fix(lyra): fixes duplicated results when search term appears on multi…

…ple properties
askorama · Aug 1, 2022 · f813865 · f813865
1 parent 137a97f
commit f813865
Show file tree

Hide file tree

Showing 2 changed files with 48 additions and 7 deletions.
diff --git a/packages/lyra/src/lyra.ts b/packages/lyra/src/lyra.ts
@@ -282,11 +282,11 @@ export function search<T extends PropertiesSchema>(
 
   const tokens = tokenize(params.term, language);
   const indices = getIndices(lyra, params.properties);
+  const uniqueDocIds = new Set<string>();
   const { limit = 10, offset = 0, exact = false } = params;
   const results: RetrievedDoc<T>[] = Array.from({
     length: limit,
   });
-  let count = 0;
 
   const timeStart = getNanosecondsTime();
 
@@ -297,13 +297,15 @@ export function search<T extends PropertiesSchema>(
     for (const index of indices) {
       const documentIDs = getDocumentIDsFromSearch(lyra, { ...params, index, term, exact });
 
-      count += documentIDs.length;
+      for (const id of documentIDs) {
+        uniqueDocIds.add(id);
+      }
 
       if (i >= limit) {
         break;
       }
 
-      for (const id of documentIDs) {
+      for (const id of uniqueDocIds) {
         if (j < offset) {
           j++;
           continue;
@@ -313,8 +315,10 @@ export function search<T extends PropertiesSchema>(
           break;
         }
 
-        const fullDoc = lyra.docs[id]!;
-        results[i] = { id, ...fullDoc };
+        if (results.findIndex(x => x?.id === id) === -1) {
+          const fullDoc = lyra.docs[id]!;
+          results[i] = { id, ...fullDoc };
+        }
         i++;
       }
     }
@@ -325,7 +329,7 @@ export function search<T extends PropertiesSchema>(
   return {
     elapsed: getNanosecondsTime() - timeStart,
     hits,
-    count,
+    count: uniqueDocIds.size,
   };
 }
 

diff --git a/packages/lyra/tests/lyra.test.ts b/packages/lyra/tests/lyra.test.ts
@@ -97,7 +97,44 @@ t.test("checkInsertDocSchema", t => {
 });
 
 t.test("lyra", t => {
-  t.plan(13);
+  t.plan(14);
+
+  t.test("should correctly search for data", t => {
+    t.plan(6);
+
+    const db = create({
+      schema: {
+        quote: "string",
+        author: "string",
+      },
+    });
+
+    insert(db, { quote: "the quick, brown fox jumps over the lazy dog. What a fox!", author: "John Doe" });
+    insert(db, { quote: "Foxes are nice animals. But I prefer having a dog.", author: "John Doe" });
+    insert(db, { quote: "I like dogs. They are the best.", author: "Jane Doe" });
+    insert(db, { quote: "I like cats. They are the best.", author: "Jane Doe" });
+
+    // Exact search
+    const result1 = search(db, { term: "fox", exact: true });
+    const result2 = search(db, { term: "dog", exact: true });
+
+    t.equal(result1.count, 1);
+    t.equal(result2.count, 2);
+
+    // Prefix search
+    const result3 = search(db, { term: "fox", exact: false });
+    const result4 = search(db, { term: "dog", exact: false });
+
+    t.equal(result3.count, 2);
+    t.equal(result4.count, 3);
+
+    // Typo-tolerant search
+    const result5 = search(db, { term: "fx", tolerance: 1 });
+    const result6 = search(db, { term: "dg", tolerance: 2 });
+
+    t.equal(result5.count, 1);
+    t.equal(result6.count, 4);
+  });
 
   t.test("should correctly insert and retrieve data", t => {
     t.plan(4);