Outline enhancement request

[danjgill]

Hi,

I am trying to use outlines with epub files, and it is not working. Chapter and PageNumber always return as -1. Perhaps I am missing a step for setting them up, or perhaps they simply aren't meant to work with reflowable files?

As an alternative, I tried to use link files, but ran into a different problem. Here Chapter and PageNumber are correct, even when I change the font size and re-Layout the document. So good job there! However, the MuPDFLink object lacks a title, or some other method of finding something to put into a combobox. I might be able to figure out the text, but every approach I take seems quite hacky.

So is it possible to either A) get the correct Chapter and PageNumbers into the Outline (including updating them when layout changes) or B) Adding a "Title" property to the MuPDFLink object that captures the text of the link.

(A) would be preferable, as sometimes the links are very short (such as in an index).

Or maybe I am missing something obvious and there is some step I am missing. If so, any direction would be appreciated.

Thanks

[arklumpus]

Hi, thanks for pointing this out! It turns out that the outline items were correctly resolved to an URI like EPUB/xhtml/section0002.xhtml, but the next step (i.e., matching that to a page number) was missing.

I just pushed version 2.0.1, which should fix this issue; please have a look (the link destinations should also update automatically if you reflow the document with a different page/font size). This should address your part A.

For part B, the problem is that a link is not strictly associated to a text element on the page, so there is no such thing as the "text" of the link. However, a link has an ActiveArea, which represents the rectangle on which the user can click to activate it; if you really wanted to get the text associated with a link, you could use this in combination with a MuPDFStructuredTextPage (see below).

Code example

using MuPDFContext ctx = new MuPDFContext();
using MuPDFDocument doc = new MuPDFDocument(ctx, @"C:\Users\Giorgio\Downloads\LinkExamples.pdf");

using MuPDFStructuredTextPage structTextPage = doc.GetStructuredTextPage(0);

foreach (MuPDFLink link in doc.Pages[0].Links)
{
    // Middle-left point of the link area.
    PointF linkStart = new PointF(link.ActiveArea.X0, (link.ActiveArea.Y0 + link.ActiveArea.Y1) * 0.5f);

    // Middle-right point of the link area.
    PointF linkEnd = new PointF(link.ActiveArea.X1, (link.ActiveArea.Y0 + link.ActiveArea.Y1) * 0.5f);

    // Get the closest text character to the start and end of the link area.
    MuPDFStructuredTextAddress? start = structTextPage.GetHitAddress(linkStart, false);
    MuPDFStructuredTextAddress? end = structTextPage.GetHitAddress(linkEnd, false);

    string linkText;

    // If both of them exist...
    if (start != null && end != null)
    {
        // ... then we can get the link text.
        linkText = structTextPage.GetText(new MuPDFStructuredTextAddressSpan(start.Value, end));
    }
    else
    {
        // ... otherwise, there is no useful link text we can retrieve.
        linkText = "<no text>";
    }

    string destination = link.Destination.Type switch
    {
        MuPDFLinkDestination.DestinationType.External => (link.Destination as MuPDFExternalLinkDestination)!.Uri,
        MuPDFLinkDestination.DestinationType.Internal => $"Page {(link.Destination as MuPDFInternalLinkDestination)!.PageNumber}",
        _ => "<Unknown destination>"
    };

    Console.WriteLine("\"{0}\" links to \"{1}\"", linkText, destination);
}

Here is the example PDF: LinkExamples.pdf

However, note that this may not always work as expected, depending on how the link is implemented in the document (see examples below).

Link examples

• In example 1, the link is a single rectangle encompassing the whole link text, so everything is fine:

  

  This produces:

  "Link text" links to "https://github.com/arklumpus/MuPDFCore"
  

• At first sight, example 2 looks exactly the same, but in this case the text is drawn as a series of paths, rather than as actual text:

  

  As a result, the link text cannot be extracted (unless you use the OCR feature):

  "<no text>" links to "https://github.com/arklumpus/MuPDFCore"
  

• In example 3, each letter has its own link area:

  

  In this case, you get a bunch of different links (the last one comes from the underline):

  "L" links to "https://github.com/arklumpus/MuPDFCore"
  "i" links to "https://github.com/arklumpus/MuPDFCore"
  "in" links to "https://github.com/arklumpus/MuPDFCore"
  "nk " links to "https://github.com/arklumpus/MuPDFCore"
  "k" links to "https://github.com/arklumpus/MuPDFCore"
  " t" links to "https://github.com/arklumpus/MuPDFCore"
  "te" links to "https://github.com/arklumpus/MuPDFCore"
  "x" links to "https://github.com/arklumpus/MuPDFCore"
  "xt" links to "https://github.com/arklumpus/MuPDFCore"
  "Link text" links to "https://github.com/arklumpus/MuPDFCore"
  

• In example 4, the link is broken by a line end:

  

  In this case, you also get two different links:

  "Link" links to "https://github.com/arklumpus/MuPDFCore"
  "text" links to "https://github.com/arklumpus/MuPDFCore"
  

In any case, I hope this helps! Please let me know if you have any more problems or questions!

[danjgill]

Hi there,

I just tried out the Outline fix and can confirm that it works. Thanks for making a great resource even better!

I think I agree on the discussion of the Links. While it is feasible to retrieve the text, it may have too many corner cases and exceptions to make it worthwhile. I do appreciate the code above however, as it may come in handy in the future for better understanding how both StructuredTextPages and Links work.

Thanks again for the speedy resolution.