pysfi 0.1.12__py3-none-any.whl → 0.1.14__py3-none-any.whl

sfi/llmquantize/llmquantize.py

@@ -8,8 +8,8 @@ import logging
 import pathlib
 import subprocess
 import sys
+from dataclasses import dataclass
 from pathlib import Path
-from types import SimpleNamespace
 
 from PySide2.QtCore import QThread, Signal, Slot
 from PySide2.QtGui import QMoveEvent, QResizeEvent
@@ -29,37 +29,65 @@ from PySide2.QtWidgets import (
     QWidget,
 )
 
-CONFIG_FILE = Path.home() / ".sfi" / "llmquantize.json"
+CONFIG_FILE = Path.home() / ".pysfi" / "llmquantize.json"
 
-logging.basicConfig(level=logging.INFO)
+logging.basicConfig(
+    level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s"
+)
 logger = logging.getLogger(__name__)
 
+__version__ = "1.0.0"
+__build__ = "20260204"
+
 
-class QuantizerConfig(SimpleNamespace):
+@dataclass
+class QuantizerConfig:
     """GGUF量化转换工具配置."""
 
     TITLE: str = "GGUF量化转换工具"
-    WIN_SIZE: list[int] = [600, 500]  # noqa: RUF012
-    WIN_POS: list[int] = [100, 100]  # noqa: RUF012
+    WIN_SIZE: list[int] = None
+    WIN_POS: list[int] = None
     LAST_INPUT_FILE: str = ""
-    SELECTED_QUANTS: list[str] = ["Q4_K_M", "Q5_K_M"]  # noqa: RUF012
+    SELECTED_QUANTS: list[str] = None
+
+    def __post_init__(self) -> None:
+        """初始化默认值并加载配置文件."""
+        # 初始化默认值
+        if self.WIN_SIZE is None:
+            self.WIN_SIZE = [600, 500]
+        if self.WIN_POS is None:
+            self.WIN_POS = [100, 100]
+        if self.SELECTED_QUANTS is None:
+            self.SELECTED_QUANTS = ["Q4_K_M", "Q5_K_M"]
 
-    def __init__(self) -> None:
         if CONFIG_FILE.exists():
             logger.info("Loading configuration from %s", CONFIG_FILE)
             try:
-                # 直接更新，忽略无效字段
-                self.__dict__.update(json.loads(CONFIG_FILE.read_text()))
-            except (json.JSONDecodeError, TypeError) as e:
+                config_data = json.loads(CONFIG_FILE.read_text())
+                # 更新实例属性，只更新存在的属性
+                for key, value in config_data.items():
+                    if hasattr(self, key):
+                        setattr(self, key, value)
+            except (json.JSONDecodeError, TypeError, AttributeError) as e:
                 logger.warning("Failed to load configuration: %s", e)
                 logger.info("Using default configuration")
         else:
             logger.info("Using default configuration")
 
     def save(self) -> None:
-        """保存配置."""
+        """保存配置到文件."""
         CONFIG_FILE.parent.mkdir(parents=True, exist_ok=True)
-        CONFIG_FILE.write_text(json.dumps(vars(self), indent=4))
+        # 将数据类转换为字典进行JSON序列化
+        config_dict = {}
+        for attr_name in dir(self):
+            if not attr_name.startswith("_") and attr_name not in ["TITLE"]:
+                try:
+                    attr_value = getattr(self, attr_name)
+                    if not callable(attr_value):
+                        config_dict[attr_name] = attr_value
+                except AttributeError:
+                    continue
+        CONFIG_FILE.write_text(json.dumps(config_dict, indent=4))
 
 
 conf = QuantizerConfig()
@@ -69,8 +97,11 @@ atexit.register(conf.save)
 def _process_gguf_stem(filename: str) -> str:
     """处理文件名, 移除可能的F16后缀.
 
+    Args:
+        filename: 输入的文件名(不含扩展名)
+
     Returns:
-        str: 处理后的文件名
+        str: 处理后的文件名, 移除了F16后缀(如果存在)
     """
     if filename.upper().endswith("-F16"):
         filename = filename[:-4]  # 移除-F16后缀
@@ -78,7 +109,13 @@ def _process_gguf_stem(filename: str) -> str:
 
 
 class QuantizationWorker(QThread):
-    """量化执行线程Worker."""
+    """量化执行线程Worker.
+
+    Attributes:
+        progress_msg_updated: 进度消息更新信号
+        progress_count_updated: 进度数值更新信号
+        is_finished: 完成信号
+    """
 
     progress_msg_updated = Signal(str)
     progress_count_updated = Signal(int)
@@ -89,6 +126,12 @@ class QuantizationWorker(QThread):
         input_file: pathlib.Path,
         quant_types: list[str],
     ) -> None:
+        """初始化量化Worker.
+
+        Args:
+            input_file: 输入的F16 GGUF文件路径
+            quant_types: 需要转换的量化类型列表
+        """
         super().__init__()
 
         self.input_file = input_file
@@ -111,7 +154,7 @@ class QuantizationWorker(QThread):
                     f"正在转换到 {quant_type} 格式...",
                 )
 
-                # 构建命令行参数（确保所有参数都是字符串）
+                # 构建命令行参数
                 cmd = [
                     "llama-quantize",
                     str(self.input_file.name),
@@ -119,7 +162,7 @@ class QuantizationWorker(QThread):
                     quant_type,
                 ]
 
-                # 执行转换命令（使用 cwd 参数避免全局目录变更）
+                # 执行转换命令
                 try:
                     process = subprocess.Popen(
                         cmd,
@@ -284,7 +327,7 @@ class GGUFQuantizerGUI(QMainWindow):
         self.setCentralWidget(main_widget)
 
     def select_file(self) -> None:
-        """选择文件."""
+        """选择F16 GGUF文件."""
         # 使用上次选择的目录作为初始目录
         initial_dir = ""
         if conf.LAST_INPUT_FILE and pathlib.Path(conf.LAST_INPUT_FILE).exists():
@@ -333,21 +376,25 @@ class GGUFQuantizerGUI(QMainWindow):
             filename = f"{_process_gguf_stem(self.input_file.stem)}-{quant_type}.gguf"
             expected_file = dir_path / filename
             if expected_file.exists():
-                # 文件已存在，标记但不禁用，允许用户选择重新生成
+                # 文件已存在，标记并禁用，防止重复生成
                 self.quant_checks[quant_type].setText(
                     f"{self.quant_types[quant_type]} (已存在)",
                 )
                 self.quant_checks[quant_type].setStyleSheet("color: orange;")
+                self.quant_checks[quant_type].setChecked(True)
+                self.quant_checks[quant_type].setEnabled(False)
             else:
                 self.quant_checks[quant_type].setText(
                     self.quant_types[quant_type],
                 )
                 self.quant_checks[quant_type].setStyleSheet("")
+                self.quant_checks[quant_type].setEnabled(True)
 
     def _scroll_to_bottom(self) -> None:
         """滚动输出框到底部."""
         scrollbar = self.output_text.verticalScrollBar()
-        scrollbar.setValue(scrollbar.maximum())
+        if scrollbar:
+            scrollbar.setValue(scrollbar.maximum())
 
     def on_quant_type_changed(self, _state: int) -> None:
         """量化类型变更时保存配置."""
@@ -369,7 +416,7 @@ class GGUFQuantizerGUI(QMainWindow):
         return super().resizeEvent(event)
 
     def start_conversion(self) -> None:
-        """开始转换."""
+        """开始转换量化任务."""
         # 检查是否已有任务在运行
         if self.worker and self.worker.isRunning():
             self.output_text.append("已有转换任务正在进行, 请等待完成")
@@ -377,7 +424,9 @@ class GGUFQuantizerGUI(QMainWindow):
             return
 
         selected_quants: list[str] = [
-            q for q, check in self.quant_checks.items() if check.isChecked()
+            q
+            for q, check in self.quant_checks.items()
+            if check.isChecked() and check.isEnabled()
         ]
 
         if not selected_quants:
@@ -390,19 +439,8 @@ class GGUFQuantizerGUI(QMainWindow):
             self._scroll_to_bottom()
             return
 
-        # 检查是否有已存在的文件将被覆盖
-        existing_files = []
-        for quant_type in selected_quants:
-            filename = f"{_process_gguf_stem(self.input_file.stem)}-{quant_type}.gguf"
-            expected_file = self.input_file.parent / filename
-            if expected_file.exists():
-                existing_files.append(filename)
-
-        if existing_files:
-            self.output_text.append("警告: 将覆盖以下已存在文件:")
-            for existing_file in existing_files:
-                self.output_text.append(f"  - {existing_file}")
-            self._scroll_to_bottom()
+        # 注意：由于selected_quants只包含启用的复选框，
+        # 所以不会包含已存在的禁用文件，无需额外检查覆盖
 
         self.convert_btn.setEnabled(False)
         self.progress_bar.setValue(0)
@@ -459,6 +497,7 @@ class GGUFQuantizerGUI(QMainWindow):
 
 
 def main() -> None:
+    """主程序入口."""
     app = QApplication(sys.argv)
 
     # 检查是否安装了llama.cpp
@@ -469,8 +508,8 @@ def main() -> None:
             check=False,
         )
     except FileNotFoundError:
-        logger.exception("错误: 未找到llama.cpp/quantize工具")
-        logger.exception(
+        logger.error("错误: 未找到llama.cpp/quantize工具")
+        logger.error(
             "请确保已编译llama.cpp并将quantize工具放在llama.cpp/目录下",
         )
         sys.exit(1)

sfi/llmserver/__init__.py

@@ -0,0 +1 @@
+ 

sfi/llmserver/llmserver.py

@@ -6,8 +6,8 @@ import logging
 import os
 import pathlib
 import sys
+from dataclasses import dataclass
 from pathlib import Path
-from types import SimpleNamespace
 from typing import ClassVar
 
 from PySide2.QtCore import QProcess, QTextStream, QUrl
@@ -36,36 +36,61 @@ from PySide2.QtWidgets import (
     QWidget,
 )
 
-CONFIG_FILE = Path.home() / ".sfi" / "llmserver.json"
+CONFIG_FILE = Path.home() / ".pysfi" / "llmserver.json"
 logging.basicConfig(level="INFO", format="%(message)s")
 logger = logging.getLogger(__name__)
 
 
-class LLMServerConfig(SimpleNamespace):
+@dataclass
+class LLMServerConfig:
     """Llama local model server configuration."""
 
     TITLE: str = "Llama Local Model Server"
     WIN_SIZE: ClassVar[list[int]] = [800, 800]
     WIN_POS: ClassVar[list[int]] = [200, 200]
     MODEL_PATH: str = ""
-
     URL: str = "http://127.0.0.1"
     LISTEN_PORT: int = 8080
     LISTEN_PORT_RNG: ClassVar[list[int]] = [1024, 65535]
     THREAD_COUNT_RNG: ClassVar[list[int]] = [1, 24]
     THREAD_COUNT: int = 4
 
-    def __init__(self) -> None:
+    _loaded_from_file: bool = False
+
+    def __post_init__(self) -> None:
         if CONFIG_FILE.exists():
             logger.info("Loading configuration from %s", CONFIG_FILE)
-            self.__dict__.update(json.loads(CONFIG_FILE.read_text()))
+            try:
+                config_data = json.loads(CONFIG_FILE.read_text())
+                # Update instance attributes with loaded values
+                for key, value in config_data.items():
+                    if hasattr(self, key):
+                        setattr(self, key, value)
+                self._loaded_from_file = True
+            except (json.JSONDecodeError, TypeError, AttributeError) as e:
+                logger.error(f"Error loading config from {CONFIG_FILE}: {e}")
         else:
             logger.info("Using default configuration")
 
     def save(self) -> None:
         """Save configuration."""
-        CONFIG_FILE.parent.mkdir(parents=True, exist_ok=True)
-        CONFIG_FILE.write_text(json.dumps(vars(self), indent=4))
+        try:
+            CONFIG_FILE.parent.mkdir(parents=True, exist_ok=True)
+            # Convert dataclass to dict for JSON serialization
+            config_dict = {}
+            for attr_name in dir(self):
+                if not attr_name.startswith("_"):
+                    try:
+                        attr_value = getattr(self, attr_name)
+                        if not callable(attr_value):
+                            config_dict[attr_name] = attr_value
+                    except AttributeError:
+                        continue
+            CONFIG_FILE.write_text(json.dumps(config_dict, indent=4))
+            logger.info(f"Configuration saved to {CONFIG_FILE}")
+        except Exception as e:
+            logger.error(f"Failed to save configuration: {e}")
+            raise
 
 
 conf = LLMServerConfig()
@@ -92,11 +117,8 @@ class LlamaServerGUI(QMainWindow):
         self.init_ui()
         self.setup_process()
 
-        model_path = conf.MODEL_PATH
-        if model_path:
-            self.model_path_input.setText(str(model_path))
-        else:
-            self.model_path_input.setPlaceholderText("Choose model file...")
+        # Apply loaded configuration to UI controls
+        self.apply_config_to_ui()
 
     def init_ui(self) -> None:
         """Initialize user interface."""
@@ -191,6 +213,19 @@ class LlamaServerGUI(QMainWindow):
         self.process.readyReadStandardError.connect(self.handle_stderr)  # type: ignore
         self.process.finished.connect(self.on_process_finished)  # type: ignore
 
+    def apply_config_to_ui(self) -> None:
+        """Apply loaded configuration to UI controls."""
+        # Set model path
+        model_path = conf.MODEL_PATH
+        if model_path:
+            self.model_path_input.setText(str(model_path))
+        else:
+            self.model_path_input.setPlaceholderText("Choose model file...")
+
+        # Set port and thread values
+        self.port_spin.setValue(conf.LISTEN_PORT)
+        self.threads_spin.setValue(conf.THREAD_COUNT)
+
     def on_config_changed(self) -> None:
         """Configuration changed."""
         conf.MODEL_PATH = self.model_path_input.text().strip()
@@ -325,6 +360,21 @@ class LlamaServerGUI(QMainWindow):
         conf.WIN_SIZE = [geometry.width(), geometry.height()]
         return super().resizeEvent(event)
 
+    def closeEvent(self, event) -> None:  # noqa: N802
+        """Handle window close event to ensure configuration is saved."""
+        try:
+            # Save current configuration
+            conf.save()
+            logger.info("Configuration saved successfully on exit")
+        except Exception as e:
+            logger.error(f"Failed to save configuration on exit: {e}")
+        finally:
+            # Also stop server if running
+            if hasattr(self, "process") and self.process.state() == QProcess.Running:
+                self.process.terminate()
+                self.process.waitForFinished(2000)  # Wait up to 2 seconds
+            event.accept()
+
 
 def main() -> None:
     """Main entry point."""