python - dict.keys() 返回的键 k 在执行 dict[k] : KeyError on existing key 时导致 KeyError

Question

下面的代码

for k in list(g_score.keys()):
    print(g_score[k])

返回 KeyError对我来说:

Traceback (most recent call last):
  File "./np.py", line 134, in <module>
    main()
  File "./np.py", line 131, in main
    NPuzzle(n).solve()
  File "./np.py", line 116, in solve
    print(g_score[k])
KeyError: 3

我不明白这怎么可能 print(list(g_score.keys()))是[4, 3, 7] . 3字典里写的很清楚。

---

对于上下文，我正在尝试对 N-Puzzle 问题实现 A* 搜索(我什至不确定 A* 是否正确实现，因为我无法克服这个错误)，并且以下State类和 solve功能:

class State:
    def __init__(self, blank_idx, puzzle, g=0, h=0):
        self.blank_idx = blank_idx
        self.puzzle = puzzle
        self.f = g + h

    def __eq__(self, other):
        return self.puzzle == other.puzzle

    def __hash__(self):
        return self.f + self.blank_idx

    def __lt__(self, other):
        return self.f < other.f

    def __repr__(self):
        return str(self.f)

...

class NPuzzle:

    # ...other stuff

    def solve(self):
      start_state = State(
          self.puzzle.index(' '),
          self.puzzle,
          0,
          self.cost(self.puzzle)
      )

      g_score = {start_state: 0}
      open_set = [start_state]
      path = {}

      while open_set:
          state = open_set[0]

          if state.puzzle == self.goal_state:
              break

          heappop(open_set)
          for next_state in self.neighbors(state):
              g = g_score[state] + 1
              if next_state in g_score and g >= g_score[next_state]:
                  continue

              path[next_state] = state
              g_score[next_state] = g
              next_state.f = g + self.cost(next_state.puzzle)
              heappush(open_set, next_state)

我第一次遇到错误是在我所在的行上:

g = g_score[state] + 1

我不确定为什么这个 KeyError正在发生，但我假设这可能与我的习惯有关 __hash()__功能。

Answer 1

好吧，原来问题是我立即更改了哈希函数所依赖的 State 实例的属性...糟糕:

State 的__hash()__ 函数是:

return self.f + self.blank_idx

我在 g_score 中存储 State 的方式如下:

g_score[next_state] = g
next_state.f = g + self.cost(next_state.puzzle)

原来上面的代码破坏了一切，因为它使用 next_state.f 将 next_state 放入 g_score，但紧接着在下一行我变异 next_state.f

像这样切换两个语句的顺序:

next_state.f = g + self.cost(next_state.puzzle)
g_score[next_state] = g

解决了我的问题。