Maps loose county names to mapit area ids

mapit_mapper.py

import json

def mappit_county_matcher(counties_json):

    source_counties = json.loads(counties_json)

    counties_map = {}
    for code in source_counties.values():
        name,type_name,type_id,mapit_id =  code[u'name'],code[u'type_name'],code[u'type'],code[u'id']
        counties_map[name] = {
            'name':name,
            'type_name':type_name,
            'type_id':type_id,
            'mapit_id':mapit_id,
            }
    
    custom_mappings = {
        "Bradford Metropolitan District Council":"Bradford City Council",
        "City of London":"City of London Corporation",                                
        "Royal Borough of Kensington and Chelsea":"Kensington and Chelsea Borough Council",         
        "Kingston-upon-Hull City Council":"Hull City Council",      
        "Oadby & Wigston District Council":"Oadby and Wigston Borough Council",              
        "South Buckinghamshire District Council":"South Bucks District Council",           
        "St Albans District Council":"St Albans City Council",       
        "Wakefield Metropolitan District Council":"Wakefield City Council",          
        "London Borough of Westminster":"Westminster City Council",                    
        "Royal Borough of Windsor and Maidenhead":"Windsor and Maidenhead Council",          
        "Edinburgh City Council":"City of Edinburgh Council",                           
        "Middlesbrough Borough Council":"Middlesbrough Council",                    
        "Rhondda Cynon Taff County Borough Council":"Rhondda Cynon Taf Council",        
        "Newcastle-Under-Lyme District Council":"Newcastle-under-Lyme Borough Council",            
        "Northumberland Council (Unitary)":"Northumberland County Council",                 
        "Bedford Council (Unitary)":"Bedford Borough Council",                         
    }
    
    def map(local_county):
        if counties_map.has_key(local_county):
            return ("exact_match",counties_map[local_county])
        
        if custom_mappings.has_key(local_county):
            local_county = custom_mappings[local_county]
        
        options = []
        if local_county.startswith('London Borough of '):
            options.append(local_county.replace('London Borough of ','') + " Borough Council")
            options.append((local_county.replace('London Borough of ','') + " Borough Council").replace('&','and'))
        if local_county.startswith('Royal Borough of '):
            options.append(local_county.replace('London Borough of ','') + " Borough Council")
        options.append(local_county.replace(" Metropolitan",''))
        options.append(local_county.replace(" Metropolitan District",''))
        options.append(local_county.replace(" County Borough",""))
        options.append(local_county.replace(" District",""))
        options.append(local_county.replace(" County",""))
        options.append(local_county.replace(" City",""))
        options.append(local_county.replace("City and Borough ",""))
        options.append(local_county.replace(" (Unitary)",""))
        options.append(local_county.replace(" & "," and "))
        options.append(local_county.replace("-"," "))
        options.append(local_county.replace(" Council"," Borough Council"))
        options.append(local_county.replace(" Council"," District Council"))
        options.append(local_county.replace(" Council"," County Council"))
        options.append(local_county.replace(" Council"," City Council"))
    
        options.append(local_county+" Council")
    
        if local_county.startswith('Borough of '):
            options.append(local_county.replace('Borough of ','')+" Borough Council")
            
        for option in options:
            if counties_map.has_key(option):
                return("fuzzy_match",counties_map[option])
                break
        else:
            return None
    
    return map
    
# councils data from: http://mapit.mysociety.org/areas/CTY,DIS,LBO,LGD,UTA,MTD,COI
mapper = mappit_county_matcher(open("/Users/ben/Desktop/counties.json").read())

# example

councils = ["Borough of Tower Hamlets", "Trafford Metropolitan Borough Council", "Tunbridge Wells Borough Council", "Uttlesford District Council", "Vale of White Horse District Council"]

print [(c,mapper(c)) for c in councils]