匹配字符串末尾的某些数字

Question

I have a vector of strings:

s <- c('abc1',   'abc2',   'abc3',   'abc11',   'abc12', 
       'abcde1', 'abcde2', 'abcde3', 'abcde11', 'abcde12', 
       'nonsense')

I would like a regular expression to match only the strings that begin with abc and end with 3 , 11 , or 12 . In other words, the regex has to exclude abc1 but not abc11 , abc2 but not abc12 , and so on.

I thought that this would be easy to do with lookahead assertions, but I haven't found a way. Is there one?

---

EDIT: Thanks to posters below for pointing out a serious ambiguity in the original post.

In reality, I have many strings. They all end in digits: some in 0, some in 9, some in the digits in between. I am looking for a regex that will match all strings except those that end with a letter followed by a 1 or a 2. (The regex should also match only those strings that start with abc , but that's an easy problem.)

I tried to use negative lookahead assertions to create such a regex. But I didn't have any success.

---

Thanks to all who replied and commented. Inspired by several of you, I ended up using this combination: grepl('^abc', s) & !grepl('[[:lower:]][12]$', s) .

Answer 1

Is this what you want?

s[grepl("abc.*(3|11|12)", s)]
[1] "abc3"    "abc11"   "abc12"   "abcde3"  "abcde11" "abcde12"

And the excluded strings are:

s[!grepl("abc.*(3|11|12)", s)]
[1] "abc1"     "abc2"     "abcde1"   "abcde2"   "nonsense"

---

Edit: As the comments indicate, there is some ambiguity in your requirements. A more comprehensive regex will test for the string start ^ and string end $ and possibly only allow alphabet characters [[:alpha:]] before the final digits:

s[grepl("^abc[[:alpha:]]*.*(3|11|12)$", s)]
[1] "abc3"    "abc11"   "abc12"   "abcde3"  "abcde11" "abcde12"

---

You can also get grep to return the values directly, by passing the argument value=TRUE , thus saving a bit of duplication in the code:

grep("^abc[[:alpha:]]*.*(3|11|12)$", s, value=TRUE)
[1] "abc3"    "abc11"   "abc12"   "abcde3"  "abcde11" "abcde12"

Answer 2

Instead of one complicated regular expression, in this case I think it's easier to use two simple regular expressions:

s <- c('abc1',   'abc2',   'abc3',   'abc11',   'abc12', 
       'abcde1', 'abcde2', 'abcde3', 'abcde11', 'abcde12', 
       'nonsense')

s[grepl("^abc", s) & grepl("(3|11|12)$", s)]

Answer 3

You could use substring in this case too:

z <- nchar(s)
s[substring(s, 1, 3) == "abc" & substring(s, z) == "3" | 
    substring(s, z-1) %in%  c("12", "11")] 

Answer 4

Looking specifically for the requested numbers gives this:

n <-  c(3,11,12)

s[sub('abc[^[:digit:]]*([[:digit:]]+)$',s, replacement='\\1') %in% n]
 [1] "abc3"    "abc11"   "abc12"   "abcde3"  "abcde11" "abcde12"

This doesn't confuse 11 for 1:

 n <-  c(3,1,12)

s[sub('abc[^[:digit:]]*([[:digit:]]+)$',s, replacement='\\1') %in% n]
 [1] "abc1"    "abc3"    "abc12"   "abcde1"  "abcde3"  "abcde12"

For your edit, not ending in 1 or 2 (and using two regular expressions)

s[grepl('^abc',s) & !(sub('.*[^[:digit:]]([[:digit:]]+)$',s, replacement='\\1') %in% c(1,2))]
[1] "abc3"    "abc11"   "abc12"   "abcde3"  "abcde11" "abcde12"