From 63c0d300c0d14b9c0b3d8667ac9f43cf30933c8f Mon Sep 17 00:00:00 2001
From: Galen <whatisgalen@gmail.com>
Date: Wed, 14 Aug 2024 18:50:31 -0700
Subject: [PATCH] create string utils with methods for normalizing, computing
 similarity score, re #10780

---
 arches/app/media/js/utils/strings.js | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 arches/app/media/js/utils/strings.js

diff --git a/arches/app/media/js/utils/strings.js b/arches/app/media/js/utils/strings.js
new file mode 100644
index 00000000000..b268ce4aa10
--- /dev/null
+++ b/arches/app/media/js/utils/strings.js
@@ -0,0 +1,22 @@
+define(['jquery', 'knockout', 'arches'], function($, ko, arches) {
+    const stringUtils = {
+        compareTwoStrings: function(str1, str2) {
+            // uses dice coefficient for string similarity score
+            if (str1.length < 2 || str2.length < 2) return 0;
+            let set1 = new Set();
+            let set2 = new Set();
+            for (let i = 0; i < str1.length - 1; i++) {
+                const bigram = str1.substr(i, 2);
+                set1.add(bigram);
+            }
+            for (let i = 0; i < str2.length - 1; i++) {
+                const bigram = str2.substr(i, 2);
+                set2.add(bigram);
+            }
+            const intersection = new Set([...set1].filter(x => set2.has(x)));
+            return (2 * intersection.size) / (set1.size + set2.size);
+        },
+        normalizeText: function(text) { return text.toLowerCase().replace(/\W+/g, ''); },
+    };
+    return stringUtils;
+});