From 6add0b91bcea9440789bff05b694d0b2491812ba Mon Sep 17 00:00:00 2001 From: Timo <46269205+timo-mart@users.noreply.github.com> Date: Sun, 12 Jul 2026 05:16:42 +0300 Subject: [PATCH] feat: Add Finnish hyphenation (#2084) --- lib/Epub/Epub/hyphenation/Hyphenator.cpp | 2 +- .../Epub/hyphenation/LanguageRegistry.cpp | 7 +- .../Epub/hyphenation/generated/hyph-fi.trie.h | 95 +++++++++++++++++++ scripts/update_hyphenation.sh | 1 + 4 files changed, 102 insertions(+), 3 deletions(-) create mode 100644 lib/Epub/Epub/hyphenation/generated/hyph-fi.trie.h diff --git a/lib/Epub/Epub/hyphenation/Hyphenator.cpp b/lib/Epub/Epub/hyphenation/Hyphenator.cpp index ed05d061..eaea7b31 100644 --- a/lib/Epub/Epub/hyphenation/Hyphenator.cpp +++ b/lib/Epub/Epub/hyphenation/Hyphenator.cpp @@ -24,7 +24,7 @@ struct Iso639Mapping { }; static constexpr Iso639Mapping kIso639Mappings[] = {{"eng", "en"}, {"fra", "fr"}, {"fre", "fr"}, {"deu", "de"}, {"ger", "de"}, {"rus", "ru"}, {"spa", "es"}, {"ita", "it"}, - {"ukr", "uk"}, {"swe", "sv"}}; + {"ukr", "uk"}, {"swe", "sv"}, {"fin", "fi"}}; // Maps a BCP-47 or ISO 639-2 language tag to a language-specific hyphenator. const LanguageHyphenator* hyphenatorForLanguage(const std::string& langTag) { diff --git a/lib/Epub/Epub/hyphenation/LanguageRegistry.cpp b/lib/Epub/Epub/hyphenation/LanguageRegistry.cpp index fc770c76..2b0b1d70 100644 --- a/lib/Epub/Epub/hyphenation/LanguageRegistry.cpp +++ b/lib/Epub/Epub/hyphenation/LanguageRegistry.cpp @@ -7,6 +7,7 @@ #include "generated/hyph-de.trie.h" #include "generated/hyph-en.trie.h" #include "generated/hyph-es.trie.h" +#include "generated/hyph-fi.trie.h" #include "generated/hyph-fr.trie.h" #include "generated/hyph-it.trie.h" #include "generated/hyph-pl.trie.h" @@ -26,8 +27,9 @@ LanguageHyphenator italianHyphenator(it_patterns, isLatinLetter, toLowerLatin); LanguageHyphenator swedishHyphenator(sv_patterns, isLatinLetter, toLowerLatin); LanguageHyphenator ukrainianHyphenator(uk_patterns, isCyrillicLetter, toLowerCyrillic); LanguageHyphenator polishHyphenator(pl_patterns, isLatinLetter, toLowerLatin); +LanguageHyphenator finnishHyphenator(fi_patterns, isLatinLetter, toLowerLatin); -using EntryArray = std::array; +using EntryArray = std::array; const EntryArray& entries() { static const EntryArray kEntries = {{{"english", "en", &englishHyphenator}, @@ -38,7 +40,8 @@ const EntryArray& entries() { {"italian", "it", &italianHyphenator}, {"polish", "pl", &polishHyphenator}, {"swedish", "sv", &swedishHyphenator}, - {"ukrainian", "uk", &ukrainianHyphenator}}}; + {"ukrainian", "uk", &ukrainianHyphenator}, + {"finnish", "fi", &finnishHyphenator}}}; return kEntries; } diff --git a/lib/Epub/Epub/hyphenation/generated/hyph-fi.trie.h b/lib/Epub/Epub/hyphenation/generated/hyph-fi.trie.h new file mode 100644 index 00000000..062a83d4 --- /dev/null +++ b/lib/Epub/Epub/hyphenation/generated/hyph-fi.trie.h @@ -0,0 +1,95 @@ +#pragma once + +#include +#include + +#include "Epub/hyphenation/SerializedHyphenationTrie.h" + +// Auto-generated by generate_hyphenation_trie.py. Do not edit manually. +alignas(4) constexpr uint8_t fi_trie_data[] = { + 0x01, 0x01, 0x16, 0x0B, 0x0C, 0x17, 0x0C, 0x15, 0x0C, 0x0B, 0x16, 0x29, 0x2B, 0x20, 0x1F, 0x0C, + 0x33, 0x02, 0x0B, 0x02, 0x0B, 0x0C, 0x01, 0x0C, 0x34, 0x0B, 0x2A, 0x0B, 0x2A, 0x0B, 0x0C, 0x20, + 0x0B, 0x21, 0xA0, 0x00, 0x41, 0xA0, 0x02, 0x51, 0x21, 0x69, 0xFD, 0x21, 0x6C, 0xFD, 0xA1, 0x00, + 0x41, 0x62, 0xFD, 0xA0, 0x01, 0xA2, 0xA1, 0x00, 0x81, 0x6F, 0xFD, 0xA3, 0x00, 0x81, 0x69, 0x6F, + 0x75, 0xF8, 0xF8, 0xF8, 0x28, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0x6C, 0x72, 0xDE, 0xDE, 0xEA, + 0xDE, 0xDE, 0xDE, 0xF2, 0xF7, 0x22, 0xA4, 0xB6, 0xCD, 0xCD, 0xA1, 0x00, 0x81, 0x61, 0xD9, 0x28, + 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x72, 0xC3, 0xC3, 0xC3, 0xC3, 0xC3, 0xC3, 0xF6, 0xFB, + 0xA2, 0x00, 0x81, 0x61, 0x65, 0xC3, 0xC3, 0x48, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0x6C, 0x72, + 0xFF, 0xAB, 0xFF, 0xAB, 0xFF, 0xAB, 0xFF, 0xAB, 0xFF, 0xAB, 0xFF, 0xAB, 0xFF, 0xE3, 0xFF, 0xF9, + 0x49, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x6C, 0x72, 0xFF, 0x92, 0xFF, 0x92, 0xFF, 0x92, + 0xFF, 0x92, 0xFF, 0x92, 0xFF, 0x92, 0xFF, 0xC5, 0xFF, 0xA6, 0xFF, 0xCA, 0x47, 0x61, 0x65, 0x69, + 0x6F, 0x75, 0x79, 0xC3, 0xFF, 0x76, 0xFF, 0x76, 0xFF, 0x76, 0xFF, 0x76, 0xFF, 0x76, 0xFF, 0x76, + 0xFF, 0xA9, 0xA0, 0x00, 0xF1, 0x21, 0x73, 0xFD, 0xA1, 0x00, 0x41, 0x75, 0xFD, 0xA0, 0x00, 0x81, + 0x43, 0x61, 0x65, 0x69, 0xFF, 0x63, 0xFF, 0x63, 0xFF, 0x63, 0xC1, 0x01, 0xA2, 0x61, 0xFF, 0x59, + 0x4A, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x6C, 0x72, 0x76, 0xFF, 0x42, 0xFF, 0xE8, 0xFF, + 0x42, 0xFF, 0x42, 0xFF, 0x42, 0xFF, 0x42, 0xFF, 0x75, 0xFF, 0xED, 0xFF, 0xF0, 0xFF, 0xFA, 0xA1, + 0x00, 0x41, 0x73, 0xCE, 0x47, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0xFF, 0xFB, 0xFF, 0x1E, + 0xFF, 0x1E, 0xFF, 0x1E, 0xFF, 0x1E, 0xFF, 0x1E, 0xFF, 0x51, 0xA0, 0x01, 0x52, 0x21, 0x6F, 0xFD, + 0x22, 0x74, 0x6E, 0xFD, 0xFD, 0xA0, 0x01, 0x73, 0x21, 0x6E, 0xFD, 0x22, 0x61, 0x6F, 0xFD, 0xFA, + 0x21, 0x61, 0xEA, 0x21, 0x6B, 0xFD, 0x21, 0x65, 0xF2, 0xA4, 0x00, 0x41, 0x6E, 0x6A, 0x69, 0x6C, + 0xE7, 0xF2, 0xFA, 0xFD, 0x21, 0x73, 0xE1, 0x21, 0x75, 0xFD, 0xA1, 0x00, 0x41, 0x64, 0xFD, 0x21, + 0x74, 0xD6, 0x21, 0x73, 0xFD, 0x22, 0x65, 0x69, 0xFA, 0xFD, 0x21, 0x61, 0xCB, 0x21, 0x6E, 0xBD, + 0x22, 0x74, 0x6F, 0xBD, 0xFD, 0x21, 0x69, 0xC0, 0x21, 0x61, 0xFD, 0xA4, 0x00, 0x41, 0x70, 0x73, + 0x74, 0x6D, 0xEA, 0xEF, 0xF5, 0xFD, 0x21, 0x69, 0xD9, 0xA1, 0x00, 0x41, 0x6C, 0xFD, 0x47, 0x61, + 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0xFF, 0xBB, 0xFF, 0xCC, 0xFE, 0xA4, 0xFF, 0xED, 0xFE, 0xA4, + 0xFF, 0xFB, 0xFE, 0xD7, 0xA0, 0x01, 0x41, 0x21, 0x73, 0xFD, 0x21, 0x75, 0xFD, 0xA1, 0x00, 0x41, + 0x72, 0xFD, 0x49, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x6C, 0x72, 0xFE, 0x80, 0xFF, 0xFB, + 0xFE, 0x80, 0xFE, 0x80, 0xFE, 0x80, 0xFE, 0x80, 0xFE, 0xB3, 0xFF, 0x2B, 0xFE, 0x94, 0x21, 0x61, + 0xDC, 0x48, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x74, 0xFF, 0x3E, 0xFE, 0x61, 0xFE, 0x61, + 0xFE, 0x61, 0xFE, 0x61, 0xFE, 0x61, 0xFE, 0x94, 0xFF, 0xFD, 0x42, 0x69, 0x65, 0xFF, 0x80, 0xFF, + 0x40, 0x42, 0x6F, 0x65, 0xFF, 0x84, 0xFF, 0x47, 0x41, 0x75, 0xFF, 0x32, 0x21, 0x74, 0xFC, 0x42, + 0x61, 0x6F, 0xFF, 0xFD, 0xFF, 0x36, 0xA4, 0x00, 0x41, 0x73, 0x6C, 0x6A, 0x70, 0xE4, 0xEB, 0xF9, + 0xF2, 0x41, 0x79, 0xFF, 0x24, 0x21, 0x74, 0xFC, 0x21, 0x69, 0xFD, 0xA1, 0x00, 0x41, 0x73, 0xFD, + 0x42, 0x2E, 0x6E, 0xFF, 0x15, 0xFF, 0x15, 0x21, 0x61, 0xF9, 0x21, 0x65, 0xFD, 0xA1, 0x00, 0x41, + 0x64, 0xFD, 0x41, 0x65, 0xFE, 0xF8, 0x21, 0x6A, 0xFC, 0x42, 0x6B, 0x74, 0xFE, 0xFC, 0xFE, 0xFC, + 0x21, 0x73, 0xF9, 0x21, 0x69, 0xFD, 0xC3, 0x00, 0x41, 0x68, 0x70, 0x73, 0xFF, 0xF0, 0xFF, 0xFD, + 0xFF, 0x24, 0x41, 0x74, 0xFF, 0x23, 0xC2, 0x00, 0x41, 0x72, 0x68, 0xFF, 0x30, 0xFF, 0xFC, 0xA0, + 0x00, 0x52, 0x21, 0x72, 0xFD, 0x21, 0x69, 0xFA, 0x21, 0x6C, 0xFD, 0xA0, 0x00, 0x61, 0x21, 0x68, + 0xFD, 0x4A, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x74, 0x70, 0x63, 0xFF, 0x95, 0xFF, 0xAA, + 0xFF, 0xBC, 0xFF, 0xD5, 0xFD, 0xC1, 0xFF, 0xE5, 0xFD, 0xF4, 0xFF, 0xF1, 0xFF, 0xF7, 0xFF, 0xFD, + 0x48, 0x61, 0x65, 0x69, 0x6F, 0x75, 0x79, 0xC3, 0x73, 0xFD, 0xA2, 0xFD, 0xA2, 0xFD, 0xA2, 0xFD, + 0xA2, 0xFD, 0xA2, 0xFD, 0xA2, 0xFD, 0xD5, 0xFF, 0xDE, 0xA0, 0x00, 0x92, 0xA0, 0x00, 0xB2, 0xA0, + 0x01, 0x01, 0xC3, 0x00, 0x61, 0x69, 0x65, 0x79, 0xFE, 0x20, 0xFE, 0x20, 0xFF, 0xFD, 0x22, 0xA4, + 0xB6, 0xF4, 0xAD, 0x25, 0x79, 0x61, 0x6F, 0x75, 0xC3, 0xA8, 0xE6, 0xE6, 0xE9, 0xFB, 0x42, 0xB6, + 0xA4, 0xFF, 0x9D, 0xFF, 0x9D, 0x46, 0x79, 0x61, 0x6F, 0x75, 0xC3, 0x65, 0xFF, 0x96, 0xFF, 0xD4, + 0xFF, 0xD4, 0xFF, 0xD7, 0xFF, 0xF9, 0xFF, 0xD7, 0x22, 0xA4, 0xB6, 0xDB, 0xED, 0xA0, 0x00, 0x72, + 0xA0, 0x00, 0x71, 0x21, 0xA4, 0xFD, 0x21, 0xC3, 0xFD, 0x21, 0xA4, 0xFD, 0x21, 0x69, 0xF4, 0xA0, + 0x01, 0x22, 0x21, 0x70, 0xFD, 0x21, 0x6F, 0xFD, 0x21, 0x69, 0xFD, 0x26, 0x61, 0x6F, 0x75, 0xC3, + 0x65, 0x6C, 0xE2, 0xE2, 0xE2, 0xEE, 0xF1, 0xFD, 0x22, 0xA4, 0xB6, 0xD8, 0xD8, 0x21, 0x61, 0xD3, + 0xA0, 0x00, 0xB1, 0x24, 0x75, 0x69, 0x65, 0x6F, 0xCD, 0xCD, 0xFD, 0xFD, 0x24, 0x61, 0x65, 0x6F, + 0x75, 0xF4, 0xF4, 0xF4, 0xF4, 0x25, 0x79, 0xC3, 0x61, 0x75, 0x69, 0xBB, 0xE3, 0xE8, 0xEE, 0xF7, + 0xA0, 0x00, 0xD2, 0x22, 0xA4, 0xB6, 0xFD, 0xFD, 0x44, 0x61, 0x65, 0x6F, 0x69, 0xFF, 0x64, 0xFF, + 0x64, 0xFF, 0x64, 0xFF, 0x64, 0x42, 0x65, 0x61, 0xFF, 0x9B, 0xFF, 0xCB, 0x21, 0x65, 0xC4, 0x22, + 0x61, 0x75, 0xC1, 0xC1, 0xA0, 0x02, 0x32, 0x21, 0x73, 0xFD, 0x21, 0x6F, 0xFD, 0x49, 0x79, 0xC3, + 0x75, 0x61, 0x65, 0x69, 0x6F, 0x73, 0x6C, 0xFF, 0x80, 0xFF, 0xD6, 0xFF, 0xDB, 0xFF, 0xB0, 0xFF, + 0xE8, 0xFF, 0xEF, 0xFF, 0xF2, 0xFD, 0x70, 0xFF, 0xFD, 0x44, 0x69, 0x65, 0x6F, 0x75, 0xFF, 0x23, + 0xFF, 0x23, 0xFF, 0x23, 0xFF, 0x23, 0x43, 0x75, 0x61, 0x65, 0xFF, 0x5A, 0xFF, 0x8A, 0xFF, 0x8A, + 0x41, 0x76, 0xFF, 0x5F, 0x21, 0x61, 0xFC, 0xA0, 0x01, 0xC2, 0x21, 0x73, 0xFD, 0x21, 0x69, 0xFD, + 0x21, 0x65, 0xFD, 0x43, 0x69, 0x6F, 0x6B, 0xFF, 0xF1, 0xFD, 0xF7, 0xFF, 0xFD, 0xA0, 0x01, 0xA4, + 0x21, 0x73, 0xFD, 0x21, 0x61, 0xFD, 0x43, 0x6E, 0x74, 0x6B, 0xFC, 0x7D, 0xFC, 0x7D, 0xFF, 0xFD, + 0x41, 0x69, 0xFC, 0x73, 0x22, 0x61, 0x6F, 0xF2, 0xFC, 0x21, 0x69, 0xFB, 0x48, 0xC3, 0x61, 0x75, + 0x65, 0x6F, 0x69, 0x6C, 0x73, 0xFF, 0x3C, 0xFF, 0xAD, 0xFF, 0xBA, 0xFF, 0x20, 0xFF, 0x20, 0xFF, + 0x50, 0xFF, 0xD7, 0xFF, 0xFD, 0x44, 0x61, 0x69, 0x75, 0x79, 0xFE, 0xB7, 0xFE, 0xB7, 0xFE, 0xB7, + 0xFE, 0xB7, 0x42, 0x61, 0x69, 0xFE, 0xEE, 0xFE, 0xEE, 0x42, 0x75, 0x61, 0xFE, 0xE7, 0xFF, 0x17, + 0x42, 0xA4, 0xB6, 0xFE, 0xE6, 0xFF, 0x30, 0x24, 0x65, 0x61, 0x75, 0xC3, 0xDE, 0xEB, 0xF2, 0xF9, + 0x43, 0x61, 0x65, 0x6F, 0xFE, 0x8C, 0xFE, 0x8C, 0xFE, 0x8C, 0x42, 0x61, 0x75, 0xFE, 0xC6, 0xFE, + 0xC6, 0x44, 0x75, 0x61, 0x65, 0x6F, 0xFE, 0xBF, 0xFE, 0xEF, 0xFE, 0xEF, 0xFE, 0xEF, 0x41, 0xB6, + 0xFE, 0xB2, 0x21, 0xC3, 0xFC, 0x42, 0xA4, 0xB6, 0xFE, 0xB1, 0xFF, 0xFD, 0x43, 0x61, 0x6F, 0x79, + 0xFE, 0xD4, 0xFE, 0xD4, 0xFE, 0xD4, 0x42, 0x61, 0x65, 0xFE, 0x56, 0xFE, 0x56, 0x26, 0x69, 0x61, + 0x75, 0xC3, 0x65, 0x6F, 0xC3, 0xCD, 0xD4, 0xE8, 0xEF, 0xF9, 0xA0, 0x01, 0x11, 0x21, 0xA4, 0xFD, + 0xA0, 0x01, 0xE2, 0x21, 0x6E, 0xFD, 0x21, 0x69, 0xFD, 0x21, 0x64, 0xFD, 0xA0, 0x02, 0x03, 0x21, + 0x61, 0xFD, 0x21, 0x72, 0xFD, 0x21, 0x75, 0xFD, 0x21, 0x75, 0xFD, 0x23, 0xC3, 0x79, 0x73, 0xE2, + 0xEE, 0xFD, 0x41, 0x72, 0xFD, 0xD9, 0x42, 0x6C, 0x68, 0xFC, 0x47, 0xFF, 0xFC, 0xC1, 0x00, 0x81, + 0x69, 0xFB, 0xA6, 0x21, 0x76, 0xFA, 0x59, 0x62, 0x64, 0x66, 0x67, 0x68, 0x6A, 0x6B, 0x6C, 0x6D, + 0x6E, 0x70, 0x72, 0x73, 0x74, 0x76, 0xC3, 0x79, 0x6F, 0x75, 0x61, 0x65, 0x69, 0x2E, 0x63, 0x71, + 0xFB, 0xAE, 0xFB, 0xC9, 0xFB, 0xE1, 0xFB, 0xFA, 0xFC, 0x16, 0xFC, 0x16, 0xFC, 0x4A, 0xFC, 0x6E, + 0xFC, 0x16, 0xFC, 0xE8, 0xFD, 0x0C, 0xFD, 0x2B, 0xFD, 0xCB, 0xFD, 0xEA, 0xFC, 0x16, 0xFE, 0x42, + 0xFE, 0x65, 0xFE, 0x8F, 0xFE, 0xC7, 0xFF, 0x36, 0xFF, 0x71, 0xFF, 0xB7, 0xFF, 0xE5, 0xFF, 0xF0, + 0xFF, 0xFD, +}; + +constexpr SerializedHyphenationPatterns fi_patterns = { + 0x496u, + fi_trie_data, + sizeof(fi_trie_data), +}; diff --git a/scripts/update_hyphenation.sh b/scripts/update_hyphenation.sh index 1f8d643b..0a52669f 100755 --- a/scripts/update_hyphenation.sh +++ b/scripts/update_hyphenation.sh @@ -25,3 +25,4 @@ process it process uk process pl process sv +process fi