What is the most efficient way to filter documents with a similar structure to the one below based on the 'expanded_url' key's value. There are two more general questions here: * How do you filter documents based on a nested key(s)? * How do you filter documents based on the value of a key that resolves to an array of objects?

gistfile1.js

{
  "created_at":	"Tue, 01 Jan 2013 12:55:09 +0000",
	"entities":	{
		"hashtags":	[{
				"text":	"Clojure",
				"indices":	[39, 47]
			}],
		"urls":	[{
				"url":	"http://t.co/ythjuG5U",
				"expanded_url":	"http://benchmarksgame.alioth.debian.org/u32/benchmark.php?test=all&lang=clojure&lang2=java",
				"display_url":	"benchmarksgame.alioth.debian.org/u32/benchmark.…",
				"indices":	[118, 138]
			}],
		"user_mentions":	[]
	},
	"from_user":	"foo",
	"from_user_id":	000000000,
	"from_user_id_str":	"000000000",
	"from_user_name":	"Foo Bar",
	"geo":	null,
	"id":	0000000000000000000,
	"id_str":	"0000000000000000000",
	"iso_language_code":	"en",
	"metadata":	{
		"result_type":	"recent"
	},
	"profile_image_url":	"http://example.com/profile_images/000000000/avatar_normal.png",
	"profile_image_url_https":	"https://si0.twimg.com/profile_images/000000000/avatar_normal.png",
	"source":	"<a href="http://twitter.com/">web</a>",
	"text":	"New \"Programming Clojure\" still claims #Clojure is at \"The speed of hand-written Java code\" when it clearly isn't :-( http://t.co/ythjuG5U",
	"to_user":	null,
	"to_user_id":	0,
	"to_user_id_str":	"0",
	"to_user_name":	null
}

Here's the error I get when running a naiive filter in the ruby console:

ruby :132 >   r.expr([286108290734252030]).map {|id| r.table(table).get(id)}.filter {|t| t[:entities][:urls].contains('expanded_url')}.run
RuntimeError: RQL: Data: 
[{
        "url":  "https://t.co/QC0jAzFB",
        "expanded_url": "https://groups.google.com/forum/?fromgroups=#!topic/vimclojure/B-UU8qctd5A",
        "display_url":  "groups.google.com/forum/?fromgro…",
        "indices":  [72, 93]
    }]
must be an object
Line: (irb):132:in `evaluate'
Query: [286108290734252030].map("_var_1048", getbykey([:default_db, "tweets"], :id, _var_1048)).filter("_var_1049", _var_1049[:entities][:urls].contains("expanded_url"))
                                                                                                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^

The only thing I can think of now is using a concatMap and checking the length.
It's definitively not optimal, but we don't have a lot of method to help walking through an array now.
I'll open an issue for that.

r.expr(['id1', 'id2']).map( function(id) { 
     return r.db('db_name').table('table_name').get(id) 
}).filter( function(valid_doc) {
     valid_doc('urls').concatMap(
         function(url) {
             r.branch(r.expr({'www.valid_url.com': true, 'www.other_url.com': true}).contain(url)
            , [true]
            , []
      )}).count().gt(0)
}).run()

Discussion on this is here: rethinkdb/rethinkdb#211