Merge branch 'release/0.3-4' into production

.Rbuildignore

@@ -2,9 +2,6 @@
 ^\.Rproj\.user$
 ^\.travis\.yml$
 .Rprofile
-inst/db
-man-roxygen
-demo/pandas
 ^\.httr-oauth$
 ^cran-comments\.md$
 ^README\.Rmd$

DESCRIPTION

@@ -1,20 +1,30 @@
 Package: tibble
 Encoding: UTF-8
-Version: 0.3-3
+Version: 0.3-4
 Title: Simple Data Frames
 Description: Provides a 'tbl_df' class that offers better checking and
-        printing capabilities than traditional data frames.
-Authors@R: c( person("Hadley", "Wickham", , "[email protected]",
-        "aut"), person("Romain", "Francois", ,
-        "[email protected]", "aut"), person("Kirill", "Müller",
-        , "[email protected]", c("aut", "cre")), person("RStudio",
-        role = "cph") )
+    printing capabilities than traditional data frames.
+Authors@R: c(
+    person("Hadley", "Wickham", , "[email protected]", "aut"),
+    person("Romain", "Francois", , "[email protected]", "aut"),
+    person("Kirill", "Müller", , "[email protected]", c("aut", "cre")),
+    person("RStudio", role = "cph")
+    )
 URL: https://github.com/krlmlr/tibble
 BugReports: https://github.com/krlmlr/tibble/issues
 Depends: R (>= 3.1.2)
-Imports: methods, assertthat, utils, lazyeval (>= 0.1.10), Rcpp
-Suggests: testthat, knitr, rmarkdown, Lahman (>= 3.0.1), magrittr,
-        microbenchmark
+Imports:
+    methods,
+    assertthat,
+    utils,
+    lazyeval (>= 0.1.10),
+    Rcpp
+Suggests:
+    testthat,
+    knitr,
+    rmarkdown,
+    Lahman (>= 3.0.1),
+    microbenchmark
 LinkingTo: Rcpp
 LazyData: yes
 License: MIT + file LICENSE

NAMESPACE

@@ -16,23 +16,19 @@ S3method(format_v,default)
 S3method(glimpse,data.frame)
 S3method(glimpse,default)
 S3method(glimpse,tbl)
-S3method(obj_type,"NULL")
-S3method(obj_type,data.frame)
-S3method(obj_type,data_frame)
-S3method(obj_type,default)
+S3method(is_vector_s3,Date)
+S3method(is_vector_s3,POSIXct)
+S3method(is_vector_s3,data.frame)
+S3method(is_vector_s3,default)
+S3method(is_vector_s3,factor)
+S3method(obj_sum,default)
 S3method(print,tbl_df)
 S3method(print,trunc_mat)
 S3method(type_sum,Date)
 S3method(type_sum,POSIXt)
-S3method(type_sum,array)
-S3method(type_sum,character)
 S3method(type_sum,data.frame)
 S3method(type_sum,default)
 S3method(type_sum,factor)
-S3method(type_sum,integer)
-S3method(type_sum,logical)
-S3method(type_sum,matrix)
-S3method(type_sum,numeric)
 export(add_row)
 export(as_data_frame)
 export(column_to_rownames)
@@ -42,9 +38,11 @@ export(dim_desc)
 export(frame_data)
 export(glimpse)
 export(has_rownames)
+export(is_vector_s3)
 export(knit_print.trunc_mat)
 export(lst)
 export(lst_)
+export(obj_sum)
 export(remove_rownames)
 export(repair_names)
 export(rownames_to_column)

NEWS.md

@@ -1,3 +1,16 @@
+Version 0.3-4 (2016-03-18)
+===
+
+- Renamed `obj_type()` to `obj_sum()`, improvements, better integration with `type_sum()`.
+- Add tests.
+- Improve documentation and vignette.
+- Internal cleanup.
+- Improve `[.tbl_df()` error message.
+- `frame_data()` returns 0-row but n-col data frame if no data.
+- Further cleanup of `repair_names()`.
+- Don't trim ws in `repair_names()` (#47).
+
+
 Version 0.3-3 (2016-03-18)
 ===
 

R/dataframe.R

@@ -1,5 +1,3 @@
-methods::setOldClass(c("tbl_df", "tbl", "data.frame"))
-
 #' Build a data frame or list.
 #'
 #' \code{data_frame} is trimmed down version of \code{\link{data.frame}} that:

R/frame-data.R

@@ -1,9 +1,12 @@
-#' Row-wise data_frame creation
+#' Row-wise tibble creation
 #'
-#' Create a row-wise \code{\link{data_frame}}.
+#' Create \code{\link{data_frame}}s laying out the data in rows, rather than
+#' in columns. This is useful for small tables of data where readability is
+#' important.
 #'
 #' @param ... Arguments specifying the structure of a \code{data_frame}.
-#'
+#'   Variable names should be formulas, and may only appear before the data.
+#' @return A \code{\link{tbl_df}}.
 #' @export
 #' @examples
 #' frame_data(
@@ -12,6 +15,14 @@
 #'   "b",   2,
 #'   "c",   3
 #' )
+#'
+#' # frame_data will create a list column if the value in each cell is
+#' # not a scalar
+#' frame_data(
+#'   ~x,  ~y,
+#'   "a", 1:3,
+#'   "b", 4:6
+#' )
 frame_data <- function(...) {
 
   dots <- list(...)
@@ -21,7 +32,9 @@ frame_data <- function(...) {
   i <- 1
   while (TRUE) {
     if (i > length(dots)) {
-      return(data_frame())
+      out <- rep(list(logical()), length(frame_names))
+      names(out) <- frame_names
+      return(as_data_frame(out))
     }
 
     el <- dots[[i]]
@@ -32,12 +45,14 @@ frame_data <- function(...) {
       break
 
     if (length(el) != 2) {
-      stop("expected a column name with a single argument; e.g. '~ name'")
+      stop("expected a column name with a single argument; e.g. '~ name'",
+        call. = FALSE)
     }
 
     candidate <- el[[2]]
     if (!(is.symbol(candidate) || is.character(candidate))) {
-        stop("expected a symbol or string denoting a column name")
+      stop("expected a symbol or string denoting a column name",
+        call. = FALSE)
     }
 
     frame_names <- c(frame_names, as.character(el[[2]]))
@@ -46,7 +61,7 @@ frame_data <- function(...) {
   }
 
   if (!length(frame_names)) {
-    stop("no column names detected in 'frame_data()' call")
+    stop("no column names detected in 'frame_data()' call", call. = FALSE)
   }
 
   frame_rest <- dots[i:length(dots)]
@@ -57,11 +72,14 @@ frame_data <- function(...) {
   # structure.
   frame_ncol <- length(frame_names)
   if (n_elements %% frame_ncol != 0) {
-    stop(sprintf(
-      "invalid 'frame_data()' specification: had %s elements and %s columns",
-      n_elements,
-      frame_ncol
-    ))
+    stop(
+      sprintf(
+        "invalid 'frame_data()' specification: had %s elements and %s columns",
+        n_elements,
+        frame_ncol
+      ),
+      call. = FALSE
+    )
   }
 
   frame_mat <- matrix(frame_rest, ncol = frame_ncol, byrow = TRUE)

R/repair-names.R

@@ -1,12 +1,12 @@
 #' Repair object names.
 #'
 #' \code{repair_names} ensures its input has non-missing and
-#' unique names. It also strips any leading or trailing spaces.
-#' Valid names are left as is.
+#' unique names (duplicated names get a numeric suffix). Valid names are
+#' left as is.
 #'
 #' @param x A named vector.
 #' @param prefix A string, the prefix to use for new column names.
-#' @param sep A string, inserted between the column name and de-duplicating
+#' @param sep A string inserted between the column name and de-duplicating
 #'    number.
 #' @return \code{x} with valid names.
 #' @export
@@ -16,33 +16,24 @@
 #' tbl <- as_data_frame(structure(list(3, 4, 5), class = "data.frame"))
 #' repair_names(tbl)
 repair_names <- function(x, prefix = "V", sep = "") {
-  if (length(x) == 0)
+  if (length(x) == 0) {
+    names(x) <- character()
     return(x)
+  }
 
-  xnames <- init_names(x)
-  blanks <- xnames == ""
+  new_names <- make_unique(names2(x), prefix = prefix, sep = sep)
+  setNames(x, new_names)
+}
 
-  # The order vector defines the order in which make.unique() should process the
-  # entries.  Blanks are initialized with the prefix. The index of the first
-  # blank entry appears twice in this vector if there's no column named like the
-  # prefix, to make sure that blank columns always start with V1 (or a higher
-  # index if appropriate).  See also the "pathological cases" test.
-  order <- c(
-    which(!blanks),
-    if (all(xnames[!blanks] != prefix) && any(blanks))
-      which.max(blanks),
-    which(blanks))
-  xnames[blanks] <- prefix
-  xnames[order] <- make.unique(xnames[order], sep = sep)
+make_unique <- function(x, prefix = "V", sep = "") {
+  blank <- x == ""
 
-  names(x) <- xnames
-  x
-}
+  # Ensure existing names are unique
+  x[!blank] <- make.unique(x[!blank], sep = sep)
 
-init_names <- function(x) {
-  xnames <- names(x)
-  if (is.null(xnames))
-    rep("", length(x))
-  else
-    ifelse(is.na(xnames), "", trim_ws(xnames))
+  # Replace blank names
+  new_vars <- setdiff(paste(prefix, seq_along(x), sep = sep), x)
+  x[blank] <- new_vars[seq_len(sum(blank))]
+
+  x
 }

R/rownames.R

@@ -1,39 +1,45 @@
-#' Row names
+#' Tools for working with row names
 #'
-#' \code{has_rownames} checks if a data frame has row names.
-#' @param df Input data frame
-#' @export
-#' @rdname rownames
+#' Generally, it is best to avoid row names, because they are basically a
+#' character column with different semantics to every other column. These
+#' functions allow to you detect if a data frame has row names
+#' (\code{has_rownames}), remove them (\code{remove_rownames}), or convert
+#' them back-and-forth between an explicit column (\code{rownames_to_column},
+#' and \code{column_to_rownames}).
+#'
+#' @param df A data frame
+#' @param var Name of column to use for rownames.
 #' @examples
 #' has_rownames(mtcars)
 #' has_rownames(iris)
+#' has_rownames(remove_rownames(mtcars))
+#'
+#' head(rownames_to_column(mtcars))
+#'
+#' mtcars_tbl <- rownames_to_column(tbl_df(mtcars))
+#' mtcars_tbl
+#' column_to_rownames(mtcars_tbl)
+#' @name rownames
+NULL
+
+
+#' @export
+#' @rdname rownames
 has_rownames <- function(df) {
   stopifnot(is.data.frame(df))
   .row_names_info(df) > 0L
 }
 
-#' \code{remove_rownames} removes all row names.
 #' @export
 #' @rdname rownames
-#' @examples
-#' rownames(remove_rownames(mtcars))
 remove_rownames <- function(df) {
   stopifnot(is.data.frame(df))
   rownames(df) <- NULL
   df
 }
 
-#' \code{rownames_to_column} convert row names to an explicit variable.
-#'
-#' @param var Name of variable to use
 #' @export
 #' @rdname rownames
-#' @importFrom stats setNames
-#' @examples
-#' rownames_to_column(mtcars)
-#'
-#' mtcars_tbl <- rownames_to_column(tbl_df(mtcars))
-#' mtcars_tbl
 rownames_to_column <- function(df, var = "rowname") {
   stopifnot(is.data.frame(df))
 
@@ -53,14 +59,8 @@ rownames_to_column <- function(df, var = "rowname") {
   new_df
 }
 
-#' \code{column_to_rownames} convert a column variable to row names. This is an
-#' inverted operation of \code{rownames_to_column}.
-#'
 #' @rdname rownames
 #' @export
-#' @examples
-#'
-#' column_to_rownames(mtcars_tbl)
 column_to_rownames <- function(df, var = "rowname") {
   stopifnot(is.data.frame(df))