blob: 34eb5de82bd60c73c2be0280156b19c937eaf1f2 [file] [log] [blame]
// Copyright 2018 the V8 project authors. All rights reserved.
// This code is governed by the BSD license found in the LICENSE file.
/*---
esid: sec-Intl.Segmenter.prototype.segment
description: Verifies the behavior for the "segment" function of the Segmenter prototype object.
info: |
Intl.Segmenter.prototype.segment( string )
features: [Intl.Segmenter]
---*/
const seg = new Intl.Segmenter([], {granularity: "sentence"})
for (const text of [
"Hello world!", // English
" Hello world! ", // English with space before/after
" Hello world? Foo bar!", // English
"Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
"Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese
"Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
"Решение Индии о покупке российских С-400 расценили как вызов США", // Russian
"הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew,
"ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
"भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi
"ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
"'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
"台北》抹黑柯P失敗?朱學恒酸:姚文智氣pupu嗆大老闆", // Chinese
"วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai
"九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese
"법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
]) {
let segments = [];
// Create another %SegmentIterator% to compare with result from the one that
// created in the for of loop.
let iter = seg.segment(text);
let prev = 0;
for (const v of seg.segment(text)) {
assert(["sep", "term"].includes(v.breakType), v.breakType);
assert.sameValue("string", typeof v.segment);
assert(v.segment.length > 0);
segments.push(v.segment);
// manually advance the iter.
assert.sameValue(iter.following(), false);
assert.sameValue(iter.breakType, v.breakType);
assert.sameValue(text.substring(prev, iter.index), v.segment);
prev = iter.index;
}
assert(iter.following());
assert.sameValue(text, segments.join(''));
}