Fulltext-Search on large context

Question

Hi there,
I'm trying to implement a product search in AIF based on a key/value-table (material no / material text) that comes with the context. Ideally the search includes fulltext search on material number and text, but a simple fulltext search on the material number would be enough as a first step.
Unfortunateley the customer is not able to cut down on the number of materials which comes with the context (~ 100,000 materials).
Up to ~ 30,000 materials the search works without problems. Using more than 35,000 - 40,000 the cpu usage increases to 100%, the search never ends and the cpu usage stays on 100% until i kill the adobe reader by task manager. Looks like a crash inside adobe reader. This behaviour is reproducible.
I have also tried to use tables of tables (1 table including 10 tables, 10000 items each) to ensure there's not a problem with out-of-range-indices. I used nodelists (xfa.resolve.nodes()...) as well as direct access on the context (xfa.resolvenode("...ABC.["index"]")...). The problem stays the same.
Can anybody tell me how to implement a full text search on about 100,000 unsorted and unstructured items with javascript in AIF?
Thanks you,
Eike

Answer 1

Maybe an interesting update to this issue: I started to measure the time spans adobe reader needed to search over a certain constant amount of items. It turned out that these times spans get longer and longer with increasing item index.
5000 items needed about 2 seconds, the next 5000 4 seconds, the next 5000 6 seconds and so on. Items 50000 - 55000 needed about 40 seconds to be searched. The search on items with an index > 60000 doesn't return at all...
Regards, Eike