webscout 8.2.4__py3-none-any.whl → 8.2.6__py3-none-any.whl

webscout/Provider/asksteve.py

@@ -1,8 +1,11 @@
-import requests
+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
-from webscout.AIbase import Provider
+from webscout.AIutel import AwesomePrompts, sanitize_stream # Import sanitize_stream
+from webscout.AIbase import Provider 
 from webscout.litagent import LitAgent
 
 class AskSteve(Provider):
@@ -36,7 +39,7 @@ class AskSteve(Provider):
             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 = requests.Session()
+        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"
@@ -73,7 +76,15 @@ class AskSteve(Provider):
             is_conversation, self.max_tokens_to_sample, filepath, update_file
         )
         self.conversation.history_offset = history_offset
-        self.session.proxies = proxies
+        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,
@@ -115,37 +126,43 @@ class AskSteve(Provider):
             "prompt": conversation_prompt
         }
 
-        def for_stream():
+
+        # 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=True,
+                stream=False, # API doesn't stream
                 timeout=self.timeout,
+                impersonate="chrome120" # Add impersonate
             )
-            if not response.ok:
-                raise Exception(
-                    f"Failed to generate response - ({response.status_code}, {response.reason}) - {response.text}"
-                )
-            
-            response_data = response.json()
-            if "candidates" in response_data and len(response_data["candidates"]) > 0:
-                text = response_data["candidates"][0]["content"]["parts"][0]["text"]
-                self.last_response.update(dict(text=text))
-                yield dict(text=text) if not raw else text
-            else:
-                raise Exception("No response generated")
+            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)
             )
+            # Return dict or raw string based on raw flag
+            return text if raw else self.last_response
 
-        def for_non_stream():
-            for _ in for_stream():
-                pass
-            return self.last_response
-
-        return for_stream() if stream else for_non_stream()
+        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,
@@ -164,23 +181,15 @@ class AskSteve(Provider):
             str: Response generated
         """
 
-        def for_stream():
-            for response in self.ask(
-                prompt, True, optimizer=optimizer, conversationally=conversationally
-            ):
-                yield self.get_message(response)
-
-        def for_non_stream():
-            return self.get_message(
-                self.ask(
-                    prompt,
-                    False,
-                    optimizer=optimizer,
-                    conversationally=conversationally,
-                )
-            )
-
-        return for_stream() if stream else for_non_stream()
+        # Since ask() doesn't truly stream, we just call it once.
+        response_data = self.ask(
+            prompt,
+            stream=False, # Always False for this API
+            raw=False,    # Get the dict back
+            optimizer=optimizer,
+            conversationally=conversationally,
+        )
+        return self.get_message(response_data)
 
     def get_message(self, response: dict) -> str:
         """Retrieves message only from response
@@ -192,12 +201,12 @@ class AskSteve(Provider):
             str: Message extracted
         """
         assert isinstance(response, dict), "Response should be of dict data-type only"
-        return response["text"]
+        return response.get("text", "") # Use .get for safety
 
 
 if __name__ == "__main__":
     from rich import print
     ai = AskSteve()
-    response = ai.chat("hi", stream=True)
+    response = ai.chat("write a short poem about AI", stream=True)
     for chunk in response:
         print(chunk, end="", flush=True) 

webscout/Provider/cerebras.py

@@ -1,10 +1,11 @@
 
 import re
-import requests
+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
+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
@@ -17,7 +18,9 @@ class Cerebras(Provider):
     AVAILABLE_MODELS = [
         "llama3.1-8b",
         "llama-3.3-70b",
-        "deepseek-r1-distill-llama-70b"
+        "deepseek-r1-distill-llama-70b",
+        "llama-4-scout-17b-16e-instruct"
+
     ]
 
     def __init__(
@@ -49,6 +52,8 @@ class Cerebras(Provider):
         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)
@@ -74,6 +79,9 @@ class Cerebras(Provider):
             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
@@ -88,7 +96,14 @@ class Cerebras(Provider):
         """Refines the input text by removing surrounding quotes."""
         return text.strip('"')
 
-    def get_demo_api_key(self, cookie_path: str) -> str:
+    @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:
@@ -114,17 +129,19 @@ class Cerebras(Provider):
         }
 
         try:
-            response = requests.post(
+            # 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()["data"]["GetMyDemoApiKey"]
+            api_key = response.json().get("data", {}).get("GetMyDemoApiKey")
             return api_key
-        except requests.exceptions.RequestException as e:
+        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.")
@@ -144,41 +161,48 @@ class Cerebras(Provider):
         }
 
         try:
-            response = requests.post(
+            # 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
+                timeout=self.timeout,
+                impersonate="chrome120" # Add impersonate
             )
             response.raise_for_status()
 
             if stream:
                 def generate_stream():
-                    for line in response.iter_lines():
-                        if line:
-                            line = line.decode('utf-8')
-                            if line.startswith('data:'):
-                                try:
-                                    data = json.loads(line[6:])
-                                    if data.get('choices') and data['choices'][0].get('delta', {}).get('content'):
-                                        content = data['choices'][0]['delta']['content']
-                                        yield content
-                                except json.JSONDecodeError:
-                                    continue
+                    # 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()
-                return response_json['choices'][0]['message']['content']
+                # 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 requests.exceptions.RequestException as e:
+        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]:
@@ -199,11 +223,23 @@ class Cerebras(Provider):
 
         try:
             response = self._make_request(messages, stream)
-            if stream:
-                return response
             
-            self.last_response = response
-            return response
+            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}")
@@ -216,14 +252,24 @@ class Cerebras(Provider):
         conversationally: bool = False,
     ) -> Union[str, Generator]:
         """Chat with the model."""
-        response = self.ask(prompt, stream, optimizer, conversationally)
+        # 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:
-            return response
-        return response
+            # 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."""
-        return 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__":
@@ -231,7 +277,7 @@ if __name__ == "__main__":
     
     # Example usage
     cerebras = Cerebras(
-        cookie_path='cookie.json',
+        cookie_path=r'cookies.json',
         model='llama3.1-8b',
         system_prompt="You are a helpful AI assistant."
     )

webscout/Provider/chatglm.py

@@ -1,11 +1,12 @@
-import requests
+from curl_cffi import CurlError
+from curl_cffi.requests import Session
 import json
 from typing import Any, Dict, Optional, Generator, List, Union
 import uuid
 
 from webscout.AIutel import Optimizers
 from webscout.AIutel import Conversation
-from webscout.AIutel import AwesomePrompts
+from webscout.AIutel import AwesomePrompts, sanitize_stream # Import sanitize_stream
 from webscout.AIbase import Provider
 from webscout import exceptions
 from webscout.litagent import LitAgent
@@ -29,7 +30,7 @@ class ChatGLM(Provider):
         plus_model: bool = True,
     ):
         """Initializes the ChatGLM API client."""
-        self.session = requests.Session()
+        self.session = Session() # Use curl_cffi Session
         self.is_conversation = is_conversation
         self.max_tokens_to_sample = max_tokens
         self.api_endpoint = "https://chatglm.cn/chatglm/mainchat-api/guest/stream"
@@ -55,7 +56,7 @@ class ChatGLM(Provider):
             if callable(getattr(Optimizers, method)) and not method.startswith("__")
         )
         self.session.headers.update(self.headers)
-        Conversation.intro = (
+        Conversation.intro = ( # type: ignore
             AwesomePrompts().get_act(
                 act, raise_not_found=True, default=None, case_insensitive=True
             )
@@ -66,7 +67,16 @@ class ChatGLM(Provider):
             is_conversation, self.max_tokens_to_sample, filepath, update_file
         )
         self.conversation.history_offset = history_offset
-        self.session.proxies = proxies
+        self.session.proxies = proxies # Assign proxies directly
+
+    @staticmethod
+    def _chatglm_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from ChatGLM stream JSON objects."""
+        if isinstance(chunk, dict):
+            parts = chunk.get('parts', [])
+            if parts and isinstance(parts[0].get('content'), list) and parts[0]['content']:
+                return parts[0]['content'][0].get('text')
+        return None
 
     def ask(
         self,
@@ -119,45 +129,45 @@ class ChatGLM(Provider):
         }
 
         def for_stream():
+            streaming_text = "" # Initialize outside try block
+            last_processed_content = "" # Track the last processed content
             try:
-                with self.session.post(
-                    self.api_endpoint, json=payload, stream=True, timeout=self.timeout
-                ) as response:
-                    response.raise_for_status()
-                    
-                    streaming_text = ""
-                    last_processed_content = "" # Track the last processed content
-                    for chunk in response.iter_lines():
-                        if chunk:
-                            decoded_chunk = chunk.decode('utf-8')
-                            if decoded_chunk.startswith('data: '):
-                                try:
-                                    json_data = json.loads(decoded_chunk[6:])
-                                    parts = json_data.get('parts', [])
-                                    if parts:
-                                        content = parts[0].get('content', [])
-                                        if content:
-                                            text = content[0].get('text', '')
-                                            new_text = text[len(last_processed_content):]
-                                            if new_text:  # Check for new content
-                                                streaming_text += new_text
-                                                last_processed_content = text
-                                                yield new_text if raw else dict(text=new_text)
-                                except json.JSONDecodeError:
-                                    continue
+                response = self.session.post(
+                    self.api_endpoint, json=payload, stream=True, timeout=self.timeout,
+                    impersonate="chrome120" # Add impersonate
+                )
+                response.raise_for_status()
+
+                # 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._chatglm_extractor, # Use the specific extractor
+                    yield_raw_on_error=False # Skip non-JSON lines or lines where extractor fails
+                )
 
+                for current_full_text in processed_stream:
+                    # current_full_text is the full text extracted by _chatglm_extractor
+                    if current_full_text and isinstance(current_full_text, str):
+                        new_text = current_full_text[len(last_processed_content):]
+                        if new_text: # Check for new content
+                            streaming_text += new_text
+                            last_processed_content = current_full_text # Update tracker
+                            yield new_text if raw else dict(text=new_text)
+
+            except CurlError as e:
+                raise exceptions.ProviderConnectionError(f"Request failed (CurlError): {e}") from e
+            except Exception as e:
+                raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred ({type(e).__name__}): {e}") from e
+            finally:
+                # Update history after stream finishes or fails
+                if streaming_text:
                     self.last_response.update(dict(text=streaming_text))
                     self.conversation.update_chat_history(
                         prompt, self.get_message(self.last_response)
                     )
 
-            except requests.exceptions.RequestException as e:
-                raise exceptions.ProviderConnectionError(f"Request failed: {e}")
-            except json.JSONDecodeError as e:
-                raise exceptions.InvalidResponseError(f"Failed to decode JSON: {e}")
-            except Exception as e:
-                raise exceptions.FailedToGenerateResponseError(f"An unexpected error occurred: {e}")
-
         def for_non_stream():
             for _ in for_stream():
                 pass

webscout/Provider/elmo.py

@@ -1,13 +1,14 @@
 from curl_cffi.requests import Session
 from curl_cffi import CurlError
 import json
-from typing import Union, Any, Dict, Generator
+from typing import Optional, Union, Any, Dict, Generator
 from webscout import exceptions
 from webscout.AIutel import Optimizers
-from webscout.AIutel import Conversation
+from webscout.AIutel import Conversation, sanitize_stream # Import sanitize_stream
 from webscout.AIutel import AwesomePrompts
 from webscout.AIbase import Provider
 from webscout.litagent import LitAgent
+import re # Import re for the extractor
 
 
 class Elmo(Provider):
@@ -84,6 +85,17 @@ class Elmo(Provider):
         )
         self.conversation.history_offset = history_offset
 
+    @staticmethod
+    def _elmo_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from the Elmo 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,
@@ -144,7 +156,7 @@ class Elmo(Provider):
         }
 
         def for_stream():
-            full_response = ""  # Initialize outside try block
+            streaming_text = "" # Initialize outside try block
             try:
                 # Use curl_cffi session post with impersonate
                 # Note: The API expects 'text/plain' but we send JSON.
@@ -159,40 +171,32 @@ class Elmo(Provider):
                 )
                 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')
-                            if line.startswith('0:'):
-                                # Extract content after '0:"' and before the closing '"'
-                                match = line.split(':"', 1)
-                                if len(match) > 1:
-                                    chunk = match[1]
-                                    if chunk.endswith('"'):
-                                        chunk = chunk[:-1]  # Remove trailing quote
-
-                                    # Handle potential escape sequences like \\n
-                                    formatted_output = chunk.encode().decode('unicode_escape')
+                # 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._elmo_extractor, # Use the specific extractor
+                    yield_raw_on_error=True
+                )
 
-                                    if formatted_output:  # Ensure content is not None or empty
-                                        full_response += formatted_output
-                                        resp = dict(text=formatted_output)
-                                        # Yield dict or raw string chunk
-                                        yield resp if not raw else formatted_output
-                        except (UnicodeDecodeError, IndexError):
-                            continue  # Ignore lines that cannot be decoded or parsed
+                for content_chunk in processed_stream:
+                    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
 
-                # Update history after stream finishes
-                self.last_response = dict(text=full_response)
-                self.conversation.update_chat_history(
-                    prompt, full_response
-                )
             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
+            finally:
+                # Update history after stream finishes
+                self.last_response = dict(text=streaming_text)
+                self.conversation.update_chat_history(
+                    prompt, streaming_text
+                )
 
         def for_non_stream():
             # Aggregate the stream using the updated for_stream logic
@@ -210,7 +214,9 @@ class Elmo(Provider):
                  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
+            # Update last_response and history *after* aggregation for non-stream
+            self.last_response = {"text": collected_text}
+            self.conversation.update_chat_history(prompt, collected_text)
             # Return the final aggregated response dict or raw string
             return collected_text if raw else self.last_response
 
@@ -265,7 +271,7 @@ class Elmo(Provider):
             str: Message extracted
         """
         assert isinstance(response, dict), "Response should be of dict data-type only"
-        return response["text"]
+        return response.get("text", "") # Use .get for safety
 
 
 if __name__ == "__main__":