Fail to analyze list comprehension code

[lantlani]

Chalice does not deploy below code with error message 'key_name'.
If I change the code from item['payload'][key_name] to item['payload']['data'], it works well.

I have used version 1.0.0b1.

@app.route('/data/{node_name}', methods=['GET'], cors=True, authorizer=authorizer)
def retrieve_data(node_name):
    key_name = 'data'
    response['Items'] = [
        { 'Timestamp' : 10, 'payload' : { 'data' : 1}},
        { 'Timestamp' : 11, 'payload' : { 'data' : 2}}
    ]
    stamp = [item['Timestamp'] for item in response['Items']]
    temps = [item['payload'][key_name] for item in response['Items']]
    hums = [item['payload'][key_name] for item in response['Items']]

[Segflow]

How response in created? is it global?
Running this code works for me

@app.route('/test')
def test_api():
    response = {}
    key_name = 'data'
    response['Items'] = [
        {'Timestamp' : 10, 'payload' : { 'data' : 1}},
        {'Timestamp' : 11, 'payload' : { 'data' : 2}}
    ]
    stamp = [item['Timestamp'] for item in response['Items']]
    temps = [item['payload'][key_name] for item in response['Items']]
    hums = [item['payload'][key_name] for item in response['Items']]
    
    return {
        'stamp': stamp,
        'temps': temps,
    }

Result: {"stamp": [10, 11], "temps": [1, 2]}

[lantlani]

@Segflow thanks for your comment.
For your question, response is local variable.
Does your environment run gen-policy your code?

I could execute my code in local option, however, I cannot run 'deploy' or 'gen-policy'.
As far as I know,
Whenever I try to deploy this code to AWS, Chalice run analyse the code.
This process does not finished with above example.

[lantlani]

I guess this issue is related to below codes.

def _get_matching_sub_namespace(self, name):
        # type: (str) -> symtable.SymbolTable
        namespaces = [
            t for t in self._symbol_table.get_sub_namespaces()
            if t.get_name() == name]
        if not namespaces:
            return
        # We're making a simplification and using the genexpr subnamespace.
        # This has potential to miss a client call but we don't do
        # inference on node.generators so this doesn't matter for now.
        child_scope = namespaces[0]
        return child_scope

To handle list comprehension for temps, it seems to use namespaces[1] instead of namespaces[0].
However, I do not have proper understanding for AST related code to fix this issue.
So I changed my code as below, I could avoid problem.
I am not sure this could be related to my python environment, which is anaconda virtual environment with python 3.6.

        temps = []
        hums = []
        for item in response['Items']:
            temps.append(item['payload'][temp_key])
            hums.append(item['payload'][hum_key])