flexllm 0.3.3__py3-none-any.whl

flexllm/msg_processors/__init__.py

@@ -0,0 +1,174 @@
+# maque.mllm.processors - 统一的高性能图像处理和消息预处理模块
+
+# 核心图像处理功能
+from .image_processor import (
+    ImageCacheConfig,
+    encode_image_to_base64,
+    encode_to_base64,
+    get_pil_image,
+    get_pil_image_sync,
+    decode_base64_to_pil,
+    decode_base64_to_file,
+    decode_base64_to_bytes,
+    encode_base64_from_local_path,
+    encode_base64_from_pil,
+)
+
+# 基础消息处理功能
+from .messages_processor import (
+    process_content_recursive,
+    messages_preprocess,
+    batch_messages_preprocess,
+    batch_process_messages,  # 别名
+)
+
+# 统一高性能处理器（推荐用于生产环境）
+from .unified_processor import (
+    UnifiedProcessorConfig,
+    UnifiedImageProcessor,
+    UnifiedMemoryCache,
+    batch_process_messages as unified_batch_process_messages,
+    unified_encode_image_to_base64,
+    get_global_unified_processor,
+    cleanup_global_unified_processor,
+)
+
+# 便捷的类接口
+from .image_processor_helper import ImageProcessor
+
+__all__ = [
+    # 图像缓存配置
+    "ImageCacheConfig",
+    # 核心图像处理
+    "encode_image_to_base64",
+    "encode_to_base64",
+    "get_pil_image",
+    "get_pil_image_sync",
+    "decode_base64_to_pil",
+    "decode_base64_to_file",
+    "decode_base64_to_bytes",
+    "encode_base64_from_local_path",
+    "encode_base64_from_pil",
+    # 基础消息处理
+    "process_content_recursive",
+    "messages_preprocess",
+    "batch_messages_preprocess",
+    "batch_process_messages",
+    # 统一高性能处理器（推荐）
+    "UnifiedProcessorConfig",
+    "UnifiedImageProcessor",
+    "UnifiedMemoryCache",
+    "unified_batch_process_messages",
+    "unified_encode_image_to_base64",
+    "get_global_unified_processor",
+    "cleanup_global_unified_processor",
+    # 便捷类接口
+    "ImageProcessor",
+]
+
+# 版本和性能信息
+__version__ = "3.0.0"  # 架构精简后的新版本
+__description__ = "统一高性能图像处理和消息预处理模块 - 精简架构，专注于统一处理器"
+
+# 性能建议
+PERFORMANCE_RECOMMENDATIONS = {
+    "unified_high_performance": {
+        "function": "unified_batch_process_messages",
+        "single_image_function": "unified_encode_image_to_base64",
+        "config": "UnifiedProcessorConfig.high_performance()",
+        "recommended_settings": {
+            "max_concurrent": 20,
+            "max_workers": 16,
+            "memory_cache_size_mb": 1000,
+        },
+        "speedup": "50-200x",
+        "description": "🚀 最高性能选择，统一处理本地文件和URL，支持自适应配置",
+    },
+    "unified_auto_detect": {
+        "function": "unified_batch_process_messages",
+        "single_image_function": "unified_encode_image_to_base64",
+        "config": "UnifiedProcessorConfig.auto_detect()",
+        "recommended_settings": "自动检测系统资源并调整",
+        "speedup": "根据系统自动优化",
+        "description": "🤖 智能配置选择，根据CPU和内存自动调整参数",
+    },
+    "unified_memory_optimized": {
+        "function": "unified_batch_process_messages",
+        "single_image_function": "unified_encode_image_to_base64",
+        "config": "UnifiedProcessorConfig.memory_optimized()",
+        "recommended_settings": {
+            "max_concurrent": 6,
+            "max_workers": 4,
+            "memory_cache_size_mb": 200,
+        },
+        "speedup": "20-60x",
+        "description": "💾 内存优化选择，适合资源受限环境",
+    },
+    "single_processing": {
+        "function": "messages_preprocess",
+        "description": "⚡ 推荐用于单个消息列表处理，简单快速",
+    },
+    "image_only": {
+        "function": "encode_image_to_base64",
+        "description": "🖼️ 仅处理单张图像时使用",
+    },
+}
+
+
+def get_performance_recommendation(use_case: str = "unified_auto_detect") -> dict:
+    """获取性能建议
+
+    Args:
+        use_case: 使用场景，可选：
+            - 'unified_auto_detect': 自适应配置（推荐）
+            - 'unified_high_performance': 最高性能配置
+            - 'unified_memory_optimized': 内存优化配置
+            - 'single_processing': 单个消息处理
+            - 'image_only': 单张图像处理
+
+    Returns:
+        性能建议字典
+    """
+    return PERFORMANCE_RECOMMENDATIONS.get(
+        use_case, PERFORMANCE_RECOMMENDATIONS["unified_auto_detect"]
+    )
+
+
+def print_performance_guide():
+    """打印性能使用指南"""
+    print("🚀 Sparrow VLLM Client 统一处理器使用指南")
+    print("=" * 60)
+    print()
+
+    for use_case, config in PERFORMANCE_RECOMMENDATIONS.items():
+        print(f"📊 {use_case.replace('_', ' ').title()}:")
+        print(f"   {config['description']}")
+        print(f"   函数: {config['function']}")
+        if "speedup" in config:
+            print(f"   性能提升: {config['speedup']}")
+        if "config" in config:
+            print(f"   配置: {config['config']}")
+        print()
+
+    print("💡 推荐: 优先使用 unified_auto_detect 获得最佳性能！")
+    print()
+    print("📝 批量消息处理示例:")
+    print("```python")
+    print(
+        "from flexllm.msg_processors import unified_batch_process_messages, UnifiedProcessorConfig"
+    )
+    print("")
+    print("# 自适应配置（推荐）")
+    print("config = UnifiedProcessorConfig.auto_detect()")
+    print("processed = await unified_batch_process_messages(")
+    print("    messages_list,")
+    print("    processor_config=config,")
+    print("    show_progress=True")
+    print(")")
+    print("```")
+    print()
+    print("🎯 精简架构特性:")
+    print("   ✅ 统一处理本地文件和URL")
+    print("   ✅ 自适应系统资源配置")
+    print("   ✅ 智能缓存系统")
+    print("   ✅ 性能监控统计")