File
Blob: test/pack-indexer.resolve.ref.worker.test.ts
| 1 | import { describe, expect, it } from "vitest"; |
| 2 | import { env } from "cloudflare:workers"; |
| 3 | import { buildPack, buildAppendOnlyDelta, buildCopyPrefixDelta } from "./util/git-pack"; |
| 4 | import { createTestCacheContext } from "./util/pack-first"; |
| 5 | import { uniqueRepoId } from "./util/test-helpers"; |
| 6 | import { |
| 7 | makeActiveCatalogRow, |
| 8 | makeCountSubrequest, |
| 9 | makeLimiter, |
| 10 | makeTracingLimiter, |
| 11 | packIndexerLog as log, |
| 12 | } from "./util/pack-indexer.helpers"; |
| 13 | |
| 14 | import { scanPack, resolveDeltasAndWriteIdx } from "@/worker/git/pack/indexer"; |
| 15 | import { computeOid, encodeGitObject } from "@/worker/git/core/objects"; |
| 16 | import { bytesToHex } from "@/worker/common/hex"; |
| 17 | import { findOidIndex } from "@/worker/git/object-store/idxView"; |
| 18 | import { packIndexKey } from "@/worker/keys"; |
| 19 | |
| 20 | describe("resolveDeltasAndWriteIdx REF_DELTA", () => { |
| 21 | it("resolves REF_DELTA with external base from the provided active catalog snapshot", async () => { |
| 22 | const baseBlobPayload = new TextEncoder().encode("external base content\n"); |
| 23 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 24 | const basePack = await buildPack([{ type: "blob", payload: baseBlobPayload }]); |
| 25 | |
| 26 | const basePackKey = "test/ref-delta-base.pack"; |
| 27 | await env.REPO_BUCKET.put(basePackKey, basePack); |
| 28 | const baseHead = await env.REPO_BUCKET.head(basePackKey); |
| 29 | |
| 30 | const baseScan = await scanPack({ |
| 31 | env, |
| 32 | packKey: basePackKey, |
| 33 | packSize: baseHead!.size, |
| 34 | limiter: makeLimiter(), |
| 35 | countSubrequest: () => {}, |
| 36 | log, |
| 37 | }); |
| 38 | const repoId = uniqueRepoId(); |
| 39 | const baseResolve = await resolveDeltasAndWriteIdx({ |
| 40 | env, |
| 41 | packKey: basePackKey, |
| 42 | packSize: baseHead!.size, |
| 43 | limiter: makeLimiter(), |
| 44 | countSubrequest: () => {}, |
| 45 | log, |
| 46 | scanResult: baseScan, |
| 47 | repoId, |
| 48 | }); |
| 49 | |
| 50 | const suffix = new TextEncoder().encode("extra data\n"); |
| 51 | const delta = buildAppendOnlyDelta(baseBlobPayload, suffix); |
| 52 | const thinPack = await buildPack([{ type: "ref-delta", baseOid: baseBlob.oid, delta }]); |
| 53 | |
| 54 | const thinPackKey = "test/ref-delta-thin.pack"; |
| 55 | await env.REPO_BUCKET.put(thinPackKey, thinPack); |
| 56 | const thinHead = await env.REPO_BUCKET.head(thinPackKey); |
| 57 | |
| 58 | const thinScan = await scanPack({ |
| 59 | env, |
| 60 | packKey: thinPackKey, |
| 61 | packSize: thinHead!.size, |
| 62 | limiter: makeLimiter(), |
| 63 | countSubrequest: () => {}, |
| 64 | log, |
| 65 | }); |
| 66 | |
| 67 | const cacheCtx = createTestCacheContext("http://localhost/test"); |
| 68 | const thinResolve = await resolveDeltasAndWriteIdx({ |
| 69 | env, |
| 70 | packKey: thinPackKey, |
| 71 | packSize: thinHead!.size, |
| 72 | limiter: makeLimiter(), |
| 73 | countSubrequest: () => {}, |
| 74 | log, |
| 75 | scanResult: thinScan, |
| 76 | activeCatalog: [ |
| 77 | makeActiveCatalogRow({ |
| 78 | packKey: basePackKey, |
| 79 | packBytes: baseHead!.size, |
| 80 | objectCount: baseScan.objectCount, |
| 81 | idxBytes: baseResolve.idxBytes, |
| 82 | }), |
| 83 | ], |
| 84 | cacheCtx, |
| 85 | repoId, |
| 86 | }); |
| 87 | |
| 88 | expect(thinScan.table.resolved[0]).toBe(1); |
| 89 | expect(thinResolve.objectCount).toBe(1); |
| 90 | |
| 91 | const expectedPayload = new Uint8Array(baseBlobPayload.length + suffix.length); |
| 92 | expectedPayload.set(baseBlobPayload, 0); |
| 93 | expectedPayload.set(suffix, baseBlobPayload.length); |
| 94 | const expectedOid = await computeOid("blob", expectedPayload); |
| 95 | expect(bytesToHex(thinScan.table.oids.subarray(0, 20))).toBe(expectedOid); |
| 96 | }); |
| 97 | |
| 98 | it("rejects a thin REF_DELTA when the external base is missing", async () => { |
| 99 | const fakeBaseOid = "ab".repeat(20); |
| 100 | const delta = buildAppendOnlyDelta( |
| 101 | new TextEncoder().encode("base\n"), |
| 102 | new TextEncoder().encode("next\n") |
| 103 | ); |
| 104 | const thinPack = await buildPack([{ type: "ref-delta", baseOid: fakeBaseOid, delta }]); |
| 105 | |
| 106 | const packKey = "test/ref-delta-missing-base.pack"; |
| 107 | await env.REPO_BUCKET.put(packKey, thinPack); |
| 108 | const head = await env.REPO_BUCKET.head(packKey); |
| 109 | |
| 110 | const scanResult = await scanPack({ |
| 111 | env, |
| 112 | packKey, |
| 113 | packSize: head!.size, |
| 114 | limiter: makeLimiter(), |
| 115 | countSubrequest: () => {}, |
| 116 | log, |
| 117 | }); |
| 118 | |
| 119 | const repoId = uniqueRepoId(); |
| 120 | const cacheCtx = createTestCacheContext("http://localhost/test"); |
| 121 | await expect( |
| 122 | resolveDeltasAndWriteIdx({ |
| 123 | env, |
| 124 | packKey, |
| 125 | packSize: head!.size, |
| 126 | limiter: makeLimiter(), |
| 127 | countSubrequest: () => {}, |
| 128 | log, |
| 129 | scanResult, |
| 130 | repoId, |
| 131 | cacheCtx, |
| 132 | }) |
| 133 | ).rejects.toThrow(/could not be resolved/); |
| 134 | |
| 135 | const idxObj = await env.REPO_BUCKET.get(packIndexKey(packKey)); |
| 136 | expect(idxObj).toBeNull(); |
| 137 | }); |
| 138 | |
| 139 | it("prefers the explicit active catalog snapshot over stale memoized pack-catalog state", async () => { |
| 140 | const baseBlobPayload = new TextEncoder().encode("external base content\n"); |
| 141 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 142 | const basePack = await buildPack([{ type: "blob", payload: baseBlobPayload }]); |
| 143 | |
| 144 | const basePackKey = "test/ref-delta-stale-memo-base.pack"; |
| 145 | await env.REPO_BUCKET.put(basePackKey, basePack); |
| 146 | const baseHead = await env.REPO_BUCKET.head(basePackKey); |
| 147 | |
| 148 | const baseScan = await scanPack({ |
| 149 | env, |
| 150 | packKey: basePackKey, |
| 151 | packSize: baseHead!.size, |
| 152 | limiter: makeLimiter(), |
| 153 | countSubrequest: () => {}, |
| 154 | log, |
| 155 | }); |
| 156 | const repoId = uniqueRepoId(); |
| 157 | const baseResolve = await resolveDeltasAndWriteIdx({ |
| 158 | env, |
| 159 | packKey: basePackKey, |
| 160 | packSize: baseHead!.size, |
| 161 | limiter: makeLimiter(), |
| 162 | countSubrequest: () => {}, |
| 163 | log, |
| 164 | scanResult: baseScan, |
| 165 | repoId, |
| 166 | }); |
| 167 | |
| 168 | const suffix = new TextEncoder().encode("active catalog wins\n"); |
| 169 | const thinPack = await buildPack([ |
| 170 | { |
| 171 | type: "ref-delta", |
| 172 | baseOid: baseBlob.oid, |
| 173 | delta: buildAppendOnlyDelta(baseBlobPayload, suffix), |
| 174 | }, |
| 175 | ]); |
| 176 | |
| 177 | const thinPackKey = "test/ref-delta-stale-memo-thin.pack"; |
| 178 | await env.REPO_BUCKET.put(thinPackKey, thinPack); |
| 179 | const thinHead = await env.REPO_BUCKET.head(thinPackKey); |
| 180 | |
| 181 | const thinScan = await scanPack({ |
| 182 | env, |
| 183 | packKey: thinPackKey, |
| 184 | packSize: thinHead!.size, |
| 185 | limiter: makeLimiter(), |
| 186 | countSubrequest: () => {}, |
| 187 | log, |
| 188 | }); |
| 189 | |
| 190 | const cacheCtx = createTestCacheContext("http://localhost/test"); |
| 191 | cacheCtx.memo = { |
| 192 | ...(cacheCtx.memo || {}), |
| 193 | packCatalog: [], |
| 194 | }; |
| 195 | |
| 196 | const thinResolve = await resolveDeltasAndWriteIdx({ |
| 197 | env, |
| 198 | packKey: thinPackKey, |
| 199 | packSize: thinHead!.size, |
| 200 | limiter: makeLimiter(), |
| 201 | countSubrequest: () => {}, |
| 202 | log, |
| 203 | scanResult: thinScan, |
| 204 | activeCatalog: [ |
| 205 | makeActiveCatalogRow({ |
| 206 | packKey: basePackKey, |
| 207 | packBytes: baseHead!.size, |
| 208 | objectCount: baseScan.objectCount, |
| 209 | idxBytes: baseResolve.idxBytes, |
| 210 | }), |
| 211 | ], |
| 212 | cacheCtx, |
| 213 | repoId, |
| 214 | }); |
| 215 | |
| 216 | expect(thinResolve.objectCount).toBe(1); |
| 217 | expect(thinScan.table.resolved[0]).toBe(1); |
| 218 | expect(cacheCtx.memo?.packCatalog?.map((row) => row.packKey)).toEqual([basePackKey]); |
| 219 | }); |
| 220 | |
| 221 | it("resolves same-pack REF_DELTA chains whose bases appear later in the pack", async () => { |
| 222 | const baseBlobPayload = new TextEncoder().encode("base\n"); |
| 223 | const midSuffix = new TextEncoder().encode("mid\n"); |
| 224 | const finalSuffix = new TextEncoder().encode("final\n"); |
| 225 | |
| 226 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 227 | |
| 228 | const midPayload = new Uint8Array(baseBlobPayload.length + midSuffix.length); |
| 229 | midPayload.set(baseBlobPayload, 0); |
| 230 | midPayload.set(midSuffix, baseBlobPayload.length); |
| 231 | const midOid = await computeOid("blob", midPayload); |
| 232 | |
| 233 | const finalPayload = new Uint8Array(midPayload.length + finalSuffix.length); |
| 234 | finalPayload.set(midPayload, 0); |
| 235 | finalPayload.set(finalSuffix, midPayload.length); |
| 236 | const finalOid = await computeOid("blob", finalPayload); |
| 237 | |
| 238 | const midDelta = buildAppendOnlyDelta(baseBlobPayload, midSuffix); |
| 239 | const finalDelta = buildAppendOnlyDelta(midPayload, finalSuffix); |
| 240 | |
| 241 | const packBytes = await buildPack([ |
| 242 | { type: "ref-delta", baseOid: midOid, delta: finalDelta }, |
| 243 | { type: "ref-delta", baseOid: baseBlob.oid, delta: midDelta }, |
| 244 | { type: "blob", payload: baseBlobPayload }, |
| 245 | ]); |
| 246 | |
| 247 | const packKey = "test/resolve-forward-ref-chain.pack"; |
| 248 | await env.REPO_BUCKET.put(packKey, packBytes); |
| 249 | const head = await env.REPO_BUCKET.head(packKey); |
| 250 | |
| 251 | const scanResult = await scanPack({ |
| 252 | env, |
| 253 | packKey, |
| 254 | packSize: head!.size, |
| 255 | limiter: makeLimiter(), |
| 256 | countSubrequest: () => {}, |
| 257 | log, |
| 258 | }); |
| 259 | |
| 260 | const repoId = uniqueRepoId(); |
| 261 | await resolveDeltasAndWriteIdx({ |
| 262 | env, |
| 263 | packKey, |
| 264 | packSize: head!.size, |
| 265 | limiter: makeLimiter(), |
| 266 | countSubrequest: () => {}, |
| 267 | log, |
| 268 | scanResult, |
| 269 | repoId, |
| 270 | }); |
| 271 | |
| 272 | expect(scanResult.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(3); |
| 273 | expect(bytesToHex(scanResult.table.oids.subarray(0, 20))).toBe(finalOid); |
| 274 | expect(bytesToHex(scanResult.table.oids.subarray(20, 40))).toBe(midOid); |
| 275 | expect(bytesToHex(scanResult.table.oids.subarray(40, 60))).toBe(baseBlob.oid); |
| 276 | }); |
| 277 | |
| 278 | it("uses an existing idx to map same-pack REF_DELTA bases during refs-only backfill", async () => { |
| 279 | const baseBlobPayload = new TextEncoder().encode("base\n"); |
| 280 | const midSuffix = new TextEncoder().encode("mid\n"); |
| 281 | const finalSuffix = new TextEncoder().encode("final\n"); |
| 282 | |
| 283 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 284 | |
| 285 | const midPayload = new Uint8Array(baseBlobPayload.length + midSuffix.length); |
| 286 | midPayload.set(baseBlobPayload, 0); |
| 287 | midPayload.set(midSuffix, baseBlobPayload.length); |
| 288 | const midOid = await computeOid("blob", midPayload); |
| 289 | |
| 290 | const finalPayload = new Uint8Array(midPayload.length + finalSuffix.length); |
| 291 | finalPayload.set(midPayload, 0); |
| 292 | finalPayload.set(finalSuffix, midPayload.length); |
| 293 | const finalOid = await computeOid("blob", finalPayload); |
| 294 | |
| 295 | const packBytes = await buildPack([ |
| 296 | { |
| 297 | type: "ref-delta", |
| 298 | baseOid: midOid, |
| 299 | delta: buildAppendOnlyDelta(midPayload, finalSuffix), |
| 300 | }, |
| 301 | { |
| 302 | type: "ref-delta", |
| 303 | baseOid: baseBlob.oid, |
| 304 | delta: buildAppendOnlyDelta(baseBlobPayload, midSuffix), |
| 305 | }, |
| 306 | { type: "blob", payload: baseBlobPayload }, |
| 307 | ]); |
| 308 | |
| 309 | const packKey = "test/resolve-ref-chain-existing-idx.pack"; |
| 310 | await env.REPO_BUCKET.put(packKey, packBytes); |
| 311 | const head = await env.REPO_BUCKET.head(packKey); |
| 312 | |
| 313 | const initialScan = await scanPack({ |
| 314 | env, |
| 315 | packKey, |
| 316 | packSize: head!.size, |
| 317 | limiter: makeLimiter(), |
| 318 | countSubrequest: () => {}, |
| 319 | log, |
| 320 | }); |
| 321 | const repoId = uniqueRepoId(); |
| 322 | const initialResolve = await resolveDeltasAndWriteIdx({ |
| 323 | env, |
| 324 | packKey, |
| 325 | packSize: head!.size, |
| 326 | limiter: makeLimiter(), |
| 327 | countSubrequest: () => {}, |
| 328 | log, |
| 329 | scanResult: initialScan, |
| 330 | repoId, |
| 331 | }); |
| 332 | |
| 333 | const midOidIndex = findOidIndex(initialResolve.idxView, midOid); |
| 334 | expect(midOidIndex).toBeGreaterThanOrEqual(0); |
| 335 | expect(midOidIndex).not.toBe(1); |
| 336 | expect(initialResolve.idxView.offsets[midOidIndex]).toBe(initialScan.table.offsets[1]); |
| 337 | |
| 338 | const backfillScan = await scanPack({ |
| 339 | env, |
| 340 | packKey, |
| 341 | packSize: head!.size, |
| 342 | limiter: makeLimiter(), |
| 343 | countSubrequest: () => {}, |
| 344 | log, |
| 345 | }); |
| 346 | |
| 347 | const backfillResolve = await resolveDeltasAndWriteIdx({ |
| 348 | env, |
| 349 | packKey, |
| 350 | packSize: head!.size, |
| 351 | limiter: makeLimiter(), |
| 352 | countSubrequest: () => {}, |
| 353 | log, |
| 354 | scanResult: backfillScan, |
| 355 | repoId, |
| 356 | activeCatalog: [], |
| 357 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 358 | writeIdx: false, |
| 359 | existingIdxView: initialResolve.idxView, |
| 360 | }); |
| 361 | |
| 362 | expect(backfillResolve.idxBytes).toBe(0); |
| 363 | expect(backfillScan.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(3); |
| 364 | expect(bytesToHex(backfillScan.table.oids.subarray(0, 20))).toBe(finalOid); |
| 365 | expect(bytesToHex(backfillScan.table.oids.subarray(20, 40))).toBe(midOid); |
| 366 | expect(bytesToHex(backfillScan.table.oids.subarray(40, 60))).toBe(baseBlob.oid); |
| 367 | }); |
| 368 | |
| 369 | it("uses another duplicate idx row for identity same-pack REF_DELTA backfill", async () => { |
| 370 | const baseBlobPayload = new TextEncoder().encode("duplicate oid base\n"); |
| 371 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 372 | const identityDelta = buildAppendOnlyDelta(baseBlobPayload, new Uint8Array(0)); |
| 373 | |
| 374 | const packBytes = await buildPack([ |
| 375 | { type: "ref-delta", baseOid: baseBlob.oid, delta: identityDelta }, |
| 376 | { type: "blob", payload: baseBlobPayload }, |
| 377 | ]); |
| 378 | |
| 379 | const packKey = "test/resolve-ref-identity-existing-idx.pack"; |
| 380 | await env.REPO_BUCKET.put(packKey, packBytes); |
| 381 | const head = await env.REPO_BUCKET.head(packKey); |
| 382 | |
| 383 | const initialScan = await scanPack({ |
| 384 | env, |
| 385 | packKey, |
| 386 | packSize: head!.size, |
| 387 | limiter: makeLimiter(), |
| 388 | countSubrequest: () => {}, |
| 389 | log, |
| 390 | }); |
| 391 | const repoId = uniqueRepoId(); |
| 392 | const initialResolve = await resolveDeltasAndWriteIdx({ |
| 393 | env, |
| 394 | packKey, |
| 395 | packSize: head!.size, |
| 396 | limiter: makeLimiter(), |
| 397 | countSubrequest: () => {}, |
| 398 | log, |
| 399 | scanResult: initialScan, |
| 400 | repoId, |
| 401 | }); |
| 402 | |
| 403 | expect(findOidIndex(initialResolve.idxView, baseBlob.oid)).toBe(0); |
| 404 | |
| 405 | const backfillScan = await scanPack({ |
| 406 | env, |
| 407 | packKey, |
| 408 | packSize: head!.size, |
| 409 | limiter: makeLimiter(), |
| 410 | countSubrequest: () => {}, |
| 411 | log, |
| 412 | }); |
| 413 | |
| 414 | const backfillResolve = await resolveDeltasAndWriteIdx({ |
| 415 | env, |
| 416 | packKey, |
| 417 | packSize: head!.size, |
| 418 | limiter: makeLimiter(), |
| 419 | countSubrequest: () => {}, |
| 420 | log, |
| 421 | scanResult: backfillScan, |
| 422 | repoId, |
| 423 | activeCatalog: [], |
| 424 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 425 | writeIdx: false, |
| 426 | existingIdxView: initialResolve.idxView, |
| 427 | }); |
| 428 | |
| 429 | expect(backfillResolve.idxBytes).toBe(0); |
| 430 | expect(backfillScan.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(2); |
| 431 | expect(bytesToHex(backfillScan.table.oids.subarray(0, 20))).toBe(baseBlob.oid); |
| 432 | expect(bytesToHex(backfillScan.table.oids.subarray(20, 40))).toBe(baseBlob.oid); |
| 433 | }); |
| 434 | |
| 435 | it("falls back to an external base for target-pack identity REF_DELTA duplicates", async () => { |
| 436 | const baseBlobPayload = new TextEncoder().encode("external duplicate oid base\n"); |
| 437 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 438 | const repoId = uniqueRepoId(); |
| 439 | |
| 440 | const basePackKey = "test/ref-delta-identity-external-base.pack"; |
| 441 | const basePack = await buildPack([{ type: "blob", payload: baseBlobPayload }]); |
| 442 | await env.REPO_BUCKET.put(basePackKey, basePack); |
| 443 | const baseHead = await env.REPO_BUCKET.head(basePackKey); |
| 444 | const baseScan = await scanPack({ |
| 445 | env, |
| 446 | packKey: basePackKey, |
| 447 | packSize: baseHead!.size, |
| 448 | limiter: makeLimiter(), |
| 449 | countSubrequest: () => {}, |
| 450 | log, |
| 451 | }); |
| 452 | const baseResolve = await resolveDeltasAndWriteIdx({ |
| 453 | env, |
| 454 | packKey: basePackKey, |
| 455 | packSize: baseHead!.size, |
| 456 | limiter: makeLimiter(), |
| 457 | countSubrequest: () => {}, |
| 458 | log, |
| 459 | scanResult: baseScan, |
| 460 | repoId, |
| 461 | }); |
| 462 | |
| 463 | const activeCatalog = [ |
| 464 | makeActiveCatalogRow({ |
| 465 | packKey: basePackKey, |
| 466 | packBytes: baseHead!.size, |
| 467 | objectCount: baseScan.objectCount, |
| 468 | idxBytes: baseResolve.idxBytes, |
| 469 | }), |
| 470 | ]; |
| 471 | const identityDelta = buildAppendOnlyDelta(baseBlobPayload, new Uint8Array(0)); |
| 472 | const targetPack = await buildPack([ |
| 473 | { type: "ref-delta", baseOid: baseBlob.oid, delta: identityDelta }, |
| 474 | { type: "ref-delta", baseOid: baseBlob.oid, delta: identityDelta }, |
| 475 | ]); |
| 476 | |
| 477 | const targetPackKey = "test/ref-delta-identity-external-target.pack"; |
| 478 | await env.REPO_BUCKET.put(targetPackKey, targetPack); |
| 479 | const targetHead = await env.REPO_BUCKET.head(targetPackKey); |
| 480 | const initialScan = await scanPack({ |
| 481 | env, |
| 482 | packKey: targetPackKey, |
| 483 | packSize: targetHead!.size, |
| 484 | limiter: makeLimiter(), |
| 485 | countSubrequest: () => {}, |
| 486 | log, |
| 487 | }); |
| 488 | const initialResolve = await resolveDeltasAndWriteIdx({ |
| 489 | env, |
| 490 | packKey: targetPackKey, |
| 491 | packSize: targetHead!.size, |
| 492 | limiter: makeLimiter(), |
| 493 | countSubrequest: () => {}, |
| 494 | log, |
| 495 | scanResult: initialScan, |
| 496 | activeCatalog, |
| 497 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 498 | repoId, |
| 499 | }); |
| 500 | |
| 501 | const backfillScan = await scanPack({ |
| 502 | env, |
| 503 | packKey: targetPackKey, |
| 504 | packSize: targetHead!.size, |
| 505 | limiter: makeLimiter(), |
| 506 | countSubrequest: () => {}, |
| 507 | log, |
| 508 | }); |
| 509 | const backfillResolve = await resolveDeltasAndWriteIdx({ |
| 510 | env, |
| 511 | packKey: targetPackKey, |
| 512 | packSize: targetHead!.size, |
| 513 | limiter: makeLimiter(), |
| 514 | countSubrequest: () => {}, |
| 515 | log, |
| 516 | scanResult: backfillScan, |
| 517 | activeCatalog, |
| 518 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 519 | repoId, |
| 520 | writeIdx: false, |
| 521 | existingIdxView: initialResolve.idxView, |
| 522 | }); |
| 523 | |
| 524 | expect(backfillResolve.idxBytes).toBe(0); |
| 525 | expect(backfillScan.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(2); |
| 526 | expect(bytesToHex(backfillScan.table.oids.subarray(0, 20))).toBe(baseBlob.oid); |
| 527 | expect(bytesToHex(backfillScan.table.oids.subarray(20, 40))).toBe(baseBlob.oid); |
| 528 | }); |
| 529 | |
| 530 | it("tries older external duplicate bases when the newest duplicate points back to the target pack", async () => { |
| 531 | const targetPayload = new TextEncoder().encode("target prefix\n"); |
| 532 | const duplicateSuffix = new TextEncoder().encode("duplicate suffix\n"); |
| 533 | const duplicatePayload = new Uint8Array(targetPayload.length + duplicateSuffix.length); |
| 534 | duplicatePayload.set(targetPayload, 0); |
| 535 | duplicatePayload.set(duplicateSuffix, targetPayload.length); |
| 536 | |
| 537 | const targetOid = await computeOid("blob", targetPayload); |
| 538 | const duplicateOid = await computeOid("blob", duplicatePayload); |
| 539 | const repoId = uniqueRepoId(); |
| 540 | |
| 541 | const olderPackKey = "test/ref-delta-cycle-older-base.pack"; |
| 542 | const olderPack = await buildPack([{ type: "blob", payload: duplicatePayload }]); |
| 543 | await env.REPO_BUCKET.put(olderPackKey, olderPack); |
| 544 | const olderHead = await env.REPO_BUCKET.head(olderPackKey); |
| 545 | const olderScan = await scanPack({ |
| 546 | env, |
| 547 | packKey: olderPackKey, |
| 548 | packSize: olderHead!.size, |
| 549 | limiter: makeLimiter(), |
| 550 | countSubrequest: () => {}, |
| 551 | log, |
| 552 | }); |
| 553 | const olderResolve = await resolveDeltasAndWriteIdx({ |
| 554 | env, |
| 555 | packKey: olderPackKey, |
| 556 | packSize: olderHead!.size, |
| 557 | limiter: makeLimiter(), |
| 558 | countSubrequest: () => {}, |
| 559 | log, |
| 560 | scanResult: olderScan, |
| 561 | repoId, |
| 562 | }); |
| 563 | const olderRow = makeActiveCatalogRow({ |
| 564 | packKey: olderPackKey, |
| 565 | packBytes: olderHead!.size, |
| 566 | objectCount: olderScan.objectCount, |
| 567 | idxBytes: olderResolve.idxBytes, |
| 568 | }); |
| 569 | |
| 570 | const targetPackKey = "test/ref-delta-cycle-target.pack"; |
| 571 | const targetPack = await buildPack([ |
| 572 | { |
| 573 | type: "ref-delta", |
| 574 | baseOid: duplicateOid, |
| 575 | delta: buildCopyPrefixDelta(duplicatePayload, targetPayload.length), |
| 576 | }, |
| 577 | ]); |
| 578 | await env.REPO_BUCKET.put(targetPackKey, targetPack); |
| 579 | const targetHead = await env.REPO_BUCKET.head(targetPackKey); |
| 580 | const targetScan = await scanPack({ |
| 581 | env, |
| 582 | packKey: targetPackKey, |
| 583 | packSize: targetHead!.size, |
| 584 | limiter: makeLimiter(), |
| 585 | countSubrequest: () => {}, |
| 586 | log, |
| 587 | }); |
| 588 | const targetResolve = await resolveDeltasAndWriteIdx({ |
| 589 | env, |
| 590 | packKey: targetPackKey, |
| 591 | packSize: targetHead!.size, |
| 592 | limiter: makeLimiter(), |
| 593 | countSubrequest: () => {}, |
| 594 | log, |
| 595 | scanResult: targetScan, |
| 596 | activeCatalog: [olderRow], |
| 597 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 598 | repoId, |
| 599 | }); |
| 600 | const targetRow = makeActiveCatalogRow({ |
| 601 | packKey: targetPackKey, |
| 602 | packBytes: targetHead!.size, |
| 603 | objectCount: targetScan.objectCount, |
| 604 | idxBytes: targetResolve.idxBytes, |
| 605 | }); |
| 606 | |
| 607 | const newerPackKey = "test/ref-delta-cycle-newer-duplicate.pack"; |
| 608 | const newerPack = await buildPack([ |
| 609 | { |
| 610 | type: "ref-delta", |
| 611 | baseOid: targetOid, |
| 612 | delta: buildAppendOnlyDelta(targetPayload, duplicateSuffix), |
| 613 | }, |
| 614 | ]); |
| 615 | await env.REPO_BUCKET.put(newerPackKey, newerPack); |
| 616 | const newerHead = await env.REPO_BUCKET.head(newerPackKey); |
| 617 | const newerScan = await scanPack({ |
| 618 | env, |
| 619 | packKey: newerPackKey, |
| 620 | packSize: newerHead!.size, |
| 621 | limiter: makeLimiter(), |
| 622 | countSubrequest: () => {}, |
| 623 | log, |
| 624 | }); |
| 625 | const newerResolve = await resolveDeltasAndWriteIdx({ |
| 626 | env, |
| 627 | packKey: newerPackKey, |
| 628 | packSize: newerHead!.size, |
| 629 | limiter: makeLimiter(), |
| 630 | countSubrequest: () => {}, |
| 631 | log, |
| 632 | scanResult: newerScan, |
| 633 | activeCatalog: [targetRow, olderRow], |
| 634 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 635 | repoId, |
| 636 | }); |
| 637 | const newerRow = makeActiveCatalogRow({ |
| 638 | packKey: newerPackKey, |
| 639 | packBytes: newerHead!.size, |
| 640 | objectCount: newerScan.objectCount, |
| 641 | idxBytes: newerResolve.idxBytes, |
| 642 | }); |
| 643 | |
| 644 | const backfillScan = await scanPack({ |
| 645 | env, |
| 646 | packKey: targetPackKey, |
| 647 | packSize: targetHead!.size, |
| 648 | limiter: makeLimiter(), |
| 649 | countSubrequest: () => {}, |
| 650 | log, |
| 651 | }); |
| 652 | const backfillResolve = await resolveDeltasAndWriteIdx({ |
| 653 | env, |
| 654 | packKey: targetPackKey, |
| 655 | packSize: targetHead!.size, |
| 656 | limiter: makeLimiter(), |
| 657 | countSubrequest: () => {}, |
| 658 | log, |
| 659 | scanResult: backfillScan, |
| 660 | activeCatalog: [newerRow, olderRow], |
| 661 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 662 | repoId, |
| 663 | writeIdx: false, |
| 664 | existingIdxView: targetResolve.idxView, |
| 665 | }); |
| 666 | |
| 667 | expect(backfillResolve.idxBytes).toBe(0); |
| 668 | expect(backfillScan.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(1); |
| 669 | expect(bytesToHex(backfillScan.table.oids.subarray(0, 20))).toBe(targetOid); |
| 670 | }); |
| 671 | |
| 672 | it("wakes deferred OFS_DELTA children after an external-base REF_DELTA resolves", async () => { |
| 673 | const baseBlobPayload = new TextEncoder().encode("base\n"); |
| 674 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 675 | const basePack = await buildPack([{ type: "blob", payload: baseBlobPayload }]); |
| 676 | |
| 677 | const basePackKey = "test/ref-delta-ofs-base.pack"; |
| 678 | await env.REPO_BUCKET.put(basePackKey, basePack); |
| 679 | const baseHead = await env.REPO_BUCKET.head(basePackKey); |
| 680 | |
| 681 | const baseScan = await scanPack({ |
| 682 | env, |
| 683 | packKey: basePackKey, |
| 684 | packSize: baseHead!.size, |
| 685 | limiter: makeLimiter(), |
| 686 | countSubrequest: () => {}, |
| 687 | log, |
| 688 | }); |
| 689 | const repoId = uniqueRepoId(); |
| 690 | const baseResolve = await resolveDeltasAndWriteIdx({ |
| 691 | env, |
| 692 | packKey: basePackKey, |
| 693 | packSize: baseHead!.size, |
| 694 | limiter: makeLimiter(), |
| 695 | countSubrequest: () => {}, |
| 696 | log, |
| 697 | scanResult: baseScan, |
| 698 | repoId, |
| 699 | }); |
| 700 | |
| 701 | const midSuffix = new TextEncoder().encode("mid\n"); |
| 702 | const finalSuffix = new TextEncoder().encode("final\n"); |
| 703 | |
| 704 | const midPayload = new Uint8Array(baseBlobPayload.length + midSuffix.length); |
| 705 | midPayload.set(baseBlobPayload, 0); |
| 706 | midPayload.set(midSuffix, baseBlobPayload.length); |
| 707 | |
| 708 | const finalPayload = new Uint8Array(midPayload.length + finalSuffix.length); |
| 709 | finalPayload.set(midPayload, 0); |
| 710 | finalPayload.set(finalSuffix, midPayload.length); |
| 711 | const finalOid = await computeOid("blob", finalPayload); |
| 712 | |
| 713 | const packBytes = await buildPack([ |
| 714 | { |
| 715 | type: "ref-delta", |
| 716 | baseOid: baseBlob.oid, |
| 717 | delta: buildAppendOnlyDelta(baseBlobPayload, midSuffix), |
| 718 | }, |
| 719 | { |
| 720 | type: "ofs-delta", |
| 721 | baseIndex: 0, |
| 722 | delta: buildAppendOnlyDelta(midPayload, finalSuffix), |
| 723 | }, |
| 724 | ]); |
| 725 | |
| 726 | const packKey = "test/resolve-ref-delta-ofs-dependent.pack"; |
| 727 | await env.REPO_BUCKET.put(packKey, packBytes); |
| 728 | const head = await env.REPO_BUCKET.head(packKey); |
| 729 | |
| 730 | const scanResult = await scanPack({ |
| 731 | env, |
| 732 | packKey, |
| 733 | packSize: head!.size, |
| 734 | limiter: makeLimiter(), |
| 735 | countSubrequest: () => {}, |
| 736 | log, |
| 737 | }); |
| 738 | |
| 739 | await resolveDeltasAndWriteIdx({ |
| 740 | env, |
| 741 | packKey, |
| 742 | packSize: head!.size, |
| 743 | limiter: makeLimiter(), |
| 744 | countSubrequest: () => {}, |
| 745 | log, |
| 746 | scanResult, |
| 747 | activeCatalog: [ |
| 748 | makeActiveCatalogRow({ |
| 749 | packKey: basePackKey, |
| 750 | packBytes: baseHead!.size, |
| 751 | objectCount: baseScan.objectCount, |
| 752 | idxBytes: baseResolve.idxBytes, |
| 753 | }), |
| 754 | ], |
| 755 | cacheCtx: createTestCacheContext("http://localhost/test"), |
| 756 | repoId, |
| 757 | }); |
| 758 | |
| 759 | expect(scanResult.table.resolved.reduce((sum, value) => sum + value, 0)).toBe(2); |
| 760 | expect(bytesToHex(scanResult.table.oids.subarray(20, 40))).toBe(finalOid); |
| 761 | }); |
| 762 | |
| 763 | it("uses the caller cache context for thin-pack external-base reads", async () => { |
| 764 | const baseBlobPayload = new TextEncoder().encode("external base content\n"); |
| 765 | const baseBlob = await encodeGitObject("blob", baseBlobPayload); |
| 766 | const basePackKey = "test/ref-delta-budget-base.pack"; |
| 767 | const basePack = await buildPack([{ type: "blob", payload: baseBlobPayload }]); |
| 768 | await env.REPO_BUCKET.put(basePackKey, basePack); |
| 769 | const baseHead = await env.REPO_BUCKET.head(basePackKey); |
| 770 | |
| 771 | const baseScan = await scanPack({ |
| 772 | env, |
| 773 | packKey: basePackKey, |
| 774 | packSize: baseHead!.size, |
| 775 | limiter: makeLimiter(), |
| 776 | countSubrequest: () => {}, |
| 777 | log, |
| 778 | }); |
| 779 | |
| 780 | const repoId = uniqueRepoId(); |
| 781 | const baseResolve = await resolveDeltasAndWriteIdx({ |
| 782 | env, |
| 783 | packKey: basePackKey, |
| 784 | packSize: baseHead!.size, |
| 785 | limiter: makeLimiter(), |
| 786 | countSubrequest: () => {}, |
| 787 | log, |
| 788 | scanResult: baseScan, |
| 789 | repoId, |
| 790 | }); |
| 791 | |
| 792 | const suffix = new TextEncoder().encode("with shared limiter\n"); |
| 793 | const thinPackKey = "test/ref-delta-budget-thin.pack"; |
| 794 | const thinPack = await buildPack([ |
| 795 | { |
| 796 | type: "ref-delta", |
| 797 | baseOid: baseBlob.oid, |
| 798 | delta: buildAppendOnlyDelta(baseBlobPayload, suffix), |
| 799 | }, |
| 800 | ]); |
| 801 | await env.REPO_BUCKET.put(thinPackKey, thinPack); |
| 802 | const thinHead = await env.REPO_BUCKET.head(thinPackKey); |
| 803 | |
| 804 | const labels: string[] = []; |
| 805 | const limiter = makeTracingLimiter(labels); |
| 806 | const counter = { count: 0 }; |
| 807 | const cacheCtx = createTestCacheContext("http://localhost/test", 10); |
| 808 | cacheCtx.memo = { |
| 809 | ...(cacheCtx.memo || {}), |
| 810 | limiter, |
| 811 | }; |
| 812 | |
| 813 | const thinScan = await scanPack({ |
| 814 | env, |
| 815 | packKey: thinPackKey, |
| 816 | packSize: thinHead!.size, |
| 817 | limiter, |
| 818 | countSubrequest: makeCountSubrequest(counter), |
| 819 | log, |
| 820 | }); |
| 821 | |
| 822 | await resolveDeltasAndWriteIdx({ |
| 823 | env, |
| 824 | packKey: thinPackKey, |
| 825 | packSize: thinHead!.size, |
| 826 | limiter, |
| 827 | countSubrequest: makeCountSubrequest(counter), |
| 828 | log, |
| 829 | scanResult: thinScan, |
| 830 | activeCatalog: [ |
| 831 | makeActiveCatalogRow({ |
| 832 | packKey: basePackKey, |
| 833 | packBytes: baseHead!.size, |
| 834 | objectCount: baseScan.objectCount, |
| 835 | idxBytes: baseResolve.idxBytes, |
| 836 | }), |
| 837 | ], |
| 838 | cacheCtx, |
| 839 | repoId, |
| 840 | }); |
| 841 | |
| 842 | expect(labels).toContain("r2:get-pack-idx"); |
| 843 | expect(labels).toContain("r2:get-range"); |
| 844 | expect(cacheCtx.memo?.subreqBudget).toBeLessThan(10); |
| 845 | }); |
| 846 | }); |