wzray/tt9
1
0
Fork 0
Code Activity

changed the injest-words script to break dashed words only up to the first dash

Browse source
This commit is contained in:
sspanak 2024-01-07 18:00:20 +02:00 • committed by Dimo Karaivanov
parent b6b8d5bed0
commit cffa66899d
1 changed files with 21 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

33
scripts/injest-words.js
View file

@ -54,26 +54,35 @@ function splitDashedWords(inputWords) {
return [];
}
const wordsSet = new Set();
const dashedRoots = new Set();
const repeatingDashedRoots = new Set();
for (const word of inputWords) {
if (!word.includes('-')) {
wordsSet.add(word);
const [root, ...others] = word.split('-');
if (root === undefined || others.length != 1) {
continue;
}
const parts = word.split('-');
let root = '';
for (let i = 0; i < parts.length - 1; i++) {
root += `${parts[i]}-`;
wordsSet.add(root);
if (dashedRoots.has(root)) {
repeatingDashedRoots.add(root);
} else {
dashedRoots.add(root);
}
wordsSet.add(parts[parts.length - 1]);
}
return Array.from(wordsSet);
const outputWords = new Set();
for (const word of inputWords) {
const [root, ...others] = word.split('-');
if (root && others.length === 1 && repeatingDashedRoots.has(root)) {
outputWords.add(`${root}-`);
outputWords.add(others.join('-'));
} else {
outputWords.add(word);
}
}
return Array.from(outputWords);
}