Unpacking/repacking mobi books

Jellby · 12-24-2009, 04:58 AM

I think I have already asked this, but I don't think I got a definitive answer/solution.

I have bought some mobipocket books and, as you probably know, they have often deficient/faulty/missing metadata and/or covers (not to talk about the text formatting). Fixing that would be easy with MobiPerl's mobi2mobi or with Calibre's ebook-meta, but sometimes the books are in the generic PRC format (not MOBI) that does not allow adding covers or whatever. When possible, I've been using mobi2html to extract the html and then recreate the book as a proper mobi file, with the correct metadata and covers, but this does not always work, just today I found a book that loses all images in this way (the resulting HTML code has only <img alt="whatever">, but no src and no image files are created).

So, I'd like a working way to extract the contents of a mobi file or to convert from PRC to MOBI. Calibre, as far as I can tell, is no good, because it processes and does all kind of modifications on the HTML/CSS code (or can this be disabled?). MobiPerl's "mobi2html --rawhtml" would be good, I guess, but it does never extract images (and links are given as "filepos", which is not quite convenient).

Is there really no better way to add covers and metadata to a PRC book?

pdurrant · 12-24-2009, 05:43 AM

PRC ebooks that are not Mobipocket ebooks can be in a variety of different formats.

What format are these ebooks, if not Mobi?

Quote:

Originally Posted by Jellby

sometimes the books are in the generic PRC format (not MOBI) that does not allow adding covers or whatever.

user_none · 12-24-2009, 07:12 AM

You can use calibre's command line ebook-convert with the --debug destination/ option. destination/input will give you the un-processed files you're looking for.

tompe · 12-24-2009, 07:37 AM

If you just run mobi2mobi and save as a new file the new file should be a MobiPocket file and images should work. So if you do not need to fox the text formatting it is unnecessary to unpack the book.

HarryT · 12-24-2009, 07:56 AM

Quote:

Originally Posted by pdurrant

PRC ebooks that are not Mobipocket ebooks can be in a variety of different formats.

What format are these ebooks, if not Mobi?

It's the "old" Mobi format - the one that Book Designer creates, for example. These books have a "file tag" of "Text/Read" rather than "Book/Mobi", and then don't store any metadata other than (I think) the book title.

Jellby · 12-24-2009, 08:00 AM

Quote:

Originally Posted by user_none

You can use calibre's command line ebook-convert with the --debug destination/ option. destination/input will give you the un-processed files you're looking for.

Yes, I've found that, thanks. But it converts the images to JPG (whereas mobi2html gave BMP files for the same book). Also, how would I recreate the mobi without involving all the usual Calibre parsing and processing?

Quote:

Originally Posted by tompe

If you just run mobi2mobi and save as a new file the new file should be a MobiPocket file and images should work. So if you do not need to fox the text formatting it is unnecessary to unpack the book.

Really? I think I tried some time with that, and it didn't work... I'll try again.

Jellby · 12-24-2009, 10:20 AM

Quote:

Originally Posted by tompe

If you just run mobi2mobi and save as a new file the new file should be a MobiPocket file and images should work. So if you do not need to fox the text formatting it is unnecessary to unpack the book.

Let's see. Just running mobi2mobi and trying to change the cover gave me:

Code:

Invalid type '-' in unpack at /home/ignacio/bin/mobiperl/MobiPerl/EXTH.pm line 268.

The doctype is MOBI, so maybe the "old PRC format" issue is not the problem. Apparently ebook-meta can set the cover for this file. For another file, however, ebook-meta fails with "ValueError: non-existent record 4294968689L", while mob2mobi silently fails to set the cover (it does not give an error, but after that it still says "ERROR: No cover image in book").

Really, this is a pain in you know where

tompe · 12-24-2009, 04:53 PM

What I meant was that you first do "mobi2mobi oldfile.prc --outfile newfile.mobi" and then try to set the meta data on newfile.mobi.

But the error message is a bit strange. mobi2mobi seem to think that the file have EXTH headers since "if ($exthflg & 0x40)" seems to become true. And that should not happen if it is not a real MobiPocket file. So you could try to add "& false" to this if condition to see if that works.

Jellby · 12-25-2009, 05:58 AM

Quote:

Originally Posted by tompe

What I meant was that you first do "mobi2mobi oldfile.prc --outfile newfile.mobi" and then try to set the meta data on newfile.mobi.

I tried that too. Same result. Could the absence of cover after adding it be related to this piece of output?

Code:

EXTH    item: 203 - hasFakeCover - 4 - 0x0000

Quote:

But the error message is a bit strange. mobi2mobi seem to think that the file have EXTH headers since "if ($exthflg & 0x40)" seems to become true. And that should not happen if it is not a real MobiPocket file. So you could try to add "& false" to this if condition to see if that works.

Yes, it seems this files are problematic. How would I do the change? I tried with "if (($exthflg & 0x40) & false)", but I got:

Code:

Bareword "false" not allowed while "strict subs" in use at /home/ignacio/bin/mobiperl/mobi2mobi line 608.

Commenting out the "use strict" line gave me the same "Invalid type" message

tompe · 12-25-2009, 06:21 AM

Aha, I probably misremembered (too much C++ lately). Use 0 instead of false.

12-24-2009, 04:58 AM	#1
Jellby frumious Bandersnatch Posts: 7,515 Karma: 18512745 Join Date: Jan 2008 Location: Spaniard in Sweden Device: Cybook Orizon, Kobo Aura	Unpacking/repacking mobi books I think I have already asked this, but I don't think I got a definitive answer/solution. I have bought some mobipocket books and, as you probably know, they have often deficient/faulty/missing metadata and/or covers (not to talk about the text formatting). Fixing that would be easy with MobiPerl's mobi2mobi or with Calibre's ebook-meta, but sometimes the books are in the generic PRC format (not MOBI) that does not allow adding covers or whatever. When possible, I've been using mobi2html to extract the html and then recreate the book as a proper mobi file, with the correct metadata and covers, but this does not always work, just today I found a book that loses all images in this way (the resulting HTML code has only <img alt="whatever">, but no src and no image files are created). So, I'd like a working way to extract the contents of a mobi file or to convert from PRC to MOBI. Calibre, as far as I can tell, is no good, because it processes and does all kind of modifications on the HTML/CSS code (or can this be disabled?). MobiPerl's "mobi2html --rawhtml" would be good, I guess, but it does never extract images (and links are given as "filepos", which is not quite convenient). Is there really no better way to add covers and metadata to a PRC book?

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Mobi books, same experience as kindle?	abvmoose	Amazon Kindle	4	08-04-2010 11:44 AM
Kindle dictionary and .mobi books?	yuriman2002	Amazon Kindle	2	03-02-2010 11:04 AM
Can Mobi books be 'exploded'	AnemicOak	Kindle Formats	26	03-18-2009 03:16 PM
new books in mobi format	sassanik	IMP	11	09-08-2008 09:40 AM
New and improved Mobi books for Gen3	HarryT	Bookeen	9	11-19-2007 05:55 AM

12-24-2009, 07:12 AM	#3
user_none Sigil & calibre developer Posts: 2,488 Karma: 1063785 Join Date: Jan 2009 Location: Florida, USA Device: Nook STR	You can use calibre's command line ebook-convert with the --debug destination/ option. destination/input will give you the un-processed files you're looking for.

12-24-2009, 07:37 AM	#4
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	If you just run mobi2mobi and save as a new file the new file should be a MobiPocket file and images should work. So if you do not need to fox the text formatting it is unnecessary to unpack the book.

12-24-2009, 04:53 PM	#8
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	What I meant was that you first do "mobi2mobi oldfile.prc --outfile newfile.mobi" and then try to set the meta data on newfile.mobi. But the error message is a bit strange. mobi2mobi seem to think that the file have EXTH headers since "if ($exthflg & 0x40)" seems to become true. And that should not happen if it is not a real MobiPocket file. So you could try to add "& false" to this if condition to see if that works.

12-25-2009, 06:21 AM	#10
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	Aha, I probably misremembered (too much C++ lately). Use 0 instead of false.

Advert

Advert