Add missing openblas and nltk_data

+ move venv to /opt/paperless

images/paperless-ngx/Dockerfile

@@ -29,12 +29,14 @@ RUN set -euxo pipefail &&\
     curl -sSLO https://github.com/paperless-ngx/paperless-ngx/releases/download/v${PAPERLESS_VERSION}/paperless-ngx-v${PAPERLESS_VERSION}.tar.xz &&\
     tar xvJf paperless-ngx-v${PAPERLESS_VERSION}.tar.xz &&\
     rm paperless-ngx-v${PAPERLESS_VERSION}.tar.xz &&\
-    python3 -m venv venv &&\
-    source venv/bin/activate &&\
     mv paperless-ngx paperless &&\
     cd paperless &&\
+    python3 -m venv venv &&\
+    source venv/bin/activate &&\
     pip --no-cache-dir install -r requirements.txt &&\
     ./src/manage.py collectstatic --no-input &&\
+    mkdir -p /usr/share/nltk_data &&\
+    python3 -c "import nltk; nltk.download('stopwords', download_dir='/usr/share/nltk_data')" &&\
     rdfind /opt
 
 FROM python:3.9-alpine
@@ -43,7 +45,7 @@ MAINTAINER [[ .docker.maintainer ]]
 ENV LANG=[[ .locale.lang ]] \
     TZ=[[ .locale.tz ]] \
     PAPERLESS_MODE=all-in-one \
-    PATH=/opt/venv/bin:${PATH} \
+    PATH=/opt/paperless/venv/bin:${PATH} \
     PAPERLESS_REDIS=redis://127.0.0.1:6379/0 \
     PAPERLESS_DBENGINE=postgresql \
     PAPERLESS_DBHOST=127.0.0.1 \
@@ -77,6 +79,7 @@ ENV LANG=[[ .locale.lang ]] \
 
 ADD https://git.lapiole.org/nomad/base_tools.git#master /
 COPY --from=builder /opt /opt
+COPY --from=builder /usr/share/nltk_data /usr/share/nltk_data
 
 RUN set -euxo pipefail &&\
     apk --no-cache add \
@@ -100,6 +103,7 @@ RUN set -euxo pipefail &&\
       libmagic \
       pngquant \
       zlib \
+      openblas \
       supervisor \
     &&\
     addgroup --gid 8994 paperless &&\

images/paperless-ngx/root/entrypoint.d/20-migrate.sh

@@ -7,7 +7,7 @@ if [ "${PAPERLESS_MODE}" != "all-in-one" -a "${PAPERLESS_MODE}" != "webserver" ]
   exit 0
 fi
 
-source /opt/venv/bin/activate
+source /opt/paperless/venv/bin/activate
 echo "Migrating database"
 cd /opt/paperless/src
 ./manage.py migrate

images/paperless-ngx/root/etc/supervisord.conf

@@ -5,7 +5,7 @@ logfile=/dev/stdout
 logfile_maxbytes=0
 
 [program:webserver]
-command=/opt/venv/bin/gunicorn -c /opt/paperless/gunicorn.conf.py --bind=%(ENV_PAPERLESS_BIND_ADDR)s:%(ENV_PAPERLESS_PORT)s paperless.asgi:application
+command=/opt/paperless/venv/bin/gunicorn -c /opt/paperless/gunicorn.conf.py --bind=%(ENV_PAPERLESS_BIND_ADDR)s:%(ENV_PAPERLESS_PORT)s paperless.asgi:application
 stdout_logfile=/proc/self/fd/1
 stdout_logfile_backups=0
 stdout_logfile_maxbytes=0
@@ -16,7 +16,7 @@ autostart=true
 autorestart=true
 
 [program:consumer]
-command=/opt/venv/bin/python3 manage.py document_consumer
+command=/opt/paperless/venv/bin/python3 manage.py document_consumer
 stdout_logfile=/proc/self/fd/1
 stdout_logfile_backups=0
 stdout_logfile_maxbytes=0
@@ -27,7 +27,7 @@ autostart=true
 autorestart=true
 
 [program:scheduler]
-command=/opt/venv/bin/celery --app paperless beat --loglevel INFO
+command=/opt/paperless/venv/bin/celery --app paperless beat --loglevel INFO
 stdout_logfile=/proc/self/fd/1
 stdout_logfile_backups=0
 stdout_logfile_maxbytes=0
@@ -38,7 +38,7 @@ autostart=true
 autorestart=true
 
 [program:task-queue]
-command=/opt/venv/bin/celery --app paperless worker --loglevel INFO
+command=/opt/paperless/venv/bin/celery --app paperless worker --loglevel INFO
 stdout_logfile=/proc/self/fd/1
 stdout_logfile_backups=0
 stdout_logfile_maxbytes=0

images/paperless-ngx/root/usr/local/bin/paperless

@@ -2,21 +2,21 @@
 
 set -euo pipefail
 
-source /opt/venv/bin/activate
+source /opt/paperless/venv/bin/activate
 
 cd /opt/paperless/src
 
 if [ "${PAPERLESS_MODE}" = "all-in-one" ]; then
   exec supervisor -c /etc/supervisord.conf -n
 elif [ "${PAPERLESS_MODE}" = "webserver" ]; then
-  exec /opt/venv/bin/gunicorn \
+  exec /opt/paperless/venv/bin/gunicorn \
          -c /opt/paperless/gunicorn.conf.py \
          --bind=${PAPERLESS_BIND_ADDR}:${PAPERLESS_PORT} \
          paperless.asgi:application 
 elif [ "${PAPERLESS_MODE}" = "consumer" ]; then
-  exec /opt/venv/bin/python3 manage.py document_consumer
+  exec /opt/paperless/venv/bin/python3 manage.py document_consumer
 elif [ "${PAPERLESS_MODE}" = "scheduler" ]; then
-  exec /opt/venv/bin/celery --app paperless beat --loglevel INFO
+  exec /opt/paperless/venv/bin/celery --app paperless beat --loglevel INFO
 elif [ "${PAPERLESS_MODE}" = "task-queue" ]; then
-  exec /opt/venv/bin/celery --app paperless worker --loglevel INFO
+  exec /opt/paperless/venv/bin/celery --app paperless worker --loglevel INFO
 fi

variables.yml

@@ -31,7 +31,7 @@ paperless:
   webserver:
 
     # The Docker image to use
-    image: danielberteaud/paperless-ngx:1.17.4-1
+    image: danielberteaud/paperless-ngx:1.17.4-4
 
     # Env vars to set in the container
     env: {}