HTML import: Making span optional

[voidray]

Currently elements have to be there for html to be loaded (otherwise there is an error). Would it make sense, to also allow formatting with <b>, <i>, <u>, <sup>, <sub> etc. when loading html? Saving to html would than use the span-tag (css), so it is only about compatibility when loading html.
I created a branch/commit in my fork with that change:
voidray@7a67e5a

What isn't adressed in this commit:

• Whitespace: For example LibreOffice Writer adds line breaks (not <br>) when a document is saved as html. A browser would remove the line breaks and add a space instead, but the control would still have the line break.
• Hyperlinks <a>: This should be adressed for all supported formats together (rtf, xaml, html). A first step could be supporting loading and saving. A second step could be adding a create/modify UI and a shortcut (CTRL+K for example). The ui could be like in other applications the LibreOffice (a window with Diplay name und actual hyperlink).

[cuikp]

I put together the html conversions pretty quickly, so while it does save/load attributes created by the RichTextBox, it does a poor job of reading in existing .html docs.
So, any additional support for reading html is appreciated. I'll create the pull request from your code and merge then.

For supporting hyperlinks (for all formats), that will be another matter to work on, but I'm not sure how far this RichTextBox control should mirror apps like LibreOffice. WPF's RTB doesn't support hyperlinks out of the box, but then again, that doesn't mean AvRichTextBox can't lol.

[cuikp]

I just discovered an additional issue with html - where there is no content (it goes through child nodes and finds nothing). That make the document entirely empty (no paragraphs), which is a no-no for RichTextBox, so next push will include a fail-safe if any document is imported without any content.

[voidray]

Can you take a look at this commit here voidray@8871e3e ?
I tested newline handling for rtf/xaml/html and for html I found the issue that the control handles <p></p> as new line, wheras LibreOffice Writer or Browsers don't. When saving as html with LibreOffice Writer empty newlines are saved as <p><br/></p>
The commit changes loading / saving HTML, so that the resulting html also looks the same in Browsers or when loading with Writer.

To reproduce the issue:
Create this text in the control, save it as html, and open it in a browser:

Line 1

Line 2

-> Collapsed line break (or in Libreoffice Writer: empty line missing)

Then create the same txt in for example LibreOffice Writer, and save as HTML. Load the html-file in the control. -> additional linebreak

[cuikp]

Seems to be ok, is there a PR for this?

[voidray]

Seems to be ok, is there a PR for this?

#35

[cuikp]

Merged.
After testing import of a more complex html file, it wasn't reading anything at all because it was so nested. So the reading has to be recursive. I added changes for this but want to make sure it meshes with your previous changes to HtmlToContent (particularly as concerns
). I also remove white-space from inner text and ignore
outside of paragraphs (where they shouldn't be for AvRichTextBox).

main...html-recursive-read

[cuikp]

Regarding hyperlinks: I extensively reworked the newly updated github to include an EditableHyperlink class. It can be added to a FlowDoc by new EditableHyperlink("link text", "www.link.com")

public EditableHyperlink(string displayText, string navigateUri) 
  {  
     Id = ++FlowDocument.InlineIdCounter; 
     Text = displayText;
     NavigateUri = navigateUri;

     //force hyperlink visual formatting
     this.Foreground = displayBrush;
     this.TextDecorations = displayDecoration;
  }

The navigation keys logic and inline context awareness had to be significantly revised to allow navigation around hyperlinks (prevent their text from being partially selected), but I think it's correct now. In addition, partial mouse selection is prevented, and Ctrl-Click opens the link.

@voidray
Next would be to support reading/writing of hyperlinks to the supported formats as you suggested. If you are keen to look into rtf/html I can work on the Xaml implementation.

[voidray]

I think it would be better if the link can be selected partially, because this also for example enables the user to change the hyperlink display-text. This is also like Word or Writer work. If the caret is at the border of the hyperlink, new text should not be part of the link. If the link needs to be changed at the start or end, then a editing dialog / overlay is needed for link and display text.

[cuikp]

Yes it appears Word allows inline editing of hyperlink text unless you have mixed selection (normal text + part of hyperlink).

Not too hard to implement I think, just have to allow entering the hyperlink if (Selection.Length == 0).

[cuikp]

latest Push has navigation more similar to what you describe.
(can edit text within hyperlink, but trying to straddle selection across hyperlink and surrounding text will select full hyperlink.
Also, you can delete up to the last character of the hyperlink, then it blocks to preserve it.

[voidray]

I just realised that the selection behavior is different between Word and Writer. For me it's fine to to have the behavior like it is in Word, I was just used to Writer.
One thing I would change is I would make it possible to click inside the link, so the caret goes to the clicked position. Currently the whole link is selected when clicked. The hand symbol should then only be displayed when holding the CTRL key.

[cuikp]

In latest Push i added a CtrlKeyOpensHyperlink bool property to the RichTextBox. So user can have it either way.

Also, hyperlinks now saved/loaded from Xaml (or XamlPackage).

[voidray]

That looks great, thanks. I can take a look at loading / saving hyperlinks with html next.

[voidray]

please take a look at the PR #38