╭─────────────╮ │scroll to top│ ╰─────────────╯ ╭ event.zig ────────────────────────────────────────────────────╮ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ ╰───────────────────────────────────────────────────────────────╯
╭────╮ ╭ clone ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╮ │http│ ssh │ │ ╰────╯ ╰ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╯
1357 var existing_cursor_maybe = try map.getCursor(key); 1358 if (existing_cursor_maybe) |*existing_cursor| { 1359 if (try bytesEqual(DB, existing_cursor, value)) { 1360 return; 1361 } 1362 } 1363 1364 var value_cursor = try map.putCursor(key); 1365 var write_buffer: [1024]u8 = undefined; 1366 var writer = try value_cursor.writer(&write_buffer); 1367 try writer.interface.writeAll(value); 1368 try writer.finish(); 1369 } 1370 1371 fn bytesEqual( 1372 comptime DB: type, 1373 cursor: *DB.Cursor(.read_only), 1374 value: []const u8, 1375 ) !bool { 1376 var read_buffer: [1024]u8 = undefined; 1377 var reader = try cursor.reader(&read_buffer); 1378 if (reader.size != value.len) { 1379 return false; 1380 } 1381 1382 var chunk_buffer: [1024]u8 = undefined; 1383 var offset: usize = 0; 1384 while (offset < value.len) { 1385 const chunk_len = @min(chunk_buffer.len, value.len - offset); 1386 try reader.interface.readSliceAll(chunk_buffer[0..chunk_len]); 1387 if (!std.mem.eql(u8, chunk_buffer[0..chunk_len], value[offset .. offset + chunk_len])) { 1388 return false; 1389 } 1390 offset += chunk_len; 1391 } 1392 1393 return true; 1394 } 1395 1396 // 1397 // CommitIterator 1398 // 1399 // differs from xit's ObjectIterator in a few ways: 1400 // 1401 // 1. iterates parents first, rather than children first 1402 // 2. `next` only returns the object id, not the full Object 1403 // 3. stores its temporary state in a file on disk to avoid OOMs 1404 // 1405 1406 pub fn CommitIterator(comptime repo_kind: rp.RepoKind, comptime repo_opts: rp.RepoOpts(repo_kind)) type { 1407 return struct { 1408 const DB = EventDB(repo_opts.hash); 1409 const db_name = "haxy-repo-events.db"; 1410 1411 repo_dir: std.Io.Dir, 1412 db_file: std.Io.File, 1413 db: *DB, 1414 // map of each object id to its children 1415 parent_to_children: DB.HashMap(.read_write), 1416 // tracking pending parents is necessary because a child can only 1417 // be returned after all of its unconsumed parents have been returned 1418 child_to_pending_parents: DB.HashMap(.read_write), 1419 // queue of object ids that are ready to be returned from `next` 1420 ready_queue: DB.ArrayList(.read_write), 1421 ready_queue_index: u64, 1422 // the object id at the tip of this branch 1423 newest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 1424 // the object id furthest back in the history that hasn't been consumed 1425 oldest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 1426 1427 pub fn init( 1428 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 1429 io: std.Io, 1430 allocator: std.mem.Allocator, 1431 consumed_object_ids: DB.HashMap(.read_only), 1432 ref: rf.Ref, 1433 ) !CommitIterator(repo_kind, repo_opts) { 1434 const db_file = try state.core.repo_dir.createFile(io, db_name, .{ .truncate = true, .lock = .exclusive, .read = true }); 1435 errdefer { 1436 db_file.close(io); 1437 state.core.repo_dir.deleteFile(io, db_name) catch {}; 1438 } 1439 1440 const buffer_ptr = try allocator.create(std.Io.Writer.Allocating); 1441 errdefer allocator.destroy(buffer_ptr); 1442 1443 buffer_ptr.* = std.Io.Writer.Allocating.init(allocator); 1444 errdefer buffer_ptr.deinit(); 1445 1446 const db_ptr = try allocator.create(DB); 1447 errdefer allocator.destroy(db_ptr); 1448 // the db is scratch state that gets deleted after iteration, so 1449 // there is nothing worth fsyncing 1450 db_ptr.* = try DB.init(.{ .io = io, .file = db_file, .buffer = buffer_ptr, .fsync = false }); 1451 1452 const map = try DB.HashMap(.read_write).init(db_ptr.rootCursor()); 1453 1454 const parent_to_children_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "parent->children")); 1455 const parent_to_children = try DB.HashMap(.read_write).init(parent_to_children_cursor); 1456 1457 const child_to_pending_parents_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "child->pending-parents")); 1458 const child_to_pending_parents = try DB.HashMap(.read_write).init(child_to_pending_parents_cursor); 1459 1460 const ready_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "ready-queue")); 1461 const ready_queue = try DB.ArrayList(.read_write).init(ready_queue_cursor); 1462 1463 var self = CommitIterator(repo_kind, repo_opts){ 1464 .repo_dir = state.core.repo_dir, 1465 .db_file = db_file, 1466 .db = db_ptr, 1467 .parent_to_children = parent_to_children, 1468 .child_to_pending_parents = child_to_pending_parents, 1469 .ready_queue = ready_queue, 1470 .ready_queue_index = 0, 1471 .newest_object_id = null, 1472 .oldest_object_id = null, 1473 }; 1474 errdefer self.deinit(io, allocator); 1475 1476 try self.collect(state, io, allocator, consumed_object_ids, ref); 1477 1478 return self; 1479 } 1480 1481 pub fn deinit(self: *CommitIterator(repo_kind, repo_opts), io: std.Io, allocator: std.mem.Allocator) void { 1482 self.db_file.close(io); 1483 self.db.core.memory.buffer.deinit(); 1484 allocator.destroy(self.db.core.memory.buffer); 1485 self.repo_dir.deleteFile(io, db_name) catch {}; 1486 allocator.destroy(self.db); 1487 } 1488 1489 pub fn next(self: *CommitIterator(repo_kind, repo_opts)) !?[hash.byteLen(repo_opts.hash)]u8 { 1490 if (self.ready_queue_index >= try self.ready_queue.count()) return null; 1491 1492 const oid = try readOidFromList(self.ready_queue.readOnly(), self.ready_queue_index); 1493 self.ready_queue_index += 1; 1494 1495 try self.enqueueChildren(&oid); 1496 return oid; 1497 } 1498 1499 fn collect( 1500 self: *CommitIterator(repo_kind, repo_opts), 1501 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 1502 io: std.Io, 1503 allocator: std.mem.Allocator, 1504 consumed_object_ids: DB.HashMap(.read_only), 1505 ref: rf.Ref, 1506 ) !void { 1507 const map = try DB.HashMap(.read_write).init(self.db.rootCursor()); 1508 1509 const walk_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "walk-queue")); 1510 const walk_queue = try DB.ArrayList(.read_write).init(walk_queue_cursor); 1511 var walk_queue_index: u64 = 0; 1512 1513 const seen_object_ids_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "seen-object-ids")); 1514 const seen_object_ids = try DB.HashSet(.read_write).init(seen_object_ids_cursor); 1515 1516 const head_oid_hex = (try rf.readRecur(repo_kind, repo_opts, state, io, .{ .ref = ref })) orelse return error.OidNotFound; 1517 var head_oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 1518 _ = try std.fmt.hexToBytes(&head_oid, &head_oid_hex); 1519 try walk_queue.append(.{ .bytes = &head_oid }); 1520 1521 while (walk_queue_index < try walk_queue.count()) { 1522 const oid = try readOidFromList(walk_queue.readOnly(), walk_queue_index); 1523 const oid_int = hash.bytesToInt(repo_opts.hash, &oid); 1524 1525 walk_queue_index += 1; 1526 1527 // if we've seen this object id, skip it 1528 if (null != try seen_object_ids.getCursor(oid_int)) { 1529 continue; 1530 } 1531 1532 try seen_object_ids.put(oid_int, .{ .bytes = &oid }); 1533 1534 var commit_object = try obj.Object(repo_kind, repo_opts).init(state, io, allocator, &std.fmt.bytesToHex(oid, .lower)); 1535 defer commit_object.deinit(); 1536 1537 // if this object id has already been consumed, skip it 1538 if (null != try consumed_object_ids.getCursor(oid_int)) { 1539 continue; 1540 } 1541 1542 if (self.newest_object_id == null) self.newest_object_id = oid; 1543 self.oldest_object_id = oid; 1544 1545 const parent_oids = commit_object.content.commit.metadata.parent_oids orelse return error.ParentOidsNotFound; 1546 var pending_parent_count: u64 = 0; 1547 for (parent_oids) |*parent_oid| { 1548 var parent_oid_bytes: [hash.byteLen(repo_opts.hash)]u8 = undefined; 1549 _ = try std.fmt.hexToBytes(&parent_oid_bytes, parent_oid); 1550 const parent_oid_int = hash.bytesToInt(repo_opts.hash, &parent_oid_bytes); 1551 1552 // if this object id has already been consumed, skip it 1553 if (null != try consumed_object_ids.getCursor(parent_oid_int)) { 1554 continue; 1555 } 1556 1557 // if this object id hasn't already been seen, add it to the walk queue 1558 if (null == try seen_object_ids.getCursor(parent_oid_int)) { 1559 try walk_queue.append(.{ .bytes = &parent_oid_bytes }); 1560 } 1561 1562 const children_cursor = try self.parent_to_children.putCursor(parent_oid_int); 1563 const children = try DB.HashSet(.read_write).init(children_cursor); 1564 try children.put(oid_int, .{ .bytes = &oid }); 1565 pending_parent_count += 1; 1566 } 1567 1568 if (pending_parent_count == 0) { 1569 try self.ready_queue.append(.{ .bytes = &oid }); 1570 } else { 1571 try self.child_to_pending_parents.put(oid_int, .{ .uint = pending_parent_count }); 1572 } 1573 } 1574 } 1575 1576 fn enqueueChildren(self: *CommitIterator(repo_kind, repo_opts), oid: *const [hash.byteLen(repo_opts.hash)]u8) !void { 1577 const oid_int = hash.bytesToInt(repo_opts.hash, oid); 1578 const children_cursor = try self.parent_to_children.getCursor(oid_int) orelse return; 1579 const children = try DB.HashSet(.read_only).init(children_cursor); 1580 var children_iter = try children.iterator(); 1581 1582 while (try children_iter.next()) |*child_cursor| { 1583 const kv_pair = try child_cursor.readKeyValuePair(); 1584 const child_oid_int = kv_pair.hash; 1585 const pending_parent_cursor = try self.child_to_pending_parents.getCursor(child_oid_int) orelse return error.CursorNotFound; 1586 const pending_parent_count = try pending_parent_cursor.readUint(); 1587 1588 if (pending_parent_count <= 1) { 1589 _ = try self.child_to_pending_parents.remove(child_oid_int); 1590 const child_oid = hash.intToBytes(hash.HashInt(repo_opts.hash), child_oid_int); 1591 try self.ready_queue.append(.{ .bytes = &child_oid }); 1592 } else { 1593 try self.child_to_pending_parents.put(child_oid_int, .{ .uint = pending_parent_count - 1 }); 1594 } 1595 } 1596 } 1597 1598 fn readOidFromList(list: DB.ArrayList(.read_only), index: u64) ![hash.byteLen(repo_opts.hash)]u8 { 1599 const oid_cursor = try list.getCursor(index) orelse return error.CursorNotFound; 1600 var oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 1601 _ = try oid_cursor.readBytes(&oid); 1602 return oid; 1603 } 1604 }; 1605 } 1606