将 Pandas DataFrame 单元格中的字典转换为 DataFrame 列

Question

我正在使用以下code 从 timeanddate.com 抓取天气数据：

import requests, re, typing
from bs4 import BeautifulSoup as soup
import contextlib
import pandas as pd

def _remove(d:list) -> list:
    return list(filter(None, [re.sub('\xa0', '', b) for b in d]))

@contextlib.contextmanager
def get_weather_data(url:str, by_url = True) -> typing.Generator[dict, None, None]:
    d = soup(requests.get(url).text if by_url else url, 'html.parser')
    _table = d.find('table', {'id':'wt-his'})
    _data = [[[i.text for i in c.find_all('th')], *[i.text for i in c.find_all('td')]] for c in _table.find_all('tr')]
    [h1], [h2], *data, _ = _data
    _h2 = _remove(h2)
    yield {tuple(_remove(h1)):[dict(zip(_h2, _remove([a, *i]))) for [[a], *i] in data]}

city = 'https://www.timeanddate.com/weather/germany/berlin/historic?hd=20210905'
with get_weather_data(city) as weather:
    #print(weather)

    df = pd.DataFrame.from_dict(weather)
    df = df['Conditions']['Comfort']

    # keep only temp, wind speed, humidity
    df0 = pd.DataFrame()

    df0[['Time','Temp','Wind']].assign(**pd.DataFrame(df0.to_dict()).T)

我想将结果表转换为 Pandas DataFrame，其中每列显示每次的温度、风速等值。对于每个时间戳，所有数据都存储在一个单元格中的字典中，因此我尝试了post 中的建议，但到目前为止没有任何效果。有没有一种简单的 Pythonic 方法可以将每个 DataFrame 行中的这些字典条目转换为列？

Answer 1

转换为列表，然后像 post 一样返回 DataFrame：

with get_weather_data(city) as weather:
    #print(weather)

    df = pd.DataFrame.from_dict(weather)
    df = df['Conditions']['Comfort']

    df0 = pd.DataFrame(df.tolist())