KindleUnpack (MobiUnpack): Extracts text, images and metadata from Kindle/Mobi files

[KevinH]

Yes, the bug is in mobi_html.py where we explicitly look for url( style pieces in the css but do not handle that case:

url_pattern = re.compile(br'''(url\(.*?\))''', re.IGNORECASE)

url_img_index_pattern = re.compile(br'''[('"]kindle:embed:([0-9|A-V]+)\?mime=image/[^\)]*["')]''', re.IGNORECASE)

font_index_pattern = re.compile(br'''[('"]kindle:embed:([0-9|A-V]+)["')]''', re.IGNORECASE)

url_css_index_pattern = re.compile(br'''kindle:flow:([0-9|A-V]+)\?mime=text/css[^\)]*''', re.IGNORECASE)

I will try to use your example see how to best to handle the missing case of a kindle:flow but with a mime=image/svg+xml.

I think we can easily handle it as a special case by adding a:

url_svg_image_pattern = re.compile(br'''kindle:flow:([0-9|A-V]+)\?mime=image/svg+xml[^\)]*''', re.IGNORECASE)

and then handling it just like we handle inlined css flow urls.

Thanks,

KevinH

[KevinH]

Should be fixed with this commit:

https://github.com/kevinhendricks/Ki...f6bdbf29157b61

Thank you for the bug report and wonderfully simple testcase!

[jhowell]

Originally Posted by KevinH

When you say round-tripped to Mobi are you referring to just an azw3 or to just an older .mobi (mobi 6) format or to the the joint mobi format (with both azw3 and mobi6 versions) or something else?

I used kindlegen v2.9 to create a dual mobi/kf8 file then I used kindleunpack to decode it. The style css file created in the mobi8/OEBPS/Styles directory shows the problem.

[jhowell]

Originally Posted by KevinH

Should be fixed with this commit...

I tried it and it works for me. Thanks!

[timofonic]

Oh. It's great to see KindleUnpack is still alive. Thanks a lot!

[raoni.kachille]

hello im trying to use the program in the comand line and it gives this following error message

(i was trying to use the gui and was getting a error message, then i tried to use directly from the cmd, like the readme tells and this message appeared.)

File "kindleunpack.py", line 1016, in <module>
sys.exit(main())
File "kindleunpack.py", line 992, in main
infile, outdir = args
ValueError: too many values to unpack (expected 2)

[pdurrant]

The command takes two arguments. The file to unpack and the directory to unpack it into. You seem to be giving it more than two arguments, but you haven't given the command you issued, so exactly what's wrong is a mystery.

[AcidWeb]

There is some new header field (or other black magic) that decide if book with field 501 = EBOK should be put in Comics not Books category.

None of known comics related fields (122, 123, 127, 128) do that. And no known configuration of EPUB trigger this categorization - Syntax in accordance with Amazon Publishing Guidelines is not enough.

[DaleDe]

Originally Posted by pdurrant

The command takes two arguments. The file to unpack and the directory to unpack it into. You seem to be giving it more than two arguments, but you haven't given the command you issued, so exactly what's wrong is a mystery.

In Linux he might have entered a wild card.

Dale

[jhowell]

Originally Posted by AcidWeb

There is some new header field (or other black magic) that decide if book with field 501 = EBOK should be put in Comics not Books category.

My best guess is that the "comic" category is not determined by the book contents, but comes from Amazon's servers.

As a test I put my Oasis in airplane mode, bought a (free) comic, downloaded it and copied it to the Oasis via USB. It did not show as a comic. I then turned off airplane mode and synced the kindle. After that the sideloaded book showed as a comic.

As another test I took a book that showed as a comic and made a copy of it with a different (unused) ASIN. The copy did not show as a comic even though the book contents were identical to the original book, other than the ASIN.

[AcidWeb]

Originally Posted by jhowell

My best guess is that the "comic" category is not determined by the book contents, but comes from Amazon's servers.

As a test I put my Oasis in airplane mode, bought a (free) comic, downloaded it and copied it to the Oasis via USB. It did not show as a comic. I then turned off airplane mode and synced the kindle. After that the sideloaded book showed as a comic.

As another test I took a book that showed as a comic and made a copy of it with a different (unused) ASIN. The copy did not show as a comic even though the book contents were identical to the original book, other than the ASIN.

You are right. Sync is editing p_type column in cc.Entries internal database: Entry:Item -> Entry:Item:Comic

Well played Amazon. Forcing remote query to check information that is available in file header.

[kovidgoyal]

Originally Posted by AcidWeb

Well played Amazon. Forcing remote query to check information that is available in file header.

Amazon has been doing that for years. The earliest example, I can remember is when they refused to display thumbnails for books unless the thumbnails are downloaded from Amazon even though the files have perfectly good thumbnails in their metadata, this was back in 2010-2011 IIRC

Amazon has been gradually trying to make side-loading less and less pleasant on their platforms for a very long time.

[AcidWeb]

Yeah. It is infuriating.

[j.p.s]

For my own use, I am considering adding a global variable named VERBOSITY (or similar) to KindleUnpack and putting some or most print statements in

if VERBOSITY>0:

blocks and adding -q and -v options. Each -q would decrement VERBOSITY and each -v would increment it. (Various blocks might use values other than 0.)

I think this would be generally useful and am curious what others think about this.

Should I open an issue on github with a similar message to discuss this?

[DiapDealer]

So long as none of the various entry point functions' parameter numbers/orders are altered (there are several existing plugins and wrapper programs that use KindleUnpack pieces via import), feel free to submit a pull request on github. But we can discuss it here, I think.

I think most of the people still using the standalone version of KindleUnpack are probably using it for debugging problematic Kindlebooks. So they tend to want as much output as possible for that purpose. For those using the various plugin-ized versions of KindleUnpack, the debug output doesn't really matter anyway.

I'm not opposed to the idea. I just don't know how utilized it would be. I always want to see all the print statements when I use kindleunpack from the command line. Needing to see everything is frankly why I'd be using the standalone version of KU from the command-line in the first place.