Merge pull request #11554 from Enmk/extractAllGroups-Horizontal-and-Vertical

extractAllGroupsHorizontal and extractAllGroupsVertical

src/DataTypes/tests/gtest_data_type_get_common_type.cpp

@@ -14,11 +14,6 @@ static bool operator==(const IDataType & left, const IDataType & right)
     return left.equals(right);
 }
 
-std::ostream & operator<<(std::ostream & ostr, const IDataType & dt)
-{
-    return ostr << dt.getName();
-}
-
 }
 
 using namespace DB;

src/Functions/extractAllGroups.cpp

@@ -1,127 +0,0 @@
-#include <Columns/ColumnString.h>
-#include <Columns/ColumnArray.h>
-#include <Columns/ColumnConst.h>
-#include <DataTypes/DataTypeArray.h>
-#include <DataTypes/DataTypeString.h>
-#include <Functions/FunctionFactory.h>
-#include <Functions/FunctionHelpers.h>
-#include <Functions/Regexps.h>
-
-#include <memory>
-#include <string>
-#include <vector>
-
-
-namespace DB
-{
-
-namespace ErrorCodes
-{
-    extern const int ARGUMENT_OUT_OF_BOUND;
-    extern const int BAD_ARGUMENTS;
-}
-
-
-/** Match all groups of given input string with given re, return array of arrays of matches.
- *
- *  SELECT extractAllGroups('abc=111, def=222, ghi=333', '("[^"]+"|\\w+)=("[^"]+"|\\w+)')
- * should produce:
- *   [['abc', '111'], ['def', '222'], ['ghi', '333']]
- */
-class FunctionExtractAllGroups : public IFunction
-{
-public:
-    static constexpr auto name = "extractAllGroups";
-    static FunctionPtr create(const Context &) { return std::make_shared<FunctionExtractAllGroups>(); }
-
-    String getName() const override { return name; }
-
-    size_t getNumberOfArguments() const override { return 2; }
-
-    bool useDefaultImplementationForConstants() const override { return false; }
-    ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; }
-
-    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
-    {
-        FunctionArgumentDescriptors args{
-            {"haystack", isStringOrFixedString, nullptr, "const String or const FixedString"},
-            {"needle", isStringOrFixedString, isColumnConst, "const String or const FixedString"},
-        };
-        validateFunctionArgumentTypes(*this, arguments, args);
-
-        /// Two-dimensional array of strings, each `row` of top array represents matching groups.
-        return std::make_shared<DataTypeArray>(std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>()));
-    }
-
-    void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override
-    {
-        const ColumnPtr column_haystack = block.getByPosition(arguments[0]).column;
-        const ColumnPtr column_needle = block.getByPosition(arguments[1]).column;
-
-        const auto needle = typeid_cast<const ColumnConst &>(*column_needle).getValue<String>();
-
-        if (needle.empty())
-            throw Exception(getName() + " length of 'needle' argument must be greater than 0.", ErrorCodes::ARGUMENT_OUT_OF_BOUND);
-
-        const auto regexp = Regexps::get<false, false>(needle);
-        const auto & re2 = regexp->getRE2();
-
-        if (!re2)
-            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
-
-        const size_t groups_count = re2->NumberOfCapturingGroups();
-
-        if (!groups_count)
-            throw Exception("There is no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
-
-        // Including 0-group, which is the whole regexp.
-        PODArrayWithStackMemory<re2_st::StringPiece, 128> matched_groups(groups_count + 1);
-
-        ColumnArray::ColumnOffsets::MutablePtr root_offsets_col = ColumnArray::ColumnOffsets::create();
-        ColumnArray::ColumnOffsets::MutablePtr nested_offsets_col = ColumnArray::ColumnOffsets::create();
-        ColumnString::MutablePtr data_col = ColumnString::create();
-
-        auto & root_offsets_data = root_offsets_col->getData();
-        auto & nested_offsets_data = nested_offsets_col->getData();
-
-        root_offsets_data.resize(input_rows_count);
-        ColumnArray::Offset current_root_offset = 0;
-        ColumnArray::Offset current_nested_offset = 0;
-
-        for (size_t i = 0; i < input_rows_count; ++i)
-        {
-            StringRef current_row = column_haystack->getDataAt(i);
-
-            // Extract all non-intersecting matches from haystack except group #0.
-            const auto * pos = current_row.data;
-            const auto * end = pos + current_row.size;
-            while (pos < end
-                && re2->Match(re2_st::StringPiece(pos, end - pos),
-                    0, end - pos, re2_st::RE2::UNANCHORED, matched_groups.data(), matched_groups.size()))
-            {
-                // 1 is to exclude group #0 which is whole re match.
-                for (size_t group = 1; group <= groups_count; ++group)
-                    data_col->insertData(matched_groups[group].data(), matched_groups[group].size());
-
-                pos = matched_groups[0].data() + matched_groups[0].size();
-
-                current_nested_offset += groups_count;
-                nested_offsets_data.push_back(current_nested_offset);
-
-                ++current_root_offset;
-            }
-
-            root_offsets_data[i] = current_root_offset;
-        }
-        ColumnArray::MutablePtr nested_array_col = ColumnArray::create(std::move(data_col), std::move(nested_offsets_col));
-        ColumnArray::MutablePtr root_array_col = ColumnArray::create(std::move(nested_array_col), std::move(root_offsets_col));
-        block.getByPosition(result).column = std::move(root_array_col);
-    }
-};
-
-void registerFunctionExtractAllGroups(FunctionFactory & factory)
-{
-    factory.registerFunction<FunctionExtractAllGroups>();
-}
-
-}

src/Functions/extractAllGroups.h

@@ -0,0 +1,238 @@
+#include <Columns/ColumnArray.h>
+#include <Columns/ColumnConst.h>
+#include <Columns/ColumnString.h>
+#include <DataTypes/DataTypeArray.h>
+#include <DataTypes/DataTypeString.h>
+#include <Functions/FunctionHelpers.h>
+#include <Functions/IFunctionImpl.h>
+#include <Functions/Regexps.h>
+
+#include <memory>
+#include <string>
+#include <vector>
+
+#include <Core/iostream_debug_helpers.h>
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int BAD_ARGUMENTS;
+}
+
+
+enum class ExtractAllGroupsResultKind
+{
+    VERTICAL,
+    HORIZONTAL
+};
+
+
+/** Match all groups of given input string with given re, return array of arrays of matches.
+ *
+ * Depending on `Impl::Kind`, result is either grouped by grop id (Horizontal) or in order of appearance (Vertical):
+ *
+ *  SELECT extractAllGroupsVertical('abc=111, def=222, ghi=333', '("[^"]+"|\\w+)=("[^"]+"|\\w+)')
+ * =>
+ *   [['abc', '111'], ['def', '222'], ['ghi', '333']]
+ *
+ *  SELECT extractAllGroupsHorizontal('abc=111, def=222, ghi=333', '("[^"]+"|\\w+)=("[^"]+"|\\w+)')
+ * =>
+ *   [['abc', 'def', 'ghi'], ['111', '222', '333']
+*/
+template <typename Impl>
+class FunctionExtractAllGroups : public IFunction
+{
+public:
+    static constexpr auto Kind = Impl::Kind;
+    static constexpr auto name = Impl::Name;
+
+    static FunctionPtr create(const Context &) { return std::make_shared<FunctionExtractAllGroups>(); }
+
+    String getName() const override { return name; }
+
+    size_t getNumberOfArguments() const override { return 2; }
+
+    bool useDefaultImplementationForConstants() const override { return false; }
+    ColumnNumbers getArgumentsThatAreAlwaysConstant() const override { return {1}; }
+
+    DataTypePtr getReturnTypeImpl(const ColumnsWithTypeAndName & arguments) const override
+    {
+        FunctionArgumentDescriptors args{
+            {"haystack", isStringOrFixedString, nullptr, "const String or const FixedString"},
+            {"needle", isStringOrFixedString, isColumnConst, "const String or const FixedString"},
+        };
+        validateFunctionArgumentTypes(*this, arguments, args);
+
+        /// Two-dimensional array of strings, each `row` of top array represents matching groups.
+        return std::make_shared<DataTypeArray>(std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>()));
+    }
+
+    void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override
+    {
+        static const auto MAX_GROUPS_COUNT = 128;
+
+        const ColumnPtr column_haystack = block.getByPosition(arguments[0]).column;
+        const ColumnPtr column_needle = block.getByPosition(arguments[1]).column;
+
+        const auto needle = typeid_cast<const ColumnConst &>(*column_needle).getValue<String>();
+
+        if (needle.empty())
+            throw Exception("Length of 'needle' argument must be greater than 0.", ErrorCodes::BAD_ARGUMENTS);
+
+        using StringPiece = typename Regexps::Regexp::StringPieceType;
+        const auto & regexp = Regexps::get<false, false>(needle)->getRE2();
+
+        if (!regexp)
+            throw Exception("There are no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
+        const size_t groups_count = regexp->NumberOfCapturingGroups();
+
+        if (!groups_count)
+            throw Exception("There are no groups in regexp: " + needle, ErrorCodes::BAD_ARGUMENTS);
+
+        if (groups_count > MAX_GROUPS_COUNT - 1)
+            throw Exception("Too many groups in regexp: " + std::to_string(groups_count)
+                            + ", max: " + std::to_string(MAX_GROUPS_COUNT - 1),
+                            ErrorCodes::BAD_ARGUMENTS);
+
+        // Including 0-group, which is the whole regexp.
+        PODArrayWithStackMemory<StringPiece, MAX_GROUPS_COUNT> matched_groups(groups_count + 1);
+
+        ColumnArray::ColumnOffsets::MutablePtr root_offsets_col = ColumnArray::ColumnOffsets::create();
+        ColumnArray::ColumnOffsets::MutablePtr nested_offsets_col = ColumnArray::ColumnOffsets::create();
+        ColumnString::MutablePtr data_col = ColumnString::create();
+
+        auto & root_offsets_data = root_offsets_col->getData();
+        auto & nested_offsets_data = nested_offsets_col->getData();
+
+        ColumnArray::Offset current_root_offset = 0;
+        ColumnArray::Offset current_nested_offset = 0;
+
+        if constexpr (Kind == ExtractAllGroupsResultKind::VERTICAL)
+        {
+            root_offsets_data.resize(input_rows_count);
+            for (size_t i = 0; i < input_rows_count; ++i)
+            {
+                StringRef current_row = column_haystack->getDataAt(i);
+
+                // Extract all non-intersecting matches from haystack except group #0.
+                const auto * pos = current_row.data;
+                const auto * end = pos + current_row.size;
+                while (pos < end
+                    && regexp->Match({pos, static_cast<size_t>(end - pos)},
+                        0, end - pos, regexp->UNANCHORED, matched_groups.data(), matched_groups.size()))
+                {
+                    // 1 is to exclude group #0 which is whole re match.
+                    for (size_t group = 1; group <= groups_count; ++group)
+                        data_col->insertData(matched_groups[group].data(), matched_groups[group].size());
+
+                    pos = matched_groups[0].data() + matched_groups[0].size();
+
+                    current_nested_offset += groups_count;
+                    nested_offsets_data.push_back(current_nested_offset);
+
+                    ++current_root_offset;
+                }
+
+                root_offsets_data[i] = current_root_offset;
+            }
+        }
+        else
+        {
+            std::vector<StringPiece> all_matches;
+            // number of times RE matched on each row of haystack column.
+            std::vector<size_t> number_of_matches_per_row;
+
+            // we expect RE to match multiple times on each row, `* 8` is arbitrary to reduce number of re-allocations.
+            all_matches.reserve(input_rows_count * groups_count * 8);
+            number_of_matches_per_row.reserve(input_rows_count);
+
+            for (size_t i = 0; i < input_rows_count; ++i)
+            {
+                size_t matches_per_row = 0;
+
+                const auto & current_row = column_haystack->getDataAt(i);
+
+                // Extract all non-intersecting matches from haystack except group #0.
+                const auto * pos = current_row.data;
+                const auto * end = pos + current_row.size;
+                while (pos < end
+                    && regexp->Match({pos, static_cast<size_t>(end - pos)},
+                        0, end - pos, regexp->UNANCHORED, matched_groups.data(), matched_groups.size()))
+                {
+                    // 1 is to exclude group #0 which is whole re match.
+                    for (size_t group = 1; group <= groups_count; ++group)
+                        all_matches.push_back(matched_groups[group]);
+
+                    pos = matched_groups[0].data() + matched_groups[0].size();
+
+                    ++matches_per_row;
+                }
+
+                number_of_matches_per_row.push_back(matches_per_row);
+            }
+
+            {
+                size_t total_matched_groups_string_len = 0;
+                for (const auto & m : all_matches)
+                    total_matched_groups_string_len += m.length();
+
+                data_col->reserve(total_matched_groups_string_len);
+            }
+
+            nested_offsets_col->reserve(matched_groups.size());
+            root_offsets_col->reserve(groups_count);
+
+            // Re-arrange `all_matches` from:
+            // [
+            //      "ROW 0: 1st group 1st match",
+            //      "ROW 0: 2nd group 1st match",
+            //      ...,
+            //      "ROW 0: 1st group 2nd match",
+            //      "ROW 0: 2nd group 2nd match",
+            //      ...,
+            //      "ROW 1: 1st group 1st match",
+            //      ...
+            // ]
+            //
+            // into column of 2D arrays:
+            // [
+            //      /* all matchig groups from ROW 0 of haystack column */
+            //      ["ROW 0: 1st group 1st match", "ROW 0: 1st group 2nd match", ...],
+            //      ["ROW 0: 2nd group 1st match", "ROW 0: 2nd group 2nd match", ...],
+            //      ...
+            // ],
+            // [
+            //      /* all matchig groups from row 1 of haystack column */
+            //      ["ROW 1: 1st group 1st match", ...],
+            //      ...
+            // ]
+
+            size_t row_offset = 0;
+            for (const auto matches_per_row : number_of_matches_per_row)
+            {
+                const size_t next_row_offset = row_offset + matches_per_row * groups_count;
+                for (size_t group_id = 0; group_id < groups_count; ++group_id)
+                {
+                    for (size_t i = row_offset + group_id; i < next_row_offset && i < all_matches.size(); i += groups_count)
+                    {
+                        const auto & match = all_matches[i];
+                        data_col->insertData(match.begin(), match.length());
+                    }
+                    nested_offsets_col->insertValue(data_col->size());
+                }
+                root_offsets_col->insertValue(nested_offsets_col->size());
+                row_offset = next_row_offset;
+            }
+        }
+        DUMP(Kind, needle, column_haystack, root_offsets_col, nested_offsets_col);
+
+        ColumnArray::MutablePtr nested_array_col = ColumnArray::create(std::move(data_col), std::move(nested_offsets_col));
+        ColumnArray::MutablePtr root_array_col = ColumnArray::create(std::move(nested_array_col), std::move(root_offsets_col));
+        block.getByPosition(result).column = std::move(root_array_col);
+    }
+};
+
+}

src/Functions/extractAllGroupsHorizontal.cpp

@@ -0,0 +1,23 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/extractAllGroups.h>
+
+namespace
+{
+
+struct HorizontalImpl
+{
+    static constexpr auto Kind = DB::ExtractAllGroupsResultKind::HORIZONTAL;
+    static constexpr auto Name = "extractAllGroupsHorizontal";
+};
+
+}
+
+namespace DB
+{
+
+void registerFunctionExtractAllGroupsHorizontal(FunctionFactory & factory)
+{
+    factory.registerFunction<FunctionExtractAllGroups<HorizontalImpl>>();
+}
+
+}

src/Functions/extractAllGroupsVertical.cpp

@@ -0,0 +1,24 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/extractAllGroups.h>
+
+namespace
+{
+
+struct VerticalImpl
+{
+    static constexpr auto Kind = DB::ExtractAllGroupsResultKind::VERTICAL;
+    static constexpr auto Name = "extractAllGroupsVertical";
+};
+
+}
+
+namespace DB
+{
+
+void registerFunctionExtractAllGroupsVertical(FunctionFactory & factory)
+{
+    factory.registerFunction<FunctionExtractAllGroups<VerticalImpl>>();
+    factory.registerAlias("extractAllGroups", VerticalImpl::Name, FunctionFactory::CaseInsensitive);
+}
+
+}

src/Functions/extractGroups.cpp

@@ -17,7 +17,6 @@ namespace DB
 
 namespace ErrorCodes
 {
-    extern const int ARGUMENT_OUT_OF_BOUND;
     extern const int BAD_ARGUMENTS;
 }
 
@@ -49,7 +48,6 @@ public:
         };
         validateFunctionArgumentTypes(*this, arguments, args);
 
-        /// Two-dimensional array of strings, each `row` of top array represents matching groups.
         return std::make_shared<DataTypeArray>(std::make_shared<DataTypeString>());
     }
 
@@ -61,7 +59,7 @@ public:
         const auto needle = typeid_cast<const ColumnConst &>(*column_needle).getValue<String>();
 
         if (needle.empty())
-            throw Exception(getName() + " length of 'needle' argument must be greater than 0.", ErrorCodes::ARGUMENT_OUT_OF_BOUND);
+            throw Exception(getName() + " length of 'needle' argument must be greater than 0.", ErrorCodes::BAD_ARGUMENTS);
 
         const auto regexp = Regexps::get<false, false>(needle);
         const auto & re2 = regexp->getRE2();

src/Functions/registerFunctionsStringRegexp.cpp

@@ -18,7 +18,8 @@ void registerFunctionMultiFuzzyMatchAny(FunctionFactory &);
 void registerFunctionMultiFuzzyMatchAnyIndex(FunctionFactory &);
 void registerFunctionMultiFuzzyMatchAllIndices(FunctionFactory &);
 void registerFunctionExtractGroups(FunctionFactory &);
-void registerFunctionExtractAllGroups(FunctionFactory &);
+void registerFunctionExtractAllGroupsVertical(FunctionFactory &);
+void registerFunctionExtractAllGroupsHorizontal(FunctionFactory &);
 
 void registerFunctionsStringRegexp(FunctionFactory & factory)
 {
@@ -37,7 +38,8 @@ void registerFunctionsStringRegexp(FunctionFactory & factory)
     registerFunctionMultiFuzzyMatchAnyIndex(factory);
     registerFunctionMultiFuzzyMatchAllIndices(factory);
     registerFunctionExtractGroups(factory);
-    registerFunctionExtractAllGroups(factory);
+    registerFunctionExtractAllGroupsVertical(factory);
+    registerFunctionExtractAllGroupsHorizontal(factory);
 }
 
 }

src/Functions/ya.make

@@ -144,7 +144,8 @@ SRCS(
     exp10.cpp
     exp2.cpp
     exp.cpp
-    extractAllGroups.cpp
+    extractAllGroupsHorizontal.cpp
+    extractAllGroupsVertical.cpp
     extract.cpp
     extractGroups.cpp
     extractTimeZoneFromFunctionArguments.cpp

tests/performance/extract.xml

@@ -6,5 +6,6 @@
     <query>SELECT count() FROM test.hits WHERE NOT ignore(extract(URL, '(\\w+=\\w+)'))</query>
     <query>SELECT count() FROM test.hits WHERE NOT ignore(extractAll(URL, '(\\w+=\\w+)'))</query>
     <query>SELECT count() FROM test.hits WHERE NOT ignore(extractGroups(URL, '(\\w+)=(\\w+)'))</query>
-    <query>SELECT count() FROM test.hits WHERE NOT ignore(extractAllGroups(URL, '(\\w+)=(\\w+)'))</query>
+    <query>SELECT count() FROM test.hits WHERE NOT ignore(extractAllGroupsVertical(URL, '(\\w+)=(\\w+)'))</query>
+    <query>SELECT count() FROM test.hits WHERE NOT ignore(extractAllGroupsHorizontal(URL, '(\\w+)=(\\w+)'))</query>
 </test>

tests/queries/0_stateless/01246_extractAllGroups.sql

@@ -1,51 +0,0 @@
--- error cases
-SELECT extractAllGroups();  --{serverError 42} not enough arguments
-SELECT extractAllGroups('hello');  --{serverError 42} not enough arguments
-SELECT extractAllGroups('hello', 123);  --{serverError 43} invalid argument type
-SELECT extractAllGroups(123, 'world');  --{serverError 43}  invalid argument type
-SELECT extractAllGroups('hello world', '((('); --{serverError 427}  invalid re
-SELECT extractAllGroups('hello world', materialize('\\w+')); --{serverError 44} non-const needle
-
-SELECT '0 groups, zero matches';
-SELECT extractAllGroups('hello world', '\\w+'); -- { serverError 36 }
-
-SELECT '1 group, multiple matches, String and FixedString';
-SELECT extractAllGroups('hello world', '(\\w+)');
-SELECT extractAllGroups('hello world', CAST('(\\w+)' as FixedString(5)));
-SELECT extractAllGroups(CAST('hello world' AS FixedString(12)), '(\\w+)');
-SELECT extractAllGroups(CAST('hello world' AS FixedString(12)), CAST('(\\w+)' as FixedString(5)));
-SELECT extractAllGroups(materialize(CAST('hello world' AS FixedString(12))), '(\\w+)');
-SELECT extractAllGroups(materialize(CAST('hello world' AS FixedString(12))), CAST('(\\w+)' as FixedString(5)));
-
-SELECT 'mutiple groups, multiple matches';
-SELECT extractAllGroups('abc=111, def=222, ghi=333 "jkl mno"="444 foo bar"', '("[^"]+"|\\w+)=("[^"]+"|\\w+)');
-
-SELECT 'big match';
-SELECT
-    length(haystack), length(matches[1]), length(matches), arrayMap((x) -> length(x), arrayMap(x -> x[1], matches))
-FROM (
-    SELECT
-        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
-        extractAllGroups(haystack, '(abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz)') AS matches
-    FROM numbers(3)
-);
-
-SELECT 'lots of matches';
-SELECT
-    length(haystack), length(matches[1]), length(matches), arrayReduce('sum', arrayMap((x) -> length(x), arrayMap(x -> x[1], matches)))
-FROM (
-    SELECT
-        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
-        extractAllGroups(haystack, '(\\w)') AS matches
-    FROM numbers(3)
-);
-
-SELECT 'lots of groups';
-SELECT
-    length(haystack), length(matches[1]), length(matches), arrayMap((x) -> length(x), arrayMap(x -> x[1], matches))
-FROM (
-    SELECT
-        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
-        extractAllGroups(haystack, repeat('(\\w)', 100)) AS matches
-    FROM numbers(3)
-);

tests/queries/0_stateless/01246_extractAllGroupsHorizontal.reference

@@ -0,0 +1,22 @@
+0 groups, zero matches
+1 group, multiple matches, String and FixedString
+[['hello','world']]
+[['hello','world']]
+[['hello','world']]
+[['hello','world']]
+[['hello','world']]
+[['hello','world']]
+mutiple groups, multiple matches
+[['abc','def','ghi','"jkl mno"'],['111','222','333','"444 foo bar"']]
+big match
+0	1	0	[]
+260	1	1	[156]
+520	1	3	[156,156,156]
+lots of matches
+0	1	0	0
+260	1	260	260
+520	1	520	520
+lots of groups
+0	100	0	[]
+260	100	2	[1,1]
+520	100	5	[1,1,1,1,1]

tests/queries/0_stateless/01246_extractAllGroupsHorizontal.sql

@@ -0,0 +1,51 @@
+-- error cases
+SELECT extractAllGroupsHorizontal();  --{serverError 42} not enough arguments
+SELECT extractAllGroupsHorizontal('hello');  --{serverError 42} not enough arguments
+SELECT extractAllGroupsHorizontal('hello', 123);  --{serverError 43} invalid argument type
+SELECT extractAllGroupsHorizontal(123, 'world');  --{serverError 43}  invalid argument type
+SELECT extractAllGroupsHorizontal('hello world', '((('); --{serverError 427}  invalid re
+SELECT extractAllGroupsHorizontal('hello world', materialize('\\w+')); --{serverError 44} non-cons needle
+
+SELECT '0 groups, zero matches';
+SELECT extractAllGroupsHorizontal('hello world', '\\w+');  -- { serverError 36 }
+
+SELECT '1 group, multiple matches, String and FixedString';
+SELECT extractAllGroupsHorizontal('hello world', '(\\w+)');
+SELECT extractAllGroupsHorizontal('hello world', CAST('(\\w+)' as FixedString(5)));
+SELECT extractAllGroupsHorizontal(CAST('hello world' AS FixedString(12)), '(\\w+)');
+SELECT extractAllGroupsHorizontal(CAST('hello world' AS FixedString(12)), CAST('(\\w+)' as FixedString(5)));
+SELECT extractAllGroupsHorizontal(materialize(CAST('hello world' AS FixedString(12))), '(\\w+)');
+SELECT extractAllGroupsHorizontal(materialize(CAST('hello world' AS FixedString(12))), CAST('(\\w+)' as FixedString(5)));
+
+SELECT 'mutiple groups, multiple matches';
+SELECT extractAllGroupsHorizontal('abc=111, def=222, ghi=333 "jkl mno"="444 foo bar"', '("[^"]+"|\\w+)=("[^"]+"|\\w+)');
+
+SELECT 'big match';
+SELECT
+    length(haystack), length(matches), length(matches[1]), arrayMap((x) -> length(x), matches[1])
+FROM (
+    SELECT
+           repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+           extractAllGroupsHorizontal(haystack, '(abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz)') AS matches
+    FROM numbers(3)
+);
+
+SELECT 'lots of matches';
+SELECT
+    length(haystack), length(matches), length(matches[1]), arrayReduce('sum', arrayMap((x) -> length(x), matches[1]))
+FROM (
+    SELECT
+        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+        extractAllGroupsHorizontal(haystack, '(\\w)') AS matches
+    FROM numbers(3)
+);
+
+SELECT 'lots of groups';
+SELECT
+    length(haystack), length(matches), length(matches[1]), arrayMap((x) -> length(x), matches[1])
+FROM (
+    SELECT
+        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+        extractAllGroupsHorizontal(haystack, repeat('(\\w)', 100)) AS matches
+    FROM numbers(3)
+);

tests/queries/0_stateless/01246_extractAllGroupsVertical.sql

@@ -0,0 +1,51 @@
+-- error cases
+SELECT extractAllGroupsVertical();  --{serverError 42} not enough arguments
+SELECT extractAllGroupsVertical('hello');  --{serverError 42} not enough arguments
+SELECT extractAllGroupsVertical('hello', 123);  --{serverError 43} invalid argument type
+SELECT extractAllGroupsVertical(123, 'world');  --{serverError 43}  invalid argument type
+SELECT extractAllGroupsVertical('hello world', '((('); --{serverError 427}  invalid re
+SELECT extractAllGroupsVertical('hello world', materialize('\\w+')); --{serverError 44} non-const needle
+
+SELECT '0 groups, zero matches';
+SELECT extractAllGroupsVertical('hello world', '\\w+'); -- { serverError 36 }
+
+SELECT '1 group, multiple matches, String and FixedString';
+SELECT extractAllGroupsVertical('hello world', '(\\w+)');
+SELECT extractAllGroupsVertical('hello world', CAST('(\\w+)' as FixedString(5)));
+SELECT extractAllGroupsVertical(CAST('hello world' AS FixedString(12)), '(\\w+)');
+SELECT extractAllGroupsVertical(CAST('hello world' AS FixedString(12)), CAST('(\\w+)' as FixedString(5)));
+SELECT extractAllGroupsVertical(materialize(CAST('hello world' AS FixedString(12))), '(\\w+)');
+SELECT extractAllGroupsVertical(materialize(CAST('hello world' AS FixedString(12))), CAST('(\\w+)' as FixedString(5)));
+
+SELECT 'mutiple groups, multiple matches';
+SELECT extractAllGroupsVertical('abc=111, def=222, ghi=333 "jkl mno"="444 foo bar"', '("[^"]+"|\\w+)=("[^"]+"|\\w+)');
+
+SELECT 'big match';
+SELECT
+    length(haystack), length(matches[1]), length(matches), arrayMap((x) -> length(x), arrayMap(x -> x[1], matches))
+FROM (
+    SELECT
+        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+        extractAllGroupsVertical(haystack, '(abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz)') AS matches
+    FROM numbers(3)
+);
+
+SELECT 'lots of matches';
+SELECT
+    length(haystack), length(matches[1]), length(matches), arrayReduce('sum', arrayMap((x) -> length(x), arrayMap(x -> x[1], matches)))
+FROM (
+    SELECT
+        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+        extractAllGroupsVertical(haystack, '(\\w)') AS matches
+    FROM numbers(3)
+);
+
+SELECT 'lots of groups';
+SELECT
+    length(haystack), length(matches[1]), length(matches), arrayMap((x) -> length(x), arrayMap(x -> x[1], matches))
+FROM (
+    SELECT
+        repeat('abcdefghijklmnopqrstuvwxyz', number * 10) AS haystack,
+        extractAllGroupsVertical(haystack, repeat('(\\w)', 100)) AS matches
+    FROM numbers(3)
+);

tests/queries/0_stateless/01275_extract_groups_check.sql

@@ -1,5 +1,5 @@
-SELECT extractGroups('hello', ''); -- { serverError 69 }
-SELECT extractAllGroups('hello', ''); -- { serverError 69 }
+SELECT extractGroups('hello', ''); -- { serverError 36 }
+SELECT extractAllGroups('hello', ''); -- { serverError 36 }
 
 SELECT extractGroups('hello', ' '); -- { serverError 36 }
 SELECT extractAllGroups('hello', ' '); -- { serverError 36 }