libdd_common/
lib.rs

1// Copyright 2021-Present Datadog, Inc. https://www.datadoghq.com/
2// SPDX-License-Identifier: Apache-2.0
3#![cfg_attr(not(test), deny(clippy::panic))]
4#![cfg_attr(not(test), deny(clippy::unwrap_used))]
5#![cfg_attr(not(test), deny(clippy::expect_used))]
6#![cfg_attr(not(test), deny(clippy::todo))]
7#![cfg_attr(not(test), deny(clippy::unimplemented))]
8
9use anyhow::Context;
10use http::uri;
11use serde::de::Error;
12use serde::{Deserialize, Deserializer, Serialize, Serializer};
13use std::sync::{Mutex, MutexGuard, RwLock, RwLockReadGuard, RwLockWriteGuard};
14use std::{borrow::Cow, ops::Deref, path::PathBuf, str::FromStr};
15
16pub mod azure_app_services;
17#[cfg(not(target_arch = "wasm32"))]
18pub mod cc_utils;
19#[cfg(not(target_arch = "wasm32"))]
20pub mod connector;
21#[cfg(feature = "reqwest")]
22pub mod dump_server;
23pub mod entity_id;
24pub mod regex_engine;
25#[macro_use]
26pub mod cstr;
27#[cfg(feature = "bench-utils")]
28pub mod bench_utils;
29pub mod config;
30pub mod error;
31pub mod http_common;
32pub mod multipart;
33#[cfg(not(target_arch = "wasm32"))]
34pub mod rate_limiter;
35pub mod tag;
36#[cfg(any(test, feature = "test-utils"))]
37pub mod test_utils;
38#[cfg(not(target_arch = "wasm32"))]
39pub mod threading;
40#[cfg(not(target_arch = "wasm32"))]
41pub mod timeout;
42pub mod unix_utils;
43
44/// Extension trait for `Mutex` to provide a method that acquires a lock, panicking if the lock is
45/// poisoned.
46///
47/// This helper function is intended to be used to avoid having to add many
48/// `#[allow(clippy::unwrap_used)]` annotations if there are a lot of usages of `Mutex`.
49///
50/// # Arguments
51///
52/// * `self` - A reference to the `Mutex` to lock.
53///
54/// # Returns
55///
56/// A `MutexGuard` that provides access to the locked data.
57///
58/// # Panics
59///
60/// This function will panic if the `Mutex` is poisoned.
61///
62/// # Examples
63///
64/// ```
65/// use libdd_common::MutexExt;
66/// use std::sync::{Arc, Mutex};
67///
68/// let data = Arc::new(Mutex::new(5));
69/// let data_clone = Arc::clone(&data);
70///
71/// std::thread::spawn(move || {
72///     let mut num = data_clone.lock_or_panic();
73///     *num += 1;
74/// })
75/// .join()
76/// .expect("Thread panicked");
77///
78/// assert_eq!(*data.lock_or_panic(), 6);
79/// ```
80pub trait MutexExt<T> {
81    fn lock_or_panic(&self) -> MutexGuard<'_, T>;
82}
83
84impl<T> MutexExt<T> for Mutex<T> {
85    #[inline(always)]
86    #[track_caller]
87    fn lock_or_panic(&self) -> MutexGuard<'_, T> {
88        #[allow(clippy::unwrap_used)]
89        self.lock().unwrap()
90    }
91}
92
93/// Extension trait for `RwLock` to provide methods that acquire read/write locks, panicking if
94/// the lock is poisoned.
95///
96/// Mirrors [`MutexExt`] for `RwLock` so callers avoid `#[allow(clippy::unwrap_used)]` at each
97/// lock site.
98///
99/// # Examples
100///
101/// ```
102/// use libdd_common::RwLockExt;
103/// use std::sync::{Arc, RwLock};
104///
105/// let data = Arc::new(RwLock::new(5));
106/// let data_clone = Arc::clone(&data);
107///
108/// std::thread::spawn(move || {
109///     let mut num = data_clone.write_or_panic();
110///     *num += 1;
111/// })
112/// .join()
113/// .expect("Thread panicked");
114///
115/// assert_eq!(*data.read_or_panic(), 6);
116/// ```
117pub trait RwLockExt<T> {
118    fn read_or_panic(&self) -> RwLockReadGuard<'_, T>;
119    fn write_or_panic(&self) -> RwLockWriteGuard<'_, T>;
120}
121
122impl<T> RwLockExt<T> for RwLock<T> {
123    #[inline(always)]
124    #[track_caller]
125    fn read_or_panic(&self) -> RwLockReadGuard<'_, T> {
126        #[allow(clippy::unwrap_used)]
127        self.read().unwrap()
128    }
129
130    #[inline(always)]
131    #[track_caller]
132    fn write_or_panic(&self) -> RwLockWriteGuard<'_, T> {
133        #[allow(clippy::unwrap_used)]
134        self.write().unwrap()
135    }
136}
137
138pub mod header {
139    #![allow(clippy::declare_interior_mutable_const)]
140    use http::{header::HeaderName, HeaderValue};
141
142    pub const APPLICATION_MSGPACK_STR: &str = "application/msgpack";
143    pub const APPLICATION_PROTOBUF_STR: &str = "application/x-protobuf";
144
145    pub const DATADOG_CONTAINER_ID: HeaderName = HeaderName::from_static("datadog-container-id");
146    pub const DATADOG_ENTITY_ID: HeaderName = HeaderName::from_static("datadog-entity-id");
147    pub const DATADOG_EXTERNAL_ENV: HeaderName = HeaderName::from_static("datadog-external-env");
148    pub const DATADOG_TRACE_COUNT: HeaderName = HeaderName::from_static("x-datadog-trace-count");
149    /// Signal to the agent to send 429 responses when a payload is dropped
150    /// If this is not set then the agent will always return a 200 regardless if the payload is
151    /// dropped.
152    pub const DATADOG_SEND_REAL_HTTP_STATUS: HeaderName =
153        HeaderName::from_static("datadog-send-real-http-status");
154    pub const DATADOG_API_KEY: HeaderName = HeaderName::from_static("dd-api-key");
155    pub const APPLICATION_JSON: HeaderValue = HeaderValue::from_static("application/json");
156    pub const APPLICATION_MSGPACK: HeaderValue = HeaderValue::from_static(APPLICATION_MSGPACK_STR);
157    pub const APPLICATION_PROTOBUF: HeaderValue =
158        HeaderValue::from_static(APPLICATION_PROTOBUF_STR);
159    pub const X_DATADOG_TEST_SESSION_TOKEN: HeaderName =
160        HeaderName::from_static("x-datadog-test-session-token");
161}
162
163#[cfg(not(target_arch = "wasm32"))]
164pub type HttpClient = http_common::GenericHttpClient<connector::Connector>;
165#[cfg(not(target_arch = "wasm32"))]
166pub type HttpResponse = http_common::HttpResponse;
167pub type HttpRequestBuilder = http::request::Builder;
168#[cfg(not(target_arch = "wasm32"))]
169pub trait Connect:
170    hyper_util::client::legacy::connect::Connect + Clone + Send + Sync + 'static
171{
172}
173#[cfg(not(target_arch = "wasm32"))]
174impl<C: hyper_util::client::legacy::connect::Connect + Clone + Send + Sync + 'static> Connect
175    for C
176{
177}
178
179// Used by tag! macro
180pub use const_format;
181
182#[derive(Clone, PartialEq, Eq, Hash, Debug, Serialize, Deserialize)]
183pub struct Endpoint {
184    #[serde(serialize_with = "serialize_uri", deserialize_with = "deserialize_uri")]
185    pub url: http::Uri,
186    pub api_key: Option<Cow<'static, str>>,
187    pub timeout_ms: u64,
188    /// Sets X-Datadog-Test-Session-Token header on any request
189    pub test_token: Option<Cow<'static, str>>,
190    /// Use the system DNS resolver when building the HTTP client. If false, the default
191    /// in-process resolver is used.
192    #[serde(default)]
193    pub use_system_resolver: bool,
194}
195
196impl Default for Endpoint {
197    fn default() -> Self {
198        Endpoint {
199            url: http::Uri::default(),
200            api_key: None,
201            timeout_ms: Self::DEFAULT_TIMEOUT,
202            test_token: None,
203            use_system_resolver: false,
204        }
205    }
206}
207
208#[derive(serde::Deserialize, serde::Serialize)]
209struct SerializedUri<'a> {
210    scheme: Option<Cow<'a, str>>,
211    authority: Option<Cow<'a, str>>,
212    path_and_query: Option<Cow<'a, str>>,
213}
214
215fn serialize_uri<S>(uri: &http::Uri, serializer: S) -> Result<S::Ok, S::Error>
216where
217    S: Serializer,
218{
219    let parts = uri.clone().into_parts();
220    let uri = SerializedUri {
221        scheme: parts.scheme.as_ref().map(|s| Cow::Borrowed(s.as_str())),
222        authority: parts.authority.as_ref().map(|s| Cow::Borrowed(s.as_str())),
223        path_and_query: parts
224            .path_and_query
225            .as_ref()
226            .map(|s| Cow::Borrowed(s.as_str())),
227    };
228    uri.serialize(serializer)
229}
230
231fn deserialize_uri<'de, D>(deserializer: D) -> Result<http::Uri, D::Error>
232where
233    D: Deserializer<'de>,
234{
235    let uri = SerializedUri::deserialize(deserializer)?;
236    let mut builder = http::Uri::builder();
237    if let Some(v) = uri.authority {
238        builder = builder.authority(v.deref());
239    }
240    if let Some(v) = uri.scheme {
241        builder = builder.scheme(v.deref());
242    }
243    if let Some(v) = uri.path_and_query {
244        builder = builder.path_and_query(v.deref());
245    }
246
247    builder.build().map_err(Error::custom)
248}
249
250/// TODO: we should properly handle malformed urls
251/// * For windows and unix schemes:
252///     * For compatibility reasons with existing implementation this parser stores the encoded path
253///       in authority section as there is no existing standard [see](https://github.com/whatwg/url/issues/577)
254///       that covers this. We need to pick one hack or another
255///     * For windows, interprets everything after windows: as path
256///     * For unix, interprets everything after unix:// as path
257/// * For file scheme implementation will simply backfill missing authority section
258pub fn parse_uri(uri: &str) -> anyhow::Result<http::Uri> {
259    if let Some(path) = uri.strip_prefix("unix://") {
260        encode_uri_path_in_authority("unix", path)
261    } else if let Some(path) = uri.strip_prefix("windows:") {
262        encode_uri_path_in_authority("windows", path)
263    } else if let Some(path) = uri.strip_prefix("file://") {
264        encode_uri_path_in_authority("file", path)
265    } else {
266        Ok(http::Uri::from_str(uri)?)
267    }
268}
269
270fn encode_uri_path_in_authority(scheme: &str, path: &str) -> anyhow::Result<http::Uri> {
271    let mut parts = uri::Parts::default();
272    parts.scheme = uri::Scheme::from_str(scheme).ok();
273
274    let path = hex::encode(path);
275
276    parts.authority = uri::Authority::from_str(path.as_str()).ok();
277    parts.path_and_query = Some(uri::PathAndQuery::from_static(""));
278    Ok(http::Uri::from_parts(parts)?)
279}
280
281pub fn decode_uri_path_in_authority(uri: &http::Uri) -> anyhow::Result<PathBuf> {
282    let path = hex::decode(uri.authority().context("missing uri authority")?.as_str())?;
283    #[cfg(unix)]
284    {
285        use std::os::unix::ffi::OsStringExt;
286        Ok(PathBuf::from(std::ffi::OsString::from_vec(path)))
287    }
288    #[cfg(not(unix))]
289    {
290        match String::from_utf8(path) {
291            Ok(s) => Ok(PathBuf::from(s.as_str())),
292            _ => Err(anyhow::anyhow!("file uri should be utf-8")),
293        }
294    }
295}
296
297impl Endpoint {
298    /// Default value for the timeout field in milliseconds.
299    pub const DEFAULT_TIMEOUT: u64 = 3_000;
300
301    /// Returns an iterator of optional endpoint-specific headers (api-key, test-token)
302    /// as (header_name, header_value) string tuples for any that are available.
303    pub fn get_optional_headers(&self) -> impl Iterator<Item = (&'static str, &str)> {
304        [
305            self.api_key.as_ref().map(|v| ("dd-api-key", v.as_ref())),
306            self.test_token
307                .as_ref()
308                .map(|v| ("x-datadog-test-session-token", v.as_ref())),
309        ]
310        .into_iter()
311        .flatten()
312    }
313
314    /// Apply standard headers (user-agent, api-key, test-token, entity headers) to an
315    /// [`http::request::Builder`].
316    pub fn set_standard_headers(
317        &self,
318        mut builder: http::request::Builder,
319        user_agent: &str,
320    ) -> http::request::Builder {
321        builder = builder.header("user-agent", user_agent);
322        for (name, value) in self.get_optional_headers() {
323            builder = builder.header(name, value);
324        }
325        for (name, value) in entity_id::get_entity_headers() {
326            builder = builder.header(name, value);
327        }
328        builder
329    }
330
331    /// Return a request builder with the following headers:
332    /// - User agent
333    /// - Api key
334    /// - Container Id/Entity Id
335    pub fn to_request_builder(&self, user_agent: &str) -> anyhow::Result<HttpRequestBuilder> {
336        let mut builder = http::Request::builder()
337            .uri(self.url.clone())
338            .header(http::header::USER_AGENT, user_agent);
339
340        // Add optional endpoint headers (api-key, test-token)
341        for (name, value) in self.get_optional_headers() {
342            builder = builder.header(name, value);
343        }
344
345        // Add entity-related headers (container-id, entity-id, external-env)
346        for (name, value) in entity_id::get_entity_headers() {
347            builder = builder.header(name, value);
348        }
349
350        Ok(builder)
351    }
352
353    #[inline]
354    pub fn from_slice(url: &str) -> Endpoint {
355        Endpoint {
356            #[allow(clippy::unwrap_used)]
357            url: parse_uri(url).unwrap(),
358            ..Default::default()
359        }
360    }
361
362    #[inline]
363    pub fn from_url(url: http::Uri) -> Endpoint {
364        Endpoint {
365            url,
366            ..Default::default()
367        }
368    }
369
370    pub fn is_file_endpoint(&self) -> bool {
371        self.url.scheme_str() == Some("file")
372    }
373
374    /// Set a custom timeout for this endpoint.
375    /// If not called, uses the default timeout of 3000ms.
376    ///
377    /// # Arguments
378    /// * `timeout_ms` - Timeout in milliseconds. Pass 0 to use the default timeout (3000ms).
379    ///
380    /// # Returns
381    /// Self with the timeout set, allowing for method chaining
382    pub fn with_timeout(mut self, timeout_ms: u64) -> Self {
383        self.timeout_ms = if timeout_ms == 0 {
384            Self::DEFAULT_TIMEOUT
385        } else {
386            timeout_ms
387        };
388        self
389    }
390
391    /// Use the system DNS resolver when building the reqwest client. Only has effect for
392    /// HTTP(S) endpoints.
393    pub fn with_system_resolver(mut self, use_system_resolver: bool) -> Self {
394        self.use_system_resolver = use_system_resolver;
395        self
396    }
397
398    /// Creates a reqwest ClientBuilder configured for this endpoint.
399    ///
400    /// This method handles various endpoint schemes:
401    /// - `http`/`https`: Standard HTTP(S) endpoints
402    /// - `unix`: Unix domain sockets (Unix only)
403    /// - `windows`: Windows named pipes (Windows only)
404    /// - `file`: File dump endpoints for debugging (spawns a local server to capture requests)
405    ///
406    /// The default in-process resolver is used for DNS (fork-safe). To use the system DNS resolver
407    /// instead (less fork-safe), set [`Endpoint::use_system_resolver`] to true via
408    /// [`Endpoint::with_system_resolver`].
409    ///
410    /// # Returns
411    /// A tuple of (ClientBuilder, request_url) where:
412    /// - ClientBuilder is configured with the appropriate transport and timeout
413    /// - request_url is the URL string to use for HTTP requests
414    ///
415    /// # Errors
416    /// Returns an error if:
417    /// - The endpoint scheme is unsupported
418    /// - Path decoding fails
419    /// - The dump server fails to start (for file:// scheme)
420    #[cfg(feature = "reqwest")]
421    pub fn to_reqwest_client_builder(&self) -> anyhow::Result<(reqwest::ClientBuilder, String)> {
422        use anyhow::Context;
423
424        // Don't use proxies, as this calls `getenv` which is unsafe and not
425        // just in theory. It can cause crashes with PHP where php-fpm's env
426        // configuration will mutate the system environment (it doesn't pass
427        // it as part of the SAPI env, it changes the actual system env).
428        let mut builder = reqwest::Client::builder()
429            .timeout(std::time::Duration::from_millis(self.timeout_ms))
430            .hickory_dns(!self.use_system_resolver)
431            .no_proxy();
432
433        let request_url = match self.url.scheme_str() {
434            // HTTP/HTTPS endpoints
435            Some("http") | Some("https") => self.url.to_string(),
436
437            // File dump endpoint (debugging) - uses platform-specific local transport
438            Some("file") => {
439                let output_path = decode_uri_path_in_authority(&self.url)
440                    .context("Failed to decode file path from URI")?;
441                let socket_or_pipe_path = dump_server::spawn_dump_server(output_path)?;
442
443                // Configure the client to use the local socket/pipe
444                #[cfg(unix)]
445                {
446                    builder = builder.unix_socket(socket_or_pipe_path);
447                }
448                #[cfg(windows)]
449                {
450                    builder = builder
451                        .windows_named_pipe(socket_or_pipe_path.to_string_lossy().to_string());
452                }
453
454                "http://localhost/".to_string()
455            }
456
457            // Unix domain sockets
458            #[cfg(unix)]
459            Some("unix") => {
460                use connector::uds::socket_path_from_uri;
461                let socket_path = socket_path_from_uri(&self.url)?;
462                builder = builder.unix_socket(socket_path);
463                format!("http://localhost{}", self.url.path())
464            }
465
466            // Windows named pipes
467            #[cfg(windows)]
468            Some("windows") => {
469                use connector::named_pipe::named_pipe_path_from_uri;
470                let pipe_path = named_pipe_path_from_uri(&self.url)?;
471                builder = builder.windows_named_pipe(pipe_path.to_string_lossy().to_string());
472                format!("http://localhost{}", self.url.path())
473            }
474
475            // Unsupported schemes
476            scheme => anyhow::bail!("Unsupported endpoint scheme: {:?}", scheme),
477        };
478
479        Ok((builder, request_url))
480    }
481}
libdd_common/lib.rs

libdd_common/
lib.rs
