google-protobuf 3.25.0 → 4.28.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f07b3e363c6a69664f352f94158f1c150d11fc6947fd8aafe1d06decbed0a545
-  data.tar.gz: dd3bb39d53484b82469238a0fd009d2049d4c1ce5732b1c141e73a1f5b23c8bb
+  metadata.gz: 89810f0fea0c365b6862bc16275aa7da8ef73c5923d82fed89ef9af70cc18850
+  data.tar.gz: fd9e7a43e25c88e31978b90384f7e5e693d90ffa13a026da771c20ec7faaabfa
 SHA512:
-  metadata.gz: '028e14308853bce11c830bd55735df413954ac79a986e46375964907ca984f5f93bf4cf0ed8632c175b5fe9946724d270a999733778e735f568fa215018275f9'
-  data.tar.gz: 3f6832e092254e27ca64fdf11367cb3a4e4a0e8f34d6e824044a8e701bcea5e88e4b8aaeba81ea8b83ab5e5a834969c074ba034e056cdc26484822064c0403f6
+  metadata.gz: bc5619f68ea52e4fa1deba15b2b87a016bbdb6ad49089d3e9e3c30b3cd1eb9e67ac1f0cc84a55503d2feb2eec92401fcdd4159b6aa99a4f4acb81c106ba91448
+  data.tar.gz: ce47defb15aec14aeb2f6117982530e5b0ec53138c0f9805cfe9c83f677e8aba37d7f1e6f787b411f396928f2b8053d8567ee48281611c481a7d128ea711dd16

data/ext/google/protobuf_c/convert.c

@@ -104,6 +104,41 @@ unknownval:
   rb_raise(rb_eRangeError, "Unknown symbol value for enum field '%s'.", name);
 }
 
+VALUE Convert_CheckStringUtf8(VALUE str) {
+  VALUE utf8 = rb_enc_from_encoding(rb_utf8_encoding());
+
+  if (rb_obj_encoding(str) == utf8) {
+    // Note: Just because a string is marked as having UTF-8 encoding does
+    // not mean that it is *valid* UTF-8.  We have to check separately
+    // whether it is valid.
+    if (rb_enc_str_coderange(str) == ENC_CODERANGE_BROKEN) {
+      // TODO: For now
+      // we only warn for this case.  We will remove the warning and throw an
+      // exception below in the 30.x release
+
+      rb_warn(
+          "String is invalid UTF-8. This will be an error in a future "
+          "version.");
+      // VALUE exc = rb_const_get_at(
+      //     rb_cEncoding, rb_intern("InvalidByteSequenceError"));
+      // rb_raise(exc, "String is invalid UTF-8");
+    }
+  } else {
+    // Note: this will not duplicate underlying string data unless
+    // necessary.
+    //
+    // This will throw an exception if the conversion cannot be performed:
+    // - Encoding::UndefinedConversionError if certain characters cannot be
+    //   converted to UTF-8.
+    // - Encoding::InvalidByteSequenceError if certain characters were invalid
+    //   in the source encoding.
+    str = rb_str_encode(str, utf8, 0, Qnil);
+    PBRUBY_ASSERT(rb_enc_str_coderange(str) != ENC_CODERANGE_BROKEN);
+  }
+
+  return str;
+}
+
 upb_MessageValue Convert_RubyToUpb(VALUE value, const char* name,
                                    TypeInfo type_info, upb_Arena* arena) {
   upb_MessageValue ret;
@@ -137,29 +172,18 @@ upb_MessageValue Convert_RubyToUpb(VALUE value, const char* name,
       }
       break;
     }
-    case kUpb_CType_String: {
-      VALUE utf8 = rb_enc_from_encoding(rb_utf8_encoding());
+    case kUpb_CType_String:
       if (rb_obj_class(value) == rb_cSymbol) {
         value = rb_funcall(value, rb_intern("to_s"), 0);
-      } else if (rb_obj_class(value) != rb_cString) {
+      } else if (!rb_obj_is_kind_of(value, rb_cString)) {
         rb_raise(cTypeError,
                  "Invalid argument for string field '%s' (given %s).", name,
                  rb_class2name(CLASS_OF(value)));
       }
 
-      if (rb_obj_encoding(value) != utf8) {
-        // Note: this will not duplicate underlying string data unless
-        // necessary.
-        value = rb_str_encode(value, utf8, 0, Qnil);
-
-        if (rb_enc_str_coderange(value) == ENC_CODERANGE_BROKEN) {
-          rb_raise(rb_eEncodingError, "String is invalid UTF-8");
-        }
-      }
-
+      value = Convert_CheckStringUtf8(value);
       ret.str_val = Convert_StringData(value, arena);
       break;
-    }
     case kUpb_CType_Bytes: {
       VALUE bytes = rb_enc_from_encoding(rb_ascii8bit_encoding());
       if (rb_obj_class(value) != rb_cString) {
@@ -204,11 +228,13 @@ upb_MessageValue Convert_RubyToUpb(VALUE value, const char* name,
           ret.uint64_val = NUM2ULL(value);
           break;
         default:
-          break;
+          rb_raise(cTypeError, "Convert_RubyToUpb(): Unexpected type %d",
+                   (int)type_info.type);
       }
       break;
     default:
-      break;
+      rb_raise(cTypeError,
+                "Convert_RubyToUpb(): Unexpected type %d", (int)type_info.type);
   }
 
   return ret;
@@ -296,7 +322,8 @@ bool Msgval_IsEqual(upb_MessageValue val1, upb_MessageValue val2,
   if (upb_Status_IsOk(&status)) {
     return return_value;
   } else {
-    rb_raise(rb_eRuntimeError, upb_Status_ErrorMessage(&status));
+    rb_raise(rb_eRuntimeError, "Msgval_IsEqual(): %s",
+             upb_Status_ErrorMessage(&status));
   }
 }
 
@@ -309,6 +336,7 @@ uint64_t Msgval_GetHash(upb_MessageValue val, TypeInfo type_info,
   if (upb_Status_IsOk(&status)) {
     return return_value;
   } else {
-    rb_raise(rb_eRuntimeError, upb_Status_ErrorMessage(&status));
+    rb_raise(rb_eRuntimeError, "Msgval_GetHash(): %s",
+             upb_Status_ErrorMessage(&status));
   }
 }