Script for comparing the different ways of "boosting" queries in Solr

compare_solr_boosts.py

#!/bin/sh/env python
#
# Simple script for comparing the various ways of "boosting" queries in Solr,
# combined with the three main query parsers (lucene, dismax, and edismax).
# The idea is to generate md5sums of the search results, so I can quickly
# figure out which boost methods work the same as the other ones.
#

import re, urllib2, urllib, md5, json

solr = "http://localhost:8081"

parsers = ['lucene','dismax','edismax']
# more parsers: ['lucene','dismax','edismax', 'boost', 'field', 'func', 'frange']

boost = 'log(relevancy_score)'
query = 'diabetes'

base_params = {'echoParams':'none','rows':1000, 'omitHeader' : 'true', \
'qf' : 'text', 'debugQuery' : 'on', 'wt' : 'json'}

def boost_query(boost_method, query):
  if (boost_method == 0): # no boost
    return {'q':query}
  if (boost_method == 1):
    return {'q':'{!boost b=%s}%s' % (boost, query)}
  elif (boost_method == 2):
    return {'q':'{!boost b=$myboost v=$myquery}', 'myboost':boost, 'myquery':query}
  elif (boost_method == 3):
    return {'q':query,'bq':boost}
  elif (boost_method == 4):
    return {'q':query,'bf':boost}
  else: # 5
    return {'q':query,'boost':boost}

for parser in parsers:
  print parser
  for boost_method in range(6):
    params = {'defType' : parser}
    params.update(base_params)
    boost_params = boost_query(boost_method, query)
    params.update(boost_params)
    encoded_params = urllib.urlencode(params)
    url = solr + '/select?' + encoded_params
    #print url
    response = json.loads(urllib2.urlopen(url).read())
    docs = response['response']
    parsed_query = response['debug']['parsedquery']
    md5sum = md5.new(str(docs)).hexdigest()
    print "    %s: %s (%s)" % (boost_method,md5sum,boost_params)
    print "        " + str(parsed_query)