+133 -1114

Cargo.lock

··· 3 3 version = 4 4 4 5 5 [[package]] 6 6 - name = "addr2line" 7 7 - version = "0.25.1" 8 8 - source = "registry+https://github.com/rust-lang/crates.io-index" 9 9 - checksum = "1b5d307320b3181d6d7954e663bd7c774a838b8220fe0593c86d9fb09f498b4b" 10 10 - dependencies = [ 11 11 - "gimli", 12 12 - ] 13 13 - 14 14 - [[package]] 15 15 - name = "adler2" 16 16 - version = "2.0.1" 17 17 - source = "registry+https://github.com/rust-lang/crates.io-index" 18 18 - checksum = "320119579fcad9c21884f5c4861d16174d0e06250625266f50fe6898340abefa" 19 19 - 20 20 - [[package]] 21 21 - name = "aho-corasick" 22 22 - version = "1.1.3" 23 23 - source = "registry+https://github.com/rust-lang/crates.io-index" 24 24 - checksum = "8e60d3430d3a69478ad0993f19238d2df97c507009a52b3c10addcd7f6bcb916" 25 25 - dependencies = [ 26 26 - "memchr", 27 27 - ] 28 28 - 29 29 - [[package]] 30 30 - name = "anes" 31 31 - version = "0.1.6" 32 32 - source = "registry+https://github.com/rust-lang/crates.io-index" 33 33 - checksum = "4b46cbb362ab8752921c97e041f5e366ee6297bd428a31275b9fcf1e380f7299" 34 34 - 35 35 - [[package]] 36 6 name = "anstream" 37 7 version = "0.6.21" 38 8 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 68 38 source = "registry+https://github.com/rust-lang/crates.io-index" 69 39 checksum = "9e231f6134f61b71076a3eab506c379d4f36122f2af15a9ff04415ea4c3339e2" 70 40 dependencies = [ 71 71 - "windows-sys 0.60.2", 41 41 + "windows-sys", 72 42 ] 73 43 74 44 [[package]] ··· 79 49 dependencies = [ 80 50 "anstyle", 81 51 "once_cell_polyfill", 82 82 - "windows-sys 0.60.2", 83 83 - ] 84 84 - 85 85 - [[package]] 86 86 - name = "anyhow" 87 87 - version = "1.0.100" 88 88 - source = "registry+https://github.com/rust-lang/crates.io-index" 89 89 - checksum = "a23eb6b1614318a8071c9b2521f36b424b2c83db5eb3a0fead4a6c0809af6e61" 90 90 - 91 91 - [[package]] 92 92 - name = "autocfg" 93 93 - version = "1.5.0" 94 94 - source = "registry+https://github.com/rust-lang/crates.io-index" 95 95 - checksum = "c08606f8c3cbf4ce6ec8e28fb0014a2c086708fe954eaa885384a6165172e7e8" 96 96 - 97 97 - [[package]] 98 98 - name = "backtrace" 99 99 - version = "0.3.76" 100 100 - source = "registry+https://github.com/rust-lang/crates.io-index" 101 101 - checksum = "bb531853791a215d7c62a30daf0dde835f381ab5de4589cfe7c649d2cbe92bd6" 102 102 - dependencies = [ 103 103 - "addr2line", 104 104 - "cfg-if", 105 105 - "libc", 106 106 - "miniz_oxide", 107 107 - "object", 108 108 - "rustc-demangle", 109 109 - "windows-link", 110 110 - ] 111 111 - 112 112 - [[package]] 113 113 - name = "base-x" 114 114 - version = "0.2.11" 115 115 - source = "registry+https://github.com/rust-lang/crates.io-index" 116 116 - checksum = "4cbbc9d0964165b47557570cce6c952866c2678457aca742aafc9fb771d30270" 117 117 - 118 118 - [[package]] 119 119 - name = "base256emoji" 120 120 - version = "1.0.2" 121 121 - source = "registry+https://github.com/rust-lang/crates.io-index" 122 122 - checksum = "b5e9430d9a245a77c92176e649af6e275f20839a48389859d1661e9a128d077c" 123 123 - dependencies = [ 124 124 - "const-str", 125 125 - "match-lookup", 126 126 - ] 127 127 - 128 128 - [[package]] 129 129 - name = "bincode" 130 130 - version = "2.0.1" 131 131 - source = "registry+https://github.com/rust-lang/crates.io-index" 132 132 - checksum = "36eaf5d7b090263e8150820482d5d93cd964a81e4019913c972f4edcc6edb740" 133 133 - dependencies = [ 134 134 - "bincode_derive", 135 135 - "serde", 136 136 - "unty", 137 137 - ] 138 138 - 139 139 - [[package]] 140 140 - name = "bincode_derive" 141 141 - version = "2.0.1" 142 142 - source = "registry+https://github.com/rust-lang/crates.io-index" 143 143 - checksum = "bf95709a440f45e986983918d0e8a1f30a9b1df04918fc828670606804ac3c09" 144 144 - dependencies = [ 145 145 - "virtue", 52 52 + "windows-sys", 146 53 ] 147 54 148 55 [[package]] ··· 152 59 checksum = "2261d10cca569e4643e526d8dc2e62e433cc8aba21ab764233731f8d369bf394" 153 60 154 61 [[package]] 155 155 - name = "bumpalo" 156 156 - version = "3.19.0" 62 62 + name = "byteorder-lite" 63 63 + version = "0.1.0" 157 64 source = "registry+https://github.com/rust-lang/crates.io-index" 158 158 - checksum = "46c5e41b57b8bba42a04676d81cb89e9ee8e859a1a66f80a5a72e1cb76b34d43" 65 65 + checksum = "8f1fe948ff07f4bd06c30984e69f5b4899c516a3ef74f34df92a2df2ab535495" 159 66 160 67 [[package]] 161 161 - name = "bytes" 162 162 - version = "1.10.1" 163 163 - source = "registry+https://github.com/rust-lang/crates.io-index" 164 164 - checksum = "d71b6127be86fdcfddb610f7182ac57211d4b18a3e9c82eb2d17662f2227ad6a" 165 165 - 166 166 - [[package]] 167 167 - name = "cast" 168 168 - version = "0.3.0" 169 169 - source = "registry+https://github.com/rust-lang/crates.io-index" 170 170 - checksum = "37b2a672a2cb129a2e41c10b1224bb368f9f37a2b16b612598138befd7b37eb5" 171 171 - 172 172 - [[package]] 173 173 - name = "cbor4ii" 174 174 - version = "0.2.14" 68 68 + name = "byteview" 69 69 + version = "0.10.0" 175 70 source = "registry+https://github.com/rust-lang/crates.io-index" 176 176 - checksum = "b544cf8c89359205f4f990d0e6f3828db42df85b5dac95d09157a250eb0749c4" 177 177 - dependencies = [ 178 178 - "serde", 179 179 - ] 71 71 + checksum = "dda4398f387cc6395a3e93b3867cd9abda914c97a0b344d1eefb2e5c51785fca" 180 72 181 73 [[package]] 182 74 name = "cfg-if" ··· 185 77 checksum = "2fd1289c04a9ea8cb22300a459a72a385d7c73d3259e2ed7dcb2af674838cfa9" 186 78 187 79 [[package]] 188 188 - name = "ciborium" 189 189 - version = "0.2.2" 190 190 - source = "registry+https://github.com/rust-lang/crates.io-index" 191 191 - checksum = "42e69ffd6f0917f5c029256a24d0161db17cea3997d185db0d35926308770f0e" 192 192 - dependencies = [ 193 193 - "ciborium-io", 194 194 - "ciborium-ll", 195 195 - "serde", 196 196 - ] 197 197 - 198 198 - [[package]] 199 199 - name = "ciborium-io" 200 200 - version = "0.2.2" 201 201 - source = "registry+https://github.com/rust-lang/crates.io-index" 202 202 - checksum = "05afea1e0a06c9be33d539b876f1ce3692f4afea2cb41f740e7743225ed1c757" 203 203 - 204 204 - [[package]] 205 205 - name = "ciborium-ll" 206 206 - version = "0.2.2" 207 207 - source = "registry+https://github.com/rust-lang/crates.io-index" 208 208 - checksum = "57663b653d948a338bfb3eeba9bb2fd5fcfaecb9e199e87e1eda4d9e8b240fd9" 209 209 - dependencies = [ 210 210 - "ciborium-io", 211 211 - "half", 212 212 - ] 213 213 - 214 214 - [[package]] 215 215 - name = "cid" 216 216 - version = "0.11.1" 217 217 - source = "registry+https://github.com/rust-lang/crates.io-index" 218 218 - checksum = "3147d8272e8fa0ccd29ce51194dd98f79ddfb8191ba9e3409884e751798acf3a" 219 219 - dependencies = [ 220 220 - "core2", 221 221 - "multibase", 222 222 - "multihash", 223 223 - "serde", 224 224 - "serde_bytes", 225 225 - "unsigned-varint 0.8.0", 226 226 - ] 227 227 - 228 228 - [[package]] 229 80 name = "clap" 230 81 version = "4.5.48" 231 82 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 256 107 "heck", 257 108 "proc-macro2", 258 109 "quote", 259 259 - "syn 2.0.106", 110 110 + "syn", 260 111 ] 261 112 262 113 [[package]] ··· 272 123 checksum = "b05b61dc5112cbb17e4b6cd61790d9845d13888356391624cbe7e41efeac1e75" 273 124 274 125 [[package]] 275 275 - name = "const-str" 276 276 - version = "0.4.3" 277 277 - source = "registry+https://github.com/rust-lang/crates.io-index" 278 278 - checksum = "2f421161cb492475f1661ddc9815a745a1c894592070661180fdec3d4872e9c3" 279 279 - 280 280 - [[package]] 281 281 - name = "core2" 282 282 - version = "0.4.0" 283 283 - source = "registry+https://github.com/rust-lang/crates.io-index" 284 284 - checksum = "b49ba7ef1ad6107f8824dbe97de947cbaac53c44e7f9756a1fba0d37c1eec505" 285 285 - dependencies = [ 286 286 - "memchr", 287 287 - ] 288 288 - 289 289 - [[package]] 290 290 - name = "criterion" 291 291 - version = "0.7.0" 292 292 - source = "registry+https://github.com/rust-lang/crates.io-index" 293 293 - checksum = "e1c047a62b0cc3e145fa84415a3191f628e980b194c2755aa12300a4e6cbd928" 294 294 - dependencies = [ 295 295 - "anes", 296 296 - "cast", 297 297 - "ciborium", 298 298 - "clap", 299 299 - "criterion-plot", 300 300 - "itertools", 301 301 - "num-traits", 302 302 - "oorandom", 303 303 - "plotters", 304 304 - "rayon", 305 305 - "regex", 306 306 - "serde", 307 307 - "serde_json", 308 308 - "tinytemplate", 309 309 - "tokio", 310 310 - "walkdir", 311 311 - ] 312 312 - 313 313 - [[package]] 314 314 - name = "criterion-plot" 315 315 - version = "0.6.0" 126 126 + name = "compare" 127 127 + version = "0.0.6" 316 128 source = "registry+https://github.com/rust-lang/crates.io-index" 317 317 - checksum = "9b1bcc0dc7dfae599d84ad0b1a55f80cde8af3725da8313b528da95ef783e338" 318 318 - dependencies = [ 319 319 - "cast", 320 320 - "itertools", 321 321 - ] 129 129 + checksum = "ea0095f6103c2a8b44acd6fd15960c801dafebf02e21940360833e0673f48ba7" 322 130 323 131 [[package]] 324 324 - name = "crossbeam-deque" 325 325 - version = "0.8.6" 132 132 + name = "crossbeam-epoch" 133 133 + version = "0.9.18" 326 134 source = "registry+https://github.com/rust-lang/crates.io-index" 327 327 - checksum = "9dd111b7b7f7d55b72c0a6ae361660ee5853c9af73f70c3c2ef6858b950e2e51" 135 135 + checksum = "5b82ac4a3c2ca9c3460964f020e1402edd5753411d7737aa39c3714ad1b5420e" 328 136 dependencies = [ 329 329 - "crossbeam-epoch", 330 137 "crossbeam-utils", 331 138 ] 332 139 333 140 [[package]] 334 334 - name = "crossbeam-epoch" 335 335 - version = "0.9.18" 141 141 + name = "crossbeam-skiplist" 142 142 + version = "0.1.3" 336 143 source = "registry+https://github.com/rust-lang/crates.io-index" 337 337 - checksum = "5b82ac4a3c2ca9c3460964f020e1402edd5753411d7737aa39c3714ad1b5420e" 144 144 + checksum = "df29de440c58ca2cc6e587ec3d22347551a32435fbde9d2bff64e78a9ffa151b" 338 145 dependencies = [ 146 146 + "crossbeam-epoch", 339 147 "crossbeam-utils", 340 148 ] 341 149 ··· 346 154 checksum = "d0a5c400df2834b80a4c3327b3aad3a4c4cd4de0629063962b03235697506a28" 347 155 348 156 [[package]] 349 349 - name = "crunchy" 350 350 - version = "0.2.4" 351 351 - source = "registry+https://github.com/rust-lang/crates.io-index" 352 352 - checksum = "460fbee9c2c2f33933d720630a6a0bac33ba7053db5344fac858d4b8952d77d5" 353 353 - 354 354 - [[package]] 355 355 - name = "data-encoding" 356 356 - version = "2.9.0" 357 357 - source = "registry+https://github.com/rust-lang/crates.io-index" 358 358 - checksum = "2a2330da5de22e8a3cb63252ce2abb30116bf5265e89c0e01bc17015ce30a476" 359 359 - 360 360 - [[package]] 361 361 - name = "data-encoding-macro" 362 362 - version = "0.1.18" 157 157 + name = "dashmap" 158 158 + version = "6.1.0" 363 159 source = "registry+https://github.com/rust-lang/crates.io-index" 364 364 - checksum = "47ce6c96ea0102f01122a185683611bd5ac8d99e62bc59dd12e6bda344ee673d" 160 160 + checksum = "5041cc499144891f3790297212f32a74fb938e5136a14943f338ef9e0ae276cf" 365 161 dependencies = [ 366 366 - "data-encoding", 367 367 - "data-encoding-macro-internal", 162 162 + "cfg-if", 163 163 + "crossbeam-utils", 164 164 + "hashbrown 0.14.5", 165 165 + "lock_api", 166 166 + "once_cell", 167 167 + "parking_lot_core", 368 168 ] 369 169 370 170 [[package]] 371 371 - name = "data-encoding-macro-internal" 372 372 - version = "0.1.16" 373 373 - source = "registry+https://github.com/rust-lang/crates.io-index" 374 374 - checksum = "8d162beedaa69905488a8da94f5ac3edb4dd4788b732fadb7bd120b2625c1976" 375 375 - dependencies = [ 376 376 - "data-encoding", 377 377 - "syn 2.0.106", 378 378 - ] 379 379 - 380 380 - [[package]] 381 381 - name = "either" 382 382 - version = "1.15.0" 383 383 - source = "registry+https://github.com/rust-lang/crates.io-index" 384 384 - checksum = "48c757948c5ede0e46177b7add2e67155f70e33c07fea8284df6576da70b3719" 385 385 - 386 386 - [[package]] 387 387 - name = "env_filter" 388 388 - version = "0.1.3" 171 171 + name = "enum_dispatch" 172 172 + version = "0.3.13" 389 173 source = "registry+https://github.com/rust-lang/crates.io-index" 390 390 - checksum = "186e05a59d4c50738528153b83b0b0194d3a29507dfec16eccd4b342903397d0" 174 174 + checksum = "aa18ce2bc66555b3218614519ac839ddb759a7d6720732f979ef8d13be147ecd" 391 175 dependencies = [ 392 392 - "log", 393 393 - "regex", 176 176 + "once_cell", 177 177 + "proc-macro2", 178 178 + "quote", 179 179 + "syn", 394 180 ] 395 181 396 182 [[package]] 397 397 - name = "env_logger" 398 398 - version = "0.11.8" 183 183 + name = "equivalent" 184 184 + version = "1.0.2" 399 185 source = "registry+https://github.com/rust-lang/crates.io-index" 400 400 - checksum = "13c863f0904021b108aa8b2f55046443e6b1ebde8fd4a15c399893aae4fa069f" 401 401 - dependencies = [ 402 402 - "anstream", 403 403 - "anstyle", 404 404 - "env_filter", 405 405 - "jiff", 406 406 - "log", 407 407 - ] 186 186 + checksum = "877a4ace8713b0bcf2a4e7eec82529c029f1d0619886d18145fea96c3ffe5c0f" 408 187 409 188 [[package]] 410 189 name = "errno" ··· 413 192 checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb" 414 193 dependencies = [ 415 194 "libc", 416 416 - "windows-sys 0.60.2", 195 195 + "windows-sys", 417 196 ] 418 197 419 198 [[package]] 420 420 - name = "fallible-iterator" 421 421 - version = "0.3.0" 422 422 - source = "registry+https://github.com/rust-lang/crates.io-index" 423 423 - checksum = "2acce4a10f12dc2fb14a218589d4f1f62ef011b2d0cc4b3cb1bba8e94da14649" 424 424 - 425 425 - [[package]] 426 426 - name = "fallible-streaming-iterator" 427 427 - version = "0.1.9" 428 428 - source = "registry+https://github.com/rust-lang/crates.io-index" 429 429 - checksum = "7360491ce676a36bf9bb3c56c1aa791658183a54d2744120f27285738d90465a" 430 430 - 431 431 - [[package]] 432 199 name = "fastrand" 433 200 version = "2.3.0" 434 201 source = "registry+https://github.com/rust-lang/crates.io-index" 435 202 checksum = "37909eebbb50d72f9059c3b6d82c0463f2ff062c9e95845c43a6c9c0355411be" 436 203 437 204 [[package]] 438 438 - name = "foldhash" 439 439 - version = "0.1.5" 440 440 - source = "registry+https://github.com/rust-lang/crates.io-index" 441 441 - checksum = "d9c4f5dac5e15c24eb999c26181a6ca40b39fe946cbe4c263c7209467bc83af2" 442 442 - 443 443 - [[package]] 444 444 - name = "futures" 445 445 - version = "0.3.31" 446 446 - source = "registry+https://github.com/rust-lang/crates.io-index" 447 447 - checksum = "65bc07b1a8bc7c85c5f2e110c476c7389b4554ba72af57d8445ea63a576b0876" 448 448 - dependencies = [ 449 449 - "futures-channel", 450 450 - "futures-core", 451 451 - "futures-executor", 452 452 - "futures-io", 453 453 - "futures-sink", 454 454 - "futures-task", 455 455 - "futures-util", 456 456 - ] 457 457 - 458 458 - [[package]] 459 459 - name = "futures-channel" 460 460 - version = "0.3.31" 461 461 - source = "registry+https://github.com/rust-lang/crates.io-index" 462 462 - checksum = "2dff15bf788c671c1934e366d07e30c1814a8ef514e1af724a602e8a2fbe1b10" 463 463 - dependencies = [ 464 464 - "futures-core", 465 465 - "futures-sink", 466 466 - ] 467 467 - 468 468 - [[package]] 469 469 - name = "futures-core" 470 470 - version = "0.3.31" 471 471 - source = "registry+https://github.com/rust-lang/crates.io-index" 472 472 - checksum = "05f29059c0c2090612e8d742178b0580d2dc940c837851ad723096f87af6663e" 473 473 - 474 474 - [[package]] 475 475 - name = "futures-executor" 476 476 - version = "0.3.31" 477 477 - source = "registry+https://github.com/rust-lang/crates.io-index" 478 478 - checksum = "1e28d1d997f585e54aebc3f97d39e72338912123a67330d723fdbb564d646c9f" 479 479 - dependencies = [ 480 480 - "futures-core", 481 481 - "futures-task", 482 482 - "futures-util", 483 483 - ] 484 484 - 485 485 - [[package]] 486 486 - name = "futures-io" 487 487 - version = "0.3.31" 488 488 - source = "registry+https://github.com/rust-lang/crates.io-index" 489 489 - checksum = "9e5c1b78ca4aae1ac06c48a526a655760685149f0d465d21f37abfe57ce075c6" 490 490 - 491 491 - [[package]] 492 492 - name = "futures-macro" 493 493 - version = "0.3.31" 205 205 + name = "fjall" 206 206 + version = "3.0.1" 494 207 source = "registry+https://github.com/rust-lang/crates.io-index" 495 495 - checksum = "162ee34ebcb7c64a8abebc059ce0fee27c2262618d7b60ed8faf72fef13c3650" 208 208 + checksum = "4f69637c02d38ad1b0f003101d0195a60368130aa17d9ef78b1557d265a22093" 496 209 dependencies = [ 497 497 - "proc-macro2", 498 498 - "quote", 499 499 - "syn 2.0.106", 210 210 + "byteorder-lite", 211 211 + "byteview", 212 212 + "dashmap", 213 213 + "flume", 214 214 + "log", 215 215 + "lsm-tree", 216 216 + "lz4_flex", 217 217 + "tempfile", 218 218 + "xxhash-rust", 500 219 ] 501 220 502 221 [[package]] 503 503 - name = "futures-sink" 504 504 - version = "0.3.31" 505 505 - source = "registry+https://github.com/rust-lang/crates.io-index" 506 506 - checksum = "e575fab7d1e0dcb8d0c7bcf9a63ee213816ab51902e6d244a95819acacf1d4f7" 507 507 - 508 508 - [[package]] 509 509 - name = "futures-task" 510 510 - version = "0.3.31" 511 511 - source = "registry+https://github.com/rust-lang/crates.io-index" 512 512 - checksum = "f90f7dce0722e95104fcb095585910c0977252f286e354b5e3bd38902cd99988" 513 513 - 514 514 - [[package]] 515 515 - name = "futures-util" 516 516 - version = "0.3.31" 222 222 + name = "flume" 223 223 + version = "0.12.0" 517 224 source = "registry+https://github.com/rust-lang/crates.io-index" 518 518 - checksum = "9fa08315bb612088cc391249efdc3bc77536f16c91f6cf495e6fbe85b20a4a81" 225 225 + checksum = "5e139bc46ca777eb5efaf62df0ab8cc5fd400866427e56c68b22e414e53bd3be" 519 226 dependencies = [ 520 520 - "futures-channel", 521 521 - "futures-core", 522 522 - "futures-io", 523 523 - "futures-macro", 524 524 - "futures-sink", 525 525 - "futures-task", 526 526 - "memchr", 527 527 - "pin-project-lite", 528 528 - "pin-utils", 529 529 - "slab", 227 227 + "spin", 530 228 ] 531 229 532 230 [[package]] ··· 538 236 "cfg-if", 539 237 "libc", 540 238 "r-efi", 541 541 - "wasi 0.14.7+wasi-0.2.4", 542 542 - ] 543 543 - 544 544 - [[package]] 545 545 - name = "gimli" 546 546 - version = "0.32.3" 547 547 - source = "registry+https://github.com/rust-lang/crates.io-index" 548 548 - checksum = "e629b9b98ef3dd8afe6ca2bd0f89306cec16d43d907889945bc5d6687f2f13c7" 549 549 - 550 550 - [[package]] 551 551 - name = "half" 552 552 - version = "2.7.0" 553 553 - source = "registry+https://github.com/rust-lang/crates.io-index" 554 554 - checksum = "e54c115d4f30f52c67202f079c5f9d8b49db4691f460fdb0b4c2e838261b2ba5" 555 555 - dependencies = [ 556 556 - "cfg-if", 557 557 - "crunchy", 558 558 - "zerocopy", 239 239 + "wasi", 559 240 ] 560 241 561 242 [[package]] 562 243 name = "hashbrown" 563 563 - version = "0.15.5" 244 244 + version = "0.14.5" 564 245 source = "registry+https://github.com/rust-lang/crates.io-index" 565 565 - checksum = "9229cfe53dfd69f0609a49f65461bd93001ea1ef889cd5529dd176593f5338a1" 566 566 - dependencies = [ 567 567 - "foldhash", 568 568 - ] 246 246 + checksum = "e5274423e17b7c9fc20b6e7e208532f9b19825d82dfd615708b70edd83df41f1" 569 247 570 248 [[package]] 571 571 - name = "hashlink" 572 572 - version = "0.10.0" 249 249 + name = "hashbrown" 250 250 + version = "0.16.1" 573 251 source = "registry+https://github.com/rust-lang/crates.io-index" 574 574 - checksum = "7382cf6263419f2d8df38c55d7da83da5c18aef87fc7a7fc1fb1e344edfe14c1" 575 575 - dependencies = [ 576 576 - "hashbrown", 577 577 - ] 252 252 + checksum = "841d1cc9bed7f9236f321df977030373f4a4163ae1a7dbfe1a51a2c1a51d9100" 578 253 579 254 [[package]] 580 255 name = "heck" ··· 583 258 checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea" 584 259 585 260 [[package]] 586 586 - name = "io-uring" 587 587 - version = "0.7.10" 588 588 - source = "registry+https://github.com/rust-lang/crates.io-index" 589 589 - checksum = "046fa2d4d00aea763528b4950358d0ead425372445dc8ff86312b3c69ff7727b" 590 590 - dependencies = [ 591 591 - "bitflags", 592 592 - "cfg-if", 593 593 - "libc", 594 594 - ] 595 595 - 596 596 - [[package]] 597 597 - name = "ipld-core" 598 598 - version = "0.4.2" 599 599 - source = "registry+https://github.com/rust-lang/crates.io-index" 600 600 - checksum = "104718b1cc124d92a6d01ca9c9258a7df311405debb3408c445a36452f9bf8db" 601 601 - dependencies = [ 602 602 - "cid", 603 603 - "serde", 604 604 - "serde_bytes", 605 605 - ] 606 606 - 607 607 - [[package]] 608 608 - name = "iroh-car" 609 609 - version = "0.5.1" 261 261 + name = "interval-heap" 262 262 + version = "0.0.5" 610 263 source = "registry+https://github.com/rust-lang/crates.io-index" 611 611 - checksum = "cb7f8cd4cb9aa083fba8b52e921764252d0b4dcb1cd6d120b809dbfe1106e81a" 264 264 + checksum = "11274e5e8e89b8607cfedc2910b6626e998779b48a019151c7604d0adcb86ac6" 612 265 dependencies = [ 613 613 - "anyhow", 614 614 - "cid", 615 615 - "futures", 616 616 - "serde", 617 617 - "serde_ipld_dagcbor", 618 618 - "thiserror 1.0.69", 619 619 - "tokio", 620 620 - "unsigned-varint 0.7.2", 266 266 + "compare", 621 267 ] 622 268 623 269 [[package]] ··· 627 273 checksum = "7943c866cc5cd64cbc25b2e01621d07fa8eb2a1a23160ee81ce38704e97b8ecf" 628 274 629 275 [[package]] 630 630 - name = "itertools" 631 631 - version = "0.13.0" 632 632 - source = "registry+https://github.com/rust-lang/crates.io-index" 633 633 - checksum = "413ee7dfc52ee1a4949ceeb7dbc8a33f2d6c088194d9f922fb8318faf1f01186" 634 634 - dependencies = [ 635 635 - "either", 636 636 - ] 637 637 - 638 638 - [[package]] 639 639 - name = "itoa" 640 640 - version = "1.0.15" 641 641 - source = "registry+https://github.com/rust-lang/crates.io-index" 642 642 - checksum = "4a5f13b858c8d314ee3e8f639011f7ccefe71f97f96e50151fb991f267928e2c" 643 643 - 644 644 - [[package]] 645 645 - name = "jiff" 646 646 - version = "0.2.15" 647 647 - source = "registry+https://github.com/rust-lang/crates.io-index" 648 648 - checksum = "be1f93b8b1eb69c77f24bbb0afdf66f54b632ee39af40ca21c4365a1d7347e49" 649 649 - dependencies = [ 650 650 - "jiff-static", 651 651 - "log", 652 652 - "portable-atomic", 653 653 - "portable-atomic-util", 654 654 - "serde", 655 655 - ] 656 656 - 657 657 - [[package]] 658 658 - name = "jiff-static" 659 659 - version = "0.2.15" 660 660 - source = "registry+https://github.com/rust-lang/crates.io-index" 661 661 - checksum = "03343451ff899767262ec32146f6d559dd759fdadf42ff0e227c7c48f72594b4" 662 662 - dependencies = [ 663 663 - "proc-macro2", 664 664 - "quote", 665 665 - "syn 2.0.106", 666 666 - ] 667 667 - 668 668 - [[package]] 669 669 - name = "js-sys" 670 670 - version = "0.3.81" 671 671 - source = "registry+https://github.com/rust-lang/crates.io-index" 672 672 - checksum = "ec48937a97411dcb524a265206ccd4c90bb711fca92b2792c407f268825b9305" 673 673 - dependencies = [ 674 674 - "once_cell", 675 675 - "wasm-bindgen", 676 676 - ] 677 677 - 678 678 - [[package]] 679 276 name = "libc" 680 277 version = "0.2.176" 681 278 source = "registry+https://github.com/rust-lang/crates.io-index" 682 279 checksum = "58f929b4d672ea937a23a1ab494143d968337a5f47e56d0815df1e0890ddf174" 683 683 - 684 684 - [[package]] 685 685 - name = "libsqlite3-sys" 686 686 - version = "0.35.0" 687 687 - source = "registry+https://github.com/rust-lang/crates.io-index" 688 688 - checksum = "133c182a6a2c87864fe97778797e46c7e999672690dc9fa3ee8e241aa4a9c13f" 689 689 - dependencies = [ 690 690 - "pkg-config", 691 691 - "vcpkg", 692 692 - ] 693 280 694 281 [[package]] 695 282 name = "linux-raw-sys" ··· 713 300 checksum = "34080505efa8e45a4b816c349525ebe327ceaa8559756f0356cba97ef3bf7432" 714 301 715 302 [[package]] 716 716 - name = "match-lookup" 717 717 - version = "0.1.1" 718 718 - source = "registry+https://github.com/rust-lang/crates.io-index" 719 719 - checksum = "1265724d8cb29dbbc2b0f06fffb8bf1a8c0cf73a78eede9ba73a4a66c52a981e" 720 720 - dependencies = [ 721 721 - "proc-macro2", 722 722 - "quote", 723 723 - "syn 1.0.109", 724 724 - ] 725 725 - 726 726 - [[package]] 727 727 - name = "memchr" 728 728 - version = "2.7.6" 729 729 - source = "registry+https://github.com/rust-lang/crates.io-index" 730 730 - checksum = "f52b00d39961fc5b2736ea853c9cc86238e165017a493d1d5c8eac6bdc4cc273" 731 731 - 732 732 - [[package]] 733 733 - name = "miniz_oxide" 734 734 - version = "0.8.9" 735 735 - source = "registry+https://github.com/rust-lang/crates.io-index" 736 736 - checksum = "1fa76a2c86f704bdb222d66965fb3d63269ce38518b83cb0575fca855ebb6316" 737 737 - dependencies = [ 738 738 - "adler2", 739 739 - ] 740 740 - 741 741 - [[package]] 742 742 - name = "mio" 743 743 - version = "1.0.4" 744 744 - source = "registry+https://github.com/rust-lang/crates.io-index" 745 745 - checksum = "78bed444cc8a2160f01cbcf811ef18cac863ad68ae8ca62092e8db51d51c761c" 746 746 - dependencies = [ 747 747 - "libc", 748 748 - "wasi 0.11.1+wasi-snapshot-preview1", 749 749 - "windows-sys 0.59.0", 750 750 - ] 751 751 - 752 752 - [[package]] 753 753 - name = "multibase" 754 754 - version = "0.9.2" 755 755 - source = "registry+https://github.com/rust-lang/crates.io-index" 756 756 - checksum = "8694bb4835f452b0e3bb06dbebb1d6fc5385b6ca1caf2e55fd165c042390ec77" 757 757 - dependencies = [ 758 758 - "base-x", 759 759 - "base256emoji", 760 760 - "data-encoding", 761 761 - "data-encoding-macro", 762 762 - ] 763 763 - 764 764 - [[package]] 765 765 - name = "multihash" 766 766 - version = "0.19.3" 303 303 + name = "lsm-tree" 304 304 + version = "3.0.1" 767 305 source = "registry+https://github.com/rust-lang/crates.io-index" 768 768 - checksum = "6b430e7953c29dd6a09afc29ff0bb69c6e306329ee6794700aee27b76a1aea8d" 306 306 + checksum = "b875f1dfe14f557f805b167fb9b0fc54c5560c7a4bd6ae02535b2846f276a8cb" 769 307 dependencies = [ 770 770 - "core2", 771 771 - "serde", 772 772 - "unsigned-varint 0.8.0", 308 308 + "byteorder-lite", 309 309 + "byteview", 310 310 + "crossbeam-skiplist", 311 311 + "enum_dispatch", 312 312 + "interval-heap", 313 313 + "log", 314 314 + "lz4_flex", 315 315 + "quick_cache", 316 316 + "rustc-hash", 317 317 + "self_cell", 318 318 + "sfa", 319 319 + "tempfile", 320 320 + "varint-rs", 321 321 + "xxhash-rust", 773 322 ] 774 323 775 324 [[package]] 776 776 - name = "num-traits" 777 777 - version = "0.2.19" 325 325 + name = "lz4_flex" 326 326 + version = "0.11.5" 778 327 source = "registry+https://github.com/rust-lang/crates.io-index" 779 779 - checksum = "071dfc062690e90b734c0b2273ce72ad0ffa95f0c74596bc250dcfd960262841" 328 328 + checksum = "08ab2867e3eeeca90e844d1940eab391c9dc5228783db2ed999acbc0a9ed375a" 780 329 dependencies = [ 781 781 - "autocfg", 782 782 - ] 783 783 - 784 784 - [[package]] 785 785 - name = "object" 786 786 - version = "0.37.3" 787 787 - source = "registry+https://github.com/rust-lang/crates.io-index" 788 788 - checksum = "ff76201f031d8863c38aa7f905eca4f53abbfa15f609db4277d44cd8938f33fe" 789 789 - dependencies = [ 790 790 - "memchr", 330 330 + "twox-hash", 791 331 ] 792 332 793 333 [[package]] ··· 803 343 checksum = "a4895175b425cb1f87721b59f0f286c2092bd4af812243672510e1ac53e2e0ad" 804 344 805 345 [[package]] 806 806 - name = "oorandom" 807 807 - version = "11.1.5" 808 808 - source = "registry+https://github.com/rust-lang/crates.io-index" 809 809 - checksum = "d6790f58c7ff633d8771f42965289203411a5e5c68388703c06e14f24770b41e" 810 810 - 811 811 - [[package]] 812 812 - name = "parking_lot" 813 813 - version = "0.12.5" 814 814 - source = "registry+https://github.com/rust-lang/crates.io-index" 815 815 - checksum = "93857453250e3077bd71ff98b6a65ea6621a19bb0f559a85248955ac12c45a1a" 816 816 - dependencies = [ 817 817 - "lock_api", 818 818 - "parking_lot_core", 819 819 - ] 820 820 - 821 821 - [[package]] 822 346 name = "parking_lot_core" 823 347 version = "0.9.12" 824 348 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 832 356 ] 833 357 834 358 [[package]] 835 835 - name = "pin-project-lite" 836 836 - version = "0.2.16" 837 837 - source = "registry+https://github.com/rust-lang/crates.io-index" 838 838 - checksum = "3b3cff922bd51709b605d9ead9aa71031d81447142d828eb4a6eba76fe619f9b" 839 839 - 840 840 - [[package]] 841 841 - name = "pin-utils" 842 842 - version = "0.1.0" 843 843 - source = "registry+https://github.com/rust-lang/crates.io-index" 844 844 - checksum = "8b870d8c151b6f2fb93e84a13146138f05d02ed11c7e7c54f8826aaaf7c9f184" 845 845 - 846 846 - [[package]] 847 847 - name = "pkg-config" 848 848 - version = "0.3.32" 849 849 - source = "registry+https://github.com/rust-lang/crates.io-index" 850 850 - checksum = "7edddbd0b52d732b21ad9a5fab5c704c14cd949e5e9a1ec5929a24fded1b904c" 851 851 - 852 852 - [[package]] 853 853 - name = "plotters" 854 854 - version = "0.3.7" 855 855 - source = "registry+https://github.com/rust-lang/crates.io-index" 856 856 - checksum = "5aeb6f403d7a4911efb1e33402027fc44f29b5bf6def3effcc22d7bb75f2b747" 857 857 - dependencies = [ 858 858 - "num-traits", 859 859 - "plotters-backend", 860 860 - "plotters-svg", 861 861 - "wasm-bindgen", 862 862 - "web-sys", 863 863 - ] 864 864 - 865 865 - [[package]] 866 866 - name = "plotters-backend" 867 867 - version = "0.3.7" 868 868 - source = "registry+https://github.com/rust-lang/crates.io-index" 869 869 - checksum = "df42e13c12958a16b3f7f4386b9ab1f3e7933914ecea48da7139435263a4172a" 870 870 - 871 871 - [[package]] 872 872 - name = "plotters-svg" 873 873 - version = "0.3.7" 874 874 - source = "registry+https://github.com/rust-lang/crates.io-index" 875 875 - checksum = "51bae2ac328883f7acdfea3d66a7c35751187f870bc81f94563733a154d7a670" 876 876 - dependencies = [ 877 877 - "plotters-backend", 878 878 - ] 879 879 - 880 880 - [[package]] 881 881 - name = "portable-atomic" 882 882 - version = "1.11.1" 883 883 - source = "registry+https://github.com/rust-lang/crates.io-index" 884 884 - checksum = "f84267b20a16ea918e43c6a88433c2d54fa145c92a811b5b047ccbe153674483" 885 885 - 886 886 - [[package]] 887 887 - name = "portable-atomic-util" 888 888 - version = "0.2.4" 359 359 + name = "proc-macro2" 360 360 + version = "1.0.101" 889 361 source = "registry+https://github.com/rust-lang/crates.io-index" 890 890 - checksum = "d8a2f0d8d040d7848a709caf78912debcc3f33ee4b3cac47d73d1e1069e83507" 362 362 + checksum = "89ae43fd86e4158d6db51ad8e2b80f313af9cc74f5c0e03ccb87de09998732de" 891 363 dependencies = [ 892 892 - "portable-atomic", 364 364 + "unicode-ident", 893 365 ] 894 366 895 367 [[package]] 896 896 - name = "proc-macro2" 897 897 - version = "1.0.101" 368 368 + name = "quick_cache" 369 369 + version = "0.6.18" 898 370 source = "registry+https://github.com/rust-lang/crates.io-index" 899 899 - checksum = "89ae43fd86e4158d6db51ad8e2b80f313af9cc74f5c0e03ccb87de09998732de" 371 371 + checksum = "7ada44a88ef953a3294f6eb55d2007ba44646015e18613d2f213016379203ef3" 900 372 dependencies = [ 901 901 - "unicode-ident", 373 373 + "equivalent", 374 374 + "hashbrown 0.16.1", 902 375 ] 903 376 904 377 [[package]] ··· 917 390 checksum = "69cdb34c158ceb288df11e18b4bd39de994f6657d83847bdffdbd7f346754b0f" 918 391 919 392 [[package]] 920 920 - name = "rayon" 921 921 - version = "1.11.0" 922 922 - source = "registry+https://github.com/rust-lang/crates.io-index" 923 923 - checksum = "368f01d005bf8fd9b1206fb6fa653e6c4a81ceb1466406b81792d87c5677a58f" 924 924 - dependencies = [ 925 925 - "either", 926 926 - "rayon-core", 927 927 - ] 928 928 - 929 929 - [[package]] 930 930 - name = "rayon-core" 931 931 - version = "1.13.0" 932 932 - source = "registry+https://github.com/rust-lang/crates.io-index" 933 933 - checksum = "22e18b0f0062d30d4230b2e85ff77fdfe4326feb054b9783a3460d8435c8ab91" 934 934 - dependencies = [ 935 935 - "crossbeam-deque", 936 936 - "crossbeam-utils", 937 937 - ] 938 938 - 939 939 - [[package]] 940 393 name = "redox_syscall" 941 394 version = "0.5.18" 942 395 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 946 399 ] 947 400 948 401 [[package]] 949 949 - name = "regex" 950 950 - version = "1.11.3" 951 951 - source = "registry+https://github.com/rust-lang/crates.io-index" 952 952 - checksum = "8b5288124840bee7b386bc413c487869b360b2b4ec421ea56425128692f2a82c" 953 953 - dependencies = [ 954 954 - "aho-corasick", 955 955 - "memchr", 956 956 - "regex-automata", 957 957 - "regex-syntax", 958 958 - ] 959 959 - 960 960 - [[package]] 961 961 - name = "regex-automata" 962 962 - version = "0.4.11" 963 963 - source = "registry+https://github.com/rust-lang/crates.io-index" 964 964 - checksum = "833eb9ce86d40ef33cb1306d8accf7bc8ec2bfea4355cbdebb3df68b40925cad" 965 965 - dependencies = [ 966 966 - "aho-corasick", 967 967 - "memchr", 968 968 - "regex-syntax", 969 969 - ] 970 970 - 971 971 - [[package]] 972 972 - name = "regex-syntax" 973 973 - version = "0.8.6" 974 974 - source = "registry+https://github.com/rust-lang/crates.io-index" 975 975 - checksum = "caf4aa5b0f434c91fe5c7f1ecb6a5ece2130b02ad2a590589dda5146df959001" 976 976 - 977 977 - [[package]] 978 402 name = "repo-stream" 979 979 - version = "0.1.1" 403 403 + version = "0.2.2" 980 404 dependencies = [ 981 981 - "bincode", 982 405 "clap", 983 983 - "criterion", 984 984 - "env_logger", 985 985 - "futures", 986 986 - "futures-core", 987 987 - "ipld-core", 988 988 - "iroh-car", 989 989 - "log", 990 990 - "multibase", 991 991 - "rusqlite", 992 992 - "serde", 993 993 - "serde_bytes", 994 994 - "serde_ipld_dagcbor", 995 995 - "tempfile", 996 996 - "thiserror 2.0.17", 997 997 - "tokio", 406 406 + "fjall", 998 407 ] 999 408 1000 409 [[package]] 1001 1001 - name = "rusqlite" 1002 1002 - version = "0.37.0" 410 410 + name = "rustc-hash" 411 411 + version = "2.1.1" 1003 412 source = "registry+https://github.com/rust-lang/crates.io-index" 1004 1004 - checksum = "165ca6e57b20e1351573e3729b958bc62f0e48025386970b6e4d29e7a7e71f3f" 1005 1005 - dependencies = [ 1006 1006 - "bitflags", 1007 1007 - "fallible-iterator", 1008 1008 - "fallible-streaming-iterator", 1009 1009 - "hashlink", 1010 1010 - "libsqlite3-sys", 1011 1011 - "smallvec", 1012 1012 - ] 1013 1013 - 1014 1014 - [[package]] 1015 1015 - name = "rustc-demangle" 1016 1016 - version = "0.1.26" 1017 1017 - source = "registry+https://github.com/rust-lang/crates.io-index" 1018 1018 - checksum = "56f7d92ca342cea22a06f2121d944b4fd82af56988c270852495420f961d4ace" 413 413 + checksum = "357703d41365b4b27c590e3ed91eabb1b663f07c4c084095e60cbed4362dff0d" 1019 414 1020 415 [[package]] 1021 416 name = "rustix" ··· 1027 422 "errno", 1028 423 "libc", 1029 424 "linux-raw-sys", 1030 1030 - "windows-sys 0.60.2", 1031 1031 - ] 1032 1032 - 1033 1033 - [[package]] 1034 1034 - name = "rustversion" 1035 1035 - version = "1.0.22" 1036 1036 - source = "registry+https://github.com/rust-lang/crates.io-index" 1037 1037 - checksum = "b39cdef0fa800fc44525c84ccb54a029961a8215f9619753635a9c0d2538d46d" 1038 1038 - 1039 1039 - [[package]] 1040 1040 - name = "ryu" 1041 1041 - version = "1.0.20" 1042 1042 - source = "registry+https://github.com/rust-lang/crates.io-index" 1043 1043 - checksum = "28d3b2b1366ec20994f1fd18c3c594f05c5dd4bc44d8bb0c1c632c8d6829481f" 1044 1044 - 1045 1045 - [[package]] 1046 1046 - name = "same-file" 1047 1047 - version = "1.0.6" 1048 1048 - source = "registry+https://github.com/rust-lang/crates.io-index" 1049 1049 - checksum = "93fc1dc3aaa9bfed95e02e6eadabb4baf7e3078b0bd1b4d7b6b0b68378900502" 1050 1050 - dependencies = [ 1051 1051 - "winapi-util", 425 425 + "windows-sys", 1052 426 ] 1053 427 1054 428 [[package]] ··· 1058 432 checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49" 1059 433 1060 434 [[package]] 1061 1061 - name = "serde" 1062 1062 - version = "1.0.228" 1063 1063 - source = "registry+https://github.com/rust-lang/crates.io-index" 1064 1064 - checksum = "9a8e94ea7f378bd32cbbd37198a4a91436180c5bb472411e48b5ec2e2124ae9e" 1065 1065 - dependencies = [ 1066 1066 - "serde_core", 1067 1067 - "serde_derive", 1068 1068 - ] 1069 1069 - 1070 1070 - [[package]] 1071 1071 - name = "serde_bytes" 1072 1072 - version = "0.11.19" 1073 1073 - source = "registry+https://github.com/rust-lang/crates.io-index" 1074 1074 - checksum = "a5d440709e79d88e51ac01c4b72fc6cb7314017bb7da9eeff678aa94c10e3ea8" 1075 1075 - dependencies = [ 1076 1076 - "serde", 1077 1077 - "serde_core", 1078 1078 - ] 1079 1079 - 1080 1080 - [[package]] 1081 1081 - name = "serde_core" 1082 1082 - version = "1.0.228" 1083 1083 - source = "registry+https://github.com/rust-lang/crates.io-index" 1084 1084 - checksum = "41d385c7d4ca58e59fc732af25c3983b67ac852c1a25000afe1175de458b67ad" 1085 1085 - dependencies = [ 1086 1086 - "serde_derive", 1087 1087 - ] 1088 1088 - 1089 1089 - [[package]] 1090 1090 - name = "serde_derive" 1091 1091 - version = "1.0.228" 435 435 + name = "self_cell" 436 436 + version = "1.2.2" 1092 437 source = "registry+https://github.com/rust-lang/crates.io-index" 1093 1093 - checksum = "d540f220d3187173da220f885ab66608367b6574e925011a9353e4badda91d79" 1094 1094 - dependencies = [ 1095 1095 - "proc-macro2", 1096 1096 - "quote", 1097 1097 - "syn 2.0.106", 1098 1098 - ] 438 438 + checksum = "b12e76d157a900eb52e81bc6e9f3069344290341720e9178cde2407113ac8d89" 1099 439 1100 440 [[package]] 1101 1101 - name = "serde_ipld_dagcbor" 1102 1102 - version = "0.6.4" 1103 1103 - source = "registry+https://github.com/rust-lang/crates.io-index" 1104 1104 - checksum = "46182f4f08349a02b45c998ba3215d3f9de826246ba02bb9dddfe9a2a2100778" 1105 1105 - dependencies = [ 1106 1106 - "cbor4ii", 1107 1107 - "ipld-core", 1108 1108 - "scopeguard", 1109 1109 - "serde", 1110 1110 - ] 1111 1111 - 1112 1112 - [[package]] 1113 1113 - name = "serde_json" 1114 1114 - version = "1.0.145" 1115 1115 - source = "registry+https://github.com/rust-lang/crates.io-index" 1116 1116 - checksum = "402a6f66d8c709116cf22f558eab210f5a50187f702eb4d7e5ef38d9a7f1c79c" 1117 1117 - dependencies = [ 1118 1118 - "itoa", 1119 1119 - "memchr", 1120 1120 - "ryu", 1121 1121 - "serde", 1122 1122 - "serde_core", 1123 1123 - ] 1124 1124 - 1125 1125 - [[package]] 1126 1126 - name = "signal-hook-registry" 1127 1127 - version = "1.4.6" 441 441 + name = "sfa" 442 442 + version = "1.0.0" 1128 443 source = "registry+https://github.com/rust-lang/crates.io-index" 1129 1129 - checksum = "b2a4719bff48cee6b39d12c020eeb490953ad2443b7055bd0b21fca26bd8c28b" 444 444 + checksum = "a1296838937cab56cd6c4eeeb8718ec777383700c33f060e2869867bd01d1175" 1130 445 dependencies = [ 1131 1131 - "libc", 446 446 + "byteorder-lite", 447 447 + "log", 448 448 + "xxhash-rust", 1132 449 ] 1133 450 1134 451 [[package]] 1135 1135 - name = "slab" 1136 1136 - version = "0.4.11" 1137 1137 - source = "registry+https://github.com/rust-lang/crates.io-index" 1138 1138 - checksum = "7a2ae44ef20feb57a68b23d846850f861394c2e02dc425a50098ae8c90267589" 1139 1139 - 1140 1140 - [[package]] 1141 452 name = "smallvec" 1142 453 version = "1.15.1" 1143 454 source = "registry+https://github.com/rust-lang/crates.io-index" 1144 455 checksum = "67b1b7a3b5fe4f1376887184045fcf45c69e92af734b7aaddc05fb777b6fbd03" 1145 456 1146 457 [[package]] 1147 1147 - name = "socket2" 1148 1148 - version = "0.6.0" 458 458 + name = "spin" 459 459 + version = "0.9.8" 1149 460 source = "registry+https://github.com/rust-lang/crates.io-index" 1150 1150 - checksum = "233504af464074f9d066d7b5416c5f9b894a5862a6506e306f7b816cdd6f1807" 461 461 + checksum = "6980e8d7511241f8acf4aebddbb1ff938df5eebe98691418c4468d0b72a96a67" 1151 462 dependencies = [ 1152 1152 - "libc", 1153 1153 - "windows-sys 0.59.0", 463 463 + "lock_api", 1154 464 ] 1155 465 1156 466 [[package]] ··· 1161 471 1162 472 [[package]] 1163 473 name = "syn" 1164 1164 - version = "1.0.109" 1165 1165 - source = "registry+https://github.com/rust-lang/crates.io-index" 1166 1166 - checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237" 1167 1167 - dependencies = [ 1168 1168 - "proc-macro2", 1169 1169 - "quote", 1170 1170 - "unicode-ident", 1171 1171 - ] 1172 1172 - 1173 1173 - [[package]] 1174 1174 - name = "syn" 1175 474 version = "2.0.106" 1176 475 source = "registry+https://github.com/rust-lang/crates.io-index" 1177 476 checksum = "ede7c438028d4436d71104916910f5bb611972c5cfd7f89b8300a8186e6fada6" ··· 1191 490 "getrandom", 1192 491 "once_cell", 1193 492 "rustix", 1194 1194 - "windows-sys 0.60.2", 493 493 + "windows-sys", 1195 494 ] 1196 495 1197 496 [[package]] 1198 1198 - name = "thiserror" 1199 1199 - version = "1.0.69" 497 497 + name = "twox-hash" 498 498 + version = "2.1.2" 1200 499 source = "registry+https://github.com/rust-lang/crates.io-index" 1201 1201 - checksum = "b6aaf5339b578ea85b50e080feb250a3e8ae8cfcdff9a461c9ec2904bc923f52" 1202 1202 - dependencies = [ 1203 1203 - "thiserror-impl 1.0.69", 1204 1204 - ] 1205 1205 - 1206 1206 - [[package]] 1207 1207 - name = "thiserror" 1208 1208 - version = "2.0.17" 1209 1209 - source = "registry+https://github.com/rust-lang/crates.io-index" 1210 1210 - checksum = "f63587ca0f12b72a0600bcba1d40081f830876000bb46dd2337a3051618f4fc8" 1211 1211 - dependencies = [ 1212 1212 - "thiserror-impl 2.0.17", 1213 1213 - ] 1214 1214 - 1215 1215 - [[package]] 1216 1216 - name = "thiserror-impl" 1217 1217 - version = "1.0.69" 1218 1218 - source = "registry+https://github.com/rust-lang/crates.io-index" 1219 1219 - checksum = "4fee6c4efc90059e10f81e6d42c60a18f76588c3d74cb83a0b242a2b6c7504c1" 1220 1220 - dependencies = [ 1221 1221 - "proc-macro2", 1222 1222 - "quote", 1223 1223 - "syn 2.0.106", 1224 1224 - ] 1225 1225 - 1226 1226 - [[package]] 1227 1227 - name = "thiserror-impl" 1228 1228 - version = "2.0.17" 1229 1229 - source = "registry+https://github.com/rust-lang/crates.io-index" 1230 1230 - checksum = "3ff15c8ecd7de3849db632e14d18d2571fa09dfc5ed93479bc4485c7a517c913" 1231 1231 - dependencies = [ 1232 1232 - "proc-macro2", 1233 1233 - "quote", 1234 1234 - "syn 2.0.106", 1235 1235 - ] 1236 1236 - 1237 1237 - [[package]] 1238 1238 - name = "tinytemplate" 1239 1239 - version = "1.2.1" 1240 1240 - source = "registry+https://github.com/rust-lang/crates.io-index" 1241 1241 - checksum = "be4d6b5f19ff7664e8c98d03e2139cb510db9b0a60b55f8e8709b689d939b6bc" 1242 1242 - dependencies = [ 1243 1243 - "serde", 1244 1244 - "serde_json", 1245 1245 - ] 1246 1246 - 1247 1247 - [[package]] 1248 1248 - name = "tokio" 1249 1249 - version = "1.47.1" 1250 1250 - source = "registry+https://github.com/rust-lang/crates.io-index" 1251 1251 - checksum = "89e49afdadebb872d3145a5638b59eb0691ea23e46ca484037cfab3b76b95038" 1252 1252 - dependencies = [ 1253 1253 - "backtrace", 1254 1254 - "bytes", 1255 1255 - "io-uring", 1256 1256 - "libc", 1257 1257 - "mio", 1258 1258 - "parking_lot", 1259 1259 - "pin-project-lite", 1260 1260 - "signal-hook-registry", 1261 1261 - "slab", 1262 1262 - "socket2", 1263 1263 - "tokio-macros", 1264 1264 - "windows-sys 0.59.0", 1265 1265 - ] 1266 1266 - 1267 1267 - [[package]] 1268 1268 - name = "tokio-macros" 1269 1269 - version = "2.5.0" 1270 1270 - source = "registry+https://github.com/rust-lang/crates.io-index" 1271 1271 - checksum = "6e06d43f1345a3bcd39f6a56dbb7dcab2ba47e68e8ac134855e7e2bdbaf8cab8" 1272 1272 - dependencies = [ 1273 1273 - "proc-macro2", 1274 1274 - "quote", 1275 1275 - "syn 2.0.106", 1276 1276 - ] 500 500 + checksum = "9ea3136b675547379c4bd395ca6b938e5ad3c3d20fad76e7fe85f9e0d011419c" 1277 501 1278 502 [[package]] 1279 503 name = "unicode-ident" ··· 1282 506 checksum = "f63a545481291138910575129486daeaf8ac54aee4387fe7906919f7830c7d9d" 1283 507 1284 508 [[package]] 1285 1285 - name = "unsigned-varint" 1286 1286 - version = "0.7.2" 1287 1287 - source = "registry+https://github.com/rust-lang/crates.io-index" 1288 1288 - checksum = "6889a77d49f1f013504cec6bf97a2c730394adedaeb1deb5ea08949a50541105" 1289 1289 - 1290 1290 - [[package]] 1291 1291 - name = "unsigned-varint" 1292 1292 - version = "0.8.0" 1293 1293 - source = "registry+https://github.com/rust-lang/crates.io-index" 1294 1294 - checksum = "eb066959b24b5196ae73cb057f45598450d2c5f71460e98c49b738086eff9c06" 1295 1295 - 1296 1296 - [[package]] 1297 1297 - name = "unty" 1298 1298 - version = "0.0.4" 1299 1299 - source = "registry+https://github.com/rust-lang/crates.io-index" 1300 1300 - checksum = "6d49784317cd0d1ee7ec5c716dd598ec5b4483ea832a2dced265471cc0f690ae" 1301 1301 - 1302 1302 - [[package]] 1303 509 name = "utf8parse" 1304 510 version = "0.2.2" 1305 511 source = "registry+https://github.com/rust-lang/crates.io-index" 1306 512 checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821" 1307 513 1308 514 [[package]] 1309 1309 - name = "vcpkg" 1310 1310 - version = "0.2.15" 1311 1311 - source = "registry+https://github.com/rust-lang/crates.io-index" 1312 1312 - checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426" 1313 1313 - 1314 1314 - [[package]] 1315 1315 - name = "virtue" 1316 1316 - version = "0.0.18" 1317 1317 - source = "registry+https://github.com/rust-lang/crates.io-index" 1318 1318 - checksum = "051eb1abcf10076295e815102942cc58f9d5e3b4560e46e53c21e8ff6f3af7b1" 1319 1319 - 1320 1320 - [[package]] 1321 1321 - name = "walkdir" 1322 1322 - version = "2.5.0" 1323 1323 - source = "registry+https://github.com/rust-lang/crates.io-index" 1324 1324 - checksum = "29790946404f91d9c5d06f9874efddea1dc06c5efe94541a7d6863108e3a5e4b" 1325 1325 - dependencies = [ 1326 1326 - "same-file", 1327 1327 - "winapi-util", 1328 1328 - ] 1329 1329 - 1330 1330 - [[package]] 1331 1331 - name = "wasi" 1332 1332 - version = "0.11.1+wasi-snapshot-preview1" 515 515 + name = "varint-rs" 516 516 + version = "2.2.0" 1333 517 source = "registry+https://github.com/rust-lang/crates.io-index" 1334 1334 - checksum = "ccf3ec651a847eb01de73ccad15eb7d99f80485de043efb2f370cd654f4ea44b" 518 518 + checksum = "8f54a172d0620933a27a4360d3db3e2ae0dd6cceae9730751a036bbf182c4b23" 1335 519 1336 520 [[package]] 1337 521 name = "wasi" ··· 1352 536 ] 1353 537 1354 538 [[package]] 1355 1355 - name = "wasm-bindgen" 1356 1356 - version = "0.2.104" 1357 1357 - source = "registry+https://github.com/rust-lang/crates.io-index" 1358 1358 - checksum = "c1da10c01ae9f1ae40cbfac0bac3b1e724b320abfcf52229f80b547c0d250e2d" 1359 1359 - dependencies = [ 1360 1360 - "cfg-if", 1361 1361 - "once_cell", 1362 1362 - "rustversion", 1363 1363 - "wasm-bindgen-macro", 1364 1364 - "wasm-bindgen-shared", 1365 1365 - ] 1366 1366 - 1367 1367 - [[package]] 1368 1368 - name = "wasm-bindgen-backend" 1369 1369 - version = "0.2.104" 1370 1370 - source = "registry+https://github.com/rust-lang/crates.io-index" 1371 1371 - checksum = "671c9a5a66f49d8a47345ab942e2cb93c7d1d0339065d4f8139c486121b43b19" 1372 1372 - dependencies = [ 1373 1373 - "bumpalo", 1374 1374 - "log", 1375 1375 - "proc-macro2", 1376 1376 - "quote", 1377 1377 - "syn 2.0.106", 1378 1378 - "wasm-bindgen-shared", 1379 1379 - ] 1380 1380 - 1381 1381 - [[package]] 1382 1382 - name = "wasm-bindgen-macro" 1383 1383 - version = "0.2.104" 1384 1384 - source = "registry+https://github.com/rust-lang/crates.io-index" 1385 1385 - checksum = "7ca60477e4c59f5f2986c50191cd972e3a50d8a95603bc9434501cf156a9a119" 1386 1386 - dependencies = [ 1387 1387 - "quote", 1388 1388 - "wasm-bindgen-macro-support", 1389 1389 - ] 1390 1390 - 1391 1391 - [[package]] 1392 1392 - name = "wasm-bindgen-macro-support" 1393 1393 - version = "0.2.104" 1394 1394 - source = "registry+https://github.com/rust-lang/crates.io-index" 1395 1395 - checksum = "9f07d2f20d4da7b26400c9f4a0511e6e0345b040694e8a75bd41d578fa4421d7" 1396 1396 - dependencies = [ 1397 1397 - "proc-macro2", 1398 1398 - "quote", 1399 1399 - "syn 2.0.106", 1400 1400 - "wasm-bindgen-backend", 1401 1401 - "wasm-bindgen-shared", 1402 1402 - ] 1403 1403 - 1404 1404 - [[package]] 1405 1405 - name = "wasm-bindgen-shared" 1406 1406 - version = "0.2.104" 1407 1407 - source = "registry+https://github.com/rust-lang/crates.io-index" 1408 1408 - checksum = "bad67dc8b2a1a6e5448428adec4c3e84c43e561d8c9ee8a9e5aabeb193ec41d1" 1409 1409 - dependencies = [ 1410 1410 - "unicode-ident", 1411 1411 - ] 1412 1412 - 1413 1413 - [[package]] 1414 1414 - name = "web-sys" 1415 1415 - version = "0.3.81" 1416 1416 - source = "registry+https://github.com/rust-lang/crates.io-index" 1417 1417 - checksum = "9367c417a924a74cae129e6a2ae3b47fabb1f8995595ab474029da749a8be120" 1418 1418 - dependencies = [ 1419 1419 - "js-sys", 1420 1420 - "wasm-bindgen", 1421 1421 - ] 1422 1422 - 1423 1423 - [[package]] 1424 1424 - name = "winapi-util" 1425 1425 - version = "0.1.11" 1426 1426 - source = "registry+https://github.com/rust-lang/crates.io-index" 1427 1427 - checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22" 1428 1428 - dependencies = [ 1429 1429 - "windows-sys 0.60.2", 1430 1430 - ] 1431 1431 - 1432 1432 - [[package]] 1433 539 name = "windows-link" 1434 540 version = "0.2.1" 1435 541 source = "registry+https://github.com/rust-lang/crates.io-index" ··· 1437 543 1438 544 [[package]] 1439 545 name = "windows-sys" 1440 1440 - version = "0.59.0" 1441 1441 - source = "registry+https://github.com/rust-lang/crates.io-index" 1442 1442 - checksum = "1e38bc4d79ed67fd075bcc251a1c39b32a1776bbe92e5bef1f0bf1f8c531853b" 1443 1443 - dependencies = [ 1444 1444 - "windows-targets 0.52.6", 1445 1445 - ] 1446 1446 - 1447 1447 - [[package]] 1448 1448 - name = "windows-sys" 1449 546 version = "0.60.2" 1450 547 source = "registry+https://github.com/rust-lang/crates.io-index" 1451 548 checksum = "f2f500e4d28234f72040990ec9d39e3a6b950f9f22d3dba18416c35882612bcb" 1452 549 dependencies = [ 1453 1453 - "windows-targets 0.53.5", 1454 1454 - ] 1455 1455 - 1456 1456 - [[package]] 1457 1457 - name = "windows-targets" 1458 1458 - version = "0.52.6" 1459 1459 - source = "registry+https://github.com/rust-lang/crates.io-index" 1460 1460 - checksum = "9b724f72796e036ab90c1021d4780d4d3d648aca59e491e6b98e725b84e99973" 1461 1461 - dependencies = [ 1462 1462 - "windows_aarch64_gnullvm 0.52.6", 1463 1463 - "windows_aarch64_msvc 0.52.6", 1464 1464 - "windows_i686_gnu 0.52.6", 1465 1465 - "windows_i686_gnullvm 0.52.6", 1466 1466 - "windows_i686_msvc 0.52.6", 1467 1467 - "windows_x86_64_gnu 0.52.6", 1468 1468 - "windows_x86_64_gnullvm 0.52.6", 1469 1469 - "windows_x86_64_msvc 0.52.6", 550 550 + "windows-targets", 1470 551 ] 1471 552 1472 553 [[package]] ··· 1476 557 checksum = "4945f9f551b88e0d65f3db0bc25c33b8acea4d9e41163edf90dcd0b19f9069f3" 1477 558 dependencies = [ 1478 559 "windows-link", 1479 1479 - "windows_aarch64_gnullvm 0.53.1", 1480 1480 - "windows_aarch64_msvc 0.53.1", 1481 1481 - "windows_i686_gnu 0.53.1", 1482 1482 - "windows_i686_gnullvm 0.53.1", 1483 1483 - "windows_i686_msvc 0.53.1", 1484 1484 - "windows_x86_64_gnu 0.53.1", 1485 1485 - "windows_x86_64_gnullvm 0.53.1", 1486 1486 - "windows_x86_64_msvc 0.53.1", 560 560 + "windows_aarch64_gnullvm", 561 561 + "windows_aarch64_msvc", 562 562 + "windows_i686_gnu", 563 563 + "windows_i686_gnullvm", 564 564 + "windows_i686_msvc", 565 565 + "windows_x86_64_gnu", 566 566 + "windows_x86_64_gnullvm", 567 567 + "windows_x86_64_msvc", 1487 568 ] 1488 569 1489 570 [[package]] 1490 571 name = "windows_aarch64_gnullvm" 1491 1491 - version = "0.52.6" 1492 1492 - source = "registry+https://github.com/rust-lang/crates.io-index" 1493 1493 - checksum = "32a4622180e7a0ec044bb555404c800bc9fd9ec262ec147edd5989ccd0c02cd3" 1494 1494 - 1495 1495 - [[package]] 1496 1496 - name = "windows_aarch64_gnullvm" 1497 572 version = "0.53.1" 1498 573 source = "registry+https://github.com/rust-lang/crates.io-index" 1499 574 checksum = "a9d8416fa8b42f5c947f8482c43e7d89e73a173cead56d044f6a56104a6d1b53" 1500 575 1501 576 [[package]] 1502 577 name = "windows_aarch64_msvc" 1503 1503 - version = "0.52.6" 1504 1504 - source = "registry+https://github.com/rust-lang/crates.io-index" 1505 1505 - checksum = "09ec2a7bb152e2252b53fa7803150007879548bc709c039df7627cabbd05d469" 1506 1506 - 1507 1507 - [[package]] 1508 1508 - name = "windows_aarch64_msvc" 1509 578 version = "0.53.1" 1510 579 source = "registry+https://github.com/rust-lang/crates.io-index" 1511 580 checksum = "b9d782e804c2f632e395708e99a94275910eb9100b2114651e04744e9b125006" 1512 581 1513 582 [[package]] 1514 583 name = "windows_i686_gnu" 1515 1515 - version = "0.52.6" 1516 1516 - source = "registry+https://github.com/rust-lang/crates.io-index" 1517 1517 - checksum = "8e9b5ad5ab802e97eb8e295ac6720e509ee4c243f69d781394014ebfe8bbfa0b" 1518 1518 - 1519 1519 - [[package]] 1520 1520 - name = "windows_i686_gnu" 1521 584 version = "0.53.1" 1522 585 source = "registry+https://github.com/rust-lang/crates.io-index" 1523 586 checksum = "960e6da069d81e09becb0ca57a65220ddff016ff2d6af6a223cf372a506593a3" 1524 587 1525 588 [[package]] 1526 589 name = "windows_i686_gnullvm" 1527 1527 - version = "0.52.6" 1528 1528 - source = "registry+https://github.com/rust-lang/crates.io-index" 1529 1529 - checksum = "0eee52d38c090b3caa76c563b86c3a4bd71ef1a819287c19d586d7334ae8ed66" 1530 1530 - 1531 1531 - [[package]] 1532 1532 - name = "windows_i686_gnullvm" 1533 590 version = "0.53.1" 1534 591 source = "registry+https://github.com/rust-lang/crates.io-index" 1535 592 checksum = "fa7359d10048f68ab8b09fa71c3daccfb0e9b559aed648a8f95469c27057180c" 1536 1536 - 1537 1537 - [[package]] 1538 1538 - name = "windows_i686_msvc" 1539 1539 - version = "0.52.6" 1540 1540 - source = "registry+https://github.com/rust-lang/crates.io-index" 1541 1541 - checksum = "240948bc05c5e7c6dabba28bf89d89ffce3e303022809e73deaefe4f6ec56c66" 1542 593 1543 594 [[package]] 1544 595 name = "windows_i686_msvc" ··· 1548 599 1549 600 [[package]] 1550 601 name = "windows_x86_64_gnu" 1551 1551 - version = "0.52.6" 1552 1552 - source = "registry+https://github.com/rust-lang/crates.io-index" 1553 1553 - checksum = "147a5c80aabfbf0c7d901cb5895d1de30ef2907eb21fbbab29ca94c5b08b1a78" 1554 1554 - 1555 1555 - [[package]] 1556 1556 - name = "windows_x86_64_gnu" 1557 602 version = "0.53.1" 1558 603 source = "registry+https://github.com/rust-lang/crates.io-index" 1559 604 checksum = "9c3842cdd74a865a8066ab39c8a7a473c0778a3f29370b5fd6b4b9aa7df4a499" 1560 605 1561 606 [[package]] 1562 607 name = "windows_x86_64_gnullvm" 1563 1563 - version = "0.52.6" 1564 1564 - source = "registry+https://github.com/rust-lang/crates.io-index" 1565 1565 - checksum = "24d5b23dc417412679681396f2b49f3de8c1473deb516bd34410872eff51ed0d" 1566 1566 - 1567 1567 - [[package]] 1568 1568 - name = "windows_x86_64_gnullvm" 1569 608 version = "0.53.1" 1570 609 source = "registry+https://github.com/rust-lang/crates.io-index" 1571 610 checksum = "0ffa179e2d07eee8ad8f57493436566c7cc30ac536a3379fdf008f47f6bb7ae1" 1572 1572 - 1573 1573 - [[package]] 1574 1574 - name = "windows_x86_64_msvc" 1575 1575 - version = "0.52.6" 1576 1576 - source = "registry+https://github.com/rust-lang/crates.io-index" 1577 1577 - checksum = "589f6da84c646204747d1270a2a5661ea66ed1cced2631d546fdfb155959f9ec" 1578 611 1579 612 [[package]] 1580 613 name = "windows_x86_64_msvc" ··· 1589 622 checksum = "f17a85883d4e6d00e8a97c586de764dabcc06133f7f1d55dce5cdc070ad7fe59" 1590 623 1591 624 [[package]] 1592 1592 - name = "zerocopy" 1593 1593 - version = "0.8.27" 625 625 + name = "xxhash-rust" 626 626 + version = "0.8.15" 1594 627 source = "registry+https://github.com/rust-lang/crates.io-index" 1595 1595 - checksum = "0894878a5fa3edfd6da3f88c4805f4c8558e2b996227a3d864f47fe11e38282c" 1596 1596 - dependencies = [ 1597 1597 - "zerocopy-derive", 1598 1598 - ] 1599 1599 - 1600 1600 - [[package]] 1601 1601 - name = "zerocopy-derive" 1602 1602 - version = "0.8.27" 1603 1603 - source = "registry+https://github.com/rust-lang/crates.io-index" 1604 1604 - checksum = "88d2b8d9c68ad2b9e4340d7832716a4d21a22a1154777ad56ea55c51a9cf3831" 1605 1605 - dependencies = [ 1606 1606 - "proc-macro2", 1607 1607 - "quote", 1608 1608 - "syn 2.0.106", 1609 1609 - ] 628 628 + checksum = "fdd20c5420375476fbd4394763288da7eb0cc0b8c11deed431a91562af7335d3"

+3 -36

Cargo.toml

··· 1 1 [package] 2 2 name = "repo-stream" 3 3 - version = "0.1.1" 3 3 + version = "0.2.2" 4 4 edition = "2024" 5 5 license = "MIT OR Apache-2.0" 6 6 - description = "Fast and robust atproto CAR file processing in rust" 6 6 + description = "A robust CAR file -> MST walker for atproto" 7 7 repository = "https://tangled.org/@microcosm.blue/repo-stream" 8 8 9 9 [dependencies] 10 10 - bincode = { version = "2.0.1", features = ["serde"] } 11 11 - futures = "0.3.31" 12 12 - futures-core = "0.3.31" 13 13 - ipld-core = { version = "0.4.2", features = ["serde"] } 14 14 - iroh-car = "0.5.1" 15 15 - log = "0.4.28" 16 16 - 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" 21 21 - thiserror = "2.0.17" 22 22 - tokio = { version = "1.47.1", features = ["rt", "sync"] } 23 23 - 24 24 - [dev-dependencies] 10 10 + fjall = "3.0.1" 25 11 clap = { version = "4.5.48", features = ["derive"] } 26 26 - criterion = { version = "0.7.0", features = ["async_tokio"] } 27 27 - env_logger = "0.11.8" 28 28 - multibase = "0.9.2" 29 29 - tempfile = "3.23.0" 30 30 - tokio = { version = "1.47.1", features = ["full"] } 31 12 32 32 - [profile.profiling] 33 33 - inherits = "release" 34 34 - debug = true 35 35 - 36 36 - # [profile.release] 37 37 - # debug = true 38 38 - 39 39 - [[bench]] 40 40 - name = "non-huge-cars" 41 41 - harness = false 42 42 - 43 43 - [[bench]] 44 44 - name = "huge-car" 45 45 - harness = false

+12 -21

benches/huge-car.rs

··· 1 1 extern crate repo_stream; 2 2 - use futures::TryStreamExt; 3 3 - use iroh_car::CarReader; 4 4 - use std::convert::Infallible; 2 2 + use repo_stream::Driver; 5 3 use std::path::{Path, PathBuf}; 6 4 7 5 use criterion::{Criterion, criterion_group, criterion_main}; ··· 20 18 }); 21 19 } 22 20 23 23 - async fn drive_car(filename: impl AsRef<Path>) { 21 21 + async fn drive_car(filename: impl AsRef<Path>) -> usize { 24 22 let reader = tokio::fs::File::open(filename).await.unwrap(); 25 23 let reader = tokio::io::BufReader::new(reader); 26 26 - let reader = CarReader::new(reader).await.unwrap(); 27 24 28 28 - let root = reader 29 29 - .header() 30 30 - .roots() 31 31 - .first() 32 32 - .ok_or("missing root") 25 25 + let mut driver = match Driver::load_car(reader, |block| block.len(), 1024) 26 26 + .await 33 27 .unwrap() 34 34 - .clone(); 35 35 - 36 36 - let stream = std::pin::pin!(reader.stream()); 37 37 - 38 38 - let (_commit, v) = 39 39 - repo_stream::drive::Vehicle::init(root, stream, |block| Ok::<_, Infallible>(block.len())) 40 40 - .await 41 41 - .unwrap(); 42 42 - let mut record_stream = std::pin::pin!(v.stream()); 28 28 + { 29 29 + Driver::Memory(_, mem_driver) => mem_driver, 30 30 + Driver::Disk(_) => panic!("not doing disk for benchmark"), 31 31 + }; 43 32 44 44 - while let Some(_) = record_stream.try_next().await.unwrap() { 45 45 - // just here for the drive 33 33 + let mut n = 0; 34 34 + while let Some(pairs) = driver.next_chunk(256).await.unwrap() { 35 35 + n += pairs.len(); 46 36 } 37 37 + n 47 38 } 48 39 49 40 criterion_group!(benches, criterion_benchmark);

+16 -22

benches/non-huge-cars.rs

··· 1 1 extern crate repo_stream; 2 2 - use futures::TryStreamExt; 3 3 - use iroh_car::CarReader; 4 4 - use std::convert::Infallible; 2 2 + use repo_stream::Driver; 5 3 6 4 use criterion::{Criterion, criterion_group, criterion_main}; 7 5 6 6 + const EMPTY_CAR: &'static [u8] = include_bytes!("../car-samples/empty.car"); 8 7 const TINY_CAR: &'static [u8] = include_bytes!("../car-samples/tiny.car"); 9 8 const LITTLE_CAR: &'static [u8] = include_bytes!("../car-samples/little.car"); 10 9 const MIDSIZE_CAR: &'static [u8] = include_bytes!("../car-samples/midsize.car"); ··· 15 14 .build() 16 15 .expect("Creating runtime failed"); 17 16 17 17 + c.bench_function("empty-car", |b| { 18 18 + b.to_async(&rt).iter(async || drive_car(EMPTY_CAR).await) 19 19 + }); 18 20 c.bench_function("tiny-car", |b| { 19 21 b.to_async(&rt).iter(async || drive_car(TINY_CAR).await) 20 22 }); ··· 26 28 }); 27 29 } 28 30 29 29 - async fn drive_car(bytes: &[u8]) { 30 30 - let reader = CarReader::new(bytes).await.unwrap(); 31 31 - 32 32 - let root = reader 33 33 - .header() 34 34 - .roots() 35 35 - .first() 36 36 - .ok_or("missing root") 31 31 + async fn drive_car(bytes: &[u8]) -> usize { 32 32 + let mut driver = match Driver::load_car(bytes, |block| block.len(), 32) 33 33 + .await 37 34 .unwrap() 38 38 - .clone(); 39 39 - 40 40 - let stream = std::pin::pin!(reader.stream()); 35 35 + { 36 36 + Driver::Memory(_, mem_driver) => mem_driver, 37 37 + Driver::Disk(_) => panic!("not benching big cars here"), 38 38 + }; 41 39 42 42 - let (_commit, v) = 43 43 - repo_stream::drive::Vehicle::init(root, stream, |block| Ok::<_, Infallible>(block.len())) 44 44 - .await 45 45 - .unwrap(); 46 46 - let mut record_stream = std::pin::pin!(v.stream()); 47 47 - 48 48 - while let Some(_) = record_stream.try_next().await.unwrap() { 49 49 - // just here for the drive 40 40 + let mut n = 0; 41 41 + while let Some(pairs) = driver.next_chunk(256).await.unwrap() { 42 42 + n += pairs.len(); 50 43 } 44 44 + n 51 45 } 52 46 53 47 criterion_group!(benches, criterion_benchmark);

car-samples/empty.car

This is a binary file and will not be displayed.

+26 -53

examples/disk-read-file/main.rs

··· 1 1 - extern crate repo_stream; 2 1 use clap::Parser; 3 3 - use repo_stream::drive::Processable; 4 4 - use serde::{Deserialize, Serialize}; 5 5 - use std::path::PathBuf; 6 6 - 7 7 - type Result<T> = std::result::Result<T, Box<dyn std::error::Error>>; 2 2 + use fjall::{Database, KeyspaceCreateOptions}; 3 3 + use std::{path::PathBuf, collections::BTreeMap}; 8 4 9 5 #[derive(Debug, Parser)] 10 6 struct Args { 11 7 #[arg()] 12 12 - car: PathBuf, 13 13 - #[arg()] 14 14 - tmpfile: PathBuf, 8 8 + db_path: PathBuf, 15 9 } 16 10 17 17 - #[derive(Clone, Serialize, Deserialize)] 18 18 - struct S(usize); 11 11 + fn main() -> Result<(), Box<dyn std::error::Error>> { 12 12 + let Args { db_path } = Args::parse(); 19 13 20 20 - impl Processable for S { 21 21 - fn get_size(&self) -> usize { 22 22 - 0 // no additional space taken, just its stack size (newtype is free) 14 14 + let db = Database::builder(db_path).open()?; 15 15 + let ks = db.keyspace("z", KeyspaceCreateOptions::default)?; 16 16 + let mut seen_keys: BTreeMap<Vec<u8>, usize> = BTreeMap::default(); 17 17 + 18 18 + print!("writing..."); 19 19 + for i in 0..250_000_usize { 20 20 + let k = i.to_be_bytes().to_vec(); 21 21 + ks.insert(k.clone(), vec![0xAA; 256])?; 22 22 + seen_keys.insert(k, i); 23 23 } 24 24 - } 25 24 26 26 - #[tokio::main] 27 27 - async fn main() -> Result<()> { 28 28 - env_logger::init(); 25 25 + println!(" done. checking keys..."); 29 26 30 30 - let Args { car, tmpfile } = Args::parse(); 31 31 - let reader = tokio::fs::File::open(car).await?; 32 32 - let reader = tokio::io::BufReader::new(reader); 27 27 + // remove every seen key that fjall actually has, to see what's left 28 28 + for guard in ks.iter() { 29 29 + seen_keys.remove(guard.key()?.as_ref()); 30 30 + } 33 31 34 34 - // let kb = 2_usize.pow(10); 35 35 - let mb = 2_usize.pow(20); 36 36 - 37 37 - let limit_mb = 32; 38 38 - 39 39 - let driver = match repo_stream::drive::load_car(reader, |block| S(block.len()), 10 * mb).await? 40 40 - { 41 41 - repo_stream::drive::Vehicle::Lil(_, _) => panic!("try this on a bigger car"), 42 42 - repo_stream::drive::Vehicle::Big(big_stuff) => { 43 43 - let disk_store = repo_stream::disk::SqliteStore::new(tmpfile.clone(), limit_mb); 44 44 - let (commit, driver) = big_stuff.finish_loading(disk_store).await?; 45 45 - log::warn!("big: {:?}", commit); 46 46 - driver 32 32 + // report the result 33 33 + if seen_keys.len() == 0 { 34 34 + println!("[ OK ] all keys found"); 35 35 + } else { 36 36 + println!("[FAIL] fjall did not have all seen_keys:"); 37 37 + for (k, i) in seen_keys { 38 38 + println!(" insert #{i} missing, key bytes: {k:?}"); 47 39 } 48 48 - }; 49 49 - 50 50 - let mut n = 0; 51 51 - let (mut rx, worker) = driver.rx(512).await?; 52 52 - 53 53 - log::debug!("walking..."); 54 54 - while let Some(pairs) = rx.recv().await { 55 55 - n += pairs.len(); 56 40 } 57 57 - log::debug!("done walking! joining..."); 58 58 - 59 59 - worker.await.unwrap().unwrap(); 60 60 - 61 61 - log::debug!("joined."); 62 62 - 63 63 - // log::info!("now is the time to check mem..."); 64 64 - // tokio::time::sleep(std::time::Duration::from_secs(22)).await; 65 65 - log::info!("bye! {n}"); 66 66 - 67 67 - std::fs::remove_file(tmpfile).unwrap(); // need to also remove -shm -wal 68 41 69 42 Ok(()) 70 43 }

+14 -6

examples/read-file/main.rs

··· 1 1 + /*! 2 2 + Read a CAR file with in-memory processing 3 3 + */ 4 4 + 1 5 extern crate repo_stream; 2 6 use clap::Parser; 7 7 + use repo_stream::{Driver, DriverBuilder}; 3 8 use std::path::PathBuf; 4 9 5 10 type Result<T> = std::result::Result<T, Box<dyn std::error::Error>>; ··· 18 23 let reader = tokio::fs::File::open(file).await?; 19 24 let reader = tokio::io::BufReader::new(reader); 20 25 21 21 - let (commit, mut driver) = 22 22 - match repo_stream::drive::load_car(reader, |block| block.len(), 1024 * 1024).await? { 23 23 - repo_stream::drive::Vehicle::Lil(commit, mem_driver) => (commit, mem_driver), 24 24 - repo_stream::drive::Vehicle::Big(_) => panic!("can't handle big cars yet"), 25 25 - }; 26 26 + let (commit, mut driver) = match DriverBuilder::new() 27 27 + .with_block_processor(|block| block.len()) 28 28 + .load_car(reader) 29 29 + .await? 30 30 + { 31 31 + Driver::Memory(commit, mem_driver) => (commit, mem_driver), 32 32 + Driver::Disk(_) => panic!("this example doesn't handle big CARs"), 33 33 + }; 26 34 27 35 log::info!("got commit: {commit:?}"); 28 36 ··· 31 39 n += pairs.len(); 32 40 // log::info!("got {rkey:?}"); 33 41 } 34 34 - log::info!("bye! {n}"); 42 42 + log::info!("bye! total records={n}"); 35 43 36 44 Ok(()) 37 45 }

+70 -2

readme.md

··· 1 1 # repo-stream 2 2 3 3 - Fast and (aspirationally) robust atproto CAR file processing in rust 3 3 + A robust CAR file -> MST walker for atproto 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) 8 8 + 9 9 + [crates-badge]: https://img.shields.io/crates/v/repo-stream.svg 10 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 12 12 + 13 13 + ```rust 14 14 + use repo_stream::{Driver, DriverBuilder, DriveError, DiskBuilder}; 15 15 + 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 + // clean up the disk store (drop tables etc) 55 55 + driver.reset_store().await?; 56 56 + } 57 57 + }; 58 58 + println!("sum of size of all records: {total_size}"); 59 59 + Ok(()) 60 60 + } 61 61 + ``` 62 62 + 63 63 + more recent todo 64 64 + 65 65 + - [ ] get an *emtpy* car for the test suite 66 66 + - [x] implement a max size on disk limit 67 67 + 68 68 + 69 69 + ----- 70 70 + 71 71 + older stuff (to clean up): 4 72 5 73 6 74 current car processing times (records processed into their length usize, phil's dev machine): ··· 27 95 -> yeah the commit is returned from init 28 96 - [ ] spec compliance todos 29 97 - [x] assert that keys are ordered and fail if not 30 30 - - [ ] verify node mst depth from key (possibly pending [interop test fixes](https://github.com/bluesky-social/atproto-interop-tests/issues/5)) 98 98 + - [x] verify node mst depth from key (possibly pending [interop test fixes](https://github.com/bluesky-social/atproto-interop-tests/issues/5)) 31 99 - [ ] performance todos 32 100 - [x] consume the serialized nodes into a mutable efficient format 33 101 - [ ] maybe customize the deserialize impl to do that directly?

-105

src/disk.rs

··· 1 1 - use rusqlite::OptionalExtension; 2 2 - use std::path::PathBuf; 3 3 - 4 4 - pub struct SqliteStore { 5 5 - path: PathBuf, 6 6 - limit_mb: usize, 7 7 - } 8 8 - 9 9 - impl SqliteStore { 10 10 - pub fn new(path: PathBuf, limit_mb: usize) -> Self { 11 11 - Self { path, limit_mb } 12 12 - } 13 13 - } 14 14 - 15 15 - impl SqliteStore { 16 16 - pub async fn get_access(&mut self) -> Result<SqliteAccess, rusqlite::Error> { 17 17 - let path = self.path.clone(); 18 18 - let limit_mb = self.limit_mb; 19 19 - let conn = tokio::task::spawn_blocking(move || { 20 20 - let conn = rusqlite::Connection::open(path)?; 21 21 - 22 22 - let sq_mb = -(2_i64.pow(10)); // negative is kibibytes for sqlite cache_size 23 23 - 24 24 - // conn.pragma_update(None, "journal_mode", "OFF")?; 25 25 - // conn.pragma_update(None, "journal_mode", "MEMORY")?; 26 26 - conn.pragma_update(None, "journal_mode", "WAL")?; 27 27 - // conn.pragma_update(None, "wal_autocheckpoint", "0")?; // this lets things get a bit big on disk 28 28 - conn.pragma_update(None, "synchronous", "OFF")?; 29 29 - conn.pragma_update(None, "cache_size", (limit_mb as i64 * sq_mb).to_string())?; 30 30 - conn.execute( 31 31 - "CREATE TABLE blocks ( 32 32 - key BLOB PRIMARY KEY NOT NULL, 33 33 - val BLOB NOT NULL 34 34 - ) WITHOUT ROWID", 35 35 - (), 36 36 - )?; 37 37 - 38 38 - Ok::<_, rusqlite::Error>(conn) 39 39 - }) 40 40 - .await 41 41 - .expect("join error")?; 42 42 - 43 43 - Ok(SqliteAccess { conn }) 44 44 - } 45 45 - } 46 46 - 47 47 - pub struct SqliteAccess { 48 48 - conn: rusqlite::Connection, 49 49 - } 50 50 - 51 51 - impl SqliteAccess { 52 52 - pub fn get_writer(&'_ mut self) -> Result<SqliteWriter<'_>, rusqlite::Error> { 53 53 - let tx = self.conn.transaction()?; 54 54 - // let insert_stmt = tx.prepare("INSERT INTO blocks (key, val) VALUES (?1, ?2)")?; 55 55 - Ok(SqliteWriter { tx: Some(tx) }) 56 56 - } 57 57 - pub fn get_reader(&'_ self) -> Result<SqliteReader<'_>, rusqlite::Error> { 58 58 - let select_stmt = self.conn.prepare("SELECT val FROM blocks WHERE key = ?1")?; 59 59 - Ok(SqliteReader { select_stmt }) 60 60 - } 61 61 - } 62 62 - 63 63 - pub struct SqliteWriter<'conn> { 64 64 - tx: Option<rusqlite::Transaction<'conn>>, 65 65 - } 66 66 - 67 67 - /// oops careful in async 68 68 - impl Drop for SqliteWriter<'_> { 69 69 - fn drop(&mut self) { 70 70 - let tx = self.tx.take(); 71 71 - tx.unwrap().commit().unwrap(); 72 72 - } 73 73 - } 74 74 - 75 75 - impl SqliteWriter<'_> { 76 76 - pub fn put(&mut self, key: Vec<u8>, val: Vec<u8>) -> rusqlite::Result<()> { 77 77 - let tx = self.tx.as_ref().unwrap(); 78 78 - let mut insert_stmt = tx.prepare_cached("INSERT INTO blocks (key, val) VALUES (?1, ?2)")?; 79 79 - insert_stmt.execute((key, val))?; 80 80 - Ok(()) 81 81 - } 82 82 - pub fn put_many( 83 83 - &mut self, 84 84 - kv: impl Iterator<Item = (Vec<u8>, Vec<u8>)>, 85 85 - ) -> rusqlite::Result<()> { 86 86 - let tx = self.tx.as_ref().unwrap(); 87 87 - let mut insert_stmt = tx.prepare_cached("INSERT INTO blocks (key, val) VALUES (?1, ?2)")?; 88 88 - for (k, v) in kv { 89 89 - insert_stmt.execute((k, v))?; 90 90 - } 91 91 - Ok(()) 92 92 - } 93 93 - } 94 94 - 95 95 - pub struct SqliteReader<'conn> { 96 96 - select_stmt: rusqlite::Statement<'conn>, 97 97 - } 98 98 - 99 99 - impl SqliteReader<'_> { 100 100 - pub fn get(&mut self, key: Vec<u8>) -> rusqlite::Result<Option<Vec<u8>>> { 101 101 - self.select_stmt 102 102 - .query_one((&key,), |row| row.get(0)) 103 103 - .optional() 104 104 - } 105 105 - }

-423

src/drive.rs

··· 1 1 - //! Consume an MST block stream, producing an ordered stream of records 2 2 - 3 3 - use crate::disk::{SqliteAccess, SqliteStore}; 4 4 - use ipld_core::cid::Cid; 5 5 - use iroh_car::CarReader; 6 6 - use serde::de::DeserializeOwned; 7 7 - use serde::{Deserialize, Serialize}; 8 8 - use std::collections::HashMap; 9 9 - use std::convert::Infallible; 10 10 - use tokio::io::AsyncRead; 11 11 - 12 12 - use crate::mst::{Commit, Node}; 13 13 - use crate::walk::{DiskTrip, Step, Trip, Walker}; 14 14 - 15 15 - /// Errors that can happen while consuming and emitting blocks and records 16 16 - #[derive(Debug, thiserror::Error)] 17 17 - pub enum DriveError { 18 18 - #[error("Error from iroh_car: {0}")] 19 19 - CarReader(#[from] iroh_car::Error), 20 20 - #[error("Failed to decode commit block: {0}")] 21 21 - BadBlock(#[from] serde_ipld_dagcbor::DecodeError<Infallible>), 22 22 - #[error("The Commit block reference by the root was not found")] 23 23 - MissingCommit, 24 24 - #[error("The MST block {0} could not be found")] 25 25 - MissingBlock(Cid), 26 26 - #[error("Failed to walk the mst tree: {0}")] 27 27 - Tripped(#[from] Trip), 28 28 - #[error("CAR file had no roots")] 29 29 - MissingRoot, 30 30 - } 31 31 - 32 32 - #[derive(Debug, thiserror::Error)] 33 33 - pub enum DiskDriveError { 34 34 - #[error("Error from iroh_car: {0}")] 35 35 - CarReader(#[from] iroh_car::Error), 36 36 - #[error("Failed to decode commit block: {0}")] 37 37 - BadBlock(#[from] serde_ipld_dagcbor::DecodeError<Infallible>), 38 38 - #[error("Storage error")] 39 39 - StorageError(#[from] rusqlite::Error), 40 40 - #[error("The Commit block reference by the root was not found")] 41 41 - MissingCommit, 42 42 - #[error("The MST block {0} could not be found")] 43 43 - MissingBlock(Cid), 44 44 - #[error("Encode error: {0}")] 45 45 - BincodeEncodeError(#[from] bincode::error::EncodeError), 46 46 - #[error("Decode error: {0}")] 47 47 - BincodeDecodeError(#[from] bincode::error::DecodeError), 48 48 - #[error("disk tripped: {0}")] 49 49 - DiskTripped(#[from] DiskTrip), 50 50 - } 51 51 - 52 52 - pub trait Processable: Clone + Serialize + DeserializeOwned { 53 53 - /// the additional size taken up (not including its mem::size_of) 54 54 - fn get_size(&self) -> usize; 55 55 - } 56 56 - 57 57 - #[derive(Debug, Clone, Serialize, Deserialize)] 58 58 - pub enum MaybeProcessedBlock<T> { 59 59 - /// A block that's *probably* a Node (but we can't know yet) 60 60 - /// 61 61 - /// It *can be* a record that suspiciously looks a lot like a node, so we 62 62 - /// cannot eagerly turn it into a Node. We only know for sure what it is 63 63 - /// when we actually walk down the MST 64 64 - Raw(Vec<u8>), 65 65 - /// A processed record from a block that was definitely not a Node 66 66 - /// 67 67 - /// Processing has to be fallible because the CAR can have totally-unused 68 68 - /// blocks, which can just be garbage. since we're eagerly trying to process 69 69 - /// record blocks without knowing for sure that they *are* records, we 70 70 - /// discard any definitely-not-nodes that fail processing and keep their 71 71 - /// error in the buffer for them. if we later try to retreive them as a 72 72 - /// record, then we can surface the error. 73 73 - /// 74 74 - /// If we _never_ needed this block, then we may have wasted a bit of effort 75 75 - /// trying to process it. Oh well. 76 76 - /// 77 77 - /// There's an alternative here, which would be to kick unprocessable blocks 78 78 - /// back to Raw, or maybe even a new RawUnprocessable variant. Then we could 79 79 - /// surface the typed error later if needed by trying to reprocess. 80 80 - Processed(T), 81 81 - } 82 82 - 83 83 - impl<T: Processable> Processable for MaybeProcessedBlock<T> { 84 84 - /// TODO this is probably a little broken 85 85 - fn get_size(&self) -> usize { 86 86 - use std::{cmp::max, mem::size_of}; 87 87 - 88 88 - // enum is always as big as its biggest member? 89 89 - let base_size = max(size_of::<Vec<u8>>(), size_of::<T>()); 90 90 - 91 91 - let extra = match self { 92 92 - Self::Raw(bytes) => bytes.len(), 93 93 - Self::Processed(t) => t.get_size(), 94 94 - }; 95 95 - 96 96 - base_size + extra 97 97 - } 98 98 - } 99 99 - 100 100 - pub enum Vehicle<R: AsyncRead + Unpin, T: Processable> { 101 101 - Lil(Commit, MemDriver<T>), 102 102 - Big(BigCar<R, T>), 103 103 - } 104 104 - 105 105 - pub async fn load_car<R: AsyncRead + Unpin, T: Processable>( 106 106 - reader: R, 107 107 - process: fn(Vec<u8>) -> T, 108 108 - max_size: usize, 109 109 - ) -> Result<Vehicle<R, T>, DriveError> { 110 110 - let mut mem_blocks = HashMap::new(); 111 111 - 112 112 - let mut car = CarReader::new(reader).await?; 113 113 - 114 114 - let root = *car 115 115 - .header() 116 116 - .roots() 117 117 - .first() 118 118 - .ok_or(DriveError::MissingRoot)?; 119 119 - log::debug!("root: {root:?}"); 120 120 - 121 121 - let mut commit = None; 122 122 - 123 123 - // try to load all the blocks into memory 124 124 - let mut mem_size = 0; 125 125 - while let Some((cid, data)) = car.next_block().await? { 126 126 - // the root commit is a Special Third Kind of block that we need to make 127 127 - // sure not to optimistically send to the processing function 128 128 - if cid == root { 129 129 - let c: Commit = serde_ipld_dagcbor::from_slice(&data)?; 130 130 - commit = Some(c); 131 131 - continue; 132 132 - } 133 133 - 134 134 - // remaining possible types: node, record, other. optimistically process 135 135 - // TODO: get the actual in-memory size to compute disk spill 136 136 - let maybe_processed = if Node::could_be(&data) { 137 137 - MaybeProcessedBlock::Raw(data) 138 138 - } else { 139 139 - MaybeProcessedBlock::Processed(process(data)) 140 140 - }; 141 141 - 142 142 - // stash (maybe processed) blocks in memory as long as we have room 143 143 - mem_size += std::mem::size_of::<Cid>() + maybe_processed.get_size(); 144 144 - mem_blocks.insert(cid, maybe_processed); 145 145 - if mem_size >= max_size { 146 146 - return Ok(Vehicle::Big(BigCar { 147 147 - car, 148 148 - root, 149 149 - process, 150 150 - max_size, 151 151 - mem_blocks, 152 152 - commit, 153 153 - })); 154 154 - } 155 155 - } 156 156 - 157 157 - // all blocks loaded and we fit in memory! hopefully we found the commit... 158 158 - let commit = commit.ok_or(DriveError::MissingCommit)?; 159 159 - 160 160 - let walker = Walker::new(commit.data); 161 161 - 162 162 - Ok(Vehicle::Lil( 163 163 - commit, 164 164 - MemDriver { 165 165 - blocks: mem_blocks, 166 166 - walker, 167 167 - process, 168 168 - }, 169 169 - )) 170 170 - } 171 171 - 172 172 - /// a paritally memory-loaded car file that needs disk spillover to continue 173 173 - pub struct BigCar<R: AsyncRead + Unpin, T: Processable> { 174 174 - car: CarReader<R>, 175 175 - root: Cid, 176 176 - process: fn(Vec<u8>) -> T, 177 177 - max_size: usize, 178 178 - mem_blocks: HashMap<Cid, MaybeProcessedBlock<T>>, 179 179 - pub commit: Option<Commit>, 180 180 - } 181 181 - 182 182 - fn encode(v: impl Serialize) -> Result<Vec<u8>, bincode::error::EncodeError> { 183 183 - bincode::serde::encode_to_vec(v, bincode::config::standard()) 184 184 - } 185 185 - 186 186 - pub fn decode<T: Processable>(bytes: &[u8]) -> Result<T, bincode::error::DecodeError> { 187 187 - let (t, n) = bincode::serde::decode_from_slice(bytes, bincode::config::standard())?; 188 188 - assert_eq!(n, bytes.len(), "expected to decode all bytes"); // TODO 189 189 - Ok(t) 190 190 - } 191 191 - 192 192 - impl<R: AsyncRead + Unpin, T: Processable + Send + 'static> BigCar<R, T> { 193 193 - pub async fn finish_loading( 194 194 - mut self, 195 195 - mut store: SqliteStore, 196 196 - ) -> Result<(Commit, BigCarReady<T>), DiskDriveError> { 197 197 - // set up access for real 198 198 - let mut access = store.get_access().await?; 199 199 - 200 200 - // move access in and back out so we can manage lifetimes 201 201 - // dump mem blocks into the store 202 202 - access = tokio::task::spawn(async move { 203 203 - let mut writer = access.get_writer()?; 204 204 - 205 205 - let kvs = self 206 206 - .mem_blocks 207 207 - .into_iter() 208 208 - .map(|(k, v)| (k.to_bytes(), encode(v).unwrap())); 209 209 - 210 210 - writer.put_many(kvs)?; 211 211 - 212 212 - drop(writer); // cannot outlive access 213 213 - Ok::<_, DiskDriveError>(access) 214 214 - }) 215 215 - .await 216 216 - .unwrap()?; 217 217 - 218 218 - let (tx, mut rx) = tokio::sync::mpsc::channel::<Vec<(Cid, MaybeProcessedBlock<T>)>>(2); 219 219 - 220 220 - let access_worker = tokio::task::spawn_blocking(move || { 221 221 - let mut writer = access.get_writer()?; 222 222 - 223 223 - while let Some(chunk) = rx.blocking_recv() { 224 224 - let kvs = chunk 225 225 - .into_iter() 226 226 - .map(|(k, v)| (k.to_bytes(), encode(v).unwrap())); 227 227 - writer.put_many(kvs)?; 228 228 - } 229 229 - 230 230 - drop(writer); // cannot outlive access 231 231 - Ok::<_, DiskDriveError>(access) 232 232 - }); // await later 233 233 - 234 234 - // dump the rest to disk (in chunks) 235 235 - log::debug!("dumping the rest of the stream..."); 236 236 - loop { 237 237 - let mut mem_size = 0; 238 238 - let mut chunk = vec![]; 239 239 - loop { 240 240 - let Some((cid, data)) = self.car.next_block().await? else { 241 241 - break; 242 242 - }; 243 243 - // we still gotta keep checking for the root since we might not have it 244 244 - if cid == self.root { 245 245 - let c: Commit = serde_ipld_dagcbor::from_slice(&data)?; 246 246 - self.commit = Some(c); 247 247 - continue; 248 248 - } 249 249 - // remaining possible types: node, record, other. optimistically process 250 250 - // TODO: get the actual in-memory size to compute disk spill 251 251 - let maybe_processed = if Node::could_be(&data) { 252 252 - MaybeProcessedBlock::Raw(data) 253 253 - } else { 254 254 - MaybeProcessedBlock::Processed((self.process)(data)) 255 255 - }; 256 256 - mem_size += std::mem::size_of::<Cid>() + maybe_processed.get_size(); 257 257 - chunk.push((cid, maybe_processed)); 258 258 - if mem_size >= self.max_size { 259 259 - // soooooo if we're setting the db cache to max_size and then letting 260 260 - // multiple chunks in the queue that are >= max_size, then at any time 261 261 - // we might be using some multiple of max_size? 262 262 - break; 263 263 - } 264 264 - } 265 265 - if chunk.is_empty() { 266 266 - break; 267 267 - } 268 268 - tx.send(chunk).await.unwrap(); 269 269 - } 270 270 - drop(tx); 271 271 - log::debug!("done. waiting for worker to finish..."); 272 272 - 273 273 - access = access_worker.await.unwrap()?; 274 274 - 275 275 - log::debug!("worker finished."); 276 276 - 277 277 - let commit = self.commit.ok_or(DiskDriveError::MissingCommit)?; 278 278 - 279 279 - let walker = Walker::new(commit.data); 280 280 - 281 281 - Ok(( 282 282 - commit, 283 283 - BigCarReady { 284 284 - process: self.process, 285 285 - access, 286 286 - walker, 287 287 - }, 288 288 - )) 289 289 - } 290 290 - } 291 291 - 292 292 - pub struct BigCarReady<T: Clone> { 293 293 - process: fn(Vec<u8>) -> T, 294 294 - access: SqliteAccess, 295 295 - walker: Walker, 296 296 - } 297 297 - 298 298 - impl<T: Processable + Send + 'static> BigCarReady<T> { 299 299 - pub async fn next_chunk( 300 300 - mut self, 301 301 - n: usize, 302 302 - ) -> Result<(Self, Option<Vec<(String, T)>>), DiskDriveError> { 303 303 - let mut out = Vec::with_capacity(n); 304 304 - (self, out) = tokio::task::spawn_blocking(move || { 305 305 - let access = self.access; 306 306 - let mut reader = access.get_reader()?; 307 307 - 308 308 - for _ in 0..n { 309 309 - // walk as far as we can until we run out of blocks or find a record 310 310 - match self.walker.disk_step(&mut reader, self.process)? { 311 311 - Step::Missing(cid) => return Err(DiskDriveError::MissingBlock(cid)), 312 312 - Step::Finish => break, 313 313 - Step::Step { rkey, data } => { 314 314 - out.push((rkey, data)); 315 315 - continue; 316 316 - } 317 317 - }; 318 318 - } 319 319 - 320 320 - drop(reader); // cannot outlive access 321 321 - self.access = access; 322 322 - Ok::<_, DiskDriveError>((self, out)) 323 323 - }) 324 324 - .await 325 325 - .unwrap()?; // TODO 326 326 - 327 327 - if out.is_empty() { 328 328 - Ok((self, None)) 329 329 - } else { 330 330 - Ok((self, Some(out))) 331 331 - } 332 332 - } 333 333 - 334 334 - pub async fn rx( 335 335 - mut self, 336 336 - n: usize, 337 337 - ) -> Result< 338 338 - ( 339 339 - tokio::sync::mpsc::Receiver<Vec<(String, T)>>, 340 340 - tokio::task::JoinHandle<Result<(), DiskDriveError>>, 341 341 - ), 342 342 - DiskDriveError, 343 343 - > { 344 344 - let (tx, rx) = tokio::sync::mpsc::channel::<Vec<(String, T)>>(1); 345 345 - 346 346 - // sketch: this worker is going to be allowed to execute without a join handle 347 347 - // ...should we return the join handle here so the caller at least knows about it? 348 348 - // yes probably for error handling?? (orrr put errors in the channel) 349 349 - let worker = tokio::task::spawn_blocking(move || { 350 350 - let mut reader = self.access.get_reader()?; 351 351 - 352 352 - loop { 353 353 - let mut out = Vec::with_capacity(n); 354 354 - 355 355 - for _ in 0..n { 356 356 - // walk as far as we can until we run out of blocks or find a record 357 357 - match self.walker.disk_step(&mut reader, self.process)? { 358 358 - Step::Missing(cid) => return Err(DiskDriveError::MissingBlock(cid)), 359 359 - Step::Finish => break, 360 360 - Step::Step { rkey, data } => { 361 361 - out.push((rkey, data)); 362 362 - continue; 363 363 - } 364 364 - }; 365 365 - } 366 366 - 367 367 - if out.is_empty() { 368 368 - break; 369 369 - } 370 370 - tx.blocking_send(out).unwrap(); 371 371 - } 372 372 - 373 373 - drop(reader); // cannot outlive access 374 374 - Ok(()) 375 375 - }); // await later 376 376 - 377 377 - Ok((rx, worker)) 378 378 - } 379 379 - } 380 380 - 381 381 - /// The core driver between the block stream and MST walker 382 382 - /// 383 383 - /// In the future, PDSs will export CARs in a stream-friendly order that will 384 384 - /// enable processing them with tiny memory overhead. But that future is not 385 385 - /// here yet. 386 386 - /// 387 387 - /// CARs are almost always in a stream-unfriendly order, so I'm reverting the 388 388 - /// optimistic stream features: we load all block first, then walk the MST. 389 389 - /// 390 390 - /// This makes things much simpler: we only need to worry about spilling to disk 391 391 - /// in one place, and we always have a reasonable expecatation about how much 392 392 - /// work the init function will do. We can drop the CAR reader before walking, 393 393 - /// so the sync/async boundaries become a little easier to work around. 394 394 - #[derive(Debug)] 395 395 - pub struct MemDriver<T: Processable> { 396 396 - blocks: HashMap<Cid, MaybeProcessedBlock<T>>, 397 397 - walker: Walker, 398 398 - process: fn(Vec<u8>) -> T, 399 399 - } 400 400 - 401 401 - impl<T: Processable> MemDriver<T> { 402 402 - /// Manually step through the record outputs 403 403 - pub async fn next_chunk(&mut self, n: usize) -> Result<Option<Vec<(String, T)>>, DriveError> { 404 404 - let mut out = Vec::with_capacity(n); 405 405 - for _ in 0..n { 406 406 - // walk as far as we can until we run out of blocks or find a record 407 407 - match self.walker.step(&mut self.blocks, self.process)? { 408 408 - Step::Missing(cid) => return Err(DriveError::MissingBlock(cid)), 409 409 - Step::Finish => break, 410 410 - Step::Step { rkey, data } => { 411 411 - out.push((rkey, data)); 412 412 - continue; 413 413 - } 414 414 - }; 415 415 - } 416 416 - 417 417 - if out.is_empty() { 418 418 - Ok(None) 419 419 - } else { 420 420 - Ok(Some(out)) 421 421 - } 422 422 - } 423 423 - }

+74 -7

src/lib.rs

··· 1 1 - //! Fast and robust atproto CAR file processing in rust 2 2 - //! 3 3 - //! For now see the [examples](https://tangled.org/@microcosm.blue/repo-stream/tree/main/examples) 1 1 + /*! 2 2 + A robust CAR file -> MST walker for atproto 3 3 + 4 4 + Small CARs have their blocks buffered in memory. If a configurable memory limit 5 5 + is reached while reading blocks, CAR reading is suspended, and can be continued 6 6 + by providing disk storage to buffer the CAR blocks instead. 7 7 + 8 8 + A `process` function can be provided for tasks where records are transformed 9 9 + into a smaller representation, to save memory (and disk) during block reading. 10 10 + 11 11 + Once blocks are loaded, the MST is walked and emitted as chunks of pairs of 12 12 + `(rkey, processed_block)` pairs, in order (depth first, left-to-right). 13 13 + 14 14 + Some MST validations are applied 15 15 + - Keys must appear in order 16 16 + - Keys must be at the correct MST tree depth 17 17 + 18 18 + `iroh_car` additionally applies a block size limit of `2MiB`. 19 19 + 20 20 + ``` 21 21 + use repo_stream::{Driver, DriverBuilder, DiskBuilder}; 22 22 + 23 23 + # #[tokio::main] 24 24 + # async fn main() -> Result<(), Box<dyn std::error::Error>> { 25 25 + # let reader = include_bytes!("../car-samples/tiny.car").as_slice(); 26 26 + let mut total_size = 0; 27 27 + 28 28 + match DriverBuilder::new() 29 29 + .with_mem_limit_mb(10) 30 30 + .with_block_processor(|rec| rec.len()) // block processing: just extract the raw record size 31 31 + .load_car(reader) 32 32 + .await? 33 33 + { 34 34 + 35 35 + // if all blocks fit within memory 36 36 + Driver::Memory(_commit, mut driver) => { 37 37 + while let Some(chunk) = driver.next_chunk(256).await? { 38 38 + for (_rkey, size) in chunk { 39 39 + total_size += size; 40 40 + } 41 41 + } 42 42 + }, 4 43 5 5 - pub mod disk; 6 6 - pub mod drive; 7 7 - pub mod mst; 8 8 - pub mod walk; 44 44 + // if the CAR was too big for in-memory processing 45 45 + Driver::Disk(paused) => { 46 46 + // set up a disk store we can spill to 47 47 + let store = DiskBuilder::new().open("some/path.db".into()).await?; 48 48 + // do the spilling, get back a (similar) driver 49 49 + let (_commit, mut driver) = paused.finish_loading(store).await?; 50 50 + 51 51 + while let Some(chunk) = driver.next_chunk(256).await? { 52 52 + for (_rkey, size) in chunk { 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 59 + } 60 60 + }; 61 61 + println!("sum of size of all records: {total_size}"); 62 62 + # Ok(()) 63 63 + # } 64 64 + ``` 65 65 + 66 66 + Disk spilling suspends and returns a `Driver::Disk(paused)` instead of going 67 67 + ahead and eagerly using disk I/O. This means you have to write a bit more code 68 68 + to handle both cases, but it allows you to have finer control over resource 69 69 + usage. For example, you can drive a number of parallel memory CAR workers, and 70 70 + separately have a different number of disk workers picking up suspended disk 71 71 + tasks from a queue. 72 72 + 73 73 + Find more [examples in the repo](https://tangled.org/@microcosm.blue/repo-stream/tree/main/examples). 74 74 + 75 75 + */

-114

src/mst.rs

··· 1 1 - //! Low-level types for parsing raw atproto MST CARs 2 2 - //! 3 3 - //! The primary aim is to work through the **tree** structure. Non-node blocks 4 4 - //! are left as raw bytes, for upper levels to parse into DAG-CBOR or whatever. 5 5 - 6 6 - use ipld_core::cid::Cid; 7 7 - use serde::Deserialize; 8 8 - 9 9 - /// The top-level data object in a repository's tree is a signed commit. 10 10 - #[derive(Debug, Deserialize)] 11 11 - // #[serde(deny_unknown_fields)] 12 12 - pub struct Commit { 13 13 - /// the account DID associated with the repo, in strictly normalized form 14 14 - /// (eg, lowercase as appropriate) 15 15 - pub did: String, 16 16 - /// fixed value of 3 for this repo format version 17 17 - pub version: u64, 18 18 - /// pointer to the top of the repo contents tree structure (MST) 19 19 - pub data: Cid, 20 20 - /// revision of the repo, used as a logical clock. 21 21 - /// 22 22 - /// TID format. Must increase monotonically. Recommend using current 23 23 - /// timestamp as TID; rev values in the "future" (beyond a fudge factor) 24 24 - /// should be ignored and not processed 25 25 - pub rev: String, 26 26 - /// pointer (by hash) to a previous commit object for this repository. 27 27 - /// 28 28 - /// Could be used to create a chain of history, but largely unused (included 29 29 - /// for v2 backwards compatibility). In version 3 repos, this field must 30 30 - /// exist in the CBOR object, but is virtually always null. NOTE: previously 31 31 - /// specified as nullable and optional, but this caused interoperability 32 32 - /// issues. 33 33 - pub prev: Option<Cid>, 34 34 - /// cryptographic signature of this commit, as raw bytes 35 35 - #[serde(with = "serde_bytes")] 36 36 - pub sig: Vec<u8>, 37 37 - } 38 38 - 39 39 - /// MST node data schema 40 40 - #[derive(Debug, Deserialize, PartialEq)] 41 41 - #[serde(deny_unknown_fields)] 42 42 - pub struct Node { 43 43 - /// link to sub-tree Node on a lower level and with all keys sorting before 44 44 - /// keys at this node 45 45 - #[serde(rename = "l")] 46 46 - pub left: Option<Cid>, 47 47 - /// ordered list of TreeEntry objects 48 48 - /// 49 49 - /// atproto MSTs have a fanout of 4, so there can be max 4 entries. 50 50 - #[serde(rename = "e")] 51 51 - pub entries: Vec<Entry>, // maybe we can do [Option<Entry>; 4]? 52 52 - } 53 53 - 54 54 - impl Node { 55 55 - /// test if a block could possibly be a node 56 56 - /// 57 57 - /// we can't eagerly decode records except where we're *sure* they cannot be 58 58 - /// an mst node (and even then we can only attempt) because you can't know 59 59 - /// with certainty what a block is supposed to be without actually walking 60 60 - /// the tree. 61 61 - /// 62 62 - /// so if a block *could be* a node, any record converter must postpone 63 63 - /// processing. if it turns out it happens to be a very node-looking record, 64 64 - /// well, sorry, it just has to only be processed later when that's known. 65 65 - pub fn could_be(bytes: impl AsRef<[u8]>) -> bool { 66 66 - const NODE_FINGERPRINT: [u8; 3] = [ 67 67 - 0xA2, // map length 2 (for "l" and "e" keys) 68 68 - 0x61, // text length 1 69 69 - b'e', // "e" before "l" because map keys have to be lex-sorted 70 70 - // 0x8?: "e" has array (0x100 upper 3 bits) of some length 71 71 - ]; 72 72 - let bytes = bytes.as_ref(); 73 73 - bytes.starts_with(&NODE_FINGERPRINT) 74 74 - && bytes 75 75 - .get(3) 76 76 - .map(|b| b & 0b1110_0000 == 0x80) 77 77 - .unwrap_or(false) 78 78 - } 79 79 - 80 80 - /// Check if a node has any entries 81 81 - /// 82 82 - /// An empty repository with no records is represented as a single MST node 83 83 - /// with an empty array of entries. This is the only situation in which a 84 84 - /// tree may contain an empty leaf node which does not either contain keys 85 85 - /// ("entries") or point to a sub-tree containing entries. 86 86 - /// 87 87 - /// TODO: to me this is slightly unclear with respect to `l` (ask someone). 88 88 - /// ...is that what "The top of the tree must not be a an empty node which 89 89 - /// only points to a sub-tree." is referring to? 90 90 - pub fn is_empty(&self) -> bool { 91 91 - self.left.is_none() && self.entries.is_empty() 92 92 - } 93 93 - } 94 94 - 95 95 - /// TreeEntry object 96 96 - #[derive(Debug, Deserialize, PartialEq)] 97 97 - #[serde(deny_unknown_fields)] 98 98 - pub struct Entry { 99 99 - /// count of bytes shared with previous TreeEntry in this Node (if any) 100 100 - #[serde(rename = "p")] 101 101 - pub prefix_len: usize, 102 102 - /// remainder of key for this TreeEntry, after "prefixlen" have been removed 103 103 - #[serde(rename = "k", with = "serde_bytes")] 104 104 - pub keysuffix: Vec<u8>, // can we String this here? 105 105 - /// link to the record data (CBOR) for this entry 106 106 - #[serde(rename = "v")] 107 107 - pub value: Cid, 108 108 - /// link to a sub-tree Node at a lower level 109 109 - /// 110 110 - /// the lower level must have keys sorting after this TreeEntry's key (to 111 111 - /// the "right"), but before the next TreeEntry's key in this Node (if any) 112 112 - #[serde(rename = "t")] 113 113 - pub tree: Option<Cid>, 114 114 - }

-471

src/walk.rs

··· 1 1 - //! Depth-first MST traversal 2 2 - 3 3 - use crate::disk::SqliteReader; 4 4 - use crate::drive::{MaybeProcessedBlock, Processable}; 5 5 - use crate::mst::Node; 6 6 - use ipld_core::cid::Cid; 7 7 - use std::collections::HashMap; 8 8 - use std::convert::Infallible; 9 9 - 10 10 - /// Errors that can happen while walking 11 11 - #[derive(Debug, thiserror::Error)] 12 12 - pub enum Trip { 13 13 - #[error("empty mst nodes are not allowed")] 14 14 - NodeEmpty, 15 15 - #[error("Failed to fingerprint commit block")] 16 16 - BadCommitFingerprint, 17 17 - #[error("Failed to decode commit block: {0}")] 18 18 - BadCommit(#[from] serde_ipld_dagcbor::DecodeError<Infallible>), 19 19 - #[error("Action node error: {0}")] 20 20 - RkeyError(#[from] RkeyError), 21 21 - #[error("Encountered an rkey out of order while walking the MST")] 22 22 - RkeyOutOfOrder, 23 23 - } 24 24 - 25 25 - /// Errors that can happen while walking 26 26 - #[derive(Debug, thiserror::Error)] 27 27 - pub enum DiskTrip { 28 28 - #[error("tripped: {0}")] 29 29 - Trip(#[from] Trip), 30 30 - #[error("storage error: {0}")] 31 31 - StorageError(#[from] rusqlite::Error), 32 32 - #[error("Decode error: {0}")] 33 33 - BincodeDecodeError(#[from] bincode::error::DecodeError), 34 34 - } 35 35 - 36 36 - /// Errors from invalid Rkeys 37 37 - #[derive(Debug, thiserror::Error)] 38 38 - pub enum RkeyError { 39 39 - #[error("Failed to compute an rkey due to invalid prefix_len")] 40 40 - EntryPrefixOutOfbounds, 41 41 - #[error("RKey was not utf-8")] 42 42 - EntryRkeyNotUtf8(#[from] std::string::FromUtf8Error), 43 43 - } 44 44 - 45 45 - /// Walker outputs 46 46 - #[derive(Debug)] 47 47 - pub enum Step<T> { 48 48 - /// We needed this CID but it's not in the block store 49 49 - Missing(Cid), 50 50 - /// Reached the end of the MST! yay! 51 51 - Finish, 52 52 - /// A record was found! 53 53 - Step { rkey: String, data: T }, 54 54 - } 55 55 - 56 56 - #[derive(Debug, Clone, PartialEq)] 57 57 - enum Need { 58 58 - Node(Cid), 59 59 - Record { rkey: String, cid: Cid }, 60 60 - } 61 61 - 62 62 - fn push_from_node(stack: &mut Vec<Need>, node: &Node) -> Result<(), RkeyError> { 63 63 - let mut entries = Vec::with_capacity(node.entries.len()); 64 64 - 65 65 - let mut prefix = vec![]; 66 66 - for entry in &node.entries { 67 67 - let mut rkey = vec![]; 68 68 - let pre_checked = prefix 69 69 - .get(..entry.prefix_len) 70 70 - .ok_or(RkeyError::EntryPrefixOutOfbounds)?; 71 71 - rkey.extend_from_slice(pre_checked); 72 72 - rkey.extend_from_slice(&entry.keysuffix); 73 73 - prefix = rkey.clone(); 74 74 - 75 75 - entries.push(Need::Record { 76 76 - rkey: String::from_utf8(rkey)?, 77 77 - cid: entry.value, 78 78 - }); 79 79 - if let Some(ref tree) = entry.tree { 80 80 - entries.push(Need::Node(*tree)); 81 81 - } 82 82 - } 83 83 - 84 84 - entries.reverse(); 85 85 - stack.append(&mut entries); 86 86 - 87 87 - if let Some(tree) = node.left { 88 88 - stack.push(Need::Node(tree)); 89 89 - } 90 90 - Ok(()) 91 91 - } 92 92 - 93 93 - /// Traverser of an atproto MST 94 94 - /// 95 95 - /// Walks the tree from left-to-right in depth-first order 96 96 - #[derive(Debug)] 97 97 - pub struct Walker { 98 98 - stack: Vec<Need>, 99 99 - prev: String, 100 100 - } 101 101 - 102 102 - impl Walker { 103 103 - pub fn new(tree_root_cid: Cid) -> Self { 104 104 - Self { 105 105 - stack: vec![Need::Node(tree_root_cid)], 106 106 - prev: "".to_string(), 107 107 - } 108 108 - } 109 109 - 110 110 - /// Advance through nodes until we find a record or can't go further 111 111 - pub fn step<T: Processable>( 112 112 - &mut self, 113 113 - blocks: &mut HashMap<Cid, MaybeProcessedBlock<T>>, 114 114 - process: impl Fn(Vec<u8>) -> T, 115 115 - ) -> Result<Step<T>, Trip> { 116 116 - loop { 117 117 - let Some(mut need) = self.stack.last() else { 118 118 - log::trace!("tried to walk but we're actually done."); 119 119 - return Ok(Step::Finish); 120 120 - }; 121 121 - 122 122 - match &mut need { 123 123 - Need::Node(cid) => { 124 124 - log::trace!("need node {cid:?}"); 125 125 - let Some(block) = blocks.remove(cid) else { 126 126 - log::trace!("node not found, resting"); 127 127 - return Ok(Step::Missing(*cid)); 128 128 - }; 129 129 - 130 130 - let MaybeProcessedBlock::Raw(data) = block else { 131 131 - return Err(Trip::BadCommitFingerprint); 132 132 - }; 133 133 - let node = 134 134 - serde_ipld_dagcbor::from_slice::<Node>(&data).map_err(Trip::BadCommit)?; 135 135 - 136 136 - // found node, make sure we remember 137 137 - self.stack.pop(); 138 138 - 139 139 - // queue up work on the found node next 140 140 - push_from_node(&mut self.stack, &node)?; 141 141 - } 142 142 - Need::Record { rkey, cid } => { 143 143 - log::trace!("need record {cid:?}"); 144 144 - let Some(data) = blocks.get_mut(cid) else { 145 145 - log::trace!("record block not found, resting"); 146 146 - return Ok(Step::Missing(*cid)); 147 147 - }; 148 148 - let rkey = rkey.clone(); 149 149 - let data = match data { 150 150 - MaybeProcessedBlock::Raw(data) => process(data.to_vec()), 151 151 - MaybeProcessedBlock::Processed(t) => t.clone(), 152 152 - }; 153 153 - 154 154 - // found node, make sure we remember 155 155 - self.stack.pop(); 156 156 - 157 157 - log::trace!("emitting a block as a step. depth={}", self.stack.len()); 158 158 - 159 159 - // rkeys *must* be in order or else the tree is invalid (or 160 160 - // we have a bug) 161 161 - if rkey <= self.prev { 162 162 - return Err(Trip::RkeyOutOfOrder); 163 163 - } 164 164 - self.prev = rkey.clone(); 165 165 - 166 166 - return Ok(Step::Step { rkey, data }); 167 167 - } 168 168 - } 169 169 - } 170 170 - } 171 171 - 172 172 - /// blocking!!!!!! 173 173 - pub fn disk_step<T: Processable>( 174 174 - &mut self, 175 175 - reader: &mut SqliteReader, 176 176 - process: impl Fn(Vec<u8>) -> T, 177 177 - ) -> Result<Step<T>, DiskTrip> { 178 178 - loop { 179 179 - let Some(mut need) = self.stack.last() else { 180 180 - log::trace!("tried to walk but we're actually done."); 181 181 - return Ok(Step::Finish); 182 182 - }; 183 183 - 184 184 - match &mut need { 185 185 - Need::Node(cid) => { 186 186 - let cid_bytes = cid.to_bytes(); 187 187 - log::trace!("need node {cid:?}"); 188 188 - let Some(block_bytes) = reader.get(cid_bytes)? else { 189 189 - log::trace!("node not found, resting"); 190 190 - return Ok(Step::Missing(*cid)); 191 191 - }; 192 192 - 193 193 - let block: MaybeProcessedBlock<T> = crate::drive::decode(&block_bytes)?; 194 194 - 195 195 - let MaybeProcessedBlock::Raw(data) = block else { 196 196 - return Err(Trip::BadCommitFingerprint.into()); 197 197 - }; 198 198 - let node = 199 199 - serde_ipld_dagcbor::from_slice::<Node>(&data).map_err(Trip::BadCommit)?; 200 200 - 201 201 - // found node, make sure we remember 202 202 - self.stack.pop(); 203 203 - 204 204 - // queue up work on the found node next 205 205 - push_from_node(&mut self.stack, &node).map_err(Trip::RkeyError)?; 206 206 - } 207 207 - Need::Record { rkey, cid } => { 208 208 - log::trace!("need record {cid:?}"); 209 209 - let cid_bytes = cid.to_bytes(); 210 210 - let Some(data_bytes) = reader.get(cid_bytes)? else { 211 211 - log::trace!("record block not found, resting"); 212 212 - return Ok(Step::Missing(*cid)); 213 213 - }; 214 214 - let data: MaybeProcessedBlock<T> = crate::drive::decode(&data_bytes)?; 215 215 - let rkey = rkey.clone(); 216 216 - let data = match data { 217 217 - MaybeProcessedBlock::Raw(data) => process(data), 218 218 - MaybeProcessedBlock::Processed(t) => t.clone(), 219 219 - }; 220 220 - 221 221 - // found node, make sure we remember 222 222 - self.stack.pop(); 223 223 - 224 224 - log::trace!("emitting a block as a step. depth={}", self.stack.len()); 225 225 - 226 226 - // rkeys *must* be in order or else the tree is invalid (or 227 227 - // we have a bug) 228 228 - if rkey <= self.prev { 229 229 - return Err(DiskTrip::Trip(Trip::RkeyOutOfOrder)); 230 230 - } 231 231 - self.prev = rkey.clone(); 232 232 - 233 233 - return Ok(Step::Step { rkey, data }); 234 234 - } 235 235 - } 236 236 - } 237 237 - } 238 238 - } 239 239 - 240 240 - #[cfg(test)] 241 241 - mod test { 242 242 - use super::*; 243 243 - // use crate::mst::Entry; 244 244 - 245 245 - fn cid1() -> Cid { 246 246 - "bafyreihixenvk3ahqbytas4hk4a26w43bh6eo3w6usjqtxkpzsvi655a3m" 247 247 - .parse() 248 248 - .unwrap() 249 249 - } 250 250 - // fn cid2() -> Cid { 251 251 - // "QmY7Yh4UquoXHLPFo2XbhXkhBvFoPwmQUSa92pxnxjQuPU" 252 252 - // .parse() 253 253 - // .unwrap() 254 254 - // } 255 255 - // fn cid3() -> Cid { 256 256 - // "bafybeigdyrzt5sfp7udm7hu76uh7y26nf3efuylqabf3oclgtqy55fbzdi" 257 257 - // .parse() 258 258 - // .unwrap() 259 259 - // } 260 260 - // fn cid4() -> Cid { 261 261 - // "QmbWqxBEKC3P8tqsKc98xmWNzrzDtRLMiMPL8wBuTGsMnR" 262 262 - // .parse() 263 263 - // .unwrap() 264 264 - // } 265 265 - // fn cid5() -> Cid { 266 266 - // "QmSnuWmxptJZdLJpKRarxBMS2Ju2oANVrgbr2xWbie9b2D" 267 267 - // .parse() 268 268 - // .unwrap() 269 269 - // } 270 270 - // fn cid6() -> Cid { 271 271 - // "QmdmQXB2mzChmMeKY47C43LxUdg1NDJ5MWcKMKxDu7RgQm" 272 272 - // .parse() 273 273 - // .unwrap() 274 274 - // } 275 275 - // fn cid7() -> Cid { 276 276 - // "bafybeiaysi4s6lnjev27ln5icwm6tueaw2vdykrtjkwiphwekaywqhcjze" 277 277 - // .parse() 278 278 - // .unwrap() 279 279 - // } 280 280 - // fn cid8() -> Cid { 281 281 - // "bafyreif3tfdpr5n4jdrbielmcapwvbpcthepfkwq2vwonmlhirbjmotedi" 282 282 - // .parse() 283 283 - // .unwrap() 284 284 - // } 285 285 - // fn cid9() -> Cid { 286 286 - // "bafyreicnokmhmrnlp2wjhyk2haep4tqxiptwfrp2rrs7rzq7uk766chqvq" 287 287 - // .parse() 288 288 - // .unwrap() 289 289 - // } 290 290 - 291 291 - #[test] 292 292 - fn test_next_from_node_empty() { 293 293 - let node = Node { 294 294 - left: None, 295 295 - entries: vec![], 296 296 - }; 297 297 - let mut stack = vec![]; 298 298 - push_from_node(&mut stack, &node).unwrap(); 299 299 - assert_eq!(stack.last(), None); 300 300 - } 301 301 - 302 302 - #[test] 303 303 - fn test_needs_from_node_just_left() { 304 304 - let node = Node { 305 305 - left: Some(cid1()), 306 306 - entries: vec![], 307 307 - }; 308 308 - let mut stack = vec![]; 309 309 - push_from_node(&mut stack, &node).unwrap(); 310 310 - assert_eq!(stack.last(), Some(Need::Node(cid1())).as_ref()); 311 311 - } 312 312 - 313 313 - // #[test] 314 314 - // fn test_needs_from_node_just_one_record() { 315 315 - // let node = Node { 316 316 - // left: None, 317 317 - // entries: vec![Entry { 318 318 - // keysuffix: "asdf".into(), 319 319 - // prefix_len: 0, 320 320 - // value: cid1(), 321 321 - // tree: None, 322 322 - // }], 323 323 - // }; 324 324 - // assert_eq!( 325 325 - // needs_from_node(node).unwrap(), 326 326 - // vec![Need::Record { 327 327 - // rkey: "asdf".into(), 328 328 - // cid: cid1(), 329 329 - // },] 330 330 - // ); 331 331 - // } 332 332 - 333 333 - // #[test] 334 334 - // fn test_needs_from_node_two_records() { 335 335 - // let node = Node { 336 336 - // left: None, 337 337 - // entries: vec![ 338 338 - // Entry { 339 339 - // keysuffix: "asdf".into(), 340 340 - // prefix_len: 0, 341 341 - // value: cid1(), 342 342 - // tree: None, 343 343 - // }, 344 344 - // Entry { 345 345 - // keysuffix: "gh".into(), 346 346 - // prefix_len: 2, 347 347 - // value: cid2(), 348 348 - // tree: None, 349 349 - // }, 350 350 - // ], 351 351 - // }; 352 352 - // assert_eq!( 353 353 - // needs_from_node(node).unwrap(), 354 354 - // vec![ 355 355 - // Need::Record { 356 356 - // rkey: "asdf".into(), 357 357 - // cid: cid1(), 358 358 - // }, 359 359 - // Need::Record { 360 360 - // rkey: "asgh".into(), 361 361 - // cid: cid2(), 362 362 - // }, 363 363 - // ] 364 364 - // ); 365 365 - // } 366 366 - 367 367 - // #[test] 368 368 - // fn test_needs_from_node_with_both() { 369 369 - // let node = Node { 370 370 - // left: None, 371 371 - // entries: vec![Entry { 372 372 - // keysuffix: "asdf".into(), 373 373 - // prefix_len: 0, 374 374 - // value: cid1(), 375 375 - // tree: Some(cid2()), 376 376 - // }], 377 377 - // }; 378 378 - // assert_eq!( 379 379 - // needs_from_node(node).unwrap(), 380 380 - // vec![ 381 381 - // Need::Record { 382 382 - // rkey: "asdf".into(), 383 383 - // cid: cid1(), 384 384 - // }, 385 385 - // Need::Node(cid2()), 386 386 - // ] 387 387 - // ); 388 388 - // } 389 389 - 390 390 - // #[test] 391 391 - // fn test_needs_from_node_left_and_record() { 392 392 - // let node = Node { 393 393 - // left: Some(cid1()), 394 394 - // entries: vec![Entry { 395 395 - // keysuffix: "asdf".into(), 396 396 - // prefix_len: 0, 397 397 - // value: cid2(), 398 398 - // tree: None, 399 399 - // }], 400 400 - // }; 401 401 - // assert_eq!( 402 402 - // needs_from_node(node).unwrap(), 403 403 - // vec![ 404 404 - // Need::Node(cid1()), 405 405 - // Need::Record { 406 406 - // rkey: "asdf".into(), 407 407 - // cid: cid2(), 408 408 - // }, 409 409 - // ] 410 410 - // ); 411 411 - // } 412 412 - 413 413 - // #[test] 414 414 - // fn test_needs_from_full_node() { 415 415 - // let node = Node { 416 416 - // left: Some(cid1()), 417 417 - // entries: vec![ 418 418 - // Entry { 419 419 - // keysuffix: "asdf".into(), 420 420 - // prefix_len: 0, 421 421 - // value: cid2(), 422 422 - // tree: Some(cid3()), 423 423 - // }, 424 424 - // Entry { 425 425 - // keysuffix: "ghi".into(), 426 426 - // prefix_len: 1, 427 427 - // value: cid4(), 428 428 - // tree: Some(cid5()), 429 429 - // }, 430 430 - // Entry { 431 431 - // keysuffix: "jkl".into(), 432 432 - // prefix_len: 2, 433 433 - // value: cid6(), 434 434 - // tree: Some(cid7()), 435 435 - // }, 436 436 - // Entry { 437 437 - // keysuffix: "mno".into(), 438 438 - // prefix_len: 4, 439 439 - // value: cid8(), 440 440 - // tree: Some(cid9()), 441 441 - // }, 442 442 - // ], 443 443 - // }; 444 444 - // assert_eq!( 445 445 - // needs_from_node(node).unwrap(), 446 446 - // vec![ 447 447 - // Need::Node(cid1()), 448 448 - // Need::Record { 449 449 - // rkey: "asdf".into(), 450 450 - // cid: cid2(), 451 451 - // }, 452 452 - // Need::Node(cid3()), 453 453 - // Need::Record { 454 454 - // rkey: "aghi".into(), 455 455 - // cid: cid4(), 456 456 - // }, 457 457 - // Need::Node(cid5()), 458 458 - // Need::Record { 459 459 - // rkey: "agjkl".into(), 460 460 - // cid: cid6(), 461 461 - // }, 462 462 - // Need::Node(cid7()), 463 463 - // Need::Record { 464 464 - // rkey: "agjkmno".into(), 465 465 - // cid: cid8(), 466 466 - // }, 467 467 - // Need::Node(cid9()), 468 468 - // ] 469 469 - // ); 470 470 - // } 471 471 - }

+34 -31

tests/non-huge-cars.rs

··· 1 1 extern crate repo_stream; 2 2 - use futures::TryStreamExt; 3 3 - use iroh_car::CarReader; 4 4 - use std::convert::Infallible; 2 2 + use repo_stream::Driver; 5 3 4 4 + const EMPTY_CAR: &'static [u8] = include_bytes!("../car-samples/empty.car"); 6 5 const TINY_CAR: &'static [u8] = include_bytes!("../car-samples/tiny.car"); 7 6 const LITTLE_CAR: &'static [u8] = include_bytes!("../car-samples/little.car"); 8 7 const MIDSIZE_CAR: &'static [u8] = include_bytes!("../car-samples/midsize.car"); 9 8 10 10 - async fn test_car(bytes: &[u8], expected_records: usize, expected_sum: usize) { 11 11 - let reader = CarReader::new(bytes).await.unwrap(); 12 12 - 13 13 - let root = reader 14 14 - .header() 15 15 - .roots() 16 16 - .first() 17 17 - .ok_or("missing root") 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 + ) { 15 15 + let mut driver = match Driver::load_car(bytes, |block| block.len(), 10 /* MiB */) 16 16 + .await 18 17 .unwrap() 19 19 - .clone(); 20 20 - 21 21 - let stream = std::pin::pin!(reader.stream()); 22 22 - 23 23 - let (_commit, v) = 24 24 - repo_stream::drive::Vehicle::init(root, stream, |block| Ok::<_, Infallible>(block.len())) 25 25 - .await 26 26 - .unwrap(); 27 27 - let mut record_stream = std::pin::pin!(v.stream()); 18 18 + { 19 19 + Driver::Memory(_commit, mem_driver) => mem_driver, 20 20 + Driver::Disk(_) => panic!("too big"), 21 21 + }; 28 22 29 23 let mut records = 0; 30 24 let mut sum = 0; 31 25 let mut found_bsky_profile = false; 32 26 let mut prev_rkey = "".to_string(); 33 33 - while let Some((rkey, size)) = record_stream.try_next().await.unwrap() { 34 34 - records += 1; 35 35 - sum += size; 36 36 - if rkey == "app.bsky.actor.profile/self" { 37 37 - found_bsky_profile = true; 27 27 + 28 28 + while let Some(pairs) = driver.next_chunk(256).await.unwrap() { 29 29 + for (rkey, size) in pairs { 30 30 + records += 1; 31 31 + sum += size; 32 32 + if rkey == "app.bsky.actor.profile/self" { 33 33 + found_bsky_profile = true; 34 34 + } 35 35 + assert!(rkey > prev_rkey, "rkeys are streamed in order"); 36 36 + prev_rkey = rkey; 38 37 } 39 39 - assert!(rkey > prev_rkey, "rkeys are streamed in order"); 40 40 - prev_rkey = rkey; 41 38 } 39 39 + 42 40 assert_eq!(records, expected_records); 43 41 assert_eq!(sum, expected_sum); 44 44 - 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 45 48 } 46 49 47 50 #[tokio::test] 48 51 async fn test_tiny_car() { 49 49 - test_car(TINY_CAR, 8, 2071).await 52 52 + test_car(TINY_CAR, 8, 2071, true).await 50 53 } 51 54 52 55 #[tokio::test] 53 56 async fn test_little_car() { 54 54 - test_car(LITTLE_CAR, 278, 246960).await 57 57 + test_car(LITTLE_CAR, 278, 246960, true).await 55 58 } 56 59 57 60 #[tokio::test] 58 61 async fn test_midsize_car() { 59 59 - test_car(MIDSIZE_CAR, 11585, 3741393).await 62 62 + test_car(MIDSIZE_CAR, 11585, 3741393, true).await 60 63 }