This terraform provider allows you to interact with APIs that may not yet have a first-class provider available.
There are a few requirements about how the API must work for this provider to be able to do its thing:
- The API is expected to support the following HTTP methods:
- POST: create an object
- GET: read an object
- PUT: update an object
- DELETE: remove an object
- An "object" in the API has a unique identifier the API will return
- Objects live under a distinct path such that for the path
/api/v1/things
...- POST on
/api/v1/things
creates a new object - GET, PUT and DELETE on
/api/v1/things/{id}
manages an existing object
- POST on
Have a look at the examples directory for some use cases
uri
(string, required): URI of the REST API endpoint. This serves as the base of all requests. Example:https://myapi.env.local/api/v1
. This can also be set with the environment variableREST_API_URI
.insecure
(boolean, optional): When using https, this disables TLS verification of the host. This can also be set with the environment variableREST_API_INSECURE
.username
(string, optional): When set, will use this username for BASIC auth to the API. This can also be set with the environment variableREST_API_USERNAME
.password
(string, optional): When set, will use this password for BASIC auth to the API. This can also be set with the environment variableREST_API_PASSWORD
.headers
(hash of strings, optional): A map of header names and values to set on all outbound requests. This is useful if you want to use a script via the 'external' provider or provide a pre-approved token or change Content-Type fromapplication/json
. Ifusername
andpassword
are set and Authorization is one of the headers defined here, the BASIC auth credentials take precedence.use_cookies
(boolean, optional): Enable cookie jar to persist session. This can also be set with the environment variableREST_API_USE_COOKIES
.timeout
(integer, optional): When set, will cause requests taking longer than this time (in seconds) to be aborted. Default is0
which means no timeout is set. This can also be set with the environment variableREST_API_TIMEOUT
.id_attribute
(string, optional): Defaults toid
. When set, this key will be used to operate on REST objects. For example, if the ID is set to 'name', changes to the API object will be to http://foo.com/bar/VALUE_OF_NAME. This value may also be a '/'-delimeted path to the id attribute if it is multple levels deep in the data (such asattributes/id
in the case of an object{ \"attributes\": { \"id\": 1234 }, \"config\": { \"name\": \"foo\", \"something\": \"bar\"}}
. Lists are also supported, f.e.attributes/items/0/id
in case of an object{ \"attributes\": { \"items\": [{"id": 1234}] } }
. This can also be set with the environment variableREST_API_ID_ATTRIBUTE
.create_method
(string, optional): Defaults toPOST
. The HTTP method used to CREATE objects of this type on the API server. This can also be set with the environment variableREST_API_CREATE_METHOD
.read_method
(string, optional): Defaults toGET
. The HTTP method used to READ objects of this type on the API server. This can also be set with the environment variableREST_API_READ_METHOD
.update_method
(string, optional): Defaults toPUT
. The HTTP method used to UPDATE objects of this type on the API server. This can also be set with the environment variableREST_API_UPDATE_METHOD
.destroy_method
(string, optional): Defaults toDELETE
. The HTTP method used to DELETE objects of this type on the API server. This can also be set with the environment variableREST_API_DESTROY_METHOD
.copy_keys
(array of strings, optional): When set, anyPUT
to the API for an object will copy these keys from the data the provider has gathered about the object. This is useful if internal API information must also be provided with updates, such as the revision of the object.write_returns_object
(boolean, optional): Set this when the API returns the object created on all write operations (POST
,PUT
). This is used by the provider to refresh internal data structures. This can also be set with the environment variableREST_API_WRO
.create_returns_object
(boolean, optional): Set this when the API returns the object created only on creation operations (POST
). This is used by the provider to refresh internal data structures. This can also be set with the environment variableREST_API_CRO
.xssi_prefix
(boolean, optional): Trim the xssi prefix from response string, if present, before parsing. This can also be set with the environment variableREST_API_XSSI_PREFIX
.rate_limit
(float, optional): Set this to limit the number of requests per second made to the API.debug
(boolean, optional): Enabling this will cause lots of debug information to be printed to STDOUT by the API client. This can be gathered by settingTF_LOG=1
environment variable. This can also be set with the environment variableREST_API_DEBUG
.
path
(string, required): The API path on top of the base URL set in the provider that represents objects of this type on the API server.create_path
(string, optional): Defaults topath
. The API path that represents where to CREATE (POST) objects of this type on the API server. The string{id}
will be replaced with the terraform ID of the object if the data contains theid_attribute
.read_path
(string, optional): Defaults topath/{id}
. The API path that represents where to READ (GET) objects of this type on the API server. The string{id}
will be replaced with the terraform ID of the object.update_path
(string, optional): Defaults topath/{id}
. The API path that represents where to UPDATE (PUT) objects of this type on the API server. The string{id}
will be replaced with the terraform ID of the object.create_method
(string, optional): Defaults tocreate_method
set on the provider. Allows per-resource override ofcreate_method
(seecreate_method
provider config documentation).read_method
(string, optional): Defaults toread_method
set on the provider. Allows per-resource override ofread_method
(seeread_method
provider config documentation).update_method
(string, optional): Defaults toupdate_method
set on the provider. Allows per-resource override ofupdate_method
(seeupdate_method
provider config documentation).destroy_method
(string, optional): Defaults todestroy_method
set on the provider. Allows per-resource override ofdestroy_method
(seedestroy_method
provider config documentation).destroy_path
(string, optional): Defaults topath/{id}
. The API path that represents where to DESTROY (DELETE) objects of this type on the API server. The string{id}
will be replaced with the terraform ID of the object.id_attribute
(string, optional): Defaults toid_attribute
set on the provider. Allows per-resource override ofid_attribute
(seeid_attribute
provider config documentation).object_id
(string, optional): Defaults to the id learned by the provider during normal operations andid_attribute
. Allows you to set the id manually. This is used in conjunction with the*_path
attributes.data
(string, required): Valid JSON data that this provider will manage with the API server. This should represent the whole API object that you want to create. The provider's information.force_new
(array of strings, optional): Any changes to these values will result in recreating the resource instead of updating.debug
(boolean, optional): Whether to emit verbose debug output while working with the API object on the server. This can be gathered by settingTF_LOG=1
environment variable.read_search
(map, optional): Custom search forread_path
. This map will takesearch_key
,search_value
,results_key
andquery_string
(see datasource config documentation).
This provider also exports the following parameters:
id
: The ID of the object that is being managed.api_data
: After data from the API server is read, this map will include k/v pairs usable in other Terraform resources as readable objects. Currently the value is the golang fmt package's representation of the value (simple primitives are set as expected, but complex types like arrays and maps contain golang formatting).api_response
: Contains the raw JSON response read back from the API server. Can be parsed withjsondecode
to allow access to deeply nested data.create_response
: Contains the raw JSON response from the initial object creation - use when an API only returns required data during create. Can be parsed withjsondecode
to allow access to deeply nested data.
Note that the *_path
elements are for very specific use cases where one might initially create an object in one location, but read/update/delete it on another path. For this reason, they allow for substitution to be done by the provider internally by injecting the id
somewhere along the path. This is similar to terraform's substitution syntax in the form of ${variable.name}
, but must be done within the provider due to structure. The only substitution available is to replace the string {id}
with the internal (terraform) id
of the object as learned by the id_attribute
.
By default data isn't considered as sensitive. If you want to hide data's
value in output you would need to set environment variable API_DATA_IS_SENSITIVE="true"
.
path
(string, required): The API path on top of the base URL set in the provider that represents objects of this type on the API server.query_string
(string, optional): An optional query string to send when performing the search.search_key
(string, required): When reading search results from the API, this key is used to identify the specific record to read. This should be a unique record such as 'name'.search_value
(string, required): The value of 'search_key' will be compared to this value to determine if the correct object was found. Example: if 'search_key' is 'name' and 'search_value' is 'foo', the record in the array returned by the API with name=foo will be used.results_key
(string, required): When issuing a GET to the path, this JSON key is used to locate the results array. The format is 'field/field/field'. Example: 'results/values'. If omitted, it is assumed the results coming back are already an array and are to be used exactly as-isid_attribute
(string, optional): Defaults toid_attribute
set on the provider. Allows per-resource override ofid_attribute
(seeid_attribute
provider config documentation).debug
(boolean, optional): Whether to emit verbose debug output while working with the API object on the server. This can be gathered by settingTF_LOG=1
environment variable.
This provider also exports the following parameters:
id
: The native ID of the API object as the API server recognizes it.api_data
: After data from the API server is read, this map will include k/v pairs usable in other terraform resources as readable objects. Currently the value is the golang fmt package's representation of the value (simple primitives are set as expected, but complex types like arrays and maps contain golang formatting).api_response
: Contains the raw JSON response read back from the API server. Can be parsed withjsondecode
to allow access to deeply nested data.
This provider supports importing existing resources into the terraform state. Import is done according to the various provider/resource configuation settings to contact the API server and obtain data. That is: if a custom read method, path, or id attribute is defined, the provider will honor those settings to pull data in.
To import data:
terraform import restapi.Name /path/to/resource
See a concrete example here
There are two standard methods of installing this provider detailed in Terraform's documentation. You can place the file in the directory of your .tf file in terraform.d/plugins/{OS}_{ARCH}/
or place it in your home directory at ~/.terraform.d/plugins/{OS}_{ARCH}/
The released binaries are named terraform-provider-restapi_vX.Y.Z-{OS}-{ARCH}
so you know which binary to install. You may need to rename the binary you use during installation to just terraform-provider-restapi_vX.Y.Z
.
Once downloaded, be sure to make the plugin executable by running chmod +x terraform-provider-restapi_vX.Y.Z-{OS}-{ARCH}
.
Pull requests are always welcome! Please be sure the following things are taken care of with your pull request:
go fmt
is run before pushing- Be sure to add a test case for new functionality (or explain why this cannot be done)
- Run the
scripts/test.sh
script to be sure everything works - Ensure new attributes can also be set by environment variables
- Golang v1.11 or newer is installed and
go
is in your path - Terraform is installed and
terraform
is in your path
To make development easy, you can use the Docker image druggeri/tdk as a development environment:
docker run -it --name tdk --rm -v "$HOME/go":/root/go druggeri/tdk
go get github.com/Mastercard/terraform-provider-restapi
cd ~/go/src/github.com/Mastercard/terraform-provider-restapi
#Hack hack hack