jdt - JSON Data Tools

Version: 0.5.6

This repository contains a handful of command-line utilities and related code libraries for

processing CSV, JSON, and [Newline Delimited JSON].(http://ndjson.org/) files.

It also contains tools for loading then into:

• MongoDB. This tool should theoretically work with Microsoft Azure Cosmos DB, but it has not been tested. Contributions along those lines are encouraged.

• FHIR servers. This tool has been tested with HAPI using FHIR R4.

Contributing

json-data-tools is free open source software. All pull requests will be considered. Please consider a code contribution for extending jdt's database import capabilities. In addition to MongoDB, it would be useful to import into MarkLogic, Microsoft Azure Cosmos DB, CouchDB, Redis, Elastic Search, etc. We encourage those vendors to add their support to this library.

Command Line Utilities

Commands

• csv2mongo - Converting a CSV into documents directly into a MongoDB database/collection.
• json2mongo - Convert a JSON file object into a record into a MongoDB database/collection.
• jsondir2ndjson - Traverse a directory path and convert all JSON documents into a single NDJSON file.
• jsonlist2ndjson - Import a JSON list file and convert it into NDJSON.
• fhirbundle2ndjson - Convert the FHIR resources within a FHIR Bundle and output them to NDJSON.
• ndjson2mongo - Import an NDJSON file into a MongoDB database/collection.
• ndjsonurl2mongo - Import an NDJSON file, at a given URL, into a MongoDB database/collection.
• ndjson2fhir - Import an NDJSON file into a FHIR server using POST/CREATE or PUT/UPDATE. This now can accept an OAuth2 token.
• ndjsonurl2fhir - Import an NDJSON file, at a given URL, into a FHIR server using POST/CREATE (experimental).
• sftp-ndjson2mongo - Download files from SFTP, then import NDJSON and CSV and import them into MongoDB.
• split_large_files - Split a large CSV, NDJSON or other file into N many lines per file.
• fhirbundle2mongo - Load a FHIR Bundle into a Mongo Database. Each FHIR Resource will be its own collection.
• fhirbundledownsize - Split a large FHIR Bundles into smaller Bundles with N entries per object/file.

Installation

You can install the tool using pip to install json-data-tools on your system.

To install with pip just type:

~$ sudo pip install jdt

Note: If you use sudo, the scripts will be installed at the system level and used by all users.

General Usage

For info about each command type:  
<command> -h

MongoDB Configuration

An accesiable MongoDB instance is not required to run MongoDB based commands. See http://docs.mongodb.org/manual/installation/ for more installation and instructions for your OS.For tools that use MongoDB, you should supply a database_name and collection_name to the command line tools. For csv2mongo as an example

~$ csv2mongo [CSVFILE] [DATABASE] [COLLECTION] 

By default, if you pass in db/collection name that already exists in your MongoDB, the import will only add to it. If you pass the -d option to a command, the specified collection contents will be deleted before importing the new data. The default value for host is the localhost (127.0.0.1) and the port is 27017.

These options can be changed with the --host and --port options respectively.

Command Reference

csv2mongo

csv2mongo convert a CSV into a MongoDB collection. The script expects the first row of data to contain header information. Any whitespace and other funky characters in the header row are auto-fixed by converting to , _, or -.

Usage:

~$ csv2mongo [CSVFILE] [DATABASE] [COLLECTION] 

Example:

~$ csv2mongo npidata_20050523-20140413.csv npi_database npi_collection

{
"num_rows_imported": 9, 
"num_csv_rows": 10, 
"code": 200, 
"message": "Completed."
}

json2mongo

json2mongo imports a JSON object file into a MongoDB document. The file is checked for validity (i.e. {}) before attempting to import it into MongoDB.

Usage:

~$ json2mongo [JSONFILE] [DATABASE] [COLLECTION] 

Example:

~$ json2mongo test.json npi nppes 

{
"num_rows_imported": 1, 
"num_file_errors": 0, 
"code": 200, 
"message": "Completed without errors."
}

jsondir2mongo

jsondir2mongo imports a directory containing files of JSON objects to MongoDB documents. It will walk through subdirectories as well, looking for JSON files. The files are checked for validity (i.e. {}) before attempting to import it each into MongoDB. Files that are not JSON objects are automatically skipped. A summary is returned when the process ends.

Usage:

~$ jsondir2mongo [JSONFILE] [DATABASE] [COLLECTION]

Example:

~$ json2dirmongo data npi nppes 

Example output:

Clearing the collection prior to import.

Start the import of the directory data into the collection test within the database csv2json.

('Getting a list of all files for importing from', '../provider-data-tools/tests')
Done creating file list. Begin file import.
{
    "num_files_attempted": 31, 
    "num_files_imported": 30, 
    "num_file_errors": 1, 
    "errors": [
        "Error writing ../provider-data-tools/tests/json_schema_test.json to Mongo. (<class 'bson.errors.InvalidDocument'>, InvalidDocument(\"key '$schema' must not start with '$'\",), <traceback object at 0x7fe1941fa3b0>)"
    ], 
    "code": 400, 
    "message": "Completed with errors."
}

In the above example, 30/31 JSON files were imported. The error reported that json_schema_test.json was an invalid JSON document and pointed out why.