LCOV - code coverage report
Current view: top level - ogr/ogrsf_frmts/arrow_common - ogr_arrow.h (source / functions) Hit Total Coverage
Test: gdal_filtered.info Lines: 36 44 81.8 %
Date: 2025-09-10 17:48:50 Functions: 15 19 78.9 %

          Line data    Source code
       1             : /******************************************************************************
       2             :  *
       3             :  * Project:  Arrow generic code
       4             :  * Purpose:  Arrow generic code
       5             :  * Author:   Even Rouault, <even.rouault at spatialys.com>
       6             :  *
       7             :  ******************************************************************************
       8             :  * Copyright (c) 2022, Planet Labs
       9             :  *
      10             :  * SPDX-License-Identifier: MIT
      11             :  ****************************************************************************/
      12             : 
      13             : #ifndef OGR_ARROW_H
      14             : #define OGR_ARROW_H
      15             : 
      16             : #include "gdal_pam.h"
      17             : #include "ogrsf_frmts.h"
      18             : #include "ogrlayerarrow.h"
      19             : 
      20             : #include <map>
      21             : #include <set>
      22             : 
      23             : #include "ogr_include_arrow.h"
      24             : 
      25             : #if defined(__clang__)
      26             : #pragma clang diagnostic push
      27             : #pragma clang diagnostic ignored "-Wweak-vtables"
      28             : #endif
      29             : 
      30             : enum class OGRArrowGeomEncoding
      31             : {
      32             :     WKB,
      33             :     WKT,
      34             : 
      35             :     // F(ixed) S(ize) L(ist) of (x,y[,z][,m]) values / Interleaved layout
      36             :     GEOARROW_FSL_GENERIC,  // only used by OGRArrowWriterLayer::m_eGeomEncoding
      37             :     GEOARROW_FSL_POINT,
      38             :     GEOARROW_FSL_LINESTRING,
      39             :     GEOARROW_FSL_POLYGON,
      40             :     GEOARROW_FSL_MULTIPOINT,
      41             :     GEOARROW_FSL_MULTILINESTRING,
      42             :     GEOARROW_FSL_MULTIPOLYGON,
      43             : 
      44             :     // Struct of (x,y,[,z][,m])
      45             :     GEOARROW_STRUCT_GENERIC,  // only used by OGRArrowWriterLayer::m_eGeomEncoding
      46             :     GEOARROW_STRUCT_POINT,
      47             :     GEOARROW_STRUCT_LINESTRING,
      48             :     GEOARROW_STRUCT_POLYGON,
      49             :     GEOARROW_STRUCT_MULTIPOINT,
      50             :     GEOARROW_STRUCT_MULTILINESTRING,
      51             :     GEOARROW_STRUCT_MULTIPOLYGON,
      52             : };
      53             : 
      54             : /************************************************************************/
      55             : /*                        OGRArrowIsGeoArrowStruct()                    */
      56             : /************************************************************************/
      57             : 
      58         260 : inline bool OGRArrowIsGeoArrowStruct(OGRArrowGeomEncoding eEncoding)
      59             : {
      60         260 :     switch (eEncoding)
      61             :     {
      62         260 :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_GENERIC:
      63             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_POINT:
      64             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_LINESTRING:
      65             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_POLYGON:
      66             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_MULTIPOINT:
      67             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_MULTILINESTRING:
      68             :         case OGRArrowGeomEncoding::GEOARROW_STRUCT_MULTIPOLYGON:
      69         260 :             return true;
      70             : 
      71           0 :         default:
      72           0 :             return false;
      73             :     }
      74             : }
      75             : 
      76             : /************************************************************************/
      77             : /*                         OGRArrowLayer                                */
      78             : /************************************************************************/
      79             : 
      80             : class OGRArrowDataset;
      81             : 
      82             : class OGRArrowLayer CPL_NON_FINAL
      83             :     : public OGRLayer,
      84             :       public OGRGetNextFeatureThroughRaw<OGRArrowLayer>
      85             : {
      86             :   public:
      87             :     struct Constraint
      88             :     {
      89             :         enum class Type
      90             :         {
      91             :             Integer,
      92             :             Integer64,
      93             :             Real,
      94             :             String,
      95             :         };
      96             :         int iField = -1;      // index to a OGRFeatureDefn OGRField
      97             :         int iArrayIdx = -1;   // index to m_poBatchColumns
      98             :         int nOperation = -1;  // SWQ_xxxx
      99             :         Type eType{};
     100             :         OGRField sValue{};
     101             :         std::string osValue{};
     102             :     };
     103             : 
     104             :   private:
     105             :     OGRArrowLayer(const OGRArrowLayer &) = delete;
     106             :     OGRArrowLayer &operator=(const OGRArrowLayer &) = delete;
     107             : 
     108             :     int m_nUseOptimizedAttributeFilter = -1;
     109             :     bool m_bSpatialFilterIntersectsLayerExtent = true;
     110             :     bool m_bUseRecordBatchBaseImplementation = false;
     111             : 
     112             :     // Modified by UseRecordBatchBaseImplementation()
     113             :     mutable struct ArrowSchema m_sCachedSchema = {};
     114             : 
     115             :     bool SkipToNextFeatureDueToAttributeFilter() const;
     116             :     void ExploreExprNode(const swq_expr_node *poNode);
     117             :     bool UseRecordBatchBaseImplementation() const;
     118             : 
     119             :     template <typename SourceOffset>
     120             :     static struct ArrowArray *
     121             :     CreateWKBArrayFromWKTArray(const struct ArrowArray *sourceArray);
     122             : 
     123             :     int GetArrowSchemaInternal(struct ArrowSchema *out) const;
     124             : 
     125             :   protected:
     126             :     OGRArrowDataset *m_poArrowDS = nullptr;
     127             :     arrow::MemoryPool *m_poMemoryPool = nullptr;
     128             :     OGRFeatureDefn *m_poFeatureDefn = nullptr;
     129             :     std::shared_ptr<arrow::Schema> m_poSchema{};
     130             :     std::string m_osFIDColumn{};
     131             :     int m_iFIDArrowColumn = -1;
     132             :     std::vector<std::vector<int>> m_anMapFieldIndexToArrowColumn{};
     133             :     std::vector<int> m_anMapGeomFieldIndexToArrowColumn{};
     134             :     std::vector<OGRArrowGeomEncoding> m_aeGeomEncoding{};
     135             : 
     136             :     //! Whether bounding box based spatial filter should be skipped.
     137             :     // This is set to true by OGRParquetDatasetLayer when there is a bounding
     138             :     // box field, as an optimization.
     139             :     bool m_bBaseArrowIgnoreSpatialFilterRect = false;
     140             : 
     141             :     //! Whether spatial filter should be skipped (by GetNextArrowArray())
     142             :     // This is set to true by OGRParquetDatasetLayer when filtering points in
     143             :     // a rectangle.
     144             :     bool m_bBaseArrowIgnoreSpatialFilter = false;
     145             : 
     146             :     //! Describe the bbox column of a geometry column
     147             :     struct GeomColBBOX
     148             :     {
     149             :         bool bIsFloat = false;
     150             :         int iArrowCol = -1;
     151             :         int iArrayIdx = -1;  // only valid when m_bIgnoredFields == true
     152             :         int iArrowSubfieldXMin = -1;
     153             :         int iArrowSubfieldYMin = -1;
     154             :         int iArrowSubfieldXMax = -1;
     155             :         int iArrowSubfieldYMax = -1;
     156             :     };
     157             : 
     158             :     //! Map from OGR geometry field index to GeomColBBOX
     159             :     std::map<int, GeomColBBOX> m_oMapGeomFieldIndexToGeomColBBOX{};
     160             : 
     161             :     const arrow::BinaryArray *m_poArrayWKB = nullptr;
     162             :     const arrow::LargeBinaryArray *m_poArrayWKBLarge = nullptr;
     163             :     const arrow::Array *m_poArrayBBOX = nullptr;
     164             :     const arrow::DoubleArray *m_poArrayXMinDouble = nullptr;
     165             :     const arrow::DoubleArray *m_poArrayYMinDouble = nullptr;
     166             :     const arrow::DoubleArray *m_poArrayXMaxDouble = nullptr;
     167             :     const arrow::DoubleArray *m_poArrayYMaxDouble = nullptr;
     168             :     const arrow::FloatArray *m_poArrayXMinFloat = nullptr;
     169             :     const arrow::FloatArray *m_poArrayYMinFloat = nullptr;
     170             :     const arrow::FloatArray *m_poArrayXMaxFloat = nullptr;
     171             :     const arrow::FloatArray *m_poArrayYMaxFloat = nullptr;
     172             : 
     173             :     //! References values in range [0, m_poSchema->field_count()-1]
     174             :     std::set<int> m_oSetBBoxArrowColumns{};
     175             : 
     176             :     bool m_bIgnoredFields = false;
     177             :     std::vector<int>
     178             :         m_anMapFieldIndexToArrayIndex{};  // only valid when m_bIgnoredFields is
     179             :                                           // set
     180             :     std::vector<int> m_anMapGeomFieldIndexToArrayIndex{};  // only valid when
     181             :         // m_bIgnoredFields is set
     182             :     int m_nRequestedFIDColumn = -1;  // only valid when m_bIgnoredFields is set
     183             : 
     184             :     int m_nExpectedBatchColumns =
     185             :         -1;  // Should be equal to m_poBatch->num_columns() (when
     186             :              // m_bIgnoredFields is set)
     187             : 
     188             :     bool m_bEOF = false;
     189             :     int64_t m_nFeatureIdx = 0;
     190             :     int64_t m_nIdxInBatch = 0;
     191             :     std::map<std::string, CPLJSONObject> m_oMapGeometryColumns{};
     192             :     mutable std::map<int, OGREnvelope> m_oMapExtents{};
     193             :     mutable std::map<int, OGREnvelope3D> m_oMapExtents3D{};
     194             :     int m_iRecordBatch = -1;
     195             :     std::shared_ptr<arrow::RecordBatch> m_poBatch{};
     196             :     // m_poBatch->columns() is a relatively costly operation, so cache its
     197             :     // result
     198             :     std::vector<std::shared_ptr<arrow::Array>>
     199             :         m_poBatchColumns{};  // must always be == m_poBatch->columns()
     200             :     mutable std::shared_ptr<arrow::Array> m_poReadFeatureTmpArray{};
     201             : 
     202             :     std::vector<Constraint> m_asAttributeFilterConstraints{};
     203             : 
     204             :     //! Whether attribute filter should be skipped.
     205             :     // This is set to true by OGRParquetDatasetLayer when it can fully translate
     206             :     // a filter, as an optimization.
     207             :     bool m_bBaseArrowIgnoreAttributeFilter = false;
     208             : 
     209             :     std::map<std::string, std::unique_ptr<OGRFieldDefn>>
     210             :     LoadGDALSchema(const arrow::KeyValueMetadata *kv_metadata);
     211             : 
     212             :     void LoadGDALMetadata(const arrow::KeyValueMetadata *kv_metadata);
     213             : 
     214             :     OGRArrowLayer(OGRArrowDataset *poDS, const char *pszLayerName);
     215             : 
     216             :     virtual std::string GetDriverUCName() const = 0;
     217             :     static bool IsIntegerArrowType(arrow::Type::type typeId);
     218             :     static bool
     219             :     IsHandledListOrMapType(const std::shared_ptr<arrow::DataType> &valueType);
     220             :     static bool
     221             :     IsHandledListType(const std::shared_ptr<arrow::BaseListType> &listType);
     222             :     static bool
     223             :     IsHandledMapType(const std::shared_ptr<arrow::MapType> &mapType);
     224             :     static bool
     225             :     IsValidGeometryEncoding(const std::shared_ptr<arrow::Field> &field,
     226             :                             const std::string &osEncoding,
     227             :                             bool bWarnIfUnknownEncoding,
     228             :                             OGRwkbGeometryType &eGeomTypeOut,
     229             :                             OGRArrowGeomEncoding &eGeomEncodingOut);
     230             :     static OGRwkbGeometryType
     231             :     GetGeometryTypeFromString(const std::string &osType);
     232             :     bool
     233             :     MapArrowTypeToOGR(const std::shared_ptr<arrow::DataType> &type,
     234             :                       const std::shared_ptr<arrow::Field> &field,
     235             :                       OGRFieldDefn &oField, OGRFieldType &eType,
     236             :                       OGRFieldSubType &eSubType, const std::vector<int> &path,
     237             :                       const std::map<std::string, std::unique_ptr<OGRFieldDefn>>
     238             :                           &oMapFieldNameToGDALSchemaFieldDefn);
     239             :     void CreateFieldFromSchema(
     240             :         const std::shared_ptr<arrow::Field> &field,
     241             :         const std::vector<int> &path,
     242             :         const std::map<std::string, std::unique_ptr<OGRFieldDefn>>
     243             :             &oMapFieldNameToGDALSchemaFieldDefn);
     244             :     std::unique_ptr<OGRFieldDomain>
     245             :     BuildDomainFromBatch(const std::string &osDomainName,
     246             :                          const std::shared_ptr<arrow::RecordBatch> &poBatch,
     247             :                          int iCol) const;
     248             :     OGRwkbGeometryType ComputeGeometryColumnTypeProcessBatch(
     249             :         const std::shared_ptr<arrow::RecordBatch> &poBatch, int iGeomCol,
     250             :         int iBatchCol, OGRwkbGeometryType eGeomType) const;
     251             :     static bool ReadWKBBoundingBox(const uint8_t *data, size_t size,
     252             :                                    OGREnvelope &sEnvelope);
     253             :     OGRFeature *ReadFeature(
     254             :         int64_t nIdxInBatch,
     255             :         const std::vector<std::shared_ptr<arrow::Array>> &poColumnArrays) const;
     256             :     OGRGeometry *ReadGeometry(int iGeomField, const arrow::Array *array,
     257             :                               int64_t nIdxInBatch) const;
     258             :     virtual bool ReadNextBatch() = 0;
     259             :     virtual void InvalidateCachedBatches() = 0;
     260             :     OGRFeature *GetNextRawFeature();
     261             : 
     262           0 :     virtual bool CanRunNonForcedGetExtent()
     263             :     {
     264           0 :         return true;
     265             :     }
     266             : 
     267             :     void SetBatch(const std::shared_ptr<arrow::RecordBatch> &poBatch);
     268             : 
     269             :     // Refreshes Constraint.iArrayIdx from iField. To be called by SetIgnoredFields()
     270             :     void ComputeConstraintsArrayIdx();
     271             : 
     272             :     static const swq_expr_node *GetColumnSubNode(const swq_expr_node *poNode);
     273             :     static const swq_expr_node *GetConstantSubNode(const swq_expr_node *poNode);
     274             :     static bool IsComparisonOp(int op);
     275             : 
     276             :     virtual bool FastGetExtent(int iGeomField, OGREnvelope *psExtent) const;
     277             :     bool FastGetExtent3D(int iGeomField, OGREnvelope3D *psExtent) const;
     278             :     static OGRErr GetExtentFromMetadata(const CPLJSONObject &oJSONDef,
     279             :                                         OGREnvelope3D *psExtent);
     280             : 
     281             :     int GetArrowSchema(struct ArrowArrayStream *,
     282             :                        struct ArrowSchema *out) override;
     283             :     int GetNextArrowArray(struct ArrowArrayStream *,
     284             :                           struct ArrowArray *out) override;
     285             : 
     286        4737 :     virtual void IncrFeatureIdx()
     287             :     {
     288        4737 :         ++m_nFeatureIdx;
     289        4737 :     }
     290             : 
     291             :     void SanityCheckOfSetBatch() const;
     292             : 
     293             :   public:
     294             :     ~OGRArrowLayer() override;
     295             : 
     296       13059 :     const OGRFeatureDefn *GetLayerDefn() const override
     297             :     {
     298       13059 :         return m_poFeatureDefn;
     299             :     }
     300             : 
     301             :     void ResetReading() override;
     302             : 
     303        1875 :     const char *GetFIDColumn() const override
     304             :     {
     305        1875 :         return m_osFIDColumn.c_str();
     306             :     }
     307        9103 :     DEFINE_GET_NEXT_FEATURE_THROUGH_RAW(OGRArrowLayer)
     308             :     OGRErr IGetExtent(int iGeomField, OGREnvelope *psExtent,
     309             :                       bool bForce) override;
     310             :     OGRErr IGetExtent3D(int iGeomField, OGREnvelope3D *psExtent,
     311             :                         bool bForce) override;
     312             :     OGRErr SetAttributeFilter(const char *pszFilter) override;
     313             : 
     314             :     OGRErr ISetSpatialFilter(int iGeomField,
     315             :                              const OGRGeometry *poGeom) override;
     316             : 
     317             :     int TestCapability(const char *pszCap) const override;
     318             : 
     319             :     bool GetArrowStream(struct ArrowArrayStream *out_stream,
     320             :                         CSLConstList papszOptions = nullptr) override;
     321             : 
     322             :     virtual std::unique_ptr<OGRFieldDomain>
     323             :     BuildDomain(const std::string &osDomainName, int iFieldIndex) const = 0;
     324             : 
     325             :     static void TimestampToOGR(int64_t timestamp,
     326             :                                const arrow::TimestampType *timestampType,
     327             :                                int nTZFlag, OGRField *psField);
     328             : };
     329             : 
     330             : /************************************************************************/
     331             : /*                         OGRArrowDataset                              */
     332             : /************************************************************************/
     333             : 
     334             : class OGRArrowDataset CPL_NON_FINAL : public GDALPamDataset
     335             : {
     336             :     std::shared_ptr<arrow::MemoryPool> m_poMemoryPool{};
     337             :     std::unique_ptr<OGRArrowLayer> m_poLayer{};
     338             :     std::vector<std::string> m_aosDomainNames{};
     339             :     std::map<std::string, int> m_oMapDomainNameToCol{};
     340             : 
     341             :   public:
     342             :     explicit OGRArrowDataset(
     343             :         const std::shared_ptr<arrow::MemoryPool> &poMemoryPool);
     344             : 
     345        1694 :     ~OGRArrowDataset() override
     346        1694 :     {
     347        1694 :         OGRArrowDataset::Close();
     348        1694 :     }
     349             : 
     350        3386 :     CPLErr Close() override
     351             :     {
     352        3386 :         m_poLayer.reset();
     353        3386 :         m_poMemoryPool.reset();
     354        3386 :         return GDALPamDataset::Close();
     355             :     }
     356             : 
     357        1692 :     inline arrow::MemoryPool *GetMemoryPool() const
     358             :     {
     359        1692 :         return m_poMemoryPool.get();
     360             :     }
     361             : 
     362         609 :     inline const std::shared_ptr<arrow::MemoryPool> &GetSharedMemoryPool() const
     363             :     {
     364         609 :         return m_poMemoryPool;
     365             :     }
     366             : 
     367             :     void SetLayer(std::unique_ptr<OGRArrowLayer> &&poLayer);
     368             : 
     369             :     void RegisterDomainName(const std::string &osDomainName, int iFieldIndex);
     370             : 
     371             :     std::vector<std::string> GetFieldDomainNames(
     372             :         CSLConstList /*papszOptions*/ = nullptr) const override;
     373             :     const OGRFieldDomain *
     374             :     GetFieldDomain(const std::string &name) const override;
     375             : 
     376             :     int GetLayerCount() const override;
     377             :     using GDALDataset::GetLayer;
     378             :     const OGRLayer *GetLayer(int idx) const override;
     379             : };
     380             : 
     381             : /************************************************************************/
     382             : /*                        OGRArrowWriterLayer                           */
     383             : /************************************************************************/
     384             : 
     385             : class OGRArrowWriterLayer CPL_NON_FINAL : public OGRLayer
     386             : 
     387             : {
     388             :   protected:
     389             :     OGRArrowWriterLayer(const OGRArrowWriterLayer &) = delete;
     390             :     OGRArrowWriterLayer &operator=(const OGRArrowWriterLayer &) = delete;
     391             : 
     392             :     arrow::MemoryPool *m_poMemoryPool = nullptr;
     393             :     bool m_bInitializationOK = false;
     394             :     std::shared_ptr<arrow::io::OutputStream> m_poOutputStream{};
     395             :     std::shared_ptr<arrow::Schema> m_poSchema{};
     396             :     OGRFeatureDefn *m_poFeatureDefn = nullptr;
     397             :     std::map<std::string, std::unique_ptr<OGRFieldDomain>> m_oMapFieldDomains{};
     398             :     std::map<std::string, std::shared_ptr<arrow::Array>>
     399             :         m_oMapFieldDomainToStringArray{};
     400             : 
     401             :     bool m_bWriteFieldArrowExtensionName = false;
     402             :     OGRArrowGeomEncoding m_eGeomEncoding = OGRArrowGeomEncoding::WKB;
     403             :     std::vector<OGRArrowGeomEncoding> m_aeGeomEncoding{};
     404             :     int m_nWKTCoordinatePrecision = -1;
     405             : 
     406             :     //! Base struct data type for GeoArrow struct geometry columns.
     407             :     // Constraint: if not empty, m_apoBaseStructGeomType.size() == m_poFeatureDefn->GetGeomFieldCount()
     408             :     std::vector<std::shared_ptr<arrow::DataType>> m_apoBaseStructGeomType{};
     409             : 
     410             :     //! Whether to use a struct field with the values of the bounding box
     411             :     // of the geometries. Used by Parquet.
     412             :     bool m_bWriteBBoxStruct = false;
     413             : 
     414             :     //! Schema fields for bounding box of geometry columns.
     415             :     // Constraint: if not empty, m_apoFieldsBBOX.size() == m_poFeatureDefn->GetGeomFieldCount()
     416             :     std::vector<std::shared_ptr<arrow::Field>> m_apoFieldsBBOX{};
     417             : 
     418             :     //! Array builers for bounding box of geometry columns.
     419             :     // m_apoBuildersBBOXStruct is for the top-level field of type struct.
     420             :     // m_apoBuildersBBOX{XMin|YMin|XMax|YMax} are for the floating-point values
     421             :     // Constraint: if not empty, m_apoBuildersBBOX{Struct|XMin|YMin|XMax|YMax}.size() == m_poFeatureDefn->GetGeomFieldCount()
     422             :     std::vector<std::shared_ptr<arrow::StructBuilder>>
     423             :         m_apoBuildersBBOXStruct{};
     424             :     std::vector<std::shared_ptr<arrow::FloatBuilder>> m_apoBuildersBBOXXMin{};
     425             :     std::vector<std::shared_ptr<arrow::FloatBuilder>> m_apoBuildersBBOXYMin{};
     426             :     std::vector<std::shared_ptr<arrow::FloatBuilder>> m_apoBuildersBBOXXMax{};
     427             :     std::vector<std::shared_ptr<arrow::FloatBuilder>> m_apoBuildersBBOXYMax{};
     428             : 
     429             :     std::string m_osFIDColumn{};
     430             :     int64_t m_nFeatureCount = 0;
     431             : 
     432             :     int64_t m_nRowGroupSize = 64 * 1024;
     433             :     arrow::Compression::type m_eCompression = arrow::Compression::UNCOMPRESSED;
     434             : 
     435             :     std::vector<std::shared_ptr<arrow::Field>> m_apoFieldsFromArrowSchema{};
     436             :     std::vector<std::shared_ptr<arrow::ArrayBuilder>> m_apoBuilders{};
     437             : 
     438             :     std::vector<uint8_t> m_abyBuffer{};
     439             : 
     440             :     std::vector<int> m_anTZFlag{};               // size: GetFieldCount()
     441             :     std::vector<OGREnvelope3D> m_aoEnvelopes{};  // size: GetGeomFieldCount()
     442             :     std::vector<std::set<OGRwkbGeometryType>>
     443             :         m_oSetWrittenGeometryTypes{};  // size: GetGeomFieldCount()
     444             : 
     445             :     bool m_bEdgesSpherical = false;
     446             : #if ARROW_VERSION_MAJOR >= 21
     447             :     bool m_bUseArrowWKBExtension = false;
     448             : #endif
     449             : 
     450             :     static OGRArrowGeomEncoding
     451             :     GetPreciseArrowGeomEncoding(OGRArrowGeomEncoding eEncodingType,
     452             :                                 OGRwkbGeometryType eGType);
     453             :     static const char *
     454             :     GetGeomEncodingAsString(OGRArrowGeomEncoding eGeomEncoding,
     455             :                             bool bForParquetGeo);
     456             : 
     457             :     virtual bool IsSupportedGeometryType(OGRwkbGeometryType eGType) const = 0;
     458             : 
     459             :     virtual std::string GetDriverUCName() const = 0;
     460             : 
     461             :     virtual bool IsFileWriterCreated() const = 0;
     462             :     virtual void CreateWriter() = 0;
     463             :     virtual bool CloseFileWriter() = 0;
     464             : 
     465             :     void CreateSchemaCommon();
     466             :     void FinalizeSchema();
     467             :     virtual void CreateSchema() = 0;
     468             : 
     469           0 :     virtual void PerformStepsBeforeFinalFlushGroup()
     470             :     {
     471           0 :     }
     472             : 
     473             :     void CreateArrayBuilders();
     474             : 
     475             :     //! Clear array builders
     476             :     void ClearArrayBuilers();
     477             : 
     478             :     virtual bool FlushGroup() = 0;
     479             :     bool FinalizeWriting();
     480             :     bool WriteArrays(std::function<bool(const std::shared_ptr<arrow::Field> &,
     481             :                                         const std::shared_ptr<arrow::Array> &)>
     482             :                          postProcessArray);
     483             : 
     484         128 :     virtual void FixupWKBGeometryBeforeWriting(GByte * /*pabyWKB*/,
     485             :                                                size_t /*nLen*/)
     486             :     {
     487         128 :     }
     488             : 
     489           0 :     virtual void FixupGeometryBeforeWriting(OGRGeometry * /* poGeom */)
     490             :     {
     491           0 :     }
     492             : 
     493             :     virtual bool IsSRSRequired() const = 0;
     494             :     bool WriteArrowBatchInternal(
     495             :         const struct ArrowSchema *schema, struct ArrowArray *array,
     496             :         CSLConstList papszOptions,
     497             :         std::function<bool(const std::shared_ptr<arrow::RecordBatch> &)>
     498             :             writeBatch);
     499             : 
     500             :     OGRErr BuildGeometry(OGRGeometry *poGeom, int iGeomField,
     501             :                          arrow::ArrayBuilder *poBuilder);
     502             : 
     503             :   public:
     504             :     OGRArrowWriterLayer(
     505             :         arrow::MemoryPool *poMemoryPool,
     506             :         const std::shared_ptr<arrow::io::OutputStream> &poOutputStream,
     507             :         const char *pszLayerName);
     508             : 
     509             :     ~OGRArrowWriterLayer() override;
     510             : 
     511             :     bool AddFieldDomain(std::unique_ptr<OGRFieldDomain> &&domain,
     512             :                         std::string &failureReason);
     513             :     std::vector<std::string> GetFieldDomainNames() const;
     514             :     const OGRFieldDomain *GetFieldDomain(const std::string &name) const;
     515             : 
     516          10 :     const char *GetFIDColumn() const override
     517             :     {
     518          10 :         return m_osFIDColumn.c_str();
     519             :     }
     520             : 
     521             :     using OGRLayer::GetLayerDefn;
     522             : 
     523        6161 :     const OGRFeatureDefn *GetLayerDefn() const override
     524             :     {
     525        6161 :         return m_poFeatureDefn;
     526             :     }
     527             : 
     528          23 :     void ResetReading() override
     529             :     {
     530          23 :     }
     531             : 
     532          23 :     OGRFeature *GetNextFeature() override
     533             :     {
     534          23 :         return nullptr;
     535             :     }
     536             : 
     537             :     int TestCapability(const char *pszCap) const override;
     538             :     OGRErr CreateField(const OGRFieldDefn *poField,
     539             :                        int bApproxOK = TRUE) override;
     540             :     OGRErr CreateGeomField(const OGRGeomFieldDefn *poField,
     541             :                            int bApproxOK = TRUE) override;
     542             :     GIntBig GetFeatureCount(int bForce) override;
     543             : 
     544         112 :     bool IsArrowSchemaSupported(const struct ArrowSchema * /*schema*/,
     545             :                                 CSLConstList /* papszOptions */,
     546             :                                 std::string & /*osErrorMsg */) const override
     547             :     {
     548         112 :         return true;
     549             :     }
     550             : 
     551             :     bool
     552             :     CreateFieldFromArrowSchema(const struct ArrowSchema *schema,
     553             :                                CSLConstList papszOptions = nullptr) override;
     554             :     bool WriteArrowBatch(const struct ArrowSchema *schema,
     555             :                          struct ArrowArray *array,
     556             :                          CSLConstList papszOptions = nullptr) override = 0;
     557             : 
     558             :   protected:
     559             :     OGRErr ICreateFeature(OGRFeature *poFeature) override;
     560             : 
     561             :     bool FlushFeatures();
     562             : 
     563             :     static void RemoveIDFromMemberOfEnsembles(CPLJSONObject &obj);
     564             :     static OGRSpatialReference IdentifyCRS(const OGRSpatialReference *poSRS);
     565             : };
     566             : 
     567             : /************************************************************************/
     568             : /*                     OGRGeoArrowWkbExtensionType                      */
     569             : /************************************************************************/
     570             : 
     571             : #if ARROW_VERSION_MAJOR >= 21
     572             : 
     573             : class OGRGeoArrowWkbExtensionType final : public arrow::ExtensionType
     574             : {
     575             :   public:
     576             :     explicit OGRGeoArrowWkbExtensionType(
     577             :         const std::shared_ptr<arrow::DataType> &storage_type,
     578             :         const std::string &metadata)
     579             :         : arrow::ExtensionType(storage_type), metadata_(metadata)
     580             :     {
     581             :     }
     582             : 
     583             :     std::string extension_name() const override
     584             :     {
     585             :         return EXTENSION_NAME_GEOARROW_WKB;
     586             :     }
     587             : 
     588             :     bool ExtensionEquals(const arrow::ExtensionType &other) const override
     589             :     {
     590             :         return extension_name() == other.extension_name() &&
     591             :                storage_type_->Equals(other.storage_type()) &&
     592             :                Serialize() == other.Serialize();
     593             :     }
     594             : 
     595             :     arrow::Result<std::shared_ptr<arrow::DataType>>
     596             :     Deserialize(std::shared_ptr<arrow::DataType> storage_type,
     597             :                 const std::string &serialized) const override
     598             :     {
     599             :         return Make(std::move(storage_type), serialized);
     600             :     }
     601             : 
     602             :     std::string Serialize() const override
     603             :     {
     604             :         return metadata_;
     605             :     }
     606             : 
     607             :     std::shared_ptr<arrow::Array>
     608             :     MakeArray(std::shared_ptr<arrow::ArrayData> data) const override
     609             :     {
     610             :         CPLAssert(data->type->id() == arrow::Type::EXTENSION);
     611             :         CPLAssert(EXTENSION_NAME_GEOARROW_WKB ==
     612             :                   static_cast<const arrow::ExtensionType &>(*data->type)
     613             :                       .extension_name());
     614             :         return std::make_shared<arrow::ExtensionArray>(data);
     615             :     }
     616             : 
     617             :     static bool IsSupportedStorageType(arrow::Type::type typeId)
     618             :     {
     619             :         // TODO: also add BINARY_VIEW if we support it some day.
     620             :         return typeId == arrow::Type::BINARY ||
     621             :                typeId == arrow::Type::LARGE_BINARY;
     622             :     }
     623             : 
     624             :     static arrow::Result<std::shared_ptr<arrow::DataType>>
     625             :     Make(std::shared_ptr<arrow::DataType> storage_type,
     626             :          const std::string &metadata)
     627             :     {
     628             :         if (!IsSupportedStorageType(storage_type->id()))
     629             :         {
     630             :             return arrow::Status::Invalid(
     631             :                 "Invalid storage type for OGRGeoArrowWkbExtensionType: ",
     632             :                 storage_type->ToString());
     633             :         }
     634             :         return std::make_shared<OGRGeoArrowWkbExtensionType>(
     635             :             std::move(storage_type), metadata);
     636             :     }
     637             : 
     638             :   private:
     639             :     std::string metadata_{};
     640             : };
     641             : 
     642             : #endif  // ARROW_VERSION_MAJOR >= 21
     643             : 
     644             : #if defined(__clang__)
     645             : #pragma clang diagnostic pop
     646             : #endif
     647             : 
     648             : #endif  // OGR_ARROW_H

Generated by: LCOV version 1.14