From 52a9a785c10f404a3300421bf5ca60beebc9a633 Mon Sep 17 00:00:00 2001
From: matevip <matevip@mate.vip>
Date: Fri, 24 Apr 2026 22:08:42 +0800
Subject: [PATCH] fix(search): bundled SearXNG sidecar actually works out of
 the box

---
 .env.example                                  |  4 ++
 docker-compose.yml                            | 14 ++++++-
 .../tool/search/SearXNGSearchProvider.java    | 42 +++++++++++++++----
 3 files changed, 50 insertions(+), 10 deletions(-)

diff --git a/.env.example b/.env.example
index 4b9d24e8..9561c692 100644
--- a/.env.example
+++ b/.env.example
@@ -29,6 +29,10 @@ JWT_SECRET=
 # 若留空，服务器会允许所有 origin 并在启动日志里 WARN。生产部署务必设置。
 MATECLAW_CORS_ALLOWED_ORIGINS=
 
+# SearXNG 会话密钥（容器内部用，留空会用开发默认值）。生产部署请设成 32+ 位随机串。
+# openssl rand -hex 32
+SEARXNG_SECRET=
+
 # ==================== 浏览器工具（可选） ====================
 #
 # Docker 镜像已经把 Chromium 打进去了，默认零配置可用。
diff --git a/docker-compose.yml b/docker-compose.yml
index 9bd2b24f..82481181 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -31,18 +31,30 @@ services:
       timeout: 5s
       retries: 5
 
-  # SearXNG 搜索引擎（keyless 搜索 provider，零配置可用）
+  # SearXNG 搜索引擎（keyless 搜索 provider）
+  #
+  # ⚠️ deploy/searxng/settings.yml is bind-mounted because the upstream image
+  # ships with JSON format disabled and the anti-bot Limiter plugin enabled —
+  # both of which would silently break mateclaw's SearXNGSearchProvider.
+  # Do NOT remove that bind mount.
   searxng:
     image: searxng/searxng:latest
     container_name: mateclaw-searxng
     restart: unless-stopped
     environment:
       - SEARXNG_BASE_URL=http://searxng:8080
+      - SEARXNG_SECRET=${SEARXNG_SECRET:-mateclaw-dev-searxng-secret-change-me}
+      - UWSGI_WORKERS=2
+      - UWSGI_THREADS=4
     volumes:
       - searxng_data:/etc/searxng
+      # This override MUST be mounted AFTER the named volume so it shadows
+      # the image's default settings.yml (enables JSON, disables limiter).
+      - ./deploy/searxng/settings.yml:/etc/searxng/settings.yml:ro
     ports:
       - "8088:8080"
     healthcheck:
+      # Healthz needs json format, so this also doubles as an integration check.
       test: ["CMD", "wget", "--spider", "-q", "http://localhost:8080/healthz"]
       interval: 30s
       timeout: 5s
diff --git a/mateclaw-server/src/main/java/vip/mate/tool/search/SearXNGSearchProvider.java b/mateclaw-server/src/main/java/vip/mate/tool/search/SearXNGSearchProvider.java
index ec3c73c1..aa05211e 100644
--- a/mateclaw-server/src/main/java/vip/mate/tool/search/SearXNGSearchProvider.java
+++ b/mateclaw-server/src/main/java/vip/mate/tool/search/SearXNGSearchProvider.java
@@ -81,24 +81,42 @@ public class SearXNGSearchProvider implements SearchProvider {
             urlBuilder.append("&time_range=").append(searchQuery.freshness().toLowerCase());
         }
 
-        String response = HttpUtil.createGet(urlBuilder.toString())
+        var resp = HttpUtil.createGet(urlBuilder.toString())
                 .header("Accept", "application/json")
                 .timeout(15000)
-                .execute()
-                .body();
+                .execute();
+        int status = resp.getStatus();
+        String response = resp.body();
+        String contentType = resp.header("Content-Type");
 
-        log.debug("SearXNG result for '{}': length={}", searchQuery.query(), response != null ? response.length() : 0);
-        return parseResponse(response, searchQuery.resolvedCount());
+        log.debug("SearXNG response for '{}': status={}, contentType={}, length={}",
+                searchQuery.query(), status, contentType, response != null ? response.length() : 0);
+        return parseResponse(response, status, contentType, searchQuery.resolvedCount(), urlBuilder.toString());
     }
 
-    private List<SearchResult> parseResponse(String response, int limit) {
+    private List<SearchResult> parseResponse(String response, int status, String contentType,
+                                             int limit, String requestUrl) {
         List<SearchResult> results = new ArrayList<>();
-        if (response == null || response.isBlank()) return results;
+        if (response == null || response.isBlank()) {
+            log.warn("SearXNG returned empty body (status={}, url={})", status, requestUrl);
+            return results;
+        }
+        if (status >= 400) {
+            log.warn("SearXNG returned HTTP {} — preview: {}", status, preview(response));
+            return results;
+        }
+        if (contentType != null && !contentType.contains("json")) {
+            // Most common cause: settings.yml has no `json` under search.formats
+            // or the Limiter plugin rewrote the response to HTML.
+            log.warn("SearXNG did not return JSON (contentType={}). Check settings.yml has search.formats including 'json' and server.limiter: false. Preview: {}",
+                    contentType, preview(response));
+            return results;
+        }
 
         try {
             JSONObject json = JSONUtil.parseObj(response);
             JSONArray items = json.getJSONArray("results");
-            if (items == null) return results;
+            if (items == null || items.isEmpty()) return results;
 
             limit = Math.min(items.size(), limit);
             for (int i = 0; i < limit; i++) {
@@ -114,11 +132,17 @@ public class SearXNGSearchProvider implements SearchProvider {
                         .build());
             }
         } catch (Exception e) {
-            log.warn("SearXNG 结果解析失败: {}", e.getMessage());
+            log.warn("SearXNG parse failed: {} — preview: {}", e.getMessage(), preview(response));
         }
         return results;
     }
 
+    private static String preview(String body) {
+        if (body == null) return "";
+        String flat = body.replaceAll("\\s+", " ").trim();
+        return flat.length() > 200 ? flat.substring(0, 200) + "..." : flat;
+    }
+
     private String extractDomain(String url) {
         try {
             return URI.create(url).getHost();