Blame · Line-by-line history
import-helper.ts
Each line is annotated with the commit that last touched it. Click any SHA to jump to that commit and see the surrounding change.
| 80bed05 | 1 | /** |
| 2 | * Small helpers for the GitHub import flow (/import). | |
| 3 | * | |
| 14c3cc8 | 4 | * Pure parsing/normalization helpers live at the top of the file. |
| 5 | * `importOneRepo` at the bottom wraps the clone + DB insert so that both | |
| 6 | * the single-repo and bulk importers can share one code path. | |
| 80bed05 | 7 | */ |
| 8 | ||
| 14c3cc8 | 9 | import { and, eq } from "drizzle-orm"; |
| 9bc1aed | 10 | import { mkdir, rm } from "fs/promises"; |
| 14c3cc8 | 11 | import { join } from "path"; |
| 12 | import { db } from "../db"; | |
| 13 | import { repositories } from "../db/schema"; | |
| 14 | import { config } from "../lib/config"; | |
| 15 | ||
| 9bc1aed | 16 | /** |
| 17 | * Count commits across ALL refs in a bare repo. Returns 0 for an empty repo | |
| 18 | * (no refs / unborn HEAD) or on any git error. | |
| 19 | * | |
| 20 | * This is the gate that makes an import trustworthy: `git clone --bare | |
| 21 | * --mirror` of an empty, unreachable, or private-without-a-valid-token | |
| 22 | * source EXITS 0 yet produces a repo with zero objects. Without this check | |
| 23 | * the import inserted a DB row and reported "success" for a phantom empty | |
| 24 | * repo — which is how a Vapron import "succeeded" while transferring nothing, | |
| 25 | * a silent failure that's actively dangerous once the user deletes the | |
| 26 | * GitHub source trusting the import worked. | |
| 27 | */ | |
| 28 | export async function countRepoCommits(destPath: string): Promise<number> { | |
| 29 | try { | |
| 30 | const proc = Bun.spawn( | |
| 31 | ["git", "-C", destPath, "rev-list", "--all", "--count"], | |
| 32 | { | |
| 33 | stdout: "pipe", | |
| 34 | stderr: "pipe", | |
| 35 | env: { ...process.env, GIT_TERMINAL_PROMPT: "0" }, | |
| 36 | } | |
| 37 | ); | |
| 38 | const out = await new Response(proc.stdout).text(); | |
| 39 | const code = await proc.exited; | |
| 40 | if (code !== 0) return 0; | |
| 41 | const n = parseInt(out.trim(), 10); | |
| 42 | return Number.isFinite(n) && n > 0 ? n : 0; | |
| 43 | } catch { | |
| 44 | return 0; | |
| 45 | } | |
| 46 | } | |
| 47 | ||
| 80bed05 | 48 | export interface ParsedGithubUrl { |
| 49 | owner: string; | |
| 50 | repo: string; | |
| 51 | } | |
| 52 | ||
| 53 | /** | |
| 54 | * Parse a GitHub URL into { owner, repo }. Accepts: | |
| 55 | * - https://github.com/foo/bar | |
| 56 | * - https://github.com/foo/bar.git | |
| 57 | * - http://github.com/foo/bar/ | |
| 58 | * - git@github.com:foo/bar.git | |
| 59 | * - github.com/foo/bar | |
| 60 | * - foo/bar | |
| 61 | * | |
| 62 | * Returns null if the URL cannot be parsed. | |
| 63 | */ | |
| 64 | export function parseGithubUrl(raw: string): ParsedGithubUrl | null { | |
| 65 | const input = (raw || "").trim(); | |
| 66 | if (!input) return null; | |
| 67 | ||
| 68 | // SSH form: git@github.com:owner/repo(.git)? | |
| 69 | const ssh = input.match(/^git@github\.com:([^/]+)\/([^/\s]+?)(?:\.git)?\/?$/i); | |
| 70 | if (ssh) return { owner: ssh[1], repo: stripDotGit(ssh[2]) }; | |
| 71 | ||
| 72 | // HTTP(S) / bare host form | |
| 73 | const http = input.match( | |
| 74 | /^(?:https?:\/\/)?(?:www\.)?github\.com\/([^/\s]+)\/([^/\s?#]+?)(?:\.git)?\/?(?:[?#].*)?$/i | |
| 75 | ); | |
| 76 | if (http) return { owner: http[1], repo: stripDotGit(http[2]) }; | |
| 77 | ||
| 78 | // owner/repo shorthand | |
| 79 | const short = input.match(/^([^/\s]+)\/([^/\s]+?)(?:\.git)?$/); | |
| 80 | if (short) return { owner: short[1], repo: stripDotGit(short[2]) }; | |
| 81 | ||
| 82 | return null; | |
| 83 | } | |
| 84 | ||
| 85 | function stripDotGit(name: string): string { | |
| 86 | return name.replace(/\.git$/i, ""); | |
| 87 | } | |
| 88 | ||
| 89 | /** | |
| 90 | * Repository names on gluecron follow GitHub's rough rules: letters, | |
| 91 | * digits, hyphens, underscores, dots. We normalize by replacing anything | |
| 92 | * else with a hyphen so an imported repo is always addressable. | |
| 93 | */ | |
| 94 | export function sanitizeRepoName(name: string): string { | |
| 2cd3bb4 | 95 | // Lowercase on create — repo slugs are canonically lowercase to avoid the |
| 96 | // "Vapron" vs "vapron" confusion. Imported names are normalized here. | |
| 97 | const cleaned = name | |
| 98 | .replace(/[^A-Za-z0-9._-]/g, "-") | |
| 99 | .replace(/^-+|-+$/g, "") | |
| 100 | .toLowerCase(); | |
| 80bed05 | 101 | return cleaned || "imported-repo"; |
| 102 | } | |
| 103 | ||
| 104 | /** | |
| 105 | * Build the clone URL that `git clone --bare --mirror` will use. When a | |
| 106 | * token is supplied we inject it so private repos are reachable. | |
| 107 | */ | |
| 108 | export function buildCloneUrl(cloneUrl: string, token: string | null): string { | |
| 109 | if (!token) return cloneUrl; | |
| 110 | return cloneUrl.replace("https://github.com/", `https://${token}@github.com/`); | |
| 111 | } | |
| 14c3cc8 | 112 | |
| 113 | /** | |
| 114 | * Strip any secret (token) from a string before it gets returned to the UI | |
| 115 | * or written to logs. Defense in depth: we already avoid putting tokens into | |
| 116 | * messages, but git/HTTP errors may echo the URL we passed in. | |
| 117 | */ | |
| 118 | export function scrubSecrets(input: string, token: string | null): string { | |
| 119 | if (!input) return input; | |
| 120 | let out = input; | |
| 121 | if (token) out = out.split(token).join("***"); | |
| 122 | // Also redact any `https://<creds>@github.com/...` form the URL may leak. | |
| 123 | out = out.replace( | |
| 124 | /https:\/\/[^@\s]+@github\.com/gi, | |
| 125 | "https://***@github.com" | |
| 126 | ); | |
| 127 | return out; | |
| 128 | } | |
| 129 | ||
| 130 | export interface ImportOneRepoInput { | |
| 131 | cloneUrl: string; | |
| 132 | targetName: string; | |
| 133 | ownerId: string; | |
| 134 | ownerUsername: string; | |
| 135 | token?: string | null; | |
| 136 | description?: string | null; | |
| 137 | isPrivate?: boolean; | |
| 138 | defaultBranch?: string; | |
| 139 | } | |
| 140 | ||
| 141 | export type ImportOneRepoStatus = "success" | "skipped-exists" | "failed"; | |
| 142 | ||
| 143 | export interface ImportOneRepoResult { | |
| 144 | status: ImportOneRepoStatus; | |
| 145 | name: string; | |
| 146 | notes: string; | |
| 147 | } | |
| 148 | ||
| 149 | /** | |
| 150 | * Clone one GitHub repo into this user's namespace and insert the DB row. | |
| 151 | * | |
| 152 | * Resilient: returns a result object instead of throwing, so bulk callers | |
| 153 | * can continue past a failure. Never includes the token in the returned | |
| 154 | * notes — all output is passed through `scrubSecrets`. | |
| 155 | */ | |
| 156 | export async function importOneRepo( | |
| 157 | input: ImportOneRepoInput | |
| 158 | ): Promise<ImportOneRepoResult> { | |
| 159 | const { | |
| 160 | cloneUrl, | |
| 161 | targetName, | |
| 162 | ownerId, | |
| 163 | ownerUsername, | |
| 164 | token = null, | |
| 165 | description = null, | |
| 166 | isPrivate = false, | |
| 167 | defaultBranch = "main", | |
| 168 | } = input; | |
| 169 | ||
| 170 | const safeName = sanitizeRepoName(targetName); | |
| 171 | ||
| 172 | try { | |
| 173 | // Uniqueness in the caller's namespace (owner+name). | |
| 174 | const [existing] = await db | |
| 175 | .select() | |
| 176 | .from(repositories) | |
| 177 | .where( | |
| 178 | and(eq(repositories.ownerId, ownerId), eq(repositories.name, safeName)) | |
| 179 | ) | |
| 180 | .limit(1); | |
| 181 | ||
| 182 | if (existing) { | |
| 9bc1aed | 183 | // A prior import can leave an EMPTY repo behind (see countRepoCommits). |
| 184 | // Don't cheerfully report "already exists" for a broken empty shell — | |
| 185 | // tell the user it's empty so they can delete + re-import to populate it. | |
| 186 | const existingCommits = await countRepoCommits(existing.diskPath); | |
| 187 | if (existingCommits === 0) { | |
| 188 | return { | |
| 189 | status: "failed", | |
| 190 | name: safeName, | |
| 191 | notes: | |
| 192 | "A repo by this name already exists but is EMPTY — a previous import left an empty shell. Delete it and re-import to populate it.", | |
| 193 | }; | |
| 194 | } | |
| 14c3cc8 | 195 | return { |
| 196 | status: "skipped-exists", | |
| 197 | name: safeName, | |
| 9bc1aed | 198 | notes: `Already exists in your namespace (${existingCommits} commits)`, |
| 14c3cc8 | 199 | }; |
| 200 | } | |
| 201 | ||
| 202 | const destPath = join(config.gitReposPath, ownerUsername, `${safeName}.git`); | |
| 203 | await mkdir(join(config.gitReposPath, ownerUsername), { recursive: true }); | |
| 204 | ||
| 205 | const authedCloneUrl = buildCloneUrl(cloneUrl, token); | |
| 206 | ||
| 207 | const proc = Bun.spawn( | |
| 208 | ["git", "clone", "--bare", "--mirror", authedCloneUrl, destPath], | |
| 209 | { | |
| 210 | stdout: "pipe", | |
| 211 | stderr: "pipe", | |
| 212 | env: { ...process.env, GIT_TERMINAL_PROMPT: "0" }, | |
| 213 | } | |
| 214 | ); | |
| 215 | const stderr = await new Response(proc.stderr).text(); | |
| 216 | const exitCode = await proc.exited; | |
| 217 | ||
| 218 | if (exitCode !== 0) { | |
| 219 | return { | |
| 220 | status: "failed", | |
| 221 | name: safeName, | |
| 222 | notes: `git clone failed: ${scrubSecrets(stderr, token).slice(0, 200)}`, | |
| 223 | }; | |
| 224 | } | |
| 225 | ||
| 9bc1aed | 226 | // Gate success on the clone having actually transferred history. `git |
| 227 | // clone --mirror` can exit 0 yet leave an empty repo; without this check | |
| 228 | // that got a DB row + "success" — a phantom empty repo. Clean up the | |
| 229 | // empty dir (so a retry isn't blocked by "already exists") and fail loud. | |
| 230 | const commitCount = await countRepoCommits(destPath); | |
| 231 | if (commitCount === 0) { | |
| 232 | await rm(destPath, { recursive: true, force: true }).catch(() => {}); | |
| 233 | return { | |
| 234 | status: "failed", | |
| 235 | name: safeName, | |
| 236 | notes: | |
| 237 | "Clone produced an EMPTY repository — nothing was imported. The source may be empty, or private and the token lacks access. No repo was created; fix access and retry.", | |
| 238 | }; | |
| 239 | } | |
| 240 | ||
| 14c3cc8 | 241 | await db.insert(repositories).values({ |
| 242 | name: safeName, | |
| 243 | ownerId, | |
| 244 | description, | |
| 245 | isPrivate, | |
| 246 | defaultBranch: defaultBranch || "main", | |
| 247 | diskPath: destPath, | |
| 248 | starCount: 0, | |
| 249 | }); | |
| 250 | ||
| 9bc1aed | 251 | return { |
| 252 | status: "success", | |
| 253 | name: safeName, | |
| 254 | notes: `Cloned + indexed (${commitCount} commits)`, | |
| 255 | }; | |
| 14c3cc8 | 256 | } catch (err) { |
| 257 | return { | |
| 258 | status: "failed", | |
| 259 | name: safeName, | |
| 260 | notes: scrubSecrets(String(err), token).slice(0, 200), | |
| 261 | }; | |
| 262 | } | |
| 263 | } |