[TIKA-3420] Set tesseract ocr langauges as docker build args

README.md

@@ -11,7 +11,11 @@ There is a minimal version, which contains only Apache Tika and it's core depend
 * Italian
 * Spanish.
 
-To install more languages simply update the apt-get command to include the package containing the language you required, or include your own custom packs using an ADD command.
+To install more languages simply use `docker-build.sh` or manually using [docker --build-arg](https://docs.docker.com/engine/reference/commandline/build/#set-build-time-variables---build-arg)
+
+For see with version is supported by tesseract on official package:
+
+    apt-cache search --names-only '^tesseract-ocr-[a-z]{3}$'
 
 ## Available Tags
 

docker-tool.sh

@@ -21,11 +21,14 @@ while getopts ":h" opt; do
   case ${opt} in
     h )
       echo "Usage:"
-      echo "    docker-tool.sh -h                      Display this help message."
-      echo "    docker-tool.sh build <TIKA_VERSION>    Builds images for <TIKA_VERSION>."
-      echo "    docker-tool.sh test <TIKA_VERSION>     Tests images for <TIKA_VERSION>."
-      echo "    docker-tool.sh publish <TIKA_VERSION>  Publishes images for <TIKA_VERSION> to Docker Hub."
-      echo "    docker-tool.sh latest <TIKA_VERSION>   Tags images for <TIKA_VERSION> as latest on Docker Hub."
+      echo "    docker-tool.sh -h                                              Display this help message."
+      echo "    docker-tool.sh build <TIKA_VERSION> [<TESSERACT_LANGUAGES>]    Builds images for <TIKA_VERSION> via special [<TESSERACT_LANGUAGES>]."
+      echo "    docker-tool.sh test <TIKA_VERSION>                             Tests images for <TIKA_VERSION>."
+      echo "    docker-tool.sh publish <TIKA_VERSION>                          Publishes images for <TIKA_VERSION> to Docker Hub."
+      echo "    docker-tool.sh latest <TIKA_VERSION>                           Tags images for <TIKA_VERSION> as latest on Docker Hub."
+      echo ""
+      echo "Note: [<TESSERACT_LANGUAGES>] is optional for full image,"
+      echo "      for change default tesseract-ocr packages."
       exit 0
       ;;
    \? )
@@ -58,13 +61,18 @@ test_docker_image() {
 shift $((OPTIND -1))
 subcommand=$1; shift
 version=$1; shift
+tesseract_languages=$@
 
 case "$subcommand" in
   build)
+    build_args="--build-arg TIKA_VERSION=${version}"
+    if [[ ! -z "$tesseract_languages" ]]; then
+      build_args="$build_args --build-arg TESSERACT_LANGUAGES='${tesseract_languages}'"
+    fi
     # Build slim version with minimal dependencies
     docker build -t apache/tika:${version} --build-arg TIKA_VERSION=${version} - < minimal/Dockerfile --no-cache
     # Build full version with OCR, Fonts and GDAL
-    docker build -t apache/tika:${version}-full --build-arg TIKA_VERSION=${version} - < full/Dockerfile --no-cache
+    docker build -t apache/tika:${version}-full ${build_args} - < full/Dockerfile --no-cache
     ;;
 
   test)

full/Dockerfile

@@ -14,9 +14,9 @@ RUN apt-get update
 
 FROM base as dependencies
 ARG JRE='openjdk-14-jre-headless'
+ARG TESSERACT_LANGUAGES='tesseract-ocr-eng tesseract-ocr-ita tesseract-ocr-fra tesseract-ocr-spa tesseract-ocr-deu'
 
-RUN DEBIAN_FRONTEND=noninteractive apt-get -y install $JRE gdal-bin tesseract-ocr \
-        tesseract-ocr-eng tesseract-ocr-ita tesseract-ocr-fra tesseract-ocr-spa tesseract-ocr-deu
+RUN DEBIAN_FRONTEND=noninteractive apt-get -y install $JRE gdal-bin tesseract-ocr $TESSERACT_LANGUAGES
 
 RUN echo ttf-mscorefonts-installer msttcorefonts/accepted-mscorefonts-eula select true | debconf-set-selections \
     && DEBIAN_FRONTEND=noninteractive apt-get install -y xfonts-utils fonts-freefont-ttf fonts-liberation ttf-mscorefonts-installer wget cabextract