[Bug #21528] Try to produce strings with valid encoding when truncating syntax errors

prism_compile.c

@@ -10706,6 +10706,21 @@ pm_parse_errors_format_line(const pm_parser_t *parser, const pm_newline_list_t *
         start += column_start;
     }
 
+    // Take care to not produce a string that is invalid in its encoding.
+    if (truncate_end) {
+        const uint8_t *end_adjusted = start;
+        while (end_adjusted < end) {
+            const size_t next_width = pm_encoding_utf_8_char_width(end_adjusted, 4);
+            // We should only get here with valid utf-8
+            RUBY_ASSERT(next_width > 0);
+
+            if (end_adjusted + next_width > end) {
+                end = end_adjusted;
+            }
+            end_adjusted += next_width;
+        }
+    }
+
     pm_buffer_append_string(buffer, (const char *) start, (size_t) (end - start));
 
     if (truncate_end) {

test/ruby/test_syntax.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: false
 require 'test/unit'
+require_relative '../lib/parser_support'
 
 class TestSyntax < Test::Unit::TestCase
   using Module.new {
@@ -1069,6 +1070,22 @@ eom
     assert_syntax_error("/(?<a>[)/=~s", /premature end of char-class/, bug20295)
   end
 
+  def test_syntax_error_message_truncation
+    omit unless ParserSupport.prism_enabled?
+
+    bug21528 = '[ruby-core:122899] [Bug #21528]'
+    begin
+      eval(<<~CODE)
+        if a
+        # 0000000000000ああああああ
+        #
+      CODE
+    rescue SyntaxError => e
+      assert(e.message.valid_encoding?)
+      assert_include(e.message, "# 0000000000000あああああ ...")
+    end
+  end
+
   def test_lineno_operation_brace_block
     expected = __LINE__ + 1
     actual = caller_lineno\