将非结构化 Json 转换为结构化 DataFrame

Question

I'm trying to read this github Json (url below) with information from football teams, games and players

Here's my sample code:

import json
import pandas as pd
import urllib.request
from pandas import json_normalize

load_path = 'https://raw.githubusercontent.com/henriquepgomide/caRtola/master/data/2021/Mercado_10.txt'
games_2021 = json.loads(urllib.request.urlopen(load_path).read().decode('latin-1'))
games_2021 = json_normalize(games_2021)
games_2021

The bad output:

The desired output can be seen in the code below:

pd.read_csv('https://raw.githubusercontent.com/henriquepgomide/caRtola/master/data/2022/rodada-0.csv')

Both urls contain the same information, but the JSON file is in a dictionary schema I guess, where the initial information is translating some of the values columns for players and teams can have, while the other link is already cleaned somehow, in a Csv structure.

Answer 1

Just normalize the 'atleta' key in the json. Or simply just construct that into a DataFrame.

import json
import requests
import pandas as pd

load_path = 'https://raw.githubusercontent.com/henriquepgomide/caRtola/master/data/2021/Mercado_10.txt'
jsonData = requests.get(load_path).json()
games_2021 = pd.json_normalize(jsonData['atletas'])


cols = [x for x in games_2021.columns if 'scout.' not in x]
games_2021 = games_2021[cols]

OR

import json
import requests
import pandas as pd

load_path = 'https://raw.githubusercontent.com/henriquepgomide/caRtola/master/data/2021/Mercado_10.txt'
jsonData = requests.get(load_path).json()
games_2021 = pd.DataFrame(jsonData['atletas']).drop('scout', axis=1)

Output:

print(games_2021)
     atleta_id  ...                                               foto
0        83817  ...  https://s.glbimg.com/es/sde/f/2021/06/04/68300...
1        95799  ...  https://s.glbimg.com/es/sde/f/2020/07/28/e1784...
2        81798  ...  https://s.glbimg.com/es/sde/f/2021/04/19/7d895...
3        68808  ...  https://s.glbimg.com/es/sde/f/2021/04/19/ca9f7...
4        92496  ...  https://s.glbimg.com/es/sde/f/2020/08/28/8c0a6...
..         ...  ...                                                ...
755      50645  ...  https://s.glbimg.com/es/sde/f/2021/06/04/fae6b...
756      69345  ...  https://s.glbimg.com/es/sde/f/2021/05/01/0f714...
757     110465  ...  https://s.glbimg.com/es/sde/f/2021/04/26/a2187...
758     111578  ...  https://s.glbimg.com/es/sde/f/2021/04/27/21a13...
759      38315  ...  https://s.glbimg.com/es/sde/f/2020/10/09/a19dc...

[760 rows x 15 columns]

Then just a matter of reading in each table and merge to get the full:

import json
import requests
import pandas as pd

load_path = 'https://raw.githubusercontent.com/henriquepgomide/caRtola/master/data/2021/Mercado_10.txt'
jsonData = requests.get(load_path).json()

atletas = pd.DataFrame(jsonData['atletas']).drop('scout', axis=1)
clubes = pd.DataFrame(jsonData['clubes'].values())
posicoes = pd.DataFrame(jsonData['posicoes'].values())
status = pd.DataFrame(jsonData['status'].values())

df = atletas.merge(clubes, how='left', left_on='clube_id', right_on='id', suffixes=['', '_clube'])
df = df.merge(posicoes, how='left', left_on='posicao_id', right_on='id', suffixes=['', '_posicao'])
df = df.merge(status, how='left', left_on='status_id', right_on='id', suffixes=['', '_status'])