Simpler parsing algorthim

Fixes #189

DESCRIPTION

@@ -14,7 +14,7 @@ License: MIT + file LICENSE
 URL: https://downlit.r-lib.org/, https://github.com/r-lib/downlit
 BugReports: https://github.com/r-lib/downlit/issues
 Depends: 
-    R (>= 3.6)
+    R (>= 4.0.0)
 Imports: 
     brio,
     desc,
@@ -40,4 +40,4 @@ Config/Needs/website: tidyverse/tidytemplate
 Config/testthat/edition: 3
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)
-RoxygenNote: 7.2.3
+RoxygenNote: 7.3.1

NEWS.md

@@ -1,5 +1,7 @@
 # downlit (development version)
 
+* Use simpler parsing algorithm for R 4.0, which avoids crash with certain UTF-8 characters (#189).
+
 # downlit 0.4.3
 
 * Fix for upcoming R-devel (#169).

R/utils.R

@@ -36,13 +36,8 @@ safe_parse <- function(text, standardise = TRUE) {
   lines <- strsplit(text, "\n", fixed = TRUE, useBytes = TRUE)[[1]]
   srcfile <- srcfilecopy("test.r", lines)
 
-  # https://github.com/gaborcsardi/rencfaq#how-to-parse-utf-8-text-into-utf-8-code
-  Encoding(text) <- "unknown"
-  con <- textConnection(text)
-  on.exit(close(con), add = TRUE)
-
   tryCatch(
-    parse(con, keep.source = TRUE, encoding = "UTF-8", srcfile = srcfile),
+    parse(text = text, keep.source = TRUE, encoding = "UTF-8", srcfile = srcfile),
     error = function(e) NULL
   )
 }

tests/testthat/test-utils.R

@@ -6,3 +6,7 @@ test_that("converts Latin1 encoded text to utf8", {
   expect_equal(Encoding(y), "UTF-8")
   expect_equal(y, "\u00fc")
 })
+
+test_that("doesn't crash on utf-8 characters", {
+  expect_equal(safe_parse("×"), NULL)
+})