Client now accepts :encoding option.

When :encoding is supplied to the client, the charset for the MySQL server
is set accordingly. Additionally, we ensure that the client sends query
strings in the encoding the connection expects.

Co-authored-by: Jean Boussier <jean.boussier@shopify.com>

CHANGELOG.md

@@ -13,6 +13,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 ### Fixed
   - Fix msec values for time columns. #61
 
+### Changed
+  - Encode query strings in encoding connection expects. #64
+
 ## 2.3.0
 
 ### Added

contrib/ruby/ext/trilogy-ruby/cext.c

@@ -30,8 +30,16 @@ struct trilogy_ctx {
     trilogy_conn_t conn;
     char server_version[TRILOGY_SERVER_VERSION_SIZE + 1];
     unsigned int query_flags;
+    VALUE encoding;
 };
 
+static void mark_trilogy(void *ptr)
+{
+    struct trilogy_ctx *ctx = ptr;
+    rb_gc_mark(ctx->encoding);
+}
+
+
 static void free_trilogy(void *ptr)
 {
     struct trilogy_ctx *ctx = ptr;
@@ -54,7 +62,7 @@ static size_t trilogy_memsize(const void *ptr) {
 static const rb_data_type_t trilogy_data_type = {
     .wrap_struct_name = "trilogy",
     .function = {
-        .dmark = NULL,
+        .dmark = mark_trilogy,
         .dfree = free_trilogy,
         .dsize = trilogy_memsize,
     },
@@ -356,12 +364,13 @@ static void authenticate(struct trilogy_ctx *ctx, trilogy_handshake_t *handshake
     }
 }
 
-static VALUE rb_trilogy_initialize(VALUE self, VALUE opts)
+static VALUE rb_trilogy_initialize(VALUE self, VALUE encoding, VALUE opts)
 {
     struct trilogy_ctx *ctx = get_ctx(self);
     trilogy_sockopt_t connopt = {0};
     trilogy_handshake_t handshake;
     VALUE val;
+    RB_OBJ_WRITE(self, &ctx->encoding, encoding);
 
     Check_Type(opts, T_HASH);
     rb_ivar_set(self, id_connection_options, opts);
@@ -817,7 +826,7 @@ static VALUE rb_trilogy_query(VALUE self, VALUE query)
 {
     struct trilogy_ctx *ctx = get_open_ctx(self);
 
-    StringValue(query);
+    query = rb_str_export_to_enc(query, rb_to_encoding(ctx->encoding));
 
     int rc = trilogy_query_send(&ctx->conn, RSTRING_PTR(query), RSTRING_LEN(query));
 
@@ -997,7 +1006,7 @@ RUBY_FUNC_EXPORTED void Init_cext()
     VALUE Trilogy = rb_const_get(rb_cObject, rb_intern("Trilogy"));
     rb_define_alloc_func(Trilogy, allocate_trilogy);
 
-    rb_define_method(Trilogy, "initialize", rb_trilogy_initialize, 1);
+    rb_define_private_method(Trilogy, "_initialize", rb_trilogy_initialize, 2);
     rb_define_method(Trilogy, "change_db", rb_trilogy_change_db, 1);
     rb_define_method(Trilogy, "query", rb_trilogy_query, 1);
     rb_define_method(Trilogy, "ping", rb_trilogy_ping, 0);

contrib/ruby/lib/trilogy.rb

@@ -107,6 +107,59 @@ class ConnectionClosed < IOError
     include ConnectionError
   end
 
+  ENCODINGS_MAP = mysql_to_rb = {
+    "big5"     => "Big5",
+    "dec8"     => nil,
+    "cp850"    => "CP850",
+    "hp8"      => nil,
+    "koi8r"    => "KOI8-R",
+    "latin1"   => "ISO-8859-1",
+    "latin2"   => "ISO-8859-2",
+    "swe7"     => nil,
+    "ascii"    => "US-ASCII",
+    "ujis"     => "eucJP-ms",
+    "sjis"     => "Shift_JIS",
+    "hebrew"   => "ISO-8859-8",
+    "tis620"   => "TIS-620",
+    "euckr"    => "EUC-KR",
+    "koi8u"    => "KOI8-R",
+    "gb2312"   => "GB2312",
+    "greek"    => "ISO-8859-7",
+    "cp1250"   => "Windows-1250",
+    "gbk"      => "GBK",
+    "latin5"   => "ISO-8859-9",
+    "armscii8" => nil,
+    "utf8"     => "UTF-8",
+    "ucs2"     => "UTF-16BE",
+    "cp866"    => "IBM866",
+    "keybcs2"  => nil,
+    "macce"    => "macCentEuro",
+    "macroman" => "macRoman",
+    "cp852"    => "CP852",
+    "latin7"   => "ISO-8859-13",
+    "utf8mb4"  => "UTF-8",
+    "cp1251"   => "Windows-1251",
+    "utf16"    => "UTF-16",
+    "cp1256"   => "Windows-1256",
+    "cp1257"   => "Windows-1257",
+    "utf32"    => "UTF-32",
+    "binary"   => "ASCII-8BIT",
+    "geostd8"  => nil,
+    "cp932"    => "Windows-31J",
+    "eucjpms"  => "eucJP-ms",
+    "utf16le"  => "UTF-16LE",
+    "gb18030"  => "GB18030",
+  }
+
+  attr_reader :encoding
+
+  def initialize(options = {})
+    mysql_encoding = options[:encoding] || "utf8mb4"
+    @encoding = Encoding.find(ENCODINGS_MAP[mysql_encoding])
+    _initialize(@encoding, **options)
+    query("SET NAMES #{mysql_encoding}")
+  end
+
   def connection_options
     @connection_options.dup.freeze
   end

contrib/ruby/test/client_test.rb

@@ -883,4 +883,34 @@ def test_trilogy_decimal_sum_query
     assert sum.is_a?(BigDecimal)
     assert_equal 8, sum
   end
+
+  def test_character_set_encoding_query
+    client = new_tcp_client(encoding: "cp932")
+
+    assert_equal "cp932", client.query("SELECT @@character_set_client").first.first
+    assert_equal "cp932", client.query("SELECT @@character_set_results").first.first
+    assert_equal "cp932", client.query("SELECT @@character_set_connection").first.first
+    assert_equal "cp932_japanese_ci", client.query("SELECT @@collation_connection").first.first
+
+    expected = "こんにちは".encode(Encoding::CP932)
+    assert_equal expected, client.query("SELECT 'こんにちは'").to_a.first.first
+  end
+
+  def test_character_set_encoding_handles_binary_query
+    client = new_tcp_client
+    expected = "\xff".b
+
+    result = client.query("SELECT _binary'#{expected}'").to_a.first.first
+    assert_equal expected, result
+    assert_equal Encoding::BINARY, result.encoding
+
+    result = client.query("SELECT '#{expected}'").to_a.first.first
+    assert_equal expected.dup.force_encoding(client.encoding), result
+    assert_equal client.encoding, result.encoding
+
+    client = new_tcp_client(encoding: "cp932")
+    result = client.query("SELECT '#{expected}'").to_a.first.first
+    assert_equal expected.dup.force_encoding(client.encoding), result
+    assert_equal client.encoding, result.encoding
+  end
 end