webscout 8.2.4__py3-none-any.whl → 8.2.5__py3-none-any.whl

webscout/Provider/HeckAI.py

@@ -120,7 +120,7 @@ class HeckAI(Provider):
 
         def for_stream():
             streaming_text = "" # Initialize outside try block
-            in_answer = False # Initialize outside try block
+            # in_answer = False # No longer needed
             try:
                 # Use curl_cffi session post with impersonate
                 response = self.session.post(
@@ -132,41 +132,23 @@ class HeckAI(Provider):
                     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 not line_bytes:
-                        continue
-                        
-                    try:
-                        line = line_bytes.decode('utf-8')
-                        # Remove "data: " prefix
-                        if line.startswith("data: "):
-                            data = line[6:]
-                        else:
-                            continue # Skip lines without the prefix
-                        
-                        # Check for control markers
-                        if data == "[ANSWER_START]":
-                            in_answer = True
-                            continue
-                            
-                        if data == "[ANSWER_DONE]":
-                            in_answer = False
-                            continue
-                            
-                        if data == "[RELATE_Q_START]" or data == "[RELATE_Q_DONE]":
-                            continue
-                            
-                        # Process content if we're in an answer section
-                        if in_answer:
-                            # Assuming 'data' is the text chunk here
-                            streaming_text += data
-                            resp = dict(text=data)
-                            # Yield dict or raw string chunk
-                            yield resp if not raw else data
-                    except UnicodeDecodeError:
-                        continue # Ignore decoding errors for specific lines
+
+                # Use sanitize_stream to process the stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value="data:",  # Prefix to remove *from lines between markers*
+                    to_json=False,        # Content is text
+                    start_marker="data: [ANSWER_START]", # Check against the raw line including prefix
+                    end_marker="data: [ANSWER_DONE]",     # Check against the raw line including prefix
+                    skip_markers=["[RELATE_Q_START]", "[RELATE_Q_DONE]"], # Skip these if they appear within answer block
+                    yield_raw_on_error=True
+                )
+
+                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
                 
                 # Update history and previous answer after stream finishes
                 self.previous_answer = streaming_text

webscout/Provider/Jadve.py

@@ -4,7 +4,7 @@ import json
 import re
 from typing import Union, Any, Dict, Optional, Generator
 
-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
@@ -97,6 +97,17 @@ class JadveOpenAI(Provider):
         )
         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,
@@ -153,41 +164,22 @@ class JadveOpenAI(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
 
-                # Pattern to match the streaming chunks format: 0:"text"
-                pattern = r'0:"(.*?)"'
-                buffer = ""
-                
-                # Iterate over bytes and decode manually
-                for line_bytes in response.iter_lines():
-                    if not line_bytes:
-                        continue
-                    
-                    try:
-                        line = line_bytes.decode('utf-8')
-                        buffer += line
-                        
-                        # Try to match chunks in the current buffer
-                        matches = re.findall(pattern, buffer)
-                        if matches:
-                            for chunk in matches:
-                                # Handle potential escape sequences like \\n
-                                decoded_chunk = chunk.encode().decode('unicode_escape')
-                                full_response_text += decoded_chunk
-                                resp = {"text": decoded_chunk}
-                                # Yield dict or raw string chunk
-                                yield resp if not raw else decoded_chunk
-                            
-                            # Remove matched parts from the buffer
-                            # Be careful with buffer modification during iteration if issues arise
-                            matched_parts = [f'0:"{match}"' for match in matches]
-                            for part in matched_parts:
-                                buffer = buffer.replace(part, '', 1)
-                        
-                        # Check if we've reached the end of the response
-                        if 'e:' in line or 'd:' in line:
-                            break
-                    except UnicodeDecodeError:
-                        continue # Ignore decoding errors for specific lines
+                # 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}
@@ -272,7 +264,8 @@ class JadveOpenAI(Provider):
             str: Extracted text.
         """
         assert isinstance(response, dict), "Response should be of dict data-type only"
-        return response["text"]
+        # Extractor handles formatting
+        return response.get("text", "")
 
 if __name__ == "__main__":
     # Ensure curl_cffi is installed

webscout/Provider/LambdaChat.py

@@ -7,8 +7,8 @@ import re
 import uuid
 from typing import Any, Dict, List, Optional, Union, Generator
 
-from webscout.AIutel import Conversation
-from webscout.AIbase import Provider
+from webscout.AIutel import Conversation, sanitize_stream
+from webscout.AIbase import Provider # Import sanitize_stream
 from webscout import exceptions
 from webscout.litagent import LitAgent
 
@@ -182,62 +182,21 @@ class LambdaChat(Provider):
         boundary += "".join(random.choice(boundary_chars) for _ in range(16))
         return boundary
     
-    def process_response(self, response, prompt: str):
-        """Process streaming response and extract content."""
-        full_text = ""
-        sources = None
+    @staticmethod
+    def _lambdachat_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Extracts content from LambdaChat stream JSON objects."""
+        if not isinstance(chunk, dict) or "type" not in chunk:
+            return None
+
         reasoning_text = ""
-        has_reasoning = False
-        
-        for line in response.iter_lines(decode_unicode=True):
-            if not line:
-                continue
-                
-            try:
-                # Parse each line as JSON
-                data = json.loads(line)
-                
-                # Handle different response types
-                if "type" not in data:
-                    continue
-                    
-                if data["type"] == "stream" and "token" in data:
-                    token = data["token"].replace("\u0000", "")
-                    full_text += token
-                    resp = {"text": token}
-                    yield resp
-                elif data["type"] == "finalAnswer":
-                    final_text = data.get("text", "")
-                    if final_text and not full_text:
-                        full_text = final_text
-                        resp = {"text": final_text}
-                        yield resp
-                elif data["type"] == "webSearch" and "sources" in data:
-                    sources = data["sources"]
-                elif data["type"] == "reasoning":
-                    has_reasoning = True
-                    if data.get("subtype") == "stream" and "token" in data:
-                        reasoning_text += data["token"]
-                    
-                    # If we have reasoning, prepend it to the next text output
-                    if reasoning_text and not full_text:
-                        resp = {"text": f"<think>\n{reasoning_text}\n</think>\n", "is_reasoning": True}
-                        yield resp
-                    
-            except json.JSONDecodeError:
-                continue
-        
-        # Update conversation history only for saving to file if needed
-        if full_text and self.conversation.file:
-            if has_reasoning:
-                full_text_with_reasoning = f"<think>\n{reasoning_text}\n</think>\n{full_text}"
-                self.last_response = {"text": full_text_with_reasoning}
-                self.conversation.update_chat_history(prompt, full_text_with_reasoning)
-            else:
-                self.last_response = {"text": full_text}
-                self.conversation.update_chat_history(prompt, full_text)
-        
-        return full_text
+        if chunk["type"] == "stream" and "token" in chunk:
+            return chunk["token"].replace("\u0000", "")
+        elif chunk["type"] == "finalAnswer":
+            return chunk.get("text")
+        elif chunk["type"] == "reasoning" and chunk.get("subtype") == "stream" and "token" in chunk:
+            # Prepend reasoning with <think> tags? Or handle separately? For now, just return token.
+            return chunk["token"] # Or potentially format as f"<think>{chunk['token']}</think>"
+        return None
     
     def ask(
         self,
@@ -296,6 +255,7 @@ class LambdaChat(Provider):
         multipart_headers["Content-Length"] = str(len(body))
         
         def for_stream():
+            streaming_text = "" # Initialize for history
             try:
                 # Try with multipart/form-data first
                 response = None
@@ -327,8 +287,21 @@ class LambdaChat(Provider):
                 
                 response.raise_for_status() # Check status after potential fallback
                 
-                # Process the streaming response
-                yield from self.process_response(response, prompt)
+                # Use sanitize_stream
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value=None, # No prefix
+                    to_json=True,     # Stream sends JSON lines
+                    content_extractor=self._lambdachat_extractor, # Use the specific extractor
+                    yield_raw_on_error=False # Skip non-JSON lines or lines where extractor fails
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by _lambdachat_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk # Aggregate text for history
+                        resp = {"text": content_chunk}
+                        yield resp if not raw else content_chunk
                 
             except (CurlError, exceptions.FailedToGenerateResponseError, Exception) as e: # Catch errors from both attempts
                 # Handle specific exceptions if needed
@@ -353,6 +326,10 @@ class LambdaChat(Provider):
                 # If we get here, all models failed
                 raise exceptions.FailedToGenerateResponseError(f"Request failed after trying fallback: {str(e)}") from e
 
+            # Update history after stream finishes
+            if streaming_text and self.conversation.file:
+                self.last_response = {"text": streaming_text}
+                self.conversation.update_chat_history(prompt, streaming_text)
 
         def for_non_stream():
             # Aggregate the stream using the updated for_stream logic

webscout/Provider/MCPCore.py

@@ -8,7 +8,7 @@ from curl_cffi import CurlError
 
 from webscout.AIutel import Optimizers
 from webscout.AIutel import Conversation
-from webscout.AIutel import AwesomePrompts
+from webscout.AIutel import AwesomePrompts, sanitize_stream 
 from webscout.AIbase import Provider
 from webscout import exceptions
 from webscout.litagent import LitAgent
@@ -146,11 +146,11 @@ class MCPCore(Provider):
             )
             return cookie_string, token
         except FileNotFoundError:
-            raise exceptions.InvalidAuthenticationError(
+            raise exceptions.FailedToGenerateResponseError(
                 f"Error: Cookies file not found at {self.cookies_path}!"
             )
         except json.JSONDecodeError:
-            raise exceptions.InvalidAuthenticationError(
+            raise exceptions.FailedToGenerateResponseError(
                 f"Error: Invalid JSON format in cookies file: {self.cookies_path}!"
             )
 
@@ -207,24 +207,21 @@ class MCPCore(Provider):
                 )
                 response.raise_for_status()
 
-                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 and len(json_data['choices']) > 0:
-                                    delta = json_data['choices'][0].get('delta', {})
-                                    content = delta.get('content')
-                                    if content:
-                                        streaming_text += content
-                                        resp = dict(text=content)
-                                        yield resp if not raw else content
-                        except (json.JSONDecodeError, UnicodeDecodeError):
-                            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
+                    skip_markers=["[DONE]"],
+                    content_extractor=lambda chunk: chunk.get('choices', [{}])[0].get('delta', {}).get('content') if isinstance(chunk, dict) else None,
+                    yield_raw_on_error=False # Skip non-JSON or lines where extractor fails
+                )
+
+                for content_chunk in processed_stream:
+                    # content_chunk is the string extracted by the content_extractor
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk
+                        yield dict(text=content_chunk) if not raw else content_chunk
 
                 self.last_response = {"text": streaming_text}
                 self.conversation.update_chat_history(prompt, self.get_message(self.last_response))

webscout/Provider/Marcus.py

@@ -5,7 +5,7 @@ from typing import Union, Any, Dict, Optional, Generator
 
 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
 
@@ -103,18 +103,18 @@ class Marcus(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:
-                            decoded_line = line_bytes.decode('utf-8')
-                            streaming_text += decoded_line # Aggregate text
-                            resp = {"text": decoded_line}
-                            # Yield dict or raw string chunk
-                            yield resp if not raw else decoded_line
-                        except UnicodeDecodeError:
-                            continue # Ignore decoding errors
+                # Use sanitize_stream to decode bytes and yield text chunks
+                processed_stream = sanitize_stream(
+                    data=response.iter_content(chunk_size=None), # Pass byte iterator
+                    intro_value=None, # No prefix
+                    to_json=False,    # It's plain text
+                    yield_raw_on_error=True
+                )
 
+                for content_chunk in processed_stream:
+                    if content_chunk and isinstance(content_chunk, str):
+                        streaming_text += content_chunk # Aggregate text
+                        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(
@@ -140,8 +140,17 @@ class Marcus(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
                 
-                # Use response.text which is already decoded
-                full_response = response.text
+                response_text_raw = response.text # Get raw text
+
+                # Process the text using sanitize_stream (even though it's not streaming)
+                processed_stream = sanitize_stream(
+                    data=response_text_raw,
+                    intro_value=None, # No prefix
+                    to_json=False     # It's plain text
+                )
+                # Aggregate the single result
+                full_response = "".join(list(processed_stream))
+
                 self.last_response = {"text": full_response}
                 self.conversation.update_chat_history(prompt, full_response)
                 # Return dict or raw string

webscout/Provider/Netwrck.py

@@ -1,10 +1,5 @@
-import time
-import uuid
-import json
 from typing import Any, Dict, Optional, Generator, Union
-from dataclasses import dataclass, asdict
-from datetime import date
-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
@@ -95,6 +90,15 @@ class Netwrck(Provider):
             if callable(getattr(Optimizers, method)) and not method.startswith("__")
         )
 
+    @staticmethod
+    def _netwrck_extractor(chunk: Union[str, Dict[str, Any]]) -> Optional[str]:
+        """Removes surrounding quotes and handles potential escapes."""
+        if isinstance(chunk, str):
+            text = chunk.strip('"')
+            # Handle potential unicode escapes if they appear
+            # text = text.encode().decode('unicode_escape') # Uncomment if needed
+            return text
+        return None
     def ask(
         self,
         prompt: str,
@@ -136,21 +140,18 @@ class Netwrck(Provider):
                 response.raise_for_status() # Check for HTTP errors
 
                 streaming_text = ""
-                # Iterate over bytes and decode manually
-                for line_bytes in response.iter_lines():
-                    if line_bytes:
-                        try:
-                            decoded_line = line_bytes.decode('utf-8').strip('"')
-                            # Handle potential escape sequences if necessary
-                            # decoded_line = decoded_line.encode().decode('unicode_escape') # Uncomment if needed
-                            streaming_text += decoded_line
-                            resp = {"text": decoded_line}
-                            # Yield dict or raw string
-                            yield resp if not raw else decoded_line
-                        except UnicodeDecodeError:
-                            # Handle potential decoding errors if chunks split mid-character
-                            continue 
-
+                # 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)
@@ -174,11 +175,19 @@ class Netwrck(Provider):
                 )
                 response.raise_for_status() # Check for HTTP errors
                 
-                # Use response.text which is already decoded
-                text = response.text.strip('"')
-                # Handle potential escape sequences if necessary
-                # text = text.encode().decode('unicode_escape') # Uncomment if needed
-                self.last_response = {"text": text}
+                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
+                )
+                # 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

webscout/Provider/OPENAI/__init__.py

@@ -25,4 +25,4 @@ from .textpollinations import *
 from .e2b import *
 from .multichat import * # Add MultiChatAI
 from .ai4chat import * # Add AI4Chat
-from .mcpcore import *
+from .mcpcore import *

webscout/Provider/OPENAI/exachat.py

@@ -35,6 +35,8 @@ MODEL_CONFIGS = {
             "gemini-2.0-flash-thinking-exp-01-21",
             "gemini-2.5-pro-exp-03-25",
             "gemini-2.0-pro-exp-02-05",
+            "gemini-2.5-flash-preview-04-17",
+
         
         ],
     },
@@ -83,6 +85,7 @@ MODEL_CONFIGS = {
     },
 }
 
+
 class Completions(BaseCompletions):
     def __init__(self, client: 'ExaChat'):
         self._client = client
@@ -292,6 +295,7 @@ class ExaChat(OpenAICompatibleProvider):
         "gemini-2.0-flash-thinking-exp-01-21",
         "gemini-2.5-pro-exp-03-25",
         "gemini-2.0-pro-exp-02-05",
+        "gemini-2.5-flash-preview-04-17",
         
         # OpenRouter Models
         "mistralai/mistral-small-3.1-24b-instruct:free",

webscout/Provider/OPENAI/scirachat.py

@@ -324,15 +324,13 @@ class SciraChat(OpenAICompatibleProvider):
     """
     
     AVAILABLE_MODELS = {
-        "scira-default": "Grok3",
-        "scira-grok-3-mini": "Grok3-mini", # thinking model
+        "scira-default": "Grok3-mini", # thinking model
+        "scira-grok-3": "Grok3", 
         "scira-vision" : "Grok2-Vision", # vision model
         "scira-4.1-mini": "GPT4.1-mini",
         "scira-qwq": "QWQ-32B",
         "scira-o4-mini": "o4-mini",
         "scira-google": "gemini 2.5 flash"
-
-
     }
     
     def __init__(

webscout/Provider/OPENAI/textpollinations.py

@@ -268,28 +268,26 @@ class TextPollinations(OpenAICompatibleProvider):
     """
 
     AVAILABLE_MODELS = [
-        "openai",              # OpenAI GPT-4.1-nano (Azure) - vision capable
-        "openai-large",        # OpenAI GPT-4.1 mini (Azure) - vision capable
-        "openai-reasoning",    # OpenAI o4-mini (Azure) - vision capable, reasoning
-        "qwen-coder",          # Qwen 2.5 Coder 32B (Scaleway)
-        "llama",               # Llama 3.3 70B (Cloudflare)
-        "llamascout",          # Llama 4 Scout 17B (Cloudflare)
-        "mistral",             # Mistral Small 3 (Scaleway) - vision capable
-        "unity",               # Unity Mistral Large (Scaleway) - vision capable, uncensored
-        "midijourney",         # Midijourney (Azure)
-        "rtist",               # Rtist (Azure)
-        "searchgpt",           # SearchGPT (Azure) - vision capable
-        "evil",                # Evil (Scaleway) - vision capable, uncensored
-        "deepseek-reasoning",  # DeepSeek-R1 Distill Qwen 32B (Cloudflare) - reasoning
-        "deepseek-reasoning-large", # DeepSeek R1 - Llama 70B (Scaleway) - reasoning
-        "phi",                 # Phi-4 Instruct (Cloudflare) - vision and audio capable
-        "llama-vision",        # Llama 3.2 11B Vision (Cloudflare) - vision capable
-        "gemini",              # gemini-2.5-flash-preview-04-17 (Azure) - vision and audio capable
-        "hormoz",              # Hormoz 8b (Modal)
-        "hypnosis-tracy",      # Hypnosis Tracy 7B (Azure) - audio capable
-        "deepseek",            # DeepSeek-V3 (DeepSeek)
-        "sur",                 # Sur AI Assistant (Mistral) (Scaleway) - vision capable
-        "openai-audio",        # OpenAI GPT-4o-audio-preview (Azure) - vision and audio capable
+        "openai",
+        "openai-large",
+        "qwen-coder",
+        "llama",
+        "llamascout",
+        "mistral",
+        "unity",
+        "midijourney",
+        "rtist",
+        "searchgpt",
+        "evil",
+        "deepseek-reasoning",
+        "deepseek-reasoning-large",
+        "phi",
+        "llama-vision",
+        "hormoz",
+        "hypnosis-tracy",
+        "deepseek",
+        "sur",
+        "openai-audio",
     ]
 
     def __init__(

webscout/Provider/OPENAI/toolbaz.py

@@ -288,6 +288,7 @@ class Toolbaz(OpenAICompatibleProvider):
         "gemini-2.0-flash-thinking",
         "gemini-2.0-flash",
         "gemini-1.5-flash",
+        "o3-mini",
         "gpt-4o-latest",
         "gpt-4o",
         "deepseek-r1",