Regex 正则表达式在捕获组时捕获所有内容'；不存在

我试图利用正则表达式捕获组从列中提取一些特定的文本。我注意到的一件事是，如果抓捕小组不存在，它会抓住一切

55 g                            <- fine  
45 Gallon                       <- fine  
Amazing Hexagonal Fish Tank.    <- not good.  how to replace with NA?  
92gallon                        <- fine  
30 o                            <- wrongly identified  
29                              <- wrongly identified 
10 gallon                       <- I thought [0-9]{2,3} would grab 2 or 3 digits?  
10 gallon                       <- only 1 of 2 tank sizes identified

下面是我用来创建带有解析文本的新列的代码（R中的data.table语法）：

55 g                            <- fine  
45 Gallon                       <- fine  
Amazing Hexagonal Fish Tank.    <- not good.  how to replace with NA?  
92gallon                        <- fine  
30 o                            <- wrongly identified  
29                              <- wrongly identified 
10 gallon                       <- I thought [0-9]{2,3} would grab 2 or 3 digits?  
10 gallon                       <- only 1 of 2 tank sizes identified

结果如下：

55 g                            <- fine  
45 Gallon                       <- fine  
Amazing Hexagonal Fish Tank.    <- not good.  how to replace with NA?  
92gallon                        <- fine  
30 o                            <- wrongly identified  
29                              <- wrongly identified 
10 gallon                       <- I thought [0-9]{2,3} would grab 2 or 3 digits?  
10 gallon                       <- only 1 of 2 tank sizes identified

---

55g我不确定您期望的确切输出是什么，但以下是我的尝试：

55 g                            <- fine  
45 Gallon                       <- fine  
Amazing Hexagonal Fish Tank.    <- not good.  how to replace with NA?  
92gallon                        <- fine  
30 o                            <- wrongly identified  
29                              <- wrongly identified 
10 gallon                       <- I thought [0-9]{2,3} would grab 2 or 3 digits?  
10 gallon                       <- only 1 of 2 tank sizes identified

x <- c('Nice 55 g fish tank with stand', 
       '45 Gallon Aquarium fish tank and Stand',
       'Amazing Hexagonal Fish Tank.', 
       '92gallon fish tank', 'Fish Tank & Stand $30 obo',
       "2007 PROLINE 29' GRAND SPORT CENTER CONSOLE", 
       '110 gallon tall fish tank',
       '20 and 10 Gallon Aquarium / Fish Tanks')

r <- regmatches(x, gregexpr('\\d{2,3}[^\n]*(?i:g\\b|gallon)', x, perl=TRUE))
unlist({r[sapply(r, length)==0] <- NA; r})

# [1] "55 g"             "45 Gallon"        NA                 "92gallon"        
# [5] NA                 NA                 "110 gallon"       "20 and 10 Gallon"

x off-topic：当通过引用赋值（使用
：=
）时，您不必重新赋值结果。也就是说，
data是的，我还没有对错误识别的数据考虑太多。。但是NA会更好！为什么每个字母都是可选的？只是因为有时人们会说90g、90gal、90gal、90g等等。我试图让代码更智能，以识别这些变化。但也许我可以这样做，它只会寻找一个数字后面跟着一个“Gg”或“Gg”。