+170 -61

Cargo.lock

··· 167 167 checksum = "46c5e41b57b8bba42a04676d81cb89e9ee8e859a1a66f80a5a72e1cb76b34d43" 168 168 169 169 [[package]] 170 170 + name = "byteorder-lite" 171 171 + version = "0.1.0" 172 172 + source = "registry+https://github.com/rust-lang/crates.io-index" 173 173 + checksum = "8f1fe948ff07f4bd06c30984e69f5b4899c516a3ef74f34df92a2df2ab535495" 174 174 + 175 175 + [[package]] 170 176 name = "bytes" 171 177 version = "1.10.1" 172 178 source = "registry+https://github.com/rust-lang/crates.io-index" 173 179 checksum = "d71b6127be86fdcfddb610f7182ac57211d4b18a3e9c82eb2d17662f2227ad6a" 180 180 + 181 181 + [[package]] 182 182 + name = "byteview" 183 183 + version = "0.10.0" 184 184 + source = "registry+https://github.com/rust-lang/crates.io-index" 185 185 + checksum = "dda4398f387cc6395a3e93b3867cd9abda914c97a0b344d1eefb2e5c51785fca" 174 186 175 187 [[package]] 176 188 name = "cast" ··· 281 293 checksum = "b05b61dc5112cbb17e4b6cd61790d9845d13888356391624cbe7e41efeac1e75" 282 294 283 295 [[package]] 296 296 + name = "compare" 297 297 + version = "0.0.6" 298 298 + source = "registry+https://github.com/rust-lang/crates.io-index" 299 299 + checksum = "ea0095f6103c2a8b44acd6fd15960c801dafebf02e21940360833e0673f48ba7" 300 300 + 301 301 + [[package]] 284 302 name = "const-str" 285 303 version = "0.4.3" 286 304 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 358 376 ] 359 377 360 378 [[package]] 379 379 + name = "crossbeam-skiplist" 380 380 + version = "0.1.3" 381 381 + source = "registry+https://github.com/rust-lang/crates.io-index" 382 382 + checksum = "df29de440c58ca2cc6e587ec3d22347551a32435fbde9d2bff64e78a9ffa151b" 383 383 + dependencies = [ 384 384 + "crossbeam-epoch", 385 385 + "crossbeam-utils", 386 386 + ] 387 387 + 388 388 + [[package]] 361 389 name = "crossbeam-utils" 362 390 version = "0.8.21" 363 391 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 380 408 ] 381 409 382 410 [[package]] 411 411 + name = "dashmap" 412 412 + version = "6.1.0" 413 413 + source = "registry+https://github.com/rust-lang/crates.io-index" 414 414 + checksum = "5041cc499144891f3790297212f32a74fb938e5136a14943f338ef9e0ae276cf" 415 415 + dependencies = [ 416 416 + "cfg-if", 417 417 + "crossbeam-utils", 418 418 + "hashbrown 0.14.5", 419 419 + "lock_api", 420 420 + "once_cell", 421 421 + "parking_lot_core", 422 422 + ] 423 423 + 424 424 + [[package]] 383 425 name = "data-encoding" 384 426 version = "2.9.0" 385 427 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 422 464 checksum = "48c757948c5ede0e46177b7add2e67155f70e33c07fea8284df6576da70b3719" 423 465 424 466 [[package]] 467 467 + name = "enum_dispatch" 468 468 + version = "0.3.13" 469 469 + source = "registry+https://github.com/rust-lang/crates.io-index" 470 470 + checksum = "aa18ce2bc66555b3218614519ac839ddb759a7d6720732f979ef8d13be147ecd" 471 471 + dependencies = [ 472 472 + "once_cell", 473 473 + "proc-macro2", 474 474 + "quote", 475 475 + "syn 2.0.106", 476 476 + ] 477 477 + 478 478 + [[package]] 425 479 name = "env_filter" 426 480 version = "0.1.3" 427 481 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 445 499 ] 446 500 447 501 [[package]] 502 502 + name = "equivalent" 503 503 + version = "1.0.2" 504 504 + source = "registry+https://github.com/rust-lang/crates.io-index" 505 505 + checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f" 506 506 + 507 507 + [[package]] 448 508 name = "errno" 449 509 version = "0.3.14" 450 510 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 455 515 ] 456 516 457 517 [[package]] 458 458 - name = "fallible-iterator" 459 459 - version = "0.3.0" 518 518 + name = "fastrand" 519 519 + version = "2.3.0" 460 520 source = "registry+https://github.com/rust-lang/crates.io-index" 461 461 - checksum = "2acce4a10f12dc2fb14a218589d4f1f62ef011b2d0cc4b3cb1bba8e94da14649" 521 521 + checksum = "37909eebbb50d72f9059c3b6d82c0463f2ff062c9e95845c43a6c9c0355411be" 462 522 463 523 [[package]] 464 464 - name = "fallible-streaming-iterator" 465 465 - version = "0.1.9" 524 524 + name = "fjall" 525 525 + version = "3.0.1" 466 526 source = "registry+https://github.com/rust-lang/crates.io-index" 467 467 - checksum = "7360491ce676a36bf9bb3c56c1aa791658183a54d2744120f27285738d90465a" 527 527 + checksum = "4f69637c02d38ad1b0f003101d0195a60368130aa17d9ef78b1557d265a22093" 528 528 + dependencies = [ 529 529 + "byteorder-lite", 530 530 + "byteview", 531 531 + "dashmap", 532 532 + "flume", 533 533 + "log", 534 534 + "lsm-tree", 535 535 + "tempfile", 536 536 + "xxhash-rust", 537 537 + ] 468 538 469 539 [[package]] 470 470 - name = "fastrand" 471 471 - version = "2.3.0" 540 540 + name = "flume" 541 541 + version = "0.12.0" 472 542 source = "registry+https://github.com/rust-lang/crates.io-index" 473 473 - checksum = "37909eebbb50d72f9059c3b6d82c0463f2ff062c9e95845c43a6c9c0355411be" 474 474 - 475 475 - [[package]] 476 476 - name = "foldhash" 477 477 - version = "0.1.5" 478 478 - source = "registry+https://github.com/rust-lang/crates.io-index" 479 479 - checksum = "d9c4f5dac5e15c24eb999c26181a6ca40b39fe946cbe4c263c7209467bc83af2" 543 543 + checksum = "5e139bc46ca777eb5efaf62df0ab8cc5fd400866427e56c68b22e414e53bd3be" 544 544 + dependencies = [ 545 545 + "spin", 546 546 + ] 480 547 481 548 [[package]] 482 549 name = "futures" ··· 608 675 609 676 [[package]] 610 677 name = "hashbrown" 611 611 - version = "0.15.5" 678 678 + version = "0.14.5" 612 679 source = "registry+https://github.com/rust-lang/crates.io-index" 613 613 - checksum = "9229cfe53dfd69f0609a49f65461bd93001ea1ef889cd5529dd176593f5338a1" 614 614 - dependencies = [ 615 615 - "foldhash", 616 616 - ] 680 680 + checksum = "e5274423e17b7c9fc20b6e7e208532f9b19825d82dfd615708b70edd83df41f1" 617 681 618 682 [[package]] 619 619 - name = "hashlink" 620 620 - version = "0.10.0" 683 683 + name = "hashbrown" 684 684 + version = "0.16.1" 621 685 source = "registry+https://github.com/rust-lang/crates.io-index" 622 622 - checksum = "7382cf6263419f2d8df38c55d7da83da5c18aef87fc7a7fc1fb1e344edfe14c1" 623 623 - dependencies = [ 624 624 - "hashbrown", 625 625 - ] 686 686 + checksum = "841d1cc9bed7f9236f321df977030373f4a4163ae1a7dbfe1a51a2c1a51d9100" 626 687 627 688 [[package]] 628 689 name = "heck" ··· 631 692 checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea" 632 693 633 694 [[package]] 695 695 + name = "interval-heap" 696 696 + version = "0.0.5" 697 697 + source = "registry+https://github.com/rust-lang/crates.io-index" 698 698 + checksum = "11274e5e8e89b8607cfedc2910b6626e998779b48a019151c7604d0adcb86ac6" 699 699 + dependencies = [ 700 700 + "compare", 701 701 + ] 702 702 + 703 703 + [[package]] 634 704 name = "io-uring" 635 705 version = "0.7.10" 636 706 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 730 800 checksum = "58f929b4d672ea937a23a1ab494143d968337a5f47e56d0815df1e0890ddf174" 731 801 732 802 [[package]] 733 733 - name = "libsqlite3-sys" 734 734 - version = "0.35.0" 735 735 - source = "registry+https://github.com/rust-lang/crates.io-index" 736 736 - checksum = "133c182a6a2c87864fe97778797e46c7e999672690dc9fa3ee8e241aa4a9c13f" 737 737 - dependencies = [ 738 738 - "pkg-config", 739 739 - "vcpkg", 740 740 - ] 741 741 - 742 742 - [[package]] 743 803 name = "linux-raw-sys" 744 804 version = "0.11.0" 745 805 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 761 821 checksum = "34080505efa8e45a4b816c349525ebe327ceaa8559756f0356cba97ef3bf7432" 762 822 763 823 [[package]] 824 824 + name = "lsm-tree" 825 825 + version = "3.0.1" 826 826 + source = "registry+https://github.com/rust-lang/crates.io-index" 827 827 + checksum = "b875f1dfe14f557f805b167fb9b0fc54c5560c7a4bd6ae02535b2846f276a8cb" 828 828 + dependencies = [ 829 829 + "byteorder-lite", 830 830 + "byteview", 831 831 + "crossbeam-skiplist", 832 832 + "enum_dispatch", 833 833 + "interval-heap", 834 834 + "log", 835 835 + "quick_cache", 836 836 + "rustc-hash", 837 837 + "self_cell", 838 838 + "sfa", 839 839 + "tempfile", 840 840 + "varint-rs", 841 841 + "xxhash-rust", 842 842 + ] 843 843 + 844 844 + [[package]] 764 845 name = "match-lookup" 765 846 version = "0.1.1" 766 847 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 892 973 checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184" 893 974 894 975 [[package]] 895 895 - name = "pkg-config" 896 896 - version = "0.3.32" 897 897 - source = "registry+https://github.com/rust-lang/crates.io-index" 898 898 - checksum = "7edddbd0b52d732b21ad9a5fab5c704c14cd949e5e9a1ec5929a24fded1b904c" 899 899 - 900 900 - [[package]] 901 976 name = "plotters" 902 977 version = "0.3.7" 903 978 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 950 1025 ] 951 1026 952 1027 [[package]] 1028 1028 + name = "quick_cache" 1029 1029 + version = "0.6.18" 1030 1030 + source = "registry+https://github.com/rust-lang/crates.io-index" 1031 1031 + checksum = "7ada44a88ef953a3294f6eb55d2007ba44646015e18613d2f213016379203ef3" 1032 1032 + dependencies = [ 1033 1033 + "equivalent", 1034 1034 + "hashbrown 0.16.1", 1035 1035 + ] 1036 1036 + 1037 1037 + [[package]] 953 1038 name = "quote" 954 1039 version = "1.0.41" 955 1040 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 1024 1109 1025 1110 [[package]] 1026 1111 name = "repo-stream" 1027 1027 - version = "0.2.0" 1112 1112 + version = "0.2.2" 1028 1113 dependencies = [ 1029 1114 "bincode", 1030 1115 "clap", 1031 1116 "criterion", 1032 1117 "env_logger", 1118 1118 + "fjall", 1033 1119 "futures", 1034 1120 "futures-core", 1035 1121 "ipld-core", 1036 1122 "iroh-car", 1037 1123 "log", 1038 1124 "multibase", 1039 1039 - "rusqlite", 1040 1125 "serde", 1041 1126 "serde_bytes", 1042 1127 "serde_ipld_dagcbor", ··· 1047 1132 ] 1048 1133 1049 1134 [[package]] 1050 1050 - name = "rusqlite" 1051 1051 - version = "0.37.0" 1135 1135 + name = "rustc-demangle" 1136 1136 + version = "0.1.26" 1052 1137 source = "registry+https://github.com/rust-lang/crates.io-index" 1053 1053 - checksum = "165ca6e57b20e1351573e3729b958bc62f0e48025386970b6e4d29e7a7e71f3f" 1054 1054 - dependencies = [ 1055 1055 - "bitflags", 1056 1056 - "fallible-iterator", 1057 1057 - "fallible-streaming-iterator", 1058 1058 - "hashlink", 1059 1059 - "libsqlite3-sys", 1060 1060 - "smallvec", 1061 1061 - ] 1138 1138 + checksum = "56f7d92ca342cea22a06f2121d944b4fd82af56988c270852495420f961d4ace" 1062 1139 1063 1140 [[package]] 1064 1064 - name = "rustc-demangle" 1065 1065 - version = "0.1.26" 1141 1141 + name = "rustc-hash" 1142 1142 + version = "2.1.1" 1066 1143 source = "registry+https://github.com/rust-lang/crates.io-index" 1067 1067 - checksum = "56f7d92ca342cea22a06f2121d944b4fd82af56988c270852495420f961d4ace" 1144 1144 + checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d" 1068 1145 1069 1146 [[package]] 1070 1147 name = "rustix" ··· 1105 1182 version = "1.2.0" 1106 1183 source = "registry+https://github.com/rust-lang/crates.io-index" 1107 1184 checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49" 1185 1185 + 1186 1186 + [[package]] 1187 1187 + name = "self_cell" 1188 1188 + version = "1.2.2" 1189 1189 + source = "registry+https://github.com/rust-lang/crates.io-index" 1190 1190 + checksum = "b12e76d157a900eb52e81bc6e9f3069344290341720e9178cde2407113ac8d89" 1108 1191 1109 1192 [[package]] 1110 1193 name = "serde" ··· 1172 1255 ] 1173 1256 1174 1257 [[package]] 1258 1258 + name = "sfa" 1259 1259 + version = "1.0.0" 1260 1260 + source = "registry+https://github.com/rust-lang/crates.io-index" 1261 1261 + checksum = "a1296838937cab56cd6c4eeeb8718ec777383700c33f060e2869867bd01d1175" 1262 1262 + dependencies = [ 1263 1263 + "byteorder-lite", 1264 1264 + "log", 1265 1265 + "xxhash-rust", 1266 1266 + ] 1267 1267 + 1268 1268 + [[package]] 1175 1269 name = "sha2" 1176 1270 version = "0.10.9" 1177 1271 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 1211 1305 dependencies = [ 1212 1306 "libc", 1213 1307 "windows-sys 0.59.0", 1308 1308 + ] 1309 1309 + 1310 1310 + [[package]] 1311 1311 + name = "spin" 1312 1312 + version = "0.9.8" 1313 1313 + source = "registry+https://github.com/rust-lang/crates.io-index" 1314 1314 + checksum = "6980e8d7511241f8acf4aebddbb1ff938df5eebe98691418c4468d0b72a96a67" 1315 1315 + dependencies = [ 1316 1316 + "lock_api", 1214 1317 ] 1215 1318 1216 1319 [[package]] ··· 1372 1475 checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821" 1373 1476 1374 1477 [[package]] 1375 1375 - name = "vcpkg" 1376 1376 - version = "0.2.15" 1478 1478 + name = "varint-rs" 1479 1479 + version = "2.2.0" 1377 1480 source = "registry+https://github.com/rust-lang/crates.io-index" 1378 1378 - checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426" 1481 1481 + checksum = "8f54a172d0620933a27a4360d3db3e2ae0dd6cceae9730751a036bbf182c4b23" 1379 1482 1380 1483 [[package]] 1381 1484 name = "version_check" ··· 1659 1762 version = "0.46.0" 1660 1763 source = "registry+https://github.com/rust-lang/crates.io-index" 1661 1764 checksum = "f17a85883d4e6d00e8a97c586de764dabcc06133f7f1d55dce5cdc070ad7fe59" 1765 1765 + 1766 1766 + [[package]] 1767 1767 + name = "xxhash-rust" 1768 1768 + version = "0.8.15" 1769 1769 + source = "registry+https://github.com/rust-lang/crates.io-index" 1770 1770 + checksum = "fdd20c5420375476fbd4394763288da7eb0cc0b8c11deed431a91562af7335d3" 1662 1771 1663 1772 [[package]] 1664 1773 name = "zerocopy"

+2 -2

Cargo.toml

··· 1 1 [package] 2 2 name = "repo-stream" 3 3 - version = "0.2.0" 3 3 + version = "0.2.2" 4 4 edition = "2024" 5 5 license = "MIT OR Apache-2.0" 6 6 description = "A robust CAR file -> MST walker for atproto" ··· 8 8 9 9 [dependencies] 10 10 bincode = { version = "2.0.1", features = ["serde"] } 11 11 + fjall = { version = "3.0.1", default-features = false } 11 12 futures = "0.3.31" 12 13 futures-core = "0.3.31" 13 14 ipld-core = { version = "0.4.2", features = ["serde"] } 14 15 iroh-car = "0.5.1" 15 16 log = "0.4.28" 16 17 multibase = "0.9.2" 17 17 - rusqlite = "0.37.0" 18 18 serde = { version = "1.0.228", features = ["derive"] } 19 19 serde_bytes = "0.11.19" 20 20 serde_ipld_dagcbor = "0.6.4"

+4

benches/non-huge-cars.rs

··· 3 3 4 4 use criterion::{Criterion, criterion_group, criterion_main}; 5 5 6 6 + const EMPTY_CAR: &'static [u8] = include_bytes!("../car-samples/empty.car"); 6 7 const TINY_CAR: &'static [u8] = include_bytes!("../car-samples/tiny.car"); 7 8 const LITTLE_CAR: &'static [u8] = include_bytes!("../car-samples/little.car"); 8 9 const MIDSIZE_CAR: &'static [u8] = include_bytes!("../car-samples/midsize.car"); ··· 13 14 .build() 14 15 .expect("Creating runtime failed"); 15 16 17 17 + c.bench_function("empty-car", |b| { 18 18 + b.to_async(&rt).iter(async || drive_car(EMPTY_CAR).await) 19 19 + }); 16 20 c.bench_function("tiny-car", |b| { 17 21 b.to_async(&rt).iter(async || drive_car(TINY_CAR).await) 18 22 });

car-samples/empty.car

This is a binary file and will not be displayed.

+6 -7

examples/disk-read-file/main.rs

··· 6 6 use clap::Parser; 7 7 use repo_stream::{DiskBuilder, Driver, DriverBuilder}; 8 8 use std::path::PathBuf; 9 9 + use std::time::Instant; 9 10 10 11 #[derive(Debug, Parser)] 11 12 struct Args { ··· 27 28 let reader = tokio::io::BufReader::new(reader); 28 29 29 30 log::info!("hello! reading the car..."); 31 31 + let t0 = Instant::now(); 30 32 31 33 // in this example we only bother handling CARs that are too big for memory 32 34 // `noop` helper means: do no block processing, store the raw blocks 33 35 let driver = match DriverBuilder::new() 34 34 - .with_mem_limit_mb(10) // how much memory can be used before disk spill 36 36 + .with_mem_limit_mb(32) // how much memory can be used before disk spill 35 37 .load_car(reader) 36 38 .await? 37 39 { ··· 48 50 49 51 // at this point you might want to fetch the account's signing key 50 52 // via the DID from the commit, and then verify the signature. 51 51 - log::warn!("big's comit: {:?}", commit); 53 53 + log::warn!("big's comit ({:?}): {:?}", t0.elapsed(), commit); 52 54 53 55 // pop the driver back out to get some code indentation relief 54 56 driver ··· 78 80 } 79 81 } 80 82 81 81 - log::info!("arrived! joining rx..."); 83 83 + log::info!("arrived! ({:?}) joining rx...", t0.elapsed()); 82 84 83 83 - // clean up the database. would be nice to do this in drop so it happens 84 84 - // automatically, but some blocking work happens, so that's not allowed in 85 85 - // async rust. 🤷‍♀️ 86 86 - join.await?.reset_store().await?; 85 85 + join.await?; 87 86 88 87 log::info!("done. n={n} zeros={zeros}"); 89 88

+50 -2

readme.md

··· 4 4 5 5 [![Crates.io][crates-badge]](https://crates.io/crates/repo-stream) 6 6 [![Documentation][docs-badge]](https://docs.rs/repo-stream) 7 7 + [![Sponsor][sponsor-badge]](https://github.com/sponsors/uniphil) 7 8 8 9 [crates-badge]: https://img.shields.io/crates/v/repo-stream.svg 9 10 [docs-badge]: https://docs.rs/repo-stream/badge.svg 11 11 + [sponsor-badge]: https://img.shields.io/badge/at-microcosm-b820f9?labelColor=b820f9&logo=githubsponsors&logoColor=fff 10 12 13 13 + ```rust 14 14 + use repo_stream::{Driver, DriverBuilder, DriveError, DiskBuilder}; 11 15 12 12 - todo 16 16 + #[tokio::main] 17 17 + async fn main() -> Result<(), DriveError> { 18 18 + // repo-stream takes any AsyncRead as input, like a tokio::fs::File 19 19 + let reader = tokio::fs::File::open("repo.car".into()).await?; 20 20 + let reader = tokio::io::BufReader::new(reader); 21 21 + 22 22 + // example repo workload is simply counting the total record bytes 23 23 + let mut total_size = 0; 24 24 + 25 25 + match DriverBuilder::new() 26 26 + .with_mem_limit_mb(10) 27 27 + .with_block_processor(|rec| rec.len()) // block processing: just extract the raw record size 28 28 + .load_car(reader) 29 29 + .await? 30 30 + { 31 31 + 32 32 + // if all blocks fit within memory 33 33 + Driver::Memory(_commit, mut driver) => { 34 34 + while let Some(chunk) = driver.next_chunk(256).await? { 35 35 + for (_rkey, size) in chunk { 36 36 + total_size += size; 37 37 + } 38 38 + } 39 39 + }, 40 40 + 41 41 + // if the CAR was too big for in-memory processing 42 42 + Driver::Disk(paused) => { 43 43 + // set up a disk store we can spill to 44 44 + let store = DiskBuilder::new().open("some/path.db".into()).await?; 45 45 + // do the spilling, get back a (similar) driver 46 46 + let (_commit, mut driver) = paused.finish_loading(store).await?; 47 47 + 48 48 + while let Some(chunk) = driver.next_chunk(256).await? { 49 49 + for (_rkey, size) in chunk { 50 50 + total_size += size; 51 51 + } 52 52 + } 53 53 + } 54 54 + }; 55 55 + println!("sum of size of all records: {total_size}"); 56 56 + Ok(()) 57 57 + } 58 58 + ``` 59 59 + 60 60 + more recent todo 13 61 14 62 - [ ] get an *emtpy* car for the test suite 15 15 - - [ ] implement a max size on disk limit 63 63 + - [x] implement a max size on disk limit 16 64 17 65 18 66 -----

+38 -96

src/disk.rs

··· 18 18 */ 19 19 20 20 use crate::drive::DriveError; 21 21 - use rusqlite::OptionalExtension; 21 21 + use fjall::config::{CompressionPolicy, PinningPolicy, RestartIntervalPolicy}; 22 22 + use fjall::{CompressionType, Database, Error as FjallError, Keyspace, KeyspaceCreateOptions}; 22 23 use std::path::PathBuf; 23 24 24 25 #[derive(Debug, thiserror::Error)] ··· 28 29 /// (The wrapped err should probably be obscured to remove public-facing 29 30 /// sqlite bits) 30 31 #[error(transparent)] 31 31 - DbError(#[from] rusqlite::Error), 32 32 + DbError(#[from] FjallError), 32 33 /// A tokio blocking task failed to join 33 34 #[error("Failed to join a tokio blocking task: {0}")] 34 35 JoinError(#[from] tokio::task::JoinError), ··· 38 39 /// limit. 39 40 #[error("Maximum disk size reached")] 40 41 MaxSizeExceeded, 41 41 - #[error("this error was replaced, seeing this is a bug.")] 42 42 - #[doc(hidden)] 43 43 - Stolen, 44 44 - } 45 45 - 46 46 - impl DiskError { 47 47 - /// hack for ownership challenges with the disk driver 48 48 - pub(crate) fn steal(&mut self) -> Self { 49 49 - let mut swapped = DiskError::Stolen; 50 50 - std::mem::swap(self, &mut swapped); 51 51 - swapped 52 52 - } 53 42 } 54 43 55 44 /// Builder-style disk store setup 45 45 + #[derive(Debug, Clone)] 56 46 pub struct DiskBuilder { 57 47 /// Database in-memory cache allowance 58 48 /// ··· 70 60 impl Default for DiskBuilder { 71 61 fn default() -> Self { 72 62 Self { 73 73 - cache_size_mb: 32, 63 63 + cache_size_mb: 64, 74 64 max_stored_mb: 10 * 1024, // 10 GiB 75 65 } 76 66 } ··· 83 73 } 84 74 /// Set the in-memory cache allowance for the database 85 75 /// 86 86 - /// Default: 32 MiB 76 76 + /// Default: 64 MiB 87 77 pub fn with_cache_size_mb(mut self, size: usize) -> Self { 88 78 self.cache_size_mb = size; 89 79 self ··· 96 86 self 97 87 } 98 88 /// Open and initialize the actual disk storage 99 99 - pub async fn open(self, path: PathBuf) -> Result<DiskStore, DiskError> { 89 89 + pub async fn open(&self, path: PathBuf) -> Result<DiskStore, DiskError> { 100 90 DiskStore::new(path, self.cache_size_mb, self.max_stored_mb).await 101 91 } 102 92 } 103 93 104 94 /// On-disk block storage 105 95 pub struct DiskStore { 106 106 - conn: rusqlite::Connection, 96 96 + #[allow(unused)] 97 97 + db: Database, 98 98 + partition: Keyspace, 107 99 max_stored: usize, 108 100 stored: usize, 109 101 } ··· 116 108 max_stored_mb: usize, 117 109 ) -> Result<Self, DiskError> { 118 110 let max_stored = max_stored_mb * 2_usize.pow(20); 119 119 - let conn = tokio::task::spawn_blocking(move || { 120 120 - let conn = rusqlite::Connection::open(path)?; 121 121 - 122 122 - let sqlite_one_mb = -(2_i64.pow(10)); // negative is kibibytes for sqlite cache_size 111 111 + let (db, partition) = tokio::task::spawn_blocking(move || { 112 112 + let db = Database::builder(path) 113 113 + // .manual_journal_persist(true) 114 114 + // .flush_workers(1) 115 115 + // .compaction_workers(1) 116 116 + .journal_compression(CompressionType::None) 117 117 + .cache_size(cache_mb as u64 * 2_u64.pow(20)) 118 118 + .temporary(true) 119 119 + .open()?; 120 120 + let opts = KeyspaceCreateOptions::default() 121 121 + .data_block_restart_interval_policy(RestartIntervalPolicy::all(8)) 122 122 + .filter_block_pinning_policy(PinningPolicy::disabled()) 123 123 + .expect_point_read_hits(true) 124 124 + .data_block_compression_policy(CompressionPolicy::disabled()) 125 125 + .manual_journal_persist(true) 126 126 + .max_memtable_size(32 * 2_u64.pow(20)); 127 127 + let partition = db.keyspace("z", || opts)?; 123 128 124 124 - // conn.pragma_update(None, "journal_mode", "OFF")?; 125 125 - // conn.pragma_update(None, "journal_mode", "MEMORY")?; 126 126 - conn.pragma_update(None, "journal_mode", "WAL")?; 127 127 - // conn.pragma_update(None, "wal_autocheckpoint", "0")?; // this lets things get a bit big on disk 128 128 - conn.pragma_update(None, "synchronous", "OFF")?; 129 129 - conn.pragma_update( 130 130 - None, 131 131 - "cache_size", 132 132 - (cache_mb as i64 * sqlite_one_mb).to_string(), 133 133 - )?; 134 134 - Self::reset_tables(&conn)?; 135 135 - 136 136 - Ok::<_, DiskError>(conn) 129 129 + Ok::<_, DiskError>((db, partition)) 137 130 }) 138 131 .await??; 139 132 140 133 Ok(Self { 141 141 - conn, 134 134 + db, 135 135 + partition, 142 136 max_stored, 143 137 stored: 0, 144 138 }) 145 139 } 146 146 - pub(crate) fn get_writer(&'_ mut self) -> Result<SqliteWriter<'_>, DiskError> { 147 147 - let tx = self.conn.transaction()?; 148 148 - Ok(SqliteWriter { 149 149 - tx, 150 150 - stored: &mut self.stored, 151 151 - max: self.max_stored, 152 152 - }) 153 153 - } 154 154 - pub(crate) fn get_reader<'conn>(&'conn self) -> Result<SqliteReader<'conn>, DiskError> { 155 155 - let select_stmt = self.conn.prepare("SELECT val FROM blocks WHERE key = ?1")?; 156 156 - Ok(SqliteReader { select_stmt }) 157 157 - } 158 158 - /// Drop and recreate the kv table 159 159 - pub async fn reset(self) -> Result<Self, DiskError> { 160 160 - tokio::task::spawn_blocking(move || { 161 161 - Self::reset_tables(&self.conn)?; 162 162 - Ok(self) 163 163 - }) 164 164 - .await? 165 165 - } 166 166 - fn reset_tables(conn: &rusqlite::Connection) -> Result<(), DiskError> { 167 167 - conn.execute("DROP TABLE IF EXISTS blocks", ())?; 168 168 - conn.execute( 169 169 - "CREATE TABLE blocks ( 170 170 - key BLOB PRIMARY KEY NOT NULL, 171 171 - val BLOB NOT NULL 172 172 - ) WITHOUT ROWID", 173 173 - (), 174 174 - )?; 175 175 - Ok(()) 176 176 - } 177 177 - } 178 140 179 179 - pub(crate) struct SqliteWriter<'conn> { 180 180 - tx: rusqlite::Transaction<'conn>, 181 181 - stored: &'conn mut usize, 182 182 - max: usize, 183 183 - } 184 184 - 185 185 - impl SqliteWriter<'_> { 186 141 pub(crate) fn put_many( 187 142 &mut self, 188 143 kv: impl Iterator<Item = Result<(Vec<u8>, Vec<u8>), DriveError>>, 189 144 ) -> Result<(), DriveError> { 190 190 - let mut insert_stmt = self 191 191 - .tx 192 192 - .prepare_cached("INSERT INTO blocks (key, val) VALUES (?1, ?2)") 193 193 - .map_err(DiskError::DbError)?; 145 145 + let mut batch = self.db.batch(); 194 146 for pair in kv { 195 147 let (k, v) = pair?; 196 196 - *self.stored += v.len(); 197 197 - if *self.stored > self.max { 148 148 + self.stored += v.len(); 149 149 + if self.stored > self.max_stored { 198 150 return Err(DiskError::MaxSizeExceeded.into()); 199 151 } 200 200 - insert_stmt.execute((k, v)).map_err(DiskError::DbError)?; 152 152 + batch.insert(&self.partition, k, v); 201 153 } 154 154 + batch.commit().map_err(DiskError::DbError)?; 202 155 Ok(()) 203 156 } 204 204 - pub fn commit(self) -> Result<(), DiskError> { 205 205 - self.tx.commit()?; 206 206 - Ok(()) 207 207 - } 208 208 - } 209 209 - 210 210 - pub(crate) struct SqliteReader<'conn> { 211 211 - select_stmt: rusqlite::Statement<'conn>, 212 212 - } 213 157 214 214 - impl SqliteReader<'_> { 215 215 - pub(crate) fn get(&mut self, key: Vec<u8>) -> rusqlite::Result<Option<Vec<u8>>> { 216 216 - self.select_stmt 217 217 - .query_one((&key,), |row| row.get(0)) 218 218 - .optional() 158 158 + #[inline] 159 159 + pub(crate) fn get(&mut self, key: &[u8]) -> Result<Option<fjall::Slice>, FjallError> { 160 160 + self.partition.get(key) 219 161 } 220 162 }

+9 -55

src/drive.rs

··· 116 116 } 117 117 118 118 /// Builder-style driver setup 119 119 + #[derive(Debug, Clone)] 119 120 pub struct DriverBuilder { 120 121 pub mem_limit_mb: usize, 121 122 } ··· 153 154 } 154 155 /// Begin processing an atproto MST from a CAR file 155 156 pub async fn load_car<R: AsyncRead + Unpin>( 156 156 - self, 157 157 + &self, 157 158 reader: R, 158 159 ) -> Result<Driver<R, Vec<u8>>, DriveError> { 159 160 Driver::load_car(reader, crate::process::noop, self.mem_limit_mb).await ··· 163 164 /// Builder-style driver intermediate step 164 165 /// 165 166 /// start from `DriverBuilder` 167 167 + #[derive(Debug, Clone)] 166 168 pub struct DriverBuilderWithProcessor<T: Processable> { 167 169 pub mem_limit_mb: usize, 168 170 pub block_processor: fn(Vec<u8>) -> T, ··· 178 180 } 179 181 /// Begin processing an atproto MST from a CAR file 180 182 pub async fn load_car<R: AsyncRead + Unpin>( 181 181 - self, 183 183 + &self, 182 184 reader: R, 183 185 ) -> Result<Driver<R, T>, DriveError> { 184 186 Driver::load_car(reader, self.block_processor, self.mem_limit_mb).await ··· 333 335 // move store in and back out so we can manage lifetimes 334 336 // dump mem blocks into the store 335 337 store = tokio::task::spawn(async move { 336 336 - let mut writer = store.get_writer()?; 337 337 - 338 338 let kvs = self 339 339 .mem_blocks 340 340 .into_iter() 341 341 .map(|(k, v)| Ok(encode(v).map(|v| (k.to_bytes(), v))?)); 342 342 343 343 - writer.put_many(kvs)?; 344 344 - writer.commit()?; 343 343 + store.put_many(kvs)?; 345 344 Ok::<_, DriveError>(store) 346 345 }) 347 346 .await??; 348 347 349 349 - let (tx, mut rx) = mpsc::channel::<Vec<(Cid, MaybeProcessedBlock<T>)>>(2); 348 348 + let (tx, mut rx) = mpsc::channel::<Vec<(Cid, MaybeProcessedBlock<T>)>>(1); 350 349 351 350 let store_worker = tokio::task::spawn_blocking(move || { 352 352 - let mut writer = store.get_writer()?; 353 353 - 354 351 while let Some(chunk) = rx.blocking_recv() { 355 352 let kvs = chunk 356 353 .into_iter() 357 354 .map(|(k, v)| Ok(encode(v).map(|v| (k.to_bytes(), v))?)); 358 358 - writer.put_many(kvs)?; 355 355 + store.put_many(kvs)?; 359 356 } 360 360 - 361 361 - writer.commit()?; 362 357 Ok::<_, DriveError>(store) 363 358 }); // await later 364 359 ··· 451 446 /// println!("{rkey}: size={}", record.len()); 452 447 /// } 453 448 /// } 454 454 - /// let store = disk_driver.reset_store().await?; 455 449 /// # Ok(()) 456 450 /// # } 457 451 /// ``` ··· 466 460 // comes out again. 467 461 let (state, res) = tokio::task::spawn_blocking( 468 462 move || -> (BigState, Result<BlockChunk<T>, DriveError>) { 469 469 - let mut reader_res = state.store.get_reader(); 470 470 - let reader: &mut _ = match reader_res { 471 471 - Ok(ref mut r) => r, 472 472 - Err(ref mut e) => { 473 473 - // unfortunately we can't return the error directly because 474 474 - // (for some reason) it's attached to the lifetime of the 475 475 - // reader? 476 476 - // hack a mem::swap so we can get it out :/ 477 477 - let e_swapped = e.steal(); 478 478 - // the pain: `state` *has to* outlive the reader 479 479 - drop(reader_res); 480 480 - return (state, Err(e_swapped.into())); 481 481 - } 482 482 - }; 483 483 - 484 463 let mut out = Vec::with_capacity(n); 485 464 486 465 for _ in 0..n { 487 466 // walk as far as we can until we run out of blocks or find a record 488 488 - let step = match state.walker.disk_step(reader, process) { 467 467 + let step = match state.walker.disk_step(&mut state.store, process) { 489 468 Ok(s) => s, 490 469 Err(e) => { 491 491 - // the pain: `state` *has to* outlive the reader 492 492 - drop(reader_res); 493 470 return (state, Err(e.into())); 494 471 } 495 472 }; 496 473 match step { 497 474 Step::Missing(cid) => { 498 498 - // the pain: `state` *has to* outlive the reader 499 499 - drop(reader_res); 500 475 return (state, Err(DriveError::MissingBlock(cid))); 501 476 } 502 477 Step::Finish => break, 503 478 Step::Found { rkey, data } => out.push((rkey, data)), 504 479 }; 505 480 } 506 506 - 507 507 - // `state` *has to* outlive the reader 508 508 - drop(reader_res); 509 481 510 482 (state, Ok::<_, DriveError>(out)) 511 483 }, ··· 530 502 tx: mpsc::Sender<Result<BlockChunk<T>, DriveError>>, 531 503 ) -> Result<(), mpsc::error::SendError<Result<BlockChunk<T>, DriveError>>> { 532 504 let BigState { store, walker } = self.state.as_mut().expect("valid state"); 533 533 - let mut reader = match store.get_reader() { 534 534 - Ok(r) => r, 535 535 - Err(e) => return tx.blocking_send(Err(e.into())), 536 536 - }; 537 505 538 506 loop { 539 507 let mut out: BlockChunk<T> = Vec::with_capacity(n); ··· 541 509 for _ in 0..n { 542 510 // walk as far as we can until we run out of blocks or find a record 543 511 544 544 - let step = match walker.disk_step(&mut reader, self.process) { 512 512 + let step = match walker.disk_step(store, self.process) { 545 513 Ok(s) => s, 546 514 Err(e) => return tx.blocking_send(Err(e.into())), 547 515 }; ··· 590 558 /// } 591 559 /// 592 560 /// } 593 593 - /// let store = join.await?.reset_store().await?; 594 561 /// # Ok(()) 595 562 /// # } 596 563 /// ``` ··· 612 579 }); 613 580 614 581 (rx, chan_task) 615 615 - } 616 616 - 617 617 - /// Reset the disk storage so it can be reused. You must call this. 618 618 - /// 619 619 - /// Ideally we'd put this in an `impl Drop`, but since it makes blocking 620 620 - /// calls, that would be risky in an async context. For now you just have to 621 621 - /// carefully make sure you call it. 622 622 - /// 623 623 - /// The sqlite store is returned, so it can be reused for another 624 624 - /// `DiskDriver`. 625 625 - pub async fn reset_store(mut self) -> Result<DiskStore, DriveError> { 626 626 - let BigState { store, .. } = self.state.take().expect("valid state"); 627 627 - Ok(store.reset().await?) 628 582 } 629 583 }

+1 -4

src/lib.rs

··· 53 53 total_size += size; 54 54 } 55 55 } 56 56 - 57 57 - // clean up the disk store (drop tables etc) 58 58 - driver.reset_store().await?; 59 56 } 60 57 }; 61 58 println!("sum of size of all records: {total_size}"); ··· 82 79 pub mod process; 83 80 84 81 pub use disk::{DiskBuilder, DiskError, DiskStore}; 85 85 - pub use drive::{DriveError, Driver, DriverBuilder}; 82 82 + pub use drive::{DriveError, Driver, DriverBuilder, NeedDisk}; 86 83 pub use mst::Commit; 87 84 pub use process::Processable;

+21

src/process.rs

··· 77 77 } 78 78 } 79 79 80 80 + impl Processable for String { 81 81 + fn get_size(&self) -> usize { 82 82 + self.capacity() 83 83 + } 84 84 + } 85 85 + 80 86 impl<Item: Sized + Processable> Processable for Vec<Item> { 81 87 fn get_size(&self) -> usize { 82 88 let slot_size = std::mem::size_of::<Item>(); ··· 85 91 direct_size + items_referenced_size 86 92 } 87 93 } 94 94 + 95 95 + impl<Item: Processable> Processable for Option<Item> { 96 96 + fn get_size(&self) -> usize { 97 97 + self.as_ref().map(|item| item.get_size()).unwrap_or(0) 98 98 + } 99 99 + } 100 100 + 101 101 + impl<Item: Processable, Error: Processable> Processable for Result<Item, Error> { 102 102 + fn get_size(&self) -> usize { 103 103 + match self { 104 104 + Ok(item) => item.get_size(), 105 105 + Err(err) => err.get_size(), 106 106 + } 107 107 + } 108 108 + }

+11 -8

src/walk.rs

··· 1 1 //! Depth-first MST traversal 2 2 3 3 - use crate::disk::SqliteReader; 3 3 + use crate::disk::DiskStore; 4 4 use crate::drive::{DecodeError, MaybeProcessedBlock}; 5 5 use crate::mst::Node; 6 6 use crate::process::Processable; ··· 19 19 #[error("Action node error: {0}")] 20 20 MstError(#[from] MstError), 21 21 #[error("storage error: {0}")] 22 22 - StorageError(#[from] rusqlite::Error), 22 22 + StorageError(#[from] fjall::Error), 23 23 #[error("Decode error: {0}")] 24 24 DecodeError(#[from] DecodeError), 25 25 } ··· 87 87 } 88 88 89 89 fn push_from_node(stack: &mut Vec<Need>, node: &Node, parent_depth: Depth) -> Result<(), MstError> { 90 90 - // empty nodes are not allowed in the MST 91 91 - // ...except for a single one for empty MST, but we wouldn't be pushing that 90 90 + // empty nodes are not allowed in the MST except in an empty MST 92 91 if node.is_empty() { 93 93 - return Err(MstError::EmptyNode); 92 92 + if parent_depth == Depth::Root { 93 93 + return Ok(()); // empty mst, nothing to push 94 94 + } else { 95 95 + return Err(MstError::EmptyNode); 96 96 + } 94 97 } 95 98 96 99 let mut entries = Vec::with_capacity(node.entries.len()); ··· 236 239 /// blocking!!!!!! 237 240 pub fn disk_step<T: Processable>( 238 241 &mut self, 239 239 - reader: &mut SqliteReader, 242 242 + reader: &mut DiskStore, 240 243 process: impl Fn(Vec<u8>) -> T, 241 244 ) -> Result<Step<T>, WalkError> { 242 245 loop { ··· 249 252 &mut Need::Node { depth, cid } => { 250 253 let cid_bytes = cid.to_bytes(); 251 254 log::trace!("need node {cid:?}"); 252 252 - let Some(block_bytes) = reader.get(cid_bytes)? else { 255 255 + let Some(block_bytes) = reader.get(&cid_bytes)? else { 253 256 log::trace!("node not found, resting"); 254 257 return Ok(Step::Missing(cid)); 255 258 }; ··· 271 274 Need::Record { rkey, cid } => { 272 275 log::trace!("need record {cid:?}"); 273 276 let cid_bytes = cid.to_bytes(); 274 274 - let Some(data_bytes) = reader.get(cid_bytes)? else { 277 277 + let Some(data_bytes) = reader.get(&cid_bytes)? else { 275 278 log::trace!("record block not found, resting"); 276 279 return Ok(Step::Missing(*cid)); 277 280 };

+16 -5

tests/non-huge-cars.rs

··· 1 1 extern crate repo_stream; 2 2 use repo_stream::Driver; 3 3 4 4 + const EMPTY_CAR: &'static [u8] = include_bytes!("../car-samples/empty.car"); 4 5 const TINY_CAR: &'static [u8] = include_bytes!("../car-samples/tiny.car"); 5 6 const LITTLE_CAR: &'static [u8] = include_bytes!("../car-samples/little.car"); 6 7 const MIDSIZE_CAR: &'static [u8] = include_bytes!("../car-samples/midsize.car"); 7 8 8 8 - async fn test_car(bytes: &[u8], expected_records: usize, expected_sum: usize) { 9 9 + async fn test_car( 10 10 + bytes: &[u8], 11 11 + expected_records: usize, 12 12 + expected_sum: usize, 13 13 + expect_profile: bool, 14 14 + ) { 9 15 let mut driver = match Driver::load_car(bytes, |block| block.len(), 10 /* MiB */) 10 16 .await 11 17 .unwrap() ··· 33 39 34 40 assert_eq!(records, expected_records); 35 41 assert_eq!(sum, expected_sum); 36 36 - assert!(found_bsky_profile); 42 42 + assert_eq!(found_bsky_profile, expect_profile); 43 43 + } 44 44 + 45 45 + #[tokio::test] 46 46 + async fn test_empty_car() { 47 47 + test_car(EMPTY_CAR, 0, 0, false).await 37 48 } 38 49 39 50 #[tokio::test] 40 51 async fn test_tiny_car() { 41 41 - test_car(TINY_CAR, 8, 2071).await 52 52 + test_car(TINY_CAR, 8, 2071, true).await 42 53 } 43 54 44 55 #[tokio::test] 45 56 async fn test_little_car() { 46 46 - test_car(LITTLE_CAR, 278, 246960).await 57 57 + test_car(LITTLE_CAR, 278, 246960, true).await 47 58 } 48 59 49 60 #[tokio::test] 50 61 async fn test_midsize_car() { 51 51 - test_car(MIDSIZE_CAR, 11585, 3741393).await 62 62 + test_car(MIDSIZE_CAR, 11585, 3741393, true).await 52 63 }