diff --git a/CMakeLists.txt b/CMakeLists.txt index ad30a13..2794a06 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -93,6 +93,8 @@ set(EXTENSION_SOURCES src/geo/tnpoint_in_out.cpp src/geo/tpose.cpp src/geo/tpose_in_out.cpp + src/geo/trgeo.cpp + src/geo/trgeo_in_out.cpp src/geo/tgeogpoint.cpp src/geo/tgeogpoint_in_out.cpp src/geo/tgeogpoint_ops.cpp diff --git a/src/geo/trgeo.cpp b/src/geo/trgeo.cpp new file mode 100644 index 0000000..15f1177 --- /dev/null +++ b/src/geo/trgeo.cpp @@ -0,0 +1,1288 @@ +#include "geo/trgeo.hpp" +#include "geo/tgeompoint_functions.hpp" +#include "duckdb/main/extension/extension_loader.hpp" +#include "duckdb/common/extension_type_info.hpp" +#include +#include +#include +#include "temporal/spanset.hpp" +#include "temporal/set.hpp" +#include "temporal/temporal_functions.hpp" +#include "geo/stbox.hpp" +#include "geo/geoset.hpp" +#include +#include "geo_util.hpp" +#include "spatial/spatial_types.hpp" +#include "mobilityduck/meos_exec_serial.hpp" + +extern "C" { + #include + #include + #include + #include + // meos_rgeo.h is deliberately NOT included: it is the only extended + // module header that declares an Interval-typed prototype + // (trgeo_append_tinstant), and at this namespace scope the MEOS + // `typedef struct Interval Interval` (from meos.h) collides with + // `duckdb::Interval` (from tydef.hpp), making the reference + // ambiguous. This port only needs four trgeo symbols, declared + // locally as extern prototypes instead. + extern char *trgeo_out(const Temporal *temp); + extern GSERIALIZED *trgeo_geom(const Temporal *temp); + extern GSERIALIZED *trgeo_start_value(const Temporal *temp); + extern GSERIALIZED *trgeo_end_value(const Temporal *temp); + // The temporal rigid geometry text/EWKT parser. trgeo_in and + // trgeo_from_mfjson are MEOS inline (#if MEOS) wrappers with no + // linkable symbol and are not declared in the installed + // meos_rgeo.h, so this port routes through the subtype-agnostic + // spatial parser tspatial_parse (a real exported symbol that + // trgeo_in itself delegates to with T_TRGEOMETRY), exactly the way + // the canonical MobilityDB SQL binds trgeometryFromText / + // trgeometry_in. Declared locally because the internal + // geo/tspatial_parser.h is not part of the installed MEOS headers. + extern Temporal *tspatial_parse(const char **str, MeosType temptype); +} + + +namespace duckdb { + +LogicalType TRGeoTypes::TRGEOMETRY() { + auto type = LogicalType(LogicalTypeId::BLOB); + type.SetAlias("TRGEOMETRY"); + return type; +} + +static inline Temporal *Trgeo_parse_text(const std::string &str) { + const char *ptr = str.c_str(); + return tspatial_parse(&ptr, T_TRGEOMETRY); +} + +/* + * Constructors +*/ + +inline void Trgeo_constructor(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_geom_str) -> string_t { + std::string input = input_geom_str.GetString(); + + Temporal *tinst = Trgeo_parse_text(input); + if (!tinst) { + throw InvalidInputException("Invalid TRGEOMETRY input: " + input); + } + + size_t data_size = temporal_mem_size(tinst); + + uint8_t *data_buffer = (uint8_t*)malloc(data_size); + if (!data_buffer) { + free(tinst); + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY data"); + } + + memcpy(data_buffer, tinst, data_size); + + string_t data_string_t(reinterpret_cast(data_buffer), data_size); + string_t stored_data = StringVector::AddStringOrBlob(result, data_string_t); + + free(data_buffer); + free(tinst); + + return stored_data; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +TInstant **temparr_extract_rg(Vector &trgeo_arr_vec, list_entry_t list_entry, int *count) { + auto &child_vector = ListVector::GetEntry(trgeo_arr_vec); + auto list_size = list_entry.length; + auto list_offset = list_entry.offset; + + if (list_size == 0) { + *count = 0; + return nullptr; + } + + *count = list_size; + + TInstant **instants = (TInstant**)malloc(sizeof(TInstant*) * list_size); + if (!instants) { + *count = 0; + return nullptr; + } + + for (idx_t i = 0; i < list_size; i++) { + auto element_idx = list_offset + i; + string_t tgeom_blob = FlatVector::GetData(child_vector)[element_idx]; + + const uint8_t *data = reinterpret_cast(tgeom_blob.GetData()); + size_t data_size = tgeom_blob.GetSize(); + + if (data_size < sizeof(void*)) { + for (idx_t j = 0; j < i; j++) { + if (instants[j]) free(instants[j]); + } + free(instants); + *count = 0; + return nullptr; + } + + uint8_t *data_copy = (uint8_t*)malloc(data_size); + if (!data_copy) { + for (idx_t j = 0; j < i; j++) { + if (instants[j]) free(instants[j]); + } + free(instants); + *count = 0; + return nullptr; + } + memcpy(data_copy, data, data_size); + + Temporal *temp = reinterpret_cast(data_copy); + if (!temp) { + free(data_copy); + for (idx_t j = 0; j < i; j++) { + if (instants[j]) free(instants[j]); + } + free(instants); + *count = 0; + return nullptr; + } + + instants[i] = (TInstant*)temp; + } + + return instants; +} + +inline void Trgeo_sequence_constructor(DataChunk &args, ExpressionState &state, Vector &result) { + // Default values + const char* default_interp = "linear"; + bool default_lower_inc = true; + bool default_upper_inc = true; + + auto count = args.size(); + auto arg_count = args.ColumnCount(); + + + auto &trgeo_arr_vec = args.data[0]; + trgeo_arr_vec.Flatten(count); + + Vector *interp_vec = nullptr; + Vector *lower_vec = nullptr; + Vector *upper_vec = nullptr; + + if (arg_count > 1) { + interp_vec = &args.data[1]; + interp_vec->Flatten(count); + } + if (arg_count > 2) { + lower_vec = &args.data[2]; + lower_vec->Flatten(count); + } + if (arg_count > 3) { + upper_vec = &args.data[3]; + upper_vec->Flatten(count); + } + + result.Flatten(count); + + auto trgeo_data = FlatVector::GetData(trgeo_arr_vec); + auto result_data = FlatVector::GetData(result); + + // Get validity masks + auto &trgeo_validity = FlatVector::Validity(trgeo_arr_vec); + auto &result_validity = FlatVector::Validity(result); + + for (idx_t i = 0; i < count; i++) { + if (!trgeo_validity.RowIsValid(i)) { + result_validity.SetInvalid(i); + continue; + } + + try { + list_entry_t list_entry = trgeo_data[i]; + + // Handle interp parameter with default + std::string interp_str = default_interp; + if (interp_vec) { + auto interp_data = FlatVector::GetData(*interp_vec); + auto &interp_validity = FlatVector::Validity(*interp_vec); + if (interp_validity.RowIsValid(i)) { + interp_str = interp_data[i].GetString(); + } + } + interpType interp = interptype_from_string(interp_str.c_str()); + + bool lower_inc = default_lower_inc; + bool upper_inc = default_upper_inc; + + if (lower_vec) { + auto lower_data = FlatVector::GetData(*lower_vec); + auto &lower_validity = FlatVector::Validity(*lower_vec); + if (lower_validity.RowIsValid(i)) { + lower_inc = lower_data[i]; + } + } + + if (upper_vec) { + auto upper_data = FlatVector::GetData(*upper_vec); + auto &upper_validity = FlatVector::Validity(*upper_vec); + if (upper_validity.RowIsValid(i)) { + upper_inc = upper_data[i]; + } + } + + // Extract array elements + int element_count; + TInstant **instants = temparr_extract_rg(trgeo_arr_vec, list_entry, &element_count); + + if (!instants || element_count == 0) { + result_validity.SetInvalid(i); + continue; + } + + TSequence *sequence_result = tsequence_make((TInstant **) instants, element_count, + lower_inc, upper_inc, interp, true); + + if (!sequence_result) { + for (int j = 0; j < element_count; j++) { + if (instants[j]) { + free(instants[j]); + } + } + free(instants); + result_validity.SetInvalid(i); + continue; + } + + size_t data_size = temporal_mem_size(reinterpret_cast(sequence_result)); + uint8_t *data_buffer = (uint8_t*)malloc(data_size); + if (!data_buffer) { + free(sequence_result); + for (int j = 0; j < element_count; j++) { + if (instants[j]) { + free(instants[j]); + } + } + free(instants); + result_validity.SetInvalid(i); + continue; + } + + memcpy(data_buffer, sequence_result, data_size); + + string_t data_string_t(reinterpret_cast(data_buffer), data_size); + result_data[i] = StringVector::AddStringOrBlob(result, data_string_t); + + free(data_buffer); + free(sequence_result); + for (int j = 0; j < element_count; j++) { + if (instants[j]) { + free(instants[j]); + } + } + free(instants); + + } catch (const std::exception& e) { + result_validity.SetInvalid(i); + } + } + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + + + +/* + * Conversions +*/ + +inline void Temporal_to_tstzspan(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + Span *timespan = temporal_to_tstzspan(temp); + + if (!timespan) { + throw InvalidInputException("Failed to extract timespan from TRGEOMETRY"); + } + + size_t span_size = sizeof(Span); + + uint8_t *span_buffer = (uint8_t*)malloc(span_size); + if (!span_buffer) { + free(timespan); + throw InvalidInputException("Failed to allocate memory for timespan data"); + } + + memcpy(span_buffer, timespan, span_size); + + string_t span_string_t(reinterpret_cast(span_buffer), span_size); + string_t stored_data = StringVector::AddStringOrBlob(result, span_string_t); + + free(span_buffer); + free(timespan); + + return stored_data; + } + ); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +/* + * Transformations +*/ + +inline void Temporal_to_tinstant(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + TInstant *inst = temporal_to_tinstant(temp); + if (!inst) { + throw InvalidInputException("Failed to convert TRGEOMETRY to TInstant"); + } + + size_t inst_size = temporal_mem_size((Temporal*)inst); + + uint8_t *inst_buffer = (uint8_t*)malloc(inst_size); + if (!inst_buffer) { + free(inst); + throw InvalidInputException("Failed to allocate memory for TInstant data"); + } + + memcpy(inst_buffer, inst, inst_size); + + string_t inst_string_t(reinterpret_cast(inst_buffer), inst_size); + string_t stored_data = StringVector::AddStringOrBlob(result, inst_string_t); + + free(inst_buffer); + free(inst); + + return stored_data; + } + ); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +inline void Temporal_set_interp(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom_vec = args.data[0]; + auto &interp_vec = args.data[1]; + + tgeom_vec.Flatten(count); + interp_vec.Flatten(count); + + BinaryExecutor::Execute( + tgeom_vec, interp_vec, result, count, + [&](string_t tgeom_str_t, string_t interp_str_t) -> string_t { + + std::string input = tgeom_str_t.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + std::string interp_str = interp_str_t.GetString(); + interpType new_interp = interptype_from_string(interp_str.c_str()); + + Temporal *result_temp = temporal_set_interp(temp, new_interp); + if (!result_temp) { + throw InvalidInputException("Failed to set interpolation"); + } + + // Serialize result back to binary + size_t result_size = temporal_mem_size(result_temp); + uint8_t *result_buffer = (uint8_t*)malloc(result_size); + if (!result_buffer) { + free(result_temp); + throw InvalidInputException("Failed to allocate memory for result"); + } + + memcpy(result_buffer, result_temp, result_size); + string_t result_string_t(reinterpret_cast(result_buffer), result_size); + string_t stored_data = StringVector::AddStringOrBlob(result, result_string_t); + + free(result_buffer); + free(result_temp); + + return stored_data; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +inline void Temporal_merge(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom1_vec = args.data[0]; + auto &tgeom2_vec = args.data[1]; + + tgeom1_vec.Flatten(count); + tgeom2_vec.Flatten(count); + + BinaryExecutor::Execute( + tgeom1_vec, tgeom2_vec, result, count, + [&](string_t tgeom1_str_t, string_t tgeom2_str_t) -> string_t { + std::string tgeom1 = tgeom1_str_t.GetString(); + + Temporal *temp1 = reinterpret_cast(const_cast(tgeom1.c_str())); + if (!temp1) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + std::string tgeom2 = tgeom2_str_t.GetString(); + + Temporal *temp2 = reinterpret_cast(const_cast(tgeom2.c_str())); + if (!temp2) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + Temporal *result_temp = temporal_merge(temp1, temp2); + if (!result_temp) { + throw InvalidInputException("Failed to merge temporal rigid geometries"); + } + + // Serialize result back to binary + size_t result_size = temporal_mem_size(result_temp); + uint8_t *result_buffer = (uint8_t*)malloc(result_size); + if (!result_buffer) { + free(result_temp); + throw InvalidInputException("Failed to allocate memory for result"); + } + + memcpy(result_buffer, result_temp, result_size); + string_t result_string_t(reinterpret_cast(result_buffer), result_size); + string_t stored_data = StringVector::AddStringOrBlob(result, result_string_t); + + free(result_buffer); + free(result_temp); + + return stored_data; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +/* + * Accessor Functions +*/ + +inline void Temporal_subtype(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom_vec = args.data[0]; + + tgeom_vec.Flatten(count); + + UnaryExecutor::Execute( + tgeom_vec, result, count, + [&](string_t tgeom_str_t) -> string_t { + std::string input = tgeom_str_t.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + const char *subtype_str = temporal_subtype(temp); + if (!subtype_str) { + throw InvalidInputException("Failed to get temporal subtype"); + } + + std::string result_str(subtype_str); + string_t stored_result = StringVector::AddString(result, result_str); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + + + +inline void Temporal_interp(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom_vec = args.data[0]; + + tgeom_vec.Flatten(count); + + UnaryExecutor::Execute( + tgeom_vec, result, count, + [&](string_t tgeom_str_t) -> string_t { + + std::string input = tgeom_str_t.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + + const char *interp_str = temporal_interp(temp); + if (!interp_str) { + throw InvalidInputException("Failed to get temporal interpolation"); + } + + std::string result_str(interp_str); + string_t stored_result = StringVector::AddString(result, result_str); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Temporal_mem_size(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom_vec = args.data[0]; + + tgeom_vec.Flatten(count); + + UnaryExecutor::Execute( + tgeom_vec, result, count, + [&](string_t tgeom_str_t) -> int32_t { + std::string input = tgeom_str_t.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + size_t mem_size = temporal_mem_size(temp); + + + return static_cast(mem_size); + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +// ---- Rigid-geometry value accessors ---- +// The trgeometry value is a moving rigid geometry: a reference geometry +// rotated/translated over time by an embedded temporal pose. It is not a +// registered DuckDB type. geom(trgeometry) returns the reference geometry +// via trgeo_geom; startValue / endValue return the geometry snapshot at +// the first / last instant via trgeo_start_value / trgeo_end_value (both +// header-declared, returning a freshly allocated GSERIALIZED the caller +// owns), surfaced as a GEOMETRY blob, mirroring how the tpose port +// surfaces its value accessors. + +inline void Trgeo_geom(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + // trgeo_geom returns a freshly allocated GSERIALIZED copy of + // the reference geometry, which the caller owns. + GSERIALIZED *gs = trgeo_geom(temp); + if (!gs) { + throw InvalidInputException("Failed to extract reference geometry from TRGEOMETRY"); + } + + string_t geometry_blob = GSerializedToGeometry(gs, state, result); + string_t stored_result = StringVector::AddStringOrBlob(result, geometry_blob); + + free(gs); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Temporal_start_value(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + // trgeo_start_value returns a freshly allocated GSERIALIZED + // copy of the rigid-geometry snapshot at the first instant. + GSERIALIZED *gs = trgeo_start_value(temp); + if (!gs) { + throw InvalidInputException("Failed to extract start value from TRGEOMETRY"); + } + + string_t geometry_blob = GSerializedToGeometry(gs, state, result); + string_t stored_result = StringVector::AddStringOrBlob(result, geometry_blob); + + free(gs); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +inline void Temporal_end_value(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + // trgeo_end_value returns a freshly allocated GSERIALIZED + // copy of the rigid-geometry snapshot at the last instant. + GSERIALIZED *gs = trgeo_end_value(temp); + if (!gs) { + throw InvalidInputException("Failed to extract end value from TRGEOMETRY"); + } + + string_t geometry_blob = GSerializedToGeometry(gs, state, result); + string_t stored_result = StringVector::AddStringOrBlob(result, geometry_blob); + + free(gs); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +inline void Temporal_lower_inc(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal* temp = reinterpret_cast(const_cast(input.c_str())); + + bool lower_inc = temporal_lower_inc(temp); + + std::string result_str = lower_inc ? "true" : "false"; + string_t stored_result = StringVector::AddString(result, result_str); + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Temporal_upper_inc(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal* temp = reinterpret_cast(const_cast(input.c_str())); + + bool upper_inc = temporal_upper_inc(temp); + + std::string result_str = upper_inc ? "true" : "false"; + string_t stored_result = StringVector::AddString(result, result_str); + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Temporal_start_instant(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + + TInstant *start_inst = temporal_start_instant(temp); + + if (!start_inst) { + throw InvalidInputException("Failed to get start_inst from temporal object"); + } + + size_t result_size = temporal_mem_size((Temporal*)start_inst); + if (result_size == 0) { + throw InvalidInputException("Invalid result size from temporal object"); + } + + uint8_t *result_buffer = (uint8_t*)malloc(result_size); + if (!result_buffer) { + free(start_inst); + throw InvalidInputException("Failed to allocate memory for result"); + } + + memcpy(result_buffer, start_inst, result_size); + string_t result_string_t(reinterpret_cast(result_buffer), result_size); + string_t stored_result = StringVector::AddStringOrBlob(result, result_string_t); + + free(result_buffer); + free(start_inst); + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Temporal_end_instant(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_vec = args.data[0]; + + UnaryExecutor::Execute( + input_vec, result, count, + [&](string_t input_str) -> string_t { + std::string input = input_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(input.c_str())); + + TInstant *end_inst = temporal_end_instant(temp); + + if (!end_inst) { + throw InvalidInputException("Failed to get end_inst from temporal object"); + } + + size_t result_size = temporal_mem_size((Temporal*)end_inst); + if (result_size == 0) { + throw InvalidInputException("Invalid result size from temporal object"); + } + + uint8_t *result_buffer = (uint8_t*)malloc(result_size); + if (!result_buffer) { + free(end_inst); + throw InvalidInputException("Failed to allocate memory for result"); + } + + memcpy(result_buffer, end_inst, result_size); + string_t result_string_t(reinterpret_cast(result_buffer), result_size); + string_t stored_result = StringVector::AddStringOrBlob(result, result_string_t); + + free(result_buffer); + free(end_inst); + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + + + +inline void Temporal_instant_n(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &tgeom_vec = args.data[0]; + auto &n_vec = args.data[1]; + + BinaryExecutor::Execute( + tgeom_vec, n_vec, result, count, + [&](string_t tgeom_str, int32_t n) -> string_t { + std::string tgeom = tgeom_str.GetString(); + + Temporal *temp = reinterpret_cast(const_cast(tgeom.c_str())); + + TInstant *inst_n = temporal_instant_n(temp, n); + if (!inst_n) { + throw InvalidInputException("Failed to get instant n from temporal object"); + } + + size_t result_size = temporal_mem_size((Temporal*)inst_n); + if (result_size == 0) { + throw InvalidInputException("Invalid result size from temporal object"); + } + + uint8_t *result_buffer = (uint8_t*)malloc(result_size); + if (!result_buffer) { + free(inst_n); + throw InvalidInputException("Failed to allocate memory for result"); + } + + memcpy(result_buffer, inst_n, result_size); + string_t result_string_t(reinterpret_cast(result_buffer), result_size); + string_t stored_result = StringVector::AddStringOrBlob(result, result_string_t); + + free(result_buffer); + free(inst_n); + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +inline void Tinstant_timestamptz(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_geom_str) -> timestamp_tz_t { + const uint8_t *data = reinterpret_cast(input_geom_str.GetData()); + size_t data_size = input_geom_str.GetSize(); + + if (data_size < sizeof(void*)) { + throw InvalidInputException("Invalid TRGEOMETRY data: insufficient size"); + } + + uint8_t *data_copy = (uint8_t*)malloc(data_size); + if (!data_copy) { + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY deserialization"); + } + memcpy(data_copy, data, data_size); + + TInstant *temp = reinterpret_cast(data_copy); + + if (!temp) { + free(data_copy); + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + TimestampTz meos_t = temp->t; + + timestamp_tz_t meos_timestamp{meos_t}; + timestamp_tz_t duckdb_t = MeosToDuckDBTimestamp(meos_timestamp); + + free(data_copy); + + return duckdb_t; + } + ); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +void TRGeoTypes::RegisterScalarFunctions(ExtensionLoader &loader) { + + auto trgeo_function = ScalarFunction( + "TRGEOMETRY", + {LogicalType::VARCHAR}, + TRGeoTypes::TRGEOMETRY(), + Trgeo_constructor + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_function); + + auto trgeoseqarr_1param= ScalarFunction( + "trgeometrySeq", + {LogicalType::LIST(TRGeoTypes::TRGEOMETRY())}, + TRGeoTypes::TRGEOMETRY(), + Trgeo_sequence_constructor + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeoseqarr_1param); + + auto trgeoseqarr_2params = ScalarFunction( + "trgeometrySeq", + {LogicalType::LIST(TRGeoTypes::TRGEOMETRY()), LogicalType::VARCHAR}, + TRGeoTypes::TRGEOMETRY(), + Trgeo_sequence_constructor + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeoseqarr_2params); + + auto trgeoseqarr_3params = ScalarFunction( + "trgeometrySeq", + {LogicalType::LIST(TRGeoTypes::TRGEOMETRY()), LogicalType::VARCHAR, LogicalType::BOOLEAN}, + TRGeoTypes::TRGEOMETRY(), + Trgeo_sequence_constructor + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeoseqarr_3params); + + auto trgeoseqarr_4params = ScalarFunction( + "trgeometrySeq", + {LogicalType::LIST(TRGeoTypes::TRGEOMETRY()), LogicalType::VARCHAR, LogicalType::BOOLEAN, LogicalType::BOOLEAN}, + TRGeoTypes::TRGEOMETRY(), + Trgeo_sequence_constructor + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeoseqarr_4params); + + auto trgeo_to_timespan_function = ScalarFunction( + "timeSpan", + {TRGeoTypes::TRGEOMETRY()}, + SpanTypes::TSTZSPAN(), + Temporal_to_tstzspan); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_to_timespan_function); + + auto trgeo_to_tinstant_function = ScalarFunction( + "trgeometryInst", + {TRGeoTypes::TRGEOMETRY()}, + TRGeoTypes::TRGEOMETRY(), + Temporal_to_tinstant); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_to_tinstant_function); + + + auto setInterp_function = ScalarFunction( + "setInterp", + {TRGeoTypes::TRGEOMETRY(), LogicalType::VARCHAR}, + TRGeoTypes::TRGEOMETRY(), + Temporal_set_interp + ); + duckdb::RegisterSerializedScalarFunction(loader, setInterp_function); + + + auto merge_function = ScalarFunction( + "merge", + {TRGeoTypes::TRGEOMETRY(), TRGeoTypes::TRGEOMETRY()}, + TRGeoTypes::TRGEOMETRY(), + Temporal_merge + ); + duckdb::RegisterSerializedScalarFunction(loader, merge_function); + + auto tempSubtype_function = ScalarFunction( + "tempSubtype", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::VARCHAR, + Temporal_subtype + ); + duckdb::RegisterSerializedScalarFunction(loader, tempSubtype_function); + + auto interp_function = ScalarFunction( + "interp", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::VARCHAR, + Temporal_interp + ); + duckdb::RegisterSerializedScalarFunction(loader, interp_function); + + auto memSize_function = ScalarFunction( + "memSize", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::INTEGER, + Temporal_mem_size + ); + duckdb::RegisterSerializedScalarFunction(loader, memSize_function); + + auto trgeo_geom_function = ScalarFunction( + "geom", + {TRGeoTypes::TRGEOMETRY()}, + GeoTypes::GEOMETRY(), + Trgeo_geom + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_geom_function); + + auto trgeo_start_value_function = ScalarFunction( + "startValue", + {TRGeoTypes::TRGEOMETRY()}, + GeoTypes::GEOMETRY(), + Temporal_start_value + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_start_value_function); + + auto trgeo_end_value_function = ScalarFunction( + "endValue", + {TRGeoTypes::TRGEOMETRY()}, + GeoTypes::GEOMETRY(), + Temporal_end_value + ); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_end_value_function); + + auto startInstant_function = ScalarFunction( + "startInstant", + {TRGeoTypes::TRGEOMETRY()}, + TRGeoTypes::TRGEOMETRY(), + Temporal_start_instant + ); + duckdb::RegisterSerializedScalarFunction(loader, startInstant_function); + + auto endInstant_function = ScalarFunction( + "endInstant", + {TRGeoTypes::TRGEOMETRY()}, + TRGeoTypes::TRGEOMETRY(), + Temporal_end_instant + ); + duckdb::RegisterSerializedScalarFunction(loader, endInstant_function); + + auto instantN_function = ScalarFunction( + "instantN", + {TRGeoTypes::TRGEOMETRY(), LogicalType::INTEGER}, + TRGeoTypes::TRGEOMETRY(), + Temporal_instant_n + ); + duckdb::RegisterSerializedScalarFunction(loader, instantN_function); + + + auto trgeo_gettimestamptz_function = ScalarFunction( + "getTimestamp", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::TIMESTAMP_TZ, + Tinstant_timestamptz); + duckdb::RegisterSerializedScalarFunction(loader, trgeo_gettimestamptz_function); + + + // =================================================================== + // Foundational trgeo surface — accessors, time/value-restrict, + // modifiers, and comparison. The MEOS C functions delegated to here + // are subtype-agnostic (they take Temporal *), so we reuse the same + // generic handlers wired for tgeompoint in temporal_functions.cpp. + // =================================================================== + + const LogicalType TGEOM = TRGeoTypes::TRGEOMETRY(); + const LogicalType TSTZ = LogicalType::TIMESTAMP_TZ; + const LogicalType IVAL = LogicalType::INTERVAL; + + // ---- Accessors ---- + loader.RegisterFunction(ScalarFunction( + "getTime", {TGEOM}, SpansetTypes::tstzspanset(), + TemporalFunctions::Temporal_time)); + loader.RegisterFunction(ScalarFunction( + "duration", {TGEOM}, IVAL, + TemporalFunctions::Temporal_duration)); + loader.RegisterFunction(ScalarFunction( + "duration", {TGEOM, LogicalType::BOOLEAN}, IVAL, + TemporalFunctions::Temporal_duration)); + loader.RegisterFunction(ScalarFunction( + "lowerInc", {TGEOM}, LogicalType::BOOLEAN, + TemporalFunctions::Temporal_lower_inc)); + loader.RegisterFunction(ScalarFunction( + "upperInc", {TGEOM}, LogicalType::BOOLEAN, + TemporalFunctions::Temporal_upper_inc)); + loader.RegisterFunction(ScalarFunction( + "numInstants", {TGEOM}, LogicalType::INTEGER, + TemporalFunctions::Temporal_num_instants)); + loader.RegisterFunction(ScalarFunction( + "instants", {TGEOM}, LogicalType::LIST(TGEOM), + TemporalFunctions::Temporal_instants)); + loader.RegisterFunction(ScalarFunction( + "numSequences", {TGEOM}, LogicalType::INTEGER, + TemporalFunctions::Temporal_num_sequences)); + loader.RegisterFunction(ScalarFunction( + "sequences", {TGEOM}, LogicalType::LIST(TGEOM), + TemporalFunctions::Temporal_sequences)); + loader.RegisterFunction(ScalarFunction( + "startSequence", {TGEOM}, TGEOM, + TemporalFunctions::Temporal_start_sequence)); + loader.RegisterFunction(ScalarFunction( + "endSequence", {TGEOM}, TGEOM, + TemporalFunctions::Temporal_end_sequence)); + loader.RegisterFunction(ScalarFunction( + "sequenceN", {TGEOM, LogicalType::INTEGER}, TGEOM, + TemporalFunctions::Temporal_sequence_n)); + loader.RegisterFunction(ScalarFunction( + "numTimestamps", {TGEOM}, LogicalType::INTEGER, + TemporalFunctions::Temporal_num_timestamps)); + loader.RegisterFunction(ScalarFunction( + "timestamps", {TGEOM}, LogicalType::LIST(TSTZ), + TemporalFunctions::Temporal_timestamps)); + loader.RegisterFunction(ScalarFunction( + "startTimestamp", {TGEOM}, TSTZ, + TemporalFunctions::Temporal_start_timestamptz)); + loader.RegisterFunction(ScalarFunction( + "endTimestamp", {TGEOM}, TSTZ, + TemporalFunctions::Temporal_end_timestamptz)); + loader.RegisterFunction(ScalarFunction( + "timestampN", {TGEOM, LogicalType::INTEGER}, TSTZ, + TemporalFunctions::Temporal_timestamptz_n)); + loader.RegisterFunction(ScalarFunction( + "segments", {TGEOM}, LogicalType::LIST(TGEOM), + TemporalFunctions::Temporal_segments)); + + // ---- Time-domain restrict / minus ---- + for (const auto &t : std::vector>{ + {TSTZ, TemporalFunctions::Temporal_at_timestamptz}, + {SetTypes::tstzset(), TemporalFunctions::Temporal_at_tstzset}, + {SpanTypes::TSTZSPAN(), TemporalFunctions::Temporal_at_tstzspan}, + {SpansetTypes::tstzspanset(), TemporalFunctions::Temporal_at_tstzspanset}}) { + loader.RegisterFunction(ScalarFunction( + "atTime", {TGEOM, t.first}, TGEOM, t.second)); + } + for (const auto &t : std::vector>{ + {TSTZ, TemporalFunctions::Temporal_minus_timestamptz}, + {SetTypes::tstzset(), TemporalFunctions::Temporal_minus_tstzset}, + {SpanTypes::TSTZSPAN(), TemporalFunctions::Temporal_minus_tstzspan}, + {SpansetTypes::tstzspanset(), TemporalFunctions::Temporal_minus_tstzspanset}}) { + loader.RegisterFunction(ScalarFunction( + "minusTime", {TGEOM, t.first}, TGEOM, t.second)); + } + + // beforeTimestamp / afterTimestamp accept timestamptz + loader.RegisterFunction(ScalarFunction( + "beforeTimestamp", {TGEOM, TSTZ}, TGEOM, + TemporalFunctions::Temporal_before_timestamptz)); + loader.RegisterFunction(ScalarFunction( + "afterTimestamp", {TGEOM, TSTZ}, TGEOM, + TemporalFunctions::Temporal_after_timestamptz)); + + // ---- Modifiers (shift / scale / shiftScale / append / insert / update / + // delete) ---- + loader.RegisterFunction(ScalarFunction( + "shiftTime", {TGEOM, IVAL}, TGEOM, + TemporalFunctions::Temporal_shift_time)); + loader.RegisterFunction(ScalarFunction( + "scaleTime", {TGEOM, IVAL}, TGEOM, + TemporalFunctions::Temporal_scale_time)); + loader.RegisterFunction(ScalarFunction( + "shiftScaleTime", {TGEOM, IVAL, IVAL}, TGEOM, + TemporalFunctions::Temporal_shift_scale_time)); + loader.RegisterFunction(ScalarFunction( + "appendInstant", {TGEOM, TGEOM}, TGEOM, + TemporalFunctions::Temporal_append_tinstant)); + loader.RegisterFunction(ScalarFunction( + "appendSequence", {TGEOM, TGEOM}, TGEOM, + TemporalFunctions::Temporal_append_tsequence)); + loader.RegisterFunction(ScalarFunction( + "insert", {TGEOM, TGEOM}, TGEOM, + TemporalFunctions::Temporal_insert)); + loader.RegisterFunction(ScalarFunction( + "insert", {TGEOM, TGEOM, LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_insert)); + loader.RegisterFunction(ScalarFunction( + "update", {TGEOM, TGEOM}, TGEOM, + TemporalFunctions::Temporal_update)); + loader.RegisterFunction(ScalarFunction( + "update", {TGEOM, TGEOM, LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_update)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, TSTZ}, TGEOM, + TemporalFunctions::Temporal_delete_timestamptz)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, TSTZ, LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_delete_timestamptz)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SetTypes::tstzset()}, TGEOM, + TemporalFunctions::Temporal_delete_tstzset)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SetTypes::tstzset(), LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_delete_tstzset)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SpanTypes::TSTZSPAN()}, TGEOM, + TemporalFunctions::Temporal_delete_tstzspan)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SpanTypes::TSTZSPAN(), LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_delete_tstzspan)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SpansetTypes::tstzspanset()}, TGEOM, + TemporalFunctions::Temporal_delete_tstzspanset)); + loader.RegisterFunction(ScalarFunction( + "deleteTime", {TGEOM, SpansetTypes::tstzspanset(), LogicalType::BOOLEAN}, TGEOM, + TemporalFunctions::Temporal_delete_tstzspanset)); + + // ---- Comparison (named functions + operators) ---- + struct CmpEntry { + const char *name; + scalar_function_t fn; + }; + const std::vector named_cmps = { + {"temporal_eq", TemporalFunctions::Temporal_eq}, + {"temporal_ne", TemporalFunctions::Temporal_ne}, + {"temporal_lt", TemporalFunctions::Temporal_lt}, + {"temporal_le", TemporalFunctions::Temporal_le}, + {"temporal_gt", TemporalFunctions::Temporal_gt}, + {"temporal_ge", TemporalFunctions::Temporal_ge}, + }; + for (const auto &c : named_cmps) { + loader.RegisterFunction(ScalarFunction( + c.name, {TGEOM, TGEOM}, LogicalType::BOOLEAN, c.fn)); + } + loader.RegisterFunction(ScalarFunction( + "temporal_cmp", {TGEOM, TGEOM}, LogicalType::INTEGER, + TemporalFunctions::Temporal_cmp)); + + // Operator forms — mirror the registrations tgeometry.cpp does. + const std::vector op_cmps = { + {"=", TemporalFunctions::Temporal_eq}, + {"<>", TemporalFunctions::Temporal_ne}, + {"<", TemporalFunctions::Temporal_lt}, + {"<=", TemporalFunctions::Temporal_le}, + {">", TemporalFunctions::Temporal_gt}, + {">=", TemporalFunctions::Temporal_ge}, + }; + for (const auto &c : op_cmps) { + loader.RegisterFunction(ScalarFunction( + c.name, {TGEOM, TGEOM}, LogicalType::BOOLEAN, c.fn)); + } +} + +void TRGeoTypes::RegisterTypes(ExtensionLoader &loader) { + loader.RegisterType( "TRGEOMETRY", TRGeoTypes::TRGEOMETRY()); +} + + +} diff --git a/src/geo/trgeo_in_out.cpp b/src/geo/trgeo_in_out.cpp new file mode 100644 index 0000000..a713e44 --- /dev/null +++ b/src/geo/trgeo_in_out.cpp @@ -0,0 +1,339 @@ +#include "geo/trgeo.hpp" +#include "duckdb/main/extension/extension_loader.hpp" +#include "duckdb/common/extension_type_info.hpp" +#include +#include +#include +#include "mobilityduck/meos_exec_serial.hpp" + +extern "C" { + #include + #include + #include + #include + // meos_rgeo.h is deliberately NOT included: its Interval-typed + // trgeo_append_tinstant prototype makes `Interval` ambiguous against + // duckdb::Interval at this namespace scope (see geo/trgeo.cpp). Only + // trgeo_out is needed here; declared locally. + extern char *trgeo_out(const Temporal *temp); + // See geo/trgeo.cpp: trgeo_in is a MEOS inline (#if MEOS) wrapper + // with no linkable symbol and is not declared in the installed + // meos_rgeo.h. This port routes text/EWKT input through the + // subtype-agnostic spatial parser tspatial_parse with T_TRGEOMETRY + // (the exact path trgeo_in itself delegates to and the canonical + // MobilityDB SQL binds trgeometryFromText / trgeometry_in to). + extern Temporal *tspatial_parse(const char **str, MeosType temptype); +} + +namespace duckdb { + +static inline Temporal *Trgeo_parse_in(const std::string &str) { + const char *ptr = str.c_str(); + return tspatial_parse(&ptr, T_TRGEOMETRY); +} + +inline void Tspatial_as_text(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_geom_str) -> string_t { + const uint8_t *data = reinterpret_cast(input_geom_str.GetData()); + size_t data_size = input_geom_str.GetSize(); + + if (data_size < sizeof(void*)) { + throw InvalidInputException("Invalid TRGEOMETRY data: insufficient size"); + } + + uint8_t *data_copy = (uint8_t*)malloc(data_size); + if (!data_copy) { + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY deserialization"); + } + memcpy(data_copy, data, data_size); + + Temporal *temp = reinterpret_cast(data_copy); + + if (!temp) { + free(data_copy); + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + char *str = tspatial_as_text(temp, 0); + + if (!str) { + free(data_copy); + throw InvalidInputException("Failed to convert TRGEOMETRY to text"); + } + + std::string result_str(str); + string_t stored_result = StringVector::AddString(result, result_str); + + free(str); + free(data_copy); + + return stored_result; + } + ); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + +inline void Tspatial_as_ewkt(DataChunk &args, ExpressionState &state, Vector &result) { + auto count = args.size(); + auto &input_geom_vec = args.data[0]; + + UnaryExecutor::Execute( + input_geom_vec, result, count, + [&](string_t input_geom_str) -> string_t { + + const uint8_t *data = reinterpret_cast(input_geom_str.GetData()); + size_t data_size = input_geom_str.GetSize(); + + if (data_size < sizeof(void*)) { + throw InvalidInputException("Invalid TRGEOMETRY data: insufficient size"); + } + + uint8_t *data_copy = (uint8_t*)malloc(data_size); + if (!data_copy) { + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY deserialization"); + } + memcpy(data_copy, data, data_size); + + Temporal *temp = reinterpret_cast(data_copy); + + if (!temp) { + free(data_copy); + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + char *ewkt = tspatial_as_ewkt(temp, 0); + + if (!ewkt) { + free(data_copy); + throw InvalidInputException("Failed to convert TRGEOMETRY to EWKT"); + } + + std::string result_str(ewkt); + string_t stored_result = StringVector::AddString(result, result_str); + + + free(ewkt); + free(data_copy); + + return stored_result; + } + ); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } +} + + +bool TrgeoFunctions::StringToTrgeo(Vector &source, Vector &result, idx_t count, CastParameters ¶meters) { + UnaryExecutor::Execute( + source, result, count, + [&](string_t input_string) -> string_t { + std::string input_str = input_string.GetString(); + + Temporal *temp = Trgeo_parse_in(input_str); + if (!temp) { + throw InvalidInputException("Invalid TRGEOMETRY input: " + input_str); + } + + size_t data_size = temporal_mem_size(temp); + uint8_t *data_buffer = (uint8_t*)malloc(data_size); + if (!data_buffer) { + free(temp); + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY data"); + } + + memcpy(data_buffer, temp, data_size); + + string_t data_string_t(reinterpret_cast(data_buffer), data_size); + string_t stored_data = StringVector::AddStringOrBlob(result, data_string_t); + + free(data_buffer); + free(temp); + + return stored_data; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } + return true; +} + +bool TrgeoFunctions::TrgeoToString(Vector &source, Vector &result, idx_t count, CastParameters ¶meters) { + UnaryExecutor::Execute( + source, result, count, + [&](string_t input_blob) -> string_t { + const uint8_t *data = reinterpret_cast(input_blob.GetData()); + size_t data_size = input_blob.GetSize(); + + if (data_size < sizeof(void*)) { + throw InvalidInputException("Invalid TRGEOMETRY data: insufficient size"); + } + + uint8_t *data_copy = (uint8_t*)malloc(data_size); + if (!data_copy) { + throw InvalidInputException("Failed to allocate memory for TRGEOMETRY deserialization"); + } + memcpy(data_copy, data, data_size); + + Temporal *temp = reinterpret_cast(data_copy); + if (!temp) { + free(data_copy); + throw InvalidInputException("Invalid TRGEOMETRY data: null pointer"); + } + + // The trgeometry value is a moving rigid geometry; trgeo_out + // is the type-specific serializer that renders the reference + // geometry plus the embedded temporal pose. The generic + // temporal_out does not understand the rigid value model. + char *str = trgeo_out(temp); + if (!str) { + free(data_copy); + throw InvalidInputException("Failed to convert TRGEOMETRY to string"); + } + + std::string output(str); + string_t stored_result = StringVector::AddString(result, output); + + free(str); + free(data_copy); + + return stored_result; + }); + + if (count == 1) { + result.SetVectorType(VectorType::CONSTANT_VECTOR); + } + return true; +} + +// ---- Spatial-temporal parsers (Binary / HexWKB / MFJSON / Text) ---- +// Used to register the `trgeometryFrom*` overloads. +// `temporal_from_wkb` and `temporal_from_hexwkb` are subtype-agnostic; +// the trgeometry text/EWKT parser routes through tspatial_parse with +// T_TRGEOMETRY. The temporal-rigid-geometry MF-JSON support already +// lives on MobilityDB master (MovingRigidGeometry typestring + dispatch), +// so no preceding MEOS parity PR is needed; MEOS does not, however, +// expose a header-declared `trgeo_from_mfjson(const char *)` symbol (it +// is an `#if MEOS` inline that calls `temporal_from_mfjson(mfjson, +// T_TRGEOMETRY)`). This port therefore routes through that +// subtype-agnostic dispatch directly, exactly as the canonical +// MobilityDB SQL binds `trgeometryFromMFJSON` to the generic +// Temporal_from_mfjson handler. The result is stored as a raw blob, +// the same format every other temporal type uses. + +inline string_t StoreTempAsBlob(Vector &result, Temporal *t) { + size_t sz = temporal_mem_size(t); + string_t stored = StringVector::AddStringOrBlob( + result, string_t(reinterpret_cast(t), sz)); + free(t); + return stored; +} + +inline void TspatialFromWkbExec(DataChunk &args, ExpressionState &, Vector &result) { + UnaryExecutor::Execute( + args.data[0], result, args.size(), + [&](string_t input) -> string_t { + if (input.GetSize() == 0) + throw InvalidInputException("fromBinary: empty WKB input"); + uint8_t *wkb = (uint8_t *)malloc(input.GetSize()); + if (!wkb) throw InternalException("fromBinary: malloc failed"); + memcpy(wkb, input.GetData(), input.GetSize()); + Temporal *t = temporal_from_wkb(wkb, input.GetSize()); + free(wkb); + if (!t) throw InvalidInputException("fromBinary: invalid MEOS-WKB"); + return StoreTempAsBlob(result, t); + }); +} + +inline void TspatialFromHexWkbExec(DataChunk &args, ExpressionState &, Vector &result) { + UnaryExecutor::Execute( + args.data[0], result, args.size(), + [&](string_t input) -> string_t { + std::string hex(input.GetData(), input.GetSize()); + Temporal *t = temporal_from_hexwkb(hex.c_str()); + if (!t) throw InvalidInputException( + "fromHexWKB: invalid hex-encoded MEOS-WKB"); + return StoreTempAsBlob(result, t); + }); +} + +inline void TrgeoFromTextExec(DataChunk &args, ExpressionState &, Vector &result) { + UnaryExecutor::Execute( + args.data[0], result, args.size(), + [&](string_t input) -> string_t { + std::string s(input.GetData(), input.GetSize()); + Temporal *t = Trgeo_parse_in(s); + if (!t) throw InvalidInputException("from*: invalid input"); + return StoreTempAsBlob(result, t); + }); +} + +inline void TrgeoFromMFJSONExec(DataChunk &args, ExpressionState &, Vector &result) { + UnaryExecutor::Execute( + args.data[0], result, args.size(), + [&](string_t input) -> string_t { + std::string s(input.GetData(), input.GetSize()); + // trgeo exposes no header-declared *_from_mfjson symbol; + // route through the generic dispatch with the T_TRGEOMETRY + // temporal type, the same path the canonical MobilityDB SQL + // binds trgeometryFromMFJSON to. + Temporal *t = temporal_from_mfjson(s.c_str(), T_TRGEOMETRY); + if (!t) throw InvalidInputException("fromMFJSON: invalid input"); + return StoreTempAsBlob(result, t); + }); +} + +void TRGeoTypes::RegisterScalarInOutFunctions(ExtensionLoader &loader){ + auto TrgeoAsText = ScalarFunction( + "asText", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::VARCHAR, + Tspatial_as_text + ); + duckdb::RegisterSerializedScalarFunction(loader, TrgeoAsText); + + auto TrgeoAsEWKT = ScalarFunction( + "asEWKT", + {TRGeoTypes::TRGEOMETRY()}, + LogicalType::VARCHAR, + Tspatial_as_ewkt + ); + duckdb::RegisterSerializedScalarFunction(loader, TrgeoAsEWKT); + + // ---- trgeometryFromBinary / FromEWKB (auto-detects format) ---- + const auto B = LogicalType::BLOB; + const auto V = LogicalType::VARCHAR; + const auto T = TRGeoTypes::TRGEOMETRY(); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromBinary", {B}, T, TspatialFromWkbExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromEWKB", {B}, T, TspatialFromWkbExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromHexWKB", {V}, T, TspatialFromHexWkbExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromHexEWKB", {V}, T, TspatialFromHexWkbExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromMFJSON", {V}, T, TrgeoFromMFJSONExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromText", {V}, T, TrgeoFromTextExec)); + duckdb::RegisterSerializedScalarFunction(loader, + ScalarFunction("trgeometryFromEWKT", {V}, T, TrgeoFromTextExec)); +} + + +void TRGeoTypes::RegisterCastFunctions(ExtensionLoader &loader) { + loader.RegisterCastFunction( LogicalType::VARCHAR, TRGeoTypes::TRGEOMETRY(), TrgeoFunctions::StringToTrgeo); + loader.RegisterCastFunction( TRGeoTypes::TRGEOMETRY(), LogicalType::VARCHAR, TrgeoFunctions::TrgeoToString); +} + +} diff --git a/src/include/geo/trgeo.hpp b/src/include/geo/trgeo.hpp new file mode 100644 index 0000000..8c72e36 --- /dev/null +++ b/src/include/geo/trgeo.hpp @@ -0,0 +1,29 @@ +#pragma once + +#include +#include "duckdb/common/exception.hpp" +#include "duckdb/common/string_util.hpp" +#include "duckdb/function/scalar_function.hpp" +#include "duckdb/main/extension/extension_loader.hpp" +#include + +namespace duckdb { + + +struct TRGeoTypes { + static LogicalType TRGEOMETRY(); + static LogicalType GEOMETRY(); + static void RegisterTypes(ExtensionLoader &loader); + static void RegisterScalarFunctions(ExtensionLoader &loader); + static void RegisterCastFunctions(ExtensionLoader &loader); + static void RegisterScalarInOutFunctions(ExtensionLoader &loader); +}; + +struct TrgeoFunctions { + static bool StringToTrgeo(Vector &source, Vector &result, idx_t count, CastParameters ¶meters); + static bool TrgeoToString(Vector &source, Vector &result, idx_t count, CastParameters ¶meters); + static bool WkbBlobToGeometry(Vector &source, Vector &result, idx_t count, CastParameters ¶meters); +}; + + +} // namespace duckdb diff --git a/src/mobilityduck_extension.cpp b/src/mobilityduck_extension.cpp index ba74b65..ac489dc 100644 --- a/src/mobilityduck_extension.cpp +++ b/src/mobilityduck_extension.cpp @@ -18,6 +18,7 @@ #include "geo/tcbuffer.hpp" #include "geo/tnpoint.hpp" #include "geo/tpose.hpp" +#include "geo/trgeo.hpp" #include "geo/tgeogpoint.hpp" #include "geo/tgeogpoint_ops.hpp" #include "temporal/span.hpp" @@ -365,6 +366,11 @@ static void LoadInternal(ExtensionLoader &loader) { TPoseTypes::RegisterCastFunctions(loader); TPoseTypes::RegisterScalarInOutFunctions(loader); + TRGeoTypes::RegisterScalarFunctions(loader); + TRGeoTypes::RegisterTypes(loader); + TRGeoTypes::RegisterCastFunctions(loader); + TRGeoTypes::RegisterScalarInOutFunctions(loader); + TGeogpointType::RegisterScalarFunctions(loader); TGeogpointType::RegisterCastFunctions(loader); TGeogpointType::RegisterScalarInOutFunctions(loader); diff --git a/test/sql/trgeo.test b/test/sql/trgeo.test new file mode 100644 index 0000000..774c5ce --- /dev/null +++ b/test/sql/trgeo.test @@ -0,0 +1,211 @@ +# name: test/sql/trgeo.test +# description: Core trgeo type port — construction, text/EWKT/MFJSON I/O +# and basic accessors. The trgeometry value is a moving rigid +# geometry: a reference geometry rotated and translated over +# time by an embedded temporal pose. +# group: [sql] + +require mobilityduck + +# Test trgeometry constructor with parentheses +query I +SELECT asText(trgeometry('Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01')) IS NOT NULL; +---- +true + +# Test trgeometry constructor without parentheses +query I +SELECT asText(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01') IS NOT NULL; +---- +true + +# Test asText round-trips through the type cast (text -> trgeometry -> text) +query I +SELECT asText(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01') + = asText(trgeometryFromText('Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01')); +---- +true + +# Test asText with continuous sequence +query I +SELECT asText(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02, Pose(Point(1 1),0.5)@2000-01-03]') IS NOT NULL; +---- +true + +# Test asText with discrete sequence +query I +SELECT asText(trgeometry '{Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.3)@2000-01-01, Pose(Point(1 1),0.5)@2000-01-02}') IS NOT NULL; +---- +true + +# Test asEWKT is non-null +query I +SELECT asEWKT(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01') IS NOT NULL; +---- +true + +# Test MFJSON / FromMFJSON round-trip is identity (format-agnostic) +query I +SELECT asText(trgeometryFromMFJSON(asMFJSON(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 2),0.5)@2000-01-01'))) + = asText(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 2),0.5)@2000-01-01'); +---- +true + +# Test MFJSON round-trip for a continuous sequence +query I +SELECT asText(trgeometryFromMFJSON(asMFJSON( + trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 2),0.5)@2000-01-01, Pose(Point(3 4),0.5)@2000-01-02]'))) + = asText(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 2),0.5)@2000-01-01, Pose(Point(3 4),0.5)@2000-01-02]'); +---- +true + +# Test trgeometryFromText constructor is non-null +query I +SELECT asText(trgeometryFromText('Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01')) IS NOT NULL; +---- +true + +# Test binary round-trip is identity +query I +SELECT asText(trgeometryFromBinary(asBinary(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'))) + = asText(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'); +---- +true + +# Test HexWKB round-trip is identity +query I +SELECT asText(trgeometryFromHexWKB(asHexWKB(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'))) + = asText(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'); +---- +true + +# Test timeSpan function +query I +SELECT timeSpan(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-03]'); +---- +[2000-01-01 00:00:00+01, 2000-01-03 00:00:00+01] + +# Test setInterp with discrete interpolation +query I +SELECT tempSubtype(setInterp(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]', 'discrete')); +---- +Sequence + +# Test merge function +query I +SELECT numInstants(merge( + trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.3)@2000-01-02]', + trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.3)@2000-01-02, Pose(Point(1 1),0.5)@2000-01-03]')); +---- +3 + +# Test tempSubtype with instant +query I +SELECT tempSubtype(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'); +---- +Instant + +# Test tempSubtype with discrete sequence +query I +SELECT tempSubtype(trgeometry '{Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.3)@2000-01-01, Pose(Point(1 1),0.5)@2000-01-02}'); +---- +Sequence + +# Test tempSubtype with continuous sequence +query I +SELECT tempSubtype(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]'); +---- +Sequence + +# Test tempSubtype with sequence set +query I +SELECT tempSubtype(trgeometry '{[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02], [Pose(Point(2 2),0.6)@2000-01-04, Pose(Point(2 2),0.6)@2000-01-05]}'); +---- +SequenceSet + +# Test memSize is positive +query I +SELECT memSize(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01') > 0; +---- +true + +# Test interp accessor +query I +SELECT interp(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]'); +---- +Linear + +# Test geom returns the reference geometry (non-null, the rigid body) +query I +SELECT ST_AsText(geom(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01')) IS NOT NULL; +---- +true + +# Test startValue returns the rigid-geometry snapshot at the first instant +query I +SELECT ST_AsText(startValue(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]')) IS NOT NULL; +---- +true + +# Test endValue returns the rigid-geometry snapshot at the last instant +query I +SELECT ST_AsText(endValue(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]')) IS NOT NULL; +---- +true + +# Test startInstant +query I +SELECT asText(startInstant(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]')) IS NOT NULL; +---- +true + +# Test endInstant +query I +SELECT asText(endInstant(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]')) IS NOT NULL; +---- +true + +# Test instantN +query I +SELECT asText(instantN(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]', 1)) IS NOT NULL; +---- +true + +# Test getTimestamp function +query I +SELECT getTimestamp(trgeometry 'Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2023-01-01 10:00:00+00'); +---- +2023-01-01 11:00:00+01 + +# Test numInstants generic accessor +query I +SELECT numInstants(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02, Pose(Point(1 1),0.5)@2000-01-03]'); +---- +3 + +# Test duration generic accessor +query I +SELECT duration(trgeometry '[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-03]'); +---- +2 days + +# Test typed-literal INSERT round-trip (avoids the sequential VARCHAR-cast path) +statement ok +CREATE TABLE trgeo_tbl (rg trgeometry); + +statement ok +INSERT INTO trgeo_tbl VALUES ('Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.5)@2000-01-01'::trgeometry); + +statement ok +INSERT INTO trgeo_tbl VALUES ('[Polygon((1 1,2 2,3 1,1 1));Pose(Point(1 1),0.2)@2000-01-01, Pose(Point(1 1),0.4)@2000-01-02]'::trgeometry); + +query I +SELECT count(*) FROM trgeo_tbl WHERE asText(rg) IS NOT NULL; +---- +2 + +query I +SELECT numInstants(rg) FROM trgeo_tbl ORDER BY numInstants(rg); +---- +1 +2