fix(corpus): return object instead of array, make second argument opt…

…ional
voyanttools · Jan 16, 2025 · 75c3495 · 75c3495
1 parent 436a8b2
commit 75c3495
Show file tree

Hide file tree

Showing 4 changed files with 361 additions and 174 deletions.
diff --git a/src/corpus.js b/src/corpus.js
@@ -1328,20 +1328,41 @@ class Corpus {
 	}
 
 	/**
-	 * Returns a list of corpus terms, filtered by the provided category.
+	 * Given a Categories instance or ID, returns an object mapping category names to corpus terms. The results can be limited to specific category names by providing one or more of them.
 	 * @param {String|Spyral.Categories} categories A categories ID or a Spyral.Categories instance.
-	 * @param {String} categoryName The name of the category within the instance.
-	 * @returns {Promise<Array>}
+	 * @param {String|Array<String>} [categoryName] One or more names of categories within the instance.
+	 * @returns {Promise<Object>}
 	 */
 	async filterByCategory(categories, categoryName) {
 		if (categories === undefined) return;
-		if (categoryName === undefined) return;
 
 		if (categories instanceof Categories === false) {
 			categories = await Categories.load(categories);
 		}
-		const catTerms = categories.getCategoryTerms(categoryName);
-		return this.terms({whiteList: catTerms});
+
+		let categoryNames = [];
+
+		if (categoryName === undefined) {
+			categoryNames = categories.getCategoryNames();
+		} else if (Util.isString(categoryName)) {
+			categoryNames = [categoryName];
+		} else {
+			categoryNames = categoryName;
+		}
+
+		const termsResults = await Promise.all(
+			categoryNames.map(key => {
+				let catTerms = categories.getCategoryTerms(key);
+				return this.terms({whiteList: catTerms});
+			})
+		);
+
+		let results = {};
+		termsResults.forEach((terms, i) => {
+			results[categoryNames[i]] = terms;
+		});
+
+		return results;
 	}
 
 	/**

diff --git a/test/corpus.js b/test/corpus.js
@@ -193,9 +193,15 @@ test('filterByCategory', async () => {
 	const corpus = await Corpus.load(corpusId);
 	fetch.once(JSON.stringify(MocksCategories.Categories));
 	const categories = await Categories.load('categories.en.txt');
-	fetch.once(JSON.stringify(Mocks.TermsWhiteList));
+	fetch.once(JSON.stringify(MocksCategories.TermsWhiteListPositive));
 	const data = await corpus.filterByCategory(categories, 'positive');
-	expect(data.length).toBe(22);
+	expect(data.positive.length).toBe(22);
+	fetch
+		.once(JSON.stringify(MocksCategories.Categories))
+		.once(JSON.stringify(MocksCategories.TermsWhiteListPositive))
+		.once(JSON.stringify(MocksCategories.TermsWhiteListNegative));
+	const data2 = await corpus.filterByCategory('categories.en.txt');
+	expect(data2.negative.length).toBe(21);
 })
 
 test('tool', () => {

diff --git a/test/mocks/categories.js b/test/mocks/categories.js
@@ -11,3 +11,329 @@ export const Categories = {
 		}`
 	}
 }
+
+
+export const TermsWhiteListPositive = {
+	"version": "5.7",
+	"voyantVersion": "2025-01-14",
+	"voyantBuild": "",
+	"duration": 3,
+	"corpusTerms": {
+	  "total": 22,
+	  "terms": [
+		{
+		  "term": "good",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 1444,
+		  "relativeFreq": 0.0018471071,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "hope",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 649,
+		  "relativeFreq": 8.301749E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "happy",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 562,
+		  "relativeFreq": 7.1888795E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "happiness",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 396,
+		  "relativeFreq": 5.065474E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "advantage",
+		  "inDocumentsCount": 7,
+		  "rawFreq": 166,
+		  "relativeFreq": 2.1234057E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "praise",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 131,
+		  "relativeFreq": 1.6756996E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "excellent",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 129,
+		  "relativeFreq": 1.6501165E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "superior",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 128,
+		  "relativeFreq": 1.6373249E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "confidence",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 103,
+		  "relativeFreq": 1.3175349E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "safe",
+		  "inDocumentsCount": 7,
+		  "rawFreq": 80,
+		  "relativeFreq": 1.0233281E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "success",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 64,
+		  "relativeFreq": 8.1866245E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "enjoy",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 57,
+		  "relativeFreq": 7.2912124E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "wonderful",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 48,
+		  "relativeFreq": 6.139968E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "joyful",
+		  "inDocumentsCount": 6,
+		  "rawFreq": 19,
+		  "relativeFreq": 2.4304041E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "fun",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 15,
+		  "relativeFreq": 1.91874E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "amazing",
+		  "inDocumentsCount": 5,
+		  "rawFreq": 14,
+		  "relativeFreq": 1.790824E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "freedom",
+		  "inDocumentsCount": 6,
+		  "rawFreq": 13,
+		  "relativeFreq": 1.6629081E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "enthusiasm",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 9,
+		  "relativeFreq": 1.151244E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "magnificent",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 8,
+		  "relativeFreq": 1.0233281E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "win",
+		  "inDocumentsCount": 4,
+		  "rawFreq": 4,
+		  "relativeFreq": 5.1166403E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "bliss",
+		  "inDocumentsCount": 2,
+		  "rawFreq": 2,
+		  "relativeFreq": 2.5583201E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "optimistic",
+		  "inDocumentsCount": 0,
+		  "rawFreq": 0,
+		  "relativeFreq": 0.0,
+		  "comparisonRelativeFreqDifference": null
+		}
+	  ]
+	}
+  }
+
+export const TermsWhiteListNegative = {
+	"version": "5.7",
+	"voyantVersion": "2.6.0",
+	"voyantBuild": "",
+	"duration": 7,
+	"corpusTerms": {
+	  "total": 21,
+	  "terms": [
+		{
+		  "term": "bad",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 182,
+		  "relativeFreq": 2.3280713E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "concern",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 116,
+		  "relativeFreq": 1.4838256E-4,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "suffering",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 76,
+		  "relativeFreq": 9.7216165E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "suffer",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 67,
+		  "relativeFreq": 8.570372E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "sad",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 63,
+		  "relativeFreq": 8.058708E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "inferior",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 51,
+		  "relativeFreq": 6.523716E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "horror",
+		  "inDocumentsCount": 7,
+		  "rawFreq": 38,
+		  "relativeFreq": 4.8608083E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "fail",
+		  "inDocumentsCount": 7,
+		  "rawFreq": 32,
+		  "relativeFreq": 4.0933122E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "despair",
+		  "inDocumentsCount": 8,
+		  "rawFreq": 30,
+		  "relativeFreq": 3.83748E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "hesitation",
+		  "inDocumentsCount": 6,
+		  "rawFreq": 24,
+		  "relativeFreq": 3.069984E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "desperate",
+		  "inDocumentsCount": 6,
+		  "rawFreq": 17,
+		  "relativeFreq": 2.174572E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "disadvantage",
+		  "inDocumentsCount": 7,
+		  "rawFreq": 16,
+		  "relativeFreq": 2.0466561E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "terrible",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 15,
+		  "relativeFreq": 1.91874E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "forbidden",
+		  "inDocumentsCount": 4,
+		  "rawFreq": 11,
+		  "relativeFreq": 1.4070761E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "destroy",
+		  "inDocumentsCount": 6,
+		  "rawFreq": 10,
+		  "relativeFreq": 1.2791601E-5,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "sadness",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 6,
+		  "relativeFreq": 7.67496E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "failure",
+		  "inDocumentsCount": 5,
+		  "rawFreq": 6,
+		  "relativeFreq": 7.67496E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "depression",
+		  "inDocumentsCount": 3,
+		  "rawFreq": 4,
+		  "relativeFreq": 5.1166403E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "disaster",
+		  "inDocumentsCount": 1,
+		  "rawFreq": 1,
+		  "relativeFreq": 1.2791601E-6,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "pessimistic",
+		  "inDocumentsCount": 0,
+		  "rawFreq": 0,
+		  "relativeFreq": 0.0,
+		  "comparisonRelativeFreqDifference": null
+		},
+		{
+		  "term": "criticize",
+		  "inDocumentsCount": 0,
+		  "rawFreq": 0,
+		  "relativeFreq": 0.0,
+		  "comparisonRelativeFreqDifference": null
+		}
+	  ]
+	}
+  }