Merge pull request #12126 from rouault/plugin_algorithm

Add a mechanism to register external algorithm, in particular for drivers

Author: rouault

Diff for: apps/data/gdal_algorithm.schema.json

@@ -79,8 +79,6 @@
       },
       "required": [
         "description",
-        "short_url",
-        "url",
         "sub_algorithms",
         "input_arguments",
         "output_arguments",

Diff for: autotest/cpp/test_gdal_algorithm.cpp

@@ -3696,4 +3696,80 @@ TEST_F(test_gdal_algorithm, raster_edit_failures_unset_metadata)
                  "edit: SetMetadataItem('foo', NULL) failed");
 }
 
+TEST_F(test_gdal_algorithm, register_plugin_algorithms)
+{
+    auto &singleton = GDALGlobalAlgorithmRegistry::GetSingleton();
+    bool flag = false;
+    singleton.DeclareAlgorithm(
+        {"foo", "bar"},
+        [&flag]() -> std::unique_ptr<GDALAlgorithm>
+        {
+            flag = true;
+            return std::make_unique<GDALContainerAlgorithm>("dummy");
+        });
+
+    {
+        EXPECT_NE(singleton.Instantiate("foo"), nullptr);
+        EXPECT_FALSE(flag);
+    }
+
+    {
+        auto got = singleton.GetDeclaredSubAlgorithmNames({"gdal"});
+        EXPECT_TRUE(std::find(got.begin(), got.end(), "foo") != got.end());
+        EXPECT_FALSE(flag);
+    }
+
+    {
+        auto got = singleton.GetDeclaredSubAlgorithmNames({"gdal", "foo"});
+        EXPECT_TRUE(std::find(got.begin(), got.end(), "bar") != got.end());
+        EXPECT_TRUE(flag);
+        flag = false;
+    }
+
+    {
+        auto got =
+            singleton.GetDeclaredSubAlgorithmNames({"gdal", "foo", "bar"});
+        EXPECT_TRUE(got.empty());
+        EXPECT_FALSE(flag);
+    }
+
+    {
+        auto got = singleton.GetDeclaredSubAlgorithmNames({"gdal", "bar"});
+        EXPECT_TRUE(got.empty());
+        EXPECT_FALSE(flag);
+    }
+
+    {
+        auto alg = singleton.InstantiateDeclaredSubAlgorithm({"gdal", "foo"});
+        ASSERT_NE(alg, nullptr);
+        EXPECT_TRUE(alg->HasSubAlgorithms());
+        EXPECT_EQ(alg->GetSubAlgorithmNames().size(), 1);
+        EXPECT_TRUE(flag);
+        flag = false;
+    }
+
+    {
+        auto alg =
+            singleton.InstantiateDeclaredSubAlgorithm({"gdal", "foo", "bar"});
+        ASSERT_NE(alg, nullptr);
+        EXPECT_TRUE(flag);
+        flag = false;
+    }
+
+    {
+        auto alg = singleton.Instantiate("foo")->InstantiateSubAlgorithm("bar");
+        ASSERT_NE(alg, nullptr);
+        EXPECT_TRUE(flag);
+    }
+
+    {
+        auto alg = singleton.InstantiateDeclaredSubAlgorithm({"gdal", "bar"});
+        ASSERT_EQ(alg, nullptr);
+    }
+
+    singleton.DeclareAlgorithm({"foo", "bar"},
+                               []() -> std::unique_ptr<GDALAlgorithm>
+                               { return nullptr; });
+}
+
 }  // namespace test_gdal_algorithm

Diff for: autotest/gcore/driver_algorithms.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env pytest
+# -*- coding: utf-8 -*-
+###############################################################################
+# Project:  GDAL/OGR Test Suite
+# Purpose:  Check that drivers that declare algorithms actually implement them
+# Author:   Even Rouault <even dot rouault @ spatialys.com>
+#
+###############################################################################
+# Copyright (c) 2025, Even Rouault <even dot rouault at spatialys.com>
+#
+# SPDX-License-Identifier: MIT
+###############################################################################
+
+
+from osgeo import gdal
+
+
+def check(alg, expected_name):
+    assert alg.GetName() == expected_name
+    if alg.HasSubAlgorithms():
+        for subalg_name in alg.GetSubAlgorithmNames():
+            subalg = alg[subalg_name]
+            assert subalg, subalg_name
+            check(subalg, subalg_name)
+
+            assert alg.InstantiateSubAlgorithm("i_do_not_exist") is None
+
+
+def test_driver_algorithms():
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]
+    if alg:
+        check(alg, "driver")

Diff for: autotest/gdrivers/pdf.py

@@ -12,6 +12,7 @@
 # SPDX-License-Identifier: MIT
 ###############################################################################
 
+import json
 import os
 import shutil
 import sys
@@ -3097,3 +3098,45 @@ def test_pdf_iso32000_invalid_srs():
 
     # Just test that this does not crash
     gdal.Open("data/pdf/invalid_srs.pdf")
+
+
+###############################################################################
+#
+
+
+@gdaltest.enable_exceptions()
+@pytest.mark.skipif(not have_read_support(), reason="no read support available")
+def test_pdf_gdal_driver_pdf_list_layer():
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]
+    assert alg.GetName() == "driver"
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["pdf"]
+    assert alg.GetName() == "pdf"
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["pdf"]["list-layers"]
+    assert alg.GetName() == "list-layers"
+    alg["input"] = "data/pdf/adobe_style_geospatial.pdf"
+    assert alg.Run()
+    assert json.loads(alg["output-string"]) == [
+        "New_Data_Frame",
+        "New_Data_Frame.Graticule",
+        "Layers",
+        "Layers.Measured_Grid",
+        "Layers.Graticule",
+    ]
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["pdf"]["list-layers"]
+    alg["input"] = "data/pdf/adobe_style_geospatial.pdf"
+    alg["output-format"] = "text"
+    assert alg.Run()
+    assert (
+        alg["output-string"]
+        == "New_Data_Frame\nNew_Data_Frame.Graticule\nLayers\nLayers.Measured_Grid\nLayers.Graticule\n"
+    )
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["pdf"]["list-layers"]
+    assert alg.GetName() == "list-layers"
+    alg["input"] = "data/byte.tif"
+    with pytest.raises(Exception, match="is not a PDF"):
+        alg.Run()

Diff for: autotest/ogr/ogr_openfilegdb_write.py

@@ -4601,3 +4601,45 @@ def test_ogr_openfilegdb_write_check_golden_file(tmp_path, src_directory):
         assert (
             open(src_filename, "rb").read() == open(out_filename, "rb").read()
         ), filename
+
+
+###############################################################################
+# Test 'gdal driver openfilegdb repack'
+
+
+@gdaltest.enable_exceptions()
+def test_ogropenfilegdb_write_gdal_driver_openfilegdb_repack(tmp_path):
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]
+    assert alg.GetName() == "driver"
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["openfilegdb"]
+    assert alg.GetName() == "openfilegdb"
+
+    out_directory = str(tmp_path / "test.gdb")
+    gdal.VectorTranslate(
+        out_directory, "data/openfilegdb/polygon_golden.gdb", format="OpenFileGDB"
+    )
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["openfilegdb"]["repack"]
+    assert alg.GetName() == "repack"
+    alg["input"] = "data/poly.shp"
+    with pytest.raises(Exception, match="is not a FileGeoDatabase"):
+        alg.Run()
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["openfilegdb"]["repack"]
+    assert alg.GetName() == "repack"
+    alg["input"] = out_directory
+    assert alg.Run()
+
+    last_pct = [0]
+
+    def my_progress(pct, msg, user_data):
+        last_pct[0] = pct
+        return True
+
+    alg = gdal.GetGlobalAlgorithmRegistry()["driver"]["openfilegdb"]["repack"]
+    assert alg.GetName() == "repack"
+    alg["input"] = out_directory
+    assert alg.Run(my_progress)
+    assert last_pct[0] == 1.0

Diff for: frmts/pdf/pdfdataset.cpp

@@ -19,10 +19,12 @@
 
 #include "gdal_pdf.h"
 
+#include "cpl_json_streaming_writer.h"
 #include "cpl_vsi_virtual.h"
 #include "cpl_spawn.h"
 #include "cpl_string.h"
 #include "gdal_frmts.h"
+#include "gdalalgorithm.h"
 #include "ogr_spatialref.h"
 #include "ogr_geometry.h"
 
@@ -7711,6 +7713,88 @@ CPLString PDFSanitizeLayerName(const char *pszName)
     return osName;
 }
 
+/************************************************************************/
+/*                    GDALPDFListLayersAlgorithm                        */
+/************************************************************************/
+
+#ifdef HAVE_PDF_READ_SUPPORT
+
+class GDALPDFListLayersAlgorithm final : public GDALAlgorithm
+{
+  public:
+    GDALPDFListLayersAlgorithm()
+        : GDALAlgorithm("list-layers",
+                        std::string("List layers of a PDF dataset"),
+                        "/drivers/raster/pdf.html")
+    {
+        AddInputDatasetArg(&m_dataset, GDAL_OF_RASTER | GDAL_OF_VECTOR);
+        AddOutputFormatArg(&m_format).SetDefault(m_format).SetChoices("json",
+                                                                      "text");
+        AddOutputStringArg(&m_output);
+    }
+
+  protected:
+    bool RunImpl(GDALProgressFunc, void *) override
+    {
+        auto poDS = dynamic_cast<PDFDataset *>(m_dataset.GetDatasetRef());
+        if (!poDS)
+        {
+            ReportError(CE_Failure, CPLE_AppDefined, "%s is not a PDF",
+                        m_dataset.GetName().c_str());
+            return false;
+        }
+        if (m_format == "json")
+        {
+            CPLJSonStreamingWriter oWriter(nullptr, nullptr);
+            oWriter.StartArray();
+            for (const auto &[key, value] : cpl::IterateNameValue(
+                     const_cast<CSLConstList>(poDS->GetMetadata("LAYERS"))))
+            {
+                CPL_IGNORE_RET_VAL(key);
+                oWriter.Add(value);
+            }
+            oWriter.EndArray();
+            m_output = oWriter.GetString();
+            m_output += '\n';
+        }
+        else
+        {
+            for (const auto &[key, value] : cpl::IterateNameValue(
+                     const_cast<CSLConstList>(poDS->GetMetadata("LAYERS"))))
+            {
+                CPL_IGNORE_RET_VAL(key);
+                m_output += value;
+                m_output += '\n';
+            }
+        }
+        return true;
+    }
+
+  private:
+    GDALArgDatasetValue m_dataset{};
+    std::string m_format = "json";
+    std::string m_output{};
+};
+
+/************************************************************************/
+/*                    GDALPDFInstantiateAlgorithm()                     */
+/************************************************************************/
+
+static GDALAlgorithm *
+GDALPDFInstantiateAlgorithm(const std::vector<std::string> &aosPath)
+{
+    if (aosPath.size() == 1 && aosPath[0] == "list-layers")
+    {
+        return std::make_unique<GDALPDFListLayersAlgorithm>().release();
+    }
+    else
+    {
+        return nullptr;
+    }
+}
+
+#endif  // HAVE_PDF_READ_SUPPORT
+
 /************************************************************************/
 /*                         GDALRegister_PDF()                           */
 /************************************************************************/
@@ -7729,6 +7813,7 @@ void GDALRegister_PDF()
 
 #ifdef HAVE_PDF_READ_SUPPORT
     poDriver->pfnOpen = PDFDataset::OpenWrapper;
+    poDriver->pfnInstantiateAlgorithm = GDALPDFInstantiateAlgorithm;
 #endif  // HAVE_PDF_READ_SUPPORT
 
     poDriver->pfnCreateCopy = GDALPDFCreateCopy;

Diff for: frmts/pdf/pdfdrivercore.cpp

@@ -234,6 +234,8 @@ void PDFDriverSetCommonMetadata(GDALDriver *poDriver)
     poDriver->SetMetadataItem(GDAL_DCAP_UPDATE, "YES");
     poDriver->SetMetadataItem(GDAL_DMD_UPDATE_ITEMS,
                               "GeoTransform SRS GCPs DatasetMetadata");
+
+    poDriver->DeclareAlgorithm({"list-layers"});
 #endif
 
     poDriver->SetMetadataItem(GDAL_DCAP_CREATE, "YES");

Diff for: gcore/gdal_priv.h

@@ -34,6 +34,7 @@ class GDALProxyDataset;
 class GDALProxyRasterBand;
 class GDALAsyncReader;
 class GDALRelationship;
+class GDALAlgorithm;
 
 /* -------------------------------------------------------------------- */
 /*      Pull in the public declarations.  This gets the C apis, and     */
@@ -2368,6 +2369,26 @@ class CPL_DLL GDALDriver : public GDALMajorObject
     GDALSubdatasetInfo *(*pfnGetSubdatasetInfoFunc)(const char *pszFileName) =
         nullptr;
 
+    typedef GDALAlgorithm *(*InstantiateAlgorithmCallback)(
+        const std::vector<std::string> &aosPath);
+    InstantiateAlgorithmCallback pfnInstantiateAlgorithm = nullptr;
+
+    virtual InstantiateAlgorithmCallback GetInstantiateAlgorithmCallback()
+    {
+        return pfnInstantiateAlgorithm;
+    }
+
+    /** Instantiate an algorithm by its full path (omitting leading "gdal").
+     * For example {"driver", "pdf", "list-layers"}
+     */
+    GDALAlgorithm *
+    InstantiateAlgorithm(const std::vector<std::string> &aosPath);
+
+    /** Declare an algorithm by its full path (omitting leading "gdal").
+     * For example {"driver", "pdf", "list-layers"}
+     */
+    void DeclareAlgorithm(const std::vector<std::string> &aosPath);
+
     //! @endcond
 
     /* -------------------------------------------------------------------- */
@@ -2517,6 +2538,8 @@ class GDALPluginDriverProxy : public GDALDriver
     RenameCallback GetRenameCallback() override;
 
     CopyFilesCallback GetCopyFilesCallback() override;
+
+    InstantiateAlgorithmCallback GetInstantiateAlgorithmCallback() override;
     //! @endcond
 
     CPLErr SetMetadataItem(const char *pszName, const char *pszValue,