[Argh]

You can crawl the item pages, store them in a large (huge) json file, host the file on aws/heroku, and make the search requests to that server as a hood rat/duct taped api.