feat(util): excerpt string

paulrobertlloyd · paulrobertlloyd · commit 8caecf6544f6 · 2024-12-22T01:08:59.000Z
diff --git a/packages/util/index.js b/packages/util/index.js
@@ -10,6 +10,13 @@ export {
 } from "./lib/date.js";
 export { sanitise } from "./lib/object.js";
 export { ISO_6709_RE } from "./lib/regex.js";
-export { md5, randomString, sha1, slugify, supplant } from "./lib/string.js";
+export {
+  excerpt,
+  md5,
+  randomString,
+  sha1,
+  slugify,
+  supplant,
+} from "./lib/string.js";
 export { getCanonicalUrl, isSameOrigin } from "./lib/url.js";
 export { isRequired } from "./lib/validation-schema.js";
diff --git a/packages/util/lib/string.js b/packages/util/lib/string.js
@@ -2,6 +2,36 @@ import { randomBytes, createHash } from "node:crypto";
 
 import slugifyString from "@sindresorhus/slugify";
 
+/**
+ * Excerpt a string
+ * @param {string} string - String to excerpt
+ * @param {number} value - Maximum number of words
+ * @param {string} locale - Locale
+ * @returns {string} Excerpted string
+ */
+export const excerpt = (string, value = 100, locale = "en") => {
+  const segmenter = new Intl.Segmenter(locale, { granularity: "word" });
+  const segments = segmenter.segment(string);
+
+  let excerpt = "";
+  let n = 0;
+  let words = 0;
+  [...segments].map((segment) => {
+    words = segment.isWordLike ? n++ : n;
+    if (words < value) {
+      excerpt += segment.segment;
+    }
+
+    return excerpt;
+  });
+
+  if (words > value) {
+    excerpt += "…";
+  }
+
+  return excerpt;
+};
+
 /**
  * Generate MD5 hashed string
  * @param {string} string - String
diff --git a/packages/util/test/unit/string.js b/packages/util/test/unit/string.js
@@ -3,6 +3,7 @@ import { Buffer } from "node:buffer";
 import { describe, it } from "node:test";
 
 import {
+  excerpt,
   md5,
   randomString,
   sha1,
@@ -11,6 +12,11 @@ import {
 } from "../../lib/string.js";
 
 describe("util/lib/string", () => {
+  it("Excerpts a string", () => {
+    assert.equal(excerpt("Well, well, well!", 1), "Well…");
+    assert.equal(excerpt("Well, indeed."), "Well, indeed.");
+  });
+
   it("Generates MD5 hashed string", () => {
     const result = md5("foo");
 
