Prefer vec_split() over split()

Author: krlmlr

NAMESPACE

@@ -54,3 +54,4 @@ importFrom(rlang,set_names)
 importFrom(rlang,warn)
 importFrom(vctrs,vec_rbind)
 importFrom(vctrs,vec_slice)
+importFrom(vctrs,vec_split)

R/compat-tidyr.R

@@ -3,8 +3,8 @@ nest_ <- function(data, key_col, nest_cols = character()) {
   key_data <- data[[key_column]]
   key_levels <- unique(key_data)
   key_factor <- factor(key_data, levels = key_levels)
-  res <- list()
-  res[[key_column]] <- key_levels
-  res[[key_col]] <- split(data[, nest_cols], key_factor)
-  new_styler_df(res)
+
+  res <- vec_split(data[, nest_cols], key_factor)
+  names(res) <- c(key_column, key_col)
+  res
 }

R/detect-alignment.R

@@ -46,7 +46,12 @@ token_is_on_aligned_line <- function(pd_flat) {
   # pos_id too expensive to construct in alignment_ensure_trailing_comma()
   pd_flat$lag_newlines <- pd_flat$pos_id <- NULL
   pd_flat$.lag_spaces <- lag(pd_flat$spaces)
-  pd_by_line <- split(pd_flat, line_idx)
+  pd_by_line_split <- vec_split(pd_flat, line_idx)
+
+  # FIXME: Why are we using names here?
+  pd_by_line <- pd_by_line_split[[2L]]
+  names(pd_by_line) <- as.character(pd_by_line_split[[1L]])
+
   pd_by_line[purrr::map_lgl(pd_by_line, ~ any(.x$stylerignore))] <- NULL
   if (length(pd_by_line) < 1L) {
     return(TRUE)

R/nest.R

@@ -100,8 +100,8 @@ shallowify <- function(pd) {
   if (cache_is_activated()) {
     order <- order(pd$line1, pd$col1, -pd$line2, -pd$col2, as.integer(pd$terminal))
     pd_parent_first <- vec_slice(pd, order)
-    pos_ids_to_keep <- pd_parent_first %>%
-      split(cumsum(pd_parent_first$parent == 0L)) %>%
+    pd_parent_first_split <- vec_split(pd_parent_first, cumsum(pd_parent_first$parent == 0L))
+    pos_ids_to_keep <- pd_parent_first_split[[2L]] %>%
       map(find_pos_id_to_keep) %>%
       unlist(use.names = FALSE)
     shallow <- vec_slice(pd, pd$pos_id %in% pos_ids_to_keep)
@@ -335,10 +335,9 @@ nest_parse_data <- function(pd_flat) {
       return(pd_flat)
     }
     pd_flat$internal <- with(pd_flat, (id %in% parent) | (parent <= 0L))
-    split_data <- split(pd_flat, pd_flat$internal)
 
-    child <- split_data$`FALSE`
-    internal <- split_data$`TRUE`
+    child <- vec_slice(pd_flat, !pd_flat$internal)
+    internal <- vec_slice(pd_flat, pd_flat$internal)
 
     internal$internal_child <- internal$child
     internal$child <- NULL

R/relevel.R

@@ -178,8 +178,8 @@ relocate_eq_assign_nest <- function(pd) {
   idx_eq_assign <- which(pd$token == "EQ_ASSIGN")
   if (length(idx_eq_assign) > 0L) {
     block_id <- find_block_id(pd)
-    blocks <- split(pd, block_id)
-    pd <- map_dfr(blocks, relocate_eq_assign_one)
+    blocks <- vec_split(pd, block_id)
+    pd <- map_dfr(blocks[[2L]], relocate_eq_assign_one)
   }
   pd
 }

R/roxygen-examples.R

@@ -8,9 +8,9 @@
 #' @inheritSection parse_transform_serialize_roxygen Hierarchy
 #' @keywords internal
 style_roxygen_code_example <- function(example, transformers, base_indention) {
-  example <- split(example, cumsum(grepl("^#' *@examples", example)))
+  example <- vec_split(example, cumsum(grepl("^#' *@examples", example)))
   purrr::map(
-    example, style_roxygen_code_example_one,
+    example[[2L]], style_roxygen_code_example_one,
     transformers = transformers, base_indention = base_indention
   ) %>%
     flatten_chr()
@@ -28,8 +28,8 @@ style_roxygen_code_example_one <- function(example_one,
   example_one <- example_one[example_one != ""]
 
   bare <- parse_roxygen(example_one)
-  one_dont <- split(bare$text, factor(cumsum(bare$text %in% dont_keywords())))
-  unmasked <- map(one_dont, style_roxygen_code_example_segment,
+  one_dont <- vec_split(bare$text, factor(cumsum(bare$text %in% dont_keywords())))
+  unmasked <- map(one_dont[[2L]], style_roxygen_code_example_segment,
     transformers = transformers,
     base_indention = base_indention
   ) %>%

R/styler-package.R

@@ -26,6 +26,7 @@
 #' @importFrom rlang abort warn seq2 is_installed "%||%" set_names
 #' @importFrom vctrs vec_rbind
 #' @importFrom vctrs vec_slice
+#' @importFrom vctrs vec_split
 ## usethis namespace: end
 NULL
 

R/stylerignore.R

@@ -24,9 +24,12 @@ env_add_stylerignore <- function(pd_flat) {
     pd_flat_temp$stylerignore,
     default = pd_flat_temp$stylerignore[1L]
   )
-  pd_flat_temp$first_pos_id_in_segment <- split(
+
+  pos_id_split <- vec_split(
     pd_flat_temp$pos_id, cumsum(is_stylerignore_switchpoint)
-  ) %>%
+  )
+
+  pd_flat_temp$first_pos_id_in_segment <- pos_id_split[[2L]] %>%
     map(~ rep(.x[1L], length(.x))) %>%
     unlist(use.names = FALSE)
   pd_flat_temp$lag_newlines <- pd_flat_temp$lag_newlines

R/transform-files.R

@@ -216,7 +216,7 @@ split_roxygen_segments <- function(text, roxygen_examples) {
   all_lines <- seq2(1L, length(text))
   active_segment <- as.integer(all_lines %in% roxygen_examples)
   segment_id <- cumsum(abs(c(0L, diff(active_segment)))) + 1L
-  separated <- split(text, factor(segment_id))
+  separated <- vec_split(text, factor(segment_id))[[2L]]
   restyle_selector <- if (roxygen_examples[1L] == 1L) {
     odd_index
   } else {
@@ -260,7 +260,7 @@ parse_transform_serialize_r <- function(text,
   )
 
   strict <- transformers$more_specs_style_guide$strict %||% TRUE
-  pd_split <- unname(split(pd_nested, pd_nested$block))
+  pd_split <- vec_split(pd_nested, pd_nested$block)[[2L]]
   pd_blank <- find_blank_lines_to_next_block(pd_nested)
 
   text_out <- vector("list", length(pd_split))

R/visit.R

@@ -231,8 +231,8 @@ enrich_terminals <- function(flattened_pd, use_raw_indention = FALSE) {
   flattened_pd$newlines <- lead(flattened_pd$lag_newlines, default = 0L)
   flattened_pd$nchar <- nchar(flattened_pd$text, type = "width")
   groups <- flattened_pd$line1
-  flattened_pd <- flattened_pd %>%
-    split(groups) %>%
+  split_pd <- vec_split(flattened_pd, groups)[[2L]]
+  flattened_pd <- split_pd %>%
     map_dfr(function(.x) {
       .x$col2 <- cumsum(.x$nchar + .x$lag_spaces)
       .x