Skip to content

gavincabbage.com/chiv | Archive relational database tables to Amazon S3

License

Notifications You must be signed in to change notification settings

gavincabbage/chiv

Repository files navigation

Title


Archive relational data to Amazon S3.


This project provides the chiv package and a simple CLI wrapper of the same name. It requires Go 1.13.

Use chiv to download relational data from a database and format it before uploading to Amazon S3. Built-in formats include CSV, YAML and JSON. Custom formats are also supported through the Formatter interface.

Getting Started

import "gavincabbage.com/chiv"

Provide a database and upload manager to upload a table to an S3 bucket in the default CSV format.

db, _ := sql.Open(config.driver, config.url)

client := s3.New(session.NewSessionWithOptions(session.Options{}))
uploader := s3manager.NewUploaderWithClient(client)

chiv.Archive(db, uploader, "table", "bucket")

Use options to configure the archival format, upload key, null placeholder, etc.

chiv.Archive(db, uploader, "table", "bucket"
    chiv.WithFormat(chiv.JSON),
    chiv.WithKey("2019/september/monthly_archive.json"),
    chiv.WithNull("empty"),
)

For multiple uploads using the same database and S3 clients, construct an Archiver. Options provided during construction of an Archiver can be overridden in individual archival calls.

a := chiv.NewArchiver(db, uploader, chiv.WithFormat(chiv.YAML))
a.Archive("first_table", "bucket")
a.Archive("second_table", "bucket")
a.Archive("second_table", "bucket", chiv.WithFormat(chiv.JSON), chiv.WithKey("second_table.json"))

Custom queries can be archived using the ArchiveRows family of functions.

rows, _ := db.Exec("SELECT * FROM table and JOIN all the things...")

chiv.ArchiveRows(rows, uploader, "bucket")

Context-aware versions are also provided, e.g. ArchiveWithContext, ArchiveRowsWithContext, etc.

See the unit and integration tests for additional examples.

Custom Formats

Custom formats can be used by implementing the FormatterFunc and Formatter interfaces. The optional Extensioner interface can be implemented to allow a Formatter to provide a default extension.

See the three built-in formats for examples.

CLI

A simple CLI wrapping the package is also included.

NAME:
   chiv - Archive relational data to Amazon S3

USAGE:
   chiv [flags...]

VERSION:
   vX.Y.Z

GLOBAL OPTIONS:
   --database value, -d value   database connection string [$DATABASE_URL]
   --table value, -t value      database table to archive
   --bucket value, -b value     upload S3 bucket name
   --driver value, -r value     database driver type: postgres or mysql (default: "postgres")
   --columns value, -c value    database columns to archive, comma-separated
   --format value, -f value     upload format: csv, yaml or json (default: "csv")
   --key value, -k value        upload key
   --extension value, -e value  upload extension
   --null value, -n value       upload null value
   --help, -h                   show usage details
   --version, -v                print the version

Design

This package ties together three components of what is essentially an ETL operation:

  • Extract database rows with a Database or Rows
    • Database may be a *sql.DB, *sql.Conn, *sql.Tx, etc.
  • Transform data into an upload format with a Formatter
  • Load the formatted data into S3 with a Uploader
    • Uploader is typically an *s3manager.Uploader

An io.Pipe is used to guarantee a cap on total memory usage by chiv itself. A benchmark is provided to profile the package's memory usage. Total memory usage of the archival process can be further controlled by configuring the S3 upload manager.

Architecture

Testing & Development

Tests are segregated by build tags.

Run unit tests:

go test -tags=unit ./...

Use docker-compose to run the full suite of unit and integration tests:

docker-compose up --exit-code-from test --build

Run benchmarks:

go test -v -p 1 -tags=benchmark,unit -run=Benchmark -benchmem -bench=.

This project uses GitHub Actions for CI. Linting, unit tests, integration tests and benchmarks are run for each push or pull request.

Tagged releases are built automatically by GoReleaser.

Contributing

Contributions in the form of comments, issues or pull requests are very welcome.

If you use this package I would love to hear from you!

About

gavincabbage.com/chiv | Archive relational database tables to Amazon S3

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published