If you look at Google’s service on the “individual search” level, it’s only a matter of access control: no security through obscurity, users just have limited access to Google’s database.

If you look at the service from a “totality of searches” point of view, it indeed resembles a DRM system (but only vaguely): there are restrictions on usage (that is, you can only view small snippets per time unit), but the full text can be extracted in an ‘analog hole’ like way (i.e. do an automated search for subsequent snippets, OCR, paste together). However, is this a real DRM system? Is cutting a file in smaller parts the same as scrambling that file, so that it can only be opened with the appropriate application?
In other words, I don’t believe that the puzzling program you describe would be illegal (under anti-circumvention provisions), as there wouldn’t be any circumvention activity going on – it’s just puzzling.

Note: in case such a puzzling program is made, it is very probable that Google will block it from abusing its index. Furthermore, I expect that the Google Book Search API (if we ever see one) will have security hooks that prevents users from extracting full book contents.

As of today, I am not aware of such software existing, but it should be not too difficult to build. Since all that stands between that happening and the protection of copyrights is Google’s DRM effort, and DRM efforts can as you know never be effective, we might find some such program available in the near future. It might be interesting to discuss if such a program would be illegal under anti-circumvention laws.

But anyway, you and other Google supporters are wrong when you say that Google is serving “only snippets”. That is only correct for the individual search. To the totality of searches, Google is serving all the parts of the puzzle, and it only takes an effort to collect them and assemble them to, indeed, download whole books.