Parquet: detect bbox geometry column when opening current Overture Ma…

…ps in dataset mode
rouault · Jul 24, 2024 · 9ef063c · 9ef063c
1 parent 3122857
commit 9ef063c
Show file tree

Hide file tree

Showing 2 changed files with 89 additions and 5 deletions.
diff --git a/autotest/ogr/ogr_parquet.py b/autotest/ogr/ogr_parquet.py
@@ -3317,13 +3317,22 @@ def test_ogr_parquet_bbox_double():
 
 
 @pytest.mark.require_geos
-def test_ogr_parquet_bbox_float32_but_no_covering_in_metadata():
+@pytest.mark.parametrize("use_dataset", [True, False])
+def test_ogr_parquet_bbox_float32_but_no_covering_in_metadata(use_dataset):
 
-    ds = ogr.Open("data/parquet/bbox_similar_to_overturemaps_2024-04-16-beta.0.parquet")
+    if use_dataset and not _has_arrow_dataset():
+        pytest.skip("Test requires build with ArrowDataset")
+
+    prefix = "PARQUET:" if use_dataset else ""
+
+    ds = ogr.Open(
+        prefix + "data/parquet/bbox_similar_to_overturemaps_2024-04-16-beta.0.parquet"
+    )
     lyr = ds.GetLayer(0)
     assert lyr.GetGeometryColumn() == "geometry"
     assert lyr.GetLayerDefn().GetFieldIndex("bbox.xmin") < 0
-    assert lyr.TestCapability(ogr.OLCFastGetExtent) == 1
+    if not use_dataset:
+        assert lyr.TestCapability(ogr.OLCFastGetExtent) == 1
     minx, maxx, miny, maxy = lyr.GetExtent()
     assert (minx, miny, maxx, maxy) == pytest.approx(
         (478315.53125, 4762880.5, 481645.3125, 4765610.5)
@@ -3337,14 +3346,16 @@ def test_ogr_parquet_bbox_float32_but_no_covering_in_metadata():
         maxy - (maxy - miny) / 2,
     ):
         f = lyr.GetNextFeature()
-        assert f.GetFID() == 8
+        if not use_dataset:
+            assert f.GetFID() == 8
         assert lyr.GetNextFeature() is None
 
     ds = None
 
     with gdaltest.config_option("OGR_PARQUET_USE_BBOX", "NO"):
         ds = ogr.Open(
-            "data/parquet/bbox_similar_to_overturemaps_2024-04-16-beta.0.parquet"
+            prefix
+            + "data/parquet/bbox_similar_to_overturemaps_2024-04-16-beta.0.parquet"
         )
         lyr = ds.GetLayer(0)
         assert lyr.GetGeometryColumn() == "geometry"

diff --git a/ogr/ogrsf_frmts/parquet/ogrparquetdatasetlayer.cpp b/ogr/ogrsf_frmts/parquet/ogrparquetdatasetlayer.cpp
@@ -163,6 +163,79 @@ void OGRParquetDatasetLayer::EstablishFeatureDefn()
     }
 
     const auto &fields = m_poSchema->fields();
+
+    // Overture Maps 2024-04-16-beta.0 almost follows GeoParquet 1.1, except
+    // they don't declare the "covering" element in the GeoParquet JSON metadata
+    if (m_oMapGeometryColumns.find("geometry") != m_oMapGeometryColumns.end() &&
+        bUseBBOX &&
+        !m_oMapGeometryColumns["geometry"].GetObj("covering").IsValid() &&
+        m_oMapGeometryColumns["geometry"].GetString("encoding") == "WKB")
+    {
+        for (int i = 0; i < m_poSchema->num_fields(); ++i)
+        {
+            const auto &field = fields[i];
+            if (field->name() == "bbox" &&
+                field->type()->id() == arrow::Type::STRUCT)
+            {
+                bool bXMin = false;
+                bool bXMax = false;
+                bool bYMin = false;
+                bool bYMax = false;
+                const auto subfields = field->Flatten();
+                if (subfields.size() == 4)
+                {
+                    for (int j = 0; j < static_cast<int>(subfields.size()); j++)
+                    {
+                        const auto &subfield = subfields[j];
+                        if (subfield->name() == "bbox.xmin")
+                            bXMin = true;
+                        else if (subfield->name() == "bbox.xmax")
+                            bXMax = true;
+                        else if (subfield->name() == "bbox.ymin")
+                            bYMin = true;
+                        else if (subfield->name() == "bbox.ymax")
+                            bYMax = true;
+                    }
+                }
+                if (bXMin && bXMax && bYMin && bYMax)
+                {
+                    CPLJSONObject oDef = m_oMapGeometryColumns["geometry"];
+                    CPLJSONObject oCovering;
+                    oDef.Add("covering", oCovering);
+                    CPLJSONObject oBBOX;
+                    oCovering.Add("bbox", oBBOX);
+                    {
+                        CPLJSONArray oArray;
+                        oArray.Add("bbox");
+                        oArray.Add("xmin");
+                        oBBOX.Add("xmin", oArray);
+                    }
+                    {
+                        CPLJSONArray oArray;
+                        oArray.Add("bbox");
+                        oArray.Add("ymin");
+                        oBBOX.Add("ymin", oArray);
+                    }
+                    {
+                        CPLJSONArray oArray;
+                        oArray.Add("bbox");
+                        oArray.Add("xmax");
+                        oBBOX.Add("xmax", oArray);
+                    }
+                    {
+                        CPLJSONArray oArray;
+                        oArray.Add("bbox");
+                        oArray.Add("ymax");
+                        oBBOX.Add("ymax", oArray);
+                    }
+                    oSetBBOXColumns.insert("bbox");
+                    m_oMapGeometryColumns["geometry"] = std::move(oDef);
+                }
+                break;
+            }
+        }
+    }
+
     for (int i = 0; i < m_poSchema->num_fields(); ++i)
     {
         const auto &field = fields[i];