Expfactory Result JSON Extraction Examples

parse.py

#!/usr/bin/env python

# parse.py
# Run this script on downloaded .json files to extract the "inner" json for an experiment,
# given that the experiment exports data in json. You can do this for one or more files at once
# after you download the script, and python 2 and python 3 should be supported.
# The result will be written in the present working directory with a file named equivalently 
# prefixed with "parsed." For example:
# stroop-results.json --> parsed_stroop-results.json

# Usage 
# python parse.py stroop-results.json
# python parse.py stroop-results.json go-no-go-task-results.json

import json
import sys
import os

def usage():
    print('Usage Examples:')
    print('python parse.py stroop.json')
    print('python parse.py stroop.json feedback.json')

if len(sys.argv) < 2:
    print('Please provide the json files to parse!')
    usage()
    sys.exit(1)


json_files = sys.argv[1:]
for json_file in json_files:
 
    if os.path.exists(json_file):

        # We load the initial json, which has one key "data"   
        with open(json_file , 'r') as fh:
            content = json.load(fh)

        # Then we load the "data" again with json. We do this because
        # it could be the case that the experiment doesn't export json!
        parsed = json.loads(content['data'])
  
        # Save to file, the same name prefixed with "parsed"
        json_output = "parsed_%s" % os.path.basename(json_file) 
        print('Writing output to %s' %json_output)
        with open(json_output, 'w') as fh:
            fh.writelines(json.dumps(parsed, sort_keys=True, indent=4))

parse.R

# Here is an equivalent (interactive) version to parse with R, and two methods (using rjson and jsonlite) are shown.

install.packages('rjson')
install.packages('jsonlite')

# Load the package required to read JSON files.
library("rjson")
library("jsonlite")

# Give the input file name to the function.
result <- rjson::fromJSON(file = "stroop-results.json")

# This needs to be loaded again.
print(result[['data']])

# This method generates a data structure
rjson::fromJSON(result[['data']])

# Here we load again with jsonlite, more of a data frame!
jsonlite::fromJSON(result[['data']])