From 2df644d3565d42cf04b0228d8621c002856ca3dc Mon Sep 17 00:00:00 2001 From: Vladimir Byko-Ianko Date: Mon, 23 Dec 2019 17:13:37 +0300 Subject: [PATCH] [track_analyzer] Moving to uint64 DataPoints number. --- .../track_analyzer/balance_csv_impl.cpp | 20 +++++++++---------- .../track_analyzer/balance_csv_impl.hpp | 6 +++--- track_analyzing/track_analyzer/utils.cpp | 10 +++++----- track_analyzing/track_analyzer/utils.hpp | 4 ++-- 4 files changed, 20 insertions(+), 20 deletions(-) diff --git a/track_analyzing/track_analyzer/balance_csv_impl.cpp b/track_analyzing/track_analyzer/balance_csv_impl.cpp index 619cd83c48..46387ce59c 100644 --- a/track_analyzing/track_analyzer/balance_csv_impl.cpp +++ b/track_analyzing/track_analyzer/balance_csv_impl.cpp @@ -36,7 +36,7 @@ void FillTable(basic_istream & tableCsvStream, MwmToDataPoints & matchedDa } void RemoveKeysSmallValue(MwmToDataPoints & checkedMap, MwmToDataPoints & additionalMap, - uint32_t ignoreDataPointNumber) + uint64_t ignoreDataPointNumber) { CHECK(AreKeysEqual(additionalMap, checkedMap), ("Mwms in |checkedMap| and in |additionalMap| should have the same set of keys.")); @@ -59,7 +59,7 @@ void RemoveKeysSmallValue(MwmToDataPoints & checkedMap, MwmToDataPoints & additi MwmToDataPointFraction GetMwmToDataPointFraction(MwmToDataPoints const & numberMapping) { CHECK(!numberMapping.empty(), ()); - uint32_t const totalDataPointNumber = ValueSum(numberMapping); + uint64_t const totalDataPointNumber = ValueSum(numberMapping); MwmToDataPointFraction fractionMapping; for (auto const & kv : numberMapping) @@ -81,7 +81,7 @@ MwmToDataPoints CalcsMatchedDataPointsToKeepDistribution( double maxRatio = 0.0; double maxRationDistributionFraction = 0.0; - uint32_t maxRationMatchedDataPointNumber = 0; + uint64_t maxRationMatchedDataPointNumber = 0; // First, let's find such mwm that all |matchedDataPoints| of it may be used and // the distribution set in |distributionFractions| will be kept. It's an mwm // on which the maximum of ratio @@ -112,7 +112,7 @@ MwmToDataPoints CalcsMatchedDataPointsToKeepDistribution( // fraction of data point in the distribution for this mwm (less or equal 1.0) it's possible to // calculate the total matched points number which may be used to keep the distribution. auto const totalMatchedPointNumberToKeepDistribution = - static_cast(maxRationMatchedDataPointNumber / maxRationDistributionFraction); + static_cast(maxRationMatchedDataPointNumber / maxRationDistributionFraction); CHECK_LESS_OR_EQUAL(totalMatchedPointNumberToKeepDistribution, ValueSum(matchedDataPoints), ()); // Having total maximum matched point number which let to keep the distribution @@ -125,7 +125,7 @@ MwmToDataPoints CalcsMatchedDataPointsToKeepDistribution( auto const & mwm = kv.first; auto const fraction = kv.second; auto const matchedDataPointsToKeepDistributionForMwm = - static_cast(fraction * totalMatchedPointNumberToKeepDistribution); + static_cast(fraction * totalMatchedPointNumberToKeepDistribution); matchedDataPointsToKeepDistribution.emplace(mwm, matchedDataPointsToKeepDistributionForMwm); if (matchedDataPointsToKeepDistributionForMwm == 0) @@ -141,7 +141,7 @@ MwmToDataPoints CalcsMatchedDataPointsToKeepDistribution( MwmToDataPoints BalancedDataPointNumber(MwmToDataPoints && distribution, MwmToDataPoints && matchedDataPoints, - uint32_t ignoreDataPointsNumber) + uint64_t ignoreDataPointsNumber) { // Removing every mwm from |distribution| and |matchedDataPoints| if it has // |ignoreDataPointsNumber| data points or less in |matchedDataPoints|. @@ -202,19 +202,19 @@ void FilterTable(MwmToDataPoints const & balancedDataPointNumbers, vector & distributionCsvStream, - basic_istream & tableCsvStream, uint32_t ignoreDataPointsNumber, + basic_istream & tableCsvStream, uint64_t ignoreDataPointsNumber, vector & balancedTable) { LOG(LINFO, ("Balancing data points...")); // Filling a map mwm to DataPoints number according to distribution csv file. MwmToDataPoints distribution; MappingFromCsv(distributionCsvStream, distribution); - uint32_t const totalDistributionDataPointsNumber = ValueSum(distribution); + uint64_t const totalDistributionDataPointsNumber = ValueSum(distribution); balancedTable.clear(); MwmToDataPoints matchedDataPoints; FillTable(tableCsvStream, matchedDataPoints, balancedTable); - uint32_t const totalMatchedDataPointsNumber = ValueSum(matchedDataPoints); + uint64_t const totalMatchedDataPointsNumber = ValueSum(matchedDataPoints); if (matchedDataPoints.empty()) { @@ -236,7 +236,7 @@ void BalanceDataPoints(basic_istream & distributionCsvStream, // Calculating how many points should have every mwm to keep the |distribution|. MwmToDataPoints const balancedDataPointNumber = BalancedDataPointNumber(move(distribution), move(matchedDataPoints), ignoreDataPointsNumber); - uint32_t const totalBalancedDataPointsNumber = ValueSum(balancedDataPointNumber); + uint64_t const totalBalancedDataPointsNumber = ValueSum(balancedDataPointNumber); // |balancedTable| is filled now with all the items from |tableCsvStream|. // Removing some items form |tableCsvStream| (if it's necessary) to correspond to diff --git a/track_analyzing/track_analyzer/balance_csv_impl.hpp b/track_analyzing/track_analyzer/balance_csv_impl.hpp index 839e429221..7f87aad864 100644 --- a/track_analyzing/track_analyzer/balance_csv_impl.hpp +++ b/track_analyzing/track_analyzer/balance_csv_impl.hpp @@ -28,7 +28,7 @@ void FillTable(std::basic_istream & tableCsvStream, MwmToDataPoints & matc /// \brief Removing every mwm from |checkedMap| and |additionalMap| if it has /// |ignoreDataPointsNumber| data points or less in |checkedMap|. void RemoveKeysSmallValue(MwmToDataPoints & checkedMap, MwmToDataPoints & additionalMap, - uint32_t ignoreDataPointNumber); + uint64_t ignoreDataPointNumber); /// \returns mapping from mwm to fraction of data points contained in the mwm. MwmToDataPointFraction GetMwmToDataPointFraction(MwmToDataPoints const & numberMapping); @@ -44,7 +44,7 @@ MwmToDataPoints CalcsMatchedDataPointsToKeepDistribution( /// \returns mapping with number of matched data points for every mwm to correspond to |distribution|. MwmToDataPoints BalancedDataPointNumber(MwmToDataPoints && distribution, MwmToDataPoints && matchedDataPoints, - uint32_t ignoreDataPointsNumber); + uint64_t ignoreDataPointsNumber); /// \breif Leaves in |table| only number of items according to |balancedDataPointNumbers|. /// \note |table| may have a significant size. It may be several tens millions records. @@ -53,7 +53,7 @@ void FilterTable(MwmToDataPoints const & balancedDataPointNumbers, std::vector & distributionCsvStream, - std::basic_istream & tableCsvStream, uint32_t ignoreDataPointsNumber, + std::basic_istream & tableCsvStream, uint64_t ignoreDataPointsNumber, std::vector & balancedTable); template diff --git a/track_analyzing/track_analyzer/utils.cpp b/track_analyzing/track_analyzer/utils.cpp index 35d2dc1b3f..a9f82148a1 100644 --- a/track_analyzing/track_analyzer/utils.cpp +++ b/track_analyzing/track_analyzer/utils.cpp @@ -98,7 +98,7 @@ void Stats::AddTracksStats(MwmToTracks const & mwmToTracks, NumMwmIds const & nu for (auto const & kv : mwmToTracks) { auto const & userToTrack = kv.second; - uint32_t dataPointNum = 0; + uint64_t dataPointNum = 0; for (auto const & userTrack : userToTrack) dataPointNum += userTrack.second.size(); @@ -111,7 +111,7 @@ void Stats::AddTracksStats(MwmToTracks const & mwmToTracks, NumMwmIds const & nu } void Stats::AddDataPoints(string const & mwmName, string const & countryName, - uint32_t dataPointNum) + uint64_t dataPointNum) { m_mwmToTotalDataPoints[mwmName] += dataPointNum; m_countryToTotalDataPoints[countryName] += dataPointNum; @@ -142,7 +142,7 @@ void MappingToCsv(string const & keyName, Stats::NameToCountMapping const & mapp struct KeyValue { string m_key; - uint32_t m_value = 0; + uint64_t m_value = 0; }; if (mapping.empty()) @@ -163,7 +163,7 @@ void MappingToCsv(string const & keyName, Stats::NameToCountMapping const & mapp sort(keyValues.begin(), keyValues.end(), [](KeyValue const & a, KeyValue const & b) { return a.m_value > b.m_value; }); - uint32_t allValues = 0; + uint64_t allValues = 0; for (auto const & kv : keyValues) allValues += kv.m_value; @@ -189,7 +189,7 @@ void MappingFromCsv(basic_istream & ss, Stats::NameToCountMapping & mappin auto const & key = row[0]; uint64_t value = 0; CHECK(strings::to_uint64(row[1], value), ()); - auto const it = mapping.insert(make_pair(key, base::checked_cast(value))); + auto const it = mapping.insert(make_pair(key, value)); CHECK(it.second, ()); } } diff --git a/track_analyzing/track_analyzer/utils.hpp b/track_analyzing/track_analyzer/utils.hpp index f32408243a..945d2dd303 100644 --- a/track_analyzing/track_analyzer/utils.hpp +++ b/track_analyzing/track_analyzer/utils.hpp @@ -19,7 +19,7 @@ class Stats public: friend std::string DebugPrint(Stats const & s); - using NameToCountMapping = std::map; + using NameToCountMapping = std::map; Stats() = default; Stats(NameToCountMapping const & mwmToTotalDataPoints, @@ -34,7 +34,7 @@ public: /// \brief Adds |dataPointNum| to |m_mwmToTotalDataPoints| and |m_countryToTotalDataPoints|. void AddDataPoints(std::string const & mwmName, std::string const & countryName, - uint32_t dataPointNum); + uint64_t dataPointNum); /// \brief Saves csv file with numbers of DataPoints for each mwm to |csvPath|. /// If |csvPath| is empty it does nothing.