| 1 | import { execFileSync } from "node:child_process"; |
| 2 | import { mkdtempSync, writeFileSync, readFileSync, mkdirSync } from "node:fs"; |
| 3 | import { tmpdir } from "node:os"; |
| 4 | import { join } from "node:path"; |
| 5 | import { describe, expect, it } from "vitest"; |
| 6 | import { |
| 7 | MemoryStore, |
| 8 | applyChanges, |
| 9 | deflate, |
| 10 | diffTrees, |
| 11 | fileDiff, |
| 12 | mergeBase, |
| 13 | mergeText, |
| 14 | mergeTrees, |
| 15 | objectsForFetch, |
| 16 | parseCommit, |
| 17 | parsePack, |
| 18 | readCommit, |
| 19 | serializeCommit, |
| 20 | utf8, |
| 21 | writePack, |
| 22 | fromUtf8, |
| 23 | listFiles, |
| 24 | hashObject, |
| 25 | inflate, |
| 26 | } from "../src/index.js"; |
| 27 | import { randomBytes } from "node:crypto"; |
| 28 | |
| 29 | const git = (cwd: string, ...args: string[]) => |
| 30 | execFileSync("git", args, { cwd, env: { ...process.env, GIT_AUTHOR_NAME: "T", GIT_AUTHOR_EMAIL: "t@x", GIT_COMMITTER_NAME: "T", GIT_COMMITTER_EMAIL: "t@x" } }); |
| 31 | |
| 32 | function makeRepo(): string { |
| 33 | const dir = mkdtempSync(join(tmpdir(), "gc-")); |
| 34 | git(dir, "init", "-q", "-b", "main"); |
| 35 | const big = Array.from({ length: 400 }, (_, i) => `line ${i} of a fairly long file to encourage deltas`).join("\n"); |
| 36 | writeFileSync(join(dir, "a.txt"), big); |
| 37 | mkdirSync(join(dir, "src/deep"), { recursive: true }); |
| 38 | writeFileSync(join(dir, "src/deep/x.js"), "console.log(1)\n"); |
| 39 | writeFileSync(join(dir, "bin.dat"), Buffer.from([0, 1, 2, 3, 0, 255])); |
| 40 | git(dir, "add", "-A"); |
| 41 | git(dir, "commit", "-q", "-m", "first"); |
| 42 | writeFileSync(join(dir, "a.txt"), big.replace("line 200", "LINE 200")); |
| 43 | git(dir, "commit", "-qam", "second"); |
| 44 | git(dir, "tag", "-a", "v1", "-m", "release"); |
| 45 | return dir; |
| 46 | } |
| 47 | |
| 48 | async function loadAll(dir: string): Promise<{ store: MemoryStore; expected: string[] }> { |
| 49 | const expected = git(dir, "rev-list", "--objects", "--all").toString().trim().split("\n").map((l) => l.split(" ")[0]!); |
| 50 | expected.push(git(dir, "rev-parse", "v1").toString().trim()); |
| 51 | const pack = execFileSync("git", ["pack-objects", "--stdout", "--delta-base-offset"], { cwd: dir, input: expected.join("\n") + "\n" }); |
| 52 | const objs = await parsePack(new Uint8Array(pack)); |
| 53 | const store = new MemoryStore(); |
| 54 | for (const o of objs) await store.write(o.type, o.content); |
| 55 | return { store, expected }; |
| 56 | } |
| 57 | |
| 58 | describe("packfiles", () => { |
| 59 | it("parses a real git pack with ofs deltas and verifies all oids", async () => { |
| 60 | const dir = makeRepo(); |
| 61 | const { store, expected } = await loadAll(dir); |
| 62 | for (const oid of expected) expect(store.objects.has(oid), oid).toBe(true); |
| 63 | }); |
| 64 | |
| 65 | it("keeps large blobs compressed, with the same oids, and still resolves deltas against them", async () => { |
| 66 | const dir = makeRepo(); |
| 67 | const expected = git(dir, "rev-list", "--objects", "--all").toString().trim().split("\n").map((l) => l.split(" ")[0]!); |
| 68 | const pack = new Uint8Array(execFileSync("git", ["pack-objects", "--stdout", "--delta-base-offset"], { cwd: dir, input: expected.join("\n") + "\n" })); |
| 69 | const plain = await parsePack(pack); |
| 70 | const lean = await parsePack(pack, undefined, { largeObjectBytes: 1000 }); |
| 71 | expect(lean.map((o) => o.oid).sort()).toEqual(plain.map((o) => o.oid).sort()); |
| 72 | const large = lean.filter((o) => o.deflated); |
| 73 | // a.txt in both commits: one is stored whole, the other is a delta against it. |
| 74 | expect(large.length).toBeGreaterThanOrEqual(1); |
| 75 | for (const o of large) { |
| 76 | const want = plain.find((p) => p.oid === o.oid)!; |
| 77 | expect(o.size).toBe(want.content.length); |
| 78 | expect(inflate(o.deflated!)).toEqual(want.content); |
| 79 | expect(o.content).toEqual(want.content); |
| 80 | } |
| 81 | for (const o of lean) expect(await hashObject(o.type, o.content)).toBe(o.oid); |
| 82 | }); |
| 83 | |
| 84 | it("parses a pack over 16 MB holding one large blob", async () => { |
| 85 | const dir = mkdtempSync(join(tmpdir(), "gc-")); |
| 86 | git(dir, "init", "-q", "-b", "main"); |
| 87 | const data = randomBytes(17 * 1024 * 1024); |
| 88 | writeFileSync(join(dir, "big.bin"), data); |
| 89 | git(dir, "add", "-A"); |
| 90 | git(dir, "commit", "-q", "-m", "big"); |
| 91 | const expected = git(dir, "rev-list", "--objects", "--all").toString().trim().split("\n").map((l) => l.split(" ")[0]!); |
| 92 | const pack = new Uint8Array(execFileSync("git", ["pack-objects", "--stdout"], { cwd: dir, input: expected.join("\n") + "\n", maxBuffer: 64 * 1024 * 1024 })); |
| 93 | expect(pack.length).toBeGreaterThan(16 * 1024 * 1024); |
| 94 | const objs = await parsePack(pack, undefined, { largeObjectBytes: 4 * 1024 * 1024 }); |
| 95 | const blob = objs.find((o) => o.type === "blob")!; |
| 96 | expect(blob.oid).toBe(git(dir, "rev-parse", "HEAD:big.bin").toString().trim()); |
| 97 | expect(blob.size).toBe(data.length); |
| 98 | expect(Buffer.from(inflate(blob.deflated!)).equals(data)).toBe(true); |
| 99 | // A damaged pack still fails its checksum. |
| 100 | pack[pack.length - 30]! ^= 1; |
| 101 | await expect(parsePack(pack, undefined, { largeObjectBytes: 4 * 1024 * 1024 })).rejects.toThrow(/checksum/); |
| 102 | }); |
| 103 | |
| 104 | it("writes packs that git accepts", async () => { |
| 105 | const dir = makeRepo(); |
| 106 | const { store } = await loadAll(dir); |
| 107 | const head = git(dir, "rev-parse", "HEAD").toString().trim(); |
| 108 | const oids = await objectsForFetch(store, [head], []); |
| 109 | const pack = await writePack( |
| 110 | await Promise.all(oids.map(async (oid) => { |
| 111 | const o = (await store.read(oid))!; |
| 112 | return { type: o.type, size: o.content.length, deflated: deflate(o.content) }; |
| 113 | })), |
| 114 | ); |
| 115 | const out = mkdtempSync(join(tmpdir(), "gc-out-")); |
| 116 | git(out, "init", "-q", "--bare"); |
| 117 | const p = join(out, "objects/pack/pack-x.pack"); |
| 118 | writeFileSync(p, pack); |
| 119 | git(out, "index-pack", p); |
| 120 | git(out, "update-ref", "refs/heads/main", head); |
| 121 | git(out, "fsck", "--strict", "main"); |
| 122 | }); |
| 123 | |
| 124 | it("sends only new objects on incremental fetch", async () => { |
| 125 | const dir = makeRepo(); |
| 126 | const { store } = await loadAll(dir); |
| 127 | const head = git(dir, "rev-parse", "HEAD").toString().trim(); |
| 128 | const first = git(dir, "rev-parse", "HEAD~1").toString().trim(); |
| 129 | const oids = await objectsForFetch(store, [head], [first]); |
| 130 | // new commit + new root tree + new a.txt blob |
| 131 | expect(oids.length).toBe(3); |
| 132 | }); |
| 133 | }); |
| 134 | |
| 135 | describe("trees, diffs and merges", () => { |
| 136 | it("builds commits from file changes and diffs them", async () => { |
| 137 | const store = new MemoryStore(); |
| 138 | const t1 = await applyChanges(store, null, [ |
| 139 | { path: "README.md", content: utf8("hello\n") }, |
| 140 | { path: "src/a.ts", content: utf8("a\n") }, |
| 141 | ]); |
| 142 | const t2 = await applyChanges(store, t1, [ |
| 143 | { path: "src/a.ts", content: utf8("a\nb\n") }, |
| 144 | { path: "README.md", delete: true }, |
| 145 | { path: "src/new/b.ts", content: utf8("x") }, |
| 146 | ]); |
| 147 | const changes = await diffTrees(store, t1, t2); |
| 148 | expect(changes.map((c) => `${c.status}:${c.path}`)).toEqual(["deleted:README.md", "modified:src/a.ts", "added:src/new/b.ts"]); |
| 149 | const d = await fileDiff(store, changes[1]!); |
| 150 | expect(d.additions).toBe(1); |
| 151 | expect((await listFiles(store, t2)).map((f) => f.path)).toEqual(["src/a.ts", "src/new/b.ts"]); |
| 152 | }); |
| 153 | |
| 154 | it("merges non-overlapping line edits and detects conflicts", () => { |
| 155 | const base = "a\nb\nc\nd\ne\n"; |
| 156 | expect(mergeText(base, "A\nb\nc\nd\ne\n", "a\nb\nc\nd\nE\n")).toBe("A\nb\nc\nd\nE\n"); |
| 157 | expect(mergeText(base, "a\nX\nc\nd\ne\n", "a\nY\nc\nd\ne\n")).toBeNull(); |
| 158 | expect(mergeText(base, "a\nb\nc\nd\ne\nf\n", "z\na\nb\nc\nd\ne\n")).toBe("z\na\nb\nc\nd\ne\nf\n"); |
| 159 | }); |
| 160 | |
| 161 | it("three-way merges trees and finds the merge base", async () => { |
| 162 | const store = new MemoryStore(); |
| 163 | const sig = { name: "T", email: "t@x", timestamp: 1, timezone: "+0000" }; |
| 164 | const commit = async (tree: string, parents: string[], ts: number) => |
| 165 | store.write("commit", serializeCommit({ tree, parents, author: { ...sig, timestamp: ts }, committer: { ...sig, timestamp: ts }, message: "m\n" })); |
| 166 | const tb = await applyChanges(store, null, [{ path: "f.txt", content: utf8("1\n2\n3\n4\n5\n") }, { path: "g.txt", content: utf8("g") }]); |
| 167 | const base = await commit(tb, [], 1); |
| 168 | const to = await applyChanges(store, tb, [{ path: "f.txt", content: utf8("one\n2\n3\n4\n5\n") }, { path: "new.txt", content: utf8("n") }]); |
| 169 | const ours = await commit(to, [base], 2); |
| 170 | const tt = await applyChanges(store, tb, [{ path: "f.txt", content: utf8("1\n2\n3\n4\nfive\n") }, { path: "g.txt", delete: true }]); |
| 171 | const theirs = await commit(tt, [base], 3); |
| 172 | expect(await mergeBase(store, ours, theirs)).toBe(base); |
| 173 | const res = await mergeTrees(store, tb, to, tt); |
| 174 | expect(res.conflicts).toEqual([]); |
| 175 | const files = await listFiles(store, res.tree!); |
| 176 | expect(files.map((f) => f.path)).toEqual(["f.txt", "new.txt"]); |
| 177 | expect(fromUtf8((await store.read(files[0]!.oid))!.content)).toBe("one\n2\n3\n4\nfive\n"); |
| 178 | expect(parseCommit((await store.read(ours))!.content).parents).toEqual([base]); |
| 179 | expect((await readCommit(store, theirs)).tree).toBe(tt); |
| 180 | }); |
| 181 | }); |