NVDA does not support line breaks in PDF files


Perry Simm
 

Hi!
When reading a PDF with NVDA in Adobe Reader, I find that NVDA does not honor line breaks. To illustrate, let's say part of the PDF looks like this:
[snip]
Line one
Line two
Line three
[/snip]
When reading this PDF with NVDA, I would get:
[snip]
Line one Line two Line three
[/snip]
Such is the basic nature of the problem I am facing. Now for some auxiliary observations:
1. When exporting the PDF as a text file, the text file ends up having correct line breaks.
2. When reading the PDF with JAWS, line breaks are correctly presented.
3. The problem occurs on three machines, two of them running Windows 10, the third running Windows 7, with both NVDA 2019.1 and current beta of 2019.2.
4. The problem persists after uninstalling and then reinstalling Adobe Reader.
5. The problem persists after repairing the Adobe Reader installation.
Since Adobe Reader itself exports the line breaks correctly, and since JAWS is also seeing them as expected, the problem is clearly with NVDA.
Cheers Perry

Sent with ProtonMail Secure Email.


David Griffith
 

I cannot answer this definitively but I have encountered similar problem on web pages and the solution then was to change some of the document layout properties of NVDA using NVDA Control and D.

Unfortunately I have gone back to a web page and examined the document layout properties pane and I cannot immediately identify which was the setting I changed at the  time.

Someone else may remember what property I am thinking of.

I hope I am not sending people off on a wild goose chase.

David Griffith.

 

 

 

Sent from Mail for Windows 10

 

From: Perry Simm via Groups.Io
Sent: 08 July 2019 10:00
To: nvda@nvda.groups.io
Subject: [nvda] NVDA does not support line breaks in PDF files

 

Hi!

When reading a PDF with NVDA in Adobe Reader, I find that NVDA does not honor line breaks. To illustrate, let's say part of the PDF looks like this:

[snip]

Line one

Line two

Line three

[/snip]

When reading this PDF with NVDA, I would get:

[snip]

Line one Line two Line three

[/snip]

Such is the basic nature of the problem I am facing. Now for some auxiliary observations:

1. When exporting the PDF as a text file, the text file ends up having correct line breaks.

2. When reading the PDF with JAWS, line breaks are correctly presented.

3. The problem occurs on three machines, two of them running Windows 10, the third running Windows 7, with both NVDA 2019.1 and current beta of 2019.2.

4. The problem persists after uninstalling and then reinstalling Adobe Reader.

5. The problem persists after repairing the Adobe Reader installation.

Since Adobe Reader itself exports the line breaks correctly, and since JAWS is also seeing them as expected, the problem is clearly with NVDA.

Cheers Perry

 

Sent with ProtonMail Secure Email.

 

 


Perry Simm
 

Hi David,
On Monday, July 8, 2019 11:51 AM, David Griffith <daj.griffith@...> wrote:

I hope I am not sending people off on a wild goose chase.

Don't you worry, sometimes a wild goose chase is a good start, and the more people are thinking about a problem, the more likely a solution will emerge. And I really hope it does emerge soon. This thing is seriously broken.
Cheers Perry


Gene
 

You are probably thinking of the number of characters per line setting.  Web pages generally don't have line breaks except where the writer specifically puts one in the text.  But browsers determine the length of lines in general by the size of the window from left to right.  In browse mode, the line length is determined by the screen-reader and I don't think has any relation to the size of the reader.  So you have the ability to set the lines to your desired length. 
 
But PDF documents may have regularly occurring line breaks.  I don't know.  
NVDA uses browse mode and evidently doesn't understand the character used by PDF documents for line breaks.  You can change the length of lines at present by working with the characters per line setting.  This won't follow actual line breaks but may make the document more as you want it.
 
Gene

----- Original Message -----
Sent: Monday, July 08, 2019 4:51 AM
Subject: Re: [nvda] NVDA does not support line breaks in PDF files

I cannot answer this definitively but I have encountered similar problem on web pages and the solution then was to change some of the document layout properties of NVDA using NVDA Control and D.

Unfortunately I have gone back to a web page and examined the document layout properties pane and I cannot immediately identify which was the setting I changed at the  time.

Someone else may remember what property I am thinking of.

I hope I am not sending people off on a wild goose chase.

David Griffith.

 

 

 

Sent from Mail for Windows 10

 

From: Perry Simm via Groups.Io
Sent: 08 July 2019 10:00
To: nvda@nvda.groups.io
Subject: [nvda] NVDA does not support line breaks in PDF files

 

Hi!

When reading a PDF with NVDA in Adobe Reader, I find that NVDA does not honor line breaks. To illustrate, let's say part of the PDF looks like this:

[snip]

Line one

Line two

Line three

[/snip]

When reading this PDF with NVDA, I would get:

[snip]

Line one Line two Line three

[/snip]

Such is the basic nature of the problem I am facing. Now for some auxiliary observations:

1. When exporting the PDF as a text file, the text file ends up having correct line breaks.

2. When reading the PDF with JAWS, line breaks are correctly presented.

3. The problem occurs on three machines, two of them running Windows 10, the third running Windows 7, with both NVDA 2019.1 and current beta of 2019.2.

4. The problem persists after uninstalling and then reinstalling Adobe Reader.

5. The problem persists after repairing the Adobe Reader installation.

Since Adobe Reader itself exports the line breaks correctly, and since JAWS is also seeing them as expected, the problem is clearly with NVDA.

Cheers Perry

 

Sent with ProtonMail Secure Email.

 

 


Perry Simm
 

Hi!
I think I figured out what's going on. In issue #7275 it was discussed that Adobe Reader doesn't let NVDA differentiate between the two kinds of line breaks, namely, author-generated line breaks such as in poetry or source code, and line breaks created by Adobe Reader automatically to make the text fit within the window. So it was decided to have NVDA simply strip all line breaks from PDF content.
Technically, NVDA is doing the right thing, and the error lies with incorrectly authored PDF files. However, since these incorrectly authored files abound on the web, I strongly urge developers to prioritize usability above technological perfectionism in this case. JAWS went down this route, and succeeded. This doesn't make this route the correct one always, but this time I believe it would do much more good than harm.
A possible compromise would be to make this behavior dependent upon the NVDA option to honor screen layout in browse mode. Why not strip line breaks when screen layout is turned off, and honor them when it is on?
I fully understand that NVDA is currently doing the technically correct thing. However, in this case it breaks the user experience and puts NVDA at a clear disadvantage compared to its competitor, a disadvantage which, in its conquest to be accepted in workplace environments, it clearly cannot afford.
Cheers Perry


Jonathan COHN
 

Because PDF is an evolving format that was initially just used to provide displaying of printed documents, but now including PDF-UA 2 tagging that provides appropriate tagging for distinguishing headings, table titles, list structure etc, often one PDF file can be used appropriately by screen readers while a similar looking file can not. First question to consider:
Were you prompted to use auto tagging when viewing the document? If so, then this means the PDF creator did not create a fully accessible version of the document. One can often find discussions about this on accessibility support discussions.
I currently don't know how to determine if the tagging of the PDF meets a specific PDF-UA standard.
HTH,

Jonathan