Yes, but that would mean duplicating work and server space for the exact same
data packaged slightly differently. In other words, the data exists in a
different form already, and is freely available to download. I honestly don't
think duplicating the test data for *all* the languages of Tesseract into an
add-on form would be advantageous for anyone.
Timothy
Timothy
----- Original Message -----
From: Simon Jaeger
To: nvda-addons@xxxxxxxxxxxxx
Sent: Sunday, March 12, 2017 11:23 PM
Subject: [nvda-addons] Re: Updating OCR add-on to new Tesseract
The non-codefactory vocalizer plugin has voices which are all separate
addons, but which are accessible from the main vocalizer plugin. You install
the synthesizer, and then you install voices separately. Could this be done for
OCR? They do get stored in a separate folder, but this would save anyone from
needing to write a downloader, as the user could just download an addon and
install it the normal way.
On 2017-03-12 19:30, James Teh wrote:
If you close your browser without a download manager, same problem. Simple
solution: don't exit NVDA while you're downloading. I don't think fixing this
is worthwhile.
On Mon, Mar 13, 2017 at 12:25 PM, Derek Riemer <driemer.riemer@xxxxxxxxx>
wrote:
The problem with this is that if NVDA exits while the download is in
progress, goodbye previous download progress.
On Sun, Mar 12, 2017 at 6:32 PM, James Teh <jamie@xxxxxxxxxxxx> wrote:
On Sat, Mar 11, 2017 at 1:48 AM, Alberto Buffolino
<a.buffolino@xxxxxxxxx> wrote:
I agree. Currently, size will be about 40MB, absolutely acceptable.
Wow. Is that just with the program and English data? That's definitely
grown quite a bit larger. :)
hmm, why will we remove them when updating? To guarantee the
compatibility with current Tesseract?
No. I mean that if the data is stored in the add-on directory, it will
get removed by NVDA when the add-on is updated. This is because when an add-on
is updated, the old add-on is completely removed and the new one installed
fresh. This guarantees that files from the old version do not interfere with
the new version. So, if you want files to remain when you update an add-on, you
must store those files outside the add-on directory.
* a download/update button to launch the specific download via
default browser (using webbrowser module, already included, very easy);
It might be easy, but it's not an acceptable user experience IMO. If we
do it using the default browser, the user has to specify the download location.
Many users will not be able to figure this out; appdata file paths require at
least basic technical knowledge. This is why I suggested that the downloader
needs to be in the add-on. This way, the add-on can just download the requested
languages and install them itself.
Ah, a question: I grabbed new tesseract.exe and requested dll from
the setup linked here, without compiling on my machine:
https://github.com/tesseract-ocr/tesseract/wiki/4.0-with-LSTM
Could this be a problem?
Not exatly sure what you mean, but if it helps, I never built Tesseract
myself either; I just use pre-built binaries.
Jamie
--
Derek Riemer: Improving the world one byte at a time!
a.. University of Colorado Boulder Department of computer science, 4th
year undergraduate student.
b.. Accessibility enthusiast.
c.. Proud user of the NVDA screen reader.
d.. Open source enthusiast.
e.. Skier.
Personal website