Question

I have a string that is a LaTeX table. I’m trying to find the n-th (let’s say third) column and wrapping everything inside, say \emph{} without matching the delimiting dollar signs.

I’m looking for the first &...& which is second column. Then find the next &...& which is second grouping and by no coincidence a third column in the table.

My dummy example works but is a bit different, because it has text between two &...&. There’s a little thing I will tackle at a later stage – I need to put & outside the \emph{} call using back- and forward-references.

xy <-  "This is &more or less& a match and here is &another one&.\nSecond line with &occurrance 1& and &occurrance 2&"
gsub("(&.*?&)|(.*?&)(.*)(&.*?&)", "\\1\\2\\3\\\\emph{\\4}", xy, perl = TRUE)
[1] "This is &more or less& a match and here is \\emph{&another one&}.\nSecond line with &occurrance 1& and \\emph{&occurrance 2&}"

When I kick it up a notch to a read set with LaTeX tables (bam!), it’s a bit different. There are no characters between two &...&, which means that one & borders two columns. Having that in mind I removed the (.*). No matter what I try, I can’t get this to work. Any tips?

library(xtable)
data(tli)
tli.table <- xtable(tli[1:5,])
x <- print.xtable(tli.table, print.results = FALSE, include.rownames = FALSE)

cat(x)
% latex table generated in R 2.15.1 by xtable 1.7-0 package
% Thu Jul 26 14:13:39 2012
\begin{table}[ht]
\begin{center}
\begin{tabular}{rlllr}
  \hline
grade & sex & disadvg & ethnicty & tlimth \\ 
  \hline
  6 & M & YES & HISPANIC &  43 \\ 
    7 & M & NO & BLACK &  88 \\ 
    5 & F & YES & HISPANIC &  34 \\ 
    3 & M & YES & HISPANIC &  65 \\ 
    8 & M & YES & WHITE &  75 \\ 
   \hline
\end{tabular}
\end{center}
\end{table}

gsub("(&.*?&)(&.*?&)", "\\1\\\\emph{\\2}", x, perl = TRUE)

Answer 1

Assuming 1^st column is n <- 1 (and not n <- 0), the regex that you should use for replacement of n-th column should be:

(?m)^(?=[^&\n\r]*&)((?:[^&]*&){n-1})\\s*([^&]*?)\\s*(&|\\\\)
                                ↑
                                └ replace this n-1 with real number

and replacement string then has to be \\1\\\\emph{\\2}\\3.

So your replacement code is:

input <- "% latex table generated in R 2.15.1 by xtable 1.7-0 package\n% Thu Jul 26 17:49:09 2012\n\\begin{table}[ht]\n\\begin{center}\n\\begin{tabular}{rlllr}\n  \\hline\ngrade & sex & disadvg & ethnicty & tlimth \\\\ \n  \\hline\n  6 & M & YES & HISPANIC &  43 \\\\ \n    7 & M & NO & BLACK &  88 \\\\ \n    5 & F & YES & HISPANIC &  34 \\\\ \n    3 & M & YES & HISPANIC &  65 \\\\ \n    8 & M & YES & WHITE &  75 \\\\ \n   \\hline\n\\end{tabular}\n\\end{center}\n\\end{table}\n"

n <- 1
regex <- paste(c('(?m)^(?=[^&\n\r]*&)((?:[^&]*&){', n-1, '})\\s*([^&]*?)\\s*(&|\\\\)'), collapse='')
cat(gsub(regex, "\\1\\\\emph{\\2}\\3", input, perl = TRUE))