import { MAX_INLINE_SIZE } from "#lib/api/blob.ts"; import { blob, knotMirrorTarget, type GitServiceConfig } from "#lib/api/gitclient.ts"; import { diffFileKind, diffFileStat, diffSideExists, getDiff, getInterdiff, listLog, type MirrorFileDiff, type MirrorFileKind } from "#lib/api/knotmirror.ts"; import type { FileContents } from "@pierre/diffs"; import type * as ShTangledGitDefs from "#lib/api/lexicons/types/sh/tangled/git/defs.ts"; import type { Did } from "@atcute/lexicons/syntax"; import type { BobbinContext } from "#lib/api/client.ts"; import type { GitTarget } from "#lib/api/gitclient.ts"; import type { DiffStat } from "#lib/api/diff.ts"; import type { Revspec } from "#lib/api/pullRoute.ts"; import type { CommitSummary } from "#lib/api/repo.ts"; type DiffSrc = ShTangledGitDefs.DiffSrc; // pierre renders roughly one row per line pair, plus a little chrome per hunk. only // used to reserve height under a pending card, so an estimate is enough const HUNK_CHROME_LINES = 2; // what the page asks for: one version's diff, or the interdiff between two of // them. `repo` is the source repo -- knotmirror unions both dids' object dirs // into one scratch repo, so a fork's head and its base share one did here export interface DiffSpec { kind: "diff"; repo: Did; version: Revspec; } export interface InterdiffSpec { kind: "interdiff"; repo: Did; oldVersion: Revspec; version: Revspec; // narrows the interdiff to the one commit carrying this change id changeId?: string; } export type RepoDiffSpec = DiffSpec | InterdiffSpec; export interface RepoDiffDeps { ctx: BobbinContext; // where the blob text comes from; the diff payload carries line numbers only git: GitTarget; } export interface RepoDiffFile { // stable across sides, used for anchors key: string; name: string; oldName?: string; kind: MirrorFileKind; stat: { insertions: number; deletions: number }; // rough rendered height, so a card can reserve space before its text lands lines: number; // Explains unrenderable files (binary, submodule, oversized); suppresses sides fetch. note?: string; sides?: { lhs: DiffSrc; rhs: DiffSrc }; } // where the two sides' text comes from. an interdiff's lhs is a tree knotmirror // rebased in memory for that one request, reachable by no ref, so its text can // only arrive inlined on the payload export type ContentsSource = { inlineOld?: false; oldRef: string; newRef: string } | { inlineOld: true; newRef: string }; export interface RepoDiff { files: RepoDiffFile[]; stat: DiffStat; contents: ContentsSource; } // one file's two sides, ready for pierre to diff export interface FilePair { oldFile: FileContents; newFile: FileContents; } export type FileContentsResult = FilePair | { note: string }; export const DIFF_TOO_LARGE_MESSAGE = "diff payload exceeds the ssr byte cap"; export class DiffTooLargeError extends Error { constructor() { super(DIFF_TOO_LARGE_MESSAGE); this.name = "DiffTooLargeError"; } } export const isDiffTooLarge = (cause: unknown): boolean => cause instanceof DiffTooLargeError || (cause instanceof Error && cause.message === DIFF_TOO_LARGE_MESSAGE); // sveltekit base64s a read of `body` but not `text()` on a clone export const cappedFetch = (fetch: typeof globalThis.fetch, maxBytes: number): typeof globalThis.fetch => async (input, init) => { const abort = new AbortController(); const signal = init?.signal ? AbortSignal.any([init.signal, abort.signal]) : abort.signal; const response = await fetch(input, { ...init, signal }); if (!response.ok) return response; const reader = response.clone().body?.getReader(); if (!reader) return response; let total = 0; for (;;) { const { done, value } = await reader.read(); if (done) return response; total += value.byteLength; if (total > maxBytes) { abort.abort(); throw new DiffTooLargeError(); } } }; export const loadDiff = (deps: RepoDiffDeps, spec: RepoDiffSpec): Promise => { if (spec.kind === "diff") return plainDiff(deps, spec.repo, spec.version); if (!spec.changeId) return interdiff(deps, spec.repo, spec.oldVersion, spec.version); return oneChange(deps, spec); }; const plainDiff = async (deps: RepoDiffDeps, repo: Did, version: Revspec): Promise => { const { diffs } = await getDiff(deps.ctx, { baseRepo: repo, baseCommit: version.base, headRepo: (version.repo ?? repo) as Did, headCommit: version.head }); // the left side is read at `base`, while knotmirror diffs from // merge_base(base, head). identical when base is the merge base, which is // what a pull version stores return buildDiff(diffs, { oldRef: version.base, newRef: version.head }); }; // an interdiff's right side is the new version's head tree, so it reads like any // other diff -- its left side comes inlined, see `ContentsSource` const interdiff = async ( deps: RepoDiffDeps, repo: Did, oldVersion: Revspec, version: Revspec ): Promise => { const { diffs } = await getInterdiff(deps.ctx, { baseRepo: (oldVersion.repo ?? repo) as Did, baseCommit1: oldVersion.base, baseCommit2: oldVersion.head, headRepo: (version.repo ?? repo) as Did, headCommit1: version.base, headCommit2: version.head }); return buildDiff(diffs, { inlineOld: true, newRef: version.head }); }; // counts for the og card, which leaves them out rather than wait on a slow or oversized diff export const loadDiffStat = ( config: GitServiceConfig & { ssrDiffMaxBytes: number }, repo: Did, version: Revspec, fetch: typeof globalThis.fetch, timeoutMs: number ): Promise => { const signal = AbortSignal.timeout(timeoutMs); const git = knotMirrorTarget(config, { repoDid: repo }, (input, init) => fetch(input, { ...init, signal }) ); const ctx = config.ssrDiffMaxBytes > 0 ? { ...git.ctx, fetch: cappedFetch(git.ctx.fetch, config.ssrDiffMaxBytes) } : git.ctx; return plainDiff({ ctx, git }, repo, version).then( ({ stat }) => stat, () => undefined ); }; // a change id names the same logical commit in both versions, so narrowing to it // means interdiffing that one commit's span on each side const oneChange = async (deps: RepoDiffDeps, spec: InterdiffSpec): Promise => { const [oldCommits, newCommits] = await Promise.all([ listLog(deps.ctx, spec.repo, spec.oldVersion), listLog(deps.ctx, spec.repo, spec.version) ]); const target = resolveChange(spec.changeId ?? "", oldCommits, newCommits, spec); switch (target.kind) { case "missing": throw new Error("Can't find a commit with that change-id in this version."); case "diff": return plainDiff(deps, spec.repo, target.version); case "interdiff": return interdiff(deps, spec.repo, target.oldVersion, target.version); } }; export type ChangeTarget = // the change is missing in the newer version, so there is nothing to interdiff | { kind: "missing" } // the change is new in the newer version, so we just diff since its parent | { kind: "diff"; version: Revspec } | { kind: "interdiff"; oldVersion: Revspec; version: Revspec }; // a root commit has no parent to diff against, so it falls back to the base of // the version it belongs to const single = (commit: CommitSummary, version: Revspec): Revspec => ({ base: commit.parent ?? version.base, head: commit.hash, repo: version.repo }); export const resolveChange = ( changeId: string, oldCommits: readonly CommitSummary[], newCommits: readonly CommitSummary[], bases: { oldVersion: Revspec; version: Revspec } ): ChangeTarget => { const find = (commits: readonly CommitSummary[]) => commits.find((commit) => commit.changeId === changeId); const to = find(newCommits); if (!to) return { kind: "missing" }; const from = find(oldCommits); if (!from) return { kind: "diff", version: single(to, bases.version) }; return { kind: "interdiff", oldVersion: single(from, bases.oldVersion), version: single(to, bases.version) }; }; // the diff endpoint answers metadata only -- `diffSrc` is path/oid/size/flags and // a hunk is aligned line numbers -- so the text comes from the blob endpoint and // pierre diffs the two sides itself const buildDiff = (diffs: MirrorFileDiff[], contents: ContentsSource): RepoDiff => { const stat: DiffStat = { insertions: 0, deletions: 0, files_changed: diffs.length }; const files = diffs.map((file): RepoDiffFile => { const fileStat = diffFileStat(file); stat.insertions += fileStat.insertions; stat.deletions += fileStat.deletions; const kind = diffFileKind(file); const name = file.rhsSrc.path || file.lhsSrc.path; const base: RepoDiffFile = { key: `${file.lhsSrc.path}→${file.rhsSrc.path}`, name, oldName: kind === "renamed" ? file.lhsSrc.path : undefined, kind, stat: fileStat, lines: renderedLines(file) }; if (file.lhsSrc.isSubmodule || file.rhsSrc.isSubmodule) { return { ...base, note: "Submodule pointer, not shown." }; } if (file.lhsSrc.isBinary || file.rhsSrc.isBinary) { return { ...base, note: "This is a binary file and will not be displayed." }; } // the payload carries each side's size, so a file the blob endpoint would // refuse never costs a request if (tooLarge(file.lhsSrc) || tooLarge(file.rhsSrc)) { return { ...base, note: "This file is too large to display." }; } return { ...base, sides: { lhs: file.lhsSrc, rhs: file.rhsSrc } }; }); return { files, stat, contents }; }; const tooLarge = (src: DiffSrc): boolean => src.size > MAX_INLINE_SIZE; const renderedLines = (file: MirrorFileDiff): number => file.hunks.reduce((total, hunk) => total + hunk.lines.length + HUNK_CHROME_LINES, 0); // 429 burst limits, 5xx, and network drops are retryable; 404 is permanent. const isTransient = (cause: unknown): boolean => { const status = (cause as { status?: unknown })?.status; if (typeof status === "number") return status === 429 || status >= 500; // a network failure carries no status at all return cause instanceof TypeError; }; const sideContents = async ( git: GitTarget, ref: string | undefined, src: DiffSrc ): Promise => { // a side that doesn't exist at this ref is an empty side, not a failed fetch if (!diffSideExists(src)) return ""; if (src.content !== undefined) return src.content; // the rebased tree an interdiff diffs against is reachable by no ref, so a // missing inline side there can never be recovered if (ref === undefined) return null; // a rate-limited or briefly failing mirror is worth retrying; a blob that is // really not there is not, so only the transient shape escapes const output = await blob(git, { ref, path: src.path }, MAX_INLINE_SIZE).catch( (cause: unknown) => { if (isTransient(cause)) throw cause; return null; } ); if (!output || output.encoding !== "utf-8") return null; return output.content ?? ""; }; export const fetchSides = async ( deps: RepoDiffDeps, contents: ContentsSource, file: RepoDiffFile ): Promise => { const sides = file.sides; if (!sides) return { note: file.note ?? "Contents could not be loaded." }; const [oldContents, newContents] = await Promise.all([ sideContents(deps.git, contents.inlineOld ? undefined : contents.oldRef, sides.lhs), sideContents(deps.git, contents.newRef, sides.rhs) ]); if (oldContents === null || newContents === null) { return { note: "Contents could not be loaded." }; } return { // pierre infers the language from the name, same as the blob page oldFile: { name: sides.lhs.path || file.name, contents: oldContents }, newFile: { name: file.name, contents: newContents } }; };