[geocoder] Fix correct rank for eponymous Town and Hamlet: calculate result ranks by kind field (in jsonl)
This commit is contained in:
Anatoly Serdtcev
LaGrunge
parent
1711eb5454
commit
fd63ea617c
12 changed files with 365 additions and 129 deletions
|
|
@ -158,6 +158,7 @@ geocore_add_library(${PROJECT_NAME} ${SRC})
|
|||
geocore_link_libraries(${PROJECT_NAME}
|
||||
base
|
||||
coding
|
||||
geocoder
|
||||
indexer
|
||||
platform
|
||||
geometry
|
||||
|
|
|
|||
|
|
@ -6,6 +6,8 @@
|
|||
#include "generator/regions/collector_region_info.hpp"
|
||||
#include "generator/regions/region_info.hpp"
|
||||
|
||||
#include "geocoder/types.hpp"
|
||||
|
||||
#include "platform/platform.hpp"
|
||||
|
||||
#include "base/geo_object_id.hpp"
|
||||
|
|
@ -165,3 +167,15 @@ UNIT_TEST(RegionInfoCollector_MergeAndSave)
|
|||
TEST(!rg.GetIsoCodeAlphaNumeric(), ());
|
||||
}
|
||||
}
|
||||
|
||||
// PlaceType tests --------------------------------------------------------------------------------
|
||||
UNIT_TEST(CollectorRegionInfo_PlaceType2KindTest)
|
||||
{
|
||||
TEST_EQUAL(static_cast<int>(PlaceType::Unknown), 0, ());
|
||||
for (auto i = 1; i < static_cast<int>(PlaceType::Count); ++i)
|
||||
{
|
||||
auto const placeType = static_cast<PlaceType>(i);
|
||||
TEST_NOT_EQUAL(
|
||||
geocoder::KindFromString(StringifyPlaceType(placeType)), geocoder::Kind::Unknown, ());
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -77,6 +77,8 @@ char const * StringifyPlaceType(PlaceType placeType)
|
|||
return "neighbourhood";
|
||||
case PlaceType::Unknown:
|
||||
return "unknown";
|
||||
case PlaceType::Count:
|
||||
UNREACHABLE();
|
||||
};
|
||||
|
||||
UNREACHABLE();
|
||||
|
|
|
|||
|
|
@ -64,6 +64,8 @@ enum class PlaceType: uint8_t
|
|||
Suburb = 12,
|
||||
Quarter = 13,
|
||||
Neighbourhood = 14,
|
||||
|
||||
Count
|
||||
};
|
||||
|
||||
PlaceType EncodePlaceType(std::string const & place);
|
||||
|
|
|
|||
|
|
@ -61,6 +61,8 @@ PlaceLevel CountrySpecifier::GetLevel(PlaceType placeType)
|
|||
return PlaceLevel::Sublocality;
|
||||
case PlaceType::Unknown:
|
||||
break;
|
||||
case PlaceType::Count:
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
return PlaceLevel::Unknown;
|
||||
|
|
|
|||
|
|
@ -59,18 +59,44 @@ double GetWeight(Type t)
|
|||
switch (t)
|
||||
{
|
||||
case Type::Country: return 10.0;
|
||||
case Type::Locality: return 5.0;
|
||||
case Type::Region: return 4.0;
|
||||
case Type::Subregion: return 4.0;
|
||||
case Type::Locality: return 5.0;
|
||||
case Type::Street: return 2.0;
|
||||
case Type::Suburb: return 1.0;
|
||||
case Type::Sublocality: return 1.0;
|
||||
case Type::Street: return 2.0;
|
||||
case Type::Building: return 0.1;
|
||||
case Type::Count: return 0.0;
|
||||
}
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
double GetWeight(Kind kind)
|
||||
{
|
||||
switch (kind)
|
||||
{
|
||||
case Kind::Country: return 10.0;
|
||||
case Kind::City: return 5.05;
|
||||
case Kind::Town: return 5.04;
|
||||
case Kind::State: return 4.05;
|
||||
case Kind::Province: return 4.01;
|
||||
case Kind::District: return 4.01;
|
||||
case Kind::County: return 4.01;
|
||||
case Kind::Municipality: return 4.0;
|
||||
case Kind::Village: return 3.0;
|
||||
case Kind::Street: return 2.0;
|
||||
case Kind::Hamlet: return 1.06;
|
||||
case Kind::Suburb: return 1.05;
|
||||
case Kind::Quarter: return 1.01;
|
||||
case Kind::Neighbourhood: return 1.0;
|
||||
case Kind::IsolatedDwelling: return 0.5;
|
||||
case Kind::Building: return 0.1;
|
||||
case Kind::Unknown: return 0.0;
|
||||
case Kind::Count: return 0.0;
|
||||
}
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
// todo(@m) This is taken from search/geocoder.hpp. Refactor.
|
||||
class ScopedMarkTokens
|
||||
{
|
||||
|
|
@ -126,6 +152,20 @@ strings::UniString MakeHouseNumber(Tokens const & tokens)
|
|||
}
|
||||
} // namespace
|
||||
|
||||
// Geocoder::Layer ---------------------------------------------------------------------------------
|
||||
Geocoder::Layer::Layer(Type type)
|
||||
: m_type{type}
|
||||
{
|
||||
}
|
||||
|
||||
void Geocoder::Layer::SetCandidates(std::vector<Candidate> && candidates)
|
||||
{
|
||||
std::sort(candidates.begin(), candidates.end(), [](auto const & a, auto const & b) {
|
||||
return a.m_totalCertainty < b.m_totalCertainty;
|
||||
});
|
||||
m_candidatesByCertainty = std::move(candidates);
|
||||
}
|
||||
|
||||
// Geocoder::Context -------------------------------------------------------------------------------
|
||||
Geocoder::Context::Context(string const & query) : m_beam(kMaxResults)
|
||||
{
|
||||
|
|
@ -378,8 +418,7 @@ void Geocoder::Go(Context & ctx, Type type) const
|
|||
subquery.push_back(ctx.GetToken(j));
|
||||
subqueryTokenIds.push_back(j);
|
||||
|
||||
Layer curLayer;
|
||||
curLayer.m_type = type;
|
||||
Layer curLayer{type};
|
||||
|
||||
// Buildings are indexed separately.
|
||||
if (type == Type::Building)
|
||||
|
|
@ -391,7 +430,7 @@ void Geocoder::Go(Context & ctx, Type type) const
|
|||
FillRegularLayer(ctx, type, subquery, curLayer);
|
||||
}
|
||||
|
||||
if (curLayer.m_entries.empty())
|
||||
if (curLayer.GetCandidatesByCertainty().empty())
|
||||
continue;
|
||||
|
||||
ScopedMarkTokens mark(ctx, type, i, j + 1);
|
||||
|
|
@ -400,7 +439,7 @@ void Geocoder::Go(Context & ctx, Type type) const
|
|||
if (type == Type::Street)
|
||||
MarkStreetSynonym(ctx, streetSynonymMark);
|
||||
|
||||
AddResults(ctx, curLayer.m_entries);
|
||||
AddResults(ctx, curLayer.GetCandidatesByCertainty());
|
||||
|
||||
ctx.GetLayers().emplace_back(move(curLayer));
|
||||
SCOPE_GUARD(pop, [&] { ctx.GetLayers().pop_back(); });
|
||||
|
|
@ -425,7 +464,7 @@ void Geocoder::FillBuildingsLayer(Context & ctx, Tokens const & subquery, vector
|
|||
|
||||
for (auto const & layer : boost::adaptors::reverse(ctx.GetLayers()))
|
||||
{
|
||||
if (layer.m_type != Type::Street && layer.m_type != Type::Locality)
|
||||
if (layer.GetType() != Type::Street && layer.GetType() != Type::Locality)
|
||||
continue;
|
||||
|
||||
// We've already filled a street/location layer and now see something that resembles
|
||||
|
|
@ -433,11 +472,14 @@ void Geocoder::FillBuildingsLayer(Context & ctx, Tokens const & subquery, vector
|
|||
// let's stay on the safer side and mark the tokens as potential house number.
|
||||
ctx.MarkHouseNumberPositionsInQuery(subqueryTokenIds);
|
||||
|
||||
auto candidates = std::vector<Candidate>{};
|
||||
|
||||
auto const & lastLayer = ctx.GetLayers().back();
|
||||
auto const forSublocalityLayer =
|
||||
lastLayer.m_type == Type::Suburb || lastLayer.m_type == Type::Sublocality;
|
||||
for (auto const & docId : layer.m_entries)
|
||||
lastLayer.GetType() == Type::Suburb || lastLayer.GetType() == Type::Sublocality;
|
||||
for (auto const & buildingOwnerCandidate : layer.GetCandidatesByCertainty())
|
||||
{
|
||||
auto const & docId = buildingOwnerCandidate.m_entry;
|
||||
m_index.ForEachRelatedBuilding(docId, [&](Index::DocId const & buildingDocId) {
|
||||
auto const & building = m_index.GetDoc(buildingDocId);
|
||||
auto const & multipleHN = building.GetNormalizedMultipleNames(
|
||||
|
|
@ -447,14 +489,23 @@ void Geocoder::FillBuildingsLayer(Context & ctx, Tokens const & subquery, vector
|
|||
if (search::house_numbers::HouseNumbersMatch(realHNUniStr, subqueryHN,
|
||||
false /* queryIsPrefix */))
|
||||
{
|
||||
if (forSublocalityLayer && !HasParent(ctx.GetLayers(), building))
|
||||
auto && parentCandidateCertainty =
|
||||
forSublocalityLayer ? FindMaxCertaintyInParentCandidates(ctx.GetLayers(), building)
|
||||
: boost::optional<double>{buildingOwnerCandidate.m_totalCertainty};
|
||||
if (!parentCandidateCertainty)
|
||||
return;
|
||||
|
||||
curLayer.m_entries.emplace_back(buildingDocId);
|
||||
static auto const buildingTokenWeight = GetWeight(Kind::Building);
|
||||
auto totalCertainty =
|
||||
*parentCandidateCertainty + buildingTokenWeight * subqueryTokenIds.size();
|
||||
|
||||
candidates.push_back({buildingDocId, totalCertainty});
|
||||
}
|
||||
});
|
||||
}
|
||||
|
||||
if (!candidates.empty())
|
||||
curLayer.SetCandidates(std::move(candidates));
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
|
@ -462,30 +513,38 @@ void Geocoder::FillBuildingsLayer(Context & ctx, Tokens const & subquery, vector
|
|||
void Geocoder::FillRegularLayer(Context const & ctx, Type type, Tokens const & subquery,
|
||||
Layer & curLayer) const
|
||||
{
|
||||
auto candidates = std::vector<Candidate>{};
|
||||
|
||||
m_index.ForEachDocId(subquery, [&](Index::DocId const & docId) {
|
||||
auto const & d = m_index.GetDoc(docId);
|
||||
if (d.m_type != type)
|
||||
return;
|
||||
|
||||
if (ctx.GetLayers().empty() || HasParent(ctx.GetLayers(), d))
|
||||
{
|
||||
if (type > Type::Locality && !IsRelevantLocalityMember(ctx, d, subquery))
|
||||
return;
|
||||
auto && parentCandidateCertainty = FindMaxCertaintyInParentCandidates(ctx.GetLayers(), d);
|
||||
if (!parentCandidateCertainty)
|
||||
return;
|
||||
|
||||
curLayer.m_entries.emplace_back(docId);
|
||||
}
|
||||
if (type > Type::Locality && !IsRelevantLocalityMember(ctx, d, subquery))
|
||||
return;
|
||||
|
||||
auto subqueryWeight =
|
||||
(d.m_kind != Kind::Unknown ? GetWeight(d.m_kind) : GetWeight(d.m_type)) * subquery.size();
|
||||
auto totalCertainty = *parentCandidateCertainty + subqueryWeight;
|
||||
|
||||
candidates.push_back({docId, totalCertainty});
|
||||
});
|
||||
|
||||
if (!candidates.empty())
|
||||
curLayer.SetCandidates(std::move(candidates));
|
||||
}
|
||||
|
||||
void Geocoder::AddResults(Context & ctx, std::vector<Index::DocId> const & entries) const
|
||||
void Geocoder::AddResults(Context & ctx, std::vector<Candidate> const & candidates) const
|
||||
{
|
||||
double certainty = 0;
|
||||
vector<size_t> tokenIds;
|
||||
vector<Type> allTypes;
|
||||
for (size_t tokId = 0; tokId < ctx.GetNumTokens(); ++tokId)
|
||||
{
|
||||
auto const t = ctx.GetTokenType(tokId);
|
||||
certainty += GetWeight(t);
|
||||
if (t != Type::Count)
|
||||
{
|
||||
tokenIds.push_back(tokId);
|
||||
|
|
@ -493,11 +552,11 @@ void Geocoder::AddResults(Context & ctx, std::vector<Index::DocId> const & entri
|
|||
}
|
||||
}
|
||||
|
||||
for (auto const & docId : entries)
|
||||
for (auto const & candidate : candidates)
|
||||
{
|
||||
auto const & docId = candidate.m_entry;
|
||||
auto const & entry = m_index.GetDoc(docId);
|
||||
|
||||
auto entryCertainty = certainty;
|
||||
auto entryCertainty = candidate.m_totalCertainty;
|
||||
|
||||
if (InCityState(entry))
|
||||
{
|
||||
|
|
@ -536,19 +595,23 @@ bool Geocoder::InCityState(Hierarchy::Entry const & entry) const
|
|||
return false;
|
||||
}
|
||||
|
||||
bool Geocoder::HasParent(vector<Geocoder::Layer> const & layers, Hierarchy::Entry const & e) const
|
||||
boost::optional<double> Geocoder::FindMaxCertaintyInParentCandidates(
|
||||
vector<Geocoder::Layer> const & layers, Hierarchy::Entry const & e) const
|
||||
{
|
||||
CHECK(!layers.empty(), ());
|
||||
if (layers.empty())
|
||||
return 0;
|
||||
|
||||
auto const & layer = layers.back();
|
||||
for (auto const & docId : layer.m_entries)
|
||||
for (auto const & candidate : layer.GetCandidatesByCertainty())
|
||||
{
|
||||
auto const & docId = candidate.m_entry;
|
||||
// Note that the relationship is somewhat inverted: every ancestor
|
||||
// is stored in the address but the nodes have no information
|
||||
// about their children.
|
||||
if (m_hierarchy.IsParentTo(m_index.GetDoc(docId), e))
|
||||
return true;
|
||||
return candidate.m_totalCertainty;
|
||||
}
|
||||
return false;
|
||||
return {};
|
||||
}
|
||||
|
||||
bool Geocoder::IsRelevantLocalityMember(Context const & ctx, Hierarchy::Entry const & member,
|
||||
|
|
@ -562,14 +625,15 @@ bool Geocoder::HasMemberLocalityInMatching(Context const & ctx, Hierarchy::Entry
|
|||
{
|
||||
for (auto const & layer : ctx.GetLayers())
|
||||
{
|
||||
auto const layerType = layer.m_type;
|
||||
auto const layerType = layer.GetType();
|
||||
if (layerType > Type::Locality)
|
||||
break;
|
||||
if (layerType != Type::Locality)
|
||||
continue;
|
||||
|
||||
for (auto const docId : layer.m_entries)
|
||||
for (auto const & candidate : layer.GetCandidatesByCertainty())
|
||||
{
|
||||
auto const & docId = candidate.m_entry;
|
||||
auto const & matchedEntry = m_index.GetDoc(docId);
|
||||
if (m_hierarchy.IsParentTo(matchedEntry, member))
|
||||
return true;
|
||||
|
|
|
|||
|
|
@ -18,6 +18,7 @@
|
|||
#include <utility>
|
||||
#include <vector>
|
||||
|
||||
#include <boost/optional.hpp>
|
||||
#include <boost/serialization/version.hpp>
|
||||
|
||||
namespace geocoder
|
||||
|
|
@ -45,11 +46,30 @@ public:
|
|||
DECLARE_EXCEPTION(Exception, RootException);
|
||||
DECLARE_EXCEPTION(OpenException, Exception);
|
||||
|
||||
// A Layer contains all entries matched by a subquery of consecutive tokens.
|
||||
struct Layer
|
||||
// Candidate contain matched entry with certainty of all matched tokens.
|
||||
struct Candidate
|
||||
{
|
||||
Type m_type = Type::Count;
|
||||
std::vector<Index::DocId> m_entries;
|
||||
Index::DocId m_entry;
|
||||
double m_totalCertainty;
|
||||
};
|
||||
|
||||
// A Layer contains all entries matched by a subquery of consecutive tokens.
|
||||
class Layer
|
||||
{
|
||||
public:
|
||||
Layer() = default;
|
||||
Layer(Type type);
|
||||
|
||||
Type GetType() const noexcept { return m_type; }
|
||||
std::vector<Candidate> const & GetCandidatesByCertainty() const noexcept
|
||||
{
|
||||
return m_candidatesByCertainty;
|
||||
}
|
||||
void SetCandidates(std::vector<Candidate> && candidates);
|
||||
|
||||
private:
|
||||
Type m_type{Type::Count};
|
||||
std::vector<Candidate> m_candidatesByCertainty;
|
||||
};
|
||||
|
||||
// This class is very similar to the one we use in search/.
|
||||
|
|
@ -158,13 +178,16 @@ private:
|
|||
Layer & curLayer) const;
|
||||
void FillRegularLayer(Context const & ctx, Type type, Tokens const & subquery,
|
||||
Layer & curLayer) const;
|
||||
void AddResults(Context & ctx, std::vector<Index::DocId> const & entries) const;
|
||||
void AddResults(Context & ctx, std::vector<Candidate> const & candidates) const;
|
||||
|
||||
bool InCityState(Hierarchy::Entry const & entry) const;
|
||||
|
||||
// Returns whether any of the paths through |layers| can be extended
|
||||
// by appending |e|.
|
||||
bool HasParent(std::vector<Geocoder::Layer> const & layers, Hierarchy::Entry const & e) const;
|
||||
// Find max certainty in parent candidates.
|
||||
// 0 - first candidate.
|
||||
// none - there is no parent from candidates.
|
||||
boost::optional<double> FindMaxCertaintyInParentCandidates(
|
||||
std::vector<Geocoder::Layer> const & layers, Hierarchy::Entry const & e) const;
|
||||
|
||||
bool IsRelevantLocalityMember(Context const & ctx, Hierarchy::Entry const & member,
|
||||
Tokens const & subquery) const;
|
||||
bool HasMemberLocalityInMatching(Context const & ctx, Hierarchy::Entry const & member) const;
|
||||
|
|
|
|||
|
|
@ -24,9 +24,9 @@ using Id = base::GeoObjectId;
|
|||
|
||||
double const kCertaintyEps = 1e-4;
|
||||
string const kRegionsData = R"#(
|
||||
C00000000004B279 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-80.1142033187951, 21.55511095]}, "properties": {"locales": {"default": {"name": "Cuba", "address": {"country": "Cuba"}}}, "rank": 2}}
|
||||
C0000000001C4CA7 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-78.7260117405499, 21.74300205]}, "properties": {"locales": {"default": {"name": "Ciego de Ávila", "address": {"region": "Ciego de Ávila", "country": "Cuba"}}}, "rank": 4}}
|
||||
C00000000059D6B5 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-78.9263054493181, 22.08185765]}, "properties": {"locales": {"default": {"name": "Florencia", "address": {"subregion": "Florencia", "region": "Ciego de Ávila", "country": "Cuba"}}}, "rank": 6}}
|
||||
C00000000004B279 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-80.1142033187951, 21.55511095]}, "properties": {"kind": "country", "locales": {"default": {"name": "Cuba", "address": {"country": "Cuba"}}}, "rank": 2}}
|
||||
C0000000001C4CA7 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-78.7260117405499, 21.74300205]}, "properties": {"kind": "province", "locales": {"default": {"name": "Ciego de Ávila", "address": {"region": "Ciego de Ávila", "country": "Cuba"}}}, "rank": 4}}
|
||||
C00000000059D6B5 {"type": "Feature", "geometry": {"type": "Point", "coordinates": [-78.9263054493181, 22.08185765]}, "properties": {"kind": "district", "locales": {"default": {"name": "Florencia", "address": {"subregion": "Florencia", "region": "Ciego de Ávila", "country": "Cuba"}}}, "rank": 6}}
|
||||
)#";
|
||||
} // namespace
|
||||
|
||||
|
|
@ -44,7 +44,7 @@ void TestGeocoder(Geocoder & geocoder, string const & query, vector<Result> && e
|
|||
TEST(actual[i].m_certainty >= 0.0 && actual[i].m_certainty <= 1.0,
|
||||
(query, actual[i].m_certainty));
|
||||
TEST_EQUAL(actual[i].m_osmId, expected[i].m_osmId, (query));
|
||||
TEST(base::AlmostEqualAbs(actual[i].m_certainty, expected[i].m_certainty, kCertaintyEps),
|
||||
TEST_NEAR(actual[i].m_certainty, expected[i].m_certainty, kCertaintyEps,
|
||||
(query, actual[i].m_certainty, expected[i].m_certainty));
|
||||
}
|
||||
}
|
||||
|
|
@ -59,8 +59,8 @@ UNIT_TEST(Geocoder_Smoke)
|
|||
base::GeoObjectId const cubaId(0xc00000000004b279);
|
||||
|
||||
TestGeocoder(geocoder, "florencia", {{florenciaId, 1.0}});
|
||||
TestGeocoder(geocoder, "cuba florencia", {{florenciaId, 1.0}, {cubaId, 0.714286}});
|
||||
TestGeocoder(geocoder, "florencia somewhere in cuba", {{cubaId, 0.714286}, {florenciaId, 1.0}});
|
||||
TestGeocoder(geocoder, "cuba florencia", {{florenciaId, 1.0}, {cubaId, 0.713776}});
|
||||
TestGeocoder(geocoder, "florencia somewhere in cuba", {{cubaId, 0.713776}, {florenciaId, 1.0}});
|
||||
}
|
||||
|
||||
UNIT_TEST(Geocoder_Hierarchy)
|
||||
|
|
@ -88,31 +88,31 @@ UNIT_TEST(Geocoder_Hierarchy)
|
|||
UNIT_TEST(Geocoder_EnglishNames)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица Новый Арбат"}}, "en": {"address": {"locality": "Moscow", "street": "New Arbat Avenue"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}}}
|
||||
11 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица Новый Арбат"}}, "en": {"address": {"locality": "Moscow", "street": "New Arbat Avenue"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
ScopedFile const regionsJsonFile("regions.jsonl", kData);
|
||||
geocoder.LoadFromJsonl(regionsJsonFile.GetFullPath());
|
||||
|
||||
TestGeocoder(geocoder, "Moscow, New Arbat", {{Id{0x11}, 1.0}, {Id{0x10}, 0.5555}});
|
||||
TestGeocoder(geocoder, "Moscow, New Arbat", {{Id{0x11}, 1.0}, {Id{0x10}, 0.558011}});
|
||||
}
|
||||
|
||||
UNIT_TEST(Geocoder_OnlyBuildings)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Some Locality"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Some Locality"}}}}}
|
||||
|
||||
21 {"properties": {"locales": {"default": {"address": {"street": "Good", "locality": "Some Locality"}}}}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"building": "5", "street": "Good", "locality": "Some Locality"}}}}}
|
||||
21 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Good", "locality": "Some Locality"}}}}}
|
||||
22 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "5", "street": "Good", "locality": "Some Locality"}}}}}
|
||||
|
||||
31 {"properties": {"locales": {"default": {"address": {"street": "Bad", "locality": "Some Locality"}}}}}
|
||||
32 {"properties": {"locales": {"default": {"address": {"building": "10", "street": "Bad", "locality": "Some Locality"}}}}}
|
||||
31 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Bad", "locality": "Some Locality"}}}}}
|
||||
32 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "10", "street": "Bad", "locality": "Some Locality"}}}}}
|
||||
|
||||
40 {"properties": {"locales": {"default": {"address": {"street": "MaybeNumbered", "locality": "Some Locality"}}}}}
|
||||
41 {"properties": {"locales": {"default": {"address": {"street": "MaybeNumbered-3", "locality": "Some Locality"}}}}}
|
||||
42 {"properties": {"locales": {"default": {"address": {"building": "3", "street": "MaybeNumbered", "locality": "Some Locality"}}}}}
|
||||
40 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "MaybeNumbered", "locality": "Some Locality"}}}}}
|
||||
41 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "MaybeNumbered-3", "locality": "Some Locality"}}}}}
|
||||
42 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "3", "street": "MaybeNumbered", "locality": "Some Locality"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -126,8 +126,8 @@ UNIT_TEST(Geocoder_OnlyBuildings)
|
|||
base::GeoObjectId const building10(0x32);
|
||||
|
||||
TestGeocoder(geocoder, "some locality", {{localityId, 1.0}});
|
||||
TestGeocoder(geocoder, "some locality good", {{goodStreetId, 1.0}, {localityId, 0.833333}});
|
||||
TestGeocoder(geocoder, "some locality bad", {{badStreetId, 1.0}, {localityId, 0.833333}});
|
||||
TestGeocoder(geocoder, "some locality good", {{goodStreetId, 1.0}, {localityId, 0.834711}});
|
||||
TestGeocoder(geocoder, "some locality bad", {{badStreetId, 1.0}, {localityId, 0.834711}});
|
||||
|
||||
TestGeocoder(geocoder, "some locality good 5", {{building5, 1.0}});
|
||||
TestGeocoder(geocoder, "some locality bad 10", {{building10, 1.0}});
|
||||
|
|
@ -142,20 +142,20 @@ UNIT_TEST(Geocoder_OnlyBuildings)
|
|||
base::GeoObjectId const numberedStreet(0x41);
|
||||
base::GeoObjectId const houseOnANonNumberedStreet(0x42);
|
||||
TestGeocoder(geocoder, "some locality maybenumbered 3",
|
||||
{{numberedStreet, 1.0}, {houseOnANonNumberedStreet, 0.864286}});
|
||||
{{numberedStreet, 1.0}, {houseOnANonNumberedStreet, 0.865248}});
|
||||
}
|
||||
|
||||
UNIT_TEST(Geocoder_MismatchedLocality)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Moscow"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Paris"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Moscow"}}}}}
|
||||
11 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Paris"}}}}}
|
||||
|
||||
21 {"properties": {"locales": {"default": {"address": {"street": "Krymskaya", "locality": "Moscow"}}}}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Moscow"}}}}}
|
||||
21 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Krymskaya", "locality": "Moscow"}}}}}
|
||||
22 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Moscow"}}}}}
|
||||
|
||||
31 {"properties": {"locales": {"default": {"address": {"street": "Krymskaya", "locality": "Paris"}}}}}
|
||||
32 {"properties": {"locales": {"default": {"address": {"building": "3", "street": "Krymskaya", "locality": "Paris"}}}}}
|
||||
31 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Krymskaya", "locality": "Paris"}}}}}
|
||||
32 {"properties": {"kind": "builidng", "locales": {"default": {"address": {"building": "3", "street": "Krymskaya", "locality": "Paris"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -174,34 +174,34 @@ UNIT_TEST(Geocoder_MismatchedLocality)
|
|||
UNIT_TEST(Geocoder_MoscowLocalityRank)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"region": "Москва"}}}, "rank": 2}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "region": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}, "rank": 4}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"street": "Ленинский проспект", "locality": "Москва", "region": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}}}
|
||||
10 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Москва"}}}, "rank": 2}}
|
||||
11 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва", "region": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}, "rank": 4}}
|
||||
12 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Ленинский проспект", "locality": "Москва", "region": "Москва"}}, "en": {"address": {"locality": "Moscow"}}}}}
|
||||
|
||||
20 {"properties": {"locales": {"default": {"address": {"region": "Тверская Область"}}}, "rank": 2}}
|
||||
21 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "region": "Тверская Область"}}}, "rank": 4}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"street": "Ленинский проспект", "locality": "Москва", "region": "Тверская Область"}}}}}
|
||||
20 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Тверская Область"}}}, "rank": 2}}
|
||||
21 {"properties": {"kind": "hamlet", "locales": {"default": {"address": {"locality": "Москва", "region": "Тверская Область"}}}, "rank": 4}}
|
||||
22 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Ленинский проспект", "locality": "Москва", "region": "Тверская Область"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
ScopedFile const regionsJsonFile("regions.jsonl", kData);
|
||||
geocoder.LoadFromJsonl(regionsJsonFile.GetFullPath());
|
||||
|
||||
TestGeocoder(geocoder, "Москва", {{Id{0x11}, 1.0}, {Id{0x21}, 0.990099}, {Id{0x10}, 0.792079}});
|
||||
TestGeocoder(geocoder, "Москва, Ленинский проспект", {{Id{0x12}, 1.0}, {Id{0x22}, 0.994475},
|
||||
{Id{0x11}, 0.558011}, {Id{0x21}, 0.552486},
|
||||
{Id{0x10}, 0.441989}});
|
||||
TestGeocoder(geocoder, "Москва", {{Id{0x11}, 1.0}, {Id{0x21}, 0.207843}, {Id{0x10}, 0.794118}});
|
||||
TestGeocoder(geocoder, "Москва, Ленинский проспект",
|
||||
{{Id{0x12}, 1.0}, {Id{0x22}, 0.556044}, {Id{0x11}, 0.56044}, {Id{0x10}, 0.445055},
|
||||
{Id{0x21}, 0.116484}});
|
||||
}
|
||||
|
||||
// Geocoder_StreetWithNumber* ----------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_StreetWithNumberInCity)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
|
||||
20 {"properties": {"locales": {"default": {"address": {"locality": "Краснокамск"}}}}}
|
||||
28 {"properties": {"locales": {"default": {"address": {"locality": "Краснокамск", "street": "улица 1905 года"}}}}}
|
||||
20 {"properties": {"kind": "town", "locales": {"default": {"address": {"locality": "Краснокамск"}}}}}
|
||||
28 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Краснокамск", "street": "улица 1905 года"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -214,8 +214,8 @@ UNIT_TEST(Geocoder_StreetWithNumberInCity)
|
|||
UNIT_TEST(Geocoder_StreetWithNumberInClassifiedCity)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -228,11 +228,11 @@ UNIT_TEST(Geocoder_StreetWithNumberInClassifiedCity)
|
|||
UNIT_TEST(Geocoder_StreetWithNumberInAnyCity)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
|
||||
20 {"properties": {"locales": {"default": {"address": {"locality": "Краснокамск"}}}}}
|
||||
28 {"properties": {"locales": {"default": {"address": {"locality": "Краснокамск", "street": "улица 1905 года"}}}}}
|
||||
20 {"properties": {"kind": "town", "locales": {"default": {"address": {"locality": "Краснокамск"}}}}}
|
||||
28 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Краснокамск", "street": "улица 1905 года"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -245,8 +245,8 @@ UNIT_TEST(Geocoder_StreetWithNumberInAnyCity)
|
|||
UNIT_TEST(Geocoder_StreetWithNumberAndWithoutStreetSynonym)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 1905 года"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -259,8 +259,8 @@ UNIT_TEST(Geocoder_StreetWithNumberAndWithoutStreetSynonym)
|
|||
UNIT_TEST(Geocoder_UntypedStreetWithNumberAndStreetSynonym)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
13 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "8 Марта"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
13 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "8 Марта"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -273,10 +273,10 @@ UNIT_TEST(Geocoder_UntypedStreetWithNumberAndStreetSynonym)
|
|||
UNIT_TEST(Geocoder_StreetWithTwoNumbers)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "4-я улица 8 Марта"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
12 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "4-я улица 8 Марта"}}}}}
|
||||
|
||||
13 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта"}}}}}
|
||||
13 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -289,9 +289,9 @@ UNIT_TEST(Geocoder_StreetWithTwoNumbers)
|
|||
UNIT_TEST(Geocoder_BuildingOnStreetWithNumber)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
13 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта"}}}}}
|
||||
15 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта", "building": "4"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
13 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта"}}}}}
|
||||
15 {"properties": {"kind": "street", "locales": {"default": {"address": {"locality": "Москва", "street": "улица 8 Марта", "building": "4"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -305,10 +305,10 @@ UNIT_TEST(Geocoder_BuildingOnStreetWithNumber)
|
|||
UNIT_TEST(Geocoder_LocalityBuilding)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Zelenograd"}}}}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"building": "2", "locality": "Zelenograd"}}}}}
|
||||
31 {"properties": {"locales": {"default": {"address": {"street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
32 {"properties": {"locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Zelenograd"}}}}}
|
||||
22 {"properties": {"kind": "builiding", "locales": {"default": {"address": {"building": "2", "locality": "Zelenograd"}}}}}
|
||||
31 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
32 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -323,44 +323,44 @@ UNIT_TEST(Geocoder_LocalityBuilding)
|
|||
UNIT_TEST(Geocoder_LocalityBuildingRankWithSuburb)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"suburb": "Арбат", "locality": "Москва"}}}}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"building": "1", "suburb": "Арбат", "locality": "Москва"}}}}}
|
||||
13 {"properties": {"locales": {"default": {"address": {"suburb": "район Северный", "locality": "Москва"}}}}}
|
||||
14 {"properties": {"locales": {"default": {"address": {"building": "1", "suburb": "район Северный", "locality": "Москва"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва"}}}}}
|
||||
11 {"properties": {"kind": "suburb", "locales": {"default": {"address": {"suburb": "Арбат", "locality": "Москва"}}}}}
|
||||
12 {"properties": {"kind": "builidng", "locales": {"default": {"address": {"building": "1", "suburb": "Арбат", "locality": "Москва"}}}}}
|
||||
13 {"properties": {"kind": "suburb", "locales": {"default": {"address": {"suburb": "район Северный", "locality": "Москва"}}}}}
|
||||
14 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "1", "suburb": "район Северный", "locality": "Москва"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
ScopedFile const regionsJsonFile("regions.jsonl", kData);
|
||||
geocoder.LoadFromJsonl(regionsJsonFile.GetFullPath());
|
||||
|
||||
TestGeocoder(geocoder, "Москва, Арбат 1", {{Id{0x12}, 1.0}, {Id{0x14}, 0.836066}});
|
||||
TestGeocoder(geocoder, "Москва, Арбат 1", {{Id{0x12}, 1.0}, {Id{0x14}, 0.830645}});
|
||||
}
|
||||
|
||||
//--------------------------------------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_LocalityAndStreetBuildingsRank)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"locality": "Zelenograd"}}}}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"building": "2", "locality": "Zelenograd"}}}}}
|
||||
31 {"properties": {"locales": {"default": {"address": {"street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
32 {"properties": {"locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
10 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Zelenograd"}}}}}
|
||||
22 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "2", "locality": "Zelenograd"}}}}}
|
||||
31 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
32 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "2", "street": "Krymskaya", "locality": "Zelenograd"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
ScopedFile const regionsJsonFile("regions.jsonl", kData);
|
||||
geocoder.LoadFromJsonl(regionsJsonFile.GetFullPath());
|
||||
|
||||
TestGeocoder(geocoder, "Zelenograd, Krymskaya 2", {{Id{0x32}, 1.0}, {Id{0x22}, 0.71831}});
|
||||
TestGeocoder(geocoder, "Zelenograd, Krymskaya 2", {{Id{0x32}, 1.0}, {Id{0x22}, 0.72028}});
|
||||
}
|
||||
|
||||
// Geocoder_Subregion* -----------------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_SubregionInLocality)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"region": "Москва"}}}, "rank": 2}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "region": "Москва"}}}, "rank": 4}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"subregion": "Северный административный округ", "locality": "Москва", "region": "Москва"}}}, "rank": 3}}
|
||||
10 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Москва"}}}, "rank": 2}}
|
||||
11 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва", "region": "Москва"}}}, "rank": 4}}
|
||||
12 {"properties": {"kind": "district", "locales": {"default": {"address": {"subregion": "Северный административный округ", "locality": "Москва", "region": "Москва"}}}, "rank": 3}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -369,20 +369,20 @@ UNIT_TEST(Geocoder_SubregionInLocality)
|
|||
|
||||
TestGeocoder(geocoder, "Северный административный округ", {{Id{0x12}, 1.0}});
|
||||
TestGeocoder(geocoder, "Москва, Северный административный округ",
|
||||
{{Id{0x12}, 1.0}, {Id{0x11}, 0.314642}, {Id{0x10}, 0.249221}});
|
||||
TestGeocoder(geocoder, "Москва", {{Id{0x11}, 1.0}, {Id{0x10}, 0.792079}});
|
||||
{{Id{0x12}, 1.0}, {Id{0x11}, 0.316181}, {Id{0x10}, 0.251085}});
|
||||
TestGeocoder(geocoder, "Москва", {{Id{0x11}, 1.0}, {Id{0x10}, 0.794118}});
|
||||
}
|
||||
|
||||
// Geocoder_NumericalSuburb* ----------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_NumericalSuburbRelevance)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"region": "Metro Manila"}}}}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"locality": "Caloocan", "region": "Metro Manila"}}}}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"suburb": "60", "locality": "Caloocan", "region": "Metro Manila"}}}}}
|
||||
20 {"properties": {"locales": {"default": {"address": {"locality": "Белгород"}}}}}
|
||||
21 {"properties": {"locales": {"default": {"address": {"street": "Щорса", "locality": "Белгород"}}}}}
|
||||
22 {"properties": {"locales": {"default": {"address": {"building": "60", "street": "Щорса", "locality": "Белгород"}}}}}
|
||||
10 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Metro Manila"}}}}}
|
||||
11 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Caloocan", "region": "Metro Manila"}}}}}
|
||||
12 {"properties": {"kind": "suburb", "locales": {"default": {"address": {"suburb": "60", "locality": "Caloocan", "region": "Metro Manila"}}}}}
|
||||
20 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Белгород"}}}}}
|
||||
21 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Щорса", "locality": "Белгород"}}}}}
|
||||
22 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "60", "street": "Щорса", "locality": "Белгород"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
|
|
@ -399,11 +399,11 @@ UNIT_TEST(Geocoder_NumericalSuburbRelevance)
|
|||
UNIT_TEST(Geocoder_Serialization)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"locales": {"default": {"address": {"country": "Россия"}}, "en": {"address": {"country": "Russia"}}}, "rank": 1}}
|
||||
11 {"properties": {"locales": {"default": {"address": {"region": "Москва", "country": "Россия"}}}, "rank": 2}}
|
||||
12 {"properties": {"locales": {"default": {"address": {"locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 4}}
|
||||
13 {"properties": {"locales": {"default": {"address": {"street": "Арбат", "locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 7}}
|
||||
15 {"properties": {"locales": {"default": {"address": {"building": "4", "street": "Арбат", "locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 8}}
|
||||
10 {"properties": {"kind": "country", "locales": {"default": {"address": {"country": "Россия"}}, "en": {"address": {"country": "Russia"}}}, "rank": 1}}
|
||||
11 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Москва", "country": "Россия"}}}, "rank": 2}}
|
||||
12 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 4}}
|
||||
13 {"properties": {"kind": "street", "locales": {"default": {"address": {"street": "Арбат", "locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 7}}
|
||||
15 {"properties": {"kind": "building", "locales": {"default": {"address": {"building": "4", "street": "Арбат", "locality": "Москва", "region": "Москва", "country": "Россия"}}}, "rank": 8}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoderFromJsonl;
|
||||
|
|
@ -462,7 +462,7 @@ UNIT_TEST(Geocoder_BigFileConcurrentRead)
|
|||
<< "{"
|
||||
<< R"("type": "Feature",)"
|
||||
<< R"("geometry": {"type": "Point", "coordinates": [0, 0]},)"
|
||||
<< R"("properties": {"locales": {"default": {)"
|
||||
<< R"("properties": {"kind": "country", "locales": {"default": {)"
|
||||
<< R"("name": ")" << i << R"(", "address": {"country": ")" << i << R"("}}}, "rank": 2})"
|
||||
<< "}\n";
|
||||
}
|
||||
|
|
@ -473,4 +473,32 @@ UNIT_TEST(Geocoder_BigFileConcurrentRead)
|
|||
|
||||
TEST_EQUAL(geocoder.GetHierarchy().GetEntries().size(), kEntryCount, ());
|
||||
}
|
||||
|
||||
//--------------------------------------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_CityVsHamletRankTest)
|
||||
{
|
||||
string const kData = R"#(
|
||||
10 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Оренбургская область"}}}}}
|
||||
11 {"properties": {"kind": "hamlet", "locales": {"default": {"address": {"locality": "Красноярск", "region": "Оренбургская область"}}}}}
|
||||
20 {"properties": {"kind": "state", "locales": {"default": {"address": {"region": "Красноярский край"}}}}}
|
||||
21 {"properties": {"kind": "city", "locales": {"default": {"address": {"locality": "Красноярск", "region": "Красноярский край"}}}}}
|
||||
)#";
|
||||
|
||||
Geocoder geocoder;
|
||||
ScopedFile const regionsJsonFile("regions.jsonl", kData);
|
||||
geocoder.LoadFromJsonl(regionsJsonFile.GetFullPath());
|
||||
|
||||
TestGeocoder(geocoder, "Красноярск", {{Id{0x21}, 1.0}, {Id{0x11}, 0.2099}});
|
||||
}
|
||||
|
||||
// Kind tests --------------------------------------------------------------------------------------
|
||||
UNIT_TEST(Geocoder_KindStringConversion)
|
||||
{
|
||||
TEST_EQUAL(static_cast<int>(Kind::Unknown), 0, ());
|
||||
for (auto i = 1; i < static_cast<int>(Kind::Count); ++i)
|
||||
{
|
||||
auto const kind = static_cast<Kind>(i);
|
||||
TEST_EQUAL(kind, KindFromString(ToString(kind)), ());
|
||||
}
|
||||
}
|
||||
} // namespace geocoder
|
||||
|
|
|
|||
|
|
@ -48,13 +48,17 @@ bool Hierarchy::Entry::DeserializeFromJSONImpl(
|
|||
if (!DeserializeAddressFromJSON(root, normalizedNameDictionaryBuilder, stats))
|
||||
return false;
|
||||
|
||||
coding::JsonValue const & properties = coding::GetJsonObligatoryField(root, "properties");
|
||||
coding::JsonValue const & defaultLocale =
|
||||
coding::GetJsonObligatoryFieldByPath(root, "properties", "locales", "default");
|
||||
coding::FromJsonObjectOptionalField(defaultLocale, "name", m_name);
|
||||
coding::GetJsonObligatoryFieldByPath(properties, "locales", "default");
|
||||
|
||||
coding::FromJsonObjectOptionalField(defaultLocale, "name", m_name);
|
||||
if (m_name.empty())
|
||||
++stats.m_emptyNames;
|
||||
|
||||
if (auto const * kind = coding::GetJsonOptionalField(properties, "kind"))
|
||||
m_kind = KindFromString(kind->GetString());
|
||||
|
||||
if (m_type == Type::Count)
|
||||
{
|
||||
LOG(LDEBUG, ("No address in an hierarchy entry:", jsonStr));
|
||||
|
|
|
|||
|
|
@ -73,6 +73,7 @@ public:
|
|||
ar & m_osmId;
|
||||
ar & m_name;
|
||||
ar & m_type;
|
||||
ar & m_kind;
|
||||
ar & m_normalizedAddress;
|
||||
}
|
||||
|
||||
|
|
@ -101,6 +102,7 @@ public:
|
|||
std::string m_name;
|
||||
|
||||
Type m_type = Type::Count;
|
||||
Kind m_kind{Kind::Unknown};
|
||||
|
||||
// The positions of entry address fields in normalized name dictionary, one per Type.
|
||||
std::array<NameDictionary::Position, static_cast<size_t>(Type::Count)> m_normalizedAddress{};
|
||||
|
|
|
|||
|
|
@ -27,4 +27,71 @@ string DebugPrint(Type type)
|
|||
{
|
||||
return ToString(type);
|
||||
}
|
||||
|
||||
char const * ToString(Kind kind)
|
||||
{
|
||||
switch (kind)
|
||||
{
|
||||
case Kind::Country:
|
||||
return "country";
|
||||
case Kind::State:
|
||||
return "state";
|
||||
case Kind::Province:
|
||||
return "province";
|
||||
case Kind::District:
|
||||
return "district";
|
||||
case Kind::County:
|
||||
return "county";
|
||||
case Kind::Municipality:
|
||||
return "municipality";
|
||||
case Kind::City:
|
||||
return "city";
|
||||
case Kind::Town:
|
||||
return "town";
|
||||
case Kind::Village:
|
||||
return "village";
|
||||
case Kind::Hamlet:
|
||||
return "hamlet";
|
||||
case Kind::IsolatedDwelling:
|
||||
return "isolated_dwelling";
|
||||
case Kind::Suburb:
|
||||
return "suburb";
|
||||
case Kind::Quarter:
|
||||
return "quarter";
|
||||
case Kind::Neighbourhood:
|
||||
return "neighbourhood";
|
||||
case Kind::Street:
|
||||
return "street";
|
||||
case Kind::Building:
|
||||
return "building";
|
||||
case Kind::Unknown:
|
||||
case Kind::Count:
|
||||
UNREACHABLE();
|
||||
};
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
Kind KindFromString(std::string const & str)
|
||||
{
|
||||
static auto const string2kindMap = std::map<std::string, Kind>{
|
||||
{"country", Kind::Country},
|
||||
{"state", Kind::State},
|
||||
{"province", Kind::Province},
|
||||
{"district", Kind::District},
|
||||
{"county", Kind::County},
|
||||
{"municipality", Kind::Municipality},
|
||||
{"city", Kind::City},
|
||||
{"town", Kind::Town},
|
||||
{"village", Kind::Village},
|
||||
{"hamlet", Kind::Hamlet},
|
||||
{"isolated_dwelling", Kind::IsolatedDwelling},
|
||||
{"suburb", Kind::Suburb},
|
||||
{"quarter", Kind::Quarter},
|
||||
{"neighbourhood", Kind::Neighbourhood},
|
||||
{"street", Kind::Street},
|
||||
{"building", Kind::Building},
|
||||
};
|
||||
auto it = string2kindMap.find(str);
|
||||
return it != string2kindMap.end() ? it->second : Kind::Unknown;
|
||||
}
|
||||
} // namespace geocoder
|
||||
|
|
|
|||
|
|
@ -7,7 +7,7 @@
|
|||
|
||||
namespace geocoder
|
||||
{
|
||||
enum : unsigned int { kIndexFormatVersion = 1 };
|
||||
enum : unsigned int { kIndexFormatVersion = 2 };
|
||||
|
||||
using Tokens = std::vector<std::string>;
|
||||
|
||||
|
|
@ -29,4 +29,31 @@ enum class Type
|
|||
|
||||
std::string ToString(Type type);
|
||||
std::string DebugPrint(Type type);
|
||||
|
||||
enum class Kind
|
||||
{
|
||||
Unknown = 0,
|
||||
|
||||
Country,
|
||||
State,
|
||||
Province,
|
||||
District,
|
||||
County,
|
||||
Municipality,
|
||||
City,
|
||||
Town,
|
||||
Village,
|
||||
Hamlet,
|
||||
IsolatedDwelling,
|
||||
Suburb,
|
||||
Quarter,
|
||||
Neighbourhood,
|
||||
Street,
|
||||
Building,
|
||||
|
||||
Count
|
||||
};
|
||||
|
||||
char const * ToString(Kind kind);
|
||||
Kind KindFromString(std::string const & str);
|
||||
} // namespace geocoder
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue