Dockerfile: use a shell script to create a host UID/GID-based container user (#27)

Author: stanislaw

Dockerfile

@@ -4,6 +4,7 @@ FROM ubuntu:24.04
 RUN apt-get update && apt-get install -y \
     curl \
     git \
+    gosu \
     python3 \
     python3-pip \
     python3-venv \
@@ -18,15 +19,6 @@ RUN wget -q -O google-chrome.deb https://dl.google.com/linux/direct/google-chrom
     && apt-get install -y ./google-chrome.deb \
     && rm google-chrome.deb
 
-# Create a new non-root user and group.
-# NOTE: It is important that a non-root user is used because otherwise the
-# Chrome Driver fails with: "User data directory is already in use"
-# https://github.com/SeleniumHQ/selenium/issues/15327#issuecomment-2688613182
-RUN groupadd -r html2print && useradd -r -m -g html2print html2print
-
-# Grant the new user sudo privileges.
-RUN echo "html2print ALL=(ALL) NOPASSWD:ALL" > /etc/sudoers.d/html2print
-
 # Create a virtual environment in the user's home directory.
 RUN python3 -m venv /opt/venv
 
@@ -47,9 +39,14 @@ RUN if [ "$HTML2PRINT_SOURCE" = "pypi" ]; then \
     fi; \
     chmod -R 777 /opt/venv;
 
-USER html2print
+# Remove the default 'ubuntu' user.
+RUN userdel -r ubuntu 2>/dev/null || true
+
+# Allow updating the UID/GID dynamically at runtime
+COPY entrypoint.sh /entrypoint.sh
+RUN chmod +x /entrypoint.sh
 
 # Set the working directory to the user's home directory.
 WORKDIR /data
 
-ENTRYPOINT ["/bin/bash"]
+ENTRYPOINT ["/entrypoint.sh"]

entrypoint.sh

@@ -0,0 +1,55 @@
+#!/bin/sh
+
+# This custom entrypoint script is needed for creating a container user with
+# a host's UID/GUI which enables sharing of the files between the container
+# and the host.
+# NOTE: It is important that a non-root user is used because otherwise the
+# Chrome Driver fails with: "User data directory is already in use"
+# https://github.com/SeleniumHQ/selenium/issues/15327#issuecomment-2688613182
+
+set -e
+
+# Ensure we have the environment variables
+if [ -z "$HOST_UID" ] || [ -z "$HOST_GID" ]; then
+    echo "HOST_UID and HOST_GID must be set!"
+    exit 1
+fi
+
+echo "html2print/docker: running a Docker container entrypoint."
+echo "html2print/docker: ensuring html2print user with UID=$HOST_UID and GID=$HOST_GID exists"
+
+# Check if a user with this UID already exists (e.g., "ubuntu")
+EXISTING_USER=$(getent passwd "$HOST_UID" | cut -d: -f1)
+
+if [ -n "$EXISTING_USER" ]; then
+    echo "error: html2print/docker: detected a wrong user: '$EXISTING_USER'. Ensure that any default users are removed from the Dockerfile. This entrypoint script is supposed to create a new user 'html2print'."
+    exit 1
+else
+    # Ensure the group exists.
+    EXISTING_GROUP=$(getent group "$HOST_GID" | cut -d: -f1)
+    if [ -z "$EXISTING_GROUP" ]; then
+        echo "html2print/docker: creating new group html2print with GID=$HOST_GID"
+        groupadd -g "$HOST_GID" html2print
+    else
+        echo "html2print/docker: group with GID=$HOST_GID already exists: $EXISTING_GROUP, reusing it."
+    fi
+
+    # Create the user.
+    echo "html2print/docker: creating new user html2print with UID=$HOST_UID"
+    useradd -m -u "$HOST_UID" -g "$HOST_GID" -s /bin/bash html2print
+
+    # Give the user root privileges. Useful for debugging.
+    echo "html2print ALL=(ALL) NOPASSWD:ALL" > /etc/sudoers.d/html2print
+fi
+
+echo "html2print/docker: show created user info:"
+id html2print
+
+# Run as the correct user. If no command is provided, run a shell.
+if [ $# -eq 0 ]; then
+    echo "html2print/docker: no command provided, opening an interactive shell."
+    exec gosu html2print /bin/bash
+else
+    # Otherwise, run the provided command.
+    exec gosu html2print "$@"
+fi

tasks.py

@@ -296,7 +296,6 @@ def run_docker(
     command_argument = (
         f'/bin/bash -c "{command}"' if command is not None else ""
     )
-    entry_point_argument = '--entrypoint=""' if command_argument else ""
 
     run_invoke(
         context,
@@ -305,8 +304,8 @@ def run_docker(
             --name html2print
             --rm
             -it
-            -v "$(pwd):/data"
-            {entry_point_argument}
+            -e HOST_UID=$(id -u) -e HOST_GID=$(id -g)
+            -v "$(pwd):/data:rw"
             {image}
             {command_argument}
         """,
@@ -326,6 +325,6 @@ def test_docker(context, image: str = "html2print:latest"):
         context,
         image=image,
         command=(
-            "cd tests/integration/01_hello_world && html2print print --debug index.html /data/output/index.pdf && cat /tmp/chromedriver.log"
+            "cd tests/integration/01_hello_world && html2print print index.html /data/output/index.pdf"
         ),
     )