English tokenization models

CMakeLists.txt

@@ -0,0 +1,15 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+set(CPACK_ARCHIVE_COMPONENT_INSTALL ON)
+set(CPACK_COMPONENTS_GROUPING IGNORE)
+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 "0.1.1")
+set(CPACK_DEBIAN_PACKAGE_ARCHITECTURE "all")
+set(CPACK_DEBIAN_FILE_NAME DEB-DEFAULT)
+
+include(CPack)
+
+add_subdirectory(TensorFlowTokenizer)

TensorFlowTokenizer/CMakeLists.txt

@@ -0,0 +1,5 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+cpack_add_component_group(Tokenizer)
+
+add_subdirectory(eng.ud)

TensorFlowTokenizer/eng.ud/CMakeLists.txt

@@ -0,0 +1,4 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+add_subdirectory(ewt)
+add_subdirectory(ud23-all)

TensorFlowTokenizer/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
+        tokenizer-eng.ud-ewt
+    DESTINATION
+        /share/apps/lima/resources/TensorFlowTokenizer/eng.ud/ewt/
+)
+
+cpack_add_component(tokenizer-eng.ud-ewt
+    DISPLAY_NAME
+        "UD English / EWT corpus"
+    GROUP
+        Tokenizer
+)
+

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

@@ -0,0 +1,19 @@
+cmake_minimum_required(VERSION 3.12.1)
+
+install(
+    FILES
+        UD_English-ALL.conf
+        UD_English-ALL.model
+    COMPONENT
+        tokenizer-eng.ud-all
+    DESTINATION
+        /share/apps/lima/resources/TensorFlowTokenizer/eng.ud/all/
+)
+
+cpack_add_component(tokenizer-eng.ud-all
+    DISPLAY_NAME
+        "UD English / All corpora"
+    GROUP
+        Tokenizer
+)
+