name-match.py

"""

We are trying to determine which offenses can be matched to a candidate. Candidates names' often do not match exactly to criminal records, but still represent the candidate. We are given a JSON file of possible criminal offenses and a candidate's aliases. We use a point system to judge how well a record matches to a candidate.

The point system starts out at 100 (100% accurate) being an exact match. We can do 4 edit types to a record name to get to an exact match of one of the candidate's aliases.

Edits:
  A middle name swapped with a first name subtracts 7 points
  A first name nickname swap subtracts 5 points
  A middle name nickname swap subtracts 3 points


Nicknames comes from nick_names.json. If a name is contained in the set of nicknames, the other names can be considered nicknames.
The candidate's aliases and offenses come from input.json.

Goal 1:
With one edit per record name list the offenses that have a score above 94.




#####################################################################################

todo: is it not a flaw to assume 100% score by default?

it might be better to first break down the candidate aliases into first, middle, last.
then build a list of alternate candidate names using the edits, and run those alt  candidate f/m/l against the edits and match exactly against the records.

building an alternative match for each record, and then checking against the aliases may not be ideal.

#####################################################################################

"""




from copy import copy
import pprint
pp = pprint.PrettyPrinter(indent=4)


def middle_name_swap(record):
    names = []
    #A middle name swapped with a first name subtracts 7 points
    if record['_middle']:
        names.append('%s %s %s' % (record['_middle'], record['_first'], record['_last']))
        return (-7, names)
    else:
        return(0, names)
    

def first_nick_swap(record):
    #A first name nickname swap subtracts 5 points
    names = []
    nicks = find_nick_names(record['_first'])
    for nick in nicks:
        names.append('%s %s %s' % (nick, record['_middle'], record['_last']))
    
    return (-5, names)


def middle_nick_swap(record):
    #A middle name nickname swap subtracts 3 points
    names = []
    nicks = find_nick_names(record['_middle'])
    for nick in nicks:
        names.append('%s %s %s' % (record['_first'], nick, record['_last']))
    
    return (-3, names)


def maiden_last_swap(record):
    #todo
    #A maiden last name swap subtracts 5 points
    return 0, []


def find_nick_names(search):
    if search:
        for nicks in nick_names:
            if search.upper() in nicks:
                return nicks
    return []


nick_names = [
  ["EDWARD", "EDDIE", "ED", "EDDY"],
  ["AL", "ALPHONSE", "ALI"],
  ["BONNIE", "BON", "BONNY", "BONAVENTURE"]
]


data = {
  "candidate": {
    "aliases": ["Bonnie Elizabeth Parker", "Bonnie Elizabeth Clyde", "Bonnie Parker"]
  },
  "records": [
    {
      "county": "San Francisco",
      "offense": "speeding 2",
      "name": "Elizabeth Bonnie Parker",
      "severity": "misdemeanor"
    },
    {
      "county": "San Francisco",
      "offense": "speeding",
      "name": "Bonnie Elizabeth Parker",
      "severity": "misdemeanor"
    },
    {
      "county": "Nevada",
      "offense": "robbery",
      "name": "Bonny Elizabeth Parker",
      "severity": "felony"
    },
    {
      "county": "Los Angeles",
      "offense": "murder",
      "name": "Elizabeth James",
      "severity": "felony"
    },
    {
      "county": "Marin",
      "offense": "assault",
      "name": "Elizabeth Parker",
      "severity": "felony"
    },
    {
      "county": "Lassen",
      "offense": "assault 2",
      "name": "Bonny Elizabeth Clyde",
      "severity": "misdemeanor"
    }
  ]
}


def filter_records():
    #iterate the records, normalize names, run edits, and tally up the scores 
    
    data['candidate']['aliases'] = [alias.upper() for alias in data['candidate']['aliases']]
    
    filtered_records = []
    for record in data['records']:
        record['score'] = 100
        record['name'] = record['name'].upper()
        record['_name_parts'] = record['name'].split(' ')

        record['_first'] = None
        record['_middle'] = None
        record['_last'] = None

        record['_aliases'] = []

        if len(record['_name_parts']) == 3:
            record['_first'] = record['_name_parts'][0]
            record['_middle'] = record['_name_parts'][1]
            record['_last'] = record['_name_parts'][2]

        if len(record['_name_parts']) == 2:
            record['_first'] = record['_name_parts'][0]
            record['_last'] = record['_name_parts'][1]

        edits = [
            middle_name_swap(record),
            first_nick_swap(record),
            middle_nick_swap(record),
            #maiden_last_swap(record),
        ]


        for score, aliases in edits:
            for alias in aliases:
                record['_aliases'].append(alias)
                if alias in data['candidate']['aliases']:
                    record['score'] += score
                    
        if record['score'] > 94:
            filtered_records.append(record)
            
    return filtered_records


pp.pprint(filter_records())