fix: improve fuzzy search sorting (#358)

* fix: improve fuzzy search sorting * test: fix tests
decentraland · Oct 5, 2023 · 4d3d1e6 · 4d3d1e6
1 parent 280c23b
commit 4d3d1e6
Show file tree

Hide file tree

Showing 4 changed files with 26 additions and 19 deletions.
diff --git a/src/ports/catalog/component.ts b/src/ports/catalog/component.ts
@@ -46,15 +46,19 @@ export function createCatalogComponent(options: {
         {}
       )
       if (filters.search) {
+        const filteredItems = []
         for (const schema of Object.values(reducedSchemas)) {
-          const filteredItemsById = await client.query<CollectionsItemDBResult>(
-            getItemIdsBySearchTextQuery(schema, filters)
-          )
-          filters.ids = [
-            ...(filters.ids ?? []),
-            ...filteredItemsById.rows.map(({ id }) => id),
-          ]
+          const filteredItemsById = await client.query<{
+            id: string
+            similarity: number
+          }>(getItemIdsBySearchTextQuery(schema, filters))
+          filteredItems.push(...filteredItemsById.rows)
         }
+        filteredItems?.sort((a, b) => b.similarity - a.similarity)
+        filters.ids = [
+          ...(filters.ids ?? []),
+          ...filteredItems.map(({ id }) => id),
+        ]
 
         if (filters.ids?.length === 0) {
           // if no items matched the search text, return empty result

diff --git a/src/ports/catalog/queries.ts b/src/ports/catalog/queries.ts
@@ -459,7 +459,10 @@ export const getItemIdsBySearchTextQuery = (
   filters: CatalogQueryFilters
 ) => {
   const { category, search, limit, offset } = filters
-  const query = SQL`SELECT items.id`
+  const query = SQL`SELECT 
+      items.id, 
+      GREATEST(similarity(word, ${search}), similarity(metadata_wearable.name, ${search}), 
+      similarity(metadata_emote.name, ${search})) as similarity`
     .append(` FROM `)
     .append(schemaVersion)
     .append(`.item_active AS items `)
@@ -469,7 +472,7 @@ export const getItemIdsBySearchTextQuery = (
     .append(getSearchWhere({ search, category }))
     .append(
       category
-        ? SQL` ORDER BY GREATEST(similarity(word, ${search})) DESC`
+        ? SQL` ORDER BY GREATEST(similarity(word, ${search}), similarity(metadata_wearable.name, ${search}), similarity(metadata_emote.name, ${search})) DESC`
         : SQL` ORDER BY GREATEST(similarity(word_wearable, ${search}), similarity(word_emote, ${search})) DESC`
     )
     .append(SQL` LIMIT ${limit} OFFSET ${offset};`)

diff --git a/src/tests/ports/catalog-queries.spec.ts b/src/tests/ports/catalog-queries.spec.ts
@@ -526,10 +526,10 @@ test('catalog utils', () => {
           expect(query.text).toContain(
             `LEFT JOIN LATERAL unnest(string_to_array(metadata_emote.name, ' ')) AS word_emote ON TRUE `
           )
-          expect(query.text).toContain(`word_wearable % $1 OR word_emote % $2 `)
+          expect(query.text).toContain(`word_wearable % $4 OR word_emote % $5 `)
           // it appears four times `JOIN LATERAL unnest(string_to_array(metadata_emote.name, ' ')) AS word ON TRUE WHERE word % $1 ORDER BY GREATEST(similarity(word, $2))`
           expect(query.values).toStrictEqual([
-            ...Array(4).fill(search),
+            ...Array(7).fill(search),
             limit,
             offset,
           ])
@@ -550,10 +550,10 @@ test('catalog utils', () => {
           expect(query.text).toContain(
             `JOIN LATERAL unnest(string_to_array(metadata_wearable.name, ' ')) AS word ON TRUE `
           )
-          expect(query.text).toContain(`word % $1 `)
+          expect(query.text).toContain(`word % $4 `)
           // it appears twice `JOIN LATERAL unnest(string_to_array(metadata_wearable.name, ' ')) AS word ON TRUE WHERE word % $1 ORDER BY GREATEST(similarity(word, $2))`
           expect(query.values).toStrictEqual([
-            ...Array(2).fill(search),
+            ...Array(7).fill(search),
             limit,
             offset,
           ])
@@ -574,10 +574,10 @@ test('catalog utils', () => {
           expect(query.text).toContain(
             `JOIN LATERAL unnest(string_to_array(metadata_emote.name, ' ')) AS word ON TRUE `
           )
-          expect(query.text).toContain(`word % $1 `)
-          // it appears twice `JOIN LATERAL unnest(string_to_array(metadata_emote.name, ' ')) AS word ON TRUE WHERE word % $1 ORDER BY GREATEST(similarity(word, $2))`
+          expect(query.text).toContain(`word % $4 `)
+          // it appears twice `JOIN LATERAL unnest(string_to_array(metadata_emote.name, ' ')) AS word ON TRUE WHERE word % $1 ORDER BY GREATEST(similarity(word, $2), similarity(metadata_wearable.name, $3), similarity(metadata_emote.name, $4))`
           expect(query.values).toStrictEqual([
-            ...Array(2).fill(search),
+            ...Array(7).fill(search),
             limit,
             offset,
           ])

diff --git a/src/tests/ports/catalog.spec.ts b/src/tests/ports/catalog.spec.ts
@@ -335,9 +335,9 @@ test('catalog component', function () {
           expect(dbClientQueryMock.mock.calls[1][0]).toEqual(
             itemIdsBySearchTextQuery
           )
-          // It's repeated 4 times due to this WHERE statement: `WHERE word_wearable % $1 OR word_emote % $2 ORDER BY GREATEST(similarity(word_wearable, $3), similarity(word_emote, $4)) DESC;`
+          // It's repeated 7 times due to the SELECT + WHERE statements: `WHERE word_wearable % $1 OR word_emote % $2 ORDER BY GREATEST(similarity(word_wearable, $3), similarity(word_emote, $4)) DESC;`
           expect(dbClientQueryMock.mock.calls[1][0].values).toEqual([
-            ...Array(4).fill(search),
+            ...Array(7).fill(search),
             limit,
             offset,
           ])
@@ -388,7 +388,7 @@ test('catalog component', function () {
           )
           // It's repeated 4 times due to this WHERE statement: `WHERE word_wearable % $1 OR word_emote % $2 ORDER BY GREATEST(similarity(word_wearable, $3), similarity(word_emote, $4)) DESC;`
           expect(dbClientQueryMock.mock.calls[1][0].values).toEqual([
-            ...Array(4).fill(search),
+            ...Array(7).fill(search),
             limit,
             offset,
           ])