wzray/tt9
1
0
Fork 0
Code Activity

added support for preventing dashed words to be broken in the word injestion scripts

Browse source
This commit is contained in:
sspanak 2024-01-08 15:12:03 +02:00 • committed by Dimo Karaivanov
parent 4b7cef763a
commit e58af0d45f
2 changed files with 38 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

53
scripts/injest-words.js
View file

@ -4,8 +4,9 @@ const { createInterface } = require('readline');
function printHelp() {
console.log(`Usage ${basename(process.argv[1])} word-list.txt`);
console.log(`Usage ${basename(process.argv[1])} word-list.txt [split-ignore-list.txt]`);
console.log('Breaks dashed words into separate words, puts multiple words on a line on new lines and deletes repeating new lines.');
console.log('The split-ignore-list is optional. Allows for not splitting certain words by dashes.');
}
@ -21,9 +22,15 @@ function validateInput() {
process.exit(2);
}
return {
fileName: process.argv[2]
if (process.argv[3] && !existsSync(process.argv[3])) {
console.error(`Failure! Could not ignore list file "${process.argv[3]}."`);
process.exit(2);
}
return {
fileName: process.argv[2],
ignoreListFileName: process.argv[3]
};
}
@ -49,15 +56,21 @@ function cleanSpecialChars(line) {
}
function splitDashedWords(inputWords) {
if (!Array.isArray(inputWords)) {
function splitDashedWords(inputWords, ignoreList) {
const ignoreWords = ignoreList instanceof Set ? ignoreList : new Set();
if (!(inputWords instanceof Set)) {
return [];
}
const dashedRoots = new Set();
const repeatingDashedRoots = new Set();
const outputWords = new Set();
for (const word of inputWords) {
if (ignoreWords.has(word)) {
continue;
}
const [root, ...others] = word.split('-');
if (root === undefined || others.length != 1) {
continue;
@ -70,8 +83,6 @@ function splitDashedWords(inputWords) {
}
}
const outputWords = new Set();
for (const word of inputWords) {
const [root, ...others] = word.split('-');
if (root && others.length === 1 && repeatingDashedRoots.has(root)) {
@ -86,22 +97,24 @@ function splitDashedWords(inputWords) {
}
async function readWords(fileName) {
const words = new Set();
async function work({ fileName }) {
const wordsSet = new Set();
const lineReader = createInterface({ input: createReadStream(fileName) });
for await (const line of lineReader) {
const newWords = cleanSpecialChars(line);
for (let i = 0; i < newWords.length; i++) {
wordsSet.add(newWords[i]);
}
if (!fileName) {
return words;
}
const wordsArray = Array.from(wordsSet);
const splitWords = splitDashedWords(wordsArray);
for await (const line of createInterface({ input: createReadStream(fileName) })) {
cleanSpecialChars(line).forEach(w => words.add(w));
}
return words;
}
async function work({ fileName, ignoreListFileName }) {
const [ words, ignoreList ] = await Promise.all([ readWords(fileName), readWords(ignoreListFileName) ]);
const splitWords = splitDashedWords(words, ignoreList);
const filteredAndSortedWords = splitWords.filter(word => word.length > 1).sort();
return filteredAndSortedWords;