catboost 1.25.1 → 1.27.0

package/inc/util/string/strip.h

@@ -1,257 +0,0 @@
-#pragma once
-
-#include "ascii.h"
-
-#include <util/generic/string.h>
-#include <util/generic/strbuf.h>
-#include <utility>
-
-template <class It>
-struct TIsAsciiSpaceAdapter {
-    bool operator()(const It& it) const noexcept {
-        return IsAsciiSpace(*it);
-    }
-};
-
-template <class It>
-TIsAsciiSpaceAdapter<It> IsAsciiSpaceAdapter(It) {
-    return {};
-}
-
-template <class TChar>
-struct TEqualsStripAdapter {
-    TEqualsStripAdapter(TChar ch)
-        : Ch(ch)
-    {
-    }
-
-    template <class It>
-    bool operator()(const It& it) const noexcept {
-        return *it == Ch;
-    }
-
-    const TChar Ch;
-};
-
-template <class TChar>
-TEqualsStripAdapter<TChar> EqualsStripAdapter(TChar ch) {
-    return {ch};
-}
-
-template <class It, class TStripCriterion>
-inline void StripRangeBegin(It& b, const It& e, TStripCriterion&& criterion) noexcept {
-    while (b < e && criterion(b)) {
-        ++b;
-    }
-}
-
-template <class It>
-inline void StripRangeBegin(It& b, const It& e) noexcept {
-    StripRangeBegin(b, e, IsAsciiSpaceAdapter(b));
-}
-
-template <class It, class TStripCriterion>
-inline void StripRangeEnd(const It& b, It& e, TStripCriterion&& criterion) noexcept {
-    while (b < e && criterion(e - 1)) {
-        --e;
-    }
-}
-
-template <class It>
-inline void StripRangeEnd(const It& b, It& e) noexcept {
-    StripRangeEnd(b, e, IsAsciiSpaceAdapter(b));
-}
-
-template <bool stripBeg, bool stripEnd>
-struct TStripImpl {
-    template <class It, class TStripCriterion>
-    static inline bool StripRange(It& b, It& e, TStripCriterion&& criterion) noexcept {
-        const size_t oldLen = e - b;
-
-        if (stripBeg) {
-            StripRangeBegin(b, e, std::forward<TStripCriterion>(criterion));
-        }
-
-        if (stripEnd) {
-            StripRangeEnd(b, e, std::forward<TStripCriterion>(criterion));
-        }
-
-        const size_t newLen = e - b;
-        return newLen != oldLen;
-    }
-
-    template <class T, class TStripCriterion>
-    static inline bool StripString(const T& from, T& to, TStripCriterion&& criterion) {
-        auto b = from.begin();
-        auto e = from.end();
-
-        if (StripRange(b, e, std::forward<TStripCriterion>(criterion))) {
-            to = T(b, e - b);
-
-            return true;
-        }
-
-        to = from;
-
-        return false;
-    }
-
-    template <class T, class TStripCriterion>
-    static inline T StripString(const T& from, TStripCriterion&& criterion) {
-        T ret;
-        StripString(from, ret, std::forward<TStripCriterion>(criterion));
-        return ret;
-    }
-
-    template <class T>
-    static inline T StripString(const T& from) {
-        return StripString(from, IsAsciiSpaceAdapter(from.begin()));
-    }
-};
-
-template <class It, class TStripCriterion>
-inline bool StripRange(It& b, It& e, TStripCriterion&& criterion) noexcept {
-    return TStripImpl<true, true>::StripRange(b, e, std::forward<TStripCriterion>(criterion));
-}
-
-template <class It>
-inline bool StripRange(It& b, It& e) noexcept {
-    return StripRange(b, e, IsAsciiSpaceAdapter(b));
-}
-
-template <class It, class TStripCriterion>
-inline bool Strip(It& b, size_t& len, TStripCriterion&& criterion) noexcept {
-    It e = b + len;
-
-    if (StripRange(b, e, std::forward<TStripCriterion>(criterion))) {
-        len = e - b;
-
-        return true;
-    }
-
-    return false;
-}
-
-template <class It>
-inline bool Strip(It& b, size_t& len) noexcept {
-    return Strip(b, len, IsAsciiSpaceAdapter(b));
-}
-
-template <class T, class TStripCriterion>
-static inline bool StripString(const T& from, T& to, TStripCriterion&& criterion) {
-    return TStripImpl<true, true>::StripString(from, to, std::forward<TStripCriterion>(criterion));
-}
-
-template <class T>
-static inline bool StripString(const T& from, T& to) {
-    return StripString(from, to, IsAsciiSpaceAdapter(from.begin()));
-}
-
-template <class T, class TStripCriterion>
-static inline T StripString(const T& from, TStripCriterion&& criterion) {
-    return TStripImpl<true, true>::StripString(from, std::forward<TStripCriterion>(criterion));
-}
-
-template <class T>
-static inline T StripString(const T& from) {
-    return TStripImpl<true, true>::StripString(from);
-}
-
-template <class T>
-static inline T StripStringLeft(const T& from) {
-    return TStripImpl<true, false>::StripString(from);
-}
-
-template <class T>
-static inline T StripStringRight(const T& from) {
-    return TStripImpl<false, true>::StripString(from);
-}
-
-template <class T, class TStripCriterion>
-static inline T StripStringLeft(const T& from, TStripCriterion&& criterion) {
-    return TStripImpl<true, false>::StripString(from, std::forward<TStripCriterion>(criterion));
-}
-
-template <class T, class TStripCriterion>
-static inline T StripStringRight(const T& from, TStripCriterion&& criterion) {
-    return TStripImpl<false, true>::StripString(from, std::forward<TStripCriterion>(criterion));
-}
-
-/// Copies the given string removing leading and trailing spaces.
-static inline bool Strip(const TString& from, TString& to) {
-    return StripString(from, to);
-}
-
-/// Removes leading and trailing spaces from the string.
-inline TString& StripInPlace(TString& s) {
-    Strip(s, s);
-    return s;
-}
-
-/// Returns a copy of the given string with removed leading and trailing spaces.
-inline TString Strip(const TString& s) Y_WARN_UNUSED_RESULT;
-inline TString Strip(const TString& s) {
-    TString ret = s;
-    Strip(ret, ret);
-    return ret;
-}
-
-template <class TChar, class TWhitespaceFunc>
-size_t CollapseImpl(TChar* s, size_t n, const TWhitespaceFunc& isWhitespace) {
-    size_t newLen = 0;
-    for (size_t i = 0; i < n; ++i, ++newLen) {
-        size_t nextNonSpace = i;
-        while (nextNonSpace < n && isWhitespace(s[nextNonSpace])) {
-            ++nextNonSpace;
-        }
-        size_t numSpaces = nextNonSpace - i;
-        if (numSpaces > 1 || (numSpaces == 1 && s[i] != ' ')) {
-            s[newLen] = ' ';
-            i = nextNonSpace - 1;
-        } else {
-            s[newLen] = s[i];
-        }
-    }
-    return newLen;
-}
-
-template <class TStringType, class TWhitespaceFunc>
-bool CollapseImpl(const TStringType& from, TStringType& to, size_t maxLen, const TWhitespaceFunc& isWhitespace) {
-    to = from;
-    maxLen = maxLen ? Min(maxLen, to.size()) : to.size();
-    for (size_t i = 0; i < maxLen; ++i) {
-        if (isWhitespace(to[i]) && (to[i] != ' ' || isWhitespace(to[i + 1]))) {
-            size_t tailSize = maxLen - i;
-            size_t newTailSize = CollapseImpl(to.begin() + i, tailSize, isWhitespace);
-            to.remove(i + newTailSize, tailSize - newTailSize);
-            return true;
-        }
-    }
-    return false;
-}
-
-bool Collapse(const TString& from, TString& to, size_t maxLen = 0);
-
-/// Replaces several consequtive space symbols with one (processing is limited to maxLen bytes)
-inline TString& CollapseInPlace(TString& s, size_t maxLen = 0) {
-    Collapse(s, s, maxLen);
-    return s;
-}
-
-/// Replaces several consequtive space symbols with one (processing is limited to maxLen bytes)
-inline TString Collapse(const TString& s, size_t maxLen = 0) Y_WARN_UNUSED_RESULT;
-inline TString Collapse(const TString& s, size_t maxLen) {
-    TString ret;
-    Collapse(s, ret, maxLen);
-    return ret;
-}
-
-void CollapseText(const TString& from, TString& to, size_t maxLen);
-
-/// The same as Collapse() + truncates the string to maxLen.
-/// @details An ellipsis is inserted at the end of the truncated line.
-inline void CollapseText(TString& s, size_t maxLen) {
-    TString to;
-    CollapseText(s, to, maxLen);
-    s = to;
-}

package/inc/util/string/strspn.h

@@ -1,65 +0,0 @@
-#pragma once
-
-#include "cstriter.h"
-
-#include <util/generic/bitmap.h>
-
-template <class TSetType>
-class TStrSpnImpl {
-public:
-    inline TStrSpnImpl(const char* b, const char* e) {
-        Init(b, e);
-    }
-
-    inline TStrSpnImpl(const char* s) {
-        Init(s, TCStringEndIterator());
-    }
-
-    //FirstOf
-    template <class It>
-    inline It FindFirstOf(It b, const char* e) const noexcept {
-        return FindFirst<false>(b, e);
-    }
-
-    template <class It>
-    inline It FindFirstOf(It s) const noexcept {
-        return FindFirst<false>(s, TCStringEndIterator());
-    }
-
-    //FirstNotOf
-    template <class It>
-    inline It FindFirstNotOf(It b, const char* e) const noexcept {
-        return FindFirst<true>(b, e);
-    }
-
-    template <class It>
-    inline It FindFirstNotOf(It s) const noexcept {
-        return FindFirst<true>(s, TCStringEndIterator());
-    }
-
-    inline void Set(ui8 b) noexcept {
-        S_.Set(b);
-    }
-
-private:
-    template <bool Result, class It1, class It2>
-    inline It1 FindFirst(It1 b, It2 e) const noexcept {
-        while (b != e && (S_.Get((ui8)*b) == Result)) {
-            ++b;
-        }
-
-        return b;
-    }
-
-    template <class It1, class It2>
-    inline void Init(It1 b, It2 e) {
-        while (b != e) {
-            this->Set((ui8)*b++);
-        }
-    }
-
-private:
-    TSetType S_;
-};
-
-using TCompactStrSpn = TStrSpnImpl<TBitMap<256>>;

package/inc/util/string/subst.h

@@ -1,56 +0,0 @@
-#pragma once
-
-#include <util/generic/fwd.h>
-
-#include <stlfwd>
-
-/* Replace all occurences of substring `what` with string `with` starting from position `from`.
- *
- * @param text      String to modify.
- * @param what      Substring to replace.
- * @param with      Substring to use as replacement.
- * @param from      Position at with to start replacement.
- *
- * @return          Number of replacements occured.
- */
-size_t SubstGlobal(TString& text, TStringBuf what, TStringBuf with, size_t from = 0);
-size_t SubstGlobal(std::string& text, TStringBuf what, TStringBuf with, size_t from = 0);
-size_t SubstGlobal(TUtf16String& text, TWtringBuf what, TWtringBuf with, size_t from = 0);
-size_t SubstGlobal(std::u16string& text, TWtringBuf what, TWtringBuf with, size_t from = 0);
-size_t SubstGlobal(TUtf32String& text, TUtf32StringBuf what, TUtf32StringBuf with, size_t from = 0);
-
-/* Replace all occurences of character `what` with character `with` starting from position `from`.
- *
- * @param text      String to modify.
- * @param what      Character to replace.
- * @param with      Character to use as replacement.
- * @param from      Position at with to start replacement.
- *
- * @return          Number of replacements occured.
- */
-size_t SubstGlobal(TString& text, char what, char with, size_t from = 0);
-size_t SubstGlobal(std::string& text, char what, char with, size_t from = 0);
-size_t SubstGlobal(TUtf16String& text, wchar16 what, wchar16 with, size_t from = 0);
-size_t SubstGlobal(std::u16string& text, wchar16 what, wchar16 with, size_t from = 0);
-size_t SubstGlobal(TUtf32String& text, wchar32 what, wchar32 with, size_t from = 0);
-
-// TODO(yazevnul):
-// - rename `SubstGlobal` to `ReplaceAll` for convenience
-// - add `SubstGlobalCopy(TStringBuf)` for convenience
-// - add `RemoveAll(text, what, from)` as a shortcut for `SubstGlobal(text, what, "", from)`
-// - rename file to `replace.h`
-
-/* Replace all occurences of substring or character `what` with string or character `with` starting from position `from`, and return result string.
- *
- * @param text      String to modify.
- * @param what      Substring/character to replace.
- * @param with      Substring/character to use as replacement.
- * @param from      Position at with to start replacement.
- *
- * @return          Result string
- */
-template <class TStringType, class TPatternType>
-Y_WARN_UNUSED_RESULT TStringType SubstGlobalCopy(TStringType result, TPatternType what, TPatternType with, size_t from = 0) {
-    SubstGlobal(result, what, with, from);
-    return result;
-}

package/inc/util/string/type.h

@@ -1,50 +0,0 @@
-#pragma once
-
-#include <util/generic/strbuf.h>
-
-Y_PURE_FUNCTION
-bool IsSpace(const char* s, size_t len) noexcept;
-
-/// Checks if a string is a set of only space symbols.
-Y_PURE_FUNCTION
-static inline bool IsSpace(const TStringBuf s) noexcept {
-    return IsSpace(s.data(), s.size());
-}
-
-/// Returns "true" if the given string is an arabic number ([0-9]+)
-Y_PURE_FUNCTION
-bool IsNumber(const TStringBuf s) noexcept;
-
-Y_PURE_FUNCTION
-bool IsNumber(const TWtringBuf s) noexcept;
-
-/// Returns "true" if the given string is a hex number ([0-9a-fA-F]+)
-Y_PURE_FUNCTION
-bool IsHexNumber(const TStringBuf s) noexcept;
-
-Y_PURE_FUNCTION
-bool IsHexNumber(const TWtringBuf s) noexcept;
-
-/* Tests if the given string is case insensitive equal to one of:
- * - "true",
- * - "t",
- * - "yes",
- * - "y",
- * - "on",
- * - "1",
- * - "da".
- */
-Y_PURE_FUNCTION
-bool IsTrue(const TStringBuf value) noexcept;
-
-/* Tests if the given string is case insensitive equal to one of:
- *  - "false",
- *  - "f",
- *  - "no",
- *  - "n",
- *  - "off",
- *  - "0",
- *  - "net".
- */
-Y_PURE_FUNCTION
-bool IsFalse(const TStringBuf value) noexcept;

package/inc/util/string/util.h

@@ -1,195 +0,0 @@
-#pragma once
-
-//THIS FILE A COMPAT STUB HEADER
-
-#include <cstring>
-#include <cstdarg>
-#include <algorithm>
-
-#include <util/system/defaults.h>
-#include <util/generic/string.h>
-#include <util/generic/strbuf.h>
-
-/// @addtogroup Strings_Miscellaneous
-/// @{
-int a2i(const TString& s);
-
-/// Removes the last character if it is equal to c.
-template <class T>
-inline void RemoveIfLast(T& s, int c) {
-    const size_t length = s.length();
-    if (length && s[length - 1] == c)
-        s.remove(length - 1);
-}
-
-/// Adds lastCh symbol to the the of the string if it is not already there.
-inline void addIfNotLast(TString& s, int lastCh) {
-    size_t len = s.length();
-    if (!len || s[len - 1] != lastCh) {
-        s.append(char(lastCh));
-    }
-}
-
-/// @details Finishes the string with lastCh1 if lastCh2 is not present in the string and lastCh1 is not already at the end of the string.
-/// Else, if lastCh2 is not equal to the symbol before the last, it finishes the string with lastCh2.
-/// @todo ?? Define, when to apply the function. Is in use several times for URLs parsing.
-inline void addIfAbsent(TString& s, char lastCh1, char lastCh2) {
-    size_t pos = s.find(lastCh2);
-    if (pos == TString::npos) {
-        //s.append((char)lastCh1);
-        addIfNotLast(s, lastCh1);
-    } else if (pos < s.length() - 1) {
-        addIfNotLast(s, lastCh2);
-    }
-}
-
-/// @}
-
-/*
- * ------------------------------------------------------------------
- *
- * A fast implementation of glibc's functions;
- *      strspn, strcspn and strpbrk.
- *
- * ------------------------------------------------------------------
- */
-struct ui8_256 {
-    // forward chars table
-    ui8 chars_table[256];
-    // reverse (for c* functions) chars table
-    ui8 c_chars_table[256];
-};
-
-class str_spn: public ui8_256 {
-public:
-    explicit str_spn(const char* charset, bool extended = false) {
-        // exteneded: if true, treat charset string more like
-        // interior of brackets [ ], e.g. "a-z0-9"
-        init(charset, extended);
-    }
-
-    /// Return first character in table, like strpbrk()
-    /// That is, skip all characters not in table
-    /// [DIFFERENCE FOR NOT_FOUND CASE: Returns end of string, not NULL]
-    const char* brk(const char* s) const {
-        while (c_chars_table[(ui8)*s])
-            ++s;
-        return s;
-    }
-
-    const char* brk(const char* s, const char* e) const {
-        while (s < e && c_chars_table[(ui8)*s])
-            ++s;
-        return s;
-    }
-
-    /// Return first character not in table, like strpbrk() for inverted table.
-    /// That is, skip all characters in table
-    const char* cbrk(const char* s) const {
-        while (chars_table[(ui8)*s])
-            ++s;
-        return s;
-    }
-
-    const char* cbrk(const char* s, const char* e) const {
-        while (s < e && chars_table[(ui8)*s])
-            ++s;
-        return s;
-    }
-
-    /// Offset of the first character not in table, like strspn().
-    size_t spn(const char* s) const {
-        return cbrk(s) - s;
-    }
-
-    size_t spn(const char* s, const char* e) const {
-        return cbrk(s, e) - s;
-    }
-
-    /// Offset of the first character in table, like strcspn().
-    size_t cspn(const char* s) const {
-        return brk(s) - s;
-    }
-
-    size_t cspn(const char* s, const char* e) const {
-        return brk(s, e) - s;
-    }
-
-    char* brk(char* s) const {
-        return const_cast<char*>(brk((const char*)s));
-    }
-
-    char* cbrk(char* s) const {
-        return const_cast<char*>(cbrk((const char*)s));
-    }
-
-    /// See strsep [BUT argument is *&, not **]
-    char* sep(char*& s) const {
-        char sep_char; // unused;
-        return sep(s, sep_char);
-    }
-
-    /// strsep + remember character that was destroyed
-    char* sep(char*& s, char& sep_char) const {
-        if (!s)
-            return nullptr;
-        char* ret = s;
-        char* next = brk(ret);
-        if (*next) {
-            sep_char = *next;
-            *next = 0;
-            s = next + 1;
-        } else {
-            sep_char = 0;
-            s = nullptr;
-        }
-        return ret;
-    }
-
-protected:
-    void init(const char* charset, bool extended);
-    str_spn() = default;
-};
-
-// an analogue of tr/$from/$to/
-class Tr {
-public:
-    Tr(const char* from, const char* to);
-
-    char ConvertChar(char ch) const {
-        return Map[(ui8)ch];
-    }
-
-    void Do(char* s) const {
-        for (; *s; s++)
-            *s = ConvertChar(*s);
-    }
-    void Do(const char* src, char* dst) const {
-        for (; *src; src++)
-            *dst++ = ConvertChar(*src);
-        *dst = 0;
-    }
-    void Do(char* s, size_t l) const {
-        for (size_t i = 0; i < l && s[i]; i++)
-            s[i] = ConvertChar(s[i]);
-    }
-    void Do(TString& str) const;
-
-private:
-    char Map[256];
-
-    size_t FindFirstChangePosition(const TString& str) const;
-};
-
-// Removes all occurrences of given character from string
-template <typename TStringType>
-void RemoveAll(TStringType& str, typename TStringType::char_type ch) {
-    size_t pos = str.find(ch); // 'find' to avoid cloning of string in 'TString.begin()'
-    if (pos == TStringType::npos)
-        return;
-
-    typename TStringType::iterator begin = str.begin();
-    typename TStringType::iterator end = begin + str.length();
-    typename TStringType::iterator it = std::remove(begin + pos, end, ch);
-    str.erase(it, end);
-}