ExoHub Logo
ExoHub

Exporting Data to Google Drive

Table of Contents

This tutorial shows how to export data from an ExoHub repository to a Google Drive folder so that collaborators — analysts, reviewers, or stakeholders who use Google Workspace — can browse, preview, and download files without installing exo or touching the command line.

Use case: you’ve processed a dataset and want to hand off the results to non-technical collaborators. Instead of packing and emailing files, you push them to a Drive folder they already have access to, keeping everything reproducible and versioned on your side.

See the https://genentech.github.io/exohub/docs/user-guide/drive-remote/ reference page for all configuration options and authentication details.

Prerequisites

1. Add a Drive Remote

Open .exohub/remotes in your repository. Add a new entry for the Drive remote alongside any existing remotes:

remotes:
  - name: s3-archive
    type: annex
    s3url: s3://exohub-sandbox-uat/tutorials/YOUR_UNIXID/mnist/_annex

  - name: s3-export
    type: export
    s3url: s3://exohub-sandbox-uat/tutorials/YOUR_UNIXID/mnist/_export

  - name: my-drive                              # any name you like
    type: drive
    drive_path: /My Drive/datasets/mnist        # folder path inside Drive
    tracking_branch: main
    include:
      - data/processed/**                       # export only processed results

Key fields:

FieldWhat it does
type: driveTells exo to use the Google Drive backend
drive_pathThe Drive folder that will receive the files (created automatically)
tracking_branchThe branch whose content is tracked for preferred-content rules
includeGlob patterns — only matching files are exported; omit to export everything

include patterns are enforced at the remote level, so both plain files (like .exohub/ metadata) and annexed files respect them. See https://genentech.github.io/exohub/docs/user-guide/drive-remote/ for exclude patterns and other options.

2. Initialize the Remote

After editing .exohub/remotes, register the new remote with git-annex:

exo init

Expected output:

✅ Initialized remote 's3-archive'
✅ Initialized remote 's3-export'
⚠️  Drive remote 'my-drive' needs authorization. Run: exo auth my-drive

The Drive remote is registered but not yet authorized. Commit the updated config:

git add .exohub/
git commit -m "feat: add Google Drive export remote"

3. Authorize Google Drive (one-time)

exo uses Google’s device flow, so you can authorize from any device — even a remote server without a browser:

exo auth my-drive

To authorize exo to access Google Drive:
  1. Go to: https://www.google.com/device
  2. Enter code: WXYZ-1234

Waiting for authorization... ✅

✅ Authorized. Token saved to ~/.config/exo/drive-tokens/my-drive.json
   You can now run: exo sync --with my-drive

Open the URL on any device (phone, laptop, browser), enter the code shown, and grant access. The CLI polls until you confirm.

Tokens are stored locally in ~/.config/exo/drive-tokens/ and are never committed to the repository. Subsequent syncs refresh the token automatically.

4. Export to Drive

Now sync your data to Drive:

exo sync --with my-drive

exo uploads every annexed file that matches the include patterns and mirrors the working-tree layout as a Drive folder hierarchy:

data/processed/train.parquet  →  /My Drive/datasets/mnist/data/processed/train.parquet
data/processed/test.parquet   →  /My Drive/datasets/mnist/data/processed/test.parquet

You can also target the Drive remote alone, without touching your S3 remotes:

exo sync --to my-drive

Export a specific path

exo sync --to my-drive --path data/processed/

Export a tagged version

exo sync --to my-drive --ref v1.0.0

5. What It Looks Like in Drive

Once the sync completes, collaborators open their Google Drive and navigate to:

My Drive / datasets / mnist / data / processed /

They see the real file names — train.parquet, test.parquet — as ordinary Drive items. They can share the folder, preview supported formats, or download directly. No ExoHub account or CLI is needed on their side.

6. Exporting Only Specific Files

Suppose your repository has multiple output directories and you only want to share one set of files with a particular audience. Update the include list:

  - name: stakeholder-drive
    type: drive
    drive_path: /My Drive/project-x/results
    tracking_branch: main
    include:
      - reports/**/*.pdf
      - summaries/*.csv

After editing, run exo init again and then sync:

exo init
exo sync --to stakeholder-drive

Only files matching those patterns are pushed to Drive — raw data files, scripts, and metadata stay out of the shared folder.

What’s Next