fix wrongly tagged encoding when inserting and selecting UTF16LE encoded values

Author: tenderlove

ext/sqlite3/sqlite3_ruby.h

@@ -19,6 +19,7 @@
 #ifdef HAVE_RUBY_ENCODING_H
 #include <ruby/encoding.h>
 
+#define USASCII_P(_obj) (rb_enc_get_index(_obj) == rb_usascii_encindex())
 #define UTF8_P(_obj) (rb_enc_get_index(_obj) == rb_utf8_encindex())
 #define UTF16_LE_P(_obj) (rb_enc_get_index(_obj) == rb_enc_find_index("UTF-16LE"))
 #define UTF16_BE_P(_obj) (rb_enc_get_index(_obj) == rb_enc_find_index("UTF-16BE"))

ext/sqlite3/statement.c

@@ -156,7 +156,7 @@ static VALUE step(VALUE self)
                     (long)sqlite3_column_bytes(stmt, i)
                 );
 #ifdef HAVE_RUBY_ENCODING_H
-                rb_enc_associate_index(str, enc_index);
+                rb_enc_associate_index(str, rb_utf8_encindex());
                 if(internal_encoding)
                   str = rb_str_export_to_enc(str, internal_encoding);
 #endif
@@ -238,22 +238,32 @@ static VALUE bind_param(VALUE self, VALUE key, VALUE value)
             SQLITE_TRANSIENT
             );
       } else {
+
+
 #ifdef HAVE_RUBY_ENCODING_H
-        if(!UTF8_P(value)) {
-              VALUE db          = rb_iv_get(self, "@connection");
-              VALUE encoding    = rb_funcall(db, rb_intern("encoding"), 0);
-              rb_encoding * enc = rb_to_encoding(encoding);
-              value = rb_str_export_to_enc(value, enc);
+        if (UTF16_LE_P(value)) {
+          status = sqlite3_bind_text16(
+              ctx->st,
+              index,
+              (const char *)StringValuePtr(value),
+              (int)RSTRING_LEN(value),
+              SQLITE_TRANSIENT
+              );
+        } else {
+          if (!UTF8_P(value) || !USASCII_P(value)) {
+              value = rb_str_encode(value, rb_enc_from_encoding(rb_utf8_encoding()), 0, Qnil);
           }
 #endif
-
         status = sqlite3_bind_text(
             ctx->st,
             index,
             (const char *)StringValuePtr(value),
             (int)RSTRING_LEN(value),
             SQLITE_TRANSIENT
             );
+#ifdef HAVE_RUBY_ENCODING_H
+        }
+#endif
       }
       break;
     case T_BIGNUM:

test/test_encoding.rb

@@ -11,6 +11,39 @@ def setup
       @db.execute(@create);
     end
 
+    def test_select_encoding_on_utf_16
+      str = "foo"
+      db = SQLite3::Database.new(':memory:'.encode('UTF-16LE'))
+      db.execute @create
+      db.execute "insert into ex (id, data) values (1, \"#{str}\")"
+
+      stmt = db.prepare 'select * from ex where data = ?'
+      ['US-ASCII', 'UTF-16LE', 'EUC-JP', 'UTF-8'].each do |enc|
+        stmt.bind_param 1, str.encode(enc)
+        assert_equal 1, stmt.to_a.length
+        stmt.reset!
+      end
+    end
+
+    def test_insert_encoding
+      str = "foo"
+      db = SQLite3::Database.new(':memory:'.encode('UTF-16LE'))
+      db.execute @create
+      stmt = db.prepare @insert
+
+      ['US-ASCII', 'UTF-16LE', 'UTF-16BE', 'EUC-JP', 'UTF-8'].each_with_index do |enc,i|
+        str2 = str.encode(enc)
+        stmt.bind_param 1, i
+        stmt.bind_param 2, str.encode(enc)
+        stmt.to_a
+        stmt.reset!
+      end
+
+      db.execute('select data from ex').flatten.each do |s|
+        assert_equal str, s
+      end
+    end
+
     def test_default_internal_is_honored
       warn_before = $-w
       $-w = false