From eab2768f79b7b417061b1d75631c6f3d754e1318 Mon Sep 17 00:00:00 2001 From: Henschi Date: Fri, 4 Apr 2025 16:32:54 +0200 Subject: adapt test to commit 0689f4f0ae2907c35c080ea15b5ad406ea9402e7 --- test/java/net/yacy/document/WordTokenizerTest.java | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'test') diff --git a/test/java/net/yacy/document/WordTokenizerTest.java b/test/java/net/yacy/document/WordTokenizerTest.java index 7f4250953..4ca2bf147 100644 --- a/test/java/net/yacy/document/WordTokenizerTest.java +++ b/test/java/net/yacy/document/WordTokenizerTest.java @@ -3,6 +3,9 @@ package net.yacy.document; import org.junit.Test; import static org.junit.Assert.*; +import org.hamcrest.CoreMatchers; +import org.hamcrest.MatcherAssert; + public class WordTokenizerTest { /** @@ -13,7 +16,8 @@ public class WordTokenizerTest { // test sentences containing 10x the word "word" String[] testTxtArr = new String[]{ " word word..... (word) [word] . 'word word' \"word word\" word ? word! ", - "word-word word . word.word@word.word ....word... word,word " + "word-word word . word.word@word.word ....word... word,word word word" + // !! 12x the word "word" because "word-word" and "word,word" is one word }; for (String testTxt : testTxtArr) { @@ -23,7 +27,10 @@ public class WordTokenizerTest { while (wt.hasMoreElements()) { StringBuilder sb = wt.nextElement(); if (sb.length() > 1) { // skip punktuation - assertEquals("word", sb.toString()); + MatcherAssert.assertThat(sb.toString(), + CoreMatchers.either(CoreMatchers.is("word")) + .or(CoreMatchers.is("word-word")) + .or(CoreMatchers.is("word,word"))); cnt++; } else { assertTrue("punktuation", SentenceReader.punctuation(sb.charAt(0))); -- cgit v1.2.3