╭─────────────╮ │scroll to top│ ╰─────────────╯ ╭ event.zig ────────────────────────────────────────────────────╮ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ ╰───────────────────────────────────────────────────────────────╯
╭────╮ ╭ clone ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╮ │http│ ssh │ │ ╰────╯ ╰ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╯
1297 if (parent_id_to_field_to_oid) |map| { 1298 if (try map.getCursor(record_key)) |field_oids_cursor| { 1299 parent_field_oids = try DB.SortedMap(.read_only).init(field_oids_cursor); 1300 } 1301 } 1302 // created only once a field actually comes from the parent 1303 var field_oids: ?DB.SortedMap(.read_write) = null; 1304 1305 // the fields both sides changed differently, space separated 1306 var conflicted_fields: [conflictedFieldsMaxLen(T)]u8 = undefined; 1307 var conflicted_len: usize = 0; 1308 1309 inline for (std.meta.fields(T), 0..) |field, i| { 1310 switch (outcome[i]) { 1311 // the target's value stands, so its oid does too 1312 .kept => {}, 1313 // the parent set this value, so its commit takes the field. a 1314 // parent with no oid took its own value from a merge, leaving 1315 // the field unattributed rather than crediting the target's 1316 // commit for a value it no longer holds. 1317 .parent => { 1318 if (field_oids == null) { 1319 field_oids = try DB.SortedMap(.read_write).init(try id_to_field_to_oid.putCursor(record_key)); 1320 } 1321 const map = field_oids orelse unreachable; 1322 const parent_oid = if (parent_field_oids) |parent_map| try parent_map.getCursor(field.name) else null; 1323 if (parent_oid) |oid_cursor| { 1324 try map.put(field.name, .{ .slot = oid_cursor.slot() }); 1325 } else { 1326 _ = try map.remove(field.name); 1327 } 1328 }, 1329 .conflicted => { 1330 if (conflicted_len > 0) { 1331 conflicted_fields[conflicted_len] = ' '; 1332 conflicted_len += 1; 1333 } 1334 @memcpy(conflicted_fields[conflicted_len..][0..field.name.len], field.name); 1335 conflicted_len += field.name.len; 1336 }, 1337 } 1338 } 1339 1340 if (conflicted_len > 0) { 1341 // each side's whole record is stored by reference, so the view can 1342 // show any of them or take theirs wholesale 1343 _ = try conflicts.remove(&merged_key); 1344 const conflict = try DB.HashMap(.read_write).init(try conflicts.putCursor(&merged_key)); 1345 try conflict.put(hash.hashInt(hash_kind, conflicted_fields_key), .{ .bytes = conflicted_fields[0..conflicted_len] }); 1346 try conflict.put(hash.hashInt(hash_kind, base_record_key), .{ .slot = baseline_slot }); 1347 try conflict.put(hash.hashInt(hash_kind, their_record_key), .{ .slot = parent_slot }); 1348 if (parent_field_oids) |map| { 1349 try conflict.put(hash.hashInt(hash_kind, their_field_to_oid_key), .{ .slot = map.slot() }); 1350 } 1351 continue; 1352 } 1353 1354 // created_order never changes, so the parent keys this record's conflict 1355 // identically, and it carries over when the target has none, unless the 1356 // parent's entry is unchanged from the baseline, meaning the target 1357 // inherited it too and resolving it removed it here. 1358 if (parent_conflicts) |map| { 1359 if (null == try conflicts.getCursor(&merged_key)) { 1360 if (try map.getSlot(&merged_key)) |conflict_slot| { 1361 const baseline_conflict_slot = if (baseline_conflicts) |baseline_map| try baseline_map.getSlot(&merged_key) else null; 1362 const resolved = if (baseline_conflict_slot) |slot| slot.eql(conflict_slot) else false; 1363 if (!resolved) try conflicts.put(&merged_key, .{ .slot = conflict_slot }); 1364 } 1365 } 1366 } 1367 } 1368 } 1369 1370 pub fn currentMoment( 1371 comptime repo_opts: rp.RepoOpts(.xit), 1372 repo: *rp.Repo(.xit, repo_opts), 1373 ) !rp.Repo(.xit, repo_opts).DB.HashMap(.read_only) { 1374 return currentMomentFromDb(repo_opts.hash, &repo.core.db); 1375 } 1376 1377 pub fn currentMomentFromDb( 1378 comptime hash_kind: hash.HashKind, 1379 db: *EventDB(hash_kind), 1380 ) !EventDB(hash_kind).HashMap(.read_only) { 1381 const DB = EventDB(hash_kind); 1382 const history = try DB.ArrayList(.read_only).init(db.rootCursor().readOnly()); 1383 const moment_cursor = try history.getCursor(-1) orelse return error.NotFound; 1384 const moment = try DB.HashMap(.read_only).init(moment_cursor); 1385 return try currentMomentFromRepoMoment(hash_kind, moment); 1386 } 1387 1388 pub fn currentMomentFromRepoMoment( 1389 comptime hash_kind: hash.HashKind, 1390 moment: EventDB(hash_kind).HashMap(.read_only), 1391 ) !EventDB(hash_kind).HashMap(.read_only) { 1392 const DB = EventDB(hash_kind); 1393 const last_object_id_cursor = try moment.getCursor(hash.hashInt(hash_kind, last_object_id_key)) orelse return error.NotFound; 1394 var last_object_id: [hash.byteLen(hash_kind)]u8 = undefined; 1395 _ = try last_object_id_cursor.readBytes(&last_object_id); 1396 const haxy_cursor = try moment.getCursor(hash.hashInt(hash_kind, materialized_key)) orelse return error.NotFound; 1397 const haxy = try DB.ArrayList(.read_only).init(haxy_cursor); 1398 const haxy_moments_cursor = try haxy.getCursor(-1) orelse return error.NotFound; 1399 const haxy_moments = try DB.HashMap(.read_only).init(haxy_moments_cursor); 1400 const haxy_moment_cursor = try haxy_moments.getCursor(hash.bytesToInt(hash_kind, &last_object_id)) orelse return error.NotFound; 1401 return try DB.HashMap(.read_only).init(haxy_moment_cursor); 1402 } 1403 1404 // build a sorted-set key from an order and event id. big-endian order makes 1405 // byte order match numeric order; the event id breaks ties. 1406 pub fn orderKey(order: u64, event_id: *const [event_id_size]u8) [@sizeOf(u64) + event_id_size]u8 { 1407 var key: [@sizeOf(u64) + event_id_size]u8 = undefined; 1408 std.mem.writeInt(u64, key[0..@sizeOf(u64)], order, .big); 1409 @memcpy(key[@sizeOf(u64)..], event_id); 1410 return key; 1411 } 1412 1413 pub fn orderKeyDesc(order: u64, event_id: *const [event_id_size]u8) [@sizeOf(u64) + event_id_size]u8 { 1414 return orderKey(std.math.maxInt(u64) - order, event_id); 1415 } 1416 1417 pub fn removedRecord( 1418 comptime T: type, 1419 comptime DB: type, 1420 comptime hash_kind: hash.HashKind, 1421 haxy_moment: DB.HashMap(.read_only), 1422 existing_maybe: ?T, 1423 ) !T { 1424 var record = existing_maybe orelse return error.EventNotFound; 1425 record.removed = true; 1426 const order_cursor = try haxy_moment.getCursor(hash.hashInt(hash_kind, event_order_key)) orelse return error.CursorNotFound; 1427 record.updated_order = try order_cursor.readUint(); 1428 return record; 1429 } 1430 1431 // add or refresh one entry in the global logical-event index 1432 pub fn indexEvent( 1433 comptime DB: type, 1434 comptime hash_kind: hash.HashKind, 1435 haxy_moment: DB.HashMap(.read_write), 1436 event_id: *const [event_id_size]u8, 1437 kind: EventKind, 1438 existing_maybe: anytype, 1439 record: anytype, 1440 ) !void { 1441 const event_key = hash.hashInt(hash_kind, event_id); 1442 const kind_map_cursor = try haxy_moment.putCursor(hash.hashInt(hash_kind, event_index_key)); 1443 const kind_map = try DB.HashMap(.read_write).init(kind_map_cursor); 1444 try kind_map.put(event_key, .{ .bytes = @tagName(kind) }); 1445 1446 const active_cursor = try haxy_moment.putCursor(hash.hashInt(hash_kind, active_event_id_set_key)); 1447 const active = try DB.SortedSet(.read_write).init(active_cursor); 1448 const removed_cursor = try haxy_moment.putCursor(hash.hashInt(hash_kind, removed_event_id_set_key)); 1449 const removed_set = try DB.SortedSet(.read_write).init(removed_cursor); 1450 1451 if (existing_maybe) |existing| { 1452 const old_key = orderKeyDesc(existing.updated_order, event_id); 1453 _ = try active.remove(&old_key); 1454 _ = try removed_set.remove(&old_key); 1455 } 1456 1457 const order_key = orderKeyDesc(record.updated_order, event_id); 1458 if (record.removed) { 1459 try removed_set.put(&order_key); 1460 } else { 1461 try active.put(&order_key); 1462 } 1463 } 1464 1465 pub fn touchThread( 1466 comptime DB: type, 1467 comptime hash_kind: hash.HashKind, 1468 haxy_moment: DB.HashMap(.read_write), 1469 thread_id: *const [event_id_size]u8, 1470 order: u64, 1471 arena: *std.heap.ArenaAllocator, 1472 ) !void { 1473 const kind_map_cursor = try haxy_moment.getCursor(hash.hashInt(hash_kind, event_index_key)) orelse return; 1474 const kind_map = try DB.HashMap(.read_only).init(kind_map_cursor); 1475 const kind = (try readEventKind(hash_kind, kind_map, thread_id)) orelse return; 1476 switch (kind) { 1477 .discuss => try Discussion.touch(DB, hash_kind, haxy_moment, thread_id, order, arena), 1478 else => {}, 1479 } 1480 } 1481 1482 pub fn readEventKind( 1483 comptime hash_kind: hash.HashKind, 1484 kind_map: anytype, 1485 id: *const [event_id_size]u8, 1486 ) !?EventKind { 1487 const cursor = try kind_map.getCursor(hash.hashInt(hash_kind, id)) orelse return null; 1488 var buffer: [64]u8 = undefined; 1489 return std.meta.stringToEnum(EventKind, try cursor.readBytes(&buffer)) orelse error.InvalidEventKind; 1490 } 1491 1492 // the event id embedded in a set entry's order key 1493 pub fn readOrderKeyId(comptime DB: type, kv_pair_cursor: DB.Cursor(.read_only)) ![event_id_size]u8 { 1494 var cursor = kv_pair_cursor; 1495 const kv_pair = try cursor.readKeyValuePair(); 1496 var order_key: [@sizeOf(u64) + event_id_size]u8 = undefined; 1497 _ = try kv_pair.key_cursor.readBytes(&order_key); 1498 return order_key[@sizeOf(u64)..][0..event_id_size].*; 1499 } 1500 1501 // split a pushed "<owner>/<repo>" path into its two components, or null if it 1502 // isn't exactly two non-empty segments. 1503 pub fn parseOwnerRepoPath(path: []const u8) ?struct { owner: []const u8, name: []const u8 } { 1504 var it = std.mem.splitScalar(u8, path, '/'); 1505 const owner = it.next() orelse return null; 1506 const name = it.next() orelse return null; 1507 if (it.next() != null) return null; 1508 if (owner.len == 0 or name.len == 0) return null; 1509 1510 return .{ .owner = owner, .name = name }; 1511 } 1512 1513 pub const CreateRepoOptions = struct { 1514 read_access: Repo.Access = .private, 1515 }; 1516 1517 // resolve a pushed `<owner>/<repo>` to the hex event id that names its on-disk 1518 // directory under the repos dir 1519 pub fn resolveOrCreateRepo( 1520 io: std.Io, 1521 allocator: std.mem.Allocator, 1522 admin_repo_path: []const u8, 1523 owner_name: []const u8, 1524 repo_name: []const u8, 1525 create_options_maybe: ?CreateRepoOptions, 1526 ) !?[event_id_size * 2]u8 { 1527 var repo = rp.Repo(.xit, admin_repo_opts).open(io, allocator, .{ .path = admin_repo_path }) catch |err| switch (err) { 1528 error.RepoNotFound => return null, 1529 else => |e| return e, 1530 }; 1531 defer repo.deinit(io, allocator); 1532 1533 var arena = std.heap.ArenaAllocator.init(allocator); 1534 defer arena.deinit(); 1535 1536 const moment = try currentMoment(admin_repo_opts, &repo); 1537 1538 // an already-registered repo reuses its event id, so a re-push (or a clone) 1539 // lands in the same repo 1540 if (try Repo.readByOwnerAndName(AdminDB, admin_repo_opts.hash, moment, &arena, owner_name, repo_name)) |found| { 1541 return std.fmt.bytesToHex(found.event_id, .lower); 1542 } 1543 1544 const create_options = create_options_maybe orelse return null; 1545 1546 // the new repo is owned by the named user, read from the name index; an 1547 // unknown owner can't own a repo, so the push is rejected 1548 const name_to_user_id_cursor = try moment.getCursor(hash.hashInt(admin_repo_opts.hash, "name->user-id")) orelse return null; 1549 const name_to_user_id = try AdminDB.HashMap(.read_only).init(name_to_user_id_cursor); 1550 const owner_id_cursor = try name_to_user_id.getCursor(hash.hashInt(admin_repo_opts.hash, owner_name)) orelse return null; 1551 var owner_user_id: [event_id_size]u8 = undefined; 1552 _ = try owner_id_cursor.readBytes(&owner_user_id); 1553 1554 const owner = (try User.readById(AdminDB, admin_repo_opts.hash, moment, &arena, &owner_user_id)) orelse unreachable; 1555 1556 var id_bytes: [event_id_size]u8 = undefined; 1557 io.random(&id_bytes); 1558 const event_id_hex = std.fmt.bytesToHex(id_bytes, .lower); 1559 1560 try consume(.server, .admin, .xit, admin_repo_opts, io, allocator, &repo, events_ref, &[_]EventWithId{.{ 1561 .id = event_id_hex, 1562 .timestamp = @intCast(std.Io.Timestamp.now(io, .real).toSeconds()), 1563 .author = .{ .name = owner.event.name, .email = owner.event.email }, 1564 .event = .{ .repo = .{ 1565 .user_id = &owner_user_id, 1566 .name = repo_name, 1567 .description = "", 1568 .read_access = create_options.read_access, 1569 } }, 1570 }}); 1571 1572 return event_id_hex; 1573 } 1574 1575 // 1576 // reading from xitdb 1577 // 1578 1579 pub fn read( 1580 comptime T: type, 1581 comptime DB: type, 1582 comptime hash_kind: hash.HashKind, 1583 arena: *std.heap.ArenaAllocator, 1584 map: DB.HashMap(.read_only), 1585 ) !T { 1586 var event: T = undefined; 1587 1588 switch (@typeInfo(T)) { 1589 .@"struct" => |struct_info| { 1590 inline for (struct_info.fields) |field| { 1591 switch (@typeInfo(field.type)) { 1592 .pointer => |pointer_info| { 1593 if (pointer_info.size == .slice and pointer_info.child == u8 and pointer_info.is_const) { 1594 @field(event, field.name) = try readBytes(DB, hash_kind, arena.allocator(), map, field.name); 1595 } else { 1596 @compileError("unsupported read field type: " ++ @typeName(field.type)); 1597 } 1598 }, 1599 .array => |array_info| { 1600 if (array_info.child == u8) { 1601 const bytes = try readBytes(DB, hash_kind, arena.allocator(), map, field.name); 1602 if (bytes.len != array_info.len) return error.InvalidByteArrayLength; 1603 @memcpy(@field(event, field.name)[0..], bytes); 1604 } else { 1605 @compileError("unsupported read field type: " ++ @typeName(field.type)); 1606 } 1607 }, 1608 .bool => @field(event, field.name) = try readBool(DB, hash_kind, map, field.name), 1609 .int => |int_info| switch (int_info.signedness) { 1610 .unsigned => @field(event, field.name) = @intCast(try readUint(DB, hash_kind, map, field.name)), 1611 .signed => @field(event, field.name) = @intCast(try readInt(DB, hash_kind, map, field.name)), 1612 }, 1613 .@"enum" => { 1614 const bytes = try readBytes(DB, hash_kind, arena.allocator(), map, field.name); 1615 @field(event, field.name) = std.meta.stringToEnum(field.type, bytes) orelse return error.InvalidEnumTag; 1616 }, 1617 .@"union" => { 1618 if (!@hasDecl(field.type, "decode")) @compileError("unsupported read field type: " ++ @typeName(field.type)); 1619 @field(event, field.name) = try field.type.decode(try readBytes(DB, hash_kind, arena.allocator(), map, field.name)); 1620 }, 1621 .optional => |optional_info| { 1622 // a missing key is null 1623 if (try map.getCursor(hash.hashInt(hash_kind, field.name))) |cursor| { 1624 if (optional_info.child == []const u8) { 1625 @field(event, field.name) = try cursor.readBytesAlloc(arena.allocator(), null); 1626 } else switch (@typeInfo(optional_info.child)) { 1627 .array => |array_info| { 1628 if (array_info.child != u8) @compileError("unsupported read field type: " ++ @typeName(field.type)); 1629 var bytes: optional_info.child = undefined; 1630 const value = try cursor.readBytes(&bytes); 1631 if (value.len != bytes.len) return error.InvalidByteArrayLength; 1632 @field(event, field.name) = bytes; 1633 }, 1634 .@"struct" => @field(event, field.name) = try read( 1635 optional_info.child, 1636 DB, 1637 hash_kind, 1638 arena, 1639 try DB.HashMap(.read_only).init(cursor), 1640 ), 1641 else => @compileError("unsupported read field type: " ++ @typeName(field.type)), 1642 } 1643 } else { 1644 @field(event, field.name) = null; 1645 } 1646 }, 1647 .@"struct" => { 1648 const cursor = try map.getCursor(hash.hashInt(hash_kind, field.name)) orelse return error.NotFound; 1649 @field(event, field.name) = try read( 1650 field.type, 1651 DB, 1652 hash_kind, 1653 arena, 1654 try DB.HashMap(.read_only).init(cursor), 1655 ); 1656 }, 1657 else => @compileError("unsupported read field type: " ++ @typeName(field.type)), 1658 } 1659 } 1660 }, 1661 else => @compileError("read expects a struct"), 1662 } 1663 1664 return event; 1665 } 1666 1667 // whether two events carry the same data 1668 fn fieldsEqual(comptime T: type, a: T, b: T) bool { 1669 switch (@typeInfo(T)) { 1670 .@"struct" => |struct_info| { 1671 inline for (struct_info.fields) |field| { 1672 if (!fieldEqual(field.type, @field(a, field.name), @field(b, field.name))) return false; 1673 } 1674 }, 1675 else => @compileError("fieldsEqual expects a struct"), 1676 } 1677 1678 return true; 1679 } 1680 1681 pub fn fieldEqual(comptime Field: type, a: Field, b: Field) bool { 1682 switch (@typeInfo(Field)) { 1683 .pointer => |pointer_info| { 1684 if (pointer_info.size == .slice and pointer_info.child == u8) { 1685 return std.mem.eql(u8, a, b); 1686 } else { 1687 @compileError("unsupported field type: " ++ @typeName(Field)); 1688 } 1689 }, 1690 .array => |array_info| { 1691 if (array_info.child == u8) { 1692 return std.mem.eql(u8, &a, &b); 1693 } else { 1694 @compileError("unsupported field type: " ++ @typeName(Field)); 1695 } 1696 }, 1697 .void => return true, 1698 .bool, .int, .@"enum" => return a == b, 1699 .@"union" => { 1700 if (std.meta.activeTag(a) != std.meta.activeTag(b)) return false; 1701 return switch (a) { 1702 inline else => |value, tag| fieldEqual(@TypeOf(value), value, @field(b, @tagName(tag))), 1703 }; 1704 }, 1705 .@"struct" => return fieldsEqual(Field, a, b), 1706 .optional => |optional_info| { 1707 if (a) |a_value| { 1708 if (b) |b_value| return fieldEqual(optional_info.child, a_value, b_value); 1709 return false; 1710 } 1711 return b == null; 1712 }, 1713 else => @compileError("unsupported field type: " ++ @typeName(Field)), 1714 } 1715 } 1716 1717 fn readBytes( 1718 comptime DB: type, 1719 comptime hash_kind: hash.HashKind, 1720 allocator: std.mem.Allocator, 1721 map: DB.HashMap(.read_only), 1722 field_name: []const u8, 1723 ) ![]const u8 { 1724 const cursor = try map.getCursor(hash.hashInt(hash_kind, field_name)) orelse return error.NotFound; 1725 return try cursor.readBytesAlloc(allocator, null); 1726 } 1727 1728 fn readUint( 1729 comptime DB: type, 1730 comptime hash_kind: hash.HashKind, 1731 map: DB.HashMap(.read_only), 1732 field_name: []const u8, 1733 ) !u64 { 1734 const cursor = try map.getCursor(hash.hashInt(hash_kind, field_name)) orelse return error.NotFound; 1735 return try cursor.readUint(); 1736 } 1737 1738 fn readInt( 1739 comptime DB: type, 1740 comptime hash_kind: hash.HashKind, 1741 map: DB.HashMap(.read_only), 1742 field_name: []const u8, 1743 ) !i64 { 1744 const cursor = try map.getCursor(hash.hashInt(hash_kind, field_name)) orelse return error.NotFound; 1745 return try cursor.readInt(); 1746 } 1747 1748 fn readBool( 1749 comptime DB: type, 1750 comptime hash_kind: hash.HashKind, 1751 map: DB.HashMap(.read_only), 1752 field_name: []const u8, 1753 ) !bool { 1754 const cursor = try map.getCursor(hash.hashInt(hash_kind, field_name)) orelse return error.NotFound; 1755 var buffer: [5]u8 = undefined; 1756 const bytes = try cursor.readBytesObject(&buffer); 1757 const format_tag = bytes.format_tag orelse return error.InvalidFormatTag; 1758 if (!std.mem.eql(u8, &format_tag, "bl")) return error.InvalidFormatTag; 1759 if (std.mem.eql(u8, bytes.value, "true")) return true; 1760 if (std.mem.eql(u8, bytes.value, "false")) return false; 1761 return error.InvalidBool; 1762 } 1763 1764 // 1765 // writing to xitdb 1766 // 1767 1768 pub fn upsert( 1769 comptime T: type, 1770 comptime DB: type, 1771 comptime hash_kind: hash.HashKind, 1772 map: DB.HashMap(.read_write), 1773 event: T, 1774 ) !void { 1775 switch (@typeInfo(T)) { 1776 .@"struct" => |struct_info| { 1777 inline for (struct_info.fields) |field| { 1778 try upsertField(DB, hash_kind, map, field.name, field.type, @field(event, field.name)); 1779 } 1780 }, 1781 else => @compileError("upsert expects a struct"), 1782 } 1783 } 1784 1785 fn upsertField( 1786 comptime DB: type, 1787 comptime hash_kind: hash.HashKind, 1788 map: DB.HashMap(.read_write), 1789 comptime field_name: []const u8, 1790 comptime Field: type, 1791 value: Field, 1792 ) !void { 1793 const key = hash.hashInt(hash_kind, field_name); 1794 1795 switch (@typeInfo(Field)) { 1796 .pointer => |pointer_info| { 1797 if (pointer_info.size == .slice and pointer_info.child == u8) { 1798 try upsertBytes(DB, hash_kind, map, key, value); 1799 } else { 1800 @compileError("unsupported upsert field type: " ++ @typeName(Field)); 1801 } 1802 }, 1803 .array => |array_info| { 1804 if (array_info.child == u8) { 1805 try upsertBytes(DB, hash_kind, map, key, &value); 1806 } else { 1807 @compileError("unsupported upsert field type: " ++ @typeName(Field)); 1808 } 1809 }, 1810 .int => |int_info| switch (int_info.signedness) { 1811 .unsigned => { 1812 if (try map.getCursor(key)) |value_cursor| { 1813 if (try value_cursor.readUint() == value) { 1814 return; 1815 } 1816 } 1817 1818 try map.put(key, .{ .uint = value }); 1819 }, 1820 .signed => { 1821 if (try map.getCursor(key)) |value_cursor| { 1822 if (try value_cursor.readInt() == value) { 1823 return; 1824 } 1825 } 1826 1827 try map.put(key, .{ .int = value }); 1828 }, 1829 }, 1830 .bool => { 1831 const bytes = if (value) "true" else "false"; 1832 if (try map.getCursor(key)) |value_cursor| { 1833 var buffer: [5]u8 = undefined; 1834 const existing = try value_cursor.readBytesObject(&buffer); 1835 if (existing.format_tag) |format_tag| { 1836 if (std.mem.eql(u8, &format_tag, "bl") and std.mem.eql(u8, existing.value, bytes)) { 1837 return; 1838 } 1839 } 1840 } 1841 1842 try map.put(key, .{ .bytes_object = .{ .value = bytes, .format_tag = "bl".* } }); 1843 }, 1844 .@"enum" => try upsertBytes(DB, hash_kind, map, key, @tagName(value)), 1845 .@"union" => { 1846 if (!@hasDecl(Field, "encode") or !@hasDecl(Field, "max_encoded_len")) { 1847 @compileError("unsupported upsert field type: " ++ @typeName(Field)); 1848 } 1849 var buffer: [Field.max_encoded_len]u8 = undefined; 1850 try upsertBytes(DB, hash_kind, map, key, try value.encode(&buffer)); 1851 }, 1852 .optional => |optional_info| { 1853 // a missing key is null 1854 if (value) |child| { 1855 if (optional_info.child == []const u8) { 1856 try upsertBytes(DB, hash_kind, map, key, child); 1857 } else switch (@typeInfo(optional_info.child)) { 1858 .array => |array_info| { 1859 if (array_info.child != u8) @compileError("unsupported upsert field type: " ++ @typeName(Field)); 1860 try upsertBytes(DB, hash_kind, map, key, &child); 1861 }, 1862 .@"struct" => try upsert( 1863 optional_info.child, 1864 DB, 1865 hash_kind, 1866 try DB.HashMap(.read_write).init(try map.putCursor(key)), 1867 child, 1868 ), 1869 else => @compileError("unsupported upsert field type: " ++ @typeName(Field)), 1870 } 1871 } else { 1872 _ = try map.remove(key); 1873 } 1874 }, 1875 .@"struct" => try upsert( 1876 Field, 1877 DB, 1878 hash_kind, 1879 try DB.HashMap(.read_write).init(try map.putCursor(key)), 1880 value, 1881 ), 1882 else => @compileError("unsupported upsert field type: " ++ @typeName(Field)), 1883 } 1884 } 1885 1886 fn upsertBytes( 1887 comptime DB: type, 1888 comptime hash_kind: hash.HashKind, 1889 map: DB.HashMap(.read_write), 1890 key: hash.HashInt(hash_kind), 1891 value: []const u8, 1892 ) !void { 1893 var existing_cursor_maybe = try map.getCursor(key); 1894 if (existing_cursor_maybe) |*existing_cursor| { 1895 if (try bytesEqual(DB, existing_cursor, value)) { 1896 return; 1897 } 1898 } 1899 1900 try map.put(key, .{ .bytes = value }); 1901 } 1902 1903 fn bytesEqual( 1904 comptime DB: type, 1905 cursor: *DB.Cursor(.read_only), 1906 value: []const u8, 1907 ) !bool { 1908 var read_buffer: [1024]u8 = undefined; 1909 var reader = try cursor.reader(&read_buffer); 1910 if (reader.size != value.len) { 1911 return false; 1912 } 1913 1914 var chunk_buffer: [1024]u8 = undefined; 1915 var offset: usize = 0; 1916 while (offset < value.len) { 1917 const chunk_len = @min(chunk_buffer.len, value.len - offset); 1918 try reader.interface.readSliceAll(chunk_buffer[0..chunk_len]); 1919 if (!std.mem.eql(u8, chunk_buffer[0..chunk_len], value[offset .. offset + chunk_len])) { 1920 return false; 1921 } 1922 offset += chunk_len; 1923 } 1924 1925 return true; 1926 } 1927 1928 // 1929 // CommitIterator 1930 // 1931 // differs from xit's ObjectIterator in a few ways: 1932 // 1933 // 1. iterates parents first, rather than children first 1934 // 2. `next` only returns the object id, not the full Object 1935 // 3. stores its temporary state in a file on disk to avoid OOMs 1936 // 1937 1938 pub fn CommitIterator(comptime repo_kind: rp.RepoKind, comptime repo_opts: rp.RepoOpts(repo_kind)) type { 1939 return struct { 1940 const DB = EventDB(repo_opts.hash); 1941 const db_name = "haxy-repo-events.db"; 1942 1943 repo_dir: std.Io.Dir, 1944 db_file: std.Io.File, 1945 db: *DB, 1946 // map of each object id to its children 1947 parent_to_children: DB.HashMap(.read_write), 1948 // tracking pending parents is necessary because a child can only 1949 // be returned after all of its unconsumed parents have been returned 1950 child_to_pending_parents: DB.HashMap(.read_write), 1951 // queue of object ids that are ready to be returned from `next` 1952 ready_queue: DB.ArrayList(.read_write), 1953 ready_queue_index: u64, 1954 // the object id at the tip of this branch 1955 newest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 1956 // the object id furthest back in the history that hasn't been consumed 1957 oldest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 1958 1959 pub fn init( 1960 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 1961 io: std.Io, 1962 allocator: std.mem.Allocator, 1963 consumed_object_ids: DB.HashMap(.read_only), 1964 ref: rf.Ref, 1965 ) !CommitIterator(repo_kind, repo_opts) { 1966 const db_file = try state.core.repo_dir.createFile(io, db_name, .{ .truncate = true, .lock = .exclusive, .read = true }); 1967 errdefer { 1968 db_file.close(io); 1969 state.core.repo_dir.deleteFile(io, db_name) catch {}; 1970 } 1971 1972 const buffer_ptr = try allocator.create(std.Io.Writer.Allocating); 1973 errdefer allocator.destroy(buffer_ptr); 1974 1975 buffer_ptr.* = std.Io.Writer.Allocating.init(allocator); 1976 errdefer buffer_ptr.deinit(); 1977 1978 const db_ptr = try allocator.create(DB); 1979 errdefer allocator.destroy(db_ptr); 1980 // the db is scratch state that gets deleted after iteration, so 1981 // there is nothing worth fsyncing 1982 db_ptr.* = try DB.init(.{ .io = io, .file = db_file, .buffer = buffer_ptr, .fsync = false }); 1983 1984 const map = try DB.HashMap(.read_write).init(db_ptr.rootCursor()); 1985 1986 const parent_to_children_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "parent->children")); 1987 const parent_to_children = try DB.HashMap(.read_write).init(parent_to_children_cursor); 1988 1989 const child_to_pending_parents_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "child->pending-parents")); 1990 const child_to_pending_parents = try DB.HashMap(.read_write).init(child_to_pending_parents_cursor); 1991 1992 const ready_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "ready-queue")); 1993 const ready_queue = try DB.ArrayList(.read_write).init(ready_queue_cursor); 1994 1995 var self = CommitIterator(repo_kind, repo_opts){ 1996 .repo_dir = state.core.repo_dir, 1997 .db_file = db_file, 1998 .db = db_ptr, 1999 .parent_to_children = parent_to_children, 2000 .child_to_pending_parents = child_to_pending_parents, 2001 .ready_queue = ready_queue, 2002 .ready_queue_index = 0, 2003 .newest_object_id = null, 2004 .oldest_object_id = null, 2005 }; 2006 errdefer self.deinit(io, allocator); 2007 2008 try self.collect(state, io, allocator, consumed_object_ids, ref); 2009 2010 return self; 2011 } 2012 2013 pub fn deinit(self: *CommitIterator(repo_kind, repo_opts), io: std.Io, allocator: std.mem.Allocator) void { 2014 self.db_file.close(io); 2015 self.db.core.memory.buffer.deinit(); 2016 allocator.destroy(self.db.core.memory.buffer); 2017 self.repo_dir.deleteFile(io, db_name) catch {}; 2018 allocator.destroy(self.db); 2019 } 2020 2021 pub fn next(self: *CommitIterator(repo_kind, repo_opts)) !?[hash.byteLen(repo_opts.hash)]u8 { 2022 if (self.ready_queue_index >= try self.ready_queue.count()) return null; 2023 2024 const oid = try readOidFromList(self.ready_queue.readOnly(), self.ready_queue_index); 2025 self.ready_queue_index += 1; 2026 2027 try self.enqueueChildren(&oid); 2028 return oid; 2029 } 2030 2031 fn collect( 2032 self: *CommitIterator(repo_kind, repo_opts), 2033 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 2034 io: std.Io, 2035 allocator: std.mem.Allocator, 2036 consumed_object_ids: DB.HashMap(.read_only), 2037 ref: rf.Ref, 2038 ) !void { 2039 const map = try DB.HashMap(.read_write).init(self.db.rootCursor()); 2040 2041 const walk_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "walk-queue")); 2042 const walk_queue = try DB.ArrayList(.read_write).init(walk_queue_cursor); 2043 var walk_queue_index: u64 = 0; 2044 2045 const seen_object_ids_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "seen-object-ids")); 2046 const seen_object_ids = try DB.HashSet(.read_write).init(seen_object_ids_cursor); 2047 2048 const head_oid_hex = (try rf.readRecur(repo_kind, repo_opts, state, io, .{ .ref = ref })) orelse return error.OidNotFound; 2049 var head_oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2050 _ = try std.fmt.hexToBytes(&head_oid, &head_oid_hex); 2051 try walk_queue.append(.{ .bytes = &head_oid }); 2052 2053 while (walk_queue_index < try walk_queue.count()) { 2054 const oid = try readOidFromList(walk_queue.readOnly(), walk_queue_index); 2055 const oid_int = hash.bytesToInt(repo_opts.hash, &oid); 2056 2057 walk_queue_index += 1; 2058 2059 // if we've seen this object id, skip it 2060 if (null != try seen_object_ids.getCursor(oid_int)) { 2061 continue; 2062 } 2063 2064 try seen_object_ids.put(oid_int, .{ .bytes = &oid }); 2065 2066 var commit_object = try obj.Object(repo_kind, repo_opts).init(state, io, allocator, &std.fmt.bytesToHex(oid, .lower)); 2067 defer commit_object.deinit(); 2068 2069 // if this object id has already been consumed, skip it 2070 if (null != try consumed_object_ids.getCursor(oid_int)) { 2071 continue; 2072 } 2073 2074 if (self.newest_object_id == null) self.newest_object_id = oid; 2075 self.oldest_object_id = oid; 2076 2077 const parent_oids = commit_object.content.commit.metadata.parent_oids orelse return error.ParentOidsNotFound; 2078 var pending_parent_count: u64 = 0; 2079 for (parent_oids) |*parent_oid| { 2080 var parent_oid_bytes: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2081 _ = try std.fmt.hexToBytes(&parent_oid_bytes, parent_oid); 2082 const parent_oid_int = hash.bytesToInt(repo_opts.hash, &parent_oid_bytes); 2083 2084 // if this object id has already been consumed, skip it 2085 if (null != try consumed_object_ids.getCursor(parent_oid_int)) { 2086 continue; 2087 } 2088 2089 // if this object id hasn't already been seen, add it to the walk queue 2090 if (null == try seen_object_ids.getCursor(parent_oid_int)) { 2091 try walk_queue.append(.{ .bytes = &parent_oid_bytes }); 2092 } 2093 2094 const children_cursor = try self.parent_to_children.putCursor(parent_oid_int); 2095 const children = try DB.HashSet(.read_write).init(children_cursor); 2096 try children.put(oid_int, .{ .bytes = &oid }); 2097 pending_parent_count += 1; 2098 } 2099 2100 if (pending_parent_count == 0) { 2101 try self.ready_queue.append(.{ .bytes = &oid }); 2102 } else { 2103 try self.child_to_pending_parents.put(oid_int, .{ .uint = pending_parent_count }); 2104 } 2105 } 2106 } 2107 2108 fn enqueueChildren(self: *CommitIterator(repo_kind, repo_opts), oid: *const [hash.byteLen(repo_opts.hash)]u8) !void { 2109 const oid_int = hash.bytesToInt(repo_opts.hash, oid); 2110 const children_cursor = try self.parent_to_children.getCursor(oid_int) orelse return; 2111 const children = try DB.HashSet(.read_only).init(children_cursor); 2112 var children_iter = try children.iterator(); 2113 2114 while (try children_iter.next()) |*child_cursor| { 2115 const kv_pair = try child_cursor.readKeyValuePair(); 2116 const child_oid_int = kv_pair.hash; 2117 const pending_parent_cursor = try self.child_to_pending_parents.getCursor(child_oid_int) orelse return error.CursorNotFound; 2118 const pending_parent_count = try pending_parent_cursor.readUint(); 2119 2120 if (pending_parent_count <= 1) { 2121 _ = try self.child_to_pending_parents.remove(child_oid_int); 2122 const child_oid = hash.intToBytes(hash.HashInt(repo_opts.hash), child_oid_int); 2123 try self.ready_queue.append(.{ .bytes = &child_oid }); 2124 } else { 2125 try self.child_to_pending_parents.put(child_oid_int, .{ .uint = pending_parent_count - 1 }); 2126 } 2127 } 2128 } 2129 2130 fn readOidFromList(list: DB.ArrayList(.read_only), index: u64) ![hash.byteLen(repo_opts.hash)]u8 { 2131 const oid_cursor = try list.getCursor(index) orelse return error.CursorNotFound; 2132 var oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2133 _ = try oid_cursor.readBytes(&oid); 2134 return oid; 2135 } 2136 }; 2137 } 2138