/*
  This is a JavaScript script for mongodb.
  It constructs a new version of the Artworks collection of the
  moma database from the one created by mongoimport and the
  Artworks.json file downloaded from the Museum of Modern Art:

  https://github.com/MuseumofModernArt/collection

  It makes the following changes:

  The auto-generated _id field is replaced by the ObjectID field

  The parallel arrays representing Artist, ConstituentID,
  etc. are replaced by an array of embedded documents
*/
conn = new Mongo("mongodb.cs.oberlin.edu")
db1 = conn.getDB("moma1")
db2 = conn.getDB("moma2")
db2.Artworks.drop()
db2.createCollection("Artworks")
cursor = db1.Artworks.find()
//cursor = db1.Artworks.find({ $where : function(){ return this.Artist.length>2 } }).limit(20)  // I used this for testing
var n = 0
var docs = []
while(cursor.hasNext()){
    doc = cursor.next()
    doc._id = doc.ObjectID

    artistArray = []
    for(i=0; i<doc.Artist.length; i++) {
	artistArray.push( doc.Artist[i].ConstituentID )
    }
    doc.Artist = artistArray
    delete doc.ConstituentID
    delete doc.ArtistBio
    delete doc.Nationality
    delete doc.BeginDate
    delete doc.EndDate
    delete doc.Gender
    docs.push(doc)
}
print("calling insertMany")
db2.Artworks.insertMany(docs)