Add demo repositories mechanism to populate user's space

src/_nebari/stages/kubernetes_services/__init__.py

@@ -221,6 +221,7 @@ class IdleCuller(schema.Base):
 
 class JupyterLab(schema.Base):
     idle_culler: IdleCuller = IdleCuller()
+    initial_repositories: typing.List[typing.Dict[str, str]] = []
 
 
 class InputSchema(schema.Base):
@@ -344,6 +345,7 @@ class CondaStoreInputVars(schema.Base):
 class JupyterhubInputVars(schema.Base):
     jupyterhub_theme: Dict[str, Any] = Field(alias="jupyterhub-theme")
     jupyterlab_image: ImageNameTag = Field(alias="jupyterlab-image")
+    initial_repositories: str = Field(alias="initial-repositories")
     jupyterhub_overrides: List[str] = Field(alias="jupyterhub-overrides")
     jupyterhub_stared_storage: str = Field(alias="jupyterhub-shared-storage")
     jupyterhub_shared_endpoint: str = Field(None, alias="jupyterhub-shared-endpoint")
@@ -488,6 +490,7 @@ def input_vars(self, stage_outputs: Dict[str, Dict[str, Any]]):
             idle_culler_settings=self.config.jupyterlab.idle_culler.dict(),
             argo_workflows_enabled=self.config.argo_workflows.enabled,
             jhub_apps_enabled=self.config.jhub_apps.enabled,
+            initial_repositories=str(self.config.jupyterlab.initial_repositories),
         )
 
         dask_gateway_vars = DaskGatewayInputVars(

src/_nebari/stages/kubernetes_services/template/jupyterhub.tf

@@ -39,6 +39,11 @@ variable "jupyterlab-profiles" {
   description = "JupyterHub profiles to expose to user"
 }
 
+variable "initial-repositories" {
+  description = "Map of folder location and git repo url to clone"
+  type        = string
+}
+
 variable "jupyterhub-hub-extraEnv" {
   description = "Extracted overrides to merge with jupyterhub.hub.extraEnv"
   type        = string
@@ -129,6 +134,7 @@ module "jupyterhub" {
   jupyterhub-hub-extraEnv        = var.jupyterhub-hub-extraEnv
 
   idle-culler-settings = var.idle-culler-settings
+  initial-repositories = var.initial-repositories
 
   jupyterlab-pioneer-enabled    = var.jupyterlab-pioneer-enabled
   jupyterlab-pioneer-log-format = var.jupyterlab-pioneer-log-format

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/configmaps.tf

@@ -95,3 +95,14 @@ resource "kubernetes_config_map" "jupyterlab-settings" {
     filename => file("${path.module}/files/jupyterlab/${filename}")
   }
 }
+
+resource "kubernetes_config_map" "git_clone_update" {
+  metadata {
+    name      = "git-clone-update"
+    namespace = var.namespace
+  }
+
+  data = {
+    "git-clone-update.sh" = "${file("${path.module}/files/extras/git_clone_update.sh")}"
+  }
+}

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/files/extras/git_clone_update.sh

@@ -0,0 +1,86 @@
+#!/bin/sh
+
+################################################################################
+# Git Clone and/or Update Script
+#
+# This script automates Git repository handling with the following features:
+#
+# 1. Clones/Updates a Git repository into a specified folder;
+# 2. Creates a `.firstrun` file in the folder to mark the script's execution, ensuring it only runs once for each folder.
+#
+# Usage: ./git_clone_update.sh "<folder1> <git_repo_url1>" [...]
+#   - <folderX>: Path to the folder where the Git repository will be cloned or updated.
+#   - <git_repo_urlX>: URL of the Git repository to clone or update.
+################################################################################
+
+# Define colors for messages and command output
+GREEN='\033[0;32m'
+RED='\033[0;31m'
+YELLOW='\033[0;33m'
+NC='\033[0m'
+
+ERROR_LOG=".git-sync-errors.txt"
+
+echo -e "${GREEN}Starting execution...${NC}"
+
+# Check if at least one pair of folder and git repo URL is provided
+if [ "$#" -lt 1 ] || [ "$1" = "--help" ]; then
+  echo "Usage: $0 \"<folder1> <git_repo_url1>\" \"<folder2> <git_repo_url2>\" ..."
+
+  # Exit with status code 0 if '--help' is provided, otherwise exit with status code 1
+  [ "$1" = "--help" ] && exit 0 || exit 1
+fi
+
+clone_update_repository() {
+  # Clone or update a Git repository into a specified folder,
+  # and create a `.firstrun` file to mark the script's execution.
+
+  local folder_path="$1"
+  local git_repo_url="$2"
+
+  local firstrun_file="$folder_path/.firstrun"
+
+  if [ -f "$firstrun_file" ]; then
+    echo -e "The script has already been run for ${folder_path}. Skipping. ${GREEN}✅${NC}"
+  else
+    if [ ! -d "$folder_path" ]; then
+      mkdir -p "$folder_path"
+    fi
+
+    if [ -d "$folder_path/.git" ]; then
+      echo -e "Updating Git repository in ${folder_path}..."
+      (cd "$folder_path" && git pull)
+    else
+      echo -e "Cloning Git repository to ${folder_path}..."
+      (git clone "$git_repo_url" "$folder_path")
+    fi
+
+    touch "$firstrun_file"
+    echo -e "Execution for ${folder_path} completed. ${GREEN}✅${NC}"
+  fi
+}
+
+
+# Iterate through pairs and run in parallel
+for pair in "$@"; do
+  # Split the pair into folder_path and git_repo_url using space as the delimiter
+  folder_path=$(echo "$pair" | cut -d ' ' -f1)
+  git_repo_url=$(echo "$pair" | cut -d ' ' -f2-)
+
+  if [ -z "$folder_path" ] || [ -z "$git_repo_url" ]; then
+    # Initialize error log
+    echo -e "${RED}Invalid argument format: \"${pair}\". Please provide folder path and Git repository URL in the correct order.${NC}" >> "$ERROR_LOG"
+  else
+    clone_update_repository "$folder_path" "$git_repo_url" || echo -e "${RED}Error executing for ${folder_path}.${NC}" >> "$ERROR_LOG"
+    chown -R 1000:100 "$folder_path" # User permissions for JupyterLab user
+  fi
+done
+
+wait
+
+if [ -s "$ERROR_LOG" ]; then
+  echo -e "${RED}Some operations failed. See errors in '${ERROR_LOG}'.${NC}"
+  chown 1000:100 "$ERROR_LOG"
+else
+  echo -e "${GREEN}All operations completed successfully. ✅${NC}"
+fi

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/files/jupyterhub/03-profiles.py

@@ -1,3 +1,4 @@
+import ast
 import copy
 import functools
 import json
@@ -232,6 +233,85 @@ def base_profile_extra_mounts():
     }
 
 
+def configure_user_provisioned_repositories(username):
+    # Define paths and configurations
+    pvc_home_mount_path = f"home/{username}"
+
+    git_repos_provision_pvc = z2jh.get_config("custom.initial-repositories")
+    git_clone_update_config = {
+        "name": "git-clone-update",
+        "configMap": {"name": "git-clone-update", "defaultMode": 511},
+    }
+
+    # Convert the string configuration to a list of dictionaries
+    def string_to_objects(input_string):
+        try:
+            result = ast.literal_eval(input_string)
+            if isinstance(result, list) and all(
+                isinstance(item, dict) for item in result
+            ):
+                return result
+            else:
+                raise ValueError(
+                    "Input string does not contain a list of dictionaries."
+                )
+        except (ValueError, SyntaxError):
+            # Return an error message if the input string is not a list of dictionaries
+            raise ValueError(f"Invalid input string format: {input_string}")
+
+    git_repos_provision_pvc = string_to_objects(git_repos_provision_pvc)
+
+    if not git_repos_provision_pvc:
+        return {}
+
+    # Define the extra pod configuration for the volumes
+    extra_pod_config = {
+        "volumes": [{"name": "git-clone-update", **git_clone_update_config}]
+    }
+
+    extras_git_clone_cp_path = f"/mnt/{pvc_home_mount_path}/.git-clone-update.sh"
+
+    BASH_EXECUTION = "./.git-clone-update.sh"
+
+    for local_repo_pair in git_repos_provision_pvc:
+        for path, remote_url in local_repo_pair.items():
+            BASH_EXECUTION += f" '{path} {remote_url}'"
+
+    EXEC_OWNERSHIP_CHANGE = " && ".join(
+        [
+            f"cp /mnt/extras/git-clone-update.sh {extras_git_clone_cp_path}",
+            f"chmod 777 {extras_git_clone_cp_path}",
+            f"chown -R 1000:100 {extras_git_clone_cp_path}",
+            f"cd /mnt/{pvc_home_mount_path}",
+            BASH_EXECUTION,
+            f"rm -f {extras_git_clone_cp_path}",
+        ]
+    )
+
+    # Define init containers configuration
+    init_containers = [
+        {
+            "name": "pre-populate-git-repos",
+            "image": "bitnami/git",
+            "command": ["sh", "-c", EXEC_OWNERSHIP_CHANGE],
+            "securityContext": {"runAsUser": 0},
+            "volumeMounts": [
+                {
+                    "mountPath": f"/mnt/{pvc_home_mount_path}",
+                    "name": "home",
+                    "subPath": pvc_home_mount_path,
+                },
+                {"mountPath": "/mnt/extras", "name": "git-clone-update"},
+            ],
+        }
+    ]
+
+    return {
+        "extra_pod_config": extra_pod_config,
+        "init_containers": init_containers,
+    }
+
+
 def configure_user(username, groups, uid=1000, gid=100):
     environment = {
         # nss_wrapper
@@ -416,6 +496,7 @@ def render_profile(profile, username, groups, keycloak_profilenames):
             profile_conda_store_mounts(username, groups),
             base_profile_extra_mounts(),
             configure_user(username, groups),
+            configure_user_provisioned_repositories(username),
             profile_kubespawner_override,
         ],
         {},

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/main.tf

@@ -74,6 +74,7 @@ resource "helm_release" "jupyterhub" {
         conda-store-service-name      = var.conda-store-service-name
         conda-store-jhub-apps-token   = var.conda-store-jhub-apps-token
         jhub-apps-enabled             = var.jhub-apps-enabled
+        initial-repositories          = var.initial-repositories
         skel-mount = {
           name      = kubernetes_config_map.etc-skel.metadata.0.name
           namespace = kubernetes_config_map.etc-skel.metadata.0.namespace

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/variables.tf

@@ -167,3 +167,9 @@ variable "jupyterlab-pioneer-log-format" {
   description = "Logging format for JupyterLab Pioneer"
   type        = string
 }
+
+variable "initial-repositories" {
+  description = "Map of folder location and git repo url to clone"
+  type        = string
+  default     = "[]"
+}