run-docker.sh

#!/usr/bin/env bash

# This command tries to start up your searchfox docker container, creating it
# from the image created in `build.docker.sh` if it doesn't already exist, and
# ensure you get a distinct bash shell in the container even if you run this
# script in multiple tabs[1].
#
# This command will attempt to generate automatic mounts for any symlinks found
# in the "trees" subdirectory.  This is intended as a means of letting you
# index local trees by symlinking them.
#
# 1: We definer the container as running a canonical bash shell.  When we run
# `docker container start CONTAINER` this spins up the shell, but you don't get
# your terminal bound to the shell until we run `docker attach CONTAINER`.
# However, if we naively run `docker attach CONTAINER` in 2 separate terminals,
# you'll just end up seeing the same single underlying bash instance mirrored to
# both terminals.  This is almost certainly not what you want, so we will
# instead run an additional bash shell in the container.

# ==============================================================================

# If you have multiple searchfox checkouts on your machine and you want to be
# able to have them operate in isolation, you can define the variables below.
# I would suggest adding them to the appropriate env/bin/activate script that
# you use; `activate` is for bash but there are different suffixed files for
# other shell types.
CONTAINER_NAME=${SEARCHFOX_DOCKER_CONTAINER_NAME:-searchfox}
IMAGE_NAME=${SEARCHFOX_DOCKER_IMAGE_NAME:-searchfox}
# Note that the volume is optional!  Also, we suffix it with "-vol" because I
# saw it in the docs and that seems reasonable to avoid having everything be
# named like exactly the same.
VOLUME_NAME=${SEARCHFOX_DOCKER_VOLUME_NAME:-searchfox-vol}

THIS_DIR=$(pwd)
# For consistency we mount the source dir at /vagrant still
INSIDE_CONTAINER_DIR=/vagrant

# connect to this port on your computer
OUTSIDE_CONTAINER_PORT=16995
# which is served at this port inside the container
INSIDE_CONTAINER_PORT=80

SHELL=bash

container_exists() {
    docker container inspect ${CONTAINER_NAME} &> /dev/null
}

volume_exists() {
    docker volume inspect ${VOLUME_NAME} &> /dev/null
}

if container_exists; then
    CONTAINER_STATE=$(docker container inspect ${CONTAINER_NAME} | jq -r '.[0].State.Status')
    # If it's already running, run a new bash command inside the container
    if [[ $CONTAINER_STATE == "running" ]]; then
        docker exec -it ${CONTAINER_NAME} ${SHELL}
    else # start the (already created) container and attach to its canonical shell
        # this will print out our container name if we don't redirect stdout
        docker container start ${CONTAINER_NAME} > /dev/null
        docker attach ${CONTAINER_NAME}
    fi
else
    # build list of additional mounts; we use process substitution because
    # piping would create a subshell; see https://mywiki.wooledge.org/BashFAQ/024
    #
    # Note that the following currently works, but it doesn't work like I
    # intended it to work.  Like, the symlink ends up working because we end up
    # mounting the actual path at its true path in the real world.  I, uh,
    # don't understand what's actually going on, although I'm guessing it
    # doesn't like trying to create a mount-point with a mount-point and is
    # ignoring the target we're specifying and using the source.
    LINKMOUNTS=()
    while read -r link; do
      LINKMOUNTS+=( --mount type=bind,source=$(readlink -f ${link}),target=/vagrant/${link} )
    done < <(find trees -type l)

    # Mount the home directory volume if it exists.  The docker docs say that
    # if there is anything already at that location prior to us passing this
    # directive, it will be copied into the volume.
    VOLMOUNTS=()
    if volume_exists; then
      VOLMOUNTS+=( --mount type=volume,source=${VOLUME_NAME},target=/home/vagrant )
    fi

    # flags:
    # - `-it`: `i` is interactive, `t` is allocate a pseudo-tty
    # - `--name`: controls the name that is used to refer to the container for other
    #   commands.  For example, `docker container stop $NAME` and
    #   `docker container rm $NAME`.
    # - `--mount`: lets us bind the current directory into the container.
    # - `-p`: specifies the port mapping to expose the nginx web-server (when
    #   running; it doesn't automatically run!) on localhost port 16995.
    docker run \
        --platform linux/amd64 \
        -it \
        --name $CONTAINER_NAME \
        --mount type=bind,source=${THIS_DIR},target=${INSIDE_CONTAINER_DIR} \
        "${LINKMOUNTS[@]}" \
        "${VOLMOUNTS[@]}" \
        -p ${OUTSIDE_CONTAINER_PORT}:${INSIDE_CONTAINER_PORT} \
        ${IMAGE_NAME} \
        ${SHELL}
fi