check for invalid header entity

__init__.py

@@ -10,7 +10,10 @@
 
 from lark import Lark, Transformer, Tree, Token
 from lark.exceptions import UnexpectedToken, UnexpectedCharacters
-from .mvd_info import MvdInfo, LARK_AVAILABLE
+try:
+    from .mvd_info import MvdInfo, LARK_AVAILABLE
+except ImportError: # in case of running module locally (e.g. test_parser.py)
+    from mvd_info import MvdInfo, LARK_AVAILABLE
 
 class ValidationError(Exception):
     pass
@@ -87,8 +90,10 @@ def build():
 subsuper_record : "(" simple_record_list ")" 
 simple_record_list:simple_record simple_record* 
 simple_record: keyword "("parameter_list?")"
-header_entity :keyword "(" parameter_list ")" ";" 
-header_entity_list: header_entity header_entity* 
+header_entity_list: file_description file_name file_schema
+file_description: "FILE_DESCRIPTION" "(" parameter_list ")" ";"
+file_name: "FILE_NAME" "(" parameter_list ")" ";"
+file_schema: "FILE_SCHEMA" "(" parameter_list ")" ";"
 id: /#[0-9]+/
 keyword: /[A-Z][0-9A-Z_]*/
 parameter: untyped_parameter|typed_parameter|omitted_parameter
@@ -285,17 +290,18 @@ def traverse(fn, x):
         attributes,
         (min(lines), max(lines)),
     )
+
+def make_header_ent(ast):
+    rule = ast.data
+    params = T(visit_tokens=True).transform(ast.children[0])
+    return rule.upper(), params
+
 
 
 def process_tree(filecontent, file_tree, with_progress, with_header=False):
     ents = defaultdict(list)
     header, data = file_tree.children
 
-    def make_header_ent(ast):
-        kw, param_list = ast.children
-        kw = kw.children[0].value
-        return kw, T(visit_tokens=True).transform(param_list)
-
     if with_header:
         header = dict(map(make_header_ent, header.children[0].children))
 
@@ -366,11 +372,6 @@ def replace_fn(match):
 
         header_tree = ast.children[0]  # HEADER section
 
-        def make_header_ent(ast):
-            kw, param_list = ast.children
-            kw = kw.children[0].value
-            return kw, T(visit_tokens=True).transform(param_list)
-
         header = dict(map(make_header_ent, header_tree.children[0].children))
         return header
 

fixtures/extended_mvd.ifc

@@ -0,0 +1,29 @@
+ISO-10303-21;
+HEADER;
+FILE_DESCRIPTION(('ViewDefinition [ReferenceView_V1.2]', 'ViewDefinition [QuantityTakeOffAddOnView]', 'Option [ExcludedObjects: Stair, Ramp, Space; SplitLevel: On]', 'ExchangeRequirement [CustomRequirement: Value1, Value2]', 'Remark [SomeKey: SomeValue; AnotherKey: AnotherValue]', 'Comment [This is a free text comment, or a comma-separated list of items]'),'2;1');
+FILE_NAME('','2022-05-04T08:08:30',(''),(''),'IfcOpenShell-0.7.0','IfcOpenShell-0.7.0','');
+FILE_SCHEMA(('IFC4'));
+ENDSEC;
+DATA;
+#1=IFCPERSON($,$,'',$,$,$,$);
+#2=IFCORGANIZATION($,'',$,$,$);
+#3=IFCPERSONANDORGANIZATION(#1,#2,$);
+#4=IFCAPPLICATION(#2,'0.7.0','IfcOpenShell-0.7.0','');
+#5=IFCOWNERHISTORY(#3,#4,$,.ADDED.,$,#3,#4,1651651710);
+#6=IFCDIRECTION((1.,0.,0.));
+#7=IFCDIRECTION((0.,0.,1.));
+#8=IFCCARTESIANPOINT((0.,0.,0.));
+#9=IFCAXIS2PLACEMENT3D(#8,#7,#6);
+#10=IFCDIRECTION((0.,1.,0.));
+#11=IFCGEOMETRICREPRESENTATIONCONTEXT($,'Model',3,1.E-05,#9,#10);
+#12=IFCDIMENSIONALEXPONENTS(0,0,0,0,0,0,0);
+#13=IFCSIUNIT(*,.LENGTHUNIT.,$,.METRE.);
+#14=IFCSIUNIT(*,.AREAUNIT.,$,.SQUARE_METRE.);
+#15=IFCSIUNIT(*,.VOLUMEUNIT.,$,.CUBIC_METRE.);
+#16=IFCSIUNIT(*,.PLANEANGLEUNIT.,$,.RADIAN.);
+#17=IFCMEASUREWITHUNIT(IFCPLANEANGLEMEASURE(0.017453292519943295),#16);
+#18=IFCCONVERSIONBASEDUNIT(#12,.PLANEANGLEUNIT.,'DEGREE',#17);
+#19=IFCUNITASSIGNMENT((#13,#14,#15,#18));
+#20=IFCPROJECT('2AyG2X0sb16Bjd4gQc07yZ',#5,'',$,$,$,$,(#11),#19);
+ENDSEC;
+END-ISO-10303-21;

fixtures/fail_invalid_header_entity.ifc

@@ -0,0 +1,46 @@
+ISO-10303-21;
+HEADER;
+FILE_DESCRIPTION(('ViewDefinition [Alignment-basedView]'),'2;1');
+FILE_NAME('Header example2.ifc', '2022-09-16T10:35:07', ('Evandro Alfieri'), ('buildingSMART Int.'), 'IFC Motor 1.0', 'Company - Application - 26.0.0.0', 'none');
+FILE_SCHEMA(('IFC4X3_ADD2'));
+FILE_INFO(
+'Report Date', '2025-04-21 18:16:37',
+'IFC Schema', 'IFC2X3',
+'MVD(s)', 'None',
+'File Name in Header', 'generated (2).ifc',
+'File Name', 'generated (2).ifc',
+'File Size', '0.41 kB',
+'File Date', '2025-04-21 18:16:37',
+'Originating System', 'PHP Script',
+'Preprocessor Version', '1.0.0',
+'Company Name', 'Your Company Name',
+'Application Name', 'IFC Generator PHP Script',
+'Application Version', '1.0.0',
+'Author', 'Your Name',
+'Organization', 'Your Company Name'
+);
+ENDSEC;
+DATA;
+#1=IFCPERSON($,$,'',$,$,$,$,$);
+#2=IFCORGANIZATION($,'',$,$,$);
+#3=IFCPERSONANDORGANIZATION(#1,#2,$);
+#4=IFCAPPLICATION(#2,'v0.7.0-6c9e130ca','IfcOpenShell-v0.7.0-6c9e130ca','');
+#5=IFCOWNERHISTORY(#3,#4,$,.NOTDEFINED.,$,#3,#4,1700419055);
+#6=IFCDIRECTION((1.,0.,0.));
+#7=IFCDIRECTION((0.,0.,1.));
+#8=IFCCARTESIANPOINT((0.,0.,0.));
+#9=IFCAXIS2PLACEMENT3D(#8,#7,#6);
+#10=IFCDIRECTION((0.,1.));
+#11=IFCGEOMETRICREPRESENTATIONCONTEXT($,'Model',3,1.E-05,#9,#10);
+#12=IFCDIMENSIONALEXPONENTS(0,0,0,0,0,0,0);
+#13=IFCSIUNIT(*,.LENGTHUNIT.,$,.METRE.);
+#14=IFCSIUNIT(*,.AREAUNIT.,$,.SQUARE_METRE.);
+#15=IFCSIUNIT(*,.VOLUMEUNIT.,$,.CUBIC_METRE.);
+#16=IFCSIUNIT(*,.PLANEANGLEUNIT.,$,.RADIAN.);
+#17=IFCMEASUREWITHUNIT(IFCPLANEANGLEMEASURE(0.017453292519943295),#16);
+#18=IFCCONVERSIONBASEDUNIT(#12,.PLANEANGLEUNIT.,'DEGREE',#17);
+#19=IFCUNITASSIGNMENT((#13,#14,#15,#18));
+#20=IFCPROJECT('0iDmeiiLP3AOllitM2Favn',#5,'',$,$,$,$,(#11),#19);
+#21=IFCSITE('3rg2jGkIH10RFhrQsGZKRk',#5,$,$,$,$,$,$,$,$,$,$,$,$);
+ENDSEC;
+END-ISO-10303-21;

test_parser.py

@@ -104,3 +104,5 @@ def test_file_mvd_attr():
     assert 'Ramp' in f.mvd.options['ExcludedObjects']
     assert f.mvd.Remark['SomeKey'] == 'SomeValue'
     assert len(f.mvd.comments) == 2
+    assert all(v in vars(f.header).keys() for v in ['file_description', 'file_name', 'file_schema'])
+    assert len(f.header.file_name) == 7