[nix] add corenlp.nix, implement in Dockerfile

devops/docker/Dockerfile

@@ -18,32 +18,27 @@ ENV PATH=/root/.local/bin:$PATH
 RUN apt-get update && \
     apt-get install --no-install-recommends -y \
       apt-transport-https \
-      autoconf \
-      automake \
-      build-essential \
+      #autoconf \
+      #automake \
+      #build-essential \
       ca-certificates \
       curl \
-      #gcc \
       git \
       gnupg2 \
-      libffi-dev \
-      libffi8 \
-      libgmp-dev \
-      libgmp10 \
-      libncurses-dev \
-      libncurses6 \
-      libnuma-dev \
-      libtinfo6 \
+      #libffi-dev \
+      #libffi8 \
+      #libgmp-dev \
+      #libgmp10 \
+      #libncurses-dev \
+      #libncurses6 \
+      #libnuma-dev \
+      #libtinfo6 \
       locales \
-      lsb-release \
+      #lsb-release \
       software-properties-common \
-      #strace \
       sudo \
-      #wget \
-      #vim \
       xz-utils \
       zlib1g-dev \
-      #openjdk-21-jdk \
       unzip && \
     apt-get clean && rm -rf /var/lib/apt/lists/* && \
     mkdir -m 0755 /nix && groupadd -r nixbld && chown root /nix && \
@@ -55,23 +50,24 @@ SHELL ["/bin/bash", "-o", "pipefail", "-c"]
 
 COPY ./shell.nix                                       /builds/gargantext/shell.nix
 COPY ./nix/pkgs.nix                                    /builds/gargantext/nix/pkgs.nix
+COPY ./nix/corenlp.nix                                 /builds/gargantext/nix/corenlp.nix
 COPY ./nix/pinned-25.05.nix                            /builds/gargantext/nix/pinned-25.05.nix
 COPY ./devops/coreNLP/build.sh                         /root/devops/coreNLP/build.sh
-COPY ./devops/coreNLP/startServer.sh                   /root/devops/coreNLP/startServer.sh
+#COPY ./devops/coreNLP/startServer.sh                   /root/devops/coreNLP/startServer.sh
 COPY ./bin/setup-ci-environment                        /builds/gargantext/bin/setup-ci-environment
 COPY ./bin/install-cabal2stack                         /builds/gargantext/bin/install-cabal2stack
 
-#RUN cd /root/devops/coreNLP; ./build.sh && \
 RUN set -o pipefail && \
+    locale-gen en_US.UTF-8 && \
     bash <(curl -L https://releases.nixos.org/nix/nix-2.26.2/install) --no-daemon && \
-    locale-gen en_US.UTF-8 && chown root -R /nix && \
+    chown root -R /nix && \
     . "$HOME/.nix-profile/etc/profile.d/nix.sh" && \
     mkdir -p "/builds/gargantext/" && chmod 777 -R "/builds/gargantext" && \
     echo "source $HOME/.nix-profile/etc/profile.d/nix.sh" >> "$HOME/.bashrc" && \
     echo `which nix-env` && \
     . $HOME/.bashrc && nix-env --version && \
-    cd /builds/gargantext && nix-shell --run "./bin/install-cabal2stack" && \
-    cd /root/devops/coreNLP && nix-shell -p jre --run "./build.sh"
+    cd /builds/gargantext && nix-shell --run "./bin/install-cabal2stack"
+    #cd /root/devops/coreNLP && nix-shell -p jre --run "./build.sh"
 
 WORKDIR "/builds/gargantext/"
 

nix/corenlp.nix

+# https://nixos.wiki/wiki/Java
+{ fetchzip,
+  stdenv,
+  writeShellScript,
+  jre,
+
+  version ? "4.5.8",
+  hash ? "sha256-HZHNyw+LrxAN4OLueAhyxRG2onPWNMm5MwJDLLzsE2M="
+}:
+
+let
+  file_name = "stanford-corenlp-${version}.zip";
+  url = "http://nlp.stanford.edu/software/${file_name}";
+  zip_file = fetchzip { inherit url hash; };
+  start_server = writeShellScript "startCoreNLPServer.sh" ''
+    PORT=9000
+
+    while getopts ':p:h' opt; do
+      case $opt in
+        (p) PORT=$OPTARG;;
+        (h) echo "$(basename $0) [-p 9000]"
+            exit 0
+            ;;
+      esac
+    done
+
+    shift "$((OPTIND - 1))"
+
+    ${jre}/bin/java -mx4g -cp "${zip_file}/*"  edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port $PORT -timeout 15000
+  '';
+in
+
+stdenv.mkDerivation rec {
+  name = "corenlp";
+  inherit version;
+  phases = "installPhase";
+  installPhase = ''
+    mkdir -p $out/bin
+    cp ${start_server} $out/bin/startCoreNLPServer.sh
+  '';
+}

nix/pkgs.nix

@@ -88,6 +88,9 @@ rec {
     '';
 
   });
+
+  corenlp = pkgs.callPackage ./corenlp.nix { };
+  
   hsBuildInputs = [
     ghc966
     cabal_install
@@ -97,6 +100,7 @@ rec {
   ];
   nonhsBuildInputs = with pkgs; [
     bzip2
+    corenlp
     czmq
     docker-compose
     git

test/drivers/hspec/Main.hs

@@ -19,18 +19,16 @@ startCoreNLPServer :: IO ProcessHandle
 startCoreNLPServer = do
   putText "calling start core nlp"
   devNull <- openFile "/dev/null" WriteMode
-  let p = proc "./startServer.sh" []
-  (_, _, _, hdl) <- (createProcess $ p { cwd = Just "devops/coreNLP/stanford-corenlp-current"
+  let p = proc "startCoreNLPServer.sh" []
+  (_, _, _, hdl) <- (createProcess $ p {
+                        cwd = Nothing
                       , delegate_ctlc = True
                       , create_group = True
                       , std_out = UseHandle devNull
                       , std_err = UseHandle devNull
                       }) `catch` \e -> case e of
                                            _ | True <- "does not exist" `isInfixOf` (T.pack . show @SomeException $ e)
-                                             -> fail $ "Cannot execute the 'startServer.sh' script. If this is the " <>
-                                                       "first time you are running the tests, you have to run " <>
-                                                       "cd devops/coreNLP && ./build.sh first. You have to run it only once, " <>
-                                                       "and then you are good to go for the time being."
+                                             -> fail $ "Cannot execute the 'startCoreNLPServer.sh' script. Make sure you are in a nix environment."
                                              | otherwise -> throwIO e
   pure hdl