Re: Needed tool for vision-impaired - was [Re: PDF Editor for Debian]

2024-06-24 Thread Richard
I wouldn't say PDFs are bad for visually impaired users. In fact, as bitmap
fonts are thankfully a thing of the past for almost everywhere, you can
zoom any document to your hearts desire. Though sometimes you need some
tricks, e.g. Evince is configured to only use 50 MB of storage by default
for caching, vastly limiting zoom capabilities. So you'll have to dig into
dconf to change that.

What you are looking for is ways to reflow text, but as a fixed layout
format, PDFs are just not meant for that. Not even the PDF/UA standard [1]
does require this, it only lays the ground rules for screen readers.
Supposedly the swiss-made "VIP PDF-Reader" was able to help, yet it seems
to have been abandoned as there doesn't seem to be any download options
anymore. And other than that, PDF readers with that capability are very
rare on any platform. No idea if anybody besides Adobe is doing that
because PDF is such a terribly complicated format.

In theory, this should all be doable with Tesseract, as it already does the
OCR part. Just nobody has bothered yet to support such use cases yet and
support an output format that can even handle more than just text.

Best
Richard

[1]: https://en.wikipedia.org/wiki/PDF/UA


Re: Needed tool for vision-impaired - was [Re: PDF Editor for Debian]

2024-06-24 Thread Nicolas George
Karen Lewellen (12024-06-24):
> Good afternoon.
> I am providing another option that might help here.
> robobraille,
> 
> www.robobraille.org
> Provides services, free of charge, that will convert pdf files  to a number
> of different formats, including .html
> They provide audio, mobi, and  convert epub files too..but I digress.
> As a test, consider sending your file to
> convert at robobraille.org
>  correctly of course.
> in the subjectline put html
> leaving the body blank, and attach the file.
> See if the .html file returned meets your needs.

Interesting.

Do you know how they fare with math? I mean real, non-trivial formulas
produced by LaTeX like you would find in
https://arxiv.org/abs/1803.05929 ?

(I know, I could test. I will if you do not know the answer.)

Regards,

-- 
  Nicolas George



Re: Needed tool for vision-impaired - was [Re: PDF Editor for Debian]

2024-06-24 Thread Karen Lewellen

Good afternoon.
I am providing another option that might help here.
robobraille,

www.robobraille.org
Provides services, free of charge, that will convert pdf files  to a 
number of different formats, including .html

They provide audio, mobi, and  convert epub files too..but I digress.
As a test, consider sending your file to
convert at robobraille.org
 correctly of course.
in the subjectline put html
leaving the body blank, and attach the file.
See if the .html file returned meets your needs.
Best,
Karen



On Mon, 24 Jun 2024, Richard Owlett wrote:


On 06/24/2024 12:35 AM, Richard wrote:

 Hello,
 this very much depends on what you are expecting it to do. In general,
 PDFs
 are only meant to be viewed - and printed - they where never meant for
 anything else. ...


Second sentence should read:

 ... only meant to be viewed by those with *NORMAL* vision ...


I'm attempting to read a USDA document.[1]
The printed version of this document is marginally readable.

Tools such as "Atril Document Viewer" provide selected magnification.
For this particular document and monitor, 150% is comfortable. Requires 
re-positioning the viewpoint 500 to 600 times to read document.


For _this_ document, Atril can select all the text on a page in a manner that 
can be pasted in a "reasonable" manner to a Pluma document.


It will:
   a. ignore actual graphics.
   b. put title/headings/??? on a separate line.
   c. all text between full page-width title/headings/??? will be
 treated as a logical unit.
It will not:
   1. put a blank line between paragraphs.
   2. put a blank line above/below lines containing title/headings/???.
   3. identify superscripts in some manner.

All this suggests that it should be able to extract text from a PDF and 
create a HTML document likely using only , , , and  in its 
.



[1] 
https://fns-prod.azureedge.us/sites/default/files/resource-files/TFP2021.pdf

_Thrifty Food Plan, 2021_
Food and Nutrition Service
August 2021
FNS-916






Needed tool for vision-impaired - was [Re: PDF Editor for Debian]

2024-06-24 Thread Richard Owlett

On 06/24/2024 12:35 AM, Richard wrote:

Hello,
this very much depends on what you are expecting it to do. In general, PDFs
are only meant to be viewed - and printed - they where never meant for
anything else. ...


Second sentence should read:

... only meant to be viewed by those with *NORMAL* vision ...


I'm attempting to read a USDA document.[1]
The printed version of this document is marginally readable.

Tools such as "Atril Document Viewer" provide selected magnification.
For this particular document and monitor, 150% is comfortable. Requires 
re-positioning the viewpoint 500 to 600 times to read document.


For _this_ document, Atril can select all the text on a page in a manner 
that can be pasted in a "reasonable" manner to a Pluma document.


It will:
   a. ignore actual graphics.
   b. put title/headings/??? on a separate line.
   c. all text between full page-width title/headings/??? will be
  treated as a logical unit.
It will not:
   1. put a blank line between paragraphs.
   2. put a blank line above/below lines containing title/headings/???.
   3. identify superscripts in some manner.

All this suggests that it should be able to extract text from a PDF and 
create a HTML document likely using only , , , and  in 
its .



[1] 
https://fns-prod.azureedge.us/sites/default/files/resource-files/TFP2021.pdf

_Thrifty Food Plan, 2021_
Food and Nutrition Service
August 2021
FNS-916