Importing Data


As an alternative to sending your data directly to the Nexosis API, you can import your data into the Nexosis platform from external sources.


Importing from AWS S3

If you have a CSV or JSON file hosted out on AWS S3, you can tell the Nexosis Platform to import that data into a DataSet.

To tell Nexosis to import data from S3, you’ll issue a POST request to /imports/s3. The body of this request should have the following information.

  • dataSetName : The name of the Dataset into which the data from the S3 file should be imported.
  • bucket : The S3 bucket containing the file to be imported.
  • path : The path to the file inside of the bucket. The data inside this file will be imported.
  • region (optional) : The AWS region in which your bucket is hosted. This field is optional, if your bucket is in the default AWS region ( us-east-1 ).
  • columns (optional) : The Column Metadata to be applied to the data being imported.
{
  "dataSetName": "Location-A",
  "bucket": "nexosis-sample-data",
  "path": "LocationA.csv",
  "region": "us-east-1",
  "columns": {}
}

The response from this POST will be some details about the import.

{
  "importId": "015d1d17-2a45-4f61-aa97-f21cc4fd656c",
  "type": "s3",
  "status": "requested",
  "dataSetName": "Location-A",
  "parameters": {
    "bucket": "nexosis-sample-data",
    "path": "LocationA.csv",
    "region": "us-east-1"
  },
  "requestedDate": "2017-07-07T12:47:23.682205+00:00",
  "statusHistory": [
    {
      "date": "2017-07-07T12:47:23.682205+00:00",
      "status": "requested"
    }
  ],
  "messages": [],
  "columns": {},
  "links": [
    {
      "rel": "data",
      "href": "https://ml.nexosis.com.com/v1/data/Location-A"
    }
  ]
}

The response will also contain a Nexosis-Import-Status header with the status of the import.

Using AWS Credentials

If your S3 resource is private you can attach an accessKeyId and secretAccessKey to the request in order to make an authenticated request. We will encrypt your keys while in use, and will delete them immediately after the import. However, we still recommend that you use an IAM account with least privilege to read only the intended resource.

Simply add these variables to the json, with everything else above remaining the same:

{
  "dataSetName": "Location-A",
  "bucket": "nexosis-sample-data",
  "path": "LocationA.csv",
  "region": "us-east-1",
  "accessKeyId": "AKIAIOSFODNN7EXAMPLE",
  "secretAccessKey": "wJalrXUtnFEMI/K7MDENG/bPxRfiCYzEXAMPLEKEY"
}

Supported File Extensions

At this time we can support the following file extensions and formats.

  • CSV
  • JSON : If you’re importing a JSON file, the contents of the JSON should be a simple array of JSON objects without any additional metadata:
[
  {
    "col1": "value11",
    "col2": "value21"
  },
  {
    "col1": "value12",
    "col2": "value22"
  }
]
  • gz : You can optionally gzip a file in one of the above formats.

See the guide on Sending Data for more examples.

Importing From Azure

Import from Azure is very similar and allows the same file types and data formats. When importing from Azure you modify the endpoint to /imports/azure and then provide Azure specific directives with the dataset name in the json request payload:

{
	"dataSetName": "MyAzureDataset",
	"connectionString": "BlobEndpoint=https://myblobendpoint.blob.core.windows.net/",
	"container": "mycontainer"
	"blob": "mydatafile.csv"
}

We recommend using SAS tokens whenever possible to provide limited access to the resource with the credentials you provide. As with all credentials, we will encrypt them and delete them after use but encourage best practices with regard to least privilege. Your SAS token then becomes part of the connection string as discussed in this article.

Also note that any folder in your storage path should be made part of the ‘blob’ entry in your json payload.

Importing By URL

Importing by URL is very straightforward and just requires a url property included in your json payload sent to the /imports/url endpoint:

{
	"dataSetName": "MyUrlDataset",
	"url": "https://example.com/data/somepayload.csv"
}

As with the imports above you’ll need to provide a url that returns JSON or CSV content, or that points to a gzip file. Urls are a convenient way to load from diverse sources such as raw github content, temporary secure urls from DropBox, OneDrive, Google Drive, and any other resource that provides secure sharing urls. Again, with an abundance of caution we recommend that you deactivate sharing links after a successful import.

Checking the status of an import

Once you’ve submitted an import, you may want to check back to see when it completes. To see the status of an import you make a GET request to /imports/{importId}.

The response from this endpoint will be in the same format as when you made the initial POST to start the import.

{
  "importId": "015d1d17-2a45-4f61-aa97-f21cc4fd656c",
  "type": "s3",
  "status": "completed",
  "dataSetName": "Location-A",
  "parameters": {
    "bucket": "nexosis-sample-data",
    "path": "LocationA.csv",
    "region": "us-east-1"
  },
  "requestedDate": "2017-07-07T12:47:23.682205+00:00",
  "statusHistory": [
    {
      "date": "2017-07-07T12:47:23.682205+00:00",
      "status": "requested"
    },
    {
      "date": "2017-07-07T12:47:24.8153474+00:00",
      "status": "started"
    },
    {
      "date": "2017-07-07T12:47:27.5910919+00:00",
      "status": "completed"
    }
  ],
  "messages": [],
  "columns": {},
  "links": [
    {
      "rel": "data",
      "href": "https://ml.nexosis.com/v1/data/Location-A"
    }
  ]
}

Listing imports

You can also query the imports you’ve run previously by issuing a GET to /imports.

You can also provide the following parameters in the query string to filter the imports you’ve run.

  • dataSetName : Limits imports to those for a particular dataset

  • requestedAfterDate : Limits imports to those requested on or after the specified date

  • requestedBeforeDate : Limits imports to those requested on or before the specified date

  • page : Zero-based page number of imports to retrieve

  • pageSize : Count of imports to retrieve in each page (max 1000)

The response from this endpoint will be an object containing an array of import records

{
  "items": [
    {
      "importId": "015d1836-ab9e-4839-be01-6dda3d710d06",
      "type": "s3",
      "status": "completed",
      "dataSetName": "s3-import-locationa",
      "parameters": {
        "bucket": "nexosis-sample-data",
        "path": "LocationA.csv",
        "region": "us-east-1"
      },
      "requestedDate": "2017-07-06T14:03:42.326925+00:00",
      "statusHistory": [
        {
          "date": "2017-07-06T14:03:42.326925+00:00",
          "status": "requested"
        },
        {
          "date": "2017-07-06T14:03:43.3510578+00:00",
          "status": "started"
        },
        {
          "date": "2017-07-06T14:03:46.5554317+00:00",
          "status": "completed"
        }
      ],
      "messages": [],
      "columns": null,
      "links": [
        {
          "rel": "data",
          "href": "https://ml.nexosis.com.com/v1/data/s3-import-locationa"
        }
      ]
    }
  ]
}

   Data CSV JSON