g1t/services/context/src/index.ts
| 1 | /** |
| 2 | * The context service: the context hub's catalog, search and scorecards. |
| 3 | * |
| 4 | * - **Catalog.** What a workspace builds and runs, as entities (projects, |
| 5 | * apps, APIs, packages, languages, owners, environments, integrations, |
| 6 | * docs) and relations between them. Built by itself: on every push to a |
| 7 | * project's default branch it reads the project's manifests and docs |
| 8 | * whose blobs changed (at most `MAX_READS` files a push) and joins them |
| 9 | * with what projects, deployments, identity and integrations know. |
| 10 | * Rebuilding is idempotent: entity ids are hashes of what they are. |
| 11 | * - **Search.** Docs, catalog entities, issues, pull requests and kept |
| 12 | * memory, embedded with Workers AI into a Vectorize index whose rows |
| 13 | * carry their workspace, project and whether they are private, so a |
| 14 | * query reads only what its reader may. Matching words in D1 answers |
| 15 | * when the index cannot, and fills in after it. |
| 16 | * - **Backfill.** Once per workspace (and again on request): the catalog |
| 17 | * for every project, memory candidates from docs, manifests and the last |
| 18 | * merged pull requests, and the search index filled. One queued job per |
| 19 | * project, capped and metered. |
| 20 | * - **Run context.** The Context section every g1t agent run starts with. |
| 21 | * |
| 22 | * Reached through service bindings: `POST /rpc/<method>`. |
| 23 | */ |
| 24 | |
| 25 | import { |
| 26 | type Backfill, |
| 27 | type CaptureItem, |
| 28 | type Catalog, |
| 29 | type ContextStatus, |
| 30 | type Entity, |
| 31 | type EntityDetail, |
| 32 | type EntityKind, |
| 33 | ENTITY_KINDS, |
| 34 | type G1tEvent, |
| 35 | type Memory, |
| 36 | type Project, |
| 37 | type ProjectDeploys, |
| 38 | type RelationKind, |
| 39 | type RunContext, |
| 40 | type Scorecard, |
| 41 | type SearchHit, |
| 42 | type SearchKind, |
| 43 | type SearchResult, |
| 44 | type ServiceBinding, |
| 45 | type User, |
| 46 | type Viewer, |
| 47 | currentWorkspaceSlug, |
| 48 | deploymentsClient, |
| 49 | fail, |
| 50 | identityClient, |
| 51 | integrationsClient, |
| 52 | memoryReviewClient, |
| 53 | ok, |
| 54 | projectsClient, |
| 55 | reposClient, |
| 56 | securityClient, |
| 57 | staleSlugs, |
| 58 | workClient, |
| 59 | type Result, |
| 60 | } from "@g1t/contracts"; |
| 61 | |
| 62 | import { assemble, authorsOf, integrationEntities, type EntityDraft, type FileRecord, type ProjectInput, type Surroundings } from "./assemble"; |
| 63 | import { extract, interesting, type FileFacts } from "./extract"; |
| 64 | import { composeRunContext, type ContextNote, type ProjectContext } from "./runcontext"; |
| 65 | import { evaluate } from "./scorecards"; |
| 66 | import { allowedKinds, indexFilter, merge, readable, type IndexMeta, type Reader } from "./visibility"; |
| 67 | |
| 68 | type Job = |
| 69 | | { type: "backfill_project"; workspace: string; slug: string } |
| 70 | | { type: "backfill_memory"; workspace: string }; |
| 71 | |
| 72 | type Env = { |
| 73 | DB: D1Database; |
| 74 | AI?: Ai; |
| 75 | VECTORS?: Vectorize; |
| 76 | JOBS: Queue<Job>; |
| 77 | REPOS: ServiceBinding; |
| 78 | PROJECTS: ServiceBinding; |
| 79 | WORK: ServiceBinding; |
| 80 | IDENTITY: ServiceBinding; |
| 81 | DEPLOYMENTS: ServiceBinding; |
| 82 | INTEGRATIONS: ServiceBinding; |
| 83 | SECURITY?: ServiceBinding; |
| 84 | }; |
| 85 | |
| 86 | /** Workers AI's embedding model: 768 dimensions, as the index was made with. */ |
| 87 | const EMBED_MODEL = "@cf/baai/bge-base-en-v1.5"; |
| 88 | /** What it costs, in millionths of a dollar per token ($0.067 per million). */ |
| 89 | const MICROS_PER_TOKEN = 0.067; |
| 90 | /** Past this many tokens in a month, a workspace's new text goes unindexed; text search still finds it. */ |
| 91 | const MONTHLY_TOKENS = 20_000_000; |
| 92 | /** The most text embedded for one row; the model reads 512 tokens. */ |
| 93 | const EMBED_CHARS = 2000; |
| 94 | const EMBED_BATCH = 50; |
| 95 | /** The most files read from a repository for one project's rebuild. */ |
| 96 | const MAX_READS = 15; |
| 97 | /** Of them, workflows. */ |
| 98 | const MAX_WORKFLOW_READS = 3; |
| 99 | /** The most projects one push rebuilds. */ |
| 100 | const MAX_PROJECTS_PER_PUSH = 5; |
| 101 | /** A backfill's caps. */ |
| 102 | const BACKFILL_PROJECTS = 50; |
| 103 | const BACKFILL_PULLS = 20; |
| 104 | const BACKFILL_ITEMS = 30; |
| 105 | /** A backfill still marked running after this long is taken to have died. */ |
| 106 | const BACKFILL_STALE_MS = 30 * 60 * 1000; |
| 107 | const ITEM_CHARS = 4000; |
| 108 | const SNIPPET_CHARS = 280; |
| 109 | /** Kinds every project shares rather than owns. */ |
| 110 | const SHARED: Set<EntityKind> = new Set(["owner", "language", "integration"]); |
| 111 | |
| 112 | const now = () => new Date().toISOString(); |
| 113 | const month = () => now().slice(0, 7); |
| 114 | |
| 115 | function isMember(viewer: Viewer, workspace: string): boolean { |
| 116 | return !!viewer?.workspaces?.some((m) => m.slug === workspace.toLowerCase()); |
| 117 | } |
| 118 | |
| 119 | async function sha(text: string): Promise<string> { |
| 120 | const digest = await crypto.subtle.digest("SHA-256", new TextEncoder().encode(text)); |
| 121 | return [...new Uint8Array(digest)].map((b) => b.toString(16).padStart(2, "0")).join(""); |
| 122 | } |
| 123 | |
| 124 | /** An entity's id: the same for the same thing, however often it is rebuilt. */ |
| 125 | export async function entityId(workspace: string, kind: string, key: string): Promise<string> { |
| 126 | return `ent_${(await sha(`${workspace}\u0000${kind}\u0000${key.toLowerCase()}`)).slice(0, 24)}`; |
| 127 | } |
| 128 | |
| 129 | function snippet(text: string | null | undefined): string { |
| 130 | const line = (text ?? "").replace(/\s+/g, " ").trim(); |
| 131 | return line.length <= SNIPPET_CHARS ? line : `${line.slice(0, SNIPPET_CHARS - 1)}…`; |
| 132 | } |
| 133 | |
| 134 | type EntityRow = { |
| 135 | id: string; |
| 136 | workspace: string; |
| 137 | kind: EntityKind; |
| 138 | key: string; |
| 139 | name: string; |
| 140 | summary: string | null; |
| 141 | project_id: string | null; |
| 142 | project: string | null; |
| 143 | repo_id: string | null; |
| 144 | private: number; |
| 145 | data: string; |
| 146 | source: string; |
| 147 | ref: string | null; |
| 148 | updated_at: string; |
| 149 | }; |
| 150 | |
| 151 | type ItemRow = { |
| 152 | id: string; |
| 153 | workspace: string; |
| 154 | kind: "doc" | "issue" | "pull"; |
| 155 | entity_id: string | null; |
| 156 | project: string | null; |
| 157 | private: number; |
| 158 | title: string; |
| 159 | text: string; |
| 160 | url: string | null; |
| 161 | by: string | null; |
| 162 | updated_at: string; |
| 163 | }; |
| 164 | |
| 165 | function toEntity(row: EntityRow): Entity { |
| 166 | let data: Record<string, unknown> = {}; |
| 167 | try { |
| 168 | data = JSON.parse(row.data); |
| 169 | } catch {} |
| 170 | return { |
| 171 | id: row.id, |
| 172 | workspace: row.workspace, |
| 173 | kind: row.kind, |
| 174 | key: row.key, |
| 175 | name: row.name, |
| 176 | summary: row.summary, |
| 177 | project: row.project, |
| 178 | private: !!row.private, |
| 179 | data, |
| 180 | source: row.source, |
| 181 | ref: row.ref, |
| 182 | updatedAt: row.updated_at, |
| 183 | }; |
| 184 | } |
| 185 | |
| 186 | /** Where a memory came from, in a few words. */ |
| 187 | function memorySource(memory: Memory): string { |
| 188 | const ref = memory.source.reference ?? ""; |
| 189 | const number = memory.source.number; |
| 190 | switch (memory.source.kind) { |
| 191 | case "doc": |
| 192 | return ref.split(":").slice(2).join(":") || "a doc"; |
| 193 | case "review": |
| 194 | return number ? `review on #${number}` : "a review"; |
| 195 | case "pr": |
| 196 | return number ? `#${number}` : "a merged pull request"; |
| 197 | case "run": |
| 198 | return number ? `agent run on #${number}` : "an agent run"; |
| 199 | default: |
| 200 | return `added by ${memory.createdBy}`; |
| 201 | } |
| 202 | } |
| 203 | |
| 204 | /** What one workspace's rebuilds share: asked of other services once. */ |
| 205 | class WorkspaceCache { |
| 206 | private members?: Promise<string[]>; |
| 207 | private deploys?: Promise<ProjectDeploys[]>; |
| 208 | private connections?: Promise<Surroundings["integrations"]>; |
| 209 | constructor( |
| 210 | private readonly env: Env, |
| 211 | readonly workspace: string, |
| 212 | readonly actor: User, |
| 213 | ) {} |
| 214 | |
| 215 | memberNames(): Promise<string[]> { |
| 216 | this.members ??= identityClient(this.env.IDENTITY) |
| 217 | .listMembers(this.workspace, this.actor) |
| 218 | .then((found) => (found.ok ? found.value.map((member) => member.username) : [])) |
| 219 | .catch(() => []); |
| 220 | return this.members; |
| 221 | } |
| 222 | |
| 223 | deployments(): Promise<ProjectDeploys[]> { |
| 224 | this.deploys ??= deploymentsClient(this.env.DEPLOYMENTS) |
| 225 | .overview(this.workspace, this.actor) |
| 226 | .then((found) => (found.ok ? found.value : [])) |
| 227 | .catch(() => []); |
| 228 | return this.deploys; |
| 229 | } |
| 230 | |
| 231 | integrations(): Promise<Surroundings["integrations"]> { |
| 232 | this.connections ??= integrationsClient(this.env.INTEGRATIONS) |
| 233 | .list(this.workspace, this.actor) |
| 234 | .then((found) => |
| 235 | found.ok |
| 236 | ? found.value |
| 237 | .filter((connection) => connection.kind !== "models") |
| 238 | .map((connection) => ({ |
| 239 | id: connection.id, |
| 240 | provider: connection.provider, |
| 241 | name: connection.name, |
| 242 | kind: connection.kind, |
| 243 | repo: connection.config.repo ?? null, |
| 244 | })) |
| 245 | : [], |
| 246 | ) |
| 247 | .catch(() => []); |
| 248 | return this.connections; |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | type ScanStats = { entities: number; candidates: number; kept: number; indexed: number }; |
| 253 | |
| 254 | class Context { |
| 255 | constructor(private readonly env: Env) {} |
| 256 | |
| 257 | private get db() { |
| 258 | return this.env.DB; |
| 259 | } |
| 260 | |
| 261 | private async workspaceActor(slug: string): Promise<User | null> { |
| 262 | const workspace = await identityClient(this.env.IDENTITY).getWorkspace(slug); |
| 263 | if (!workspace) return null; |
| 264 | return { |
| 265 | id: workspace.id, |
| 266 | username: workspace.slug, |
| 267 | kind: "workspace", |
| 268 | verified: true, |
| 269 | workspaces: [{ slug: workspace.slug, role: "member" }], |
| 270 | }; |
| 271 | } |
| 272 | |
| 273 | // ---- Search index -------------------------------------------------------- |
| 274 | |
| 275 | /** Embeds and stores rows in the search index, within the workspace's monthly allowance. Never throws. */ |
| 276 | private async index(workspace: string, rows: { id: string; text: string; meta: IndexMeta }[]): Promise<number> { |
| 277 | const { AI, VECTORS } = this.env; |
| 278 | if (!AI || !VECTORS || rows.length === 0) return 0; |
| 279 | try { |
| 280 | const used = await this.db |
| 281 | .prepare("SELECT tokens FROM usage WHERE workspace = ? AND month = ?") |
| 282 | .bind(workspace, month()) |
| 283 | .first<{ tokens: number }>(); |
| 284 | if ((used?.tokens ?? 0) >= MONTHLY_TOKENS) return 0; |
| 285 | let stored = 0; |
| 286 | for (let at = 0; at < rows.length; at += EMBED_BATCH) { |
| 287 | const batch = rows.slice(at, at + EMBED_BATCH); |
| 288 | const texts = batch.map((row) => row.text.slice(0, EMBED_CHARS)); |
| 289 | const embedded = (await AI.run(EMBED_MODEL, { text: texts })) as { data?: number[][] }; |
| 290 | const vectors = (embedded.data ?? []).map((values, i) => ({ |
| 291 | id: batch[i].id, |
| 292 | values, |
| 293 | metadata: batch[i].meta as unknown as Record<string, VectorizeVectorMetadata>, |
| 294 | })); |
| 295 | if (vectors.length) await VECTORS.upsert(vectors); |
| 296 | stored += vectors.length; |
| 297 | await this.meter(workspace, texts.reduce((sum, text) => sum + Math.ceil(text.length / 4), 0)); |
| 298 | } |
| 299 | return stored; |
| 300 | } catch (error) { |
| 301 | console.error("could not index", rows.length, "rows for", workspace, error); |
| 302 | return 0; |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | private async unindex(ids: string[]): Promise<void> { |
| 307 | if (!this.env.VECTORS || ids.length === 0) return; |
| 308 | try { |
| 309 | await this.env.VECTORS.deleteByIds(ids); |
| 310 | } catch (error) { |
| 311 | console.error("could not take", ids.length, "rows out of the index", error); |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | private async meter(workspace: string, tokens: number): Promise<void> { |
| 316 | await this.db |
| 317 | .prepare( |
| 318 | `INSERT INTO usage (workspace, month, tokens, cost_micros) VALUES (?1, ?2, ?3, ?4) |
| 319 | ON CONFLICT (workspace, month) DO UPDATE SET tokens = tokens + ?3, cost_micros = cost_micros + ?4`, |
| 320 | ) |
| 321 | .bind(workspace, month(), tokens, Math.ceil(tokens * MICROS_PER_TOKEN)) |
| 322 | .run(); |
| 323 | } |
| 324 | |
| 325 | private async embedQuery(workspace: string, query: string): Promise<number[] | null> { |
| 326 | if (!this.env.AI || !this.env.VECTORS) return null; |
| 327 | try { |
| 328 | const embedded = (await this.env.AI.run(EMBED_MODEL, { text: [query.slice(0, EMBED_CHARS)] })) as { data?: number[][] }; |
| 329 | await this.meter(workspace, Math.ceil(query.length / 4)); |
| 330 | return embedded.data?.[0] ?? null; |
| 331 | } catch (error) { |
| 332 | console.error("could not embed a query", error); |
| 333 | return null; |
| 334 | } |
| 335 | } |
| 336 | |
| 337 | // ---- Building the catalog -------------------------------------------------- |
| 338 | |
| 339 | /** The files of a project worth reading, with their blobs, found in a few tree reads. */ |
| 340 | private async candidates(project: Project, actor: User, ref: string): Promise<{ files: { path: string; hash: string }[]; siblings: string[]; head: string | null } | null> { |
| 341 | if (project.source.kind !== "hosted") return null; |
| 342 | const repos = reposClient(this.env.REPOS); |
| 343 | const { repo, rootDir: root } = project.source; |
| 344 | const at = (path: string) => [root, path].filter(Boolean).join("/"); |
| 345 | const top = await repos.tree(repo, actor, ref, root); |
| 346 | if (!top.ok) return null; |
| 347 | const files: { path: string; hash: string }[] = []; |
| 348 | const siblings = top.value.entries.map((entry) => entry.name); |
| 349 | const blobs = (entries: { name: string; hash: string; kind: string }[], dir: string) => { |
| 350 | for (const entry of entries) { |
| 351 | if (entry.kind !== "blob" && entry.kind !== "exec") continue; |
| 352 | const path = dir ? `${dir}/${entry.name}` : entry.name; |
| 353 | if (interesting(path)) files.push({ path, hash: entry.hash }); |
| 354 | } |
| 355 | }; |
| 356 | blobs(top.value.entries, ""); |
| 357 | const dirs = new Set(top.value.entries.filter((entry) => entry.kind === "tree").map((entry) => entry.name)); |
| 358 | const look = async (dir: string) => { |
| 359 | const found = await repos.tree(repo, actor, ref, at(dir)).catch(() => null); |
| 360 | return found?.ok ? found.value.entries : []; |
| 361 | }; |
| 362 | for (const dir of ["docs", "doc", "runbooks"]) if (dirs.has(dir)) blobs(await look(dir), dir); |
| 363 | for (const dir of [".g1t", ".github"]) { |
| 364 | if (!dirs.has(dir)) continue; |
| 365 | const inside = await look(dir); |
| 366 | blobs(inside, dir); |
| 367 | if (inside.some((entry) => entry.name === "workflows" && entry.kind === "tree")) blobs(await look(`${dir}/workflows`), `${dir}/workflows`); |
| 368 | } |
| 369 | return { files, siblings, head: top.value.head?.hash ?? null }; |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * Builds one project's place in the catalog from its default branch (or |
| 374 | * `commit`), reading only files whose blobs changed unless `force`. |
| 375 | * Indexes what changed and sends what its docs say to memory. |
| 376 | */ |
| 377 | async scan(project: Project, cache: WorkspaceCache, commit: string | null, force: boolean): Promise<ScanStats> { |
| 378 | const stats: ScanStats = { entities: 0, candidates: 0, kept: 0, indexed: 0 }; |
| 379 | if (project.source.kind !== "hosted") return stats; |
| 380 | const { repo, repoId, rootDir, defaultBranch } = project.source; |
| 381 | const ref = commit ?? defaultBranch; |
| 382 | const found = await this.candidates(project, cache.actor, ref); |
| 383 | if (!found) return stats; |
| 384 | const workspace = project.workspace; |
| 385 | |
| 386 | // What each file says: stored for unchanged blobs, read for the rest. |
| 387 | const stored = new Map( |
| 388 | ( |
| 389 | await this.db.prepare("SELECT path, hash, facts FROM files WHERE project_id = ?").bind(project.id).all<{ path: string; hash: string; facts: string }>() |
| 390 | ).results.map((row) => [row.path, row]), |
| 391 | ); |
| 392 | const files: FileRecord[] = []; |
| 393 | const changed: string[] = []; |
| 394 | const writes: D1PreparedStatement[] = []; |
| 395 | let reads = 0; |
| 396 | let workflowReads = 0; |
| 397 | const repos = reposClient(this.env.REPOS); |
| 398 | for (const file of found.files) { |
| 399 | const before = stored.get(file.path); |
| 400 | const workflow = file.path.includes("/workflows/"); |
| 401 | const fresh = before && before.hash === file.hash && !force; |
| 402 | const canRead = reads < MAX_READS && (!workflow || workflowReads < MAX_WORKFLOW_READS); |
| 403 | if (fresh || !canRead) { |
| 404 | if (before) files.push({ path: file.path, facts: JSON.parse(before.facts) as FileFacts }); |
| 405 | continue; |
| 406 | } |
| 407 | reads++; |
| 408 | if (workflow) workflowReads++; |
| 409 | const blob = await repos.blob(repo, cache.actor, found.head ?? ref, [rootDir, file.path].filter(Boolean).join("/")).catch(() => null); |
| 410 | if (!blob?.ok || blob.value.text == null) continue; |
| 411 | const facts = extract(file.path, blob.value.text, { project: project.name, siblings: found.siblings }); |
| 412 | files.push({ path: file.path, facts }); |
| 413 | changed.push(file.path); |
| 414 | writes.push( |
| 415 | this.db |
| 416 | .prepare( |
| 417 | `INSERT INTO files (project_id, path, hash, facts, read_at) VALUES (?, ?, ?, ?, ?) |
| 418 | ON CONFLICT (project_id, path) DO UPDATE SET hash = excluded.hash, facts = excluded.facts, read_at = excluded.read_at`, |
| 419 | ) |
| 420 | .bind(project.id, file.path, file.hash, JSON.stringify(facts), now()), |
| 421 | ); |
| 422 | } |
| 423 | const present = new Set(found.files.map((file) => file.path)); |
| 424 | for (const path of stored.keys()) { |
| 425 | if (!present.has(path)) writes.push(this.db.prepare("DELETE FROM files WHERE project_id = ? AND path = ?").bind(project.id, path)); |
| 426 | } |
| 427 | |
| 428 | // What g1t knows about it besides its files. |
| 429 | const [members, deploys, integrations, graph, log] = await Promise.all([ |
| 430 | cache.memberNames(), |
| 431 | cache.deployments(), |
| 432 | cache.integrations(), |
| 433 | projectsClient(this.env.PROJECTS).graph(project.id).catch(() => ({ dependsOn: [], usedBy: [] })), |
| 434 | repos.log(repo, cache.actor, found.head ?? ref, 100).catch(() => null), |
| 435 | ]); |
| 436 | const deploy = deploys.find((d) => d.slug === project.slug) ?? null; |
| 437 | const around: Surroundings = { |
| 438 | owners: authorsOf(log?.ok ? log.value : [], members), |
| 439 | dependsOn: graph.dependsOn.map((dep) => ({ slug: dep.slug, as: dep.as })), |
| 440 | deploy: deploy |
| 441 | ? { |
| 442 | enabled: deploy.enabled, |
| 443 | production: deploy.production ? { url: deploy.production.url, commit: deploy.production.commit, deployedAt: deploy.production.deployedAt } : null, |
| 444 | previews: deploy.previews, |
| 445 | latest: deploy.latest ? { status: deploy.latest.status, kind: deploy.latest.kind, error: deploy.latest.error, createdAt: deploy.latest.createdAt } : null, |
| 446 | } |
| 447 | : null, |
| 448 | integrations, |
| 449 | }; |
| 450 | const input: ProjectInput = { |
| 451 | id: project.id, |
| 452 | workspace, |
| 453 | slug: project.slug, |
| 454 | name: project.name, |
| 455 | description: project.description, |
| 456 | private: project.private, |
| 457 | repoId, |
| 458 | repo, |
| 459 | rootDir, |
| 460 | defaultBranch, |
| 461 | }; |
| 462 | const built = assemble(input, files, around); |
| 463 | const drafts: EntityDraft[] = [...built.entities, ...integrationEntities(integrations)]; |
| 464 | |
| 465 | // Entities: upserted; the project's own that it no longer has, removed. |
| 466 | const previous = new Map( |
| 467 | ( |
| 468 | await this.db |
| 469 | .prepare("SELECT id, name, summary FROM entities WHERE workspace = ? AND (project_id = ? OR project_id IS NULL)") |
| 470 | .bind(workspace, project.id) |
| 471 | .all<{ id: string; name: string; summary: string | null }>() |
| 472 | ).results.map((row) => [row.id, row]), |
| 473 | ); |
| 474 | const ids: string[] = []; |
| 475 | const toIndex: { id: string; text: string; meta: IndexMeta }[] = []; |
| 476 | const at = now(); |
| 477 | for (const draft of drafts) { |
| 478 | const id = await entityId(workspace, draft.kind, draft.key); |
| 479 | ids.push(id); |
| 480 | const shared = SHARED.has(draft.kind); |
| 481 | const source = draft.kind === "app" || draft.kind === "environment" ? "deployments" : draft.kind === "integration" ? "integrations" : "scan"; |
| 482 | writes.push( |
| 483 | this.db |
| 484 | .prepare( |
| 485 | `INSERT INTO entities (id, workspace, kind, key, name, summary, project_id, project, repo_id, private, data, source, ref, updated_at) |
| 486 | VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?) |
| 487 | ON CONFLICT (workspace, kind, key) DO UPDATE SET name = excluded.name, summary = excluded.summary, |
| 488 | project_id = excluded.project_id, project = excluded.project, repo_id = excluded.repo_id, private = excluded.private, |
| 489 | data = excluded.data, source = excluded.source, ref = excluded.ref, updated_at = excluded.updated_at`, |
| 490 | ) |
| 491 | .bind( |
| 492 | id, |
| 493 | workspace, |
| 494 | draft.kind, |
| 495 | draft.key, |
| 496 | draft.name, |
| 497 | draft.summary, |
| 498 | shared ? null : project.id, |
| 499 | shared ? null : project.slug, |
| 500 | shared ? null : repoId, |
| 501 | shared ? 0 : project.private ? 1 : 0, |
| 502 | JSON.stringify(draft.data), |
| 503 | source, |
| 504 | draft.ref, |
| 505 | at, |
| 506 | ), |
| 507 | ); |
| 508 | const before = previous.get(id); |
| 509 | if (force || !before || before.name !== draft.name || before.summary !== draft.summary) { |
| 510 | toIndex.push({ |
| 511 | id, |
| 512 | text: `${draft.kind} ${draft.name}. ${draft.summary ?? ""}`, |
| 513 | meta: { |
| 514 | workspace, |
| 515 | kind: draft.kind, |
| 516 | project: shared ? "" : project.slug, |
| 517 | private: shared ? false : project.private, |
| 518 | title: draft.name, |
| 519 | snippet: snippet(draft.summary), |
| 520 | url: draft.ref ?? "", |
| 521 | source: "catalog", |
| 522 | by: "", |
| 523 | at, |
| 524 | }, |
| 525 | }); |
| 526 | } |
| 527 | } |
| 528 | const gone = ( |
| 529 | await this.db |
| 530 | .prepare("SELECT id FROM entities WHERE project_id = ? AND id NOT IN (SELECT value FROM json_each(?))") |
| 531 | .bind(project.id, JSON.stringify(ids)) |
| 532 | .all<{ id: string }>() |
| 533 | ).results.map((row) => row.id); |
| 534 | if (gone.length) { |
| 535 | writes.push(this.db.prepare("DELETE FROM entities WHERE id IN (SELECT value FROM json_each(?))").bind(JSON.stringify(gone))); |
| 536 | writes.push(this.db.prepare("DELETE FROM items WHERE entity_id IN (SELECT value FROM json_each(?))").bind(JSON.stringify(gone))); |
| 537 | } |
| 538 | |
| 539 | // Relations: the project's own, replaced whole. |
| 540 | writes.push(this.db.prepare("DELETE FROM relations WHERE project_id = ?").bind(project.id)); |
| 541 | for (const relation of built.relations) { |
| 542 | const [from, to] = await Promise.all([entityId(workspace, relation.from.kind, relation.from.key), entityId(workspace, relation.to.kind, relation.to.key)]); |
| 543 | writes.push( |
| 544 | this.db |
| 545 | .prepare("INSERT OR REPLACE INTO relations (workspace, from_id, kind, to_id, project_id, updated_at) VALUES (?, ?, ?, ?, ?, ?)") |
| 546 | .bind(workspace, from, relation.kind, to, project.id, at), |
| 547 | ); |
| 548 | } |
| 549 | |
| 550 | // Docs that changed: their pieces, for search. |
| 551 | const repoPath = `/${repo.namespace}/${repo.name}`; |
| 552 | const chunkIds: string[] = []; |
| 553 | for (const file of files) { |
| 554 | const doc = file.facts.doc; |
| 555 | if (!doc || (!changed.includes(file.path) && !force)) continue; |
| 556 | const docId = await entityId(workspace, "doc", `${project.slug}:${doc.path}`); |
| 557 | writes.push(this.db.prepare("DELETE FROM items WHERE entity_id = ?").bind(docId)); |
| 558 | const url = `${repoPath}/blob/${defaultBranch}/${[rootDir, doc.path].filter(Boolean).join("/")}`; |
| 559 | doc.chunks.forEach((text, i) => { |
| 560 | const id = `${docId}:${i}`; |
| 561 | chunkIds.push(id); |
| 562 | writes.push( |
| 563 | this.db |
| 564 | .prepare( |
| 565 | `INSERT OR REPLACE INTO items (id, workspace, kind, entity_id, project_id, project, repo_id, private, title, text, url, by, updated_at) |
| 566 | VALUES (?, ?, 'doc', ?, ?, ?, ?, ?, ?, ?, ?, NULL, ?)`, |
| 567 | ) |
| 568 | .bind(id, workspace, docId, project.id, project.slug, repoId, project.private ? 1 : 0, `${doc.title} (${doc.path})`, text, url, at), |
| 569 | ); |
| 570 | toIndex.push({ |
| 571 | id, |
| 572 | text: `${doc.title}\n${text}`, |
| 573 | meta: { workspace, kind: "doc", project: project.slug, private: project.private, title: `${doc.title} (${doc.path})`, snippet: snippet(text), url, source: doc.path, by: "", at }, |
| 574 | }); |
| 575 | }); |
| 576 | } |
| 577 | writes.push( |
| 578 | this.db |
| 579 | .prepare( |
| 580 | `INSERT INTO scans (project_id, workspace, repo_id, "commit", tests, scanned_at) VALUES (?, ?, ?, ?, ?, ?) |
| 581 | ON CONFLICT (project_id) DO UPDATE SET workspace = excluded.workspace, "commit" = excluded."commit", tests = excluded.tests, scanned_at = excluded.scanned_at`, |
| 582 | ) |
| 583 | .bind(project.id, workspace, repoId, found.head, built.tests ? 1 : 0, at), |
| 584 | ); |
| 585 | for (let i = 0; i < writes.length; i += 50) await this.db.batch(writes.slice(i, i + 50)); |
| 586 | await this.unindex(gone); |
| 587 | stats.entities = drafts.length; |
| 588 | stats.indexed = await this.index(workspace, toIndex); |
| 589 | |
| 590 | // What changed files say worth remembering, as memory candidates. |
| 591 | const items: CaptureItem[] = built.hints |
| 592 | .filter((hint) => force || changed.includes(hint.path)) |
| 593 | .map((hint) => ({ |
| 594 | scope: "project", |
| 595 | repoId, |
| 596 | kind: hint.kind, |
| 597 | text: hint.text, |
| 598 | confidence: hint.confidence, |
| 599 | source: "doc", |
| 600 | reference: `doc:${repoId}:${[rootDir, hint.path].filter(Boolean).join("/")}`, |
| 601 | evidence: hint.evidence, |
| 602 | })); |
| 603 | for (let i = 0; i < items.length; i += 50) { |
| 604 | const captured = await memoryReviewClient(this.env.WORK) |
| 605 | .captureMemories(workspace, items.slice(i, i + 50), "g1t") |
| 606 | .catch(() => null); |
| 607 | stats.candidates += captured?.added ?? 0; |
| 608 | stats.kept += captured?.kept ?? 0; |
| 609 | } |
| 610 | return stats; |
| 611 | } |
| 612 | |
| 613 | // ---- Issues, pull requests and memory in search ------------------------ |
| 614 | |
| 615 | private async repoOf(repoId: string): Promise<{ namespace: string; name: string } | null> { |
| 616 | const response = await this.env.REPOS.fetch("https://repos/rpc/path_by_id", { |
| 617 | method: "POST", |
| 618 | headers: { "content-type": "application/json" }, |
| 619 | body: JSON.stringify({ id: repoId }), |
| 620 | }); |
| 621 | return response.ok ? ((await response.json()) as { namespace: string; name: string } | null) : null; |
| 622 | } |
| 623 | |
| 624 | /** Stores and indexes issues and pull requests as search rows. */ |
| 625 | private async putItems( |
| 626 | workspace: string, |
| 627 | project: Project | null, |
| 628 | repoId: string, |
| 629 | rows: { kind: "issue" | "pull"; number: number; title: string; body: string | null; by: string | null; updatedAt: string; url: string; private: boolean }[], |
| 630 | ): Promise<number> { |
| 631 | const writes: D1PreparedStatement[] = []; |
| 632 | const toIndex: { id: string; text: string; meta: IndexMeta }[] = []; |
| 633 | for (const row of rows) { |
| 634 | const id = `${row.kind}:${repoId}#${row.number}`; |
| 635 | const title = `#${row.number} ${row.title}`; |
| 636 | const text = (row.body ?? "").slice(0, ITEM_CHARS); |
| 637 | writes.push( |
| 638 | this.db |
| 639 | .prepare( |
| 640 | `INSERT OR REPLACE INTO items (id, workspace, kind, entity_id, project_id, project, repo_id, private, title, text, url, by, updated_at) |
| 641 | VALUES (?, ?, ?, NULL, ?, ?, ?, ?, ?, ?, ?, ?, ?)`, |
| 642 | ) |
| 643 | .bind(id, workspace, row.kind, project?.id ?? null, project?.slug ?? null, repoId, row.private ? 1 : 0, title, text, row.url, row.by, row.updatedAt), |
| 644 | ); |
| 645 | toIndex.push({ |
| 646 | id, |
| 647 | text: `${row.title}\n${text}`, |
| 648 | meta: { workspace, kind: row.kind, project: project?.slug ?? "", private: row.private, title, snippet: snippet(text || row.title), url: row.url, source: row.kind === "issue" ? "issue" : "pull request", by: row.by ?? "", at: row.updatedAt }, |
| 649 | }); |
| 650 | } |
| 651 | for (let i = 0; i < writes.length; i += 50) await this.db.batch(writes.slice(i, i + 50)); |
| 652 | return this.index(workspace, toIndex); |
| 653 | } |
| 654 | |
| 655 | private async indexIssueOrPull(kind: "issue" | "pull", repoId: string, number: number): Promise<void> { |
| 656 | const path = await this.repoOf(repoId); |
| 657 | if (!path) return; |
| 658 | const workspace = path.namespace.toLowerCase(); |
| 659 | const actor = await this.workspaceActor(workspace); |
| 660 | if (!actor) return; |
| 661 | const [repo, projects] = await Promise.all([reposClient(this.env.REPOS).get(path, actor), projectsClient(this.env.PROJECTS).byRepo(repoId)]); |
| 662 | if (!repo.ok || repo.value.forkOf) return; |
| 663 | const work = workClient(this.env.WORK); |
| 664 | const base = `/${path.namespace}/${path.name}`; |
| 665 | if (kind === "issue") { |
| 666 | const found = await work.getIssue(path, number, actor); |
| 667 | if (!found.ok) return; |
| 668 | const issue = found.value.issue; |
| 669 | await this.putItems(workspace, projects[0] ?? null, repoId, [ |
| 670 | { kind, number, title: issue.title, body: issue.body, by: issue.author.username, updatedAt: issue.updatedAt, url: `${base}/issues/${number}`, private: repo.value.isPrivate }, |
| 671 | ]); |
| 672 | } else { |
| 673 | const found = await work.getPull(path, number, actor); |
| 674 | if (!found.ok) return; |
| 675 | const pull = found.value.pull as { title: string; body: string | null; agent: string; updatedAt?: string; author?: { username: string } }; |
| 676 | await this.putItems(workspace, projects[0] ?? null, repoId, [ |
| 677 | { kind, number, title: pull.title, body: pull.body, by: pull.author?.username ?? pull.agent, updatedAt: pull.updatedAt ?? now(), url: `${base}/pull/${number}`, private: repo.value.isPrivate }, |
| 678 | ]); |
| 679 | } |
| 680 | } |
| 681 | |
| 682 | /** A memory in search while it is kept, and out of it otherwise. */ |
| 683 | private async indexMemories(workspace: string, memories: Memory[]): Promise<number> { |
| 684 | const kept = memories.filter((memory) => (memory.status ?? "kept") === "kept"); |
| 685 | await this.unindex(memories.filter((memory) => !kept.includes(memory)).map((memory) => `memory:${memory.id}`)); |
| 686 | const projects = await projectsClient(this.env.PROJECTS) |
| 687 | .list(workspace, (await this.workspaceActor(workspace)) ?? null) |
| 688 | .then((found) => (found.ok ? found.value : [])) |
| 689 | .catch(() => [] as Project[]); |
| 690 | const slugOf = (memory: Memory) => |
| 691 | memory.scope === "project" && memory.repo |
| 692 | ? (projects.find( |
| 693 | (project) => |
| 694 | project.source.kind === "hosted" && |
| 695 | project.primary && |
| 696 | project.source.repo.namespace.toLowerCase() === memory.repo!.namespace.toLowerCase() && |
| 697 | project.source.repo.name.toLowerCase() === memory.repo!.name.toLowerCase(), |
| 698 | )?.slug ?? "") |
| 699 | : ""; |
| 700 | return this.index( |
| 701 | workspace, |
| 702 | kept.map((memory) => ({ |
| 703 | id: `memory:${memory.id}`, |
| 704 | text: `${memory.kind}: ${memory.text}`, |
| 705 | meta: { |
| 706 | workspace, |
| 707 | kind: "memory", |
| 708 | project: slugOf(memory), |
| 709 | // Memory is for members, whatever its project. |
| 710 | private: true, |
| 711 | title: `${memory.kind[0].toUpperCase()}${memory.kind.slice(1)}${memory.scope === "workspace" ? " (workspace)" : ""}`, |
| 712 | snippet: snippet(memory.text), |
| 713 | url: memory.repo ? `/${memory.repo.namespace}/${memory.repo.name}/memory` : `/${workspace}/-/memory`, |
| 714 | source: memorySource(memory), |
| 715 | by: memory.createdBy, |
| 716 | at: memory.updatedAt, |
| 717 | }, |
| 718 | })), |
| 719 | ); |
| 720 | } |
| 721 | |
| 722 | // ---- Events -------------------------------------------------------------- |
| 723 | |
| 724 | async onEvent(event: G1tEvent): Promise<void> { |
| 725 | switch (event.type) { |
| 726 | case "git.push": { |
| 727 | if (!event.data.defaultBranch) return; |
| 728 | const projects = (await projectsClient(this.env.PROJECTS).byRepo(event.data.repoId)).slice(0, MAX_PROJECTS_PER_PUSH); |
| 729 | if (projects.length === 0) return; |
| 730 | const actor = await this.workspaceActor(projects[0].workspace); |
| 731 | if (!actor) return; |
| 732 | const cache = new WorkspaceCache(this.env, projects[0].workspace, actor); |
| 733 | for (const project of projects) await this.scan(project, cache, event.data.after, false); |
| 734 | return; |
| 735 | } |
| 736 | case "issue.opened": |
| 737 | case "issue.updated": |
| 738 | case "issue.closed": |
| 739 | return this.indexIssueOrPull("issue", event.data.repoId, event.data.number); |
| 740 | case "pull.ready": |
| 741 | case "pull.merged": |
| 742 | return this.indexIssueOrPull("pull", event.data.repoId, event.data.number); |
| 743 | case "memory.changed": { |
| 744 | const { memoryId, workspace, status } = event.data; |
| 745 | if (status !== "kept") return this.unindex([`memory:${memoryId}`]); |
| 746 | const memories = await memoryReviewClient(this.env.WORK).memoriesById(workspace, [memoryId]); |
| 747 | await this.indexMemories(workspace, memories); |
| 748 | return; |
| 749 | } |
| 750 | case "workspace.renamed": { |
| 751 | const current = await currentWorkspaceSlug(this.env.IDENTITY, event.data); |
| 752 | for (const old of staleSlugs(event.data, current)) { |
| 753 | await this.db.batch( |
| 754 | ["entities", "relations", "items", "scans", "usage"].map((table) => |
| 755 | this.db.prepare(`UPDATE OR IGNORE ${table} SET workspace = ? WHERE workspace = ?`).bind(current, old), |
| 756 | ), |
| 757 | ); |
| 758 | await this.db.prepare("DELETE FROM backfills WHERE workspace = ?").bind(old).run(); |
| 759 | } |
| 760 | // The index's rows carry the slug: build them again under the new one. |
| 761 | const actor = await this.workspaceActor(current); |
| 762 | if (actor) await this.backfill({ actor, workspace: current }); |
| 763 | return; |
| 764 | } |
| 765 | default: |
| 766 | return; |
| 767 | } |
| 768 | } |
| 769 | |
| 770 | // ---- Backfill ------------------------------------------------------------ |
| 771 | |
| 772 | private async backfillRow(workspace: string): Promise<Backfill | null> { |
| 773 | const row = await this.db.prepare("SELECT * FROM backfills WHERE workspace = ?").bind(workspace).first<Record<string, unknown>>(); |
| 774 | if (!row) return null; |
| 775 | return { |
| 776 | workspace, |
| 777 | status: row.status as Backfill["status"], |
| 778 | by: String(row.by), |
| 779 | projects: Number(row.projects), |
| 780 | done: Number(row.done), |
| 781 | entities: Number(row.entities), |
| 782 | candidates: Number(row.candidates), |
| 783 | kept: Number(row.kept), |
| 784 | indexed: Number(row.indexed), |
| 785 | error: (row.error as string | null) ?? null, |
| 786 | startedAt: String(row.started_at), |
| 787 | finishedAt: (row.finished_at as string | null) ?? null, |
| 788 | }; |
| 789 | } |
| 790 | |
| 791 | async backfill(a: { actor: User; workspace: string }): Promise<Result<Backfill>> { |
| 792 | const workspace = a.workspace.toLowerCase(); |
| 793 | if (!isMember(a.actor, workspace)) return fail("forbidden", "Only members can rebuild a workspace's context."); |
| 794 | const running = await this.backfillRow(workspace); |
| 795 | if (running?.status === "running" && Date.now() - Date.parse(running.startedAt) < BACKFILL_STALE_MS) return ok(running); |
| 796 | const actor = (await this.workspaceActor(workspace)) ?? a.actor; |
| 797 | const listed = await projectsClient(this.env.PROJECTS).list(workspace, actor); |
| 798 | if (!listed.ok) return listed; |
| 799 | const projects = listed.value.filter((project) => project.source.kind === "hosted").slice(0, BACKFILL_PROJECTS); |
| 800 | const at = now(); |
| 801 | await this.db |
| 802 | .prepare( |
| 803 | `INSERT OR REPLACE INTO backfills (workspace, status, by, projects, done, entities, candidates, kept, indexed, error, started_at, finished_at) |
| 804 | VALUES (?, ?, ?, ?, 0, 0, 0, 0, 0, NULL, ?, ?)`, |
| 805 | ) |
| 806 | .bind(workspace, projects.length ? "running" : "done", a.actor.username, projects.length, at, projects.length ? null : at) |
| 807 | .run(); |
| 808 | const jobs: { body: Job }[] = [ |
| 809 | ...projects.map((project) => ({ body: { type: "backfill_project", workspace, slug: project.slug } as Job })), |
| 810 | { body: { type: "backfill_memory", workspace } }, |
| 811 | ]; |
| 812 | for (let i = 0; i < jobs.length; i += 100) await this.env.JOBS.sendBatch(jobs.slice(i, i + 100)); |
| 813 | return ok((await this.backfillRow(workspace))!); |
| 814 | } |
| 815 | |
| 816 | async runJob(job: Job): Promise<void> { |
| 817 | const actor = await this.workspaceActor(job.workspace); |
| 818 | if (!actor) return; |
| 819 | if (job.type === "backfill_memory") { |
| 820 | const memories = await memoryReviewClient(this.env.WORK).searchMemories(job.workspace, null, { limit: 100 }); |
| 821 | const indexed = await this.indexMemories(job.workspace, memories); |
| 822 | await this.db.prepare("UPDATE backfills SET indexed = indexed + ? WHERE workspace = ?").bind(indexed, job.workspace).run(); |
| 823 | return; |
| 824 | } |
| 825 | const stats: ScanStats = { entities: 0, candidates: 0, kept: 0, indexed: 0 }; |
| 826 | let error: string | null = null; |
| 827 | try { |
| 828 | const found = await projectsClient(this.env.PROJECTS).get(job.workspace, job.slug, actor); |
| 829 | if (found.ok && found.value.source.kind === "hosted") { |
| 830 | const project = found.value; |
| 831 | const source = project.source as Extract<Project["source"], { kind: "hosted" }>; |
| 832 | const cache = new WorkspaceCache(this.env, job.workspace, actor); |
| 833 | Object.assign(stats, await this.scan(project, cache, null, true)); |
| 834 | if (project.primary) { |
| 835 | // Decisions from merged pull requests, and people's corrections in their reviews. |
| 836 | const seeded = await memoryReviewClient(this.env.WORK).seedFromPulls(source.repoId, BACKFILL_PULLS).catch(() => null); |
| 837 | stats.candidates += seeded?.added ?? 0; |
| 838 | stats.kept += seeded?.kept ?? 0; |
| 839 | const work = workClient(this.env.WORK); |
| 840 | const base = `/${source.repo.namespace}/${source.repo.name}`; |
| 841 | const [issues, pulls] = await Promise.all([ |
| 842 | work.listIssues(source.repo, actor).catch(() => null), |
| 843 | work.listPulls(source.repo, actor, "closed").catch(() => null), |
| 844 | ]); |
| 845 | stats.indexed += await this.putItems(job.workspace, project, source.repoId, [ |
| 846 | ...(issues?.ok ? issues.value.slice(0, BACKFILL_ITEMS) : []).map((issue) => ({ |
| 847 | kind: "issue" as const, |
| 848 | number: issue.number, |
| 849 | title: issue.title, |
| 850 | body: issue.body, |
| 851 | by: issue.author.username, |
| 852 | updatedAt: issue.updatedAt, |
| 853 | url: `${base}/issues/${issue.number}`, |
| 854 | private: project.private, |
| 855 | })), |
| 856 | ...(pulls?.ok ? pulls.value.filter((pull) => pull.status === "merged").slice(0, BACKFILL_ITEMS) : []).map((pull) => ({ |
| 857 | kind: "pull" as const, |
| 858 | number: pull.number, |
| 859 | title: pull.title, |
| 860 | body: pull.body, |
| 861 | by: (pull as { author?: { username: string } }).author?.username ?? pull.agent, |
| 862 | updatedAt: pull.mergedAt ?? now(), |
| 863 | url: `${base}/pull/${pull.number}`, |
| 864 | private: project.private, |
| 865 | })), |
| 866 | ]); |
| 867 | } |
| 868 | } |
| 869 | } catch (caught) { |
| 870 | error = String(caught).slice(0, 300); |
| 871 | console.error("backfill of", job.workspace, job.slug, "failed", caught); |
| 872 | } |
| 873 | await this.db |
| 874 | .prepare( |
| 875 | `UPDATE backfills SET done = done + 1, entities = entities + ?, candidates = candidates + ?, kept = kept + ?, indexed = indexed + ?, |
| 876 | error = COALESCE(?, error), |
| 877 | status = CASE WHEN done + 1 >= projects THEN 'done' ELSE status END, |
| 878 | finished_at = CASE WHEN done + 1 >= projects THEN ? ELSE finished_at END |
| 879 | WHERE workspace = ?`, |
| 880 | ) |
| 881 | .bind(stats.entities, stats.candidates, stats.kept, stats.indexed, error, now(), job.workspace) |
| 882 | .run(); |
| 883 | } |
| 884 | |
| 885 | // ---- Reading ------------------------------------------------------------- |
| 886 | |
| 887 | /** Who is reading, and what of the workspace they may see. */ |
| 888 | private async reader(workspace: string, viewer: Viewer): Promise<Reader> { |
| 889 | const member = isMember(viewer, workspace); |
| 890 | if (member) return { workspace, member, visible: new Set() }; |
| 891 | const listed = await projectsClient(this.env.PROJECTS).list(workspace, viewer).catch(() => null); |
| 892 | return { workspace, member, visible: new Set(listed?.ok ? listed.value.filter((p) => !p.private).map((p) => p.slug) : []) }; |
| 893 | } |
| 894 | |
| 895 | private visibleRow(row: { private: number; project: string | null }, reader: Reader): boolean { |
| 896 | return readable({ workspace: reader.workspace, kind: "entity", project: row.project ?? "", private: !!row.private }, reader); |
| 897 | } |
| 898 | |
| 899 | async status(a: { workspace: string; viewer: Viewer }): Promise<Result<ContextStatus>> { |
| 900 | const workspace = a.workspace.toLowerCase(); |
| 901 | if (!isMember(a.viewer, workspace)) return fail("not_found", "There is no such workspace."); |
| 902 | let backfill = await this.backfillRow(workspace); |
| 903 | // The hub is never empty for long: a workspace's first look starts it. |
| 904 | if (!backfill) { |
| 905 | const actor = await this.workspaceActor(workspace); |
| 906 | if (actor) { |
| 907 | const started = await this.backfill({ actor: { ...actor, username: "g1t" }, workspace }); |
| 908 | backfill = started.ok ? started.value : null; |
| 909 | } |
| 910 | } |
| 911 | const [counts, usage] = await Promise.all([ |
| 912 | this.db.prepare("SELECT kind, COUNT(*) AS n FROM entities WHERE workspace = ? GROUP BY kind").bind(workspace).all<{ kind: EntityKind; n: number }>(), |
| 913 | this.db.prepare("SELECT tokens, cost_micros FROM usage WHERE workspace = ? AND month = ?").bind(workspace, month()).first<{ tokens: number; cost_micros: number }>(), |
| 914 | ]); |
| 915 | return ok({ |
| 916 | backfill, |
| 917 | counts: Object.fromEntries(counts.results.map((row) => [row.kind, row.n])), |
| 918 | usage: { month: month(), tokens: usage?.tokens ?? 0, costMicros: usage?.cost_micros ?? 0 }, |
| 919 | semantic: !!(this.env.AI && this.env.VECTORS), |
| 920 | }); |
| 921 | } |
| 922 | |
| 923 | async catalog(a: { workspace: string; viewer: Viewer; kind?: EntityKind | null; project?: string | null }): Promise<Result<Catalog>> { |
| 924 | const workspace = a.workspace.toLowerCase(); |
| 925 | const reader = await this.reader(workspace, a.viewer); |
| 926 | let sql = "SELECT * FROM entities WHERE workspace = ?"; |
| 927 | const params: unknown[] = [workspace]; |
| 928 | if (a.kind) { |
| 929 | sql += " AND kind = ?"; |
| 930 | params.push(a.kind); |
| 931 | } |
| 932 | if (a.project) { |
| 933 | sql += " AND (project = ? OR project IS NULL)"; |
| 934 | params.push(a.project.toLowerCase()); |
| 935 | } |
| 936 | sql += " ORDER BY kind, name COLLATE NOCASE LIMIT 2000"; |
| 937 | const rows = (await this.db.prepare(sql).bind(...params).all<EntityRow>()).results.filter((row) => this.visibleRow(row, reader)); |
| 938 | const ids = new Set(rows.map((row) => row.id)); |
| 939 | const relations = ( |
| 940 | await this.db.prepare("SELECT from_id, kind, to_id FROM relations WHERE workspace = ? LIMIT 10000").bind(workspace).all<{ from_id: string; kind: RelationKind; to_id: string }>() |
| 941 | ).results |
| 942 | .filter((row) => ids.has(row.from_id) && ids.has(row.to_id)) |
| 943 | .map((row) => ({ from: row.from_id, kind: row.kind, to: row.to_id })); |
| 944 | const built = await this.db.prepare("SELECT MAX(scanned_at) AS at FROM scans WHERE workspace = ?").bind(workspace).first<{ at: string | null }>(); |
| 945 | return ok({ entities: rows.map(toEntity), relations, builtAt: built?.at ?? null }); |
| 946 | } |
| 947 | |
| 948 | async entity(a: { workspace: string; viewer: Viewer; kind: EntityKind; id: string }): Promise<Result<EntityDetail>> { |
| 949 | const workspace = a.workspace.toLowerCase(); |
| 950 | if (!ENTITY_KINDS.includes(a.kind)) return fail("invalid", `kind is one of ${ENTITY_KINDS.join(", ")}.`); |
| 951 | const reader = await this.reader(workspace, a.viewer); |
| 952 | const row = |
| 953 | (await this.db.prepare("SELECT * FROM entities WHERE workspace = ? AND id = ?").bind(workspace, a.id).first<EntityRow>()) ?? |
| 954 | (await this.db.prepare("SELECT * FROM entities WHERE workspace = ? AND kind = ? AND key = ? COLLATE NOCASE").bind(workspace, a.kind, a.id).first<EntityRow>()); |
| 955 | if (!row || row.kind !== a.kind || !this.visibleRow(row, reader)) return fail("not_found", `There is no such ${a.kind} in ${workspace}'s catalog.`); |
| 956 | const [out, into] = await Promise.all([ |
| 957 | this.db |
| 958 | .prepare("SELECT r.kind AS rel, e.* FROM relations r JOIN entities e ON e.id = r.to_id WHERE r.from_id = ? LIMIT 200") |
| 959 | .bind(row.id) |
| 960 | .all<EntityRow & { rel: RelationKind }>(), |
| 961 | this.db |
| 962 | .prepare("SELECT r.kind AS rel, e.* FROM relations r JOIN entities e ON e.id = r.from_id WHERE r.to_id = ? LIMIT 200") |
| 963 | .bind(row.id) |
| 964 | .all<EntityRow & { rel: RelationKind }>(), |
| 965 | ]); |
| 966 | const relations = [ |
| 967 | ...out.results.filter((r) => this.visibleRow(r, reader)).map((r) => ({ kind: r.rel, direction: "out" as const, entity: toEntity(r) })), |
| 968 | ...into.results.filter((r) => this.visibleRow(r, reader)).map((r) => ({ kind: r.rel, direction: "in" as const, entity: toEntity(r) })), |
| 969 | ]; |
| 970 | return ok({ entity: toEntity(row), relations }); |
| 971 | } |
| 972 | |
| 973 | async search(a: { workspace: string; viewer: Viewer; query: string; project?: string | null; kinds?: SearchKind[] | null; limit?: number | null }): Promise<Result<SearchResult>> { |
| 974 | const workspace = a.workspace.toLowerCase(); |
| 975 | const query = (a.query ?? "").trim().slice(0, 500); |
| 976 | if (!query) return fail("invalid", "Say what to search for."); |
| 977 | const reader = await this.reader(workspace, a.viewer); |
| 978 | if (!reader.member && reader.visible.size === 0) return ok({ query, hits: [], mode: "text" }); |
| 979 | const limit = Math.min(Math.max(a.limit ?? 20, 1), 50); |
| 980 | const kinds = allowedKinds(reader, a.kinds); |
| 981 | const wants = (kind: SearchKind) => !kinds || kinds.includes(kind); |
| 982 | const project = a.project?.toLowerCase() || null; |
| 983 | |
| 984 | // Semantic: the index, filtered to what this reader may see. |
| 985 | let semantic: SearchHit[] = []; |
| 986 | let mode: SearchResult["mode"] = "text"; |
| 987 | const vector = await this.embedQuery(workspace, query); |
| 988 | if (vector && this.env.VECTORS) { |
| 989 | try { |
| 990 | const found = await this.env.VECTORS.query(vector, { topK: 20, returnMetadata: "all", filter: indexFilter(reader, { project, kinds }) as VectorizeVectorMetadataFilter }); |
| 991 | mode = "semantic"; |
| 992 | semantic = found.matches |
| 993 | .map((match) => ({ id: match.id, score: match.score, meta: match.metadata as unknown as IndexMeta })) |
| 994 | .filter((match) => match.meta && readable(match.meta, reader)) |
| 995 | .map((match) => ({ |
| 996 | kind: match.meta.kind as SearchKind, |
| 997 | id: match.id.startsWith("memory:") ? match.id.slice(7) : match.id, |
| 998 | title: match.meta.title, |
| 999 | snippet: match.meta.snippet, |
| 1000 | project: match.meta.project || null, |
| 1001 | url: match.meta.url || null, |
| 1002 | score: match.score, |
| 1003 | source: match.meta.source, |
| 1004 | by: match.meta.by || null, |
| 1005 | updatedAt: match.meta.at || null, |
| 1006 | })); |
| 1007 | // Memory changes after it is indexed: show only what is still kept. |
| 1008 | const memoryIds = semantic.filter((hit) => hit.kind === "memory").map((hit) => hit.id); |
| 1009 | if (memoryIds.length) { |
| 1010 | const kept = new Set( |
| 1011 | (await memoryReviewClient(this.env.WORK).memoriesById(workspace, memoryIds)) |
| 1012 | .filter((memory) => (memory.status ?? "kept") === "kept") |
| 1013 | .map((memory) => memory.id), |
| 1014 | ); |
| 1015 | semantic = semantic.filter((hit) => hit.kind !== "memory" || kept.has(hit.id)); |
| 1016 | } |
| 1017 | } catch (error) { |
| 1018 | console.error("semantic search failed; matching words instead", error); |
| 1019 | } |
| 1020 | } |
| 1021 | |
| 1022 | // Text: every word, in the catalog, docs, issues and pull requests, and memory. |
| 1023 | const words = query.toLowerCase().split(/\s+/).filter(Boolean).slice(0, 6); |
| 1024 | const like = (columns: string) => words.map(() => `(${columns}) LIKE ?`).join(" AND "); |
| 1025 | const patterns = words.map((word) => `%${word.replace(/[%_]/g, "")}%`); |
| 1026 | const text: SearchHit[] = []; |
| 1027 | const entityKinds = ENTITY_KINDS.filter(wants); |
| 1028 | if (entityKinds.length) { |
| 1029 | const rows = await this.db |
| 1030 | .prepare( |
| 1031 | `SELECT * FROM entities WHERE workspace = ? AND kind IN (SELECT value FROM json_each(?)) ${project ? "AND (project = ? OR project IS NULL)" : ""} |
| 1032 | AND ${like("lower(name || ' ' || COALESCE(summary, ''))")} LIMIT 30`, |
| 1033 | ) |
| 1034 | .bind(workspace, JSON.stringify(entityKinds), ...(project ? [project] : []), ...patterns) |
| 1035 | .all<EntityRow>(); |
| 1036 | for (const row of rows.results.filter((r) => this.visibleRow(r, reader))) { |
| 1037 | text.push({ kind: row.kind, id: row.id, title: row.name, snippet: snippet(row.summary), project: row.project, url: row.ref, score: 0.3, source: "catalog", by: null, updatedAt: row.updated_at }); |
| 1038 | } |
| 1039 | } |
| 1040 | const itemKinds = (["doc", "issue", "pull"] as const).filter(wants); |
| 1041 | if (itemKinds.length) { |
| 1042 | const rows = await this.db |
| 1043 | .prepare( |
| 1044 | `SELECT * FROM items WHERE workspace = ? AND kind IN (SELECT value FROM json_each(?)) ${project ? "AND project = ?" : ""} |
| 1045 | AND ${like("lower(title || ' ' || text)")} ORDER BY updated_at DESC LIMIT 30`, |
| 1046 | ) |
| 1047 | .bind(workspace, JSON.stringify(itemKinds), ...(project ? [project] : []), ...patterns) |
| 1048 | .all<ItemRow>(); |
| 1049 | for (const row of rows.results.filter((r) => this.visibleRow(r, reader))) { |
| 1050 | text.push({ kind: row.kind, id: row.id, title: row.title, snippet: snippet(row.text), project: row.project, url: row.url, score: 0.2, source: row.kind === "doc" ? "doc" : row.kind, by: row.by, updatedAt: row.updated_at }); |
| 1051 | } |
| 1052 | } |
| 1053 | if (reader.member && wants("memory")) { |
| 1054 | const memories = await memoryReviewClient(this.env.WORK).searchMemories(workspace, query, { limit: 10 }).catch(() => [] as Memory[]); |
| 1055 | for (const memory of memories) { |
| 1056 | text.push({ |
| 1057 | kind: "memory", |
| 1058 | id: memory.id, |
| 1059 | title: `${memory.kind[0].toUpperCase()}${memory.kind.slice(1)}${memory.scope === "workspace" ? " (workspace)" : ""}`, |
| 1060 | snippet: snippet(memory.text), |
| 1061 | project: null, |
| 1062 | url: memory.repo ? `/${memory.repo.namespace}/${memory.repo.name}/memory` : `/${workspace}/-/memory`, |
| 1063 | score: memory.pinned ? 0.35 : 0.25, |
| 1064 | source: memorySource(memory), |
| 1065 | by: memory.createdBy, |
| 1066 | updatedAt: memory.updatedAt, |
| 1067 | }); |
| 1068 | } |
| 1069 | } |
| 1070 | return ok({ query, hits: merge(semantic, text, limit), mode }); |
| 1071 | } |
| 1072 | |
| 1073 | async scorecards(a: { workspace: string; viewer: Viewer }): Promise<Result<Scorecard[]>> { |
| 1074 | const workspace = a.workspace.toLowerCase(); |
| 1075 | if (!isMember(a.viewer, workspace)) return fail("forbidden", "Scorecards are for members of the workspace."); |
| 1076 | const listed = await projectsClient(this.env.PROJECTS).list(workspace, a.viewer); |
| 1077 | if (!listed.ok) return listed; |
| 1078 | const [entities, scans, deploys, security] = await Promise.all([ |
| 1079 | this.db |
| 1080 | .prepare("SELECT * FROM entities WHERE workspace = ? AND kind IN ('project', 'doc')") |
| 1081 | .bind(workspace) |
| 1082 | .all<EntityRow>(), |
| 1083 | this.db.prepare("SELECT project_id, tests FROM scans WHERE workspace = ?").bind(workspace).all<{ project_id: string; tests: number }>(), |
| 1084 | deploymentsClient(this.env.DEPLOYMENTS) |
| 1085 | .overview(workspace, a.viewer) |
| 1086 | .then((found) => (found.ok ? found.value : [])) |
| 1087 | .catch(() => [] as ProjectDeploys[]), |
| 1088 | this.env.SECURITY |
| 1089 | ? securityClient(this.env.SECURITY) |
| 1090 | .workspace(workspace, a.viewer) |
| 1091 | .then((found) => (found.ok ? found.value : null)) |
| 1092 | .catch(() => null) |
| 1093 | : Promise.resolve(null), |
| 1094 | ]); |
| 1095 | const tests = new Map(scans.results.map((row) => [row.project_id, !!row.tests])); |
| 1096 | const cards: Scorecard[] = []; |
| 1097 | for (const project of listed.value) { |
| 1098 | if (project.source.kind !== "hosted") continue; |
| 1099 | const own = entities.results.filter((row) => row.project_id === project.id); |
| 1100 | const entry = own.find((row) => row.kind === "project"); |
| 1101 | const data = entry ? toEntity(entry).data : {}; |
| 1102 | const deploy = deploys.find((d) => d.slug === project.slug); |
| 1103 | const repoId = project.source.repoId; |
| 1104 | const findings = security?.find((repo) => repo.repoId === repoId); |
| 1105 | const rules = evaluate({ |
| 1106 | name: project.name, |
| 1107 | owners: Array.isArray(data.owners) ? (data.owners as string[]) : [], |
| 1108 | docs: own.filter((row) => row.kind === "doc").map((row) => String(toEntity(row).data.path ?? "")), |
| 1109 | tests: tests.get(project.id) ?? false, |
| 1110 | testCommand: Array.isArray(data.testCommands) && data.testCommands.length ? String(data.testCommands[0]) : null, |
| 1111 | deploy: deploy |
| 1112 | ? { |
| 1113 | enabled: deploy.enabled, |
| 1114 | production: deploy.production ? { url: deploy.production.url } : null, |
| 1115 | latest: deploy.latest ? { kind: deploy.latest.kind, status: deploy.latest.status, error: deploy.latest.error } : null, |
| 1116 | } |
| 1117 | : null, |
| 1118 | secretFindings: security ? (findings?.secrets ?? 0) : null, |
| 1119 | }); |
| 1120 | const applies = rules.filter((rule) => rule.status !== "na"); |
| 1121 | cards.push({ |
| 1122 | project: project.slug, |
| 1123 | name: project.name, |
| 1124 | repo: project.source.repo, |
| 1125 | passed: applies.filter((rule) => rule.status === "pass").length, |
| 1126 | total: applies.length, |
| 1127 | rules, |
| 1128 | }); |
| 1129 | } |
| 1130 | return ok(cards); |
| 1131 | } |
| 1132 | |
| 1133 | /** The Context section for an agent starting work. Never fails a run: on any trouble, nothing. */ |
| 1134 | async runContext(a: { repoId: string; task?: string; budget?: number }): Promise<RunContext> { |
| 1135 | try { |
| 1136 | const projects = (await projectsClient(this.env.PROJECTS).byRepo(a.repoId)).slice(0, 2); |
| 1137 | if (projects.length === 0) return { text: null, sources: [] }; |
| 1138 | const workspace = projects[0].workspace; |
| 1139 | const actor = await this.workspaceActor(workspace); |
| 1140 | if (!actor) return { text: null, sources: [] }; |
| 1141 | const cache = new WorkspaceCache(this.env, workspace, actor); |
| 1142 | const deploys = await cache.deployments(); |
| 1143 | const live = new Map(deploys.map((d) => [d.slug, d])); |
| 1144 | const contexts: ProjectContext[] = []; |
| 1145 | for (const project of projects) { |
| 1146 | if (project.source.kind !== "hosted") continue; |
| 1147 | const rows = (await this.db.prepare("SELECT * FROM entities WHERE workspace = ? AND project_id = ?").bind(workspace, project.id).all<EntityRow>()).results.map(toEntity); |
| 1148 | const entry = rows.find((row) => row.kind === "project"); |
| 1149 | const graph = await projectsClient(this.env.PROJECTS).graph(project.id).catch(() => ({ dependsOn: [], usedBy: [] })); |
| 1150 | const deploy = live.get(project.slug); |
| 1151 | contexts.push({ |
| 1152 | slug: project.slug, |
| 1153 | name: project.name, |
| 1154 | repo: `${project.source.repo.namespace}/${project.source.repo.name}`, |
| 1155 | rootDir: project.source.rootDir, |
| 1156 | languages: Array.isArray(entry?.data.languages) ? (entry!.data.languages as string[]) : [], |
| 1157 | packages: rows.filter((row) => row.kind === "package").map((row) => row.name).slice(0, 5), |
| 1158 | testCommands: Array.isArray(entry?.data.testCommands) ? (entry!.data.testCommands as string[]) : [], |
| 1159 | owners: Array.isArray(entry?.data.owners) ? (entry!.data.owners as string[]) : [], |
| 1160 | dependsOn: graph.dependsOn.map((dep) => ({ slug: dep.slug, as: dep.as, url: live.get(dep.slug)?.production?.url ?? null })), |
| 1161 | usedBy: graph.usedBy.map((dep) => ({ slug: dep.slug })), |
| 1162 | environments: deploy?.enabled |
| 1163 | ? [{ name: "Production", url: deploy.production?.url ?? null, status: deploy.latest?.kind === "production" ? deploy.latest.status : deploy.production ? "ready" : null }] |
| 1164 | : [], |
| 1165 | docs: rows.filter((row) => row.kind === "doc").map((row) => String(row.data.path ?? row.name)), |
| 1166 | }); |
| 1167 | } |
| 1168 | const slugs = new Set(["", ...projects.map((project) => project.slug)]); |
| 1169 | const review = memoryReviewClient(this.env.WORK); |
| 1170 | // The memories closest to the task, less the pinned ones every run already has. |
| 1171 | let memories: ContextNote[] = []; |
| 1172 | const task = (a.task ?? "").trim(); |
| 1173 | const vector = task ? await this.embedQuery(workspace, task.slice(0, 1500)) : null; |
| 1174 | if (vector && this.env.VECTORS) { |
| 1175 | const found = await this.env.VECTORS.query(vector, { topK: 20, returnMetadata: "all", filter: { workspace, kind: "memory" } }).catch(() => null); |
| 1176 | const ids = (found?.matches ?? []) |
| 1177 | .filter((match) => slugs.has(String((match.metadata as { project?: string } | undefined)?.project ?? "")) && match.score >= 0.5) |
| 1178 | .map((match) => match.id.slice("memory:".length)); |
| 1179 | if (ids.length) { |
| 1180 | const byId = new Map((await review.memoriesById(workspace, ids)).map((memory) => [memory.id, memory])); |
| 1181 | memories = ids |
| 1182 | .map((id) => byId.get(id)) |
| 1183 | .filter((memory): memory is Memory => !!memory && (memory.status ?? "kept") === "kept" && !memory.pinned) |
| 1184 | .slice(0, 6) |
| 1185 | .map((memory) => ({ id: memory.id, kind: memory.kind, text: memory.text, source: memorySource(memory) })); |
| 1186 | } |
| 1187 | } |
| 1188 | const shown = new Set(memories.map((note) => note.id)); |
| 1189 | const decisions = (await review.searchMemories(workspace, null, { repoIds: [a.repoId], limit: 100 }).catch(() => [] as Memory[])) |
| 1190 | .filter((memory) => memory.kind === "decision" && !memory.pinned && !shown.has(memory.id)) |
| 1191 | .sort((x, y) => y.updatedAt.localeCompare(x.updatedAt)) |
| 1192 | .slice(0, 3) |
| 1193 | .map((memory) => ({ id: memory.id, kind: memory.kind, text: memory.text, source: memorySource(memory) })); |
| 1194 | return composeRunContext({ projects: contexts, memories, decisions, budget: Math.min(Math.max(a.budget ?? 4000, 500), 12_000) }); |
| 1195 | } catch (error) { |
| 1196 | console.error("no run context for", a.repoId, error); |
| 1197 | return { text: null, sources: [] }; |
| 1198 | } |
| 1199 | } |
| 1200 | } |
| 1201 | |
| 1202 | export default { |
| 1203 | async fetch(request: Request, env: Env): Promise<Response> { |
| 1204 | const match = new URL(request.url).pathname.match(/^\/rpc\/([a-z_]+)$/); |
| 1205 | if (request.method !== "POST" || !match) return new Response("Not found\n", { status: 404 }); |
| 1206 | const service = new Context(env); |
| 1207 | const args = (await request.json().catch(() => ({}))) as any; |
| 1208 | switch (match[1]) { |
| 1209 | case "catalog": |
| 1210 | return Response.json(await service.catalog(args)); |
| 1211 | case "entity": |
| 1212 | return Response.json(await service.entity(args)); |
| 1213 | case "search": |
| 1214 | return Response.json(await service.search(args)); |
| 1215 | case "scorecards": |
| 1216 | return Response.json(await service.scorecards(args)); |
| 1217 | case "backfill": |
| 1218 | return Response.json(await service.backfill(args)); |
| 1219 | case "status": |
| 1220 | return Response.json(await service.status(args)); |
| 1221 | case "run_context": |
| 1222 | return Response.json(await service.runContext(args)); |
| 1223 | default: |
| 1224 | return new Response("Unknown method\n", { status: 404 }); |
| 1225 | } |
| 1226 | }, |
| 1227 | |
| 1228 | async queue(batch: MessageBatch<G1tEvent | Job>, env: Env): Promise<void> { |
| 1229 | const service = new Context(env); |
| 1230 | for (const message of batch.messages) { |
| 1231 | try { |
| 1232 | if (batch.queue === "g1t-context-jobs") await service.runJob(message.body as Job); |
| 1233 | else await service.onEvent(message.body as G1tEvent); |
| 1234 | message.ack(); |
| 1235 | } catch (error) { |
| 1236 | console.error("context could not handle", (message.body as { type?: string }).type, error); |
| 1237 | message.retry(); |
| 1238 | } |
| 1239 | } |
| 1240 | }, |
| 1241 | } satisfies ExportedHandler<Env, G1tEvent | Job>; |