ocd-mkwsi

#!/usr/bin/python

import urllib, urllib2, base64, json, sys, manatee

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

corp = sys.argv[1]
wsstrip = int(sys.argv[2])
usr = sys.argv[3]
apikey = sys.argv[4]


attrs = dict(corpname=corp, format='json', clustercolls='1', api_key=apikey,
             minsim=0.05)

def unimap (relname, grfile):
    lastgr = None
    for line in grfile:
        if line.startswith("="):
            lastgr = line[1:-1].split("/")
            continue
        elif not line.startswith("*UNIMAP"):
            continue
        else:
            line = line.replace("*UNIMAP","").strip().split("/")
            try:
                i = line.index(relname)
            except ValueError:
                continue
            return lastgr[i]

gramrels = {"-n": 'modifier', "-v": 'object_of', "-j": 'n_modifies'}
try:
    c = manatee.Corpus(corp)
except manatee.CorpInfoNotFound:
    c = manatee.Corpus(corp.split("/")[-1])
grfile = open(c.get_conf("WSDEF"))
for g in gramrels:
    gramrels[g] = unimap(gramrels[g], grfile)

for h in sys.stdin:
    entry = json.loads(h)
    attrs['lemma'] = entry["hw"]
    lempos = attrs['lemma']
    if ' ' in lempos:
        print json.dumps(entry)
        continue
    if wsstrip:
        attrs['lpos'] = attrs['lemma'][-wsstrip:]
        attrs['lemma'] = attrs['lemma'][:-wsstrip]
        attrs['selgrlist'] = gramrels.get(attrs['lpos'], "")

    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)
    clusters = []
    if not "Gramrels" in data:
        print json.dumps(entry)
        continue
    for g in data["Gramrels"]:
        for w in g["Words"]:
            if not "lempos" in w:
                continue
            cluster = [{"lempos": w["lempos"], "conclink": w["conclink"], "relconclink": '[ws("%s","%s","%s")]' % (lempos, g["name"].replace('"',r'\"'), w["lempos"])}]
            for c in w.get("Clust",[]):
                cluster.append({"lempos": c["lempos"], "conclink": c["conclink"], "relconclink": '[ws("%s","%s","%s")]' % (lempos, g["name"].replace('"',r'\"'), c["lempos"])})
            clusters.append (cluster)
#    changed = True
#    while changed:
#        changed = False
#        for cindex in xrange(len(clusters)):
#            for i in xrange(cindex + 1, len(clusters)):
#                if (set([x["lempos"] for x in clusters[cindex]]) & set([x["lempos"] for x in clusters[i]])):
#                    clusters[cindex].extend(clusters[i])
#                    changed = True
#                    break
#            if changed:
#                del clusters[cindex]
#                break
    senses = []
    for c in clusters:
        senses.append({"relconclink" : "|".join ([x["relconclink"].encode('utf8') for x in c]),
                       "conclink": "|".join ([x["conclink"].encode('utf8') for x in c]),
                       "colls": [x["lempos"].encode('utf8') for x in c]})
    entry["senses"] = senses
    entry["colls_wsstrip"] = 2
    print json.dumps(entry)