Python - nested comprehension list

Question

I have the following structure:

input = {
'clark': {
          'elements':['id[3]_element[1]','id[3]_element[2]','id[3]_element[3]'],
          'nickname': 'superman',
          'registered': 'Y'
         },
'rodger': {
          'elements':['id[2]_element[1]','id[2]_element[2]','id[2]_element[3]'],
          'nickname': 'rodjeuur',
          'registered': 'N'
         },
'ratman': {
          'elements':['id[6]_element[1]','id[6]_element[2]','id[6]_element[3]'],
          'nickname': 'rat',
          'registered': 'Y'
         },
'marylin': {
          'elements':['id[90]_element[1]','id[90]_element[2]','id[90]_element[3]'],
          'nickname': 'monroe',
          'registered': 'Y'
         },
'snoopy': {
          'elements':['id[1]_element[1]','id[1]_element[2]','id[1]_element[3]'],
          'nickname': 'lame',
          'registered': 'N'
         },
'dummy_field[1]': 'dogfood',
'dummy_field[2]': 'monkeyjump',
'id[3]_element[1]':{
    'feature': 'vision glass',
    'value': 0.12,
    'color': 'blue'
},
'id[3]_element[2]':{
    'feature': 'height',
    'value': 1.88
},
'id[3]_element[3]':{
    'feature': 'cookingskill',
    'value': 0
},
'id[6]_element[1]':{
    'feature': 'rat teeth',
    'value': 25,
    'similarity':'luis suarez'
},
'id[6]_element[2]':{
    'feature': 'height',
    'value': 1.70
},
'id[6]_element[3]':{
    'feature': 'cookingskill',
    'value': 0.5
},
'id[2]_element[1]':{
    'feature': 'tennis raquet',
    'value': 500000,
    'particularity':'second arm'
},
'id[2]_element[2]':{
    'feature': 'height',
    'value': 1.78
},
'id[2]_element[3]':{
    'feature': 'trickshot',
    'value': 120
},
'id[1]_element[1]':{
    'feature': 'mood',
    'value': 0,
    'particularity':'depressed'
},
'id[1]_element[2]':{
    'feature': 'height',
    'value': 0.45
},
'id[1]_element[3]':{
    'feature': 'sadness',
    'value': 10000000
},
'id[90]_element[1]':{
    'feature': 'charm',
    'value': 500000
},
'id[90]_element[2]':{
    'feature': 'height',
    'value': 1.72
},
'id[90]_element[3]':{
    'feature': 'cookingskills',
    'value': 0
}
}

And a list over which to iterate:

lnames=['clark','ratman','snoopy','marylin']

I would like to obtain the following list of dictionnaries as result:

output=[{'vision glass':0.12, 'height':1.88, 'cookingskill':0},
    {'rat teeth':25, 'height':1.70, 'cookingskill':0.5},
    {'mood':0, 'height':0.45, 'sadness':10000000},
    {'charm':500000, 'height':1.72, 'cookingskills':0}
]

I started the very basic code:

output=[{.... for el in input[name]['elements']} for name in lnames]

I can reach each name elements and iterate over them to access the different 'features' and 'values', but I do not know how to group them into dictionnaries for a given name.

Do anyone has an idea?

EDIT

Now supposing I have the same input except there is an '@' in the field 'elements' of each name:

input = {
'clark': {
          'elements':['@id[3]_element[1]','@id[3]_element[2]','@id[3]_element[3]'],
          'nickname': 'superman',
          'registered': 'Y'
         },
'rodger': {
          'elements':['@id[2]_element[1]','@id[2]_element[2]','@id[2]_element[3]'],
          'nickname': 'rodjeuur',
          'registered': 'N'
         },
'ratman': {
          'elements':['@id[6]_element[1]','@id[6]_element[2]','@id[6]_element[3]'],
          'nickname': 'rat',
          'registered': 'Y'
         },
'marylin': {
          'elements':['@id[90]_element[1]','@id[90]_element[2]','@id[90]_element[3]'],
          'nickname': 'monroe',
          'registered': 'Y'
         },
'snoopy': {
          'elements':['@id[1]_element[1]','@id[1]_element[2]','@id[1]_element[3]'],
          'nickname': 'lame',
          'registered': 'N'
         },
'dummy_field[1]': 'dogfood',
'dummy_field[2]': 'monkeyjump',
'id[3]_element[1]':{
    'feature': 'vision glass',
    'value': 0.12,
    'color': 'blue'
},
'id[3]_element[2]':{
    'feature': 'height',
    'value': 1.88
},
'id[3]_element[3]':{
    'feature': 'cookingskill',
    'value': 0
},
'id[6]_element[1]':{
    'feature': 'rat teeth',
    'value': 25,
    'similarity':'luis suarez'
},
'id[6]_element[2]':{
    'feature': 'height',
    'value': 1.70
},
'id[6]_element[3]':{
    'feature': 'cookingskill',
    'value': 0.5
},
'id[2]_element[1]':{
    'feature': 'tennis raquet',
    'value': 500000,
    'particularity':'second arm'
},
'id[2]_element[2]':{
    'feature': 'height',
    'value': 1.78
},
'id[2]_element[3]':{
    'feature': 'trickshot',
    'value': 120
},
'id[1]_element[1]':{
    'feature': 'mood',
    'value': 0,
    'particularity':'depressed'
},
'id[1]_element[2]':{
    'feature': 'height',
    'value': 0.45
},
'id[1]_element[3]':{
    'feature': 'sadness',
    'value': 10000000
},
'id[90]_element[1]':{
    'feature': 'charm',
    'value': 500000
},
'id[90]_element[2]':{
    'feature': 'height',
    'value': 1.72
},
'id[90]_element[3]':{
    'feature': 'cookingskills',
    'value': 0
}
}

Using the folowing code:

import re

def at(x):
    return input[re.sub(r'@','',x)] 

I believed I could proceed with dictionary comprehension like this:

output=[{at(el)['feature']:at(el)['value'] for el in input[name]['elements']} for name in lnames]

But I got the error NameError: global name 'at' is not defined. When I use list comprehension instead of dictionary comprehension, it works , but:

1 - I still wonder what this name error means since it is linked to dictionary comprehension. 2 - this is not the expected result.

The edit is not so complex compared to the initial answer, but points 1/ really annoys me!

Answer 1

A comprehensive solution with existence checks

output = [
    {
        input[element]['feature'] : input[element]['value']
        for element in input.get(name,{}).get('elements',[])
    }
    for name in lnames
]

Answer 2

You were almost there. It should be:

output=[{input[el]['feature']:input[el]['value']
  for el in input[name]['elements']} for name in lnames]

Answer 3

It sometimes helps to write out regular loops, then convert to a list comprehension. After doing so, I got the following code:

output = [{input[element]['feature']: input[element]['value'] 
           for element in input[name]['elements']}
          for name in lnames]