ai-lib-python 0.5.0__py3-none-any.whl → 0.7.0__py3-none-any.whl

ai_lib_python/__init__.py

@@ -1,43 +1,62 @@
-"""
-ai-lib-python: Official Python Runtime for AI-Protocol
-
-The canonical Pythonic implementation for unified AI model interaction.
-Core principle: All logic is operators, all configuration is protocol.
-"""
-
-from ai_lib_python.client import AiClient, AiClientBuilder, CallStats, ChatResponse
-from ai_lib_python.errors import AiLibError, ProtocolError, TransportError
-from ai_lib_python.types.events import StreamingEvent
-from ai_lib_python.types.message import (
-    ContentBlock,
-    Message,
-    MessageContent,
-    MessageRole,
-)
-from ai_lib_python.types.tool import ToolCall, ToolDefinition
-
-__version__ = "0.5.0"
-
-__all__ = [
-    # Client
-    "AiClient",
-    "AiClientBuilder",
-    # Errors
-    "AiLibError",
-    "CallStats",
-    "ChatResponse",
-    "ContentBlock",
-    # Types - Message
-    "Message",
-    "MessageContent",
-    "MessageRole",
-    "ProtocolError",
-    # Types - Events
-    "StreamingEvent",
-    "ToolCall",
-    # Types - Tool
-    "ToolDefinition",
-    "TransportError",
-    # Version
-    "__version__",
-]
+"""AI-Protocol 官方 Python 运行时：提供统一的多厂商 AI 模型交互接口。
+
+ai-lib-python: Official Python Runtime for AI-Protocol.
+
+The canonical Pythonic implementation for unified AI model interaction.
+Core principle: All logic is operators, all configuration is protocol.
+"""
+from __future__ import annotations
+
+from ai_lib_python._features import (
+    HAS_AUDIO,
+    HAS_KEYRING,
+    HAS_TELEMETRY,
+    HAS_TOKENIZER,
+    HAS_VISION,
+    HAS_WATCHDOG,
+    require_extra,
+)
+from ai_lib_python.client import AiClient, AiClientBuilder, CallStats, ChatResponse
+from ai_lib_python.errors import AiLibError, ProtocolError, TransportError
+from ai_lib_python.types.events import StreamingEvent
+from ai_lib_python.types.message import (
+    ContentBlock,
+    Message,
+    MessageContent,
+    MessageRole,
+)
+from ai_lib_python.types.tool import ToolCall, ToolDefinition
+
+__version__ = "0.7.0"
+
+__all__ = [
+    # Client
+    "AiClient",
+    "AiClientBuilder",
+    # Feature flags
+    "HAS_AUDIO",
+    "HAS_KEYRING",
+    "HAS_TELEMETRY",
+    "HAS_TOKENIZER",
+    "HAS_VISION",
+    "HAS_WATCHDOG",
+    "require_extra",
+    # Errors
+    "AiLibError",
+    "CallStats",
+    "ChatResponse",
+    "ContentBlock",
+    # Types - Message
+    "Message",
+    "MessageContent",
+    "MessageRole",
+    "ProtocolError",
+    # Types - Events
+    "StreamingEvent",
+    "ToolCall",
+    # Types - Tool
+    "ToolDefinition",
+    "TransportError",
+    # Version
+    "__version__",
+]

ai_lib_python/_features.py

@@ -0,0 +1,51 @@
+"""运行时特性检测：检查可选依赖以确定可用的能力模块。
+
+Runtime feature detection for optional extras.
+
+Checks availability of optional dependencies to determine which
+capability modules can be used.
+"""
+from __future__ import annotations
+
+
+def _check_import(module_name: str) -> bool:
+    """Check if a module is importable."""
+    try:
+        __import__(module_name)
+        return True
+    except ImportError:
+        return False
+
+
+# Capability feature flags
+HAS_VISION: bool = _check_import("PIL")
+HAS_AUDIO: bool = _check_import("soundfile")
+HAS_TELEMETRY: bool = _check_import("opentelemetry")
+HAS_TOKENIZER: bool = _check_import("tiktoken")
+HAS_KEYRING: bool = _check_import("keyring")
+HAS_WATCHDOG: bool = _check_import("watchdog")
+
+
+# Map pip package names to import module names (when they differ)
+_PACKAGE_TO_MODULE: dict[str, str] = {
+    "pillow": "PIL",
+}
+
+
+def require_extra(extra_name: str, package_name: str) -> None:
+    """Raise ImportError with installation hint if extra is not available.
+
+    Args:
+        extra_name: Name of the pip extra (e.g., 'vision')
+        package_name: Name of the required package (e.g., 'pillow')
+
+    Raises:
+        ImportError: With installation instructions when package is not available.
+    """
+    module_name = _PACKAGE_TO_MODULE.get(package_name, package_name)
+    if _check_import(module_name):
+        return
+    raise ImportError(
+        f"The '{extra_name}' extra is required for this feature. "
+        f"Install it with: pip install ai-lib-python[{extra_name}]"
+    )

ai_lib_python/batch/__init__.py

@@ -1,15 +1,15 @@
-"""
-Batch processing module for ai-lib-python.
-
-Provides request batching and batch execution utilities.
-"""
-
-from ai_lib_python.batch.collector import BatchCollector, BatchConfig
-from ai_lib_python.batch.executor import BatchExecutor, BatchResult
-
-__all__ = [
-    "BatchCollector",
-    "BatchConfig",
-    "BatchExecutor",
-    "BatchResult",
-]
+"""
+Batch processing module for ai-lib-python.
+
+Provides request batching and batch execution utilities.
+"""
+
+from ai_lib_python.batch.collector import BatchCollector, BatchConfig
+from ai_lib_python.batch.executor import BatchExecutor, BatchResult
+
+__all__ = [
+    "BatchCollector",
+    "BatchConfig",
+    "BatchExecutor",
+    "BatchResult",
+]

ai_lib_python/batch/collector.py

@@ -1,244 +1,244 @@
-"""
-Batch collector for grouping requests.
-
-Collects requests and groups them for batch execution.
-"""
-
-from __future__ import annotations
-
-import asyncio
-import time
-from dataclasses import dataclass, field
-from typing import TYPE_CHECKING, Any, Generic, TypeVar
-
-if TYPE_CHECKING:
-    from collections.abc import Awaitable, Callable
-
-T = TypeVar("T")
-R = TypeVar("R")
-
-
-@dataclass
-class BatchConfig:
-    """Configuration for batch collection.
-
-    Attributes:
-        max_batch_size: Maximum requests per batch
-        max_wait_ms: Maximum wait time before flushing batch
-        group_by: Function to determine grouping key
-    """
-
-    max_batch_size: int = 10
-    max_wait_ms: float = 100.0
-    group_by: Callable[[Any], str] | None = None
-
-    @classmethod
-    def default(cls) -> BatchConfig:
-        """Create default configuration."""
-        return cls()
-
-    @classmethod
-    def for_embeddings(cls) -> BatchConfig:
-        """Create configuration optimized for embeddings."""
-        return cls(
-            max_batch_size=100,
-            max_wait_ms=50.0,
-        )
-
-    @classmethod
-    def for_chat(cls) -> BatchConfig:
-        """Create configuration for chat completions."""
-        return cls(
-            max_batch_size=5,
-            max_wait_ms=10.0,
-        )
-
-
-@dataclass
-class PendingRequest(Generic[T]):
-    """A pending request waiting for batch execution.
-
-    Attributes:
-        data: Request data
-        future: Future to resolve with result
-        added_at: Timestamp when added
-        group_key: Grouping key
-    """
-
-    data: T
-    future: asyncio.Future[Any]
-    added_at: float = field(default_factory=time.time)
-    group_key: str = "_default_"
-
-
-class BatchCollector(Generic[T, R]):
-    """Collects requests for batch processing.
-
-    Accumulates requests until batch size or time limit is reached,
-    then triggers batch execution.
-
-    Example:
-        >>> async def process_batch(items):
-        ...     return [f"result_{i}" for i in range(len(items))]
-        ...
-        >>> collector = BatchCollector(
-        ...     config=BatchConfig(max_batch_size=5),
-        ...     executor=process_batch,
-        ... )
-        >>> result = await collector.add("request1")
-    """
-
-    def __init__(
-        self,
-        config: BatchConfig | None = None,
-        executor: Callable[[list[T]], Awaitable[list[R]]] | None = None,
-    ) -> None:
-        """Initialize batch collector.
-
-        Args:
-            config: Batch configuration
-            executor: Function to execute batches
-        """
-        self._config = config or BatchConfig.default()
-        self._executor = executor
-        self._pending: dict[str, list[PendingRequest[T]]] = {}
-        self._lock = asyncio.Lock()
-        self._timers: dict[str, asyncio.Task[None]] = {}
-        self._running = True
-
-    def set_executor(
-        self, executor: Callable[[list[T]], Awaitable[list[R]]]
-    ) -> None:
-        """Set the batch executor function.
-
-        Args:
-            executor: Function to execute batches
-        """
-        self._executor = executor
-
-    async def add(self, data: T) -> R:
-        """Add a request to the batch.
-
-        Args:
-            data: Request data
-
-        Returns:
-            Result from batch execution
-        """
-        if not self._running:
-            raise RuntimeError("Batch collector is stopped")
-
-        if self._executor is None:
-            raise RuntimeError("No executor set")
-
-        # Determine group key
-        group_key = (
-            self._config.group_by(data) if self._config.group_by else "_default_"
-        )
-
-        # Create future for result
-        loop = asyncio.get_event_loop()
-        future: asyncio.Future[R] = loop.create_future()
-
-        async with self._lock:
-            # Add to pending
-            if group_key not in self._pending:
-                self._pending[group_key] = []
-
-            self._pending[group_key].append(
-                PendingRequest(data=data, future=future, group_key=group_key)
-            )
-
-            # Check if batch is full
-            if len(self._pending[group_key]) >= self._config.max_batch_size:
-                await self._flush_group(group_key)
-            else:
-                # Start timer if not already running
-                if group_key not in self._timers or self._timers[group_key].done():
-                    self._timers[group_key] = asyncio.create_task(
-                        self._timer_flush(group_key)
-                    )
-
-        return await future
-
-    async def _timer_flush(self, group_key: str) -> None:
-        """Flush group after timeout."""
-        await asyncio.sleep(self._config.max_wait_ms / 1000.0)
-
-        async with self._lock:
-            if self._pending.get(group_key):
-                await self._flush_group(group_key)
-
-    async def _flush_group(self, group_key: str) -> None:
-        """Flush a specific group.
-
-        Args:
-            group_key: Group to flush
-        """
-        if group_key not in self._pending or not self._pending[group_key]:
-            return
-
-        # Get pending requests
-        requests = self._pending.pop(group_key)
-
-        # Cancel timer
-        if group_key in self._timers:
-            self._timers[group_key].cancel()
-            del self._timers[group_key]
-
-        # Extract data
-        data_list = [r.data for r in requests]
-
-        try:
-            # Execute batch
-            results = await self._executor(data_list)
-
-            # Resolve futures
-            for request, result in zip(requests, results, strict=False):
-                if not request.future.done():
-                    request.future.set_result(result)
-
-        except Exception as e:
-            # Reject all futures
-            for request in requests:
-                if not request.future.done():
-                    request.future.set_exception(e)
-
-    async def flush(self) -> None:
-        """Flush all pending batches."""
-        async with self._lock:
-            for group_key in list(self._pending.keys()):
-                await self._flush_group(group_key)
-
-    async def stop(self) -> None:
-        """Stop the collector and flush pending requests."""
-        self._running = False
-        await self.flush()
-
-        # Cancel all timers
-        for timer in self._timers.values():
-            timer.cancel()
-        self._timers.clear()
-
-    def get_pending_count(self, group_key: str | None = None) -> int:
-        """Get count of pending requests.
-
-        Args:
-            group_key: Group to count (all if None)
-
-        Returns:
-            Number of pending requests
-        """
-        if group_key:
-            return len(self._pending.get(group_key, []))
-        return sum(len(p) for p in self._pending.values())
-
-    @property
-    def config(self) -> BatchConfig:
-        """Get batch configuration."""
-        return self._config
-
-    @property
-    def is_running(self) -> bool:
-        """Check if collector is running."""
-        return self._running
+"""
+Batch collector for grouping requests.
+
+Collects requests and groups them for batch execution.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import time
+from dataclasses import dataclass, field
+from typing import TYPE_CHECKING, Any, Generic, TypeVar
+
+if TYPE_CHECKING:
+    from collections.abc import Awaitable, Callable
+
+T = TypeVar("T")
+R = TypeVar("R")
+
+
+@dataclass
+class BatchConfig:
+    """Configuration for batch collection.
+
+    Attributes:
+        max_batch_size: Maximum requests per batch
+        max_wait_ms: Maximum wait time before flushing batch
+        group_by: Function to determine grouping key
+    """
+
+    max_batch_size: int = 10
+    max_wait_ms: float = 100.0
+    group_by: Callable[[Any], str] | None = None
+
+    @classmethod
+    def default(cls) -> BatchConfig:
+        """Create default configuration."""
+        return cls()
+
+    @classmethod
+    def for_embeddings(cls) -> BatchConfig:
+        """Create configuration optimized for embeddings."""
+        return cls(
+            max_batch_size=100,
+            max_wait_ms=50.0,
+        )
+
+    @classmethod
+    def for_chat(cls) -> BatchConfig:
+        """Create configuration for chat completions."""
+        return cls(
+            max_batch_size=5,
+            max_wait_ms=10.0,
+        )
+
+
+@dataclass
+class PendingRequest(Generic[T]):
+    """A pending request waiting for batch execution.
+
+    Attributes:
+        data: Request data
+        future: Future to resolve with result
+        added_at: Timestamp when added
+        group_key: Grouping key
+    """
+
+    data: T
+    future: asyncio.Future[Any]
+    added_at: float = field(default_factory=time.time)
+    group_key: str = "_default_"
+
+
+class BatchCollector(Generic[T, R]):
+    """Collects requests for batch processing.
+
+    Accumulates requests until batch size or time limit is reached,
+    then triggers batch execution.
+
+    Example:
+        >>> async def process_batch(items):
+        ...     return [f"result_{i}" for i in range(len(items))]
+        ...
+        >>> collector = BatchCollector(
+        ...     config=BatchConfig(max_batch_size=5),
+        ...     executor=process_batch,
+        ... )
+        >>> result = await collector.add("request1")
+    """
+
+    def __init__(
+        self,
+        config: BatchConfig | None = None,
+        executor: Callable[[list[T]], Awaitable[list[R]]] | None = None,
+    ) -> None:
+        """Initialize batch collector.
+
+        Args:
+            config: Batch configuration
+            executor: Function to execute batches
+        """
+        self._config = config or BatchConfig.default()
+        self._executor = executor
+        self._pending: dict[str, list[PendingRequest[T]]] = {}
+        self._lock = asyncio.Lock()
+        self._timers: dict[str, asyncio.Task[None]] = {}
+        self._running = True
+
+    def set_executor(
+        self, executor: Callable[[list[T]], Awaitable[list[R]]]
+    ) -> None:
+        """Set the batch executor function.
+
+        Args:
+            executor: Function to execute batches
+        """
+        self._executor = executor
+
+    async def add(self, data: T) -> R:
+        """Add a request to the batch.
+
+        Args:
+            data: Request data
+
+        Returns:
+            Result from batch execution
+        """
+        if not self._running:
+            raise RuntimeError("Batch collector is stopped")
+
+        if self._executor is None:
+            raise RuntimeError("No executor set")
+
+        # Determine group key
+        group_key = (
+            self._config.group_by(data) if self._config.group_by else "_default_"
+        )
+
+        # Create future for result
+        loop = asyncio.get_event_loop()
+        future: asyncio.Future[R] = loop.create_future()
+
+        async with self._lock:
+            # Add to pending
+            if group_key not in self._pending:
+                self._pending[group_key] = []
+
+            self._pending[group_key].append(
+                PendingRequest(data=data, future=future, group_key=group_key)
+            )
+
+            # Check if batch is full
+            if len(self._pending[group_key]) >= self._config.max_batch_size:
+                await self._flush_group(group_key)
+            else:
+                # Start timer if not already running
+                if group_key not in self._timers or self._timers[group_key].done():
+                    self._timers[group_key] = asyncio.create_task(
+                        self._timer_flush(group_key)
+                    )
+
+        return await future
+
+    async def _timer_flush(self, group_key: str) -> None:
+        """Flush group after timeout."""
+        await asyncio.sleep(self._config.max_wait_ms / 1000.0)
+
+        async with self._lock:
+            if self._pending.get(group_key):
+                await self._flush_group(group_key)
+
+    async def _flush_group(self, group_key: str) -> None:
+        """Flush a specific group.
+
+        Args:
+            group_key: Group to flush
+        """
+        if group_key not in self._pending or not self._pending[group_key]:
+            return
+
+        # Get pending requests
+        requests = self._pending.pop(group_key)
+
+        # Cancel timer
+        if group_key in self._timers:
+            self._timers[group_key].cancel()
+            del self._timers[group_key]
+
+        # Extract data
+        data_list = [r.data for r in requests]
+
+        try:
+            # Execute batch
+            results = await self._executor(data_list)
+
+            # Resolve futures
+            for request, result in zip(requests, results, strict=False):
+                if not request.future.done():
+                    request.future.set_result(result)
+
+        except Exception as e:
+            # Reject all futures
+            for request in requests:
+                if not request.future.done():
+                    request.future.set_exception(e)
+
+    async def flush(self) -> None:
+        """Flush all pending batches."""
+        async with self._lock:
+            for group_key in list(self._pending.keys()):
+                await self._flush_group(group_key)
+
+    async def stop(self) -> None:
+        """Stop the collector and flush pending requests."""
+        self._running = False
+        await self.flush()
+
+        # Cancel all timers
+        for timer in self._timers.values():
+            timer.cancel()
+        self._timers.clear()
+
+    def get_pending_count(self, group_key: str | None = None) -> int:
+        """Get count of pending requests.
+
+        Args:
+            group_key: Group to count (all if None)
+
+        Returns:
+            Number of pending requests
+        """
+        if group_key:
+            return len(self._pending.get(group_key, []))
+        return sum(len(p) for p in self._pending.values())
+
+    @property
+    def config(self) -> BatchConfig:
+        """Get batch configuration."""
+        return self._config
+
+    @property
+    def is_running(self) -> bool:
+        """Check if collector is running."""
+        return self._running