Tesseract: Publish Tesseract 4.0 to nuget.org

[badre429]

the feature/321-Tesseract-4 barche has a working code for version Tesseract 4 wrapper for .net standards,
just publish the package to nuget

[seriouz]

@charlesw That would be very awesome! Please push the 321-Tesseract-4 branch as a nuget beta package.
Cheers!

[charlesw]

Agreed.

On Mon., 10 Sep. 2018, 23:53 seriouz, [e-mail] wrote:

@charlesw https://github.com/charlesw That would be very awesome!
Please push the 321-Tesseract-4 branch as a nuget beta package.
Cheers!

—
You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
https://github.com/charlesw/tesseract/issues/428#issuecomment-419919810,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAPzyAI95MuWWaDjL7PNeed-clIETdoIks5uZm6_gaJpZM4VGsuO
.

[charlesw]

As in yes, I'll push a beta release. Might as well release the latest 3.05
version too. It's 🐝 a while 🙂

Might take a couple of days though, quite busy with other priorities ATM.

On Tue., 11 Sep. 2018, 08:34 Charles Weld, [e-mail] wrote:

Agreed.

On Mon., 10 Sep. 2018, 23:53 seriouz, [e-mail] wrote:

@charlesw https://github.com/charlesw That would be very awesome!
Please push the 321-Tesseract-4 branch as a nuget beta package.
Cheers!

—
You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub
https://github.com/charlesw/tesseract/issues/428#issuecomment-419919810,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAPzyAI95MuWWaDjL7PNeed-clIETdoIks5uZm6_gaJpZM4VGsuO
.

[seriouz]

@charlesw Thank you very much! Appreciated!

[arthurvb]

@charlesw Are you still planning of doing this? Otherwise I can look in too it; I tried to publish it to Tesseract-test (only netstandard). I will of course remove the package on your request.

[charlesw]

Yes, still plan on doing this just haven't got around to doing it. Sorry.

I have no issues with you publishing your own copy.

On Mon., 1 Oct. 2018, 05:23 arthurvb, [e-mail] wrote:

@charlesw https://github.com/charlesw Are you still planning of doing
this? Otherwise I can look in too it; I tried to publish it to
Tesseract-test (only netstandard). I will of course remove the package on
you request.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/charlesw/tesseract/issues/428#issuecomment-425744718,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAPzyFv9_Xj-DclREBGtXms4H2wFGvG4ks5ugRolgaJpZM4VGsuO
.

[arthurvb]

FYI: I just created a package that now works for OSX and Windows, and should works as well on Linux if we would add the binaries:
https://github.com/arthurvb/tesseract/tree/feature/321-Tesseract-4-osx
I removed BaseAPIProcessPage since that gave errors

[arthurvb]

Nuget package is at https://www.nuget.org/packages/Tesseract-test/4.0.0-beta15

[ilochray]

Thank you for the package. When I installed it, I could only add it to a project with a .Net target framework of 4.6.1. Is this a minimum requirement?

[arthurvb]

@ilochray Probably because I selected netstandard2.0 as output (I got it running on both .net core and .net framework 4.7.2)

[badre429]

@charlesw any update after tesseract v4.0 is RTM

[Nonobis]

@arthurvb : I trying to use your test package in asp.netCore apps (2.1), ... before official package. But, i got this error :
Failed to find library "liblept1760.dll" for platform x64.

NB :

• WebApi Project in NetCore 2.1
• The dll are exising in x86 & x64 directory.
• Same error in Any Cpu / x64.

An idea ?

[arthurvb]

@Nonobis Probably it is the wrong version (just unlisted all the wrong ones). Check if you use version 15: https://www.nuget.org/packages/Tesseract-test/4.0.0-beta15

[Nonobis]

@arthurvb : got it working with lastest package and TesseractEnviornment.CustomSearchPath. thx

[charlesw]

Quick update on release of 'official' nuget package.

I've done some testing on my machine and unfortunately a number of tests are failing, mainly around detecting orientation, at this time it does look like the issue is with the tesseract library but haven't yet worked out if it's because of how we're compiling it or what. Holding of doing an official release for now until I've looked into this more however could do a 4.0 beta release if that helps others that don't require this functionality.

[SimonChris]

@arthurvb I am also trying to use this package, but the PixConverter class seems to be missing, so I can't convert my Bitmap. I am using 4.0.0-beta15.

[richardtallent]

An "official" beta release would be awesome! I've been dead in the water with 3.x recently (.NET framework version issue maybe? The exception is vague), so between that and the newer engine, I've been eager to try v4, and everything I'm using it for is portrait orientation.

[unarix]

Im using 4.0.0-beta15, and get this exception:
"Exception has been thrown by the target of an invocation."
when try to do:
"using (var engine = new TesseractEngine(@"./tessdata", "eng", EngineMode.Default))"

(core 2.1 - linux)

[arthurvb]

@SimonChris You should be able to do var pix = Pix.LoadFromFile(tmpLocation). This location can contain png-file. Not perfect.. but it does work.
@unarix I have to say that I only tested it multiple times on OSX and Windows. It could be that the library file is not compatible with the linux version (maybe 32bit vs. 64bit?)

[charlesw]

.net core support doesn't work on Linux atm. The issue appears that for
some reason some of the method signatures from the tesseract library don't
load resulting in errors. Others have reported that commenting them out
resolves the issue (I believe ProcessPage is the problematic one) .

On Tue., 27 Nov. 2018, 05:17 arthurvb <[email protected] wrote:

@SimonChris https://github.com/SimonChris You should be able to do var
pix = Pix.LoadFromFile(tmpLocation). This location can contain png-file.
Not perfect.. but it does work.
@unarix https://github.com/unarix I have to say that I only tested it
multiple times on OSX and Windows. It could be that the library file is not
compatible with the linux version (maybe 32bit vs. 64bit?)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/charlesw/tesseract/issues/428#issuecomment-441742366,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAPzyNG_CQamkz26AH_s8WjE2yzZBNTqks5uzDBPgaJpZM4VGsuO
.

[refkmd]

Hi @charlesw
First of all - Happy New Year :-)
Second - is there any news on an official Nuget Package release with Tesseract 4.0...?
We are really waiting for this, because we are excited to see the improvements in out environment ;-)

[charlesw]

Hi, this is the next priority. However very busy with work ATM so don't
have a lot of time to spend on this.

Sorry I can't give you an ETA at the moment, but stay tuned :)

On Tue., 8 Jan. 2019, 19:09 refkmd <[email protected] wrote:

Hi @charlesw https://github.com/charlesw
First of all - Happy New Year :-)
Second - is there any news on an official Nuget Package release with
Tesseract 4.0...?
We are really waiting for this, because we are excited to see the
improvements in out environment ;-)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/charlesw/tesseract/issues/428#issuecomment-452209474,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AAPzyDjMPZd_aLgJsV5NZg6kk8rDITy6ks5vBFJBgaJpZM4VGsuO
.

[richardtallent]

Just a quick note in case it's useful -- while waiting for this library to move to v4, we're currently doing command-line tesseract... painful but works. Today we got a PDF with one page upside down and several landscape, and tesseract totally failed to recognize the differing orientations. I don't recall it having that issue in v3, but this is an unusual PDF.

So it may be an issue with v4 itself, not with your library's use of it. The improvements in recognition in v4 are worth the hassle of dealing with orientation issues manually, so if that's the only concern at the moment, it may be worth just getting the new version out there and as users, we'll figure out a workaround for any orientation detection issues (or keep using v3 if it's a huge problem--it wouldn't be for us).

[refkmd]

Hi @charlesw
Now it's almost 3 months since I wrote last ;-)
You answered back, that official Nuget Package release with Tesseract 4.0 was next priority, so could you please give a status and an expected time frame?
We a really waiting for this release and it dosen't seams like I/we are the only ones :-)

[aurnoi1]

I support the request of @refkmd. Just a little update on your progression :) . And thanks a lot for your work.

[richardtallent]

I'm also interested to know where things stand. But as a maintainer of several open-source projects myself, I know it's an often thankless job crammed into minimal free time.

So... is there anything we in the community can be doing to help or move things forward? A certain set of issues to try to narrow or close, etc.?

[prasanthtp]

Those who want to do a local nuget for teseract 4.0;
I did the following and it worked for me;

1. Clone the teseract 4.0 feature branch
2. Do a release build in visual studio
3. Navigate to your BinRelease folder and copy the full path
4. in visual studio > nuget packagemanager console
5. type the following command and press enter Install-Package Tesseract -Source ""

DONE

Or you can create a private nuet feed in azure devops , Its free

[VasylSulakov]

Hi,
By no means I can launch tesseract-feature-321-Tesseract-4, it would always fail with "error-1" thing (could not be initialized). I've tried all the possible combinations of traineddata files (Data Files for Version 4.00 (November 29, 2016), Updated Data Files for Version 4.00 (September 15, 2017) - best, fast, legacy), but no luck.

[LorisFriedel]

Hello!

That would be awesome to have the Tesseract 4.0.0 release available on nuget! Is the full procedure to generate a release of C# tesseract wrapper described by the "Compling tesseract and leptonica.txt" file or is there some other steps to perform?

BTW, great job, this is really good work and thank you for that!

[Doomer3D]

Enjoy!
https://www.nuget.org/packages/Genesis.Tesseract4/

[chucker]

@Doomer3D FWIW, it doesn't work for me.

I have a project that works fine with the current release build.

Switching the NuGet package to Genesis.Tesseract4 4.0.2.11, I got a DllNotFoundException on libtesseract. I tried fiddling with TesseractEnviornment.CustomSearchPath, but that didn't help and shouldn't have, since the DLL _was_ copied to bin\…\x64.

After that, I tried switching to Tesseract-test 4.0.0-beta16 instead, and that worked out of the box. So I think something's not quite right with your NuGet package. (I haven't tried cloning and debugging yet.)

[Doomer3D]

@chucker, thanks for report.

It is my first Nuget package, ofc, it may be bugged, but I just created new .NET Core 2 solution and added Genesis.Tesseract4 package, and its works:

Can you send me your solution or project file for reproduce?

P.S. Sorry for my english, it is not my native languange.

[MichaelPeter]

Hello Doomer3D, could you also add a target for netstandard 2.1 in the Genesis.Tesseract package?

I have a 2.1 project but the x86 and the x64 directories are only copied to the netstandard 2.0 directory not the netstandard 2.1.