Get only visible text from document.body

[Kevin Jones]

Hello,

I am interested in get all the text from document.body, but only visible text, ie, that which the user actually can read on the webpage.

I have experimented with textContent and TreeWalker/NodeFilter.SHOW_TEXT, but they include “text” within script tags and style tags, etc. One thing I haven’t tried would be to clone the node and getElementsByTagName(“noscript/script/syle”) and set innerHTML = “” on those nodes. But even if it worked it seems it would be inefficient and more prone to bugs (ie, are there other tags which contain text that does not show up on the screen?).

I was wondering if Firefox has an API which would accomplish this. Obviously the browser has already done this work, so I wondered if there is an API which exposes that text. This would be far more efficient I expect, and more reliable. I have not found anything in this vein in my searches.

Or in lieu of that, any other suggestions which would efficiently accomplish this goal.

Thank you,
Allasso

[patrickjdempsey]

Text.wholeText?
https://developer.mozilla.org/en-US/doc ... /wholeText

[Kevin Jones]

Text.wholeText?
https://developer.mozilla.org/en-US/doc ... /wholeText

No, that returns all that other junk as well. Thanks for the suggestion.

[patrickjdempsey]

I wonder if there's a way to simulate "Select All, Copy"? That would do exactly what you want, and I think even preserve links.

[Kevin Jones]

Wow, I just saw this (Mozillazine does not always send update emails), and that is exactly what I did:

Code: Select all

    _getSelectionController = function(aWindow) {
    
      // 'display: none' iframes don't have a selection controller, see bug 493658
      if (!aWindow.innerWidth || !aWindow.innerHeight)
        return null;

      // Yuck. See bug 138068.
      var Ci = Components.interfaces;
      var docShell = aWindow.QueryInterface(Ci.nsIInterfaceRequestor)
                            .getInterface(Ci.nsIWebNavigation)
                            .QueryInterface(Ci.nsIDocShell);

      var controller = docShell.QueryInterface(Ci.nsIInterfaceRequestor)
                               .getInterface(Ci.nsISelectionDisplay)
                               .QueryInterface(Ci.nsISelectionController);
      return controller;
      
    }

    let win = <content window or iframe for whatever you want to select>
    let controller = this._getSelectionController(win);

    controller.selectAll();
    let selection = controller.getSelection(controller.SELECTION_NORMAL); // controller.SELECTION_NORMAL === 1
    let selectionString = selection.toString();
    selection.removeAllRanges();  // clear the selection
    selectionString = selectionString.replace(/\s/g," ").replace(/  +/g," ");  // this may not be necessary
 

Caveat is that it will also select text for accessibility items, such as the 'alt' attribute in 'img' tags. This may or may not be desired.

This code and much more can be found in Finder.jsm.

Thanks, Patrick.