Mobiperl Mobiperl - Perl tools for handling MobiPocket files

[DaleDe]

Originally Posted by HarryT

Hi Dale,

The "BookDesigner" program, which many people here use, creates "MobiPocket" books using the TEXt/REAd descriptors in the PRC header. You'll find hundreds of such books on this site.

Yuck, breaking Palm rules is not a good thing. BookDesigner is a mixed blessing at best. I really have a love/hate relationship with that program. Maybe the .pdb extension can help to distinguish real PalmDoc files from those fake BookDesigner ones. There is a clear type for MobiPocket files, it is BOOKMOBI in that field.

Dale

[DMcCunney]

Originally Posted by DaleDe

Yuck, breaking Palm rules is not a good thing. BookDesigner is a mixed blessing at best. I really have a love/hate relationship with that program. Maybe the .pdb extension can help to distinguish real PalmDoc files from those fake BookDesigner ones. There is a clear type for MobiPocket files, it is BOOKMOBI in that field.

Well, the PDB extension will tell you it's a Palm Database. But the content of a Palm Database file can be arbitrary. You need to look at the header record of the file for the REAd creator ID and TEXt type to be sure it's a PalmDOC file.
______
Dennis

[wallcraft]

Originally Posted by DaleDe

TEXt/REAd should not contain html. Any application doing this should have a different type. This is the PalmDOC format and should never contain any extra stuff or many PalmDOC readers would be broken. Have you seen an example of this and do you know what application created it?

MobiPocket (incorrectly) used TEXt/REAd for its initial PRC format, and I'm pretty sure it contained HTML as well as images.

[tompe]

Originally Posted by nrapallo

That's it!

Is there a way to use '--rawhtml' and NOT print $text to STDOUT, but rather re-direct (re-open) it internally for the my substitution code to work.

I think what is happening here is that printf to STDOUT in binmode 'actually' generates the para boundaries I want to use. They don't seem to be there when the text is first used i.e my $text = $pdb->text;

Just save it in a variable. "my $rawhtml = $text;" and do it outside the $opt_rawhtml if statement.

binmode is used because Windows add chacacters to the output without it. So the filepos pointer will not be correct so I cannot debug without binmode.

[DaleDe]

Originally Posted by DMcCunney

Well, the PDB extension will tell you it's a Palm Database. But the content of a Palm Database file can be arbitrary. You need to look at the header record of the file for the REAd creator ID and TEXt type to be sure it's a PalmDOC file.
______
Dennis

Those two together will probably identify it ok most of the time although I have seen some real PalmDOC files with .prc extensions. It is terrible when software doesn't obey the rules like BD.

Dale

[DMcCunney]

Originally Posted by DaleDe

Those two together will probably identify it ok most of the time although I have seen some real PalmDOC files with .prc extensions. It is terrible when software doesn't obey the rules like BD.

Ignore the extension and look at the header record in the database.

While all Palm OS programs are PRC files, all PRC files are not programs. Consider Mobipocket, who uses the PRC extension for ebooks on Palm devices. (The Mobi reader doesn't care about the extension, and will find and locate Mobi books with a PDB extension.)

I believe someone once commented that Mobi actually did use the "resource" database format, but I don't recall why.
______
Dennis

[nrapallo]

Originally Posted by tompe

Just save it in a variable. "my $rawhtml = $text;" and do it outside the $opt_rawhtml if statement.

binmode is used because Windows add chacacters to the output without it. So the filepos pointer will not be correct so I cannot debug without binmode.

After 'mobi2html' fixes all the filepos pointers, what if I wanted (need) those Windows characters in the STDOUT output? Could I simulate it using the variable $rawhtml without printf $text?

What is the [char] code that is used in the .pdb (text) for para breaks? It is '\n' in the rawhtml text output to STDOUT.

[tompe]

Originally Posted by nrapallo

After 'mobi2html' fixes all the filepos pointers, what if I wanted (need) those Windows characters in the STDOUT output? Could I simulate it using the variable $rawhtml without printf $text?

What is the [char] code that is used in the .pdb (text) for para breaks? It is '\n' in the rawhtml text output to STDOUT.

You have misunderstood how it works. The construction to a HTML tree will remove "\n". So the $text variable will contain what you want which is the fixed HTML code (filepos fixed) with "\n" in it. If you just print that string in Windows the "\n" will be converted to the Windows line break convention with two characters.

I assume it is "\n" or "\n\n" that is used for paragraph break. But that is something that must be defined in the original format.

But I do not understan how you can have filepos pointers in something that is not an HTML file.

[nrapallo]

Originally Posted by tompe

But I do not understan how you can have filepos pointers in something that is not an HTML file.

You're right, the .pdb text has no filepos pointer; I used that term loosely and 'out of context'.

I will go back and study the 'mobi' part of the code to better understand how $pdb->text gets loaded/created.

[DaleDe]

Originally Posted by DMcCunney

Ignore the extension and look at the header record in the database.

While all Palm OS programs are PRC files, all PRC files are not programs. Consider Mobipocket, who uses the PRC extension for ebooks on Palm devices. (The Mobi reader doesn't care about the extension, and will find and locate Mobi books with a PDB extension.)

I believe someone once commented that Mobi actually did use the "resource" database format, but I don't recall why.
______
Dennis

I guess you jumped in here without reading the context. The problem we are trying to solve is that a program called Book Designer actually codes the PalmDOC type record in the database into actual MobiPocket files. It is a bad error but we are looking for a work around. We already know how they are supposed to work.

Dale

[HarryT]

Originally Posted by DaleDe

I guess you jumped in here without reading the context. The problem we are trying to solve is that a program called Book Designer actually codes the PalmDOC type record in the database into actual MobiPocket files. It is a bad error but we are looking for a work around. We already know how they are supposed to work.

Dale

The best "workaround" is not to use BD to create MobiPocket books. What I do myself these days is to use BD to create the book and the Sony Reader version of it, then export from BD in HTML format and use "MobiPocket Creator" to create the Mobi-format book. I've written a tutorial describing exactly how to do this in the "Mobi Book Uploads" forum section.

[Ortep]

I'm experimenting with mobi2html. Most of the time it looks fine, but sometimes I get really ugly results. The punctuation is replaced by something that looks like high ascii. Is it something I'm doing wrong?

An example:

           “Sorry, no change, miss,� the driver said. The girl stared at him, at the dollar, and the coinbox for another moment, and now the driver was beginning to get impatient. “Miss, are you going to—�

When I look at the original is is formatted like this:

___________“Sorry, no change, miss,” the driver said.__The girl stared at him, at the dollar, and the coinbox for another moment, and now the driver was beginning to get impatient.__“Miss, are you going to—”

(Multiple spaces are replaced by under scores by me because they do not show here)

When I recreate the mobi file the text contains the "           “" and not the correct punctuation

[Gudy]

That pretty much is the correct punctuation, albeit in UTF-8 encoding. Is the content encoding set correctly in the html file? There should be a line like the following somewhere near the beginning of the file:

<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />

If there isn't, it should be considered a bug in mobi2html. If there is, either get a better browser or a tool to convert UTF-8 into something more palatable, like e.g. numeric html entities.

[Ortep]

Originally Posted by Gudy

That pretty much is the correct punctuation, albeit in UTF-8 encoding. Is the content encoding set correctly in the html file? There should be a line like the following somewhere near the beginning of the file:

<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />

If there isn't, it should be considered a bug in mobi2html. If there is, either get a better browser or a tool to convert UTF-8 into something more palatable, like e.g. numeric html entities.

The line isn't there. Inserting it does not help. And can you recommend a better browser? IE6, IE7 and Firefox give exactly the same output. But also Mobicreator does the same when I recreate the mobi file. Even when I force it to use UTF-8. So I guess I'll have to wait what tompe has it say

[Gudy]

Originally Posted by Ortep

Inserting it does not help. And can you recommend a better browser? IE6, IE7 and Firefox give exactly the same output.

Hmm, I just checked, and Mozilla, which is essentially the same as Firefox, does the right thing, i.e. it displays crap without that line, and correct punctuation with it. Without it, your browser will use whatever is the default encoding for your platform and locale, which is bound to be CP1250 or similar under Windows and ISO-8859-1 or so under Linux.

I don't know how much HTML you know, so forgive me for perhaps stating the bleeding obvious, but the line must be placed between the <head> and </head> elements in the HTML, It won't work anywhere else. Directly behind the <head> is usually best, just look at the source html for this site for an example (Ctrl-U in Mozilla, I don't know what Firefox uses).