intermediate changes

ref:cde9a383711a11544ce7e107a78147fb96cc4029

7 files changed, 1130 insertions, 0 deletions

diff --git a/library/cpp/deprecated/split/delim_string_iter.cpp b/library/cpp/deprecated/split/delim_string_iter.cpp
new file mode 100644
index 0000000000..af418c5bfb
--- /dev/null
+++ b/library/cpp/deprecated/split/delim_string_iter.cpp
@@ -0,0 +1,45 @@
+#include "delim_string_iter.h"
+
+//
+// TKeyValueDelimStringIter
+//
+
+void TKeyValueDelimStringIter::ReadKeyAndValue() {
+    TStringBuf currentToken(*DelimIter);
+
+    size_t pos = currentToken.find('=');
+    if (pos == TString::npos) {
+        ChunkValue.Clear();
+        ChunkKey = currentToken;
+    } else {
+        ChunkKey = currentToken.SubStr(0, pos);
+        ChunkValue = currentToken.SubStr(pos + 1);
+    }
+}
+
+TKeyValueDelimStringIter::TKeyValueDelimStringIter(const TStringBuf str, const TStringBuf delim)
+    : DelimIter(str, delim)
+{
+    if (DelimIter.Valid())
+        ReadKeyAndValue();
+}
+
+bool TKeyValueDelimStringIter::Valid() const {
+    return DelimIter.Valid();
+}
+
+TKeyValueDelimStringIter& TKeyValueDelimStringIter::operator++() {
+    ++DelimIter;
+    if (DelimIter.Valid())
+        ReadKeyAndValue();
+
+    return *this;
+}
+
+const TStringBuf& TKeyValueDelimStringIter::Key() const {
+    return ChunkKey;
+}
+
+const TStringBuf& TKeyValueDelimStringIter::Value() const {
+    return ChunkValue;
+}
diff --git a/library/cpp/deprecated/split/delim_string_iter.h b/library/cpp/deprecated/split/delim_string_iter.h
new file mode 100644
index 0000000000..8e4ca171a0
--- /dev/null
+++ b/library/cpp/deprecated/split/delim_string_iter.h
@@ -0,0 +1,185 @@
+#pragma once
+
+#include <util/generic/algorithm.h>
+#include <util/generic/strbuf.h>
+#include <util/generic/yexception.h>
+#include <util/string/cast.h>
+#include <util/system/yassert.h>
+
+#include <iterator>
+
+class TDelimStringIter {
+public:
+    using value_type = TStringBuf;
+    using difference_type = ptrdiff_t;
+    using pointer = const TStringBuf*;
+    using reference = const TStringBuf&;
+    using iterator_category = std::forward_iterator_tag;
+
+    inline TDelimStringIter(const char* begin, const char* strEnd, TStringBuf delim)
+        : TDelimStringIter(TStringBuf(begin, strEnd), delim)
+    {
+    }
+
+    inline TDelimStringIter(TStringBuf str, TStringBuf delim)
+        : IsValid(true)
+        , Str(str)
+        , Delim(delim)
+    {
+        UpdateCurrent();
+    }
+
+    inline TDelimStringIter()
+        : IsValid(false)
+    {
+    }
+
+    inline explicit operator bool() const {
+        return IsValid;
+    }
+
+    // NOTE: this is a potentially unsafe operation (no overrun check)
+    inline TDelimStringIter& operator++() {
+        if (Current.end() != Str.end()) {
+            Str.Skip(Current.length() + Delim.length());
+            UpdateCurrent();
+        } else {
+            Str.Clear();
+            Current.Clear();
+            IsValid = false;
+        }
+        return *this;
+    }
+
+    inline void operator+=(size_t n) {
+        for (; n > 0; --n) {
+            ++(*this);
+        }
+    }
+
+    inline bool operator==(const TDelimStringIter& rhs) const {
+        return (IsValid == rhs.IsValid) && (!IsValid || (Current.begin() == rhs.Current.begin()));
+    }
+
+    inline bool operator!=(const TDelimStringIter& rhs) const {
+        return !(*this == rhs);
+    }
+
+    inline TStringBuf operator*() const {
+        return Current;
+    }
+
+    inline const TStringBuf* operator->() const {
+        return &Current;
+    }
+
+    // Get & advance
+    template <class T>
+    inline bool TryNext(T& t) {
+        if (IsValid) {
+            t = FromString<T>(Current);
+            operator++();
+            return true;
+        } else {
+            return false;
+        }
+    }
+
+    template <class T>
+    inline TDelimStringIter& Next(T& t) // Get & advance
+    {
+        if (!TryNext(t))
+            ythrow yexception() << "No valid field";
+        return *this;
+    }
+
+    template <class T>
+    inline T GetNext() {
+        T res;
+        Next(res);
+        return res;
+    }
+
+    inline const char* GetBegin() const {
+        return Current.begin();
+    }
+
+    inline const char* GetEnd() const {
+        return Current.end();
+    }
+
+    inline bool Valid() const {
+        return IsValid;
+    }
+
+    // contents from next token to the end of string
+    inline TStringBuf Cdr() const {
+        return Str.SubStr(Current.length() + Delim.length());
+    }
+
+    inline TDelimStringIter IterEnd() const {
+        return TDelimStringIter();
+    }
+
+private:
+    inline void UpdateCurrent() {
+        // it is much faster than TStringBuf::find
+        size_t pos = std::search(Str.begin(), Str.end(), Delim.begin(), Delim.end()) - Str.begin();
+        Current = Str.Head(pos);
+    }
+
+private:
+    bool IsValid;
+
+    TStringBuf Str;
+    TStringBuf Current;
+    TStringBuf Delim;
+};
+
+//example: for (TStringBuf field: TDelimStroka(line, "@@")) { ... }
+struct TDelimStroka {
+    TStringBuf S;
+    TStringBuf Delim;
+
+    inline TDelimStroka(TStringBuf s, TStringBuf delim)
+        : S(s)
+        , Delim(delim)
+    {
+    }
+
+    inline TDelimStringIter begin() const {
+        return TDelimStringIter(S, Delim);
+    }
+
+    inline TDelimStringIter end() const {
+        return TDelimStringIter();
+    }
+};
+
+inline TDelimStringIter begin_delim(const TString& str, TStringBuf delim) {
+    return TDelimStringIter(str, delim);
+}
+
+inline TDelimStringIter begin_delim(TStringBuf str, TStringBuf delim) {
+    return TDelimStringIter(str.begin(), str.end(), delim);
+}
+
+inline TDelimStringIter end_delim(const TString& /*str*/, TStringBuf /*delim*/) {
+    return TDelimStringIter();
+}
+
+class TKeyValueDelimStringIter {
+public:
+    TKeyValueDelimStringIter(const TStringBuf str, const TStringBuf delim);
+    bool Valid() const;
+    TKeyValueDelimStringIter& operator++();
+    const TStringBuf& Key() const;
+    const TStringBuf& Value() const;
+
+private:
+    TDelimStringIter DelimIter;
+    TStringBuf ChunkKey, ChunkValue;
+
+private:
+    void ReadKeyAndValue();
+};
diff --git a/library/cpp/deprecated/split/delim_string_iter_ut.cpp b/library/cpp/deprecated/split/delim_string_iter_ut.cpp
new file mode 100644
index 0000000000..18a8b2a160
--- /dev/null
+++ b/library/cpp/deprecated/split/delim_string_iter_ut.cpp
@@ -0,0 +1,99 @@
+#include "delim_string_iter.h"
+#include <util/generic/vector.h>
+#include <library/cpp/testing/unittest/registar.h>
+
+/// Test that TDelimStringIter build on top of given string and delimeter will produce expected sequence
+static void AssertStringSplit(const TString& str, const TString& delim, const TVector<TString>& expected) {
+    TDelimStringIter it(str, delim);
+
+    // test iterator invariants
+    for (const auto& expectedString : expected) {
+        UNIT_ASSERT(it.Valid());
+        UNIT_ASSERT(bool(it));
+        UNIT_ASSERT_STRINGS_EQUAL(it->ToString(), expectedString);
+        ++it;
+    }
+    UNIT_ASSERT(!it.Valid());
+};
+
+Y_UNIT_TEST_SUITE(TDelimStrokaIterTestSuite) {
+    Y_UNIT_TEST(SingleCharacterAsDelimiter) {
+        AssertStringSplit(
+            "Hello words!", " ", {"Hello", "words!"});
+    }
+
+    Y_UNIT_TEST(MultipleCharactersAsDelimiter) {
+        AssertStringSplit(
+            "0, 1, 1, 2, 3, 5, 8, 13, 21, 34", "1, ", {"0, ", "", "2, 3, 5, 8, 13, 2", "34"});
+    }
+
+    Y_UNIT_TEST(NoDelimitersPresent) {
+        AssertStringSplit("This string could be yours", "\t", {"This string could be yours"});
+    }
+
+    Y_UNIT_TEST(Cdr) {
+        TDelimStringIter it("a\tc\t", "\t");
+        UNIT_ASSERT_STRINGS_EQUAL(*it, "a");
+        UNIT_ASSERT_STRINGS_EQUAL(it.Cdr(), "c\t");
+        ++it;
+        UNIT_ASSERT_STRINGS_EQUAL(it.Cdr(), "");
+    }
+
+    Y_UNIT_TEST(ForIter) {
+        TVector<TStringBuf> expected = {"1", "", "3@4", ""};
+        TVector<TStringBuf> got;
+
+        for (TStringBuf x : TDelimStroka("1@@@@3@4@@", "@@")) {
+            got.push_back(x);
+        }
+
+        UNIT_ASSERT_EQUAL(got, expected);
+    }
+}
+
+static void AssertKeyValueStringSplit(
+    const TStringBuf str,
+    const TStringBuf delim,
+    const TVector<std::pair<TStringBuf, TStringBuf>>& expected) {
+    TKeyValueDelimStringIter it(str, delim);
+
+    for (const auto& expectedKeyValue : expected) {
+        UNIT_ASSERT(it.Valid());
+        UNIT_ASSERT_STRINGS_EQUAL(it.Key(), expectedKeyValue.first);
+        UNIT_ASSERT_STRINGS_EQUAL(it.Value(), expectedKeyValue.second);
+        ++it;
+    }
+    UNIT_ASSERT(!it.Valid());
+}
+
+Y_UNIT_TEST_SUITE(TKeyValueDelimStringIterTestSuite) {
+    Y_UNIT_TEST(SingleCharacterAsDelimiter) {
+        AssertKeyValueStringSplit(
+            "abc=123,cde=qwer", ",",
+            {{"abc", "123"},
+             {"cde", "qwer"}});
+    }
+
+    Y_UNIT_TEST(MultipleCharactersAsDelimiter) {
+        AssertKeyValueStringSplit(
+            "abc=xyz@@qwerty=zxcv", "@@",
+            {{"abc", "xyz"},
+             {"qwerty", "zxcv"}});
+    }
+
+    Y_UNIT_TEST(NoDelimiters) {
+        AssertKeyValueStringSplit(
+            "abc=zz", ",",
+            {{"abc", "zz"}});
+    }
+
+    Y_UNIT_TEST(EmptyElements) {
+        AssertKeyValueStringSplit(
+            "@@abc=zxy@@@@qwerty=y@@", "@@",
+            {{"", ""},
+             {"abc", "zxy"},
+             {"", ""},
+             {"qwerty", "y"},
+             {"", ""}});
+    }
+}
diff --git a/library/cpp/deprecated/split/split_iterator.cpp b/library/cpp/deprecated/split/split_iterator.cpp
new file mode 100644
index 0000000000..32262d25bd
--- /dev/null
+++ b/library/cpp/deprecated/split/split_iterator.cpp
@@ -0,0 +1,318 @@
+#include "split_iterator.h"
+
+#include <util/system/yassert.h>
+
+#include <cctype>
+#include <cstring>
+#include <cstdlib>
+
+/****************** TSplitDelimiters2 ******************/
+
+TSplitDelimiters::TSplitDelimiters(const char* s) {
+    memset(Delims, 0, sizeof(Delims));
+    while (*s)
+        Delims[(ui8) * (s++)] = true;
+}
+
+/****************** TSplitBase ******************/
+TSplitBase::TSplitBase(const char* str, size_t length)
+    : Str(str)
+    , Len(length)
+{
+}
+
+TSplitBase::TSplitBase(const TString& s)
+    : Str(s.data())
+    , Len(s.size())
+{
+}
+
+/****************** TDelimitersSplit ******************/
+
+TDelimitersSplit::TDelimitersSplit(const char* str, size_t length, const TSplitDelimiters& delimiters)
+    : TSplitBase(str, length)
+    , Delimiters(delimiters)
+{
+}
+
+TDelimitersSplit::TDelimitersSplit(const TString& s, const TSplitDelimiters& delimiters)
+    : TSplitBase(s)
+    , Delimiters(delimiters)
+{
+}
+
+size_t TDelimitersSplit::Begin() const {
+    size_t pos = 0;
+    while ((pos < Len) && Delimiters.IsDelimiter(Str[pos]))
+        ++pos;
+    return pos;
+}
+
+TSizeTRegion TDelimitersSplit::Next(size_t& pos) const {
+    size_t begin = pos;
+    while ((pos < Len) && !Delimiters.IsDelimiter(Str[pos]))
+        ++pos;
+    TSizeTRegion result(begin, pos);
+
+    while ((pos < Len) && Delimiters.IsDelimiter(Str[pos]))
+        ++pos;
+
+    return result;
+}
+
+TDelimitersSplit::TIterator TDelimitersSplit::Iterator() const {
+    return TIterator(*this);
+}
+
+/****************** TDelimitersStrictSplit ******************/
+
+TDelimitersStrictSplit::TDelimitersStrictSplit(const char* str, size_t length, const TSplitDelimiters& delimiters)
+    : TSplitBase(str, length)
+    , Delimiters(delimiters)
+{
+}
+
+TDelimitersStrictSplit::TDelimitersStrictSplit(const TString& s, const TSplitDelimiters& delimiters)
+    : TSplitBase(s)
+    , Delimiters(delimiters)
+{
+}
+
+TDelimitersStrictSplit::TIterator TDelimitersStrictSplit::Iterator() const {
+    return TIterator(*this);
+}
+
+TSizeTRegion TDelimitersStrictSplit::Next(size_t& pos) const {
+    size_t begin = pos;
+    while ((pos < Len) && !Delimiters.IsDelimiter(Str[pos]))
+        ++pos;
+    TSizeTRegion result(begin, pos);
+
+    if (pos < Len)
+        ++pos;
+
+    return result;
+}
+
+size_t TDelimitersStrictSplit::Begin() const {
+    return 0;
+}
+
+/****************** TScreenedDelimitersSplit ******************/
+
+TScreenedDelimitersSplit::TScreenedDelimitersSplit(const TString& s, const TSplitDelimiters& delimiters, const TSplitDelimiters& screens)
+    : TSplitBase(s)
+    , Delimiters(delimiters)
+    , Screens(screens)
+{
+}
+
+TScreenedDelimitersSplit::TScreenedDelimitersSplit(const char* str, size_t length, const TSplitDelimiters& delimiters, const TSplitDelimiters& screens)
+    : TSplitBase(str, length)
+    , Delimiters(delimiters)
+    , Screens(screens)
+{
+}
+
+TScreenedDelimitersSplit::TIterator TScreenedDelimitersSplit::Iterator() const {
+    return TIterator(*this);
+}
+
+TSizeTRegion TScreenedDelimitersSplit::Next(size_t& pos) const {
+    size_t begin = pos;
+    bool screened = false;
+    while (pos < Len) {
+        if (Screens.IsDelimiter(Str[pos]))
+            screened = !screened;
+        if (Delimiters.IsDelimiter(Str[pos]) && !screened)
+            break;
+        ++pos;
+    }
+    TSizeTRegion result(begin, pos);
+
+    if (pos < Len)
+        ++pos;
+
+    return result;
+}
+
+size_t TScreenedDelimitersSplit::Begin() const {
+    return 0;
+}
+
+/****************** TDelimitersSplitWithoutTags ******************/
+
+TDelimitersSplitWithoutTags::TDelimitersSplitWithoutTags(const char* str, size_t length, const TSplitDelimiters& delimiters)
+    : TSplitBase(str, length)
+    , Delimiters(delimiters)
+{
+}
+
+TDelimitersSplitWithoutTags::TDelimitersSplitWithoutTags(const TString& s, const TSplitDelimiters& delimiters)
+    : TSplitBase(s)
+    , Delimiters(delimiters)
+{
+}
+
+size_t TDelimitersSplitWithoutTags::SkipTag(size_t pos) const {
+    Y_ASSERT('<' == Str[pos]);
+    while ((pos < Len) && ('>' != Str[pos]))
+        ++pos;
+    return pos + 1;
+}
+
+size_t TDelimitersSplitWithoutTags::SkipDelimiters(size_t pos) const {
+    while (true) {
+        while ((pos < Len) && Delimiters.IsDelimiter(Str[pos]) && ('<' != Str[pos]))
+            ++pos;
+        if (pos < Len) {
+            if ('<' != Str[pos])
+                break;
+            else
+                pos = SkipTag(pos);
+        } else
+            break;
+    }
+    return pos;
+}
+
+size_t TDelimitersSplitWithoutTags::Begin() const {
+    size_t pos = 0;
+    pos = SkipDelimiters(pos);
+    return pos;
+}
+
+TSizeTRegion TDelimitersSplitWithoutTags::Next(size_t& pos) const {
+    size_t begin = pos;
+    while ((pos < Len) && !Delimiters.IsDelimiter(Str[pos]) && ('<' != Str[pos]))
+        ++pos;
+    TSizeTRegion result(begin, pos);
+
+    pos = SkipDelimiters(pos);
+
+    return result;
+}
+
+TDelimitersSplitWithoutTags::TIterator TDelimitersSplitWithoutTags::Iterator() const {
+    return TIterator(*this);
+}
+
+/****************** TCharSplit ******************/
+
+TCharSplit::TCharSplit(const char* str, size_t length)
+    : TSplitBase(str, length)
+{
+}
+
+TCharSplit::TCharSplit(const TString& s)
+    : TSplitBase(s)
+{
+}
+
+TCharSplit::TIterator TCharSplit::Iterator() const {
+    return TIterator(*this);
+}
+
+TSizeTRegion TCharSplit::Next(size_t& pos) const {
+    TSizeTRegion result(pos, pos + 1);
+    ++pos;
+    return result;
+}
+
+size_t TCharSplit::Begin() const {
+    return 0;
+}
+
+/****************** TCharSplitWithoutTags ******************/
+
+TCharSplitWithoutTags::TCharSplitWithoutTags(const char* str, size_t length)
+    : TSplitBase(str, length)
+{
+}
+
+TCharSplitWithoutTags::TCharSplitWithoutTags(const TString& s)
+    : TSplitBase(s)
+{
+}
+
+size_t TCharSplitWithoutTags::SkipTag(size_t pos) const {
+    Y_ASSERT('<' == Str[pos]);
+    while ((pos < Len) && ('>' != Str[pos]))
+        ++pos;
+    return pos + 1;
+}
+
+size_t TCharSplitWithoutTags::SkipDelimiters(size_t pos) const {
+    while (true) {
+        if (pos < Len) {
+            if ('<' != Str[pos])
+                break;
+            else
+                pos = SkipTag(pos);
+        } else
+            break;
+    }
+    return pos;
+}
+
+size_t TCharSplitWithoutTags::Begin() const {
+    size_t pos = 0;
+    pos = SkipDelimiters(pos);
+    return pos;
+}
+
+TSizeTRegion TCharSplitWithoutTags::Next(size_t& pos) const {
+    size_t begin = pos++;
+    TSizeTRegion result(begin, pos);
+
+    pos = SkipDelimiters(pos);
+
+    return result;
+}
+
+TCharSplitWithoutTags::TIterator TCharSplitWithoutTags::Iterator() const {
+    return TIterator(*this);
+}
+
+TSubstringSplitDelimiter::TSubstringSplitDelimiter(const TString& s)
+    : Matcher(s)
+    , Len(s.size())
+{
+}
+
+/****************** TSubstringSplit ******************/
+
+TSubstringSplit::TSubstringSplit(const char* str, size_t length, const TSubstringSplitDelimiter& delimiter)
+    : TSplitBase(str, length)
+    , Delimiter(delimiter)
+{
+}
+
+TSubstringSplit::TSubstringSplit(const TString& str, const TSubstringSplitDelimiter& delimiter)
+    : TSplitBase(str)
+    , Delimiter(delimiter)
+{
+}
+
+TSubstringSplit::TIterator TSubstringSplit::Iterator() const {
+    return TIterator(*this);
+}
+
+TSizeTRegion TSubstringSplit::Next(size_t& pos) const {
+    const char* begin = Str + pos;
+    const char* end = Str + Len;
+    const char* delim;
+    if (Delimiter.Matcher.SubStr(begin, end, delim)) {
+        TSizeTRegion result(pos, delim - begin + pos);
+        pos += delim - begin + Delimiter.Len;
+        return result;
+    } else {
+        TSizeTRegion result(pos, end - begin + pos);
+        pos += end - begin;
+        return result;
+    }
+}
+
+size_t TSubstringSplit::Begin() const {
+    return 0;
+}
diff --git a/library/cpp/deprecated/split/split_iterator.h b/library/cpp/deprecated/split/split_iterator.h
new file mode 100644
index 0000000000..0eacc29228
--- /dev/null
+++ b/library/cpp/deprecated/split/split_iterator.h
@@ -0,0 +1,317 @@
+#pragma once
+
+#include <library/cpp/deprecated/kmp/kmp.h>
+#include <util/string/cast.h>
+#include <util/string/util.h>
+#include <util/string/builder.h>
+
+#include <util/system/yassert.h>
+#include <util/system/defaults.h>
+#include <util/generic/strbuf.h>
+#include <util/generic/string.h>
+#include <util/generic/vector.h>
+#include <util/generic/yexception.h>
+
+#include <cstdio>
+
+template <typename T>
+struct TNumPair {
+    T Begin;
+    T End;
+
+    TNumPair() = default;
+
+    TNumPair(T begin, T end)
+        : Begin(begin)
+        , End(end)
+    {
+        Y_ASSERT(begin <= end);
+    }
+
+    T Length() const {
+        return End - Begin + 1;
+    }
+
+    bool operator==(const TNumPair& r) const {
+        return (Begin == r.Begin) && (End == r.End);
+    }
+
+    bool operator!=(const TNumPair& r) const {
+        return (Begin != r.Begin) || (End != r.End);
+    }
+};
+
+using TSizeTRegion = TNumPair<size_t>;
+using TUi32Region = TNumPair<ui32>;
+
+template <>
+inline TString ToString(const TUi32Region& r) {
+    return TStringBuilder() << "(" << r.Begin << ", " << r.End << ")";
+}
+
+template <>
+inline TUi32Region FromString(const TString& s) {
+    TUi32Region result;
+    sscanf(s.data(), "(%" PRIu32 ", %" PRIu32 ")", &result.Begin, &result.End);
+    return result;
+}
+
+class TSplitDelimiters {
+private:
+    bool Delims[256];
+
+public:
+    explicit TSplitDelimiters(const char* s);
+
+    Y_FORCE_INLINE bool IsDelimiter(ui8 ch) const {
+        return Delims[ch];
+    }
+};
+
+template <class Split>
+class TSplitIterator;
+
+class TSplitBase {
+protected:
+    const char* Str;
+    size_t Len;
+
+public:
+    TSplitBase(const char* str, size_t length);
+    TSplitBase(const TString& s);
+
+    Y_FORCE_INLINE const char* GetString() const {
+        return Str;
+    }
+
+    Y_FORCE_INLINE size_t GetLength() const {
+        return Len;
+    }
+
+private:
+    // we don't own Str, make sure that no one calls us with temporary object
+    TSplitBase(TString&&) = delete;
+};
+
+#ifdef _MSC_VER
+#pragma warning(push)
+#pragma warning(disable : 4512)
+#endif
+
+class TDelimitersSplit: public TSplitBase {
+private:
+    const TSplitDelimiters& Delimiters;
+
+public:
+    using TIterator = TSplitIterator<TDelimitersSplit>;
+    friend class TSplitIterator<TDelimitersSplit>;
+
+    TDelimitersSplit(const char* str, size_t length, const TSplitDelimiters& delimiters);
+    TDelimitersSplit(const TString& s, const TSplitDelimiters& delimiters);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Delimiters, make sure that no one calls us with temporary object
+    TDelimitersSplit(const char*, size_t, TSplitDelimiters&&) = delete;
+    TDelimitersSplit(const TString&, TSplitDelimiters&&) = delete;
+    TDelimitersSplit(TString&&, const TSplitDelimiters&) = delete;
+};
+
+class TDelimitersStrictSplit: public TSplitBase {
+private:
+    const TSplitDelimiters& Delimiters;
+
+public:
+    using TIterator = TSplitIterator<TDelimitersStrictSplit>;
+    friend class TSplitIterator<TDelimitersStrictSplit>;
+
+    TDelimitersStrictSplit(const char* str, size_t length, const TSplitDelimiters& delimiters);
+    TDelimitersStrictSplit(const TString& s, const TSplitDelimiters& delimiters);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Delimiters, make sure that no one calls us with temporary object
+    TDelimitersStrictSplit(const char*, size_t, TSplitDelimiters&&) = delete;
+    TDelimitersStrictSplit(const TString&, TSplitDelimiters&&) = delete;
+    TDelimitersStrictSplit(TString&&, const TSplitDelimiters&) = delete;
+};
+
+class TScreenedDelimitersSplit: public TSplitBase {
+private:
+    const TSplitDelimiters& Delimiters;
+    const TSplitDelimiters& Screens;
+
+public:
+    using TIterator = TSplitIterator<TScreenedDelimitersSplit>;
+    friend class TSplitIterator<TScreenedDelimitersSplit>;
+
+    TScreenedDelimitersSplit(const char*, size_t, const TSplitDelimiters& delimiters, const TSplitDelimiters& screens);
+    TScreenedDelimitersSplit(const TString& s, const TSplitDelimiters& delimiters, const TSplitDelimiters& screens);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Delimiters and Screens, make sure that no one calls us with temporary object
+    TScreenedDelimitersSplit(TString&&, const TSplitDelimiters&, const TSplitDelimiters&) = delete;
+    TScreenedDelimitersSplit(const TString&, TSplitDelimiters&&, const TSplitDelimiters&) = delete;
+    TScreenedDelimitersSplit(const TString&, const TSplitDelimiters&, TSplitDelimiters&&) = delete;
+};
+
+class TDelimitersSplitWithoutTags: public TSplitBase {
+private:
+    const TSplitDelimiters& Delimiters;
+    size_t SkipTag(size_t pos) const;
+    size_t SkipDelimiters(size_t pos) const;
+
+public:
+    using TIterator = TSplitIterator<TDelimitersSplitWithoutTags>;
+    friend class TSplitIterator<TDelimitersSplitWithoutTags>;
+
+    TDelimitersSplitWithoutTags(const char* str, size_t length, const TSplitDelimiters& delimiters);
+    TDelimitersSplitWithoutTags(const TString& s, const TSplitDelimiters& delimiters);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Delimiters, make sure that no one calls us with temporary object
+    TDelimitersSplitWithoutTags(const char*, size_t, TSplitDelimiters&&) = delete;
+    TDelimitersSplitWithoutTags(const TString&, TSplitDelimiters&&) = delete;
+    TDelimitersSplitWithoutTags(TString&&, const TSplitDelimiters&) = delete;
+};
+
+class TCharSplit: public TSplitBase {
+public:
+    using TIterator = TSplitIterator<TCharSplit>;
+    friend class TSplitIterator<TCharSplit>;
+
+    TCharSplit(const char* str, size_t length);
+    TCharSplit(const TString& s);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Str, make sure that no one calls us with temporary object
+    TCharSplit(TString&&) = delete;
+};
+
+#ifdef _MSC_VER
+#pragma warning(pop)
+#endif
+
+class TCharSplitWithoutTags: public TSplitBase {
+private:
+    size_t SkipTag(size_t pos) const;
+    size_t SkipDelimiters(size_t pos) const;
+
+public:
+    using TIterator = TSplitIterator<TCharSplitWithoutTags>;
+    friend class TSplitIterator<TCharSplitWithoutTags>;
+
+    TCharSplitWithoutTags(const char* str, size_t length);
+    TCharSplitWithoutTags(const TString& s);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Str, make sure that no one calls us with temporary object
+    TCharSplitWithoutTags(TString&&) = delete;
+};
+
+class TSubstringSplitDelimiter {
+public:
+    TKMPMatcher Matcher;
+    size_t Len;
+
+    TSubstringSplitDelimiter(const TString& s);
+};
+
+class TSubstringSplit: public TSplitBase {
+private:
+    const TSubstringSplitDelimiter& Delimiter;
+
+public:
+    using TIterator = TSplitIterator<TSubstringSplit>;
+    friend class TSplitIterator<TSubstringSplit>;
+
+    TSubstringSplit(const char* str, size_t length, const TSubstringSplitDelimiter& delimiter);
+    TSubstringSplit(const TString& str, const TSubstringSplitDelimiter& delimiter);
+    TIterator Iterator() const;
+    TSizeTRegion Next(size_t& pos) const;
+    size_t Begin() const;
+
+private:
+    // we don't own Delimiters, make sure that no one calls us with temporary object
+    TSubstringSplit(TString&&, const TSubstringSplitDelimiter&) = delete;
+    TSubstringSplit(const TString&, TSubstringSplitDelimiter&&) = delete;
+};
+
+template <class TSplit>
+class TSplitIterator {
+protected:
+    const TSplit& Split;
+    size_t Pos;
+    TString* CurrentStroka;
+
+public:
+    TSplitIterator(const TSplit& split)
+        : Split(split)
+        , Pos(Split.Begin())
+        , CurrentStroka(nullptr)
+    {
+    }
+
+    virtual ~TSplitIterator() {
+        delete CurrentStroka;
+    }
+
+    inline TSizeTRegion Next() {
+        Y_ENSURE(!Eof(), TStringBuf("eof reached"));
+        return Split.Next(Pos);
+    }
+
+    TStringBuf NextTok() {
+        if (Eof())
+            return TStringBuf();
+        TSizeTRegion region = Next();
+        return TStringBuf(Split.Str + region.Begin, region.End - region.Begin);
+    }
+
+    const TString& NextString() {
+        if (!CurrentStroka)
+            CurrentStroka = new TString();
+        TSizeTRegion region = Next();
+        CurrentStroka->assign(Split.Str, region.Begin, region.Length() - 1);
+        return *CurrentStroka;
+    }
+
+    inline bool Eof() const {
+        return Pos >= Split.Len;
+    }
+
+    TString GetTail() const {
+        return TString(Split.Str + Pos);
+    }
+
+    void Skip(size_t count) {
+        for (size_t i = 0; i < count; ++i)
+            Next();
+    }
+};
+
+using TSplitTokens = TVector<TString>;
+
+template <typename TSplit>
+void Split(const TSplit& split, TSplitTokens* words) {
+    words->clear();
+    TSplitIterator<TSplit> it(split);
+    while (!it.Eof())
+        words->push_back(it.NextString());
+}
diff --git a/library/cpp/deprecated/split/split_iterator_ut.cpp b/library/cpp/deprecated/split/split_iterator_ut.cpp
new file mode 100644
index 0000000000..be5069c4be
--- /dev/null
+++ b/library/cpp/deprecated/split/split_iterator_ut.cpp
@@ -0,0 +1,152 @@
+#include "split_iterator.h"
+
+#include <library/cpp/testing/unittest/registar.h>
+
+class TSplitIteratorTest: public TTestBase {
+    UNIT_TEST_SUITE(TSplitIteratorTest);
+    UNIT_TEST(TestDelimiters);
+    UNIT_TEST(TestDelimitersSplit);
+    UNIT_TEST(TestDelimitersStrictSplit);
+    UNIT_TEST(TestTail);
+    UNIT_TEST(TestScreenedDelimitersSplit);
+    UNIT_TEST(TestSubstringDelimiter);
+    UNIT_TEST_SUITE_END();
+
+public:
+    void TestDelimiters();
+    void TestDelimitersSplit();
+    void TestDelimitersStrictSplit();
+    void TestTail();
+    void TestScreenedDelimitersSplit();
+    void TestSubstringDelimiter();
+};
+
+void TSplitIteratorTest::TestDelimiters() {
+    TSplitDelimiters delims("@");
+    for (int i = 0; i < 256; ++i)
+        if ('@' != i) {
+            UNIT_ASSERT(!delims.IsDelimiter((ui8)i));
+        } else {
+            UNIT_ASSERT(delims.IsDelimiter((ui8)i));
+        }
+}
+
+void TSplitIteratorTest::TestDelimitersSplit() {
+    {
+        TString s = "1a3b45cd";
+        TSplitDelimiters delims("abcd");
+        TDelimitersSplit split(s, delims);
+        TSplitTokens tokens;
+        Split(split, &tokens);
+        TSplitTokens pattern = {"1", "3", "45"};
+        UNIT_ASSERT(tokens == pattern);
+    }
+    {
+        TString s = "aaaaaa";
+        TSplitDelimiters delims("abcd");
+        TDelimitersSplit split(s, delims);
+        TSplitTokens tokens;
+        Split(split, &tokens);
+        TSplitTokens pattern = {};
+        UNIT_ASSERT(tokens == pattern);
+    }
+}
+
+void TSplitIteratorTest::TestDelimitersStrictSplit() {
+    {
+        TString s = "grp@2";
+        TSplitDelimiters delims("@");
+        TDelimitersStrictSplit split(s, delims);
+        TSplitTokens tokens;
+        Split(split, &tokens);
+        TSplitTokens pattern = {"grp", "2"};
+        UNIT_ASSERT(tokens == pattern);
+    }
+
+    {
+        TString s = "@grp@2@@";
+        TSplitDelimiters delims("@");
+        TDelimitersStrictSplit split(s, delims);
+        TSplitTokens tokens;
+        Split(split, &tokens);
+        TSplitTokens pattern = {"", "grp", "2", ""};
+        UNIT_ASSERT(tokens == pattern);
+    }
+}
+
+void TSplitIteratorTest::TestTail() {
+    TString s = "grp@2@4";
+    TSplitDelimiters delims("@");
+    TDelimitersSplit split(s, delims);
+    TDelimitersSplit::TIterator it = split.Iterator();
+    UNIT_ASSERT_EQUAL(it.GetTail(), "grp@2@4");
+    it.Next();
+    UNIT_ASSERT_EQUAL(it.GetTail(), "2@4");
+    it.Next();
+    UNIT_ASSERT_EQUAL(it.GetTail(), "4");
+    it.Next();
+    UNIT_ASSERT_EQUAL(it.GetTail(), "");
+}
+
+void TSplitIteratorTest::TestScreenedDelimitersSplit() {
+    {
+        const TString s = "77.88.58.91 - - [28/Aug/2008:00:08:07 +0400] \"GET /export/mordashka.tgz HTTP/1.1\" 304 - \"-\" \"libwww-perl/5.805\" \"news.yandex.ru,80\" \"-\" \"-\" 1219867687 \"0\" 3283 2";
+        const TSplitDelimiters delims(" ");
+        const TSplitDelimiters screens("\"[]");
+        const TScreenedDelimitersSplit splitter(s, delims, screens);
+        TScreenedDelimitersSplit::TIterator it = splitter.Iterator();
+        UNIT_ASSERT_EQUAL(it.NextString(), "77.88.58.91");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "[28/Aug/2008:00:08:07 +0400]");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"GET /export/mordashka.tgz HTTP/1.1\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "304");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"libwww-perl/5.805\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"news.yandex.ru,80\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "1219867687");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"0\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "3283");
+        UNIT_ASSERT_EQUAL(it.NextString(), "2");
+    }
+    {
+        const TString s = "77.88.58.91 - - [28/Aug/2008:00:08:07 +0400] \"GET /export/mordashka.tgz HTTP/1.1\" 304 - \"-\" \"libwww-perl/5.805\" \"news.yandex.ru,80\" \"-\" \"-\" 1219867687 \"0\" 3283 2";
+        const TSplitDelimiters delims(" ");
+        const TSplitDelimiters screens("\"[]");
+        const TScreenedDelimitersSplit splitter(s.Data(), s.Size(), delims, screens);
+        TScreenedDelimitersSplit::TIterator it = splitter.Iterator();
+        UNIT_ASSERT_EQUAL(it.NextString(), "77.88.58.91");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "[28/Aug/2008:00:08:07 +0400]");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"GET /export/mordashka.tgz HTTP/1.1\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "304");
+        UNIT_ASSERT_EQUAL(it.NextString(), "-");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"libwww-perl/5.805\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"news.yandex.ru,80\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"-\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "1219867687");
+        UNIT_ASSERT_EQUAL(it.NextString(), "\"0\"");
+        UNIT_ASSERT_EQUAL(it.NextString(), "3283");
+        UNIT_ASSERT_EQUAL(it.NextString(), "2");
+    }
+}
+
+void TSplitIteratorTest::TestSubstringDelimiter() {
+    const TString s = "a@@bb@@cc@c.d@@r";
+    static const TSubstringSplitDelimiter delimiter("@@");
+    const TSubstringSplit splitter(s, delimiter);
+    TSubstringSplit::TIterator it = splitter.Iterator();
+    UNIT_ASSERT_EQUAL(it.NextString(), "a");
+    UNIT_ASSERT_EQUAL(it.NextString(), "bb");
+    UNIT_ASSERT_EQUAL(it.NextString(), "cc@c.d");
+    UNIT_ASSERT_EQUAL(it.NextString(), "r");
+    UNIT_ASSERT(it.Eof());
+}
+
+UNIT_TEST_SUITE_REGISTRATION(TSplitIteratorTest);
diff --git a/library/cpp/deprecated/split/ya.make b/library/cpp/deprecated/split/ya.make
new file mode 100644
index 0000000000..946e685ac8
--- /dev/null
+++ b/library/cpp/deprecated/split/ya.make
@@ -0,0 +1,14 @@
+LIBRARY()
+
+OWNER(wrg0ababd)
+
+SRCS(
+    delim_string_iter.cpp
+    split_iterator.cpp
+)
+
+PEERDIR(
+    library/cpp/deprecated/kmp
+)
+
+END()