wip

provider/docs/src/corpus/search/embeddings.test.ts

@@ -1,14 +1,15 @@
 import { describe, expect, test } from 'vitest'
 import { indexCorpus, type CorpusSearchResult } from '..'
+import { noopCache } from '../cache/cache'
 import { corpusData } from '../data'
 import { doc } from '../index.test'
 import { embeddingsSearch, embedTextInThisScope, similarity } from './embeddings'
 
 describe('embeddingsSearch', () => {
     test('finds matches', async () => {
-        expect(await embeddingsSearch(await indexCorpus(corpusData([doc(1, 'a'), doc(2, 'b')])), 'b')).toEqual<
-            CorpusSearchResult[]
-        >([{ doc: 2, chunk: 0, score: 1, excerpt: 'b' }])
+        expect(
+            await embeddingsSearch(await indexCorpus(corpusData([doc(1, 'a'), doc(2, 'b')])), 'b', { cache: noopCache })
+        ).toEqual<CorpusSearchResult[]>([{ doc: 2, chunk: 0, score: 1, excerpt: 'b' }])
     })
 })
 

provider/docs/src/corpus/search/keyword.test.ts

@@ -1,15 +1,18 @@
 import { describe, expect, test } from 'vitest'
 import { indexCorpus, type CorpusSearchResult } from '..'
+import { noopCache } from '../cache/cache'
 import { corpusData } from '../data'
 import { doc } from '../index.test'
 import { keywordSearch } from './keyword'
 import { calculateTFIDF } from './tfidf'
 
 describe('keywordSearch', () => {
     test('finds matches', async () => {
-        expect(keywordSearch(await indexCorpus(corpusData([doc(1, 'aaa'), doc(2, 'bbb')])), 'bbb')).toEqual<
-            CorpusSearchResult[]
-        >([
+        expect(
+            await keywordSearch(await indexCorpus(corpusData([doc(1, 'aaa'), doc(2, 'bbb')])), 'bbb', {
+                cache: noopCache,
+            })
+        ).toEqual<CorpusSearchResult[]>([
             {
                 doc: 2,
                 chunk: 0,

provider/docs/src/corpus/search/tfidf.test.ts

@@ -1,35 +1,35 @@
 import { describe, expect, test } from 'vitest'
 import { indexCorpus } from '..'
 import { corpusData } from '../data'
-import { calculateTFIDF, createTFIDFIndex } from './tfidf'
+import { calculateTFIDF, computeTFIDF, createTFIDFIndex } from './tfidf'
 
-describe('createIndexForTFIDF', async () => {
+describe('createTFIDFIndex', async () => {
     const data = corpusData([
         { id: 1, text: 'a b c c c' },
         { id: 2, text: 'b c d' },
         { id: 3, text: 'c d e' },
     ])
     const docIDs = data.docs.map(({ id }) => id)
     const index = await indexCorpus(data)
-    const tfidf = createTFIDFIndex(index.docs)
+    const tfidfIndex = createTFIDFIndex(index.docs)
 
     test('term in 1 doc', () => {
-        expect(docIDs.map(docID => tfidf('a', docID, 0))).toEqual([
+        expect(docIDs.map(docID => computeTFIDF('a', docID, 0, tfidfIndex))).toEqual([
             calculateTFIDF({ termOccurrencesInChunk: 1, chunkTermLength: 5, totalChunks: 3, termChunkFrequency: 1 }),
             0,
             0,
         ])
     })
 
     test('term in all docs', () => {
-        expect(docIDs.map(docID => tfidf('c', docID, 0))).toEqual([
+        expect(docIDs.map(docID => computeTFIDF('c', docID, 0, tfidfIndex))).toEqual([
             calculateTFIDF({ termOccurrencesInChunk: 3, chunkTermLength: 5, totalChunks: 3, termChunkFrequency: 3 }),
             calculateTFIDF({ termOccurrencesInChunk: 1, chunkTermLength: 3, totalChunks: 3, termChunkFrequency: 3 }),
             calculateTFIDF({ termOccurrencesInChunk: 1, chunkTermLength: 3, totalChunks: 3, termChunkFrequency: 3 }),
         ])
     })
 
     test('unknown term', () => {
-        expect(docIDs.map(docID => tfidf('x', docID, 0))).toEqual([0, 0, 0])
+        expect(docIDs.map(docID => computeTFIDF('x', docID, 0, tfidfIndex))).toEqual([0, 0, 0])
     })
 })