Bulk Data Client

If you want to use this in JS/TS projects, aor if you would like to contribute to this projects see the API Docs

Usage examples

Note: these example are using an open server. Protected server examples would work the same but you need to set your clientId, privateKey and tokenEndpoint in the configuration file. The fhirUrl option can also set in the config file to keep the examples shorter.

Patient-level export

node . -f https://bulk-data.smarthealthit.org/fhir

System-level export

node . --global

Group-level export

node . -g myGroupId

Passing export parameters Group-level export

node . --_since 2010-03 --_type Patient, Observations

Patient-level export with debugging information printed to the console

export NODE_DEBUG=app-request; node . -f https://builk-data.smarthealthit.org/fhir 

For more options see the CLI parameters and configuration options below.

Installation

Prerequisites: Git and NodeJS 15+

1. Clone the repository

   git clone https://github.com/smart-on-fhir/bulk-data-client.git

2. Go into the project directory

   cd bulk-data-client

3. Make sure you are using NodeJS 15+. If you use nvm run:

   nvm use

4. Install dependencies

   npm i

Configuration

A configuration file will have to be created for every server you want to connect to. The way to do that is:

1. Copy the default config file and give it new name:

   cp config/defaults.js config/my-config-1.js

2. Edit your configuration open your newly created configuration file, read the description in the comments for every option and set whatever you need. Feel free to remove unneeded options! See below for details.

The configuration works by loading the default values from config/defaults.js, then merging that with your custom config (overriding the defaults), and finally merging with any CLI parameters (a subset of the config options can be passed as CLI parameters).

Configuration File Options

The Bulk Data Client uses js configuration files, but you can think of them as JSON configuration objects. The only reason to use JS is to allow for comments and type hinting. Below are all the options that can be set in a configuration file.

• string fhirUrl - FHIR server base URL. Can be overridden by the -f or --fhir-url CLI parameter.

• string tokenUrl - The Bulk Data server token URL (use "none" for open servers and "" to try to auto-detect it)

• object privateKey - The private key (as JWK) used to sign authentication tokens. This is not needed for open servers

• string clientId - This is not needed for open servers

• number accessTokenLifetime - The access token lifetime in seconds. Note that the authentication server may ignore or restrict this to its own boundaries

• string reporter - The default reporter is "cli". That works well in terminal and renders some fancy stuff like progress bars. However, this does not look good when your STDOUT ends up in log files. For example, if you are using this tool as part of some kind of pipeline and want to maintain clean logs, then consider changing this to "text". Can be overridden from terminal parameter --reporter.

  Running an export using the (default) cli reporter produces output looking like this:

  Running the same export using the text reporter produces output looking like this:

• string _outputFormat - The value of the _outputFormat parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -F or --_outputFormat

• string _since - The value of the _since parameter for Bulk Data kick-off requests. Can also be partial date like "2002", "2020-03" etc. Can be anything that Moment can parse. Will be ignored if empty or falsy. See https://momentjs.com/docs/#/parsing/. Can be overridden from terminal parameter -F or --_outputFormat

• string _type - The value of the _type parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -t or --_type

• string _elements - The value of the _elements parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -e or --_elements

• string patient - The value of the patient parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -p or --patient

• string includeAssociatedData - The value of the includeAssociatedData parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -i or --includeAssociatedData

• string _typeFilter - The value of the _typeFilter parameter for Bulk Data kick-off requests. Will be ignored if empty or falsy. Can be overridden from terminal parameter -q or --_typeFilter

• boolean global - By default this client will make patient-level exports. If this is set to true, it will make system-level exports instead. Ignored if group is set! Can be overridden from terminal parameter --global

• string group - Id of FHIR group to export. If set, the client will make group-level exports. Can be overridden from terminal parameter -g or --group

• boolean lenient - If true, adds handling=lenient to the prefer request header. This may enable a "retry" option after certain errors. It can also be used to signal the server to silently ignore unsupported parameters. Can be overridden from terminal parameter --lenient

• object requests - Custom options for every request, EXCLUDING the authorization request and any upload requests (in case we use remote destination). Many options are available so be careful what you specify here! See https://github.com/sindresorhus/got/blob/main/documentation/2-options.md. Example:

  requests: {
      https: {
          rejectUnauthorized: true // reject self-signed certs
      },
      timeout: 20000, // 20 seconds custom timeout
      headers: {
          "x-client-id": "whatever" // pass custom headers
      }
  }

• number parallelDownloads - How many downloads to run in parallel. This will speed up the download but can also overload the server. Don't be too greedy and don't set this to more than 10!

• boolean saveManifest - In some cases it might be useful to also save the export manifest file along with the downloaded NDJSON files.

• number ndjsonMaxLineLength - While parsing NDJSON files every single (non-empty) line is parsed as JSON. It is recommended to set a reasonable limit for the line length so that a huge line does not consume the entire memory. This is the maximal acceptable line length expressed as number characters.

• boolean ndjsonValidateFHIRResourceType - If true, verifies that every single JSON object extracted for the NDJSON file has a resourceType property, and that this property equals the expected type reported in the export manifest.

• boolean ndjsonValidateFHIRResourceCount - If the server reports the file count in the export manifest, verify that the number of resources found in the file matches the count reported by the server.

• boolean addDestinationToManifest - The original export manifest will have an url property for each file, containing the source location. It his is set to true, add a destination property to each file containing the path (relative to the manifest file) to the saved file. This is ONLY used if saveManifest is set to true.

• boolean forceStandardFileNames - Sometimes a server may use weird names for the exported files. For example, a HAPI server will use random numbers as file names. If this is set to true files will be renamed to match the standard naming convention - {fileNumber}.{ResourceType}.ndjson.

• boolean downloadAttachments - If this is set to false, external attachments found in DocumentReference resources will not be downloaded. The DocumentReference resources will still be downloaded but no further processing will be done.

• number inlineDocRefAttachmentsSmallerThan - In DocumentReference resources, any attachment elements having an url (instead of inline data) and a size below this number will be downloaded and put inline as base64 data. Then the size property will be updated and the url will be removed. Ignored if downloadAttachments is set to false!

  • To always disable this, set it to 0
  • To always enable this, set it to Infinity (bad idea!)
  • To inline files smaller than 5 MB set it to 5 * 1024 * 1024

• string[] inlineDocRefAttachmentTypes - If an attachment can be inlined (based on its size and the value of the inlineDocRefAttachmentsSmallerThan option), then its mime type will be compared with this list. Only files of listed types will be inlined and the rest will be downloaded into "attachment" subfolder. Example: ["text/plain", "application/pdf"]. Ignored if downloadAttachments is set to false!

• boolean pdfToText - If this is true, attachments of type PDF that are being inlined will first be converted to text and then inlined as base64. Ignored if downloadAttachments is set to false!

• string destination - Examples:

  • s3://bucket-name/optional-subfolder/ - Upload to S3
  • ./downloads - Save to local folder (relative to the config file)
  • downloads - Save to local folder (relative to the config file)
  • /path/to/downloads - Save to local folder (absolute path)
  • file:///path/to/downloads - Save to local folder (file url)
  • http://destination.dev - POST to http
  • http://username:[email protected] - POST to http with basic auth
  • "" - do nothing
  • "none" - do nothing

  Can be overridden from terminal parameter -d or --destination

• string awsRegion - Example: us-east-1. Only used if destination points to S3. The AWS SDK will first look for this in the shared config file (~/.aws/config). Then the SDK will look for an AWS_REGION environment variable. Finally, you can override both of these if you set the awsRegion variable in your bulk-data client config file.

• string awsAccessKeyId - Only used if destination points to S3. The AWS SDK will first look for this in the shared credentials file (~/.aws/credentials). You can override this if you set the awsAccessKeyId variable in your bulk-data client config file, but only if you also set the awsSecretAccessKey.

• string awsSecretAccessKey - Only needed if destination points to S3. The AWS SDK will first look for this in the shared credentials file (~/.aws/credentials). You can override this if you set the awsSecretAccessKey variable in your bulk-data client config file, but only if you also set the awsAccessKeyId.

• object log - Optional logging options (see below)

• boolean log.enabled - Set this to false to disable logging. Optional (defaults to true).

• string log.file - Path to the log file. Absolute, or relative to process CWD. If not provided, the file will be called log.ndjson and will be stored in the downloads folder.

• object log.metadata - Key/value pairs to be added to every log entry. Can be used to add useful information (for example which site imported this data).

• number retryAfterMSec - If the server does not provide Retry-after header use this number of milliseconds before checking the status again.

• complex logResponseHeaders - ResponseHeaders to include in error logs for debugging purposes.

  • As for the complex type, valid values are "all" | "none" | string | RegExp | (string | RegExp)[]
  • When "all" is specified, all responseHeaders are returned. When "none" is specified, no responseHeaders are returned. Otherwise, log any responseHeaders matches against 1...* strings/regexp

• object fileDownloadRetry - A subset of got retry configuration object, determining retry behavior when downloading files.

  • For most scenarios, an object with only a limit: number property will be sufficient. This determines how many times a file download will be retried before failing. Each subsequent attempt will delay using an exponential backoff.
  • For more details on options, see https://github.com/sindresorhus/got/blob/main/documentation/7-retry.md.

Environment Variables

There are two environment that can be passed to the client to modify it's behavior.

• AUTO_RETRY_TRANSIENT_ERRORS - Typically, if the server replies with an error as OperationOutcome having a transient code, the user is asked if (s)he wants to retry. However, if the client runs as part of some kind of automated pipeline (with no human interaction), the we don't want to ask question which no one could answer. AUTO_RETRY_TRANSIENT_ERRORS can be set to truthy or falsy value to pre-answer questions like these.
• SHOW_ERRORS - When an error is thrown, if it contains additional details the user is asked if (s)he wants to see those. Similarly to AUTO_RETRY_TRANSIENT_ERRORS, setting SHOW_ERRORS to boolean-like value will make it so that those error details are always shown or hidden and will avoid having to show question prompts.

Example of running in non-interactive mode:

AUTO_RETRY_TRANSIENT_ERRORS=1 SHOW_ERRORS=1 node . --config myConfigFile.js --reporter text

CLI Parameters

Note that you can pass a --help parameter to see this listed in your terminal

short	long	description
-f	--fhir-url	FHIR server base URL. Must be set either as parameter or in the configuration file.
-F	--_outputFormat	The output format you expect.
-s	--_since	Only include resources modified after this date
-t	--_type	Zero or more resource types to download. If omitted downloads everything.
-e	--_elements	Zero or more FHIR elements to include in the downloaded resources
-p	--patient	Zero or more patient IDs to be included. Implies --post
-i	--includeAssociatedData	String of comma delimited values. When provided, server with support for the parameter and requested values SHALL return a pre-defined set of metadata associated with the request.
-q	--_typeFilter	Experimental _typeFilter parameter passed as is to the server
	--global	Global (system-level) export
	--post	Use POST kick-off requests
-g	--group	Group ID - only include resources that belong to this group. Ignored if --global is set
	--lenient	Sets a "Prefer: handling=lenient" request header to tell the server to ignore unsupported parameters
-d	--destination	Download destination. See config/defaults.js for examples
	--config	Relative path to config file
	--reporter	Reporter to use to render the output. "cli" renders fancy progress bars and tables. "text" is better for log files. Defaults to "cli"
-c	--custom	Custom parameters to be passed to the kick-off endpoint. Example: -c a=1 b=c
	--status	If a status request fails for some reason the client will exit. However, if the status endpoint is printed in the output, you can retry by passing it as --status option here

Features

• Patient-level export
• System-level export
• Group-level export
• All Bulk Data v2 parameters
• Token endpoint auto detection
• Support multiple reporters
• Parallel downloads
• Save manifest
• NDJSON line limit
• NDJSON ResourceType validation
• NDJSON count validation
• Add destination to manifest
• Force standard file names
• Inline DocumentReference attachments
• PDF to Text
• Destination none
• Destination S3
• Destination file://
• Destination directory path
• Destination http
• multiple config files
• tests (~64% coverage)
• Custom kick-off params
• Logging