fix: do not break subsequent exclamation points or question marks in …

…`nlp/sentencize` PR-URL: #5380 Closes: #3013 Closes: stdlib-js/metr-issue-tracker#1 Reviewed-by: Athan Reines <[email protected]>
stdlib-js · Feb 22, 2025 · 0830450 · 0830450 · stdlib-bot · Feb 22, 2025
1 parent ecfa20e
commit 0830450
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 1 deletion.
diff --git a/lib/node_modules/@stdlib/nlp/sentencize/lib/main.js b/lib/node_modules/@stdlib/nlp/sentencize/lib/main.js
@@ -68,7 +68,8 @@ function isEndOfSentence( tokens, i ) {
 	if (
 		( token === '!' || token === '?' ) &&
 		!RE_PREFIXES.test( tokens[ im1 ] ) &&
-		!RE_SUFFIXES.test( tokens[ ip1 ] )
+		!RE_SUFFIXES.test( tokens[ ip1 ] ) &&
+		( tokens[ ip1 ] !== '!' && tokens[ ip1 ] !== '?' )
 	) {
 		return true;
 	}

diff --git a/lib/node_modules/@stdlib/nlp/sentencize/test/test.js b/lib/node_modules/@stdlib/nlp/sentencize/test/test.js
@@ -289,6 +289,28 @@ tape( 'the function splits a string into an array of sentences (unfinished last
 	t.end();
 });
 
+tape( 'the function splits a string into an array of sentences (multiple punctuation marks)', function test( t ) {
+	var expected;
+	var actual;
+	var str;
+
+	str = 'HAPPY BIRTHDAY!!! Have an awesome day!';
+	expected = [ 'HAPPY BIRTHDAY!!!', 'Have an awesome day!' ];
+	actual = sentencize( str );
+	t.deepEqual( actual, expected, 'returns an array of sentences' );
+
+	str = 'What?? How can that be??';
+	expected = [ 'What??', 'How can that be??' ];
+	actual = sentencize( str );
+	t.deepEqual( actual, expected, 'returns an array of sentences' );
+
+	str = 'How dare you!?!';
+	expected = [ 'How dare you!?!' ];
+	actual = sentencize( str );
+	t.deepEqual( actual, expected, 'returns an array of sentences' );
+	t.end();
+});
+
 tape( 'the function returns an empty array if provided an empty string', function test( t ) {
 	var out = sentencize( '' );
 	t.equal( isArray( out ), true, 'returns an array' );