How does Google Play choose page number / how to influence it?

Question

Our book has quite simple page numbering: 1, 2, 3 and so on. (We don't use any special numbering for the first pages, e.g.: the book title is on the page 1, technical information about the book is on the page 4, table of contents starts on the page 5, the main content starts on the page 12. Maybe that's unusual, but the book is already printed, and we can't change it.)

We have the PDF and want to publish on Google Play Books. The problem is that Google Play assigns page numbers in a different way: i, ii, iii, iv, 1, 2, 3 and so on — such page numbers are displayed in the Google Play user interface controls: on the trackbar (bottom) and on the “Contents” pop-up (top-right). So, the actual numbering in the book (visible in the corners of pages and in the in-book table-of-contents) and numbering in the Google Play UI controls (trackbar and TOC pop-up) don't match.

I know PDF allows to specify \PageLabels. We successfully specified them in the PDF (other PDF viewers display them correctly), but Google Play simply ignores these specifications (we even tried several unusual styles like “cover, zzz-I, zzz-II, zzz-III, 1, 2, …” for the sake of experiment — just in case Google Play ignores only the trivial specifications like “1, 2, …” but respects others — but it ignores any).

I would think that the problem is unsolvable and that Google Play is simply hardcoded to set “I, II, III, IV, 1, 2, …”, but the thing is that when we did some experiments with submitting a smaller PDF into Google Play, the “I, II, 1, 2, …” (without III and IV) numbering was used! So, it looks like Google Play does support multiple ways (and maybe even supports “1, 2, …”) — but we simply don't know what influences its choice.

_{Some time ago I assumed that it renumbers pages in the way that a page referred by the first item of TOC (I mean a TOC in the metadata) gets number 1. So, I thought that if I artificially start metadata-TOC with a “Bla-bla-bla — page 1”, pages won't get renumbered. But it appeared to be not true.}

So, any ideas how I can force Google Play Books to use “1, 2, …”-style numbering for PDF book?

Answer 1

_{I've promised to write semi-solution to my own question. So…}

Empirically, as of 2020-12, Google Play uses the following criteria for PDF books to choose where Roman numbering (“i”, “ii”, etc) will change to Arabic numbering (“1”, “2”, etc):

• If the PDF has no table-of-contents in metadata: the first page after ii (Roman two) that has “enough text” on it, will be interpreted as “1” (Arabic one).

• I the PDF has table-of-contents in metadata: the first page after “ii” (Roman two) that has “enough text” on it and is referred by the table-of-contents, will be interpreted as “1” (Arabic one).

  To be precise, a page must comply to all of the three criteria (not just one or two of them):

  1. It must be not first or second page of the book (probably Google Play just doesn't take into account that somebody could upload a book without a front cover).
  2. It must be referred by at least one entry from the metadata-stored table-of-contents.
  3. It must have “enough text” on it (see below).

So, returning to my specific case:

• Initially, the book was uploaded without its front cover and back cover. I was unable to make all page numbers Arabic – at least first two pages must have Roman numbering. Thus I prepended the PDF with the front cover and an empty page (obviously, being forced to have two extra pages in the beginning anyway, I preferred to have meaningful ones; but see notes about the automatic empty page below).
• I have several internal title pages at the beginning of the book (the book is a Ukrainian translation of an English book, so it has Ukrainian title page, then English title page, then more detailed Ukrainian title page); after I inserted the front cover and the empty page in the beginning of the PDF, the internal title pages are still there (right after the front cover and empty page). From Google Play's point of view, these internal title pages have ”not enough text” (and, besides that, were not pointed by table-of-contents entries), so it was a puzzle for to force Google Play to reduce amount of Roman-numbered pages to two (it tended to have 3–6 of them). To force Google Play to have only two Roman-numbered pages, I needed to apply both of the following measures:
  1. I created a fake table-of-contents entry (I named it “Title and technical pages”) pointing to the first internal title page.
  2. I added almost-invisible fake text on the first internal title page.

About “enough text on the page”/“not enough text on the page”

I didn't experiment with it a lot. I know that adding white (invisible unless user clicks on it and displaying text layer is turned on in the book settings) text of normal/large font size in such amounts that it covers most of the page space, works. I know that adding tiny-font-size text (even huge amounts of it) in the middle of page doesn't work. Finally, I've ended up by adding white text that doesn't intersect with the proper (visible) page content but covers most of the page space and is somewhat-meaningful (like “sorry, this text is here for technical reasons” and etc).

About an empty page after the front cover

Google Play has feature to automatically add an empty page after the front cover page (“Book info” / “Settings” / “Show advanced settings” / “Insert a blank page after the cover?”). Theoretically, I thought that, when using this feature, it would be enough to prepend only one page (front cover) to my PDF (the empty page would be inserted automatically, and the rest of the pages of my PDF would be numbered as “1”, “2”, etc if I add fake TOC entry & fake text on first of them). But in practice, turning this option on doesn't decrease number of forcefully-Roman PDF pages from 2 to 1 but instead seems to increase it from 2 to 3. I have no idea why, maybe some bug. So I've ended up by inserting an empty after the front cover into PDF manually and not using this option.