import-kepr2.ts

7.8 KB · raw

1// One-time Kepr 2 import (§28), deleted after cutover; the only bucket writer.2// --dry-run writes nothing.3import { dirname, join } from "node:path";4import { blake3 } from "@noble/hashes/blake3.js";5import { bytesToHex } from "@noble/hashes/utils.js";6import { SQL } from "bun";7import { openBucket } from "../src/bucket/reader.ts";8import { checkRelease } from "../src/bucket/repo.ts";9import { compareVersions } from "../src/releases/releases.ts";10import { openKeprDb } from "../src/server/db.ts";1112type Row = Record<string, string | number | null>;13const env = (name: string) => process.env[name] ?? "";14const dry = process.argv.includes("--dry-run");15const pg = new SQL(env("KEPR2_DATABASE_URL"));16const kepr = openKeprDb(env("KEPR_DATA"));17const folder = env("KOH_FOLDER");18const VARS = "ENDPOINT BUCKET REGION ACCESS_KEY_ID SECRET_ACCESS_KEY".split(" ");19const [endpoint, name, region, accessKeyId, secretAccessKey] = VARS.map((v) => env(`KOH_S3_${v}`));20const s3 = { endpoint, region, accessKeyId, secretAccessKey };21const bucket = openBucket(folder ? { kind: "folder", path: folder } : { kind: "s3", name, ...s3 });22/** put_if_absent (§9.1); throws when `key` holds other bytes. */23async function put(key: string, body: string | Uint8Array): Promise<void> {24	const bytes = typeof body === "string" ? new TextEncoder().encode(body) : body;25	const had = (await bucket.get(key)) ?? (dry ? null : await write(key, bytes));26	console.log(had ? "have" : dry ? "would put" : "put", key);27	if (had && !Bun.deepEquals(had, bytes)) throw new Error(`${key} holds other bytes`);28}29// Bun's S3 client can't write conditionally, so a presigned PUT does;30// a 412 returns the racer's bytes.31async function write(key: string, bytes: Uint8Array): Promise<Uint8Array | null> {32	if (folder) return Bun.write(join(folder, key), bytes).then(() => null);33	const url = new Bun.S3Client({ ...s3, bucket: name }).presign(key, { method: "PUT" });34	const res = await fetch(url, { method: "PUT", body: bytes, headers: { "If-None-Match": "*" } });35	if (res.ok || res.status === 412) return res.ok ? null : bucket.get(key);36	throw new Error(`put ${key}: ${res.status}`);37}38function insert(table: string, row: Row, tail: string): number | null {39	const cols = Object.keys(row);40	const sql = `INSERT INTO ${table} (${cols})41		VALUES (${cols.map((c) => `$${c}`)})42		${tail}43		RETURNING rowid AS id`;44	const id = kepr.query<{ id: number }, [Row]>(sql).get(row)?.id ?? null;45	console.log(table, JSON.stringify(row).slice(0, 80), id ? "set" : "have");46	return id;47}48const errors: string[] = [],49	writes: [string, string | Uint8Array][] = [];50kepr.run("BEGIN");51const upsert = "ON CONFLICT DO UPDATE SET public = excluded.public, category = excluded.category";52for (const repo of await pg<Row[]>`53	SELECT54		name,55		(private = 0)::int AS public,56		trim(split_part(COALESCE(categories, ''), ',', 1)) AS category57	FROM repos`) {58	insert("repos", repo, upsert);59}60// Kepr 2's four statuses become two (Phase 5); "enhancement" is an idea.61// Kepr 2 wrote the operator's replies as "operator" (NULL before accounts);62// a third account's reply has no author kind here.63for (const { replies, ...i } of await pg<(Row & { replies: Row[] })[]>`64	SELECT65		i.repo,66		i.number,67		i.title,68		i.description AS body,69		CASE WHEN i.type IN ('bug', 'question') THEN i.type ELSE 'idea' END AS type,70		COALESCE(i.filed_ver, '') AS version,71		COALESCE(NULLIF(i.reporter_name, ''), '@' || a.username, '') AS reporter_name,72		COALESCE(NULLIF(i.contact_email, ''), i.contact, '') AS contact_email,73		i.filed_at,74		CASE75			WHEN i.status IN ('resolved', 'wontfix', 'closed') THEN 'closed'76			ELSE 'open'77		END AS status,78		CASE79			WHEN i.closed_save ~ '^[0-9a-f]{12}$' THEN i.closed_save80			ELSE ''81		END AS fixed_in,82		CASE83			WHEN i.status IN ('resolved', 'wontfix', 'closed') THEN COALESCE(i.closed_at, i.filed_at)84		END AS closed_at,85		(86			SELECT COALESCE(87				json_agg(88					json_build_object(89						'author',90						CASE91							WHEN e.account_id IS NULL92								OR e.account_id = 'operator'93								OR o.tier = 'operator'94								THEN 'operator'95							WHEN e.account_id = i.account_id THEN 'reporter'96						END,97						'body', e.body,98						'created_at', e.created_at99					)100					ORDER BY e.created_at, e.id101				),102				'[]'103			)104			FROM issue_entries e105			LEFT JOIN accounts o ON o.id = e.account_id106			WHERE e.issue_id = i.id107				AND e.entry_type = 'reply'108				AND e.body IS NOT NULL109		) AS replies110	FROM issues i111	LEFT JOIN accounts a ON a.id = i.account_id112	ORDER BY i.repo, i.number`) {113	const id = insert("issues", i, "ON CONFLICT DO NOTHING");114	if (replies.some((r) => !r.author)) errors.push(`${i.repo}#${i.number}: third-party reply`);115	else {116		for (const r of id ? replies : []) {117			const name = r.author === "reporter" ? i.reporter_name : "";118			insert("issue_entries", { issue_id: id, ...r, name }, "");119		}120	}121}122// Kepr 2 counted Nix fetches as installs (kept 90 days).123// Its archive downloads have no home in Kepr 3.124for (const count of await pg<Row[]>`125	SELECT126		repo_id AS repo,127		COUNT(*) AS count128	FROM engagement_events129	WHERE event_type = 'nix_install'130		AND repo_id IS NOT NULL131	GROUP BY repo_id`) {132	insert("installs", count, "ON CONFLICT DO NOTHING");133}134// Only what Kepr 2 served: successful and partial builds, a file stored as135// the repo's name as <repo>-<platform>, older uploads named by platform;136// builds off hid a repo's. Save, notes: newest build's.137type Release = { repo: string; version: string; hidden: boolean; record: Row; stored: string[] };138const newest = new Map<string, string>();139for (const { repo, version, hidden, record, stored } of await pg<Release[]>`140	SELECT141		a.repo_id AS repo,142		a.version,143		bool_or(r.builds_enabled = 0) AS hidden,144		json_agg(145			concat_ws(146				'/',147				'releases',148				a.repo_id,149				a.version,150				a.platform,151				a.filename152			)153			ORDER BY a.platform154		) AS stored,155		json_build_object(156			'format', 1,157			'repo', a.repo_id,158			'version', a.version,159			'save', (array_agg(b.save_id ORDER BY b.started_at DESC))[1],160			'published_at', max(COALESCE(b.finished_at, a.built_at)),161			'machine', 'kepr2',162			'notes', COALESCE((array_agg(b.release_notes ORDER BY b.started_at DESC))[1], ''),163			'files',164			json_agg(165				json_build_object(166					'platform', a.platform,167					'name',168					CASE169						WHEN a.filename = a.repo_id THEN a.repo_id || '-' || a.platform170						ELSE a.filename171					END,172					'size', a.size_bytes,173					'blake3', a.blake3,174					'sha256', a.sha256175				)176				ORDER BY a.platform177			)178		) AS record179	FROM release_artifacts a180	JOIN build_runs b ON b.id = a.build_run181	JOIN repos r ON r.name = a.repo_id182	WHERE b.status IN ('success', 'partial')183	GROUP BY a.repo_id, a.version184	ORDER BY a.repo_id, a.version`) {185	const prefix = `${repo}/releases/${version}`;186	try {187		for (const [n, f] of checkRelease(repo, version, record).files.entries()) {188			const path = join(env("KEPR2_DATA"), stored[n] ?? "");189			const file = Bun.file(path);190			const bytes = await ((await file.exists()) ? file : Bun.file(dirname(path))).bytes();191			if (bytesToHex(blake3(bytes)) !== f.blake3) throw new Error(`${f.platform}: hash mismatch`);192			writes.push([`${prefix}/${f.platform}/${f.name}`, bytes]);193		}194		writes.push([`${prefix}/release.json`, JSON.stringify(record)]);195		if (hidden) insert("release_settings", { repo, version, hidden: 1 }, "ON CONFLICT DO NOTHING");196		if (compareVersions(version, newest.get(repo) ?? "") > 0) newest.set(repo, version);197	} catch (error) {198		errors.push(`${prefix}: ${error instanceof Error ? error.message : error}`);199	}200}201// Fail before any write (kepr.db is not yet written); writes never overwrite, so a re-run finishes one.202if (errors.length > 0) throw new Error(`not imported, nothing written:\n${errors.join("\n")}`);203for (const [repo, version] of newest) writes.push([`${repo}/releases/latest`, version]);204for (const [key, body] of writes) await put(key, body);205kepr.run(dry ? "ROLLBACK" : "END");206await pg.close();