╭─────────────╮ │scroll to top│ ╰─────────────╯ ╭ event.zig ────────────────────────────────────────────────────╮ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ │ ╰───────────────────────────────────────────────────────────────╯
🬐 🬠 🬞🬂🬏 🬟 🬑 🬟 🬑 ╭a40064bb..╮ a40064bb.. see-em: 🬋🬋🬋 🬋🬋🬋 🬋 🬋 │files │ commits refs discuss issues new login 🬐 🬠 🬐 🬠 🬑 🬟 🬰 ╰──────────╯ (599) user
╭────╮ ╭ clone ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╮ │http│ ssh │ │ ╰────╯ ╰ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ╯
.. embed/ event/ test/ ui/ command.zig diff3.zig ╭─────────────────────────────────╮ │event.zig │ ╰─────────────────────────────────╯ find.zig fork.zig lib.zig main.zig main_wasm.zig markdown.zig patch.zig progress.zig push.zig search.zig search_commit.zig search_thread.zig serve.zig serve_common.zig serve_http.zig serve_ssh.zig serve_ssh_protocol.zig test.zig testnet.zig try.zig ui.zig web.zig
2193 comptime hash_kind: hash.HashKind, 2194 map: DB.HashMap(.read_write), 2195 key: hash.HashInt(hash_kind), 2196 value: []const u8, 2197 ) !void { 2198 var existing_cursor_maybe = try map.getCursor(key); 2199 if (existing_cursor_maybe) |*existing_cursor| { 2200 if (try bytesEqual(DB, existing_cursor, value)) { 2201 return; 2202 } 2203 } 2204 2205 try map.put(key, .{ .bytes = value }); 2206 } 2207 2208 fn bytesEqual( 2209 comptime DB: type, 2210 cursor: *DB.Cursor(.read_only), 2211 value: []const u8, 2212 ) !bool { 2213 switch (cursor.slot().tag) { 2214 .bytes, .short_bytes => {}, 2215 else => return false, 2216 } 2217 var read_buffer: [1024]u8 = undefined; 2218 var reader = try cursor.reader(&read_buffer); 2219 if (reader.size != value.len) { 2220 return false; 2221 } 2222 2223 var chunk_buffer: [1024]u8 = undefined; 2224 var offset: usize = 0; 2225 while (offset < value.len) { 2226 const chunk_len = @min(chunk_buffer.len, value.len - offset); 2227 try reader.interface.readSliceAll(chunk_buffer[0..chunk_len]); 2228 if (!std.mem.eql(u8, chunk_buffer[0..chunk_len], value[offset .. offset + chunk_len])) { 2229 return false; 2230 } 2231 offset += chunk_len; 2232 } 2233 2234 return true; 2235 } 2236 2237 // 2238 // CommitIterator 2239 // 2240 // differs from xit's ObjectIterator in a few ways: 2241 // 2242 // 1. iterates parents first, rather than children first 2243 // 2. `next` only returns the object id, not the full Object 2244 // 3. stores its temporary state in a file on disk to avoid OOMs 2245 // 2246 2247 pub fn CommitIterator(comptime repo_kind: rp.RepoKind, comptime repo_opts: rp.RepoOpts(repo_kind)) type { 2248 return struct { 2249 const DB = EventDB(repo_opts.hash); 2250 // a plain file is used for the scratch db because each write outside 2251 // a transaction is synced anyway, so buffering would gain nothing 2252 const TempDB = xit.xitdb.Database(.file, hash.HashInt(repo_opts.hash)); 2253 const db_name = "haxy-repo-events.db"; 2254 2255 repo_dir: std.Io.Dir, 2256 db_file: std.Io.File, 2257 db: *TempDB, 2258 // map of each object id to its children 2259 parent_to_children: TempDB.HashMap(.read_write), 2260 // tracking pending parents is necessary because a child can only 2261 // be returned after all of its unconsumed parents have been returned 2262 child_to_pending_parents: TempDB.HashMap(.read_write), 2263 // queue of object ids that are ready to be returned from `next` 2264 ready_queue: TempDB.ArrayList(.read_write), 2265 ready_queue_index: u64, 2266 // the object id at the tip of this branch 2267 newest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 2268 // the object id furthest back in the history that hasn't been consumed 2269 oldest_object_id: ?[hash.byteLen(repo_opts.hash)]u8, 2270 2271 pub fn init( 2272 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 2273 io: std.Io, 2274 allocator: std.mem.Allocator, 2275 consumed_object_ids: DB.HashMap(.read_only), 2276 ref: rf.Ref, 2277 ) !CommitIterator(repo_kind, repo_opts) { 2278 const db_file = try state.core.repo_dir.createFile(io, db_name, .{ .truncate = true, .lock = .exclusive, .read = true }); 2279 errdefer { 2280 db_file.close(io); 2281 state.core.repo_dir.deleteFile(io, db_name) catch {}; 2282 } 2283 2284 // cursors point at the db, so it needs a stable address 2285 const db_ptr = try allocator.create(TempDB); 2286 errdefer allocator.destroy(db_ptr); 2287 // the db is scratch state that gets deleted after iteration, so 2288 // there is nothing worth fsyncing 2289 db_ptr.* = try TempDB.init(.{ .io = io, .file = db_file, .fsync = false }); 2290 2291 const map = try TempDB.HashMap(.read_write).init(db_ptr.rootCursor()); 2292 2293 const parent_to_children_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "parent->children")); 2294 const parent_to_children = try TempDB.HashMap(.read_write).init(parent_to_children_cursor); 2295 2296 const child_to_pending_parents_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "child->pending-parents")); 2297 const child_to_pending_parents = try TempDB.HashMap(.read_write).init(child_to_pending_parents_cursor); 2298 2299 const ready_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "ready-queue")); 2300 const ready_queue = try TempDB.ArrayList(.read_write).init(ready_queue_cursor); 2301 2302 var self = CommitIterator(repo_kind, repo_opts){ 2303 .repo_dir = state.core.repo_dir, 2304 .db_file = db_file, 2305 .db = db_ptr, 2306 .parent_to_children = parent_to_children, 2307 .child_to_pending_parents = child_to_pending_parents, 2308 .ready_queue = ready_queue, 2309 .ready_queue_index = 0, 2310 .newest_object_id = null, 2311 .oldest_object_id = null, 2312 }; 2313 errdefer self.deinit(io, allocator); 2314 2315 try self.collect(state, io, allocator, consumed_object_ids, ref); 2316 2317 return self; 2318 } 2319 2320 pub fn deinit(self: *CommitIterator(repo_kind, repo_opts), io: std.Io, allocator: std.mem.Allocator) void { 2321 self.db_file.close(io); 2322 self.repo_dir.deleteFile(io, db_name) catch {}; 2323 allocator.destroy(self.db); 2324 } 2325 2326 pub fn next(self: *CommitIterator(repo_kind, repo_opts)) !?[hash.byteLen(repo_opts.hash)]u8 { 2327 if (self.ready_queue_index >= try self.ready_queue.count()) return null; 2328 2329 const oid = try readOidFromList(self.ready_queue.readOnly(), self.ready_queue_index); 2330 self.ready_queue_index += 1; 2331 2332 try self.enqueueChildren(&oid); 2333 return oid; 2334 } 2335 2336 fn collect( 2337 self: *CommitIterator(repo_kind, repo_opts), 2338 state: rp.Repo(repo_kind, repo_opts).State(.read_only), 2339 io: std.Io, 2340 allocator: std.mem.Allocator, 2341 consumed_object_ids: DB.HashMap(.read_only), 2342 ref: rf.Ref, 2343 ) !void { 2344 const map = try TempDB.HashMap(.read_write).init(self.db.rootCursor()); 2345 2346 const walk_queue_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "walk-queue")); 2347 const walk_queue = try TempDB.ArrayList(.read_write).init(walk_queue_cursor); 2348 var walk_queue_index: u64 = 0; 2349 2350 const seen_object_ids_cursor = try map.putCursor(hash.hashInt(repo_opts.hash, "seen-object-ids")); 2351 const seen_object_ids = try TempDB.HashSet(.read_write).init(seen_object_ids_cursor); 2352 2353 const head_oid_hex = (try rf.readRecur(repo_kind, repo_opts, state, io, .{ .ref = ref })) orelse return error.OidNotFound; 2354 var head_oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2355 _ = try std.fmt.hexToBytes(&head_oid, &head_oid_hex); 2356 try walk_queue.append(.{ .bytes = &head_oid }); 2357 2358 while (walk_queue_index < try walk_queue.count()) { 2359 const oid = try readOidFromList(walk_queue.readOnly(), walk_queue_index); 2360 const oid_int = hash.bytesToInt(repo_opts.hash, &oid); 2361 2362 walk_queue_index += 1; 2363 2364 // if we've seen this object id, skip it 2365 if (null != try seen_object_ids.getCursor(oid_int)) { 2366 continue; 2367 } 2368 2369 try seen_object_ids.put(oid_int, .{ .bytes = &oid }); 2370 2371 var commit_object = try obj.Object(repo_kind, repo_opts).init(state, io, allocator, &std.fmt.bytesToHex(oid, .lower)); 2372 defer commit_object.deinit(); 2373 2374 // if this object id has already been consumed, skip it 2375 if (null != try consumed_object_ids.getCursor(oid_int)) { 2376 continue; 2377 } 2378 2379 if (self.newest_object_id == null) self.newest_object_id = oid; 2380 self.oldest_object_id = oid; 2381 2382 const parent_oids = commit_object.content.commit.metadata.parent_oids orelse return error.ParentOidsNotFound; 2383 var pending_parent_count: u64 = 0; 2384 for (parent_oids) |*parent_oid| { 2385 var parent_oid_bytes: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2386 _ = try std.fmt.hexToBytes(&parent_oid_bytes, parent_oid); 2387 const parent_oid_int = hash.bytesToInt(repo_opts.hash, &parent_oid_bytes); 2388 2389 // if this object id has already been consumed, skip it 2390 if (null != try consumed_object_ids.getCursor(parent_oid_int)) { 2391 continue; 2392 } 2393 2394 // if this object id hasn't already been seen, add it to the walk queue 2395 if (null == try seen_object_ids.getCursor(parent_oid_int)) { 2396 try walk_queue.append(.{ .bytes = &parent_oid_bytes }); 2397 } 2398 2399 const children_cursor = try self.parent_to_children.putCursor(parent_oid_int); 2400 const children = try TempDB.HashSet(.read_write).init(children_cursor); 2401 try children.put(oid_int, .{ .bytes = &oid }); 2402 pending_parent_count += 1; 2403 } 2404 2405 if (pending_parent_count == 0) { 2406 try self.ready_queue.append(.{ .bytes = &oid }); 2407 } else { 2408 try self.child_to_pending_parents.put(oid_int, .{ .uint = pending_parent_count }); 2409 } 2410 } 2411 } 2412 2413 fn enqueueChildren(self: *CommitIterator(repo_kind, repo_opts), oid: *const [hash.byteLen(repo_opts.hash)]u8) !void { 2414 const oid_int = hash.bytesToInt(repo_opts.hash, oid); 2415 const children_cursor = try self.parent_to_children.getCursor(oid_int) orelse return; 2416 const children = try TempDB.HashSet(.read_only).init(children_cursor); 2417 var children_iter = try children.iterator(); 2418 2419 while (try children_iter.next()) |*child_cursor| { 2420 const kv_pair = try child_cursor.readKeyValuePair(); 2421 const child_oid_int = kv_pair.hash; 2422 const pending_parent_cursor = try self.child_to_pending_parents.getCursor(child_oid_int) orelse return error.CursorNotFound; 2423 const pending_parent_count = try pending_parent_cursor.readUint(); 2424 2425 if (pending_parent_count <= 1) { 2426 _ = try self.child_to_pending_parents.remove(child_oid_int); 2427 const child_oid = hash.intToBytes(hash.HashInt(repo_opts.hash), child_oid_int); 2428 try self.ready_queue.append(.{ .bytes = &child_oid }); 2429 } else { 2430 try self.child_to_pending_parents.put(child_oid_int, .{ .uint = pending_parent_count - 1 }); 2431 } 2432 } 2433 } 2434 2435 fn readOidFromList(list: TempDB.ArrayList(.read_only), index: u64) ![hash.byteLen(repo_opts.hash)]u8 { 2436 const oid_cursor = try list.getCursor(index) orelse return error.CursorNotFound; 2437 var oid: [hash.byteLen(repo_opts.hash)]u8 = undefined; 2438 _ = try oid_cursor.readBytes(&oid); 2439 return oid; 2440 } 2441 }; 2442 } 2443