add cached fuzzy file search and debounce unified picker

This commit is contained in:
Shantur Rathore
2025-11-19 16:43:28 +00:00
parent 7e95005d8c
commit 629d098add
12 changed files with 635 additions and 271 deletions

View File

@@ -20,6 +20,7 @@
"@fastify/static": "^7.0.4",
"commander": "^12.1.0",
"fastify": "^4.28.1",
"fuzzysort": "^2.0.4",
"pino": "^9.4.0",
"undici": "^6.19.8",
"zod": "^3.23.8"

View File

@@ -103,6 +103,8 @@ export interface WorkspaceFileResponse {
contents: string
}
export type WorkspaceFileSearchResponse = FileSystemEntry[]
export interface InstanceData {
messageHistory: string[]
}
@@ -112,6 +114,7 @@ export interface BinaryRecord {
path: string
label: string
version?: string
/** Indicates that this binary will be picked when workspaces omit an explicit choice. */
isDefault: boolean
lastValidatedAt?: string

View File

@@ -0,0 +1,61 @@
import assert from "node:assert/strict"
import { beforeEach, describe, it } from "node:test"
import type { FileSystemEntry } from "../../api-types"
import {
clearWorkspaceSearchCache,
getWorkspaceCandidates,
refreshWorkspaceCandidates,
WORKSPACE_CANDIDATE_CACHE_TTL_MS,
} from "../search-cache"
describe("workspace search cache", () => {
beforeEach(() => {
clearWorkspaceSearchCache()
})
it("expires cached candidates after the TTL", () => {
const workspacePath = "/tmp/workspace"
const startTime = 1_000
refreshWorkspaceCandidates(workspacePath, () => [createEntry("file-a")], startTime)
const beforeExpiry = getWorkspaceCandidates(
workspacePath,
startTime + WORKSPACE_CANDIDATE_CACHE_TTL_MS - 1,
)
assert.ok(beforeExpiry)
assert.equal(beforeExpiry.length, 1)
assert.equal(beforeExpiry[0].name, "file-a")
const afterExpiry = getWorkspaceCandidates(
workspacePath,
startTime + WORKSPACE_CANDIDATE_CACHE_TTL_MS + 1,
)
assert.equal(afterExpiry, undefined)
})
it("replaces cached entries when manually refreshed", () => {
const workspacePath = "/tmp/workspace"
refreshWorkspaceCandidates(workspacePath, () => [createEntry("file-a")], 5_000)
const initial = getWorkspaceCandidates(workspacePath)
assert.ok(initial)
assert.equal(initial[0].name, "file-a")
refreshWorkspaceCandidates(workspacePath, () => [createEntry("file-b")], 6_000)
const refreshed = getWorkspaceCandidates(workspacePath)
assert.ok(refreshed)
assert.equal(refreshed[0].name, "file-b")
})
})
function createEntry(name: string): FileSystemEntry {
return {
name,
path: name,
absolutePath: `/tmp/${name}`,
type: "file",
size: 1,
modifiedAt: new Date().toISOString(),
}
}

View File

@@ -0,0 +1,66 @@
import path from "path"
import type { FileSystemEntry } from "../api-types"
export const WORKSPACE_CANDIDATE_CACHE_TTL_MS = 30_000
interface WorkspaceCandidateCacheEntry {
expiresAt: number
candidates: FileSystemEntry[]
}
const workspaceCandidateCache = new Map<string, WorkspaceCandidateCacheEntry>()
export function getWorkspaceCandidates(rootDir: string, now = Date.now()): FileSystemEntry[] | undefined {
const key = normalizeKey(rootDir)
const cached = workspaceCandidateCache.get(key)
if (!cached) {
return undefined
}
if (cached.expiresAt <= now) {
workspaceCandidateCache.delete(key)
return undefined
}
return cloneEntries(cached.candidates)
}
export function refreshWorkspaceCandidates(
rootDir: string,
builder: () => FileSystemEntry[],
now = Date.now(),
): FileSystemEntry[] {
const key = normalizeKey(rootDir)
const freshCandidates = builder()
if (!freshCandidates || freshCandidates.length === 0) {
workspaceCandidateCache.delete(key)
return []
}
const storedCandidates = cloneEntries(freshCandidates)
workspaceCandidateCache.set(key, {
expiresAt: now + WORKSPACE_CANDIDATE_CACHE_TTL_MS,
candidates: storedCandidates,
})
return cloneEntries(storedCandidates)
}
export function clearWorkspaceSearchCache(rootDir?: string) {
if (typeof rootDir === "undefined") {
workspaceCandidateCache.clear()
return
}
const key = normalizeKey(rootDir)
workspaceCandidateCache.delete(key)
}
function cloneEntries(entries: FileSystemEntry[]): FileSystemEntry[] {
return entries.map((entry) => ({ ...entry }))
}
function normalizeKey(rootDir: string) {
return path.resolve(rootDir)
}

View File

@@ -0,0 +1,184 @@
import fs from "fs"
import path from "path"
import fuzzysort from "fuzzysort"
import type { FileSystemEntry } from "../api-types"
import { clearWorkspaceSearchCache, getWorkspaceCandidates, refreshWorkspaceCandidates } from "./search-cache"
const DEFAULT_LIMIT = 100
const MAX_LIMIT = 200
const MAX_CANDIDATES = 8000
const IGNORED_DIRECTORIES = new Set(
[".git", ".hg", ".svn", "node_modules", "dist", "build", ".next", ".nuxt", ".turbo", ".cache", "coverage"].map(
(name) => name.toLowerCase(),
),
)
export type WorkspaceFileSearchType = "all" | "file" | "directory"
export interface WorkspaceFileSearchOptions {
limit?: number
type?: WorkspaceFileSearchType
refresh?: boolean
}
interface CandidateEntry {
entry: FileSystemEntry
key: string
}
export function searchWorkspaceFiles(
rootDir: string,
query: string,
options: WorkspaceFileSearchOptions = {},
): FileSystemEntry[] {
const trimmedQuery = query.trim()
if (!trimmedQuery) {
throw new Error("Search query is required")
}
const normalizedRoot = path.resolve(rootDir)
const limit = normalizeLimit(options.limit)
const typeFilter: WorkspaceFileSearchType = options.type ?? "all"
const refreshRequested = options.refresh === true
let entries: FileSystemEntry[] | undefined
try {
if (!refreshRequested) {
entries = getWorkspaceCandidates(normalizedRoot)
}
if (!entries) {
entries = refreshWorkspaceCandidates(normalizedRoot, () => collectCandidates(normalizedRoot))
}
} catch (error) {
clearWorkspaceSearchCache(normalizedRoot)
throw error
}
if (!entries || entries.length === 0) {
clearWorkspaceSearchCache(normalizedRoot)
return []
}
const candidates = buildCandidateEntries(entries, typeFilter)
if (candidates.length === 0) {
return []
}
const matches = fuzzysort.go<CandidateEntry>(trimmedQuery, candidates, {
key: "key",
limit,
})
if (!matches || matches.length === 0) {
return []
}
return matches.map((match) => match.obj.entry)
}
function collectCandidates(rootDir: string): FileSystemEntry[] {
const queue: string[] = [""]
const entries: FileSystemEntry[] = []
while (queue.length > 0 && entries.length < MAX_CANDIDATES) {
const relativeDir = queue.pop() || ""
const absoluteDir = relativeDir ? path.join(rootDir, relativeDir) : rootDir
let dirents: fs.Dirent[]
try {
dirents = fs.readdirSync(absoluteDir, { withFileTypes: true })
} catch {
continue
}
for (const dirent of dirents) {
const entryName = dirent.name
const lowerName = entryName.toLowerCase()
const relativePath = relativeDir ? `${relativeDir}/${entryName}` : entryName
const absolutePath = path.join(absoluteDir, entryName)
if (dirent.isDirectory() && IGNORED_DIRECTORIES.has(lowerName)) {
continue
}
let stats: fs.Stats
try {
stats = fs.statSync(absolutePath)
} catch {
continue
}
const isDirectory = stats.isDirectory()
if (isDirectory && !IGNORED_DIRECTORIES.has(lowerName)) {
if (entries.length < MAX_CANDIDATES) {
queue.push(relativePath)
}
}
const entryType: FileSystemEntry["type"] = isDirectory ? "directory" : "file"
const normalizedPath = normalizeRelativeEntryPath(relativePath)
const entry: FileSystemEntry = {
name: entryName,
path: normalizedPath,
absolutePath: path.resolve(rootDir, normalizedPath === "." ? "" : normalizedPath),
type: entryType,
size: entryType === "file" ? stats.size : undefined,
modifiedAt: stats.mtime.toISOString(),
}
entries.push(entry)
if (entries.length >= MAX_CANDIDATES) {
break
}
}
}
return entries
}
function buildCandidateEntries(entries: FileSystemEntry[], filter: WorkspaceFileSearchType): CandidateEntry[] {
const filtered: CandidateEntry[] = []
for (const entry of entries) {
if (!shouldInclude(entry.type, filter)) {
continue
}
filtered.push({ entry, key: buildSearchKey(entry) })
}
return filtered
}
function normalizeLimit(limit?: number) {
if (!limit || Number.isNaN(limit)) {
return DEFAULT_LIMIT
}
const clamped = Math.min(Math.max(limit, 1), MAX_LIMIT)
return clamped
}
function shouldInclude(entryType: FileSystemEntry["type"], filter: WorkspaceFileSearchType) {
return filter === "all" || entryType === filter
}
function normalizeRelativeEntryPath(relativePath: string): string {
if (!relativePath) {
return "."
}
let normalized = relativePath.replace(/\\+/g, "/")
if (normalized.startsWith("./")) {
normalized = normalized.replace(/^\.\/+/, "")
}
if (normalized.startsWith("/")) {
normalized = normalized.replace(/^\/+/g, "")
}
return normalized || "."
}
function buildSearchKey(entry: FileSystemEntry) {
return entry.path.toLowerCase()
}

View File

@@ -19,6 +19,16 @@ const WorkspaceFileContentQuerySchema = z.object({
path: z.string(),
})
const WorkspaceFileSearchQuerySchema = z.object({
q: z.string().trim().min(1, "Query is required"),
limit: z.coerce.number().int().positive().max(200).optional(),
type: z.enum(["all", "file", "directory"]).optional(),
refresh: z
.string()
.optional()
.transform((value) => (value === undefined ? undefined : value === "true")),
})
export function registerWorkspaceRoutes(app: FastifyInstance, deps: RouteDeps) {
app.get("/api/workspaces", async () => {
return deps.workspaceManager.list()
@@ -57,6 +67,22 @@ export function registerWorkspaceRoutes(app: FastifyInstance, deps: RouteDeps) {
}
})
app.get<{
Params: { id: string }
Querystring: { q?: string; limit?: string; type?: "all" | "file" | "directory"; refresh?: string }
}>("/api/workspaces/:id/files/search", async (request, reply) => {
try {
const query = WorkspaceFileSearchQuerySchema.parse(request.query ?? {})
return deps.workspaceManager.searchFiles(request.params.id, query.q, {
limit: query.limit,
type: query.type,
refresh: query.refresh,
})
} catch (error) {
return handleWorkspaceError(error, reply)
}
})
app.get<{
Params: { id: string }
Querystring: { path?: string }
@@ -70,6 +96,7 @@ export function registerWorkspaceRoutes(app: FastifyInstance, deps: RouteDeps) {
})
}
function handleWorkspaceError(error: unknown, reply: FastifyReply) {
if (error instanceof Error && error.message === "Workspace not found") {
reply.code(404)

View File

@@ -3,6 +3,8 @@ import { EventBus } from "../events/bus"
import { ConfigStore } from "../config/store"
import { BinaryRegistry } from "../config/binaries"
import { FileSystemBrowser } from "../filesystem/browser"
import { searchWorkspaceFiles, WorkspaceFileSearchOptions } from "../filesystem/search"
import { clearWorkspaceSearchCache } from "../filesystem/search-cache"
import { WorkspaceDescriptor, WorkspaceFileResponse, FileSystemEntry } from "../api-types"
import { WorkspaceRuntime } from "./runtime"
import { Logger } from "../logger"
@@ -43,6 +45,11 @@ export class WorkspaceManager {
return browser.list(relativePath)
}
searchFiles(workspaceId: string, query: string, options?: WorkspaceFileSearchOptions): FileSystemEntry[] {
const workspace = this.requireWorkspace(workspaceId)
return searchWorkspaceFiles(workspace.path, query, options)
}
readFile(workspaceId: string, relativePath: string): WorkspaceFileResponse {
const workspace = this.requireWorkspace(workspaceId)
const browser = new FileSystemBrowser({ rootDir: workspace.path })
@@ -55,14 +62,17 @@ export class WorkspaceManager {
}
async create(folder: string, name?: string): Promise<WorkspaceDescriptor> {
const id = `${Date.now().toString(36)}`
const binary = this.options.binaryRegistry.resolveDefault()
const workspacePath = path.isAbsolute(folder) ? folder : path.resolve(this.options.rootDir, folder)
clearWorkspaceSearchCache(workspacePath)
this.options.logger.info({ workspaceId: id, folder: workspacePath, binary: binary.path }, "Creating workspace")
const proxyPath = `/workspaces/${id}/instance`
const descriptor: WorkspaceRecord = {
id,
path: workspacePath,
@@ -120,6 +130,7 @@ export class WorkspaceManager {
}
this.workspaces.delete(id)
clearWorkspaceSearchCache(workspace.path)
if (!wasRunning) {
this.options.eventBus.publish({ type: "workspace.stopped", workspaceId: id })
}