tesseract @3.02.02 version upgrade

[mdbecque@…]

The 3.02 version upgrade.

[cooljeanius (Eric Gallager)]

The livecheck should also be fixed:

Local-Admins-MacBook-Pro:~ ericgallager$ port -v livecheck tesseract
Error: cannot check if tesseract was updated (Couldn't resolve host 'googlecode:tesseract-ocr')

[cooljeanius (Eric Gallager)]

Cc Me!

[larryv (Lawrence Velázquez)]

Thanks. In the future, please Cc relevant port maintainers.

[larryv (Lawrence Velázquez)]

Replying to egall@…:

The livecheck should also be fixed

Open a separate ticket for this.

[cooljeanius (Eric Gallager)]

Replying to larryv@…:

Replying to egall@…:

The livecheck should also be fixed

Open a separate ticket for this.

OK, #38873. I just thought this would be relevant here because usually it's useful to post a livecheck result as evidence that a port should be updated.

[markemer (Mark Anderson)]

I've been meaning to do this since the last time the Portfile broke, so I'll get right on it.

[mdbecque@…]

I was able to build 3.02.02 with leptonica-1.69 with libgif 4.1.6 : libjpeg 9 : libpng 1.5.15 : libtiff 4.0.3 : zlib 1.2.7 with the attached port file. Hope this helps.

[mf2k (Frank Schima)]

​As per the guide, please instead attach a unified diff of the Portfile vs. trunk.

[mdbecque@…]

Attempted diff file.

[markemer (Mark Anderson)]

Ok, sorry I've been off this so long. I think the diff file gets us most of the way there, thanks. I think it needs a tweak or two. Will do so soon.

Edit: Actually, I take that back. This checks out from SVN, but googlecode has the 3.02 files we need. I am, however, still on the case.

[markemer (Mark Anderson)]

Ok, so this is more complicated than I thought. This patch out from SVN, but googlecode has the 3.02 files we need. So, I did that, but now I need to check to see if we still need to hack autoconf, or if we can configure like normal people.

[markemer (Mark Anderson)]

And now, for more complication - There are a lot of new languages that need to be added as subports. The guys at the tesseract don't like making my life easy. :-P

[markemer (Mark Anderson)]

Ok, I have not given up on this, I'm currently on 10.9 which is making my life tough ala macports, but I'm closing in on a proper 3.02.02 build, which has been a huge pain in the ass. I'm going to post it soon, after I'm sure it works on 10.9 and 10.8 which is all I have access to easily.

[markemer (Mark Anderson)]

Portfile diff with all the new langs an update to 3.02.02

[markemer (Mark Anderson)]

I added a new portfile for 3.02.02 with all the new languages and checksums. I'm not sure if we still need the old patches. It builds on my machine without them, but I forget exactly why we needed them, and I think it was something I could not replicate. If someone could try that out, that'd be cool.

[mdbecque@…]

Replying to emer@…:

I added a new portfile for 3.02.02 with all the new languages and checksums. I'm not sure if we still need the old patches. It builds on my machine without them, but I forget exactly why we needed them, and I think it was something I could not replicate. If someone could try that out, that'd be cool.

I used your diff to patch the old portfile. 3.02.02 built and deployed like a charm on my machine. tesseract-eng installed without a problem.

[markemer (Mark Anderson)]

I think we're good. We can move ahead on this I think.

[mf2k (Frank Schima)]

As with all version increases, the revision line needs to be deleted.

[mdbecque@…]

New patch with revision number deleted.

[markemer (Mark Anderson)]

Indeed it does, good catch. Still works for me.

[ryandesign (Ryan Carsten Schmidt)]

I'm working on committing this; stand by.

[ryandesign (Ryan Carsten Schmidt)]

The patch was reversed but that's easy to deal with.

Replying to emer@…:

I'm not sure if we still need the old patches. It builds on my machine without them, but I forget exactly why we needed them, and I think it was something I could not replicate. If someone could try that out, that'd be cool.

The only reason we were patching configure.ac and running autoreconf was to fix a failure when $CC has a space in it, e.g. when ccache is being used (which it is on my system). Running autoreconf necessitated making the empty m4 directory. This bug in the configure script is fixed in 3.02 so we no longer need this patch nor to run autoreconf nor to create the m4 directory.

The only reason we were patching strngs.h was to remove the UTF-8 byte order marker from the file, which older GCC compilers failed on. In 3.02 there is no UTF-8 BOM on this file so we don't need to patch it.

There was a special-case section in the language subport code for dealing with version 3.00 language files. Since the port now doesn't have any of those anymore I removed the special-case code.

Some of the subports appear to conflict with one another. For example, tesseract-eng fails to activate, saying:

Error: org.macports.activate for port tesseract-eng returned: Image error: /opt/local/share/tessdata/eng.cube.params is being used by the active tesseract-ara port.  Please deactivate this port first, or use 'port -f activate tesseract-eng' to force the activation.

Why does tesseract-ara install files for other languages?

$ port cont tesseract-ara
Port tesseract-ara contains:
  /opt/local/share/tessdata/ara.cube.bigrams
  /opt/local/share/tessdata/ara.cube.fold
  /opt/local/share/tessdata/ara.cube.lm
  /opt/local/share/tessdata/ara.cube.nn
  /opt/local/share/tessdata/ara.cube.params
  /opt/local/share/tessdata/ara.cube.size
  /opt/local/share/tessdata/ara.cube.word-freq
  /opt/local/share/tessdata/ara.traineddata
  /opt/local/share/tessdata/eng.cube.params
  /opt/local/share/tessdata/fra.cube.params
  /opt/local/share/tessdata/hin.cube.params
  /opt/local/share/tessdata/ita.cube.params
  /opt/local/share/tessdata/rus.cube.params
  /opt/local/share/tessdata/spa.cube.params

Similarly, tesseract-fra fails to activate:

Error: org.macports.activate for port tesseract-fra returned: Image error: /opt/local/share/tessdata/dan-frak.traineddata is being used by the active tesseract-dan port.  Please deactivate this port first, or use 'port -f activate tesseract-fra' to force the activation.

And so on. I've committed the port as is in r112497 but we need to figure out how we want to handle this: either marking the ports as conflicting with one another, or making them not conflict. I'll file a new ticket.