pastk/organicmaps
1
Fork 0
forked from organicmaps/organicmaps
Code Pull requests Activity

Add SponsoredDataset as an abstruction.

Browse source
This commit is contained in:
Sergey Magidovich 2016-08-25 12:36:48 +03:00
parent 78ea6a4038
commit 389ddf670f
10 changed files with 438 additions and 306 deletions

245
generator/booking_dataset.cpp
View file

@ -3,133 +3,46 @@
#include "generator/booking_scoring.hpp"
#include "generator/feature_builder.hpp"
#include "platform/local_country_file_utils.hpp"
#include "platform/platform.hpp"
#include "indexer/classificator.hpp"
#include "indexer/ftypes_matcher.hpp"
#include "geometry/distance_on_sphere.hpp"
#include "base/logging.hpp"
#include "base/string_utils.hpp"
#include "std/fstream.hpp"
#include "std/iostream.hpp"
#include "std/sstream.hpp"
namespace generator
{
namespace
{
string EscapeTabs(string const & str)
{
stringstream ss;
for (char c : str)
{
if (c == '\t')
ss << "\\t";
else
ss << c;
}
return ss.str();
}
} // namespace
// BookingDataset::AddressMatcher::AddressMatcher()
// {
// vector<platform::LocalCountryFile> localFiles;
BookingDataset::BookingId const BookingDataset::kInvalidHotelIndex = BookingId(numeric_limits<BookingId::RepType>::max());
// Platform & platform = GetPlatform();
// platform::FindAllLocalMapsInDirectoryAndCleanup(platform.WritableDir(), 0 /* version */,
// -1 /* latestVersion */, localFiles);
BookingDataset::Hotel::Hotel(string const & src)
{
vector<string> rec;
strings::ParseCSVRow(src, '\t', rec);
CHECK(rec.size() == FieldsCount(), ("Error parsing hotels.tsv line:", EscapeTabs(src)));
// for (platform::LocalCountryFile const & localFile : localFiles)
// {
// LOG(LINFO, ("Found mwm:", localFile));
// try
// {
// m_index.RegisterMap(localFile);
// }
// catch (RootException const & ex)
// {
// CHECK(false, ("Bad mwm file:", localFile));
// }
// }
strings::to_uint(rec[Index(Fields::Id)], id.Get());
strings::to_double(rec[Index(Fields::Latitude)], lat);
strings::to_double(rec[Index(Fields::Longtitude)], lon);
// m_coder = make_unique<search::ReverseGeocoder>(m_index);
// }
name = rec[Index(Fields::Name)];
address = rec[Index(Fields::Address)];
// void BookingDataset::AddressMatcher::operator()(Hotel & hotel)
// {
// search::ReverseGeocoder::Address addr;
// m_coder->GetNearbyAddress(MercatorBounds::FromLatLon(hotel.lat, hotel.lon), addr);
// hotel.street = addr.GetStreetName();
// hotel.houseNumber = addr.GetHouseNumber();
// }
strings::to_uint(rec[Index(Fields::Stars)], stars);
strings::to_uint(rec[Index(Fields::PriceCategory)], priceCategory);
strings::to_double(rec[Index(Fields::RatingBooking)], ratingBooking);
strings::to_double(rec[Index(Fields::RatingUsers)], ratingUser);
descUrl = rec[Index(Fields::DescUrl)];
strings::to_uint(rec[Index(Fields::Type)], type);
translations = rec[Index(Fields::Translations)];
}
ostream & operator<<(ostream & s, BookingDataset::Hotel const & h)
{
s << fixed << setprecision(7);
return s << "Id: " << h.id << "\t Name: " << h.name << "\t Address: " << h.address
<< "\t lat: " << h.lat << " lon: " << h.lon;
}
BookingDataset::AddressMatcher::AddressMatcher()
{
vector<platform::LocalCountryFile> localFiles;
Platform & platform = GetPlatform();
platform::FindAllLocalMapsInDirectoryAndCleanup(platform.WritableDir(), 0 /* version */,
-1 /* latestVersion */, localFiles);
for (platform::LocalCountryFile const & localFile : localFiles)
{
LOG(LINFO, ("Found mwm:", localFile));
try
{
m_index.RegisterMap(localFile);
}
catch (RootException const & ex)
{
CHECK(false, ("Bad mwm file:", localFile));
}
}
m_coder = make_unique<search::ReverseGeocoder>(m_index);
}
void BookingDataset::AddressMatcher::operator()(Hotel & hotel)
{
search::ReverseGeocoder::Address addr;
m_coder->GetNearbyAddress(MercatorBounds::FromLatLon(hotel.lat, hotel.lon), addr);
hotel.street = addr.GetStreetName();
hotel.houseNumber = addr.GetHouseNumber();
}
BookingDataset::BookingDataset(string const & dataPath, string const & addressReferencePath)
{
if (dataPath.empty())
return;
ifstream dataSource(dataPath);
if (!dataSource.is_open())
{
LOG(LERROR, ("Error while opening", dataPath, ":", strerror(errno)));
return;
}
LoadHotels(dataSource, addressReferencePath);
}
BookingDataset::BookingDataset(istream & dataSource, string const & addressReferencePath)
{
LoadHotels(dataSource, addressReferencePath);
}
BookingDataset::BookingId BookingDataset::GetMatchingHotelId(FeatureBuilder1 const & fb) const
{
if (CanBeBooking(fb))
return MatchWithBooking(fb);
return kInvalidHotelIndex;
}
bool BookingDataset::CanBeBooking(FeatureBuilder1 const & fb) const
bool BookingDataset::NecessaryMatchingConditionHolds(FeatureBuilder1 const & fb) const
{
if (fb.GetName(StringUtf8Multilang::kDefaultCode).empty())
return false;
@ -137,49 +50,8 @@ bool BookingDataset::CanBeBooking(FeatureBuilder1 const & fb) const
return ftypes::IsHotelChecker::Instance()(fb.GetTypes());
}
BookingDataset::Hotel const & BookingDataset::GetHotelById(BookingId const id) const
{
auto const it = m_hotels.find(id);
CHECK(it != end(m_hotels), ("Got wrong hotel id:", id));
return it->second;
}
BookingDataset::Hotel & BookingDataset::GetHotelById(BookingId const id)
{
auto it = m_hotels.find(id);
CHECK(it != end(m_hotels), ("Got wrong hotel id:", id));
return it->second;
}
vector<BookingDataset::BookingId> BookingDataset::GetNearestHotels(
ms::LatLon const & latLon, size_t const limit,
double const maxDistance /* = 0.0 */) const
{
namespace bgi = boost::geometry::index;
vector<BookingId> indexes;
for_each(bgi::qbegin(m_rtree, bgi::nearest(TPoint(latLon.lat, latLon.lon), limit)),
bgi::qend(m_rtree), [this, &latLon, &indexes, maxDistance](TValue const & v)
{
auto const & hotel = GetHotelById(v.second);
double const dist = ms::DistanceOnEarth(latLon.lat, latLon.lon, hotel.lat, hotel.lon);
if (maxDistance != 0.0 && dist > maxDistance /* max distance in meters */)
return;
indexes.emplace_back(v.second);
});
return indexes;
}
void BookingDataset::BuildHotels(function<void(FeatureBuilder1 &)> const & fn) const
{
for (auto const & item : m_hotels)
BuildHotel(item.second, fn);
}
void BookingDataset::BuildHotel(Hotel const & hotel,
function<void(FeatureBuilder1 &)> const & fn) const
void BookingDataset::BuildObject(SponsoredDataset::Object const & hotel,
function<void(FeatureBuilder1 &)> const & fn) const
{
FeatureBuilder1 fb;
FeatureParams params;
@ -276,68 +148,23 @@ void BookingDataset::BuildHotel(Hotel const & hotel,
fn(fb);
}
void BookingDataset::LoadHotels(istream & src, string const & addressReferencePath)
{
m_hotels.clear();
m_rtree.clear();
for (string line; getline(src, line);)
{
Hotel hotel(line);
m_hotels.emplace(hotel.id, hotel);
}
if (!addressReferencePath.empty())
{
LOG(LINFO, ("Reference addresses for booking objects", addressReferencePath));
Platform & platform = GetPlatform();
string const backupPath = platform.WritableDir();
platform.SetWritableDirForTests(addressReferencePath);
AddressMatcher addressMatcher;
size_t matchedNum = 0;
size_t emptyAddr = 0;
for (auto & item : m_hotels)
{
auto & hotel = item.second;
addressMatcher(hotel);
if (hotel.address.empty())
++emptyAddr;
if (hotel.IsAddressPartsFilled())
++matchedNum;
}
LOG(LINFO,
("Num of hotels:", m_hotels.size(), "matched:", matchedNum, "empty addresses:", emptyAddr));
platform.SetWritableDirForTests(backupPath);
}
for (auto const & item : m_hotels)
{
auto const & hotel = item.second;
TBox b(TPoint(hotel.lat, hotel.lon), TPoint(hotel.lat, hotel.lon));
m_rtree.insert(make_pair(b, hotel.id));
}
}
BookingDataset::BookingId BookingDataset::MatchWithBooking(FeatureBuilder1 const & fb) const
BookingDataset::ObjectId BookingDataset::FindMatchingObjectIdImpl(FeatureBuilder1 const & fb) const
{
auto const name = fb.GetName(StringUtf8Multilang::kDefaultCode);
if (name.empty())
return kInvalidHotelIndex;
return kInvalidObjectId;
// Find |kMaxSelectedElements| nearest values to a point.
auto const bookingIndexes = GetNearestHotels(MercatorBounds::ToLatLon(fb.GetKeyPoint()),
kMaxSelectedElements, kDistanceLimitInMeters);
auto const bookingIndexes = GetNearestObjects(MercatorBounds::ToLatLon(fb.GetKeyPoint()),
kMaxSelectedElements, kDistanceLimitInMeters);
for (auto const j : bookingIndexes)
{
if (booking_scoring::Match(GetHotelById(j), fb).IsMatched())
if (booking_scoring::Match(GetObjectById(j), fb).IsMatched())
return j;
}
return kInvalidHotelIndex;
return kInvalidObjectId;
}
} // namespace generator

107
generator/booking_dataset.hpp
View file

@ -2,6 +2,8 @@
#include "indexer/index.hpp"
#include "generator/sponsored_dataset.hpp"
#include "search/reverse_geocoder.hpp"
#include "base/newtype.hpp"
@ -12,7 +14,6 @@
#include "boost/geometry/index/rtree.hpp"
#include "std/function.hpp"
#include "std/limits.hpp"
#include "std/map.hpp"
#include "std/string.hpp"
@ -20,102 +21,36 @@ class FeatureBuilder1;
namespace generator
{
class BookingDataset
class BookingDataset : public SponsoredDatasetBase
{
public:
NEWTYPE(uint32_t, BookingId);
static double constexpr kDistanceLimitInMeters = 150;
static size_t constexpr kMaxSelectedElements = 3;
static BookingId const kInvalidHotelIndex;
// class AddressMatcher
// {
// Index m_index;
// unique_ptr<search::ReverseGeocoder> m_coder;
struct Hotel
// public:
// AddressMatcher();
// void operator()(Hotel & hotel);
// };
explicit BookingDataset(string const & dataPath, string const & addressReferencePath = string())
: SponsoredDatasetBase(dataPath, addressReferencePath)
{
enum class Fields
{
Id = 0,
Latitude = 1,
Longtitude = 2,
Name = 3,
Address = 4,
Stars = 5,
PriceCategory = 6,
RatingBooking = 7,
RatingUsers = 8,
DescUrl = 9,
Type = 10,
Translations = 11,
}
Counter
};
BookingId id{kInvalidHotelIndex};
double lat = 0.0;
double lon = 0.0;
string name;
string address;
string street;
string houseNumber;
uint32_t stars = 0;
uint32_t priceCategory = 0;
double ratingBooking = 0.0;
double ratingUser = 0.0;
string descUrl;
uint32_t type = 0;
string translations;
static constexpr size_t Index(Fields field) { return static_cast<size_t>(field); }
static constexpr size_t FieldsCount() { return static_cast<size_t>(Fields::Counter); }
explicit Hotel(string const & src);
inline bool IsAddressPartsFilled() const { return !street.empty() || !houseNumber.empty(); }
};
class AddressMatcher
explicit BookingDataset(istream & dataSource, string const & addressReferencePath = string())
: SponsoredDatasetBase(dataSource, addressReferencePath)
{
Index m_index;
unique_ptr<search::ReverseGeocoder> m_coder;
}
public:
AddressMatcher();
void operator()(Hotel & hotel);
};
explicit BookingDataset(string const & dataPath, string const & addressReferencePath = string());
explicit BookingDataset(istream & dataSource, string const & addressReferencePath = string());
/// @return an id of a matched hotel or kInvalidHotelIndex on failure.
BookingId GetMatchingHotelId(FeatureBuilder1 const & fb) const;
/// @return true if |fb| is a hotel with a name.
bool CanBeBooking(FeatureBuilder1 const & fb) const;
inline size_t Size() const { return m_hotels.size(); }
Hotel const & GetHotelById(BookingId id) const;
Hotel & GetHotelById(BookingId id);
vector<BookingId> GetNearestHotels(ms::LatLon const & latLon, size_t limit,
double maxDistance = 0.0) const;
bool MatchByName(string const & osmName, vector<size_t> const & bookingIndexes) const;
void BuildHotels(function<void(FeatureBuilder1 &)> const & fn) const;
bool NecessaryMatchingConditionHolds(FeatureBuilder1 const & fb) const override;
protected:
map<BookingId, Hotel> m_hotels;
void BuildObject(Object const & hotel, function<void(FeatureBuilder1 &)> const & fn) const override;
using TPoint = boost::geometry::model::point<float, 2, boost::geometry::cs::cartesian>;
using TBox = boost::geometry::model::box<TPoint>;
using TValue = pair<TBox, BookingId>;
// Create the rtree using default constructor.
boost::geometry::index::rtree<TValue, boost::geometry::index::quadratic<16>> m_rtree;
void BuildHotel(Hotel const & hotel, function<void(FeatureBuilder1 &)> const & fn) const;
void LoadHotels(istream & path, string const & addressReferencePath);
/// @return an id of a matched hotel or kInvalidHotelIndex on failure.
BookingId MatchWithBooking(FeatureBuilder1 const & e) const;
ObjectId FindMatchingObjectIdImpl(FeatureBuilder1 const & e) const override;
};
ostream & operator<<(ostream & s, BookingDataset::Hotel const & h);
NEWTYPE_SIMPLE_OUTPUT(BookingDataset::BookingId);
} // namespace generator

12
generator/booking_quality_check/booking_quality_check.cpp
View file

@ -101,7 +101,7 @@ public:
void operator()(FeatureBuilder1 & fb) override
{
if (m_bookingDataset.CanBeBooking(fb))
if (m_bookingDataset.NecessaryMatchingConditionHolds(fb))
m_features.emplace(fb.GetMostGenericOsmId(), fb);
}
@ -139,11 +139,11 @@ feature::GenerateInfo GetGenerateInfo()
struct SampleItem
{
enum MatchStatus {Uninitialized, Yes, No};
using BookingId = BookingDataset::BookingID;
using ObjectId = SponsoredDataset::ObjectId;
SampleItem() = default;
SampleItem(osm::Id const & osmId, BookingId const bookingId, MatchStatus const match = Uninitialized)
SampleItem(osm::Id const & osmId, ObjectId const bookingId, MatchStatus const match = Uninitialized)
: m_osmId(osmId)
, m_bookingId(bookingId)
, m_match(match)
@ -151,7 +151,7 @@ struct SampleItem
}
osm::Id m_osmId;
BookingId m_bookingId = BookingDataset::kInvalidHotelIndex;
ObjectId m_bookingId = SponsoredDataset::kInvalidObjectId;
MatchStatus m_match = Uninitialized;
};
@ -179,7 +179,7 @@ SampleItem ReadSampleItem(string const & str)
"due to wrong number of fields."));
item.m_osmId = ReadDebuggedPrintedOsmId(parts[0]);
if (!strings::to_uint(parts[1], item.m_bookingId))
if (!strings::to_uint(parts[1], item.m_bookingId.Get()))
MYTHROW(ParseError, ("Can't make uint32 from string:", parts[1]));
item.m_match = ReadMatchStatus(parts[2]);
@ -219,7 +219,7 @@ void GenerateFactors(BookingDataset const & booking, map<osm::Id, FeatureBuilder
{
for (auto const & item : sampleItems)
{
auto const & hotel = booking.GetHotelById(item.m_bookingId);
auto const & hotel = booking.GetObjectById(item.m_bookingId);
auto const & feature = features.at(item.m_osmId);
auto const score = booking_scoring::Match(hotel, feature);

2
generator/booking_scoring.cpp
View file

@ -126,7 +126,7 @@ bool BookingMatchScore::IsMatched() const
return GetMatchingScore() > kOptimalThreshold;
}
BookingMatchScore Match(BookingDataset::Hotel const & h, FeatureBuilder1 const & fb)
BookingMatchScore Match(BookingDataset::Object const & h, FeatureBuilder1 const & fb)
{
BookingMatchScore score;

2
generator/booking_scoring.hpp
View file

@ -17,6 +17,6 @@ struct BookingMatchScore
double m_nameSimilarityScore{};
};
BookingMatchScore Match(BookingDataset::Hotel const & h, FeatureBuilder1 const & fb);
BookingMatchScore Match(BookingDataset::Object const & h, FeatureBuilder1 const & fb);
} // namespace booking_scoring
} // namespace generator

2
generator/generator.pro
View file

@ -36,6 +36,7 @@ SOURCES += \
region_meta.cpp \
routing_generator.cpp \
search_index_builder.cpp \
sponsored_dataset.cpp \
srtm_parser.cpp \
statistics.cpp \
tesselator.cpp \
@ -72,6 +73,7 @@ HEADERS += \
region_meta.hpp \
routing_generator.hpp \
search_index_builder.hpp \
sponsored_dataset.hpp \
srtm_parser.hpp \
statistics.hpp \
tag_admixer.hpp \

6
generator/osm_source.cpp
View file

@ -349,8 +349,8 @@ public:
[](Place const & p1, Place const & p2) { return p1.IsEqual(p2); },
[](Place const & p1, Place const & p2) { return p1.IsBetterThan(p2); });
}
else if (m_bookingDataset.GetMatchingHotelId(fb) !=
generator::BookingDataset::kInvalidHotelIndex)
else if (m_bookingDataset.FindMatchingObjectId(fb) !=
generator::SponsoredDatasetBase::kInvalidObjectId)
{
m_skippedElements << DebugPrint(fb.GetMostGenericOsmId()) << endl;
@ -389,7 +389,7 @@ public:
DumpSkippedElements();
// Emit all booking objecs to the map.
m_bookingDataset.BuildHotels([this](FeatureBuilder1 & fb) { Emit(fb); });
m_bookingDataset.BuildOsmObjects([this](FeatureBuilder1 & fb) { Emit(fb); });
m_places.ForEach([this](Place const & p)
{

180
generator/sponsored_dataset.cpp Normal file
View file

@ -0,0 +1,180 @@
#include "generator/sponsored_dataset.hpp"
#include "platform/platform.hpp"
#include "geometry/distance_on_sphere.hpp"
#include "base/logging.hpp"
#include "base/string_utils.hpp"
#include "std/limits.hpp"
#include "std/fstream.hpp"
#include "std/iostream.hpp"
namespace generator
{
namespace
{
string EscapeTabs(string const & str)
{
stringstream ss;
for (char c : str)
{
if (c == '\t')
ss << "\\t";
else
ss << c;
}
return ss.str();
}
} // namespace
SponsoredDataset::Object::Object(string const & src)
{
vector<string> rec;
strings::ParseCSVRow(src, '\t', rec);
CHECK(rec.size() == FieldsCount(), ("Error parsing hotels.tsv line:", EscapeTabs(src)));
strings::to_uint(rec[Index(Fields::Id)], id.Get());
strings::to_double(rec[Index(Fields::Latitude)], lat);
strings::to_double(rec[Index(Fields::Longtitude)], lon);
name = rec[Index(Fields::Name)];
address = rec[Index(Fields::Address)];
strings::to_uint(rec[Index(Fields::Stars)], stars);
strings::to_uint(rec[Index(Fields::PriceCategory)], priceCategory);
strings::to_double(rec[Index(Fields::RatingBooking)], ratingBooking);
strings::to_double(rec[Index(Fields::RatingUsers)], ratingUser);
descUrl = rec[Index(Fields::DescUrl)];
strings::to_uint(rec[Index(Fields::Type)], type);
translations = rec[Index(Fields::Translations)];
}
ostream & operator<<(ostream & s, SponsoredDataset::Object const & h)
{
s << fixed << setprecision(7);
return s << "Id: " << h.id << "\t Name: " << h.name << "\t Address: " << h.address
<< "\t lat: " << h.lat << " lon: " << h.lon;
}
SponsoredDataset::ObjectId const SponsoredDataset::kInvalidObjectId =
SponsoredDataset::ObjectId(numeric_limits<SponsoredDataset::ObjectId::RepType>::max());
SponsoredDatasetBase::SponsoredDatasetBase(string const & dataPath, string const & addressReferencePath)
{
if (dataPath.empty())
return;
ifstream dataSource(dataPath);
if (!dataSource.is_open())
{
LOG(LERROR, ("Error while opening", dataPath, ":", strerror(errno)));
return;
}
LoadData(dataSource, addressReferencePath);
}
SponsoredDatasetBase::SponsoredDatasetBase(istream & dataSource, string const & addressReferencePath)
{
LoadData(dataSource, addressReferencePath);
}
SponsoredDataset::Object const & SponsoredDatasetBase::GetObjectById(SponsoredDataset::ObjectId id) const
{
auto const it = m_hotels.find(id);
CHECK(it != end(m_hotels), ("Got wrong object id:", id));
return it->second;
}
SponsoredDataset::Object & SponsoredDatasetBase::GetObjectById(SponsoredDataset::ObjectId id)
{
auto const it = m_hotels.find(id);
CHECK(it != end(m_hotels), ("Got wrong object id:", id));
return it->second;
}
void SponsoredDatasetBase::BuildOsmObjects(function<void(FeatureBuilder1 &)> const & fn) const
{
for (auto const & item : m_hotels)
BuildObject(item.second, fn);
}
SponsoredDatasetBase::ObjectId SponsoredDatasetBase::FindMatchingObjectId(FeatureBuilder1 const & fb) const
{
if (NecessaryMatchingConditionHolds(fb))
return FindMatchingObjectIdImpl(fb);
return kInvalidObjectId;
}
vector<SponsoredDataset::ObjectId> SponsoredDatasetBase::GetNearestObjects(
ms::LatLon const & latLon, size_t const limit,
double const maxDistance /* = 0.0 */) const
{
namespace bgi = boost::geometry::index;
vector<ObjectId> indexes;
for_each(bgi::qbegin(m_rtree, bgi::nearest(TPoint(latLon.lat, latLon.lon), limit)),
bgi::qend(m_rtree), [this, &latLon, &indexes, maxDistance](TValue const & v)
{
auto const & object = GetObjectById(v.second);
double const dist = ms::DistanceOnEarth(latLon.lat, latLon.lon, object.lat, object.lon);
if (maxDistance != 0.0 && dist > maxDistance /* max distance in meters */)
return;
indexes.emplace_back(v.second);
});
return indexes;
}
void SponsoredDatasetBase::LoadData(istream & src, string const & addressReferencePath)
{
m_hotels.clear();
m_rtree.clear();
for (string line; getline(src, line);)
{
Object hotel(line);
m_hotels.emplace(hotel.id, hotel);
}
if (!addressReferencePath.empty())
{
LOG(LINFO, ("Reference addresses for booking objects", addressReferencePath));
Platform & platform = GetPlatform();
string const backupPath = platform.WritableDir();
platform.SetWritableDirForTests(addressReferencePath);
//TODO(mgsergio): AddressMatcher addressMatcher;
size_t matchedNum = 0;
size_t emptyAddr = 0;
for (auto & item : m_hotels)
{
auto & hotel = item.second;
// TODO(mgsergio): addressMatcher(hotel);
if (hotel.address.empty())
++emptyAddr;
if (hotel.IsAddressPartsFilled())
++matchedNum;
}
LOG(LINFO,
("Num of hotels:", m_hotels.size(), "matched:", matchedNum, "empty addresses:", emptyAddr));
platform.SetWritableDirForTests(backupPath);
}
for (auto const & item : m_hotels)
{
auto const & hotel = item.second;
TBox b(TPoint(hotel.lat, hotel.lon), TPoint(hotel.lat, hotel.lon));
m_rtree.insert(make_pair(b, hotel.id));
}
}
} // namespace generator

136
generator/sponsored_dataset.hpp Normal file
View file

@ -0,0 +1,136 @@
#pragma once
#include "indexer/index.hpp"
#include "search/reverse_geocoder.hpp"
#include "base/newtype.hpp"
#include "boost/geometry.hpp"
#include "boost/geometry/geometries/point.hpp"
#include "boost/geometry/geometries/box.hpp"
#include "boost/geometry/index/rtree.hpp"
#include "std/function.hpp"
#include "std/limits.hpp"
#include "std/map.hpp"
#include "std/string.hpp"
class FeatureBuilder1;
namespace generator
{
class SponsoredDataset
{
public:
NEWTYPE(uint32_t, ObjectId);
static double constexpr kDistanceLimitInMeters = 150;
static size_t constexpr kMaxSelectedElements = 3;
static ObjectId const kInvalidObjectId;
struct Object
{
enum class Fields
{
Id = 0,
Latitude = 1,
Longtitude = 2,
Name = 3,
Address = 4,
Stars = 5,
PriceCategory = 6,
RatingBooking = 7,
RatingUsers = 8,
DescUrl = 9,
Type = 10,
Translations = 11,
Counter
};
ObjectId id{kInvalidObjectId};
double lat = 0.0;
double lon = 0.0;
string name;
string address;
string street;
string houseNumber;
uint32_t stars = 0;
uint32_t priceCategory = 0;
double ratingBooking = 0.0;
double ratingUser = 0.0;
string descUrl;
uint32_t type = 0;
string translations;
static constexpr size_t Index(Fields field) { return static_cast<size_t>(field); }
static constexpr size_t FieldsCount() { return static_cast<size_t>(Fields::Counter); }
explicit Object(string const & src);
inline bool IsAddressPartsFilled() const { return !street.empty() || !houseNumber.empty(); }
};
// class AddressMatcher
// {
// Index m_index;
// unique_ptr<search::ReverseGeocoder> m_coder;
// public:
// AddressMatcher();
// void operator()(Hotel & hotel);
// };
virtual ~SponsoredDataset() = default;
// TODO(mgsergio): Comment /// @return an id of a matched hotel or kInvalidHotelIndex on failure.
virtual ObjectId FindMatchingObjectId(FeatureBuilder1 const & e) const = 0;
virtual size_t Size() const = 0;
virtual void BuildOsmObjects(function<void(FeatureBuilder1 &)> const & fn) const = 0;
};
ostream & operator<<(ostream & s, SponsoredDataset::Object const & h);
NEWTYPE_SIMPLE_OUTPUT(SponsoredDataset::ObjectId);
class SponsoredDatasetBase : public SponsoredDataset
{
public:
explicit SponsoredDatasetBase(string const & dataPath, string const & addressReferencePath = string());
explicit SponsoredDatasetBase(istream & dataSource, string const & addressReferencePath = string());
size_t Size() const override { return m_hotels.size(); }
Object const & GetObjectById(ObjectId id) const;
Object & GetObjectById(ObjectId id);
vector<ObjectId> GetNearestObjects(ms::LatLon const & latLon, size_t limit,
double maxDistance = 0.0) const;
/// @return true if |fb| satisfies some necesary conditions to match one or serveral
/// objects from dataset.
virtual bool NecessaryMatchingConditionHolds(FeatureBuilder1 const & fb) const = 0;
/// @return an id of a matched object or kInvalidObjectId on failure.
ObjectId FindMatchingObjectId(FeatureBuilder1 const & e) const override;
void BuildOsmObjects(function<void(FeatureBuilder1 &)> const & fn) const override;
protected:
map<ObjectId, Object> m_hotels;
using TPoint = boost::geometry::model::point<float, 2, boost::geometry::cs::cartesian>;
using TBox = boost::geometry::model::box<TPoint>;
using TValue = pair<TBox, ObjectId>;
// Create the rtree using default constructor.
boost::geometry::index::rtree<TValue, boost::geometry::index::quadratic<16>> m_rtree;
virtual void BuildObject(Object const & object, function<void(FeatureBuilder1 &)> const & fn) const = 0;
void LoadData(istream & src, string const & addressReferencePath);
/// @return an id of a matched object or kInvalidObjectId on failure.
virtual ObjectId FindMatchingObjectIdImpl(FeatureBuilder1 const & e) const = 0;
};
} // namespace generator

52
generator/sponsored_object.hpp Normal file
View file

@ -0,0 +1,52 @@
#pragma once
NEWTYPE(uint32_t, ObjectId);
NEWTYPE_SIMPLE_OUTPUT(ObjectId);
static double constexpr kDistanceLimitInMeters = 150;
static size_t constexpr kMaxSelectedElements = 3;
static ObjectId const kInvalidObjectId;
struct SponsoredObject
{
enum class Fields
{
Id = 0,
Latitude = 1,
Longtitude = 2,
Name = 3,
Address = 4,
Stars = 5,
PriceCategory = 6,
RatingBooking = 7,
RatingUsers = 8,
DescUrl = 9,
Type = 10,
Translations = 11,
Counter
};
ObjectId id{kInvalidObjectId};
double lat = 0.0;
double lon = 0.0;
string name;
string address;
string street;
string houseNumber;
uint32_t stars = 0;
uint32_t priceCategory = 0;
double ratingBooking = 0.0;
double ratingUser = 0.0;
string descUrl;
uint32_t type = 0;
string translations;
static size_t constexpr Index(Fields field) { return static_cast<size_t>(field); }
static size_t constexpr FieldsCount() { return static_cast<size_t>(Fields::Counter); }
explicit SponsoredObject(string const & src);
inline bool IsAddressFilled() const { return !street.empty() || !houseNumber.empty(); }
};
ostream & operator<<(ostream & s, BookingDataset::Hotel const & h);