Converts the sentence beginnings assessment to use the HTML parser

Yoast · Jan 17, 2025 · 79e772f · 79e772f
1 parent f7f95dc
commit 79e772f
Show file tree

Hide file tree

Showing 12 changed files with 290 additions and 195 deletions.
diff --git a/packages/yoastseo/spec/fullTextTests/testTexts/sk/slovakPaper.js b/packages/yoastseo/spec/fullTextTests/testTexts/sk/slovakPaper.js
@@ -136,9 +136,8 @@ const expectedResults = {
 	},
 	sentenceBeginnings: {
 		isApplicable: true,
-		score: 3,
-		resultText: "<a href='https://yoa.st/35f' target='_blank'>Consecutive sentences</a>: The text contains 3 consecutive sentences " +
-			"starting with the same word. <a href='https://yoa.st/35g' target='_blank'>Try to mix things up</a>!",
+		score: 9,
+		resultText: "<a href='https://yoa.st/35f' target='_blank'>Consecutive sentences</a>: There is enough variety in your sentences. That's great!",
 	},
 	imageKeyphrase: {
 		isApplicable: true,

diff --git a/packages/yoastseo/spec/languageProcessing/helpers/sentence/getSentencesFromTreeSpec.js b/packages/yoastseo/spec/languageProcessing/helpers/sentence/getSentencesFromTreeSpec.js
@@ -1,7 +1,9 @@
+import { cloneDeep } from "lodash";
 import Paper from "../../../../src/values/Paper";
 import getSentencesFromTree from "../../../../src/languageProcessing/helpers/sentence/getSentencesFromTree";
 import buildTree from "../../../specHelpers/parse/buildTree";
 import EnglishResearcher from "../../../../src/languageProcessing/languages/en/Researcher";
+import filterTree from "../../../../src/parse/build/private/filterTree";
 
 describe( "test to get sentences from the tree", () => {
 	let researcher;
@@ -14,7 +16,7 @@ describe( "test to get sentences from the tree", () => {
 			"</h3><h4>A subheading 4</h4></div>" );
 		researcher.setPaper( paper );
 		buildTree( paper, researcher );
-		expect( getSentencesFromTree( paper ) ).toEqual( [
+		expect( getSentencesFromTree( paper.getTree() ) ).toEqual( [
 			{
 				sourceCodeRange: { startOffset: 8, endOffset: 49 },
 				parentStartOffset: 8,
@@ -98,7 +100,7 @@ describe( "test to get sentences from the tree", () => {
 			" width=\"200\" height=\"300\" /> elephant[/caption]", { shortcodes: [ "caption" ] } );
 		researcher.setPaper( paper );
 		buildTree( paper, researcher );
-		expect( getSentencesFromTree( paper ) ).toEqual( [
+		expect( getSentencesFromTree( paper.getTree() ) ).toEqual( [
 			{
 				isParentFirstSectionOfBlock: false,
 				parentAttributeId: "",
@@ -154,7 +156,7 @@ describe( "test to get sentences from the tree", () => {
 			" is a small mammal native to the eastern Himalayas and southwestern China</blockquote>" );
 		researcher.setPaper( paper );
 		buildTree( paper, researcher );
-		expect( getSentencesFromTree( paper ) ).toEqual( [
+		expect( getSentencesFromTree( paper.getTree() ) ).toEqual( [
 			{ sourceCodeRange: { startOffset: 3, endOffset: 19 },
 				text: "A cute red panda",
 				tokens: [
@@ -183,7 +185,7 @@ describe( "test to get sentences from the tree", () => {
 			"is this giant panda</p> </div> </div>" );
 		researcher.setPaper( paper );
 		buildTree( paper, researcher );
-		expect( getSentencesFromTree( paper ) ).toEqual(
+		expect( getSentencesFromTree( paper.getTree() ) ).toEqual(
 			[
 				{
 					text: "What is giant panda",
@@ -285,6 +287,17 @@ describe( "test to get sentences from the tree", () => {
 			" is a small mammal native to the eastern Himalayas and southwestern China</blockquote>" );
 		researcher.setPaper( paper );
 		buildTree( paper, researcher );
-		expect( getSentencesFromTree( paper ) ).toEqual( [] );
+		expect( getSentencesFromTree( paper.getTree() ) ).toEqual( [] );
+	} );
+	it( "should return empty array if all sentences appear in filtered nodes", () => {
+		const paper = new Paper( "<ol><li>This is a test.</li><li>This is another test.</li></ol>" );
+		researcher.setPaper( paper );
+		buildTree( paper, researcher );
+		expect( getSentencesFromTree( paper.getTree() ).length ).toEqual( 2 );
+
+		let clonedTree = cloneDeep( paper.getTree() );
+		const additionalFilters = [ ( node ) => node.name === "ol" ];
+		clonedTree = filterTree( clonedTree, additionalFilters );
+		expect( getSentencesFromTree( clonedTree ).length ).toEqual( 0 );
 	} );
 } );
diff --git a/packages/yoastseo/spec/languageProcessing/helpers/sentence/sentencesLengthSpec.js b/packages/yoastseo/spec/languageProcessing/helpers/sentence/sentencesLengthSpec.js
@@ -11,7 +11,7 @@ describe( "A test to count sentence lengths.", function() {
 		const mockResearcher = new EnglishResearcher( mockPaper );
 		buildTree( mockPaper, mockResearcher );
 
-		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper ), mockResearcher );
+		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper.getTree() ), mockResearcher );
 
 		expect( sentenceLengths.length ).toEqual( 1 );
 		expect( sentenceLengths[ 0 ].sentenceLength ).toEqual( 2 );
@@ -24,7 +24,7 @@ describe( "A test to count sentence lengths.", function() {
 		const mockResearcher = new EnglishResearcher( mockPaper );
 		buildTree( mockPaper, mockResearcher );
 
-		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper ), mockResearcher );
+		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper.getTree() ), mockResearcher );
 
 		expect( sentenceLengths.length ).toEqual( 2 );
 		expect( sentenceLengths[ 0 ].sentenceLength ).toEqual( 3 );
@@ -40,7 +40,7 @@ describe( "A test to count sentence lengths.", function() {
 		const mockResearcher = new EnglishResearcher( mockPaper );
 		buildTree( mockPaper, mockResearcher );
 
-		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper ), mockResearcher );
+		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper.getTree() ), mockResearcher );
 
 		expect( sentenceLengths.length ).toEqual( 2 );
 		expect( sentenceLengths[ 0 ].sentenceLength ).toEqual( 7 );
@@ -53,7 +53,7 @@ describe( "A test to count sentence lengths.", function() {
 		const mockResearcher = new EnglishResearcher( mockPaper );
 		buildTree( mockPaper, mockResearcher );
 
-		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper ), mockResearcher );
+		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper.getTree() ), mockResearcher );
 
 		expect( sentenceLengths.length ).toEqual( 2 );
 		expect( sentenceLengths[ 0 ].sentenceLength ).toEqual( 3 );
@@ -70,7 +70,7 @@ describe( "A test to count sentence lengths.", function() {
 		const mockJapaneseResearcher = new JapaneseResearcher( mockPaper );
 		buildTree( mockPaper, mockJapaneseResearcher );
 
-		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper ), mockJapaneseResearcher );
+		const sentenceLengths = sentencesLength( getSentencesFromTree( mockPaper.getTree() ), mockJapaneseResearcher );
 
 		expect( sentenceLengths.length ).toEqual( 2 );
 		expect( sentenceLengths[ 0 ].sentenceLength ).toEqual( 15 );