Fix OSM→OSW handling for ext-only geometries and node reference export

CHANGELOG.md

@@ -1,5 +1,11 @@
 # Change log
 
+### 0.3.3
+- Fix OSM→OSW export classification so canonical OSM tags are used for semantic recognition and `ext:*` tags are preserved as extensions instead of being treated as feature-defining tags.
+- Fix closed ext-only ways such as `ext:demolished:building=yes` to emit polygon output without falling through to point geometry construction.
+- Fix GeoJSON node/point export so topology endpoints remain in `nodes.geojson`, keeping edge `_u_id`/`_v_id` references OSW-compliant on roundtrip validation.
+- Add regression coverage for bug 3477 and serializer/compliance cases around ext-only geometries and remapped node references.
+
 ### 0.3.2
 - Fix duplicate polygon `_id` generation in OSM→OSW export by assigning sequential IDs per feature type.
 - Remap edge `_u_id`/`_v_id` and zone `_w_id` references to exported node IDs so references stay consistent after ID normalization.

requirements.txt

@@ -5,4 +5,4 @@ shapely~=2.0.2
 pyproj~=3.6.1
 coverage~=7.5.1
 ogr2osm==1.2.0
-python-osw-validation==0.3.1
+python-osw-validation==0.3.5

setup.py

@@ -26,7 +26,13 @@
     },
     long_description_content_type='text/markdown',
     url='https://github.com/TaskarCenterAtUW/TDEI-python-lib-osw-formatter',
-    install_requires=install_requires,
+    install_requires=[
+        'osmium~=3.6.0',
+        'asyncio~=3.4.3',
+        'networkx~=3.2',
+        'shapely~=2.0.2',
+        'ogr2osm==1.2.0'
+    ],
     packages=find_packages(where='src'),
     classifiers=[
         'Programming Language :: Python :: 3',

src/osm_osw_reformatter/serializer/osm/osm_graph.py

@@ -150,13 +150,20 @@ def way(self, w):
         if self.progressbar:
             self.progressbar.update(1)
 
-        if not self.line_filter(w.tags):
+        tags = dict(w.tags)
+        if not self.line_filter(tags):
             return
 
         d = {}
-        tags = dict(w.tags)
+        normalizer = OSWLineNormalizer(tags)
+
+        is_closed = len(w.nodes) > 2 and w.nodes[0].ref == w.nodes[-1].ref
+        if is_closed and normalizer.is_custom() and not (
+            normalizer.is_fence() or normalizer.is_tree_row()
+        ):
+            return
 
-        d2 = {**d, **OSWLineNormalizer(tags).normalize()}
+        d2 = {**d, **normalizer.normalize()}
 
         ndref = []
         for i in range(len(w.nodes)):
@@ -264,13 +271,23 @@ def area(self, a):
         if self.progressbar:
             self.progressbar.update(1)
 
-        if not self.polygon_filter(a.tags):
+        tags = dict(a.tags)
+        if not self.polygon_filter(tags):
             return
 
         d = {}
-        tags = dict(a.tags)
+        normalizer = OSWPolygonNormalizer(tags)
+        line_normalizer = OSWLineNormalizer(tags)
+        zone_normalizer = OSWZoneNormalizer(tags)
+
+        if normalizer.is_custom() and (
+            line_normalizer.is_fence()
+            or line_normalizer.is_tree_row()
+            or zone_normalizer.is_pedestrian()
+        ):
+            return
 
-        d2 = {**d, **OSWPolygonNormalizer(tags).normalize()}
+        d2 = {**d, **normalizer.normalize()}
 
         exteriors_count = 0
         for exterior in a.outer_rings():
@@ -534,7 +551,7 @@ def construct_geometries(self, progressbar: Optional[callable] = None) -> None:
 
                 if progressbar:
                     progressbar.update(1)
-            elif OSWPolygonNormalizer.osw_polygon_filter(d):
+            elif "ndref" in d and "indref" in d:
                 ndref = d.get("ndref")
                 indref = d.get("indref", [])
                 if not ndref:
@@ -547,7 +564,7 @@ def construct_geometries(self, progressbar: Optional[callable] = None) -> None:
 
                 if progressbar:
                     progressbar.update(1)
-            elif OSWLineNormalizer.osw_line_filter(d):
+            elif "ndref" in d:
                 ndref = d.get("ndref")
                 if not ndref:
                     continue
@@ -653,11 +670,28 @@ def _remap_node_ref(ref, node_id_map):
         for n, d in self.G.nodes(data=True):
             d_copy = {**d}
             source_id = _source_id(n)
+            geometry_obj = d_copy.pop("geometry")
+            geometry = mapping(geometry_obj)
+            geometry_type = geometry_obj.geom_type
+            is_topology_node = geometry_type == "Point" and self.G.degree(n) > 0
+
+            if is_topology_node:
+                _assign_ids(d_copy, node_id_counter, source_id)
+                node_id_map[n] = d_copy["_id"]
+                node_id_counter += 1
 
-            if OSWPointNormalizer.osw_point_filter(d):
+                if 'lon' in d_copy:
+                    d_copy.pop('lon')
+
+                if 'lat' in d_copy:
+                    d_copy.pop('lat')
+
+                node_features.append(
+                    {'type': 'Feature', 'geometry': geometry, 'properties': d_copy}
+                )
+            elif geometry_type == "Point" and OSWPointNormalizer.osw_point_filter(d):
                 _assign_ids(d_copy, point_id_counter, source_id)
                 point_id_counter += 1
-                geometry = mapping(d_copy.pop("geometry"))
 
                 if "lon" in d_copy:
                     d_copy.pop("lon")
@@ -668,26 +702,23 @@ def _remap_node_ref(ref, node_id_map):
                 point_features.append(
                     {"type": "Feature", "geometry": geometry, "properties": d_copy}
                 )
-            elif OSWLineNormalizer.osw_line_filter(d):
+            elif geometry_type == "LineString":
                 _assign_ids(d_copy, line_id_counter, source_id)
                 line_id_counter += 1
-                geometry = mapping(d_copy.pop("geometry"))
 
                 line_features.append(
                     {"type": "Feature", "geometry": geometry, "properties": d_copy}
                 )
-            elif OSWZoneNormalizer.osw_zone_filter(d):
+            elif geometry_type == "Polygon" and OSWZoneNormalizer.osw_zone_filter(d):
                 _assign_ids(d_copy, zone_id_counter, source_id)
                 zone_id_counter += 1
-                geometry = mapping(d_copy.pop("geometry"))
 
                 zone_features.append(
                     {"type": "Feature", "geometry": geometry, "properties": d_copy}
                 )
-            elif OSWPolygonNormalizer.osw_polygon_filter(d):
+            elif geometry_type == "Polygon":
                 _assign_ids(d_copy, polygon_id_counter, source_id)
                 polygon_id_counter += 1
-                geometry = mapping(d_copy.pop("geometry"))
 
                 polygon_features.append(
                     {"type": "Feature", "geometry": geometry, "properties": d_copy}
@@ -697,8 +728,6 @@ def _remap_node_ref(ref, node_id_map):
                 node_id_map[n] = d_copy["_id"]
                 node_id_counter += 1
 
-                geometry = mapping(d_copy.pop('geometry'))
-
                 if 'lon' in d_copy:
                     d_copy.pop('lon')
 

src/osm_osw_reformatter/serializer/osw/osw_normalizer.py

@@ -20,9 +20,7 @@
 
 
 def _tag_value(tags, key):
-    if key in tags:
-        return tags.get(key, "")
-    return tags.get(f"ext:{key}", "")
+    return tags.get(key, "")
 
 
 def _tags_to_dict(tags):
@@ -43,6 +41,25 @@ def _tags_to_dict(tags):
         return {}
 
 
+def _feature_tags(tags):
+    tag_dict = _tags_to_dict(tags)
+    internal_keys = {
+        "geometry",
+        "indref",
+        "lat",
+        "length",
+        "lon",
+        "ndref",
+        "osm_id",
+        "segment",
+    }
+    return {
+        k: v
+        for k, v in tag_dict.items()
+        if k not in internal_keys and not str(k).startswith("_")
+    }
+
+
 def _has_only_ext_tags(tags):
     if not tags:
         return False
@@ -278,7 +295,7 @@ def _normalize_kerb(self, keep_keys = {}, defaults = {}):
     def is_kerb(self):
         kerb_value = _tag_value(self.tags, "kerb")
         barrier_value = _tag_value(self.tags, "barrier")
-        has_kerb_key = "kerb" in self.tags or "ext:kerb" in self.tags
+        has_kerb_key = "kerb" in self.tags
         return (kerb_value in self.KERB_VALUES) or (
             barrier_value == "kerb" and (not has_kerb_key or kerb_value == "yes")
         )
@@ -361,7 +378,7 @@ def is_tree(self):
         return _tag_value(self.tags, "natural") == "tree"
 
     def is_custom(self):
-        tag_dict = _tags_to_dict(self.tags)
+        tag_dict = _feature_tags(self.tags)
         return _has_only_ext_tags(tag_dict)
 
 class OSWLineNormalizer:
@@ -405,7 +422,7 @@ def is_tree_row(self):
         return _tag_value(self.tags, "natural") == "tree_row"
 
     def is_custom(self):
-        tag_dict = _tags_to_dict(self.tags)
+        tag_dict = _feature_tags(self.tags)
         return _has_only_ext_tags(tag_dict)
 
 class OSWPolygonNormalizer:
@@ -556,13 +573,13 @@ def _normalize_polygon(self, keep_keys={}, defaults = {}):
         return new_tags
 
     def is_building(self):
-        return _tag_value(self.tags, "building") in self.BUILDING_VALUES
+        return self.tags.get("building", "") in self.BUILDING_VALUES
 
     def is_wood(self):
         return _tag_value(self.tags, "natural") == "wood"
 
     def is_custom(self):
-        tag_dict = _tags_to_dict(self.tags)
+        tag_dict = _feature_tags(self.tags)
         return _has_only_ext_tags(tag_dict)
 
 class OSWZoneNormalizer:

src/osm_osw_reformatter/version.py

@@ -1 +1 @@
-__version__ = '0.3.2'
+__version__ = '0.3.3'

tests/unit_tests/test_files/bug_3477.xml

@@ -0,0 +1,16 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<osm version="0.6" generator="test" upload="false">
+  <node id="1" visible="true" version="1" changeset="1" timestamp="2025-02-28T06:06:01Z" user="Cy Rossignol" uid="1" lat="47.6151489" lon="-122.3206583"/>
+  <node id="2" visible="true" version="1" changeset="1" timestamp="2025-02-28T06:06:01Z" user="Cy Rossignol" uid="1" lat="47.6149868" lon="-122.3206509"/>
+  <node id="3" visible="true" version="1" changeset="1" timestamp="2025-02-28T06:06:01Z" user="Cy Rossignol" uid="1" lat="47.6149898" lon="-122.3203263"/>
+  <node id="4" visible="true" version="1" changeset="1" timestamp="2025-02-28T06:06:01Z" user="Cy Rossignol" uid="1" lat="47.6151507" lon="-122.3203266"/>
+
+  <way id="301846" visible="true" version="2" changeset="816" timestamp="2026-01-07T22:03:22Z" user="Amy Bordenave" uid="8">
+    <nd ref="1"/>
+    <nd ref="4"/>
+    <nd ref="3"/>
+    <nd ref="2"/>
+    <nd ref="1"/>
+    <tag k="ext:demolished:building" v="yes"/>
+  </way>
+</osm>

tests/unit_tests/test_osm2osw/test_osm2osw.py

@@ -14,6 +14,7 @@
 TEST_INCLINE_FILE = os.path.join(ROOT_DIR, 'test_files/incline-test.xml')
 TEST_INVALID_NODE_TAGS_FILE = os.path.join(ROOT_DIR, 'test_files/node_with_invalid_tags.xml')
 TEST_TREE_FILE = os.path.join(ROOT_DIR, 'test_files/tree-test.xml')
+TEST_BUG_3477_FILE = os.path.join(ROOT_DIR, 'test_files/bug_3477.xml')
 
 
 def is_valid_float(value):
@@ -322,6 +323,33 @@ async def run_test():
 
         asyncio.run(run_test())
 
+    def test_bug_3477_ext_only_closed_way_emits_polygon(self):
+        osm_file_path = TEST_BUG_3477_FILE
+
+        async def run_test():
+            osm2osw = OSM2OSW(osm_file=osm_file_path, workdir=OUTPUT_DIR, prefix='bug3477')
+            result = await osm2osw.convert()
+            self.assertTrue(result.status)
+            self.assertEqual(len(result.generated_files), 1)
+
+            polygon_file = result.generated_files[0]
+            self.assertTrue(polygon_file.endswith('.graph.polygons.geojson'))
+
+            with open(polygon_file) as f:
+                geojson = json.load(f)
+
+            self.assertEqual(geojson.get("$schema"), OSW_SCHEMA_ID)
+            self.assertEqual(len(geojson.get("features", [])), 1)
+
+            feature = geojson["features"][0]
+            self.assertEqual(feature["geometry"]["type"], "Polygon")
+            self.assertEqual(feature["properties"].get("ext:demolished:building"), "yes")
+
+            for file_path in result.generated_files:
+                os.remove(file_path)
+
+        asyncio.run(run_test())
+
 
 if __name__ == '__main__':
     unittest.main()