feat: search engine

Co-authored-by: Max Mahn <[e-mail]>

ts/searchEngine/index.test.ts

···
         expect(index.getPagesForKeyword("pineapples")).toBeArrayOfSize(2);
     });
 });
+
+describe("Search Algorithm", () => {
+    let index: SearchIndex;
+    beforeEach(() => {
+        index = new SearchIndex();
+        index.addPage(
+            "https://www.beans.com",
+            "beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans beans",
+        );
+        index.addPage("https://www.beans-are-ok.com", "beans are ok I guess");
+        index.addPage(
+            "https://www.example.com/cats",
+            "This is a sample web page about cats",
+        );
+        index.addPage(
+            "https://www.example.com/dogs",
+            "This is a sample web page about dogs and training",
+        );
+        index.addPage(
+            "https://www.training.com",
+            "This is a general training website",
+        );
+        index.addPage(
+            "https://www.pineapple-world.com",
+            "We have lots of pineapples. You've never seen this many pineapples before.",
+        );
+        index.addPage(
+            "https://www.pineapple-is-my-favorite-fruit.com",
+            "I love pineapples, it's all I eat. I mean I REALLY LOVE PINEAPPLES",
+        );
+        index.addPage(
+            "https://www.example.com/ml",
+            "This is a page about machine learning",
+        );
+    });
+
+    it("should return relevant pages for a single keyword search", () => {
+        const results = index.search("cats");
+        expect(results).toContain("https://www.example.com/cats");
+    });
+
+    it("should return relevant pages for a multi keyword search", () => {
+        const results = index.search("dogs training");
+        expect(results).toContainAllValues([
+            "https://www.example.com/dogs",
+            "https://www.training.com",
+        ]);
+        expect(results.indexOf("https://www.example.com/dogs")).toBe(0);
+        expect(results.indexOf("https://www.training.com")).toBe(1);
+    });
+
+    it("should return relevant pages for a phrase search", () => {
+        const results = index.search("machine learning");
+        expect(results).toContain("https://www.example.com/ml");
+    });
+
+    it("should rank results properly by relevance", () => {
+        const results = index.search("beans");
+        expect(results.indexOf("https://www.beans.com")).toBe(0);
+        expect(results.indexOf("https://www.beans-are-ok.com")).toBe(1);
+        const results2 = index.search("beans beans");
+        expect(results2.indexOf("https://www.beans.com")).toBe(0);
+        expect(results2.indexOf("https://www.beans-are-ok.com")).toBe(1);
+    });
+});

ts/searchEngine/index.ts

···
             .sort((a, b) => a[1] - b[1])
             .map(([url, _]) => url);
     }
+
+    search(query: string): string[] {
+        const urls = new Map<string, number>();
+        const keys = this.extractKeywords(query);
+        for (const [key, count] of keys.entries()) {
+            const pages = this.index.get(key);
+            if (!pages) continue;
+            for (const [url, v] of pages) {
+                urls.set(
+                    url,
+                    urls.has(url)
+                        ? (urls.get(url) || 0) + v * count
+                        : v * count,
+                );
+            }
+        }
+        return Array.from(urls.entries())
+            .sort((a, b) => b[1] - a[1])
+            .map(([url, _]) => url);
+    }
 }