From 70af9c12fc460b6c502ff28380429bb95d2eda95 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Pawe=C5=82=20Maciusiak?=
 <158472457+pmacius@users.noreply.github.com>
Date: Mon, 5 Aug 2024 15:08:43 +0200
Subject: [PATCH] OPSEXP-2726 Add T-Engine Tika image Dockerfile (#18)

---
 .github/workflows/build_and_test.yml |  6 ++++++
 .gitignore                           |  2 +-
 docker-bake.hcl                      | 32 ++++++++++++++++++++++++++--
 tengine/imagemagick/Dockerfile       |  4 ++--
 tengine/imagemagick/entrypoint.sh    |  2 +-
 tengine/libreoffice/Dockerfile       |  4 ++--
 tengine/tika/Dockerfile              | 30 ++++++++++++++++++++++++++
 tengine/tika/README.md               | 22 +++++++++++++++++++
 tengine/tika/artifacts.json          | 14 ++++++++++++
 tengine/tika/entrypoint.sh           |  2 ++
 test/helm/test-overrides.yaml        | 13 ++++++-----
 11 files changed, 118 insertions(+), 13 deletions(-)
 create mode 100644 tengine/tika/Dockerfile
 create mode 100644 tengine/tika/README.md
 create mode 100644 tengine/tika/artifacts.json
 create mode 100755 tengine/tika/entrypoint.sh

diff --git a/.github/workflows/build_and_test.yml b/.github/workflows/build_and_test.yml
index 3d60c022..3a85dc99 100644
--- a/.github/workflows/build_and_test.yml
+++ b/.github/workflows/build_and_test.yml
@@ -37,6 +37,8 @@ jobs:
       - name: Checkout
         uses: actions/checkout@692973e3d937129bcbf40652eb9f2f61becf3332 # v4.1.7
 
+      - uses: Alfresco/alfresco-build-tools/.github/actions/free-hosted-runner-disk-space@v5.34.0
+
       - name: Setup nexus authentication
         run: |
           echo "machine nexus.alfresco.com" >> ~/.netrc
@@ -74,6 +76,10 @@ jobs:
         run: |
           docker save -o /tmp/${{ env.ARTIFACT_NAME }}.tar $(docker images --format "{{.Repository}}:{{.Tag}}" | grep alfresco)
 
+      - name: Check disk space
+        run: df -h
+        if: always()
+
       - name: Upload images as artifact
         uses: actions/upload-artifact@0b2256b8c012f0828dc542b3febcab082c67f72b # v4.3.4
         with:
diff --git a/.gitignore b/.gitignore
index 14d0d311..78355de2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,5 +4,5 @@ dist
 *.jar
 *.rpm
 *.gz
-
+*.tgz
 test/helm/enterprise-integration-test-values.yaml
diff --git a/docker-bake.hcl b/docker-bake.hcl
index f48491e7..784b9592 100644
--- a/docker-bake.hcl
+++ b/docker-bake.hcl
@@ -15,7 +15,7 @@ group "ats" {
 }
 
 group "tengines" {
-  targets = ["tengine_libreoffice", "tengine_imagemagick"]
+  targets = ["tengine_libreoffice", "tengine_imagemagick", "tengine_tika"]
 }
 
 variable "LABEL_VENDOR" {
@@ -310,7 +310,7 @@ variable "ALFRESCO_LIBREOFFICE_USER_NAME" {
 }
 
 variable "ALFRESCO_LIBREOFFICE_USER_ID" {
-  default = "33002"
+  default = "33003"
 }
 
 target "tengine_libreoffice" {
@@ -332,3 +332,31 @@ target "tengine_libreoffice" {
   tags = ["localhost/alfresco-libreoffice:latest"]
   output = ["type=docker"]
 }
+
+variable "ALFRESCO_TIKA_USER_NAME" {
+  default = "tika"
+}
+
+variable "ALFRESCO_TIKA_USER_ID" {
+  default = "33004"
+}
+
+target "tengine_tika" {
+  dockerfile = "./tengine/tika/Dockerfile"
+  inherits = ["java_base"]
+  contexts = {
+    java_base = "target:java_base"
+  }
+  args = {
+    ALFRESCO_TIKA_GROUP_NAME = "${ALFRESCO_GROUP_NAME}"
+    ALFRESCO_TIKA_GROUP_ID = "${ALFRESCO_GROUP_ID}"
+    ALFRESCO_TIKA_USER_NAME = "${ALFRESCO_TIKA_USER_NAME}"
+    ALFRESCO_TIKA_USER_ID = "${ALFRESCO_TIKA_USER_ID}"
+  }
+  labels = {
+    "org.opencontainers.image.title" = "${PRODUCT_LINE} Transform Engine Tika"
+    "org.opencontainers.image.description" = "Alfresco Transform Engine Tika"
+  }
+  tags = ["localhost/alfresco-tika:latest"]
+  output = ["type=docker"]
+}
diff --git a/tengine/imagemagick/Dockerfile b/tengine/imagemagick/Dockerfile
index 83c7ff0d..55c8dd13 100644
--- a/tengine/imagemagick/Dockerfile
+++ b/tengine/imagemagick/Dockerfile
@@ -11,7 +11,7 @@ ARG ALFRESCO_IMAGEMAGICK_GROUP_ID
 ARG ALFRESCO_IMAGEMAGICK_USER_NAME
 ARG ALFRESCO_IMAGEMAGICK_USER_ID
 
-ADD ${IMAGEMAGICK_COMMON_PATH}/alfresco-transform-imagemagick*.jar /usr/bin/app.jar
+ADD ${IMAGEMAGICK_COMMON_PATH}/alfresco-transform-imagemagick*.jar /opt/app.jar
 ADD ${IMAGEMAGICK_COMMON_PATH}/imagemagick-distribution*-el9.rpm /tmp/imagemagick-distribution-el9.rpm
 ADD ${IMAGEMAGICK_COMMON_PATH}/libs/imagemagick-distribution*-libs-el9.rpm /tmp/imagemagick-distribution-libs-el9.rpm
 
@@ -21,7 +21,7 @@ RUN yum install -y $IMAGEMAGICK_DEP_RPM_URL && \
 
 RUN groupadd -g ${ALFRESCO_IMAGEMAGICK_GROUP_ID} ${ALFRESCO_IMAGEMAGICK_GROUP_NAME} && \
     useradd -u ${ALFRESCO_IMAGEMAGICK_USER_ID} -g ${ALFRESCO_IMAGEMAGICK_GROUP_NAME} ${ALFRESCO_IMAGEMAGICK_USER_NAME} && \
-    chgrp ${ALFRESCO_IMAGEMAGICK_GROUP_NAME} /usr/bin/app.jar
+    chgrp ${ALFRESCO_IMAGEMAGICK_GROUP_NAME} /opt/app.jar
 
 USER ${ALFRESCO_IMAGEMAGICK_USER_NAME}
 
diff --git a/tengine/imagemagick/entrypoint.sh b/tengine/imagemagick/entrypoint.sh
index 9837e5a3..1f50ab1c 100755
--- a/tengine/imagemagick/entrypoint.sh
+++ b/tengine/imagemagick/entrypoint.sh
@@ -1,3 +1,3 @@
 #!/bin/bash
 
-exec java $JAVA_OPTS $JAVA_OPTS_CONTAINER_FLAGS -jar /usr/bin/app.jar
+exec java $JAVA_OPTS $JAVA_OPTS_CONTAINER_FLAGS -jar /opt/app.jar
diff --git a/tengine/libreoffice/Dockerfile b/tengine/libreoffice/Dockerfile
index d493ba41..1a262cca 100644
--- a/tengine/libreoffice/Dockerfile
+++ b/tengine/libreoffice/Dockerfile
@@ -18,11 +18,11 @@ RUN --mount=type=bind,source=${LIBREOFFICE_COMMON_PATH},target=/dist \
     rm -rf /tmp/dist && \
     yum clean all
 
-ADD ${LIBREOFFICE_COMMON_PATH}/alfresco-transform-libreoffice*.jar /usr/bin/app.jar
+ADD ${LIBREOFFICE_COMMON_PATH}/alfresco-transform-libreoffice*.jar /opt/app.jar
 
 RUN groupadd -g ${ALFRESCO_LIBREOFFICE_GROUP_ID} ${ALFRESCO_LIBREOFFICE_GROUP_NAME} && \
     useradd -u ${ALFRESCO_LIBREOFFICE_USER_ID} -g ${ALFRESCO_LIBREOFFICE_GROUP_NAME} ${ALFRESCO_LIBREOFFICE_USER_NAME} && \
-    chgrp ${ALFRESCO_LIBREOFFICE_GROUP_NAME} /usr/bin/app.jar
+    chgrp ${ALFRESCO_LIBREOFFICE_GROUP_NAME} /opt/app.jar
 
 USER ${ALFRESCO_LIBREOFFICE_USER_NAME}
 
diff --git a/tengine/tika/Dockerfile b/tengine/tika/Dockerfile
new file mode 100644
index 00000000..84b96bdd
--- /dev/null
+++ b/tengine/tika/Dockerfile
@@ -0,0 +1,30 @@
+FROM java_base
+
+ENV JAVA_OPTS_CONTAINER_FLAGS=-XX:MaxRAMPercentage=80
+ENV TIKA_COMMON_PATH=tengine/tika
+ENV TIKA_DEP_RPM_URL=https://dl.fedoraproject.org/pub/epel/epel-release-latest-9.noarch.rpm
+
+EXPOSE 8090
+
+ARG ALFRESCO_TIKA_GROUP_NAME
+ARG ALFRESCO_TIKA_GROUP_ID
+ARG ALFRESCO_TIKA_USER_NAME
+ARG ALFRESCO_TIKA_USER_ID
+
+RUN yum install -y ${TIKA_DEP_RPM_URL} && \
+    yum -y install perl-Image-ExifTool && \
+    yum clean all
+
+ADD ${TIKA_COMMON_PATH}/alfresco-transform-tika*.jar /opt/app.jar
+
+RUN groupadd -g ${ALFRESCO_TIKA_GROUP_ID} ${ALFRESCO_TIKA_GROUP_NAME} && \
+    useradd -u ${ALFRESCO_TIKA_USER_ID} -g ${ALFRESCO_TIKA_GROUP_NAME} ${ALFRESCO_TIKA_USER_NAME} && \
+    chgrp ${ALFRESCO_TIKA_GROUP_NAME} /opt/app.jar
+
+USER ${ALFRESCO_TIKA_USER_NAME}
+
+HEALTHCHECK --interval=20s --timeout=10s --retries=3 --start-period=30s \
+  CMD curl -f http://localhost:8090/live || exit 1
+
+COPY ${TIKA_COMMON_PATH}/entrypoint.sh /entrypoint.sh
+CMD ["/entrypoint.sh"]
diff --git a/tengine/tika/README.md b/tengine/tika/README.md
new file mode 100644
index 00000000..203782f2
--- /dev/null
+++ b/tengine/tika/README.md
@@ -0,0 +1,22 @@
+# Runtime variables
+
+Sets of variables configurable with your docker image
+
+## tika
+
+```yaml
+
+transform-core-tika:
+    image: localhost/alfresco-tika:YOUR-TAG
+    environment:
+      JAVA_OPTS: 
+      ACTIVEMQ_URL: nio://activemq:61616
+      ACTIVEMQ_USER: admin  
+      ACTIVEMQ_PASSWORD: admin
+      FILE_STORE_URL: http://shared-file-store:8099/alfresco/api/-default-/private/sfs/versions/1/file
+
+```
+
+- `JAVA_OPTS` - Additional java options
+- `ACTIVEMQ_URL` - The URL for Alfresco ActiveMQ.
+- `FILE_STORE_URL` -  Alfresco Shared FileStore endpoint.
diff --git a/tengine/tika/artifacts.json b/tengine/tika/artifacts.json
new file mode 100644
index 00000000..452776ac
--- /dev/null
+++ b/tengine/tika/artifacts.json
@@ -0,0 +1,14 @@
+{
+  "artifacts": {
+    "acs23": [
+      {
+        "name": "alfresco-transform-tika",
+        "version": "5.1.3",
+        "path": "tengine/tika",
+        "classifier": ".jar",
+        "repository": "releases",
+        "group": "org.alfresco"
+      }
+    ]
+  }
+}
diff --git a/tengine/tika/entrypoint.sh b/tengine/tika/entrypoint.sh
new file mode 100755
index 00000000..158a2a1e
--- /dev/null
+++ b/tengine/tika/entrypoint.sh
@@ -0,0 +1,2 @@
+#!/bin/bash
+exec java $JAVA_OPTS $JAVA_OPTS_CONTAINER_FLAGS -jar /opt/app.jar
diff --git a/test/helm/test-overrides.yaml b/test/helm/test-overrides.yaml
index 8a279185..7ada7d0d 100644
--- a/test/helm/test-overrides.yaml
+++ b/test/helm/test-overrides.yaml
@@ -24,11 +24,14 @@ alfresco-transform-service:
     image:
       repository: localhost/alfresco-imagemagick
       tag: latest
-  #libreoffice:
-    #image:
-    #  repository: localhost/alfresco-libreoffice
-    #  tag: latest
-  #tika:
+  # libreoffice:
+  #   image:
+  #     repository: localhost/alfresco-libreoffice
+  #     tag: latest
+  tika:
+    image:
+      repository: localhost/alfresco-tika
+      tag: latest
   #transformmisc:
   transformrouter:
     image: