Merge pull request #38722 from ClickHouse/fix-rank-corr-functions

Fix a bug in `rankCorr` function

src/AggregateFunctions/AggregateFunctionRankCorrelation.h

@@ -31,8 +31,8 @@ struct RankCorrelationData : public StatisticalSample<Float64, Float64>
         RanksArray ranks_y;
         std::tie(ranks_y, std::ignore) = computeRanksAndTieCorrection(this->y);
 
-        /// In our case sizes of both samples are equal.
-        const auto size = this->size_x;
+        /// Sizes can be non-equal due to skipped NaNs.
+        const auto size = std::min(this->size_x, this->size_y);
 
         /// Count d^2 sum
         Float64 answer = 0;

src/AggregateFunctions/StatCommon.h

@@ -31,8 +31,8 @@ std::pair<RanksArray, Float64> computeRanksAndTieCorrection(const Values & value
     /// Save initial positions, than sort indices according to the values.
     std::vector<size_t> indexes(size);
     std::iota(indexes.begin(), indexes.end(), 0);
-    ::sort(indexes.begin(), indexes.end(),
-                [&] (size_t lhs, size_t rhs) { return values[lhs] < values[rhs]; });
+    std::sort(indexes.begin(), indexes.end(),
+        [&] (size_t lhs, size_t rhs) { return values[lhs] < values[rhs]; });
 
     size_t left = 0;
     Float64 tie_numenator = 0;
@@ -74,12 +74,18 @@ struct StatisticalSample
 
     void addX(X value, Arena * arena)
     {
+        if (isNaN(value))
+            return;
+
         ++size_x;
         x.push_back(value, arena);
     }
 
     void addY(Y value, Arena * arena)
     {
+        if (isNaN(value))
+            return;
+
         ++size_y;
         y.push_back(value, arena);
     }

tests/queries/0_stateless/01848_http_insert_segfault.sh

@@ -1,7 +1,9 @@
 #!/usr/bin/env bash
+# Tags: no-tsan
+# Sometimes is takes longer than 60 seconds under TSan.
 
- CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
- # shellcheck source=../shell_config.sh
- . "$CUR_DIR"/../shell_config.sh
+CUR_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CUR_DIR"/../shell_config.sh
 
- ${CLICKHOUSE_LOCAL} -q "select col1, initializeAggregation('argMaxState', col2, insertTime) as col2, now() as insertTime FROM generateRandom('col1 String, col2 Array(Float64)') LIMIT 1000000 FORMAT CSV"  | ${CLICKHOUSE_CURL} -s 'http://localhost:8123/?query=INSERT%20INTO%20non_existing_table%20SELECT%20col1%2C%20initializeAggregation(%27argMaxState%27%2C%20col2%2C%20insertTime)%20as%20col2%2C%20now()%20as%20insertTime%20FROM%20input(%27col1%20String%2C%20col2%20Array(Float64)%27)%20FORMAT%20CSV' --data-binary @- | grep -q "Table default.non_existing_table doesn't exist" && echo 'Ok.' || echo 'FAIL' ||:
+${CLICKHOUSE_LOCAL} --query "select col1, initializeAggregation('argMaxState', col2, insertTime) as col2, now() as insertTime FROM generateRandom('col1 String, col2 Array(Float64)') LIMIT 1000000 FORMAT CSV"  | ${CLICKHOUSE_CURL} -s 'http://localhost:8123/?query=INSERT%20INTO%20non_existing_table%20SELECT%20col1%2C%20initializeAggregation(%27argMaxState%27%2C%20col2%2C%20insertTime)%20as%20col2%2C%20now()%20as%20insertTime%20FROM%20input(%27col1%20String%2C%20col2%20Array(Float64)%27)%20FORMAT%20CSV' --data-binary @- | grep -q "Table default.non_existing_table doesn't exist" && echo 'Ok.' || echo 'FAIL' ||:

tests/queries/0_stateless/02347_rank_corr_nan.reference

@@ -0,0 +1 @@
+nan

tests/queries/0_stateless/02347_rank_corr_nan.sql

@@ -0,0 +1 @@
+SELECT rankCorr(number, nan) FROM numbers(10);