How to store business ID in elasticsearch? - twitter

Question

I'm trying to store tweets in some Elasticsearch index using Spring Data Elasticsearch (for tweet requesting , I'm using twitter4j).
I have followed some basic example and I'm using this basic annotated POJO (metadatas with complex type have been removed) :
#Document(indexName = "twitter", type = "tweet")
public class StorableTweet {
#Id
private long id;
private String createdAt;
private String text;
private String source;
private boolean isTruncated;
private long inReplyToStatusId;
private long inReplyToUserId;
private boolean isFavorited;
private boolean isRetweeted;
private int favoriteCount;
private String inReplyToScreenName;
private String userScreenName = null;
// Getters/setters removed
}
To store a tweet using this model, I use :
public interface TweetRepository extends ElasticsearchRepository<StorableTweet, Long> {
}
and in my storing service :
tweetRepository.save(storableTweet);
It works fine, but my tweet Id is stored in "_id" (why not) and some other number coming from nowhere is stored in "id" (why....?) :
{
"_index": "twitter",
"_type": "tweet",
**"_id": "655008947099840512"**, <-- this is the real tweet id
"_version": 1,
"found": true,
"_source":
{
**"id": 655008947099840500**, <-- this number comes from nowhere
"createdAt": "Fri Oct 16 15:14:37 CEST 2015",
"text": "tweet text(...)",
"source": "Twitter for iPhone",
"inReplyToStatusId": -1,
"inReplyToUserId": -1,
"favoriteCount": 0,
"inReplyToScreenName": null,
"user": "971jml",
"favorited": false,
"retweeted": false,
"truncated": false
}
}
What I would like is either my tweet id stored in "_id" (and no "id" field), either my tweet id stored in "id" an having a generated number in "_id", and get rid of this random useless number in "id".
EDIT
mapping :
{
"twitter":
{
"mappings":
{
"tweet":
{
"properties":
{
"createdAt":
{
"type": "string"
},
"favoriteCount":
{
"type": "long"
},
"favorited":
{
"type": "boolean"
},
"inReplyToScreenName":
{
"type": "string"
},
"inReplyToStatusId":
{
"type": "long"
},
"inReplyToUserId":
{
"type": "long"
},
"retweeted":
{
"type": "boolean"
},
"source":
{
"type": "string"
},
"text":
{
"type": "string"
},
"truncated":
{
"type": "boolean"
},
"tweetId":
{
"type": "long"
},
"user":
{
"type": "string"
}
}
}
}
}
}
EDIT 2 : It looks like the problem is not about #Id annotation but about "long" type. Some other longs (not all) are transformed (a few units more or less) when stored into elasticsearch via Spring Data Elasticsearch.