search.ipynb: Broken A* visualizations

[kylekyle]

A few of the visualizations in the A* search (section 9) are broken. This all happened undera fresh installation of anaconda 3 (python 3.6.5). It seems to be related to the best_first_graph_search function in search.py, but I'm not positive.

Either way, the widget that provides a dropdown of algorithms to visualize on the Romania map fails with the following error:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
/anaconda3/lib/python3.6/site-packages/ipywidgets/widgets/interaction.py in update(self, *args)
    249                     value = widget.get_interact_value()
    250                     self.kwargs[widget._kwarg] = value
--> 251                 self.result = self.f(**self.kwargs)
    252                 show_inline_matplotlib_plots()
    253                 if self.auto_display and self.result is not None:

~/Desktop/AI/aima-python/notebook.py in visualize_callback(Visualize)
   1020                 user_algorithm = algorithm[algo_dropdown.value]
   1021 
-> 1022                 iterations, all_node_colors, node = user_algorithm(problem)
   1023                 solution = node.solution()
   1024                 all_node_colors.append(final_path_colors(all_node_colors[0], problem, solution))

TypeError: 'Node' object is not iterable

A few of the 8-puzzle (but not all) examples that follow the Romania one are also broken, all with errors similar to this:

~/Desktop/AI/aima-python/search.py in <lambda>(n)
    410     else in your Problem subclass."""
    411     h = memoize(h or problem.h, 'h')
--> 412     return best_first_graph_search(problem, lambda n: n.path_cost + h(n))
    413 
    414 # ______________________________________________________________________________

~/Desktop/AI/aima-python/utils.py in memoized_fn(obj, *args)
    374                 return getattr(obj, slot)
    375             else:
--> 376                 val = fn(obj, *args)
    377                 setattr(obj, slot, val)
    378                 return val

<ipython-input-63-594d40b3aa01> in linear(node)
      1 # Heuristics for 8 Puzzle Problem
      2 def linear(node):
----> 3     return sum([1 if node.state[i] != goal[i] else 0 for i in range(8)])
      4 
      5 def manhattan(node):

<ipython-input-63-594d40b3aa01> in <listcomp>(.0)
      1 # Heuristics for 8 Puzzle Problem
      2 def linear(node):
----> 3     return sum([1 if node.state[i] != goal[i] else 0 for i in range(8)])
      4 
      5 def manhattan(node):

TypeError: 'NoneType' object is not subscriptable

[dmeoli]

This is resolved on current master. The original break was that the visualization/dropdown called raw astar_search (returns a Node) into a 3-tuple unpack. The notebook now defines astar_search_graph wrapping best_first_graph_search_for_vis (returning iterations, all_node_colors, node), and the dropdown maps every algorithm to its *_for_vis variant; the 8-puzzle heuristic A* cells also run without the old NoneType error. Executing search.ipynb end-to-end shows no error in the A* cells. Closing as resolved.