Log and validate invalid attribute counts

src/ifcopenshell-python/ifcopenshell/validate.py

@@ -256,11 +256,12 @@ def assert_valid(
         return True
 
 
-def log_internal_cpp_errors(filename: str, logger: Logger) -> None:
+def log_internal_cpp_errors(f: ifcopenshell.file, filename: str, logger: Logger) -> None:
     import re
     import bisect
 
     chr_offset_re = re.compile(r"at offset (\d+)\s*")
+    for_instance_re = re.compile(r"\s*for instance #(\d+)\s*")
 
     log = ifcopenshell.get_log()
     msgs = list(map(json.loads, filter(None, log.split("\n"))))
@@ -286,6 +287,23 @@ def log_internal_cpp_errors(filename: str, logger: Logger) -> None:
                 else:
                     logger.error("For instance:\n    %s\n%s", line, m)
 
+    instance_messages = [for_instance_re.findall(m["message"]) for m in msgs]
+    if instance_messages:
+        for instid, msg in zip(instance_messages, msgs):
+            if instid:
+                m = for_instance_re.sub("", msg["message"])
+                try:
+                    inst = f[int(instid[0])]
+                except:
+                    inst = None
+                if hasattr(logger, "set_state"):
+                    logger.set_state("instance", inst)
+                    logger.set_state("attribute", None)
+                    logger.error(m)
+                elif inst:
+                    logger.error("For instance:\n    %s\n%s", inst, m)
+                else:
+                    logger.error(m)
 
 entity_attribute_map: dict[tuple[str, str], tuple[entity_type, tuple[attribute]]] = {}
 
@@ -368,7 +386,7 @@ def validate(f: Union[ifcopenshell.file, str], logger: Logger, express_rules=Fal
                 logger.error(f"Unsupported schema: {schema_name}")
                 return
 
-        log_internal_cpp_errors(filename, logger)
+        log_internal_cpp_errors(f, filename, logger)
 
     schema = ifcopenshell.ifcopenshell_wrapper.schema_by_name(f.schema_identifier)
     used_guids: dict[str, ifcopenshell.entity_instance] = dict()
@@ -490,7 +508,7 @@ def validate(f: Union[ifcopenshell.file, str], logger: Logger, express_rules=Fal
         # Re capturing the log when validate() is finished
         # iterating over every instance so that all attribute counts
         # are verified.
-        log_internal_cpp_errors(filename, logger)
+        log_internal_cpp_errors(f, filename, logger)
 
     # Restore the original value for 'use_attribute_value_derived'
     ifcopenshell.ifcopenshell_wrapper.set_feature("use_attribute_value_derived", attribute_value_derived_org)

src/ifcparse/IfcFile.cpp

@@ -180,7 +180,7 @@ namespace {
     }
 }
 
-IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl) {
+IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size) {
     std::vector<const IfcParse::parameter_type*> parameter_types;
 
     if ((decl != nullptr) && (decl->as_type_declaration() != nullptr)) {
@@ -197,8 +197,11 @@ IfcEntityInstanceData IfcParse::parse_context::construct(int name, unresolved_re
         );
     }
 
-    if ((decl != nullptr) && (tokens_.size() != parameter_types.size())) {
+    if (((decl != nullptr) && (tokens_.size() != parameter_types.size())) ||
-        // warning
+        expected_size && *expected_size != tokens_.size())
+    {
+        size_t expected = expected_size ? *expected_size : parameter_types.size();
+        Logger::Warning("Expected " + std::to_string(expected) + " attribute values, found " + std::to_string(tokens_.size()) + " for instance #" + std::to_string(name));
     }
 
     if (tokens_.empty()) {

src/ifcparse/IfcFile.h

@@ -91,7 +91,7 @@ struct parse_context {
 
     void push(IfcUtil::IfcBaseClass* inst);
 
-    IfcEntityInstanceData construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl);
+    IfcEntityInstanceData construct(int name, unresolved_references& references_to_resolve, const IfcParse::declaration* decl, boost::optional<size_t> expected_size);
 };
 
 /// This class provides several static convenience functions and variables

src/ifcparse/IfcParse.cpp

@@ -719,7 +719,7 @@ void IfcParse::IfcFile::load(unsigned entity_instance_name, const IfcParse::enti
                     tokens->Next();
                     load(0, nullptr, ps, -1);
                     const auto *decl = schema_->declaration_by_name(TokenFunc::asStringRef(next));
-                    auto* simple_type_instance = schema_->instantiate(decl, ps.construct(-1, references_to_resolve, decl));
+                    auto* simple_type_instance = schema_->instantiate(decl, ps.construct(-1, references_to_resolve, decl, boost::none));
                     //@todo decide addEntity(((IfcUtil::IfcBaseClass*)*entity));
                     context.push(simple_type_instance);
                     simple_type_instance->file_ = this;
@@ -748,7 +748,7 @@ IfcEntityInstanceData IfcParse::read(unsigned int i, IfcFile* f) {
     parse_context pc;
     f->tokens->Next();
     f->load(i, ty->as_entity(), pc, -1);
-    return IfcEntityInstanceData(pc.construct(i, f->references_to_resolve, ty));
+    return IfcEntityInstanceData(pc.construct(i, f->references_to_resolve, ty, boost::none));
 }
 
 void IfcParse::IfcFile::try_read_semicolon() const {
@@ -1300,7 +1300,7 @@ void IfcFile::initialize_(IfcParse::IfcSpfStream* s) {
             parse_context ps;
             tokens->Next();
             load(current_id, entity_type->as_entity(), ps, -1);
-            instance = schema_->instantiate(entity_type, ps.construct(current_id, references_to_resolve, entity_type));
+            instance = schema_->instantiate(entity_type, ps.construct(current_id, references_to_resolve, entity_type, boost::none));
             instance->file_ = this;
             instance->id_ = current_id;
 

src/ifcparse/IfcSpfHeader.cpp

@@ -41,12 +41,7 @@ namespace {
         parse_context pc;
         f->tokens->Next();
         f->load(-1, nullptr, pc, -1);
-        return pc.construct(-1, f->references_to_resolve, nullptr);
+        return pc.construct(-1, f->references_to_resolve, nullptr, s);
-        /*std::ostringstream oss;
-        ent.toString(oss);
-        auto osss = oss.str();
-        std::wcout << osss.c_str() << std::endl;*/
-        // return ent;
     }
 }