Add jupyterlab-pioneer (#2127)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Chuck McAndrew <[email protected]>

src/_nebari/stages/kubernetes_services/__init__.py

@@ -196,6 +196,15 @@ class Monitoring(schema.Base):
     enabled: bool = True
 
 
+class JupyterLabPioneer(schema.Base):
+    enabled: bool = False
+    log_format: typing.Optional[str] = None
+
+
+class Telemetry(schema.Base):
+    jupyterlab_pioneer: JupyterLabPioneer = JupyterLabPioneer()
+
+
 class JupyterHub(schema.Base):
     overrides: typing.Dict = {}
 
@@ -284,6 +293,7 @@ class InputSchema(schema.Base):
     conda_store: CondaStore = CondaStore()
     argo_workflows: ArgoWorkflows = ArgoWorkflows()
     monitoring: Monitoring = Monitoring()
+    telemetry: Telemetry = Telemetry()
     jupyterhub: JupyterHub = JupyterHub()
     jupyterlab: JupyterLab = JupyterLab()
     jhub_apps: JHubApps = JHubApps()
@@ -354,6 +364,13 @@ class MonitoringInputVars(schema.Base):
     monitoring_enabled: bool = Field(alias="monitoring-enabled")
 
 
+class TelemetryInputVars(schema.Base):
+    jupyterlab_pioneer_enabled: bool = Field(alias="jupyterlab-pioneer-enabled")
+    jupyterlab_pioneer_log_format: typing.Optional[str] = Field(
+        alias="jupyterlab-pioneer-log-format"
+    )
+
+
 class ArgoWorkflowsInputVars(schema.Base):
     argo_workflows_enabled: bool = Field(alias="argo-workflows-enabled")
     argo_workflows_overrides: List[str] = Field(alias="argo-workflows-overrides")
@@ -484,6 +501,11 @@ def input_vars(self, stage_outputs: Dict[str, Dict[str, Any]]):
             monitoring_enabled=self.config.monitoring.enabled,
         )
 
+        telemetry_vars = TelemetryInputVars(
+            jupyterlab_pioneer_enabled=self.config.telemetry.jupyterlab_pioneer.enabled,
+            jupyterlab_pioneer_log_format=self.config.telemetry.jupyterlab_pioneer.log_format,
+        )
+
         argo_workflows_vars = ArgoWorkflowsInputVars(
             argo_workflows_enabled=self.config.argo_workflows.enabled,
             argo_workflows_overrides=[json.dumps(self.config.argo_workflows.overrides)],
@@ -499,6 +521,7 @@ def input_vars(self, stage_outputs: Dict[str, Dict[str, Any]]):
             **dask_gateway_vars.dict(by_alias=True),
             **monitoring_vars.dict(by_alias=True),
             **argo_workflows_vars.dict(by_alias=True),
+            **telemetry_vars.dict(by_alias=True),
         }
 
     def check(

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

@@ -130,4 +130,6 @@ module "jupyterhub" {
 
   idle-culler-settings = var.idle-culler-settings
 
+  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

@@ -11,12 +11,24 @@ locals {
   )
 }
 
+locals {
+  jupyter-pioneer-config-py-template = templatefile("${path.module}/files/jupyter/jupyter_jupyterlab_pioneer_config.py.tpl", {
+    log_format = var.jupyterlab-pioneer-log-format != null ? var.jupyterlab-pioneer-log-format : ""
+    }
+  )
+}
+
 
 resource "local_file" "jupyter_server_config_py" {
   content  = local.jupyter-notebook-config-py-template
   filename = "${path.module}/files/jupyter/jupyter_server_config.py"
 }
 
+resource "local_file" "jupyter_jupyterlab_pioneer_config_py" {
+  content  = local.jupyter-pioneer-config-py-template
+  filename = "${path.module}/files/jupyter/jupyter_jupyterlab_pioneer_config.py"
+}
+
 
 resource "kubernetes_config_map" "etc-ipython" {
   metadata {
@@ -31,19 +43,31 @@ resource "kubernetes_config_map" "etc-ipython" {
 }
 
 
+locals {
+  etc-jupyter-config-data = merge(
+    {
+      "jupyter_server_config.py" = local_file.jupyter_server_config_py.content,
+    },
+    var.jupyterlab-pioneer-enabled ? {
+      # quotes are must here, as terraform would otherwise think py is a property of
+      # a defined resource jupyter_jupyterlab_pioneer_config
+      "jupyter_jupyterlab_pioneer_config.py" = local_file.jupyter_jupyterlab_pioneer_config_py.content
+    } : {}
+  )
+}
+
 resource "kubernetes_config_map" "etc-jupyter" {
   depends_on = [
-    local_file.jupyter_server_config_py
+    local_file.jupyter_server_config_py,
+    local_file.jupyter_jupyterlab_pioneer_config_py
   ]
 
   metadata {
     name      = "etc-jupyter"
     namespace = var.namespace
   }
 
-  data = {
-    "jupyter_server_config.py" : local_file.jupyter_server_config_py.content
-  }
+  data = local.etc-jupyter-config-data
 }
 
 

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/jupyterhub/files/jupyter/jupyter_jupyterlab_pioneer_config.py.tpl

@@ -0,0 +1,60 @@
+import logging
+import json
+
+
+default_log_format = "%(asctime)s %(levelname)9s %(lineno)4s %(module)s: %(message)s"
+log_format = ${log_format}
+
+logging.basicConfig(
+    level=logging.INFO,
+    format=log_format if log_format else default_log_format
+)
+
+logger = logging.getLogger(__name__)
+
+CUSTOM_EXPORTER_NAME = "MyCustomExporter"
+
+
+def my_custom_exporter(args):
+    """Custom exporter to log JupyterLab events to command line."""
+    logger.info(json.dumps(args.get("data")))
+    return {
+        "exporter": CUSTOM_EXPORTER_NAME,
+        "message": ""
+    }
+
+
+c.JupyterLabPioneerApp.exporters = [
+    {
+        # sends telemetry data to the browser console
+        "type": "console_exporter",
+    },
+    {
+        # sends telemetry data (json) to the python console jupyter is running on
+        "type": "custom_exporter",
+        "args": {
+            "id": CUSTOM_EXPORTER_NAME
+            # add additional args for your exporter function here
+        },
+    }
+]
+
+c.JupyterLabPioneerApp.custom_exporter = {
+    CUSTOM_EXPORTER_NAME: my_custom_exporter,
+}
+
+c.JupyterLabPioneerApp.activeEvents = [
+    {"name": "ActiveCellChangeEvent", "logWholeNotebook": False},
+    {"name": "CellAddEvent", "logWholeNotebook": False},
+    {"name": "CellEditEvent", "logWholeNotebook": False},
+    {"name": "CellExecuteEvent", "logWholeNotebook": False},
+    {"name": "CellRemoveEvent", "logWholeNotebook": False},
+    {"name": "ClipboardCopyEvent", "logWholeNotebook": False},
+    {"name": "ClipboardCutEvent", "logWholeNotebook": False},
+    {"name": "ClipboardPasteEvent", "logWholeNotebook": False},
+    {"name": "NotebookHiddenEvent", "logWholeNotebook": False},
+    {"name": "NotebookOpenEvent", "logWholeNotebook": False},
+    {"name": "NotebookSaveEvent", "logWholeNotebook": False},
+    {"name": "NotebookScrollEvent", "logWholeNotebook": False},
+    {"name": "NotebookVisibleEvent", "logWholeNotebook": False},
+]

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

@@ -157,3 +157,13 @@ variable "argo-workflows-enabled" {
   description = "Enable Argo Workflows"
   type        = bool
 }
+
+variable "jupyterlab-pioneer-enabled" {
+  description = "Enable JupyterLab Pioneer for telemetry"
+  type        = bool
+}
+
+variable "jupyterlab-pioneer-log-format" {
+  description = "Logging format for JupyterLab Pioneer"
+  type        = string
+}

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

@@ -39,12 +39,20 @@ variable "conda-store-default-namespace" {
   type        = string
 }
 
-
 variable "argo-workflows-enabled" {
   description = "Enable Argo Workflows"
   type        = bool
 }
 
+variable "jupyterlab-pioneer-enabled" {
+  description = "Enable JupyterLab Pioneer for telemetry"
+  type        = bool
+}
+
+variable "jupyterlab-pioneer-log-format" {
+  description = "Logging format for JupyterLab Pioneer"
+  type        = string
+}
 variable "jhub-apps-enabled" {
   description = "Enable JupyterHub Apps"
   type        = bool