Enforce header entity length in grammar

__main__.py

@@ -1,7 +1,7 @@
 import sys
 import json
 import argparse
-from . import parse, ValidationError
+from . import parse, CollectedValidationErrors
 
 def main():
     parser = argparse.ArgumentParser(description="Parse and validate STEP file.")
@@ -22,11 +22,11 @@ def main():
         if not args.json:
             print("Valid", file=sys.stderr)
         exit(0)
-    except ValidationError as exc:
+    except CollectedValidationErrors as exc:
         if not args.json:
             print(exc, file=sys.stderr)
         else:
-            json.dump(exc.asdict(), sys.stdout)
+            json.dump([e.asdict() for e in exc.errors], sys.stdout, indent=2)
         exit(1)
 
 if __name__ == '__main__':

fixtures/fail_multiple_duplicate_ids.ifc

@@ -0,0 +1,29 @@
+ISO-10303-21;
+HEADER;
+FILE_DESCRIPTION(('ViewDefinition [CoordinationView]'),'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.,$,.RADIAN.);
+#18=IFCMEASUREWITHUNIT(IFCPLANEANGLEMEASURE(0.017453292519943295),#16);
+#18=IFCCONVERSIONBASEDUNIT(#12,.PLANEANGLEUNIT.,'DEGREE',#17);
+#19=IFCUNITASSIGNMENT((#13,#14,#15,#18));
+#19=IFCPROJECT('2AyG2X0sb16Bjd4gQc07yZ',#5,'',$,$,$,$,(#11),#19);
+ENDSEC;
+END-ISO-10303-21;

fixtures/fail_multiple_wrong_header_fields.ifc

@@ -0,0 +1,30 @@
+ISO-10303-21;
+HEADER;
+FILE_DESCRIPTION(('ViewDefinition [ReferenceView_V1.2]', 'ExchangeRequirement [Any]'));
+FILE_NAME('Header.ifc','2025-02-13T15:58:45',('tricott'),('Trimble Inc.'),'TrimBimToIFC rel. 4.0.2','Example - Example - 2025.0','IFC4 model', '');
+FILE_SCHEMA(('IFC4'));
+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;

fixtures/fail_too_many_header_entity_fields.ifc

@@ -0,0 +1,30 @@
+ISO-10303-21;
+HEADER;
+FILE_DESCRIPTION(('ViewDefinition [ReferenceView_V1.2]', 'ExchangeRequirement [Any]'),'2;1');
+FILE_NAME('Header.ifc','2025-02-13T15:58:45',('tricott'),('Trimble Inc.'),'TrimBimToIFC rel. 4.0.2','Example - Example - 2025.0','IFC4 model', '');
+FILE_SCHEMA(('IFC4'));
+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;

parser/errors.py

@@ -0,0 +1,108 @@
+from lark.exceptions import UnexpectedToken
+
+class _ValidationError(Exception):
+    def __init__(self, *args, **kwargs):
+        if self.__class__ is _ValidationError:
+            raise TypeError("Do not raise _ValidationError directly.")
+        super().__init__(*args, **kwargs)
+
+class ErrorCollector:
+    def __init__(self):
+        self.errors = []
+
+    def add(self, error):
+        self.errors.append(error)
+
+    def raise_if_any(self):
+        if self.errors:
+            raise CollectedValidationErrors(self.errors)
+
+class CollectedValidationErrors(_ValidationError):
+    def __init__(self, errors):
+        self.errors = errors
+
+    def asdict(self, with_message=True):
+        return [e.asdict(with_message=with_message) for e in self.errors]
+
+    def __str__(self):
+        return f"{len(self.errors)} validation error(s) collected:\n" + "\n\n".join(str(e) for e in self.errors)
+
+class SyntaxError(_ValidationError):
+    def __init__(self, filecontent, exception):
+        self.filecontent = filecontent
+        self.exception = exception
+
+    def asdict(self, with_message=True):
+        return {
+            "type": (
+                "unexpected_token"
+                if isinstance(self.exception, UnexpectedToken)
+                else "unexpected_character"
+            ),
+            "lineno": self.exception.line,
+            "column": self.exception.column,
+            "found_type": self.exception.token.type.lower(),
+            "found_value": self.exception.token.value,
+            "expected": sorted(x for x in self.exception.accepts if "__ANON" not in x),
+            "line": self.filecontent.split("\n")[self.exception.line - 1],
+            **({"message": str(self)} if with_message else {}),
+        }
+
+    def __str__(self):
+        d = self.asdict(with_message=False)
+        if len(d["expected"]) == 1:
+            exp = d["expected"][0]
+        else:
+            exp = f"one of {' '.join(d['expected'])}"
+
+        sth = "character" if d["type"] == "unexpected_character" else ""
+
+        return f"On line {d['lineno']} column {d['column']}:\nUnexpected {sth}{d['found_type']} ('{d['found_value']}')\nExpecting {exp}\n{d['lineno']:05d} | {d['line']}\n        {' ' * (self.exception.column - 1)}^"
+
+
+class DuplicateNameError(_ValidationError):
+    def __init__(self, filecontent, name, linenumbers):
+        self.name = name
+        self.filecontent = filecontent
+        self.linenumbers = linenumbers
+
+    def asdict(self, with_message=True):
+        return {
+            "type": "duplicate_name",
+            "name": self.name,
+            "lineno": self.linenumbers[0],
+            "line": self.filecontent.split("\n")[self.linenumbers[0] - 1],
+            **({"message": str(self)} if with_message else {}),
+        }
+
+    def __str__(self):
+        d = self.asdict(with_message=False)
+
+        def build():
+            yield f"On line {d['lineno']}:\nDuplicate instance name #{d['name']}"
+            yield f"{d['lineno']:05d} | {d['line']}"
+            yield " " * 8 + "^" * len(d["line"].rstrip())
+
+        return "\n".join(build())
+
+
+class HeaderFieldError(_ValidationError):
+    def __init__(self, field, found_len, expected_len):
+        self.field = field
+        self.found_len = found_len
+        self.expected_len = expected_len
+
+    def asdict(self, with_message=True):
+        return {
+            "type": "invalid_header_field",
+            "field": self.field,
+            "expected_field_count": self.expected_len,
+            "actual_field_count": self.found_len,
+            **({"message": str(self)} if with_message else {}),
+        }
+
+    def __str__(self):
+        return (
+            f"Invalid number of parameters for HEADER field '{self.field}'. "
+            f"Expected {self.expected_len}, found {self.found_len}."
+        )

parser/file.py

@@ -0,0 +1,108 @@
+import types
+import re 
+import numbers
+import itertools
+
+from .parse import parse, ParseResult
+from .grammar import HEADER_FIELDS
+from .transformer import entity_instance
+
+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 file:
+    """
+    A somewhat compatible interface (but very limited) to ifcopenshell.file
+    """
+
+    def __init__(self, result:ParseResult):
+        self.header_ = result.header
+        self.data_ = result.entities
+
+    @property
+    def schema_identifier(self) -> str:
+        return self.header_["FILE_SCHEMA"][0][0]
+
+    @property
+    def schema(self) -> str:
+        """General IFC schema version: IFC2X3, IFC4, IFC4X3."""
+        prefixes = ("IFC", "X", "_ADD", "_TC")
+        reg = "".join(f"(?P<{s}>{s}\\d+)?" for s in prefixes)
+        match = re.match(reg, self.schema_identifier)
+        version_tuple = tuple(
+            map(
+                lambda pp: int(pp[1][len(pp[0]) :]) if pp[1] else None,
+                ((p, match.group(p)) for p in prefixes),
+            )
+        )
+        return "".join(
+            "".join(map(str, t)) if t[1] else ""
+            for t in zip(prefixes, version_tuple[0:2])
+        )
+
+    @property
+    def schema_version(self) -> tuple[int, int, int, int]:
+        """Numeric representation of the full IFC schema version.
+
+        E.g. IFC4X3_ADD2 is represented as (4, 3, 2, 0).
+        """
+        schema = self.wrapped_data.schema
+        version = []
+        for prefix in ("IFC", "X", "_ADD", "_TC"):
+            number = re.search(prefix + r"(\d)", schema)
+            version.append(int(number.group(1)) if number else 0)
+        return tuple(version)
+
+
+    @property
+    def header(self):
+        header = {}
+        for field_name, namedtuple_class in HEADER_FIELDS.items():
+            field_data = self.header_.get(field_name.upper(), [])
+            header[field_name.lower()] = namedtuple_class(*field_data)
+
+        return types.SimpleNamespace(**header)
+
+
+    @property
+    def mvd(self):
+        if not LARK_AVAILABLE or MvdInfo is None:
+            return None
+        return MvdInfo(self.header)
+
+    def __getitem__(self, key: numbers.Integral) -> entity_instance:
+        return self.by_id(key)
+
+    def by_id(self, id: int) -> entity_instance:
+        """Return an IFC entity instance filtered by IFC ID.
+
+        :param id: STEP numerical identifier
+        :type id: int
+
+        :raises RuntimeError: If `id` is not found or multiple definitions exist for `id`.
+
+        :rtype: entity_instance
+        """
+        ns = self.data_.get(id, [])
+        if len(ns) == 0:
+            raise RuntimeError(f"Instance with id {id} not found")
+        elif len(ns) > 1:
+            raise RuntimeError(f"Duplicate definition for id {id}")
+        return ns[0]
+
+    def by_type(self, type: str) -> list[entity_instance]:
+        """Return IFC objects filtered by IFC Type and wrapped with the entity_instance class.
+        :rtype: list[entity_instance]
+        """
+        type_lc = type.lower()
+        return list(
+            filter(
+                lambda ent: ent.type.lower() == type_lc,
+                itertools.chain.from_iterable(self.data_.values()),
+            )
+        )
+
+def open(fn, only_header= False) -> file:
+    return file(parse(filename=fn, only_header=only_header))