I apologize for not directly answering your question or providing a solution
for your particular need but the title invites comment. (I suspect that the
solution would not be as simple as imagined, once you get rid of
non-relevant matches, insert proper pdf links, and translate the notebook
back into the original pdf document. Of course, the person who wrote the
original pdf could have provided links if he was in a position to do so and
cared more about his readers.)
Why don't people actually write smart textbooks with Mathematica?
One maxim would be: Don't jerk the reader around! Bring the material that a
reader needs to the point that he needs it. Once you move to another point
in the document you lose sight of the original material. Even if your
proposal succeeds it doesn't solve that problem.
Wouldn't it be nice if the reader became a reader/user and with the document
also had routines so he could start doing calculations and variations
immediately? Wouldn't it be nice if he could just recall an equation to his
place in the document? Or maybe the material could be displayed in a
separate window on the side of the notebook. Wouldn't it be nice if basic
axioms or theorems were available in a separate window where they could be
constantly referred to, or better yet used, by pasting them into the
document? Wouldn't it be nice if complex proofs or derivations could be
presented as a unity in one place, where one could flip back and forth
between various sections of the proof, or put several sections up
simultaneously?
Wouldn't it be nice if documents didn't have so many misprints, omissions,
or errors? Wouldn't it be nice if the material in the document had been
proofed by actual evaluation?
Suppose you could produce such a document and also publish a hard copy
version with the basic material without the nice usage features.
It's possible to do all of this now. It's called a Mathematica Application
and can be produced with Mathematica and Workbench. All one needs to read
and use the document is Mathematica. It's not that expensive. Such documents
are orders or magnitude better than conventional practice. It's the future.
The time and effort spent in trying to provide Mathematica solutions to
final non-Mathematica documents seems to me to be a poor use of time and
resources. They all fall short in some way or introduce extra problems.
David Park
djm...@comcast.net
http://home.comcast.net/~djmpark/index.html