search for match between objects

instructions.md

Input Data Given json contains 20K articles. Each article contains set of attributes with one set value.

Recommendation Engine Requirements Calculate similarity of articles identified by sku based on their attributes values. Attributes with name higher in alphabet (a is higher than z) is weighted with heavier weight.

Absolute weighting values are up to engineer. Relative order as described above should be maintained.

Recommendation request example sku-123 > ENGINE > 10 most similar skus based on criteria described above with their corresponding weights.

Language requirements Not specified

OOP requirements Clean structured reusable code

**Expected delivery format tgz file containing solution with simple instructions how to run data import and how to execute recommendation request

#################################### Solution :

python match.py -i test-data.json -m sku-123

match.py

#!/usr/bin/env python
from __future__ import division
import json
from pprint import pprint
import sys, getopt
import re


class Match(object):
    def __init__(self, match, ratio , matchIn):
        self.match = match
        self.ratio = ratio
        self.matchIn = matchIn

    def __repr__(self):
        return '{}: {} has weight {} and matched in {}\n'.format(self.__class__.__name__,
                                  self.match,
                                  self.ratio,
                                  self.matchIn)

    def __cmp__(self, other):
        return self.ratio.__gt__(other.ratio)

# Calculate similarity as weight between @param match and @param against based on their attributes values.
# weight is 0 if no match found
# weight value is increased when match found
# big weight value mean the similarity is very close
def sku_match(match , against):
    weight = 0
    matchIn = []
    for idx, att in enumerate(against):
        a_att = match[att]
        b_att = against[att]
        if( a_att is not None and b_att == a_att):
            matchIn.append(att)
            # weight is increased
            # weight of attribute determined by integer value {char} at att-{char}
            _weight = ord(att.split('-')[1])
            # substract by 96 'a ascii -1' and to make a is more weight than b 1/SUBSTRACT
            _weight = 1 / (_weight - 96)
            weight = weight + 1 + _weight

    return {'weight' : weight , 'matchIn' : matchIn}

## brute force through skus and compraing search value with skus and assigning match ratio value
## extract top ten ratio values
def similar(file , match):
    print 'search for match with ', match , ' in ' , file
    matched = [];
    with open(file) as data_file:
        skus = json.load(data_file)

    _a = skus[match]
    for sku in skus:
        if(match == sku) : continue
        _mat = sku_match(_a , skus[sku])
        if(_mat['weight'] > 0) :
            matched.append(Match(sku, _mat['weight'] , _mat['matchIn']))

    if (len(matched) > 0):
        # extract top ten elements based on ratio
        topTen = sorted(matched, key = lambda x : x.ratio, reverse = True)[:10]
        print 'top matched docs :'
        for item in topTen :
            print item
    else: print 'no match found'

if __name__ == '__main__':
    argv = sys.argv[1:]
    try:
      opts, args = getopt.getopt(argv,"hi:m:t:",["ifile=","match="])
    except getopt.GetoptError:
      print 'match.py -i <inputfile.json> -m <match>'
      sys.exit(2)
    for opt, arg in opts:
      if opt == '-h':
         print 'match.py -i <inputfile.json> -m <match>'
         sys.exit()
      elif opt in ("-i", "--ifile"):
         inputfile = arg
      elif opt in ("-m", "--match"):
         match = arg
    similar(inputfile , match );