Fixing tagging automatically with the accessibility tool causes document changes

[showalters]

I have run into this issue several times now. A user will send me a document to post on our website, I'll run the accessibility checker and then go in and attempt to fix the issues. The first issue is Tagged PDF- Failed. After I run the fix and move to the next issue (usually Primary language failed or Title failed) the document will change. I have had documents where letters or whole words disappear, letters or numbers appear at the end of every line, background colors come to the foreground and either cover text/images or make the text/image disappear totally.

After doing a lot of testing, we have identified what may be causing the issue. We have 35 different departments and no standard for documents, so some people when creating PDFs use Save as PDF with Microsoft Office or another PDF maker doing the conversion from its native format to PDF. Some people are using Nitro Pro. And some are using Acrobat to convert the documents. It seems that unless you use Acrobat Pro and choose the Make Accessible option when converting, it can cause (but not always) the issues I described.

We are attempting to work with our IT department to ensure that everyone who has to convert documents to PDFs for our website, will have access to Acrobat Pro, and if they don't we are asking them to send us the orignal Word document so that we can do it ourselves correctly. However, we have some programs that spit out documents already in PDF format that are not ADA compliant, and that is where there is a big issue. I have included a document (09-28-20 ZHM Agenda) that is from that kind of program. If you run the accessibility checker, then fix the tagging, then fix the language, if you scroll to another page, then back you will see that various letters and numbers have disappeared from the document. As there is no "orignal Word document", I can't fix this issue. The only think I have found to do is to Export the document to a Word Document, HOPE that the formatting stays the same, then convert it back with Acrobat using the Make Accessible option.

My question is why is Acrobat changing the document after it tags it? Why is it deleting (hiding?) letters and/or numbers, adding characters, bringing background colors forward to hide text/images? And why is it only doing it to some documents and not others? The same person who provided the agenda attached, provided one for the next day, using the same program, same format, but Acrobat didn't change the document at all. I am attaching that one as well. (09-29-20 ZHM Agenda)

[Dave__M]

I suspect the reason all those changes happen is due to the way those PDFs were created.  If the PDF is image based, or partially image based (both text and images of text), tagging becomes a major operation, for either the software and then the user who has to remediate the mess.  And two PDFs can look identical to the eye, and be structurally different as night and day.  In my experience, starting with a Word doc, and using the PDF Maker (Acrobat ribbon) will get you the closest to a compliant PDF.  Even then, the Word doc has to be a well-formed document.  Some people type, some Word Process with styles and headings. The latter will yield a more compliant document, or at least closer.

My best,

Dave

[showalters]

I totally agree it is the way that the PDFs are created, but for the two that I attached, they were both created from the same program (an agenda maker program) and not created in Word. The progran spits out the PDF. Why would one have issues and one not? The only "images" on either documents are our County logo and the wheelchair icon, everything else is text. Both of the ones I included are the original as presented to me, before I ran the accessibilty checker and fixed the issues.

As I said, we are working with our IT department so that people turning in documents for our website will have access to the Acrobat PDF maker, but that doesn't help when it is not Word, PowerPoint or Excel, and it is a program that the Acrobat PDF maker can't be integrated into. How do I fix those when autotagging "breaks" them?

[Dave__M]

You definitely have some peculiarities happening here.  I opened up both PDF files.  Both have 'selectable text'.  But on one, I can perform a Find command, and yield appropriate results.  The other... it doesn't behave as it should!  See below:

When I have the time, I'll see if I can figure out what's going on {in the throws of a project currently}. The Find command certainly should have found a text match!

My best,

Dave

[Prim-A11y]

I'm having a similar issue when trying to remediate an academic article. Everything seems to tag somewhat properly, except elements of the graphs disappear after running Make Accessible, particularly shading on bars. I initially thought they were being hidden by reading order changing the structure, as I've run into that a lot with InDesign documents, but trying Send to Back or Bring to Front doesn't seem to change anything. It seems as though entire graph elements are being removed, or at least hidden in a way that I can't access.

[Bevi Chagnon - PubCom.com]

I don't know why the Make Accessible is causing elements to either disappear or become hidden. I'm suspecting the latter.

Did you examine the Content pane and see if the elements are there or not? Most likely the missing elements are behind another element, and the usual front/back utility doesn't quite drill down deep enough to fix them.

[Prim-A11y]

I did check the content panel, and it looked like there were two elements for each bar, but I couldn't find a way to regain the shading. The solution I've been using has been to edit the visible items in Illustrator and add the shading back manually.

[Bevi Chagnon - PubCom.com]

Just like with tags in the Tag Tree, you can slide elements in the Content up and down in its tree.

I'd try dragging the bars up higher in the Contents tree until they become visible again.

I do not recommend bringing the graphics back into Illustrator because PDFs, once they've been tagged, can't be edited. It's really destroy the tagged content beyond repair.

Q: what was the source program from which the PDF was exported? Source platform?

[Prim-A11y]

I don't know or have access to the source of the document, and changing the order in Content doesn't change anything visually.

How exactly does it destroy the tagged content? It seems as though everything is still contained within the Figure tags they were in before.