python之将字符串转换为 float 以求和时忽略所有非浮点值_编程设计

我正在尝试将所有字符串值转换为浮点值以进行添加，但是我的 csv 中存在字符串，并且它返回此错误:

L.append([i, j, sum(i for i in map(float, filter(None,  
k)) if i in {0.5, 1, 2})]) 
ValueError: could not convert string to float: 'l0g0dim'

我怎样才能忽略带有字母、逗号和其他不可转换字符的字符串并毫无问题地对它们求和？

示例文件 3.csv

1754|2014-06-13 07:00:00|0|0.5|0 
1754|2014-06-13 08:00:00|0|2|0.5 
1754|2014-06-13 09:00:00|0|a0|b0|2 
1278|2014-01-26 18:00:00|light subcoatal draft ...|0|0|2|0.5 
1754|2014-06-04 19:00:00|0|leg dim|0|0 
1754|2014-06-13 10:00:00|0|(C) fins|0|0

代码

    import csv 
    import re 
    import time 
    from io import StringIO 
 
replacements = ( 
("(B)", "0"), ("(D)", "1"), ("Entrée air absente", "2"), 
 
("+", "0.5"), ("++", "1"), ("+++", "2"), 
 
("(S) +", "0.5"), ("(S) expi. ++", "1"), ("(S) +++", "2"), 
 
("100", "0"), ("99", "0"), ("98", "0"), ("97", "0"), ("96", "0"), 
("95", "0"), ("94", "1"), ("93", "1"), 
("92", "1"), ("91", "1"), ("90", "1"), ("89", "1"), 
("Bruits de transmission", "0"), ("Fatigué/épuisé", "0"), (" *", "0"), ("tirage sous costal", "0"), ("léger BAN", "0"),   ) 
 
replvalues = dict(replacements) 
regex = "|".join(map(re.escape,                                  
replvalues.keys())) 
repl = lambda x: replvalues.get(x.string[x.start():x.end()]) 
 
with open("file3.csv", "r", encoding='utf-8') as f_in, \ 
    open("file4.csv", "w", encoding='utf-8') as f_out: 
 
 
    for i in f_in: 
        line = re.sub(regex, repl, i) 
        f_out.write(line) 
 
with open("file4.csv", "r", encoding='utf-8') as f_in, \ 
     open("file5.csv", "w", encoding='utf-8') as f_out: 
L = [] 
with f_in as fin: 
    reader = csv.reader(fin, delimiter='|') 
    for i, j, *k in reader: 
      L.append([i, j, sum(i for i in map(float, filter(None, k)) if i in {0.5, 1, 2})]) 
    f_out.write(L) 
print(L)

请您参考如下方法:

为什么过滤两次？

filter(None, k)           # filter 
(... if i in {0.5, 1, 2}) # another filter

您可以过滤一次并通过检查一组字符串中的适当值(CSV 中的预期值)简单地删除非浮点值:

L.append([i, j, sum(float(i) for x in k if x in {'0.5', '1', '2'})]) 
                                       #|<-      filter       ->|

不兼容的字符串值，例如字符串 'leg dim'永远不要通过过滤器，因此不要进入最后的 float称呼。