MobileRead Forums - View Single Post

DMSmillie · 03-08-2011, 10:42 AM

Hi overholt and welcome to MobileRead.

Not sure how much or how little you know about the insides of an EPUB file, so apologies if the following is stuff you already know.

An EPUB file is basically just a ZIP file with the file extension changed from .zip to .epub. The book content is in one or more HTML files inside the EPUB file (there are a few other files, in XML, that provide information about the book, about the file structure inside the EPUB, and specifying a logical table of contents and navigation structure for the book). As ATDrake says, the first three errors are validation errors in the HTML code in the book content files. The "dir" attribute referred to in the fourth error is an HTML attribute used to specify if the language being used in the content is written from left to right (such as most European languages) or right to left (such as many of the languages used in the Middle East and Asia). There are only two allowed values for this attribute ("ltr" or "rtl") so presumably the error indicates that it appears somewhere in the HTML code with some other value.

These problems aren't anything to do with the numbers of heading levels or hyperlinks in your original Word doc. The problem is the conversion route you're using to get from Word to EPUB. In the first step, you're using MobiPocket Creator to convert the Word doc to MOBI format (the .prc file). Then you're converting the MOBI file to an EPUB file. The problem with this sequence, in terms of producing an EPUB that will pass Epubcheck, is that the MOBI format doesn't support or use the same level of HTML as that required for an EPUB. So you're going from a feature-rich format (Word) to a relatively simple and not HTML standards compliant format (MOBI/prc) then trying to step back up to a feature-rich format which has to be HTML standards compliant in order to pass Epubcheck. The EPUB will probably work OK, so if it's just for personal use, you might decide that you can live with it not passing Epubcheck. However it won't be acceptable for commercial publication (e.g. with Apple in the iBookstore).

Solutions?

Something you could try would be to open the EPUB in Sigil, with the HTML Tidy option switched on (the default setting), then re-save the EPUB. That might be all that's needed to sort out the HTML errors that are currently being flagged.

Longer term, though, I'd strongly recommend changing the process you're using to arrive at an EPUB version - going from Word to MOBI then to EPUB means you'll almost certainly lose some of the formatting and layout that you've created in Word, since MOBI supports a lot less in terms of formatting richness than EPUB does.

An alternative might be to save your finished Word doc as "Web Page, Filtered" to create an HTML version of your doc, then import that into Sigil and use it to create your EPUB version. If you then also want a MOBI version, I'd recommend using Amazon's Kindlegen app to convert the EPUB file to a MOBI file.

Hope that helps a little!

03-08-2011, 10:42 AM	#3
DMSmillie Enquiring Mind Posts: 562 Karma: 42350 Join Date: Aug 2010 Location: London, UK Device: Kindle 3 (WiFi)	Hi overholt and welcome to MobileRead. Not sure how much or how little you know about the insides of an EPUB file, so apologies if the following is stuff you already know. An EPUB file is basically just a ZIP file with the file extension changed from .zip to .epub. The book content is in one or more HTML files inside the EPUB file (there are a few other files, in XML, that provide information about the book, about the file structure inside the EPUB, and specifying a logical table of contents and navigation structure for the book). As ATDrake says, the first three errors are validation errors in the HTML code in the book content files. The "dir" attribute referred to in the fourth error is an HTML attribute used to specify if the language being used in the content is written from left to right (such as most European languages) or right to left (such as many of the languages used in the Middle East and Asia). There are only two allowed values for this attribute ("ltr" or "rtl") so presumably the error indicates that it appears somewhere in the HTML code with some other value. These problems aren't anything to do with the numbers of heading levels or hyperlinks in your original Word doc. The problem is the conversion route you're using to get from Word to EPUB. In the first step, you're using MobiPocket Creator to convert the Word doc to MOBI format (the .prc file). Then you're converting the MOBI file to an EPUB file. The problem with this sequence, in terms of producing an EPUB that will pass Epubcheck, is that the MOBI format doesn't support or use the same level of HTML as that required for an EPUB. So you're going from a feature-rich format (Word) to a relatively simple and not HTML standards compliant format (MOBI/prc) then trying to step back up to a feature-rich format which has to be HTML standards compliant in order to pass Epubcheck. The EPUB will probably work OK, so if it's just for personal use, you might decide that you can live with it not passing Epubcheck. However it won't be acceptable for commercial publication (e.g. with Apple in the iBookstore). Solutions? Something you could try would be to open the EPUB in Sigil, with the HTML Tidy option switched on (the default setting), then re-save the EPUB. That might be all that's needed to sort out the HTML errors that are currently being flagged. Longer term, though, I'd strongly recommend changing the process you're using to arrive at an EPUB version - going from Word to MOBI then to EPUB means you'll almost certainly lose some of the formatting and layout that you've created in Word, since MOBI supports a lot less in terms of formatting richness than EPUB does. An alternative might be to save your finished Word doc as "Web Page, Filtered" to create an HTML version of your doc, then import that into Sigil and use it to create your EPUB version. If you then also want a MOBI version, I'd recommend using Amazon's Kindlegen app to convert the EPUB file to a MOBI file. Hope that helps a little!