╭─────────────╮ │scroll to top│ ╰─────────────╯ ╭ event.zig ────────────────────────────────────────────────────╮ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ ╰───────────────────────────────────────────────────────────────╯
╭────╮ ╭ clone ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╮ │http│ ssh │ │ ╰────╯ ╰ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╯
2139 .array => |array_info| { 2140 if (array_info.child == u8) { 2141 return std.mem.eql(u8, &a, &b); 2142 } else { 2143 @compileError("unsupported field type: " ++ @typeName(Field)); 2144 } 2145 }, 2146 .void => return true, 2147 .bool, .int, .@"enum" => return a == b, 2148 .@"union" => { 2149 if (std.meta.activeTag(a) != std.meta.activeTag(b)) return false; 2150 return switch (a) { 2151 inline else => |value, tag| fieldEqual(@TypeOf(value), value, @field(b, @tagName(tag))), 2152 }; 2153 }, 2154 .@"struct" => return fieldsEqual(Field, a, b), 2155 .optional => |optional_info| { 2156 if (a) |a_value| { 2157 if (b) |b_value| return fieldEqual(optional_info.child, a_value, b_value); 2158 return false; 2159 } 2160 return b == null; 2161 }, 2162 else => @compileError("unsupported field type: " ++ @typeName(Field)), 2163 } 2164 } 2165 2166 // 2167 // writing to xitdb 2168 // 2169 2170 pub fn upsert( 2171 comptime T: type, 2172 comptime DB: type, 2173 comptime hash_kind: hash.HashKind, 2174 map: DB.HashMap(.read_write), 2175 event: T, 2176 ) !void { 2177 switch (@typeInfo(T)) { 2178 .@"struct" => |struct_info| { 2179 inline for (struct_info.fields) |field| { 2180 try upsertField(DB, hash_kind, map, field.name, field.type, @field(event, field.name)); 2181 } 2182 }, 2183 else => @compileError("upsert expects a struct"), 2184 } 2185 } 2186 2187 fn upsertField( 2188 comptime DB: type, 2189 comptime hash_kind: hash.HashKind, 2190 map: DB.HashMap(.read_write), 2191 comptime field_name: []const u8, 2192 comptime Field: type, 2193 value: Field, 2194 ) !void { 2195 const key = hash.hashInt(hash_kind, field_name); 2196 2197 switch (@typeInfo(Field)) { 2198 .pointer => |pointer_info| { 2199 if (pointer_info.size == .slice and pointer_info.child == u8) { 2200 try upsertBytes(DB, hash_kind, map, key, value); 2201 } else { 2202 @compileError("unsupported upsert field type: " ++ @typeName(Field)); 2203 } 2204 }, 2205 .array => |array_info| { 2206 if (array_info.child == u8) { 2207 try upsertBytes(DB, hash_kind, map, key, &value); 2208 } else { 2209 @compileError("unsupported upsert field type: " ++ @typeName(Field)); 2210 } 2211 }, 2212 .int => |int_info| switch (int_info.signedness) { 2213 .unsigned => { 2214 if (try map.getCursor(key)) |value_cursor| { 2215 if (try value_cursor.readUint() == value) { 2216 return; 2217 } 2218 } 2219 2220 try map.put(key, .{ .uint = value }); 2221 }, 2222 .signed => { 2223 if (try map.getCursor(key)) |value_cursor| { 2224 if (try value_cursor.readInt() == value) { 2225 return; 2226 } 2227 } 2228 2229 try map.put(key, .{ .int = value }); 2230 }, 2231 }, 2232 .bool => { 2233 const bytes = if (value) "true" else "false"; 2234 if (try map.getCursor(key)) |value_cursor| { 2235 var buffer: [5]u8 = undefined; 2236 const existing = try value_cursor.readBytesObject(&buffer); 2237 if (existing.format_tag) |format_tag| { 2238 if (std.mem.eql(u8, &format_tag, "bl") and std.mem.eql(u8, existing.value, bytes)) { 2239 return; 2240 } 2241 } 2242 } 2243 2244 try map.put(key, .{ .bytes_object = .{ .value = bytes, .format_tag = "bl".* } }); 2245 }, 2246 .@"enum" => try upsertBytes(DB, hash_kind, map, key, @tagName(value)), 2247 .@"union" => switch (value) { 2248 inline else => |payload, tag| { 2249 if (@TypeOf(payload) == void) { 2250 try upsertBytes(DB, hash_kind, map, key, @tagName(tag)); 2251 } else { 2252 const tag_key = hash.hashInt(hash_kind, @tagName(tag)); 2253 if (try map.getCursor(key)) |cursor| { 2254 switch (cursor.slot().tag) { 2255 .hash_map => { 2256 const variants = try DB.HashMap(.read_only).init(cursor); 2257 if (try variants.getCursor(tag_key) == null) try map.put(key, .{ .slot = null }); 2258 }, 2259 else => try map.put(key, .{ .slot = null }), 2260 } 2261 } 2262 const variants = try DB.HashMap(.read_write).init(try map.putCursor(key)); 2263 if (@typeInfo(@TypeOf(payload)) == .optional) { 2264 // keep the variant even when its payload is null 2265 if (payload == null) return try variants.put(tag_key, .{ .slot = null }); 2266 } 2267 if (try variants.getCursor(tag_key)) |cursor| { 2268 if (cursor.slot().tag == .none) _ = try variants.remove(tag_key); 2269 } 2270 try upsertField(DB, hash_kind, variants, @tagName(tag), @TypeOf(payload), payload); 2271 } 2272 }, 2273 }, 2274 .optional => |optional_info| { 2275 // a missing key is null 2276 if (value) |child| { 2277 try upsertField(DB, hash_kind, map, field_name, optional_info.child, child); 2278 } else { 2279 _ = try map.remove(key); 2280 } 2281 }, 2282 .@"struct" => try upsert( 2283 Field, 2284 DB, 2285 hash_kind, 2286 try DB.HashMap(.read_write).init(try map.putCursor(key)), 2287 value, 2288 ), 2289 else => @compileError("unsupported upsert field type: " ++ @typeName(Field)), 2290 } 2291 } 2292 2293 fn upsertBytes( 2294 comptime DB: type, 2295 comptime hash_kind: hash.HashKind, 2296 map: DB.HashMap(.read_write), 2297 key: hash.HashInt(hash_kind), 2298 value: []const u8, 2299 ) !void { 2300 var existing_cursor_maybe = try map.getCursor(key); 2301 if (existing_cursor_maybe) |*existing_cursor| { 2302 if (try bytesEqual(DB, existing_cursor, value)) { 2303 return; 2304 } 2305 } 2306 2307 try map.put(key, .{ .bytes = value }); 2308 } 2309 2310 fn bytesEqual( 2311 comptime DB: type, 2312 cursor: *DB.Cursor(.read_only), 2313 value: []const u8, 2314 ) !bool { 2315 switch (cursor.slot().tag) { 2316 .bytes, .short_bytes => {}, 2317 else => return false, 2318 } 2319 var read_buffer: [1024]u8 = undefined; 2320 var reader = try cursor.reader(&read_buffer); 2321 if (reader.size != value.len) { 2322 return false; 2323 } 2324 2325 var chunk_buffer: [1024]u8 = undefined; 2326 var offset: usize = 0; 2327 while (offset < value.len) { 2328 const chunk_len = @min(chunk_buffer.len, value.len - offset); 2329 try reader.interface.readSliceAll(chunk_buffer[0..chunk_len]); 2330 if (!std.mem.eql(u8, chunk_buffer[0..chunk_len], value[offset .. offset + chunk_len])) { 2331 return false; 2332 } 2333 offset += chunk_len; 2334 } 2335 2336 return true; 2337 } 2338 2339 // 2340 // CommitIterator 2341 // 2342 // differs from xit's ObjectIterator in a few ways: 2343 // 2344 // 1. iterates parents first, rather than children first 2345 // 2. `next` only returns the object id, not the full Object 2346 // 3. stores its temporary state in a file on disk to avoid OOMs 2347 // 2348 2349 pub fn CommitIterator(comptime repo_kind: rp.RepoKind, comptime repo_opts: rp.RepoOpts(repo_kind)) type { 2350 return struct { 2351 const DB = EventDB(repo_opts.hash); 2352 // a plain file is used for the scratch db because each write outside 2353 // a transaction is synced anyway, so buffering would gain nothing 2354 const TempDB = xit.xitdb.Database(.file, hash.HashInt(repo_opts.hash)); 2355 const db_name = "haxy-repo-events.db"; 2356 2357 repo_dir: std.Io.Dir, 2358 db_file: std.Io.File, 2359 db: *TempDB, 2360 // map of each object id to its children 2361 parent_to_children: TempDB.HashMap(.read_write), 2362 // tracking pending parents is necessary because a child can only 2363 // be returned after all of its unconsumed parents have been returned 2364 child_to_pending_parents: TempDB.HashMap(.read_write), 2365 // queue of object ids that are ready to be returned from `next` 2366 ready_queue: TempDB.ArrayList(.read_write), 2367 ready_queue_index: u64, 2368 // the object id at the tip of this branch 2369 newest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 2370 // the object id furthest back in the history that hasn't been consumed 2371 oldest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 2372 2373 pub fn init( 2374 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 2375 io: std.Io, 2376 allocator: std.mem.Allocator, 2377 consumed_object_ids: DB.HashMap(.read_only), 2378 ref: rf.Ref, 2379 ) !CommitIterator(repo_kind, repo_opts) { 2380 const db_file = try state.core.repo_dir.createFile(io, db_name, .{ .truncate = true, .lock = .exclusive, .read = true }); 2381 errdefer { 2382 db_file.close(io); 2383 state.core.repo_dir.deleteFile(io, db_name) catch {}; 2384 } 2385 2386 // cursors point at the db, so it needs a stable address 2387 const db_ptr = try allocator.create(TempDB); 2388 errdefer allocator.destroy(db_ptr); 2389 // the db is scratch state that gets deleted after iteration, so 2390 // there is nothing worth fsyncing 2391 db_ptr.* = try TempDB.init(.{ .io = io, .file = db_file, .fsync = false }); 2392 2393 const map = try TempDB.HashMap(.read_write).init(db_ptr.rootCursor()); 2394 2395 const parent_to_children_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "parent->children")); 2396 const parent_to_children = try TempDB.HashMap(.read_write).init(parent_to_children_cursor); 2397 2398 const child_to_pending_parents_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "child->pending-parents")); 2399 const child_to_pending_parents = try TempDB.HashMap(.read_write).init(child_to_pending_parents_cursor); 2400 2401 const ready_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "ready-queue")); 2402 const ready_queue = try TempDB.ArrayList(.read_write).init(ready_queue_cursor); 2403 2404 var self = CommitIterator(repo_kind, repo_opts){ 2405 .repo_dir = state.core.repo_dir, 2406 .db_file = db_file, 2407 .db = db_ptr, 2408 .parent_to_children = parent_to_children, 2409 .child_to_pending_parents = child_to_pending_parents, 2410 .ready_queue = ready_queue, 2411 .ready_queue_index = 0, 2412 .newest_object_id = null, 2413 .oldest_object_id = null, 2414 }; 2415 errdefer self.deinit(io, allocator); 2416 2417 try self.collect(state, io, allocator, consumed_object_ids, ref); 2418 2419 return self; 2420 } 2421 2422 pub fn deinit(self: *CommitIterator(repo_kind, repo_opts), io: std.Io, allocator: std.mem.Allocator) void { 2423 self.db_file.close(io); 2424 self.repo_dir.deleteFile(io, db_name) catch {}; 2425 allocator.destroy(self.db); 2426 } 2427 2428 pub fn next(self: *CommitIterator(repo_kind, repo_opts)) !?[hash.byteLen(repo_opts.hash)]u8 { 2429 if (self.ready_queue_index >= try self.ready_queue.count()) return null; 2430 2431 const oid = try readOidFromList(self.ready_queue.readOnly(), self.ready_queue_index); 2432 self.ready_queue_index += 1; 2433 2434 try self.enqueueChildren(&oid); 2435 return oid; 2436 } 2437 2438 fn collect( 2439 self: *CommitIterator(repo_kind, repo_opts), 2440 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 2441 io: std.Io, 2442 allocator: std.mem.Allocator, 2443 consumed_object_ids: DB.HashMap(.read_only), 2444 ref: rf.Ref, 2445 ) !void { 2446 const map = try TempDB.HashMap(.read_write).init(self.db.rootCursor()); 2447 2448 const walk_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "walk-queue")); 2449 const walk_queue = try TempDB.ArrayList(.read_write).init(walk_queue_cursor); 2450 var walk_queue_index: u64 = 0; 2451 2452 const seen_object_ids_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "seen-object-ids")); 2453 const seen_object_ids = try TempDB.HashSet(.read_write).init(seen_object_ids_cursor); 2454 2455 const head_oid_hex = (try rf.readRecur(repo_kind, repo_opts, state, io, .{ .ref = ref })) orelse return error.OidNotFound; 2456 var head_oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2457 _ = try std.fmt.hexToBytes(&head_oid, &head_oid_hex); 2458 try walk_queue.append(.{ .bytes = &head_oid }); 2459 2460 while (walk_queue_index < try walk_queue.count()) { 2461 const oid = try readOidFromList(walk_queue.readOnly(), walk_queue_index); 2462 const oid_int = hash.bytesToInt(repo_opts.hash, &oid); 2463 2464 walk_queue_index += 1; 2465 2466 // if we've seen this object id, skip it 2467 if (null != try seen_object_ids.getCursor(oid_int)) { 2468 continue; 2469 } 2470 2471 try seen_object_ids.put(oid_int, .{ .bytes = &oid }); 2472 2473 var commit_object = try obj.Object(repo_kind, repo_opts).init(state, io, allocator, &std.fmt.bytesToHex(oid, .lower)); 2474 defer commit_object.deinit(); 2475 2476 // if this object id has already been consumed, skip it 2477 if (null != try consumed_object_ids.getCursor(oid_int)) { 2478 continue; 2479 } 2480 2481 if (self.newest_object_id == null) self.newest_object_id = oid; 2482 self.oldest_object_id = oid; 2483 2484 const parent_oids = commit_object.content.commit.metadata.parent_oids orelse return error.ParentOidsNotFound; 2485 var pending_parent_count: u64 = 0; 2486 for (parent_oids) |*parent_oid| { 2487 var parent_oid_bytes: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2488 _ = try std.fmt.hexToBytes(&parent_oid_bytes, parent_oid); 2489 const parent_oid_int = hash.bytesToInt(repo_opts.hash, &parent_oid_bytes); 2490 2491 // if this object id has already been consumed, skip it 2492 if (null != try consumed_object_ids.getCursor(parent_oid_int)) { 2493 continue; 2494 } 2495 2496 // if this object id hasn't already been seen, add it to the walk queue 2497 if (null == try seen_object_ids.getCursor(parent_oid_int)) { 2498 try walk_queue.append(.{ .bytes = &parent_oid_bytes }); 2499 } 2500 2501 const children_cursor = try self.parent_to_children.putCursor(parent_oid_int); 2502 const children = try TempDB.HashSet(.read_write).init(children_cursor); 2503 try children.put(oid_int, .{ .bytes = &oid }); 2504 pending_parent_count += 1; 2505 } 2506 2507 if (pending_parent_count == 0) { 2508 try self.ready_queue.append(.{ .bytes = &oid }); 2509 } else { 2510 try self.child_to_pending_parents.put(oid_int, .{ .uint = pending_parent_count }); 2511 } 2512 } 2513 } 2514 2515 fn enqueueChildren(self: *CommitIterator(repo_kind, repo_opts), oid: *const [hash.byteLen(repo_opts.hash)]u8) !void { 2516 const oid_int = hash.bytesToInt(repo_opts.hash, oid); 2517 const children_cursor = try self.parent_to_children.getCursor(oid_int) orelse return; 2518 const children = try TempDB.HashSet(.read_only).init(children_cursor); 2519 var children_iter = try children.iterator(); 2520 2521 while (try children_iter.next()) |*child_cursor| { 2522 const kv_pair = try child_cursor.readKeyValuePair(); 2523 const child_oid_int = kv_pair.hash; 2524 const pending_parent_cursor = try self.child_to_pending_parents.getCursor(child_oid_int) orelse return error.CursorNotFound; 2525 const pending_parent_count = try pending_parent_cursor.readUint(); 2526 2527 if (pending_parent_count <= 1) { 2528 _ = try self.child_to_pending_parents.remove(child_oid_int); 2529 const child_oid = hash.intToBytes(hash.HashInt(repo_opts.hash), child_oid_int); 2530 try self.ready_queue.append(.{ .bytes = &child_oid }); 2531 } else { 2532 try self.child_to_pending_parents.put(child_oid_int, .{ .uint = pending_parent_count - 1 }); 2533 } 2534 } 2535 } 2536 2537 fn readOidFromList(list: TempDB.ArrayList(.read_only), index: u64) ![hash.byteLen(repo_opts.hash)]u8 { 2538 const oid_cursor = try list.getCursor(index) orelse return error.CursorNotFound; 2539 var oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2540 _ = try oid_cursor.readBytes(&oid); 2541 return oid; 2542 } 2543 }; 2544 } 2545