Currently, link text is not associated with the page linked to, only with the page the link occurs on. This means search results for things like images are weaker than they should be.

One way to handle this is to create for each link processed a "micro document" which is stored in the inverted word document index. The main document and all of the inlink documents are grouped together using the group iterator of Bug 8.

Fixed with commit 7fb47f3