Add row limit setting of data viewer and support Apache Arrow Table (#945)

* add row limit setting of data viewer

* add setting to convert Apache Arrow Table to a data frame and display it in the data viewer

* fix

* remove the show_arrow_table option

* refactoring: functionize the processing to trunc data

* add the overall number of rows to the title

Author: eitsupi

R/session/vsc.R

@@ -31,6 +31,7 @@ load_settings <- function() {
     vsc.browser = setting(session$viewers$viewColumn$browser, Disable = FALSE),
     vsc.viewer = setting(session$viewers$viewColumn$viewer, Disable = FALSE),
     vsc.page_viewer = setting(session$viewers$viewColumn$pageViewer, Disable = FALSE),
+    vsc.row_limit = session$data$rowLimit,
     vsc.view = setting(session$viewers$viewColumn$view, Disable = FALSE),
     vsc.helpPanel = setting(session$viewers$viewColumn$helpPanel, Disable = FALSE)
   ))
@@ -346,7 +347,7 @@ if (show_view) {
     }
 
     if (is.data.frame(data)) {
-      nrow <- nrow(data)
+      .nrow <- nrow(data)
       colnames <- colnames(data)
       if (is.null(colnames)) {
         colnames <- sprintf("V%d", seq_len(ncol(data)))
@@ -357,14 +358,14 @@ if (show_view) {
         rownames <- rownames(data)
         rownames(data) <- NULL
       } else {
-        rownames <- seq_len(nrow)
+        rownames <- seq_len(.nrow)
       }
       colnames <- c("(row)", colnames)
       fields <- sprintf("x%d", seq_along(colnames))
       data <- c(list(" " = rownames), .subset(data))
       names(data) <- fields
       class(data) <- "data.frame"
-      attr(data, "row.names") <- .set_row_names(nrow)
+      attr(data, "row.names") <- .set_row_names(.nrow)
       columns <- .mapply(get_column_def,
         list(colnames, fields, data),
         NULL
@@ -379,11 +380,25 @@ if (show_view) {
   }
 
   show_dataview <- function(x, title, uuid = NULL,
-                            viewer = getOption("vsc.view", "Two")) {
+                            viewer = getOption("vsc.view", "Two"),
+                            row_limit = abs(getOption("vsc.row_limit", 0))) {
+    as_truncated_data <- function(.data) {
+      .nrow <- nrow(.data)
+      if (row_limit != 0 && row_limit < .nrow) {
+        title <<- sprintf("%s (limited to %d/%d)", title, row_limit, .nrow)
+        .data <- utils::head(.data, n = row_limit)
+      }
+      return(.data)
+    }
+
     if (missing(title)) {
       sub <- substitute(x)
       title <- deparse(sub, nlines = 1)
     }
+    if (inherits(x, "ArrowTabular")) {
+      x <- as_truncated_data(x)
+      x <- as.data.frame(x)
+    }
     if (is.environment(x)) {
       all_names <- ls(x)
       is_promise <- rlang::env_binding_are_lazy(x, all_names)
@@ -438,6 +453,7 @@ if (show_view) {
       }
     }
     if (is.data.frame(x) || is.matrix(x)) {
+      x <- as_truncated_data(x)
       data <- dataview_table(x)
       file <- tempfile(tmpdir = tempdir, fileext = ".json")
       jsonlite::write_json(data, file, na = "string", null = "null", auto_unbox = TRUE)

package.json

@@ -1448,6 +1448,11 @@
           "default": true,
           "markdownDescription": "Emulate the RStudio API for addin support and other {rstudioapi} calls. Changes the option `vsc.rstudioapi` in R. Requires `#r.sessionWatcher#` to be set to `true`."
         },
+        "r.session.data.rowLimit": {
+          "type": "integer",
+          "default": 0,
+          "markdownDescription": "The maximum number of rows to be displayed in the data viewer. `0` means no limit. Changes the option `vsc.row_limit` in R. Requires `#r.sessionWatcher#` to be set to `true`."
+        },
         "r.session.viewers.viewColumn": {
           "type": "object",
           "markdownDescription": "Which view column should R-related webviews be displayed? Requires `#r.sessionWatcher#` to be set to `true`.",