R CMD check fixes (#22)

* wrap slow example in dontrun

* fix Data masking and tidy selection NOTEs

* fix false positive warning about pkg startup message

* move pandoc to Suggests to remove 'too many deps' Note

Author: t-kalinowski

DESCRIPTION

@@ -30,15 +30,15 @@ Imports:
     xml2,
     methods,
     S7,
-    reticulate (>= 1.41.0.9001),
-    pandoc,
+    reticulate (>= 1.42.0),
     commonmark,
     blob,
     cli,
     curl,
     withr,
     dotty
 Suggests:
+    pandoc,
     ellmer,
     knitr,
     readr,

R/ragnar-package.R

@@ -31,6 +31,7 @@ dotty::.
     file <- tempfile(fileext = ".txt")
     on.exit(unlink(file))
     writeLines("hi", file)
+    (writeLines)("hi", file)
     convert <- init_markitdown()$convert
     DocumentConverterResult <- convert(file)
     if (inherits(DocumentConverterResult, "python.builtin.object")) {

R/read-html.R

@@ -325,6 +325,7 @@ ragnar_read_document <- function(x, ...,
 #' @export
 #'
 #' @examples
+#' \dontrun{
 #' ragnar_find_links("https://r4ds.hadley.nz/base-R.html")
 #' ragnar_find_links("https://ellmer.tidyverse.org/")
 #' ragnar_find_links("https://ellmer.tidyverse.org/", depth = 2)
@@ -335,6 +336,7 @@ ragnar_read_document <- function(x, ...,
 #'   children_only = "https://github.com/Snowflake-Labs/sfquickstarts",
 #'   depth = 1
 #' )
+#' }
 ragnar_find_links <- function(x, depth = 0L, children_only = TRUE, progress = TRUE, ...,
                               url_filter = identity) {
 

R/read-markdown.R

@@ -272,6 +272,7 @@ markdown_segment_text <- function(text, split_by = c("h1", "h2", "h3", "pre", "p
   ## TOOD: probably better to use commonmark instead of pandoc here.
   tmp_html <- tempfile(fileext = ".html")
   on.exit(unlink(tmp_html))
+  check_installed("pandoc")
   pandoc::pandoc_convert(text = text, to = "html", output = tmp_html)
   html_text3(doc = read_html(tmp_html, encoding = "UTF-8"),
              split_tags = split_by)
@@ -403,9 +404,11 @@ ragnar_read <- function(x, ..., split_by_tags = NULL, frame_by_tags = NULL) {
     frame[["tag"]] <- NULL
   }
 
-  frame <- frame |>
-    dplyr::mutate(origin = x, hash = hash) |>
-    dplyr::select(origin, hash, text, dplyr::everything())
+  frame[["origin"]] <- x
+  frame[["hash"]] <- hash
+
+  # reorder columns
+  frame <- frame[unique(c("origin", "hash", names(frame), "text"))]
 
   as_tibble(frame)
 }

R/retrieve.R

@@ -150,8 +150,8 @@ ragnar_retrieve_vss_and_bm25 <- function(store, text, top_k = 3, ...) {
   # pivot to wide format
   out <- tidyr::pivot_wider(
     out,
-    names_from = metric_name,
-    values_from = metric_value
+    names_from = "metric_name",
+    values_from = "metric_value"
   )
 
   # TODO: come up with a nice reordering that doesn't involve too much compute.

R/store.R

@@ -267,7 +267,7 @@ ragnar_store_update <- function(store, chunks) {
     DBI::dbWriteTable(
       store@.con,
       "tmp_chunks",
-      chunks |> dplyr::select(origin, hash) |> dplyr::distinct(),
+      chunks |> dplyr::select("origin", "hash") |> dplyr::distinct(),
       temporary = TRUE,
       overwrite = TRUE
     )