import time
from datetime import datetime
import urllib2

def get_docs(data_type, today):
    # curl '192.168.3.153:9200/_cat/indices/metadata-dis-2017.05.16-*-93001?v'
    url = 'http://192.168.3.153:9200/_cat/indices/%s-dis-%s-*' % (data_type, today)
    out = urllib2.urlopen(urllib2.Request(url))
    data = out.read()
    docs_cnt = 0
    for line in data.split('\n'):
        if line:
            docs_cnt += int(line.split()[5])
    return docs_cnt


def main():
    today = datetime.now().strftime("%Y.%m.%d")
    init_data = {"event": 0, "metadata": 0}
    docs_cnt, docs_cnt2, total_speed = dict(init_data), dict(init_data), dict(init_data)
    data_types = docs_cnt.keys()
    cnt = 0
    sleepy_time = 20
    for data_type in data_types:
        try:
            got_doc_cnt = get_docs(data_type, today)
        except:
            got_doc_cnt = docs_cnt[data_type]
        docs_cnt[data_type] = got_doc_cnt
    while True:
        time.sleep(sleepy_time)
        for data_type in data_types:
            try:
                got_doc_cnt = get_docs(data_type, today)
            except:
                got_doc_cnt = docs_cnt[data_type]
            docs_cnt2[data_type] = got_doc_cnt
        cnt += 1
        for data_type in data_types:
            speed = (docs_cnt2[data_type]-docs_cnt[data_type])/(sleepy_time+0.0)
            total_speed[data_type] += speed
            print "cnt=%d %s speed = %.1f total_speed=%.1f" % (cnt, data_type, speed, total_speed[data_type]/cnt)
        docs_cnt = dict(docs_cnt2)
        
if __name__ == "__main__":
    main()