google-protobuf 3.25.6 → 4.30.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7ed4bd55ed50c29cd40fecbbd93d294546fe6f1d942c7440ec1fce2a26e2bc04
-  data.tar.gz: 9f499fbc0539610fe54b04da07fe65323b40edfc4220e0c2234ecfb9d1166108
+  metadata.gz: 3da1fe7a0b07a7333405dc10a634bede28ecd81b754e9c82f47d4d7d58c13400
+  data.tar.gz: d3795db46eb0c181a1aaceb7c4a8b7f8b7ae1fc333180bfb54ece1fdda9cf0c6
 SHA512:
-  metadata.gz: 35d44491d7a2163b7ce827e242ab59851d13ec5bb99be8feb0e0560945f62b54e7efd38a4aeaaf5721678b69d300fd769a0694294b63447f6cbca7db9f8d2aaf
-  data.tar.gz: 42035d41193b5343016af24fd141e7d27259d265e4b10dc15a467244a06a4d8d8daf3acc45865bcffa946162fce4fbadc4b31fbba406b4c52b30350a55e82935
+  metadata.gz: e25a6404869843044042d492b5612c2bb6bbf8d11cee513de10f5c2d0d41ed5869f965b68971f6727624965cb90afcf1c8433998bce5b9ba21a379815b46be63
+  data.tar.gz: 18f281a695a40fa48cc5d3243d8691e12c53596199887e7f2bc026f3e64ad7a36ca4fc1e2a6158f32519307fd0bfaac2cd14d63b3c433d42daeb0586859a84fe

data/ext/google/protobuf_c/convert.c

@@ -104,6 +104,34 @@ 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) {
+      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 +165,18 @@ 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) {
@@ -204,7 +221,8 @@ 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: