gdal vector check-geometry: support --include-field ALL

dbaston · web-flow · commit df9974498c4a · 2026-02-10T15:38:19.000+01:00
Harmonizes behavior between gdal vector check-geometry and gdal raster pixel-info. Changes the gdal raster pixel-info behavior such that fields are added to the output layer in the order they are specified, not the order they occur in the source layer.
diff --git a/apps/gdalalg_raster_pixel_info.cpp b/apps/gdalalg_raster_pixel_info.cpp
@@ -270,7 +270,7 @@ bool GDALRasterPixelInfoAlgorithm::RunStep(GDALPipelineStepRunContext &)
     const bool bIsGeoJSON = EQUAL(m_format.c_str(), "GeoJSON");
 
     OGRLayer *poSrcLayer = nullptr;
-    std::set<int> anSrcFieldIndicesToInclude;
+    std::vector<int> anSrcFieldIndicesToInclude;
     std::vector<int> anMapSrcToDstFields;
 
     if (poVectorSrcDS)
@@ -315,29 +315,11 @@ bool GDALRasterPixelInfoAlgorithm::RunStep(GDALPipelineStepRunContext &)
                         poVectorSrcDS->GetDescription());
             return false;
         }
-        if (m_includeFields.size() == 1 && m_includeFields[0] == "ALL")
-        {
-            const int nSrcFieldCount =
-                poSrcLayer->GetLayerDefn()->GetFieldCount();
-            for (int i = 0; i < nSrcFieldCount; ++i)
-                anSrcFieldIndicesToInclude.insert(i);
-        }
-        else if (!m_includeFields.empty() &&
-                 !(m_includeFields.size() == 1 && m_includeFields[0] == "NONE"))
+
+        if (!GetFieldIndices(m_includeFields, OGRLayer::ToHandle(poSrcLayer),
+                             anSrcFieldIndicesToInclude))
         {
-            for (const std::string &osFieldName : m_includeFields)
-            {
-                const int nIdx = poSrcLayer->GetLayerDefn()->GetFieldIndex(
-                    osFieldName.c_str());
-                if (nIdx < 0)
-                {
-                    ReportError(CE_Failure, CPLE_AppDefined,
-                                "Field '%s' does not exist in layer '%s'",
-                                osFieldName.c_str(), poSrcLayer->GetName());
-                    return false;
-                }
-                anSrcFieldIndicesToInclude.insert(nIdx);
-            }
+            return false;
         }
 
         if (m_posCrs.empty())
diff --git a/apps/gdalalg_vector_check_geometry.cpp b/apps/gdalalg_vector_check_geometry.cpp
@@ -32,7 +32,10 @@ GDALVectorCheckGeometryAlgorithm::GDALVectorCheckGeometryAlgorithm(
                                       standaloneStep)
 {
     AddArg("include-field", 0,
-           _("Fields from input layer to include in output"), &m_includeFields);
+           _("Fields from input layer to include in output (special values: "
+             "ALL and NONE)"),
+           &m_includeFields)
+        .SetDefault("NONE");
 
     AddArg("include-valid", 0,
            _("Include valid inputs in output, with empty geometry"),
@@ -348,19 +351,11 @@ bool GDALVectorCheckGeometryAlgorithm::RunStep(GDALPipelineStepRunContext &)
             }
 
             std::vector<int> includeFieldIndices;
-            for (const auto &fieldName : m_includeFields)
+            if (!GetFieldIndices(m_includeFields,
+                                 OGRLayer::ToHandle(poSrcLayer),
+                                 includeFieldIndices))
             {
-                auto iSrcField =
-                    poSrcLayerDefn->GetFieldIndex(fieldName.c_str());
-                if (iSrcField == -1)
-                {
-                    ReportError(
-                        CE_Failure, CPLE_AppDefined,
-                        "Specified field '%s' does not exist in layer '%s'",
-                        fieldName.c_str(), poSrcLayer->GetDescription());
-                    return false;
-                }
-                includeFieldIndices.push_back(iSrcField);
+                return false;
             }
 
             outDS->AddLayer(*poSrcLayer,
diff --git a/autotest/utilities/test_gdalalg_vector_check_geometry.py b/autotest/utilities/test_gdalalg_vector_check_geometry.py
@@ -64,8 +64,15 @@ def test_gdalalg_vector_check_geometry(alg, polys):
 
     dst_ds = alg["output"].GetDataset()
     dst_lyr = dst_ds.GetLayer(0)
+    dst_defn = dst_lyr.GetLayerDefn()
+
     assert dst_lyr.GetName() == "error_location"
-    assert dst_lyr.GetLayerDefn().GetGeomType() == ogr.wkbMultiPoint
+    assert dst_defn.GetGeomType() == ogr.wkbMultiPoint
+
+    field_names = [
+        dst_defn.GetFieldDefn(i).GetName() for i in range(dst_defn.GetFieldCount())
+    ]
+    assert field_names == ["error"]
 
     errors = [f for f in dst_lyr]
 
@@ -498,23 +505,46 @@ def test_gdalalg_vector_check_geometry_no_geometry_field(alg):
         alg.Run()
 
 
-def test_gdalalg_vector_check_geometry_include_field(alg):
+@pytest.mark.parametrize(
+    "include_field",
+    (
+        ["EAS_ID", "AREA"],
+        ["AREA", "EAS_ID"],
+        ["AREA", "AREA", "EAS_ID"],
+        "ALL",
+        None,
+        "NONE",
+    ),
+)
+def test_gdalalg_vector_check_geometry_include_field(alg, include_field):
 
     alg["input"] = "../ogr/data/poly.shp"
-    alg["include-field"] = ["EAS_ID", "AREA"]
+    if include_field is not None:
+        alg["include-field"] = include_field
     alg["output-format"] = "stream"
     alg["include-valid"] = True
 
     assert alg.Run()
     dst_ds = alg["output"].GetDataset()
     dst_lyr = dst_ds.GetLayer(0)
     dst_defn = dst_lyr.GetLayerDefn()
-    assert dst_defn.GetFieldDefn(0).GetName() == "EAS_ID"
-    assert dst_defn.GetFieldDefn(1).GetName() == "AREA"
+    dst_fields = [
+        dst_defn.GetFieldDefn(i).GetName() for i in range(dst_defn.GetFieldCount())
+    ]
+
+    if include_field in (None, "NONE"):
+        assert dst_fields == ["error"]
+    else:
+        unique_include_fields = list({f: True for f in include_field}.keys())
+
+        if include_field == "ALL":
+            assert dst_fields == ["AREA", "EAS_ID", "PRFEDEA", "error"]
+        else:
+            assert dst_fields == unique_include_fields + ["error"]
 
-    f = dst_lyr.GetNextFeature()
-    assert f["EAS_ID"] == 168
-    assert f["AREA"] == 215229.266
+        f = dst_lyr.GetNextFeature()
+        assert f["EAS_ID"] == 168
+        assert f["AREA"] == 215229.266
 
 
 def test_gdalalg_vector_check_geometry_include_field_error(alg):
@@ -523,5 +553,5 @@ def test_gdalalg_vector_check_geometry_include_field_error(alg):
     alg["include-field"] = "does_not_exist"
     alg["output-format"] = "stream"
 
-    with pytest.raises(Exception, match="Specified field .* does not exist"):
+    with pytest.raises(Exception, match="Field .* does not exist"):
         alg.Run()
diff --git a/doc/source/programs/gdal_vector_check_geometry.rst b/doc/source/programs/gdal_vector_check_geometry.rst
@@ -58,7 +58,8 @@ Program-Specific Options
 
    .. versionadded:: 3.12.1
 
-   Optional field(s) to copy from the input features to the output.
+   Optional field(s) to copy from the input features to the output. Since GDAL 3.13, the value ``ALL`` can be used to include all fields
+   from the source layer.
 
 .. option:: --include-valid
 
diff --git a/gcore/gdalalgorithm.cpp b/gcore/gdalalgorithm.cpp
@@ -6852,6 +6852,52 @@ GDALAlgorithm::GetAutoComplete(std::vector<std::string> &args,
     return ret;
 }
 
+/************************************************************************/
+/*                   GDALAlgorithm::GetFieldIndices()                   */
+/************************************************************************/
+
+bool GDALAlgorithm::GetFieldIndices(const std::vector<std::string> &names,
+                                    OGRLayerH hLayer, std::vector<int> &indices)
+{
+    VALIDATE_POINTER1(hLayer, __func__, false);
+
+    const OGRLayer &layer = *OGRLayer::FromHandle(hLayer);
+
+    if (names.size() == 1 && names[0] == "ALL")
+    {
+        const int nSrcFieldCount = layer.GetLayerDefn()->GetFieldCount();
+        for (int i = 0; i < nSrcFieldCount; ++i)
+        {
+            indices.push_back(i);
+        }
+    }
+    else if (!names.empty() && !(names.size() == 1 && names[0] == "NONE"))
+    {
+        std::set<int> fieldsAdded;
+        for (const std::string &osFieldName : names)
+        {
+
+            const int nIdx =
+                layer.GetLayerDefn()->GetFieldIndex(osFieldName.c_str());
+
+            if (nIdx < 0)
+            {
+                CPLError(CE_Failure, CPLE_AppDefined,
+                         "Field '%s' does not exist in layer '%s'",
+                         osFieldName.c_str(), layer.GetName());
+                return false;
+            }
+
+            if (fieldsAdded.insert(nIdx).second)
+            {
+                indices.push_back(nIdx);
+            }
+        }
+    }
+
+    return true;
+}
+
 /************************************************************************/
 /*              GDALAlgorithm::ExtractLastOptionAndValue()              */
 /************************************************************************/
diff --git a/gcore/gdalalgorithm_cpp.h b/gcore/gdalalgorithm_cpp.h
@@ -2995,6 +2995,10 @@ class CPL_DLL GDALAlgorithmRegistry
     std::pair<std::vector<std::pair<GDALAlgorithmArg *, std::string>>, size_t>
     GetArgNamesForCLI() const;
 
+    /** Get the indices of fields to be included, recognizing the special values "ALL" and "NONE" */
+    static bool GetFieldIndices(const std::vector<std::string> &osFieldNames,
+                                OGRLayerH hLayer, std::vector<int> &anIndices);
+
     //! @cond Doxygen_Suppress
     std::string GetUsageForCLIEnd() const;
     //! @endcond
