identifiers: Add MatrixToUri parsing

crates/ruma-identifiers-validation/Cargo.toml

@@ -15,3 +15,4 @@ compat = []
 
 [dependencies]
 thiserror = "1.0.26"
+url = "2.2.2"

crates/ruma-identifiers-validation/src/error.rs

@@ -1,5 +1,7 @@
 //! Error conditions.
 
+use std::str::Utf8Error;
+
 /// An error encountered when trying to parse an invalid ID string.
 #[derive(Copy, Clone, Debug, Eq, Hash, PartialEq, thiserror::Error)]
 #[non_exhaustive]
@@ -30,6 +32,14 @@ pub enum Error {
     #[error("key ID version contains invalid characters")]
     InvalidKeyVersion,
 
+    /// The string isn't a valid Matrix ID.
+    #[error("invalid matrix ID: {0}")]
+    InvalidMatrixId(#[from] MatrixIdError),
+
+    /// The string isn't a valid Matrix.to URI.
+    #[error("invalid matrix.to URI: {0}")]
+    InvalidMatrixToRef(#[from] MatrixToError),
+
     /// The mxc:// isn't a valid Matrix Content URI.
     #[error("invalid Matrix Content URI: {0}")]
     InvalidMxcUri(#[from] MxcUriError),
@@ -38,6 +48,14 @@ pub enum Error {
     #[error("server name is not a valid IP address or domain name")]
     InvalidServerName,
 
+    /// The string isn't a valid URI.
+    #[error("invalid URI")]
+    InvalidUri,
+
+    /// The string isn't valid UTF-8.
+    #[error("invalid UTF-8")]
+    InvalidUtf8,
+
     /// The ID exceeds 255 bytes (or 32 codepoints for a room version ID).
     #[error("ID exceeds 255 bytes")]
     MaximumLengthExceeded,
@@ -52,6 +70,18 @@ pub enum Error {
     MissingLeadingSigil,
 }
 
+impl From<Utf8Error> for Error {
+    fn from(_: Utf8Error) -> Self {
+        Self::InvalidUtf8
+    }
+}
+
+impl From<url::ParseError> for Error {
+    fn from(_: url::ParseError) -> Self {
+        Self::InvalidUri
+    }
+}
+
 /// An error occurred while validating an MXC URI.
 #[derive(Copy, Clone, Debug, Eq, Hash, PartialEq, thiserror::Error)]
 #[non_exhaustive]
@@ -75,3 +105,41 @@ pub enum MxcUriError {
     #[error("invalid Server Name")]
     ServerNameMalformed,
 }
+
+/// An error occurred while validating a `MatrixId`.
+#[derive(Copy, Clone, Debug, Eq, Hash, PartialEq, thiserror::Error)]
+#[non_exhaustive]
+pub enum MatrixIdError {
+    /// The string is missing a room ID or alias.
+    #[error("missing room ID or alias")]
+    MissingRoom,
+
+    /// The string contains no identifier.
+    #[error("no identifier")]
+    NoIdentifier,
+
+    /// The string contains too many identifiers.
+    #[error("too many identifiers")]
+    TooManyIdentifiers,
+
+    /// The string contains an unknown identifier.
+    #[error("unknown identifier")]
+    UnknownIdentifier,
+
+    /// The string contains two identifiers that cannot be paired.
+    #[error("unknown identifier pair")]
+    UnknownIdentifierPair,
+}
+
+/// An error occurred while validating an MXC URI.
+#[derive(Copy, Clone, Debug, Eq, Hash, PartialEq, thiserror::Error)]
+#[non_exhaustive]
+pub enum MatrixToError {
+    /// String did not start with `https://matrix.to/#/`.
+    #[error("base URL is not https://matrix.to/#/")]
+    WrongBaseUrl,
+
+    /// String has an unknown additional argument.
+    #[error("unknown additional argument")]
+    UnknownArgument,
+}

crates/ruma-identifiers/CHANGELOG.md

@@ -15,6 +15,7 @@ Breaking changes:
 Improvements:
 
 * Add `host`, `port` and `is_ip_literal` methods to `ServerName`
+* Add `matrix_uri::MatrixToUri` to build `matrix.to` URIs
 
 Bug fixes:
 

crates/ruma-identifiers/Cargo.toml

@@ -30,6 +30,7 @@ ruma-serde = { version = "0.5.0", path = "../ruma-serde", optional = true }
 ruma-serde-macros = { version = "0.5.0", path = "../ruma-serde-macros" }
 # Renamed so we can have a serde feature.
 serde1 = { package = "serde", version = "1.0.126", optional = true, features = ["derive"] }
+url = "2.2.2"
 uuid = { version = "0.8.2", optional = true, features = ["v4"] }
 
 [dev-dependencies]

crates/ruma-identifiers/src/matrix_uri.rs

@@ -1,8 +1,13 @@
 //! Matrix URIs.
 
-use std::fmt;
+use std::{convert::TryFrom, fmt};
 
-use percent_encoding::{percent_encode, AsciiSet, CONTROLS};
+use percent_encoding::{percent_decode_str, percent_encode, AsciiSet, CONTROLS};
+use ruma_identifiers_validation::{
+    error::{MatrixIdError, MatrixToError},
+    Error,
+};
+use url::Url;
 
 use crate::{EventId, RoomAliasId, RoomId, RoomOrAliasId, ServerName, UserId};
 
@@ -48,6 +53,54 @@ pub enum MatrixId {
 }
 
 impl MatrixId {
+    /// Try parsing a `&str` with sigils into a `MatrixId`.
+    ///
+    /// The identifiers are expected to start with a sigil and to be percent
+    /// encoded. Slashes at the beginning and the end are stripped.
+    ///
+    /// For events, the room ID or alias and the event ID should be separated by
+    /// a slash and they can be in any order.
+    pub(crate) fn parse_with_sigil(s: &str) -> Result<Self, Error> {
+        let s = if let Some(stripped) = s.strip_prefix('/') { stripped } else { s };
+        let s = if let Some(stripped) = s.strip_suffix('/') { stripped } else { s };
+        if s.is_empty() {
+            return Err(MatrixIdError::NoIdentifier.into());
+        }
+
+        if s.matches('/').count() > 1 {
+            return Err(MatrixIdError::TooManyIdentifiers.into());
+        }
+
+        if let Some((first_raw, second_raw)) = s.split_once('/') {
+            let first = percent_decode_str(first_raw).decode_utf8()?;
+            let second = percent_decode_str(second_raw).decode_utf8()?;
+
+            match first.as_bytes()[0] {
+                b'!' | b'#' if second.as_bytes()[0] == b'$' => {
+                    let room_id = <&RoomOrAliasId>::try_from(first.as_ref())?;
+                    let event_id = <&EventId>::try_from(second.as_ref())?;
+                    Ok((room_id, event_id).into())
+                }
+                b'$' if matches!(second.as_bytes()[0], b'!' | b'#') => {
+                    let room_id = <&RoomOrAliasId>::try_from(second.as_ref())?;
+                    let event_id = <&EventId>::try_from(first.as_ref())?;
+                    Ok((room_id, event_id).into())
+                }
+                _ => Err(MatrixIdError::UnknownIdentifierPair.into()),
+            }
+        } else {
+            let id = percent_decode_str(s).decode_utf8()?;
+
+            match id.as_bytes()[0] {
+                b'@' => Ok(<&UserId>::try_from(id.as_ref())?.into()),
+                b'!' => Ok(<&RoomId>::try_from(id.as_ref())?.into()),
+                b'#' => Ok(<&RoomAliasId>::try_from(id.as_ref())?.into()),
+                b'$' => Err(MatrixIdError::MissingRoom.into()),
+                _ => Err(MatrixIdError::UnknownIdentifier.into()),
+            }
+        }
+    }
+
     /// Construct a string with sigils from `self`.
     ///
     /// The identifiers will start with a sigil and be percent encoded.
@@ -130,6 +183,30 @@ impl MatrixToUri {
     pub fn via(&self) -> &[Box<ServerName>] {
         &self.via
     }
+
+    /// Try parsing a `&str` into a `MatrixToUri`.
+    pub fn parse(s: &str) -> Result<Self, Error> {
+        let without_base = if let Some(stripped) = s.strip_prefix(MATRIX_TO_BASE_URL) {
+            stripped
+        } else {
+            return Err(MatrixToError::WrongBaseUrl.into());
+        };
+
+        let url = Url::parse(MATRIX_TO_BASE_URL.trim_end_matches("#/"))?.join(without_base)?;
+
+        let id = MatrixId::parse_with_sigil(url.path())?;
+        let mut via = vec![];
+
+        for (key, value) in url.query_pairs() {
+            if key.as_ref() == "via" {
+                via.push(ServerName::parse(value)?);
+            } else {
+                return Err(MatrixToError::UnknownArgument.into());
+            }
+        }
+
+        Ok(Self { id, via })
+    }
 }
 
 impl fmt::Display for MatrixToUri {
@@ -151,7 +228,14 @@ impl fmt::Display for MatrixToUri {
 
 #[cfg(test)]
 mod tests {
-    use crate::{event_id, room_alias_id, room_id, server_name, user_id};
+    use matches::assert_matches;
+    use ruma_identifiers_validation::{
+        error::{MatrixIdError, MatrixToError},
+        Error,
+    };
+
+    use super::{MatrixId, MatrixToUri};
+    use crate::{event_id, room_alias_id, room_id, server_name, user_id, RoomOrAliasId};
 
     #[test]
     fn display_matrixtouri() {
@@ -182,4 +266,204 @@ mod tests {
             "https://matrix.to/#/%21ruma%3Anotareal.hs/%24event%3Anotareal.hs"
         );
     }
+
+    #[test]
+    fn parse_valid_matrixid_with_sigil() {
+        assert_eq!(
+            MatrixId::parse_with_sigil("@user:imaginary.hs").expect("Failed to create MatrixId."),
+            MatrixId::User(user_id!("@user:imaginary.hs").into())
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("!roomid:imaginary.hs").expect("Failed to create MatrixId."),
+            MatrixId::Room(room_id!("!roomid:imaginary.hs").into())
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("#roomalias:imaginary.hs")
+                .expect("Failed to create MatrixId."),
+            MatrixId::RoomAlias(room_alias_id!("#roomalias:imaginary.hs").into())
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("!roomid:imaginary.hs/$event:imaginary.hs")
+                .expect("Failed to create MatrixId."),
+            MatrixId::Event(
+                <&RoomOrAliasId>::from(room_id!("!roomid:imaginary.hs")).into(),
+                event_id!("$event:imaginary.hs").into()
+            )
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("#roomalias:imaginary.hs/$event:imaginary.hs")
+                .expect("Failed to create MatrixId."),
+            MatrixId::Event(
+                <&RoomOrAliasId>::from(room_alias_id!("#roomalias:imaginary.hs")).into(),
+                event_id!("$event:imaginary.hs").into()
+            )
+        );
+        // Invert the order of the event and the room.
+        assert_eq!(
+            MatrixId::parse_with_sigil("$event:imaginary.hs/!roomid:imaginary.hs")
+                .expect("Failed to create MatrixId."),
+            MatrixId::Event(
+                <&RoomOrAliasId>::from(room_id!("!roomid:imaginary.hs")).into(),
+                event_id!("$event:imaginary.hs").into()
+            )
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("$event:imaginary.hs/#roomalias:imaginary.hs")
+                .expect("Failed to create MatrixId."),
+            MatrixId::Event(
+                <&RoomOrAliasId>::from(room_alias_id!("#roomalias:imaginary.hs")).into(),
+                event_id!("$event:imaginary.hs").into()
+            )
+        );
+        // Starting with a slash
+        assert_eq!(
+            MatrixId::parse_with_sigil("/@user:imaginary.hs").expect("Failed to create MatrixId."),
+            MatrixId::User(user_id!("@user:imaginary.hs").into())
+        );
+        // Ending with a slash
+        assert_eq!(
+            MatrixId::parse_with_sigil("!roomid:imaginary.hs/")
+                .expect("Failed to create MatrixId."),
+            MatrixId::Room(room_id!("!roomid:imaginary.hs").into())
+        );
+        // Starting and ending with a slash
+        assert_eq!(
+            MatrixId::parse_with_sigil("/#roomalias:imaginary.hs/")
+                .expect("Failed to create MatrixId."),
+            MatrixId::RoomAlias(room_alias_id!("#roomalias:imaginary.hs").into())
+        );
+    }
+
+    #[test]
+    fn parse_matrixid_no_identifier() {
+        assert_eq!(MatrixId::parse_with_sigil("").unwrap_err(), MatrixIdError::NoIdentifier.into());
+        assert_eq!(
+            MatrixId::parse_with_sigil("/").unwrap_err(),
+            MatrixIdError::NoIdentifier.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixid_too_many_identifiers() {
+        assert_eq!(
+            MatrixId::parse_with_sigil(
+                "@user:imaginary.hs/#room:imaginary.hs/$event1:imaginary.hs"
+            )
+            .unwrap_err(),
+            MatrixIdError::TooManyIdentifiers.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixid_unknown_identifier_pair() {
+        assert_eq!(
+            MatrixId::parse_with_sigil("!roomid:imaginary.hs/@user:imaginary.hs").unwrap_err(),
+            MatrixIdError::UnknownIdentifierPair.into()
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("#roomalias:imaginary.hs/notanidentifier").unwrap_err(),
+            MatrixIdError::UnknownIdentifierPair.into()
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("$event:imaginary.hs/$otherevent:imaginary.hs").unwrap_err(),
+            MatrixIdError::UnknownIdentifierPair.into()
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("notanidentifier/neitheristhis").unwrap_err(),
+            MatrixIdError::UnknownIdentifierPair.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixid_missing_room() {
+        assert_eq!(
+            MatrixId::parse_with_sigil("$event:imaginary.hs").unwrap_err(),
+            MatrixIdError::MissingRoom.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixid_unknown_identifier() {
+        assert_eq!(
+            MatrixId::parse_with_sigil("event:imaginary.hs").unwrap_err(),
+            MatrixIdError::UnknownIdentifier.into()
+        );
+        assert_eq!(
+            MatrixId::parse_with_sigil("notanidentifier").unwrap_err(),
+            MatrixIdError::UnknownIdentifier.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixtouri_valid_uris() {
+        let matrix_to = MatrixToUri::parse("https://matrix.to/#/%40jplatte%3Anotareal.hs")
+            .expect("Failed to create MatrixToUri.");
+        assert_eq!(matrix_to.id(), &user_id!("@jplatte:notareal.hs").into());
+
+        let matrix_to = MatrixToUri::parse("https://matrix.to/#/%23ruma%3Anotareal.hs")
+            .expect("Failed to create MatrixToUri.");
+        assert_eq!(matrix_to.id(), &room_alias_id!("#ruma:notareal.hs").into());
+
+        let matrix_to =
+            MatrixToUri::parse("https://matrix.to/#/%21ruma%3Anotareal.hs?via=notareal.hs")
+                .expect("Failed to create MatrixToUri.");
+        assert_eq!(matrix_to.id(), &room_id!("!ruma:notareal.hs").into());
+        assert_eq!(matrix_to.via(), &vec![server_name!("notareal.hs").to_owned()]);
+
+        let matrix_to =
+            MatrixToUri::parse("https://matrix.to/#/%23ruma%3Anotareal.hs/%24event%3Anotareal.hs")
+                .expect("Failed to create MatrixToUri.");
+        assert_eq!(
+            matrix_to.id(),
+            &(room_alias_id!("#ruma:notareal.hs"), event_id!("$event:notareal.hs")).into()
+        );
+
+        let matrix_to =
+            MatrixToUri::parse("https://matrix.to/#/%21ruma%3Anotareal.hs/%24event%3Anotareal.hs")
+                .expect("Failed to create MatrixToUri.");
+        assert_eq!(
+            matrix_to.id(),
+            &(room_id!("!ruma:notareal.hs"), event_id!("$event:notareal.hs")).into()
+        );
+        assert!(matrix_to.via().is_empty());
+    }
+
+    #[test]
+    fn parse_matrixtouri_wrong_base_url() {
+        assert_eq!(MatrixToUri::parse("").unwrap_err(), MatrixToError::WrongBaseUrl.into());
+        assert_eq!(
+            MatrixToUri::parse("https://notreal.to/#/").unwrap_err(),
+            MatrixToError::WrongBaseUrl.into()
+        );
+    }
+
+    #[test]
+    fn parse_matrixtouri_wrong_identifier() {
+        assert_matches!(
+            MatrixToUri::parse("https://matrix.to/#/notanidentifier").unwrap_err(),
+            Error::InvalidMatrixId(_)
+        );
+        assert_matches!(
+            MatrixToUri::parse("https://matrix.to/#/").unwrap_err(),
+            Error::InvalidMatrixId(_)
+        );
+        assert_matches!(
+            MatrixToUri::parse(
+                "https://matrix.to/#/%40jplatte%3Anotareal.hs/%24event%3Anotareal.hs"
+            )
+            .unwrap_err(),
+            Error::InvalidMatrixId(_)
+        );
+    }
+
+    #[test]
+    fn parse_matrixtouri_unknown_arguments() {
+        assert_eq!(
+            MatrixToUri::parse(
+                "https://matrix.to/#/%21ruma%3Anotareal.hs?via=notareal.hs&custom=data"
+            )
+            .unwrap_err(),
+            MatrixToError::UnknownArgument.into()
+        )
+    }
 }