| 1 | #!/usr/bin/env python |
|---|
| 2 | |
|---|
| 3 | import urllib, urllib2, cookielib |
|---|
| 4 | import simplejson |
|---|
| 5 | import Cookie |
|---|
| 6 | |
|---|
| 7 | username = '<username>' |
|---|
| 8 | password = '<password>' |
|---|
| 9 | corp = 'bnc' |
|---|
| 10 | |
|---|
| 11 | root_url = 'http://beta.sketchengine.co.uk' |
|---|
| 12 | |
|---|
| 13 | base_url = '%s/auth/preloaded_corpus/%s/ske/' % (root_url, corp) |
|---|
| 14 | method = 'view' |
|---|
| 15 | |
|---|
| 16 | # creating query string |
|---|
| 17 | attrs = dict(q='', pagesize='1', format='json') |
|---|
| 18 | # query_list can be read from a file, ... |
|---|
| 19 | query_list = ['[lemma="test"]', |
|---|
| 20 | '[lemma="drug"][lemma="test"]', |
|---|
| 21 | '[lemma="blood"][lemma="test"]', |
|---|
| 22 | '[lemma="test"][lemma="result"]' |
|---|
| 23 | ] |
|---|
| 24 | |
|---|
| 25 | # authentication |
|---|
| 26 | cj = cookielib.CookieJar() |
|---|
| 27 | opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj)) |
|---|
| 28 | login_data = urllib.urlencode({ 'username' : username, |
|---|
| 29 | 'password' : password, |
|---|
| 30 | 'submit' : 'ok', |
|---|
| 31 | }) |
|---|
| 32 | data = opener.open('%s/login/' % root_url) |
|---|
| 33 | data = opener.open('%s/login/' % root_url, login_data) |
|---|
| 34 | |
|---|
| 35 | for query in query_list: |
|---|
| 36 | attrs['q'] = ['q' + query, 'r100'] |
|---|
| 37 | encoded_attrs = urllib.quote(simplejson.JSONEncoder().encode(attrs)) |
|---|
| 38 | url = base_url + method + '?json=%s' % encoded_attrs |
|---|
| 39 | # json data receiving |
|---|
| 40 | file = opener.open(url) |
|---|
| 41 | data = file.read() |
|---|
| 42 | file.close() |
|---|
| 43 | |
|---|
| 44 | # now, in the 'data' variable, there is a json string that can be parsed |
|---|
| 45 | # for json syntax (e.g. by simplejson) |
|---|
| 46 | json_obj = simplejson.loads(data) |
|---|
| 47 | |
|---|
| 48 | print query + '\t' + str(json_obj.get('concsize', '0')) |
|---|
| 49 | |
|---|