tesseract-ocr · stweil · May 23, 2019
diff --git a/.gitmodules b/.gitmodules
@@ -7,3 +7,6 @@
 [submodule "test"]
 	path = test
 	url = https://github.com/tesseract-ocr/test
+[submodule "tessdata"]
+	path = tessdata
+	url = https://github.com/tesseract-ocr/tessconfigs
diff --git a/Makefile.am b/Makefile.am
@@ -1,4 +1,9 @@
 ## run autogen.sh to create Makefile.in from this file
+
+# Default location for tessdata directory.
+# It can be overriden with configure option --datarootdir=DIR.
+datadir = @datarootdir@/tessdata
+
 ACLOCAL_AMFLAGS = -I m4
 
 if ENABLE_TRAINING
@@ -18,17 +23,34 @@ training:
 	@echo "Need to reconfigure project, so there are no errors"
 endif
 
-.PHONY: doc install-langs ScrollView.jar install-jars training
+.PHONY: doc install-langs ScrollView.jar install-jars install-tessdata training
 
 SUBDIRS = src/arch src/ccutil src/viewer src/cutil src/opencl src/ccstruct
 SUBDIRS += src/dict src/classify src/wordrec src/textord src/lstm
-SUBDIRS += src/ccmain src/api . tessdata doc unittest
+SUBDIRS += src/ccmain src/api . doc unittest
 
 EXTRA_DIST = README.md LICENSE
 EXTRA_DIST += aclocal.m4 config configure.ac autogen.sh
 EXTRA_DIST += tesseract.pc.in $(TRAINING_SUBDIR) java doc
 EXTRA_DIST += CMakeLists.txt tesseract.pc.cmake cmake VERSION src/vs2010 cppan.yml
 
+# Files for tessdata.
+TESSDATA_FILES = $(top_srcdir)/tessdata/pdf.ttf
+
+# Files for tessdata/configs.
+CONFIG_FILES = tessdata/pdf.ttf
+CONFIG_FILES = inter makebox box.train unlv ambigs.train lstm.train lstmdebug
+CONFIG_FILES += api_config kannada box.train.stderr quiet logfile digits get.images
+CONFIG_FILES += lstmbox wordstrbox
+# Configurations for OCR output.
+CONFIG_FILES += alto hocr pdf tsv txt
+CONFIG_FILES += linebox rebox strokewidth bigram
+
+# Files for tessdata/tessconfigs.
+TESSDATA_CONFIG_FILES = ${CONFIG_FILES:%=$(top_srcdir)/tessdata/configs/%}
+TESSCONFIG_FILES = batch batch.nochop nobatch matdemo segdemo msdemo
+TESSDATA_TESSCONFIG_FILES = ${TESSCONFIG_FILES:%=$(top_srcdir)/tessdata/tessconfigs/%}
+
 DIST_SUBDIRS = $(SUBDIRS) $(TRAINING_SUBDIR)
 
 uninstall-hook:
@@ -50,6 +72,14 @@ ScrollView.jar:
 install-jars:
 	@cd "$(top_builddir)/java" && $(MAKE) $@
 
+install-tessdata:
+	mkdir -p $(DESTDIR)$(datadir)
+	$(INSTALL) -m 644 $(TESSDATA_FILES) $(DESTDIR)$(datadir)
+	mkdir -p $(DESTDIR)$(datadir)/configs
+	$(INSTALL) -m 644 $(TESSDATA_CONFIG_FILES) $(DESTDIR)$(datadir)/configs
+	mkdir -p $(DESTDIR)$(datadir)/tessconfigs
+	$(INSTALL) -m 644 $(TESSDATA_TESSCONFIG_FILES) $(DESTDIR)$(datadir)/tessconfigs
+
 doc:
 	-srcdir="$(top_srcdir)" builddir="$(top_builddir)" \
 	version="@PACKAGE_VERSION@" name="@PACKAGE_NAME@" \

diff --git a/configure.ac b/configure.ac
@@ -482,9 +482,6 @@ AC_CONFIG_FILES([src/lstm/Makefile])
 AC_CONFIG_FILES([src/textord/Makefile])
 AC_CONFIG_FILES([src/viewer/Makefile])
 AC_CONFIG_FILES([src/wordrec/Makefile])
-AC_CONFIG_FILES([tessdata/Makefile])
-AC_CONFIG_FILES([tessdata/configs/Makefile])
-AC_CONFIG_FILES([tessdata/tessconfigs/Makefile])
 AC_CONFIG_FILES([unittest/Makefile])
 AC_CONFIG_FILES([java/Makefile])
 AC_CONFIG_FILES([java/com/Makefile])

diff --git a/tessdata b/tessdata
diff --git a/tessdata/Makefile.am b/tessdata/Makefile.am
diff --git a/tessdata/configs/Makefile.am b/tessdata/configs/Makefile.am
diff --git a/tessdata/configs/alto b/tessdata/configs/alto
diff --git a/tessdata/configs/ambigs.train b/tessdata/configs/ambigs.train
diff --git a/tessdata/configs/api_config b/tessdata/configs/api_config
diff --git a/tessdata/configs/bazaar b/tessdata/configs/bazaar
diff --git a/tessdata/configs/bigram b/tessdata/configs/bigram
diff --git a/tessdata/configs/box.train b/tessdata/configs/box.train
diff --git a/tessdata/configs/box.train.stderr b/tessdata/configs/box.train.stderr
diff --git a/tessdata/configs/digits b/tessdata/configs/digits
diff --git a/tessdata/configs/get.images b/tessdata/configs/get.images
diff --git a/tessdata/configs/hocr b/tessdata/configs/hocr
diff --git a/tessdata/configs/inter b/tessdata/configs/inter
diff --git a/tessdata/configs/kannada b/tessdata/configs/kannada
diff --git a/tessdata/configs/linebox b/tessdata/configs/linebox
diff --git a/tessdata/configs/logfile b/tessdata/configs/logfile
diff --git a/tessdata/configs/lstm.train b/tessdata/configs/lstm.train
diff --git a/tessdata/configs/lstmbox b/tessdata/configs/lstmbox
diff --git a/tessdata/configs/lstmdebug b/tessdata/configs/lstmdebug
diff --git a/tessdata/configs/makebox b/tessdata/configs/makebox
diff --git a/tessdata/configs/pdf b/tessdata/configs/pdf
diff --git a/tessdata/configs/quiet b/tessdata/configs/quiet
diff --git a/tessdata/configs/rebox b/tessdata/configs/rebox
diff --git a/tessdata/configs/strokewidth b/tessdata/configs/strokewidth
diff --git a/tessdata/configs/tsv b/tessdata/configs/tsv
diff --git a/tessdata/configs/txt b/tessdata/configs/txt
diff --git a/tessdata/configs/unlv b/tessdata/configs/unlv
diff --git a/tessdata/configs/wordstrbox b/tessdata/configs/wordstrbox
diff --git a/tessdata/eng.user-patterns b/tessdata/eng.user-patterns
diff --git a/tessdata/eng.user-words b/tessdata/eng.user-words
diff --git a/tessdata/pdf.ttf b/tessdata/pdf.ttf
diff --git a/tessdata/tessconfigs/Makefile.am b/tessdata/tessconfigs/Makefile.am
diff --git a/tessdata/tessconfigs/batch b/tessdata/tessconfigs/batch
diff --git a/tessdata/tessconfigs/batch.nochop b/tessdata/tessconfigs/batch.nochop
diff --git a/tessdata/tessconfigs/matdemo b/tessdata/tessconfigs/matdemo
diff --git a/tessdata/tessconfigs/msdemo b/tessdata/tessconfigs/msdemo
diff --git a/tessdata/tessconfigs/nobatch b/tessdata/tessconfigs/nobatch
diff --git a/tessdata/tessconfigs/segdemo b/tessdata/tessconfigs/segdemo