gdal vector create: add a --fid option, and other tweaks/fixes discovered along the way

rouault · rouault · commit 683b85e6ddbf · 2026-04-01T16:34:20.000+02:00
diff --git a/apps/gdalalg_vector_create.cpp b/apps/gdalalg_vector_create.cpp
@@ -47,11 +47,7 @@ GDALVectorCreateAlgorithm::GDALVectorCreateAlgorithm(bool standaloneStep)
               // Note: this is required despite SetAddDefaultArguments(false)
               .SetAddUpsertArgument(false)
               .SetAddSkipErrorsArgument(false)
-              .SetAddAppendLayerArgument(false)
-
-              ),
-      m_crs(), m_geometryType(), m_geometryFieldName("geom"),
-      m_schemaJsonOrPath(), m_fieldDefinitions(), m_fieldStrDefinitions()
+              .SetAddAppendLayerArgument(false))
 {
 
     AddVectorInputArgs(false);
@@ -63,16 +59,17 @@ GDALVectorCreateAlgorithm::GDALVectorCreateAlgorithm(bool standaloneStep)
     auto &geomFieldNameArg =
         AddArg("geometry-field", 0,
                _("Name of the geometry field to create (if supported by the "
-                 "output "
-                 "format)"),
+                 "output format)"),
                &m_geometryFieldName)
             .SetMetaVar("GEOMETRY-FIELD")
-            .SetDefault("geom");
+            .SetDefault(m_geometryFieldName);
 
     AddArg("crs", 0, _("Set CRS"), &m_crs)
         .AddHiddenAlias("srs")
         .SetIsCRSArg(/*noneAllowed=*/false);
 
+    AddArg("fid", 0, _("FID column name"), &m_fidColumnName);
+
     constexpr auto inputMutexGroup = "like-schema-field";
 
     // Apply mutex to GDAL_ARG_NAME_INPUT
@@ -104,19 +101,12 @@ GDALVectorCreateAlgorithm::GDALVectorCreateAlgorithm(bool standaloneStep)
                 ((!m_geometryFieldName.empty() &&
                   geomFieldNameArg.IsExplicitlySet()) ||
                  !m_geometryType.empty() || !m_fieldDefinitions.empty() ||
-                 !m_crs.empty()))
+                 !m_crs.empty() || !m_fidColumnName.empty()))
             {
                 ReportError(CE_Failure, CPLE_AppDefined,
                             "When --schema or --like is specified, "
-                            "--geometry-field, --geometry-type, "
-                            "--field and --crs options must not be specified.");
-                return false;
-            }
-            if (!m_geometryType.empty() && m_crs.empty())
-            {
-                ReportError(CE_Failure, CPLE_AppDefined,
-                            "When --geometry-type is specified, --crs must "
-                            "also be specified");
+                            "--geometry-field, --geometry-type, --field, "
+                            "--crs and --fid options must not be specified.");
                 return false;
             }
             return true;
@@ -131,8 +121,9 @@ bool GDALVectorCreateAlgorithm::RunStep(GDALPipelineStepRunContext &)
 {
 
     const std::string &datasetName = m_outputDataset.GetName();
-    auto outputLayerName =
-        m_outputLayerName.empty() ? datasetName : m_outputLayerName;
+    const std::string outputLayerName =
+        m_outputLayerName.empty() ? CPLGetBasenameSafe(datasetName.c_str())
+                                  : m_outputLayerName;
 
     std::unique_ptr<GDALDataset> poDstDS;
     poDstDS.reset(GDALDataset::Open(datasetName.c_str(),
@@ -265,13 +256,6 @@ bool GDALVectorCreateAlgorithm::RunStep(GDALPipelineStepRunContext &)
         return false;
     }
 
-    if (EQUAL(poDstDriver->GetDescription(), "ESRI Shapefile") &&
-        EQUAL(CPLGetExtensionSafe(poDstDS->GetDescription()).c_str(), "shp") &&
-        poDstDS->GetLayerCount() <= 1)
-    {
-        outputLayerName = CPLGetBasenameSafe(outputLayerName.c_str());
-    }
-
     // An OGR_SCHEMA has been provided
     if (!oSchemaOverride.GetLayerOverrides().empty())
     {
@@ -353,6 +337,7 @@ bool GDALVectorCreateAlgorithm::RunStep(GDALPipelineStepRunContext &)
                                              : userSpecifiedNewName;
 
             if (!CreateLayer(poDstDS.get(), outputLayerNewName,
+                             oLayerOverride.GetFIDColumnName(),
                              oLayerOverride.GetFieldDefinitions(),
                              oLayerOverride.GetGeomFieldDefinitions()))
             {
@@ -370,7 +355,8 @@ bool GDALVectorCreateAlgorithm::RunStep(GDALPipelineStepRunContext &)
         {
             const OGRwkbGeometryType eDstType =
                 OGRFromOGCGeomType(m_geometryType.c_str());
-            if (eDstType == wkbUnknown)
+            if (eDstType == wkbUnknown &&
+                !STARTS_WITH_CI(m_geometryType.c_str(), "GEOMETRY"))
             {
                 ReportError(CE_Failure, CPLE_AppDefined,
                             "Unsupported geometry type: '%s'.",
@@ -403,16 +389,8 @@ bool GDALVectorCreateAlgorithm::RunStep(GDALPipelineStepRunContext &)
             }
         }
 
-        if (EQUAL(poDstDriver->GetDescription(), "ESRI Shapefile") &&
-            EQUAL(CPLGetExtensionSafe(poDstDS->GetDescription()).c_str(),
-                  "shp") &&
-            poDstDS->GetLayerCount() <= 1)
-        {
-            outputLayerName = CPLGetBasenameSafe(poDstDS->GetDescription());
-        }
-
-        if (!CreateLayer(poDstDS.get(), outputLayerName, GetOutputFields(),
-                         geometryFieldDefinitions))
+        if (!CreateLayer(poDstDS.get(), outputLayerName, m_fidColumnName,
+                         GetOutputFields(), geometryFieldDefinitions))
         {
             ReportError(CE_Failure, CPLE_AppDefined,
                         "Cannot create layer '%s'.", outputLayerName.c_str());
@@ -452,6 +430,7 @@ std::vector<OGRFieldDefn> GDALVectorCreateAlgorithm::GetOutputFields() const
 /************************************************************************/
 bool GDALVectorCreateAlgorithm::CreateLayer(
     GDALDataset *poDstDS, const std::string &layerName,
+    const std::string &fidColumnName,
     const std::vector<OGRFieldDefn> &fieldDefinitions,
     const std::vector<OGRGeomFieldDefn> &geometryFieldDefinitions) const
 {
@@ -517,10 +496,19 @@ bool GDALVectorCreateAlgorithm::CreateLayer(
 
     if (!poDstLayer)
     {
-
-        poDstLayer = poDstDS->CreateLayer(
-            layerName.c_str(), poGeomFieldDefn.get(),
-            CPLStringList(GetLayerCreationOptions()).List());
+        CPLStringList aosCreationOptions(GetLayerCreationOptions());
+        if (aosCreationOptions.FetchNameValue("FID") == nullptr &&
+            !fidColumnName.empty())
+        {
+            auto poDstDriver = poDstDS->GetDriver();
+            if (poDstDriver && poDstDriver->HasLayerCreationOption("FID"))
+            {
+                aosCreationOptions.SetNameValue("FID", fidColumnName.c_str());
+            }
+        }
+        poDstLayer =
+            poDstDS->CreateLayer(layerName.c_str(), poGeomFieldDefn.get(),
+                                 aosCreationOptions.List());
     }
 
     if (!poDstLayer)
diff --git a/apps/gdalalg_vector_create.h b/apps/gdalalg_vector_create.h
@@ -63,15 +63,17 @@ class GDALVectorCreateAlgorithm /* non final */
      */
     bool CreateLayer(
         GDALDataset *poDstDS, const std::string &layerName,
+        const std::string &fidColumnName,
         const std::vector<OGRFieldDefn> &fieldDefinitions,
         const std::vector<OGRGeomFieldDefn> &geometryFieldDefinitions) const;
 
-    std::string m_crs;
-    std::string m_geometryType;
-    std::string m_geometryFieldName;
-    std::string m_schemaJsonOrPath;
-    std::vector<OGRFieldDefn> m_fieldDefinitions;
-    std::vector<std::string> m_fieldStrDefinitions;
+    std::string m_crs{};
+    std::string m_fidColumnName{};
+    std::string m_geometryType{};
+    std::string m_geometryFieldName{"geom"};
+    std::string m_schemaJsonOrPath{};
+    std::vector<OGRFieldDefn> m_fieldDefinitions{};
+    std::vector<std::string> m_fieldStrDefinitions{};
 };
 
 /************************************************************************/
diff --git a/autotest/utilities/test_gdalalg_vector_create.py b/autotest/utilities/test_gdalalg_vector_create.py
@@ -485,15 +485,19 @@ def test_gdalalg_vector_create_layer_names(
         ),
         (
             {"schema": r"{}", "geometry-type": "Point"},
-            " When --schema or --like is specified, --geometry-field, --geometry-type, --field and --crs options must not be specified.",
+            "When --schema or --like is specified, .*options must not be specified.",
         ),
         (
             {"schema": r"{}", "geometry-field": "geomm"},
-            " When --schema or --like is specified, --geometry-field, --geometry-type, --field and --crs options must not be specified.",
+            "When --schema or --like is specified, .*options must not be specified.",
         ),
         (
             {"schema": r"{}", "crs": "EPSG:4326"},
-            " When --schema or --like is specified, --geometry-field, --geometry-type, --field and --crs options must not be specified.",
+            "When --schema or --like is specified, .*options must not be specified.",
+        ),
+        (
+            {"schema": r"{}", "fid": "fid"},
+            "When --schema or --like is specified, .*options must not be specified.",
         ),
     ],
 )
@@ -612,3 +616,48 @@ def test_gdalalg_vector_create_datetime_timezone_round_trip(tmp_vsimem):
     assert (
         json_field_defn["timezone"] == "+11:30"
     ), "Unexpected timezone information in exported schema"
+
+
+@pytest.mark.require_driver("GPKG")
+def test_gdalalg_vector_create_geometry_field_without_crs(tmp_vsimem):
+
+    with gdal.alg.vector.create(
+        output=tmp_vsimem / "out.gpkg", geometry_type="GEOMETRY"
+    ) as alg:
+        out_ds = alg.Output()
+        out_lyr = out_ds.GetLayer(0)
+        assert out_lyr.GetName() == "out"
+        assert out_lyr.GetFIDColumn() == "fid"
+        assert out_lyr.GetGeometryColumn() == "geom"
+        assert out_lyr.GetGeomType() == ogr.wkbUnknown
+        assert out_lyr.GetSpatialRef() is None
+        assert out_lyr.GetLayerDefn().GetFieldCount() == 0
+
+
+@pytest.mark.require_driver("GPKG")
+def test_gdalalg_vector_create_implicit_fid_from_source(tmp_vsimem):
+
+    src_ds = gdal.GetDriverByName("GPKG").CreateVector(tmp_vsimem / "src.gpkg")
+    src_ds.CreateLayer("test", options=["FID=my_fid"])
+
+    with gdal.alg.vector.create(input=src_ds, output=tmp_vsimem / "out.gpkg") as alg:
+        out_ds = alg.Output()
+        out_lyr = out_ds.GetLayer(0)
+        assert out_lyr.GetName() == "test"
+        assert out_lyr.GetFIDColumn() == "my_fid"
+        assert out_lyr.GetGeometryColumn() == "geom"
+        assert out_lyr.GetGeomType() == ogr.wkbUnknown
+        assert out_lyr.GetSpatialRef() is None
+        assert out_lyr.GetLayerDefn().GetFieldCount() == 0
+
+
+@pytest.mark.require_driver("GPKG")
+def test_gdalalg_vector_create_explicit_fid(tmp_vsimem):
+
+    with gdal.alg.vector.create(output=tmp_vsimem / "out.gpkg", fid="my_fid") as alg:
+        out_ds = alg.Output()
+        out_lyr = out_ds.GetLayer(0)
+        assert out_lyr.GetName() == "out"
+        assert out_lyr.GetFIDColumn() == "my_fid"
+        assert out_lyr.GetLayerDefn().GetFieldCount() == 0
+        assert out_lyr.GetLayerDefn().GetGeomFieldCount() == 0
diff --git a/doc/source/programs/gdal_vector_create.rst b/doc/source/programs/gdal_vector_create.rst
@@ -118,6 +118,13 @@ Program-Specific Options
 
    Mutually exclusive with :option:`--schema` and with :option:`--like`.
 
+.. option:: --fid <FID>
+
+   Defines the name of the Feature Identifier (FID) column, for drivers that
+   support setting it (those which declare a ``FID`` layer creation option)
+
+   Mutually exclusive with :option:`--schema` and with :option:`--like`.
+
 
 Standard Options
 ----------------
@@ -151,4 +158,4 @@ Examples
     :title: Create a new vector dataset with a layer named `countries_new` based on the layer `countries` of an existing dataset
 
     .. command-output:: gdal vector create --like ../data/poly.gpkg --input-layer poly --output-layer areas_new ./areas.gpkg
-        :cwd: ../../data
+        :cwd: ../../data
diff --git a/ogr/ogr_schema_override.cpp b/ogr/ogr_schema_override.cpp
@@ -66,6 +66,9 @@ bool OGRSchemaOverride::LoadFromJSON(const std::string &osJSON,
                     oLayerOverride.SetLayerName(osLayerName);
                     oLayerOverride.SetFullOverride(bSchemaFullOverride);
 
+                    oLayerOverride.SetFIDColumnName(
+                        oLayer.GetString("fidColumnName"));
+
                     if (oLayerFields.Size() > 0 && !osLayerName.empty())
                     {
                         for (const auto &oField : oLayerFields)
@@ -348,19 +351,21 @@ bool OGRSchemaOverride::LoadFromJSON(const std::string &osJSON,
                             const auto osGeomFieldName =
                                 oGeometryField.GetString("name");
                             oGeomFieldOverride.SetFieldName(osGeomFieldName);
-                            const CPLString oGeometryType(
+                            const CPLString osGeometryType(
                                 CPLString(oGeometryField.GetString("type"))
                                     .tolower());
-                            if (!oGeometryType.empty())
+                            if (!osGeometryType.empty())
                             {
                                 const OGRwkbGeometryType eType =
-                                    OGRFromOGCGeomType(oGeometryType.c_str());
-                                if (eType == wkbUnknown)
+                                    OGRFromOGCGeomType(osGeometryType.c_str());
+                                if (eType == wkbUnknown &&
+                                    !cpl::starts_with(osGeometryType,
+                                                      "geometry"))
                                 {
                                     CPLError(CE_Failure, CPLE_AppDefined,
                                              "Unsupported geometry field type: "
                                              "%s for geometry field %s",
-                                             oGeometryType.c_str(),
+                                             osGeometryType.c_str(),
                                              osGeomFieldName.c_str());
                                     return false;
                                 }
@@ -632,6 +637,12 @@ void OGRLayerSchemaOverride::SetLayerName(const std::string &osLayerName)
     m_osLayerName = osLayerName;
 }
 
+void OGRLayerSchemaOverride::SetFIDColumnName(
+    const std::string &osFIDColumnName)
+{
+    m_osFIDColumnName = osFIDColumnName;
+}
+
 void OGRLayerSchemaOverride::AddNamedFieldOverride(
     const std::string &osFieldName, const OGRFieldDefnOverride &oFieldOverride)
 {
@@ -649,6 +660,11 @@ const std::string &OGRLayerSchemaOverride::GetLayerName() const
     return m_osLayerName;
 }
 
+const std::string &OGRLayerSchemaOverride::GetFIDColumnName() const
+{
+    return m_osFIDColumnName;
+}
+
 const std::map<std::string, OGRFieldDefnOverride> &
 OGRLayerSchemaOverride::GetNamedFieldOverrides() const
 {
@@ -706,9 +722,9 @@ void OGRLayerSchemaOverride::SetFullOverride(bool bIsFullOverride)
 
 bool OGRLayerSchemaOverride::IsValid() const
 {
-    bool isValid =
-        !m_osLayerName.empty() &&
-        (!m_oNamedFieldOverrides.empty() || !m_aoUnnamedFieldOverrides.empty());
+    bool isValid = !m_osLayerName.empty() &&
+                   (!m_oNamedFieldOverrides.empty() ||
+                    !m_aoUnnamedFieldOverrides.empty() || m_bIsFullOverride);
     for (const auto &oFieldOverrideIter : m_oNamedFieldOverrides)
     {
         isValid &= !oFieldOverrideIter.first.empty();
diff --git a/ogr/ogr_schema_override.h b/ogr/ogr_schema_override.h
@@ -261,13 +261,17 @@ class CPL_DLL OGRLayerSchemaOverride
 
     void SetLayerName(const std::string &osLayerName);
 
+    void SetFIDColumnName(const std::string &osFIDColumnName);
+
     void AddNamedFieldOverride(const std::string &osFieldName,
                                const OGRFieldDefnOverride &oFieldOverride);
 
     void AddUnnamedFieldOverride(const OGRFieldDefnOverride &oFieldOverride);
 
     const std::string &GetLayerName() const;
 
+    const std::string &GetFIDColumnName() const;
+
     const std::map<std::string, OGRFieldDefnOverride> &
     GetNamedFieldOverrides() const;
 
@@ -293,6 +297,7 @@ class CPL_DLL OGRLayerSchemaOverride
 
   private:
     std::string m_osLayerName{};
+    std::string m_osFIDColumnName{};
     std::map<std::string, OGRFieldDefnOverride> m_oNamedFieldOverrides{};
     std::vector<OGRFieldDefnOverride> m_aoUnnamedFieldOverrides{};
     std::vector<OGRGeomFieldDefnOverride> m_aoGeomFieldOverrides{};
