NHL Schedule: JSON to csv/html

parseSchedule.py

import json

print("Parse start")

# http://live.nhl.com/GameData/SeasonSchedule-20152016.json
sourcefile = "C:/Users/TOM/DataNHL/original/SeasonSchedule.json"
targetfile = "C:/Users/TOM/DataNHL/final/parsed_SeasonSchedule.csv"
htmltargetfile = "C:/Users/TOM/DataNHL/final/parsed_SeasonSchedule.htm"

header_row='NHL_GAME_ID' \
            '|SEASON_ID' \
            '|SUBSEASON_ID' \
            '|GAME_ID' \
            '|YEAR_GAME_DT' \
            '|MONTH_GAME_DT' \
            '|DAY_GAME_DT' \
            '|HOUR_GAME_TM' \
            '|MINUTE_GAME_TM' \
            '|SECOND_GAME_TM' \
            '|AWAY_ID' \
            '|HOME_ID'

html_header_row = '<table border="1" cellpadding="4" cellspacing="0"><tr bgcolor="#CCCCCC">' \
            '<th>NHL_GAME_ID</th>' \
            '<th>SEASON_ID</th>' \
            '<th>SUBSEASON_ID</th>' \
            '<th>GAME_ID</th>' \
            '<th>YEAR_GAME_DT</th>' \
            '<th>MONTH_GAME_DT</th>' \
            '<th>DAY_GAME_DT</th>' \
            '<th>HOUR_GAME_TM</th>' \
            '<th>MINUTE_GAME_TM</th>' \
            '<th>SECOND_GAME_TM</th>' \
            '<th>AWAY_ID</th>' \
            '<th>HOME_ID</th>' \
            '</tr>'

with open(sourcefile,'r') as infile \
        , open(targetfile, 'w') as outfile \
        , open(htmltargetfile, 'w') as htmloutfile:

    json_data = json.load(infile)
    outfile.write(header_row + '\n')
    htmloutfile.write(html_header_row + '\n')

    i=0
    for d in json_data:
        nhl_id=str(d['id']) # json field
        year_id=int(nhl_id[0:4])
        season_id=str(year_id) + str(year_id+1)
        subseason_id=nhl_id[4:6]
        game_id=nhl_id[6:10]

        game_datetime=d['est']  # json field
        split_game_datetime=game_datetime.split(' ')

        game_date=split_game_datetime[0]
        game_year=game_date[0:4]
        game_month=game_date[4:6]
        game_day=game_date[6:8]

        game_time=split_game_datetime[1].split(':')
        game_hour=game_time[0]
        game_minute=game_time[1]
        game_second=game_time[2]

        team_away=d['a']  # json field
        team_home=d['h']  # json field

        outfile.write('{nhlid}'
                      '|{seasonid}'
                      '|{subseasonid}'
                      '|{gameid}'
                      '|{gameyear}'
                      '|{gamemonth}'
                      '|{gameday}'
                      '|{gamehour}'
                      '|{gameminute}'
                      '|{gamesecond}'
                      '|{teamaway}'
                      '|{teamhome}'
                      '\n'.format(
            nhlid=nhl_id
            , seasonid=season_id
            , subseasonid=subseason_id
            , gameid=game_id
            , gameyear=game_year
            , gamemonth=game_month
            , gameday=game_day
            , gamehour=game_hour
            , gameminute=game_minute
            , gamesecond=game_second
            , teamaway=team_away
            , teamhome=team_home
        ))

        htmloutfile.write('<tr>'
                          '<td>{nhlid}</td>'
                          '<td>{seasonid}</td>'
                          '<td>{subseasonid}</td>'
                          '<td>{gameid}</td>'
                          '<td>{gameyear}</td>'
                          '<td>{gamemonth}</td>'
                          '<td>{gameday}</td>'
                          '<td>{gamehour}</td>'
                          '<td>{gameminute}</td>'
                          '<td>{gamesecond}</td>'
                          '<td>{teamaway}</td>'
                          '<td>{teamhome}</td>'
                          '</tr>'
                          '\n'.format(
            nhlid=nhl_id
            , seasonid=season_id
            , subseasonid=subseason_id
            , gameid=game_id
            , gameyear=game_year
            , gamemonth=game_month
            , gameday=game_day
            , gamehour=game_hour
            , gameminute=game_minute
            , gamesecond=game_second
            , teamaway=team_away
            , teamhome=team_home
        ))

        i = i+1

    htmloutfile.write('</table>\n')

print(str(i) + " : records written")
print("Parse end")