Copy and backup your photos and videos.
Copy between Google Photos, Flickr, AWS S3, and local directories.
Requires Go 1.21+
# macOS / Linux
./setup.sh
# Windows
setup.batEach command will tell you what you need to do. Credentials are saved to ~/.config/photo-copy/.
./photo-copy config flickr # Flickr API key + OAuth
./photo-copy config google # Google OAuth credentials for upload
./photo-copy config s3 # AWS credentials for S3# Download all photos from Flickr
./photo-copy flickr download ../flickr-photos
# Upload local photos to Flickr
./photo-copy flickr upload ../photosThe Google Photos API only allows access to photos the app itself uploaded, so downloading your full library requires Google Takeout (a manual zip export from Google).
# Upload local photos to Google Photos (requires 'config google' setup)
./photo-copy google upload ../photos
# Download: export your library via Google Takeout, then extract the zips
./photo-copy google import-takeout ../takeout-zips ../google-photos# Upload local photos to S3
./photo-copy s3 upload ../photos --bucket my-bucket --prefix photos/
# Download photos from S3
./photo-copy s3 download ../photos --bucket my-bucket --prefix photos/All transfers are resumable — if a download or upload is interrupted, re-running the same command picks up where it left off:
- Flickr downloads — A
transfer.logfile in the output directory tracks each successfully downloaded file. Already-downloaded files are skipped on restart. - Google Photos uploads — An upload log file tracks completed uploads the same way.
- S3 uploads/downloads — Handled by rclone, which compares source and destination and only transfers changed or missing files.
Files that fail during transfer are not marked as completed in the log, so re-running the same command will automatically retry them while skipping files that already succeeded.
- Flickr — Requests are throttled to 1/second (staying under Flickr's 3,600 requests/hour API limit). HTTP 429 and 5xx errors are retried up to 5 times with exponential backoff (2s, 4s, 8s, 16s, 32s), honoring the
Retry-Afterheader when present. This applies to both API calls and photo downloads. - Google Photos — Subject to a 10,000 uploads/day limit, enforced in code.
Every transfer automatically prints a summary and writes a detailed report file when it finishes. The summary includes file counts (succeeded, skipped, failed), total size transferred, and elapsed time. Any errors are re-listed so they're easy to spot.
Post-transfer validation checks for potential issues:
- Count verification — Compares the expected file count (from the service API) against the actual number of files succeeded, skipped, and failed. A mismatch means some files were unaccounted for.
- Zero-size file detection — Scans the output directory after downloads for empty files that may indicate incomplete transfers.
- Transfer log consistency (Flickr) — Verifies that every entry in
transfer.loghas a corresponding file on disk, catching orphaned log entries from interrupted downloads.
A report file (photo-copy-report-{service}-{operation}-{timestamp}.txt) is written to the transfer directory with the full breakdown of counts, errors, and validation warnings.
Add --debug to any command for verbose logging:
./photo-copy flickr download ../photos --debugJPEG, PNG, TIFF, GIF, HEIC, WebP, MP4, MOV, AVI, MKV
See CLAUDE.md for some details on the project.
Install golangci-lint:
brew install golangci-lintRun linting and unit tests:
golangci-lint run ./...
go test ./...Integration tests exercise CLI commands end-to-end against mock HTTP servers
for Flickr and Google Photos. They use a build tag and don't run with
go test ./...:
go test ./internal/cli/ -tags integrationS3 integration testing is not included — S3 operations delegate to a rclone subprocess, and rclone's own test coverage handles that layer. S3 unit tests cover command arg building, config generation, and binary resolution.
To update the bundled rclone binaries:
./rclone-bin/update-rclone.sh v1.68.2