libs/minikin/Hyphenator.cpp

/*
 * Copyright (C) 2015 The Android Open Source Project
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include "minikin/Hyphenator.h"

#include <algorithm>
#include <memory>
#include <string>
#include <vector>

#include <unicode/uchar.h>
#include <unicode/uscript.h>

#include "minikin/Characters.h"

namespace minikin {

// The following are structs that correspond to tables inside the hyb file format

struct AlphabetTable0 {
    uint32_t version;
    uint32_t min_codepoint;
    uint32_t max_codepoint;
    uint8_t data[1];  // actually flexible array, size is known at runtime
};

struct AlphabetTable1 {
    uint32_t version;
    uint32_t n_entries;
    uint32_t data[1];  // actually flexible array, size is known at runtime

    static uint32_t codepoint(uint32_t entry) { return entry >> 11; }
    static uint32_t value(uint32_t entry) { return entry & 0x7ff; }
};

struct Trie {
    uint32_t version;
    uint32_t char_mask;
    uint32_t link_shift;
    uint32_t link_mask;
    uint32_t pattern_shift;
    uint32_t n_entries;
    uint32_t data[1];  // actually flexible array, size is known at runtime
};

struct Pattern {
    uint32_t version;
    uint32_t n_entries;
    uint32_t pattern_offset;
    uint32_t pattern_size;
    uint32_t data[1];  // actually flexible array, size is known at runtime

    // accessors
    static uint32_t len(uint32_t entry) { return entry >> 26; }
    static uint32_t shift(uint32_t entry) { return (entry >> 20) & 0x3f; }
    const uint8_t* buf(uint32_t entry) const {
        return reinterpret_cast<const uint8_t*>(this) + pattern_offset + (entry & 0xfffff);
    }
};

struct Header {
    uint32_t magic;
    uint32_t version;
    uint32_t alphabet_offset;
    uint32_t trie_offset;
    uint32_t pattern_offset;
    uint32_t file_size;

    // accessors
    const uint8_t* bytes() const { return reinterpret_cast<const uint8_t*>(this); }
    uint32_t alphabetVersion() const {
        return *reinterpret_cast<const uint32_t*>(bytes() + alphabet_offset);
    }
    const AlphabetTable0* alphabetTable0() const {
        return reinterpret_cast<const AlphabetTable0*>(bytes() + alphabet_offset);
    }
    const AlphabetTable1* alphabetTable1() const {
        return reinterpret_cast<const AlphabetTable1*>(bytes() + alphabet_offset);
    }
    const Trie* trieTable() const { return reinterpret_cast<const Trie*>(bytes() + trie_offset); }
    const Pattern* patternTable() const {
        return reinterpret_cast<const Pattern*>(bytes() + pattern_offset);
    }
};

// static
Hyphenator* Hyphenator::loadBinary(const uint8_t* patternData, size_t minPrefix, size_t minSuffix,
                                   const std::string& locale) {
    HyphenationLocale hyphenLocale = HyphenationLocale::OTHER;
    if (locale == "pl") {
        hyphenLocale = HyphenationLocale::POLISH;
    } else if (locale == "ca") {
        hyphenLocale = HyphenationLocale::CATALAN;
    } else if (locale == "sl") {
        hyphenLocale = HyphenationLocale::SLOVENIAN;
    }
    return new Hyphenator(patternData, minPrefix, minSuffix, hyphenLocale);
}

Hyphenator::Hyphenator(const uint8_t* patternData, size_t minPrefix, size_t minSuffix,
                       HyphenationLocale hyphenLocale)
        : mPatternData(patternData),
          mMinPrefix(minPrefix),
          mMinSuffix(minSuffix),
          mHyphenationLocale(hyphenLocale) {}

void Hyphenator::hyphenate(const U16StringPiece& word, HyphenationType* out) const {
    const size_t len = word.size();
    const size_t paddedLen = len + 2;  // start and stop code each count for 1
    if (mPatternData != nullptr && len >= mMinPrefix + mMinSuffix &&
        paddedLen <= MAX_HYPHENATED_SIZE) {
        uint16_t alpha_codes[MAX_HYPHENATED_SIZE];
        const HyphenationType hyphenValue = alphabetLookup(alpha_codes, word);
        if (hyphenValue != HyphenationType::DONT_BREAK) {
            hyphenateFromCodes(alpha_codes, paddedLen, hyphenValue, out);
            return;
        }
        // TODO: try NFC normalization
        // TODO: handle non-BMP Unicode (requires remapping of offsets)
    }
    // Note that we will always get here if the word contains a hyphen or a soft hyphen, because the
    // alphabet is not expected to contain a hyphen or a soft hyphen character, so alphabetLookup
    // would return DONT_BREAK.
    hyphenateWithNoPatterns(word, out);
}

// This function determines whether a character is like U+2010 HYPHEN in
// line breaking and usage: a character immediately after which line breaks
// are allowed, but words containing it should not be automatically
// hyphenated using patterns. This is a curated set, created by manually
// inspecting all the characters that have the Unicode line breaking
// property of BA or HY and seeing which ones are hyphens.
bool Hyphenator::isLineBreakingHyphen(uint32_t c) {
    return (c == 0x002D ||  // HYPHEN-MINUS
            c == 0x058A ||  // ARMENIAN HYPHEN
            c == 0x05BE ||  // HEBREW PUNCTUATION MAQAF
            c == 0x1400 ||  // CANADIAN SYLLABICS HYPHEN
            c == 0x2010 ||  // HYPHEN
            c == 0x2013 ||  // EN DASH
            c == 0x2027 ||  // HYPHENATION POINT
            c == 0x2E17 ||  // DOUBLE OBLIQUE HYPHEN
            c == 0x2E40);   // DOUBLE HYPHEN
}

EndHyphenEdit editForThisLine(HyphenationType type) {
    switch (type) {
        case HyphenationType::BREAK_AND_INSERT_HYPHEN:
        case HyphenationType::BREAK_AND_INSERT_HYPHEN_AT_CURRENT_AND_NEXT_LINE:
            return EndHyphenEdit::INSERT_HYPHEN;
        case HyphenationType::BREAK_AND_INSERT_ARMENIAN_HYPHEN:
            return EndHyphenEdit::INSERT_ARMENIAN_HYPHEN;
        case HyphenationType::BREAK_AND_INSERT_MAQAF:
            return EndHyphenEdit::INSERT_MAQAF;
        case HyphenationType::BREAK_AND_INSERT_UCAS_HYPHEN:
            return EndHyphenEdit::INSERT_UCAS_HYPHEN;
        case HyphenationType::BREAK_AND_REPLACE_WITH_HYPHEN:
            return EndHyphenEdit::REPLACE_WITH_HYPHEN;
        case HyphenationType::BREAK_AND_INSERT_HYPHEN_AND_ZWJ:
            return EndHyphenEdit::INSERT_ZWJ_AND_HYPHEN;
        case HyphenationType::DONT_BREAK:  // Hyphen edit for non breaking case doesn't make sense.
        default:
            return EndHyphenEdit::NO_EDIT;
    }
}

StartHyphenEdit editForNextLine(HyphenationType type) {
    switch (type) {
        case HyphenationType::BREAK_AND_INSERT_HYPHEN_AT_NEXT_LINE:
        case HyphenationType::BREAK_AND_INSERT_HYPHEN_AT_CURRENT_AND_NEXT_LINE:
            return StartHyphenEdit::INSERT_HYPHEN;
        case HyphenationType::BREAK_AND_INSERT_HYPHEN_AND_ZWJ:
            return StartHyphenEdit::INSERT_ZWJ;
        case HyphenationType::DONT_BREAK:  // Hyphen edit for non breaking case doesn't make sense.
        default:
            return StartHyphenEdit::NO_EDIT;
    }
}

static UScriptCode getScript(uint32_t codePoint) {
    UErrorCode errorCode = U_ZERO_ERROR;
    const UScriptCode script = uscript_getScript(static_cast<UChar32>(codePoint), &errorCode);
    if (U_SUCCESS(errorCode)) {
        return script;
    } else {
        return USCRIPT_INVALID_CODE;
    }
}

static inline int32_t getJoiningType(UChar32 codepoint) {
    return u_getIntPropertyValue(codepoint, UCHAR_JOINING_TYPE);
}

// Assumption for caller: location must be >= 2 and word[location] == CHAR_SOFT_HYPHEN.
// This function decides if the letters before and after the hyphen should appear as joining.
static inline HyphenationType getHyphTypeForArabic(const U16StringPiece& word, size_t location) {
    ssize_t i = location;
    int32_t type = U_JT_NON_JOINING;
    while (static_cast<size_t>(i) < word.size() &&
           (type = getJoiningType(word[i])) == U_JT_TRANSPARENT) {
        i++;
    }
    if (type == U_JT_DUAL_JOINING || type == U_JT_RIGHT_JOINING || type == U_JT_JOIN_CAUSING) {
        // The next character is of the type that may join the last character. See if the last
        // character is also of the right type.
        i = location - 2;  // Skip the soft hyphen
        type = U_JT_NON_JOINING;
        while (i >= 0 && (type = getJoiningType(word[i])) == U_JT_TRANSPARENT) {
            i--;
        }
        if (type == U_JT_DUAL_JOINING || type == U_JT_LEFT_JOINING || type == U_JT_JOIN_CAUSING) {
            return HyphenationType::BREAK_AND_INSERT_HYPHEN_AND_ZWJ;
        }
    }
    return HyphenationType::BREAK_AND_INSERT_HYPHEN;
}

HyphenationType Hyphenator::hyphenationTypeBasedOnScriptAndLocale(uint32_t codePoint) const {
    // Note: It's not clear what the best hyphen for Hebrew is. While maqaf is the "correct" hyphen
    // for Hebrew, modern practice may have shifted towards Western hyphens. We use normal hyphens
    // for now to be safe.  BREAK_AND_INSERT_MAQAF is already implemented, so if we want to switch
    // to maqaf for Hebrew, we can simply add a condition here.
    const UScriptCode script = getScript(codePoint);
    if (script == USCRIPT_KANNADA || script == USCRIPT_MALAYALAM || script == USCRIPT_TAMIL ||
        script == USCRIPT_TELUGU) {
        // Grantha is not included, since we don't support non-BMP hyphenation yet.
        return HyphenationType::BREAK_AND_DONT_INSERT_HYPHEN;
    } else if (script == USCRIPT_ARMENIAN) {
        return HyphenationType::BREAK_AND_INSERT_ARMENIAN_HYPHEN;
    } else if (script == USCRIPT_CANADIAN_ABORIGINAL) {
        return HyphenationType::BREAK_AND_INSERT_UCAS_HYPHEN;
    } else if (isRepeatHyphen(script, mHyphenationLocale)) {
        return HyphenationType::BREAK_AND_INSERT_HYPHEN_AT_CURRENT_AND_NEXT_LINE;
    } else {
        return HyphenationType::BREAK_AND_INSERT_HYPHEN;
    }
}

// Use various recommendations of UAX #14 Unicode Line Breaking Algorithm for hyphenating words
// that didn't match patterns, especially words that contain hyphens or soft hyphens (See sections
// 5.3, Use of Hyphen, and 5.4, Use of Soft Hyphen).
void Hyphenator::hyphenateWithNoPatterns(const U16StringPiece& word, HyphenationType* out) const {
    out[0] = HyphenationType::DONT_BREAK;
    for (size_t i = 1; i < word.size(); i++) {
        const uint16_t prevChar = word[i - 1];
        if (i > 1 && isLineBreakingHyphen(prevChar)) {
            // Break after hyphens, but only if they don't start the word.

            if ((prevChar == CHAR_HYPHEN_MINUS || prevChar == CHAR_HYPHEN) &&
                isRepeatHyphen(getScript(word[i]), mHyphenationLocale)) {
                // In Polish and Slovenian, hyphens get repeated at the next line. To be safe,
                // we will do this only if the next character is Latin.
                out[i] = HyphenationType::BREAK_AND_INSERT_HYPHEN_AT_NEXT_LINE;
            } else {
                out[i] = HyphenationType::BREAK_AND_DONT_INSERT_HYPHEN;
            }
        } else if (i > 1 && prevChar == CHAR_SOFT_HYPHEN) {
            // Break after soft hyphens, but only if they don't start the word (a soft hyphen
            // starting the word doesn't give any useful break opportunities). The type of the break
            // is based on the script of the character we break on.
            if (getScript(word[i]) == USCRIPT_ARABIC) {
                // For Arabic, we need to look and see if the characters around the soft hyphen
                // actually join. If they don't, we'll just insert a normal hyphen.
                out[i] = getHyphTypeForArabic(word, i);
            } else {
                out[i] = hyphenationTypeBasedOnScriptAndLocale(word[i]);
            }
        } else if (prevChar == CHAR_MIDDLE_DOT && mMinPrefix < i && i <= word.size() - mMinSuffix &&
                   ((word[i - 2] == 'l' && word[i] == 'l') ||
                    (word[i - 2] == 'L' && word[i] == 'L')) &&
                   mHyphenationLocale == HyphenationLocale::CATALAN) {
            // In Catalan, "l·l" should break as "l-" on the first line
            // and "l" on the next line.
            out[i] = HyphenationType::BREAK_AND_REPLACE_WITH_HYPHEN;
        } else {
            out[i] = HyphenationType::DONT_BREAK;
        }
    }
}

HyphenationType Hyphenator::alphabetLookup(uint16_t* alpha_codes,
                                           const U16StringPiece& word) const {
    const Header* header = getHeader();
    HyphenationType result = HyphenationType::BREAK_AND_INSERT_HYPHEN;
    // TODO: check header magic
    uint32_t alphabetVersion = header->alphabetVersion();
    if (alphabetVersion == 0) {
        const AlphabetTable0* alphabet = header->alphabetTable0();
        uint32_t min_codepoint = alphabet->min_codepoint;
        uint32_t max_codepoint = alphabet->max_codepoint;
        alpha_codes[0] = 0;  // word start
        for (size_t i = 0; i < word.size(); i++) {
            uint16_t c = word[i];
            if (c < min_codepoint || c >= max_codepoint) {
                return HyphenationType::DONT_BREAK;
            }
            uint8_t code = alphabet->data[c - min_codepoint];
            if (code == 0) {
                return HyphenationType::DONT_BREAK;
            }
            if (result == HyphenationType::BREAK_AND_INSERT_HYPHEN) {
                result = hyphenationTypeBasedOnScriptAndLocale(c);
            }
            alpha_codes[i + 1] = code;
        }
        alpha_codes[word.size() + 1] = 0;  // word termination
        return result;
    } else if (alphabetVersion == 1) {
        const AlphabetTable1* alphabet = header->alphabetTable1();
        size_t n_entries = alphabet->n_entries;
        const uint32_t* begin = alphabet->data;
        const uint32_t* end = begin + n_entries;
        alpha_codes[0] = 0;
        for (size_t i = 0; i < word.size(); i++) {
            uint16_t c = word[i];
            auto p = std::lower_bound(begin, end, c << 11);
            if (p == end) {
                return HyphenationType::DONT_BREAK;
            }
            uint32_t entry = *p;
            if (AlphabetTable1::codepoint(entry) != c) {
                return HyphenationType::DONT_BREAK;
            }
            if (result == HyphenationType::BREAK_AND_INSERT_HYPHEN) {
                result = hyphenationTypeBasedOnScriptAndLocale(c);
            }
            alpha_codes[i + 1] = AlphabetTable1::value(entry);
        }
        alpha_codes[word.size() + 1] = 0;
        return result;
    }
    return HyphenationType::DONT_BREAK;
}

/**
 * Internal implementation, after conversion to codes. All case folding and normalization
 * has been done by now, and all characters have been found in the alphabet.
 * Note: len here is the padded length including 0 codes at start and end.
 **/
void Hyphenator::hyphenateFromCodes(const uint16_t* codes, size_t len, HyphenationType hyphenValue,
                                    HyphenationType* out) const {
    static_assert(sizeof(HyphenationType) == sizeof(uint8_t), "HyphnationType must be uint8_t.");
    // Reuse the result array as a buffer for calculating intermediate hyphenation numbers.
    uint8_t* buffer = reinterpret_cast<uint8_t*>(out);

    const Header* header = getHeader();
    const Trie* trie = header->trieTable();
    const Pattern* pattern = header->patternTable();
    uint32_t char_mask = trie->char_mask;
    uint32_t link_shift = trie->link_shift;
    uint32_t link_mask = trie->link_mask;
    uint32_t pattern_shift = trie->pattern_shift;
    size_t maxOffset = len - mMinSuffix - 1;
    for (size_t i = 0; i < len - 1; i++) {
        uint32_t node = 0;  // index into Trie table
        for (size_t j = i; j < len; j++) {
            uint16_t c = codes[j];
            uint32_t entry = trie->data[node + c];
            if ((entry & char_mask) == c) {
                node = (entry & link_mask) >> link_shift;
            } else {
                break;
            }
            uint32_t pat_ix = trie->data[node] >> pattern_shift;
            // pat_ix contains a 3-tuple of length, shift (number of trailing zeros), and an offset
            // into the buf pool. This is the pattern for the substring (i..j) we just matched,
            // which we combine (via point-wise max) into the buffer vector.
            if (pat_ix != 0) {
                uint32_t pat_entry = pattern->data[pat_ix];
                int pat_len = Pattern::len(pat_entry);
                int pat_shift = Pattern::shift(pat_entry);
                const uint8_t* pat_buf = pattern->buf(pat_entry);
                int offset = j + 1 - (pat_len + pat_shift);
                // offset is the index within buffer that lines up with the start of pat_buf
                int start = std::max((int)mMinPrefix - offset, 0);
                int end = std::min(pat_len, (int)maxOffset - offset);
                for (int k = start; k < end; k++) {
                    buffer[offset + k] = std::max(buffer[offset + k], pat_buf[k]);
                }
            }
        }
    }
    // Since the above calculation does not modify values outside
    // [mMinPrefix, len - mMinSuffix], they are left as 0 = DONT_BREAK.
    for (size_t i = mMinPrefix; i < maxOffset; i++) {
        // Hyphenation opportunities happen when the hyphenation numbers are odd.
        out[i] = (buffer[i] & 1u) ? hyphenValue : HyphenationType::DONT_BREAK;
    }
}

bool Hyphenator::isRepeatHyphen(UScriptCode script,
                                HyphenationLocale locale) const {
    return script == USCRIPT_LATIN &&
            (locale == HyphenationLocale::POLISH ||locale == HyphenationLocale::SLOVENIAN);
}

}  // namespace minikin