yomichan/ext/bg/js/translator.js

678 lines
24 KiB
JavaScript
Raw Normal View History

2016-03-20 17:52:27 +00:00
/*
* Copyright (C) 2016-2020 Yomichan Authors
2016-03-20 17:52:27 +00:00
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
2020-01-01 17:00:31 +00:00
* along with this program. If not, see <https://www.gnu.org/licenses/>.
2016-03-20 17:52:27 +00:00
*/
2020-03-11 02:30:36 +00:00
/* global
* Deinflector
2020-03-28 21:51:58 +00:00
* TextSourceMap
2020-03-11 02:30:36 +00:00
* dictEnabledSet
* dictTagBuildSource
* dictTagSanitize
* dictTagsSort
* dictTermsCompressTags
* dictTermsGroup
* dictTermsMergeByGloss
* dictTermsMergeBySequence
* dictTermsSort
* dictTermsUndupe
* jp
2020-03-11 02:30:36 +00:00
*/
2016-03-24 02:25:32 +00:00
2016-03-20 17:52:27 +00:00
class Translator {
2020-03-31 00:39:04 +00:00
constructor(database) {
this._database = database;
this._deinflector = null;
this._tagCache = new Map();
2016-03-26 21:16:21 +00:00
}
2017-07-10 21:10:58 +00:00
async prepare() {
const reasons = await this._fetchJsonAsset('/bg/lang/deinflect.json');
this._deinflector = new Deinflector(reasons);
2016-03-20 17:52:27 +00:00
}
2016-03-21 01:45:37 +00:00
clearDatabaseCaches() {
this._tagCache.clear();
}
async findTerms(mode, text, details, options) {
switch (mode) {
case 'group':
return await this._findTermsGrouped(text, details, options);
case 'merge':
return await this._findTermsMerged(text, details, options);
case 'split':
return await this._findTermsSplit(text, details, options);
case 'simple':
return await this._findTermsSimple(text, details, options);
default:
return [[], 0];
}
}
async findKanji(text, options) {
const dictionaries = dictEnabledSet(options);
const kanjiUnique = new Set();
for (const c of text) {
kanjiUnique.add(c);
}
const definitions = await this._database.findKanjiBulk([...kanjiUnique], dictionaries);
if (definitions.length === 0) {
return definitions;
}
if (definitions.length > 1) {
definitions.sort((a, b) => a.index - b.index);
}
for (const definition of definitions) {
const tags = await this._expandTags(definition.tags, definition.dictionary);
tags.push(dictTagBuildSource(definition.dictionary));
dictTagsSort(tags);
const stats = await this._expandStats(definition.stats, definition.dictionary);
definition.tags = tags;
definition.stats = stats;
}
await this._buildKanjiMeta(definitions, dictionaries);
return definitions;
}
// Private
async _getSequencedDefinitions(definitions, mainDictionary) {
const [definitionsBySequence, defaultDefinitions] = dictTermsMergeBySequence(definitions, mainDictionary);
const sequenceList = [];
const sequencedDefinitions = [];
for (const [key, value] of definitionsBySequence.entries()) {
sequenceList.push(key);
sequencedDefinitions.push({definitions: value, rawDefinitions: []});
}
for (const definition of await this._database.findTermsBySequenceBulk(sequenceList, mainDictionary)) {
sequencedDefinitions[definition.index].rawDefinitions.push(definition);
}
return {sequencedDefinitions, defaultDefinitions};
}
async _getMergedSecondarySearchResults(text, expressionsMap, secondarySearchDictionaries) {
2020-02-22 16:30:24 +00:00
if (secondarySearchDictionaries.size === 0) {
return [];
2019-10-19 16:24:38 +00:00
}
const expressionList = [];
const readingList = [];
2019-10-19 16:24:38 +00:00
for (const expression of expressionsMap.keys()) {
if (expression === text) { continue; }
2019-10-19 16:24:38 +00:00
for (const reading of expressionsMap.get(expression).keys()) {
expressionList.push(expression);
readingList.push(reading);
2019-10-19 16:24:38 +00:00
}
}
const definitions = await this._database.findTermsExactBulk(expressionList, readingList, secondarySearchDictionaries);
for (const definition of definitions) {
const definitionTags = await this._expandTags(definition.definitionTags, definition.dictionary);
definitionTags.push(dictTagBuildSource(definition.dictionary));
definition.definitionTags = definitionTags;
const termTags = await this._expandTags(definition.termTags, definition.dictionary);
definition.termTags = termTags;
}
if (definitions.length > 1) {
definitions.sort((a, b) => a.index - b.index);
}
return definitions;
2019-10-19 16:24:38 +00:00
}
async _getMergedDefinition(text, dictionaries, sequencedDefinition, defaultDefinitions, secondarySearchDictionaries, mergedByTermIndices) {
const result = sequencedDefinition.definitions;
const rawDefinitionsBySequence = sequencedDefinition.rawDefinitions;
for (const definition of rawDefinitionsBySequence) {
const definitionTags = await this._expandTags(definition.definitionTags, definition.dictionary);
definitionTags.push(dictTagBuildSource(definition.dictionary));
definition.definitionTags = definitionTags;
const termTags = await this._expandTags(definition.termTags, definition.dictionary);
definition.termTags = termTags;
}
const definitionsByGloss = dictTermsMergeByGloss(result, rawDefinitionsBySequence);
const secondarySearchResults = await this._getMergedSecondarySearchResults(text, result.expressions, secondarySearchDictionaries);
2017-10-15 02:19:16 +00:00
dictTermsMergeByGloss(result, defaultDefinitions.concat(secondarySearchResults), definitionsByGloss, mergedByTermIndices);
2020-02-15 21:17:11 +00:00
for (const definition of definitionsByGloss.values()) {
dictTagsSort(definition.definitionTags);
result.definitions.push(definition);
}
dictTermsSort(result.definitions, dictionaries);
const expressions = [];
2019-12-27 20:08:55 +00:00
for (const [expression, readingMap] of result.expressions.entries()) {
for (const [reading, termTagsMap] of readingMap.entries()) {
const termTags = [...termTagsMap.values()];
2019-11-27 03:01:54 +00:00
const score = termTags.map((tag) => tag.score).reduce((p, v) => p + v, 0);
expressions.push(this._createExpression(expression, reading, dictTagsSort(termTags), this._scoreToTermFrequency(score)));
}
}
result.expressions = expressions;
result.expression = Array.from(result.expression);
result.reading = Array.from(result.reading);
return result;
}
2017-10-17 08:16:38 +00:00
async _findTermsGrouped(text, details, options) {
2019-11-05 01:09:34 +00:00
const dictionaries = dictEnabledSet(options);
const [definitions, length] = await this._findTermsInternal(text, dictionaries, details, options);
2019-11-05 00:35:57 +00:00
const definitionsGrouped = dictTermsGroup(definitions, dictionaries);
await this._buildTermMeta(definitionsGrouped, dictionaries);
2019-11-05 00:35:57 +00:00
if (options.general.compactTags) {
for (const definition of definitionsGrouped) {
dictTermsCompressTags(definition.definitions);
}
}
2019-11-05 01:09:34 +00:00
return [definitionsGrouped, length];
2019-11-05 00:35:57 +00:00
}
async _findTermsMerged(text, details, options) {
2019-11-05 01:09:34 +00:00
const dictionaries = dictEnabledSet(options);
const secondarySearchDictionaries = new Map();
for (const [title, dictionary] of dictionaries.entries()) {
if (!dictionary.allowSecondarySearches) { continue; }
secondarySearchDictionaries.set(title, dictionary);
}
const [definitions, length] = await this._findTermsInternal(text, dictionaries, details, options);
const {sequencedDefinitions, defaultDefinitions} = await this._getSequencedDefinitions(definitions, options.general.mainDictionary);
const definitionsMerged = [];
const mergedByTermIndices = new Set();
for (const sequencedDefinition of sequencedDefinitions) {
const result = await this._getMergedDefinition(
text,
dictionaries,
sequencedDefinition,
defaultDefinitions,
secondarySearchDictionaries,
mergedByTermIndices
);
definitionsMerged.push(result);
}
const strayDefinitions = defaultDefinitions.filter((definition, index) => !mergedByTermIndices.has(index));
for (const groupedDefinition of dictTermsGroup(strayDefinitions, dictionaries)) {
// from dictTermsMergeBySequence
const {reasons, score, expression, reading, source, dictionary} = groupedDefinition;
const compatibilityDefinition = {
reasons,
score,
expression: [expression],
reading: [reading],
expressions: [this._createExpression(groupedDefinition.expression, groupedDefinition.reading)],
source,
dictionary,
definitions: groupedDefinition.definitions
};
definitionsMerged.push(compatibilityDefinition);
}
await this._buildTermMeta(definitionsMerged, dictionaries);
2017-10-12 06:59:09 +00:00
if (options.general.compactTags) {
for (const definition of definitionsMerged) {
dictTermsCompressTags(definition.definitions);
}
}
2019-11-05 01:09:34 +00:00
return [dictTermsSort(definitionsMerged), length];
}
async _findTermsSplit(text, details, options) {
2019-11-05 01:09:34 +00:00
const dictionaries = dictEnabledSet(options);
const [definitions, length] = await this._findTermsInternal(text, dictionaries, details, options);
2017-09-14 06:22:41 +00:00
await this._buildTermMeta(definitions, dictionaries);
2017-09-14 06:22:41 +00:00
2019-11-05 01:09:34 +00:00
return [definitions, length];
2017-07-10 21:10:58 +00:00
}
async _findTermsSimple(text, details, options) {
const dictionaries = dictEnabledSet(options);
const [definitions, length] = await this._findTermsInternal(text, dictionaries, details, options);
dictTermsSort(definitions);
return [definitions, length];
}
async _findTermsInternal(text, dictionaries, details, options) {
text = this._getSearchableText(text, options);
if (text.length === 0) {
return [[], 0];
}
2019-11-05 01:52:08 +00:00
const deinflections = (
details.wildcard ?
await this._findTermWildcard(text, dictionaries, details.wildcard) :
await this._findTermDeinflections(text, dictionaries, options)
2019-11-05 01:52:08 +00:00
);
2016-03-21 03:34:50 +00:00
2017-07-10 21:10:58 +00:00
let definitions = [];
for (const deinflection of deinflections) {
for (const definition of deinflection.definitions) {
const definitionTags = await this._expandTags(definition.definitionTags, definition.dictionary);
definitionTags.push(dictTagBuildSource(definition.dictionary));
const termTags = await this._expandTags(definition.termTags, definition.dictionary);
2017-09-13 01:29:16 +00:00
2019-12-27 20:08:55 +00:00
const {expression, reading} = definition;
const furiganaSegments = jp.distributeFurigana(expression, reading);
2019-12-27 20:08:55 +00:00
2017-07-10 21:10:58 +00:00
definitions.push({
source: deinflection.source,
2019-12-22 19:07:30 +00:00
rawSource: deinflection.rawSource,
2017-07-10 21:10:58 +00:00
reasons: deinflection.reasons,
score: definition.score,
id: definition.id,
dictionary: definition.dictionary,
2019-12-27 20:08:55 +00:00
expression,
reading,
furiganaSegments,
2017-07-10 21:10:58 +00:00
glossary: definition.glossary,
definitionTags: dictTagsSort(definitionTags),
termTags: dictTagsSort(termTags),
sequence: definition.sequence
2017-07-10 21:10:58 +00:00
});
2017-01-08 19:18:55 +00:00
}
2017-07-10 21:10:58 +00:00
}
definitions = dictTermsUndupe(definitions);
definitions = dictTermsSort(definitions, dictionaries);
2017-01-08 19:18:55 +00:00
2017-07-10 21:10:58 +00:00
let length = 0;
for (const definition of definitions) {
2019-12-22 19:07:30 +00:00
length = Math.max(length, definition.rawSource.length);
2017-07-10 21:10:58 +00:00
}
2019-11-05 01:09:34 +00:00
return [definitions, length];
2016-03-21 01:45:37 +00:00
}
async _findTermWildcard(text, dictionaries, wildcard) {
const definitions = await this._database.findTermsBulk([text], dictionaries, wildcard);
2019-11-05 01:52:08 +00:00
if (definitions.length === 0) {
return [];
}
return [{
source: text,
2019-12-22 19:07:30 +00:00
rawSource: text,
2019-11-05 01:52:08 +00:00
term: text,
rules: 0,
definitions,
reasons: []
}];
}
async _findTermDeinflections(text, dictionaries, options) {
const deinflections = this._getAllDeinflections(text, options);
2019-08-31 01:06:21 +00:00
if (deinflections.length === 0) {
return [];
}
const uniqueDeinflectionTerms = [];
const uniqueDeinflectionArrays = [];
const uniqueDeinflectionsMap = new Map();
for (const deinflection of deinflections) {
const term = deinflection.term;
let deinflectionArray = uniqueDeinflectionsMap.get(term);
if (typeof deinflectionArray === 'undefined') {
deinflectionArray = [];
uniqueDeinflectionTerms.push(term);
uniqueDeinflectionArrays.push(deinflectionArray);
uniqueDeinflectionsMap.set(term, deinflectionArray);
}
deinflectionArray.push(deinflection);
}
const definitions = await this._database.findTermsBulk(uniqueDeinflectionTerms, dictionaries, null);
2019-08-31 01:06:21 +00:00
for (const definition of definitions) {
const definitionRules = Deinflector.rulesToRuleFlags(definition.rules);
for (const deinflection of uniqueDeinflectionArrays[definition.index]) {
const deinflectionRules = deinflection.rules;
if (deinflectionRules === 0 || (definitionRules & deinflectionRules) !== 0) {
deinflection.definitions.push(definition);
}
}
2019-08-31 01:06:21 +00:00
}
2019-11-27 03:01:54 +00:00
return deinflections.filter((e) => e.definitions.length > 0);
2019-08-31 01:06:21 +00:00
}
_getAllDeinflections(text, options) {
2019-12-22 19:07:30 +00:00
const translationOptions = options.translation;
const collapseEmphaticOptions = [[false, false]];
switch (translationOptions.collapseEmphaticSequences) {
case 'true':
collapseEmphaticOptions.push([true, false]);
break;
case 'full':
collapseEmphaticOptions.push([true, false], [true, true]);
break;
}
2019-12-22 18:09:35 +00:00
const textOptionVariantArray = [
this._getTextOptionEntryVariants(translationOptions.convertHalfWidthCharacters),
this._getTextOptionEntryVariants(translationOptions.convertNumericCharacters),
this._getTextOptionEntryVariants(translationOptions.convertAlphabeticCharacters),
this._getTextOptionEntryVariants(translationOptions.convertHiraganaToKatakana),
this._getTextOptionEntryVariants(translationOptions.convertKatakanaToHiragana),
collapseEmphaticOptions
2019-12-22 18:09:35 +00:00
];
2016-03-21 03:34:50 +00:00
2019-08-31 01:06:21 +00:00
const deinflections = [];
2019-12-22 18:09:35 +00:00
const used = new Set();
for (const [halfWidth, numeric, alphabetic, katakana, hiragana, [collapseEmphatic, collapseEmphaticFull]] of this._getArrayVariants(textOptionVariantArray)) {
2019-12-22 18:09:35 +00:00
let text2 = text;
2020-03-28 21:51:58 +00:00
const sourceMap = new TextSourceMap(text2);
2019-12-22 19:07:30 +00:00
if (halfWidth) {
2020-03-28 21:51:58 +00:00
text2 = jp.convertHalfWidthKanaToFullWidth(text2, sourceMap);
2019-12-22 19:07:30 +00:00
}
if (numeric) {
2020-03-14 20:38:12 +00:00
text2 = jp.convertNumericToFullWidth(text2);
}
2019-12-22 19:07:30 +00:00
if (alphabetic) {
2020-03-28 21:51:58 +00:00
text2 = jp.convertAlphabeticToKana(text2, sourceMap);
2019-12-22 19:07:30 +00:00
}
2020-01-24 01:49:41 +00:00
if (katakana) {
text2 = jp.convertHiraganaToKatakana(text2);
2020-01-24 01:49:41 +00:00
}
if (hiragana) {
text2 = jp.convertKatakanaToHiragana(text2);
}
if (collapseEmphatic) {
text2 = jp.collapseEmphaticSequences(text2, collapseEmphaticFull, sourceMap);
}
2019-12-22 18:09:35 +00:00
for (let i = text2.length; i > 0; --i) {
const text2Substring = text2.substring(0, i);
if (used.has(text2Substring)) { break; }
used.add(text2Substring);
for (const deinflection of this._deinflector.deinflect(text2Substring)) {
2020-03-28 21:51:58 +00:00
deinflection.rawSource = sourceMap.source.substring(0, sourceMap.getSourceLength(i));
2019-12-22 19:07:30 +00:00
deinflections.push(deinflection);
}
2019-08-31 01:06:21 +00:00
}
}
return deinflections;
}
_getTextOptionEntryVariants(value) {
2019-12-22 19:07:30 +00:00
switch (value) {
case 'true': return [true];
case 'variant': return [false, true];
default: return [false];
}
}
async _buildTermMeta(definitions, dictionaries) {
2019-08-31 01:06:21 +00:00
const terms = [];
for (const definition of definitions) {
if (definition.expressions) {
terms.push(...definition.expressions);
} else {
terms.push(definition);
}
}
if (terms.length === 0) {
return;
}
2019-08-31 01:06:21 +00:00
// Create mapping of unique terms
const expressionsUnique = [];
const termsUnique = [];
const termsUniqueMap = new Map();
2019-08-31 01:06:21 +00:00
for (let i = 0, ii = terms.length; i < ii; ++i) {
const term = terms[i];
const expression = term.expression;
let termList = termsUniqueMap.get(expression);
if (typeof termList === 'undefined') {
termList = [];
2019-08-31 01:06:21 +00:00
expressionsUnique.push(expression);
termsUnique.push(termList);
termsUniqueMap.set(expression, termList);
2019-08-31 01:06:21 +00:00
}
termList.push(term);
// New data
term.frequencies = [];
term.pitches = [];
2019-08-31 01:06:21 +00:00
}
const metas = await this._database.findTermMetaBulk(expressionsUnique, dictionaries);
2020-01-25 03:24:05 +00:00
for (const {expression, mode, data, dictionary, index} of metas) {
switch (mode) {
case 'freq':
for (const term of termsUnique[index]) {
const frequencyData = this._getFrequencyData(expression, data, dictionary, term);
if (frequencyData === null) { continue; }
term.frequencies.push(frequencyData);
2020-01-25 03:24:05 +00:00
}
break;
case 'pitch':
for (const term of termsUnique[index]) {
const pitchData = await this._getPitchData(expression, data, dictionary, term);
if (pitchData === null) { continue; }
term.pitches.push(pitchData);
}
break;
2017-09-23 15:46:34 +00:00
}
2017-09-14 06:22:41 +00:00
}
}
async _buildKanjiMeta(definitions, dictionaries) {
const kanjiList = [];
for (const definition of definitions) {
kanjiList.push(definition.character);
definition.frequencies = [];
}
const metas = await this._database.findKanjiMetaBulk(kanjiList, dictionaries);
for (const {character, mode, data, dictionary, index} of metas) {
switch (mode) {
case 'freq':
definitions[index].frequencies.push({character, frequency: data, dictionary});
break;
}
}
}
async _expandTags(names, title) {
const tagMetaList = await this._getTagMetaList(names, title);
return tagMetaList.map((meta, index) => {
const name = names[index];
const tag = dictTagSanitize(Object.assign({}, meta !== null ? meta : {}, {name}));
return dictTagSanitize(tag);
});
2017-09-13 23:42:04 +00:00
}
2017-09-18 02:16:08 +00:00
async _expandStats(items, title) {
const names = Object.keys(items);
const tagMetaList = await this._getTagMetaList(names, title);
2017-09-18 02:16:08 +00:00
2020-02-15 18:12:03 +00:00
const statsGroups = new Map();
for (let i = 0; i < names.length; ++i) {
const name = names[i];
const meta = tagMetaList[i];
if (meta === null) { continue; }
const category = meta.category;
2020-02-15 18:12:03 +00:00
let group = statsGroups.get(category);
if (typeof group === 'undefined') {
group = [];
statsGroups.set(category, group);
}
const stat = Object.assign({}, meta, {name, value: items[name]});
2017-09-21 06:15:02 +00:00
group.push(dictTagSanitize(stat));
2017-09-18 02:16:08 +00:00
}
2020-02-15 18:12:03 +00:00
const stats = {};
const sortCompare = (a, b) => a.notes - b.notes;
2020-02-15 18:12:03 +00:00
for (const [category, group] of statsGroups.entries()) {
group.sort(sortCompare);
stats[category] = group;
2017-09-22 06:20:51 +00:00
}
2017-09-18 02:57:39 +00:00
return stats;
2017-09-18 02:16:08 +00:00
}
async _getTagMetaList(names, title) {
const tagMetaList = [];
let cache = this._tagCache.get(title);
2020-02-14 01:24:54 +00:00
if (typeof cache === 'undefined') {
cache = new Map();
this._tagCache.set(title, cache);
2020-02-14 01:24:54 +00:00
}
for (const name of names) {
const base = this._getNameBase(name);
2020-02-14 01:24:54 +00:00
let tagMeta = cache.get(base);
if (typeof tagMeta === 'undefined') {
tagMeta = await this._database.findTagForTitle(base, title);
2020-02-14 01:24:54 +00:00
cache.set(base, tagMeta);
}
2020-02-14 01:24:54 +00:00
tagMetaList.push(tagMeta);
}
return tagMetaList;
}
_getFrequencyData(expression, data, dictionary, term) {
if (data !== null && typeof data === 'object') {
const {frequency, reading} = data;
const termReading = term.reading || expression;
if (reading !== termReading) { return null; }
return {expression, frequency, dictionary};
}
return {expression, frequency: data, dictionary};
}
async _getPitchData(expression, data, dictionary, term) {
const reading = data.reading;
const termReading = term.reading || expression;
if (reading !== termReading) { return null; }
const pitches = [];
for (let {position, tags} of data.pitches) {
tags = Array.isArray(tags) ? await this._getTagMetaList(tags, dictionary) : [];
pitches.push({position, tags});
}
return {reading, pitches, dictionary};
}
_createExpression(expression, reading, termTags=null, termFrequency=null) {
const furiganaSegments = jp.distributeFurigana(expression, reading);
2019-12-27 20:08:55 +00:00
return {
expression,
reading,
furiganaSegments,
termTags,
termFrequency
};
}
_scoreToTermFrequency(score) {
if (score > 0) {
return 'popular';
} else if (score < 0) {
return 'rare';
} else {
return 'normal';
}
}
_getNameBase(name) {
const pos = name.indexOf(':');
return (pos >= 0 ? name.substring(0, pos) : name);
}
2019-12-22 18:09:35 +00:00
*_getArrayVariants(arrayVariants) {
2019-12-22 18:09:35 +00:00
const ii = arrayVariants.length;
let total = 1;
for (let i = 0; i < ii; ++i) {
total *= arrayVariants[i].length;
}
for (let a = 0; a < total; ++a) {
const variant = [];
let index = a;
for (let i = 0; i < ii; ++i) {
const entryVariants = arrayVariants[i];
variant.push(entryVariants[index % entryVariants.length]);
index = Math.floor(index / entryVariants.length);
}
yield variant;
}
}
_getSearchableText(text, options) {
if (!options.scanning.alphanumeric) {
let newText = '';
for (const c of text) {
if (!jp.isCodePointJapanese(c.codePointAt(0))) {
break;
}
newText += c;
}
text = newText;
}
return text;
}
async _fetchJsonAsset(url) {
const response = await fetch(chrome.runtime.getURL(url), {
method: 'GET',
mode: 'no-cors',
cache: 'default',
credentials: 'omit',
redirect: 'follow',
referrerPolicy: 'no-referrer'
});
if (!response.ok) {
throw new Error(`Failed to fetch ${url}: ${response.status}`);
}
return await response.json();
}
2016-03-20 17:52:27 +00:00
}