                text = "".join(list(processed_stream))
-
-                self.last_response = {"text": text} # Store processed text
-                self.conversation.update_chat_history(prompt, text)
-
-                # Return dict or raw string
-                return text if raw else self.last_response
-
-            except CurlError as e: # Catch CurlError
+                response.raise_for_status()
+                response_text_raw = response.text
+                self.last_response = {"text": response_text_raw}
+                self.conversation.update_chat_history(prompt, response_text_raw)
+                return response_text_raw if raw else self.last_response
+            except CurlError as e:
                 raise exceptions.FailedToGenerateResponseError(f"Network error (CurlError): {str(e)}") from e
-            except Exception as e: # Catch other potential exceptions (like HTTPError)
+            except Exception as e:
                 err_text = getattr(e, 'response', None) and getattr(e.response, 'text', '')
                 raise exceptions.FailedToGenerateResponseError(f"Unexpected error ({type(e).__name__}): {str(e)} - {err_text}") from e
 

webscout/Provider/OPENAI/BLACKBOXAI.py

@@ -1,6 +1,5 @@
 # from pickle import NONE
 import requests
-import requests
 import random
 import string
 import base64
@@ -1044,6 +1043,3 @@ if __name__ == "__main__":
     )
     for chunk in response:
         print(chunk.choices[0].delta.content, end='', flush=True)
-    print()
-    print("Proxies on instance:", client.proxies)
-    print("Proxies on session:", client.session.proxies)