Merge pull request #3636 from OpenNeuroOrg/download-bug-fixes

fix(cli): Checkout latest snapshot by default and allow version for download action

Author: nellh

cli/src/commands/download.ts

@@ -3,6 +3,8 @@ import type { CommandOptions } from "@cliffy/command"
 import { readConfig } from "../config.ts"
 import { logger } from "../logger.ts"
 import { getRepoAccess } from "./git-credential.ts"
+import { getLatestSnapshotVersion } from "../graphq.ts"
+import { version } from "node:os"
 
 export const download = new Command()
   .name("download")
@@ -11,10 +13,12 @@ export const download = new Command()
   .option(
     "-d, --draft",
     "Download a draft instead of the latest version snapshot.",
+    { conflicts: ["version"] },
   )
   .option(
-    "-v, --version",
+    "-v, --version <version:string>",
     "Download a specific version.",
+    { conflicts: ["draft"] },
   )
   .action(downloadAction)
 
@@ -44,14 +48,23 @@ export async function downloadAction(
 
   console.log("Downloading...")
 
-  // Clone main/master and git-annex branches
+  let version
+  if (options.version) {
+    version = options.version
+  } else if (!options.draft) {
+    version = await getLatestSnapshotVersion(datasetId)
+  }
+
+  // Clone the repo
   worker.postMessage({
     "command": "clone",
+    "version": version,
   })
 
   // Setup any git-annex remotes required for downloads
   worker.postMessage({
     "command": "remote-setup",
+    "version": version,
   })
 
   // Close after all tasks are queued

cli/src/graphq.ts

@@ -92,3 +92,26 @@ export async function prepareUpload(
 
 export async function finishUpload() {
 }
+
+export async function getLatestSnapshotVersion(datasetId: string) {
+  const query = `
+  query($datasetId: ID!) {
+    dataset(id: $datasetId) {
+      latestSnapshot {
+        id
+        tag
+      }
+    }
+  }
+  `
+  const res = await request(query, { datasetId })
+  const body = await res.json()
+  if (body.errors) {
+    throw new ResponseError(JSON.stringify(body.errors))
+  }
+  if (body.data) {
+    return body.data.dataset.latestSnapshot.tag
+  } else {
+    throw new QueryError("Invalid response")
+  }
+}

cli/src/worker/git.ts

@@ -1,4 +1,8 @@
-import type { GitWorkerEventAdd } from "./types/git-context.ts"
+import type {
+  GitWorkerEventAdd,
+  GitWorkerEventClone,
+  GitWorkerEventRemoteSetup,
+} from "./types/git-context.ts"
 import type { GitAnnexAttributes, GitAnnexBackend } from "../gitattributes.ts"
 import { matchGitAttributes, parseGitAttributes } from "../gitattributes.ts"
 import { dirname, join } from "@std/path"
@@ -28,11 +32,12 @@ async function done() {
 /**
  * Clone or fetch the draft
  */
-async function update() {
+async function update(event: GitWorkerEventClone) {
+  const version = event.data?.version
   try {
     await context.fs.promises.access(join(context.repoPath, ".git"))
     logger.info(
-      `Fetching ${context.datasetId} draft from "${context.repoEndpoint}"`,
+      `Fetching ${context.datasetId} from "${context.repoEndpoint}"`,
     )
     // Reset first to be sure that an interrupted run didn't create untracked files
     await resetWorktree(context, await getDefault(context))
@@ -49,19 +54,26 @@ async function update() {
         theirs: "origin/git-annex",
       },
     )
-    logger.info("Checkout latest HEAD.")
     // Make sure the default branch checkout is updated
     const defaultBranch = await getDefault(context)
+    const ref = version || defaultBranch
+    logger.info(`Checking out ${ref} branch.`)
     await git.checkout({
       ...context.config(),
-      ref: defaultBranch, // Checkout main
+      ref, // Checkout main
       noUpdateHead: false, // Make sure we update the local branch HEAD
     })
   } catch (_err) {
     logger.info(
-      `Cloning ${context.datasetId} draft from "${context.repoEndpoint}"`,
+      `Cloning ${context.datasetId} from "${context.repoEndpoint}"`,
     )
+    // Get the default branch
     await git.clone(context.config())
+    // Checkout a specific version if needed
+    if (version) {
+      logger.info(`Checking out ${version}.`)
+      await git.checkout({ ...context.config(), ref: version })
+    }
     await git.fetch({ ...context.config(), ref: "git-annex" })
     try {
       await git.branch({
@@ -232,14 +244,14 @@ async function createAnnexBranch() {
 /**
  * Generate a commit for remote.log updates if needed
  */
-async function remoteSetup() {
-  await commitAnnexBranch()
+async function remoteSetup(event: GitWorkerEventRemoteSetup) {
+  await commitAnnexBranch(event.data?.version)
 }
 
 /**
  * Generate one commit for all pending git-annex branch changes
  */
-async function commitAnnexBranch() {
+async function commitAnnexBranch(version?: string) {
   // Find the UUID of this repository if it exists already
   const expectedRemote = "OpenNeuro" // TODO - This could be more flexible?
   let uuid
@@ -347,7 +359,7 @@ async function commitAnnexBranch() {
       }
     }
   } finally {
-    const defaultBranch = await getDefault(context)
+    const defaultBranch = version || await getDefault(context)
     await git.checkout({
       ...context.config(),
       ref: defaultBranch,
@@ -556,7 +568,7 @@ self.onmessage = (event: GitWorkerEvent) => {
     )
     setupLogging(event.data.logLevel)
   } else if (event.data.command === "clone") {
-    workQueue.enqueue(update)
+    workQueue.enqueue(update, event)
   } else if (event.data.command === "clearWorktree") {
     workQueue.enqueue(clearWorktree)
   } else if (event.data.command === "add") {
@@ -566,7 +578,7 @@ self.onmessage = (event: GitWorkerEvent) => {
   } else if (event.data.command === "push") {
     workQueue.enqueue(push)
   } else if (event.data.command === "remote-setup") {
-    workQueue.enqueue(remoteSetup)
+    workQueue.enqueue(remoteSetup, event)
   } else if (event.data.command === "done") {
     workQueue.enqueue(done)
   }

cli/src/worker/types/git-context.ts

@@ -104,7 +104,26 @@ export interface GitWorkerEventAdd {
   }
 }
 
+/** Clone event to clone or fetch a version */
+export interface GitWorkerEventClone {
+  data: {
+    command: "clone"
+    // Version to clone
+    version?: string
+  }
+}
+
+export interface GitWorkerEventRemoteSetup {
+  data: {
+    command: "remoteSetup"
+    // Version to checkout after setup
+    version?: string
+  }
+}
+
 export type GitWorkerEvent =
   | GitWorkerEventSetup
   | GitWorkerEventGeneric
   | GitWorkerEventAdd
+  | GitWorkerEventClone
+  | GitWorkerEventRemoteSetup