Adjust searchToMikroOrmQuery function to reduce the amount of irrel…

…evant results (#2075)

This is done by a combination of an and- & or-query.
It makes a search more restrictive. For example a search of `Red Shirt`
won't give all products containing `red` or `shirt` but rather returns
all products that have the words `red` and `shirt` in some column. The
words don't have to be in the same column.

---------

Co-authored-by: Johannes Obermair <48853629+johnnyomair@users.noreply.github.com>

.changeset/tasty-garlics-repair.md

@@ -0,0 +1,7 @@
+---
+"@comet/cms-api": minor
+---
+
+Adjust `searchToMikroOrmQuery` function to reduce the amount of irrelevant results
+
+This is done by using a combination of AND- and OR-queries. For example, a search of `red shirt` won't give all products containing `red` OR `shirt` but rather returns all products that have the words `red` AND `shirt` in some column. The words don't have to be in the same column.

packages/api/cms-api/src/common/filter/mikro-orm.spec.ts

@@ -1,38 +1,96 @@
 import { BooleanFilter } from "./boolean.filter";
 import { DateFilter } from "./date.filter";
-import { filtersToMikroOrmQuery, filterToMikroOrmQuery, searchToMikroOrmQuery } from "./mikro-orm";
+import { filtersToMikroOrmQuery, filterToMikroOrmQuery, searchToMikroOrmQuery, splitSearchString } from "./mikro-orm";
 import { NumberFilter } from "./number.filter";
 import { StringFilter } from "./string.filter";
 
+describe("splitSearchString", () => {
+    it("should split a simple space-separated string", () => {
+        const input = "This is a test";
+        const expected = ["%This%", "%is%", "%a%", "%test%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle quoted strings as single tokens", () => {
+        const input = 'This is a "quoted string"';
+        const expected = ["%This%", "%is%", "%a%", "%quoted string%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle escaped quotes within quoted strings", () => {
+        const input = 'This is a "quoted \\"string\\""';
+        const expected = ["%This%", "%is%", "%a%", '%quoted "string"%'];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle single quotes", () => {
+        const input = "This is a 'quoted string'";
+        const expected = ["%This%", "%is%", "%a%", "%quoted string%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle escaped quotes within single quoted strings", () => {
+        const input = "This is a 'quoted \\'string\\''";
+        const expected = ["%This%", "%is%", "%a%", "%quoted 'string'%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle mixed quotes", () => {
+        const input = "This \"is a\" 'test'";
+        const expected = ["%This%", "%is a%", "%test%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle empty strings", () => {
+        const input = "";
+        const expected: string[] = [];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle strings with special characters", () => {
+        const input = "This is a test with % and _ characters";
+        const expected = ["%This%", "%is%", "%a%", "%test%", "%with%", "%\\%%", "%and%", "%\\_%", "%characters%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+
+    it("should handle strings with only special characters", () => {
+        const input = "% _";
+        const expected = ["%\\%%", "%\\_%"];
+        expect(splitSearchString(input)).toEqual(expected);
+    });
+});
+
 describe("searchToMikroOrmQuery", () => {
     it("should work", async () => {
         expect(searchToMikroOrmQuery("foo", ["title", "description"])).toStrictEqual({
-            $or: [{ title: { $ilike: "%foo%" } }, { description: { $ilike: "%foo%" } }],
+            $and: [{ $or: [{ title: { $ilike: "%foo%" } }, { description: { $ilike: "%foo%" } }] }],
         });
     });
     it("should escape %", async () => {
         expect(searchToMikroOrmQuery("fo%o", ["title", "description"])).toStrictEqual({
-            $or: [{ title: { $ilike: "%fo\\%o%" } }, { description: { $ilike: "%fo\\%o%" } }],
+            $and: [{ $or: [{ title: { $ilike: "%fo\\%o%" } }, { description: { $ilike: "%fo\\%o%" } }] }],
         });
     });
     it("should escape _", async () => {
         expect(searchToMikroOrmQuery("fo_o", ["title", "description"])).toStrictEqual({
-            $or: [{ title: { $ilike: "%fo\\_o%" } }, { description: { $ilike: "%fo\\_o%" } }],
+            $and: [{ $or: [{ title: { $ilike: "%fo\\_o%" } }, { description: { $ilike: "%fo\\_o%" } }] }],
         });
     });
     it("should split by spaces", async () => {
         expect(searchToMikroOrmQuery("foo bar", ["title", "description"])).toStrictEqual({
-            $or: [
-                { title: { $ilike: "%foo%" } },
-                { title: { $ilike: "%bar%" } },
-                { description: { $ilike: "%foo%" } },
-                { description: { $ilike: "%bar%" } },
+            $and: [
+                {
+                    $or: [{ title: { $ilike: "%foo%" } }, { description: { $ilike: "%foo%" } }],
+                },
+                {
+                    $or: [{ title: { $ilike: "%bar%" } }, { description: { $ilike: "%bar%" } }],
+                },
             ],
         });
     });
     it("should ignore leading and trailing spaces", async () => {
         expect(searchToMikroOrmQuery(" a ", ["title"])).toStrictEqual({
-            $or: [{ title: { $ilike: "%a%" } }],
+            $and: [{ $or: [{ title: { $ilike: "%a%" } }] }],
         });
     });
 });

packages/api/cms-api/src/common/filter/mikro-orm.ts

@@ -150,7 +150,7 @@ export function filtersToMikroOrmQuery(
     return genericFilter(filter);
 }
 
-const splitSearchString = (search: string) => {
+export const splitSearchString = (search: string) => {
     // regex to match all single tokens or quotes in a string => "This is a 'quoted string'" will result in ["This", "is", "a", "quoted string"]
     // it will also take escaped quotes (prepended with a backslash => \) into account
     const regex = /(["'])(?:(?=(\\?))\2.)*?\1|\S+/g;
@@ -169,9 +169,11 @@ const splitSearchString = (search: string) => {
 export function searchToMikroOrmQuery(search: string, fields: string[]): ObjectQuery<any> {
     const quotedSearchParts = splitSearchString(search);
 
-    const ors = [];
-    for (const field of fields) {
-        for (const quotedSearch of quotedSearchParts) {
+    const ands = [];
+
+    for (const quotedSearch of quotedSearchParts) {
+        const ors = [];
+        for (const field of fields) {
             // eslint-disable-next-line @typescript-eslint/no-explicit-any
             const or: any = {};
             let nestedFilter = or;
@@ -181,8 +183,9 @@ export function searchToMikroOrmQuery(search: string, fields: string[]): ObjectQ
             nestedFilter.$ilike = quotedSearch;
             ors.push(or);
         }
+        ands.push({ $or: ors });
     }
     return {
-        $or: ors,
+        $and: ands,
     };
 }