Ability to replace empty <p> with <br> during html2mark conversion

[dmitrymurashenkov]

This is a typical case when converting html from rich text editor like TinyMCE to markdown. User presses enter key 2 times and gets 2 paragraphs in editor which visually look like several newlines in html:

<p>1</p>
<p></p>
<p>2</p>

We expect it to be processed similar to:

1<br><br><br>2

which gives:

1

<br />

2

but actual result is:

1

2

Note that in real case the empty <p> tag looks like this:

<p>&nbsp;</p>

So strictly it is not empty, but is still omitted.

Options used:

MutableDataSet options = new MutableDataSet();
options.set(FlexmarkHtmlParser.OUTPUT_ATTRIBUTES_ID, false);
options.set(FlexmarkHtmlParser.THEMATIC_BREAK, "-------------------");
options.set(FlexmarkHtmlParser.SETEXT_HEADINGS, false);
options.set(FlexmarkHtmlParser.BR_AS_EXTRA_BLANK_LINES, true);
options.set(FlexmarkHtmlParser.BR_AS_PARA_BREAKS, true);
options.set(FlexmarkHtmlParser.DIV_AS_PARAGRAPH, true);
options.set(TableFormatOptions.FORMAT_TABLE_CAPTION, TableCaptionHandling.REMOVE);

[dmitrymurashenkov]

Found workaround for my particular case with nbsp in the empty paragraph:

options.set(FlexmarkHtmlParser.NBSP_TEXT, " ");

This gives:

1

 

2

Which is rendered as several blank lines too.

[vsch]

@dmitrymurashenkov, this is a bug of not implementing BR_AS_EXTRA_BLANK_LINES if the paragraph element is empty.

1

<br />

2
.
<p>1</p>
<p></p>
<p>2</p>

Release with the fix coming shortly.

[vsch]

Fix for this is available. Repo updated, maven updated but may take a while to show up in maven central.