predictionio_import.py

# -.- coding: utf-8 -.-

import json
import urllib2
import predictionio

client = predictionio.Client(appkey="XXXXXXXXXXXXXXXXXXXXXXX")

total = 6028561
size = 1000
iter_ = 715

while iter_ * size < total:
        from_ = iter_ * size
        data = urllib2.urlopen("XXXXXXXXXXXXX/_search?size=%d&from=%d" % (size, from_))
        data = json.load(data)

        for video in data['hits']['hits']:
                video = video.get("_source")
                print u" ".join(["import", video.get("_id"), "tags", u",".join(video.get("g"))])

                if not video.get("_id") or not video.get("g"):
                        break

                try:
                        client.create_item(video.get("_id"), map(lambda g: g.encode("ascii", "ignore").replace("\t", "").replace(",", ""), video.get("g")))
                except predictionio.ItemNotCreatedError, e:
                        print "Error importing item"

        iter_ = iter_ + 1
        print "Batch %d completed" % iter_

requirements.txt

PredictionIO==0.7.0
argparse==1.2.1
wsgiref==0.1.2