I'm a total novice to regex, and have a hard time wrapping my head around it. Right now I have a column filled with strings, but the only relevant text to my analysis is between quotation marks. I've tried this:
response$text <- stri_extract_all_regex(response$text, '"\\S+"')
but when I view response$text, the output comes out like this:
"\"caring\""
How do I change my regex expression so that instead the output reads:
caring
You can use
library(stringi)
response$text <- stri_extract_all_regex(response$text, '(?<=")[^\\s"]+(?=")')
Or, with stringr:
library(stringr)
response$text <- str_extract_all(response$text, '(?<=")[^\\s"]+(?=")')
However, with several words inside quotes, I'd rather use stringr::str_match_all:
library(stringr)
matches <- str_match_all(response$text, '"([^\\s"]+)"')
response$text <- lapply(matches, function(x) x[,2])
See this regex demo.
With the capturing group approach used in "([^\\s"]+)" it becomes possible to avoid overlapping matches between quoted substrings, and str_match_all becomes handy since the matches it returns contain the captured substrings as well (unlike *extract* functions).
Related
I have looked at lots of posts here on SO with suggestions on REGEX patterns to grab texts from parentheses. However, from what I have looked into I cannot find a solution that works.
For example, I have had a look at the following:
R - Regular Expression to Extract Text Between Parentheses That Contain Keyword, Extract text in parentheses in R, regex to pickout some text between parenthesis [duplicate]
In the following order, here were the top answers solutions (with some amendments):
pattern1= '\\([^()]*[^()]*\\)'
pattern2= '(?<=\\()[^()]*(?=\\))'
pattern3= '.*\\((.*)\\).*'
all_patterns = c(pattern1, pattern2, pattern3)
I have used the following:
sapply(all_patterns , function(x)stringr::str_extract('I(data^2)', x))
\\([^()]*[^()]*\\) (?<=\\()[^()]*(?=\\)) .*\\((.*)\\).*
"(data^2)" "data^2" "I(data^2)"
None of these seem to only grab the characters within the brackets, so how can I just grab the characters inside brackets?
Expected output:
data
With str_extract, it would extract all those characters matched in the patterns. Instead, use a regex lookaround to match one or more characters that are not a ^ or the closing bracket ()) ([^\\^\\)]+) that succeeds an opening bracket ((?<=\\() - these are escaped (\\) as they are metacharacters
library(stringr)
str_extract('I(data^2)', '(?<=\\()[^\\^\\)]+')
# [1] "data"
Here is combinations of str_extract and str_remove
library(stringr)
str_extract(str_remove('I(data^2)', '.\\('), '\\w*')
[1] "data"
I have column names in the following format:
col= c('UserLanguage','Q48','Q21...20','Q22...21',"Q22_4_TEXT...202")
I would like to get the column names without everything that is after ...
[1] "UserLanguage" "Q48" "Q21" "Q22" "Q22_4_TEXT"
I am not sure how to code it. I found this post here but I am not sure how to specify the pattern in my case.
You can use gsub.
gsub("\\...*","",col)
#[1] "UserLanguage" "Q48" "Q21" "Q22" "Q22_4_TEXT"
Or you can use stringr
library(stringr)
str_remove(col, "\\...*")
Since . matches any character, we need to "escape" (\) to specify exactly what we want to match in the regular expression (and not use the special behavior of the .). So, to match a period, we would need \.. However, the backslash (\) is used to escape special behavior (e.g., escape symbol in strings) in regexps. So, to create the regular expression, we need an additional backslash, \\. In this case, we want to match additional periods, so we can add those here, hence \\.... Then, * specifies that the previous expression (everything the three periods) may occur 0 or more times.
You could sub and capture the first word in each column:
col <- c("UserLanguage", "Q48", "Q21...20", "Q22...21", "Q22_4_TEXT...202")
sub("^(\\w+).*$", "\\1", col)
[1] "UserLanguage" "Q48" "Q21" "Q22" "Q22_4_TEXT"
The regex pattern used here says to match:
^ from the start of the input
(\w+) match AND capture the first word
.* then consume the rest
$ end of the input
Then, using sub we replace with \1 to retain just the first word.
There is a vector of strings that looks like the following (text with two or more substrings in quotation marks):
vec <- 'ab"cd"efghi"j"kl"m"'
The text within the first pair of quotation marks (cd) contains a useful identifier (cd is the desired output). I have been studying how to use regular expressions but I haven't learned how to find the first and second occurrences of something like quotation marks.
Here's how I have been getting cd:
tmp <- strsplit(vec,split="")[[1]]
paste(tmp[(which(tmp=='\"')[1]+1):(which(tmp=='\"')[2]-1)],collapse="")
"cd"
My question is, is there another way to find "cd" using regular expressions? in order to learn more how to use them. I prefer base R solutions but will accept an answer using packages if that's the only way. Thanks for your help.
Match everything except " then capture everything upto next " and replace captured group by itself.
gsub( '[^"]*"([^"]*).*', '\\1', vec)
[1] "cd"
For detailed explanation of regex you can see this demo
I have this vector Target <- c( "tes_1123_SS1G_340T01", "tes_23_SS2G_340T021". I want to remove anything before SS and anything after T0 (including T0).
Result I want in one line of code:
SS1G_340 SS2G_340
Code I have tried:
gsub("^.*?SS|\\T0", "", Target)
We can use str_extract
library(stringr)
str_extract(Target, "SS[^T]*")
#[1] "SS1G_340" "SS2G_340"
Try this:
gsub(".*(SS.*)T0.*","\\1",Target)
[1] "SS1G_340" "SS2G_340"
Why it works:
With regex, we can choose to keep a pattern and remove everything outside of that pattern with a two-step process. Step 1 is to put the pattern we'd like to keep in parentheses. Step 2 is to reference the number of the parentheses-bound pattern we'd like to keep, as sometimes we might have multiple parentheses-bound elements. See the example below for example:
gsub(".*(SS.*)+(T0.*)","\\1",Target)
[1] "SS1G_340" "SS2G_340"
Note that I've put the T0.* in parentheses this time, but we still get the correct answer because I've told gsub to return the first of the two parentheses-bound patterns. But now see what happens if I use \\2 instead:
gsub(".*(SS.*)+(T0.*)","\\2",Target)
[1] "T01" "T021"
The .* are wild cards by the way. If you'd like to learn more about using regex in R, here's a reference that can get you started.
I looked around both here and elsewhere, I found many similar questions but none which exactly answer mine. I need to clean up naming conventions, specifically replace/remove certain words and phrases from a specific column/variable, not the entire dataset. I am migrating from SPSS to R, I have an example of the code to do this in SPSS below, but I am not sure how to do it in R.
EG:
"Acadia Parish" --> "Acadia" (removes Parish and space before Parish)
"Fifth District" --> "Fifth" (removes District and space before District)
SPSS syntax:
COMPUTE county=REPLACE(county,' Parish','').
There are only a few instances of this issue in the column with 32,000 cases, and what needs replacing/removing varies and the cases can repeat (there are dozens of instances of a phrase containing 'Parish'), meaning it's much faster to code what needs to be removed/replaced, it's not as simple or clean as a regular expression to remove all spaces, all characters after a specific word or character, all special characters, etc. And it must include leading spaces.
I have looked at the replace() gsub() and other similar commands in R, but they all involve creating vectors, or it seems like they do. What I'd like is syntax that looks for characters I specify, which can include leading or trailing spaces, and replaces them with something I specify, which can include nothing at all, and if it does not find the specific characters, the case is unchanged.
Yes, I will end up repeating the same syntax many times, it's probably easier to create a vector but if possible I'd like to get the syntax I described, as there are other similar operations I need to do as well.
Thank you for looking.
> x <- c("Acadia Parish", "Fifth District")
> x2 <- gsub("^(\\w*).*$", "\\1", x)
> x2
[1] "Acadia" "Fifth"
Legend:
^ Start of pattern.
() Group (or token).
\w* One or more occurrences of word character more than 1 times.
.* one or more occurrences of any character except new line \n.
$ end of pattern.
\1 Returns group from regexp
Maybe I'm missing something but I don't see why you can't simply use conditionals in your regex expression, then trim out the annoying white space.
string <- c("Arcadia Parish", "Fifth District")
bad_words <- c("Parish", "District") # Write all the words you want removed here!
bad_regex <- paste(bad_words, collapse = "|")
trimws( sub(bad_regex, "", string) )
# [1] "Arcadia" "Fifth"
dataframename$varname <- gsub(" Parish","", dataframename$varname)