function to return all S3 methods applicable to an object

Question

Has anyone put together/found a good method for listing all the S3 methods available for a given object? The built-in methods() function will give all available methods for a specified class, or for a specified generic function, but not for an object.

The example I have in mind is a glm object, which is of (minor?) class "glm" but also inherits from "lm"

g <- glm(y~x,data=data.frame(x=1:10,y=1:10))
class(g)
## [1] "glm" "lm"

There are 35 methods for class "lm" and 22 for "glm". I'm interested in an answer that combines the results of

lapply(class(g),function(x) methods(class=x))

in a sensible way, so that I can immediately see (for example) that there is a glm-specific method for add1, but that the method for alias is inherited from the lm class.

Does someone have a slick way to do this, or does it already exist?

PS Steve Walker's S3-S4-reference class glossary shows that this works automatically for reference classes, where we have to use an object to get the methods (x$getRefClass()$methods()).

Answer 1

Here's an attempt to replicate the "standard" behavior

classMethods <- function(cl) {
    if(!is.character(cl)) {
        cl<-class(cl)
    }
    ml<-lapply(cl, function(x) {
        sname <- gsub("([.[])", "\\\\\\1", paste0(".", x, "$"))
        m <- methods(class=x)
        data.frame(
            m=as.vector(m), 
            c=x, n=sub(sname, "", as.vector(m)),
            attr(m,"info"),
            stringsAsFactors=F
        )
    })
    df<-do.call(rbind, ml)
    df<-df[!duplicated(df$n),]
    structure(df$m, 
        info=data.frame(visible=df$visible, from=df$from), 
        class="MethodsFunction")
}

And then you can try it out with

g <- glm(y~x,data=data.frame(x=1:10,y=1:10))
classMethods(g)
#or classMethods(c("glm","lm"))

and that will return

 [1] add1.glm*           anova.glm           confint.glm*        cooks.distance.glm*
 [5] deviance.glm*       drop1.glm*          effects.glm*        extractAIC.glm*    
 [9] family.glm*         formula.glm*        influence.glm*      logLik.glm*        
[13] model.frame.glm     nobs.glm*           predict.glm         print.glm          
[17] residuals.glm       rstandard.glm       rstudent.glm        summary.glm        
[21] vcov.glm*           weights.glm*        alias.lm*           case.names.lm*     
[25] dfbeta.lm*          dfbetas.lm*         dummy.coef.lm*      hatvalues.lm       
[29] kappa.lm            labels.lm*          model.matrix.lm     plot.lm            
[33] proj.lm*            qr.lm*              simulate.lm*        variable.names.lm* 

   Non-visible functions are asterisked

It's not as elegant or short as Josh's, but I think its a good recreation of the default behavior. It's funny to see that the methods function is itself mostly just a grep across all known function names. I borrowed the gsub stuff from there.

Answer 2

Here's a function that will at least tell you which S3 methods an object will initially trigger:

findMethodsS3 <- function(object) {
    x <- unlist(lapply(class(object),function(x) methods(class=x)))   
    sort(x[!duplicated(tools::file_path_sans_ext(x))])
}

findMethodsS3(g)
#  [1] "add1.glm"           "alias.lm"           "anova.glm"         
#  [4] "case.names.lm"      "confint.glm"        "cooks.distance.glm"
#  [7] "deviance.glm"       "dfbeta.lm"          "dfbetas.lm"        
# [10] "drop1.glm"          "dummy.coef.lm"      "effects.glm"       
# [13] "extractAIC.glm"     "family.glm"         "formula.glm"       
# [16] "hatvalues.lm"       "influence.glm"      "kappa.lm"          
# [19] "labels.lm"          "logLik.glm"         "model.frame.glm"   
# [22] "model.matrix.lm"    "nobs.glm"           "plot.lm"           
# [25] "predict.glm"        "print.glm"          "proj.lm"           
# [28] "qr.lm"              "residuals.glm"      "rstandard.glm"     
# [31] "rstudent.glm"       "simulate.lm"        "summary.glm"       
# [34] "variable.names.lm"  "vcov.glm"           "weights.glm"