AZW3 to EPUB Conversion issue

[Cougar281]

I usually purchase books from Books A Million, but every now and then, I'll purchase books from Amazon, usually because I can't find it elsewhere. I've purchased a few books in the past from Amazon and had no issue getting them converted to EPUB so that they would work on my Sony PRS-T1, but I have one right now, Seveneves, that I haven't been able to get to convert right.

When viewed using the Calibre reader, all three formats (AZW3, EPUB and PDF) look fine. But when uploaded to my reader, the EPUB stops at page 4 (which seems to be the end of the Table of Contents) and won't advance further. If I use the 'Navigate Page' feature and navigate to page 5, I get a blank page that has 'Page 2-13 of 712' at the bottom. Going straight to pages 6-13 give the exact same blank page, but navigating to page 14 brings me to the page that should be right after the TOC. The PDF seemed ok at first, but I fairly quickly found some gaps that I determined were quite a bit of missing text buy comparing it with the same copy in the Calibre reader.

Does anyone have any thoughts on why this particular book seems to be giving me trouble, and maybe a way to get it to convert properly?

[kovidgoyal]

Try converting it to TXT and reading that.

[Cougar281]

Bringing this back up - while converting to TXT 'worked', it was all kinds of messed up, with the text being huge and spaced 'oddly', and the page count more than doubled. I discovered that with the epub, once I navigated directly to page 14, I was able to read the while book as an epub without further issue. But if at all possible, I'd really like to get it converted such that the epub is 'perfect', so if I choose to read it again at some point, it 'just works' like all of my other books. I just tried again using 5.23.0 and, interestingly, while the output size was different (1,057K for the older one and 1,044K for the new one), I got the same issue. Is there something I could send you like debug output that might help determine why this particular book is having issues converting to epub?

[kovidgoyal]

This is an issue with your epub reader, you should send its developers your epub file and ask them to fix their reader.

[rantanplan]

Modern azw files contain the original epub file it was based on. So instead of converting the file, you can try to use https://github.com/kevinhendricks/KindleUnpack to simply get the original file.

[JSWolf]

Originally Posted by Cougar281

I usually purchase books from Books A Million, but every now and then, I'll purchase books from Amazon, usually because I can't find it elsewhere. I've purchased a few books in the past from Amazon and had no issue getting them converted to EPUB so that they would work on my Sony PRS-T1, but I have one right now, Seveneves, that I haven't been able to get to convert right.

When viewed using the Calibre reader, all three formats (AZW3, EPUB and PDF) look fine. But when uploaded to my reader, the EPUB stops at page 4 (which seems to be the end of the Table of Contents) and won't advance further. If I use the 'Navigate Page' feature and navigate to page 5, I get a blank page that has 'Page 2-13 of 712' at the bottom. Going straight to pages 6-13 give the exact same blank page, but navigating to page 14 brings me to the page that should be right after the TOC. The PDF seemed ok at first, but I fairly quickly found some gaps that I determined were quite a bit of missing text buy comparing it with the same copy in the Calibre reader.

Does anyone have any thoughts on why this particular book seems to be giving me trouble, and maybe a way to get it to convert properly?

I know what your problem is. The problem is that your ePub has errors. You have to go into the Calibre editor and fix the errors. RMSDK (reading software on the T1) is encountering the errors and not displaying your ePub correctly. You will need to install the epubcheck plugin and using both the Calibre editor's error checker and epubcheck, you have to get both to say there are no errors.

If you don't know HTML/CSS, now's a good time to learn.

[deback]

I would open the ePub file in the Calibre editor and click on the Beautify Files button. This might fix the problem (or it might not, but it's worth trying). Also, click on the Check Files button to see if Calibre finds any errors. If it does, click over on the right to automatically fix all errors found.

[Cougar281]

Thanks JSWolf and deback. I'll give that a shot.

kovidgoyal, as JSWolf and deback suggested, it's not the reader - it's the book. However, how the reader is handling it is another matter. On my PRS-T1, it gets to a point and simply won't go any further unless I manually advance to a specific 'working' page. On my older PRS-600, the same issue is present with the book, but rather than halt at the damage, it pushes through and ends up advancing to the first usable page. The same happens with the Kobo Clara HD I just bought.

[Cougar281]

I installed epubcheck and after a bunch of poking around to figure out how to run it, it didn't seem to do anything. When I went to 'Edit' the book, then 'plugins', followed by 'Run epubcheck', it seemed to indicate that it was running. But nothing after that.

So I looked at the files. As I'm sure you're aware, there are basically a bunch of HTML files inside an epub. Looking at what was in this file, there was a 'part0002_split_000.html' and a 'part0002_split_001.html'. The 'part0002_split_000.html' seemed to be the proper and complete TOC when viewed in the 'file preview', but the 'part0002_split_001.html', viewed in the 'file preview', was blank. The next file, 'part0003.html', was the beginning of the book. So on a 'whim', I deleted 'part0002_split_001.html' and saved the book, then sent it to my PRS-T1, since that's the one that REALLY didn't like the error. Works perfect now (or at least seems to).

[DNSB]

@cougar281: Did you see a window similar to the one on the left side of the attached image (I keep epubcheck docked with file preview and live CSS, your location will vary.

[DNSB]

Interesting. A duplicated message.

[JSWolf]

@Cougar281, to run the epubcheck plugin, you need to have Javascript installed. Install the 64-bit version if running a 64-bit OS.

[Cougar281]

Originally Posted by DNSB

@cougar281: Did you see a window similar to the one on the left side of the attached image (I keep epubcheck docked with file preview and live CSS, your location will vary.

I'm not exactly sure what you're referring to, but if you mean the file browser/test and all of the html/xhtml pages below it, yes. But I did not get the 'Check Book' window above it or the 'EpubCheck' Window to the right.

I do have Java 8 update 221, both 32-bit and 64-bit installed on my machine.

I think I resolved the issue with my epub by deleting the 'part0002_split_001.html' file from it, but for grins, I installed just the latest version of java on a VM with Calibre and installed the plugin, and it worked - so either having both the 32-bit and 64-bit versions of java confused things, or the ABSOLUTE latest version is required (Java sucks so bad.... I cringe any time it's required). Whatever the case, it seems there are a 'few' errors. From skimming it, it seems pretty much the entirety of 'part0002_split_001.html' is one big error. But the list is pretty long... So I clicked 'tools' > 'Fix HTML - All Files' and saved the file, but no change on the device. Maybe I just didn't do it right, but as just removing the 'part0002_split_001.html' seems to have fixed it, might just stick with the version where I deleted the 'part0002_split_001.html' file... Oddly, out of all the epub books I have, mostly native epubs but some converted from other formats, this is the ONLY one I've had any issue with.

[JSWolf]

Originally Posted by Cougar281

I'm not exactly sure what you're referring to, but if you mean the file browser/test and all of the html/xhtml pages below it, yes. But I did not get the 'Check Book' window above it or the 'EpubCheck' Window to the right.

I do have Java 8 update 221, both 32-bit and 64-bit installed on my machine.

I think I resolved the issue with my epub by deleting the 'part0002_split_001.html' file from it, but for grins, I installed just the latest version of java on a VM with Calibre and installed the plugin, and it worked - so either having both the 32-bit and 64-bit versions of java confused things, or the ABSOLUTE latest version is required (Java sucks so bad.... I cringe any time it's required). Whatever the case, it seems there are a 'few' errors. From skimming it, it seems pretty much the entirety of 'part0002_split_001.html' is one big error. But the list is pretty long... So I clicked 'tools' > 'Fix HTML - All Files' and saved the file, but no change on the device. Maybe I just didn't do it right, but as just removing the 'part0002_split_001.html' seems to have fixed it, might just stick with the version where I deleted the 'part0002_split_001.html' file... Oddly, out of all the epub books I have, mostly native epubs but some converted from other formats, this is the ONLY one I've had any issue with.

If I am not mistaken, having the latest Java is what did it. But you really should dump the 32-bit version and install the latest version in your non-VM environment.