YQL-16239 Move purecalc to public

9 files changed, 349 insertions, 0 deletions

diff --git a/library/cpp/containers/CMakeLists.txt b/library/cpp/containers/CMakeLists.txt
index 43fcbe83466..40f50138673 100644
--- a/library/cpp/containers/CMakeLists.txt
+++ b/library/cpp/containers/CMakeLists.txt
@@ -20,5 +20,6 @@ add_subdirectory(ring_buffer)
 add_subdirectory(sorted_vector)
 add_subdirectory(stack_array)
 add_subdirectory(stack_vector)
+add_subdirectory(str_hash)
 add_subdirectory(str_map)
 add_subdirectory(top_keeper)
diff --git a/library/cpp/containers/str_hash/CMakeLists.darwin-x86_64.txt b/library/cpp/containers/str_hash/CMakeLists.darwin-x86_64.txt
new file mode 100644
index 00000000000..627814f0ed6
--- /dev/null
+++ b/library/cpp/containers/str_hash/CMakeLists.darwin-x86_64.txt
@@ -0,0 +1,19 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-containers-str_hash)
+target_link_libraries(cpp-containers-str_hash PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-charset
+  cpp-containers-str_map
+)
+target_sources(cpp-containers-str_hash PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/containers/str_hash/str_hash.cpp
+)
diff --git a/library/cpp/containers/str_hash/CMakeLists.linux-aarch64.txt b/library/cpp/containers/str_hash/CMakeLists.linux-aarch64.txt
new file mode 100644
index 00000000000..cd723cbea23
--- /dev/null
+++ b/library/cpp/containers/str_hash/CMakeLists.linux-aarch64.txt
@@ -0,0 +1,20 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-containers-str_hash)
+target_link_libraries(cpp-containers-str_hash PUBLIC
+  contrib-libs-linux-headers
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-charset
+  cpp-containers-str_map
+)
+target_sources(cpp-containers-str_hash PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/containers/str_hash/str_hash.cpp
+)
diff --git a/library/cpp/containers/str_hash/CMakeLists.linux-x86_64.txt b/library/cpp/containers/str_hash/CMakeLists.linux-x86_64.txt
new file mode 100644
index 00000000000..cd723cbea23
--- /dev/null
+++ b/library/cpp/containers/str_hash/CMakeLists.linux-x86_64.txt
@@ -0,0 +1,20 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-containers-str_hash)
+target_link_libraries(cpp-containers-str_hash PUBLIC
+  contrib-libs-linux-headers
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-charset
+  cpp-containers-str_map
+)
+target_sources(cpp-containers-str_hash PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/containers/str_hash/str_hash.cpp
+)
diff --git a/library/cpp/containers/str_hash/CMakeLists.txt b/library/cpp/containers/str_hash/CMakeLists.txt
new file mode 100644
index 00000000000..f8b31df0c11
--- /dev/null
+++ b/library/cpp/containers/str_hash/CMakeLists.txt
@@ -0,0 +1,17 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+if (CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64" AND NOT HAVE_CUDA)
+  include(CMakeLists.linux-aarch64.txt)
+elseif (CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
+  include(CMakeLists.darwin-x86_64.txt)
+elseif (WIN32 AND CMAKE_SYSTEM_PROCESSOR STREQUAL "AMD64" AND NOT HAVE_CUDA)
+  include(CMakeLists.windows-x86_64.txt)
+elseif (CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64" AND NOT HAVE_CUDA)
+  include(CMakeLists.linux-x86_64.txt)
+endif()
diff --git a/library/cpp/containers/str_hash/CMakeLists.windows-x86_64.txt b/library/cpp/containers/str_hash/CMakeLists.windows-x86_64.txt
new file mode 100644
index 00000000000..627814f0ed6
--- /dev/null
+++ b/library/cpp/containers/str_hash/CMakeLists.windows-x86_64.txt
@@ -0,0 +1,19 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-containers-str_hash)
+target_link_libraries(cpp-containers-str_hash PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-charset
+  cpp-containers-str_map
+)
+target_sources(cpp-containers-str_hash PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/containers/str_hash/str_hash.cpp
+)
diff --git a/library/cpp/containers/str_hash/str_hash.cpp b/library/cpp/containers/str_hash/str_hash.cpp
new file mode 100644
index 00000000000..12986385334
--- /dev/null
+++ b/library/cpp/containers/str_hash/str_hash.cpp
@@ -0,0 +1,60 @@
+#include "str_hash.h"
+
+#include <library/cpp/charset/ci_string.h>
+#include <util/stream/output.h>
+#include <util/stream/input.h>
+
+HashSet::HashSet(const char** array, size_type size) {
+    Resize(size);
+    while (*array && **array)
+        AddPermanent(*array++);
+}
+
+void HashSet::Read(IInputStream* input) {
+    TString s;
+
+    while (input->ReadLine(s)) {
+        AddUniq(TCiString(s).c_str());
+    }
+}
+
+void HashSet::Write(IOutputStream* output) const {
+    for (const auto& it : *this) {
+        *output << it.first << "\n";
+    }
+}
+
+#ifdef TEST_STRHASH
+#include <ctime>
+#include <fstream>
+#include <cstdio>
+#include <cstdlib>
+
+using namespace std;
+
+int main(int argc, char* argv[]) {
+    if (argc < 2) {
+        printf("usage: stoplist <stop-words file ...\n");
+        exit(EXIT_FAILURE); // FreeBSD: EX_USAGE
+    }
+    Hash hash;
+    hash.Read(cin);
+    for (--argc, ++argv; argc > 0; --argc, ++argv) {
+        ifstream input(argv[0]);
+        if (!input.good()) {
+            perror(argv[0]);
+            continue;
+        }
+        TCiString s;
+        while (input >> s) {
+            if (!hash.Has(s))
+                cout << s << "\n";
+            else
+                cout << "[[" << s << "]]"
+                     << "\n";
+        }
+    }
+    return EXIT_SUCCESS; // EX_OK
+}
+
+#endif
diff --git a/library/cpp/containers/str_hash/str_hash.h b/library/cpp/containers/str_hash/str_hash.h
new file mode 100644
index 00000000000..25f960dbb5f
--- /dev/null
+++ b/library/cpp/containers/str_hash/str_hash.h
@@ -0,0 +1,181 @@
+#pragma once
+
+#include <library/cpp/containers/str_map/str_map.h>
+#include <library/cpp/charset/ci_string.h>
+#include <util/system/yassert.h>
+#include <util/memory/tempbuf.h>
+
+#include <memory>
+
+class IInputStream;
+class IOutputStream;
+
+template <class T, class Alloc = std::allocator<const char*>>
+class Hash;
+
+struct yvoid {
+    yvoid() = default;
+};
+
+template <typename T, class Alloc>
+class Hash: public string_hash<T, ci_hash, ci_equal_to, Alloc> {
+    using ci_string_hash = string_hash<T, ci_hash, ci_equal_to, Alloc>;
+
+protected:
+    using ci_string_hash::pool;
+
+public:
+    using size_type = typename ci_string_hash::size_type;
+    using const_iterator = typename ci_string_hash::const_iterator;
+    using iterator = typename ci_string_hash::iterator;
+    using value_type = typename ci_string_hash::value_type;
+    using ci_string_hash::begin;
+    using ci_string_hash::end;
+    using ci_string_hash::find;
+    using ci_string_hash::size;
+
+    Hash()
+        : ci_string_hash()
+    {
+    }
+    explicit Hash(size_type theSize)
+        : ci_string_hash(theSize, theSize * AVERAGEWORD_BUF)
+    {
+    }
+    Hash(const char** strings, size_type size = 0, T* = 0); // must end with NULL or "\0"
+    virtual ~Hash();
+    bool Has(const char* s, size_t len, T* pp = nullptr) const;
+    bool Has(const char* s, T* pp = nullptr) const {
+        const_iterator it;
+        if ((it = find(s)) == end())
+            return false;
+        else if (pp)
+            *pp = (*it).second;
+        return true;
+    }
+    void Add(const char* s, T data) {
+        // in fact it is the same insert_unique as in AddUnique.
+        // it's impossible to have _FAST_ version of insert() in 'hash_map'
+
+        // you have to use 'hash_mmap' to get the _kind_ of desired effect.
+        // BUT still there will be "Checks" inside -
+        // to make the same keys close to each other (see insert_equal())
+        this->insert_copy(s, data);
+    }
+    bool AddUniq(const char* s, T data) {
+        return this->insert_copy(s, data).second;
+    }
+    // new function to get rid of allocations completely! -- e.g. in constructors
+    void AddPermanent(const char* s, T data) {
+        this->insert(value_type(s, data));
+    }
+    T Detach(const char* s) {
+        iterator it = find(s);
+        if (it == end())
+            return T();
+        T data = (*it).second;
+        this->erase(it);
+        return data;
+    }
+    size_type NumEntries() const {
+        return size();
+    }
+    bool ForEach(bool (*func)(const char* key, T data, void* cookie), void* cookie = nullptr);
+    void Resize(size_type theSize) {
+        this->reserve(theSize);
+        // no pool resizing here.
+    }
+    virtual void Clear();
+    char* Pool() {
+        if (pool.Size() < 2 || pool.End()[-2] != '\0')
+            pool.Append("\0", 1);
+        return pool.Begin();
+    }
+};
+
+template <class T, class Alloc>
+Hash<T, Alloc>::Hash(const char** array, size_type theSize, T* data) {
+    // must end with NULL or "\0"
+    Y_ASSERT(data != nullptr);
+    Resize(theSize);
+    while (*array && **array)
+        AddPermanent(*array++, *data++);
+}
+
+template <class T, class Alloc>
+bool Hash<T, Alloc>::Has(const char* s, size_t len, T* pp) const {
+    TTempArray<char> buf(len + 1);
+    char* const allocated = buf.Data();
+    memcpy(allocated, s, len);
+    allocated[len] = '\x00';
+    return Has(allocated, pp);
+}
+
+template <class T, class Alloc>
+Hash<T, Alloc>::~Hash() {
+    Clear();
+}
+
+template <class T, class Alloc>
+void Hash<T, Alloc>::Clear() {
+    ci_string_hash::clear_hash(); // to make the key pool empty
+}
+
+template <class T, class Alloc>
+bool Hash<T, Alloc>::ForEach(bool (*func)(const char* key, T data, void* cookie), void* cookie) {
+    for (const_iterator it = begin(); it != end(); ++it)
+        if (!func((*it).first, (*it).second, cookie))
+            return false;
+    return true;
+}
+
+class HashSet: public Hash<yvoid> {
+public:
+    HashSet(const char** array, size_type size = 0);
+    HashSet()
+        : Hash<yvoid>()
+    {
+    }
+    void Read(IInputStream* input);
+    void Write(IOutputStream* output) const;
+    void Add(const char* s) {
+        // in fact it is the same insert_unique as in AddUnique.
+        // it's impossible to have _FAST_ version of insert() in 'hash_map'
+
+        // you have to use 'hash_mmap' to get the _kind_ of desired effect.
+        // BUT still there will be "Checks" inside -
+        // to make the same keys close to each other (see insert_equal())
+        insert_copy(s, yvoid());
+    }
+    bool AddUniq(const char* s) {
+        return insert_copy(s, yvoid()).second;
+    }
+    // new function to get rid of allocations completely! -- e.g. in constructors
+    void AddPermanent(const char* s) {
+        insert(value_type(s, yvoid()));
+    }
+};
+
+template <class T, class HashFcn = THash<T>, class EqualKey = TEqualTo<T>, class Alloc = std::allocator<T>>
+class TStaticHash: private THashMap<T, T, HashFcn, EqualKey> {
+private:
+    using TBase = THashMap<T, T, HashFcn, EqualKey>;
+
+public:
+    TStaticHash(T arr[][2], size_t size) {
+        TBase::reserve(size);
+        while (size) {
+            TBase::insert(typename TBase::value_type(arr[0][0], arr[0][1]));
+            arr++;
+            size--;
+        }
+    }
+    T operator[](const T& key) const { // !!! it is not lvalue nor it used to be
+        typename TBase::const_iterator it = TBase::find(key);
+        if (it == TBase::end())
+            return nullptr;
+        return it->second;
+    }
+};
+
+using TStHash = TStaticHash<const char*, ci_hash, ci_equal_to>;
diff --git a/library/cpp/containers/str_hash/ya.make b/library/cpp/containers/str_hash/ya.make
new file mode 100644
index 00000000000..f7e24316b98
--- /dev/null
+++ b/library/cpp/containers/str_hash/ya.make
@@ -0,0 +1,12 @@
+LIBRARY()
+
+PEERDIR(
+    library/cpp/charset
+    library/cpp/containers/str_map
+)
+
+SRCS(
+    str_hash.cpp
+)
+
+END()