PDF driver: add 'gdal driver pdf list-layers'

Author: rouault

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: 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");