mcp_servers/coding/tools/web_search/
mod.rs1pub mod search_client;
2
3pub use search_client::{BraveSearchClient, SearchClient, SearchParams};
4
5#[cfg(test)]
6pub use search_client::FakeSearchClient;
7
8use schemars::JsonSchema;
9use serde::{Deserialize, Serialize};
10use std::future::Future;
11use std::time::Duration;
12use url::Url;
13
14use crate::coding::error::WebSearchError;
15
16const DEFAULT_COUNT: u32 = 10;
17const MAX_COUNT: u32 = 20;
18const MAX_RATE_LIMIT_RETRIES: u32 = 3;
19const MAX_RETRY_AFTER: Duration = Duration::from_mins(1);
20const DEFAULT_RETRY_DELAY: Duration = Duration::from_secs(2);
21
22#[derive(Debug, Clone, Serialize, Deserialize, JsonSchema)]
24#[serde(rename_all = "camelCase")]
25pub struct WebSearchInput {
26 pub query: String,
28
29 pub count: Option<u32>,
31
32 #[serde(alias = "allowed_domains")]
34 pub allowed_domains: Option<Vec<String>>,
35
36 #[serde(alias = "blocked_domains")]
38 pub blocked_domains: Option<Vec<String>>,
39}
40
41#[derive(Debug, Clone, Serialize, Deserialize, JsonSchema)]
43#[serde(rename_all = "camelCase")]
44pub struct WebSearchOutput {
45 pub results: Vec<SearchResult>,
47
48 pub query: String,
50}
51
52#[derive(Debug, Clone, Serialize, Deserialize, JsonSchema)]
54#[serde(rename_all = "camelCase")]
55pub struct SearchResult {
56 pub title: String,
58
59 pub url: String,
61
62 #[serde(alias = "description")]
64 pub snippet: String,
65}
66
67pub struct WebSearcher {
69 client: Box<dyn SearchClient>,
70}
71
72impl WebSearcher {
73 pub fn with_client(client: impl SearchClient + 'static) -> Self {
74 Self { client: Box::new(client) }
75 }
76
77 pub fn try_new() -> Result<Self, WebSearchError> {
78 Ok(Self::with_client(BraveSearchClient::new()?))
79 }
80
81 pub async fn search<T, U>(&self, args: WebSearchInput, mut on_retry: T) -> Result<WebSearchOutput, WebSearchError>
83 where
84 T: FnMut(Duration) -> U,
85 U: Future<Output = ()>,
86 {
87 let query = args.query.trim();
88
89 if query.is_empty() {
90 return Err(WebSearchError::InvalidQuery("Search query cannot be empty".to_string()));
91 }
92
93 let count = args.count.unwrap_or(DEFAULT_COUNT).min(MAX_COUNT);
94
95 let query_owned = query.to_string();
96 let mut retries = 0;
97 let mut results = loop {
98 let params = SearchParams { query: query_owned.clone(), count };
99
100 match self.client.search(params).await {
101 Ok(results) => break results,
102 Err(WebSearchError::RateLimited { retry_after, .. }) if retries < MAX_RATE_LIMIT_RETRIES => {
103 let delay = retry_after.unwrap_or(DEFAULT_RETRY_DELAY).min(MAX_RETRY_AFTER);
104 retries += 1;
105 on_retry(delay).await;
106 tokio::time::sleep(delay).await;
107 }
108 Err(error) => return Err(error),
109 }
110 };
111
112 results.truncate(count as usize);
113 results.retain(|result| {
114 let url = Url::parse(&result.url).ok();
115 let host = url.as_ref().and_then(Url::host_str);
116 let in_any =
117 |domains: &[String]| host.is_some_and(|host| domains.iter().any(|d| host.ends_with(d.as_str())));
118 args.allowed_domains.as_deref().is_none_or(in_any) && !args.blocked_domains.as_deref().is_some_and(in_any)
119 });
120
121 Ok(WebSearchOutput { results, query: query.to_string() })
122 }
123}
124
125#[cfg(test)]
126mod tests {
127 use std::sync::{Arc, Mutex};
128 use std::time::Duration;
129
130 use super::*;
131
132 fn make_result(url: &str) -> SearchResult {
133 SearchResult { title: "Test".to_string(), url: url.to_string(), snippet: "Test description".to_string() }
134 }
135
136 fn input(query: &str) -> WebSearchInput {
137 WebSearchInput { query: query.to_string(), count: None, allowed_domains: None, blocked_domains: None }
138 }
139
140 fn strs(vals: &[&str]) -> Vec<String> {
141 vals.iter().map(std::string::ToString::to_string).collect()
142 }
143
144 fn searcher_with(query: &str, results: Vec<SearchResult>) -> WebSearcher {
145 WebSearcher::with_client(FakeSearchClient::default().with_results(query, results))
146 }
147
148 fn rate_limited(retry_after: Option<Duration>) -> WebSearchError {
149 WebSearchError::RateLimited { message: "too many requests".to_string(), retry_after }
150 }
151
152 #[tokio::test]
153 async fn test_domain_filtering() {
154 let urls = ["https://example.com/page", "https://test.org/page", "https://sub.example.com:8080/page"];
155 let cases = [
156 (Some(strs(&["example.com"])), None, vec![urls[0], urls[2]]),
157 (None, Some(strs(&["example.com"])), vec![urls[1]]),
158 (Some(strs(&["example.com", "test.org"])), Some(strs(&["test.org"])), vec![urls[0], urls[2]]),
159 ];
160
161 for (allowed_domains, blocked_domains, expected) in cases {
162 let searcher = searcher_with("rust", urls.iter().map(|url| make_result(url)).collect());
163 let query = WebSearchInput { allowed_domains, blocked_domains, ..input("rust") };
164 let output = searcher.search(query, |_| async {}).await.unwrap();
165 let urls: Vec<_> = output.results.iter().map(|r| r.url.as_str()).collect();
166 assert_eq!(urls, expected);
167 }
168 }
169
170 #[tokio::test]
171 async fn test_empty_query_returns_error() {
172 let searcher = WebSearcher::with_client(FakeSearchClient::default());
173 let result = searcher.search(input(" "), |_| async {}).await;
174 assert!(matches!(result, Err(WebSearchError::InvalidQuery(_))));
175 }
176
177 #[tokio::test(start_paused = true)]
178 async fn retries_rate_limited_search_after_retry_after_period() {
179 let client = FakeSearchClient::default()
180 .with_results("rust", vec![make_result("https://example.com")])
181 .with_failures("rust", [rate_limited(Some(Duration::ZERO))]);
182 let searcher = WebSearcher::with_client(client);
183
184 let retry_delays = Arc::new(Mutex::new(Vec::new()));
185 let notified_delays = retry_delays.clone();
186 let output = searcher
187 .search(input("rust"), move |delay| {
188 let notified_delays = notified_delays.clone();
189 async move {
190 notified_delays.lock().unwrap().push(delay);
191 }
192 })
193 .await
194 .unwrap();
195
196 assert_eq!(output.results.len(), 1);
197 assert_eq!(*retry_delays.lock().unwrap(), vec![Duration::ZERO]);
198 }
199
200 #[tokio::test(start_paused = true)]
201 async fn retries_rate_limited_search_when_retry_after_header_is_missing() {
202 let client = FakeSearchClient::default()
203 .with_results("rust", vec![make_result("https://example.com")])
204 .with_failures("rust", [rate_limited(None)]);
205 let searcher = WebSearcher::with_client(client);
206
207 let retry_delays = Arc::new(Mutex::new(Vec::new()));
208 let notified_delays = retry_delays.clone();
209 let output = searcher
210 .search(input("rust"), move |delay| {
211 let notified_delays = notified_delays.clone();
212 async move {
213 notified_delays.lock().unwrap().push(delay);
214 }
215 })
216 .await
217 .unwrap();
218
219 assert_eq!(output.results.len(), 1);
220 assert_eq!(retry_delays.lock().unwrap().len(), 1, "should have retried once");
221 }
222
223 #[tokio::test(start_paused = true)]
224 async fn returns_rate_limited_error_after_retry_budget_is_exhausted() {
225 let mut failures = vec![rate_limited(Some(Duration::ZERO)); 4];
226 failures.push(WebSearchError::ApiError("retry budget was exceeded".to_string()));
227 let searcher = WebSearcher::with_client(FakeSearchClient::default().with_failures("rust", failures));
228 let result = searcher.search(input("rust"), |_| async {}).await;
229
230 assert!(matches!(result, Err(WebSearchError::RateLimited { .. })));
231 }
232
233 #[tokio::test]
234 async fn test_count_limiting() {
235 let searcher =
236 searcher_with("test query", (0..30).map(|i| make_result(&format!("https://example.com/{i}"))).collect());
237
238 let output = searcher.search(WebSearchInput { count: Some(15), ..input("test query") }, |_| async {}).await;
239 assert_eq!(output.unwrap().results.len(), 15);
240 }
241
242 #[test]
243 fn search_result_deserializes_brave_description_as_snippet() {
244 let result: SearchResult = serde_json::from_value(serde_json::json!({
245 "title": "Rust Language",
246 "url": "https://rust-lang.org",
247 "description": "A systems language"
248 }))
249 .unwrap();
250
251 assert_eq!(result.snippet, "A systems language");
252 }
253
254 #[test]
255 fn web_search_input_accepts_snake_case_domains() {
256 let args: WebSearchInput = serde_json::from_value(serde_json::json!({
257 "query": "rust",
258 "count": 5,
259 "allowed_domains": ["docs.rs"],
260 "blocked_domains": ["example.com"]
261 }))
262 .unwrap();
263
264 assert_eq!(args.query, "rust");
265 assert_eq!(args.count, Some(5));
266 assert_eq!(args.allowed_domains, Some(vec!["docs.rs".to_string()]));
267 assert_eq!(args.blocked_domains, Some(vec!["example.com".to_string()]));
268 }
269}