python 通过正则表达式和列表理解从字符串中提取数字值

Question

I want to extract this

3.76    2.35    3.30    5.08     NaN    8.44    10.00
3.76    2.35    3.30    4.99    6.63    8.42    10.00
1.50    1.50    1.60    2.00    2.60    3.35    3.85
NaN      NaN    NaN     NaN     NaN     0.00    0.00

from the following return of an bs4 operation:

[<td class="font-bold">Ergebnis je Aktie (unverwässert, nach Steuern)</td>, <td>3,76</td>, 
<td>2,35</td>, <td>3,30</td>, <td>5,08</td>, <td>-</td>, <td>8,44</td>, <td>10,00</td>, <td class="font-
bold">Ergebnis je Aktie (verwässert, nach Steuern)</td>, <td>3,76</td>, <td>2,35</td>, <td>3,30</td>,
 <td>4,99</td>, <td>6,63</td>, <td>8,42</td>, <td>10,00</td>, <td class="font-bold">Dividende je 
Aktie</td>, <td>1,50</td>, <td>1,50</td>, <td>1,60</td>, <td>2,00</td>, <td>2,60</td>, <td>3,35</td>,
 <td>3,85</td>, <td class="font-bold">Gesamtdividendenausschüttung in Mio.</td>, <td>-</td>, <td>-</td>,
 <td>-</td>, <td>-</td>, <td>-</td>, <td>0,00</td>, <td>0,00</td>]

I tried something like

def get_table_entries(element, len_colums):    
        #--------------------------------
        #
        _re_digits = re.compile("-?\d+\.?\d+")
        #--------------------------------
        # find all table entries
        entries = []
        temp = element.findAll("td")
        temp = str(temp)
        #print(temp)
        #--------------------------------
        # replace elements and extract digits from string
        temp = temp.replace('.', '') 
        temp = temp.replace(',', '.')

        print(temp)
        entries += [ n for n in _re_digits.findall(temp)]
        #--------------------------------
        # reshape output array to fit original table shape and return entries
        print(entries)
        entries = np.reshape(entries, (-1, len_colums))

        return entries

But this solution also kicks the minus in <td>-</td> I want to transform into NaN. But still when i keep the minus and replace it via temp = temp.replace('-', 'NaN') I will get an error in the following list comprehension.

Answer 1

Perhaps simplest will be to define a helper function:

def to_float(s): 
    if s == "-": 
        return float("nan") 
    else: 
        return float(s.replace(",", ".")) 

And then just write a basic loop over cells:

values = []
for elem in soup.find_all("td"): 
    try: 
        values.append(to_float(elem.text)) 
    except ValueError: 
        pass 

Now it will be easy to convert to numpy array of desired shape:

>>> np.array(values).reshape(-1, 7)
array([[ 3.76,  2.35,  3.3 ,  5.08,   nan,  8.44, 10.  ],
       [ 3.76,  2.35,  3.3 ,  4.99,  6.63,  8.42, 10.  ],
       [ 1.5 ,  1.5 ,  1.6 ,  2.  ,  2.6 ,  3.35,  3.85],
       [  nan,   nan,   nan,   nan,   nan,  0.  ,  0.  ]])