events2csv.py

#!/usr/bin/python
import os.path
import csv
import cjson
import sys
import gzip
import ntpath

def write_event(prefix, event_type, headers, event):
    data = prepare_data(headers, event)
    path = "%s_%s.csv" % (prefix, event_type)
    with open(path, 'ab') as fp:
        cw = csv.writer(fp, lineterminator="\n", quoting=csv.QUOTE_NONNUMERIC)

        # write headers only to empty file
        if os.stat(path)[6] == 0:
            cw.writerow(headers)

        cw.writerow(data)

def prepare_data(headers, event):
    result = []
    for header in headers:
        if "." in header:
            # handle 2nd level fields
            keys = header.split(".")
            if keys[0] in event:
                if keys[1] == "revenue":
                    revenue = event[keys[0]][keys[1]]
                    if revenue == {}:
                        result.append("")
                    else:
                        for revenue_key in revenue:
                            result.append("%s %d" % (revenue_key, revenue[revenue_key]))
                else:
                    #print " Looking for '"+keys[0]+"' sub '"+keys[1]+"'"
                    if keys[1] in event[keys[0]]:
                        #print " Found '"+keys[0]+"' sub '"+keys[1]+"' : '"+str(event[keys[0]][keys[1]])+"'"
                        result.append(event[keys[0]][keys[1]])
                    else:
                        #print "Can't find '"+keys[0]+"' sub '"+keys[1]+"'"
                        result.append("")                    
            else:
                if keys[0] in event:
                    if keys[1] in event[keys[0]]:
                        result.append(event[keys[0]][keys[1]])
                    else:
                        result.append("")
                else:
                    result.append("")
        else:
            # handle 1st level fields
            if header in event:
                result.append(event[header])
            else:
                result.append("")

    return tuple(result)

#TODO: check with collectors validation logic for missing fields
def get_csv_header(event_type):
    header = []
    common_data_fields = ["data.event_id","data.session_id", "data.user_id", "data.build",
        "data.device","data.platform","data.os_version",
        "data.client_ts"]
    common_fields = ["country_code","arrival_ts","game_id","ip"]
    common_user_meta_fields = ["user_meta.install_ts","user_meta.revenue"]

    if event_type == "quality":
        header =  common_data_fields
        header += ["data.value","data.event_id"]
        header += common_fields
        header += ["user_meta.platform","data.device",
                   "user_meta.os_major", "user_meta.os_minor",
                   "user_meta.sdk_version"]
        header += common_user_meta_fields
    elif event_type == "design":
        header =  common_data_fields
        header += ["data.value"]
        header += common_fields        
        header += common_user_meta_fields
    elif event_type == "progression":
        header =  common_data_fields
        header += ["data.attempt_num"]
        header += common_fields        
        header += common_user_meta_fields   
    elif event_type == "session_end":
        header =  common_data_fields
        header += ["data.attempt_num"]
        header += common_fields        
        header += common_user_meta_fields        
    elif event_type == "error":
        header =  common_data_fields
        header += ["data.severity","data.x","data.y","data.z",
                   "data.message"]
        header += common_fields
        #header += ["user_meta.gender"]
        header += common_user_meta_fields
    elif event_type == "user":
        header =  common_data_fields
        header += ["data.os_major","data.os_minor",
                   "data.platform","data.sdk_version"]
        header += common_fields
        #header += ["user_meta.platform",
        #           "user_meta.os_major", "user_meta.os_minor",
        #           "user_meta.sdk_version"]
        header += common_user_meta_fields
    elif event_type == "business":
        header =  common_data_fields
        header += ["data.event_id"]
        header += common_fields + ["currency", "amount"]
        header += common_user_meta_fields

    return header

def main():
    if len(sys.argv) < 2:
        print "Usage: ./events2csv.py <source_file>"
    else:
        source_name = sys.argv[1]
        if os.path.isfile(source_name):

            print "Uncompressing file..."
            with gzip.open(source_name) as f:
                file_prefix = os.path.splitext(
                    ntpath.basename(source_name))[0].replace(".json", "")

                print "Reading events from file..."
                i = 0
                for raw_event in f:
                    event = cjson.decode(raw_event.encode("ascii", "ignore"))
                    event_data=event["data"]

                    # for entry in event:
                    #    print "Event entry " + entry
                    # print "Event category '" + event_data["category"] +"' ID '" + event_data["event_id"] + "'"
                    headers = get_csv_header(event_data["category"])

                    # report progress
                    sys.stdout.write("\r%d rows written..." % i)
                    sys.stdout.flush()

                    write_event(file_prefix, event_data["category"], headers, event)
                    i = i + 1
                print "\nDone"
        else:
            print "Supplied source file does not exists!"

if __name__ == '__main__':
    main()