ocd-mkdefs

#!/usr/bin/python

import urllib, urllib2, json, sys

base_url = 'https://api.sketchengine.co.uk/corpus/'
method = 'view'

corpora = sys.argv[1].split(',')
wsstrip = int(sys.argv[2])
usr = sys.argv[3]
apikey = sys.argv[4]

def build_lempos(s):
    s = s.split(" ")
    s = map(lambda x: x + wsstrip * '.', s)
    return "|".join (s)

def get_definitions(w, q):
    defs = []
    for corp in corpora:
        attrs = dict(corpname=corp, format='json', api_key=apikey, structs='g',
                     change_gdex=False, viewmode='sen',
                     q=['q'+q, 'p 0 0>0 1 [ws("%s", "definitions", ".*")]' % build_lempos(w),
                        ])
        encoded_attrs = urllib.quote(json.dumps(attrs))
        url = base_url + method + '?username=%s;json=%s' % (usr, encoded_attrs)
        request = urllib2.Request(url)
        file = urllib2.urlopen(request)
        data = file.read()
        file.close()
        data = json.loads(data)

        if 'error' in data and data['error'] == u'Empty result':
            continue
        for line in data['Lines']:
            d = ''
            d += ''.join([x['str'] for x in line['Left']])
            d += ''.join([x['str'] for x in line['Kwic']])
            d += ''.join([x['str'] for x in line['Right']])
            defs.append(d)
        if len(defs) > 9:
            break
    return defs

for ll in sys.stdin:
    entry = json.loads(ll)
    if "senses" in entry:
        for s in entry["senses"]:
            s["defs"] = get_definitions(entry["hw"], s["relconclink"])
    else:
        entry["defs"] = get_definitions(entry["hw"], entry["conclink"])
    print json.dumps(entry)