webscout 8.2.7__py3-none-any.whl → 8.2.9__py3-none-any.whl

webscout/Provider/OPENAI/utils.py

@@ -0,0 +1,318 @@
+from typing import List, Dict, Optional, Any, Union, Literal
+from enum import Enum
+import time
+import uuid
+from webscout.Provider.OPENAI.pydantic_imports import (
+    BaseModel, Field, field_validator, model_validator, field_serializer, model_serializer,
+    StrictStr, StrictInt, StrictFloat, StrictBool
+)
+
+# --- OpenAI Response Structure Mimics ---
+# Moved here for reusability across different OpenAI-compatible providers
+
+class ToolCallType(str, Enum):
+    """Type of tool call."""
+    FUNCTION = "function"
+
+class FunctionCall(BaseModel):
+    """Function call specification."""
+    name: StrictStr
+    arguments: StrictStr
+
+class ToolFunction(BaseModel):
+    """Function specification in a tool."""
+    name: StrictStr
+    arguments: StrictStr
+
+class ToolCall(BaseModel):
+    """Tool call specification."""
+    id: StrictStr
+    type: StrictStr
+    function: ToolFunction
+
+class CompletionUsage(BaseModel):
+    """Token usage information."""
+    prompt_tokens: StrictInt
+    completion_tokens: StrictInt
+    total_tokens: StrictInt
+    prompt_tokens_details: Optional[Dict[str, Any]] = None
+
+class ChoiceDelta(BaseModel):
+    """Delta content in streaming response."""
+    content: Optional[StrictStr] = None
+    function_call: Optional[FunctionCall] = None
+    role: Optional[StrictStr] = None
+    tool_calls: Optional[List[ToolCall]] = None
+
+class ChatCompletionMessage(BaseModel):
+    """Chat message in completion response."""
+    role: StrictStr
+    content: Optional[StrictStr] = None
+    function_call: Optional[FunctionCall] = None
+    tool_calls: Optional[List[ToolCall]] = None
+
+class Choice(BaseModel):
+    """Choice in completion response."""
+    index: StrictInt
+    message: Optional[ChatCompletionMessage] = None
+    delta: Optional[ChoiceDelta] = None
+    finish_reason: Optional[StrictStr] = None
+    logprobs: Optional[Dict[str, Any]] = None
+
+class ModelData(BaseModel):
+    """OpenAI model info response."""
+    id: StrictStr
+    object: StrictStr = "model"
+    created: StrictInt = int(time.time())
+    owned_by: StrictStr = "webscout"
+    permission: Optional[List[Dict[str, Any]]] = None
+    root: Optional[StrictStr] = None
+    parent: Optional[StrictStr] = None
+
+class ModelList(BaseModel):
+    """OpenAI model list response."""
+    data: List[ModelData]
+    object: StrictStr = "list"
+
+
+# @dataclass
+# class EmbeddingData(BaseModel):
+#     """Single embedding data."""
+#     embedding: List[float]
+#     index: int
+#     object: str = "embedding"
+
+# @dataclass
+# class EmbeddingResponse(BaseModel):
+#     """OpenAI embeddings response."""
+#     data: List[EmbeddingData]
+#     model: str
+#     usage: CompletionUsage
+#     object: str = "list"
+
+# @dataclass
+# class FineTuningJob(BaseModel):
+#     """OpenAI fine-tuning job."""
+#     id: str
+#     model: str
+#     created_at: int
+#     status: str
+#     training_file: str
+#     hyperparameters: Dict[str, Any]
+#     object: str = "fine_tuning.job"
+#     finished_at: Optional[int] = None
+#     validation_file: Optional[str] = None
+#     trained_tokens: Optional[int] = None
+#     result_files: Optional[List[str]] = None
+#     organization_id: Optional[str] = None
+
+# @dataclass
+# class FineTuningJobList(BaseModel):
+#     """OpenAI fine-tuning job list response."""
+#     data: List[FineTuningJob]
+#     object: str = "list"
+#     has_more: bool = False
+
+# @dataclass
+# class File(BaseModel):
+#     """OpenAI file."""
+#     id: str
+#     bytes: int
+#     created_at: int
+#     filename: str
+#     purpose: str
+#     object: str = "file"
+#     status: str = "uploaded"
+#     status_details: Optional[str] = None
+
+# @dataclass
+# class FileList(BaseModel):
+#     """OpenAI file list response."""
+#     data: List[File]
+#     object: str = "list"
+
+# @dataclass
+# class DeletedObject(BaseModel):
+#     """OpenAI deleted object response."""
+#     id: str
+#     object: str = "deleted_object"
+#     deleted: bool = True
+
+# @dataclass
+# class ImageData(BaseModel):
+#     """OpenAI generated image."""
+#     url: Optional[str] = None
+#     b64_json: Optional[str] = None
+#     revised_prompt: Optional[str] = None
+
+# @dataclass
+# class ImageResponse(BaseModel):
+#     """OpenAI image generation response."""
+#     data: List[ImageData]
+#     created: int = int(time.time())
+
+class ChatCompletion(BaseModel):
+    """Chat completion response."""
+    model: StrictStr
+    choices: List[Choice]
+    id: StrictStr = Field(default_factory=lambda: f"chatcmpl-{str(uuid.uuid4())}")
+    created: StrictInt = Field(default_factory=lambda: int(time.time()))
+    object: StrictStr = "chat.completion"
+    system_fingerprint: Optional[StrictStr] = None
+    usage: Optional[CompletionUsage] = None
+
+class ChatCompletionChunk(BaseModel):
+    """Streaming chat completion response chunk."""
+    model: StrictStr
+    choices: List[Choice]
+    id: StrictStr = Field(default_factory=lambda: f"chatcmpl-{str(uuid.uuid4())}")
+    created: StrictInt = Field(default_factory=lambda: int(time.time()))
+    object: StrictStr = "chat.completion.chunk"
+    system_fingerprint: Optional[StrictStr] = None
+    usage: Optional[Dict[str, Any]] = None  # Add usage field for streaming chunks
+
+
+# --- Helper Functions ---
+
+def format_prompt(messages: List[Dict[str, Any]], add_special_tokens: bool = False,
+                 do_continue: bool = False, include_system: bool = True) -> str:
+    """
+    Format a series of messages into a single string, optionally adding special tokens.
+
+    Args:
+        messages: A list of message dictionaries, each containing 'role' and 'content'.
+        add_special_tokens: Whether to add special formatting tokens.
+        do_continue: If True, don't add the final "Assistant:" prompt.
+        include_system: Whether to include system messages in the formatted output.
+
+    Returns:
+        A formatted string containing all messages.
+    """
+    # Helper function to convert content to string
+    def to_string(value) -> str:
+        if isinstance(value, str):
+            return value
+        elif isinstance(value, dict):
+            if "text" in value:
+                return value.get("text", "")
+            return ""
+        elif isinstance(value, list):
+            return "".join([to_string(v) for v in value])
+        return str(value)
+
+    # If there's only one message and no special tokens needed, just return its content
+    if not add_special_tokens and len(messages) <= 1:
+        return to_string(messages[0]["content"])
+
+    # Filter and process messages
+    processed_messages = [
+        (message["role"], to_string(message["content"]))
+        for message in messages
+        if include_system or message.get("role") != "system"
+    ]
+
+    # Format each message as "Role: Content"
+    formatted = "\n".join([
+        f'{role.capitalize()}: {content}'
+        for role, content in processed_messages
+        if content.strip()
+    ])
+
+    # Add final prompt for assistant if needed
+    if do_continue:
+        return formatted
+
+    return f"{formatted}\nAssistant:"
+
+
+def get_system_prompt(messages: List[Dict[str, Any]]) -> str:
+    """
+    Extract and concatenate all system messages.
+
+    Args:
+        messages: A list of message dictionaries.
+
+    Returns:
+        A string containing all system messages concatenated with newlines.
+    """
+    return "\n".join([m["content"] for m in messages if m["role"] == "system"])
+
+
+def get_last_user_message(messages: List[Dict[str, Any]]) -> str:
+    """
+    Get the content of the last user message in the conversation.
+
+    Args:
+        messages: A list of message dictionaries.
+
+    Returns:
+        The content of the last user message as a string.
+    """
+    for message in reversed(messages):
+        if message["role"] == "user":
+            if isinstance(message["content"], str):
+                return message["content"]
+            # Handle complex content structures
+            if isinstance(message["content"], dict) and "text" in message["content"]:
+                return message["content"]["text"]
+            if isinstance(message["content"], list):
+                text_parts = []
+                for part in message["content"]:
+                    if isinstance(part, dict) and part.get("type") == "text":
+                        text_parts.append(part.get("text", ""))
+                    elif isinstance(part, str):
+                        text_parts.append(part)
+                return "".join(text_parts)
+    return ""
+
+
+# --- Token Counter ---
+
+def count_tokens(text_or_messages: Any) -> int:
+    """
+    Count tokens in a string or a list of messages using tiktoken if available, else fallback to webstoken's WordTokenizer.
+
+    Args:
+        text_or_messages: A string or a list of messages (string or any type).
+        model: Optional model name for tiktoken encoding.
+
+    Returns:
+        int: Number of tokens.
+    """
+    try:
+        import tiktoken
+        # Use tiktoken if available
+        if isinstance(text_or_messages, str):
+            enc = tiktoken.encoding_for_model("gpt-4o")
+            return len(enc.encode(text_or_messages))
+        elif isinstance(text_or_messages, list):
+            enc = tiktoken.encoding_for_model("gpt-4o")
+            total = 0
+            for m in text_or_messages:
+                # Remove .get('content', '') and treat m as string or convert to string
+                if isinstance(m, str):
+                    total += len(enc.encode(m))
+                else:
+                    total += len(enc.encode(str(m)))
+            return total
+        else:
+            return 0
+    except ImportError:
+        # Fallback to webstoken's WordTokenizer
+        try:
+            from webstoken import WordTokenizer
+        except ImportError:
+            return 0
+        tokenizer = WordTokenizer()
+        if isinstance(text_or_messages, str):
+            return len(tokenizer.tokenize(text_or_messages))
+        elif isinstance(text_or_messages, list):
+            total = 0
+            for m in text_or_messages:
+                if isinstance(m, str):
+                    total += len(tokenizer.tokenize(m))
+                else:
+                    total += len(tokenizer.tokenize(str(m)))
+            return total
+        else:
+            return 0