Problems in a regular expression to extract names using stringr

Question

I cannot fully understand why my regular expression does not work to extract the info I want. I have an unlisted vector that looks like this:

   text <- c("Senator, 1.4balbal", "rule 46.1, declares",
             "Town, 24", "A Town with a Long Name, 23", "THIS IS A DOCUMENT,23)

I would like to create a regular expression to extract only the name of the "Town", even if the town has a long name as the one written in the vector ("A Town with a Long Name"). I have tried this to extract the name of the town:

   reg.town <- "[[:alpha:]](.+?)+,(.+?)\\d{2}" 
   towns<- unlist(str_extract_all(example, reg.prov))

but I extract everything around the ",".

Thanks in advance,

What is the expected output for the `text` you provided? How can you identify a town name (what are the requirements)? — Wiktor Stribiżew, Mar 28 '17 at 10:40
Maybe you are looking for `str_extract(text, "^\\p{Lu}[\\s\\p{L}]*")`? Can each vector item contain only one town? — Wiktor Stribiżew, Mar 28 '17 at 10:43

score 2 · Accepted Answer · answered Mar 28 '17 at 10:46

2

It looks like a town name starts with a capital letter ([[:upper:]]), ends with a comma (or continues to the end of text if there is no comma) ([^,]+) and should be at the start of the input text (^). The corresponding regex in this case would be:

^[[:upper:]][^,]+

Demo: https://regex101.com/r/QXYtyv/1

answered Mar 28 '17 at 10:46

Dmitry Egorov

9,542
3
22
40

Thanks! you got what I mean. However, for whatever reason, I do not obtain the names of the towns from my document. I obtain characters that look like the last one in my vector in my updated question. – Adrian del rio rodriguez Mar 28 '17 at 11:50

score 1 · Answer 2 · answered Mar 28 '17 at 12:47

1

I have solve the problem thanks to @Dmitry Egorov 's demo post in the comment. the regular expression is this one ([[:upper:]].+?, [[:digit:]])

Thanks for your quick replies!!

answered Mar 28 '17 at 12:47

Adrian del rio rodriguez

97
10

Please check my answer, I believe it is a better option since it shows how to use a base R function to get the matches you need. – Wiktor Stribiżew May 08 '17 at 09:07
Yes, I agree, but Dmitry's link was really usefull for other projects that I have done =) – Adrian del rio rodriguez May 08 '17 at 12:41
1

BTW, see [`\b\p{Lu}[^,]++(?=, \d|$)`](https://regex101.com/r/CYSMS2/1). It might even be a better pattern. – Wiktor Stribiżew May 08 '17 at 12:43

score 1 · Answer 3 · answered May 08 '17 at 09:06

You may use the following regex:

> library(stringr)
> text <- c("Senator, 1.4balbal", "rule 46.1, declares", "Town, 24", "A Town with a Long Name, 23", "THIS IS A DOCUMENT,23")
> towns <- unlist(str_extract_all(text, "\\b\\p{Lu}[^,]++(?=, \\d)"))
> towns
[1] "Senator"                 "Town"                   
[3] "A Town with a Long Name"

The regex matches:

\\b - a leading word boundary
\\p{Lu} - an uppercase letter
[^,]++ - 1+ chars other than a , (possessively, due to ++ quantifier, with no backtracking into this pattern for a more efficient matching)
(?=, \\d) - a positive lookahead that requires a ,, then a space and then any digit to appear immediately after the last non-, symbol matched with [^,]++.

Note you may get the same results with base R using the same regex with a PCRE option enabled:

> towns_baseR <- unlist(regmatches(text, gregexpr("\\b\\p{Lu}[^,]++(?=, \\d)", text, perl=TRUE)))
> towns_baseR
[1] "Senator"                 "Town"                   
[3] "A Town with a Long Name"
>

Problems in a regular expression to extract names using stringr

3 Answers3