I have multiple strings, and I want to extract the part that matches. In practice my strings are directories, and I need to choose where to write a file, which is the location that matches in all strings. For example, if you have a vector with three strings:
data.dir <- c("C:\\data\\files\\subset1\\", "C:\\data\\files\\subset3\\", "C:\\data\\files\\subset3\\")
...the part that matches in all strings is "C:\data\files\". How can I extract this?
strsplit and intersect the overlapping parts recursively using Reduce. You can then piece it back together by paste-ing.
paste(Reduce(intersect, strsplit(data.dir, "\\\\")), collapse="\\")
#[1] "C:\\data\\files"
As @g-grothendieck notes, this will fail in certain circumstances like:
data.dir <- c("C:\\a\\b\\c\\", "C:\\a\\X\\c\\")
An ugly hack might be something like:
tail(
Reduce(
intersect,
lapply(strsplit(data.dir, "\\\\"),
function(x) sapply(1:length(x), function(y) paste(x[1:y], collapse="\\") )
)
),
1)
...which will deal with either case.
Alternatively, use dirname if you only ever have one extra directory level:
unique(dirname(data.dir))
#[1] "C:/data/files"
g contains the character positions to successive backslashes in data.dir[1]. From this create a logical vector ok whose ith element is TRUE if the first g[i] characters of all elements in data.dir are the same, i.e. all elements of substr(data.dir, 1, g[i]) are the same. If ok[1] is TRUE then there is a non-zero length common prefix whose length is given by the first g[k] characters of data.dir[1] where k (which equals rle(ok)$lengths[1]) is the leading number of TRUE values in ok; otherwise, there is no common prefix so return "".
g <- gregexpr("\\", data.dir[1], fixed = TRUE)[[1]]
ok <- sapply(g, function(i) all(substr(data.dir[1], 1, i) == substr(data.dir, 1, i)))
if (ok[1]) substr(data.dir[1], 1, g[rle(ok)$lengths[1]]) else ""
For data.dir defined in the question the last line gives:
[1] "C:\\data\\files\\"
