[R] Grep with wildcards across multiple columns

[William Dunlap]

grep(pattern, textVector) returns of the integer indices of the elements
of textVector that match the pattern.  E.g.,
  > grep("T", c("One","Two","Three","Four"))
  [1] 2 3

The '&' operator naturally operates on logical vectors of the same length
(If you give it numbers it silently converts 0 to FALSE and  other numbers
to TRUE.)

The two don't fit together.  You could use grepl(), which returns a logical
vector the length of textVector, as in
   grepl(p1,v1) & grepl(p2,v2)
to figure which entries in the table have v1 matching p1 and v2 matching p2.

Or, you could use
  intersect(grep(p1,v1), grep(p2,v2))
if you want to stick with integer indices.

By the way, the eval(parse(text=paste(...))) business is a good way to make
hard-to-read code and hard-to-read means hard-to-fix.  Just write out the
expression.
  > paste(
  +   "dt2 <- dt[", "grep('", par.fund, "', fund) & ",
  +   "grep('", par.func, "', func) & grep('", par.obj, "', obj)",
  +   ", sum(amount), by=c('code', 'year')]" , sep="")
  [1] "dt2 <- dt[grep('^1.E$', fund) & grep('^1.....$', func) & grep('^...$', obj), sum(amount), by=c('code', 'year')]"

Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com


> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On Behalf
> Of Bush, Daniel P. DPI
> Sent: Thursday, March 14, 2013 2:43 PM
> To: 'r-help at r-project.org'
> Subject: [R] Grep with wildcards across multiple columns
> 
> I have a fairly large data set with six variables set up like the following dummy:
> 
> # Create fake data
> df <- data.frame(code   = c(rep(1001, 8), rep(1002, 8)),
>                  year   = rep(c(rep(2011, 4), rep(2012, 4)), 2),
>                  fund   = rep(c("10E", "10E", "10E", "27E"), 4),
>                  func   = rep(c("110000", "122000", "214000", "158000"), 4),
>                  obj    = rep("100", 16),
>                  amount = round(rnorm(16, 50000, 10000)))
> 
> What I would like to do is sum the amount variable by code and year, filtering rows using
> different wildcard searches in each of three columns: "1?E" in fund, "1??????" in func,
> and "???" in obj. I'm OK turning these into regular expressions:
> 
> # Set parameters
> par.fund <- "10E"; par.func <- "100000"; par.obj <- "000"
> par.fund <- glob2rx(gsub("0", "?", par.fund))
> par.func <- glob2rx(gsub("0", "?", par.func))
> par.obj <- glob2rx(gsub("0", "?", par.obj))
> 
> The problem occurs when I try to apply multiple greps across columns. I'd prefer to use
> data.table since it's so much faster than plyr and I have 159 different sets of parameters
> to run through, but I get the same error setting it up either way:
> 
> # Doesn't work
> library(data.table)
> dt <- data.table(df)
> eval(parse(text=paste(
>   "dt2 <- dt[", "grep('", par.fund, "', fund) & ",
>   "grep('", par.func, "', func) & grep('", par.obj, "', obj)",
>   ", sum(amount), by=c('code', 'year')]" , sep="")))
> # Warning message:
> #   In grep("^1.E$", fund) & grep("^1.....$", func) :
> #   longer object length is not a multiple of shorter object length
> 
> # Also doesn't work
> library(plyr)
> eval(parse(text=paste(
>   "df2 <- ddply(df[grep('", par.fund, "', df$fund) & ",
>   "grep('", par.func, "', df$func) & grep('", par.obj, "', df$obj), ]",
>   ", .(code, year), summarize, amount = sum(amount))" , sep="")))
> # Warning message:
> #   In grep("^1.E$", df$fund) & grep("^1.....$", df$func) :
> #   longer object length is not a multiple of shorter object length
> 
> Clearly, the problem is how I'm trying to combine greps in subsetting rows, but I haven't
> been able to find a solution that works. Any thoughts-preferably something that works
> with data.table?
> 
> DB
> 
> Daniel Bush
> School Finance Consultant
> School Financial Services
> Wisconsin Department of Public Instruction
> PO Box 7841 | Madison, WI 53707-7841
> daniel.bush -at- dpi.wi.gov | sfs.dpi.wi.gov
> Ph: 608-267-9212 | Fax: 608-266-2840
> 
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.