STAC FastAPI Elasticsearch (sfes)

Elasticsearch backend for stac-fastapi

Join our Gitter page

Check out the public Postman documentation Postman

Check out the examples folder for deployment options, ex. running sfes from pip in docker

For changes, see the Changelog

Development Environment Setup

To install the classes in your local Python env, run:

pip install -e 'stac_fastapi/elasticsearch[dev]'

Pre-commit

Install pre-commit.

Prior to commit, run:

pre-commit run --all-files

Building

docker-compose build

Running API on localhost:8080

docker-compose up

By default, docker-compose uses Elasticsearch 8.x. However, most recent 7.x versions should also work. If you wish to use a different version, put the following in a file named .env in the same directory you run docker-compose from:

ELASTICSEARCH_VERSION=7.17.1

To create a new Collection:

curl -X "POST" "http://localhost:8080/collections" \
     -H 'Content-Type: application/json; charset=utf-8' \
     -d $'{
  "id": "my_collection"
}'

Note: this "Collections Transaction" behavior is not part of the STAC API, but may be soon.

Collection pagination

The collections route handles optional limit and token parameters. The links field that is returned from the /collections route contains a next link with the token that can be used to get the next page of results.

curl -X "GET" "http://localhost:8080/collections?limit=1&token=example_token"

Testing

make test

Ingest sample data

make ingest

Elasticsearch Mappings

Mappings apply to search index, not source.

Managing Elasticsearch Indices

This section covers how to create a snapshot repository and then create and restore snapshots with this.

Create a snapshot repository. This puts the files in the elasticsearch/snapshots in this git repo clone, as the elasticsearch.yml and docker-compose files create a mapping from that directory to /usr/share/elasticsearch/snapshots within the Elasticsearch container and grant permissions on using it.

curl -X "PUT" "http://localhost:9200/_snapshot/my_fs_backup" \
     -H 'Content-Type: application/json; charset=utf-8' \
     -d $'{
            "type": "fs",
            "settings": {
                "location": "/usr/share/elasticsearch/snapshots/my_fs_backup"
            }
}'

The next step is to create a snapshot of one or more indices into this snapshot repository. This command creates a snapshot named my_snapshot_2 and waits for the action to be completed before returning. This can also be done asynchronously, and queried for status. The indices parameter determines which indices are snapshotted, and can include wildcards.

curl -X "PUT" "http://localhost:9200/_snapshot/my_fs_backup/my_snapshot_2?wait_for_completion=true" \
     -H 'Content-Type: application/json; charset=utf-8' \
     -d $'{
  "metadata": {
    "taken_because": "dump of all items",
    "taken_by": "pvarner"
  },
  "include_global_state": false,
  "ignore_unavailable": false,
  "indices": "items_my-collection"
}'

To see the status of this snapshot:

curl http://localhost:9200/_snapshot/my_fs_backup/my_snapshot_2

To see all the snapshots:

curl http://localhost:9200/_snapshot/my_fs_backup/_all

To restore a snapshot, run something similar to the following. This specific command will restore any indices that match items_* and rename them so that the new index name will be suffixed with -copy.

curl -X "POST" "http://localhost:9200/_snapshot/my_fs_backup/my_snapshot_2/_restore?wait_for_completion=true" \
     -H 'Content-Type: application/json; charset=utf-8' \
     -d $'{
  "include_aliases": false,
  "include_global_state": false,
  "ignore_unavailable": true,
  "rename_replacement": "items_$1-copy",
  "indices": "items_*",
  "rename_pattern": "items_(.+)"
}'

Now the item documents have been restored in to the new index (e.g., my-collection-copy), but the value of the collection field in those documents is still the original value of my-collection. To update these to match the new collection name, run the following Elasticsearch Update By Query command, substituting the old collection name into the term filter and the new collection name into the script parameter:

curl -X "POST" "http://localhost:9200/items_my-collection-copy/_update_by_query" \
     -H 'Content-Type: application/json; charset=utf-8' \
     -d $'{
    "query": {
        "match_all": {}
},
  "script": {
    "lang": "painless",
    "params": {
      "collection": "my-collection-copy"
    },
    "source": "ctx._source.collection = params.collection"
  }
}'

Then, create a new collection through the api with the new name for each of the restored indices:

curl -X "POST" "http://localhost:8080/collections" \
     -H 'Content-Type: application/json' \
     -d $'{
  "id": "my-collection-copy"
}'

Voila! You have a copy of the collection now that has a resource URI (/collections/my-collection-copy) and can be correctly queried by collection name.