aboutsummaryrefslogtreecommitdiff
path: root/ext/bg/js/translator.js
diff options
context:
space:
mode:
authorAlex Yatskov <alex@foosoft.net>2017-09-23 12:42:02 -0700
committerAlex Yatskov <alex@foosoft.net>2017-09-23 12:42:02 -0700
commit327027d820759c62f50454822c11c0b8859c1d04 (patch)
tree3a13a25634b7eaf702634a3d6baeb8806c48ab02 /ext/bg/js/translator.js
parent2eb85cb835a4aece7839eba25c0030e9eb186f85 (diff)
parentd13cb09fae53834baaac254d7b55d02a26f4a5c0 (diff)
Merge branch 'dev'
Diffstat (limited to 'ext/bg/js/translator.js')
-rw-r--r--ext/bg/js/translator.js114
1 files changed, 108 insertions, 6 deletions
diff --git a/ext/bg/js/translator.js b/ext/bg/js/translator.js
index 1be485c7..21881cf3 100644
--- a/ext/bg/js/translator.js
+++ b/ext/bg/js/translator.js
@@ -37,8 +37,26 @@ class Translator {
}
async findTermsGrouped(text, dictionaries, alphanumeric) {
+ const titles = Object.keys(dictionaries);
const {length, definitions} = await this.findTerms(text, dictionaries, alphanumeric);
- return {length, definitions: dictTermsGroup(definitions, dictionaries)};
+
+ const definitionsGrouped = dictTermsGroup(definitions, dictionaries);
+ for (const definition of definitionsGrouped) {
+ await this.buildTermFrequencies(definition, titles);
+ }
+
+ return {length, definitions: definitionsGrouped};
+ }
+
+ async findTermsSplit(text, dictionaries, alphanumeric) {
+ const titles = Object.keys(dictionaries);
+ const {length, definitions} = await this.findTerms(text, dictionaries, alphanumeric);
+
+ for (const definition of definitions) {
+ await this.buildTermFrequencies(definition, titles);
+ }
+
+ return {length, definitions};
}
async findTerms(text, dictionaries, alphanumeric) {
@@ -51,17 +69,18 @@ class Translator {
const cache = {};
const titles = Object.keys(dictionaries);
- let deinflections = await this.findTermsDeinflected(text, titles, cache);
+ let deinflections = await this.findTermDeinflections(text, titles, cache);
const textHiragana = jpKatakanaToHiragana(text);
if (text !== textHiragana) {
- deinflections = deinflections.concat(await this.findTermsDeinflected(textHiragana, titles, cache));
+ deinflections = deinflections.concat(await this.findTermDeinflections(textHiragana, titles, cache));
}
let definitions = [];
for (const deinflection of deinflections) {
for (const definition of deinflection.definitions) {
- const tags = definition.tags.map(tag => dictTagBuild(tag, definition.tagMeta));
+ const tags = await this.expandTags(definition.tags, definition.dictionary);
tags.push(dictTagBuildSource(definition.dictionary));
+
definitions.push({
source: deinflection.source,
reasons: deinflection.reasons,
@@ -87,7 +106,7 @@ class Translator {
return {length, definitions};
}
- async findTermsDeinflected(text, titles, cache) {
+ async findTermDeinflections(text, titles, cache) {
const definer = async term => {
if (cache.hasOwnProperty(term)) {
return cache[term];
@@ -117,11 +136,94 @@ class Translator {
}
for (const definition of definitions) {
- const tags = definition.tags.map(tag => dictTagBuild(tag, definition.tagMeta));
+ const tags = await this.expandTags(definition.tags, definition.dictionary);
tags.push(dictTagBuildSource(definition.dictionary));
+
definition.tags = dictTagsSort(tags);
+ definition.stats = await this.expandStats(definition.stats, definition.dictionary);
+
+ definition.frequencies = [];
+ const metas = await this.database.findKanjiMeta(definition.character, titles);
+ for (const meta of metas) {
+ if (meta.mode === 'freq') {
+ definition.frequencies.push({
+ character: meta.character,
+ frequency: meta.data,
+ dictionary: meta.dictionary
+ });
+ }
+ }
}
return definitions;
}
+
+ async buildTermFrequencies(definition, titles) {
+ let metas = await this.database.findTermMeta(definition.expression, titles);
+ if (metas.length === 0) {
+ metas = await this.database.findTermMeta(definition.reading, titles);
+ }
+
+ definition.frequencies = [];
+ for (const meta of metas) {
+ if (meta.mode === 'freq') {
+ definition.frequencies.push({
+ expression: meta.expression,
+ frequency: meta.data,
+ dictionary: meta.dictionary
+ });
+ }
+ }
+ }
+
+ async expandTags(names, title) {
+ const tags = [];
+ for (const name of names) {
+ const base = name.split(':')[0];
+ const meta = await this.database.findTagForTitle(base, title);
+
+ const tag = {name};
+ for (const prop in meta || {}) {
+ if (prop !== 'name') {
+ tag[prop] = meta[prop];
+ }
+ }
+
+ tags.push(dictTagSanitize(tag));
+ }
+
+ return tags;
+ }
+
+ async expandStats(items, title) {
+ const stats = {};
+ for (const name in items) {
+ const base = name.split(':')[0];
+ const meta = await this.database.findTagForTitle(base, title);
+ const group = stats[meta.category] = stats[meta.category] || [];
+
+ const stat = {name, value: items[name]};
+ for (const prop in meta || {}) {
+ if (prop !== 'name') {
+ stat[prop] = meta[prop];
+ }
+ }
+
+ group.push(dictTagSanitize(stat));
+ }
+
+ for (const category in stats) {
+ stats[category].sort((a, b) => {
+ if (a.notes < b.notes) {
+ return -1;
+ } else if (a.notes > b.notes) {
+ return 1;
+ } else {
+ return 0;
+ }
+ });
+ }
+
+ return stats;
+ }
}