gsubfn (version 0.7)

strapply: Apply a function over a string or strings.

Description

Similar to "'gsubfn'" except instead of performing substitutions it returns the output of "'FUN'".

Usage

strapply(X, pattern, FUN = function(x, ...) x, backref, ..., empty,
	ignore.case = FALSE, perl = FALSE, engine,
	simplify = FALSE, USE.NAMES, combine = c)
strapplyc(X, pattern, backref, ignore.case = FALSE, simplify = FALSE, USE.NAMES, engine)

Arguments

X

list or (atomic) vector of character strings to be used.

pattern

character string containing a regular expression (or character string for "'fixed = TRUE')" to be matched in the given character vector.

FUN

a function, formula, character string, list or proto object to be applied to each element of "'X'". See discussion in gsubfn.

backref

See gsubfn.

empty

If there is no match to a string return this value.

ignore.case

If TRUE then case is ignored in the pattern argument.

perl

If TRUE then engine="R" is used with perl regular expressions.

engine

This argument defaults to getOption("gsubfn.engine"). If that option has not been set engine defaults to the "R" engine if (1) FUN is a proto object or if (2) the R installation does not have tcltk capability. If the "R" default does not apply then it defaults to the "tcl" engine.

…

optional arguments to "'gsubfn'".

simplify

logical or function. If logical, should the result be simplified to a vector or matrix, as in "sapply" if possible? If function, that function is applied to the result with each component of the result passed as a separate argument. Typically if the form is used it will typically be specified as rbind.

USE.NAMES

logical; if "'TRUE'" and if "'X'" is character, use "'X'" as 'names' for the result unless it had names already. Default is FALSE.

combine

combine is a function applied to the components of the result of FUN. The default is "c". "list" is another common choice. The default may change to be "list" in the future.

Value

A list of character strings.

Details

If FUN is a function then for each character string in "X" the pattern is repeatedly matched, each such match along with back references, if any, are passed to the function "FUN" and the output of FUN is returned as a list. If FUN is a formula or proto object then it is interpreted to the way discussed in gsubfn.

If FUN is a proto object or if perl=TRUE is specified then engine="R" is used and the engine argument is ignored.

If backref is not specified and engine="R" is specified or implied then a heuristic is used to calculate the number of backreferences. The primary situation that can fool it is if there are parentheses in the string that are not back references. In those cases the user will have to specify backref. If engine="tcl" then an exact algorithm is used and the problem sentence never occurs.

strapplyc is like strapply but specialized to FUN=c for speed. If the "tcl" engine is not available then it calls strapply and there will be no speed advantage.

See Also

See gsubfn. For regular expression syntax used in tcl see http://www.tcl.tk/man/tcl8.6/TclCmd/re_syntax.htm and for regular expression syntax used in R see the help page for regex.

Examples

# NOT RUN {
strapply("12;34:56,89,,12", "[0-9]+")

# separate leading digits from rest of string
# creating a 2 column matrix: digits, rest
s <- c("123abc", "12cd34", "1e23")
t(strapply(s, "^([[:digit:]]+)(.*)", c, simplify = TRUE)) 

# same but create matrix
strapply(s, "^([[:digit:]]+)(.*)", c, simplify = rbind)

# running window of 5 characters using 0-lookahead perl regexp
# Note that the three ( in the regexp will fool it into thinking there
# are three backreferences so specify backref explicitly.
x <- "abcdefghijkl"
strapply(x, "(.)(?=(....))",  paste0, backref = -2, perl = TRUE)[[1]]

# Note difference.  First gives character vector.  Second is the same.
# Third has same elements but is a list.
# Fourth gives list of two character vectors. Fifth is the same.
strapply("a:b c:d", "(.):(.)", c)[[1]]
strapply("a:b c:d", "(.):(.)", list, simplify = unlist) # same

strapply("a:b c:d", "(.):(.)", list)[[1]]

strapply("a:b c:d", "(.):(.)", c, combine = list)[[1]]
strapply("a:b c:d", "(.):(.)", c, combine = list, simplify = c) # same

# find second CPU_SPEED value given lines of config file
Lines <- c("DEVICE = 'PC'", "CPU_SPEED = '1999', '233'")
parms <- strapply(Lines, "[^ ',=]+", c, USE.NAMES = TRUE, 
	simplify = ~ lapply(list(...), "[", -1))
parms$CPU_SPEED[2]

# return first two words in each string
p <- proto(fun = function(this, x) if (count <=2) x)
strapply(c("the brown fox", "the eager beaver"), "\\w+", p)

# }
# NOT RUN {
# convert to chron
library(chron)
x <- c("01/15/2005 23:32:45", "02/27/2005 01:22:30")
x.chron <- strapply(x, "(../../....) (..:..:..)",  chron, simplify = c)

# time parsing of all 275,546 words from James Joyce's Ulysses
joyce <- readLines("http://www.gutenberg.org/files/4300/4300-8.txt") 
joycec <- paste(joyce, collapse = " ") 
system.time(s <- strapplyc(joycec, "\\w+")[[1]]) 
length(s) # 275546 
# }
# NOT RUN {
# }