ci: activate Tika's thumbnail preset wherever Tika runs

The preset ships inert in tika-pipes-core, so PUT /unpack/preset/thumbnail
answers 404 until a config names it, and the thumbnailer then finds no
thumbnail for any document. The acceptance stack and the example
deployment now pass a config that activates it, and the deployment reads
the shared OC_TIKA_URL so the thumbnailer uses the Tika that is already
there. The acceptance image goes to 4.1, the release the preset is in.

.woodpecker.star

@@ -8,7 +8,7 @@ docker_repo_slug = "opencloudeu/opencloud"
 
 # images
 ALPINE_GIT = "alpine/git:latest"
-APACHE_TIKA = "apache/tika:4.0.0-full"
+APACHE_TIKA = "apache/tika:4.1.0-full"
 CHKO_DOCKER_PUSHRM = "chko/docker-pushrm:1"
 CODACY_COVERAGE_REPORTER = "codacy/codacy-coverage-reporter:14.1.3"
 COLLABORA_CODE = "collabora/code:24.04.5.1.1"
@@ -3447,6 +3447,17 @@ def tikaService():
         # tika 4 discovers its plugins relative to the image working directory,
         # the workspace default would leave the pipes fetchers empty
         "directory": "/opt/tika-server",
+        # the thumbnail preset ships inert; the config is what activates it
+        "entrypoint": [
+            "java",
+            "-cp",
+            "/opt/tika-server/*:/opt/tika-server/lib/*:/tika-extras/*",
+            "org.apache.tika.server.core.TikaServerCli",
+            "-h",
+            "0.0.0.0",
+            "-c",
+            "%s/tests/config/woodpecker/tika-config.json" % dirs["base"],
+        ],
         "detach": True,
     }, {
         "name": "wait-for-tika-service",

devtools/deployments/opencloud_full/config/tika/tika-config.json

@@ -0,0 +1,12 @@
+{
+  // -h and -p still come from the command line; a config file has to carry the block
+  "server": {
+    "port": 9998,
+    "host": "0.0.0.0"
+  },
+  // the thumbnail preset ships inert in tika-pipes-core; activating it is what serves
+  // PUT /unpack/preset/thumbnail, which the webdav thumbnailer asks for
+  "presets": {
+    "thumbnail": true
+  }
+}

devtools/deployments/opencloud_full/tika.yml

@@ -3,6 +3,10 @@ services:
   tika:
     image: ${TIKA_IMAGE:-apache/tika:latest-full}
     # release notes: https://tika.apache.org
+    # the thumbnail preset ships inert; the config is what activates it
+    command: ['-c', '/tika-config.json']
+    volumes:
+      - ./config/tika/tika-config.json:/tika-config.json
     networks:
       opencloud-net:
     restart: always
@@ -13,5 +17,6 @@ services:
     environment:
       # fulltext search
       SEARCH_EXTRACTOR_TYPE: tika
-      SEARCH_EXTRACTOR_TIKA_TIKA_URL: http://tika:9998
+      # the search extractor and the webdav thumbnailer both read it
+      OC_TIKA_URL: http://tika:9998
       FRONTEND_FULL_TEXT_SEARCH_ENABLED: "true"

tests/acceptance/docker/src/tika.yml

@@ -5,4 +5,7 @@ services:
       - tika
     command: tika:9998
   tika:
-    image: apache/tika:4.0.0-full
+    image: apache/tika:4.1.0-full
+    command: ['-c', '/tika-config.json']
+    volumes:
+      - ../../../config/woodpecker/tika-config.json:/tika-config.json

tests/config/woodpecker/tika-config.json

@@ -0,0 +1,12 @@
+{
+  // -h and -p still come from the command line; a config file has to carry the block
+  "server": {
+    "port": 9998,
+    "host": "0.0.0.0"
+  },
+  // the thumbnail preset ships inert in tika-pipes-core; activating it is what serves
+  // PUT /unpack/preset/thumbnail, which the webdav thumbnailer asks for
+  "presets": {
+    "thumbnail": true
+  }
+}