Better strategy for replacing text

Fixes #34

DESCRIPTION

@@ -17,6 +17,7 @@ Imports:
     brio,
     fansi,
     rlang,
+    vctrs,
     yaml
 Suggests:
     rmarkdown,

R/highlight.R

@@ -18,7 +18,8 @@
 #' @examples
 #' cat(highlight("1 + 1"))
 highlight <- function(text, classes = classes_chroma(), pre_class = NULL) {
-  text <- gsub("\t", "  ", text)
+  text <- gsub("\t", "  ", text, fixed = TRUE)
+  text <- gsub("\r", "", text, fixed = TRUE)
   parsed <- parse_data(text)
   if (is.null(parsed)) {
     return(NA_character_)
@@ -39,21 +40,13 @@ highlight <- function(text, classes = classes_chroma(), pre_class = NULL) {
   # Update input - basic idea from prettycode
   changed <- !is.na(out$href) | !is.na(out$class) | out$text != out$escaped
   changes <- out[changed, , drop = FALSE]
-  changes_by_line <- split(changes, changes$line1)
-  lines <- strsplit(text, "\r?\n")[[1]]
-
-  for (change in changes_by_line) {
-    i <- change$line1[[1]]
-    new <- style_token(change$escaped, change$href, change$class)
-
-    lines[[i]] <- replace_in_place(
-      lines[[i]],
-      start = change$col1,
-      end = change$col2,
-      replacement = new
-    )
-  }
-  out <- paste0(lines, collapse = "\n")
+
+  loc <- line_col(text)
+  start <- vctrs::vec_match(data.frame(line = changes$line1, col = changes$col1), loc)
+  end <- vctrs::vec_match(data.frame(line = changes$line2, col = changes$col2), loc)
+
+  new <- style_token(changes$escaped, changes$href, changes$class)
+  out <- replace_in_place(text, start, end, replacement = new)
 
   if (is.null(pre_class)) {
     return(out)
@@ -88,6 +81,16 @@ replace_in_place <- function(str, start, end, replacement) {
   paste0(pieces, collapse = "")
 }
 
+line_col <- function(x) {
+  char <- strsplit(x, "")[[1]]
+
+  nl <- char == "\n"
+  line <- cumsum(c(TRUE, nl[-length(char)]))
+  col <- sequence(rle(line)$lengths)
+
+  data.frame(line, col)
+}
+
 parse_data <- function(text) {
   stopifnot(is.character(text), length(text) == 1)
 

tests/testthat/test-downlit-html.txt

@@ -9,6 +9,7 @@
 <span class="co"># This is a comment</span>
 
 <span class="kw">stats</span>::<span class="fu"><a href="https://rdrr.io/r/stats/median.html">median</a></span>()
+
 </pre>
 <p><code><a href="https://rdrr.io/r/stats/median.html">stats::median()</a></code></p>
 

tests/testthat/test-downlit-md.txt

@@ -5,6 +5,7 @@
 > cat(downlit_md_string("```\nbase::t(1)\n```"))
 <pre class='chroma'>
 <span class='k'>base</span>::<span class='nf'><a href='https://rdrr.io/r/base/t.html'>t</a></span>(<span class='m'>1</span>)
+
 </pre>
 
 > # No transforms

tests/testthat/test-highlight.R

@@ -45,6 +45,11 @@ test_that("can parse code with carriage returns", {
   expect_equal(lines[[2]], "<span class='m'>2</span>")
 })
 
+test_that("syntax can span multiple lines", {
+  expect_equal(highlight("f(\n\n)"), "<span class='nf'>f</span>(\n\n)")
+  expect_equal(highlight("'\n\n'"), "<span class='s'>'\n\n'</span>")
+})
+
 test_that("unparsable code returns NULL", {
   expect_equal(highlight("<"), NA_character_)
   # but pure comments still highlighted