From b6712602e27e0378e120ef44f0575d56c76d855a Mon Sep 17 00:00:00 2001
From: Thomas Krijnen <t.krijnen@gmail.com>
Date: Thu, 28 Aug 2025 11:15:05 +0200
Subject: [PATCH] Retain exact attribute counts in rdb as parsed

---
 src/ifcparse/IfcFile.cpp              | 10 ++++++----
 src/ifcparse/IfcFile.h                |  2 ++
 src/ifcparse/storage.h                |  2 +-
 src/serializers/RocksDbSerializer.cpp |  4 ++++
 4 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/src/ifcparse/IfcFile.cpp b/src/ifcparse/IfcFile.cpp
index 819582f931..7ffcf82a28 100644
--- a/src/ifcparse/IfcFile.cpp
+++ b/src/ifcparse/IfcFile.cpp
@@ -225,7 +225,7 @@ namespace {
     }
 }
 
-IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size, int resolve_reference_index) {
+IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size, int resolve_reference_index, bool coerce_attribute_count) {
     std::vector<const IfcParse::parameter_type*> parameter_types;
     std::unique_ptr<IfcParse::named_type> transient_named_type;
 
@@ -257,9 +257,11 @@ IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_re
         return IfcEntityInstanceData(in_memory_attribute_storage(0));
     }
 
-    in_memory_attribute_storage storage(decl != nullptr
+    in_memory_attribute_storage storage(coerce_attribute_count
+        ? (decl != nullptr
         ? (std::min)(parameter_types.size(), tokens_.size())
-        : tokens_.size()
+        : tokens_.size())
+		: tokens_.size()
     );
 
     auto it = tokens_.begin();
@@ -635,7 +637,7 @@ std::optional<std::tuple<size_t, const IfcParse::declaration*, IfcEntityInstance
                 Logger::Status(ss.str(), false);
             }
 
-            auto data = ps.construct(current_id, references_to_resolve_, entity_type, boost::none, -1);
+            auto data = ps.construct(current_id, references_to_resolve_, entity_type, boost::none, -1, coerce_attribute_count);
 
             return_value.emplace(
                 (size_t)current_id,
diff --git a/src/ifcparse/IfcFile.h b/src/ifcparse/IfcFile.h
index 00db611f68..18df6563ae 100644
--- a/src/ifcparse/IfcFile.h
+++ b/src/ifcparse/IfcFile.h
@@ -101,6 +101,8 @@ private:
     int yielded_header_instances_ = 0;
 
 public:
+	bool coerce_attribute_count = true;
+
     operator bool() const {
         return good_ && !lexer_->stream->eof;
     }
diff --git a/src/ifcparse/storage.h b/src/ifcparse/storage.h
index 9a26039e1f..2a1e2778c4 100644
--- a/src/ifcparse/storage.h
+++ b/src/ifcparse/storage.h
@@ -184,7 +184,7 @@ namespace IfcParse {
 
         void push(IfcUtil::IfcBaseClass* inst);
 
-        IfcEntityInstanceData construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size, int resolve_reference_index);
+        IfcEntityInstanceData construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size, int resolve_reference_index, bool coerce_attribute_count=true);
     };
 
     namespace impl {
diff --git a/src/serializers/RocksDbSerializer.cpp b/src/serializers/RocksDbSerializer.cpp
index a9273017c1..d749d5f31c 100644
--- a/src/serializers/RocksDbSerializer.cpp
+++ b/src/serializers/RocksDbSerializer.cpp
@@ -96,6 +96,10 @@ void RocksDbSerializer::write_streaming_() {
 
 	IfcParse::InstanceStreamer streamer(input_filename);
 
+	// We do not want to coerce attribute counts here, because we want
+	// to store exactly what is in the file for validation purposes
+	streamer.coerce_attribute_count = false;
+
 	while (streamer) {
 		auto inst = streamer.read_instance();
 		if (inst) {