Mobiperl Mobiperl - Perl tools for handling MobiPocket files

[kovidgoyal]

You can specify the encoding when using mobigen. I haven't bothered with getting the filepos exactly right since it's not that critical. As long as you make sure you dont insert an anchor inside another <> you're fine.

If I had to guess I'd say filepos represents byte position not character position, since in the cp1252 encoding every character is one byte and that was the original encoding mobigen used, back when it was prcgen.

[tompe]

Originally Posted by kovidgoyal

You can specify the encoding when using mobigen. I haven't bothered with getting the filepos exactly right since it's not that critical. As long as you make sure you dont insert an anchor inside another <> you're fine.

If I had to guess I'd say filepos represents byte position not character position, since in the cp1252 encoding every character is one byte and that was the original encoding mobigen used, back when it was prcgen.

I have confirmed that it is byte position. My original test file was corrupt.

Yes, i assume you can just put an anchor somewhere but I wanted to re-use the existing anchor if it existed...

[tompe]

Verson 0.0.32 is available at

http://www.ida.liu.se/~tompe/mobiperl/

No Windows binaries for this version yet. Feel free to contribute them...

I removed mobi2imp from the distribution since it is distributed in another thread and it is to much work to keep it updated for me. If things stabilize I can put it back if people think it is a good idea.

I added some instruction for Mac on the web site. Please give feedback how to make them better if you try to use them.

Changes in 0.0.32:

• html2mobi now works if images that are refered to in the html file does not exist.
  
• Added flag --fixhtmlbr to html2mobi. This flag will try to fix html files where two <br> have been uses instead of <p></p>.
  
• Removed mobi2imp from distribution since it is distributed seperately in MobileRead forum. Duplications are bad and confusing.
  
• Fixed so that mobi2html works with utf-8 encoded files.
  
• Added meta tag with charset in the html generated by mobi2html.
  
• Added instructions of how to install Perl and modules on Mac on the web page.

[badgoodDeb]

Oops -- meant to be a PM.

[tompe]

There is a version 0.0.33 now since there was a bug in the utf-8 fix that caused books with codepage 1252 converted to HTML not to be displayed correctly.

[tompe]

Just so the best version is available if somebody do a Windows version...

I did a 0.0.34 and the change is:

• png files did not work for html2mobi. Changed so that png files are converted to jpg before it is packed into the mobi file.

[Ortep]

Originally Posted by tompe

I have confirmed that it is byte position. My original test file was corrupt.

You were working late. Great job!!!

If there is corruption in the file, it is something that was caused by the software used in creating it. I have 9 others with the same problem.

The corruption has no (direct) effect when I use them on my Cybook.

The only thing I did notice was that the cover was not visible. So that was when I started to experiment with mobi2mobi. And when I didn't like the result I used mobi2html in an attempt to start from scratch.

Perhaps I should contact the writer and tell him about it

[tompe]

Originally Posted by Ortep

If there is corruption in the file, it is something that was caused by the software used in creating it. I have 9 others with the same problem.

The corruption has no (direct) effect when I use them on my Cybook.

The only noticeable effect is probably that the links will jump to the wrong file position but you will only notice that if it jumps inside a format tag or something similar and in the case of extra bytes the jump will be too short and then you will probably not notice it.

[IceHand]

Any chance that you'll include huffdic compression support for mobi2html?

[tompe]

Originally Posted by IceHand

Any chance that you'll include huffdic compression support for mobi2html?

Maybe. But I would have to convert the code and it is not trivial. If somebody read the Python code and write a "mathematical" specification of the algorithm I would definitely implement it. I saw that kovidgoyal had distributed the code under GPL so I can just choose to trust that he is correct in this. Otherwise I had some concern about the status of the resulting code. I will not do the work if I cannot distribute the code under GPL. A written description of the algorithm would also make totally clear that it is no problem to distribute the code under GPL.

[Gudy]

Originally Posted by tompe

Verson 0.0.32 is available at

http://www.ida.liu.se/~tompe/mobiperl/

No Windows binaries for this version yet. Feel free to contribute them...

Here's the Windows binary build for 0.0.34. I once again used HarryT's Sherlock Holmes Omnibus for testing, and something's not right: mobi2html converts the non-breaking space character (& nbsp; / & #xA0; / &# 160; ) into the replacement character (& #xFFFD; / & #65533; ), i.e. the Unicode equivalent of "unknown character".

Also, you may want to add the Encode package both to the installation instructions on your web site and the mh target in the Makefile.

[tompe]

Originally Posted by Gudy

Here's the Windows binary build for 0.0.34. I once again used HarryT's Sherlock Holmes Omnibus for testing, and something's not right: mobi2html converts the non-breaking space character (& nbsp; / & #xA0; / &# 160; ) into the replacement character (& #xFFFD; / & #65533; ), i.e. the Unicode equivalent of "unknown character".

Also, you may want to add the Encode package both to the installation instructions on your web site and the mh target in the Makefile.

I had hoped that Encode was included in the core files. Maybe I can try to use another function. I will check with the Sherlock Holmes Omnibus and see what the problem is.

[JSWolf]

tompe, can you also check your code against purchased Mobipocket eBooks as well?

[tompe]

Originally Posted by tompe

I had hoped that Encode was included in the core files. Maybe I can try to use another function. I will check with the Sherlock Holmes Omnibus and see what the problem is.

The problem is that this is an old file so it does not have the information about codepage and I had maybe wrongly assumed that if it was not specified it was UTF-8...

I have now changed the code so that if not "codepage" is specified it does nothing. If 1252 is specified it add the meta header for that. If 65001 (UTF-8) is specified I add meta header for that and decode the byte stream with before sending it to HTML::TreeBuilder.

It worked for the Sherlock Holmes omnibus. I will check the UTF-8 example I have and maybe do a new release later today...

[JSWolf]

Please include a Windows edition too.