The Postgres development platform. Supabase gives you a dedicated Postgres database to build your web, mobile, and AI applications.
TypeScript aialternativeauthdatabasedenoembeddingsexamplefirebasenextjsoauth2pgvectorpostgispostgrespostgresqlpostgrestrealtimesupabasevectorswebsockets
import { at } from 'lodash-es'
/**
* Resolves a dot/bracket path within a shared-data dataset. If the resolved
* value is an object with `value`/`unit` fields, returns `${value} ${unit}`
* (trimmed); otherwise returns the resolved primitive as-is.
*
* Pure: no `shared-data` import. Callers supply the dataset so this util can
* be reused by the React `<SharedData>` component (Next.js bundle) and by the
* build-time markdown-schema handler (tsx) without each having to navigate
* `shared-data`'s ESM/CJS interop independently.
*/
export function resolveSharedDataPath(dataset: unknown, path: string): string | number | undefined {
const selected = at(dataset as any, [path])[0]
if (typeof selected === 'object' && selected !== null) {
return `${(selected as any).value ?? ''} ${(selected as any).unit ?? ''}`.trim()
}
return selected
}
type LogSourceSchema = {
name: string
reference: string
fields: { path: string; type: string }[]
}
const LOG_COLUMNS = new Map([
['id', 'String'],
['timestamp', 'DateTime64'],
['event_message', 'String'],
['severity_text', 'String'],
['source', 'String'],
])
/** One field mapping for the HTML reference and its Markdown export. */
export function getLogFieldReference(schemas: LogSourceSchema[]) {
return schemas.map((schema) => {
const fields = [...schema.fields]
for (const [path, type] of LOG_COLUMNS) {
if (!fields.some((field) => field.path === path)) fields.push({ path, type })
}
return {
...schema,
fields: fields
.sort((a, b) => a.path.localeCompare(b.path))
.map((field) => {
const key = field.path
.replace(/^metadata\./, '')
.replace(/\\/g, '\\\\')
.replace(/'/g, "''")
return {
...field,
queryField: LOG_COLUMNS.has(field.path) ? field.path : `log_attributes['${key}']`,
queryType: LOG_COLUMNS.get(field.path) ?? 'String',
}
}),
}
})
}