ruby
diff --git a/‎prism_compile.c
Lines changed: 10 additions & 9 deletions b/‎prism_compile.c
Lines changed: 10 additions & 9 deletions
diff --git a/‎test/ruby/test_compile_prism.rb
Lines changed: 6 additions & 0 deletions b/‎test/ruby/test_compile_prism.rb
Lines changed: 6 additions & 0 deletions
@@ -200,23 +200,24 @@ parse_string_encoded(const pm_node_t *node, const pm_string_t *string, const pm_
 }
 
 static inline ID
-parse_symbol(const uint8_t *start, const uint8_t *end, const pm_parser_t *parser)
+parse_symbol(const uint8_t *start, const uint8_t *end, const char *encoding)
 {
-    rb_encoding *enc = rb_enc_from_index(rb_enc_find_index(parser->encoding->name));
+    rb_encoding *enc = rb_enc_from_index(rb_enc_find_index(encoding));
     return rb_intern3((const char *) start, end - start, enc);
 }
 
 static inline ID
-parse_string_symbol(const pm_string_t *string, const pm_parser_t *parser)
+parse_string_symbol(const pm_symbol_node_t *symbol, const pm_parser_t *parser)
 {
-    const uint8_t *start = pm_string_source(string);
-    return parse_symbol(start, start + pm_string_length(string), parser);
+    const char *encoding = symbol->base.flags & PM_SYMBOL_FLAGS_FORCED_UTF8_ENCODING ? "UTF-8" : parser->encoding->name;
+    const uint8_t *start = pm_string_source(&symbol->unescaped);
+    return parse_symbol(start, start + pm_string_length(&symbol->unescaped), encoding);
 }
 
 static inline ID
 parse_location_symbol(const pm_location_t *location, const pm_parser_t *parser)
 {
-    return parse_symbol(location->start, location->end, parser);
+    return parse_symbol(location->start, location->end, parser->encoding->name);
 }
 
 static int
@@ -395,7 +396,7 @@ pm_static_literal_value(const pm_node_t *node, const pm_scope_node_t *scope_node
       case PM_STRING_NODE:
         return parse_string(&((pm_string_node_t *) node)->unescaped, parser);
       case PM_SYMBOL_NODE:
-        return ID2SYM(parse_string_symbol(&((pm_symbol_node_t *) node)->unescaped, parser));
+        return ID2SYM(parse_string_symbol((pm_symbol_node_t *)node, parser));
       case PM_TRUE_NODE:
         return Qtrue;
       default:
@@ -1870,7 +1871,7 @@ pm_compile_pattern(rb_iseq_t *iseq, pm_scope_node_t *scope_node, const pm_node_t
                 const pm_node_t *key = ((const pm_assoc_node_t *) element)->key;
                 assert(PM_NODE_TYPE_P(key, PM_SYMBOL_NODE));
 
-                VALUE symbol = ID2SYM(parse_string_symbol(&((const pm_symbol_node_t *) key)->unescaped, scope_node->parser));
+                VALUE symbol = ID2SYM(parse_string_symbol((const pm_symbol_node_t *)key, scope_node->parser));
                 rb_ary_push(keys, symbol);
             }
         }
@@ -1915,7 +1916,7 @@ pm_compile_pattern(rb_iseq_t *iseq, pm_scope_node_t *scope_node, const pm_node_t
                 const pm_node_t *key = assoc->key;
                 assert(PM_NODE_TYPE_P(key, PM_SYMBOL_NODE));
 
-                VALUE symbol = ID2SYM(parse_string_symbol(&((const pm_symbol_node_t *) key)->unescaped, scope_node->parser));
+                VALUE symbol = ID2SYM(parse_string_symbol((const pm_symbol_node_t *)key, scope_node->parser));
                 ADD_INSN(ret, &line.node, dup);
                 ADD_INSN1(ret, &line.node, putobject, symbol);
                 ADD_SEND(ret, &line.node, rb_intern("key?"), INT2FIX(1));
 
@@ -779,6 +779,12 @@ def test_StringNode
 
     def test_SymbolNode
       assert_prism_eval(":pit")
+
+      # Test UTF-8 symbol in a US-ASCII file
+      assert_prism_eval(<<~'RUBY', raw: true)
+        # -*- coding: us-ascii -*-
+        :"\u{e9}"
+      RUBY
     end
 
     def test_XStringNode
Original file line number	Diff line number	Diff line change
`@@ -200,23 +200,24 @@ parse_string_encoded(const pm_node_t node, const pm_string_t string, const pm_`
`200`	`200`	`}`
`201`	`201`
`202`	`202`	`static inline ID`
`203`		`-parse_symbol(const uint8_t start, const uint8_t end, const pm_parser_t *parser)`
	`203`	`+parse_symbol(const uint8_t start, const uint8_t end, const char *encoding)`
`204`	`204`	`{`
`205`		`- rb_encoding *enc = rb_enc_from_index(rb_enc_find_index(parser->encoding->name));`
	`205`	`+ rb_encoding *enc = rb_enc_from_index(rb_enc_find_index(encoding));`
`206`	`206`	`return rb_intern3((const char *) start, end - start, enc);`
`207`	`207`	`}`
`208`	`208`
`209`	`209`	`static inline ID`
`210`		`-parse_string_symbol(const pm_string_t string, const pm_parser_t parser)`
	`210`	`+parse_string_symbol(const pm_symbol_node_t symbol, const pm_parser_t parser)`
`211`	`211`	`{`
`212`		`- const uint8_t *start = pm_string_source(string);`
`213`		`- return parse_symbol(start, start + pm_string_length(string), parser);`
	`212`	`+ const char *encoding = symbol->base.flags & PM_SYMBOL_FLAGS_FORCED_UTF8_ENCODING ? "UTF-8" : parser->encoding->name;`
	`213`	`+ const uint8_t *start = pm_string_source(&symbol->unescaped);`
	`214`	`+ return parse_symbol(start, start + pm_string_length(&symbol->unescaped), encoding);`
`214`	`215`	`}`
`215`	`216`
`216`	`217`	`static inline ID`
`217`	`218`	`parse_location_symbol(const pm_location_t location, const pm_parser_t parser)`
`218`	`219`	`{`
`219`		`- return parse_symbol(location->start, location->end, parser);`
	`220`	`+ return parse_symbol(location->start, location->end, parser->encoding->name);`
`220`	`221`	`}`
`221`	`222`
`222`	`223`	`static int`
`@@ -395,7 +396,7 @@ pm_static_literal_value(const pm_node_t node, const pm_scope_node_t scope_node`
`395`	`396`	`case PM_STRING_NODE:`
`396`	`397`	`return parse_string(&((pm_string_node_t *) node)->unescaped, parser);`
`397`	`398`	`case PM_SYMBOL_NODE:`
`398`		`- return ID2SYM(parse_string_symbol(&((pm_symbol_node_t *) node)->unescaped, parser));`
	`399`	`+ return ID2SYM(parse_string_symbol((pm_symbol_node_t *)node, parser));`
`399`	`400`	`case PM_TRUE_NODE:`
`400`	`401`	`return Qtrue;`
`401`	`402`	`default:`
`@@ -1870,7 +1871,7 @@ pm_compile_pattern(rb_iseq_t iseq, pm_scope_node_t scope_node, const pm_node_t`
`1870`	`1871`	`const pm_node_t key = ((const pm_assoc_node_t ) element)->key;`
`1871`	`1872`	`assert(PM_NODE_TYPE_P(key, PM_SYMBOL_NODE));`
`1872`	`1873`
`1873`		`- VALUE symbol = ID2SYM(parse_string_symbol(&((const pm_symbol_node_t *) key)->unescaped, scope_node->parser));`
	`1874`	`+ VALUE symbol = ID2SYM(parse_string_symbol((const pm_symbol_node_t *)key, scope_node->parser));`
`1874`	`1875`	`rb_ary_push(keys, symbol);`
`1875`	`1876`	`}`
`1876`	`1877`	`}`
`@@ -1915,7 +1916,7 @@ pm_compile_pattern(rb_iseq_t iseq, pm_scope_node_t scope_node, const pm_node_t`
`1915`	`1916`	`const pm_node_t *key = assoc->key;`
`1916`	`1917`	`assert(PM_NODE_TYPE_P(key, PM_SYMBOL_NODE));`
`1917`	`1918`
`1918`		`- VALUE symbol = ID2SYM(parse_string_symbol(&((const pm_symbol_node_t *) key)->unescaped, scope_node->parser));`
	`1919`	`+ VALUE symbol = ID2SYM(parse_string_symbol((const pm_symbol_node_t *)key, scope_node->parser));`
`1919`	`1920`	`ADD_INSN(ret, &line.node, dup);`
`1920`	`1921`	`ADD_INSN1(ret, &line.node, putobject, symbol);`
`1921`	`1922`	`ADD_SEND(ret, &line.node, rb_intern("key?"), INT2FIX(1));`