Fix charpos when the stored string is shrunk

Author: nobu

ext/jruby/org/jruby/ext/strscan/RubyStringScanner.java

@@ -236,8 +236,12 @@ public IRubyObject charpos(ThreadContext context) {
 
         ByteList strBL = str.getByteList();
         int strBeg = strBL.begin();
+        int strSize = strBL.getRealSize();
+        int pos = curr;
+
+        if (pos > strSize) pos = strSize;
 
-        return runtime.newFixnum(StringSupport.strLength(strBL.getEncoding(), strBL.unsafeBytes(), strBeg, strBeg + curr));
+        return runtime.newFixnum(StringSupport.strLength(strBL.getEncoding(), strBL.unsafeBytes(), strBeg, strBeg + pos));
     }
 
     private IRubyObject extractRange(Ruby runtime, int beg, int end) {

ext/strscan/strscan.c

@@ -79,7 +79,7 @@ struct strscanner
 #define CURPTR(s) (S_PBEG(s) + (s)->curr)
 #define S_RESTLEN(s) (S_LEN(s) - (s)->curr)
 
-#define EOS_P(s) ((s)->curr >= RSTRING_LEN(p->str))
+#define EOS_P(s) ((s)->curr >= RSTRING_LEN((s)->str))
 
 #define GET_SCANNER(obj,var) do {\
     (var) = check_strscan(obj);\
@@ -573,10 +573,12 @@ static VALUE
 strscan_get_charpos(VALUE self)
 {
     struct strscanner *p;
+    const char *s;
 
     GET_SCANNER(self, p);
 
-    return LONG2NUM(rb_enc_strlen(S_PBEG(p), CURPTR(p), rb_enc_get(p->str)));
+    s = EOS_P(p) ? S_PEND(p) : CURPTR(p);
+    return LONG2NUM(rb_enc_strlen(S_PBEG(p), s, rb_enc_get(p->str)));
 }
 
 /*

lib/strscan/truffleruby.rb

@@ -59,7 +59,7 @@ def pos=(new_pos)
   end
   alias_method :pointer=, :pos=
 
-  def charpos = Primitive.string_byte_index_to_character_index(@string, @pos)
+  def charpos = (eos? ? @string.size : Primitive.string_byte_index_to_character_index(@string, @pos))
 
   def rest = @string.byteslice(@pos, @string.bytesize)
 

test/strscan/test_stringscanner.rb

@@ -231,6 +231,14 @@ class << string
     assert_equal(8, scanner.charpos)
   end
 
+  def test_charpos_when_shrunk
+    s = "\u{e9}" * 64
+    sc = StringScanner.new(s)
+    sc.scan(/(?:\u{e9})+/)
+    s.replace("z")
+    assert_equal(s.length, sc.charpos)
+  end
+
   def test_concat
     s = create_string_scanner('a'.dup)
     s.scan(/a/)