tag concordance

gistfile1.txt

import json

f = open('c:\\users\\jon\\all_annos.json','r')
s = f.read().decode('utf-8')
j = json.loads(s)

s = ''

all_tags = {}

for row in j:
    try:
        user = row['user'].replace('acct:','')
        if row.has_key('tags'):
            tags = row['tags']
            for tag in tags:
                if all_tags.has_key(tag):
                    all_tags[tag] += 1
                else:
                    all_tags[tag] = 1
        uri = row['uri']
        line = '%s\t%s\t%s\n' % (user, uri, ','.join(tags))
        s += line
    except:
        print row['uri']

f = open('c:\\users\\jon\\all_tags.txt','w')
f.write(s.encode('utf-8'))
f.close()

s = ''

keys = all_tags.keys()
keys.sort()
for key in keys:
    s += '%02d: %s\n' % (all_tags[key], key)

f = open('c:\\users\\jon\\tag_concordance.txt','w')
f.write(s.encode('utf-8'))
f.close()