Specific CFI Problems via the XPATH reference

[MaxDaten]

Hi,

I'm currently debugging a problem with a specific CFI generated by me.

First the relevant HTML section (not the complete HTML document!)

      <h2 id="h2-5" title="Kapitel 5"><a id="page_39"></a><span class="small">01100010110010</span> <b>5.</b> <span class="small">10010111010100</span></h2>

And now two CFIs. With the first CFI I can successfully reference theh2 tag in the document but with the second not the first whitespace before the first opening b tag. The corresponding XPath is inspected here

epub.js/src/epubcfi.js

Line 511 in d8ba903

startContainer = doc.evaluate(xpath, doc, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null).singleNodeValue;

• Working
  • CFI "epubcfi(/6/18!/4/352[h2-5]"
  • XPath: "./*/*[2]/*[position()=176 and @id='h2-5']/*[1]
• NOT working
  • CFI: "epubcfi(/6/18!/4/352[h2-5]/5:0)"
  • XPath: "./*/*[2]/*[position()=176 and @id='h2-5']/text()[2]"

I narrowed down the problem to a mismatch in the CFI and XPath spec. In CFI empty character data chunks are still indexed:

Consecutive (_potentially-empty_) chunks of character data are each assigned odd indices (i.e., starting at 1, followed by 3, etc.).

http://www.idpf.org/epub/linking/cfi/epub-cfi.html#sec-path-child-ref

In XPATH only character data with at least one character is grouped into a text node.

Character data is grouped into text nodes. As much character data as possible is grouped into each text node: a text node never has an immediately following or preceding sibling that is a text node. The string-value of a text node is the character data. _A text node always has at least one character of data._

https://www.w3.org/TR/xpath/#section-Text-Nodes

Now the generated XPath (

epub.js/src/epubcfi.js

Line 448 in d8ba903

EPUBJS.EpubCFI.prototype.generateXpathFromSteps = function(steps) {

) can reference text nodes that are not possible to reference.

Maybe a fallback to the parent node is a possible solution?

[fchasen]

Sorry this slipped me by, thanks so much for writing this up.

I'll need to look into this a bit more, but I agree that a good approach for now is fallback to the parent.

[khcpietro]

Is there any progress here? @MaxDaten pointed out exactly. CFI and Xpath treat empty chunk differently.

[khcpietro]

Here is my workaround.
You can append space chunk between tags after html content loaded.

document.body.innerHTML = document.body.innerHTML.replaceAll('><', '> <')

Sadly, this changes epub content itself, and can cause performance issue. However, If you more care about consistency between CFI and Xpath, you can try.