Getting names of objects passed with "..."

I use:

foo <- function(...) {
  args <- list(...);
  names(args);
}

/Henrik
Is there a tidy way to get the names of objects passed to a function via the
"..." argument?

rbind/cbind does what I want:

test.func1 <- function(...) {
   nms <- rownames(rbind(..., deparse.level=1))
   print(nms)
}

x <- "some stuff"
second <- "more stuff"
test.func1(first=x, second)
[1] "first"  "second"

The usual 'deparse(substitute())' doesn't do it:

test.func2 <- function(...) {
   nms <- deparse(substitute(...))
   print(nms)
}
test.func2(first=x, second)
[1] "x"

I'm using "nms <- rownames(rbind(...))" as a workaround, which works, but
there must be a neater way!

rbind/cbind are .Internal, so I can't pinch code from there.

Thanks,  Mike.

--
View this message in context: http://www.nabble.com/Getting-names-of-objects-passed-with-%22...%22-tf3850318.html#a10906614
Sent from the R help mailing list archive at Nabble.com.

______________________________________________
R-help at stat.math.ethz.ch 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.

I use:

foo <- function(...) {
 args <- list(...);
 names(args);
}
But that does not do what was asked: it gets the argument names, not the 
object names.  (Did you actually try it?)  It looks from the example that 
he wants the argument name if there is one otherwise the 
deparsed argument value, but clarification would be helpful.
/Henrik

On 6/1/07, Mike Meredith <mmeredith at wcs.org> wrote:
Is there a tidy way to get the names of objects passed to a function via the
"..." argument?

rbind/cbind does what I want:

test.func1 <- function(...) {
   nms <- rownames(rbind(..., deparse.level=1))
   print(nms)
}

x <- "some stuff"
second <- "more stuff"
test.func1(first=x, second)
[1] "first"  "second"

The usual 'deparse(substitute())' doesn't do it:

test.func2 <- function(...) {
   nms <- deparse(substitute(...))
   print(nms)
}
test.func2(first=x, second)
[1] "x"

I'm using "nms <- rownames(rbind(...))" as a workaround, which works, but
there must be a neater way!

rbind/cbind are .Internal, so I can't pinch code from there.

Thanks,  Mike.

--
View this message in context: http://www.nabble.com/Getting-names-of-objects-passed-with-%22...%22-tf3850318.html#a10906614
Sent from the R help mailing list archive at Nabble.com.

______________________________________________
R-help at stat.math.ethz.ch 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.

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595
Thanks, Henrik, but 'foo' doesn't do what I want:

x <- "some stuff"
second <- "more stuff"

foo(first=x, second)
[1] "first" ""

Brian's right:
...he wants the argument name if there is one otherwise the 
deparsed argument value, but clarification would be helpful. 
The function using this compares estimates of animal densities, CIs, etc
using different models, with one object containing the results of one model.
It extracts key results and AIC from these objects and does a summary
matrix, with lowest AIC at the top, so the row names need to reflect the
model used.

If the object name is sufficiently explanatory -- eg. point.est.hazardRate
-- then the deparsed argument value is fine as row name. But we need the
option to be more specific if necessary, eg. with "halfNormal=x1,
hazardRate=x2". Just like 'rbind', in fact.

Thanks,  Mike
View this message in context: http://www.nabble.com/Getting-names-of-objects-passed-with-%22...%22-tf3850318.html#a10910245
Sent from the R help mailing list archive at Nabble.com.
How about

foo <- function(...)
{
    m <- as.list(match.call(expand.dots=TRUE))[-1]
    nm <- names(m)
    for(i in seq_along(m)) if(!nchar(nm[i])) nm[i] <- deparse(m[[i]])
    nm
}

Such things are hard to do from R level, hence the use of match.call to do 
it at C level.

Thanks, Henrik, but 'foo' doesn't do what I want:

x <- "some stuff"
second <- "more stuff"

foo(first=x, second)
[1] "first" ""

Brian's right:
...he wants the argument name if there is one otherwise the
deparsed argument value, but clarification would be helpful.
The function using this compares estimates of animal densities, CIs, etc
using different models, with one object containing the results of one model.
It extracts key results and AIC from these objects and does a summary
matrix, with lowest AIC at the top, so the row names need to reflect the
model used.

If the object name is sufficiently explanatory -- eg. point.est.hazardRate
-- then the deparsed argument value is fine as row name. But we need the
option to be more specific if necessary, eg. with "halfNormal=x1,
hazardRate=x2". Just like 'rbind', in fact.

Thanks,  Mike

Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595
Thanks very much to all of you. It looks like 'match.call' is the key, and
both Brian's and Gabor's solutions work fine. --- Mike.
View this message in context: http://www.nabble.com/Getting-names-of-objects-passed-with-%22...%22-tf3850318.html#a10913978
Sent from the R help mailing list archive at Nabble.com.

Thanks very much to all of you. It looks like 'match.call' is the key, and
both Brian's and Gabor's solutions work fine. --- Mike.
Brian's is shorter but I think the one in my post is a bit more robust:
f1 <- function(...) {
+    m <- as.list(match.call(expand.dots=TRUE))[-1]
+    nm <- names(m)
+    for(i in seq_along(m)) if(!nchar(nm[i])) nm[i] <- deparse(m[[i]])
+    nm
+ }
f2 <- function(...) {
+ x <- list(...)
+ if (is.null(names(x))) names(x) <- ""
+ names(x)[names(x) == ""] <- NA
+ mc <- match.call()[-1]
+ ifelse(is.na(names(x)), as.character(mc), names(x))
+ }
f1(sin, cos)
Error in if (!nchar(nm[i])) nm[i] <- deparse(m[[i]]) :
        argument is of length zero
f2(sin, cos)
[1] "sin" "cos"
Sorry, I responded a bit too hastily last night, without testing the two
functions properly.
Brian's is shorter but I think the one in my post is a bit more robust:
Indeed! 'f1' only works if at least one of the arguments is named. Otherwise
'nm' is NULL and 'nchar(nm[i])' fails. 'f2' seems to have all combinations
covered.  Thanks again!!
f1 <- function(...) {
+    m <- as.list(match.call(expand.dots=TRUE))[-1]
+    nm <- names(m)
+    for(i in seq_along(m)) if(!nchar(nm[i])) nm[i] <- deparse(m[[i]])
+    nm
+ }
f2 <- function(...) {
+ x <- list(...)
+ if (is.null(names(x))) names(x) <- ""
+ names(x)[names(x) == ""] <- NA
+ mc <- match.call()[-1]
+ ifelse(is.na(names(x)), as.character(mc), names(x))
+ }
f1(sin, cos)
Error in if (!nchar(nm[i])) nm[i] <- deparse(m[[i]]) :
        argument is of length zero
f2(sin, cos)
[1] "sin" "cos"

______________________________________________
R-help at stat.math.ethz.ch 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.
View this message in context: http://www.nabble.com/Getting-names-of-objects-passed-with-%22...%22-tf3850318.html#a10922269
Sent from the R help mailing list archive at Nabble.com.