Exception handling for regular expressions in R - regex

Question

I've found several related questions, but haven't found one that I solves my problem yet, please let me know if I'm missing a question that addresses this.
Essentially I want to use a regular expression to find a pattern but with an exception based on the preceding characters. For example, I have the following text object ("muffins") as a vector and I want to match the names ("Sarah","Muffins", and "Bob").:
muffins
[1] "Dear Sarah,"
[2] "I love your dog, Muffins, who is adorable and very friendly. However, I cannot say I enjoy the \"muffins\" he regularly leaves in my front yard. Please consider putting him on a leash outside and properly walking him like everyone else in the neighborhood."
[3] "Sincerely,"
[4] "Bob"
My approach was the search for capitalized words and then exclude words capitalized for grammatical reasons, such as the beginning of a sentence.
pattern = "\\b[[:upper:]]\\w+\\b"
m = gregexpr(pattern,muffins)
regmatches(muffins,m)
This pattern gets me most of the way, returning:
[[1]]
[1] "Dear" "Sarah"
[[2]]
[1] "Muffins" "However" "Please"
[[3]]
[1] "Sincerely"
[[4]]
[1] "Win"
and I can identify some of the sentence beginnings with:
pattern2 = "[.]\\s[[:upper:]]\\w+\\b"
m = gregexpr(pattern2,muffins)
regmatches(muffins,m)
but I can't seem to do both simultaneously, where I say I want pattern where pattern2 is not the case.
I've tried several combinations that I thought would work, but with little success. A few of the ones I tried:
pattern2 = "(?<![.]\\s[[:upper:]]\\w+\\b)(\\b[[:upper:]]\\w+\\b)"
pattern2 = "(^[.]\\s[[:upper:]]\\w+\\b)(\\b[[:upper:]]\\w+\\b)"
Any advice or insight would be greatly appreciated!

Answer 1

You maybe looking for a negative look-behind.
pattern = "(?<!\\.\\s)\\b[[:upper:]]\\w+\\b"
m = gregexpr(pattern,muffins, perl=TRUE)
regmatches(muffins,m)
# [[1]]
# [1] "Dear" "Sarah"
#
# [[2]]
# [1] "Muffins"
#
# [[3]]
# [1] "Sincerely"
#
# [[4]]
# [1] "Bob"
The look behind part (?<!\\.\\s) makes sure there's not a period and a space immediately before the match.

Answer 2

The below regex would match only the names Bob, Sarah and Muffins,
(?<=^)[A-Z][a-z]+(?=$)|(?<!\. )[A-Z][a-z]+(?=,[^\n])|(?<= )[A-Z][a-z]+(?=,$)
DEMO

Answer 3

Trying to use regular expressions to identify names becomes a problem. There is no hope of working reliably. It is very complicated to match names from arbitrary data. If extracting these names is your goal, you need to approach this in a different way instead of simply trying to match an uppercase letter followed by word characters.
Considering your vector is as you posted in your question:
x <- c('Dear Sarah,',
'I love your dog, Muffins, who is adorable and very friendly. However, I cannot say I enjoy the "muffins" he regularly leaves in my front yard. Please consider putting him on a leash outside and properly walking him like everyone else in the neighborhood.',
'Sincerely',
'Bob')
m = regmatches(x, gregexpr('(?<!\\. )[A-Z][a-z]{1,7}\\b(?! [A-Z])', x, perl=T))
Filter(length, m)
# [[1]]
# [1] "Sarah"
# [[2]]
# [1] "Muffins"
# [[3]]
# [1] "Bob"