CODEWITHSUNDEEP
pythonelasticsearch
slim
slim

Reputation: 2583

How can I convert the 64 bit timestamp returned from elasticsearch into a datetime in python?

My date histogram query returns something like this:

"facets": {
  "hist": {
     "_type": "date_histogram",
     "entries": [
        {
           "time": 1385856000000,
           "count": 1884
        },
        {
           "time": 1385942400000,
           "count": 1902
        },

How can I take that time value and get a date string: '2014-02-16'?

Here's what I have so far:

def getFiletime(dt):
    microseconds = int(dt) / 10
    seconds, microseconds = divmod(microseconds, 1000000)
    days, seconds = divmod(seconds, 86400)
    return datetime.datetime(1601, 1, 1) + datetime.timedelta(days, seconds, microseconds)

I get back: 

1601-01-02 14:37:23.520000
1601-01-02 14:37:32.160000
1601-01-02 14:37:40.800000
1601-01-02 14:37:49.440000
1601-01-02 14:37:58.080000
1601-01-02 14:38:06.720000
1601-01-02 14:38:15.360000

Any ideas? I just copied the getFiletime function from the internet. I don't think it's meant for what I'm doing, but I know it's on the right track. I tried putting the "format" specifier in my elastic search query but that doesn't return the timestamp as a string like the documentation states. Any help would be appreciated and thank you for reading!

Upvotes: 3

Views: 4661

Answers (3)

AALAP JETHWA
AALAP JETHWA

Reputation: 160

A simple function:

from datetime import datetime
import arrow

def timestamp(timestamp):
    date = datetime.fromtimestamp(timestamp /  1_000_000.0)
    return formattdatetime(date)

def formattdatetime(date):
    date = arrow.get(date, "YYYY-MM-DD")
    return arrow.get(date)

Upvotes: 2

egpbos
egpbos

Reputation: 502

These aren't 64bit timestamps, just regular (unix epoch) ones with millisecond resolution:

 import datetime
 datetime.datetime.utcfromtimestamp(1385942400000 / 1000.).strftime('%Y-%m-%d')

gives 2013-12-02

(thanks to @knutwalker's comment above)

Upvotes: 1

jxstanford
jxstanford

Reputation: 3397

I had the same challenge and created this method to solve the same problem. 

def to_es_date(d):
    s = d.strftime('%Y-%m-%dT%H:%M:%S.')
    s += '%03d' % int(round(d.microsecond / 1000.0))
    s += d.strftime('%z')
    return s

It takes a datetime with or without timezone, and returns a string that is recognized by ElasticSearch (v1.1.1 for sure). I'm not sure if the padding is required or not, but it seemed prudent.

In [5]: from datetime import datetime

In [6]: import pytz

In [7]: %paste
def to_es_date(d):
        s = d.strftime('%Y-%m-%dT%H:%M:%S.')
        s += '%03d' % int(round(d.microsecond / 1000.0))
        s += d.strftime('%z')
        return s

In [8]: to_es_date(datetime.now())
Out[8]: '2014-04-30T03:56:02.893'

In [9]: to_es_date(datetime.now(tz=pytz.utc))
Out[9]: '2014-04-30T03:56:20.134+0000'

Upvotes: 1

Related Questions