Refector: Do the npm compile as a Build step

This removes the requirement of npm on the host system and removes an
additional manual step during the building of the docker image.

Updated the documentation accordingly and fixed a typo.

Dockerfile

@@ -1,3 +1,12 @@
+FROM node AS compile-frontend
+
+COPY . /src
+
+WORKDIR /src/src-ui
+RUN npm install
+RUN ./node_modules/.bin/ng build --prod
+
+
 FROM ubuntu:20.04 AS jbig2enc
 
 WORKDIR /usr/src/jbig2enc
@@ -88,7 +97,7 @@ RUN cd docker \
 COPY gunicorn.conf.py ../
 
 # copy app
-COPY src/ ./
+COPY --from=compile-frontend /src/src/ ./
 
 # add users, setup scripts
 RUN addgroup --gid 1000 paperless \

compile-frontend.sh

@@ -1,7 +0,0 @@
-#!/bin/bash
-
-set -e
-
-cd src-ui
-npm install
-./node_modules/.bin/ng build --prod

docs/setup.rst

@@ -260,9 +260,7 @@ Build the Docker image yourself
         webserver:
             build: .
 
-4.  Run the ``compile-frontend.sh`` script. This requires ``node`` and ``npm >= v15``.
-
-5.  Follow steps 3 to 8 of :ref:`setup-docker_hub`. When asked to run
+4.  Follow steps 3 to 8 of :ref:`setup-docker_hub`. When asked to run
     ``docker-compose pull`` to pull the image, do
 
     .. code:: shell-session
@@ -684,7 +682,7 @@ configuring some options in paperless can help improve performance immensely:
     your documents before feeding them into paperless. Some scanners are able to
     do this! You might want to even specify ``skip_noarchive`` to skip archive
     file generation for already ocr'ed documents entirely.
-*   If you want to perform OCR on the the device, consider using ``PAPERLESS_OCR_CLEAN=none``.
+*   If you want to perform OCR on the device, consider using ``PAPERLESS_OCR_CLEAN=none``.
     This will speed up OCR times and use less memory at the expense of slightly worse
     OCR results.
 *   Set ``PAPERLESS_OPTIMIZE_THUMBNAILS`` to 'false' if you want faster consumption