示例#1
0
    if extension not in ('csv', 'html', 'xls', 'json', 'xlsx'):
        sys.stderr.write("invalid filename: %s filename\n" % filename)

    print extension

    try:
        open(filename,'w').write(ds.__getattribute__(extension))
        sys.stdout.write("saved %s \n" % filename)
    except Exception, e:
        sys.stderr.write(str(e))



if __name__ == '__main__':

    c = mc.open_connection('mongolab')


    sys.stderr.write('loading events..')
    a = list(c.find({}).limit(LIMIT))
    sys.stderr.write('loaded %d events\n' % len(a))

    unique_keys = set()

    for doc in a:
        for k in doc.keys():
            unique_keys.add(k)


    headers = list(map(str,list(unique_keys)))
    pprint.pprint(headers)
示例#2
0
import sys
from collections import defaultdict
import mongo_connect as mc




def key_count(collection):

    sys.stderr.write('loading collection...')
    a = list(collection.find({}))
    sys.stderr.write('loaded %d records\n' % len(a))
    sys.stderr.write('counting keys...')

    keycount = defaultdict(int)

    for i, doc in enumerate(a):
        for key in doc.keys():
            keycount[key] += 1
        if i % 1000 == 0:
            print i, doc

    return dict(keycount)

if __name__ == '__main__':
    c = mc.open_connection(account='mongolab')
    keycount = key_count(c)