Improve model for BinData (#11)

* Add special-case hacks for BinData class

In the schema BinData is an extension of string, but modeling this in python
as a subclass of str with extra constructor args is not feasible. Instead
this change moves the string content to a `value` attribute which aligns
with how our to_dict converter presents element text content.

Fixes #9

* Remove xfail from many sample xml files and tidy

* Now that BinData parses properly we can remove the xfail marker from
  about half of the sample ome-xml files
* Refactor filename stem extraction
* Invert SHOULD_PASS to SHOULD_FAIL for easier updating
* Remove +x mode bit from sample xml files

src/ome_autogen.py

@@ -76,7 +76,10 @@ def local_import(item_type):
 
 def make_dataclass(component) -> List[str]:
     lines = ["from pydantic.dataclasses import dataclass", ""]
-    if isinstance(component, XsdType):
+    # FIXME: Refactor to remove BinData special-case.
+    if component.local_name == "BinData":
+        base_type = None
+    elif isinstance(component, XsdType):
         base_type = component.base_type
     else:
         base_type = component.type.base_type
@@ -105,6 +108,9 @@ def make_dataclass(component) -> List[str]:
         lines += ["_no_default = object()", ""]
 
     lines += ["@dataclass", f"class {component.local_name}{base_name}:"]
+    # FIXME: Refactor to remove BinData special-case.
+    if component.local_name == "BinData":
+        lines.append("    value: str")
     lines += members.lines(
         indent=1,
         force_defaults=" = _no_default  # type: ignore"
@@ -497,7 +503,8 @@ def _abstract_class(self) -> List[str]:
         return lines
 
     def lines(self) -> str:
-        if not self.is_complex:
+        # FIXME: Refactor to remove BinData special-case.
+        if not self.is_complex and self.elem.local_name != "BinData":
             lines = self._simple_class()
         elif self.elem.abstract:
             lines = self._abstract_class()

testing/test_autogen.py

@@ -23,8 +23,27 @@ def model(tmp_path_factory):
     return importlib.import_module(target_dir.name)
 
 
-SHOULD_PASS = {"example.ome", "small.ome"}
-SHOULD_RAISE = {"bad.ome"}
+SHOULD_FAIL = {
+    "ROI",
+    "commentannotation",
+    "filter",
+    "hcs",
+    "instrument",
+    "instrument-units-alternate",
+    "instrument-units-default",
+    "mapannotation",
+    "metadata-only",
+    "spim",
+    "tagannotation",
+    "timestampannotation",
+    "timestampannotation-posix-only",
+    "transformations-downgrade",
+    "transformations-upgrade",
+    "xmlannotation-body-space",
+    "xmlannotation-multi-value",
+    "xmlannotation-svg",
+}
+SHOULD_RAISE = {"bad"}
 
 
 def mark_xfail(fname):
@@ -35,16 +54,19 @@ def mark_xfail(fname):
         ),
     )
 
+def true_stem(p):
+    return p.name.partition(".")[0]
+
 
 params = [
-    f if f.stem in SHOULD_PASS.union(SHOULD_RAISE) else mark_xfail(f)
-    for f in TESTING_DIR.glob("data/*.xml")
+    mark_xfail(f) if true_stem(f) in SHOULD_FAIL else f
+    for f in TESTING_DIR.glob("data/*.ome.xml")
 ]
 
 
-@pytest.mark.parametrize("xml", params, ids=lambda x: x.stem)
+@pytest.mark.parametrize("xml", params, ids=true_stem)
 def test_convert_schema(model, xml):
-    if xml.stem in SHOULD_RAISE:
+    if true_stem(xml) in SHOULD_RAISE:
         with pytest.raises(XMLSchemaValidationError):
             assert from_xml(xml, model.OME)
     else: