Question

I’ve got a Rails 3.2 app running on Ruby 1.9.3 that returns JSON data stored in a MongoDB database. The data seems to be stored correctly in mongo, e.g. (look at the name attribute):

{ "_id" : ObjectId("4f986cbe4c8086fdc9000002"), "created_at" : ISODate("2012-04-25T21:31:45.474Z"), "updated_at" : ISODate("2012-04-26T22:07:23.901Z"), "creator_id" : ObjectId("4f6b4d3c4c80864381000001"), "updater_id" : null, "name" : "Trädgår'n", "sort" : "tradgarn", "address" : "Nya Allén 11", "coordinates" : [ 11.9764791, 57.7045625 ], "phone" : "46031102080", "url" : "http://www.profilrestauranger.se/tradgarn/", "user_ids" : [ ] }

But when I issue a request that returns this record, I get something like this back (now look at the name attribute):

{"address":"Nya All\u00e9n 11","coordinates":[11.9764791,57.7045625],"created_at":"2012-04-25T23:31:45+02:00","id":"4f986cbe4c8086fdc9000002","name":"Tr\u00e4dg\u00e5r'n","phone":"46031102080","sort":"tradgarn","updated_at":"2012-04-27T00:07:23+02:00","url":"http://www.profilrestauranger.se/tradgarn/"}

The response headers for anyone interested:

HTTP/1.1 200 OK
Server: nginx/1.2.0
Date: Thu, 26 Apr 2012 22:41:13 GMT
Content-Type: application/json; charset=utf-8
Content-Length: 909
Connection: keep-alive
Status: 200 OK
Access-Control-Allow-Origin: *
Access-Control-Allow-Methods: POST, GET, OPTIONS
Access-Control-Max-Age: 1000
Access-Control-Allow-Headers: *,x-requested-with
X-UA-Compatible: IE=Edge
ETag: "d2a95f06bec10d8087c3188280292d3c"
Cache-Control: max-age=0, private, must-revalidate
X-Request-Id: fdd042568195df279e59affe45bdcd37
X-Runtime: 0.037134

I cannot seem to figure out why or where the encoding is getting messed up? What gives? Help? 🙂

Answer 1

The issue is indeed one of JSON UTF-8 encoding. The #to_json method is escaping unicode characters. This can be observed by something like:

user.to_json

#  => "{\"created_at\":\"2012-04-19T18:48:01Z\",\"email\":\"tr\\u00e4dg\\[e-mail]\",\"id\":10,\"updated_at\":\"2012-04-27T18:37:10Z\"}"

When parsed, however, this is converted back to how you would expect it. It is possible, however, to generate the JSON using JSON.generate, with which the #as_json method can be used, along with any options for the construction. This doesn’t escape the unicode. To do such:

JSON.generate(user.as_json)

#  => "{\"created_at\":\"2012-04-19T18:48:01Z\",\"email\":\"trädgå[e-mail]\",\"id\":10,\"updated_at\":\"2012-04-27T18:37:10Z\"}"