| 1 | // © 2021 and later: Unicode, Inc. and others. |
| 2 | // License & terms of use: https://www.unicode.org/copyright.html |
| 3 | |
| 4 | // emojiprops.h |
| 5 | // created: 2021sep03 Markus W. Scherer |
| 6 | |
| 7 | #ifndef __EMOJIPROPS_H__ |
| 8 | #define __EMOJIPROPS_H__ |
| 9 | |
| 10 | #include "unicode/utypes.h" |
| 11 | #include "unicode/ucptrie.h" |
| 12 | #include "unicode/udata.h" |
| 13 | #include "unicode/uobject.h" |
| 14 | #include "uset_imp.h" |
| 15 | |
| 16 | U_NAMESPACE_BEGIN |
| 17 | |
| 18 | class EmojiProps : public UMemory { |
| 19 | public: |
| 20 | // @internal |
| 21 | EmojiProps(UErrorCode &errorCode) { load(errorCode); } |
| 22 | ~EmojiProps(); |
| 23 | |
| 24 | static const EmojiProps *getSingleton(UErrorCode &errorCode); |
| 25 | static UBool hasBinaryProperty(UChar32 c, UProperty which); |
| 26 | static UBool hasBinaryProperty(const char16_t *s, int32_t length, UProperty which); |
| 27 | |
| 28 | void addPropertyStarts(const USetAdder *sa, UErrorCode &errorCode) const; |
| 29 | void addStrings(const USetAdder *sa, UProperty which, UErrorCode &errorCode) const; |
| 30 | |
| 31 | enum { |
| 32 | // Byte offsets from the start of the data, after the generic header, |
| 33 | // in ascending order. |
| 34 | // UCPTrie=CodePointTrie, follows the indexes |
| 35 | IX_CPTRIE_OFFSET, |
| 36 | IX_RESERVED1, |
| 37 | IX_RESERVED2, |
| 38 | IX_RESERVED3, |
| 39 | |
| 40 | // UCharsTrie=CharsTrie |
| 41 | IX_BASIC_EMOJI_TRIE_OFFSET, |
| 42 | IX_EMOJI_KEYCAP_SEQUENCE_TRIE_OFFSET, |
| 43 | IX_RGI_EMOJI_MODIFIER_SEQUENCE_TRIE_OFFSET, |
| 44 | IX_RGI_EMOJI_FLAG_SEQUENCE_TRIE_OFFSET, |
| 45 | IX_RGI_EMOJI_TAG_SEQUENCE_TRIE_OFFSET, |
| 46 | IX_RGI_EMOJI_ZWJ_SEQUENCE_TRIE_OFFSET, |
| 47 | IX_RESERVED10, |
| 48 | IX_RESERVED11, |
| 49 | IX_RESERVED12, |
| 50 | IX_TOTAL_SIZE, |
| 51 | |
| 52 | // Not initially byte offsets. |
| 53 | IX_RESERVED14, |
| 54 | IX_RESERVED15, |
| 55 | IX_COUNT // 16 |
| 56 | }; |
| 57 | |
| 58 | // Properties in the code point trie. |
| 59 | enum { |
| 60 | // https://www.unicode.org/reports/tr51/#Emoji_Properties |
| 61 | BIT_EMOJI, |
| 62 | BIT_EMOJI_PRESENTATION, |
| 63 | BIT_EMOJI_MODIFIER, |
| 64 | BIT_EMOJI_MODIFIER_BASE, |
| 65 | BIT_EMOJI_COMPONENT, |
| 66 | BIT_EXTENDED_PICTOGRAPHIC, |
| 67 | // https://www.unicode.org/reports/tr51/#Emoji_Sets |
| 68 | BIT_BASIC_EMOJI |
| 69 | }; |
| 70 | |
| 71 | private: |
| 72 | static UBool U_CALLCONV |
| 73 | isAcceptable(void *context, const char *type, const char *name, const UDataInfo *pInfo); |
| 74 | /** Input i: One of the IX_..._TRIE_OFFSET indexes into the data file indexes[] array. */ |
| 75 | static int32_t getStringTrieIndex(int32_t i) { |
| 76 | return i - IX_BASIC_EMOJI_TRIE_OFFSET; |
| 77 | } |
| 78 | |
| 79 | void load(UErrorCode &errorCode); |
| 80 | UBool hasBinaryPropertyImpl(UChar32 c, UProperty which) const; |
| 81 | UBool hasBinaryPropertyImpl(const char16_t *s, int32_t length, UProperty which) const; |
| 82 | |
| 83 | UDataMemory *memory = nullptr; |
| 84 | UCPTrie *cpTrie = nullptr; |
| 85 | const char16_t *stringTries[6] = { nullptr, nullptr, nullptr, nullptr, nullptr, nullptr }; |
| 86 | }; |
| 87 | |
| 88 | U_NAMESPACE_END |
| 89 | |
| 90 | #endif // __EMOJIPROPS_H__ |
| 91 | |