# Create a Pipe object that stores a value and allows command chaining with `$`. ```r Pipe(value = NULL) ``` ## Arguments - `value`: value to pipe (default is `NULL`) ## Returns Pipe object ## Details `Pipe()` function creates a Pipe object that provides object-like command chaining mechanism, which avoids using external operator and can be cleaner than operator-based pipline. `Pipe()` creates a Pipe object that allows using `$` to perform first-argument piping, call `.()` to evaluate an expression with `.` or symbol defined by lambda expression, for side effect, or simply extract an element from the stored value. `$value` or `[]` ends a pipeline and extracts its final value. The functionality of Pipe object fully covers that of the pipe operator `%>>%` and provides more features. For example, Pipe object supports directly subsetting `$value` by `[...]`, extracting element by `[[...]]`, and assigning value by `$item <-`, `[...] <-`, and `[[...]] <-`. A typical usage of Pipe object is to start with `Pipe()` and end with `$value` or `[]`. `print()` and `str()` are implemented for `Pipe` object. Use `header = FALSE` to suppress Pipe header message in printed results. Use `options(Pipe.header = FASLE)` to suppress it globally. If the Pipe object is used in more than one pipelines, a recommended usage is to name the object specially so that it is easy to distinguish the Pipe object from the value it stores. For example, it can start with `p`. ## Examples ```r ## Not run: # Pipe as first-argument using $ Pipe(rnorm(100))$mean() Pipe(rnorm(100))$plot(col="red") # Extract the value from the Pipe object using [] Pipe(rnorm(100))$c(4,5) [] # Pipe to an exrepssion with . or symbol defined in # lambda expression to represent the object Pipe(rnorm(100))$.(1 + .) [] Pipe(rnorm(100))$.(x ~ 1 + x) [] # Pipe for side effect Pipe(rnorm(100))$ .(~ cat("number:",length(.),"\n"))$ summary() Pipe(rnorm(100))$ .(~ x ~ cat("number:",length(x),"\n"))$ summary() # Assignment Pipe(rnorm(100))$ .(~ x)$ mean() Pipe(rnorm(100))$ .(~ x <- length(.))$ mean() Pipe(rnorm(100))% .(x <- c(min(.),max(.)))$ mean() # Extract element with \code{.(name)} Pipe(mtcars)$lm(formula = mpg ~ cyl + wt)$.(coefficients) # Command chaining Pipe(rnorm(100,mean=10))$ log()$ diff()$ plot(col="red") Pipe(rnorm(100))$ density(kernel = "rect")$ plot(col = "blue") # Store an continue piping pipe1 <- Pipe(rnorm(100,mean=10))$log()$diff() pipe1$plot(col="red") # Subsetting, extracting, and assigning p <- Pipe(list(a=1,b=2)) p["a"] p[["a"]] p$a <- 2 p["b"] <- NULL p[["a"]] <- 3 p[length(.)] # . = p$value # Data manipulation with dplyr library(dplyr) Pipe(mtcars)$ select(mpg,cyl,disp,hp)$ filter(mpg <= median(mpg))$ mutate(rmpg = mpg / max(mpg))$ group_by(cyl)$ do(data.frame(mean=mean(.$rmpg),median=median(.$rmpg))) [] # Graphics with ggvis library(ggvis) Pipe(mtcars)$ ggvis(~ mpg, ~ wt)$ layer_points() # Data manipulation with rlist library(rlist) Pipe(list(1,2,3))$ list.map(. + 1)$ list.filter(. <= 5)$ list.sort(.) [] # Lazy evaluation p1 <- Pipe(mtcars)$ ggvis(~ mpg, ~ wt) p1$layer_points() p1$layer_bars() # Stored Pipe f1 <- Pipe(rnorm(100))$plot f1(col="red") f1(col="green") ## End(Not run) ```

Pipe function