pre-CRAN flight check

DESCRIPTION

@@ -25,6 +25,7 @@ Depends:
 Imports:
     Rcpp,
     data.table,
+    dplyr,
     dtplyr
 LinkingTo: Rcpp
 RoxygenNote: 6.0.1

NAMESPACE

@@ -5,5 +5,6 @@ export(stream_in)
 export(validate)
 importFrom(Rcpp,sourceCpp)
 importFrom(data.table,rbindlist)
+importFrom(dplyr,tbl_df)
 importFrom(dtplyr,tbl_dt)
 useDynLib(ndjson, .registration=TRUE)

NEWS.md

@@ -1,6 +1,7 @@
 0.5.0
 =====================
 * Updated core ndjson file to take care of some buffer overflow vulns
+* Optionally returns a `tbl_df` vs a `tbl_dt` (data.table is used for speed on list rbind)
 * Fixed CRAN check errors
 
 0.4.0

R/ndjson-package.R

@@ -7,4 +7,5 @@
 #' @importFrom Rcpp sourceCpp
 #' @importFrom data.table rbindlist
 #' @importFrom dtplyr tbl_dt
+#' @importFrom dplyr tbl_df
 NULL

R/ndjson.R

@@ -3,8 +3,12 @@
 #' Given a file of streaming JSON (ndjson) this function reads in the records
 #' and creates a flat \code{data.table} / \code{tbl_dt} from it.
 #'
+#' @md
 #' @param path path to file (supports "\code{gz}" files)
-#' @return \code{tbl_dt}
+#' @param cls the package uses \code{data.table::rbindlist} for speed but
+#'        that's not always the best return type for everyone, so you have
+#'        option of keeping it a `tbl_dt` via "`dt`" or converting it to a `tbl`
+#' @return \code{tbl_dt} or \code{tbl} or \{data.frame}
 #' @export
 #' @references \url{http://ndjson.org/}
 #' @examples
@@ -13,9 +17,11 @@
 #'
 #' gzf <- system.file("extdata", "testgz.json.gz", package="ndjson")
 #' nrow(stream_in(gzf))
-stream_in <- function(path) {
+stream_in <- function(path, cls = c("dt", "tbl")) {
+  cls <- match.arg(cls, c("dt", "tbl"))
   tmp <- .Call('ndjson_internal_stream_in', path.expand(path), PACKAGE='ndjson')
-  dtplyr::tbl_dt(data.table::rbindlist(tmp, fill=TRUE))
+  tmp <- dtplyr::tbl_dt(data.table::rbindlist(tmp, fill=TRUE))
+  if (cls == "tbl") dplyr::tbl_df(tmp) else tmp
 }
 
 #' Validate ndjson file
@@ -42,12 +48,18 @@ validate <- function(path, verbose=FALSE) {
 
 #' Flatten a character vector of individual JSON lines into a \code{tbl_dt}
 #'
+#' @md
 #' @param x character vector of individual JSON lines to flatten
-#' @return \code{tbl_dt}
+#' @param cls the package uses \code{data.table::rbindlist} for speed but
+#'        that's not always the best return type for everyone, so you have
+#'        option of keeping it a `tbl_dt` via "`dt`" or converting it to a `tbl`
+#' @return \code{tbl_dt} or \code{tbl} or \{data.frame}
 #' @export
 #' @examples
 #' flatten('{"top":{"next":{"final":1,"end":true},"another":"yes"},"more":"no"}')
-flatten <- function(x) {
+flatten <- function(x, cls = c("dt", "tbl")) {
+  cls <- match.arg(cls, c("dt", "tbl"))
   tmp <- .Call('ndjson_internal_flatten', x, PACKAGE='ndjson')
-  dtplyr::tbl_dt(data.table::rbindlist(tmp, fill=TRUE))
+  tmp <- dtplyr::tbl_dt(data.table::rbindlist(tmp, fill=TRUE))
+  if (cls == "tbl") dplyr::tbl_df(tmp) else tmp
 }

cran-comments.md

@@ -1,7 +1,7 @@
 ## Test environments
 * local OS X install, R 3.4.0 (clang)
 * local ubuntu 16.04, R 3.4.0 (g++-7)
-* win-builder (devel and release)
+* win-builder (devel and release) https://win-builder.r-project.org/P5u60xh8LPrA/
 
 ## R CMD check results
 
@@ -16,3 +16,4 @@ None
 ## General notes
 
 - Fixed CRAN gcc 7 toolchain issues (linux)
+- Update core C++ library the pkg uses

man/flatten.Rd

@@ -4,13 +4,17 @@
 \alias{flatten}
 \title{Flatten a character vector of individual JSON lines into a \code{tbl_dt}}
 \usage{
-flatten(x)
+flatten(x, cls = c("dt", "tbl"))
 }
 \arguments{
 \item{x}{character vector of individual JSON lines to flatten}
+
+\item{cls}{the package uses \code{data.table::rbindlist} for speed but
+that's not always the best return type for everyone, so you have
+option of keeping it a \code{tbl_dt} via "\code{dt}" or converting it to a \code{tbl}}
 }
 \value{
-\code{tbl_dt}
+\code{tbl_dt} or \code{tbl} or {data.frame}
 }
 \description{
 Flatten a character vector of individual JSON lines into a \code{tbl_dt}

man/stream_in.Rd

@@ -4,13 +4,17 @@
 \alias{stream_in}
 \title{Stream in & flatten an ndjson file into a \code{tbl_dt}}
 \usage{
-stream_in(path)
+stream_in(path, cls = c("dt", "tbl"))
 }
 \arguments{
 \item{path}{path to file (supports "\code{gz}" files)}
+
+\item{cls}{the package uses \code{data.table::rbindlist} for speed but
+that's not always the best return type for everyone, so you have
+option of keeping it a \code{tbl_dt} via "\code{dt}" or converting it to a \code{tbl}}
 }
 \value{
-\code{tbl_dt}
+\code{tbl_dt} or \code{tbl} or {data.frame}
 }
 \description{
 Given a file of streaming JSON (ndjson) this function reads in the records

tests/testthat/test-ndjson.R

@@ -4,9 +4,13 @@ test_that("we can do something", {
   f <- system.file("extdata", "test.json", package="ndjson")
   expect_that(validate(f), equals(TRUE))
   expect_that(nrow(stream_in(f)), equals(100))
+  expect_that(stream_in(f), is_a("tbl_dt"))
+  expect_that(stream_in(f, "tbl"), is_a("tbl_df"))
 
   gzf <- system.file("extdata", "testgz.json.gz", package="ndjson")
   expect_that(validate(gzf), equals(TRUE))
   expect_that(nrow(stream_in(gzf)), equals(100))
+  expect_that(stream_in(gzf), is_a("tbl_dt"))
+  expect_that(stream_in(gzf, "tbl"), is_a("tbl_df"))
 
 })