使用Python中的BeautifulSoup提取表标记值？

Question

我正在尝试编写一个Python脚本来从此页面上的表中提取一些标记值：https://azure.microsoft.com/en-us/pricing/details/virtual-machines/windows/

我已经包含了HTML源代码的截图，但是我无法弄清楚如何提取第6,7,8和9列的价格数据。下面是我已编写的代码。

import requests
import pandas as pd
from bs4 import BeautifulSoup

url = 'https://azure.microsoft.com/en-us/pricing/details/virtual-machines/windows/'

response = requests.get(url)

soup = BeautifulSoup(response.content, 'html.parser')

table1 = soup.find_all('table', class_= 'sd-table')

#writing the first few columns to text file

with open('examplefile.txt', 'w') as r:
    for row in table1.find_all('tr'):
        for cell in row.find_all('td'):
            r.write(cell.text.ljust(5))
        r.write('\n')

我最终试图提取每行的所有值并将其保存到Pandas DataFrame或CSV中。谢谢。

Answer 1

表值似乎嵌入在可以使用json.loads获取的JSON字符串中。然后我们可以通过指示国家地区的"regional"密钥来获得价值。

它有点复杂，但至少它获得了我们放入数据帧的值，如下所示：

import requests
from bs4 import BeautifulSoup
import json
import pandas as pd
import os
import numpy as np

# force maximum dataframe column width
pd.set_option('display.max_colwidth', 0)

url = 'https://azure.microsoft.com/en-us/pricing/details/virtual-machines/windows/'

response = requests.get(url)
soup = BeautifulSoup(response.content, 'html.parser')
tables = soup.find_all('div', {'class': 'row row-size3 column'})

region = 'us-west-2' # Adjust your region here

def parse_table_as_dataframe(table):
    data = []
    header = []
    c5 = c6 = c7 = c8 = []

    rows = []
    columns = []

    name = table.h3.text

    try:
        # This part gets the first word in each column header so the table
        # fits reasonably in the display, adjust to your preference 
        header = [h.text.split()[0].strip() for h in table.thead.find_all('th')][1::]
    except AttributeError:
        return 'N/A'

    for row in table.tbody.find_all('tr'):
        for c in row.find_all('td')[1::]:
            if c.text.strip() not in (u'', u'$-') :
                if 'dash' in c.text.strip():
                    columns.append('-') # replace "‐ &dash:" with a `-`
                else:
                    columns.append(c.text.strip())  
            else:
                try:
                    data_text = c.span['data-amount']
                    # data = json.loads(data_text)['regional']['asia-pacific-southeast']
                    data = json.loads(data_text)['regional'][region]
                    columns.append(data)
                except (KeyError, TypeError):
                    columns.append('N/A')



    num_rows = len(table.tbody.find_all('tr'))
    num_columns = len(header)

    # For debugging
    # print(len(columns), columns)
    # print(num_rows, num_columns)

    df = pd.DataFrame(np.array(columns).reshape(num_rows, num_columns), columns=header)
    return df

for n, table in enumerate(tables):
    print(n, table.h3.text)
    print(parse_table_as_dataframe(table))

从页面中获取24个数据帧，每个表一个：

0 B-series
  Instance Core        RAM Temporary    Pay      One    Three        3
0  B1S      1    1.00 GiB   2 GiB     0.017  0.01074  0.00838  0.00438
1  B2S      2    4.00 GiB   8 GiB     0.065  0.03483  0.02543  0.01743
2  B1MS     1    2.00 GiB   4 GiB     0.032  0.01747  0.01271  0.00871
3  B2MS     2    8.00 GiB   16 GiB    0.122  0.06165  0.04289  0.03489
4  B4MS     4    16.00 GiB  32 GiB    0.229  0.12331  0.08579  0.06979
5  B8MS     8    32.00 GiB  64 GiB    0.438  0.24661  0.17157  0.13957

...

...

23 H-series
  Instance Core         RAM  Temporary    Pay      One    Three        3
0  H8       8    56.00 GiB   1,000 GiB  1.129  0.90579  0.72101  0.35301
1  H16      16   112.00 GiB  2,000 GiB  2.258  1.81168  1.44205  0.70605
2  H8m      8    112.00 GiB  1,000 GiB  1.399  1.08866  0.84106  0.47306
3  H16m     16   224.00 GiB  2,000 GiB  2.799  2.17744  1.68212  0.94612
4  H16mr    16   224.00 GiB  2,000 GiB  3.012  2.32162  1.77675  1.04075
5  H16r     16   112.00 GiB  2,000 GiB  2.417  1.91933  1.51267  0.77667

Answer 2

熊猫可以用read_html独自处理这个问题。然后，您可以在结果框架内清理数据类型等。返回一个匹配数组 - 这是一般的想法：

import pandas as pd

url = 'https://azure.microsoft.com/en-us/pricing/details/virtual-machines/windows/'

dfs = pd.read_html(url, attrs={'class':'sd-table'})

print dfs[0]

希望有所帮助！

Answer 3

soup = find_all ('table', {'class':'sd-table'})