jsonToCsvConverterGoogleTimeline

jsonToCsvConverterGoogleTimeline.py

#!/usr/bin/python3

import json
import csv
import sys
import datetime
import os


def make_reader(in_json):
    # Open location history data
    with open(in_json) as f:
        json_data = json.load(f)

    print("JSON data loaded.")

    # Process the JSON data
    if 'timelineObjects' in json_data:
        timeline_objects = json_data['timelineObjects']
    else:
        print("Expected 'timelineObjects' in JSON data.")
        return

    # Process placeVisit entries
    for item in timeline_objects:
        if 'placeVisit' in item:
            visit = item['placeVisit']
            location = visit['location']
            duration = visit['duration']

            try:
                dt_start = datetime.datetime.strptime(duration['startTimestamp'], '%Y-%m-%dT%H:%M:%SZ')
            except ValueError:
                dt_start = datetime.datetime.strptime(duration['startTimestamp'], '%Y-%m-%dT%H:%M:%S.%fZ')

            try:
                dt_end = datetime.datetime.strptime(duration['endTimestamp'], '%Y-%m-%dT%H:%M:%SZ')
            except ValueError:
                dt_end = datetime.datetime.strptime(duration['endTimestamp'], '%Y-%m-%dT%H:%M:%S.%fZ')

            longitude = location['longitudeE7'] / 10000000.0
            latitude = location['latitudeE7'] / 10000000.0
            place_id = location['placeId']
            address = location.get('address', 'N/A')
            location_confidence = location.get('locationConfidence', 'N/A')

            yield [dt_start, dt_start.strftime('%Y-%m-%d'), dt_start.strftime('%H:%M:%S'),
                   dt_end, dt_end.strftime('%Y-%m-%d'), dt_end.strftime('%H:%M:%S'),
                   longitude, latitude, place_id, address, location_confidence, 'N/A', 'N/A']

    # Process activitySegment entries
    for item in timeline_objects:
        if 'activitySegment' in item:
            segment = item['activitySegment']
            start_location = segment['startLocation']
            end_location = segment['endLocation']
            duration = segment['duration']
            activity_type = segment['activityType']
            distance = segment.get('distance', 0)

            try:
                dt_start = datetime.datetime.strptime(duration['startTimestamp'], '%Y-%m-%dT%H:%M:%SZ')
            except ValueError:
                dt_start = datetime.datetime.strptime(duration['startTimestamp'], '%Y-%m-%dT%H:%M:%S.%fZ')

            try:
                dt_end = datetime.datetime.strptime(duration['endTimestamp'], '%Y-%m-%dT%H:%M:%SZ')
            except ValueError:
                dt_end = datetime.datetime.strptime(duration['endTimestamp'], '%Y-%m-%dT%H:%M:%S.%fZ')

            start_longitude = start_location['longitudeE7'] / 10000000.0
            start_latitude = start_location['latitudeE7'] / 10000000.0
            end_longitude = end_location['longitudeE7'] / 10000000.0
            end_latitude = end_location['latitudeE7'] / 10000000.0

            yield [dt_start, dt_start.strftime('%Y-%m-%d'), dt_start.strftime('%H:%M:%S'),
                   dt_end, dt_end.strftime('%Y-%m-%d'), dt_end.strftime('%H:%M:%S'),
                   start_longitude, start_latitude, 'N/A', 'N/A', 'N/A', 'N/A',
                   end_longitude, end_latitude, activity_type, distance]


def getFullPath(inPath):
    if not os.path.isabs(inPath):
        # we need to set up the absolute path
        script_path = os.path.abspath(__file__)
        path, file = os.path.split(script_path)
        inPath = os.path.join(path, inPath)
    return inPath


# Read the Parameters
if len(sys.argv) < 3:
    print("Usage: script.py <input_json_file> <output_csv_file>")
    sys.exit(1)

in_file = sys.argv[1]
out_file = sys.argv[2]

in_file = getFullPath(in_file)
out_file = getFullPath(out_file)

features = []
# add the Headers
features.append(['Start Datetime', 'Start Date', 'Start Time', 'End Datetime', 'End Date', 'End Time',
                 'Longitude', 'Latitude', 'Place ID', 'Address', 'Location Confidence',
                 'Start Longitude', 'Start Latitude', 'End Longitude', 'End Latitude', 'Activity Type', 'Distance'])

print("Reading {0}".format(in_file))

reader = make_reader(in_file)

record_count = 0
for r in reader:
    print(r)  # Print each record to see what is being processed
    features.append(r)
    record_count += 1

print(f'Read {record_count} Records')

# write this data
with open(out_file, 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerows(features)

print(f'CSV output written to {out_file}')

This should only work for the old timeline exports only, right? The ones exported with Takeout, which would include a JSON file for each month of the year, and would include the timelineObjects array, and then divide into placeVisit and activitySegment. However, the new timeline export through the Android Location settings has a very different JSON structure, showing semanticSegments and more than the two children types. How do we parse this new JSON to CSV?