from __future__ import annotations """页面识别条件组合器。""" from dataclasses import dataclass from typing import Iterable, Union from module.atom.gif import RuleGif from module.atom.image import RuleImage from module.atom.ocr import RuleOcr from tasks.GameUi.common import RecognizerLike, invoke_task_callable class Matcher: """页面识别条件的抽象基类。""" def evaluate(self, task) -> bool: """执行一次识别判断。 Args: task: 当前运行时任务对象。 Returns: 识别是否成功。 """ raise NotImplementedError @dataclass(frozen=True) class AtomMatcher(Matcher): """最小识别单元。 Attributes: target: 单个识别目标,可以是 Rule 元对象或自定义函数。 """ target: RecognizerLike def evaluate(self, task) -> bool: """执行单个识别目标的判断。""" target = self.target if callable(target) and not isinstance(target, (RuleImage, RuleGif, RuleOcr)): return bool(invoke_task_callable(target, task)) if isinstance(target, RuleOcr): return bool(task.ocr_appear(target)) if isinstance(target, (RuleImage, RuleGif)): return bool(task.appear(target)) return False @dataclass(frozen=True) class AnyMatcher(Matcher): """任一子条件命中即成功。""" children: tuple[Matcher, ...] def evaluate(self, task) -> bool: """按“或”语义执行识别。""" return any(child.evaluate(task) for child in self.children) @dataclass(frozen=True) class AllMatcher(Matcher): """所有子条件都命中才成功。""" children: tuple[Matcher, ...] def evaluate(self, task) -> bool: """按“且”语义执行识别。""" return all(child.evaluate(task) for child in self.children) @dataclass(frozen=True) class NotMatcher(Matcher): """对子条件结果取反。""" child: Matcher def evaluate(self, task) -> bool: """执行取反识别。""" return not self.child.evaluate(task) def ensure_matcher(target: Union[Matcher | RecognizerLike | Iterable[RecognizerLike] | None]) -> Matcher | None: """将任意识别输入规范化为 `Matcher`。 Args: target: 原始识别定义,可以是 `Matcher`、Rule 元对象、可调用对象或可迭代集合。 Returns: 规范化后的 `Matcher`;传入 `None` 时返回 `None`。 """ if target is None: return None if isinstance(target, Matcher): return target if isinstance(target, (list, tuple, set)): return AnyMatcher(tuple(ensure_matcher(item) for item in target if item is not None)) return AtomMatcher(target) def any_of(*targets: RecognizerLike | Matcher) -> Matcher: """创建“任一命中即可”的组合识别条件。 Args: *targets: 需要参与组合的识别条件。 Returns: `AnyMatcher` 实例。 """ return AnyMatcher(tuple(ensure_matcher(target) for target in targets if target is not None)) def all_of(*targets: RecognizerLike | Matcher) -> Matcher: """创建“全部命中才成功”的组合识别条件。 Args: *targets: 需要参与组合的识别条件。 Returns: `AllMatcher` 实例。 """ return AllMatcher(tuple(ensure_matcher(target) for target in targets if target is not None)) def not_(target: RecognizerLike | Matcher) -> Matcher: """创建取反识别条件。 Args: target: 需要取反的识别条件。 Returns: `NotMatcher` 实例。 """ return NotMatcher(ensure_matcher(target)) def collect_rule_images(target: Union[Matcher | RecognizerLike | Iterable[RecognizerLike] | None]) -> tuple[RuleImage, ...]: """提取 matcher 树中可批量预取的 `RuleImage` 原子节点。""" images: list[RuleImage] = [] seen = set() def visit(node): if node is None: return if isinstance(node, RuleImage): cache_key = id(node) if cache_key in seen: return seen.add(cache_key) images.append(node) return if isinstance(node, AtomMatcher): visit(node.target) return if isinstance(node, (AnyMatcher, AllMatcher)): for child in node.children: visit(child) return if isinstance(node, NotMatcher): visit(node.child) return if isinstance(node, Matcher): return if isinstance(node, (list, tuple, set)): for item in node: visit(item) visit(target) return tuple(images)