Merge pull request #240 from nmfs-opensci/test

rm jupyter-ai, set ubuntu to jammy not nobel, upgrade to 4.4.2, gh-scoped-creds in RStudio terminal

Author: eeholmes

Dockerfile

@@ -3,21 +3,20 @@ FROM ghcr.io/nmfs-opensci/py-rocket-base/base-image:latest
 LABEL org.opencontainers.image.maintainers="eli.holmes@noaa.gov"
 LABEL org.opencontainers.image.author="eli.holmes@noaa.gov"
 LABEL org.opencontainers.image.source=https://github.com/nmfs-opensci/py-rocket-base
-LABEL org.opencontainers.image.description="Python (3.12), R (4.4.1), Desktop and Publishing tools"
+LABEL org.opencontainers.image.description="Python (3.12), R (4.4.2), Desktop and Publishing tools"
 LABEL org.opencontainers.image.licenses=Apache2.0
-LABEL org.opencontainers.image.version=2025.02.19
+LABEL org.opencontainers.image.version=2025.03.03
 
 USER root
 
 # Define environment variables
 # DISPLAY Tell applications where to open desktop apps - this allows notebooks to pop open GUIs
+# Set QUARTO_VERSION due to Jupyter Lab bug with version 1.6 that won't all qmd to open
 ENV REPO_DIR="/srv/repo" \
     DISPLAY=":1.0" \
-    R_VERSION="4.4.1" \
-    QUARTO_VERSION="1.5.57"
-# The latest rocker will set CRAN to 'latest' but we need a date stamped version for reproducibility
-# So pull the latest and use one earlier
-ARG R_VERSION_PULL="4.4.2"
+    R_VERSION="4.4.2" \
+    QUARTO_VERSION="1.5.57" \
+    UBUNTU_VERSION="jammy"
 ENV LANG=en_US.UTF-8
 ENV LC_ALL=en_US.UTF-8
 
@@ -47,7 +46,10 @@ RUN mkdir -p /pyrocket_scripts && \
 RUN /pyrocket_scripts/install-conda-packages.sh ${REPO_DIR}/environment.yml
 
 # Install R, RStudio via Rocker scripts. Requires the prefix for a rocker Dockerfile
-RUN R_VERSION_PULL=$R_VERSION_PULL /pyrocket_scripts/install-rocker.sh "verse_${R_VERSION}"
+# Set the R_VERSION_PULL variable to specify what branch or release. If need to use a release use
+# R_VERSION_PULL="R4.3.3" for example; R_VERSION_PULL="master" is getting the current master branch
+# Be aware that if R_VERSION_PULL is set to the latest release, CRAN repo will use "latest" and date will not be pinned.
+RUN R_VERSION_PULL="master" /pyrocket_scripts/install-rocker.sh "verse_${R_VERSION}"
 
 # Install Zotero; must be run before apt since zotero apt install requires this is run first
 RUN wget -qO- https://raw.githubusercontent.com/retorquere/zotero-deb/master/install.sh | bash 
@@ -83,8 +85,9 @@ RUN mkdir -p ${XDG_CONFIG_HOME} && \
 # Fix home permissions. Not needed in JupyterHub with persistent memory but needed if not used in that context
 RUN /pyrocket_scripts/fix-home-permissions.sh
 
-# Create a symlink for python to python3 for all users
+# Create a symlink for python to python3 and gh-scoped-creds for all users; need for RStudio sinc conda not on path
 RUN ln -s /usr/bin/python3 /usr/local/bin/python
+RUN ln -s /srv/conda/envs/notebook/bin/gh-scoped-creds /usr/local/bin/gh-scoped-creds
 
 # Set up the start command 
 USER ${NB_USER}

book/r-packages.qmd

@@ -31,14 +31,15 @@ There are a few ways to get around this.
 
 ### Default CRAN repository
 
-The default CRAN repository is set in ${R_HOME}/etc/Rprofile.site and the CRAN environmental variable to
+The default CRAN repository is set in `${R_HOME}/etc/Rprofile.site` and the CRAN environmental variable is set to the date pinned version associated with the R version in the image. For example, `https://p3m.dev/cran/__linux__/jammy/2024-10-30`.
+
+To set a different repo, edit `${R_HOME}/etc/Rprofile.site` if you are changing an image or `~/.Rprofile` if setting user defaults:
 ```markdown
-https://p3m.dev/cran/__linux__/jammy/latest
+options(repos = c(CRAN = 'https://p3m.dev/cran/__linux__/jammy/latest'), download.file.method = 'libcurl')
 ```
-
-To set a different repo, edit ${R_HOME}/etc/Rprofile.site or specify the repo to use in `install.packages()` like so
+You can also specify the repo to use in `install.packages()` like so
 ```markdown
-repo <- "https://p3m.dev/cran/__linux__/jammy/2024-05-13"
+repo <- "https://p3m.dev/cran/__linux__/jammy/latest"
 list.of.packages <- c("maps", "mapdata", "RColorBrewer")
 install.packages(list.of.packages, repos=repo)
 ```

environment.yml

@@ -26,7 +26,7 @@ dependencies:
 
   # Extra Jupyter tools
   - jupyterlab-git
-  - jupyter-ai
+  #- jupyter-ai
   - jupyter-book
   - jupyter-offlinenotebook
   - jupyter-sshd-proxy

scripts/install-rocker.sh

@@ -34,13 +34,10 @@ R_DOCKERFILE="$1"
 export PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
 
 # Copy in the rocker files. Work in ${REPO_DIR} to make sure I don't clobber anything
-# R_VERSION_PULL may be different than R_VERSION. Specifically pulling latest but using R_VERSION prior to latest
-# So that CRAN repo is pinned to a date.
 cd ${REPO_DIR}
 ROCKER_DOCKERFILE_NAME="${R_DOCKERFILE}.Dockerfile"
-# Pull a tag (release) or pull the latest master (stable)
-# TAR_NAME="R${R_VERSION_PULL}"
-TAR_NAME="master"
+# Pull a tag (release) or pull the latest master (stable); R_VERSION_PULL is defined in Dockerfile when this script is called.
+TAR_NAME=${R_VERSION_PULL}
 # For degugging use: wget https://github.com/eeholmes/rocker-versioned2/archive/refs/tags/R4.4.1.tar.gz
 # wget https://github.com/rocker-org/rocker-versioned2/archive/refs/tags/R${R_VERSION_PULL}.tar.gz
 if [[ "$TAR_NAME" == "master" ]]; then
@@ -63,9 +60,15 @@ cmd=""
 while IFS= read -r line; do
     if [[ "$line" == ENV* ]]; then
         var_assignment=$(echo "$line" | sed 's/^ENV //g')
+        # Special handling for CRAN
+        if [[ "$var_assignment" =~ ^CRAN=.*__linux__/([^/]+)/(.+)$ ]]; then
+            var_assignment="CRAN=https://p3m.dev/cran/__linux__/${UBUNTU_VERSION}/${BASH_REMATCH[2]//\"/}"
+        fi
+        # Special handling for DEFAULT_USER
         if [[ "$var_assignment" == DEFAULT_USER* ]]; then
             var_assignment="DEFAULT_USER=${NB_USER}"
         fi
+        echo "Processed ENV variable: $var_assignment"
         eval "export $var_assignment"
         echo "export $var_assignment" >> "${REPO_DIR}/env.txt"
 
@@ -100,3 +103,35 @@ if command -v tlmgr &> /dev/null; then
     tlmgr install collection-latexrecommended
     tlmgr install pdfcol tcolorbox eurosym upquote adjustbox titling enumitem ulem soul rsfs
 fi
+
+# Make sure the env vars set in the rocker Dockerfile are in Renviron.site
+ENV_FILE="${REPO_DIR}/env.txt"
+RENVIRO_SITE="${R_HOME}/etc/Renviron.site"
+
+# Ensure the file exists before processing
+if [[ -f "$ENV_FILE" ]]; then
+    echo "Appending environment variables from $ENV_FILE to $RENVIRO_SITE..."
+
+    awk -F '=' '
+    /^export / {
+        gsub(/"/, "", $2);  # Remove double quotes around values
+        if ($1 ~ /PATH/) {
+            print substr($1, 8) "=\"" ENVIRON["PATH"] ":" $2 "\""
+        } else {
+            print substr($1, 8) "=\"" $2 "\""
+        }
+    }' "$ENV_FILE" > "$RENVIRO_SITE"
+
+    echo "Done."
+else
+    echo "Warning: $ENV_FILE not found. No changes made."
+fi
+
+# Ensure jovyan can modify Rprofile.site and Renviron.site because start will need to this
+# to set the gh-scoped-cred variables if they are present
+chown ${NB_USER}:staff ${R_HOME}/etc/Rprofile.site
+chmod g+w ${R_HOME}/etc/Rprofile.site
+chown ${NB_USER}:staff ${R_HOME}/etc/Renviron.site
+chmod g+w ${R_HOME}/etc/Renviron.site
+
+echo "Updated permissions for Rprofile.site and Renviron.site"

start

@@ -11,6 +11,15 @@ if [ -f "${REPO_DIR}/env.txt" ]; then
 fi
 # End - Set any environment variables here
 
+# check for gh-scoped-creds needed variables so that can use this in RStudio
+# These are based on Jupyter Hub config so will not know before server starts what these are.
+for var in GH_SCOPED_CREDS_CLIENT_ID GH_SCOPED_CREDS_APP_URL; do
+    if [[ -n "${!var:-}" ]]; then
+        echo "${var}=\"${!var}\"" >> "${R_HOME}/etc/Renviron.site"
+        echo "Added ${var} to Renviron.site"
+    fi
+done
+
 # Run child start scripts in a subshell to contain its environment
 # ${REPO_DIR}/childstart/ is created by setup-start.sh
 if [ -d "${REPO_DIR}/childstart/" ]; then