fix(plugin-match-highlight): support tolerance parameter in searchWit…

…hHighlight (#489)

packages/plugin-match-highlight/src/index.ts

@@ -8,6 +8,7 @@ import {
   save,
   search
 } from '@orama/orama'
+import { boundedLevenshtein } from '@orama/orama/internals';
 
 export interface Position {
   start: number
@@ -89,20 +90,39 @@ export async function searchWithHighlight<T extends AnyOrama, ResultDocument = T
 ): Promise<SearchResultWithHighlight<ResultDocument>> {
   const result = await search(orama, params, language)
   const queryTokens: string[] = await orama.tokenizer.tokenize(params.term ?? '', language)
-  const hits = result.hits.map((hit: AnyDocument) =>
-    Object.assign(hit, {
-      positions: Object.fromEntries(
-        Object.entries<any>((orama as OramaWithHighlight<T>).data.positions[hit.id]).map(([propName, tokens]) => [
-          propName,
-          Object.fromEntries(
-            Object.entries(tokens).filter(([token]) => queryTokens.find(queryToken => token.startsWith(queryToken)))
-          )
-        ])
-      )
-    })
-  )
 
-  result.hits = hits
+  let hitsWithPosition = []
+  for (const hit of result.hits) {
+    const hitPositions = Object.entries<any>((orama as OramaWithHighlight<T>).data.positions[hit.id])
+
+    let hits: AnyDocument[] = []
+    for (const [propName, tokens] of hitPositions) {
+      const matchWithSearchTokens = []
+
+      const tokenEntries = Object.entries(tokens)
+      for (const tokenEntry of tokenEntries) {
+        const [token] = tokenEntry
+
+        for (const queryToken of queryTokens) {
+          if (params.tolerance) {
+            const distance = await boundedLevenshtein(token, queryToken, params.tolerance)
+            if (distance.isBounded) {
+              matchWithSearchTokens.push(tokenEntry)
+              break
+            }
+          } else if (token.startsWith(queryToken)) {
+            matchWithSearchTokens.push(tokenEntry)
+            break
+          }
+        }
+      }
+      hits.push([propName, Object.fromEntries(matchWithSearchTokens)])
+    }
+
+    hitsWithPosition.push(Object.assign(hit, { positions: Object.fromEntries(hits) }))
+  }
+
+  result.hits = hitsWithPosition
 
   // @ts-ignore
   return result

packages/plugin-match-highlight/test/index.test.ts

@@ -68,6 +68,33 @@ t.test('should retrieve positions', async t => {
     t.same(results.hits[0].positions, { text: { hello: [{ start: 0, length: 5 }] } });
 });
 
+t.test('should retrieve positions also with typo, if tolerance is used', async t => {
+    const schema = {
+        title: 'string',
+        summary: 'string',
+        id: 'string',
+        slug: 'string'
+    } as const
+
+    const db = await create({ schema, components: { afterInsert } })
+
+    await insert(db, {
+        title: 'Introduction to React',
+        summary: 'React is a popular JavaScript library for building user interfaces, primarily for single-page applications. By utilizing a component-based architecture, it allows developers to build reusable UI components and manage the state of an application seamlessly. This introduction covers its core philosophies, JSX, and the virtual DOM.',
+        id: '1a2b3c',
+        slug: 'introduction-to-react'
+    })
+
+    const results = await searchWithHighlight(db, { term: 'reat', tolerance: 1 })
+
+    t.same(results.hits[0].positions, { 
+        title: { react: [{ start: 16, length: 5 } ] },
+        summary: { react: [ { start: 0, length: 5 } ] },
+        id: {},
+        slug: {}
+    })
+});
+
 t.test('should work with texts containing constructor and __proto__ properties', async t => {
     const schema = {
         text: 'string',