donnemartin/interactive-coding-challenges · error · KeyError

neighbor not found

Error message

neighbor not found

What it means

Raised by Node.remove_neighbor(neighbor) when neighbor.key is not present in self.adj_nodes, meaning no edge from this node to that neighbor exists. Because the method deletes from both adj_weights and adj_nodes, removing a non-existent adjacency would corrupt the dicts, so it fails fast with KeyError.

Source

Thrown at graphs_trees/graph/graph_solution.ipynb:205

    "\n",
    "    def __repr__(self):\n",
    "        return str(self.key)\n",
    "\n",
    "    def __lt__(self, other):\n",
    "        return self.key < other.key\n",
    "\n",
    "    def add_neighbor(self, neighbor, weight=0):\n",
    "        if neighbor is None or weight is None:\n",
    "            raise TypeError('neighbor or weight cannot be None')\n",
    "        neighbor.incoming_edges += 1\n",
    "        self.adj_weights[neighbor.key] = weight\n",
    "        self.adj_nodes[neighbor.key] = neighbor\n",
    "\n",
    "    def remove_neighbor(self, neighbor):\n",
    "        if neighbor is None:\n",
    "            raise TypeError('neighbor cannot be None')\n",
    "        if neighbor.key not in self.adj_nodes:\n",
    "            raise KeyError('neighbor not found')\n",
    "        neighbor.incoming_edges -= 1\n",
    "        del self.adj_weights[neighbor.key]\n",
    "        del self.adj_nodes[neighbor.key]\n",
    "\n",
    "\n",
    "class Graph:\n",
    "\n",
    "    def __init__(self):\n",
    "        self.nodes = {}  # Key = key, val = Node\n",
    "\n",
    "    def add_node(self, key):\n",
    "        if key is None:\n",
    "            raise TypeError('key cannot be None')\n",
    "        if key not in self.nodes:\n",
    "            self.nodes[key] = Node(key)\n",
    "        return self.nodes[key]\n",
    "\n",
    "    def add_edge(self, source_key, dest_key, weight=0):\n",

View on GitHub (pinned to 358f2cc604)

Solutions

  1. Guard the call: check neighbor.key in node.adj_nodes before removing
  2. For undirected graphs, remove from both endpoints' adj lists exactly once, in the correct order
  3. Track edge existence in one place (e.g., a set of (src,dst) tuples) instead of manually managing both directions

Example fix

# before
node.remove_neighbor(other)  # KeyError: neighbor not found

# after
if other.key in node.adj_nodes:
    node.remove_neighbor(other)
Defensive patterns

Strategy: validation

Validate before calling

if neighbor is not None and neighbor.key in node.adj_nodes:
    node.remove_neighbor(neighbor)

Type guard

def edge_exists(node, neighbor):
    return neighbor is not None and neighbor.key in node.adj_nodes

Try / catch

try:
    node.remove_neighbor(neighbor)
except KeyError:
    pass  # edge absent; treat as no-op or log

Prevention

When it happens

Trigger: Calling remove_neighbor for a node never passed to add_neighbor/add_edge; removing an undirected edge by calling remove_neighbor on only one side (or the wrong side); double-removal of the same edge.

Common situations: Graph teardown code assuming symmetric adjacency; replaying edge-removal logs where some edges were already applied; desync between adj_weights and adj_nodes after partial manual mutation.

Related errors


AI-assisted analysis of donnemartin/interactive-coding-challenges@358f2cc604 (2026-08-28). Data as JSON: /api/errors/53629f3a16e7ae94. Report an issue: GitHub.