# <img className="page-icon" src="/img/icons/s3.svg" alt="" /> Amazon S3

Go from files in a bucket to a tested data contract in about five minutes. Works with S3 and any S3-compatible endpoint (e.g. MinIO), for data in CSV, JSON, Delta, or Parquet format.

For tables in an Amazon S3 table bucket, use the dedicated [Amazon S3 Tables guide](./s3-tables.md) with `type: iceberg` instead.

## 1. Install

```bash
uv tool install --python python3.11 --upgrade 'datacontract-cli[s3]'
```

See [Installation](../installation.md) for pip, pipx, and Docker.

## 2. Authenticate

The easiest way is to sign in to AWS once — the CLI picks the session up, so no key is stored anywhere:

```bash
aws sso login   # or any other way of getting an AWS session
```

Prefer static keys? Set them directly and they are used instead:

```bash
# .env
DATACONTRACT_S3_REGION=eu-central-1
DATACONTRACT_S3_ACCESS_KEY_ID=AKIAXV5Q5QABCDEFGH
DATACONTRACT_S3_SECRET_ACCESS_KEY=93S7LRrJcqLaaaa/XXXXXXXXXXXXX
```

Public buckets need neither: with nothing configured the objects are read unsigned.

## 3. Create a contract from your files

Import the schema straight from the bucket. This also generates a ready-to-test `servers` block:

```bash
datacontract import s3 \
  --source s3://my-bucket/orders/*.json \
  --output datacontract.yaml
```

The format is taken from the file suffix; pass `--format` for Delta tables, which have none. Add `--endpoint-url` for an S3-compatible store such as MinIO.

## 4. Test the actual data

```bash
datacontract test datacontract.yaml
```

```
Testing datacontract.yaml
Server: production (type=s3, format=json, location=s3://my-bucket/orders/*.json)
╭────────┬─────────────────────────────────────────────────┬─────────────────┬─────────╮
│ Result │ Check                                           │ Field           │ Details │
├────────┼─────────────────────────────────────────────────┼─────────────────┼─────────┤
│ passed │ Check that field 'order_id' is present          │ orders.order_id │         │
│ passed │ Check that field order_id has no missing values │ orders.order_id │         │
│  ...   │                                                 │                 │         │
╰────────┴─────────────────────────────────────────────────┴─────────────────┴─────────╯
🟢 data contract is valid. Run 17 checks. Took 3.6 seconds.
```

## 5. Let it catch a violation

The contract becomes valuable when it detects drift. Tighten an expectation — for example, mark a field as `required: true` that occasionally arrives empty, or add a quality rule:

```yaml
schema:
  - name: orders
    # ...
    quality:
      - type: sql
        description: No order has a negative total
        query: SELECT COUNT(*) FROM orders WHERE order_total < 0
        mustBe: 0
```

Run `datacontract test datacontract.yaml` again: every violation is listed as an error, and the command exits with code `1` — ready for [CI/CD and scheduled runs](../scheduling/index.md) so you catch drift before your consumers do.

## Reference

All authentication options and the data type handling per file format: **[S3 Reference](../reference/s3.md)**.

CSV files are read as UTF-8 unless the server declares an `encoding` (ODCS v3.2.0), such as `ISO-8859-1` or `UTF-16`. JSON files are always read as UTF-8.

## Troubleshooting

- **`403 Forbidden` / `Access Denied`** — the key pair lacks `s3:GetObject`/`s3:ListBucket` on the location, or `DATACONTRACT_S3_REGION` doesn't match the bucket's region.
- **`No files found that match the pattern`** — check the `location` glob; it matches object keys, not directories.
- **MinIO / S3-compatible storage fails to connect** — set `endpointUrl` in the `servers` block; the CLI then switches to path-style addressing.