Merge branch 'master' of https://github.com/ropensci/plotly into feature/hovermode

Author: cpsievert

CONTRIBUTING.md

@@ -0,0 +1,35 @@
+# Contributing to plotly.js
+
+## Opening issues
+
+Search for existing and closed issues. If your problem or idea is not addressed yet, [please open a new issue](https://github.com/ropensci/plotly/issues/new)
+
+Bug reports __must__ have a [reproducible example](http://adv-r.had.co.nz/Reproducibility.html) and include the output of `devtools::session_info()` (instead of `sessionInfo()`).
+
+GitHub issues are great for reporting bugs and requesting new features, but implementation questions are better suited for Stack Overflow (tagged
+[`plotly`](https://stackoverflow.com/questions/tagged/plotly)) or on
+community.plot.ly (tagged [`plotly-js`](http://community.plot.ly/c/plotly-js)).
+
+## Development guidelines
+
+If you'd like to contribute changes to plotly, we use [the GitHub flow](https://guides.github.com/introduction/flow/index.html) for proposing, submitting, reviewing, and accepting changes. If you haven't done this before, Hadley Wickham provides a nice overview of git (<http://r-pkgs.had.co.nz/git.html>), as well as best practices for submitting pull requests (<http://r-pkgs.had.co.nz/git.html#pr-make>). We also recommend using his style guide when writing code (<http://adv-r.had.co.nz/Style.html>).
+
+If your pull request fixes a bug, or implements a new feature, it's a good idea to write a test (<http://r-pkgs.had.co.nz/tests.html>) to demonstrate it's working. If you'd like to closely simulate the tests that run when you submit your pull request, open R under your local plotly git repo, then do the following:
+
+```r
+# the pull request number is arbitrary when running locally
+Sys.setenv('TRAVIS_PULL_REQUEST' = '1')
+Sys.setenv('TRAVIS_COMMIT' = substr(system('git rev-parse HEAD', intern = T), 1, 7))
+devtools::load_all(); source('tests/testthat.R', chdir = TRUE)
+```
+
+You can also build a ggplot2/plotly comparison table:
+
+```r
+Sys.setenv('PLOTLY_TEST' = 'TRUE')
+devtools::load_all(); source('tests/testthat.R', chdir = TRUE)
+```
+
+## Code of Conduct
+
+We want to encourage a warm, welcoming, and safe environment for contributing to this project. See the [code of conduct](https://github.com/ropensci/plotly/blob/master/CONDUCT.md) for more information.

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: plotly
 Title: Create Interactive Web Graphics via Plotly's JavaScript Graphing Library
-Version: 2.0.18
+Version: 2.1.3
 Authors@R: c(person("Carson", "Sievert", role = c("aut", "cre"),
     email = "[email protected]"),
     person("Chris", "Parmer", role = c("aut", "cph"),

NAMESPACE

@@ -4,6 +4,7 @@ S3method(print,figure)
 S3method(print,plotly)
 export("%>%")
 export(add_trace)
+export(as.widget)
 export(config)
 export(embed_notebook)
 export(get_figure)

NEWS

@@ -1,3 +1,23 @@
+2.1.3 -- 12 Jan 2015
+
+Fix size and alpha translation for geom_point. Fixes #386
+
+2.1.2 -- 11 Jan 2015
+
+Upgraded to plotlyjs 1.4.1. For a list of changes, see https://github.com/plotly/plotly.js/releases/tag/v1.4.1
+
+2.1.1 -- 11 Jan 2015
+
+Upgraded to plotlyjs 1.4. For a list of changes, see https://github.com/plotly/plotly.js/releases/tag/v1.4.0
+
+2.1.0 -- 29 Dec 2015
+
+plot_ly() now defaults to inherit=FALSE and plotly_build() is now idempotent. Fixes #280 and #277. See #368 for details.
+
+2.0.19 -- 23 Dec 2015
+
+Added as.widget() function for conveniency in converting plotly object to htmlwidget objects. See #294.
+
 2.0.18 -- 22 Dec 2015
 
 Fix #365

R/plotly.R

@@ -65,7 +65,7 @@
 #' 
 plot_ly <- function(data = data.frame(), ..., type = "scatter",
                     group, color, colors, symbol, symbols, size,
-                    width = NULL, height = NULL, inherit = TRUE, 
+                    width = NULL, height = NULL, inherit = FALSE, 
                     evaluate = FALSE) {
   # "native" plotly arguments
   argz <- substitute(list(...))
@@ -170,7 +170,7 @@ layout <- function(p = last_plot(), ...,
     enclos = parent.frame()
   )
   p <- last_plot(p)
-  p$layout <- c(p$layout, list(layout))
+  p$layout <- c(p$layout, list(layout = layout))
   if (evaluate) p <- plotly_build(p)
   hash_plot(data, p)
 }
@@ -265,9 +265,6 @@ plotly_build <- function(l = last_plot()) {
   # ggplot objects don't need any special type of handling
   if (is.ggplot(l)) return(gg2list(l))
   l <- get_plot(l)
-  # plots without NSE don't need it either
-  nmz <- c(lapply(l$data, names), lapply(l$layout, names), lapply(l$style, names))
-  if (!all(c("args", "env") %in% unlist(nmz))) return(structure(l, class = unique("plotly", class(l))))
   # assume unnamed list elements are data/traces
   nms <- names(l)
   idx <- nms %in% ""
@@ -276,14 +273,22 @@ plotly_build <- function(l = last_plot()) {
   } else if (any(idx)) {
     c(data = c(l$data, l[idx]), l[!idx])
   } else l
-  dats <- list()
+  # carry over properties, if necessary (but don't carry over evaluation envir)
+  if (length(l$data) > 1 && isTRUE(l$data[[1]]$inherit)) {
+    d <- l$data[[1]]
+    d <- d[!names(d) %in% c("env", "enclos")]
+    for (i in seq.int(2, length(l$data))) {
+      l$data[[i]] <- modifyList(l$data[[i]], d)
+    }
+  }
+  # 'x' is the same as 'l', but with arguments evaluated
+  # this is ugly, but I think it is necessary, since we don't know how many 
+  # traces we have until we evaluate args and call traceify() (or similar)
+  x <- list()
   for (i in seq_along(l$data)) {
     d <- l$data[[i]]
-    # if appropriate, evaluate trace arguments in a suitable environment
-    idx <- names(d) %in% c("args", "env", "enclos")
-    if (sum(idx) == 3) {
-      dat <- c(d[!idx], eval(d$args, as.list(d$env, all.names = TRUE), d$enclos))
-      dat[c("args", "env", "enclos")] <- NULL
+    if (should_eval(d)) {
+      dat <- do_eval(d)
       # start processing specially named arguments
       s <- dat[["size"]]
       if (!is.null(s)) {
@@ -309,49 +314,30 @@ plotly_build <- function(l = last_plot()) {
       has_group <- !is.null(dat[["group"]])
       if (has_color) {
         title <- as.list(d$args)[["color"]] %||% as.list(d$args)[["z"]] %||% ""
-        dats <- c(dats, colorize(dat, title))
+        x$data <- c(x$data, colorize(dat, title))
       }
       # TODO: add a legend title (is this only possible via annotations?!?)
-      if (has_symbol) dats <- c(dats, symbolize(dat))
-      if (has_group) dats <- c(dats, traceify(dat, "group"))
-      if (!has_color && !has_symbol && !has_group) dats <- c(dats, list(dat))
+      if (has_symbol) x$data <- c(x$data, symbolize(dat))
+      if (has_group) x$data <- c(x$data, traceify(dat, "group"))
+      if (!has_color && !has_symbol && !has_group) x$data <- c(x$data, list(dat))
     } else {
-      dats <- c(dats, list(d))
-    }
-  }
-  x <- list(data = dats)
-  # carry over properties/data from first trace (if appropriate)
-  if (length(x$data) > 1 && isTRUE(l$data[[1]]$inherit)) {
-    for (i in seq.int(2, length(x$data))) {
-      x$data[[i]] <- modifyList(x$data[[1]], x$data[[i]])
+      x$data <- c(x$data, list(d))
     }
   }
-  # layout() tacks on an unnamed list element to potentially pre-existing
-  # layout(s). Note that ggplotly() will return a named list 
-  # of length n >= 1 (so we need to carefully merge them ).
+  # it's possible have nested layouts (e.g., plot_ly() %>% layout() %>% layout())
   nms <- names(l$layout)
-  if (!is.null(nms) && any(idx <- nms %in% "")) {
-    # TODO: does this always preserve the correct order to layouts?
-    # (important since we use modifyList at a later point)
-    l$layout <- c(list(l$layout[!idx]), l$layout[idx])
-  }
+  idx <- nms %in% "layout"
+  l$layout <- c(list(l$layout[!idx]), setNames(l$layout[idx], NULL))
   for (i in seq_along(l$layout)) {
-    layout <- l$layout[[i]]
-    idx <- names(layout) %in% c("args", "env", "enclos")
-    x$layout[[i]] <- if (sum(idx) == 3) {
-      c(layout[!idx], eval(layout$args, as.list(layout$env, all.names = TRUE), layout$enclos)) 
-    } else {
-      layout
-    }
+    x$layout[[i]] <- perform_eval(l$layout[[i]])
   }
   x$layout <- Reduce(modifyList, x$layout)
   # if style is not null, use it to modify existing traces
   if (!is.null(l$style)) {
     for (i in seq_along(l$style)) {
-      sty <- l$style[[i]]
-      idx <- names(sty) %in% c("args", "env", "enclos")
-      new_sty <- if (sum(idx) == 3) c(sty[!idx], eval(sty$args, as.list(sty$env, all.names = TRUE), sty$enclos)) else sty
-      for (k in sty$traces) x$data[[k]] <- modifyList(x$data[[k]], new_sty)
+      sty <- perform_eval(l$style[[i]])
+      for (k in l$style[[i]]$traces) 
+        x$data[[k]] <- modifyList(x$data[[k]], sty)
     }
   }
   # add appropriate axis title (if they don't already exist)
@@ -365,26 +351,6 @@ plotly_build <- function(l = last_plot()) {
   if (!is.null(a) && !is.null(names(a))) {
     x$layout$annotations <- list(x$layout$annotations)
   }
-  # search for keyword args in traces and place them at the top level
-  kwargs <- lapply(x$data, function(z) z[get_kwargs()])
-  # 'top-level' keywords args take precedence
-  kwargs <- Reduce(modifyList, c(kwargs, list(x[get_kwargs()])))
-  # empty keyword arguments can cause problems
-  kwargs <- kwargs[sapply(kwargs, length) > 0]
-  # try our damndest to assign a sensible filename
-  if (is.null(kwargs$filename)) {
-    kwargs$filename <- 
-      as.character(kwargs$layout$title) %||% 
-      paste(
-        c(kwargs$layout$xaxis$title, 
-          kwargs$layout$yaxis$title, 
-          kwargs$layout$zaxis$title), 
-        collapse = " vs. "
-      ) %||%
-      "plot from api" 
-  }
-  # tack on keyword arguments
-  x <- c(x, kwargs)
   # traces shouldn't have any names
   x$data <- setNames(x$data, NULL)
   # if this is a non-line scatter trace and no hovermode exists, 

R/plotly_POST.R

@@ -26,7 +26,14 @@
 
 plotly_POST <- function(x, filename, fileopt = "new", world_readable = TRUE) {
   x <- plotly_build(x)
-  if (!missing(filename)) x$filename <- filename
+  x$filename <- if (!missing(filename)) { 
+    filename
+  } else {
+    # try our damndest to assign a sensible filename
+    x$filename %||% as.character(x$layout$title) %||% 
+      paste(c(x$layout$xaxis$title, x$layout$yaxis$title, x$layout$zaxis$title), 
+            collapse = " vs. ") %||% "plot from api" 
+  }
   if (!is.null(x$fileopt)) 
     warning("fileopt was specified in the wrong place. Please specify in plotly_POST()")
   x$fileopt <- fileopt

R/print.R

@@ -1,33 +1,37 @@
 #' Print a plotly object
 #' 
 #' @param x a plotly object
-#' @param ... additional arguments (currently ignored)
+#' @param ... additional arguments
 #' @export
 #' @importFrom htmlwidgets createWidget
 #' @importFrom htmlwidgets sizingPolicy
 print.plotly <- function(x, ...) {
-  w <- toWidget(x)
-  get("print.htmlwidget", envir = asNamespace("htmlwidgets"))(w)
+  if (!inherits(x, "htmlwidget")) x <- as.widget(x)
+  get("print.htmlwidget", envir = asNamespace("htmlwidgets"))(x, ...)
 }
 
 #' Print a plotly object in a knitr doc
 #' 
 #' @param x a plotly object
 #' @param options knitr options.
-#' @param ... additional arguments (currently ignored)
+#' @param ... additional arguments
 #' @export
 knit_print.plotly <- function(x, options, ...) {
-  w <- toWidget(x)
-  get("knit_print.htmlwidget", envir = asNamespace("htmlwidgets"))(w, options = options)
+  if (!inherits(x, "htmlwidget")) x <- as.widget(x)
+  get("knit_print.htmlwidget", envir = asNamespace("htmlwidgets"))(x, options = options, ...)
 }
 
 #' Convert a plotly object to an htmlwidget object
 #' 
-#' Users shouldn't need to use this function. It's exported for internal reasons.
-#' 
 #' @param x a plotly object.
+#' @param ... other options passed onto \code{htmlwidgets::createWidget}
+#' @export
+#' @examples \dontrun{
+#' p <- plot_ly(mtcars, x = mpg, y = disp, mode = "markers")
+#' htmlwidgets::saveWidget(as.widget(p), "index.html")
+#' }
 #' 
-toWidget <- function(x) {
+as.widget <- function(x, ...) {
   p <- plotly_build(x)
   # set some margin defaults if none are provided
   p$layout$margin <- modifyList(
@@ -41,13 +45,17 @@ toWidget <- function(x) {
     x = p,
     width = p$width,
     height = p$height,
-    htmlwidgets::sizingPolicy(
+    sizingPolicy = htmlwidgets::sizingPolicy(
       padding = 5, 
       browser.fill = TRUE
-    )
+    ),
+    ...
   )
 }
 
+# for legacy reasons
+toWidget <- as.widget
+
 #' Print a plotly figure object
 #' 
 #' @param x a plotly figure object

R/process.R

@@ -28,6 +28,7 @@ process.figure <- function(resp) {
   fig <- con$payload$figure
   fig$url <- sub("apigetfile/", "~", resp$url)
   fig <- add_boxed(fig)
+  fig$data[[1]]$inherit <- FALSE
   # any reasonable way to return a data frame?
   hash_plot(data.frame(), fig)
 }

R/shiny.R

@@ -26,6 +26,6 @@ plotlyOutput <- function(outputId, width = "100%", height = "400px") {
 renderPlotly <- function(expr, env = parent.frame(), quoted = FALSE) {
   if (!quoted) { expr <- substitute(expr) } # force quoted
   # https://github.com/ramnathv/htmlwidgets/issues/166#issuecomment-153000306
-  expr <- as.call(list(call(":::", quote("plotly"), quote("toWidget")), expr))
+  expr <- call("as.widget", expr)
   shinyRenderWidget(expr, plotlyOutput, env, quoted = TRUE)
 }

R/trace_generation.R

@@ -590,6 +590,10 @@ geom2trace <- list(
     )
   },
   point=function(data, params){
+    # params contains unique values, but we need _all_ values from the data
+    for (i in names(params)) {
+      if (length(params[[i]]) > 1) params[[i]] <- data[[i]]
+    }
     L <- list(
       x = data$x,
       y = data$y,

R/utils.R

@@ -190,6 +190,25 @@ plotly_headers <- function(type = "main") {
   httr::add_headers(.headers = h)
 }
 
+
+perform_eval <- function(x) {
+  if (should_eval(x)) do_eval(x) else x
+}
+
+# env/enclos are special properties specific to the R API 
+# if they appear _and_ are environments, then evaluate arguments
+# (sometimes figures return these properties but evaluation doesn't make sense)
+should_eval <- function(x) { 
+  any(vapply(x[c("env", "enclos")], is.environment, logical(1))) 
+}
+
+# perform evaluation of arguments, keeping other list elements
+do_eval <- function(x) {
+  y <- c(x, eval(x$args, as.list(x$env, all.names = TRUE), x$enclos))
+  y[c("args", "env", "enclos")] <- NULL
+  y
+}
+
 # try to write environment variables to an .Rprofile
 cat_profile <- function(key, value, path = "~") {
   r_profile <- file.path(normalizePath(path, mustWork = TRUE),

README.md

@@ -20,7 +20,7 @@ devtools::install_github("ropensci/plotly")
 
 ## Introduction
 
-If you use [ggplot2](http://cran.r-project.org/package=ggplot2), simply call `ggplotly()` to convert your ggplot to an interactive, web-based version!
+If you use [ggplot2](http://cran.r-project.org/package=ggplot2), use `ggplotly()` to convert your ggplot to an interactive, web-based version!
 
 ```r
 library(plotly)
@@ -50,19 +50,15 @@ The `ggplotly()` function converts a ggplot object to a plotly object, so if you
 layout(gg, hovermode = "closest")
 ```
 
-## Learn more
+## Documentation
 
-* [An overview of plotly's R API](https://cdn.rawgit.com/ropensci/plotly/master/vignettes/intro.html)
-* Peruse the examples on plotly's [R homepage](https://plot.ly/r) and [ggplot2 homepage](https://plot.ly/ggplot2)
+* [An introduction to plotly's R API](https://cran.r-project.org/web/packages/plotly/vignettes/intro.html)
+* Examples and vignettes on plotly's R homepage - <https://plot.ly/r>
+* The complete figure reference guide - <https://plot.ly/r/reference>
 
 ## Contributing
 
-- We love collaboration! See the [wiki](https://github.com/ropensci/plotly/wiki/Development-guidelines) and the [code of conduct](https://github.com/ropensci/plotly/blob/master/CONDUCT.md) for more information.
-
-## Stay in touch
-
-- <[email protected]>
-- [@plotlygraphs](https://twitter.com/plotlygraphs)
+Please read through our [contributing guidelines](https://github.com/ropensci/plotly/blob/master/CONTRIBUTING.md). Included are directions for opening issues, asking questions, contributing changes to plotly, and our code of conduct. 
 
 ---