uset.h

Go to the documentation of this file.

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
*   Copyright (C) 2002-2014, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
*******************************************************************************
*   file name:  uset.h
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2002mar07
*   created by: Markus W. Scherer
*
*   C version of UnicodeSet.
*/
 
 
#ifndef __USET_H__
#define __USET_H__
 
#include "unicode/utypes.h"
#include "unicode/uchar.h"
 
#if U_SHOW_CPLUSPLUS_API
#include <string_view>
#include "unicode/char16ptr.h"
#include "unicode/localpointer.h"
#include "unicode/unistr.h"
#endif   // U_SHOW_CPLUSPLUS_API
 
#ifndef USET_DEFINED
 
#ifndef U_IN_DOXYGEN
#define USET_DEFINED
#endif
typedef struct USet USet;
#endif
 
enum {
    USET_IGNORE_SPACE = 1,
 
    USET_CASE_INSENSITIVE = 2,
 
    USET_ADD_CASE_MAPPINGS = 4,
 
    USET_SIMPLE_CASE_INSENSITIVE = 6
};
 
typedef enum USetSpanCondition {
    USET_SPAN_NOT_CONTAINED = 0,
    USET_SPAN_CONTAINED = 1,
    USET_SPAN_SIMPLE = 2,
#ifndef U_HIDE_DEPRECATED_API
    USET_SPAN_CONDITION_COUNT
#endif  // U_HIDE_DEPRECATED_API
} USetSpanCondition;
 
enum {
    USET_SERIALIZED_STATIC_ARRAY_CAPACITY=8
};
 
typedef struct USerializedSet {
    const uint16_t *array;
    int32_t bmpLength;
    int32_t length;
    uint16_t staticArray[USET_SERIALIZED_STATIC_ARRAY_CAPACITY];
} USerializedSet;
 
/*********************************************************************
 * USet API
 *********************************************************************/
 
U_CAPI USet* U_EXPORT2
uset_openEmpty(void);
 
U_CAPI USet* U_EXPORT2
uset_open(UChar32 start, UChar32 end);
 
U_CAPI USet* U_EXPORT2
uset_openPattern(const UChar* pattern, int32_t patternLength,
                 UErrorCode* ec);
 
U_CAPI USet* U_EXPORT2
uset_openPatternOptions(const UChar* pattern, int32_t patternLength,
                 uint32_t options,
                 UErrorCode* ec);
 
U_CAPI void U_EXPORT2
uset_close(USet* set);
 
#if U_SHOW_CPLUSPLUS_API
 
U_NAMESPACE_BEGIN
 
U_DEFINE_LOCAL_OPEN_POINTER(LocalUSetPointer, USet, uset_close);
 
U_NAMESPACE_END
 
#endif
 
U_CAPI USet * U_EXPORT2
uset_clone(const USet *set);
 
U_CAPI UBool U_EXPORT2
uset_isFrozen(const USet *set);
 
U_CAPI void U_EXPORT2
uset_freeze(USet *set);
 
U_CAPI USet * U_EXPORT2
uset_cloneAsThawed(const USet *set);
 
U_CAPI void U_EXPORT2
uset_set(USet* set,
         UChar32 start, UChar32 end);
 
U_CAPI int32_t U_EXPORT2 
uset_applyPattern(USet *set,
                  const UChar *pattern, int32_t patternLength,
                  uint32_t options,
                  UErrorCode *status);
 
U_CAPI void U_EXPORT2
uset_applyIntPropertyValue(USet* set,
                           UProperty prop, int32_t value, UErrorCode* ec);
 
U_CAPI void U_EXPORT2
uset_applyPropertyAlias(USet* set,
                        const UChar *prop, int32_t propLength,
                        const UChar *value, int32_t valueLength,
                        UErrorCode* ec);
 
U_CAPI UBool U_EXPORT2
uset_resemblesPattern(const UChar *pattern, int32_t patternLength,
                      int32_t pos);
 
U_CAPI int32_t U_EXPORT2
uset_toPattern(const USet* set,
               UChar* result, int32_t resultCapacity,
               UBool escapeUnprintable,
               UErrorCode* ec);
 
U_CAPI void U_EXPORT2
uset_add(USet* set, UChar32 c);
 
U_CAPI void U_EXPORT2
uset_addAll(USet* set, const USet *additionalSet);
 
U_CAPI void U_EXPORT2
uset_addRange(USet* set, UChar32 start, UChar32 end);
 
U_CAPI void U_EXPORT2
uset_addString(USet* set, const UChar* str, int32_t strLen);
 
U_CAPI void U_EXPORT2
uset_addAllCodePoints(USet* set, const UChar *str, int32_t strLen);
 
U_CAPI void U_EXPORT2
uset_remove(USet* set, UChar32 c);
 
U_CAPI void U_EXPORT2
uset_removeRange(USet* set, UChar32 start, UChar32 end);
 
U_CAPI void U_EXPORT2
uset_removeString(USet* set, const UChar* str, int32_t strLen);
 
U_CAPI void U_EXPORT2
uset_removeAllCodePoints(USet *set, const UChar *str, int32_t length);
 
U_CAPI void U_EXPORT2
uset_removeAll(USet* set, const USet* removeSet);
 
U_CAPI void U_EXPORT2
uset_retain(USet* set, UChar32 start, UChar32 end);
 
U_CAPI void U_EXPORT2
uset_retainString(USet *set, const UChar *str, int32_t length);
 
U_CAPI void U_EXPORT2
uset_retainAllCodePoints(USet *set, const UChar *str, int32_t length);
 
U_CAPI void U_EXPORT2
uset_retainAll(USet* set, const USet* retain);
 
U_CAPI void U_EXPORT2
uset_compact(USet* set);
 
U_CAPI void U_EXPORT2
uset_complement(USet* set);
 
U_CAPI void U_EXPORT2
uset_complementRange(USet *set, UChar32 start, UChar32 end);
 
U_CAPI void U_EXPORT2
uset_complementString(USet *set, const UChar *str, int32_t length);
 
U_CAPI void U_EXPORT2
uset_complementAllCodePoints(USet *set, const UChar *str, int32_t length);
 
U_CAPI void U_EXPORT2
uset_complementAll(USet* set, const USet* complement);
 
U_CAPI void U_EXPORT2
uset_clear(USet* set);
 
U_CAPI void U_EXPORT2
uset_closeOver(USet* set, int32_t attributes);
 
U_CAPI void U_EXPORT2
uset_removeAllStrings(USet* set);
 
U_CAPI UBool U_EXPORT2
uset_isEmpty(const USet* set);
 
U_CAPI UBool U_EXPORT2
uset_hasStrings(const USet *set);
 
U_CAPI UBool U_EXPORT2
uset_contains(const USet* set, UChar32 c);
 
U_CAPI UBool U_EXPORT2
uset_containsRange(const USet* set, UChar32 start, UChar32 end);
 
U_CAPI UBool U_EXPORT2
uset_containsString(const USet* set, const UChar* str, int32_t strLen);
 
U_CAPI int32_t U_EXPORT2
uset_indexOf(const USet* set, UChar32 c);
 
U_CAPI UChar32 U_EXPORT2
uset_charAt(const USet* set, int32_t charIndex);
 
U_CAPI int32_t U_EXPORT2
uset_size(const USet* set);
 
U_CAPI int32_t U_EXPORT2
uset_getRangeCount(const USet *set);
 
#ifndef U_HIDE_DRAFT_API
 
U_CAPI int32_t U_EXPORT2
uset_getStringCount(const USet *set);
 
U_CAPI const UChar* U_EXPORT2
uset_getString(const USet *set, int32_t index, int32_t *pLength);
 
#endif  // U_HIDE_DRAFT_API
 
U_CAPI int32_t U_EXPORT2
uset_getItemCount(const USet* set);
 
U_CAPI int32_t U_EXPORT2
uset_getItem(const USet* set, int32_t itemIndex,
             UChar32* start, UChar32* end,
             UChar* str, int32_t strCapacity,
             UErrorCode* ec);
 
U_CAPI UBool U_EXPORT2
uset_containsAll(const USet* set1, const USet* set2);
 
U_CAPI UBool U_EXPORT2
uset_containsAllCodePoints(const USet* set, const UChar *str, int32_t strLen);
 
U_CAPI UBool U_EXPORT2
uset_containsNone(const USet* set1, const USet* set2);
 
U_CAPI UBool U_EXPORT2
uset_containsSome(const USet* set1, const USet* set2);
 
U_CAPI int32_t U_EXPORT2
uset_span(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition);
 
U_CAPI int32_t U_EXPORT2
uset_spanBack(const USet *set, const UChar *s, int32_t length, USetSpanCondition spanCondition);
 
U_CAPI int32_t U_EXPORT2
uset_spanUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition);
 
U_CAPI int32_t U_EXPORT2
uset_spanBackUTF8(const USet *set, const char *s, int32_t length, USetSpanCondition spanCondition);
 
U_CAPI UBool U_EXPORT2
uset_equals(const USet* set1, const USet* set2);
 
/*********************************************************************
 * Serialized set API
 *********************************************************************/
 
U_CAPI int32_t U_EXPORT2
uset_serialize(const USet* set, uint16_t* dest, int32_t destCapacity, UErrorCode* pErrorCode);
 
U_CAPI UBool U_EXPORT2
uset_getSerializedSet(USerializedSet* fillSet, const uint16_t* src, int32_t srcLength);
 
U_CAPI void U_EXPORT2
uset_setSerializedToOne(USerializedSet* fillSet, UChar32 c);
 
U_CAPI UBool U_EXPORT2
uset_serializedContains(const USerializedSet* set, UChar32 c);
 
U_CAPI int32_t U_EXPORT2
uset_getSerializedRangeCount(const USerializedSet* set);
 
U_CAPI UBool U_EXPORT2
uset_getSerializedRange(const USerializedSet* set, int32_t rangeIndex,
                        UChar32* pStart, UChar32* pEnd);
 
#if U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
#ifndef U_HIDE_DRAFT_API
 
namespace U_HEADER_ONLY_NAMESPACE {
 
// Note: Not U_COMMON_API, and not a subclass of UMemory, because this is a header-only class,
// not intended to be used via export from the ICU DLL.
 
class USetCodePointIterator {
public:
    USetCodePointIterator(const USetCodePointIterator &other) = default;
 
    bool operator==(const USetCodePointIterator &other) const {
        // No need to compare rangeCount & end given private constructor
        // and assuming we don't compare iterators across the set being modified.
        // And comparing rangeIndex is redundant with comparing c.
        // We might even skip comparing uset.
        // Unless we want operator==() to be "correct" for more than iteration.
        return uset == other.uset && c == other.c;
    }
 
    bool operator!=(const USetCodePointIterator &other) const { return !operator==(other); }
 
    UChar32 operator*() const { return c; }
 
    USetCodePointIterator &operator++() {
        if (c < end) {
            ++c;
        } else if (rangeIndex < rangeCount) {
            UErrorCode errorCode = U_ZERO_ERROR;
            int32_t result = uset_getItem(uset, rangeIndex, &c, &end, nullptr, 0, &errorCode);
            if (U_SUCCESS(errorCode) && result == 0) {
                ++rangeIndex;
            } else {
                c = end = U_SENTINEL;
            }
        } else {
            c = end = U_SENTINEL;
        }
        return *this;
    }
 
    USetCodePointIterator operator++(int) {
        USetCodePointIterator result(*this);
        operator++();
        return result;
    }
 
private:
    friend class USetCodePoints;
 
    USetCodePointIterator(const USet *uset, int32_t rangeIndex, int32_t rangeCount)
            : uset(uset), rangeIndex(rangeIndex), rangeCount(rangeCount),
                c(U_SENTINEL), end(U_SENTINEL) {
        // Fetch the first range.
        operator++();
    }
 
    const USet *uset;
    int32_t rangeIndex;
    int32_t rangeCount;
    UChar32 c, end;
};
 
class USetCodePoints {
public:
    USetCodePoints(const USet *uset) : uset(uset), rangeCount(uset_getRangeCount(uset)) {}
 
    USetCodePoints(const USetCodePoints &other) = default;
 
    USetCodePointIterator begin() const {
        return USetCodePointIterator(uset, 0, rangeCount);
    }
 
    USetCodePointIterator end() const {
        return USetCodePointIterator(uset, rangeCount, rangeCount);
    }
 
private:
    const USet *uset;
    int32_t rangeCount;
};
 
struct CodePointRange {
    struct iterator {
        iterator(UChar32 c) : c(c) {}
 
        bool operator==(const iterator &other) const { return c == other.c; }
        bool operator!=(const iterator &other) const { return !operator==(other); }
 
        UChar32 operator*() const { return c; }
 
        iterator &operator++() {
            ++c;
            return *this;
        }
 
        iterator operator++(int) {
            return c++;
        }
 
        UChar32 c;
    };
 
    CodePointRange(UChar32 start, UChar32 end) : rangeStart(start), rangeEnd(end) {}
    CodePointRange(const CodePointRange &other) = default;
    size_t size() const { return (rangeEnd + 1) - rangeStart; }
    iterator begin() const { return rangeStart; }
    iterator end() const { return rangeEnd + 1; }
 
    UChar32 rangeStart;
    UChar32 rangeEnd;
};
 
class USetRangeIterator {
public:
    USetRangeIterator(const USetRangeIterator &other) = default;
 
    bool operator==(const USetRangeIterator &other) const {
        // No need to compare rangeCount given private constructor
        // and assuming we don't compare iterators across the set being modified.
        // We might even skip comparing uset.
        // Unless we want operator==() to be "correct" for more than iteration.
        return uset == other.uset && rangeIndex == other.rangeIndex;
    }
 
    bool operator!=(const USetRangeIterator &other) const { return !operator==(other); }
 
    CodePointRange operator*() const {
        if (rangeIndex < rangeCount) {
            UChar32 start, end;
            UErrorCode errorCode = U_ZERO_ERROR;
            int32_t result = uset_getItem(uset, rangeIndex, &start, &end, nullptr, 0, &errorCode);
            if (U_SUCCESS(errorCode) && result == 0) {
                return CodePointRange(start, end);
            }
        }
        return CodePointRange(U_SENTINEL, U_SENTINEL);
    }
 
    USetRangeIterator &operator++() {
        ++rangeIndex;
        return *this;
    }
 
    USetRangeIterator operator++(int) {
        USetRangeIterator result(*this);
        ++rangeIndex;
        return result;
    }
 
private:
    friend class USetRanges;
 
    USetRangeIterator(const USet *uset, int32_t rangeIndex, int32_t rangeCount)
            : uset(uset), rangeIndex(rangeIndex), rangeCount(rangeCount) {}
 
    const USet *uset;
    int32_t rangeIndex;
    int32_t rangeCount;
};
 
class USetRanges {
public:
    USetRanges(const USet *uset) : uset(uset), rangeCount(uset_getRangeCount(uset)) {}
 
    USetRanges(const USetRanges &other) = default;
 
    USetRangeIterator begin() const {
        return USetRangeIterator(uset, 0, rangeCount);
    }
 
    USetRangeIterator end() const {
        return USetRangeIterator(uset, rangeCount, rangeCount);
    }
 
private:
    const USet *uset;
    int32_t rangeCount;
};
 
class USetStringIterator {
public:
    USetStringIterator(const USetStringIterator &other) = default;
 
    bool operator==(const USetStringIterator &other) const {
        // No need to compare count given private constructor
        // and assuming we don't compare iterators across the set being modified.
        // We might even skip comparing uset.
        // Unless we want operator==() to be "correct" for more than iteration.
        return uset == other.uset && index == other.index;
    }
 
    bool operator!=(const USetStringIterator &other) const { return !operator==(other); }
 
    std::u16string_view operator*() const {
        if (index < count) {
            int32_t length;
            const UChar *uchars = uset_getString(uset, index, &length);
            // assert uchars != nullptr;
            return {ConstChar16Ptr(uchars), static_cast<uint32_t>(length)};
        }
        return {};
    }
 
    USetStringIterator &operator++() {
        ++index;
        return *this;
    }
 
    USetStringIterator operator++(int) {
        USetStringIterator result(*this);
        ++index;
        return result;
    }
 
private:
    friend class USetStrings;
 
    USetStringIterator(const USet *uset, int32_t index, int32_t count)
            : uset(uset), index(index), count(count) {}
 
    const USet *uset;
    int32_t index;
    int32_t count;
};
 
class USetStrings {
public:
    USetStrings(const USet *uset) : uset(uset), count(uset_getStringCount(uset)) {}
 
    USetStrings(const USetStrings &other) = default;
 
    USetStringIterator begin() const {
        return USetStringIterator(uset, 0, count);
    }
 
    USetStringIterator end() const {
        return USetStringIterator(uset, count, count);
    }
 
private:
    const USet *uset;
    int32_t count;
};
 
class USetElementIterator {
public:
    USetElementIterator(const USetElementIterator &other) = default;
 
    bool operator==(const USetElementIterator &other) const {
        // No need to compare rangeCount & end given private constructor
        // and assuming we don't compare iterators across the set being modified.
        // We might even skip comparing uset.
        // Unless we want operator==() to be "correct" for more than iteration.
        return uset == other.uset && c == other.c && index == other.index;
    }
 
    bool operator!=(const USetElementIterator &other) const { return !operator==(other); }
 
    UnicodeString operator*() const {
        if (c >= 0) {
            return UnicodeString(c);
        } else if (index < totalCount) {
            int32_t length;
            const UChar *uchars = uset_getString(uset, index - rangeCount, &length);
            // assert uchars != nullptr;
            return UnicodeString(uchars, length);
        } else {
            return UnicodeString();
        }
    }
 
    USetElementIterator &operator++() {
        if (c < end) {
            ++c;
        } else if (index < rangeCount) {
            UErrorCode errorCode = U_ZERO_ERROR;
            int32_t result = uset_getItem(uset, index, &c, &end, nullptr, 0, &errorCode);
            if (U_SUCCESS(errorCode) && result == 0) {
                ++index;
            } else {
                c = end = U_SENTINEL;
            }
        } else if (c >= 0) {
            // assert index == rangeCount;
            // Switch from the last range to the first string.
            c = end = U_SENTINEL;
        } else {
            ++index;
        }
        return *this;
    }
 
    USetElementIterator operator++(int) {
        USetElementIterator result(*this);
        operator++();
        return result;
    }
 
private:
    friend class USetElements;
 
    USetElementIterator(const USet *uset, int32_t index, int32_t rangeCount, int32_t totalCount)
            : uset(uset), index(index), rangeCount(rangeCount), totalCount(totalCount),
                c(U_SENTINEL), end(U_SENTINEL) {
        if (index < rangeCount) {
            // Fetch the first range.
            operator++();
        }
        // Otherwise don't move beyond the (index - rangeCount)-th string.
    }
 
    const USet *uset;
    int32_t index;
    int32_t rangeCount;
    int32_t totalCount;
    UChar32 c, end;
};
 
class USetElements {
public:
    USetElements(const USet *uset)
        : uset(uset), rangeCount(uset_getRangeCount(uset)),
            stringCount(uset_getStringCount(uset)) {}
 
    USetElements(const USetElements &other) = default;
 
    USetElementIterator begin() const {
        return USetElementIterator(uset, 0, rangeCount, rangeCount + stringCount);
    }
 
    USetElementIterator end() const {
        return USetElementIterator(uset, rangeCount + stringCount, rangeCount, rangeCount + stringCount);
    }
 
private:
    const USet *uset;
    int32_t rangeCount, stringCount;
};
 
}  // namespace U_HEADER_ONLY_NAMESPACE
 
#endif  // U_HIDE_DRAFT_API
#endif  // U_SHOW_CPLUSPLUS_API || U_SHOW_CPLUSPLUS_HEADER_API
 
#endif  // __USET_H__