Sony PRS T-1 won't open epub created from zip file

[steinarb]

I have created an ePub from a zip file from the Yahoo! Hadoop Tutorial. The ePub shows up fine in the Calibre ebook reader, and in fbreader.

But when I try to read it in my Sony PRS T-1 reader, the reader immedately returns to the Home menu. I.e. the ePub file won't display in the Sony.

Has anyone seen something similar? Is there a way to get debugging output from the Sony? Is there a "lint"-like utility for ePub files?

I have attached a zip file containing the export of the book from calibre, i.e. containg the original zip file, an opf file, a cover, and the created ePub file.

I created the zip file by downloading and unpacking the zip file on the tutorial's home page, removing all of the example code (reducing it in size from 600+MB to 0.3MB), and adding a toc.html as indicated by the calibre documentation.

Note that the cover says calibre 0.8.51, but I am currently using calibre 0.9.5.

The output profile in calibre is set to "Sony reader".

Thanks!


- steinar

[DoctorOhh]

Since there is no calibre question here. This may be better served in the Epub or Sigil forums.

[steinarb]

Originally Posted by DoctorOhh

Since there is no calibre question here. This may be better served in the Epub or Sigil forums.

Let me clarify: epubs produced by calibre from zip files, are broken with respect to the Sony Reader. This goes for both the PRS T-1 and the PRS T-2 (they exhibit the same behaviour).

I have had little success with this conversion (zip->epub), and have other example documents with the same issue. I just tried to limit the question to one particular file to help pinpoint the issue.

Or to be even clearer: this is probably a calibre bug, but I would like to figure out more about what's actually wrong with the produced epub file, before reporting it.

I have one successful conversion, but that's copyrighted (the 6 sample chapters of "Captain Vorpatril's Alliance" by Lois McMaster Bujold). I can't remember if I did an HTML to epub conversion or a zip to epub conversion (is it possible to import a multi-file HTML document? I will check when I get home). Also this was a document without images, and no hyperlinks outside of the document navigation.

[DoctorOhh]

Originally Posted by steinarb

I can't remember if I did an HTML to epub conversion or a zip to epub conversion (is it possible to import a multi-file HTML document? I will check when I get home). Also this was a document without images, and no hyperlinks outside of the document navigation.

Dragging the html index page (or TOC) to calibre should grab up all extra files the index points to and combine them into 1 zip file.

Originally Posted by steinarb

Let me clarify: epubs produced by calibre from zip files, are broken with respect to the Sony Reader. This goes for both the PRS T-1 and the PRS T-2 (they exhibit the same behaviour).

I have had little success with this conversion (zip->epub), and have other example documents with the same issue. I just tried to limit the question to one particular file to help pinpoint the issue.

Or to be even clearer: this is probably a calibre bug, but I would like to figure out more about what's actually wrong with the produced epub file, before reporting it.

Zip(html) to epub is a mainstay of calibre and has been working great for years. One way to get converted ePubs to crash is to increase the default Split files larger than: 260K setting under Preferences - Output options - Epub output. Your epub though doesn't seem to have any large html chunks so this is not your problem.

Open up a bug report, follow these guidelines for describing the problem and providing data, then attach the file to the bug report.

Sorry I couldn't help further.

Good Luck.

[steinarb]

Originally Posted by DoctorOhh

Zip(html) to epub is a mainstay of calibre and has been working great for years.

Yes. The resulting epub works fine in ebook-reader, and FbReader on a PC. Today I've opened the ePub in FbReader and Aldiko on an Android tablet (Samsung Galaxy Tab 2 7.0). It opened in both places, but worked better in FbReader than in Aldiko.

I've also tried it in iBooks on an iPad, and it worked very well there. Better than in the android readers. By "better" I mean that there are fewer empty or almost-empty pages, and just flicking through the pages made more sense (I don't understand why this should be different across readers, but I've seen the same effect before).

But it fails on my PRS-T1.

I wonder if it will work in the previous generation of Sony readers...? I had a PRS-950, but it died the eInk-death that sometimes afflict the ebook readers.

Originally Posted by DoctorOhh

One way to get converted ePubs to crash is to increase the default Split files larger than: 260K setting under Preferences - Output options - Epub output. Your epub though doesn't seem to have any large html chunks so this is not your problem.

Right. I remember reading something about that. But as you said, the individual HTML pages here are quite small.

Thanks for your help, so far.

[steinarb]

Originally Posted by DoctorOhh

Dragging the html index page (or TOC) to calibre should grab up all extra files the index points to and combine them into 1 zip file.

I tried this now, and the zip file was restructured compared to the original: all HTML files were on the same level. So I hopefully tried a new zip->epub conversion.

But the resulting ePub still fails to open on the PRS-T1. It just exits back to the menu, showing this ePub as the most recently read document.

[theducks]

Originally Posted by steinarb

I tried this now, and the zip file was restructured compared to the original: all HTML files were on the same level. So I hopefully tried a new zip->epub conversion.

But the resulting ePub still fails to open on the PRS-T1. It just exits back to the menu, showing this ePub as the most recently read document.

Validae the EPUB with Flightcrew or at http://validator.w3.org/

This should indicate if it is malformed (but not oversized segments)

[kovidgoyal]

SONY readers are among the les srobust set of EPUB readers in terms of what markup trips them up. I suggest converting the file to MOBI and then converting the MOBI to epub, that will dumb down the markup enough that the SONY should handle it. You may lose some formatting that MOBI cannot handle, however.

[Agama]

Originally Posted by steinarb

when I try to read it in my Sony PRS T-1 reader, the reader immedately returns to the Home menu.

Kovid is correct here about Sony readers. My reader is unforgiving with less than perfect ePubs. I've generally only caused imperfections by using Tweak Book, (not had any problems directly from calibre conversions), and have had symptoms similar to those you describe.

Having spent some time tweaking I was reluctant to lose my efforts by doing another conversion so I checked my book at ePub Validator http://validator.idpf.org/ which located the problems with pinpoint accuracy. The output can be a bit verbose but it certainly shows what's wrong.

This may give you a means of fixing your own ePubs and also the info necessary to see whether a change in calibre is required.

[theducks]

Sony uses ADE (Mobile reader) as do many devices like my PEz.
It forgives errors, NOT.
Go down to the EPUB forum and read the sticky ADE bugs http://www.mobileread.com/forums/showthread.php?t=94733 (the behavior is a bug, because CSS errors are to be ignored, not cause a crash)

forget a semicolon and your whole CSS will/can be ignored (and it is real fun to find without using the validator service .

[steinarb]

Originally Posted by kovidgoyal

SONY readers are among the les srobust set of EPUB readers in terms of what markup trips them up. I suggest converting the file to MOBI and then converting the MOBI to epub, that will dumb down the markup enough that the SONY should handle it. You may lose some formatting that MOBI cannot handle, however.

This worked: converting ePub->MOBI->ePub gave an ePub that I was able to open in the PRS-T1.

Thanks, Kovid!