ksaa 2025.5.16.1__tar.gz → 2025.5.23.0__tar.gz

{ksaa-2025.5.16.1/ksaa.egg-info → ksaa-2025.5.23.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ksaa
-Version: 2025.5.16.1
+Version: 2025.5.23.0
 Summary: Kotones Auto Assistant(kaa) is a script for game 'Gakuen Idol M@ster' that automates the dailies and produces.
 Requires-Python: ~=3.10
 Description-Content-Type: text/markdown

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/WHATS_NEW.md

@@ -1,5 +1,23 @@
 # 更新日志
 ## kaa
+### v2025.5.23.0
+脚本：
+* [修复] 修复 OCR 错误导致无法开始 MASTER 培育（#b4b14fa）
+* [修复] 修复标题包含“gakumas”的窗口会被误认为游戏窗口（#ac3be6a）
+* [修复] 修改“技能卡自选强化”的选择技能卡逻辑（#a7b9dac）
+* [新增] 为 DMM 控制适配 HostProtocol（#68bf47d）
+* [重构] 重构竞赛（#dbe9cae）
+* [重构] 使用基础类中的矩形与点重构任务（#b434278）
+
+界面：
+* [新增] 新增命令行参数 --kill-dmm 与 --kill-game（#397c45b）
+* [重构] 重构 save_config() 函数，移除冗长参数（#9a34749）
+
+框架：
+* [新增] 整合雷电模拟器与 MuMu 模拟器控制到 kaa 中（#548ba04）
+* [新增] 新增 Mumu12Host & LeidianHost 模拟器控制（#6b5f972）
+* [新增] 新增一些基础数据类（#2999367）
+
 ### v2025.5.16.1
 修复了上一个版本打包时缺失了部分文件的问题。
 

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/__init__.py

@@ -17,13 +17,11 @@ from .backend.context import (
     wait
 )
 from .util import (
-    Rect,
     cropped,
     AdaptiveWait,
     Countdown,
     Interval,
     until,
-    crop_rect,
 )
 from .backend.color import (
     hsv_cv2web,

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/bot.py

@@ -8,7 +8,7 @@ from typing_extensions import Self
 from dataclasses import dataclass, field
 from typing import Any, Literal, Callable, Generic, TypeVar, ParamSpec
 
-from kotonebot.ui import user
+from kotonebot.client import Device
 from kotonebot.client.host.protocol import Instance
 from kotonebot.backend.context import init_context, vars
 from kotonebot.backend.context import task_registry, action_registry, Task, Action
@@ -130,49 +130,34 @@ class KotoneBot:
             logger.debug(f'Loading sub-module: {name}')
             try:
                 importlib.import_module(name)
-            except Exception as e:
+            except Exception:
                 logger.error(f'Failed to load sub-module: {name}')
-                logger.exception(f'Error: ')
+                logger.exception('Error: ')
         
         logger.info('Tasks and actions initialized.')
         logger.info(f'{len(task_registry)} task(s) and {len(action_registry)} action(s) loaded.')
 
-    def check_backend(self):
-        from kotonebot.client.host import create_custom
-        from kotonebot.config.manager import load_config
-        # HACK: 硬编码
-        config = load_config(self.config_path, type=self.config_type)
-        config = config.user_configs[0]
-        logger.info('Checking backend...')
-        if config.backend.type == 'custom' and config.backend.check_emulator:
-            exe = config.backend.emulator_path
-            if exe is None:
-                user.error('「检查并启动模拟器」已开启但未配置「模拟器 exe 文件路径」。')
-                raise ValueError('Emulator executable path is not set.')
-            if not os.path.exists(exe):
-                user.error('「模拟器 exe 文件路径」对应的文件不存在！请检查路径是否正确。')
-                raise FileNotFoundError(f'Emulator executable not found: {exe}')
-            self.backend_instance = create_custom(
-                adb_ip=config.backend.adb_ip,
-                adb_port=config.backend.adb_port,
-                adb_emulator_name=config.backend.adb_emulator_name,
-                exe_path=exe,
-                emulator_args=config.backend.emulator_args
-            )
-            if not self.backend_instance.running():
-                logger.info('Starting custom backend...')
-                self.backend_instance.start()
-                logger.info('Waiting for custom backend to be available...')
-                self.backend_instance.wait_available()
-            else:
-                logger.info('Custom backend "%s" already running.', self.backend_instance)
+    def _on_create_device(self) -> Device:
+        """
+        抽象方法，用于创建 Device 类，在 `run()` 方法执行前会被调用。
+
+        所有子类都需要重写该方法。
+        """
+        raise NotImplementedError('Implement `_create_device` before using Kotonebot.')
+
+    def _on_after_init_context(self):
+        """
+        抽象方法，在 init_context() 被调用后立即执行。
+        """
+        pass
 
     def run(self, tasks: list[Task], *, by_priority: bool = True):
         """
         按优先级顺序运行所有任务。
         """
-        self.check_backend()
-        init_context(config_path=self.config_path, config_type=self.config_type)
+        d = self._on_create_device()
+        init_context(config_path=self.config_path, config_type=self.config_type, target_device=d)
+        self._on_after_init_context()
         vars.interrupted.clear()
 
         if by_priority:

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/color.py

@@ -7,8 +7,8 @@ import cv2
 from cv2.typing import MatLike
 
 from .core import unify_image
+from ..primitives import RectTuple, Rect
 from .debug import result as debug_result, debug, color as debug_color
-from ..util import Rect
 
 RgbColorTuple = tuple[int, int, int]
 RgbColorStr = str
@@ -131,6 +131,7 @@ def find(
     * rgb_dist:
         计算图片中每个点的颜色到目标颜色的欧氏距离，并以 442 为最大值归一化到 0-1 之间。
     """
+    _rect = rect.xywh if rect else None
     ret = None
     ret_similarity = 0
     found_color = None
@@ -167,8 +168,8 @@ def find(
     # 寻找结果
     matches: np.ndarray = dist <= (1 - threshold)
     # 只在rect范围内搜索
-    if rect is not None:
-        x, y, w, h = rect
+    if _rect is not None:
+        x, y, w, h = _rect
         search_area = matches[y:y+h, x:x+w]
         if search_area.any():
             # 在裁剪区域中找到最小距离的点
@@ -199,8 +200,8 @@ def find(
                 (min(result_image.shape[1], x+20), min(result_image.shape[0], y+20)),
                 (255, 0, 0), 2)
         # 绘制搜索范围
-        if rect is not None:
-            x, y, w, h = rect
+        if _rect is not None:
+            x, y, w, h = _rect
             # 红色圈出rect
             cv2.rectangle(result_image, (x, y), (x+w, y+h), (0, 0, 255), 2)
         debug_result(
@@ -219,7 +220,7 @@ def color_distance_map(
     image: MatLike | str,
     color: RgbColor,
     *,
-    rect: Rect | None = None,
+    rect: RectTuple | None = None,
 ) -> np.ndarray:
     """
     计算图像中每个像素点到目标颜色的HSL距离，并返回归一化后的距离矩阵。
@@ -266,7 +267,7 @@ def color_distance_map(
     dist = np.sqrt((h_diff * 2)**2 + l_diff**2 + s_diff**2) / np.sqrt(6)
     return dist
 
-def _rect_intersection(rect1: Rect, rect2: Rect) -> Rect | None:
+def _rect_intersection(rect1: RectTuple, rect2: RectTuple) -> RectTuple | None:
     """
     计算两个矩形的交集区域。
     
@@ -311,6 +312,7 @@ def find_all(
     :param max_results: 最大返回结果数量。如果为 None，则返回所有结果。
     :return: 结果列表。
     """
+    _rect: RectTuple | None = rect.xywh if rect is not None else None
     # 计算距离矩阵
     dist = color_distance_map(image, color)
     # 筛选满足要求的点，二值化
@@ -319,8 +321,8 @@ def find_all(
     def filter_by_point(binary: np.ndarray, target_color: RgbColor, dist: np.ndarray, rect: Rect | None = None, max_results: int | None = None) -> list[FindColorPointResult]:
         results = []
         
-        if rect is not None:
-            x, y, w, h = rect
+        if _rect is not None:
+            x, y, w, h = _rect
             search_area = binary[y:y+h, x:x+w]
             local_dist = dist[y:y+h, x:x+w]
             
@@ -370,8 +372,8 @@ def find_all(
             assert len(contour_rect) == 4
             
             # 如果指定了rect，计算轮廓外接矩形与rect的交集
-            if rect is not None:
-                intersection = _rect_intersection(contour_rect, rect)
+            if _rect is not None:
+                intersection = _rect_intersection(contour_rect, _rect)
                 if intersection is None:
                     continue
                     
@@ -433,8 +435,8 @@ def find_all(
                 (min(result_image.shape[1], x+10), min(result_image.shape[0], y+10)),
                 (255, 0, 0), 1)
         # 绘制搜索范围
-        if rect is not None:
-            x, y, w, h = rect
+        if _rect is not None:
+            x, y, w, h = _rect
             cv2.rectangle(result_image, (x, y), (x+w, y+h), (0, 0, 255), 2)
         
         debug_result(
@@ -466,10 +468,11 @@ def dominant_color(
     :param count: 提取的颜色数量。默认为 1。
     :param rect: 提取范围。如果为 None，则在整个图像中提取。
     """
+    _rect: RectTuple | None = rect.xywh if rect is not None else None
     # 载入/裁剪图像
     img = unify_image(image)
-    if rect is not None:
-        x, y, w, h = rect
+    if _rect is not None:
+        x, y, w, h = _rect
         img = img[y:y+h, x:x+w]
     
     pixels = np.float32(img.reshape(-1, 3))
@@ -497,8 +500,8 @@ def dominant_color(
     if debug.enabled:
         origin_image = unify_image(image)
         result_image = origin_image.copy()
-        if rect is not None:
-            x, y, w, h = rect
+        if _rect is not None:
+            x, y, w, h = _rect
             cv2.rectangle(result_image, (x, y), (x + w, y + h), (0, 0, 255), 2)
         debug_result(
             'color.dominant_color',

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/context/context.py

@@ -25,7 +25,6 @@ import cv2
 from cv2.typing import MatLike
 
 from kotonebot.client.device import Device
-from kotonebot.util import Rect
 import kotonebot.backend.image as raw_image
 from kotonebot.backend.image import (
     TemplateMatchResult,
@@ -52,6 +51,7 @@ from kotonebot.backend.core import Image, HintBox
 from kotonebot.errors import KotonebotWarning
 from kotonebot.client.factory import DeviceImpl
 from kotonebot.backend.preprocessor import PreprocessorProtocol
+from kotonebot.primitives import Rect
 
 OcrLanguage = Literal['jp', 'en']
 ScreenshotMode = Literal['auto', 'manual', 'manual-inherit']
@@ -174,11 +174,30 @@ def is_manual_screenshot_mode() -> bool:
 
 class ContextGlobalVars:
     def __init__(self):
-        self.auto_collect: bool = False
-        """遇到未知P饮料/卡片时，是否自动截图并收集"""
+        self.__vars = dict[str, Any]()
         self.interrupted: Event = Event()
         """用户请求中断事件"""
 
+    def __getitem__(self, key: str) -> Any:
+        return self.__vars[key]
+
+    def __setitem__(self, key: str, value: Any) -> None:
+        self.__vars[key] = value
+
+    def __delitem__(self, key: str) -> None:
+        del self.__vars[key]
+
+    def __contains__(self, key: str) -> bool:
+        return key in self.__vars
+
+    def get(self, key: str, default: Any = None) -> Any:
+        return self.__vars.get(key, default)
+
+    def set(self, key: str, value: Any) -> None:
+        self.__vars[key] = value
+
+    def clear(self):
+        self.__vars.clear()
 
 class ContextStackVars:
     stack: list['ContextStackVars'] = []
@@ -730,7 +749,13 @@ class ContextDevice(Device):
 
 
 class Context(Generic[T]):
-    def __init__(self, config_path: str, config_type: Type[T], screenshot_impl: Optional[DeviceImpl] = None):
+    def __init__(
+        self,
+        config_path: str,
+        config_type: Type[T],
+        screenshot_impl: Optional[DeviceImpl] = None,
+        device: Optional[Device] = None
+    ):
         self.__ocr = ContextOcr(self)
         self.__image = ContextImage(self)
         self.__color = ContextColor(self)
@@ -744,7 +769,7 @@ class Context(Generic[T]):
         if screenshot_impl is None:
             screenshot_impl = self.config.current.backend.screenshot_impl
         logger.info(f'Using "{screenshot_impl}" as screenshot implementation')
-        self.__device = ContextDevice(create_device(f'{ip}:{port}', screenshot_impl))
+        self.__device = ContextDevice(device or create_device(f'{ip}:{port}', screenshot_impl))
 
     def inject(
         self,
@@ -803,11 +828,12 @@ class Context(Generic[T]):
     def config(self) -> 'ContextConfig[T]':
         return self.__config
 
+@deprecated('使用 Rect 类的实例方法代替')
 def rect_expand(rect: Rect, left: int = 0, top: int = 0, right: int = 0, bottom: int = 0) -> Rect:
     """
     向四个方向扩展矩形区域。
     """
-    return (rect[0] - left, rect[1] - top, rect[2] + right + left, rect[3] + bottom + top)
+    return Rect(rect.x1 - left, rect.y1 - top, rect.w + right + left, rect.h + bottom + top)
 
 def use_screenshot(*args: MatLike | None) -> MatLike:
     for img in args:
@@ -854,7 +880,8 @@ def init_context(
     config_path: str = 'config.json',
     config_type: Type[T] = dict[str, Any],
     force: bool = False,
-    screenshot_impl: Optional[DeviceImpl] = None
+    screenshot_impl: Optional[DeviceImpl] = None,
+    target_device: Device | None = None,
 ):
     """
     初始化 Context 模块。
@@ -867,11 +894,17 @@ def init_context(
         若为 `True`，则忽略已存在的 Context 实例，并重新创建一个新的实例。
     :param screenshot_impl: 截图实现。
         若为 `None`，则使用默认配置文件中指定的截图实现。
+    :param target_device: 目标设备
     """
     global _c, device, ocr, image, color, vars, debug, config
     if _c is not None and not force:
         return
-    _c = Context(config_path=config_path, config_type=config_type, screenshot_impl=screenshot_impl)
+    _c = Context(
+        config_path=config_path,
+        config_type=config_type,
+        screenshot_impl=screenshot_impl,
+        device=target_device
+    )
     device._FORWARD_getter = lambda: _c.device # type: ignore
     ocr._FORWARD_getter = lambda: _c.ocr # type: ignore
     image._FORWARD_getter = lambda: _c.image # type: ignore

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/core.py

@@ -1,14 +1,13 @@
 import logging
 from functools import cache
-from typing import Callable, overload, TYPE_CHECKING
+from typing import Callable
 
 import cv2
 from cv2.typing import MatLike
 
 from kotonebot.util import cv2_imread
+from kotonebot.primitives import RectTuple, Rect, Point
 from kotonebot.errors import ResourceFileMissingError
-if TYPE_CHECKING:
-    from kotonebot.util import Rect
 
 class Ocr:
     def __init__(
@@ -67,20 +66,7 @@ class Image:
             return f'<Image: "{self.name}" at {self.path}>'
 
 
-class HintBox(tuple[int, int, int, int]):
-    def __new__(
-        cls,
-        x1: int,
-        y1: int,
-        x2: int,
-        y2: int,
-        *,
-        source_resolution: tuple[int, int],
-    ):
-        w = x2 - x1
-        h = y2 - y1
-        return super().__new__(cls, [x1, y1, w, h])
-    
+class HintBox(Rect):
     def __init__(
         self,
         x1: int,
@@ -92,11 +78,7 @@ class HintBox(tuple[int, int, int, int]):
         description: str | None = None,
         source_resolution: tuple[int, int],
     ):
-        self.x1 = x1
-        self.y1 = y1
-        self.x2 = x2
-        self.y2 = y2
-        self.name = name
+        super().__init__(x1, y1, x2 - x1, y2 - y1, name=name)
         self.description = description
         self.source_resolution = source_resolution
 
@@ -109,17 +91,12 @@ class HintBox(tuple[int, int, int, int]):
         return self.y2 - self.y1
     
     @property
-    def rect(self) -> 'Rect':
+    def rect(self) -> RectTuple:
         return self.x1, self.y1, self.width, self.height
 
-class HintPoint(tuple[int, int]):
-    def __new__(cls, x: int, y: int):
-        return super().__new__(cls, (x, y))
-    
+class HintPoint(Point):
     def __init__(self, x: int, y: int, *, name: str | None = None, description: str | None = None):
-        self.x = x
-        self.y = y
-        self.name = name
+        super().__init__(x, y, name=name)
         self.description = description
 
     def __repr__(self) -> str:

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/dispatch.py

@@ -11,7 +11,7 @@ from typing_extensions import deprecated
 from dataclasses import dataclass
 
 from kotonebot.backend.ocr import StringMatchFunction
-from kotonebot.util import Rect, is_rect
+from kotonebot.primitives import Rect, is_rect
 
 from .core import Image
 

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/image.py

@@ -1,16 +1,16 @@
 import os
 from logging import getLogger
-from typing import NamedTuple, Protocol, TypeVar, Sequence, runtime_checkable
+from typing import NamedTuple, Protocol, Sequence, runtime_checkable
 
 import cv2
 import numpy as np
-from cv2.typing import MatLike, Size
+from cv2.typing import MatLike, Rect as CvRect
 from skimage.metrics import structural_similarity
 
 from .core import Image, unify_image
-from ..util import Rect, Point
-from .debug import result as debug_result, debug, img
 from .preprocessor import PreprocessorProtocol
+from kotonebot.primitives import Point as KbPoint, Rect as KbRect, Size as KbSize
+from .debug import result as debug_result, debug, img
 
 logger = getLogger(__name__)
 
@@ -24,46 +24,46 @@ class TemplateNoMatchError(Exception):
 @runtime_checkable
 class ResultProtocol(Protocol):
     @property
-    def rect(self) -> Rect:
+    def rect(self) -> KbRect:
         """结果区域。左上角坐标和宽高。"""
         ...
 
 
 class TemplateMatchResult(NamedTuple):
     score: float
-    position: Point
+    position: KbPoint
     """结果位置。左上角坐标。"""
-    size: Size
+    size: KbSize
     """输入模板的大小。宽高。"""
 
     @property
-    def rect(self) -> Rect:
-        """结果区域。左上角坐标和宽高。"""
-        return (self.position[0], self.position[1], self.size[0], self.size[1])
+    def rect(self) -> KbRect:
+        """结果区域。"""
+        return KbRect(self.position[0], self.position[1], self.size[0], self.size[1])
     
     @property
-    def right_bottom(self) -> Point:
+    def right_bottom(self) -> KbPoint:
         """结果右下角坐标。"""
-        return (self.position[0] + self.size[0], self.position[1] + self.size[1])
+        return KbPoint(self.position[0] + self.size[0], self.position[1] + self.size[1])
 
 class MultipleTemplateMatchResult(NamedTuple):
     score: float
-    position: Point
+    position: KbPoint
     """结果位置。左上角坐标。"""
-    size: Size
+    size: KbSize
     """命中模板的大小。宽高。"""
     index: int
     """命中模板在列表中的索引。"""
 
     @property
-    def rect(self) -> Rect:
+    def rect(self) -> KbRect:
         """结果区域。左上角坐标和宽高。"""
-        return (self.position[0], self.position[1], self.size[0], self.size[1])
+        return KbRect(self.position[0], self.position[1], self.size[0], self.size[1])
     
     @property
-    def right_bottom(self) -> Point:
+    def right_bottom(self) -> KbPoint:
         """结果右下角坐标。"""
-        return (self.position[0] + self.size[0], self.position[1] + self.size[1])
+        return KbPoint(self.position[0] + self.size[0], self.position[1] + self.size[1])
 
     @classmethod
     def from_template_match_result(cls, result: TemplateMatchResult, index: int):
@@ -76,13 +76,13 @@ class MultipleTemplateMatchResult(NamedTuple):
 
 class CropResult(NamedTuple):
     score: float
-    position: Point
-    size: Size
+    position: KbPoint
+    size: KbSize
     image: MatLike
 
     @property
-    def rect(self) -> Rect:
-        return (self.position[0], self.position[1], self.size[0], self.size[1])
+    def rect(self) -> KbRect:
+        return KbRect(self.position[0], self.position[1], self.size[0], self.size[1])
 
 def _draw_result(image: MatLike, matches: Sequence[ResultProtocol] | ResultProtocol | None) -> MatLike:
     """在图像上绘制匹配结果的矩形框。"""
@@ -92,7 +92,7 @@ def _draw_result(image: MatLike, matches: Sequence[ResultProtocol] | ResultProto
         matches = [matches]
     result_image = image.copy()
     for match in matches:
-        cv2.rectangle(result_image, match.rect, (0, 0, 255), 2)
+        cv2.rectangle(result_image, match.rect.xywh, (0, 0, 255), 2)
     return result_image
 
 def _img2str(image: MatLike | str | Image | None) -> str:
@@ -229,8 +229,8 @@ def template_match(
         
         matches.append(TemplateMatchResult(
             score=score,
-            position=(int(x), int(y)),
-            size=(int(w), int(h))
+            position=KbPoint(int(x), int(y)),
+            size=KbSize(int(w), int(h))
         ))
         
         # 如果达到最大结果数，提前结束
@@ -242,7 +242,7 @@ def template_match(
 def hist_match(
     image: MatLike | str,
     template: MatLike | str,
-    rect: Rect | None = None,
+    rect: CvRect | None = None,
     threshold: float = 0.9,
 ) -> bool:
     """

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/ocr.py

@@ -15,8 +15,9 @@ from thefuzz import fuzz as _fuzz
 from rapidocr_onnxruntime import RapidOCR
 
 
+from ..util import lf_path
+from ..primitives import Rect, Point
 from .core import HintBox, Image, unify_image
-from ..util import Rect, lf_path
 from .debug import result as debug_result, debug
 
 logger = logging.getLogger(__name__)
@@ -65,16 +66,16 @@ class OcrResultList(list[OcrResult]):
         将所有识别结果合并为一个大结果。
         """
         if not self:
-            return OcrResult('', (0, 0, 0, 0), 0, (0, 0, 0, 0))
+            return OcrResult('', Rect(0, 0, 0, 0), 0, Rect(0, 0, 0, 0))
         text = [r.text for r in self]
         confidence = sum(r.confidence for r in self) / len(self)
         points = []
         for r in self:
-            points.append((r.rect[0], r.rect[1]))
-            points.append((r.rect[0] + r.rect[2], r.rect[1]))
-            points.append((r.rect[0], r.rect[1] + r.rect[3]))
-            points.append((r.rect[0] + r.rect[2], r.rect[1] + r.rect[3]))
-        rect = bounding_box(points)
+            points.append(Point(r.rect.x1, r.rect.y1))
+            points.append(Point(r.rect.x1 + r.rect.w, r.rect.y1))
+            points.append(Point(r.rect.x1, r.rect.y1 + r.rect.h))
+            points.append(Point(r.rect.x1 + r.rect.w, r.rect.y1 + r.rect.h))
+        rect = Rect(xywh=bounding_box(points))
         text = '\n'.join(text)
         if remove_newlines:
             text = text.replace('\n', '')
@@ -245,7 +246,7 @@ def _draw_result(image: 'MatLike', result: list[OcrResult]) -> 'MatLike':
     for r in result:
         # 画矩形框
         draw.rectangle(
-            [r.rect[0], r.rect[1], r.rect[0] + r.rect[2], r.rect[1] + r.rect[3]], 
+            [r.rect.x1, r.rect.y1, r.rect.x1 + r.rect.w, r.rect.y1 + r.rect.h], 
             outline=(255, 0, 0), 
             width=2
         )
@@ -257,8 +258,8 @@ def _draw_result(image: 'MatLike', result: list[OcrResult]) -> 'MatLike':
         text_height = text_bbox[3] - text_bbox[1]
         
         # 计算文本位置
-        text_x = r.rect[0]
-        text_y = r.rect[1] - text_height - 5 if r.rect[1] > text_height + 5 else r.rect[1] + r.rect[3] + 5
+        text_x = r.rect.x1
+        text_y = r.rect.y1 - text_height - 5 if r.rect.y1 > text_height + 5 else r.rect.y1 + r.rect.h + 5
         
         # 添加padding
         padding = 4
@@ -313,7 +314,7 @@ class Ocr:
         :return: 所有识别结果
         """
         if rect is not None:
-            x, y, w, h = rect
+            x, y, w, h = rect.xywh
             img = img[y:y+h, x:x+w]
         original_img = img
         if pad:
@@ -338,8 +339,8 @@ class Ocr:
             # result_rect (x, y, w, h)
             if rect is not None:
                 original_rect = (
-                    result_rect[0] + rect[0] - pos_in_padded_img[0],
-                    result_rect[1] + rect[1] - pos_in_padded_img[1],
+                    result_rect[0] + rect.x1 - pos_in_padded_img[0],
+                    result_rect[1] + rect.y1 - pos_in_padded_img[1],
                     result_rect[2],
                     result_rect[3]
                 )
@@ -352,8 +353,8 @@ class Ocr:
             confidence = float(r[2])
             ret.append(OcrResult(
                 text=text,
-                rect=result_rect,
-                original_rect=original_rect,
+                rect=Rect(xywh=result_rect),
+                original_rect=Rect(xywh=original_rect),
                 confidence=confidence
             ))
         ret = OcrResultList(ret)
@@ -392,7 +393,7 @@ class Ocr:
         """
         if hint is not None:
             warnings.warn("使用 `rect` 参数代替")
-            if ret := self.find(img, text, rect=hint):
+            if ret := self.find(img, text, rect=Rect(xywh=hint.rect)):
                 logger.debug(f"find: {text} SUCCESS [hint={hint}]")
                 return ret
             logger.debug(f"find: {text} FAILED [hint={hint}]")
@@ -430,7 +431,7 @@ class Ocr:
         # HintBox 处理
         if hint is not None:
             warnings.warn("使用 `rect` 参数代替")
-            result = self.find_all(img, texts, rect=hint, pad=pad)
+            result = self.find_all(img, texts, rect=Rect(xywh=hint.rect), pad=pad)
             if all(result):
                 return result
 

{ksaa-2025.5.16.1 → ksaa-2025.5.23.0}/kotonebot/backend/preprocessor.py

@@ -1,4 +1,3 @@
-
 from typing import Protocol, Literal
 
 import cv2