:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import type { MarkdownNotebookAskAIRequest } from 'lib/components/MarkdownNotebook'
import { SCRATCHPAD_NOTEBOOK } from '~/models/notebooksModel'
import { AccessControlLevel, UserType } from '~/types'
import type { NotebookRunStatusResponseApi } from 'products/notebooks/frontend/generated/api.schemas'
import { NotebookType } from '../types'
import type { InlineAICompletion } from './MarkdownNotebookInlineAI'
export type NotebookInlineAIRequestedProperties = {
conversation_id: string
notebook_short_id: string | null
source: MarkdownNotebookAskAIRequest['source']
prompt_length: number
has_selection: boolean
}
export type NotebookInlineAIFinishedProperties = NotebookInlineAIRequestedProperties & {
status: InlineAICompletion['status']
result_kind: InlineAICompletion['kind']
has_artifact: boolean
duration_ms: number | null
}
/**
* Builds the `notebook inline ai requested` event payload. It carries lengths and flags only, because
* the prompt and the selected markdown can contain customer data.
*/
export function buildNotebookInlineAIRequestedEvent(
request: MarkdownNotebookAskAIRequest,
notebookShortId: string | null
): NotebookInlineAIRequestedProperties {
return {
conversation_id: request.conversationId,
notebook_short_id: notebookShortId,
source: request.source,
prompt_length: request.query.length,
has_selection: !!request.selectedMarkdown,
}
}
export function buildNotebookInlineAIFinishedEvent(
request: MarkdownNotebookAskAIRequest,
completion: InlineAICompletion,
notebookShortId: string | null,
startedAtMs: number | undefined,
finishedAtMs: number
): NotebookInlineAIFinishedProperties {
return {
...buildNotebookInlineAIRequestedEvent(request, notebookShortId),
status: completion.status,
result_kind: completion.kind,
has_artifact: completion.hasArtifact,
duration_ms: startedAtMs === undefined ? null : Math.max(0, Math.round(finishedAtMs - startedAtMs)),
}
}
export type NotebookOpenedProperties = {
short_id: string
is_creator: boolean
user_access_level?: AccessControlLevel
access_source: 'direct' | 'shared_link'
node_count: number
}
/**
* Builds the `notebook opened` event payload, or `null` when the loaded notebook is not a real
* persisted notebook (scratchpad / template) and so should not count as a human open.
*/
export function buildNotebookOpenedEvent(
notebook: NotebookType | null,
user: UserType | null,
isShared: boolean
): NotebookOpenedProperties | null {
const shortId = notebook?.short_id
if (!notebook || !shortId || shortId === SCRATCHPAD_NOTEBOOK.short_id || shortId.startsWith('template-')) {
return null
}
return {
short_id: shortId,
is_creator: !!user && notebook.created_by?.uuid === user.uuid,
user_access_level: notebook.user_access_level,
access_source: isShared ? 'shared_link' : 'direct',
node_count: notebook.content?.content?.length ?? 0,
}
}
export type NotebookRunAllStartedProperties = {
short_id: string
cell_count: number
}
export type NotebookRunAllFinishedProperties = {
short_id: string
cell_count: number
python_cell_count: number
completed_count: number
outcome: string
duration_ms: number
}
/** No code and no cell content in the properties: a run's payload is counts and an outcome. */
export function buildNotebookRunStartedEvent(
shortId: string,
cellCount: number
): ['notebook run all started', NotebookRunAllStartedProperties] {
return ['notebook run all started', { short_id: shortId, cell_count: cellCount }]
}
export function buildNotebookRunFinishedEvent(
shortId: string,
run: NotebookRunStatusResponseApi,
durationMs: number
): ['notebook run all finished', NotebookRunAllFinishedProperties] {
return [
'notebook run all finished',
{
short_id: shortId,
cell_count: run.cell_count,
python_cell_count: run.cells.filter((cell) => cell.cell_type === 'python').length,
completed_count: run.cells.filter((cell) => cell.status === 'done').length,
outcome: run.status,
duration_ms: durationMs,
},
]
}