BREAKING. feat: sets threshold as 0 by default (#784)

oramasearch · Sep 10, 2024 · bfc484a · bfc484a
1 parent 20cd8a4
commit bfc484a
Show file tree

Hide file tree

Showing 8 changed files with 37 additions and 15 deletions.
diff --git a/packages/orama/src/components/algorithms.ts b/packages/orama/src/components/algorithms.ts
@@ -5,7 +5,7 @@ import { InternalDocumentID } from './internal-document-id-store.js'
 export function prioritizeTokenScores(
   arrays: TokenScore[][],
   boost: number,
-  threshold = 1,
+  threshold = 0,
   keywordsCount: number
 ): TokenScore[] {
   if (boost === 0) {

diff --git a/packages/orama/src/methods/search-fulltext.ts b/packages/orama/src/methods/search-fulltext.ts
@@ -34,7 +34,7 @@ export async function fullTextSearch<T extends AnyOrama, ResultDocument = TypedD
   const vectorProperties = Object.keys(orama.data.index.vectorIndexes)
 
   const shouldCalculateFacets = params.facets && Object.keys(params.facets).length > 0
-  const { limit = 10, offset = 0, term, properties, threshold = 1, distinctOn, includeVectors = false } = params
+  const { limit = 10, offset = 0, term, properties, threshold = 0, distinctOn, includeVectors = false } = params
   const isPreflight = params.preflight === true
 
   const { index, docs } = orama.data

diff --git a/packages/orama/src/methods/search-hybrid.ts b/packages/orama/src/methods/search-hybrid.ts
@@ -134,7 +134,7 @@ async function getFullTextSearchIDs<T extends AnyOrama, ResultDocument = TypedDo
   const timeStart = await getNanosecondsTime()
   params.relevance = Object.assign(defaultBM25Params, params.relevance ?? {})
 
-  const { term = '', properties, threshold = 1 } = params
+  const { term = '', properties, threshold = 0 } = params
 
   const { index, docs } = orama.data
   const tokens = await orama.tokenizer.tokenize(term, language)

diff --git a/packages/orama/src/types.ts b/packages/orama/src/types.ts
@@ -430,7 +430,7 @@ export interface SearchParamsFullText<T extends AnyOrama, ResultDocument = Typed
   /**
    * Threshold to use for refining the search results.
    * The threshold is a number between 0 and 1 that represents the minimum score of the documents to return.
-   * By default, the threshold is 1.
+   * By default, the threshold is 0.
    *
    * Full documentation: https://docs.oramasearch.com/open-source/usage/search/threshold
    *

diff --git a/packages/orama/tests/boosting.test.ts b/packages/orama/tests/boosting.test.ts
@@ -12,7 +12,7 @@ t.test('boosting', (t) => {
         id: 'string',
         title: 'string',
         description: 'string'
-      }
+      } as const
     })
 
     await insert(db, {
@@ -29,11 +29,13 @@ t.test('boosting', (t) => {
     })
 
     const { hits: hits1 } = await search(db, {
-      term: 'computer for browsing and movies'
+      term: 'computer for browsing and movies',
+      threshold: 1,
     })
 
     const { hits: hits2 } = await search(db, {
       term: 'computer for browsing and movies',
+      threshold: 1,
       boost: {
         title: 2.5
       }

diff --git a/packages/orama/tests/dataset.test.ts b/packages/orama/tests/dataset.test.ts
@@ -39,7 +39,7 @@ t.test('orama.dataset', async (t) => {
         first: 'string',
         second: 'string'
       }
-    },
+    } as const,
     sort: {
       enabled: false
     },
@@ -67,6 +67,7 @@ t.test('orama.dataset', async (t) => {
     const s1 = await search(db, {
       term: 'august',
       exact: true,
+      threshold: 1,
       properties: ['categories.first'],
       limit: 10,
       offset: 0
@@ -75,6 +76,7 @@ t.test('orama.dataset', async (t) => {
     const s2 = await search(db, {
       term: 'january, june',
       exact: true,
+      threshold: 1,
       properties: ['categories.first'],
       limit: 10,
       offset: 0
@@ -83,6 +85,7 @@ t.test('orama.dataset', async (t) => {
     const s3 = await search(db, {
       term: 'january/june',
       exact: true,
+      threshold: 1,
       properties: ['categories.first'],
       limit: 10,
       offset: 0
@@ -100,16 +103,19 @@ t.test('orama.dataset', async (t) => {
   t.test('should correctly search long strings', async (t) => {
     const s1 = await search(db, {
       term: 'e into the',
+      threshold: 1,
       properties: ['description']
     })
 
     const s2 = await search(db, {
       term: 'The Roman armies',
+      threshold: 1,
       properties: ['description']
     })
 
     const s3 = await search(db, {
       term: 'the King of Epirus, is taken',
+      threshold: 1,
       properties: ['description']
     })
 
@@ -124,6 +130,7 @@ t.test('orama.dataset', async (t) => {
     const s1 = removeVariadicData(
       await search(db, {
         term: 'war',
+        threshold: 1,
         exact: true,
         // eslint-disable-next-line
         // @ts-ignore
@@ -136,6 +143,7 @@ t.test('orama.dataset', async (t) => {
     const s2 = removeVariadicData(
       await search(db, {
         term: 'war',
+        threshold: 1,
         exact: true,
         properties: ['description'],
         limit: 10,
@@ -146,6 +154,7 @@ t.test('orama.dataset', async (t) => {
     const s3 = removeVariadicData(
       await search(db, {
         term: 'war',
+        threshold: 1,
         exact: true,
         properties: ['description'],
         limit: 10,
@@ -155,6 +164,7 @@ t.test('orama.dataset', async (t) => {
 
     const s4 = await search(db, {
       term: 'war',
+      threshold: 1,
       exact: true,
       properties: ['description'],
       limit: 2240,
@@ -163,6 +173,7 @@ t.test('orama.dataset', async (t) => {
 
     const s5 = await search(db, {
       term: 'war',
+      threshold: 1,
       exact: true,
       properties: ['description'],
       limit: 10,
@@ -204,6 +215,7 @@ t.test('orama.dataset', async (t) => {
   t.test('should correctly delete documents', async (t) => {
     const documentsToDelete = await search(db, {
       term: 'war',
+      threshold: 1,
       exact: true,
       properties: ['description'],
       limit: 10,
@@ -216,6 +228,7 @@ t.test('orama.dataset', async (t) => {
 
     const newSearch = await search(db, {
       term: 'war',
+      threshold: 1,
       exact: true,
       properties: ['description'],
       limit: 10,

diff --git a/packages/orama/tests/facets.test.ts b/packages/orama/tests/facets.test.ts
@@ -13,7 +13,7 @@ t.test('facets', (t) => {
           tag: 'string',
           isFavorite: 'boolean'
         }
-      }
+      } as const
     })
 
     await insert(db, {
@@ -73,6 +73,7 @@ t.test('facets', (t) => {
 
     const results = await search(db, {
       term: 'work time',
+      threshold: 1,
       facets: {
         'meta.isFavorite': {
           true: true,
@@ -99,7 +100,7 @@ t.test('facets', (t) => {
         name: 'string',
         price: 'number',
         category: 'string'
-      }
+      } as const
     })
 
     await insert(db, {
@@ -140,6 +141,7 @@ t.test('facets', (t) => {
 
     const results = await search(db, {
       term: 'groceries',
+      threshold: 1,
       properties: ['category'],
       facets: {
         price: {
@@ -165,7 +167,7 @@ t.test('facets', (t) => {
       schema: {
         category: 'enum',
         colors: 'enum[]'
-      }
+      } as const
     })
 
     await insertMultiple(db, [
@@ -235,10 +237,10 @@ t.test('facets', (t) => {
       schema: {
         author: 'string',
         quote: 'string'
-      }
+      } as const
     })
 
-    const quotes = []
+    const quotes: any[] = []
 
     for (let i = 0; i < orderedAuthors.length; i++) {
       quotes.push({ author: orderedAuthors[i], quote: 'Be the change you wish to see in the world' })
@@ -254,6 +256,7 @@ t.test('facets', (t) => {
 
     const results = await search(db, {
       term: 'person',
+      threshold: 1,
       facets: {
         author: {}
       }
@@ -268,6 +271,7 @@ t.test('facets', (t) => {
 
     const results = await search(db, {
       term: 'person',
+      threshold: 1,
       facets: {
         author: {
           limit: orderedAuthors.length + 1
@@ -284,6 +288,7 @@ t.test('facets', (t) => {
 
     const results = await search(db, {
       term: 'person',
+      threshold: 1,
       facets: {
         author: {
           limit: orderedAuthors.length - 1

diff --git a/packages/orama/tests/search.test.ts b/packages/orama/tests/search.test.ts
@@ -54,7 +54,8 @@ t.test('search method', (t) => {
       const docIdsShouldMatch = allDocs.filter((d) => /coffee/.test(d.document.name as string)).map((d) => d.id)
 
       const result = await search(db, {
-        term: 'coffee'
+        term: 'coffee',
+        threshold: 1,
       })
 
       const matchedIds = result.hits.map((d) => d.id)
@@ -107,8 +108,8 @@ t.test('search method', (t) => {
       t.equal(result6.count, 4)
 
       // Long string search (Tests for https://github.com/askorama/orama/issues/159 )
-      const result7 = await search(db, { term: 'They are the best' })
-      const result8 = await search(db, { term: 'Foxes are nice animals' })
+      const result7 = await search(db, { term: 'They are the best', threshold: 1 })
+      const result8 = await search(db, { term: 'Foxes are nice animals', threshold: 1 })
 
       t.equal(result7.count, 2)
       t.equal(result8.count, 2)
@@ -661,6 +662,7 @@ t.test('search method', (t) => {
 
     const result = await search(db, {
       term: '',
+      threshold: 1,
       properties: ['animal']
     })