Initial commit of tesseract.

This includes tessdata as a nice to have (though tesseract doesn't work
very well without the trained data). This also includes the dependency
leptonica.

3 files changed, 129 insertions, 0 deletions

diff --git a/tessdata/.footprint b/tessdata/.footprint
new file mode 100644
index 0000000..bc73f29
--- /dev/null
+++ b/tessdata/.footprint
@@ -0,0 +1,110 @@
+drwxr-xr-x	root/root	usr/
+drwxr-xr-x	root/root	usr/share/
+drwxr-xr-x	root/root	usr/share/tessdata/
+-rw-r--r--	root/root	usr/share/tessdata/afr.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/amh.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ara.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/asm.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/aze.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/aze_cyrl.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/bel.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ben.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/bod.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/bos.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/bul.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/cat.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ceb.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ces.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/chi_sim.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/chi_tra.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/chr.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/cym.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/dan.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/dan_frak.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/deu.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/deu_frak.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/dzo.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ell.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/eng.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/enm.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/epo.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/equ.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/est.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/eus.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/fas.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/fin.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/fra.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/frk.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/frm.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/gle.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/glg.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/grc.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/guj.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/hat.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/heb.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/hin.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/hrv.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/hun.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/iku.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ind.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/isl.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ita.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ita_old.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/jav.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/jpn.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kan.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kat.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kat_old.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kaz.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/khm.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kir.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kor.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/kur.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/lao.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/lat.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/lav.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/lit.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/mal.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/mar.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/mkd.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/mlt.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/msa.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/mya.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/nep.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/nld.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/nor.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ori.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/osd.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/pan.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/pol.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/por.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/pus.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ron.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/rus.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/san.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/sin.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/slk.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/slk_frak.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/slv.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/spa.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/spa_old.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/sqi.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/srp.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/srp_latn.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/swa.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/swe.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/syr.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tam.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tel.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tgk.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tgl.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tha.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tir.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/tur.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/uig.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/ukr.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/urd.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/uzb.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/uzb_cyrl.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/vie.traineddata
+-rw-r--r--	root/root	usr/share/tessdata/yid.traineddata
diff --git a/tessdata/.md5sum b/tessdata/.md5sum
new file mode 100644
index 0000000..354f5a7
--- /dev/null
+++ b/tessdata/.md5sum
@@ -0,0 +1 @@
+b25e830d203af5c863081af3f684b53a  3.04.00.tar.gz
diff --git a/tessdata/Pkgfile b/tessdata/Pkgfile
new file mode 100644
index 0000000..0b4475d
--- /dev/null
+++ b/tessdata/Pkgfile
@@ -0,0 +1,18 @@
+# Description:  Training data for tesseract OCR
+# URL:          https://github.com/tesseract-ocr/tessdata
+# Maintainer:   Aaron Ball, nullspoon at oper dot io
+# Depends on:
+# Nice to have: tesseract
+
+name=tessdata
+version=3.04.00
+release=1
+source=(https://github.com/tesseract-ocr/${name}/archive/${version}.tar.gz)
+
+build() {
+  cd ${name}-${version}
+
+  # Install all traineddata files 
+  mkdir -p ${PKG}/usr/share/tessdata/
+  install -m 644 *.traineddata ${PKG}/usr/share/tessdata/
+}