Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.
Python agent-contextagent-memoryagent-pluginsagentic-ragcontext-databasedsh-pluginknowledge-basememorymemory-managementself-evolvingvector-database
export function fileNameFromUri(uri: string): string {
const trimmed = uri.endsWith('/') ? uri.slice(0, -1) : uri
const index = trimmed.lastIndexOf('/')
if (index < 0) return trimmed
return trimmed.slice(index + 1) || trimmed
}
const DIRECTORY_SEMANTIC_SIDECARS = new Set(['.abstract.md', '.overview.md'])
export function isDirectorySemanticSidecarUri(uri: string): boolean {
return DIRECTORY_SEMANTIC_SIDECARS.has(fileNameFromUri(uri))
}
/**
* L0/L1 hits all share one of two hidden filenames, so a bare basename loses
* the identity of the directory they summarize. Keep ordinary result names
* compact while adding exactly one parent segment for semantic sidecars.
*/
export function retrievalResultNameFromUri(uri: string): string {
const name = fileNameFromUri(uri)
if (!DIRECTORY_SEMANTIC_SIDECARS.has(name)) return name
const parentName = fileNameFromUri(parentUri(uri))
return parentName ? `${parentName}/${name}` : name
}
export function normalizeDirUri(uri: string): string {
const value = uri.trim()
if (!value) {
return 'viking://'
}
if (value === 'viking://') {
return value
}
return value.endsWith('/') ? value : `${value}/`
}
export function normalizeFileUri(uri: string): string {
const value = uri.trim()
if (!value) {
return 'viking://'
}
if (value === 'viking://') {
return value
}
return value.endsWith('/') ? value.slice(0, -1) : value
}
export function parentUri(uri: string): string {
const normalized = normalizeDirUri(uri)
if (normalized === 'viking://') {
return normalized
}
const body = normalized.slice('viking://'.length, -1)
if (!body.includes('/')) {
return 'viking://'
}
return `viking://${body.slice(0, body.lastIndexOf('/') + 1)}`
}
export function joinUri(baseUri: string, child: string): string {
const raw = child.trim()
if (!raw) {
return normalizeDirUri(baseUri)
}
if (raw.startsWith('viking://')) {
return raw
}
const normalizedBase = normalizeDirUri(baseUri)
return `${normalizedBase}${raw.replace(/^\//, '')}`
}
/**
* Matches `viking://` URIs inside arbitrary text. Stops at whitespace, quotes,
* brackets and the punctuation that commonly trails a URI in prose. Shared so
* the sessions view and the filesystem extract URIs identically.
*/
export const VIKING_URI_RE = /viking:\/\/[^\s,,。;;'"`<>()\]}\\]+/g
/**
* Extract the first `viking://` URI from a string (or accept an already-clean
* URI) and strip trailing punctuation that bleeds in from surrounding prose.
*/
export function cleanVikingUri(value: string): string {
const trimmed = value.trim()
if (trimmed.toLowerCase().startsWith('viking://')) {
return trimmed.replace(/[\\,。;;,.]+$/u, '')
}
const match = value.match(VIKING_URI_RE)
return (match?.[0] ?? trimmed).trim().replace(/[\\,。;;,.]+$/u, '')
}