fixed cacheLength thing, isLeaf option is no longer user-definable

src/DocumentProcessor.js

@@ -25,7 +25,7 @@ export const DocumentProcessor = ops => {
 
   const processValueUnknownType = (unknown, key) =>
     // eslint-disable-next-line
-        new Promise(async resolve => {
+    new Promise(async resolve => {
       if (unknown === null) return resolve([null, '1.00'])
       if (isEmptyObject(unknown)) return resolve([[], '1.00'])
       if (Number.isInteger(unknown)) return resolve([unknown, unknown])
@@ -42,7 +42,6 @@ export const DocumentProcessor = ops => {
       if (isString(doc)) doc = { body: doc }
 
       // Docs with no _id are auto-assigned an ID
-      // if (!doc.hasOwnProperty('_id')) doc._id = ops.idGenerator.next().value
       if (!Object.prototype.hasOwnProperty.call(doc, '_id')) {
         doc._id = ops.idGenerator.next().value
       }

src/main.js

@@ -17,29 +17,31 @@ export class Main {
           yield Date.now() + '-' + i++
         }
       })(),
-      isLeaf: node =>
-        Array.isArray(node) &&
-        node.length === 2 &&
-        node.every(
-          item =>
-            typeof item === 'string' ||
-            typeof item === 'number' ||
-            item === null
-        ),
       skipFields: [],
       ngrams: {},
       replace: {},
       storeRawDocs: true,
       stopwords: [],
       storeVectors: true, // TODO: make a test for this being false
-      tokenAppend: '#',
       tokenSplitRegex: /[\p{L}\d]+/gu,
       tokenizer: tokenization.tokenizationPipeline,
       ...ops
     }
-    this.INDEX = new InvertedIndex(ops)
+    this.INDEX = new InvertedIndex({
+      ...ops,
+      // isLeaf must be like so and is not a user defined option
+      isLeaf: node =>
+        Array.isArray(node) &&
+        node.length === 2 &&
+        node.every(
+          item =>
+            typeof item === 'string' ||
+            typeof item === 'number' ||
+            item === null
+        )
+    })
     // Now that constructor is not async- not sure where this should be called...
-    this._CACHE = new LRUCache({ max: 1000 })
+    this._CACHE = new LRUCache({ max: ops.cacheLength })
     this.r = new Reader(ops, this._CACHE, this.INDEX)
     this.w = new Writer(ops, this._CACHE, this.INDEX)
     validateVersion(this.INDEX)

src/write.js

@@ -63,6 +63,7 @@ export class Writer {
   }
 
   DELETE_RAW (...docIds) {
+    // TODO: should we be using ops.docExistsSpace instead of DOC_RAW?
     return Promise.all(docIds.map(id => this.#ii.STORE.del(['DOC_RAW', id])))
   }
 

test/src/CACHE-test.js

@@ -6,12 +6,85 @@ const sandbox = 'test/sandbox/'
 const indexName = sandbox + 'CACHE'
 const global = {}
 
-test('create a search index', async t => {
+const data = [
+  {
+    _id: 0,
+    make: 'Tesla',
+    manufacturer: 'Volvo',
+    brand: 'Volvo',
+    colour: 'yellow'
+  },
+  {
+    _id: 1,
+    make: 'BMW',
+    manufacturer: 'Volvo',
+    brand: 'Volvo',
+    colour: 'red'
+  },
+  {
+    _id: 2,
+    make: 'Tesla',
+    manufacturer: 'Tesla',
+    brand: 'Volvo',
+    colour: 'blue'
+  },
+  {
+    _id: 3,
+    make: 'Tesla',
+    manufacturer: 'Volvo',
+    brand: 'BMW',
+    colour: 'red'
+  },
+  {
+    _id: 4,
+    make: 'Volvo',
+    manufacturer: 'Volvo',
+    brand: 'Volvo',
+    colour: 'red'
+  },
+  {
+    _id: 5,
+    make: 'Volvo',
+    manufacturer: 'Tesla',
+    brand: 'Volvo',
+    colour: 'blue'
+  },
+  {
+    _id: 6,
+    make: 'Tesla',
+    manufacturer: 'Tesla',
+    brand: 'BMW',
+    colour: 'yellow'
+  },
+  {
+    _id: 7,
+    make: 'BMW',
+    manufacturer: 'Tesla',
+    brand: 'Tesla',
+    colour: 'yellow'
+  },
+  {
+    _id: 8,
+    make: 'Volvo',
+    manufacturer: 'BMW',
+    brand: 'Tesla',
+    colour: 'blue'
+  },
+  {
+    _id: 9,
+    make: 'BMW',
+    manufacturer: 'Tesla',
+    brand: 'Volvo',
+    colour: 'red'
+  }
+]
+
+test('create a search index', t => {
   t.plan(1)
   try {
-    global[indexName] = await new SearchIndex({
-      name: indexName,
-      cacheLength: 5 // TODO: this needs to be updated
+    global[indexName] = new SearchIndex({
+      name: indexName
+      //      cacheLength: 5 // TODO: this needs to be updated
     })
     t.ok(global[indexName])
   } catch (e) {
@@ -20,79 +93,6 @@ test('create a search index', async t => {
 })
 
 test('can add data', t => {
-  const data = [
-    {
-      _id: 0,
-      make: 'Tesla',
-      manufacturer: 'Volvo',
-      brand: 'Volvo',
-      colour: 'yellow'
-    },
-    {
-      _id: 1,
-      make: 'BMW',
-      manufacturer: 'Volvo',
-      brand: 'Volvo',
-      colour: 'red'
-    },
-    {
-      _id: 2,
-      make: 'Tesla',
-      manufacturer: 'Tesla',
-      brand: 'Volvo',
-      colour: 'blue'
-    },
-    {
-      _id: 3,
-      make: 'Tesla',
-      manufacturer: 'Volvo',
-      brand: 'BMW',
-      colour: 'red'
-    },
-    {
-      _id: 4,
-      make: 'Volvo',
-      manufacturer: 'Volvo',
-      brand: 'Volvo',
-      colour: 'red'
-    },
-    {
-      _id: 5,
-      make: 'Volvo',
-      manufacturer: 'Tesla',
-      brand: 'Volvo',
-      colour: 'blue'
-    },
-    {
-      _id: 6,
-      make: 'Tesla',
-      manufacturer: 'Tesla',
-      brand: 'BMW',
-      colour: 'yellow'
-    },
-    {
-      _id: 7,
-      make: 'BMW',
-      manufacturer: 'Tesla',
-      brand: 'Tesla',
-      colour: 'yellow'
-    },
-    {
-      _id: 8,
-      make: 'Volvo',
-      manufacturer: 'BMW',
-      brand: 'Tesla',
-      colour: 'blue'
-    },
-    {
-      _id: 9,
-      make: 'BMW',
-      manufacturer: 'Tesla',
-      brand: 'Volvo',
-      colour: 'red'
-    }
-  ]
-
   t.plan(1)
   global[indexName].PUT(data).then(t.pass)
 })
@@ -319,3 +319,51 @@ test('cache is filling up again', t => {
     '{"QUERY":["boooom",null]}'
   )
 })
+
+test('test cacheLength', t => {
+  t.plan(9)
+  const cacheLengthIndex = new SearchIndex({
+    name: sandbox + 'cacheLength-test',
+    cacheLength: 5
+  })
+  t.ok(cacheLengthIndex)
+  cacheLengthIndex
+    .PUT(data)
+    .then(t.ok)
+    .then(() => cacheLengthIndex.QUERY('one'))
+    .then(res => cacheLengthIndex.QUERY('two'))
+    .then(res => cacheLengthIndex.QUERY('two'))
+    .then(res => cacheLengthIndex.QUERY('two'))
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.DICTIONARY())
+    .then(res => cacheLengthIndex.DOCUMENTS())
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.SEARCH('tesla'))
+    .then(res => cacheLengthIndex.SEARCH('tesla'))
+    .then(res => cacheLengthIndex.SEARCH(['tesla']))
+    .then(res => cacheLengthIndex.QUERY('three'))
+    .then(res => cacheLengthIndex.QUERY('four'))
+    .then(res => t.pass('done'))
+    .then(() => {
+      const keys = [
+        '{"QUERY":["four",null]}',
+        '{"QUERY":["three",null]}',
+        '{"SEARCH":[["tesla"],null]}',
+        '{"SEARCH":["tesla",null]}',
+        '{"DOCUMENTS":[]}'
+      ]
+      // cache only has 5 entries since cacheLength:5
+      let cacheSize = 0
+      const cKeys = cacheLengthIndex._CACHE.keys()
+      keys.forEach(item => {
+        cacheSize++
+        const cKeysNext = cKeys.next().value
+        t.deepEquals(cKeysNext, item)
+      })
+      t.equal(cacheSize, 5)
+    })
+    .catch(t.error)
+})