bit_string.{to_string, is_utf8}

Author: lpil

CHANGELOG.md

@@ -3,7 +3,7 @@
 ## Unreleased
 
 - `bit_string` module created with `from_string`, `byte_size`, `append`,
-  `part`, `int_to_u32` and `int_from_u32`.
+  `part`, `to_string`, `is_utf8`, `int_to_u32` and `int_from_u32` functions.
 - The `bit_builder` module has been introduced with `prepend`, `append`,
   `prepend_builder`, `append_builder`, `prepend_string`, `append_string`,
   `concat`, `from_bit_string`, `to_bit_string`, and `byte_size` functions.

src/gleam/bit_string.gleam

@@ -5,7 +5,7 @@
 pub type BitString =
   BitString
 
-/// Convert a utf8 String type into a raw BitString type.
+/// Convert a UTF-8 String type into a raw BitString type.
 ///
 pub external fn from_string(String) -> BitString =
   "gleam_stdlib" "identity"
@@ -52,3 +52,31 @@ pub external fn int_to_u32(Int) -> Result(BitString, Nil) =
 ///
 pub external fn int_from_u32(BitString) -> Result(Int, Nil) =
   "gleam_stdlib" "bit_string_int_from_u32"
+
+/// Test to see whether a bit string is valid UTF-8.
+///
+pub fn is_utf8(bits: BitString) -> Bool {
+  case bits {
+    <<>> -> True
+    <<c:utf8, rest:binary>> -> {
+      // TODO: https://github.com/gleam-lang/gleam/issues/704
+      let _ = c
+      is_utf8(rest)
+    }
+    _ -> False
+  }
+}
+
+external fn unsafe_to_string(BitString) -> String =
+  "gleam_stdlib" "identity"
+
+/// Convert a bit string to a string.
+///
+/// Returns an error if the bit string is valid UTF-8 data.
+///
+pub fn to_string(bits: BitString) -> Result(String, Nil) {
+  case is_utf8(bits) {
+    True -> Ok(unsafe_to_string(bits))
+    False -> Error(Nil)
+  }
+}

src/gleam_stdlib.erl

@@ -202,6 +202,6 @@ regex_scan(Regex, String) ->
     end.
 
 base_decoded4(S) ->
-  try {ok, base64:decode(S)} catch
-    error:badarith -> {error, nil}
+  try {ok, base64:decode(S)}
+  catch error:badarith -> {error, nil}
   end.

test/gleam/bit_string_test.gleam

@@ -69,3 +69,25 @@ pub fn u32_test() {
     bit_string.int_from_u32(bit_string.from_string("12345")),
   )
 }
+
+pub fn to_string_test() {
+  <<>>
+  |> bit_string.to_string
+  |> should.equal(Ok(""))
+
+  <<"":utf8>>
+  |> bit_string.to_string
+  |> should.equal(Ok(""))
+
+  <<"Hello":utf8>>
+  |> bit_string.to_string
+  |> should.equal(Ok("Hello"))
+
+  <<"ø":utf8>>
+  |> bit_string.to_string
+  |> should.equal(Ok("ø"))
+
+  <<65535:16>>
+  |> bit_string.to_string
+  |> should.equal(Error(Nil))
+}