FR: guess page sides when title page but not page sides marked

[mangtronix]

When the title page is specified but page sides are not specified we could assume that the title page is on the dominant side (right for LTR books) and infer the sided-ness of the other pages.

[mangtronix (mang)]

We currently assume that the first page is right-hand-side and that the page stack (accessible pages) is contiguous. Munging with the page sides will require us to be able to display synthetic "empty" pages.

e.g. inferring page sides based off the title page could cause the first page image to be LHS or make a gap in the pages to display in 2up mode. We want to support this but the underlying part isn't implemented yet.

[mangtronix (mang)]

Bonnie can you do a check on the google + scandata books and an assortment of random books and see that the pages show up on the correct sides? For LTR books the even numbered pages should show up on the left.

[Bonnie Real (bonnie-archive)]

Scooted through a couple dozen google + scandata and can confirm evensies are lefties.

One exception here, which mang notes is marked incorrectly as it has no title page:
http://www-mang.us.archive.org/GnuBook/?id=americanandengl04michgoog#9

I also looked at some random books and the structure holds up there, too.

[mangtronix (mang)]

Fix released. The page side is now inferred for the *goog books based off the title page. If the marked page is not RHS or the image stack contains sequences other than simple page pairs (e.g. an extra tissue page in between pages) the left/right pairs might get off.

[mangtronix (mang)]

Here's the checkin log:

/var/cvs/petabox/www/datanode/GnuBook/GnuBookJSIA.php,v <-- datanode/GnuBook/GnuBookJSIA.php
new revision: 1.16; previous revision: 1.15
/var/cvs/petabox/www/petabox/GnuBook/GnuBook.js,v <-- petabox/GnuBook/GnuBook.js
new revision: 1.73; previous revision: 1.72

[Hank Bromley (hank-archive)]

Sorry to be so late with this question, but are we sure we want to limit this to google books? We're now making scandata for every book that doesn't already have it - besides google books, that would include opensource contributions and bulk ingests (like the current ingests of the Rosetta Project materials and the Digital Library of India scans). Is there a reason not to infer handSide's based on location of title page for those books, too?

(The key restriction appears to be this test in GnuBookJSIA.php:

preg_match('/goog$/', $id)

.)

[mangtronix (mang)]

Reopening. Now that other books are in a similar situation we should enable it for all books.