Mobiperl Mobiperl - Perl tools for handling MobiPocket files - Page 44

tompe · 01-03-2009, 09:39 PM

OK, I see the problem now. The prc file does not contain HTML code and I think the MobiPocket format requires HTML code. And if you do mobi2html followed by html2mobi you loose the formatting.

So I think the mistake was to generate files of type TEXt with the content as ordinary text.

This is the same problem as for the file in the other thread.

My lit2mobi can of course be used in a batch job.

nrapallo · 01-03-2009, 10:48 PM

Quote:

Originally Posted by tompe

OK, I see the problem now. The prc file does not contain HTML code and I think the MobiPocket format requires HTML code. And if you do mobi2html followed by html2mobi you loose the formatting.

So I think the mistake was to generate files of type TEXt with the content as ordinary text.

This is the same problem as for the file in the other thread.

My lit2mobi can of course be used in a batch job.

My Mobi2IMP can handle that TEXt/REAd .prc because I included a simple detection for text files (in bold) as follows:

Code:

my $text = $pdb->text;

if (defined $opt_rawhtml) {
    binmode (STDOUT);
    print $text;
}
if (substr ($text, 0, 6) ne "<html>" && substr ($text, 0, 6) ne "<HTML>" ) {                                       
    open TEMPFILE, ">$explodedir/$htmlfile.txt" or die "Cannot create .txt file";
    binmode (TEMPFILE);
    print TEMPFILE $text;
    close TEMPFILE;

    open TEMPFILE, "$explodedir/$htmlfile.txt" or die "Cannot open .txt file";
    my @lines = <TEMPFILE>;               # Read it into an array
    close TEMPFILE;                       # Close the file

    #unlink "$explodedir/$htmlfile.txt";

    $text = "<HTML><HEAD></HEAD>\n<BODY>\n" ;

    my $line;
    foreach $line (@lines) {              # assign @lines to $line, one at a time
       $text .= "<P>" . $line . "</P>" unless (length($line) == 1);  
    }

    $text .= "\n</BODY>\n</HTML>\n";

}

tompe · 01-04-2009, 12:04 AM

Yes, it would be possible to build a txt2mobi command that do something like this. But in mobi2mobi I did not want to touch the content data at all.

But a better approach seems to be to use some external program like

http://txt2html.sourceforge.net/

to make a html file from a text file and then run mobigen or html2mobi on the html file.

Here is a Windows program for converting text to html:

http://www.softpedia.com/get/Interne...onverter.shtml

Dr Dave · 01-04-2009, 12:08 AM

Quote:

Originally Posted by tompe

My lit2mobi can of course be used in a batch job.

Can you provide a batch script? I have seen examples of how to do it with titles that are all one "word", but the overwhelming majority of mine are multiple words and I can't seem to get a script generated to handle all of them in a given directory.

Thanx again!

tompe · 01-04-2009, 12:15 AM

Quote:

Originally Posted by Dr Dave

Can you provide a batch script? I have seen examples of how to do it with titles that are all one "word", but the overwhelming majority of mine are multiple words and I can't seem to get a script generated to handle all of them in a given directory.

That I have to let somebody that know Windows answer. In Unix I would do:

Code:

find . -name "*.lit" -exec lit2mobi \{\} \;

and I suppose this would work in Windows in a Cygwin shell.

Also notice that lit2mobi have some (undocumented, read the source) flags to fix strange html code in a lit file.

tompe · 01-07-2009, 09:12 PM

I did check in the fix for multibyte character overlap discussed in another thread. I till have an offset of one for file position using mobi2html on files using multibyte character overlap.

nrapallo · 01-07-2009, 10:28 PM

Quote:

Originally Posted by tompe

I did check in the fix for multibyte character overlap discussed in another thread. I till have an offset of one for file position using mobi2html on files using multibyte character overlap.

I incorporated those changes into my Mobi2IMP and now the Feedbooks.com ebook, Rosenbaum - The Ant King and Other Stories.mobi, discussed here properly decodes and creates valid links to all filepos positions.

Great debugging!

EDIT: 8-Jan-2009 added results from beta Mobi2IMP as well as beta Mobi2IMP.pl for your review.

tompe · 01-08-2009, 06:33 AM

Quote:

Originally Posted by nrapallo

I incorporated those changes into my Mobi2IMP and now the Feedbooks.com ebook, Rosenbaum - The Ant King and Other Stories.mobi, discussed here properly decodes and creates valid links to all filepos positions.

You do not get filepos pointing to the position before "<"? I got warnings for that but it seems to be off by one only.

pdurrant · 01-21-2009, 07:37 AM

Images in Mobipocket files:

I recently came across a problem with Calibre's import on Mobipocket files.
http://calibre.kovidgoyal.net/ticket/1645

Kovid says the problem was that mobi2mobi can create non-sequential image records. That is, there can be some non-image records between the first image in the file (pointed to in the header) and the last image.

I'm not sure this is a bug myself, although if Mobipocket itself doesn't create files with images mixed in with non image data, perhaps mobi2mobi shouldn't either.

The real problem is, I can't see a clean way to fix the issue. Any ideas?

tompe · 01-22-2009, 07:22 PM

Quote:

Originally Posted by pdurrant

Images in Mobipocket files:

I recently came across a problem with Calibre's import on Mobipocket files.
http://calibre.kovidgoyal.net/ticket/1645

Kovid says the problem was that mobi2mobi can create non-sequential image records. That is, there can be some non-image records between the first image in the file (pointed to in the header) and the last image.

I'm not sure this is a bug myself, although if Mobipocket itself doesn't create files with images mixed in with non image data, perhaps mobi2mobi shouldn't either.

The real problem is, I can't see a clean way to fix the issue. Any ideas?

I do not get what the problem is. I thought that all records after the text is images. As I recall adding a thumb nail will add a new record and set the pointer. Why should this be wrong? Since we do not have a spec we can not know that it is wrong. Does some debugged code fails to handle this?

pdurrant · 01-23-2009, 04:28 AM

It seems that there can be some non-image records after the image records. So then adding more images results in the images not all being in consecutive records.

As you say, we don't have a spec for Mobipocket, but since Mobipocket Creator always put images in consecutive records, I can see Kovid's point.

However, since he's going to change Calibre to work with non-consecutive images , I don't think a change in MobiPerl is necessary (or perhaps even possible). I just wanted to find out what other people thought.

Quote:

Originally Posted by tompe

I do not get what the problem is. I thought that all records after the text is images. As I recall adding a thumb nail will add a new record and set the pointer. Why should this be wrong? Since we do not have a spec we can not know that it is wrong. Does some debugged code fails to handle this?

Jellby · 01-23-2009, 04:33 AM

Quote:

Originally Posted by pdurrant

As you say, we don't have a spec for Mobipocket, but since Mobipocket Creator always put images in consecutive records, I can see Kovid's point.

But all other mobipocket readers read those mobiperl-created files without problems, don't they?

tompe · 01-23-2009, 05:03 AM

Quote:

Originally Posted by Jellby

But all other mobipocket readers read those mobiperl-created files without problems, don't they?

I thought so. It would be interesting to know if this is not the case.

It is no problem to put them in another order but since I do not know if the meta data contains any pointer to the non image records I on purpose made sure that mobi2mobi did not change anything in the data records.

pdurrant · 01-23-2009, 07:40 AM

This is why I said I couldn't see any clean way to fix the 'problem'. We can't move any records because, as you say, we don't know what other data might expect them to be just where they are.

Quote:

Originally Posted by tompe

It is no problem to put them in another order but since I do not know if the meta data contains any pointer to the non image records I on purpose made sure that mobi2mobi did not change anything in the data records.

kovidgoyal · 01-23-2009, 01:10 PM

Quote:

Originally Posted by pdurrant

This is why I said I couldn't see any clean way to fix the 'problem'. We can't move any records because, as you say, we don't know what other data might expect them to be just where they are.

Actually, if you move them and something breaks, that will be an excellent way to figure out more about the format.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Mobi2Mobi Mobi2Mobi v0.13 - GUI for Mobiperl tools	Jad	Kindle Formats	476	03-15-2015 05:51 PM
Tools for Editing Kindle .mobi Files?	GJN	Kindle Formats	33	12-26-2013 02:05 PM
Handy Perl Script to convert HTML0 files to smartquotes	maggotb0y	Sony Reader	0	04-12-2007 11:49 AM
PRS-500 Perl tools to generate Reader content	TadW	Sony Reader Dev Corner	0	01-08-2007 05:55 AM
gmail copy (gmcp) - Perl script to copy files to/from Gmail	Colin Dunstan	Lounge	0	09-04-2004 01:24 PM

01-03-2009, 09:39 PM	#646
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	OK, I see the problem now. The prc file does not contain HTML code and I think the MobiPocket format requires HTML code. And if you do mobi2html followed by html2mobi you loose the formatting. So I think the mistake was to generate files of type TEXt with the content as ordinary text. This is the same problem as for the file in the other thread. My lit2mobi can of course be used in a batch job.

01-04-2009, 12:04 AM	#648
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	Yes, it would be possible to build a txt2mobi command that do something like this. But in mobi2mobi I did not want to touch the content data at all. But a better approach seems to be to use some external program like http://txt2html.sourceforge.net/ to make a html file from a text file and then run mobigen or html2mobi on the html file. Here is a Windows program for converting text to html: http://www.softpedia.com/get/Interne...onverter.shtml

01-07-2009, 09:12 PM	#651
tompe Grand Sorcerer Posts: 7,452 Karma: 7185064 Join Date: Oct 2007 Location: Linköpng, Sweden Device: Kindle Voyage, Nexus 5, Kindle PW	I did check in the fix for multibyte character overlap discussed in another thread. I till have an offset of one for file position using mobi2html on files using multibyte character overlap.

01-21-2009, 07:37 AM	#654
pdurrant The Grand Mouse 高貴的老鼠 Posts: 73,964 Karma: 315160596 Join Date: Jul 2007 Location: Norfolk, England Device: Kindle Oasis	Images in Mobipocket files: I recently came across a problem with Calibre's import on Mobipocket files. http://calibre.kovidgoyal.net/ticket/1645 Kovid says the problem was that mobi2mobi can create non-sequential image records. That is, there can be some non-image records between the first image in the file (pointed to in the header) and the last image. I'm not sure this is a bug myself, although if Mobipocket itself doesn't create files with images mixed in with non image data, perhaps mobi2mobi shouldn't either. The real problem is, I can't see a clean way to fix the issue. Any ideas?