How can I create the tsvector data for large text files WITHOUT loading the text into a column of the DB?
I did it using 2 tables (one with the text inserted with a perl program) and copying the tsvectors and removing the table with the text, but it is not easy to add data this way.
The text files are "frozen" HTML documents so I don't need a trigger to update the tsvector, all I store is a URL of the text.
So what I need is a way to point to a file (maybe there is a special data type?) and create the tsvector from the document.