//! `rdsh search-web`: SearXNG by default, or explicitly selected Ollama search.
//! SearXNG needs no API key and retains its HTML-based HTTP client; a local
//! rustxng sidecar is the same protocol, so it rides the same client. Ollama
//! uses the hosted HTTPS API with `OLLAMA_API_KEY`; providers never fall back
//! to one another, so a local search cannot silently become a cloud request.
mod ollama;
#[derive(Clone, Copy, Debug, clap::ValueEnum)]
pub enum Provider {
Searxng,
Ollama,
}
/// Which SearXNG-compatible engine `search-web` should resolve.
///
/// `Auto` keeps the historical environment order and only then tries a local
/// rustxng. An explicit `Searxng` never spawns rustxng, so a scripted run
/// keeps today's behaviour exactly.
#[derive(Clone, Copy, Debug, PartialEq, Eq, clap::ValueEnum)]
pub enum Engine {
Auto,
Searxng,
Rustxng,
}
/// How one engine serves results: SearXNG renders result markup, while
/// rustxng keeps that markup incompatible on purpose and publishes JSON.
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
enum Format {
Html,
Json,
}
fn format_for(engine: Engine) -> Format {
match engine {
Engine::Rustxng => Format::Json,
Engine::Searxng | Engine::Auto => Format::Html,
}
}
/// Parse the JSON result document rustxng serves (and SearXNG serves when the
/// json format is enabled): {"results":[{"url","title","content"}...]}.
fn parse_json(body: &str, limit: usize) -> anyhow::Result<Vec<Hit>> {
let value: serde_json::Value = serde_json::from_str(body)
.map_err(|error| anyhow::anyhow!("the search engine did not return JSON: {error}"))?;
let Some(results) = value.get("results").and_then(serde_json::Value::as_array) else {
anyhow::bail!("the search engine JSON has no results array");
};
let mut hits = Vec::new();
for item in results {
let Some(url) = item.get("url").and_then(serde_json::Value::as_str) else {
continue;
};
if url.is_empty() || !url.starts_with("http") {
continue;
}
let title = item
.get("title")
.and_then(serde_json::Value::as_str)
.unwrap_or(url);
let content = item
.get("content")
.and_then(serde_json::Value::as_str)
.unwrap_or("");
hits.push(Hit {
title: title.trim().to_string(),
url: url.to_string(),
content: content.trim().to_string(),
});
if hits.len() >= limit {
break;
}
}
Ok(hits)
}
pub fn help(locale: crate::locale::Locale) -> &'static str {
match locale {
crate::locale::Locale::English => "Enable: rdsh features on search-web\nOllama: set OLLAMA_API_KEY, then use --provider ollama (up to 10 results).\nEngines: --engine auto|searxng|rustxng. auto resolves SEARXNG_URL, then RUSTXNG_URL, then a local rustxng, then http://127.0.0.1:8888.\nLocal rustxng: set RUSTXNG_BIN (or install rustxng) and rdsh starts the sidecar itself; --engine searxng never spawns it.\nWeb Search API service guide: Tavily https://www.tavily.com/\nTavily quickstart: https://docs.tavily.com/documentation/quickstart\nBuilt-in providers: searxng, ollama. Tavily is a service reference.\nSearXNG down? --auto-start runs docker-compose.searxng.yml (loopback only).\nDocs: https://github.com/jimoto-no-llm/rustdsh/blob/main/docs/WEB-SEARCH.md",
crate::locale::Locale::Chinese => "启用：rdsh features on search-web\nOllama：设置 OLLAMA_API_KEY 后使用 --provider ollama（最多10条结果）。\n搜索引擎：--engine auto|searxng|rustxng。auto 依次使用 SEARXNG_URL、RUSTXNG_URL、本地 rustxng，最后是 http://127.0.0.1:8888。\n本地 rustxng：设置 RUSTXNG_BIN（或安装 rustxng）后由 rdsh 启动该服务；--engine searxng 不会启动它。\nWeb Search API 服务指南：Tavily https://www.tavily.com/\nTavily 入门：https://docs.tavily.com/documentation/quickstart\n内置提供方：searxng、ollama。Tavily 为服务介绍。\nSearXNG 未启动时：--auto-start 会运行 docker-compose.searxng.yml（仅回环地址）。\n文档：https://github.com/jimoto-no-llm/rustdsh/blob/main/docs/WEB-SEARCH.md",
crate::locale::Locale::Japanese => "有効化: rdsh features on search-web\nOllama: OLLAMA_API_KEY を設定し --provider ollama を指定（最大10件）。\nエンジン: --engine auto|searxng|rustxng。auto は SEARXNG_URL、RUSTXNG_URL、ローカル rustxng、http://127.0.0.1:8888 の順に解決します。\nローカル rustxng: RUSTXNG_BIN を設定（または rustxng を導入）すると rdsh がサイドカーを起動します。--engine searxng は起動しません。\nWeb Search API サービスの案内: Tavily https://www.tavily.com/\nTavily 入門: https://docs.tavily.com/documentation/quickstart\n組み込み検索先: searxng、ollama。Tavily はサービスの紹介です。\nSearXNG が停止中: --auto-start が docker-compose.searxng.yml を起動します（ループバック限定）。\n利用手順: https://github.com/jimoto-no-llm/rustdsh/blob/main/docs/WEB-SEARCH.md",
}
}
pub(crate) fn default_base() -> String {
match std::env::var("SEARXNG_URL") {
Ok(s) => {
let s = s.trim_end_matches('/').to_string();
if s.is_empty() {
"http://127.0.0.1:8888".to_string()
} else {
s
}
}
Err(_) => {
let u = crate::rdsh_config::load().search.searxng_url;
let u = u.trim().trim_end_matches('/').to_string();
if u.is_empty() {
"http://127.0.0.1:8888".to_string()
} else {
u
}
}
}
}
/// Wrapper around the loopback health probe: the unit tests swap this so the
/// resolution order can be checked without a listener.
fn engine_health(base: &str) -> bool {
crate::rustxng::health(base)
}
/// Resolve the SearXNG-compatible base URL for one search.
///
/// `Auto` walks SEARXNG_URL, RUSTXNG_URL, a local rustxng (started on demand),
/// then the historical http://127.0.0.1:8888 default. `Searxng` still honours
/// SEARXNG_URL and then the `rdsh.json` search.searxng_url override, but never
/// starts rustxng. `Rustxng` uses RUSTXNG_URL, else a local rustxng, and never
/// falls through to a different engine.
pub(crate) fn resolve_base(engine: Engine) -> anyhow::Result<String> {
resolve_base_with(engine, engine_health)
}
fn resolve_base_with(engine: Engine, health: impl Fn(&str) -> bool) -> anyhow::Result<String> {
match engine {
Engine::Searxng => Ok(default_base()),
Engine::Rustxng => match crate::rustxng::base_url() {
Some(url) => Ok(url),
None => crate::rustxng::ensure().map_err(anyhow::Error::new),
},
Engine::Auto => {
if let Some(url) = env_url("SEARXNG_URL") {
return Ok(url);
}
if let Some(url) = env_url("RUSTXNG_URL") {
return Ok(url);
}
// A healthy instance is reused; a missing binary or a process that
// never answered falls back to the historical default, whose own
// connection error names both variables.
if let Ok(base) = crate::rustxng::ensure() {
if health(&base) {
return Ok(base);
}
}
Ok(default_base())
}
}
}
/// One trimmed, trailing-slash-free environment URL, when it is set.
fn env_url(name: &str) -> Option<String> {
let value = std::env::var(name).ok()?;
let value = value.trim().trim_end_matches('/').to_string();
(!value.is_empty()).then_some(value)
}
pub(crate) fn encode_query(s: &str) -> String {
use std::fmt::Write as _;
let mut out = String::with_capacity(s.len());
for b in s.bytes() {
if b.is_ascii_alphanumeric() || b"-_.~".contains(&b) {
out.push(b as char);
} else if b == b' ' {
out.push('+');
} else {
let _ = write!(out, "%{b:02X}");
}
}
out
}
/// Split an `http://host[:port][/prefix]` base URL.
pub(crate) fn split_base(
base: &str,
locale: crate::locale::Locale,
) -> anyhow::Result<(String, u16, String)> {
// The configured endpoint is a human-selected capability, not a URL from
// search results. Preserve remote instances, but never interpolate request
// separators, userinfo, queries, or fragments into the HTTP request.
anyhow::ensure!(
!base.chars().any(|c| c.is_control() || c.is_whitespace())
&& !base.contains(['@', '?', '#']),
"{}",
crate::locale::native_message(locale, crate::locale::NativeMessage::InvalidSearxUrl)
);
let rest = base.strip_prefix("http://").ok_or_else(|| {
anyhow::anyhow!(
"{}: {base}",
crate::locale::native_message(locale, crate::locale::NativeMessage::OnlyHttpSearxUrl)
)
})?;
let (hostport, prefix) = match rest.find('/') {
Some(i) => (&rest[..i], rest[i..].to_string()),
None => (rest, String::new()),
};
let (host, port_text) = if hostport.starts_with('[') {
let end = hostport.find(']').ok_or_else(|| {
anyhow::anyhow!(
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::InvalidIpv6SearxHost
)
)
})?;
hostport[1..end]
.parse::<std::net::Ipv6Addr>()
.map_err(|_| {
anyhow::anyhow!(
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::InvalidIpv6SearxHost
)
)
})?;
(&hostport[..=end], &hostport[end + 1..])
} else {
let end = hostport.find(':').unwrap_or(hostport.len());
let host = &hostport[..end];
anyhow::ensure!(
!host.is_empty()
&& host
.bytes()
.all(|b| b.is_ascii_alphanumeric() || b".-_".contains(&b)),
"{}",
crate::locale::native_message(locale, crate::locale::NativeMessage::InvalidSearxHost)
);
(host, &hostport[end..])
};
let port = if port_text.is_empty() {
80
} else {
port_text
.strip_prefix(':')
.and_then(|s| s.parse::<u16>().ok())
.filter(|port| *port != 0)
.ok_or_else(|| {
anyhow::anyhow!(
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::InvalidSearxPort
)
)
})?
};
Ok((host.to_string(), port, prefix))
}
/// Loopback check for the SearXNG warning: 127/8, ::1 and localhost names.
/// Bracketed IPv6 literals (as kept by split_base) are unwrapped first.
pub(crate) fn is_loopback_host(host: &str) -> bool {
let h = host
.strip_prefix('[')
.and_then(|s| s.strip_suffix(']'))
.unwrap_or(host);
h == "localhost"
|| h.parse::<std::net::Ipv4Addr>()
.is_ok_and(|ip| ip.is_loopback())
|| h.parse::<std::net::Ipv6Addr>()
.is_ok_and(|ip| ip.is_loopback())
}
#[derive(Debug, Clone, PartialEq)]
pub(crate) struct Hit {
pub(crate) title: String,
pub(crate) url: String,
pub(crate) content: String,
}
fn strip_tags(s: &str) -> String {
let mut out = String::with_capacity(s.len());
let mut tag = false;
let mut prev_space = true;
for c in s.chars() {
match c {
'<' => tag = true,
'>' => {
tag = false;
prev_space = true;
}
_ if tag => {}
_ if c.is_whitespace() => {
if !prev_space {
out.push(' ');
prev_space = true;
}
}
_ => {
out.push(c);
prev_space = false;
}
}
}
out.trim().to_string()
}
fn decode_entities(s: &str) -> String {
let mut out = s
.replace("&", "&")
.replace("<", "<")
.replace(">", ">")
.replace(""", "\"")
.replace("'", "'")
.replace("'", "'");
let mut search = 0;
while let Some(offset) = out[search..].find("&#") {
let i = search + offset;
let rest = &out[i + 2..];
let Some(end) = rest.find(';') else {
break;
};
let num = &rest[..end];
let ch = if num.len() > 8 || num.contains('&') {
None
} else if let Some(hex) = num.strip_prefix(['x', 'X']) {
u32::from_str_radix(hex, 16).ok().and_then(char::from_u32)
} else {
num.parse::<u32>().ok().and_then(char::from_u32)
};
match ch {
Some(c) => {
out.replace_range(i..i + 2 + end + 1, &c.to_string());
search = i + c.len_utf8();
}
None => search = i + 2,
}
}
out
}
fn clean(s: &str) -> String {
decode_entities(&strip_tags(s))
}
/// First `href` + link text inside an `<article>` block whose `class`
/// attribute marks it as a result. Only the start tag's class tokens decide:
/// body text (title/URL/snippet) must never affect classification.
fn parse_html(h: &str, limit: usize) -> Vec<Hit> {
let mut hits = Vec::new();
let mut rest = h;
while let Some(a) = rest.find("<article") {
let block = &rest[a..];
let end = block
.find("</article>")
.map(|i| i + 10)
.unwrap_or(block.len());
let blk = &block[..end];
rest = &block[end.min(block.len())..];
// Classify by the article start tag's class tokens only: title, URL
// and snippet text must never affect the decision (#382).
let head = &blk[..blk
.find('>')
.map(|i| i + 1)
.unwrap_or(blk.len())
.min(blk.len())];
let class = attr(head, "class").unwrap_or_default();
let tokens: Vec<&str> = class.split_whitespace().collect();
if !tokens.iter().any(|t| t.contains("result")) {
continue;
}
if tokens.iter().any(|t| t.contains("correction")) {
continue;
}
let href = match find_href(blk) {
Some(u) => u,
None => continue,
};
let title = match find_link_text(blk) {
Some(t) => clean(&t),
None => continue,
};
if title.is_empty() || href.is_empty() {
continue;
}
let content = find_snippet(blk).map(|s| clean(&s)).unwrap_or_default();
hits.push(Hit {
title,
url: href,
content,
});
if hits.len() >= limit.max(1) {
break;
}
}
hits
}
fn find_href(blk: &str) -> Option<String> {
let mut rest = blk;
while let Some(a) = rest.find("<a") {
let tag = &rest[a..];
let close = tag.find('>')?;
let head = &tag[..close];
if let Some(u) = attr(head, "href").filter(|u| !u.is_empty() && !u.starts_with('#')) {
return Some(decode_entities(&u));
}
rest = &tag[close + 1..];
}
None
}
/// Text of the same anchor `find_href` returns (first link with a real href).
fn find_link_text(blk: &str) -> Option<String> {
let mut rest = blk;
while let Some(a) = rest.find("<a") {
let tag = &rest[a..];
let close = tag.find('>')?;
let head = &tag[..close];
let ok = attr(head, "href").is_some_and(|u| !u.is_empty() && !u.starts_with('#'));
let after = &tag[close + 1..];
if ok {
let end = after.find("</a>").unwrap_or(after.len());
return Some(after[..end].to_string());
}
rest = after;
}
None
}
fn attr(tag_head: &str, name: &str) -> Option<String> {
// Matches name="..." or name='...' without quote escapes in source.
const SQ: u8 = b'\'';
let mut rest = tag_head;
while let Some(i) = rest.find(name) {
let after = &rest[i + name.len()..];
let b = after.as_bytes();
if b.len() < 2 || b[0] != b'=' || (b[1] != b'"' && b[1] != SQ) {
let step = after.chars().next().map_or(0, char::len_utf8);
rest = &after[step..];
continue;
}
let v = &after[2..];
let end = v.find(b[1] as char).unwrap_or(v.len());
return Some(v[..end].to_string());
}
None
}
fn find_snippet(blk: &str) -> Option<String> {
let mut rest = blk;
while let Some(i) = rest.find("<p class=") {
let tag = &rest[i..];
let gt = tag.find('>')?;
if tag[..gt].contains("content") {
let after = &tag[gt + 1..];
let end = after.find("</p>").unwrap_or(after.len());
return Some(after[..end].to_string());
}
rest = &tag[gt + 1..];
}
None
}
/// The connection-failure sentence. It names both variables that can point at
/// a SearXNG instance and says that installing rustxng gives rdsh a local
/// engine it can start itself.
fn unreachable_engine_message(base: &str, locale: crate::locale::Locale) -> String {
match locale {
crate::locale::Locale::English => format!(
"cannot reach SearXNG at {base}; is it running? Set SEARXNG_URL or RUSTXNG_URL, or install the rustxng binary (RUSTXNG_BIN) so rdsh can start the local engine"
),
crate::locale::Locale::Chinese => format!(
"无法连接 {base} 的 SearXNG；它是否正在运行？请设置 SEARXNG_URL 或 RUSTXNG_URL，或安装 rustxng（RUSTXNG_BIN）以便 rdsh 启动本地引擎"
),
crate::locale::Locale::Japanese => format!(
"{base} の SearXNG に接続できません。起動していますか？SEARXNG_URL か RUSTXNG_URL を設定するか、rustxng（RUSTXNG_BIN）を導入すると rdsh がローカルエンジンを起動できます"
),
}
}
fn fetch_with(
base: &str,
query: &str,
locale: crate::locale::Locale,
format: Format,
) -> anyhow::Result<Vec<Hit>> {
use std::io::{Read, Write};
use std::net::ToSocketAddrs;
let (host, port, prefix) = split_base(base, locale)?;
if !is_loopback_host(&host) {
// Plain-http to a non-loopback host is observable on the network.
// Advisory only: remote instances stay usable via an SSH tunnel.
eprintln!("[rdsh] warning: SearXNG host {host} is not loopback; prefer an SSH tunnel to localhost");
}
let mut target = format!("{prefix}/search?q={}", encode_query(query));
if format == Format::Json {
target.push_str("&format=json");
}
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(15);
let mut stream = None;
for address in format!("{host}:{port}").to_socket_addrs()? {
let remaining = deadline.saturating_duration_since(std::time::Instant::now());
if remaining.is_zero() {
break;
}
if let Ok(connected) = std::net::TcpStream::connect_timeout(&address, remaining) {
stream = Some(connected);
break;
}
}
let mut s =
stream.ok_or_else(|| anyhow::anyhow!("{}", unreachable_engine_message(base, locale)))?;
let ua = format!("rdsh/{}", env!("CARGO_PKG_VERSION"));
// Virtual-hosted instances need the port to route; the default port is
// omitted per convention. Bracketed IPv6 literals pass through as-is.
let host_header = if port == 80 {
host.clone()
} else {
format!("{host}:{port}")
};
let accept = if format == Format::Json {
"application/json"
} else {
"text/html"
};
let request = format!("GET {target} HTTP/1.0\r\nHost: {host_header}\r\nUser-Agent: {ua}\r\nAccept: {accept}\r\nConnection: close\r\n\r\n");
let mut pending = request.as_bytes();
while !pending.is_empty() {
let remaining = deadline.saturating_duration_since(std::time::Instant::now());
anyhow::ensure!(
!remaining.is_zero(),
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::SearxRequestTimedOut
)
);
s.set_write_timeout(Some(remaining))?;
let n = s.write(pending)?;
anyhow::ensure!(
n != 0,
"{}",
crate::locale::native_message(locale, crate::locale::NativeMessage::SearxRequestClosed)
);
pending = &pending[n..];
}
let mut raw = Vec::new();
let mut chunk = [0u8; 8192];
loop {
let remaining = deadline.saturating_duration_since(std::time::Instant::now());
if remaining.is_zero() {
anyhow::bail!(
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::SearxResponseTimedOut
)
);
}
s.set_read_timeout(Some(remaining))?;
let n = s.read(&mut chunk)?;
if n == 0 {
break;
}
if raw.len().saturating_add(n) > 2 * 1024 * 1024 {
anyhow::bail!(
"{}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::SearxResponseTooLarge
)
);
}
raw.extend_from_slice(&chunk[..n]);
}
let text = String::from_utf8_lossy(&raw);
let status = text
.lines()
.next()
.and_then(|line| {
let mut parts = line.split_whitespace();
let protocol = parts.next()?;
if !matches!(protocol, "HTTP/1.0" | "HTTP/1.1") {
return None;
}
parts.next()?.parse::<u16>().ok()
})
.ok_or_else(|| {
anyhow::anyhow!(
"{} {base}",
crate::locale::native_message(
locale,
crate::locale::NativeMessage::BadSearxResponse
)
)
})?;
if !(200..300).contains(&status) {
anyhow::bail!(
"{} {status} {} {base}",
crate::locale::native_message(locale, crate::locale::NativeMessage::SearxReturnedHttp),
match locale {
crate::locale::Locale::English => "from",
crate::locale::Locale::Chinese => "，来源",
crate::locale::Locale::Japanese => "、送信元",
}
);
}
let body = match text.find("\r\n\r\n") {
Some(i) => &text[i + 4..],
None => anyhow::bail!(
"{} {base}",
crate::locale::native_message(locale, crate::locale::NativeMessage::BadSearxResponse)
),
};
if format == Format::Json {
return parse_json(body, 50);
}
ensure_results_page(body, base, locale)?;
Ok(parse_html(body, 50))
}
/// HTML search client used by the stub tests.
#[cfg(test)]
fn fetch(base: &str, query: &str, locale: crate::locale::Locale) -> anyhow::Result<Vec<Hit>> {
fetch_with(base, query, locale, Format::Html)
}
/// Accept recognizable empty SearXNG pages; preserve locale on invalid pages.
fn ensure_results_page(
body: &str,
base: &str,
locale: crate::locale::Locale,
) -> anyhow::Result<()> {
if body.contains("<article") || looks_like_searxng(body) {
return Ok(());
}
anyhow::bail!(
"{} {base}{}",
crate::locale::native_message(locale, crate::locale::NativeMessage::NoSearxResultsPage),
match locale {
crate::locale::Locale::English => " (is SEARXNG_URL right?)",
crate::locale::Locale::Chinese => "（SEARXNG_URL 是否正确？）",
crate::locale::Locale::Japanese => "（SEARXNG_URL は正しいですか？）",
}
);
}
/// The real HTML search client against one base URL. Shared with the rustxng
/// unit test, which points it at its own TCP stub.
#[cfg(test)]
pub(crate) fn fetch_via_stub(
base: &str,
query: &str,
locale: crate::locale::Locale,
) -> anyhow::Result<Vec<Hit>> {
fetch(base, query, locale)
}
fn looks_like_searxng(body: &str) -> bool {
body.contains("id=\"results\"") || body.contains("content=\"searxng")
}
pub fn cmd_search_web(
query: &str,
limit: usize,
json: bool,
provider: Provider,
engine: Engine,
auto_start: bool,
locale: crate::locale::Locale,
) -> anyhow::Result<()> {
let (base, hits) = match provider {
Provider::Searxng => {
let base = resolve_base(engine)?;
// Opt-in only: the search still runs (and reports its own
// connection error) when a container cannot be started, except for
// the explicit --auto-start failures below, which say why. A
// rustxng base keeps --auto-start out of the compose path.
if auto_start && engine != Engine::Rustxng {
let outcome = crate::searxng::ensure_running(&base, true, locale);
if let Some(notice) = outcome.notice(&base, locale) {
eprintln!("{notice}");
}
if let Some(failure) = outcome.start_failure(&base, locale) {
anyhow::bail!("{failure}");
}
}
let format = format_for(engine);
let hits = match fetch_with(&base, query, locale, format) {
Ok(hits) => hits,
Err(error) => {
// An auto-detected rustxng base also serves JSON rather
// than the SearXNG result markup, so one JSON retry is
// worth it unless the operator pinned SearXNG.
if format == Format::Html && engine != Engine::Searxng {
match fetch_with(&base, query, locale, Format::Json) {
Ok(hits) => hits,
Err(_) => return Err(error),
}
} else {
return Err(error);
}
}
};
(base, hits)
}
Provider::Ollama => (
ollama::ENDPOINT.to_string(),
ollama::search(query, limit, locale)?,
),
};
let shown: Vec<&Hit> = hits.iter().take(limit.max(1)).collect();
if json {
println!(
"{}",
serde_json::json!({
"query": query,
"engine": base,
"results": shown
.iter()
.map(|r| {
serde_json::json!({
"title": r.title,
"url": r.url,
"content": r.content,
})
})
.collect::<Vec<_>>(),
})
);
return Ok(());
}
use std::io::Write as _;
let out = std::io::stdout();
let mut out = std::io::BufWriter::new(out.lock());
for r in &shown {
let _ = writeln!(out, "- {}\n {}", r.title, r.url);
if !r.content.is_empty() {
let snippet: String = r.content.chars().take(240).collect();
let _ = writeln!(out, " {snippet}");
}
}
let _ = out.flush();
match locale {
crate::locale::Locale::English => eprintln!("[rdsh] {} web hit(s) via {base}", shown.len()),
crate::locale::Locale::Chinese => {
eprintln!("[rdsh] {} 个网页匹配，来源 {base}", shown.len())
}
crate::locale::Locale::Japanese => {
eprintln!("[rdsh] {} 件のWeb一致、経由 {base}", shown.len())
}
}
Ok(())
}
#[cfg(test)]
mod tests {
use super::*;
// 0 件でも正しい SearXNG インスタンスなら成功扱いにする（設定ミスと誤診しない）。
#[test]
fn empty_searxng_page_is_not_a_configuration_error() {
let page =
"<html><body><div id=\"results\"><p>No results were found.</p></div></body></html>";
assert!(ensure_results_page(
page,
"http://127.0.0.1:8888",
crate::locale::Locale::English
)
.is_ok());
assert!(parse_html(page, 50).is_empty());
}
#[test]
fn searxng_generator_meta_is_accepted() {
let page = "<html><head><meta name=\"generator\" content=\"searxng/2026\"></head></html>";
assert!(ensure_results_page(
page,
"http://127.0.0.1:8888",
crate::locale::Locale::English
)
.is_ok());
}
#[test]
fn a_plain_page_is_still_reported_as_a_wrong_url() {
let page = "<html><body><h1>hello</h1></body></html>";
assert!(ensure_results_page(
page,
"http://127.0.0.1:8888",
crate::locale::Locale::English
)
.is_err());
}
#[test]
fn empty_results_and_invalid_pages_retain_the_requested_language() {
use crate::locale::Locale;
for (locale, expected) in [
(Locale::English, "no SearXNG results page"),
(Locale::Chinese, "以下地址没有 SearXNG 结果页"),
(Locale::Japanese, "SearXNG の結果ページがありません"),
] {
assert!(ensure_results_page(
"<div id=\"results\"></div>",
"http://127.0.0.1:8888",
locale
)
.is_ok());
let error = ensure_results_page(
"<html>wrong service</html>",
"http://127.0.0.1:8888",
locale,
)
.unwrap_err()
.to_string();
assert!(error.contains(expected), "{error}");
assert!(error.contains("http://127.0.0.1:8888"));
}
}
// Engine selection: auto falls back to the historical default when neither
// variable is set and no local rustxng answers.
#[test]
fn auto_engine_falls_back_to_the_default() {
let _lock = engine_env_lock();
let dir = std::env::temp_dir().join(format!(
"rdsh-websearch-auto-{}-{}",
std::process::id(),
nondet_nonce()
));
std::fs::create_dir_all(&dir).unwrap();
let _env = EngineEnvGuard::apply(&[
("SEARXNG_URL", None),
("RUSTXNG_URL", None),
("RUSTXNG_PORT", None),
("DSH_HOME", Some(dir.to_string_lossy().as_ref())),
("HOME", Some(dir.to_string_lossy().as_ref())),
// No PATH entry, so the lookup cannot reach a real install.
("PATH", Some(dir.to_string_lossy().as_ref())),
]);
// With nothing configured and no local engine the historical default
// is the answer, and the health probe is never consulted.
assert_eq!(
resolve_base_with(Engine::Auto, |_| panic!("no probe without a local engine")).unwrap(),
"http://127.0.0.1:8888"
);
let _ = std::fs::remove_dir_all(&dir);
}
/// A unique tag for temp directories shared by the resolution tests.
fn nondet_nonce() -> u128 {
std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.map(|d| d.as_nanos())
.unwrap_or_default()
}
// SEARXNG_URL is consulted before RUSTXNG_URL, and both before the local
// sidecar (the health probe must never be reached).
#[test]
fn auto_engine_prefers_searxng_url_then_rustxng_url() {
let _lock = engine_env_lock();
let _env = EngineEnvGuard::apply(&[
("SEARXNG_URL", Some("http://127.0.0.1:9999/")),
("RUSTXNG_URL", Some("http://127.0.0.1:7777")),
]);
assert_eq!(
resolve_base_with(Engine::Auto, |_| panic!("no probe when an env URL is set")).unwrap(),
"http://127.0.0.1:9999"
);
std::env::remove_var("SEARXNG_URL");
assert_eq!(
resolve_base_with(Engine::Auto, |_| panic!("no probe when RUSTXNG_URL is set"))
.unwrap(),
"http://127.0.0.1:7777"
);
}
// --engine searxng keeps today's behaviour and never starts rustxng, even
// when RUSTXNG_URL points somewhere.
#[test]
fn explicit_searxng_never_spawns_rustxng() {
let _lock = engine_env_lock();
let _env = EngineEnvGuard::apply(&[
("RUSTXNG_URL", Some("http://127.0.0.1:7777")),
("SEARXNG_URL", Some("http://127.0.0.1:9999")),
]);
assert_eq!(
resolve_base_with(Engine::Searxng, |_| panic!(
"searxng must not probe rustxng"
))
.unwrap(),
"http://127.0.0.1:9999"
);
std::env::remove_var("SEARXNG_URL");
let base = resolve_base_with(Engine::Searxng, |_| {
panic!("searxng must not probe rustxng")
})
.unwrap();
assert!(!base.contains("7777"), "{base}");
}
// --engine rustxng uses RUSTXNG_URL and never another engine.
#[test]
fn explicit_rustxng_uses_only_the_configured_or_local_engine() {
let _lock = engine_env_lock();
let _env = EngineEnvGuard::apply(&[
("RUSTXNG_URL", Some("http://127.0.0.1:7777")),
("SEARXNG_URL", Some("http://127.0.0.1:9999")),
]);
assert_eq!(
resolve_base_with(Engine::Rustxng, |_| panic!(
"a configured URL needs no probe"
))
.unwrap(),
"http://127.0.0.1:7777"
);
// With nothing configured the local engine decides; a missing binary is
// an explicit error rather than a silent fallback.
std::env::remove_var("RUSTXNG_URL");
let dir = std::env::temp_dir().join(format!(
"rdsh-websearch-resolve-{}-{}",
std::process::id(),
std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.map(|d| d.as_nanos())
.unwrap_or_default()
));
std::fs::create_dir_all(&dir).unwrap();
std::env::set_var("DSH_HOME", &dir);
std::env::remove_var("RUSTXNG_PORT");
let result = resolve_base_with(Engine::Rustxng, |_| false);
std::env::remove_var("RUSTXNG_BIN");
assert!(result.is_err(), "rustxng must not fall back to searxng");
let _ = std::fs::remove_dir_all(&dir);
}
/// Serialises the environment-touching resolution tests in this module.
fn engine_env_lock() -> std::sync::MutexGuard<'static, ()> {
static LOCK: std::sync::OnceLock<std::sync::Mutex<()>> = std::sync::OnceLock::new();
LOCK.get_or_init(|| std::sync::Mutex::new(()))
.lock()
.unwrap_or_else(|poisoned| poisoned.into_inner())
}
/// Sets variables for one test and restores them on drop.
struct EngineEnvGuard {
saved: Vec<(String, Option<String>)>,
}
impl EngineEnvGuard {
fn apply(pairs: &[(&str, Option<&str>)]) -> Self {
let mut saved = Vec::new();
for (name, value) in pairs {
saved.push(((*name).to_string(), std::env::var(name).ok()));
match value {
Some(value) => std::env::set_var(name, value),
None => std::env::remove_var(name),
}
}
Self { saved }
}
}
impl Drop for EngineEnvGuard {
fn drop(&mut self) {
for (name, value) in &self.saved {
match value {
Some(value) => std::env::set_var(name, value),
None => std::env::remove_var(name),
}
}
}
}
#[test]
fn query_encoding() {
assert_eq!(encode_query("a b+c"), "a+b%2Bc");
assert_eq!(encode_query("日本語"), "%E6%97%A5%E6%9C%AC%E8%AA%9E");
}
#[test]
fn base_splitting() {
assert_eq!(
split_base("http://127.0.0.1:8888", crate::locale::Locale::English).unwrap(),
("127.0.0.1".to_string(), 8888, String::new())
);
assert_eq!(
split_base("http://h/searx", crate::locale::Locale::English).unwrap(),
("h".to_string(), 80, "/searx".to_string())
);
assert!(split_base("https://h/", crate::locale::Locale::English).is_err());
assert_eq!(
split_base("http://[::1]/searx", crate::locale::Locale::English).unwrap(),
("[::1]".to_string(), 80, "/searx".to_string())
);
assert_eq!(
split_base(
"http://searx.example:8888/prefix",
crate::locale::Locale::English
)
.unwrap(),
("searx.example".to_string(), 8888, "/prefix".to_string())
);
}
#[test]
fn base_rejects_http_request_injection_and_malformed_authorities() {
for base in [
"http://127.0.0.1:8888/prefix\r\nInjected: yes",
"http://127.0.0.1:8888/path HTTP/1.0",
"http://localhost\t:8888",
"http://user@localhost:8888",
"http://localhost:8888/path?other=query",
"http://localhost:8888/#fragment",
"http://[::1]extra:8888",
"http://[bad]:8888",
"http://::1",
"http://localhost:0",
"http://localhost:65536",
"http://localhost:",
"http:///path",
] {
assert!(
split_base(base, crate::locale::Locale::English).is_err(),
"{base:?}"
);
}
// The query is encoded separately and cannot create a second request.
assert_eq!(encode_query("\r\nInjected: yes"), "%0D%0AInjected%3A+yes");
}
#[test]
fn rustxng_uses_the_json_format() {
assert_eq!(format_for(Engine::Rustxng), Format::Json);
assert_eq!(format_for(Engine::Searxng), Format::Html);
assert_eq!(format_for(Engine::Auto), Format::Html);
}
#[test]
fn json_results_are_parsed_and_limited() {
let body = r#"{"results":[{"url":"https://a.example/1","title":"A","content":"one"},{"url":"https://b.example/2","title":"B"},{"title":"no url"},{"url":"https://c.example/3","title":"C","content":"three"}]}"#;
let hits = parse_json(body, 2).expect("json body");
assert_eq!(hits.len(), 2);
assert_eq!(hits[0].title, "A");
assert_eq!(hits[0].url, "https://a.example/1");
assert_eq!(hits[1].content, "");
}
#[test]
fn json_without_results_is_an_error() {
assert!(parse_json("<html>not json</html>", 5).is_err());
assert!(parse_json("{\"nope\":1}", 5).is_err());
}
#[test]
fn fetch_encodes_query_and_preserves_configured_prefix() {
use std::io::{Read, Write};
let listener = std::net::TcpListener::bind("127.0.0.1:0").unwrap();
listener.set_nonblocking(true).unwrap();
let address = listener.local_addr().unwrap();
let server = std::thread::spawn(move || {
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(3);
let mut stream = loop {
match listener.accept() {
Ok((stream, _)) => break stream,
Err(e) if e.kind() == std::io::ErrorKind::WouldBlock => {
assert!(
std::time::Instant::now() < deadline,
"no fixture connection"
);
std::thread::sleep(std::time::Duration::from_millis(5));
}
Err(e) => panic!("fixture accept: {e}"),
}
};
// Accepted sockets inherit O_NONBLOCK from the listener on BSD.
stream.set_nonblocking(false).unwrap();
stream
.set_read_timeout(Some(std::time::Duration::from_secs(3)))
.unwrap();
let mut request = Vec::new();
while !request.ends_with(b"\r\n\r\n") {
let mut byte = [0];
assert_eq!(stream.read(&mut byte).unwrap(), 1);
request.push(byte[0]);
assert!(request.len() < 8192);
}
let body = "<article class=\"result\"><a href=\"https://example.invalid/\">DUMMY_RESULT</a></article>";
write!(
stream,
"HTTP/1.0 200 OK\r\nContent-Length: {}\r\n\r\n{body}",
body.len()
)
.unwrap();
String::from_utf8(request).unwrap()
});
let hits = fetch(
&format!("http://{address}/prefix"),
"hello\r\nInjected: yes",
crate::locale::Locale::English,
);
let request = server.join().unwrap();
assert_eq!(hits.unwrap()[0].title, "DUMMY_RESULT");
assert!(request.starts_with("GET /prefix/search?q=hello%0D%0AInjected%3A+yes HTTP/1.0\r\n"));
assert!(!request.contains("\r\nInjected:"));
// Ephemeral port: the Host header must carry it for vhost routing.
assert!(request.contains(&format!("\r\nHost: {address}\r\n")));
}
#[test]
fn loopback_classification() {
for good in ["127.0.0.1", "127.1.2.3", "localhost", "::1", "[::1]"] {
assert!(is_loopback_host(good), "{good:?}");
}
for bad in [
"192.168.1.1",
"10.0.0.1",
"example.com",
"",
"127.0.0.1:8888",
] {
assert!(!is_loopback_host(bad), "{bad:?}");
}
}
#[test]
fn html_parsing() {
let h = "<article class=\"result\"><h3><a href=\"https://e.com/x?a=1&b=2\">A <b>B</b></a></h3><p class=\"content\">one <two></p></article><article class=\"result result-correction\"><a href=\"https://e.com/y\">no</a></article>";
let hits = parse_html(h, 10);
assert_eq!(hits.len(), 1);
assert_eq!(hits[0].title, "A B");
assert_eq!(hits[0].url, "https://e.com/x?a=1&b=2");
assert_eq!(hits[0].content, "one <two>");
}
#[test]
fn html_parsing_keeps_correction_words_in_title_url_snippet() {
// Regression test for #382: the word "correction" in body text must
// not exclude a normal result; only the start tag's class tokens decide.
let h = concat!(
"<article class=\"result\"><h3><a href=\"https://example.invalid/a\">",
"Quantum error correction</a></h3><p class=\"content\">Useful paper</p></article>",
"<article class=\"result\"><h3><a href=\"https://example.invalid/correction-notes\">",
"Quantum paper</a></h3><p class=\"content\">Useful paper</p></article>",
"<article class=\"result\"><h3><a href=\"https://example.invalid/b\">",
"Quantum paper</a></h3><p class=\"content\">About error correction</p></article>",
"<article class=\"result result-correction\"><h3><a href=\"https://example.invalid/d\">",
"Did you mean?</a></h3></article>",
);
let hits = parse_html(h, 10);
assert_eq!(hits.len(), 3);
assert_eq!(hits[0].title, "Quantum error correction");
assert_eq!(hits[0].url, "https://example.invalid/a");
assert_eq!(hits[1].url, "https://example.invalid/correction-notes");
assert_eq!(hits[2].content, "About error correction");
}
#[test]
fn html_parsing_skips_utf8_attribute_name_candidates() {
for character in ["é", "→", "🦀"] {
let url = format!("https://example.invalid/{character}");
let title = format!("結果 {character}");
let html = format!(
"<article class=\"result\"><a title=\"href{character}\" href=\"{url}\">{title}</a></article>"
);
assert_eq!(
parse_html(&html, 1),
vec![Hit {
title,
url,
content: String::new(),
}],
"failed to skip a {character:?} candidate"
);
}
}
#[test]
fn attribute_name_at_end_does_not_panic() {
assert_eq!(attr("href", "href"), None);
}
#[test]
fn malformed_numeric_entity_does_not_panic_or_hide_later_entities() {
assert_eq!(decode_entities("AA"), "AA");
assert_eq!(decode_entities("AAB"), "AAB");
assert_eq!(decode_entities("bad &#xZ; then A"), "bad &#xZ; then A");
}
}