Add MorphoSyntax and fastText embeddings

CMakeLists.txt

@@ -1,15 +1,19 @@
 cmake_minimum_required(VERSION 3.12.1)
 
 set(CPACK_ARCHIVE_COMPONENT_INSTALL ON)
-set(CPACK_COMPONENTS_GROUPING IGNORE)
+set(CPACK_COMPONENTS_GROUPING ONE_PER_GROUP)
 set(CPACK_DEB_COMPONENT_INSTALL ON)
 set(CPACK_GENERATOR "DEB")
 set(CPACK_DEBIAN_PACKAGE_MAINTAINER "LIMA Publisher")
 set(CPACK_PACKAGE_NAME lima-deep-models)
-set(CPACK_PACKAGE_VERSION "0.1.2")
+set(CPACK_PACKAGE_VERSION "0.1.3")
 set(CPACK_DEBIAN_PACKAGE_ARCHITECTURE "all")
 set(CPACK_DEBIAN_FILE_NAME DEB-DEFAULT)
 
 include(CPack)
 
+cpack_add_component_group(English)
+cpack_add_component_group(French)
+
 add_subdirectory(TensorFlowTokenizer)
+add_subdirectory(TensorFlowMorphoSyntax)

TensorFlowMorphoSyntax/CMakeLists.txt

@@ -0,0 +1,4 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+add_subdirectory(eng.ud)
+add_subdirectory(fre.ud)

TensorFlowMorphoSyntax/eng.ud/CMakeLists.txt

@@ -0,0 +1,20 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+add_subdirectory(ewt)
+
+install(
+    FILES
+        cc.en.300.bin
+    COMPONENT
+        morphosyntax-eng.ud
+    DESTINATION
+        share/apps/lima/resources/TensorFlowMorphoSyntax/eng.ud/
+)
+
+cpack_add_component(morphosyntax-eng.ud
+    DISPLAY_NAME
+        "UD English / FastText Embeddings"
+    GROUP
+        English
+)
+

TensorFlowMorphoSyntax/eng.ud/ewt/CMakeLists.txt

@@ -0,0 +1,19 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+install(
+    FILES
+        UD_English-EWT.conf
+        UD_English-EWT.model
+    COMPONENT
+        morphosyntax-eng.ud-ewt
+    DESTINATION
+        share/apps/lima/resources/TensorFlowMorphoSyntax/eng.ud/ewt/
+)
+
+cpack_add_component(morphosyntax-eng.ud-ewt
+    DISPLAY_NAME
+        "UD English / EWT corpus"
+    GROUP
+        English
+)
+

TensorFlowMorphoSyntax/fre.ud/CMakeLists.txt

@@ -0,0 +1,20 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+add_subdirectory(sequoia)
+
+install(
+    FILES
+        cc.fr.300.bin
+    COMPONENT
+        morphosyntax-fre.ud
+    DESTINATION
+        share/apps/lima/resources/TensorFlowMorphoSyntax/fre.ud/
+)
+
+cpack_add_component(morphosyntax-fre.ud
+    DISPLAY_NAME
+        "UD French / FastText Embeddings"
+    GROUP
+        French
+)
+

TensorFlowMorphoSyntax/fre.ud/sequoia/CMakeLists.txt

@@ -0,0 +1,19 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+install(
+    FILES
+        UD_French-Sequoia.conf
+        UD_French-Sequoia.model
+    COMPONENT
+        morphosyntax-fre.ud-sequoia
+    DESTINATION
+        share/apps/lima/resources/TensorFlowMorphoSyntax/fre.ud/sequoia/
+)
+
+cpack_add_component(morphosyntax-fre.ud-sequoia
+    DISPLAY_NAME
+        "UD French / Sequoia corpus"
+    GROUP
+        French
+)
+

TensorFlowTokenizer/CMakeLists.txt

@@ -1,5 +1,4 @@
 cmake_minimum_required(VERSION 3.12.1)
 
-cpack_add_component_group(Tokenizer)
-
 add_subdirectory(eng.ud)
+add_subdirectory(fre.ud)

TensorFlowTokenizer/eng.ud/ewt/CMakeLists.txt

@@ -14,6 +14,6 @@ cpack_add_component(tokenizer-eng.ud-ewt
     DISPLAY_NAME
         "UD English / EWT corpus"
     GROUP
-        Tokenizer
+        English
 )
 

TensorFlowTokenizer/eng.ud/ud23-all/CMakeLists.txt

@@ -14,6 +14,6 @@ cpack_add_component(tokenizer-eng.ud-all
     DISPLAY_NAME
         "UD English / All corpora"
     GROUP
-        Tokenizer
+        English
 )
 

build.sh

@@ -1,3 +1,4 @@
+rm -rf build
 mkdir -p build
 cd build
 cmake -DCMAKE_INSTALL_PREFIX:PATH=/usr ..