Capture more information when running cells

src/gdpr.ts

@@ -778,6 +778,14 @@
      "startKernel": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"% Time spent in starting the kernel.","owner":"donjayamanne","isMeasurement":true},
      "startupCode": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"% Time spent in executing startup code.","owner":"donjayamanne","isMeasurement":true},
      "updateConnection": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"% Time spent in updating the kernel connection.","owner":"donjayamanne","isMeasurement":true},
+     "codeCellCount": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Number of code cells in the notebook.","owner":"donjayamanne","isMeasurement":true},
+     "mdCellCount": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Number of md cells in the notebook.","owner":"donjayamanne","isMeasurement":true},
+     "codeCellCharLength": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total char length of all text in all code cells.","owner":"donjayamanne","isMeasurement":true},
+     "mdCellCharLength": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total char length of all text in all md cells.","owner":"donjayamanne","isMeasurement":true},
+     "outputCount": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total number of outputs in all cells.","owner":"donjayamanne","isMeasurement":true},
+     "outputsByteSize": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total bytes of all outputs in all cells.","owner":"donjayamanne","isMeasurement":true},
+     "attachmentCount": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total number of attachments","owner":"donjayamanne","isMeasurement":true},
+     "attachmentCharLength": {"classification":"SystemMetaData","purpose":"FeatureInsight","comment":"Total number of chars in the attachment (generally these are base64 encoded strings).","owner":"donjayamanne","isMeasurement":true},
      "${include}": [
        "${F1}"
 

src/kernels/telemetry/notebookTelemetry.ts

@@ -1,13 +1,15 @@
 // Copyright (c) Microsoft Corporation.
 // Licensed under the MIT License.
 
-import { workspace, type NotebookDocument, type Uri } from 'vscode';
+import { NotebookCellKind, workspace, type NotebookDocument, type Uri } from 'vscode';
 import { DisposableStore } from '../../platform/common/utils/lifecycle';
 import { isUri } from '../../platform/common/utils/misc';
 import { once } from '../../platform/common/utils/functional';
 import { sendKernelTelemetryEvent } from './sendKernelTelemetryEvent';
 import { Telemetry } from '../../platform/common/constants';
 import type { Environment } from '@vscode/python-extension';
+import { getCellMetadata } from '../../platform/common/utils';
+import { traceWarning } from '../../platform/logging';
 
 /* eslint-disable @typescript-eslint/no-use-before-define */
 
@@ -245,7 +247,40 @@ type SpawnSummary = {
     envVars: number;
     interruptHandle: number;
 };
-
+type NotebookSummary = {
+    /**
+     * Number of code cells in the notebook.
+     */
+    codeCellCount: number;
+    /**
+     * Number of md cells in the notebook.
+     */
+    mdCellCount: number;
+    /**
+     * Total char length of all text in all code cells.
+     */
+    codeCellCharLength: number;
+    /**
+     * Total char length of all text in all md cells.
+     */
+    mdCellCharLength: number;
+    /**
+     * Total number of outputs in all cells.
+     */
+    outputCount: number;
+    /**
+     * Total bytes of all outputs in all cells.
+     */
+    outputsByteSize: number;
+    /**
+     * Total number of attachments
+     */
+    attachmentCount: number;
+    /**
+     * Total number of chars in the attachment (generally these are base64 encoded strings).
+     */
+    attachmentCharLength: number;
+};
 const sendTelemetry = once(function (
     notebook: NotebookDocument,
     info: {
@@ -436,15 +471,66 @@ const sendTelemetry = once(function (
         const duration = measures.spawnCompletedAfter - measures.spawnStartedAfter;
         computeSummary(spawnSummary, spawnSummaryParts, duration, measures.openedAfter);
     }
+    const notebookSummary = computeNotebookSummary(notebook);
     const allMeasures: BriefSummary & StartupSummary & PostKernelStartupSummary & SpawnSummary = {
         ...briefSummary,
         ...startupSummary,
         ...postKernelStartSummary,
-        ...spawnSummary
+        ...spawnSummary,
+        ...notebookSummary
     };
     sendKernelTelemetryEvent(notebook.uri, Telemetry.NotebookFirstStartBreakDown, allMeasures, info);
 });
 
+function computeNotebookSummary(notebook: NotebookDocument) {
+    const notebookSummary: NotebookSummary = {
+        attachmentCharLength: 0,
+        attachmentCount: 0,
+        codeCellCharLength: 0,
+        codeCellCount: 0,
+        mdCellCharLength: 0,
+        mdCellCount: 0,
+        outputCount: 0,
+        outputsByteSize: 0
+    };
+
+    notebook.getCells().forEach((cell) => {
+        const lastChar = cell.document.lineAt(cell.document.lineCount - 1).range.end;
+        const length = cell.document.offsetAt(lastChar);
+        if (cell.kind === NotebookCellKind.Markup) {
+            notebookSummary.mdCellCount += 1;
+            notebookSummary.mdCellCharLength += length;
+            try {
+                const metadata = getCellMetadata(cell);
+                const attachments = (metadata.attachments || {}) as unknown as Record<string, string>;
+                Object.keys(attachments).forEach((key) => {
+                    notebookSummary.attachmentCount += 1;
+                    const attachment = attachments[key] as unknown as Record<string, string>;
+                    if (typeof attachment === 'object') {
+                        Object.keys(attachment).forEach((mime) => {
+                            const value = attachment[mime];
+                            if (value && typeof value === 'string') {
+                                notebookSummary.attachmentCharLength += value.length;
+                            }
+                        });
+                    }
+                });
+            } catch (error) {
+                traceWarning(`Error parsing attachments in cell metadata`, error);
+            }
+        } else {
+            notebookSummary.codeCellCount += 1;
+            notebookSummary.codeCellCharLength += length;
+            notebookSummary.outputCount += cell.outputs.length;
+            notebookSummary.outputsByteSize += cell.outputs.reduce(
+                (acc, output) => acc + output.items.reduce((itemTotal, item) => itemTotal + item.data.byteLength, 0),
+                0
+            );
+        }
+    });
+
+    return notebookSummary;
+}
 const sendTelemetryForFirstAutoSelectedKernel = once(function (
     notebook: NotebookDocument,
     info: {

src/platform/common/utils.ts

@@ -457,14 +457,14 @@ type JupyterCellMetadata = Pick<nbformat.IRawCell, 'id' | 'metadata' | 'attachme
 
 export function getCellMetadata(cell: NotebookCell): JupyterCellMetadata {
     if (useCustomMetadata()) {
-        const metadata: JupyterCellMetadata = cell.metadata.custom || {};
+        const metadata: JupyterCellMetadata = JSON.parse(JSON.stringify(cell.metadata.custom || {})) || {};
         const cellMetadata = metadata as nbformat.IRawCell;
         // metadata property is never optional.
         cellMetadata.metadata = cellMetadata.metadata || {};
 
         return metadata;
     } else {
-        const metadata: JupyterCellMetadata = cell.metadata.metadata || {};
+        const metadata: JupyterCellMetadata = JSON.parse(JSON.stringify(cell.metadata || {})) || { metadata: {} };
         // metadata property is never optional.
         metadata.metadata = metadata.metadata || {};
         return metadata;

src/telemetry.ts

@@ -2900,6 +2900,38 @@ export class IEventNamePropertyMapping {
              * % Time spent in updating the kernel connection.
              */
             updateConnection: number;
+            /**
+             * Number of code cells in the notebook.
+             */
+            codeCellCount: number;
+            /**
+             * Number of md cells in the notebook.
+             */
+            mdCellCount: number;
+            /**
+             * Total char length of all text in all code cells.
+             */
+            codeCellCharLength: number;
+            /**
+             * Total char length of all text in all md cells.
+             */
+            mdCellCharLength: number;
+            /**
+             * Total number of outputs in all cells.
+             */
+            outputCount: number;
+            /**
+             * Total bytes of all outputs in all cells.
+             */
+            outputsByteSize: number;
+            /**
+             * Total number of attachments
+             */
+            attachmentCount: number;
+            /**
+             * Total number of chars in the attachment (generally these are base64 encoded strings).
+             */
+            attachmentCharLength: number;
         } & ResourceSpecificTelemetryProperties
     > = {
         owner: 'donjayamanne',
@@ -3008,6 +3040,46 @@ export class IEventNamePropertyMapping {
                 classification: 'SystemMetaData',
                 purpose: 'FeatureInsight',
                 isMeasurement: true
+            },
+            codeCellCount: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            mdCellCount: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            codeCellCharLength: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            mdCellCharLength: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            outputCount: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            outputsByteSize: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            attachmentCount: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
+            },
+            attachmentCharLength: {
+                classification: 'SystemMetaData',
+                purpose: 'FeatureInsight',
+                isMeasurement: true
             }
         }
     };