fix(search): default web search to bing (#1619)

Summary: - add Bing as explicit default web_search provider - keep explicit DuckDuckGo configuration supported - update docs/help/config examples Validation: CI green before merge.
2026-05-14 03:31:15 -05:00
parent d46f7415b3
commit a3f88bf6cf
9 changed files with 117 additions and 50 deletions
@@ -5,6 +5,13 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

+## [Unreleased]
+
+### Changed
+
+- **Bing is the default `web_search` backend.** DuckDuckGo remains selectable
+  with `[search] provider = "duckduckgo"` and keeps its Bing fallback path.
+
 ## [0.8.36] - 2026-05-14

 ### Added
@@ -220,12 +220,14 @@ max_subagents = 10 # optional (1-20)
 # ─────────────────────────────────────────────────────────────────────────────────
 # Web Search Provider
 # ─────────────────────────────────────────────────────────────────────────────────
-# Choose which backend `web_search` uses. Default is DuckDuckGo HTML scraping
-# with Bing fallback — no API key needed but may be blocked in some regions.
+# Choose which backend `web_search` uses. Default is Bing HTML scraping — no
+# API key needed. DuckDuckGo remains selectable and still falls back to Bing
+# when its HTML endpoint returns a bot challenge or no parseable results.
 # Switch to Tavily or Bocha for reliable search in mainland China.
 #
 # [search]
-# provider = "duckduckgo"   # duckduckgo | tavily | bocha
+# provider = "bing"         # bing | duckduckgo | tavily | bocha
+#                            # duckduckgo: HTML scrape with Bing fallback
 #                            # tavily: https://tavily.com — AI search, needs api_key
 #                            # bocha:  https://bochaai.com — 博查AI搜索，国内友好，需api_key
 # api_key = "tvly-YOUR_KEY"  # required for tavily and bocha
@@ -5,6 +5,13 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

+## [Unreleased]
+
+### Changed
+
+- **Bing is the default `web_search` backend.** DuckDuckGo remains selectable
+  with `[search] provider = "duckduckgo"` and keeps its Bing fallback path.
+
 ## [0.8.36] - 2026-05-14

 ### Added
@@ -525,8 +525,10 @@ impl SnapshotsConfig {
 #[derive(Debug, Clone, Copy, Default, PartialEq, Eq, Deserialize, Serialize)]
 #[serde(rename_all = "snake_case")]
 pub enum SearchProvider {
-    /// DuckDuckGo HTML scraping with Bing fallback. No API key needed.
+    /// Bing HTML scraping. No API key needed.
    #[default]
+    Bing,
+    /// DuckDuckGo HTML scraping with Bing fallback. No API key needed.
    #[serde(alias = "duckduckgo")]
    DuckDuckGo,
    /// Tavily AI Search API (<https://tavily.com>). Requires api_key.
@@ -539,6 +541,7 @@ impl SearchProvider {
    #[must_use]
    pub fn as_str(self) -> &'static str {
        match self {
+            Self::Bing => "bing",
            Self::DuckDuckGo => "duckduckgo",
            Self::Tavily => "tavily",
            Self::Bocha => "bocha",
@@ -549,10 +552,10 @@ impl SearchProvider {
 /// Web search provider configuration (`[search]` table in config.toml).
 #[derive(Debug, Clone, Deserialize, Default)]
 pub struct SearchConfig {
-    /// Search provider: `duckduckgo` | `tavily` | `bocha`. Default: `duckduckgo`.
+    /// Search provider: `bing` | `duckduckgo` | `tavily` | `bocha`. Default: `bing`.
    #[serde(default)]
    pub provider: Option<SearchProvider>,
-    /// API key for Tavily or Bocha. Not required for DuckDuckGo.
+    /// API key for Tavily or Bocha. Not required for Bing or DuckDuckGo.
    #[serde(default)]
    pub api_key: Option<String>,
 }
@@ -905,9 +908,9 @@ pub struct Config {
    #[serde(default)]
    pub snapshots: Option<SnapshotsConfig>,

-    /// Web search provider configuration. When absent, defaults to DuckDuckGo
-    /// with Bing fallback. Set `provider` to `tavily` or `bocha` and provide
-    /// an `api_key` to use those services instead.
+    /// Web search provider configuration. When absent, defaults to Bing.
+    /// Set `provider` to `duckduckgo`, `tavily`, or `bocha` to use those
+    /// services instead; Tavily and Bocha also require an `api_key`.
    #[serde(default)]
    pub search: Option<SearchConfig>,

@@ -3468,6 +3471,27 @@ mod tests {
        assert!(runtime.trusts_proxy_fakeip_host("raw.githubusercontent.com"));
    }

+    #[test]
+    fn search_provider_defaults_to_bing() {
+        assert_eq!(SearchProvider::default(), SearchProvider::Bing);
+    }
+
+    #[test]
+    fn explicit_duckduckgo_search_provider_is_preserved() {
+        let config: Config = toml::from_str(
+            r#"
+            [search]
+            provider = "duckduckgo"
+            "#,
+        )
+        .expect("search config");
+
+        assert_eq!(
+            config.search.and_then(|search| search.provider),
+            Some(SearchProvider::DuckDuckGo)
+        );
+    }
+
    struct EnvGuard {
        home: Option<OsString>,
        userprofile: Option<OsString>,
@@ -162,9 +162,9 @@ pub struct EngineConfig {
    pub strict_tool_mode: bool,
    /// Workshop / large-tool-output routing (#548). `None` disables routing.
    pub workshop: Option<crate::tools::large_output_router::WorkshopConfig>,
-    /// Which search backend `web_search` should use. Default: DuckDuckGo.
+    /// Which search backend `web_search` should use. Default: Bing.
    pub search_provider: crate::config::SearchProvider,
-    /// API key for Tavily or Bocha. `None` for DuckDuckGo.
+    /// API key for Tavily or Bocha. `None` for Bing or DuckDuckGo.
    pub search_api_key: Option<String>,
 }

@@ -157,10 +157,10 @@ pub struct ToolContext {
    /// routing (e.g. in sub-agents and test contexts to avoid recursion).
    pub large_output_router: Option<crate::tools::large_output_router::LargeOutputRouter>,

-    /// Which search backend `web_search` should use. Default: DuckDuckGo with
-    /// Bing fallback. Set via `[search] provider` in config.toml.
+    /// Which search backend `web_search` should use. Default: Bing. Set via
+    /// `[search] provider` in config.toml.
    pub search_provider: crate::config::SearchProvider,
-    /// API key for Tavily or Bocha. `None` for DuckDuckGo.
+    /// API key for Tavily or Bocha. `None` for Bing or DuckDuckGo.
    pub search_api_key: Option<String>,

    /// Per-session workshop variable store (#548). Holds the raw content of
@@ -1,11 +1,11 @@
-//! Web search tool backed by multiple providers: DuckDuckGo (HTML scrape
-//! with Bing fallback), Tavily API, and Bocha (博查) API.
+//! Web search tool backed by multiple providers: Bing HTML scrape, DuckDuckGo
+//! (HTML scrape with Bing fallback), Tavily API, and Bocha (博查) API.
 //!
 //! This is the primary web search surface for agents. For browsing workflows
 //! (page open, click, screenshot) use a direct URL approach instead.
 //!
 //! Set `[search]` in config.toml to switch providers:
-//!   provider = "tavily"  # requires api_key
+//!   provider = "duckduckgo"  # or tavily/bocha
 //!   api_key = "tvly-..."

 use super::spec::{
@@ -124,7 +124,7 @@ impl ToolSpec for WebSearchTool {
    }

    fn description(&self) -> &'static str {
-        "Search the web and return ranked results with URLs and snippets. Default backend is DuckDuckGo (with Bing fallback); set `[search] provider = \"tavily\" | \"bocha\"` in config.toml to switch backends. Use this instead of scraping search engines with `curl` in `exec_shell`. For a known canonical URL, prefer `fetch_url` directly."
+        "Search the web and return ranked results with URLs and snippets. Default backend is Bing; set `[search] provider = \"duckduckgo\" | \"tavily\" | \"bocha\"` in config.toml to switch backends. Use this instead of scraping search engines with `curl` in `exec_shell`. For a known canonical URL, prefer `fetch_url` directly."
    }

    fn input_schema(&self) -> Value {
@@ -181,7 +181,8 @@ impl ToolSpec for WebSearchTool {
        let max_results = max_results.clamp(1, MAX_RESULTS);
        let timeout_ms = optional_u64(&input, "timeout_ms", DEFAULT_TIMEOUT_MS).min(60_000);

-        // Dispatch to the configured search provider.
+        // Dispatch to the configured API-backed search providers before
+        // building the HTML-scraping client used by Bing/DuckDuckGo.
        match context.search_provider {
            SearchProvider::Tavily => {
                let decider = context.network_policy.as_ref();
@@ -197,18 +198,10 @@ impl ToolSpec for WebSearchTool {
                    .run_bocha_search(&query, max_results, timeout_ms, context)
                    .await;
            }
-            SearchProvider::DuckDuckGo => {
-                // fall through to existing DuckDuckGo + Bing fallback logic
-            }
+            SearchProvider::Bing | SearchProvider::DuckDuckGo => {}
        }

-        // Per-domain network policy gate (#135). The "host" for web search is
-        // the upstream search engine domain — DuckDuckGo first, Bing on
-        // fallback. We gate DuckDuckGo here; Bing is gated separately inside
-        // `run_bing_search` so a deny on one engine doesn't block the other.
        let decider = context.network_policy.as_ref();
-        check_policy(decider, DUCKDUCKGO_HOST)?;
-
        let client = reqwest::Client::builder()
            .timeout(Duration::from_millis(timeout_ms))
            .user_agent(USER_AGENT)
@@ -217,6 +210,18 @@ impl ToolSpec for WebSearchTool {
                ToolError::execution_failed(format!("Failed to build HTTP client: {e}"))
            })?;

+        if matches!(context.search_provider, SearchProvider::Bing) {
+            check_policy(decider, BING_HOST)?;
+            let results = run_bing_search(&client, &query, max_results).await?;
+            return search_tool_result(query, "bing", results, None);
+        }
+
+        // Per-domain network policy gate (#135). The "host" for web search is
+        // the upstream search engine domain — DuckDuckGo first, Bing on
+        // fallback. We gate DuckDuckGo here; Bing is gated separately inside
+        // the fallback path so a deny on one engine doesn't block the other.
+        check_policy(decider, DUCKDUCKGO_HOST)?;
+
        let encoded = url_encode(&query);
        let url = format!("https://html.duckduckgo.com/html/?q={encoded}");
        let resp = client
@@ -244,7 +249,7 @@ impl ToolSpec for WebSearchTool {
        }

        let mut results = parse_duckduckgo_results(&body, max_results);
-        let mut source = "duckduckgo".to_string();
+        let mut source = "duckduckgo";
        let mut message_suffix = None;
        if results.is_empty() {
            let duckduckgo_blocked = is_duckduckgo_challenge(&body);
@@ -254,7 +259,7 @@ impl ToolSpec for WebSearchTool {
            match run_bing_search(&client, &query, max_results).await {
                Ok(fallback_results) if !fallback_results.is_empty() => {
                    results = fallback_results;
-                    source = "bing".to_string();
+                    source = "bing";
                    message_suffix = Some(if duckduckgo_blocked {
                        "DuckDuckGo returned a bot challenge; used Bing fallback"
                    } else {
@@ -274,26 +279,36 @@ impl ToolSpec for WebSearchTool {
                Ok(_) | Err(_) => {}
            }
        }
-        let message = if results.is_empty() {
-            "No results found".to_string()
-        } else if let Some(suffix) = message_suffix {
-            format!("Found {} result(s). {suffix}", results.len())
-        } else {
-            format!("Found {} result(s)", results.len())
-        };

-        let response = WebSearchResponse {
-            query,
-            source,
-            count: results.len(),
-            message,
-            results,
-        };
-
-        ToolResult::json(&response).map_err(|e| ToolError::execution_failed(e.to_string()))
+        search_tool_result(query, source, results, message_suffix)
    }
 }

+fn search_tool_result(
+    query: String,
+    source: &'static str,
+    results: Vec<WebSearchEntry>,
+    message_suffix: Option<&str>,
+) -> Result<ToolResult, ToolError> {
+    let message = if results.is_empty() {
+        "No results found".to_string()
+    } else if let Some(suffix) = message_suffix {
+        format!("Found {} result(s). {suffix}", results.len())
+    } else {
+        format!("Found {} result(s)", results.len())
+    };
+
+    let response = WebSearchResponse {
+        query,
+        source: source.to_string(),
+        count: results.len(),
+        message,
+        results,
+    };
+
+    ToolResult::json(&response).map_err(|e| ToolError::execution_failed(e.to_string()))
+}
+
 impl WebSearchTool {
    /// Search via Tavily AI Search API (<https://tavily.com>).
    async fn run_tavily_search(
@@ -591,16 +606,16 @@ async fn run_bing_search(
        .header("Accept-Language", "en-US,en;q=0.9")
        .send()
        .await
-        .map_err(|e| ToolError::execution_failed(format!("Bing fallback request failed: {e}")))?;
+        .map_err(|e| ToolError::execution_failed(format!("Bing search request failed: {e}")))?;

    let status = resp.status();
    let body = resp.text().await.map_err(|e| {
-        ToolError::execution_failed(format!("Failed to read Bing fallback response: {e}"))
+        ToolError::execution_failed(format!("Failed to read Bing search response: {e}"))
    })?;

    if !status.is_success() {
        return Err(ToolError::execution_failed(format!(
-            "Bing fallback failed: HTTP {}",
+            "Bing search failed: HTTP {}",
            status.as_u16()
        )));
    }
@@ -620,6 +620,18 @@ You can also override features for a single run:

 Use `deepseek-tui features list` to inspect known flags and their effective state.

+## Web Search Provider
+
+`web_search` uses Bing by default and does not require an API key. DuckDuckGo
+remains selectable for users who explicitly want it, and Tavily or Bocha can be
+selected when an API-backed provider is preferred.
+
+```toml
+[search]
+provider = "bing" # bing | duckduckgo | tavily | bocha
+# api_key = "tvly-YOUR_KEY" # required for tavily and bocha
+```
+
 ## Local Media Attachments

 Use `@path/to/file` in the composer to add local text file or directory context
@@ -35,7 +35,7 @@ chosen over the available shell equivalent. Companion to `crates/tui/src/prompts
 |---|---|
 | `grep_files` | Regex search file contents within the workspace; structured matches + context lines. Pure-Rust (`regex` crate), no `rg`/`grep` shell-out. |
 | `file_search` | Fuzzy-match filenames (not contents). Use when you know roughly the name. |
-| `web_search` | DuckDuckGo (with Bing fallback); ranked snippets + `ref_id` for citation. |
+| `web_search` | Bing by default; DuckDuckGo, Tavily, and Bocha are selectable in config. Ranked snippets + `ref_id` for citation. |
 | `fetch_url` | Direct HTTP GET on a known URL. Faster than `web_search` when the link is already known. HTML stripped to text by default. |

 ### Shell