mirror of https://gitee.com/bigwinds/arangodb
408 lines
14 KiB
C++
408 lines
14 KiB
C++
////////////////////////////////////////////////////////////////////////////////
|
|
/// DISCLAIMER
|
|
///
|
|
/// Copyright 2014-2016 ArangoDB GmbH, Cologne, Germany
|
|
/// Copyright 2004-2014 triAGENS GmbH, Cologne, Germany
|
|
///
|
|
/// Licensed under the Apache License, Version 2.0 (the "License");
|
|
/// you may not use this file except in compliance with the License.
|
|
/// You may obtain a copy of the License at
|
|
///
|
|
/// http://www.apache.org/licenses/LICENSE-2.0
|
|
///
|
|
/// Unless required by applicable law or agreed to in writing, software
|
|
/// distributed under the License is distributed on an "AS IS" BASIS,
|
|
/// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
/// See the License for the specific language governing permissions and
|
|
/// limitations under the License.
|
|
///
|
|
/// Copyright holder is ArangoDB GmbH, Cologne, Germany
|
|
///
|
|
/// @author Dr. Frank Celler
|
|
////////////////////////////////////////////////////////////////////////////////
|
|
|
|
#include "MMFilesGeoIndex.h"
|
|
|
|
#include "Aql/Ast.h"
|
|
#include "Aql/AstNode.h"
|
|
#include "Aql/SortCondition.h"
|
|
#include "Basics/StaticStrings.h"
|
|
#include "Basics/VelocyPackHelper.h"
|
|
#include "Geo/GeoUtils.h"
|
|
#include "GeoIndex/Near.h"
|
|
#include "Indexes/IndexIterator.h"
|
|
#include "Logger/Logger.h"
|
|
#include "VocBase/LogicalCollection.h"
|
|
|
|
#include <velocypack/Iterator.h>
|
|
#include <velocypack/StringRef.h>
|
|
#include <velocypack/velocypack-aliases.h>
|
|
|
|
using namespace arangodb;
|
|
|
|
template <typename CMP = geo_index::DocumentsAscending>
|
|
struct NearIterator final : public IndexIterator {
|
|
/// @brief Construct an RocksDBGeoIndexIterator based on Ast Conditions
|
|
NearIterator(LogicalCollection* collection, transaction::Methods* trx,
|
|
MMFilesGeoIndex const* index, geo::QueryParams&& params)
|
|
: IndexIterator(collection, trx), _index(index), _near(std::move(params)) {
|
|
estimateDensity();
|
|
}
|
|
|
|
~NearIterator() {}
|
|
|
|
char const* typeName() const override { return "s2-index-iterator"; }
|
|
|
|
/// internal retrieval loop
|
|
template <typename F>
|
|
inline bool nextToken(F&& cb, size_t limit) {
|
|
if (_near.isDone()) {
|
|
// we already know that no further results will be returned by the index
|
|
TRI_ASSERT(!_near.hasNearest());
|
|
return false;
|
|
}
|
|
|
|
while (limit > 0 && !_near.isDone()) {
|
|
while (limit > 0 && _near.hasNearest()) {
|
|
if (std::forward<F>(cb)(_near.nearest())) {
|
|
limit--;
|
|
}
|
|
_near.popNearest();
|
|
}
|
|
// need to fetch more geo results
|
|
if (limit > 0 && !_near.isDone()) {
|
|
TRI_ASSERT(!_near.hasNearest());
|
|
performScan();
|
|
}
|
|
}
|
|
return !_near.isDone();
|
|
}
|
|
|
|
bool nextDocument(DocumentCallback const& cb, size_t limit) override {
|
|
return nextToken(
|
|
[this, &cb](geo_index::Document const& gdoc) -> bool {
|
|
bool result = true; // updated by the callback
|
|
if (!_collection->readDocumentWithCallback(_trx, gdoc.token, [&](LocalDocumentId const&, VPackSlice doc) {
|
|
geo::FilterType const ft = _near.filterType();
|
|
if (ft != geo::FilterType::NONE) { // expensive test
|
|
geo::ShapeContainer const& filter = _near.filterShape();
|
|
TRI_ASSERT(!filter.empty());
|
|
geo::ShapeContainer test;
|
|
Result res = _index->shape(doc, test);
|
|
TRI_ASSERT(res.ok() && !test.empty()); // this should never fail here
|
|
if (res.fail() ||
|
|
(ft == geo::FilterType::CONTAINS && !filter.contains(&test)) ||
|
|
(ft == geo::FilterType::INTERSECTS && !filter.intersects(&test))) {
|
|
result = false; // skip
|
|
return;
|
|
}
|
|
}
|
|
cb(gdoc.token, doc); // return result
|
|
result = true;
|
|
})) {
|
|
return false; // skip
|
|
}
|
|
return result;
|
|
},
|
|
limit);
|
|
}
|
|
|
|
bool next(LocalDocumentIdCallback const& cb, size_t limit) override {
|
|
return nextToken(
|
|
[this, &cb](geo_index::Document const& gdoc) -> bool {
|
|
geo::FilterType const ft = _near.filterType();
|
|
if (ft != geo::FilterType::NONE) {
|
|
geo::ShapeContainer const& filter = _near.filterShape();
|
|
TRI_ASSERT(!filter.empty());
|
|
bool result = true; // updated by the callback
|
|
if (!_collection->readDocumentWithCallback(_trx, gdoc.token, [&](LocalDocumentId const&, VPackSlice doc) {
|
|
geo::ShapeContainer test;
|
|
Result res = _index->shape(doc, test);
|
|
TRI_ASSERT(res.ok()); // this should never fail here
|
|
if (res.fail() ||
|
|
(ft == geo::FilterType::CONTAINS && !filter.contains(&test)) ||
|
|
(ft == geo::FilterType::INTERSECTS && !filter.intersects(&test))) {
|
|
result = false;
|
|
} else {
|
|
result = true;
|
|
}
|
|
})) {
|
|
return false;
|
|
}
|
|
return result;
|
|
}
|
|
cb(gdoc.token); // return result
|
|
return true;
|
|
},
|
|
limit);
|
|
}
|
|
|
|
void reset() override { _near.reset(); }
|
|
|
|
private:
|
|
// we need to get intervals representing areas in a ring (annulus)
|
|
// around our target point. We need to fetch them ALL and then sort
|
|
// found results in a priority list according to their distance
|
|
void performScan() {
|
|
MMFilesGeoIndex::IndexTree const& tree = _index->tree();
|
|
// list of sorted intervals to scan
|
|
std::vector<geo::Interval> const scan = _near.intervals();
|
|
|
|
auto it = tree.begin();
|
|
for (size_t i = 0; i < scan.size(); i++) {
|
|
geo::Interval const& interval = scan[i];
|
|
TRI_ASSERT(interval.range_min <= interval.range_max);
|
|
|
|
// intervals are sorted and likely consecutive, try to avoid seeks
|
|
// by checking whether we are in the range already
|
|
bool seek = true;
|
|
if (i > 0) {
|
|
TRI_ASSERT(scan[i - 1].range_max < interval.range_min);
|
|
if (it == tree.end()) { // no more valid keys after this
|
|
break;
|
|
} else if (it->first > interval.range_max) {
|
|
continue; // beyond range already
|
|
} else if (interval.range_min <= it->first) {
|
|
seek = false; // already in range: min <= key <= max
|
|
TRI_ASSERT(it->first <= interval.range_max);
|
|
}
|
|
}
|
|
|
|
if (seek) { // try to avoid seeking at all cost
|
|
it = tree.lower_bound(interval.range_min);
|
|
}
|
|
|
|
while (it != tree.end() && it->first <= interval.range_max) {
|
|
_near.reportFound(it->second.documentId, it->second.centroid);
|
|
it++;
|
|
}
|
|
}
|
|
|
|
_near.didScanIntervals(); // calculate next bounds
|
|
}
|
|
|
|
/// find the first indexed entry to estimate the # of entries
|
|
/// around our target coordinates
|
|
void estimateDensity() {
|
|
MMFilesGeoIndex::IndexTree const& tree = _index->tree();
|
|
if (!tree.empty()) {
|
|
S2CellId cell = S2CellId(_near.origin());
|
|
auto it = tree.upper_bound(cell);
|
|
if (it == tree.end()) {
|
|
it = tree.lower_bound(cell);
|
|
}
|
|
if (it != tree.end()) {
|
|
_near.estimateDensity(it->second.centroid);
|
|
}
|
|
}
|
|
}
|
|
|
|
private:
|
|
MMFilesGeoIndex const* _index;
|
|
geo_index::NearUtils<CMP> _near;
|
|
};
|
|
|
|
typedef NearIterator<geo_index::DocumentsAscending> LegacyIterator;
|
|
|
|
MMFilesGeoIndex::MMFilesGeoIndex(TRI_idx_iid_t iid, LogicalCollection& collection,
|
|
arangodb::velocypack::Slice const& info,
|
|
std::string const& typeName)
|
|
: MMFilesIndex(iid, collection, info), geo_index::Index(info, _fields), _typeName(typeName) {
|
|
TRI_ASSERT(iid != 0);
|
|
_unique = false;
|
|
_sparse = true;
|
|
TRI_ASSERT(_variant != geo_index::Index::Variant::NONE);
|
|
}
|
|
|
|
size_t MMFilesGeoIndex::memory() const { return _tree.bytes_used(); }
|
|
|
|
/// @brief return a JSON representation of the index
|
|
void MMFilesGeoIndex::toVelocyPack(VPackBuilder& builder,
|
|
std::underlying_type<arangodb::Index::Serialize>::type flags) const {
|
|
TRI_ASSERT(_variant != geo_index::Index::Variant::NONE);
|
|
builder.openObject();
|
|
// Basic index
|
|
MMFilesIndex::toVelocyPack(builder, flags);
|
|
_coverParams.toVelocyPack(builder);
|
|
builder.add("geoJson", VPackValue(_variant == geo_index::Index::Variant::GEOJSON));
|
|
// geo indexes are always non-unique
|
|
builder.add(arangodb::StaticStrings::IndexUnique, arangodb::velocypack::Value(false));
|
|
// geo indexes are always sparse.
|
|
builder.add(arangodb::StaticStrings::IndexSparse, arangodb::velocypack::Value(true));
|
|
builder.close();
|
|
}
|
|
|
|
/// @brief Test if this index matches the definition
|
|
bool MMFilesGeoIndex::matchesDefinition(VPackSlice const& info) const {
|
|
TRI_ASSERT(_variant != geo_index::Index::Variant::NONE);
|
|
TRI_ASSERT(info.isObject());
|
|
#ifdef ARANGODB_ENABLE_MAINTAINER_MODE
|
|
auto typeSlice = info.get(arangodb::StaticStrings::IndexType);
|
|
TRI_ASSERT(typeSlice.isString());
|
|
arangodb::velocypack::StringRef typeStr(typeSlice);
|
|
TRI_ASSERT(typeStr == oldtypeName());
|
|
#endif
|
|
auto value = info.get(arangodb::StaticStrings::IndexId);
|
|
|
|
if (!value.isNone()) {
|
|
// We already have an id.
|
|
if (!value.isString()) {
|
|
// Invalid ID
|
|
return false;
|
|
}
|
|
// Short circuit. If id is correct the index is identical.
|
|
arangodb::velocypack::StringRef idRef(value);
|
|
return idRef == std::to_string(_iid);
|
|
}
|
|
|
|
if (_unique != basics::VelocyPackHelper::getBooleanValue(info, arangodb::StaticStrings::IndexUnique,
|
|
false)) {
|
|
return false;
|
|
}
|
|
|
|
if (_sparse != basics::VelocyPackHelper::getBooleanValue(info, arangodb::StaticStrings::IndexSparse,
|
|
true)) {
|
|
return false;
|
|
}
|
|
|
|
value = info.get(arangodb::StaticStrings::IndexFields);
|
|
|
|
if (!value.isArray()) {
|
|
return false;
|
|
}
|
|
|
|
size_t const n = static_cast<size_t>(value.length());
|
|
if (n != _fields.size()) {
|
|
return false;
|
|
}
|
|
|
|
if (n == 1) {
|
|
bool geoJson1 = basics::VelocyPackHelper::getBooleanValue(info, "geoJson", false);
|
|
bool geoJson2 = _variant == geo_index::Index::Variant::GEOJSON;
|
|
if (geoJson1 != geoJson2) {
|
|
return false;
|
|
}
|
|
}
|
|
|
|
// This check takes ordering of attributes into account.
|
|
std::vector<arangodb::basics::AttributeName> translate;
|
|
for (size_t i = 0; i < n; ++i) {
|
|
translate.clear();
|
|
VPackSlice f = value.at(i);
|
|
if (!f.isString()) {
|
|
// Invalid field definition!
|
|
return false;
|
|
}
|
|
arangodb::velocypack::StringRef in(f);
|
|
TRI_ParseAttributeString(in, translate, true);
|
|
if (!basics::AttributeName::isIdentical(_fields[i], translate, false)) {
|
|
return false;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
Result MMFilesGeoIndex::insert(transaction::Methods& trx, LocalDocumentId const& documentId,
|
|
velocypack::Slice const& doc,
|
|
arangodb::Index::OperationMode mode) {
|
|
// covering and centroid of coordinate / polygon / ...
|
|
size_t reserve = _variant == Variant::GEOJSON ? 8 : 1;
|
|
std::vector<S2CellId> cells;
|
|
cells.reserve(reserve);
|
|
S2Point centroid;
|
|
Result res = geo_index::Index::indexCells(doc, cells, centroid);
|
|
|
|
if (res.fail()) {
|
|
if (res.is(TRI_ERROR_BAD_PARAMETER)) {
|
|
res.reset(); // Invalid, no insert. Index is sparse
|
|
}
|
|
return res;
|
|
}
|
|
// LOG_TOPIC("0e6a2", ERR, Logger::ENGINES) << "Inserting #cells " << cells.size() << "
|
|
// doc: " << doc.toJson() << " center: " << centroid.toString();
|
|
TRI_ASSERT(!cells.empty());
|
|
TRI_ASSERT(S2::IsUnitLength(centroid));
|
|
|
|
IndexValue value(documentId, std::move(centroid));
|
|
for (S2CellId cell : cells) {
|
|
_tree.insert(std::make_pair(cell, value));
|
|
}
|
|
|
|
return res;
|
|
}
|
|
|
|
Result MMFilesGeoIndex::remove(transaction::Methods& trx, LocalDocumentId const& documentId,
|
|
velocypack::Slice const& doc,
|
|
arangodb::Index::OperationMode mode) {
|
|
// covering and centroid of coordinate / polygon / ...
|
|
size_t reserve = _variant == Variant::GEOJSON ? 8 : 1;
|
|
std::vector<S2CellId> cells(reserve);
|
|
S2Point centroid;
|
|
Result res = geo_index::Index::indexCells(doc, cells, centroid);
|
|
|
|
if (res.fail()) { // might occur if insert is rolled back
|
|
if (res.is(TRI_ERROR_BAD_PARAMETER)) {
|
|
res.reset(); // Invalid, no remove. Index is sparse
|
|
}
|
|
return res;
|
|
}
|
|
// LOG_TOPIC("1255b", ERR, Logger::ENGINES) << "Removing #cells " << cells.size() << "
|
|
// doc: " << doc.toJson();
|
|
TRI_ASSERT(!cells.empty());
|
|
|
|
for (S2CellId cell : cells) {
|
|
auto it = _tree.lower_bound(cell);
|
|
while (it != _tree.end() && it->first == cell) {
|
|
if (it->second.documentId == documentId) {
|
|
it = _tree.erase(it);
|
|
} else {
|
|
++it;
|
|
}
|
|
}
|
|
}
|
|
return res;
|
|
}
|
|
|
|
/// @brief creates an IndexIterator for the given Condition
|
|
std::unique_ptr<IndexIterator> MMFilesGeoIndex::iteratorForCondition(
|
|
transaction::Methods* trx, arangodb::aql::AstNode const* node,
|
|
arangodb::aql::Variable const* reference, IndexIteratorOptions const& opts) {
|
|
TRI_ASSERT(!isSorted() || opts.sorted);
|
|
TRI_ASSERT(node != nullptr);
|
|
|
|
geo::QueryParams params;
|
|
params.sorted = opts.sorted;
|
|
params.ascending = opts.ascending;
|
|
params.pointsOnly = pointsOnly();
|
|
params.limit = opts.limit;
|
|
geo_index::Index::parseCondition(node, reference, params);
|
|
|
|
// FIXME: <Optimize away>
|
|
params.sorted = true;
|
|
if (params.filterType != geo::FilterType::NONE) {
|
|
TRI_ASSERT(!params.filterShape.empty());
|
|
params.filterShape.updateBounds(params);
|
|
}
|
|
// </Optimize away>
|
|
|
|
TRI_ASSERT(!opts.sorted || params.origin.is_valid());
|
|
// params.cover.worstIndexedLevel < _coverParams.worstIndexedLevel
|
|
// is not necessary, > would be missing entries.
|
|
params.cover.worstIndexedLevel = _coverParams.worstIndexedLevel;
|
|
if (params.cover.bestIndexedLevel > _coverParams.bestIndexedLevel) {
|
|
// it is unnessesary to use a better level than configured
|
|
params.cover.bestIndexedLevel = _coverParams.bestIndexedLevel;
|
|
}
|
|
|
|
// why does this have to be shit?
|
|
if (params.ascending) {
|
|
return std::make_unique<NearIterator<geo_index::DocumentsAscending>>(&_collection, trx, this, std::move(params));
|
|
} else {
|
|
return std::make_unique<NearIterator<geo_index::DocumentsDescending>>(&_collection, trx, this, std::move(params));
|
|
}
|
|
}
|
|
|
|
void MMFilesGeoIndex::unload() {
|
|
_tree.clear(); // TODO: do we need load?
|
|
}
|