Headings going awry

[MikeMJ]

I am trying to convert a number of Microsoft Word documents (.doc or .docx files) to e-books. I have tried saving the documents as .RTF files, and also as "filtered HTML" files. With either format, converting to EPUB (using Calibre 0.8.45) generates a readable file, but with significant errors.

The errors seem to be related to changes that are applied to the major headings. In the original HTML file, a heading looks like:

<h1><a name="_Toc244191671"></a>New York Bumper Stickers</h1>

However, in the HTML file in the Debug\Parsed folder, the heading has been changed to:

<h1><a name="&lt;i&gt;Toc244191671”&gt;&lt;/a&gt;New York Bumper Stickers&lt;/h1&gt;

As a result:

• The paragraphs that follow the <h1> are appearing in the large font of the <h1>
• The reference from the <a> is appearing as text
• The text of the <h1> is not visible

In other cases, a very similar heading changes from:

<h1><a name="_Toc244191667"></a><a name="_Toc104266959">You know you’re from </a>Jersey when …</h1>

to:

<h1 style="margin-top:1em;margin-bottom:1em;"><a name="&lt;i&gt;Toc244191667”&gt;&lt;/a&gt;&lt;a name=">Toc104266959”&gt;You know you’re from </a>Jersey when …</h1>

In this case, the text of the anchor is visible, but, because the </h1> is not corrupted, the subsequent text is properly formatted.

Can anybody tell me why some of the HTML tags are being corrupted? Also, why is the text:

&lt;i&gt;

being inserted before the text of the anchor name attribute?

[JSWolf]

My suggestion is to just delete the <a name="_Toc244191671"> type code in the headers. You don't actually need it.

[dwig]

Originally Posted by MikeMJ

I am trying to...
"_Toc...

the heading has been changed to:
"&lt;i&gt;Toc...

Can anybody tell me why some of the HTML tags are being corrupted? Also, why is the text:

"&lt;i&gt;

being inserted before the text of the anchor name attribute?

Quite obviously, the underscore preceding the "Toc" is giving calibre grief. Calibre seems to be incorrectly interpreting it as the antique (still frequently seen) use of underscores wrapping a portion of text to indicate italics. Calibre is replacing the underscore with the <i> tag. Later processing is then converting the < to &lt and > to &gt.

One fix is to sweep the document with S&R to remove the underscore (e.g. replace "_Toc" with "Toc") before conversion.

[kovidgoyal]

Turn off heuristics in the conversion settings.

[JSWolf]

I do prefer my solution to actually remove the <a> from the code. It's not needed and just adds bloat.

[theducks]

Originally Posted by JSWolf

I do prefer my solution to actually remove the <a> from the code. It's not needed and just adds bloat.

I believe you need Anchors

IF

You don't start a new file at chapter boundries

By default, TOF is the anchor point.

[MikeMJ]

I knew there had to be a simple answer. Turning off Heuristics solved the problem -- without any editing.

Thank you, Kovid!

[JSWolf]

But do you really want to leave in code you don't actually need?