feat: improve repo query disambiguation

src/sources/web.ts

@@ -8,6 +8,46 @@ type SearxResult = {
   publishedDate?: string;
 };
 
+function buildQueries(query: string) {
+  const cleaned = query.trim();
+  if (!cleaned) return [];
+  if (cleaned.includes("site:")) return [cleaned];
+
+  const repoLike = cleaned.match(/\b([a-z0-9_.-]+\/[a-z0-9_.-]+)\b/i)?.[1];
+
+  return [
+    cleaned,
+    ...(repoLike ? [`site:github.com ${repoLike}`] : []),
+    ...(repoLike ? [] : [`site:github.com ${cleaned}`]),
+    `site:stackoverflow.com ${cleaned}`,
+    `${cleaned} official docs`,
+  ];
+}
+
+function normalizeResults(list: SearxResult[], depth: Depth): RawResult[] {
+  const limit = depth === "quick" ? 5 : 10;
+  const deduped = new Map<string, SearxResult>();
+
+  for (const item of list) {
+    const key = item.url ?? `${item.title ?? "untitled"}:${item.content ?? ""}`;
+    if (!deduped.has(key)) deduped.set(key, item);
+  }
+
+  return Array.from(deduped.values())
+    .slice(0, limit)
+    .map((item, i, arr) => ({
+      id: `web-${i}`,
+      type: "web",
+      title: item.title ?? item.url ?? "Untitled",
+      snippet: (item.content ?? item.url ?? "").slice(0, 700),
+      url: item.url,
+      relevance: Math.max(0.1, 1 - i / Math.max(1, arr.length)),
+      timestamp: item.publishedDate
+        ? Date.parse(item.publishedDate) || undefined
+        : undefined,
+    } satisfies RawResult));
+}
+
 export async function searchWeb(
   query: string,
   baseUrl: string | undefined,
@@ -23,49 +63,45 @@ export async function searchWeb(
   }
 
   try {
-    const url = new URL("search", baseUrl.endsWith("/") ? baseUrl : `${baseUrl}/`);
-    url.searchParams.set("q", query);
-    url.searchParams.set("format", "json");
-    const res = await fetch(url, {
-      headers: {
-        Accept: "application/json",
-      },
-    });
+    const collected: SearxResult[] = [];
 
-    if (!res.ok) {
-      return failure(
-        source,
-        "request_failed",
-        res.status === 403
-          ? "SearXNG search failed with status 403. Enable JSON output on the instance."
-          : `SearXNG search failed with status ${res.status}.`,
-      );
-    }
-    const body = (await res.json()) as { results?: SearxResult[] };
-    if (!Array.isArray(body.results)) {
-      return failure(
-        source,
-        "invalid_response",
-        "SearXNG search returned an invalid payload.",
-      );
+    for (const candidate of buildQueries(query)) {
+      const url = new URL("search", baseUrl.endsWith("/") ? baseUrl : `${baseUrl}/`);
+      url.searchParams.set("q", candidate);
+      url.searchParams.set("format", "json");
+      const res = await fetch(url, {
+        headers: {
+          Accept: "application/json",
+        },
+      });
+
+      if (!res.ok) {
+        return failure(
+          source,
+          "request_failed",
+          res.status === 403
+            ? "SearXNG search failed with status 403. Enable JSON output on the instance."
+            : `SearXNG search failed with status ${res.status}.`,
+        );
+      }
+      const body = (await res.json()) as { results?: SearxResult[] };
+      if (!Array.isArray(body.results)) {
+        return failure(
+          source,
+          "invalid_response",
+          "SearXNG search returned an invalid payload.",
+        );
+      }
+
+      collected.push(...body.results);
+      if (collected.length > 0) break;
     }
 
-    const limit = depth === "quick" ? 5 : 10;
-    const list = body.results.slice(0, limit);
+    const results = normalizeResults(collected, depth);
 
     return {
       source,
-      results: list.map((item, i) => ({
-        id: `web-${i}`,
-        type: source,
-        title: item.title ?? item.url ?? "Untitled",
-        snippet: (item.content ?? item.url ?? "").slice(0, 700),
-        url: item.url,
-        relevance: Math.max(0.1, 1 - i / Math.max(1, list.length)),
-        timestamp: item.publishedDate
-          ? Date.parse(item.publishedDate) || undefined
-          : undefined,
-      } satisfies RawResult)),
+      results,
     };
   } catch (error) {
     return failure(

tests/index.test.ts

@@ -398,4 +398,117 @@ describe("opensearch tool", () => {
     });
     expect(body.sources[0]?.snippet).toContain("self-hosted SearXNG");
   });
+
+  it("retries web search with developer-focused fallback queries", async () => {
+    const fetchMock = vi
+      .fn<typeof fetch>()
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({ results: [] }),
+      } as Response)
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({
+          results: [
+            {
+              title: "GitHub issue result",
+              url: "https://github.com/example/repo/issues/1",
+              content: "Relevant developer-facing result.",
+            },
+          ],
+        }),
+      } as Response);
+    globalThis.fetch = fetchMock;
+
+    const hooks = await createHooks();
+    await hooks.config?.({
+      opensearch: {
+        sources: {
+          session: false,
+          web: { enabled: true, url: "https://search.example/" },
+          code: false,
+        },
+        depth: "quick",
+        synth: false,
+      },
+    } as never);
+
+    const tool = hooks.tool?.opensearch;
+    if (!tool) throw new Error("opensearch tool missing");
+
+    const output = await tool.execute(
+      { query: "opencode plugin hooks", sources: ["web"] },
+      createToolContext("/tmp/opensearch").context,
+    );
+    const body = JSON.parse(output) as {
+      status: string;
+      sources: Array<{ title: string; url?: string }>;
+      meta: { sources_yielded: number };
+    };
+
+    expect(body.status).toBe("raw");
+    expect(body.meta.sources_yielded).toBe(1);
+    expect(body.sources[0]).toMatchObject({
+      title: "GitHub issue result",
+      url: "https://github.com/example/repo/issues/1",
+    });
+    expect(fetchMock).toHaveBeenCalledTimes(2);
+    expect(decodeURIComponent(String(fetchMock.mock.calls[1]?.[0]))).toContain(
+      "site:github.com+opencode+plugin+hooks",
+    );
+  });
+
+  it("prefers owner-repo GitHub disambiguation for repo-like queries", async () => {
+    const fetchMock = vi
+      .fn<typeof fetch>()
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({ results: [] }),
+      } as Response)
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({
+          results: [
+            {
+              title: "OpenSearch repository",
+              url: "https://github.com/kagan-sh/opensearch",
+              content: "The exact repository match.",
+            },
+          ],
+        }),
+      } as Response);
+    globalThis.fetch = fetchMock;
+
+    const hooks = await createHooks();
+    await hooks.config?.({
+      opensearch: {
+        sources: {
+          session: false,
+          web: { enabled: true, url: "https://search.example/" },
+          code: false,
+        },
+        depth: "quick",
+        synth: false,
+      },
+    } as never);
+
+    const tool = hooks.tool?.opensearch;
+    if (!tool) throw new Error("opensearch tool missing");
+
+    const output = await tool.execute(
+      { query: "kagan-sh/opensearch", sources: ["web"] },
+      createToolContext("/tmp/opensearch").context,
+    );
+    const body = JSON.parse(output) as {
+      status: string;
+      sources: Array<{ url?: string }>;
+    };
+
+    expect(body.status).toBe("raw");
+    expect(body.sources[0]?.url).toBe("https://github.com/kagan-sh/opensearch");
+    expect(fetchMock).toHaveBeenCalledTimes(2);
+    expect(decodeURIComponent(String(fetchMock.mock.calls[1]?.[0]))).toContain(
+      "site:github.com+kagan-sh/opensearch",
+    );
+  });
 });