Compare commits

...

7 commits

11 changed files with 942 additions and 593 deletions

View file

@ -1,6 +1,6 @@
[package]
name = "acrate-hostmeta"
version = "0.1.0"
version = "0.2.0"
edition = "2021"
[dependencies]

286
acrate-hostmeta/src/any.rs Normal file
View file

@ -0,0 +1,286 @@
use thiserror::Error;
use crate::jrd::{GetJRDError, ResourceDescriptorJRD};
use crate::xrd::{GetXRDError, ResourceDescriptorXRD};
#[derive(Debug, Clone)]
pub enum ResourceDescriptor {
/// The resource descriptor is in JRD format.
JRD(ResourceDescriptorJRD),
/// The resource descriptor is in XRD format.
XRD(ResourceDescriptorXRD),
}
impl ResourceDescriptor {
/// Attempt to get a [`ResourceDescriptor`] at the given URL in various ways.
///
/// In order, this method attempts:
///
/// 1. HTTPS [XRD](ResourceDescriptorJRD::get)
/// 2. HTTPS [JRD](ResourceDescriptorJRD::get)
/// 3. HTTPS [JRD](ResourceDescriptorJRD::get) with .json path suffix
/// 4. HTTP [XRD](ResourceDescriptorJRD::get)
/// 5. HTTP [JRD](ResourceDescriptorJRD::get)
/// 6. HTTP [JRD](ResourceDescriptorJRD::get) with .json path suffix
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::any::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/host-meta".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::get(&client, url)
/// .await
/// .expect("resource descriptor to be discovered correctly");
/// # })
/// ```
///
pub async fn get(client: &reqwest::Client, mut url: reqwest::Url) -> Result<Self, GetError> {
use GetError::*;
log::debug!("Discovering resource descriptor document at: {url}");
log::trace!("Unsetting URL query...");
url.set_query(None);
log::trace!("Unsetting URL fragment...");
url.set_fragment(None);
log::trace!("Setting URL scheme to HTTPS...");
url.set_scheme("https")
.map_err(UrlManipulation)?;
log::trace!("Cloning URL for HTTPS XRD retrieval...");
let https_xrd_url = url.clone();
log::trace!("Attempting HTTPS XRD retrieval...");
let https_xrd = match ResourceDescriptorXRD::get(client, https_xrd_url).await {
Ok(data) => {
log::trace!("HTTPS XRD retrieval was successful, returning...");
return Ok(Self::XRD(data))
}
Err(err) => {
log::warn!("HTTPS XRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTPS JRD retrieval...");
let https_jrd_url = url.clone();
log::trace!("Attempting HTTPS JRD retrieval...");
let https_jrd = match ResourceDescriptorJRD::get(client, https_jrd_url).await {
Ok(data) => {
log::trace!("HTTPS JRD retrieval was successful, returning...");
return Ok(Self::JRD(data))
}
Err(err) => {
log::warn!("HTTPS JRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTPS JRD .json retrieval...");
let mut https_jrdj_url = url.clone();
log::trace!("Altering URL path for HTTPS JRD .json retrieval...");
https_jrdj_url.set_path(
&format!("{}.json", https_jrdj_url.path())
);
log::trace!("Attempting HTTPS JRD .json retrieval...");
let https_jrdj = match ResourceDescriptorJRD::get(client, https_jrdj_url).await {
Ok(data) => {
log::trace!("HTTPS JRD .json retrieval was successful, returning...");
return Ok(Self::JRD(data))
}
Err(err) => {
log::warn!("HTTPS JRD .json retrieval failed.");
err
}
};
log::trace!("Setting URL scheme to HTTP...");
url.set_scheme("http")
.map_err(UrlManipulation)?;
log::trace!("Cloning URL for HTTP XRD retrieval...");
let http_xrd_url = url.clone();
log::trace!("Attempting HTTP XRD retrieval...");
let http_xrd = match ResourceDescriptorXRD::get(client, http_xrd_url).await {
Ok(data) => {
log::trace!("HTTP XRD retrieval was successful, returning...");
return Ok(Self::XRD(data))
}
Err(err) => {
log::warn!("HTTP XRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTP JRD retrieval...");
let http_jrd_url = url.clone();
log::trace!("Attempting HTTP JRD retrieval...");
let http_jrd = match ResourceDescriptorJRD::get(client, http_jrd_url).await {
Ok(data) => {
log::trace!("HTTP JRD retrieval was successful, returning...");
return Ok(Self::JRD(data))
}
Err(err) => {
log::warn!("HTTP JRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTP JRD .json retrieval...");
let mut http_jrdj_url = url.clone();
log::trace!("Altering URL path for HTTPS JRD .json retrieval...");
http_jrdj_url.set_path(
&format!("{}.json", http_jrdj_url.path())
);
log::trace!("Attempting HTTP JRD .json retrieval...");
let http_jrdj = match ResourceDescriptorJRD::get(client, http_jrdj_url).await {
Ok(data) => {
log::trace!("HTTP JRD .json retrieval was successful, returning...");
return Ok(Self::JRD(data))
}
Err(err) => {
log::warn!("HTTP JRD .json retrieval failed.");
err
}
};
Err(Fetch(GetFetchFailures {
https_xrd,
https_jrd,
https_jrdj,
http_xrd,
http_jrd,
http_jrdj,
}))
}
/// Well-known path for host-meta documents.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-2>
///
pub const WELLKNOWN_HOSTMETA_PATH: &'static str = "/.well-known/host-meta";
/// Attempt to discover a host-meta document at the given base URL.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::any::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let base: reqwest::Url = "https://junimo.party".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::discover_hostmeta(&client, base)
/// .await
/// .expect("host-meta to be discovered correctly");
/// })
/// ```
///
pub async fn discover_hostmeta(client: &reqwest::Client, mut base: reqwest::Url) -> Result<Self, GetError> {
log::debug!("Attempting discovery of host-meta document of: {base}");
log::trace!("Adding well-known path to the given base URL: {}", Self::WELLKNOWN_HOSTMETA_PATH);
base.set_path(Self::WELLKNOWN_HOSTMETA_PATH);
log::trace!("Starting discovery at: {base}");
Self::get(client, base)
.await
}
pub fn jrd(self) -> ResourceDescriptorJRD {
self.into()
}
pub fn xrd(self) -> ResourceDescriptorXRD {
self.into()
}
}
impl From<ResourceDescriptorJRD> for ResourceDescriptor {
fn from(value: ResourceDescriptorJRD) -> Self {
Self::JRD(value)
}
}
impl From<ResourceDescriptorXRD> for ResourceDescriptor {
fn from(value: ResourceDescriptorXRD) -> Self {
Self::XRD(value)
}
}
impl From<ResourceDescriptor> for ResourceDescriptorJRD {
fn from(value: ResourceDescriptor) -> Self {
match value {
ResourceDescriptor::JRD(jrd) => jrd,
ResourceDescriptor::XRD(xrd) => xrd.into(),
}
}
}
impl From<ResourceDescriptor> for ResourceDescriptorXRD {
fn from(value: ResourceDescriptor) -> Self {
match value {
ResourceDescriptor::JRD(jrd) => jrd.into(),
ResourceDescriptor::XRD(xrd) => xrd,
}
}
}
/// Error occurred during [`ResourceDescriptor::discover`].
#[derive(Debug, Error)]
pub enum GetError {
/// Manipulation of the provided base [`reqwest::Url`] failed.
///
/// See [reqwest::Url::set_scheme] for possible causes.
#[error("manipulation of the provided URL failed")]
UrlManipulation(()),
/// All attempts of fetching a resource descriptor document failed.
#[error("fetchign the resource descriptor document failed")]
Fetch(GetFetchFailures),
}
/// Request errors occurred during [`ResourceDescriptor::discover`].
#[derive(Debug, Error)]
#[error("all attempts of fetching the resource descriptor document failed")]
pub struct GetFetchFailures {
/// HTTPS XRD retrieval.
pub https_xrd: GetXRDError,
/// HTTPS JRD retrieval.
pub https_jrd: GetJRDError,
/// HTTPS JRD with .json extension retrieval.
pub https_jrdj: GetJRDError,
/// HTTPS XRD retrieval.
pub http_xrd: GetXRDError,
/// HTTP JRD retrieval.
pub http_jrd: GetJRDError,
/// HTTP JRD with .json extension retrieval.
pub http_jrdj: GetJRDError,
}

251
acrate-hostmeta/src/jrd.rs Normal file
View file

@ -0,0 +1,251 @@
use std::collections::HashMap;
use serde::{Serialize, Deserialize};
use thiserror::Error;
use crate::xrd::{ResourceDescriptorLinkXRD, ResourceDescriptorPropertyXRD, ResourceDescriptorTitleXRD, ResourceDescriptorXRD};
/// A resource descriptor object in JRD format.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4>
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorJRD {
/// The resource this document refers to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.1>
///
pub subject: Option<String>,
/// Other names the resource described by this document can be referred to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.2>
///
#[serde(default)]
pub aliases: Vec<String>,
/// Additional information about the resource described by this document.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.3>
///
#[serde(default)]
pub properties: HashMap<String, Option<String>>,
/// Links established between the [`Self::subject`] and other resources.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4>
///
#[serde(default)]
pub links: Vec<ResourceDescriptorLinkJRD>,
}
/// A link element, which puts the subject resource in relation with another.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
///
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorLinkJRD {
/// The kind of relation established by the subject with the attached resource.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.1>
///
pub rel: String,
/// The media type of the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.2>
///
pub r#type: Option<String>,
/// URI to the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.3>
///
pub href: Option<String>,
/// Titles of the resource put in relation in various languages.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.4>
///
#[serde(default)]
pub titles: HashMap<String, String>,
/// Additional information about the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.5>
///
#[serde(default)]
pub properties: HashMap<String, Option<String>>,
/// Template to fill to get the URL to resource-specific information.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-4.2>
///
pub template: Option<String>,
}
impl ResourceDescriptorJRD {
/// Get a [`ResourceDescriptorJRD`] from an URL.
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::jrd::ResourceDescriptorJRD;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/nodeinfo".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptorJRD::get(&client, url)
/// .await
/// .expect("JRD to be processed correctly");
/// # })
/// ```
///
pub async fn get(client: &reqwest::Client, url: reqwest::Url) -> Result<Self, GetJRDError> {
use GetJRDError::*;
log::debug!("Getting JRD document at: {url}");
log::trace!("Building request...");
let request = {
log::trace!("Creating new request...");
let mut request = reqwest::Request::new(reqwest::Method::GET, url);
log::trace!("Setting request headers...");
let headers = request.headers_mut();
log::trace!("Setting `Accept: application/jrd+json, application/json`...");
let _ = headers.insert(
reqwest::header::ACCEPT,
"application/jrd+json, application/json".parse().unwrap()
);
request
};
log::trace!("Sending request...");
let response = client.execute(request)
.await
.map_err(Request)?;
log::trace!("Checking `Content-Type` of the response...");
let content_type = response
.headers()
.get(reqwest::header::CONTENT_TYPE)
.ok_or(ContentTypeMissing)?;
log::trace!("Extracting MIME type from the `Content-Type` header...");
let mime_type = crate::utils::extract_mime_from_content_type(content_type)
.ok_or(ContentTypeInvalid)?;
log::trace!("Ensuring MIME type is acceptable for JRD parsing...");
if !(mime_type == "application/json" || mime_type == "application/jrd+json") {
log::error!("MIME type `{mime_type}` is not acceptable for JRD parsing.");
return Err(ContentTypeInvalid);
}
log::trace!("Attempting to parse response as JSON...");
let data = response.json::<Self>()
.await
.map_err(Parse)?;
Ok(data)
}
}
impl From<ResourceDescriptorXRD> for ResourceDescriptorJRD {
fn from(value: ResourceDescriptorXRD) -> Self {
Self {
subject: value.subject,
aliases: value.aliases,
properties: value.properties.into_iter()
.map(From::from)
.collect(),
links: value.links.into_iter()
.map(From::from)
.collect(),
}
}
}
impl From<ResourceDescriptorLinkXRD> for ResourceDescriptorLinkJRD {
fn from(value: ResourceDescriptorLinkXRD) -> Self {
Self {
rel: value.rel,
r#type: value.r#type,
href: value.href,
titles: HashMap::from_iter(
value.titles.into_iter()
.map(From::from)
),
properties: HashMap::from_iter(
value.properties.into_iter()
.map(From::from)
),
template: value.template,
}
}
}
impl From<ResourceDescriptorPropertyXRD> for (String, Option<String>) {
fn from(value: ResourceDescriptorPropertyXRD) -> Self {
(value.r#type, value.value)
}
}
impl From<ResourceDescriptorTitleXRD> for (String, String) {
fn from(value: ResourceDescriptorTitleXRD) -> Self {
(value.language, value.value)
}
}
/// Error occurred during [`ResourceDescriptor::get_jrd`].
#[derive(Debug, Error)]
pub enum GetJRDError {
/// The HTTP request failed.
#[error("the HTTP request failed")]
Request(reqwest::Error),
/// The `Content-Type` header of the response is missing.
#[error("the Content-Type header of the response is missing")]
ContentTypeMissing,
/// The `Content-Type` header of the response is invalid.
#[error("the Content-Type header of the response is invalid")]
ContentTypeInvalid,
/// The document failed to be parsed as JSON by [`reqwest`].
#[error("the document failed to be parsed as JSON")]
Parse(reqwest::Error),
}

View file

@ -1,568 +1,12 @@
//! Resource descriptior deserializer.
//! Resource descriptior handler.
//!
//! # Specification
//!
//! - <https://datatracker.ietf.org/doc/html/rfc6415>
//! - <https://datatracker.ietf.org/doc/html/rfc7033>
use std::collections::HashMap;
pub mod jrd;
pub mod xrd;
pub mod any;
use serde::Deserialize;
use thiserror::Error;
/// A resource descriptor object.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4>
///
#[derive(Debug, Clone, Deserialize)]
pub struct ResourceDescriptor {
/// The resource this document refers to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.1>
///
#[serde(alias = "Subject")]
pub subject: Option<String>,
/// Other names the resource described by this document can be referred to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.2>
///
#[serde(alias = "Alias")]
pub aliases: Option<Vec<String>>,
/// Additional information about the resource described by this document.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.3>
///
#[serde(alias = "Property")]
pub properties: Option<Vec<ResourceDescriptorProperty>>,
/// Links established between the [`Self::subject`] and other resources.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4>
///
#[serde(alias = "Link")]
pub links: Option<Vec<ResourceDescriptorLink>>,
}
/// A link element, which puts the subject resource in relation with another.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
///
#[derive(Debug, Clone, Deserialize)]
pub struct ResourceDescriptorLink {
/// The kind of relation established by the subject with the attached resource.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.1>
///
#[serde(alias = "@rel")]
pub rel: String,
/// The media type of the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.2>
///
#[serde(alias = "@type")]
pub r#type: Option<String>,
/// URI to the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.3>
///
#[serde(alias = "@href")]
pub href: Option<String>,
/// Titles of the resource put in relation in various languages.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.4>
///
pub titles: Option<Vec<HashMap<String, String>>>,
/// Additional information about the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.5>
///
pub properties: Option<Vec<ResourceDescriptorProperty>>,
/// Template to fill to get the URL to resource-specific information.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-4.2>
///
#[serde(alias = "@template")]
pub template: Option<String>,
}
/// A property element, which describes a certain aspect of the subject resource.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.3>
///
#[derive(Debug, Clone, Deserialize)]
pub struct ResourceDescriptorProperty {
/// The property identifier, or type.
#[serde(alias = "@type")]
pub r#type: String,
/// The property value.
pub value: Option<String>,
}
impl ResourceDescriptor {
/// Get a JRD (JSON [`ResourceDescriptor`]).
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/nodeinfo".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::get_jrd(&client, url)
/// .await
/// .expect("JRD to be processed correctly");
/// # })
/// ```
///
pub async fn get_jrd(client: &reqwest::Client, url: reqwest::Url) -> Result<Self, GetJRDError> {
use GetJRDError::*;
log::debug!("Getting JRD document at: {url}");
log::trace!("Building request...");
let request = {
log::trace!("Creating new request...");
let mut request = reqwest::Request::new(reqwest::Method::GET, url);
log::trace!("Setting request headers...");
let headers = request.headers_mut();
log::trace!("Setting `Accept: application/json`...");
let _ = headers.insert(reqwest::header::ACCEPT, "application/json".parse().unwrap());
request
};
log::trace!("Sending request...");
let response = client.execute(request)
.await
.map_err(Request)?;
log::trace!("Checking `Content-Type` of the response...");
let content_type = response
.headers()
.get(reqwest::header::CONTENT_TYPE)
.ok_or(ContentTypeMissing)?;
log::trace!("Extracting MIME type from the `Content-Type` header...");
let mime_type = extract_mime_from_content_type(content_type)
.ok_or(ContentTypeInvalid)?;
log::trace!("Ensuring MIME type is acceptable for JRD parsing...");
if !(mime_type == "application/json" || mime_type == "application/jrd+json") {
log::error!("MIME type `{mime_type}` is not acceptable for JRD parsing.");
return Err(ContentTypeInvalid)
}
log::trace!("Attempting to parse response as JSON...");
let data = response.json::<Self>()
.await
.map_err(Parse)?;
Ok(data)
}
/// Get a XRD (Extensible [`ResourceDescriptor`]).
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/host-meta".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::get_xrd(&client, url)
/// .await
/// .expect("XRD to be processed correctly");
/// # })
/// ```
///
pub async fn get_xrd(client: &reqwest::Client, url: reqwest::Url) -> Result<Self, GetXRDError> {
use GetXRDError::*;
log::debug!("Getting host-meta XRD document at: {url}");
log::trace!("Building request...");
let request = {
log::trace!("Creating new request...");
let mut request = reqwest::Request::new(reqwest::Method::GET, url);
log::trace!("Setting request headers...");
let headers = request.headers_mut();
log::trace!("Setting `Accept: application/xrd+xml`...");
let _ = headers.insert(reqwest::header::ACCEPT, "application/xrd+xml".parse().unwrap());
request
};
log::trace!("Sending request...");
let response = client.execute(request)
.await
.map_err(Request)?;
log::trace!("Checking `Content-Type` of the response...");
let content_type = response
.headers()
.get(reqwest::header::CONTENT_TYPE)
.ok_or(ContentTypeMissing)?;
log::trace!("Extracting MIME type from the `Content-Type` header...");
let mime_type = extract_mime_from_content_type(content_type)
.ok_or(ContentTypeInvalid)?;
log::trace!("Ensuring MIME type is acceptable for XRD parsing...");
if mime_type != "application/xrd+xml" {
log::error!("MIME type `{mime_type}` is not acceptable for XRD parsing.");
return Err(ContentTypeInvalid)
}
log::trace!("Attempting to parse response as text...");
let data = response.text()
.await
.map_err(Decode)?;
log::trace!("Parsing response as XML...");
let data = quick_xml::de::from_str::<Self>(&data)
.map_err(Parse)?;
Ok(data)
}
/// Attempt to discover a [`ResourceDescriptor`] at the given URL in various ways.
///
/// In order, this method attempts:
///
/// 1. HTTPS [XRD](Self::get_xrd)
/// 2. HTTPS [JRD](Self::get_jrd)
/// 3. HTTPS [JRD](Self::get_jrd) with .json path extension
/// 4. HTTP [XRD](Self::get_xrd)
/// 5. HTTP [JRD](Self::get_jrd)
/// 6. HTTP [JRD](Self::get_jrd) with .json path extension
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/host-meta".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::discover(&client, url)
/// .await
/// .expect("resource descriptor to be discovered correctly");
/// # })
/// ```
///
pub async fn discover(client: &reqwest::Client, mut url: reqwest::Url) -> Result<Self, ResourceDescriptorDiscoveryError> {
use ResourceDescriptorDiscoveryError::*;
log::debug!("Discovering resource descriptor document at: {url}");
log::trace!("Unsetting URL query...");
url.set_query(None);
log::trace!("Unsetting URL fragment...");
url.set_fragment(None);
log::trace!("Setting URL scheme to HTTPS...");
url.set_scheme("https")
.map_err(UrlManipulation)?;
log::trace!("Cloning URL for HTTPS XRD retrieval...");
let https_xrd_url = url.clone();
log::trace!("Attempting HTTPS XRD retrieval...");
let https_xrd = match Self::get_xrd(client, https_xrd_url).await {
Ok(data) => {
log::trace!("HTTPS XRD retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTPS XRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTPS JRD retrieval...");
let https_jrd_url = url.clone();
log::trace!("Attempting HTTPS JRD retrieval...");
let https_jrd = match Self::get_jrd(client, https_jrd_url).await {
Ok(data) => {
log::trace!("HTTPS JRD retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTPS JRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTPS JRD .json retrieval...");
let mut https_jrdj_url = url.clone();
log::trace!("Altering URL path for HTTPS JRD .json retrieval...");
https_jrdj_url.set_path(
&format!("{}.json", https_jrdj_url.path())
);
log::trace!("Attempting HTTPS JRD .json retrieval...");
let https_jrdj = match Self::get_jrd(client, https_jrdj_url).await {
Ok(data) => {
log::trace!("HTTPS JRD .json retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTPS JRD .json retrieval failed.");
err
}
};
log::trace!("Setting URL scheme to HTTP...");
url.set_scheme("http")
.map_err(UrlManipulation)?;
log::trace!("Cloning URL for HTTP XRD retrieval...");
let http_xrd_url = url.clone();
log::trace!("Attempting HTTP XRD retrieval...");
let http_xrd = match Self::get_xrd(client, http_xrd_url).await {
Ok(data) => {
log::trace!("HTTP XRD retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTP XRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTP JRD retrieval...");
let http_jrd_url = url.clone();
log::trace!("Attempting HTTP JRD retrieval...");
let http_jrd = match Self::get_jrd(client, http_jrd_url).await {
Ok(data) => {
log::trace!("HTTP JRD retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTP JRD retrieval failed.");
err
}
};
log::trace!("Cloning URL for HTTP JRD .json retrieval...");
let mut http_jrdj_url = url.clone();
log::trace!("Altering URL path for HTTPS JRD .json retrieval...");
http_jrdj_url.set_path(
&format!("{}.json", http_jrdj_url.path())
);
log::trace!("Attempting HTTP JRD .json retrieval...");
let http_jrdj = match Self::get_jrd(client, http_jrdj_url).await {
Ok(data) => {
log::trace!("HTTP JRD .json retrieval was successful, returning...");
return Ok(data)
}
Err(err) => {
log::warn!("HTTP JRD .json retrieval failed.");
err
}
};
Err(
ResourceDescriptorDiscoveryError::Fetch(
ResourceDescriptorDiscoveryFailures {
https_xrd,
https_jrd,
https_jrdj,
http_xrd,
http_jrd,
http_jrdj,
}
)
)
}
/// Well-known path for host-meta documents.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-2>
///
pub const WELLKNOWN_HOSTMETA_PATH: &str = "/.well-known/host-meta";
/// Attempt to discover a host-meta document at the given base URL.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::ResourceDescriptor;
///
/// let client = reqwest::Client::new();
/// let base: reqwest::Url = "https://junimo.party".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptor::discover_hostmeta(&client, base)
/// .await
/// .expect("host-meta to be discovered correctly");
/// })
/// ```
///
pub async fn discover_hostmeta(client: &reqwest::Client, mut base: reqwest::Url) -> Result<Self, ResourceDescriptorDiscoveryError> {
base.set_path(Self::WELLKNOWN_HOSTMETA_PATH);
Self::discover(client, base)
.await
}
}
/// Error occurred during [`ResourceDescriptor::discover`].
#[derive(Debug, Error)]
pub enum ResourceDescriptorDiscoveryError {
/// Manipulation of the provided base [`reqwest::Url`] failed.
///
/// See [reqwest::Url::set_scheme] for possible causes.
#[error("manipulation of the provided URL failed")]
UrlManipulation(()),
/// All attempts of fetching a resource descriptor document failed.
#[error("fetchign the resource descriptor document failed")]
Fetch(ResourceDescriptorDiscoveryFailures),
}
/// Request errors occurred during [`ResourceDescriptor::discover`].
#[derive(Debug, Error)]
#[error("all attempts of fetching the resource descriptor document failed")]
pub struct ResourceDescriptorDiscoveryFailures {
/// HTTPS XRD retrieval.
pub https_xrd: GetXRDError,
/// HTTPS JRD retrieval.
pub https_jrd: GetJRDError,
/// HTTPS JRD with .json extension retrieval.
pub https_jrdj: GetJRDError,
/// HTTPS XRD retrieval.
pub http_xrd: GetXRDError,
/// HTTP JRD retrieval.
pub http_jrd: GetJRDError,
/// HTTP JRD with .json extension retrieval.
pub http_jrdj: GetJRDError,
}
/// Error occurred during [`ResourceDescriptor::get_xrd`].
#[derive(Debug, Error)]
pub enum GetXRDError {
/// The HTTP request failed.
#[error("the HTTP request failed")]
Request(reqwest::Error),
/// The `Content-Type` header of the response is missing.
#[error("the Content-Type header of the response is missing")]
ContentTypeMissing,
/// The `Content-Type` header of the response is invalid.
#[error("the Content-Type header of the response is invalid")]
ContentTypeInvalid,
/// The document failed to be decoded as text.
#[error("the document failed to be decoded as text")]
Decode(reqwest::Error),
/// The document failed to be parsed as XML by [`quick_xml`].
#[error("the document failed to be parsed as XML")]
Parse(quick_xml::DeError),
}
/// Error occurred during [`ResourceDescriptor::get_jrd`].
#[derive(Debug, Error)]
pub enum GetJRDError {
/// The HTTP request failed.
#[error("the HTTP request failed")]
Request(reqwest::Error),
/// The `Content-Type` header of the response is missing.
#[error("the Content-Type header of the response is missing")]
ContentTypeMissing,
/// The `Content-Type` header of the response is invalid.
#[error("the Content-Type header of the response is invalid")]
ContentTypeInvalid,
/// The document failed to be parsed as JSON by [`reqwest`].
#[error("the document failed to be parsed as JSON")]
Parse(reqwest::Error),
}
/// Extract the MIME type from the value of the `Content-Type` header.
fn extract_mime_from_content_type(value: &reqwest::header::HeaderValue) -> Option<String> {
let value = value.to_str().ok()?;
match value.split_once("; ") {
None => Some(value.to_string()),
Some((mime, _)) => Some(mime.to_string()),
}
}
mod utils;

View file

@ -0,0 +1,8 @@
/// Extract the MIME type from the value of the `Content-Type` header.
pub fn extract_mime_from_content_type(value: &reqwest::header::HeaderValue) -> Option<String> {
let value = value.to_str().ok()?;
match value.split_once("; ") {
None => Some(value.to_string()),
Some((mime, _)) => Some(mime.to_string()),
}
}

292
acrate-hostmeta/src/xrd.rs Normal file
View file

@ -0,0 +1,292 @@
use serde::{Serialize, Deserialize};
use thiserror::Error;
use crate::jrd::{ResourceDescriptorJRD, ResourceDescriptorLinkJRD};
/// A resource descriptor object in XRD format.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4>
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorXRD {
/// The resource this document refers to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.1>
///
#[serde(rename = "Subject")]
pub subject: Option<String>,
/// Other names the resource described by this document can be referred to.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.2>
///
#[serde(rename = "Alias")]
#[serde(default)]
pub aliases: Vec<String>,
/// Additional information about the resource described by this document.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.3>
///
#[serde(rename = "Property")]
#[serde(default)]
pub properties: Vec<ResourceDescriptorPropertyXRD>,
/// Links established between the [`Self::subject`] and other resources.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4>
///
#[serde(rename = "Link")]
#[serde(default)]
pub links: Vec<ResourceDescriptorLinkXRD>,
}
/// A link element, which puts the subject resource in relation with another.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-3.1.1>
///
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorLinkXRD {
/// The kind of relation established by the subject with the attached resource.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.1>
///
#[serde(rename = "@rel")]
pub rel: String,
/// The media type of the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.2>
///
#[serde(rename = "@type")]
pub r#type: Option<String>,
/// URI to the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.3>
///
#[serde(rename = "@href")]
pub href: Option<String>,
/// Titles of the resource put in relation in various languages.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.4>
///
#[serde(default)]
pub titles: Vec<ResourceDescriptorTitleXRD>,
/// Additional information about the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.5>
///
#[serde(default)]
pub properties: Vec<ResourceDescriptorPropertyXRD>,
/// Template to fill to get the URL to resource-specific information.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc6415#section-4.2>
///
#[serde(rename = "@template")]
pub template: Option<String>,
}
/// A title of the resource put in relation.
///
/// # Specification
///
/// - <https://datatracker.ietf.org/doc/html/rfc7033#section-4.4.4.4>
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorTitleXRD {
/// The language of the title.
#[serde(rename = "@lang")]
pub language: String,
/// The title itself.
pub value: String,
}
#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct ResourceDescriptorPropertyXRD {
/// The property identifier, or type.
#[serde(alias = "@type")]
pub r#type: String,
/// The property value.
pub value: Option<String>,
}
impl ResourceDescriptorXRD {
/// Get a [`ResourceDescriptorXRD`] from an URL.
///
/// # Notes
///
/// This follows redirects until the redirect chain is 10 hops; see [`reqwest::redirect`] for more info.
///
/// # Examples
///
/// ```
/// # tokio_test::block_on(async {
/// use acrate_hostmeta::xrd::ResourceDescriptorXRD;
///
/// let client = reqwest::Client::new();
/// let url: reqwest::Url = "https://junimo.party/.well-known/host-meta".parse()
/// .expect("URL to be valid");
///
/// let rd = ResourceDescriptorXRD::get(&client, url)
/// .await
/// .expect("XRD to be processed correctly");
/// # })
/// ```
///
pub async fn get(client: &reqwest::Client, url: reqwest::Url) -> Result<Self, GetXRDError> {
use GetXRDError::*;
log::debug!("Getting host-meta XRD document at: {url}");
log::trace!("Building request...");
let request = {
log::trace!("Creating new request...");
let mut request = reqwest::Request::new(reqwest::Method::GET, url);
log::trace!("Setting request headers...");
let headers = request.headers_mut();
log::trace!("Setting `Accept: application/xrd+xml`...");
let _ = headers.insert(reqwest::header::ACCEPT, "application/xrd+xml".parse().unwrap());
request
};
log::trace!("Sending request...");
let response = client.execute(request)
.await
.map_err(Request)?;
log::trace!("Checking `Content-Type` of the response...");
let content_type = response
.headers()
.get(reqwest::header::CONTENT_TYPE)
.ok_or(ContentTypeMissing)?;
log::trace!("Extracting MIME type from the `Content-Type` header...");
let mime_type = crate::utils::extract_mime_from_content_type(content_type)
.ok_or(ContentTypeInvalid)?;
log::trace!("Ensuring MIME type is acceptable for XRD parsing...");
if mime_type != "application/xrd+xml" {
log::error!("MIME type `{mime_type}` is not acceptable for XRD parsing.");
return Err(ContentTypeInvalid)
}
log::trace!("Attempting to parse response as text...");
let data = response.text()
.await
.map_err(Decode)?;
log::trace!("Parsing response as XML...");
let data = quick_xml::de::from_str::<Self>(&data)
.map_err(Parse)?;
Ok(data)
}
}
impl From<ResourceDescriptorJRD> for ResourceDescriptorXRD {
fn from(value: ResourceDescriptorJRD) -> Self {
Self {
subject: value.subject,
aliases: value.aliases,
properties: value.properties.into_iter()
.map(From::from)
.collect(),
links: value.links.into_iter()
.map(From::from)
.collect(),
}
}
}
impl From<(String, Option<String>)> for ResourceDescriptorPropertyXRD {
fn from(value: (String, Option<String>)) -> Self {
Self {
r#type: value.0,
value: value.1,
}
}
}
impl From<ResourceDescriptorLinkJRD> for ResourceDescriptorLinkXRD {
fn from(value: ResourceDescriptorLinkJRD) -> Self {
Self {
rel: value.rel,
r#type: value.r#type,
href: value.href,
titles: value.titles.into_iter()
.map(From::from)
.collect(),
properties: value.properties.into_iter()
.map(From::from)
.collect(),
template: value.template,
}
}
}
impl From<(String, String)> for ResourceDescriptorTitleXRD {
fn from(value: (String, String)) -> Self {
Self {
language: value.0,
value: value.1,
}
}
}
/// Error occurred during [`ResourceDescriptor::get_xrd`].
#[derive(Debug, Error)]
pub enum GetXRDError {
/// The HTTP request failed.
#[error("the HTTP request failed")]
Request(reqwest::Error),
/// The `Content-Type` header of the response is missing.
#[error("the Content-Type header of the response is missing")]
ContentTypeMissing,
/// The `Content-Type` header of the response is invalid.
#[error("the Content-Type header of the response is invalid")]
ContentTypeInvalid,
/// The document failed to be decoded as text.
#[error("the document failed to be decoded as text")]
Decode(reqwest::Error),
/// The document failed to be parsed as XML by [`quick_xml`].
#[error("the document failed to be parsed as XML")]
Parse(quick_xml::DeError),
}

View file

@ -23,40 +23,94 @@ fn make_client() -> reqwest::Client {
.expect("reqwest client to build")
}
macro_rules! test_discover_hostmeta {
($id:ident, $url:literal) => {
test_discover_hostmeta!($id, $url,);
};
($id:ident, $url:literal, $($tag:meta),*) => {
mod $id {
use acrate_hostmeta::*;
use super::*;
#[tokio::test]
$(#[$tag])*
async fn $id() {
init_log();
let client = make_client();
#[tokio::test]
$(#[$tag])*
async fn test() {
init_log();
let client = make_client();
let base: reqwest::Url = $url.parse()
.expect("a valid URL");
let base: reqwest::Url = $url.parse()
.expect("a valid URL");
let doc = acrate_hostmeta::any::ResourceDescriptor::discover_hostmeta(&client, base)
.await
.expect("host-meta discovery to succeed");
let doc = ResourceDescriptor::discover_hostmeta(&client, base)
.await
.expect("host-meta discovery to succeed");
log::info!("Parsed host-meta document: {doc:#?}");
}
log::info!("Parsed host-meta document: {doc:#?}");
}
};
}
test_discover_hostmeta!(akkoma, "https://junimo.party");
test_discover_hostmeta!(mastodon, "https://mastodon.social");
test_discover_hostmeta!(misskey, "https://misskey.io");
test_discover_hostmeta!(iceshrimpnet, "https://ice.frieren.quest");
test_discover_hostmeta!(gotosocial, "https://alpha.polymaths.social");
test_discover_hostmeta!(bridgyfed, "https://fed.brid.gy");
test_discover_hostmeta!(threads, "https://threads.net", ignore = "does not support host-meta");
test_discover_hostmeta!(snac, "https://ngoa.giao.loan", ignore = "does not support host-meta");
test_discover_hostmeta!(hollo, "https://hollo.social", ignore = "does not support host-meta");
macro_rules! test_de_ser_jrd {
($id:ident, $file:literal) => {
test_de_ser_jrd!($id, $file,);
};
($id:ident, $file:literal, $($tag:meta),*) => {
const JRD_DOCUMENT: &'static str = include_str!($file);
#[test]
$(#[$tag])*
fn $id() {
init_log();
log::info!("Starting document: {:#?}", JRD_DOCUMENT);
let de: acrate_hostmeta::jrd::ResourceDescriptorJRD = serde_json::from_str(JRD_DOCUMENT)
.expect("document to be deserialized successfully");
log::info!("Serialized document: {de:#?}");
let ser = serde_json::to_string(&de)
.expect("document to be serialized successfully");
log::info!("Deserialized document: {ser:#?}");
}
};
}
macro_rules! test_de_ser_xrd {
($id:ident, $file:literal) => {
test_de_ser_xrd!($id, $file,);
};
($id:ident, $file:literal, $($tag:meta),*) => {
const XRD_DOCUMENT: &'static str = include_str!($file);
#[test]
$(#[$tag])*
fn $id() {
init_log();
let client = make_client();
log::info!("Starting document: {:#?}", XRD_DOCUMENT);
let de: acrate_hostmeta::xrd::ResourceDescriptorXRD = quick_xml::de::from_str(XRD_DOCUMENT)
.expect("document to be deserialized successfully");
log::info!("Serialized document: {de:#?}");
let ser = quick_xml::se::to_string(&de)
.expect("document to be serialized successfully");
log::info!("Deserialized document: {ser:#?}");
}
};
}
test_discover_hostmeta!(test_discover_hostmeta_junimo_party, "https://junimo.party");
test_discover_hostmeta!(test_discover_hostmeta_mastodon_social, "https://mastodon.social");
test_discover_hostmeta!(test_discover_hostmeta_misskey_io, "https://misskey.io");
test_discover_hostmeta!(test_discover_hostmeta_ice_frieren_quest, "https://ice.frieren.quest");
test_discover_hostmeta!(test_discover_hostmeta_alpha_polymaths_social, "https://alpha.polymaths.social");
test_discover_hostmeta!(test_discover_hostmeta_fed_brid_gy, "https://fed.brid.gy");
test_discover_hostmeta!(test_discover_hostmeta_threads_net, "https://threads.net", ignore = "does not support host-meta");
test_discover_hostmeta!(test_discover_hostmeta_ngoa_giao_loan, "https://ngoa.giao.loan", ignore = "does not support host-meta");
test_discover_hostmeta!(test_discover_hostmeta_hollo_social, "https://hollo.social", ignore = "does not support host-meta");
test_de_ser_jrd!(test_de_ser_sample_junimo_party, "samples/junimo_party.nodeinfo.jrd.json");

View file

@ -0,0 +1 @@
<?xml version="1.0" encoding="UTF-8"?><XRD xmlns="http://docs.oasis-open.org/ns/xri/xrd-1.0"><Link type="application/xrd+xml" template="https://junimo.party/.well-known/webfinger?resource={uri}" rel="lrdd" /></XRD>

View file

@ -0,0 +1,12 @@
{
"links": [
{
"rel": "http://nodeinfo.diaspora.software/ns/schema/2.0",
"href": "https://junimo.party/nodeinfo/2.0.json"
},
{
"rel": "http://nodeinfo.diaspora.software/ns/schema/2.1",
"href": "https://junimo.party/nodeinfo/2.1.json"
}
]
}

View file

@ -4,7 +4,7 @@ version = "0.1.0"
edition = "2021"
[dependencies]
acrate-hostmeta = { version = "0.1.0", path = "../acrate-hostmeta" }
acrate-hostmeta = { path = "../acrate-hostmeta" }
log = "0.4.22"
reqwest = { version = "0.12.9", features = ["json", "stream"] }
serde = { version = "1.0.214", features = ["derive"] }

View file

@ -172,7 +172,7 @@ impl NodeInfo {
/// # Specification
///
/// - <https://github.com/jhass/nodeinfo/blob/main/PROTOCOL.md#discovery>
pub const WELLKNOWN_NODEINFO_PATH: &str = "/.well-known/nodeinfo";
pub const WELLKNOWN_NODEINFO_PATH: &'static str = "/.well-known/nodeinfo";
/// Discover and get the latest NodeInfo version available given a certain base URL.
///
@ -206,12 +206,13 @@ impl NodeInfo {
base.set_path(Self::WELLKNOWN_NODEINFO_PATH);
log::trace!("Discovering NodeInfo document locations...");
let discovery = acrate_hostmeta::ResourceDescriptor::discover(client, base)
let discovery = acrate_hostmeta::any::ResourceDescriptor::get(client, base)
.await
.map_err(Discovery)?;
.map_err(Get)?
.jrd();
log::trace!("Getting a list of NodeInfo document links...");
let mut links = discovery.links.unwrap_or_default();
let mut links = discovery.links;
links.sort_unstable_by_key(|o| o.rel.clone()); // TODO: Performance can be improved.
links.reverse();
@ -314,7 +315,7 @@ impl NodeInfo {
pub enum NodeInfoGetWellknownError {
/// The discovery of possible locations for NodeInfo documents failed.
#[error("the discovery of possible locations for NodeInfo documents failed")]
Discovery(acrate_hostmeta::ResourceDescriptorDiscoveryError),
Get(acrate_hostmeta::any::GetError),
/// No compatible NodeInfo documents were detected at the given URL.
#[error("no compatible NodeInfo documents were detected at the given URL")]
Unsupported,