paradizelost/paperless-ngx
1
0
Fork 0
mirror of https://github.com/paperless-ngx/paperless-ngx.git synced 2025-04-02 13:45:10 -05:00
Code Issues Packages Projects Releases Wiki Activity

Bakes the NLTK data into the image (~60mb)

Browse Source
This commit is contained in:
Trenton H 2022-12-02 12:02:56 -08:00
parent 3f6e3a2750
commit 8da3ae2c53
3 changed files with 5 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile
View File

@ -228,6 +228,10 @@ RUN set -eux \
&& python3 -m pip install --no-cache-dir --upgrade wheel \ && python3 -m pip install --no-cache-dir --upgrade wheel \
&& echo "Installing Python requirements" \ && echo "Installing Python requirements" \
&& python3 -m pip install --default-timeout=1000 --no-cache-dir --requirement requirements.txt \ && python3 -m pip install --default-timeout=1000 --no-cache-dir --requirement requirements.txt \
&& echo "Installing NLTK data" \
&& python3 -W ignore::RuntimeWarning -m nltk.downloader -d "/usr/local/share/nltk_data" snowball_data \
&& python3 -W ignore::RuntimeWarning -m nltk.downloader -d "/usr/local/share/nltk_data" stopwords \
&& python3 -W ignore::RuntimeWarning -m nltk.downloader -d "/usr/local/share/nltk_data" punkt \
&& echo "Cleaning up image" \ && echo "Cleaning up image" \
&& apt-get -y purge ${BUILD_PACKAGES} \ && apt-get -y purge ${BUILD_PACKAGES} \
&& apt-get -y autoremove --purge \ && apt-get -y autoremove --purge \

30
docker/docker-entrypoint.sh
View File

@ -53,30 +53,6 @@ map_folders() {
export CONSUME_DIR="${PAPERLESS_CONSUMPTION_DIR:-/usr/src/paperless/consume}" export CONSUME_DIR="${PAPERLESS_CONSUMPTION_DIR:-/usr/src/paperless/consume}"
} }
nltk_data () {
# Store the NLTK data outside the Docker container
local -r nltk_data_dir="${DATA_DIR}/nltk"
local -r truthy_things=("yes y 1 t true")
# If not set, or it looks truthy
if [[ -z "${PAPERLESS_ENABLE_NLTK}" ]] || [[ "${truthy_things[*]}" =~ ${PAPERLESS_ENABLE_NLTK,} ]]; then
# Download or update the snowball stemmer data
python3 -W ignore::RuntimeWarning -m nltk.downloader -d "${nltk_data_dir}" snowball_data
# Download or update the stopwords corpus
python3 -W ignore::RuntimeWarning -m nltk.downloader -d "${nltk_data_dir}" stopwords
# Download or update the punkt tokenizer data
python3 -W ignore::RuntimeWarning -m nltk.downloader -d "${nltk_data_dir}" punkt
else
echo "Skipping NLTK data download"
fi
}
custom_container_init() { custom_container_init() {
# Mostly borrowed from the LinuxServer.io base image # Mostly borrowed from the LinuxServer.io base image
# https://github.com/linuxserver/docker-baseimage-ubuntu/tree/bionic/root/etc/cont-init.d # https://github.com/linuxserver/docker-baseimage-ubuntu/tree/bionic/root/etc/cont-init.d
@ -157,8 +133,6 @@ initialize() {
echo "Creating directory ${tmp_dir}" echo "Creating directory ${tmp_dir}"
mkdir -p "${tmp_dir}" mkdir -p "${tmp_dir}"
nltk_data
set +e set +e
echo "Adjusting permissions of paperless files. This may take a while." echo "Adjusting permissions of paperless files. This may take a while."
chown -R paperless:paperless ${tmp_dir} chown -R paperless:paperless ${tmp_dir}
@ -191,10 +165,6 @@ install_languages() {
for lang in "${langs[@]}"; do for lang in "${langs[@]}"; do
pkg="tesseract-ocr-$lang" pkg="tesseract-ocr-$lang"
# English is installed by default
#if [[ "$lang" == "eng" ]]; then
# continue
#fi
if dpkg -s "$pkg" &>/dev/null; then if dpkg -s "$pkg" &>/dev/null; then
echo "Package $pkg already installed!" echo "Package $pkg already installed!"

2
src/paperless/settings.py
View File

@ -123,7 +123,7 @@ THUMBNAIL_DIR = os.path.join(MEDIA_ROOT, "documents", "thumbnails")
DATA_DIR = __get_path("PAPERLESS_DATA_DIR", os.path.join(BASE_DIR, "..", "data")) DATA_DIR = __get_path("PAPERLESS_DATA_DIR", os.path.join(BASE_DIR, "..", "data"))
NLTK_DIR = os.path.join(DATA_DIR, "nltk") NLTK_DIR = __get_path("PAPERLESS_NLTK_DIR", "/usr/local/share/nltk_data")
TRASH_DIR = os.getenv("PAPERLESS_TRASH_DIR") TRASH_DIR = os.getenv("PAPERLESS_TRASH_DIR")