Yeah so you dont have to write all the code to get the umbraco content into elastic. You also dont have to write the client to query it either.
Basically we have examine which is the search and indexing in umbraco. It wraps around lucene.net. Examine is extensible.
The package link i sent you is an examine provider for elastic search so instead of the engine under being lucene its now elastic (although elastic is powered also by lucene). There is also an azure search provider examinex although that one is paid.
The elastic one only does content and media however for the media it only does stub information like filename size extension not the actual content of the media. So in theory you can use examine events and test is current item being indexed media item, if it is then test is it pdf and if it is extract the pdf lib of your choice then inject the extracted content in. That way you can get actual pdf content.
I did create a composition for examine pdf indexer which uses textsharp and i swapped out the textsharp engine with apache tika. Apache tika can extract most file formats, its a bit on the heavy side as its written in java and uses IKVM but it works really well. See
https://www.nuget.org/packages/TikaOnDotNet/
