Question

是否可以在以下向量上使用正则表达式：

u<-c("first1","sec2","thir33","fourth4","fifth25","sixth16",
     "seven7","eight8","nine9","ten10","eleven11")

获得：

[1] "first.1" "sec.2" "thir3.3" "fourth.4" "fifth2.5" "sixth1.6" "seven.7" "eight.8"
[9] "nine.9" "ten.10" "eleven.11"

这就像我得到的那样接近：

gsub("([A-Za-z]*)([1]{0,1})([0-9]$)","\\1\\.\\2\\3",u)
#[1] "first.1"   "sec.2"     "thir3.3"   "fourth.4"  "fifth2.5"  "sixth.16"  "seven.7"   "eight.8"   "nine.9"    "ten.10"   
#[11] "eleven.11"

注意第六个元素不正确：“six.16”应为“six1.6”。

Answer 1

我没有看到内部正则表达式方法“知道”或有权访问向量中的位置，但是当然可以将其传入并在模式中使用其“as.character”强制值。

sapply(seq_along(u), function(x) sub(
                                     paste("(^.+)(", as.character(x), "$)", sep=""),
                                     "\\1.\\2", u[x]) )
 [1] "first.1"   "sec.2"     "thir3.3"   "fourth.4"  "fifth2.5"  "sixth1.6"  "seven.7"   "eight.8"   "nine.9"   
[10] "ten.10"    "eleven.11"

Answer 2

这不是特别漂亮，但您可以通过以下步骤一步完成：

gsub("([A-Za-z]+)(10|11)?(?:(\\d)(\\d))?([0-9]{0,1}?)$","\\1\\3\\.\\2\\4\\5",u)

或者，您可以将其分解为几个步骤。首先取一位数，然后分别处理2位数的情况。

v <- gsub("([A-Za-z]+)(\\d)$","\\1.\\2",u)
v <- gsub("([A-Za-z]+)(10|11)$","\\1.\\2",v)
v <- gsub("([A-Za-z]+\\d)(\\d)$","\\1.\\2",v)

Answer 3

使用DWin的答案作为跳跃点，你可以通过知道元素1：9,10：99,100：999等等都可以获得一些速度（假设你真正的问题解决了更长的向量）以相同的方式。

所以，获取一些更大的数据

u<-c("first1","sec2","thir33","fourth4","fifth25","sixth16",
     "seven7","eight8","nine9","ten10","eleven11")
u[12:101981]<-NA

set.seed(1)
for(i in 12:101981)u[i]<-paste0(paste(sample(c(LETTERS,1:9),5),collapse=""),i)

lengthu<-length(u)
maxLength<-nchar(lengthu)
theStart<-10^(seq_len(maxLength)-1)
theEnd<-c(theStart[-1]-1,lengthu)

然后使用sapply而不是u中的每个元素，而是使用长度为maxLength的序列

tempans<-sapply(seq_len(maxLength),function(x){
  sub(paste0("(^.*)(\\d{",x,"})"),"\\1.\\2",u[theStart[x]:theEnd[x]])
})
tail(unlist(tempans))
# [1] "DWY96.101976" "UWFCO.101977" "UR5L8.101978" "XBQ9V.101979" "48MTI.101980"
# [6] "75LIS.101981"

head(unlist(tempans))
# [1] "first.1"  "sec.2"    "thir3.3"  "fourth.4" "fifth2.5" "sixth1.6"

正则表达式中r的最后一位或两位数

3 个答案: