plumber() and Plumber$new() will read source file as UTF-8 encoded

NEWS.md

@@ -2,11 +2,14 @@ plumber 0.4.7
 --------------------------------------------------------------------------------
 * Add support for swagger for mounted routers (@bradleyhd, [#274](https://github.com/trestletech/plumber/issues/274)).
 * BUGFIX: A multiline POST body is now collapsed to a single line ([#270](https://github.com/trestletech/plumber/issues/270)).
+* The source files used in plumber must use the UTF-8 encoding if they contain
+non-ASCII characters (@shrektan, [#312](https://github.com/trestletech/plumber/pull/312)).
 * SECURITY: Wrap `jsonlite::fromJSON` to ensure that `jsonlite` never reads
   input as a remote address (such as a file path or URL) and attempts to parse
   that. The only known way to exploit this behavior in plumber unless an
   API were using encrypted cookies and an attacker knew the encryption key in
-  order to craft arbitrary cookies.
+  order to craft arbitrary cookies. ([#325](https://github.com/trestletech/plumber/pull/325))
+
 
 plumber 0.4.6
 --------------------------------------------------------------------------------

R/plumber.R

@@ -2,6 +2,11 @@
 #' @import stringi
 NULL
 
+# Hard code UTF-8 file encoding
+# Removes encoding headache at minor cost of setting encoding in editor
+# https://github.com/trestletech/plumber/pull/312
+utf8Encoding <- "UTF-8"
+
 # used to identify annotation flags.
 verbs <- c("GET", "PUT", "POST", "DELETE", "HEAD", "OPTIONS", "PATCH")
 enumerateVerbs <- function(v){
@@ -50,7 +55,7 @@ plumb <- function(file, dir="."){
     on.exit(setwd(old))
 
     # Expect that entrypoint will provide us with the router
-    x <- source(entrypoint)
+    x <- source(entrypoint, encoding=utf8Encoding)
 
     # source returns a list with value and visible elements, we want the (visible) value object.
     pr <- x$value
@@ -177,10 +182,19 @@ plumber <- R6Class(
       private$notFoundHandler <- default404Handler
 
       if (!is.null(file)){
-        private$lines <- readLines(file)
-        private$parsed <- parse(file, keep.source=TRUE)
-
-        source(file, local=private$envir, echo=FALSE, keep.source=TRUE)
+        con <- file(file, encoding=utf8Encoding)
+        on.exit(close(con), add=TRUE)
+
+        # Read lines directly
+        private$lines <- readLines(con)
+        # "...produces an object of the descendant class ‘srcfilecopy’, 
+        #   which saves the source lines in a character vector" (?srcfilecopy)
+        srcfile <- srcfilecopy(file, private$lines, isFile=TRUE)
+        # "When ‘keep.source’ is ‘TRUE’, if ‘text’ is used, 
+        #   ‘srcfile’ will be set to a ‘srcfilecopy’ containing the text" (?parse)
+        private$parsed <- parse(text=private$lines, srcfile=srcfile, keep.source=TRUE)
+
+        source(con, local=private$envir, echo=FALSE, keep.source=TRUE)
 
         for (i in 1:length(private$parsed)){
           e <- private$parsed[i]

tests/testthat/files/multibytes.R

@@ -0,0 +1,4 @@
+#* @get /echo
+function() {
+  "中文消息"
+}

tests/testthat/test-multibytes.R

@@ -0,0 +1,12 @@
+context("multibytes source file")
+
+test_that("support files with multibytes", {
+  # on Windows, the default encoding is not UTF-8. So, plumber has to
+  # tell R to use UTF-8 encoding when reading the source file.
+  r <- plumber$new("files/multibytes.R")
+  req <- make_req("GET", "/echo")
+  res <- PlumberResponse$new()
+  out <- r$serve(req, res)$body
+  expect_identical(out, jsonlite::toJSON("\u4e2d\u6587\u6d88\u606f"))
+  expect_equal(Encoding(out), utf8Encoding)
+})