github-actions[bot] commented on code in PR #69032:
URL: https://github.com/apache/doris/pull/69032#discussion_r4390298187
##########
be/src/exprs/function/geo/functions_geo.cpp:
##########
@@ -18,32 +18,1387 @@
#include "exprs/function/geo/functions_geo.h"
#include <glog/logging.h>
+#include <s2/mutable_s2shape_index.h>
+#include <s2/s2boolean_operation.h>
+#include <s2/s2closest_edge_query.h>
+#include <s2/s2closest_point_query.h>
+#include <s2/s2contains_point_query.h>
+#include <s2/s2earth.h>
+#include <s2/s2lax_polygon_shape.h>
+#include <s2/s2loop.h>
+#include <s2/s2loop_measures.h>
#include <s2/s2point.h>
+#include <s2/s2point_index.h>
+#include <s2/s2point_vector_shape.h>
+#include <s2/s2pointutil.h>
+#include <s2/s2polygon.h>
+#include <s2/s2polyline.h>
+#include <s2/s2predicates.h>
+#include <s2/s2shape_measures.h>
#include <algorithm>
+#include <array>
+#include <bit>
#include <boost/iterator/iterator_facade.hpp>
+// cpp_int needs no quadmath wrappers. Boost's float128 numeric_limits
specialization
+// conflicts with libstdc++ 15's native specialization in the GCC performance
build.
+#if defined(__GNUC__) && !defined(__clang__) && __GNUC__ >= 15
+#define BOOST_CSTDFLOAT_NO_LIBQUADMATH_SUPPORT
+#endif
+#include <boost/multiprecision/cpp_int.hpp>
+#if defined(__GNUC__) && !defined(__clang__) && __GNUC__ >= 15
+#undef BOOST_CSTDFLOAT_NO_LIBQUADMATH_SUPPORT
+#endif
+#include <cmath>
+#include <cstdint>
+#include <limits>
+#include <map>
+#include <numbers>
+#include <optional>
+#include <set>
#include <utility>
+#include "common/cast_set.h"
#include "common/compiler_util.h"
#include "core/assert_cast.h"
#include "core/block/block.h"
#include "core/block/column_with_type_and_name.h"
#include "core/column/column.h"
#include "core/column/column_array.h"
#include "core/column/column_execute_util.h"
+#include "core/column/column_map.h"
#include "core/column/column_nullable.h"
+#include "core/column/column_struct.h"
+#include "core/column/column_varbinary.h"
+#include "core/custom_allocator.h"
#include "core/data_type/data_type_array.h"
+#include "core/data_type/data_type_map.h"
#include "core/data_type/data_type_nullable.h"
#include "core/data_type/data_type_number.h"
+#include "core/data_type/data_type_spatial.h"
#include "core/data_type/data_type_string.h"
+#include "core/data_type/data_type_struct.h"
+#include "core/data_type/data_type_varbinary.h"
#include "core/data_type/define_primitive_type.h"
#include "core/string_ref.h"
#include "exprs/function/geo/geo_common.h"
#include "exprs/function/geo/geo_types.h"
+#include "exprs/function/geo/wkb_parse.h"
#include "exprs/function/simple_function_factory.h"
+#include "exprs/function/string_hex_util.h"
namespace doris {
+static bool is_spatial_type(const DataTypePtr& type) {
+ const auto primitive_type = remove_nullable(type)->get_primitive_type();
+ return primitive_type == TYPE_GEOMETRY || primitive_type == TYPE_GEOGRAPHY;
+}
+
+static Status validate_geography_semantics(const DataTypePtr& type, const
char* function_name) {
+ const auto& nested_type = remove_nullable(type);
+ if (!is_spatial_type(nested_type)) {
+ return Status::OK();
+ }
+
+ const auto* spatial_type = dynamic_cast<const
DataTypeSpatial*>(nested_type.get());
+ DCHECK(spatial_type != nullptr);
+ if (spatial_type->get_primitive_type() == TYPE_GEOGRAPHY &&
+ spatial_type->crs() == "OGC:CRS84" && spatial_type->algorithm() ==
"spherical") {
+ return Status::OK();
+ }
+ return Status::NotSupported(
+ "Function {} requires GEOGRAPHY(OGC:CRS84, spherical) for spatial
inputs",
+ function_name);
+}
+
+static std::unique_ptr<GeoShape> decode_geo_shape(StringRef value, const
DataTypePtr& type) {
+ if (!is_spatial_type(type)) {
+ return GeoShape::from_encoded(value.data, value.size);
+ }
+
+ GeoParseStatus status;
+ auto shape = GeoShape::from_wkb_bytes(value.data, value.size, status);
+ return status == GeoParseStatus::GEO_PARSE_OK ? std::move(shape) : nullptr;
+}
+
+static bool any_geo_argument_is_null(const Block& block, const ColumnNumbers&
arguments,
+ size_t row) {
+ return std::ranges::any_of(arguments, [&](auto argument) {
+ return block.get_by_position(argument).column->is_null_at(row);
+ });
+}
+
+static std::unique_ptr<GeoShape> decode_geography_calculation_shape(StringRef
value);
+
+static Status decode_calculation_shape(StringRef value, const DataTypePtr&
type,
+ std::unique_ptr<GeoShape>* shape,
+ const char* function_name) {
+ *shape = is_spatial_type(type) ? decode_geography_calculation_shape(value)
+ : decode_geo_shape(value, type);
+ // Typed admission rejects unsupported and EMPTY calculation shapes before
distance
+ // helpers can expose infinity/max sentinels. Raw WKB storage/accessors
remain unchanged.
+ if (is_spatial_type(type) && *shape == nullptr) {
+ return Status::NotSupported("Function {} does not support this
Geography WKB shape",
+ function_name);
+ }
+ return Status::OK();
+}
+
+// Calculations use canonical positions; raw spatial accessors still return
the original WKB.
+static S2Point canonical_geography_point(const S2Point& point) {
+ const auto coordinate = S2LatLng(point);
+ const double latitude = coordinate.lat().radians();
+ // All longitudes at a pole and +/-180 at the date line denote the same
position.
+ if (std::abs(latitude) == M_PI_2) {
+ return {0, 0, latitude > 0 ? 1.0 : -1.0};
+ }
+ if (coordinate.lng().radians() == M_PI) {
+ // Flip only sin(longitude): rebuilding latitude can change its last
bit.
+ return {point.x(), -point.y(), point.z()};
+ }
+ return point;
+}
+
+// Error-free TwoSum preserves tiny differences from a half-turn (e.g. -1e-15
+ 180).
+static std::pair<double, double> geography_degree_sum(double a, double b) {
+ const double high = a + b;
+ const double offset = high - a;
+ return {high, (a - (high - offset)) + (b - offset)};
+}
+
+// Reduce the exact degree sum before converting to radians. In particular,
+// 180 + (-1e-300) must retain its small sine instead of becoming sin(pi).
+static std::pair<long double, long double> geography_degree_sincos(double a,
double b = 0,
+ bool half =
false) {
+ const auto [high, low] = geography_degree_sum(a, b);
+ int quadrant;
+ const long double reduced =
+ std::remquo(static_cast<long double>(high), half ? 180.0L : 90.0L,
&quadrant) +
+ static_cast<long double>(low);
+ const auto radians = reduced * (std::numbers::pi_v<long double> / 180) /
(half ? 2 : 1);
+ const auto sine = std::sin(radians);
+ const auto cosine = std::cos(radians);
+ switch ((quadrant % 4 + 4) % 4) {
+ case 0:
+ return {sine, cosine};
+ case 1:
+ return {cosine, -sine};
+ case 2:
+ return {-sine, -cosine};
+ default:
+ return {-cosine, sine};
+ }
+}
+
+// Unlike cross(ToPoint(a), ToPoint(b)), these product identities do not
subtract
+// rounded nearly opposite vectors. Longitude-midpoint coordinates give
+// (-sin(dLon/2)*sin(latA+latB), cos(dLon/2)*sin(latA-latB),
+// cos(latA)*cos(latB)*sin(dLon)) for the oriented normal.
+struct GeographyRawArc {
+ S2Point normal;
+ S2Point midpoint;
+
+ GeographyRawArc(const GeoCoordinate& a, const GeoCoordinate& b) {
+ // Wide exponents retain products of the smallest binary64 degree
residuals.
+ static_assert(std::numeric_limits<long double>::digits >= 64 &&
+ std::numeric_limits<long double>::min_exponent <= -4096);
+ const auto [sd, cd] = geography_degree_sincos(b.x, -a.x, true);
+ const auto [sm, cm] = geography_degree_sincos(a.x, b.x, true);
+ const auto rotate_unit = [&](long double x, long double y, long double
z) {
+ const auto norm = std::sqrt(x * x + y * y + z * z);
+ DCHECK_GT(norm, 0);
+ return S2Point(static_cast<double>((cm * x - sm * y) / norm),
+ static_cast<double>((sm * x + cm * y) / norm),
+ static_cast<double>(z / norm));
+ };
+ normal = rotate_unit(-sd * geography_degree_sincos(a.y, b.y).first,
+ cd * geography_degree_sincos(a.y, -b.y).first,
+ geography_degree_sincos(a.y).second *
+ geography_degree_sincos(b.y).second *
+ geography_degree_sincos(b.x, -a.x).first);
+ const auto [sl, cl] = geography_degree_sincos(a.y, b.y, true);
+ const auto [st, ct] = geography_degree_sincos(b.y, -a.y, true);
+ midpoint = rotate_unit(cd * cl * ct, -sd * sl * st, sl * ct);
+ }
+};
+
+// Raw identity must survive ToPoint rounding: 58 and 58.00000000000001 degrees
+// produce the same S2Point, but only one is an endpoint of the original edge.
+struct GeographyVertex {
+ GeoCoordinate coordinate;
+ S2Point point;
+
+ GeographyVertex() : GeographyVertex({.x = 0, .y = 0}) {}
+ explicit GeographyVertex(GeoCoordinate value) : coordinate(value) {
+ if (std::abs(coordinate.y) == 90) {
+ coordinate.x = 0;
+ point = {0, 0, coordinate.y > 0 ? 1.0 : -1.0};
+ } else {
+ if (coordinate.x == 180) {
+ coordinate.x = -180;
+ }
+ point = S2LatLng::FromDegrees(coordinate.y,
coordinate.x).ToPoint();
+ }
+ }
+ bool operator==(const GeographyVertex& other) const {
+ return coordinate.x == other.coordinate.x && coordinate.y ==
other.coordinate.y;
+ }
+ bool operator<(const GeographyVertex& other) const {
+ return std::pair {coordinate.x, coordinate.y} <
+ std::pair {other.coordinate.x, other.coordinate.y};
+ }
+};
+
+static bool geography_antipodal(const GeographyVertex& a, const
GeographyVertex& b) {
+ return a.coordinate.y == -b.coordinate.y &&
+ (std::abs(a.coordinate.y) == 90 ||
+ geography_degree_sum(std::max(a.coordinate.x, b.coordinate.x),
+ -std::min(a.coordinate.x, b.coordinate.x)) ==
+ std::pair {180.0, 0.0});
+}
+
+struct GeographyCoordinates {
+ GeoShapeType type = GeoShapeType::GEO_SHAPE_ANY;
+ DorisVector<DorisVector<GeographyVertex>> sequences;
+ GeoParseStatus status;
+
+ explicit GeographyCoordinates(StringRef wkb) {
+ GeoCoordinateListList raw;
+ status = WkbParse::calculation_coordinates(wkb.data, wkb.size, &raw,
&type);
+ if (status != GeoParseStatus::GEO_PARSE_OK) {
+ return;
+ }
+ for (const auto& sequence : raw.list) {
+ auto& vertices = sequences.emplace_back();
+ for (const auto& coordinate : sequence->list) {
+ vertices.emplace_back(coordinate);
+ }
+ }
+ }
+};
+
+static bool geography_stable_points(const DorisVector<GeographyVertex>&
sequence,
+ std::vector<S2Point>* points) {
+ for (size_t i = 0; i + 1 < sequence.size(); ++i) {
+ const auto& a = sequence[i];
+ const auto& b = sequence[i + 1];
+ // (0,0)->(180,0) has no unique arc and would normalize a zero raw
normal.
+ if (geography_antipodal(a, b)) {
+ return false;
+ }
+ points->push_back(a.point);
+ // Stabilize before S2 admission: rounded (30,58)->(-150,-58-e) can be
+ // exactly antipodal despite its unique raw arc passing through the
south pole.
+ if (a.point.DotProd(b.point) < 0) {
+ points->push_back(GeographyRawArc(a.coordinate,
b.coordinate).midpoint);
+ }
+ }
+ // A line's final vertex is real, including its closing edge; only rings
remove it.
+ points->push_back(sequence.back().point);
+ points->erase(std::ranges::unique(*points).begin(), points->end());
+ return true;
+}
+
+static std::unique_ptr<GeoShape> decode_geography_line(const
GeographyCoordinates& raw,
+ StringRef value) {
+ std::vector<S2Point> points;
+ if (!geography_stable_points(raw.sequences.front(), &points)) {
+ return nullptr;
+ }
+ if (points.size() < 2) {
+ // Canonical pole aliases such as (0,90),(90,90) collapse to one
calculation
+ // point. Preserve their prior admission; the old parser still rejects
a
+ // truly duplicate-only line such as (1,1),(1,1).
+ GeoParseStatus status;
+ return GeoShape::from_wkb_bytes(value.data, value.size, status);
+ }
+ auto polyline = std::make_unique<S2Polyline>(points, S2Debug::DISABLE);
+ if (!polyline->IsValid()) {
+ return nullptr;
+ }
+ return GeoLine::create_unique(std::move(polyline));
+}
+
+static std::unique_ptr<GeoShape> decode_geography_calculation_shape(StringRef
value) {
+ GeographyCoordinates raw(value);
+ if (raw.status != GeoParseStatus::GEO_PARSE_OK) {
+ return nullptr;
+ }
+ if (raw.type == GeoShapeType::GEO_SHAPE_POINT) {
+ GeoParseStatus status;
+ return GeoShape::from_wkb_bytes(value.data, value.size, status);
+ }
+ if (raw.type == GeoShapeType::GEO_SHAPE_LINE_STRING) {
+ return decode_geography_line(raw, value);
+ }
+ std::vector<std::unique_ptr<S2Loop>> loops;
+ std::vector<const S2Loop*> input_loops;
+ for (const auto& ring : raw.sequences) {
+ std::vector<S2Point> points;
+ // Validate holes against stable raw arcs, as with typed line
admission.
+ if (!geography_stable_points(ring, &points)) {
+ return nullptr;
+ }
+ // Validated WKB rings are closed; only polygon admission removes that
endpoint.
+ DCHECK(points.front() == points.back());
+ points.pop_back();
+ if (points.size() < 3) {
+ return nullptr;
+ }
+ auto loop = std::make_unique<S2Loop>(points, S2Debug::DISABLE);
+ if (!loop->IsValid()) {
+ return nullptr;
+ }
+ loop->Normalize();
+ if (!loops.empty() && !loops.front()->Contains(*loop)) {
+ return nullptr;
+ }
+ input_loops.push_back(loop.get());
+ loops.push_back(std::move(loop));
+ }
+ auto polygon = std::make_unique<S2Polygon>(std::move(loops),
S2Debug::DISABLE);
+ if (!polygon->IsValid()) {
+ return nullptr;
+ }
+ std::vector<int> depths;
+ for (const auto* loop : input_loops) {
+ depths.push_back(loop->depth());
+ }
+ return GeoPolygon::create_unique(std::move(polygon), std::move(depths));
+}
+
+#ifdef BE_TEST
+static thread_local size_t geography_exact_proofs = 0;
+static thread_local size_t geography_candidate_visits = 0;
+#endif
+
+// Test coplanarity of the original degree ordinates, not independently
rounded S2 vectors.
+// For example (45,-45), (90,0), (135,45) share an arc although ToPoint's
determinant is nonzero.
+static bool geography_common_great_circle(const GeoCoordinate& a, const
GeoCoordinate& b,
+ const GeoCoordinate& c) {
+#ifdef BE_TEST
+ ++geography_exact_proofs;
+#endif
+ using boost::multiprecision::cpp_int;
+ const std::array<double, 6> ordinates {a.y, a.x, b.y, b.x, c.y, c.x};
+ std::array<uint64_t, 6> significands;
+ std::array<int, 6> exponents;
+ int denominator_bits = 0;
+ for (size_t i = 0; i < ordinates.size(); ++i) {
+ const auto bits = std::bit_cast<uint64_t>(ordinates[i]);
+ const int biased = (bits >> 52) & 0x7ff;
+ uint64_t significand = bits & ((uint64_t {1} << 52) - 1);
+ int exponent = biased ? biased - 1075 : -1074;
+ if (biased) {
+ significand |= uint64_t {1} << 52;
+ }
+ if (significand != 0) {
+ const int zeros = std::countr_zero(significand);
+ significand >>= zeros;
+ exponent += zeros;
+ denominator_bits = std::max(denominator_bits, -exponent);
+ } else {
+ exponent = 0;
+ }
+ significands[i] = significand;
+ exponents[i] = exponent;
+ }
+ std::array<cpp_int, 6> angles;
+ for (size_t i = 0; i < angles.size(); ++i) {
+ angles[i] = cpp_int(significands[i]) << (denominator_bits +
exponents[i]);
+ if (std::signbit(ordinates[i])) {
+ angles[i] = -angles[i];
+ }
+ }
+ // Binary-rational degrees have denominator 2^q. Exp(i*angle) is a power
of a
+ // root of unity of order 360*2^q. Its minimal polynomial is Phi_90(t),
with
+ // t = z^(2^(q+2)), and Phi_90(t) = t^24+t^21-t^15-t^12-t^9+t^3+1.
+ // Expand cos(lat_i)cos(lat_j)sin(lat_k)sin(lon_j-lon_i), cyclically, into
+ // 48 integer-coefficient powers. Equal residues modulo 2^(q+2) share one
+ // polynomial; reduction by Phi_90 proves zero exactly, including tiny
nonzero controls.
+ const int shift = denominator_bits + 2;
+ const cpp_int basis = cpp_int(1) << shift;
+ const cpp_int period = 90 * basis;
+ std::map<cpp_int, std::array<int, 90>> polynomials;
+ for (int i = 0; i < 3; ++i) {
+ const int j = (i + 1) % 3;
+ const int k = (i + 2) % 3;
+ for (int mask = 0; mask < 16; ++mask) {
+ std::array<int, 4> signs;
+ for (int bit = 0; bit < 4; ++bit) {
+ signs[bit] = (mask & (1 << bit)) ? 1 : -1;
+ }
+ cpp_int power = signs[0] * angles[2 * i] + signs[1] * angles[2 *
j] +
+ signs[2] * angles[2 * k] +
+ signs[3] * (angles[2 * j + 1] - angles[2 * i + 1]);
+ power %= period;
+ if (power < 0) {
+ power += period;
+ }
+ const int quotient = (power >> shift).convert_to<int>();
+ const cpp_int residue = power - (cpp_int(quotient) << shift);
+ polynomials[residue][quotient] += signs[2] * signs[3];
+ }
+ }
+ for (auto& [residue, coefficients] : polynomials) {
+ for (int i = 89; i >= 24; --i) {
+ const int coefficient = coefficients[i];
+ coefficients[i] = 0;
+ coefficients[i - 3] -= coefficient;
+ coefficients[i - 9] += coefficient;
+ coefficients[i - 12] += coefficient;
+ coefficients[i - 15] += coefficient;
+ coefficients[i - 21] -= coefficient;
+ coefficients[i - 24] -= coefficient;
+ }
+ if (std::ranges::any_of(coefficients, [](int value) { return value !=
0; })) {
+ return false;
+ }
+ }
+ return true;
+}
+
+// Keep each atomic segment in its own shape. Pinned S2 ProcessEdge1 uses a
chain's
+// global start as a local offset in one endpoint path, so multiple chains per
line
+// shape are unsafe. Closed AA shapes restore internal vertices in the OPEN
model.
+class GeographyLineShape final : public S2Shape {
+public:
+ GeographyLineShape(const S2Point& a, const S2Point& b) {
_edge.emplace_back(a, b); }
+ int num_edges() const override { return 1; }
+ Edge edge(int i) const override {
+ DCHECK_EQ(i, 0);
+ return _edge.front();
+ }
+ int dimension() const override { return 1; }
+ ReferencePoint GetReferencePoint() const override { return
ReferencePoint::Contained(false); }
+ int num_chains() const override { return 1; }
+ Chain chain(int i) const override {
+ DCHECK_EQ(i, 0);
+ return {0, 1};
+ }
+ Edge chain_edge(int i, int j) const override {
+ DCHECK_EQ(i, 0);
+ return edge(j);
+ }
+ ChainPosition chain_position(int i) const override {
+ DCHECK_EQ(i, 0);
+ return {0, 0};
+ }
+
+private:
+ // Owning storage is checked/accounted by the current query's allocator.
+ DorisVector<Edge> _edge;
+};
+
+class GeographyEdgeNodes {
+public:
+ using LineEdge = std::pair<GeographyVertex, GeographyVertex>;
+ using LineEdges = std::set<LineEdge, std::less<>,
CustomStdAllocator<LineEdge>>;
+ using PolygonEdges = std::array<LineEdges, 2>;
+
+ GeographyCoordinates left;
+ GeographyCoordinates right;
+
+ GeographyEdgeNodes(StringRef left_wkb, StringRef right_wkb) :
left(left_wkb), right(right_wkb) {
+ DCHECK_EQ(left.status, GeoParseStatus::GEO_PARSE_OK);
+ DCHECK_EQ(right.status, GeoParseStatus::GEO_PARSE_OK);
+ DorisMap<double, DorisVector<GeographyVertex>> longitude_vertices;
+ std::set<GeographyVertex, std::less<>,
CustomStdAllocator<GeographyVertex>> coordinates;
+ for (const auto* shape : {&left, &right}) {
+ for (const auto& sequence : shape->sequences) {
+ for (const auto& point : sequence) {
+ if (!coordinates.insert(point).second) {
+ continue;
+ }
+ const auto& coordinate = point.coordinate;
+ longitude_vertices[coordinate.x].push_back(point);
+ if (std::abs(coordinate.y) != 90) {
+ _meridians[meridian(coordinate.x)][coordinate.x <
0].push_back(
+ {coordinate.y, point});
+ }
+ }
+ }
+ }
+ // Keep dense longitude groups separately so a circle can skip its
known
+ // endpoint longitudes before enumerating nearby points. At sqrt(N)
points
+ // per group there are at most sqrt(N) extra indexes, even with
unrelated
+ // vertices padding a near-antipodal spoke input.
+ const auto dense_size =
+ std::max(size_t {2},
static_cast<size_t>(std::ceil(std::sqrt(coordinates.size()))));
+ for (const auto& [longitude, points] : longitude_vertices) {
+ auto& index =
+ points.size() >= dense_size ? _dense_longitudes[longitude]
: _vertex_index;
+ for (const auto& point : points) {
+ index.Add(point.point, point);
+ }
+ }
+ for (auto& entry : _meridians) {
+ for (auto& side : entry.second) {
+ side.push_back({-90, GeographyVertex({.x = 0, .y = -90})});
+ side.push_back({90, GeographyVertex({.x = 0, .y = 90})});
+ std::ranges::sort(side, {}, &Vertex::latitude);
+ }
+ }
+ }
+
+ bool contains_point(const GeographyVertex& start, const GeographyVertex&
end,
+ const GeographyVertex& point) const {
+ // Preserve exact endpoints, including date-line and pole aliases.
+ if (point == start || point == end) {
+ return true;
+ }
+ if (start == end) {
+ return false;
+ }
+ const auto& a = start.coordinate;
+ const auto& b = end.coordinate;
+ const auto& c = point.coordinate;
+ const auto key = meridian(std::abs(a.y) == 90 ? b.x : a.x);
+ bool member = false;
+ if (meridian_edge(start, end, [&](const auto& vertices, double from,
double to) {
+ // Opposite-longitude edges can pass through a pole; the pole
belongs
+ // to both legs, whereas other positions must be on the
selected side.
+ const bool same_side =
+ std::abs(c.y) == 90 ||
+ (meridian(c.x) == key && &_meridians.at(key)[c.x < 0]
== &vertices);
+ member |= same_side && c.y >= std::min(from, to) && c.y <=
std::max(from, to);
+ })) {
+ return member;
+ }
+ const bool wraps = geography_degree_sum(std::max(a.x, b.x),
-std::min(a.x, b.x)) >
+ std::pair {180.0, 0.0};
+ const bool within = wraps ? (c.x < std::min(a.x, b.x) || c.x >
std::max(a.x, b.x))
+ : (c.x > std::min(a.x, b.x) && c.x <
std::max(a.x, b.x));
+ // Near-antipodal endpoint conversion can rotate the S2 plane
substantially.
+ // POINT(45 0) is on (-44.99999999999997 -45,134.99999999999997 45);
+ // decide from raw degrees rather than accepting a larger distance
tolerance.
+ return within && geography_common_great_circle(a, b, c);
+ }
+
+ bool line_contains_point(const GeographyCoordinates& shape,
+ const GeographyVertex& point) const {
+ const auto& line = shape.sequences.front();
+ for (size_t i = 0; i + 1 < line.size(); ++i) {
+ if (contains_point(line[i], line[i + 1], point)) {
+ return true;
+ }
+ }
+ return false;
+ }
+
+ bool line_boundary_touches(const GeographyCoordinates& line,
+ const GeographyCoordinates& other) const {
+ const auto& vertices = line.sequences.front();
+ // Closed lines have no boundary. An open endpoint must belong to the
other raw line:
+ // a rounded coincidence across the gap 58 -> 58+ulp is not a contact.
+ return vertices.front() != vertices.back() &&
+ (line_contains_point(other, vertices.front()) ||
+ line_contains_point(other, vertices.back()));
+ }
+
+ void append_edge(const GeographyVertex& start, const GeographyVertex& end,
+ DorisVector<GeographyVertex>* output) const {
+ output->push_back(start);
+ if (start == end) {
+ return;
+ }
+ if (meridian_edge(start, end, [&](const auto& vertices, double from,
double to) {
+ append_range(vertices, from, to, output);
+ if (std::abs(to) == 90) {
+ const GeographyVertex pole({.x = 0, .y = to});
+ if (end != pole) {
+ output->push_back(pole);
+ }
+ }
+ })) {
+ return;
+ }
+ const auto& a = start.coordinate;
+ const auto& b = end.coordinate;
+ const bool wraps = geography_degree_sum(std::max(a.x, b.x),
-std::min(a.x, b.x)) >
+ std::pair {180.0, 0.0};
+ const bool east = (a.x < b.x) != wraps;
+ // A near-antipodal rounded edge can miss true raw-coordinate nodes.
Two
+ // stable half-arcs recall those nodes without a hemispherical search.
+ const auto midpoint = GeographyRawArc(a, b).midpoint;
+ DorisVector<GeographyVertex> on_edge;
+ for (const auto& endpoint : {start, end}) {
+ visit_edge_candidates(
+ endpoint.point, midpoint, [](double) { return false; },
+ [&](const GeographyVertex& point) {
+ const auto& c = point.coordinate;
+ // Longitude is strictly monotone along every
non-meridian minor arc.
+ // Check the original interval (including date-line
wrap), not a distance
+ // tolerance that could append a point just beyond an
endpoint.
+ const bool within =
+ wraps ? (c.x < std::min(a.x, b.x) || c.x >
std::max(a.x, b.x))
+ : (c.x > std::min(a.x, b.x) && c.x <
std::max(a.x, b.x));
+ if (within && geography_common_great_circle(a, b, c)) {
+ on_edge.push_back(point);
+ }
+ });
+ }
+ std::ranges::sort(on_edge, [&](const GeographyVertex& x, const
GeographyVertex& y) {
+ const double x_lon = x.coordinate.x;
+ const double y_lon = y.coordinate.x;
+ if ((x_lon < a.x) != (y_lon < a.x)) {
+ return east ? x_lon >= a.x : x_lon < a.x;
+ }
+ return east ? x_lon < y_lon : x_lon > y_lon;
+ });
+ on_edge.erase(std::ranges::unique(on_edge).begin(), on_edge.end());
+ output->insert(output->end(), on_edge.begin(), on_edge.end());
+ }
+
+ // Calculation-only points never enter the raw-degree identity/noding map.
+ std::optional<S2Point> arc_midpoint(const GeographyVertex& a, const
GeographyVertex& b) const {
+ if (a.point.DotProd(b.point) < 0) {
+ return GeographyRawArc(a.coordinate, b.coordinate).midpoint;
+ }
+ return std::nullopt;
+ }
+
+ void add_line(const GeographyCoordinates& shape, MutableS2ShapeIndex*
index,
+ LineEdges* raw_line_edges = nullptr) const {
+ using Interval = std::pair<double, double>;
+ DorisMap<const DorisVector<Vertex>*, DorisVector<Interval>> intervals;
+ LineEdges raw_edges;
+ const auto& line = shape.sequences.front();
+ for (size_t i = 0; i + 1 < line.size(); ++i) {
+ const auto& a = line[i];
+ const auto& b = line[i + 1];
+ if (a == b) {
+ continue;
+ }
+ if (!meridian_edge(a, b, [&](const auto& vertices, double from,
double to) {
+ intervals[&vertices].emplace_back(std::min(from, to),
std::max(from, to));
+ })) {
+ raw_edges.emplace(std::min(a, b), std::max(a, b));
+ }
+ }
+ LineEdges edges;
+ auto add_edge = [&](const GeographyVertex& a, const GeographyVertex&
b) {
+ // Equality must distinguish (0,58)->(0,59) from
(0,58+ulp)->(0,59),
+ // even when conversion collapses the raw 58->58+ulp subedge to
one S2Point.
+ if (raw_line_edges != nullptr) {
+ raw_line_edges->emplace(std::min(a, b), std::max(a, b));
+ }
+ // Rounded adjacent vertices can share one S2Point while remaining
distinct raw
+ // WKB vertices, for example 58 and 58.00000000000001 degrees. The
raw edge is
+ // retained for identity, but its zero-length S2 representation
must not become
+ // an OPEN/AA boundary marker at the original line endpoint.
+ if (a.point == b.point) {
+ return;
+ }
+ edges.emplace(std::min(a, b), std::max(a, b));
+ };
+ // Merge latitude coverage before splitting: N repeated or overlapping
edges
+ // and M collinear vertices produce O(N+M) scratch, never an N*M
vertex walk.
+ for (auto& [vertices, ranges] : intervals) {
+ std::ranges::sort(ranges);
+ auto emit = [&](double from, double to) {
+ auto begin = std::ranges::lower_bound(*vertices, from, {},
&Vertex::latitude);
+ const auto end = std::ranges::upper_bound(*vertices, to, {},
&Vertex::latitude);
+ for (auto next = begin + 1; next != end; ++begin, ++next) {
+ add_edge(begin->point, next->point);
+ }
+ };
+ double from = ranges.front().first;
+ double to = ranges.front().second;
+ for (const auto& range : ranges) {
+ if (range.first > to) {
+ emit(from, to);
+ from = range.first;
+ }
+ to = std::max(to, range.second);
+ }
+ emit(from, to);
+ }
+ add_circle_edges(raw_edges, add_edge);
+ add_line_shapes(line, edges, index);
+ }
+
+private:
+ void add_line_shapes(const DorisVector<GeographyVertex>& line, const
LineEdges& edges,
+ MutableS2ShapeIndex* index) const {
+ const auto& first = line.front();
+ const auto& last = line.back();
+ auto add = [&](const S2Point& a, const S2Point& b) {
+ index->Add(std::make_unique<GeographyLineShape>(a, b));
+ };
+ if (edges.empty()) {
+ // A raw pole-alias line may collapse to one canonical calculation
position.
+ add(first.point, first.point);
+ }
+ std::set<S2Point, std::less<>, CustomStdAllocator<S2Point>>
interior_vertices;
+ for (const auto& [a, b] : edges) {
+ const auto midpoint = arc_midpoint(a, b);
+ if (midpoint) {
+ add(a.point, *midpoint);
+ add(*midpoint, b.point);
+ // The synthetic split is interior, not a new OPEN line
boundary:
+ // a crossing at this point must not become ST_Touches=true.
+ interior_vertices.insert(*midpoint);
+ } else {
+ add(a.point, b.point);
+ }
+ interior_vertices.insert(a.point);
+ interior_vertices.insert(b.point);
+ }
+ for (const auto& point : interior_vertices) {
+ if (first == last || (point != first.point && point !=
last.point)) {
+ add(point, point);
+ }
+ }
+ }
+
+ struct Vertex {
+ double latitude;
+ GeographyVertex point;
+ };
+ struct Circle {
+ GeographyVertex a;
+ GeographyVertex b;
+ S2Point normal;
+ DorisVector<std::pair<double, double>> ranges;
+ DorisMap<double, GeographyVertex> vertices;
+ };
+
+ template <typename Edges, typename AddEdge>
+ void add_circle_edges(const Edges& edges, AddEdge&& add_edge) const {
+ DorisVector<Circle> circles;
+ S2PointIndex<size_t> normal_index;
+ const double rounding = 64 * std::numeric_limits<double>::epsilon();
+ for (const auto& [a, b] : edges) {
+ const auto normal = GeographyRawArc(a.coordinate,
b.coordinate).normal;
+ size_t group = circles.size();
+ S2ClosestPointQuery<size_t> query(&normal_index);
+ // Raw trigonometric products avoid the epsilon/sin(length) error
+ // amplification. Normals only recall candidates; exact proof
decides.
+
query.mutable_options()->set_conservative_max_distance(S1Angle::Radians(rounding));
+ S2ClosestPointQuery<size_t>::PointTarget target(normal);
+ for (const auto& candidate : query.FindClosestPoints(&target)) {
+#ifdef BE_TEST
+ ++geography_candidate_visits;
+#endif
+ const auto& circle = circles[candidate.data()];
+ const auto& x = circle.a.coordinate;
+ const auto& y = circle.b.coordinate;
+ const auto on_circle = [&](const GeographyVertex& point) {
+ return point == circle.a || point == circle.b ||
+ geography_common_great_circle(x, y,
point.coordinate);
+ };
+ if (on_circle(a) && on_circle(b)) {
+ group = candidate.data();
+ break;
+ }
+ }
+ if (group == circles.size()) {
+ normal_index.Add(normal, group);
+ normal_index.Add(-normal, group);
+ circles.push_back({a, b, normal, {}, {}});
+ }
+ auto& circle = circles[group];
+ const double lo = std::min(a.coordinate.x, b.coordinate.x);
+ const double hi = std::max(a.coordinate.x, b.coordinate.x);
+ // Longitude cut points are interval sentinels, not fabricated
spatial
+ // vertices. For 170->-170 the real wrap edge is emitted below.
+ if (geography_degree_sum(hi, -lo) > std::pair {180.0, 0.0}) {
+ circle.ranges.emplace_back(-180, lo);
+ circle.ranges.emplace_back(hi, 180);
+ } else {
+ circle.ranges.emplace_back(lo, hi);
+ }
+ // Keep every raw endpoint, especially original open-line endpoints
+ // inside a larger merged interval: OPEN must still exclude those
points.
+ circle.vertices.emplace(a.coordinate.x, a);
+ circle.vertices.emplace(b.coordinate.x, b);
+ }
+ for (auto& circle : circles) {
+ collect_circle_vertices(circle);
+ emit_circle_edges(circle, add_edge);
+ }
+ }
+
+ template <typename SkipLongitude, typename Visit>
+ void visit_edge_candidates(const S2Point& a, const S2Point& b,
SkipLongitude&& skip_longitude,
+ Visit&& visit) const {
+ const auto query_index = [&](const S2PointIndex<GeographyVertex>&
index) {
+ S2ClosestPointQuery<GeographyVertex> query(&index);
+ query.mutable_options()->set_conservative_max_distance(
+ S1Angle::Radians(64 *
std::numeric_limits<double>::epsilon()));
+ S2ClosestPointQuery<GeographyVertex>::EdgeTarget target(a, b);
+ for (const auto& candidate : query.FindClosestPoints(&target)) {
+#ifdef BE_TEST
+ ++geography_candidate_visits;
+#endif
+ visit(candidate.data());
+ }
+ };
+ query_index(_vertex_index);
+ for (const auto& [longitude, index] : _dense_longitudes) {
+#ifdef BE_TEST
+ ++geography_candidate_visits;
+#endif
+ // A non-meridian circle has only one nonpolar point at each
longitude.
+ // Dense endpoint groups must be skipped before the S2 query,
including
+ // when unrelated vertices make them a minority of the complete
input.
+ if (!skip_longitude(longitude)) {
+ query_index(index);
+ }
+ }
+ }
+
+ void collect_circle_vertices(Circle& circle) const {
+ // Query a whole circle once, using four non-antipodal quarter arcs.
+ const auto u = S2::Ortho(circle.normal);
+ const auto v = circle.normal.CrossProd(u).Normalize();
+ const std::array<S2Point, 4> quarter {u, v, -u, -v};
+ std::set<GeographyVertex, std::less<>,
CustomStdAllocator<GeographyVertex>> seen;
+ for (size_t i = 0; i < quarter.size(); ++i) {
+ visit_edge_candidates(
+ quarter[i], quarter[(i + 1) % 4],
+ [&](double longitude) { return
circle.vertices.contains(longitude); },
+ [&](const GeographyVertex& point) {
+ const auto& coordinate = point.coordinate;
+ if (seen.insert(point).second &&
!circle.vertices.contains(coordinate.x) &&
+ geography_common_great_circle(circle.a.coordinate,
circle.b.coordinate,
+ coordinate)) {
+ circle.vertices.emplace(coordinate.x, point);
+ }
+ });
+ }
+ }
+
+ template <typename AddEdge>
+ static void emit_circle_edges(Circle& circle, AddEdge&& add_edge) {
+ std::ranges::sort(circle.ranges);
+ DorisVector<std::pair<double, double>> merged;
+ for (const auto& range : circle.ranges) {
+ if (merged.empty() || range.first > merged.back().second) {
+ merged.push_back(range);
+ } else {
+ merged.back().second = std::max(merged.back().second,
range.second);
+ }
+ }
+ for (const auto& [lo, hi] : merged) {
+ const auto begin = circle.vertices.lower_bound(lo);
+ const auto end = circle.vertices.upper_bound(hi);
+ // Each original interval retains an endpoint, including either
side
+ // of the date-line cut. Merging and node collection never remove
it.
+ DCHECK(begin != end);
+ auto previous = begin;
+ for (auto next = std::next(begin); next != end; ++previous,
++next) {
+ add_edge(previous->second, next->second);
+ }
+ }
+ if (merged.front().first == -180 && merged.back().second == 180) {
+ add_edge(circle.vertices.rbegin()->second,
circle.vertices.begin()->second);
+ }
+ }
+
+ static std::pair<double, double> meridian(double longitude) {
+ return longitude >= 0 ? std::pair {longitude, 0.0} :
geography_degree_sum(longitude, 180);
+ }
+ template <typename Append>
+ bool meridian_edge(const GeographyVertex& start, const GeographyVertex&
end,
+ Append&& append) const {
+ const auto& a = start.coordinate;
+ const auto& b = end.coordinate;
+ const bool a_pole = std::abs(a.y) == 90;
+ const bool b_pole = std::abs(b.y) == 90;
+ const auto key = meridian(a_pole ? b.x : a.x);
+ if (!a_pole && !b_pole && key != meridian(b.x)) {
+ return false;
+ }
+ // Exact opposite endpoints (e.g. (0,10),(180,-10)) have no unique
minor arc.
+ if (a.y == -b.y && ((a_pole && b_pole) || (a.x < 0) != (b.x < 0))) {
+ throw Exception(ErrorCode::NOT_IMPLEMENTED_ERROR,
+ "Geography calculations do not support antipodal
edges");
+ }
+ const bool a_side = (a_pole ? b.x : a.x) < 0;
+ const bool b_side = (b_pole ? a.x : b.x) < 0;
+ const auto& group = _meridians.at(key);
+ if (a_side == b_side) {
+ append(group[a_side], a.y, b.y);
+ } else {
+ // (10,80)->(-170,80) goes via the north pole; retain raw latitude
on each side.
+ const double latitude = a.y > -b.y ? 90 : -90;
+ append(group[a_side], a.y, latitude);
+ append(group[b_side], latitude, b.y);
+ }
+ return true;
+ }
+ static void append_range(const DorisVector<Vertex>& vertices, double from,
double to,
+ DorisVector<GeographyVertex>* output) {
+ const auto first =
+ std::ranges::upper_bound(vertices, std::min(from, to), {},
&Vertex::latitude);
+ const auto last =
+ std::ranges::lower_bound(vertices, std::max(from, to), {},
&Vertex::latitude);
+ if (from < to) {
+ for (auto it = first; it != last; ++it) {
+ output->push_back(it->point);
+ }
+ } else {
+ auto current = last;
+ while (current != first) {
+ output->push_back((--current)->point);
+ }
+ }
+ }
+ DorisMap<double, S2PointIndex<GeographyVertex>> _dense_longitudes;
+ S2PointIndex<GeographyVertex> _vertex_index;
+ DorisMap<std::pair<double, double>, std::array<DorisVector<Vertex>, 2>>
_meridians;
+};
+
+static void geography_shape_index(const GeoShape& shape, const
GeographyCoordinates& raw,
+ const GeographyEdgeNodes& nodes,
MutableS2ShapeIndex* index,
+ GeographyEdgeNodes::PolygonEdges*
raw_polygon_edges = nullptr) {
+ switch (shape.type()) {
+ case GeoShapeType::GEO_SHAPE_POINT:
+ index->Add(std::make_unique<S2PointVectorShape>(
+ std::vector<S2Point> {raw.sequences.front().front().point}));
+ break;
+ case GeoShapeType::GEO_SHAPE_LINE_STRING:
+ nodes.add_line(raw, index);
+ break;
+ case GeoShapeType::GEO_SHAPE_POLYGON: {
+ const auto& depths = assert_cast<const
GeoPolygon&>(shape).calculation_ring_depths();
+ DCHECK_EQ(depths.size(), raw.sequences.size());
+ std::vector<S2LaxPolygonShape::Loop> loops;
+ for (size_t i = 0; i < raw.sequences.size(); ++i) {
+ auto& vertices = loops.emplace_back();
+ const auto& ring = raw.sequences[i];
+ DorisVector<GeographyVertex> edge;
+ for (size_t j = 0; j + 1 < ring.size(); ++j) {
+ edge.clear();
+ nodes.append_edge(ring[j], ring[j + 1], &edge);
+ edge.push_back(ring[j + 1]);
+ for (size_t k = 0; k + 1 < edge.size(); ++k) {
+ // Keep raw apex 58 distinct from 58+ulp before S2
conversion. Depth parity
+ // preserves shell/hole roles; duplicate aliases and
synthetic points add no edge.
+ if (raw_polygon_edges != nullptr && edge[k] != edge[k +
1]) {
+ (*raw_polygon_edges)[depths[i] %
2].emplace(std::min(edge[k], edge[k + 1]),
+
std::max(edge[k], edge[k + 1]));
+ }
+ vertices.push_back(edge[k].point);
+ // Use the same stable long arcs for interiors and
distances.
+ // The raw (-45+e,-45)->(135-e,45) edge otherwise changes
sides
+ // around POINT(45 0.000001) when its midpoint is inserted.
+ if (const auto midpoint = nodes.arc_midpoint(edge[k],
edge[k + 1])) {
+ vertices.push_back(*midpoint);
+ }
+ }
+ }
+ // A thin near-antipodal ring may have been reversed by
normalization
+ // of the rounded edge. Restore the smaller-region shell
orientation
+ // after stabilization while preserving the admitted hole depths.
+ // Span normalization also retains canonical pole-alias
degeneracies.
+ if (S2::IsNormalized(vertices) == (depths[i] % 2 != 0)) {
+ std::ranges::reverse(vertices);
+ }
+ }
+ index->Add(std::make_unique<S2LaxPolygonShape>(loops));
+ break;
+ }
+ default:
+ throw Exception(ErrorCode::INTERNAL_ERROR, "Unexpected Geography
calculation shape");
+ }
+}
+
+#ifdef BE_TEST
+size_t geography_calculation_edges_for_test(const std::string& left, const
std::string& right,
+ size_t* exact_proofs, size_t*
candidate_visits) {
+ geography_exact_proofs = 0;
+ geography_candidate_visits = 0;
+ const auto a = decode_geography_calculation_shape(StringRef(left));
+ const auto b = decode_geography_calculation_shape(StringRef(right));
+ const GeographyEdgeNodes nodes {StringRef(left), StringRef(right)};
+ MutableS2ShapeIndex a_index;
+ MutableS2ShapeIndex b_index;
+ geography_shape_index(*a, nodes.left, nodes, &a_index);
+ geography_shape_index(*b, nodes.right, nodes, &b_index);
+ size_t edges = 0;
+ for (const auto* shape : a_index) {
+ edges += shape->num_edges();
+ }
+ for (const auto* shape : b_index) {
+ edges += shape->num_edges();
+ }
+ if (exact_proofs != nullptr) {
+ *exact_proofs = geography_exact_proofs;
+ }
+ if (candidate_visits != nullptr) {
+ *candidate_visits = geography_candidate_visits;
+ }
+ return edges;
+}
+#endif
+
+static S1ChordAngle geography_conversion_error() {
+ // S2LatLng::ToPoint and degrees-to-radians rounding can separate
positions on
+ // the same meridian. This is about 1e-8 metres, not a user snap radius.
+ return S1ChordAngle(S1Angle::Radians(8 *
std::numeric_limits<double>::epsilon()));
+}
+
+static bool geography_point_relation(std::string_view name, const GeoShape&
lhs,
+ const GeoShape& rhs, const
GeographyEdgeNodes& nodes) {
+ const auto& other = lhs.type() == GeoShapeType::GEO_SHAPE_POINT ? rhs :
lhs;
+ const auto& point_raw = lhs.type() == GeoShapeType::GEO_SHAPE_POINT ?
nodes.left : nodes.right;
+ const auto& other_raw = lhs.type() == GeoShapeType::GEO_SHAPE_POINT ?
nodes.right : nodes.left;
+ const auto& point = point_raw.sequences.front().front();
+ bool interior = false;
+ bool boundary = false;
+ if (other.type() == GeoShapeType::GEO_SHAPE_POINT) {
+ interior = point == other_raw.sequences.front().front();
+ } else if (other.type() == GeoShapeType::GEO_SHAPE_LINE_STRING) {
+ const auto& line = other_raw.sequences.front();
+ const bool member = nodes.line_contains_point(other_raw, point);
+ const auto& first = line.front();
+ const auto& last = line.back();
+ boundary = first != last && (point == first || point == last);
+ interior = member && !boundary;
+ } else {
+ DCHECK(other.type() == GeoShapeType::GEO_SHAPE_POLYGON);
+ for (const auto& ring : other_raw.sequences) {
+ for (size_t j = 0; j + 1 < ring.size(); ++j) {
+ boundary = boundary || nodes.contains_point(ring[j], ring[j +
1], point);
+ }
+ }
+ if (!boundary) {
+ MutableS2ShapeIndex index;
+ geography_shape_index(other, other_raw, nodes, &index);
+ interior = MakeS2ContainsPointQuery(&index).Contains(point.point);
+ }
+ }
+ if (name == "st_intersects") {
+ return interior || boundary;
+ }
+ if (name == "st_disjoint") {
+ return !interior && !boundary;
+ }
+ if (name == "st_touches") {
+ return boundary;
+ }
+ return interior;
+}
+
+static Status geography_relation(std::string_view name, const GeoShape& lhs,
const GeoShape& rhs,
+ StringRef left_wkb, StringRef right_wkb,
bool* result) {
+ const auto& container = name == "st_within" ? rhs : lhs;
+ const auto& value = name == "st_within" ? lhs : rhs;
+ const bool containment = name == "st_contains" || name == "st_within";
+ // A point cannot contain a line or polygon; this is a defined false
result.
+ if (containment && container.type() == GeoShapeType::GEO_SHAPE_POINT &&
+ value.type() != GeoShapeType::GEO_SHAPE_POINT) {
+ *result = false;
+ return Status::OK();
+ }
+ if (name == "st_equals" && lhs.type() != rhs.type()) {
+ *result = false;
+ return Status::OK();
+ }
+ if (name == "st_overlaps" &&
+ (lhs.type() != rhs.type() || lhs.type() ==
GeoShapeType::GEO_SHAPE_POINT)) {
+ *result = false;
+ return Status::OK();
+ }
+ if (lhs.type() == GeoShapeType::GEO_SHAPE_POINT ||
+ rhs.type() == GeoShapeType::GEO_SHAPE_POINT) {
+ // Point equality has the same canonical position test as point
containment.
+ const GeographyEdgeNodes nodes(left_wkb, right_wkb);
+ *result = geography_point_relation(name, lhs, rhs, nodes);
+ return Status::OK();
+ }
+ MutableS2ShapeIndex left;
+ MutableS2ShapeIndex right;
+ const GeographyEdgeNodes nodes(left_wkb, right_wkb);
+ GeographyEdgeNodes::LineEdges left_edges;
+ GeographyEdgeNodes::LineEdges right_edges;
+ GeographyEdgeNodes::PolygonEdges left_polygon_edges;
+ GeographyEdgeNodes::PolygonEdges right_polygon_edges;
+ const bool raw_polygon_coverage = lhs.type() ==
GeoShapeType::GEO_SHAPE_POLYGON &&
+ rhs.type() ==
GeoShapeType::GEO_SHAPE_POLYGON &&
+ (containment || name == "st_equals" ||
name == "st_overlaps");
+ // (0,58)->(0,59) covers a longer raw interval than (0,58+ulp)->(0,59),
+ // even when their S2 endpoints collide. Reuse the shared noding for these
relations.
+ const bool raw_line_coverage =
+ container.type() == GeoShapeType::GEO_SHAPE_LINE_STRING &&
+ value.type() == GeoShapeType::GEO_SHAPE_LINE_STRING &&
+ (containment || name == "st_equals" || name == "st_overlaps" ||
name == "st_touches");
+ if (raw_line_coverage) {
+ nodes.add_line(name == "st_within" ? nodes.right : nodes.left, &left,
&left_edges);
+ nodes.add_line(name == "st_within" ? nodes.left : nodes.right, &right,
&right_edges);
+ } else {
+ geography_shape_index(container, name == "st_within" ? nodes.right :
nodes.left, nodes,
+ &left, raw_polygon_coverage ?
&left_polygon_edges : nullptr);
+ geography_shape_index(value, name == "st_within" ? nodes.left :
nodes.right, nodes, &right,
+ raw_polygon_coverage ? &right_polygon_edges :
nullptr);
+ }
+ S2BooleanOperation::Options options;
+ options.set_polygon_model(S2BooleanOperation::PolygonModel::CLOSED);
+ options.set_polyline_model(S2BooleanOperation::PolylineModel::CLOSED);
+ options.set_polyline_loops_have_boundaries(false);
+ const auto shares_raw_line_edge = [&] {
+ return std::ranges::any_of(left_edges,
+ [&](const auto& edge) { return
right_edges.contains(edge); });
+ };
+ if (name == "st_equals") {
+ *result = left_edges == right_edges && left_polygon_edges ==
right_polygon_edges &&
+ S2BooleanOperation::Equals(left, right, options);
+ return Status::OK();
+ }
+ // Different raw triangle apexes can produce the same S2 region. Boundary
identity decides
+ // equality, but cannot decide which raw region contains the other; do not
accept a false bool.
+ if (raw_polygon_coverage && left_polygon_edges != right_polygon_edges &&
+ S2BooleanOperation::Equals(left, right, options)) {
+ return Status::NotSupported(
+ "{} does not support distinct raw Polygon boundaries with the
same spherical "
+ "representation",
+ name);
+ }
+ if (containment && container.type() == GeoShapeType::GEO_SHAPE_LINE_STRING
&&
+ value.type() == GeoShapeType::GEO_SHAPE_POLYGON) {
+ *result = false;
+ return Status::OK();
+ }
+ const auto covers = [&](bool reverse) {
+ const auto& a = reverse ? right_edges : left_edges;
+ const auto& b = reverse ? left_edges : right_edges;
+ return (!raw_line_coverage || std::ranges::includes(a, b)) &&
+ S2BooleanOperation::Contains(reverse ? right : left, reverse ?
left : right,
+ options);
+ };
+ // Rounded indexes alone would let the shorter raw line contain the longer
one.
+ if (containment && !covers(false)) {
+ *result = false;
+ return Status::OK();
+ }
+ auto interior_options = options;
+ interior_options.set_polygon_model(S2BooleanOperation::PolygonModel::OPEN);
+
interior_options.set_polyline_model(S2BooleanOperation::PolylineModel::OPEN);
+ if (containment) {
+ // CLOSED Contains is covers: a polygon must not contain a line solely
on its boundary.
+ *result = S2BooleanOperation::Intersects(left, right,
interior_options);
+ return Status::OK();
+ }
+ if (name == "st_overlaps") {
+ // Neither equality nor containment is a partial overlap.
+ if (covers(false) || covers(true)) {
+ *result = false;
+ return Status::OK();
+ }
+ if (lhs.type() == GeoShapeType::GEO_SHAPE_POLYGON) {
+ *result = S2BooleanOperation::Intersects(left, right,
interior_options);
+ return Status::OK();
+ }
+ // A crossing point has no shared raw edge. Conversely, a positive raw
overlap
+ // such as (0,58)->(0,58+ulp) must survive even when S2 rounds it to
one point.
+ *result = shares_raw_line_edge();
+ return Status::OK();
+ }
+ bool intersects = S2BooleanOperation::Intersects(left, right, options);
+ // A nearby disjoint curve must not become a boundary contact through the
legacy
+ // conversion-distance fallback used below by Intersects/Disjoint.
+ if (name == "st_touches") {
+ *result = intersects && !S2BooleanOperation::Intersects(left, right,
interior_options);
+ // CLOSED/OPEN alone treats a raw gap or a positive overlap 58->58+ulp
as one
+ // endpoint contact. A shared raw edge has intersecting interiors, so
cannot touch.
+ if (*result && lhs.type() == GeoShapeType::GEO_SHAPE_LINE_STRING &&
Review Comment:
[P2] Validate raw polygon contact before accepting Touches
For default GEOGRAPHY imported as WKB, let L=`LINESTRING(0 57,0 58)` and
P=`POLYGON((0 58.00000000000001,1 59,-1 59,0 58.00000000000001))`. They are
disjoint: L ends below P's southernmost latitude. The pinned [degree
conversion](https://github.com/google/s2geometry/blob/v0.10.0/src/s2/s1angle.h#L290-L292)
nevertheless maps 58 and 58.00000000000001 to the same radians
(`0x1.03258031ade7cp+0`), so L's endpoint and P's apex become identical
S2Points.
The short polygon edges get no midpoint, and `append_edge` excludes L's
same-longitude endpoint from the sloped edges. Thus the rounded CLOSED indexes
meet only at that apex, while their OPEN interiors are disjoint under the [S2
boundary
models](https://github.com/google/s2geometry/blob/v0.10.0/src/s2/s2boolean_operation.h#L97-L140).
The test above accepts contact, and this raw-contact correction is skipped
because the operands are not both lines. `ST_Touches(L,P)` and its reverse
therefore return true for disjoint inputs.
Confirm original-coordinate polygon boundary contact, or reject unresolved
rounded contacts explicitly, and cover both orders plus true-contact controls.
This is the mixed line/polygon Touches consumer: existing 4388539598's
line-pair fix runs for two lines, and 4388539591's polygon equality/containment
guard does not run here. Evidence is source and binary64 arithmetic, not a
Doris runtime test.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
