Reputation: 8169
I often create specific sweave files to do analyses and produce pdf documents, however, sometimes I want to take an existing R script file and run it while producing a pdf of the commands and output together. I know there is a tangle command that pulls the R commands out of an rnw sweave file to create an R script, but I want the opposite.
Edit: I have considered creating a sweave file that just sources my R script, but I'd need to edit the sweave file each time, and it wouldn't deal very well with plots and graphics.
so for an example R script like this (pseudo code, might not run)
head(mcars)
1+1
x<-3
describe(mcars)
p <- ggplot(mtcars, aes(wt, mpg))
p + geom_point()
print(p)
It would generate
<<echo=TRUE, result=latex>>=
print(xtable(head(mcars)))
@
<<echo=TRUE, include=TRUE>>=
1+1
x<-3
@
<<echo=TRUE, result=latex>>=
print(xtable(describe(mcars)))
@
<<echo=TRUE, fig=TRUE, include=TRUE>>=
p <- ggplot(mtcars, aes(wt, mpg))
p + geom_point()
print(p)
@
Anyone done this ?
Upvotes: 3
Views: 1873
Reputation: 37754
No, but it's not too hard to build one off of source(file, echo=TRUE)
. With the below functions sourced, the following function creates a LaTeX file. Try it out!
latexR("code.R")
I tried it with the following code.R
file:
c("hi",
"there")
1+1
a <- 5
plot(1:a) # a plot
# hi there
data.frame(a=1:5, b=LETTERS[1:5])
Below are the functions. The latex
S3 method is how you would get it to output special LaTeX formatting for a given object type, otherwise it just outputs the output verbatim. It also sets the default graphics device to save to pdf and output the filename.
latex <- function(obj) {
UseMethod("latex")
}
latex.data.frame <- function(obj) {
print(xtable(obj), floating=FALSE)
}
latex.default <- function(obj) {
cat("\\begin{Soutput}\n")
if (isS4(obj))
methods::show(obj)
else print(obj)
cat("\\end{Soutput}\n")
}
fig <- function(...) {
i <<- i+1
n <- paste(nam, formatC(i, width=3, flag="0"), sep="-")
f <- paste(n, "pdf", sep=".")
cat("\\includegraphics{", n, "}\n",sep="")
pdf(file=f, ...)
}
stopfig <- function() { if(dev.cur()!=1) dev.off() }
latexR <- function(file, base=sub(".R","", file)) {
eval.with.vis <- function(expr, envir = parent.frame(), enclos = if (is.list(envir) ||
is.pairlist(envir)) parent.frame() else baseenv())
.Internal(eval.with.vis(expr, envir, enclos))
on.exit(sink())
sink(paste(base,"tex",sep="."))
cat("\\documentclass{article}\n\\usepackage{Sweave}\n\\begin{document}\n")
environment(fig) <- list2env(list(i=0, nam=base))
options(device=fig)
exprs <- parse(file)
srcrefs <- attr(exprs,"srcref")
lastshown <- 0
for(i in seq_along(exprs)) {
srcref <- srcrefs[[i]]
dep <- getSrcLines(attr(srcref, "srcfile"), lastshown+1, srcref[3L])
lastshown <- srcref[3L]
cat("\\begin{Sinput}\n", paste(dep, collapse="\n"), "\n\\end{Sinput}\n", sep="")
yy <- eval.with.vis(exprs[i])
stopfig()
if(yy$visible) latex(yy$value)
}
cat("\\end{document}")
}
Upvotes: 2
Reputation: 30114
I have a package named knitr
under development which may be suitable for you, but I need to warn you that it is still at an early stage of development, so bugs may exist. Read the section on Code Reference in the incomplete manual https://github.com/downloads/yihui/knitr/knitr-manual.pdf
Basically you write an empty chunk in Sweave like
<<my-chunk-label>>=
@
and you use the same label in an external R script (e.g. foo.R
)
## @knitr my-chunk-label
1 + 1; x <- 3
The chunk option ref
is used to specify the path of the script (e.g. ref=foo.R
); knitr
will read the R script and match labels with the Rnw document.
Upvotes: 6