SI4T & Elasticsearch indexing Spanish text

Question

Not strictly a Tridion problem I know, but I'm reaching out to those who might have experience with encoding problems and SI4T.

https://stackoverflow.com/questions/27545834/elasticsearch-indexing-spanish-text-odd-results

Basically my Spanish text is appearing in Elasticsearch as double-encoded - I'm using the JEST library to index my documents and every log I've looked at my text looks ok.

The impact is I have to convert the UTF-8 text before rendering it back out. I just can't track down where its being double encoded.

Answer 1

You mention in your comment on stackoverflow that the JEST client log shows everything correctly. This tells us that the sending side has it right before firing off the request with the data. So this leaves the http request itself (perhaps the client has an encoding setting?) and the encoding config on the receiving side.

Take a snoop at the Request data being sent across the network via a tool like Fiddler and see if it is double-encoded. If it is, then it's the sending side that is packaging the data into the request that's doing it. Otherwise, there has to be some encoding mismatch on the ElasticSearch side. So look at the search configs there and also the settings of the web server that hosts the engine.